{"_id":"q-en-astropy__astropy-11693-17bb2019924814a4b9f0352623cc9b397e57044b8ca4c9f7fefe791e4009ad89","text":"'WCS.allworld2pix' failed to converge when plotting WCS with non linear distortions <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --When trying to plot an image with a WCS as projection that contains non linear Distortions it fails with a error. When I add as parameter to the call at line 326 of I get the good enough looking plot below: ! It would be nice if there was a way of getting that plot without having to hack the library code like that. <!-- What actually happened. --<!-- Was the output confusing or poorly described? --The call to plotting the grid fails with the following error (last few lines, can provide more if necessary): <!-- Ideally a code example could be provided so we can run it ourselves. --<!-- If you are pasting code, use triple backticks ( from import WCS, Sip import numpy as np import as plt wcs = WCS(naxis=2) a = [[ 0.+00, 0.+00, 6.-07, -1.-10], [ 0.+00, 9.-06, -1.-07, 0.+00], [ 7.-06, 2.-09, 0.+00, 0.+00], [-1.-07, 0.+00, 0.+00, 0.+00]] b = [[ 0.+00, 0.+00, 1.-05, -1.-07], [ 0.+00, 5.-06, -1.-09, 0.+00], [ 1.-05, -1.-07, 0.+00, 0.+00], [ 3.-09, 0.+00, 0.+00, 0.+00]] crpix = [1221., 994.] ap = bp = ((4, 4)) = Sip(a, b, ap, bp, crpix) plt.subplot(projection=wcs) (((1944, 2592))) (color='white', ls='solid') import platform; print(platform.platform()) Linux-5.11.10-arch1-1-x8664-with-glibc2.33 import sys; print(\"Python\", sys.version) Python 3.9.2 (default, Feb 20 2021, 18:40:11) [GCC 10.2.0] import numpy; print(\"Numpy\", numpy.version) Numpy 1.20.2 import astropy; print(\"astropy\", astropy.version) astropy 4.3.dev690+g7811614f8 import scipy; print(\"Scipy\", scipy.version) Scipy 1.6.1 import matplotlib; print(\"Matplotlib\", matplotlib.version) Matplotlib 3.3.4 pixel = self.allworld2pix(worldarrays, 0) ~\/work\/develop\/env\/lib\/python3.9\/site- in worldtopixelvalues(self, worldarrays) 324 325 def worldtopixelvalues(self, worldarrays): --326 pixel = self.allworld2pix(worldarrays, 0) 327 return pixel[0] if self.pixelndim == 1 else tuple(pixel) 328 ~\/work\/develop\/env\/lib\/python3.9\/site- in wrapper(args, kwargs) 534 (message, warningtype, stacklevel=2) 535 --536 return function(args, kwargs) 537 538 return wrapper ~\/work\/develop\/env\/lib\/python3.9\/site- in allworld2pix(self, tolerance, maxiter, adaptive, detectdivergence, quiet, args, kwargs) 1886 raise ValueError(\"No basic WCS settings were created.\") 1887 -1888 return self.arrayconverter( 1889 lambda args, kwargs: 1890 self.allworld2pix( ~\/work\/develop\/env\/lib\/python3.9\/site- in arrayconverter(self, func, sky, radecorder, args) 1335 \"a 1-D array for each axis, followed by an origin.\") 1336 -1337 return returnlistofarrays(axes, origin) 1338 1339 raise TypeError( ~\/work\/develop\/env\/lib\/python3.9\/site- in returnlistofarrays(axes, origin) 1289 if radecorder and sky == 'input': 1290 xy = self.denormalizesky(xy) -1291 output = func(xy, origin) 1292 if radecorder and sky == 'output': 1293 output = self.normalizesky(output) ~\/work\/develop\/env\/lib\/python3.9\/site- in <lambda(args, kwargs) 1888 return self.arrayconverter( 1889 lambda args, kwargs: -1890 self.allworld2pix( 1891 args, tolerance=tolerance, maxiter=maxiter, 1892 adaptive=adaptive, detectdivergence=detectdivergence, ~\/work\/develop\/env\/lib\/python3.9\/site- in allworld2pix(self, world, origin, tolerance, maxiter, adaptive, detectdivergence, quiet) 1869 slowconv=ind, divergent=None) 1870 else: -1871 raise NoConvergence( 1872 \"'WCS.allworld2pix' failed to \" 1873 \"converge to the requested accuracy.\\n\" NoConvergence: 'WCS.all_world2pix' failed to converge to the requested accuracy. After 20 iterations, the solution is diverging at least for one input point. ) around your code snippet. --<!-- If necessary, sanitize your screen output to be pasted so you do not reveal secrets like tokens and passwords. --Here is the code to reproduce the problem: <!-- Even if you do not think this is necessary, it is useful information for the maintainers. Please run the following snippet and paste the output below: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"Numpy\", numpy.version) import astropy; print(\"astropy\", astropy.version) import scipy; print(\"Scipy\", scipy.version) import matplotlib; print(\"Matplotlib\", matplotlib.version) --"}
{"_id":"q-en-astropy__astropy-12057-a60bb5063305ebc8884284067c66dca1bfc707e0231323da9ef687632392df95","text":"Add helpers to convert between different types of uncertainties Currently there no easy way to convert from an arbitrary uncertainty class to a different uncertainty class. This would be useful to be able to pass NDData objects to external libraries\/tools which assume, for example, that uncertainties will always stored as variances. Here's some really scrappy code I bunged together quickly for my purposes (probably buggy, I need to properly test it), but what are peoples opinions on what's the best API\/design\/framework for such a system?"}
{"_id":"q-en-astropy__astropy-12318-6db372c785e844eaa78b207f9d1024131e4407ac7bce97c08769f2e14f3170a2","text":"BlackBody bolometric flux is wrong if scale has units of dimensionlessunscaled The class has the wrong bolometric flux if argument is passed as a Quantity with units, but the correct bolometric flux if is simply a float. <!-- Provide a general description of the bug. --Expected output from sample code: Actual output from sample code: Sample code:"}
{"_id":"q-en-astropy__astropy-12544-2c27deb15727d656f35ce94691a2d30f5e1d07363b9b43d0c7fe4fad42ccd305","text":"Can Table masking be turned off? <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Provide a general description of the feature you would like. --<!-- If you want to, you can suggest a draft design or API. --<!-- This way we have a deeper discussion on the feature. --As of Astropy 5, when encounters values such as , it automatically creates a and the whole table becomes a . While this might be useful for individual end-users, it is very inconvenient for intermediate data in pipelines. Here's the scenario: data are being passed via files and . A downstream function needs to replace with valid values. Previously those values could be easily identified (e.g. and replaced. However, now additional work is need to look \"underneath\" the mask, extracting the actual values, replacing them, and then possibly creating a new, unmasked column, or even an entirely new table. Ideally, a keyword like would disable this behavior, for people who don't need this masking."}
{"_id":"q-en-astropy__astropy-12825-5fd82bd1729d08671ed5e14e714a859f36b9a1ec6d9ffc5d38255f00997ad44e","text":"SkyCoord in Table breaks aggregate on groupby When putting a column of s in a , does not work on : fails with This happens irregardless of the aggregation function. Aggregation works, only fails to aggregate columns where operation does not make sense. and"}
{"_id":"q-en-astropy__astropy-12880-960b15832070f2ce90be0d5644c1af18fa025af994b7cf3d39d64f92b4ef960b","text":"No longer able to read BinnedTimeSeries with datetime column saved as ECSV after upgrading from 4.2.1 -5.0+ <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --<!-- Provide a general description of the bug. --Hi, merged in PR breaks my ability to read an ECSV file created using Astropy v 4.2.1, BinnedTimeSeries class's write method, which has a datetime64 column. Downgrading astropy back to 4.2.1 fixes the issue because the strict type checking in line 177 of is not there. Is there a reason why this strict type checking was to ECSV? Is there a way to preserve reading and writing of ECSV files created with BinnedTimeSeries across versions? I am happy to make a PR on this if the strict type checking is allowed to be scaled back or we can add datetime64 as an allowed type. <!-- What did you expect to happen. --The file is read into a object from ecsv file without error. <!-- What actually happened. --<!-- Was the output confusing or poorly described? --ValueError is produced and the file is not read because does not accept the datetime64 column. <!-- Ideally a code example could be provided so we can run it ourselves. --<!-- If you are pasting code, use triple backticks ( ValueError Traceback (most recent call last) Input In [3], in <module---49 tsrbin = ('', format='') File ~\/Apps\/miniconda3\/envs\/py310latest\/lib\/python3.10\/site-, in (self, filename, timebinstartcolumn, timebinendcolumn, timebinsizecolumn, timebinsizeunit, timeformat, timescale, format, args, kwargs) 230 \"\"\" 231 Read and parse a file and returns a . 232 (...) 279 280 \"\"\" 282 try: 283 284 # First we try the readers defined for the BinnedTimeSeries class --285 return super().read(filename, format=format, args, kwargs) 287 except TypeError: 288 289 # Otherwise we fall back to the default Table readers 291 if timebinstartcolumn is None: File ~\/Apps\/miniconda3\/envs\/py310latest\/lib\/python3.10\/site-, in (self, args, kwargs) 59 units = ('units', None) 60 descriptions = ('descriptions', None) ---62 out = (cls, args, kwargs) 64 # For some readers (e.g., ), the returned class is not 65 # guaranteed to be the same as the desired output . If so, 66 # try coercing to desired class without copying ( 67 # would normally do a copy). The normal case here is swapping 68 # Table <=QTable. 69 if cls is not File ~\/Apps\/miniconda3\/envs\/py310latest\/lib\/python3.10\/site-, in (self, cls, format, cache, args, kwargs) 195 format = self.getvalidformat( 196 'read', cls, path, fileobj, args, kwargs) 198 reader = self.getreader(format, cls) --199 data = reader(args, kwargs) 201 if not isinstance(data, cls): 202 # User has read with a subclass where only the parent class is 203 # registered. This returns the parent class, so try coercing 204 # to desired subclass. 205 try: File ~\/Apps\/miniconda3\/envs\/py310latest\/lib\/python3.10\/site-, in ioread(format, filename, kwargs) 16 format = (r'^ascii.', '', format) 17 kwargs['format'] = format ---18 return read(filename, kwargs) File ~\/Apps\/miniconda3\/envs\/py310latest\/lib\/python3.10\/site-, in read(table, guess, kwargs) 374 else: 375 reader = getreader(newkwargs) --376 dat = (table) 377 ({'kwargs': copy.deepcopy(newkwargs), 378 'Reader': , 379 'status': 'Success with specified Reader class ' 380 '(no guessing)'}) 382 # Static analysis (pyright) indicates might be left undefined, so just 383 # to be sure define it at the beginning and check here. File ~\/Apps\/miniconda3\/envs\/py310latest\/lib\/python3.10\/site-, in (self, table) 1340 (, ) 1342 # Get the table column definitions -1343 () 1345 # Make sure columns are valid 1346 (, self.strictnames, self.guessing) File ~\/Apps\/miniconda3\/envs\/py310latest\/lib\/python3.10\/site-, in EcsvHeader.getcols(self, lines) 175 = headercols[]['datatype'] 176 if not in ECSVDATATYPES: --177 raise ValueError(f'datatype {!r} of column {!r} ' 178 f'is not in allowed values {ECSVDATATYPES}') 180 # Subtype is written like \"int64[2,null]\" and we want to split this 181 # out to \"int64\" and [2, None]. 182 subtype = col.subtype ValueError: datatype 'datetime64' of column 'timebinstart' is not in allowed values ('bool', 'int8', 'int16', 'int32', 'int64', 'uint8', 'uint16', 'uint32', 'uint64', 'float16', 'float32', 'float64', 'float128', 'string') <!-- Even if you do not think this is necessary, it is useful information for the maintainers. Please run the following snippet and paste the output below: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"Numpy\", numpy.version) import erfa; print(\"pyerfa\", erfa.version) import astropy; print(\"astropy\", astropy.version) import scipy; print(\"Scipy\", scipy.version) import matplotlib; print(\"Matplotlib\", matplotlib.version) --(For the version that does not work) Python 3.10.2 (main, Feb 1 2022, 19:28:35) [GCC 9.4.0] Numpy 1.22.2 pyerfa 2.0.0.1 astropy 5.0.1 Scipy 1.8.0 Matplotlib 3.5.1 (For the version that does work) Python 3.7.11 (default, Jul 27 2021, 14:32:16) [GCC 7.5.0] Numpy 1.20.3 pyerfa 2.0.0.1 astropy 4.2.1 Scipy 1.7.0 Matplotlib 3.4.2"}
{"_id":"q-en-astropy__astropy-12891-87f18488e000f63a51e25653ad5d40caa18bbfa82bc4a81254511fd8f2a73219","text":"The keyword argument of is not supported for instances. <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Provide a general description of the feature you would like. --<!-- If you want to, you can suggest a draft design or API. --<!-- This way we have a deeper discussion on the feature. --Apologies if there is a duplicate somewhere, I scoured all the issues for this problem and I couldn't find it mentioned yet. The keyword argument was to and all the other -based functions in version 1.20.0 of numpy, but it doesn't seem to work yet with . Does anyone know if there is some limitation in that is preventing this feature from being implemented? If not, I could put some time towards updating to support this feature. <!-- Add any other context or screenshots about the feature request here. --<!-- This part is optional. --"}
{"_id":"q-en-astropy__astropy-12907-4d00b55531b654c13f089639989a04f825d0dd58eab7b09492f133b9ff2d32ab","text":"Modeling's does not compute separability correctly for nested CompoundModels Consider the following model: It's separability matrix as you might expect is a diagonal: If I make the model more complex: The output matrix is again, as expected, the outputs and inputs to the linear models are separable and independent of each other. If however, I nest these compound models: Suddenly the inputs and outputs are no longer separable? This feels like a bug to me, but I might be missing something?"}
{"_id":"q-en-astropy__astropy-12962-4d6c31950f3923d748cf998f705a7ad66bc56a400f2bb97d31b5b6f5bc64f5e9","text":"Convert CCDData to ImageHDU As far as I can tell, currently there's no way to directly convert a object to an object. If we write it to a file using it will always create a file where the first HDU is a that contains the , followed by optionally some s that contain mask or uncertainty. If we instead use , it will return an equivalent to the file it writes with , that is, the is stored in the first element of the , which is always a . This is somewhat unexpected given that you can already do it the other way around (that is, convert a object to a object): The problem is that if we then want to use this as an extension to another FITS file, there's no obvious way to get an which would allow us to do that. As far as I can tell, there's also no trivial way to convert a to a . We could manually create a new by copying the data from the , as well as its relevant cards and so on... but this seems unnecessarily complicated. I propose the following interfaces: This seems somewhat similar to the situation with and . In that case, we can also specify an parameter when reading: From here we can use:"}
{"_id":"q-en-astropy__astropy-13032-cc516fa612c49295f7d66f7d90cc4ddb6aea5a204cb6475bd5767c0265e434d1","text":"Incorrect ignored usage in <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --<!-- Provide a general description of the bug. --Providing inputs to does not always work as expected. Running the following code: Produces: This is incorrect. It instead should produce: Somehow the status of the input is being accounted for during the validation which occurs during the construction of the bounding box; however, it is getting \"lost\" somehow resulting in the weird behavior we see above. Oddly enough ignoring does not have an issue. E.G. this code: Produces: as expected. This is present in both astropy 5.03 and astropy develop"}
{"_id":"q-en-astropy__astropy-13033-8a583992a75168fdc958b7325888f656a165d695a4755a2e96d1ee80295a87e0","text":"TimeSeries: misleading exception when required column check fails. <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --<!-- Provide a general description of the bug. --For a object that has additional required columns (in addition to ), when codes mistakenly try to remove a required column, the exception it produces is misleading. <!-- What did you expect to happen. --An exception that informs the users required columns are missing. The actual exception message is confusing: <!-- Ideally a code example could be provided so we can run it ourselves. --<!-- If you are pasting code, use triple backticks (python from import Time from astropy.timeseries import TimeSeries time=Time((, ), format='jd') ts = TimeSeries(time=time, data = {\"flux\": [99.9, 99.8, 99.7]}) ts.requiredcolumns = [\"time\", \"flux\"] ts.remove_column(\"flux\") Windows-10-10.0.-SP0 Python 3.9.10 (main, Feb 1 2022, 21:21:54) [MSC v.1929 64 bit (AMD64)] Numpy 1.22.3 pyerfa 2.0.0.1 astropy 5.0.3 Scipy 1.8.0 Matplotlib 3.5.1"}
{"_id":"q-en-astropy__astropy-13068-b5be77b17ea026d4041bfe3ce2d56085ee51de2315b336805decf240d1a83e60","text":"Time from not precise Hello, I encounter difficulties with Time. I'm working on a package to perform photometry and occultation. For this last case, data need times values accurately estimated. Of course, data coming from different camera will will have different time format in the header. to manage this without passing long time to build a time parser, i decided to use Time object which do exactly what i need. The problem is, i dont arrive to make accurate conversion between different format using Time. let's do an exemple: now let's do reverse as you can see i don't fall back on the same value and the difference is quite high. I would like to fall back on the original one. thank you in advance"}
{"_id":"q-en-astropy__astropy-13073-ca7c16488d559bde6fa5d6abda0be36cb1e524104e4fe5dca97f8f56fc2603df","text":"Document reading True\/False in ASCII table as bool not str <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Provide a general description of the feature you would like. --<!-- If you want to, you can suggest a draft design or API. --<!-- This way we have a deeper discussion on the feature. -- showed a use case for having an ASCII table column consisting of only \"True\" and \"False\" be read as instead of (current behavior). That issue discusses reasons to maintain the current behavior, but there are simple workarounds discussed there that should be brought out to the narrative docs as examples for users. I'd suggest the following as a recommendation for users: This new information could go in the existing section on in the read documentation. <!-- Add any other context or screenshots about the feature request here. --<!-- This part is optional. -- Control dtype with (converters API needs better doc or refactoring) I cannot find a way to control the dtype of the output table when reading a file into a . Consider the following MWE, with 3 numerical columns, while one of them would preferably be kept as a string: Reading in the same with and then converting to a Table works, but it should ideally be supported directly."}
{"_id":"q-en-astropy__astropy-13075-c25a4d5314d68f9db5ffd23ada46c415ba6b64d74c62a4d2f23b8d1e1fb69d5f","text":"Register format with nice mathjax Cosmology can now read and write to files. It would be nice to register with a method for exporting a Cosmology to a HTML table. There are good examples of IO with Cosmology at and documentation at I'm thinking the (..., format='html'). Edit: also, the mathjax representation of each parameter can be stored on the corresponding Parameter object, like how units have the argument in ."}
{"_id":"q-en-astropy__astropy-13132-63311e049bac1a2b8176c196e68ec771aa2a192c254ba8be04a99df1d8710297","text":"Add arrayfunc for <!-- This comments are hidden when you submit the pull request, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- If you are new or need to be re-acquainted with Astropy contributing workflow, please see . There is even a practical example at . --<!-- Astropy coding style guidelines can be found here: Our testing infrastructure enforces to follow a subset of the PEP8 to be followed. You can check locally whether your changes have followed these by running the following command: tox -e codestyle --<!-- Please just have a quick search on GitHub to see if a similar pull request has already been posted. We have old closed pull requests that might provide useful code or ideas that directly tie in with your pull request. --<!-- We have several automatic features that run when a pull request is open. They can appear daunting but do not worry because maintainers will help you navigate them, if necessary. --<!-- Provide a general description of what your pull request does. Complete the following sentence and add relevant details as you see fit. --<!-- In addition please ensure that the pull request title is descriptive and allows maintainers to infer the applicable subpackage(s). --<!-- READ THIS FOR MANUAL BACKPORT FROM A MAINTAINER: Apply \"skip-basebranch-check\" label before you open the PR! --xref This provides some numpy array functions for objects. Most notably, one can now do the following without an errror(!): This still needs: [ ] Tests [ ] What's new [ ] API docs??? but opening now for feedback and a full CI run. <!-- If the pull request closes any open issues you can add this. If you replace <Issue Numberwith a number, GitHub will automatically link it. If this pull request is unrelated to any issues, please remove the following line. --<!-- This section is to be filled by package maintainer(s) who will review this pull request. --This checklist is meant to remind the package maintainer(s) who will review this pull request of some common things to look for. This list is not exhaustive. [x] Do the proposed changes actually accomplish desired goals? ? ? ? . [ ] Did the CI pass? If no, are the failures related? If you need to run daily and weekly cron jobs as part of the PR, please apply the label. [ ] Is a change log needed? If yes, did the change log check pass? If no, add the label. If this is a manual backport, use the label unless special changelog handling is necessary. [ ] Is a milestone set? Milestone must be set but check might be missing; do not let the green checkmark fool you. [ ] At the time of adding the milestone, if the milestone set requires a backport to release branch(es), apply the appropriate label(s) before merge."}
{"_id":"q-en-astropy__astropy-13162-50716b742e6ce4ecab89488803f758e7789d1c51d28a45ae513780201e54c920","text":"Angle bug for (d, m, s) tuple input (deprecate dmstodegrees) does not handle the sign correctly for a tuple input if : <!-- Provide a general description of the bug. --fix for the issue (Angle bug for (d, m, s) tuple input (deprecate dmstodegrees)) fix for the issue Two solutions are proposed. code for solution 1 is presented in this pull request."}
{"_id":"q-en-astropy__astropy-13234-18d1177ad14d975eece085fb9ff075b5cb6af08f7fbdd66a7c150689097b8096","text":"Structured column serialization round-trip fails with field name of \"name\" <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --<!-- Provide a general description of the bug. --A structured column with a field name of cannot be round-tripped through ECSV. Along with this suggests a tweak to the serialization format is needed. Perhaps: cc: <!-- What did you expect to happen. --It should work! <!-- Ideally a code example could be provided so we can run it ourselves. --<!-- If you are pasting code, use triple backticks (python import io import numpy as np from import Table, Column dtype = ([('z', 'f8'), ('name', 'f8'), ('y', 'i4')]) t = Table() t['c'] = Column([(1, 2, 3), (4, 5, 6)], dtype=dtype) out = io.StringIO() t.write(out, format='') print(out.getvalue()) t2 = (out.getvalue(), format='') c.z c.name c.y 0 2.0 3 0 5.0 6 Traceback (most recent call last): File in <modulet2 = (out.getvalue(), format='') File in call out = (cls, args, kwargs) File in read data = reader(args, kwargs) File in ioread return read(filename, **kwargs) File in read dat = (table) File in read table = self.outputter(, ) File in call out = serialize.constructmixinsfromcolumns(out) File in constructmixinsfromcolumns constructmixinfromcolumns(newname, objattrs, out) File in constructmixinfromcolumns dataattrsmap[val['name']] = name TypeError: unhashable type: 'SerializedColumn' macOS-10.16-x8664-i386-64bit Python 3.8.12 (default, Oct 12 2021, 06:23:56) [Clang 10.0.0 ] Numpy 1.22.2 pyerfa 2.0.0.1 astropy 5.1.dev956+g1d10de9d45. Scipy 1.8.0 Matplotlib 3.5.1"}
{"_id":"q-en-astropy__astropy-13236-bfeb4dbdf139d67d2d439b5dee2b86b5e08b53cb6e453e1b4c492235ed0ccb5e","text":"Consider removing auto-transform of structured column into NdarrayMixin <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Provide a general description of the feature you would like. --<!-- If you want to, you can suggest a draft design or API. --<!-- This way we have a deeper discussion on the feature. --Currently if you add a structured to a Table, it gets turned into an (via the code below). While this mostly works, I am not sure this is necessary or desirable any more after . Basically the original rational for was that structured dtype didn't quite work, in particular for serialization. So we pushed that out to a mixin class which would signal to unified I\/O that it might not be supported. Proposal: Add a FutureWarning here telling the user to wrap in and that in the future (5.2) the structured array will be as a . Change the behavior in 5.2 by removing this clause. This is not critical for 5.1 but if we have the opportunity due to other (critical) bugfixes it might be nice to save 6 months in the change process. cc:"}
{"_id":"q-en-astropy__astropy-13306-80acfd0b25d81c75aebd088fb83870e6ba15ecf8afef147d3aa7c8595746e84d","text":"vstack'ing structured array tables fails with casting error <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --<!-- Provide a general description of the bug. --Using on tables containing columns backed by numpy structured arrays fails. <!-- Ideally a code example could be provided so we can run it ourselves. --<!-- If you are pasting code, use triple backticks (python a=Table([dict(field1='test',field2=(1.,0.5,1.5))]) b=Table([dict(field1='foo')]) ((a,b)) # works a=Table([dict(field1='test',field2=(1.,0.5,1.5))],dtype=[str,[('val','f4'),('min','f4'),('max','f4')]]) ((a,b)) # fails Traceback (most recent call last): Input In [45] in <cell line: ((a,b)) File in vstack out = vstack(tables, jointype, colnamemap, metadataconflicts) File in vstack col[idx0:idx1] = array[name] File in setitem [index] = value TypeError: Cannot cast array data from dtype([('val', '<f4'), ('min', '<f4'), ('max', '<f4')]) to dtype('V12') according to the rule 'unsafe' macOS-12.3.1-x86_64-i386-64bit Python 3.10.4 (main, Apr 26 2022, 19:42:59) [Clang 13.1.6 (clang-1316.0.21.2)] Numpy 1.22.3 pyerfa 2.0.0.1 astropy 5.2.dev92+gf0e2129aa Scipy 1.7.3 Matplotlib 3.5.2"}
{"_id":"q-en-astropy__astropy-13398-39786b3c6f84541ada36a8f298b1743531cd073a439a63ddf3863ff825ad5dff","text":"A direct approach to ITRS to Observed transformations that stays within the ITRS. <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Provide a general description of the feature you would like. --<!-- If you want to, you can suggest a draft design or API. --<!-- This way we have a deeper discussion on the feature. --We have experienced recurring issues raised by folks that want to observe satellites and such (airplanes?, mountains?, neighboring buildings?) regarding the apparent inaccuracy of the ITRS to AltAz transform. I tire of explaining the problem of geocentric versus topocentric aberration and proposing the entirely nonintuitive solution laid out in . So, for the latest such issue (), I came up with a more direct approach. This approach stays entirely within the ITRS and merely converts between ITRS, AltAz, and HADec coordinates. I have put together the makings of a pull request that follows this approach for transforms between these frames (i.e. ITRS<-AltAz, ITRS<-HADec). One feature of this approach is that it treats the ITRS position as time invariant. It makes no sense to be doing an ITRS-ITRS transform for differing between the input and output frame, so the of the output frame is simply adopted. Even if it ends up being in the case of an or output frame where that is the default. This is because the current ITRS-ITRS transform refers the ITRS coordinates to the SSB rather than the rotating ITRF. Since ITRS positions tend to be nearby, any transform from one time to another leaves the poor ITRS position lost in the wake of the Earth's orbit around the SSB, perhaps millions of kilometers from where it is intended to be. Would folks be receptive to this approach? If so, I will submit my pull request. <!-- Add any other context or screenshots about the feature request here. --<!-- This part is optional. --Here is the basic concept, which is tested and working. I have yet to add refraction, but I can do so if it is deemed important to do so:"}
{"_id":"q-en-astropy__astropy-13404-2fd876b5b3df1240638e7b27508d389513437ea4eea4d25eef95ef0deb618de9","text":"Masked doesn't work on some structured arrays The dispatch function doesn't work on arrays with strutcured dtypes. Masked arrays with structured dtypes can be printed no prob, like their non-masked counterparts. It errors because of the structured dtype. Here's an example: While the following works: macOS-10.16-x8664-i386-64bit Python 3.9.12 (main, Jun 1 2022, 06:36:29) [Clang 12.0.0 ] Numpy 1.22.3 pyerfa 2.0.0.1 astropy 5.2.dev131+ga2da0589eb. Scipy 1.8.1 Matplotlib 3.5.2"}
{"_id":"q-en-astropy__astropy-13417-44ba502f2a45f74548286f12c8244ddc0be96cecbc5b46a03988b9345ee63ff7","text":"FITS problem reading binary table with variable length columns I want to read a certain FITS file (), which is part of a Multi-Beam-FITS measurement set () as used by several radio observatories around the world. The file has a binary table extension with variable length columns. Usually this works fine, but this particular example has some columns with a \"1PJ(1)\" type (and one row only), which seems to lead to problems when reading with Here it appears already, that the content of the record (\"(1, 1)\"] is smaller than the 'itemsize' (109). In fact, accessing the first two columns works, but all var-length columns raise an error: I checked the file with , which results in zero warnings and errors. Thanks a lot for your help! Multi-dimensional arrays in variable-length array columns of a FITS binary table cause file corruption may create corrupted files when writing a to a file, if that table contains a variable-length array (VLA) column with arrays that have two (or more) dimensions. No warnings or errors are raised while writing, yet the resulting file may be unreadable to . Being able to write any n-dimensional arrays to a VLA column, writing that to a file and then successfully reading the column (round-trip). The resulting file is partially or even completely corrupted. a two-dimensional array and place it on a array with a VLA column with that array a from that column and write it to a file the file back Produces the following error: _ Playing around with it a bit more, I could produce some other weird behaviors. Outputs the following: while the expected result would be: So it seems that everything that doesn't fit in the first dimension is going out of bounds and writing over the next array. This explains why it can also heavily corrupt the file. _ Reading the FITS standard, I get the impression that multi-dimensional VLAs should be possible, so this seems like an unexpected behavior. At the very least, if multi-dimensional VLAs aren't meant to be supported, should be throwing errors. Right now it's simply failing silently. Windows-10-10.0.-SP0 Python 3.9.9 (tags\/v3.9.9:, Nov 15 2021, 18:08:50) [MSC v.1929 64 bit (AMD64)] Numpy 1.22.2 pyerfa 2.0.0.1 astropy 5.0.1 Scipy 1.7.1"}
{"_id":"q-en-astropy__astropy-13438-c83cb34bbed50646d183a0744b88bea67c61432cf6aceecbfa0823b48e48ec08","text":"[Security] Jquery 3.1.1 is vulnerable to untrusted code execution <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --<!-- Provide a general description of the bug. --Passing HTML from untrusted sources - even after sanitizing it - to one of jQuery's DOM manipulation methods (i.e. .html(), .append(), and others) may execute untrusted code (see and ) <!-- What did you expect to happen. --Update jquery to the version 3.5 or newer in <!-- What actually happened. --<!-- Was the output confusing or poorly described? --jquery version 3.1.1 is distributed with the latest astropy release <!-- ### Steps to Reproduce <!-- Ideally a code example could be provided so we can run it ourselves. --<!-- If you are pasting code, use triple backticks (python --<!--### System Details <!-- Even if you do not think this is necessary, it is useful information for the maintainers. Please run the following snippet and paste the output below: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"Numpy\", numpy.version) import erfa; print(\"pyerfa\", erfa.version) import astropy; print(\"astropy\", astropy.version) import scipy; print(\"Scipy\", scipy.version) import matplotlib; print(\"Matplotlib\", matplotlib.version) --"}
{"_id":"q-en-astropy__astropy-13453-a6d37b3e8aa6a5e92dfc42d0b82763867a286ea92b61456ceb9f8ae218b363da","text":"ASCII table output to HTML does not support supplied \"formats\" <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --<!-- Provide a general description of the bug. --When writing out an astropy table to HTML format, the option to the  method seems to be ignored. It does work when writing out to other formats, e.g., rst, CSV, MRT, etc. <!-- What did you expect to happen. --I expect the HTML table output to respect the formatting given by the argument. <!-- What actually happened. --<!-- Was the output confusing or poorly described? --The argument seems to be ignored and the output is not formatted as required. <!-- Ideally a code example could be provided so we can run it ourselves. --<!-- If you are pasting code, use triple backticks (python from import Table from io import StringIO t = Table([(1.-24, 3.-15), (2, 4)], names=('a', 'b')) tc = t.copy() # copy table with StringIO() as sp: (sp, format=\"html\", formats={\"a\": lambda x: f\"{x:.2e}\"}) print(sp.getvalue()) <html<head<meta charset=\"utf-8\"\/<meta content=\"text\/html;charset=UTF-8\" http-equiv=\"Content-type\"\/<\/head<body<table<thead<tr<tha<\/th<thb<\/th<\/tr<\/thead<tr<td1.-24<\/td<td2<\/td<\/tr<tr<td3.-15<\/td<td4<\/td<\/tr<\/table<\/body<\/htmlpython with StringIO() as sp: (sp, format=\"csv\", formats={\"a\": lambda x: f\"{x:.2e}\"}) print(sp.getvalue()) a,b 24e-24,2 23e-15,4 python with StringIO() as sp: (sp, format=\"\", formats={\"a\": lambda x: f\"{x:.2e}\"}) print(sp.getvalue()) ======== = a b ======== = 24e-24 2 23e-15 4 ======== = ) around your code snippet. --<!-- If necessary, sanitize your screen output to be pasted so you do not reveal secrets like tokens and passwords. --Outputting a HTML table gives the numbers to the full number of decimal places. Instead, outputting to a CSV table: or, e.g., rsrt: gives the formatting as expected. <!-- Even if you do not think this is necessary, it is useful information for the maintainers. Please run the following snippet and paste the output below: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"Numpy\", numpy.version) import erfa; print(\"pyerfa\", erfa.version) import astropy; print(\"astropy\", astropy.version) import scipy; print(\"Scipy\", scipy.version) import matplotlib; print(\"Matplotlib\", matplotlib.version) --Linux-5.4.0-121-generic-x86_64-with-glibc2.31 Python 3.9.12 (main, Jun 1 2022, 11:38:51) [GCC 7.5.0] Numpy 1.22.4 pyerfa 2.0.0.1 astropy 5.1 Scipy 1.8.1 Matplotlib 3.5.2"}
{"_id":"q-en-astropy__astropy-13462-c0075e6f7e273a8ded8b2e2a91d8476d639a0580a010d42f706b185049a3d293","text":"TST: failed in pyinstaller (computed error is different depending on the order of the arguments) First failing log (2022-03-13): Last successful log (2022-03-12): Looks like this test was in . Any idea how to fix, or ?"}
{"_id":"q-en-astropy__astropy-13465-85df052bdb119fa40f477764b515708b467022dd8ba859248b29f369b844671d","text":"rtol for FITSDiff not working as expected. I have question about the rtol parameter for FITSDiff, when I create a report it appears that the numbers cited as being different are within the given relative tolerance. I couldn't figure out why so I thought this may be a bug, apologies if I'm missing something super obvious here! Here's how to recreate the issue using FITSdiff, I included a zip file containing the two fits file and an example logfile."}
{"_id":"q-en-astropy__astropy-13469-1eee612d7d2b2c640c41932bbf45d880b0f7f82c5587a31e14f96f0d9df4736e","text":"Can't convert a list of Astropy tables to a NumPy array of tables I recently stumbled upon where someone likes to convert a list of Tables to a NumPy array. By default, NumPy will convert the Table along the way, resulting in the wrong data structure. Using a specific , however, fails with This error leads directly to the source of , which explicitly checks for any to be not , which will raise the error. The reasoning behind that is clear, as given in the comments below. But I wonder if an exception is reasonable for here, and let that pass through. For a single Table, this may be odd, but not necessarily incorrect. And for a list of Tables, to be converted to an array, this may be helpful."}
{"_id":"q-en-astropy__astropy-13477-ab933bcc680ee0c33077d5b10642e5440f434c228a8a972aebd878a963c5d38e","text":"Comparing Frame with data and SkyCoord with same data raises exception <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --instances and instances with data are somewhat used interchangebly and I am still not sure after all this time spending with astropy what is preferable when... So it's a bit surprising to me, that comparing a frame with data to a instance with exactly the same data raises an exception: python In [1]: from astropy.coordinates import SkyCoord, ICRS In [2]: frame = ICRS(\"0d\", \"0d\") In [3]: coord = SkyCoord(frame) In [4]: frame == coord TypeError Traceback (most recent call last) Input In [4], in <cell line: 1() ----1 frame == coord File , in (self, value) 1651 def eq(self, value): 1652 \"\"\"Equality operator for frame. 1653 1654 This implements strict equality and requires that the frames are 1655 equivalent and that the representation data are exactly equal. 1656 \"\"\" -1657 isequiv = self.isequivalentframe(value) 1659 if is None and is None: 1660 # For Frame with no data, == compare is same as isequivalentframe() 1661 return isequiv File , in BaseCoordinateFrame.isequivalentframe(self, other) 1358 return True 1359 elif not isinstance(other, BaseCoordinateFrame): -1360 raise TypeError(\"Tried to do isequivalentframe on something that \" 1361 \"isn't a frame\") 1362 else: 1363 return False TypeError: Tried to do isequivalentframe on something that isn't a frame <!-- What did you expect to happen. --Compare to true \/ false depending on data. Exception"}
{"_id":"q-en-astropy__astropy-13572-54f8d2fa6aafb9c2ccced444c87cf3d6926e8200a9e2bf3ece572e34beaa632b","text":"Problem in function nutationmatrix in Recently, when I try to call function nutationmatrix in astropy.coordinates.earthorientation, error occurs with following info: Angle instances require units equivalent to 'rad', so cannot set it to '0'. Then, I checked the code of def nutationmatrix as follows: In its return sentence, the third argument of 'rotationmatrix' should be , rather than False. Any response?"}
{"_id":"q-en-astropy__astropy-13579-b979367ff0c93f23e0423f08edc6740e89e98726e1959a1555b027ee57623618","text":"Inconsistent behavior of in <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --<!-- Provide a general description of the bug. --I have a 3D WCS with dimensions corresponding to space, space, and wavelength and what some might call a non-trivial PCij matrix that couples the spectral and spatial dimensions. I find that when I perform a worldtopixel on the full (unsliced) WCS, I get back the expected result. However, when I perform that same worldtopixel operation on a single wavelength slice (i.e. a 2D slice with dimensions corresponding to space, space), my worldtopixel returns an erroneous result for one of the dimensions. This issue was originally posted as sunpy\/ndcube, but I've moved it here as it seems to be an issue with rather than anything specific to . <!-- Ideally a code example could be provided so we can run it ourselves. --<!-- If you are pasting code, use triple backticks (python import numpy as np import from astropy.coordinates import SkyCoord import as u nx = 100 ny = 25 nz = 2 wcsheader = { 'WCSAXES': 3, 'CRPIX1': (nx + 1)\/2, 'CRPIX2': (ny + 1)\/2, 'CRPIX3': 1.0, 'PC11': 0.0, 'PC12': -1.0, 'PC13': 0.0, 'PC21': 1.0, 'PC22': 0.0, 'PC23': -1.0, 'CDELT1': 5, 'CDELT2': 5, 'CDELT3': 0.055, 'CUNIT1': 'arcsec', 'CUNIT2': 'arcsec', 'CUNIT3': 'Angstrom', 'CTYPE1': 'HPLN-TAN', 'CTYPE2': 'HPLT-TAN', 'CTYPE3': 'WAVE', 'CRVAL1': 0.0, 'CRVAL2': 0.0, 'CRVAL3': 1.05, } fitswcs = (header=wcsheader) python pt = SkyCoord(, , (fitswcs)) fitswcs.worldtopixel(pt, 1.05u.angstrom) (array(49.5), array(12.), array(2.-15)) python llslicedwcs = (fitswcs, 0) hlslicedwcs = (llslicedwcs) hlslicedwcs.worldtopixel(pt) (array(1.+11), array(12.)) python px,py,pz = fitswcs.worldtopixel(pt, 1.05u.Angstrom) fitswcs.pixeltoworld(px, py, pz) [<SkyCoord (Helioprojective: obstime=None, rsun=.0 km, observer=None): (Tx, Ty) in arcsec (1.-27, 0.), <SpectralCoord 1.05e-10 m] hlslicedwcs.pixeltoworld(px, py) <SkyCoord (Helioprojective: obstime=None, rsun=.0 km, observer=None): (Tx, Ty) in arcsec (1.-27, 0.)macOS-10.16-x8664-i386-64bit Python 3.9.7 (default, Sep 16 2021, 08:50:36) [Clang 10.0.0 ] Numpy 1.21.5 pyerfa 2.0.0.1 astropy 5.1 Scipy 1.8.0 Matplotlib 3.5.1"}
{"_id":"q-en-astropy__astropy-13638-3a147cd51586e19fc5da061ed1146251faf4bf469e7a7fb4884f29769e73c25e","text":"throws exception with <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --<!-- Provide a general description of the bug. --The decorator throws a when used on a function that returns a with and a return type annotation. <!-- What did you expect to happen. --For the function to return a with with the appropriate units or to throw an exception if the output units are of the wrong type. <!-- What actually happened. --<!-- Was the output confusing or poorly described? --Using the decorator results in a gives <!-- Even if you do not think this is necessary, it is useful information for the maintainers. Please run the following snippet and paste the output below: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"Numpy\", numpy.version) import erfa; print(\"pyerfa\", erfa.version) import astropy; print(\"astropy\", astropy.version) import scipy; print(\"Scipy\", scipy.version) import matplotlib; print(\"Matplotlib\", matplotlib.version) --"}
{"_id":"q-en-astropy__astropy-13668-df0087e692e6d5f275e4999b475c3ad3a59aea79850b3942468fc86ead1f3533","text":"wcslint crashes on valid WCS calls an underlying function here: Looks like all it does is tries to create a object with the header and report warnings, so the bug is either inside or it is a matter of updating on how validator calls in more complicated cases: File: File: File: (Reported by"}
{"_id":"q-en-astropy__astropy-13731-b81dfe210251e44278709b3c2939916b19d62e589538c0bcaea686b2ec58d977","text":"parses fractional days in year-month-day format incorrectly results in : the fractional days are interpreted as fractional seconds ( is what I hoped for). The format is perhaps not the best format, but it is used, and since Astropy does not raise an exception, but silently returns an incorrect result, this may lead to errors. The issue can be traced to , which will interpret anything right of the last dot as a fractional second. Since matching to regexes or formats is done afterwards, there is no (easy) way to catch this through a subformat before the fractional second get stripped. I'd be happy to try and put in a PR for this (if it's indeed a bug), but I'll need to know whether to raise an exception, or implement a proper parser for this format (provided it doesn't clash with other interpretations). Some suggestions on the best way to attack this issue (or at what point in the code) are welcome as well."}
{"_id":"q-en-astropy__astropy-13734-03898002c5f3dee0a5d59476624e07487ad01123e9bdf633e15e7560c9295e03","text":"Add option to input\/output column units for fixed width tables Extend the reader to include a keyword arg that will specify that there is a row of unit specifiers after the column name specifiers (or at the top of the header if there are no column names). This will apply for both reading and writing fixed width tables. This allows for outputting a table to a file in a format like with , and then reading back that table with no information loss."}
{"_id":"q-en-astropy__astropy-13745-d9cc47482916ad7eed91eca6ac4d4378bdc2022717b0a8e7c6c8c1ea46646bd1","text":"float32 representation of pi\/2 is rejected by <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --The closest float32 value to pi\/2 is by accident slightly larger than pi\/2: Astropy checks using float64 precision, rejecting \"valid\" alt values (e.g. float32 values read from files): Be lenient? E.g. only make the comparison up to float 32 precision? See error above See snippet above. <!-- Even if you do not think this is necessary, it is useful information for the maintainers. Please run the following snippet and paste the output below: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"Numpy\", numpy.version) import erfa; print(\"pyerfa\", erfa.version) import astropy; print(\"astropy\", astropy.version) import scipy; print(\"Scipy\", scipy.version) import matplotlib; print(\"Matplotlib\", matplotlib.version) --"}
{"_id":"q-en-astropy__astropy-13803-d9cc47482916ad7eed91eca6ac4d4378bdc2022717b0a8e7c6c8c1ea46646bd1","text":"float32 representation of pi\/2 is rejected by <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --The closest float32 value to pi\/2 is by accident slightly larger than pi\/2: Astropy checks using float64 precision, rejecting \"valid\" alt values (e.g. float32 values read from files): Be lenient? E.g. only make the comparison up to float 32 precision? See error above See snippet above. <!-- Even if you do not think this is necessary, it is useful information for the maintainers. Please run the following snippet and paste the output below: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"Numpy\", numpy.version) import erfa; print(\"pyerfa\", erfa.version) import astropy; print(\"astropy\", astropy.version) import scipy; print(\"Scipy\", scipy.version) import matplotlib; print(\"Matplotlib\", matplotlib.version) --"}
{"_id":"q-en-astropy__astropy-13838-88d72475c114151cc8d805703634a3ab3ad1a888c5b0572fd24ac75e5e57c3d9","text":"Printing tables doesn't work correctly with 0-length array cells <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --I have data in form of a list of dictionaries. Each dictionary contains some items with an integer value and some of these items set the length for 1 or more array values. I am creating a Table using the attribute and feeding to it the list of dictionaries. As long as I create a table until the first event with data in the array fields the table gets printed correctly. If I fill the table only with events with null array data (but the rest of the fields have something to show) I get an IndexError. <!-- What did you expect to happen. --The table should print fine also when there are only \"bad\" events <!-- What actually happened. --<!-- Was the output confusing or poorly described? --I get the following error Traceback <!-- Ideally a code example could be provided so we can run it ourselves. --<!-- If you are pasting code, use triple backticks ( events = [{\"A\":0,\"B\":0, \"C\"([], dtype=np.uint64)}, {\"A\":1,\"B\":0, \"C\"([], dtype=np.uint64)}, {\"A\":2,\"B\":2, \"C\"([0,1], dtype=np.uint64)}] Showing just the first event prints the column names as a column, <img width=\"196\" alt=\"image\" src=\"https:\/\/user-\"Printing the first 2 throws the Traceback above Plotting all 3 events works <img width=\"177\" alt=\"image\" src=\"https:\/\/user-\"<!-- Even if you do not think this is necessary, it is useful information for the maintainers. Please run the following snippet and paste the output below: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"Numpy\", numpy.version) import erfa; print(\"pyerfa\", erfa.version) import astropy; print(\"astropy\", astropy.version) import scipy; print(\"Scipy\", scipy.version) import matplotlib; print(\"Matplotlib\", matplotlib.version) --macOS-11.7-x86_64-i386-64bit Python 3.9.13 (main, May 27 2022, 17:00:52) [Clang 13.0.1 ] Numpy 1.23.3 pyerfa 2.0.0.1 astropy 5.1 Scipy 1.9.1 Matplotlib 3.6.0"}
{"_id":"q-en-astropy__astropy-13842-9f95aeb8b12a57b3e121ae59f449da015db938603c23c1b34237e3a27546aad8","text":"Creating a mixin column in a new table from columns of another table renames columns in original table. Consider the following code, where a subset of columns from another table should be included in a new table with new names, prerably without copying the actual payload data: If any of the two classes or both are a , not a , the code works as expected. Data in the columns is not copied, but column names in original table stay the same. Column names do change in both tables: See above. <!-- Even if you do not think this is necessary, it is useful information for the maintainers. Please run the following snippet and paste the output below: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"Numpy\", numpy.version) import erfa; print(\"pyerfa\", erfa.version) import astropy; print(\"astropy\", astropy.version) import scipy; print(\"Scipy\", scipy.version) import matplotlib; print(\"Matplotlib\", matplotlib.version) --(also tested with current branch)"}
{"_id":"q-en-astropy__astropy-13933-8e58e34e6d9c72b2547012592afc620a0803446b120e9df7f60e8c22e8c2f25f","text":"Unpickled Angle.tostring fails <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --<!-- Provide a general description of the bug. --Pickling and unpickling an Angle object causes the tostring function to fail claiming hourangle and degree units cannot be represented in sexagesimal notation. <!-- Ideally a code example could be provided so we can run it ourselves. --<!-- If you are pasting code, use triple backticks (python import astropy.coordinates import pickle ang = astropy.coordinates.Angle(0.25 * ) pang = ((ang)) ang.tostring() pang.tostring() <!-- Even if you do not think this is necessary, it is useful information for the maintainers. Please run the following snippet and paste the output below: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"Numpy\", numpy.version) import erfa; print(\"pyerfa\", erfa.version) import astropy; print(\"astropy\", astropy.version) import scipy; print(\"Scipy\", scipy.version) import matplotlib; print(\"Matplotlib\", matplotlib.version) --macOS-10.15.7-x86_64-i386-64bit Python 3.10.6 (main, Aug 22 2022, 20:43:44) [Clang 13.0.1 ] Numpy 1.23.4 pyerfa 2.0.0.1 astropy 5.1 Scipy 1.9.3 Matplotlib 3.6.1"}
{"_id":"q-en-astropy__astropy-13977-32c71ca5ed71b5e70605c1fa4ae05e557b1f3ffa94242b3d1531d2976f02a64f","text":"Should return instead of raising if the inputs are incompatible? I'm trying to implement a duck type of . If you are interested, the project is available . I'm running into trouble trying to coerce my duck type to use the reflected versions of the arithmetic operators if the left operand is not an instance of the duck type and they have equivalent but different units. Consider the following minimal working example of my duck type. If I do an operation like It works as expected. Or I can do and it still works properly. But if the left operand has different units I get the following error: I would argue that should really return in this instance, since it would allow for to be called instead of the error being raised. I feel that the current behavior is also inconsistent with the which specify that should be returned if the requested operation is not implemented. What does everyone think? I am more than happy to open a PR to try and solve this issue if we think it's worth pursuing."}
{"_id":"q-en-astropy__astropy-14042-c36052239530c170362fdd292fed07c2849506f1cc1c64ac7466eba878b32ea1","text":"Degrees Celsius should be supported by FITS units <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --The FITS standards says that units should follow IAU recommendations. These note that: However, astropy does not support for fits: See"}
{"_id":"q-en-astropy__astropy-14096-9d646028881e5bdaf51e534757406b24655eba6cc37e86d0f89bc4af5a024d68","text":"Subclassed SkyCoord gives misleading attribute access message I'm trying to subclass , and add some custom properties. This all seems to be working fine, but when I have a custom property ( below) that tries to access a non-existent attribute () below, the error message is misleading because it says doesn't exist, where it should say doesn't exist. raises"}
{"_id":"q-en-astropy__astropy-14163-1da610d896c9e49e18e60f66ba8157c51dfa9e3cea5864923138f5c1ea9f9317","text":"Should calling on instances with incompatible units return ? <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --Using on instances with incompatible units raises a . I would've expected this function just to return in this case. Do we think it's really necessary to halt if the units are incompatible? <!-- What actually happened. --An exception was raised. <!-- Ideally a code example could be provided so we can run it ourselves. --<!-- If you are pasting code, use triple backticks (python np.arrayequal([1, 2, 3] u.mm, [1, 2, 3] u.s) Traceback (most recent call last): File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\\", line 980, in tovalue scale = (unit) File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\\", line 1129, in to raise UnitConversionError(f\"'{self!r}' is not a scaled version of '{other!r}'\") 'Unit(\"s\")' is not a scaled version of 'Unit(\"mm\")' During handling of the above exception, another exception occurred: Traceback (most recent call last): File \"C:\\Users\\royts\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\IPython\\core\\\", line 3378, in runcode exec(codeobj, self.userglobalns, ) File \"<ipython-input-13-\", line 1, in <modulenp.arrayequal([1, 2, 3] u.mm, [1, 2, 3] u.s) File \"<arrayfunction internals\", line 180, in arrayequal File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\\", line 1844, in arrayfunction args, kwargs, unit, out = functionhelper(args, *kwargs) File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\quantityhelper\\\", line 566, in arrayequal args, unit = quantities2arrays(a1, a2) File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\quantityhelper\\\", line 351, in quantities2arrays arrays = tuple((q.toownunit(arg)) for arg in args) File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\quantityhelper\\\", line 351, in <genexprarrays = tuple((q.toownunit(arg)) for arg in args) File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\\", line 1652, in toownunit value = value.tovalue(unit) File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\\", line 983, in tovalue value = self.tovalue(unit, equivalencies) File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\\", line 889, in tovalue return ( File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\\", line 1165, in to return self.getconverter(Unit(other), equivalencies)(value) File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\\", line 1094, in getconverter raise exc File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\\", line 1077, in getconverter return self.applyequivalencies( File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\\", line 1054, in applyequivalencies raise UnitConversionError(f\"{unitstr} and {other_str} are not convertible\") 's' (time) and 'mm' (length) are not convertible Windows-10-10.0.-SP0 Python 3.10.8 (tags\/v3.10.8:, Oct 11 2022, 16:50:30) [MSC v.1933 64 bit (AMD64)] Numpy 1.23.5 pyerfa 2.0.0.1 astropy 5.3.dev89+g4989e530b Scipy 1.9.1 Matplotlib 3.6.0"}
{"_id":"q-en-astropy__astropy-14182-2adf9a22bbb7587e377a970f6a03a91b219982c07cba8b913e390baf7bd0c3f1","text":"Please support header rows in RestructuredText output It would be great if the following would work: RestructuredText output is a great way to fill autogenerated documentation with content, so having this flexible makes the life easier"}
{"_id":"q-en-astropy__astropy-14213-8e3dd4e66f709aa1286c487f52b28f551bebea4be679822a20fdfac4e8d24be3","text":"The argument to -like functions does not accept instances of <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --When using , , etc. family of functions, with instances of , the argument only accepts instances of . I would have expected that the argument needs to be an instance of with compatible units. An is raised if is an instance of . <!-- Ideally a code example could be provided so we can run it ourselves. --<!-- If you are pasting code, use triple backticks (python import numpy as np import as u a = (21) u.m np.histogram(a, range=[.25, .75] u.m) Traceback (most recent call last): File \"C:\\Users\\royts\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\IPython\\core\\\", line 3378, in runcode exec(codeobj, self.userglobalns, ) File \"<ipython-input-8-\", line 1, in <modulenp.histogram(a, range=[.2, .7] u.m) File \"<arrayfunction internals\", line 200, in histogram File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\\", line 1844, in arrayfunction result = super().arrayfunction(function, types, args, kwargs) File \"C:\\Users\\royts\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\numpy\\lib\\\", line 824, in histogram keep = (tmpa = firstedge) File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\\", line 699, in arrayufunc raise e File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\\", line 644, in arrayufunc converters, unit = convertersandunit(function, method, inputs) File \"C:\\Users\\royts\\Kankelborg-Group\\astropy\\astropy\\units\\quantityhelper\\\", line 200, in convertersandunit raise UnitConversionError( Can only apply 'lessequal' function to dimensionless quantities when other argument is not a quantity (unless the latter is all zero\/infinity\/nan). Windows-10-10.0.-SP0 Python 3.10.8 (tags\/v3.10.8:, Oct 11 2022, 16:50:30) [MSC v.1933 64 bit (AMD64)] Numpy 1.24.0 pyerfa 2.0.0.1 astropy 5.3.dev128+ge512a6799 Scipy 1.9.1 Matplotlib 3.6.0"}
{"_id":"q-en-astropy__astropy-14253-2a9a47dbe306d13104005db715cdb8f7f6ff4742071330e3468fe14e069084e6","text":"When should be linked to a new object? Mostly for - I noticed that in the way we have set up , is passed on not just for views (where it should be), but also for copies (implicitly in arithmetic operations, etc.). Which ones are reasonable? Just thinking about whether, e.g., should be propagated, I'd think: Surely for views & reshapes: , , etc. insertions: Probably for selection of scalars: or in (for columns this returns a scalar without ) copies: and equivalents equivalent unit changes: , , , etc. Probably not for operations real unit changes (including in-place??; ) What do you think? p.s. Currently, all of the above happen, in part because I use in , something which I don't think we had really considered when we made the change in . But that also means that in principle it may not too hard to fairly finely define the behaviour."}
{"_id":"q-en-astropy__astropy-14295-178805533918852811a1524ebb7a322a88979dc3252b7156b3ddfd8ca9e19605","text":"Presence of SIP keywords leads to ignored PV keywords. <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --I am working on updating the fits header for one of our telescopes. We wanted to represent the distortions in SIP convention and the projection to be 'CAR'. While working on this, I noticed if SIP coefficients are present in the header and\/or '-SIP' is to CTYPEia keywords, astropy treats the PV keywords (PV10, PV11 and PV12) as \"redundant SCAMP distortions\". Earlier I could not figure out why the projection weren't going as I expected, and I suspected a bug in astropy wcs. After talking to Mark Calabretta about the difficulties I'm facing, that suspicion only grew stronger. The header was working as expected with WCSLIB but was giving unexpected behavior in astropy wcs. The following would be one example header - When you parse the wcs information from this header, the image should be centered at ra=120 and dec=29 with lines of constant ra and dec looking like the following image generated using wcslib - ! If I parse the wcs information using astropy wcs, it throws the following warning - And the resulting grid is different. Code - Grid - ! The astropy wcs grid\/solution does not change whethere we keep or remove the PV keywords. Furthermore, the astropy grid can be recreated in wcslib, by removing the PV keywords. ! <!-- Ideally a code example could be provided so we can run it ourselves. --<!-- If you are pasting code, use triple backticks () around your code snippet. --<!-- If necessary, sanitize your screen output to be pasted so you do not reveal secrets like tokens and passwords. --the header the header using the graticule the PV keywords and run again will find the same graticule indicating that PV keywords are completely ignored. Additionally, the graticules can be compared with the wcsgrid utility of wcslib. <!-- Even if you do not think this is necessary, it is useful information for the maintainers. Please run the following snippet and paste the output below: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"Numpy\", numpy.version) import erfa; print(\"pyerfa\", erfa.version) import astropy; print(\"astropy\", astropy.version) import scipy; print(\"Scipy\", scipy.version) import matplotlib; print(\"Matplotlib\", matplotlib.version) --Linux-6.0.11-200.fc36.x8664-x8664-with-glibc2.35 Python 3.9.12 (main, Apr 5 2022, 06:56:58) [GCC 7.5.0] Numpy 1.21.5 pyerfa 2.0.0 astropy 5.1 Scipy 1.7.3 Matplotlib 3.5.1 Remove heuristic code to handle PTF files which is causing a bug <!-- This comments are hidden when you submit the pull request, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- If you are new or need to be re-acquainted with Astropy contributing workflow, please see . There is even a practical example at . --<!-- Astropy coding style guidelines can be found here: Our testing infrastructure enforces to follow a subset of the PEP8 to be followed. You can check locally whether your changes have followed these by running the following command: tox -e codestyle --<!-- Please just have a quick search on GitHub to see if a similar pull request has already been posted. We have old closed pull requests that might provide useful code or ideas that directly tie in with your pull request. --<!-- We have several automatic features that run when a pull request is open. They can appear daunting but do not worry because maintainers will help you navigate them, if necessary. --<!-- Provide a general description of what your pull request does. Complete the following sentence and add relevant details as you see fit. --<!-- In addition please ensure that the pull request title is descriptive and allows maintainers to infer the applicable subpackage(s). --<!-- READ THIS FOR MANUAL BACKPORT FROM A MAINTAINER: Apply \"skip-basebranch-check\" label before you open the PR! --Currently the function remove any PV keywords when SIP distortions are present and no radial terms are present which should not be the case. This function was put in place for solving but it causes the bug . We can either keep adding heuristic code to fix the edge cases as they come up with or remove and let the user deal with non-standard files. I'm opening a pull request for the latter following the discusison in . <!-- If the pull request closes any open issues you can add this. If you replace <Issue Numberwith a number, GitHub will automatically link it. If this pull request is unrelated to any issues, please remove the following line. -- <!-- This section is to be filled by package maintainer(s) who will review this pull request. --This checklist is meant to remind the package maintainer(s) who will review this pull request of some common things to look for. This list is not exhaustive. [ ] Do the proposed changes actually accomplish desired goals? ? ? ? . . [ ] Is a change log needed? If yes, did the change log check pass? If no, add the label. If this is a manual backport, use the label unless special changelog handling is necessary. [ ] Is this a big PR that makes a \"What's new?\" entry worthwhile and if so, is (1) a \"what's new\" entry included in this PR and (2) the \"whatsnew-needed\" label applied? [ ] Is a milestone set? Milestone must be set but check might be missing; do not let the green checkmark fool you. [ ] At the time of adding the milestone, if the milestone set requires a backport to release branch(es), apply the appropriate label(s) before merge."}
{"_id":"q-en-astropy__astropy-14309-86d58ba2e60cf8478b73b1fc760dea165779bd2cfbbf786809b222cba96b497c","text":"IndexError: tuple index out of range in identifyformat (io.registry) <!-- This comments are hidden when you submit the issue, so you do not need to remove them! --<!-- Please be sure to check out our contributing guidelines, . Please be sure to check out our code of conduct, . --<!-- Please have a search on our GitHub repository to see if a similar issue has already been posted. If a similar issue is closed, have a quick look to see if you are satisfied by the resolution. If not please go ahead and open an issue! --<!-- Please check that the development version still produces the same bug. You can install development version with pip install command. --Cron tests in HENDRICS using identifyformat have started failing in (e.g. ) with this error: As per a Slack conversation with and this should be related to Citing When is a string without a FITS extension, the function was returning None, now it executes <!-- Ideally a code example could be provided so we can run it ourselves. --<!-- If you are pasting code, use triple backticks ( In [1]: from import identifyformat In [3]: from import Table In [4]: identifyformat(\"write\", Table, \"\", None, [], {}) IndexError Traceback (most recent call last) Cell In [4], line 1 ----1 identifyformat(\"write\", Table, \"\", None, [], {}) File ~\/opt\/anaconda3\/envs\/py310\/lib\/python3.10\/site-, in makeiofunc.<locals.wrapper(registry, args, kwargs) 50 registry = defaultregistry 51 # get and call bound method from registry instance ---52 return getattr(registry, methodname)(args, kwargs) File ~\/opt\/anaconda3\/envs\/py310\/lib\/python3.10\/site-, in UnifiedIORegistryBase.identifyformat(self, origin, dataclassrequired, path, fileobj, args, kwargs) 311 for dataformat, dataclass in self.identifiers: 312 if self.isbestmatch(dataclassrequired, dataclass, self.identifiers): --313 if self.identifiers(dataformat, dataclass): 316 (dataformat) 318 return validformats File ~\/opt\/anaconda3\/envs\/py310\/lib\/python3.10\/site-, in is_fits(origin, filepath, fileobj, args, kwargs) 68 if ().endswith( 69 (\".fits\", \"\", \".fit\", \"\", \".fts\", \"\") 70 ): 71 return True ---72 return isinstance(args[0], (HDUList, TableHDU, BinTableHDU, GroupsHDU)) IndexError: tuple index out of range <!-- Even if you do not think this is necessary, it is useful information for the maintainers. Please run the following snippet and paste the output below: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"Numpy\", numpy.version) import erfa; print(\"pyerfa\", erfa.version) import astropy; print(\"astropy\", astropy.version) import scipy; print(\"Scipy\", scipy.version) import matplotlib; print(\"Matplotlib\", matplotlib.version) --"}
{"_id":"q-en-astropy__astropy-14365-e98a6aec632d1281bdf994c858bf32ce8efefafb4c4d5a5abf1a0ae898c7cc34","text":"Table format assumes QDP commands are upper case assumes that commands in a QDP file are upper case, for example, for errors they must be \"READ SERR 1 2\" whereas QDP itself is not case sensitive and case use \"read serr 1 2\". As many QDP files are created by hand, the expectation that all commands be all-caps should be removed. The following qdp file should read into a with errors, rather than crashing. Create a QDP file: Running \"qdp \" works just fine. Python 3.10.9 (main, Dec 7 2022, 02:03:23) [Clang 13.0.0 (clang-1300.0.29.30)] astropy 5.1 Numpy 1.24.1 pyerfa 2.0.0.1 Scipy 1.10.0 Matplotlib 3.6.3"}
{"_id":"q-en-astropy__astropy-14369-e5c55f65c55220adcbc6009f6f2ba0d68658a00d8fbf0b4fb7c53c6a9a863da9","text":"Incorrect units read from MRT (CDS format) files with When reading MRT files (formatted according to the CDS standard which is also the format recommended by AAS\/ApJ) with , incorrectly parses composite units. According to CDS standard the units should be SI without spaces (http:\/\/vizier.u-). Thus a unit of (surface brightness for a continuum measurement) should be written as . When I use these types of composite units with the reader the units do not come out correct. Specifically the order of the division seems to be jumbled. The units in the resulting Table should be the same as in the input MRT file. Get astropy package from pip Using the following MRT as input: And then reading the table I get: For the SBCONT column the second is in the wrong place, and for SBLINE kpc2 is in the wrong place."}
{"_id":"q-en-astropy__astropy-14371-c3e05dadcb179cb4b555d2c974d00f1d4ca1e113b6422f23bb59c55379eba464","text":"Add ` Or at least use the maximum precision of the matrix dtype instead of the arbitrarily chosen 1e-15."}
{"_id":"q-en-astropy__astropy-14379-795aae96f996e9bdb4ce0c5599f598e1bdc35da8e4d805676692c68e71070da2","text":"Space between value and unit Currently, ` doesn't include a space between the value and unit: I think there are cases where it would make sense to allow a space to be included, so this is a feature request to add a boolean keyword argument to optionally add a space. Note that Quantity does include a space by default so maybe actually we should just change the default and not add an option?"}
{"_id":"q-en-astropy__astropy-14413-8c404342b42936e08da441fc16411bae9f20f9dfc09dbcb5b17ce993d5a90286","text":"Unicode and console unit representations sometimes include an extraneous space As noted in , for units typset in or format, a space is included in front, unlike for regular units, yet it is sometimes omitted if a unit scale factor is present. should never start with a space, independent of . From and Any astropy really."}
{"_id":"q-en-astropy__astropy-14439-a2175d3ed1307ceebebd3738ce654fbfa40d096bdd47638dc294ada11d185100","text":"Regarding FITS standard definition for 'Jy\/beam' Astropy unit definition for Jy\/beam currently exports it as 'beam-1 Jy' 'beam-1 Jy' This is contrary to how most radio astronomy packages define the unit. 'Jy\/beam' seems to be the accepted convention. The space after beam-1 makes parsing needlessly cumbersome as well. Is this something that can be fixed? See related issues opened in SpectralCube and SoFiA2."}
{"_id":"q-en-astropy__astropy-14484-92b519611fce8d4412e873cb3ceab1975dbc2c9ed33348fc7d39dbf2364d6cd8","text":"New Quantity warning starting with yesterday's numpy-dev Starting today, CI tests with and started failing due a new warning. I've extracted a MWE showing the warning: The warning is: The warning is not emitted with and stable (1.24.2). CC:"}
{"_id":"q-en-astropy__astropy-14508-bae0d90490c5997108003f5350d118f353a4c0de4b7dcdd9615c5cb59c0b6bbe","text":"may use a string representation of floats that is larger than necessary In some scenarios, may use a string representation of floats that is larger than necessary, which can force comments to be truncated. Due to this, there are some keyword\/value\/comment combinations that are impossible to create via even though they are entirely possible in FITS. Being able to create any valid FITS Card via . contains the following card in the header: We can read the header of this file and get this card without any issue: However, we have problems creating a object with exactly the same contents of : Essentially the value \"0.\" is being unnecessarily expanded to \"0.\", which forces the comment to be truncated. I've investigated the source code and the root issue is the function which creates a of when is used as the input: It seems to me that before doing , we should attempt to use the string representation provided by Python (in other words ), and we should only attempt to format it ourselves if the resulting string does not fit in 20 characters. However, since this is fairly deep in the code, it's possible this would have side-effects that I am not aware of. Windows-10-10.0.-SP0 Python 3.10.10 (tags\/v3.10.10:, Feb 7 2023, 17:20:36) [MSC v.1929 64 bit (AMD64)] astropy 5.2.1 Numpy 1.24.2 pyerfa 2.0.0.1 Scipy 1.10.0 Matplotlib 3.6.3"}
{"_id":"q-en-astropy__astropy-14528-68bd76809f093e4b5f9b1f1e42ef4196b1c2f4b5ccfdd9046bd5318e9e8f5325","text":"creates a corrupt FITS files if a contains zero-dimensional data accepts a ndarray with shape (zero-dimensional) as a data array. This later causes issues when writing to a file because assumes that the data has at least 1 dimension, resulting in a corrupt FITS file. should never silently create a corrupt FITS file. Minimal reproducible example: Although this causes no errors\/warnings, the resulting file is not valid FITS and will fail to properly open with . Windows-10-10.0.-SP0 Python 3.10.10 (tags\/v3.10.10:, Feb 7 2023, 17:20:36) [MSC v.1929 64 bit (AMD64)] astropy 5.2.1 Numpy 1.24.2 pyerfa 2.0.0.1 Scipy 1.10.0 Matplotlib 3.6.3"}
{"_id":"q-en-astropy__astropy-14539-96602704b39deefd2e70e54f3fccec37161219c01f28b45c3dc790899919c2c3","text":"may sometimes report differences between identical files In some scenarios, may report differences between identical files, even when comparing the same file to itself. This may be caused by improper handling of VLAs (variable-length arrays). only reports differences in files if they exist. Comparing a file to itself should never yield a difference. Prints out: I suspect the handling of VLAs is the culprit here as I couldn't reproduce the bug without using at least one VLA column. Windows-10-10.0.-SP0 Python 3.10.10 (tags\/v3.10.10:, Feb 7 2023, 17:20:36) [MSC v.1929 64 bit (AMD64)] astropy 5.2.1 Numpy 1.24.2 pyerfa 2.0.0.1 Scipy 1.10.0 Matplotlib 3.6.3"}
{"_id":"q-en-astropy__astropy-14566-fac0a5277416d62a66b4c7fe933d00a56930365aa1ad1bc509162f0580e1f058","text":"Problem with \"decimalyear\" applied to MaskedColumn type The \"decimalyear\" type produces odd errors when applied to the MaskedColumn type. The \"jyear\" type does not, and appears to behave properly. I would expect \"decimalyear\" and \"jyear\" to work similarly, although the difference between them is not clear from the documentation. Linux-5.19.0-35-generic-x86_64-with-glibc2.35 Python 3.9.13 (main, Aug 25 2022, 23:26:10) [GCC 11.2.0] astropy 5.1"}
{"_id":"q-en-astropy__astropy-14578-3ff5dba4a14a4d7f90864e7d359b7175bff4d20ea469df51ca54e54b1c1da7ee","text":"Writing a Table to FITS fails if the table contains objects The following works fine: whereas the following fails: with This behaviour is seen whenever a Table contains an object, i.e. does not know how to deal with . I wonder if we want the Table API to write objects to files by their string representation as a default, or otherwise provide a more meaningful error message?"}
{"_id":"q-en-astropy__astropy-14590-70653929cd8d3a59e43001306862672f7074c44830595a5be05722e3eaf81f81","text":"TST: check fails with numpy-dev (TypeError: cannot write to unmasked output) Started popping up in numpy-dev jobs. is investigating."}
{"_id":"q-en-astropy__astropy-14598-9ecd41ea2b4da9a68c70362f133e1d2a59c8204e22bd7c59651bba70d47ae874","text":"Inconsistency in double single-quote ('') management in FITS Card The management of single-quotes in FITS cards seem correct, except sometimes when dealing with null strings, i.e. double single quotes (), which sometimes are transformed into single single quotes (). E.g.: gives If the null string is included in a larger value, the issue occurs at a different position: gives All card values should be handled properly. Linux-5.10.0-1029-oem-x86_64-with-glibc2.29 Python 3.8.10 (default, Mar 13 2023, 10:26:41) [GCC 9.4.0] astropy 5.2.1 Numpy 1.23.5 pyerfa 2.0.0 Scipy 1.10.0 Matplotlib 3.6.2"}
{"_id":"q-en-astropy__astropy-14628-31b7f41fc09b5b93b1da94e4f37e6550fb397d8e2db810826b75daca42c3fab3","text":"Make getting a topocentric ITRS position easier Right now, it is not easy to create ITRS coordinates for sources relative to a given location (rather than geocentric), to the level that we have specific instructions on how to calculate relative and then put these into an : This has led to numerous issues, the latest of which is It would be nice if as part of it would be possible to get a topocentric rather than a geocentric position. In , and (and below) to extend to take not just an but also a argument, with an implementation along : See . Labeling this a good first issue since it is easy code wise. However, writing the tests and documentation will require understanding of how ITRS and the associated coordinate transformations work."}
{"_id":"q-en-astropy__astropy-14701-d9cc66ee48f25fc7ba00032bbe7a2ae23544735c2918c47b1dd7bac239ec1a47","text":"Register Cosmology can now read and write to files. It would be nice to register with a method for exporting a Cosmology to a Latex table. There are good examples of IO with Cosmology at and documentation at I'm thinking the H0(..., format='latex')`."}
{"_id":"q-en-astropy__astropy-14702-57a838e927277fc908470324d29674f63c66572c3adff20726e3faa79669cdf7","text":"BUG: change representation of vs table.Table More often than not it's rather confusing (and annoying) that a VOTable and a Table look exactly the same, but obviously, they don't behave the same way and don't have the same methods available, etc. I would suggest to change the votable case of to something else, e.g. ."}
{"_id":"q-en-astropy__astropy-14907-3de14674ec8eb634ad47ba31fc49f21dbb642bade80548f0942fdb071124eb36","text":"TST: testtablegroupby[True] and testgroupbymasked[True] failed with numpy 1.25rc1 I see this in the predeps job that pulls in numpy 1.25rc1. Example log: Hard to discern between the other 100+ failures from and I do not understand why we didn't catch this earlier in devdeps. , does this look familiar to you? TST: testtablegroupby[True] and testgroupbymasked[True] failed with numpy 1.25rc1 I see this in the predeps job that pulls in numpy 1.25rc1. Example log: Hard to discern between the other 100+ failures from and I do not understand why we didn't catch this earlier in devdeps. , does this look familiar to you?"}
{"_id":"q-en-astropy__astropy-14938-c1a90e8b3381574a2595ebd4bedb4f847e8bf526dc1bd602b7514d339356261a","text":"Add registration label ‘’ to Cosmology IO The Cosmology write methods that leverage Table should have the same keys. Table has both “latex” and “”, so too should Cosmology. Register the method a second time, under No response"}
{"_id":"q-en-astropy__astropy-14966-a8bbc2198f5513444f292182e960bda5ba8953181221c80a1ade2710548293be","text":"QTable: Group keys don't have a unit When grouping a QTable with a column that contains a quantity, the keys only contain a float with a value, but not the full quantity: The keys should be complete, i.e. for a quantity column it should be a quantity with the proper unit. No response Linux-6.1.0-9-amd64-x86_64-with-glibc2.36 (Debian bookworm) Python 3.11.4 (main, Jun 7 2023, 10:13:09) [GCC 12.2.0] astropy 5.2.1 (also checked with 5.3) Numpy 1.24.2 pyerfa 2.0.0.3 Scipy 1.10.1 Matplotlib 3.6.3"}
{"_id":"q-en-astropy__astropy-14991-69c421444db61627464719aef88ff7d9a0b3a8bcb11d9af54cb1857695011805","text":"Error in distance calculations for w0wz Cosmologies I believe that the equation used to calculate the dedensityscale in is incorrect. Line 205 has After manually calculating the integral\/checking wolfram, I don't think it should be a negative in the exponent and should read: package from '...' run '...' error occurs. import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import astropy; print(\"astropy\", astropy.version) import numpy; print(\"Numpy\", numpy.version) import erfa; print(\"pyerfa\", erfa.version) import scipy; print(\"Scipy\", scipy.version) import matplotlib; print(\"Matplotlib\", matplotlib.version)"}
{"_id":"q-en-astropy__astropy-14995-7320b55f60352bff48433550840cac6c1295b05aa9f4e37d96af73563bbb8242","text":"In v5.3, NDDataRef mask propagation fails when one of the operand does not have a mask This applies to v5.3. It looks like when one of the operand does not have a mask, the mask propagation when doing arithmetic, in particular with fails. This is not a problem in v5.2. I don't know enough about how all that works, but it seems from the error that the operand without a mask is set as a mask of None's and then the bitwise_or tries to operate on an integer and a None and fails. When one of the operand does not have mask, the mask that exists should just be copied over to the output. Or whatever was done in that situation in v5.2 where there's no problem. This is with v5.3. With v5.2, there are no errors. import sys; print(\"Python\", sys.version) Python 3.10.11 (main, May 10 2023, 19:07:22) [Clang 14.0.6 ] import astropy; print(\"astropy\", astropy.version) astropy 5.3 import numpy; print(\"Numpy\", numpy.version) Numpy 1.24.3 import erfa; print(\"pyerfa\", erfa.version) pyerfa 2.0.0.3 import scipy; print(\"Scipy\", scipy.version) Scipy 1.10.1 import matplotlib; print(\"Matplotlib\", matplotlib.version) Matplotlib 3.7.1"}
{"_id":"q-en-astropy__astropy-6938-d440dcd13fc68a007e9796c986b8aea19184c570f9f9e99612ea2e421fc0ccfd","text":"Possible bug in related to D exponents I came across the following code in : I think this may be incorrect because as far as I can tell (it returns a copy). Commenting out this code doesn't cause any tests to fail so I think this code isn't being tested anyway."}
{"_id":"q-en-astropy__astropy-7008-39b3f590a6e25089915a08cdc6ecc8601b238120a61f3ff8633b3e94886f3ad6","text":"Context manager for constant versions For some use cases it would be helpful to have a context manager to set the version set of the constants. E.g., something like `"}
{"_id":"q-en-astropy__astropy-7166-23351b069fdb68f8a2c7e136acf595069dacdba34fa433fbf8288c94fa32e479","text":"InheritDocstrings metaclass doesn't work for properties Inside the InheritDocstrings metaclass it uses which returns for properties."}
{"_id":"q-en-astropy__astropy-7336-908fb7fe7947954c48932c988e79c2ffc4a84ede70f281192190c2ec59cc4bfa","text":"units.quantityinput decorator fails for constructors with type hinted return value -None I am using the decorator with typing hints for constructors, however when I add the correct return value for the constructor () then I get an exception, because has no attribute . The issue can be reproduced with the following file: which results in the following error: This has been tested on Fedora 27 with python 3.6.3, astropy 2.0.2 and numpy 1.13.3 all from Fedora's repository. The issue can be circumvented by not adding the return type typing hint. Unfortunately, then a static type checker cannot infer that this function returns nothing. Maybe the decorator could explicitly check whether None is returned and then omit the unit check."}
{"_id":"q-en-astropy__astropy-7441-4f9976353e9d5dc21de157190964ba5b674b1cda9590e3ac9439aa38e3317523","text":"should support conversion to Because of the inheritance has a method which is useless. It should have a method which returns a object or objects. Conversion to (for ) and (for ) would also be great."}
{"_id":"q-en-astropy__astropy-7671-06fd81ee9dbb051d8076a731e60ec3045ecb44b0dcb37abe45fab69accc8b8e9","text":"minversion failures The change in PR causes to fail in certain cases, e.g.: apparently because of a bug in LooseVersion (): Note that without the \".3\" it doesn't fail: and using pkgresources.parseversion (which was removed) works: CC:"}
{"_id":"q-en-astropy__astropy-7737-08fa9444ee64ddac9138a503309aa233c9023f12ff11d2b467ff7a36c926d6ec","text":"biweightlocation of a constant array returns nan Currently the robust mean estimator returns for an array with zero variance. eg: This is primarily because of a 0\/0 division in the code (case when the calculated mad of array in denominator becomes zero). Wouldn't it be better to catch this special case and return the median, instead of returning nan?"}
{"_id":"q-en-astropy__astropy-7746-e79ce75a811b46275785915caea224a8b77aabbdf6bfbfbeceafd0c74905856e","text":"Issue when passing empty lists\/arrays to WCS transformations The following should not fail but instead should return empty lists\/arrays:"}
{"_id":"q-en-astropy__astropy-7858-5ce040bce2d09739d69b10c1a0d419a7edef39aec0c52ffebbb995d388bb2e13","text":"Issue when transforming a single scalar coordinate with a 1D WCS The following example illustrates a bug when dealing with single scalar coordinates in 1D WCSes: - would you have a chance to look at this?"}
{"_id":"q-en-astropy__astropy-7973-e91016ab874b5fe7b5f8b5e70e591ff40b98ecf961cd7f8a85dbfd6b9e3625d5","text":"Record data size in the WCS object It is clear that there are practical reasons to have a record of the original data size in the WCS object. This in the past has been recorded in public attributes and and subsequently in the private and . There's along thread on why this should or should not be done in . More recently expanded this attribute to more than 2 axes. It also broke the ability to set the private attributes which was subsequently fixed in as a stop gap measure. We need to fix this permanently starting with v 1.3. There are two proposed ways to do it. public attributes, populated by the FITS header keywords. a subclass of which adds those and leave the original WCS to represent strictly the FITS WCS standard and nothing more. Either way a deprecation process will accompany the change. The purpose of this issue is to collect votes. I don't want to start a new long discussion. Keep in mind that is available for reference and only share new arguments. Since 1.3 is approaching we need to make a fast decision. So please share your preference here and volunteer for the work if possible."}
{"_id":"q-en-astropy__astropy-8005-be3f73ea53cd71b10aa10bb3944435e18552d2d39762898eb7959c88e7b5dcc4","text":"Cosmologies used in equivalencies The thermodynamic temperature equivalency appears to default to the Planck 2015 cosmology value: whereas in the equivalency in , the default cosmology is used: and this is currently WMAP9. This seems inconsistent, so can we try and make things more consistent for the 3.1 release? (I see this as bug-like, so I think it can be addressed after feature freeze). Note that this is related to - if we change the default cosmology to Planck 2015, then we can change the equivalency to use the default cosmology without changing results (for now)."}
{"_id":"q-en-astropy__astropy-8251-648be00d293608f5769e03e07e7028789c47de8b5efeba628a03531b1e86e815","text":"FITS-standard unit parsing fails on some types of exponents Why don't these work: When these all do: The non-working versions give e.g.: which is not how I would interpret the . Tested on 2.0.7 and 3.0.3"}
{"_id":"q-en-astropy__astropy-8263-b5e317ecdc8b711d1b5fbf6e7e29d5bf822cb7ae1e0a1cd2f9610c9a165b23f2","text":"units: 's \/ m' and 's \/ m' are not convertible To paraphrase from , exceptions like (and a few other cases with first, second and third powers of second over meter, curiously) started popping up in our travis tests as seen at https:\/\/travis- . For a brief overview, running and getting into : I've managed to figure out that this is localized to 3.1. As suggested by on slack, I ran with and I was able to identify the first bad commit as I'm currently trying to find out the exact cause of this bug in that commit, but I figured it's best to submit this issue here before I accidentally close this tab or something."}
{"_id":"q-en-astropy__astropy-8292-db9c28ec53f91ab473ca4876684ca30c5e187d79ad7b9f14edb0f64231d3e085","text":"Problem with the part of unit equivalencies? In the newly equivalencies: We notice that the implementation of seems to be wrong, as highlighted in the following figure: ! If , and , should it be equivalent to 140 Mpc, instead of 70Mpc? I can reproduce this so it is not a typo..."}
{"_id":"q-en-astropy__astropy-8339-e9ff3d1353a31f0f69e94f57cbe0cfb42c8634a5298572b2f22c3c90e37664cd","text":"ncpprior referenced before assignment in Bayesian Blocks There is a bug in the bayesian blocks algorithm of It's not a big deal so I show you below how to solve it directly. When I call: I get: You can fix this just by changing: adding an else sentence EDIT: Text formatting ncpprior referenced before assignment in Bayesian Blocks There is a bug in the bayesian blocks algorithm of It's not a big deal so I show you below how to solve it directly. When I call: I get: You can fix this just by changing: adding an else sentence EDIT: Text formatting"}
{"_id":"q-en-astropy__astropy-8519-2f7debb9a3130fb90e09b88a08b4aca8ce1c9f2cb7212f4f5d226258f5eb0845","text":"Adding\/subtracting ABmag Quantities loses the \"type\" of magnitude The following code raises a , because it appears the math operation lost track of the \"type\" of magnitude. and are both , so I would expect their difference to also be an ABmag. prints the following, and then raises: If the is changed to , the exception is different:"}
{"_id":"q-en-astropy__astropy-8707-df426390c28245253b16da77e1f960511700340a699594e15705092dd5486d51","text":"Header.fromstring does not accept Python 3 bytes According to , the method \"...creates an HDU header from a byte string containing the entire header data.\" By \"byte string\" here it really means the type which on Python 2 could be raw binary data, but on Python 3 explicitly is not. In fact it does work on Python 3's unicode s, but here it assumes that the data can be ASCII-encoded. Its counterpart, will work with files opened in text or binary mode. So probably the simplest solution for now (as opposed to adding new methods or something like that) is to change to accept unicode or bytes string types. likely needs a similar treatment."}
{"_id":"q-en-astropy__astropy-8747-b418caf81ee8cd25479d04d619c89c4c5f332d046d3362bd6663eb8bc1768b42","text":"Support new clip ufunc Starting with numpy 1.17, will be based on a , which means we can ensure it works properly with . (Until we do so, we might also get failures.)"}
{"_id":"q-en-astropy__astropy-8872-6c8e3e5d749e0e6338568c17abb96a53cb4faeafe9e242dae121242c2d388b73","text":"float16 quantities get upgraded to float64 automatically When trying to create a from a (not something I actually intended to do, I was experimenting while investigating other issue) it gets upgraded automatically to , which is something that does not happen with other float types: However: Somewhat related:"}
{"_id":"q-en-django__django-10087-f44e0343b51f233dffe0f16fc0a26a2b9ceae4239f7a075edcaaaf086e75f649","text":"Misleading sqlmigrate \"App 'apps.somethings' does not have migrations.\" error message Description This ticket is very similar to As shown above, validation should be sqlmigrate."}
{"_id":"q-en-django__django-10097-d60cd6230b125695d226fcf60ae5344a53e4203d7c9527c479bfcdba82dbf1dd","text":"Make URLValidator reject invalid characters in the username and password Description (last modified by Tim Bell) Since , core.validators.URLValidator accepts URLs with usernames and passwords. RFC 1738 section 3.1 requires \"Within the user and password field, any \":\", \" or \"\/\" must be encoded\"; however, those characters are currently accepted without being %-encoded. That allows certain invalid URLs to pass validation incorrectly. (The issue originates in Diego Perini's ​gist, from which the implementation in was derived.) An example URL that should be invalid is ; furthermore, many of the test cases in would be rendered valid under the current implementation by appending a query string of the form to them. I note Tim Graham's concern about adding complexity to the validation regex. However, I take the opposite position to Danilo Bargen about invalid URL edge cases: it's not fine if invalid URLs (even so-called \"edge cases\") are accepted when the regex could be fixed simply to reject them correctly. I also note that a URL of the form above was encountered in a production setting, so that this is a genuine use case, not merely an academic exercise. Pull request: ​ Make URLValidator reject invalid characters in the username and password Description (last modified by Tim Bell) Since , core.validators.URLValidator accepts URLs with usernames and passwords. RFC 1738 section 3.1 requires \"Within the user and password field, any \":\", \" or \"\/\" must be encoded\"; however, those characters are currently accepted without being %-encoded. That allows certain invalid URLs to pass validation incorrectly. (The issue originates in Diego Perini's ​gist, from which the implementation in was derived.) An example URL that should be invalid is ; furthermore, many of the test cases in would be rendered valid under the current implementation by appending a query string of the form to them. I note Tim Graham's concern about adding complexity to the validation regex. However, I take the opposite position to Danilo Bargen about invalid URL edge cases: it's not fine if invalid URLs (even so-called \"edge cases\") are accepted when the regex could be fixed simply to reject them correctly. I also note that a URL of the form above was encountered in a production setting, so that this is a genuine use case, not merely an academic exercise. Pull request: ​"}
{"_id":"q-en-django__django-10213-7071a822fdcac193d80abe07cdb656f9e1f20f1ab2de2672ce8fedd04f94352f","text":"Add an option to django-admin to always colorize output Description With Django management commands, it is currently possible disable colors with the --no-colors flag. What I'd like to have is basically the other side of the coin: a --force-colors flag that instructs Django to output ANSI color sequences in cases it would disable colors by default (typically, when the output is piped to another command, as documented). My real world use-case is the following one: I have a custom Django command to import data. I run this command myself, and I'd like to send a colored log (HTML seems perfect for this) to the data curators. I can use the ​ utility for this, but that doesn't work since Django disable colors when the output is piped. Other *nix commands have a special flag for this exact use-case, for example $ ls --color=always"}
{"_id":"q-en-django__django-10287-fc44be0f2a3dc4d3ee5d2c46675b56bd7819b8fab7b3c51ff3202f315ada4084","text":"Add validation of related fields and lookups in model Meta.ordering Description When the ordering class member in Meta of a model contains a field from a related model, and that field does not exist, django's makemigrations does not throw an error. However, if it is a direct field member of the same class, makemigrations does throw an error. Example below tested on Django 2.0.5 from import models class Agreement(models.Model): agreementid = models.AutoField(verbosename='ID', serialize=False, autocreated=True, primarykey=True) #class Meta: # generates error in makemigrations # app.Agreement: (models.E015) 'ordering' refers to the nonexistent field 'id'. # ordering = ['id'] class Order(models.Model): agreement = models.ForeignKey(Agreement, models.DONOTHING) class Meta: # does not generate error in makemigrations # but does so during runtime # e.g. [x for x in ()] ordering = ['agreement_id']"}
{"_id":"q-en-django__django-10316-0ca7db74145e28bd0de10cca100c4c28a2f3b4c5292f054d1758573ba75abf10","text":"diffsettings raises misleading exception message if using settings.configure() Description If, rather than using the env var DJANGOSETTINGSMODULE one uses settings.configure(...), attempting to call diffsettings can fail because it calls () unconditionally, with the exception Requested settings, but settings are not configured. You must either define the environment variable DJANGOSETTINGSMODULE or call settings.configure() before accessing settings. were the call instead: if not settings.configured: () things would work correctly."}
{"_id":"q-en-django__django-10390-a8122eafa8083ffb396aee2b17004bf79f625df667a932ff067946f2aaacc3e5","text":"Trunc() should allow passing isdst resolution to avoid NonExistentTimeError\/AmbiguousTimeError Description (last modified by Alexander Holmbäck) When Trunc() truncates to a nonexisting or ambiguous datetime, the exception raised by pytz remains unhandled. The expected behavior would, IMO, be to not check the validity of truncated dates. This test for example: import datetime import pytz from import Trunc from import TestCase from import timezone from .models import Log class TestTruncateToInvalidTime(TestCase): def testtruncatetodstendsstockholm(self): tzinfo = pytz.timezone('Europe\/Stockholm') timestamp = datetime.datetime(2018, 10, 28, 2, tzinfo=tzinfo) (timestamp=timestamp) logs = Log.objects.annotate(day=Trunc('timestamp', 'hour')).all() timezone.activate(tzinfo) self.assertEqual(logs[0], 28) Results in the following error: ERROR: testtruncatetodstendsstockholm () Traceback (most recent call last): File \"\", line 47, in testtruncatetodstendsstockholm self.assertEqual(logs[0], 28) File \"\", line 303, in getitem qs.fetchall() File \"\", line 1190, in fetchall self.resultcache = list(self.iterableclass(self)) File \"\", line 64, in iter for row in compiler.resultsiter(results): File \"\", line 1013, in applyconverters value = converter(value, expression, connection) File \"\", line 225, in convertvalue value = timezone.makeaware(value, ) File \"\", line 270, in makeaware return timezone.localize(value, isdst=is_dst) File \"\/home\/alex\/.virtualenvs\/djangodev\/lib\/python3.6\/site-\", line 363, in localize raise AmbiguousTimeError(dt) pytz.exceptions.AmbiguousTimeError: 2018-10-28 02:00:00"}
{"_id":"q-en-django__django-10426-26b21b90e8bdd1659268e941f24ae6c8d94bb87e020aaf7c45804144db6792a0","text":"Add app_label validation to showmigrations Description The app label validation was to some commands as above tickets. But showmigrations command doesn't contain it. So I think the validation should be to showmigrations command."}
{"_id":"q-en-django__django-10531-20cfbee57de3b120317dac9c53d906922aeb06ff0d572abbd81b678ece7824ad","text":"Models history doesn't use verbose names Description The history for a model object (within the admin section) should show human-readable messages, favoring verbose names over field names. However, this is not currently the case. For example, consider a model with the following class variable: pubdate = models.DateTimeField(\"date published\") Changing the publication date for an object of that model, will display \"Changed pubdate.\" in its admin history, rather than \"Change date published.\" as one would expect (as older versions of Django did)."}
{"_id":"q-en-django__django-10554-625f6e5c0eb5bc426676e9deaa790752f162ff12bc47c7f0b3c06b6f0269726d","text":"Union queryset with ordering breaks on ordering with derived querysets Description (last modified by Sergei Maertens) May be related to Simple reproduction (the exact models are not relevant I think): Dimension.objects.valueslist('id', flat=True) <QuerySet [10, 11, 12, 13, 14, 15, 16, 17, 18]qs = ( (pkin=[10, 11]) .union((pkin=[16, 17]) .orderby('order') ) qs <QuerySet [<Dimension: boeksoort, <Dimension: grootboek, <Dimension: kenteken, <Dimension: activa]qs.orderby().valueslist('pk', flat=True) <QuerySet [16, 11, 10, 17]qs [breaks] Traceback: Traceback (most recent call last): File \"<input\", line 1, in <moduleqs File \"\/home\/bbt\/.virtualenvs\/ispnext\/lib\/python3.6\/site-\", line 248, in repr data = list(self[:REPROUTPUTSIZE + 1]) File \"\/home\/bbt\/.virtualenvs\/ispnext\/lib\/python3.6\/site-\", line 272, in iter self.fetchall() File \"\/home\/bbt\/.virtualenvs\/ispnext\/lib\/python3.6\/site-\", line 1179, in fetchall self.resultcache = list(self.iterableclass(self)) File \"\/home\/bbt\/.virtualenvs\/ispnext\/lib\/python3.6\/site-\", line 53, in iter results = compiler.executesql(chunkedfetch=self.chunkedfetch, chunksize=self.chunksize) File \"\/home\/bbt\/.virtualenvs\/ispnext\/lib\/python3.6\/site-\", line 1068, in executesql cursor.execute(sql, params) File \"\/home\/bbt\/.virtualenvs\/ispnext\/lib\/python3.6\/site-\", line 100, in execute return super().execute(sql, params) File \"\/home\/bbt\/.virtualenvs\/ispnext\/lib\/python3.6\/site-\", line 68, in execute return self.executewithwrappers(sql, params, many=False, executor=self.execute) File \"\/home\/bbt\/.virtualenvs\/ispnext\/lib\/python3.6\/site-\", line 77, in executewithwrappers return executor(sql, params, many, context) File \"\/home\/bbt\/.virtualenvs\/ispnext\/lib\/python3.6\/site-\", line 85, in execute return (sql, params) File \"\/home\/bbt\/.virtualenvs\/ispnext\/lib\/python3.6\/site-\", line 89, in exit raise djexcvalue.withtraceback(traceback) from excvalue File \"\/home\/bbt\/.virtualenvs\/ispnext\/lib\/python3.6\/site-\", line 85, in execute return (sql, params) ORDER BY position 4 is not in select list LINE 1: ...dimensionsdimension\".\"id\" IN (16, 17)) ORDER BY (4) ASC LIM... ^ Evaluating the qs instead of creating a new qs makes the code proceed as expected. [ for dim in qs]"}
{"_id":"q-en-django__django-10606-21c29fc179a48842e4db975840a634474b7d3af5989cac5787ea85522e32c24f","text":"Using RelatedOnlyFieldListFilter with reverse ManyToMany crashes Description (last modified by Tim Graham) Using RelatedOnlyFieldListFilter with a reverse ManyToMany relation causes this exception: getchoices() got an unexpected keyword argument 'limitchoicesto' This method in ForeignObjectRel.getchoices is missing the parameter that Field.get_choices has. Pull Request: ​ Demo of how to trigger bug: ​"}
{"_id":"q-en-django__django-10643-96244e0984c067d69eada8a131e3061c9fffe40865b2b6d914402f88deff7e38","text":"Allow icontains lookup to accept uuids with or without dashes Description We have Django 2.1 project with model admin which includes an UUIDField in listdisplay and searchfields. The UUID is displayed with dashes on changelist (e.g. \"-6852-47be-82be-\") and if the user cut'n'paste it to the search field, I would expect admin to find it. This works however only on Postgres but fails on Oracle. I can understand why this happens (Oracle backend stores uuid as string) and I believe I can workaround it by customizing getsearchresults but I think should be internal thing that Django handles gracefully - search should be possible by the value as displayed in admin."}
{"_id":"q-en-django__django-10680-4286857edc867c855ade1ce3ff60dc1bcea0c605fd2937464a13021b99ecd76e","text":"Refactor AutoField logic into a mixin, implement checks and validators. Description Currently AutoField inherits from Field and BigAutoField from AutoField. In effect they largely redefine IntegerField and BigIntegerField respectively, but add in the auto field \"behaviour\". As a result they do not perform some of the system checks, e.g. maxlength warning, nor the validation checks, e.g. range checks, that the integer fields do. The proposal is to move all the auto field \"behaviour\" into a new AutoFieldMixin and fix AutoField and BigAutoField to inherit from this new mixin and IntegerField and BigIntegerField respectively. Many attributes and methods would be nicely inherited from the correct parent field type without requiring redefinition: description emptystringsallowed defaulterrormessages getprepvalue() topython() AutoField and BigAutoField could also inherit the following checks from IntegerField: IntegerField.checkmaxlengthwarning() AutoField and BigAutoField could also perform minimum and maximum value validation checks inherited from IntegerField. This should be backwards compatible and potentially will make it easier to define new types of auto fields based on other fields in the future."}
{"_id":"q-en-django__django-10730-6402c553e3d07d983559a8fe25c18b1892a11f052df1f854b48f0ae2eadfb74a","text":"Infinite loop in ExceptionReporter.gettracebackframes() Description The following code generates a cause\/context cycle ( is excvalue): except WrapperException as exc: raise The ​while exc_value loop then never terminates."}
{"_id":"q-en-django__django-10737-0132c906d479fdbb9aa495fa2c25023ce8fc2a627d18ee2d71ff04d67b9fa31a","text":"Raised FieldError's should include the field that caused the error Description Currently there are many places in ​ that raise a FieldError however frustratingly the field that caused the error is not included. I'm proposing that, where possible, all raised FieldError's raised should include, as part of the error message, the field that caused the error."}
{"_id":"q-en-django__django-10853-e347ef92db438d6169823cba66206ab624df759216394df3ce68db1f446fd203","text":"SQLite (pre 3.25.0) does not support window functions, raises OperationalError Description Window functions are supported in SQLite 3.25.0 and newer, but Python 3.6.7 and 3.7.1 only ships with SQLite 3.21.0. Window function syntax is invalid for older versions. As per the title, window functions therefore aren't supported, but Django doesn't check the SQLite version or availability of window functions. Instead, when the generated SQL is executed, the sqlite3 Python library raises the SQLite syntax error as sqlite3.OperationalError, which in turn is reraised as I believe this is not intended behaviour, as it is incredibly confusing, and not documented. Typically, if a database feature is not supported, Django will explicitly raise an error when attempting it, rather than allowing the SQL execution to fail. It is also normally documented. The following code raises an exception (it should work for any model): from import F, Window from import RowNumber MyModel.objects.annotate(rn=Window(expression=RowNumber(), orderby=[F('pk')])) Basic Python code that will also raise sqlite3.OperationalError: import sqlite3 conn = sqlite3.connect(\":memory:\") c = () c.execute(\"CREATE TABLE t0(x INTEGER PRIMARY KEY, y TEXT)\") c.execute(\"INSERT INTO t0 VALUES (1, 'aaa'), (2, 'ccc'), (3, 'bbb')\") c.execute(\"SELECT x, y, rownumber() OVER (ORDER BY y) AS row_number FROM t0 ORDER BY x\") Tested on master branch (commit ) with Python 3.6.6 64-bit (Windows 10 x64). This likely also affects 2.0 and 2.1 branches."}
{"_id":"q-en-django__django-10880-6f34cda154ce08f93273c4b3ac1ad7d508dcdae91fc2780a2a67748a5e799365","text":"Query syntax error with condition and distinct combination Description A Count annotation containing both a Case condition and a distinct=True param produces a query error on Django 2.2 (whatever the db backend). A space is missing at least (... COUNT(DISTINCTCASE WHEN ...)."}
{"_id":"q-en-django__django-10910-5c20f9b4c2eccde99aff64ff29c2299bfbe9b3210087a70012fde29f02873972","text":"Using database functions with tzinfo=datetime.timezone(datetime.timedelta(...)) results in an incorrect query Description I haven’t checked this bug with other databases, but it definitely works improperly with postgres. Django ORM create incorrect query when I use timezone determined like \"timezone(timedelta(hours=somehours))\". \"timezone(timedelta(hours=5))\" in query will look like \"UTC+05:00\", but postgres doesn't know this timezone name and handle it as POSIX style. \"UTC\" part will be interpreted as some zone abbreviation and timezone will be shifted by 5 hours to the west (positive shift is shift to the west in accordance with POSIX standart), i.e. actually timezone will be equal to UTC-5. From ​ : \"In addition to the timezone names and abbreviations, PostgreSQL will accept POSIX-style time zone specifications of the form STDoffset or STDoffsetDST, where STD is a zone abbreviation, offset is a numeric offset in hours west from UTC\" Checked with: django==2.1.5 psycopg2==2.7.6.1 postgreSQL==10.6 Using the following example model: class test(models.Model): class Meta: dbtable = 'testtimezones' datetime = models.DateTimeField() Sample of bug is bellow: from datetime import timezone, timedelta from import ExtractWeekDay from import test from import ExtractHour from pytz import timezone as pytztimezone print(test.objects.annotate(hour=ExtractHour('datetime')).values('datetime', 'hour').get()) {'datetime': datetime.datetime(2018, 1, 1, 7, 0, tzinfo=<UTC), 'hour': 7} tz = timezone(timedelta(hours=5)) print(tz) UTC+05:00 print(test.objects.annotate(hour=ExtractHour('datetime', tzinfo=tz)).values('datetime', 'hour').get()) {'datetime': datetime.datetime(2018, 1, 1, 7, 0, tzinfo=<UTC), 'hour': 2} print(test.objects.annotate(hour=ExtractHour('datetime', tzinfo=tz)).values('datetime', 'hour').query) SELECT \"testtimezones\".\"datetime\", EXTRACT('hour' FROM \"testtimezones\".\"datetime\" AT TIME ZONE 'UTC+05:00') AS \"hour\" FROM \"testtimezones\" tz2 = pytz_timezone('Asia\/Yekaterinburg') print(tz2) Asia\/Yekaterinburg print(test.objects.annotate(hour=ExtractHour('datetime', tzinfo=tz2)).values('datetime', 'hour').get()) {'datetime': datetime.datetime(2018, 1, 1, 7, 0, tzinfo=<UTC), 'hour': 12}"}
{"_id":"q-en-django__django-10914-c6226eae94baeb4318d69387f106e17188c895da8165a8614cd76876c17d02d9","text":"Set default FILEUPLOADPERMISSION to 0o644. Description Hello, As far as I can see, the ​File Uploads documentation page does not mention any permission issues. What I would like to see is a warning that in absence of explicitly configured FILEUPLOADPERMISSIONS, the permissions for a file uploaded to FileSystemStorage might not be consistent depending on whether a MemoryUploadedFile or a TemporaryUploadedFile was used for temporary storage of the uploaded data (which, with the default FILEUPLOADHANDLERS, in turn depends on the uploaded data size). The tempfile.NamedTemporaryFile + sequence causes the resulting file permissions to be 0o0600 on some systems (I experience it here on CentOS 7.4.1708 and Python 3.6.5). In all probability, the implementation of Python's built-in tempfile module explicitly sets such permissions for temporary files due to security considerations. I found mentions of this issue ​on GitHub, but did not manage to find any existing bug report in Django's bug tracker."}
{"_id":"q-en-django__django-10924-1ee4c6d5e236363a13dc1e69b9c8a45c52258d3ea8a31651b80c13bb89d28295","text":"Allow FilePathField path to accept a callable. Description I have a special case where I want to create a model containing the path to some local files on the server\/dev machine. Seeing as the place where these files are stored is different on different machines I have the following: import os from import settings from import models class LocalFiles(models.Model): name = models.CharField(maxlength=255) file = models.FilePathField((settings.LOCALFILEDIR, 'exampledir')) Now when running makemigrations it will resolve the path based on the machine it is being run on. Eg: \/home\/<username\/serverfiles\/exampledir I had to manually change the migration to include the () part to not break this when running the migration on production\/other machine."}
{"_id":"q-en-django__django-10939-3a732b122e4cc82481b07b0020a3f346731dd73b5adea7e4ee5cd5214d59af2f","text":"ModelAdmin with custom widgets, inlines, and filterhorizontal can merge media in broken order Description when a modeadmin have a inline with a filed has its own media js no need jquery , and have a one to many field show filterhorizontal, the problem appear. there will be MediaOrderConflictWarning and load before jquery."}
{"_id":"q-en-django__django-10957-52b7b5d3918c671cd89fce9195b72449a2f5f5a034ae01d3e2956039d1ca82b4","text":"Deprecate ugettext(), ugettextlazy(), ugettextnoop(), ungettext(), and ungettextlazy() Description Along the lines of (Cleanups when no supported version of Django supports Python 2 anymore), the legacy functions in -- ugettext(), ugettextlazy(), ugettextnoop(), ungettext(), and ungettextlazy() -- are simple aliases that remain for Python 2 Unicode backwards compatibility. As other compatibility layers have been cleaned up, these shims can be deprecated for removal."}
{"_id":"q-en-django__django-10973-d40142cf52de60302da8cefddefd17f8c6fa62c07dfd656c21721e1fa1e5befa","text":"Use and PGPASSWORD for client in postgres backend Description ​ was in python 3.5 (which is the minimum version since Django 2.1). This function allows you to pass a custom environment for the subprocess. Using this in to set PGPASSWORD simplifies the code and makes it more reliable."}
{"_id":"q-en-django__django-10997-d9ca76d2836540e96249446d7b120603501175b2f8b2d521ac80524e50769ef2","text":"Show applied datetime in showmigrations Description (last modified by Timothy Schilling) My idea is to add the applied datetime value to the showmigrations command. I've run into the case where I'm working on a branch that involves a number of migrations across various apps, but then have to switch to a different branch which has different migrations. It can be troublesome to determine which migrations are new and need to be rolled back. I've recently started looking at the django_migrations table sorted on the applied column to determine which I've run recently. This would make switching between branches involving conflicting migrations easier. There was some brief discussion ​here. I've initially implemented this so that it would only apply to the --list option with a --verbosity of 2 and above. ​Here's what I have so far. I wasn't sure how to handle backporting. ​PR Edited to strikeout old PR and reference the one to origin."}
{"_id":"q-en-django__django-10999-bc33a4304ffe5175f3a3a15a431c9ad7559baed43d417d2877375d7eafce88a3","text":"Fix parseduration() for some negative durations Description The ​ defines: standarddurationre = re.compile( r'^' r'(?:(?P<days-?\\d+) (days?, )?)?' r'((?:(?P<hours-?\\d+):)(?=\\d+:\\d+))?' r'(?:(?P<minutes-?\\d+):)?' r'(?P<seconds-?\\d+)' r'(?:.(?P<microseconds\\d{1,6})\\d{0,6})?' r'$' ) that doesn't match to negative durations, because of the <hoursdefinition final (lookahead) part does not have '-?' in it. The following will work: r'((?:(?P<hours-?\\d+):)(?=-?\\d+:-?\\d+))?' (Thanks to Konstantin Senichev for finding the fix.)"}
{"_id":"q-en-django__django-11001-dc84b8bcea42b702953a343b869f01bfc002b1f6a10c4dcca466c237792b7ced","text":"Incorrect removal of orderby clause created as multiline RawSQL Description Hi. The SQLCompiler is ripping off one of my \"order by\" clause, because he \"thinks\" the clause was already \"seen\" (in SQLCompiler.getorderby()). I'm using expressions written as multiline RawSQLs, which are similar but not the same. The bug is located in SQLCompiler.getorderby(), somewhere around line computing part of SQL query without ordering: withoutordering = (sql).group(1) The sql variable contains multiline sql. As a result, the self.orderingparts regular expression is returning just a line containing ASC or DESC words. This line is to seen set, and because my raw queries have identical last lines, only the first clasue is returing from SQLCompiler.getorderby(). As a quick\/temporal fix I can suggest making sql variable clean of newline characters, like this: sqloneline = ' '.join(('\\n')) withoutordering = (sqloneline).group(1) Note: beware of unicode (Py2.x u'') and EOL dragons (\\r). Example of my query: return ().orderby( RawSQL(, []).desc(), RawSQL(, []).asc(), RawSQL(, []).desc()) The is returing accordingly: ' then 2 else 1 end)' ' else null end' ' else null end' Second RawSQL with a else null end part is removed from query. The fun thing is that the issue can be solved by workaround by adding a space or any other char to the last line. So in case of RawSQL I can just say, that current implementation of avoiding duplicates in order by clause works only for special\/rare cases (or does not work in all cases). The bug filed here is about wrong identification of duplicates (because it compares only last line of SQL passed to order by clause). Hope my notes will help you fixing the issue. Sorry for my english."}
{"_id":"q-en-django__django-11003-471ea6d32bd860aa8afa99db104d9c56003ee27a0c2d20a84f80915ec006ade4","text":"Initialising disabled ModelChoiceField yields 'Select a valid choice'-error despite initialised option being valid Description (last modified by Mariusz Felisiak) I have a form with a ModelChoiceField that gets initialised to a specific value using getinitial in that form's View. This value is a valid choice for that Model. I don't want the user to be able to change the option on the form, but it needs to be displayed nonetheless. When I set disabled=True on that field in , submitting the form yields the following error: <ul class=\"errorlist\"<lifieldname<ul class=\"errorlist\"<liSelect a valid choice. That choice is not one of the available choices.<\/li<\/ul<\/li<\/ul. Firstly, I would like to comment on the general quality of the error message, as it is not very useful: It does not return which choice it considers invalid. Including this information would make the message much more informative, and would avoid sending people on a wild goose chase to discover what the message could possibly mean. Secondly, if a field is disabled but does contain a valid choice, validating the form should work and not trigger an error. Edit: Adding the tofield_name option to the form field fixes the problem. However, when disabled=True is not present, this is not required. This is probably related to the bugfix for this bug:"}
{"_id":"q-en-django__django-11011-a6e42dbe4306054ce0fff304967b91b19537860c79f2432b71ff78425a3d73da","text":"Make FileResponse always set Content-Disposition header. Description (last modified by Piotr Kunicki) FileResponse currently sets the Content-Disposition header only if as_attachment is true. Setting it explicitly to, e.g. 'inline; filename=\"\"' in the other case would allow the browser to set a default name for that inline file in case a user attempts to download it with the 'Save image as...' option. That filename value is also visible in the title of the tab when image is being viewed directly in Firefox (at least v56). Created a pull request: ​"}
{"_id":"q-en-django__django-11019-9c62e60c01c0e57b05a1de1e093787ac10741db895e6ff0d98c7d8efdef9d031","text":"Merging 3 or more media objects can throw unnecessary MediaOrderConflictWarnings Description Consider the following form definition, where text-editor- depends on text- but all other JS files are independent: from django import forms class ColorPicker(forms.Widget): class Media: js = ['color-'] class SimpleTextWidget(forms.Widget): class Media: js = ['text-'] class FancyTextWidget(forms.Widget): class Media: js = ['text-', 'text-editor-', 'color-'] class MyForm(forms.Form): background_color = forms.CharField(widget=ColorPicker()) intro = forms.CharField(widget=SimpleTextWidget()) body = forms.CharField(widget=FancyTextWidget()) Django should be able to resolve the JS files for the final form into the order text-, text-editor-, color- However, accessing MyForm().media results in: MediaOrderConflictWarning: Detected duplicate Media files in an opposite order: text-editor- text- MediaOrderConflictWarning, Media(css={}, js=['text-editor-', 'color-', 'text-']) The MediaOrderConflictWarning is a result of the order that the additions happen in: ColorPicker().media + SimpleTextWidget().media produces Media(css={}, js=['color-', 'text-']), which (wrongly) imposes the constraint that color- must appear before text- The final result is particularly unintuitive here, as it's worse than the \"naïve\" result produced by Django 1.11 before order-checking was (color-, text-, text-editor-), and the pair of files reported in the warning message seems wrong too (aren't color- and text- the wrong-ordered ones?)"}
{"_id":"q-en-django__django-11034-2f042a59103edef77315d5676dcaae0db747c99b8b815aea5287f5d989c05e71","text":"Disabled field in admin use hard coded labelsuffix Description In the class helpers.AdminReadonlyField, the labeltag method has hard coded \":\" It is nor configurable neither able to tranlate (in French we use a space before \":\")."}
{"_id":"q-en-django__django-11039-e490ee9b8f3b5bbfcb95c7693e3726923336ec69645f412160f073786b55b9b0","text":"sqlmigrate wraps it's outpout in BEGIN\/COMMIT even if the database doesn't support transactional DDL Description (last modified by Simon Charette) The migration executor only adds the outer BEGIN\/COMMIT ​if the migration is atomic and ​the schema editor can rollback DDL but the current sqlmigrate logic only takes into consideration. The issue can be addressed by Changing sqlmigrate ​assignment of self.outputtransaction to consider connection.features.canrollbackddl as well. Adding a test in based on ​an existing test for non-atomic migrations that mocks connection.features.canrollbackddl to False instead of overdidding MIGRATION_MODULES to point to a non-atomic migration. I marked the ticket as easy picking because I included the above guidelines but feel free to uncheck it if you deem it inappropriate."}
{"_id":"q-en-django__django-11044-f26f664379aeca83d672f1385bda9c2402d64d3336e4f9db23467759e06f5bf9","text":"Make Form data\/files initialize with an empty MultiValueDict rather than dict Description You might have a look here: ​ None is converted to a regular dict but not to a QueryDict. Methods of the form might rely on the API of a QueryDict such as 'iterlists' or 'getlist' which a regular dict doesn't provide."}
{"_id":"q-en-django__django-11049-a058e4205daf6eec2a7d62bec31464e0082be5cf86c6b1ee92a993dee1348b62","text":"Correct expected format in invalid DurationField error message Description If you enter a duration \"14:00\" into a duration field, it translates to \"00:14:00\" which is 14 minutes. The current error message for invalid DurationField says that this should be the format of durations: \"[DD] [HH:[MM:]]ss[.uuuuuu]\". But according to the actual behaviour, it should be: \"[DD] [[HH:]MM:]ss[.uuuuuu]\", because seconds are mandatory, minutes are optional, and hours are optional if minutes are provided. This seems to be a mistake in all Django versions that support the DurationField. Also the duration fields could have a default help_text with the requested format, because the syntax is not self-explanatory."}
{"_id":"q-en-django__django-11053-902d830198ebe40b3fa2fc4d8096864db739fb14b1c230fecae9c2eb1a8a5337","text":"Disallow non-uppercase settings in settings.configure() Description Contradictory to the documentation, settings.configure() and UserSettingsHolder.getattr() allow non-upper settings to be set and retrieved, respectively. This allows for the following to happen: import types from import settings settings.configure(types.SimpleNamespace(foo=\"bar\"), baz=\"qux\") print(, ) # Doesn't error"}
{"_id":"q-en-django__django-11057-7987a327a2e5096e5e473da2fea0a08de56df921e12031e2f2065415778bc39c","text":"admin check for AuthenticationMiddleware should allow subclasses Description I am attempting an upgrade from 2.1 to 2.2b1. I have a custom middleware in place for authentication which essentially wraps around the base AuthenticationMiddleware and sets\/unsets a cookie upon login and logout. This was working with Django 2.1 as my authentication middleware. Now, when I attempt to start my application, the check fails with admin.E408: ?: (admin.E408) '' must be in MIDDLEWARE in order to use the admin application. Is this a bug\/regression, or have I been using the middleware incorrectly this whole time?"}
{"_id":"q-en-django__django-11062-a411065b7f5f2c00cedf12ccd10ea2eb53a6f6ca92c8b61b49140e708400f2c9","text":"Using an annotated field calculated with in aggregate results in ProgrammingError Description Aggregating most annotated fields works as expected, but if I put a DateTimeField through Extract during the annotate step, I get a ProgrammingError when trying to aggregate. class MyModel(models.Model): usagetime = models.DateTimeField() usage = models.FloatField() I would like to take the whole queryset, and calculate hourly usages. I figured using the transform would suit my needs well. This is the sample piece of code that, in a perfect scenario, would give me a dictionary filled with key value pairs, where key is the hour, and value is the sum of usages measured in that hour. houraggregates = {} for i in range(24): houraggregates['{}{}'.format(\"am\" if i < 12 else \"pm\", i)] = Sum(\"usage\", filter=Q(hour=i)) usages = MyModel.objects.annotate(hour=Extract(\"usagetime\", \"hour\")).aggregate(**houraggregates) Unfortunately, I get the following error: Traceback (most recent call last): File \"\/home\/jan\/project\/env\/lib\/python3.6\/site-\", line 85, in execute return (sql, params) psycopg2.ProgrammingError: column \"col2\" does not exist LINE 1: ...CT \"packagemymodel\".\"id\" AS Col1, EXTRACT('hour' FROM \"col2\" A... This occured to me while using Django 2.1.7. It doesn't work on 2.2b1, but I have tested this solution on Django 1.8 and it works, which is why I am filing this bug report. My Python version is 3.6.7 and I'm using PostgreSQL 10.6."}
{"_id":"q-en-django__django-11066-bb1157cb95aa5c6cde2d5d89ec393da1a0ad3179765d3c6494876f9bd54ec448","text":"() doesn't save the content type on the correct database Description The commit in question: ​ The specific lines in question: ​ with (using=db): (updatefields={'model'}) The issue: For some background, we run a dynamic database router and have no \"real\" databases configured in the settings file, just a default sqlite3 backend which is never actually generated or used. We forked the management command and modified it to accept a dictionary containing database connection parameters as the --database argument. The dynamic database router is based on, and very similar to this: ​ This has worked beautifully for all migrations up until this point. The issue we're running into is that when attempting to run a migration which contains a call to migrations.RenameModel, and while specifying the database parameters to the migrate command, the migration fails with an OperationalError, stating that no such table: djangocontenttypes exists. After having exhaustively stepped through the traceback, it appears that even though the call is wrapped in the with (using=db) context manager, the actual database operation is being attempted on the default database (which in our case does not exist) rather than the database specified via (on line 15 of the same file) and thus fails loudly. So, I believe that: (updatefields={'model'}) should be (using=db, update_fields={'model'})"}
{"_id":"q-en-django__django-11070-fd465e2b185b9e9e02c19809c68dda23c37a0f0aeb303f213218067276a1c6f2","text":"Add autocomplete attribute to fields Description (last modified by CHI Cheng) Add autocomplete=username\/email\/current-password\/new-password to builtin forms. Pull request: ​ The most useful one is autocomplete=new-password, which prevents browsers prefill with current password, Chrome will also suggest a random strong password for users who turned on account sync. Related docs: ​ ​ ​"}
{"_id":"q-en-django__django-11085-898cc93b071e19340bc10275a91bd24a2776a749fb362f9c77956d24e133fda9","text":"Custom model metaclasses cannot access the attribute dict in init Description In Django <=2.2, it is possible for models to define a custom metaclass (as a subclass of ) and access the attribute dict of the class being defined: from import models class PageBase(): def init(cls, name, bases, dct): super(PageBase, cls).init(name, bases, dct) if 'magic' in dct: print(\"enabling magic on %s\" % (name)) class Page(models.Model, metaclass=PageBase): magic = True title = models.CharField(maxlength=255) As of commit , this fails because all attributes without a contributeto_class method are popped from the dict in . (This pattern is used by Wagtail's Page model ​ , so this is causing various failures against django stable\/2.2.x.)"}
{"_id":"q-en-django__django-11087-9ba7b40869878f9e83ddb29155c05dc64a566b98e41c07ad2fbcec568ddb855c","text":"Optimize .delete() to use only required fields. Description Hi! We're in the process of upgrading our Django 1.11 installation from Python 2.7 to Python 3.6, however are hitting an unexpected UnicodeDecodeError during a .delete() run by our daily data purging management command. STR: Have an existing Django 1.11 project running under Python 2.7.15 that uses mysqlclient-python v1.3.13 to connect to MySQL server v5.7.23, with Django's DATABASES options including 'charset': 'utf8mb4' (​) Update to Python 3.6.8 Run the daily cycledata Django management command against the dev instance's DB: ​ ​ Expected: That the cycledata management command succeeds, like it did under Python 2. Actual: Traceback (most recent call last): File \"\", line 16, in <moduleexecutefromcommandline() File \"\", line 364, in executefromcommandline utility.execute() File \"\", line 356, in execute self.fetchcommand(subcommand).runfromargv() File \"\", line 988, in nrwrapperBaseCommandrunfromargv return wrapped(args, kwargs) File \"\", line 283, in runfromargv self.execute(args, cmdoptions) File \"\", line 330, in execute output = (args, options) File \"\", line 139, in literalwrapper return wrapped(args, kwargs) File \"\", line 62, in handle options['sleeptime']) File \"\", line 461, in cycledata (guidin=jobschunk).delete() File \"\", line 619, in delete collector.collect(delquery) File \"\", line 223, in collect field.remotefield.ondelete(self, field, subobjs, ) File \"\", line 17, in CASCADE , ) File \"\", line 222, in collect elif subobjs: File \"\", line 254, in bool self.fetchall() File \"\", line 1121, in fetchall self.resultcache = list(self.iterableclass(self)) File \"\", line 53, in iter results = compiler.executesql(chunkedfetch=self.chunkedfetch) File \"\", line 899, in executesql raise originalexception File \"\", line 889, in executesql cursor.execute(sql, params) File \"\", line 64, in execute return (sql, params) File \"\", line 101, in execute return (query, args) File \"\", line 25, in execute args, *kwargs) File \"\", line 250, in execute self.errorhandler(self, exc, value) File \"\", line 50, in defaulterrorhandler raise errorvalue File \"\", line 247, in execute res = (query) File \"\", line 413, in query self.postgetresult() File \"\", line 417, in postgetresult = self.fetchrow(0) File \"\", line 385, in fetchrow return (size, self.fetchtype) File \"\", line 231, in stringdecoder return s.decode(db.encoding) UnicodeDecodeError: 'utf-8' codec can't decode byte 0xed in position 78: invalid continuation byte The exception occurs during the .delete() of Jobs, here: ​ Enabling debug logging of Django's DB backend, shows the generated SQL to be: SELECT FROM job WHERE (job.repositoryid = 1 AND job.submittime < '2018-10-21 11:03:32.') LIMIT 1; args=(1, '2018-10-21 11:03:32.') SELECT , failureline.jobguid, failureline.repositoryid, failureline.joblogid, , , , failureline.subtest, , failureline.expected, failureline.message, failureline.signature, , , failureline.stackwalkstdout, failureline.stackwalkstderr, failureline.bestclassificationid, failureline.bestisverified, failureline.created, failureline.modified FROM failureline WHERE failureline.jobguid IN ('-b854-4300-969a-\/0'); args=('-b854-4300-969a-\/0',) SELECT , job.repositoryid, , job.projectspecificid, job.autoclassifystatus, job.coalescedtoguid, job.signatureid, job.buildplatformid, job.machineplatformid, job.machineid, job.optioncollectionhash, job.jobtypeid, job.jobgroupid, job.productid, job.failureclassificationid, , , , , job.submittime, job.starttime, job.endtime, job.lastmodified, job.runningeta, , FROM job WHERE IN ('-b854-4300-969a-\/0'); args=('-b854-4300-969a-\/0',) SELECT , , , , FROM joblog WHERE IN (); args=(,) [2019-02-18 11:03:33,403] DEBUG [] (0.107) SELECT , failureline.jobguid, failureline.repositoryid, failureline.joblogid, , , , failureline.subtest, , failureline.expected, failureline.message, failureline.signature, , , failureline.stackwalkstdout, failureline.stackwalkstderr, failureline.bestclassificationid, failureline.bestisverified, failureline.created, failureline.modified FROM failureline WHERE failureline.joblogid IN (, ); args=(, ) SELECT , , , textlogstep.started, textlogstep.finished, textlogstep.startedlinenumber, textlogstep.finishedlinenumber, FROM textlogstep WHERE IN (); args=(,) SELECT , , , textlogerror.linenumber FROM textlogerror WHERE IN (); args=(,) Querying the textlogerror table for those ids shows there to be junk values in its line field. These are from data inserted when using Python 2.7, which presumably wasn't validating the unicode escape sequences being used. There appear to be two issues here: mysqlclient-python's behaviour differs depending on Python version - under Python 3 it defaults useunicode to True, which means it attempts to decode the line field but fails (since it doesn't use 'replace' or 'ignore'). This seems like something that the Django ORM should try to protect against (eg by setting useunicode to the same value on all Python versions and handling the unicode conversion itself), given it generally handles any implementation differences in layers lower than the ORM. the UnicodeDecodeError is occurring for a field () that is not actually needed for the .delete() (it's not a primary key etc), so Django shouldn't be fetching that field regardless when making the textlogerror SELECT query (Plus ideally Django would support cascade deletes, so we wouldn't need to use the current .delete() approach; ticket ) Fixing issue (2) would presumably also improve .delete() performance. Related: ​"}
{"_id":"q-en-django__django-11088-391836e48343f7e8ac27f76211c0626d8e8f3dace603b4e99e9464b8e998ee62","text":"default='' (non-bytestring) on BinaryField crashes some migration operations Description (last modified by James) Description Initial migration has a default value '' for BinaryField. Later, change default value to b'' and migrate. Trying to undo this migration fails. It seems like '' is allowed during migration, but not in reverse migration. Related issue Default value for BinaryField Reproduce Python 3.6.0, Django 1.10.6, Postgres 9.5.4 startproject djangoproject startapp firstapp class TableOne(models.Model): field1 = models.BinaryField(default = '') makemigrations firstapp migrate firstapp 0001 Modify class TableOne(models.Model): field1 = models.BinaryField(default = b'') migrate firstapp 0002 migrate firstapp 0001 Error: TypeError: can't escape str to binary Traceback (most recent call last): File \"\", line 22, in <moduleexecutefromcommandline() File \"C:\\Py\\py364\\lib\\site-packages\\django\\core\\management\\\", line 367, in executefromcommandline utility.execute() File \"C:\\Py\\py364\\lib\\site-packages\\django\\core\\management\\\", line 359, in execute self.fetchcommand(subcommand).runfromargv() File \"C:\\Py\\py364\\lib\\site-packages\\django\\core\\management\\\", line 294, in runfromargv self.execute(args, cmdoptions) File \"C:\\Py\\py364\\lib\\site-packages\\django\\core\\management\\\", line 345, in execute output = (args, options) File \"C:\\Py\\py364\\lib\\site-packages\\django\\core\\management\\commands\\\", line 204, in handle fakeinitial=fakeinitial, File \"C:\\Py\\py364\\lib\\site-packages\\django\\db\\migrations\\\", line 119, in migrate state = self.migrateallbackwards(plan, fullplan, fake=fake) File \"C:\\Py\\py364\\lib\\site-packages\\django\\db\\migrations\\\", line 194, in migrateallbackwards self.unapplymigration(states[migration], migration, fake=fake) File \"C:\\Py\\py364\\lib\\site-packages\\django\\db\\migrations\\\", line 264, in unapplymigration state = migration.unapply(state, schemaeditor) File \"C:\\Py\\py364\\lib\\site-packages\\django\\db\\migrations\\\", line 178, in unapply operation.databasebackwards(self.applabel, schemaeditor, fromstate, tostate) File \"C:\\Py\\py364\\lib\\site-packages\\django\\db\\migrations\\operations\\\", line 210, in databasebackwards self.databaseforwards(applabel, schemaeditor, fromstate, tostate) File \"C:\\Py\\py364\\lib\\site-packages\\django\\db\\migrations\\operations\\\", line 205, in databaseforwards schemaeditor.alterfield(frommodel, fromfield, tofield) File \"C:\\Py\\py364\\lib\\site-packages\\django\\db\\backends\\base\\\", line 506, in alterfield olddbparams, newdbparams, strict) File \"C:\\Py\\py364\\lib\\site-packages\\django\\db\\backends\\postgresql\\\", line 118, in alterfield newdbparams, strict, File \"C:\\Py\\py364\\lib\\site-packages\\django\\db\\backends\\base\\\", line 660, in alterfield params, File \"C:\\Py\\py364\\lib\\site-packages\\django\\db\\backends\\base\\\", line 112, in execute cursor.execute(sql, params) File \"C:\\Py\\py364\\lib\\site-packages\\django\\db\\backends\\\", line 80, in execute return super(CursorDebugWrapper, self).execute(sql, params) File \"C:\\Py\\py364\\lib\\site-packages\\django\\db\\backends\\\", line 65, in execute return (sql, params) TypeError: can't escape str to binary Notes site-packages\\django\\db\\backends\\base\\ def effectivedefault(self, field): determines default as an empty <class 'str', when (default = '') Possible Fix? site-packages\\django\\db\\backends\\base\\ ~line 197 def effectivedefault(self, field): if field.hasdefault(): default = field.getdefault() if field.getinternaltype() == \"BinaryField\" and not default: default = six.binarytype() elif not and and field.emptystringsallowed: if field.getinternaltype() == \"BinaryField\": default = six.binarytype() else: default = six.texttype() elif getattr(field, 'auto_now', False)"}
{"_id":"q-en-django__django-11095-afd82a2c8451129edbc4dc9f8f33a48db68f4eb2c587c906fb3d18a25dc3a4e8","text":"add ModelAdmin.getinlines() hook to allow set inlines based on the request or model instance. Description add ModelAdmin.getinlines() hook to allow set inlines based on the request or model instance. Currently, We can override the method getinlineinstances to do such a thing, but a for loop should be copied to my code. So I wished add a hook get_inlines(request, obj=None)"}
{"_id":"q-en-django__django-11096-aa6e0aca0453d3f0a26ba46b8e7e4358b28721e228bd44defc6c1329f2e18600","text":"Support the use of adminorderfield on properties in the admin. Description The ModelAdmin class allows to specify properties from the model in listdisplay. Normally non-fields cannot be sorted. This can be changed by setting adminorderfield. This doesn't work on properties, while it is possible to do so with shortdescription. The ​code that checks for shortdescription checks if it handles a property and returns if available. The files that check for adminorderfield do not handle the case that it might be a property. The relevant files are probably: (​GitHub) (​ GitHub)"}
{"_id":"q-en-django__django-11099-65d3afc1c991bf158b049b13e41808c8d0a1a8f7c03a53110866f4293d2057e3","text":"UsernameValidator allows trailing newline in usernames Description ASCIIUsernameValidator and UnicodeUsernameValidator use the regex r'^[\\w. The intent is to only allow alphanumeric characters as well as ., +, and -. However, a little known quirk of Python regexes is that $ will also match a trailing newline. Therefore, the user name validators will accept usernames which end with a newline. You can avoid this behavior by instead using \\A and \\Z to terminate regexes. For example, the validator regex could be changed to r'\\A[\\w. in order to reject usernames that end with a newline. I am not sure how to officially post a patch, but the required change is trivial - using the regex above in the two validators in"}
{"_id":"q-en-django__django-11115-5e2b582912279d21ceecef9cefec4008bb02e71c75081a9590485633c3957f5b","text":"Make it easier to track down the offending models for AlreadyRegistered exceptions Description I've just updated 20+ packages locally in a project's requirements file, one of which has caused: [2016-10-18 15:00:18,667] ERROR [django.request:256] Internal Server Error: \/browserid\/csrf\/ Traceback (most recent call last): File \"\/home\/vagrant\/venv\/local\/lib\/python2.7\/site-\", line 223, in getresponse response = middlewaremethod(request, response) File \"\/home\/vagrant\/venv\/local\/lib\/python2.7\/site-\", line 129, in processresponse panel.generatestats(request, response) File \"\/home\/vagrant\/venv\/local\/lib\/python2.7\/site-\", line 41, in generatestats match = resolve() File \"\/home\/vagrant\/venv\/local\/lib\/python2.7\/site-\", line 521, in resolve return getresolver(urlconf).resolve(path) File \"\/home\/vagrant\/venv\/local\/lib\/python2.7\/site-\", line 365, in resolve for pattern in self.urlpatterns: File \"\/home\/vagrant\/venv\/local\/lib\/python2.7\/site-\", line 401, in urlpatterns patterns = getattr(self.urlconfmodule, \"urlpatterns\", self.urlconfmodule) File \"\/home\/vagrant\/venv\/local\/lib\/python2.7\/site-\", line 395, in urlconfmodule self.urlconfmodule = importmodule(self.urlconfname) File \"\", line 37, in importmodule import(name) File \"\", line 12, in <modulebrowseridadmin.copyregistry() File \"\/home\/vagrant\/venv\/local\/lib\/python2.7\/site-\", line 39, in copyregistry self.register(model, ) File \"\/home\/vagrant\/venv\/local\/lib\/python2.7\/site-\", line 90, in register raise AlreadyRegistered('The model %s is already registered' % ) AlreadyRegistered: The model Site is already registered This model isn't defined in the project's own models so in this particular case it must be django-browserid now clashing with one of the packages that was updated. Rather than having to bisect, it would be much more helpful if this exception gave more details about the already-registered model, so I know which package\/app is clashing with the later app's attempts to re-register it."}
{"_id":"q-en-django__django-11119-b888c4106cc9fd84ed4e4afa26fb140a579e0ada1da95646c6a260e22bc3c9c5","text":"Engine.rendertostring() should honor the autoescape attribute Description In Engine.rendertostring, a Context is created without specifying the engine autoescape attribute. So if you create en engine with autoescape=False and then call its rendertostring() method, the result will always be autoescaped. It was probably overlooked in []."}
{"_id":"q-en-django__django-11129-4535eabba9c9171d55e597ce3dc0a26d1ab0cb2c67a33779a5ceae09c3f76f52","text":"Incorrect error message with () that requires a join on an annotated F expression Description (last modified by Asif Saifuddin Auvi) I ran into a bit of a nasty error yesterday on Django 1.11, Postgres 9.5 where I was trying to do an update using an F expression where the value in that expression referred to an annotation of an annotation, each using a Case statement inside the annotation. The error message it was giving me was something like the following: ProgrammingError Traceback (most recent call last) <ipython-input-1-29e48364245cin <module() ----1 MyModel.objects.causingproblems() in causingproblems(self) 833 ).update( --834 myfield=F('myannotatedfield'), 835 ) 836 in update(self, **kwargs) 645 query.annotations = None 646 with (, savepoint=False): --647 rows = query.getcompiler().executesql(CURSOR) 648 self.resultcache = None 649 return rows in executesql(self, resulttype) 1189 related queries are not available. 1190 \"\"\" -1191 cursor = super(SQLUpdateCompiler, self).executesql(resulttype) 1192 try: 1193 rows = cursor.rowcount if cursor else 0 in executesql(self, resulttype, chunkedfetch) 884 # silencing when dropping Python 2 compatibility. 885 pass --886 raise originalexception 887 888 if resulttype == CURSOR: ProgrammingError: missing FROM-clause entry for table \"myappmymodel\" LINE 1: END, \"afield\" = CASE WHEN CASE WHEN (\"myappmy... Spent a while trying to debug where it was going wrong, ended up narrowing down the problem to trying to update a field using an F expression on an annotated field which included a join in it. I'm not sure if better error message would have helped in my case but in the much simpler test attached it'd be useful if the first and second assertions raised the same or similar error message."}
{"_id":"q-en-django__django-11133-bb0b622b0bc692130a389a9034cb42558292054676a29123a800856ff8fa0f4e","text":"HttpResponse doesn't handle memoryview objects Description I am trying to write a BinaryField retrieved from the database into a HttpResponse. When the database is Sqlite this works correctly, but Postgresql returns the contents of the field as a memoryview object and it seems like current Django doesn't like this combination: from import HttpResponse response = HttpResponse(\"My Content\") response.content response = HttpResponse(b\"My Content\") response.content response = HttpResponse(memoryview(b\"My Content\")) response.content"}
{"_id":"q-en-django__django-11138-47413ea335e92f2beb2e19bb9ba726a8773bbb1b728eaade97bd19ea82af2000","text":"TIMEZONE value in DATABASES settings is not used when making dates timezone-aware on MySQL, SQLite, and Oracle. Description (last modified by Victor Talpaert) (We assume the mysql backends) I can set TIMEZONE several times in , one for the global django app, and one for each database (see ​ (ref1)) Typical usage would be for a legacy database where datetimes are not stored in UTC. No date lookup Querying my database takes this setting into account, e.g. : In USETZ = True TIMEZONE = 'Europe\/Paris' # tz1 DATABASES = { 'legacy': { 'ENGINE': '', 'OPTIONS': { 'readdefaultfile': '', }, 'TIMEZONE': 'Europe\/Paris', # tz2 }, 'default' : { 'ENGINE': '', 'OPTIONS': { 'readdefaultfile': '', }, } } In the shell dt = timezone.makeaware(datetime.datetime(2017, 7, 6, 20, 50)) dt datetime.datetime(2017, 7, 6, 20, 50, tzinfo=<DstTzInfo 'Europe\/Paris' CEST+2:00:00 DST) (mydatetimefield=dt).exists() True This works because my database reads '2017-07-06 20:50:00' With date lookup Related doc ​ (ref2) But this does not work, while it logically should (()).exists() False The related SQL query from DEBUG is : SELECT (1) AS FROM WHERE DATE(CONVERTTZ(., 'UTC', 'Europe\/Paris')) = '2017-07-06' LIMIT 1; () Note that I haven't filled the timezone table in MySQL, so the result should be True in this case, but could be False close to midnight. Related doc is ​ Two things are wrong. First, conversion should be from Paris to Paris, instead of UTC to Paris. The conversion should go from the database timezone tz2 to the django app one tz1. Indeed from ref1 and ref2: When USETZ is True and the database doesn’t support time zones (e.g. SQLite, MySQL, Oracle), Django reads and writes datetimes in local time according to this option if it is set and in UTC if it isn’t. When USETZ is True, fields are converted to the current time zone before filtering Secondly, when tz1 == tz2, there should be no need to use CONVERTTZ and the query will work without timezone tables in MySQL. The explicit queries are : mysqlSELECT (1) AS FROM WHERE . = '2017-07-06 20:50:00' LIMIT 1; +---+ +---+ +---+ 1 row in set (0.00 sec) mysqlSELECT (1) AS FROM WHERE DATE(.) = '2017-07-06' LIMIT 1; +---+ +---+ +---+ 1 row in set (0.00 sec) I understand that the date lookup can have some history, but I find the behaviour illogical and undesired. Would you agree there is a problem here? EDIT : line where 'UTC' is forced disregarding the database setting ​ PS: ​stackoverflow question"}
{"_id":"q-en-django__django-11141-763b5437546434496c6918708ede541338552790c20a4b79e9c473178d39c620","text":"Allow migrations directories without files Description (last modified by Tim Graham) Background: In python 3 a package with no is implicitly a namespace package, so it has no file attribute. The migrate command currently checks for existence of a file attribute on the migrations package. This check was introduced in , because the file attribute was used in migration file discovery. However, in migration file discovery was changed to use pkgutil.itermodules (), instead of direct filesystem access. pkgutil. itermodules() uses the package's path list, which exists on implicit namespace packages. As a result, the file check is no longer needed, and in fact prevents migrate from working on namespace packages (implicit or otherwise). Related work:"}
{"_id":"q-en-django__django-11149-d75588145634da66ab77cfd5a623b2081e80e04c5f5a9bc4ade8109b91b33372","text":"Admin inlines for auto-created ManyToManyFields are editable if the user only has the view permission Description From Replying to Will Gordon: This seems to have regressed in (at least) 2.1. I have 2 view only permissions. I have a ManyToManyField represented in my main model as a TabularInline. But, my user with view only permissions can now add or remove these items at will! I am having the same issue, so I assume this is a bug. I did not find Will had created a separate ticket. class Photo(models.Model): pass class Report(models.Model): photos = models.ManyToManyField(Photo) class ReportPhotoInlineModelAdmin(admin.TabularInline): model = showchangelink = True"}
{"_id":"q-en-django__django-11155-c540e0b308f29bdb3a664ce49782527634a8321555ce8b2548ec005c0c3ea043","text":"Support setting Secure, HttpOnly, SameSite on the language cookie Description I propose to add the following settings, with the following default values: LANGUAGECOOKIESECURE = False LANGUAGECOOKIEHTTPONLY = False LANGUAGECOOKIESAMESITE = None The default values maintain the current behavior. These settings do not provide much security value, since the language is not secret or sensitive. This was also discussed briefly here: ​ The reasons I'd like to add them are: Sometimes auditors require them. I personally prefer to set them unless I have a reason not to. Browsers are starting to strongly nudge toward HttpOnly and Secure when possible, e.g. ​"}
{"_id":"q-en-django__django-11163-9254583e1eeac65c9976501156db68c89e2fec47e1c9782315be896e19802336","text":"modeltodict() should return an empty dict for an empty list of fields. Description Been called as modeltodict(instance, fields=[]) function should return empty dict, because no fields were requested. But it returns all fields The problem point is if fields and f.name not in fields: which should be if fields is not None and f.name not in fields: PR: ​"}
{"_id":"q-en-django__django-11165-d798dea29ff741f60bdf0f1c9ace2a0fea23a4d952d004c6b2f4b5f6cb4bceee","text":"New HTTPRequest.headers not usable in templates because of hyphens Description With the release of 2.2, I took the opportunity from the new ​HTTPRequest.headers object to clean up old code using e.g. request.META['HTTPXREALIP'] to request.headers['X-Real-IP']. However, this new approach does not work with templates, as ​variable lookups cannot use hyphens. Could the object contain a parallel set of keys in underscored variables? e.g. request.headers['foo-bar'] is also available in request.headers['foobar'] and can be looked up with {{ request.headers.foo_bar }} in a template?"}
{"_id":"q-en-django__django-11166-230aad66feb43c37e8912d85d12f74c436092365e020055c1b62a54ae0a9c530","text":"Admin app has too hard a dependency on sessions app Description Since (), released in 2.2, the admin app checks whether the django.contrib.sessions app is in INSTALLEDAPPS. Some projects may have opted to use a replacement session management app such as ​ – the admin app claims to be incompatible with such a configuration, even if it actually means \"I'm going to need some session management that works like django.contrib.sessions\". Maybe it would be better to get rid of the app check and do what's being done for various middleware in the checks function anyway, e.g. something like if not containssubclass('django.contrib.sessions.middleware.SessionMiddleware', settings.MIDDLEWARE): (checks.Error( \"'django.contrib.sessions.middleware.SessionMiddleware' must \" \"be in MIDDLEWARE in order to use the admin application.\", id='admin.E4XX', )) – this would be out-of-the-box compatible with e.g. Qsessions. The obvious workaround is to just re-add django.contrib.sessions back into INSTALLEDAPPS which kinda works, but has the mild but unfortunate side effect of forcibly enabling the model and migrations, (re-)adding a useless django_session table into the database upon migration."}
{"_id":"q-en-django__django-11169-e0398db1e6c892a81d14efe81dee906e630e8767d820ce06c19eab513fc8154b","text":"Add new system check message when custom error handler '' cannot be imported Description (last modified by Alasdair Nicol) checks for the signatures of custom error handlers. When the '' cannot be imported, it raises ModuleNotFoundError or ViewDoesNotExist, as seen in this Stack Overflow question: ​ I suggest we catch the exception, and add another check code, e.g. urls.E008: The custom ` cannot be imported."}
{"_id":"q-en-django__django-11170-a6c05d988ba0c6e668bf8596a93771ffd0f17cbb7f825735e16cff5494d03ae3","text":"UnicodeDecodeError when loading debug templates. Description While studying Django, I had an error where the page was not properly printed. In the log window, UnicodeDecodeError was continuously printed, and in the template, the sentence 'A server error occured. Please contact the administrator' was printed So when I checked the file of Django 2.2version, I found that the dotdotdot wrapped by the <spantag on the 239th line was changed to ellipsis. Apparently, the version of Django 2.1.8 was a dotdotdot. So I took steps to change the 239th line's ellipsis to dotdotdot. Or, when reading the file from inside the file, the encoding format was set to utf-8. This enabled me to resolve the error. Did you intentionally modify the technicalhtml file?"}
{"_id":"q-en-django__django-11177-8aa7256a177512846e30af46ec9d2cb91c4e1e708839f138fd78882eb82041ff","text":"Django 2.2 autoreloader is failing intermittently (not using watchman) Description Since upgrading to Django 2.2 the autoreloader is working differently and is not always picking up changes to .py files to reload runserver. It seems to work consistently once on a file, but multiple changes to the same file will not necessarily trigger a reload. It's definitely a change in the behavior to previous versions of Django (up to 2.1.7) I'm not using watchman. I'm running runserver in a python 3.7 slim docker container like runserver 0.0.0.0:8001 As a workaround I have tried to add an edit a different .py -- that doesn't seem to help consistently. I've also noticed that the problem seems to sometimes go away by itself -- perhaps there is some change to the timing of the autoreloader which is not clear? For example and ignore interval of a certain duration."}
{"_id":"q-en-django__django-11179-4708d52b6439f5bc5b5c21baa042c116741726f52d8cd086db2773c1b0633f20","text":"delete() on instances of models without any dependencies doesn't clear PKs. Description Deleting any model with no dependencies not updates the PK on the model. It should be set to None after .delete() call. See Should update the model line 280."}
{"_id":"q-en-django__django-11185-6da05900df281aa26011a043bb9189ab5082518ffe80b25ba3cc5cfe71763eca","text":"(keep_parents=True) does not preserve all parent reverse relationships Description (last modified by Stephen Brown) There was a partial fix for this in , but it doesn't cater for relationships toward parents of parents, and so on."}
{"_id":"q-en-django__django-11194-ff610a18498a7309f49c9968c0b3d8752530d1467fb0b46acc75e511935e8f32","text":"Allow specifying a Feed language different from settings.LANGUAGECODE Description RSS uses settings.LANGUAGECODE as \"language tag\" for the RSS feed I couldnt find a way to change it.. so I made this patch ​"}
{"_id":"q-en-django__django-11205-5a3e415cf64dab111689bd334e6cb55818106d4f8ddca87ee2561e82491b60f0","text":"Inefficient SQL generated when counting a ManyToMany Description When calling count() on an unfiltered many to many relation, a useless join is included in the SQL that makes it much slower than it should be. On my dataset, the difference is 1000ms to 100ms, because an index-only scan can be used. This is the SQL that is currently generated: SELECT COUNT() AS \"count\" FROM \"appfoo\" INNER JOIN \"appfoobar\" ON (\"appfoo\".\"id\" = \"appfoobar\".\"fooid\") WHERE \"appfoobar\".\"fooid\" = ?; This is the SQL that should be generated: SELECT COUNT() AS \"count\" FROM \"appfoobar\" WHERE \"appfoobar\".\"foo_id\" = ?; This optimization can only be applied when there are no filters applied, because then the join is used to satisfy the filters. In the no-filters case, only the through table needs to be consulted."}
{"_id":"q-en-django__django-11206-2dc4f73590928734fc1fc20d24826b42e430ea78eaed7c209af7caaaa3d242cd","text":"renders small decimals in exponential notation. Description When using utils.numberformat with decimalpos, extremely small numbers get displayed using exponential notation. from import format as nformat nformat(Decimal('1e-199'), '.', decimalpos=2) '0.00' nformat(Decimal('1e-200'), '.', decimalpos=2) '1.00e-200' This is caused by a hardcoded cut-off point in the internal logic, but I would argue that when a decimal_pos argument is supplied and the number to be formatted is smaller in absolute size than what can be encoded using the provided number of decimal positions, the returned string should be 0.0000...000 instead."}
{"_id":"q-en-django__django-11211-72129bdf85c97333879f23ebb04a68d65729b5f78ac868348a6d88ed7e0e850e","text":"Prefetch related is not working when used GFK for model that uses UUID field as PK. Description How to reproduce: create model with UUID as primary key class Foo(models.Model): id = models.UUIDField(primarykey=True, default=uuid.uuid4, editable=False) ... create another model with GFK to model Foo class Bar(models.Model): foocontenttype = models.ForeignKey( ContentType, relatedname='actor', ondelete=models.CASCADE, dbindex=True ) fooobjectid = models.CharField(maxlength=255, dbindex=True) foo = GenericForeignKey('foocontenttype', 'fooobjectid') ... and try to get queryset with prefetch related (django orm engine return None for attribute foo): ().prefetch_related('foo') Thanks a lot for your attention! Also i wanna point out some related bug report from third party library in which previously i with that issue, maybe it would useful – ​"}
{"_id":"q-en-django__django-11214-bb0ca5df519f579012e204dcf5df08159d7cb7b6268fa206a977f5a010f6a13b","text":"Migration re-add check constraint continuously when check condition contains a range object. Description A CheckConstraint with a Q(xin=range(y, z)) condition is repeatedly deleted and re- when running makemigrations. models.CheckConstraint( check=models.Q(monthin=range(1, 13)), name='checkvalidmonth', ) The generated migration looks like this, so I suspect that the issue is because the range is converted into a tuple: operations = [ migrations.RemoveConstraint( modelname='monthlybudget', name='checkvalidmonth', ), migrations.AddConstraint( modelname='monthlybudget', constraint=models.CheckConstraint(check=models.Q(month_in=(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12)), name='checkvalid_month'), ), ] A sample project with this issue can be found here: ​ I'm open to fixing this issue myself, but I would like to clarify what would be a correct fix to this issue. I see at least two possible solutions, maybe three: Keep the range method call in the generated migration file Disallow using ranges in check constraints (At least on PostgreSQL, we could use a range expression in the database too.)"}
{"_id":"q-en-django__django-11216-2a8a60f3a3abf64173f9879f14a2b8a2fd9eabb5449fd425705ba6d87a69f4cd","text":"Prevent ManifestStaticFilesStorage from leaving behind intermediate files Description Currently when using ManifestStaticFilesStorage, collectstatic generates duplicate versions of some files. For example looking at the output for for Django 1.11.5, there is: This is exacerbated when using something like WhiteNoise's CompressedStaticFilesMixin, which then has to spend extra time generating gzip and Brotli compressed versions of every file (or else try and work around it: ​evansd\/whitenoise). This was called unavoidable\/working as intended according to: ​ However now that it's looking like CachedStaticFilesStorage will end up being removed (​mailing list thread; or at the very least we're discouraging people from using it, since it's buggy in several scenarios) - the intermediate files needn't be left behind. Even before CachedStaticFilesStorage ends up being removed, we could perhaps add a keepintermediatefiles property to HashedFilesMixin, that is set to False for CachedStaticFilesStorage and True for ManifestStaticFilesStorage, allowing us to fix the latter in the meantime."}
{"_id":"q-en-django__django-11234-8446bf7cfa5da1aa138c311e12c1a59c42565f37f10982e50457e4904a1ba409","text":"Prefetch related is not working when used GFK for model that uses FK as PK. Description Steps to reproduce Create Base model with an AutoField primary key class Base(models.Model): title = models.TextField() Create a model where the primary key is also a foreign key class Extended(models.Model): base = models.OneToOneField(Base, ondelete=models.CASCADE, primarykey=True) Create model with GenericForeignKey class Comment(models.Model): contenttype = models.ForeignKey(ContentType, ondelete=models.CASCADE) objectpk = models.TextField() contentobject = GenericForeignKey(ctfield=\"contenttype\", fkfield=\"objectpk\") Prefetch the GenericForeignKey field contentobject expecting it to have a value but get None instead. base = (title=\"foo\") extended = (base=base) (contentobject=extended) comment = Comment.objects.prefetchrelated(\"contentobject\").get() print(comment.content_object)"}
{"_id":"q-en-django__django-11239-92629e3ca1af581c6d8768c409bc5689c30755dda0c05c1dc89b3bed93eda775","text":"Add support for postgresql client certificates and key to dbshell. Description This bug is very similar to the A common security procedure for DB access is to require mutual TLS for the DB connection. This involves specifying a server certificate, client certificate, and client key when connecting. Django already supports this configuration, it looks like this: DATABASES = { 'default': { 'ENGINE': '', 'NAME': ('POSTGRESDBNAME'), 'USER': ('POSTGRESDBUSER'), 'HOST': 'postgres', 'PORT': '5432', 'SCHEMA': ('POSTGRESDBSCHEMA'), 'OPTIONS': { 'sslmode': 'verify-ca', 'sslrootcert': ('POSTGRESCLISSLCA', ''), 'sslcert': ('POSTGRESCLISSLCRT', ''), 'sslkey': ('POSTGRESCLISSLKEY', '') } } } However the dbshell command does not support the client cert params. Should be a trivial fix to add in support for the other 'ssl' parameters required here."}
{"_id":"q-en-django__django-11244-a89b9f9c0647f78d13152940dbde33c1ec41c52aad9e604aa3ae84ec92931368","text":"Remove the LANGUAGESBIDI<=LANGUAGES check. Description (Adding Nick Pope to Cc: as author of the commit referenced below) Since ​ there is a system check which verifies that LANGUAGESBIDI is a subset of LANGUAGES. This breaks almost all installations of Django using a custom LANGUAGES list which do not also override LANGUAGESBIDI -- probably many installations. All of them will either have to add a LANGUAGESBIDI override or silence translation.E005 when updating. If this is intentional then this change has to be mentioned in the release notes and documented somewhere. However, I don't really see the need to verify that LANGUAGESBIDI is a subset of LANGUAGES and propose that the easiest and also the best way to solve this is to remove the translation.E005 check again. Here's a test which currently fails but shouldn't in my opinion: diff --git index .. --- +++ -92,3 +92,7 class TranslationCheckTests(SimpleTestCase): self.assertEqual(checklanguagesettingsconsistent(None), [ Error(msg, id='translation.E005'), ]) + def testlanguageswithoutbidi(self): with self.settings(LANGUAGECODE='en', LANGUAGES=[('en', 'English')]): self.assertEqual(checklanguagesettingsconsistent(None), []) Remove the LANGUAGESBIDI<=LANGUAGES check. Description (Adding Nick Pope to Cc: as author of the commit referenced below) Since ​ there is a system check which verifies that LANGUAGESBIDI is a subset of LANGUAGES. This breaks almost all installations of Django using a custom LANGUAGES list which do not also override LANGUAGESBIDI -- probably many installations. All of them will either have to add a LANGUAGESBIDI override or silence translation.E005 when updating. If this is intentional then this change has to be mentioned in the release notes and documented somewhere. However, I don't really see the need to verify that LANGUAGESBIDI is a subset of LANGUAGES and propose that the easiest and also the best way to solve this is to remove the translation.E005 check again. Here's a test which currently fails but shouldn't in my opinion: diff --git index .. --- +++ -92,3 +92,7 class TranslationCheckTests(SimpleTestCase): self.assertEqual(checklanguagesettingsconsistent(None), [ Error(msg, id='translation.E005'), ]) + def testlanguageswithoutbidi(self): with self.settings(LANGUAGECODE='en', LANGUAGES=[('en', 'English')]): self.assertEqual(checklanguagesettingsconsistent(None), [])"}
{"_id":"q-en-django__django-11260-a88bc7a4ace8b746662992d13ffd11dfaeb41ae0b97c740f9a793744d92b6fb5","text":"inspectdb generates unique ForeignKey instead of OneToOneField. Description (last modified by Mariusz Felisiak) inspectdb generates unique ForeignKey instead of OneToOneField that caused raising fields.W342 warnings."}
{"_id":"q-en-django__django-11265-e44b3bc1b3f92389df0b0efafe50ff5f63c0d49cb7d9ac8a5b56e05af610000f","text":"Using exclude on annotated FilteredRelation doesn't work Description It looks like using exclude on queryset with annotated FilteredRelation give a FieldError on the annotation name. For exemple, in Django tests () if we change this : def testwithjoin(self): self.assertSequenceEqual( Author.objects.annotate( bookalice=FilteredRelation('book', condition=Q(booktitleiexact='poem by alice')), ).filter(bookaliceisnull=False), [self.author1] ) to this def testwithjoin(self): self.assertSequenceEqual( Author.objects.annotate( bookalice=FilteredRelation('book', condition=Q(booktitleiexact='poem by alice')), ).exclude(bookaliceisnull=False), [] ) You get the error : Traceback (most recent call last): File \"\", line 59, in testPartExecutor yield File \"\", line 605, in run testMethod() File \"\", line 99, in testwithjoinexclude ).filter(~Q(bookalice_isnull=False)), File \"\/home\/lucas\/dev\/overmind\/venvs\/release\/lib\/python3.6\/site-\", line 844, in filter return self.filterorexclude(False, args, kwargs) File \"\/home\/lucas\/dev\/overmind\/venvs\/release\/lib\/python3.6\/site-\", line 862, in filterorexclude (Q(args, kwargs)) File \"\/home\/lucas\/dev\/overmind\/venvs\/release\/lib\/python3.6\/site-\", line 1263, in addq clause, = (qobject, self.usedaliases) File \"\/home\/lucas\/dev\/overmind\/venvs\/release\/lib\/python3.6\/site-\", line 1281, in addq currentnegated, allowjoins, splitsubq) File \"\/home\/lucas\/dev\/overmind\/venvs\/release\/lib\/python3.6\/site-\", line 1287, in addq splitsubq=splitsubq, File \"\/home\/lucas\/dev\/overmind\/venvs\/release\/lib\/python3.6\/site-\", line 1204, in buildfilter return self.splitexclude(filterexpr, canreuse, e.nameswithpath) File \"\/home\/lucas\/dev\/overmind\/venvs\/release\/lib\/python3.6\/site-\", line 1604, in splitexclude query.addfilter(filterexpr) File \"\/home\/lucas\/dev\/overmind\/venvs\/release\/lib\/python3.6\/site-\", line 1249, in addfilter (Q(**{filterclause[0]: filterclause[1]})) File \"\/home\/lucas\/dev\/overmind\/venvs\/release\/lib\/python3.6\/site-\", line 1263, in addq clause, = (qobject, self.usedaliases) File \"\/home\/lucas\/dev\/overmind\/venvs\/release\/lib\/python3.6\/site-\", line 1287, in addq splitsubq=splitsubq, File \"\/home\/lucas\/dev\/overmind\/venvs\/release\/lib\/python3.6\/site-\", line 1164, in buildfilter lookups, parts, reffedexpression = self.solvelookuptype(arg) File \"\/home\/lucas\/dev\/overmind\/venvs\/release\/lib\/python3.6\/site-\", line 1028, in solvelookuptype , field, , lookupparts = self.namestopath(lookupsplitted, self.getmeta()) File \"\/home\/lucas\/dev\/overmind\/venvs\/release\/lib\/python3.6\/site-\", line 1389, in namestopath \"Choices are: %s\" % (name, \", \".join(available))) Cannot resolve keyword 'bookalice' into field. Choices are: book, contentobject, contenttype, contenttypeid, favoritebooks, id, name, objectid As far as I understand, the function splitexclude(self, filterexpr, canreuse, nameswith_path) seams to be the faulty one. A new query is created without all extra datas from the original query."}
{"_id":"q-en-django__django-11270-fee1b215fbc9a6a0d9033c9f26f52d7c2d5c054c35f104c6497e259d8eab1917","text":"Add validation of target argument in startapp command. Description When someone calls startapp command with a target argument, The app will overlay on the target directory. If the target directory has invalid name, the app can not be imported. So, I think it would be good to add validation on target directory's name like app name validation."}
{"_id":"q-en-django__django-11276-adb8806107e55f2fd6217772a1a1f51663b1acd90a40bd71fe0ba01254059997","text":"Use Python stdlib () to in () Description The function () partially duplicates the Python stdlib function (). We can replace this duplication with wider community developed version. () has been available since Python 3.2: ​ This function is also faster than Django's. As Python bug ​ concludes, using .replace() can be faster than .translate(). This function gets called numerous times when rendering templates. After making the change locally, I saw the following improvement: master: $ python -m timeit -s 'from import escape' 'escape(copyright)' loops, best of 5: 4.03 usec per loop branch: $ python -m timeit -s 'from import escape' 'escape(copyright)' loops, best of 5: 2.45 usec per loop One small concern, () converts ' to &#x27; rather than &. These values are functionally equivalent HTML, but I'll mention it as a backwards incompatible change as the literal text has changed"}
{"_id":"q-en-django__django-11278-dde22f51f97caf92316a87d793d3138bde3d61e32e0c461da9b9cfba980609b0","text":"Add a system check for uniqueness of partial indexes and constraints names. Description (last modified by Mariusz Felisiak) name is a mandatory field for constraints (check and unique) and partial indexes that must be unique in the database scope. We should add a system check for uniqueness of names. Based on discussion in ."}
{"_id":"q-en-django__django-11279-ad6149273e3eff1e5cbfb6ff7f4fa4d476e99191f8f57e5f8d7399e61abf430c","text":"Allow applabel and class to be specified in the name argument for indexes and constraints. Description Allow %(applabel)s and %(class)s to be specified in name argument for BaseConstraint (CheckConstraint, UniqueConstraint) and Index: %(class)s should be replaced by the lowercased name of the child class that the field is used in, '%(app_label)s should be replaced by the lowercased name of the app the child class is contained within. This should allow for specifying check constraints and partial indexes (which name's are mandatory) in abstract models. Based on discussion in ."}
{"_id":"q-en-django__django-11283-e4fe44204b63c3fee37d0180fd9fe7bf150179c11ea3ce795a927a02c1d3d8ec","text":"Migration auth.0011updateproxypermissions fails for models recreated as a proxy. Description (last modified by Mariusz Felisiak) I am trying to update my project to Django 2.2. When I launch python migrate, I get this error message when migration auth.0011updateproxypermissions is applying (full stacktrace is available ​here): duplicate key value violates unique constraint \"idx18141authpermissioncontenttypeid01ab375auniq\" DETAIL: Key (co.ntenttypeid, codename)=(12, addagency) already exists. It looks like the migration is trying to re-create already existing entries in the authpermission table. At first I though it cloud because we recently renamed a model. But after digging and deleting the entries associated with the renamed model from our database in the auth_permission table, the problem still occurs with other proxy models. I tried to update directly from 2.0.13 and 2.1.8. The issues appeared each time. I also deleted my venv and recreated it without an effect. I searched for a ticket about this on the bug tracker but found nothing. I also posted this on ​django-users and was asked to report this here."}
{"_id":"q-en-django__django-11292-6691f695934ac27266013b59cb117c29653d1eb8567163be7d51b6d5f5fd6da7","text":"Add --skip-checks option to management commands. Description Management commands already have skip_checks stealth option. I propose exposing this option on the command line. This would allow users to skip checks when running a command from the command line. Sometimes in a development environment, it is nice to move ahead with a task at hand rather than getting side tracked fixing a system check."}
{"_id":"q-en-django__django-11294-64ad52fe0fdf8342b847c51fcb7b78afe7c2ff05a7e6c4153e5676c44aad586c","text":"Pluralize filter sometimes returns singular form instead of an empty string for invalid inputs Description Filters are documented to return either their input unchanged, or the empty string, whatever makes most sense, when they're used incorrectly. The pluralize filter returns the empty string in such cases, for instance when it receives more than two forms (singular, plural). However, it returns the singular form instead of the empty string when it's passed an object that isn't a number, a string or a list. Failing test case: --- +++ -622,6 +622,9 class DefaultFiltersTests(TestCase): self.assertEqual(pluralize(2,'y,ies'), 'ies') self.assertEqual(pluralize(0,'y,ies,error'), '') def testpluralizeerror(self): self.assertEqual(pluralize(object, 'y,ies'), '') def testphone2numeric(self): self.assertEqual(phone2numericfilter('0800 flowers'), '0800 ') I understand that the implementation is crafted to avoid isinstance checks, but in this case we really want different logic depending on the type of the input. I think the filter should be rewritten with the following pseudo-code: if the value is a number: return singular if value is 1 else plural if the value is a string: return singular if value is '1' else plural if the value has a length (needs a try\/except TypeError): return singular if length is 1 else plural return '' I discovered this while working on ."}
{"_id":"q-en-django__django-11298-58d7e815a380ee11ce269e425d32a3877a83731542afef1d3105fe87a443ce73","text":"Allow ManyToManyField using a intermediary table to be defined as symmetrical. Description Thanks to the work made by Collin Anderson in I think we can remove the check \"fields.E332 Many-to-many fields with intermediate tables must not be symmetrical.\" with a little adjustment. This change was discussed in the django-dev mailing list ​!topic\/django-developers\/BuT0-Uq8pyc. This would let have class Person(models.Model): name = models.CharField(maxlength=20) friends = models.ManyToManyField('self', through='Friendship') class Friendship(models.Model): first = models.ForeignKey(Person, models.CASCADE, relatedname=\"+\") second = models.ForeignKey(Person, models.CASCADE) friendshipdate = models.DateTimeField() and just do something like (anna, throughdefaults={'friendshipdate': date.datetime(...)}) where currently we would have to do (anna, throughdefaults={'friendshipdate': date.datetime(...)}) (joe, throughdefaults={'friendship_date': date.datetime(...)})"}
{"_id":"q-en-django__django-11299-196e00529929b94a05f72887b6ce9a3dd29768d1ceb1724ed428b58e6042e3c3","text":"CheckConstraint with OR operator generates incorrect SQL on SQLite and Oracle. Description (last modified by Michael Spallino) Django is incorrectly including the fully qualified field name(e.g. “mytable”.”myfield”) in part of the check constraint. This only appears to happen when there is a combination of OR and AND clauses in the CheckConstraint. Including the fully qualified field name fails the migration because when we drop the old table and swap the name of the staging table in place, the constraint fails with a malformed schema exception (on sqlite) saying that the field doesn’t exist on the table. It appears that this has to do with the AND clause items using Col while the OR clause uses SimpleCol. Here is an example of this behavior: class TestConstraint(models.Model): field1 = models.IntegerField(blank=True, null=True) flag = models.BooleanField(blank=False, null=False) class Meta: constraints = [ models.CheckConstraint(check=models.Q(flagexact=True, field1isnull=False) | models.Q(flagexact=False,), name='field1hasvalueifflagset'), ] class Migration(migrations.Migration): dependencies = [ ('app', '0001initial'), ] operations = [ migrations.CreateModel( name='TestConstraint', fields=[ ('id', models.AutoField(autocreated=True, primarykey=True, serialize=False, verbosename='ID')), ('field1', models.IntegerField(blank=True, null=True)), ('flag', models.BooleanField()), ], ), migrations.AddConstraint( modelname='testconstraint', constraint=models.CheckConstraint(check=models.Q(models.Q(('field1isnull', False), ('flagexact', True)), ('flagexact', False), connector='OR'), name='field1hasvalueifflagset'), ), ] This is the sql that the migration is going to try and execute: BEGIN; -- Create model TestConstraint CREATE TABLE \"apptestconstraint\" (\"id\" integer NOT NULL PRIMARY KEY AUTOINCREMENT, \"field1\" integer NULL, \"flag\" bool NOT NULL); -- Create constraint field1hasvalueifflagset on model testconstraint CREATE TABLE \"newapptestconstraint\" (\"id\" integer NOT NULL PRIMARY KEY AUTOINCREMENT, \"field1\" integer NULL, \"flag\" bool NOT NULL, CONSTRAINT \"field1hasvalueifflagset\" CHECK (((\"newapptestconstraint\".\"field1\" IS NOT NULL AND \"newapptestconstraint\".\"flag\" = 1) OR \"flag\" = 0))); INSERT INTO \"newapptestconstraint\" (\"id\", \"field1\", \"flag\") SELECT \"id\", \"field1\", \"flag\" FROM \"apptestconstraint\"; DROP TABLE \"apptestconstraint\"; ALTER TABLE \"newapptestconstraint\" RENAME TO \"apptestconstraint\"; COMMIT; The ALTER TABLE fails with the following: malformed database schema (apptestconstraint) - no such column: newapptestconstraint.field1. The proper CREATE TABLE query should look like this: CREATE TABLE \"newapptestconstraint\" (\"id\" integer NOT NULL PRIMARY KEY AUTOINCREMENT, \"field1\" integer NULL, \"flag\" bool NOT NULL, CONSTRAINT \"field1hasvalueifflagset\" CHECK (((\"field1\" IS NOT NULL AND \"flag\" = 1) OR \"flag\" = 0)));"}
{"_id":"q-en-django__django-11323-7ad7ccd0b3b04fd0863be3dcdf899f4a238880adc057b44785fafeb984a6b302","text":"Required SelectDateWidget renders invalid HTML Description SelectDateWidget in required field renders an invalid HTML. According to standard ​ every select with required attribute must have a placeholder option, i.e. first option must have an empty string as a value. That is not a case of SelectDateWidget. Example from django import forms class FooForm(forms.Form): adate = forms.DateField(widget=forms.SelectDateWidget) form = FooForm() str(form) # ...<select name=\"adatemonth\" required id=\"idadatemonth\"<option value=\"1\"January<\/option..."}
{"_id":"q-en-django__django-11333-3181786cd389b26c9edb05198d9bc445bf6247682441f61b463aafa4316f51e8","text":"Optimization: Multiple URLResolvers may be unintentionally be constructed by calls to Description Multiple URLResolvers may be constructed by if has not yet been called, resulting in multiple expensive calls to URLResolver.populate. ​ constructs a new URLResolver, and caches it using functools.lrucache. URLResolver instances can pre-compute a large amount of information about routes in ​, and they store those caches as instance variables. ​ is called with when ​we first handle a request in . getresolver has a number of call-sites. Most notably, ​. Like the other call-sites, reverse calls getresolver with the result of geturlconf. If reverse (or anything else using getresolver) is called both before (e.g. at import time) and after a request is handled, getresolver will be called with different values. Initially it will be called with None, and later if will be called with settings.ROOTURLCONF, because request handling calls seturlconf. In an application with a large number of routes, URLResolver.populate can be expensive, so calling it twice and storing those caches twice is wasteful. My proposed solution is just to modify ​ to look up settings.ROOT_URLCONF before the memoized function call. I'm planning to contribute a fix, as soon as I can get the CLA signed."}
{"_id":"q-en-django__django-11334-ecd1223abda3b150abf7dc39b6543f5da4710dba4e9aa5cfa981c65ab15314f5","text":"Django's template library tags cant use already decorated things like lrucache because of getfullargspec Description Django's template library tags cant use already decorated things like lrucache because of getfullargspec. I have a tag that requires to be lrucached but i cant use it without an helper. The above exception was the direct cause of the following exception: Traceback (most recent call last): File \"\", line 917, in bootstrapinner () File \"\", line 865, in run (, ) File \"\", line 54, in wrapper fn(args, kwargs) File \"\", line 117, in innerrun (displaynumerrors=True) File \"\", line 390, in check includedeploymentchecks=includedeploymentchecks, File \"\", line 377, in runchecks return checks.runchecks(**kwargs) File \"\", line 72, in runchecks newerrors = check(appconfigs=appconfigs) File \"\", line 79, in checkdependencies for engine in (): File \"\", line 90, in all return [self[alias] for alias in self] File \"\", line 90, in <listcompreturn [self[alias] for alias in self] File \"\", line 81, in getitem engine = enginecls(params) File \"\", line 25, in init options['libraries'] = self.gettemplatetaglibraries(libraries) File \"\", line 43, in gettemplatetaglibraries libraries = getinstalledlibraries() File \"\", line 108, in getinstalledlibraries for name in getpackagelibraries(pkg): File \"\", line 121, in getpackagelibraries module = importmodule(entry[1]) File \"\", line 127, in importmodule return bootstrap.gcdimport(name[level:], package, level) File \"<frozen importlib.bootstrap\", line 1006, in gcdimport File \"<frozen importlib.bootstrap\", line 983, in findandload File \"<frozen importlib.bootstrap\", line 967, in findandloadunlocked File \"<frozen importlib.bootstrap\", line 677, in loadunlocked File \"<frozen importlib.bootstrapexternal\", line 728, in execmodule File \"<frozen importlib.bootstrap\", line 219, in callwithframesremoved File \"\", line 25, in <module File \"\", line 132, in simpletag return dec(func) File \"\", line 109, in dec params, varargs, varkw, defaults, kwonly, kwonlydefaults, = getfullargspec(func) File \"\", line 1132, in getfullargspec raise TypeError('unsupported callable') from ex TypeError: unsupported callable It can be solved with unwrapping the function like did."}
{"_id":"q-en-django__django-11354-808522268331101c5b78b8fe765ae0b7a2e9dd6c9890ddca36363e78da14c3d0","text":"() does not with work raw sql annotations on inherited model fields Description Consider these models class BaseItem(models.Model): title = models.CharField(maxlength=32) class Item(BaseItem): pass If I use a RawSQL annotation of Item's queryset that includes one of the fields defined in BaseItem and call .count() on annotated queryset, it fails with the error: (1054, \"Unknown column 'title' in 'field list'\") (MySQL 5.7) code to reproduce the bug with given models: queryset = () queryset = queryset.annotate(title2=RawSQL(\"title\", ())) () # crashes I have tracked down what causes this bug. Query.getaggregation method drops INNER JOIN required to select the title field. Specifically, this code drops it: if not innerquery.distinct: # If the inner query uses default select and it has some # aggregate annotations, then we must make sure the inner # query is grouped by the main model's primary key. However, # clearing the select clause can alter results if distinct is # used. if innerquery.defaultcols and hasexistingannotations: innerquery.groupby = [(innerquery.getinitialalias())] innerquery.defaultcols = False Code is taken from Django 1.8 but 1.11 looks the same. defaultcols is set to False and the INNER JOIN is dropped. Quick fix is to add a condition for setting defaultcols to False: if not innerquery.distinct: # If the inner query uses default select and it has some # aggregate annotations, then we must make sure the inner # query is grouped by the main model's primary key. However, # clearing the select clause can alter results if distinct is # used. if innerquery.defaultcols and hasexistingannotations: innerquery.groupby = [(innerquery.getinitialalias())] if not hasexistingannotations: innerquery.defaultcols = False I don't know if it could be done in a nicer way. I was able to reproduce this in 1.8.18 and 1.11.0"}
{"_id":"q-en-django__django-11356-8c2aac25797490b8144c901dd660cfb713e5005babbd82636531d463d2f85693","text":"ondelete attribute must be callable. Description If you set ondelete=None as a ForeignKey field parameter you might get the following error: File \"\", line 1823, in getdeletedobjects return getdeletedobjects(objs, request, self.adminsite) File \"\", line 134, in getdeletedobjects collector.collect(objs) File \"\", line 197, in collect return super().collect(objs, sourceattr=sourceattr, **kwargs) File \"\", line 221, in collect field.remotefield.ondelete(self, field, subobjs, ) TypeError: 'NoneType' object is not callable I believe that we could validate the ondelete value to prevent such behaviour. Or at least tell that None is not a valid ondelete value. Refs ​"}
{"_id":"q-en-django__django-11359-aa6c3d8d35775360d354e2fde759fca05f3ce19445dbcc2e714206d9acf3861a","text":"Automatically resolve Value's outputfield for stdlib types. Description Hi, I have a model of AModel. AModel has a SearchVectorField named searchvector. I want to update this vector by indexing a string that is not in any other field. from import Value from import SearchVector (pk=1).update(searchvector=SearchVector(Value(\"a string to be indexed and inserted to searchvector field\"))) This code generates this error: FieldError: Cannot resolve expression type, unknown output_field It seemed to be a bug since I found similar usages in forums.."}
{"_id":"q-en-django__django-11374-4d6b0fcf1fc7d71ddf938209f3ec6b7103c97de02919e5049faf1da43ee827cf","text":"Unexpected behavior for Description The function has been changed to give unexpected result for tuple values (and other iterable objects) in the case when no iterations is expected: (dict(a=('a','b')), doseq=False) 'a=%5B%27a%27%2C+%27b%27%5D' One would expect the same as the standard library version (Note the first and last characters has been replaced by square brackets): (dict(a=('a', 'b')), doseq=False) 'a=%28%27a%27%2C+%27b%27%29' If the value is a list, the result if what one would expect: (dict(a=['a','b']), doseq=False) 'a=%5B%27a%27%2C+%27b%27%5D' (dict(a=['a', 'b']), doseq=False) 'a=%5B%27a%27%2C+%27b%27%5D' Note: This is a problem when one has objects that has a str method defined, returning the value one would want to be in the urlencode result, but the object by coincidence is also iterable."}
{"_id":"q-en-django__django-11377-09a6d67d4e908faca4ec36276d2f5b91b6622f62ced123caef744b5309f81edb","text":"Deprecation message crashes when using a query expression in Model.ordering. Description Since updating to Django 2.2 our test suite fails because the newly introduced deprecation warning which warns about Meta.ordering being ignored from Django 3.1 onwards leads to errors when a query expression is used. Take a model definition like this as an example: class Book name = models.CharField(maxlength=255) class Meta: ordering = [F('name',).asc()] The error happens here: File \"\", line 558, in assql \"', '\".join(self.metaordering) TypeError: sequence item 0: expected str instance, OrderBy found A quick and dirty way around that problem is to join the string representations of all the list items instead of concatenating them directly: ( \"%s QuerySet won't use Meta.ordering in Django 3.1. \" \"Add .orderby('%s') to retain the current query.\" % ( , \"', '\".join([str(f) for f in self.metaordering]) ), ) Unfortunately this doesn't generate real source code compatible with .orderby() because the quotation marks are not correct. Maybe someone else has a clean solution on how to fix this? Book QuerySet won't use Meta.ordering in Django 3.1. Add .orderby('name', 'OrderBy(F(price), descending=False)') to retain the current query. - - Book QuerySet won't use Meta.ordering in Django 3.1. Add .orderby('name', OrderBy(F('price'), descending=False)) to retain the current query. A regression test is available here: ​"}
{"_id":"q-en-django__django-11383-21c98ee7a24aa66e152fd4c7f9293961cb10e66a8e51011e6d3684774f8783a4","text":"Saving parent object after setting on child leads to unexpected data loss Description (last modified by Erwin Junge) When saving a parent object after setting it on a child object and then saving the child object, no error is thrown but the FK relation is saved with a NULL value. Failing testcase: # Create parent and child, save parent, save child, parentid should be set p = Parent() c = Child(parent=p) p.save() c.save() c.refreshfrom_db() self.assertIs(c.parent, p) Patch available: ​"}
{"_id":"q-en-django__django-11389-ca245485d67860bf20f08252c3401825968e0ee6e1d16f5767b368b8bb70bf26","text":"Allow SessionStore's to be easily overridden to make dynamic the session cookie age Description If I want to make dynamic my SESSIONCOOKIEAGE setting based on certain parameters of the session I need to reimplement in my SessionStore subclasses the following methods: getexpiryage getexpirydate just to override the points where settings.SESSIONCOOKIEAGE is used."}
{"_id":"q-en-django__django-11396-5158c20f503a0e141b77712097b995c022872f044795fb42c0656930cfaf6d85","text":"Cannot order query by constant value on PostgreSQL Description (last modified by Sven R. Kunze) MyModel.objects.annotate(mycolumn=Value('asdf')).orderby('mycolumn').valueslist('id') ProgrammingError: non-integer constant in ORDER BY LINE 1: \".\"id\" FROM \"mymodel\" ORDER BY 'asdf' ASC... Does it qualify as a bug this time? ;-)"}
{"_id":"q-en-django__django-11399-c5218c1b200b280f97a1bc5d5c952c66ac32b398927de8a231194ce92611fd89","text":"lazy() class preparation is not being cached correctly. Description Doing self.prepared = True changes the instance, but the intention is to change the class variable: ​ This makes functions like gettextlazy, formatlazy and reverselazy a lot slower than they ought to be. Regressed in Django 1.8 (). Using this micro-benchmark on Python 3.7: import cProfile from import lazy def identity(x): return x lazyidentity = lazy(identity, int) (\"for i in range(): str(lazyidentity(1))\") Before: function calls in 0.208 seconds Ordered by: standard name ncalls tottime percall cumtime percall filename:lineno(function) 1 0.010 0.010 0.208 0.208 <string:1(<module) 0.001 0.000 0.001 0.000 (identity) 0.005 0.000 0.010 0.000 (str) 0.004 0.000 0.188 0.000 (wrapper) 0.007 0.000 0.185 0.000 (init) 0.089 0.000 0.178 0.000 (prepareclass) 0.004 0.000 0.005 0.000 (cast) 1 0.000 0.000 0.208 0.208 {built-in method } 0.087 0.000 0.087 0.000 {built-in method builtins.hasattr} 46 0.000 0.000 0.000 0.000 {built-in method builtins.setattr} 1 0.000 0.000 0.000 0.000 {method 'disable' of 'lsprof.Profiler' objects} 0.002 0.000 0.002 0.000 {method 'mro' of 'type' objects} After: function calls in 0.025 seconds Ordered by: standard name ncalls tottime percall cumtime percall filename:lineno(function) 1 0.008 0.008 0.025 0.025 <string:1(<module) 0.001 0.000 0.001 0.000 (identity) 0.005 0.000 0.009 0.000 (str) 0.003 0.000 0.008 0.000 (wrapper) 0.005 0.000 0.005 0.000 (init) 1 0.000 0.000 0.000 0.000 (prepareclass) 0.004 0.000 0.005 0.000 (_cast) 1 0.000 0.000 0.025 0.025 {built-in method } 84 0.000 0.000 0.000 0.000 {built-in method builtins.hasattr} 46 0.000 0.000 0.000 0.000 {built-in method builtins.setattr} 1 0.000 0.000 0.000 0.000 {method 'disable' of 'lsprof.Profiler' objects} 1 0.000 0.000 0.000 0.000 {method 'mro' of 'type' objects}"}
{"_id":"q-en-django__django-11400-321998145ea4c8e6d1a46e94e2207a9937b786fe438ef3f1217c7deb78d736c9","text":"Ordering problem in admin.RelatedFieldListFilter and admin.RelatedOnlyFieldListFilter Description RelatedFieldListFilter doesn't fall back to the ordering defined in Ordering gets set to an empty tuple in ​ and unless ordering is defined on the related model's ModelAdmin class it stays an empty tuple. IMHO it should fall back to the ordering defined in the related model's Meta.ordering field. RelatedOnlyFieldListFilter doesn't order the related model at all, even if ordering is defined on the related model's ModelAdmin class. That's because the call to field.getchoices ​ omits the ordering kwarg entirely."}
{"_id":"q-en-django__django-11405-a3485b6a696b9927142d7dc4c1b7886b8a84da874a78e12effd322ddc65bee8a","text":"Queryset ordering and Meta.ordering are mutable on expressions with reverse(). Description Queryset order and Meta.ordering are mutable with reverse(). Bug revealed by running --reverse (reproduced at ). It seems that test in wasn't correct because order mutates on queryset execution in ​SQLCompiler.getorderby()."}
{"_id":"q-en-django__django-11417-756fc4f22c1430944820895328fef83627fc1e7c0345058167cc83352f542892","text":"Update mail backend to use modern standard library parsing approach. Description uses from the standard lib. On Python 3, () does the same, and is less error-prone."}
{"_id":"q-en-django__django-11422-a0b3a28e0b88dac3f7a33909fb056308865cc14753401d13d6e5ff2f0bff73ff","text":"Autoreloader with StatReloader doesn't track changes in Description (last modified by Mariusz Felisiak) This is a bit convoluted, but here we go. Environment (OSX 10.11): $ python -V Python 3.6.2 $ pip -V pip 19.1.1 $ pip install Django==2.2.1 Steps to reproduce: Run a server python runserver Edit the file, e.g. add print(): def main(): print('sth') os.environ.setdefault('DJANGOSETTINGSMODULE', 'ticket_30479.settings') ... Under 2.1.8 (and prior), this will trigger the auto-reloading mechanism. Under 2.2.1, it won't. As far as I can tell from the log lines, it never sees the itself."}
{"_id":"q-en-django__django-11423-000aff7df44b125e77878b1220ca88e5c5d6bca4cf99027804c1d9712486e907","text":"GenericRelation and prefetchrelated: wrong caching with cyclic prefetching. Description Hello I encountered an issue with GenericRelations. Here is an example to reproduce this issue: ​ Just do a migrate and runserver. The code showing the error is started automatically in Please start with --noreload not to have the output twice. Whats the problem? I have a generic model (Tag) that have a contentobject. Then there are multiple (in the example 2) models, Book and CD, that have exactly one tag assigned. In the real application (OpenSlides) this invariant is ensured in other places; in the example the objects are created in a way, that this invariant holds. All these content objects have a property tag, that should return the one assigned tag. This is done by adding a helper field tags=GenericRelation(Tag) and the property accesses ()[0]. The .all()[0] instead of a simple .get() is required for the prefetching to work. See in the example. Now all tags should be loaded because in OpenSlides they would be serialized. For each tag the contentobject is accessed as well as Because this would result in many DB queries (in the real application about Tags, and in sum content objects) the models are prefetched with: Tag.objects.prefetchrelated(\"contentobject\", \"contentobjecttag\") (Note: The executed code is in ). This results in a constant amount of queries (4 in this case) instead of something proportional to the amount of objects. In the example you can set N, the amount of objects created, to a higher amount to verify, that the amount of queries stays constant. What is expected: If I have a tag tag, should be equal to tag. Output from the example (with N=1): Got 'Tag to book0': -the content object: Book0 -the content objects tag (should be the same as 'Tag to book0'!):Tag to book0 Got 'Tag to cd0': -the content object: CD0 -the content objects tag (should be the same as 'Tag to cd0'!):Tag to book0 This is not the case: 'Tag to cd1' -'cd1' -'Tag to book1'. I tracked this a bit showing, that prefetchedobjectscache holds the wrong value, which is accessed through .all() -.getqueryset() where the cached\/prefetched result is taken. Thanks!"}
{"_id":"q-en-django__django-11428-7c4672c27c5064a0c6e6688425cc21d0d2af5fd25a508d088bc9f2610ee03717","text":"Autoreloader crashes on re-raising exceptions with custom signature. Description (last modified by Alan Trick) How to reproduce: In , put the following code, and update or the settings to have this app config be used. from import AppConfig class MyException(Exception): def init(self, value: str, otherthing: str): super().init(value) = otherthing class Config(AppConfig): name = \"myapp\" verbosename = \"My App\" def ready(self): raise MyException(\"foo\", \"bar\") The problem is that tries to construct a new exception of the same type, with 1 argument (the original exception). The consequence is that you just get a TypeError exception about init() missing 1 required positional argument: 'otherthing' and it completely masks the original exception. Note that this behavior was changed in , it used to just re-raise the exception value. I don't know why it was changed. I noticed this issue as a result of ​"}
{"_id":"q-en-django__django-11433-7934226f85ccd5f88939c9b1bff7db24b692335f0801201963d30dac956a7118","text":"Allow to overwrite fields' default values. Description See comments here: ​ Currently, when submitting a form, if 'somefield' isn't in the data payload (e.g. it wasn't included in the form, perhaps because its value is derived from another field), and 'somefield' has a default value on the model, it cannot be overwritten with 'self.cleaneddata'. This does not really follow the paradigm of modifying data in 'cleaned_data'. It requires the user to copy and overwrite the raw data submitted with the form."}
{"_id":"q-en-django__django-11446-c94d8e4c8b35cae4e8a6d3680e7c1d25e5a72013a5fb3e8a6dfa2d088939024b","text":"Default error webpages are not correctly-formed html pages. Description The default page served for the 404 error in \"DEBUG=False\" mode is (django 2.2.1): <h1Not Found<\/h1<pThe requested resource was not found on this server.<\/pI would expect that by default, a full webpage is sent to the user, thus: <html<body<h1Not Found<\/h1<pThe requested resource was not found on this server.<\/p<\/body<\/htmlIn \"DEBUG=True\" mode, the webpage served is correct html: <!DOCTYPE html<html lang=\"en\"... <\/html"}
{"_id":"q-en-django__django-11451-78167e6ad60e1f85374ced7033aa2fdbe7a347bf00f5475223107d482496f161","text":"ModelBackend.authenticate() shouldn't make a database query when username is None Description It's easier to explain my issue by adding a comment in the current implementation of ModelBackend.authenticate(): def authenticate(self, request, username=None, password=None, **kwargs): if username is None: username = (UserModel.USERNAMEFIELD) # At this point, username and password can be None, # typically if credentials are provided for another backend. # Continuing makes a useless database query and runs # the password hasher needlessly (which is expensive). try: user = UserModel.defaultmanager.getbynaturalkey(username) except UserModel.DoesNotExist: # Run the default password hasher once to reduce the timing # difference between an existing and a nonexistent user (). UserModel().setpassword(password) else: ... My suggestion is to shortcut with: if username is None or password is None: return I noticed this when writing assertNumQueries tests in django-sesame, which provides another authentication backend. I saw this query: sql = SELECT \"authuser\".\"id\", \"authuser\".\"password\", \"authuser\".\"lastlogin\", \"authuser\".\"issuperuser\", \"authuser\".\"username\", \"authuser\".\"firstname\", \"authuser\".\"lastname\", \"authuser\".\"email\", \"authuser\".\"isstaff\", \"authuser\".\"isactive\", \"authuser\".\"datejoined\" FROM \"authuser\" WHERE \"authuser\".\"username\" IS NULL params = () which doesn't make sense: username isn't a nullable field. I thought about timing issues. authenticate() attempts to mask timing differences between existing and non-existing users. I don't think that concern extends to different authentication backends. Since they run different code, they will have timing differences anyway. Currently, in the scenario I'm describing, users are paying the time cost of UserModel().setpassword(password), then of their other authentication backend, so there's a timing difference. With the change I'm proposing, they're only paying the time cost of their other authentication backend."}
{"_id":"q-en-django__django-11457-3d4d12724589c7eaf466619e9f8ae4ccefefd1514db6c376f4139d5654679930","text":"Improve exceptions about mixed types in Expressions. Description The ​source which raises the exception has enough information to say both what types were found, and which of those were unexpected, and probably have a useful repr() In the test suite, the unexpected output types encountered seem to be DurationField and IntegerField, so a more thorough message might be something like: Expression repr(self) contained mixed types: DateField, DurationField. DurationField was unexpected; you must set the outputfield= for this Expression to either DurationField(), DateField() or ... (??? I dunno, some concrete explanation of what the outputfield has to be\/implement if you're not going to use any of the builtins) The merit of including the repr is arguable, as the Expression may not what the user put in (eg: in the test suite it always seems to be a CombinedExpression(lhs, connector, rhs)) but it gives more of a hint in a query which contains multiple expressions (either nested or separate) as to which one is actually causing the problem vs just being told \"something was wrong. Put an output_field= everywhere until it ceases, your guess is as good as mine\"; at the very least the word Expression could be replaced with the class name which is actually raising it."}
{"_id":"q-en-django__django-11477-eebc853ec8a130e765eb929f036afd77f49876727ccfa20aa858c24570c97011","text":"translate_url() creates an incorrect URL when optional named groups are missing in the URL pattern Description There is a problem when translating urls with absent 'optional' arguments (it's seen in test case of the patch)"}
{"_id":"q-en-django__django-11490-cb42b4f11ced28fc5eba4e6bbc074d2ecb214c2064c9cb96c2b62d433bc816b2","text":"Composed queries cannot change the list of columns with values()\/valueslist(). Description Composed queries cannot change the list of columns when values()\/valueslist() is evaluated multiple times, e.g. (name='a', order=2) qs1 = () print((qs1).valueslist('name', 'order').get()) ('a', 2) print((qs1).valueslist('order').get()) ('a', 2) (see ​)."}
{"_id":"q-en-django__django-11501-d79a85bce084ca8f49bff294ea117bdbc5d5d5d6c1f5e9ec92e0eb2f5c211a18","text":"Make createsuperuser inspect environment variables for username and password Description The createsuperuser management command is not quite suitable for scripting, even with the --no-input flag, as it doesn't set a password. The management command should inspect some environment variables (e.g. DJANGOSUPERUSER{()} and DJANGOSUPERUSER_PASSWORD) to pick up the username, password and possibly all other required fields."}
{"_id":"q-en-django__django-11514-11140ebb14caf8938420b2ed5435a415b41ca6e341f8bc7c58c761e2b08d4fbb","text":"Add Cache-Control: private to nevercache decorator. Description If a Django user wants to ensure that a resource is not cached. The user might use nevercache decorator, however, sometimes it doesn't work as he or she expected, which means the resource is cached by CDN. The reason why is that CDN providers cache the resource. For example, Fastly needs to set Cache-Control: private for HTTP header with the origin response. The document is below. ​ Currently Django's never_cache lacks Cache-Control: private, so I suggest to add this header to a response. Thanks,"}
{"_id":"q-en-django__django-11517-4e9dbd541b0709ce6e714ad6e4dceea7849dd23e8eac5201bafc458b8ce30b46","text":"callcommand raises ValueError when subparser dest is passed in options. Description If a management command contains subparsers: class Command(BaseCommand): def addarguments(self, parser): subparsers = parser.addsubparsers(title=\"subcommands\", dest=\"subcommand\", required=True) foo = subparsers.addparser(\"foo\") foo.setdefaults(method=self.onfoocommand) foo.addargument(\"--bar\") In Django, 1.11, this could be called using callcommand('mycommand', 'foo', bar=True) With the additional argument validation in callcommand, this generates ValueError: min() arg is an empty sequence at line 124 in because the SubParsersAction.optionstrings is an empty array. parseargs += [ '{}={}'.format(min(opt.optionstrings), argoptions[]) for opt in parser.actions if opt.required and in options ] If the subcommand parser is not tagged as required, TypeError: Unknown option(s) for mycommand command: bar occurs downstream. The same occurs if the subcommand is passed as an option: call_command('mycommand', subcommand='foo', bar=True)"}
{"_id":"q-en-django__django-11525-3a9516ddcdb6d5ead55195176eb7737bcd5a90bf1a8ae160adbe18d0407611aa","text":"Raise exceptions in mailadmins()\/mailmanagers() when settings are not in expected formats. Description Hi, First time writing a ticket so I apologize if I do anything improperly here. This issue just arose on a project I've been working on, and it goes as follows: Our MANAGERS setting was set like so: MANAGERS = ['', ''] And after calling mail_managers, the result was: smtplib.SMTPRecipientsRefused: {'=?utf-8?q?h?=': (550, b'5.1.1 <=?utf-8?q?h?=: Recipient address rejected: User unknown in local recipient table'), '=?utf-8?q?u?=': (550, b'5.1.1 <=?utf-8?q?u?=: Recipient address rejected: User unknown in local recipient table')} After some investigation it became clear that this setting was in the improper format, but that was only because of ​this StackOverflow post. It would be nice if Django failed early if this setting was detected but improperly set, rather than waiting until the consequences become apparent. Thank you, Kevin"}
{"_id":"q-en-django__django-11527-40c9406c24455777034ed4bdf291f15ca2c557548acd724b2a294885e8d81964","text":"sqlsequencereset should inform that no sequences found. Description This just came up on IRC, because someone was expecting sqlsequencereset to provide resets for the auto-increment values for an SQLite table. Running python sqlsequencereset <myappprovides no output if there are no results returned by (see ​here) So the user has no idea if it failed, succeeded, or they got the invocation wrong (assuming they're not familiar enough with Django to know that invoking it wrong will raise a CommandError). I'd suggest it should avoid ambiguity, so if len(statements) == 0 it should raise CommandError and say there's nothing to do. Bonus points if it also sniffs the connection backend to know if there is anything it could do, and if there's not, report that sqlsequencereset isn't necessary\/available for that backend."}
{"_id":"q-en-django__django-11539-8c43ebdcd0a6172be7324ff292e47dc2817465608a50b8d442a9f7f419b17123","text":"Move index name checks from into system checks. Description (last modified by Mariusz Felisiak) Index names assertions should be moved to system checks to keep code cleaner and more consistent."}
{"_id":"q-en-django__django-11543-6ea7472b78d5a07d3dbad25a2171510c34afdc1991b5639c1a98573a3f9448aa","text":"runserver fails to close connection if --nothreading specified. Description (last modified by Carlton Gibson) Client: Chrome 75.0.3770.100\/Firefox 67.0.4 on macOS 10.14.5. Server: macOS 10.14.5., Python 3.7.3, Django 2.2.3 Running runserver with the --nothreading option may stop responding. This is because Web browser uses multiple connection, and all of them has Connection: keep-alive header by default. When the first request is finished, wsgi server continue to read the socket first request used because the connection is keep-alive. So, the second connection is kept waiting without accepted by wsgi server, until the fist connection is closed. But the first connection will not be closed by browser for very long time."}
{"_id":"q-en-django__django-11550-b24b77ca7cee678486a6151e715f04312a4160f63997cba4cfda456f93e4fc1e","text":"orderby() on union() querysets results with wrong ordering when the same field type is presented multiple times. Description When doing a union of 2 querysets and then doing an orderby on the resulting queryset, if we order on a field whose type is present multiple time, the ordering will be incorrect if the field we sort on is not the 1st field of the query. Explicitly settings values with .values('field1', 'field2') on the base querysets or after the union has no effect. For instance, with this model, with 2 DecimalFields and one BooleanField: class Listing(models.Model): saleprice = models.DecimalField('Sale price', maxdigits=10, decimalplaces=2) yearlyrent = models.DecimalField('Yearly rent', maxdigits=10, decimalplaces=2) toto = models.BooleanField() qs1 = () qs2 = () qs3 = (qs2) print(qs3.orderby('saleprice').query) print(qs3.orderby('yearlyrent').query) print(qs3.orderby('toto').query) From the debugging I did, it seems to come from ​this commit: If I revert def eq back to what it was in Django 2.1 (​) it works as normal again. The difference between the two methods that can explain this is that in Django 2.1, we have a check on the actual field instances thanks to otherargs[1], but in 2.2, because of identity[2][1] which is the class of the field, we can't distinguish two fields of the same type (please refer to the respective implementations to know the values of otherargs and identity). ​Sample projet to reproduce (sqlite db included) (​Model, ​test file). Steps: Install Django 2.2 Run DJANGOSETTINGSMODULE=testunion.settings python test- You will see the queries for qs3.orderby('saleprice') and qs3.orderby('yearly_rent') They are exactly the same whereas they should be different (one with ORDER BY (1) and the other with ORDER BY (2))."}
{"_id":"q-en-django__django-11551-e323a582b4ef89dd63a9d0286666fac9d071b82ddd9333ec6f7e9754b17f06e3","text":"admin.E108 is raised on fields accessible only via instance. Description (last modified by ajcsimons) As part of startup django validates the ModelAdmin's listdisplay list\/tuple for correctness (). Having upgraded django from 2.07 to 2.2.1 I found that a ModelAdmin with a list display that used to pass the checks and work fine in admin now fails validation, preventing django from starting. A PositionField from the django-positions library triggers this bug, explanation why follows. from import models from position.Fields import PositionField class Thing(models.Model) number = models.IntegerField(default=0) order = PositionField() from django.contrib import admin from .models import Thing class ThingAdmin(admin.ModelAdmin) listdisplay = ['number', 'order'] Under 2.2.1 this raises an incorrect admin.E108 message saying \"The value of listdisplay[1] refers to 'order' which is not a callable...\". Under 2.0.7 django starts up successfully. If you change 'number' to 'nonumber' or 'order' to 'noorder' then the validation correctly complains about those. The reason for this bug is commit ​ which was proposed and accepted as a fix for bug The problem is while it fixed that bug it broke the functionality of checklistdisplayitem in other cases. The rationale for that change was that after field=getattr(model, item) field could be None if item was a descriptor returning None, but subsequent code incorrectly interpreted field being None as meaning getattr raised an AttributeError. As this was done after trying field = (item) and that failing that meant the validation error should be returned. However, after the above change if hasattr(model, item) is false then we no longer even try field = (item) before returning an error. The reason hasattr(model, item) is false in the case of a PositionField is its get method throws an exception if called on an instance of the PositionField class on the Thing model class, rather than a Thing instance. For clarity, here are the various logical tests that checklistdisplayitem needs to deal with and the behaviour before the above change, after it, and the correct behaviour (which my suggested patch exhibits). Note this is assuming the first 2 tests callable(item) and hasattr(obj, item) are both false (corresponding to item is an actual function\/lambda rather than string or an attribute of ThingAdmin). hasattr(model, item) returns True or False (which is the same as seeing if getattr(model, item) raises AttributeError) (item) returns a field or raises FieldDoesNotExist Get a field from somewhere, could either be from getattr(model,item) if hasattr was True or from getfield. Is that field an instance of ManyToManyField? Is that field None? (True in case of bug ) hasattr getfield field is None? field ManyToMany? 2.0 returns 2.2 returns Correct behaviour Comments True ok False False [] [] [] - True ok False True E109 E109 E109 - True ok True False E108 [] [] good bit of fix, 2.0 was wrong True raises False False [] [] [] - True raises False True E109 [] E109 Another bug introduced by fix, fails to check if ManyToMany in getfield raise case True raises True False E108 [] [] good bit of fix, 2.0 was wrong False ok False False [] E108 [] bad bit of fix, bug hit with PositionField False ok False True [] E108 E109 both 2.0 and 2.2 wrong False ok True False [] E108 [] bad fix False raises False False E108 E108 E108 - False raises False True E108 E108 E108 impossible condition, we got no field assigned to be a ManyToMany False raises True False E108 E108 E108 impossible condition, we got no field assigned to be None The following code exhibits the correct behaviour in all cases. The key changes are there is no longer a check for hasattr(model, item), as that being false should not prevent us form attempting to get the field via getfield, and only return an E108 in the case both of them fail. If either of those means or procuring it are successful then we need to check if it's a ManyToMany. Whether or not the field is None is irrelevant, and behaviour is contained within the exception catching blocks that should cause it instead of signalled through a variable being set to None which is a source of conflation of different cases. def checklistdisplayitem(self, obj, item, label): if callable(item): return [] elif hasattr(obj, item): return [] else: try: field = (item) except FieldDoesNotExist: try: field = getattr(, item) except AttributeError: return [ checks.Error( \"The value of '%s' refers to '%s', which is not a callable, \" \"an attribute of '%s', or an attribute or method on '%s.%s'.\" % ( label, item, , , , ), , id='admin.E108', ) ] if isinstance(field, models.ManyToManyField): return [ checks.Error( \"The value of '%s' must not be a ManyToManyField.\" % label, , id='admin.E109', ) ] return []"}
{"_id":"q-en-django__django-11555-d8c608d4f9fe672015313fe81915bf4ca6f63a44f9ed4aad89775fd9c8e7ad28","text":"orderby() a parent model crash when Meta.ordering contains expressions. Description (last modified by Jonny Fuller) Hi friends, During testing I discovered a strange bug when using a query expression for ordering during multi-table inheritance. You can find the full write up as well as reproducible test repository ​ The bug occurs because the field is an OrderBy object, not a string, during getorder_dir. The linked stacktrace should make the issue obvious, but what I don't understand is why it only fails during test db setup, not during repl or script use. I wish I could help more and come up with a real solution. Hopefully, this is enough for someone wiser to find the culprit."}
{"_id":"q-en-django__django-11559-d8c608d4f9fe672015313fe81915bf4ca6f63a44f9ed4aad89775fd9c8e7ad28","text":"orderby() a parent model crash when Meta.ordering contains expressions. Description (last modified by Jonny Fuller) Hi friends, During testing I discovered a strange bug when using a query expression for ordering during multi-table inheritance. You can find the full write up as well as reproducible test repository ​ The bug occurs because the field is an OrderBy object, not a string, during getorder_dir. The linked stacktrace should make the issue obvious, but what I don't understand is why it only fails during test db setup, not during repl or script use. I wish I could help more and come up with a real solution. Hopefully, this is enough for someone wiser to find the culprit."}
{"_id":"q-en-django__django-11560-969cd221d32c5751c572011d3028cbecea2ed29c6e7c8f212bc1c73598a6ee11","text":"Raise ValueError in Extract lookups that don't work properly with DurationField. Description Lookups on ExtractYear on a DurationField fails because ExtractYear has an optimisation where it compares the source date with a range of dates. class MyModel(models.Model): duration = models.DurationField() MyModel.objects.annotate(year=ExtractYear('duration')).filter(year__gt=1) The queryset generated looks like SELECT * FROM mymodel WHERE duration '0001-01-01' and it fails because interval are not comparable with dates"}
{"_id":"q-en-django__django-11564-1eddab4e443911e534efa2a10fc782aeb7b3ff9a3e6aedf56a4734c0b74d3900","text":"Add support for SCRIPTNAME in STATICURL and MEDIAURL Description (last modified by Rostyslav Bryzgunov) By default, {% static '...' %} tag just appends STATICURL in the path. When running on sub-path, using SCRIPTNAME WSGI param, it results in incorrect static URL - it doesn't prepend SCRIPTNAME prefix. This problem can be solved with prepending SCRIPTNAME to STATICURL in but that doesn't work when SCRIPTNAME is a dynamic value. This can be easily into default Django static tag and django.contrib.staticfiles tag as following: def render(self, context): url = (context) # Updating url here with request.META['SCRIPTNAME'] if self.varname is None: return url context[self.varname] = url return '' On more research I found that FileSystemStorage and StaticFilesStorage ignores SCRIPT_NAME as well. We might have to do a lot of changes but I think it's worth the efforts."}
{"_id":"q-en-django__django-11583-797a89309e1af52446d56c4ac4b9ada131c923b5f9596a7404ce76d09aef9bef","text":"Auto-reloading with StatReloader very intermittently throws \"ValueError: embedded null byte\". Description Raising this mainly so that it's tracked, as I have no idea how to reproduce it, nor why it's happening. It ultimately looks like a problem with Pathlib, which wasn't used prior to 2.2. Stacktrace: Traceback (most recent call last): File \"\" ... executefromcommandline() File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 381, in executefromcommandline utility.execute() File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 375, in execute self.fetchcommand(subcommand).runfromargv() File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 323, in runfromargv self.execute(args, cmdoptions) File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 60, in execute super().execute(args, options) File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 364, in execute output = (args, options) File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 95, in handle (options) File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 102, in run autoreload.runwithreloader(self.innerrun, options) File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 577, in runwithreloader startdjango(reloader, mainfunc, args, kwargs) File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 562, in startdjango (djangomainthread) File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 280, in run self.runloop() File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 286, in runloop next(ticker) File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 326, in tick for filepath, mtime in self.snapshotfiles(): File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 342, in snapshotfiles for file in self.watchedfiles(): File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 241, in watchedfiles yield from iterallpythonmodulefiles() File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 103, in iterallpythonmodulefiles return itermodulesandfiles(modules, frozenset(errorfiles)) File \"\/Userz\/kez\/path\/to\/venv\/lib\/python3.6\/site-\", line 132, in itermodulesandfiles (path.resolve().absolute()) File \"\", line 1120, in resolve s = self.flavour.resolve(self, strict=strict) File \"\", line 346, in resolve return resolve(base, str(path)) or sep File \"\", line 330, in resolve target = accessor.readlink(newpath) File \"\", line 441, in readlink return os.readlink(path) ValueError: embedded null byte I did print(path) before os.readlink(path) in pathlib and ended up with: \/Users\/kez \/Users It always seems to be \/Users which is last It may have already printed \/Users as part of another .resolve() multiple times (that is, the order is not deterministic, and it may have traversed beyond \/Users successfully many times during startup. I don't know where to begin looking for the rogue null byte, nor why it only exists sometimes. Best guess I have is that there's a mountpoint in \/Users to a samba share which may not have been connected to yet? I dunno. I have no idea if it's fixable without removing the use of pathlib (which tbh I think should happen anyway, because it's slow) and reverting to using and friends. I have no idea if it's fixed in a later Python version, but with no easy way to reproduce ... dunno how I'd check. I have no idea if it's something specific to my system (pyenv, OSX 10.11, etc)"}
{"_id":"q-en-django__django-11584-e964cd034ea5223a1089ff3447e71ebafe7601252beb7586fa113df707747e14","text":"[FATAL] FileNotFoundError with runserver command inside Docker container Description Summary Trying to run the development server in a container with volume-mounted source is throwing a FileNotFoundError. I've verified that the issue is consistently reproducible with Django==2.2.3 and not present in Django==2.1.4. Trace INFO changed, reloading. INFO Watching for file changes with StatReloader Performing system checks... Traceback (most recent call last): File \"\", line 21, in <modulemain() File \"\", line 17, in main executefromcommandline() File \"\/usr\/local\/lib\/python3.6\/site-\", line 381, in executefromcommandline utility.execute() File \"\/usr\/local\/lib\/python3.6\/site-\", line 375, in execute self.fetchcommand(subcommand).runfromargv() File \"\/usr\/local\/lib\/python3.6\/site-\", line 323, in runfromargv self.execute(args, cmdoptions) File \"\/usr\/local\/lib\/python3.6\/site-\", line 60, in execute super().execute(args, options) File \"\/usr\/local\/lib\/python3.6\/site-\", line 364, in execute output = (args, options) File \"\/usr\/local\/lib\/python3.6\/site-\", line 95, in handle (options) File \"\/usr\/local\/lib\/python3.6\/site-\", line 102, in run autoreload.runwithreloader(self.innerrun, options) File \"\/usr\/local\/lib\/python3.6\/site-\", line 587, in runwithreloader startdjango(reloader, mainfunc, args, kwargs) File \"\/usr\/local\/lib\/python3.6\/site-\", line 572, in startdjango (djangomainthread) File \"\/usr\/local\/lib\/python3.6\/site-\", line 289, in run (sender=self) File \"\/usr\/local\/lib\/python3.6\/site-\", line 175, in send for receiver in self.livereceivers(sender) File \"\/usr\/local\/lib\/python3.6\/site-\", line 175, in <listcompfor receiver in self.livereceivers(sender) File \"\/usr\/local\/lib\/python3.6\/site-\", line 16, in watchfortranslationchanges absolutepath = path.absolute() File \"\", line 1129, in absolute obj = self.fromparts([()] + , init=False) FileNotFoundError: [Errno 2] No such file or directory Configuration Dockerfile FROM python:3.6.7-alpine3.7 RUN mkdir \/code WORKDIR \/code RUN apk add postgresql-dev libffi-dev build-base musl-dev RUN apk add linux-headers ADD . RUN pip install -r EXPOSE 3031 ADD cs \/code docker-compose version: '3.7' services: db: image: postgres volumes: .\/pgdata:\/var\/lib\/postgresql\/data ports: \"5432:5432\" environment: POSTGRESPASSWORD: postgres POSTGRESUSER: postgres POSTGRESDB: postgres app: build: context: . volumes: .\/cs\/:\/code\/ ports: \"8000:8000\" env_file: .env command: [\"python\", \"\", \"runserver\", \"0.0.0.0:8000\"]"}
{"_id":"q-en-django__django-11591-0b8d7fb7672641fc43f979052684ae5c3c69f5bf96e4421b4e14543c0c7921a6","text":"Raise a descriptive error on unsupported operations following (), intersection(), and difference(). Description The documentation for () says, \"In addition, only LIMIT, OFFSET, and ORDER BY (i.e. slicing and order_by()) are allowed on the resulting QuerySet.\", however, there isn't any strict enforcement about this -- some operations like () might appear to work. See and for confusion about this."}
{"_id":"q-en-django__django-11592-58538233ea41f042ea13621cd93e9ba1bd182d4d3a3e9c806034958f80e77d68","text":"Start passing FileResponse.blocksize to wsgi.filewrapper. Description (last modified by Chris Jerdonek) I noticed that Django's FileResponse class has a blocksize attribute which can be customized by subclassing: ​ but it's not passed to wsgi.filewrapper. Only the filelike object is passed: response = environ (from: ​ )"}
{"_id":"q-en-django__django-11603-03308b43adc164b8b88f4aa93822bcd4b77459b77419958661bffc69913377a4","text":"Add DISTINCT support for Avg and Sum aggregates. Description As an extension of , aggregates should be supported for other general aggregates such as Avg and Sum. Before 2.2, these aggregations just ignored the parameter, but now throw an exception. This change would just involve setting these classes as allowing DISTINCT, and could also be applied to Min and Max (although pointless)."}
{"_id":"q-en-django__django-11605-0ac791d00dc66bd2fd8dd54c6adb7c0c6492224cd1fe03ad4116ab01b84a6c8d","text":"Filter by window expression should raise a descriptive error. Description Django has a check that filter does not contain window expressions. But it is shallow, neither right side of the expression nor combined expressions are checked. class Employee(models.Model): grade = models.IntegerField() Employee.objects.annotate( prevgrade=Window(expression=Lag('grade')) ).filter(prevgrade=F('grade')) Employee.objects.annotate( prevgrade=Window(expression=Lag('grade')) ).filter(grade=F('prevgrade')) Employee.objects.annotate( prevgrade=Window(expression=Lag('grade')), decgrade=F('prevgrade') - Value(1) ).filter(decgrade=F('grade'))"}
{"_id":"q-en-django__django-11612-463709532a8bf48a712632ac48245528122c0215f78c9a65b9b9e0cb6448fa6c","text":"SQLite3 migrations can fail when used quoted dbtable. Description (last modified by Maciej Olko) If model's Meta dbtable is quoted, e.g. '\"tablewithquotedname\"', SQLite3 migration with this table creation with can fail with near \"tablewithquotedname\": syntax error. I suppose following generated query causes the error: CREATE TABLE \"new\"tablewithquotedname\"\" (\"objid\" integer NOT NULL PRIMARY KEY, \"objnum\" varchar(20) NULL, \"countryid\" integer NOT NULL REFERENCES \"countries\" (\"countryid\") DEFERRABLE INITIALLY DEFERRED) To reproduce table with quoted name should have at least one foreign key. Django documentation says it supports quoted names (​). Quoted names can also be used with Django’s other supported database backends; except for Oracle, however, the quotes have no effect. Traceback: Traceback (most recent call last): File \"…\", line 82, in execute return (sql) File \"…\", line 382, in execute return Database.Cursor.execute(self, query) sqlite3.OperationalError: near \"tablewithquotedname\": syntax error The above exception was the direct cause of the following exception: Traceback (most recent call last): File \"\", line 15, in <moduleexecutefromcommandline() … File \"…\", line 234, in handle fakeinitial=fakeinitial, File \"…\", line 117, in migrate state = self.migrateallforwards(state, plan, fullplan, fake=fake, fakeinitial=fakeinitial) File \"…\", line 147, in migrateallforwards state = self.applymigration(state, migration, fake=fake, fakeinitial=fakeinitial) File \"…\", line 245, in applymigration state = (state, schemaeditor) File \"…\", line 124, in apply operation.databaseforwards(self.applabel, schemaeditor, oldstate, projectstate) File \"…\", line 112, in databaseforwards field, File \"…\", line 327, in addfield self.remaketable(model, createfield=field) File \"…\", line 279, in remaketable self.createmodel(newmodel) File \"…\", line 307, in createmodel self.execute(sql, params or None) File \"…\", line 137, in execute cursor.execute(sql, params) File \"…\", line 99, in execute return super().execute(sql, params) File \"…\", line 67, in execute return self.executewithwrappers(sql, params, many=False, executor=self.execute) File \"…\", line 76, in executewithwrappers return executor(sql, params, many, context) File \"…\", line 84, in execute return (sql, params) File \"…\", line 89, in exit raise djexcvalue.withtraceback(traceback) from excvalue File \"…\", line 82, in execute return (sql) File \"…\", line 382, in execute return Database.Cursor.execute(self, query) near \"tablewithquoted_name\": syntax error"}
{"_id":"q-en-django__django-11618-4338e9f0ea05777913b6a85914263a20716a27c8a0f5be0537b582f9724bafba","text":"Cloaking PermissionErrors raised in ManifestFilesMixin.readmanifest(). Description While using the ManifestStaticFilesStorage, I encountered the ValueError shown below. <trimFile \"\/<some venv\/site-\", line 134, in url hashedname = hashednamefunc(*args) File \"\/<some venv\/site-\", line 422, in storedname raise ValueError(\"Missing staticfiles manifest entry for '%s'\" % cleanname) ValueError: Missing staticfiles manifest entry for '' Further debugging showed that staticfilestorage.hashedfiles was empty. This was odd to me because existed on my system in its proper location in STATICROOT. Additionally, the '' existed as a key in the so I was very confused. I did more debugging of the manifest loading process and hit the root problem. Reading the manifest file raised a PermissionError because the user account running the Django app did not have proper permission to read from where the file was stored on the filesystem. Unfortunately, ​ catches any OSError and proceeds by returning no content. This was in a Vagrant virtual machine so I switched to the root user to bypass any permission errors and the manifest loading worked. I confirmed that I was able to load a page without any issue. I'd like to suggest that ManifestFilesMixin.read_manifest either a) not catch OSError at all to let the system fail so a developer can fix it, b) have more granular error handling of different OSError exception subclasses, or c) do some kind of logging to hint to the app developer that there was a problem. I'm making this suggestion because catching this error and proceeding left the app in an unrecoverable state. The ValueError listed at the beginning of the issue was actually a side effect of the deeper underlying PermissionError problem from attempting to read the manifest file. If there is any more info you need from me, please let me know. Thanks!"}
{"_id":"q-en-django__django-11620-fcfbf94e0e30fdf1d016b1f9e4a0e0cfd8252b347a0af2b4a7fbecb4846551a1","text":"When DEBUG is True, raising Http404 in a path converter's topython method does not result in a technical response Description This is the response I get (plain text): A server error occurred. Please contact the administrator. I understand a ValueError should be raised which tells the URL resolver \"this path does not match, try next one\" but Http404 is what came to my mind intuitively and the error message was not very helpful. One could also make a point that raising a Http404 should be valid way to tell the resolver \"this is indeed the right path but the current parameter value does not match anything so stop what you are doing and let the handler return the 404 page (including a helpful error message when DEBUG is True instead of the default 'Django tried these URL patterns')\". This would prove useful for example to implement a path converter that uses getobjector404."}
{"_id":"q-en-django__django-11622-4fd688c52aff66970ef8e156cd636292c549d932761aa386c57cf0054eac035a","text":"Add a helpful exception for invalid values passed to AutoField\/FloatField\/IntegerField. Description (last modified by Nick Pope) When a large model is updated and saved with invalid values, Django produces a traceback deep within the ORM, with no clue which field assignment caused the error. Developers are with: \"TypeError: int() argument must be a string, a bytes-like object or a number, not 'tuple'\" This change displays the field name which makes spotting errors a lot easier."}
{"_id":"q-en-django__django-11630-64c0885899a1f1de7c6e5f2787ed3d67fd57155e76395ef2e38206810f380ead","text":"Django throws error when different apps with different models have the same name table name. Description Error message: tablename: (models.E028) dbtable 'table_name' is used by multiple models: base.ModelName, app2.ModelName. We have a Base app that points to a central database and that has its own tables. We then have multiple Apps that talk to their own databases. Some share the same table names. We have used this setup for a while, but after upgrading to Django 2.2 we're getting an error saying we're not allowed 2 apps, with 2 different models to have the same table names. Is this correct behavior? We've had to roll back to Django 2.0 for now."}
{"_id":"q-en-django__django-11638-109cba24c4d7771681019a8b5b65982a86c7ec35893b876051a7d593c0ac4b64","text":"Improve Exception Message In Test Client and urlencode() when None is passed as data. Description (last modified by Keith Gray) I am upgrading to 2.2.1 from 2.0.5 and my tests are failing due to a change in the test client: ​ It now throws an exception if a None value is provided in data given to a POST. I would like to propose an improvement to the message to display the offending Key and Value that generate the exception. I have a proposed change in my fork on github: ​"}
{"_id":"q-en-django__django-11666-b5fe2f1a24533ad2193114913122af1d51dc8e3648f0e99e2b86a363d1b76202","text":"Allowing patchvaryheaders() caching utility to handle '' value. Description Function \"patchvaryheaders\", simply appends new headers to list. If view code sets Vary header to asterisk, the resulting header (after applying SessionMiddleware and LocaleMiddleware) looks like this: Vary: , Accept-Language, Cookie This is unnecessary and possible violates HTTP spec: The \"Vary\" header field in a response describes what parts of a request message, aside from the method, Host header field, and request target, might influence the origin server's process for selecting and representing this response. The value consists of either a single asterisk (\"\") or a list of header field names (case-insensitive). Vary = \"\" \/ 1#field-name (from ​) I am using Django to implement REST API, so I'd like it to speak robust HTTP, that works with all present and future caching libraries, — even if widely used browsers and Nginx can correctly interpret current form of the header."}
{"_id":"q-en-django__django-11669-98f0c17dabd5dfd44427b8f803019607a61fb098990a8071f24eeb4a62a903a2","text":"Stop TemplateView automatically passing kwargs into the context Description Only TemplateView pushes to the context. ListView does not, I yet have to check others. This is inconsistency and, I think, it should be fixed."}
{"_id":"q-en-django__django-11677-f84442a776ff797ca878c4e04afbd6871b9d02fab5da4022b9e6c8a74fe8c941","text":"Nested OuterRef not looking on the right model for the field. Description (last modified by Aaron Lisman) I ran into this and made a test case for it. It seems similar to the test case above it. I'm not sure what's different about it and causing it to fail. Let me know if the query is just built wrong. diff --git index .. --- +++ -83,6 +83,28 class SimulationRun(models.Model): return \"%s (%s to %s)\" % (self.midpoint, , ) +class Customer(models.Model): name = models.TextField() + +class Item(models.Model): pass + +class Invoice(models.Model): INVOICE = 'invoice' EXPENSE = 'expense' + KINDCHOICES = ( (INVOICE, 'Invoice'), (EXPENSE, 'Expense'), ) + kind = models.CharField(choices=KINDCHOICES, maxlength=255, default=None) owner = models.ForeignKey(Customer, models.CASCADE) items = models.ManyToManyField(Item, relatedname='invoices') + class UUIDPK(models.Model): id = models.UUIDField(primarykey=True, default=uuid.uuid4) diff --git index .. --- +++ -24,7 +24,7 from import Approximate from .models import ( UUID, UUIDPK, Company, Employee, Experiment, Number, Result, SimulationRun, Time, Time, Customer, Item, Invoice ) -536,6 +536,55 class BasicExpressionsTests(TestCase): # This is a contrived example. It exercises the double OuterRef form. self.assertCountEqual(outer, [first, second, third]) def testnestedsubqueryouterref3(self): customer = (name='Test Customer') othercustomer = (name='Ohter Customer') + unexpensedinvoice = (kind=Invoice.INVOICE, owner=customer) unexpenseditem1 = () (unexpenseditem1) + partiallyexpensedinvoice = (kind=Invoice.INVOICE, owner=customer) expense1 = (kind=Invoice.EXPENSE, owner=customer) expenseditem1 = () unexpenseditem2 = () (expenseditem1, unexpenseditem2) (expenseditem1) + expensedinvoice = (kind=Invoice.INVOICE, owner=customer) (kind=Invoice.EXPENSE, owner=customer) expenseditem2 = () (expenseditem2) (expenseditem2) + otherinvoice = (kind=Invoice.INVOICE, owner=othercustomer) (unexpenseditem1) otherexpense = (kind=Invoice.EXPENSE, owner=othercustomer) (unexpenseditem1) + inner = ( kind=Invoice.EXPENSE, owner=OuterRef(OuterRef('owner')), items=OuterRef('id'), ) middle = ( invoices=OuterRef('id'), ).annotate( expensed=Exists(inner), ).filter( expensed=False, ) outer = ( kind=Invoice.INVOICE, owner=customer, ).annotate( unexpensed=Exists(middle), ) + self.assertTrue(().unexpensed) self.assertTrue(().unexpensed) self.assertFalse(().unexpensed) def testnestedsubqueryouterrefwithautofield(self): first = (time='09:00') second = (time='17:00') Cannot resolve keyword 'owner' into field. Choices are: expensed, id, invoices"}
{"_id":"q-en-django__django-11680-87ef1073c6ed29de4cc7cf2ad679f72310460741dc98270530077b19a7e9784d","text":"Remove UPDATE query when saving a new model instance with a primary key that has a default Description (last modified by user0007) Using a model's instance: class Account(models.Model): id = models.UUIDField( primarykey=True, default=uuid.uuid4, editable=False ) title = models.TextField() account = Account() = \"abc\" () \"appaccount\" SET \"title\" = \\'\\', WHERE \"appaccount\".\"id\" = \\'-150e-419f-b493-\\'::uuid', INTO \"appaccount\" (\"title\", \"id\") VALUES (\\'abc\\', \\'-214a-4798-a0fa-\\'::uuid) Using a model's manager method: (title=\"abc\") INTO \"appaccount\" (\"title\", \"id\") VALUES (\\'abc\\', \\'-214a-4798-a0fa-\\'::uuid) Using a model's instance with forceinsert argument: account = Account() = \"abc\" (forceinsert=true) INTO \"appaccount\" (\"title\", \"id\") VALUES (\\'abc\\', \\'-214a-4798-a0fa-\\'::uuid) Related issue?"}
{"_id":"q-en-django__django-11688-7d2bf591612a6bde14833ece600a3bbeeba7fdda80334fd0c0642743de61801f","text":"path converters don't handle spaces well. Description This came up for someone on IRC last week, but I can't see that they raised a ticket about it. Correct: from import routetoregex routetoregex(\"<uuid:test\") ('^(?P<test[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12})', {'test': at 0x1055e8c88}) Also working correctly: from import routetoregex routetoregex(\"<uuid:2\") ImproperlyConfigured: URL route '<uuid:2' uses parameter name '2' which isn't a valid Python identifier. however, constructing a valid looking converter reference apparently hits neither the happy nor the unhappy path, and also I presume passes any system checks in place that might otherwise warn about the sensitivity: from import routetoregex routetoregex(\"<uuid: test\") # note the preceeding space ('^\\<uuid\\:\\ test\\', {}) the regex is invalid the kwargs dictionary is (sort of rightly) empty. the same is true with \"test \" and \"te st\" Unless I'm misunderstanding the code therein, \"test \", \" test\" and \"te st\" should all be hitting the invalid identifier part, and personally I feel like leading\/trailing spaces at least could just be sanitised (stripped) as they're almost certainly accidental or for display purposes. Tested in a shell against master @ ."}
{"_id":"q-en-django__django-11692-61604d4931f1fe4590842aed983af1c892c613a20d34e47da5e50633c66bbf6b","text":"Can't use OuterRef in union Subquery Description When you make a QuerySet using the union method or the | operator, the QuerySet passed into the union method cannot reference OuterRef even when wrapped with Subquery. For example: cls = ( checklistisnull=False, part=OuterRef('id') ).values('checklistcustomer', 'created') ots = ( ownershiptransferisnull=False, part=OuterRef('id') ).values('ownershiptransfershiptargetcontact', 'created') return self.annotate( owner=Subquery((ots).values('owner')[:1]) ) Returns this error: ValueError This queryset contains a reference to an outer query and may only be used in a subquery. I get the same error with this statement: return self.annotate( owner=Subquery((cls | ots).values('owner')[:1]) ) (As an aside, I also get an error when I try to apply an orderby clause.)"}
{"_id":"q-en-django__django-11695-69e81b95a921fc4ec96eab44aac7ed0e6522cb431b8b04a9e2ad63f6ee9bff8c","text":"Rate-limit autocomplete widgets Ajax requests Description (last modified by Federico Jaramillo Martínez) The current implementation of the Ajax autocomplete widget using Select2 in Django triggers a request for every key-press. This creates unnecessary load on servers. This patch rate-limit the requests by adding a delay using the option provided by Select2. ​"}
{"_id":"q-en-django__django-11701-76f064cfbbeac30facfea440b4477bf0b7d6d2a21cc936c4583a00ee85fd0d09","text":"Admin search with a null character crashes with \"A string literal cannot contain NUL (0x00) characters.\" on PostgreSQL Description Input following URL to browser URL field and access. ​http:\/\/localhost\/admin\/auth\/user\/?q=%00 Crash with following Error. Environment: Request Method: GET Request URL: http:\/\/localhost\/admin\/auth\/user\/?q=%00 Django Version: 2.1.4 Python Version: 3.6.7 Installed Applications: ['', 'django.contrib.contenttypes', 'django.contrib.sessions', 'django.contrib.messages', 'django.contrib.staticfiles', 'djangoextensions', 'restframework', 'select2', 'corsheaders', .......] Installed Middleware: ['django.middleware.security.SecurityMiddleware', 'django.contrib.sessions.middleware.SessionMiddleware', 'corsheaders.middleware.CorsMiddleware', '', '', '', 'django.contrib.messages.middleware.MessageMiddleware', 'django.middleware.clickjacking.XFrameOptionsMiddleware'] Traceback: File \"\" in inner response = getresponse(request) File \"\" in getresponse response = self.processexceptionbymiddleware(e, request) File \"\" in getresponse response = wrappedcallback(request, callbackargs, callbackkwargs) File \"\" in wrapper return self.adminsite.adminview(view)(args, kwargs) File \"\" in wrappedview response = viewfunc(request, args, kwargs) File \"\" in wrappedviewfunc response = viewfunc(request, args, kwargs) File \"\" in inner return view(request, args, kwargs) File \"\" in wrapper return boundmethod(args, kwargs) File \"\" in wrappedview response = viewfunc(request, args, *kwargs) File \"\" in changelistview cl = self.getchangelistinstance(request) File \"\" in getchangelistinstance sortableby, File \"\" in init self.getresults(request) File \"\" in getresults resultcount = File \"\" in get res = [] = (instance) File \"\" in count return () File \"\" in count return () File \"\" in getcount number = obj.getaggregation(using, ['count'])['count'] File \"\" in getaggregation result = compiler.executesql(SINGLE) File \"\" in executesql cursor.execute(sql, params) File \"\" in execute return super().execute(sql, params) File \"\" in execute return self.executewithwrappers(sql, params, many=False, executor=self.execute) File \"\" in executewithwrappers return executor(sql, params, many, context) File \"\" in _execute return (sql, params) Exception Type: ValueError at \/admin\/auth\/user\/ Exception Value: A string literal cannot contain NUL (0x00) characters."}
{"_id":"q-en-django__django-11707-4c71dd44713b075bf61d68b6cb92afa95c6657ea377b93b87167e60f8e373427","text":"Pickling a QuerySet evaluates the querysets given to Subquery in annotate. Description I wrote a test case for modeled after the test from bug which is very similar. def testpicklesubqueryquerysetnotevaluated(self): \"\"\" Verifies that querysets passed into Subquery expressions are not evaluated when pickled \"\"\" groups = Group.objects.annotate( hasevent=models.Exists((groupid=models.OuterRef('id'))) ) with self.assertNumQueries(0): (()) The Subquery class (which is the base for Exists) only stores the underlying query object and throws the QuerySet away (as of this ​commit, although I don't think it worked before that). So in theory it shouldn't be pickling the QuerySet. However, the QuerySet is still stored on the instance within the constructorargs attribute by the decorator on the BaseExpression base class. So when the inner query object gets pickled, so does the QuerySet, which attempts to evaluate it. In this case, it gets the error \"ValueError: This queryset contains a reference to an outer query and may only be used in a subquery.\" since the inner queryset is being evaluated independently when called from pickle: it calls QuerySet.getstate(). I'm not sure what the best solution is here. I made a patch that adds the following override to getstate to the SubQuery class, which fixes the problem and passes all other Django tests on my machine. I'll submit a PR shortly, but welcome any better approaches since I'm not sure what else that may effect. class Subquery(Expression): ... def getstate(self): objdict = super().getstate() ('constructorargs', None) return obj_dict"}
{"_id":"q-en-django__django-11727-2051497ee275d3856ce2e5b659989333bce8a1782a345904add52711a2754f57","text":"Allow hiding the \"Save and Add Another\" button with a context variable Description To provide better adjustability, to introduce new context var - showsaveandaddanother. E.g. if I want to hide button \"Save and add another\", I can just modify extra_context - write False to the variable. For other buttons - \"Save\" and \"Save and continue editing\", this already works exactly in this manner."}
{"_id":"q-en-django__django-11728-2a9e3f2625c657ded0bbebfddd114fdc6c39e83f340e69c245340d9f1b35875c","text":"simplifyregexp() doesn't replace trailing groups. Description replacenamedgroups() fails to replace the final named group if the urlpattern passed in is missing a trailing '\/'. For example, with input r'entries\/(?P<pk[^\/.]+)\/relationships\/(?P<relatedfield\\w+)' the \"related_field\" does not get properly replaced. A workaround is to tack on a '\/' at the end and then it works. Code that reproduces this is attached. This function is used downstream in Django REST Framework. See issue ​6888"}
{"_id":"q-en-django__django-11734-7550a49ba3acdaac5f1e005ab844de34121ef80252a490499aa79ee96cdab66b","text":"OuterRef in exclude() or ~Q() uses wrong model. Description The following test ( to tests\/queries\/testqscombinators) fails when trying to exclude results using OuterRef() def testexistsexclude(self): # filter() qs = Number.objects.annotate( foo=Exists( (tagscategoryid=OuterRef('pk')) ) ).filter(foo=True) print(qs) # works # exclude() qs = Number.objects.annotate( foo =Exists( Item.objects.exclude(tagscategoryid=OuterRef('pk')) ) ).filter(foo=True) print(qs) # crashes # filter(~Q()) qs = Number.objects.annotate( foo =Exists( (~Q(tags_categoryid=OuterRef('pk'))) ) ).filter(foo=True) print(qs) # crashes It results in the following error ValueError: This queryset contains a reference to an outer query and may only be used in a subquery"}
{"_id":"q-en-django__django-11740-0ecc9aecb1a65c5a2dadef1383112e0f44f6c072ff27c8ef951ec2050ad29308","text":"Change uuid field to FK does not create dependency Description (last modified by Simon Charette) Hi! I am new in django community, so please help me, because i really dont know is it really \"bug\". I have a django project named \"testproject\" and two apps: testapp1, testapp2. It will be simpler to understand, with this example: class App1(models.Model): id = models.UUIDField(primarykey=True, unique=True, default=uuid.uuid4, editable=False, verbosename=('identifier')) text = models.CharField(maxlength=100, verbosename=('text')) anotherapp = models.UUIDField(null=True, blank=True, verbosename=('another app')) class App2(models.Model): id = models.UUIDField(primarykey=True, unique=True, default=uuid.uuid4, editable=False, verbosename=('identifier')) text = models.CharField(maxlength=100, verbosename=('text')) First model named \"App1\" has UUID field named \"anotherapp\": anotherapp = models.UUIDField(null=True, blank=True, verbosename=('another app')) After some time i change field from UUID to FK, like this: anotherapp = models.ForeignKey(App2, null=True, blank=True, ondelete=models.SETNULL, verbosename=('another app')) And as result i create new migration, but Migration class was unexpected result, because it does not create any \"dependencies\" for App2, because of FK. I think the correct solution will be create dependency for App2. This project use django version 2.2 and postgresql. Attach archive with sources. Project contains small test, after running him, you will get exception like this: ValueError: Related model 'testapp2.App2' cannot be resolved. So is it problem in django or maybe i dont understand something ? Here is my post in django users: ​!searchin\/django-users\/Django3A20uuid20to20does20create$20dependency%7Csort:date\/django-users\/-h9LZxFomLU\/yz-NLi1cDgAJ Regards, Viktor Lomakin"}
{"_id":"q-en-django__django-11742-1a17dc66d57a5c96da52c61eb48ecf291aed1742bd14d0e96bbe17b2009e6260","text":"Add check to ensure maxlength fits longest choice. Description There is currently no check to ensure that Field.maxlength is large enough to fit the longest value in Field.choices. This would be very helpful as often this mistake is not noticed until an attempt is made to save a record with those values that are too long."}
{"_id":"q-en-django__django-11749-0b40cfe3616e99cbcc0bfc55cfcaeda949d5794e14ae152eb2406c2e21d57c0f","text":"callcommand fails when argument of required mutually exclusive group is passed in kwargs. Description This error Error: one of the arguments --shop-id --shop is required is raised when I run callcommand('mycommand', shopid=1) the argument 'shopid' is part of a required mutually exclusive group: shop = parser.addmutuallyexclusivegroup(required=True) shop.addargument('--shop-id', nargs='?', type=int, default=None, dest='shopid') shop.addargument('--shop', nargs='?', type=str, default=None, dest='shopname') However, everything is fine when I call this command in this way: callcommand('mycommand, '--shop-id=1') In django sources I found that only those keyword arguments of callcommand are passed to the parser that are defined as required: parseargs += [ '{}={}'.format(min(opt.optionstrings), argoptions[]) for opt in parser.actions if opt.required and in options ] but in this special case both of them individually are not required, they are actually part of a group that is required. And the code of callcommand does nothing with groups defined in the parser."}
{"_id":"q-en-django__django-11751-a74c7ce38e5edd69901f47b9873acb65f6da4f36165f98f0e02e42bb38fef4a3","text":"Make security headers default. Description (last modified by Adam Johnson) Following my security headers talk at DjangoCon Europe and its related blog post ( ​ ), I'd like to make Django use more of these security headers by default on new projects. They're always harder to roll out on existing projects than to just bake in to the new project template. On current master, running python check --deploy on a fresh project created with startproject yields these warnings: System check identified some issues: WARNINGS: ?: (security.W004) You have not set a value for the SECUREHSTSSECONDS setting. If your entire site is served only over SSL, you may want to consider setting a value and enabling HTTP Strict Transport Security. Be sure to read the documentation first; enabling HSTS carelessly can cause serious, irreversible problems. ?: (security.W006) Your SECURECONTENTTYPENOSNIFF setting is not set to True, so your pages will not be served with an 'X-Content-Type-Options: nosniff' header. You should consider enabling this header to prevent the browser from identifying content types incorrectly. ?: (security.W007) Your SECUREBROWSERXSSFILTER setting is not set to True, so your pages will not be served with an 'X-XSS-Protection: 1; mode=block' header. You should consider enabling this header to activate the browser's XSS filtering and help prevent XSS attacks. ?: (security.W008) Your SECURESSLREDIRECT setting is not set to True. Unless your site should be available over both SSL and non-SSL connections, you may want to either set this setting True or configure a load balancer or reverse-proxy server to redirect all connections to HTTPS. ?: (security.W012) SESSIONCOOKIESECURE is not set to True. Using a secure-only session cookie makes it more difficult for network traffic sniffers to hijack user sessions. ?: (security.W016) You have '' in your MIDDLEWARE, but you have not set CSRFCOOKIESECURE to True. Using a secure-only CSRF cookie makes it more difficult for network traffic sniffers to steal the CSRF token. ?: (security.W018) You should not have DEBUG set to True in deployment. ?: (security.W019) You have 'django.middleware.clickjacking.XFrameOptionsMiddleware' in your MIDDLEWARE, but XFRAMEOPTIONS is not set to 'DENY'. The default is 'SAMEORIGIN', but unless there is a good reason for your site to serve other parts of itself in a frame, you should change it to 'DENY'. ?: (security.W020) ALLOWEDHOSTS must not be empty in deployment. System check identified 9 issues (0 silenced). Three of these come from security headers that we could activate by default in the settings SECURECONTENTTYPENOSNIFF, SECUREBROWSERXSSFILTER, and XFRAME_OPTIONS. I'd like to propose making them default in the startproject settings and even changing their global defaults (through a deprecation period) so they are activated by default."}
{"_id":"q-en-django__django-11754-026c2825d5959e1b8fd4b1eac77d4fb35db7375de9afa37f9dc6a67f02695cec","text":"Allow using ExceptionReporter subclass in Description (last modified by Carlton Gibson) allows using an ExceptionReporter subclass with AdminEmailHandler. Ideally we'd make the similar available for the 500 debug error view. ​Currently the use of is hardcoded. Move this to a parameter Provide an example of using, e.g., functools.partial to configure a subclass when specifying handler500. Updated for comment:5 Add ExceptionReporter to the documentation, explaining the relation between the exception reporter class and the filter it uses, and showing a simple override of gettracebackdata(). Add a DEFAULTEXCEPTIONREPORTER setting to allow specifying this. (At that point we could review deprecating DEFAULTEXCEPTIONREPORTER_FILTER, as was discussed on — but maybe that's unnecessary. Thoughts?)"}
{"_id":"q-en-django__django-11772-b3cbc0700c687dc44d1fd8876b21420eaf628d96a3abb8e4af6c23bc5a10d837","text":"Template Cache \"maketemplatefragmentkey\" function speed up + simplify (also discussing switch to alternate hashes) Description (last modified by Daniel) The maketemplatefragmentkey function in has the following (minor) issues: Using for varyon args, is not needed any more - it was originally to make the unhashed strings safe to send to memcached and similar restricted systems. But since the value is hashed, this is now adding nothing. (See ​). Use of the MD5 hashing function is disallowed on certain (odd) systems, not being FIPS compliant. See (​). Creates a string of all joined varyon args to send to the hashing function, rather than using the hashlib .update() method. Here is a version solving these, switching to SHA256, and speeding up the function quite a bit: ​ And PR: ​ And here's the repo showing performance improvement: ​ Which seems to be faster in every case. The downside of this is that the cache key is now different from before. The tests have been updated to the new values. There are other cache key generating functions used in other places which use MD5 still - if switching to SHA256 it would make sense to me to change those at the same time, meaning only one time invalidating keys on upgrade. Thoughts?"}
{"_id":"q-en-django__django-11790-96fb3178107d4ee2b747aaee98943912775fd4ae962c087fa432a8e7f4cbe5c5","text":"AuthenticationForm's username field doesn't set maxlength HTML attribute. Description AuthenticationForm's username field doesn't render with maxlength HTML attribute anymore. Regression introduced in and . ​!topic\/django-developers\/qnfSqro0DlA ​"}
{"_id":"q-en-django__django-11797-d38389aa2770aa0736d90387fb501fbe1c6f3c758354daf136cc7f9b61f602ff","text":"Filtering on query result overrides GROUP BY of internal query Description from import models a = (email__isnull=True).values('email').annotate(m=Max('id')).values('m') print(a.query) # good print(a[:1].query) # good b = (id=a[:1]) print(b.query) # GROUP BY U0.\"id\" should be GROUP BY U0.\"email\""}
{"_id":"q-en-django__django-11810-f748675753e73bafd7dde3fb265a444220b6733ad26fc2ea79352ceef6a94c15","text":"Chaining selectrelated mutates original QuerySet. Description (last modified by Darren Maki) When creating a new QuerySet from an existing QuerySet that has had 'selectrelated' applied, if you apply another 'selectrelated' to the new QuerySet it will mutate the original QuerySet to also have the extra 'selectrelated'. from import models class ModelA(models.Model): pass class ModelB(models.Model): pass class ModelC(models.Model): modela = models.ForeignKey('foobar.ModelA', ondelete=models.CASCADE) modelb = models.ForeignKey('foobar.ModelB', ondelete=models.CASCADE) query1 = ModelC.objects.selectrelated('modela') print('QUERY 1:', str()) query2 = query1.selectrelated('modelb') print('QUERY 2:', str()) print('QUERY 1:', str()) if str() == str(): print('\\n!!! The two queries are the same !!!\\n') output QUERY 1: SELECT \"foobarmodelc\".\"id\", \"foobarmodelc\".\"modelaid\", \"foobarmodelc\".\"modelbid\", \"foobarmodela\".\"id\" FROM \"foobarmodelc\" INNER JOIN \"foobarmodela\" ON (\"foobarmodelc\".\"modelaid\" = \"foobarmodela\".\"id\") QUERY 2: SELECT \"foobarmodelc\".\"id\", \"foobarmodelc\".\"modelaid\", \"foobarmodelc\".\"modelbid\", \"foobarmodela\".\"id\", \"foobarmodelb\".\"id\" FROM \"foobarmodelc\" INNER JOIN \"foobarmodela\" ON (\"foobarmodelc\".\"modelaid\" = \"foobarmodela\".\"id\") INNER JOIN \"foobarmodelb\" ON (\"foobarmodelc\".\"modelbid\" = \"foobarmodelb\".\"id\") QUERY 1: SELECT \"foobarmodelc\".\"id\", \"foobarmodelc\".\"modelaid\", \"foobarmodelc\".\"modelbid\", \"foobarmodela\".\"id\", \"foobarmodelb\".\"id\" FROM \"foobarmodelc\" INNER JOIN \"foobarmodela\" ON (\"foobarmodelc\".\"modelaid\" = \"foobarmodela\".\"id\") INNER JOIN \"foobarmodelb\" ON (\"foobarmodelc\".\"modelbid\" = \"foobarmodelb\".\"id\") !!! The two queries are the same !!! The expectation is that the original QuerySet is not mutated, and the two queries are different. This behavior also happens with 'prefetchrelated'. Since the QuerySet methods call '()', and state that they return 'a new QuerySet instance', this behavior does not seem correct."}
{"_id":"q-en-django__django-11815-ad72117e8f7e737ffa4c2ba6502d3ad6ac577068e9cc98fef2a8b7b9f4d23a51","text":"Migrations uses value of enum object instead of its name. Description (last modified by oasl) When using Enum object as a default value for a CharField, the generated migration file uses the value of the Enum object instead of the its name. This causes a problem when using Django translation on the value of the Enum object. The problem is that, when the Enum object value get translated to the users language, the old migration files raise an error stating that the Enum does not have the corresponding value. (because the Enum value is translated to another language) Example: Let say we have this code in from enum import Enum from import gettextlazy as from import models class Status(Enum): GOOD = ('Good') # 'Good' will be translated BAD = ('Bad') # 'Bad' will be translated def str(self): return class Item(models.Model): status = models.CharField(default=Status.GOOD, maxlength=128) In the generated migration file, the code will be: ... ('status', models.CharField(default=Status('Good'), maxlength=128)) ... After the translation, 'Good' will be translated to another word and it will not be part of the Status Enum class any more, so the migration file will raise the error on the previous line: ValueError: 'Good' is not a valid Status Shouldn't the code generated by the migration uses the name of the Status Enum 'GOOD', not the value of it, since it is changeable? It should be: ('status', models.CharField(default=Status['GOOD'], max_length=128)) This will be correct regardless of the translated word"}
{"_id":"q-en-django__django-11820-c6cbbda24080bc4ca8b5458c520b0d0e3de38cde08af38de5984713b76fd2cbb","text":"models.E015 is raised when Meta.ordering contains \"pk\" of a related field. Description models.E015 is raised when Meta.ordering contains pk of a related field, e.g.: test_app.SomeModel: (models.E015) 'ordering' refers to the nonexistent field, related field, or lookup 'optionpk'. Regression in ."}
{"_id":"q-en-django__django-11823-69252342df0d65c96aaac16b47c66635f7cbd281646db7d752ae8d6f4531be4c","text":"cachecontrol() \"maxage\" overrides cachepage() \"timeout\" Description If you decorate a view with both cachecontrol(maxage=3600) and cachepage(timeout=), the server side cache uses the maxage value instead of the timeout value. The comments in UpdateCacheMiddleware.processresponse() indicate it's trying to set the timeout by first looking for the max-age header before reverting to the default cachetimeout : Try to get the timeout from the \"max-age\" section of the \"Cache- Control\" header before reverting to using the default cachetimeout length. However, cachepage(timeout=) is explicitly setting the cachetimeout so that is what should be used. In summary, if a user wants the client-side cache to be shorter-lived than the server-side cache, it is currently not possible using these two decorators."}
{"_id":"q-en-django__django-11829-c068cbabf04a62727e8c2a057d4564fdd7a64441298ec345fca2d6d5a317ce69","text":"patchcachecontrol should special case \"no-cache\" Description (last modified by Tim Graham) From my cursory reading of ​, it looks like patchcachecontrol needs to special case \"no-cache\". no-cache If the no-cache directive does not specify a field-name, then a cache MUST NOT use the response to satisfy a subsequent request without successful revalidation with the origin server. This allows an origin server to prevent caching even by caches that have been configured to return stale responses to client requests. If the no-cache directive does specify one or more field-names, then a cache MAY use the response to satisfy a subsequent request, subject to any other restrictions on caching. However, the specified field-name(s) MUST NOT be sent in the response to a subsequent request without successful revalidation with the origin server. This allows an origin server to prevent the re-use of certain header fields in a response, while still allowing caching of the rest of the response. For example, to integrate a site that uses \"Vary: Cookie\" with AWS CloudFront, one must use 'Cache-Control: no-cache=\"Set-Cookie\"' if a response does not vary by cookie. (I've confirmed this with AWS support as of 10\/31\/2014). patchcachecontrol does not treat \"no-cache\" as a list. If you call patchcachecontrol(response, nocache=\"Set-Cookie\") and then patchcachecontrol(response, nocache=\"foo\"), you end up with 'Cache-Control: no-cache=\"foo\"' Also, nocache=True should take precedence over nocache=\"foo\" regardless of the order it is applied. I found Ticket which proposes to add \"no-cache\" to Just wanted to link it here since they are related."}
{"_id":"q-en-django__django-11848-5525b52e586d17d654ae837377ad046f04c47dbb12eba2cd7c921c474b975dad","text":"two digit year check is incorrect Description (last modified by Ad Timmering) RFC 850 does not mention this, but in RFC 7231 (and there's something similar in RFC 2822), there's the following quote: Recipients of a timestamp value in rfc850-date format, which uses a two-digit year, MUST interpret a timestamp that appears to be more than 50 years in the future as representing the most recent year in the past that had the same last two digits. Current logic is hard coded to consider 0-69 to be in 2000-2069, and 70-99 to be 1970-1999, instead of comparing versus the current year."}
{"_id":"q-en-django__django-11880-e2fed0f51a01d59aee2bcb4897610dd0d258b0df7214cc581ff307b3e26bec62","text":"Form Field’s deepcopy does not (deep)copy the error messages. Description The deepcopy method defined for the formfields (​) performs a shallow copy of self and does not include additional treatment for the error_messages dictionary. As a result, all copies of the same field share the same dictionary and any modification of either the dictionary or the error message itself for one formfield is immediately reflected on all other formfiels. This is relevant for Forms and ModelForms that modify the error messages of their fields dynamically: while each instance of the specific form (e.g., ProfileForm) is expected to have a set of fields “sealed” away from other instances of the same ProfileForm (​), in fact all these instances share the same error messages, resulting in incorrectly raised errors. Confirmed for versions of Django going back to 1.11."}
{"_id":"q-en-django__django-11883-98efafbd3cd47ebbd872b3f50c9d5dffac53945008822ed8cac759da2bc2ede4","text":"Make () return whether or not it suceeded. Description It can be quite useful when dealing with complex caching\/locking systems or simply for logging purposes. Memcache clients already returns this value and it should be straigtforward to implement for file, inmemory, and database backend based on the number of returned rows. Redis del operation also returns the number of keys it successfully deleted so it should be implementable if it's eventually as discussed on the mailing list."}
{"_id":"q-en-django__django-11885-8dcfb2ff7ad461997f402dedcc74dde86dfd2b75d895998803433afab69b51df","text":"Combine fast delete queries Description When emulating ON DELETE CASCADE via ondelete=models.CASCADE the deletion.Collector will try to perform fast queries which are DELETE FROM table WHERE IN .... There's a few conditions required for this fast path to be taken but when this happens the collection logic should combine such queries by table to reduce the number of roundtrips to the database. For example, given the following models class Person(models.Model): friends = models.ManyToManyField('self') class User(models.Model): pass class Entry(models.Model): createdby = models.ForeignKey(User) updatedby = models.ForeignKey(User) Issuing a () or () will result in 3 queries of the form DELETE FROM personfriends WHERE fromid = :id DELETE FROM personfriends WHERE toid = :id DELETE FROM person WHERE id = :id DELETE FROM entry WHERE createdbyid = :id DELETE FROM entry WHERE updatedby = :id DELETRE FROM user WHERE id = :id But both queries (or N queries depending on the number of foreign relationships) can be combined into a single one by using OR DELETE FROM personfriends WHERE fromid = :id OR toid = :id DELETE FROM person WHERE id = :id DELETE FROM entry WHERE createdbyid = :id OR updatedby = :id DELETE FROM user WHERE id = :id"}
{"_id":"q-en-django__django-11891-9ef7e51c7407240bec5cde061def365b6ca2e6ef6614b299cb4a5424f906590f","text":"ConditionalGetMiddleware returns 304 if ETag is the same but Last-Modified has changed. Description (last modified by Mariusz Felisiak) ConditionalGetMiddleware in combination with apache x-sendfile (django-sendfile) doesn't work properly. Each response gets a ETag generated based on response.content which is an empty string in the case of a x-sendfile response, so each time the file is accessed, the ETag generated by ConditionalGetMiddleware is the same. Regardless of the changed file\/changed mtime. In getconditionalresponse() the ETag (which is always the same hash of empty string) is checked first and returns a 304 because it ignores Last-Modified time. Django shouldn't return 304 if ETag is the same but Last-Modified has changed. Related with ."}
{"_id":"q-en-django__django-11893-1bbe55ca25e05fd0812b42d0f264d7649ca6406f423e9e7538b14c292d88105a","text":"DateTimeField doesn't accept ISO 8601 formatted date string Description DateTimeField doesn't accept ISO 8601 formatted date string. Differene is that ISO format allows date and time separator to be capital T letter. (Format being YYYY-MM-DDTHH:MM:SS. Django expects to have only space as a date and time separator."}
{"_id":"q-en-django__django-11894-e1c2607e55dbb5be11b15a42a66b74cecf5e38419bbe28975a81326891096fd4","text":"Explicitly SameSite: None cookies. Description When setting cookies (with .setcookie and setsigned_cookie) the default value for the samesite argument is None but the problem here is that Django doesn't do anything with the None value. This behaviour used to be fine because most browsers assumed that a missing samesite attribute on a cookie meant None. But now, Chrome has issued a warning that samesite has to be explicitly set to None else the cookies won't be sent for cross-origin requests. ​"}
{"_id":"q-en-django__django-11903-164d914fc957432c92f8ed870000387db46991e2edf0420e0ce8cb5efdbbbb00","text":"ManagementUtility.fetchcommand prints \"No Django settings specified.\" even if they are. Description fetchcommand(...) currently ​does the following: if ('DJANGOSETTINGSMODULE'): # If is missing due to misconfigured settings, the # following line will retrigger an ImproperlyConfigured exception # (getcommands() swallows the original one) so the user is # informed about it. settings.INSTALLEDAPPS else: (\"No Django settings specified.\\n\") which doesn't account for settings being set via a UserSettingsHolder by doing settings.configure(...) But the parent execute method ​correctly checks if settings.configured: I've not checked deeply, but I don't think the intent or outcome depends specifically on the LazySettings having been configured via a Settings through a named module import, and it would seem that if settings.configured: could\/should apply here too."}
{"_id":"q-en-django__django-11905-554ae854675972ba956d5a883f755adef20051521bf7c279255dd4228d428827","text":"Prevent using isnull lookup with non-boolean value. Description (last modified by Mariusz Felisiak) isnull should not allow for non-boolean values. Using truthy\/falsey doesn't promote INNER JOIN to an OUTER JOIN but works fine for a simple queries. Using non-boolean values is ​undocumented and untested. IMO we should raise an error for non-boolean values to avoid confusion and for consistency."}
{"_id":"q-en-django__django-11910-16582bc2ca946fc7d572b9968c69ac773899e6d6f0f15b244e64152485fcc700","text":"ForeignKey's tofield parameter gets the old field's name when renaming a PrimaryKey. Description Having these two models class ModelA(models.Model): fieldwrong = models.CharField('field1', maxlength=50, primarykey=True) # I'm a Primary key. class ModelB(models.Model): fieldfk = models.ForeignKey(ModelA, blank=True, null=True, ondelete=models.CASCADE) ... migrations applyed ... the ModelA.fieldwrong field has been renamed ... and Django recognizes the \"renaming\" class ModelA(models.Model): fieldfixed = models.CharField('field1', maxlength=50, primarykey=True) # I'm a Primary key. Attempts to tofield parameter. The tofield points to the oldname (fieldtypo) and not to the new one (\"fieldfixed\") class Migration(migrations.Migration): dependencies = [ ('app1', '0001initial'), ] operations = [ migrations.RenameField( modelname='modela', oldname='fieldwrong', newname='fieldfixed', ), migrations.AlterField( modelname='modelb', name='modela', field=models.ForeignKey(blank=True, null=True, , to='app1.ModelB', tofield='field_wrong'), ), ]"}
{"_id":"q-en-django__django-11911-31b7cab9236483236c23f26e8e2510811cde270b68f1f9c0b587b8036d849830","text":"\"migrate --plan\" outputs \"IRREVERSIBLE\" on RunPython operations without docstrings. Description Given a migration like: from import migrations def forward(apps, schemaeditor): pass def reverse(apps, schemaeditor): pass class Migration(migrations.Migration): operations = [ migrations.RunPython(forward, reverse) ] migrate --plan will output: Planned operations: example.0001initial Raw Python operation -IRREVERSIBLE The migration should not be described as \"irreversible\". This error is in the definition of describeoperation in , reproduced below with line numbers from 2.2.6 tag. 343 344 def describeoperation(operation, backwards): 345 \"\"\"Return a string that describes a migration operation for --plan.\"\"\" 346 prefix = '' 347 if hasattr(operation, 'code'): 348 code = operation.reversecode if backwards else 349 action = if code else '' 350 elif hasattr(operation, 'sql'): 351 action = operation.reversesql if backwards else 352 else: 353 action = '' 354 if backwards: 355 prefix = 'Undo ' 356 if action is None: 357 action = 'IRREVERSIBLE' 358 iserror = True 359 else: 360 action = str(action).replace('\\n', '') 361 iserror = False 362 if action: 363 action = ' -' + action 364 truncated = Truncator(action) 365 return prefix + operation.describe() + (40), iserror Line 349 uses the docstring as the output string. Line 356 tests that value and sets action = 'IRREVERSIBLE' on line 357 because the dosctring is None. It would appear that the intention is to use a docstring to describe the operation, if available, and leave it blank otherwise. However, because it tests against code instead of it actually sets action = None resulting in 'IRREVERSIBLE' being displayed. Proposed Solutions below For a localized fix, I believe line 349 should be replaced by if code: action = if else '' else: action = None However, a more holistic view suggests that displaying \"IRREVERSIBLE\" isn't really the correct thing to do. \"IRREVERSIBLE\" is set when iserror is also set to True and seems to be trying to indicate that the migration operation is invalid rather than irreversible. That is, if code\/reversecode is None (line 348) or sql\/reversesql is None (line 351) the migration can't run. Since sql and code are required parameters for their respective Operations, action should only possibly be None in the reverse case, which seems to be what this code is trying to capture and explain. Given that, a better approach would probably make use of the reversible property defined on RunSQL and RunPython operations. This is a little verbose and could probably be pared down, but I think it has the right logic: def describeoperation(operation, backwards): \"\"\"Return a string that describes a migration operation for --plan.\"\"\" prefix = '' action = '' iserror = False if backwards: prefix = 'Undo ' if hasattr(operation, 'reversible') and not operation.reversible: action = 'INVALID' iserror = True elif hasattr(operation, 'reversecode'): action = if else '' elif hasattr(operation, 'reversesql'): action = if else '' else: if hasattr(operation, 'code'): action = if else '' elif hasattr(operation, 'sql'): action = if else '' action = ' -' + str(action).replace('\\n', '') truncated = Truncator(action) return prefix + operation.describe() + (40), is_error"}
{"_id":"q-en-django__django-11916-8824d1f5db2aa81f520508d6895629b6f0820d8652efb5b97b6f148725c52250","text":"Make prefetchrelated faster by lazily creating related querysets Description In one project of mine I will need to prefetch the following things for each \"machine\": computerdataoperatingsystem, identifiers. computerdata is one-to-one to machine, operatingsystem is manytomany, and identifiers are many-to-one. The data is distributed in a way that any one machine have on average one operating system, and a couple of identifiers. Fetching data results in this profile: 1 0.000 0.000 6.835 6.835 (<module) 1 0.000 0.000 6.795 6.795 (executefromcommandline) 1 0.000 0.000 6.795 6.795 (execute) 1 0.000 0.000 6.207 6.207 (runfromargv) 1 0.000 0.000 6.199 6.199 (execute) 1 0.000 0.000 6.072 6.072 (handle) 10\/2 0.016 0.002 6.069 3.034 (fetchall) 6\/1 0.000 0.000 6.043 6.043 (iter) 1 0.000 0.000 5.837 5.837 (prefetchrelatedobjects) 1 0.009 0.009 5.837 5.837 (prefetchrelatedobjects) 3 0.080 0.027 5.819 1.940 (prefetchonelevel) 4640 0.018 0.000 3.917 0.001 (all) 4646 0.014 0.000 3.206 0.001 (filter) 4648 0.037 0.000 3.193 0.001 (filterorexclude) 4648 0.031 0.000 2.661 0.001 (addq) 4648 0.053 0.000 2.401 0.001 (addq) 4648 0.144 0.000 2.284 0.000 (buildfilter) 2320 0.040 0.000 2.076 0.001 (getqueryset) 2320 0.063 0.000 1.823 0.001 (getqueryset) 0.068 0.000 1.052 0.000 (iterator) 1 0.023 0.023 0.993 0.993 (getprefetchqueryset) 9299 0.067 0.000 0.841 0.000 (clone) 4649 0.086 0.000 0.752 0.000 (setupjoins) 9299 0.226 0.000 0.738 0.000 (clone) 4644 0.177 0.000 0.668 0.000 (getlookupconstraint) 1 0.000 0.000 0.577 0.577 (fetchcommand) 0.330 0.000 0.548 0.000 (init) 127\/79 0.007 0.000 0.447 0.006 {import_} 4645 0.012 0.000 0.443 0.000 (using) 0.017 0.000 0.433 0.000 (resultsiter) <SNIP5 0.197 0.039 0.197 0.039 {method 'execute' of '' objects} If I am reading this correctly, the actual data fetching costs 0.2 seconds of the total runtime of 6.8 seconds. (In reality the ratio is 0.2 vs 2 seconds due to overhead of profiling not affecting the execute time but having a big effect on other parts). The big \"failure\" in above profile is the creation of related querysets: 4640 0.018 0.000 3.917 0.001 (all) this takes more than half (approx 57%) of the total runtime. Every cycle here is wasted - we don't ever need the related queryset when using the prefetched results. I see two options here: Allow assigning the results to somewhere else than () (that is, a plain list you can name). This would naturally get rid of the overhead, but then you will need to alter code to explicitly use the named list when that is available. Somehow lazily instantiate the .all() queryset. If prefetch is in effect calling () will not create a queryset, it just creates a proxy which when iterated gives the related objects, but works otherwise like the related queryset (not like manager). I prefer option 2 as this doesn't require any new APIs or changes to user code to take advantage of this feature. However creating a proxy object that works like the related queryset except for iteration, and which doesn't involve actually creating that queryset will likely be an interesting technical challenge (for example it would be nice to ensure isinstance((), QuerySet) == True. Solving this challenge will likely speed up some prefetch_related queries by 50% or so."}
{"_id":"q-en-django__django-11951-05ee7104c12646df269ead7687e6604a64f1f4d50f8c7dcf5ccedce8fcc63dff","text":"bulkcreate batchsize param overrides the compatible batch size calculation Description (last modified by Ahmet Kucuk) At this line: ​ batchsize param overrides compatible batch size calculation. This looks like a bug as bulkupdate properly picks the minimum of two: ​ I suggest using similar batchsize = min(batchsize, maxbatchsize) if batchsize else maxbatchsize logic in bulkcreate as well. I am happy to open a PR for it."}
{"_id":"q-en-django__django-11964-6f77bf3853354675734447f94f5985a95546d336890b952314bec4ce44ad50f3","text":"The value of a TextChoices\/IntegerChoices field has a differing type Description If we create an instance of a model having a CharField or IntegerField with the keyword choices pointing to IntegerChoices or TextChoices, the value returned by the getter of the field will be of the same type as the one created by enum.Enum (enum value). For example, this model: from import models from import gettextlazy as class MyChoice(models.TextChoices): FIRSTCHOICE = \"first\", (\"The first choice, it is\") SECONDCHOICE = \"second\", (\"The second choice, it is\") class MyObject(models.Model): mystrvalue = models.CharField(maxlength=10, choices=MyChoice.choices) Then this test: from import TestCase from import MyObject, MyChoice class EnumTest(TestCase): def setUp(self) -None: self.myobject = (mystrvalue=MyChoice.FIRSTCHOICE) def testcreatedobjectisstr(self): myobject = self.myobject self.assertIsInstance(myobject.mystrvalue, str) self.assertEqual(str(myobject.mystrvalue), \"first\") def testretrievedobjectisstr(self): myobject = () self.assertIsInstance(myobject.mystrvalue, str) self.assertEqual(str(myobject.mystrvalue), \"first\") And then the results: (django30-venv) django30 test Creating test database for alias 'default'... System check identified no issues (0 silenced). F. FAIL: testcreatedobjectisstr () Traceback (most recent call last): File \"\", line 14, in testcreatedobjectisstr self.assertEqual(str(myobject.mystrvalue), \"first\") AssertionError: 'MyChoice.FIRSTCHOICE' != 'first' MyChoice.FIRSTCHOICE first Ran 2 tests in 0.002s FAILED (failures=1) We notice when invoking str(...) we don't actually get the value property of the enum value which can lead to some unexpected issues, especially when communicating to an external API with a freshly created instance that will send MyEnum.MyValue, and the one that was retrieved would send myvalue."}
{"_id":"q-en-django__django-11983-61066a9e8270b558a671a13e17f3c240072f261e48e176405253ae4a6ac44a9b","text":"Admin's datehierarchy excludes 31 october when using timezone with DST in northern hemisphere. Description introduced a subtle bug where it accidentally excludes 31 october in the admin datehierarchy filter after selecting october. The underlying reason is that the generated sql has < 2019-10-31 as opposed to < 2019-11-01 as it should be. This in turn is caused by applying the timezone for 2019-10-01 (with DST in countries in the northern hemisphere) to the date used for 2019-11-01. This causes the date conversion to actually operate on 2019-10-31 23:00 instead of 2019-11-01 00:00. It's a bit hard to explain in words, PR incoming that hopefully explains better in code."}
{"_id":"q-en-django__django-11997-ec07e5425d64f5b2a7c15f6dfcca19b23ae0e8e541dc001481dee4c647e336c5","text":"The floatformat filter sometimes returns \"-0\" instead of \"0\". Description For values between 0 and -0.5, the floatformat filter returns \"-0\" where I would expect it to return \"0\". For example: $ python -m django --version 5 $ shell Python 3.5.3 (default, Sep 27 2018, 17:25:39) [GCC 6.3.0 ] on linux Type \"help\", \"copyright\", \"credits\" or \"license\" for more information. (InteractiveConsole) from django.template.defaultfilters import floatformat floatformat(-0.1, 0) '-0' floatformat(-0.01, 1) '-0.0' floatformat(-0.001, 2) '-0.00' floatformat(-0.4, 0) '-0' If others agree that this is a bug, I'll submit a patch."}
{"_id":"q-en-django__django-11999-0ecff4852ec211ed2c15bba3089efbaa9e48b81fdc653b45ff3a2fdfe24c6caa","text":"Cannot override getFOOdisplay() in Django 2.2+. Description I cannot override the getFIELDdisplay function on models since version 2.2. It works in version 2.1. Example: class FooBar(models.Model): foobar = models.CharField((\"foo\"), choices=[(1, 'foo'), (2, 'bar')]) def str(self): return self.getfoobardisplay() # This returns 'foo' or 'bar' in 2.2, but 'something' in 2.1 def getfoobardisplay(self): return \"something\" What I expect is that I should be able to override this function."}
{"_id":"q-en-django__django-12009-5dd80d0665c6bc467249bf5319a1fddd5c6baa596a81caca0be1a4dbbbeced68","text":"Django installs \/usr\/bin\/django-admin and \/usr\/bin\/django- Description Django (since 1.7) installs \/usr\/bin\/django-admin and \/usr\/bin\/django- Both of them execute (). \/usr\/bin\/django- does it directly, while \/usr\/bin\/django-admin does it through pkgresources module of Setuptools. \/usr\/bin\/django- #!\/usr\/bin\/python3.4 from import management if name == \"main\": management.executefromcommandline() \/usr\/bin\/django-admin: #!\/usr\/bin\/python3.4 requires = 'Django==1.7' import sys from pkgresources import loadentrypoint if name == 'main': ( loadentrypoint('Django==1.7', 'consolescripts', 'django-admin')() ) \/usr\/lib64\/python3.4\/site-packages\/Django-1.7- [consolescripts] django-admin = Installation of \/usr\/bin\/django- is caused by scripts=['django\/bin\/django-'] in Installation of \/usr\/bin\/django-admin is caused by entrypoints={'consolescripts': ['django-admin = ',]} in I think that it would suffice to install only one of these scripts."}
{"_id":"q-en-django__django-12039-b01dc71d23e75d499d12fb99e2ecbfb4339439338c04fbe0533db1d44d710404","text":"Use proper whitespace in CREATE INDEX statements Description (last modified by Hannes Ljungberg) Creating an index through: index = Index( fields=['-name’], name='idx' ) Will generate the valid but not so pretty CREATE INDEX statement: CREATE INDEX \"idx\" ON \"schemaauthor\" (\"name\"DESC) The following would be expected: CREATE INDEX \"idx\" ON \"schemaauthor\" (\"name\" DESC) This was partially fixed for indexes using opclasses in but it introduced a new quirk when opclasses is used without explicit ordering: index = Index( fields=['name’], name='idx' opclasses=['textpatternops’] ) Will result in: CREATE INDEX \"idx\" ON \"schemaauthor\" (“name” textpatternops ) Note the whitespace after textpatternops. When used with a descending order it will look correct. Unfortunately in the fix in it was assumed that the colsuffixes passed to would be empty for ascending order but instead it will contain empty strings and thus causing this bug. See: ​ The expected output would be: CREATE INDEX \"idx\" ON \"schemaauthor\" (“name” textpattern_ops)"}
{"_id":"q-en-django__django-12049-04a88bade2db1adc77b91b006a9be1d1351ab0c14e1e1f33a86be512f0494b43","text":"Applied migration detection may fail when using a case-insensitive collation Description (last modified by Tim Graham) Hello, I'm using this guide ​ for my studies with camelCasing together with Django (yes, I'm still trying to keep the naming convention we have inside our DB, also for the model's names) Now, I'm really new to Django and I don't know if it's intended but this part of code inside ' is doing case sensitive comparison to check if a column is already present in a database columnnames = [ for column in self.connection.introspection.gettabledescription((), table) ] if not in columnnames: return False, projectstate so if my migration file contains something like this migrations.AddField( modelname='city', name='countrycode', field=models.ForeignKey(dbcolumn='countryCode', , to='myDB.country'), and I run python3 migrate --database myDB --fake-initial myfirstapp it fires an error saying that that table already exists (1050, \"Table 'city' already exists\") If I run python3 migrate --database myDB --fake myfirstapp it correctly fakes myfirstapp The myDB collation is case insensitive, while MySql is running with the ' --lower-case-table-names=0' option"}
{"_id":"q-en-django__django-12050-bffe96d16924c612d374e72e1bbbd38fcabce2e0557bb410fe32537c53c132f8","text":"Query.resolvelookupvalue coerces value of type list to tuple Description Changes introduced in cause an input value list to be coerced to tuple breaking exact value queries. This affects ORM field types that are dependent on matching input types such as PickledField. The expected iterable return type should match input iterable type."}
{"_id":"q-en-django__django-12062-52c71a927e56a042094b6e2a8f891bc91a1413ab1abe63b7cd94fabf999ceda6","text":"Allow disabling of all migrations during tests Description As an extension to a setting DATABASE['TEST']['MIGRATE'] = False should disable all migrations on that particular database. This can be done by hooking into () and returning None."}
{"_id":"q-en-django__django-12073-d319d7cb994f4e76d37606fd7fcd740ed0a1e0adf3ad468b3e41d8db56ba1831","text":"Deprecate the barely documented InvalidQuery exception. Description The exception is ​only mentioned once by name in the documentation without reference to its defining module. It's used for the documented usage and ​abused for ​field deferring select related misuse. I suggest we replace the documented usage by raising FieldDoesNotExist instead and the latter undocumented abuse by raising FieldError while providing a deprecation shim to warn on usages of except InvalidQuery."}
{"_id":"q-en-django__django-12091-000536dd84c2d3e729d9929b728709500c973e11bb6a737dbc55bfff3de5460a","text":"Deprecate Description (last modified by Mariusz Felisiak) As discussed on ​this django-developers thread this should be deprecated. It inspects the non-standard header X-Requested-Wiith that is set by jQuery and maybe other frameworks. However jQuery's popularity, especially for making requests, is decreasing thanks to changes such as the new fetch() JS API. Also in the cases this property is used to determine the kind of content to send to a client, it'd be better to inspect the HTTP standard Accept header. For these reasons Flask has deprecated its similar property isxhr."}
{"_id":"q-en-django__django-12113-6d8afc8e690a3282935748bfc37169185f1a98b5fa29d0e4e07ba1b8515857eb","text":"adminviews.testmultidb fails with persistent test SQLite database. Description (last modified by Mariusz Felisiak) I've tried using persistent SQLite databases for the tests (to make use of --keepdb), but at least some test fails with: sqlite3.OperationalError: database is locked This is not an issue when only using TEST[\"NAME\"] with \"default\" (which is good enough in terms of performance). diff --git index .. --- +++ -15,9 +15,15 DATABASES = { 'default': { 'ENGINE': '', 'TEST': { 'NAME': 'testdefault.sqlite3' }, }, 'other': { 'ENGINE': '', 'TEST': { 'NAME': 'testother.sqlite3' }, } } % adminviews.testmultidb -v 3 --keepdb --parallel 1 … Operations to perform: Synchronize unmigrated apps: adminviews, auth, contenttypes, messages, sessions, staticfiles Apply all migrations: admin, sites Running pre-migrate handlers for application contenttypes Running pre-migrate handlers for application auth Running pre-migrate handlers for application sites Running pre-migrate handlers for application sessions Running pre-migrate handlers for application admin Running pre-migrate handlers for application adminviews Synchronizing apps without migrations: Creating tables... Running deferred SQL... Running migrations: No migrations to apply. Running post-migrate handlers for application contenttypes Running post-migrate handlers for application auth Running post-migrate handlers for application sites Running post-migrate handlers for application sessions Running post-migrate handlers for application admin Running post-migrate handlers for application adminviews System check identified no issues (0 silenced). ERROR ERROR: setUpClass (adminviews.testmultidb.MultiDatabaseTests) Traceback (most recent call last): File \"…\", line 84, in execute return (sql, params) File \"…\", line 391, in execute return Database.Cursor.execute(self, query, params) sqlite3.OperationalError: database is locked The above exception was the direct cause of the following exception: Traceback (most recent call last): File \"…\", line 1137, in setUpClass cls.setUpTestData() File \"…\", line 40, in setUpTestData username='admin', password='something', email='', File \"…\", line 158, in createsuperuser return self.createuser(username, email, password, extrafields) File \"…\", line 141, in createuser () File \"…\", line 66, in save super().save(args, kwargs) File \"…\", line 741, in save forceupdate=forceupdate, updatefields=updatefields) File \"…\", line 779, in savebase forceupdate, using, updatefields, File \"…\", line 870, in savetable result = self.doinsert(cls.basemanager, using, fields, updatepk, raw) File \"…\", line 908, in doinsert using=using, raw=raw) File \"…\", line 82, in managermethod return getattr(self.getqueryset(), name)(args, **kwargs) File \"…\", line 1175, in insert return query.getcompiler(using=using).executesql(returnid) File \"…\", line 1321, in executesql cursor.execute(sql, params) File \"…\", line 67, in execute return self.executewithwrappers(sql, params, many=False, executor=self.execute) File \"…\", line 76, in executewithwrappers return executor(sql, params, many, context) File \"…\", line 84, in execute return (sql, params) File \"…\", line 89, in exit raise djexcvalue.withtraceback(traceback) from excvalue File \"…\", line 84, in execute return (sql, params) File \"…\", line 391, in execute return Database.Cursor.execute(self, query, params) database is locked"}
{"_id":"q-en-django__django-12121-ac7741ac9eff6178bbada826a50760997529a65e5aa5785d32a54a8700062953","text":"Feature\/docs: how should url converters decline to match for a named route? Description It is sometimes convenient to have multiple instances of a named route, where the correct one is chosen based on whether the URL converters tourl call succeeds. For example, the attached file has routes like this: path('export\/foo\/<foo:obj', index, name='export'), path('export\/bar\/<bar:obj', index, name='export'), I then wanted to put {% url \"export\" somefooorbar %} in a generic export template and have the correct URL inserted. My first attempt to do this was to raise ValueError in tourl for non-matching values, hoping that it would work the same as topython where the docs specify that \"A ValueError is interpreted as no match.\" That didn't work -- nothing catches the ValueError in tourl -- so I found the workaround demonstrated in the attachment: if tourl returns an empty string (or some string that doesn't match the converter's regex), then the later regex check in reversewithprefix will detect the non-match and everything seems to work out. So this is either a feature request or a documentation check. I'm thinking either: reversewithprefix could be updated so tourl works the same as topython, and a ValueError indicates no match (which I think just means wrapping try: ... except ValueError: continue around the appropriate line), or the docs could be updated to indicate that, in to_url, a converter should return a non-regex-matching string such as the empty string in order to decline a match."}
{"_id":"q-en-django__django-12122-f35fc53f58d72a3016c44f975450d148272e166f1cc02a1c15640fc1d578aa23","text":"template filter |date:\"r\" not valid RFC 2822 formatted when LANGUAGECODE different than english Description Documentation says template filter date with argument 'r' returns a valid RFC 2822 formatted date. But setting a LANGUAGECODE different than english makes the date returned not valid because the day abbreviation is translated into the LANGUAGE_CODE language. Perhaps there should be two arguments for this: one for valid RFC 2822 dates and another one for the actual 'r' argument (RFC 2822 translated)."}
{"_id":"q-en-django__django-12125-bef0697cbc8aa00f523f3af4b279c9c5d77f68e521fbe0ee63958e9115555c16","text":"makemigrations produces incorrect path for inner classes Description When you define a subclass from as an inner class of some other class, and use this field inside a class, then when you run makemigrations, a migrations file is created which refers to the inner class as if it were a top-level class of the module it is in. To reproduce, create the following as your model: class Outer(object): class Inner(models.CharField): pass class A(models.Model): field = Outer.Inner(maxlength=20) After running makemigrations, the generated migrations file contains the following: migrations.CreateModel( name='A', fields=[ ('id', models.AutoField(autocreated=True, primarykey=True, serialize=False, verbosename='ID')), ('field', (maxlength=20)), ], ), Note the , which should have been The real life case involved an EnumField from django-enumfields, defined as an inner class of a Django Model class, similar to this: import enum from enumfields import Enum, EnumField class Thing(models.Model): class State(Enum): on = 'on' off = 'off' state = EnumField(enum=State) This results in the following migrations code: migrations.CreateModel( name='Thing', fields=[ ('id', models.AutoField(autocreated=True, primarykey=True, serialize=False, verbosename='ID')), ('state', (, max_length=10)), ], ), This refers to , instead of to"}
{"_id":"q-en-django__django-12132-b1160739ce75c1fe2b8a5df3e89c100692662c09659fc1c81ed4254a1bb09b80","text":"Add subdomains of localhost to ALLOWEDHOSTS in DEBUG mode Description (last modified by thenewguy) It would minimize configuration for new projects if ALLOWEDHOSTS += .localhost? when DEBUG=True Chrome resolves *.localhost to localhost without modifying any host files or DNS Referencing the project this way makes it easy to test subdomains -static.localhost, uploads.localhost, www.localhost, etc Updated Conversation on developer mailing list resulted in decision to reopen and accept ticket. The conversation turned up that this behavior is spec compliant per ​"}
{"_id":"q-en-django__django-12143-5344614ee3f7eef5b0d24fa60098f5d15cd047b5be08082312d7f424f09a5e7f","text":"Possible data loss in admin changeform view when using regex special characters in formset prefix Description (last modified by Baptiste Mispelon) While browsing the code in [1] (working on an unrelated ticket), I came across that line: pkpattern = re.compile(r'{}-\\d+-{}$'.format(prefix, )) Generating a regex like this using string formatting can cause problems when the arguments contain special regex characters. is probably safe (I'm not 100% sure about this) since it has to follow Python's syntax rules about identifiers. However prefix has no such restrictions [2] and could contain any number of special regex characters. The fix is quite straightforward (use ()) but it's hard to tell if there might be other occurrences of a similar pattern in Django's code. Some quick grepping (using git grep -E '(recompilematch))' -- 'django\/**.py') currently yields about 200 results. I had a superficial glance through the list and didn't spot other instances of the same usage pattern. EDIT I forgot to mention, but this bug is technically a regression (introduced in ). [1] ​ [2] ​"}
{"_id":"q-en-django__django-12148-9e2632676ce8fc4400bf92c13e8f3c2cffed42bb33c7d27c3237e7719dd4f8a5","text":"reverse() and getabsoluteurl() may return different values for same FlatPage Description (last modified by Tim Graham) The FlatPage model implements getabsoluteurl() without using reverse(). The comment suggests, that this handles SCRIPT_NAME issues, but the link in the admin interface does not work, if you are using a prefix for the flatpages urls. The templatetag for resolving a flatpage works just fine."}
{"_id":"q-en-django__django-12153-107ce1192607c883f738ae3f4eaaf0bf39252aec1d0a84c08afe58a8f6f1b2c5","text":"0011updateproxypermissions crashes in multi database environment. Description (last modified by haudoing) The tutorial said that we can omit to set the default database if default doesn't makes sense ​ But the following migration script doesn't work after configuration with empty default database ​ on line 42, it use with (): ( permissionsquery, contenttype=oldcontenttype, ).update(contenttype=newcontent_type) This will brake the migration if default database doesn't set Tracebacks raise ImproperlyConfigured(\"settings.DATABASES is improperly configured. \" settings.DATABASES is improperly configured. Please supply the ENGINE value. Check settings documentation for more details."}
{"_id":"q-en-django__django-12155-29208094d589d99b919eefe1f0ca3497b74e1f8dbd20a8e171ef639eaaade97e","text":"docutils reports an error rendering view docstring when the first line is not empty Description Currently admindoc works correctly only with docstrings where the first line is empty, and all Django docstrings are formatted in this way. However usually the docstring text starts at the first line, e.g.: def test(): \"\"\"test tests something. \"\"\" and this cause an error: Error in \"default-role\" directive: no content permitted. .. default-role:: cmsreference The culprit is this code in trim_docstring: indent = min(len(line) - len(()) for line in lines if ()) The problem is that the indentation of the first line is 0. The solution is to skip the first line: indent = min(len(line) - len(()) for line in lines[1:] if ()) Thanks."}
{"_id":"q-en-django__django-12161-00766f5c5981c18caba3b848bb08fa8e48b6cdee09cddf9f9e048d010b331e3e","text":"Support callable values in throughdefaults. Description Ticket gave us throughdefaults but unlike the defaults argument of getorcreate [1] or the default argument of any model field, it doesn't allow callable values. Callable values are passed through without being evaluated so the exact behavior depends on the fields. With a CharField for example, the repr() of the function will be saved to the database which is most likely not what the user is expecting. I took a look at the original ticket and pull request but couldn't find a mention of this use-case (ctrl+F for the word \"callable\") so it seems that it was an oversight rather than a deliberate design decision. Code-wise, fixing this is actually pretty straightforward and doesn't seem to cause any regression (see attached pull request). [1] ​"}
{"_id":"q-en-django__django-12172-0d9f6437ec777b8fa48bb3c6f77bded9491316cc64d5082d72b4d3798ddb8c1c","text":"Add ability to override \"async unsafe\" checks. Description It's been reported that Jupyter, at least, executes apparently-synchronous code in an async environment (​) and we're going to have people running headlong into this soon. The \"right\" way of wrapping code in synctoasync works, but is an undue burden on notebook authors as it would have to be in every cell, so it's suggested that we add a flag that disables the async-unsafe check. Either a setting or an environment variable could work; I slightly prefer an environment variable (as it's hard to forget about) provided this works well with Jupyter."}
{"_id":"q-en-django__django-12184-efb6fe34e58b249a18daf572117521331a13d7af734d86c72f30ae9cc4f4e746","text":"Optional URL params crash some view functions. Description My use case, running fine with Django until 2.2: URLConf: urlpatterns += [ ... repath(r'^module\/(?P<format(htmlxml))?\/?$', views.modules, name='modules'), ] View: def modules(request, format='html'): ... return render(...) With Django 3.0, this is now producing an error: Traceback (most recent call last): File \"\/l10n\/venv\/lib\/python3.6\/site-\", line 34, in inner response = getresponse(request) File \"\/l10n\/venv\/lib\/python3.6\/site-\", line 115, in getresponse response = self.processexceptionbymiddleware(e, request) File \"\/l10n\/venv\/lib\/python3.6\/site-\", line 113, in getresponse response = wrappedcallback(request, callbackargs, *callbackkwargs) Exception Type: TypeError at \/module\/ Exception Value: modules() takes from 1 to 2 positional arguments but 3 were given"}
{"_id":"q-en-django__django-12185-728b00c53e70821d266b5415ce9e7ff872ea9087ea12265d68c12da435c2573e","text":"Window expression are not allowed in conditional statements used only in the SELECT clause. Description Django raises NotSupportedError when using window expressions in conditional statements used only in the SELECT clause, e.g. Employee.objects.annotate( lag=Window( expression=Lag(expression='salary', offset=1), partitionby=F('department'), orderby=[F('salary').asc(), F('name').asc()], ), ischanged=Case( When(salary=F('lag'), then=Value(False)), default=Value(True), outputfield=BooleanField() ), ) The SQL standard disallows referencing window functions in the WHERE clause but in this case it's only used in the SELECT clause so this should be possible. Thanks utapyngo for the report. Regression in ."}
{"_id":"q-en-django__django-12187-a8b28176b27a4f3623221184421a8b8d60ba1bcad124bb9b2a2d1e235d7799ac","text":"Allow configuration of where to save staticfiles manifest. Description A standard Django deploy has all staticfiles accessible to all users. This is understandable, if undesirable. By itself this is not a huge problem since those on the public Internet don't know the filenames of all of the files a deployment has, and fuskering the entire possible namespace isn't feasible and is also detectable. However, deployments that make use of ManifestStaticFilesStorage will most likely expose a master list of all static files to anyone who wants to look. It's not a huge security risk because you shouldn't be depending on security through obscurity, but there's certainly a leg up given when there's a master list of all files. Due to the way ManifestStaticFilesStorage is setup, the manifest ends up in the directory of publicly served files. If the files are stored locally this can be fixed by blacklisting the file from webserver access and only letting Django itself read the file off the local filesystem. This is the approach I've taken once I discovered the issue - I have a server deployment running Apache serving files on the local filesystem, but have CloudFront in front of that which fetches from Apache if the cache misses. I've since blacklisted the staticfiles manifest and invalidated any cached copies in CloudFront. Here's what I consider the risks of having a publicly exposed staticfiles manifest: Easily find trade secrets in JavaScript files meant to be used only internally by staff users Find hardcoded secrets in internal files - anything in the static tree gets listed here, even pre-processed files like coffee or less if the developers use django-compressor Find potential attack vectors by finding normally unlisted files that are exploitable which could be used to form URLs in phishing emails Possible novel way to fingerprint Django versions using the easy master list of files, could be used to quickly identify potentially vulnerable Django servers All that said, I don't have a great solution to the problem that Django itself could implement. Currently Django writes the manifest to the staticfiles root so it's always going to be readable unless you take extra steps. The real stickler is deployments that use something like ​S3BotoStorage which in effect needs Django to be able to access the manifest remotely. My understanding of that setup (I don't use it) would be that on load Django is going to read the manifest from S3, so it needs to be accessible over the web by default. Further steps could be taken to make it only accessible to Django itself, but that requires user action. Potential solutions: Encrypt the manifest on disk, decrypt on load into memory - loses human readability for debugging purposes but hides it from prying eyes by default Fast-track ticket to make staticfiles storage configuration allow passing options to storage - use options to change manifest path somewhere non-public or configure a secret header to use with S3 to only give Django access to the file. On a related note, this discovery has made me extra paranoid about the exposure of internal files meant for staff only and now I'm looking at a way to formalize restricted access to the files. With the exposure of the staticfiles manifest it's clear much of the business logic we use (in JavaScript under admin) is by default visible to the Web if you know the URL."}
{"_id":"q-en-django__django-12193-512772382634066102679e755854ff69dca687af059bce0fa7654c5ee713ce1a","text":"SplitArrayField with BooleanField always has widgets checked after the first True value. Description (last modified by Peter Andersen) When providing a SplitArrayField BooleanField with preexisting data, the finalattrs dict is updated to include 'checked': True after the for loop has reached the first True value in the initial data array. Once this occurs every widget initialized after that defaults to checked even though the backing data may be False. This is caused by the CheckboxInput widget's getcontext() modifying the attrs dict passed into it. This is the only widget that modifies the attrs dict passed into its get_context(). CheckboxInput setting attrs['checked'] to True: ​"}
{"_id":"q-en-django__django-12196-8671d6eb196eea1e1e9ca76a72df131aa4fb56c8b7f5236c9e6188ec49c91358","text":"Add a safeguard to debug decorators (sensitivevariables\/sensitivepostparameters) to prevent incorrect usage. Description While trying to reproduce ticket:#comment:5, I noticed that Django happily lets you write this kind of code: # incorrect usage, should be def ispasswordok(password): return len(password) 8 It's very easy to miss that you forgot the (). Most of the time it's not really dangerous because the decorated function will be unusable but in this case, the consequences are pretty nasty: bool(ispasswordok('asdf')) True # you would expect False because len('asdf') < 8 I propose adding some code to both sensitivevariables() and sensitivepostparameters() that catches this misuse to prevent users from decorating their functions incorrectly. Because both decorators take either no arguments or only string arguments, it's not too hard to detect the error with something like this: def sensitive_variables(*variables): if len(variables) == 1 and callable(variables[0]): raise TypeError(...) # ... This should be fully backwards compatible and in most cases it will raise the error at import time which should make things easier to fix for those who've incorrectly used the decorator. (I've confirmed with the security team that this does not need to be treated as a security issue)"}
{"_id":"q-en-django__django-12198-67a8b2aa63633900f1fe3aa8324351b95a83d41ca98f5ee0750d652acb32dc09","text":"Allow sensitivevariables() to preserve the signature of its decorated function Description When the method authenticate of a custom AuthenticationBackend is decorated with sensitivevariables, inspect.getcallargs will always match. Calling the authenticate function will attempt to call this backend with any set of credentials and will raise an uncaught TypeError for an unmatching backend. Authentication with such decorated backends used to work in version 1.6."}
{"_id":"q-en-django__django-12209-830227388494ef3d0180593d5c8fdc7b15330d8a5032e2319fd3915892deb3be","text":"Change in behaviour when saving a model instance with an explcit pk value if the pk field has a default Description (last modified by Reupen Shah) Consider the following model: from uuid import uuid4 from import models class Sample(models.Model): id = models.UUIDField(primarykey=True, default=uuid4) name = models.CharField(blank=True, maxlength=100) In Django 2.2 and earlier, the following commands would result in an INSERT followed by an UPDATE: s0 = () s1 = Sample(, name='Test 1') () However, in Django 3.0, this results in two INSERTs (naturally the second one fails). The behaviour also changes if default=uuid4 is removed from the id field. This seems related to The change in behaviour also has the side effect of changing the behaviour of the loaddata management command when the fixture contains explicit pk values and the objects already exist (e.g. when loading the fixture multiple times). Perhaps the intention was to only change the behaviour if an explicit pk value was not set on the model instance being saved? (At least, that would be more backwards-compatible behaviour...)"}
{"_id":"q-en-django__django-12212-2abb3989832e652382a8216edc5b6125896a58c2b1d04e86dff62430cdd64315","text":"DeserializationError local variable 'pk' referenced before assignment (which hides real error) Description The first error is this: Environment: Request Method: GET Request URL: http:\/\/localhost:8000\/admin\/artcollection\/artobject\/2298\/history\/734\/ Django Version: 1.9.6 Python Version: 2.7.10 Installed Applications: ['saskartsboard.artcollection', 'saskartsboard.options', 'saskartsboard.descriptiveandvisualdictionary', 'grappelli', 'versatileimagefield', 'select2', 'ckeditor', '', '', 'django.contrib.contenttypes', 'django.contrib.sessions', 'django.contrib.messages', 'django.contrib.staticfiles', 'reversion'] Installed Middleware: ['django.middleware.security.SecurityMiddleware', 'django.contrib.sessions.middleware.SessionMiddleware', '', '', '', '', 'django.contrib.messages.middleware.MessageMiddleware', 'django.middleware.clickjacking.XFrameOptionsMiddleware'] Traceback: File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in getresponse response = self.processexceptionbymiddleware(e, request) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in getresponse response = wrappedcallback(request, callbackargs, callbackkwargs) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in wrappedview response = viewfunc(request, args, kwargs) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in wrappedviewfunc response = viewfunc(request, args, kwargs) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in inner return view(request, args, kwargs) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in revisionview return self.revisionformview(request, version, self.revisionformtemplate or self.gettemplatelist(\"\"), context) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in revisionformview (delete=True) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in revert saferevert(versionset) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in saferevert () File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in revert () File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in objectversion return list(serializers.deserialize(, data, ignorenonexistent=True))[0] File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in Deserializer six.reraise(DeserializationError, DeserializationError(e), sys.excinfo()[2]) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in Deserializer for obj in PythonDeserializer(objects, options): File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in Deserializer raise base.DeserializationError.WithData(e, d['model'], d.get('pk'), pk) Exception Type: DeserializationError at \/admin\/artcollection\/artobject\/2298\/history\/734\/ Exception Value: local variable 'pk' referenced before assignment Which is caused by referring to the 'pk' variable on line 142, which is only set in the first iteration of the for loop on line 139. If you add a raise before line 142 you can see the real exception which is: Environment: Request Method: GET Request URL: http:\/\/localhost:8000\/admin\/artcollection\/artobject\/2298\/history\/734\/ Django Version: 1.9.6 Python Version: 2.7.10 Installed Applications: ['saskartsboard.artcollection', 'saskartsboard.options', 'saskartsboard.descriptiveandvisualdictionary', 'grappelli', 'versatileimagefield', 'select2', 'ckeditor', '', '', 'django.contrib.contenttypes', 'django.contrib.sessions', 'django.contrib.messages', 'django.contrib.staticfiles', 'reversion'] Installed Middleware: ['django.middleware.security.SecurityMiddleware', 'django.contrib.sessions.middleware.SessionMiddleware', '', '', '', '', 'django.contrib.messages.middleware.MessageMiddleware', 'django.middleware.clickjacking.XFrameOptionsMiddleware'] Traceback: File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in getresponse response = self.processexceptionbymiddleware(e, request) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in getresponse response = wrappedcallback(request, callbackargs, callbackkwargs) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in wrappedview response = viewfunc(request, args, kwargs) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in wrappedviewfunc response = viewfunc(request, args, kwargs) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in inner return view(request, args, kwargs) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in revisionview return self.revisionformview(request, version, self.revisionformtemplate or self.gettemplatelist(\"\"), context) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in revisionformview (delete=True) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in revert saferevert(versionset) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in saferevert () File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in revert () File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in objectversion return list(serializers.deserialize(, data, ignorenonexistent=True))[0] File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in Deserializer six.reraise(DeserializationError, DeserializationError(e), sys.excinfo()[2]) File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in Deserializer for obj in PythonDeserializer(objects, options): File \"\/media\/mausoleum\/Work\/Development\/saskartsboard-collections\/env\/local\/lib\/python2.7\/site-\" in Deserializer for pk in fieldvalue: Exception Type: DeserializationError at \/admin\/artcollection\/artobject\/2298\/history\/734\/ Exception Value: 'NoneType' object is not iterable Which is caused by not dealing with None values. I'm not exactly certain why the None value is set for this field value, but it is. My best guess at this point is that this field used to be a ForeignKey but has been migrated migrated into a ManyToMany and when the object was first saved, the FK was None, but now the ManyToMany is empty. If you believe this is an issue with django reversion, I will happily move it there, but I'm not so sure. Changing line 139 to: for pk in fieldvalue or []: Seems to get around the error."}
{"_id":"q-en-django__django-12225-8a5576b9addd4d699d24ebd2d9fd438cdf60f52ec4fef09c5b5976e0cd6ec2d8","text":"Improve error message for admin.E202. Description If an inline has mutliple foreign keys to the same parent model, you get an error message like so: (admin.E202) 'account.PaymentApplication' has more than one ForeignKey to 'account.Invoice'. This error message should recommend specifying fk_name."}
{"_id":"q-en-django__django-12231-07daf5bf2d29fd042aab2be456fc5357c1b5622a8997d49e0441e8386a84936f","text":"Related Manager set() should prepare values before checking for missing elements. Description To update a complete list of foreignkeys, we use set() method of relatedmanager to get a performance gain and avoid remove and add keys not touched by user. But today i noticed our database removes all foreignkeys and adds them again. After some debugging i found the issue in this condition: ​ Our form returns all Foreignkeys as list of strings in cleaneddata, but the strings do not match the pk (int). After converting the strings to int, before calling set(), fixes the problem. Question: How to avoid this issue? Maybe all code usages of set() are using lists of strings, maybe not. I dont know at the moment. Is is possible Django fixes this issue? Should Django fix the issue? Maybe strings should raise an exception?"}
{"_id":"q-en-django__django-12237-4f9ef9172970014f311f906c8d1f16019bff4654e24eae144776ec4e1bedc957","text":"slugify() doesn't return a valid slug for \"İ\". Description While working on an international project, we discovered that the turkish\/azerbaijani letter İ can not be properly processed when SlugField and slugify are run with allowunicode=True. The project itself runs with Django 2.2.6 and Wagtail 2.6.2. I first talked about this in the Wagtail Support Channel and while researching further, discovered that this is a Django\/Python related issue. This was tested on Python 3.6 and on Python 3.7. (quick shoutout to Matt Wescott of Wagtail Fame for being a sparing partner in this) There is a rather detailed analysis (README) in a git repo I created ​ - it was also the basis for my initial call for help in wagtail's support channel. Meanwhile I have extended it with a Django-only project, as to be a 100% sure this has nothing to do with Wagtail. I was not able to find anything similar in trac. While I encourage whoever is reading this to actually read the README in the git repo, I want to honor your time and will try to provide a more concise version of the bug here. Explanation from import models from import slugify class Page(models.Model): title = models.CharField(maxlength=255) slug = models.SlugField(allowunicode=True) def str(self): return Using this in a shell\/test like a (Model)Form might: from import slugify page = Page(title=\"Hello İstanbul\") = slugify(, allowunicode=True) page.fullclean() fullclean() then raises {'slug': Enter a valid 'slug' consisting of Unicode letters, numbers, underscores, or hyphens.?} Why is that? slugify does the following internally: (r'[^\\w\\s-]', '', value).strip().lower() Thing is, Python's .lower() of the İ in İstanbul looks like this: [(character) for character in 'İ'.lower()] ['LATIN SMALL LETTER I', 'COMBINING DOT ABOVE'] So, while slugify() finishes, the result is then passed into SlugField which uses the slugunicodere (^[-\\w]+\\Z). However, the Combining Dot Above is not a valid \\w: [(character, (character), (character)) for character in 'İ'.lower()] [ ('i', 'LATIN SMALL LETTER I', <re.Match object; span=(0, 1), match='i'), # EUREKA!! ('̇', 'COMBINING DOT ABOVE', None) ] So that's why the ValidationError is raised. Proposed Solution The culprit seems to be the order in which lower() is called in slugify. The assumption that the lowercase version of a string is still a valid slugunicodere, does not seem to hold true. Hence, instead of doing this in slugify() (r'[^\\w\\s-]', '', value).strip().lower() It might be better to do it like this (r'[^\\w\\s-]', '', ()).strip() Is Python the actual culprit? Yeah it might be. Matt ( urged me to also take a look if Python might be doing this wrong. The İ is the Latin Capital Letter I with Dot Above. It's codepoint is U+0130 According to the chart for the Latin Extended-A set (​), it's lowercase version is U+0069. U+0069 lives in the C0 Controls and Basic Latin set (​). Lo and behold: it is the Latin small letter I. So a latin lowercase i. Does this really mean that Python is doing something weird here by adding the Combining dot above? Honestly, I can't imagine that and I am probably missing an important thing here because my view is too naive. I hope this shorter explanation makes sense. If it does not, please try to read through the detailed analysis in the repo (​). If that also does not make a ton of sense, let me know. In any case, thank you for taking the time to read this bug report. Looking forward to feedback and thoughts. I am happy to oblige in any capacity necessary."}
{"_id":"q-en-django__django-12262-64748826777ff1a491153180584a6a74282edde872a57ad7561e8f802f31c755","text":"Custom template tags raise TemplateSyntaxError when keyword-only arguments with defaults are provided. Description (last modified by P-Seebauer) When creating simple tags without variable keyword args, but an keyword argument with a default value. It's not possible to supply any other variable. def hello(, greeting='hello'): return f'{greeting} world' {% hello greeting='hi' %} Raises “'hello' received unexpected keyword argument 'greeting'” Also supplying a keyword argument a second time raises the wrong error message: #tag def hi(, greeting): return f'{greeting} world' {% hi greeting='hi' greeting='hello' %} Raises “'hi' received unexpected keyword argument 'greeting'” instead of \"'hi' received multiple values for keyword argument 'greeting'\" Same goes for inclusion tags (is the same code) I already have a fix ready, will push it after creating the ticket (that I have a ticket# for the commit). Is actually for all versions since the offending line is from 2.0…"}
{"_id":"q-en-django__django-12273-a443c3743c192b6eec223816036b7ffca436f3be2dce93841a13519ff2b41ac9","text":"Resetting primary key for a child model doesn't work. Description In the attached example code setting the primary key to None does not work (so that the existing object is overwritten on save()). The most important code fragments of the bug example: from import models class Item(models.Model): # uid = models.UUIDField(primarykey=True, default=uuid.uuid4, editable=False) uid = models.AutoField(primarykey=True, editable=False) f = models.BooleanField(default=False) def reset(self): = None self.f = False class Derived(Item): pass class SaveTestCase(TestCase): def setUp(self): self.derived = (f=True) # create the first object item = () obj1 = item.derived () () # the first object is overwritten def testftrue(self): obj = () self.assertTrue(obj.f) Django 2.1.2"}
{"_id":"q-en-django__django-12276-1ed172293138e20978a119f4ebee85c817e477bf2530253fd7f54f62049698aa","text":"FileInput shouldn't display required attribute when initial data exists. Description (last modified by thenewguy) I think that ClearableFileInput.userequiredattribute() (​) should be moved to FileInput.userequiredattribute() so that required is not output on the html input element that represents FileInput when a file is already set (e.g. already saved on a model instance that is being edited). Maybe I am overlooking a use case where this is not desirable? I can not think of one."}
{"_id":"q-en-django__django-12281-f5e46a495f5a1484fc3515a05a5e51079a29862da235a7cb5cbe59e049a53cd0","text":"admin.E130 (duplicate name attributes of actions) should specify which were duplicated. Description The fact that the name is used is somewhat an implementation detail, and there's no guarantee the user has enough of an understanding of python to know what that attribute is, let alone how to fix it. This just came up on IRC because a user had defined actions = [deleteselected] where deleteselected was a reference to their own callable, but shares the name of the base one (and by specifying the actions = they were assuming that they were wholesale replacing the actions list, where that may not be true for site-wide actions) so errored ... but they only had define a list of len(...) == 1 so how can there be a duplicate (is their thought process)? The error message should specify those names that occur 2(rather than just check len(...) vs len(set(...))), and ought ideally to explain where the duplicate comes from (ie: AdminSite-wide). Related ticket about E130: (+ those it references) but is about the replacement strategy rather than the error message itself."}
{"_id":"q-en-django__django-12284-cc3b8e33524f643f178b08beb21cc6c17a34b70fb95f4dacc6e677a604ffa4f1","text":"Model.getFOOdisplay() does not work correctly with inherited choices. Description (last modified by Mariusz Felisiak) Given a base model with choices A containing 3 tuples Child Model inherits the base model overrides the choices A and adds 2 more tuples getfoodisplay does not work correctly for the new tuples Example: class A(models.Model): foochoice = [(\"A\",\"output1\"),(\"B\",\"output2\")] fieldfoo = models.CharField(maxlength=254,choices=foochoice) class Meta: abstract = True class B(A): foochoice = [(\"A\",\"output1\"),(\"B\",\"output2\"),(\"C\",\"output3\")] fieldfoo = models.CharField(maxlength=254,choices=foochoice) Upon invoking getfieldfoo_display() on instance of B , For value \"A\" and \"B\" the output works correctly i.e. returns \"output1\" \/ \"output2\" but for value \"C\" the method returns \"C\" and not \"output3\" which is the expected behaviour"}
{"_id":"q-en-django__django-12286-2b2076fb7f36b656d6b6fad1033fe8ce0f3b7e4396a56f8c79f040e6a61f2cc6","text":"translation.E004 shouldn't be raised on sublanguages when a base language is available. Description According to Django documentation: If a base language is available but the sublanguage specified is not, Django uses the base language. For example, if a user specifies de-at (Austrian German) but Django only has de available, Django uses de. However, when using Django 3.0.2, if my has LANGUAGECODE = \"de-at\" I get this error message: SystemCheckError: System check identified some issues: ERRORS: ?: (translation.E004) You have provided a value for the LANGUAGECODE setting that is not in the LANGUAGES setting. If using LANGUAGE_CODE = \"es-ar\" Django works fine (es-ar is one of the translations provided out of the box)."}
{"_id":"q-en-django__django-12299-5d12fbfeab7e16f76ff01fa3822844cbe6e3cace6a416006301fb5ddfce16522","text":"Raise a descriptive error on update()\/delete() operations following (), intersection(), and difference(). Description (last modified by Joon Hwan ) bfilter() seems to merge but does not apply to the actual update q = () q = q.union(M.objects.afilter()) print(q) q = q.union(M.objects.bfilter()) print(q) q.update(name='foo') <QuerySet [<M: M object (675)]<QuerySet [<M: M object (675), <M: M object (773)]UPDATE \"m\" SET \"name\" = \"foo\" WHERE ORDER BY U0.\"id\" ASC LIMIT 1); args=(...)"}
{"_id":"q-en-django__django-12304-a3154bcbebb436d6c913c2a926ac7e8a381a44573467dc76504f77da03f55c58","text":"Enumeration Types are not usable in templates. Description (last modified by Mariusz Felisiak) The new ​enumeration types are great but can't be used in Django templates due to their being callable. For example this doesn't work: {% if student.yearinschool == YearInSchool.FRESHMAN %} This is because YearInSchool, being a class, is callable, and Django Templates always call callables with no arguments. The call fails because the required value argument is missing. The easy solution would be to declare donotcallintemplates = True on the various Choices classes."}
{"_id":"q-en-django__django-12306-02167a146aeec855f58909bb037f5b46dc40eb796e6067fd684f645f528eb6e8","text":"Named groups in choices are not properly validated in case of non str typed values. Description In case of using typed choices and string value to store it (in my case it is multiple values stored in char field as JSON) it is possible to catch error while run makemigrations (checkchoices error): main.MultiValueFieldModel.multivaluefieldintegerwithgroupedchoices: (fields.E005) 'choices' must be an iterable containing (actual value, human readable name) tuples. Looking deeper into the django code, we see actual error message: 'int' object is not iterable and it happens in this block of code (​): if self.maxlength is not None and groupchoices: choicemaxlength = max( choicemaxlength, (len(value) for value, in groupchoices if isinstance(value, str)), ) If we have CharField (any other with maxlength defined) and grouped choices with non str typed values like: choices=( ('one', ((1, 'One',), (11, 'Eleven',),),), ('two', ((2, 'Two',), (22, 'Twenty two',),),), ) we will have the situation, when max function receives only one integer value (choicemaxlength), because (len(value) for value, in groupchoices if isinstance(value, str)) will return empty generator, and that is why error 'int' object is not iterable raises (max function waits iterable if there is only one argument). Code block: choicemaxlength = max( choicemaxlength, (len(value) for value, in groupchoices if isinstance(value, str)), ) in this case works like: choicemaxlength = max( choicemaxlength, [], ) which is incorrect. The simples solution is to add one additional argument to max function, which will be usefull only in this partucular situation: choicemaxlength = max( choicemaxlength, 0, (len(value) for value, in group_choices if isinstance(value, str)), )"}
{"_id":"q-en-django__django-12308-a412e5d3d1db6597f6d4f1485c32dcd4d67a7ab9b4c81d08c5103836ba98c861","text":"JSONField are not properly displayed in admin when they are readonly. Description JSONField values are displayed as dict when readonly in the admin. For example, {\"foo\": \"bar\"} would be displayed as {'foo': 'bar'}, which is not valid JSON. I believe the fix would be to add a special case in to call the prepare_value of the JSONField (not calling directly to take care of the InvalidJSONInput case)."}
{"_id":"q-en-django__django-12313-abc9f67dc7d0205c23e51d8c62ac3d905e377ffd178033de0135e1d58da37270","text":"makemigrations does not detect\/like model name case changes Description Starting with class Evidence(models.Model): rubrictype = models.ForeignKey('Rubrictype') class Rubrictype(models.Model): typecode = models.CharField(maxlength=1) Make the initial migration: $ makemigrations Migrations for 'asmigrations': Create model Evidence Create model Rubrictype Add field rubrictype to evidence Change the name of Rubrictype to RubricType: class Evidence(models.Model): rubrictype = models.ForeignKey('RubricType') class RubricType(models.Model): typecode = models.CharField(maxlength=1) Generate the migration: $ makemigrations Migrations for 'asmigrations': Alter field rubrictype on evidence Django does not detect the name change on the RubricType model itself. No confirmation is requested for the name change and no operation is generated. The problem is that any subsequent makemigrations run will generate the same operation ad infinitum: $ makemigrations Migrations for 'asmigrations': Alter field rubrictype on evidence If instead the name is changed to RubricXype: class Evidence(models.Model): rubrictype = models.ForeignKey('RubricXype') class RubricXype(models.Model): typecode = models.CharField(maxlength=1) the corresponding migration becomes $ makemigrations Did you rename the asmigrations.Rubrictype model to RubricXype? [y\/N] y Migrations for 'asmigrations': Rename model Rubrictype to RubricXype This migration generates a RenameModel operation only and any subsequent makemigrations runs will properly report \"No changes detected\". So it appears the change detector does not pick up on capitalization changes in model names. Trying to work around by adding a migrations.RenameModel( oldname='Rubrictype', newname='RubricType', ) to the auto generated operations results in a ValueError exception when makemigrations is run again: $ makemigrations Traceback (most recent call last): File \"\", line 10, in <moduleexecutefromcommandline() File \"\/home\/svencoenye\/developer\/djangotest\/lib\/python2.7\/site-\", line 385, in executefromcommandline utility.execute() File \"\/home\/svencoenye\/developer\/djangotest\/lib\/python2.7\/site-\", line 377, in execute self.fetchcommand(subcommand).runfromargv() File \"\/home\/svencoenye\/developer\/djangotest\/lib\/python2.7\/site-\", line 288, in runfromargv self.execute(args, ) File \"\/home\/svencoenye\/developer\/djangotest\/lib\/python2.7\/site-\", line 338, in execute output = (args, options) File \"\/home\/svencoenye\/developer\/djangotest\/lib\/python2.7\/site-\", line 111, in handle convertapps=applabels or None, File \"\/home\/svencoenye\/developer\/djangotest\/lib\/python2.7\/site-\", line 42, in changes changes = self.detectchanges(convertapps, graph) File \"\/home\/svencoenye\/developer\/djangotest\/lib\/python2.7\/site-\", line 109, in detectchanges self.oldapps = (ignoreswappable=True) File \"\/home\/svencoenye\/developer\/djangotest\/lib\/python2.7\/site-\", line 89, in render model=lookupmodel, ValueError: Lookup failed for model referenced by field asmigrations.Evidence.rubrictype: asmigrations.RubricType The sequence of the operations does not matter. Neither does substituting the RenameModel operation for the AlterField operation. (Looking at the next to last entry in the traceback, the autodetector seems to be looking for the new name in the oldapps state?) It is possible, however, to go the long way around and use two separate migrations: Rubrictype -RubricXype. RubricXype -RubricType works without getting the migration state stuck and does not throw an exception."}
{"_id":"q-en-django__django-12325-6cfd0806081accb774e0ac9b99ecf08e90b5c77bd75efbaf37cfcb66a346fdf9","text":"pk setup for MTI to parent get confused by multiple OneToOne references. Description class Document(models.Model): pass class Picking(Document): documentptr = models.OneToOneField(Document, ondelete=models.CASCADE, parentlink=True, relatedname='+') origin = models.OneToOneField(Document, relatedname='picking', ondelete=models.PROTECT) produces Add parentlink=True to class Picking(Document): origin = models.OneToOneField(Document, relatedname='picking', ondelete=models.PROTECT) documentptr = models.OneToOneField(Document, ondelete=models.CASCADE, parentlink=True, relatedname='+') Works First issue is that order seems to matter? Even if ordering is required \"by design\"(It shouldn't be we have explicit parentlink marker) shouldn't it look from top to bottom like it does with managers and other things?"}
{"_id":"q-en-django__django-12343-c14e0ed3317996d7fdeb99b77eea94f9c183a76af69894fb9fe0055f82994214","text":"Admin: Render foreign key models as links for readonly users Description In the admin UI, when viewing a model for which you have view only permission, foreign key \/ m2m fields are rendered as plaintext representation of the target object. It would be nicer to render those as links instead so that a readonly user can navigate through the relationships. The link should only be rendered if the user has permission to view the target model. django-developers discussion: ​!topic\/django-developers\/XFoaohDpqZE"}
{"_id":"q-en-django__django-12360-83457aa691a43ab30771fcdc8085566e306a6e79e872965409e96bdc3e89dbc8","text":"Add system check for the length of auth permissions codenames. Description I stumbled across this while performing some migrations on models with rather... descriptive names (my original model was dynamically created). Anyway, it looks like in cases where a model name is just under the 100 character limit, and is used, the codenames generated for the default permissions (ie. add, change, delete, view) can exceed the maximum 100 characters on the Permission model. As an example, the model below having a 98 character name allows migrations to be generated, but upon running the migration, a database error is raised when it tries to create permissions with codenames above 100 characters. The model: class SomeExcessivelyDescriptiveModelWithAnAbsolutelyRidiculousNameThatCouldntEverNeedToBeNearlyThisLong(models.Model): field1 = models.CharField(maxlength=25) The error: value too long for type character varying(100) While I'm aware that you can override the default permissions by setting Meta.defaultpermissions on the model (this is how I fixed my app), and that the majority of people would never need model names this long, I figured it might be worth adding validation for this case since it uses 's default behaviour. Also, the error message originally came up while using a flush command, which seemed counterintuitive."}
{"_id":"q-en-django__django-12364-a45b3263ffcbf4e598eba6975d91595505c5d84625332df7292217fe0a5b6691","text":"Detection of existing total ordering in admin changelist should take into account UniqueConstraints without conditions. Description I've been fiddling with db indexes lately to improve the performance of an admin view. Eventually I found this PR ​ which ensures the records displayed by ChangeList are deterministically ordered. Among other things, the code looks for the presence of a uniquetogether attribute on the Meta class of the model to determine if the model is already totally ordered. I think that getdeterministicordering() should check for the presence of UniqueConstraints as well, which currently aren’t considered by the discovery algorithm. I noticed the docs currently advise against using unique_together in favor of UniqueConstraint, suggesting the former may be deprecated in the future, so this fix would change that part of the code accordingly. I'm willing to submit a PR for this. Cheers, Fabio"}
{"_id":"q-en-django__django-12394-0a9c7deb332bd2ea651aedc7f2f6026329c198f8e9a53c9ee84bf42a06c19f70","text":"Raising error about protected related objects can crash. Description (last modified by Matthias Kestenholz) ERROR: testprotectvia () Traceback (most recent call last): File \"\", line 59, in testPartExecutor yield File \"\", line 605, in run testMethod() File \"\", line 99, in testprotectvia p.delete() File \"\", line 941, in delete collector.collect([self], keepparents=keepparents) File \"\", line 300, in collect error.protectedobjects[0], TypeError: '' object is not subscriptable Pull request follows. (By the way, this came up while testing ​!msg\/django-developers\/WmgqJnQ6ioE\/b52uACoPAgAJ \/ the JSONField GSoC pull request. Apart from this crash everything worked fine!)"}
{"_id":"q-en-django__django-12406-18fb55b3aae2489226a25c2f0f0d3ad351b8c0d1a6135305c4d6eb819f2bc015","text":"ModelForm RadioSelect widget for foreign keys should not present a blank option if blank=False on the model Description Unlike the select widget, where a blank option is idiomatic even for required fields, radioselect has an inherent unfilled state that makes the \"-------\" option look suspiciously like a valid choice. class TestRun(models.Model): datafile = models.ForeignKey(BatchData, ondelete=models.SETNULL, null=True, blank=False) class TestRunForm(ModelForm): class Meta: model = TestRun fields = ['datafile'] widgets = {'datafile': RadioSelect()} renders {{testrunform.datafile}} as <ul id=\"iddatafile\"<li<label for=\"iddatafile0\"<input checked=\"checked\" id=\"iddatafile0\" name=\"datafile\" type=\"radio\" value=\"\"--------- <\/label<\/li<li<label for=\"iddatafile1\"<input id=\"iddatafile1\" name=\"datafile\" type=\"radio\" value=\"1\"First Data File <\/label<\/li<\/ulInstead, there should be no checked option for RadioSelect's <inputtags when rendering a new form from a model if blank is not a valid selection."}
{"_id":"q-en-django__django-12407-6ee1657e27b48f578ab6a6ff12cd400512958edadcee9237fe6df14793971964","text":"{% include %} uses gettemplate where it could selecttemplate Description It'd be nice if the Include template tag was sensible enough to allow fallbacks by selecting the most appropriate template, as things like render\/rendertoresponse\/rendertostring do. It's tripped me up on more than one occasion, and it seems a trivial feature to support, from my limited testing. from django.template import Template, Context tmpl = Template('{% include var %}') ctx = Context({'var':''}) ctx [{'var': ''}] (ctx) ... some HTML output ... ({'var':['', '']}) {'var': ['', '']} (ctx) Traceback (most recent call last): File \"<console\", line 1, in <moduleFile \"\", line 140, in render return (context) File \"\", line 134, in render return (context) File \"\", line 823, in render bit = self.rendernode(node, context) File \"\", line 74, in rendernode return (context) File \"\", line 165, in render template = gettemplate(templatename) File \"\", line 145, in gettemplate template, origin = findtemplate(templatename) File \"\", line 138, in findtemplate raise TemplateDoesNotExist(name) TemplateDoesNotExist: ['', ''] The 'fix' is to change ​this line from gettemplate to selecttemplate, though this might now be slightly complicated by the recent changes in to allow for rendering of Template instances. Changing to selecttemplate on 1.4 yields the results I'd expect: from django.template import Template, Context tmpl = Template('{% include var %}') ctx = Context({'var':['', '']}) (ctx) ... some HTML output ... Both the above shell sessions assume is in INSTALLED_APPS."}
{"_id":"q-en-django__django-12419-a8f4f19965d32cbf521bef2d2da1ea6515f2216a06000c40375d0c718012e87d","text":"Add secure default SECUREREFERRERPOLICY \/ Referrer-policy header Description the ability for the SECUREREFERRERPOLICY setting to set Referrer-Policy, released in Django 3.0. I propose we change the default for this to \"same-origin\" to make Django applications leak less information to third party sites. The main risk of breakage here would be linked websites breaking, if they depend on verification through the Referer header. This is a pretty fragile technique since it can be spoofed. Documentation: ​ The MDN support grid is out of date: ​"}
{"_id":"q-en-django__django-12430-365e638e6776505ac98e83719c349e11c499f42f30c4ab94addb5c0bca518e15","text":"Possible data loss when using caching from async code. Description CacheHandler use instead of , hence it's a chance of data corruption if someone tries to use caching from async code. There is a potential race condition if two coroutines touch the same cache object at exactly the same time."}
{"_id":"q-en-django__django-12431-ceba5660eafcf2e29163d1a7f57e032179e07a90e49d34dd96e81bd92bf54c96","text":"FileResponse with temporary file closing connection. Description (last modified by Oskar Persson) I think I might've found a regression in . When I run the following tests (in their defined order) against Postgres I get the error below. import tempfile from import getusermodel from import FileResponse from import TestCase User = getusermodel() class MyTests(TestCase): def setUp(self): = (username='user') def testfirst(self): with tempfile.TemporaryFile() as f: return FileResponse(f) def testsecond(self): pass Running tests... .E ERROR [0.003s]: testsecond (responses.testfileresponse.MyTests) Traceback (most recent call last): File \"\", line 238, in cursor return self.preparecursor(self.createcursor(name)) File \"\", line 26, in inner return func(args, kwargs) File \"\", line 231, in createcursor cursor = () psycopg2.InterfaceError: connection already closed The above exception was the direct cause of the following exception: Traceback (most recent call last): File \"\", line 19, in setUp = (username='user') File \"\", line 82, in managermethod return getattr(self.getqueryset(), name)(args, kwargs) File \"\", line 433, in create (forceinsert=True, ) File \"\", line 66, in save super().save(args, kwargs) File \"\", line 746, in save forceupdate=forceupdate, updatefields=updatefields) File \"\", line 784, in savebase forceupdate, using, updatefields, File \"\", line 887, in savetable results = self.doinsert(cls.basemanager, using, fields, returningfields, raw) File \"\", line 926, in doinsert using=using, raw=raw, File \"\", line 82, in managermethod return getattr(self.getqueryset(), name)(args, kwargs) File \"\", line 1204, in insert return query.getcompiler(using=using).executesql(returningfields) File \"\", line 1382, in executesql with () as cursor: File \"\", line 26, in inner return func(args, kwargs) File \"\", line 260, in cursor return () File \"\", line 238, in cursor return self.preparecursor(self.createcursor(name)) File \"\", line 90, in exit raise djexcvalue.withtraceback(traceback) from excvalue File \"\", line 238, in cursor return self.preparecursor(self.createcursor(name)) File \"\", line 26, in inner return func(args, kwargs) File \"\", line 231, in create_cursor cursor = () connection already closed"}
{"_id":"q-en-django__django-12441-d3edc9ab0846b2a78e02d205b3920c56870c8ecafeaf150f799ebc139fe36393","text":"Calling a form method htmloutput modifies the dict for NONFIELDERRORS if there are hidden field with errors Description Each time the htmloutput method of a form is called, it appends the errors of the hidden field errors to the NONFIELDERRORS (all) entry. This happen for example when the form methods asp() astable() asul() are called multiple time, or any other method that themselves call one of them. For example, a test form with an hidden input field that add errors during the clean call. Python 3.6.5 (default, Apr 25 2018, 14:26:36) Type 'copyright', 'credits' or 'license' for more information IPython 6.4.0 -- An enhanced Interactive Python. Type '?' for help. In [1]: import django In [2]: django.version Out[2]: '2.1.7' In [3]: from django import forms ...: In [4]: class TestForm(forms.Form): ...: hiddeninput = forms.CharField(widget=forms.HiddenInput) ...: ...: def clean(self): ...: self.adderror(None, 'Form error') ...: self.adderror('hiddeninput', 'Hidden input error') ...: In [5]: testform = TestForm({}) In [6]: Out[6]: {'hiddeninput': ['This field is required.', 'Hidden input error'], 'all': ['Form error']} In [7]: print(testform.astable()) <tr<td colspan=\"2\"<ul class=\"errorlist nonfield\"<liForm error<\/li<li(Hidden field hiddeninput) This field is required.<\/li<li(Hidden field hiddeninput) Hidden input error<\/li<\/ul<input type=\"hidden\" name=\"hiddeninput\" id=\"idhiddeninput\"<\/td<\/trIn [8]: Out[8]: {'hiddeninput': ['This field is required.', 'Hidden input error'], 'all': ['Form error', '(Hidden field hiddeninput) This field is required.', '(Hidden field hiddeninput) Hidden input error']} In [9]: print(testform.astable()) <tr<td colspan=\"2\"<ul class=\"errorlist nonfield\"<liForm error<\/li<li(Hidden field hiddeninput) This field is required.<\/li<li(Hidden field hiddeninput) Hidden input error<\/li<li(Hidden field hiddeninput) This field is required.<\/li<li(Hidden field hiddeninput) Hidden input error<\/li<\/ul<input type=\"hidden\" name=\"hiddeninput\" id=\"idhiddeninput\"<\/td<\/trIn [10]: Out[10]: {'hiddeninput': ['This field is required.', 'Hidden input error'], 'all': ['Form error', '(Hidden field hiddeninput) This field is required.', '(Hidden field hiddeninput) Hidden input error', '(Hidden field hiddeninput) This field is required.', '(Hidden field hiddeninput) Hidden input error']} In [11]: testform.nonfielderrors() Out[11]: ['Form error', '(Hidden field hiddeninput) This field is required.', '(Hidden field hiddeninput) Hidden input error', '(Hidden field hiddeninput) This field is required.', '(Hidden field hiddeninput) Hidden input error'] This bug affects probably also version 2.2. A simple fix would be to use a copy of the error list before adding the hidden field errors in the file --- 2019-03-17 18:59:04. +0100 +++ 2019-03-17 19:00:08. +0100 -194,7 +194,7 def htmloutput(self, normalrow, errorrow, rowender, helptexthtml, errorsonseparaterow): \"Output HTML. Used by astable(), asul(), asp().\" toperrors = self.nonfielderrors() # Errors that should be displayed above all fields. toperrors = self.nonfielderrors().copy() # Errors that should be displayed above all fields. output, hidden_fields = [], [] for name, field in ():"}
{"_id":"q-en-django__django-12453-de81bea21185dd4647df6fbe08dc3e6256c1bbe1383c30ec04442a63e1fc1881","text":"fails to restore objects due to ordering constraints Description I hit this problem in a fairly complex projet and haven't had the time to write a minimal reproduction case. I think it can be understood just by inspecting the code so I'm going to describe it while I have it in mind. Setting serializedrollback = True on a TransactionTestCase triggers ​rollback emulation. In practice, for each database: BaseDatabaseCreation.createtestdb calls connection.testserializedcontents = connection.creation.serializedbtostring() TransactionTestCase.fixturesetup calls connection.creation.deserializedbfromstring(connection.testserializedcontents) (The actual code isn't written that way; it's equivalent but the symmetry is less visible.) serializedbtostring orders models with serializers.sortdependencies and serializes them. The sorting algorithm only deals with natural keys. It doesn't do anything to order models referenced by foreign keys before models containing said foreign keys. That wouldn't be possible in general because circular foreign keys are allowed. deserializedbfromstring deserializes and saves models without wrapping in a transaction. This can result in integrity errors if an instance containing a foreign key is saved before the instance it references. I'm suggesting to fix it as follows: diff --git index .. --- +++ -4,7 +4,7 import time from import apps from import settings from import serializers -from import router +from import router, transaction from import StringIO from import input -128,8 +128,9 class BaseDatabaseCreation(object): the serializedbtostring method. \"\"\" data = StringIO(data) for obj in serializers.deserialize(\"json\", data, ): () with (): for obj in serializers.deserialize(\"json\", data, ): () def getdatabasedisplaystr(self, verbosity, databasename): \"\"\" Note that loaddata doesn't have this problem because it wraps everything in a transaction: def handle(self, fixturelabels, *options): # ... with (): self.loaddata(fixturelabels) # ... This suggest that the transaction was just forgotten in the implementation of deserializedbfromstring. It should be possible to write a deterministic test for this bug because the order in which serializedbto_string serializes models depends on the app registry, and the app registry uses OrderedDict to store apps and models in a deterministic order."}
{"_id":"q-en-django__django-12458-e652219127e2aa46368de32d55514c68e8192440048d08db9dddcfbde8d99773","text":"Serialization dependency sorting disallows circular references unneccesarily. Description The core.serialization.sortdependencies() function takes a list of apps and\/or models, and resolves this into a sorted flat list of models, ready to be serialized in that order. This function is intended to make natural foreign keys work, by serializing models referenced by a natural key before the referencing models. When deserializing, this guarantees that natural key references can be resolved, because there are no \"forward references\". Furthermore, when a circular reference using natural keys is present, this function raises an exception (e.g. \"Can't resolve dependencies for someapp.SomeModel in serialized app list\") and prevents serialization from completing, since there is no way to guarantee a model ordering that will have no forward references. Note that this ordering is only needed when natural keys are involved, since data is intended to be loaded in a transaction without constraint checks, so numerical foreign keys can be in the wrong order, as long as all referenced data is present at the end of the transaction. This does not work with natural keys, since those are resolved by Python code that needs the referenced objects present in the database to resolve them. However, this sorting is not actually strictly necessary in all cases where it is applied. When circular references are involved, this then actually prevents serialization for no good reason. In particular, this is the case: When running dumpdata without natural keys enabled (which is the default). Even though natural keys might be defined in the models (which causes the sorting and exception), no natural keys will be present in the dumped data, so no ordering is needed. When dumping data intended for loading with loaddata (which I think is the primary usecase for dumpdata?). loaddata will (since 17 months ago in v2.2, see ) automatically handle forward references by deferring setting fields that reference natural keys that are not yet. In this case, sorting is still useful, to prevent forward references where possible, but when there are circular references, it is acceptable to ignore some dependencies rather than preventing serialization from happening alltogether. When serializing data for tests for serializedrollback=True (in ). This is a serialization that does not use natural keys, so no ordering is needed at all. Note that this serialization happens always (unlike deserialization only happens with serializedrollback=True), so AFAIU this effectively prevents any tests from working on a database with circular references with natural keys defined. The fix for these issues seems to be rather simple: For dumpdata without usenaturalforeignkeys, skip the ordering and just serialize all models in arbitrary order. AFAICS usenaturalprimarykeys is not relevant here, since that only controls omitting the numerical primary key. For dumpdata with usenaturalforeignkeys, do the ordering but do not bail out when there are circular references (instead just ignore some dependencies and produce a best-effort ordering). For test database serialization, also skip the ordering and serialize in arbitrary order. Note that this would remove two of the three calls to sortdependencies() and allow loops in the last remaining instance. This means that sort_dependencies could be modified to allow loops unconditionally, or we could add an argument and default to disallowing loops in case any code outside of django is using this function? Note that is a related, but different issue, concerning the deserialization of data in testcases. I've been working on fixing this and that related issue today and have a basic version working, with testcases (which proved to be quite a challenge, since testing the test subsystem is a bit tricky...). I'll do some additional testing and cleanup and submit a PR soon. Also note that the circular-reference exception was already disabled for self-referencing models in . The fix for that issue simply ignores self-referencing models for sorting, without taking any additional measures to sort instances to prevent problems in deserialization (this code was when the deferred deserialization did not exist yet), so I wonder how much value this exception still has."}
{"_id":"q-en-django__django-12464-faaef4392319b9fb9e6ef5975a54c2563dcdd4fc3b6b08b23f65e92afded1d0d","text":"DISTINCT with GROUPCONCAT() and multiple expressions raises NotSupportedError on SQLite. Description Contrary to what is suggested in ​lines 60-64 of , SQLite does support DISTINCT on aggregate functions. One such example is GROUPCONCAT, which is quite similar to PostgreSQL's STRINGAGG. I can't find any canonical links which provide a useful explanation of GROUPCONCAT, but this should be good enough: ​ I came across this issue when trying to create my own GroupConcat function subclassing Aggregate (similar to the ​StringAgg implementation from postgres) and noticed that skipping the check in would allow my queries to run as advertised. My code for GroupConcat is: from import Value from import Aggregate class GroupConcat(Aggregate): function = 'GROUPCONCAT' template = '%(function)s(%(distinct)s %(expressions)s)' allowdistinct = True def init(self, expression, delimiter=None, extra): if delimiter is not None: self.allowdistinct = False delimiterexpr = Value(str(delimiter)) super().init(expression, delimiterexpr, extra) else: super().init(expression, extra) def assqlite(self, compiler, connection, extracontext): return super().assql( compiler, connection, function=self.function, template=self.template, **extracontext ) For the record, as the code above suggests, a separate issue is that GROUP_CONCAT only allows DISTINCT when a delimiter isn't specified. After some discussion on #django, an argument was raised in favor of changing the message to say \"Django doesn't support...\", but I would imagine that skipping the check entirely would simply result in an OperationalError for malformed queries while still allowing users to extend the ORM as needed."}
{"_id":"q-en-django__django-12469-66294af80cad4dd23c9916471e055ccf371ae2c932c0f4e3f62f8301bb53be86","text":"Admin datehierarchy filter by month displays an extra day at timezone boundary. Description (last modified by Lavrenov Ivan) When I authorized by user with not-UTC timezone, like America\/LosAngeles , and open filter by date in month, I see one extra day, that follows to the first day of the previous month"}
{"_id":"q-en-django__django-12470-c25707085e266f81bd0eb9029e94f0c40d43d8a2f8c7a5af34a0cd5a552a7adf","text":"Inherited model doesn't correctly order by \"-pk\" when specified on Parent.Meta.ordering Description Given the following model definition: from import models class Parent(models.Model): class Meta: ordering = [\"-pk\"] class Child(Parent): pass Querying the Child class results in the following: print(().query) SELECT \"myappparent\".\"id\", \"myappchild\".\"parentptrid\" FROM \"myappchild\" INNER JOIN \"myappparent\" ON (\"myappchild\".\"parentptrid\" = \"myappparent\".\"id\") ORDER BY \"myapp_parent\".\"id\" ASC The query is ordered ASC but I expect the order to be DESC."}
{"_id":"q-en-django__django-12477-1ec7c3c0324365136cce7a82d907691398076c86f5e13ad9b2817dfee72ced37","text":"fields.E310-E311 should take into account UniqueConstraints without conditions. Description Hello, I'm trying to create migration with this kind of model. class AppUsers(models.Model): name = models.CharField(...) uid = models.CharField(...) source = models.ForeignKey(...) class Meta: constraints = [models.UniqueConstraint(fields=['uid', 'source'], name='appusersuniqueness')] When I start makemigrations command in I've fields.E310 ​ error It says that I should add uniquetogether field in Meta options: (fields.E310) No subset of the fields 'uid', 'source' on model 'AppUsers' is unique. HINT: Add unique=True on any of those fields or add at least a subset of them to a uniquetogether constraint. If I change Meta options to uniquetogether constraint migration passes with no errors. class AppUsers(models.Model): name = models.CharField(...) uid = models.CharField(...) source = models.ForeignKey(...) class Meta: uniquetogether = [['uid', 'source']] As mentioned in docs ​ unique_together may be deprecated in the future. So I think nobody wants to face this issue when this will be deprecated :) Thanks, Pavel"}
{"_id":"q-en-django__django-12484-4d23343d51fc5c5d327eb4a795e300ae60978b4bf5b560c43b752ca642a621f4","text":"system checks: admin.E002 could provide a hint but doesn't Description Currently the output is: myapp.MyCustomUserModel: (auth.E002) The field named as the 'USERNAMEFIELD' for a custom user model must not be included in 'REQUIREDFIELDS'. because I accidentally had: USERNAMEFIELD = \"email\" EMAILFIELD = \"email\" REQUIREDFIELDS = (USERNAMEFIELD, \"full_name\",) Ignoring the fact that Django knows it's wrong to have it in there, and could easily just skip it or warn if it's not set, it doesn't make use of the hints available in system checks. I'd like to suggest that a hint could be provided which says which field it's moaning about, something like (open to better wording): HINT: Your username field is currently set to \"email\", you should remove \"email\" from your required fields definition. It's a stupidly little thing, but having not had to make a custom user from scratch for a while, my eyes glazed over the not in must not be and I was confused for all of 2 minutes before actually reading it properly."}
{"_id":"q-en-django__django-12485-dc2ba324eac8c590ba54f9332533fa8918b7867cf0b0296d6bee0ba4537a160c","text":"MultiPartParser support double quotes Description Although the rfc2231 document does not indicate that values can be wrapped in double quotes. However, some third-party tools wrap the value in double quotation marks when wrapping HTTP requests (such as the filename of the file uploaded by PostmanCanary). This results in double quotes for the filename at the end of Django parsing. Postman request body: ---------------------------- Content-Disposition: form-data; name=\"file\"; filename=\".txt\"; filename*=\"UTF-8''\" Content-Type: text\/plain test ------------------------------ django got filename is .txt\" not .txt This is not a bug of Django. But I suggest Django do some compatibility processing. CPython is also compatible with this situation. ​ ​"}
{"_id":"q-en-django__django-12486-e3a97e8423a70e387373da978dafa706eca1677b1b50417556bcdbced6900c88","text":"() incorrectly formats large\/tiny floats in scientific notation Description (last modified by Tim Graham) For floats with values larger than 1e16 or smaller than 1e-5, their string representation uses scientific notation in Python, which causes to return an erroneous output. from import format format(0., '.', 2) '9e-19.00' format(1e16, '.', 2, thousandsep=',', grouping=3, forcegrouping=True) '1e,+16.00' This is similar to but that was only fixed for Decimal types."}
{"_id":"q-en-django__django-12496-ecb89361940e40e5b727e69e9e0bd8c8f6bfaf02f98525b4a2eab187c8f19ce1","text":"Child model updates parent model with empty fields making an extra query in multi-inheritance when parent model has custom PK Description While creating a new model object (using multi-inheritance model =Child(Parent)), Django does an extra update query setting parent model fields to empty values. This situation occurs only if we define a custom primary key in a parent model (eg. as an UUID field). An example without custom primary key (correct behavior): class Parent(models.Model): title = models.TextField() class Child(Parent): body = models.TextField() () INTO \"appparent\" (\"title\") VALUES ('') RETURNING \"appparent\".\"id\" INTO \"appchild\" (\"parentptrid\", \"body\") VALUES (1, '') An example with custom primary key (incorrect behavior): class Parent(models.Model): id = models.UUIDField( primarykey=True, default=uuid.uuid4, editable=False ) title = models.TextField() class Child(Parent): body = models.TextField() () \"appparent\" SET \"title\" = '' WHERE \"appparent\".\"id\" = '-ae7b-48a6-a2e0-'::uuid INTO \"appparent\" (\"id\", \"title\") VALUES ('-ae7b-48a6-a2e0-'::uuid, '') INTO \"appchild\" (\"parentptrid\", \"body\") VALUES ('-ae7b-48a6-a2e0-'::uuid, '') Python 3.6, PostgreSQL 9.6"}
{"_id":"q-en-django__django-12497-6ca925a258ff8c78f7c570b08de5cd355a248e3e4ffb4abe105c15c323c5087a","text":"Wrong hint about recursive relationship. Description (last modified by Matheus Cunha Motta) When there's more than 2 ForeignKeys in an intermediary model of a m2m field and no through_fields have been set, Django will show an error with the following hint: hint=( 'If you want to create a recursive relationship, ' 'use ForeignKey(\"%s\", symmetrical=False, through=\"%s\").' But 'symmetrical' and 'through' are m2m keyword arguments, not ForeignKey. This was probably a small mistake where the developer thought ManyToManyField but typed ForeignKey instead. And the symmetrical=False is an outdated requirement to recursive relationships with intermediary model to self, not required since 3.0. I'll provide a PR with a proposed correction shortly after. Edit: fixed description."}
{"_id":"q-en-django__django-12503-b97523f44a231f9d961883c50e27033cf4a3b5f98bd1db5a88bd62429d648260","text":"makemessages doesn't provide feedback when no locale is specified Description (last modified by Cristóbal Mackenzie) makemessages requires that one of three flags be passed to specify locales for message building: --locale to explicitly specify locales, --exclude to specify locales to exclude, or --all to build message files for all locales. When non of these flags are present, the command doesn't show any errors for the user. According to the source code, it should raise CommandError, but that never happens because of a bug in an if statement that checks if a locale has been specified. I've already fixed this in my fork and have submitted a small PR. ​ Please point out if there are any other necessary steps to move this forward. Thanks!"}
{"_id":"q-en-django__django-12504-c67d0bcf515a878101f941466c69cebeff090144608742b46d7dd30bf378dc1e","text":"Logout link should be protected Description There is a logout link in admin app. It is link, not a form. Therefore it is not CSRF-protected. Probably it is not so important to protect logout from CSRF attack, because this fact cannot be used to do anything harmful. So this is just a request for purity. Another reason is that GET request should never change invernal state of the system."}
{"_id":"q-en-django__django-12513-6ce405e77d96d14689043c4bdc536d239953763145a7ed0cec3128c374257569","text":"Deprecate providingargs argument from Signal Description The argument is and always has been purely documentational. It provides no functionality or checking. Therefore, these values are stored in memory for no real use. Documentation can be handled just as easily by a code comment or real documentation articles. On a more practical level, I rarely signals, so I recently had to look up their interface. Seeing this providingargs argument required me to think about how to best use it, only to realize it actually goes unused. We can remove this cognitive distraction from the docs. It has caused other small confusion in the past: ."}
{"_id":"q-en-django__django-12517-f206d57a262b049a6627b188f9ae6d79a25ee5401f84bec24415384c5815e5f2","text":"Inconsistent datetime logging from runserver. Description In Django 1.11 and higher, the runserver logging can sometimes be inconsistent. [16\/Apr\/2018 13:32:35] \"GET \/some\/local\/url HTTP\/1.1\" 200 7927 [2018-04-16 13:32:35,745] - Broken pipe from ('127.0.0.1', ) This is because logging from WSGIRequestHandler uses servertime as calculated using BaseHTTPServer.logdatetimestring. On the other hand, WSGIServer uses logging without providing a servertime. This gets \"fixed\" in using self.formatTime(record, self.datefmt), which uses a completely different format. Currently we make this at least consistent by providing the datefmt parameter when constructing the logger, but it would be better if they were coded to be in sync (and configurable?). (Looking into it further, it looks like we should be using %(asctime)s instead of %(servertime)s, but would be good if that were the suggested default. In ​ we see %(server_time)s.)"}
{"_id":"q-en-django__django-12518-f61e8abe75dedeba00473095087fc56208f0d52e4f50880f39e2bac64646a575","text":"sqlmigrate doesn't allow inspecting migrations that have been squashed Description This project for another ticket can be used to reproduce: ​ When running sqlmigrate to pick up migration 0001 in this project, it complains that two migrations have that prefix: $ python sqlmigrate testapp 0001 CommandError: More than one migration matches '0001' in app 'testapp'. Please be more specific. But when trying to be more specific, it's not possible to load it: $ python sqlmigrate testapp 0001initial Traceback (most recent call last): File \"\", line 21, in <modulemain() File \"\", line 17, in main executefromcommandline() File \"\", line 401, in executefromcommandline utility.execute() File \"\", line 395, in execute self.fetchcommand(subcommand).runfromargv() File \"\", line 328, in runfromargv self.execute(args, cmdoptions) File \"\", line 30, in execute return super().execute(args, options) File \"\", line 369, in execute output = (args, *options) File \"\", line 64, in handle plan = [([targets[0]], options['backwards'])] KeyError: ('testapp', '0001initial') It would be nice to: A) catch this error and report a nice message as well B) allow inspection of individual migrations that have been involved in a squash. Normally the workflow is to remove the individual migrations some time after committing the squash, but until that is done it could be useful to see their sql."}
{"_id":"q-en-django__django-12519-ce2b2238223500d8e77ae5a8fbddba1c66aa03e3b186adcf43603905d8f509dc","text":"Subquery annotations are omitted in group by query section if multiple annotation are declared Description (last modified by Johannes Maron) Sadly there is more regression in Django 3.0.2 even after . Background: It's the same query as . I tried upgrading to Django 3.0.2 and now I get duplicate results. Even tho they query should be distinct. Where on 2.2 the queryset yields 490 results, it's 519 on 3.0. A quick diff on the queries still reveals a different grouped by section: This is the new query on 3.0.2: SELECT DISTINCT \"campsoffer\".\"id\", \"campsoffer\".\"title\", \"campsoffer\".\"slug\", \"campsoffer\".\"isactive\", \"campsoffer\".\"modified\", \"campsoffer\".\"created\", \"campsoffer\".\"providerid\", \"campsoffer\".\"activitytype\", \"campsoffer\".\"description\", \"campsoffer\".\"highlights\", \"campsoffer\".\"importantinformation\", \"campsoffer\".\"minage\", \"campsoffer\".\"maxage\", \"campsoffer\".\"food\", \"campsoffer\".\"video\", \"campsoffer\".\"accommodation\", \"campsoffer\".\"accommodationtype\", \"campsoffer\".\"roomtype\", \"campsoffer\".\"roomsizemin\", \"campsoffer\".\"roomsizemax\", \"campsoffer\".\"externalurl\", \"campsoffer\".\"applicationform\", \"campsoffer\".\"caseload\", \"campsoffer\".\"fieldtrips\", MIN(T4.\"retailprice\") AS \"minretailprice\", (SELECT U0.\"id\" FROM \"campsservicepackage\" U0 INNER JOIN \"campsregion\" U2 ON (U0.\"regionid\" = U2.\"id\") WHERE (U0.\"companyid\" = 1 AND U0.\"option\" = \"campsoffer\".\"activitytype\" AND STContains(U2.\"locations\", T4.\"position\")) LIMIT 1) AS \"inpackage\", \"campsprovider\".\"id\", \"campsprovider\".\"title\", \"campsprovider\".\"slug\", \"campsprovider\".\"isactive\", \"campsprovider\".\"modified\", \"campsprovider\".\"created\", \"campsprovider\".\"logo\", \"campsprovider\".\"description\", \"campsprovider\".\"video\", \"campsprovider\".\"externalurl\", \"campsprovider\".\"terms\", \"campsprovider\".\"cancellationpolicy\", \"campsprovider\".\"privacypolicy\", \"campsprovider\".\"applicationform\" FROM \"campsoffer\" LEFT OUTER JOIN \"campsbookingoption\" ON (\"campsoffer\".\"id\" = \"campsbookingoption\".\"offerid\") INNER JOIN \"campsprovider\" ON (\"campsoffer\".\"providerid\" = \"campsprovider\".\"id\") INNER JOIN \"campsbookingoption\" T4 ON (\"campsoffer\".\"id\" = T4.\"offerid\") WHERE (\"campsoffer\".\"isactive\" = True AND \"campsprovider\".\"isactive\" = True AND T4.\"end\" = STATEMENTTIMESTAMP() AND T4.\"isactive\" = True AND \"campsoffer\".\"maxage\" = 5 AND \"campsoffer\".\"minage\" <= 13 AND (SELECT U0.\"id\" FROM \"campsservicepackage\" U0 INNER JOIN \"campsregion\" U2 ON (U0.\"regionid\" = U2.\"id\") WHERE (U0.\"companyid\" = 1 AND U0.\"option\" = \"campsoffer\".\"activitytype\" AND STContains(U2.\"locations\", T4.\"position\")) LIMIT 1) IS NOT NULL) GROUP BY \"campsoffer\".\"id\", T4.\"position\", \"campsprovider\".\"id\" ORDER BY \"campsoffer\".\"created\" ASC And what it was (and should be) on 2.2.9: SELECT DISTINCT \"campsoffer\".\"id\", \"campsoffer\".\"title\", \"campsoffer\".\"slug\", \"campsoffer\".\"isactive\", \"campsoffer\".\"modified\", \"campsoffer\".\"created\", \"campsoffer\".\"providerid\", \"campsoffer\".\"activitytype\", \"campsoffer\".\"description\", \"campsoffer\".\"highlights\", \"campsoffer\".\"importantinformation\", \"campsoffer\".\"minage\", \"campsoffer\".\"maxage\", \"campsoffer\".\"food\", \"campsoffer\".\"video\", \"campsoffer\".\"accommodation\", \"campsoffer\".\"accommodationtype\", \"campsoffer\".\"roomtype\", \"campsoffer\".\"roomsizemin\", \"campsoffer\".\"roomsizemax\", \"campsoffer\".\"externalurl\", \"campsoffer\".\"applicationform\", \"campsoffer\".\"caseload\", \"campsoffer\".\"fieldtrips\", MIN(T4.\"retailprice\") AS \"minretailprice\", (SELECT U0.\"id\" FROM \"campsservicepackage\" U0 INNER JOIN \"campsregion\" U2 ON (U0.\"regionid\" = U2.\"id\") WHERE (U0.\"companyid\" = 1 AND U0.\"option\" = (\"campsoffer\".\"activitytype\") AND STContains(U2.\"locations\", (T4.\"position\"))) LIMIT 1) AS \"inpackage\", \"campsprovider\".\"id\", \"campsprovider\".\"title\", \"campsprovider\".\"slug\", \"campsprovider\".\"isactive\", \"campsprovider\".\"modified\", \"campsprovider\".\"created\", \"campsprovider\".\"logo\", \"campsprovider\".\"description\", \"campsprovider\".\"video\", \"campsprovider\".\"externalurl\", \"campsprovider\".\"terms\", \"campsprovider\".\"cancellationpolicy\", \"campsprovider\".\"privacypolicy\", \"campsprovider\".\"applicationform\" FROM \"campsoffer\" LEFT OUTER JOIN \"campsbookingoption\" ON (\"campsoffer\".\"id\" = \"campsbookingoption\".\"offerid\") INNER JOIN \"campsprovider\" ON (\"campsoffer\".\"providerid\" = \"campsprovider\".\"id\") INNER JOIN \"campsbookingoption\" T4 ON (\"campsoffer\".\"id\" = T4.\"offerid\") WHERE (\"campsoffer\".\"isactive\" = True AND \"campsprovider\".\"isactive\" = True AND T4.\"end\" = (STATEMENTTIMESTAMP()) AND T4.\"isactive\" = True AND (SELECT U0.\"id\" FROM \"campsservicepackage\" U0 INNER JOIN \"campsregion\" U2 ON (U0.\"regionid\" = U2.\"id\") WHERE (U0.\"companyid\" = 1 AND U0.\"option\" = (\"campsoffer\".\"activitytype\") AND STContains(U2.\"locations\", (T4.\"position\"))) LIMIT 1) IS NOT NULL) GROUP BY \"campsoffer\".\"id\", (SELECT U0.\"id\" FROM \"campsservicepackage\" U0 INNER JOIN \"campsregion\" U2 ON (U0.\"regionid\" = U2.\"id\") WHERE (U0.\"companyid\" = 1 AND U0.\"option\" = (\"campsoffer\".\"activitytype\") AND STContains(U2.\"locations\", (T4.\"position\"))) LIMIT 1), \"campsprovider\".\"id\" ORDER BY \"camps_offer\".\"created\" ASC"}
{"_id":"q-en-django__django-12532-0b0f09618335a8085dfe289849c4eab0b3097e3b37c84613d890a46b8ffde92a","text":"forms.ModelMultipleChoiceField should use \"invalidlist\" as error message key Description The MultipleChoiceField uses \"invalidlist\", but ModelMultipleChoiceField uses \"list\" as the key for the similar error message."}
{"_id":"q-en-django__django-12553-88e2bac41d9b1f90403c8e42087a50c91c03cdb11380f64c740b166fec140f9c","text":"Increase default password salt size in BasePasswordHasher. Description (last modified by Jon Moroney) I've made a patch for this here ​ Which changes the default salt size from ~71 bits to ~131 bits The rational is that modern guidance suggests a 128 bit minimum on salt sizes OWASP: ​ Python: ​ NIST: ​ In the case of NIST this is technically a hard requirement."}
{"_id":"q-en-django__django-12556-5d70bc9449b4478e6dd7851133b386eae89b5ead996687130d0b59b186530054","text":"Deprecate using getrandomstring without an explicit length Description currently has a default length value (12). I think we should force callers to specify the length value and not count on a default."}
{"_id":"q-en-django__django-12568-d40aa45a340a860d6f3be900bfff6d8df35fa6daf189297bf8c987de5c0f8914","text":"Django humanize's intword filter does not accept negative numbers. Description Django's humanize intword filter does not work with negative numbers. I have created a solution using absolute value. Here is my pull request: ​"}
{"_id":"q-en-django__django-12588-96cdfb16e6f517ccd339ce6e8d88d04974e64f89a765d1af02d421b64ed6c7b4","text":"Add option to removestalecontenttypes to remove entries for nonexistent apps. Description (last modified by Javier Buzzi) Add an option (disabled by default) to removestalecontenttypes command to remove entries also for nonexistent apps. Based on ​discussion. ​PR"}
{"_id":"q-en-django__django-12589-4708a2f37bbd78c5b8f80ad8b7788f1c75247f8db87fcb9a8c5acd7b18686ad9","text":"Django 3.0: \"GROUP BY\" clauses error with tricky field annotation Description Let's pretend that we have next model structure with next model's relations: class A(models.Model): bs = models.ManyToManyField('B', relatedname=\"a\", through=\"AB\") class B(models.Model): pass class AB(models.Model): a = models.ForeignKey(A, ondelete=models.CASCADE, relatedname=\"aba\") b = models.ForeignKey(B, ondelete=models.CASCADE, relatedname=\"abb\") status = models.IntegerField() class C(models.Model): a = models.ForeignKey( A, null=True, blank=True, ondelete=models.SETNULL, relatedname=\"c\", verbosename=(\"a\") ) status = models.IntegerField() Let's try to evaluate next query abquery = (a=OuterRef(\"pk\"), b=1) filterconditions = Q(pk=1) | Q(abab=1) query = A.objects.\\ filter(filterconditions).\\ annotate( status=Subquery((\"status\")), ccount=Count(\"c\"), ) answer = (\"status\").annotate(totalcount=Count(\"status\")) print() print(answer) On Django 3.0.4 we have an error column reference \"status\" is ambiguous and query is next: SELECT (SELECT U0.\"status\" FROM \"testappab\" U0 WHERE (U0.\"aid\" = \"testappa\".\"id\" AND U0.\"bid\" = 1)) AS \"status\", COUNT((SELECT U0.\"status\" FROM \"testappab\" U0 WHERE (U0.\"aid\" = \"testappa\".\"id\" AND U0.\"bid\" = 1))) AS \"totalcount\" FROM \"testappa\" LEFT OUTER JOIN \"testappab\" ON (\"testappa\".\"id\" = \"testappab\".\"aid\") LEFT OUTER JOIN \"testappc\" ON (\"testappa\".\"id\" = \"testappc\".\"aid\") WHERE (\"testappa\".\"id\" = 1 OR \"testappab\".\"bid\" = 1) GROUP BY \"status\" However, Django 2.2.11 processed this query properly with the next query: SELECT (SELECT U0.\"status\" FROM \"testappab\" U0 WHERE (U0.\"aid\" = (\"testappa\".\"id\") AND U0.\"bid\" = 1)) AS \"status\", COUNT((SELECT U0.\"status\" FROM \"testappab\" U0 WHERE (U0.\"aid\" = (\"testappa\".\"id\") AND U0.\"bid\" = 1))) AS \"totalcount\" FROM \"testappa\" LEFT OUTER JOIN \"testappab\" ON (\"testappa\".\"id\" = \"testappab\".\"aid\") LEFT OUTER JOIN \"testappc\" ON (\"testappa\".\"id\" = \"testappc\".\"aid\") WHERE (\"testappa\".\"id\" = 1 OR \"testappab\".\"bid\" = 1) GROUP BY (SELECT U0.\"status\" FROM \"testappab\" U0 WHERE (U0.\"aid\" = (\"testappa\".\"id\") AND U0.\"b_id\" = 1)) so, the difference in \"GROUP BY\" clauses (as DB provider uses \"\", postgresql 11)"}
{"_id":"q-en-django__django-12591-f55d58c6499ce794b5b3fd75c75144dbc7b4d5bb639ee05b8edd3777d974ec89","text":"Can't replace global admin actions with specialized ones per-admin Description landed in 2.2 (discussion in ), which makes it impossible to replace a generic site-wide action (such as the built-in deleteselected) with a new one. It fails with the admin.E130 system check error. We're seeing this with the qsessions app, which has to delete its session objects in non-bulk mode in order to clear caches: ​ (For this particular use case, it seems a fix is to instead override modeladmin.deletequeryset within qsessions's SessionAdmin, as that's what the built-in delete_selected action does per ​ .)"}
{"_id":"q-en-django__django-12613-cd22f8b4e583492974de74ae6b85f67f5f469c276a3df718325ba8529d1d9239","text":"XML serializer doesn't handle JSONFields. Description I have code: data = serializers.serialize(\"xml\", queryset, fields=fields) if I choose specific fields, which are not JSONField, it is ok. But if I choose field, which is JSONField, I receive error File \"\/Users\/ustnv\/PycharmProjects\/fpgnko\/venv\/lib\/python3.6\/site-\", line 128, in serialize s.serialize(queryset, **options) File \"\/Users\/ustnv\/PycharmProjects\/fpgnko\/venv\/lib\/python3.6\/site-\", line 107, in serialize self.handlefield(obj, field) File \"\/Users\/ustnv\/PycharmProjects\/fpgnko\/venv\/lib\/python3.6\/site-\", line 79, in handlefield (field.valuetostring(obj)) File \"\/Users\/ustnv\/PycharmProjects\/fpgnko\/venv\/lib\/python3.6\/site-\", line 25, in characters if content and (r'[\\x00-\\x08\\x0B-\\x0C\\x0E-\\x1F]', content): File \"\", line 182, in search return compile(pattern, flags).search(string) TypeError: expected string or bytes-like object"}
{"_id":"q-en-django__django-12627-eb0e6427d53399c490d8be92973bb07acef7902ae8cb248bfcc462212412905b","text":"makepassword shouldn't accept values other than bytes or string as an argument Description (last modified by iamdavidcz) Currently makepassword function accepts almost every Python object as an argument. This is a strange behaviour and it results directly from forcebytes casting objects to str. We should throw the TypeError when passing anything but bytes or str to makepassword. Reasons: programmers unaware of this strange behaviour can accidentally create weak passwords (potential security issue) other libraries raise the TypeError in the same cases (eg. Werkzeug, passlib) it's inconsistent with the documentation that says: It takes one mandatory argument: the password in plain-text. it's inconsistent with validatepassword behaviour (passing anything but bytes or str to validatepassword raises the TypeError with default settings.AUTHPASSWORDVALIDATORS). Discussion: ​!topic\/django-developers\/1Ap0zDjFa4E"}
{"_id":"q-en-django__django-12630-22c90c95b7659962a6e3b2c14642d0f62e09db2a08293df7dcb6883410fa97ee","text":"Add --check flag to migrate. Description (last modified by thenewguy) It would be helpful if there was a flag for migrate that acted similar to makemigrations --check that could be used to stop CI from deploying an application automatically when unapplied migrations exist. This is different from makemigrations --check because the new command flag would tell us if we need to run any migrations where as makemigrations --check tells us if we need to create any migrations. One currently needs to parse output. It seems like this would be universally useful."}
{"_id":"q-en-django__django-12663-93a567eb72bfe8d854932b12a7646371892158a1589ee48777b2b794cdc180cb","text":"Using SimpleLazyObject with a nested subquery annotation fails. Description (last modified by Jordan Ephron) Prior to it was possible to use a SimpleLazyObject in a queryset as demonstrated below. This new behavior appears to be a regression. Models from import User from import models class A(models.Model): pass class B(models.Model): a = models.ForeignKey(A, ondelete=models.CASCADE) class C(models.Model): owner = models.ForeignKey(User, ondelete=models.CASCADE) TestCase from import User from import OuterRef, Subquery from import TestCase from import SimpleLazyObject from ..models import A, B, C class BugTestCase(TestCase): def testbug(self): owneruser = ( (a=OuterRef(\"pk\")) .annotate(owneruser=Subquery((\"owner\"))) .values(\"owneruser\") ) user = SimpleLazyObject(lambda: User.objects.createuser(\"testuser\")) A.objects.annotate(owneruser=Subquery(owneruser)).filter( owneruser=user ) Sorry for the somewhat arbitrary testcase, hopefully it's sufficient to repro this issue. Results Traceback (most recent call last): File \"\", line 20, in testbug owneruser=user File \"\", line 881, in filter return self.filterorexclude(False, args, kwargs) File \"\", line 899, in filterorexclude (Q(args, kwargs)) File \"\", line 1297, in addq clause, = (qobject, self.usedaliases) File \"\", line 1325, in addq splitsubq=splitsubq, simplecol=simplecol, File \"\", line 1214, in buildfilter condition = self.buildlookup(lookups, reffedexpression, value) File \"\", line 1123, in buildlookup lookup = lookupclass(lhs, rhs) File \"\", line 20, in init = self.getpreplookup() File \"\", line 70, in getpreplookup return () File \"\", line 968, in getprepvalue return int(value) TypeError: int() argument must be a string, a bytes-like object or a number, not 'SimpleLazyObject'"}
{"_id":"q-en-django__django-12669-00d5a33e92fd5c558b1509324f1b207f416d25067e510ff8d3f7ca6de3865275","text":"Add proper field validation to QuerySet.orderby. Description When you annotate a QuerySet with a uuid key, the orderby functionality breaks for the uuid column because the uuid is \"not a valid orderby argument\". Changing the constant by allowing a \"-\" from ORDERPATTERN = re.compile(r'\\?|[-+]?[.\\w]+$') to ORDERPATTERN = re.compile(r'\\?|[-+]?[.-\\w]+$') fixes this in PostgreSQL. Is there a reason the former pattern was used, is it incompatible with other dbs?"}
{"_id":"q-en-django__django-12671-60a8907149654a6f24929f203abc0d7180e14ee308cb92fb6913ab34d71c7775","text":"Allow empty message in management command stdout and stderr proxies. Description Django management commands wrap stdout and stderr in an OutputWrapper that adds a \\n at the end of the text provided as the out argument. I suggest allowing () and () to add a newline to respectively stdout and stderr. Currently, it fails because msg is a positional argument. ​PR"}
{"_id":"q-en-django__django-12700-08d87928d5ead54abd1e81f46ff49169c19d9e10a3d7a8a76131870b8f838762","text":"Settings are cleaned insufficiently. Description Posting publicly after checking with the rest of the security team. I just ran into a case where () would return several un-cleansed values. Looking at cleansesetting() I realized that we ​only take care of s but don't take other types of iterables into account but ​return them as-is. Example: In my I have this: MYSETTING = { \"foo\": \"value\", \"secret\": \"value\", \"token\": \"value\", \"something\": [ {\"foo\": \"value\"}, {\"secret\": \"value\"}, {\"token\": \"value\"}, ], \"else\": [ [ {\"foo\": \"value\"}, {\"secret\": \"value\"}, {\"token\": \"value\"}, ], [ {\"foo\": \"value\"}, {\"secret\": \"value\"}, {\"token\": \"value\"}, ], ] } On Django 3.0 and below: import pprint from import getsafesettings (getsafesettings()[\"MYSETTING\"]) {'else': [[{'foo': 'value'}, {'secret': 'value'}, {'token': 'value'}], [{'foo': 'value'}, {'secret': 'value'}, {'token': 'value'}]], 'foo': 'value', 'secret': '', 'something': [{'foo': 'value'}, {'secret': 'value'}, {'token': 'value'}], 'token': ''} On Django 3.1 and up: from import SafeExceptionReporterFilter import pprint (SafeExceptionReporterFilter().getsafesettings()[\"MYSETTING\"]) {'else': [[{'foo': 'value'}, {'secret': 'value'}, {'token': 'value'}], [{'foo': 'value'}, {'secret': 'value'}, {'token': 'value'}]], 'foo': 'value', 'secret': '', 'something': [{'foo': 'value'}, {'secret': 'value'}, {'token': 'value'}], 'token': ''}"}
{"_id":"q-en-django__django-12708-f931813c27d6b7fe7683fab2a6b0c5209ffd4bd26048eb485508bf9814c82755","text":"Migration crashes deleting an indextogether if there is a uniquetogether on the same fields Description Happens with Django 1.11.10 Steps to reproduce: 1) Create models with 2 fields, add 2 same fields to uniquetogether and to indextogether 2) Delete indextogether -Fail It will fail at , line 378, in deletecomposedindex(), ValueError: Found wrong number (2) of constraints for as this one will find two constraints, the uniq and the idx one. No way to get out of this... The worst in my case is that happened as I wanted to refactor my code to use the \"new\" (Dj 1.11) Options.indexes feature. I am actually not deleting the index, just the way it is declared in my code. I think there are 2 different points here: 1) The deletion of indextogether should be possible alone or made coherent (migrations side?) with uniquetogether 2) Moving the declaration of an index should not result in an index re-creation"}
{"_id":"q-en-django__django-12713-c2a9c1f13af5272c563d96cf72a6c83a388d6111c33e8c8eac8faa31a86aeb65","text":"Allow overridding widget in formfieldformanytomany(). Description (last modified by Mariusz Felisiak) It does not work when I set widget param to function formfieldformanytomany(). This is different from the formfieldforforeignkey() function."}
{"_id":"q-en-django__django-12733-7c8d013c070aed667156531d95c34bc371079a6b12cc8c352a9dc140dfa56b42","text":"Use PostgreSQL TRUNCATE … RESTART IDENTITY keyword to reset sequences in sqlflush() Description Rather than executing an additional query per truncated table, can truncate and reset sequences in a single query by using the RESTART IDENTITY syntax. My project uses the sqlflush() operation internally and profiling shows that it can consume a large percentage of the runtime. Reducing the number of queries to one should help provide a small performance improvement."}
{"_id":"q-en-django__django-12734-bd3fe24266b1c4d130c2eefbc8cc1a536b89247db2b5518a61878daae8b293c9","text":"Migration doesn't detect precision changes in fields that ManyToMany points to. Description In my case was: class Vulnerability(models.Model): cveid = models.CharField(maxlength=15, primarykey=True) app = models.ManyToManyField(AppVersion) class Meta: managed = True Later, i changed cveid maxlength to 100 and did migration: operations = [ migrations.AlterField( modelname='vulnerability', name='cveid', field=models.CharField(maxlength=100, primarykey=True, serialize=False), ), ] In result: cveid field length was changed, but vulnerabilityid field length in table vulnerabilityapp remain unchanged"}
{"_id":"q-en-django__django-12741-c184f517094d5adbc99c4c494ae1fc0299d1256a564a1bb1d10499ffeaccb316","text":"Simplify signature of Description The current signature is: def executesqlflush(self, using, sqllist): The using argument can be dropped and inferred by the calling instance: def executesqlflush(self, sqllist): Some internal ises of this method are already doing: (, sql_flush)"}
{"_id":"q-en-django__django-12747-5f6b2d216113fa6a2be62e6ba13036bf46e21a75d21026d4dc966c04fdcb8bd3","text":"QuerySet.Delete - inconsistent result when zero objects deleted Description The result format of the QuerySet.Delete method is a tuple: (X, Y) X - is the total amount of deleted objects (including foreign key deleted objects) Y - is a dictionary specifying counters of deleted objects for each specific model (the key is the of the model and the value is counter of deleted objects of this model). Example: <class 'tuple': (2, {'myapp.FileAccess': 1, 'myapp.File': 1}) When there are zero objects to delete in total - the result is inconsistent: For models with foreign keys - the result will be: <class 'tuple': (0, {}) For \"simple\" models without foreign key - the result will be: <class 'tuple': (0, {'myapp.BlockLibrary': 0}) I would expect there will be no difference between the two cases: Either both will have the empty dictionary OR both will have dictionary with model-label keys and zero value."}
{"_id":"q-en-django__django-12754-a5621acab2fd53a04ae94059b87c072b0729dc8fd1db493d17deff5eb6cb2bd0","text":"FieldError when migrating field to new model subclass. Description Analogous to . If creating a model subclass and moving a field onto it in the same step, makemigrations works but migrate dies with Local field 'title' in class 'Book' clashes with field of the same name from base class 'Readable'. For example, take this model: from import models class Readable(models.Model): title = models.CharField(maxlength=200) And change to this: from import models class Readable(models.Model): pass class Book(Readable): title = models.CharField(maxlength=200) The migration generates with CreateModel for Book, then RemoveField for But running it produces the error. Reversing the order of the migration operations makes it pass. The auto-detector should be able to use this order."}
{"_id":"q-en-django__django-12774-e5ed6e53cc69100706485433ebc87bd85fbc1a75522d79c12e55bf1241c4f8ac","text":"Allow () for fields with total UniqueConstraints. Description If a field is unique by UniqueConstraint instead of unique=True running inbulk() on that field will fail. Consider: class Article(models.Model): slug = models.CharField(maxlength=255) class Meta: constraints = [ models.UniqueConstraint(fields=[\"slug\"], name=\"%(applabel)s%(class)sslugunq\") ] (fieldname=\"slug\") Traceback (most recent call last): File \"\", line 90, in runcode exec(code, ) File \"<console\", line 1, in <moduleFile \"\/app\/venv\/lib\/python3.8\/site-\", line 82, in managermethod return getattr(self.getqueryset(), name)(args, *kwargs) File \"\/app\/venv\/lib\/python3.8\/site-\", line 680, in inbulk raise ValueError(\"inbulk()'s fieldname must be a unique field but %r isn't.\" % fieldname) ValueError: inbulk()'s field_name must be a unique field but 'slug' isn't. It should be pretty simple to fix this and I have a patch if accepted."}
{"_id":"q-en-django__django-12796-07d934bc8472b671e73c2c01874e73340a3d2515bc7120976765fba39b9aa525","text":"Allow makemigrations to skip database consistency checks Description Currently makemigrations always requires an active database connection, due to it executing loader.checkconsistenthistory() here: ​ Sometimes you are making a change on a machine that doesn't have a database set up or is configured to use a database on a host that isn't resolvable (i.e a postgres host that's only resolvable within docker-compose). If you run makemigrations on such a machine it will fail while attempting to check the migrations consistency, which is quite annoying. The solution seems to either add a backend for such cases or start a database server locally. I'd like to add a flag to skip the consistency check in these situations, or fail gracefully if the connection to the database isn't available."}
{"_id":"q-en-django__django-12803-d394f585641fbd7876622f22d44fc82ed7f081d02edaa725ebec1d1b3aef5d86","text":"ManifestFilesMixin.filehash() returning None get's included in hashed filename as 'None'. Description (last modified by Mariusz Felisiak) When returning a string from a custom ManifestFilesMixin.filehash() implementation, the resulting file name is <filepath.<customhash.<extas expected, whereas returning None results in <filepathNone.<ext. ​Discussion on django-developers supports this behaviour being unintended. Behavior appears to have been introduced with which split the file hashing into a separate method. The following test, when included in the teststorage.TestCollectionManifestStorage test class demonstrates the bug: def testhashednameunchangedwhenfilehashisNone(self): with ('django.contrib.staticfiles.storage.ManifestStaticFilesStorage.filehash', returnvalue=None): self.assertEqual(storage.staticfilesstorage.hashedname(''), '') As suggested by the name of my test, my opinion is that the correct behaviour should be that if filehash returns None, then no hash is inserted into the filename and it therefore remains unchanged. With that in mind, a possible solution is to change the following lines in the hashedname() method (~line 100 in staticfiles.storage): if filehash is not None: filehash = \".%s\" % filehash hashedname = (path, \"%s%s%s\" % (root, filehash, ext)) to if filehash is None: filehash = \"\" else: filehash = \".%s\" % filehash hashedname = (path, \"%s%s%s\" % (root, file_hash, ext))"}
{"_id":"q-en-django__django-12821-79e1488db9a3c82a55151ecce060e8c404a15ecdb868edb7f88dfe6cf3649c15","text":"Stop minifying only some admin static assets Description Here is a list of JavaScript files in the admin app and their size: 20K 15K 13K 8.8K 7.6K 6.7K 5.9K 5.4K 5.3K 5.2K 3.2K 1.9K 1.5K 1.1K 911 878 674 569 495 379 363 Some things to notice: Only 4 out of 17 files are minified. The largest file, , is only 20 KB. The largest file is also not included in the list of minified files. All uncompressed files are smaller than the size of the 3 font assets, each ~80 KB. All uncompressed files are smaller than the minified jQuery, ~87 KB. I'm not sure if there is a deliberate or historical reason that only a fraction of the static assets are minified, but it looks like it could be an oversight. The approach is inconsistent. Minifying is step a contributor must manually do. The documentation for doing so is here: ​ This is a step that is easy to forget, myself included. Whether or not one remembers to compress static assets will also affect the outcome of tests. I suggest we drop the minificaiton of admin files altogether. For such small files, it doesn't seem worth it to add a build step and inconsistently at that. In a typical production scenarios, the static assets will be cached and possibly compressed. Compressing static assets largely removes the size gains of minification. Additionally, there are third party apps to fill the role of static asset optimization. I think we should continue to distribute the vendored libraries minified, however as they are not manually handled during typical contributions."}
{"_id":"q-en-django__django-12830-f2037851818b8e4b6817745043627828d2313eb2c6f5a6c10401b67d529a81ea","text":"Add an absolutemax parameter to formsetfactory Description The documentation at ​ seems to indicate (if I understood it correctly) that the purpose of the maxnum parameter is to prevent that someone sends a manipulated, excessively large value in the hidden form field that states the number of (extra) forms that are submitted, whereas it is not (directly) related to the number of forms that are actually POSTed, or initialized via parameter initials. However, following the example at that page, with MyInitials being a list of e.g. 1500 initial values and request.POST containing more than 1500 formsets: ArticleFormSet = formsetfactory(ArticleForm, extra=0) formset1 = ArticleFormSet(initial=MyInitials) formset2 = ArticleFormSet(request.POST) Now, accessing [1000] throws an IndexError exception. The maxnum is at its default value of 1000, but in the above context, it is not expected that formset1 or formset2 is reduced to maxnum forms -- I'd have expected each to have the full number of forms as initialized. Related thread at django-users: ​"}
{"_id":"q-en-django__django-12851-60edf49d8b92869d47b8bce37509091cd9c77fefa2758c1c9824799464bdf2c4","text":"Remove ifequal from the template language. Description No modern project uses ifequal. No one recommends it. I argue it is taking up valuable bytes in the project. Let's remove it."}
{"_id":"q-en-django__django-12855-eb83ceef1c3eac3733873886139837b98aa112cfae5f9068974a8a9f70ce7cbe","text":"Deprecate (). Description The docs for ​ say: This function is an alias to (). It’s likely to be deprecated in a future release. It looks like the change was made in this ​commit back in 2016 (Django 2.0). Given some years have passed, is it now the time to deprecate this function?"}
{"_id":"q-en-django__django-12856-8b8bf072c3037bec0d4c89cebf9caa56d890ed70d88e2751d76e1e1acc142138","text":"Add check for fields of UniqueConstraints. Description (last modified by Marnanel Thurman) When a model gains a UniqueConstraint, makemigrations doesn't check that the fields named therein actually exist. This is in contrast to the older uniquetogether syntax, which raises models.E012 if the fields don't exist. In the attached demonstration, you'll need to uncomment \"withuniquetogether\" in in order to show that uniquetogether raises E012."}
{"_id":"q-en-django__django-12858-9b325e37aced12c5f88855602bb97cd80789b3e837dcbce228970f118b830d47","text":"models.E015 is raised when ordering uses lookups that are not transforms. Description check SystemCheckError: System check identified some issues: ERRORS: app.Stock: (models.E015) 'ordering' refers to the nonexistent field, related field, or lookup 'supplyproductparentisnull'. However this ordering works fine: list(Stock.objects.orderby('supplyproductparentisnull').valueslist('pk', flat=True)[:5]) [1292, 1293, 1300, 1295, 1294] list(Stock.objects.orderby('-supplyproductparentisnull').valueslist('pk', flat=True)[:5]) [108, 109, 110, 23, 107] I believe it was fine until was implemented. is a foreign key to Supply, Supply.product is a foreign key to Product, is a ForeignKey('self', models.CASCADE, null=True)"}
{"_id":"q-en-django__django-12869-117aefd0fd3f62305e92984ac02b153f54802e63d30c9a8d2c3519ffb088efc5","text":"collectstatic doesn't run staticfiles checks. Description Piecing this together from a user who is having trouble with staticfiles (who doesn't, at least once) on IRC. the system checks framework has a checkfinders which is invoked if django.contrib.staticfiles is in your INSTALLEDAPPS which defers checks to each individually configured finder as () - this accounts for running the following line: if not isinstance(settings.STATICFILESDIRS, (list, tuple)): which is intended to catch the problem scenario of STATICFILESDIRS = () being edited to STATICFILESDIRS = (\"mypath\") with no trailing comma to make it a tuple, rather than an iterable string. However, the collectstatic management command has requiressystemchecks = False so it appears possible to edit the value and directly run python collectstatic without the type being checked or the error raised. Naively, I'm assuming that something like the following needs to occur: for finder in getfinders(): () # ignoring NotImplementedError # then somehow surface any errors back to stderr as if I've not delved deeply into the \"surface errors\" part."}
{"_id":"q-en-django__django-12906-03e24e63b39474e258eb2b48cc6aca29698b43d800dcb7a5c969ed78ac8948eb","text":"Document admin's requirement on django.template.contextprocessors.request context processor. Description Since commit , the admin templates use the implied request variable normally by django.template.contextprocessors.request. As Django templates silence errors, this went unnoticed during testing, and won't immediately break the templates, but certain expected rendering features won't work. Django should document this change: In the release notes (provide a deprecation period where it is a warning only) In the admin docs In the system check framework as a warning (but eventually an error)"}
{"_id":"q-en-django__django-12908-b6afb4444f5c6477dd5799429dba08859f09d6b4429d2d3a7f0bd9a7644f1132","text":"Union queryset should raise on distinct(). Description (last modified by Sielc Technologies) After using .annotate() on 2 different querysets and then .union() .distinct() will not affect the queryset def setUp(self) -None: user = self.getorcreateadminuser() (user, name=\"Sam1\") (user, name=\"Sam2 acid\") (user, name=\"Sam3\") (user, name=\"Sam4 acid\") (user, name=\"Dub\") (user, name=\"Dub\") (user, name=\"Dub\") = user def testunionannotateddiffdistinct(self): qs = () qs1 = (name='Dub').annotate(rank=Value(0, IntegerField())) qs2 = (name='Sam1').annotate(rank=Value(1, IntegerField())) qs = (qs2) qs = qs.orderby('name').distinct('name') # THIS DISTINCT DOESN'T WORK self.assertEqual((), 2) expected to get wrapped union SELECT DISTINCT ON () * FROM (SELECT ... UNION SELECT ...) AS siebox_sample"}
{"_id":"q-en-django__django-12912-637f930e73beb5652cce796f04e9e8d80e5a1eec76c1093673ee3bf99e9f7e39","text":"Alias used in aggregate filtering is incorrect. Description With the following queryset: IndicatorValue.objects .values(\"freight\") .annotate( loadingtime=Min(\"datetime\", filter=Q(type=IndicatorValue.TYPEFREIGHTCREATED)) - Max(\"datetime\", filter=Q(type=IndicatorValue.TYPEFREIGHTCOMPLETED)), hastoploading=Exists(OrderItemResult.objects.orderby().filter(orderlineorderfreight=OuterRef(\"freight\"), loadingarmloadingtype=LoadingArm.LOADINGTYPETOP, ).values('pk')), hasbottomloading=Exists(OrderItemResult.objects.orderby().filter(orderlineorderfreight=OuterRef(\"freight\"), loadingarmloadingtype=LoadingArm.LOADINGTYPEBOTTOM, ).values('pk')) ) .aggregate( topmin=Min(\"loadingtime\", filter=Q(hastoploading=True, hasbottomloading=False)) ) I get the following SQL generated for the aggregate (notice that both alias used are the same in the SQL, whereas they are not in the queryset): MIN(\"loadingtime\") FILTER (WHERE (\"hastoploading\" = false AND \"hastoploading\" = true)) The full SQL generated is: SELECT MIN(\"loadingtime\") FILTER (WHERE (\"hastoploading\" = false AND \"hastoploading\" = true)) FROM (SELECT \"indicatorsindicatorvalue\".\"freightid\" AS Col1, (MIN(\"indicatorsindicatorvalue\".\"datetime\") FILTER (WHERE \"indicatorsindicatorvalue\".\"type\" = \\'freightcreated\\') - MAX(\"indicatorsindicatorvalue\".\"datetime\") FILTER (WHERE \"indicatorsindicatorvalue\".\"type\" = \\'freightcompleted\\')) AS \"loadingtime\", EXISTS(SELECT U0.\"id\" FROM \"ordersorderitemresult\" U0 INNER JOIN \"loadingterminalloadingarm\" U1 ON (U0.\"loadingarmid\" = U1.\"id\") INNER JOIN \"ordersorderitem\" U2 ON (U0.\"orderlineid\" = U2.\"id\") INNER JOIN \"ordersorder\" U3 ON (U2.\"orderid\" = U3.\"id\") WHERE (U1.\"loadingtype\" = \\'TOP\\' AND U3.\"freightid\" = \"indicatorsindicatorvalue\".\"freightid\")) AS \"hastoploading\", EXISTS(SELECT U0.\"id\" FROM \"ordersorderitemresult\" U0 INNER JOIN \"loadingterminalloadingarm\" U1 ON (U0.\"loadingarmid\" = U1.\"id\") INNER JOIN \"ordersorderitem\" U2 ON (U0.\"orderlineid\" = U2.\"id\") INNER JOIN \"ordersorder\" U3 ON (U2.\"orderid\" = U3.\"id\") WHERE (U1.\"loadingtype\" = \\'BOTTOM\\' AND U3.\"freightid\" = \"indicatorsindicatorvalue\".\"freightid\")) AS \"hasbottomloading\" FROM \"indicatorsindicatorvalue\" WHERE \"indicatorsindicatorvalue\".\"deleted\" IS NULL GROUP BY \"indicatorsindicatorvalue\".\"freightid\", \"hastoploading\", \"hasbottomloading\") subquery It works fine with Django 2.2 (which does not use alias there if I'm not mistaken)."}
{"_id":"q-en-django__django-12915-4e2d16baa3d55e302d70d3cd5f0c189cb5dfb75ee87e164d6f64eaf3d0d19707","text":"Add getresponseasync for ASGIStaticFilesHandler Description It looks like the StaticFilesHandlerMixin is missing the the async response function. Without this, when trying to use the ASGIStaticFilesHandler, this is the traceback: Exception inside application: 'NoneType' object is not callable Traceback (most recent call last): File \"...\/lib\/python3.7\/site-\", line 30, in asgi await (scope, receive, send) File \"\", line 86, in call return await super().call(scope, receive, send) File \"\", line 161, in call response = await self.getresponseasync(request) File \"\", line 148, in getresponseasync response = await self.middlewarechain(request) TypeError: 'NoneType' object is not callable"}
{"_id":"q-en-django__django-12928-6e078be5b3a953318b9de81366853d88b92a4de73d31248deb52f1e5510703dc","text":"Implement autoreload behaviour for cached template loader Description It would be nice to be able get the speed benefit of the cached template loader during development, but without the downside of having to restart the server every time you change a template. It turns out it's possible with just a few changes. Because it's not really possible to configure the cached template loader I did have to base this patch on the fix for . Enabling autoreloading for the cached template loader would work like this: TEMPLATES = [{ 'BACKEND': '', 'DIRS': [(BASEDIR, 'templates')], 'APPDIRS': True 'OPTIONS': { 'cache_templates': True, 'autoreload': DEBUG } }]"}
{"_id":"q-en-django__django-12933-5f392b49608d83ee2f280419d0e3c8ffebf574fdfcff836ed9f8c27d8033fbb7","text":"Raise CommandError when clearsessions is called on a backend not implementing clearexpired() Description Formal ticket for the issue raised by François Freitag in ​PR. clearsessions writes to stderr directly when a backend has not implemented clearexpired(). François notes that the recommended behavior is to raise CommandError."}
{"_id":"q-en-django__django-12936-a21d6e50ea523f8d3c08040f316efce6c7a4c005bc6b965fa7654d1716674263","text":"() doesn't work properly for resolved subqueries. Description () doesn't work properly for resolved subqueries, basically all resolved subqueries are now equal. Regression in ."}
{"_id":"q-en-django__django-12951-02712ccda2882d57af813ea29b6a140ad02482bd71d55575497c9d82b82ba5f3","text":"Cannot mix Exists expression with keyword arguments to When Description (last modified by Ryan Heard) I don't seem to be able to provide an Exists expression to When alongside keyword arguments like you can with filter. For instance, consider: class State(models.Model): pass class County(models.Model): name = CharField(maxlength=50) state = ForeignKey(State, relatedname='counties') I can execute the following query just fine: ( Exists((counties=OuterRef('pk'), name=\"Texas\")), name=\"Dallas\", ) But a similar query using When does not work: County.objects.annotate( status=Case( When(Exists((counties=OuterRef('pk'), name=\"Texas\")), name=\"Dallas\", then=Value(\"DALLAS COUNTY\")), default=Value(\"ELSEWHERE\"), )) TypeError: When() supports a Q object, a boolean expression, or lookups as a condition. Instead the arguments must be wrapped in a Q object: County.objects.annotate( status=Case( When(Q(Exists((counties=OuterRef('pk'), name=\"Texas\")), name=\"Dallas\"), then=Value(\"DALLAS COUNTY\")), default=Value(\"ELSEWHERE\"), )) This is inconvenient and inconsistent with how filter works, as shown. When's init method can be modified to allow similar input as filter. ​Code is in a branch in my repo, but as this is my first time contributing to Django, I want to make sure I open a ticket and get feedback first. Also I wasn't sure how to classify this. I wasn't sure if it was a bug, as I wasn't sure if it was designed this way."}
{"_id":"q-en-django__django-12953-717a9adf544a60d5409b9cbb9ef87927e7d77afb1d2c77a01c51f94efbf7b295","text":"Check that and UniqueConstraint.condition don't span joins. Description (last modified by Mariusz Felisiak) Similar to but for check and condition. Not everyone is familiar with the fact database level constraint cannot span across tables and might be tempted to do class Person(models.Model): age = models.PositiveSmallIntegerField() parent = models.ForeignKey(self) class Meta: constraints = { CheckConstraint( name='ageltparent', check=Q(agelt=parentage) ), } Which we'll happily create migrations for but we'll then crash because we prevent JOINs when resolving check."}
{"_id":"q-en-django__django-12957-a4ac0a939d3498dba1542588dc56d91f971c0957dc2b4bc64431407e6f8cbee4","text":"Geodjango js template should use for float values to avoid DECIMALSEPARATOR ruin the js syntax Description should use |safe on float values to avoid DECIMALSEPARATOR (and probably other settings in this category) ruin the js syntax by adding unexpected characters instead of dot."}
{"_id":"q-en-django__django-12961-5e77ccc0f9fa61da9a39fe7cdffdb1266d83962f0e38c41b90e646f38d7017de","text":"orderby() with expressions crashes on union() querysets. Description (last modified by Mariusz Felisiak) I have read the recent tickets about unions and orderby (, , ) , and my bug is slightly different, so I hope it's not a duplicate. Let's consider two similar models: class EntityA(models.Model): namea = models.CharField(maxlength=128, null=True) dta = models.DateTimeField(null=True) class EntityB(models.Model): nameb = models.CharField(maxlength=128, null=True) dtb = models.DateTimeField(null=True) (namea=\"a\") (namea=\"qwerty\", ()) (nameb=\"random\", ()) (nameb=\"b\") qsa = (name=F(\"namea\"), dt=F(\"dta\")) qsb = (name=F(\"nameb\"), dt=F(\"dtb\")) queryset = (qsb) I can use a simple ORDER BY clause: queryset.orderby(\"-dt\") And everything will work, no problem here. What I actually want is the same query, but with a NULLS LAST Usually the query becomes: queryset.orderby(F(\"dt\").desc(nulls_last=True)) but that raises a DatabaseError: ORDER BY term does not match any column in the result set. I know unions can handle only a few clauses, but ORDER BY is one of them, so I'm unsure whether this is the expected behaviour or not. If it's expected, then the raised exception could be more explicit."}
{"_id":"q-en-django__django-12965-83b4119b78172f3a00ab45e97f909e728c0591bbd24611024a19a7fa56f0b8b5","text":"().delete() subquery usage performance regression Description Lock tests are failing with Django-MySQL on Django 3.1: ​ . The tests run ().delete(). Django 3.0 generates this SQL: DELETE FROM Django 3.1 generates this SQL: DELETE FROM WHERE . IN (SELECT . FROM ) The subquery is a blocker for using LOCK TABLES along with delete() - as per ​the mysql docs: You cannot refer to a locked table multiple times in a single query using the same name. Use aliases instead, and obtain a separate lock for the table and each alias: mysqlLOCK TABLE t WRITE, t AS t1 READ; mysqlINSERT INTO t SELECT FROM t; ERROR 1100: Table 't' was not locked with LOCK TABLES mysqlINSERT INTO t SELECT FROM t AS t1; Since there's no alias on the subquery, there's no way to lock it. Additionally this change is a performance regression. I benchmarked with MariaDB 10.3 and filling an InnoDB a table of 100k rows via the [sequence storage engine ​]. Using the old DELETE FROM, deletion takes 0.2 seconds: root [19]create table t(c int primary key); Query OK, 0 rows affected (0.079 sec) root [16]insert into t select from seq1to100000; Query OK, rows affected (0.252 sec) Records: Duplicates: 0 Warnings: 0 root [17]delete from t; Query OK, rows affected (0.200 sec) Using DELETE FROM WHERE id IN (...), deletion takes 7.5 seconds: root [18]drop table t; Query OK, 0 rows affected (0.008 sec) root [19]create table t(c int primary key); Query OK, 0 rows affected (0.079 sec) root [20]insert into t select from seq1to_100000; Query OK, rows affected (0.594 sec) Records: Duplicates: 0 Warnings: 0 root [21]delete from t where c in (select c from t); Query OK, rows affected (7.543 sec) root [22]drop table t; Query OK, 0 rows affected (0.013 sec) Yes in theory the optimizer should be able to find this, and it may even be fixed in later MySQL\/MariaDB versions. But I think we shouldn't change the SQL here."}
{"_id":"q-en-django__django-12973-6d088ee971245d332162df8cf90c7d3f401b1f774c156c298c95b728af71e12f","text":"Add hint to the E410 about AuthenticationMiddleware. Description Given an empty MIDDLEWARE in your project, you try and run the admin. The system checks framework kicks in, giving you: (admin.E408) '' must be in MIDDLEWARE in order to use the admin application. (admin.E409) 'django.contrib.messages.middleware.MessageMiddleware' must be in MIDDLEWARE in order to use the admin application. (admin.E410) 'django.contrib.sessions.middleware.SessionMiddleware' must be in MIDDLEWARE in order to use the admin application. You add each of those middlewares to the stack, and run the application: File \"\", line 23, in processrequest ) % (\"CLASSES\" if settings.MIDDLEWARE is None else \"\") AssertionError: The Django authentication middleware requires session middleware to be installed. Edit your MIDDLEWARE setting to insert 'django.contrib.sessions.middleware.SessionMiddleware' before '' Ignoring the fact that it should be a system check, and it's an unfortunate side-effect of spanning 2 separate contrib apps that it isn't, if the order of the errors output was updated, we'd not see this exception in this specific scenario."}
{"_id":"q-en-django__django-12983-f54081b6cf78a19c161f9beb2802b86d4499e4746d85e7b461b43d05396fedb7","text":"Make () strip dashes and underscores Description (last modified by Elinaldo do Nascimento Monteiro) Bug generation slug Example: from import text text.slugify(\"This is a test ---\") output: this-is-a-test- Improvement after correction from import text text.slugify(\"_This is a test ---\") output: this-is-a-test ​PR"}
{"_id":"q-en-django__django-13012-4594053687327831e7365a4afaea2743452293f54c802bc693711158e1d9ec46","text":"Constant expressions of an ExpressionWrapper object are incorrectly placed at the GROUP BY clause Description I have a function that expects an arbitrary Query expression and constructs a query on a Postgres db def execQuery(expr): expr = ExpressionWrapper(expr, outputfield=IntegerField()) return Model.objects.annotate(exprres=expr).values('exprres', 'columna').annotate(sum=Sum('columnb')) However, when the given expr is a constant expression (e.g., Value(3)), Django generates an SQL query that contains this constant expression in its GROUP BY clause. SELECT \"model\".\"columna\", 3 AS \"exprres\", SUM(\"model\".\"columnb\") AS \"sum\" FROM \"model\" GROUP BY \"model\".\"columna\", 3 This leads to an exception because in Postgres, the query above is invalid: aggregate functions are not allowed in GROUP BY LINE 1: SELECT \"model\".\"columna\", 3 AS \"exprres\", SUM(\"model\".\"col... Note that when the given query expression is not wrapped by the ExpressionWrapper object, Django correctly identifies and omits the constant from the GROUP BY clause. For example, the query below runs correctly. def execQuery(expr): return Model.objects.annotate(exprres=Value(3, outputfield=IntegerField())).values('exprres', 'columna').annotate(sum=Sum('columnb')) SELECT \"model\".\"columna\", 3 AS \"exprres\", SUM(\"model\".\"columnb\") AS \"sum\" FROM \"model\" GROUP BY \"model\".\"columna\""}
{"_id":"q-en-django__django-13022-cb47682705a68ad2797c54b6387b7aaaf1d2e772eaf80ba6833cf570ad69adda","text":"Memcached key validation raises InvalidCacheKey with clunky message. Description On Django 2.2.13 the code for memcachekeywarnings in has a bad format string that results in raising an exception rather than just producing a warning. This can be reproduced with a memcached key with a space in it, e.g. \"foo bar\". This code was present before the 2.2.13 release, but becomes more exposed with that release, since it begins validating cache keys. I think it's as simple as removing the , CacheKeyWarning."}
{"_id":"q-en-django__django-13023-bade7321a4af63baac789057d7b6315f20941e838a14453dcd7f7736c940f2b7","text":"DecimalField.topython() raises TypeError on dict values. Description A call to DecimalField.topython() with a dictionary as the value parameter produces TypeError instead of ValidationError. This is a problem, for example, when you try to save a model object, and a decimal field got set to a dictionary by mistake. The TypeError exception that comes back makes it hard to track the problem to the field if the object has a lot of fields. I am proposing a patch to fix it: ​"}
{"_id":"q-en-django__django-13028-d3ff9b653075cd11545b47907738df7a0b5b07e96e9153b25cdfb012513d1cf4","text":"Queryset raises NotSupportedError when RHS has filterable=False attribute. Description (last modified by Nicolas Baccelli) I'm migrating my app to django 3.0.7 and I hit a strange behavior using a model class with a field labeled filterable class ProductMetaDataType(models.Model): label = models.CharField(maxlength=255, unique=True, blank=False, null=False) filterable = models.BooleanField(default=False, verbosename=(\"filterable\")) class Meta: applabel = \"adminpricing\" verbosename = (\"product meta data type\") verbosenameplural = (\"product meta data types\") def str(self): return class ProductMetaData(models.Model): id = models.BigAutoField(primarykey=True) product = models.ForeignKey( Produit, null=False, blank=False, ondelete=models.CASCADE ) value = models.TextField(null=False, blank=False) marketplace = models.ForeignKey( Plateforme, null=False, blank=False, ondelete=models.CASCADE ) datecreated = models.DateTimeField(null=True, ) metadatatype = models.ForeignKey( ProductMetaDataType, null=False, blank=False, ondelete=models.CASCADE ) class Meta: applabel = \"adminpricing\" verbosename = (\"product meta data\") verbosenameplural = (\"product meta datas\") Error happened when filtering ProductMetaData with a metadatatype : (value=\"Dark Vador\", metadatatype=self.brandmetadata) Error traceback : Traceback (most recent call last): File \"\", line 481, in testcheckpolicywarningbyfields for p in ( File \"\/usr\/local\/lib\/python3.8\/site-\", line 82, in managermethod return getattr(self.getqueryset(), name)(args, kwargs) File \"\/usr\/local\/lib\/python3.8\/site-\", line 904, in filter return self.filterorexclude(False, args, kwargs) File \"\/usr\/local\/lib\/python3.8\/site-\", line 923, in filterorexclude (Q(args, *kwargs)) File \"\/usr\/local\/lib\/python3.8\/site-\", line 1351, in addq clause, = (qobject, self.usedaliases) File \"\/usr\/local\/lib\/python3.8\/site-\", line 1378, in addq childclause, neededinner = self.buildfilter( File \"\/usr\/local\/lib\/python3.8\/site-\", line 1264, in buildfilter self.checkfilterable(value) File \"\/usr\/local\/lib\/python3.8\/site-\", line 1131, in checkfilterable raise NotSupportedError( ProductMetaDataType is disallowed in the filter clause. I changed label to filterabletest and it fixed this issue This should be documented or fix."}
{"_id":"q-en-django__django-13030-82444958496f6879ebe101b157c0367c10389e1658d9c3e8a9632bbc1df44f07","text":"Avoid passing NULL to the IN lookup Description Currently prefetchrelated on a FK passes the NULL through to the database for e.g. authorid IN (NULL, 2). Passing NULL is always unnecessary, since it's not allowed in FK's. There's a small risk from passing NULL that it could lead to incorrect with complex prefetch querysets using PK refs because of NULL's weirdness in SQL. For example with these models: from import models class Author(models.Model): pass class Book(models.Model): author = models.ForeignKey(Author, null=True, ondelete=models.DONOTHING) Prefetching authors on Books, when at least one Book has author=None, uses IN (..., NULL, ...) in the query: In [1]: from import Author, Book In [2]: a1 = () In [3]: (author=a1) Out[3]: <Book: Book object (3)In [4]: (author=None) Out[4]: <Book: Book object (4)In [5]: Book.objects.prefetchrelated('author') Out[5]: <QuerySet [<Book: Book object (3), <Book: Book object (4)]In [6]: from import connection In [7]: connection.queries Out[7]: [{'sql': 'INSERT INTO \"coreauthor\" (\"id\") VALUES (NULL)', 'time': '0.001'}, {'sql': 'INSERT INTO \"corebook\" (\"authorid\") VALUES (2)', 'time': '0.001'}, {'sql': 'INSERT INTO \"corebook\" (\"authorid\") VALUES (NULL)', 'time': '0.001'}, {'sql': 'SELECT \"corebook\".\"id\", \"corebook\".\"authorid\" FROM \"corebook\" LIMIT 21', 'time': '0.000'}, {'sql': 'SELECT \"coreauthor\".\"id\" FROM \"coreauthor\" WHERE \"coreauthor\".\"id\" IN (NULL, 2)', 'time': '0.000'}] Maybe this could generally be extended to use of _in with non-nullable fields?"}
{"_id":"q-en-django__django-13032-02a6ce2133d4bfe8eed8960a1d2e248ea98be79132ef41fc980316ac94040b28","text":"Improve makemigrations warning message when calling without an active database connection. Description I was looking at the gis install instructions and I came across an error when running makemigrations. (Error is I had not entered password correctly). However, the error message that is generated is a bit odd, it has a full stop on a new line and shows ( I was also able to get the same error message on a clean project, with a postgres database (not gis) and an incorrect password. I'm not sure if this is a 'bug' but it doesn't look quite right? (gis) PS C:\\Users\\smith\\gis\\geodjangopython .\\ makemigrations c:\\users\\smith\\pycharmprojects\\django2\\django\\core\\management\\commands\\ RuntimeWarning: Got an error checking a consistent migration history performed for database connection 'default': fe_sendauth: no password supplied . ( No changes detected (gis) PS C:\\Users\\smith\\gis\\geodjango"}
{"_id":"q-en-django__django-13033-ba92eb7d037e83e5275b41c1913533f8c1a4d64923a86960fc89b4570e3e0339","text":"Self referencing foreign key doesn't correctly order by a relation \"id\" field. Description Initially discovered on 2.2.10 but verified still happens on 3.0.6. Given the following models: class OneModel(models.Model): class Meta: ordering = (\"-id\",) id = models.BigAutoField(primarykey=True) root = models.ForeignKey(\"OneModel\", ondelete=models.CASCADE, null=True) oneval = models.BigIntegerField(null=True) class TwoModel(models.Model): id = models.BigAutoField(primarykey=True) record = models.ForeignKey(OneModel, ondelete=models.CASCADE) twoval = models.BigIntegerField(null=True) The following queryset gives unexpected results and appears to be an incorrect SQL query: qs = (recordonevalin=[1,2,3]) qs = qs.orderby(\"recordrootid\") print() SELECT \"oriontwomodel\".\"id\", \"oriontwomodel\".\"recordid\", \"oriontwomodel\".\"twoval\" FROM \"oriontwomodel\" INNER JOIN \"oriononemodel\" ON (\"oriontwomodel\".\"recordid\" = \"oriononemodel\".\"id\") LEFT OUTER JOIN \"oriononemodel\" T3 ON (\"oriononemodel\".\"rootid\" = T3.\"id\") WHERE \"oriononemodel\".\"oneval\" IN (1, 2, 3) ORDER BY T3.\"id\" DESC The query has an unexpected DESCENDING sort. That appears to come from the default sort order on the OneModel class, but I would expect the orderby() to take prececence. The the query has two JOINS, which is unnecessary. It appears that, since is a foreign key to itself, that is causing it to do the unnecessary extra join. In fact, testing a model where root is a foreign key to a third model doesn't show the problem behavior. Note also that the queryset with orderby(\"recordroot\") gives the exact same SQL. This queryset gives correct results and what looks like a pretty optimal SQL: qs = (recordonevalin=[1,2,3]) qs = qs.orderby(\"recordrootid\") print() SELECT \"oriontwomodel\".\"id\", \"oriontwomodel\".\"recordid\", \"oriontwomodel\".\"twoval\" FROM \"oriontwomodel\" INNER JOIN \"oriononemodel\" ON (\"oriontwomodel\".\"recordid\" = \"oriononemodel\".\"id\") WHERE \"oriononemodel\".\"oneval\" IN (1, 2, 3) ORDER BY \"oriononemodel\".\"rootid\" ASC So is this a potential bug or a misunderstanding on my part? Another queryset that works around the issue and gives a reasonable SQL query and expected results: qs = (recordonevalin=[1,2,3]) qs = qs.annotate(rootid=F(\"recordrootid\")) qs = qs.orderby(\"rootid\") print() SELECT \"oriontwomodel\".\"id\", \"oriontwomodel\".\"recordid\", \"oriontwomodel\".\"twoval\" FROM \"oriontwomodel\" INNER JOIN \"oriononemodel\" ON (\"oriontwomodel\".\"recordid\" = \"oriononemodel\".\"id\") WHERE \"oriononemodel\".\"oneval\" IN (1, 2, 3) ORDER BY \"oriononemodel\".\"zeroid\" ASC ASCENDING sort, and a single INNER JOIN, as I'd expect. That actually works for my use because I need that output column anyway. One final oddity; with the original queryset but the inverted sort orderby(): qs = (recordonevalin=[1,2,3]) qs = qs.orderby(\"-recordrootid\") print() SELECT \"oriontwomodel\".\"id\", \"oriontwomodel\".\"recordid\", \"oriontwomodel\".\"twoval\" FROM \"oriontwomodel\" INNER JOIN \"oriononemodel\" ON (\"oriontwomodel\".\"recordid\" = \"oriononemodel\".\"id\") LEFT OUTER JOIN \"oriononemodel\" T3 ON (\"oriononemodel\".\"rootid\" = T3.\"id\") WHERE \"oriononemodel\".\"oneval\" IN (1, 2, 3) ORDER BY T3.\"id\" ASC One gets the query with the two JOINs but an ASCENDING sort order. I was not under the impression that sort orders are somehow relative to the class level sort order, eg: does specifing orderby(\"-record_rootid\") invert the class sort order? Testing that on a simple case doesn't show that behavior at all. Thanks for any assistance and clarification."}
{"_id":"q-en-django__django-13066-a4277132e4cfd03d27435815d62e9cf9f4a04dee7dcbda6f6152a2879f9aad39","text":"Argon2id should be supported and become the default variety for Argon2PasswordHasher Description (last modified by Si Feng) There were three important changes in the upstream ​argon2-cffi library since Django 1.10 was released with Argon2 support: (Nov 10, 2016) argon2id support was : ​ (Apr 9, 2018) argon2id became its default type: ​ (Aug 18, 2018) its hasher's default memory cost changed from 512 to , and parallelism from 2 to 8, per ​RFC draft recommendations: ​ When Django 1.10 was released, only argon2d and argon2i were available, hence the hard-coded argon2i variety in Argon2PasswordHasher. Though Django does not use its hasher, the memorycost = 512 and parallelism = 2 in Argon2PasswordHasher were simply copied from argon2-cffi's hasher back then. Now we should sync Django with upstream updates."}
{"_id":"q-en-django__django-13077-89584949312769b69ec1c082fad6420311b3dd5b213ac3a3a94731952321ee9a","text":"django-admin runserver mostly does not work on Windows Description I frequently encounter the problem that running \"django-admin runserver\" on Windows fails: <some-venv\\scripts\\ can't open file '<some-venv\\test\\Scripts\\django-admin': [Errno 2] No such file or directory The command works if run with --noreload. There are probably other conditions that must be met for it to fail, or work. I think it makes a difference, for example, if Django was installed with pip or setuptools. The .exe \"console scripts \" are different between the two; one type has a bit of readable Python at the end, the other does not. What I do know is that the problem is either caused or enabled by the generated console scripts (the .exe wrappers in venv\/Scripts) stripping the .exe extension from their own [0] when passing that to Python to run. If I sabotage this by changing the regex in the file itself (where present), or by adding the extension again in before running the command line, it works fine. To be honest, I have no idea whether this is a bug in Django, or pip, or setuptools, or distutils, or something else entirely. I have spent the last two days trying to figure out what exactly goes wrong, and where, with little success. I'm reporting it here because Django is the only place I could find where it is possible to kludge the code into working. I'll be happy to take my business elsewhere if someone can point the way."}
{"_id":"q-en-django__django-13085-40e0653792c4d31caf6a67256a82089218b5a88b8a0b023623e4dbe25e17336c","text":"compilemessages needlessly runs msgfmt on unchanged .po files Description I have a project where running django-admin compilemessages takes 1.75 seconds. Running it again, when all the .mo files already exists and are up-to-date, also takes 1.75 seconds. I propose that is changed so that it only invokes msgfmt when it would do anything useful. This can be implemented by checking the mtime of the .po file and the corresponding .mo file. (If statting the .mo file fails, treat that as if the mtime was 0.) Only submit the command to the executor if the mtime of the .po file is greater than that of the .mo file. In effect: don't do anything if the .mo file is newer than the .po file. There is one issue with this: the way the code currently uses the iswritable function. Since it modifies the mtime of the .mo file, you would have to perform the stat of the .mo file before you check if it is writable. (Or, you could just remove the iswritable function and its use. That feature is, in my opinion, of dubious value, and it doesn't appear to be documented.) After I made the changes above, the runtime in the common case where nothing needs to be done was reduced from 1.75 seconds to 0.2 seconds. (Unfortunately, I doubt that I will be able to get a Corporate Contributor License Agreement signed, so I can unfortunately not contribute my change.) 75 seconds may not be much, but when a CI system does it repeatedly, it adds up."}
{"_id":"q-en-django__django-13089-fc894d636e6a75bd20dd430c375e265cdd0ee57ff4da3da162c08db209b842eb","text":"sometimes fails with 'NoneType' object is not subscriptable Description (last modified by Guillermo Bonvehí) I'm sporadically getting some cache errors using database backend. The error is: 'NoneType' object is not subscriptable And the backtrace: \/usr\/local\/lib\/python3.7\/site-→ getresponse \/usr\/local\/lib\/python3.7\/site-→ render \/usr\/local\/lib\/python3.7\/site-→ callback \/usr\/local\/lib\/python3.7\/site-→ processresponse \/usr\/local\/lib\/python3.7\/site-→ learncachekey \/usr\/local\/lib\/python3.7\/site-→ set \/usr\/local\/lib\/python3.7\/site-→ baseset \/usr\/local\/lib\/python3.7\/site-→ cull This is using Django 2.2.11 but I see the same code is in master. ​ cursor.execute( () % table, [cullnum]) cursor.execute(\"DELETE FROM %s \" \"WHERE cachekey < %%s\" % table, [cursor.fetchone()[0]]) From what I can understand, the cursor after running () command is not returning any data, so cursor.fetchone()[0] afterwards fails. I guess a simple check to see if it contains data would be enough, may apply for an easy picking. Edit: Wording"}
{"_id":"q-en-django__django-13097-608c39d068a7a80d5a8099ac012dd637068260d0a7eb633a88912f397e7f3013","text":"formsets with candelete=True shouldn't add delete field to extra forms Description Current behavior of formsets with candelete=True is to add a delete field to every form. This behavior differs from that expected, however (why would one want a delete option on an \"add\" form?), as well as that of the builtin admin. I've included a patch on , but haven't bothered with patching tests yet."}
{"_id":"q-en-django__django-13109-35ba09080d1b38e3192116b500b36d76e88dc471ff5d41dcea1bd2720fcacfc1","text":"ForeignKey.validate() should validate using the base manager. Description ForeignKey.validate() should validate using the base manager instead of the default manager. Consider the models: class ArticleManager(models.Manage): def getqueryset(self): qs = super().getqueryset() return (archived=False) class Article(models.Model): title = models.CharField(maxlength=100) archived = models.BooleanField(default=False) # Don't include archived articles by default. objects = ArticleManager() class FavoriteAricles(models.Model): article = models.ForeignKey(Article, ondelete=models.CASCADE) In the example, now consider a form that allows users to pick a favorite article including archived articles. class FavoriteAriclesForm(forms.ModelForm): class Meta: model = FavoriteArticle fields = 'all' def init(self, args, kwargs): super().init(args, kwargs) # Use the base manager instead of the default manager to allow archived articles. ['article'].queryset = () The above form will never validate as True when a user selects an archived article. This is because the ForeignKey validation always uses defaultmanager instead of basemanager. The user facing error message is \"article instance with id 123 does not exist.\" (quite confusing to typical users). The code for this validation is here: ​ The FavoriteAriclesForm is specifically designed to use a different manager, but the ForeignKey validation makes this difficult. In this example scenario, it is not acceptable to change the model's default manager as the default should avoid archived articles in other typical scenarios. Suggested solution: the ForeignKey validation should use the basemanager instead which does not include the default filters."}
{"_id":"q-en-django__django-13111-101015e601c13c20605db18b725bc7cb6c1c45a13cf9a253b85cad550f7d0be2","text":"Support \"%V\" format in WeekArchiveView. Description (Docs for WeekArchiveView are misleading about %W) - closed 4 years ago mentioned support for %V week format. Since python 3.6, %G, %u and %V ISO 8601 formatters were to strptime. WeekArchiveView should add %V to the list of accepted week formatters. This would require as well the special case to change the year format to %G, or simply ValueError in datefrom_string should mention the message passed from datetime.datetime.strptime: ISO week directive '%V' is incompatible with the year directive '%Y'. Use the ISO year '%G'."}
{"_id":"q-en-django__django-13112-d877245412d85c7f4f0ebc016157e1441a5964bf95e410c89cdb1942ac10d171","text":"makemigrations crashes for ForeignKey with mixed-case app name. Description When i run \"python3 migrate\" on Django 3.1b1 shows me that error (Please, note that the code works well in 3.0) ValueError: The field DJRegLogin.Content.category was declared with a lazy reference to 'djreglogin.category', but app 'djreglogin' isn't installed. (Conflict Part) class Category(models.Model): title = models.CharField(maxlength=100, dbindex=True) slug = models.SlugField(maxlength=100, dbindex=True) class Meta: verbosename = 'Category' verbosenameplural = 'Categories' def str(self): return def getabsoluteurl(self): return reverse('viewblogcategory', None, kwargs={'slug': }) class Content(models.Model): title = models.CharField(maxlength=100, unique=True) slug = models.SlugField(maxlength=100, unique=True) body = RichTextField(configname='default') posted = models.DateTimeField(dbindex=True, autonowadd=True) sites = models.ManyToManyField(Site) ip = models.GenericIPAddressField(editable=False) category = models.ForeignKey(Category, ondelete=models.CASCADE) user = models.ForeignKey(User, ondelete=models.CASCADE, null=False, blank=False, editable=False) status = models.CharField(maxlength=10, choices=STATUSCHOICES, default='draft') def str(self): return def getabsoluteurl(self): return reverse('viewblogpost', None, kwargs={'slug': }) (Related to issue part) INSTALLEDAPPS = [ 'DJRegLogin', '', '', 'django.contrib.contenttypes', 'django.contrib.sessions', 'django.contrib.messages', 'django.contrib.staticfiles', 'socialdjango', 'ckeditor', '', 'django.contrib.flatpages', 'django.contrib.sitemaps', ] from import AppConfig class DJRegLoginConfig(AppConfig): name = 'DJRegLogin' verbosename = \"Contents\""}
{"_id":"q-en-django__django-13115-959400a9c3ed169673c29bf9bbb874c0c14e0d75670e179f5fda5e5a49a28100","text":"Add errors when UniqueConstraint.deferrable is combined with index specific parameters Description UniqueConstraint should throw an error when deferrable is combined with parameters which require the usage of an explicit unique index through CREATE UNIQUE INDEX. This was missed when UniqueConstraint.include and UniqueConstraint.opclasses was . We should also add documentation about this incompatibility."}
{"_id":"q-en-django__django-13118-a9a874cc05c22b644db39faba16b89cf36a6f0ea62e9a602cffa0d24c7f2ff47","text":"SQL generated for negated F() expressions is incorrect Description Consider the following model definition. from import models class Rectangle(models.Model): length = models.IntegerField(null=True) width = models.IntegerField(null=True) We make the following queries: Q1) Rectangles that are squares. Q2a) Rectangles that are not squares (length != width). Q2b) Rectangles that are not squares (width != length). Queries Q2a and Q2b semantically mean the same. However, the ORM generates different SQL for these two queries. import django () from import F from import Rectangle print '--- Q1: Get the rectangles that are squares' print (length=F('width')).values('pk').query print '--- Q2a: Get the rectangles that are not squares' print Rectangle.objects.exclude(length=F('width')).values('pk').query print '--- Q2b: Get the rectangles that are not squares' print Rectangle.objects.exclude(width=F('length')).values('pk').query The generated SQL is --- Q1: Get the rectangles that are squares SELECT \"testapprectangle\".\"id\" FROM \"testapprectangle\" WHERE \"testapprectangle\".\"length\" = (\"testapprectangle\".\"width\") --- Q2a: Get the rectangles that are not squares SELECT \"testapprectangle\".\"id\" FROM \"testapprectangle\" WHERE NOT (\"testapprectangle\".\"length\" = (\"testapprectangle\".\"width\") AND \"testapprectangle\".\"length\" IS NOT NULL) --- Q2b: Get the rectangles that are not squares SELECT \"testapprectangle\".\"id\" FROM \"testapprectangle\" WHERE NOT (\"testapprectangle\".\"width\" = (\"testapprectangle\".\"length\") AND \"testapprectangle\".\"width\" IS NOT NULL) Note the asymmetry between Q2a and Q2b. These queries will return different results. Reddit user \/u\/charettes set up this useful SQL fiddle with the above mentioned schema and test values: ​!12\/\/4 Here's my reddit post on this issue: ​"}
{"_id":"q-en-django__django-13121-fac757c7b8d31125a3e61f37f086f38b0581071d2a7109185389a73597592f6b","text":"durations-only expressions doesn't work on SQLite and MySQL Description class Experiment(models.Model): estimatedtime = models.DurationField() list(Experiment.objects.annotate(duration=F('estimatedtime') + datime.timedelta(1))) Traceback (most recent call last): File \"\", line 1218, in testdurationexpressions list(Experiment.objects.annotate(duration=F('estimatedtime') + delta)) File \"\", line 269, in iter self.fetchall() File \"\", line 1172, in fetchall self.resultcache = list(self.iterableclass(self)) File \"\", line 63, in iter for row in compiler.resultsiter(results): File \"\", line 998, in applyconverters value = converter(value, expression, connection) File \"\", line 571, in convertdurationfield_value value = str(decimal.Decimal(value) \/ decimal.Decimal()) decimal.InvalidOperation: [<class 'decimal.ConversionSyntax']"}
{"_id":"q-en-django__django-13128-f807312f0373fcf6930dc23d6408193ad7c0499e63e01a08be8a3ff0593d8ce2","text":"make temporal subtraction work without ExpressionWrapper Description class Experiment(models.Model): start = models.DateTimeField() end = models.DateTimeField() Experiment.objects.annotate( delta=F('end') - F('start') + Value(datetime.timedelta(), outputfield=DurationField()) ) This gives: Expression contains mixed types: DateTimeField, DurationField. You must set outputfield."}
{"_id":"q-en-django__django-13145-117c2d0197a06f7bc97e3df1c8502b71605b6329eb246a58954f84f478ff8a55","text":"timesince 'depth' parameter Description (last modified by Toby Such) Discussed ​here. The timesince function is a bit limiting as it is right now. A depth parameter should be to configure how many values are shown. The depth parameter should be defaulted to 2 as this is how the current implementation behaves. The existing rule of values having to be adjacent to one another should still remain. The logic for calculating the time since, before formatting should also be pulled out of the function and placed in its own so that custom implementations can be created. For example: With a depth of one it should always display as \"1 week\" or \"3 years\" etc. With a depth of two: \"1 week, 3 days\" or \"3 years, 7 months\" With a depth of three: \"1 week, 3 days, 5 hours\" or \"3 years, 7 months, 2 weeks\""}
{"_id":"q-en-django__django-13158-f9ee5b5dd7b67d52eeb96607ffbc452060fe508b8b1e66f65e07e1b1e69f9c28","text":"() on combined queries returns all results. Description I came across this issue on Stack Overflow. I'm not 100% sure it's a bug, but it does seem strange. With this code (excuse the bizarre example filtering): class Publication(models.Model): pass class Article(models.Model): publications = models.ManyToManyField(to=Publication, blank=True, null=True) class ArticleForm(forms.ModelForm): publications = forms.ModelMultipleChoiceField( (idlt=2) | (idgt=5), required=False, ) class Meta: model = Article fields = [\"publications\"] class ArticleAdmin(admin.ModelAdmin): form = ArticleForm This works well. However, changing the ModelMultipleChoiceField queryset to use union() breaks things. publications = forms.ModelMultipleChoiceField( (idlt=2).union( (idgt=5) ), required=False, ) The form correctly shows only the matching objects. However, if you submit this form while empty (i.e. you didn't select any publications), ALL objects matching the queryset will be . Using the OR query, NO objects are , as I'd expect."}
{"_id":"q-en-django__django-13162-5b0c2b30182f7ec37e55a1692d7ce0d42771f2829246b470a5fd2a524e50b1ec","text":"Improve default name of merge migrations. Description Currently, merge migrations filenames are created with a timestamp. For example: This name is more opaque than necessary. When one reads it, it isn't immediately clear which migrations were merged. One must inspect the file to find that information. Instead, I suggest the default filename should be created by combining the files being merged. This way, it includes the information without requiring one to inspect the file. This information is also useful in the migrate command's output. As it is most likely to merge two migrations files, this should remain a reasonable length. For example: If preferable, we could decide to join the names in other ways too. For example, separate names by _ (two underscores) or exclude prefix numbers. To start, I'll go with the easiest approach unless there is strong preference for a different naming scheme."}
{"_id":"q-en-django__django-13165-b1577b742e483c63c3dc4cb0882777a8ec8c9209d01922f491720fe296d714bf","text":"ExpressionWrapper loses outputfield for combined expression without an outputfield. Description I have the following model. class Mallets(models.Model): id = models.AutoField(primarykey=True,blank=True, null=True) hindsight = models.ForeignKey(Hindsight, models.DONOTHING, blank=True, null=True) believer = models.IntegerField(blank=True, null=True) damnably = models.IntegerField(blank=True, null=True) issue = models.IntegerField(blank=True, null=True) glover = models.TextField(blank=True, null=True) # This field type is a guess. class Meta: dbtable = 'mallets' and I perform the following query on Django 3.2 sheer = ExpressionWrapper((F('issue') \/ F('id')), outputfield=FloatField()) lacquer = ExpressionWrapper(Avg(F('sheer'), outputfield=FloatField()), outputfield=TextField()) q = ('default') ret = q.annotate(sheer=sheer).values('sheer').annotate(lacquer=Sum(F('believer'))).orderby('sheer').first() Django however throws the following exception Traceback (most recent call last): File \"\", line 25, in <moduleret2 = ret1.annotate(sheer=sheer).values('sheer').annotate(lacquer=Sum('believer')).orderby('sheer').first() File \"\", line 678, in first for obj in (self if self.ordered else self.orderby('pk'))[:1]: File \"\", line 287, in iter self.fetchall() File \"\", line 1305, in fetchall self.resultcache = list(self.iterableclass(self)) File \"\", line 111, in iter for row in compiler.resultsiter(chunkedfetch=self.chunkedfetch, chunksize=self.chunksize): File \"\", line 1113, in resultsiter results = self.executesql(MULTI, chunkedfetch=chunkedfetch, chunksize=chunksize) File \"\", line 1148, in executesql sql, params = () File \"\", line 498, in assql extraselect, orderby, groupby = self.presqlsetup() File \"\", line 60, in presqlsetup groupby = self.getgroupby( + extraselect, orderby) File \"\", line 142, in getgroupby sql, params = expr.selectformat(self, sql, params) File \"\", line 388, in selectformat if hasattr(self.outputfield, 'selectformat'): File \"\", line 48, in get res = [] = (instance) File \"\", line 269, in outputfield outputfield = self.resolveoutputfield() File \"\", line 309, in resolveoutputfield , Expression contains mixed types: IntegerField, AutoField. You must set output_field. Note that when I run the query above on Django 3.0.8, this query runs as expected. So is this a regression in Django 3.2?"}
{"_id":"q-en-django__django-13170-6e103bd1983052801e0cc2aa6b576b644cf46b4052293bf9179863dde44eac51","text":"Support nested relations in FilteredRelation's condition Description As the documentation explains, FilteredRelation's condition do not support nested relations: Restaurant.objects.annotate( ... pizzaswithtoppingsstartswithn=FilteredRelation( ... 'pizzastoppings', ... condition=Q(pizzastoppingsnamestartswith='n'), ... ), ... ) Traceback (most recent call last): ... ValueError: FilteredRelation's condition doesn't support nested relations (got 'pizzastoppingsname__startswith'). It would be great to support nested relations in FilteredRelation's condition (I encountered this limitation multiple times recently)."}
{"_id":"q-en-django__django-13192-ebee4bd9dd4d1be677fcd4d90c9d79322a752b857b23441b0d85c473029fe7f2","text":"Emails name over 75 characters are incompatible with the latest versions of python. Description In the process of sending an email the addresses is sanatized: =def sanitizeaddress(addr, encoding) The Name portion is encoded via the Header class which will introduce newlines at 75 characters. Unfortunately the most recent python security update no longer allows that to happen. So when Address(nm, addrspec=addr) is called in sanitize_address a new error is raised from the Python Standard library. The update to python can be found here: ​ Essentially Django can no longer send emails with names longer then 75 chracters."}
{"_id":"q-en-django__django-13195-af77b91c5654a96c0303871e15d37e7be473dfb134e2a71c50da68eeb6b37327","text":"HttpResponse.deletecookie() should preserve cookie's samesite. Description We noticed we were getting this warning message from Firefox: 'Cookie “messages” will be soon rejected because it has the “sameSite” attribute set to “none” or an invalid value, without the “secure” attribute. To know more about the “sameSite“ attribute, read ​' We are getting this from the messages system but it doesn't look like an issue with the messages app. Here is the cookie header for messages on the POST: Set-Cookie: messages=(... encoded message text ...); HttpOnly; Path=\/; SameSite=Lax This has SameSite set. But the POST returns a 304 and the following GET's cookie header is this: Set-Cookie: messages=\"\"; expires=Thu, 01 Jan 1970 00:00:00 GMT; Max-Age=0; Path=\/ This looks like it is just expiring the cookie so the browser will delete it. As we were digging in to what might be causing this we noticed that messages is using the response's deletecookie method to expire the cookie if there is no message data. HttpResponseBase's deletecookie method doesn't seem like it setting the Samesite setting on Set-Cookie headers. It also is only setting 'Secure' if the cookie's key begins with 'Secure' or 'Host'. Chrome and Firefox will soon begin ignoring Set-Cookie headers with Samesite=None that aren't marked 'Secure'. This could result in Chrome and Firefox ignoring all cookies deleted through HttpResponseBase's deletecookie method if the cookie key does not start with 'Secure' or 'Host'. For testing I modified deletecookie to look like this: def deletecookie(self, key, path='\/', domain=None): # Most browsers ignore the Set-Cookie header if the cookie name starts # with Host- or Secure- and the cookie doesn't use the secure flag. self.setcookie( key, maxage=0, path=path, expires='Thu, 01 Jan 1970 00:00:00 GMT', domain=domain if domain is not None else settings.SESSIONCOOKIEDOMAIN, secure=settings.SESSIONCOOKIESECURE or key.startswith(('Secure-', 'Host-')), httponly=settings.SESSIONCOOKIEHTTPONLY or None, samesite=settings.SESSIONCOOKIESAMESITE, ) Definitely wouldn't want to use the session cookie settings for everything but changing this stopped the warnings from coming in on Firefox. I copied the kwargs from the messages code. Thought this might be worth a report."}
{"_id":"q-en-django__django-13199-af77b91c5654a96c0303871e15d37e7be473dfb134e2a71c50da68eeb6b37327","text":"HttpResponse.deletecookie() should preserve cookie's samesite. Description We noticed we were getting this warning message from Firefox: 'Cookie “messages” will be soon rejected because it has the “sameSite” attribute set to “none” or an invalid value, without the “secure” attribute. To know more about the “sameSite“ attribute, read ​' We are getting this from the messages system but it doesn't look like an issue with the messages app. Here is the cookie header for messages on the POST: Set-Cookie: messages=(... encoded message text ...); HttpOnly; Path=\/; SameSite=Lax This has SameSite set. But the POST returns a 304 and the following GET's cookie header is this: Set-Cookie: messages=\"\"; expires=Thu, 01 Jan 1970 00:00:00 GMT; Max-Age=0; Path=\/ This looks like it is just expiring the cookie so the browser will delete it. As we were digging in to what might be causing this we noticed that messages is using the response's deletecookie method to expire the cookie if there is no message data. HttpResponseBase's deletecookie method doesn't seem like it setting the Samesite setting on Set-Cookie headers. It also is only setting 'Secure' if the cookie's key begins with 'Secure' or 'Host'. Chrome and Firefox will soon begin ignoring Set-Cookie headers with Samesite=None that aren't marked 'Secure'. This could result in Chrome and Firefox ignoring all cookies deleted through HttpResponseBase's deletecookie method if the cookie key does not start with 'Secure' or 'Host'. For testing I modified deletecookie to look like this: def deletecookie(self, key, path='\/', domain=None): # Most browsers ignore the Set-Cookie header if the cookie name starts # with Host- or Secure- and the cookie doesn't use the secure flag. self.setcookie( key, maxage=0, path=path, expires='Thu, 01 Jan 1970 00:00:00 GMT', domain=domain if domain is not None else settings.SESSIONCOOKIEDOMAIN, secure=settings.SESSIONCOOKIESECURE or key.startswith(('Secure-', 'Host-')), httponly=settings.SESSIONCOOKIEHTTPONLY or None, samesite=settings.SESSIONCOOKIESAMESITE, ) Definitely wouldn't want to use the session cookie settings for everything but changing this stopped the warnings from coming in on Firefox. I copied the kwargs from the messages code. Thought this might be worth a report."}
{"_id":"q-en-django__django-13218-763b5437546434496c6918708ede541338552790c20a4b79e9c473178d39c620","text":"Allow migrations directories without files Description (last modified by Tim Graham) Background: In python 3 a package with no is implicitly a namespace package, so it has no file attribute. The migrate command currently checks for existence of a file attribute on the migrations package. This check was introduced in , because the file attribute was used in migration file discovery. However, in migration file discovery was changed to use pkgutil.itermodules (), instead of direct filesystem access. pkgutil. itermodules() uses the package's path list, which exists on implicit namespace packages. As a result, the file check is no longer needed, and in fact prevents migrate from working on namespace packages (implicit or otherwise). Related work:"}
{"_id":"q-en-django__django-13220-2bbca031a5833432982b3cbd64a8a7c67c89086f390b25755e0fd50636a7c5af","text":"Allow ValidationErrors to equal each other when created identically Description (last modified by kamni) Currently ValidationErrors () that have identical messages don't equal each other, which is counter-intuitive, and can make certain kinds of testing more complicated. Please add an eq method that allows two ValidationErrors to be compared. Ideally, this would be more than just a simple self.messages == other.messages. It would be most helpful if the comparison were independent of the order in which errors were raised in a field or in nonfielderrors."}
{"_id":"q-en-django__django-13230-db043e661952acfe11f81df09e5e984b5e118de977c335340ff3d075bb1b4582","text":"Add support for itemcomments to syndication framework Description Add comments argument to feed.additem() in so that itemcomments can be defined directly without having to take the detour via itemextra_kwargs . Additionally, comments is already explicitly mentioned in the feedparser, but not implemented in the view."}
{"_id":"q-en-django__django-13233-a6c2c67b04a867bbb72e3babce70f8469f136f0adb41fe6aa565395ed47ea155","text":"The attribute of image fields doesn't point to concrete model. Description In Django 3.1 and before, one could use the model attribute of image fields to find the concrete model the image field belongs to. This isn't possible in 3.2 anymore, and I bisected the change to the fix of . I found this while investigating a CI failure of django-imagefield ​https:\/\/travis- I'm not sure whether this is a bug or whether it is an intentional change. If it is the later, is there an alternative to find the concrete model an image field belongs to? I'm classifying this as a bug because the change made model and field introspection harder than it was before. Also, since behavior changed may possibly not be classified as a cleanup\/optimization anymore..."}
{"_id":"q-en-django__django-13236-fbd8623f5fb430baad87110edefcf23da50235a5147cc8c971061c8e4a62ea3b","text":"RenameField with dbcolumn should be a noop. Description (last modified by Iuri de Silvio) RenameField with dbcolumn should be a noop because it is only the Django column that changed, the database still have the same db_column."}
{"_id":"q-en-django__django-13237-3c963e3f360c70cb99d3104e35fc97833c9ab42f6416f779ea35cc78369cd90b","text":"AlterField with dbcolumn addition should be a noop. Description (last modified by Iuri de Silvio) When I change pink = models.Integer(default=0) to pink = models.Integer(default=0, dbcolumn=\"pink\") the migration drop\/create the same constraints when it is an FK or even reconstruct the table (SQLite), but nothing really changed. The constraint drop\/create is a blocking operation for PostgreSQL, so it is an undesirable and unexpected behavior."}
{"_id":"q-en-django__django-13240-cffda7b7a48f99ced3a545af2d03e7c8ac3bd3dcf12705e937821f3099c561cf","text":"Change Settings to raise ImproperlyConfigured on SECRETKEY; not initialization Description Since ticket , initializing settings without a SECRETKEY raises a an ImproperlyConfigured during settings initialization. Instead, I think the error should be raised when the setting is accessed as Settings.SECRETKEY. My use case, my project has a number of management commands that run in a non-production, minimally configured environment. These management commands do not require SECRETKEY, however, the environment is forced to provide one. As a workaround this environment has been generating a random secret key each run. If Django were to instead raise the error on SECRET_KEY access, this workaround would not be necessary."}
{"_id":"q-en-django__django-13250-02700f6271548a086ad54bfda35ab8969e2c2318be819d114388970a3063fa93","text":"JSONField's contains and containedby lookups don't work with nested values on SQLite. Description SQLite doesn't provide a native way for testing containment of JSONField. The current implementation works only for basic examples without supporting nested structures and doesn't follow \"the general principle that the contained object must match the containing object as to structure and data contents, possibly after discarding some non-matching array elements or object key\/value pairs from the containing object\". I'm not sure if it's feasible to emulate it in Python. Some (not really complicated) examples that don't work: diff --git index .. --- +++ -449,9 +449,14 class TestQuerying(TestCase): tests = [ ({}, [2:5] + [6:8]), ({'baz': {'a': 'b', 'c': 'd'}}, [[7]]), ({'baz': {'a': 'b'}}, [[7]]), ({'baz': {'c': 'd'}}, [[7]]), ({'k': True, 'l': False}, [[6]]), ({'d': ['e', {'f': 'g'}]}, [[4]]), ({'d': ['e']}, [[4]]), ([1, [2]], [[5]]), ([1], [[5]]), ([[2]], [[5]]), ({'n': [None]}, [[4]]), ({'j': None}, [[4]]), ]"}
{"_id":"q-en-django__django-13251-a6cbf9f53d067a29a40b9f759230ab3825f531124b1ef0285110fb8771bd38b6","text":"Filtering on a field named raises a TypeError Description Filtering on a model with a field named negate raises a TypeError. For example: class Foo(models.Model): negate = models.BooleanField() (negate=True) raises TypeError: filterorexclude() got multiple values for argument 'negate' negate is not documented as a reserved argument for .filter(). I'm currently using .filter(negate_exact=True) as a workaround."}
{"_id":"q-en-django__django-13265-d431d694418d2cf08056073c40d994a606b561f9e91167770d6509a12f7ccb30","text":"AlterOrderWithRespectTo() with ForeignKey crash when order is included in Index(). Description class Meta: dbtable = 'lookimage' orderwithrespectto = 'look' indexes = [ models.Index(fields=['look', 'order']), models.Index(fields=['createdat']), models.Index(fields=['updatedat']), ] migrations.CreateModel( name='LookImage', fields=[ ('id', models.AutoField(autocreated=True, primarykey=True, serialize=False, verbosename='ID')), ('look', models.ForeignKey(, relatedname='images', to='posts.Look', verbosename='LOOK')), ('imageurl', models.URLField(blank=True, maxlength=, null=True)), ('image', models.ImageField(maxlength=2000, uploadto='')), ('deleted', models.DateTimeField(editable=False, null=True)), ('createdat', models.DateTimeField(autonowadd=True)), ('updatedat', models.DateTimeField(autonow=True)), ], ), migrations.AddIndex( modelname='lookimage', index=models.Index(fields=['look', 'order'], name='lookimagelookideaff30idx'), ), migrations.AddIndex( modelname='lookimage', index=models.Index(fields=['createdat'], name='lookimagecreatedf746cfidx'), ), migrations.AddIndex( modelname='lookimage', index=models.Index(fields=['updatedat'], name='lookimageupdatedaceaf9idx'), ), migrations.AlterOrderWithRespectTo( name='lookimage', orderwithrespectto='look', ), I orderswithrespectto in new model class's Meta class and also made index for 'order' field by combining with other field. And a new migration file based on the model looks like the code above. The problem is operation AlterOrderWithRespectTo after AddIndex of 'order' raising error because 'order' field had not been created yet. It seems to be AlterOrderWithRespectTo has to proceed before AddIndex of '_order'."}
{"_id":"q-en-django__django-13267-3ac6cd4f01db82ed308934f3dfba29a85996f58c9a691846d6c353b70dff79c4","text":"Instantiating an abstract model with a string ForeignKey fails with TypeError: isinstance() arg 2 must be a class, type, or tuple of classes and types Description Trying to instantiate an abstract model with a foreign key worked without throwing any warnings in Django 1.8. In Django 1.9.8, this code: class UserProfile(Model): user = models.ForeignKey('') class Meta(object): applabel = 'core' abstract = True UserProfile() raises this exception: Traceback (most recent call last): File \"<ipython-input-7-\", line 2, in <moduleUserProfile() File \"\", line 432, in init val = field.getdefault() File \"\", line 905, in getdefault if isinstance(fielddefault, ): TypeError: isinstance() arg 2 must be a class, type, or tuple of classes and types While the code mentions that abstract models can not be instantiated [1], I couldn't find mention of this in the docs. Abstract models without foreign keys can still be instantiated perfectly fine. I couldn't find any description of this change in the Release Notes, so not sure if I missed it there or if this is maybe an undocumented backwards-incompatible change. [1] ​"}
{"_id":"q-en-django__django-13279-ae5b4da15261778ecfc2c97ed7199b5995ce14216cb888e741f8579268aadb08","text":"Session data cannot be decoded during the transition to Django 3.1. Description In () we've changed format for session data, that's why setting DEFAULTHASHINGALGORITHM to 'sha1' is not enough to support running multiple instances of the same project during the transition to Django 3.1. We could use the legacy encode() when DEFAULTHASHINGALGORITHM == 'sha1' (it's a bit hacky)."}
{"_id":"q-en-django__django-13281-ace6a43d2a3871012940788a8ac1893874552f84d58c72c3e0aaf0cd8fd292c3","text":"FK field caching behavior change between 1.11.x and 2.x Description Whilst upgrading a codebase from 1.11.x to 2.0\/2.2 I noticed a weird change in behavior of FK fields when copying model instances. At the bottom of the post there is a testcase that succeeds on 1.11.x and fails on 2.x I think the commit that changed the behavior is So my question is two fold: Is the behavior in =2.0 correct? It seems quite unexpected. What is the recommended way to clone a model instance? To date we have been using copy() in a similar fashion to the test without issue. deepcopy seems to work fine in =2.0 but we haven’t done too much testing yet. Test (placed in ): import copy from import TestCase from .models import Bar, Foo class ForeignKeyCachingBehaviorTest(TestCase): def testcopy(self): foo1 = (a='foo1', d=1) foo2 = (a='foo2', d=2) bar1 = (a=foo1, b='bar1') bar2 = (bar1) = None bar2.a = foo2 # bar2 points to foo2 self.assertEqual(bar2.a, foo2) self.assertEqual(, ) # bar1 is unchanged and must still point to foo1 # These fail on Django = 2.0 self.assertEqual(bar1.a, foo1) self.assertEqual(, ) and executed that via: python3.6 --parallel 1 modelfields In ​ Simon suggests: ..... should make sure to make a deep-copy of now that fields are cached in which I will attempt to implement."}
{"_id":"q-en-django__django-13287-87d81505d7635900c2e7709c2e77f5ad2b46c5afa943afdb38a796004c675536","text":"App with defaultappconfig and without or with an empty crashes. Description (last modified by Iuri de Silvio) If I don't have an and the defaultappconfig is in , it fails. Traceback (most recent call last): File \"\", line 22, in <modulemain() File \"\", line 18, in main executefromcommandline() File \"\", line 401, in executefromcommandline utility.execute() File \"\", line 377, in execute () File \"\", line 24, in setup apps.populate(settings.INSTALLEDAPPS) File \"\", line 91, in populate appconfig = (entry) File \"\", line 157, in create if newentry == appconfigname: UnboundLocalError: local variable 'appconfigname' referenced before assignment If the is there, but the defaultappconfig is in , it fails too. Traceback (most recent call last): File \"\", line 381, in inner return func(args, *kwargs) File \"\", line 541, in testexplicitdefaultappconfigwithemptyapps with self.settings(INSTALLEDAPPS=['apps.explicitdefaultconfigwithemptyapps']): File \"\", line 336, in enter return () File \"\", line 410, in enable apps.setinstalledapps(self.options['INSTALLEDAPPS']) File \"\", line 355, in setinstalledapps self.populate(installed) File \"\", line 91, in populate appconfig = (entry) File \"\", line 160, in create if newentry == appconfigname: UnboundLocalError: local variable 'appconfig_name' referenced before assignment Looks like a regression in"}
{"_id":"q-en-django__django-13295-c162ff52e2f5a710137c2ebe95f8b638e876312a17da85fdc818599d8216bfce","text":"Silent failure when saving non-concrete field with updatefields Description If you have a non-concrete field which wraps a couple of concrete fields (a bit like GenericForeignKey, though my codebase doesn't uses this for compound data types rather than relations), then when you pass the name of that non-concrete field to save(updatefields=('mynonconcretefield',)), the model will be saved without saving that field and yet no error will be emitted. I think that this could be because the check for the validity of the updatefields is done against (​) while the later check for which fields to actually save is done using meta.localconcretefields (​). Ideally, I would like a way for a non-concrete field to be able to specify the underlying concrete fields which should be saved on its behalf. That's clearly rather complex though (and would likely have impacts beyond just update_fields) -- a simpler solution would be to error about this case so that the developer knows something is amis."}
{"_id":"q-en-django__django-13297-e24117172dc74d88160d2732af4f46d543dd637af2ff105e3ccfdfd039548afd","text":"TemplateView.getcontextdata()'s kwargs returns SimpleLazyObjects that causes a crash when filtering. Description Example Code that works in 3.0, but not in 3.1: class OfferView(TemplateView): templatename = \"\" def getcontextdata(self, kwargs): offerslug = (\"offerslug\", \"\") offer = getobjector404(Account, slug=offerslug) return {\"offer\": offer, \"offerslug\": offerslug} In order to make this work in 3.1, you have to explicitly convert the result of () to a string to get the SimpleLazyObject to resolve: class OfferView(TemplateView): templatename = \"\" def getcontextdata(self, kwargs): offerslug = (\"offerslug\", \"\") offer = getobjector404(Account, slug=str(offerslug)) return {\"offer\": offer, \"offerslug\": offerslug} The error generated if you don't is: Error binding parameter 0 - probably unsupported type from , line 144, in quoteparamsforlastexecutedquery In both cases, the looks like: path( \"\/offers\/<slug:offerslug\/\", (), name=\"offerview\", ), When debugging, I found that offerslug (coming in from ) was of type 'SimpleLazyObject' in Django 3.1, and when I explicitly converted it to a string, getobjector_404 behaved as expected. This is using Python 3.7.8 with SQLite."}
{"_id":"q-en-django__django-13300-5e64d94d0c3e601acc4ad2e1107ef4cb6688cd0d64f367023b1bf4baa4b5ce8a","text":"Use for subqueries Description If you write a QuerySet call like so in Django... ().exists() the query run will be like so. SELECT 1 AS \"a\" FROM \"myappmymodel\" LIMIT 1; If you use the Exists() function to filter with a subquery like so... (Exists(())) The subquery will be run like so. ... WHERE EXISTS(SELECT \"myappmyothermodel\".\"id\", ... FROM \"myappmyothermodel\"); It would be nice if the queries generated for Exists() used SELECT 1 like .exists() does, where possible. In an app I work on, I have one query in particular that is 15KB in size, but only around 8KB if I apply .annotate(1=Value(1, outputfield=IntegerField())).valueslist('_1') to all of the subqueries. That change alone is enough to make my queries much easier to debug."}
{"_id":"q-en-django__django-13301-c095b0796da1a439873d704f625a663541e931630d3676a115ae3a70895e8c5f","text":"Default username in createsuperuser command doesn't respect the --database option. Description The createsuperuser command in interactive mode suggests to leave username blank and use the default name (). The default name is validated to not be already used by an another user. This validation executes a query on User model using default database and not using the database option passed to the command. This is the problem when you are using multiple databases."}
{"_id":"q-en-django__django-13315-a481cafe2e47ad5ef09a6f0182c617ef21b0dc79fbc11588c661af8357790dab","text":"limitchoicesto on a ForeignKey can render duplicate options in formfield Description If you pass a Q object as limitchoicesto on a ForeignKey field involving a join, you may end up with duplicate options in your form. See regressiontest in patch for a clear view on the problem."}
{"_id":"q-en-django__django-13321-6a5377a50b8a34141f1e2863e8a69e43b10a0c3e1c3577ead6c51b2955099ccf","text":"Decoding an invalid session data crashes. Description (last modified by Matt Hegarty) Hi I recently upgraded my staging server to 3.1. I think that there was an old session which was still active. On browsing to any URL, I get the crash below. It looks similar to ​this issue. I cannot login at all with Chrome - each attempt to access the site results in a crash. Login with Firefox works fine. This is only happening on my Staging site, which is running Gunicorn behind nginx proxy. Internal Server Error: \/overview\/ Traceback (most recent call last): File \"\/usr\/local\/lib\/python3.8\/site-\", line 215, in getsession return self.sessioncache AttributeError: 'SessionStore' object has no attribute 'sessioncache' During handling of the above exception, another exception occurred: Traceback (most recent call last): File \"\/usr\/local\/lib\/python3.8\/site-\", line 118, in decode return (sessiondata, salt=self.keysalt, serializer=self.serializer) File \"\/usr\/local\/lib\/python3.8\/site-\", line 135, in loads base64d = TimestampSigner(key, salt=salt).unsign(s, maxage=maxage).encode() File \"\/usr\/local\/lib\/python3.8\/site-\", line 201, in unsign result = super().unsign(value) File \"\/usr\/local\/lib\/python3.8\/site-\", line 184, in unsign raise BadSignature('Signature \"%s\" does not match' % sig) Signature \"xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx\" does not match During handling of the above exception, another exception occurred: Traceback (most recent call last): File \"\/usr\/local\/lib\/python3.8\/site-\", line 47, in inner response = getresponse(request) File \"\/usr\/local\/lib\/python3.8\/site-\", line 179, in getresponse response = wrappedcallback(request, callbackargs, callbackkwargs) File \"\/usr\/local\/lib\/python3.8\/site-\", line 73, in view return self.dispatch(request, args, kwargs) File \"\/usr\/local\/lib\/python3.8\/site-\", line 50, in dispatch if not File \"\/usr\/local\/lib\/python3.8\/site-\", line 240, in inner () File \"\/usr\/local\/lib\/python3.8\/site-\", line 376, in setup self.wrapped = self.setupfunc() File \"\/usr\/local\/lib\/python3.8\/site-\", line 38, in verifyuser user.otpdevice = None File \"\/usr\/local\/lib\/python3.8\/site-\", line 270, in setattr () File \"\/usr\/local\/lib\/python3.8\/site-\", line 376, in setup self.wrapped = self.setupfunc() File \"\/usr\/local\/lib\/python3.8\/site-\", line 23, in = SimpleLazyObject(lambda: getuser(request)) File \"\/usr\/local\/lib\/python3.8\/site-\", line 11, in getuser request.cacheduser = auth.getuser(request) File \"\/usr\/local\/lib\/python3.8\/site-\", line 174, in getuser userid = getusersessionkey(request) File \"\/usr\/local\/lib\/python3.8\/site-\", line 58, in getusersessionkey return getusermodel()(request.session[SESSIONKEY]) File \"\/usr\/local\/lib\/python3.8\/site-\", line 65, in getitem return self.session[key] File \"\/usr\/local\/lib\/python3.8\/site-\", line 220, in getsession self.sessioncache = () File \"\/usr\/local\/lib\/python3.8\/site-\", line 44, in load return (s.sessiondata) if s else {} File \"\/usr\/local\/lib\/python3.8\/site-\", line 122, in decode return self.legacydecode(sessiondata) File \"\/usr\/local\/lib\/python3.8\/site-\", line 126, in legacydecode encodeddata = base64.b64decode(('ascii')) File \"\", line 87, in b64decode return binascii.a2b_base64(s) binascii.Error: Incorrect padding"}
{"_id":"q-en-django__django-13325-844be33c7ec0f1f3eda5900ed969299308a11db77aa6a15cb3cf043b648c3215","text":"Inconsistent key validation checks in cache backends. Description (last modified by Nick Pope) The fix for CVE2020- ensured that validatekey() was called for most cache-related operations to avoid a potential get\/set key-clash. There are some other operations that are not properly validated in some of the backend (sub)classes: () BaseMemcachedCache.deletemany() () () () () The fix to this should also include a test to ensure that self.validatekey(key) is called for all operations to avoid this issue in future. Note that this was originally raised via the security mailing list, but the decision was to handle this by raising a pull request. The main concern here is the potential for data loss in the unvalidated delete() and deletemany() operations."}
{"_id":"q-en-django__django-13341-dcb4cf32201d6eda58fc2df188e577324202b1ccce1c305620460e485f04f35a","text":"Stop TemplateView automatically passing kwargs into the context Description Only TemplateView pushes to the context. ListView does not, I yet have to check others. This is inconsistency and, I think, it should be fixed. TemplateView.getcontextdata()'s kwargs returns SimpleLazyObjects that causes a crash when filtering. Description Example Code that works in 3.0, but not in 3.1: class OfferView(TemplateView): templatename = \"\" def getcontextdata(self, kwargs): offerslug = (\"offerslug\", \"\") offer = getobjector404(Account, slug=offerslug) return {\"offer\": offer, \"offerslug\": offerslug} In order to make this work in 3.1, you have to explicitly convert the result of () to a string to get the SimpleLazyObject to resolve: class OfferView(TemplateView): templatename = \"\" def getcontextdata(self, kwargs): offerslug = (\"offerslug\", \"\") offer = getobjector404(Account, slug=str(offerslug)) return {\"offer\": offer, \"offerslug\": offerslug} The error generated if you don't is: Error binding parameter 0 - probably unsupported type from , line 144, in quoteparamsforlastexecutedquery In both cases, the looks like: path( \"\/offers\/<slug:offerslug\/\", (), name=\"offerview\", ), When debugging, I found that offerslug (coming in from ) was of type 'SimpleLazyObject' in Django 3.1, and when I explicitly converted it to a string, getobjector_404 behaved as expected. This is using Python 3.7.8 with SQLite."}
{"_id":"q-en-django__django-13343-00febe05d18be177ef660dae4f81a8fc11d23e129d78656c5e4b72b81a0f388e","text":"FileField with a callable storage does not deconstruct properly Description A FileField with a callable storage parameter should not actually evaluate the callable when it is being deconstructed. The documentation for a FileField with a callable storage parameter, states: You can use a callable as the storage parameter for or This allows you to modify the used storage at runtime, selecting different storages for different environments, for example. However, by evaluating the callable during deconstuction, the assumption that the Storage may vary at runtime is broken. Instead, when the FileField is deconstructed (which happens during makemigrations), the actual evaluated Storage is inlined into the deconstucted FileField. The correct behavior should be to return a reference to the original callable during deconstruction. Note that a FileField with a callable upload_to parameter already behaves this way: the deconstructed value is simply a reference to the callable. This bug was introduced in the initial implementation which allowed the storage parameter to be callable: ​ , which fixed the ticket"}
{"_id":"q-en-django__django-13344-0baf1b1540e64e03131b2dfa004e377767be867f32752141a38dfe250bd1867f","text":"Coroutine passed to the first middleware's processresponse() instead of HttpResponse. Description Like the title says, using ASGI (+ uvicorn in my case), the first middleware (according to the list in ) receives a coroutine as its response parameter, while all other middlewares down the line receive a object. This seems to have caused an issue in the django-cors-headers package which is often placed first in order: ​ How to reproduce: Set up a django 3.1 project with an async server (uvicorn in my case) Create a dummy class-based middleware that prints the types of arguments it receives in its processresponse method: class DummyMiddleware(MiddlewareMixin): def processresponse(self, request, response): print(, ) Set up the middleware as the first one in MIDDLEWARE = [ 'djangouvicorn_test.middleware.DummyMiddleware', 'django.middleware.security.SecurityMiddleware', ... Launch the server and perform any request, observe console output: <class ''<class 'coroutine'Move the middleware down on the list, restart the server and perform a request again: <class ''<class ''"}
{"_id":"q-en-django__django-13346-00ef94eaeccd9aeadb93cf8bfc059308c96e386567e9ad0d7496fa5a8d8a5d22","text":"On MySQL, Oracle, and SQLite, in lookup doesn't work on key transforms. Description I am currently rewriting our app where we will start using models.JSONField instead of I noticed that the in operator is not reacting the same way is it does on other fields. firstfilter = {‘ourfieldkeyin': [0]} firstitems = (firstfilter) len(firstitems) 0 secondfilter = {'ourfield_key': 0} seconditems = (secondfilter) len(seconditems ) 312 I would expect that both filters would give me the same queryset but this is not the case."}
{"_id":"q-en-django__django-13347-099bfae716d31d709846755ab274d788011e2573374090b65eb40fd3b31535d8","text":"SafeExceptionReporterFilter does not recurse into dictionaries with non-string keys Description SafeExceptionReporterFilter has provisions for recursively cleaning settings by descending into lists \/ tuples \/ dictionaries - which is great! However, recursing on dictionaries only works if the keys of the dictionary are strings. For instance it will fail to sanitize the following example: SOMESETTING = {1: {'login': 'cooper', 'password': 'secret'}} The reason for this is that cleansesetting starts by trying to apply a the hidden_settings regex to the key before attempting to recurse into the value: ​"}
{"_id":"q-en-django__django-13350-094681eb3b162d467e36675b1aed1f0b64dee33b44952b0fca25f90fdc2f3091","text":"Temporary files do not get deleted on canceled upload request. Description (last modified by Mariusz Felisiak) Temporary files do not get deleted when upload (request) is canceled by client (e.g. browser is closed before upload is finished)."}
{"_id":"q-en-django__django-13354-829d0ac836b88acab33e7bce0214feafed503d15e13c674a93cdacf47e85e1cc","text":"MigrationOptimizer mangles operation order if app name contains uppercase letters. Description I am aware that app names are preferably all-lowercase according to ​PEP 8, but uppercase letters are nevertheless valid. Steps to reproduce: Create a new project and an app with uppercase letters in the app name : django-admin startproject mysite cd mysite python startapp MyApp Add 'MyApp' to the INSTALLEDAPPS in Edit : from import models class RefModel(models.Model): pass class BaseModel(models.Model): r = models.ForeignKey(RefModel, ondelete=models.PROTECT) class SubModel(BaseModel): pass Run python makemigrations . In the resulting migration file, the create operation for SubModel comes before the create operation for BaseModel, which is wrong. Run python migrate , which will fail with this error: Operations to perform: Apply all migrations: MyApp, admin, auth, contenttypes, sessions Running migrations: Applying MyApp.0001initial...Traceback (most recent call last): File \"\", line 22, in <modulemain() File \"\", line 18, in main executefromcommandline() File \"\/home\/koen\/.virtualenvs\/dtest\/lib\/python3.8\/site-\", line 401, in executefromcommandline utility.execute() File \"\/home\/koen\/.virtualenvs\/dtest\/lib\/python3.8\/site-\", line 395, in execute self.fetchcommand(subcommand).runfromargv() File \"\/home\/koen\/.virtualenvs\/dtest\/lib\/python3.8\/site-\", line 330, in runfromargv self.execute(args, cmdoptions) File \"\/home\/koen\/.virtualenvs\/dtest\/lib\/python3.8\/site-\", line 371, in execute output = (args, options) File \"\/home\/koen\/.virtualenvs\/dtest\/lib\/python3.8\/site-\", line 85, in wrapped res = handlefunc(args, *kwargs) File \"\/home\/koen\/.virtualenvs\/dtest\/lib\/python3.8\/site-\", line 243, in handle postmigratestate = executor.migrate( File \"\/home\/koen\/.virtualenvs\/dtest\/lib\/python3.8\/site-\", line 117, in migrate state = self.migrateallforwards(state, plan, fullplan, fake=fake, fakeinitial=fakeinitial) File \"\/home\/koen\/.virtualenvs\/dtest\/lib\/python3.8\/site-\", line 147, in migrateallforwards state = self.applymigration(state, migration, fake=fake, fakeinitial=fakeinitial) File \"\/home\/koen\/.virtualenvs\/dtest\/lib\/python3.8\/site-\", line 227, in applymigration state = (state, schemaeditor) File \"\/home\/koen\/.virtualenvs\/dtest\/lib\/python3.8\/site-\", line 114, in apply operation.stateforwards(self.applabel, projectstate) File \"\/home\/koen\/.virtualenvs\/dtest\/lib\/python3.8\/site-\", line 80, in stateforwards state.addmodel(ModelState( File \"\/home\/koen\/.virtualenvs\/dtest\/lib\/python3.8\/site-\", line 95, in addmodel self.reloadmodel(applabel, modelname) File \"\/home\/koen\/.virtualenvs\/dtest\/lib\/python3.8\/site-\", line 156, in reloadmodel (relatedmodels) File \"\/home\/koen\/.virtualenvs\/dtest\/lib\/python3.8\/site-\", line 189, in reload (statestoberendered) File \"\/home\/koen\/.virtualenvs\/dtest\/lib\/python3.8\/site-\", line 310, in rendermultiple raise InvalidBasesError( Cannot resolve bases for [<ModelState: 'MyApp.SubModel'] This can happen if you are inheriting models from an app with migrations (e.g. ) in an app with no migrations; see for more This bug does not occur if the app name is all-lowercase. Digging into the code, I found that the MigrationOptimizer will combine two operations (Create model BaseModel and add ForeignKey-field r to BaseModel) without taking into account that SubModel depends on BaseModel."}
{"_id":"q-en-django__django-13355-0567fa014742ed344dd0152b0983317f7cf29a580fcf2ad6a0ae265f1d2012ef","text":"Optimize Description While working with another project that make extensive use of I discovered that the fix for ticket has unintended consequences on the performance of If the number of Media objects grows beyond a certain point (not sure it may be machine specific) then the performance of all subsequent calls becomes terrible. This was causing page load times of several minutes on my development machine. I agree that you need to delay as long as possible as intends but it seems that there probably should be an upper bound on this so that performance does not suddenly decrease. Here is some sample code that can reproduce the issue: from import Media import datetime def createmedia(MediaClass): return MediaClass(css={'all': ['']}, js=['']) start = () media = createmedia(Media) for i in range(): media = media + createmedia(Media) print(' additions took: %s' % (() - start)) On my machine several runs of this code result in times between 1:35 - 1:44 (eg. 1 minute 35 seconds to 1 minute 44 seconds). However, taking away one zero from the number of media objects runs in under a second. Near as I can tell this has to do with the memory used to store these arrays and when it gets past a certain point the performance is awful. Since I am not sure if it is machine specific, for reference my machine is a i7-8700 with 64 GB RAM. Here is a sample that has a modified Media class that does not have theses issues: from import Media import datetime def createmedia(MediaClass): return MediaClass(css={'all': ['']}, js=['']) class CustomMedia(Media): def add(self, other): combined = CustomMedia() if len(self.csslists) + len(other.csslists) 1000: combined.csslists = [, ] else: combined.csslists = self.csslists + other.csslists if len(self.jslists) + len(other.jslists) 1000: combined.jslists = [, ] else: combined.jslists = self.jslists + other.jslists return combined start = () media = createmedia(CustomMedia) for i in range(): media = media + createmedia(CustomMedia) print(' additions took: %s' % (() - start)) With this change it again runs in under a second. If I increase the number of loops the performance seems to change at a much more expected level. I set an upper limit on the length allowed before a merge occurs. I'm not sure if the number of additions allowed should be a setting that can be adjusted to meet specific needs or if something that is just \"reasonably\" high is sufficient. It does appear that limiting the total number of items in the list to about 1000 works best on my machine. I'm also not sure that this is the best solution. Thanks for your consideration."}
{"_id":"q-en-django__django-13363-9b008d85c8c9433e27763a86659286264a9166eb7c04127b8018bc3229a78372","text":"Add support for tzinfo parameter to TruncDate() and TruncTime(). Description (last modified by Joe Jackson) Description TruncDate inherits from TruncBase, which includes the TimeZone mixin. This should allow a developer to pass in a tzinfo object to be used when converting TruncDate, but it actually uses the return value from getcurrenttimezonename() unconditionally and completely discards the passed in timezone info object. The result is that attempting to aggregate by date doesn't work for timezones other than the global For example I can't have the django app be in UTC and pass the \"America\/NewYork\" timezone in. Here's the offending line: ​ Note, that a similar issue is happening in TruncTime. Here's the method I would expect it to use: ​ Example class TimeSlots(models.Model): startat = models.DateTimeField() tz = pytz.timezone(\"America\/NewYork\") report = ( TimeSlots.objects.annotate(startdate=TruncDate(\"startat\", tzinfo=tz)) .values(\"startdate\") .annotate(timeslotcount=Count(\"id\")) .values(\"startdate\", \"timeslotcount\") ) I would expect this to work, but currently the results are wrong for any timezone other than the one returned by Workaround There was a workaround for me. I was able to use TruncDay and then convert the DateTimes returned outside of the database, but I found no way to convert from DateTime to Date in the database. Maybe a Cast would work, but I would expect TruncDate to work. Patch ​PR"}
{"_id":"q-en-django__django-13369-985971455fbe6041bca692736fbbbbd6885f2d830ab998ec1d7bafd34a3a317c","text":"Django fails with FieldError: Expression contains mixed types: IntegerField, AutoField. You must set outputfield. Description I have the following query. expr = Value(3) * F('id') o = ('default') res = o.values('fieldname').annotate(expr=expr).values('expr') print(res) Unfortunately, this query crashes with a FieldError exception. The track trace is Traceback (most recent call last): File \"\", line 28, in <moduleprint(res) File \"\", line 287, in iter self.fetchall() File \"\", line 1316, in fetchall self.resultcache = list(self.iterableclass(self)) File \"\", line 111, in iter for row in compiler.resultsiter(chunkedfetch=self.chunkedfetch, chunksize=self.chunksize): File \"\", line 1115, in resultsiter results = self.executesql(MULTI, chunkedfetch=chunkedfetch, chunksize=chunksize) File \"\", line 1147, in executesql sql, params = () File \"\", line 498, in assql extraselect, orderby, groupby = self.presqlsetup() File \"\", line 55, in presqlsetup self.setupquery() File \"\", line 46, in setupquery , self.klassinfo, self.annotationcolmap = self.getselect() File \"\", line 267, in getselect sql, params = col.selectformat(self, sql, params) File \"\", line 385, in selectformat if hasattr(self.outputfield, 'selectformat'): File \"\", line 48, in get res = [] = (instance) File \"\", line 266, in outputfield outputfield = self.resolveoutputfield() File \"\", line 463, in resolveoutputfield return super().resolveoutputfield() File \"\", line 306, in resolveoutputfield , Expression contains mixed types: IntegerField, AutoField. You must set output_field. This should be a regression bug because the previous query works in Django 3.1"}
{"_id":"q-en-django__django-13371-655a2d94be93c526d64d60054fe5f565fbc84d1903b6874e0a0f495fb72204a0","text":"is not pickleable. Description (last modified by Mariusz Felisiak) The new named parameter of QuerySet.valueslist() was released In Django 2.0 (). But resulted namedtuple-s can't be pickled: class ModelA(models.Model): value = models.CharField(maxlength=12) In [12]: row = ModelA.objects.values_list('id', 'value', named=True).first() In [14]: type(row) Out[14]: In [16]: (row) PicklingError: Can't pickle <class '': attribute lookup Row on failed In particular, as a result, such requests do not work with cacheops package."}
{"_id":"q-en-django__django-13386-07fd9dcc394a572b6244cc832efb25e3c2560ef2245b949ff73f8199fde0fe44","text":"Convert maxage to an int in setcookie() Description The max-age attribute of cookie is supposed to be an integer ​ I think it would be helpful to convert the maxage parameter of setcookie() to an integer for the user. The benefit is simply that there are some cookie parsers that don't handle decimals gracefully. It's pretty easy to pass in a float without understanding the consequences. I spent a good chunk of time today trying to track down the problem. Things to consider: Do we only convert floats where the decimal part is 0? Or do we round or truncate? If we can't successfully convert to an int, do we throw an exception, or just pass in the original value?"}
{"_id":"q-en-django__django-13401-ac3e44691e598dfb3d5c5cddc8519c2931c213ceef5bfe48de313af61ffa2e42","text":"Abstract model field should not be equal across models Description Consider the following models: class A(models.Model): class Meta: abstract = True myfield = IntegerField() class B(A): pass class C(A): pass If I pull the fields of B and C into a shared set, one will be de-duplicated away, because they compare as equal. I found this surprising, though in practice using a list was sufficient for my need. The root of the issue is that they compare equal, as fields only consider self.creationcounter when comparing for equality. len({B.meta.getfield('myfield'), C.meta.getfield('myfield')}) == 1 B.meta.getfield('myfield') == C.meta.getfield('myfield') We should adjust eq so that if the is different, they will compare unequal. Similarly, it is probably wise to adjust hash and lt to match. When adjusting lt, it may be wise to order first by self.creationcounter so that cases not affected by this equality collision won't be re-ordered. In my experimental branch, there was one test that broke if I ordered them by model first. I brought this up on IRC django-dev to check my intuitions, and those conversing with me there seemed to agree that the current behavior is not intuitive."}
{"_id":"q-en-django__django-13406-a707ef95d7b38f6dd85f9dbaba176184e93cbab897337bfcaeccc8b8de31093c","text":"Queryset with values()\/valueslist() crashes when recreated from a pickled query. Description I am pickling query objects () for later re-evaluation as per ​ However, when I tried to rerun a query that combines values and annotate for a GROUP BY functionality, the result is broken. Normally, the result of the query is and should be a list of dicts, but in this case instances of the model are returned, but their internal state is broken and it is impossible to even access their .id because of a AttributeError: 'NoneType' object has no attribute 'attname' error. I created a minimum reproducible example. from import models class Toy(models.Model): name = models.CharField(maxlength=16) material = models.CharField(maxlength=16) price = models.PositiveIntegerField() crashing code import pickle from import Sum from import Toy (name='foo', price=10, material='wood') (name='bar', price=20, material='plastic') (name='baz', price=100, material='wood') prices = ('material').annotate(totalprice=Sum('price')) print(prices) print(type(prices[0])) prices2 = () = (()) print(type(prices2[0])) print(prices2) The type of prices[0] is reported as 'dict', which is ok, the type of prices2[0] is reported as '<class \"models.Toy\"', which is wrong. The code then crashes when trying to print the evaluated queryset with the following: Traceback (most recent call last): File \"\", line 19, in <moduleprint(prices2) File \"\/home\/beda\/virtualenvs\/celus\/lib\/python3.6\/site-\", line 253, in repr return '<%s %r' % (, data) File \"\/home\/beda\/virtualenvs\/celus\/lib\/python3.6\/site-\", line 519, in repr return '<%s: %s' % (, self) File \"\/home\/beda\/virtualenvs\/celus\/lib\/python3.6\/site-\", line 522, in str return '%s object (%s)' % (, ) File \"\/home\/beda\/virtualenvs\/celus\/lib\/python3.6\/site-\", line 569, in getpkval return getattr(self, ) File \"\/home\/beda\/virtualenvs\/celus\/lib\/python3.6\/site-\", line 133, in get val = self.checkparentchain(instance, self.fieldname) File \"\/home\/beda\/virtualenvs\/celus\/lib\/python3.6\/site-\", line 150, in checkparentchain return getattr(instance, link_field.attname) AttributeError: 'NoneType' object has no attribute 'attname' From my point of view it seems as though Django retrieves the correct data from the database, but instead of returning them as a dict, it tries to create model instances from them, which does not work as the data has wrong structure."}
{"_id":"q-en-django__django-13410-e0b416f7ac082b76c7ba659c7d5cf6920ae2a62340895a41604fc1adb992c974","text":"Bug in posix implementation of Description The posix version of locks (the version which supports import fcntl) has a bug. The code attempts to return True to indicate success or failure acquiring a lock, but instead it always returns False. The reason is that cpython fcntl module returns None if successful, and raises an OSError to indicate failure (see ​). Anyone interested in using the non-blocking (i.e. locks.LOCKSNB) requires a valid return value to know if they have successfully acquired the lock. I believe the correct implementation should be the following: diff --git index .. --- +++ -107,9 +107,15 else: return True else: def lock(f, flags): ret = (fd(f), flags) return ret == 0 try: (fd(f), flags) return True except OSError: return False def unlock(f): ret = (fd(f), fcntl.LOCKUN) return ret == 0 try: (fd(f), fcntl.LOCK_UN) return True except OSError: return False"}
{"_id":"q-en-django__django-13413-674f875d6d3803b77cc4122d810ed889d667aa11d38e963f938423f63d471e01","text":"BooleanFieldListFilter doesn't respect field choices. Description If I have such construction: class BoolTest(models.Model): BOUGHTCHOICES = ( (False, 'Pending'), (True, 'Bought') ) bought = models.BooleanField( verbosename=\"Fancy Boolean\", default=False, choices=BOUGHTCHOICES) class BoolTestAdmin(admin.ModelAdmin): listfilter = ('bought',) (BoolTest, BoolTestAdmin) The boolean Filter text is not modified to fit choices param Example (in FR): FILTRE Par Fancy Boolean Tout Oui Non Should be : FILTRE Par Fancy Boolean Tout Bought Pending"}
{"_id":"q-en-django__django-13417-305db860ecce12b672cb0221835374686a09cfb03b466748d69dc2a1ebaf4e08","text":"QuerySet.ordered property is incorrect for GROUP BY queries on models with Meta.ordering. Description Using the annotate function on a queryset doesn't keep the default ordering set in model's meta class. A property should say whether the queryset will be ordered or not. I wanted to use the qs.ordered property for this but it seems to stay truthy, even if the resulting SQL query will not have an ORDER BY clause. Example: qs = () ​ ​ qs.ordered # =True # =True ​ qs2 = Foo.objects.annotate(Count(\"pk\")).all() ​ ​ qs2.ordered # =True # =True If it can help : I'm using PostgreSQL"}
{"_id":"q-en-django__django-13426-de0124345e825623435cb66c30d101c30121b4265189c10f437c679caf074fe2","text":"Combined queryset crash on combined querysets with ordering. Description I have the following models. from import models class M1(models.Model): id = models.AutoField(primarykey=True,blank=True, null=True) f1 = models.TextField(blank=True, null=True) class Meta: managed = False dbtable = 'm1' class M2(models.Model): id = models.AutoField(primarykey=True,blank=True, null=True) f2 = models.TextField(blank=True, null=True) class Meta: managed = False dbtable = 'm2' class M3(models.Model): id = models.AutoField(primarykey=True,blank=True, null=True) f3 = models.TextField(blank=True, null=True) class Meta: managed = False dbtable = 'm3' Based on these models, I perform the following query. o1 = ('default') o2 = ('default') u1 = (o2) q = u1.orderby('-f2') o3 = ('default') res = q.union(o3) print(()) Unfortunately, this query crashes with a TypeError exception. The track trace is Traceback (most recent call last): File \"\", line 19, in <moduleprint(()) File \"\", line 411, in count return () File \"\", line 517, in getcount number = obj.getaggregation(using, ['count'])['count'] File \"\", line 485, in getaggregation outerquery.addsubquery(innerquery, using) File \"\", line 162, in addsubquery self.subquery, self.subparams = query.getcompiler(using).assql(withcolaliases=True) File \"\", line 507, in assql result, params = self.getcombinatorsql(combinator, ) File \"\", line 443, in getcombinatorsql if compiler.getorderby(): File \"\", line 368, in getorderby for idx, (selexpr, , col_alias) in enumerate(): TypeError: 'NoneType' object is not iterable"}
{"_id":"q-en-django__django-13431-67c75ba7d3d5915f2cb00e8665ef646229fb82f0d404dad02d62e03f2e858f37","text":"QuerySet.aggregate() mixes annotated fields names. Description I have the following model (simplified for opening the issue). class T(models.Model): id = models.AutoField(primary=True) foo = models.IntegerField() and I perform the following query on MySQL (simplified for opening the issue) T.objects.annotate(anon=F('foo')).aggregate(foo=Max(F('anon')), sum=Sum(F('foo'))) this produces the following SQL query SELECT MAX(), SUM() FROM (SELECT . AS FROM ) subquery that causes the following exception (1054, \"Unknown column 'foo' in 'field list'\") I would expect django to generate SELECT MAX(), SUM() FROM (SELECT . AS FROM ) subquery"}
{"_id":"q-en-django__django-13447-6f3f01f6d05bfcd96dc85cdc2230cdcb5622a6db1546c561f23bc2de0b7a7090","text":"Added model class to applist context Description (last modified by Raffaele Salmaso) I need to manipulate the applist in my custom admin view, and the easiest way to get the result is to have access to the model class (currently the dictionary is a serialized model). In addition I would make the buildappdict method public, as it is used by the two views index and appindex."}
{"_id":"q-en-django__django-13448-5d4a8ec08a5cb7d54b0b1790cf95bb85a4d4dee86df00ea9c17dda9007493792","text":"Test runner setupdatabases crashes with \"TEST\": {\"MIGRATE\": False}. Description I'm trying to upgrade a project from Django 3.0 to Django 3.1 and wanted to try out the new \"TEST\": {\"MIGRATE\": False} database setting. Sadly I'm running into an issue immediately when running test. Removing the \"TEST\": {\"MIGRATE\": False} line allows the tests to run. So this is not blocking the upgrade for us, but it would be nice if we were able to use the new feature to skip migrations during testing. For reference, this project was recently upgraded from Django 1.4 all the way to 3.0 so there might be some legacy cruft somewhere that triggers this. Here's the trackeback. I'll try to debug this some more. Traceback (most recent call last): File \"\/usr\/local\/lib\/python3.6\/site-\", line 84, in execute return (sql, params) relation \"djangoadminlog\" does not exist LINE 1: \", \"djangoadminlog\".\"changemessage\" FROM \"djangoad... ^ The above exception was the direct cause of the following exception: Traceback (most recent call last): File \"\/usr\/local\/lib\/python3.6\/site-\", line 1156, in executesql cursor.execute(sql, params) File \"\/usr\/local\/lib\/python3.6\/site-\", line 66, in execute return self.executewithwrappers(sql, params, many=False, executor=self.execute) File \"\/usr\/local\/lib\/python3.6\/site-\", line 75, in executewithwrappers return executor(sql, params, many, context) File \"\/usr\/local\/lib\/python3.6\/site-\", line 84, in execute return (sql, params) File \"\/usr\/local\/lib\/python3.6\/site-\", line 90, in exit raise djexcvalue.withtraceback(traceback) from excvalue File \"\/usr\/local\/lib\/python3.6\/site-\", line 84, in execute return (sql, params) relation \"djangoadminlog\" does not exist LINE 1: \", \"djangoadminlog\".\"changemessage\" FROM \"djangoad... ^ During handling of the above exception, another exception occurred: Traceback (most recent call last): File \"\", line 15, in <modulemain() File \"\", line 11, in main executefromcommandline() File \"\/usr\/local\/lib\/python3.6\/site-\", line 401, in executefromcommandline utility.execute() File \"\/usr\/local\/lib\/python3.6\/site-\", line 395, in execute self.fetchcommand(subcommand).runfromargv() File \"\/usr\/local\/lib\/python3.6\/site-\", line 23, in runfromargv super().runfromargv(argv) File \"\/usr\/local\/lib\/python3.6\/site-\", line 330, in runfromargv self.execute(args, cmdoptions) File \"\/usr\/local\/lib\/python3.6\/site-\", line 371, in execute output = (args, options) File \"\/usr\/local\/lib\/python3.6\/site-\", line 53, in handle failures = testrunner.runtests(testlabels) File \"\/usr\/local\/lib\/python3.6\/site-\", line 695, in runtests oldconfig = self.setupdatabases(aliases=databases) File \"\/usr\/local\/lib\/python3.6\/site-\", line 616, in setupdatabases self.parallel, kwargs File \"\/usr\/local\/lib\/python3.6\/site-\", line 174, in setupdatabases serialize=connection.settingsdict['TEST'].get('SERIALIZE', True), File \"\/usr\/local\/lib\/python3.6\/site-\", line 78, in createtestdb self.connection.testserializedcontents = self.serializedbtostring() File \"\/usr\/local\/lib\/python3.6\/site-\", line 121, in serializedbtostring serializers.serialize(\"json\", getobjects(), indent=None, stream=out) File \"\/usr\/local\/lib\/python3.6\/site-\", line 128, in serialize s.serialize(queryset, options) File \"\/usr\/local\/lib\/python3.6\/site-\", line 90, in serialize for count, obj in enumerate(queryset, start=1): File \"\/usr\/local\/lib\/python3.6\/site-\", line 118, in getobjects yield from queryset.iterator() File \"\/usr\/local\/lib\/python3.6\/site-\", line 360, in iterator yield from self.iterableclass(self, chunkedfetch=usechunkedfetch, chunksize=chunksize) File \"\/usr\/local\/lib\/python3.6\/site-\", line 53, in iter results = compiler.executesql(chunkedfetch=self.chunkedfetch, chunksize=self.chunksize) File \"\/usr\/local\/lib\/python3.6\/site-\", line 1159, in executesql () cursor \"djangocurs139860821038912sync_1\" does not exist"}
{"_id":"q-en-django__django-13449-48706698bfda2f6970b4c9a34e930897b6ded8a15b7140918dda61263ef360ea","text":"Lag() with DecimalField crashes on SQLite. Description On Django 3.0.7 with a SQLite database using the following model: from import models class LagTest(models.Model): modified = models.DateField() data = models.FloatField() amount = models.DecimalField(decimalplaces=4, maxdigits=7) and the following query from import F from import Lag from import Window from import LagTest w = Window(expression=Lag('amount',7), partitionby=[F('modified')], orderby=F('modified').asc()) q = ().annotate(w=w) generates the following error: In [12]: print(q) OperationalError Traceback (most recent call last) C:\\ProgramData\\Anaconda3\\envs\\djbase\\lib\\site-packages\\django\\db\\backends\\ in execute(self, sql, params, ignoredwrapperargs) 85 else: ---86 return (sql, params) 87 C:\\ProgramData\\Anaconda3\\envs\\djbase\\lib\\site-packages\\django\\db\\backends\\sqlite3\\ in execute(self, query, params) 395 query = self.convertquery(query) --396 return Database.Cursor.execute(self, query, params) 397 OperationalError: near \"OVER\": syntax error The above exception was the direct cause of the following exception: OperationalError Traceback (most recent call last) <ipython-input-12-996617e96a38in <module----1 print(q) C:\\ProgramData\\Anaconda3\\envs\\djbase\\lib\\site-packages\\django\\db\\models\\ in repr(self) 250 251 def repr(self): --252 data = list(self[:REPROUTPUTSIZE + 1]) 253 if len(data) REPROUTPUTSIZE: 254 data[-1] = \"...(remaining elements truncated)...\" C:\\ProgramData\\Anaconda3\\envs\\djbase\\lib\\site-packages\\django\\db\\models\\ in iter(self) 274 - Responsible for turning the rows into model objects. 275 \"\"\" --276 self.fetchall() 277 return iter(self.resultcache) 278 C:\\ProgramData\\Anaconda3\\envs\\djbase\\lib\\site-packages\\django\\db\\models\\ in fetchall(self) 1259 def fetchall(self): 1260 if self.resultcache is None: -1261 self.resultcache = list(self.iterableclass(self)) 1262 if self.prefetchrelatedlookups and not self.prefetchdone: 1263 self.prefetchrelatedobjects() C:\\ProgramData\\Anaconda3\\envs\\djbase\\lib\\site-packages\\django\\db\\models\\ in iter(self) 55 # Execute the query. This will also fill , klassinfo, 56 # and annotations. ---57 results = compiler.executesql(chunkedfetch=self.chunkedfetch, chunksize=self.chunksize) 58 select, klassinfo, annotationcolmap = (, compiler.klassinfo, 59 compiler.annotationcolmap) C:\\ProgramData\\Anaconda3\\envs\\djbase\\lib\\site-packages\\django\\db\\models\\sql\\ in executesql(self, resulttype, chunkedfetch, chunksize) 1150 cursor = () 1151 try: -1152 cursor.execute(sql, params) 1153 except Exception: 1154 # Might fail for server-side cursors (e.g. connection closed) C:\\ProgramData\\Anaconda3\\envs\\djbase\\lib\\site-packages\\django\\db\\backends\\ in execute(self, sql, params) 98 def execute(self, sql, params=None): 99 with self.debugsql(sql, params, uselastexecutedquery=True): --100 return super().execute(sql, params) 101 102 def executemany(self, sql, paramlist): C:\\ProgramData\\Anaconda3\\envs\\djbase\\lib\\site-packages\\django\\db\\backends\\ in execute(self, sql, params) 66 67 def execute(self, sql, params=None): ---68 return self.executewithwrappers(sql, params, many=False, executor=self.execute) 69 70 def executemany(self, sql, paramlist): C:\\ProgramData\\Anaconda3\\envs\\djbase\\lib\\site-packages\\django\\db\\backends\\ in executewithwrappers(self, sql, params, many, executor) 75 for wrapper in reversed(): 76 executor = functools.partial(wrapper, executor) ---77 return executor(sql, params, many, context) 78 79 def execute(self, sql, params, ignoredwrapperargs): C:\\ProgramData\\Anaconda3\\envs\\djbase\\lib\\site-packages\\django\\db\\backends\\ in execute(self, sql, params, ignoredwrapperargs) 84 return (sql) 85 else: ---86 return (sql, params) 87 88 def executemany(self, sql, paramlist, ignoredwrapperargs): C:\\ProgramData\\Anaconda3\\envs\\djbase\\lib\\site-packages\\django\\db\\ in exit(self, exctype, excvalue, traceback) 88 if djexctype not in (DataError, IntegrityError): 89 self.wrapper.errorsoccurred = True ---90 raise djexcvalue.withtraceback(traceback) from excvalue 91 92 def call(self, func): C:\\ProgramData\\Anaconda3\\envs\\djbase\\lib\\site-packages\\django\\db\\backends\\ in execute(self, sql, params, ignoredwrapperargs) 84 return (sql) 85 else: ---86 return (sql, params) 87 88 def executemany(self, sql, paramlist, ignoredwrapperargs): C:\\ProgramData\\Anaconda3\\envs\\djbase\\lib\\site-packages\\django\\db\\backends\\sqlite3\\ in execute(self, query, params) 394 return Database.Cursor.execute(self, query) 395 query = self.convertquery(query) --396 return Database.Cursor.execute(self, query, params) 397 398 def executemany(self, query, paramlist): OperationalError: near \"OVER\": syntax error The generated SQL query is: SELECT \"test1lagtest\".\"id\", \"test1lagtest\".\"modified\", \"test1lagtest\".\"data\", \"test1lagtest\".\"amount\", CAST(LAG(\"test1lagtest\".\"amount\", 7) AS NUMERIC) OVER (PARTITION BY \"test1lagtest\".\"modified\" ORDER BY \"test1lagtest\".\"modified\" ASC) AS \"w\" FROM \"test1lagtest\" I believe this fails as the CAST() statement ends after LAG whereas it should be around the whole statement up until \"w\" This only applies where the lagged field is a DecimalField e.g. w = Window(expression=Lag('data',7), partitionby=[F('modified')], orderby=F('modified').asc()) works correctly. I can override it by adding outputfield=FloatField() to the Lag function e.g. w = Window(expression=Lag('amount',7,outputfield=FloatField()), partitionby=[F('modified')], orderby=F('modified').asc())"}
{"_id":"q-en-django__django-13454-6800372905242cd49a377735ac3d817915d54305bf261bf530e3aec3878dc07a","text":"Using EmptyFieldListFilter with GenericForeignKey and GenericRelation crashes. Description I have a GenericForeignKey in my model: contenttyperesourcecontract = models.ForeignKey( ContentType, ondelete=models.CASCADE, blank=True, null=True, ) resourcecontractid = models.PositiveIntegerField(blank=True, null=True) resourcecontract = GenericForeignKey('contenttyperesource', 'resourcecontractid') and I want to use the new admin.EmptyFieldListFilter in my model admin: class myAdmin(admin.ModelAdmin): listfilter = (('resourcecontract', admin.EmptyFieldListFilter),) But when I try to run it I get a \"'GenericForeignKey' object has no attribute 'emptystringsallowed'\". It will work fine if I use the resourcecontractid field. Would it make sense to extend the GenericForeignKey to use those emptystrings_allowed attributes from fields that are used in this generic relation?"}
{"_id":"q-en-django__django-13458-e06cb420a6e132dd601dee41374e14c8e394c7b5015e6c8bfacb41dfc82800b0","text":"Incorrect messaging when validatemin\/validatemax and minnum == maxnum. Description When using Django Formset, the error messaging isn't checking for the condition where both minnum and maxnum are equal and both validatemin and validatemax are set to true. Code highlighting: class TestForm(forms.Form): msg = forms.CharField() testformset = formsetfactory( TestForm, minnum=2, maxnum=2, validatemin=True, validatemax=True) When formset is created in the following way and both validate flags are set True the following error messages show up If the supplied forms are less than two - please submit 2 or more forms expected please submit 2 forms similarly the reverse is also true when the forms are more than two it gives the error message saying please submit 2 or fewer forms expected please submit 2 forms This was a ​ bug reported on Wagtail and after investigating a little I noticed the incorrect messaging was coming from ​ this part in the validation"}
{"_id":"q-en-django__django-13460-da9dd3f76886289bfd639d5ccbd2410c770ac8fad7f421664cfc971f0e07bb81","text":"String formatting error when passing floats as values in {% blocktrans %} tags Description With the following template code: {% blocktrans count counter=person.distanceinmiles|floatformat:0 %}{{ counter }} mile away{% plural %}{{ counter }} miles away{% endblocktrans %} And a russian translation: #, python-format msgid \"%(counter)s mile away\" msgidplural \"%(counter)s miles away\" msgstr[0] \"На расстоянии %(counter)s мили\" msgstr[1] \"На расстоянии %(counter)s миль\" msgstr[2] \"На расстоянии %(counter)s миль\" Rendering the template fails with a String formatting error: \"TypeError: not all arguments converted during string formatting\". This happens because gettext string formatting fails when a float is passed. Removing the floatformat and casting the value as int works fine. This could be improved by either: Swallow TypeError and throw a friendlier message saying the type might not be compatible with gettext's string formatting (not a fan of swallowing TypeError though, we need to make sure nothing else in dotranslate throws such an exception) Actively checking the type's compatibility with the string format"}
{"_id":"q-en-django__django-13466-d46760c326faa85343090a52ba137f98f709e0656608742b735522af8a7f4d4d","text":"Required mutually exclusive groups don't work with boolean arguments. Description I have the following management command (called ): from import BaseCommand class Command(BaseCommand): def addarguments(self, parser): group = parser.addmutuallyexclusivegroup(required=True) group.addargument('--value', type=str) group.addargument('--flag', action='storetrue') def handle(self, args, *options): pass Running test --flag or --value foo, the command works as expected. Using callcommand to call the command from my code fails if I pass --flag, e.g.: from import callcommand callcommand('test', flag=True) This gets me the following traceback: argparse.ArgumentError: argument --flag: ignored explicit argument 'True' During handling of the above exception, another exception occurred: Traceback (most recent call last): File \"<input\", line 1, in <moduleFile \"\", line 147, in callcommand defaults = parser.parseargs(args=parseargs) File \"\", line 57, in parseargs return super().parseargs(args, namespace) File \"\", line 1780, in parseargs args, argv = self.parseknownargs(args, namespace) File \"\", line 1819, in parseknownargs (str(err)) File \"\", line 63, in error raise CommandError(\"Error: %s\" % message) Error: argument --flag: ignored explicit argument 'True'"}
{"_id":"q-en-django__django-13484-822c16f5a1e7a60781973d93c97fcf6d39c0b8e2981efcea3eded66c5b3381b5","text":"Queryset crashes when recreated from a pickled query with FilteredRelation used in aggregation. Description I am pickling query objects () for later re-evaluation as per ​ However, when I tried to rerun a query that contains a FilteredRelation inside a filter, I get an missing FROM-clause entry for table \"t3\" error. I created a minimum reproducible example. from import models class Publication(models.Model): title = models.CharField(maxlength=64) class Session(models.Model): TYPECHOICES = (('A', 'A'), ('B', 'B')) publication = models.ForeignKey(Publication, ondelete=models.CASCADE) sessiontype = models.CharField(choices=TYPECHOICES, default='A', maxlength=1) place = models.CharField(maxlength=16) value = models.PositiveIntegerField(default=1) The actual code to cause the crash: import pickle from import FilteredRelation, Q, Sum from import Publication, Session p1 = (title='Foo') p2 = (title='Bar') (publication=p1, sessiontype='A', place='X', value=1) (publication=p1, sessiontype='B', place='X', value=2) (publication=p2, sessiontype='A', place='X', value=4) (publication=p2, sessiontype='B', place='X', value=8) (publication=p1, sessiontype='A', place='Y', value=1) (publication=p1, sessiontype='B', place='Y', value=2) (publication=p2, sessiontype='A', place='Y', value=4) (publication=p2, sessiontype='B', place='Y', value=8) qs = ().annotate( relevantsessions=FilteredRelation('session', condition=Q(sessionsessiontype='A')) ).annotate(x=Sum('relevantsessionsvalue')) print(list(qs)) qs2 = () = (()) print(list(qs2)) In the crashing query, there seems to be a difference in the tablemap attribute - this is probably where the t3 table is coming from. Please let me know if there is any more info required for hunting this down. Cheers Beda p.s.- I also tried in Django 3.1 and the behavior is the same. p.p.s.- just to make sure, I am not interested in ideas on how to rewrite the query - the above is a very simplified version of what I use, so it would probably not be applicable anyway."}
{"_id":"q-en-django__django-13490-5e9a921be7f7b23f9681315d17add5b875ddbfd234b0cd16aacf48c1d28344ef","text":"Raise a descriptive error on get() with filters operation following (), intersection(), and difference(). Description The QuerySet resulting from q1.difference(q2) does not allow to retrieve one specific object of that resulting QuerySet. Minimal example: from import models class Test(models.Model): name = models.CharField(max_length=30) python shell from import Test qs1 = () qs1 <QuerySet [<Test: Test object (1), <Test: Test object (2), <Test: Test object (3)]qs2 = (id=1) qs2 <QuerySet [<Test: Test object (1)]qs3 = qs1.difference(qs2) qs3 <QuerySet <Test: Test object (2), <Test: Test object (3) Traceback (most recent call last): File \"<console\", line 1, in <moduleFile \"\/lib\/python3.8\/site-\", line 433, in get raise ( get() returned more than one Test -- it returned 2! Django version: 3.1.2 Python version: 3.8.5 OS: Arch Linux I also experienced this in the regular request\/view-context, with other Django versions (2.2) and other python versions (3.7). Sorry if this is the expected behavior, a known bug or if I missed something which changes the behavior only on my system. If you need more information, I'll be happy to assist."}
{"_id":"q-en-django__django-13512-a9b7aee59960bf773d6e76f8823ffd590ac687d6d49a462c77f220665655f15d","text":"Admin doesn't display properly unicode chars in JSONFields. Description (last modified by ZhaoQi99) import json print ('') \"\\u4e2d\\u56fd\" use ASCII encoding by default when serializing Chinese. So when we edit a JsonField which contains Chinese character in Django admin,it will appear in ASCII characters. I have try to fix this this problem in ​ it works prefectly."}
{"_id":"q-en-django__django-13513-f6a8a3fad8ee69f739e3cd50f97a02a65d4da7e12e8c474676b85fcadbd36aa2","text":"debug error view doesn't respect exc.suppresscontext (PEP 415) Description Consider the following view that raises an exception: class TestView(View): def get(self, request, args, *kwargs): try: raise RuntimeError('my error') except Exception as exc: raise ValueError('my new error') from None Even though the raise is from None, unlike the traceback Python shows, the debug error view still shows the RuntimeError. This is because the explicitorimplicitcause() function inside gettracebackframes() doesn't respect exc.suppresscontext, which was introduced in Python 3.3's PEP 415: ​ def gettracebackframes(self): def explicitorimplicitcause(excvalue): explicit = getattr(excvalue, 'cause', None) implicit = getattr(excvalue, 'context', None) return explicit or implicit Instead, it should be something more like (simplifying also for Python 3): def explicitorimplicitcause(excvalue): return ( or (None if excvalue.suppresscontext else exc_value.context) )"}
{"_id":"q-en-django__django-13516-a1ad13fc98176224a571a5fca29c5134eea9972ecb3027c2901ec7f5071e837b","text":"flush() on management commands doesn't work. Description flush() is notably called during migrate command; it doesn't work, and a long migration effectively prints to stderr no relevant information up until the end: Operations to perform: Apply all migrations: myapp Running migrations: Then nothing more, but the migration is being done. Then at the end of the real migration, the rest is flushed: Applying myapp.0002auto202008171030... OK Expected behavior: Operations to perform: Apply all migrations: myapp Running migrations: Applying myapp.0002auto202008171030... then work then OK"}
{"_id":"q-en-django__django-13528-93df0532e896e5561b07617a69c975b76cbaab6d7e718fce239ce6ecb1b95c1e","text":"Incorrect separators when chaining floatformat to intcomma in some locales Description When u use floatvalue \"2\" and intcomma together in a template the output of intcomma won't be internationalized. Since intcomma wont work with decimals in django 1.5.1 i tried to convert a decimal to a float in a template, but it wont give me the excepted output. When i have the value of 1000.11 it should be 1000,11 in germany, with intcomma(float(1000,11)) i get 1.000,11. But when i use Decimal(1000,11)intcomma, i will get 1,000,11. Thats a bug or maybe an unwanted behavior."}
{"_id":"q-en-django__django-13530-a10af4fa1879d03e77323d1b2b5271454921ac3c69b63b8986c2c6bdaffcd928","text":"Using KeyTransform for JSONField produces invalid SQL in various places. Description (last modified by Igor Jerosimić) Using KeyTransform in ordering attribute of ArrayAgg function produces invalid SQL. I don't know if it matters but I'm using Postgres for DB. from import models class Parent(models.Model): name = models.CharField(default='test') class Child(models.Model): parent = models.ForeignKey( Parent, ondelete=models.SETNULL, relatedname='children', ) data = models.JSONField(default=dict) parent = () (parent=parent, data={'en': 'English', 'fr': 'French'}) Parent.objects.annotate( childrenarray=ArrayAgg( KeyTextTransform('en', 'childrendata'), distinct=True, ordering=[KeyTransform('en', 'childrendata')], ), ).all() Produces invalid SQL in the ORDER BY section: ARRAY_AGG(DISTINCT (\"children\".\"data\" -'default') ORDER BY None(\"children\".\"data\")) NOTE: This was working fine before Django 3.1."}
{"_id":"q-en-django__django-13537-a8cf4bd5aa4d00e0e9e0c1aff5af712b944ba06fc9ba9b0ce2fce3cc894e524c","text":"MySQL: dbshell does not get charset from DATABASES setting Description I noticed that dbshell doesn't respect the databaseoptions. I ran into an issue with an application we are creating that needs to support mysql and postgre at least, we execute some sql scripts that get piped to dbshell (to avoid hardcoding psql -U xxx or mysql -u xxx and creating 2 scripts every time). When running an utf8 database with utf8 as our charset in databaseoptions, we ran into some weird encoding issues. The solution for us was to learn to respect the encoding settings in Are you opposed to something like this? Attaching small patch that fixes our problem. Let me know if it needs extending to support other backends or database_options."}
{"_id":"q-en-django__django-13551-2eb2f8260fd7c3f24f4aee9f69f470664997e736fab3defcf2d4bbc73eac2682","text":"Changing user's email could invalidate password reset tokens Description Sequence: Have account with email address foo Password reset request for that email (unused) foo account changes their email address Password reset email is used The password reset email's token should be rejected at that point, but in fact it is allowed. The fix is to add the user's email address into ​PasswordResetTokenGenerator.makehashvalue() Nothing forces a user to even have an email as per AbstractBaseUser. Perhaps the token generation method could be factored out onto the model, ala getsessionauthhash()."}
{"_id":"q-en-django__django-13553-0289b3edc180b0f220ed7f8cd3567752cbefb8eb7d485c589d2e496a1207cd04","text":"QuerySet.orderby() crashes on union() queryset with a single non-empty query. Description (last modified by Hannes Ljungberg) I noticed this error while implementing a dynamic union like this: union = reduce( lambda qs1, qs2: (qs2), querysets, (), ) If len(querysets) == 1 it will result in a pretty weird query throwing a database error when both querysets are ordered, another example recreating this in the test suite: qs1 = ().orderby(‘pk’) qs2 = ().union(qs1).orderby(‘pk’) Executing qs2 result in the following query: (SELECT \"queriesnumber\".\"id\", \"queriesnumber\".\"num\", \"queriesnumber\".\"othernum\", \"queriesnumber\".\"anothernum\" FROM \"queriesnumber\" ORDER BY \"queriesnumber\".\"id\" ASC) ORDER BY (1) ASC Result in the following error on PostgresSQL: multiple ORDER BY clauses not allowed LINE 1: \" ORDER BY \"queriesnumber\".\"id\" DESC) ORDER BY (1) ASC And sqlite: ORDER BY not allowed in subqueries of compound statements. MySQL seems to accept this query but it's most likely not the query the user expects the be executed. My proposal is to simply return the non-empty query and make the union a no-op set in this case."}
{"_id":"q-en-django__django-13556-717655496678b527b3825d88189e1f1fa2cd86b3eb8df4f65f076ab143b98882","text":"Allowing null values while registering function on transaction oncommit Description So if I were to try this command:- transaction.oncommit(None) and run any sample tests which has simple django db post_save signal in the flow and it has functions which run on transaction commit, the code would crash with error TypeError: 'NoneType' object is not callable. So how about if we do not allow it to register as none beforehand or allow it to exit gracefully?"}
{"_id":"q-en-django__django-13560-8e09af2078a874ce89965799897063d0646edf0488679698a472ae57e9ffac0c","text":"ProtectedError\/RestrictedError raised from Collector.collect contains iterators. Description (last modified by Kwist) (\"Cannot delete some instances of model 'A' because they are referenced through protected foreign keys: 'A.protect', 'B.protect'.\", object at 0x7f96b5cb6c40) As a result, the repr of exception in Sentry is not informative."}
{"_id":"q-en-django__django-13568-e6a432861cb24c2833e3a2957880af6d2e20402233827505208601139dd0576a","text":"Skip auth.E003 system check for USERNAMEFIELD with total UniqueConstraints. Description Defining a user model like this: class User(AbstractBaseUser): username = models.CharField(maxlength=30) USERNAMEFIELD = \"username\" class Meta: constraints = [UniqueConstraint(fields=[\"username\"], name=\"userusernameunq\")] Will trigger auth.E003: auth.User: (auth.E003) 'User.username' must be unique because it is named as the 'USERNAMEFIELD'. Sometimes it’s not preferable to set the field as unique with unique=True as it will create an extra implicit *like index for CharField and TextField on PostgresSQL. The system check should be extended to check for the presence of USERNAMEFIELD in Model._meta.constraints. Not really sure if this classifies as a bug."}
{"_id":"q-en-django__django-13569-3810738f3eba00d7f3094ad23677435ed9e3b69952c2630fb476dac4d4d5d3d5","text":"orderby('?') unexpectedly breaking queryset aggregation Description Steps to reproduce: class Thing(models.Model): pass class Related(models.Model): models.ForeignKey(Thing) With data t = () rs = [(thing=t) for in range(2)] The following query works as expected. The aggregation with Count produces a GROUP BY clause on Thing.objects.annotate(rc=Count('related')).orderby('rc').values('id', 'rc') <QuerySet [{'id': 1, 'rc': 2}]This also works as expected (at least to me). Although there is an aggregation, ordering by related means that the grouping will be broken down. Thing.objects.annotate(rc=Count('related')).orderby('related').values('id', 'rc') <QuerySet [{'id': 1, 'rc': 1}, {'id': 1, 'rc': 1}]But the following seems wrong to me. Thing.objects.annotate(rc=Count('related')).orderby('?').values('id', 'rc') <QuerySet [{'id': 1, 'rc': 1}, {'id': 1, 'rc': 1}]The random function call has nothing to do with the aggregation, and I see no reason it should break it. Dumping the query seems that indeed the random call breaks the group by call: (I simpilfied the table names a little) print(Thing.objects.annotate(rc=Count('related')).orderby('?').values('id', 'rc').query) SELECT \"thing\".\"id\", COUNT(\"related\".\"id\") AS \"rc\" FROM \"thing\" LEFT OUTER JOIN \"related\" ON (\"thing\".\"id\" = \"related\".\"thingid\") GROUP BY \"thing\".\"id\", RANDOM() ORDER BY RANDOM() ASC I dug into the SQL compiler, and it seems to me the problem is inside , where the compiler combines all non-aggregate, non-ref orderby expressions into groupby. I patched it like this for expr, (sql, params, isref) in orderby: if expr.containsaggregate: continue if isref: continue ([ exp for exp in expr.getsource_expressions() if not isinstance(exp, Random) ]) and things seem to work correctly. No failed tests against SQLite3 with default settings."}
{"_id":"q-en-django__django-13578-93561407104efc848d2af6bbe865a183943a2c448f6b8d58b30075310b7563c8","text":"ManagementForm exception in case of bad prefix should be easier to understand Description If user adds multiple formsets with prefixes, and specifies the prefix incorrectly when binding the form and validating: someformset = SomeFormSet(request.POST, 'articles') instead of: someformset = SomeFormSet(request.POST, prefix='articles') Django \"suppresses\" the original exception and raises only relatively unhelpful \"ManagementForm data is missing or has been tampered with\". In file , line 57: if or form = ManagementForm(, , ) if not form.isvalid(): raise ValidationError('ManagementForm data is missing or has been tampered with') Suggestion: include in output, because for such a small bug in code, it can take a really long time find it. {'INITIALFORMS': [u'This field is required.'], 'MAXNUMFORMS': [u'This field is required.'], 'TOTALFORMS': [u'This field is required.']}"}
{"_id":"q-en-django__django-13585-0d6d76f41dbcb11b201b1138811b3d8e491b28224be379e76c6be29ef824f7ae","text":"Password reset token incompatibility. Description As noted here ​ the hashing for password reset tokens has changed between 3.0 and 3.1 and work has been done to ensure existing tokens will still work (at least until 4.0). However the encoding of the token creation time has also changed. Specifically from days since 1\/1\/01 to seconds since 1\/1\/01. And it appears no work has been done to support tokens with the older values. So a token generated on Oct 1, 2020 will come through as 7213 days which will then get interpreted as 7213 seconds, aka 2am Jan 1, 2001. So while exiting tokens in the wild will pass crypto validation they will all show as expired if your PASSWORDRESETTIMEOUT is less than ~20 years. The code base I'm working on uses these tokens (perhaps unwisely) in some email links that are expected to have a 3 month lifetime and an upgrade from 3.0 to 3.1 looks likely to render all the tokens in the wild expired which is suboptimal."}
{"_id":"q-en-django__django-13589-09745e62858b996c2181720299c261994fe62a039e0962044b75a8aac99f8c3e","text":"Saving parent object after setting on child leads to unexpected data loss in bulkcreate(). Description Example: class Country(models.Model): name = models.CharField(maxlength=255) isotwoletter = models.CharField(maxlength=2) description = models.TextField() class City(models.Model): name = models.CharField(maxlength=255) country = models.ForeignKey(Country, ondelete=models.CASCADE) class BulkCreateTests(TestCase): def testfkbug(self): countrynl = Country(name='Netherlands', isotwoletter='NL') countrybe = Country(name='Belgium', isotwoletter='BE') city = City(country=countrybe, name='Brussels') # (1) () # (2) () This results in an integrity error: ERROR: testfkbug () Traceback (most recent call last): File \"\", line 85, in execute return (sql, params) psycopg2.IntegrityError: null value in column \"countryid\" violates not-null constraint DETAIL: Failing row contains (1, Brussels, null). The above exception was the direct cause of the following exception: ... I wonder wether there's a reason that this doesn't work. If setting a related object on a model instance automatically sets instance.relatedobjectid, you'd expect this behavior to continue working if the related object receives its primary key after the referencing instance was initialized. Of course, switching lines (1) and (2) makes it all work, but this behavior means that bulk creating with related objects (with postgres returning the primary keys, bless it's heart) becomes more complex than it should be, forcing the user to use mappings or even set the foreign keys themselves: for countrydata, citydata in data: country = Country(countrydata) (country) city = City(country=country, citydata) (city) Country.objects.bulkcreate(countries) for city in cities: city.countryid = City.objects.bulk_create(cities) Ideally, the main instance would, when saved, 'reach into' its related objects and set the foreign key field like the loop does. Is there a reason this can't work?"}
{"_id":"q-en-django__django-13590-61a11270b275dfd158b2f49133f05d96aa19bb1895b8257c1299288b3b718afe","text":"Upgrading 2.23.0 causes named tuples used as arguments to range to error. Description I noticed this while upgrading a project from 2.2 to 3.0. This project passes named 2-tuples as arguments to range queryset filters. This works fine on 2.2. On 3.0 it causes the following error: TypeError: new_() missing 1 required positional argument: 'far'. This happens because goes into the tuple elements to resolve lookups and then attempts to reconstitute the tuple with the resolved elements. When it attempts to construct the new tuple it preserves the type (the named tuple) but it passes a iterator to it's constructor. NamedTuples don't have the code path for copying an iterator, and so it errors on insufficient arguments. The fix is to * expand the contents of the iterator into the constructor."}
{"_id":"q-en-django__django-13592-d8bc00a989207e485197b58f58fce5886b4f318fb54271efc2499a19373f03ac","text":"ManyToManyField does not respect the PositiveBigIntegerField in m2m intermediate table. Description When creating a relation between 2 models using PositiveBigIntegerField on Postgresql, the relation table is created using regular ints as the column type. This in turn leads to out of bound error when using large numbers for ids. from import models class Node(models.Model): id = models.PositiveBigIntegerField(primarykey=True) point = models.PointField() class Relation(models.Model): id = models.PositiveBigIntegerField(primarykey=True) nodes = models.ManyToManyField(Node) The generated table will look like this: Column Description -------------+---------+-----------+----------+------------------------------------------------+---------+--------------+------------- id relationid nodeid As you can see, the PositiveBigInteger is not respected and a regular int is set"}
{"_id":"q-en-django__django-13607-4cb6310bf3a94dd4601218e06aff91e187c0ff95949bdc3598c4c413332f7869","text":"BaseFormSet.isvalid() produces ValidationError when there is no management form Description I was torn between reporting this as a bug or a feature request, but then I thought if I make it a feature request it will most likely break a lot of Django apps and hence I guess that means it's more of a bug... Anyway so the line in question is (in Django version 1.6.1): for i in range(0, self.totalformcount()): the self.totalformcount() executes this line (in Django version 1.6.1): return min(self.managementform.cleaneddata[TOTALFORMCOUNT], self.absolutemax) ..which then raises this exception (in Django version 1.6.1): raise ValidationError( ('ManagementForm data is missing or has been tampered with'), code='missingmanagementform', ) That stack trace occurs if\/when a user submits a formset after stripping out the management form hidden fields. I have been using Django for a few years now and have never come across an exception being raised by a form\/formset isvalid() call before. So my point is that I believe this exception should never be allowed to leave the BaseFormSet.isvalid() call, because it is an oddball behaviour compared to the rest of the isvalid() implementations. I.e. I believe there should be a check in BaseFormSet.isvalid() which checks for the presence of a valid management form (first) and returns False if it is not present, as opposed to raising an exception. Yes I could wrap the isvalid() call in a try\/catch, but I believe this is an unnecessary hack caused by a bad design deviation of the implementation of the BaseFormSet.is_valid() method. I didn't bother creating a patch and test cases, because I have a feeling this will get rejected or something like that, but I just thought I should bring this up, as I can't find mention of it anywhere and it seems important to me."}
{"_id":"q-en-django__django-13615-205323c630dd04bb7afa81aa9db38bc647e3e4acee4315c1f7213757cc9ee70d","text":"Add warning in makemessages command if the localecode with flag is not correct Description (last modified by Sanyam Khurana) Hey Calude, What about normalizing the directory name to something that would just work. For example, No matter, if the developer is doing all these: python makemessages -l zhcn python makemessages -l zhCN python makemessages -l ZHCN python makemessages -l ZH-CN etc. we, just normalize the directory name to zhCN and it would work. I'm still about to read the code of makemessages command and probably if there are any more checks than just this, then we'll have to figure out another way all together."}
{"_id":"q-en-django__django-13616-3bce7b385d8f9ee0aebea07346bb0e6117e50920855e882359b0c1d7cfe9917c","text":"Unclear error raised by makemessages when \/locale does not exist under app directory Description I got this error message: CommandError: Unable to find a locale path to store translations for file I had to google the error message and got to this: ​ It would be better if the error message was simple and obvious so you could solve the problem yourself directly instead. I thought the error message would be better if it said 'We are looking for a directory called \"locale\"\" but after that change the command fails silently anyway. Maybe LOCALE_PATHS should also include the directory where is by default? Or some other error message should exist..."}
{"_id":"q-en-django__django-13617-43cc1f170c0ccb739a06904a02238a64505cb1c0472ca9ef438a7c21938414ba","text":"Aggregation over subquery annotation GROUP BY produces wrong results Description (last modified by Christian Klus) Starting in django 3.0.7, specifically after patch I noticed some of my more complex queries returning incorrect results. I think I've narrowed it down to a simpler test case: Example query: ().annotate( pubyear=TruncYear('pubdate') ).orderby().values('pubyear').annotate( totalpages=Sum('pages'), toprating=Subquery( ( pubdateyear=OuterRef('pubyear') ).orderby('rating').values('rating')[:1] ) ).values('pubyear', 'totalpages', 'toprating') Generated SQL on 3.0.6: SELECT djangodatetrunc('year', \"aggregationregressbook\".\"pubdate\") AS \"pubyear\", SUM(\"aggregationregressbook\".\"pages\") AS \"totalpages\", ( SELECT U0.\"rating\" FROM \"aggregationregressbook\" U0 WHERE djangodateextract('year', U0.\"pubdate\") = djangodatetrunc('year', \"aggregationregressbook\".\"pubdate\") ORDER BY U0.\"rating\" ASC LIMIT 1 ) AS \"toprating\" FROM \"aggregationregressbook\" GROUP BY djangodatetrunc('year', \"aggregationregressbook\".\"pubdate\"), \"toprating\" Generated SQL on current master: SELECT djangodatetrunc('year', \"aggregationregressbook\".\"pubdate\", NULL, NULL) AS \"pubyear\", SUM(\"aggregationregressbook\".\"pages\") AS \"totalpages\", ( SELECT U0.\"rating\" FROM \"aggregationregressbook\" U0 WHERE djangodateextract('year', U0.\"pubdate\") = djangodatetrunc('year', \"aggregationregressbook\".\"pubdate\", NULL, NULL) ORDER BY U0.\"rating\" ASC LIMIT 1 ) AS \"toprating\" FROM \"aggregationregressbook\" GROUP BY djangodatetrunc('year', \"aggregationregressbook\".\"pubdate\", NULL, NULL), ( SELECT U0.\"rating\" FROM \"aggregationregressbook\" U0 WHERE djangodateextract('year', U0.\"pubdate\") = djangodatetrunc('year', \"aggregationregressbook\".\"pubdate\", NULL, NULL) ORDER BY U0.\"rating\" ASC LIMIT 1 ), \"aggregationregressbook\".\"pubdate\" I see two separate issues here: \"aggregationregressbook\".\"pubdate\" is being to the group by clause incorrectly (this issue probably predates the patch mentioned above) Even though the subquery is in the select statement, the alias is not being used and instead the subquery is reevaluated. This nearly doubles the cost of one of my queries that is experiencing this problem. I don't know much about the ORM internals, but here is my naive patch for the second issue: diff --git index .. --- +++ -2220,7 +2220,7 class Query(BaseExpression): # the selected fields anymore. groupby = [] for expr in self.groupby: if isinstance(expr, Ref) and not in fieldnames: if isinstance(expr, Ref) and not in fieldnames + annotationnames: expr = self.annotations[] (expr) self.groupby = tuple(groupby) I'd appreciate anyone with deeper knowlege of the ORM to chime in and let me know if I'm on the right track. Tests are passing locally. The resulting query on master: SELECT djangodatetrunc('year', \"aggregationregressbook\".\"pubdate\", NULL, NULL) AS \"pubyear\", SUM(\"aggregationregressbook\".\"pages\") AS \"totalpages\", ( SELECT U0.\"rating\" FROM \"aggregationregressbook\" U0 WHERE djangodateextract('year', U0.\"pubdate\") = djangodatetrunc('year', \"aggregationregressbook\".\"pubdate\", NULL, NULL) ORDER BY U0.\"rating\" ASC LIMIT 1 ) AS \"toprating\" FROM \"aggregationregressbook\" GROUP BY djangodatetrunc('year', \"aggregationregressbook\".\"pubdate\", NULL, NULL), \"top_rating\""}
{"_id":"q-en-django__django-13620-c5adef142f55b52af9a6a617bed8e962383b0557f7c40141bc6b78639701362e","text":"Support for list arguments inside exclusive required groups Description (last modified by Mark Gajdosik) Are there any plans to add support for the following? from import BaseCommand class Command(BaseCommand): def addarguments(self, parser): group = parser.addmutuallyexclusivegroup(required=True) group.addargument('--foo', nargs='+', type=int) def handle(self, args, *options): pass When calling the above using callcommand: callcommand('callcommandtest', foo=[1, 2, 3]) callcommand('callcommandtest', '--foo=1', '--foo=2', '--foo=3') I can't think of any workarounds other than setting type=str (somehow, that works fine) and coercing manually. Thank you!"}
{"_id":"q-en-django__django-13658-2179c3374fb1f5108e8bc07248527d1a91dddf8bb553db3634cc72183682e3b6","text":"ManagementUtility instantiates CommandParser without passing already-computed prog argument Description ManagementUtility ​goes to the trouble to parse the program name from the argv it's passed rather than from def init(self, argv=None): = argv or [:] self.progname = ([0]) if self.progname == '': self.progname = 'python -m django' But then when it needs to parse --pythonpath and --settings, it ​uses the program name from parser = CommandParser(usage='%(prog)s subcommand [options] [args]', addhelp=False, allowabbrev=False) Above \"%(prog)s\" ​refers to [0]. Instead, it should refer to self.progname. This can fixed as follows: parser = CommandParser( prog=self.progname, usage='%(prog)s subcommand [options] [args]', addhelp=False, allowabbrev=False) I'm aware that executefromcommandline is a private API, but it'd be really convenient for me if it worked properly in my weird embedded environment where [0] is ​incorrectly None. If passing my own argv to executefromcommand_line avoided all the ensuing exceptions, I wouldn't have to modify [0] globally as I'm doing in the meantime."}
{"_id":"q-en-django__django-13660-e095f1e32ad7018ac715323ae2e63ce615ccbace08a8fe921e7457225531b618","text":"shell command crashes when passing (with -c) the python code with functions. Description The examples below use Python 3.7 and Django 2.2.16, but I checked that the code is the same on master and works the same in Python 3.8. Here's how ​python -c works: $ python -c <<EOF \" import django def f(): print(django.version) f()\" EOF 16 Here's how ​python -m django shell -c works (paths shortened for clarify): $ python -m django shell -c <<EOF \" import django def f(): print(django.version) f()\" EOF Traceback (most recent call last): File \"{sys.baseprefix}\", line 193, in runmoduleasmain \"main\", modspec) File \"{sys.baseprefix}\", line 85, in runcode exec(code, runglobals) File \"{}\/lib\/python3.7\/site-\", line 9, in <modulemanagement.executefromcommandline() File \"{}\/lib\/python3.7\/site-\", line 381, in executefromcommandline utility.execute() File \"{}\/lib\/python3.7\/site-\", line 375, in execute self.fetchcommand(subcommand).runfromargv() File \"{}\/lib\/python3.7\/site-\", line 323, in runfromargv self.execute(args, cmdoptions) File \"{}\/lib\/python3.7\/site-\", line 364, in execute output = (args, options) File \"{}\/lib\/python3.7\/site-\", line 86, in handle exec(options['command']) File \"<string\", line 5, in <moduleFile \"<string\", line 4, in f NameError: name 'django' is not defined The problem is in the ​usage of ​exec: def handle(self, **options): # Execute the command and exit. if options['command']: exec(options['command']) return # Execute stdin if it has anything to read and exit. # Not supported on Windows due to () limitations. if sys.platform != 'win32' and not () and ([], [], [], 0)[0]: exec(()) return exec should be passed a dictionary containing a minimal set of globals. This can be done by just passing a new, empty dictionary as the second argument of exec."}
{"_id":"q-en-django__django-13665-2808ba566924fd1fbebddbc40483adb5386ff8181547c4d03ca18ee2119b86a4","text":"Dabase creation backend should use basemanager to serialize database Description class Service(Model): objects = CustomManagerThatFillterOutSomeRecords() class CustomManagerThatFillterOutSomeRecords(Manager): def getqueryset(self): return super().getqueryset().exclude(pk=1) class TestService(TransactionTestCase): serializedrollback = True def testsomething(self): pass Assume we have a migration that creates few records of Service. from import callcommand from import migrations def loadfixtures(, *_): callcommand('loaddata', '') class Migration(migrations.Migration): dependencies = [] operations = [ migrations.RunPython( loadfixtures, , ) ] Then TestService will fail as serializedbtostring by default use defaultmanager that is CustomManagerThatFillterOutSomeRecords. Here is proposed fix: ​"}
{"_id":"q-en-django__django-13667-13482743a1442bd1c1fd98af0994b59ad7ec10b4a5d4a759a4984c97e1ea5d6a","text":"Augment () optimizations to .union().exists(). Description (last modified by Simon Charette) The method performs optimization by ​clearing the select clause, ​dropping ordering, and limiting the number of results to 1 if possible. A similar optimization can be applied for combined queries when using () as some query planers (e.g. MySQL) are not smart enough to prune changes down into combined queries. For example, given (otherauthors).exists() the currently generated is SELECT 1 FROM (SELECT FROM authors WHERE ... ORDER BY ... UNION SELECT FROM authors WHERE ... ORDER BY ...) LIMIT 1; But some planers won't be smart enough to realize that both * and ORDER BY are not necessary and fetch all matching rows. In order to help them we should generate the following SQL SELECT 1 FROM (SELECT 1 FROM authors WHERE ... LIMIT 1 UNION SELECT 1 FROM authors WHERE ... LIMIT 1) LIMIT 1; This can already be done manually through filteredauthors.orderby().values(Value(1))[:1].union(otherauthors.orderby().values(Value(1))[:1]).exists() but that involves a lot of boilerplate. Note that the optimization is only possible in this form for union and not for intersection and difference since they require both sets to be unaltered."}
{"_id":"q-en-django__django-13670-11744c20bdb79cd728e617e44da93b2b65632e4089194055fbb3e9d1636cbe72","text":"dateformat.y() doesn't support years < 1000. Description (last modified by Sam) When using the the dateformat of django with a date before 999 (or 99 and 9 for similar matters) and the format character \"y\" no leading zero will be printed. This is not consistent with the way the python datetime module and PHP handle that character \"y\" in format strings: django (version 3.1): import datetime from import dateformat (datetime.datetime(123, 4, 5, 6, 7), \"y\") '3' python (version 3.8): import datetime datetime.datetime(123, 4, 5, 6, 7).strftime(\"%y\") '23' php (version 7.4): echo date(\"y\", strtotime(\"0123-04-05 06:07:00\")) 23 I have a pull-request ready for this: ​"}
{"_id":"q-en-django__django-13671-f23adfbc2230610dc8b7f94899c1454d2abac309fd04d8d944d0dbc36e5970b0","text":"Allow cache.getorset() to cache a None result Description (last modified by Phill Tornroth) getorset docstring says \"If the key does not exist, add the key and set it to the default value.\" -- but that's not quite what it does. It will perform a set if the key doesn't exist, or if the cached value is None. I think in order to be doing what it says on the tin it'd need to be: if self.has_key(key, version=version): return (key, version=version) else: if callable(default): default = default() (key, default, timeout=timeout, version=version) # Fetch the value again to avoid a race condition if another # caller a value between the first get() and the add() # above. return (key, default, version=version) I'd find this useful in cases where None was an expensive result to arrive at. If there's spiritual alignment with the suggestion, I'm happy to prepare and submit a change with tests."}
{"_id":"q-en-django__django-13682-27fec166649bd639b313607d4a84daba8cc2fea6c20b2324746829040108ff37","text":"Improve error messages from forgetting to call .asview() on a CBV Description (last modified by Angus Holder) We can detect early-on that the user has forgotten to call .asview() on their CBV when passing it into path(). For: urlpatterns = [ path('home', HomeView) ] The error currently happens only when you first load the route (rather than when constructing the routes), and looks like Internal Server Error: \/ Traceback (most recent call last): File \"C:\\Users\\Angus.virtualenvs\\django-WBTbdxDv\\lib\\site-packages\\django\\core\\handlers\\\", line 47, in inner response = getresponse(request) File \"C:\\Users\\Angus.virtualenvs\\django-WBTbdxDv\\lib\\site-packages\\django\\core\\handlers\\\", line 179, in getresponse response = wrappedcallback(request, callbackargs, *callbackkwargs) TypeError: init() takes 1 positional argument but 2 were given Which is especially hard to work out given that the traceback doesn't even include any of the user's own code, and this is an error that's easy for beginners to run into when first using CBVs. My PR changes it to fail early, inside the call to (), with a clear error: URL route 'foo' should pass in 'EmptyCBView.as_view()' instead of 'EmptyCBView' Pull request: ​"}
{"_id":"q-en-django__django-13684-2acd3916f20e09a8a4b73c9923741ada6f7204fccfc09909c75331305e2e2774","text":"Misleading error when loading a database backend fails with ImportError Description For example, in the following case, djangocockroachdb is installed but has an import error itself. Traceback (most recent call last): File \"\", line 110, in loadbackend return importmodule('' % backendname) File \"\", line 127, in importmodule return bootstrap.gcdimport(name[level:], package, level) File \"<frozen importlib.bootstrap\", line 1030, in gcdimport File \"<frozen importlib.bootstrap\", line 1007, in findandload File \"<frozen importlib.bootstrap\", line 972, in findandloadunlocked File \"<frozen importlib.bootstrap\", line 228, in callwithframesremoved File \"<frozen importlib.bootstrap\", line 1030, in gcdimport File \"<frozen importlib.bootstrap\", line 1007, in findandload File \"<frozen importlib.bootstrap\", line 986, in findandloadunlocked File \"<frozen importlib.bootstrap\", line 680, in loadunlocked File \"<frozen importlib.bootstrapexternal\", line 790, in execmodule File \"<frozen importlib.bootstrap\", line 228, in callwithframesremoved File \"\/home\/tim\/code\/cockroach-\", line 2, in <modulefrom .lookups import patchlookups File \"\/home\/tim\/code\/cockroach-\", line 1, in <modulefrom import HasKeyLookup, KeyTransform ModuleNotFoundError: No module named '' The above exception was the direct cause of the following exception: Traceback (most recent call last): File \"\", line 10, in <moduleexecutefromcommandline() File \"\", line 381, in executefromcommandline utility.execute() File \"\", line 357, in execute () File \"\", line 24, in setup apps.populate(settings.INSTALLEDAPPS) File \"\", line 114, in populate appconfig.importmodels() File \"\", line 211, in importmodels self.modelsmodule = importmodule(modelsmodulename) File \"\", line 127, in importmodule return bootstrap.gcdimport(name[level:], package, level) File \"<frozen importlib.bootstrap\", line 1030, in gcdimport File \"<frozen importlib.bootstrap\", line 1007, in findandload File \"<frozen importlib.bootstrap\", line 986, in findandloadunlocked File \"<frozen importlib.bootstrap\", line 680, in loadunlocked File \"<frozen importlib.bootstrapexternal\", line 790, in execmodule File \"<frozen importlib.bootstrap\", line 228, in callwithframesremoved File \"\", line 12, in <moduleclass Question(models.Model): File \"\", line 117, in new newclass.addtoclass('meta', Options(meta, applabel)) File \"\", line 321, in addtoclass value.contributetoclass(cls, name) File \"\", line 204, in contributetoclass self.dbtable = truncatename(self.dbtable, ()) File \"\", line 28, in getattr return getattr(connections[DEFAULTDBALIAS], item) File \"\", line 201, in getitem backend = loadbackend(db['ENGINE']) File \"\", line 121, in loadbackend raise ImproperlyConfigured( 'djangocockroachdb' isn't an available database backend. Try using '', where XXX is one of: 'mysql', 'oracle', 'postgresql', 'sqlite3' I'm not sure there's anything that could be done besides rephrasing the error message to say something like this: 'djangocockroachdb' isn't an available database backend or couldn't be imported. Check the above exception. In this case, \"Try using...\" isn't so helpful either. Perhaps we could only include that if backend_name begins with django.."}
{"_id":"q-en-django__django-13689-5215cc1adede418ab6309ac27c57e6c00f6d3528cf7c7139aed7b43d3d04c0cc","text":"Aggregating when grouping on an ExpressionWrapper omits the expression from the group by Description I ran into this with Postgres on Django 3.1.3, I'm not sure what other versions it exists on. print( Fred.objects.annotate( bobidisnull=ExpressionWrapper( Q(bobid=None), outputfield=BooleanField() ) ).values( \"bobidisnull\" ).annotate( idcount=Count(\"id\", distinct=True) ).values( \"bobidisnull\", \"idcount\" ).query ) SELECT \"mainfred\".\"bobid\" IS NULL AS \"bobidisnull\", COUNT(DISTINCT \"mainfred\".\"id\") AS \"idcount\" FROM \"mainfred\" GROUP BY \"mainfred\".\"bobid\" On the last line there the group by has dropped the \"IS NULL\""}
{"_id":"q-en-django__django-13691-6d2991a8b12851ae8ca0bcd8eb10faca5ef0b46f5f8a71682b980069344b54f3","text":"Error with values with JSON path lookup in Sqlite when value is numeric. Description (last modified by Gordon Wrigley) This Django 3.1.3, I only see this with Sqlite, it works fine with MySQL and Postgres. When I do a path lookup with values like (\"myjsonfieldposition\") if there is an integer, float or bool in \"position\" then I get a JSON decode error. Strings, nones, dicts and lists all work, fetching the top level dict works and filtering on the path lookup works. TypeError: the JSON object must be str, bytes or bytearray, not float .....venv\\databrowser\\lib\\site-packages\\django\\db\\models\\ in iter self.fetchall() .....venv\\databrowser\\lib\\site-packages\\django\\db\\models\\ in fetchall self.resultcache = list(self.iterableclass(self)) .....venv\\databrowser\\lib\\site-packages\\django\\db\\models\\ in iter for row in compiler.resultsiter(chunkedfetch=self.chunkedfetch, chunksize=self.chunksize): .....venv\\databrowser\\lib\\site-packages\\django\\db\\models\\sql\\ in applyconverters value = converter(value, expression, connection) .....venv\\databrowser\\lib\\site-packages\\django\\db\\models\\fields\\ in fromdb_value return (value, cls=self.decoder)"}
{"_id":"q-en-django__django-13693-445b8ae8341ec583873bfda214dd34dc3f923ad804ac185bc1fccd7a9ed59223","text":"django-admin runserver and getchildarguments() crashes on Windows and Python < 3.8. Description \"django-admin runserver\" fails with the following error: Traceback (most recent call last): File \"c:\\users\\someone\\appdata\\local\\programs\\python\\python37\\lib\\\", line 193, in runmoduleasmain \"main\", modspec) File \"c:\\users\\someone\\appdata\\local\\programs\\python\\python37\\lib\\\", line 85, in runcode exec(code, runglobals) File \"C:\\Users\\someone\\AppData\\Local\\Programs\\Python\\Python37\\Scripts\\django-\\\", line 7, in <moduleFile \"c:\\users\\someone\\appdata\\local\\programs\\python\\python37\\lib\\site-packages\\django\\core\\management\\\", line 401, in executefromcommandline utility.execute() File \"c:\\users\\someone\\appdata\\local\\programs\\python\\python37\\lib\\site-packages\\django\\core\\management\\\", line 395, in execute self.fetchcommand(subcommand).runfromargv() File \"c:\\users\\someone\\appdata\\local\\programs\\python\\python37\\lib\\site-packages\\django\\core\\management\\\", line 330, in runfromargv self.execute(args, cmdoptions) File \"c:\\users\\someone\\appdata\\local\\programs\\python\\python37\\lib\\site-packages\\django\\core\\management\\commands\\\", line 61, in execute super().execute(args, options) File \"c:\\users\\someone\\appdata\\local\\programs\\python\\python37\\lib\\site-packages\\django\\core\\management\\\", line 371, in execute output = (args, options) File \"c:\\users\\someone\\appdata\\local\\programs\\python\\python37\\lib\\site-packages\\django\\core\\management\\commands\\\", line 96, in handle (options) File \"c:\\users\\someone\\appdata\\local\\programs\\python\\python37\\lib\\site-packages\\django\\core\\management\\commands\\\", line 103, in run autoreload.runwithreloader(self.innerrun, options) File \"c:\\users\\someone\\appdata\\local\\programs\\python\\python37\\lib\\site-packages\\django\\utils\\\", line 616, in runwithreloader exitcode = restartwithreloader() File \"c:\\users\\someone\\appdata\\local\\programs\\python\\python37\\lib\\site-packages\\django\\utils\\\", line 244, in restartwithreloader p = (args, env=newenviron, closefds=False) File \"c:\\users\\someone\\appdata\\local\\programs\\python\\python37\\lib\\\", line 488, in run with Popen(popenargs, kwargs) as process: File \"c:\\users\\someone\\appdata\\local\\programs\\python\\python37\\lib\\\", line 800, in init restoresignals, startnewsession) File \"c:\\users\\someone\\appdata\\local\\programs\\python\\python37\\lib\\\", line 1148, in executechild args = list2cmdline(args) File \"c:\\users\\someone\\appdata\\local\\programs\\python\\python37\\lib\\\", line 555, in list2cmdline needquote = (\" \" in arg) or (\"\\t\" in arg) or not arg TypeError: argument of type 'WindowsPath' is not iterable Environment: Windows 10 Python: 3.7.6 Django: 3.1.3 Steps to reproduce: django-admin startproject mysite set PYTHONPATH=<thelocaldirdjango-admin runserver --settings=mysite.settings Apparently returns WindowsPath(\"C:\\Users\\someone\\AppData\\Local\\Programs\\Python\\Python37\\Scripts\\django-\") as the first argument, and subprocess.Popen expects a string and is not able to use Path. The following monkey-patch fixes the error: def getchildargumentsoverride(): args = autoreload.getchildarguments() for i, arg in enumerate(args): if isinstance(arg, Path): args[i] = str(arg) return args autoreload.getchildarguments = autoreload.getchildarguments autoreload.getchildarguments = getchildargumentsoverride"}
{"_id":"q-en-django__django-13708-65bd5688b5bbb94cb079d90c908e374050498eb296cc651092051f3356d774e3","text":"Add a 'durable' flag to () Description As discussed ​on Twitter following ​David Seddon's blog post. () guarantees the wrapped database operations is atomic - at the end of its wrapped block, operations within that block will all be applied, or all rolled back. In some situations it's also useful to guarantee that the wrapped operations are durable - at the end of the wrapped block, all operations have definitely been committed. atomic() doesn't guarantee this at the moment since its use may be wrapped by another atomic() higher in the stack, delaying the commit. David Seddon's technique to guarantee durability is to wrap atomic() to first check if an atomic() is already active for the current connection (in other words, connection.get_autocommit() returns False), and raise an error if so. On Twitter, Ian Foote suggested adding a durable flag to atomic() to add this behaviour to Django, to which Aymeric Augustin said it sounds like a good idea."}
{"_id":"q-en-django__django-13710-78163a3196f37a5e49756fb2c6682437a134164e6b52f67ca5b9b2a560bec342","text":"Use Admin Inline verbosename as default for Inline verbosenameplural Description Django allows specification of a verbosename and a verbosenameplural for Inline classes in admin views. However, verbosenameplural for an Inline is not currently based on a specified verbosename. Instead, it continues to be based on the model name, or an a verbosename specified in the model's Meta class. This was confusing to me initially (I didn't understand why I had to specify both name forms for an Inline if I wanted to overrule the default name), and seems inconsistent with the approach for a model's Meta class (which does automatically base the plural form on a specified verbosename). I propose that verbosenameplural for an Inline class should by default be based on the verbosename for an Inline if that is specified. I have written a patch to implement this, including tests. Would be happy to submit that."}
{"_id":"q-en-django__django-13714-bcf48f1988a7c7a294992679621f5456dee9127e1676a7f54c439362d936e7c1","text":"SQLite3 bad filepath raising as JSON extension error. Description Applying migrations on an installation using spatialite backend (probably sqlite too) where the path to the database file does not exist causes a confusing JSON field error. Take this following DATABASES configuration. Note that the dev.sqlite3 file does not exist. DATABASES = { 'default': { 'ENGINE': '', 'NAME': '\/home\/user\/app\/dev.sqlite3', } } When running migrate the following exception is raised SystemCheckError: System check identified some issues: ERRORS: rumble_common.RumbleEvent: (fields.E180) SQLite does not support JSONFields. We assert the JSON extension is present with the following code block, as per import sqlite3 conn = sqlite3.connect(':memory:') cursor = () cursor.execute('SELECT JSON(\\'{\"a\": \"b\"}\\')') The issue is resolved by changing DATABASES['default']['NAME'] to point to the right path. The issue is that the error is misleading, and leads developers down the wrong rabbit hole Hope thats descriptive enough, thanks all! Extra: Django 3.1.3 Debian Buster Docker environment"}
{"_id":"q-en-django__django-13722-14a966e8c7e54286dd23ba1b714433c823c4f18640b528058d6bac1a2bde1ac7","text":"Add a hook to customize the admin's formsets parameters Description New feature that adds a method on InlineModelAdmin for providing initial data for the inline formset. By default there is no implementation, although one could be implemented to use GET parameters like getchangeforminitial_data, but it wouldn't be trivial due to the list nature of formset initial data."}
{"_id":"q-en-django__django-13741-7c6a3534a5f7052c46abc504cb30fc8a83e675c023d74b244190d6979675fa30","text":"Set disabled prop on ReadOnlyPasswordHashField Description Currently the defines a cleanpassword method that returns the initial password value to prevent (accidental) changes to the password value. It is also documented that custom forms for the User model need to define this method: ​ A while ago the forms.Field base class gained the ​disabled argument to: [disable] a form field using the disabled HTML attribute so that it won’t be editable by users. Even if a user tampers with the field’s value submitted to the server, it will be ignored in favor of the value from the form’s initial data. It seems to me that this property could be set to True be default on the ReadOnlyPasswordHashField used to display the password hash. This way the cleanpassword is no longer necessary and the potential pitfall when using the ReadOnlyPasswordHashField without implementing clean_password is removed."}
{"_id":"q-en-django__django-13743-134d906fd97d30df8741174100a8702b6b5d6293e0fb804344a414df3fb25308","text":"ConnectionAbortedError should be treated as a broken pipe error in the development server Description People using Windows have been complaining about very long stack traces in the development server for a long time: ​ ​ These happen under normal interaction with the development server using Chrome. I have fixed one issue in CPython in wsgiref, but I belive that we should also fix isbrokenpipe_error in"}
{"_id":"q-en-django__django-13744-b83dacc3a51387c2a94ea4843badef552972b4e52ae7d73ed4e3a62a5fa3b215","text":"Deprecate MemcachedCache. Description python-memcached is not maintained anymore (see ​python-memcached) and it makes difficulties in fixing some issues (e.g. ). Moreover we a cache backend for pymemcache () so we have a good builtin alternative. I think it's time to deprecate the backend in Django 3.2 and remove it in Django 4.1."}
{"_id":"q-en-django__django-13757-94f23950c62f378103d285aaa157187b8bd81fbee8d905bf859d9bb5589bcf4a","text":"Using isnull=True on a KeyTransform should not match JSON null on SQLite and Oracle Description The KeyTransformIsNull lookup borrows the logic from HasKey for isnull=False, which is correct. If isnull=True, the query should only match objects that do not have the key. The query is correct for MariaDB, MySQL, and PostgreSQL. However, on SQLite and Oracle, the query also matches objects that have the key with the value null, which is incorrect. To confirm, edit tests.modelfields.testjsonfield.TestQuerying.testisnullkey. For the first assertion, change self.assertSequenceEqual( (valueaisnull=True), [:3] + [5:], ) to self.assertSequenceEqual( (valuejisnull=True), [:4] + [5:], ) The test previously only checks with valuea which could not catch this behavior because the value is not JSON null."}
{"_id":"q-en-django__django-13768-6ee9b2b0d3d91689089c1b1f9149f54fdd38b7ff8d69e9e2c3a25e63a2f0bd70","text":"Log exceptions handled in Signal.sendrobust() Description As pointed out by ​Haki Benita on Twitter, by default Signal.sendrobust() doesn't have any log messages for exceptions raised in receivers. Since Django logs exceptions in other similar situations, such as missing template variables, I think it would be worth adding a logger.exception() call in the except clause of send_robust() . Users would then see such exceptions in their error handling tools, e.g. Sentry, and be able to figure out what action to take from there. Ultimately any expected exception should be caught with a try in the receiver function."}
{"_id":"q-en-django__django-13773-364d1759d3d9253a92e0d42c3f70f986f89f90b6e23b9b09cef9df37ec08cd58","text":"squashmigrations optimizer crashes when fields' names are swapped using a temporary name Description If you rename fields using a pattern like a-c; b-a; c-b (such as if previously DateTimeFields using autonow and autonowadd had been mixed-up) and then attempt to squashmigrations with an optimization barrier between the CreateModel and RenameFields, the migration optimizer will attempt to create a CreateModel operation object with two fields using the same name and fail. I'll attach a migration file that triggers the failure. I believe the root cause of this issue is that allows itself to optimize through (i.e be moved to the right of, I may have gotten this terminology wrong) other RenameField operations that reference oldname."}
{"_id":"q-en-django__django-13774-e5a90ee389e14f82525ec8af6cee250eb60051bc3b219ab34262c55375fed103","text":"prefetchrelatedobjects() does not work for reused model instances. Description (last modified by Dennis Kliban) Our project processes instances in a stream. In some cases the instances are repeated. In these cases, we discovered that prefetchrelatedobjects() does not set the toattr on all of the instances if the first instance in the list already has it set. When Django determines that the very first instance in the list isfetched[0], it does not call into the the prefetchonelevel()[1]. The attributed specified in the toattr parameter is only set in the prefetchonelevel() method[2]. isfetched is set by looking for the to_attr attribute on the instance[3]. [0] ​ [1] ​ [2] ​ [3] ​"}
{"_id":"q-en-django__django-13786-95a7a0ab4a4d7375f967310d95d8e2e78a83026ed825fa4b73e31a6a35875329","text":"squashmigrations does not unset model options when optimizing CreateModel and AlterModelOptions Description When an operation resembling AlterModelOptions(name=\"testmodel\", options={}) is squashed into the corresponding CreateModel operation, model options are not cleared on the resulting new CreateModel operation object. () sets the new options as options={self.options, operation.options} in this case ( line 144 on commit ), with no logic to remove options not found in operation.options as is found in AlterModelOptions.stateforwards(). I believe this issue still exists on the master branch based on my reading of the code, but I've only tested against 2.2."}
{"_id":"q-en-django__django-13791-f192bc282767abb1d7ab307b084cdace6acacb9e1632600400f06042f4767a85","text":"parseduration() ISO string sign is ignored when the timedelta only has days Description I'm pretty sure that this is a bug even though I'm not an expert on the ISO 8601 standard. The sign of a timedelta string will be ignored by if the input string only contains days. Compare the following (notice the minus signs): In [4]: timedelta(days=-1) Out[4]: datetime.timedelta(days=-1) In [5]: td = timedelta(days=-1) In [6]: durationisostring(td) Out[6]: '-P1DT00H00M00S' In [7]: parseduration(durationisostring(td)) Out[7]: datetime.timedelta(days=1) # <-- Why is this 1 and not -1? In [8]: td = timedelta(days=-1, microseconds=1) In [9]: durationisostring(td) Out[9]: '-P0DT23H59M59.999999S' In [10]: parseduration(durationisostring(td)) Out[10]: datetime.timedelta(days=-1, microseconds=1) I guess the problem is in line 147 that reads return days + sign * datetime.timedelta(kw). However, if datetime.timedelta(kw) ends up being zero (timedelta(0)) then the sign multiplication ends up in zero, not -0. This is just a preliminary quick look though and maybe the problem is something else."}
{"_id":"q-en-django__django-13794-9c13c0002a81b73ff3508efd77ba82f1dfac848c465c4dd2c2c158be8ac95cdd","text":"add filter is unable to concatenate strings with lazy string Description If you try to concatenate a string with a lazy string with the add template filter, the result is always the empty string because the add filter generates an exception (TypeError: can only concatenate str (not \"proxy\") to str)."}
{"_id":"q-en-django__django-13797-5bb362ed564e3dabd4d830471c9865948d43f04fde86ff6cfb2744d0985fcfc5","text":"Add support for fixtures compression in dumpdata Description Since loaddata command support compressed fixtures [1] loading, I would propose to add fixtures compression support in dumpdata command. [1] ​"}
{"_id":"q-en-django__django-13800-2ba2c7d50571af03161eb60d6c9724d90355cf48a8bf24a98c1d9b785065380d","text":"Not RFC 6265 compliant cookies in contrib.messages. Description (last modified by Nico Giefing) Hi A Customer of mine is using a WAF which is handling Cookies as it is described tin the RFC: ​ The issue now is that Django is trying to use an escape-character in cookie-Values which is not supported in the RFC an example of such a cookie: messages=\\\"123\\\\\"NOTRECEIVED\\\"\" Please consider to get this fixed so there can be a protection of this system. Regards, Nico"}
{"_id":"q-en-django__django-13807-1467c4ae8438354cfda25d4266709535346f2a4b21032e32e07c66ae72b5c265","text":"loaddata crashes on SQLite when table names are SQL keywords. Description Steps to reproduce: Create a Model called Order. (order is a SQL reserved word) Create fixtures for the model Use loaddata to load the fixture. Notice that it fails with the following error. This is because the table name order is not quoted properly (0.000) PRAGMA foreignkeycheck(order); args=None Traceback (most recent call last): File \"python3.7\/site-\", line 82, in execute return (sql) File \"python3.7\/site-\", line 411, in execute return Database.Cursor.execute(self, query) sqlite3.OperationalError: near \"order\": syntax error Root Cause File: python3.7\/site- line 327 Function: checkconstraints Details: due to missing back ticks around %s in the SQL statement PRAGMA foreignkeycheck(%s) Here in checkconstraints line 327 in context if tablenames is None: violations = cursor.execute('PRAGMA foreignkeycheck').fetchall() else: violations = chain.fromiterable( cursor.execute('PRAGMA foreignkeycheck(%s)' % tablename).fetchall() for tablename in tablenames ) And here line 333 for tablename, rowid, referencedtablename, foreignkeyindex in violations: foreignkey = cursor.execute( 'PRAGMA foreignkeylist(%s)' % tablename ).fetchall()[foreignkey_index] Issue confirmed in 0 2"}
{"_id":"q-en-django__django-13808-c5774403267002a3735dbb77695c567eddb76f3e8dc4fe3b6957851fc093ba6d","text":"Allow postgresql database connections to use postgres services Description (last modified by levihb) Postgres offers a way to make database connections through the use of services, which are basically equivalent to MySQL's options files. Server, database, username, etc information is stored by default in and takes a very similar format to MySQL cnf files: [myalias] host=10.0.19.10 user=postgres dbname=postgres port=5432 And password can be stored in under a different format. I think being able to just add them to the DATABASES config would be useful, similar to how you can add MySQL cnf files. psycopg2 supports it just fine through the service argument\/string connect(service='myalias') connect('service=myalias'). At the moment it can be like this: DATABASES = { 'default': { 'ENGINE': '', 'NAME': 'postgres', 'OPTIONS': {'service': 'myalias'} } } Which works, however it involves repeating the database name. I don't think the database name should be repeated twice because it couples the config and the service file together, and makes it harder to just move it between different environments. I think ideally you would just specify the service, either like this: DATABASES = { 'default': { 'ENGINE': '', 'OPTIONS': {'service': 'myalias'} } } Or maybe a better way would be?: DATABASES = { 'default': { 'ENGINE': '', 'SERVICE': 'my_alias } } It seems like something that would be super easy to add. I don't mind creating a pull request for it, but would like to know why it hasn't been , and how it would be recommended to add it."}
{"_id":"q-en-django__django-13809-963a318b817fcd8b6f41a29f3cdc07a2b6151732b02715f2917e27989cd14353","text":"Add --skip-checks option to the runserver command. Description Rationale: It would be consistent with other management commands performing system checks It would help people like me who would rather have checks enabled exclusively in CI\/CD than wait 15-20 seconds for each project reload during development Related StackOverflow question: ​"}
{"_id":"q-en-django__django-13810-333dda11167c666f9bfaf905ea1760634e508612732fb9cc97deaca3b7714ee5","text":"MiddlewareNotUsed leaves undesired side effects when loading middleware in ASGI context Description I experienced strange issues when working with ​ASGI , ​django-debug-toolbar and my own small middleware. It was hard problem to debug, I uploaded an example project here: ​ (the name is misleading - I initially thought it's a bug with django-debug-toolbar). The SESSIONFILEPATH setting is intentionally broken to cause a 500 error. When starting the application and accessing \/admin (any location really, but I wanted to leave it at a minimum and didn't add any views) it gives TypeError: object HttpResponse can't be used in 'await' expression. Commenting out asgidjangotoolbarbug.middleware.DummyMiddleware fixes the issue (in that I receive a 500 ImproperlyConfigured exception). I'm not sure about the overall role of django-debug-toolbar here - removing it causes Daphne to return a 500 error page but without debug information and there's no traceback in console either. I decided to leave it since it helped me approximate the causes of issue. I notice that in ​ while MiddlewareNotUsed causes the loop to skip futher processing and go to next middleware, it does leave handler variable overwritten with output of self.adaptmethodmode(). On next pass, this handler is passed to next middleware instance, disregarding all the previous checks for (lack of) async support. This likely causes the middleware chain to be \"poisoned\" from this point onwards, resulting in last middleware in response cycle to return an HttpResponse as a synchronous middleware would, instead of coroutine that is expected. This is probably avoided by adding async support to my middleware, but unless I'm missing something ​docs indicate it should work as it is. It is my intention that it's applied only on synchronous requests, so I didn't make it async compatible on purpose. If it's intentional in Django that every middleware needs to support async if the application is run as ASGI app, the documentation should probably state that clearly. Though it kinda defeats the purpose of having async_capable = False flag in the first place."}
{"_id":"q-en-django__django-13814-370866a5cfc198408a88e2900c782fe5f30cb676450946b901eae4207d504311","text":"Raising InvalidTemplateLibrary completely masks out real exception in getpackagelibraries Description Summary In , function getpackagelibraries on line 119 completely masks out the ImportError and raises InvalidTemplateLibrary. This makes it incredibly difficult to debug application issues. Probably better not to handle the exception in the first place since it only raises another type and inner exception looses the stack trace. To reproduce Create two apps e.g. formutils and reports. Write a template tag in reports e.g. reports.templatetags.reporttags. ( Add a simple module in formutils e.g. In , import a none-existent module e.g. from import flatatt (was removed in django 1.4) import in reporttags e.g. from formutils.widgets import CalendarWidget A quick way to reproduce the error would be to register some models with admin and navigate to \/admin The following error will be raised in getpackagelibraries: InvalidTemplateLibrary at \/admin\/login\/ Invalid template library specified. ImportError raised when trying to load 'reports.templatetags.reporttags': No module named util Request Method: GET Request URL: ​http:\/\/localhost:2017\/admin\/login\/?next=\/admin\/ Django Version: 1.10.1 Exception Type: InvalidTemplateLibrary Exception Value: Invalid template library specified. ImportError raised when trying to load 'reports.templatetags.reporttags': No module named util Exception Location: D:\\repo\\django110\\lib\\site-packages\\django\\template\\backends\\ in getpackagelibraries, line 130 However, if the exception was not caught and \"wrongly\" re-raised as an InvalidTemplateLibrary, the following errors would be printed: ImportError at \/admin\/login\/ No module named util Request Method: GET Request URL: ​http:\/\/localhost:2017\/admin\/login\/?next=\/admin\/ Django Version: 1.10.1 Exception Type: ImportError Exception Value: No module named util Exception Location: D:\\repo\\projects\\evincehr\\apps\\formutils\\ in <module, line 3 The second behavior is more appropriate to debugging the error and the error would be quickly found."}
{"_id":"q-en-django__django-13820-a512eb81f8c8898a67db085abb544420442150a6e1281c4389e222a2afe3e032","text":"Permit migrations in non-namespace packages that don't have file Description Summary This feature request, for which I will post a PR shortly, aims to improve the specificity of the migration loader's check for and rejection of ​PEP-420 namespace packages. I am NOT asking to allow namespace packages for apps' migrations. I merely want to make the existing check more compliant with Python's documented import API. This would remove one impediment to using Django in so-called frozen Python environments (such as those mentioned in ) that do not set ​file on regular packages by default. This narrow proposal does not change Django's behavior at all for normal Python environments. The only change for frozen environments is that Django will learn how to find existing migrations. In particular, at this time I am not proposing to enable any other Django feature that does not already work in frozen environments. I would love for this feature to land in Django 3.2. Details I initially broached this idea on the ​django-developers mailing list. This is my second ticket related to frozen Python environments, the first being . The ​current implementation of the migration loader's no-namespace-package check in skips searching for migrations in a module m if getattr(m, 'file', None) is false. The trouble with this implementation is that namespace packages are not the only modules with no file. Indeed, the Python ​documentation states that file is optional. If set, this attribute's value must be a string. The import system may opt to leave file unset if it has no semantic meaning (e.g. a module loaded from a database). However, Python's ​documentation also states Namespace packages do not use an ordinary list for their path attribute. They instead use a custom iterable type.... The class of namespace packages' path in CPython is ​NamespacePath, but that is a CPython implementation detail. Instead, I propose to augment getattr(m, 'file', None) with and isinstance(m.path, list)."}
{"_id":"q-en-django__django-13821-8a5efb60356cd972346e5837bbbc1a2d17e758cc5a4d715283f221128ba7b664","text":"Drop support for SQLite < 3.9.0 Description (last modified by Tim Graham) Indexes on expressions (see ) and the SQLITEENABLEJSON1 compile-time option are supported on ​SQLite 3.9.0+. Ubuntu Xenial ships with SQLite 3.11.0 (which will still by supported by Django) and will EOL in April 2021. Debian Jessie ships with 3.8.7 and was EOL June 30, 2020. SQLite 3.9.0 was released in October 2015. SQLite version support seems like a similar situation as GEOS libraries which we generally support about 5 years after released."}
{"_id":"q-en-django__django-13822-c7a614f7abc8dd2b22847a1b33cfcab370c16df35a96d0092e9b2cd5433e012c","text":"fields.E305 is raised on ManyToManyFields with relatedname='+' in models in different apps but with the same name. Description (last modified by Aleksey Ruban) Django raises an error during creation a db migration if two models with the same name refer to the same model in m2m field. relatedname='+' or 'foo+' don't impact anything. In some my project there are 50 apps and almost each one has a model with the same name. So I have to come up with a related name and write it in for each m2m field. Just try to make a migration for my test project ​"}
{"_id":"q-en-django__django-13824-bb0ab95fc64ae9691b54a0ca141040d2c14d642da7b021c78caa14bb97e34391","text":"Django adds spurious \"\/\" prefix to settings.STATICURL=\"http:\/\/server\/\" Description (last modified by Adam Hooper) Here's a piece of settings from a totally reasonable, sensible, okay Docker integration-test environment STATICURL = \"http:\/\/minio\/static\/\" Django 3.1 will implicitly add \"\/\" to the URL, so my URLs look like \/http:\/\/minio\/static\/images\/app- The features and bugs that interact here: commit , adding feature , prepends SCRIPTNAME to STATICURL when STATICURL isn't a URL. bug and : according to Django, \"​http:\/\/minio\/static\/\" isn't a valid URL. (It is.) Top me, the easiest fix is to address \/ . Or to make STATICURL use some logic that is different from URLValidator."}
{"_id":"q-en-django__django-13837-355786c7ae8b315816f024d46f4ea3d1964e57730edf0574c4a06becbacc5e0a","text":"Allow autoreloading of Description (last modified by William Schwartz) ​ detects if Python was launched as python -m django. Currently it detects only when ​-m was passed specifically django (and only in Python environments in which file is set on modules, which is ​not true of all Python environments). Like , this ticket aims to remove one impediment to creating Django-based command-line utilities that have their own ​main sub-module while overriding Django's built-in management commands—in this case, runserver. The fix, which I have submitted in the ​attached PR, is to use Python's ​documented way of determining if -m was used in getchild_arguments: The top-level main module is always the entry point of a ​complete Python program. is not None ​if and only if Python was launched with -m or the name of a \"directory, zipfile or other entry.\" In the latter cases, the ​documentation says If the script name refers to a directory or zipfile, the script name is to the start of and the file in that location is executed as the main module. Hence (which is ​usually but not always main.package) exists and is the empty string when Python is started with the name of a directory or zip file. Therefore Python was started with -m pkg if and only if == \"pkg\". Following this algorithm is guaranteed to work as long as Python obeys its own documentation, and has the side benefit of avoiding use of file."}
{"_id":"q-en-django__django-13841-f94f07bcc5715956d6a54dac8855be607760b56ffa2c850cf74c0e286ad523a0","text":"Access file lazily rather than at module level Description (last modified by William Schwartz) So-called frozen Python environments (such as those mentioned in ) that do not set all modules' ​file variable, which ​need not be defined, cannot even import Django (without some workarounds) because a small number of Django modules use file at the module level, in a class defined at the module level, or in a function that is called automatically upon import. Five modules that use file like this are likely to be imported when using Django and thereby cause a frozen Python to crash with a NameError or similar exception. Importing can be avoided only by avoiding both forms and the ORM altogether as it's imported from Importing might be avoidable if DEBUG=False or by avoiding all of the views and URLs APIs. 's getgitchangeset is called when django is imported in pre-alpha development versions. Importing is only avoidable by not using the Auth app. uses file to find Django's localization files upon activation; this avoidable only by setting USEI18N=False. Dealing with transreal is sufficiently thorny (and, being an English speaker with English-speaking clients, I can avoid it for now) that I will not address it further here except to say that it might need to be part of the larger discussion at . What this ticket is not I am not proposing removing use of file at this time. That would require a longer discussion of intended semantics such as . This ticket is only about removing use of file at the module (or class definition) level in Django application code (not test code). Further I am not proposing banning use of file at the module level at this time, hence minimal new tests and no update to the Django coding style documentation. That too would require a longer conversation. Proposed fixes I have pushed ​PR GH- to address the four of those modules other than transreal. I dealt with each module's use of file in separate commits to make them easier to discuss and separate\/cherry-pick if needed. Below I link to the individual commits as I discuss each of the four modules. These first two are fairly easy, but the second two may require further consideration. (​) Remove the undocumented module constant ROOT and replace its single use. (​) Treat the lack of module-global file the same as a failure of git log by returning None from getgitchangeset. (​) The module-level constant CURRENTDIR is used only in the module itself and is undocumented, so I'm assuming it's an obscure private symbol that no one will miss. I've replaced it with a module-level private function builtintemplatepath that refactors and centralizes finding built-in templates for the entire module. The one tricky part is that the htmltemplatepath and texttemplatepath attributes I didn't want to disturb 's goal of making the template paths easily override-able, so I avoided calling builtintemplatepath in the class definition by making detecting the presence of the attributes in init and setting defaults there. Alternatives include making the attributes properties with setters or cached properties without setters. (​) The CommonPasswordValidator-class constant DEFAULTPASSWORDLISTPATH is used only in one place, the class's instance constructor. While the nature of DEFAULTPASSWORDLISTPATH is not documented, its existence is inside the docs for the ​constructor's signature. I've changed DEFAULTPASSWORDLISTPATH from a class constant into an instance attribute. Another possibility is making DEFAULTPASSWORDLIST_PATH be a"}
{"_id":"q-en-django__django-13884-f0579f4d3094c09a10d22c3429c812a14a9c83182b70d6d6c511a693fd292c9b","text":"i18n.setlanguage unquotes nexturl and produces wrong url for url params containing \"&\" character Description (last modified by Johannes Maron) When changing the language and the current URL parameter include a parameter value with an encoded \"&\" like ?paramter=some%20%26%20thing the redirect response from setlangauge is ?paramter=some%20&%20thing where I would still expect the same URL from as in the beginning. I've written a Django test that shows this bug: def testsetlanguageurlparams(): from import RequestFactory from import setlanguage rf = RequestFactory() request = (\"\", next=\"\") request.META['HTTPREFERER'] = '\/someurl\/?paramter=some%20%26%20thing' response = setlanguage(request) assert == '\/someurl\/?paramter=some%20%26%20thing' i18n.setlanguage unquotes nexturl and produces wrong url for url params containing \"&\" character Description (last modified by Johannes Maron) When changing the language and the current URL parameter include a parameter value with an encoded \"&\" like ?paramter=some%20%26%20thing the redirect response from setlangauge is ?paramter=some%20&%20thing where I would still expect the same URL from as in the beginning. I've written a Django test that shows this bug: def testsetlanguageurlparams(): from import RequestFactory from import setlanguage rf = RequestFactory() request = (\"\", next=\"\") request.META['HTTPREFERER'] = '\/someurl\/?paramter=some%20%26%20thing' response = setlanguage(request) assert == '\/someurl\/?paramter=some%20%26%20thing'"}
{"_id":"q-en-django__django-13886-a9a293c148dbed8e59bc9116a1d51481ab667fb3eb1e1944645e10718d5dfe17","text":"Add system checks for invalid model field names for functional indexes in Meta.indexes. Description (last modified by Mariusz Felisiak) The current system checks don't cover invalid models fields in functional indexes. This is not straightforward because resolving expressions with non-existent throws FieldError, so we cannot just collect columns from Query.gencols(). Follow up to . I attached tests."}
{"_id":"q-en-django__django-13924-5a263959ad9405016c7d42a124a3b589249c1e19fc639d0bc26f833e396a87f5","text":"Migrations are marked applied even if deferred SQL fails to execute Description The changes introduced in to address fail to account for the possibility of the schema editor accumulation of deferred SQL which is run at time."}
{"_id":"q-en-django__django-13925-f9d6a912e9e544539411d69ea1e7ea447d6c96372dc305ef6ee03804c795366c","text":"models.W042 is raised on inherited manually specified primary key. Description I have models which inherit from other models, and they should inherit the primary key. This works fine with Django 3.1. However, if I install Django 3.2 alpha, when I run makemigrations I get the following error messages: System check identified some issues: WARNINGS: accounts.ReservedUsername: (models.W042) Auto-created primary key used when not defining a primary key type, by default ''. HINT: Configure the DEFAULTAUTOFIELD setting or the SpeedyCoreAccountsConfig.defaultautofield attribute to point to a subclass of AutoField, e.g. ''. accounts.User: (models.W042) Auto-created primary key used when not defining a primary key type, by default ''. HINT: Configure the DEFAULTAUTOFIELD setting or the SpeedyCoreAccountsConfig.defaultautofield attribute to point to a subclass of AutoField, e.g. ''. blocks.Block: (models.W042) Auto-created primary key used when not defining a primary key type, by default ''. HINT: Configure the DEFAULTAUTOFIELD setting or the AppConfig.defaultautofield attribute to point to a subclass of AutoField, e.g. ''. contactbyform.Feedback: (models.W042) Auto-created primary key used when not defining a primary key type, by default ''. HINT: Configure the DEFAULTAUTOFIELD setting or the SpeedyCoreContactByFormConfig.defaultautofield attribute to point to a subclass of AutoField, e.g. ''. coremessages.ReadMark: (models.W042) Auto-created primary key used when not defining a primary key type, by default ''. HINT: Configure the DEFAULTAUTOFIELD setting or the SpeedyCoreMessagesConfig.defaultautofield attribute to point to a subclass of AutoField, e.g. ''. friendship.Block: (models.W042) Auto-created primary key used when not defining a primary key type, by default ''. HINT: Configure the DEFAULTAUTOFIELD setting or the AppConfig.defaultautofield attribute to point to a subclass of AutoField, e.g. ''. friendship.Follow: (models.W042) Auto-created primary key used when not defining a primary key type, by default ''. HINT: Configure the DEFAULTAUTOFIELD setting or the AppConfig.defaultautofield attribute to point to a subclass of AutoField, e.g. ''. friendship.Friend: (models.W042) Auto-created primary key used when not defining a primary key type, by default ''. HINT: Configure the DEFAULTAUTOFIELD setting or the AppConfig.defaultautofield attribute to point to a subclass of AutoField, e.g. ''. friendship.FriendshipRequest: (models.W042) Auto-created primary key used when not defining a primary key type, by default ''. HINT: Configure the DEFAULTAUTOFIELD setting or the AppConfig.defaultautofield attribute to point to a subclass of AutoField, e.g. ''. likes.UserLike: (models.W042) Auto-created primary key used when not defining a primary key type, by default ''. HINT: Configure the DEFAULTAUTOFIELD setting or the AppConfig.defaultautofield attribute to point to a subclass of AutoField, e.g. ''. uploads.Image: (models.W042) Auto-created primary key used when not defining a primary key type, by default ''. HINT: Configure the DEFAULTAUTOFIELD setting or the AppConfig.defaultautofield attribute to point to a subclass of AutoField, e.g. ''. These models should not use auto-created primary keys! I already defined the primary key in the ancestor of the model. For example class Entity which class User inherits from. It looks to me like a bug in Django 3.2 alpha."}
{"_id":"q-en-django__django-13933-87886fc1a3bc9e4d6f2d5eba4dbc7b9c9c78740bd4b18ce71d6b9d4c7b4cdf96","text":"ModelChoiceField does not provide value of invalid choice when raising ValidationError Description (last modified by Aaron Wiegel) Compared with ChoiceField and others, ModelChoiceField does not show the value of the invalid choice when raising a validation error. Passing in parameters with the invalid value and modifying the default error message for the code invalidchoice should fix this. From source code: class ModelMultipleChoiceField(ModelChoiceField): \"\"\"A MultipleChoiceField whose choices are a model QuerySet.\"\"\" widget = SelectMultiple hiddenwidget = MultipleHiddenInput defaulterrormessages = { 'invalidlist': ('Enter a list of values.'), 'invalidchoice': ('Select a valid choice. %(value)s is not one of the' ' available choices.'), 'invalidpkvalue': ('“%(pk)s” is not a valid value.') } ... class ModelChoiceField(ChoiceField): \"\"\"A ChoiceField whose choices are a model QuerySet.\"\"\" # This class is a subclass of ChoiceField for purity, but it doesn't # actually use any of ChoiceField's implementation. defaulterrormessages = { 'invalidchoice': _('Select a valid choice. That choice is not one of' ' the available choices.'), } ..."}
{"_id":"q-en-django__django-13952-25284c8e8c26811d814222ec6bfe20e2bee178f398d26d8b36c3da908962c3ce","text":"Migrate signals verbose stdout emissions are not capturable Description The migrate command takes a --verbosity flag that is passed down to emitpremigratesignal and emitpostmigratesignal functions but these are not provided which stdout the output should be directed to. This makes testing migrate -v2 through call_command pollute when it should be directed to the provided stdout as discovered in ​"}
{"_id":"q-en-django__django-13964-76d20a0d0da0b9bfec3ffb3cd09a0c6a8a6e9fe6eef9f04bcf02ff9a6e69c90b","text":"Saving parent object after setting on child leads to data loss for parents with non-numeric primary key. Description (last modified by Charlie DeTar) Given a model with a foreign key relation to another model that has a non-auto CharField as its primary key: class Product(models.Model): sku = models.CharField(primarykey=True, maxlength=50) class Order(models.Model): product = models.ForeignKey(Product, ondelete=models.CASCADE) If the relation is initialized on the parent with an empty instance that does not yet specify its primary key, and the primary key is subsequently defined, the parent does not \"see\" the primary key's change: with (): order = Order() order.product = Product() = \"foo\" () () assert (productid=\"\").exists() # Succeeds, but shouldn't assert (product=order.product).exists() # Fails Instead of productid being populated with , it is set to emptystring. The foreign key constraint which would enforce the existence of a product with sku=\"\" is deferred until the transaction commits. The transaction does correctly fail on commit with a ForeignKeyViolation due to the non-existence of a product with emptystring as its primary key. On the other hand, if the related unsaved instance is initialized with its primary key before assignment to the parent, it is persisted correctly: with (): order = Order() order.product = Product(sku=\"foo\") () () assert (product=order.product).exists() # succeeds Committing the transaction also succeeds. This may have something to do with how the Order.productid field is handled at assignment, together with something about handling fetching of auto vs non-auto primary keys from the related instance."}
{"_id":"q-en-django__django-13992-8119f092c5716fb83eef003336c4023e711ea44badac938cc620e6159256a133","text":"Delete distinct produces an unexpected and potentially harmful SQL Description (last modified by egism) I was looking for a way to delete the first Comment of each Post (a sample domain). Since I know that every new Post starts with a system generated comment I decided to go with: Comment.objects.orderby('postid', 'createdat').distinct('postid').delete() Before proceeding I tested it with: Comment.objects.orderby('postid', 'createdat').distinct('postid').count() Made sure the result actually contains what I needed and proceeded with the delete(). The result was rather surprising. I was notified that the whole table was wiped clean. I then checked the actual SQL that was executed and it was a simple DELETE FROM comments;. As an ORM user, I would say it is the worst outcome possible and I would at least expect an error in such a case or ideally a SQL of what I was trying to achieve. At the same time, count and delete produces an inconsistent result which is even more mistaking. Potential solutions: raise an error with a decent explanation produce a desired SQL according to the query Since I have never submitted a change to Django, I have a very limited knowledge of the ORM and its intricacies. I could give it a try and issue a patch for this with some guidance."}
{"_id":"q-en-django__django-13995-837193e6b22e7dcdc12e888ced7c96c0cf021af1210ce43cfd263368abda5cd1","text":"Add in admindocs Description (last modified by Nat S Dunn) From the documentation (​): \"The models section of the admindocs page describes each model in the system along with all the fields, properties, and methods available on it.\" Currently, properties decorated with are not included. Please include them. And possibly include other (or all) descriptors\/attributes."}
{"_id":"q-en-django__django-14007-9a9132100856ffc4aaffc63d5343649f947d615485ff2d89e756325682019b6c","text":"Database converters (fromdbvalue) not called for returningfields on insert Description Making a subclass of BigAutoField, I've found that, unlike all other query pathways, on insert the returned integer is not passed through any database converters defined for the field - including the fromdbvalue hook. This means that a field which would normally use a wrapper class has instead a plain integer. Take this field: class MyAutoField(models.BigAutoField): def fromdbvalue(self, value, expression, connection): if value is None: return None return MyIntWrapper(value) def getprepvalue(self, value): if value is None: return None return int(value) And a model that uses it: class AutoModel(models.Model): id = MyAutoField(primarykey=True) Queried instances have the wrapper class for id: am = () <MyIntWrapper: 1But on creation, the returned integer is directly set as an attribute on the class: am2 = () 2 This also affects bulkcreate on backends that support fetching the primary key value: ams = [AutoModel()] AutoModel.objects.bulkcreate(ams) [<AutoModel: AutoModel object (2)] ams[0].id 2"}
{"_id":"q-en-django__django-14011-881c423836cd2a8263ce16f30faffbc65a2e7ccc35cf9b32cc8b283d09143d58","text":"LiveServerTestCase's ThreadedWSGIServer doesn't close database connections after each thread Description In Django 2.2.17, I'm seeing the reappearance of after it was fixed in 1.11. is the issue where the following error will occur at the conclusion of a test run when destroytestdb() is called: OperationalError: database \"testmyapp\" is being accessed by other users This error happens when not all of the database connections are closed. In my case today, I'm seeing this when running a single test that is a LiveServerTestCase. I see it in approximately half of my test runs, so it's not wholly deterministic (it's a race condition). There weren't a whole lot of changes in the LiveServerTestCase-related code between 1.11 and 2.2, so I looked at them individually. Issue threading support to LiveServerTestCase. One of the changes it made ​was changing LiveServerThread to use ThreadedWSGIServer instead of WSGIServer. LiveServerThread is used by LiveServerTestCase. When I tried modifying LiveServerThread to use the old WSGIServer, I could no longer reproduce the above error. My changes were as follows: class NonThreadedLiveServerThread(LiveServerThread): def createserver(self): return WSGIServer((, ), QuietWSGIRequestHandler, allowreuseaddress=False) class MyTest(LiveServerTestCase): serverthread_class = NonThreadedLiveServerThread The CPython docs ​describe ThreadingMixIn as defining an attribute \"which indicates whether or not the server should wait for thread termination.\" Consistent with what I described above, Aymeric said the following on ticket , seeming to foreshadow issues like this one: more threading will certainly create more race conditions on shutdown, especially when it comes to the database connections — it's taken months to eliminate most from LiveServerTestCase, and I'm sure there are still some left,"}
{"_id":"q-en-django__django-14014-c90f06929b75a2eae2d8c4e174647848f737c876f471268bcd1b24a6b48ad6a9","text":"Introspection doesn't return column order for unique constraints on SQLite. Description Introspection doesn't return column order for unique constraints on SQLite."}
{"_id":"q-en-django__django-14016-550c3d5ed669739fcb07a3198fd8c20924c92f15d56fe37187ea3867ce21536f","text":"\"TypeError: cannot pickle\" when applying | operator to a Q object Description (last modified by Daniel Izquierdo) Using a reference to a non-pickleable type of object such as dictkeys in a Q object makes the | operator fail: from import Q Q(xin={}.keys()) <Q: (AND: ('xin', dictkeys([])))Q() | Q(xin={}.keys()) Traceback (most recent call last): ... TypeError: cannot pickle 'dict_keys' object Even though this particular example could be solved by doing Q() | Q(xin={}) it still feels like using .keys() should work. I can work on a patch if there's agreement that this should not crash."}
{"_id":"q-en-django__django-14017-a267ce819ef002ba3aeed71f5ad7341349c47d844eabc34657eaf63280145d51","text":"Q(...) & Exists(...) raises a TypeError Description Exists(...) & Q(...) works, but Q(...) & Exists(...) raise a TypeError Here's a minimal example: In [3]: Exists(()) & Q() Out[3]: <Q: (AND: object at 0x7fc18dd0ed90, (AND: ))In [4]: Q() & Exists(()) TypeError Traceback (most recent call last) <ipython-input-4-21d3dea0fcb9in <module----1 Q() & Exists(()) ~\/Code\/venv\/ecom\/lib\/python3.8\/site- in and(self, other) 90 91 def and(self, other): ---92 return self.combine(other, self.AND) 93 94 def invert(self): ~\/Code\/venv\/ecom\/lib\/python3.8\/site- in combine(self, other, conn) 71 def _combine(self, other, conn): 72 if not isinstance(other, Q): ---73 raise TypeError(other) 74 75 # If the other Q() is empty, ignore it and just use . TypeError: object at 0x7fc18dd21400The & (and |) operators should be commutative on Q-Exists pairs, but it's not I think there's a missing definition of rand somewhere."}
{"_id":"q-en-django__django-14019-9a39b3fcedba751e201fe75e56cad7bb1d8b7e41720e13c89560b0a277367e0d","text":"Make Index and Constraint repr consistent. Description (last modified by Hannes Ljungberg) Index, UniqueConstraint, ExclusionConstraint and CheckConstraint currently have slightly different formatting on their repr methods. We should keep them consistent."}
{"_id":"q-en-django__django-14026-e6cfdf395507b946e4f672cb4609e73c44b68506fcd5520a1d3afe574c799d36","text":"Support a default value for Sum (and possibly other aggregation functions) Description By default, annotate(sumfield = Sum(...)) results in sumfield being NULL if there were no values to sum. In most cases, 0 would be a better option here, for proper sorting in a later order_by, and for comparisons using lt\/gt\/gte\/lte. A monkeypatch to implement default values for NULL using COALESCE is available here: ​"}
{"_id":"q-en-django__django-14030-ca75305a188022a605c56e53fd4c82c16d1d10b15e26cc60cf63dacbbf19b514","text":"Admin nevercache decorators needs methoddecorator Description The login and logout views in Django's admin site are using the nevercache decorator, but they are both instance methods, so they need to use the methoddecorator. This is not a bug because the nevercache decorator is only operating on the response. The first argument to the decorator is supposed to be the request, but it is in-fact the adminsite instance (self). All the arguments are then passed to the view function and the decorator operate on the response. If you try to use a different decorator that uses the request (such as requiredhttpmethods) you will fail. AttributeError: 'CustomAdminSite' object has no attribute 'method' Related issue from long time ago:"}
{"_id":"q-en-django__django-14031-52c3ae691c8494cb324835182e77977fb9d584b60a346ae39230fb6cfa8e7516","text":"Ticket breaks autocomplete for inherited models Description Unfortunately, Closed ticket breaks autocomplete for the case of inherited models. For example, we have the following code: from import models class Tag(models.Model): pass class Foo(models.Model): tags = models.ManyToManyField(Tag) class Bar(Foo): pass from django.contrib import admin from . import models class Foo(admin.ModelAdmin): autocompletefields = ('tags',) class Bar(admin.ModelAdmin): autocompletefields = ('tags',) Now, autocomplete for admin.Foo will work but not for admin.Bar because () calculates a wrong value of a variable tofieldname, namely fooptr instead of id, whereupon following look up at ().filter(**{'%sin' % tofieldname: selectedchoices}) raises an exception because models.Tag does not have foo_ptr."}
{"_id":"q-en-django__django-14034-12076e59e077e4a7d36cc1be652453d5648fd07f2c0425074d75100a560978f7","text":"MultiValueField ignores a required value of a sub field Description (last modified by Takayuki Hirai) A field and a form definition: from import ( Form, CharField, MultiValueField, MultiWidget, ) class MF(MultiValueField): widget = MultiWidget def init(self): fields = [ CharField(required=False), CharField(required=True), ] widget = (widgets=[ f.widget for f in fields ], attrs={}) super(MF, self).init( fields=fields, widget=widget, requireallfields=False, required=False, ) def compress(self, value): return [] class F(Form): mf = MF() When the form is passed empty values for both sub fields, form.isvalid() == True. But I expected isvalid() returns False, because one of the sub fields is set as required. f = F({ 'mf0': '', 'mf1': '', }) assert f.isvalid() == True # I expect this should return False On the other hand, When one of its sub field is passed a non-empty value, form.isvalid() == False f = F({ 'mf0': 'xxx', 'mf1': '', }) assert f.is_valid() == Flase If above behavior is not expected, please fix this problem."}
{"_id":"q-en-django__django-14043-4808bf4f2040b0bbd4e0828d3fc83ead85db9888facd0c16bfe829a36d613e44","text":"Add support for PostgreSQL passfile to dbshell. Description (last modified by Mariusz Felisiak) The dbshell management commands all carry the risk of leaking passwords through process information (as noted in a comment in ). As of Django 3.2, there is the settingstocmdargsenv class method, which provides an API to generate the environment needed to call the utility. Using the environment is somewhat more secure, but the environment of processes can potentially still be read. Both MySQL and PostgreSQL advise against using the respective environment variables. Specifying ​a password file works for connections but dbshell doesn't support it, see comment. I want to propose a way to solve this. I already did this in django-dbbackup, which also happened to construct a command line before: ​ The mechanism is that in addition to the environment and args, a temporary file is generated. For PostgreSQL, this is a file in .pgpass format; for MySQL, it could be an options file. I wrapped that handling in a neat context manager. For Django itself, I did a quick shot at PostgreSQL as well, as attached in the patch. The patch is not complete, and is only intended as a base for discussion. If we find consensus about the mechanism, I will happily complete it and extend to the other backends."}
{"_id":"q-en-django__django-14053-a66a0027ec07446c88443bccfc961caa216556c9a1d23001c780668189fc4b50","text":"HashedFilesMixin's postprocess() yields multiple times for the same file Description As part of fixing , the implementation of HashedFilesMixin (used by both ManifestStaticFilesStorage and CachedStaticFilesStorage) was changed such that it performs several passes against the found files, therefore ensuring that nested references between the files are correctly handled. Performing these several passes is both necessary and not a problem in itself, however at present postprocess() returns (via yield) the same original filename multiple times back to collectstatic's collect(). For example using Django 1.11.5 with the app enabled: $ collectstatic --noinput | grep '' Copying '\/home\/vagrant\/python\/lib\/python2.7\/site-' Post-processed '' as '' Post-processed '' as '' Post-processed '' as '' ...whereas I would have only expected: $ collectstatic --noinput | grep '' Copying '\/home\/vagrant\/python\/lib\/python2.7\/site-' Post-processed '' as '' The problem with this is that: 1) collectstatic's collect() assumes that the number of yields is the number of files that were post-processed. As such, by yielding multiple times for the same original file, the stats shown at the end (eg \"X files copied, ..., Y post-processed\") are wrong, since there can be more files post processed than were copied 2) For anyone subclassing ManifestStaticFilesStorage who handles the yielded files as they come in, duplicate work is performed. For example WhiteNoise ends up compressing the same file multiple times, increasing deploy times due to expensive Brotli compression. And I'm guessing S3 backends similarly might upload multiple times. 3) Even if it were argued that all files should be yielded, this isn't what is actually happening since only some of the intermittent files are yielded (compare the \"Post-processed ...\" output to the file list in -- the instance is missing). Note that this issue whilst related to is actually different for two reasons: 1) Even if intermediate files need to be left around for now, IMO they still shouldn't be passed back to collectstatic and\/or subclasses (they are a lower-level implementation detail) 2) The duplicate yields occur even for assets that don't need adjusting during the second pass. For example: $ collectstatic --noinput | grep '' Post-processed '' as '' Post-processed '' as '' Post-processed '' as '' This issue was actually mentioned in the PR that the feature: ​"}
{"_id":"q-en-django__django-14056-2d06d317f2a2adf41813ccc5b7a190cd0ddc3f5dfecb4151c8110551567b02bf","text":"Make warn (rather than blow up) on missing directories Description At present if the STATICFILES_DIRS setting contains references to directories which do not exist then the whole command will die with an OSError. A situation I've seen bite a few newcomers to Django is that they will have an empty static directory which (being empty) won't get tracked by git. This means that collectstatic works when they run it locally, but blows up with a (to them) cryptic error when they try to deploy. If we made collectstatic simply log a warning in the case of non-existent directories and continue processing then we could avoid this problem and remove one possible source of frustration for newcomers. If this approach seems acceptable to others I am happy to submit an appropriate patch."}
{"_id":"q-en-django__django-14059-991a92ce9954ce16505528960ecd265340b19bfa114356b01da157536879a666","text":"Can create model instance with conflicting args and kwargs value for the same field Description (last modified by Anton Samarchyan) Discovered while working on To reproduce it, add the following test (it's using ) from import six def testcannotspecifysamefieldwithargsandkwargstoo(self): six.assertRaisesRegex( self, TypeError, \"init() got multiple values for argument 'headline'\", Article, None, # id 'args based headline', headline='kwargs based headline', pubdate=datetime(2005, 7, 31), )"}
{"_id":"q-en-django__django-14071-92ba68a94c853963fad1b5571352918101e0c126dd57613b0abf6523fdb728b0","text":"Admin's rawidfield check admin.E002 doesn't catch .attname mis-references Description Since admin.E002 relies on models.Options.getfield which allows retrieval of fields by both name and attname referring to fields by attname while only name ​is taken into consideration allows the check to pass while rawidfields is not honoured. e.g. class BookAdmin(ModelAdmin): rawidfields = ['authorid'] passes admin.E002 but the author field won't use the rawid feature. The checkrawidfieldsitem method should also make sure to check == fieldname on field retrieval success and return refertomissingfield(field=field_name, option=label, obj=obj, id='admin.E002') when it's not the case."}
{"_id":"q-en-django__django-14077-250ac93ba6f19d4e36a10d54e1afa5a9d061a75747f3d4d2b00d01859b4b2ff2","text":"()\/valueslist() with JSONField returns integers instead of booleans on SQLite. Description (last modified by Matthew Cornell) I have a model with a JSONField: class PredictionData(models.Model): data = models.JSONField() One of the rows contains this dict: {'value': True}. I'm querying the model's JSON using 'datavalue': PredictionData.objects.valueslist('data', 'datavalue') I get correct results for postgres (a boolean) but incorrect for sqlite3 (an int). For this query, sqlite3 wrongly returns: ({'value': True}, 1) whereas postgres correctly returns ({'value': True}, True) Same behavior with False\/0. versions: Python 3.9.1 sqlite3.sqlite_version # '3.33.0' django.version # '3.1.7'"}
{"_id":"q-en-django__django-14089-168665c84d16b1a48df5345936008aaeb105e12825e05b78546b59f284c71624","text":"Allow calling reversed() on an OrderedSet Description Currently, ​OrderedSet isn't reversible (i.e. allowed to be passed as an argument to Python's ​reversed()). This would be natural to support given that OrderedSet is ordered. This should be straightforward to add by adding a reversed() method to OrderedSet."}
{"_id":"q-en-django__django-14109-5a49208b4b0b7fe0d1b2f176f649324daf0ef25f150a988f785d73625b3b40a0","text":"Change automatic migration naming from date-based to operation-based Description Following and a ​discussion on django-developers with broad consensus, change the way migrations are automatically named from date-based to operation-based. That is never name migrations based upon the current date (auto_YYYYMMDD) and instead always based on the operations they contain, rather than the current behaviour which uses either style."}
{"_id":"q-en-django__django-14122-6156eee491e5e85bd9014e69efbb080935a05a9bdf7c11bb2af602ad8a54ab6f","text":"Meta.ordering fields must not be included in GROUP BY clause Description This continues (closed) [1] ticket. I beleave it was not properly fixed in commit []. While commit [] removes ORDER BY when Meta.ordering is used it still does populates GROUP BY with Meta.ordering fields thus leads to wrong aggregation. PR with test case was at [2]. [1] [2] ​​"}
{"_id":"q-en-django__django-14124-228dc2befaf60b147b236ed314934fc3bfcdd9e8b394dcbf79e6fa3098a147e3","text":"handler500 as a Class-based view raises SystemCheckError Description (last modified by Daniyal Abbasi) Setting handler500 as a Class-Based view raises the following error which running checks. $ python check SystemCheckError: System check identified some issues: ERRORS: ?: (urls.E007) The custom handler500 view '' does not take the correct number of arguments (request). In my root , I have the following configuration, handler404 = () handler500 = () I believe this is due to the function checkcustomerrorhandlers in The signature variable in this function is expected to match (request, exception) for all handlers except for handler500 which is expected to have only (request). A positional argument, templatename is also present. While using class based views, we get two positional arguments (self, request) and then it recieves args and kwargs. The check is permitting other handlers as the number of arguments coincidentally match. I suggest a fix in the checkcustomerror_handlers which first checks if the handler are function based or class based, and then it preceed the check with the appropriate number of arguments."}
{"_id":"q-en-django__django-14140-0dd02d6ae25690049bfca721d77d7221404f92da45c0aed79226a2e3c5fc9c24","text":"Combining Q() objects with boolean expressions crashes. Description (last modified by jonathan-golorry) Currently Q objects with 1 child are treated differently during deconstruct. from import Q Q(x=1).deconstruct() ('', (), {'x': 1}) Q(x=1, y=2).deconstruct() ('', (('x', 1), ('y', 2)), {}) This causes issues when deconstructing Q objects with a non-subscriptable child. from import getusermodel from import Exists Q(Exists(getusermodel()(username='jim'))).deconstruct() Traceback (most recent call last): File \"<console\", line 1, in <moduleFile \"...\", line 90, in deconstruct kwargs = {child[0]: child[1]} TypeError: 'Exists' object is not subscriptable Patch ​ removes the special case, meaning single-child Q objects deconstruct into args instead of kwargs. A more backward-compatible approach would be to keep the special case and explicitly check that the child is a length-2 tuple, but it's unlikely that anyone is relying on this undocumented behavior."}
{"_id":"q-en-django__django-14149-929d993d52d680d5cb44746df4d7a391d4f142ac83a1ef84de7e4d02e8afd9f8","text":"Validate the type of ALLOWEDHOSTS Description Python has soft behavior for conducting iteration process over lists and over strings making them look the same: for charoritem in strorlist: -- can be character or list item It would be better if it would have more strict behavior, for example, for char in (): -- now can be only of string type and class would not have method and for list for item in somelist.listitems(): -- class would not have method This soft behavior usually leads to many nasty bugs to appear. Our two software engineers from our team wasted about 1 hour debugging the issue with ALLOWEDHOSTS being initialized with string in which is included at the end of Django was matching each separate character of ALLOWEDHOSTS string against the \"Host:\" header from an incoming HTTP request. An obvious self-suggesting solution is to add a new system check that will check the type of ALLOWED_HOSTS if it is string or not and notify the developer about possible improper configuration. I think blacklist checking (string or not) is more appropiate here, but I can be wrong."}
{"_id":"q-en-django__django-14151-f30d01014405e01295329fd38a0a5ea31f02c71c5bb280d615a33e3ae87f32fa","text":"CsrfViewMiddleware assumes referer header can be parsed Description Django's CsrfViewMiddleware assumes that the HTTP referer header is valid when checking it. Specifically, it doesn't handle the case of urlparse() raising a ValueError in this line (e.g. for urls like 'https:\/\/['): ​"}
{"_id":"q-en-django__django-14155-ac949ff82f4168ceb9a340a51ddbd020764af0a54d7dd9a46f14e0e718a3315e","text":"() doesn't handle functools.partial() nicely. Description (last modified by Nick Pope) When a partial function is passed as the view, the repr shows the func argument as functools.partial which isn't very helpful, especially as it doesn't reveal the underlying function or arguments provided. Because a partial function also has arguments provided up front, we need to handle those specially so that they are accessible in repr. ISTM that we can simply unwrap functools.partial objects in ()."}
{"_id":"q-en-django__django-14164-055b9e69553757760f257bd01abc88534f209979298b55ff2a653b844edf5f5b","text":"tolocale should be idempotent Description In summary: from import tolocale tolocale('en-us') 'enUS' # <- Fine tolocale(tolocale('en-us')) 'enus' # <- Bad Typically, this breaks using e.g. with a locale code like ptBR or zh_Hans. of course, we could document that override takes a language code, not a locale code, but if we can support both without much effort, this would be a lot better."}
{"_id":"q-en-django__django-14169-119545912882cb7a5003daccb178330436d4c6bda932a6ae4354fcb07b450e0f","text":"Extra dot in cloned test database file names on SQLite when using --parallel. Description When asking Django to create on-disk (instead of in-memory) sqlite cloned test databases by using the --parallel flag, the cloned file names are created as, e.g., \"testdb1..sqlite3\", \"testdb2..sqlite3\", etc. (with two dots instead of one). If the specified test database name lacks any extension, e.g. just 'testdb', then the cloned file names have a trailing dot: \"testdb1.\", \"testdb2.\", \"testdb3.\", etc. This is due to this line: ​ gettestdbclonesettings() which constructs the file name using the string: '{}{}.{}'.format(root, suffix, ext) However, () already includes the dot in the returned extension ('ext'). Removing the dot from the format string seems the only change needed to fix it: '{}{}{}'.format(root, suffix, ext) From the github file history it looks like this quirk has been there since the --parallel flag was first introduced (commit in Sep 2015). To reproduce the issue: In , force on-disk instead of in-memory test databases by specifying any test db name: DATABASES = { 'default': { ENGINE: '', NAME: 'db.sqlite3', TEST: { 'NAME': testdb.sqlite3', } } } Create any model with migrations, and more than one TestCase class (to ensure --parallel creates clones of the test database). And on the command line, run tests with --parallel and --keepdb to see the db files generated. This should be run on a multi-core processor. test --parallel --keepdb"}
{"_id":"q-en-django__django-14170-ea106cd2eb2e03d5bd5bf715b16cdc0ec8ce43f32dc9cb8a5d8290a25412e9d0","text":"Query optimization in YearLookup breaks filtering by \"isoyear\" Description (last modified by Florian Demmer) The optimization to use BETWEEN instead of the EXTRACT operation in ​YearLookup is also registered for the ​\"isoyear\" lookup, which breaks the functionality provided by ​ExtractIsoYear when used via the lookup. This has unfortunately been broken ever since ExtractIsoYear was introduced in ​Django 2.2 via and wasn't easy to track down since ExtractIsoYear when used by itself eg. in an annotation works perfectly fine. Just when using the lookup in a filter, the optimization is used (even when explicitly using an annotation): qs = DTModel.objects.annotate(extracted=ExtractIsoYear('startdate')).only('id') print() SELECT \"dbfunctionsdtmodel\".\"id\", EXTRACT('isoyear' FROM \"dbfunctionsdtmodel\".\"startdate\") AS \"extracted\" FROM \"dbfunctionsdtmodel\" print((extracted=2020).query) SELECT \"dbfunctionsdtmodel\".\"id\", EXTRACT('isoyear' FROM \"dbfunctionsdtmodel\".\"startdate\") AS \"extracted\" FROM \"dbfunctionsdtmodel\" WHERE \"dbfunctionsdtmodel\".\"startdate\" BETWEEN 2020-01-01 AND 2020-12-31 print((startdateisoyear=2020).only('id').query) SELECT \"dbfunctionsdtmodel\".\"id\" FROM \"dbfunctionsdtmodel\" WHERE \"dbfunctionsdtmodel\".\"startdate\" BETWEEN 2020-01-01 AND 2020-12-31 This results in the wrong data being returned by filters using isoyear. This PR fixes the behaviour, reverts the invalid changes to the tests and extends one test to catch this problem: ​"}
{"_id":"q-en-django__django-14179-1f034d60618f5fafc06c32d36e7f65e748cc8ff2b6709e0e357e9d8b3f362824","text":"Handle request.gethost() raising DisallowedHost in CsrfViewMiddleware.originverified() Description Currently, on this line, CsrfViewMiddleware.originverified() doesn't handle request.gethost() raising DisallowedHost: ​ Since Django was previously fixed to handle request.gethost() raising DisallowedHost elsewhere in CsrfViewMiddleware.processview() (see ticket ), it seems like it should be handled here, too."}
{"_id":"q-en-django__django-14182-d2cd2d2320e72c6db248f26e31da53f27e87b64c8cea3e36824a4803d627179c","text":"Add support for precision argument to Round Description Django's Round function currently only supports rounding to the nearest integer. If you need to round to a more specific precision you need to roll out your own solution. But as far as I can tell, all backends natively supported by Django support a second argument to Round: Postgres: ​ Sqlite: ​ MySQL: ​ Oracle: ​ In my project (postgres only) my work around was to declare my own custom function: class Round(Func): function = 'ROUND' arity = 2 But having this built-in would be quite useful."}
{"_id":"q-en-django__django-14199-b64aa50485c7276001c617cff6b91fc7589fa417196035f09a402bdcd440eaee","text":"Value() with DecimalField crashes on SQLite. Description If you try to annotate anything to a decimal field Django will pass a string to createdecimalfromfloat function. ​These are the offending lines example code: ().annotate(shearlength=Value(1, output_field=DecimalField()))"}
{"_id":"q-en-django__django-14238-7c4117145f6e47a37a3118e26d98e62f99be905a2894ac8ed10bd5c08756d3ad","text":"DEFAULTAUTOFIELD subclass check fails for subclasses of BigAutoField and SmallAutoField. Description Set DEFAULTAUTOFIELD = \"\" , with contents of from import models class MyBigAutoField(models.BigAutoField): pass class MyModel(models.Model): pass Django then crashes with: Traceback (most recent call last): File \"\", line 21, in <modulemain() File \"\", line 17, in main executefromcommandline() File \"\/...\/venv\/lib\/python3.9\/site-\", line 419, in executefromcommandline utility.execute() File \"\/...\/venv\/lib\/python3.9\/site-\", line 395, in execute () File \"\/...\/venv\/lib\/python3.9\/site-\", line 24, in setup apps.populate(settings.INSTALLEDAPPS) File \"\/...\/venv\/lib\/python3.9\/site-\", line 114, in populate appconfig.importmodels() File \"\/...\/venv\/lib\/python3.9\/site-\", line 301, in importmodels self.modelsmodule = importmodule(modelsmodulename) File \"\", line 127, in importmodule return bootstrap.gcdimport(name[level:], package, level) File \"<frozen importlib.bootstrap\", line 1030, in gcdimport File \"<frozen importlib.bootstrap\", line 1007, in findandload File \"<frozen importlib.bootstrap\", line 986, in findandloadunlocked File \"<frozen importlib.bootstrap\", line 680, in loadunlocked File \"<frozen importlib.bootstrapexternal\", line 790, in execmodule File \"<frozen importlib.bootstrap\", line 228, in callwithframesremoved File \"\", line 8, in <moduleclass MyModel(models.Model): File \"\/...\/venv\/lib\/python3.9\/site-\", line 320, in new newclass.prepare() File \"\/...\/venv\/lib\/python3.9\/site-\", line 333, in prepare opts.prepare(cls) File \"\/...\/venv\/lib\/python3.9\/site-\", line 285, in prepare pkclass = self.getdefaultpkclass() File \"\/...\/venv\/lib\/python3.9\/site-\", line 246, in getdefaultpkclass raise ValueError( ValueError: Primary key '' referred by DEFAULTAUTOFIELD must subclass AutoField. This can be fixed in AutoFieldMeta.subclasscheck by allowing subclasses of those classes in the subclasses property."}
{"_id":"q-en-django__django-14241-941eafee615598c65891fe009c6832b7b45706366d1ceb5f5b4f3c7245a6ae45","text":"()\/valueslist() crash on combined querysets ordered by unannotated columns. Description (last modified by Iuri de Silvio) Django 3.2 fails with this query: qs1 = () qs2 = () (qs2).valueslist('name').first() It worked until Django 3.1.8. This commit[1] to be exactly. ​ This is the broken generated query. In the second query, it fetches from the first table. SQL SELECT \"queriescelebrity\".\"name\", \"queriescelebrity\".\"id\" AS \"orderbycol2\" FROM \"queriescelebrity\" UNION SELECT \"queriesreservedname\".\"name\", \"queriescelebrity\".\"id\" AS \"orderbycol2\" -- HERE IS THE PROBLEM FROM \"queriesreservedname\" ORDER BY (2) ASC LIMIT 1 Before, it was: SQL SELECT \"queriescelebrity\".\"name\", \"queriescelebrity\".\"id\" FROM \"queriescelebrity\" UNION SELECT \"queriesreservedname\".\"name\", \"queriesreservedname\".\"id\" FROM \"queriesreservedname\" ORDER BY (2) ASC LIMIT 1"}
{"_id":"q-en-django__django-14266-e8dcbc6c9ea5b9dbd2099757a0d3423b7dab866fe1604c6502bdc710e850c357","text":"CookieStorage for contrib.messages crashes after upgrade to django 3.2 Description (last modified by Jan Pieter Waagmeester) After upgrading to django 3.2, a previously stored cookie for contrib.messages crashes in ​ Django Version: 3.2 Python Version: 3.8.2 Traceback (most recent call last): File \"\/home\/obs\/virtualenv\/lib\/python3.8\/site-\", line 47, in inner response = getresponse(request) File \"\/home\/obs\/virtualenv\/lib\/python3.8\/site-\", line 119, in call response = self.processresponse(request, response) File \"\/home\/obs\/virtualenv\/lib\/python3.8\/site-\", line 23, in processresponse unstoredmessages = (response) File \"\/home\/obs\/virtualenv\/lib\/python3.8\/site-\", line 127, in update messages = self.loadedmessages + self.queuedmessages File \"\/home\/obs\/virtualenv\/lib\/python3.8\/site-\", line 79, in loadedmessages messages, allretrieved = () File \"\/home\/obs\/virtualenv\/lib\/python3.8\/site-\", line 25, in get messages, allretrieved = () File \"\/home\/obs\/virtualenv\/lib\/python3.8\/site-\", line 86, in get messages = (data) File \"\/home\/obs\/virtualenv\/lib\/python3.8\/site-\", line 175, in decode return (data, serializer=MessageSerializer) File \"\/home\/obs\/virtualenv\/lib\/python3.8\/site-\", line 195, in unsignobject data = b64decode(base64d) File \"\/home\/obs\/virtualenv\/lib\/python3.8\/site-\", line 68, in b64decode return base64.urlsafeb64decode(s + pad) File \"\", line 133, in urlsafeb64decode return b64decode(s) File \"\", line 87, in b64decode return binascii.a2bbase64(s) Exception Type: Error at \/user\/login\/ Exception Value: Invalid base64-encoded string: number of data characters (369) cannot be 1 more than a multiple of 4 (redacted) contents of the 'messages' cookie: '[[\"jsonmessage\",0,25,\"Successfully signed in as ' '\"],[\"jsonmessage\",0,25,\"Successfully ' 'signed in as jieter.\"],[\"jsonmessage\",0,25,\"Ingelogd als ' '\"],[\"jsonmessage\",0,25,\"Ingelogd ' 'als ' '\"],[\"jsonmessage\",0,20,\"Bevestigingsmail ' 'verzonden naar \"],[\"jsonmessage\",0,25,\"Ingelogd ' 'als ' '\"]]:1lTkj1:j3PlpYSKiqPTMAB6p2Q00eE8j6k7n0Sg-IpXG7Y')"}
{"_id":"q-en-django__django-14267-0dd02d6ae25690049bfca721d77d7221404f92da45c0aed79226a2e3c5fc9c24","text":"Combining Q() objects with boolean expressions crashes. Description (last modified by jonathan-golorry) Currently Q objects with 1 child are treated differently during deconstruct. from import Q Q(x=1).deconstruct() ('', (), {'x': 1}) Q(x=1, y=2).deconstruct() ('', (('x', 1), ('y', 2)), {}) This causes issues when deconstructing Q objects with a non-subscriptable child. from import getusermodel from import Exists Q(Exists(getusermodel()(username='jim'))).deconstruct() Traceback (most recent call last): File \"<console\", line 1, in <moduleFile \"...\", line 90, in deconstruct kwargs = {child[0]: child[1]} TypeError: 'Exists' object is not subscriptable Patch ​ removes the special case, meaning single-child Q objects deconstruct into args instead of kwargs. A more backward-compatible approach would be to keep the special case and explicitly check that the child is a length-2 tuple, but it's unlikely that anyone is relying on this undocumented behavior."}
{"_id":"q-en-django__django-14271-218c895fc52fd3692942ad699773d81c089fdd97fb4aa1e14ff1bf9f466aac2f","text":"Cannot combine two queryset in a subquery Description (last modified by Raffaele Salmaso) [Sample project ​ and run query] Django 3.2 fails this query (a combined queryset in a subquery): import datetime as dt from decimal import Decimal from import settings from import models from import Case, OuterRef, Q, Subquery, Value, When from import timezone class UserQuerySet(models.QuerySet): def annotateactivesubscriptionid(self): return self.annotate( activesubscriptioniddb=Subquery( () .annotate( planorder=Case( When(plancode=\"BASE\", then=Value(1)), default=Value(0), outputfield=models.PositiveSmallIntegerField(), ) ) .filter(user=OuterRef(\"id\")) .orderby(\"planorder\", \"-id\") .values(\"id\")[:1] ) ) class User(models.Model): objects = models.Manager.fromqueryset(UserQuerySet)() class Plan(models.Model): code = models.CharField(verbosename=\"Codice\", maxlength=255) class SubscriptionQuerySet(models.QuerySet): def willberenewedtoday(self): today = () return (startdatelte=today).exclude(usersubscriptionsstartdate=today).distinct() def active(self): return (enabled=True).distinct() | self.willberenewedtoday() class Subscription(models.Model): user = models.ForeignKey(User, verbosename=\"Utente\", ondelete=models.CASCADE, relatedname=\"subscriptions\") plan = models.ForeignKey(Plan, ondelete=models.CASCADE, verbosename=\"Piano di abbonamento\") startdate = models.DateField(verbosename=\"Data di inizio\", ) enabled = models.BooleanField(verbosename=\"Abilitato\", default=True) objects = models.Manager.fromqueryset(SubscriptionQuerySet)() print(User.objects.annotateactivesubscriptionid().count()) with django 3.1.8 SELECT \"subqueryuser\".\"id\", ( SELECT \"subquery\".\"id\" FROM ( SELECT DISTINCT U0.\"id\", CASE WHEN (U2.\"code\" = BASE) THEN 1 ELSE 0 END FROM \"subquerysubscription\" U0 INNER JOIN \"subqueryplan\" U2 ON (U0.\"planid\" = U2.\"id\") WHERE ( ( U0.\"enabled\" OR ( U0.\"startdate\" <= 2021 - 04 - 13 AND NOT ( U0.\"userid\" IN ( SELECT U2.\"userid\" FROM \"subquerysubscription\" U2 WHERE U2.\"startdate\" = 2021 - 04 - 13 ) ) ) ) AND U0.\"userid\" = \"subqueryuser\".\"id\" ) ORDER BY CASE WHEN (U2.\"code\" = BASE) THEN 1 ELSE 0 END ASC, U0.\"id\" DESC LIMIT 1 ) subquery ) AS \"activesubscriptioniddb\" FROM \"subqueryuser\" with django 3.2 ( SELECT \"subqueryuser\".\"id\", ( SELECT \"subquery\".\"id\" FROM ( SELECT DISTINCT U0.\"id\", CASE WHEN (U2.\"code\" = BASE) THEN 1 ELSE 0 END FROM \"subquerysubscription\" U0 INNER JOIN \"subqueryplan\" U2 ON (U0.\"planid\" = U2.\"id\") WHERE ( ( U0.\"enabled\" OR ( U0.\"startdate\" <= 2021 - 04 - 13 AND NOT ( EXISTS( SELECT (1) AS \"a\" FROM \"subquerysubscription\" V2 WHERE ( V2.\"startdate\" = 2021 - 04 - 13 AND V2.\"userid\" = V0.\"userid\" ) LIMIT 1 ) ) ) ) AND U0.\"userid\" = \"subqueryuser\".\"id\" ) ORDER BY CASE WHEN (U2.\"code\" = BASE) THEN 1 ELSE 0 END ASC, U0.\"id\" DESC LIMIT 1 ) subquery ) AS \"activesubscriptioniddb\" FROM \"subqueryuser\" Traceback (most recent call last): File \".venvs\/django32\/lib\/python3.8\/site-\", line 84, in execute return (sql, params) File \".venvs\/django32\/lib\/python3.8\/site-\", line 423, in execute return Database.Cursor.execute(self, query, params) sqlite3.OperationalError: no such column: The above exception was the direct cause of the following exception: Traceback (most recent call last): File \"\", line 22, in <modulemain() File \"\", line 18, in main executefromcommandline() File \".venvs\/django32\/lib\/python3.8\/site-\", line 419, in executefromcommandline utility.execute() File \".venvs\/django32\/lib\/python3.8\/site-\", line 413, in execute self.fetchcommand(subcommand).runfromargv() File \".venvs\/django32\/lib\/python3.8\/site-\", line 354, in runfromargv self.execute(args, cmdoptions) File \".venvs\/django32\/lib\/python3.8\/site-\", line 398, in execute output = (args, options) File \"\/home\/raf\/src\/fiscozen\/django-\", line 11, in handle print(()) File \".venvs\/django32\/lib\/python3.8\/site-\", line 412, in count return () File \".venvs\/django32\/lib\/python3.8\/site-\", line 526, in getcount number = obj.getaggregation(using, ['count'])['count'] File \".venvs\/django32\/lib\/python3.8\/site-\", line 511, in getaggregation result = compiler.executesql(SINGLE) File \".venvs\/django32\/lib\/python3.8\/site-\", line 1175, in executesql cursor.execute(sql, params) File \".venvs\/django32\/lib\/python3.8\/site-\", line 98, in execute return super().execute(sql, params) File \".venvs\/django32\/lib\/python3.8\/site-\", line 66, in execute return self.executewithwrappers(sql, params, many=False, executor=self.execute) File \".venvs\/django32\/lib\/python3.8\/site-\", line 75, in executewithwrappers return executor(sql, params, many, context) File \".venvs\/django32\/lib\/python3.8\/site-\", line 84, in execute return (sql, params) File \".venvs\/django32\/lib\/python3.8\/site-\", line 90, in exit raise djexcvalue.withtraceback(traceback) from excvalue File \".venvs\/django32\/lib\/python3.8\/site-\", line 84, in execute return (sql, params) File \".venvs\/django32\/lib\/python3.8\/site-\", line 423, in execute return Database.Cursor.execute(self, query, params) no such column: V0.user_id Tested with 3.2, ​ (next 3.2.1) and ​ (main)"}
{"_id":"q-en-django__django-14282-89e582bbf4fd5def0d2a278b71212d86a76ba1bb36ea4bc973de559c2876486c","text":"Cannot run makemigrations management command without a SECRETKEY Description I believe intended to fix this issue. Steps to reproduce: (mktemp -d) $ python -m venv venv $ source venv\/bin\/activate $ pip install 'Django=3.2' $ python -m django startproject foo $ sed -ri '\/SECRETKEY\/d' # Remove SECRETKEY from settings $ PYTHONPATH=foo DJANGOSETTINGS_MODULE=\"foo.settings\" python -m django makemigrations --check The output is attached."}
{"_id":"q-en-django__django-14291-fc2f0cf12217e1139e534f045c5576737141775849079d87f32ead3291ff4715","text":"caches.W002 check does not support tuples in STATICFILESDIRS Description The caches.W002 check defined here: ​ inspects the values of each entry in STATICFILESDIRS here: ​ and passes them to pathlib.Path(staticfilesdir) which expects a string, however according to the documentation each entry in STATICFILESDIRS may be either a string or a tuple: ​ If a STATICFILESDIRS entry is provided as a tuple, this check fails with: TypeError: expected str, bytes or os.PathLike object, not tuple in in parse_args at line 680"}
{"_id":"q-en-django__django-14311-7aa540e3b1c74afc066aa614fe54e55f6b390fb14447a8d099b910d7d2f9b35e","text":"Allow autoreloading of Description (last modified by Mariusz Felisiak) The original fix [1] only attempted to deal with -m where bar is a package and exists under foo\/bar. When a dotted name for a module (for example, where resides under foo\/bar) is specified like -m , the resulting arguments end up being -m , which is uncalled for. [1] ​ Fixed detection when started non-django modules with \"python -m\" in autoreloader."}
{"_id":"q-en-django__django-14315-c429e6579561b7e7f9a93ffb6717f730220954255b908e52047633472b1435c3","text":"database client runshell doesn't respect os.environ values in some cases Description (last modified by Konstantin Alekseev) postgresql client returns empty dict instead of None for env as a result os.environ is not used and empty env passed to subprocess. Bug introduced in ​ PR ​"}
{"_id":"q-en-django__django-14324-5857e84747bed4e94b84cb2cf43eb4fe7a74bddc5e1ab485ec365f32297d3a87","text":"Migration autodetector changes relatedname for self-referential ManyToManyField. Description Migration autodetector no longer adds a model name to the relatedname attribute for self-referential ManyToManyField, e.g. for a field class MyModel2(models.Model): field3 = models.ManyToManyField('self') it creates a migration with relatedname='field3rel+' instead of relatedname='mymodel2field3+'. Regression in (see )."}
{"_id":"q-en-django__django-14334-083fcdd05e546d4f6fb21be151fa88e4be4b6d190b5b0694312ca16648347ba3","text":"LocaleMiddleware not recognising properly zh-Hant-HK from the accept-language header Description If both zh-hans and zh-hant are in settings.LANGUAGES (and in this order) the LocaleMiddleware is choosing the first one instead of the second as would be expected. The actual faulty code seems to be in"}
{"_id":"q-en-django__django-14336-693a2104195064468d7ad2247724b3e53b0729a8d577daa52c98350c9f8ffa3c","text":"case sensitive issue on subquery aliases generated Description When generating aliases for subqueries, django uses 2 different cases: 'Col%d' ​ and 'col%d' ​ Here is a concrete example, using --v 2 --debug-sql SELECT COUNT(*) FROM (SELECT \"subquery\".\"col1\", \"subquery\".\"col2\", \"subquery\".\"col3\", \"subquery\".\"col4\" FROM (SELECT DISTINCT \"queriestag\".\"id\" AS Col1, \"queriestag\".\"name\" AS Col2, \"queriestag\".\"parentid\" AS Col3, \"queriestag\".\"categoryid\" AS Col4, \"queriesnamedcategory\".\"name\" AS Col5 FROM \"queriestag\" LEFT OUTER JOIN \"queriesnamedcategory\" ON (\"queriestag\".\"categoryid\" = \"queriesnamedcategory\".\"dumbcategoryptrid\") ORDER BY \"queriesnamedcategory\".\"name\" ASC LIMIT 3) subquery) subquery; This would fail on a database with case sensitive column names. It will be trivial to fix if deemed necessary."}
{"_id":"q-en-django__django-14341-b614f6e4dcc5bc0cb4129cc18ac2647d9a64be5479cea60f32f59e9c3f41c898","text":"Database uses cursor after it is closed Description (last modified by ecogels) The return bool(cursor.rowcount) is outside of the with block, so the cursor will have been closed at that point. From the DB API 2.0 spec: \"The cursor will be unusable from this point forward\" ​ As the main backend drivers don't mind it I suppose that is is open to interpretation. PR ​"}
{"_id":"q-en-django__django-14349-73ffab6e07af3de403a89deaf40c823044abf37e1551b7ff32235b61bd58d222","text":"URLValidator tests failing on Python versions patched for bpo- Description On Python versions with a fix for ​bpo- (i.e. 3.10.0b1 and the 3.9 git branch, not released yet) the following tests fail: FAIL: testvalidators () [URLValidator] (value='\\n') Traceback (most recent call last): File \"\", line 59, in testPartExecutor yield File \"\", line 546, in subTest yield File \"\/tmp\/portage\/dev-python\/django-3.2.1\/work\/Django-\", line 328, in testvalidators validator(value) File \"\", line 203, in exit self.raiseFailure(\"{} not raised\".format(excname)) File \"\", line 135, in raiseFailure raise self.testcase.failureException(msg) AssertionError: ValidationError not raised FAIL: testvalidators () [URLValidator] (value='http:\/\/[::ffff:192.9.5.5]\\n') Traceback (most recent call last): File \"\", line 59, in testPartExecutor yield File \"\", line 546, in subTest yield File \"\/tmp\/portage\/dev-python\/django-3.2.1\/work\/Django-\", line 328, in testvalidators validator(value) File \"\", line 203, in exit self.raiseFailure(\"{} not raised\".format(excname)) File \"\", line 135, in raiseFailure raise self.testcase.failureException(msg) AssertionError: ValidationError not raised FWICS, the project is that django rejects URLs based on the split URL components. However, the bpo- fix changes URL splitting behavior to strip all instances of LF, CR and tab characters before splitting, so they never reach the validator. I'm not sure what the best fix is. One option is to reject URLs containing the forbidden characters early. Another is to go with the new recommendation and assume that LF, CR and tabs are to stripped silently."}
{"_id":"q-en-django__django-14351-fd52c35596df7608e40df1c29aa289361c2297af4796a242f3ba00f065454772","text":"Q object or appears to get all dunder related's default columns and queryset raises ProgrammingError. Description There appears to be a difference in how Q object aliases are setup, when OR'd. The getdefaultcolumns for this agentpropertygroupsidin only uses 1, where as getdefaultcolumns for this agentpropertygroupsin gets all fields, which later results in a \" subquery must return only one column\" error. ( Q(agentpropertygroupsidin=propertygroups.valueslist(\"id\", flat=True)) | Q(agentpropertygroupscount=0) ).distinct() ( Q(agentpropertygroupsin=propertygroups) | Q(agentpropertygroupscount=0) ).distinct() Here is the error: return (sql, params) File \"\/venv\/lib\/python3.6\/site-\", line 90, in exit raise djexcvalue.withtraceback(traceback) from excvalue File \"\/venv\/lib\/python3.6\/site-\", line 84, in execute return (sql, params) subquery must return only one column LINE 1: ...ativemovingaverage\".\"id\", T5.\"propertygroupid\", (SELECT U0... ^ For example, I was able to force it to work by checking the cols[0].alias checking that it was 'U0' so that the cols, selectfields, and klassinfo would only return the field needed within the Q object or # if cols: selectlist = [] # these two lines, just to hack a debug fix if cols[0].alias == 'U0': cols = [cols[0]] Was working ( 2.2.5 ), now not working ( 3.2 ): propertygroups = PropertyGroup.objects.agentgroups(managementagent) queryset = self.annotate(Count(\"agentpropertygroups\")) return ( Q(agentpropertygroupsin=propertygroups) | Q(agentpropertygroupscount=0) ).distinct() now working: qs = blah propertygroups = PropertyGroup.objects.agentgroups(managementagent) queryset = qs.annotate(Count(\"agentpropertygroups\")) ( Q(agentpropertygroupsidin=propertygroups.valueslist(\"id\", flat=True)) | Q(agentpropertygroupscount=0) ).distinct() the generated sql SELECT COUNT(*) FROM ( SELECT DISTINCT \"thingmanagerticketratingcumulativemovingaverage\".\"id\" AS Col1, \"thingmanagerticketratingcumulativemovingaverage\".\"created\" AS Col2, \"thingmanagerticketratingcumulativemovingaverage\".\"updated\" AS Col3, \"thingmanagerticketratingcumulativemovingaverage\".\"createby\" AS Col4, \"thingmanagerticketratingcumulativemovingaverage\".\"updateby\" AS Col5, \"thingmanagerticketratingcumulativemovingaverage\".\"tenantobjs\" AS Col6, \"thingmanagerticketratingcumulativemovingaverage\".\"date\" AS Col7, \"thingmanagerticketratingcumulativemovingaverage\".\"average\" AS Col8, \"thingmanagerticketratingcumulativemovingaverage\".\"datapoints\" AS Col9, \"thingmanagerticketratingcumulativemovingaverage\".\"agentid\" AS Col10, COUNT(\"managermanagementagentpropertygroup\".\"propertygroupid\") AS \"agentpropertygroupscount\" FROM \"thingmanagerticketratingcumulativemovingaverage\" INNER JOIN \"managermanagementagent\" ON (\"thingmanagerticketratingcumulativemovingaverage\".\"agentid\" = \"managermanagementagent\".\"id\") LEFT OUTER JOIN \"managermanagementagentpropertygroup\" ON (\"managermanagementagent\".\"id\" = \"managermanagementagentpropertygroup\".\"managementagentid\") LEFT OUTER JOIN \"managermanagementagentpropertygroup\" T5 ON (\"managermanagementagent\".\"id\" = T5.\"managementagentid\") GROUP BY \"thingmanagerticketratingcumulativemovingaverage\".\"id\", T5.\"propertygroupid\", ( -- the issue is right here SELECT U0.\"id\", U0.\"created\", U0.\"updated\", U0.\"createby\", U0.\"updateby\", U0.\"tenantobjs\", U0.\"name\" -- the issue is the line above FROM \"propertypropertygroup\" U0 INNER JOIN \"managermanagementagentpropertygroup\" U1 ON (U0.\"id\" = U1.\"propertygroupid\") WHERE U1.\"managementagentid\" = %s) HAVING ( T5.\"propertygroupid\" IN ( SELECT U0.\"id\" FROM \"propertypropertygroup\" U0 INNER JOIN \"managermanagementagentpropertygroup\" U1 ON (U0.\"id\" = U1.\"propertygroupid\") WHERE U1.\"managementagentid\" = %s) OR COUNT(\"managermanagementagentpropertygroup\".\"propertygroupid\") = %s) ); The sub select which causes the error: SELECT U0.\"id\", U0.\"created\", U0.\"updated\", U0.\"createby\", U0.\"updateby\", U0.\"tenantobjs\", U0.\"name\" Looking into how th Q object looks and how the generated columns look: <Q: (OR: ('agentpropertygroupsin', <PropertyGroupQuerySet []), ('agentpropertygroupscount', 0)),) {} -try: (Pdb) c uhoh {'model': <class '', 'selectfields': [0, 1, 2, 3, 4, 5, 6]} [(Col(U0, ), ('U0.\"id\"', []), None), (Col(U0, property.PropertyGroup.created), ('U0.\"created\"', []), None), (Col(U0, property.PropertyGroup.updated), ('U0.\"updated\"', []), None), (Col(U0, property.PropertyGroup.createby), ('U0.\"createby\"', []), None), (Col(U0, property.PropertyGroup.updateby), ('U0.\"updateby\"', []), None), (Col(U0, property.PropertyGroup.tenantobjs), ('U0.\"tenantobjs\"', []), None), (Col(U0, ), ('U0.\"name\"', []), None)] {'model': <class '', 'selectfields': [0, 1, 2, 3, 4, 5, 6]} {} <Q: (OR: ('agentpropertygroupsidin', <PropertyGroupQuerySet []), ('agentpropertygroupscount', 0)),) {} -try: (Pdb) c uhoh {'model': <class '', 'selectfields': [0]} [(Col(U0, ), ('U0.\"id\"', []), None)] {'model': <class '', 'selectfields': [0]} {} extraselect [] The sub select query: (Pdb) print(self) SELECT U0.\"id\", U0.\"created\", U0.\"updated\", U0.\"createby\", U0.\"updateby\", U0.\"tenantobjs\", U0.\"name\" FROM \"propertypropertygroup\" U0 INNER JOIN \"managermanagementagentpropertygroup\" U1 ON (U0.\"id\" = U1.\"propertygroupid\") WHERE U1.\"managementagentid\" = 342 (Pdb) pprint() {'annotationselectcache': None, 'constructorargs': ((<class '',), {}), 'db': None, 'extraselectcache': None, 'filteredrelations': {}, 'lookupjoins': ['propertypropertygroup', 'managermanagementagentpropertygroup', 'managermanagementagent'], 'aliascols': True, 'aliasmap': {'U0': object at 0x7fc1efd77208, 'U1': object at 0x7fc1efd77828, 'U2': object at 0x7fc1efd777f0}, 'aliasprefix': 'U', 'aliasrefcount': {'U0': 1, 'U1': 1, 'U2': 0}, 'annotationselectmask': None, 'annotations': {}, 'basetable': 'U0', 'combinator': None, 'combinatorall': False, 'combinedqueries': (), 'containsaggregate': False, 'defaultcols': True, 'defaultordering': False, 'deferredloading': (frozenset(), True), 'distinct': False, 'distinctfields': (), 'explainformat': None, 'explainoptions': {}, 'explainquery': False, 'externalaliases': {'managermanagementagent': False, 'managermanagementagentpropertygroup': False, 'thingmanagerticketratingcumulativemovingaverage': False, 'propertypropertygroup': False}, 'extra': {}, 'extraorderby': (), 'extraselectmask': None, 'extratables': (), 'filterissticky': False, 'groupby': None, 'highmark': None, 'lowmark': 0, 'maxdepth': 5, 'model': <class '', 'orderby': (), 'select': (), 'selectfornokeyupdate': False, 'selectforupdate': False, 'selectforupdatenowait': False, 'selectforupdateof': (), 'selectforupdateskiplocked': False, 'selectrelated': False, 'standardordering': True, 'subqaliases': frozenset({'T', 'U'}), 'subquery': True, 'tablemap': {'managermanagementagent': ['U2'], 'managermanagementagentpropertygroup': ['U1'], 'propertypropertygroup': ['U0']}, 'usedaliases': {'managermanagementagentpropertygroup', 'propertypropertygroup'}, 'valuesselect': (), 'where': <WhereNode: (AND: object at 0x7fc1efd77860), 'whereclass': <class ''}"}
{"_id":"q-en-django__django-14368-9e11c086ca1fa69b99fb51bf8233561eba65f46e6c39bb492fd7b9071f90866c","text":"Support spaces separaters in ISO-8601 datetimes between times and timezone strings Description (last modified by Ben Wilber) GitHub Pull Request: ​ According to ISO-8601, there can be any number of whitespace characters between the time strings and timezone strings. Unfortunately the spec isn't public, but here's the link anyway ​ Examples: This is a valid ISO-8601 datetime string: 2012-04-23T10:20:30.400-02 parses this correctly. This is also a valid ISO-8601 datetime string: 2012-04-23T10:20:30.400 -02 does not parse this correctly and returns None, However, python-dateutil parses it correctly. The difference is that Django uses a (brittle) regex to parse ISO-8601 datetime strings, and python-dateutil does not. ​ ​ I recommend that Django: 1) Depend on python-dateutil for datetime string parsing OR 2) Inline python-dateutils' parsing functions As far as I know there is no regex that can parse the full spec of ISO-8601 datetime strings. In the meantime, this is a patch to support (valid) whitespace characters between the seconds\/millseconds part and the timezone string."}
{"_id":"q-en-django__django-14372-1cbbdc72c43e662f06355170db3e9be7026ffb501b206c33fd2d07829ac2bb2f","text":"Saving a FileField raises SuspiciousFileOperation in some scenarios. Description I came across this issue today when I was updating Django from 3.2.0 -3.2.1. It's directly caused by: ​ Starting from 3.2.1, Django requires that only the basename is passed to method, because otherwise it raises a new exception: SuspiciousFileOperation: File name ... includes path elements The issue is that in FileField.presave, a full path is passed to , causing the exception to be raised. Correct me if I'm wrong, but file-like objects always contain the full path to the file in the name attribute (the built-in Django File class even uses it to reopen the file if it was closed), and so it seems to be a bug in Django itself. Steps to reproduce: modelinstance.fileattribute = File(open(path, 'rb')) () I also created a PR with the fix: ​"}
{"_id":"q-en-django__django-14373-71d8a4fa552c7e3e94821c3ee1c03698d67528a7de142c31feb96761dad760c6","text":"DateFormat.Y() is not zero-padded. Description The Y specifier for is supposed to always return a four-digit year padded with zeros. This doesn't seem to be the case for year < 1000."}
{"_id":"q-en-django__django-14376-88be40ae3699599e5133b59fbafc1384a5a262329e6af473f03f3692af17a54b","text":"MySQL backend uses deprecated \"db\" and \"passwd\" kwargs. Description The \"db\" and \"passwd\" usage can be seen at ​ in main. mysqlclient recently marked these two kwargs as deprecated (see ​) in favor of \"database\" and \"password\" respectively. mysqlclient support for \"database\" and \"password\" in 1.3.8 with ​ Django 2.2, 3.1, and 3.2 all require a minimum version of mysqlclient newer than 1.3.8, so a fix for this could be backported to all currently supported versions of Django."}
{"_id":"q-en-django__django-14382-eb466172f679fbd743098523b04e1774694a8beb7a8798542a22e78896fdd4d7","text":"django-admin startapp with trailing slash in directory name results in error Description Bash tab-completion appends trailing slashes to directory names. django-admin startapp name directory\/ results in the error: CommandError: '' is not a valid app directory. Please make sure the directory is a valid identifier. The error is caused by ​line 77 of by calling basename() on the path with no consideration for a trailing slash: self.validatename((target), 'directory') Removing potential trailing slashes would solve the problem: self.validatename((()), 'directory')"}
{"_id":"q-en-django__django-14385-43300d58f746cd65117744086666c090344c3492fc060af0d585eafa3644c0a4","text":"simplify_regex only removes some metacharacters from regex patterns Description Input Pattern Expected Output Actual Output r'^\\b(?P<slug\\w+)\\B' \/<slug\/\\b<slug\\B r'\\Ab\/\\Z' \/b\/ \/\\Ab\/\\Z"}
{"_id":"q-en-django__django-14387-7c76eafb0e2f26abf749f02c138d046e5ab585f5d321a03e14bc1daa8101a792","text":"Incorrect SQL generation filtering OR-combined queries Description I'm running the just-released Django 3.2.1 and am seeing what I think is incorrect SQL generation involving this model (cut down for brevity): from import models as dbmodels class Buss(dbmodels.Model): MAXLENGTH = 25 CHOICES = [('Universal', 'Universal'), ('GB', 'GB'), ('US', 'US'), ('Company', 'Company')] jurisdiction = dbmodels.CharField(maxlength=MAXLENGTH, choices=CHOICES) name = dbmodels.CharField(maxlength=MAXLENGTH) class Meta: uniquetogether = [('jurisdiction', 'name')] I have a function which returns a queryset by combining 3 sets of busses using the \"|\" OR operator: from import Buss def jurisdictionqs(forjurisdiction): # Get busses identified by \"jurisdictionfor\", and add other busses from 'Universal' and 'Company' where they don't clash. qs = (jurisdiction=forjurisdiction) if forjurisdiction != 'Universal': qs = qs | (jurisdiction='Universal'). \\ exclude(namein=qs.valueslist('name', flat=True)) if forjurisdiction != 'Company': qs = qs | (jurisdiction='Company'). \\ exclude(namein=qs.valueslist('name', flat=True)) return qs In use, the function seems to work as expected: In [7]: (jurisdiction='GB').count() Out[7]: 8 In [11]: (jurisdiction_in=['GB','Universal','Company']).count() Out[11]: 37 In [12]: jurisdictionqs('GB').count() Out[12]: 34 However, if the OR'd queryset is further filtered, the results are unpredictable. For example, this works: In [13]: jurisdictionqs('GB').filter(jurisdiction='US').count() Out[13]: 0 but this - where the filter is by the original \"GB\" - returns 34 instead of 8: In [14]: jurisdictionqs('GB').filter(jurisdiction='GB').count() Out[14]: 34 I can see that the SQL from the function looks OK: str(jurisdictionqs('GB').query) SELECT \"paiyrollbuss\".\"id\", \"paiyrollbuss\".\"jurisdiction\", \"paiyrollbuss\".\"name\", \"paiyrollbuss\".\"description\" FROM \"paiyrollbuss\" WHERE ( \"paiyrollbuss\".\"jurisdiction\" = GB OR (\"paiyrollbuss\".\"jurisdiction\" = Universal AND NOT (\"paiyrollbuss\".\"name\" IN (SELECT U0.\"name\" FROM \"paiyrollbuss\" U0 WHERE U0.\"jurisdiction\" = GB)) ) OR (\"paiyrollbuss\".\"jurisdiction\" = Company AND NOT (\"paiyrollbuss\".\"name\" IN (SELECT V0.\"name\" FROM \"paiyrollbuss\" V0 WHERE (V0.\"jurisdiction\" = GB OR (V0.\"jurisdiction\" = Universal AND NOT (V0.\"name\" IN (SELECT U0.\"name\" FROM \"paiyrollbuss\" U0 WHERE U0.\"jurisdiction\" = GB)) )))) ) ) In the working case, the above SQL is changed to end as follows: str(jurisdictionqs('GB').filter(jurisdiction='US').query) SELECT ...WHERE (... AND \"paiyrollbuss\".\"jurisdiction\" = US) but in the broken case, the original SQL is returned! str(jurisdictionqs('GB').filter(jurisdiction='GB').query) SELECT \"paiyrollbuss\".\"id\", \"paiyrollbuss\".\"jurisdiction\", \"paiyrollbuss\".\"name\", \"paiyrollbuss\".\"description\" FROM \"paiyrollbuss\" WHERE (\"paiyrollbuss\".\"jurisdiction\" = GB OR (\"paiyrollbuss\".\"jurisdiction\" = Universal AND NOT (\"paiyrollbuss\".\"name\" IN (SELECT U0.\"name\" FROM \"paiyrollbuss\" U0 WHERE U0.\"jurisdiction\" = GB))) OR (\"paiyrollbuss\".\"jurisdiction\" = Company AND NOT (\"paiyrollbuss\".\"name\" IN (SELECT V0.\"name\" FROM \"paiyrollbuss\" V0 WHERE (V0.\"jurisdiction\" = GB OR (V0.\"jurisdiction\" = Universal AND NOT (V0.\"name\" IN (SELECT U0.\"name\" FROM \"paiyrollbuss\" U0 WHERE U0.\"jurisdiction\" = GB)))))))) AFAIK, it is legal to add a .filter() to this kind of query, so I think this is a bug. On the mailing list (​), there was a suggestion that using Q() might have helped but I could not see how to use Q() with \"exclude\"."}
{"_id":"q-en-django__django-14395-396f3f5db8990a234882081c0375c482e13e4b4630b16769fdc9ab0aed6323ed","text":"CacheHandler initialize unused caches. Description After the commit: ​ logic of the method \"all\" from CacheHandler class was changed. Before: def all(self): return getattr(, 'caches', {}).values() This method returned connections that were created in getitem Now: def all(self): return [self[alias] for alias in self] Connections return for all \"CACHES\" from (in case of absence - they are forcibly created in self[alias]) Which version of this method seems to be right? In my case this unnecessary mass initialization of custom diskcache-classes leads to io-lags. Snippet that helped me: import def cachegetitem(self, alias, existsonly=False): try: return getattr(self.connections, alias) except AttributeError: if alias not in self.settings: raise self.exceptionclass(f\"The connection '{alias}' doesn't exist.\") if existsonly: return conn = self.createconnection(alias) setattr(self.connections, alias, conn) return conn def cacheall(self): connections = [self.getitem(alias, existsonly=True) for alias in self] return [conn for conn in connections if conn is not None] = cacheall = cachegetitem"}
{"_id":"q-en-django__django-14396-77d2a2ab8c8d0e2bd20e489db6225bbf43635b225b886e212998e14a7a5a5b36","text":"Make Sitemap's protocol default to \"https\". Description I think it is time to change the default of Sitemap.protocol to https. ​ Of course slowly with a clear deprecation timeline. What do you think?"}
{"_id":"q-en-django__django-14399-04329517d5fee1e7de2c87020695028e60fc6478d560584893d35f89a3e67cce","text":"Documentation: hypercorn and static files Description Coming from the age-old problem of service static files, the usual process looks like this: 1) develop and test using runserver and everything just works fine 2) deploy code using WSGI or ASGI as described in the docs 3) find out that static files are missing Specifically referring to ​ As there is a dedicated documentation page for hypercorn, it doesn't look like there's a need for thinking of serving static files. A friend of mine suggested to use whitenoise: ​ Would it make sense to integrate this into the Django docs? To be transparent here, I started also different threads on different channels but it seems like nobody really wants to tackle this issue, so I thought addressing the issue at least via Django sounds reasonable because it's a Web framework: here: ​ and there: ​ from another guy: ​ As of now, I addressed my real-world setup by setting up a \"mini\"-nginx for now, serving static files and proxying hypercorn, but that does not feel like a holistic solution; also when it comes to automated deployment, permissions, principles such as \"test as you fly, fly as you test\" etc. it's a lot more brittle."}
{"_id":"q-en-django__django-14404-87c5eff946a30dc3e93bcfd527cd616c1d4edb79ed7b7f6bf03074e777518469","text":"catchallview() does not support FORCESCRIPTNAME. Description (last modified by SlavaSkvortsov) catchallview returns redirect to '%s\/' % request.path_info (script name cut off there) instead of '%s\/' % (with the script name) Patch - ​"}
{"_id":"q-en-django__django-14407-15be67355f8cc7b97812cfa5a767a20287c19f92defa159583fc5e4c91579f83","text":"Template changes cause dev server to reload Description Django 3.2 has changed the autoreload behavior of the dev server, and it now reloads on template file changes. Reverting to 3.1 fixes the issue. I believe this is related to and ​ Template settings: DEBUG = True TEMPLATES = [ { \"BACKEND\": \"\", \"DIRS\": [(BASEDIR, \"templates\")], \"APPDIRS\": True, \"OPTIONS\": { \"debug\": DEBUG, \"contextprocessors\": [ \"\", \"django.template.contextprocessors.request\", \"\", \"django.contrib.messages.context_processors.messages\", ], }, }, ] Given that it can take several seconds for the dev server to reload, this change can be disruptive to template authoring."}
{"_id":"q-en-django__django-14411-397f9b39c8959b8a6082779dad5ad5e0f5ef5453dd14c51c3b0c6cf2c5b762b4","text":"Label for ReadOnlyPasswordHashWidget points to non-labelable element. Description (last modified by David Sanders) In the admin, the label element for the ReadOnlyPasswordHashWidget widget has a 'for' attribute which points to a non-labelable element, since the widget just renders text, not an input. There's no labelable element for the widget, so the label shouldn't have a 'for' attribute."}
{"_id":"q-en-django__django-14416-92a7958254ba8ddbb0393947e26476818c12dcdda7104d29b4dcd2366095110a","text":"An incorrect language is sometimes displayed - redirects adding a language code to a path can be inappropriate cached by HTTP caches Description We have a multi-lingual site using LocaleMiddleware. With a CDN and caching reverse proxy in use, the following sequence of events can happen: Fetch ​ with Accept-Language header 'de-DE,de;q=0.9'. A redirect to ​ is returned. Fetch it again with Accept-Language header 'fr-FR,fr;q=0.9'. The cached redirect is returned by HTTP caches and the German page is displayed instead of the French one. LocaleMiddleware is issuing a redirect based on the detected language from the Accept-Language header or from the cookie chosen by settings.LANGUAGECOOKIENAME but is not telling caches that it has used these headers. Adding a Vary header fixes this."}
{"_id":"q-en-django__django-14434-6e8090b187529481990edf60e2d4c25507a992c8cc129b5ee165b1df8583f44e","text":"Statement created by createuniquesql makes referencescolumn always false Description This is due to an instance of Table is passed as an argument to Columns when a string is expected."}
{"_id":"q-en-django__django-14441-37e983cd388d3a3394aca86edd6a98a0d186de82e44c214e5777aa4754581121","text":"Prevent getimagedimensions() crash on nonexistent images. Description When using the getimagedimensions(), If a non existing file\/path is passed, the function crashes"}
{"_id":"q-en-django__django-14444-9d9c0e6d6f54d27f13aeb15acaa48131164386eacf6bcaadd57ee0567c34bb36","text":"Make covering and deferrable UniqueConstraint on SQLite a noop. Description Creating a constraint like UniqueConstraint(fields=['name'], name='nameunqcovering', include=['weight']) on SQLite will issue the warning models.W039 stating that a constraint won't be created even though it is."}
{"_id":"q-en-django__django-14447-5d78ffc2676b65b891996a722f273825fb683fa00ad209e775fa643faa99edf4","text":"Database cache counts the DB size twice at a performance penalty Description We have a lot of entries in the DB cache, and I've noticed that the following query shows up in my slow query log kind of a lot (Postgresql is slow at counting things): SELECT COUNT() FROM cachetable; This query is being run by the DB cache twice for every cache update in order to determine if culling is needed. First, in the cache setting code, it runs: cursor.execute(\"SELECT COUNT() FROM %s\" % table) num = cursor.fetchone()[0] now = () now = now.replace(microsecond=0) if num self.maxentries: (db, cursor, now) (​) Then in (the last line above) it runs: cursor.execute(\"DELETE FROM %s WHERE expires < %%s\" % table, [(now)]) cursor.execute(\"SELECT COUNT() FROM %s\" % table) num = cursor.fetchone()[0] if num self.maxentries: # Do culling routine here... (​) The idea is that if the MAXENTRIES setting is exceeded, it'll cull the DB cache down by some percentage so it doesn't grow forever. I think that's fine, but given that the SELECT COUNT() query is slow, I wonder two things: Would a refactor to remove the second query be a good idea? If you pass the count from the first query into the cull method, you can then do: def cull(self, db, cursor, now, count): ... cursor.execute(\"DELETE FROM %s WHERE expires < %%s\" % table, [(now)]) deletedcount = cursor.rowcount num = count - deletedcount if num self.maxentries: # Do culling routine here... That seems like a simple win. Is it reasonable to not run the culling code every time that we set a value? Like, could we run it every tenth time or every 100th time or something? If this is a good idea, does anybody have a proposal for how to count this? I'd be happy just doing it on a mod of the current millisecond, but there's probably a better way (randint?). Would a setting be a good idea here? We already have MAXENTRIES and CULLFREQUENCY. CULLFREQUENCY is \"the fraction of entries that are culled when MAXENTRIES is reached.\" That sounds more like it should have been named CULLRATIO (regrets!), but maybe a new setting for this could be called \"CULLEVERY_X\"? I think the first change is a no-brainer, but both changes seem like wins to me. Happy to implement either or both of these, but wanted buy-in first."}
{"_id":"q-en-django__django-14451-3b6aece1707101b73fa0f924f623bfb5394a9162374c480a7db39b3357a7cd92","text":"adding support for self closing tags in syndication feeds Description the code for syndication feeds was written in a time where self closing tags didn't exist. but today, if an element has no text between the tags, the standard is that it should be a self closing tag. python this functionality in 3.2 to XMLGenerator but django still doesn't use this in generating syndication feeds. this is the matter of passing a single argument to the SimplerXMLGenerator that django uses for generating feeds. since SimplerXMLGenerator directly inherits from pythons XMLGenrator, passing the arguments works with this class too . this is my attempt at making it right ​"}
{"_id":"q-en-django__django-14453-ab6effc2958a8ff50ca0c21412439948aa0b563ffd40215e06cc4a53de0431bb","text":"Added message when user mispells 'urlpatterns' in some 'urls' module Description I found this kind of error when I mispelled urlspattern instead of urlpatterns inside my file. So the console was throwing an error, but this error do not helped me to found the problem. Check it: The included URLconf '<module '' from ''' does not appear to have any patterns in it. If you see valid patterns in the file then the issue is probably caused by a circular import. The problem is not with a circular import, but with the mispelled urlpatterns variable itself, so I'm doing this ticket. OBS.: I have already created a pull request for this: ​ I appreciate any feedback. Thanks, Igor"}
{"_id":"q-en-django__django-14471-885d5116a1a765cabe9ac91e7d8706ad74147eeb411da21e81b89d2291fac822","text":"Reject requests earlier if the CSRF cookie token has the wrong format Description (last modified by Chris Jerdonek) (This issue is similar to but for the cookie token rather than for the non-cookie token.) I noticed in CsrfViewMiddleware.processview() that if the CSRF cookie has the wrong format (i.e. wrong length or contains invalid characters), then the code will do a fair amount of unnecessary work. Specifically, the code will proceed inside gettoken() at ​this line to use Python's secrets module twice to generate both a new token and a mask for the token. But this new token will only be used for the purposes of later calling comparemaskedtokens() in a way that will be guaranteed to fail (since the cookie being used will be brand new and so won't match). And then it will call comparemaskedtokens() with that value. Instead, if the CSRF cookie is found at that line to have the wrong format, the middleware could reject the request outright similar to how does it if the token has the wrong format (as well as similar to how the code currently handles a missing cookie in the ​lines after). I think this will simplify CsrfViewMiddleware and make it easier to understand because it will eliminate a number of steps that aren't needed for security. In particular, one thing this will do is cut down on the number of places where getnewcsrftoken() is called, which will make it clearer where a new value is really needed \/ used. Similar to , it will also make troubleshooting easier because the rejection messages will be more specific. I think this could be implemented as follows. After is merged, ​gettoken() could be changed to allow InvalidTokenFormat to bubble up instead of handling it. Then the InvalidTokenFormat exception could be handled differently in the two places gettoken() is called: (1) In processrequest(), it could be handled by calling getnewcsrftoken() (gettoken()'s current behavior). (2) In process_view(), it could be handled similar to how handles it. Namely, reject the request using the InvalidTokenFormat's reason string."}
{"_id":"q-en-django__django-14493-e42572fd35cc08cfc497b34368a4f38c16c659f73d84982b92cca2548abc0680","text":"ManifestStaticFilesStorage crashes with maxpostprocesspasses = 0. Description To reproduce: Derive a custom class from ManifestStaticFilesStorage and set maxpostprocesspasses to 0: class MyManifestStaticFilesStorage(ManifestStaticFilesStorage): maxpostprocesspasses = 0 STATICFILESSTORAGE = \"MyManifestStaticFilesStorage\" run collectstatic File \"lib\/python3.7\/site-\", line 188, in handle collected = self.collect() File \"lib\/python3.7\/site-\", line 128, in collect for originalpath, processedpath, processed in processor: File \"lib\/python3.7\/site-\", line 403, in postprocess yield from super().postprocess(args, *kwargs) File \"lib\/python3.7\/site-\", line 251, in postprocess if substitutions: UnboundLocalError: local variable 'substitutions' referenced before assignment The error can also be seen easily in the code: ​ subtitutions is only set if the loop is entered at least once. (The motivation to set maxpostprocesspasses to 0 is to have Django not produce invalid CSS as described here: )"}
{"_id":"q-en-django__django-14495-dd4ce23a950764119775f99976d001f89d646c2ec518d048333ac26810c1678c","text":"KeyError when trying to migrate backward to a replaced migration Description Django exhibits some internal confusion regarding whether replaced migrations exist or not. Consider this simple app with two migrations and squashed migration replacing both: $ ls testproj\/migrations\/ When it comes to disambiguating input, Django seems to believe that the replaced migrations still need to be considered: $ migrate testproj 0001 CommandError: More than one migration matches '0001' in app 'testproj'. Please be more specific. But if you actually try to disambiguate and specify one of the replaced migrations, Django no longer thinks it exists (and isn't very graceful about telling you so): $ migrate testproj 0001initial Traceback (most recent call last): File \"\", line 10, in <moduleexecutefromcommandline() File \"\", line 330, in executefromcommandline utility.execute() File \"\", line 322, in execute self.fetchcommand(subcommand).runfromargv() File \"\", line 347, in runfromargv self.execute(args, cmdoptions) File \"\", line 398, in execute output = (args, options) File \"\", line 135, in handle plan = executor.migrationplan(targets) File \"\", line 50, in migrationplan [target].children KeyError: ('testproj', '0001initial') There could be several different approaches to fixing this, but my feeling is that Django shouldn't prevent you from migrating to a replaced migration. If a migration still exists on disk, even if it's been squashed and you've fully migrated the squashed set, you should be able to migrate back to a state within the squashed set. It seems like there might be production rollback cases where that could be important, and I don't see in principle why it shouldn't be possible. If that turns out to be impractical, then I think Django oughtn't bother you about resolving ambiguities with migration names it won't let you migrate to anyway. And the \"nonexistent\" error for this case should be nicer than a raw KeyError. (In Django 1.7 the error was \"ValueError: Node ('testproj17', '0001_initial') not a valid node\", which is perhaps a bit better, but not much.)"}
{"_id":"q-en-django__django-14500-459880e0960275d9cd5321308672c38698d75b8ae332c6af787e1b60b09572a3","text":"Squashed migration is not marked as unapplied Description (last modified by Markus Holtermann) When unapplying a squashed migration and the replaced migration files are still around, the MigrationExecutor mark the squash migration as unapplied, too, not only the replaced migrations."}
{"_id":"q-en-django__django-14508-e5d20aaad9c5ca12480e1d49ff547776635cd07ab3e9b41618a05121d6ca4b0e","text":"Overwriting a property with field during model inheritance. Description Documentation says (in ​ paragraph) that: This restriction only applies to attributes which are Field instances. Normal Python attributes can be overridden if you wish. It also only applies to the name of the attribute as Python sees it: if you are manually specifying the database column name, you can have the same column name appearing in both a child and an ancestor model for multi-table inheritance (they are columns in two different database tables). However.. I came up today with setup like this: 1 from import models 2 3 # Create your models here. 4 5 class SomeTestModel(models.Model): 6 somefield = models.CharField(maxlength=100) 7 8 class Meta: 9 abstract = True 10 11 12 def otherfield(self): 13 return \"[OTHER] %s\" % self.somefield 14 15 16 17 class OtherModel(SomeTestModel): 18 otherfield = models.CharField(maxlength=100) 19 20 21 class AndMoreOther(SomeTestModel): 22 notimportantfield = models.CharField(max_length=100) And then if you do: from import * o = OtherModel() Traceback (most recent call last): File \"<console\", line 1, in <moduleFile \"\", line 357, in init setattr(self, field.attname, val) AttributeError: can't set attribute Since my models where a lot bigger and more complicate, it took me almost all day to figure out that the problem was a from a base model, and my suggestion is that there should be at least a warning somewhere (during model's init perhaps) that could be more precise about why attribute couldn't been set. (or attribute to which object (either Model or Field). I tried it on 1.2 and 1.4 pre-alpha SVN- To reproduce you just need to put the from above in some app."}
{"_id":"q-en-django__django-14513-63973e9173dc8de62819952c4556ade566b9c1cbbd6439e3f597b3684242a774","text":"Better Indication of Squash Migration State in showmigrations Description In the discussion of (​) it became clear that there was a disconnect between the current output of showmigrations and the actual recorded applied state of squashed migrations. Currently if all of the replaced\/original migrations have been run, showmigrations will output that the related squashed migration has been applied with an [X] in the output even if that has not yet been recorded by the migration recorder. However, it is currently a requirement that migrate be run to record this applied state for the squashed migration before the original migrations are removed. If a deployment process is looking for an empty [ ] to know to run the migration then this may trip it up. This case is to consider an output for showmigrations which can indicate that this migration has only been \"soft\" applied, that is applied but not recorded yet. Changes to the planner for such an output may also impact ."}
{"_id":"q-en-django__django-14518-8bd7d6ba57a2925ea09798017fa7f0bc80001673f0240ae37dddffe5ea17472c","text":"Include in CsrfViewMiddleware's bad CSRF token message where the token is from Description (last modified by Chris Jerdonek) Currently, if CsrfViewMiddleware encounters a bad CSRF token, it will reject the request with a message like-- \"CSRF token incorrect\" \"CSRF token has incorrect length\" I noticed that it would be relatively easy to include in these messages whether the token was obtained from POST data or a custom header, which would be useful for troubleshooting. The messages are specified ​here in the code. The new messages could look e.g. like-- \"CSRF token (from POST) incorrect\" \"CSRF token (from 'X-CSRFToken' header) has incorrect length\" The changes to CsrfViewMiddlewareTestMixin proposed in would make these cases easy to test."}
{"_id":"q-en-django__django-14534-cfc2d4d0ea26283493b1f76e358b5a151a987e59ff689474a6320ca54a8f4551","text":"BoundWidget.idforlabel ignores id set by ChoiceWidget.options Description If you look at the implementation of BoundField.subwidgets class BoundField: ... def subwidgets(self): id = ('id') or attrs = {'id': id} if id else {} attrs = self.buildwidgetattrs(attrs) return [ BoundWidget(, widget, ) for widget in (self.htmlname, (), attrs=attrs) ] one sees that (self.htmlname, (), attrs=attrs) returns a dict and assigns it to widget. Now widget['attrs']['id'] contains the \"id\" we would like to use when rendering the label of our CheckboxSelectMultiple. However BoundWidget.idforlabel() is implemented as class BoundWidget: ... def idforlabel(self): return 'id%s%s' % (['name'], ['index']) ignoring the id available through ['attrs']['id']. This re-implementation for rendering the \"id\" is confusing and presumably not intended. Nobody has probably realized that so far, because rarely the autoid-argument is overridden when initializing a form. If however we do, one would assume that the method BoundWidget.idforlabel renders that string as specified through the autoid format-string. By changing the code from above to class BoundWidget: ... def idforlabel(self): return ['attrs']['id'] that function behaves as expected. Please note that this error only occurs when rendering the subwidgets of a widget of type CheckboxSelectMultiple. This has nothing to do with the method BoundField.idforlabel()."}
{"_id":"q-en-django__django-14539-59e61eceb3abe7433c195868057d41023095f7bef80bc6cfe87e8297b78a2183","text":"urlize() does not handle html escaped string and trailing punctuation correctly Description Example: urlize('Search for ! and see.') 'Search for <a href=\"\"<\/a! and see.' 'Search for <a href=\"\"<\/alt! and see.'"}
{"_id":"q-en-django__django-14558-668632a0d99c1cf7d14c15a961c5f650dd63145c9b582a61b7dac609745f3cb1","text":"JSONField.bounddata raises TypeError when data is None Description Hi, When a bound form with no value provided for the JSONField is rendered, it will currently crash. class JSONForm(Form): jsonfield = JSONField(required=False) form = JSONForm({}) assert form.as_p() raises TypeError: the JSON object must be str, bytes or bytearray, not NoneType A fix has been created already by here: ​"}
{"_id":"q-en-django__django-14559-8f8f58430cbd3302e55b667f22233cef7ed75b0156e71f287843747baf4e09fc","text":"Include number of rows matched in bulkupdate() return value Description Currently, bulkupdate() returns None, unlike update(), which returns ​the number of rows matched. It looks like it would be easy to add the same functionality to bulkupdate() since bulkupdate() simply calls update() repeatedly: ​ I.e. the return values could simply be and returned."}
{"_id":"q-en-django__django-14580-ca8dfebc0977e0d78128c5ffc773deee159b60c1259952cb5ba9f28526dc940b","text":"Missing import statement in generated migration (NameError: name 'models' is not defined) Description I found a bug in Django's latest release: 3.2.4. Given the following contents of from import models class MyField(models.TextField): pass class MyBaseModel(models.Model): class Meta: abstract = True class MyMixin: pass class MyModel(MyMixin, MyBaseModel): name = MyField(primarykey=True) The makemigrations command will generate the following migration file: import from import migrations class Migration(migrations.Migration): initial = True dependencies = [ ] operations = [ migrations.CreateModel( name='MyModel', fields=[ ('name', (primarykey=True, serialize=False)), ], options={ 'abstract': False, }, bases=(, models.Model), ), ] Which will then fail with the following error: File \"\", line 7, in <moduleclass Migration(migrations.Migration): File \"\", line 23, in Migration bases=(, models.Model), NameError: name 'models' is not defined Expected behavior: Django generates a migration file that is valid Python. Actual behavior: Django generates a migration file that is missing an import statement. I think this is a bug of the module , but I'm not sure. I will be happy to assist with debugging. Thanks for your attention, Jaap Joris"}
{"_id":"q-en-django__django-14584-d69448a082068cf54781b33373a786fa0a925c9fb05c4fdf04a72856016edf1a","text":"logging output should include the database alias Description (last modified by David Winterbottom) As this is essential information when working with database routing. PR: ​"}
{"_id":"q-en-django__django-14599-e13ad02f05a6d0c8be53a67847f3f9506de45d0544996d4e515dcb997ecc72bc","text":"CsrfViewMiddleware.processresponse()'s csrfcookieneedsreset and csrfcookieset logic isn't right Description I noticed that the csrfcookieneedsreset and csrfcookieset logic inside CsrfViewMiddleware.processresponse() isn't right: ​ Consequently-- self.settoken(request, response) can get called twice in some circumstances, even if response.csrfcookieset is true at the beginning, and the cookie can fail to be reset in some circumstances, even if csrfcookieneeds_reset is true at the beginning. (I previously let know about this issue, and they said it was okay to resolve this publicly.)"}
{"_id":"q-en-django__django-14602-723acbf8b8f5d25ff0b2a292b5a2402060aafeb95deb36d7cadc30a14e00be93","text":"Tighten up the regular expression used by parsetime to accept less 'invalid' options. Description As per discussion in the ticket and on the Github comments for same, currently the timere allows for some variations which it arguably shouldn't. For the historical record's sake, the current regex is: (?P<hour\\d{1,2}):(?P<minute\\d{1,2})(?::(?P<second\\d{1,2})(?:\\d{0,6})?)? where you can see a whole lot of it ends up optional, and there are some ways in which that can be made to accept what we'd probably call 'invalid' (though strictly speaking the result is correct for the input portions): from import parsetime parsetime('0:5: ') (0, 5) If possible, we should derive examples of which strings might current pass and decide which, if any of them, shouldn't be accepted. It's probably also fine to leave the whole thing as-is (be liberal in what you accept etc) and just add them as necessary to the examples of valid inputs, so in future it doesn't come up again beyond \"thats just an accepted quirk\""}
{"_id":"q-en-django__django-14608-0417a8ef8012fe8ef9e5a17fc63880407002883be80837136aae383cc76c660e","text":"Add CSS class for non form errors in FormSets Description (last modified by Ties Jan Hefting) Forms add the nonfield CSS class for non field errors in ErrorList instances. This is documented in a section on ​rendering form error messages. Similarly, in FormSets I'd expect to see the nonform CSS class for non form errors. This would allow a custom ErrorList to make a distinction in form field errors, non field errors (forms) and non form errors (FormSets) when rendering error messages. Therefore I'd suggest to add this nonform CSS class and document it for developers to use."}
{"_id":"q-en-django__django-14631-3413341d979195806991abe7614482ee3902692d1a3fd2ff82bcd06f2a9c4696","text":"BaseForm's cleanfields() and changeddata should access values via BoundField Description (last modified by Chris Jerdonek) While working on , I noticed that ​BaseForm.cleanfields() and ​BaseForm.changeddata don't currently access their values through a BoundField object. It would be better for consistency if they did, and to reduce the number of code paths. One consequence of the current code is that form.cleanfields() can return a different value from form[name].initial when they should be the same. This case is almost, but not quite, covered by ​testdatetimecleaninitialcallabledisabled() (the test can be adjusted to cover this case). As part of this ticket and in line with accessing data through the BoundField objects, I noticed that the code would also be simpler if the per-field logic of changeddata() were moved into a method of the BoundField class. It could be called something like bf.didchange(). This would be more appropriate because whether form data changed for a field is a property of its BoundField (as it depends on the underlying form data), as opposed to the unbound field. With this change, the method could change from its current ~20 lines to something like this-- def changeddata(self): return [name for name, bf in self.bounditems() if bf.didchange()] A similar change could be made to BaseForm.clean_fields()."}
{"_id":"q-en-django__django-14634-247dcefbb7583cf5bb9fb6b009944198cffbdd61a0c9f4b4130b64bdb4ded824","text":"Allow delete to provide a success message through a mixin. Description Add a mixin to show a message on successful object deletion."}
{"_id":"q-en-django__django-14641-59af35186d7a0ee3871e211652e29397d14aaec567dd0248609d8ea5c96cf62b","text":"Move special-case logic in BoundField.initial() to BaseForm.getinitialforfield() Description This is another follow-up to ticket . Currently, BoundField.initial() has logic to special-case time and datetime objects: ​ I noticed that this logic can be moved to BaseForm.getinitialforfield(), and in particular under the if callable(value) block: ​ Eventually, I think it could make sense to go further and move some of this logic to a new method of the Field class, which could permit the special-casing to be handled by overriding in sub-classes that use times and datetimes."}
{"_id":"q-en-django__django-14645-214fc8242c6c962b8b26c43a0fa2a0b9617839f0755ac9dd6dc07eb1a0f64fa2","text":"decimal.InvalidOperation error on DecimalField form field Description I have a form with DecimalField and maxvalue argument: class OrderForm(forms.ModelForm): sum = DecimalField(maxvalue=12) class Meta: model = Order fields = ['sum'] class Order(models.Model): sum = models.DecimalField( 'Sum', maxdigits=18, decimalplaces=2, default=0 ) If I pass \"NaN\" value to this form it will fail with decimal.InvalidOperation error. ​"}
{"_id":"q-en-django__django-14664-7406ececf5b55bfd032aa769ec351db8141fae870b4a47200603db3c20a8de15","text":"ResolverMatch with some views is not pickleable Description given something like the following: def myfbv(request): return HttpResponse('yay') urlpatterns = [ url('whatever', myfbv, name='myfbv'), ] It is possible to do the following: from import resolve from pickle import dumps, loads loads(dumps(resolve('whatever'))) and end up with a working ResolverMatch However, given a Class Based View (mapped to a urlconf via ()), or something like , you get something like the following: dumps(resolve('\/admin\/myapp\/')) [...] PicklingError: Can't pickle <function appindex at 0x109f05de8: it's not found as PicklingError: Can't pickle <function Homepage at 0x109f16b90: it's not the same object as Both of which are raised by pickle's saveglobal(self, obj, name, pack) which recognises that it's a module+name combo (thus should be in scope) but isn't the same object in identity (if x is not y) Ordinarily, this is not a problem, but resolvermatch is set onto a request, and I'm using the with multiprocessing, which requires the ability to pickle data to send back and forth, and evidently somewhere within the TemplateResponses I'm dealing with, the request is being serialised (if I had to guess -- probably in the context) and is sometimes failing (depending on the type of view mounted) Ideally, every ResolverMatch should be serialisable, or none should be, instead of the current situation where the project's urlconf may denote success or failure."}
{"_id":"q-en-django__django-14667-95dc861b484526cfcc25dcfe3e10a17da984887915a6b575484fa3ebad18e1d0","text":"() doesn't clear deferred field when chaining with only(). Description Considering a simple Company model with four fields: id, name, tradenumber and country. If we evaluate a queryset containing a .defer() following a .only(), the generated sql query selects unexpected fields. For example: (\"name\").defer(\"name\") loads all the fields with the following query: SELECT \"company\".\"id\", \"company\".\"name\", \"company\".\"tradenumber\", \"company\".\"country\" FROM \"company\" and (\"name\").defer(\"name\").defer(\"country\") also loads all the fields with the same query: SELECT \"company\".\"id\", \"company\".\"name\", \"company\".\"trade_number\", \"company\".\"country\" FROM \"company\" In those two cases, i would expect the sql query to be: SELECT \"company\".\"id\" FROM \"company\" In the following example, we get the expected behavior: (\"name\", \"country\").defer(\"name\") only loads \"id\" and \"country\" fields with the following query: SELECT \"company\".\"id\", \"company\".\"country\" FROM \"company\""}
{"_id":"q-en-django__django-14672-c90669b684330d9ac98e80d061863550240d0805c119d014104c13fbad3c0d24","text":"Missing call on in Description In 3.2 identity property has been to all ForeignObjectRel to make it possible to compare them. A hash is derived from said identity and it's possible because identity is a tuple. To make limitchoicesto hashable (one of this tuple elements), ​there's a call to makehashable. It happens that throughfields can be a list. In such case, this makehashable call is missing in ​ManyToManyRel. For some reason it only fails on checking proxy model. I think proxy models have 29 checks and normal ones 24, hence the issue, but that's just a guess. Minimal repro: class Parent(models.Model): name = models.CharField(maxlength=256) class ProxyParent(Parent): class Meta: proxy = True class Child(models.Model): parent = models.ForeignKey(Parent, ondelete=models.CASCADE) manytomanyfield = models.ManyToManyField( to=Parent, through=\"ManyToManyModel\", throughfields=['child', 'parent'], relatedname=\"something\" ) class ManyToManyModel(models.Model): parent = models.ForeignKey(Parent, ondelete=models.CASCADE, relatedname='+') child = models.ForeignKey(Child, ondelete=models.CASCADE, relatedname='+') secondchild = models.ForeignKey(Child, ondelete=models.CASCADE, null=True, default=None) Which will result in File \"\", line 23, in <modulemain() File \"\", line 19, in main executefromcommandline() File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 419, in executefromcommandline utility.execute() File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 413, in execute self.fetchcommand(subcommand).runfromargv() File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 354, in runfromargv self.execute(args, cmdoptions) File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 393, in execute () File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 419, in check allissues = checks.runchecks( File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 76, in runchecks newerrors = check(appconfigs=appconfigs, databases=databases) File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 34, in checkallmodels ((kwargs)) File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 1277, in check cls.checkfieldnameclashes(), File \"\/home\/tom\/PycharmProjects\/djangbrokenm2mprojectProject\/venv\/lib\/python3.8\/site-\", line 1465, in checkfieldnameclashes if f not in usedfields: File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 140, in hash return hash(self.identity) TypeError: unhashable type: 'list' Solution: Add missing makehashable call on self.throughfields in ManyToManyRel. Missing call on in Description In 3.2 identity property has been to all ForeignObjectRel to make it possible to compare them. A hash is derived from said identity and it's possible because identity is a tuple. To make limitchoicesto hashable (one of this tuple elements), ​there's a call to makehashable. It happens that throughfields can be a list. In such case, this makehashable call is missing in ​ManyToManyRel. For some reason it only fails on checking proxy model. I think proxy models have 29 checks and normal ones 24, hence the issue, but that's just a guess. Minimal repro: class Parent(models.Model): name = models.CharField(maxlength=256) class ProxyParent(Parent): class Meta: proxy = True class Child(models.Model): parent = models.ForeignKey(Parent, ondelete=models.CASCADE) manytomanyfield = models.ManyToManyField( to=Parent, through=\"ManyToManyModel\", throughfields=['child', 'parent'], relatedname=\"something\" ) class ManyToManyModel(models.Model): parent = models.ForeignKey(Parent, ondelete=models.CASCADE, relatedname='+') child = models.ForeignKey(Child, ondelete=models.CASCADE, relatedname='+') secondchild = models.ForeignKey(Child, ondelete=models.CASCADE, null=True, default=None) Which will result in File \"\", line 23, in <modulemain() File \"\", line 19, in main executefromcommandline() File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 419, in executefromcommandline utility.execute() File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 413, in execute self.fetchcommand(subcommand).runfromargv() File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 354, in runfromargv self.execute(args, cmdoptions) File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 393, in execute () File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 419, in check allissues = checks.runchecks( File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 76, in runchecks newerrors = check(appconfigs=appconfigs, databases=databases) File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 34, in checkallmodels ((kwargs)) File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 1277, in check cls.checkfieldnameclashes(), File \"\/home\/tom\/PycharmProjects\/djangbrokenm2mprojectProject\/venv\/lib\/python3.8\/site-\", line 1465, in checkfieldnameclashes if f not in usedfields: File \"\/home\/tom\/PycharmProjects\/brokenm2mproject\/venv\/lib\/python3.8\/site-\", line 140, in hash return hash(self.identity) TypeError: unhashable type: 'list' Solution: Add missing makehashable call on self.throughfields in ManyToManyRel."}
{"_id":"q-en-django__django-14681-3a542b5fb31f16c0fc34587f004cd0b92062a367bbba62d57f60d1b9388d5c19","text":"CSRF failure incorrectly reported on upload when there is a problem with storage Description Minimal reproduction app is in the attachment, although the only changes from the default template are: MEDIAURL = '\/media\/' MEDIAROOT = 'media\/' FILEUPLOADMAXMEMORYSIZE = 1024 * 1024 FILEUPLOADTEMPDIR = MEDIAROOT + 'tmp' class File(models.Model): file = models.FileField() from .models import File (File) Required setup for the attached app: python migrate python createsuperuser Steps to reproduce 1) runserver 2) navigate and login to \/admin\/ 3) navigate to \/admin\/app\/file\/add\/ Scenario 1. default state - file uploads works as expected Scenario 2. remove media\/tmp directory - file uploads works only for files that fit in FILEUPLOADMAXMEMORYSIZE, error otherwise (see below) Scenario 3. remove whole media directory - error reported for all file uploads (see below) Exact error message: Forbidden (403) CSRF verification failed. Request aborted. Reason given for failure: CSRF token missing or incorrect. Expected behaviour: Filesystem error or similar reporting incorrect media storage setup. Comment: Yes, the setup in this scenario is invalid to begin with, but the error message has nothing to do with the actual problem. I suspect a real problem with an underlying filesystem will also get covered in the same way."}
{"_id":"q-en-django__django-14717-9b95a68be09e118e635718ad3718b89b468773bf26f59525f25181dc90295283","text":"Time-related checkfixdefaultvalue() methods can be optimized \/ simplified and have a bug Description I noticed that three of the checkfixdefaultvalue() method definitions in can be simplified. Here is one of them: ​ For example, in each of them, () is called even when the return value isn't needed \/ won't be used."}
{"_id":"q-en-django__django-14722-cd27a52f3ee0ccda1cedab4a4c6ed1f75689ddde07389ba2fede1d294de1ce90","text":"Moving a unique constraint from uniquetogether to generate an invalid migration. Description You can see a demo example to show the bug at . I met a problem when I convert a uniquetogether to the unique=True attribute. first commit, everything is ok I create a model file. \/\/ class MyModel(models.Model): name = models.CharField(maxlength=32) class Meta: uniquetogether = (\"name\", ) the migrations file looks like this. \/\/ from import migrations, models class Migration(migrations.Migration): dependencies = [ ('testapp', '0001initial'), ] operations = [ migrations.AlterField( modelname='mymodel', name='name', field=models.CharField(maxlength=32, unique=True), ), migrations.AlterUniqueTogether( name='mymodel', uniquetogether=set(), ), ] second commit: then I remove the uniquetogether and add unique=True to the field model file class MyModel(models.Model): name = models.CharField(maxlength=32, unique=True) class Meta: pass # uniquetogether = (\"name\", ) 0002 migrations file class Migration(migrations.Migration): dependencies = [ ('testapp', '0001initial'), ] operations = [ migrations.AlterField( modelname='mymodel', name='name', field=models.CharField(maxlength=32, unique=True), ), migrations.AlterUniqueTogether( name='mymodel', uniquetogether=set(), ), ] However, when I apply the migrations, an error occurs; wangx python3 migrate Operations to perform: Apply all migrations: admin, auth, contenttypes, sessions, testapp Running migrations: Applying contenttypes.0001initial... OK Applying auth.0001initial... OK Applying admin.0001initial... OK Applying admin.0002logentryremoveautoadd... OK Applying admin.0003logentryaddactionflagchoices... OK Applying contenttypes.0002removecontenttypename... OK Applying auth.0002alterpermissionnamemaxlength... OK Applying auth.0003alteruseremailmaxlength... OK Applying auth.0004alteruserusernameopts... OK Applying auth.0005alteruserlastloginnull... OK Applying auth.0006requirecontenttypes0002... OK Applying auth.0007altervalidatorsadderrormessages... OK Applying auth.0008alteruserusernamemaxlength... OK Applying auth.0009alteruserlastnamemaxlength... OK Applying auth.0010altergroupnamemaxlength... OK Applying auth.0011updateproxypermissions... OK Applying sessions.0001initial... OK Applying testapp.0001initial... OK Applying testapp.0002auto202004221247...Traceback (most recent call last): File \"\/usr\/local\/lib\/python3.6\/dist-\", line 86, in execute return (sql, params) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 74, in execute return (query, args) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 209, in execute res = (query) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 315, in query (q) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 239, in query (self, query) MySQLdb.exceptions.OperationalError: (1061, \"Duplicate key name 'testappmymodelnameba5e2bd2uniq'\") The above exception was the direct cause of the following exception: Traceback (most recent call last): File \"\", line 21, in <modulemain() File \"\", line 17, in main executefromcommandline() File \"\/usr\/local\/lib\/python3.6\/dist-\", line 401, in executefromcommandline utility.execute() File \"\/usr\/local\/lib\/python3.6\/dist-\", line 395, in execute self.fetchcommand(subcommand).runfromargv() File \"\/usr\/local\/lib\/python3.6\/dist-\", line 328, in runfromargv self.execute(args, cmdoptions) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 369, in execute output = (args, options) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 83, in wrapped res = handlefunc(args, *kwargs) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 233, in handle fakeinitial=fakeinitial, File \"\/usr\/local\/lib\/python3.6\/dist-\", line 117, in migrate state = self.migrateallforwards(state, plan, fullplan, fake=fake, fakeinitial=fakeinitial) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 147, in migrateallforwards state = self.applymigration(state, migration, fake=fake, fakeinitial=fakeinitial) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 245, in applymigration state = (state, schemaeditor) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 124, in apply operation.databaseforwards(self.applabel, schemaeditor, oldstate, projectstate) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 249, in databaseforwards schemaeditor.alterfield(frommodel, fromfield, tofield) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 565, in alterfield olddbparams, newdbparams, strict) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 745, in alterfield self.execute(self.createuniquesql(model, [])) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 142, in execute cursor.execute(sql, params) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 100, in execute return super().execute(sql, params) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 68, in execute return self.executewithwrappers(sql, params, many=False, executor=self.execute) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 77, in executewithwrappers return executor(sql, params, many, context) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 86, in execute return (sql, params) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 90, in exit raise djexcvalue.withtraceback(traceback) from excvalue File \"\/usr\/local\/lib\/python3.6\/dist-\", line 86, in execute return (sql, params) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 74, in execute return (query, args) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 209, in execute res = (query) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 315, in query (q) File \"\/usr\/local\/lib\/python3.6\/dist-\", line 239, in query (self, query) (1061, \"Duplicate key name 'testappmymodelnameba5e2bd2uniq'\") I check the sql for these migrations, it shows: wangx python3 sqlmigrate testapp 0001 -- Create model MyModel CREATE TABLE ( integer AUTOINCREMENT NOT NULL PRIMARY KEY, varchar(32) NOT NULL); ALTER TABLE ADD CONSTRAINT UNIQUE (); wangx python3 sqlmigrate testapp 0002 -- Alter field name on mymodel ALTER TABLE ADD CONSTRAINT UNIQUE (); -- Alter uniquetogether for mymodel (0 constraint(s)) ALTER TABLE DROP INDEX ; it looks like django will first create the index for unique=True second drop the index for uniquetogether=('name', ) but the program for creating index name generates the same index name :testappmymodelnameba5e2bd2uniq, so when django create the same index, Duplicate key name error occurs."}
{"_id":"q-en-django__django-14725-bc5e1351c20fcefca3eba572b76a713ba9b7a42d17a3a7586b073e6494298e17","text":"Provide a way for model formsets to disallow new object creation Description Model formsets don't provide a way to create an \"edit only\" view of objects. We see users trying to use extra=0 to accomplish this, but that's not reliable as extra is merely meant for the extra number of forms to display. You can add more forms with Javascript (or just send additional post data)."}
{"_id":"q-en-django__django-14727-c4cf73217e437cdf7f2939e050bca5edd6c47605abb8b98f33f745c5de5c60c3","text":"Naming an incompletely applied squashed migration as a migration target fails with bare NodeNotFoundError Description In Line 205-208 in replacement migrations (created with squash) are checked if they can be applied. If any of the to be replaced migrations isn't already applied the replacement migration is not to the nodes list. This leads to the fact that if some of the migrations are removed or not completely applied before the squash is and there is a dependency on the replacement migration, the user gets a 'NodeNotFoundError' where the replacement migration that is not being applied because of line 206 is the missing one. This is very confusing to the user, raising a warning in line 208 would inform the user that the squashed migration can not be applied because not all the 'child' migrations are applied. Had to debug into that to figure that out."}
{"_id":"q-en-django__django-14730-22c0f236bc07ff550189c5297e3b6a82e704fa4571c00d805715cbcf1aebfaf7","text":"Prevent developers from defining a relatedname on symmetrical ManyToManyFields Description In ManyToManyField, if the symmetrical argument is passed, or if it's a self-referential ManyToMany relationship, the related field on the target model is not created. However, if a developer passes in the relatedname not understanding this fact, they may be confused until they find the information about symmetrical relationship. Thus, it is proposed to raise an error when the user defines a ManyToManyField in this condition."}
{"_id":"q-en-django__django-14733-df1fb192b49fdea70a2a149bc204456929d2060e7762e37992f6b3880554cf63","text":"Allow overriding of deletion widget in formsets Description In Django 3.0 orderingwidget and getorderingwidget() were introduced (see ). The typical use case was to easily override the ORDER field in formsets that are updated in the frontend. For the exact same use case, I'd find it useful to see deletionwidget and getdeletionwidget() getting introduced. Discussion ​initiated here for this feature."}
{"_id":"q-en-django__django-14751-e7df196e2418f86c0209867b2752d5ef2ee5144e6c667be3c0f0ef140818357f","text":"Make makemigrations scriptable \/ script-friendly Description Currently, the makemigrations management command doesn't lend itself well to scripting. For example, it writes its progress output to stdout rather than stderr. Also, there doesn't appear to be a structured \/ programmatic way to figure out what files it has created. My use case is that in my development environment, I'd like to be able to run makemigrations in a Docker container, find out what files were (e.g. from makemigrations's output), and then copy those files from the Docker container to my development machine so they can be to source control. Currently, there doesn't seem to be an easy way to do this. One way, for example, is to manually read makemigrations's output to find out what apps were affected, and then inspect the directories yourself for the new files. Better, for example, would be if makemigrations could write the paths to the created files to stdout."}
{"_id":"q-en-django__django-14752-8401197aae1167e5c3f565cae95be99c432cce3698b5819d0576a9c085d7ca2c","text":"Refactor AutocompleteJsonView to support extra fields in autocomplete response Description (last modified by mrts) Adding data attributes to items in ordinary non-autocomplete foreign key fields that use forms.widgets.Select-based widgets is relatively easy. This enables powerful and dynamic admin site customizations where fields from related models are updated immediately when users change the selected item. However, adding new attributes to autocomplete field results currently requires extending and fully overriding the () method. Here's an example: class MyModelAdmin(admin.ModelAdmin): def geturls(self): return [ path('autocomplete\/', (adminsite=self.adminsite)) if ('autocomplete\/') else url for url in super().geturls() ] class CustomAutocompleteJsonView(AutocompleteJsonView): def get(self, request, args, *kwargs): , self.modeladmin, self.sourcefield, tofieldname = self.processrequest(request) if not self.hasperm(request): raise PermissionDenied self.objectlist = self.getqueryset() context = self.getcontextdata() return JsonResponse({ 'results': [ {'id': str(getattr(obj, tofieldname)), 'text': str(obj), 'notes': } # <-- customization here for obj in context['objectlist'] ], 'pagination': {'more': context['pageobj'].hasnext()}, }) The problem with this is that as () keeps evolving, there's quite a lot of maintenance overhead required to catch up. The solutions is simple, side-effect- and risk-free: adding a result customization extension point to get() by moving the lines that construct the results inside JsonResponse constructor to a separate method. So instead of return JsonResponse({ 'results': [ {'id': str(getattr(obj, tofieldname)), 'text': str(obj)} for obj in context['objectlist'] ], 'pagination': {'more': context['pageobj'].hasnext()}, }) there would be return JsonResponse({ 'results': [ self.serializeresult(obj, tofieldname) for obj in context['objectlist'] ], 'pagination': {'more': context['pageobj'].hasnext()}, }) where serializeresult() contains the original object to dictionary conversion code that would be now easy to override: def serializeresult(self, obj, tofieldname): return {'id': str(getattr(obj, tofieldname)), 'text': str(obj)} The example CustomAutocompleteJsonView from above would now become succinct and maintainable: class CustomAutocompleteJsonView(AutocompleteJsonView): def serializeresult(self, obj, tofieldname): return super.serializeresult(obj, tofield_name) | {'notes': } What do you think, is this acceptable? I'm more than happy to provide the patch."}
{"_id":"q-en-django__django-14762-5710d62447388bfcf31b56a2ee53919f1a571dbaa1b49d72e63fee5a14a54915","text":"prefetchrelated() for deleted GenericForeignKey is not consistent. Description prefetchrelated called for GenericForeignKey sets contenttypeid and objectid to None, if the foreign object doesn't exist. This behaviour is not documented. GenericForignKey is often used for audit records, so it can keep links to non-existing objects. Probably prefetchrelated shouldn't touch original values of objectid and contenttypeid and only set contentobject to None. from import User from import GenericForeignKey from import ContentType from import models class TaggedItem(models.Model): tag = models.SlugField() contenttype = models.ForeignKey(ContentType, ondelete=models.CASCADE) objectid = models.PositiveIntegerField() contentobject = GenericForeignKey('contenttype', 'objectid') guido = (username='Guido') t = TaggedItem(contentobject=guido, tag='test') t.save() () tags1 = (tag='test') tags1[0].contentobject # returns None tags1[0].objectid # returns 1 tags1[0].contenttypeid # returns X tags2 = (tag='test').prefetchrelated(\"contentobject\") tags2[0].contentobject # returns None tags2[0].objectid # returns None tags2[0].contenttype_id # returns None"}
{"_id":"q-en-django__django-14765-c917e9cbd60fc0a6377c57aa08d65de120848942058f6a1599e68a76d601236b","text":"() can assume its realapps argument is a set Description ​PR made all calls to () pass realapps as a set. In ​() now, then, instead of checking that real_apps is a set and converting it to a set if not, it can just assert that it's a set when non-None. (Presumably the construction of new ProjectState objects is part of Django's internal API.) I had made this comment on the PR, but it wasn't important enough to hold up the PR because another PR was depending on it getting merged."}
{"_id":"q-en-django__django-14771-021c046a2007caad8a753ef8d3e3d6a8e663d2b099a99fd76dddcce785807041","text":"Auto-reloader should pass -X options (for cpython implementation) Description Windows OS $ winpty python -m django startproject myproject $ cd myproject\/ $ winpty python -m django startapp myapp $ vi # demo for xoptions ... $ cat -n 1 from import AppConfig 2 3 class MyAppConfig(AppConfig): 4 defaultautofield = '' 5 name = 'myapp' 6 7 # myapp global initialdemo ... 8 with open(\"\", mode=\"r\") as stream: 9 print(\"=== %s\" % stream.encoding) $ vi # INSTALLEDAPPS $ winpty python -X utf8 runserver 0.0.0.0:8005 -v3 === UTF-8 === cp936 Watching for file changes with StatReloader Performing system checks... ... ... $ winpty python -X utf8 runserver 0.0.0.0:8005 -v3 --noreload === UTF-8 Performing system checks... ... ... Refer: ​ ​"}
{"_id":"q-en-django__django-14779-c97d630b852e5caf3dd79d58fb8e55272c496a93568f353a8739fc34991839e5","text":"Custom tags with missing context param and no other params throw an unhelpful IndexError Description Given a simpletag or inclusiontag with takescontext=True, where the context param has been forgotten: def simpletagwithoutcontextparameter(arg): return \"Expected result\" the parsebits function checks for this case and throws an informative TemplateSyntaxError. However, in the case that the tag takes no other parameters: def simpletagnoparamswithoutcontextparameter(): return \"Expected result\" the checking code fails at the point where it looks at params[0], throwing an opaque IndexError instead. Traceback (most recent call last): File \"\", line 179, in testsimpletagnoparamsmissingcontext ('{% load custom %}{% simpletagnoparamswithoutcontextparameter %}') File \"\", line 156, in fromstring return Template(templatecode, engine=self) File \"\", line 155, in init self.nodelist = self.compilenodelist() File \"\", line 199, in compilenodelist return () File \"\", line 502, in parse raise (token, e) File \"\", line 500, in parse compiledresult = compilefunc(self, token) File \"\", line 119, in compilefunc args, kwargs = parsebits( File \"\", line 246, in parsebits if params[0] == 'context': IndexError: list index out of range (PR to follow)"}
{"_id":"q-en-django__django-14785-e7e238a833aac0a6afc9bc019982b39951f5d63b25db3194ae3c4d6828b24ba6","text":"NaN can be stored in DecimalField but cannot be retrieved Description (last modified by dennisvang) Description If, for whatever reason, a NaN value (either float('nan'), , or ) is stored in a DecimalField using sqlite3, the object cannot be retrieved from the database. Attempts to do so will raise TypeError: argument must be int or float This issue also breaks e.g. the admin changelist view. Steps to reproduce Create a brand new project using python 3.8.10 and django 3.2.6 with the default sqlite3 backend (optionally with numpy 1.21.2). Create a model with a DecimalField: class MyModel(models.Model): value = models.DecimalField(maxdigits=10, decimalplaces=5) Programmatically create a model instance with value=float('nan') (or , or ), then try to retrieve the object from the database (or refresh from database). obj = (value=float('nan')) obj.refreshfromdb() Visiting the admin change view (or changelist view) for the model will also raise the error. Traceback: Internal Server Error: \/nanbug\/mymodel\/1\/change\/ Traceback (most recent call last): File \"\", line 47, in inner response = getresponse(request) File \"\", line 181, in getresponse response = wrappedcallback(request, callbackargs, callbackkwargs) File \"\", line 616, in wrapper return self.adminsite.adminview(view)(args, kwargs) File \"\", line 130, in wrappedview response = viewfunc(request, args, kwargs) File \"\", line 44, in wrappedviewfunc response = viewfunc(request, args, kwargs) File \"\", line 232, in inner return view(request, args, kwargs) File \"\", line 1660, in changeview return self.changeformview(request, objectid, formurl, extracontext) File \"\", line 43, in wrapper return boundmethod(args, kwargs) File \"\", line 130, in wrappedview response = viewfunc(request, *args, kwargs) File \"\", line 1540, in changeformview return self.changeformview(request, objectid, formurl, extracontext) File \"\", line 1561, in changeformview obj = self.getobject(request, unquote(objectid), tofield) File \"\", line 763, in getobject return ({ objectid}) File \"\", line 431, in get num = len(clone) File \"\", line 262, in len self.fetchall() File \"\", line 1324, in fetchall self.resultcache = list(self.iterableclass(self)) File \"\", line 68, in iter for row in compiler.resultsiter(results): File \"\", line 1122, in applyconverters value = converter(value, expression, connection) File \"\", line 313, in converter return createdecimal(value).quantize(quantizevalue, context=expression.output_field.context) TypeError: argument must be int or float"}
{"_id":"q-en-django__django-14787-8748ecdad547c97b0ac487d3b8171bb09f3010681142068ab0c5a62f223aaccf","text":"methoddecorator() should preserve wrapper assignments Description the function that is passed to the decorator is a partial object and does not have any of the attributes expected from a function i.e. name, module etc... consider the following case def logger(func): def inner(args, kwargs): try: result = func(args, kwargs) except Exception as e: result = str(e) finally: (f\"{} called with args: {args} and kwargs: {kwargs} resulting: {result}\") return inner class Test: def helloworld(self): return \"hello\" Test().testmethod() This results in the following exception AttributeError: 'functools.partial' object has no attribute 'name'"}
{"_id":"q-en-django__django-14792-3b84e1ef5071e20c237a956a02a5a25e2df34ab18d3fa933b947767b63c9480a","text":"Reverse time zone conversion in Trunc()\/Extract() database functions. Description When using a time zone of \"Etc\/GMT-10\" (or similar) for a Trunc class tzinfo, it appears there's a different behavior as of Django 3.2 in the resulting database query. I think it's due to a change in the return value of timezone.gettimezonename() that's called by the TimezoneMixin. On Django 3.1 the TimezoneMixin method gettzname() returns \"+10\" for a \"Etc\/GMT-10\" time zone after calling ​gettimezonename(). This later becomes \"-10\" in the resulting query due to the return value of preparetznamedelta() of the Postgres DatabaseOperations class, i.e. the time zone 10 hours east from UTC. SELECT ... DATETRUNC(\\'day\\', \"mymodel\".\"startat\" AT TIME ZONE \\'-10\\') AS \"date\" ... On Django 3.2 the TimezoneMixin method gettzname() returns \"Etc\/GMT-10\" for a \"Etc\/GMT-10\" time zone after calling ​gettimezonename(). This later, incorrectly, becomes \"Etc\/GMT+10\" in the resulting query due to the return value of preparetznamedelta() of the Postgres DatabaseOperations class, i.e. the time zone 10 hours west from UTC, which is the opposite direction from the behavior in Django 3.1. SELECT ... DATETRUNC(\\'day\\', \"mymodel\".\"startat\" AT TIME ZONE \\'Etc\/GMT+10\\') AS \"date\" ... timezone.gettimezonename(pytz.timezone(\"Etc\/GMT-10\")) '+10' timezone.gettimezone_name(pytz.timezone(\"Etc\/GMT-10\")) 'Etc\/GMT-10' The above is the same when using Python's zoneinfo.ZoneInfo() too."}
{"_id":"q-en-django__django-14805-52c491a0945daeab82000813869efc504b300095b428c947c38add604e664158","text":"Add output to makemigrations and migrate commands with --noinput Description The --noinput option to makemigrations seems to also mean \"no output\". It would be nice for scripting purposes if there could be some kind of output. When writing a script that helps manage migrations, it would be nice to be able to detect that the migrations needed input from the developer so that the script can fail gracefully and inform the developer that manual intervention is required. Right now the options seem to be to not use the --noinput options and have the prompt interrupt the script, or to use the option and have no idea when migrations failed because they required input. I'm not very picky as far as how this functionality should be implemented goes. I would think that a little warning like \"No migrations run because input was required but --noinput option was specified.\" would be sufficient. This could be output to stdout or stderr, either all the time or perhaps only if an additional option was set."}
{"_id":"q-en-django__django-14812-06e892d4e4313f4c6ed6afd99f62df7fa712038c70fe841747dcc05ad07625df","text":"ModelAdmin for proxy model with InlineModelAdmin for proxy superclass reference results in admin.E202 Description (last modified by Lucas Weyne) This is similar to , but in this case, the is a model with references to a proxy superclass Assume the following Django models: class Reporter(models.Model): name = models.CharField(maxlength=50) class Journalist(Reporter): class Meta: proxy = True class SpecialJournalist(Journalist): class Meta: proxy = True class Article(models.Model): journalist = models.ForeignKey(Journalist, ondelete=models.CASCADE) Register model admins as follows (exemplary): class ArticleInline(admin.TabularInline): model = Article fkname = 'journalist' class SpecialJournalistAdmin(admin.ModelAdmin): inlines = [ArticleInline] This will result in the following error: <class 'ArticleInline': (admin.E202) fkname 'journalist' is not a ForeignKey to 'SpecialJournalist'. This problem occurs on this check this check: ​ A ValueError is raised because the result for () does not include Journalist: () [<class 'Reporter']"}
{"_id":"q-en-django__django-14832-81b55880f7cb722ddafc6b2c539853ee263378151b67fb9205b50d91273a5aeb","text":"Make floatformat template filter independent of USEL10N and allow forcing unlocalized format. Description (last modified by Tim Graham) Specifically, when the filter is used within the {% localize [on|off] %} block with the localization setting opposed to the value of USEL10N ('on' when USEL10N = False or 'off' when USEL10N = True), the localization setting has not effect. This is due to the use of formats.numberformat() without its usel10n parameter, by the numberformat template filter (e.g. ​, ​). The value of the use_l10n parameter shall be taken out of the template rendering context. But I do not see any easy solution to this, as filters do not take context..."}
{"_id":"q-en-django__django-14855-23337e251c345abbaa911266385519efacab594b5b316fdea3054398d667143a","text":"Wrong URL generated by getadminurl for readonly field in custom Admin Site Description When a model containing a ForeignKey field is viewed (or edited) in a custom Admin Site, and that ForeignKey field is listed in readonlyfields, the url generated for the link is \/admin\/... instead of \/custom-admin\/.... This appears to be caused by the following line in getadminurl: url = reverse(urlname, args=[quote()]) Other parts of the admin use the currentapp keyword parameter to identify the correct current name of the Admin Site. (See responseadd as just one example) I have been able to correct this specific issue by replacing the above line with: url = reverse( urlname, args=[quote()], ) However, I don't know if there are any side effects and I have not yet run the full suite of tests on this. Mostly looking for feedback whether I'm on the right track."}
{"_id":"q-en-django__django-14861-00548af9e2ad9450179436f3f7ff5c5646d8b869d82a89ff51aeef9106d50cd4","text":"Conditionally changing ModelAdmin inlines based on object's field breaks when changing object and new inlines should appear. Description Minimal example: class Parent(models.Model): showinlines = models.BooleanField(default=False) class Child(models.Model): parent = models.ForeignKey(Parent, ondelete=models.CASCADE) class ChildInline(admin.StackedInline): model = Child class ParentAdmin(admin.ModelAdmin): def getinlines(self, request, obj): if obj is not None and obj.showinlines: return [ChildInline] return [] Create Parent objects in either initial state and it works as you'd expect, where ChildInline is rendered when showinlines is True. When showinlines is True, you can also set it to False from the admin site, and the ChildInline disappears as expected. But when showinlines is False, you cannot re-enable it. Saving the object fails due to a validation error in the new ChildInline that didn't exist before saving: (Hidden field TOTALFORMS) This field is required. (Hidden field INITIALFORMS) This field is required. ManagementForm data is missing or has been tampered with. Missing fields: childset-TOTALFORMS, childset-INITIAL_FORMS. You may need to file a bug report if the issue persists."}
{"_id":"q-en-django__django-14871-319685ba9518424b9848c1baf93722c081a423247401426a8fc7e222bad9f042","text":"Select2 doesn't load translations with subtags. Description For example, when using the setting LANGUAGE_CODE=\"pt-BR\", the translation of select2 is not applied, the static file i18n is not found. This is due to the fact that some languages are converted to lowercase. ​"}
{"_id":"q-en-django__django-14878-588a17bfabafa5ea96eba798f55c70a74de9885462b56543d01fdd921e92a937","text":"Primary key constraints aren't detected on SQLite. Description (last modified by jgr88) While creating models with \"inspectdb\" i discovered, that there are currently some issues with SQLite3 Databases. PrimaryKeys ain't detected properly Datatype double ain't detected properly Datatype unsigned int ain't detected properly Reproduce these issues by creating a SQLite3 Database: CREATE TABLE \"test\" ( pId INTEGER NOT NULL, doubleField DOUBLE NOT NULL, uInt UNSIGNED INTEGER NOT NULL, PRIMARY KEY(pId) ) I a pullrequest: ​"}
{"_id":"q-en-django__django-14880-59953a622ef99d64e2dafbbf8401b31ceafa1280d4a5e2a384fe0c23a30a70cb","text":"Improve error messages for reverse accessor clashes. Description refer: ​ RelatedField.checkclashes() provides feedback when it finds a clash, but fails to mentioned what the clashing name was. This cost me some significant time to track because of inadequate feedback and would have become immediately clear had the feedback listed the clashing name. A proposed patch appears above, but alas this impacts some unit tests as well. Happy to add fixes to those to the patch, but have been requested to file and issue here."}
{"_id":"q-en-django__django-14890-372fdbac0a58877903ef43f13be60626d4567cb59954e0032df9ef0298e8fb0f","text":"\"&\" and \"|\" operators are silently ignored after (), intersection(), and difference(). Description This looks like a similar issue to the one fixed in Example: class MyModel(models.Model): name = models.CharField() for name in ['a', 'b', 'c']: (name=name) group1 = (name='a') group2 = (name='b') group3 = (name='c') combinedgroup = (group2) groupand = combinedgroup & group1 groupor = combinedgroup | group 3 In this example, combinedgroup, groupand and groupor all have the same SQL. These operators should raise an exception if they can not be applied after combinator functions."}
{"_id":"q-en-django__django-14915-d22226556336bb8594a5ec9f5b684bd559616b2cb9cec162eae732658f53cc64","text":"ModelChoiceIteratorValue is not hashable. Description Recently I migrated from Django 3.0 to Django 3.1. In my code, I add custom data-* attributes to the select widget options. After the upgrade some of those options broke. Error is {TypeError}unhashable type: 'ModelChoiceIteratorValue'. Example (this one breaks): def createoption(self, name, value, label, selected, index, subindex=None, attrs=None): context = super().createoption(name, value, label, selected, index, subindex, attrs) if not value: return context if value in self.showfields: # This is a dict {1: ['firstname', 'lastname']} context['attrs']['data-fields'] = (self.showfields[value]) However, working with arrays is not an issue: def createoption(self, name, value, label, selected, index, subindex=None, attrs=None): context = super().createoption(name, value, label, selected, index, subindex, attrs) if not value: return context if value in allowed_values: # This is an array [1, 2] ..."}
{"_id":"q-en-django__django-14916-9b880ba9c5b9da1d2d8f42d4a106ab878d0a452194f654ea294a62d6dfd2fb86","text":"Investigate feasibility of improving WhereNode clone performance Description (last modified by Keryn Knight) This relates to insofar as it's about figuring out how shallow or deep a copy is necessary for cloning Node instances\/subclasses. And a bit to which acknowledges the need to avoid cloning sometimes due to it being costly. The PR I made for optimising Q combining\/inverting (​) wants to introduce a copy method onto Q objects to indicate what level of operation is needed. Unlike that PR, it's decidedly not possible to take the same approach with WhereNode and just do = () (things broke horribly) because of attributes like containsaggregate and outputfield which are cached properties on the WhereNode instance. Also resolved gets set after the fact by resolveexpression. But, as far as I can tell, the looping\/testing\/child cloning may be removable. Perhaps once upon a time they were required and other components changed such that it's now safe to consider. I can't readily say, but initial experimentation suggests it's OK to look at, at least in theory. There is a PR already for this, here: ​ which is currently marked draft\/WIP because I couldn't easily run the whole test suite locally and needed to see if the more exotic parts caused problems. They didn't, which was unexpected, so here I am. The PR currently replaces the method like so: clone = (children=[], connector=self.connector, negated=self.negated) for child in self.children: if hasattr(child, 'clone'): (()) else: (child) with: clone = (children=None, connector=self.connector, negated=self.negated) clone.children = self.children[:] But I think that Q and WhereNode can both just return a shallow copy by only aliasing the children property, at which point I think the copy method could be implemented on Node directly and mirror the existing deepcopy method. By preferring those stdlib names it draws a line in the sand over what level of copy should be expected. The existing clone and copy methods can become aliases to same. Before any changes If we now begin to examine the before and after with as much data as I can easily gather. First a bit of prelude: In [1]: from import WhereNode In [2]: from import QuerySet In [3]: from import Query In [4]: from import User In [5]: x = () In [6]: y = (username='test', email='').exclude(username='test').filter(email='nottest') In [7]: %loadext lineprofiler Now let's establish a ballpark for how much time is spent where, I've stripped the line profiling down to just the relevant bits for clarity. Queryset with no clauses (x) In [8]: %lprun -f -f -f -f -f for in range(1000): x.chain() Function: Line # Hits Time Per Hit % Time Line Contents 1325 def chain(self, **kwargs): ... 1330 1000 .0 88.2 97.8 obj = () Function: Line # Hits Time Per Hit % Time Line Contents 1337 def clone(self): ... 1342 1000 .0 80.8 86.3 chained = () 1343 1000 7240.0 7.2 7.7 c = (, query=chained, , )) Function: Line # Hits Time Per Hit % Time Line Contents 341 def chain(self, klass=None): ... 346 2000 .0 58.5 94.8 obj = () Function: Line # Hits Time Per Hit % Time Line Contents 290 def clone(self): ... 304 2000 .0 9.1 27.0 = () Function: Line # Hits Time Per Hit % Time Line Contents 142 def clone(self): ... 148 4000 6227.0 1.6 63.7 clone = ( 149 2000 1170.0 0.6 12.0 children=[], connector=self.connector, negated=self.negated) 150 2000 1407.0 0.7 14.4 for child in self.children: 151 if hasattr(child, 'clone'): 152 (()) 153 else: 154 (child) 155 2000 969.0 0.5 9.9 return clone Queryset with where clauses (y) Youu can see the cost begin to go up: In [9]: %lprun -f -f -f -f -f for in range(1000): y.chain() Function: Line # Hits Time Per Hit % Time Line Contents 290 def clone(self): ... 304 2000 .0 30.0 59.1 = () Function: Line # Hits Time Per Hit % Time Line Contents 142 def clone(self): 148 8000 .0 1.4 33.1 clone = ( 149 4000 2379.0 0.6 6.9 children=[], connector=self.connector, negated=self.negated) 150 7241.0 0.5 20.9 for child in self.children: 151 5553.0 0.6 16.0 if hasattr(child, 'clone'): 152 2000 1255.0 0.6 3.6 (()) 153 else: 154 8000 5069.0 0.6 14.6 (child) 155 4000 1715.0 0.4 4.9 return clone and the timings for those 2 (x and y) querysets, plus an example of prefetching related data: In [10]: %timeit x.chain() 43 µs ± 52 ns per loop (mean ± std. dev. of 7 runs, loops each) In [11]: %timeit y.chain() 24 µs ± 35.3 ns per loop (mean ± std. dev. of 7 runs, loops each)) In [12]: %timeit () 79 µs ± 38.2 ns per loop (mean ± std. dev. of 7 runs, loops each) In [13]: %prun for in range(): y.chain() function calls ( primitive calls) in 1.506 seconds ncalls tottime percall cumtime percall filename:lineno(function) 0.357 0.000 1.007 0.000 (clone) \/ 0.301 0.000 0.539 0.000 (clone) <-- WhereNode 0.172 0.000 1.344 0.000 (clone) 0.068 0.000 1.086 0.000 (chain) 0.058 0.000 1.415 0.000 (chain) In [14]: %timeit tuple(User.objects.prefetchrelated('groupspermissions', 'userpermissions')) 2 ms ± 117 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)) After replacing the clone method Given those data points, here are the same bits of line profiling, showing only those where the needle has moved: Queryset with no clauses (x) Function: Line # Hits Time Per Hit % Time Line Contents 290 def clone(self): ... 304 2000 .0 7.1 22.8 = () Function: Line # Hits Time Per Hit % Time Line Contents 142 def clone(self): ... 148 2000 5586.0 2.8 71.0 clone = (children=None, connector=self.connector, negated=self.negated) 149 2000 1412.0 0.7 18.0 clone.children = self.children[:] 150 2000 865.0 0.4 11.0 return clone Queryset with where clauses (y) Function: at line 290 Line # Hits Time Per Hit % Time Line Contents 290 def clone(self): ... 304 2000 .0 6.9 24.7 = () Function: Line # Hits Time Per Hit % Time Line Contents 142 def clone(self): ... 148 2000 5273.0 2.6 69.0 clone = (children=None, connector=self.connector, negated=self.negated) 149 2000 1498.0 0.7 19.6 clone.children = self.children[:] 150 2000 871.0 0.4 11.4 return clone And the timings; in the empty case (x), it's basically just paying the cost of class creation, but where there are clauses it is more consistent (ie: nearer a constant time vs linear growth). In [10]: %timeit x.chain() 51 µs ± 47.6 ns per loop (mean ± std. dev. of 7 runs, loops each)) In [11]: %timeit y.chain() 25 µs ± 36.4 ns per loop (mean ± std. dev. of 7 runs, loops each) In [12]: %timeit () 16 µs ± 9.09 ns per loop (mean ± std. dev. of 7 runs, loops each) In [13]: %prun for in range(): y.chain() function calls in 1.087 seconds ncalls tottime percall cumtime percall filename:lineno(function) 0.355 0.000 0.606 0.000 (clone) 0.169 0.000 0.939 0.000 (clone) 0.068 0.000 0.139 0.000 (clone) <--- WhereNode 0.067 0.000 0.683 0.000 (chain) 0.058 0.000 1.009 0.000 (chain) In [14]: %timeit tuple(User.objects.prefetchrelated('groupspermissions', 'userpermissions')) 9 ms ± 345 µs per loop (mean ± std. dev. of 7 runs, 10 loops each) Gut feeling says the \"final\" implementation could be something like the following, directly on the Node class. def copy(self): clone = self.newinstance(children=None, connector=self.connector, negated=self.negated) clone.children = self.children[:] return clone copy = copy clone = copy def deepcopy(self, memo): ... clone.children = copy.deepcopy(self.children, memo) ... which would remove the need for methods on either subclass. Note also that aliasing the methods that way rather than doing def copy(): return () as WhereNode currently does is also slightly quicker. We're talking 0.10µs - small potatoes, but it doesn't hurt to have + it's 1 fewer stack frame. A final note then: if this properly shallow clone is done, I suspect resolveexpression can be altered too, to avoid calling resolvenode if it's already been done on self - the shallow copy would ensure that anything which had previously been resolved is kept around and doesn't need doing again, I think. Other stats: Of the 14,000 tests I can run easily, only 1 time was resolved in when clone was called (the value was True). containsaggregate turned up 209 times (207 with a value of False, 2 with True), and output_field only appeared 2 times (both BooleanField). Those are the only extra attribute I spotted."}
{"_id":"q-en-django__django-14919-41677cef35f8d0b9cb0c3738f8ec545e585b812e920e9145962080b054381932","text":"Do not ignore transaction durability errors within TestCase Description (last modified by Krzysztof Jagiełło) Currently there is a discrepancy in how durable atomic blocks are handled in TransactionTestCase vs TestCase. Using the former, nested durable atomic blocks will, as expected, result in a RuntimeError. Using the latter however, the error will go unnoticed as the durability check is turned off. I have some issues with this behaviour in a codebase where we heavily utilize TestCase and where we recently started to introduce durable atomic blocks – the durability errors do not surface until the code hits staging\/production. The solution could be to switch over to TransactionTestCase for the test classes that hit code paths with durable atomic blocks, but having to identify which tests could be affected by this issue is a bit inconvenient. And then there is the performance penalty of using TransactionTestCase. So, to the issue at hand. The durability check is disabled for TestCase because otherwise durable atomic blocks would fail immediately as TestCase wraps its tests in transactions. We could however add a marker to the transactions created by TestCase, keep a stack of active transactions and make the durability check take the stack of transactions with their respective markers into account. This way we could easily detect when a durable atomic block is directly within a transaction created by TestCase and skip the durability check only for this specific scenario. To better illustrate what I am proposing here, I have prepared a PoC patch. Let me know what you think! Patch: ​"}
{"_id":"q-en-django__django-14935-932115d115dafd7b03530e1e462856d11c7eb64a32c0123ebf8111aa655e460d","text":"Add makemigrations warning for unique fields with callable defaults Description Callables on properties for ModelFields are used for various reasons. One use case is to autocreate random file names or user passwords if not present. The migration seems to call them only once because after the migration every \"Buchung\" has the same wlanpassword. My Model: def randomwlankey(): return ''.join(random.SystemRandom().choice(\"1234567890abcdefghkmnpqrstuvwxyz\") for i in range(9)) class Buchung(models.Model): [...] wlanpasswort = models.CharField(maxlength=10, default=randomwlankey) The generated migration: from future import unicodeliterals from import models, migrations import class Migration(migrations.Migration): dependencies = [ ('main', '0001initial'), ] operations = [ migrations.AddField( modelname='buchung', name='wlanpasswort', field=models.CharField(, maxlength=10), preserve_default=True, ), ]"}
{"_id":"q-en-django__django-14954-28d6856baf5c39368bd24f456043447a377edd451c4b6cfd0d9de970fb882d16","text":"createsuperuser doesn't work in non-interactive mode if a ManyToManyField is in REQUIREDFIELDS. Description ForeignKey support to REQUIREDFIELDS in createsuperuser command but this support is not working in non-interactive mode. The buggy line is ​this line. If value is an integer, () simply returns it after validation while ​ on the next line would expect a model instance for the ForeignKey field. If you go one step further and override createsuperuser to pass an instance of the model, then () raises an error because ​ expects an integer. There may be the same problem with ManyToManyField."}
{"_id":"q-en-django__django-14960-4c916576277c4d6f1a1e25b1c059b2990e94ae9837043e0cbdfdffc313ac5e38","text":"createsuperuser does not validate REQUIREDFIELDS values in interactive mode when passed by command-line. Description createsuperuser command will alway perform a field validation using when creating a new user. In non-interactive mode, it is done ​here. In interactive mode, it is performed in ​ when the value is requested from input. But if the valued was passed using -- command, the field is never validated. The consequence is it ends up passing a non-validated string to ​ instead of a integer corresponding to an actual PK in DB."}
{"_id":"q-en-django__django-14969-f4dddec01381717907b31f70292774243c43c5de6a319387ededd202465b8179","text":"Raise an error if a string is passed into hasperms() instead of a list Description (last modified by lieryan) A colleague made this error recently doing a user.hasperms(\"foobar\") instead of the correct user.hasperms([\"foobar\"]) or user.hasperm(\"foobar\"). The code initially appeared to work fine since in Python, str is an iterable that returned individual characters as string when iterated over. We checked for str in particular rather than enforcing it to be a list, since permlist may actually be tuple, set, generators, or other iterables. An alternative way this could be fixed is to just silently behave like hasperm() if perm_list is actually a string rather than raising an error, but that'll probably enforce a bad habit. Pull request in Github (​)."}
{"_id":"q-en-django__django-14983-9a49c8828b7d907356da2ec9d6db985ba34fd923830b89e74f322d70af3549bb","text":"makemigrations generates \"wrong\" numbered migration file if squashed migrations are in place Description When an app has migrations 0001initial and 0002auto201412021234 that are squashed to 0001squashed0002auto201412021234, a new call to makemigrations will generate a migration file called 0002auto201412022345 instead of 0003auto201412022345 which is quite irritating as long as 0002auto201412021234 is still around. It does make sense though when only 0001squashed0002auto20141202_1234 is left. Although the latter case eventually hits every project, I'd prefer the former."}
{"_id":"q-en-django__django-14996-93c34104201418d45dfed7ee424f45686d14e9497716b0b52e53021dd627ae04","text":"Renaming field and providing prior field name to dbcolumn should be an SQL noop Description (last modified by Jacob Walls) Renaming a field and setting the prior implicit field name as the dbcolumn to avoid db operations creates a migration emitting unnecessary SQL. Similar to , which handled a very similar scenario but where there is no field rename, I would expect a SQL noop. I tested with SQLite and MySQL 5.7.31. class Apple(models.Model): core = models.BooleanField() class Apple(models.Model): corerenamed = models.BooleanField(dbcolumn='core') Was renamed to apple.corerenamed (a BooleanField)? [y\/N] y Migrations for 'renamez': Rename field core on apple to corerenamed Alter field corerenamed on apple python sqlmigrate renamez 0002 showing unnecessary SQL: BEGIN; -- Rename field core on apple to corerenamed ALTER TABLE \"renamezapple\" RENAME COLUMN \"core\" TO \"corerenamed\"; -- Alter field corerenamed on apple ALTER TABLE \"renamezapple\" RENAME COLUMN \"corerenamed\" TO \"core\"; COMMIT; Without renaming the field, follow the same flow and get an AlterField migration without SQL, which is what intended: BEGIN; -- Alter field core on apple COMMIT;"}
{"_id":"q-en-django__django-14997-4b2fe7b7b54dd638f5b24c49eea3e775b19bf2eac89d33753e88002fbde76a59","text":"Remaking table with unique constraint crashes on SQLite. Description In Django 4.0a1, this model: class Tag(models.Model): name = models.SlugField(helptext=\"The tag key.\") value = models.CharField(maxlength=150, helptext=\"The tag value.\") class Meta: ordering = [\"name\", \"value\"] constraints = [ models.UniqueConstraint( \"name\", \"value\", name=\"uniquenamevalue\", ) ] def str(self): return f\"{}={}\" with these migrations, using sqlite: class Migration(migrations.Migration): initial = True dependencies = [ ] operations = [ migrations.CreateModel( name='Tag', fields=[ ('id', models.BigAutoField(autocreated=True, primarykey=True, serialize=False, verbosename='ID')), ('name', models.SlugField(helptext='The tag key.')), ('value', models.CharField(helptext='The tag value.', maxlength=200)), ], options={ 'ordering': ['name', 'value'], }, ), migrations.AddConstraint( modelname='tag', constraint=models.UniqueConstraint(('name'), ('value'), name='uniquenamevalue'), ), ] class Migration(migrations.Migration): dependencies = [ ('myapp', '0001initial'), ] operations = [ migrations.AlterField( modelname='tag', name='value', field=models.CharField(helptext='The tag value.', maxlength=150), ), ] raises this error: migrate Operations to perform: Apply all migrations: admin, auth, contenttypes, myapp, sessions Running migrations: Applying Traceback (most recent call last): File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\backends\\\", line 84, in execute return (sql, params) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\backends\\sqlite3\\\", line 416, in execute return Database.Cursor.execute(self, query, params) sqlite3.OperationalError: the \".\" operator prohibited in index expressions The above exception was the direct cause of the following exception: Traceback (most recent call last): File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\core\\management\\\", line 373, in runfromargv self.execute(args, cmdoptions) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\core\\management\\\", line 417, in execute output = (args, options) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\core\\management\\\", line 90, in wrapped res = handlefunc(args, *kwargs) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\core\\management\\commands\\\", line 253, in handle postmigratestate = executor.migrate( File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\migrations\\\", line 126, in migrate state = self.migrateallforwards(state, plan, fullplan, fake=fake, fakeinitial=fakeinitial) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\migrations\\\", line 156, in migrateallforwards state = self.applymigration(state, migration, fake=fake, fakeinitial=fakeinitial) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\migrations\\\", line 236, in applymigration state = (state, schemaeditor) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\migrations\\\", line 125, in apply operation.databaseforwards(self.applabel, schemaeditor, oldstate, projectstate) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\migrations\\operations\\\", line 225, in databaseforwards schemaeditor.alterfield(frommodel, fromfield, tofield) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\backends\\sqlite3\\\", line 140, in alterfield super().alterfield(model, oldfield, newfield, strict=strict) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\backends\\base\\\", line 618, in alterfield self.alterfield(model, oldfield, newfield, oldtype, newtype, File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\backends\\sqlite3\\\", line 362, in alterfield self.remaketable(model, alterfield=(oldfield, newfield)) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\backends\\sqlite3\\\", line 303, in remaketable self.execute(sql) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\backends\\base\\\", line 151, in execute cursor.execute(sql, params) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\backends\\\", line 98, in execute return super().execute(sql, params) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\backends\\\", line 66, in execute return self.executewithwrappers(sql, params, many=False, executor=self.execute) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\backends\\\", line 75, in executewithwrappers return executor(sql, params, many, context) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\backends\\\", line 84, in execute return (sql, params) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\\", line 90, in exit raise djexcvalue.withtraceback(traceback) from excvalue File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\backends\\\", line 84, in execute return (sql, params) File \"D:\\Projects\\Development\\sqliteerror.venv\\lib\\site-packages\\django\\db\\backends\\sqlite3\\\", line 416, in execute return Database.Cursor.execute(self, query, params) the \".\" operator prohibited in index expressions"}
{"_id":"q-en-django__django-14999-f773abbe3e6c748798ed51df65fb1bf6209324b40b5e9dc3af569e2e6079d01e","text":"RenameModel with dbtable should be a noop. Description A RenameModel operation that already has dbtable defined must be a noop. In Postgres, it drops and recreates foreign key constraints. In sqlite it recreates the table (as expected for a table renaming)."}
{"_id":"q-en-django__django-15018-a11026c9444875a48ca1d102af4afa3bac1375b07f259aa31a9b33b232b68072","text":"callcommand() fails when required mutually exclusive arguments use the same . Description I have a command which accepts two different ways to specify a time -- either as a timestamp or as a duration in the future: pause (--for duration | --until time) class Command(BaseCommand): def addarguments(self, parser) -None: group = parser.addmutuallyexclusivegroup(required=True) group.addargument('--for', dest='until', action='store', type=parsedurationtotime) group.addargument('--until', action='store', type=parsetime) def handle(self, until: datetime, **): pass This works fine on the command line, however there doesn't seem to be a way to make this work through callcommand. Specifically there are two sides to the failure: while I can provide an until value (as a string, which is processed by parsetime) there is no mechanism to pass a for value if that's how I want to spell the input the for value is always required and attempts to parse the (string) until value passed, which then errors since the input formats are very different"}
{"_id":"q-en-django__django-15022-e11eb7b55f7de3dcb8e357384f9d855acd97a3e9a0830b8c6376b4215da4e004","text":"Unnecessary joins in admin changelist query Description Django 1.2.5 Models: class Client(models.Model): name = models.CharField(('name'), maxlength=256) name2 = models.CharField(('unofficial or obsolete name'), maxlength=256, blank=True, null=True) contactperson = models.CharField(('contact person'), maxlength=256, blank=True, null=True) ... class ClientOffice(models.Model): name = models.CharField(('name'), maxlength=256) name2 = models.CharField(('unofficial or obsolete name'), maxlength=256, blank=True, null=True) ... client = models.ForeignKey(Client, verbosename=('client')) ... and admin options like these: class ClientAdmin(admin.ModelAdmin): searchfields = ('name', 'name2', 'contact_person', 'clientofficename', 'clientofficename2') ... Numbers: () () Now, if we try searching for clients in admin by a search query containig several words (3), got django\/admin stalled. The problem is going to be that each word in the search query leads to additional JOIN in final SQL query beacause of qs = (...) pattern. The attached patch is for Django 1.2.5, but adopting for the current SVN trunk is trivial."}
{"_id":"q-en-django__django-15031-39fffd3e0c75c5f98b7d1697280263876a1d19128e4bf0187b08288a8488f030","text":"Django admin allows filtering using the field lookups such as \"in\", but it is impossible to include a value that contains a comma Description The admin site allows you to filter the queryset in the changelist in a plenty of different ways. Notably, it allows you to filter the records by multiple values (if the field's value is one of the specified value options, then such record is considered matching). For example, you can test it with a query string like this: \/admin\/auth\/user\/?username_in=johnny,viola,gordon Unfortunately, there is a big limitation at the moment: you can't include a value option that contains a comma (or a few). The function that splits the string is preparelookup_value, found in ​"}
{"_id":"q-en-django__django-15037-6d4d9c1a723081467279705e06b93159a05ff7b2e3647e21c473ae03106aa992","text":"Foreign key to a specific field is not handled in inspectdb Description (last modified by Tim Graham) if you have a DB like that CREATE TABLE foo ( id serial primary key, otherid int UNIQUE); CREATE TABLE bar ( id serial primary key, otherid int, constraint myconst FOREIGN KEY(otherid) references foo(otherid) ); the generated model for the bar table will have the otherid be a FK to foo and not foo(otherid). I'm attaching a potential fix for this. Sorry I had no time for the UTs."}
{"_id":"q-en-django__django-15038-3adef1ef3b98bba5845b11e111b24c32f0e1160702c2ff5a2ce20178743bbb6a","text":"--squashed-name can overwrite existing migration files Description squashmigrations --squashed-name initial myapp 0002 will overwrite If that migration has already been applied anywhere, the code and the database will be in an inconsistent state. Furthermore, the 0002... migration likely depends on 0001initial, so the new replaces = …0002… line to introduces a CircularDependencyError. squashmigrations should exit with an error instead of overwriting existing migration files. --squashed-name can overwrite existing migration files Description squashmigrations --squashed-name initial myapp 0002 will overwrite If that migration has already been applied anywhere, the code and the database will be in an inconsistent state. Furthermore, the 0002... migration likely depends on 0001initial, so the new replaces = …0002… line to introduces a CircularDependencyError. squashmigrations should exit with an error instead of overwriting existing migration files."}
{"_id":"q-en-django__django-15044-8ea0ed890001f088d0c5eaffea63f2cfcc9823930a5bb6fbb079214e019e0c3b","text":"CacheMiddleware and FetchFromCacheMiddleware are not thread safe. Description CacheMiddleware persist = caches[cachealias] on startup and it is not thread safe. ​ I found that after some production errors with pylibmc and uwsgi threaded. Created a small project to reproduce it. Nothing fancy, just pylibmc cache and a cached view. It fails even with development server, with concurrent requests. Traceback (most recent call last): File \"versions\/pylibmcbug\/lib\/python3.9\/site-\", line 47, in inner response = getresponse(request) File \"versions\/pylibmcbug\/lib\/python3.9\/site-\", line 181, in getresponse response = wrappedcallback(request, callbackargs, *callbackkwargs) File \"versions\/pylibmcbug\/lib\/python3.9\/site-\", line 122, in wrappedview result = middleware.processrequest(request) File \"versions\/pylibmcbug\/lib\/python3.9\/site-\", line 145, in processrequest cachekey = getcachekey(request, self.keyprefix, 'GET', ) File \"versions\/pylibmcbug\/lib\/python3.9\/site-\", line 362, in getcachekey headerlist = (cachekey) File \"versions\/pylibmcbug\/lib\/python3.9\/site-\", line 77, in get return (key, default) pylibmc.ConnectionError: error 3 from memcachedget(): (0x7f290400bd60) FAILURE, poll() returned a value that was not dealt with, host: localhost: - Looking for git history, it is this way since 2010. ​"}
{"_id":"q-en-django__django-15052-0e2f8a562e76f8fd6ccaf6b33245ea637a8cf2c38b1f781c667f43db2cec7fa1","text":"Aggregate filtered by an Exists subquery crashes Description For example: (\"publisher\").aggregate( maxrating=Max( \"rating\", filter=Exists( (book=OuterRef(\"pk\")), ), ) Will crash with the following traceback: Traceback (most recent call last): File \"\", line 146, in testfilteredaggregatewithexists aggregate = ('publisher').aggregate( File \"\", line 405, in aggregate return query.getaggregation(, kwargs) File \"\", line 501, in getaggregation result = compiler.executesql(SINGLE) File \"\", line 1189, in executesql sql, params = () File \"\", line 531, in assql extraselect, orderby, groupby = self.presqlsetup() File \"\", line 59, in presqlsetup self.setupquery() File \"\", line 50, in setupquery , self.klassinfo, self.annotationcolmap = self.getselect() File \"\", line 267, in getselect sql, params = self.compile(col) File \"\", line 463, in compile sql, params = (self, self.connection) File \"\", line 90, in assql return sql, params + filterparams TypeError: can only concatenate list (not \"tuple\") to list The following patch should fix the issue: diff --git index .. --- +++ -87,7 +87,7 class Aggregate(Func): compiler, connection, template=template, filter=filtersql, extracontext ) return sql, params + filterparams return sql, (params, filterparams) else: copy = () = None"}
{"_id":"q-en-django__django-15061-62abf57b374eade6beec2c354daa01a9304a4c033d8e5335bcc17e49f1f484a9","text":"Remove \"for = ...\" from MultiWidget's <label. Description The instance from Raw MultiWidget class generate idforlabel like f'{id}0' It has not sense. For example ChoiceWidget has self.addidindex and I can decide it myself, how I will see labelid - with or without index. I think, it is better to remove completely idforlabel method from MultiWidget Class."}
{"_id":"q-en-django__django-15062-1c918ac2fbad0408d55e3d955aaea2f79172649531ac791fb3d91694e3638f56","text":"Case() and ExpressionWrapper() doesn't work with DecimalField on SQLite. Description I noticed that, on sqlite, some comparisons against DecimalField annotations behave unexpectedly, in particular when wrapping a DecimalField value in a Case\/When or ExpressionWrapper. I suspect that there might be some inconsistencies in the type conversions here somehow. I've created a few testcase to illustrate the problem on current git main: ​ and see below. def test00comparefield(self): \"\"\"Comparing a Case annotation wrapping a field to a literal works.\"\"\" (a='', d=1) try: (dgt=0).get() finally: from import connection print(connection.queries[-1]['sql']) def test01compareannotationvalueliteral(self): \"\"\"Comparing a literal annotation using Value to a literal works.\"\"\" # Fields are not actually used here (a='', d=0) try: Foo.objects.annotate( x=models.Value(1, (maxdigits=1, decimalplaces=0)), ).filter(xgt=0).get() finally: from import connection print(connection.queries[-1]['sql']) def test02compareannotationexpressionwrapperliteral(self): \"\"\"Comparing a literal annotation using ExpressionWraper and Value to a literal works.\"\"\" # Fields are not actually used here (a='', d=0) try: Foo.objects.annotate( x=models.ExpressionWrapper( models.Value(1), (maxdigits=1, decimalplaces=0), ), ).filter(xgt=0).get() finally: from import connection print(connection.queries[-1]['sql']) def test03comparecaseannotation(self): \"\"\"Comparing a Case annotation wrapping a field to a literal works.\"\"\" (a='', d=1) try: Foo.objects.annotate( x=models.Case(models.When(a='', then=models.F('d'))), ).filter(xgt=0).get() finally: from import connection print(connection.queries[-1]['sql']) test00comparefield compares a field directly with a literal, which works. test01compareannotationvalueliteral adds a literal annotation using just Value and then compares it, which also works. test02compareannotationexpressionwrapperliteral adds a literal annotation using Value wrapped in ExpressionWrapper, which does not work becomes a literal int, rather than a string like the compared value. test03comparecaseannotation wraps the field in a case\/when and then compares it, which also does not work (maybe the CASE changes the type?) Running these testcases against sqlite gives: SELECT \"modelfieldsfoo\".\"id\", \"modelfieldsfoo\".\"a\", \"modelfieldsfoo\".\"d\" FROM \"modelfieldsfoo\" WHERE \"modelfieldsfoo\".\"d\" '0' LIMIT 21 .SELECT \"modelfieldsfoo\".\"id\", \"modelfieldsfoo\".\"a\", \"modelfieldsfoo\".\"d\", CAST('1' AS NUMERIC) AS \"x\" FROM \"modelfieldsfoo\" WHERE CAST('1' AS NUMERIC) '0' LIMIT 21 .SELECT \"modelfieldsfoo\".\"id\", \"modelfieldsfoo\".\"a\", \"modelfieldsfoo\".\"d\", 1 AS \"x\" FROM \"modelfieldsfoo\" WHERE 1 '0' LIMIT 21 ESELECT \"modelfieldsfoo\".\"id\", \"modelfieldsfoo\".\"a\", \"modelfieldsfoo\".\"d\", CASE WHEN \"modelfieldsfoo\".\"a\" = '' THEN \"modelfieldsfoo\".\"d\" ELSE NULL END AS \"x\" FROM \"modelfieldsfoo\" WHERE CASE WHEN (\"modelfieldsfoo\".\"a\" = '') THEN \"modelfieldsfoo\".\"d\" ELSE NULL END '0' LIMIT 21 E.s........... ERROR: test02compareannotationexpressionwrapperliteral (modelfields.testdecimalfield.DecimalFieldTests) Comparing a literal annotation using ExpressionWraper and Value to a literal works. Traceback (most recent call last): File \"\", line 437, in inner return func(args, kwargs) File \"\", line 142, in test02compareannotationexpressionwrapperliteral Foo.objects.annotate( File \"\", line 441, in get raise ( Foo matching query does not exist. ERROR: test03comparecaseannotation (modelfields.testdecimalfield.DecimalFieldTests) Comparing a Case annotation wrapping a field to a literal works. Traceback (most recent call last): File \"\", line 437, in inner return func(args, kwargs) File \"\", line 154, in test03comparecaseannotation Foo.objects.annotate( File \"\", line 441, in get raise ( Foo matching query does not exist. Note in the printed queries that the 0 value that is compared with is a string in the query ('0'), while the ExpressionWrappered value is just a plain number (1). The Value without ExpressionWrapper has a cast to NUMERIC that I suspect makes that case work? Also note that the DEBUG stuff and try-catch is only to capture the query that is being generated, it is not needed to trigger the problem. I tried printing the attribute first, but that seems to hide the quotes around the literal 0 in the query, so took me a while to realize what was happening. The numbers in the testcase names are to ensure the SQL queries are printed in a recognizable order. All four testcases work on MySQL, there the 0 value is written in the query without the quotes, and the cast to NUMERIC is also missing. I suspect this issue is highly similar to , and can a fix can probably build on the fix for that issue."}
{"_id":"q-en-django__django-15087-b2f9102b04e0bbd4537ac30205d187288c5e27bb95efaa972212c653b82960a3","text":"SQLite schema introspection should use PRAGMA instead of parsing to retrieve foreign keys Description The SQLite backend's DatabaseIntrospection.getrelations uses a complex combination of regexes to extract foreign key constraints from DDL while ​PRAGMA foreignkeylist works just fine as ​getforeignkey_constraints."}
{"_id":"q-en-django__django-15098-54c50cf90498be7927884287d28e979623f76541a120a48fa59da4add0c149f5","text":"Internationalisation didn't support language locale containing both script and region. Description The i18npatterns didn't work with locale contains both script and region, like en-latn-us. Given LANGUAGECODE = 'en-us' LANGUAGES = [ ('en-us', \"English\"), ('en-latn-us', \"Latin English\"), ('en-Latn-US', \"BCP 47 case format\"), ] from import i18npatterns from import HttpResponse def bangiah(request): return HttpResponse('U!') urlpatterns += i18npatterns( path('', bangiah), ) The response of http:\/\/localhost:8000\/en-us\/ is 200 U!. The response of http:\/\/localhost:8000\/en-lat-us\/ is 404 not found. The response of http:\/\/localhost:8000\/en-Latn-US\/ is 404 not found. Steps to Reproduce Start a new project with django-admin startproject tshi and cd tshi\/ Append to as follows LANGUAGES = [ ('en-us', \"English\"), ('en-latn-us', \"Latin English\"), ('en-Latn-US', \"BCP 47 case format\"), ] MIDDLEWARE += [ '', ] Edit by appending follows from import i18npatterns from import HttpResponse def bangiah(request): return HttpResponse('U!') urlpatterns += i18npatterns( path('', bangiah), ) python migrate python runserver The results The response of http:\/\/localhost:8000\/en-us\/ is 200 U!. The response of http:\/\/localhost:8000\/en-lat-us\/ is 404 not found. The response of http:\/\/localhost:8000\/en-Latn-US\/ is 404 not found. Expect to happen instead The response of http:\/\/localhost:8000\/en-latn-us\/ and http:\/\/localhost:8000\/en-Latn-US\/ should be 200 U!. The en-Latn-US tag follows format defined in ​RFC 5646. It's ​documented that the language part is always in lowercase, following ​Accept-Language. ​Accept-Language is following ​Content-Language Header, which is following ​RFC 5646. The ​RFC 5646 defined langtag as follow: langtag = language [\"-\" script] [\"-\" region] (\"-\" variant) (\"-\" extension) [\"-\" privateuse] language = 23ALPHA ; shortest ISO 639 code [\"-\" extlang] ; sometimes followed by ; extended language subtags \/ 4ALPHA ; or reserved for future use \/ 58ALPHA ; or registered language subtag extlang = 3ALPHA ; selected ISO 639 codes *2(\"-\" 3ALPHA) ; permanently reserved script = 4ALPHA ; ISO code region = 2ALPHA ; ISO 3166-1 code \/ 3DIGIT ; UN M.49 code I have confirmed that this issue can be reproduced as described on a fresh Django project Python version: 3.7.5 Django version: 3.2.7"}
{"_id":"q-en-django__django-15102-89202f99cdf1c3dbc880f5654b004c42ad02af9b4805b4ad5207ea091459f632","text":"startproject should honor umask Description Ticket fixed the problem of not being made executable by copying all the permission bits (not just the executable flags). This means that the user's umask doesn't work, e.g.: $ umask 077 $ touch foo $ ls -l foo -rw------- 1 talex talex 0 2007-05-12 13:27 foo $ PYTHONPATH=trunk .\/trunk\/django\/bin\/django- startproject mysite $ ls -l -rw-r--r-- 1 talex talex 2804 2007-05-12 13:28 I discovered this whilst trying to make a Zero Install package for Django. Everything in the Zero Install cache is read-only, so startproject fails with: File \"\", line 799, in startproject fp = open(mainsettingsfile, 'w') IOError: [Errno 13] Permission denied: '' Thanks,"}
{"_id":"q-en-django__django-15103-b672acfd1636506079c554f9de69972a6d3a25aebfa38c0e02fd8bf8ff5930f3","text":"Make the elementid argument of jsonscript optional Description I recently had a use-case where I wanted to use json_script but I didn't need any id for it (I was including the <scriptinside a <templateso I didn't need an id to refer to it). I can't see any reason (security or otherwise) for the id to be required and making it optional doesn't seem to break any tests."}
{"_id":"q-en-django__django-15104-b485c4503d01bca69bbbc5b041b5cc678ef25479b6c51ad3eaa2f53467af1630","text":"KeyError with migration autodetector and FK field with hardcoded reference Description Hi, I encountered this issue on an old Django project (probably 10 years old) with tons of models and probably a lot of questionable design decisions. The symptom is that running our test suite in verbose mode doesn't work: $ python test -v 2 Creating test database for alias 'default' ('testproject')... Operations to perform: Synchronize unmigrated apps: [... about 40 apps] Apply all migrations: (none) Synchronizing apps without migrations: Creating tables... Creating table authpermission Creating table authgroup Creating table authuser Creating table djangocontenttype Creating table djangosession Creating table djangoadminlog [... 100 or so more tables] Running deferred SQL... Running migrations: No migrations to apply. Traceback (most recent call last): File \"\", line 17, in <moduleexecutefromcommandline() File \"\", line 401, in executefromcommandline utility.execute() File \"\", line 395, in execute self.fetchcommand(subcommand).runfromargv() File \"\", line 23, in runfromargv super().runfromargv(argv) File \"\", line 330, in runfromargv self.execute(args, cmdoptions) File \"\", line 371, in execute output = (args, options) File \"\", line 53, in handle failures = testrunner.runtests(testlabels) File \"\", line 697, in runtests oldconfig = self.setupdatabases(aliases=databases) File \"\", line 618, in setupdatabases self.parallel, kwargs File \"\", line 174, in setupdatabases serialize=connection.settingsdict['TEST'].get('SERIALIZE', True), File \"\", line 77, in createtestdb runsyncdb=True, File \"\", line 168, in callcommand return command.execute(args, defaults) File \"\", line 371, in execute output = (args, options) File \"\", line 85, in wrapped res = handlefunc(args, kwargs) File \"\", line 227, in handle changes = autodetector.changes() File \"\", line 43, in changes changes = self.detectchanges(convertapps, graph) File \"\", line 160, in detectchanges self.generaterenamedmodels() File \"\", line 476, in generaterenamedmodels modelfieldsdef = self.onlyrelationagnosticfields() File \"\", line 99, in onlyrelationagnosticfields del deconstruction[2]['to'] KeyError: 'to' I finally did some digging and found that the culprit is a custom ForeignKey field that hardcodes its to argument (and thus also removes it from its deconstructed kwargs). It seems that the autodetector doesn't like that. Here's a self-contained reproduction test to replicate the issue: from import models from import MigrationAutodetector from import ModelState, ProjectState from import TestCase class CustomFKField(models.ForeignKey): def init(self, args, kwargs): kwargs['to'] = 'testapp.HardcodedModel' super().init(*args, kwargs) def deconstruct(self): name, path, args, kwargs = super().deconstruct() del kwargs[\"to\"] return name, path, args, kwargs class ReproTestCase(TestCase): def testreprodution(self): before = ProjectState() before.addmodel(ModelState('testapp', 'HardcodedModel', [])) after = ProjectState() after.addmodel(ModelState('testapp', 'HardcodedModel', [])) after.addmodel(ModelState('testapp', 'TestModel', [('custom', CustomFKField(ondelete=models.CASCADE))])) changes = MigrationAutodetector(before, after).detectchanges() self.assertEqual(len(changes['testapp']), 1) While I'll happily admit that my custom field's design might be questionable, I don't think it's incorrect and I think the autodetector is at fault here. Changing del deconstruction[2]['to'] to deconstruction[2].pop('to', None) on the line indicated by the traceback makes my test suite run again, in all its glorious verbosity. Seems like an innocent enough fix to me."}
{"_id":"q-en-django__django-15108-eac75c092192bcc2906914df895d882b69f62d623a95625c0e72e3f48a4e7ccf","text":"Window(orderby) should allow usage of descending string syntax to be used Description (last modified by Simon Charette) The QuerySet.orderby and ​some aggregates ordering kwarg allows for the leading dash syntax to be used but Window.orderby doesn't as it solely wraps the provided orderby in ExpressionList(expressions=orderby). This makes for an inconsistent API so I suggest we reuse the logic in in As a related note it seems most of the logic of OrderableAggMixin could be simplified by using ExpressionList. It's a shame that we used ordering and not orderby as a kwarg for OrderableAggMixin as it's now inconsistent. Also not sure how much of a public API the OrderBy expression is but I wish it was initially named Sort (or Ordering?) so that we could define class OrderBy(ExpressionList): template = 'ORDER BY %(expressions)s' def init(self, expressions, extra): expressions = [ (Sort(F(expr[1:]), descending=True) if isinstance(expr, str) and expr[0] == '-' else expr) for expr in expressions ] super().init(expressions, *extra) And then simply use this abstraction in Window and Postgres orderable aggregates. Assigning to myself as I plan to have a look at this in next few days."}
{"_id":"q-en-django__django-15111-406f6733c8fb3c145062eef11cf9a1c2a2ead101d89f849c01f21bf3e2ef0bb8","text":"django-admin start[project|app] doesn't send proper user agent header when HTTP(S) url is sent. Description (last modified by rsp2k) By default, when fetching a remote template, django-admin start[app|project] uses the default urllib User Agent which causes some sites to block requests (namely gitlab, since they use cloudflare See ​). This patch sets the 'User Agent' header to Django\/DJANGO_VERSION."}
{"_id":"q-en-django__django-15127-7cd158ee2cce0ce8fef5f7090a6d89be3ed6636905cfc20d4ec4eba2ec63d421","text":"LEVELTAGS not updated when using Description When reading messages inside tests, new message tags created using is not updated. That causes the property results to be an empty string and not know the new tags."}
{"_id":"q-en-django__django-15128-415d13d5512dcda7970684f3d954a00d8658db263f53bc53c1cd05d269cf0acf","text":"Query.changealiases raises an AssertionError Description Python Version: 3.9.2 Django Version: 2.2.24, 3.2.9 (reproduced using two different versions) Code to Reproduce from import models class Foo(models.Model): qux = models.ForeignKey(\"app.Qux\", ondelete=models.CASCADE, relatedname=\"foos\") class Bar(models.Model): foo = models.ForeignKey(\"app.Foo\", ondelete=models.CASCADE, relatedname=\"bars\") anotherfoo = models.ForeignKey(\"app.Foo\", ondelete=models.CASCADE, relatedname=\"otherbars\") baz = models.ForeignKey(\"app.Baz\", ondelete=models.CASCADE, relatedname=\"bars\") class Baz(models.Model): pass class Qux(models.Model): bazes = models.ManyToManyField(\"app.Baz\", relatedname=\"quxes\") from import Q from import Foo, Qux qux = () qs1 = () qs2 = ( Q(()) | Q(()) ) qs2 | qs1 qs1 | qs2 Description I have encountered this bug during working on a project, recreated the code to reproduce as simple as I can. I have also examined the reason behind this bug, as far as I understand the reason is that during an or operation of two QuerySets, in Query.combine method of the variable combined, if rhs's Query currently have sequential aliases (e.g. T4 and T5) and related tablenames also exist in lhs.tablemap, calling Query.tablealias in will result in creation of aliases T5 for T4 and T6 for T5, thus changemap's keys intersect with changemap's values, so the AssertionError above is raised. Expectation Could you please fix this bug? Maybe aliasmap of rhs can be provided to and Query.tablealias, and suffix (number) of the new alias might be incremented until it is not in rhs.aliasmap, to prevent intersection between changemap's keys and values. Assertion in the first line of QuerySet.changealiases is not documented via a comment. As far as I understand, it is there because if keys and values intersects it means that an alias might be changed twice (e.g. first T4 -T5, and then T5 -T6) according to their order in the changemap. IMHO there can be a comment about what it assures, or an explanation can be to the AssertionError (like the assertions in the Query.combine method). It seems like QuerySet's OR operation is not commutative (they can create different queries, even though the results are the same), IMHO this can be explicitly declared on the documentation."}
{"_id":"q-en-django__django-15135-50685e6b2736173767e1889f0c5f3e64e5ddb7a64a571b0a5a43f5a52b094204","text":"Saving parent object after setting on child leads to unexpected data loss in bulkupdate(). Description Consider following example: class Child(models.Model): pass class Parent(models.Model): child = models.ForeignKey(Child, ondelete=models.CASCADE, null=True) parent = (child=None) = Child() () Parent.objects.bulkupdate([parent], fields=[\"child\"]) Expected behavior: parent model instance was updated with the ID to the child in the database. Actual behavior: Parent model is still referencing Null. There should probably be some check for ForeignKeys in the bulkupdate logic, and if one is updated, the ID of the child model should be re-copied to the child_id field that is actually being written to the database."}
{"_id":"q-en-django__django-15139-08e7fbacc04df96a886a51d789564282cb2acbd0d446424b4f7f4bf7a93e23c2","text":"Deprecate PickleSerializer and move it out of core Description Pickle serializer has long been known to be dangerous. This is mitigated by requiring MAC on pickle in cookies, but nevertheless, RCEs continue to happen: ​ To further discourage it's use, we should consider deprecating PickleSerializer and moving it into a third party package."}
{"_id":"q-en-django__django-15154-16e51569a477fc5d1af1a3206112e3ad356013707c8216fce33badbb24d94f60","text":"Detecting uniqueness doesn't work for models with functional unique constraints. Description When creating a new object from Django Administration site, I'm getting an error \"Tag with this already exists.\". (Tag is my model name) I'm using on this model the new functional unique constraints introducted in Django 4.0 (​). I suppose this is related to the contraints put on the model. However, creating a Tag object with code - e.g. (...) - works fine. It only fails in Django admin. I'm not sure if it's a bug or misunderstanding\/misconfiguration on my side? This is my model: class Tag(models.Model): \"\"\"Basic tag model.\"\"\" # When using getorcreate, we need to pass tag name as a default value, like this: # Tag.objects.getorcreate(defaults={'name': tag}, nameiexact=tag, ) user = models.ForeignKey(settings.AUTHUSERMODEL, ondelete=models.CASCADE, verbosename=('user')) name = models.CharField(maxlength=50, dbindex=False, verbosename=('name'), blank=False, # It's the default behavior, but let's be explicit validators=[RegexValidator(regex=r'[A-z0-9À-ž\\s]+', message='This field accepts only letters, digits and space.')]) slug = models.SlugField(maxlength=50, verbosename=('slug')) class Meta: ordering = ['name'] # Make sure tag name: # - are unique per user # - are case insensitive (prevent adding \"test\" if \"Test\" already exists) # - aren't empty constraints = [models.UniqueConstraint(fields=['user', 'name'], name='%(applabel)s%(class)snameuniqueperuser'), models.UniqueConstraint(Lower('name'), name='%(applabel)s%(class)snamecaseinsensitive'), models.CheckConstraint(check=models.Q(namelengthgt=0), name='%(applabel)s%(class)snamenotempty')] (snip) This is the admin configuration: class TagAdmin(admin.ModelAdmin): listdisplay = ('pk', 'user', 'name', 'slug') listdisplaylinks = ['pk'] fields = ('user', 'name',) listfilter = ('useremail',) search_fields = ('name',)"}
{"_id":"q-en-django__django-15161-f142e606bbbf33b2f3055cc46a18d103ff70f42f690e3dc2836d9cb7ca69f250","text":"Use simplified paths for deconstruct of expressions Description Previously F() deconstructed to: (). But since it can also be imported from , ​PR changed it to deconstruct to (). This simplifies generated migration code where it will be referenced only as from import models \/ models.F(). As Mariusz pointed out on the PR, the same technique can be applied to other expressions, further simplifying generated migrations."}
{"_id":"q-en-django__django-15166-dc56e947015d7266cb3c733fd9945846860e5742b5b8333ec90ccd1a1bcf6e47","text":"DatabaseCache backend doesn't quote all fields in queries Description Snowflake requires all fields to be quoted, otherwise they're treated as uppercase. The attached patch works with stable\/3.2.x, but I'll have to review it once ​django-snowflake development is caught up to Django's main branch."}
{"_id":"q-en-django__django-15180-4635d0a601ed50344fb39b3c139f535a27059b2ea85ea7f3b89ac952bb4797ac","text":"path()\/repath() should raise a TypeError when kwargs is not a dict. Description Apparently, however many years into using Django, I'm still capable of making a \"newbie\" mistake and getting confused. So perhaps other actual new users encounter similar, especially given the lack of typing specifiers. I defined a URL like so: urlpatterns = [ path(\"path\/to\/thing\", (), \"myview\"), ] which ... well, you either spot the issue immediately or you don't, and end up with the following. If you try and resolve() the path (eg: by making a request in your browser), you'll get something like: In [3]: resolve(\"\/path\/to\/thing\") in resolve(path, urlconf) 22 if urlconf is None: 23 urlconf = geturlconf() ---24 return getresolver(urlconf).resolve(path) 25 26 in resolve(self, path) 586 for pattern in self.urlpatterns: 587 try: --588 submatch = pattern.resolve(newpath) 589 except Resolver404 as e: 590 self.extendtried(tried, pattern, e.args[0].get('tried')) in resolve(self, path) 388 newpath, args, kwargs = match 389 # Pass any extrakwargs as kwargs. --390 (self.defaultargs) 391 return ResolverMatch(self.callback, args, kwargs, , route=str(self.pattern)) 392 ValueError: dictionary update sequence element has length 1; 2 is required The crux of the issue being that I meant to give the URL a name, and it's a super unfortunate history that kwargs comes before the name argument (because nearly everyone gives a URL a name, but passing static kwargs is comparatively infrequent). So what's actually happened is that kwargs = \"myview\" and eventually self.defaultargs = \"myview\". If I update to path(\"path\/to\/thing\", (), \"myview\", name=\"myview\"), leaving the type incorrect, I can get the following error via reverse, too: In [4]: reverse(\"myview\") in reverse(viewname, urlconf, args, kwargs, currentapp) 84 resolver = getnsresolver(nspattern, resolver, tuple(())) 85 ---86 return resolver.reversewithprefix(view, prefix, args, kwargs) 87 88 in reversewithprefix(self, lookupview, prefix, args, **kwargs) 669 if set(kwargs).symmetricdifference(params).difference(defaults): 670 continue --671 if any((k, v) != v for k, v in ()): 672 continue 673 candidatesubs = kwargs AttributeError: 'str' object has no attribute 'items' Both of these suggest that either there should be a type-guard in path to assert it's dict-ish (if not None), or a system check on URLPattern to raise a friendly message. Well, they actually continue to suggest to me that everything after the view argument should be keyword-only, or that kwargs should come later, but I suspect those to be a harder sell ;) This is specifically around the kwargs, but it doesn't look like there's any guarding on the name either, and I feel like a name of {'test': 'test'} (i.e. accidentally swapped both positionals) is likely to bite & cause an issue somewhere."}
{"_id":"q-en-django__django-15199-2392e49a441b912479af479841160404d8e943661327e6d690ddc955f5c71f89","text":"Deprecate cachedproperty's name argument Description Django 2.2 is the last version to support Python 3.5 where cachedproperty's name argument is required. Following the release of Django 4.0, most apps will drop support for Django 2.2 (and hence Python 3.5), so Django 4.0 can deprecate the name argument without giving warnings that aren't actionable as long as apps want to keep support for Python 3.5."}
{"_id":"q-en-django__django-15202-3c5aeb3d6d38b544bea0a42f6afbb04742d9391c30f07ac03f013870eb5f0cc8","text":"URLField throws ValueError instead of ValidationError on clean Description forms.URLField( ).clean('\/\/\/\/] results in: ValueError: Invalid IPv6 URL Traceback (most recent call last): File \"\", line 22, in TestOneInput File \"\", line 350, in testformsURLField File \"\", line 151, in clean File \"\", line 136, in runvalidators File \"\", line 130, in call File \"\", line 440, in urlsplit"}
{"_id":"q-en-django__django-15204-2eeb1d4383adb4c7a7c5192e1c324b044da2e00301854567cda0b9565e35c5a6","text":"fails to handle broken data Description (last modified by Florian Apolloner) The actual input string was 'P3(3D' === Uncaught Python exception: === ValueError: could not convert string to float: '3(3' Traceback (most recent call last): File \"\", line 22, in TestOneInput File \"\", line 294, in testformsDurationField File \"\", line 149, in clean File \"\", line 502, in topython File \"\", line 154, in parse_duration File \"\", line 154, in <dictcomp"}
{"_id":"q-en-django__django-15206-22d3ed8fb1d73c44dce0411fc6f5e9959de5a65a3b446ffb569e110446cf507a","text":"nevercache()\/cachecontrol() decorators raise error on duck-typed requests. Description The cache decorators cachecontrol, nevercache and sensitivepostparameters no longer work with Django REST framework because they strictly check for an HttpRequest instance."}
{"_id":"q-en-django__django-15213-50e721590e98c963522ba65b72980483b1c64ae2712c240d1feee4d1190c1729","text":"ExpressionWrapper for ~Q(pkin=[]) crashes. Description (last modified by Stefan Brand) Problem Description I'm reducing some Q objects (similar to what is described in ticket:. Everything is fine for the case where the result is ExpressionWrapper(Q(pkin=[])). However, when I reduce to ExpressionWrapper(~Q(pkin=[])) the query breaks. Symptoms Working for ExpressionWrapper(Q(pkin=[])) print(queryset.annotate(foo=ExpressionWrapper(Q(pkin=[]), outputfield=BooleanField())).values(\"foo\").query) SELECT 0 AS \"foo\" FROM \"table\" Not working for ExpressionWrapper(~Q(pkin=[])) print(queryset.annotate(foo=ExpressionWrapper(~Q(pkin=[]), outputfield=BooleanField())).values(\"foo\").query) SELECT AS \"foo\" FROM \"table\""}
{"_id":"q-en-django__django-15240-e460451ac74e70b2eeddc28c48249dac7de238b4f51f17d85bdf52b10fa235d9","text":"Delete nonexistent migrations from djangomigrations table Description Django adds a djangomigrations table to the database which list all of the migrations that have been applied (and when). With the introduction of squashmigrations, it is possible for this table to contain a lot of old migrations that no longer exist. This can be problematic if naming duplication occurs: Example: I have an app with: myapp\/migrations\/ I squash and delete replaced migrations: myapp\/migrations\/ I create a new migration and use poor naming: myapp\/migrations\/ My new migration never runs because the djangomigrations table thinks it has already been applied. I propose truncation of the djangomigrations table so that it includes only migrations that actually exist in the django project. This could be done automatically (when executor runs, or inside the migrate mcommand). Or have its own mcommand that requires it be run manually. I prefer the automatic approach though. Pros: Cleans up old data that just bloats the database. Protects users from the trap mentioned above where a new migration is created with the same name as one that was applied in the past. Cons: A loss of historical information. Note: Need to be careful with implementation to avoid a possible new trap if a user squashes migrations and then proceeds to delete the replaced migrations before running the squashed migrations on their database -django will think squashed migrations haven't been applied and will attempt to reapply them. This can be remedied simply by not removing migrations mentioned in replaces lists of other migrations from djangomigrations (ie. we'd consider replaced migrations as still existing, even if their actual files have already been removed)."}
{"_id":"q-en-django__django-15248-a212de34660f049e2fae48d3ec32ed6b64777eef71b5fce19f8298991342f32d","text":"Add ability to know where an object was deleted from to pre\/post delete signals Description Since deleting objects deletes related objects, it would be good to know the origin of the delete. It lets the signal listeners know why an object is being delete. If it's deleted from a (), you know which instance initiated the delete and if it's from a () you know which queryset initiated the delete, with the models in that queryset. Using this you know if the instance is being deleted directly or because it's related to another instance."}
{"_id":"q-en-django__django-15252-eb6b1ccb036eac0f2a7b3db483ee68c4027646d7b75f55b54aea0809cc311804","text":"MigrationRecorder does not obey dbrouter allowmigrate rules Description Hi, We have a multi-db setup. We have one connection that is for the django project, and several connections that talk to other dbs for information (ie models with managed = False). Django should only create tables in the first connection, never in any of the other connections. We have a simple router that does the following: class Router(object): def allowmigrate(self, db, model): if db == 'default': return True return False Current Behaviour We run our functional tests and the migrate command is called against each connection when the test databases are created (see , setupdatabases, line 300-ish, which calls , createtestdb, line 377-ish) When this migrate runs, it tries to apply our migrations, which tries to record that a migration has been applied (see , applymigration, which has several calls to self.recorder.recordapplied). The first thing that recordapplied does is a call to self.ensureschema() (see , recordapplied, lien 66-ish). ensureschema checks to see if the Migration model is in the tables in the connection. If it does not find the table then it tries to create the table. I believe that this is incorrect behaviour when a dbrouter has been provided. If using the router above, my expectation would be that the table is not created on any connection other than the 'default' connection. Looking at the other methods on the MigrationRecorder, I would expect that there will be similar issues with appliedmigrations and record_unapplied."}
{"_id":"q-en-django__django-15268-4aa818cde8a9930c95302290d99f934832c058af2dd54c57aadc7dffec7f01b2","text":"Optimize multiple AlterFooTogether operations into one Description Hi, In we split the AlterFooTogether (AlterUniqueTogether and AlterIndexTogether) operations into two types of operations. First, a migration will have operations to remove constraints, and then other operations adds the new constraints. This allows field alterations to work as expected during in between operations. In some cases, this introduced two operations that can actually easily be reduced to one. See for instance the test case: ​ Example: operations = [ migrations.AlterUniqueTogether( name='mymodel', uniquetogether=set(), ), migrations.AlterIndexTogether( name='mymodel', indextogether=set(), ), migrations.AlterUniqueTogether( name='mymodel', uniquetogether={(\"col\",)}, ), migrations.AlterIndexTogether( name='mymodel', indextogether={(\"col\",)}, ), ] should be optimized to operations = [ migrations.AlterUniqueTogether( name='mymodel', uniquetogether={(\"col\",)}, ), migrations.AlterIndexTogether( name='mymodel', indextogether={(\"col\",)}, ), ] So that we don't do two operations on each constraint, but only one."}
{"_id":"q-en-django__django-15272-3197e982ac527286212fd761bdaac8ee1c3d8f74f40a412da5e624812d36c776","text":"Add a management command to optimize a migration Description (last modified by Raphael Gaschignard) Because the migration optimizer still has a bit of trouble with reducing operations, you often want to edit squashed migrations and re-run it through the optimizer. The attached patch contains an implementation of a management command, optimizemigration, that will help speed up this process. optimizemigration appname migrationname reads a single migration, pass it through the optimizer, and then rewrite the result to disk. Unlike squashmigrations, this ignores things like the squashed-ness of the migration."}
{"_id":"q-en-django__django-15277-0d56d1001330c9d8b4e05d968cce1dbd25b60153485cc065710820976a09f414","text":"Micro-optimisation for Value.resolveoutputfield (by modifying ) Description Currently, when you do something like annotate(x=Value('test')) that will eventually probably call down into Value.resolveoutputfield() and run the following code: if isinstance(, str): return fields.CharField() which is innocuous enough. However, CharField currently expects that self.maxlength is always a non null value of sensible data, and AFAIK this is caught for users at system-check time as a requirement for use. So what currently happens is that the CharField internally gets granted a MaxLengthValidator which cannot work and must be demonstrably extraneous (i.e. validators aren't used the outputfield, at least for Value) x = Value('test') y = x.resolveoutputfield() y.validators [ at 0x105e3d940] y.clean('1', modelinstance=None) in compare(self, a, b): TypeError: '' not supported between instances of 'int' and 'NoneType' Further compounding this is that MaxLengthValidator is decorated by (both directly and indirectly via BaseValidator ...?). So, baseline (as of ): In [1]: from import Value In [2]: x = Value('test') In [3]: %timeit x.resolveoutputfield() 1 µs ± 39.6 ns per loop (mean ± std. dev. of 7 runs, loops each) (Note: a previous run was faster at 7.6µs, so normal CPU workfload flux is in effect). We can see how much of the time is because of (​see my comment here on a PR about deconstructible being a source to potentially optimise away) by just commenting it out from both validator classes: In [1]: from import Value In [2]: x = Value('test') In [3]: %timeit x.resolveoutputfield() 96 µs ± 130 ns per loop (mean ± std. dev. of 7 runs, loops each) But ignoring the class instantiation altogether is faster, easier and more correct at this juncture: In [1]: from import Value In [2]: x = Value('test') In [3]: %timeit x.resolveoutputfield() 86 µs ± 45.4 ns per loop (mean ± std. dev. of 7 runs, loops each) So roughly a 2µs improvement. How do we get to that? Change the to: if self.maxlength is not None: (validators.MaxLengthValidator(self.max_length)) which incidentally and happily is the same process taken by for precedent. I have a branch locally with this change, and all existing tests currently pass. I'll push it to CI once I get a ticket number out of this submission, and see if it causes any issues elsewhere, and we can decide if it can be accepted from there."}
{"_id":"q-en-django__django-15278-b9895825830b69efb925f35e64a04fa1fca8bc6c0b06c5a3ae7c0fd1bb99b4e6","text":"Adding nullable OneToOneField crashes on SQLite. Description This new sqlite3 error has cropped up between building django-oauth-toolkit between Django 4.0 and main branch for migrations.AddField of a OneToOneField (see ​): self = object at 0x10b8038b0query = 'ALTER TABLE \"oauth2provideraccesstoken\" ADD COLUMN \"sourcerefreshtokenid\" bigint NULL UNIQUE REFERENCES \"oauth2providerrefreshtoken\" (\"id\") DEFERRABLE INITIALLY DEFERRED' params = [] def execute(self, query, params=None): if params is None: return Database.Cursor.execute(self, query) query = self.convertquery(query) E Cannot add a UNIQUE column Here's the relevant migration snippet: migrations.AddField( modelname='AccessToken', name='sourcerefreshtoken', field=models.OneToOneField(blank=True, null=True, , to=oauth2settings.REFRESHTOKENMODEL, relatedname=\"refreshedaccess_token\"), ), I see there have been a lot of sqlite3 changes in since the 4.0 release...."}
{"_id":"q-en-django__django-15280-9c7256b211874d4db290c7b3885d672128ff734cc5839a143f119ca72ec2cb78","text":"Deferred fields incorrect when following prefetches back to the \"parent\" object Description Given the following models: class User(models.Model): email = models.EmailField() kind = models.CharField( maxlength=10, choices=[(\"ADMIN\", \"Admin\"), (\"REGULAR\", \"Regular\")] ) class Profile(models.Model): fullname = models.CharField(maxlength=255) user = models.OneToOneField(User, ondelete=models.CASCADE) I'd expect the following test case to pass: def testonlyrelatedqueryset(self): user = ( email=\"\", kind=\"ADMIN\", ) (user=user, fullname=\"Test Tester\") queryset = (\"email\").prefetchrelated( Prefetch( \"profile\", queryset=Profile.objects.prefetchrelated( Prefetch(\"user\", (\"kind\")) ), ) ) with self.assertNumQueries(3): user = () with self.assertNumQueries(0): self.assertEqual(, \"ADMIN\") The second assertNumQueries actually fails with: AssertionError: 1 != 0 : 1 queries executed, 0 expected Captured queries were: \"testsuser\".\"id\", \"testsuser\".\"kind\" FROM \"testsuser\" WHERE \"testsuser\".\"id\" = 1 This is exactly the query I'd expect to see if kind on the inner User queryset had been deferred, which it hasn't. The three queries executed when iterating the main queryset (ie when executing user = ()) look correct: \"testsuser\".\"id\", \"testsuser\".\"email\" FROM \"testsuser\" ORDER BY \"testsuser\".\"id\" ASC LIMIT 1 \"testsprofile\".\"id\", \"testsprofile\".\"fullname\", \"testsprofile\".\"userid\" FROM \"testsprofile\" WHERE \"testsprofile\".\"userid\" IN (1) \"testsuser\".\"id\", \"testsuser\".\"kind\" FROM \"testsuser\" WHERE \"testsuser\".\"id\" IN (1) Printing () returns {'kind'}. It looks to me like Django is correctly evaluating the set of deferred fields when executing the \"inner\" User queryset, but somehow the instances are inheriting the set of fields they \"think\" have been deferred from the outer User queryset, so when the attribute is accessed it causes a database query to be executed. It appears that this also happens if the relationship between Profile and User is a ForeignKey rather than a OneToOneField (in that case, a query is executed when accessing ()[0]). I'm happy to attempt to tackle this if someone can (a) confirm it's actually a bug and (b) point me in the right direction! Thanks :)"}
{"_id":"q-en-django__django-15292-c4f40a164d4b0d1627d5709887537d663a08524b0d51c93e09a93c9b76483c46","text":"Technical 404 debug page reported incorrect view name for CBVs. Description Technical 404 debug page reported incorrect view name for CBVs, e.g. Raised by: <class ''instead of Raised by: Regression in . Thanks Keryn Knight for the report."}
{"_id":"q-en-django__django-15297-309603e3b82b05ff8df416710271ab7890a07141ffb0f0de5a706a11cb0e4896","text":"ResolverMatch repr is incorrect for Class Based Views Description The following test applies cleanly to 3.2.9, and AFAIK would apply roughly correctly all the way back to when CBVs were introduced (I can't easily test without going back to a super old Python and finding the test cases, which have moved around): \"\"\" add to class: \"\"\" def testclassbasedrepr(self): self.assertEqual( repr(resolve('\/redirect\/')), \"ResolverMatch(, \" \"args=(), kwargs={}, urlname=None, appnames=[], \" \"namespaces=[], route=redirect\/)\", ) The funcpath as AFAIK always been a representation to the fully qualified dotted callable where possible, that is for a CBV it's the CBV module + the class name. As of 4.0, the funcpath has become , I believe because of removing the use of updatewrapper and intentionally not setting the name and qualname in favour using the viewclass attribute, as per the comment viewclass should be used to robustly determine the name of the view (see ​Pull Request) Unfortunately I think that means the detection of class based views in ResolverMatch no longer works correctly, and this can probably only be resolved by making ResolverMatch CBV aware again by embedding detection of viewclass therein. Noted it as a question in ​this PR for ticket , but hoisting it here properly to be considered separately. The fix appears to ostensibly the same as for (see ​PR) class ResolverMatch: def init(...): # ... if hasattr(func, 'viewclass'): func = func.viewclass if not hasattr(func, 'name'): # ... I have a branch which I shall push shortly to confirm it works generally."}
{"_id":"q-en-django__django-15315-84d6268f9c7a9d45fbeed24bac8164f31ee2f77a5d461d7ea1f27da367639277","text":"Model () should be immutable. Description changes value when a field is assigned to a model class. This code crashes with an AssertionError: from import models f = models.CharField(max_length=200) d = {f: 1} class Book(models.Model): title = f assert f in d The bug was introduced in . It's unlikely to have been encountered because there are few use cases to put a field in a dict before it's assigned to a model class. But I found a reason to do so whilst implementing and the behaviour had me stumped for a little. IMO we can revert the hash change from . Objects with the same hash are still checked for equality, which was fixed in that ticket. But it's bad if an object's hash changes, since it breaks its use in dicts."}
{"_id":"q-en-django__django-15316-2f1f9597ef7bf56cbf02699931a4c87281cd72726928fbcb7de9a2a11b4a59eb","text":"simplifyregex() doesn't handle non-capturing groups Description (last modified by Mariusz Felisiak) While using Django REST Framework's Schema generator, I found out they're using simplifyregex(); however, current version has a few shortcomings, namely non-capturing groups are broken. simplifyregex() doesn't handle non-capturing groups Description (last modified by Mariusz Felisiak) While using Django REST Framework's Schema generator, I found out they're using simplifyregex(); however, current version has a few shortcomings, namely non-capturing groups are broken."}
{"_id":"q-en-django__django-15318-2b49820436a858cb0098e5c778d215fa0b96b6312da7e513d07dd590cb9bf272","text":"Unify reverse foreign key and m2m unsaved model querying Description Currently when querying unsaved reverse relations the behavior differs. Using model: class Foo(models.Model): fk = models.ForeignKey('self', relatedname='fkrev') m2m = models.ManyToManyField('self') and test case: print(Foo()()) print(Foo()()) We get [] from the first filter, but an error ValueError: \"<Foo: Foo object\" needs to have a value for field \"fromfoo\" before this many-to-many relationship can be used. from the second filter. So, m2m fields can't be filtered if the object isn't saved, but reverse fk fields can be filtered. There is a (slightly stale) patch for which makes fk fields and m2m fields work consistently. The changes in behavior are: Nullable many-to-many and foreign key relations will return an empty queryset when the relation field is null. For many-to-many this was previously an error (no change for foreign keys). Trying to add objects to a foreign key relation when the relation field is null is now an error (no change for m2m). Trying to use a relation of any kind when the object isn't saved is now an error (no change for m2m). I think we can squeeze these changes in as bug-fixes. These are slight backwards compatibility changes, but to me it seems that almost always the changes will be visible only in code that isn't working as intended. If these are seen as something likely breaking working code, then I don't see any way to make the APIs consistent. The patch is available from: ​"}
{"_id":"q-en-django__django-15320-4cab07e6d0f6d7f0c8b6bcbc29b849e375a90eb7063c77f9e7e89dd4c84f3fc0","text":"() generates invalid SQL. Description (last modified by M1ha Shvn) Since ​this commit (...) method returns incorrect SQL removing first and last symbols instead of absent breakets. Adding Subquery().query.subquery = True attribute fixes the problem. From my point of view, it should be set in Subquery constructor. from import connection from import App q = Subquery(()) print(str(q.query)) print(q.assql(q.query.getcompiler('default'), connection)) q.query.subquery = True print(q.assql(q.query.getcompiler('default'), connection)) ('(SELECT \"appsapp\".\"id\", \"appsapp\".\"name\" FROM \"apps_app\")', ())"}
{"_id":"q-en-django__django-15324-e2fc72e0683f34942b151bd525da11ad798f394cf66ecbf7ad9fedc95aa1ff87","text":"File upload crash when a file extension contains null characters. Description A 2.5M file uploaded with a raw null byte anyplace after the . in its filename means that Django attempts to create a tempfile with that same \"extension,\" which errors out with ValueError: embedded null byte. It's almost certainly a violation of RFC to have a filename with a raw null byte in it, but it shouldn't result in a 500 when parsing the form. Here's code to generate a bad request: #!\/usr\/bin\/env python3 import io import requests contents = io.StringIO(\".\" (1024 1024 * 3)) files = {\"docfile\": (b\"!\", contents, \"text\/plain\")} req = requests.Request(\"POST\", \"http:\/\/localhost:8000\/\", files=files, data={}) prepared = req.prepare() body = assert isinstance(body, bytes) = body.replace(b\"!\", b\"\\x00\") requests.Session().send(prepared) produces an error with the view: from django import forms from import HttpResponseRedirect from django.shortcuts import render from import csrfexempt class UploadFileForm(forms.Form): docfile = forms.FileField() def index(request): if == 'POST': form = UploadFileForm(request.POST, request.FILES) if form.is_valid(): print(repr(request.FILES['docfile'])) return HttpResponseRedirect('\/') else: print(\"Not valid!\") return HttpResponseRedirect('\/') else: form = UploadFileForm() return render(request, '', {'form': form}) I'm not sure what the goal is of preserving the \"extension\" of the uploaded file in the tempfile that is made; if that's important enough a behaviour to keep, some amount of escaping on the parsed-out extension may be necessary."}
{"_id":"q-en-django__django-15334-466c4a46fe4e548edadaf1afbbdb6f04efca26ad57ec689ce44f5e010be757cf","text":"Support prefetchrelated() with Queryset.iterator() Description (last modified by Asif Saifuddin Auvi) I was surprised when I found out that prefetchrelated calls are ignored when using Queryset.iterator. I noticed in the docs here ​ that it is because the \"these two optimizations do not make sense together.\" That may have been true in the past, but it is definitely not the case now. The iterator allows chunking (by default 2000) and it would be very helpful to prefetch related for each chunk."}
{"_id":"q-en-django__django-15342-98d5e2be2ff09a0cc58766fd21ebc32988c7b32795a3979841cc344e9b97488a","text":"Migration autodetector crashes on models with field named order, but not using orderwithrespectto. Description (last modified by Fabian Büchler) The commit ​ introduced a new function ModelState.getfield in This converts the field name order to the one defined in options['orderwithrespectto'] automatically, which fails if the model has a field order but isn't using Meta.orderwithrespectto. That is the case for models generated by django-simple-history (​) for models that are originally using Meta.orderwithrespectto: the resulting historical records model has only order but is not using the Meta option. This shows when running migrate or makemigrations: $ makemigrations --dry-run Waiting for port 'mysql:3306' timeout 1s (attempt 1\/60) Port 'mysql:3306' is open Traceback (most recent call last): File \"\", line 42, in <modulemain() File \"\", line 36, in main executefromcommandline() File \"\/usr\/local\/lib\/python3.8\/site-\", line 425, in executefromcommandline utility.execute() File \"\/usr\/local\/lib\/python3.8\/site-\", line 419, in execute self.fetchcommand(subcommand).runfromargv() File \"\/usr\/local\/lib\/python3.8\/site-\", line 373, in runfromargv self.execute(args, cmdoptions) File \"\/usr\/local\/lib\/python3.8\/site-\", line 417, in execute output = (args, options) File \"\/usr\/local\/lib\/python3.8\/site-\", line 90, in wrapped res = handlefunc(args, *kwargs) File \"\/usr\/local\/lib\/python3.8\/site-\", line 172, in handle changes = autodetector.changes( File \"\/usr\/local\/lib\/python3.8\/site-\", line 43, in changes changes = self.detectchanges(convertapps, graph) File \"\/usr\/local\/lib\/python3.8\/site-\", line 189, in detectchanges self.generatealteredfields() File \"\/usr\/local\/lib\/python3.8\/site-\", line 928, in generatealteredfields oldfield = [applabel, oldmodelname].getfield(oldfieldname) File \"\/usr\/local\/lib\/python3.8\/site-\", line 689, in getfield self.options['orderwithrespectto'] KeyError: 'orderwithrespectto' I believe this could be solved using a bit more defensive code, like: def getfield(self, fieldname): if fieldname == 'order' and 'orderwithrespectto' in self.options: fieldname = self.options['orderwithrespectto'] return [fieldname]"}
{"_id":"q-en-django__django-15347-1ec14526f28d4fb5ebfc530f1adad6935b2de3a9427a85d841275fc4be4ed6a7","text":"Messages framework incorrectly serializes\/deserializes extratags when it's an empty string Description When a message is serialised and then deserialised with any of the built in storage backends, then extratags==\"\" is converted to extratags==None. This is because MessageEncoder checks for the truthyness of extratags rather than checking it is not None. To replicate this bug from import settings settings.configure() # Just to allow the following import from import Message from import MessageEncoder, MessageDecoder originalmessage = Message(10, \"Here is a message\", extratags=\"\") encodedmessage = MessageEncoder().encode(originalmessage) decodedmessage = MessageDecoder().decode(encodedmessage) originalmessage.extratags == \"\" True decodedmessage.extratags is None True Effect of the bug in application behaviour This error occurred in the wild with a template tag similar to the following: {% if x not in message.extratags %} When the message was displayed as part of a redirect, it had been serialised and deserialized which meant that extratags was None instead of the empty string. This caused an error. It's important to note that this bug affects all of the standard API (, etc. all have a default value of extra_tags equal to \"\")."}
{"_id":"q-en-django__django-15352-48d18ffee209e0ace947be20ce4e785df897027114b6c4cdc8a9dfdd404b19b8","text":"Session ID should be cleansed from error reporting Description the session ID should be cleansed when reporting errors, just like other credentials. A patch is available at ​ See also and ​ A quick github search yielded multiple occasions where session IDs ended up in public bug reports: ​ ​ ​ ​ I am sure you could find many more. This could potentially be exploited by automatically searching for such requests and hijacking the associated accounts."}
{"_id":"q-en-django__django-15368-dcdb7d2827966344415f599a7087828ae3b02f77165194a14a3cefaccf865cb9","text":"bulkupdate() does not work with plain F('...') expressions. Description Repro: assign plain F(...) to some model instance field save with bulkupdate Example: Code highlighting: from import SelfRef o = ().first() o.c8 = F('name') # model has char fields 'c8' and 'name' SelfRef.objects.bulkupdate([o], ['c8']) 1 o.refreshfromdb() o.c8 'F(name)' from import connection connection.queries[-2] {'sql': 'UPDATE \"exampleappselfref\" SET \"c8\" = CASE WHEN (\"exampleappselfref\".\"id\" = ) THEN \\'F(name)\\' ELSE NULL END WHERE \"exampleappselfref\".\"id\" IN ()', 'time': '0.001'} The created SQL contains the string repr of F(), instead of resolving to the column name. Looking at the source code, the culprit seems to be a too narrow type check in ​ It works, if the type check gets replaced by one of these: Code highlighting: if not hasattr(attr, 'resolve_expression'): ... if not isinstance(attr, (Expression, F)): ..."}
{"_id":"q-en-django__django-15370-c3766430780632be6a9fdf81e09eed19c7766cd31eb61cbe026cb9b8c82d80d4","text":"Introduce empty slots protocol for SafeString & SafeData Description (last modified by Keryn Knight) This is a case-by-case proposal ultimately referencing Because SafeString is used a lot and is otherwise supposed to be treatable as a untainted str we should be able to (AFAIK) update it + it's inheritance chain to use slots = () whilst still allowing custom subclasses of either to add additional attributes. By defining slots as empty on SafeString (and SafeData) we'd avoid creation of a dict on the instance, which mirrors the str() behaviour. According to pympler, currently in Python 3.10 using the following back of the napkins strings: In [4]: s = \"test\" # this might be interned, as a short string? In [5]: s2 = \"test\" 100 In [6]: s3 = SafeString(\"test\") In [7]: s4 = SafeString(\"test\" 100) we get: In [8]: asizeof(s) # str Out[8]: 56 In [9]: asizeof(s2) # str Out[9]: 456 In [10]: asizeof(s3) # SafeString Out[10]: 208 In [11]: asizeof(s4) # SafeString Out[11]: 608 But if we swap out the implementation to be slots'd, it looks more like: In [8]: asizeof(s) # str Out[8]: 56 In [9]: asizeof(s2) # str Out[9]: 456 In [10]: asizeof(s3) # SafeString Out[10]: 104 In [11]: asizeof(s4) # SafeString Out[11]: 504 So we're \"saving\" 104 bytes per SafeString created, by the look of it. I presume it to be some fun implementation detail of something somewhere that it is allegedly accounting for more than 64 bytes, which is the asizeof({}) A quick and dirty check over the test suite suggests that for me locally, running tests in 512.912s accounted for 949.0 MB of SafeStrings, checked by just incrementing a global integer of bytes (using and --parallel=1) and piping that to filesizeformat, so y'know, room for error. After the patch, the same tests accounted for 779.4 MB of SafeString, \"saving\" 170 MB overall. The only functionality this would preclude -- as far as I know -- is no longer being able to bind arbitrary values to an instance like so: s = SafeString('test') s.test = 1 which would raise AttributeError if slots were , just like trying to assign attributes to str() directly does. I don't believe this will have any marked performance change, as neither SafeString nor SafeData actually have any extra attributes, only methods. I have a branch which implements this, and tests pass for me locally."}
{"_id":"q-en-django__django-15375-1d15a1e5abad33796e673f6abf266f524b86aa3b8c3ffa251e79430549293ba8","text":"aggregate() with 'default' after annotate() crashes. Description I saw this on a PostgreSQL project and reproduced it with SQLite. Django 4.0.1. Annotate (anything) then aggregate works fine: $ shell Python 3.10.2 (main, Jan 21 2022, 19:45:54) [Clang 13.0.0 (clang-1300.0.29.30)] Type 'copyright', 'credits' or 'license' for more information IPython 7.30.1 -- An enhanced Interactive Python. Type '?' for help. In [1]: from import In [2]: from import In [3]: from import * In [4]: () Out[4]: 95 In [5]: Book.objects.annotate(idx=F(\"id\")).aggregate(Sum(\"id\")) Out[5]: {'idsum': 4560} But add the aggregate classes’ default argument (new in 4.0), and it breaks: In [6]: Book.objects.annotate(idx=F(\"id\")).aggregate(Sum(\"id\", default=0)) OperationalError Traceback (most recent call last) ... OperationalError: near \"FROM\": syntax error The generated SQL: In [7]: %debug 414 return Database.Cursor.execute(self, query) 415 query = self.convertquery(query) --416 return Database.Cursor.execute(self, query, params) 417 418 def executemany(self, query, paramlist): ipdbquery 'SELECT FROM (SELECT \"corebook\".\"id\" AS \"idx\", COALESCE(SUM(\"corebook\".\"id\"), ?) AS \"idsum\" FROM \"core_book\") subquery' ipdbparams (0,) ipdbThe “long form” using Coalesce works: In [8]: Book.objects.annotate(idx=F(\"id\")).aggregate(x=Coalesce(Sum(\"id\"), 0)) Out[8]: {'x': 4560}"}
{"_id":"q-en-django__django-15380-46e446cab9c6e8ce9dd53c5ab70e7436bb88ed2113a7f7521b0c017bb8ff33ef","text":"Migration autodetector crashes when renaming a model and field. Description Migration autodetector crashes when renaming a model and field in a single step: $ python makemigrations Did you rename the testone.MyModel model to MyModel2? [y\/N] y Traceback (most recent call last): File \"\", line 22, in <modulemain() File \"\", line 18, in main executefromcommandline() File \"\", line 419, in executefromcommandline utility.execute() File \"\", line 413, in execute self.fetchcommand(subcommand).runfromargv() File \"\", line 354, in runfromargv self.execute(args, cmdoptions) File \"\", line 398, in execute output = (args, options) File \"\", line 89, in wrapped res = handlefunc(args, *kwargs) File \"\", line 172, in handle changes = autodetector.changes( File \"\", line 43, in changes changes = self.detectchanges(convertapps, graph) File \"\", line 182, in detectchanges self.generaterenamedfields() File \"\", line 823, in generaterenamedfields newmodelstate = [applabel, oldmodelname] KeyError: ('testone', 'mymodel') Reported by HoskeOwl. Regression in ."}
{"_id":"q-en-django__django-15382-a65a9c17068bd4dfbd8629a85d930e30e0a2dcfcd3db5a69e5658cd4a9f2fb44","text":"filter on exists-subquery with empty queryset removes whole WHERE block Description (last modified by Tobias Bengfort) qs = (~models.Exists(()), name='test') qs <QuerySet []print() EmptyResultSet With django-debug-toolbar I can still see the query, but there WHERE block is missing completely. This seems to be very similar to ."}
{"_id":"q-en-django__django-15388-ee34a523faeefe404ba04c9401f3da5664ac822718f4d99e778cc45339f36ee1","text":"Dev Server fails to restart after adding BASEDIR to TEMPLATES[0]['DIRS'] in settings Description Repro steps: $ pip install -U django $ django-admin startproject <nameOpen , copy the BASEDIR variable from line 16 and paste it into the empty DIRS list on line 57 $ runserver Back in your IDE, save a file and watch the dev server NOT restart. Back in , remove BASE_DIR from the templates DIRS list. Manually CTRL-C your dev server (as it won't restart on its own when you save), restart the dev server. Now return to your file, re-save it, and notice the development server once again detects changes and restarts. This bug prevents the dev server from restarting no matter where you make changes - it is not just scoped to edits to"}
{"_id":"q-en-django__django-15400-4937e8eca537e3e1269be68e0a1f0d4f04fc6ae3f0f2658c11c13594fb4a2cdb","text":"SimpleLazyObject doesn't implement radd Description Technically, there's a whole bunch of magic methods it doesn't implement, compared to a complete proxy implementation, like that of wrapt.ObjectProxy, but radd being missing is the one that's biting me at the moment. As far as I can tell, the implementation can't just be radd = newmethodproxy() because that doesn't exist, which is rubbish. radd = newmethodproxy(operator.attrgetter(\"radd\")) also won't work because types may not have that attr, and attrgetter doesn't supress the exception (correctly) The minimal implementation I've found that works for me is: def radd(self, other): if self.wrapped is empty: () return other + self._wrapped"}
{"_id":"q-en-django__django-15401-924c94ed5415400179776b6b7c3f80e8c44537da57bcd3da31bd84f24d43c226","text":"Deprecate passing unsaved objects to related filters. Description Consider this filter: (relatedobj=bar) Where 'bar' is an unsaved object instance. In Django 1.11, this would always return an empty QuerySet (since no Foo object is related to unsaved 'bar'). In Django 2.0 through 2.2, this is equivalent to doing (which can return a non-empty QuerySet): (relatedobj=None) I found a somewhat related issue that touches on this subject: My questions: What is the intended behaviour? In the aforementioned issue Simon Charette suggests that unsaved objects should be prevented from being used in related filters. I agree with that. Is this documented anywhere? I couldn't find anything. At the very least this should be documented somewhere."}
{"_id":"q-en-django__django-15413-32df680f146d5efae234b0a489b81fa720e3a3f1cb7737c2abe79763ddec784d","text":"\"Synchronous middleware ... adapted\" is not logged for synchronous middleware Description This is a bug in log messages that contradicts the guide in a warning in the docs; the feature works. The guide at ​, emphasis mine: Warning You will only get the benefits of a fully-asynchronous request stack if you have no synchronous middleware loaded into your site. If there is a piece of synchronous middleware, then Django must use a thread per request to safely emulate a synchronous environment for it. Middleware can be built to support both sync and async contexts. Some of Django’s middleware is built like this, but not all. To see what middleware Django has to adapt, you can turn on debug logging for the django.request logger and look for log messages about “Synchronous middleware … adapted”. The test for \"Synchronous middleware ... adapted\" is instead testing an async middleware over a sync method: ​ 'middlewareexceptions.middleware.asyncpaymentmiddleware', ]) def testasyncmiddleware(self): with self.assertLogs('django.request', 'DEBUG') as cm: response = ('\/middlewareexceptions\/view\/') self.assertEqual(response.statuscode, 402) self.assertEqual( cm.records[0].getMessage(), \"Synchronous middleware \" \"middlewareexceptions.middleware.asyncpaymentmiddleware \" \"adapted.\", ) About the existing implementation: BaseHandler.loadmiddleware passes the middleware name as name to BaseHandler.adaptmethodmode: ​ adaptedhandler = self.adaptmethodmode( middlewareisasync, handler, handlerisasync, debug=settings.DEBUG, name='middleware %s' % middlewarepath, ) BaseHandler.adaptmethodmode adapts the method and treats name as the method name rather than the middleware name; when the middleware name is used, it implies a method has been adapted for the middleware, not that the middleware was adapted: if debug and not name: name = name or 'method %s()' % method.qualname if isasync: if not methodisasync: if debug: ('Synchronous %s adapted.', name) return synctoasync(method, threadsensitive=True) elif methodisasync: if debug: ('Asynchronous %s adapted.', name) return asynctosync(method) Proposed fix: Handle middleware name and method name separately within BaseHandler.adaptmethodmode: def adaptmethodmode( self, isasync, method, methodisasync=None, debug=False, name=None, ): \"\"\" Adapt a method to be in the correct \"mode\": If isasync is False: Synchronous methods are left alone Asynchronous methods are wrapped with asynctosync If isasync is True: Synchronous methods are wrapped with synctoasync() Asynchronous methods are left alone \"\"\" methodname = None if methodisasync is None: methodisasync = asyncio.iscoroutinefunction(method) if debug and not name: name = name or 'method %s()' % method.qualname methodname = 'method %s()' % method.qualname if isasync: if not methodisasync: if debug: ('Synchronous %s adapted.', name) if name: ('Asynchronous %s adapted.', name) else: ('Synchronous %s adapted.', methodname) return synctoasync(method, threadsensitive=True) elif methodisasync: if debug: ('Asynchronous %s adapted.', name) if name: ('Synchronous %s adapted.', name) else: ('Asynchronous %s adapted.', methodname) return asynctosync(method)"}
{"_id":"q-en-django__django-15414-3a8d16fe3da3d095ab73bab2cbc56d5d5152d745b3fd8aacdd74c12c75548b2b","text":"QuerySet.bulkupdate() should perform atomic writes against routed dbforwrite. Description Discovered in but the summary is that bulkupdate doesn't set self.forwrite = True prior to accessing so the latter is actually dbforread and that includes the block creation."}
{"_id":"q-en-django__django-15421-6de47f03221f582a9c385444056f77f6c3bc6dc0eb362cafacb081f08fa2f946","text":"Allow parallel test runner to work with Windows\/macOS process start method. Description (last modified by Brandon Navra) Python 3.8 on MacOS has changed the default start method for the multiprocessing module from fork to spawn: ​ When running tests with the --parallel flag, this causes the worker processes to fail with Apps aren't loaded yet. as they no longer have a copy of the parent memory state. It can also cause the workers to fail to find the cloned dbs ( {{ FATAL: database \"xxx1\" does not exist}} ) as the db test prefix is missing. I have attached a patch which changes (the worker initialiser for ParallelTestSuite) to run () and set the db name to one with the test prefix."}
{"_id":"q-en-django__django-15423-5ecbcac1ac2e51d427640ed34dd23dbd797b2cb60b28b28a103cb90675cb904f","text":"LazyObject defines attribute that don't exist on wrapped object Description LazyObject defines magic methods (getitem, iter) which may be missing from the wrapped object. This leads to the following errors: somevariable = if hasattr(somevariable, \"getitem\"): foo = somevariable[\"foo\"] # raises TypeError: 'User' object has no attribute 'getitem' if hasattr(somevariable, \"iter\"): for item in some_variable: # raises TypeError: 'User' object is not iterable"}
{"_id":"q-en-django__django-15433-b844c3e27ea910196a4db9ce4c876bc94acd5cfa193694300088e268156e5d9f","text":"ManyToManyField to lowercased swappable setting causes generating infinite migrations. Description If I create a custom user model that extends AbstractUser and then try to add a ManyToManyField that references this custom User model, django keeps making the same AlterField migration over and over again unnecessarily. I've attached a Git repository that I used to reproduce this issue with very simple code. You can see the two erroneous migrations after the initial migration. If I use the built in user, there is no issue. It seems to appear once I extend AbstractUser. Git repository: ​ Here is the accounts app code. from import settings from import AbstractUser from import models class User(AbstractUser): pass class Test(models.Model): members = models.ManyToManyField(settings.AUTHUSERMODEL) Here is one of the erroneous migrations. Notice it depends on 0002altertestmembers which is an erroneous migrations file that looks just like this one. from import settings from import migrations, models class Migration(migrations.Migration): dependencies = [ ('accounts', '0002altertestmembers'), ] operations = [ migrations.AlterField( modelname='test', name='members', field=models.ManyToManyField(to=settings.AUTHUSER_MODEL), ), ]"}
{"_id":"q-en-django__django-15438-782f885a482f352efc6d2b1790eae5c6ad15a07d91d192a54af93b4a23318713","text":"Fallback to a more generic language variant for Select2 translations. Description When using for example the setting LANGUAGECODE=\"de-ch\", the autocompletefield widget will be in English as select2 does not provide the specific regional translation for \"de-ch\". However the existing translation language \"de\" would be a much better approximation. I suggest using the language without region from the LANGUAGE_CODE in case an exact match can not be found. This is not a duplicate of ."}
{"_id":"q-en-django__django-15442-2913a59fde5c3b29a48c7ddc06378c2fc1fd7f9b88a18447a3011ec277cd73c5","text":"forces evaluation of lazy objects Description Consider the following example: from import marksafe from import activate, ugettextlazy as s = marksafe((\"username\")) tpl = Template('{{ s }}') activate('fr') print((Context({'s': s}))) I would expect this to output nom d'utilisateur (which is the french translation of username) but what happens instead is that it outputs username. The reason for this is that marksafe will force the evaluation of the lazy string provided by ugettextlazy when it's called. Unfortunately, the solution to this it trickier than simply wrapping marksafe with , because marksafe can operate both on bytes and text (and allowlazy needs to know the type of object return by the wrapped function). I wrote some tests and a proposed solution on my branch: ​"}
{"_id":"q-en-django__django-15467-13a6eb137eed891702cf32d8082a91ad090af3c34ee32b20ef98f146e5743cd0","text":"ModelAdmin with defined radiofields override emptylabel Description ModelAdmin drops my \"emptylabel\" and set \"defaultemptylabel\". For example: class MyModelAdmin(ModelAdmin): radiofields = 'myfield', def formfieldforforeignkey(self, dbfield, args, kwargs): if == 'myfield': kwargs['emptylabel'] = \"I WANT TO SET MY OWN EMPTY LABEL\" return super().formfieldforforeignkey(dbfield, args, kwargs) You get never the \"I WANT TO SET MY OWN EMPTY LABEL\" How to fix it: In django\\contrib\\admin\\, row 234: kwargs['emptylabel'] = ('None') if else None Should be changed on: kwargs['emptylabel'] = (('emptylabel') or ('None')) if else None"}
{"_id":"q-en-django__django-15474-22856481b32f0294658cd74feb9709d04267b965472c75e0f14d417328f506a3","text":"Unapplying ContentType migration 0002 does not populate legacy name field on non-default database Description Given a database with the alias other that has all contenttypes migrations applied, roll the state back to initial: $ migrate --database=other contenttypes 0001 All ContentType rows in the other database will be null instead of their intended value."}
{"_id":"q-en-django__django-15481-71dbcf2618ed79cebea0476f0cf4c0ee5c142b12f13e300e5cb4ed9ded1c636b","text":"setcookie and setsignedcookie should accept timedelta object for maxage argument Description This already works for getsignedcookie: request.getsignedcookie(\"mykey\", maxage=timedelta(days=3)) This is due to the underlying behaviour of TimestampSigner, which was fixed to do this in . But for setcookie and setsignedcookie it accepts only a number: response = HttpResponse() response.setcookie(\"mykey\", maxage=timedelta(days=3)) TypeError: int() argument must be a string, a bytes-like object or a number, not 'datetime.timedelta'"}
{"_id":"q-en-django__django-15483-3621d4be58c260282b5562eacc234c605927479ffbe97a8dfc8226bbafb26675","text":"AppAdmin class for customizing app listing in admin index Description See ​ for details. This supplements the app directive. As discussed with brosner and jkocherhans in #django-dev: <brosnerit looks reasonable, but haven't spent much time thinking about it <jkocherhansmrts: I think this is clearly backwards incompatible with the current nfa api and has to go in pre 1.0 if it goes in at all <jkocherhansI'm a big -1 on the order attribute and -0 on models (maybe just a different syntax), but the other stuff seems reasonable <mrtsjkocherhans: what's wrong with ordering? <jkocherhansit just feels like the wrong place to specify it <jkocherhansit's a global issue, and an issue any particular app should handle <mrtsmy use case: I have a lot of functionality exposed to somewhat dumb users <mrtsand they have trouble finding the right bits in the admin interface ordering is only used in context of admin index I would like to put the important apps to top and collapse the rest <jkocherhansexactly. what should 3rd party apps put there? therein lies my objection. <mrtswell, I'd say decouple admin from models (as nfa already does) and don't specify any admin options at all -- users are free to customize things with AppAdmin <jkocherhansI guess not if using a AppAdmin class is optional. I was originally thinking it would replace model registration with an admin site. <mrtsjkocherhans: yeah, that's what I kinda meant... it looks more coherent this way jkocherhans: and it may solve some of the issues register() currently has <jkocherhansmrts: I'm gonna have to let it sit for awhile. I'm trying to think of what else an AdminApp class would do besides being a coathanger for a few attributes, nothing is coming to mind. <mrtsjkocherhans: but jezdez has a point -- it would also provide easy bridging for app instances Example syntax follows. class BarModelAdmin(admin.ModelAdmin): description = 'A bar is a bar is a bar' ... class FooAppAdmin(admin.AppAdmin): app = settings.INSTALLED_APPS[0] name = \"Foo\" # overrides app().name description = \"An application that does foo\" style = {'classes' : ('collapse',)} order = 1 models = ( # model order in this list determines their display order in app block (BarModel, BarModelAdmin), (BazModel, None), # use default ModelAdmin, don't show description ) (FooAppAdmin) # no need for the tedious for model in [A, B, C, D]: (model)"}
{"_id":"q-en-django__django-15491-4dee5db7f97a32c6071c45f567e84cca2f52d6729a2346a29a0d7b191f563d2e","text":"Builtin csrffailure() view uses wrong charset Description When Django detects wrong CSRF token, it shows an error using view That file ends with return HttpResponseForbidden(t.render(c), contenttype=\"text\/html;\") When the template (CSRFFAILURETEMPLATENAME) is written using UTF-8, it is rendered incorrectly. I suggest changing that line to return HttpResponseForbidden(t.render(c), contenttype=\"text\/html;\"+ f\" charset={settings.DEFAULTCHARSET};\") or perhaps leaving out the content_type entirely. Currently I'm using a workaround, by adding <meta http-equiv=\"Content-type\" content=\"text\/html; charset=utf-8\" \/to the template's HEAD, but it seems to me that the suggested fix is a better solution."}
{"_id":"q-en-django__django-15492-cca86cd7f37365fb2212ad3e0daa47896ea55518f175c6e4011c8c2056cf4e12","text":"Allow syncing user attributes on every authentication with RemoteUserBackend. Description If using a custom RemoteUserBackend, it is possible to want to synchronize any changes from the remote system back into the Django user records whenever authentication happens. Currently, if any user attributes change in the remote system there is no easy way to reflect these changes back into the users in the Django system. The goal of this feature is to introduce a new method in the class called synchronizeuser with a method signature equal to that of configureuser (which it complements) that will be called in the authenticate method of said class right after fetching the user from the database (if any), regardless of whether the user was unknown and created or otherwise. Implementors can then override this method and implement data synchronization between the remote user and the Django user that will be applied on every user authentication attempt."}
{"_id":"q-en-django__django-15497-c4f8080e1483befcad59cd7e58e8fcacc64dbf0d17903a3bc0b8faff8cbf4a55","text":"Add support for multiple values for the x-forwarded-proto header Description When Django is deployed behind more than one proxy, the proxy behavior is sometimes to list the protocol as a comma-separated list. However, currently, Django expects only one value for the x-forwarded-proto header, instead of parsing it as a list of values and setting the protocol accordingly. x-forwarded-proto is a non-standard header, so there isn't a specification for its use, but different reverse-proxy vendors do use it in different ways, and some append the protocol as a comma-separated value from left-to-right (left being the furthermost proxy and rightmost being the closest). Similar issues have been raised and implemented in other projects, for example: Tornado: Issue: ​ Implementation: ​ Ruby: Issue: ​ Implemenation: ​ Reactor-Netty: ​ Implementation: ​ Most implementation use the leftmost-value or rightmost value. I would expect that provided that you are certain that the initial proxy can be trusted, that the left-most value makes the most sense, since it represent the original value at the entry-point for the HTTP request which is often where TLS is being terminated. Common example of this behavior is when using mulitple AWS proxies such as API Gateway proxying to an elastic load balancer."}
{"_id":"q-en-django__django-15498-8e5d326b4f06ea937ff2340078eaa775beb4ea6c88d93e7bf290063f292fe23e","text":"Fix handling empty string for If-Modified-Since header Description Empty string used to be ignored for If-Modified-Since header, but now raises exception since . Fix handling empty string for If-Modified-Since header Description Empty string used to be ignored for If-Modified-Since header, but now raises exception since ."}
{"_id":"q-en-django__django-15499-1d82ece2e149ad3c479cd446f7de35790f806b24112e6d0b6c8dc615fba8f990","text":"Optimize CreateModel + AlterModelManagers to CreateModel Description During migration optimization, CreateModel + AlterModelOptions is reduced to just CreateModel, with the model options. Similarly, CreateModel + AlterModelManagers can become just CreateModel."}
{"_id":"q-en-django__django-15503-3c83b2e4e771ac0b6217eae1ee5e808b826154c9e5e417e2b71bf9f93e129797","text":"haskey, haskeys, and hasanykeys JSONField() lookups don't handle numeric keys on SQLite, MySQL, and Oracle. Description (last modified by TheTerrasque) Problem When using models.​JSONField() ​haskey lookup with numerical keys on SQLite database it fails to find the keys. Versions: Django: 4.0.3 Python: 3.9.6 (tags\/v3.9.6:, Jun 28 2021, 15:26:21) [MSC v.1929 64 bit (AMD64)] on win32 sqlite3.version: '2.6.0' sqlite3.sqliteversion: '3.35.5' Example: Database DATABASES = { 'default': { 'ENGINE': '', 'NAME': 'db.sqlite3', } } Model class JsonFieldHasKeyTest(models.Model): data = models.JSONField() Test from import TestCase from .models import JsonFieldHasKeyTest class JsonFieldHasKeyTestCase(TestCase): def setUp(self) -None: test = JsonFieldHasKeyTest(data={'foo': 'bar'}) test2 = JsonFieldHasKeyTest(data={'1111': 'bar'}) () () def testjsonfieldhaskey(self): c1 = (datahaskey='foo').count() c2 = (datahaskey='1111').count() self.assertEqual(c1, 1, \"Should have found 1 entry with key 'foo'\") self.assertEqual(c2, 1, \"Should have found 1 entry with key '1111'\") Result FAIL: testjsonfieldhaskey () Traceback (most recent call last): File \"H:\\Files\\Projects\\Electaco\\Webservice\\elecserve\\markers\\\", line 16, in testjsonfieldhaskey self.assertEqual(c2, 1, \"Should have found 1 entry with key '1111'\") AssertionError: 0 != 1 : Should have found 1 entry with key '1111' Additional info This has been tested on SQLite and Postgresql backend, it works on postgresql but fails on sqlite."}
{"_id":"q-en-django__django-15521-ae4b8960f00da443d8972bcecb956561afad629ba46f52f66e4f68a201e2a023","text":"makemessages command skips the nl-nl-x-informal folder Description There is a mismatch on the DjangoTranslation folder and the makemessages command for the value nl-nl-x-informal DjangoTranslation uses the tolocale method to determine the language folder to read the tolocale translates nl-nl-x-informal correctly to the nlNL-x-informal folder. However makemessages skips the nlNL-x-informal folder and displays the following message invalid locale nlNL-x-informal, did you mean nlNLxinformal? ​ The check for - in the locale should only be for the first section a.k.a. nl_NL"}
{"_id":"q-en-django__django-15525-23e401ee3ce9a0c1bfb5a39dd2efe53e0993496172c3e9aa3ce493b94811bce9","text":"loaddata fails on non-default database when natural keys uses foreign keys. Description (last modified by François Granade) I've got a one-to-many relationship between two models Book and Author, that define a natural keys in both models. I'm loading some data from a fixture. It works in the default database, but when I use it a second database, then I get an exception. I'm relatively new to natural keys and to serializers, but I wouldn't expect things to work differently in the default DB and others ? I've committed a test project here: ​ (The problem doesn't appear if the data is already present in the default DB) The error: % cat | loaddata --database other --format json - Traceback (most recent call last): File \"\/Users\/francois\/Library\/Caches\/pypoetry\/virtualenvs\/exportbug-PGt-cwXF-py3.9\/lib\/python3.9\/site-\", line 187, in get relobj = (instance) File \"\/Users\/francois\/Library\/Caches\/pypoetry\/virtualenvs\/exportbug-PGt-cwXF-py3.9\/lib\/python3.9\/site-\", line 15, in getcachedvalue return [cachename] KeyError: 'author' During handling of the above exception, another exception occurred: Traceback (most recent call last): File \"\/Users\/francois\/Library\/Caches\/pypoetry\/virtualenvs\/exportbug-PGt-cwXF-py3.9\/lib\/python3.9\/site-\", line 70, in Deserializer yield from PythonDeserializer(objects, options) File \"\/Users\/francois\/Library\/Caches\/pypoetry\/virtualenvs\/exportbug-PGt-cwXF-py3.9\/lib\/python3.9\/site-\", line 174, in Deserializer obj = base.buildinstance(Model, data, using) File \"\/Users\/francois\/Library\/Caches\/pypoetry\/virtualenvs\/exportbug-PGt-cwXF-py3.9\/lib\/python3.9\/site-\", line 332, in buildinstance naturalkey = Model(data).naturalkey() File \"\/Users\/francois\/lmad\/src\/django-\", line 33, in naturalkey return (,) + () File \"\/Users\/francois\/Library\/Caches\/pypoetry\/virtualenvs\/exportbug-PGt-cwXF-py3.9\/lib\/python3.9\/site-\", line 205, in get relobj = self.getobject(instance) File \"\/Users\/francois\/Library\/Caches\/pypoetry\/virtualenvs\/exportbug-PGt-cwXF-py3.9\/lib\/python3.9\/site-\", line 168, in getobject return ((instance)) File \"\/Users\/francois\/Library\/Caches\/pypoetry\/virtualenvs\/exportbug-PGt-cwXF-py3.9\/lib\/python3.9\/site-\", line 496, in get raise ( Author matching query does not exist. the model: from import models class AuthorManager(models.Manager): def getbynaturalkey(self, name): return (name=name) class Author(models.Model): id = models.AutoField(primarykey=True) name = models.CharField(maxlength=255, unique=True) objects = AuthorManager() def naturalkey(self): return (,) def str(self): return f\"{} {}\" class BookManager(models.Manager): def getbynaturalkey(self, title, author): # OR title, author ?? return (title=title, authorname=author) class Book(models.Model): id = models.AutoField(primarykey=True) title = models.CharField(maxlength=255) author = models.ForeignKey(Author, models.DONOTHING, relatedname=\"books\") objects = BookManager() def naturalkey(self): return (,) + () naturalkey.dependencies = [\"testbug.Author\"] class Meta: uniquetogether = [[\"title\", \"author\"]] def str_(self): return f\"{}: '{}' by {}\" the data (generated with from import serializers; from import Book, Author; print(serializers.serialize(\"json\", list(()) + list(()), indent=2, usenaturalforeignkeys=True, usenaturalprimary_keys=True)) in the shell): [ { \"model\": \"\", \"fields\": { \"name\": \"JR Tolkien\" } }, { \"model\": \"\", \"fields\": { \"title\": \"The Ring\", \"author\": [ \"JR Tolkien\" ] } } ]"}
{"_id":"q-en-django__django-15526-1532361e6a6219c5b395238b8019d0a62880d9ad4c56a03c0d7f01a32370200d","text":"viewonsite redirect does not work for custom admin site. Description All reverse calls in have a currentapp keyword argument, so they resolve with the current (possibly custom) admin site – except the reverse call in getviewonsiteurl. This lead to custom admin sites using the default admin to redirect via `viewon_site. This is clearly a bug. When the default admin is handled with extra protections outside of Django (e.g. additional basic auth), users of the custom admin cannot use the \"view on site\" link because it routes through the default admin. Patch is easy and already here: ​ I'll try to provide a regression test as well."}
{"_id":"q-en-django__django-15554-a5e3a13f51fc27c5e4583ddcdbef05ba5ffd86f9cdc4c7f67281ce1a489139a4","text":"Using multiple FilteredRelation with different filters but for same relation is ignored. Description (last modified by lind-marcus) I have a relation that ALWAYS have at least 1 entry with isall=True and then I have an optional entry that could have isall=False but instead have zone set. I'm trying to use FilteredRelation together with Case(When()) to ensure that it use the zone level one (if exist) and fall back on \"All\" if zone do not exist. from import FilteredRelation ( relationzone=FilteredRelation( \"myrelationnested\", condition=Q(myrelationnestedzone=F(\"zone\")) ), relationall=FilteredRelation( \"myrelationnested\", condition=Q(myrelationnestedisall=True) ), pricezone=F(\"relationzoneprice\") ).annotate( pricefinal=Case( When( pricezoneisnull=True, then=F(\"relationallprice\"), ), default=F(\"pricezone\") ) ) I noticed that when using multiple FilteredRelation with the same relation (myrelationnested) it actually just generates a single SQL JOIN (if inspecting the raw SQL) and ignores the other. So in this case if I do print(str()) I would only see a join for relationzone. Not for relation_all. Is this intended behavior or should I be able to do the thing above?"}
{"_id":"q-en-django__django-15560-53443e7aca31eb6132f6e1c10b77e3e6b95b78a01946010e1522dc04f513cf7a","text":"createsuperuser does not validate usernames that use a UniqueConstraint. Description With a custom User model that uses a UniqueConstraint instead of unique=True, the createsuperuser command does not validate usernames at all. class CustomUser(AbstractBaseUser): customusername = models.CharField(maxlength=255) USERNAMEFIELD = 'customusername' class Meta: constraints = [ models.UniqueConstraint(fields=['customusername'], name='uniqueusernames'), ] Running createsuperuser to create a user with a username that already exists then results in an IntegrityError: IntegrityError: duplicate key value violates unique constraint \"uniqueusernames\" DETAIL: Key (customusername)=(foo) already exists."}
{"_id":"q-en-django__django-15561-42e2c9aeee25d08f7a70d4687a938f1a15b6b8af74140bf32a600e06e6b8ce73","text":"AlterField operation should be noop when adding\/changing choices on SQLite. Description while writing a test case for i found that for sqlite, even a seemingly db-transparent change like adding choices still generates sql (new table + insert + drop + rename) even though this shouldn't be needed. on e.g. postgres the same migration generates no sql"}
{"_id":"q-en-django__django-15563-2c865990983c88217135e94d064d20285e220c0543b4664055f2b2c1133359b4","text":"Wrong behavior on queryset update when multiple inheritance Description Queryset update has a wrong behavior when queryset class inherits multiple classes. The update happens not on child class but on other parents class instances. Here an easy example to show the problem: class Base(models.Model): baseid = models.AutoField(primarykey=True) fieldbase = models.IntegerField() class OtherBase(models.Model): otherbaseid = models.AutoField(primarykey=True) fieldotherbase = models.IntegerField() class Child(Base, OtherBase): pass Then in django shell: In [1]: (fieldotherbase=100) <QuerySet [{'otherbaseid': 1, 'fieldotherbase': 100}]In [2]: (fieldotherbase=101) <QuerySet [{'otherbaseid': 2, 'fieldotherbase': 101}]In [3]: (fieldbase=0, fieldotherbase=0) <Child: Child object (1)In [4]: (fieldbase=1, fieldotherbase=1) <Child: Child object (2)In [5]: (fieldotherbase=55) SELECT \"applianceschild\".\"baseptrid\" FROM \"applianceschild\" Execution time: 0.000647s [Database: default] UPDATE \"appliancesotherbase\" SET \"fieldotherbase\" = 55 WHERE \"appliancesotherbase\".\"otherbaseid\" IN (1, 2) Execution time: 0.001414s [Database: default] Out[5]: 2 In [6]: ('fieldotherbase') <QuerySet [{'fieldotherbase': 0}, {'fieldotherbase': 1}]In [7]: (otherbaseidin=[1,2]).values('fieldotherbase') <QuerySet [{'fieldotherbase': 55}, {'fieldotherbase': 55}]As seen on the above code, updating Child fields from second parent has no effect. Worse is that OtherBase fields where modifed because query is using primiary keys from Base class."}
{"_id":"q-en-django__django-15569-b832c9b3275e8723ee73661094b9db431bfd8ce37139a4c210267a02d136dc7b","text":"RegisterLookupMixin.unregisterlookup() should clear the lookup cache. Description (last modified by Himanshu Balasamanta) In current source code, in the unregisterlookup method, ​, the cache is not cleared, which should be done, as it is done in registerlookup, ​ Corresponding to this change, minor changes need to be brought in the test. The PR generated is ​"}
{"_id":"q-en-django__django-15572-6dc657b00ae31c6cef152afe5ade3a612a77f52e16482779c50a92fb4b882556","text":"Django 3.2.4+ autoreload breaks on empty string in TEMPLATES DIRS. Description Django versions 3.2.3 changes the way template dirs are handled, they are now normalized using pathlib.Path. People having an invalid value in TEMPLATESDIRS? will notice that autoreload stops working. \"DIRS\": (\"TEMPLATESDIRS\", \"\").split(\",\") # wrong, should be filter(None, (\"TEMPLATESDIRS\", \"\").split(\",\")) or anything else that produces this: \"DIRS\": [''] # wrong will break autoreload. This happens because was previously comparing the empty string to a directory, and would never match. Now the normalization transforms the empty string into the root of the project. The result is that templatechanged() will now always return True, preventing the autoreload when the app code changes Change that produced the regression Commits in main and stable\/3.2.x: ​ ​ Previous reports"}
{"_id":"q-en-django__django-15576-c0abbf0989ddc03368c3c8f3c51c174ebc47a581cd412e5064f220e6dc2667f5","text":"clear select fields for of distinct, non-sliced querysets Description The fix () for bug introduced an inefficiency in .exists() on distinct querysets in that now all distinct querysets are selecting all their fields. I propose the fields should be left on the queryset only if (q.distinct and (q.highmark is not None or q.lowmark is not None)). As I started writing the test for this, I realized I would have to inspect the query itself, which seems... like a bad idea for a test. Would this be reasonable to submit the (one-line-change) patch without a test, or does someone have a suggestion for a way to make the test that's less brittle?"}
{"_id":"q-en-django__django-15586-87759a3780227b66562a05084fd4ed1a5c0647423040b2ccd84103300f2991c6","text":"Enable cached template loader in development. Description (last modified by Carlton Gibson) Following the changes to allow auto-reloading cached templates in , there was discussion on ​PR to enable the cached template loader by default even in development. (Folks not wanting that would specify loaders.)"}
{"_id":"q-en-django__django-15607-84b6646c05fd30e0656ffa81e738becbfb8501f5575b1443ada6af20b11c2a99","text":"Unnecessary redirect in LogoutView when ?next=... contains \"unsafe\" URL Description Reproduction instructions Set LOGOUTREDIRECTURL Wire () at \/logout\/ in the URLconf Add this form to any template: <form action=\"\/logout\/?next=http:\/\/evil\/\" method=\"POST\"<input type=\"submit\" value=\"Logout\"{% csrftoken %}<\/formLog in, then use the form to log out Expected result You are logged out; the next parameter is ignored; you are redirected to LOGOUTREDIRECTURL Actual result There is an intermediary, useless redirect; see the logs of the development server: [16\/Apr\/2022 19:05:38] \"POST \/logout\/?next=http:\/\/evil\/ HTTP\/1.1\" 302 0 [16\/Apr\/2022 19:05:38] \"GET \/logout\/ HTTP\/1.1\" 302 0 [16\/Apr\/2022 19:05:38] \"GET \/en\/ HTTP\/1.1\" 200 I noticed this via code inspection. The implementation of LogoutView.getnextpage seemed a bit weird to me. This stems from ​ which predates the introduction of LOGOUTREDIRECTURL. From the user's perspective, the behavior is correct. There's just an extra round-trip and needlessly complicated code."}
{"_id":"q-en-django__django-15613-468545e965ecb568ad46ebec9bdb13e6e1e777a4549a573ccd702726ce0d0596","text":"Inconsistent \/ Unexpected handling of assigning unsaved model to Generic Foreign Key Description addresses the issue of assigned an unsaved model to a ForeignKey or OneToOneField (raises error when save() called), however the same logic \/ pattern does not apply to GFKs. Given: class ModelA(models.Model): name = models.CharField(maxlength=20) class ModelB(models.Model): gfkctype = models.ForeignKey(ContentType, ondelete=models.PROTECT) gfkid = models.PositiveIntegerField() gfk = GenericForeignKey('gfkctype', 'gfkid') class ModelC(models.Model): fk = models.ForeignKey(ModelA, ondelete=models.CASCADE) Foreign Key Behaviour: In [2]: a = ModelA(name='Model A') In [3]: c = ModelC(fk=a) In [4]: c.fk Out[4]: <ModelA: ModelA object (None)In [5]: c.save() ... ValueError: save() prohibited to prevent data loss due to unsaved related object 'fk'. In [6]: a.save() (0.016) INSERT INTO \"testappmodela\" (\"name\") VALUES ('Model A'); args=['Model A'] In [7]: c.fk Out[7]: <ModelA: ModelA object (1)In [8]: c.save() (0.016) INSERT INTO \"testappmodelc\" (\"fkid\") VALUES (1); args=[1] GFK behaviour: In [9]: a2 = ModelA(name='Model A2') In [10]: b = ModelB(gfk=a2) In [11]: b.gfk Out[11]: <ModelA: ModelA object (None)In [12]: b.save() (0.000) INSERT INTO \"testappmodelb\" (\"gfkctypeid\", \"gfkid\") VALUES (9, NULL); args=[9, None] IntegrityError: NOT NULL constraint failed: In [14]: () (0.015) INSERT INTO \"testappmodela\" (\"name\") VALUES ('Model A2'); args=['Model A2'] In [15]: b.gfk (0.000) SELECT \"testappmodela\".\"id\", \"testappmodela\".\"name\" FROM \"testappmodela\" WHERE \"testappmodela\".\"id\" IS NULL LIMIT 21; args=() None In [17]: b.gfkctype Out[17]: <ContentType: testapp | model aTwo observations: No check on b.gfk and b.gfkid value during save() which could lead to silent data loss if b.gfkid is nullable. When a2 is saved, accessing b.gfk now does a redundant DB query to try and find ModelA instance with PK = None, then then returns None value (effectively un-assigning a2 model), while keeping b.gfkctype intact. This is because the new pk of a2 is different to the existing gfkid (pkval) of the GFK field (None) What should be done: Modify () or Model.preparerelatedfieldsforsave() to also perform verification check for GFK fields Modify () to handle case of pkval = None (update fkfield value using PK value of GFK model if present, do not perform redundant DB query on pk=None, return previously assigned (then saved) model instead of None)"}
{"_id":"q-en-django__django-15620-94ec269cf922035522e2e37c212125671cbe847a01d12891cdfa930b231a818f","text":"Add SQL comment to describe deliberately no-op migration operations Description Currently when a field migration is a no-op, the operation description is output in SQL, but nothing else. This can be confusing as to which operations are no-ops. It could be clearer if we output an extra SQL comment when there are deliberately no statements to execute for a given migration operation. Take for example this output: BEGIN; -- Alter field name on Author ALTER ...; -- Alter field title on Book COMMIT; The field has an operation applied, whilst needs no changes to the database. This isn't exactly clear from the output - is the COMMIT part of the change? It could be clearer as: BEGIN; -- Alter field name on Author ALTER ...; -- Alter field name on Author -- (no-op) COMMIT; (Or perhaps more verbose wording, like \"no SQL to execute\") I think this can help especially when there are consecutive operations with no-op SQL: BEGIN; -- Alter field name on Author -- (no-op) -- Alter field name on Author -- (no-op) COMMIT; (Inspired by , where the OP suggested dropping such migration operation header comments.)"}
{"_id":"q-en-django__django-15629-2e5e3d789224a862f867f0ec6d5e275b8724dca0df9b7ee7a984615fbd477d10","text":"Errors with dbcollation – no propagation to foreignkeys Description (last modified by typonaut) Using dbcollation with a pk that also has referenced fks in other models causes foreign key constraint errors in MySQL. With the following models: class Account(models.Model): id = ShortUUIDField(primarykey=True, dbcollation='utf8bin', dbindex=True, maxlength=22) … class Address(models.Model): id = ShortUUIDField(primarykey=True, dbcollation='utf8bin', dbindex=True, maxlength=22) account = models.OneToOneField(Account, ondelete=models.CASCADE) … class Profile(models.Model): id = ShortUUIDField(primarykey=True, dbcollation='utf8bin', dbindex=True, maxlength=22) … account = models.ForeignKey('Account', verbosename=('account'), null=True, blank=True, ondelete=models.CASCADE) … etc Where has been changed from models.BigAutoField if makemigrations is run then it produces sqlmigrate output like this: ALTER TABLE MODIFY varchar(22) COLLATE ; ALTER TABLE MODIFY varchar(22) NOT NULL; ALTER TABLE MODIFY varchar(22) NULL; ALTER TABLE ADD CONSTRAINT FOREIGN KEY () REFERENCES (); ALTER TABLE ADD CONSTRAINT FOREIGN KEY () REFERENCES (); With this SQL the ADD CONSTRAINT queries fail. This is because the COLLATE should also be present in the bmanageaddress.accountid and bmanageprofile.accountid modification statements. Like this: ALTER TABLE MODIFY varchar(22) COLLATE ; ALTER TABLE MODIFY varchar(22) NOT NULL COLLATE ; ALTER TABLE MODIFY varchar(22) NULL COLLATE ; ALTER TABLE ADD CONSTRAINT FOREIGN KEY () REFERENCES (); ALTER TABLE ADD CONSTRAINT FOREIGN KEY () REFERENCES (); In the latter case the ADD CONSTRAINT statements run without error. The collation of the pk must match the collation of the fk otherwise an error will occur."}
{"_id":"q-en-django__django-15630-b61550a1cb9d8610d69e6713eddc3752cd7f4e0615ae52bfcdf1358fa3a0667a","text":"Unnecessary column in a GROUP BY clause with () Description I've got a question about the interaction between exists() and group by. For example something like: ('field').annotate(cnt=Count('id')).filter(cnt_gt=1).exists() The corresp. query with PostgreSQL looks like this: SELECT (1) AS \"a\" FROM \"appmodel\" GROUP BY \"appmodel\".\"field\", (1) HAVING COUNT(\"appmodel\".\"id\") 1 LIMIT 1 exists() (see ​) clears the SELECT clause and replaces it by (if I understand correctly) a hardcoded value 1 (as \"a\"), along with a limit of 1, which makes sense to me. But getgroupby() (see ​) pushes this hardcoded value to the GROUP BY clause and we end up with the query above. Now, on PostgreSQL, that works, but to me it sounds like it works by luck\/lucky robustness... and certainly the same query without the , (1) in the GROUP BY clause yields the same result. The problem is that outside of PostgreSQL that GROUP BY clause can be downright invalid... Note that if I alter exists() to use {'a': 2} instead of {'a': 1}, it does not work anymore (on PostgreSQL), because (2) in the SELECT clause means the hardcoded number 2 while (2) in the GROUP BY clause presumably (??) refers to the 2nd expr of the SELECT clause, but we only got one... My feeling is that getgroupby() should avoid adding extra\/raw sql (or at least extra\/raw pure values, if we can detect that...) to the group by expressions? NB: the existing\/old ticket that seems most related to my question would probably be:"}
{"_id":"q-en-django__django-15643-8695ed0068ad9d457d00e5381afc1bae01a54f664f546cf1b80b91276e5404eb","text":"Migrations changes to implicit primary key when primary key is altered on SQLite. Description Migrations drops explicit primary key defined on a model when it's altered."}
{"_id":"q-en-django__django-15648-af52ab8e9d61175371dc2916d68bba4c4d3f07a5c89fa703243b55a15d60a641","text":"views.Feed methods cannot be decorated Description If one applies a decorator on a method which is called by getdynamicattr a TypeError like this occurs: Exception Type: TypeError at \/blog\/feed\/ Exception Value: itemlink() takes exactly 2 arguments (1 given) I think this is because getdynamicattr tries to count the function's arguments, but decorators usally get defined with the args, *kwargs syntax, so this trick does not work here. if code.coargcount == 2: # one argument is 'self' return attr(obj) else: return attr() I think the best approach would be to remove one of the two methods. IMHO We should have either attr(item) or attr() not both, as \"there should be one, and preferably only one, obvious way to do it\"."}
{"_id":"q-en-django__django-15651-2fb8b9319f4db4b4c4cc03332396f0529637fce9690cf307909a8522890fdcbc","text":"Implement RenameIndex in a backwards compatible way Description In order to eventually deprecate indextogether we need a way to deal with old projects that have unnamed indexes. This proves to be a non-trivial problem. Andrew and I came up with these things to consider. RenameIndex(model, newname, oldname=None, oldfields=None) where exactly one of oldname and oldfield is given (oldname ^ oldfields) If the oldname is given we use RENAME INDEX if available Otherwise look at the state and drop existing indexes and create new the index with new name On MySQL (or other DBs) that don't support RENAME INDEX, provide SQL query to look up index name from informationschema by field names and pass in to DROP INDEX. If more than one index is found while identifying with field names, migrations must error out with an AmbiguityError If the autodetector finds an old, unnamed index and a new, named one matching field signature, issue a RenameIndex operation For backwards operations with unnamed old indexes, RenameIndex is a noop."}
{"_id":"q-en-django__django-15666-af4a71d318ef0632e6d5d31111c930e0968c7cc5d4a076c7d8d5e1b06f410873","text":"Query Expression in ordering of a related object fails Description (last modified by Tim Graham) Since 2.0, according to the doc (​), we can use QueryExpression objects in the Model.Meta.ordering field. Using: from import models class Musician(models.Model): firstname = models.CharField(maxlength=50) lastname = models.CharField(maxlength=50) instrument = models.CharField(maxlength=100, null=True, blank=True) class Meta: ordering = [models.F('instrument').asc(nullslast=True)] class Album(models.Model): artist = models.ForeignKey(Musician, ondelete=models.CASCADE) name = models.CharField(maxlength=100) releasedate = models.DateField() numstars = models.IntegerField() class Meta: ordering = ['artist'] () ... TypeError: 'OrderBy' does not support indexing When reaching ​, the compiler tries to use the related model, but at line 679, item can be an OrderBy object. Thus the failure."}
{"_id":"q-en-django__django-15669-74fcee5ea1d2ef85cae0f1b04c81ed88aa5d6cb925b9e06fe3dfddb860cbefc8","text":"Create --update flag for makemigrations management command, mimicking South's one. Description When I was developing with South I was able to use the --update flag for the schemamigration management command (​) to refine the latest migration. This was very convenient for iterative development. Could we have an equivalent of the --update flag for Django=1.7 makemigrations? (I've taken it from ​)"}
{"_id":"q-en-django__django-15671-05528e04cdd5f121db19adafbad8f014d5cffcc0f7b7d1ddcba0b1fe79683699","text":"Allow overriding FormSet's error messages for too few and too many forms. Description Moving the messages to defaulterrormessages enables overriding them via the errormessages argument when instantiating the FormSet. I would prefer being able to override them via inlineformsetfactory, but that is for another day."}
{"_id":"q-en-django__django-15678-5c16a9ae69264f33b082f6b524ba179f0469f00831fb2d7d9fae2a09527fed41","text":"Deprecate CryptPasswordHasher. Description CryptPasswordHasher was 15 years ago mainly to support legacy UNIX apps. It's almost undocumented, not recommended, and supported only on UNIX. Moreover crypt module was deprecated in Python 3.11 (see ​). We should deprecate it in Django 4.1 and remove in Django 5.0."}
{"_id":"q-en-django__django-15682-2541049c367f4c44e702d6fda7fa77f9542f691cbb1115e8bdf7732c50b2cafb","text":"Depracate passing False to OrderBy's nullsfirst and nullslast. Description Consider the following: In [11]: [tv.publishedat for tv in TemplateVersion.objects.orderby(F(\"publishedat\").desc(nullsfirst=True))] Out[11]: [None, datetime.datetime(2022, 2, 25, 13, 0, 12, , tzinfo=<UTC), datetime.datetime(2022, 2, 21, 10, 18, 0, , tzinfo=<UTC)] In [12]: [tv.publishedat for tv in TemplateVersion.objects.orderby(F(\"publishedat\").desc(nullsfirst=False))] Out[12]: [None, datetime.datetime(2022, 2, 25, 13, 0, 12, , tzinfo=<UTC), datetime.datetime(2022, 2, 21, 10, 18, 0, , tzinfo=<UTC)] In [13]: [tv.publishedat for tv in TemplateVersion.objects.orderby(F(\"publishedat\").desc(nullslast=True))] Out[13]: [datetime.datetime(2022, 2, 25, 13, 0, 12, , tzinfo=<UTC), datetime.datetime(2022, 2, 21, 10, 18, 0, , tzinfo=<UTC), None] In [14]: [tv.publishedat for tv in TemplateVersion.objects.orderby(F(\"publishedat\").desc(nullslast=False))] Out[14]: [None, datetime.datetime(2022, 2, 25, 13, 0, 12, , tzinfo=<UTC), datetime.datetime(2022, 2, 21, 10, 18, 0, , tzinfo=<UTC)] Observe how nulls_first=False still puts the nulls first. This happens because they both default False and when they are both False it lets the DB decide. This is surprising behaviour, it also makes changing the null positioning based on a variable more awkward than it needs to be. I think it would be better if they defaulted to None, let the DB decide when both are None and when one is not None do the ordering that implies."}
{"_id":"q-en-django__django-15689-0d0e05872d40af8814dc01c6c21b91717b9ca2ec7a2f7594c70e3d891513e234","text":"APPENDSLASH adds significant latency to all requests not ending in \/ (even if successful) Description Originally, APPENDSLASH worked by looking for 404 responses and replacing them with redirects, so as not to unnecessarily impact the performance of successful responses. However, commit in 1.9.5\/1.10 changed this to check shouldredirectwithslash() on every request, resulting in a moderately expensive extra urlconf lookup for every request not ending with \/, whether or not it succeeds as written. This performance impact was not considered in the commit message or the corresponding ticket , so I assume it was an oversight. That ticket asserted “This doesn't really make sense, since the two settings are not interdependent”, which is incorrect—performance was the reason for the interdependence. The overhead was found to be significant enough in Zulip to merit ​subclassing CommonMiddleware to skip it in certain conditions. Here’s a ​minimal test project with an exaggerated number of routes so the overhead can be easily observed. $ runserver $ wrk http:\/\/127.0.0.1:8000\/url9999 Running 10s test @ http:\/\/127.0.0.1:8000\/url9999 2 threads and 10 connections Thread Stats Avg Stdev Max +\/- Stdev Latency 232.40ms 73.85ms 570.86ms 69.16% Req\/Sec 21.70 9.47 40.00 63.35% 426 requests in 10.01s, 64.90KB read Requests\/sec: 42.56 Transfer\/sec: 6.48KB $ sed -i 's\/# APPENDSLASH = False\/APPENDSLASH = False\/' $ wrk http:\/\/127.0.0.1:8000\/url9999 Running 10s test @ http:\/\/127.0.0.1:8000\/url9999 2 threads and 10 connections Thread Stats Avg Stdev Max +\/- Stdev Latency 139.80ms 52.07ms 352.19ms 69.09% Req\/Sec 36.46 12.23 60.00 58.12% 714 requests in 10.01s, 108.79KB read Requests\/sec: 71.32 Transfer\/sec: 10.87KB"}
{"_id":"q-en-django__django-15695-fb9328dce2a71e400f1b933cfd87d751abb9d05bea1de670e3d32ba805afdc0a","text":"RenameIndex() crashes when unnamed index is moving backward and forward. Description RenameIndex() should restore the old auto-generated name when an unnamed index for uniquetogether is moving backward. Now re-applying RenameIndex() crashes. For example: diff --git index .. a b class OperationTests(OperationTestBase): with connection.schemaeditor() as editor, self.assertNumQueries(0): operation.databasebackwards(applabel, editor, newstate, projectstate) self.assertIndexNameExists(tablename, \"newponytestidx\") 2991 # Re-apply renaming. 2992 with connection.schemaeditor() as editor: 2993 operation.databaseforwards(applabel, editor, projectstate, newstate) 2994 self.assertIndexNameExists(tablename, \"newponytestidx\") # Deconstruction. definition = operation.deconstruct() self.assertEqual(definition[0], \"RenameIndex\") crashes on PostgreSQL: relation \"newponytest_idx\" already exists"}
{"_id":"q-en-django__django-15698-372ef103f9f58352487757bd17b80d9e6b09afa0ff7069d59fdf0b200267a2cf","text":"Templates crash when calling methods for built-in types. Description (last modified by Daniel) Found during a 2.2 -3.2 upgrade: Given a template: {{ foo }} where foo is non-existant, it returns nothing, empty. (That's good) {{ }} also empty (Also good) {% include '' with bar=foo %} and then in having: {{ }} results in File \"\/Users\/daniel\/src\/django-bug-test\/.v\/lib\/python3.8\/site-\", line 861, in resolvelookup signature = inspect.signature(current) File \"\", line 3093, in signature return Signature.fromcallable(obj, followwrapped=followwrapped) File \"\", line 2842, in fromcallable return signaturefromcallable(obj, sigcls=cls, File \"\", line 2296, in signaturefromcallable return signaturefrombuiltin(sigcls, obj, File \"\", line 2107, in signaturefrombuiltin raise ValueError(\"no signature found for builtin {!r}\".format(func)) Exception Type: ValueError at \/ Exception Value: no signature found for builtin <built-in method count of str object at 0x1100ff2f0On django 2.2, this would not crash, but resulted in empty (as I expected). this seems to fix it for me: diff --git index .. --- +++ -913,15 +913,19 def resolvelookup(self, context): try: # method call (assuming no args required) current = current() except TypeError: signature = inspect.signature(current) try: () except TypeError: # arguments were required current = ( ) # invalid method call signature = inspect.signature(current) except ValueError: # python builtins might not have signature current = else: raise try: () except TypeError: # arguments were required current = ( ) # invalid method call else: raise except Exception as e: templatename = getattr(context, \"templatename\", None) or \"unknown\" ("}
{"_id":"q-en-django__django-15703-619d492ebfb5e434e086b5067354bde00311beaff3d78f576ba0184021818292","text":"Deprecate Model.Meta.indextogether in favour of Model.Meta.indexes Description Anything that indextogether does can be done by indexes and the Index class."}
{"_id":"q-en-django__django-15731-1a350705d9f0ef955e01b0012773d26de20e5be6020dec3ef4a7256e75725c08","text":"inspect.signature() returns incorrect signature on manager methods. Description (last modified by Shiva Kumar) inspect.signature returns incorrect signature information when used on queryset methods import inspect from import models class Person(models.Model): name = models.CharField(maxlength=100) print(inspect.signature(Person.objects.bulkcreate)) ipython and jupyter seem to internally use inspect.signature to show documentation when using the <obj? command and they too show incorrect signature information: The issue is due to the code at ​ Although we are ensuring the decorated method has the right name and docstring on lines 87 and 88, complete metadata is not copied. The fix is to use instead of manually assigning name and docstring. wraps will take care of all the metadata and inspect.signature will return the expected output. If the bug is acknowledged please assign the ticket to me, I would like to raise a PR for this."}
{"_id":"q-en-django__django-15732-654bb50156715147ac739e6f5e997dd471898e16598c1912d383913563786bcf","text":"Cannot drop uniquetogether constraint on a single field with its own unique=True constraint Description I have an erroneous uniquetogether constraint on a model's primary key (uniquetogether = (('id',),)) that cannot be dropped by a migration. Apparently the migration tries to find all unique constraints on the column and expects there to be only one, but I've got two — the primary key and the uniquetogether constraint: Indexes: \"foobarpkey\" PRIMARY KEY, btree (id) \"foobarid1c3b3088c74c3b17uniq\" UNIQUE CONSTRAINT, btree (id) Database is PostgreSQL, if that makes any difference."}
{"_id":"q-en-django__django-15737-d8fbe9c71443e07c250b9d4ff95a1b83825748e131f5135e276f4482a31c88ed","text":"Avoid unnecessary clear of cached reference Description (last modified by Barry Johnson) Consider this case of ORM models \"Parent\" and \"Child\", where Child has a foreign key reference to Parent (and the database can return generated IDs following insert operations): parent = Parent(name='parentobject') child = Child(parent=parent) () () print() The print statement will cause an unnecessary lazy read of the parent object. In the application where this behavior was first observed, the application was creating thousands of parent and child objects using bulkcreate(). The subsequent lazy reads occurred when creating log entries to record the action, and thousands of unwanted SELECT queries. Closed ticket solved a problem with potential data loss in this situation by essentially executing child.parentid = while preparing the child object to be saved. However, when the child's ForeignKeyDeferredAttrbute \"parentid\" changes value from None to the parent's ID, the child's internal cache containing the reference to \"parent\" is cleared. The subsequent reference to then must do a lazy read and reload parent from the database. A workaround to avoid this lazy read is to explicitly update both the \"parentid\" and \"parent\" cache entry by adding this non-intuitive statement: = after executing () But it appears that a simple change could avoid clearing the cache in this narrow case. Within Model.preparerelatedfieldsforsave(), replace setattr(self, field.attname, ) with setattr(self, , obj) This suggested code has -not- been tested. This change would set the associated \"parent_id\" attribute while ensuring that the currently referenced object remains referenced."}
{"_id":"q-en-django__django-15738-f32744fa3c87892068818e8eb5cb20dee56cdbdcc693e3c5cc0b0c1e076a316f","text":"Models migration with change field foreign to many and deleting unique together. Description (last modified by Simon Charette) I have models like class Authors(models.Model): projectdataset = models.ForeignKey( ProjectDataSet, ondelete=models.PROTECT ) state = models.IntegerField() startdate = models.DateField() class Meta: uniquetogether = (('projectdataset', 'state', 'startdate'),) and class DataSet(models.Model): name = models.TextField(maxlength=50) class Project(models.Model): datasets = models.ManyToManyField( DataSet, through='ProjectDataSet', ) name = models.TextField(maxlength=50) class ProjectDataSet(models.Model): \"\"\" Cross table of data set and project \"\"\" dataset = models.ForeignKey(DataSet, ondelete=models.PROTECT) project = models.ForeignKey(Project, ondelete=models.PROTECT) class Meta: uniquetogether = (('dataset', 'project'),) when i want to change field projectdataset in Authors model from foreign key field to many to many field I must delete a uniquetogether, cause it can't be on many to many field. Then my model should be like: class Authors(models.Model): projectdataset = models.ManyToManyField( ProjectDataSet, ) state = models.IntegerField() startdate = models.DateField() But when I want to do a migrations. python3 makemigrations python3 migrate I have error: ValueError: Found wrong number (0) of constraints for appauthors(projectdataset, state, startdate) The database is on production, so I can't delete previous initial migrations, and this error isn't depending on database, cause I delete it and error is still the same. My solve is to first delete unique_together, then do a makemigrations and then migrate. After that change the field from foreign key to many to many field, then do a makemigrations and then migrate. But in this way I have 2 migrations instead of one. I attachment with this project, download it and then do makemigrations and then migrate to see this error."}
{"_id":"q-en-django__django-15741-663b0e23cba3e73227a8e9fc237ae23bda46d9c5640c3601402995305fe8af08","text":"should allow lazy parameter Description Commit [] (fixing ) triggered a regression when the date template filter (possibly others are affected too) receives a lazy string, like in somedate|date:('Y-m-d'). This fails with: TypeError: getattr(): attribute name must be string in"}
{"_id":"q-en-django__django-15742-2ffe1c6e7e867e033cc688aa1bb122e6521a23ec30e1b6ce393696d84a02c0c3","text":"Blocktranslate asvar escapes variables, but stores the result as str instance, leading to double escaping Description In the docs, this snippet is given as an example usage of blocktranslate with the asvar argument (here: ​ {% blocktranslate asvar thetitle %}The title is {{ title }}.{% endblocktranslate %} <title{{ thetitle }}<\/title<meta name=\"description\" content=\"{{ thetitle }}\"However, this template is buggy when title is a string, which I'd argue is a common use case. title will be escaped when formatting the content of the blocktranslate block, but the \"was escaped\" information is discarded, and thetitle will be a str instance with escaped content. When later using the thetitle variable, it will be conditionally escaped. Since it is a str, it will be escaped, so control characters are escaped again, breaking their display on the final page. Minimal example to reproduce (can be put in any view): from django.template import Template, Context templatecontent = \"\"\" {% blocktranslate asvar thetitle %}The title is {{ title }}.{% endblocktranslate %} <title{{ thetitle }}<\/title<meta name=\"description\" content=\"{{ thetitle }}\"\"\"\" rendered = Template(templatecontent).render(Context({\"title\": \"<& Title\"})) assert \"&amp;lt;\" not in rendered, \"was escaped two times\" I'd argue that blocktranslate should: Either assign a SafeString instance to prevent future escaping or not escape the variables used within the translation, and store them marked as unsafe (= as str instance)"}
{"_id":"q-en-django__django-15744-e698c6e1d329ad209d9322f04b53443d9b4820efcadde1bea65ba56cc2f3023a","text":"Request body is closed prematurely by the ASGI handler Description The following PR ​ introduces I\/O operation on closed file., due to closing the ASGI request body before getting the response, e.g. accessing in a view."}
{"_id":"q-en-django__django-15747-51a6d455b05b75d136332d4935caf75682e09c0f2870b25149f06d2d0e67ddc3","text":"() raises FieldError on queryset ordered by an annotated field. Description A FieldError results if I try to perform a custom a custom action on data in the admin interface IF that data is sorted by an annotated field. The action is obviously not acting on an annotated field as those fields do not exist in the database, but raises a FieldError saying it 'Cannot resolve keyword 'fieldthatdataiscurrentlysortedby' into field.' and then lists the choices for fields (all of which are the raw model fields, not the annotated fields in the custom queryset). My admin model: class PositionAdmin(admin.ModelAdmin): def init(self, args, kwargs): super().init(args, kwargs) def getqueryset(self, request): qs = super(PositionAdmin, self).getqueryset(request).prefetchrelated( 'orders').selectrelated( 'portfolio', 'currentposition').selectrelated( 'currentposition').annotate( Count('orders', distinct=True), Count('shortrequests'), totalpositionvalue=ExpressionWrapper( F('price') F('quantity'), outputfield=IntegerField()), diffshares=Coalesce( F('quantity') F('currentpositionquantity'), F('quantity')), diffvalue=ExpressionWrapper( F('diffshares') F('price'), outputfield=IntegerField()), orderswitherrors=Count( Case(When(~Q(orderserror=''), then=1))), nonacceptedorders=Count( Case(When(Q(ordersstatus=''), then=1))), ordertoday=Count( Case(When( orderscreatedat_gte=brokeragetodaystart(), then=1)))) return qs # Custom Action def approvepositionfortrading(self, request, queryset): try: numposapproved = (approved=True) except FieldError: self.messageuser( request, \"You cannot perform actions when you have sorted by \" \"this column. Please remove your column sortings and then try \" \"your action again.\", level=messages.ERROR) else: if numposapproved == 1: messagebit = \"1 position was was\" else: messagebit = \"%s positions were\" % numposapproved self.messageuser( request, \"%s successfully approved for trading.\" % messagebit) I had to write code to handle the error so that the user will know how to proceed. However, this seems like bad behavior. Django can successfully perform that action (changing the 'Approved' field to 'True') on the data if it is sorted by any core model field, or if sorted by no field at all. However, if the data is sorted by a field resulting from the annotation of my queryset, then I get the FieldError when I try to perform the action. This seems like a bug."}
{"_id":"q-en-django__django-15752-417ec9ed86845f1940020f4fb1cdfaab92056217656235158a0d058df8fddced","text":"Deprecate makerandompassword(). Description (last modified by Carlton Gibson) It's unused since , see also Carlton's comment."}
{"_id":"q-en-django__django-15766-e18c85497d946260931247a6296fe169a8fcc19aa12666dc669bea74f41db624","text":"Supporting robust oncommit handlers. Description (last modified by Josh Smeaton) I recently tracked down an issue in my application where some oncommit handlers didn't execute because one of the previous handlers raised an exception. There appears to be no way to execute oncommit handlers robustly as you're able to do with signals [0] using sendrobust. I could sprinkle try\/catches around the place, but I'd like to avoid doing so because not all functions that are used as handlers should always swallow exceptions, but could do so when run as oncommit handlers. Targeting which handlers can be robust or not would be really useful, for example: def updatesearch(user): # if updating search fails, it's fine, we'll bulk update later anyway transaction.oncommit(lambda: (user), robust=True) def triggerbackgroundtaskone(user): # if this task fails, we want to crash transaction.oncommit(lambda: ()) Here if search fails to update it doesn't prevent the background task from being scheduled. I'm proposing to add a robust kwarg that defaults to False, for backward compatibility, but allows a user to tag specific handlers as such. [0] ​"}
{"_id":"q-en-django__django-15774-a67184aad62810f0fd77fe0b90893e73a2775a053f79dc3e04488e5be4a1a1a4","text":"Django detects HTTP Accept-Language header in case-sensitive manner Description This issue was originally discussed in django-developers: ​!topic\/django-developers\/1Y9LZSAOSnE Per ​w3c, ​rfc2616 and ​bcp47, Language tags should be parsed in case-insensitive, however, I noticed that Django detects HTTP Accept-Language headers in case-sensitive manner. For example, the following headers: Chrome: Accept-Language: zh-TW,zh;q=0.8,en-US;q=0.6,en;q=0.4 Firefox: Accept-Language: zh-tw,zh;q=0.8,en-us;q=0.5,en;q=0.3 Django will correctly display Traditional Chinese for Chrome, but won't for Firefox because of lower-cased TW. The fix contains two parts: Fix potential case-sensitive places in code to follow case-insensitive (for example parseacceptlang_header()) Fix ​documentation, correct the sentence \"Browsers send the names of the languages they accept in the Accept-Language HTTP header using this format. Examples: it, de-at, es, pt-br. Both the language and the country parts are in lower case. \", which obviously incorrect, Chrome uses tags like zh-TW, pt-BR."}
{"_id":"q-en-django__django-15781-159abc4f42b7ce41354f4977a9e837560bf7e96bd0ca80f41d18f2a8afa80466","text":"Customizable management command formatters. Description With code like: class Command(BaseCommand): help = Help output is: $ help tzktimport usage: tzktimport [-h] [--api API] [--version] [-v {0,1,2,3}] [--settings SETTINGS] [--pythonpath PYTHONPATH] [--traceback] [--no-color] [--force-color] [--skip-checks] blockchain target Import a contract from tzkt Example usage: tzktimport 'Tezos Mainnet' KT1HTDtMBRCKoNHjfWEEvXneGQpCfPAt6BRe positional arguments: blockchain Name of the blockchain to import into target Id of the contract to import When that was expected: $ help tzktimport usage: tzktimport [-h] [--api API] [--version] [-v {0,1,2,3}] [--settings SETTINGS] [--pythonpath PYTHONPATH] [--traceback] [--no-color] [--force-color] [--skip-checks] blockchain target Import a contract from tzkt Example usage: tzktimport 'Tezos Mainnet' KT1HTDtMBRCKoNHjfWEEvXneGQpCfPAt6BRe positional arguments: blockchain Name of the blockchain to import into target Id of the contract to import"}
{"_id":"q-en-django__django-15789-9e13f23da750739dc60f6c6717e4a037e97bc141ccba24763b42adcd1827fbee","text":"Add an encoder parameter to (). Description I have a use case where I want to customize the JSON encoding of some values to output to the template layer. It looks like is a good utility for that, however the JSON encoder is hardcoded to DjangoJSONEncoder. I think it would be nice to be able to pass a custom encoder class. By the way, is not documented (only its template filter counterpart is), would it be a good thing to add to the docs?"}
{"_id":"q-en-django__django-15790-051f8f9276d55f2516e29c025cfafbfbc58d98a2084449aa774f50bfd1a2de89","text":"checkfortemplatetagswiththesamename with libraries in TEMPLATES Description I didn't explore this thoroughly, but I think there might be an issue with the checkfortemplatetagswiththesamename when you add a template tag library into TEMPLATES['OPTIONS']['librairies']. I'm getting an error like: (templates.E003) 'mytags' is used for multiple template tag modules: '', 'someapp.templatetags.my_tags'"}
{"_id":"q-en-django__django-15799-154fd07595e2b9de2db52b12cdeaa74d1895084841022f98fb08354965c36afe","text":"SelectMultiple in ModelAdminForm display help text when allowmultipleselected is False. Description In AdminForm Help text on render for SelectMultiple widget don't check, if widget.allowmultipleselected = False. Widget himself on render checks it if self.allowmultipleselected: context['widget']['attrs']['multiple'] = True But helptext for widget, whose is rendered behind widget - don't checks it. There we check only \"isinstance\" if (isinstance(, SelectMultiple) ann not isinstance(, (CheckboxSelectMultiple, AutocompleteSelectMultiple))): ... # do some stuff with help text as a result I get \"msg\", which should not be."}
{"_id":"q-en-django__django-15814-f02c0d945e3437d90e66173ab03d7e5b839c3facc00badfe809a166da513c4f3","text":"() after selectrelated() crash on proxy models. Description When I optimize a query using selectrelated() and only() methods from the proxy model I encounter an error: Windows 10; Python 3.10; Django 4.0.5 Traceback (most recent call last): File \"D:\\study\\djangocollege\\\", line 22, in <modulemain() File \"D:\\study\\djangocollege\\\", line 18, in main executefromcommandline() File \"D:\\Anaconda3\\envs\\django\\lib\\site-packages\\django\\core\\management\\\", line 446, in executefromcommandline utility.execute() File \"D:\\Anaconda3\\envs\\django\\lib\\site-packages\\django\\core\\management\\\", line 440, in execute self.fetchcommand(subcommand).runfromargv() File \"D:\\Anaconda3\\envs\\django\\lib\\site-packages\\django\\core\\management\\\", line 414, in runfromargv self.execute(args, cmdoptions) File \"D:\\Anaconda3\\envs\\django\\lib\\site-packages\\django\\core\\management\\\", line 460, in execute output = (args, options) File \"D:\\study\\djangocollege\\project\\users\\management\\commands\\\", line 9, in handle objs = list(AnotherModel.objects.selectrelated(\"custom\").only(\"customname\").all()) File \"D:\\Anaconda3\\envs\\django\\lib\\site-packages\\django\\db\\models\\\", line 302, in len self.fetchall() File \"D:\\Anaconda3\\envs\\django\\lib\\site-packages\\django\\db\\models\\\", line 1507, in fetchall self.resultcache = list(self.iterableclass(self)) File \"D:\\Anaconda3\\envs\\django\\lib\\site-packages\\django\\db\\models\\\", line 71, in iter relatedpopulators = getrelatedpopulators(klassinfo, select, db) File \"D:\\Anaconda3\\envs\\django\\lib\\site-packages\\django\\db\\models\\\", line 2268, in getrelatedpopulators relcls = RelatedPopulator(relklassinfo, select, db) File \"D:\\Anaconda3\\envs\\django\\lib\\site-packages\\django\\db\\models\\\", line 2243, in init = () ValueError: 'id' is not in list Models: class CustomModel(models.Model): name = models.CharField(maxlength=16) class ProxyCustomModel(CustomModel): class Meta: proxy = True class AnotherModel(models.Model): custom = models.ForeignKey( ProxyCustomModel, ondelete=models.SETNULL, null=True, blank=True, ) Command: class Command(BaseCommand): def handle(self, args, *options): list(AnotherModel.objects.selectrelated(\"custom\").only(\"customname\").all()) At in 745 line there is snippet: opts = If I replace it by opts = all works as expected."}
{"_id":"q-en-django__django-15819-ce583822762eaea5c2b27f9eb5ae4139874990311f8c4a962206178ef37a6ab1","text":"inspectdb should generate relatedname on same relation links. Description Hi! After models generation with inspectdb command we have issue with relations to same enities module.Model.field1: (fields.E304) Reverse accessor for 'module.Model.field1' clashes with reverse accessor for 'module.Model.field2'. HINT: Add or change a relatedname argument to the definition for 'module.Model.field1' or 'module.Model.field2'. Maybe we can autogenerate relatedname='attributename' to all fields in model if related Model was used for this table"}
{"_id":"q-en-django__django-15828-4d0cfd3b20d21f6cbb38a4cab1b0e2ab0860bf4de76c2c41f2985d700511ed37","text":"BaseConstraint.deconstruct() and eq operators don't take violationerrormessage into account. Description Thanks Stéphane \"Twidi\" Angel for the report. Regression in ."}
{"_id":"q-en-django__django-15851-57ca404b0f4632e316b96b9a76c0fc54ef5fffbe1a8cb1e0f811a90609cf40e5","text":"dbshell additional parameters should be passed before dbname on PostgreSQL. Description psql expects all options to proceed the database name, if provided. So, if doing something like ` dbshell -- -c \"select from sometable;\" one will get this: $ dbshell -- -c \"select from sometable;\" psql: warning: extra command-line argument \"-c\" ignored psql: warning: extra command-line argument \"select * from sometable;\" ignored psql (10.21) Type \"help\" for help. somedatabase=It appears the args list just need to be constructed in the proper order, leaving the database name for the end of the args list."}
{"_id":"q-en-django__django-15863-ad1f21928a6cc34992ec8407ec5f7520e4f814ffcf832ce645a842833366c071","text":"Filter floatformat drops precision in decimal numbers Description I discovered that floatformat template filter may drop precision when used for Decimal numbers. MWE: from decimal import Decimal from django import setup from import settings from django.template import Template, Context TEMPLATES = [ { 'BACKEND': '', }, ] settings.configure(TEMPLATES=TEMPLATES) setup() t = Template('{{ value|floatformat:20 }}') c = Context({'value': Decimal('42.')}) print(t.render(c)) . I traced the bug to incorrect conversion to Decimal within the floatformat implementation that can't really work for Decimal numbers. Decimal numbers are converted to float instead. Pull request is prepared ​"}
{"_id":"q-en-django__django-15869-bf192642063b7ea66f46041ec8e84d32db30f5fff54a80753f31c2fe1a0fc4fa","text":"Deprecate lengthis template filter in favor of length. Description (last modified by Nick Pope) The lengthis template filter is a vestige from the days of the {% ifequal %} and {% ifnotequal %} tags before {% if %} arrived with support for comparison with operators. Even the example in the documentation (see ​here) is poor: {{ value|lengthis:\"4\" }} will only return one of three possible values - True, False, or \"\", the empty string being for errors in the provided values. It seems to me that it would be better to encourage use of the length template filter with the {% if %} template tag which can provide more flexibility: {# Before: #} {% if value|lengthis:\"4\" %}...{% endif %} {{ value|length_is:\"4\" }} ← This is of dubious use given the restricted set of possible output values. {# After: #} {% if value|length == 4 %}...{% endif %} {% if value|length == 4 %}True{% else %}False{% endif %} ← Much clearer but also allows customising the output values."}
{"_id":"q-en-django__django-15902-fdd93fca487e6f8ff3cd86e112cff9232ec393c3349ae0adae58f9ec6fe6b13c","text":"\"\" deprecation warning raised for ManagementForm's Description I have a project where I never render forms with the {{ form }} expression. However, I'm still getting the new template deprecation warning because of the formset management form production, during which the template used is insignificant (only hidden inputs are produced). Is it worth special-casing this and avoid producing the warning for the management forms?"}
{"_id":"q-en-django__django-15916-b3f0357a5bcb8009523b522dc0984bf3f82b1d9770d701e12bbdc3467e383b9e","text":"Allow ModelForm meta to specify formfieldcallback. Description (last modified by Klaas-Jan Gorter) The function returns a form class based on the class it recieves as form argument. As an additional argument it accepts a formfieldcallback function. When no callback is provided the class uses no callback instead of the formfieldcallback of the base form provided. Example: from django import forms form import models class MyModel(forms.Model): active = models.BooleanField() name = models.CharField(maxlength=64, blank=True, null=True) def allrequired(field, kwargs): formfield = field.formfield(kwargs) formfield.required = True return formfield class MyForm(forms.ModelForm): formfieldcallback = allrequired class Meta: model = MyModel formfieldcallback = allrequired fields = ['active', 'name'] FactoryForm = forms.modelformfactory(MyModel, form=MyForm) The expected behavior would be that the FactoryForm uses the formfieldcallback specified in the Meta attribute of MyForm and that therefore the fields would be required in both the FactoryForm and MyForm. However, under the current behavior of modelformfactory the formfieldcallback is overwritten (with the default argument None) before the new class is constructed and in FactoryForm the fields are not required. I believe this is a bug, because this behavior has been observed before in Ticket in Django 1.3. The test that was proposed there was incorrect, because under the expected behavior the callback should have been called four times not two times as was asserted. (I believe this test has been removed from version 2, because I find no equivalent test in tests\/modelformsetsregress.)"}
{"_id":"q-en-django__django-15925-f331fcc3a9fab43fb43cb3c84ce52b78e53bfaf73e33663fe437994e3f502acc","text":"RemoveField on indexed fields crashes on SQLite 3.35.5+ Description (last modified by cessor) Description I encountered the following error with django 4.1 in my Gitlab CI\/CD Pipeline. When I bumped django versions from 4.0.7 to 4.1. my pipeline broke during the testing stage; specifically during db migrations. I have not changed any other source code. Steps to reproduce Minimal example attached. Run make green to see that it works with 4.0.7, run make red to see that it does not work with 4.1. It will build and exercise a docker container which installs all dependencies in isolation and sets up an example django app and run migrations. Manual steps: Install django 4.1 Create a new project Create an app Install app in project Create a model Add field on model, set dbindex=True Make migrations: $ python makemigrations Remove field from model Make migrations: $ python makemigrations Apply migrations: $ python migrate The migration fails with the following error (for an app called web, with a model called Entity with a field called attribute for example): Running migrations: Applying contenttypes.0001initial... OK ... Applying sessions.0001initial... OK Applying web.0001initial... OK Applying web.0002removeentityattribute...Traceback (most recent call last): File \"\/usr\/local\/lib\/python3.10\/site-\", line 89, in execute return (sql, params) File \"\/usr\/local\/lib\/python3.10\/site-\", line 357, in execute return Database.Cursor.execute(self, query, params) sqlite3.OperationalError: error in index webentityattributed22c3fcb after drop column: no such column: attribute Details The above steps create a set of migrations where at the end a RemoveField migration is produced. Applying this migration fails for fields which had dbindex=True. The example I attached uses a SlugField where db_index defaults to True, setting this parameter to False will apply the migration without this error. I reproduced the error with the following field types: TextField, IntegerField, SlugField, CharField, URLField"}
{"_id":"q-en-django__django-15930-bc2633b738505539ea01100074e1d68a3ff34126dfc5571c0615a444d1a9c52e","text":"Case() crashes with ~Q(pkin=[]). Description The following code generates a syntax error. User.objects.annotate( a=Case( When(~Q(pkin=[]), then=Value(True)), default=Value(False), outputfield=BooleanField(), ) ).orderby(\"-a\").values(\"pk\") The error is: ProgrammingError: syntax error at or near \"THEN\" LINE 1: \".\"id\" FROM \"usersuser\" ORDER BY CASE WHEN THEN true ... The generated SQL is: SELECT \"usersuser\".\"id\" FROM \"users_user\" ORDER BY CASE WHEN THEN True ELSE False END ASC I expected behavior to annotate all rows with the value True since they all match. Relevant because ~Q(pkin=[]) is a sentinel value that is sometimes returned by application code."}
{"_id":"q-en-django__django-15957-746d91aea04971f9e39791675317ad7a65fcc534d363f8bc6c03c801fd56f932","text":"Prefetch objects don't work with slices Description ​Prefetch() objects does not work with sliced querysets. For example the following code results in AssertionError: Cannot filter a query once a slice has been taken.: Category.objects.prefetchrelated(Prefetch( 'postset', ()[:3], toattr='exampleposts', )) This behavior is also mentioned in ​this StackOverflow answer. On the other hand it does not seem to be documented in Django Docs. Why is it needed? My use case seems to be a common one: I want to display a list of categories while displaying couple of example objects from each category next to it. If I'm not mistaken there isn't currently an efficient way of doing this. Prefetching without slicing would prefetch all objects (and there may be thousands of them) instead of the three examples that are needed."}
{"_id":"q-en-django__django-15969-eaa82d7777a59677505fb9656e806bd5a42b1987c55597deaccfbf551aa0239d","text":"Performance issues with on large tables Description Hello, I have the following models configuration: Parent model Child model, with a parentid foreign key to a Parent model, set with ondelete=models.SETNULL Each Parent can have a lot of children, in my case roughly 30k. I'm starting to encounter performance issues that make my jobs timeout, because the SQL queries simply timeout. I've enabled query logging, and noticed something weird (that is certainly that way on purpose, but I don't understand why). SELECT FROM \"parent\" WHERE \"parent\".\"id\" = 'parent123'; SELECT FROM \"children\" WHERE \"children\".\"parentid\" IN ('parent123'); UPDATE \"children\" SET \"parentid\" = NULL WHERE \"children\".\"id\" IN ('child1', 'child2', 'child3', ..., 'child30000'); DELETE FROM \"parent\" WHERE \"parent\".\"id\" IN ('parent123'); I would have expected the update condition to simply be WHERE \"children\".\"parentid\" = 'parent123', but for some reason it isn't. In the meantime, I'll switch to on_delete=models.CASCADE, which in my case does the trick, but I was curious about the reason why this happens in the first place. Thanks in advance"}
{"_id":"q-en-django__django-15973-d49298cd92549b101da63cc0d8805c438ee6a48ed56abb172282489d7e9e30c7","text":"Defining the \"through\" model in a many-to-many field in another app causes \"AttributeError: 'str' object has no attribute 'meta'\" on migration Description I tried migrating my apps into the database, the three relevant apps are called: \"fonte\", \"fontevariavel\" and \"variavel\". fonte and variavel models have a many-to-many relationship (field being defined on \"fonte\"). The many-to-many field uses fontevariavel model as the \"through\" argument. Below are the models when I define them on separate apps. class FonteModel(Model): nome = TextField(unique=True) descricao = TextField() datainicial = DateField() datafinal = DateField(blank=True, null=True) variaveis = ManyToManyField(\"variavel.VariavelModel\", through=\"fontevariavel.FonteVariavelModel\") def str(self): return class Meta: dbtable = \"fontes\" verbosename = \"Fonte\" verbosenameplural = \"Fontes\" class VariavelModel(Model): nome = TextField(unique=True) descricao = TextField() class Meta: dbtable = 'variaveis' verbosename = 'Variável' verbosenameplural = 'Variáveis' class FonteVariavelModel(Model): variavel = ForeignKey('variavel.VariavelModel', ondelete=CASCADE) fonte = ForeignKey('fonte.FonteModel', ondelete=CASCADE) class Meta: dbtable = 'fontesvariaveis' verbosename = 'Fonte' verbosenameplural = 'Fontes' Generated migration file for Fonte from import migrations, models class Migration(migrations.Migration): initial = True dependencies = [ ('variavel', 'first'), ] operations = [ migrations.CreateModel( name='FonteModel', fields=[ ('id', models.BigAutoField(autocreated=True, primarykey=True, serialize=False, verbosename='ID')), ('nome', models.TextField(unique=True)), ('descricao', models.TextField()), ('datainicial', models.DateField()), ('datafinal', models.DateField(blank=True, null=True)), ('variaveis', models.ManyToManyField(through='fontevariavel.FonteVariavelModel', to='variavel.variavelmodel')), ], options={ 'verbosename': 'Fonte', 'verbosenameplural': 'Fontes', 'dbtable': 'fontes', }, ), ] If I put \"fontevariavel\" model inside \"fonte\"'s , it works, but if I do the same for \"variavel\" and continue having FonteVariavelModel in a different app, it continues not working, so the problem must be with exclusively with the ManyToMany intermediary model. Here is the trace: Applying fonte.0001initial...Traceback (most recent call last): File \"\/home\/elysium\/tutes\/django-test-stuff\/django-\", line 22, in <modulemain() File \"\/home\/elysium\/tutes\/django-test-stuff\/django-\", line 18, in main executefromcommandline() File \"\", line 446, in e xecutefromcommandline utility.execute() File \"\", line 440, in e xecute self.fetchcommand(subcommand).runfromargv() File \"\", line 402, in runf romargv self.execute(args, cmdoptions) File \"\", line 448, in execu te output = (args, options) File \"\", line 96, in wrappe d res = handlefunc(args, *kwargs) File \"\", line 3 49, in handle postmigratestate = executor.migrate( File \"\", line 135, in mig rate state = self.migrateallforwards( File \"\", line 167, in mi grateallforwards state = self.applymigration( File \"\", line 252, in app lymigration state = (state, schemaeditor) File \"\", line 130, in ap ply operation.databaseforwards( File \"\", line 96 , in databaseforwards schemaeditor.createmodel(model) File \"\", line 453, in cr eatemodel if AttributeError: 'str' object has no attribute 'meta' Putting everything in the same file also works."}
{"_id":"q-en-django__django-15987-08d3cd1bc4a2a93752886a38f288086748105b0392f850a164bacf93570a6963","text":"Fixture dirs duplicates undetected if dir is Path instance Description When FIXTURE_DIRS contains Path instances, the duplicate check in loaddata does not detect duplicates."}
{"_id":"q-en-django__django-15993-f773abbe3e6c748798ed51df65fb1bf6209324b40b5e9dc3af569e2e6079d01e","text":"RenameModel with dbtable should be a noop. Description A RenameModel operation that already has dbtable defined must be a noop. In Postgres, it drops and recreates foreign key constraints. In sqlite it recreates the table (as expected for a table renaming)."}
{"_id":"q-en-django__django-15995-29d1ea2f7b50711b90a9bf12d9b04c0788c8224b5fce289dcf198cc90a0e82a9","text":"Too aggressive pk control in createreversemanytoonemanager Description In the context of , Django now requires an instance pk to even instanciate a related manager []. Now I have a use case where I need to introspect the model used by a related manager (MyModel()) and Django 4.1 refuses that with ValueError: 'MyModel' instance needs to have a primary key value before this relationship can be used. My opinion is that is is too aggressive of a check and would suggest to let the init succeed even if the instance has no pk. Other calls to checkfk_val in the class seems sufficient to me to safeguard against shooting in the foot."}
{"_id":"q-en-django__django-15996-c1dbd910c94562a8e9168f3780969171a65effab1ba206d1d936174b73b1b0d1","text":"Support for serialization of combination of Enum flags. Description (last modified by Willem Van Onsem) If we work with a field: regexflags = models.IntegerField(default=re.UNICODE | re.IGNORECASE) This is turned into a migration with: default=re.RegexFlag[None] This is due to the fact that the EnumSerializer aims to work with the .name of the item, but if there is no single item for the given value, then there is no such name. In that case, we can use enum.decompose to obtain a list of names, and create an expression to create the enum value by \"ORing\" the items together."}
{"_id":"q-en-django__django-16002-bfc99e27bef7c9496f4b809febf95fe23f64a1f9e100c243e16cac4b6f0713fa","text":"\"NaN\" can be stored in DecimalField but cannot be retrieved Description (last modified by Xabier Bello) Same as ticket , but I managed to trigger it anyway: Steps to reproduce Create a brand new project using python 3.10 and django 4.1 with the default sqlite3 backend. Create a model with a DecimalField: class MyModel(models.Model): value = models.DecimalField(maxdigits=10, decimalplaces=5) Programmatically create a model instance with value=\"nan\", obj = (value=\"nan\") () Then try to retrieve the object from the database (or refresh from database): (pk=1) Traceback Traceback (most recent call last): File \"\", line 9, in (pk=1) File \"\/lib64\/python3.10\/site-\", line 85, in managermethod return getattr(self.getqueryset(), name)(args, *kwargs) File \"\/lib64\/python3.10\/site-\", line 646, in get num = len(clone) File \"\/lib64\/python3.10\/site-\", line 376, in len self.fetchall() File \"\/lib64\/python3.10\/site-\", line 1866, in fetchall self.resultcache = list(self.iterableclass(self)) File \"\/lib64\/python3.10\/site-\", line 117, in iter for row in compiler.resultsiter(results): File \"\/lib64\/python3.10\/site-\", line 1333, in applyconverters value = converter(value, expression, connection) File \"\/lib64\/python3.10\/site-\", line 344, in converter return createdecimal(value).quantize( TypeError: argument must be int or float The value \"nan\" (and maybe \"inf\" also) skip the validation in DecimalField.topython, because is not None, and is not instance of float. But decimal.Decimal(\"nan\") works without triggering the exception, so NaN gets stored in the DB."}
{"_id":"q-en-django__django-16027-1417b41833a860bf8e807f32118ff2ada7a53b632191be5960bd095751f789c4","text":"timesince - wrong results for 11 months + several weeks Description (last modified by אורי) Hi, I'm using timesince to format how much time passed since the user last visited my website. The code is: (\"On {date} ({timesince} ago)\").format( date=formats.dateformat(value=lastvisitdate), timesince=timesince(d=lastvisitdate, now=today) ) Now I created a test to test these times, and I noticed that for a year minus a week, the result is \"(11\\u00A0months, 4\\u00A0weeks ago)\" (why the \"\\u00A0\" and not a space?), and for a year minus 2 weeks, the result is \"(11\\u00A0months, 3\\u00A0weeks ago)\": user18 = ActiveUserFactory() user18.profile.lastvisit -= (relativedelta(years=1) - relativedelta(weeks=1)) user18.saveuserandprofile() self.assertIs(expr1={'en': \"(11\\u00A0months, 4\\u00A0weeks ago)\", 'he': \"(לפני 11\\u00A0חודשים, 4\\u00A0שבועות)\"}[self.languagecode] in user18.profile.lastvisitstr, expr2=True) user19 = ActiveUserFactory() user19.profile.lastvisit -= (relativedelta(years=1) - relativedelta(weeks=2)) user19.saveuserandprofile() self.assertIs(expr1={'en': \"(11\\u00A0months, 3\\u00A0weeks ago)\", 'he': \"(לפני 11\\u00A0חודשים, 3\\u00A0שבועות)\"}[self.languagecode] in user19.profile.lastvisitstr, expr2=True) Now, a year is 365 days, a year minus one week is 358 days, which is 11 months and 3 weeks. I think the problem is because each month is considered as 30 days, so 11 months are 330 days. But 11 months are about 334 days actually, so we receive a result of 11 months and 4 weeks, instead of 11 months and 3 weeks. A fix would be to change the number of days in a month to 30.4 (the average), optionally only for more than 2 months (because it makes sense to calculate exactly 30 days for the first 2 months). Also, it's important to calculate the number of days in 11 (or any number) of months as an integer, so that the result will not display hours and minutes (if depth is big enough)."}
{"_id":"q-en-django__django-16032-40317e94063cb4d90f5cc1b4b3efc97decda6de33598429dfed98c612fcc6880","text":"in doesn't clear selected fields on the RHS when () is used after annotate(). Description Here is a test case to reproduce the bug, you can add this in def testannotationandaliasfilterinsubquery(self): longbooksqs = ( ( pagesgt=400, ) .annotate(bookannotate=Value(1)) .alias(bookalias=Value(1)) ) publisherbooksqs = ( ( book_in=longbooksqs ) .values(\"name\") ) self.assertCountEqual( publisherbooks_qs, [ {'name': 'Apress'}, {'name': 'Sams'}, {'name': 'Prentice Hall'}, {'name': 'Morgan Kaufmann'} ] ) You should get this error: sub-select returns 10 columns - expected 1"}
{"_id":"q-en-django__django-16037-3e6f46b83f21c03dd8ace11f19d392b188b2c623c27f62aad9a27f2c63d97cd9","text":"QuerySet.annotate() with subquery and aggregation crashes. Description I updated one app from 4.0 to 4.1.1 and found a issue with one annotation using Count. I tested with SQLite and PostgreSQL, and both raised exception. The same app works with 4.0.7. Exception with SQLite: sub-select returns 13 columns - expected 1 Traceback (most recent call last): File \"\", line 89, in execute return (sql, params) File \"\", line 357, in execute return Database.Cursor.execute(self, query, params) sqlite3.OperationalError: sub-select returns 13 columns - expected 1 The above exception was the direct cause of the following exception: Traceback (most recent call last): File \"\", line 1225, in exists return () File \"\", line 592, in hasresults return compiler.hasresults() File \"\", line 1363, in hasresults return bool(self.executesql(SINGLE)) File \"\", line 1395, in executesql cursor.execute(sql, params) File \"\", line 103, in execute return super().execute(sql, params) File \"\", line 67, in execute return self.executewithwrappers( File \"\", line 80, in executewithwrappers return executor(sql, params, many, context) File \"\", line 84, in execute with File \"\", line 91, in exit raise djexcvalue.withtraceback(traceback) from excvalue File \"\", line 89, in execute return (sql, params) File \"\", line 357, in execute return Database.Cursor.execute(self, query, params) sub-select returns 13 columns - expected 1 Exception with Postgres: subquery must return only one column LINE 1: \".\"postid\") GROUP BY \"networkpost\".\"id\", (SELECT U0... ^ Traceback (most recent call last): File \"\", line 89, in execute return (sql, params) subquery must return only one column LINE 1: \".\"postid\") GROUP BY \"networkpost\".\"id\", (SELECT U0... ^ The above exception was the direct cause of the following exception: Traceback (most recent call last): File \"\", line 1225, in exists return () File \"\", line 592, in hasresults return compiler.hasresults() File \"\", line 1363, in hasresults return bool(self.executesql(SINGLE)) File \"\", line 1395, in executesql cursor.execute(sql, params) File \"\", line 103, in execute return super().execute(sql, params) File \"\", line 67, in execute return self.executewithwrappers( File \"\", line 80, in executewithwrappers return executor(sql, params, many, context) File \"\", line 84, in execute with File \"\", line 91, in exit raise djexcvalue.withtraceback(traceback) from excvalue File \"\", line 89, in execute return (sql, params) subquery must return only one column LINE 1: \".\"postid\") GROUP BY \"networkpost\".\"id\", (SELECT U0... ^ The exception is raised by annotate(likes=Count(\"likedby\")) in method fetchallposts. class PostManager(models.Manager): def requestdata(self, requestuser): likedbyuser = Value(False) isfollowing = Value(False) isowner = Case(When(, then=True), default=False) if requestuser.isauthenticated: # Check if the user has liked the post in each row of the query likedbyuser = Exists((id=OuterRef(\"id\"))) isfollowing = Exists( (id=OuterRef(\"userid\")) ) return isowner, likedbyuser, isfollowing def fetchallposts(self, requestuser) -QuerySet[Post]: isowner, likedbyuser, isfollowing = self.requestdata(requestuser) return ( self.selectrelated() .prefetchrelated( Prefetch( \"comments\", queryset=Comment.objects.selectrelated().filter(reply=False), ), # filter related \"comments\" inside the post QuerySet ) .orderby(\"-publicationdate\") .annotate(isfollowing=isfollowing) .annotate(isowner=isowner) .annotate(likes=Count(\"likedby\")) # Doesn't work on 4.1 .annotate(likedbyuser=likedbyuser) ) def fetchfollowingposts(self, requestuser: User) -QuerySet[Post]: return self.fetchallposts(requestuser).filter( () ) Models class User(AbstractUser): id: int posts: RelatedManager[Post] likedposts: RelatedManager[Post] comments: RelatedManager[Comment] about = models.CharField(blank=True, maxlength=255) photo = models.ImageField( blank=True, null=True, uploadto=uploadpath, validators=[filevalidator], ) following = models.ManyToManyField( \"self\", relatedname=\"followers\", symmetrical=False ) objects: CustomUserManager = CustomUserManager() # Related fields # posts = ManyToOne(\"Post\", relatedname=\"user\") # likedposts = ManyToMany(\"Post\", relatedname=\"likedby\") # comments = ManyToOne(\"Comment\", relatedname=\"user\") def save(self, args, kwargs): \"\"\" fullclean is not called automatically on save by Django \"\"\" self.fullclean() super().save(args, kwargs) def str(self): return f\"{self.username}\" # type: ignore class Post(models.Model): id: int comments: RelatedManager[Comment] userid: int user = models.ForeignKey( settings.AUTHUSERMODEL, ondelete=models.CASCADE, relatedname=\"posts\" ) text = models.CharField(maxlength=200) publicationdate = models.DateTimeField(autonowadd=True) edited = models.BooleanField(default=False) lastmodified = models.DateTimeField(autonowadd=True) likedby = models.ManyToManyField( settings.AUTHUSERMODEL, relatedname=\"likedposts\", blank=True ) # Related Fields # comments = ManyToOne(\"Comment\", relatedname=\"post\") objects: PostManager = PostManager() class Meta: ordering = [\"-publicationdate\"] def str(self): return f\"{}\" class Comment(models.Model): id: int replies: RelatedManager[Comment] postid: int post = models.ForeignKey(Post, ondelete=models.CASCADE, relatedname=\"comments\") userid: int user = models.ForeignKey( settings.AUTHUSERMODEL, ondelete=models.CASCADE, relatedname=\"comments\" ) text = models.CharField(maxlength=200) publicationdate = models.DateTimeField(autonowadd=True) reply = models.BooleanField(default=False) parentcommentid: int parentcomment = models.ForeignKey( \"self\", ondelete=models.CASCADE, null=True, blank=True, relatedname=\"replies\" ) class Meta: ordering = [\"-publicationdate\"] def save(self, args, kwargs): self.fullclean() if self.parentcomment is not None: if != raise ValidationError(\"Parent comment must be from the same post.\") = True super().save(args, kwargs) def str__(self): return f\"{} - reply: {}\" # type: ignore"}
{"_id":"q-en-django__django-16041-abceca8f53d84c7abbeda9640053dbfc4f73af0ad3473fdbc4cb31f2330a877c","text":"Rendering emptyform crashes when emptypermitted is passed to formkwargs Description Issue When explicitly setting formkwargs = {'emptypermitted':True} or formkwargs = {'emptypermitted':False} , a KeyError occurs when rendering a template that uses a formset's emptyform. Expected Behavior emptypermitted is ignored for formset.emptyform since emptypermitted is irrelevant for emptyform, as emptyform is not meant to be used to pass data and therefore does not need to be validated. Steps to Reproduce from django.shortcuts import render from .models import MyModel def testview(request): context = {} ff = modelformsetfactory(MyModel, fields = ['afield']) context['formset'] = ff( queryset = (), formkwargs = {'emptypermitted':True} # or formkwargs = {'emptypermitted':False} ) return render(request, '', context) from import path, include from .views import testview urlpatterns = [ path('test', testview) ] {% extends \"\" %} {% block content %} <form id=\"my-form\" method=\"post\"{% csrftoken %} {{ formset }} <input type=\"submit\" value=\"Save\"<\/form{{ formset.emptyform }} {% endblock %}"}
{"_id":"q-en-django__django-16046-51382bdfe8c0323070ac835ab7a744b2756bd607fe89c396aed23b3666bd097b","text":"Fix \"string index out of range\" when null Description When: if strnumber[0] == \"-\" encounters a number field that's null when formatting for the admin listdisplay this causes an IndexError: string index out of range I can attach the proposed fix here, or open a pull request on GitHub if you like?"}
{"_id":"q-en-django__django-16053-5139c92b4233afca6394b7a98938bc836144d7f20f8aa6c6a8f03b9a03d6a42c","text":"ContentTypes and Permissions mix English with other languages in str Description The string representation of and was changed in commit ​ to solve . However, the str function mixes the model’s applabel (which is not localized) with the model’s verbosename (which is localized). This results in weirdly looking strings, part of which is in English and part in a different language, and maybe even different alphabet. The comment does not clarify why the application’s applabel was chosen and not verbosename (via (self.applabel).verbose_name). In my opinion, either the whole of the ContentType’s representation should use localized labels or none of them should be localized."}
{"_id":"q-en-django__django-16067-5dcbe237de4cfa48680d69b04720314d2bc46c69ce7999d1cf713d37faac9f24","text":"DecimalValidator fails to validate 0 in scientific notation (0E+1 or 0E+2) Description (last modified by Shiplu Mokaddim) In 1.11.29 DecimalValidator treats 0E+1 as valid decimal from import DecimalField DecimalField(maxdigits=8, decimalplaces=7).clean('0E+1') Decimal('0E+1') But in 2.0.13 it raises ValidatorError. DecimalField(maxdigits=8, decimalplaces=7).clean('0E+1') Traceback (most recent call last): File \"<stdin\", line 1, in <moduleFile \"\/Users\/amokaddim\/e\/fh\/lib\/python3.8\/site-\", line 150, in clean self.runvalidators(value) File \"\/Users\/amokaddim\/e\/fh\/lib\/python3.8\/site-\", line 141, in runvalidators raise ValidationError(errors) <exception str() failedThis was introduced in ​ Is there any way I can prevent this validation error from happening in Django 2.0? Any flag or option that will prevent this? Calling Decimal('0E+1').normalize() inside to_python method solves the problem. But that's a workaround! Here is a reproducible test case. from unittest import TestCase from decimal import Decimal from import DecimalValidator from import ValidationError class Test(TestCase): def test(self): value = '0E+1' validator = DecimalValidator(8, 7) try: validator(Decimal(value)) except ValidationError: (\"'{}' is an invalid decimal\".format(value))"}
{"_id":"q-en-django__django-16070-b09a0c00ac31df1a3873212efe73431c932b225f309395e50354f4faf6d440a6","text":"QuerySet.orderby() silently skips non-existing fields on related fields with Meta.ordering. Description Compare the following desirable behavior: ().orderby(\"nonexistingfield\") Traceback (most recent call last): File \"<console\", line 1, in <moduleFile \"\", line 1149, in orderby (*fieldnames) File \"\", line 2016, in addordering self.namestopath((LOOKUPSEP), ) File \"\", line 1562, in namestopath raise FieldError(\"Cannot resolve keyword '%s' into field. \" Cannot resolve keyword 'nonexistingfield' into field. Choices are: [redacted] with the following undesirable behavior: ().orderby(\"someforeignkeynonexisting_field\") <QuerySet .... [ i.e. shows some results ]"}
{"_id":"q-en-django__django-16072-d3cfbd887394194ea89479511b9ecc3bec159d19cae51e78c8dd60cf5647034f","text":"updateorcreate should utilize updatefields on update Description updateorcreate should update only the fields in default on update, not all fields. While it is concurrency-safe to update the whole model since updateorcreate fetches the object via selectforupdate it is still unnecessary to re transmit all fields back to the database (probably results in more wal\/binary logs written etc etc…). In the end updateorcreate (and most likely getorcreate) might be written in more efficient ways depending on the database backend -- but updatefields seems to be a rather low-hanging fruit."}
{"_id":"q-en-django__django-16076-1e3d38a6f813cca258324473e6abc9dd1a5e34082cbd4c5b8e2870a3532857a8","text":"Registering lookups on relation fields should be supported. Description (last modified by Thomas) Hello, I have a model, let's call it Parent, with a field called objectid. I have another model, let's call it Child, which has a ForeignKey field called parentobject[id] pointing to Parent.objectid. I need to do a lookup on Child where the FK starts with a certain character (it's a normalized value so, in the context of my app, it makes sense... also, I didn't design this schema and changing it is not a possibility ATM). The problem is that if I do: qs = (parentobjectidstartswith='c') I get: Related Field got invalid lookup: startswith The only way I could make it work is: qs = (parentobjectobjectidstartswith='c') but it forces a join between the table and the view and that's a no-no in my case (way too costly). Here's the MCVE (tested on Python 3.9 + Django 4.0.7 and Python 3.10 + Django 4.1.1): import django () from import models class Parent(models.Model): class Meta: applabel = 'test' objectid = models.CharField('Object ID', maxlength=20, unique=True) class Child(models.Model): class Meta: applabel = 'test' parentobject = models.ForeignKey( Parent, tofield='objectid', relatedname='%(class)sset', ondelete=models.CASCADE ) if name == 'main': qs = (parentobjectidstartswith='c') # fails with qs = (parentobjectobjectidstartswith='c') # works but forces a costly join And the error: Traceback (most recent call last): File \"\", line 26, in <moduleqs = (parentobjectidstartswith='c') File \"\/opt\/src\/venv\/lib\/python3.10\/site-\", line 85, in managermethod return getattr(self.getqueryset(), name)(args, kwargs) File \"\/opt\/src\/venv\/lib\/python3.10\/site-\", line 1420, in filter return self.filterorexclude(False, args, kwargs) File \"\/opt\/src\/venv\/lib\/python3.10\/site-\", line 1438, in filterorexclude clone.filterorexcludeinplace(negate, args, kwargs) File \"\/opt\/src\/venv\/lib\/python3.10\/site-\", line 1445, in filterorexcludeinplace (Q(args, kwargs)) File \"\/opt\/src\/venv\/lib\/python3.10\/site-\", line 1532, in addq clause, = (qobject, self.usedaliases) File \"\/opt\/src\/venv\/lib\/python3.10\/site-\", line 1562, in addq childclause, neededinner = self.buildfilter( File \"\/opt\/src\/venv\/lib\/python3.10\/site-\", line 1478, in buildfilter condition = self.buildlookup(lookups, col, value) File \"\/opt\/src\/venv\/lib\/python3.10\/site-\", line 1292, in build_lookup raise FieldError( Related Field got invalid lookup: startswith Thanks for your help, Regards,"}
{"_id":"q-en-django__django-16082-de4cfd14b9bb11c2ac61b81529eba4fe9f62fd35d5a572338e1b0b75977d1ad4","text":"Resolve output_field when combining numeric expressions with MOD operator. Description When writing a Django expression for a query that does MOD, if the types of the query are different (Decimal and Integer), it doesn't resolve the result to a Decimal type, like it does for other mathematical operators."}
{"_id":"q-en-django__django-16100-18607ba748f421ce927b3ae88186a14a22d204e179ded95cca45c6c3f15d1737","text":"Add transaction handling to Changelist listeditable processing. Description It seems that changelistview in Django admin is missing a transaction. Since the view may change data in database, it should be wrapped in a transaction to prevent unexpected states in case of errors."}
{"_id":"q-en-django__django-16111-e045908f93b3b64d05e0c793f0e3cee9f85af0a6ac2ee491238abeff191efa26","text":"Add support for microseconds to Now() on MySQL and SQLite. Description Add support for microseconds to Now() on MySQL and SQLite. ​PR"}
{"_id":"q-en-django__django-16116-02b1f7ceca3d02424e194db3d56f943da487313767dd2ad6c11ad08db52e467d","text":"makemigrations --check generating migrations is inconsistent with other uses of --check Description To script a check for missing migrations but without actually intending to create the migrations, it is necessary to use both --check and --dry-run, which is inconsistent with migrate --check and optimizemigration --check, which just exit (after possibly logging a bit). I'm suggesting that makemigrations --check should just exit without making migrations. The choice to write the migrations anyway was not discussed AFAICT on ticket: or ​ Noticed when reading ​PR to adjust the documentation of migrate --check. I think the current documentation is silent on this question."}
{"_id":"q-en-django__django-16117-9ac8ad349b4c5f4aeb6254f4aa748f702eddeb1b1a31ef4fc3301eb8397ce733","text":"Generated migration file is not detected by django because of the name of newly generated migration file Description After a new project is created we run: python migrate to sync with our auth database models. Lets create an app called myapp, register this app on settings. Now we create a model inside myapp as: class MyModel(models.Model): name = models.CharField(maxlength=100) age = models.IntegerField() Lets run makemigrations and migrate command for the app. This will generate a migration file with name inside myapp\/migrations\/ folder. Now let us create a constraints for the model as: class MyModel(models.Model): name = models.CharField(maxlength=100) age = models.IntegerField() class Meta: constraints = [ models.CheckConstraint( check=models.Q(agegte=1), name=\"Age should not \" ) ] When we CheckConstraint for our previous model, we need to run makemigrations and migrate to reflect the new changes on database. Running makemigrations will generate a migration file inside user\/migrations\/ with name 0002mymodelage should not When we try to run migrate, django will not be able to detect this file and NO changes will be applied to existing database. If we run makemigrations again, the same file will be generated again. No matter how many time we run makemigrations, the same file will be generated again and again. The newly generated migration file is not detected by django migrate and showmigrations command. This is because of the name of migration file. The new migration file name contains serveral dots and because of this name, migrate\/showmigration is not able to find the file. There are mutiple solutions for this: First: Renaming the generated migration file. Remove . (dot) (excluding the .py)from the newly generated migration file. This will make sure that the file is recognized as python file by django migrate\/showmigrations command. Second: We can change the name of our constraint name inside the migration as : class Meta: constraints = [ models.CheckConstraint( check=models.Q(agegte=1), name=\"Age should not be less than one\" ) ] The only difference between the constraint that was above and the constraint that is implemented now is in the value of name parameter for models.Q(). We removed the exisiting \"dot\" . from the name. Running makemigration will generate a migration file with name 0002mymodelage should not be less than Now running the migrate\/showmigrations command will find the new migration file. This example is also commited in the repository​ One can look at the commit history and checkout to proper stage as needed , described by commit message on the repository commit history. Third: I have replaced the name of the migration file in django source code so that .(dot) gets replaced by some character. I am also attaching the associated pull request ​ Using third approach makes sure that no one has to go through the step First and Second as described above. The solution would work out of the box."}
{"_id":"q-en-django__django-16120-1b22a8d6a84fc318c5b9d535c63f22f4f84da59a3b121da3a482e7991323e161","text":"migrate --check still emits signals if database is up to date Description (last modified by Jacob Walls) premigrate and postmigrate signals are emitted for migrate --check, but only if the database is up-to-date. A related side effect is that the logs also look like a like a real run, as it says \"Operations to perform:\" and \"Running migrations:\". The last sentence clarifies that nothing has been applied, but there is still the potential for a half-second of \"what have I just done?\". % python migrate --check Operations to perform: Apply all migrations: admin, auth, contenttypes, sessions, social_django Running migrations: No migrations to apply."}
{"_id":"q-en-django__django-16136-d72f85cea4f7f8d7d0dd89f926c7b0052a702b559ef7fc042dbe5dc3b0584b91","text":"object HttpResponseNotAllowed can't be used in 'await' expression Description When defining a simple View subclass with only an async \"post\" method, GET requests to this view cause the following exception: [29\/Sep\/2022 07:50:48] \"GET \/demo HTTP\/1.1\" 500 Method Not Allowed (GET): \/demo Internal Server Error: \/demo Traceback (most recent call last): File \"\", line 55, in inner response = getresponse(request) File \"\", line 197, in getresponse response = wrappedcallback(request, callbackargs, callbackkwargs) File \"\", line 218, in call return () File \"\", line 451, in result return self.getresult() File \"\", line 403, in getresult raise self.exception File \"\", line 284, in mainwrap result = await self.awaitable(args, kwargs) TypeError: object HttpResponseNotAllowed can't be used in 'await' expression This can be easily reproduced with an empty project (no external dependencies) started with Django 4.1.1 and python 3.10.6. Basic view to reproduce the bug: from import View from import HttpResponse class Demo(View): \"\"\"This basic view supports only POST requests\"\"\" async def post(self, request): return HttpResponse(\"ok\") URL pattern to access it: from import path from views import Demo urlpatterns = [ path(\"demo\", ()), ] Start the local dev server ( runserver) and open ​http:\/\/127.0.0.1:8000\/demo in the browser. Server crash with 500 error with the given traceback."}
{"_id":"q-en-django__django-16139-51f0c4ce21e43704a83dc18244baad3fbffc57ea0d5845f28ba84881c0f31050","text":"Accessing UserAdmin via tofield leads to link to PasswordResetForm being broken (404) Description (last modified by Simon Kern) Accessing the UserAdmin via another model's Admin that has a reference to User (with tofield set, e.g., tofield=\"uuid\") leads to the UserAdmin being accessed via an url that looks similar to this one: ...\/user\/-3333-4444-5555-\/change\/?tofield=uuid However the underlying form looks like this: Code highlighting: class UserChangeForm(forms.ModelForm): password = ReadOnlyPasswordHashField( label=(\"Password\"), helptext=( \"Raw passwords are not stored, so there is no way to see this \" \"user’s password, but you can change the password using \" '<a href=\"{}\"this form<\/a.' ), ) ... ... def init(self, args, kwargs): super().init(args, kwargs) password = (\"password\") if password: password.helptext = (\"..\/password\/\") ... ... This results in the link to the PasswordResetForm being wrong and thus ending up in a 404. If we drop the assumption that UserAdmin is always accessed via its pk, then we're good to go. It's as simple as replacing password.helptext = (\"..\/password\/\") with password.helptext = (f\"..\/..\/{}\/password\/\") I've opened a pull request on GitHub for this Ticket, please see: ​PR"}
{"_id":"q-en-django__django-16142-0d49e576d87b1d0421482f5c345905f37412426f013b8cb30a14a6072f73c901","text":"getlanguagefromrequest should not fallback to settings.LANGUAGECODE Description (last modified by sergioisidoro) I'm writing a middleware to fallback to a different language depending on the TLD of the domain of the HTTPHOST However, I noticed that getlanguagefromrequest falls back to the settings default language, which will almost always take precedence in this case. This is quite confusing, since settings.LANGUAGECODE is not \"from the request\", but from the application configuration, and it feels that the responsibility of falling back to the default language should lie in the Middleware, not in this function. Solution \/ Summary: getlanguagefromrequest should return None, to communicate to the middleware that there was no language from request, and that the middleware should fallback to the default. Otherwise if the getlanguagefromrequest returns \"EN\" we don't know if \"EN\" is actually a request preference, or because it came from the default settings.LANGUAGECODE"}
{"_id":"q-en-django__django-16143-8035311e287ab0dfbea534c78d0671348366ce792422000a472d37c0518d6c55","text":"TransactionManagmentError isn't helpful for tracking down cause Description The error: \"TransactionManagementError: An error occurred in the current transaction. You can't execute queries until the end of the 'atomic' block.\" thrown from \", line 372, in validatenobrokentransaction doesn't provide enough information to help track down the problem. The exception is thrown if self.needsrollback is True, but the underlying reason that self.needsrollback has been set True has been lost, since it could have happened a long time previously. Transactions should keep track of why needsrollback has been set, and use that to provide a more helpful error. Background: I'm seeing this error being thrown when a task is run on a Celery queue, but not when run manually. Since it's via Celery, dropping into the debugger is impossible, and thus I'm back to trying to dump useful debug info."}
{"_id":"q-en-django__django-16145-350c8add27f9e6088507323fab7c0a25644aeac0393dc0cb3315232f6ec92269","text":"'s \"Starting development server at <address\" doesn't work Description According to ​tutorial running python runserver 0:8000 is the same as python runserver 0.0.0.0:8000 but it's output $ python runserver 0:8000 Watching for file changes with StatReloader ... Starting development server at http:\/\/0:8000\/ ... So that you can't use link \"​http:\/\/0:8000\/\" in your browser. Output should be \"Starting development server at ​http:\/\/0.0.0.0:8000\/\" when providing \"0:8000\" in command line in order to stay consistent with docs."}
{"_id":"q-en-django__django-16208-ac419dfcafea2ec2ae2784dfcb8223292e2a798953ffe4fe1088b36c51a83293","text":"Log when DB transactions are commited and rollbacked. Description Background: I was debugging database calls today with the log. Problem: The BEGIN SQL calls show up in the logs, but there is no way to see when the transaction is commited or if it is rolled back. As easy solution would be to log commits and rollbacks as well."}
{"_id":"q-en-django__django-16229-9466e53986769fcaa783c34cb6114bde36928f7fe42bdcb2ddf66621888360fd","text":"ModelForm fields with callable defaults don't correctly propagate default values Description When creating an object via the admin, if an inline contains an ArrayField in error, the validation will be bypassed (and the inline dismissed) if we submit the form a second time (without modification). go to \/admin\/myapp\/thing\/add\/ type anything in plop submit -it shows an error on the inline submit again -no errors, plop become unfilled class Thing(models.Model): pass class RelatedModel(models.Model): thing = models.ForeignKey(Thing, ondelete=models.CASCADE) plop = ArrayField( models.CharField(maxlength=42), default=list, ) class RelatedModelForm(forms.ModelForm): def clean(self): raise ValidationError(\"whatever\") class RelatedModelInline(admin.TabularInline): form = RelatedModelForm model = RelatedModel extra = 1 class ThingAdmin(admin.ModelAdmin): inlines = [ RelatedModelInline ] It seems related to the hidden input containing the initial value: <input type=\"hidden\" name=\"initial-relatedmodelset-0-plop\" value=\"test\" id=\"initial-relatedmodelset-0-idrelatedmodelset-0-plop\"I can fix the issue locally by forcing showhidden_initial=False on the field (in the form init)"}
{"_id":"q-en-django__django-16254-760ed2fd4d50d58beb2aa29439251d1a7b1b6c9ad10d15d8b01c33e9bdc6aaf3","text":"Adding ManyToManyField on SQLite rebuilds table. Description Hey there, While updating the ​django-migration-linter for Django 4.1 (yeah, a bit late to the party, but I was busy eh :P), I bumped into what seems to be a regression. On SQLite, when adding a ManyToManyField to a table, it seems to rebuild the table - whereas it did not in Django 4.0. From my understanding, rebuilding the table is not necessary in this case. Or perhaps I'm missing something. Steps to reproduce: 1\/ Before models: class A(models.Model): pass class B(models.Model): pass (with it's boring migration) 2\/ After models: class A(models.Model): pass class B(models.Model): manytomany = models.ManyToManyField(A) Which, expectedly, generates the migration: from import migrations, models class Migration(migrations.Migration): dependencies = [(\"appaddmanytomanyfield\", \"0001initial\")] operations = [ migrations.AddField( modelname=\"b\", name=\"manytomany\", field=models.ManyToManyField(to=\"appaddmanytomanyfield.A\"), ) ] All good up until here. Now the \"regression\", in Django 4.0, a sqlmigrate generates: BEGIN; -- Add field manytomany to b CREATE TABLE \"appaddmanytomanyfieldbmanytomany\" (\"id\" integer NOT NULL PRIMARY KEY AUTOINCREMENT, \"bid\" integer NOT NULL REFERENCES \"appaddmanytomanyfieldb\" (\"id\") DEFERRABLE INITIALLY DEFERRED, \"aid\" integer NOT NULL REFERENCES \"appaddmanytomanyfielda\" (\"id\") DEFERRABLE INITIALLY DEFERRED); CREATE UNIQUE INDEX \"appaddmanytomanyfieldbmanytomanybidaid3e15251duniq\" ON \"appaddmanytomanyfieldbmanytomany\" (\"bid\", \"aid\"); CREATE INDEX \"appaddmanytomanyfieldbmanytomanybid953b185b\" ON \"appaddmanytomanyfieldbmanytomany\" (\"bid\"); CREATE INDEX \"appaddmanytomanyfieldbmanytomanyaid4b44832a\" ON \"appaddmanytomanyfieldbmanytomany\" (\"aid\"); COMMIT; whereas in Django 4.1: BEGIN; -- Add field manytomany to b CREATE TABLE \"newappaddmanytomanyfieldb\" (\"id\" integer NOT NULL PRIMARY KEY AUTOINCREMENT, \"x\" integer NOT NULL); CREATE TABLE \"appaddmanytomanyfieldbmanytomany\" (\"id\" integer NOT NULL PRIMARY KEY AUTOINCREMENT, \"bid\" integer NOT NULL REFERENCES \"appaddmanytomanyfieldb\" (\"id\") DEFERRABLE INITIALLY DEFERRED, \"aid\" integer NOT NULL REFERENCES \"appaddmanytomanyfielda\" (\"id\") DEFERRABLE INITIALLY DEFERRED); INSERT INTO \"newappaddmanytomanyfieldb\" (\"id\", \"x\") SELECT \"id\", \"x\" FROM \"appaddmanytomanyfieldb\"; DROP TABLE \"appaddmanytomanyfieldb\"; ALTER TABLE \"new_appaddmanytomanyfieldb\" RENAME TO \"appaddmanytomanyfieldb\"; CREATE UNIQUE INDEX \"appaddmanytomanyfieldbmanytomanybidaid3e15251duniq\" ON \"appaddmanytomanyfieldbmanytomany\" (\"bid\", \"aid\"); CREATE INDEX \"appaddmanytomanyfieldbmanytomanybid953b185b\" ON \"appaddmanytomanyfieldbmanytomany\" (\"bid\"); CREATE INDEX \"appaddmanytomanyfieldbmanytomanyaid4b44832a\" ON \"appaddmanytomanyfieldbmanytomany\" (\"a_id\"); COMMIT; I could bisect it down to this commit (from and this ​PR). In the diff we see that the # Special-case implicit M2M tables comment and its code were removed. That's potentially a lead for a fix here I guess :) (On a side note, this commit introduced another regression . But that's not related to the issue at hand) Thank you!"}
{"_id":"q-en-django__django-16255-0c788a3906fcbe8d6fce7122c641babdcbfd7be0a547220553b1351bc8491eb8","text":"Sitemaps without items raise ValueError on callable lastmod. Description When sitemap contains not items, but supports returning lastmod for an item, it fails with a ValueError: Traceback (most recent call last): File \"\/usr\/local\/lib\/python3.10\/site-\", line 55, in inner response = getresponse(request) File \"\/usr\/local\/lib\/python3.10\/site-\", line 197, in getresponse response = wrappedcallback(request, callbackargs, callbackkwargs) File \"\/usr\/local\/lib\/python3.10\/site-\", line 133, in wrappedview response = viewfunc(request, args, kwargs) File \"\/usr\/local\/lib\/python3.10\/site-\", line 34, in inner response = func(request, args, *kwargs) File \"\/usr\/local\/lib\/python3.10\/site-\", line 76, in index sitelastmod = site.getlatestlastmod() File \"\/usr\/local\/lib\/python3.10\/site-\", line 170, in getlatestlastmod return max([self.lastmod(item) for item in ()]) Exception Type: ValueError at Exception Value: max() arg is an empty sequence Something like this might be a solution: def getlatestlastmod(self): if not hasattr(self, \"lastmod\"): return None if callable(self.lastmod): try: return max([self.lastmod(item) for item in ()]) except TypeError: except (TypeError, ValueError): return None else: return self.lastmod"}
{"_id":"q-en-django__django-16260-1a890d6dc1339f618ff7912e98fd7f07d877a09040e16ffe7f81b20fcfd04870","text":"model.refreshfromdb() doesn't clear cached generic foreign keys Description (last modified by pascal chambon) In my code, Users have a generic foreign key like so: controlledentitycontenttype = models.ForeignKey(ContentType, blank=True, null=True, ondelete=models.CASCADE) controlledentityobjectid = models.PositiveIntegerField(blank=True, null=True) controlledentity = GenericForeignKey(\"controlledentitycontenttype\", \"controlledentityobjectid\") However, in unit-tests, when I refresh a user instance, the controlledentity relation isn't cleared from cache, as can be seen here with IDs: oldcontrolledentity = authenticateduser.controlledentity authenticateduser.refreshfromdb() newcontrolledentity = authenticateduser.controlledentity assert id(oldcontrolledentity) != id(newcontrolledentity) # FAILS And this leads to subtle bugs like non-transitive equalities in tests : assert authenticateduser.controlledentity == fixtures.client1project2organization3 assert fixtures.client1project2organization3.getpricingplan() == pricingplan assert authenticateduser.controlledentity.getpricingplan() == pricingplan # FAILS Calling \"authenticateduser.controlledentity.refreshfromdb()\" solved this particular bug, but \"authenticateduser.refreshfromdb()\" isn't enough. Tested under Django3.2.13 but the code of refreshfromdb() hasn't changed since then in Git's main branch (except few cosmetic adjustments on code format). I'm a bit lost in the code of refreshfromdb(), but I've just seen that the generic relation appears once in this loop, but is not considered as \"cached\" in the if() branch. for field in #print(\"%% CLEARING RELATED FIELD\", field) if field.iscached(self): #print(\"%% DONE\") # not called field.deletecachedvalue(self)"}
{"_id":"q-en-django__django-16281-e394a25ae2303bad3c574b8815ba3c750b4113ed7aad728658c7fd8b602a9aa2","text":"Migration changing ManyToManyField target to 'self' doesn't work correctly Description Steps to reproduce: Create Models: class Bar(models.Model): pass class Foo(models.Model): bar = models.ManyToManyField('Bar', blank=True) Migrate: makemigrations app migrate Change type of the ManyToManyField to Foo: class Bar(models.Model): pass class Foo(models.Model): bar = models.ManyToManyField('Foo', blank=True) Migrate (see above) In the admin page, navigate to \"add Foo\", click save You should see an OperationalError, \"no such column: appfoobar.fromfooid\""}
{"_id":"q-en-django__django-16302-c6a375897b601335003576cbfd71baf13abe01d9a1e6f131c54f58aa37a1819d","text":"Cannot define CharField with unlimited length Description Model validation throws an error on CharField with a null maxlength: class Test(Model): charfield = CharField(maxlength=None) One or more models did not validate: \"charfield\": CharFields require a \"maxlength\" attribute that is a positive integer. CharField should allow maxlength=None, which intuitively means there is no maximum length. This is a perfectly valid use case. Postgres, for example, supports varchar\/text columns without a length limit, but Django appears to have no way to define such a column in a model class. The model validation code looks like this (​): if isinstance(f, models.CharField): try: maxlength = int(f.maxlength) if maxlength <= 0: e.add(opts, '\"%s\": CharFields require a \"maxlength\" attribute that is a positive integer.' % f.name) except (ValueError, TypeError): e.add(opts, '\"%s\": CharFields require a \"maxlength\" attribute that is a positive integer.' % f.name) It should be changed to something this: if isinstance(f, models.CharField) and f.maxlength is not None: ... The FileField does not happen to throw this error because it is not a derivative of CharField. However, the SQL generated for FileField is not correct when max_length=None, so that would need to be addressed as well."}
{"_id":"q-en-django__django-16306-c502a676371e057141f19e7b928d890b0e6dc24b06d47de770185faefc61f8e8","text":"BaseForm.getitem() does unneeded work in the happy path Description I noticed that in the \"happy path,\" BaseForm.getitem() does unneeded work: ​ It can just return self.boundfieldscache[name] at the beginning and handle KeyError, instead of accessing followed by checking for the presence of name in self.boundfieldscache before doing so each time."}
{"_id":"q-en-django__django-16311-527e0262ad1b2652d60df0a4e999bb07c47c4374e052d3efb4526cd71b7d488e","text":"Mitigate the BREACH attack Description (last modified by Nick Pope) The BREACH attack (​) was published in 2013. The Django project responded soon after (​) suggesting users to basically stop using gzip. CSRF masking was implemented in 2016 (). In April 2022, a paper called \"Heal The Breach\" was published, suggesting a mitigation that does not depend on masking specific tokens or injecting data into HTML. It is rather a generic and effective mitigation. It suggests adding randomness to the compressed response by injecting random bytes in the gzip filename field of the gzip stream: ​ Telling users to disable gzip is not great for bandwidth consumption. I propose that Django should implement \"Heal The Breach\" with sensible default."}
{"_id":"q-en-django__django-16315-04558d831dd5dce6b6960bf61ab516290dce2338fc5cc970e5dd2d6969005a4a","text":"QuerySet.bulkcreate() crashes on mixed case columns in uniquefields\/updatefields. Description Not sure exactly how to phrase this, but when I I'm calling bulkupdate on the manager for a class with dbcolumn set on fields the SQL is invalid. Ellipses indicate other fields excluded for clarity. class ActivityBlackListed(models.Model): \"\"\" Originally sourced from ActivityBlackListed in \/home\/josh\/PNDSInterimMIS- (13 records) \"\"\" class Meta: dbtable = \"ActivityBlackListed\" blacklistid = models.IntegerField(primarykey=True, dbcolumn=\"BlacklistID\") sectorid = models.IntegerField(null=True, blank=True, dbcolumn=\"SectorID\") ... qs.bulkcreate(instances, updateconflicts=True, updatefields=[\"sectorid\", ...], uniquefields=[\"blacklistid\"]) The \"INSERT\" code does take into account the dbcolumns INSERT INTO \"ActivityBlackListed\" (\"BlacklistID\",...) VALUES (%s, ...), The code which is generated for \"ON CONFLICT\" uses the field name and not the dbcolumn which leads to a syntax error 'ON CONFLICT(\"blacklistid\") DO UPDATE SET \"sectorid\" = EXCLUDED.\"sectorid\", ... PostgreSQL returns ERROR: column \"blacklistid\" does not exist at character 1508 What should be generated is I think: 'ON CONFLICT(\"BlacklistID\") DO UPDATE SET \"SectorID\" = EXCLUDED.\"SectorID\", ..."}
{"_id":"q-en-django__django-16317-dd5748135fcb2c6d02dc952ad7f295d41330432c7d0ba0eb6febc66ffc95e42f","text":"QuerySet.bulkcreate() crashes on \"pk\" in uniquefields. Description (last modified by Mariusz Felisiak) QuerySet.bulkcreate() crashes on \"pk\" in uniquefields which should be allowed. File \"\", line 89, in _execute return (sql, params) column \"pk\" does not exist LINE 1: ...S (3127, 3, 3, 'c'), (3128, 4, 4, 'd') ON CONFLICT(\"pk\") DO ... Bug in ."}
{"_id":"q-en-django__django-16322-e6b07fe5725df559e81a366cda158a0fa3f5a4ada0cc81498b495b7ae9e787dc","text":"Django subtly produces incorrect query when the same keyword appears in both aggregate() and annotate() Description I have the following query: Model.objects.annotate(foo=F('column')).aggregate(foo=Sum(F('foo'))) Initially, I was running this query on SQLite and Django was producing a result (i.e. 0). When I switched to MySQL this query crushed with the following exception: Traceback (most recent call last): File \"\", line 86, in execute return (sql, params) File \"\", line 74, in execute return (query, args) File \"\", line 209, in execute res = (query) File \"\", line 315, in query (q) File \"\", line 239, in query (self, query) MySQLdb.exceptions.OperationalError: (1054, \"Unknown column 'foo' in 'field list'\") The above exception was the direct cause of the following exception: Traceback (most recent call last): File \"\", line 15, in <moduleret2 = ('mysql').annotate(foo=F('column')).aggregate(foo=Sum(F('foo'))) File \"\", line 384, in aggregate return query.getaggregation(, kwargs) File \"\", line 502, in getaggregation result = compiler.executesql(SINGLE) File \"\", line 1151, in executesql cursor.execute(sql, params) File \"\", line 100, in execute return super().execute(sql, params) File \"\", line 68, in execute return self.executewithwrappers(sql, params, many=False, executor=self.execute) File \"\", line 77, in executewithwrappers return executor(sql, params, many, context) File \"\", line 86, in execute return (sql, params) File \"\", line 90, in exit raise djexcvalue.withtraceback(traceback) from excvalue File \"\", line 86, in execute return (sql, params) File \"\", line 74, in execute return (query, args) File \"\", line 209, in execute res = (query) File \"\", line 315, in query (q) File \"\", line 239, in query (self, query) (1054, \"Unknown column 'foo' in 'field list'\") After examining the SQL query produced by Django, I realized that the query is not correct. Specifically, Django produced the following SQL query: SELECT SUM() AS FROM Instead of SELECT SUM() FROM (SELECT . AS FROM ) subquery The issue appears when a keyword in aggregate function is the same with one of the keywords in annotate function. I initially thought that Django does not have any restriction on the keywords used in aggregate (i.e. a keyword in aggregate can be the same with the name of a column from the inspected model). If you think that aggregate should not conflict with the annotate, then Django should produce a warning as running this query in SQLite subtly produces an incorrect result."}
{"_id":"q-en-django__django-16333-5f1089ccc3f40fe0ea492ea6796528713e1ae9dbf5015b74372bd8d805c4ee86","text":"UserCreationForm should save data from ManyToMany form fields Description When using with a custom User model which has ManyToManyField fields, the data in all related form fields (e.g. a ModelMultipleChoiceField) is not saved. This is because unlike its parent class , (commit=True) omits to call self.save_m2m(). This has been discussed on the #django-developers mailing list ​ and I'm ready to work on a PR."}
{"_id":"q-en-django__django-16343-e9e98bd68e030c509a5ba202c9c46139e244ac651ed7e8886918b343184d52e9","text":"Deprecate passing positional arguments to Signer. Description We discovered a vulnerability in one of our applications recently which was caused by an inaccurate instantiation of The developer intended to use the user's email address as the salt for the Signing instance but instead caused it to be used as the key. Here's an example code block that demonstrates the problem: signer = Signer(self.context['request']) signeddata = signer.signobject(dict( licensenumber='...', productid='...', device_count='...' )) In our case, this signed data was then being used to verify a later request and generate an active license. This meant that an attacker could feasibly generate their own licenses if they realised that their email address was the key. The fix for this was to add salt= in front of the email variable. It occurred to us that this is a relatively easy mistake to make and could be avoided if the signature of was changed thusly: def init(self, key=None, sep=':', salt=None, algorithm=None): def init(self, , key=None, sep=':', salt=None, algorithm=None): That is, adding a after self to force the developer to name the parameters."}
{"_id":"q-en-django__django-16366-911ac1573e08cfa6e5d8e1fa9b3104b69ac17edfbb5122e87feb5dc6b72b44c4","text":"Drop support for SQLite < 3.21.0 Description SQLite 3.15+ supports functions in partial indexes. SQLite 3.20+ can defer constraint checks and supports PRAGMA foreign key checks. Ubuntu Xenial ships with SQLite 3.22.0 (which will still by supported by Django) and will EOL in April 2023. Debian Buster ships with 3.27.2 and will EOL in June 2024. Python 3.7 ships with 3.21.0. ​SQLite 3.21.0 was released in October 2017. SQLite version support seems like a similar situation as GEOS libraries which we generally support about 5 years after released."}
{"_id":"q-en-django__django-16369-493d307ad346de5bb4f0482d863589b5630fdb18caa282dd2d604169dfedfd3f","text":"Choose which items are displayed per language in Sitemap Description The current implementation of Sitemap is : if we use i18n, then we display a cartesian product between some items and some languages. There is no way to use the provided i18n automation if we want to display some items depending on the language (for instance non-translated blog articles). I precise in my case, urls are translated, so given a language the url may not exist or raise an error."}
{"_id":"q-en-django__django-16379-8b851657db8ddc2c09933a540f87a3417caf11b02127964dcb5f604ae60ef633","text":"FileBasedCache haskey is susceptible to race conditions Description (last modified by Marti Raudsepp) I received the exception from Django's cache framework: FileNotFoundError: [Errno 2] No such file or directory: '\/app\/var\/cache\/.djcache' [...] File \"\", line 229, in getorset (key, default, timeout=timeout, version=version) File \"\", line 26, in add if (key, version): File \"\", line 94, in haskey with open(fname, \"rb\") as f: The code is: def haskey(self, key, version=None): fname = self.keytofile(key, version) if (fname): with open(fname, \"rb\") as f: return not self.isexpired(f) return False Between the exists() check and open(), it's possible for the file to be deleted. In fact, the is_expired() method itself deletes the file if it finds it to be expired. So if many threads race to read an expired cache at once, it's not that unlikely to hit this window."}
{"_id":"q-en-django__django-16398-eb892d90c10b81457d6c494644f51917b0caa0e42816b6c1355c0690cc936470","text":"QuerySet.selectrelated() with multiple filtered relations to the OneToOneField sets the last one. Description ​ ​ should use partial, just like the remotesetter"}
{"_id":"q-en-django__django-16400-8a2a11f4986f92780d85c3f3e65c6a55f2c4d82d63a7eb90f0757078461655d4","text":"migrate management command does not respect database parameter when adding Permissions. Description (last modified by Vasanth) When invoking migrate with a database parameter, the migration runs successfully. However, there seems to be a DB read request that runs after the migration. This call does not respect the db param and invokes the db router . When naming the db as a parameter, all DB calls in the context of the migrate command are expected to use the database specified. I came across this as I am currently using a thread-local variable to get the active DB with a custom DB router for a multi-tenant service . Minimal example Setup the custom middleware and custom DB Router as show below. Then run any DB migration. We see that \"read {}\" is being printed before the exception message. Ideally none of this code must be called as the DB was specified during management command. from threading import local from import settings localstate = local() class InvalidTenantException(Exception): pass class TenantSubdomainMiddleware: def init(self, getresponse): self.getresponse = getresponse def call(self, request): ## Get Subdomain host = request.gethost().split(\":\")[0] localstate.subdomain = ( # We assume single level of subdomain : # HOSTIP : used to for local dev. host if host in settings.HOSTIP else (\".\")[0] ) response = self.getresponse(request) return response class TenantDatabaseRouter: def defaultdb(self): subdomain = getattr(localstate, \"subdomain\", None) if subdomain is not None and subdomain in settings.TENANTMAP: dbname = settings.TENANTMAP[localstate.subdomain] return dbname else: raise InvalidTenantException() def dbforread(self, model, hints): print(\"read\", hints) return self.defaultdb() def dbforwrite(self, model, hints): print(\"write\", hints) return self.defaultdb() def allowrelation(self, obj1, obj2, hints): return None def allowmigrate(self, db, applabel, modelname=None, hints): return None MIDDLEWARE = [ \"utils.tenantdbrouter.TenantSubdomainMiddleware\", \"django.middleware.security.SecurityMiddleware\", ... ] TENANTMAP = {\"localhost\":\"default\", \"tenant1\":\"default\"} DATABASEROUTERS = [\"utils.tenantdbrouter.TenantDatabaseRouter\"]"}
{"_id":"q-en-django__django-16408-15399e329d4f48ceeb4d1d17198bc6c115041ec708e96861f15df452d1c1b958","text":"Multi-level FilteredRelation with selectrelated() may set wrong related object. Description test case: def testwrongselectrelated(self): with self.assertNumQueries(3): p = list(PoolStyle.objects.annotate( tournamentpool=FilteredRelation('pooltournamentpool'), ).selectrelated('tournamentpool')) self.assertEqual(p[0].pool.tournament, p[0].tournamentpool.tournament) result: FAIL: testwrongselectrelated () Traceback (most recent call last): File \"D:\\Work\\django\\tests\\knownrelatedobjects\\\", line 171, in testwrongselectrelated self.assertEqual(p[0].pool.tournament, p[0].tournamentpool.tournament) AssertionError: <Tournament: Tournament object (1)!= <PoolStyle: PoolStyle object (1)"}
{"_id":"q-en-django__django-16411-11ee01e1781a352e4751be35ade7f53c7a63c9831d6c99813c904c885b01260c","text":"ManifestStaticFilesStorage should expose a \"hash\" of the manifest file. Description It would be great if ManifestFilesMixin could expose a manifest_hash that changes whenever anything in the manifest itself changes. This would allow SPAs (or applications enhanced with htmx\/unpoly etc) to send along a header in responses ala X-Asset-Hash and allow them to perform a full reload if the header changes. I'll be opening a PR for discussion soon."}
{"_id":"q-en-django__django-16429-1c1b9a94478e935b3e2b7b29a9db5cb2a3834f7348267707a239c9c28d1b02d8","text":"timesince() raises TypeError with USETZ=True and 1 month interval. Description (last modified by Sage Abdullah) As of , calling timesince() with a datetime object that's one month (or more) in the past and the USETZ setting is set to True results in the following crash: TypeError: can't subtract offset-naive and offset-aware datetimes Test: ... class TimesinceTests(TestCase): ... def testlongintervalwithtz(self): now = () d = now - datetime.timedelta(days=31) self.assertEqual(timesince(d), \"1\\xa0month\") I believe this is because the pivot instantiated here: ​ does not take into account the datetime object's tzinfo. Adding 0, d.tzinfo arguments to the datetime.datetime call seems to fix this. Happy to send a PR."}
{"_id":"q-en-django__django-16454-93f4e096cd78056b598f2c6851611b8afe9932234f12546eba3fe609f3ce3815","text":"Management command subparsers don’t retain error formatting Description Django management commands use a subclass of argparse.ArgumentParser, CommandParser, that takes some extra arguments to improve error formatting. These arguments are not copied into subparsers, created via CommandParser.addsubparsers().addparser(). Missing arguments to subparsers thus end as stack traces on the CLI, rather than human-facing usage messages. For example take this command with a subparser: from import BaseCommand class Command(BaseCommand): def addarguments(self, parser): subparsers = parser.addsubparsers(required=True) create = subparsers.addparser(\"create\") create.addargument(\"name\") def handle(self, args, *options): pass Missing the required subparser name argument gives the usage message, as for any normal argument: $ cheeses usage: cheeses [-h] [--version] [-v {0,1,2,3}] [--settings SETTINGS] [--pythonpath PYTHONPATH] [--traceback] [--no-color] [--force-color] [--skip-checks] {create} ... cheeses: error: the following arguments are required: {create} But missing the name argument to create fails with a stacktrace: $ cheeses create Traceback (most recent call last): File \"\", line 21, in <modulemain() ... File \"\", line 2131, in parseknownargs (('the following arguments are required: %s') % File \"\", line 72, in error raise CommandError(\"Error: %s\" % message) Error: the following arguments are required: name We can correct this by ensuring that the subparser action returned by add_subparsers() copies the relevant arguments through to constructed subparsers. (Originally reported by Mark Gregson on django-developers: ​ )"}
{"_id":"q-en-django__django-16485-d190a96cf6356a4885db93e24f40dd51871d73473b5bd400e8d47292d136cc01","text":"floatformat() crashes on \"0.00\". Description from decimal import Decimal from django.template.defaultfilters import floatformat floatformat('0.00', 0) floatformat(Decimal('0.00'), 0) Both throw ValueError: valid range for prec is [1, MAX_PREC]"}
{"_id":"q-en-django__django-16491-37e1268b57dcb18ef90bc6acd5775005e1e45337ca2ca1a25dbbdb74f2cfed0c","text":"Exists annotations can return non-boolean results (i.e. None) if used with an empty QuerySet. Description I suspect this is following on from, but potentially separate to -- because that ticket starts out using Django 3.2 to observe that an empty queryset (EmptyQuerySet or whatever, via none()) can short circuit evaluation to be 0 as ... in the SQL query, which is exactly the same problem I observed. However, as far as I can tell, the result of an Exists(()) can still return values outside of True\/False, namely, None. Using Django main as of (from Wed Jan 11 ... 2023), here's the behaviour on both postgres and sqlite. In both scenarios I'm using 3.10.2 with psycopg2==2.9.3 and sqlite3.sqliteversion is 3.37.0 and sqlite3.version is 2.6.0. IPython outputs 8 and 11 are the problems. class A(models.Model): pass class B(models.Model): pass In [1]: from import A, B In [2]: (\"pg\").create() Out[2]: <A: A object (1)In [3]: (\"pg\").create() Out[3]: <B: B object (1)In [4]: (\"sqlite\").create() Out[4]: <A: A object (1)In [4]: (\"sqlite\").create() Out[4]: <B: B object (1)In [5]: from import Exists In [6]: (\"sqlite\").annotate(shouldbebool=Exists(())).first().shouldbebool Out[6]: True In [7]: (\"sqlite\").annotate(shouldbebool=Exists((pk=))).first().shouldbebool Out[7]: False In [8]: (\"sqlite\").annotate(shouldbebool=Exists(())).first().shouldbebool In [9]: (\"pg\").annotate(shouldbebool=Exists(())).first().shouldbebool Out[9]: True In [10]: (\"pg\").annotate(shouldbebool=Exists((pk=))).first().shouldbebool Out[10]: False In [11]: (\"pg\").annotate(shouldbebool=Exists(())).first().shouldbebool And the queries, which are the same for postgres & sqlite: {'sql': 'SELECT \"appa\".\"id\", EXISTS(SELECT 1 AS \"a\" FROM \"appb\" LIMIT 1) AS \"shouldbebool\" FROM \"appa\" ORDER BY \"appa\".\"id\" ASC LIMIT 1', 'time': '0.001'}, {'sql': 'SELECT \"appa\".\"id\", EXISTS(SELECT 1 AS \"a\" FROM \"appb\" U0 WHERE U0.\"id\" = LIMIT 1) AS \"shouldbebool\" FROM \"appa\" ORDER BY \"appa\".\"id\" ASC LIMIT 1', 'time': '0.001'}, {'sql': 'SELECT \"appa\".\"id\", NULL AS \"shouldbebool\" FROM \"appa\" ORDER BY \"appa\".\"id\" ASC LIMIT 1', 'time': '0.001'} Given Exists has an outputfield of BooleanField and that definition doesn't have null=True as an argument, it seems incongruent from both an expectations (\"exists sounds boolean\") and implementation (\"it doesn't say it could be null\") standpoint. Whilst the problem exists in main, it has also changed behaviour (presumably or potentially unexpectedly) since 3.2, where postgres and sqlite actually do different things, hence we tested both above. So main is now consistent, but I'd personally argue it's consistently wrong (for a given value thereof, no judgement made!) In 3.2.16, under sqlite, using annotate(x=Exists(y.none())) returns False but on main it now returns None (see above) -- the 3.2 behaviour is correct for my expectations In [4]: (\"sqlite\").annotate(shouldbebool=Exists(())).first().shouldbebool Out[4]: False In [5]: connections['sqlite'].queries Out[5]: {'sql': 'SELECT \"appa\".\"id\", 0 AS \"shouldbebool\" FROM \"appa\" ORDER BY \"appa\".\"id\" ASC LIMIT 1', 'time': '0.000'} In 3.2.16 with postgres we get neither None nor False but the integer 0 instead: In [4]: (\"pg\").annotate(shouldbebool=Exists(())).first().shouldbebool Out[4]: 0 In [5]: connections['pg'].queries Out[5]: {'sql': 'SELECT \"appa\".\"id\", 0 AS \"shouldbebool\" FROM \"appa\" ORDER BY \"app_a\".\"id\" ASC LIMIT 1', 'time': '0.001'} So we can observe that under 3.2 using the 0 AS ... behaviour sqlite appears to behave correctly (returning False) postgres appears to behave incorrectly (failing to cast a raw integer to a boolean) And under main using the NULL AS ... behaviour sqlite no longer behaves the same, returning None where I'd expect False (or even True given the way SQL EXISTS works...) postgres behaves differently, 0 is now None but I'd still expect True or False as the output."}
{"_id":"q-en-django__django-16493-9af8349616814a45a3bb88ffafb5987ccd9285268db3e4db89c9e0be4f638ff0","text":"Callable storage on FileField fails to deconstruct when it returns defaultstorage Description If the storage argument on a FileField is set to a callable that returns defaultstorage, it is omitted from the deconstructed form of the field, rather than being included as a reference to the callable as expected. For example, given a model definition: from import FileSystemStorage, defaultstorage from import models import random otherstorage = FileSystemStorage(location='\/media\/other') def getstorage(): return ([defaultstorage, otherstorage]) class MyModel(models.Model): myfile = models.FileField(storage=getstorage) repeatedly running makemigrations will randomly generate a migration that alternately includes or omits on the FileField definition. This case was overlooked in the fix for - the deconstruct method tests if self.storage is not defaultstorage to determine whether to add the storage kwarg, but at this point self.storage is the evaluated version, so it wrongly returns false for a callable that returns defaultstorage."}
{"_id":"q-en-django__django-16501-fa0ae3221a766f4fbe37fda8ddcb5cb13fc5129a019ae3f344f7284231d42606","text":"Validation of UniqueConstraint with ordered expressions crashes. Description I this model: class Foo(models.Model): name = models.CharField(maxlength=300, unique=True) class Meta: constraints = [ UniqueConstraint( Lower(\"name\").desc(), name=\"uniquelowername\" ) ] and this admin class: class FooAdmin(admin.ModelAdmin): searchfields = (\"name\",) I based the UniqueConstraint on the documentation (​), which shows UniqueConstraint(Lower('name').desc(), 'category', name='uniquelowernamecategory') When I visit the admin and click \"add foo\", enter name \"foo1\", and click Save, I get a stack trace: Request Method: POST Request URL: http:\/\/localhost:8000\/admin\/myapp\/foo\/add\/ Django Version: 4.1.5 Python Version: 3.10.9 ... Traceback (most recent call last): File \"...\/lib\/python3.10\/site-\", line 89, in execute return (sql, params) The above exception (syntax error at or near \"DESC\" LINE 1: ...werappfoo\" WHERE LOWER(\"myappfoo\".\"name\") DESC = (LO... ^ ) was the direct cause of the following exception: File \"...\/lib\/python3.10\/site-\", line 55, in inner response = getresponse(request) File \"...\/lib\/python3.10\/site-\", line 197, in getresponse response = wrappedcallback(request, callbackargs, callbackkwargs) File \"...\/lib\/python3.10\/site-\", line 686, in wrapper return self.adminsite.adminview(view)(args, kwargs) File \"...\/lib\/python3.10\/site-\", line 133, in wrappedview response = viewfunc(request, args, kwargs) File \"...\/lib\/python3.10\/site-\", line 62, in wrappedviewfunc response = viewfunc(request, args, kwargs) File \"...\/lib\/python3.10\/site-\", line 242, in inner return view(request, args, kwargs) File \"...\/lib\/python3.10\/site-\", line 1890, in addview return self.changeformview(request, None, formurl, extracontext) File \"...\/lib\/python3.10\/site-\", line 46, in wrapper return boundmethod(args, kwargs) File \"...\/lib\/python3.10\/site-\", line 133, in wrappedview response = viewfunc(request, args, *kwargs) File \"...\/lib\/python3.10\/site-\", line 1750, in changeformview return self.changeformview(request, objectid, formurl, extracontext) File \"...\/lib\/python3.10\/site-\", line 1796, in changeformview formvalidated = form.isvalid() File \"...\/lib\/python3.10\/site-\", line 205, in isvalid return self.isbound and not File \"...\/lib\/python3.10\/site-\", line 200, in errors self.fullclean() File \"...\/lib\/python3.10\/site-\", line 439, in fullclean self.postclean() File \"...\/lib\/python3.10\/site-\", line 492, in postclean self.instance.fullclean(exclude=exclude, validateunique=False) File \"...\/lib\/python3.10\/site-\", line 1472, in fullclean self.validateconstraints(exclude=exclude) File \"...\/lib\/python3.10\/site-\", line 1423, in validateconstraints constraint.validate(modelclass, self, exclude=exclude, using=using) File \"...\/lib\/python3.10\/site-\", line 347, in validate if (): File \"...\/lib\/python3.10\/site-\", line 1226, in exists return () File \"...\/lib\/python3.10\/site-\", line 592, in hasresults return compiler.hasresults() File \"...\/lib\/python3.10\/site-\", line 1366, in hasresults return bool(self.executesql(SINGLE)) File \"...\/lib\/python3.10\/site-\", line 1398, in executesql cursor.execute(sql, params) File \"...\/lib\/python3.10\/site-\", line 230, in execute return (, sql, params) File \"...\/lib\/python3.10\/site-\", line 154, in record return method(sql, params) File \"...\/lib\/python3.10\/site-\", line 103, in execute return super().execute(sql, params) File \"...\/lib\/python3.10\/site-\", line 67, in execute return self.executewithwrappers( File \"...\/lib\/python3.10\/site-\", line 80, in executewithwrappers return executor(sql, params, many, context) File \"...\/lib\/python3.10\/site-\", line 84, in execute with File \"...\/lib\/python3.10\/site-\", line 91, in exit raise djexcvalue.withtraceback(traceback) from excvalue File \"...\/lib\/python3.10\/site-\", line 89, in execute return (sql, params) Exception Type: ProgrammingError at \/admin\/myapp\/foo\/add\/ Exception Value: syntax error at or near \"DESC\" LINE 1: ...myappfoo\" WHERE LOWER(\"myappfoo\".\"name\") DESC = (LO... ^ This happens on the Postgres backend (psycopg2 2.9.5). I also get this error using the sqlite backend. Exception Type: OperationalError at \/admin\/myapp\/foo\/add\/ Exception Value: near \"DESC\": syntax error"}
{"_id":"q-en-django__django-16502-e2464401b7bcad7ed6bd8ab6126fd73ce0cfdf384eaf8372e4ad37540151e6e2","text":"After runserver returns response body for HTTP HEAD requests Description For compliance with RFC 2616, section 4.3, response bodies must not be returned for HEAD requests. In , the stripping of the response bodies was removed from Django in favour of letting the server perform the body removal, since the common servers (gunicorn, mod_wsgi etc) already do so. However it appears that runserver does not strip the body, contrary to: As such, starting in Django 1.10 the responses from runserver for HEAD requests are no longer compliant with the spec. (In certain configurations this also results in \"Broken pipe\" error messages in runserver output, since compliant user agents expect to be able to terminate the connection after the headers are sent.) STR: 1) mkvirtualenv django-test 2) pip install 'Django1.10,<1.11' 3) django-admin startproject django-test 4) cd django-test 5) runserver 6) In another terminal, run curl -iX HEAD http:\/\/127.0.0.1:8000\/ 7) Observe response from curl Expected: HTTP\/1.0 200 OK Date: Fri, 07 Apr 2017 14:56:39 GMT Server: WSGIServer\/0.2 CPython\/3.4.5 Content-Type: text\/html X-Frame-Options: SAMEORIGIN Actual: HTTP\/1.0 200 OK Date: Fri, 07 Apr 2017 14:56:39 GMT Server: WSGIServer\/0.2 CPython\/3.4.5 Content-Type: text\/html X-Frame-Options: SAMEORIGIN <!DOCTYPE html<html lang=\"en\"<head<meta http-equiv=\"content-type\" content=\"text\/html; charset=utf-8\"<meta name=\"robots\" content=\"NONE,NOARCHIVE\"<titleWelcome to Django<\/title... Tested with Python 2.7.13 and 3.4.5. Doesn't reproduce under Django 1.9.13."}
{"_id":"q-en-django__django-16511-e861e86d83d94bbc6ddf2027aba68fbede82bb4eec8e613791207772bba6c52c","text":"Support create defaults for updateorcreate Description I proposed the idea of extending updateorcreate to support specifying a different set of defaults for the create operation on the . There seems to be consensus it's a positive add to Django. Adam raised concerns with my proposed approach of adding a createdefaults parameter to the function since this would conflict with any fields on a model named, createdefaults. Jeff did a code search on github for that term and didn't find any matches. I suspect if someone where using a field named createdefaults, it would be a JSON or object type field. Those don't seem like reasonable candidates to be part of a UniqueConstraint, which should be underlying the look-up arguments to updateorcreate. I do like the idea of having a separate parameter for createdefaults, but if we must preserve 100% backwards compatibility, Adam's suggestion of having defaults be set to another object makes the most sense. My blocking question is, which approach should I take? From the forum post: I’ve run into a use-case in which it’d be helpful to have the ability to specify a different set of defaults for the update operation compared to the create operation. While I don’t expect my particular use case to translate, here’s a more generic one. Given the following Record model: class Record(models.Model): someid = models.CharField(unique=True) createdby = models.ForeignKey(User, ...) modifiedby = models.ForeignKey(User, null=True, blank=True, ...) When a record is created, we would want to set createdby, but if it’s being updated, we’d want to set modifiedby. This use case can’t be solved by using updateorcreate, unless it allows for us to specify a different set of default values. Record.objects.updateorcreate( someid=somevalue, defaults={\"modifiedby\": user}, createdefaults={\"createdby\": user}, )"}
{"_id":"q-en-django__django-16514-cea502749dd8fc9f4296df3751d73f73ec175dac4804e4a2268d09502bc2e64c","text":"Admin Site shall filter LogEntries by registered models Description This ticket has been discussed here: ​ If an implementer registers more than one AdminSite, one might expect that only changes on models registered at that specific site are logged. This currently is not the case, all registered sites show all entries of LogEntry. It is confusing for users to access a specific site and see log entries for models they can't access."}
{"_id":"q-en-django__django-16517-99ec163c7dfd7dd1e5045a3aeea13899397f74f718e4274cbf064267d9ff5d77","text":"Mixed-case views\/templates names causes 404 on :view:\/:template: directive. Description ​ Using a class based view, class OrderSearch(LoginRequiredMixin, UserPassesTestMixin, ListView): add a doc comment such as causes a 404 when you click on the link in the docs Page not found (404) Request Method: GET Request URL: ​ Raised by: I'm not sure exactly where orders becomes lowercase, but I thought it might have something to do with the getview_func"}
{"_id":"q-en-django__django-16527-0ca790738b5e1ff387e0457bece317e4d7b963209cc0ba3ae3e0d60beae31fb1","text":"\"showsaveasnew\" in admin can add without this permission Description (last modified by Mariusz Felisiak) At \"\" file, line 102, I think you must put one more verification for this tag: \"and hasaddpermission\", because \"saveasnew\" is a add modification. I rewrite this for my project: \"showsaveasnew\": not ispopup and hasaddpermission # This line that I put!!! and haschangepermission and change and save_as,"}
{"_id":"q-en-django__django-16532-c56f3b24c944c683b1fea7b53cd0ba451a21ac61390fc9565a5edc5bf810924c","text":"Duplicate model names in M2M relationship causes RenameModel migration failure Description Example code is here: ​ I have a django project with two apps, incidents and vault, that both have a model named Incident. The vault Incident model has an M2M involving the incidents Incident model. When the table is created for this M2M relationship the automatic field names are \"fromincidentid\" and \"toincidentid\" since models have the same names. If I then try to use a RenameModel in a migration... operations = [ migrations.RenameModel( oldname='Incident', newname='Folder', ), ] it fails with this traceback: Tracking file by folder pattern: migrations Operations to perform: Apply all migrations: admin, auth, contenttypes, incidents, sessions, vault Running migrations: Applying vault.0002renameincidentfolder...Traceback (most recent call last): File \"\/Users\/zacmiller\/PycharmProjects\/virtualenvs\/edemo\/lib\/python3.10\/site-\", line 668, in getfield return self.fieldsmap[fieldname] KeyError: 'incident' During handling of the above exception, another exception occurred: Traceback (most recent call last): File \"\/Users\/zacmiller\/Library\/Application Support\/JetBrains\/Toolbox\/apps\/PyCharm-P\/ch-\", line 52, in <moduleruncommand() File \"\/Users\/zacmiller\/Library\/Application Support\/JetBrains\/Toolbox\/apps\/PyCharm-P\/ch-\", line 46, in runcommand runmodule(managefile, None, 'main', True) File \"\", line 209, in runmodule return runmodulecode(code, initglobals, runname, modspec) File \"\", line 96, in runmodulecode runcode(code, modglobals, initglobals, File \"\", line 86, in runcode exec(code, runglobals) File \"\", line 22, in <modulemain() File \"\", line 18, in main executefromcommandline() File \"\/Users\/zacmiller\/PycharmProjects\/virtualenvs\/edemo\/lib\/python3.10\/site-\", line 446, in executefromcommandline utility.execute() File \"\/Users\/zacmiller\/PycharmProjects\/virtualenvs\/edemo\/lib\/python3.10\/site-\", line 440, in execute self.fetchcommand(subcommand).runfromargv() File \"\/Users\/zacmiller\/PycharmProjects\/virtualenvs\/edemo\/lib\/python3.10\/site-\", line 402, in runfromargv self.execute(args, cmdoptions) File \"\/Users\/zacmiller\/PycharmProjects\/virtualenvs\/edemo\/lib\/python3.10\/site-\", line 448, in execute output = (args, options) File \"\/Users\/zacmiller\/PycharmProjects\/virtualenvs\/edemo\/lib\/python3.10\/site-\", line 96, in wrapped res = handlefunc(args, *kwargs) File \"\/Users\/zacmiller\/PycharmProjects\/virtualenvs\/edemo\/lib\/python3.10\/site-\", line 349, in handle postmigratestate = executor.migrate( File \"\/Users\/zacmiller\/PycharmProjects\/virtualenvs\/edemo\/lib\/python3.10\/site-\", line 135, in migrate state = self.migrateallforwards( File \"\/Users\/zacmiller\/PycharmProjects\/virtualenvs\/edemo\/lib\/python3.10\/site-\", line 167, in migrateallforwards state = self.applymigration( File \"\/Users\/zacmiller\/PycharmProjects\/virtualenvs\/edemo\/lib\/python3.10\/site-\", line 252, in applymigration state = (state, schemaeditor) File \"\/Users\/zacmiller\/PycharmProjects\/virtualenvs\/edemo\/lib\/python3.10\/site-\", line 130, in apply operation.databaseforwards( File \"\/Users\/zacmiller\/PycharmProjects\/virtualenvs\/edemo\/lib\/python3.10\/site-\", line 422, in databaseforwards (), File \"\/Users\/zacmiller\/PycharmProjects\/virtualenvs\/edemo\/lib\/python3.10\/site-\", line 670, in getfield raise FieldDoesNotExist( Incidentincidents has no field named 'incident'"}
{"_id":"q-en-django__django-16569-12b6db04d1fec8d8972d3bc2cf6a655732e7d2d5a7490e7a1f2d5e6e30eb7923","text":"Formsets' addfields() method fails in some circumstances if the argument index is None. Description Formsets' addfields() method fails in some circumstances if the argument index is None. When a FormSet has the attributes self.candelete == True and self.candeleteextra == False, calling the addfields() method on that FormSet fails if the argument index is None. This occurs for example when calling FormSet.emptyform(). The result is that the method raises the exception TypeError: '<' not supported between instances of 'NoneType' and 'int'. Code example: MyFormSet = forms.formsetfactory( form=MyForm, candelete=True, candeleteextra=False, ) myformset = MyFormSet( initial=None, ) print(myformset.emptyform) The reason this happens is that in in line 493 of index is compared to initialformcount: if self.candelete and (self.candeleteextra or index < initialformcount): Checking for index not None should fix the issue: if self.candelete and (self.candeleteextra or (index is not None and index < initialformcount)): How to Reproduce A self-contained example to reproduce this bug is as follows: #!\/usr\/bin\/env python3 import os import django from django import forms class MyForm(forms.Form): myfield = forms.CharField() if name == \"main\": settingsfile = ((file))[0] ( DEBUG=True, MIDDLEWARECLASSES=[], ROOTURLCONF=settingsfile, ) () MyFormSet = forms.formsetfactory( form=MyForm, candelete=True, candeleteextra=False, ) myformset = MyFormSet( initial=None, ) print(myformset.emptyform)"}
{"_id":"q-en-django__django-16578-dd24a351aac95b9ad585819e6a9a41e5bb2fcfe898ab8677fa7b9244083f92e4","text":"Make BaseConstraint arguments keyword-only Description As suggested in ​ I think we should change the signature of BaseConstraint to use keyword-only arguments as a separate cleanup name and violationerrormessage are already keyword-only in all the BaseConstraint subclasses in Django code base."}
{"_id":"q-en-django__django-16588-b0cb13dde5d6077a2dfd25358c60d0f6642700f66a0d878b6ed064b578e87541","text":"floatformat() crashes on \"0.0000\" Description Similar to the current Django code (both 4.2b1 and latest git) crash on using the floatformat template filter with 0 values. from decimal import Decimal from django.template.defaultfilters import floatformat floatformat('0.0000', 2) floatformat(Decimal('0.0000'), 2) Both throw ValueError: valid range for prec is [1, MAXPREC] Using git bisect I tracked the bug to commit . When one uses \"0.0000\": floatformat:2 the current code results in a precision of 0 which is not allowed: tupl = d.astuple() # with d being \"0.0000\" returns (sign=0,digits=(0,),exponent=-4) units = len(tupl[1]) # equals 1 units += -tupl[2] if m else tupl[2] # 1 + (-4) prec = abs(p) + units + 1 # 2 (number of requested decimals) + (-3) + 1 equals 0 roundedd = d.quantize(exp, ROUNDHALF_UP, Context(prec=prec))"}
{"_id":"q-en-django__django-16595-d27bbed86eff60c8c26c0dfcc0a2e89db9170bfe8a8703e3273990a7a9e65966","text":"Migration optimizer does not reduce multiple AlterField Description Let's consider the following operations: operations = [ migrations.AddField( modelname=\"book\", name=\"title\", field=models.CharField(maxlength=256, null=True), ), migrations.AlterField( modelname=\"book\", name=\"title\", field=models.CharField(maxlength=128, null=True), ), migrations.AlterField( modelname=\"book\", name=\"title\", field=models.CharField(maxlength=128, null=True, helptext=\"help\"), ), migrations.AlterField( modelname=\"book\", name=\"title\", field=models.CharField(maxlength=128, null=True, helptext=\"help\", default=None), ), ] If I run the optimizer, I get only the AddField, as we could expect. However, if the AddField model is separated from the AlterField (e.g. because of a non-elidable migration, or inside a non-squashed migration), none of the AlterField are reduced: optimizer.optimize(operations[1:], \"books\") [<AlterField modelname='book', name='title', field=, <AlterField modelname='book', name='title', field=, <AlterField modelname='book', name='title', field=] Indeed, the does not consider the the case where operation is also an AlterField. Is this behaviour intended? If so, could it be documented? Otherwise, would it make sense to add something like if isinstance(operation, AlterField) and self.issamefieldoperation( operation ): return [operation]"}
{"_id":"q-en-django__django-16597-63da2bff315aa507e9881e72b2a555dc60887c4fdbed3a4e2e82d7f27eec4d21","text":"Subquery referencing WINDOW expression breaks query compilation Description introduced the ability to filter on WINDOW expressions but when a WINDOW expression is referred to in a subquery, the query compiler fails: ValueError: Need either using or connection I assume it has to do with self.connection not being passed on to innerquery.getcompiler in getqualifysql introduced here: ​ Below is the full traceback. A dummy test is available here: ​ Traceback (most recent call last): File \"\", line 57, in testPartExecutor yield File \"\", line 623, in run self.callTestMethod(testMethod) File \"\", line 579, in callTestMethod if method() is not None: File \"\", line 1025, in testfiltersubquery list( File \"\", line 398, in iter self.fetchall() File \"\", line 1881, in fetchall self.resultcache = list(self.iterableclass(self)) File \"\", line 91, in iter results = compiler.executesql( File \"\", line 1545, in executesql sql, params = () File \"\", line 732, in assql extraselect, orderby, groupby = self.presqlsetup( File \"\", line 84, in presqlsetup self.setupquery(withcolaliases=withcolaliases) File \"\", line 73, in setupquery , self.klassinfo, self.annotationcolmap = self.getselect( File \"\", line 296, in getselect sql, params = self.compile(col) File \"\", line 542, in compile sql, params = (self, self.connection) File \"\", line 1536, in assql subquerysql, sqlparams = (compiler, connection) File \"\", line 1150, in assql sql, params = self.getcompiler(connection=connection).assql() File \"\", line 751, in assql result, params = self.getqualifysql() File \"\", line 676, in getqualifysql innerquerycompiler = innerquery.getcompiler( File \"\", line 298, in get_compiler raise ValueError(\"Need either using or connection\") ValueError: Need either using or connection"}
{"_id":"q-en-django__django-16599-b037708280aa3755ca81630bc6d6b489e26b74ece56c7cde225dd568a5c80db9","text":"Querying with an integer larger than SQLite supports crashes with OverflowError Description Hi, i don't know, you accept this behavior Overflow on SQLite or there is max length here? Step : ​http:\/\/127.0.0.1:8000\/admin\/auth\/user\/\/change\/ Error : OverflowError at \/admin\/auth\/user\/\/change\/ Python int too large to convert to SQLite INTEGER Request Method: GET Request URL: http:\/\/127.0.0.1:8000\/admin\/auth\/user\/\/change\/ Django Version: 1.10.2 Exception Type: OverflowError Exception Value: Python int too large to convert to SQLite INTEGER Exception Location: C:\\Python27\\lib\\site-packages\\django\\db\\backends\\sqlite3\\ in quoteparamsforlastexecutedquery, line 129 Python Executable: C:\\Python27\\ Python Version: 2.7.11 Python Path: ['E:\/Programmer Language\/NodeJS\/untitled1', 'C:\\Users\\RaminFP\\.IntelliJIdea14\\config\\plugins\\python\\helpers\\pydev', 'C:\\Python27\\lib\\site-packages\\six-1.10.0-', 'C:\\Python27\\lib\\site-packages\\pisa-3.0.33-', 'C:\\Python27\\lib\\site-packages\\sqlacodegen-1.1.6-', 'C:\\Python27\\lib\\site-packages\\inflect-0.2.5-', 'C:\\Python27\\lib\\site-packages\\0x10c_asm-0.0.2-', 'C:\\Python27\\lib\\site-packages\\jpype1-0.6.1-py2.7-', 'C:\\Python27\\lib\\site-packages\\simpleaes-1.0-', 'C:\\Python27\\lib\\site-packages\\celery-3.1.23-', 'C:\\Python27\\lib\\site-packages\\kombu-3.0.35-', 'C:\\Python27\\lib\\site-packages\\anyjson-0.3.3-', 'C:\\Python27\\lib\\site-packages\\pythonnet-2.1.0.dev1-py2.7-', 'E:\\Programmer Language\\NodeJS\\untitled1', 'C:\\Windows\\SYSTEM32\\', 'C:\\Python27\\DLLs', 'C:\\Python27\\lib', 'C:\\Python27\\lib\\plat-win', 'C:\\Python27\\lib\\lib-tk', 'C:\\Python27', 'C:\\Python27\\lib\\site-packages', 'c:\\python27\\lib\\site-packages'] Server time: Fri, 28 Oct 2016 11:43:05 +0000"}
{"_id":"q-en-django__django-16600-b82acc39325d52c46339d2f985c6bbfab2b3148a051f424fd1ab5b32beec765f","text":"Field position reference for aggregate ends up in group-by clause Description Changeset introduced a regression. Aggregate queries are rejected by the database due to the aggregated field being to the GROUP BY clause. It was difficult for me to pin down, especially because it looks like the error only occurs on the second evaluation of the query. The first query is executed just fine and doesn't contain the position reference to the aggregated field in the GROUP BY, only the second one. Below is a test to reproduce the behaviour: diff --git index .. a b class AggregationTests(TestCase): lambda b: (b.name, b.authorCount), ) 1324 def testquotingaggregateorderbyf(self): 1325 author = (name=\"Peter Norvig\") 1326 qs = ( 1327 () 1328 .annotate(num=Count(\"authors\")) 1329 .orderby(F(\"num\").desc()) 1330 ) 1331 list(qs.iterator()) 1332 list(qs.iterator()) 1333 def test_stddev(self): self.assertEqual( Book.objects.aggregate(StdDev(\"pages\")),"}
{"_id":"q-en-django__django-16603-12ba2548f3ce6b935b204b8bc83843e02bc63ca73d4f03bffa493f691a954d27","text":"ASGI http.disconnect not handled on requests with body. Description Noticed whilst reviewing ​PR for , we're not handling the ASGI http.disconnect message correctly. Since it's only dealt with whilst reading the request body, http.disconnect is not processed on a request that includes a body. ​ async def readbody(self, receive): \"\"\"Reads an HTTP body from an ASGI connection.\"\"\" # Use the tempfile that auto rolls-over to a disk file as it fills up. bodyfile = tempfile.SpooledTemporaryFile( maxsize=settings.FILEUPLOADMAXMEMORYSIZE, mode=\"w+b\" ) while True: message = await receive() if message[\"type\"] == \"http.disconnect\": # This is the only place is handled. () # Early client disconnect. raise RequestAborted() # Add a body chunk from the message, if provided. if \"body\" in message: (message[\"body\"]) # Quit out if that's the end. if not (\"morebody\", False): break (0) return bodyfile http.disconnect is designed for long-polling — so we imagine a client opening a request, with a request body, and then disconnecting before the response is generated. The protocol server (Daphne\/uvicorn\/...) will send the http.diconnect message, but it's not handled. This test fails on main (at — 4.2 pre-alpha) diff --git index .. --- +++ -188,6 +188,18 class ASGITest(SimpleTestCase): with self.assertRaises(asyncio.TimeoutError): await communicator.receiveoutput() async def testdisconnectwithbody(self): application = getasgiapplication() scope = self.asyncrequestfactory.basescope(path=\"\/\") communicator = ApplicationCommunicator(application, scope) await communicator.sendinput({ \"type\": \"http.request\", \"body\": b\"some body\", }) await communicator.sendinput({\"type\": \"http.disconnect\"}) with self.assertRaises(asyncio.TimeoutError): await communicator.receiveoutput() async def testwrongconnectiontype(self): application = getasgiapplication() scope = self.asyncrequestfactory.basescope(path=\"\/\", type=\"other\") To handle this correctly it looks like we'd need something like Channel's ​ to keep receiving from the input queue whilst dispatching the request."}
{"_id":"q-en-django__django-16612-8ab57b8c2ed52536b9a0b6bdfdbd3c9e4be39a2c4add739f5a2cb7954584ba5b","text":"AdminSite.catchallview() drops query string in redirects Description introduced AdminSite.catchallview(). However, in the process it broke the ability to redirect with settings.APPENDSLASH = True when there are query strings. Provided URL: ​http:\/\/127.0.0.1:8000\/admin\/auth\/foo?id=123 Expected redirect: ​http:\/\/127.0.0.1:8000\/admin\/auth\/foo\/?id=123 Actual redirect: ​http:\/\/127.0.0.1:8000\/admin\/auth\/foo\/ This seems to be because the redirect in question does not include the query strings (such as via request.META['QUERYSTRING']): return HttpResponsePermanentRedirect(\"%s\/\" % ) ​"}
{"_id":"q-en-django__django-16614-45e1e2de3da338bd3ce28ef70fff47f314a47c233620364bf49c94179cad18f5","text":"Make URLField assume \"https\". Description In the assumption is made that the http (no S) is a good default scheme for URLs that do not specify a scheme when submitted. Entering in a URLField will give as cleaned data. Ref: ​ I think URLField should assume the safe option https. I've notified the security team, and they didn't see this as a security issue."}
{"_id":"q-en-django__django-16629-02a6594f486f2aafca65d8a6d8c5bc2bf3433aa02fb1a1c7f97300d7092dca24","text":"Added support for direct usage of Choices classes on model fields Description (last modified by T. Franzel) Hi, I would like to propose a feature addition on how Choices are handled when used on model fields. Currently, Field.choices only accepts iterables. This has 2 shortcommings imho: Rejecting a Choices class as argument to Field(choices=...) seems counter-intuitive. Providing the class directly currently results in a fields.E005 error. To make this more pythonic, the field should also accept the Choice class directly and deal with the variation internally. I really can't come up with a scenario where a user would want a different behavior or rather that a user would be surprised by the implicit resolution. forcing the user to expand the Choices class manually, essentially all meta information is lost. Downstream packages may benefit from this lost information. Specifically, as maintainer of ​drf-spectcular (OpenAPI generator for DRF), I am interested in the name of the choice set (e.g. Suit, Vehicle, Gender) and potentially also the docstring. This would greatly improve OpenAPI generation of choice sets and take out the unnecessary guesswork to find a proper name. (And if anyone wonders, the model field name is not a good candidate for a choice set name.) This PR allows to use Choices classes directly as argument, while being transparent. No behavioral changes otherwise. I marked this as dev, but it would be awesome if it could still slip into 4.2. Not sure if the feature window is still open, but this is more or less a trivial and backwards-compatible change with little risk. PR is still missing some docs, which I will write if this is considered. class Suit(models.IntegerChoices): \"\"\" All possible card categories in a deck \"\"\" DIAMOND = 1, (\"Diamond\") SPADE = 2, (\"Spade\") HEART = 3, (\"Heart\") CLUB = 4, (\"Club\") class Choiceful(models.Model): # CURRENTLY: fromenumold = models.IntegerField(choices=Suit.choices) # NEW: raised an fields.E005 prior to proposed PR. Now, retains reference # to class and transparently resolves via implicit call from_new = models.IntegerField(choices=Suit)"}
{"_id":"q-en-django__django-16631-06c2d5d276401ae643f40376a10fbb367f128f6653d3ff0d594128ed610b2c0d","text":"SECRETKEYFALLBACKS is not used for sessions Description I recently rotated my secret key, made the old one available in SECRETKEYFALLBACKS and I'm pretty sure everyone on our site is logged out now. I think the docs for ​SECRETKEYFALLBACKS may be incorrect when stating the following: In order to rotate your secret keys, set a new SECRETKEY and move the previous value to the beginning of SECRETKEYFALLBACKS. Then remove the old values from the end of the SECRETKEYFALLBACKS when you are ready to expire the sessions, password reset tokens, and so on, that make use of them. When looking at the Django source code, I see that the ​saltedhmac function uses the SECRETKEY by default and the ​AbstractBaseUser.getsessionauthhash method does not call salted_hmac with a value for the secret keyword argument."}
{"_id":"q-en-django__django-16635-52c61a1019050cddd18c1a72aaa1907560d4496a340f618aad7b70bbe00e4600","text":"Migrations tries to add constraint before adding a foreign key. Description (last modified by Raphael Beekmann) Hello, I have a model, already created through previous migrations, and in a new migration I a new field with an UniqueConstraint. The migrations script try to create the constraint first and then the new field, resulting an error : NewModel has no field named 'category' To reproduce the bug : Create a project with two models linked together with a One-to-Many relation and an unique constraint : class Type(models.Model): name = models.CharField(maxlength=10) class Model(models.Model): name = models.CharField(maxlength=10) type = models.ForeignKey(Type, ondelete=models.SETNULL, null=True) date = models.DateField(autonow=True) class Meta: constraints = ( models.UniqueConstraint(fields=('date', 'type'), name='uniquetypefordate'), ) Create a migration file with makemigrations Add a new model with another One-to-Many relation and unique constraint. The models looks like this : class Type(models.Model): name = models.CharField(maxlength=10) class Category(models.Model): name = models.CharField(maxlength=10) class Model(models.Model): name = models.CharField(maxlength=10) type = models.ForeignKey(Type, ondelete=models.SETNULL, null=True) category = models.ForeignKey(Category, ondelete=models.SETNULL, null=True) date = models.DateField(autonow=True) class Meta: constraints = ( models.UniqueConstraint(fields=('date', 'type'), name='uniquetypefordate'), models.UniqueConstraint(fields=('date', 'category'), name='uniquecategoryfordate'), ) Create a new migration file. The order of the migration's steps are incorrect and the migration crash : class Migration(migrations.Migration): dependencies = [ ('app', '0001initial'), ] operations = [ migrations.CreateModel( name='Category', fields=[ ('id', models.BigAutoField(autocreated=True, primarykey=True, serialize=False, verbosename='ID')), ('name', models.CharField(maxlength=10)), ], ), migrations.AddConstraint( modelname='model', constraint=models.UniqueConstraint(fields=('date', 'category'), name='uniquecategoryfordate'), ), migrations.AddField( modelname='model', name='category', field=models.ForeignKey(null=True, , to='app.category'), ), ]"}
{"_id":"q-en-django__django-16642-02f56dca4f5405ae6f693729dd73e8ebda7436a2218c48eb670a2a5e95ffbdde","text":"Improper guessing of Mime Type for \"br\" and \"Z\" file types Description Below FileResponse will set the content type as text\/html, even if the last file extension is \"Z' or \"br\". FileResponse(open('', 'rb')) FileResponse(open('', 'rb'))"}
{"_id":"q-en-django__django-16649-ba25b2d4a612e5fdd7db47697a79355124b2ca5c180373fde52afb85de45fa0f","text":"Querysets: annotate() columns are forced into a certain position which may disrupt union() Description (Reporting possible issue found by a user on #django) Using values() to force selection of certain columns in a certain order proved useful unioning querysets with union() for the aforementioned user. The positioning of columns with annotate() is not controllable with values() and has the potential to disrupt union() unless this fact is known and the ordering done in a certain way to accommodate it. I'm reporting this mainly for posterity but also as a highlight that perhaps this should be mentioned in the documentation. I'm sure there are reasons why the annotations are appended to the select but if someone feels that this is changeable then it would be a bonus outcome."}
{"_id":"q-en-django__django-16657-ce290f073e3e08b9b564f800b32ad6267ede0dae89cc62973683f69dc4cc89a6","text":"() on querysets in descending order by annotations. Description When I execute Model.objects.annotate(messagelength=Length('message')).orderby('-messagelength').update(text=\"Can I come on board?\") I get the error FieldError: Cannot resolve keyword 'messagelength' into field. Choices are: message, id, text, x because the code ignores the descending case."}
{"_id":"q-en-django__django-16661-62b6fb3c8014a8457cd2cb0c244fb1091eba1c0703c0a3e8d8df4c5d6302afa4","text":"ModelAdmin.lookupallowed() incorrectly raises DisallowedModelAdminLookup lookup with foreign key as primary key Description (last modified by Tim Graham) Wrote a failing test for to demonstrate - same test\/code passes on 1.8 def testlookupallowedforeignprimary(self): class Country(models.Model): name = models.CharField(maxlength=256) class Place(models.Model): country = models.ForeignKey(Country, models.CASCADE) class Restaurant(models.Model): place = models.OneToOneField(Place, models.CASCADE, primarykey=True) class Waiter(models.Model): restaurant = models.ForeignKey(Restaurant, models.CASCADE) class WaiterAdmin(ModelAdmin): listfilter = [ 'restaurantplacecountry', ] ma = WaiterAdmin(Waiter, ) self.assertIs(ma.lookupallowed('restaurantplacecountry', 'testvalue'), True) I think this is caused by the admin thinking that having a foreign key field as a primary key is the same as concrete inheritance. So when you try and check lookups for restaurantplacecountry it thinks 'place' is the concrete parent of 'restaurant' and shortcuts it to restaurantcountry which isn't in 'listfilter'. And you can't add restaurantcountry to list_filter because country isn't actually on restaurant."}
{"_id":"q-en-django__django-16662-89703df4ce25a6a278c008a1ce1e90e9a440f0d28965d3ecb9058ead6f4b992e","text":"Migration import ordering violates coding style and isort defaults Description New migration files are generated with imports sorted by module, independent of import style. For example: import datetime from import migrations, models import time The ​Django coding style specifies: Place all import module statements before from module import objects in each section. This guidance is the same as what isort does by default, ​as documented here. Newly generated migrations can fail isort for this reason. This would mean migration files should instead be generated like this: import datetime import time from import migrations, models For reference, previous issues related to migration import sorting: , ."}
{"_id":"q-en-django__django-16667-4561346794a80c1d6cdc7899ec2df6b4cc99ea6abfcb937cef586a10372caa2b","text":"SelectDateWidget can crash with OverflowError. Description Given a relatively common view like this: from django import forms from import SelectDateWidget from import HttpResponse class ReproForm(forms.Form): mydate = forms.DateField(widget=SelectDateWidget()) def reproview(request): form = ReproForm(request.GET) # for ease of reproducibility if form.isvalid(): return HttpResponse(\"ok\") else: return HttpResponse(\"not ok\") urlpatterns = [path('repro\/', views.reproview, name='repro')] A user can trigger a server crash, reproducible by running locally and visiting ​http:\/\/127.0.0.1:8000\/repro\/?mydateday=1&mydatemonth=1&mydateyear=, which results in [...] - ERROR - django.request: Internal Server Error: \/repro\/ Traceback (most recent call last): [...] File \"[...]\/site-\", line 1160, in valuefromdatadict datevalue = (int(y), int(m), int(d)) OverflowError: signed integer is greater than maximum This can be triggered similarly for a post request. The issue happens as part of the validation logic run in form.isvalid, specifically, when calling the SelectDateWidget.valuefromdatadict, where the user-controlled value is converted into a date without guarding against a possible OverflowError. Specifically, y, m and d are user controlled, and the code does this: date_value = (int(y), int(m), int(d)) When large integers (larger than sys.maxsize) are supplied to date's constructor it will throw an OverflowError: import datetime, sys (sys.maxsize+1, 3, 4) Traceback (most recent call last): File \"<stdin\", line 1, in <moduleOverflowError: Python int too large to convert to C long"}
{"_id":"q-en-django__django-16670-b8868e5f2f70b7b14417126162afa6acbf84c802b2729c6c347ec7152fb6a6f3","text":"ASGIStaticFilesHandler raises warning. Description (last modified by Carlton Gibson) async support to StreamingHttpResponse for Django 4.2. With Django 4.2rc1 and Daphne, ASGIStaticFilesHandler raises a warning about the incorrect iterator type being used: Warning: StreamingHttpResponse must consume synchronous iterators in order to serve them asynchronously. Use an asynchronous iterator instead. This is because FileResponse is not async compatible (nor is that likely, since neither is open()). A new project with: INSTALLEDAPPS = [ \"daphne\", ... ] ASGIAPPLICATION = \"\" runserver, and then visit any page serving static files such as \/admin\/. The fix is to have ASGIStaticFilesHandler adapt the iterator in the same way StreamingHttpResponse does."}
{"_id":"q-en-django__django-16686-1d0985dd151ac1d06b66cc32e1fdc3435b548bbd77d370dae83fc4f77ae8ccf9","text":"values() doesn't mention annotations as valid choices on resolving error Description When doing Book.objects.annotate(annotation=Value(1)).values(\"annotationtype\") The resulting FieldError doesn't mention annotation as a valid choice FieldError: Cannot resolve keyword 'annotationtypo' into field. Choices are: age, book, bookcontactset, friends, id, name"}
{"_id":"q-en-django__django-16693-b5b8c519dd96758be6b4c1f5b8a207db2914e262424860b176e54b8b0358f9fb","text":"Improve error message for incorrect context processors Description If you write a template context processor but return the wrong type, for example None by missing the return statement: def data(request): data = {\"something\": True} Then currently the error message is very mysterious: Internal Server Error: \/ Traceback (most recent call last): ... File \"\/....\/site-\", line 61, in render return (context) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\/....\/site-\", line 173, in render with context.bindtemplate(self): File \"\", line 137, in enter return next() ^^^^^^^^^^^^^^ File \"\/....\/site-\", line 254, in bindtemplate (processor(self.request)) TypeError: 'NoneType' object is not iterable If a view returns the wrong type, Django raises a nice error message: The view didn't return an HttpResponse object. It returned None instead. I suggest we do the same for context processors. If we use try\/except around the () line, it will not slow down normal processing any noticeable amount, thanks to Python 3.11's “zero-cost” exception handling: ​"}
{"_id":"q-en-django__django-16707-74669d0dc29e9570c520eaa6c0a2a66a857976a9c990d22d61994139998a964d","text":"DateField.inputformats cannot be printed Description I am trying to display the input formats for This seems to fail at the moment when using str, while repr works. Example code: from import DateField print(repr(DateField().inputformats)) print(DateField().input_formats) This will result in the following error: Traceback (most recent call last): File \"<console\", line 1, in <moduleTypeError: str returned non-string (type list) I would have expected the string representation to be available as well instead of failing with an internal Python error."}
{"_id":"q-en-django__django-16735-f8a7569c15e22bbd64e60323fd9f5cf2e06c8f3efec5283071a4ab6359e71e6f","text":"i18npatterns() not respecting prefixdefaultlanguage=False Description (last modified by Oussama Jarrousse) In my django project file I have the following setup: from import i18npatterns from django.contrib import admin from import include from import path urlpatterns = [] urlpatterns += i18npatterns( path(\"admin\/\", ), prefixdefaultlanguage=False ) In versions Django==4.1.7 (or prior), I was able to navigating to \/admin\/ without having to add the language prefix. Django==4.2.0, navigating to \/admin\/ will cause a HTTP 302 and only \/en\/admin\/ works... although prefixdefault_language=False is explicitly defined. This change broke my API upon backend packages upgrade from 4.1.7 to 4.2.0"}
{"_id":"q-en-django__django-16745-15e1a5af0d3e4bbed0af2127f00deaa254f165b456748dca500348ed873d2f8f","text":"StepValueValidator does not take into account minvalue Description If you define a number input with <input type=\"number\" min=1 step=2, client side this will only allow positive odd numbers. We could generate the same input in a Django form with IntegerField(minvalue=1, stepsize=2) and Field.localize is False, which would then use MinValueValidator and StepValueValidator. We then get into a problem as StepValueValidator always uses 0 as the base, so stepsize=2 only even numbers are allowed. This then conflicts with the client side validation, and the user cannot submit any value for the input. I'm unsure if this is a bug or whether this is just a configuration problem, apologies if so, but the behaviour does seem to me to conflict with how min and step is handled by browsers."}
{"_id":"q-en-django__django-16746-af5aee1e803a72e66ea99a9144c197c81b0984e411234fce8e4e8ebcdc8c9b2e","text":"Error messages customization and internationalization in Paginator class Description There is no ability to change error messages raised by Paginator.validate_number method."}
{"_id":"q-en-django__django-16749-74ffa13c180820549803a56fb38b23a4e799be374a629cdf328d465eee46f02f","text":"ASGIRequest doesn't respect settings.FORCESCRIPTNAME. Description For example, I have settings.FORCESCRIPTNAME = '\/some-prefix' I start a django server with command: daphne And I navigate to the ​http:\/\/localhost:8000\/admin\/login, and see the login form action url is \"\/admin\/login\" which is wrong, which should be \"\/some-prefix\/admin\/login\""}
{"_id":"q-en-django__django-16750-eb9649e949e018b2a150c0fc5168262df385928db6cf676464ab604efbd8bd2b","text":"Annotating with Chr raises ValueError Description (last modified by Jacob Walls) Author.objects.annotate(firstletter=Chr(Value(77))) File \"\", line 369, in <lambdaelse int(value) ValueError: invalid literal for int() with base 10: 'M' I'm suggesting in Chr we can set outputfield = CharField() to prevent this."}
{"_id":"q-en-django__django-16757-4cf155532db0e6ce2e06c196d7ed18a2979d695bae0a0d5ed6a39c4494ab7d16","text":"Admin check for reversed foreign key used in \"listdisplay\" Description Currently the admin site checks and reports an admin.E109 system check error when a ManyToManyField is declared within the listdisplay values. But, when a reversed foreign key is used there is no system error reported: System check identified no issues (0 silenced). April 10, 2023 - 19:04:29 Django version 5.0.dev20230410064954, using settings 'projectfromrepo.settings' Starting development server at http:\/\/0.0.0.0:8000\/ Quit the server with CONTROL-C. and visiting the admin site for the relevant model results in the following error: TypeError: createreversemanytoonemanager.() missing 1 required keyword-only argument: 'manager' [10\/Apr\/2023 19:30:40] \"GET \/admin\/testapp\/question\/ HTTP\/1.1\" 500 Ideally, using a reversed foreign key would also result in a system check error instead of a 500 response. To reproduce, follow the Question and Choice models from the Django Tutorial and add the following to the QuestionAdmin: listdisplay = [\"questiontext\", \"choiceset\"] Then, visit \/admin\/testapp\/question\/ and the following traceback is returned: Internal Server Error: \/admin\/testapp\/question\/ Traceback (most recent call last): File \"\", line 681, in getfield return self.fieldsmap[fieldname] ~~~^^^^^^^^^^^^ KeyError: 'choiceset' During handling of the above exception, another exception occurred: Traceback (most recent call last): File \"\", line 288, in lookupfield f = getnongfkfield(opts, name) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 319, in getnongfkfield field = opts.getfield(name) ^^^^^^^^^^^^^^^^^^^^ File \"\", line 683, in getfield raise FieldDoesNotExist( Question has no field named 'choiceset' During handling of the above exception, another exception occurred: Traceback (most recent call last): File \"\", line 55, in inner response = getresponse(request) ^^^^^^^^^^^^^^^^^^^^^ File \"\", line 220, in getresponse response = () ^^^^^^^^^^^^^^^^^ File \"\", line 111, in render self.content = self.renderedcontent ^^^^^^^^^^^^^^^^^^^^^ File \"\", line 89, in renderedcontent return (context, self.request) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 61, in render return (context) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 171, in render return (context) ^^^^^^^^^^^^^^^^^^^^^ File \"\", line 163, in render return (context) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 1001, in render return SafeString(\"\".join([node.renderannotated(context) for node in self])) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 1001, in <listcompreturn SafeString(\"\".join([node.renderannotated(context) for node in self])) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 962, in renderannotated return (context) ^^^^^^^^^^^^^^^^^^^^ File \"\", line 157, in render return (context) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 163, in render return (context) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 1001, in render return SafeString(\"\".join([node.renderannotated(context) for node in self])) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 1001, in <listcompreturn SafeString(\"\".join([node.renderannotated(context) for node in self])) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 962, in renderannotated return (context) ^^^^^^^^^^^^^^^^^^^^ File \"\", line 157, in render return (context) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 163, in render return (context) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 1001, in render return SafeString(\"\".join([node.renderannotated(context) for node in self])) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 1001, in <listcompreturn SafeString(\"\".join([node.renderannotated(context) for node in self])) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 962, in renderannotated return (context) ^^^^^^^^^^^^^^^^^^^^ File \"\", line 63, in render result = (context) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 1001, in render return SafeString(\"\".join([node.renderannotated(context) for node in self])) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 1001, in <listcompreturn SafeString(\"\".join([node.renderannotated(context) for node in self])) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 962, in renderannotated return (context) ^^^^^^^^^^^^^^^^^^^^ File \"\", line 63, in render result = (context) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 1001, in render return SafeString(\"\".join([node.renderannotated(context) for node in self])) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 1001, in <listcompreturn SafeString(\"\".join([node.renderannotated(context) for node in self])) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 962, in renderannotated return (context) ^^^^^^^^^^^^^^^^^^^^ File \"\", line 45, in render return super().render(context) ^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 258, in render dict = (resolvedargs, resolvedkwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 340, in resultlist \"results\": list(results(cl)), ^^^^^^^^^^^^^^^^^ File \"\", line 316, in results yield ResultList(None, itemsforresult(cl, res, None)) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 307, in init super().init(items) File \"\", line 217, in itemsforresult f, attr, value = lookupfield(fieldname, result, cl.modeladmin) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\", line 301, in lookupfield value = attr() ^^^^^^ TypeError: createreversemanytoonemanager.() missing 1 required keyword-only argument: 'manager' [10\/Apr\/2023 19:30:40] \"GET \/admin\/testapp\/question\/ HTTP\/1.1\" 500"}
{"_id":"q-en-django__django-16759-60db6aea71d28f8c8815078fd5221cd5db379a6b1086cca00c928759f53058b4","text":"lookupallowed fails to consider dynamic listfilter Description Currently, lookupallowed iterates over self.listfilter to determine valid params. This is technically incorrect since the introduction of getlistfilter() on ModelAdmin in 1.5, because it is possible to define a ModelAdmin such that self.listfilter is () but getlistfilter yields SimpleListFilter classes. To correct it, the above code would need to change from: for filteritem in self.listfilter: to for filteritem in self.getlistfilter(request): The problem is that now lookupallowed needs to accept request so that it can pass it back to getlistfilter In Django itself, that's actually reasonably acceptable as a change, because ​it's used infrequently - the only place it's actually used is in ​ChangeList.getfilters, which has access to the request. However, it is overridden ​in the wild without accept args, *kwargs, so it'd not be easy to provide a clean upgrade path."}
{"_id":"q-en-django__django-16801-927d7a36b74543de253b6d59cfd73c24f487685035aa228c705b826d851d6767","text":"ImageField unnecessarily adds a postinit signal handler to the model Description While debugging some performance issues in a Django app, I found a codepath where most of the time was being spent on initializing Django models after fetching from the DB. It turns out that 30% of the time was being spent on evaluating postinit signals because we were using ImageField. However, the postinit signal handler is a noop because we don't use the widthfield \/ heightfield. If widthfield and heightfield are not set, removing the postinit signal should have no effect since the signal handler will return right away. Removing this signal handler gave us a 30-40% speedup on initializing models where ImageField was used."}
{"_id":"q-en-django__django-16802-c16bd131cbd487e3deec6396472905693923f550f5afa9dd4543e423c85ecfff","text":"\"random\" template filter crashes on an empty list. Description first\/last filters applied to a empty list do not raise anything and just returns a empty page. Example: {{ listvar|last }} Rendered with: def index(request): return render(request, '', context={\"listvar\": []}) Will result in empty page: [26\/Apr\/2023 09:15:49] \"GET \/ HTTP\/1.1\" 200 0 As I undestand, 0 means no any characters in response content. The similar situation is with the first tag: {{ listvar|first }} While the random tag raises the IndexError and causes the server to response 500. Code snippet: {{ listvar|random }} In console we have: Internal Server Error: \/ Traceback (most recent call last): File \"\", line 47, in inner response = getresponse(request) File \"\", line 181, in getresponse response = wrappedcallback(request, callbackargs, callbackkwargs) File \"\/home\/alex\/Документы\/django-\", line 8, in index return render(request, '', context={\"listvar\": []}) File \"\", line 19, in render content = loader.rendertostring(templatename, context, request, using=using) File \"\", line 62, in rendertostring return (context, request) File \"\", line 61, in render return (context) File \"\", line 176, in render return (context) File \"\", line 168, in render return (context) File \"\", line 977, in render return SafeString(''.join([ File \"\", line 978, in <listcompnode.renderannotated(context) for node in self File \"\", line 938, in renderannotated return (context) File \"\", line 1038, in render output = self.filterexpression.resolve(context) File \"\", line 728, in resolve newobj = func(obj, argvals) File \"\", line 616, in random return (value) File \"\", line 347, in choice return seq[self._randbelow(len(seq))] IndexError: list index out of range [26\/Apr\/2023 09:30:03] \"GET \/ HTTP\/1.1\" 500 Django version 4.0.1, python version Python 3.9.2 Expected behavior: Since all first, last and random tags returns an element from the list, their conduct in the case when the list is empty must be the same: either all 3 of them should raise an error or all 3 should return empty content."}
{"_id":"q-en-django__django-16810-4889c27b7901088a94c946dd6d7f8f8a9a26e58c2ecc37332eb6bc0a56df08a3","text":"Translatable URL patterns raise 404 for non-English default language when prefixdefaultlanguage=False is used. Description A simple django project with instruction to replicate the bug can be found here: ​github repo In brief: prefixdefaultlanguage = False raises HTTP 404 for the default unprefixed pages if LANGUAGECODE is not \"en\". I think the problem is that the function getlanguagefrompath in returns None in case of failure instead of LANGUAGECODE: ​diff in 4.2 Consequently, other mechanisms are used to get the language (cookies or headers) that do not work neither. Related issue with my last comment adding some extra context: It is the first time I contribute to django, I hope the bug report is OK. I am also willing to write the patch and test if required."}
{"_id":"q-en-django__django-16816-4c3287387d9d7f14009d0686b4332f048f79d3ce7a8fdd76f4be9c64bee3a058","text":"Error E108 does not cover some cases Description (last modified by Baha Sdtbekov) I have two models, Question and Choice. And if I write listdisplay = [\"choice\"] in QuestionAdmin, I get no errors. But when I visit \/admin\/polls\/question\/, the following trace is returned: Internal Server Error: \/admin\/polls\/question\/ Traceback (most recent call last): File \"\", line 334, in labelforfield field = getnongfkfield(, name) File \"\", line 310, in getnongfkfield raise FieldDoesNotExist() During handling of the above exception, another exception occurred: Traceback (most recent call last): File \"\", line 55, in inner response = getresponse(request) File \"\", line 220, in getresponse response = () File \"\", line 111, in render self.content = self.renderedcontent File \"\", line 89, in renderedcontent return (context, self.request) File \"\", line 61, in render return (context) File \"\", line 175, in render return (context) File \"\", line 167, in render return (context) File \"\", line 1005, in render return SafeString(\"\".join([node.renderannotated(context) for node in self])) File \"\", line 1005, in <listcompreturn SafeString(\"\".join([node.renderannotated(context) for node in self])) File \"\", line 966, in renderannotated return (context) File \"\", line 157, in render return (context) File \"\", line 167, in render return (context) File \"\", line 1005, in render return SafeString(\"\".join([node.renderannotated(context) for node in self])) File \"\", line 1005, in <listcompreturn SafeString(\"\".join([node.renderannotated(context) for node in self])) File \"\", line 966, in renderannotated return (context) File \"\", line 157, in render return (context) File \"\", line 167, in render return (context) File \"\", line 1005, in render return SafeString(\"\".join([node.renderannotated(context) for node in self])) File \"\", line 1005, in <listcompreturn SafeString(\"\".join([node.renderannotated(context) for node in self])) File \"\", line 966, in renderannotated return (context) File \"\", line 63, in render result = (context) File \"\", line 1005, in render return SafeString(\"\".join([node.renderannotated(context) for node in self])) File \"\", line 1005, in <listcompreturn SafeString(\"\".join([node.renderannotated(context) for node in self])) File \"\", line 966, in renderannotated return (context) File \"\", line 63, in render result = (context) File \"\", line 1005, in render return SafeString(\"\".join([node.renderannotated(context) for node in self])) File \"\", line 1005, in <listcompreturn SafeString(\"\".join([node.renderannotated(context) for node in self])) File \"\", line 966, in renderannotated return (context) File \"\", line 45, in render return super().render(context) File \"\", line 258, in render dict = (resolvedargs, *resolvedkwargs) File \"\", line 326, in resultlist headers = list(resultheaders(cl)) File \"\", line 90, in resultheaders text, attr = labelforfield( File \"\", line 362, in labelfor_field raise AttributeError(message) AttributeError: Unable to lookup 'choice' on Question or QuestionAdmin [24\/Apr\/2023 15:43:32] \"GET \/admin\/polls\/question\/ HTTP\/1.1\" 500 I suggest that error E108 be updated to cover this case as well For reproduce see ​github"}
{"_id":"q-en-django__django-16819-699a6615bebc804b5b2d54875f74ce0704be7c391d5e23c45ddf36479ef24559","text":"Reduce Add\/RemoveIndex migration operations. Description We should reduce AddIndex\/RemoveIndex operations when optimizing migration operations."}
{"_id":"q-en-django__django-16820-095106eba0691d0714746b7dd958f4dafe16e6288ecff0e0481b11b75627f3cd","text":"Squashing migrations with Meta.indextogether -indexes transition should remove deprecation warnings. Description Squashing migrations with Meta.indextogether -Meta.indexes transition should remove deprecation warnings. As far as I'm aware, it's a 4.2 release blocker because you cannot get rid of the index_together deprecation warnings without rewriting migrations, see comment."}
{"_id":"q-en-django__django-16824-106d7959ed24314a33d8b1dffc01629aa78866739de8439f85ed66af9aa92c5f","text":"dbshell management command can crash if DATABASES[\"name\"] is a PosixPath Description With a sqlite database configured in like this: DATABASES = { \"default\": { \"ENGINE\": \"\", \"NAME\": BASEDIR \/ \"db.sqlite3\", } } exiting dbshell using Control - d sometimes results in this exception: Traceback (most recent call last): File \"\/home\/ubuntu\/virtualenv\/lib\/python3.11\/site-\", line 30, in handle (options[\"parameters\"]) File \"\/home\/ubuntu\/virtualenv\/lib\/python3.11\/site-\", line 28, in runshell (args, env=env, check=True) File \"\", line 571, in run raise CalledProcessError(retcode, , subprocess.CalledProcessError: Command '['sqlite3', PosixPath('\/home\/ubuntu\/planning-poker\/db.sqlite3')]' returned non-zero exit status 1. During handling of the above exception, another exception occurred: Traceback (most recent call last): File \"\/home\/ubuntu\/planning-\", line 30, in <modulemain() File \"\/home\/ubuntu\/planning-\", line 26, in main executefromcommandline() File \"\/home\/ubuntu\/virtualenv\/lib\/python3.11\/site-\", line 442, in executefromcommandline utility.execute() File \"\/home\/ubuntu\/virtualenv\/lib\/python3.11\/site-\", line 436, in execute self.fetchcommand(subcommand).runfromargv() File \"\/home\/ubuntu\/virtualenv\/lib\/python3.11\/site-\", line 412, in runfromargv self.execute(args, cmd_options) File \"\/home\/ubuntu\/virtualenv\/lib\/python3.11\/site-\", line 458, in execute output = (args, options) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\/home\/ubuntu\/virtualenv\/lib\/python3.11\/site-\", line 44, in handle \" \".join(e.cmd), ^^^^^^^^^^^^^^^ TypeError: sequence item 1: expected str instance, PosixPath found coercing each item in e.cmd to string should fix this."}
{"_id":"q-en-django__django-16830-4dd67692c4b4abd604cbabb909bdd98e0dcca86b975685be3e9b88b0530d55d5","text":"forceinsert flag is not passed when saving parents on inherited models. Description We're using non-abstract model inheritance (table per model class) and issuing our own primary keys. When saving we pass forceinsert=True to prevent the extra UPDATE statement that precedes the INSERT. The forceinsert flag is respected on the child table but not on the parent. So given: class ParentModel(models.Model): id = models.BigIntegerField(primarykey=True) class ChildModel(ParentModel): pass ChildModel(id=1).save(forceinsert=True) We'll see queries: UPDATE appparentmodel (no rows affected) INSERT appparentmodel INSERT appchildmodel This is because Model.savebase doesn't pass forceinsert along to Model.saveparents, and onto Model.savetable. Doing so would prevent the extra UPDATE and respect the spirit of the forceinsert feature. This is a change I've made locally and seems to work \/ is pretty straightforward. I'm curious though if there's intent behind not carrying the forceinsert functionality through for parents. I couldn't find any discussion on the topic. For context about why this is particularly problematic in our case -- we're using MySQL w\/ Innodb and Innodb will take out different exclusive locks for the UPDATE and INSERT in question -- so if you're creating new ChildModel instances in parallel you can get deadlocks when multiple threads issue query and then need a lock with insert intention in order to get ."}
{"_id":"q-en-django__django-16858-dbea97a3d9e14a725b2d1099fd8438fcbaab2f9f5c6053b6fe3d94f4fdd007f9","text":"no longer called for JSONField Description (last modified by Julie Rymer) Hello, I just upgraded from django 4.1 to 4.2 and I have a custom JSONField with a getprepvalue() override that stopped working. After searching a bit, I saw that was because JSONField.getprepvalue() is no longer called in 4.2 (​). I think this issue need a resolution either: JSONField should call getprepvalue() like all other fields type, because this is the method that the ​documentation tell us to override in custom fields. Otherwise we need to override getdbprepvalue() which is heavier and does not have the same purpose. I think simply replacing (value, self.encoder) with (self.getprepvalue(value), self.encoder) in JSONField.getdbprepvalue() would fix this If there is a good reason to no longer call getprepvalue(), this exception for JSONField should be clearly documented in ​custom getprepvalue() doc. It should also be to ​Backwards incompatible changes in 4.2 release note because I got stuck with this issue with no warning when migrating. PS: seems to be related but in fact is about Django 3.2 so it is not the current issue"}
{"_id":"q-en-django__django-16865-5231e08cd1fded028c0e2bc61c8a4e4ede510a9827f815381c3dcc6148de77ad","text":"makemigrations --update should respect the --name option. Description (last modified by David Sanders) This may be intentional behaviour but the docs don't mention this so creating a ticket to update docs or correct behaviour: If you create a migration with a custom name: $ makemigrations --name foo Migrations for 'updaterename': Create model Foo then running --update will change the name \"foo\" to the autogenerated one based on the operations: $ makemigrations --update Migrations for 'updaterename': Create model Foo Deleted My opinion is that it shouldn't as it violates the principle of least astonishment even though the --name argument wasn't supplied. EDIT: This is my first time using --update and here are a few other observations which could indicate that it requires broader discussion: It doesn't utilise the --name argument so it's impossible to customise the name during --update It'd be nice to provide --no-optimize option to --update, here's my use-case: 3-step non-null field addition. After doing nullable step 1, elidable data migration step 2, I want to merge the step 3 non-null update into the migration but --update optimizes this into a single step. Perhaps --update requires a rethink?"}
{"_id":"q-en-django__django-16873-de4e87412a8e8e6b51b050ba6bf3c8700c345d168ae3107f3f801b816f3cf28a","text":"Template filter should not escape the joining string if is Description Consider the following template code snippet: {% autoescape off %} {{ somelist|join:somevar }} {% endautoescape %} in this case, the items inside somelist will not be escaped (matching the expected behavior) but somevar will forcibly be escaped. From the docs for autoescape or join I don't think this is expected behavior. The following testcase illustrates what I think is a bug in the join filter (run inside the templatetests\/filtertests folder): from django.template.defaultfilters import escape from import SimpleTestCase from ..utils import setup class RegressionTests(SimpleTestCase): '{{ somelist|join:somevar }}'}) def testjoin01(self): somelist = [\"<pHello World!<\/p\", \"beta & me\", \"<scriptHi!<\/script\"] somevar = \"<br\/\" output = (\"join01\", {\"somelist\": somelist, \"somevar\": somevar}) self.assertEqual(output, escape((somelist))) '{% autoescape off %}{{ somelist|join:somevar }}{% endautoescape %}'}) def testjoin02(self): somelist = [\"<pHello World!<\/p\", \"beta & me\", \"<scriptHi!<\/script\"] somevar = \"<br\/\" output = (\"join02\", {\"somelist\": somelist, \"somevar\": somevar}) self.assertEqual(output, (somelist)) Result of this run in current main is: .F FAIL: testjoin02 (templatetests.filtertests.testregression.RegressionTests.testjoin02) Traceback (most recent call last): File \"\", line 443, in inner return func(args, *kwargs) ^^^^^^^^^^^^^^^^^^^^^ File \"\", line 58, in inner func(self) File \"\", line 21, in testjoin02 self.assertEqual(output, (somelist)) AssertionError: '<pHello World!<\/p&lt;br\/&gt;beta & me&lt;br\/&gt;<scriptHi!<\/script' != '<pHello World!<\/p<br\/beta & me<br\/<scriptHi!<\/script' Ran 2 tests in 0.007s"}
{"_id":"q-en-django__django-16877-89eaf8bf714d532a1a058321768be9f71f874973269657da0aea0151540aae6e","text":"New template filter Description Following , and after some conversations within the security team, it seems appropriate to provide a new template filter escapeseq which would be to escape what safeseq is to safe. An example of usage would be: {{ somelistjoin:\",\" }} where each item of somelist is escaped before applying the join operation. This usage makes sense in a context where autoescape is off."}
{"_id":"q-en-django__django-16879-f98b8aafbfa3b91492f8407c3277ad2e71c75f8d7f2e757a8ab6abdb1af56291","text":"Case-When aggregation over aggregated fields doesn't work since 4.2 Description (last modified by Denis Roldán) This same QuerySet was working on 4.1.X and prior versions and stopped working on 4.2.X: annotatedusers = usersqs.annotate( totalthings=Subquery( (userid=OuterRef(\"pk\")).annotate( totalobjs=F(\"total\") ).values(\"totalobjs\") ) ) annotatedusers.aggregate( sumtotalobjs=Sum(\"totalthings\"), avgconversionrate=Case( When( sumtotalobjs=0, then=0, ), default=Round( (Sum(\"sumtotalconfirmedobjs\") \/ Sum(\"sumtotalobjs\")) 100, 2 ), outputfield=FloatField(), ) ) As you can see sumtotalobjs is an aggregated field that is also used on a second field to calculate the conversion rate. To avoid a zero division problem, we were using a Case-When clause over that field. It works well on any 4.1 and prior versions but stopped working since 4.2, raising a FieldError like: Cannot resolve keyword 'sumtotalobjs' into field The bug is reproducible with an extra test on the django aggregation test suite: def testreferencedgroupbyaggregationoverannotation(self): totalbooksqs = ( (authorspk=OuterRef(\"pk\")) .orderby() .values(\"pk\") .annotate(total=Count(\"pk\")) .values(\"total\") ) annotatedauthors = Author.objects.annotate( totalbooks=Subquery(totalbooksqs.annotate( totalbooks=F(\"total\") ).values(\"totalbooks\")), totalbooksa=Subquery(( nameistartswith=\"a\" ).annotate( totalbooksa=F(\"total\") ).values(\"totalbooksa\")), ).values( \"pk\", \"totalbooks\", \"totalbooksa\", ).orderby(\"-totalbooks\") totals = annotatedauthors.aggregate( sumtotalbooks=Sum(\"totalbooks\"), sumtotalbooksa=Sum(\"totalbooksa\"), aovertotalrate=Case( When( sumtotalbooks=0, then=0, ), default=Round( (Sum(\"totalbooksa\") \/ Sum(\"totalbooks\")) 100, 2 ), outputfield=FloatField(), ), ) self.assertEqual(totals['sumtotalbooks'], 3) self.assertEqual(totals['sumtotalbooksa'], 0) self.assertEqual(totals['aovertotalrate'], 0) Thanks for the support!"}
{"_id":"q-en-django__django-16883-2cbd1c057a56b507cf5835c6bc204b94d30b7f169a4745b3525eb622fcd87271","text":"Allow to override table2model conversion for inspectdb command Description When implementing a custom inspectdb command by inheriting from inspectdb.Command users should be able to override the table name to model conversion logic. Public method for column names already exists: ​ This would allow overriding the default behaviour of command in a custom inspectdb, for example when table names are already PascalCase (i.e. schema.FooBarBuzz), etc. With default implementation, it will be converted into class SchemaFoobarbuzz(models.Model). Proposed PR: ​"}
{"_id":"q-en-django__django-16888-240065f21995541d8fb98d1e502e8c2347eb064088382ef3833a9a54ca1a4379","text":"Querying for decimals larger than maxdigits crashes on SQLite Description Introduced in: Refs -- Improved adapting DecimalField values to decimal. Description: I have DecimalField with maxdigits=4. Previously, querying for something larger (e.g. ) would execute the sql and return ObjectNotFound. Now, in 4.2, it throws a decimal.InvalidOperation error, as it tries to quantize the value to have 4 digits. I understand that it doesn't make sense to query for a larger number, but the error that occurs was pretty confusing to me. Also, it is not as easy to check in my application, because I don't have easy access to the maxdigits parameter of the field. In my opinion, the backend should either accept larger values and always return \"not found\", or the error should be more descriptive, so that it can be caught specifically. Testcase: placed in tests\/backends folder and used for git bisect import decimal from import models from import TestCase class DecimalModel(models.Model): decfield = models.DecimalField(decimalplaces=0, maxdigits=4) class InvalidDecimalQuery(TestCase): def testinvaliddecimalquery(self): try: (decfield='') except decimal.InvalidOperation: (\"Too large decimal query caused exception.\") except DecimalModel.DoesNotExist: pass Stacktrace: Traceback (most recent call last): File \"lib\/python3.10\/site-\", line 87, in managermethod return getattr(self.getqueryset(), name)(args, *kwargs) File \"lib\/python3.10\/site-\", line 633, in get num = len(clone) File \"lib\/python3.10\/site-\", line 380, in len self.fetchall() File \"lib\/python3.10\/site-\", line 1881, in fetchall self.resultcache = list(self.iterableclass(self)) File \"lib\/python3.10\/site-\", line 91, in iter results = compiler.executesql( File \"lib\/python3.10\/site-\", line 1547, in executesql sql, params = () File \"lib\/python3.10\/site-\", line 762, in assql self.compile() if is not None else (\"\", []) File \"lib\/python3.10\/site-\", line 544, in compile sql, params = (self, self.connection) File \"lib\/python3.10\/site-\", line 145, in assql sql, params = compiler.compile(child) File \"lib\/python3.10\/site-\", line 544, in compile sql, params = (self, self.connection) File \"lib\/python3.10\/site-\", line 357, in assql return super().assql(compiler, connection) File \"lib\/python3.10\/site-\", line 225, in assql rhssql, rhsparams = self.processrhs(compiler, connection) File \"lib\/python3.10\/site-\", line 126, in processrhs return self.getdbpreplookup(value, connection) File \"lib\/python3.10\/site-\", line 254, in getdbpreplookup else [getdbprepvalue(value, connection, prepared=True)], File \"lib\/python3.10\/site-\", line 1761, in getdbprepvalue return ( File \"lib\/python3.10\/site-\", line 574, in adaptdecimalfieldvalue return utils.formatnumber(value, maxdigits, decimalplaces) File \"lib\/python3.10\/site-\", line 304, in formatnumber value = value.quantize( decimal.InvalidOperation: [<class 'decimal.InvalidOperation']"}
{"_id":"q-en-django__django-16899-20ca762feffd29f78ad79dd13dc0664cc0974fd78211e9410b3c13b8228e6d95","text":"ModelAdmin: Error message for readonlyfields's check does not include the field name Description When subclassing a ModelAdmin, the current error message for the readonlyfields would indicate the index of the value at fault but it will not include the field's name (from the test suite): The value of 'readonlyfields[0]' is not a callable, an attribute of 'CityInline', or an attribute of 'adminchecks.City'. Other fields like listeditable, rawidfields, listdisplay, etc. would also include this value: The value of 'listeditable[0]' refers to 'originalrelease', which is not contained in 'listdisplay'. It would be good if we can unify this and include the field name in the readonlyfields checks, it also eases the understanding of the error when using the framework."}
{"_id":"q-en-django__django-16901-7283cb477cfae7cbe456d634d4ace3acdaefb3a5a2c411142e8c8b1b94e757bc","text":"On databases lacking XOR, Q(…) ^ Q(…) ^ Q(…) wrongly interpreted as exactly-one rather than parity Description On databases that don’t natively support XOR, such as PostgreSQL, Django generates incorrect fallback SQL for Q(…) ^ Q(…) ^ Q(…) with more than 2 arguments. The ​correct interpretation, and the interpretation of databases natively supporting XOR (e.g. ​MySQL), is that a ^ b ^ c is true when an odd number of the arguments are true. But Django’s fallback interpretation is that a ^ b ^ c is true when exactly one argument is true: from import Q from import Client (Q(id=37)).count() 1 (Q(id=37) ^ Q(id=37)).count() 0 (Q(id=37) ^ Q(id=37) ^ Q(id=37)).count() 0 (Q(id=37) ^ Q(id=37) ^ Q(id=37) ^ Q(id=37)).count() 0 (Q(id=37) ^ Q(id=37) ^ Q(id=37) ^ Q(id=37) ^ Q(id=37)).count() 0 (Expected: 1, 0, 1, 0, 1.) This was introduced in ."}
{"_id":"q-en-django__django-16902-9a365f6e67a8c3a770e93fef436919990054874f775a7dc8957586de79321454","text":"Right() function on Oracle and SQLite returns improper value when the length is zero. Description Hi I have found that the RIGHT database function on Oracle returns the whole string instead of empty string when the given length is 0. You can't explicitly give 0 to the RIGHT function, but it may be computed by the database. Basic example (you can use any model with a CharField): from import Right, Length MyModel.objects.annotate(suffix=Right(\"foo\", Length(\"foo\") - Length(\"foo\"))) On PostgreSQL this will return an empty string under the field suffix, but on Oracle this will return the whole contents of the field foo. This is because Django uses the SUBSTR function on Oracle by multiplying the given length value by -1 and giving it as a position argument. I think it is not intended behavior and it should return the empty string as PostgreSQL does. Or at least be documented as a Note in the Right function documentation."}
{"_id":"q-en-django__django-16903-214f15273a670a5cf05d6be3a96532ba21ed13cda4d191697d5b377b1229735c","text":"Diamond inheritance causes duplicated PK error when creating an object, if the primary key field has a default. Description Hi, I'm not sure if this is a bug or an unsupported feature. But I looked into the source code and now have a pretty good understanding of what is happening. My business code uses a diamond shape inheritance to model different types of user posts: UserPost, ImagePost, VideoPost, and MixedContentPost. The inheritance goes like this: both ImagePost and VideoPost extend from UserPost, and the MixedContentPost inherits from ImagePost and VideoPost. All of them are concrete models I read the doc and expected it to work, similar to the example class Piece(models.Model): pass class Article(Piece): articlepiece = models.OneToOneField(Piece, ondelete=models.CASCADE, parentlink=True) ... class Book(Piece): bookpiece = models.OneToOneField(Piece, ondelete=models.CASCADE, parentlink=True) ... class BookReview(Book, Article): pass However, I found out that the doc's example only works when these models use a primary key field that does not have a default. In my case, we are using a UUIDField as the primary key with a default of uuid4. Trying to create a BookReview in our case, causes a UNIQUE constraint failed error, because django tries to create the Piece object twice, with the same uuid. The same behavior is found if I used a AutoField on Piece, with a custom default function, such as id = 99 def increment(): global id id += 1 return id This default function makes no sense in practice, but I just use it here to illustrate the root cause of the problem: The savetable method in has a like this: # Skip an UPDATE when adding an instance and primary key has a default. if ( not raw and not forceinsert and and and is not NOTPROVIDED ): forceinsert = True When a default is not present, which is the case of the documentation example, Django will first insert the first instance of the Piece object, and then for the second one, since forceinsert is False, savetable tries an update and goes through. Therefore there is not duplicate. However, if a default is present, then the second Piece becomes an insertion as well (because and is not NOTPROVIDED, forceinsert is True). This causes a duplication on the primary key. On the other hand, insertparent does an in-order traversal calling savetable on each node, so even in the no-default pk case, it is calling a redundant update on the root node after the insertion.. So which function is at fault? The source code savetable assumes that if you are calling it with a default pk then you can skip an update. This assumption looks weird to me: why only when there IS a default pk you can skip update? Why not just skip update as long as we know we are inserting? (via ) Is it just to make it special so that AutoField works? If savetable's responsibility is to try updating before inserting, except when the params force it to do an update or insert, then it shouldn't override that behavior by this self-assumeption within it. I think the solution is to simply move the check to savebase. And don't do this check in saveparents. Like this: def savebase( self, raw=False, forceinsert=False, forceupdate=False, using=None, updatefields=None, ): \"\"\" Handle the parts of saving which should be done only once per save, yet need to be done in raw saves, too. This includes some sanity checks and signal sending. The 'raw' argument is telling savebase not to save any parent models and not to do any changes to the values before save. This is used by fixture loading. \"\"\" using = using or router.dbforwrite(, instance=self) assert not (forceinsert and (forceupdate or updatefields)) assert updatefields is None or updatefields cls = origin = # Skip proxies, but keep the origin as the proxy model. if cls = meta = if not meta.autocreated: ( sender=origin, instance=self, raw=raw, using=using, updatefields=updatefields, ) # A transaction isn't needed if one query is issued. if meta.parents: contextmanager = (using=using, savepoint=False) else: contextmanager = transaction.markforrollbackonerror(using=using) with contextmanager: parentinserted = False if not raw: parentinserted = self.saveparents(cls, using, updatefields) # Skip an UPDATE when adding an instance and primary key has a default. if ( not raw and not forceinsert and and and is not NOTPROVIDED ): forceinsert = True updated = self.savetable( raw, cls, forceinsert or parentinserted, forceupdate, using, updatefields, ) # Store the database on which the object was saved = using # Once saved, this is no longer a to-be- instance. = False # Signal that the save is complete if not meta.autocreated: ( sender=origin, instance=self, created=(not updated), updatefields=updatefields, raw=raw, using=using, ) savebase.alters_data = True I have never contributed to Django before. If you think I'm right on this one I'll look into creating a PR."}
{"_id":"q-en-django__django-16910-05ea1867af3260df7862d6dc9cf968b1c2287bbe0d60ab467c2441b1a8f6c2f1","text":"() doesn't work with selectrelated() on a reverse OneToOneField relation. Description On Django 4.2 calling only() with selectrelated() on a query using the reverse lookup for a OneToOne relation does not generate the correct query. All the fields from the related model are still included in the generated SQL. Sample models: class Main(models.Model): mainfield1 = models.CharField(blank=True, maxlength=45) mainfield2 = models.CharField(blank=True, maxlength=45) mainfield3 = models.CharField(blank=True, maxlength=45) class Secondary(models.Model): main = models.OneToOneField(Main, primarykey=True, relatedname='secondary', ondelete=models.CASCADE) secondaryfield1 = models.CharField(blank=True, maxlength=45) secondaryfield2 = models.CharField(blank=True, maxlength=45) secondaryfield3 = models.CharField(blank=True, maxlength=45) Sample code: Main.objects.selectrelated('secondary').only('mainfield1', 'secondary_secondaryfield1') Generated query on Django 4.2.1: SELECT \"bugtestmain\".\"id\", \"bugtestmain\".\"mainfield1\", \"bugtestsecondary\".\"mainid\", \"bugtestsecondary\".\"secondaryfield1\", \"bugtestsecondary\".\"secondaryfield2\", \"bugtestsecondary\".\"secondaryfield3\" FROM \"bugtestmain\" LEFT OUTER JOIN \"bugtestsecondary\" ON (\"bugtestmain\".\"id\" = \"bugtestsecondary\".\"mainid\") Generated query on Django 4.1.9: SELECT \"bugtestmain\".\"id\", \"bugtestmain\".\"mainfield1\", \"bugtestsecondary\".\"mainid\", \"bugtestsecondary\".\"secondaryfield1\" FROM \"bugtestmain\" LEFT OUTER JOIN \"bugtestsecondary\" ON (\"bugtestmain\".\"id\" = \"bugtestsecondary\".\"main_id\")"}
{"_id":"q-en-django__django-16920-4100a6debf45e260b9eeceeb5e66f1dd69ff94484468a58d83453aedf4a1b8e9","text":"Fields’ help text and errors should be associated with input Description (last modified by Thibaud Colas) With Django’s default field rendering, all field errors are rendered as a list above the field’s label, and help text is rendered after the field’s form element. Example with asp: <ul class=\"errorlist\"<liThis field is required.<\/li<\/ul<p<label for=\"iddurationrequired\"Duration required:<\/label<input type=\"text\" name=\"durationrequired\" required=\"\" id=\"iddurationrequired\"<span class=\"helptext\"Help<\/span<\/pOne problem for screen reader users is that the association between the errors and the field, and between the help text and the field, is only communicated visually. This is a failure of either WCAG 2.1 level A ​SC 1.3.1: Info and Relationships, or ​SC 3.3.2: Labels or Instructions. More importantly, it just makes it harder than necessary for screen reader users to make use of help text, and to identify error messages. The fix is relatively straightforward – using aria-describedby, as documented in the (non-normative) ​ARIA1 Using the aria-describedby property to provide a descriptive label for user interface controls technique. Here is another well-known accessibility-oriented UI library that implements this technique: ​GOV.UK design system – text input with error message. Here is what implementing aria-describedby would look like in the same example as above: <div class=\"errorlist\" id=\"iddurationrequirederrorlist\"<pThis field is required.<\/p<\/div<p<label for=\"iddurationrequired\"Duration required:<\/label<input type=\"text\" name=\"durationrequired\" required=\"\" id=\"iddurationrequired\" aria-describedby=\"iddurationrequirederrorlist iddurationrequiredhelptext\"<span class=\"helptext\" id=\"iddurationrequired_helptext\"Help<\/span<\/pWe have additional id attributes, aria-describedby, and errorlist is no longer a <ul. Result in VoiceOver: Unfortunately I tried to have this with the errorlist kept as a ul, but it wasn’t announced by VoiceOver. I haven’t heard of this limitation before so am not sure why that might be the case – I’d appreciate others taking a look if possible."}
{"_id":"q-en-django__django-16938-8b6994d8d13fe94dbd3b3f64b94ad21802f4573f81a31b3bdb22e45cad24c67c","text":"Serialization of m2m relation fails with custom manager using selectrelated Description Serialization of many to many relation with custom manager using selectrelated cause FieldError: Field cannot be both deferred and traversed using selectrelated at the same time. Exception is raised because performance optimalization . Workaround is to set simple default manager. However I not sure if this is bug or expected behaviour. class TestTagManager(Manager): def getqueryset(self): qs = super().getqueryset() qs = qs.selectrelated(\"master\") # follow master when retrieving object by default return qs class TestTagMaster(models.Model): name = models.CharField(maxlength=120) class TestTag(models.Model): # default = Manager() # solution is to define custom default manager, which is used by RelatedManager objects = TestTagManager() name = models.CharField(maxlength=120) master = models.ForeignKey(TestTagMaster, ondelete=models.SETNULL, null=True) class Test(models.Model): name = models.CharField(maxlength=120) tags = models.ManyToManyField(TestTag, blank=True) Now when serializing object from import serializers from import TestTag, Test, TestTagMaster tagmaster = (name=\"master\") tag = (name=\"tag\", master=tagmaster) test = (name=\"test\") (tag) () serializers.serialize(\"json\", [test]) Serialize raise exception because is not possible to combine selectrelated and only. File \"\/opt\/venv\/lib\/python3.11\/site-\", line 134, in serialize s.serialize(queryset, **options) File \"\/opt\/venv\/lib\/python3.11\/site-\", line 167, in serialize self.handlem2mfield(obj, field) File \"\/opt\/venv\/lib\/python3.11\/site-\", line 88, in handlem2mfield self.current[] = [m2mvalue(related) for related in m2miter] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\/opt\/venv\/lib\/python3.11\/site-\", line 88, in <listcompself.current[] = [m2mvalue(related) for related in m2miter] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\/opt\/venv\/lib\/python3.11\/site-\", line 516, in iterator yield from iterable File \"\/opt\/venv\/lib\/python3.11\/site-\", line 91, in iter results = compiler.executesql( ^^^^^^^^^^^^^^^^^^^^^ File \"\/opt\/venv\/lib\/python3.11\/site-\", line 1547, in executesql sql, params = () ^^^^^^^^^^^^^ File \"\/opt\/venv\/lib\/python3.11\/site-\", line 734, in assql extraselect, orderby, groupby = self.presqlsetup( ^^^^^^^^^^^^^^^^^^^ File \"\/opt\/venv\/lib\/python3.11\/site-\", line 84, in presqlsetup self.setupquery(withcolaliases=withcolaliases) File \"\/opt\/venv\/lib\/python3.11\/site-\", line 73, in setupquery , self.klassinfo, self.annotationcolmap = self.getselect( ^^^^^^^^^^^^^^^^ File \"\/opt\/venv\/lib\/python3.11\/site-\", line 279, in getselect relatedklassinfos = self.getrelatedselections(select, selectmask) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\/opt\/venv\/lib\/python3.11\/site-\", line 1209, in getrelatedselections if not selectrelateddescend(f, restricted, requested, selectmask): ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File \"\/opt\/venv\/lib\/python3.11\/site-\", line 347, in selectrelateddescend raise FieldError( Field cannot be both deferred and traversed using select_related at the same time."}
{"_id":"q-en-django__django-16948-d05fd60a82faeae42425918a55d193d10a33d9ca6e7749a46bc3dec7bd07d67c","text":"Deprecate formathtml calls without args or kwargs Description (last modified by Michael Howitz) In my experience, a common misuse of formathtml is to format the HTML before calling it: formathtml(f\"<i{name}<\/i\") This makes it act like marksafe, allowing data through without escaping. It provides a false sense of security since formathtml is meant to be the \"safe way\". I propose we deprecate calls to formathtml that don’t pass args or kwargs, and eventually raise a TypeError for such cases. (Following improvement to format_html docs in .)"}
{"_id":"q-en-django__django-16950-830f64930aef9c0e4f2e7cdeab7d99dfe88be30766dfacc4ca1410d9dba61479","text":"Django Admin with Inlines not using UUIDField default value Description (last modified by Joseph Metzinger) Hello, I am a long time django user, first time bug reporter, so please let me know if I need to do anything else to help get this bug fixed :) I am using Django 3.1.3 and python 3.8.5 and have cerated a toy project to illustrate the bug. I have the following models: class UUIDModel(models.Model): pkid = models.BigAutoField(primarykey=True, editable=False) id = models.UUIDField(default=uuid.uuid4, editable=False, unique=True) class Meta: abstract = True class Thing(UUIDModel): name = models.CharField(maxlength=191) class SubThing(models.Model): name = models.CharField(maxlength=191) thing = models.ForeignKey( 'bugapp.Thing', tofield='id', ondelete = models.CASCADE, relatedname='subthings', ) And the following file: class SubThingInline(admin.StackedInline): model = SubThing class ThingAdmin(admin.ModelAdmin): list_display = ('name',) ordering = ('pkid',) inlines = (SubThingInline,) When logging into the admin, if you delete all of the entries for \"subthings\", add a name, and save the model, it will work. As soon as you try to add a subthing alongside the main Thing, it fails with the following exception: ​ It shows that the value of \"id\" in the Thing model is being set to null. I believe this is a bug in django. Thanks!"}
{"_id":"q-en-django__django-16952-4428ee82f46a4f31ca00401634091fa579a2a77bcc9194a5607568b8f477aa50","text":"Creating objects with nested MTI crashes. Description (last modified by Mariusz Felisiak) Checking ​PR I noticed that creating objects with more complicated multi-table inheritance crashes. For example: diff --git index .. a b class ItalianRestaurant(Restaurant): servesgnocchi = models.BooleanField(default=False) 109class ItalianRestaurantManyParents(ItalianRestaurant, Place): 110 pass 111 112 109113class Supplier(Place): customers = models.ManyToManyField(Restaurant, relatedname=\"provider\") diff --git index .. a b from .models import ( 1414 GrandChild, 1515 GrandParent, 1616 ItalianRestaurant, 17 ItalianRestaurantManyParents, 1718 MixinModel, 1819 Parent, 1920 ParkingLot, … … class ModelInheritanceTests(TestCase): # accidentally found). self.assertSequenceEqual((), []) 153 def testcreatediamondmticommonparents(self): 154 with self.assertNumQueries(4): 155 ( 156 name=\"Ristorante Miron\", 157 address=\"1234 W. Ash\", 158 ) 159 def testupdateparentfiltering(self): \"\"\" Updating a field of a model subclass doesn't issue an UPDATE crashes with: File \"\", line 155, in testcreatediamondmticommonparents ( File \"\", line 87, in managermethod return getattr(self.getqueryset(), name)(args, *kwargs) File \"\", line 650, in create (forceinsert=True, ) File \"\", line 814, in save self.savebase( File \"\", line 876, in savebase parentinserted = self.saveparents(cls, using, updatefields) File \"\", line 928, in saveparents setattr(self, field.attname, self.getpkval(parent._meta)) AttributeError: 'OneToOneField' object has no attribute 'attname'"}
{"_id":"q-en-django__django-16983-8dfd3ca02184264ce9ea2edf01426e4109e1af8f81b5935ac17d547c627b3686","text":"Add system check for filterhorizontal\/filtervertical on ManyToManyFields with intermediary models. Description Hi team, I'm a huge fan of Django and have been using it since 0.95 but I stumbled over this one. Neither of ​ and ​ call out the requirement to not use ManyToManyField(through=\"\") In the same way: ​ doesn't call out the consequence that filterhorizontal and filtervertical will stop working if one goes down the pathway of: ManyToManyField(through=\"\") I just wasted half a day chasing this down."}
{"_id":"q-en-django__django-17029-b70c30434c14b02d1bd822837470f28968e7007ac283c60cafd75d35ae8d1b08","text":"Apps.clearcache() does not clear getswappablesettingsname cache. Description We use apps.clearcache() in django-stubs to be able to reset the previous state on consequential mypy runs. Code: ​ But, looks like we don't clear all the object's cache this way, because getswappablesettingsname (which is a functools.lrucachewrapper) is not cleared. I think that this is not correct. .clearcache doc states: Clear all internal caches, for methods that alter the app registry. Looks like that is not the case. I propose to add: self.getswappablesettingsname.cacheclear() line to def clearcache. If others agree, I will make a PR. Original discussion: ​"}
{"_id":"q-en-django__django-17045-baaec30d626bf086ca3ec9b48fb275f6cca115127b401e720a990a51490cceb8","text":"Add a system check for unmatched URL paths' < \/ Description Currently, unmatched angle brackets are silently ignored: path('<int:question_id\/vote\/', , name='vote'), This leads to a frustrating debugging experience, where a 404 or NoReverseMatch could occur and only the eagle-eyed could spot why. Similar to the ^\/$ system check in , I propose a new check in () for unmatched < or . It's rather unlikely those characters are legitimately used in URLs. (Inspired by ​a new user making this mistake and asking for help on the forum.)"}
{"_id":"q-en-django__django-17046-34fe35b7a82ca367cb08111f9244336cd3ffb345a85264e8c1432ac505716f95","text":"Deleting objects after searching related many to many field crashes the admin page Description Minimal reproduction: class Post(models.Model): title = models.String(...) authors = models.ManyToMany(\"User\", ...) class User(models.Model): email = models.String(...) class PostAdmin(admin.ModelAdmin): searchfields = (\"title\", \"authorsemail\") then opening the admin site, opening the post page that contains only one post (any title and author assigned) and entering a search term (e.g the first 2 characters of the title), selecting the post and then using the delete action results in an Internal Sever Error 500 with an error\/stack-trace: Internal Server Error: \/admin\/post\/post\/ Traceback (most recent call last): File \"\", line 47, in inner response = getresponse(request) File \"\", line 181, in getresponse response = wrappedcallback(request, callbackargs, callbackkwargs) File \"\", line 616, in wrapper return self.adminsite.adminview(view)(args, kwargs) File \"\", line 130, in wrappedview response = viewfunc(request, args, kwargs) File \"\", line 44, in wrappedviewfunc response = viewfunc(request, args, kwargs) File \"\", line 241, in inner return view(request, args, kwargs) File \"\", line 43, in wrapper return boundmethod(args, kwargs) File \"\", line 130, in wrappedview response = viewfunc(request, args, *kwargs) File \"\", line 1737, in changelistview response = self.responseaction(request, queryset=cl.getqueryset(request)) File \"\", line 1406, in responseaction response = func(self, request, queryset) File \"\", line 45, in deleteselected modeladmin.deletequeryset(request, queryset) File \"\", line 1107, in deletequeryset () File \"\", line 728, in delete raise TypeError('Cannot call delete() after .distinct().') TypeError: Cannot call delete() after .distinct(). \"POST \/admin\/post\/post\/?q=my HTTP\/1.1\" 500 I can confirm that pip install django==3.1.8 fixes the error, and after having a look at the diff between stable\/3.2.x and 3.1.8, I suspect the \"regression\" comes about from the work done on preserving the filters on delete or something along those lines - I haven't done a thorough investigation yet. Presumably .distinct() is being called because of the search involving the many to many field. I am using a Postgres database."}
{"_id":"q-en-django__django-17051-8e3e05ff7fa3cacdb00375db336710e39bcd7152355438ab80d9403f6c83b1b1","text":"Allow returning IDs in QuerySet.bulkcreate() when updating conflicts. Description Currently, when using bulkcreate with a conflict handling flag turned on (e.g. ignoreconflicts or updateconflicts), the primary keys are not set in the returned queryset, as documented in bulkcreate. While I understand using ignoreconflicts can lead to PostgreSQL not returning the IDs when a row is ignored (see ​this SO thread), I don't understand why we don't return the IDs in the case of updateconflicts. For instance: MyModel.objects.bulkcreate([MyModel(...)], updateconflicts=True, updatefields=[...], uniquefields=[...]) generates a query without a RETURNING part: INSERT INTO \"mymodel\" (...) VALUES (...) ON CONFLICT(...) DO UPDATE ... If I append the RETURNING clause, the query is indeed valid and the ID is returned (checked with PostgreSQL). I investigated a bit and ​this in Django source is where the returningfields gets removed. I believe we could discriminate the cases differently so as to keep those returningfields in the case of updateconflicts. This would be highly helpful when using bulkcreate as a bulk upsert feature."}
{"_id":"q-en-django__django-17065-d9c9e3142cc53ee10b3171b9fe2ec6e8177c094e03225f84a3be76956aa10475","text":"BoundField.aswidget() ignores aria-describedby in attrs argument Description (last modified by Sage Abdullah) BoundField.aswidget() ignores aria-describedby that is passed in the attrs argument. Use case: In Wagtail, we have our own mechanism for rendering form fields (called \"Panels\") that is built on top of Django's Forms API. We use BoundField.aswidget() in our rendering process to render only the form field's widget, while its help text element is rendered separately via our Panels API with a custom markup (including HTML id) that conforms to our design system. We've been passing additional attributes to BoundField.aswidget() to improve accessibility, including aria-describedby, to establish the relationship between the field widget rendered by Django and our help text element. As of , Django automatically adds aria-describedby to form fields with a helptext. The logic in buildwidgetattrs() (used by aswidget()) checks for an existing aria-describedby in the field's widget's attrs before automatically generating one. However, it does not take into account the possibility of an existing aria-describedby in the attrs argument. A workaround on Wagtail's side could be to modify the widget's attrs before using aswidget(), but this is not ideal as the widget is customisable by the developer and we would have to copy the widget instance to avoid modifying the existing widget instance (which might be shared across form fields). I believe Django should check for aria-describedby in attrs first, before falling back to and the auto-generated value. Test case: class BoundFieldTests(SimpleTestCase): def testaswidgetwithcustomariadescribedby(self): class TestForm(Form): data = CharField(helptext=\"Some help text\") form = TestForm({\"data\": \"some value\"}) self.assertHTMLEqual( form[\"data\"].aswidget(attrs={\"aria-describedby\": \"customhelptextid\"}), \"\"\" <input type=\"text\" name=\"data\" value=\"some value\" aria-describedby=\"customhelptextid\" required id=\"iddata\"\"\"\", ) Patch: diff --git index .. --- +++ -290,10 +290,11 class BoundField(RenderableFieldMixin): # If a custom aria-describedby attribute is given and helptext is # used, the custom aria-described by is preserved so user can set the # desired order. if customariadescribedbyid := (\"aria-describedby\"): attrs[\"aria-describedby\"] = customariadescribedbyid elif and self.idforlabel: attrs[\"aria-describedby\"] = f\"{self.idforlabel}helptext\" if not (\"aria-describedby\"): if customariadescribedbyid := (\"aria-describedby\"): attrs[\"aria-describedby\"] = customariadescribedbyid elif and self.idforlabel: attrs[\"aria-describedby\"] = f\"{self.idforlabel}_helptext\" return attrs Happy to submit a PR if this is accepted."}
{"_id":"q-en-django__django-17066-7da4f6773e91f8aa14274fab670e0b7d9cf31b79f6b6b89bd1b18ae177089d33","text":"Migration serializer for sets results in non-deterministic order. Description We are using quite a lot of complex indextogether \/ uniquetogether constraints on our models, and the output in the generated migrations is flip-flopping all the time like follows, causing spurious diffs in our checkouts: migrations.AlterUniqueTogether( uniquetogether={(\"tenant\", \"dealer\"), (\"tenant\", \"order\")}, uniquetogether={(\"tenant\", \"order\"), (\"tenant\", \"dealer\")}, This is happening because these constraints are normalized to sets internally in the ModelState, which kind of makes sense, but unfortunately set iteration order (unlike dicts!) is unstable in Python 3 due to hash randomization. However, migrations serializer doesn't have any special facilities for ensuring stable output for sets and this is what causes annoying diffs for us all the time. I suggest to add a trivial serializer specifically for unordered sequences which ensures stable output no matter the iteration order. Stability can be achieved by sorting elements in the set by their string representation. This only affects the writer output, and doesn't interfere with the rest of Django in any way, so this change only improves developer experience, but has no effect on the performance and\/or reliability. I hope that even though it's apparently not a major problem for most users you would still accept the fix to ensure stable migration writer output for the rest of us."}
{"_id":"q-en-django__django-17084-252b16d000de33f237bc504d520d063a0ce20a3529c757e5594b40e5f0a90ed6","text":"Cannot use aggregate over window functions since 4.2 Description (last modified by younes-chaoui) After upgrading to Django 4.2, I encountered an exception when executing ORM queries that involve aggregates over Window functions. The specific error was aggregate function calls cannot contain window function calls Dependencies : psycopg2 version: 2.9.3 django version: 4.2.3 PostgreSQL version: 13.4 Example Code: queryset = queryset.annotate( cumulDJR=Coalesce(Window(Sum(\"DJR\"), orderby=F(\"date\").asc()), 0.0) ) aggregate = queryset.aggregate( DJRtotal=Sum(\"DJR\"), cumulDJRtotal=Sum(\"cumulDJR\") )"}
{"_id":"q-en-django__django-17087-7414d35cca55d350b36fd15491f5c6bdb9d305782db82cb1c25a1d75c504fa8a","text":"Class methods from nested classes cannot be used as Field.default. Description (last modified by Mariusz Felisiak) Given the following model: class Profile(models.Model): class Capability(models.TextChoices): BASIC = (\"BASIC\", \"Basic\") PROFESSIONAL = (\"PROFESSIONAL\", \"Professional\") def default(cls) -list[str]: return [cls.BASIC] capabilities = ArrayField( models.CharField(choices=Capability.choices, maxlength=30, blank=True), null=True, default=Capability.default ) The resulting migration contained the following: # ... migrations.AddField( modelname='profile', name='capabilities', (basefield=models.CharField(blank=True, choices=[('BASIC', 'Basic'), ('PROFESSIONAL', 'Professional')], maxlength=30), , null=True, size=None), ), # ... As you can see, migrations.AddField is passed as argument \"default\" a wrong value \"\", which leads to an error when trying to migrate. The right value should be \"\"."}
{"_id":"q-en-django__django-5158-9788ad95f98821cff49d5307a6f7c3b678906c670642566f9d8effc840a216cf","text":"Unknown django-admin command raises \"settings not configured\" when no project exists Description When you try to execute a command that does not exist outside a Django project (e.g. createproject instead of startproject...), Django raise a ImproperlyConfigured that does not fit with the real error: (testdjango):~\/testdjango$ django- not-a-real-command test Traceback (most recent call last): File \"\/root\/testdjango\/bin\/django-\", line 5, in <modulemanagement.executefromcommandline() File \"\/root\/testdjango\/local\/lib\/python2.7\/site-\", line 338, in executefromcommandline utility.execute() File \"\/root\/testdjango\/local\/lib\/python2.7\/site-\", line 330, in execute self.fetchcommand(subcommand).runfromargv() File \"\/root\/testdjango\/local\/lib\/python2.7\/site-\", line 182, in fetchcommand settings.INSTALLEDAPPS File \"\/root\/testdjango\/local\/lib\/python2.7\/site-\", line 48, in getattr (name) File \"\/root\/testdjango\/local\/lib\/python2.7\/site-\", line 42, in setup % (desc, ENVIRONMENTVARIABLE)) Requested setting INSTALLEDAPPS, but settings are not configured. You must either define the environment variable DJANGOSETTINGS_MODULE or call settings.configure() before accessing settings. The expected output is: (testdjango):~\/testdjango$ django- not-a-real-command test Unknown command: 'not-a-real-command' Type 'django- help' for usage."}
{"_id":"q-en-django__django-5470-635b3ca14017da261f6c219659607c31d57fa4db355f1b488d04320b810365f5","text":"Set script prefix in () to allow its usage outside of requests Description The script prefix for doesn't get set to anything when being called through , because of course it doesn't know what that value should be. This is a problem if you're rendering views (or otherwise reversing urls) from a command (as one of my sites does to send emails). This is solvable by calling setscriptprefix from , but that feels kind of dirty since it's then about to be rewritten in the WSGI handler. I don't know what a good solution to this would be. Perhaps it would be nice to be able to set a global default script path somewhere that would then get incorporated into the default values of things like LOGIN_URL. Maybe just a note in the documentation would be good. It took me a while to figure out, because I haven't been able to find anything else about this online. (I guess that non-\/ script paths are uncommon and reversing urls from is also uncommon, so both together are very uncommon.)"}
{"_id":"q-en-django__django-7188-e4a8e14fe3777e92a9e60645075d464f5ecaef77eb12a0aeb2c94400e9e6fd77","text":"Allow auth machinery to be used without installing auth app Description Django projects should be able to use the auth machinery (like and ) without having in INSTALLEDAPPS See ​!topic\/django-developers\/2DzLBbk8w-w This ticket is for resolving the current issue in 1.9. I also want to write a test or two to prevent this issue from happening in the future."}
{"_id":"q-en-django__django-7475-7d65f1d3dc0ac4bb2cdf6e2c50a9456c9f9b96eb9deca8560a9cb046c0353cf7","text":"diffsettings should allow to compare to another settings module (not only to default Django settings) Description Rationale: I use a hierarchy of settings modules, where imports imports I want to see what I customized, e.g. diffsettings --"}
{"_id":"q-en-django__django-7530-2b590994cda2560495709ff700ce39d1e8fa54001ced43e06175c2765321449b","text":"makemigrations router.allowmigrate() calls for consistency checks use incorrect (applabel, model) pairs Description As reported in ticket:#comment:14, I makemigrations incorrectly calls allow_migrate() for each app with all the models in the project rather than for each app with the app's models. It broke the router I use because it was passing invalid combinations for shards since not all shards have the same models. [​​ PR]"}
{"_id":"q-en-django__django-8119-a9b8faae79feb6ea07f429a62a9c8302fcf1c817d6f366f7024ad5819ed138dc","text":"Allow expressions in .filter() calls Description Expressions in filter calls will allow 3rd party apps to create query syntax extensions (for example .filter(F('some_field').lower() == 'anssi')) style). In addition, having the ability to use expressions everywhere unifies the ORM experience."}
{"_id":"q-en-django__django-8326-7b0c04466ded4adafab2bdf50f650b7496e49e06ebec9925a9b5e59a1f531412","text":"Add \"unified diff\" output mode to diffsettings management command Description Most people are used to unified diffs vs. the somewhat strange \"###\" formatting of diffsettings, so this adds an optional —output=unified."}
{"_id":"q-en-django__django-8630-d930ecd8d8bc370571ae1afbcc2c8dd424901587f8197b5f81c138bad95f3aff","text":"Add nextpage to LoginView Description LogoutView has a nextpage attribute used to override settings.LOGOUTREDIRECTURL. It would be nice if LoginView had the same mechanism."}
{"_id":"q-en-django__django-8961-5b71c8020e0654f9e7e626f13b3adc4e308a5482e88405e9c593072c2f8a3af7","text":"Improve program name for Django's command-line utility. Description By default, according to the ​documentation there are three ways to execute Django's command-line utility: $ django-admin <command[options] $ <command[options] $ python -m django <command[options] When executing the help command, however, the way these are rendered may be incorrect: $ django-admin help Type 'django-admin help <subcommand' for help on a specific subcommand. ... $ help Type ' help <subcommand' for help on a specific subcommand. ... $ python -m django help Type ' help <subcommand' for help on a specific subcommand. ... It should also be noted that will typically not work and is required: $ help bash: command not found On Windows it will work, but only if the file type is associated with the Python interpreter and not an editor. So, the issues here are: should be replaced by python -m django (or possibly use (sys.executable)) Should be replaced with on non-Windows platforms? What about Windows? What should happen with the option in the documentation?"}
{"_id":"q-en-django__django-9003-705afa8b261b9baec7acd4302e402ad6e84391730d406dd0e2aafa9f079c1396","text":"Non-deterministic crash in () Description Hi, There's a logical error somewhere in Query.combine or friends. I have a pretty complicated query with many-to-many self-joins and ORs, and after the upgrade to Django 1.9 and Python 3 combine() sometimes, but not always, crashes with the assertion error assert set(()).intersection(set(())) == set() Inspecting the changemap, it does indeed contain a circular reference (BTW shouldn't you use \"not\" to test for an empty set rather than comparing with set()?). At first, I didn't understand how it could crash sometimes and sometimes not - we're talking about the same query being executed through a view. But when I examine changemap, its content does indeed change from reload to reload - I suppose this may have something to do with the order of dicts\/sets the combine() algorithm is using. Here comes some code, I cut out a bunch of unused stuff, but otherwise it's the same: class Invoice(models.Model): customer = models.ForeignKey(Customer) datecreated = models.DateField(, dbindex=True) datesent = models.DateField(null=True, blank=True) datedue = models.DateField(null=True, blank=True) datepaid = models.DateField(null=True, blank=True) datecredited = models.DateField(null=True, blank=True) datecollect = models.DateField(null=True, blank=True) invoicetype = models.CharField(default=\"invoice\", maxlength=32) reminders = models.ManyToManyField(\"Invoice\", relatedname=\"remindedset\", blank=True) remindercounter = models.IntegerField(null=True, blank=True) That's the model, and in the view: import datetime from import Q date = () invoices = ( Q(datecreatedlte=date), Q(datepaidgt=date) | Q(datepaid=None), Q(datecreditedgt=date) | Q(datecredited=None), customer=1, ) filteredinvoices = () notdue = Q(dateduegte=date) | Q(datedue=None) notremindedyet = ~Q(remindersdatecreatedlte=date) notcollected = Q(datecollectgt=date) | Q(datecollect=None) filteredinvoices |= (notdue, notcollected, datesentlte=date, invoicetype=\"invoice\") filteredinvoices |= (notcollected, notremindedyet, datesentlte=date, dateduelt=date, invoicetype=\"invoice\") for r in [1, 2, 3]: qs = (notcollected, remindersdatecreatedlte=date, remindersremindercounter=r, invoicetype=\"invoice\") for i in range(r + 1, 3 + 1): qs = (~Q(remindersremindercounter=i) | Q(remindersremindercounter=i, remindersdatecreated_gt=date)) filteredinvoices |= qs I realize it's pretty complicated but I'm not sure what's essential and what's not. I hope someone knowledgeable of how combine() works can figure out why ordering in some cases matters."}
{"_id":"q-en-django__django-9296-bc898ee720c370448058a079fe2c5fd4a0e16fffd2d2c1c871ea1309df061036","text":"Paginator just implement the iter function Description (last modified by Alex Gaynor) Right now, when you want to iter into all the pages of a Paginator object you to use the pagerange function. It would be more logical and naturel to use the normal python of doing that by implementing the iter function like that: def iter(self): for pagenum in self.pagerange: yield (pagenum)"}
{"_id":"q-en-django__django-9703-7b9d6105705b03ccd86e5fae4307ebb36049ecf03c6762b93ed9242c058c8c97","text":"Allow management command invocation to suggest commands for mistyped commands Description Too often I can't remember the full mgmt command name, but can remember a part of it. A little enhancement would save me time. Attaching a screenshot and a patch. Thanks."}
{"_id":"q-en-django__django-9871-bb5d9029d608b354e9ae93565f0ad578ffd3082dc7a655d5c9a704237b1b2dd4","text":"Reorder management command arguments in --help output to prioritize command-specific arguments Description Currently if you run a custom management command with --help, you will get output that looks like: I have highlighted in yellow the useful information specific to the command that is not boilerplate. Notice that most of this yellow text is at the end of the output, with the boilerplate dominating what the user reads first. I propose reordering the options in the output so that the useful information is at the beginning rather than the end, so that it looks like the following: Discussion on django-developers: ​!topic\/django-developers\/PByZfN_IccE"}
{"_id":"q-en-matplotlib__matplotlib-13859-66dc95bc1d45b85f27c2aabcc36c499c130aac2d98ffe348a4d2033a1ae80cd3","text":"Zero-width figure crashes libpng Bug summary Zero-width figure crashes libpng. This happens when using ` or saving to png. Code for reproduction Actual outcome Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: ubuntu \/ conda Matplotlib version: 3.0.2, conda 3.0.2-py37h54297110 same at 3.0.3-py37h54297110 libpng 1.6.35-hbc830470, same with 1.6.36-hbc83047 Apparently I broke \"conda list\" on my machine so getting all the versions seems a bit tricky."}
{"_id":"q-en-matplotlib__matplotlib-13908-fc5ae2a7dd1618d8664bf32fb131318329a120848c47b071668a31310e85c038","text":"Minor ticklabels are missing at positions of major ticks. <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--Bug summary Minor ticklabels are missing at positions of major ticks. Code for reproduction Actual outcome The above code run with current master produces ! The minor ticklabels showing the hours are missing. Expected outcome The expected outcome would be the same as when running the code with matplotlib 3.0.2 or below: ! I would expect to see the hours throughout. Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: Win8 Matplotlib version: master Matplotlib backend (): any Python version: 3.6"}
{"_id":"q-en-matplotlib__matplotlib-13913-459066bea69d96525b471f49213de21cf3c9d6222cffa0d2e9cb0ca8e5c55e96","text":"returning erroneous value on strings of integers <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--Bug summary returns on strings containing integers, even though these are invalid colours when passed to or equivalent. Code for reproduction <!--A minimum code snippet required to reproduce the bug. Please make sure to minimize the number of dependencies required, and provide any necessary plotted data. Avoid using threads, as Matplotlib is (explicitly) not thread-safe.--Actual outcome <!--The output produced by the above code, which may be a screenshot, console output, etc.--Expected outcome <!--A description of the expected outcome from the code snippet--<!--If this used to work in an earlier version of Matplotlib, please note the version it used to work on--I would expect either the call to successfully recognise these as colours, or the calls to return . The latter is what I would expect of these two. Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: Arch Linux Matplotlib version: 3.0.2 Matplotlib backend (): Qt5Agg Python version: Python 3.7.1 Jupyter version (if applicable): N\/A Other libraries: 1.15.0 installed from . The example code runs without any errors on Windows, Python 3.6, 2.2.2. <!--Please tell us how you installed matplotlib and python e.g., from source, pip, conda--<!--If you installed from conda, please specify which channel you used if not the default--"}
{"_id":"q-en-matplotlib__matplotlib-13980-15b34c019ac652c8564f3120be4104335fc14b268282eb8c04da8890f0fe970f","text":"Non-sensical negative radial scale minimum autoset in polar plot When plotting a set of data on a polar plot, the default bottom ylimit might not be zero unexpectedly from the perspective of the user, resulting in confusion about the meaning of the plot, especially for a person (like me) unfamiliar with the concept of a polar plot where r=0 is not at the very center point of the plot. In a Jupyter Notebook (-0., 1.) ! I ran across this when plotting data and wondering if I had a bug in my analysis somewhere that was giving me a hole around the origin. It took me some time to figure out that the problem was simply in the axis scaling as I expected the version on the left (which seems sensible to me as the default) and not the version on the right which has a hole in the middle. Matplotlib version Operating system: Windows 10, also Ubuntu Linux Matplotlib version: 3.0.2 from pip Matplotlib backend (): inline Python version: 3.7, 3.6 Jupyter version (if applicable): JupyterLab 0.35.4"}
{"_id":"q-en-matplotlib__matplotlib-13983-0d9324d0a2b71665633bc6535e1093cf7802e09c9f491059d65ea8f5a767d3fc","text":"Remove()ing a shared axes prevents the remaining axes from using unit-provided formatters Consider One gets ! i.e. the call to prevented the axs[0] from acquiring the correct tick formatter and locator. Interestingly, using doesn't exhibit the same bug. Looks like the problem comes from where the call to (etc.), which basically call (to update the axis seen by the formatter) also make Matplotlib believe that we had a user-provided formatter (isDefaultmajloc = False, etc.) which should not be overridden by the unit framework. mpl master (ca. 3.0.2)"}
{"_id":"q-en-matplotlib__matplotlib-13984-8b716d1e1e40d7f09402bd42677918fb32ce6654d47645148b321a102bceecdc","text":"Tick mark color cannot be set on Axes3D As , the method does not change the color of tick marks on , only the color of tick labels. Several workarounds were proposed, and according to one comment, this used to work as expected in version 1.3.1. Here is code that tries to change the colors of all the axes but fails to get the tick marks:"}
{"_id":"q-en-matplotlib__matplotlib-13989-92d61fbb758c170a1b8e16b98050df611bebee493453ebe4d25042f5045a8497","text":"hist() no longer respects range=... when density=True <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--Bug summary <!--A short 1-2 sentences that succinctly describes the bug--Code for reproduction <!--A minimum code snippet required to reproduce the bug. Please make sure to minimize the number of dependencies required, and provide any necessary plotted data. Avoid using threads, as Matplotlib is (explicitly) not thread-safe.--Actual outcome <!--The output produced by the above code, which may be a screenshot, console output, etc.--Expected outcome Some array where the first value is 0 and the last one is 1. Note that this bug doesn't happen if density=False. Bisects to (). Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: linux Matplotlib version: master Matplotlib backend (): any Python version: 37 Jupyter version (if applicable): no Other libraries: numpy 1.16.2 <!--Please tell us how you installed matplotlib and python e.g., from source, pip, conda--<!--If you installed from conda, please specify which channel you used if not the default--"}
{"_id":"q-en-matplotlib__matplotlib-14043-33b5542fe357a2ca0b860d927d50a1bb91b1ab653e80a895e4c244897c16f59f","text":"bar plot yerr lines\/caps should respect zorder Bug summary Bar plot error bars break when zorder is greater than 1. Actual outcome ! Matplotlib version Operating system: Arch Linux Matplotlib version: 2.2.3 Matplotlib backend (): Python version: 3.6 Jupyter version (if applicable): 5.7.0 Conda default channel Possible related issue:"}
{"_id":"q-en-matplotlib__matplotlib-14471-c6fe8f61279bfc0b9aa9b8033f92f1cbaec84efaa75768a54ffcb8e0ae074293","text":"Existing FigureCanvasQT objects destroyed by call to Bug summary For a number of years, I have been maintaining an interactive application that embeds subclassed FigureCanvasQT objects within a PyQt application. Up until Matplotlib v3.0.3., it was possible to create standard Matplotlib PyQt figures, i.e., using within an embedded IPython shell, that coexist with the subclassed canvas objects. Now in Matplotlib v3.1.0, a call to destroys all the other canvas objects. Unfortunately, I cannot debug this within Visual Studio since I am currently unable to install Matplotlib from the source. By going through the code line by line, I can confirm that the windows are destroyed when calling , but not when calling , but I can't work out what triggers the destruction of the other windows. I presume the subclassed canvasses are not being registered somewhere, and assumes they are no longer active, but I am hoping someone will show me where to look. This may not be a Matplotlib bug, but it's certainly an unwelcome side effect in my application. Code for reproduction If you have installed () and launch it, you can reproduce this from within the embedded IPython shell with the following, which creates a new subclassed window and then attempts to open a regular window.: There are two figure numbers, because NeXpy automatically creates a window with a figure number of 1 when it is launched. Actual outcome A new window with an updated figure number is created but all other windows not created by are destroyed. Expected outcome In Matplotlib v3.0.3, a new window is created by the PyQt5 backend without destroying anything else. Matplotlib version Operating system: Mac OS v10.14.5 Matplotlib version: 3.1.0 Matplotlib backend: Qt5Agg Python version: 3.7.2 Jupyter version (if applicable): 1.0.0 Other libraries: <!--Please tell us how you installed matplotlib and python e.g., from source, pip, conda--<!--If you installed from conda, please specify which channel you used if not the default--"}
{"_id":"q-en-matplotlib__matplotlib-14623-781d5d1a6c625f1f5b3ea01ae589e52d63e24c02d744deb067004f7bda7b1e77","text":"Inverting an axis using its limits does not work for log scale Bug summary Starting in matplotlib 3.1.0 it is no longer possible to invert a log axis using its limits. Code for reproduction Actual outcome The yaxis is only inverted for the scale. ! ! Expected outcome I would expect the yaxis to be inverted for both the scale. Matplotlib version Operating system: Linux and MacOS Matplotlib version: 3.1.0 Python version: 3.7.3 Python and matplotlib have been installed using conda."}
{"_id":"q-en-matplotlib__matplotlib-17810-7d0aa26165e93078c4f28f464621f18efcc4fa814b7fc1250d0957f6f0cfe388","text":"and fig.savefig interfere with each other and raise StopIteration The following code snippet raises StopIteration: The same also happens if the figure is saved before the animation. The interesting thing is that the two parts of the script work fine when run alone, so there must be some interference between the two. The output produced by the above code: Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: Ubunut 18.04 Matplotlib version: 3.2.2 Matplotlib backend (): TkAgg Python version: 3.7.7 Jupyter version (if applicable): No Jupyter Other libraries: Matplotlib installed using pip inside a conda environment"}
{"_id":"q-en-matplotlib__matplotlib-18869-35f70f202ffc22474a63a4d81626376d14bbe495e6186217f4a79310a4896ac5","text":"Add easily comparable version info to toplevel <!-- Welcome! Thanks for thinking of a way to improve Matplotlib. Before creating a new feature request please search the issues for relevant feature requests. --Currently matplotlib only exposes . For quick version checks, exposing either a tuple (which can be compared with other tuples) or a instance (which can be properly compared with other strings) would be a small usability improvement. (In practice I guess boring string comparisons will work just fine until we hit mpl 3.10 or 4.10 which is unlikely to happen soon, but that feels quite dirty :)) <!-- Provide a clear and concise description of the problem this feature will solve. For example: I'm always frustrated when [...] because [...] I would like it if [...] happened when I [...] because [...] Here is a sample image of what I am asking for [...] --I guess I slightly prefer , but exposing just a tuple is much more common in other packages (and perhaps simpler to understand). The hardest(?) part is probably just bikeshedding this point :-) <!-- Provide a clear and concise description of a way to accomplish what you want. For example: Add an option so that when [...] [...] will happen --is a pretty common thing (citation needed). <!-- Add any other context or screenshots about the feature request here. You can also include links to examples of other programs that have something similar to your request. For example: Another project [...] solved this by [...] --"}
{"_id":"q-en-matplotlib__matplotlib-19763-2b780460dab3af5d5a1d6e31fee46e988d48e62b6480c50149fc5d2273502237","text":"Multicursor disappears when not moving on nbagg with useblit=False + burns CPU <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--Bug summary When on the nbagg backend if you stop moving the mouse the multicursor will disappear. The same example works fine on the qt backend. Additionally I noticed that when I add the multicursor my cpu usage jumps and the kernel busy indicator constantly flashes on and off. Showing the plot without the multicursor: ! and with the multicursor (just displaying, not interacting with the plot): ! That usage is pretty stable and my laptop's fan goes wild. The issue with the dissappearing was originally noticed by in Code for reproduction <!--A minimum code snippet required to reproduce the bug. Please make sure to minimize the number of dependencies required, and provide any necessary plotted data. Avoid using threads, as Matplotlib is (explicitly) not thread-safe.--Actual outcome ! and the high CPU usage Expected outcome Red line doesn't disappear + my CPU doesn't get crushed. <!--A description of the expected outcome from the code snippet--<!--If this used to work in an earlier version of Matplotlib, please note the version it used to work on--Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: Ubuntu Matplotlib version (): '3.3.4.post2456+gfd23bb238' Matplotlib backend (): nbagg Python version: '3.9.1 (default, Jan 26 2021, 01:34:10) \\n[GCC 9.3.0]' Jupyter version (if applicable): Notebook 6.2.0 - IPython 7.20.0 dev instlal of maptlotlib + conda-forge for the others"}
{"_id":"q-en-matplotlib__matplotlib-20374-6121124c04141ca2aeb4d6e624cabcb3970ca49d1aa4c538a22ae37395a84491","text":"plotdirective is confused by include directives, part 2 (context option) Whenever a file is included in via the RST directive, Sphinx rebuilds the page; but in plotdirective, is not considered 'out of date' (as judged by the function), because the modification time of was not changed. This discrepancy is part of the reason why exists; and while working on it (at PR ) I discovered also that figures with the option set will get confused. A specific example is shown below. It's quite hard to encounter in real life but it is directly relevant to matplotlib's tests as there is a very similar construct in plots 6-9 of matplotlib's . a.rst b.rst the four files above in a directory and into it. the docs the first time using . in any way. the docs again. The third plot in throws an error. The reason for this, as suggested above, is because of the function. When is invoked again, Sphinx decides that both and must be recompiled. Now: Plot 1 is not considered out of date, because the image file already exists and was not modified. So the code is never run and is never saved to the context. Plot 2 is there to ensure that figures are closed prior to Plot 3, so that Plot 3 never generates an image file. Plot 3 is considered out of date, because there is no image file that corresponds to it. Thus it is run again, and doesn't see in the context, hence the warning. There shouldn't be any errors. One easy way to accomplish this is to make sure that Sphinx re-runs all code snippets which are context-dependent, whenever a file is recompiled. That is, if a plot directive has :context: on, then the code should always be considered out of date regardless of the file modification times. This will lead to some excessive regeneration of plots whenever included files are modified. For example, in the above code, whenever is modified, Plots 1 and 3 will always be re-created, even if is untouched. But IMO this is more sensible behaviour than the current bug. It would also be in line with what happens if any part of is modified, including the text outside the plot directives: all the plots in would be re-created. This doesn't change the case where neither nor are modified, because in that case Sphinx will never attempt to recompile either file and plot_directive will never be called."}
{"_id":"q-en-matplotlib__matplotlib-20470-c32181aa49aa2e303822e3a13e21dff90c4b8e0f44617cf7308e8cf03847eebc","text":"Handle and label not created for Text with label Bug summary Text accepts a keyword argument but neither its handle nor its label is created and to the legend. Code for reproduction Actual outcome ! Expected outcome I expect a legend entry for the text. Matplotlib version Matplotlib version: 3.3.3"}
{"_id":"q-en-matplotlib__matplotlib-20488-b9f27522da0e673c6c2e0fe0c876f945470a5aaaca97af5078bfbc18894b681d","text":"testhugerangelog is failing... <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--is failing quite a few of the CI runs with a Value Error. I cannot reproduce locally, so I assume there was a numpy change somewhere... This test came in"}
{"_id":"q-en-matplotlib__matplotlib-20518-856102d83166451cdf174895201801d440c6d5826b59edadb770be417163aa5a","text":"Sketch params ignored when using PGF backend Bug summary Calls to are ignored by the PGF backend and do not have any influence in the resulting pgf or pdf file. Code for reproduction Actual outcome ! Expected outcome ! Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--* Matplotlib version: '3.0.2' I am working on a fix for this."}
{"_id":"q-en-matplotlib__matplotlib-20584-3f4d09ba1dfdfe4af0f1efa76853d0283b2efe02781bd3e9944534812cdc66c6","text":"setsegments(getsegments()) makes lines coarse After plotting with , I would like to retrieve the lines and manipulate them. Unfortunately, I noticed that the result is much coarser than without manipulation. In fact, a simple has this effect. I would have expected this does nothing at all. MWE: This is with mpl 3.4.2."}
{"_id":"q-en-matplotlib__matplotlib-20676-3d9cc3ea2736121580449b1a95880b62a1d684020ee317eb870f3178ae682010","text":"interactive SpanSelector incorrectly forces axes limits to include 0 <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--Bug summary Code for reproduction <!--A minimum code snippet required to reproduce the bug. Please make sure to minimize the number of dependencies required, and provide any necessary plotted data. Avoid using threads, as Matplotlib is (explicitly) not thread-safe.--Actual outcome The axes xlimits are expanded to include x=0. Expected outcome The axes xlimits remain at (10, 20) + margins, as was the case in Matplotlib 3.4 (with replaced by its old name ). attn Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: linux Matplotlib version (): master (3.5.0.dev1362+g57489bf19b) Matplotlib backend (): qt5agg Python version: 39 Jupyter version (if applicable): no Other libraries: <!--Please tell us how you installed matplotlib and python e.g., from source, pip, conda--<!--If you installed from conda, please specify which channel you used if not the default--"}
{"_id":"q-en-matplotlib__matplotlib-20761-f25b4a7fab791129de83e3b2bf923d6aaf10b0e45b65f1177b54bc8f19d0c6b1","text":"[Bug]: subfigure position shifts on y-axis when x kwarg to supxlabel Location of subfigure shifts lower on y-axis when 'x' kwarg is used for supxlabel for that subfigure. I've also posted to StackOverflow: Body of subfigure shifts downward (lower on y-axis) and covers supxlabel ! subfigure position doesn't change. supxlabel shifts to right. ! Windows 10 Pro 2 No response 5 16 No response conda conda-forge"}
{"_id":"q-en-matplotlib__matplotlib-20788-7f50a54d7afbaa238ddae0e0b07dd3b503f90a678c8b155974608061e4ab259c","text":"[Bug]: Colorbar creation from pcolormesh with cell specific alpha values When I try to take advantage of the new ability to set cell specific alpha values in pcolormesh - and then use the resulting QuadMesh object to create a colorbar it generates an error detailed below. A heat-map with a color bar beside it. Ubuntu inside singularity container 2 agg 5 not used none pip No response_"}
{"_id":"q-en-matplotlib__matplotlib-20805-71cd5e1de081de02cc7407ef406af981428d3669dbd6bfda64bd2b109a58b6c8","text":"[Bug]: spines and ticklabels Hello everyone, I am not sure if this is a bug or just how the spines function () is supposed to work. Basically, if I set the spines command \"after\" the tickparams specs, the \"labelrotation\" commands is not followed but everything else is. Any suggestions? Seems kind of weird that only some of the tickparams are executed and not others. Thank you! Key aspect of code below: [\"top\"].setposition((\"axes\", 1.05)) ax1.tickparams(axis='x', colors=\"green\", gridcolor='g',labelsize='small', labelrotation = 45) [\"top\"].setposition((\"axes\", 1.05)) See code above. I would expect the labels to be rotated regardless of where the spine command is placed. Windows or MacOs 2 No response 6 No response No response pip No response_"}
{"_id":"q-en-matplotlib__matplotlib-20816-834f68de47acaeeca142afc51e4ac45a2746b40c49a21a7b87151ef1dd8517f2","text":"Add ability to disable callbacks temporarily It may be useful to update some attributes on an object that has callbacks temporarily and we don't want those callback signals being processed during that time. I guess we could add a context manager on CallbackRegistry? Originally posted by in"}
{"_id":"q-en-matplotlib__matplotlib-20826-9e37510ee921dacf9436b50be9d80d5b53666cc39b01bf8880125e4aaa9c5d26","text":"() adds extra ticks, un-hides shared-axis tick labels Bug summary When using shared axes (e.g. from ), calling causes ticks and tick labels to be shown that should be hidden. The axes are still linked, though (e.g. adjusting the plotting range on one subplot adjusts the others as well). This is a behavior change between matplotlib 3.4.1 and 3.4.2. Code for reproduction This code produces different results with matplotlib 3.4.1 and 3.4.2: This example is of course silly, but I use the general pattern when making animations with FuncAnimation, where my plotting function is a complex module which doesn't facilitate blitting, so I clear and re-use the axes for each frame of the animation. Actual outcome This is the plot produced with matplotlib 3.4.2: ! The presence of tick labels that should be hidden by virtue of the shared axes is the clearest problem in this plot, but there are also ticks that appear along the top and right side of each subplot which are not present in the example below (and not part of the default plotting style, IIRC). The top and right-side ticks also appear when not using multiple subplots, so I think the shared-axis aspect reveals another symptom but is not a core part of this bug. If the call is removed, the plot produced with matplotlib 3.4.2 appears identical to the 3.4.1 plot below. Expected outcome This is the plot produced with matplotlib 3.4.1: ! Matplotlib version Operating system: Ubuntu 20.04 Matplotlib version (): 3.4.2 Matplotlib backend (): module:\/\/matplotlibinline.backendinline Python version: 3.8.10 Jupyter version (if applicable): jupyter core 4.7.1, jupyter lab 3.0.16 Other libraries: I've installed matplotlib (3.4.2-py38h578d9bd0) via conda from conda-forge"}
{"_id":"q-en-matplotlib__matplotlib-20859-5fb7e9601ae5a4999ae1c443b683453b4d20ab6e89398bcbe8f04bbd8901512c","text":"Adding a legend to a doesn't work <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--Bug summary <!--A short 1-2 sentences that succinctly describes the bug--Adding a legend to a doesn't work Code for reproduction <!--A minimum code snippet required to reproduce the bug. Please make sure to minimize the number of dependencies required, and provide any necessary plotted data. Avoid using threads, as Matplotlib is (explicitly) not thread-safe.--Actual outcome <!--The output produced by the above code, which may be a screenshot, console output, etc.--Expected outcome <!--A description of the expected outcome from the code snippet--<!--If this used to work in an earlier version of Matplotlib, please note the version it used to work on--I'd expect this to work and produce a legend. The example is of course a bit contrived but it would be useful to allow a legend per subfigure Changing L437 here to check against fixes it. I can make a PR at some point but wanted to flag the issue here in case anyone gets to it first. Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: macOS 11.4 Matplotlib version (): 3.4.2.post1350+gdba02be18e Matplotlib backend (): TkAgg Python version: Python 3.8.3 <!--Please tell us how you installed matplotlib and python e.g., from source, pip, conda--<!--If you installed from conda, please specify which channel you used if not the default--"}
{"_id":"q-en-matplotlib__matplotlib-21042-a30e8bf69447c244a7f85b4f66a380dfed1dcd261a29f0d9f5537f5b7285ca5b","text":"MatplotlibDeprecationWarning when updating rcparams Bug summary Updating RC params in code produces a deprecation warning. In order to control rcParams for an internal library I am reading in parameters and popping some that I don't want used. I am using a variation on the code below. I think I have traced it to the _validators dictionary in Code for reproduction This code reproduces the error. Actual outcome Expected outcome I would expect Matplotlib to not load deprecated rcParams Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: Windows Matplotlib version: 3.4.2 Matplotlib backend: 'Qt5Agg' Python version: 3.9.4 Jupyter version (if applicable): Other libraries: Setup a conda environment and just specified matplotlib from conda-forge"}
{"_id":"q-en-matplotlib__matplotlib-21318-f8ee10c2d167e5f4d81eb1b1340c2bccbdd3857417f8eef80114262d9abe4d61","text":"RuntimeError: adjustable='datalim' is not allowed when both axes are shared. <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--Bug summary Matplotlib 2.2.2 fails to plot the figure described below. The code works with Matplotlib 2.0.2. The issue seems to be the combination of `. Code for reproduction Actual outcome With Matplotlib 2.2.2, the figure is not drawn and the following exception is raised: Expected outcome Drawn figure, no error, as with Matplotlib 2.0.2 Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: Windows 7 64 bits Matplotlib version: 2.2.2 Matplotlib backend (): Qt5Agg Python version: 3.6 Jupyter version (if applicable): Other libraries: Anaconda 5.2 <!--Please tell us how you installed matplotlib and python e.g., from source, pip, conda--<!--If you installed from conda, please specify which channel you used if not the default--"}
{"_id":"q-en-matplotlib__matplotlib-21443-b20ecd522b5c381ae59cd6e50f9711ad99b9cd43d215749c11cbc0dc37af4227","text":"[Bug]: axes(position = [...]) behavior when setting axes position with the position data is not being incorporated. The two axes completely overlap ! Would expect two separate axes (these were created by adding and , which should not be necessary) ! Windows 2 Qt5Agg 8 No response No response conda No response"}
{"_id":"q-en-matplotlib__matplotlib-21481-02ad685470097af13b7a02dcc1b22a73a25ee680142342a4c9aee9e2aab9c5b9","text":"[Bug]: Subfigure breaks for some slices when using When creating a figure with you cannot use arbitrary gridspecs to create subfigures as it throws an error at some point ( I think once the layout manager actually takes effect?). This happened immediately on the call in and only on the call on No error. Should be the same as with Ubuntu 2428+g8daad3364a QtAgg 2 No response No response source No response_"}
{"_id":"q-en-matplotlib__matplotlib-21490-3cbd11d6622aa1e0ca0ec78b8a31bfb2728fa798c7aedcb725e7c57746927156","text":"[Bug]: Line2D should copy its inputs Currently, Line2D doesn't copy its inputs if they are already arrays. Most of the time, in-place modifications to the input arrays do not affect the draw line, because there is a cache that doesn't get invalidated, but in some circumstances, it is possible for these modifications to affect the drawn line. Instead, Line2D should just copy its inputs. This was rejected in on a memory-saving argument, but note that AxesImage (which would typically have much bigger (2D) inputs than Line2D (which has 1D inputs)) does a copy, which if anything is much worse memory-wise. (1) ! (2) (same as (1)) ! (3) (different, but only x got updated, not y) ! Modifying a posteriori should not affect the Line2D. Compare e.g. with AxesImage: Both images are identical. linux 5b1 mplcairo 39 No response No response source No response"}
{"_id":"q-en-matplotlib__matplotlib-21542-8a7d4022b70f1e8815d1334fc8bdc6b908504827ad3d88b56cfad81ff75888b5","text":"[ENH]: use new style format strings for colorbar ticks At the moment, the default format strings in colorbar are old style ones, as in their init there is: which is a different convention from the one of a normal axis, which was introduced in . As in we pass the colorbar's formatter to the long axis, the statement above may be removed to keep the default logic only in . Right now one can pass a callable directly (although that's not documented), and the default behaviour from is triggered. However, I guess making this change for format strings would imply a deprecation cycle, as it breaks current behaviour. Another option would be to check in for what kind of string format we've been passed (unsure how, although there must be way). No response_"}
{"_id":"q-en-matplotlib__matplotlib-21550-4a180699baed5363f2bb63238a5dc31e805b333d819770b2aa4759b0a83be690","text":"[Bug]: this example shows ok on matplotlib-3.4.3, but not in matplotlib-3.5.0 master of october 30th the display is not working well if swaping matplotlib-3.4.3 with matplotlib-3.5.0.dev2445+gb09aad279b, all the rest being strictly equal. it was also bad with rc1, so I tested with last master, thanks to the artefact generation on jupyterlab ` ! ! Windows 10 matplotlib-3.5.0.dev2445+gb09aad279b-cp39-cp39- module:\/\/matplotlibinline.backendinline Python 3.9.7 1 wheels from cgohlke when binaries, except the matplotlib-master from pip list: <detailsPackage Version --------------------------------- ------------------ adodbapi 2.6.1.3 affine 2.3.0 aiofiles 0.6.0 aiohttp 3.7.4.post0 aiosqlite 0.17.0 alabaster 0.7.12 algopy 0.5.7 altair 4.1.0 altair-data-server 0.4.1 altair-transform 0.2.0 altair-widgets 0.2.2 altgraph 0.17.2 amply 0.1.4 aniso8601 7.0.0 ansiwrap 0.8.4 anyio 3.3.4 appdirs 1.4.4 argon2-cffi 21.1.0 arrow 1.2.1 asciitree 0.3.3 asgi-csrf 0.9 asgiref 3.4.1 asn1crypto 1.4.0 asteval 0.9.25 astor 0.8.1 astroid 2.6.6 astroML 1.0.1 astropy 4.3.1 async-generator 1.10 async-timeout 3.0.1 atomicwrites 1.4.0 attrs 21.2.0 autopep8 1.5.7 Babel 2.9.1 backcall 0.2.0 backports-abc 0.5 1.1.0 baresql 0.7.6 base58 2.0.0 bcrypt 3.2.0 beautifulsoup4 4.10.0 binaryornot 0.4.4 black 21.9b0 bleach 4.1.0 blinker 1.4 blis 0.7.5 blosc 1.10.6 bloscpack 0.16.0 bokeh 2.4.1 botorch 0.4.0 Bottleneck 1.3.2 bqplot 0.12.31 branca 0.4.2 brewer2mpl 1.4.1 Brotli 1.0.9 cachelib 0.3.0 cachetools 4.2.4 Cartopy 0.20.1 catalogue 2.0.6 certifi 2021.10.8 cffi 1.15.0 cftime 1.5.1.1 chardet 4.0.0 charset-normalizer 2.0.7 click 7.1.2 click-default-group 1.2.2 click-plugins 1.1.1 cligj 0.7.2 cloudpickle 2.0.0 clrmagic 0.0.1a2 colorama 0.4.4 colorcet 2.0.6 cookiecutter 1.7.3 coverage 6.0.2 cramjam 2.4.0 cryptography 35.0.0 csvs-to-sqlite 1.2 cvxopt 1.2.7 cvxpy 1.1.15 cx-Freeze 6.5.3 cycler 0.11.0 cymem 2.0.6 Cython 0.29.24 cytoolz 0.11.0 dash 2.0.0 dash-core-components 2.0.0 dash-html-components 2.0.0 dash-table 5.0.0 dask 2021.10.0 dask-glm 0.2.0 dask-image 0.6.0 dask-labextension 5.1.0 dask-ml 2021.10.17 dask-searchcv 0.2.0 databases 0.4.1 datasette 0.59.1 datasette-graphql 1.5 datashader 0.13.0 datashape 0.5.2 dateparser 1.1.0 dateutils 0.6.12 0.5.4b1 debugpy 1.5.1 decorator 4.4.2 defusedxml 0.7.1 Deprecated 1.2.13 deprecation 2.1.0 descartes 1.1.0 diff-match-patch distlib 0.3.3 distributed 2021.10.0 docopt 0.6.2 docrepr 0.1.1 docutils 0.17.1 ecos 2.0.7.post1 emcee 3.1.1 entrypoints 0.3 et-xmlfile 1.1.0 fast-histogram 0.10 fastai 2.5.3 fastapi 0.70.0 fastcore 1.3.26 fastdownload 0.0.5 fasteners 0.16.3 fastparquet 0.7.1 fastprogress 1.0.0 feather-format 0.4.1 filelock 3.3.2 Fiona 1.8.20 flake8 3.9.2 Flask 2.0.2 flask-accepts 0.18.4 Flask-Compress 1.10.1 Flask-Cors 3.0.10 Flask-Mail 0.9.1 flask-restx 0.5.1 Flask-Session 0.4.0 Flask-SQLAlchemy 2.5.1 flaskerize 0.14.0 flatbuffers 2.0 flit 3.4.0 flitcore 3.4.0 folium 0.12.1 fonttools 4.27.1 formlayout 1.2.1a1 fs 2.4.13 fsspec 2021.10.1 future 0.18.2 fuzzywuzzy 0.18.0 GDAL 3.3.3 geographiclib 1.52 geopandas 0.10.2 geopy 2.2.0 geoviews 1.9.2 gitdb 4.0.9 GitPython 3.1.24 gmpy2 2.0.8 gpytorch 1.5.1 graphene 2.1.9 graphql-core 2.3.1 graphql-relay 2.0.1 great-expectations 0.13.36 greenlet 1.1.2 guidata 1.8.1a0 guiqwt 3.0.7 h11 0.12.0 h2 4.1.0 h5py 3.5.0 HeapDict 1.0.1 holoviews 1.14.6 hpack 4.0.0 html5lib 1.1 httpcore 0.13.7 httpie 2.6.0 httpx 0.20.0 hupper 1.10.3 husl 4.0.3 hvplot 0.7.3 Hypercorn 0.11.2 hyperframe 6.0.1 hypothesis 6.24.0 ibis-framework 1.4.0 idlex 1.18 idna 3.1 imageio 2.10.1 imageio-ffmpeg 0.4.2 imagesize 1.2.0 imbalanced-learn 0.8.1 importlib-metadata 4.8.0 inflection 0.5.1 iniconfig 1.1.1 intake 0.6.2 intervaltree 3.0.2 ipycanvas 0.9.1 ipykernel 6.4.2 ipyleaflet 0.14.0 ipympl 0.8.2 ipython 7.29.0 ipython-genutils 0.2.0 ipython-sql 0.4.1b1 ipywidgets 7.6.5 isort 5.9.3 itsdangerous 2.0.1 janus 0.6.2 jedi 0.18.0 Jinja2 3.0.2 jinja2-time 0.2.0 joblib 1.1.0 json5 0.9.6 jsonpatch 1.32 jsonpointer 2.1 jsonschema 4.1.2 julia 0.5.7 jupyter 1.0.0 jupyter-bokeh 3.0.4 jupyter-client 6.2.0 jupyter-console 6.4.0 jupyter-core 4.9.1 jupyter-lsp 1.5.0 jupyter-packaging 0.11.0 jupyter-server 1.11.1 jupyter-server-mathjax 0.2.3 jupyter-server-proxy 3.1.0 jupyter-sphinx 0.3.2 jupyterlab 3.2.1 jupyterlab-git 0.33.0 jupyterlab-launcher 0.13.1 jupyterlab-lsp 3.9.1 jupyterlab-pygments 0.1.2 jupyterlab-server 2.8.2 jupyterlab-widgets 1.0.2Note: you may need to restart the kernel to use updated packages. keyring 23.2.1 kiwisolver 1.3.2 lazy-object-proxy 1.6.0 llvmlite 0.37.0 lmfit 1.0.3 locket 0.2.1 loky 3.0.0 lxml 4.6.3 lz4 3.1.3 Markdown 3.3.4 MarkupSafe 2.0.1 marshmallow 3.12.1 matplotlib 3.4.3 matplotlib-inline 0.1.3 maturin 0.11.5 mccabe 0.6.1 mercantile 1.2.1 mergedeep 1.3.4 metakernel 0.27.5 mistune 0.8.4 mizani 0.7.3 mkl-service 2.4.0 mlxtend 0.18.0 moviepy 1.0.3 mpl-scatter-density 0.7 mpld3 0.5.5 mpldatacursor 0.7.1 mpmath 1.2.1 msgpack 1.0.2 msvc-runtime 14.29. multidict 5.2.0 multipledispatch 0.6.0 munch 2.5.0 murmurhash 1.0.6 mypy 0.910 mypy-extensions 0.4.3 mysql-connector-python 8.0.26 nbclassic 0.3.4 nbclient 0.5.4 nbconvert 6.2.0 nbconvertreportlab 0.2 nbdime 3.1.1 nbformat 5.1.3 nbval 0.9.6 nest-asyncio 1.5.1 netCDF4 1.5.8 networkx 2.6.3 NLopt 2.7.0 nltk 3.6.5 notebook 6.4.5 numba 0.54.1 numcodecs 0.9.1 numdifftools 0.9.40 numexpr 2.7.3 numpy 1.20.3+mkl numpydoc 1.1.0 oct2py 5.2.0 octave-kernel 0.32.0 onnxruntime 1.9.0 openpyxl 3.0.9 orjson 3.6.4 osqp 0.6.2.post0 outcome 1.1.0 packaging 21.2 palettable 3.3.0 pandas 1.3.4 pandas-datareader 0.10.0 pandocfilters 1.5.0 panel 0.12.4 papermill 2.3.3 param 1.12.0 parambokeh 0.2.3 paramiko 2.8.0 paramnb 2.0.4 parso 0.8.2 partd 1.2.0 pathspec 0.9.0 pathy 0.6.1 patsy 0.5.2 pdfrw 0.4 pdvega 0.2.1.dev0 pefile 2021.9.3 pep8 1.7.1 pexpect 4.8.0 pg8000 1.21.1 pickleshare 0.7.5 Pillow 8.4.0 PIMS 0.5 Pint 0.18 pip 21.3.1 pipdeptree 2.2.0 pkginfo 1.7.1 platformdirs 2.4.0 plotly 5.3.1 plotnine 0.8.0 pluggy 1.0.0 ply 3.11 portpicker 1.4.0 poyo 0.5.0 ppci 0.5.8 preshed 3.0.6 prettytable 2.2.1 priority 2.0.0 proglog 0.1.9 prometheus-client 0.12.0 promise 2.3 prompt-toolkit 3.0.21 protobuf 4.0.0rc1 psutil 5.8.0 ptpython 3.0.20 ptyprocess 0.7.0 PuLP 2.3 py 1.10.0 py-lru-cache 0.1.4 pyaml 20.4.0 pyarrow 6.0.0 PyAudio 0.2.11 pybars3 0.9.7 pybind11 2.8.1 pycodestyle 2.7.0 pycosat 0.6.3 pycparser 2.20 pyct 0.4.8 pydantic 1.8.2 pydeck 0.7.1 pydocstyle 6.1.1 pyepsg 0.4.0 pyerfa 2.0.0 pyflakes 2.3.1 pyflux 0.4.17 pygame 2.0.3 pygbm 0.1.0 Pygments 2.10.0 pyhdf 0.10.3 pyinstaller 4.5.1 pyinstaller-hooks-contrib 2021.3 pylint 2.9.6 pyls-spyder 0.4.0 pymc 2.3.8 PyMeta3 0.5.1 pymongo 3.12.1 PyNaCl 1.4.0 pynndescent 0.5.5 pyodbc 4.0.32 PyOpenGL 3.1.5 pypandoc 1.5 pyparsing 2.4.7 pyproj 3.2.1 PyQt5 5.15.1 PyQt5-sip 12.8.1 pyqt5-tools 5.15.1.1.7.5.post3 pyqtgraph 0.12.2 PyQtWebEngine 5.15.1 pyrsistent 0.18.0 pyserial 3.5 pyshp 2.1.3 PySocks 1.7.1 pystache 0.5.4 pytest 6.2.5 python-baseconv 1.2.2 python-dateutil 2.8.2 python-dotenv 0.19.1 python-hdf4 0.10.0+dummy python-Levenshtein 0.12.2 python-lsp-black 1.0.0 python-lsp-jsonrpc 1.0.0 python-lsp-server 1.2.4 python-multipart 0.0.5 python-picard 0.7 python-slugify 5.0.2 python-snappy 0.6.0 pythonnet 2.5.2 PythonQwt 0.9.2 pytz 2021.3 pyviz-comms 2.1.0 PyWavelets 1.1.1 pywin32 302 pywin32-ctypes 0.2.0 pywinpty 1.1.5 pywinusb 0.4.2 PyYAML 6.0 pyzmq 22.3.0 pyzo 4.11.3a1 QDarkStyle 3.0.2 qdldl 0.1.5.post0 qpsolvers 1.7.0 qstylizer 0.2.1 QtAwesome 1.0.3 qtconsole 5.1.1 QtPy 2.0.0.dev0 quadprog 0.1.8 quantecon 0.5.1 Quart 0.15.1 rasterio 1.2.10 readme-renderer 30.0 redis 3.5.3 regex 2021.10.23 reportlab 3.6.2 requests 2.26.0 requests-toolbelt 0.9.1 requests-unixsocket 0.2.0 rfc3986 1.5.0 rise 5.7.1 rope 0.21.0 rpy2 3.4.0.dev0 Rtree 0.9.7 0.17.15 0.2.6 Rx 1.6.1 scikit-fuzzy 0.4.1 scikit-image 0.18.3 scikit-learn 1.0.1 scikit-optimize 0.9.0 scilab2py 0.6.2 scipy 1.7.1 scramp 1.4.1 scs 2.1.4 seaborn 0.11.2 Send2Trash 1.8.0 setuptools 58.3.0 setuptools-scm 6.3.2 shap 0.39.0 Shapely 1.8.0 simpervisor 0.4 simplegeneric 0.8.1 simplejson 3.17.5 simpy 4.0.1 six 1.16.0 sklearn-contrib-lightning 0.6.1 slicer 0.0.7 slicerator 1.0.0 smart-open 5.2.1 smmap 5.0.0 snakeviz 2.1.0 sniffio 1.2.0 snowballstemmer 2.1.0 snuggs 1.4.7 sortedcontainers 2.4.0 sounddevice 0.4.3 soupsieve 2.2.1 spacy 3.1.3 spacy-legacy 3.0.8 Sphinx 4.2.0 sphinx-rtd-theme 1.0.0 sphinxcontrib-applehelp 1.0.2 sphinxcontrib-devhelp 1.0.2 sphinxcontrib-htmlhelp 2.0.0 sphinxcontrib-jsmath 1.0.1 sphinxcontrib-qthelp 1.0.3 sphinxcontrib-serializinghtml 1.1.5 spyder 5.1.5 spyder-kernels 2.1.3 SQLAlchemy 1.4.26 sqlite-bro 0.12.1 sqlite-fts4 1.0.1 sqlite-utils 3.17.1 sqlparse 0.4.2 srsly 2.4.2 starlette 0.16.0 statsmodels 0.13.0 streamlit 1.1.0 streamz 0.6.3 supersmoother 0.4 sympy 1.9 tables 3.6.1 tabulate 0.8.9 tblib 1.7.0 tenacity 8.0.1 termcolor 1.1.0 terminado 0.12.1 testpath 0.5.0 text-unidecode 1.3 textdistance 4.2.2 textwrap3 0.9.2 thinc 8.0.12 threadpoolctl 3.0.0 three-merge 0.1.1 thrift 0.15.0 tifffile 2021.10.12 tinycss2 1.1.0 toml 0.10.2 tomli 1.2.2 tomliw 0.4.0 tomlkit 0.7.2 toolz 0.11.1 torch 1.10.0 torchaudio 0.10.0 torchvision 0.11.1 tornado 6.1 tqdm 4.62.3 traitlets 5.1.1 traittypes 0.2.1 tranquilizer 0.5.1a1 trio 0.19.0 trio-asyncio 0.12.0 twine 3.4.2 twitter 1.19.2 typer 0.4.0 typing-extensions 3.10.0.2 tzlocal 2.1 ujson 4.2.0 umap-learn 0.5.1 uncertainties 3.1.6 urllib3 1.26.7 uvicorn 0.15.0 validators 0.18.2 vega 3.5.0 vega-datasets 0.9.0 virtualenv 20.9.0 ViTables 3.0.2 voila 0.2.16 voila-gridstack 0.2.0 wasabi 0.8.2 wasmer 1.0.0 wasmercompilercranelift 1.0.0 wasmercompilersinglepass 1.0.0 watchdog 2.1.5 wcwidth 0.2.5 webencodings 0.5.1 websocket-client 1.2.1 Werkzeug 2.0.2 wheel 0.37.0 widgetsnbextension 3.5.2 winpython 4.6. winrt 1.0..1 wordcloud 1.8.1 wrapt 1.12.1 wsproto 1.0.0 xarray 0.19.0 XlsxWriter 3.0.1 xlwings 0.24.7 yapf 0.31.0 yarl 1.7.0 zarr 2.10.2 zict 2.0.0 zipp 3.6.0 zstandard 0.16.0 <\/detailspip No response_"}
{"_id":"q-en-matplotlib__matplotlib-21559-361cfe2c44bc38ff3df20060df9b45cb5149be5da0cf8798394bf0702a9a2cad","text":"[Bug]: eventplot cannot handle multiple datetime-based series I am having the exact same issue as in (which apparently was resolved at some point, but it's not working for me) with the latest version of matplotlib as passing a list of s to is not working. TypeError Traceback (most recent call last) ~\\AppData\\Local\\ in <module4 ts1 = [(2021, 1, 15), (2021, 1, 16), (2021, 1, 18), (2021, 1, 22), (2021, 1, 26), (2021, 1, 28), (2021, 1, 30), (2021, 2, 1), (2021, 2, 3), (2021, 2, 7), (2021, 2, 8), (2021, 2, 11), (2021, 2, 15)] 5 ts2 = [(2021, 1, 1), (2021, 1, 2), (2021, 1, 3), (2021, 1, 4), (2021, 1, 5), (2021, 1, 6), (2021, 1, 7), (2021, 1, 8), (2021, 1, 9), (2021, 1, 10), (2021, 1, 11), (2021, 1, 12), (2021, 1, 13), (2021, 1, 14), (2021, 1, 15), (2021, 1, 16), (2021, 1, 17), (2021, 1, 18), (2021, 1, 19), (2021, 1, 20), (2021, 1, 25), (2021, 1, 26), (2021, 1, 27), (2021, 1, 28), (2021, 1, 29), (2021, 1, 30), (2021, 1, 31), (2021, 2, 12)] ----6 plt.eventplot([ts1, ts2], colors=\"black\") c:\\Users\\User\\Documents\\venv\\lib\\site-packages\\matplotlib\\ in eventplot(positions, orientation, lineoffsets, linelengths, linewidths, colors, linestyles, data, kwargs) 2784 linelengths=1, linewidths=None, colors=None, 2785 linestyles='solid', , data=None, kwargs): -2786 return gca().eventplot( 2787 positions, orientation=orientation, lineoffsets=lineoffsets, 2788 linelengths=linelengths, linewidths=linewidths, colors=colors, c:\\Users\\User\\Documents\\venv\\lib\\site-packages\\matplotlib\\ in inner(ax, data, args, kwargs) 1359 def inner(ax, args, data=None, kwargs): 1360 if data is None: -1361 return func(ax, map(sanitizesequence, args), kwargs) 1362 1363 bound = (ax, *args, kwargs) c:\\Users\\User\\Documents\\venv\\lib\\site-packages\\matplotlib\\axes\\ in eventplot(self, positions, orientation, lineoffsets, linelengths, linewidths, colors, linestyles, kwargs) 1321 zip(positions, lineoffsets, linelengths, linewidths, 1322 colors, linestyles): -1323 coll = mcoll.EventCollection(position, 1324 orientation=orientation, 1325 lineoffset=lineoffset, c:\\Users\\User\\Documents\\venv\\lib\\site-packages\\matplotlib\\ in init(self, positions, orientation, lineoffset, linelength, linewidth, color, linestyle, antialiased, kwargs) 1619 self.lineoffset = lineoffset 1620 self.setorientation(orientation) -1621 self.setpositions(positions) 1622 1623 def getpositions(self): c:\\Users\\User\\Documents\\venv\\lib\\site-packages\\matplotlib\\ in setpositions(self, positions) 1638 posidx = 0 if self.ishorizontal() else 1 1639 segments = ((len(positions), 2, 2)) -1640 segments[:, :, posidx] = (positions)[:, None] 1641 segments[:, 0, 1 - posidx] = lineoffset + linelength \/ 2 1642 segments[:, 1, 1 - posidx] = lineoffset - linelength \/ 2 TypeError: float() argument must be a string or a number, not '' An eventplot with one time series per \"row\" in the y-axis. Windows 3 No response 7 No response No response pip No response"}
{"_id":"q-en-matplotlib__matplotlib-21568-f9f0b1ebf76d88a4f8ef861820bcc434846b1a2bb512f2d562422e47d31af9ab","text":"[Bug]: Datetime axis with usetex is unclear The spacing for a datetime axis when using is unclear in matplotlib version 3.4 when comparing it to 3.3. Example of how it look in 3.3.4: ! Example of how it look in 3.4.3: ! The ideal case would be to have the spacing from version 3.3 in a tex format. Windows 3 No response No response No response No response conda conda-forge"}
{"_id":"q-en-matplotlib__matplotlib-21570-558f14c421a420e360cfd5a178b337734e8286500e5d14f5bf33e200a9d20349","text":"[Bug]: upon deepcopy of a object A deepcopy of a object results in . This issue surfaced in a bug report of : Nevertheless, the current issue is unrelated to . I'm just mentioning this to illustrate a relevant use case leading to this problem. Disclaimer: this is taken from a message of on Either a deepcopy of the figure or a meaningful error message explaining that this operation is not supported. Fedora Linux 3 Qt5Agg 7 6, (not installed) None Linux package manager (Debian\/Fedora\/etc.) No response"}
{"_id":"q-en-matplotlib__matplotlib-21617-083275b807c509c446ac611729a66ba1b4c59e2e7adb2d712e337af50f6c6a78","text":"[Bug]: Text sometimes is missing when figure saved to EPS I'm using cartopy to render data in polar projection. After an update I noticed that labels and title is not anymore rendered in saved eps file, though it was rendered in Jupyter notebook. I managed to simplify the code and found that matplotlib=3.3.4 does not suffer from the problem, while matplotlib=3.4.3 does. The testing environment was obtained by calls Might be related to An eps file with no message. (matplotlib v3.4.3) ! An eps file with message. (matplotlib v3.3.4) ! Debian GNU\/Linux bookworm, Linux Mint 20.2 Uma 3 Qt5Agg Python 3.7.12 not used cartopy=0.18.0 conda conda-forge"}
{"_id":"q-en-matplotlib__matplotlib-22711-6fbedab3dd1648e406ad264687deb37d0fb33177c04b7384cce088faabbc763a","text":"[Bug]: cannot give init value for RangeSlider widget I think should be commented in \/matplotlib\/widgets. py\", line 915, in setval as it prevents to initialized value for RangeSlider range slider with user initial values error can be removed by commenting this line OSX 1 No response 8 No response_ pip"}
{"_id":"q-en-matplotlib__matplotlib-22719-5cf2b71d4c1edbc048a63774ea35a45f043bbfef81d684019b50ccae1a06c176","text":"[Bug]: Confusing deprecation warning when empty data passed to axis with category units I'm seeing a when using calling axes methods on empty data structures for axes that are using string unit converters. I think this is either a false alarm or a non-actionable warning. ([], []) Here's the full traceback if I force the warning to be an error: <details<\/detailsAdditionally, the problem is not solved by doing what the warning message suggests: <details<\/detailsI would expect this to either (1) continue producing artists with no data, or (2) more accurately describe what the problem is and how to avoid it. Looking at the traceback, it seems like it's catching exceptions too broadly and issuing a generic warning. If passing empty data structures through unit converters is now deprecated, it should be possible to detect that specific case. But I can't quite follow the API change note here: The traceback appears inconsistent with the claim that is never called with a unit-less value and that provides an alternate, supported interface: So it feels like maybe whatever is changing behind the scenes failed to anticipate the \"empty data\" edge case? 1"}
{"_id":"q-en-matplotlib__matplotlib-22734-a3a9454cdcbdeb55c7bdec7f85b3085c8363553e455840acf3a36cc0aa7b1993","text":"[Bug]: tripcolor ignores clim shows that tripcolor ignores clim. Colorbar\/colormapping goes from nearly 0 to nearly 1. Colorbar\/colormapping goes from 0 to 0.5. This is the same issue as \/: kwargs should be handled a bit further down in the implementation of tripcolor() (just before calling scalenorm). arch linux 1920+gdfd83c2c5d mplcairo 310 ENOSUCHLIB git checkout"}
{"_id":"q-en-matplotlib__matplotlib-22767-d9e1d21af900f9632f0d32e8ae2924ab2b6bebfdc6a474bf091ef480cfe886a6","text":"[Bug]: Issue with findnearestcontour in The default for the indices keyword in the findnearestcontour function within is defined incorrectly in the code. Line 1377 should be \"indices = range(len())\" instead of \"indices = range(len()).\" Traceback (most recent call last): File \"<stdin\", line 1, in <moduleFile \"\/opt\/local\/Library\/Frameworks\/Python.framework\/Versions\/3.8\/lib\/python3.8\/site-\", line 1388, in findnearestcontour con = self.collections[icon] IndexError: list index out of range (0, 0, 397, 23., 14., 49.) No response Mac OS 12.3.1 1 TkAgg 12 No response pip"}
{"_id":"q-en-matplotlib__matplotlib-22815-712d00ed753116b05202358c6783978394cce9e17c5082ffaa676f059fe072fa","text":"[Bug]: makenormfromscale should create picklable classes even when used in-line. The new helper dynamically generates a norm class from a scale class. Currently, in the codebase, it is only used as a decorator to create \"toplevel\" classes (e.g., it is used to generate LogNorm from LogScale, etc.), but it can also be used within other functions to dynamically generate a norm class based on a user-given arbitrary scale (see for an example of application, which is however not necessary to understand for what follows). In the latter case, the dynamically generate class is currently not picklable (because pickling of classes relies on the existence of global names, see e.g. ). It would be generally useful to get rid of this restriction, which can be done by implementing the ; there's already other examples in the codebase of dynamically generated classes that use the same mechanism). I'm tagging this as \"good first issue\" because there's no API design and the eng goal is clear, but medium (perhaps hard) difficulty because it requires somewhat sophisticated understanding of the details of the pickling process. (Note the additional confusion here: there's two classes that are both named -- the original one and the dynamically generated one -- but they are different.) A correct round-trippable pickle. No response master (unreleased, pre 3.5) No response No response No response No response source No response"}
{"_id":"q-en-matplotlib__matplotlib-22835-a6054f96f8ae9d82fb7a5863d1daaffa96e3f34cbe9e7fe9632375a9a141bc17","text":"[Bug]: scalar mappable formatcursordata crashes on BoundarNorm In 3.5.0 if you do: and mouse over the image, it crashes with and interaction stops. Not sure if we should have a special check here, a try-except, or actually just make BoundaryNorm approximately invertible. main 3.5.0 [Bug]: scalar mappable formatcursordata crashes on BoundarNorm In 3.5.0 if you do: and mouse over the image, it crashes with and interaction stops. Not sure if we should have a special check here, a try-except, or actually just make BoundaryNorm approximately invertible. main 3.5.0"}
{"_id":"q-en-matplotlib__matplotlib-22865-74c16253a7bf37dd4110c5f93b1caa9a8e65d96df1c273db10865053595a9a92","text":"[Bug]: Colorbar with drawedges=True and extend='both' does not draw edges at extremities When creating a matplotlib colorbar, it is possible to set drawedges to True which separates the colors of the colorbar with black lines. However, when the colorbar is extended using extend='both', the black lines at the extremities do not show up. ! ! No response No response 1 No response No response No response No response"}
{"_id":"q-en-matplotlib__matplotlib-22871-039967f6fe56e64b1025dcc5cd9e17a68b4782dbbd3c3ea5ceafd2ab80f7d293","text":"[Bug]: ConciseDateFormatter not showing year anywhere when plotting <12 months When I plot < 1 year and January is not included in the x-axis, the year doesn't show up anywhere. This bug is different from bug (fixed in ). ! I expect the year \"2021\" to show in the offset, to the right of the x-axis I'm using Spyder IDE, v5.1.5 Windows 10 3 Qt5Agg 1 No response conda"}
{"_id":"q-en-matplotlib__matplotlib-22883-3c831c9505a88b40173817221093370800a2f6b1b3e27e1482ca025b869625d7","text":"Missing prefix on f-strings Some strings looks like they're meant to be f-strings but are missing the prefix meaning variable interpolation won't happen. I found this issue automatically. I'm a bot. Beep Boop . See other issues I found in your repo"}
{"_id":"q-en-matplotlib__matplotlib-22926-6fbedab3dd1648e406ad264687deb37d0fb33177c04b7384cce088faabbc763a","text":"[Bug]: cannot give init value for RangeSlider widget I think should be commented in \/matplotlib\/widgets. py\", line 915, in setval as it prevents to initialized value for RangeSlider range slider with user initial values error can be removed by commenting this line OSX 1 No response 8 No response_ pip"}
{"_id":"q-en-matplotlib__matplotlib-22929-b8701238fffde003dab09224f66305040e451218361c974420f12a4785f4e6fb","text":"[Bug]: barlabel fails with nan errorbar values appears not to be robust to bars with missing (nan) values when also including error values. This issue is similar to , but occurs in each of three cases: Case 1. When a dependent value is missing. Case 2. When an error value is missing. Case 3. When both a dependent value and an error value are missing. The error seems to happen here, but I don't know the code well enough to pinpoint what should change to fix this: runcell('Case 3: Missing dependent and error values', 'C:\/Users\/jam\/Documents\/GitHub\/ci-greedy-agents-') Traceback (most recent call last): File \"C:\\ProgramData\\Miniconda3\\lib\\site-packages\\spyderkernels\\\", line 356, in compatexec exec(code, globals, locals) File \"c:\\users\\jam\\documents\\github\\ci-greedy-agents-base\\\", line 27, in <moduleax.barlabel(bars) File \"C:\\ProgramData\\Miniconda3\\lib\\site-packages\\matplotlib\\axes\\\", line 2641, in barlabel endpt = err[:, 1].max() if dat = 0 else err[:, 1].min() IndexError: too many indices for array: array is 1-dimensional, but 2 were indexed Maybe either raise an error telling me what I should do instead, or have the code resolve whatever the source is on the backend? Ideally, I think the following should happen: Case 1. Raise an error that there is no value to apply the errorbar value to. Cases 2 & 3. Ignore the missing value and move on to the next. No response Windows 10.1 1 module:\/\/matplotlibinline.backendinline 5 Spyder 5.3.0 conda"}
{"_id":"q-en-matplotlib__matplotlib-22931-7c88298497d59a630722c4f02b0078de060370a803b40aaa3fcd28991562fdc9","text":"[Bug]: Regression in dashes due to Previously working plotting code has broken. It fails on \/ but passes on the parent . No line shown, error while drawing Line shown ! Maybe we use a weird\/bad\/incorrect convention in our code by creating and we can certainly work around it\/fix it, but it used to work at least! At a minimum it might be nice to raise a nicer error if possible... Ubuntu 22.04 QtAgg (PyQt6) 10.4 No response git checkout"}
{"_id":"q-en-matplotlib__matplotlib-22945-54e43b89ab62219affa53cb7b3e484407e9677fdbb18d70246f3c47815ec9cbe","text":"[Bug]: Regression in animation from On things are fine, on the next commit from we get an error with previously working animation code. No error No response Ubuntu 22.04 QtAgg (PyQt6) 10.4 No response git checkout"}
{"_id":"q-en-matplotlib__matplotlib-22991-c4a57a9d2447c1832e66a3b26122f0d5df7e7c04c64fd42f783b08992d7b6a05","text":"[Bug]: Can't use on axes created using subfigures Using on an axes created using subfigures result in an error. It should set the given ax as the current axe, as done by the code below: No response UBuntu 20.04 1 QtAgg 10 No response pip"}
{"_id":"q-en-matplotlib__matplotlib-23031-75e4c4d0d98bcc4756b7665d4c952a1022a8ab91c62c5ee904a72b92771b1bee","text":"[MNT]: Require that matplotlibrc\/style files use utf-8 (or have an encoding cookie) Currently, matplotlibrc and style files are read with the locale encoding, since . There's even a test for it in , which reads But actually, we probably never really supported non-ascii encodings (such as utf-32-be), because if you try to import matplotlib in such a context, we will fail much earlier, when trying to read the default matplotlibrc file: gives (the test doesn't see that because the default matplotlibrc file has already been imported at this point...). This behavior also means that style files are actually not shareable between systems that use incompatible encodings. Given that was implemented in response to , which is about the Py2\/Py3 unicode transition and not any user actually requesting support for non-standard encodings, I think we should just drop any intent of reading matplotlibrc\/style files using the user locale, and instead spec them as being utf-8 (or, if we want to be super-flexible, support encoding cookies as in \/ -- but I'd say it's probably not worth it?). No response"}
{"_id":"q-en-matplotlib__matplotlib-23047-16eb94046b9be1df9638cf22bced89d50678d7fa9af15e47726523eaa96b8295","text":"[Bug]: Gaps and overlapping areas between bins when using float16 When creating a histogram out of float16 data, the bins are also calculated in float16. The lower precision can cause two errors: 1) Gaps between certain bins. 2) Two neighboring bins overlap each other (only visible when alpha < 1) ! ! Created by ((np.float32), bins=100, alpha=0.5) () Possible solution Calculate the bins in float32: Determine minimal and maximal value in float16. Convert min and max to float32. Calculate the bin edges. Theoretical possible, but unwanted solution Convert data into float32 before calculating the histogram. This behavior does not make a lot of sense, as float16 is mostly used because of memory limitations (arrays with billions of values can easily take several gigabytes). Windows 10 3 TkAgg 1 No response pip Refactor hist for less numerical errors Should help with Idea is to do computation on the edges rather than the widths and then do diff on the result. This may be numerically better (or not...). Or rather, it is probably numerically worse, but will give visually better results... Probably the alternative approach of providing a flag to \/, making sure that adjacent bars are actually exactly adjacent may be a better approach, but I wanted to see what comes out of this first... <!-- Please mark any checkboxes that do not apply to this PR as [N\/A]. --Tests and Styling [ ] Has pytest style unit tests (and passes). compliant (install and run ). Documentation [ ] New features are documented, with examples if plot related. [ ] New features have an entry in (follow instructions in there). [ ] API changes documented in (follow instructions in there). without error). <!-- Thank you so much for your PR! To help us review your contribution, please consider the following points: A development guide is available at Help with git and github is available at Do not create the PR out of main, but out of a separate branch. The PR title should summarize the changes, for example \"Raise ValueError on non-numeric input to setxlim\". Avoid non-descriptive titles such as \"Addresses issue \". The summary should provide at least 1-2 sentences describing the pull request in detail (Why is this change required? What problem does it solve?) and link to any relevant issues. If you are contributing fixes to docstrings, please pay attention to In particular, note the difference between using single backquotes, double backquotes, and asterisks in the markup. We understand that PRs can sometimes be overwhelming, especially as the reviews start coming in. Please let us know if the reviews are unclear or the recommended next step seems overly demanding, if you would like help in addressing a reviewer's comments, or if you have been waiting too long to hear back on your PR. --"}
{"_id":"q-en-matplotlib__matplotlib-23049-aa61cde4c9366b3a037e9dccccbd444e262d3868e091e50b22ac3b6a76543213","text":"[MNT]: Should () get a minor keyword argument Extracted as remaining question from . Currently does not support a minor kwarg, in contrast to . it's not strictly necessary because pyplot may have less functionality than the OOP interface; but it doesn't hurt either. No response_"}
{"_id":"q-en-matplotlib__matplotlib-23057-3343e69e661ab34747d4ae5aef4069550bdb25eef45f61b9a0e142a599aafa26","text":"[Bug]: Figures fail to redraw with IPython A regression between release versions 3.5.1 and 3.5.2 causes figures to fail to redraw after an initial plot is using the interface in an interactive IPython session. This has been observed with both and . The figure will show the first plot drawn, but subsequent calls to and fail to update an on-screen figure until is invoked. This has been observed with IPython versions 8.3.0 (current) and 8.2.0. Both the Qt5 and Tk backends exhibit the same issues. the first command, a figure appears with a line shown. the second command, the figure does not update. the command, the figure updates to show both the and lines. the first command, a figure appears with a line shown. (This is as expected.) the second command, the figure updates with the addition of a line. (This is the deviation.) command should produce no visible change in the figure. This regression has been bisected to commit . The testbed is a current Void Linux system running Python 3.10.4, including the system package for a GUI. (As noted above, this bug is also present with the Qt5 backend.) All packages were installed in a virtual environment. The output of is: (Note that the funny version comes from a local git repo checked out to the problematic commit.) Void Linux x86_64 2 TkAgg, Qt5Agg 10.4 None pip"}
{"_id":"q-en-matplotlib__matplotlib-23088-861d520eb58dd81b533a774955e95a8ae10fc43af165f63f9c05424728b81855","text":"[Bug]: Confusing error messages Basically, plotting from a dataframe failed because of a keyerror but the message I received was regarding formatting using a string. The failure happened silently, causing me to spend over an hour tracking down a type because I had no clue where to start. Traceback (most recent call last): File \"<stdin\", line 1, in <moduleFile \"\/home\/bbriscoe\/thirdparty\/phel-1.2.0\/linuxx86649.4.0\/miniconda3-4.9.2\/lib\/python3.9\/site-\", line 1605, in plot lines = [self.getlines(args, data=data, **kwargs)] File \"\/home\/bbriscoe\/thirdparty\/phel-1.2.0\/linuxx86649.4.0\/miniconda3-4.9.2\/lib\/python3.9\/site-\", line 315, in call yield from self.plotargs(this, kwargs) File \"\/home\/bbriscoe\/thirdparty\/phel-1.2.0\/linuxx86649.4.0\/miniconda3-4.9.2\/lib\/python3.9\/site-\", line 452, in plotargs linestyle, marker, color = processplotformat(fmt) File \"\/home\/bbriscoe\/thirdparty\/phel-1.2.0\/linuxx86649.4.0\/miniconda3-4.9.2\/lib\/python3.9\/site-\", line 188, in processplotformat raise ValueError( ValueError: Illegal format string \"correctlySpelledHeader\"; two color symbols The actual failure is happening when the df and key are passed into this as data and value respectively. mpl.replacer(data,value): ----try: --------# if key isn't a string don't bother --------if isinstance(value, str): --------# try to use getitem --------value = data[value] <-----------------------Key Error because of typo ----except Exception: --------# key does not exist, silently fall back to key --------pass ----return sanitizesequence(value) As you can see from the comment, this happens silently. And as you can see from the Traceback provided the error you finally receive is regarding a formatting string. So this caused quite a bit of confusion, because I was looking everywhere except my header spellings. I feel like this shouldn't happen 'silently', it at least deseves a warning, perhaps: ----except Exception: --------('KeyError generated when attempting to access data using provided str') side note: the docstring says it returns data[value] or data back. in reality it passes back data[value] or value back. Not sure what the purpose is for allowing this to slide through, but either the behavior is wrong or the docstring is. No response Ubuntu 20.04 2 Qt5Agg 1 No response No response"}
{"_id":"q-en-matplotlib__matplotlib-23111-2f85c989f4a4a725be651a552addde0dac70966ddc2d413dc5ea670fa576cee3","text":"[Bug]: Imshow with all negative values leads to math domain errors. When using imshow to display a numpy array filled with identical negative values hovering the mouse over the displayed image throws math domain errors. Hovering the mouse over the plot spews math domain errors into console and stops the value and coordinate under the mouse from being displayed in the top right. ! The coordinates and values in the top right of the window are missing. ! The value area should be displayed correctly. Perhaps the delta in cbook\\\", line 2228, in gsigdigits (math.log10(delta))) if math.isfinite(value) else 0 is negative in this case. The errors do not occur if a single value in the array is different(negativ or positive doesn't matter). i.e. Will not error. The errors also do not occur if setclim() is used. Windows 10 .1645 1 QtAgg Python 3.9.12 No response conda"}
{"_id":"q-en-matplotlib__matplotlib-23140-750ba11a177ee56231e8ff1cc2adc948a0caf710333673fc18aa5b0ced65c59b","text":"Legend Title Left Alignment I'm trying to align my legend title on the left of the legend box. I assumed this would work: leg = ax.getlegend() titleinst = leg.gettitle() titleinst.sethorizontalalignment('left') However only this worked: leg = ax.getlegend() = \"left\" Seems like something is not working as expected, especially since I'm accessing something hidden."}
{"_id":"q-en-matplotlib__matplotlib-23174-7f43ae98f34d7e457fb8312e379df96551f6cf4ae0367d0d327ad54486b0210f","text":"[Bug]: Crash when adding clabels to subfigures Adding a clabel to a contour plot of a subfigure results in a traceback. The two subfigures appearing side by side, each showing the Contour Demo example No response Gentoo 2 QtAgg 13 No response Linux package manager"}
{"_id":"q-en-matplotlib__matplotlib-23188-967c0599f3a1ddc7051ed4c97df5713a368cb9dba35d9f897f0b0b3b2685ce41","text":"[MNT]: default date limits... The default date limits are 2000-01-01 to 2010-01-01. This leads to problems as a default if folks add for instance day tickers without specifying the limits. See for instance: We can change these to 1970-01-01 to 1970-01-02. For the default date epoch, this would yield default limits of 0-1 in our float conversion. If folks use the old epoch (0000-01-01), it would still work, but of course be something much larger than 0-1. This should only affect empty date axes, not any axes that folks have actually data to. If I make this change on main, the only tests that fail are empty axes tests."}
{"_id":"q-en-matplotlib__matplotlib-23198-e1268b3b7ba688711ae2a65c702d8044a9df7dfd9f068ec9fe2e6d593a286a94","text":"Inconsistency in keyword-arguments ncol\/ncols, nrow\/nrows I find it quite inconsistent that one sometimes has to specify and sometimes . For example: while (Likewise for \/)"}
{"_id":"q-en-matplotlib__matplotlib-23203-5c6311b489cc04e26df9876d40236353bfc0f4adaad74fde63d35f3dfcea83ac","text":"[Bug]: colorbar ignores keyword panchor=False seems to ignore the keyword setting My reading of the is that the axes anchor should still be \"N\" after I add the colorbar. Though given , it's possible I have misunderstood! No response No response branch (3.6.0.dev2341+g3df958c760), 3.5.2 and some older (3.3+) versions. No response No response No response No response"}
{"_id":"q-en-matplotlib__matplotlib-23266-5c966948c136de2ed309b7eebd7b0b98b372ae40d8a34b391848a5dfdf482998","text":"[ENH]: contour kwarg for negativelinestyle if you contour a negative quantity, it gets dashed lines. Leaving aside whether this is a good default or not, the only way to toggle this is via . I think this should be togglable via kwarg, though I appreciate that overlaps with and only is activated with monochrome contours. (I actually think the default should be False, FWIW - this surprises me every time, and I make quite a few contour plots)."}
{"_id":"q-en-matplotlib__matplotlib-23267-14d20134f76cec53dbb19e262f150ddc4b6ac9f06d7af737af371c5d9fc0297b","text":"[ENH]: Colorbar should support location kwarg that sets both orientation and ticklocation When colorbar autocreates an Axes, one can pass , which also sets the colorbar's orientation and the ticklocation (left for a left colorbar, right for a right colorbar, etc.). When one instead passes a manually created Axes (e.g. using insetaxes, as suggested by the example), the kwarg is not accepted (because things are directly passed to the Colorbar class); one needs to explicitly set and (the latter is not even documented by ): Add a kwarg to the Colorbar constructor which sets both and , and is mutually exclusive with them. ... or at least better document the workaround."}
{"_id":"q-en-matplotlib__matplotlib-23288-3cce379156fa05ff9f8352710fdae4161f7168320b66f4e9c7c3d34c8a5ab05a","text":"[Bug]: URL-area not rotated in PDFs The URL-sensitive area is not rotated in the PDF output Note that the link area is still the horizontal part as if the text was not rotated (this makes sense from reading the code). Clicking on the text, not where the non-rotated text would have been would activate the URL. In this is described in 12.5.6.5 From PDF version 1.6 it is possible to specify a \"QuadPoints\", i.e. a \"rectangle\" with four corners rather than just x, y, height, width as the current Rect has. However it says: it, the region specified by the Rect entry should be used. QuadPoints shall be ignored if any coordinate in the array lies outside the region specified by Rect. So one would also need to provide a larger Rect, which, for viewers not supporting QuadPoints will lead to that the total rectangle outlined by the rotated text will be clickable. This also holds for mathtexts. No response main No response No response No response_ git checkout"}
{"_id":"q-en-matplotlib__matplotlib-23299-247a49ab24aa9e15454436e129b56291af8fd6d26203a82a375f0abcc014f5f4","text":"[Bug]: getbackend() clears figures from if they were created under rccontext calling removes all figures from if the first figure in was created in an . The figure should not be missing from . Consequences of this are, e.g, doesn't work because can't find it. No response_ Xubuntu 2 QtAgg Python 3.10.4 n\/a conda"}
{"_id":"q-en-matplotlib__matplotlib-23314-9204b10581c9ff981427b54268cf60f1a1d4d00897d16a64bb4f4631c59da087","text":"[Bug]: setvisible() not working for 3d projection in the subplot projection=\"3d\" the setvisible function doesn't work even if the value is set to False the subplot remains visible which should not happen if the value is set to False the subplot is not visible if the value is set to False No response No response 2 Qt5Agg 10 No response No response"}
{"_id":"q-en-matplotlib__matplotlib-23332-3a906b03e8eda4aac9b3f06044308dbc0c6a2ae845e1e3982f67c02411b43187","text":"Certain non-hashable parameters to text() give cryptic error messages <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--Bug summary Per the title. See as well. Code for reproduction <!--A minimum code snippet required to reproduce the bug. Please make sure to minimize the number of dependencies required, and provide any necessary plotted data. Avoid using threads, as Matplotlib is (explicitly) not thread-safe.--all fail with Actual outcome see above Expected outcome Error out in the setter instead. Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: linux Matplotlib version: (): master Matplotlib backend (): any Python version: 38 Jupyter version (if applicable): Other libraries: <!--Please tell us how you installed matplotlib and python e.g., from source, pip, conda--<!--If you installed from conda, please specify which channel you used if not the default--"}
{"_id":"q-en-matplotlib__matplotlib-23348-0760326d2f2378a0f72c7655709980a4894f0108fec7cb30df1e891f4846e394","text":"MultiCursor should be able to bind to axes in more than one figure... Multicursor only works if all the axes are in the same figure... Originally posted by in and possible solution: Originally posted by in"}
{"_id":"q-en-matplotlib__matplotlib-23412-57a862f70e0e82076ad1c8336ee74893e535fb560b8a44634a059d2a24a60656","text":"[Bug]: offset dash linestyle has no effect in patch objects When setting the linestyle on a patch object using a dash tuple the offset has no effect. <img width=\"874\" alt=\"Screen Shot 2022-05-04 at 4 45 33 PM\" src=\"https:\/\/user-\"the patch edge lines overlap, not adhering to the offset. Haven't been able to get any patch objects to have a proper offset on the edge line style but the expected outcome is shown here with Line2D objects <img width=\"580\" alt=\"Screen Shot 2022-05-04 at 4 59 25 PM\" src=\"https:\/\/user-\"I have tried the Ellipse patch object as well and found the same issue. I also reproduced in Ubuntu 18.04 VM running matplotlib 3.5.0 with agg backend. OS\/X 4 MacOSX Python 3.8.8 No response conda"}
{"_id":"q-en-matplotlib__matplotlib-23476-a5233e80bd7eedd50dd566caf5318f70efb050938d65abdd63d6459ceb7512b4","text":"[Bug]: DPI of a figure is doubled after unpickling on M1 Mac When a figure is unpickled, it's dpi is doubled. This behaviour happens every time and if done in a loop it can cause an . This seems to happen only on M1 MacBooks and the version of python doesn't matter. OS\/X 2 MacOSX 12 No response pip"}
{"_id":"q-en-matplotlib__matplotlib-23516-59d016abc3124b5a961261c6d569be940039e11488554071a22b4713c024a658","text":"[Bug]: scatter plot color settings discarded unless c given When making an animation of a scatter plot, if you don't set (the color value parameter) when initializing the artist, the color settings are ignored. Even though is passed to , the points use the default \"viridis\" color map. I would expect the points to use the \"gray\" color map. If you modify the above code to use: it works as expected. It seems like all color-related settings, including cmap, vmin, vmax, etc. are discarded unless is given during the first call to . This workaround (passing an empty ) isn't problematic, but I found the default behavior quite unintuitive and it took me the better part of a day to figure out what was happening, so I figured it would be worth reporting. Possible solutions: Raise an error\/warning if \/\/ are given but is not Store parameters like \/\/ even if they aren't immediately used These changes should probably happen in in . According to , wrote most of this logic. macOS 12.4 2 MacOSX 13 N\/A from source ()"}
{"_id":"q-en-matplotlib__matplotlib-23562-3f92c0aaf2593a148ee986514fa3f76ceba15b120a9e07281390008c4146a341","text":"'Poly3DCollection' object has no attribute '_facecolors2d' The following minimal example demonstrates the issue: It fails on the last line with the following traceback: Tested with mpl versions 1.3.1 and 1.4.2. Sent here by Benjamin, from the mpl users mailing list (mail with the same title). Sorry for dumping this without more assistance, I'm not yet at a python level where I can help in debugging, I think (well, it seems daunting)."}
{"_id":"q-en-matplotlib__matplotlib-23563-6e03ad689c14d7f9d5150b2fcc66264ada2901e869ee2a05e8e857cf591f817b","text":"[Bug]: 'Line3D' object has no attribute 'verts3d' I use matplotlib 3D to visualize some lines in 3D. When I first run the following code, the code can run right. But, if I give a numpy array, it will report the error 'input operand has more dimensions than allowed by the axis remapping'. The point is when next I give and other variables an int number, the AttributeError: 'Line3D' object has no attribute 'verts3d' will appear and can not be fixed whatever I change the variables or delete them. The error can be only fixed when I restart the kernel of ipython console. I don't know why it happens, so I come here for help. Traceback (most recent call last): File \"\/home\/hanyaning\/anaconda3\/envs\/SBeA\/lib\/python3.8\/site-\", line 3444, in runcode exec(codeobj, self.userglobalns, ) File \"<ipython-input-80-\", line 20, in () File \"\/home\/hanyaning\/anaconda3\/envs\/SBeA\/lib\/python3.8\/site-\", line 368, in show return (args, kwargs) File \"\", line 29, in call (kwargs) File \"\", line 112, in show () File \"\", line 68, in show (self) File \"\/home\/hanyaning\/anaconda3\/envs\/SBeA\/lib\/python3.8\/site-\", line 436, in draw (self.renderer) File \"\/home\/hanyaning\/anaconda3\/envs\/SBeA\/lib\/python3.8\/site-\", line 73, in drawwrapper result = draw(artist, renderer, args, **kwargs) File \"\/home\/hanyaning\/anaconda3\/envs\/SBeA\/lib\/python3.8\/site-\", line 50, in drawwrapper return draw(artist, renderer) File \"\/home\/hanyaning\/anaconda3\/envs\/SBeA\/lib\/python3.8\/site-\", line 2803, in draw mimage.drawlistcompositingimages( File \"\/home\/hanyaning\/anaconda3\/envs\/SBeA\/lib\/python3.8\/site-\", line 132, in drawlistcompositingimages a.draw(renderer) File \"\/home\/hanyaning\/anaconda3\/envs\/SBeA\/lib\/python3.8\/site-\", line 50, in drawwrapper return draw(artist, renderer) File \"\/home\/hanyaning\/anaconda3\/envs\/SBeA\/lib\/python3.8\/site-\", line 469, in draw super().draw(renderer) File \"\/home\/hanyaning\/anaconda3\/envs\/SBeA\/lib\/python3.8\/site-\", line 50, in drawwrapper return draw(artist, renderer) File \"\/home\/hanyaning\/anaconda3\/envs\/SBeA\/lib\/python3.8\/site-\", line 3082, in draw mimage.drawlistcompositingimages( File \"\/home\/hanyaning\/anaconda3\/envs\/SBeA\/lib\/python3.8\/site-\", line 132, in drawlistcompositingimages a.draw(renderer) File \"\/home\/hanyaning\/anaconda3\/envs\/SBeA\/lib\/python3.8\/site-\", line 50, in drawwrapper return draw(artist, renderer) File \"\/home\/hanyaning\/anaconda3\/envs\/SBeA\/lib\/python3.8\/site-\", line 215, in draw xs3d, ys3d, zs3d = self.verts3d AttributeError: 'Line3D' object has no attribute 'verts3d' Some 3D lines No response Local: windows + pycharm, Remote: Ubuntu 20.04 0 module:\/\/backendinteragg 12 No response_ pip"}
{"_id":"q-en-matplotlib__matplotlib-23740-36b4f2fe8335b297504812d051c51d8f97e0b8bc01ba2afda80a46e4718c4759","text":"[Bug]: colorbar with unattached mappables can't steal space This is something I noticed downstream in networkx: Formerly, space for a was stolen from the current ; that was deprecated and now in 3.6, it is stolen from the mappable's . But if the mappable is not to an , it fails with a somewhat unrelated-looking error. Either we should switch to the current Axes if the mappable doesn't have one (probably not desired given the previous change), or raise a clearer error message when this happens. No response No response 0rc1 No response No response No response pip"}
{"_id":"q-en-matplotlib__matplotlib-23742-b2e71597f188e593e3ffba180ab05be5b8cbe43ef2a76b41243c3a6060e08daa","text":"[Bug]: Bug with toolbar instantiation in notebook In MNE-Python we have an abstraction layer for widgets+toolbars. Until today's latest install it was working fine. Now it fails with: See I'll work on boiling it down to a MWE tomorrow hopefully. Could also be a bug with traitlets. But the code that runs on the CIs is essentially this in a notebook I think: <details<summaryFull traceback<\/summary<\/detailsNo error EDIT: ipympl 0.9.2 Ubuntu (GH actions), or macOS M1 (locally) 0rc1 Notebook 10 EDIT: 6.4.11 pip"}
{"_id":"q-en-matplotlib__matplotlib-23913-cd9dde810fe992fe28db650de71b82042f3c98039eab3f026dc3a97be30b6b82","text":"legend draggable as keyword <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--There is not keyword to make legend draggable at creation <!--A short 1-2 sentences that succinctly describes the bug--Is there a code reason why one can not add a \"draggable=True\" keyword to the init function for Legend? This would be more handy than having to call it after legend creation. And, naively, it would seem simple to do. But maybe there is a reason why it would not work?"}
{"_id":"q-en-matplotlib__matplotlib-23964-79f61c9bb86f08b69a34f4751c2941a9ab798bdec04ad15c786ca84e356dfe8f","text":"[Bug]: Text label with empty line causes a \"TypeError: cannot unpack non-iterable NoneType object\" in PostScript backend When saving a figure with the PostScript backend, a happens if the figure contains a multi-line text label with an empty line (see example). $ .\/venv\/Scripts\/python Traceback (most recent call last): File \"C:\\temp\\matplotlibsaveps\\\", line 7, in <modulefigure.savefig('') File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\\", line 3272, in savefig (fname, kwargs) File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\\", line 2338, in printfigure result = printmethod( File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\\", line 2204, in <lambdaprintmethod = (meth)(lambda args, kwargs: meth( File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\api\\\", line 410, in wrapper return func(innerargs, innerkwargs) File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\backends\\\", line 869, in printps printer(fmt, outfile, dpi=dpi, dsccomments=dsccomments, File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\backends\\\", line 927, in printfigure (renderer) File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\\", line 74, in drawwrapper result = draw(artist, renderer, args, kwargs) File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\\", line 51, in drawwrapper return draw(artist, renderer) File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\\", line 3069, in draw mimage.drawlistcompositingimages( File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\\", line 131, in drawlistcompositingimages a.draw(renderer) File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\\", line 51, in drawwrapper return draw(artist, renderer) File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\axes\\\", line 3106, in draw mimage.drawlistcompositingimages( File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\\", line 131, in drawlistcompositingimages a.draw(renderer) File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\\", line 51, in drawwrapper return draw(artist, renderer) File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\\", line 1995, in draw (self, renderer) File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\\", line 51, in drawwrapper return draw(artist, renderer) File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\\", line 736, in draw textrenderer.drawtext(gc, x, y, cleanline, File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\backends\\\", line 248, in wrapper return meth(self, args, **kwargs) File \"C:\\temp\\matplotlibsaveps\\venv\\lib\\site-packages\\matplotlib\\backends\\\", line 673, in drawtext for psname, xsnames in stream: TypeError: cannot unpack non-iterable NoneType object The figure can be saved as without error. seems to happen if a text label or title contains a linebreak with an empty line works without error for other backends such as PNG, PDF, SVG, Qt works with matplotlib<=3.5.3 adding before line 669 of seems to fix the bug Windows 0 No response 13 No response pip"}
{"_id":"q-en-matplotlib__matplotlib-23987-c025e003d34473108592da53292b4f50d26aee4b5f01ed4a9cd06f956c66964b","text":"[Bug]: Constrained layout UserWarning even when False When using layout settings such as or , a UserWarning is produced due to incompatibility with constrainedlayout, even if constrainedlayout = False. This was not the case in previous versions. The plot works fine but the warning is generated UserWarning: This figure was using a layout engine that is incompatible with subplotsadjust and\/or tightlayout; not calling subplotsadjust. plt.subplotsadjust(wspace=0) no warning Warning disappears when constrainedlayout=False is removed OS\/X 0 No response No response No response conda"}
{"_id":"q-en-matplotlib__matplotlib-24026-8efacd17f75c6fb7d5c427a38076170c0b67a25d1fce0980443c04cea1dd6596","text":"stackplot should not change Axes cycler Usecase: I am producing various types of plots (some use rectangle collections, some regular plot-lines, some stacked plots) and wish to keep the colors synchronized across plot types for consistency and ease of comparison. While and support supplying a alias, stackplot throws a ValueError. For example: Originally posted by in"}
{"_id":"q-en-matplotlib__matplotlib-24088-846899918627c0d660861fab6254e7425ae40e12e57ede5b11d56b6122f7192c","text":"[Bug]: ValueError: Unable to determine Axes to steal space for Colorbar. started raising an error when trying to add a colorbar to : ValueError: Unable to determine Axes to steal space for Colorbar. Either provide the cax argument to use as the Axes for the Colorbar, provide the ax argument to steal space from it, or add mappable to an Axes. In : ! In : ! macOS 12.6 0 10 pip"}
{"_id":"q-en-matplotlib__matplotlib-24111-7b64cc7e67684ed2babc7da486f1c7e319e43db1f8d0740aeeb977aae124348f","text":"[ENH]: Default to call ? While addressing the deprecation in 3.6: I noticed that isn't directly migrate-able It appears from the source that defaults to so it would be nice if could default to that as well. Otherwise, it would be nice if this was better documented."}
{"_id":"q-en-matplotlib__matplotlib-24149-b485f679b89775c57b0a38475a0b22cbd3709adf2c1023c2b8f1f489f18ca7f4","text":"[Bug]: raises for all-nan data on matplotlib 3.6.1 raises an exception in 3.6.1 when passed only nan data. This irrevocably breaks seaborn's histogram function (which draws and then removes a \"phantom\" bar to trip the color cycle). On 3.6.0 this returns a with one Rectangle, having for and . I assume it's related to this bullet in the release notes: Fix barplot being empty when first element is NaN But I don't know the context for it to investigate further (could these link to PRs?) Further debugging: So it's about the x position specifically. Macos 1 No response No response No response pip"}
{"_id":"q-en-matplotlib__matplotlib-24177-74edd353c69bcf1516f03f210080713d8538a3109d9cb5b16a80464e8c262eee","text":"[Bug]: density not auto-scaled when using histtype='step' I need to plot a histogram of some data (generated by in binary format) from my work using the function. I noted that the histogram's density axis (when setting ) is not automatically adjusted to fit the whole histogram. I played with different combinations of parameters, and noted that the densities changes if you rescale the whole data array, which is counterintuitive as rescaling the data should only affect the x-axis values. I noted that if you set , the issue will occur, but is otherwise okay for other s. I started a github repo for testing this issue . The file is the data generated from my program. Here's the histograms generated using some simulated data. You can play with the and parameters in the code to see the differences. When , I got ! When , sometimes the randomised array would lead to identical left and right panel ... ! No response OS\/X 0 No response 10.4 No response pip"}
{"_id":"q-en-matplotlib__matplotlib-24189-dc63c7552e7bba1ba30c6b96e55a3ddbde59b7c7f7f1305413f2addcbaade80f","text":"[Bug]: Error using widthratios with nested mosaic in subplotmosaic() plt.subplotmosaic() fails with a confusing error message when passing widthratios (or heightratios) with a nested list mosaic, unless all outer and inner mosaics have the same number of columns (or rows). ! From what I can tell, this happens because the layout is built recursively, passing the same gridspeckw to subgridspec() at each level of nesting. I realize that the use of widthratios \/ heightratios \/ gridspeckw with nested list mosaics could get complicated, but it would be nice to be able to at least specify them for the outer list, or in some other way implement this feature for nested list layouts. If not, maybe include a helpful error message that explains this limitation so the user doesn't think they have specified the wrong number of ratios. Windows 10 0 QtAgg 10 No response_ conda"}
{"_id":"q-en-matplotlib__matplotlib-24224-c5cee514a7b8e5d84ad67abf1379a575326ef67fa76cbbe4c0d048ec56c6bce7","text":"[Doc]: document and ? These two methods have a decent doc-string, but are not in the documentation. They were 2006-08-10 Maybe add them to the documentation? It would make sense to add them towards the end of the page and include those and all functions after the last Locator in a new subsection \"Miscellaneous date functions\" or something. Edit: looking at the source, it is not really clear why they are not there..."}
{"_id":"q-en-matplotlib__matplotlib-24250-2fa2345fcfc9c1511f5d26cf6c43e92aa3748b25c9c3a2e9219376efbbe82c34","text":"[Bug]: pick events do not forward mouseevent-key on Linux Somehow on keyboard-buttons are not forwarded with ... on everything works as expected. When clicking on one of the datapoints while pressing the buttons and the following happens: on Windows I get: on CentOS Linux I get: The same output for both OS Windows \/ Linux CentOS7 1 Qt5Agg 13 conda"}
{"_id":"q-en-matplotlib__matplotlib-24257-e7b4fc3ed87da13cf48c9a7033c723e6c3d1e15bfaf695a146ddcd7c84b14e7d","text":"Document how to distribute style files in python packages <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--Style files (actually, style dicts) can be distributed as python packages, as discussed e.g. in This has the advantage of being relatively easy to do and that it works \"as is\" for essentially all versions of matplotlib (if your matplotlib is too old to have (pre 1.5...) you can still use ). In today's call we agreed that this approach should be documented and that a template package (similar to ) could be created. It was also pointed out during the call that this approach makes it easy to distribute colormaps together with the style, or add any additional custom logic; this may be worth documenting too."}
{"_id":"q-en-matplotlib__matplotlib-24265-bec636263ea2726f06d7262c3299db275333d21871383b6e31cb7a5960f90e59","text":"[Bug]: Setting ['seaborn-colorblind'] result in key error on matplotlib v3.6.1 I have code that executes: Using version 3.4.3 of matplotlib, this works fine. I recently installed my code on a machine with matplotlib version 3.6.1 and upon importing my code, this generated a key error for line saying \"seaborn-colorblind\" was a bad key. Traceback (most recent call last): KeyError: 'seaborn-colorblind' seaborn-colorblind should be set as the matplotlib library style and I should be able to continue plotting with that style. Bug occurs with matplotlib version 3.6.1 Bug does not occur with matplotlib version 3.4.3 Tested on MacOSX and Ubuntu (same behavior on both) OS\/X 1 MacOSX 7 No response_ pip"}
{"_id":"q-en-matplotlib__matplotlib-24334-01cdcb8239615647d2c589b304fc22220f9aea53401f519d63da3af2fb7f9d47","text":"[ENH]: Axes.setxticks\/Axis.setticks only validates kwargs if ticklabels are set, but they should Per the doc of : This means that in e.g. , the incorrect silently do nothing; they are not even validated (because has not been passed). We should at least check that are valid Text properties in all cases; we could even consider making any kwargs an error if is not set."}
{"_id":"q-en-matplotlib__matplotlib-24362-f95d314b0212beb4a7ce36d82b9ab7f4dca4f55aa02dbc673a754db67bfeb43a","text":"[Bug]: sharex and sharey don't accept 0 and 1 as bool values When using or in place of or in or arguments of an error is raised. We get the following error : Note that using instead of produces the same error (albeit with the following warning : but this is expected and not part of the present issue) I expected values 1 and 0 to be understood as bool. Suggested fix : Maybe not accepting 1 or 0 was done on purpose, but I did not find it very clear from the error message as and are accepted but not listed. I am happy to chat about an other fix, if this one doesn't do the trick. I can also create a PR in case you think this fix is good enough ! Linux 5.10.0-19-amd64 SMP Debian 5.10.149-2 600+g0b6d3703ff TkAgg 10.0 Not applicable git checkout"}
{"_id":"q-en-matplotlib__matplotlib-24403-7a02f7e78034c45bac261af446f44e7c561afdd9676206f4a36d3d3bbc974dd4","text":"[ENH]: Use instead of in the error message I mistakenly supplied as the argument for [matplitlib claimed for illegal color name like this: I was not aware that the argument actually contained a trailing newline so I was very confused. The error message would be nicer if it outputs user's input via . For example, in this case the error message can be easily replced with: so that we may now get an easy-to-troubleshoot error like this: This kind of improvement can be applied to many other places."}
{"_id":"q-en-matplotlib__matplotlib-24431-0cf36783aa5dd4ff82d059513497b627f7898f7d5cb06494fb989d57065a4035","text":"eventplot: allow a list of alpha channels as in the case with colors I'm not sure if it's a bug or a feature. It's both. works while throws an error matplotlib v3.3.3 Python 3.7.6"}
{"_id":"q-en-matplotlib__matplotlib-24570-1040f3cec7aa77defaae10bc87b26747ad311577be29b30661e635bcd80d8c23","text":"[Bug]: in is reversed For the parameter in , the options and seems reversed ! ! No response No response 2 No response No response No response No response"}
{"_id":"q-en-matplotlib__matplotlib-24619-9a0a1a194bd7bd990fae980740f27a8c4fce464b171a9fb540fccecb0b3452b6","text":"[Bug]: integer colours for pcolorfast \/ quadmesh I get an error when passing a byte\/integer array to pcolorfast to code the colors as RGBA. It also fails when data type is a plot in in some bluish colour fixes: 1) in , line 321: should be replaced by to allow for unsigned int values as well 2) in line 343: should be replaced by a test including - for 'i' and 'u'. It may be sufficient to comment it out as a quick fix as it is definitively more broken having it in. even with this it does not quite work 5.15.13-200.fc35.x86_64 1 gtk3 10.1 0 pip"}
{"_id":"q-en-matplotlib__matplotlib-24627-8996322fe7e1ab7c9f8c41560e2e5d255c6a6203bdd23064fa68c2377129c2d7","text":"cla(), clf() should unset the and attributes of deparented artists mpl2.0b3: Removing an artist from its axes unsets its attribute, but clearing the axes does not do so."}
{"_id":"q-en-matplotlib__matplotlib-24637-251a70351b97a3e2254af9ca3664d455baf85af1104226938af531bd4636b075","text":"AnnotationBbox gid not passed to renderer Hi, I'm creating matplotlib figures that contain images using AnnotationBbox (following the examples here ) and my aim is to set the artist gid associated with each image so I can access them later when saved to an svg. I can use setgid but when I save to an svg, the gid label for the images are not included. A similar issue has been discussed here , where a solution was applied for all known instances of missing gid's. Could it be that the AnnotationBbox artist has been missed by this fix? Example code: which prints: but produces an svg file that contains the image with no gid label (attached here as a txt file since svg is not supported): stock image used: ! Versions <!--Please specify your platform and versions of the relevant libraries you are using:--matplotlib version 3.3.4 python version 3.7.7 <!--Please tell us how you installed matplotlib and python e.g., from source, pip, conda--<!--If you installed from conda, please specify which channel you used if not the default--Thanks, Lauren"}
{"_id":"q-en-matplotlib__matplotlib-24691-48f093aab7b10438e27cdca4458c4316046abb5c4802a4be7562e7162e1928b5","text":"[ENH]: set facecolor and edgecolor alpha separately I like making diagrams where my patch edge is solid but the fill is semi-transparent - in my case usually to fade out the color, but it's also useful to emphasize boundaries ! Since alpha applies to the whole patch, the way I do this now is by converting my HTML colors into RGBs and appending an A ETA: apparently I could also just do Allow alpha to take a tuple for patch methods. As an example of prior art, we allow vectorized\/separable alphas in imshow as of"}
{"_id":"q-en-matplotlib__matplotlib-24749-67a4cbfa921f0dee62b5b2668f1cf6f0936232c6832357995b6bb716b93eb5af","text":"[Bug]: contour raises IndexError if Z is specified as keyword argument raises when is specified as . <img width=\"1675\" alt=\"Screen Shot 2022-12-15 at 8 23 39 PM\" src=\"https:\/\/user-\"<img width=\"713\" alt=\"Screen Shot 2022-12-15 at 8 24 46 PM\" src=\"https:\/\/user-\"No response OS\/X 2 module:\/\/matplotlibinline.backendinline Python 3.8.15 2 pip"}
{"_id":"q-en-matplotlib__matplotlib-24768-d637f755c732134e13e7dfb52dd83f4e8731420c699ae0b62cdefc26c8c8faad","text":"[Bug]: pcolormesh(rasterized=True) conflicts with setrasterizationzorder() According to the , a color plot can be rasterized in two ways: The two ways cannot be used together.  The bug appears in version 3.5.1 and commit , but not version 3.1.2. The most immediate cause seems to be reentrance tracking being dropped from and in commit : However, these are probably not the right places to fix this bug. Ubuntu 20.04, 22.04 2, 3.5.1, 3.7.0.dev447+g2d18bba0ea agg 10, 3.10.6 No response_ git checkout"}
{"_id":"q-en-matplotlib__matplotlib-24849-3c864434fd9f4d76fbdc8b15e8f5bba8be84bb3aca33e8bca6e3acd0c64b6a44","text":"[Bug]: gapcolor not supported for LineCollections doesn't have a or , so gapcolor doesn't work in plotting methods that return LineCollections (like vlines or hlines). ! I think the easiest fix is probably add and to LineCollection, modeled on and 1121+g509315008c"}
{"_id":"q-en-matplotlib__matplotlib-24870-f8afb5d53f93c04d9cb7c537dd5dfff5bfd3bcbc201b6ee760d667b5691a9a0a","text":"[ENH]: Auto-detect bool arrays passed to contour()? I find myself fairly regularly calling to draw the boundary line between True and False regions on a boolean 2d array. Without , one gets the default 8 levels which go at 0, 0.15, 0.3, 0.45, 0.6, 0.75, 0.9, 1.05 resulting in all the contour lines being drawn on top of one another; but clearly(?), for boolean inputs, the only choice that makes sense is to have a single level at 0.5 (or rather, anywhere between 0 and 1). ! Autodetect boolean inputs to contour, and default levels to [0.5] in that case. I guess the closest similar kind of autodetection in the library is for imshow, which auto-switches between 0-1 float RGBA arrays and 0-255 uint8 RGBA arrays (when given a 3D array as input). Thoughts?"}
{"_id":"q-en-matplotlib__matplotlib-24912-941c685e5bcbaa3cdf922f14bbd7b33daf662efdd538baf24f8d1ce348b2121b","text":"[ENH]: Allow override of contour level autoscaling In Matplotlib 3, when using a list of values for the argument in , the list of values is overridden in the case that all requested levels fall outside the data range. While this may be desirable for casually browsing data when the user is unfamiliar with the data range, it causes serious problems for batch applications where the user legitimately intends to use their list of levels but does not know whether every input array will produce contours. Example: The above prints when is an array of values ranging from 0 to 50 (i.e., the requested contour level of 100 is outside the data range). As a result, the plot contains erroneous contours around near-zero values, presumably due to floating point precision. This is a consequence of the change described here (): Add a kwarg to that overrides the autoscaling behavior. When the kwarg is set, it would trigger a flag in ():"}
{"_id":"q-en-matplotlib__matplotlib-24924-9491e15c220de496b7b9d221aa9ba5aa8e2b022680e7d46c398c0897dcf6afdb","text":"[Bug]: Cannot toggle settightlayout As of calling does not disable the tight layout algorithm. AssertionError clean exit I'm pretty sure just needs to set in the falsey case. Attn: No response main No response No response No response No response"}
{"_id":"q-en-matplotlib__matplotlib-24970-c60c15ce79d6edd731ab589113c9c08dfb58a24693c6b96a54ed63b45990b7ae","text":"[Bug]: NumPy 1.24 deprecation warnings Starting NumPy 1.24 I observe several deprecation warnings. No warnings. No response ArchLinux 2 QtAgg Python 3.10.9 No response Linux package manager"}
{"_id":"q-en-matplotlib__matplotlib-24971-fa21d7cd32e8ccbe12617402369f552b226d421e0486a63d211d8d2d3d8e680c","text":"[Bug]: compressed layout setting can be forgotten on second save I'm not sure whether this is really a bug or I'm just using an inconsistent combination of options. Under some specific circumstances (see below) compressed layout is not applied the second time a figure is saved. ! ! Both images should look like the first. If I do not set the , all is well. If I do not set in my calls to , the images are identical (but I have too much white space top and bottom). Maybe there is a better option than when using compressed layout? For context, my real example is a script that makes several figures. For most of them I want constrained layout, so I set that once in the for convenience. Only one figure needs \"compressed\", and I am saving twice because I want both a png and a pdf. Fixed it in my current example by just reverting the setting for the one figure. RHEL7 2 and main QtAgg 9 and 3.11 No response conda"}
{"_id":"q-en-matplotlib__matplotlib-25052-545f6cef66c07ce3a8cabef90f8374bba9657e25a4dd078ac6e26d5074c7c6fd","text":"[Doc]: bbox parameter The documentation states that should be an instance of . When I try to use it like this I obtain the following error traceback below using matplotlib v3.6.2, whereas it works if I define it as e.g. which I guess (but just because I tried empirically) is equivalent to the method of Bbox. Either let the parameter be a proper BBox (my suggestion) or specify that the usage is that of"}
{"_id":"q-en-matplotlib__matplotlib-25079-a4c0c7a164c698b59780601f3ba5c306856390d503164f4ceb1893d478b91521","text":"[Bug]: Setting norm with existing colorbar fails with 3.6.3 Setting the norm to a after the colorbar has been created (e.g. in interactive code) fails with an value in matplotlib 3.6.3. The same code worked in previous matplotlib versions. Not that vmin and vmax are explicitly set to values valid for and no negative values (or values == 0) exist in the input data. Works, colorbar and mappable are updated with new norm. No response Linux 3 (works with 3.6.2) Multpiple backends tested, same error in all (Qt5Agg, TkAgg, agg, ...) 15 not in jupyter conda"}
{"_id":"q-en-matplotlib__matplotlib-25085-2ac3fdab9b0badaddcfb4ee0ff5513bbc2541ff6fdc218548f8bda60b55a40ae","text":"[Bug]: Widget blitting broken when saving as PDF When running a test doc build for 3.7.0rc1, I build the PDF, which runs everything with the PDF backend. So either the PDF backend does not correctly mark itself as not supporting blitting, or the blitting is not turned off correctly in the button widgets. Docs build without warning. No response Fedora 37 v3.7.x PDF 11.1 No response git checkout"}
{"_id":"q-en-matplotlib__matplotlib-25122-708c83722e3dc92ce19777541bba7b0d4b09543a1d846c0696e8a9746aa9bbe5","text":"[Bug]: Windows correction is not correct in Windows correction is not correct in : The is not needed, and give wrong result for window with negative value, such as . For reference, the implementation of scipy can be found here : 4372. 0 No response No response latest No response No response No response None"}
{"_id":"q-en-matplotlib__matplotlib-25126-9bf5c499c5863008f244de28ae58a4067b5263930a8545e29d5e93f4452e4408","text":"[Bug]: (x,y) disappears after changing yscale The output of (x,y) disappears while changing yscale from 'log' (initial scale for the y axis) to 'linear'. The expected outcome is generated with matplotlib==3.5.3 fillbetween and hist are working fine. MacOSX 3 MacOSX 13 No response_ pip"}
{"_id":"q-en-matplotlib__matplotlib-25238-f81ba1b5f1b558f2590d17960a30d9afe2f17670b7348a00ad23f2b6ec82fd3e","text":"[MNT]: FFMpegWriter does not check if out path exists when initialized. FFMpegWriter does not ensure the outputpath exists when initialized (). This leads to a broken pipe error with no mention of a non-existent path which can be misleading to the user. When setup is called, check the output path exists and if not, throw an error to inform the user."}
{"_id":"q-en-matplotlib__matplotlib-25281-9576fa98b95fa7b2b73134127a165f99a58f27b9173bdae3217b9a064225266d","text":"[Bug]: Validation not performed for argument to When passing non-str values to , validation is not performed. So even for invalid inputs, errors are raised only when we call Errors should be raised when invalid arguments are passed to . Similar to what we get when we pass an invalid string value as shown: Do you know why this bug is happening? No validation is done when setting values for . We do check strings on line 473, which is why we don't face this issue there. Windows 2 'TkAgg' 7 No response pip"}
{"_id":"q-en-matplotlib__matplotlib-25287-3701b2c2b5fce4eb2547d7de26d33917cb9f6df5281ba3663d0ed52c7179816b","text":"[Bug]: offsetText is colored based on instead of tick.labelcolor In version 3.6.3, when setting ytick.labelcolor \/ xtick.labelcolor in styles \/ rcParams, it does not change the color of the exponent label as well. It will be colored based on \/ ! ! The following patch seems to fix it for my simple usecases: No response 3 No response No response No response None"}
{"_id":"q-en-matplotlib__matplotlib-25311-5a5ad1e79142da8bf2f98158c32c928db3c9a0f824e6a58facfc3c648fb7ab6e","text":"[Bug]: Unable to pickle figure with draggable legend I am unable to pickle figure with draggable legend. Same error comes for draggable annotations. Pickling successful No response Windows 10 0 No response 10 No response pip"}
{"_id":"q-en-matplotlib__matplotlib-25332-a19af7d83a407c80ca4d1d48ba208e0eb6af5ad2f1bc96f3e860e05487cd27ec","text":"[Bug]: Unable to pickle figure with aligned labels Unable to pickle figure after calling Pickling successful No response Windows 0 No response No response No response_ None"}
{"_id":"q-en-matplotlib__matplotlib-25334-4898e1a40758745caf0db87baafe9ac0fcc3b4764d0744d2e2faa2a8bd63e79d","text":"plt.contour with all NaNs fails assertion in Bug summary Passing an array with all NaN values into causes the following assertion to fail in : This is actually triggered by a test-case in the xarray test suite, but I imagine it hasn't been noticed (yet) because release builds of matplotlib typically disable assertion checks. Code for reproduction Actual outcome Failed assertion (see referenced line above, tracked down with ) Expected outcome I would expect to see the empty plot (and long lists of warnings) that are currently shown if assertions are disabled: ! Matplotlib version Operating system: Linux Matplotlib version: 3.0.3 Matplotlib backend (): agg Python version: 3.6 Python, matplotlib, etc are installed from source"}
{"_id":"q-en-matplotlib__matplotlib-25340-f266887bddb3e91938f33e631b5629b12d5ff8bca0e3cb45fa74fbd2ccdd745b","text":"[Bug]: setval of rangeslider sets incorrect value The setval() method of a range slider doesn't set the value correctly with values close to the minimal and maximal values of the range slider. With values in the middle, everything works fine. ! The values for the rangesliders \"wrong\" and \"wrong2\" are not set correctly Expected values are: (1, 2) and (9, 10) Actual values are (1, 3.25) and (7.75, 10) The result of using setval should be the same, as when using the valinit parameter of the constructor. The problem also occurred on Manjaro with: Python version: 3.10.9 Matplotlib version: 3.6.2 Matplotlib backend: QtAgg Installation of matplotlib via Linux package manager Windows 10 2 TkAgg 11.0 No response_ pip"}
{"_id":"q-en-matplotlib__matplotlib-25346-75d126c6fc64ff0f5de5b9f2e3f6ffd62698008d097d3de797a96c48257864e1","text":"[Bug]: constrained layout with wrapped titles When titles are long and wrapped, constrained layout doesn't leave enough room for them. ! Enough space for the titles. There's no problem if the titles have an explicit new line (\"\\n\"), so this seems to be specific to the wrapping. RHEL7 0 and main QtAgg 11.0 N\/A conda"}
{"_id":"q-en-matplotlib__matplotlib-25404-07ad81302ba06b4a384460d6696049d2ce8de0c47aede9b4232c0d8b850f7b04","text":"[Bug]: AttributeError: 'LassoSelector' object has no attribute 'props' I used the LassoSelector object to select the single point in the scatterplot. But when I try to update the line color of LassoSelector with the setprops function, I get an error like this AttributeError: 'LassoSelector' object has no attribute 'props'. This warning does not occur when the comment symbol is placed at the beginning of the line \"(props)\" in the \"set_ props\" function of the matplotlib library's code."}
{"_id":"q-en-matplotlib__matplotlib-25405-ae38a30a3ee82ae81bdae1d774ec703b6fff371221d6c7cbc086f2b5afb0592c","text":"[Bug]: LogLocator with subs argument fragile. Logarithmic tick markers do not appear if the y-axis scales a sufficient range, with the keyword argument of set to non-trivial values. ! ! I expect to see ticks in both cases, as appears in the \"Good Plot\". The problem is isolated to . This returns correct values in the first case (e.g., ([1.e-01 2.e-01 5.e-01 1.e+00 2.e+00 5.e+00 1.e+01 2.e+01 5.e+01 1.e+02 2.e+02 5.e+02])([]) in the second case. Windows 2 module:\/\/backendinteragg 13 No response conda [Bug]: LogLocator with subs argument fragile. Logarithmic tick markers do not appear if the y-axis scales a sufficient range, with the keyword argument of set to non-trivial values. ! ! I expect to see ticks in both cases, as appears in the \"Good Plot\". The problem is isolated to . This returns correct values in the first case (e.g., ([1.e-01 2.e-01 5.e-01 1.e+00 2.e+00 5.e+00 1.e+01 2.e+01 5.e+01 1.e+02 2.e+02 5.e+02])([]) in the second case. Windows 2 module:\/\/backendinteragg 13 No response conda"}
{"_id":"q-en-matplotlib__matplotlib-25425-2610cf0cb88e2fa2da02e24734b60c14c69ebe0a56db7af4b0188d1710135839","text":"[ENH]: Add as alias for in AxesImage, or make that include depth too changed to . While this indeed corresponds to , this is not consistent with the numpy API, where and . Perhaps we can consider 1) tweaking to instead, and add as an alias for ? Alternatively, could return (i.e., including whether the data is colormapped, RGB, or RGBA), and we could use that* info in ? (displaying whether the data is colormapped\/RGB\/RGBA seems reasonably useful) No response_"}
{"_id":"q-en-matplotlib__matplotlib-25430-029c0bb1042262d6cfc9a5c01b9d82813a4db859a869354ff8506e8c7bf23398","text":"[Bug]: savefig + jpg + metadata fails with inscrutable error message If we call with a with a extension, with the kwarg specified, the error message is inscrutable. Either metadata should be , the argument ignored, or a more informative error message. No response OS\/X 1 MacOSX Python 3.11.2 No response pip"}
{"_id":"q-en-matplotlib__matplotlib-25433-da8a098e6391a80a9d00059f156328d068bd0d1766cb31dce65dea68e62cb65b","text":"[Bug]: using clf and in range slider onchanged callback blocks input to widgets When using clear figure, adding new widgets and then redrawing the current figure in the onchanged callback of a range slider the inputs to all the widgets in the figure are blocked. When doing the same in the button callback onclicked, everything works fine. The widgets can't receive any input from a mouse click, when redrawing in the onchanged callback of a range Slider. When using a button, there is no problem. The range slider callback onchanged behaves the same as the button callback onclicked. The problem also occurred on Manjaro with: Python version: 3.10.9 Matplotlib version: 3.6.2 Matplotlib backend: QtAgg Installation of matplotlib via Linux package manager Windows 10 2 TkAgg 11.0 No response pip"}
{"_id":"q-en-matplotlib__matplotlib-25442-3b62c68b6e181e379692ea2c4c62223167d11fefeff903a07e83e714c5d5eedd","text":"[Bug]: Attribute Error combining matplotlib 3.7.1 and mplcursor on data selection If you combine mplcursor and matplotlib 3.7.1, you'll get an after clicking a few data points. Henceforth, selecting a new data point will trigger the same traceback. Otherwise, it works fine. No terminal output Using matplotlib 3.7.0 or lower works fine. Using a conda install or pip install doesn't affect the output. Windows 11 and Windwos 10 1 QtAgg 16 No response conda"}
{"_id":"q-en-matplotlib__matplotlib-25479-ea87290cb67c49f5edef54e02e9d5379e8d89aced3c947594e05689681112144","text":"Confusing (broken?) colormap name handling Consider the following example in which one creates and registers a new colormap and attempt to use it with the interface. Everything OK so far. Note the difference in the names and . Now when we try to use the new colormap things start to go wrong. As seen from the error message, it's that is looked up instead of the registered colormap name, . Manually looking up works just fine: For this to work as I had expected, one has to make sure that the colormap name and the registered name are the same due to some sort of \"double internal name lookup tables\" in matplotlib. I found this problem to be very confusing at first since I imported a colormap from another module, registered it, and tried to use it with no luck, e.g. something like: at which point, I expected to be able to refer to my newly registered colormap by the name ."}
{"_id":"q-en-matplotlib__matplotlib-25498-9aa4aab92634c9698a03033d0a2396bc790ca5192b0a6cc3475a030320fa76b2","text":"Update colorbar after changing How can I update a colorbar, after I changed the norm instance of the colorbar? has now effect and throws a -Exception. Consider this example: Output for :"}
{"_id":"q-en-matplotlib__matplotlib-25499-d8d49d5c5b46295ce45c1330bd824ece177fdfb6b6c4f1737264ebc5bc87ddf0","text":"[Bug]: Setting bboxinches to a Bbox in fig.savefig resizes colorbar Setting bboxinches in fig.savefig to a specified Bbox rather than \"tight\" resizes the colorbar relative to when bboxinches is not set—the resulting colorbar is a lot larger than it should be. ! ! Calling fig.savefig without bboxinches set first and then calling it again with bbox_inches set produces expected outcome. Calling () prior to savefig also works. MacOS 12.2.1 (Monterey) 1 inline (for interactive), png for plot, also a problem with pdf 10 Jupyter lab 3.2.9 conda"}
{"_id":"q-en-matplotlib__matplotlib-25547-cd87834b8241ab07b659336e917946cac680202872ab428a04d03c91967f0d56","text":"UserWarning thrown when all values are \"bad\", but not when only some are <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--Bug summary If all values passed as colors are \"bad\" (), a warning is thrown from numpy about converting masked values to null. If some of the values are \"bad\", this does not happen. Code for reproduction This does not warn: This does: This is difficult to hide since since the warning is not thrown when the figure is being constructed, but when it is shown. Here's a full traceback to show this: <details<summaryFull traceback <\/summary<\/detailsExpected outcome I wouldn't expect the second to warn if the first doesn't. Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: macOS Matplotlib version: 3.3.1 Matplotlib backend (): MacOSX and Python version: 3.8.5 Jupyter version (if applicable): Replicates in python repl, but also was happening in ipython and jupyter lab Other libraries: Installed with pip [Bug]: ax.errorbar raises for all-nan data on matplotlib 3.6.2 The function raises a error when contains only values. No crash, similar to the case where only some values are NaN. This happens because unconditionally looks for a first finite element in and . Debian 2 TkAgg 2 No response pip"}
{"_id":"q-en-matplotlib__matplotlib-25551-cfc2c3b51d1b9884a967b2cb6ed95c8e6f2edc209c0ba8d9f81668b105a88e4a","text":"handling of zeros in log-scale changes irreversibly after user zoom matplotlib 2.0b1 (but also present before that) Call e.g. and type \"k\" and \"l\" to switch both axes to log scale. The points at the two ends, where one of the coordinate is zero, are cropped out. This has been the behavior for a long time and seems reasonable. Now come back to linear scale (\"k\", \"l\") and zoom in with the zoom tool to one of the ends of the segment, e.g. xlims=(-0.1, 0.4), ylims=(8.6, 9.1) or something similar. Switching again to log scale now leads to a badly scaled plot, because the axis now goes all the way to (float epsilon). Even this is not unreasonable: the user effectively set his own axes limits, rather than the autocomputed ones, and these limits are invalid in log scale. The problem comes when you go back to the original limits (either \"back\" or \"h\" (\"home\")): even then, log-scale is broken (both axes to to instead of cropping the extremities); in fact, it seems impossible to restore the original behavior of autocomputed axes limits."}
{"_id":"q-en-matplotlib__matplotlib-25565-2fe949e5583658c59fb2ab6edd613040531a7c29a4f30eed3649c3ddeccd5758","text":"[Bug]: legend for Poly3dCollection fails When plotting an iso surface using Poly3dCollection, Supposed to insert a legend. ! on assumes is a , having the attribute . Consequently, a tuple will be generated if the 's size is not zero. In this case with , is already a and checking for will result in an error. A quick fix would be to have to make sure that colors has . MacOS 2 MacOSX 11.2 No response pip"}
{"_id":"q-en-matplotlib__matplotlib-25624-8758252ddcffd8ce98a718e5ffff7c8102965341a223672a8f1fd16760f84821","text":"tight layout kwargs have no effect if rc autolayout setting is set (MPL 1.5.3) Bug summary kwargs passed to tight layout do not have an effect if rc the figure.autolayout setting is set Code for reproduction No padding is inserted in the following example if figure.autolayout is set to True in the rc Matplotlib version Tested with a conda installed Matplotlib 1.5.3 on Ubuntu Linux."}
{"_id":"q-en-matplotlib__matplotlib-25631-a54223443fdb628c74964cd916bbea97b488c0354790c28dd1758212276aa73c","text":"[Bug]: does not work for writer.grabframe() will make snowflake movie in the example. The example runs fine after removing . https:\/\/user- https:\/\/user- No response No response 1 No response No response No response None"}
{"_id":"q-en-matplotlib__matplotlib-25640-12e1418110324696bb1aba1e4f656e37db4cc8feb14ca9b409c4072098456f2a","text":"PGF output: Contour labels extend beyond figure boundary Bug summary When using contour plots with the PGF backend, contour labels close to the axes can protrude beyond the axis boundary. This was originally posted as a . Code for reproduction (See the [ example][1].) Actual outcome ! Expected outcome The contour label at the top should be clipped, just as in the first image shown in the [ example][1]. Matplotlib version Operating system: Ubuntu 17.10 Matplotlib version: 2.2.2 Matplotlib backend: PGF ( shows , though) Python version: 3.6.3 was installed through . [1]:"}
{"_id":"q-en-matplotlib__matplotlib-25651-80e9f136b790c9a63eadd5318024f9097baf40dc5be2fe68a28a6fdf8f9e6539","text":"[MNT]: numdecs parameter in takes a parameter numdecs, which is not described in its docstring. I also can't find anywhere the parameter is used in the code. numdec (no s) is used within , but is calculated when needed If numdecs really isn't used, maybe remove it. Otherwise describe it in the docstring."}
{"_id":"q-en-matplotlib__matplotlib-25667-61130c8be7f7ef94ce49dd9fdf89add571df04b47aa0dcdc3e21ca4d6e494894","text":"[Bug]: bar\/barh don't trigger datetime units doesn't check the units of bottom\/left parameters to see if the axis needs a different converter. Same applies for : ! This works fine: ! main"}
{"_id":"q-en-matplotlib__matplotlib-25712-2722c8842e052bb8b6df8c04e161f4628a3c7b6d8c30c819852f7e34414011fa","text":"[Bug]: fillbetween{x} does not respect Axes transform Using an axes transform with and incorrectly sets the axes limits if the Axes coordinates are larger than the data coordinates. Note that code is slightly modified from the , but with the y-data values and threshold reduced by a factor of 10. What you get a plot where the y-limits have been expanded as if I've plotted y-data spanning between (0,1), but get a fill that covers the entire axis space. <img src=\"https:\/\/user-\" width=50%Should look like the , but with y axis labels reduced by a factor of 10. My guess is that the y-axis limits are being set by the \/ values in data coordinates before the transform is applied to actually fill the regions. You will get the expected result as long as the provided Axes coordinate values are less than the extreme values of the y-data itself. For example gives a correct result. <img src=\"https:\/\/user-\" width=50%But this issue means that you can't span the axes using this technique if your plotted data does not already span y=(0,1). Windows 10 1 module:\/\/matplotlibinline.backendinline 16 3 conda"}
{"_id":"q-en-matplotlib__matplotlib-25746-4c077e61b2933d72df330c68801a6fad01099a748c24185d59a3d53fe1ce0c89","text":"Add fontfamily\/labelfont to tickparams <!-- Welcome! Thanks for thinking of a way to improve Matplotlib. Before creating a new feature request please search the issues for relevant feature requests. --There is no simple way of assigning a fontfamily to tick labels without needing to either override the default fontfamily or setx\/yticklabels. We currently have access to color and size, kwarg to change the font would fit here nicely. <!-- For example: I'm always frustrated when [...] because [...] I would like it if [...] happened when I [...] because [...] Here is a sample image of what I am asking for [...] --Add a fontfamily\/labelfont kwarg to tick_params <!-- Provide a clear and concise description of a way to accomplish what you want. For example: Add an option so that when [...] [...] will happen --<!-- Add any other context or screenshots about the feature request here. You can also include links to examples of other programs that have something similar to your request. For example: Another project [...] solved this by [...] --"}
{"_id":"q-en-matplotlib__matplotlib-25772-a495124597c4d333d1b2c1fe9632ea74a5b36351e3fc92cc9d4de48a09457136","text":"[Bug]: VSCode matplotlib interactive mode cannot import Qt bindings Running in VS Code interactive mode, throws an exception . However, PySide6 is installed. <details<summaryError trace<\/summary<\/detailsRun in a VS Code interactive cell ipykernel: 6.22.0 ipython: 8.12.0 ipywidgets: 8.0.6 Windows 10 and Windows 11 (2 separate computers) 1 Qt5Agg 10.10 No response pip"}
{"_id":"q-en-matplotlib__matplotlib-25779-7e29eda10804c53dd637e87463e65e7d53b3f03efa3bdf75d9190158f8dfbf12","text":"Plot ellipse with arrow showing rotation I'm trying to plot an together with an arrow showing a rotation clockwise or counterclockwise as it is shown in that image. ! That can ben implement by another optional argument? No response Plot ellipse with arrow showing rotation I'm trying to plot an together with an arrow showing a rotation clockwise or counterclockwise as it is shown in that image. ! That can ben implement by another optional argument? No response"}
{"_id":"q-en-matplotlib__matplotlib-25785-4e4c82f12a90444c006e8472902465b221a7f3cd1f84722c0aec9f60395503f6","text":"automatic papersize selection by ps backend is almost certainly broken No minimal example, but the relevant chunk () is Note that the sorting is by name, which means that the size is the first one among \"a9, a8, ..., a2, a10, a1, b9, b8, ..., b2, b10, b1\" (in that order) that is larger than the requested size -- which makes no sense."}
{"_id":"q-en-matplotlib__matplotlib-25794-9119cce1dffda0800421c6ef34c6bea55632ceeab53974c0216a56a8086c474c","text":"Raise when both singular and plural scatter attributes are specified accepts both singular and plural forms of the and attributes. The plural forms are documented in the function signature, but the singular forms actually take precedence if both are specified. This adds some complexity for downstream libraries and confusion for their users (cf. ). Small change: Matplotlib could raise when both the singular and plural forms are specified. Larger change: I will confess that I don't know why the plural forms of the kwargs exist. If there's not a strong reason for the duplication, perhaps they could be deprecated, or at least \"formally discouraged\"? Scatter does a lot of argument checking on the \/ parameters (too much at times, ), so there's some local precedence for a lot of handholding. On the other hand, matplotlib generally doesn't raise when both long- and short-forms of kwargs are given edgecolorec`)."}
{"_id":"q-en-matplotlib__matplotlib-25859-6be971b49e11ada8b5a0fbf5832529f4d6a11982e22f74b750ea609c478e95df","text":"[MNT]: Do not accept arbitrary positional parameters in Figure.addaxes() ! seen at What are the parameters? This is not readable I'm surprised this works I suspect extra positional parameters are ignored. Documentation: Check whether the parameters do something. If so, document more clearly. if not deprecate extra positional parameters."}
{"_id":"q-en-matplotlib__matplotlib-25960-93eb8e183acbd6214b15451d1dbf555014d31b470d5e4f824ed927ffe5ba3ea6","text":"[Bug]: wspace and hspace in subfigures not working and in do nothing. Same figure independently of the values of hspace and wspace. No response OS\/X 1 MacOSX Python 3.10.9 No response conda"}
{"_id":"q-en-matplotlib__matplotlib-26011-cef42be199bcf5eba2d61a3b9d979f8ea2dd45c59de5aff55f9a27e054e4d808","text":"xlimchanged not emitted on shared axis <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--Bug summary When an axis is shared with another its registered \"xlimchanged\" callbacks does not get called when the change is induced by a shared axis (via sharex=). In the setxlim for sibling axis are called with emit=False: I'm very new to matplotlib, so perhaps there is a good reason for this? emit=False seems to disable both continued \"inheritance\" of axis (why?) and triggering of change callbacks (looking at the code above). It seems like one would at least want to trigger the xlim_changed callbacks as they would be intended to react to any change in axis limits. Edit: Setting emit=True seems to introduce a recursion issue (not sure why but as inheritance seems to be passed along anyway it doesn't really matter). Moving the callback call to outside of the \"if emit:\"-statement seems to solve the issue as far as I can see when trying it out. Any reason to keep it inside the if-statement?"}
{"_id":"q-en-matplotlib__matplotlib-26020-77ce0a4cb95ab978f2030046291232a0a02555218700a3d0b18daa410b08241a","text":"Error creating AxisGrid with non-default axis class <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--Bug summary Creating using cartopy as raises . Seems to be due to different behaviour of attr. for and other axes instances (like ) where is only a callable. The error is raised in method when trying to access keys from attr. Code for reproduction <!--A minimum code snippet required to reproduce the bug. Please make sure to minimize the number of dependencies required, and provide any necessary plotted data. Avoid using threads, as Matplotlib is (explicitly) not thread-safe.--Actual outcome <!--The output produced by the above code, which may be a screenshot, console output, etc.--Expected outcome <!--A description of the expected outcome from the code snippet--<!--If this used to work in an earlier version of Matplotlib, please note the version it used to work on--Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: Ubuntu 18.04.4 LTS Matplotlib version: 3.1.2 (conda-forge) Matplotlib backend: Qt5Agg Python version: 3.7.6 Jupyter version (if applicable): Other libraries:"}
{"_id":"q-en-matplotlib__matplotlib-26024-8f1f20d179bca286a33443ca945650b3c8ab46327dbb9426cf6cc80b88081479","text":"[ENH]: Missing mathematical operations Just browsed the available mathematical operators and compared with the ones defined. (One can probably do a similar thing with other groups of symbols.) The following are missing (as in not defined in in , in hex): For the corresponding symbols, see: For LaTeX names, see: One should probably be a bit discriminate when adding these, but at least those in standard LaTeX (like = ) and those from AMS should be supported."}
{"_id":"q-en-matplotlib__matplotlib-26089-4235b1682a2fc208c1010d23cc3e0579c87b34d7d2940a0ad3905fcecc3d95e7","text":"[ENH]: Add public method to update object's loc property . I'm working on the issue The object's property can only be set at initialization time. There is no public method to update the property when the object has been created. It can now be understood as implemented as follows: Desired implementation No response"}
{"_id":"q-en-matplotlib__matplotlib-26101-f6150b0941b9013cf099fda902f7cdce057ba8867685dbac9e074ac833e9a495","text":"[Bug]: Star marker (using mathtext) is not center-aligned Is there any other way to center-align mathtext markers? ! All markers using mathtext were not center-aligned center-aligned markers (whether mathtext is used or not) No response No response 1 No response No response No response pip"}
{"_id":"q-en-matplotlib__matplotlib-26113-3b52c12052c4966b6fec54423b42f1fe3c6be53b4f423f8c2bb3a4bd6fb5b034","text":"Inconsistent behavior of hexbins mincnt parameter, depending on C parameter <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--Bug summary Different behavior of s parameter, depending on whether the parameter is supplied. Code for reproduction See below for a full snippet. Actual outcome <!--The output produced by the above code, which may be a screenshot, console output, etc.--With no parameter specified, a value of works as I intuitively expect: it plots only gridpoints that have at least 1 datum. With specified but not specified, I can kind of understand why it defaults to only gridpoints that have at least one data point, as otherwise the has to yield a sensible output for an empty array. Expected outcome However, with I'd expect the same gridpoints to be plotted, whether is supplied or not... Additional resources The most recent commit that changed how I should interpret : The lines in current code that deal with when is : The lines in current code that deal with when is not : Resolution Although it might mean a breaking change, I'd prefer to see the behavior of being applied also when isn't None (i.e. , rather than the current ). I'm happy to supply a PR if the matplotlib maintainers agree. Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: Linux 4.15.0-38-generic Matplotlib version: 3.0.2 Matplotlib backend (): Python version: 3.6.7 (default, Oct 22 2018, 11:32:17) Jupyter version (if applicable): Other libraries: numpy: 1.15.3 <!--Please tell us how you installed matplotlib and python e.g., from source, pip, conda--<!--If you installed from conda, please specify which channel you used if not the default--Inconsistent behavior of hexbins mincnt parameter, depending on C parameter <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--Bug summary Different behavior of s parameter, depending on whether the parameter is supplied. Code for reproduction See below for a full snippet. Actual outcome <!--The output produced by the above code, which may be a screenshot, console output, etc.--With no parameter specified, a value of works as I intuitively expect: it plots only gridpoints that have at least 1 datum. With specified but not specified, I can kind of understand why it defaults to only gridpoints that have at least one data point, as otherwise the has to yield a sensible output for an empty array. Expected outcome However, with I'd expect the same gridpoints to be plotted, whether is supplied or not... Additional resources The most recent commit that changed how I should interpret : The lines in current code that deal with when is : The lines in current code that deal with when is not : Resolution Although it might mean a breaking change, I'd prefer to see the behavior of being applied also when isn't None (i.e. , rather than the current ). I'm happy to supply a PR if the matplotlib maintainers agree. Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: Linux 4.15.0-38-generic Matplotlib version: 3.0.2 Matplotlib backend (): Python version: 3.6.7 (default, Oct 22 2018, 11:32:17) Jupyter version (if applicable): Other libraries: numpy: 1.15.3 <!--Please tell us how you installed matplotlib and python e.g., from source, pip, conda--<!--If you installed from conda, please specify which channel you used if not the default--"}
{"_id":"q-en-matplotlib__matplotlib-26122-0ecffd78f8b0ff1b13325f73c295d0f81a7366089b4f0622c5856e511c8f57b9","text":"imshow() should not modify axes aspect if transform != ax.transData. Currently, imshow() automatically updates the axes aspect via the kwarg; its default, None, means , which is \"equal\" by default (i.e., square image pixels). If the kwarg is also passed, and set to something else[1] than (the default), then setting the aspect is clearly not useful (the image is not going to be drawn in data coordinates so it should not affect the relative size of x- and y-data). In that case, the default of should just mean \"don't modify the aspect\". [1] Really, this should be \"something that does not contains transData as a branch\", as in . The current behavior is the reason why and need to explicitly set the aspect to \"auto\" in or after the last imshow() call (otherwise, some head-scratching occurs). On the other hand, making this change would once again lead to some seriously non-obvious interaction between parameters (the meaning of depends on the value of ), which I'm not sure is great either :\/ Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: linux Matplotlib version: master\/any Matplotlib backend (): qt5agg Python version: 37"}
{"_id":"q-en-matplotlib__matplotlib-26160-e90d6486759aff03a98f2d247cc81f730c764e579f6f15f8e63c7b96948c7cca","text":"[ENH]: Add setters for AxLine.xy1, .xy2, .slope Currently the control points \/ slope of the artist returned by axline() cannot be (publically) modified after being instantiated. It would be nice if the relevant properties (xy1, xy2, slope) had setters (following normal Artist design). For simplicity it is probably enough if we don't let one set xy2 if slope is set and vice-versa (i.e. whether axline is specified by 2 points or by point-and-slope is locked in). Note that while I do have a use case for changing a previously set xy1\/xy2, wanting to switch between the two different representations seems rarer to me(?) This would likely also make AxLine public. No response_"}
{"_id":"q-en-matplotlib__matplotlib-26184-087e3b27edc6ba5f58fd4ff0b732388946591301a997f9ab5c8249a5d990b420","text":"[Bug]: AnnotationBbox does not return correct windowextent before first draw I’m trying to use a constrained layout in a visualization that contains an artist that is an instance of AnnotationBbox, and matplotlib raises a warning saying constrained layout is not applied. The visual effect is not evident in this simple example, but it becomes very clear once we have multiple panels. UserWarning: constrainedlayout not applied because axes sizes collapsed to zero. Try making figure larger or axes decorations smaller. No warning should appear The following works without any warning The problem with the constrained layout is more evident if I have two or more panels. One way of fixing it (i.e. getting rid of the warning and bad functionality) is to do ab.setinlayout(False) before doing ax.addartist(ab). This was first posted on Discourse Ubuntu 22 2 No response No response No response_ pip"}
{"_id":"q-en-matplotlib__matplotlib-26208-fe8c7bd2f51639508855918a6eefe4a4048f36f99c17a361c3cd4fe9a955637f","text":"[Bug]: dataLims get replaced by inf for charts with twinx if ax1 is a stackplot Bringing this over from Discourse In Matplotlib 3.4.0 and later versions, when using twin x-axis (two-y-axis charts), the data limits (dataLims) of the first axis (ax1) get changed to ±inf when plotting a stackplot on the second axis (ax2), which is unexpected. This prints It caught me off guard that the ax1 dataLims get changed to ±inf. It’s interesting that, if you swap the plot order (i.e. do plot on ax1 and stackplot on ax2, the dataLims don’t get replaced by infs: [-22. 26.585 ] \/ [-2. 0. ] \/ ). To not change ax1 dataLims, since I made no changes to it, like with matplotlib versions prior to 3.4.0. I went throught he changelogs and couldn't find (or perhaps missed it) that this behavior change was intentional. No response Windows 10 0 through 3.7.1 9 for old versions, 3.11.3 for new versions No response_ pip"}
{"_id":"q-en-matplotlib__matplotlib-26223-9c72529a20135031f8787c6fd7c9820f7a1a1f881997ef3a1d42bcad9035fa06","text":"[Bug]: pcolormesh writing to input mask When receives a masked array, it seems to be writing back to the mask. Since numpy 1.24 this now causes to fail if the mask is read-only. No error The code above runs fine with numpy v1.23, although the output from was already read-only at that version. From numpy release notes, this looks like the likely reason for the change: Aside from the new error, if a user passes a masked array that has nans or infs at the unmasked points, we are modifying their input array with the call to . I guess we just need to take a copy somewhere? RHEL7 1 QtAgg 11.3 N\/A conda"}
{"_id":"q-en-matplotlib__matplotlib-26232-330a39ed2cd3f96177b042b9e56731834ecb30d14ce06726bce2712c531d19d6","text":"[Bug]: pcolor writing to read-only input mask When the parameter or is a masked array with a read-only mask, fails with Traceback (most recent call last): File \"<stdin\", line 1, in <moduleFile \"\/Library\/Python\/3.9\/lib\/python\/site-\", line 2773, in pcolormesh ret = gca().pcolormesh( File \"\/Library\/Python\/3.9\/lib\/python\/site-\", line 1442, in inner return func(ax, map(sanitizesequence, args), kwargs) File \"\/Library\/Python\/3.9\/lib\/python\/site-\", line 6220, in pcolormesh X, Y, C, shading = self.pcolorargs('pcolormesh', args, File \"\/Library\/Python\/3.9\/lib\/python\/site-\", line 5713, in pcolorargs X, Y = [cbook.safemaskedinvalid(a) for a in [X, Y]] File \"\/Library\/Python\/3.9\/lib\/python\/site-\", line 5713, in <listcompX, Y = [cbook.safemaskedinvalid(a) for a in [X, Y]] File \"\/Library\/Python\/3.9\/lib\/python\/site-\", line 715, in safemaskedinvalid xm = (x, copy=False) File \"\/Library\/Python\/3.9\/lib\/python\/site-\", line 2360, in maskedinvalid res = maskedwhere(~(np.isfinite(a)), a, copy=copy) File \"\/Library\/Python\/3.9\/lib\/python\/site-\", line 1942, in maskedwhere = shrinkmask(cond) File \"\/Library\/Python\/3.9\/lib\/python\/site-\", line 3516, in mask self.setmask(value) File \"\/Library\/Python\/3.9\/lib\/python\/site-\", line 3462, in setmask_ = mask ValueError: array is read-only No error The error still exists because I missed the following code in fixing : And I will fix it as well if allowed. OS\/X 1 MacOSX 9 No response_ None"}
{"_id":"q-en-matplotlib__matplotlib-26249-7db91db8a2e6af1189fd7535c741521a493388ea0194fb541ebd4448dca51e96","text":"[Bug]: ax.scatter (projection='3d') - incorrect handling of NaN In axis 3D projection NaN values are not handled correctly, apparently the values are masked out (as it should be) but the mask is not applied to a color array that may not have NaN in the same position. A plot with the first and 3rd data point. Unconditionally reproducible. I have not seen this before, but I may never have called it this way before. Fedora 38 1 TkAgg 11.4 IPython 8.14.0 pip"}
{"_id":"q-en-matplotlib__matplotlib-26278-c5fca2ee4aeb05573f3f43db925f6bc096b1191d804bc0b7410ecf3544d63455","text":"Cleaning up kwargs in ContourSet This is a continuation of a mailing list thread where we talked about how to clip a plot inside a polygon. It is a very useful application for people who are visualizing data on maps since often times we want to plot everything inside one region (country, state or province). Currently for many types of plots this is not that hard to do using the clippath keyword for most of the plotting functions, since the kwargs are usually used to modify properties of the artists to be generated by the plotting function. For example, suppose that I had a polygon defining the border of a country, poly, and some data to overlay on top. Does what I want because the kwargs of pcolor let me modify the underlying PolyCollection instance. However, there are a few plotting functions where I cannot do this, most notably in contour \/ contourf: Will work but the clippath kwarg gets completely ignored. To get the result I want, I need to store the output of contourf and use the setclippath method on each collection instance: So I looked at the code in and realized that no kwargs get passed when instantiating the collections. mentioned that this might call for an overhaul of how the kwargs get passed into a ContourSet. His suggestion was either adding a setclippath method directly to ContourSet, or a more thorough change of how the kwargs are getting passed so they are more consistent with the other plotting functions. Ideally, I would prefer the latter case since then for my usage case I could always get what I want just by passing in the kwarg directly. Additionally it would make the functionality of contour(f) more similar to the other plotting functions, ie some of the kwargs can be passed to the collections. Any thoughts on this?"}
{"_id":"q-en-matplotlib__matplotlib-26285-df894e15622df0159276d2350f92837b7d022c057c3b7bf3dc7485a025c8179f","text":"[Bug]: setticks provides mysterious error message errors with . Two issues here: is incorrect for , I should have done . It's a bit annoying that is the kwarg for some things and for . Second, the error message is somewhat annoying as I would have expected this call to work or give me an error for an incorrect kwarg. No response No response main No response No response No response None"}
{"_id":"q-en-matplotlib__matplotlib-26291-6d514556b995da56a06f4971981533623817a38dffd2607c130355e22a10a2be","text":"[Bug]: Error while creating inset axes using Unable to create the inset axes in a plot using the code (following the first example on the website as posted posted below. I was expecting to add an empty box towards the top right of the first subplot (with axes ) in the figure, as shown in the demo on the website. No response Arch linux: 6.4.2-arch1-1 2 module:\/\/matplotlibinline.backend_inline Python 3.8.17 Jupyter lab: 3.6.5 conda"}
{"_id":"q-en-matplotlib__matplotlib-26300-ce8e7414c93126c802d9148d48e270e82676991b248ca8fa030684485af87720","text":"[Bug]: calling fig.tightlayout multiple times Calling multiple times warns. no-warning. does not show up in 3.7.1, does show up in 3.7.2. Have not bisected this yet (or checked main). From looking at the code I suspect \/ \/ which is from me . xref I suspect the fix is to not warn if we set the place holder due to . Arch 2 any 11 No response_ pip"}
{"_id":"q-en-matplotlib__matplotlib-26311-c73438bc608a263eae60b79dcb1bde091c5bf2c88f77177ca1e7acdcaae9bc3c","text":"[Bug]: labels can't be placed at start of contours For some combinations of contour shape and fontsize, the automatic label placement tries to put the label right at the start of the contour. This is not currently possible on . With v3.7.1 I get ! The fix is easy: Writing a test seems harder. I tried pasting the above code into a test, and it passed against . I assume that is because the tests have different \"screen space\" than when I just run it as a script. Marking as \"release critical\" because this is a regression. RHEL7 main QtAgg 11.3 N\/A git checkout"}
{"_id":"q-en-matplotlib__matplotlib-26341-f088f505947c185b3b839f74f40ee2408c6306f4c4d1f5f32a50b193f1cd64e0","text":"[MNT]: Make cyclers indexable and rely on indexing them rather than Currently the propcycle code (in processplotvarargs) creates an over the Cycler instance to yield the successive line properties. objects are opaque, which creates some difficulties e.g. in parsescattercolorargs which needs to use self.getpatchesforfill.getnextcolor to workaround the impossibility to peek at the next color in the cycle without advancing the iterator, and also with pickling (currently we just completely drop the cycler state when pickling\/unpickling). An alternative would be to drop the use of and instead simply store in processplotvarargs both the Cycler object and an integer index, which simply gets incremented at each use, and add support for indexing Cyclers (perhaps something like or forcing the caller to explicitly write , to avoid confusion with the fact that returns the finite, non-cycled length). This would both make peeking at the next color easier, and directly solve the issue of picklability. No response"}
{"_id":"q-en-matplotlib__matplotlib-26342-29ffeff78cb5b0a1705d9ba82eba410e8bd03153ab94410670488040ea976255","text":"[ENH]: ContourSet.set_paths To get contour labelling working with its special transforms, Cartopy has a where it replaces all the paths on the with transformed versions. This currently looks like which doesn’t smell very good. The above would smell better as"}
{"_id":"q-en-matplotlib__matplotlib-26399-2fe88051ff5050ddd7218041eb5fcf28405f48af29140373b62540acdc7c3833","text":"[Bug]: ContourSet.antialiased attribute not present The new does not have an attribute. This causes failures in . With v3.7.1, I can access this attribute. Marking as release critical, as this is a regression. RHEL7 main QtAgg 11.4 N\/A git checkout"}
{"_id":"q-en-matplotlib__matplotlib-26466-461bf65779489e35c83afb19679fb801fa379928bce8bc82b16b47122c2e9d96","text":"Updating an array passed as the xy parameter to annotate updates the anottation Bug summary When an array is used as the xy kwarg for an annotation that includes arrows, changing the array after calling the function changes the arrow position. It is very likely that the same array is kept instead of a copy. Code for reproduction Actual outcome ! Expected outcome Both arrows should be horizontal Matplotlib version Operating system: Debian 9 Matplotlib version: '3.0.3' Matplotlib backend: Qt5Agg Python version:'3.5.3' Jupyter version (if applicable): Other libraries: Numpy 1.17.3 Matplotlib was installed using pip"}
{"_id":"q-en-matplotlib__matplotlib-26472-611c6fd30e66d7b0d9d0486b3d7c370ea7786bd33e16ae853da1ea00b8ec31de","text":"Switching to inline backend closes GUI windows <!--To help us understand and resolve your issue, please fill out the form to the best of your ability.--<!--You can feel free to delete the sections that do not apply.--Bug summary When a plot is made with the backend on and then the backend is changed to , the active plot window(s) are closed. This makes it very difficult to switch cleanly between backends within a notebook. The two mediocre workarounds I can see are: always specify the backend before plotting (since you can't automatically \"switch back\") use and switch back to after the user is done with the GUI figure. This has the downside of locking the Notebook execution while the figure is live. Code for reproduction The linked gist shows what I'd like to accomplish which is a context manager that enables matplotlib plotting in a GUI window from within a Notebook that is otherwise using the backend. Basically I want the notebook to use inline (for a variety of reasons), but I occasionally wish I could interact with the data in a separate figure. As a side note, easy switching between and \/ backends would also scratch part of this itch, but the interactive notebook backends still don't enable full screen usage or easily throwing a plot on another monitor. has full screen figured out, so that seems doable. Here's the context manager I wish worked: Actual outcome The outcome of the above code is that a plot window flashes into existence for a split second, then is closed when the call is processed. Expected outcome It would be great if plots created with the qt5 backend could stay visible while other plots with the inline backend were also being created. If use (to establish a reliable event loop) and create a bunch of Qt windows by other means, they live concurrently with the Notebook as long as the kernel is alive. I'd like matplotlib GUI figures to be able to live on regardless of the current backend. Matplotlib version <!--Please specify your platform and versions of the relevant libraries you are using:--Operating system: Matplotlib version: 3.1.1 Matplotlib backend (): and Python version: 3.7.3 Jupyter version (if applicable): <!--Please tell us how you installed matplotlib and python e.g., from source, pip, conda--<!--If you installed from conda, please specify which channel you used if not the default--"}
{"_id":"q-en-matplotlib__matplotlib-26479-d2287f0600229785a782b8e53392f11a43bf02900116c19c52d2da029572fcd0","text":"Postscript backend gives wrong page sizes Bug summary When creating a Figure of exactly A4 size, the PS backend chooses \"letter\" as document type, leading to additional borders of the output in x direction and undesired cropping in y direction. Code for reproduction Actual outcome When not specifying the papertype explicitly, the PS backend chooses \"letter\" format as can be seen from the resulting postscript output. It should, instead, choose A4 format. When specifying the papertype explicitly, the output looks fine. Expected outcome The PS backend should choose A4 format if the Figure is exactly this size. Anyway, nothing should ever be cropped off the Figure. If the Figure does not fit one papertype, the next larger one should be chosen. I also do not understand why the output of the PS backend is restricted to a handfull of explicit paper sizes in the first place. Postscript does well support arbitrary page sizes. Can someone explain why matplotlib chose to only support the given sizes? This is not transparent to the user, who expects to get output of exactly the size of his\/her desired Figure object. Matplotlib version Operating system: Ubuntu 19.04 Matplotlib version: 3.1.1 Matplotlib backend: PS Python version: 3.7.4 Jupyter version: 1.0.0 Other libraries: Matplotlib was installed via conda."}
{"_id":"q-en-matplotlib__matplotlib-26532-61930ae401e251431c52389f3da89e46d3a9f5e69b1428c0af260f72bcd12c94","text":"Typo in Poly3DCollection constructor There is a typo in that causes a exception whenever the function is called with . should be"}
{"_id":"q-en-mwaskom__seaborn-2389-5a865d070fef9c94aa3a945a9f3372e3c1f04c2590af22275ae2e2b9c3733d82","text":"ValueError: fill value must be in categories In the preprocesscolors function, there is the code to replace na's with background color as the comment said, using , however, if the original colors do not contain the 'white' category, this line would raise the Pandas ValueError:fill value must be in categories in ."}
{"_id":"q-en-mwaskom__seaborn-2576-094eba2fd2741e504fe23844861a33b992c242ea17fffefe237282595991fede","text":"lmplot(sharey=False) not working The following code behaves as if . (edit: actually, it does not behave the same, but it is still not rescaling the plots individually the way it should) If you do this, it suddenly works: Versions of seaborn and matplotlib: ! lmplot(sharey=False) not working The following code behaves as if . (edit: actually, it does not behave the same, but it is still not rescaling the plots individually the way it should) If you do this, it suddenly works: Versions of seaborn and matplotlib: ! Allow xlim as parameter for lmplot Seaborn versions: latest dev version and 0.11.1 doesn't seem to accept the parameter, although FacetGrid does. Use case: when , the regression lines are extrapolated until they touch the current xlims. If one afterwards want to extend these xlims, the regression lines are floating again. A workaround is either to call FacetGrid and regplot separately, or to set very wide xmargins via the rcParams. Example code."}
{"_id":"q-en-mwaskom__seaborn-2813-6f616bf1004e07c7344e4bd6737500792212d867875ed8d55346db92e413f62b","text":"histplot stat=count does not count all data points import as plt import seaborn as sns import numpy as np (style=\"whitegrid\") dataa = [1, 2, 3] datab = [2.4, 2.5, 2.6] sns.histplot((dataa), color=\"red\", binwidth=0.01, stat=\"count\") sns.histplot((datab), color=\"blue\", binwidth=0.01, stat=\"count\") This produces The data point 2.6 is omitted in the output produced by histplot. The problem also exists, if the first sns.histplot command is removed. Interestingly, it has been pointed out to me that the following command works: but as I said, the single command also does not work."}
{"_id":"q-en-mwaskom__seaborn-2846-520a6f7b25b824a33050e3e581605310047ccb402e830bcc0338ea0a4c028141","text":"relplot does not handle numpy-types for dimensional variables Test case: ! Note how it handles fine\\, but does not create two columns (or error\/warn in any useful way). handles this better: ! and solve the problem of initializing a from vector data differently. ~I do not remember if this ever worked in and is a regression (quite possibly not) and, if not, whether that was a failure of implementation or the result of a decision to punt on a particularly tricky issue. If the latter, it should at least give feedback about why it is not working.~ It looks like this never worked. the internal name used for here, , shows up in the y label, which I'm also not sure that we want. Note that there is no y axis label for the , although the internal name for the column variable is used because titles include that in their title template."}
{"_id":"q-en-mwaskom__seaborn-2848-ffc953df52e0b5c530945e4911c8daa98f1b70296ef67d5b3b6ee953df8094cb","text":"PairGrid errors with assigned in In seaborn version 0.9.0 I was able to use the following Code to plot scatterplots across a PairGrid with categorical hue. The reason I am not using the \"hue\" keyword in creating the PairGrid is, that I want one regression line (with regplot) and not one regression per hue-category. However, since I updated to searbon 0.11.1 the following Error message occurs: My further observations are: the error does not occur when using the \"hue\" keyword when creating PairGrid the error does not occur for numerical values for hue changing the dtype to \"categorical\" does not help Edit: I tried all versions between 0.9.0 and the current release (0.11.1) and the error only occurs in the current release. If I use 0.11.0, the plot seems to work."}
{"_id":"q-en-mwaskom__seaborn-2853-d6d5b35be679a2986ecda1e1954095e3382af80c35b9cb06e193be6e5d70a1ad","text":"xestimator bars now inherit scatterkws alpha x_estimator error bars were previously always opaque, but now inherit alpha parameter from scatterplot settings (if present), since the error bars replace the scatterplot."}
{"_id":"q-en-mwaskom__seaborn-2946-3eebdf880146900adf314ed29d1c50a0d6a5c8eeeb46bb65db8bd96bb6145018","text":"Regression: and raise when using a custom estimator This may be related to . According to (and ) documentation, the can be any callable that maps vector to scalar. However, the following example raises with on my Windows and Mac machines (and doesn't raise with ): <details<summaryException<\/summary<\/detailsIt does work, however, when changing the function to use the builtin func: The same error is raised when changing the example code to use , or when using a different numpy aggregator within the custom function."}
{"_id":"q-en-mwaskom__seaborn-2979-e4e38bcf37cf95378d8aa5bde5915f80030e816a1937105bd52536be03d4e3ad","text":"Visibility of internal axis labels is wrong with wrapped pair plot ! The top two subplots should have distinct x labels. Visibility of internal axis labels is wrong with wrapped pair plot ! The top two subplots should have distinct x labels."}
{"_id":"q-en-mwaskom__seaborn-2996-abae9420ee823174191b596978fa3d6644203eca72e4ba023b9705589752412a","text":"Rename layout(algo=) to layout(engine=) Matplotlib has settled on this term with the new method in 3.6 so might as well be consistent with them. The new API also ha some implications for how the parameter should be documented \/ typed."}
{"_id":"q-en-mwaskom__seaborn-3010-a6ddacdf5e8bbe4af2d6a285f4366606649f830bb10820438b51e0dabd4cd41d","text":"PolyFit is not robust to missing data <details<summaryTraceback<\/summary<\/details"}
{"_id":"q-en-mwaskom__seaborn-3069-4d3a971f45d5a3747737777f9b54320ccdc6e513a5a43fb4d31c726cdabf701c","text":"Nominal scale should be drawn the same way as categorical scales Three distinctive things happen on the categorical axis in seaborn's categorical plots: scale is drawn to +\/- 0.5 from the first and last tick, rather than using the normal margin logic grid is not shown, even when it otherwise would be with the active style on the y axis, the axis is inverted It probably makes sense to have scales (including inferred ones) do this too. Some comments on implementation: is actually trickier than you'd think; I may have posted an issue over in matplotlib about this at one point, or just discussed on their gitter. I believe the suggested approach is to add an invisible artist with sticky edges and set the margin to 0. Feels like a hack! I might have looked into setting the sticky edges on the spine artist at one point? straightforward to do in . Always a good idea? How do we defer to the theme if the user wants to force a grid? Should the grid be something that is set in the scale object itself straightforward to implement but I am not exactly sure where would be best."}
{"_id":"q-en-mwaskom__seaborn-3187-64faeae11cff0236c26128e7f3c0b23f593a6140ac11c71cccd1a9512936de68","text":"Wrong legend values of large ranges As of 0.12.1, legends describing large numbers that were created using with an offset are formatted without their multiplicative offset value. An example: The code creates the following plot: ! which is wrong because is in the order of 1E6. The issue also reproduces if you create the mentioned plot using . I believe the issue stems from not using the offset value of the used to generate the tick labels: Examining the code of suggests the issue also depends on the following rcParam settings: However, I did not test it. The offset value can be safely retrieved from all formatters and based on that it can be used to create the legend title and\/or labels."}
{"_id":"q-en-mwaskom__seaborn-3202-b1d0af642b52efd8e32fc30c148b0e5582331454c45db659a66bcb57a522cfe3","text":"Objects interface raises if faceting on partially-crossed row and column In the objects interface, one can facet two variables using rows and columns. When the faceted categories are not fully crossed, it raises: <details<summaryThe trace<\/summary<\/detailsI expect a behavior that is similar to , where the facets that contain no data are empty: !"}
{"_id":"q-en-mwaskom__seaborn-3216-9a143cfeb4bfcc0c854c27c01f7cedecdff4d44ffeecb301f76a8bbd448a42b4","text":"Figure title being removed by seaborn objects API when plotting on subfigures I recently came across an odd behaviour with the seaborn objects API when using subfigures. Here is a minimal example : which results in the title missing from the image : ! Commenting the results in the title reappearing. I have done a bit of digging and found that changing line 186 from the file from to seems to solve the issue. Is there a specific reason to why it fetches the parent figure currently, since Subfigure is supposed to be a drop-in replacement for Figure ? I also expect this will not have the intended behaviour if we deal with subfigures of subfigures."}
{"_id":"q-en-mwaskom__seaborn-3217-633e64a2e68bcf59d8d22a6a45732bf950233c2cddfbb13099bc92e34892e409","text":"Width computation after histogram slightly wrong with log scale Note the slight overlap here: ! It becomes nearly imperceptible with more bins: ! This is not about ; has it too: !"}
{"_id":"q-en-mwaskom__seaborn-3276-906723bdd9db9192aebcdc73a321da43d1775d11f2696da0bf2e2d2c9442da1f","text":"has different behaviour than without specifying the argument Hi, I'm noticing a strange behaviour when passing to a heatmap and I believe it's due to these lines: Specifically, if I use I get something like this: ! but when I use , and are lost: ! I'm happy to send a PR if this issue isn't addressed anywhere. Ștefan"}
{"_id":"q-en-mwaskom__seaborn-3394-bb38ee071aa03d32f7f20ef940b7c1484a86b0607314ceacd6a090317f885d44","text":"pd.NA reverses axis ordering When plotting with , axis ordering get reversed into descending. does not produce this issue NAs should be excluded without reversing axis order !"}
{"_id":"q-en-mwaskom__seaborn-3407-0f835b06e9d7741fbdc2fb941f8e344224132f2e412c3a5f3545cc7e17f2f483","text":"pairplot raises KeyError with MultiIndex DataFrame When trying to pairplot a MultiIndex DataFrame, raises a : MRE: Output: A workaround is to \"flatten\" the columns:"}
{"_id":"q-en-pallets__flask-4045-cf4bfa230c5c2e5d4e098928b54f4127f2fa09c88a9d521c7fe6d52fadb0aab7","text":"Raise error when blueprint name contains a dot This is required since every dot is now significant since blueprints can be nested. An error was already for endpoint names in 1.0, but should have been for this as well."}
{"_id":"q-en-pallets__flask-4160-0ca691f8259e7edbcf24e6e64c606ed44eba7f1467657835191573b1037b5159","text":"handle Decimal in json encoder The removal () decreased the flask encoding capabilities as the built-in doesn't cover cases like types. The solution seems to be: overwrite the flask app encoder with from , but this incorporates a problem for users that relies on both and as doesn't handle , while flask encoder does. The solution I found is to build a custom encoder that tests if the value is and gives it to , otherwise, handles it with the default flask app encoder. My suggestion is to incorporate a simple test in the flask encoder to add coverage, that would remove any residual dependency on . The str(decimal) was taken from: The custom encoder: The expected behavior is to work with both and as it used to work on Flask version 1.1.2 Environment: Python version: 3.8.10 Flask version: 2.0.1 and 1.1.2 Simplejson(Optional) version: 3.17.2"}
{"_id":"q-en-pallets__flask-4169-6db82c678547e1290dc5e51562328d175fb15674cae9a0c665d1114fce648322","text":"Exceptions are sometimes replaced with \"TypeError: exceptions must derive from BaseException\" I expected something about a RuntimeError. here worked for me Environment: Python version: 3.9.5 Flask version: 2.0.1"}
{"_id":"q-en-pallets__flask-4544-b9ad23f8e0d39bb9d96ae495b7f05626e40c9c2cde7da02463eb02d9313b5b38","text":"rejects When trying the development server, I find that works fine, but fails with: Isn't this somewhat counterintuitive? In ,  function insists the certificate file should have been specified, and it feels too restrictive. But I'm not familiar with Click to contribute a pretty PR... Environment: Python version: 3.9.10 Flask version: 2.0.3"}
{"_id":"q-en-pallets__flask-4575-5e3a0800d4375bc1a9f0f95b61fd5db8a37fe0e9e532bc71cbc351a114313d60","text":"Move to the app object Add a method to the app object. Similar to functions like , should look for a and call its method. This will allow applications to override the redirect behavior."}
{"_id":"q-en-pallets__flask-4935-cc5bf2f3f67bbff195c3d0ffcf70e5d965bdb5199e1cfed42ccf39defeb28455","text":"Nested blueprints are not respected when mounted on subdomains Hello, and thanks for all your work Nested blueprints work perfectly fine when using . However, when mounting the parent blueprint using a subdomain, the child routes are not accessible. The index route works as expected: So does the parent route in the subdomain: But the child responds with a 404: If we change the for when registering the blueprint however everything works as expected: This was surprising to me as I expected the same nesting to apply regardless of whether the parent is mounted using a subdomain or a URL prefix. Am I missing something? Environment: Python version: 3.10 Flask version: 2.2.2"}
{"_id":"q-en-pallets__flask-4992-b91dacac11b498f6b2fac8340bf9d5b43354862687220df7aa33adc38ad4b6ff","text":"Add a file mode parameter to flask.Config.fromfile() Python 3.11 introduced native TOML support with the package. This could work nicely with the method as an easy way to load TOML config files: However, takes an object readable in binary mode, while opens a file in text mode, resulting in this error: We can get around this with a more verbose expression, like loading from a file opened with the built-in function and passing the to : But adding a file mode parameter to would enable the use of a simpler expression. E.g.:"}
{"_id":"q-en-pallets__flask-5014-b232a6036fe57de5e8bda5e96cad77d4e4a85b55f2382b5f25142ee06057a93d","text":"Require a non-empty name for Blueprints Things do not work correctly if a Blueprint is given an empty name (e.g. ). It would be helpful if a was raised when trying to do that."}
{"_id":"q-en-pallets__flask-5063-1c16d5d85b944840dd2bfbf9f2642425b4e3f04d2060239129117c4808be2c28","text":"Flask routes to return domain\/sub-domains information Currently when checking flask routes it provides all routes but it is no way to see which routes are assigned to which subdomain. Default server name: SERVERNAME: '' Domains (sub-domains): Adding blueprints: app.registerblueprint(adminblueprint,urlprefix='',subdomain='admin') app.registerblueprint(testsubdomainblueprint,urlprefix='',subdomain='test') Feature request It will be good to see something like below (that will make more clear which route for which subdomain, because now need to go and check configuration). If it is not possible to fix routes, can you add or tell which method(s) should be used to get below information from flask?"}
{"_id":"q-en-psf__requests-1142-03c9ebf32372ec30f1ab0c83f10f5d921654ebd98d3b71a7419ca28abd404958","text":"is ALWAYS sending content length Hi, It seems like that always adds 'content-length' header to the request. I think that the right behavior is not to add this header automatically in GET requests or add the possibility to not send it. For example returns 503 for every get request that contains 'content-length' header. Thanks, Oren"}
{"_id":"q-en-psf__requests-1327-33363807c846f78a5774ad150212f83eabbec297d638a7e8dbae1b18b3514838","text":"Transport adapters don't work as advertised Having: We cannot currently reliably do:: We should, this is documented: Instead, because of the following might or might not work as expected: This is because there is already a default adapter for in the form of . Depending on the (seemingly random) order of keys in the dictionary, for some combinations of keys it will work, for others it won't. There are a couple. As usual, a compromise between backwards compatibility, clarity and performance. Change so that it sorts the keys every time it fetches an adapter. The advantage is that is still a vanilla dictionary, if you consider it a public API. Something like Django's which would sort the keys on insertion. The advantage is that is still a mapping and now is faster. The disadvantage is the maintenance cost of a custom data structure. Also, mild backwards incompatibility if someone somewhere used the fact that is exactly of the type (unlikely, isn't it?). Having and as the public API for adapters, we might just as well implement the sorting while mounting and just maintain as a list of tuples. The advantage is that it's the fastest solution with no additional maintenance burden. The disadvantage is backwards incompatibility. This is another design issue. It appears that sorting first from the longest key to shortest, and then alphabetically (for keys of the same length), is the way. What should we do about it? By the way, this test adapter in the example above ."}
{"_id":"q-en-psf__requests-1339-a93d0141fd81f65193c00fab0907f87e8e40f3561255925c6bd0e4415cfeae92","text":"CaseInsensitiveDict setitem faulty for differing-case keys I think I figured out the cause of the issue described here: It appears that if one isn't careful to use a common case convention when assigning a header after it has been assigned before, the header will fail to work. The request will use the old header value... I think I know why, too. Here's an example: The output: Notice how in both passes, only the first assignment actually \"works\". It stores the data for all three assignments, but will only return the value set by the initial key. I believe the only change necessary would be to the setitem method on CaseInsensitiveDict: Broken: Fixed: When I use the corrected module, the CaseInsensitiveDict behaves as I believe it should: Note that the value for that key is the last one assigned, instead of the first. If needed, I can put together a pull request for this, but it might take some time since I've never done it before :)"}
{"_id":"q-en-psf__requests-1537-251e938b2826f2419848592e01a017314e863a06823d877fe1a39a3c37cd252d","text":"multipart\/form-data and datetime data I raise an bug that you already fix in the past on this issue : or I tried the same methodology with that code : With the 1.2.0 version, no error is raised. With 1.2.3 version, I have that traceback : My original problem was with a python datetime in the data dict Thanks,"}
{"_id":"q-en-psf__requests-1635-539406d772f44d672f84e056da07e4ed32dab3df27d77308dcef67a98a3764a4","text":"Cookies not persisted when set via functional API. Cookies set as part of a call to (or any of the top level methods that call it) are not persisted, including on redirects. Expected behaviour: Actual behaviour: And, a super extra bonus bug: even though we definitely sent the cookie on the first request."}
{"_id":"q-en-psf__requests-1657-d7c25f808305571ef7088bf566171c7ed197d0d462c72c9b04d093a2d792abc5","text":"Session hooks broken Request hooks are being with session hooks; since both hook dicts have a list as the value, one simply overwrites the other."}
{"_id":"q-en-psf__requests-1689-54421f1ccf08863601a31f715958697200ce5d5d5dad7a345b4a145ebb0ba99b","text":"Problem POST'ing png file because of UnicodeError Here is the code I'm using: This raises the error: This problem is caused by the fact that the content-length header is actually a unicode object. When the actual body of the request is being constructed, python attempts to coerce the entire request into unicode resulting in the decode error. After tracing it, the cause is the following lines: where is declared in"}
{"_id":"q-en-psf__requests-1713-99eb739f4b55358943068c306ec565ae803633ae61a1f755662179fa034d2887","text":"Regression 2.0.1: Using MozillaCookieJar does not work Could not find an issue raised for this, not sure if this was an expected change either. This is reproducible on master. Existing code fails on update to . The cause seems to be triggered by the change at The parameter expects either or . Treating as a dict triggers the error in this instance. The following code highlights the issue: The following is thrown:"}
{"_id":"q-en-psf__requests-1724-d3782f6a920df2152bcc3398f01ac519eb973d22b538c04fc31bee75468378f9","text":"Unicode method names cause UnicodeDecodeError for some requests in Python 2.7.2 The following example works fine: But the following example (using instead of ) produces a UnicodeDecodeError: My guess is that is infecting the header with unicode when it should be a string. This is because is simply: My requests version is 1.2.3, but I see the same being used in the current source. Unicode method names cause UnicodeDecodeError for some requests in Python 2.7.2 The following example works fine: But the following example (using instead of ) produces a UnicodeDecodeError: My guess is that is infecting the header with unicode when it should be a string. This is because is simply: My requests version is 1.2.3, but I see the same being used in the current source. Unicode method names cause UnicodeDecodeError for some requests in Python 2.7.2 The following example works fine: But the following example (using instead of ) produces a UnicodeDecodeError: My guess is that is infecting the header with unicode when it should be a string. This is because is simply: My requests version is 1.2.3, but I see the same being used in the current source."}
{"_id":"q-en-psf__requests-1766-31776b23a770a485682234f62624175e71488620a8999c3eb698aff98e7809b0","text":"quote qop options in Digest Auth Based on RFC2617 (), the value of 'qop-options' directive should be quoted with double quotes: curl comamnd-line tool also appends these quotes. You can see this by . Unfortunately, some minor server-side implementations seem to be sensitive on this difference."}
{"_id":"q-en-psf__requests-1768-ec10f08f3525975c819f91fbe00bf525bc08dfb39ab1da12367c6d7754231f36","text":"HTTP Basic Auth credentials extracted from URL are %-encoded I was relying on credential auto-extraction from the hostname [1] to perform HTTP Basic Auth but I was getting \"401 Unauthorized\": the spaces in the password were substituted by . Manually extracting them with and passing them to the parameter as a tuple works. [1] (second paragraph)"}
{"_id":"q-en-psf__requests-1776-fcaa7483121cdf26249ce6d07c29df07560cabf1aedae9e6193c3ad5fe696e25","text":"Request cookies should not be persisted to session After the fix for , cookies sent with a request are now incorrectly persisted to the session. Specifically, problem lies here: Removing that breaks the test case for though, still investigating a solution."}
{"_id":"q-en-psf__requests-1888-42373db936a5da5d23f2b3ed911a5cfee92afc00688328b85f639067b2f78232","text":"301 redirect broken with latest pyopenssl\/SNI With the latest pyopenssl on Windows 64bit: I get an exception raised when ing a response to a HTTPS request. I see that after the redirect is received the returned URL is . Then requests passes the response to which uses the url to make a new request. This leads to a Unicode string being passed to urllib3 and eventually pyopenssl. And because in pyopenssl they now check that the data is of type bytes, an exception is thrown. I Wrote this test: and this is the result of"}
{"_id":"q-en-psf__requests-1921-0005cee1849ae3191a431e0fc7b0bdd5d70a6416d4f9fd06d2170ac65db0dd41","text":"Removing a default header of a session say that you can prevent sending a session header by setting the headers value to None in the method's arguments. You would expect (as ) that this would work for session's default headers, too: What happens is that \"None\" gets sent as the value of header. For the reference, here is a way that works:"}
{"_id":"q-en-psf__requests-1944-77044d60b6f24ac609daad2f7013b660f74966735af0fc26908b4d5421836072","text":"Why decode the response body of a redirect? Requests fails on the URL , which is a 301 redirect to . The issue seems to be that the server's initial 301 response has a header falsely claiming that the response body (a simple HTML page) is gzipped, when it's actually uncompressed. When resolving redirects, Requests does (in ): which attempts to decode One could legitimately say this is the server's problem. However, conceptually, why decode the response body of a redirect, which won't get returned? Other programs (Chromium, Firefox, ) don't do this. For example, gives an error, as expected, when not following redirects: whereas it works if you add the flag (follow redirects)."}
{"_id":"q-en-psf__requests-1962-9fdcb5fda9e58a9c6870b966d3b560a8a57aa0a65a8474305b650c506fcd072e","text":"request.history can be either a list or a tuple IMHO r.history should always be a list for least surprise. In some cases, it is returned as a tuple: Thanks!"}
{"_id":"q-en-psf__requests-1963-702f58551cc5ba0c588c7227b0dc33755e741d7c97e8c0c5b879a1d94e8cca45","text":"copies the original request for all subsequent requests, can cause incorrect method selection Consider the following redirection chain: The intermediate 303 See Other has caused the POST to be converted to a GET. The subsequent 307 should preserve the GET. However, because starts each iteration by copying the original request object, Requests will issue a POST!"}
{"_id":"q-en-psf__requests-2148-67f4cded64a369af8e9fe2d77c207f321f7e30c433b0a52b6344c102ac4998d8","text":"exception not caught\/wrapped in a requests exception (ConnectionError perhaps?) I just noticed a case where I had a socket reset on me, and was raised to me as a raw socket error as opposed to something like a requests.exceptions.ConnectionError: Not sure if this is by accident or design... in general, I guess I'd expect a requests exception when using requests, but I can start looking for socket errors and the like as well."}
{"_id":"q-en-psf__requests-2153-339d67936d634e33366419e2cb7f58de2fda562f30c4b8e92d4a4c102c1d5ca5","text":"Uncaught socket.timeout during post Hallo requests devs (and thanks for an awesome lib) During a specific I most of the time get a (on timeouts) but I also sometimes get a exception. Since there is a requests exception for it, I assume that the socket exception was supposed to be caught and the requests one raise instead. The full stack trace is: The development is for a plugin for the music service Wimp for the project, which means that I could post code to reproduce, but you will not be able to run it without a Sonos speaker and a Wimp subscription. I understand if this difficulty to reproduce may mean that you cannot work with this issue. Thanks in advance Kenneth"}
{"_id":"q-en-psf__requests-2317-a4603ba8365f1b421fc67550a0bfb7946bb22dc8b6a223830a472c7e0816ba6f","text":"method = builtinstr(method) problem In is a command: method = builtinstr(method) Converts method from b’GET’ to \"b'GET’\" Which is the literal string, no longer a binary string. When requests tries to use the method \"b'GET’”, it gets a 404 Not Found response. I am using python3.4 and python-neutronclient (2.3.9) with requests (2.4.3). neutronclient is broken because it uses this \"args = utils.safeencodelist(args)\" command which converts all the values to binary string, including method. I'm not sure if this is a bug with neutronclient or a bug with requests, but I'm starting here. Seems if requests handled the method value being a binary string, we wouldn't have any problem. Also, I tried in python2.6 and this bug doesn't exist there. Some difference between 2.6 and 3.4 makes this not work right."}
{"_id":"q-en-psf__requests-2393-336924f7e3c05f28ae3421dac43ef506cb91b6021418dc443477995beaa1f38d","text":"Requests unable to follow\/retrieve links with percent in url A simple (url) doesn't work for the following: !;&placement=%pplacement=!;"}
{"_id":"q-en-psf__requests-2413-bef892fabac7ae07fb2481fac553285d3c7cd5aad2a4e43fdc72d20f838f8667","text":"Requests 2.5.1 doesn't recognize unicode filenames for uploads After merge of , to allow filenames to be types, unicode filenames are no longer recognized under Python 2. This checks that the filename is a , which has different behaviour on Python 2 and Python 3: In , is defines as , which is non-unicode in Python 2 and unicode in Python 3. Perhaps the check should be against basestring, or is this change in behaviour intended?"}
{"_id":"q-en-psf__requests-2466-195265fbc902fa644973281f8aaed422f229593d250a61b67f828139a7004803","text":"Issue with pyinstaller Can't seem to wrap my head around this. having a problem compiling a windows exe in python 2.7.9 that uses the requests library and can't find anything on google about the specific error. My script runs fine from the interpreter but when i use pyinstaller, i get :ImportError: No module named '' I can also compile windows executables that don't use requests just fine."}
{"_id":"q-en-psf__requests-2617-5e9dd7e3fccaa758dee0a73f9461d6f9dee8984ee9c1fb418d41d99cddd569b0","text":"Prepared requests containing binary files will not send when unicodeliterals is imported The above program works perfectly in python3, and in python2 when is not imported. If the request isn't prepared it works without a problem unfortunately, I require both prepared requests and in my project. The exception raised is: Traceback (most recent call last): File \"\", line 15, in (main()) File \"\", line 12, in main requests.Session().send(prepared) File \"\", line 573, in send r = (request, kwargs) File \"\", line 370, in send timeout=timeout File \"\", line 544, in urlopen body=body, headers=headers) File \"\", line 349, in makerequest conn.request(method, url, httplibrequestkw) File \"\", line 1001, in request self.sendrequest(method, url, body, headers) File \"\", line 1035, in sendrequest self.endheaders(body) File \"\", line 997, in endheaders self.sendoutput(messagebody) File \"\", line 848, in sendoutput msg += message_body UnicodeDecodeError: 'ascii' codec can't decode byte 0xff in position 109: ordinal not in range(128)"}
{"_id":"q-en-psf__requests-2674-a2d42d7eb3920000415f368570104f820da035c8a8e684819db4ed4b5f7e28f2","text":"urllib3 exceptions passing through requests API I don't know if it's a design goal of requests to hide urllib3's exceptions and wrap them around requests.exceptions types. (If it's not IMHO it should be, but that's another discussion) If it is, I have at least two of them passing through that I have to catch in addition to requests' exceptions. They are requests.packages.urllib3.exceptions.DecodeError and requests.packages.urllib3.exceptions.TimeoutError (this one I get when a proxy timeouts) Thanks!"}
{"_id":"q-en-psf__requests-2754-c75e705d21366647a2f39f67f2dbde8c6b7d3bd5408e342ea8df63ecc0034986","text":".htaccesss redirect to non ASCII folder does not work Hello, I have the following setup on a shared hoster: Apache 2.2.15 A Japanese language . (.minna; xn--q9jyb4c) IDN domain. A blog which is in the subfolder (blog) A redirect in the .htaccess file like this: So I usually open the domain http:\/\/test. and the server redirects to http:\/\/test.\/. This works fine in Firefox etc. With requests, I get the following error (Python 3.4 with Requests 2.7.0 on a Japanese Ubuntu 15.04): So I guess the request lib gets a redirect from a server with Japanese characters, but then fails to convert the characters correctly. If I do directly it works, only the redirect does not."}
{"_id":"q-en-psf__requests-2821-6d48e79d8968c46b7ad10f631a92f9feaa6e27149bf1d797ea4f5466c1b699a4","text":"TypeError after 2.7.0 -2.8.0 upgrade (cannot make memory view...) I'm running into this traceback after upgrading to 2.8.0 from 2.7.0: The problem goes away after downgrading to 2.7.0. A full traceback can be found . Here are the versions of relevant packages: And a full list of packages: Any ideas?"}
{"_id":"q-en-psf__requests-2873-28c0ae898f8dce57bc4f168743c7733c3c6d67958e90a08337b2f87f147f5d40","text":"Post request hangs in certain cases when body is a StringIO This is related to a report for the . The following hangs: After a cursory look, it looks like the request isn't fully formed so the server doesn't attempt to send a response which leaves the client hanging. If we call , this works. A bit more counterintuitively, this also works:"}
{"_id":"q-en-psf__requests-2931-9735327c4c7cd8655ff3dee1197b16dad0552cc489310b008364b24310a2b2df","text":"Request with binary payload fails due to calling tonativestring Introduced with This works with 2.8.1, but not with 2.9."}
{"_id":"q-en-psf__requests-3362-9255c5ba349395ca7821aa51d151e0c6d220e55ae795fa9e2da54f9d763b5754","text":"Uncertain about content\/text vs itercontent(decodeunicode=True\/False) When requesting an application\/json document, I'm seeing returning bytes, whereas returns unicode. My understanding was that both should return a unicode object. In essence, I thought \"itercontent\" was equivalent to \"itertext\" when decode_unicode was True. Have I misunderstood something? I can provide an example if needed. For reference, I'm using python 3.5.1 and requests 2.10.0. Thanks!"}
{"_id":"q-en-psf__requests-3718-d49a1ba83216efa214360b5590fa3830ce9cd3492fc371a906a15a65c8b1025c","text":"AttributeError: 'NoneType' object has no attribute 'read' Hello :) After a recent upgrade for our project to 2.12.1 we encounter an exception in our test suites which seems to be caused by . Build: Relevant part: happens on Windows and Linux. Thanks in advance :)"}
{"_id":"q-en-psf__requests-3738-d14c222b2a870294b10d8cd8dfc20a8a2e29d03e04ef19e0713540e5f9d12193","text":"Requests v2.12.2 does not add parameters to URLs with schemes it does not understand. Follows from . When we patched to ignore all unrecognised schemes instead of just those that didn't begin , we stopped handling parameters for those URLs. This may break some more benign uses such as , which wanted to add parameters to their URLs. While this is inline with our intended policies (we do not understand URLs that have schemes we don't know anything about), this, along with the IDNA behaviour in v2.12.{0,1} will probably have broken a whole bunch of people using these non-standard URL schemes. We should consider whether Session objects should have a registry of places to look for URL preparation based on scheme. This will allow people to opt-in to the HTTP-like processing of URLs, as well as to register their own."}
{"_id":"q-en-psf__requests-4106-0c64d7fd30e7fd2dce02436ace3d0aea580b1d6b443a5a403db841faeb7b7dbc","text":"AttributeError: module 'requests.packages' has no attribute 'urllib3' This seems to have broken requests.packages. requests.packages.urllib3 to be the urllib3 package AttributeError: module 'requests.packages' has no attribute 'urllib3' $ python -m"}
{"_id":"q-en-psf__requests-4356-d870db5f20fa49ace30d0c5605b7fa61954446a6311d3f0deabefe32c1e768c7","text":"Misleading exception with invalid protocol in proxy variable When the value of or variable(s) accidentally miss one '\/' in the protocol, a traceback is thrown to the user which doesn't pin point that the issue is with the proxy configuration. A better exception An exception which doesn't pin point exactly what went wrong. I am not sure what is the correct place to fix this. Should the fix\/check be in requests, urllib3, or urlparse?"}
{"_id":"q-en-psf__requests-4718-57339d8b0f882f90b184dc7ce4e73bc45e644d708e35f4096cc6546e8786a4ed","text":"Should Authorization header be cleared in https -http redirect? This may be considered intentional behaviour (in which case feel free to close this), but if a request is made to an https endpoint with authorization and it redirects to http on the same host, the Authorization header is not stripped and will be exposed on the wire. rebuild_auth would strip the Authorization header if the scheme is changed from https to http. The credentials that were intended to be sent over TLS were transmitted in plaintext with the redirected request. Run an HTTPS server on localhost:4443 that replies with a 302 redirect to , and a plain HTTP server (or netcat) on localhost:8000. Then run The basic auth credentials are sent in plaintext to (the is just because I had a self-signed cert). Here's the code I used for the SSL server:"}
{"_id":"q-en-psf__requests-5087-301829d186d5268027dd96bd04dacc2edeba189acc30fce44f4c41314b1c96d3","text":"Accessing response.content twice removes forgets read error I had a hard debugging time today because an error in the response stream is only reported when accessing for the first time. This is especially irritating when running code in a debugger. If accessing the first time raises an exception I would expect that accessing again would also raise an exception (ideally the same). Instead after raising on the first get, getting again returns an empty string. Here is a patch with a new test case for this: . Basically, it boils down to this: Output of my test case: $ python -m Edit: Oops, I used which actually called into system python 2.7. Here comes the real data: Thanks for looking into this!"}
{"_id":"q-en-psf__requests-5414-b672ac79f5d9e28024abe139152ec9a30078b1b2f13bf8001170d331a50853d0","text":"Getting raises UnicodeError Attempting to get e.g. results in a . It seems like the intention so far has been to raise instead (see e.g. ). I see there was some hesitation in fixing a similar issue () and would like to add that even catching the error just to rethrow as a requests exception would be beneficial. Based on PR : $ python -m"}
{"_id":"q-en-psf__requests-6028-1706a5892b2e8b8ba3f5c25a984f42a1403eb27c697ac7063928d4cf1f3fc30b","text":"Proxy authentication bug <!-- Summary. --When using proxies in python 3.8.12, I get an error 407. Using any other version of python works fine. I am assuming it could be to do with this <!-- What you expected. --I should get a status of 200. <!-- What happened instead. --I get a status code of 407."}
{"_id":"q-en-psf__requests-774-e0615a652cc6e1addd24921cf2afed8fe92c9e3149ca6beec5cd5758ec358bc4","text":"Catch UnicodeError coming from Here's a Python 2.6 Requests 0.13.1 traceback. Looks like the call to needs a try\/except."}
{"_id":"q-en-psf__requests-863-770d3124a68fd3dcb30d977abbc56136c1d75b6d7cadb59b0114ac1b757e2657","text":"Allow lists in the dict values of the hooks argument Currently the Request class has a .registerhook() method but it parses the dictionary it expects from it's hooks argument weirdly: the argument can only specify one hook function per hook. If you pass in a list of hook functions per hook the code in () will wrap the list in a list which then fails when the hooks are consumed (since a list is not callable). This is especially annoying since you can not use multiple hooks from a session. The only way to get multiple hooks now is to create the request object without sending it, then call .registerhook() multiple times and then finally call .send(). This would all be much easier if () parsed the hooks parameter in a way that it accepts lists as it's values."}
{"_id":"q-en-pydata__xarray-2905-9748a110c2cd9559e79ccee3c0ac780c4ea3b168415a2aa9355c8f86469df98e","text":"Variable.setitem coercing types on objects with a values property The current behavior prevents storing objects inside arrays of even when only performing non-broadcasted assignments if the RHS has a property. Many libraries produce objects with a property that gets coerced as a result. The use case I had in prior versions was to store instances from the curve fitting library , when fitting had be performed over an axis of a or . Ideally: Breaking changed introduced going from -as a result of , namely the change on line <detailsINSTALLED VERSIONS commit: None python: python-bits: 64 OS: Darwin OS-release: 16.7.0 machine: x8664 processor: i386 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 xarray: 0.10.1 pandas: 0.20.3 numpy: 1.13.1 scipy: 0.19.1 netCDF4: 1.3.0 h5netcdf: None h5py: 2.7.0 Nio: None zarr: None bottleneck: None cyordereddict: None dask: 0.15.2 distributed: None matplotlib: 2.0.2 cartopy: None seaborn: 0.8.1 setuptools: 38.4.0 pip: 9.0.1 conda: None pytest: 3.3.2 IPython: 6.1.0 sphinx: None <\/detailsThank you for your help! If I can be brought to better understand any constraints to adjacent issues, I can consider drafting a fix for this."}
{"_id":"q-en-pydata__xarray-2922-6bcdb6d14fa8e3282ec56ef6ec9707906b48a9ad1e22763b7e50f008fe493596","text":"add average function It would be nice to be able to do to compute weighted averages (e.g. for geo data). Of course this would require the axes to be in a predictable order. Or to give a weight per dimension..."}
{"_id":"q-en-pydata__xarray-3095-79f4ac0d246bc43a425f0a8ad6c997d21f9f4a6554c7e3a4d9dcefc0819410cb","text":"REGRESSION: copy(deep=True) casts unicode indices to object (deep=True) and (deep=True\/False) accidentally cast IndexVariable's with dtype='<U*' to object. Same applies to () and copy.deepcopy(). This is a regression in xarray = 0.12.2. xarray 0.12.1 and earlier are unaffected."}
{"_id":"q-en-pydata__xarray-3114-001ecd9596a756abc65044025a995433b77e79b4abb56da26601c095be30cd66","text":"expanddims() modifies to write only, upon manually reverting this flag back, attempting to set a single inner value using .loc will instead set all of the inner array values I am using the newly updated expanddims API that was recently updated with this PR . However the flag setting behaviour can also be observed using the old API syntax. When applying the operation 'expanddims({'z': 3}, -1)' on a DataArray the underlying Numpy array flags are changed. 'CCONTIGUOUS' is set to False, and 'WRITEABLE' is set to False, and 'OWNDATA' is set to False. Upon changing 'WRITEABLE' back to True, when I try to set a single value in the DataArray using the '.loc' operator it will instead set all the values in that selected inner array. I am new to Xarray so I can't be entirely sure if this expected behaviour. Regardless I would expect that adding a new dimension to the array would not make that array 'read-only'. I would also not expect the '.loc' method to work differently to how it would otherwise. It's also not congruent with the Numpy 'expanddims' operation. Because when I call the operation 'np.expanddims(nparr, axis=-1)' the 'CCONTIGUOUS ' and 'WRITEABLE ' flags will not be modified. Here is a similar flow of operations that demonstrates the behaviour I would expect from the DataArray after applying 'expanddims': <detailsINSTALLED VERSIONS commit: None python: 3.6.7 (default, Dec 29 2018, 12:05:36) [GCC 4.2.1 Compatible Apple LLVM 10.0.0 (clang-1000.11.45.5)] python-bits: 64 OS: Darwin OS-release: 18.2.0 machine: x8664 processor: i386 byteorder: little LCALL: None LANG: enAU.UTF-8 LOCALE: enAU.UTF-8 libhdf5: 1.10.2 libnetcdf: 4.4.1.1 xarray: 0.12.1 pandas: 0.24.2 numpy: 1.16.2 scipy: 1.2.1 netCDF4: 1.5.0 pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: 1.0.3.4 nctime_axis: None PseudonetCDF: None rasterio: None cfgrib: 0.9.6.1.post1 iris: None bottleneck: None dask: None distributed: None matplotlib: 3.0.3 cartopy: 0.17.0 seaborn: None setuptools: 39.0.1 pip: 10.0.1 conda: None pytest: None IPython: None sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-3151-ef20860c0157dd8bee0ad8e47feac19617e3ae2a84852ce4a3d1ab4b939e598f","text":"xr.combinebycoords raises ValueError if identical coordinates are non-monotonic <!-- In order for the maintainers to efficiently understand and prioritize issues, we ask you post a \"Minimal, Complete and Verifiable Example\" (MCVE): --should return without error. Running the example with raises an error: The documentation for says that \"Non-coordinate dimensions will be ignored, as will any coordinate dimensions which do not vary between each dataset\". This is not the case with the current implementation, since identical coordinate dimensions are still required to be monotonic. <detailsINSTALLED VERSIONS commit: None python: 3.7.1 (v3.7.1:, Oct 20 2018, 14:57:15) [MSC v.1915 64 bit (AMD64)] python-bits: 64 OS: Windows OS-release: 10 machine: AMD64 processor: Intel64 Family 6 Model 94 Stepping 3, GenuineIntel byteorder: little LCALL: None LANG: None LOCALE: None.None libhdf5: None libnetcdf: None xarray: 0.12.3 pandas: 0.24.2 numpy: 1.16.4 scipy: 1.3.0 netCDF4: None pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: None nctime_axis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: None dask: None distributed: None matplotlib: 3.1.1 cartopy: None seaborn: 0.9.0 numbagg: None setuptools: 39.0.1 pip: 10.0.1 conda: None pytest: None IPython: 7.1.1 sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-3156-4162fe0ad6f633e294460b5baba6838d213d712a246b641e18fd98c0b025458a","text":"groupby on empty DataArray raises StopIteration This seems similar to and so apologies if it is a duplicate, but I have a minimal example where it happens on an empty DataArray: Using groupby on an empty DataArray or Dataset raises . It should raise a more meaningful error. In particular, I had this issue in a function I was calling inside of a generator, so the StopIteration just broke out of the generator and it took some digging to figure out what was going wrong in my code. <detailsINSTALLED VERSIONS commit: None python: python-bits: 64 OS: Linux OS-release: 3.10.0-862.14.4.el7.x8664 machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.utf8 LOCALE: enUS.UTF-8 xarray: 0.10.7 pandas: 0.23.0 numpy: 1.14.3 scipy: 1.1.0 netCDF4: 1.4.0 h5netcdf: 0.6.1 h5py: 2.7.1 Nio: None zarr: None bottleneck: 1.2.1 cyordereddict: None dask: 0.17.5 distributed: 1.21.8 matplotlib: 2.2.2 cartopy: None seaborn: 0.8.1 setuptools: 39.1.0 pip: 10.0.1 conda: None pytest: 3.5.1 IPython: 6.4.0 sphinx: 1.7.4 <\/details"}
{"_id":"q-en-pydata__xarray-3159-cfd827f8ea1b65dcb23852b1440f7cf4a45dab933293bd8d82f6078a73dcba1f","text":"Allow passing a default value (instead of ndarray) for data argument for DataArray Hi, First of all, thanks a lot for the amazing module. It seems when I create a DataArray, I have to pass a numpy.ndarray with a correct size for the argument. It works well when I already have some data, but sometimes I want to create an \"empty\" DataArray with known coordinates and fill up the data later. For these cases, it would be great if xarray allows passing just a value for the argument, and fill up all the elements of the array with the value. For example, with pandas, I can do: and the resulting would be: Thanks a lot!"}
{"_id":"q-en-pydata__xarray-3305-fa21d9b53b15aa8698c2cf447e817b2e3aaf29ea31ac166a1c96340b305c5578","text":"DataArray.quantile does not honor <!-- In order for the maintainers to efficiently understand and prioritize issues, we ask you post a \"Minimal, Complete and Verifiable Example\" (MCVE): --returns <detailsINSTALLED VERSIONS commit: python: 3.6.8 (default, Dec 30 2018, 01:22:34) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 4.15.0-60-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enCA.UTF-8 LOCALE: enCA.UTF-8 libhdf5: 1.10.2 libnetcdf: 4.6.1 xarray: pandas: 0.23.4 numpy: 1.16.1 scipy: 1.1.0 netCDF4: 1.3.1 pydap: installed h5netcdf: None h5py: None Nio: None zarr: None cftime: 1.0.3.4 nctimeaxis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.2.1 dask: 0.19.0 distributed: 1.23.0 matplotlib: 3.0.2 cartopy: 0.17.0 seaborn: None numbagg: None setuptools: 41.0.0 pip: 9.0.1 conda: None pytest: 4.4.0 IPython: 7.0.1 sphinx: 1.7.1 <\/details"}
{"_id":"q-en-pydata__xarray-3364-bda68187262144c31e8cefb26438ac872deb3c2e856fe2d3709f4ce7aa98508a","text":"Ignore missing variables when concatenating datasets? Several users ( now myself) have wondered about how to concatenate xray Datasets with different variables. With the current , you need to awkwardly create dummy variables filled with in datasets that don't have them (or drop mismatched variables entirely). Neither of these are great options -- should have an option (the default?) to take care of this for the user. This would also be more consistent with , which takes a more relaxed approach to matching dataframes with different variables (it does an outer join)."}
{"_id":"q-en-pydata__xarray-3406-9d27646e662637c7cca3e13d18cc81808fd5179240fb0d067ac82b47fa7c6e41","text":"groupby().apply() on variable with NaNs raises IndexError This results in an IndexError. This is mildly confusing, it took me a little while to figure out the NaN's were to blame. I'm guessing the NaN doesn't get filtered out everywhere. The traceback: My assumption was that it would throw out the values that fall within the NaN group, like: Out: <details<\/details"}
{"_id":"q-en-pydata__xarray-3520-a9db2851fb5c92602d10cb84d64cfa794baecfab7e9066d46058dfa9f73fbdbd","text":"selection from MultiIndex does not work properly Should select the array <detailsINSTALLED VERSIONS commit: None python: 3.6.8 (default, Dec 30 2018, 01:22:34) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 3.10.0-957.10.1.el7.x8664 machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 libhdf5: 1.10.4 libnetcdf: 4.6.1 xarray: 0.14.0 pandas: 0.24.2 numpy: 1.15.4 scipy: 1.2.1 netCDF4: 1.4.2 pydap: None h5netcdf: None h5py: 2.9.0 Nio: None zarr: None cftime: 1.0.3.4 nctimeaxis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.2.1 dask: None distributed: None matplotlib: 3.0.2 cartopy: None seaborn: 0.9.0 numbagg: None setuptools: 40.8.0 pip: 19.0.3 conda: None pytest: 5.0.0 IPython: 7.3.0 sphinx: None <\/detailsSorry for being quiet for a long time. I hope I could send a fix for this in a few days..."}
{"_id":"q-en-pydata__xarray-3527-5ddc0dcd0c711bf32a009a4fa35ab41ad8b2415f4586144e3131abf70b9754e6","text":"DatasetGroupBy does not implement quantile The docs claim works on grouped datasets, but that does not seem to be the case: this was found while trying to silence the nit-picky sphinx warnings in"}
{"_id":"q-en-pydata__xarray-3635-cca6d3068fe5376e9a308ba1dc3454f79e18eac94d82c543c8c2df204ce18edb","text":"\"ValueError: Percentiles must be in the range [0, 100]\" <!-- In order for the maintainers to efficiently understand and prioritize issues, we ask you post a \"Minimal, Complete and Verifiable Example\" (MCVE): --By wrapping (xref: ) we also get the numpy error. However, the error message is wrong as xarray needs it to be in 0..1. BTW: thanks for , makes my life easier! Edit: uses internally. <detailsINSTALLED VERSIONS commit: None python: 3.7.3 (default, Jul 1 2019, 21:52:21) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 4.12.14-lp151.28.36-default machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enGB.UTF-8 LOCALE: enUS.UTF-8 libhdf5: 1.10.5 libnetcdf: 4.7.1 xarray: 0.14.1+28.gf2b2f9f6 (current master) pandas: 0.25.2 numpy: 1.17.3 scipy: 1.3.1 netCDF4: 1.5.3 pydap: None h5netcdf: 0.7.4 h5py: 2.10.0 Nio: None zarr: None cftime: 1.0.4.2 nctime_axis: 1.2.0 PseudoNetCDF: None rasterio: 1.1.1 cfgrib: None iris: None bottleneck: 1.2.1 dask: 2.6.0 distributed: 2.6.0 matplotlib: 3.1.2 cartopy: 0.17.0 seaborn: 0.9.0 numbagg: None setuptools: 41.4.0 pip: 19.3.1 conda: None pytest: 5.2.2 IPython: 7.9.0 sphinx: 2.2.1 <\/details"}
{"_id":"q-en-pydata__xarray-3637-513927ab1fa6ef6eeabb5d47a5d6d8bca2fba15adfab586375edff8e99af5980","text":"Confusing error message when attribute not equal during concat If two or more s are concatentated and they have list attributes that are not equal an exception is raised about arrays not being truth values. I guess the expected result would be that the list attribute is not included in the resulting DataArray's attributes. <details<\/detailsconcat fails with attrs that are dictionaries with ndarrays Fails with the following error: This is a problem because the following code actually executes properly Equivalence for an array within a nested dictionary as an attribute is evaluated differently than an array attribute, which is non-intuitive. This bug is related to but is additionally pointing out a difference in evaluation for more complex attributes. The output of the code sample should concatenate successfully with the nested dictionary attribute, or a more easily interpretable error should be thrown telling me I'm dumb for using dictionaries in attributes. (See ) <detailsINSTALLED VERSIONS commit: None python: python-bits: 64 OS: Linux OS-release: 4.15.0-23-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 xarray: 0.10.9 pandas: 0.23.4 numpy: 1.15.3 scipy: 1.1.0 netCDF4: None h5netcdf: None h5py: None Nio: None zarr: None cftime: None PseudonetCDF: None rasterio: None iris: None bottleneck: None cyordereddict: None dask: None distributed: None matplotlib: 3.0.0 cartopy: None seaborn: None setuptools: 40.4.3 pip: 9.0.1 conda: None pytest: None IPython: 7.0.1 sphinx: None <\/detailsconcat fails with attrs that are dictionaries with ndarrays Fails with the following error: This is a problem because the following code actually executes properly Equivalence for an array within a nested dictionary as an attribute is evaluated differently than an array attribute, which is non-intuitive. This bug is related to but is additionally pointing out a difference in evaluation for more complex attributes. The output of the code sample should concatenate successfully with the nested dictionary attribute, or a more easily interpretable error should be thrown telling me I'm dumb for using dictionaries in attributes. (See ) <detailsINSTALLED VERSIONS commit: None python: python-bits: 64 OS: Linux OS-release: 4.15.0-23-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 xarray: 0.10.9 pandas: 0.23.4 numpy: 1.15.3 scipy: 1.1.0 netCDF4: None h5netcdf: None h5py: None Nio: None zarr: None cftime: None PseudonetCDF: None rasterio: None iris: None bottleneck: None cyordereddict: None dask: None distributed: None matplotlib: 3.0.0 cartopy: None seaborn: None setuptools: 40.4.3 pip: 9.0.1 conda: None pytest: None IPython: 7.0.1 sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-3649-bd57823bdddea291e489462a84b91f3b4f8ca390138596f67973dc8c3ba4c4b0","text":"combinebycoords should allow for missing panels in hypercube Currently, it throws the following error: This is b\/c calls , which mandates that the passed datasets form a complete hypercube. This check functiono also serves the purpose of mandating that the dimension depths are the same. Could we pull that part out as a separate function and, for , only call this first part but NOT mandate that the hypercube is complete? The expected behavior, in my mind, should be to simply replace the missing tiles of the hypercube with . I'll file a PR to this effect and welcome comments. <detailsINSTALLED VERSIONS commit: None python: 3.7.3 (default, Dec 6 2019, 08:54:18) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 4.14.150+ machine: x8664 processor: x8664 byteorder: little LCALL: enUS.UTF-8 LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 libhdf5: 1.10.5 libnetcdf: 4.7.1 xarray: 0.14.1 pandas: 0.25.3 numpy: 1.17.3 scipy: 1.3.2 netCDF4: 1.5.3 pydap: None h5netcdf: 0.7.4 h5py: 2.10.0 Nio: None zarr: 2.3.2 cftime: 1.0.4.2 nctime_axis: 1.2.0 PseudoNetCDF: None rasterio: 1.1.1 cfgrib: None iris: 2.2.0 bottleneck: None dask: 2.8.1 distributed: 2.8.1 matplotlib: 3.1.2 cartopy: 0.17.0 seaborn: 0.9.0 numbagg: None setuptools: 42.0.2.post20191201 pip: 19.3.1 conda: 4.8.0 pytest: 5.3.1 IPython: 7.10.1 sphinx: 2.2.2 <\/details"}
{"_id":"q-en-pydata__xarray-3677-dbc13d5aa968125312491c1fb2d17147630f90e31ae18c64553c017e9f8d29ad","text":"Merging dataArray into dataset using dataset method fails While it's possible to merge a dataset and a dataarray object using the top-level function, if you try the same thing with the method it fails. Output:"}
{"_id":"q-en-pydata__xarray-3733-cbe51558e5a55941710d3e12fc4c73c18b5f424eb45c4e05e95f071c49cf9f44","text":"Implement polyfit? Fitting a line (or curve) to data along a specified axis is a long-standing need of xarray users. There are many blog posts and SO questions about how to do it: The main use case in my domain is finding the temporal trend on a 3D variable (e.g. temperature in time, lon, lat). Yes, you can do it with applyufunc, but applyufunc is inaccessibly complex for many users. Much of our existing API could be removed and replaced with applyufunc calls, but that doesn't mean we should do it. I am proposing we add a Dataarray method called . It would work like this: Numpy's function is already vectorized in the sense that it accepts 1D x and 2D y, performing the fit independently over each column of y. To extend this to ND, we would just need to reshape the data going in and out of the function. We do this already in . For dask, we could simply require that the dimension over which the fit is calculated be contiguous, and then call mapblocks. Thoughts?"}
{"_id":"q-en-pydata__xarray-3812-1b6ec6fd1abd58207ce5d84be721966184b4b1de8cbc1a6c285b4aaccd524d3f","text":"Turn on reprhtml by default? I just wanted to open this to discuss turning the reprhtml on by default. This PR it as a style option, but I suspect that more people will use if it is on by default. Does that seem like a reasonable change?"}
{"_id":"q-en-pydata__xarray-3905-40a8950d2c809bc56bca5a3694a99fb8737cf8e375d1340eecf596d7d6d1bad5","text":"Truncate array repr based on line count <!-- In order for the maintainers to efficiently understand and prioritize issues, we ask you post a \"Minimal, Complete and Verifiable Example\" (MCVE): --I thought we might have had an issue (and maybe solved it?) but couldn't find it anywhere. Forgive me if I'm duplicating. With larger arrays, it's much more reasonable: <!-- this should explain why the current behavior is a problem and why the expected output is a better solution --Something like 40 lines is probably a reasonable place to truncate? <detailsINSTALLED VERSIONS commit: None python: 3.7.3 (default, Jul 1 2019, 21:52:21) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: [...] machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.utf8 LOCALE: enUS.UTF-8 libhdf5: 1.10.5 libnetcdf: 4.7.1 xarray: 0.15.0 pandas: 0.25.3 numpy: 1.17.3 scipy: 1.3.2 netCDF4: 1.5.3 pydap: None h5netcdf: 0.7.4 h5py: 2.10.0 Nio: None zarr: None cftime: 1.0.4.2 nctime_axis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.2.1 dask: 2.7.0 distributed: 2.7.0 matplotlib: 3.1.2 cartopy: None seaborn: 0.9.0 numbagg: None setuptools: 41.6.0.post20191101 pip: 19.3.1 conda: None pytest: 5.2.2 IPython: 7.9.0 sphinx: 2.2.1 <\/details"}
{"_id":"q-en-pydata__xarray-3976-c50575cecdc6a2d67da98ca41884e444de25d36c5b00656d0528f3244ad341ad","text":"In-place addition of arrays with the same coords but in a different order I have two DataArrays with the same dimension, but the index is in a different order. Adding them with works fine, but the in-place addition fails. is working fine. I would expect to do the same. The in-place addition fails: <details<summaryOutput of <\/summaryINSTALLED VERSIONS commit: None python: 3.8.2 (default, Mar 26 2020, 15:53:00) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 4.19.112-1-MANJARO machine: x8664 processor: byteorder: little LCALL: None LANG: frFR.UTF-8 LOCALE: frFR.UTF-8 libhdf5: None libnetcdf: None xarray: 0.15.0 pandas: 1.0.3 numpy: 1.18.1 scipy: None netCDF4: None pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: None nctime_axis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: None dask: None distributed: None matplotlib: None cartopy: None seaborn: None numbagg: None setuptools: 46.1.1.post20200323 pip: 20.0.2 conda: None pytest: None IPython: None sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-3979-9778f594ca9a7717c33258bdfe5f93a9f31fb3738767b1758dd8eaa3d0d3a81f","text":"xr.fulllike (often) fails when other is chunked and fillvalue is non-scalar I've been running into some issues when using , when my is a chunked dask array, and the is a numpy array. Now, I just checked, mentions only scalar in the signature. However, this is a very convenient way to get all the coordinates and dimensions attached to an array like this, so it feels like desirable functionality. And as I mention below, both numpy and dask function similary, taking much more than just scalars. This results in an error: Expected is a DataArray with the dimensions and coords of , and the numpy array of as its data. The issue lies here: Calling with the given number of chunks results in it trying to to apply the for every individual chunk. As one would expect, if I set to the size of a single chunk it doesn't error: It does fail on a similarly chunked dask array (since it's applying it for every chunk): The most obvious solution would be to force it down the route, since all the values already exist in memory anyway. So maybe another type check does the trick. However, accepts quite a variety of arguments for the fill value (scalars, numpy arrays, lists, tuples, ranges). The dask docs mention only a scalar in the signature for: As does However, in all cases, they still broadcast automatically... So kind of undefined behavior of a blocked ? <details<summaryOutput of <\/summaryINSTALLED VERSIONS commit: None python: 3.7.6 (default, Jan 7 2020, 21:48:41) [MSC v.1916 64 bit (AMD64)] python-bits: 64 OS: Windows OS-release: 10 machine: AMD64 processor: Intel64 Family 6 Model 158 Stepping 9, GenuineIntel byteorder: little LCALL: None LANG: en LOCALE: None.None libhdf5: 1.10.5 libnetcdf: 4.7.3 xarray: 0.15.1 pandas: 0.25.3 numpy: 1.17.5 scipy: 1.3.1 netCDF4: 1.5.3 pydap: None h5netcdf: None h5py: 2.10.0 Nio: None zarr: 2.4.0 cftime: 1.0.4.2 nctimeaxis: None PseudoNetCDF: None rasterio: 1.1.2 cfgrib: None iris: None bottleneck: 1.3.2 dask: 2.9.2 distributed: 2.10.0 matplotlib: 3.1.2 cartopy: None seaborn: 0.10.0 numbagg: None setuptools: 46.1.3.post20200325 pip: 20.0.2 conda: None pytest: 5.3.4 IPython: 7.13.0 sphinx: 2.3.1 <\/detailsxr.fulllike (often) fails when other is chunked and fillvalue is non-scalar I've been running into some issues when using , when my is a chunked dask array, and the is a numpy array. Now, I just checked, mentions only scalar in the signature. However, this is a very convenient way to get all the coordinates and dimensions attached to an array like this, so it feels like desirable functionality. And as I mention below, both numpy and dask function similary, taking much more than just scalars. This results in an error: Expected is a DataArray with the dimensions and coords of , and the numpy array of as its data. The issue lies here: Calling with the given number of chunks results in it trying to to apply the for every individual chunk. As one would expect, if I set to the size of a single chunk it doesn't error: It does fail on a similarly chunked dask array (since it's applying it for every chunk): The most obvious solution would be to force it down the route, since all the values already exist in memory anyway. So maybe another type check does the trick. However, accepts quite a variety of arguments for the fill value (scalars, numpy arrays, lists, tuples, ranges). The dask docs mention only a scalar in the signature for: As does However, in all cases, they still broadcast automatically... So kind of undefined behavior of a blocked ? <details<summaryOutput of <\/summaryINSTALLED VERSIONS commit: None python: 3.7.6 (default, Jan 7 2020, 21:48:41) [MSC v.1916 64 bit (AMD64)] python-bits: 64 OS: Windows OS-release: 10 machine: AMD64 processor: Intel64 Family 6 Model 158 Stepping 9, GenuineIntel byteorder: little LCALL: None LANG: en LOCALE: None.None libhdf5: 1.10.5 libnetcdf: 4.7.3 xarray: 0.15.1 pandas: 0.25.3 numpy: 1.17.5 scipy: 1.3.1 netCDF4: 1.5.3 pydap: None h5netcdf: None h5py: 2.10.0 Nio: None zarr: 2.4.0 cftime: 1.0.4.2 nctimeaxis: None PseudoNetCDF: None rasterio: 1.1.2 cfgrib: None iris: None bottleneck: 1.3.2 dask: 2.9.2 distributed: 2.10.0 matplotlib: 3.1.2 cartopy: None seaborn: 0.10.0 numbagg: None setuptools: 46.1.3.post20200325 pip: 20.0.2 conda: None pytest: 5.3.4 IPython: 7.13.0 sphinx: 2.3.1 <\/details"}
{"_id":"q-en-pydata__xarray-3993-29a3137e45abc44dad89f8eefebd8558e6fce25bf9f8aeb63b133d4f68f17831","text":"DataArray.integrate has a 'dim' arg, but Dataset.integrate has a 'coord' arg This is just a minor gripe but I think it should be fixed. The API syntax is inconsistent: It should definitely be - IMO it doesn't make sense to integrate or differentiate over a dim because a dim by definition has no information about the distance between grid points. I think because the distinction between dims and coords is one of the things that new users have to learn about, we should be strict to not confuse up the meanings in the documentation\/API. The discussion on the original PR , so I think this was just an small oversight. The only question is whether it requires a deprecation cycle?"}
{"_id":"q-en-pydata__xarray-4075-293eb751aec77d18ead06212465e4d35b9f1e71eeb28a35a6b5a6497d94f0783","text":"[bug] when passing boolean weights to weighted mean <!-- A short summary of the issue, if appropriate --<!-- In order for the maintainers to efficiently understand and prioritize issues, we ask you post a \"Minimal, Complete and Verifiable Example\" (MCVE): --Returns Passing a boolean array as weights to the weighted mean returns the wrong result because the are not properly normalized (in this case). Internally the is calculated as i.e. the dot product of two boolean arrays. This yields: We'll need to convert it to int or float: which is correct <details<summaryOutput of <ttxr.showversions()<\/tt<\/summaryINSTALLED VERSIONS commit: None python: 3.7.6 (default, Mar 23 2020, 23:03:20) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 5.3.0-51-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 libhdf5: 1.10.6 libnetcdf: 4.7.4 xarray: 0.15.1 pandas: 1.0.3 numpy: 1.18.1 scipy: 1.4.1 netCDF4: 1.5.3 pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: 1.1.1.2 nctimeaxis: None PseudoNetCDF: None rasterio: 1.1.3 cfgrib: None iris: None bottleneck: None dask: 2.16.0 distributed: 2.16.0 matplotlib: 3.2.1 cartopy: 0.17.0 seaborn: None numbagg: None setuptools: 46.1.3.post20200325 pip: 20.1 conda: None pytest: 5.4.1 IPython: 7.13.0 sphinx: 3.0.3 <\/details"}
{"_id":"q-en-pydata__xarray-4094-b0460d733539f4179f6fe68264da75c1fd405df34b91954f57c42771eee2fa77","text":"tounstackeddataset broken for single-dim variables <!-- A short summary of the issue, if appropriate --A working roundtrip. I need to stack a bunch of variables and later unstack them again, however this doesn't work if the variables only have a single dimension. <details<summaryOutput of <ttxr.showversions()<\/tt<\/summaryINSTALLED VERSIONS commit: None python: 3.7.3 (default, Mar 27 2019, 22:11:17) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 4.15.0-96-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enGB.UTF-8 LOCALE: enGB.UTF-8 libhdf5: 1.10.4 libnetcdf: 4.6.2 xarray: 0.15.1 pandas: 1.0.3 numpy: 1.17.3 scipy: 1.3.1 netCDF4: 1.4.2 pydap: None h5netcdf: None h5py: 2.10.0 Nio: None zarr: None cftime: 1.0.4.2 nctimeaxis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: None dask: 2.10.1 distributed: 2.10.0 matplotlib: 3.1.1 cartopy: None seaborn: 0.10.0 numbagg: None setuptools: 41.0.0 pip: 19.0.3 conda: 4.8.3 pytest: 5.3.5 IPython: 7.9.0 sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-4182-e675e3184497d1428a429e17286246345e40d06c9e2a20c2a06d9cdb3eddd4fa","text":"Pre-expand data and attributes in DataArray\/Variable HTML repr? Given that a major purpose for plotting an array is to look at data or attributes, I wonder if we should expand these sections by default? I worry that clicking on icons to expand sections may not be easy to discover This would also be consistent with the text repr, which shows these sections by default (the Dataset repr is already consistent by default between text and HTML already) Currently the HTML repr for DataArray\/Variable looks like this: ! To see array data, you have to click on the ! icon: ! (thanks to for making this a little bit more manageably sized in !) There's also a really nice repr for nested dask arrays: !"}
{"_id":"q-en-pydata__xarray-4184-52205817c2658d3f60b0f5bededf507db851b85733bf00ddaf107ea7664a0769","text":"Stack + toarray before toxarray is much faster that a simple toxarray I was seeing some slow performance around on MultiIndexed series, and found that unstacking one of the dimensions before running , and then restacking with was ~30x faster. This time difference is consistent with larger data sizes. To reproduce: Create a series with a MultiIndex, ensuring the MultiIndex isn't a simple product: Two approaches for getting this into xarray; 1 - Simple : This takes 536 ms 2 - unstack in pandas first, and then use to do the equivalent of a restack: This takes 17.3 ms To confirm these are identical: A default operation is much slower than a (potentially) equivalent operation that's not the default. I need to look more at what's causing the issues. I think it's to do with the , but I'm unclear why it's so much faster in the alternative route, and whether there's a fix that we can make to make the default path fast. <detailsINSTALLED VERSIONS commit: None python: python-bits: 64 OS: Linux OS-release: 4.9.93-linuxkit-aufs machine: x8664 processor: x8664 byteorder: little LCALL: enUS.UTF-8 LANG: enUS.utf8 LOCALE: None.None xarray: 0.10.9 pandas: 0.23.4 numpy: 1.15.2 scipy: 1.1.0 netCDF4: None h5netcdf: None h5py: None Nio: None zarr: None cftime: None PseudonetCDF: None rasterio: None iris: None bottleneck: 1.2.1 cyordereddict: None dask: None distributed: None matplotlib: 2.2.3 cartopy: 0.16.0 seaborn: 0.9.0 setuptools: 40.4.3 pip: 18.0 conda: None pytest: 3.8.1 IPython: 5.8.0 sphinx: None <\/detailstoxarray() result is incorrect when one of multi-index levels is not sorted <!-- Please include a self-contained copy-pastable example that generates the issue if possible. Please be concise with code posted. See guidelines below on how to provide a good bug report: Craft Minimal Bug Reports: Minimal Complete Verifiable Examples: Bug reports that follow these guidelines are easier to diagnose, and so are often handled much more quickly. --What happened: toxarray() sorts multi-index level values and returns result for the sorted values but it doesn't sort levels or expects levels to be sorted resulting in completely incorrect order of data for the displayed coordinates. What you expected to happen: Should account for the order of levels in the original index. Minimal Complete Verifiable Example: Anything else we need to know?: Environment: <details<summaryOutput of <ttxr.showversions()<\/tt<\/summary<!-- Paste the output here xr.showversions() here --INSTALLED VERSIONS commit: None python: 3.8.2 (default, Apr 24 2020, 08:20:52) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 5.4.7-100.fc30.x8664 machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enGB.UTF-8 LOCALE: enGB.UTF-8 libhdf5: 1.10.4 libnetcdf: 4.6.3 xarray: 0.15.1 pandas: 1.0.5 numpy: 1.19.0 scipy: 1.5.0 netCDF4: 1.5.3 pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: 1.1.3 nctimeaxis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.3.2 dask: 2.19.0 distributed: 2.19.0 matplotlib: 3.2.2 cartopy: None seaborn: 0.10.1 numbagg: installed setuptools: 46.3.0.post20200513 pip: 20.1 conda: None pytest: 5.4.3 IPython: 7.15.0 sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-4248-b2a849fdcd0299c5b6dfc28559e8876b76f94b586078bec78be1452c09088efb","text":"Feature request: show units in dataset overview Here's a hypothetical dataset: It would be really nice if the units of the coordinates and of the data variables were shown in the repr, for example as:"}
{"_id":"q-en-pydata__xarray-4356-8f62af52c34d49100fa92dc7d7589189135f10c3ea61d80942f21b77d6c09f03","text":"sum: mincount is not available for reduction with more than one dimensions Is your feature request related to a problem? Please describe. with errors when passing more than one dim: Describe the solution you'd like The logic to calculate the number of valid elements is here: I think this can be fixed by replacing with Additional context Potentially relevant for"}
{"_id":"q-en-pydata__xarray-4419-7db62982880464fbefde2025b5cb141a36d995c3b7cbb135f4debf51f4bea8d1","text":"concat changes variable order A \"Minimal, Complete and Verifiable Example\" will make it much easier for maintainers to help you: Case 1: Creation of Dataset without Coordinates yields (assumed correct) output of: Case 2: Creation of Dataset with Coordinates yields (assumed false) output of: changes the dimension order for as well as to an alphanumerically sorted representation. should not change the dimension order in any case. <detailsINSTALLED VERSIONS commit: None python: 3.7.1 (default, Nov 13 2018, 18:33:04) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 4.12.14-lp150.12.48-default machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: deDE.UTF-8 LOCALE: deDE.UTF-8 libhdf5: 1.10.4 libnetcdf: 4.6.2 xarray: 0.11.3 pandas: 0.24.1 numpy: 1.16.1 scipy: 1.2.0 netCDF4: 1.4.2 pydap: None h5netcdf: 0.6.2 h5py: 2.9.0 Nio: None zarr: None cftime: 1.0.3.4 PseudonetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.2.1 cyordereddict: None dask: None distributed: None matplotlib: 3.0.2 cartopy: 0.17.0 seaborn: None setuptools: 40.8.0 pip: 19.0.2 conda: None pytest: 4.2.0 IPython: 7.2.0 sphinx: None <\/details[BUG] inverts coordinates order <!-- A short summary of the issue, if appropriate --Following the issue Merging two datasets using inverts the coordinates order. The concat function should not invert the coordinates but maintain the original order. <details<summaryINSTALLED VERSIONS <\/summarycommit: None python: 3.6.8 (default, May 7 2019, 14:58:50) [GCC 8.3.0] python-bits: 64 OS: Linux OS-release: 4.15.0-88-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: C.UTF-8 LOCALE: enUS.UTF-8 libhdf5: None libnetcdf: None xarray: 0.15.1 pandas: 1.0.3 numpy: 1.18.2 scipy: None netCDF4: None pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: None nctime_axis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: None dask: None distributed: None matplotlib: 3.2.0 cartopy: None seaborn: None numbagg: None setuptools: 46.1.3 pip: 9.0.1 conda: None pytest: None IPython: 7.13.0 sphinx: 2.4.3<\/details"}
{"_id":"q-en-pydata__xarray-4423-d90dad2fd1f66ffedeb524816073414bc5cde3cd492f1b00cdac3bae8bc53f4e","text":"(mincount=1) errors for integer data <!-- Please include a self-contained copy-pastable example that generates the issue if possible. Please be concise with code posted. See guidelines below on how to provide a good bug report: Craft Minimal Bug Reports: Minimal Complete Verifiable Examples: Bug reports that follow these guidelines are easier to diagnose, and so are often handled much more quickly. --What happened: returns a if has an integer dtype. Of course mincount is not necessary for integer data as it cannot contain . What you expected to happen: should be ignored Minimal Complete Verifiable Example: Anything else we need to know?: Full traceback <details<\/detailsEnvironment: <details<summaryOutput of <ttxr.showversions()<\/tt<\/summary<!-- Paste the output here xr.showversions() here --INSTALLED VERSIONS commit: python: 3.7.3 (default, Jul 1 2019, 21:52:21) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 5.4.0-42-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 libhdf5: 1.10.6 libnetcdf: 4.7.4 xarray: 0.15.2.dev64+g2542a63f pandas: 0.25.3 numpy: 1.17.3 scipy: 1.3.1 netCDF4: 1.5.4 pydap: installed h5netcdf: 0.7.4 h5py: 2.10.0 Nio: None zarr: 2.3.2 cftime: 1.0.4.2 nctime_axis: None PseudoNetCDF: installed rasterio: 1.1.0 cfgrib: 0.9.5.4 iris: None bottleneck: 1.2.1 dask: 2.6.0 distributed: 2.6.0 matplotlib: 3.3.1 cartopy: 0.18.0 seaborn: 0.9.0 numbagg: None pint: None setuptools: 49.6.0.post20200814 pip: 19.3.1 conda: None pytest: 5.2.2 IPython: 7.17.0 sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-4442-71fb5749fb557bd6a0ebb75345c5d751eef6821bb2d2801b443359755b28d7c2","text":"todataframe\/toseries fails when one out of more than one dims are stacked \/ multiindex When a dataarray has one multiindex dimension, as produced by stack, and has other dimesnions as well, toseries fails to create an combined multiindex. I would expect a series\/dataframe with a multiindex with names a,b,c. Instead I get <detailslib\/python2.7\/site- in isnanew(obj) 115 # hack (for now) because MI registers as ndarray 116 elif isinstance(obj, ABCMultiIndex): --117 raise NotImplementedError(\"isna is not defined for MultiIndex\") 118 elif isinstance(obj, (ABCSeries, np.ndarray, ABCIndexClass, 119 ABCExtensionArray)): NotImplementedError: isna is not defined for MultiIndex <\/detailsOn the other hand, when there is only one dimension, which is stacked, toseries and todataframe work <detailsINSTALLED VERSIONS commit: None python: 2.7.15 (default, May 1 2018, 23:32:55) [GCC 7.2.0] python-bits: 64 OS: Linux OS-release: 3.13.0-48-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: en_GB.UTF-8 LOCALE: None.None libhdf5: 1.8.17 libnetcdf: 4.4.1 xarray: 0.11.3 pandas: 0.23.4 numpy: 1.12.1 scipy: 0.19.1 netCDF4: 1.2.8 pydap: None h5netcdf: None h5py: 2.6.0 Nio: None zarr: None cftime: None PseudonetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.2.0 cyordereddict: None dask: 0.17.3 distributed: 1.21.0 matplotlib: 2.2.2 cartopy: None seaborn: 0.7.1 setuptools: 0.6 pip: 19.0.1 conda: None pytest: 3.0.5 IPython: 5.8.0 sphinx: 1.5.1 <\/details"}
{"_id":"q-en-pydata__xarray-4493-8c63bb04dcafec04e828729a87033fb06d00f2887221f1c2e67a3c47634ef3f2","text":"causes chunked dask DataArray to evalute its values eagerly What happened: Used to update a chunked dask DataArray, but the DataArray is no longer chunked after the update. What you expected to happen: The chunked DataArray should still be chunked after the update Minimal Complete Verifiable Example: Environment: <details<summaryOutput of <ttxr.show_versions()<\/tt<\/summary<\/detailsDataset constructor with DataArray triggers computation Is it intentional that creating a Dataset with a DataArray and dimension names for a single variable causes computation of that variable? In other words, why does cause the dask array to compute? A longer example:"}
{"_id":"q-en-pydata__xarray-4629-6240309bc29bec997548f2bd83e98ca50b3b95204aded8355a8758bd90ed0350","text":"merge(combineattrs='override') does not copy attrs but instead references attrs from the first object <!-- Please include a self-contained copy-pastable example that generates the issue if possible. Please be concise with code posted. See guidelines below on how to provide a good bug report: Craft Minimal Bug Reports: Minimal Complete Verifiable Examples: Bug reports that follow these guidelines are easier to diagnose, and so are often handled much more quickly. --What happened: After a merge, an attribute value change in the merged product is reflected in the first source. What you expected to happen: After a merge, the attrs of the merged product should be able to be changed without having any effect on the sources. Minimal Complete Verifiable Example: Anything else we need to know?: I believe the issue is with the line for combineattrs == \"override\": . This should be changed to , like it is for the other combineattrs cases. Environment: <details<summaryOutput of <ttxr.showversions()<\/tt<\/summary<!-- Paste the output here xr.showversions() here --INSTALLED VERSIONS commit: None python: 3.6.12 (default, Sep 15 2020, 12:49:50) [GCC 4.8.5 (Red Hat 4.8.5-37)] python-bits: 64 OS: Linux OS-release: 3.10.0-1160.6.1.el7.x8664 machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 libhdf5: None libnetcdf: None xarray: 0.16.1 pandas: 1.1.4 numpy: 1.19.4 scipy: 1.5.3 netCDF4: None pydap: None h5netcdf: None h5py: None Nio: None zarr: 2.5.0 cftime: None nctime_axis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: None dask: 2.30.0 distributed: 2.30.0 matplotlib: 3.3.2 cartopy: None seaborn: None numbagg: None pint: None setuptools: 50.3.2 pip: 20.2.4 conda: None pytest: None IPython: None sphinx: 3.3.0 <\/details"}
{"_id":"q-en-pydata__xarray-4683-15653f70d7ee5c9d591e49f9b19400384322ad9474e6f3bd77f34d997c0b68bb","text":"astype method lost its order parameter What happened: I upgraded from xarray 0.15.1 to 0.16.2 and the method seems to have lost the parameter. What you expected to happen: I was expecting to get the same result as with xarray 0.15.1: Anything else we need to know?: Looking at the documentation it seems it disappeared between 0.16.0 and 0.16.1. The documentation at still has this snippet (which was identical to the documentation from at ) while seems to lack that part."}
{"_id":"q-en-pydata__xarray-4684-6d5502df56312c5eea7958723ae13a1a32c68ba632f453a9f4bd93a5038f6b37","text":"Millisecond precision is lost on datetime64 during IO roundtrip <!-- A short summary of the issue, if appropriate --I have millisecond-resolution time data as a coordinate on a DataArray. That data loses precision when round-tripping through disk. <!-- In order for the maintainers to efficiently understand and prioritize issues, we ask you post a \"Minimal, Complete and Verifiable Example\" (MCVE): -- Unzip the data. It will result in a pickle file. output: <!-- this should explain why the current behavior is a problem and why the expected output is a better solution --As you can see, I lose millisecond precision in this data. (The same happens when I use millisecond in the encoding). <details<summaryOutput of <ttxr.showversions()<\/tt<\/summary<!-- Paste the output here xr.showversions() here --INSTALLED VERSIONS commit: None python: 3.7.6 (default, Jan 7 2020, 22:05:27) [Clang 9.0.1 ] python-bits: 64 OS: Darwin OS-release: 19.4.0 machine: x8664 processor: i386 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: None.UTF-8 libhdf5: 1.10.5 libnetcdf: 4.7.3 xarray: 0.15.1 pandas: 1.0.1 numpy: 1.18.1 scipy: 1.4.1 netCDF4: 1.5.3 pydap: None h5netcdf: 0.8.0 h5py: 2.10.0 Nio: None zarr: None cftime: 1.0.4.2 nctime_axis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: None dask: 2.11.0 distributed: 2.14.0 matplotlib: 3.1.3 cartopy: None seaborn: 0.10.0 numbagg: None setuptools: 45.2.0.post20200209 pip: 20.0.2 conda: None pytest: 5.3.5 IPython: 7.12.0 sphinx: 2.4.3 <\/details"}
{"_id":"q-en-pydata__xarray-4687-93964edd38fff0ad88f541a1229b235c3227327dec5802025e197de34c49ea18","text":"not preserving attributes <!-- Please include a self-contained copy-pastable example that generates the issue if possible. Please be concise with code posted. See guidelines below on how to provide a good bug report: Craft Minimal Bug Reports: Minimal Complete Verifiable Examples: Bug reports that follow these guidelines are easier to diagnose, and so are often handled much more quickly. --What happened: Using on a DataArray with attributes results in a new DataArray without attributes. What you expected to happen: Attributes should be preserved or at least there should be a choice (e.g. pass kwargs to so can be passed). Minimal Complete Verifiable Example: Anything else we need to know?: Apart from loosing attributes the dtype is not conserved. In the example the resulting DataArray has dtype np.int64 instead of np.int8. As far as I can see this might not be an xarray but a numpy problem. Environment: <details<summaryOutput of <ttxr.showversions()<\/tt<\/summary<!-- Paste the output here xr.showversions() here --INSTALLED VERSIONS commit: None python: 3.7.8 (default, Jul 31 2020, 02:25:08) [GCC 7.5.0] python-bits: 64 OS: Linux OS-release: 4.14.11--generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 libhdf5: 1.10.6 libnetcdf: 4.7.4 xarray: 0.16.0 pandas: 1.1.2 numpy: 1.19.1 scipy: 1.5.2 netCDF4: 1.5.4 pydap: None h5netcdf: 0.8.1 h5py: 2.10.0 Nio: None zarr: 2.4.0 cftime: 1.2.1 nctimeaxis: None PseudoNetCDF: None rasterio: 1.1.5 cfgrib: None iris: None bottleneck: None dask: 2.25.0 distributed: 2.25.0 matplotlib: 3.3.1 cartopy: 0.18.0 seaborn: None numbagg: None pint: None setuptools: 49.6.0.post20200814 pip: 20.2.3 conda: None pytest: 6.0.1 IPython: 7.18.1 sphinx: 3.2.1 () drops attributes <!-- A short summary of the issue, if appropriate --<!-- In order for the maintainers to efficiently understand and prioritize issues, we ask you post a \"Minimal, Complete and Verifiable Example\" (MCVE): --<!-- this should explain why the current behavior is a problem and why the expected output is a better solution --I would expect the attributes to remain in the data array. <details<summaryOutput of <ttxr.showversions()<\/tt<\/summary<!-- Paste the output here xr.showversions() here --INSTALLED VERSIONS commit: None python: 3.8.2 (default, Apr 24 2020, 08:20:52) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 5.4.0-33-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 libhdf5: None libnetcdf: None xarray: 0.15.1 pandas: 1.0.4 numpy: 1.18.4 scipy: 1.4.1 netCDF4: None pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: None nctimeaxis: None PseudoNetCDF: None rasterio: 1.1.4 cfgrib: None iris: None bottleneck: None dask: 2.16.0 distributed: None matplotlib: 3.2.1 cartopy: None seaborn: None numbagg: None setuptools: 46.2.0 pip: 20.1 conda: None pytest: None IPython: 7.14.0 sphinx: 3.0.4 <\/details"}
{"_id":"q-en-pydata__xarray-4695-766ccfe0bb8564c5fd8745029d1f8192e3d3f46e6147216c48dc59ca6361dda6","text":"Naming a dimension \"method\" throws error when calling \".loc\" The name of the dimension should be irrelevant. The error message suggests that at some point the are given to another method in unsanitized form. Edit: Updated to xarray 0.12 from conda-forge channel. The bug is still present. <detailsINSTALLED VERSIONS commit: None python: 3.6.8 (default, Dec 30 2018, 01:22:34) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 4.18.0-16-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 libhdf5: 1.10.4 libnetcdf: 4.6.1 xarray: 0.12.0 pandas: 0.24.2 numpy: 1.16.2 scipy: 1.2.1 netCDF4: 1.4.2 pydap: None h5netcdf: None h5py: 2.9.0 Nio: None zarr: None cftime: 1.0.3.4 nctimeaxis: None PseudonetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.2.1 dask: None distributed: None matplotlib: 3.0.3 cartopy: None seaborn: None setuptools: 40.8.0 pip: 19.0.3 conda: 4.6.8 pytest: None IPython: 7.3.0 sphinx: 1.8.5 <\/detailsNaming a dimension \"method\" throws error when calling \".loc\" The name of the dimension should be irrelevant. The error message suggests that at some point the are given to another method in unsanitized form. Edit: Updated to xarray 0.12 from conda-forge channel. The bug is still present. <detailsINSTALLED VERSIONS commit: None python: 3.6.8 (default, Dec 30 2018, 01:22:34) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 4.18.0-16-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 libhdf5: 1.10.4 libnetcdf: 4.6.1 xarray: 0.12.0 pandas: 0.24.2 numpy: 1.16.2 scipy: 1.2.1 netCDF4: 1.4.2 pydap: None h5netcdf: None h5py: 2.9.0 Nio: None zarr: None cftime: 1.0.3.4 nctimeaxis: None PseudonetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.2.1 dask: None distributed: None matplotlib: 3.0.3 cartopy: None seaborn: None setuptools: 40.8.0 pip: 19.0.3 conda: 4.6.8 pytest: None IPython: 7.3.0 sphinx: 1.8.5 <\/details"}
{"_id":"q-en-pydata__xarray-4750-c1467f669f7629fe02973bdb27770430a31289dc84e07e3df3142df1bb0e93b6","text":"Limit number of data variables shown in repr <!-- Please include a self-contained copy-pastable example that generates the issue if possible. Please be concise with code posted. See guidelines below on how to provide a good bug report: Craft Minimal Bug Reports: Minimal Complete Verifiable Examples: Bug reports that follow these guidelines are easier to diagnose, and so are often handled much more quickly. --What happened: xarray feels very unresponsive when using datasets with 2000 data variables because it has to print all the 2000 variables everytime you print something to console. What you expected to happen: xarray should limit the number of variables printed to console. Maximum maybe 25? Same idea probably apply to dimensions, coordinates and attributes as well, pandas only shows 2 for reference, the first and last variables. Minimal Complete Verifiable Example: Anything else we need to know?: Out of scope brainstorming: Though printing 2000 variables is probably madness for most people it is kind of nice to show all variables because you sometimes want to know what happened to a few other variables as well. Is there already an easy and fast way to create subgroup of the dataset, so we don' have to rely on the dataset printing everything to the console everytime? Environment: <details<summaryOutput of <ttxr.showversions()<\/tt<\/summaryxr.showversions() INSTALLED VERSIONS commit: None python: 3.8.5 (default, Sep 3 2020, 21:29:08) [MSC v.1916 64 bit (AMD64)] python-bits: 64 OS: Windows OS-release: 10 libhdf5: 1.10.4 libnetcdf: None xarray: 0.16.2 pandas: 1.1.5 numpy: 1.17.5 scipy: 1.4.1 netCDF4: None pydap: None h5netcdf: None h5py: 2.10.0 Nio: None zarr: None cftime: None nctimeaxis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.3.2 dask: 2020.12.0 distributed: 2020.12.0 matplotlib: 3.3.2 cartopy: None seaborn: 0.11.1 numbagg: None pint: None setuptools: 51.0.0.post20201207 pip: 20.3.3 conda: 4.9.2 pytest: 6.2.1 IPython: 7.19.0 sphinx: 3.4.0 <\/details"}
{"_id":"q-en-pydata__xarray-4758-63d46d6cd314cbf0168bbf2023f3e96eaf034bfcc3669163ab13693ceb6d5f35","text":"CFTime offsets missing for milli- and micro-seconds <!-- A short summary of the issue, if appropriate --The smallest cftime offset defined in is \"second\" (S), but the precision of cftime objects goes down to the millisecond (L) and microsecond (U). They should be easily . PR adds a that supports the two, but they are currently untested as cannot generate an index. <!-- In order for the maintainers to efficiently understand and prioritize issues, we ask you post a \"Minimal, Complete and Verifiable Example\" (MCVE): --<!-- this should explain why the current behavior is a problem and why the expected output is a better solution --An error gets raised : . <details<summaryOutput of <ttxr.showversions()<\/tt<\/summary<!-- Paste the output here xr.showversions() here --INSTALLED VERSIONS commit: None python: 3.8.2 (default, Apr 24 2020, 08:20:52) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 5.6.13-arch1-1 machine: x8664 processor: byteorder: little LCALL: None LANG: frCA.utf8 LOCALE: frCA.UTF-8 libhdf5: 1.10.5 libnetcdf: 4.7.4 xarray: 0.15.2.dev9+g6378a711. pandas: 1.0.3 numpy: 1.18.4 scipy: 1.4.1 netCDF4: 1.5.3 pydap: None h5netcdf: None h5py: 2.10.0 Nio: None zarr: None cftime: 1.1.1.2 nctime_axis: 1.2.0 PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.3.2 dask: 2.16.0 distributed: 2.16.0 matplotlib: 3.2.1 cartopy: None seaborn: None numbagg: None pint: 0.11 setuptools: 46.1.3.post20200325 pip: 20.0.2 conda: None pytest: 5.4.1 IPython: 7.13.0 sphinx: 3.0.2 <\/details"}
{"_id":"q-en-pydata__xarray-4767-8d22c141125093e83f58126052ad92752a56a3c177ba6c676d6445108d3e3874","text":"DataArray transpose inconsistent with Dataset Ellipsis usage This works: This doesn't (subset air): The error message is a bit inaccurate too since I do have Ellipsis included; might be related to two calls of:"}
{"_id":"q-en-pydata__xarray-4802-a227f3d3ea4feef77cc7999d95de9b7a567df3accc0e456494c70ecfb29d524c","text":"Decodecf fails when scalefactor is a length-1 list Some datasets I work with have and encoded as length-1 lists. The following code worked as of Xarray 0.16.1 In 0.16.2 (just released) and current master, it fails with this error I'm very confused, because this feels quite similar to , and I thought it was resolved . However, the behavior is different with . That works fine--no error. How might I end up with a dataset with as a python list? It happens when I open a netcdf file using the engine (documented by in ) and then write it to zarr. The numpy array gets encoded as a list in the zarr json metadata. This problem would go away if we could resolve the discrepancies between the two engines' treatment of scalar attributes."}
{"_id":"q-en-pydata__xarray-4819-f8042fa5656ef2b606ef4d13d1554824456f64fd95c894e274a57eb26c13d8b1","text":"dropsel indices in dimension that doesn't have coordinates? <!-- Please do a quick search of existing issues to make sure that this has not been asked before. --Is your feature request related to a problem? Please describe. I am trying to drop particular indices from a dimension that doesn't have coordinates. Following: , but leaving out the coordinate labels: gives me an error. Describe the solution you'd like I would think should exist and work in analogy to as does to . Describe alternatives you've considered As far as I know, I could either create coordinates especially to in order to drop, or rebuild a new dataset. Both are not congenial. (I'd be grateful to know if there is actually a straightforward way to do this I've overlooked."}
{"_id":"q-en-pydata__xarray-4827-510d09c2efb386d94c184814635d23f967fcc3f060e2303fbf5cda8ebec6a3f0","text":"Option for combineattrs with conflicting values silently dropped currently supports four options for merging : It would be nice to have an option to combine attrs from all objects like \"noconflicts\", but that drops attributes with conflicting values rather than raising an error. We might call this or . This is similar to how xarray currently handles conflicting values for and would be more suitable to consider for the default behavior of and other functions\/methods that merge coordinates (e.g., applyufunc, concat, where, binary arithmetic). cc"}
{"_id":"q-en-pydata__xarray-4879-b0e8f7959c2fd545b66b07ad9d328a8946d4d9eac7ef5aca41c3f12e00154c46","text":"jupyter repr caching deleted netcdf file What happened: Testing xarray data storage in a jupyter notebook with varying data sizes and storing to a netcdf, i noticed that opendataset\/array (both show this behaviour) continue to return data from the first testing run, ignoring the fact that each run deletes the previously created netcdf file. This only happens once the was used to display the xarray object. But once in error mode, even the previously fine printed objects are then showing the wrong data. This was hard to track down as it depends on the precise sequence in jupyter. What you expected to happen: when i use , the resulting object should reflect reality on disk. Minimal Complete Verifiable Example: When executed in a cell with print for display, all is fine: but as soon as one cell used the jupyter repr: all future file reads, even after executing the test function again and even using and not , show the data from the last repr use. Anything else we need to know?: Here's a notebook showing the issue: Environment: <details<summaryOutput of <ttxr.showversions()<\/tt<\/summaryINSTALLED VERSIONS commit: None python: 3.7.6 (default, Jun 1 2020, 18:57:50) [GCC 7.5.0] python-bits: 64 OS: Linux OS-release: 5.4.0-40-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 libhdf5: 1.10.6 libnetcdf: 4.7.4 xarray: 0.16.0 pandas: 1.0.5 numpy: 1.19.0 scipy: 1.5.1 netCDF4: 1.5.3 pydap: None h5netcdf: None h5py: 2.10.0 Nio: None zarr: None cftime: 1.2.1 nctimeaxis: None PseudoNetCDF: None rasterio: 1.1.5 cfgrib: None iris: None bottleneck: None dask: 2.21.0 distributed: 2.21.0 matplotlib: 3.3.0 cartopy: 0.18.0 seaborn: 0.10.1 numbagg: None pint: None setuptools: 49.2.0.post20200712 pip: 20.1.1 conda: installed pytest: 6.0.0rc1 IPython: 7.16.1 sphinx: 3.1.2 <\/details"}
{"_id":"q-en-pydata__xarray-4911-b751fa708424f6309335e1278ea229784b62370cedb632d72fa4d3cc84ad9e4d","text":"Sum and prod with mincount forces evaluation If I use the method on a lazy array with then evaluation is forced. If there is some limitation of the implementation which means it cannot be to the computation graph for lazy evaluation then this should be mentioned in the docs. Minimal Complete Verifiable Example: What happened: ` What you expected to happen: No output or exceptions, as the result1 and result2 arrays are not printed or saved. Environment: <details<summaryOutput of <ttxr.showversions()<\/tt<\/summaryINSTALLED VERSIONS commit: None python: 3.9.1 (default, Feb 6 2021, 06:49:13) [GCC 10.2.0] python-bits: 64 OS: Linux OS-release: 5.10.15-arch1-1 machine: x8664 processor: byteorder: little LCALL: None LANG: enNZ.UTF-8 LOCALE: enNZ.UTF-8 libhdf5: 1.12.0 libnetcdf: 4.7.4 xarray: 0.16.2 pandas: 1.2.1 numpy: 1.20.0 scipy: 1.6.0 netCDF4: 1.5.5.1 pydap: None h5netcdf: 0.9.0 h5py: 3.1.0 Nio: None zarr: None cftime: 1.4.1 nctime_axis: None PseudoNetCDF: None rasterio: 1.2.0 cfgrib: None iris: None bottleneck: 1.3.2 dask: 2020.12.0 distributed: 2020.12.0 matplotlib: 3.3.4 cartopy: 0.18.0 seaborn: None numbagg: None pint: None setuptools: 53.0.0 pip: 20.3.1 conda: None pytest: 6.2.1 IPython: 7.19.0 sphinx: 3.4.3 <\/details"}
{"_id":"q-en-pydata__xarray-4939-e430bcafd8703df3eadb6548b7cfd14011e73733737b821f94c1aa79881ac831","text":"DataArrayCoarsen does not have a map or reduce function I'm trying to count unique samples when resampling to a square kilometre from a 5x5m input grid. I'd like to be able to apply the function with to give me a new dimension with the original integer values and their counts. In order to resample along spatial dimensions I assume I need to use , unfortunately the object does not yet implement either a or function for applying an arbitrary function when coarsening. <!-- In order for the maintainers to efficiently understand and prioritize issues, we ask you post a \"Minimal, Complete and Verifiable Example\" (MCVE): --outputs; N.B. has both and while has . Would it make sense for all three to have the same interface? <detailsINSTALLED VERSIONS commit: None python: 3.7.4 (default, Sep 7 2019, 18:27:02) [Clang 10.0.1 (clang-1001.0.46.4)] python-bits: 64 OS: Darwin OS-release: 19.3.0 machine: x8664 processor: i386 byteorder: little LCALL: None LANG: enAU.UTF-8 LOCALE: enAU.UTF-8 libhdf5: None libnetcdf: None xarray: 0.15.0 pandas: 1.0.0 numpy: 1.18.1 scipy: 1.4.1 netCDF4: None pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: None nctime_axis: None PseudoNetCDF: None rasterio: 1.1.2 cfgrib: None iris: None bottleneck: None dask: 2.10.1 distributed: 2.10.0 matplotlib: 3.1.2 cartopy: None seaborn: None numbagg: None setuptools: 40.8.0 pip: 19.0.3 conda: None pytest: 5.3.5 IPython: 7.11.1 sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-4940-9a0eafb214736ea24b961abc9305acf579c08179dcf4898702354bdb839672c5","text":"Calling () drops coordinates This is a similar issue to bug . <!-- In order for the maintainers to efficiently understand and prioritize issues, we ask you post a \"Minimal, Complete and Verifiable Example\" (MCVE): --Output (with extra stuff snipped for brevity): It's also worth mentioning that the data arrays contained in the dataset also loose their coordinates during this operation. I.E: <detailsINSTALLED VERSIONS commit: None python: 3.7.4 (tags\/v3.7.4:, Jul 8 2019, 20:34:20) [MSC v.1916 64 bit (AMD64)] python-bits: 64 OS: Windows OS-release: 10 machine: AMD64 processor: Intel64 Family 6 Model 158 Stepping 11, GenuineIntel byteorder: little LCALL: None LANG: None LOCALE: None.None libhdf5: 1.10.5 libnetcdf: 4.7.2 xarray: 0.14.0 pandas: 0.25.2 numpy: 1.17.3 scipy: 1.3.1 netCDF4: 1.5.3 pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: 1.0.4.2 nctime_axis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: None dask: 2.7.0 distributed: None matplotlib: 3.1.1 cartopy: None seaborn: None numbagg: None setuptools: 40.8.0 pip: 19.3 conda: None pytest: None IPython: 7.8.0 sphinx: None None <\/details"}
{"_id":"q-en-pydata__xarray-4966-29cb69ff715e2376141840bbc7767901abf487ff690d8f8eb334ee8eb5942caa","text":"Handling of signed bytes from OPeNDAP via pydap netCDF3 only knows signed bytes, but there's of adding an attribute to the variable to be able to store unsigned bytes non the less. This convention is handled by xarray. OPeNDAP only knows unsigned bytes, but there's which is used by the thredds server and the netCDF-c library of adding an attribute to the variable to be able to store signed bytes non the less. This hack is not handled by xarray, but maybe should be handled symmetrically at the same place (i.e. ). As descibed in the \"hack\", netCDF-c handles this internally, but pydap doesn't. This is why the variant returns (correctly according to the hack) negative values and the variant doesn't. However, as returns a warning at exactly the location referenced above, I think that this is the place where it should be fixed. If you agree, I could prepare a PR to implement the fix. Handling of signed bytes from OPeNDAP via pydap netCDF3 only knows signed bytes, but there's of adding an attribute to the variable to be able to store unsigned bytes non the less. This convention is handled by xarray. OPeNDAP only knows unsigned bytes, but there's which is used by the thredds server and the netCDF-c library of adding an attribute to the variable to be able to store signed bytes non the less. This hack is not handled by xarray, but maybe should be handled symmetrically at the same place (i.e. ). As descibed in the \"hack\", netCDF-c handles this internally, but pydap doesn't. This is why the variant returns (correctly according to the hack) negative values and the variant doesn't. However, as returns a warning at exactly the location referenced above, I think that this is the place where it should be fixed. If you agree, I could prepare a PR to implement the fix."}
{"_id":"q-en-pydata__xarray-4994-1d546e0736832212dabe1b15eda93554e66ce25f7dc93bc7e76191056ef553f5","text":"Date missing in datetime accessor What happened: I wonder if there is a reason, why there is no attribute in the datetime accessor. What you expected to happen: As the attribute is supported I would expect the same for the attribute Minimal Complete Verifiable Example: Suggestion: A simple addition of in should do the trick. Happy to do a PR. Anything else we need to know?: Environment: <details<summaryOutput of <ttxr.showversions()<\/tt<\/summaryINSTALLED VERSIONS commit: None python: 3.8.6 (default, Oct 7 2020, 19:08:05) [GCC 7.5.0] python-bits: 64 OS: Linux OS-release: 2.6.32-754.33.1.el6.x8664 machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 libhdf5: 1.10.6 libnetcdf: 4.7.4 xarray: 0.17.0 pandas: 1.2.1 numpy: 1.20.0 scipy: 1.6.0 netCDF4: 1.5.5.1 pydap: None h5netcdf: None h5py: None Nio: None zarr: 2.6.1 cftime: 1.4.1 nctime_axis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: None dask: 2021.02.0 distributed: 2021.02.0 matplotlib: 3.3.4 cartopy: 0.18.0 seaborn: 0.11.1 numbagg: None pint: 0.16.1 setuptools: 49.6.0.post20210108 pip: 21.0.1 conda: None pytest: None IPython: 7.20.0 sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-5033-68ee81b033fc92d7425e8fc63f789f7e67f88b37f07d87caaef1d10ab47cdefd","text":"Simplify adding custom backends <!-- Please do a quick search of existing issues to make sure that this has not been asked before. --Is your feature request related to a problem? Please describe. I've been working on opening custom hdf formats in xarray, reading up on the apiv2 it is currently only possible to declare a new external plugin in but that doesn't seem easy or intuitive to me. Describe the solution you'd like Why can't we simply be allowed to add functions to the engine parameter? Example: This seems like a small function change to me from my initial quick look because there's mainly a bunch of string checks in the normal case until we get to the registered backend functions, if we send in a function instead in the engine-parameter we can just bypass those checks."}
{"_id":"q-en-pydata__xarray-5131-c134585b18390454f7fddbd56d5602400464172a204c98f517fa5638a730edfc","text":"Trailing whitespace in DatasetGroupBy text representation When displaying a DatasetGroupBy in an interactive Python session, the first line of output contains a trailing whitespace. The first example in the documentation demonstrate this: There is a trailing whitespace in the first line of output which is \"DatasetGroupBy, grouped over 'letters' \". This can be seen more clearly by converting the object to a string (note the whitespace before ): While this isn't a problem in itself, it causes an issue for us because we use flake8 in continuous integration to verify that our code is correctly formatted and we also have doctests that rely on DatasetGroupBy textual representation. Flake8 reports a violation on the trailing whitespaces in our docstrings. If we remove the trailing whitespaces, our doctests fail because the expected output doesn't match the actual output. So we have conflicting constraints coming from our tools which both seem reasonable. Trailing whitespaces are forbidden by flake8 because, among other reasons, they lead to noisy git diffs. Doctest want the expected output to be exactly the same as the actual output and considers a trailing whitespace to be a significant difference. We could configure flake8 to ignore this particular violation for the files in which we have these doctests, but this may cause other trailing whitespaces to creep in our code, which we don't want. Unfortunately it's not possible to just add comments to get flake8 to ignore the violation only for specific lines because that creates a difference between expected and actual output from doctest point of view. Flake8 doesn't allow to disable checks for blocks of code either. Is there a reason for having this trailing whitespace in DatasetGroupBy representation? Whould it be OK to remove it? If so please let me know and I can make a pull request."}
{"_id":"q-en-pydata__xarray-5180-728447f22ac35611d4f598311e4d0cc567a2c09361e08cae792013e9a0d7db98","text":"opendataset uses cftime, not datetime64, when calendar attribute is \"Gregorian\" <!-- Please include a self-contained copy-pastable example that generates the issue if possible. Please be concise with code posted. See guidelines below on how to provide a good bug report: Craft Minimal Bug Reports: Minimal Complete Verifiable Examples: Bug reports that follow these guidelines are easier to diagnose, and so are often handled much more quickly. --What happened: I used to open a NetCDF file whose coordinate had the attribute set to . All dates were within the Timestamp-valid range. The resulting dataset represented the co-ordinate as a . What you expected to happen: I expected the dataset to represent the co-ordinate as a , as documented and . Minimal Complete Verifiable Example: Anything else we need to know?: The says: In practice, we are getting some s even for times which are interpretable and representable as s. In particular, we have some NetCDF files in which the variable has a attribute with a value of (with a capital ‘G’). CF conventions : However, xarray regards as a non-standard calendar and falls back to . If (as in the example) is to , the times are read as s. Suggested fix: in , change ‘’ to ‘’. Environment: <details<summaryOutput of <ttxr.showversions()<\/tt<\/summaryINSTALLED VERSIONS commit: None python: 3.9.2 (default, Feb 21 2021, 05:02:46) [GCC 9.3.0] python-bits: 64 OS: Linux OS-release: 5.8.0-48-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enGB.UTF-8 LOCALE: enGB.UTF-8 libhdf5: 1.10.6 libnetcdf: 4.7.4 xarray: 0.17.1.dev39+g45b4436b pandas: 1.2.3 numpy: 1.20.2 scipy: None netCDF4: 1.5.6 pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: 1.4.1 nctimeaxis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: None dask: None distributed: None matplotlib: None cartopy: None seaborn: None numbagg: None pint: None setuptools: 49.6.0.post20210108 pip: 21.0.1 conda: None pytest: None IPython: None sphinx: None <!-- Paste the output here xr.show_versions() here --<\/details"}
{"_id":"q-en-pydata__xarray-5187-5803fff15fb671fb2ecb4b28814a58ca7f392cca65c82ad8f937ea0dec2f933b","text":"bfill behavior dask arrays with small chunk size bfill methods seems to miss nans when dask array chunk size is small. Resulting array still has nan present (see 'small chunk' section of code) absence of nans INSTALLED VERSIONS commit: None python: python-bits: 64 OS: Linux OS-release: 4.15.0-43-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enCA.UTF-8 LOCALE: en_CA.UTF-8 xarray: 0.11.0 pandas: 0.23.4 numpy: 1.15.4 scipy: None netCDF4: None h5netcdf: None h5py: None Nio: None zarr: None cftime: None PseudonetCDF: None rasterio: None iris: None bottleneck: 1.2.1 cyordereddict: None dask: 1.0.0 distributed: 1.25.2 matplotlib: None cartopy: None seaborn: None setuptools: 40.6.3 pip: 18.1 conda: None pytest: None IPython: None sphinx: None"}
{"_id":"q-en-pydata__xarray-5362-82cf8a7ae496215919abf77835afe341b5aa109229c4f857a84f2a0a3616296a","text":"Should weighted operations raise an error when dimensions don't exist? <!-- Please include a self-contained copy-pastable example that generates the issue if possible. Please be concise with code posted. See guidelines below on how to provide a good bug report: Craft Minimal Bug Reports: Minimal Complete Verifiable Examples: Bug reports that follow these guidelines are easier to diagnose, and so are often handled much more quickly. --What happened: Weighted operations don't raise an error when the dimensions passed don't exist. What you expected to happen: This is not really a bug, but I find it a bit confusing because it's not consistent with the same \"unweighted\" operation. Minimal Complete Verifiable Example: Environment: <details<summaryOutput of <ttxr.showversions()<\/tt<\/summary<!-- Paste the output here xr.showversions() here --INSTALLED VERSIONS commit: None python: 3.9.4 (default, May 10 2021, 22:13:33) [GCC 9.3.0] python-bits: 64 OS: Linux OS-release: 3.10.0-1062.18.1.el7.x8664 machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enGB.UTF-8 LOCALE: ('enGB', 'UTF-8') libhdf5: 1.10.6 libnetcdf: 4.7.4 xarray: 0.18.1.dev30+g2578fc3 pandas: 1.2.4 numpy: 1.20.2 scipy: 1.6.3 netCDF4: 1.5.6 pydap: installed h5netcdf: 0.11.0 h5py: 3.2.1 Nio: None zarr: 2.8.1 cftime: 1.4.1 nctimeaxis: 1.2.0 PseudoNetCDF: None rasterio: 1.2.3 cfgrib: 0.9.9.0 iris: None bottleneck: 1.3.2 dask: 2021.05.0 distributed: 2021.05.0 matplotlib: 3.4.2 cartopy: 0.19.0.post1 seaborn: 0.11.1 numbagg: installed pint: None setuptools: 49.6.0.post20210108 pip: 21.1.1 conda: None pytest: None IPython: 7.23.1 sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-5365-e7c3ef9dc787f2b2b0c6e789c88e445e848b9442e99d799bba9018ea624ac424","text":"Feature request: vector cross product xarray currently has the function for calculating arbitrary dot products which is indeed very handy. Sometimes, especially for physical applications I also need a vector cross product. I' wondering whether you would be interested in having as a wrapper of I currently use the following implementation: Do you want such a function (and possibly associated methods) in the namespace, or should it be in some other package? I didn't find a package which would be a good fit as this is close to core numpy functionality and isn't as domain specific as some geo packages. I'm not aware of some \"xrphysics\" package. I could make a PR if you'd want to have it in directly. <detailsINSTALLED VERSIONS commit: None python: 3.7.3 (default, Mar 27 2019, 22:11:17) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 4.9.0-9-amd64 machine: x8664 processor: byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 libhdf5: 1.10.4 libnetcdf: 4.6.1 xarray: 0.12.3 pandas: 0.24.2 numpy: 1.16.4 scipy: 1.3.0 netCDF4: 1.4.2 pydap: None h5netcdf: 0.7.4 h5py: 2.9.0 Nio: None zarr: None cftime: 1.0.3.4 nctimeaxis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.2.1 dask: 2.1.0 distributed: 2.1.0 matplotlib: 3.1.0 cartopy: None seaborn: 0.9.0 numbagg: None setuptools: 41.0.1 pip: 19.1.1 conda: 4.7.11 pytest: 5.0.1 IPython: 7.6.1 sphinx: 2.1.2 <\/details"}
{"_id":"q-en-pydata__xarray-5580-f8eaa832ed20cf3eff03647af348159abb56646b1960c78ea3b4b2998fa2846f","text":"Increase default This must have been introduced into somewhere around version 0.17. First of all this limit breaks backwards compatibility in the output format with or something similar on the console. Second, the default of 12 is much too low imo and makes no sense, in particular since terminals usually have a scrollback buffer and notebook cells can be made scrollable. I use frequently to check that all variables made it into the data set correctly, which is meaningless when lines are skipped with this default limit. And it broke my doctests that I wrote to do exactly that (thanks for that btw.). So if not removed, could the default be at least increased to a sensible number like 100 or 1000 or ? (I'd personally prefer much higher or even no limit, but I guess that is not an option.) Cheers."}
{"_id":"q-en-pydata__xarray-5662-588949ec5b2a57dd742382b4acc628a4ef5555471cd57e16347d1cf537fc4baf","text":"Limit number of displayed dimensions in repr <!-- Please include a self-contained copy-pastable example that generates the issue if possible. Please be concise with code posted. See guidelines below on how to provide a good bug report: Craft Minimal Bug Reports: Minimal Complete Verifiable Examples: Bug reports that follow these guidelines are easier to diagnose, and so are often handled much more quickly. --What happened: Dimension doesn't seem to be limited when there are too many of them. See example below. This slows down the repr significantly and is quite unreadable to me. What you expected to happen: To be limited so that it aligns with whatever the maximum line length is for variables. It's also fine if it continues on a couple of rows below in similar fashion to variables. Minimal Complete Verifiable Example: This is probably a bit of an edge case. My real datasets usually have around 12 \"dimensions\" and coords, +2000 variables, 50 attrs. Anything else we need to know?: Environment: <details<summaryOutput of <ttxr.showversions()<\/tt<\/summaryINSTALLED VERSIONS commit: None python: 3.8.8 (default, Feb 20 2021, 15:50:08) [MSC v.1916 64 bit (AMD64)] python-bits: 64 OS: Windows OS-release: 10 byteorder: little LCALL: None LANG: en libhdf5: 1.10.6 libnetcdf: None xarray: 0.18.2 pandas: 1.2.4 numpy: 1.20.3 scipy: 1.6.3 netCDF4: None pydap: None h5netcdf: None h5py: 3.2.1 Nio: None zarr: None cftime: None nctimeaxis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.3.2 dask: 2021.05.0 distributed: 2021.05.0 matplotlib: 3.4.2 cartopy: None seaborn: 0.11.1 numbagg: None pint: None setuptools: 49.6.0.post20210108 pip: 21.1.2 conda: 4.10.1 pytest: 6.2.4 IPython: 7.24.1 sphinx: 4.0.2 <\/details"}
{"_id":"q-en-pydata__xarray-5682-fa5200799f50c796da87762f45b999be96abe6270962cb279af58e9376aef553","text":"Complex LaTeX expressions in s aren't rendered correctly when plotting <!-- Please include a self-contained copy-pastable example that generates the issue if possible. Please be concise with code posted. See guidelines below on how to provide a good bug report: Craft Minimal Bug Reports: Minimal Complete Verifiable Examples: Bug reports that follow these guidelines are easier to diagnose, and so are often handled much more quickly. --What happened: When I try to give a variable a that's a complex latex expression and then plot that variable the expression doesn't get rendered by latex What you expected to happen: I expected the name to get rendered by latex Minimal Complete Verifiable Example: In the example below I'm plotting a variable with a complex via xarray and then plotting it again (in a separate figure) using only matplotlib and manually setting . The matplotlib-only version works fine (right), but the xarray version doesn't render (left). ! Anything else we need to know?: Environment: <details<summaryOutput of <ttxr.showversions()<\/tt<\/summaryINSTALLED VERSIONS commit: None python: 3.9.2 (default, Mar 3 2021, 20:02:32) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 5.10.53-1-MANJARO machine: x8664 processor: byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: enUS.UTF-8 libhdf5: 1.10.6 libnetcdf: 4.6.1 xarray: 0.17.0 pandas: 1.2.3 numpy: 1.19.2 scipy: 1.5.3 netCDF4: 1.5.6 pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: 1.2.1 nctimeaxis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.3.2 dask: 2021.04.0 distributed: 2021.04.0 matplotlib: 3.3.4 cartopy: 0.18.0 seaborn: None numbagg: None pint: 0.17 setuptools: 52.0.0.post20210125 pip: 21.0.1 conda: None pytest: None IPython: 7.22.0 sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-5731-269111d7f85b959419bfac8e2399e7e3b31fbb49effa93562e7e07692ee9c549","text":"Dask error on What happened: When I use on two Dataarrays I get a error. Notice that dask is not installed in my environement. What you expected to happen: Obtain the correlation values without dask interfering (as it should be optional in my understanding) Minimal Complete Verifiable Example: Results in: Environment:"}
{"_id":"q-en-pydata__xarray-6386-4df0688f26c93f3980127763eee10552c7881c5cea179dc5399ba20f0e4375a1","text":"Dataset groupby returning DataArray broken in some cases Got a TypeError when resampling a dataset along a dimension, mapping a function to each group. The function returns a DataArray. Failed with : This worked before the merging of . A DataArray was returned as expected. In the docstring of it is not made clear that the passed function should return a dataset, but the opposite is also not said. This worked before and I think the issues comes from , which introduced different signatures for (which is called in my case) and (which is expected by the new ). If the function passed to should only return s then I believe a more explicit error is needed, as well as some notice in the docs and a breaking change entry in the changelog. If s should be accepted, then we have a regression here. I may have time to help on this. <detailsINSTALLED VERSIONS commit: None python: 3.9.6 (default, Jul 11 2021, 03:39:48) [GCC 9.3.0] python-bits: 64 OS: Linux OS-release: 5.16.13-arch1-1 machine: x8664 processor: byteorder: little LCALL: None LANG: frCA.utf8 LOCALE: ('frCA', 'UTF-8') libhdf5: 1.12.0 libnetcdf: 4.7.4 xarray: 2022.3.1.dev16+g3ead17ea pandas: 1.4.0 numpy: 1.20.3 scipy: 1.7.1 netCDF4: 1.5.7 pydap: None h5netcdf: 0.11.0 h5py: 3.4.0 Nio: None zarr: 2.10.0 cftime: 1.5.0 nctime_axis: 1.3.1 PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.3.2 dask: 2021.08.0 distributed: 2021.08.0 matplotlib: 3.4.3 cartopy: None seaborn: None numbagg: None fsspec: 2021.07.0 cupy: None pint: 0.18 sparse: None setuptools: 57.4.0 pip: 21.2.4 conda: None pytest: 6.2.5 IPython: 8.0.1 sphinx: 4.1.2 <\/details"}
{"_id":"q-en-pydata__xarray-6394-45f938fffcf3a54c0d2e4769cfb9d23e91309cb5a5dee6fffa7af53ce6c3ee7f","text":"DataArray groupby returning Dataset broken in some cases This is a the reverse problem of , the method seems broken when the mapped function returns a Dataset (which worked before ). No response I guess the same solution as could be used! <detailsINSTALLED VERSIONS commit: None python: 3.9.6 (default, Jul 11 2021, 03:39:48) [GCC 9.3.0] python-bits: 64 OS: Linux OS-release: 5.16.13-arch1-1 machine: x8664 processor: byteorder: little LCALL: None LANG: frCA.utf8 LOCALE: ('frCA', 'UTF-8') libhdf5: 1.12.0 libnetcdf: 4.7.4 xarray: 2022.3.1.dev16+g3ead17ea pandas: 1.4.0 numpy: 1.20.3 scipy: 1.7.1 netCDF4: 1.5.7 pydap: None h5netcdf: 0.11.0 h5py: 3.4.0 Nio: None zarr: 2.10.0 cftime: 1.5.0 nctime_axis: 1.3.1 PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.3.2 dask: 2021.08.0 distributed: 2021.08.0 matplotlib: 3.4.3 cartopy: None seaborn: None numbagg: None fsspec: 2021.07.0 cupy: None pint: 0.18 sparse: None setuptools: 57.4.0 pip: 21.2.4 conda: None pytest: 6.2.5 IPython: 8.0.1 sphinx: 4.1.2 <\/details"}
{"_id":"q-en-pydata__xarray-6400-915388425b23bb79d45e2977cba2b3c4b6cf1d29b998f87d2de00001de255c68","text":"Very poor html repr performance on large multi-indexes <!-- Please include a self-contained copy-pastable example that generates the issue if possible. Please be concise with code posted. See guidelines below on how to provide a good bug report: Craft Minimal Bug Reports: Minimal Complete Verifiable Examples: Bug reports that follow these guidelines are easier to diagnose, and so are often handled much more quickly. --What happened: We have catestrophic performance on the html repr of some long multi-indexed data arrays. Here's a case of it taking 12s. Minimal Complete Verifiable Example: Anything else we need to know?: I thought we'd fixed some issues here: Environment: <details<summaryOutput of <ttxr.showversions()<\/tt<\/summaryINSTALLED VERSIONS commit: None python: 3.8.10 (default, May 9 2021, 13:21:55) [Clang 12.0.5 (clang-1205.0.22.9)] python-bits: 64 OS: Darwin OS-release: 20.4.0 machine: x8664 processor: i386 byteorder: little LCALL: None LANG: None LOCALE: ('enUS', 'UTF-8') libhdf5: None libnetcdf: None xarray: 0.18.2 pandas: 1.2.4 numpy: 1.20.3 scipy: 1.6.3 netCDF4: None pydap: None h5netcdf: None h5py: None Nio: None zarr: 2.8.3 cftime: 1.4.1 nctimeaxis: None PseudoNetCDF: None rasterio: 1.2.3 cfgrib: None iris: None bottleneck: 1.3.2 dask: 2021.06.1 distributed: 2021.06.1 matplotlib: 3.4.2 cartopy: None seaborn: 0.11.1 numbagg: 0.2.1 pint: None setuptools: 56.0.0 pip: 21.1.2 conda: None pytest: 6.2.4 IPython: 7.24.0 sphinx: 4.0.1 <\/details"}
{"_id":"q-en-pydata__xarray-6461-0ff48c5164180d7c1ef5726eab460363d1ce48fbeb117b5deb4de0a87b729482","text":"with scalar as second argument fails with keepattrs=True fails with The workaround is to pass No response No response No response No response xarray 2022.3.0"}
{"_id":"q-en-pydata__xarray-6548-d9347e1f3a044bd0ccc2ee654fd4e04e916156586272d4eaf790bb918070d55e","text":"xr.polyval first arg requires name attribute I have some polynomial coefficients and want to evaluate them at some values using . As described in the docstring\/docu I created a 1D coordinate DataArray and pass it to but it raises a KeyError (see example). I expected that the polynomial would be evaluated at the given points. No response I assume that the \"standard\" workflow is to obtain the argument from an existing DataArrays coordinate, where the name would be correctly set already. However, that is not clear from the description, and also prevents my \"manual\" workflow. It could be that the problem will be solved by replacing the coord DataArray argument by an explicit Index in the future. <detailsINSTALLED VERSIONS commit: None python: 3.9.10 (main, Mar 15 2022, 15:56:56) [GCC 7.5.0] python-bits: 64 OS: Linux OS-release: 3.10.0-1160.49.1.el7.x8664 machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: ('enUS', 'UTF-8') libhdf5: 1.12.0 libnetcdf: 4.7.4 xarray: 2022.3.0 pandas: 1.4.2 numpy: 1.22.3 scipy: None netCDF4: 1.5.8 pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: 1.6.0 nctimeaxis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: None dask: None distributed: None matplotlib: 3.5.1 cartopy: 0.20.2 seaborn: None numbagg: None fsspec: None cupy: None pint: None sparse: None setuptools: 58.1.0 pip: 22.0.4 conda: None pytest: None IPython: 8.2.0 sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-6598-b4b15229149560bfa505bd7b4bb88c333989003332776840663b632d64aea627","text":"xarray improperly decodes times from a NetCDF when it is a uint improperly decodes times from a NetCDF when it is a . The generates a NetCDF file with the right time ('goodtime') and the wrong time ('time') (use ) to be properly decoded (see ). [X] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray. [X] Complete example — the example is self-contained, including all data and the text of any traceback. , returning the result. [X] New issue — a search of GitHub Issues suggests this is not a duplicate. No response <detailsINSTALLED VERSIONS commit: None python: 3.8.13 (default, Mar 25 2022, 06:04:10) [GCC 10.3.0] python-bits: 64 OS: Linux OS-release: 3.10.0-1160.62.1.el7.x8664 machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: ('enUS', 'UTF-8') libhdf5: 1.10.6 libnetcdf: 4.8.0 xarray: 2022.3.0 pandas: 1.4.2 numpy: 1.22.3 scipy: 1.7.0 netCDF4: 1.5.7 pydap: None h5netcdf: 1.0.0 h5py: 3.3.0 Nio: None zarr: 2.11.3 cftime: 1.6.0 nctimeaxis: None PseudoNetCDF: None rasterio: None cfgrib: 0.9.10.1 iris: None bottleneck: 1.3.4 dask: 2022.04.1 distributed: 2022.4.1 matplotlib: 3.5.1 cartopy: None seaborn: 0.11.2 numbagg: None fsspec: 2021.06.1 cupy: None pint: 0.19.1 sparse: None setuptools: 62.1.0 pip: 22.0.4 conda: 4.12.0 pytest: 7.1.1 IPython: 8.2.0 sphinx: 4.5.0 <\/details"}
{"_id":"q-en-pydata__xarray-6599-a0ddad63a2ad77578d9333cbbed4900f8218984011ceee4df9ba9569e8183757","text":"with timedelta64 coordinates produces wrong results I'm not sure if this is a bug or an expected breaking change, but I'm not able to reproduce the results generated by using a timedelta64 coordinate. The results are correct in , whereas they are wrong in the latest unreleased version (, ). Both the stable and latest functions should return the same results. [X] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray. [X] Complete example — the example is self-contained, including all data and the text of any traceback. , returning the result. [X] New issue — a search of GitHub Issues suggests this is not a duplicate. No response <detailsINSTALLED VERSIONS commit: None python: 3.10.4 (main, Mar 24 2022, 17:43:32) [Clang 12.0.1 ] python-bits: 64 OS: Darwin OS-release: 21.4.0 machine: x8664 processor: i386 byteorder: little LCALL: None LANG: None LOCALE: (None, 'UTF-8') libhdf5: None libnetcdf: None xarray: 2022.3.0 or 2022.3.1.dev102+g6bb2b855 pandas: 1.4.2 numpy: 1.22.3 scipy: 1.8.0 netCDF4: None pydap: None h5netcdf: None h5py: None Nio: None zarr: 2.11.3 cftime: None nctime_axis: None PseudoNetCDF: None rasterio: 1.2.10 cfgrib: None iris: None bottleneck: None dask: 2022.05.0 distributed: 2022.5.0 matplotlib: None cartopy: None seaborn: None numbagg: None fsspec: 2022.3.0 cupy: None pint: None sparse: None setuptools: 62.2.0 pip: 22.1 conda: None pytest: 7.1.2 IPython: None sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-6601-873b2a8868d490480819887d1847565d82d6da7df23793e7b92dc8eb8828d778","text":"returns objects with different dimension order I noticed that the dimension order of the object returned by the latest (, unreleased) is different compared to . For example, the following code returns different results. Is this the expected behaviour? If yes, is it worth mentioning this change in what's new\/breaking changes? cc:"}
{"_id":"q-en-pydata__xarray-6721-71dcfb174624d32f2f75363684615476f25dde2042fea8ee47f9db1f0a8ecf9d","text":"Accessing chunks on zarr backed xarray seems to load entire array into memory When running the following example it appears the entire dataset is loaded into memory when accessing the attribute: According to accessing the chunks attribute should simply inspect the attribute on the underlying DataArrays. No response No response <detailsINSTALLED VERSIONS commit: None python: 3.9.12 (main, Mar 24 2022, 23:24:38) [Clang 12.0.1 ] python-bits: 64 OS: Darwin OS-release: 21.2.0 machine: arm64 processor: arm byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: ('enUS', 'UTF-8') libhdf5: None libnetcdf: None xarray: 2022.3.0 pandas: 1.4.2 numpy: 1.21.2 scipy: 1.8.0 netCDF4: None pydap: None h5netcdf: None h5py: None Nio: None zarr: 2.8.1 cftime: None nctime_axis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.3.4 dask: 2022.04.0 distributed: 2022.4.0 matplotlib: 3.4.3 cartopy: None seaborn: None numbagg: None fsspec: 2022.3.0 cupy: None pint: None sparse: None setuptools: 62.0.0 pip: 22.0.4 conda: None pytest: 7.1.1 IPython: 8.2.0 sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-6744-3cf46849068850e780ba142f617123bd58aaf97b248afd9137165d6860c26295","text":"\"center\" kwarg ignored when manually iterating over DataArrayRolling <div type='discussions-op-text'<supOriginally posted by ckingdon95 June 29, 2022<\/supHello, I am trying to manually iterate over a DataArrayRolling object, as described in the documentation. I am confused why the following two code chunks do not produce the same sequence of values. I would like to be able to manually iterate over a DataArrayRolling object, and still be given center-justified windows. Is there a way to do this? This returns the following values, as expected: Whereas when I do it manually, it is not equivalent: This returns Is this an issue with the window iterator? If it is not an issue, then is there a way for me to get the center-justified windows in the manual iteration? <\/div"}
{"_id":"q-en-pydata__xarray-6798-b7c72a3d6a9d8888caed89f583439c5de59fe01d4f1c2e940852a490feb4c817","text":"Dropping a MultiIndex variable raises an error after explicit indexes refactor With the latest released version of Xarray, it is possible to delete all variables corresponding to a MultiIndex by simply deleting the name of the MultiIndex. After the explicit indexes refactor (i.e,. using the \"main\" development branch) this now raises error about how this would \"corrupt\" index state. This comes up when using and and possibly some other methods. This is not hard to work around, but we may want to consider this bug a blocker for the next Xarray release. I found the issue surfaced in several projects when attempting to use the new version of Xarray inside Google's codebase. CC in case you have any thoughts to share. For now, we should preserve the behavior of deleting the variables corresponding to MultiIndex levels, but should issue a deprecation warning encouraging users to explicitly delete everything. No response <detailsINSTALLED VERSIONS commit: python: 3.10.4 (main, Mar 24 2022, 17:42:03) [Clang 12.0.1 ] python-bits: 64 OS: Darwin OS-release: 21.4.0 machine: arm64 processor: arm byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: ('enUS', 'UTF-8') libhdf5: 1.12.1 libnetcdf: 4.8.1 xarray: 0.18.3.dev137+g96c56836 pandas: 1.4.2 numpy: 1.22.3 scipy: 1.8.0 netCDF4: 1.5.8 pydap: None h5netcdf: None h5py: None Nio: None zarr: 2.11.3 cftime: 1.6.0 nctimeaxis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: None dask: 2022.04.1 distributed: 2022.4.1 matplotlib: None cartopy: None seaborn: None numbagg: None fsspec: 2022.3.0 cupy: None pint: None sparse: None setuptools: 62.1.0 pip: 22.0.4 conda: None pytest: 7.1.1 IPython: 8.2.0 sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-6804-06e33caf34cd972d50d584cb8487c70e0802aac656666452dfec714e89ca1964","text":"Use pytorch as backend for xarrays I would be interested in using pytorch as a backend for xarrays - because: a) pytorch is very similar to numpy - so the conceptual overhead is small b) [most helpful] enable having a GPU as the underlying hardware for compute - which would provide non-trivial speed up c) it would allow seamless integration with deep-learning algorithms and techniques Any thoughts on what the interest for such a feature might be ? I would be open to implementing parts of it - so any suggestions on where I could start ? Thanks"}
{"_id":"q-en-pydata__xarray-6823-30f61b966883fb1261d1623998f9dc23ea68e45903adec4181e474fefbc24667","text":"RuntimeError when formatting sparse-backed DataArray in f-string On upgrading from xarray 2022.3.0 to 2022.6.0, f-string formatting of sparse-backed DataArray raises an exception. Code does not error, or A breaking change is listed in the section of the docs.[X] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray. [X] Complete example — the example is self-contained, including all data and the text of any traceback. , returning the result. [X] New issue — a search of GitHub Issues suggests this is not a duplicate. Along with the versions below, I have confirmed the error occurs with both sparse 0.12 and sparse 0.13. <detailsINSTALLED VERSIONS commit: None python: 3.10.4 (main, Jun 29 2022, 12:14:53) [GCC 11.2.0] python-bits: 64 OS: Linux OS-release: 5.15.0-41-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enCA.UTF-8 LOCALE: ('enCA', 'UTF-8') libhdf5: 1.10.7 libnetcdf: 4.8.1 xarray: 2022.6.0 pandas: 1.4.2 numpy: 1.22.4 scipy: 1.8.0 netCDF4: 1.5.8 pydap: None h5netcdf: 0.12.0 h5py: 3.6.0 Nio: None zarr: None cftime: 1.5.2 nctimeaxis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.3.2 dask: 2022.01.0+dfsg distributed: 2022.01.0+ds.1 matplotlib: 3.5.1 cartopy: 0.20.2 seaborn: 0.11.2 numbagg: None fsspec: 2022.01.0 cupy: None pint: 0.18 sparse: 0.13.0 flox: None numpy_groupies: None setuptools: 62.1.0 pip: 22.0.2 conda: None pytest: 6.2.5 IPython: 7.31.1 sphinx: 4.5.0 <\/details"}
{"_id":"q-en-pydata__xarray-6857-9037342019cb0c32792f41e132585bb600c8d27ec5e03076db8e85e77763ff42","text":"Testing DataArray equality using built-in '==' operator leads to mutilated dictionary In previous versions of xarray, testing numerical equivalence of two DataArrays was possible using the built-in operator '==' and without side affects. Now in version 2022.6.0, when one DataArray lacks an attribute that the other DataArray has, the DataArray with the attribute is mutilated during comparison leading to an empty attrs dictionary. DataArray1 == DataArray2 should not have side affects. [X] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray. [X] Complete example — the example is self-contained, including all data and the text of any traceback. , returning the result. [X] New issue — a search of GitHub Issues suggests this is not a duplicate. No response <detailsINSTALLED VERSIONS commit: None python: 3.9.13 (tags\/v3.9.13:, May 17 2022, 16:36:42) [MSC v.1929 64 bit (AMD64)] python-bits: 64 OS: Windows OS-release: 10 machine: AMD64 processor: Intel64 Family 6 Model 85 Stepping 4, GenuineIntel byteorder: little LCALL: None LANG: None LOCALE: ('EnglishUnited States', '1252') libhdf5: None libnetcdf: None xarray: 2022.6.0 pandas: 1.4.3 numpy: 1.23.1 scipy: 1.9.0 netCDF4: None pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: None nctimeaxis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: None dask: None distributed: None matplotlib: 3.5.2 cartopy: None seaborn: None numbagg: None fsspec: None cupy: None pint: None sparse: None flox: None numpygroupies: None setuptools: 63.2.0 pip: 22.2.1 conda: None pytest: 7.1.2 IPython: 8.4.0 sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-6882-65c214c54c42daccd6a1183a18c284d8541fb7f6038f7eaa331fdf63d32e007b","text":"Cannot open NetCDF file if dimension with time coordinate has length 0 ( when decoding CF datetime) If a data set has a zero-sized coordinate that is a time index, reading fails. A is triggered when xarray tries to decode the array, as shown below:"}
{"_id":"q-en-pydata__xarray-6889-1e5fb69d6b07b0c652acfa69820fd95af693d41e188786d2b2bdceca6d4d7a52","text":"Alignment of dataset with MultiIndex fails after applying After applying the function to a dataset with a Multiindex, a lot of functions related to indexing are broken. For example, it is not possible to apply to itself anymore. The error is raised in the alignment module. It seems that the function does not find indexes that belong to the same dimension. I expected the alignment to be functional and that these basic functions work. [X] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray. [X] Complete example — the example is self-contained, including all data and the text of any traceback. , returning the result. [X] New issue — a search of GitHub Issues suggests this is not a duplicate. No response <detailsINSTALLED VERSIONS commit: None python: 3.10.5 (main, Jun 14 2022, 07:04:59) [GCC 10.3.0] python-bits: 64 OS: Linux OS-release: 5.15.0-41-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: ('enUS', 'UTF-8') libhdf5: 1.12.1 libnetcdf: 4.8.1 xarray: 2022.6.0 pandas: 1.4.2 numpy: 1.21.6 scipy: 1.8.1 netCDF4: 1.6.0 pydap: None h5netcdf: None h5py: 3.6.0 Nio: None zarr: None cftime: 1.5.1.1 nctimeaxis: None PseudoNetCDF: None rasterio: 1.2.10 cfgrib: None iris: None bottleneck: 1.3.4 dask: 2022.6.1 distributed: 2022.6.1 matplotlib: 3.5.1 cartopy: 0.20.2 seaborn: 0.11.2 numbagg: None fsspec: 2022.3.0 cupy: None pint: None sparse: 0.13.0 flox: None numpygroupies: None setuptools: 61.2.0 pip: 22.1.2 conda: 4.13.0 pytest: 7.1.2 IPython: 7.33.0 sphinx: 5.0.2 \/home\/fabian\/.miniconda3\/lib\/python3.10\/site- UserWarning: Setuptools is replacing distutils. (\"Setuptools is replacing distutils.\") <\/details"}
{"_id":"q-en-pydata__xarray-6938-97888fb9f6ad5cd1a7661899df169a95f9cbf81d15c140c1f67a92a34ee5afd4","text":"can modify original object This is kind of a convoluted example, but something I ran into. It appears that in certain cases can modify the original object, here the of a data variable that was swapped into being a dimension coordinate variable. I expected it not to modify the original object. [X] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray. [X] Complete example — the example is self-contained, including all data and the text of any traceback. , returning the result. [X] New issue — a search of GitHub Issues suggests this is not a duplicate. No response_ More experiments in . <details<\/details"}
{"_id":"q-en-pydata__xarray-6992-2b0606d1cb9d3ca8f24e26d10a0dc2547a5f235a8614abc7fbdaef72779a6986","text":"index refactor: more than on Dataset assumes that everything that is in and not in is a \"data variable\". However, since the index refactor we can end up with more than which breaks a number of stuff (e.g. the repr). Well it seems this assumption is now wrong. [ ] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray. [ ] Complete example — the example is self-contained, including all data and the text of any traceback. , returning the result. [ ] New issue — a search of GitHub Issues suggests this is not a duplicate. The error comes from here Bisected to - which probably does not help too much. <details<\/details"}
{"_id":"q-en-pydata__xarray-6999-2a21d99e4204d0865a4b758c5b97f54dedfbb6c2fdb071dc312562a6d4060aa4","text":"[Bug]: renamevars to dimension coordinate does not create an index We used to make a coordinate a dimension coordinate (instead of ). This results in the coordinate correctly being displayed as a dimension coordinate (with the *) but it does not create an index, such that further operations like fail with a strange . I expect one of two things to be true: does not allow setting dimension coordinates (raises Error and tells you to use setindex) checks for this occasion and sets the index correctly No response No response INSTALLED VERSIONS commit: None python: 3.9.1 (default, Jan 13 2021, 15:21:08) [GCC 4.8.5 (Red Hat 4.8.5-44)] python-bits: 64 OS: Linux OS-release: 3.10.0-1160.49.1.el7.x8664 machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: ('enUS', 'UTF-8') libhdf5: 1.12.0 libnetcdf: 4.7.4 xarray: 0.20.2 pandas: 1.3.5 numpy: 1.21.5 scipy: 1.7.3 netCDF4: 1.5.8 pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: 1.5.1.1 nctime_axis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: None dask: None distributed: None matplotlib: 3.5.1 cartopy: None seaborn: None numbagg: None fsspec: None cupy: None pint: None sparse: None setuptools: 49.2.1 pip: 22.0.2 conda: None pytest: 6.2.5 IPython: 8.0.0 sphinx: None"}
{"_id":"q-en-pydata__xarray-7003-4474202a425eed3b9f34ad0186e5b67d3cdad387a9578c768c79567202934a8c","text":"Indexes.getunique() TypeError with pandas indexes I also just tested the method that you mentioned and maybe noticed a related issue here, which I'm not sure is wanted \/ expected. Taking the above dataset , accessing this function results in an error: However, for it works: Originally posted by in"}
{"_id":"q-en-pydata__xarray-7101-523a7e69c89dd5ca571cba55433a1f840baa06e8f7325d1bb332269326ef81bb","text":"Broken state when using assigncoords with multiindex I was trying to assign coordinates on a dataset that had been created by using stack. After assigning the coordinates, the dataset was in a state where its length was coming out as less than zero, which caused all sorts of issues. I think the issue is with the updating of , perhaps in I expected to just be able to assign the coords and then print the array to see the result. [X] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray. [X] Complete example — the example is self-contained, including all data and the text of any traceback. , returning the result. [X] New issue — a search of GitHub Issues suggests this is not a duplicate. Here's a test (I put it in but maybe there is a better spot) <detailsINSTALLED VERSIONS commit: python: 3.8.13 (default, Mar 25 2022, 06:04:14) [Clang 12.0.1 ] python-bits: 64 OS: Darwin OS-release: 21.5.0 machine: arm64 processor: arm byteorder: little LCALL: None LANG: enAU.UTF-8 LOCALE: ('enAU', 'UTF-8') libhdf5: 1.12.2 libnetcdf: 4.8.1 xarray: 0.1.dev4312+ge678a1d. pandas: 1.5.0 numpy: 1.22.4 scipy: 1.9.1 netCDF4: 1.6.1 pydap: installed h5netcdf: 1.0.2 h5py: 3.7.0 Nio: None zarr: 2.13.2 cftime: 1.6.2 nctimeaxis: 1.4.1 PseudoNetCDF: 3.2.2 rasterio: 1.3.1 cfgrib: 0.9.10.1 iris: 3.3.0 bottleneck: 1.3.5 dask: 2022.9.1 distributed: 2022.9.1 matplotlib: 3.6.0 cartopy: 0.21.0 seaborn: 0.12.0 numbagg: 0.2.1 fsspec: 2022.8.2 cupy: None pint: 0.19.2 sparse: 0.13.0 flox: 0.5.9 numpy_groupies: 0.9.19 setuptools: 65.4.0 pip: 22.2.2 conda: None pytest: 7.1.3 IPython: None sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-7120-eb3ddadd4f3df781fb8a9d9c277750aa72855f48a87c50e30d83c80fa9fc6507","text":"Raise nicer error if passing a list of dimension names to transpose Hello, in xarray 0.20.1, I am getting the following error The expected result is No response No response No response <detailsINSTALLED VERSIONS commit: None python: 3.9.12 (main, Apr 5 2022, 06:56:58) [GCC 7.5.0] python-bits: 64 OS: Linux OS-release: 3.10.0-1160.42.2.el7.x8664 machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS LOCALE: ('enUS', 'ISO8859-1') libhdf5: 1.12.1 libnetcdf: 4.8.1 xarray: 0.20.1 pandas: 1.4.1 numpy: 1.21.5 scipy: 1.8.0 netCDF4: 1.5.7 pydap: None h5netcdf: 999 h5py: 3.6.0 Nio: None zarr: None cftime: 1.5.1.1 nctimeaxis: 1.4.0 PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.3.4 dask: 2022.02.1 distributed: 2022.2.1 matplotlib: 3.5.1 cartopy: 0.18.0 seaborn: 0.11.2 numbagg: None fsspec: 2022.02.0 cupy: None pint: 0.18 sparse: 0.13.0 setuptools: 61.2.0 pip: 21.2.4 conda: None pytest: None IPython: 8.2.0 sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-7147-0746b5b77aaecbfd67f4a9649ad18f466f3c50c6d6884b354076ac20c616c706","text":"Time decoding error message does not include the problematic variable's name If any variable in a Dataset has times that cannot be represented as objects, an error message will be raised. However, this error message will not indicate the problematic variable's name. It would be nice if it did, because it would make it easier for users to determine the source of the error. cc: xref: Unidata\/cftime This is a minimal example of the issue. The error message gives no indication that is the problem:"}
{"_id":"q-en-pydata__xarray-7150-be8fda750ed69b703cdfce1a12979b26c64f1e683e3e46138c646bdc196d78cc","text":"xarray.opendataset has issues if the dataset returned by the backend contains a multiindex As a follow up of this comment: I'm currently trying to implement a custom backend that allows me to also handle multiindices when loading a NetCDF dataset using . I'm using the following two functions to convert the dataset to a NetCDF compatible version and back again: Here is a small code example: While the manual patching for saving is currently still required, I tried to at least work around the function call in by creating a custom NetCDF Backend: but this works: So I'm guessing is performing some operation on the dataset returned by the backend, and one of those leads to a failure if there is a multiindex already contained. I expected that it doesn't matter wheter is called inside the backend or afterwards, and the same dataset will be returned each time. [X] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray. [X] Complete example — the example is self-contained, including all data and the text of any traceback. , returning the result. [X] New issue — a search of GitHub Issues suggests this is not a duplicate. No response I'm also open to other suggestions how I could simplify the usage of multiindices, maybe there is an approach that doesn't require a custom backend at all? <detailsINSTALLED VERSIONS commit: None python: 3.8.10 (default, Jan 28 2022, 09:41:12) [GCC 9.3.0] python-bits: 64 OS: Linux OS-release: 5.10.102.1-microsoft-standard-WSL2 machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: C.UTF-8 LOCALE: ('enUS', 'UTF-8') libhdf5: 1.10.5 libnetcdf: 4.6.3 xarray: 2022.9.0 pandas: 1.5.0 numpy: 1.23.3 scipy: 1.9.1 netCDF4: 1.5.4 pydap: None h5netcdf: None h5py: 3.7.0 Nio: None zarr: None cftime: 1.6.2 nctimeaxis: None PseudoNetCDF: None rasterio: 1.3.2 cfgrib: None iris: None bottleneck: None dask: None distributed: None matplotlib: 3.6.0 cartopy: 0.19.0.post1 seaborn: None numbagg: None fsspec: None cupy: None pint: None sparse: 0.13.0 flox: None numpy_groupies: None setuptools: 65.3.0 pip: 22.2.2 conda: None pytest: 7.1.3 IPython: 8.5.0 sphinx: 4.5.0 <\/details"}
{"_id":"q-en-pydata__xarray-7203-e97f75c192e0fed30f550981f4dff7c52a51624994d894a9a17585fc9ff49c59","text":"Avoid loading any data for reprs For \"small\" datasets, we load in to memory when displaying the repr. For cloud backed datasets with large number of \"small\" variables, this can use a lot of time sequentially loading O(100) variables just for a repr. Fast reprs! This dataset has 48 \"small\" variables On this repr takes 36.4s If I comment the condition I get 6μs. [x] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray. [x] Complete example — the example is self-contained, including all data and the text of any traceback. , returning the result. [x] New issue — a search of GitHub Issues suggests this is not a duplicate. No response No response <detailsINSTALLED VERSIONS commit: None python: 3.10.4 (main, Mar 24 2022, 17:43:32) [Clang 12.0.1 ] python-bits: 64 OS: Darwin OS-release: 21.5.0 machine: x8664 processor: i386 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: ('enUS', 'UTF-8') libhdf5: None libnetcdf: None xarray: 2022.3.0 pandas: 1.4.2 numpy: 1.22.4 scipy: 1.8.1 netCDF4: None pydap: None h5netcdf: None h5py: None Nio: None zarr: 2.11.3 cftime: None nctime_axis: None PseudoNetCDF: None rasterio: 1.2.10 cfgrib: None iris: None bottleneck: None dask: 2022.05.2 distributed: None matplotlib: 3.5.2 cartopy: 0.20.2 seaborn: 0.11.2 numbagg: None fsspec: 2022.5.0 cupy: None pint: None sparse: None setuptools: 62.3.2 pip: 22.1.2 conda: None pytest: None IPython: 8.4.0 sphinx: 4.5.0 <\/details"}
{"_id":"q-en-pydata__xarray-7229-00064fbb5e0ca85fb23bd427cc0eb5c4fd7d0728a165cd675d74d60dbbad8a3b","text":"overwrites coordinate attributes had some unintended consequences for , where coordinate attributes are getting overwritten by variable attributes. I guess this has been broken since . Coordinate attributes should be preserved. [X] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray. [X] Complete example — the example is self-contained, including all data and the text of any traceback. , returning the result. [X] New issue — a search of GitHub Issues suggests this is not a duplicate. I'm struggling to figure out how the simple change in brought this about. I tried tracing my way through the various merge functions but there are a lot of layers. Happy to submit a PR if someone has an idea for an obvious fix. <detailsINSTALLED VERSIONS commit: None python: 3.9.13 (main, May 27 2022, 16:56:21) [GCC 10.3.0] python-bits: 64 OS: Linux OS-release: 5.15.0-52-generic machine: x8664 processor: x8664 byteorder: little LCALL: None LANG: enUS.UTF-8 LOCALE: ('enUS', 'UTF-8') libhdf5: 1.12.2 libnetcdf: 4.8.1 xarray: 2022.10.0 pandas: 1.4.3 numpy: 1.23.4 scipy: 1.9.3 netCDF4: 1.6.1 pydap: None h5netcdf: 1.0.2 h5py: 3.7.0 Nio: None zarr: 2.13.3 cftime: 1.6.2 nctimeaxis: 1.4.1 PseudoNetCDF: None rasterio: 1.3.3 cfgrib: 0.9.10.2 iris: None bottleneck: 1.3.5 dask: 2022.10.0 distributed: 2022.10.0 matplotlib: 3.6.1 cartopy: 0.21.0 seaborn: None numbagg: None fsspec: 2022.10.0 cupy: None pint: 0.19.2 sparse: 0.13.0 flox: 0.6.1 numpy_groupies: 0.9.19 setuptools: 65.5.0 pip: 22.3 conda: None pytest: 7.1.3 IPython: 8.5.0 sphinx: None <\/details"}
{"_id":"q-en-pydata__xarray-7233-c449576585020b07fa4ae79979acbcb9d14dda17db69ab6cbe9e7aa515912cb7","text":"ds.Coarsen.construct demotes non-dimensional coordinates to variables demotes non-dimensional coordinates to variables All variables that were coordinates before the coarsen.construct stay as coordinates afterwards. [X] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray. [X] Complete example — the example is self-contained, including all data and the text of any traceback. , returning the result. [X] New issue — a search of GitHub Issues suggests this is not a duplicate. No response No response_"}
{"_id":"q-en-pydata__xarray-7347-1191cc4d03b2e276d871c1bad7dda963f20bf5abf1c092901d0b37d56bd98a51","text":"assigncoords reset all dimension coords to default (pandas) index See should preserve the index of coordinates that are not updated or not part of a dropped multi-coordinate index. See [X] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray. [X] Complete example — the example is self-contained, including all data and the text of any traceback. , returning the result. [X] New issue — a search of GitHub Issues suggests this is not a duplicate. No response No response_ <detailsXarray version 2022.11.0 <\/details"}
{"_id":"q-en-pydata__xarray-7391-24b45e74321dbcb08c4135ce241cc50283a257b51866fc6ee177d538c21f0980","text":"binary ops ignore option When doing arithmetic operations on two Dataset operands, the option is ignored and therefore attributes not kept. Minimal example: This is not true for DataArrays\/Variables which do take into account. Datasets to behave the same as DataArray\/Variables, and keep attributes during binary operations when option is set. PR is inbound."}
{"_id":"q-en-pydata__xarray-7393-e9769c1eecb964cadac114e837387cb8d2508722b2e87a86ac38d5ffab2724f5","text":"stack casts int32 dtype coordinate to int64 The code example below results in , because the data type of the coordinate is changed from 'i4' to 'i8'. I expect the result to be . Creating a MultiIndex should not change the data type of the Indexes from which it is built. [X] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray. [X] Complete example — the example is self-contained, including all data and the text of any traceback. , returning the result. [X] New issue — a search of GitHub Issues suggests this is not a duplicate. No response No response <detailsINSTALLED VERSIONS commit: None python: 3.10.8 (main, Oct 13 2022, 10:17:43) [Clang 14.0.0 (clang-1400.0.29.102)] python-bits: 64 OS: Darwin OS-release: 21.6.0 machine: x8664 processor: i386 byteorder: little LCALL: None LANG: None LOCALE: (None, 'UTF-8') libhdf5: 1.12.2 libnetcdf: 4.9.0 xarray: 2022.10.0 pandas: 1.5.1 numpy: 1.23.4 scipy: 1.9.3 netCDF4: 1.6.1 pydap: None h5netcdf: None h5py: 3.7.0 Nio: None zarr: None cftime: 1.6.2 nctimeaxis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: None dask: 2022.10.2 distributed: None matplotlib: 3.6.1 cartopy: 0.21.0 seaborn: None numbagg: None fsspec: 2022.10.0 cupy: None pint: None sparse: None flox: None numpygroupies: None setuptools: 65.5.0 pip: 22.1.2 conda: None pytest: None IPython: 8.6.0 sphinx: None (\"Setuptools is replacing distutils.\") <\/details"}
{"_id":"q-en-pydata__xarray-7400-a735479f590c55b0aa911a51cd7c7c80108c8a126e9bc9cde452c2c4048c5d48","text":"Add defaults during concat 508 [x] [x] Tests [x] Passes [x] Fully documented, including for all changes and for new API Contined on issue by removing exception when concat two datasets with disjoint variables and instead add the missing variable with"}
{"_id":"q-en-pylint-dev__pylint-4175-cc7a9dfd53097a1a7f9e9950ea4b7aa8e91d238b2d3ca9cc2ed7a932e745258c","text":"Pylint 2.7.0 seems to ignore the min-similarity-lines setting <!-- Hi there! Thank you for discovering and submitting an issue. Before you submit this, make sure that the issue doesn't already exist or if it is not closed. Is your issue fixed on the preview release?: pip install pylint astroid --pre -U --two Python source files that share 8 common lines min-similarity-lines=40 in the pylint config pylint 2.7.0 on the source files Before pylint 2.7.0, the min-similarity-lines setting was honored and caused shorter similar lines to be accepted. Starting with pylint 2.7.0, the min-similarity-lines setting seems to be ignored and the common lines are always reported as an issue R0801, even when the min-similarity-lines setting is significantly larger than the number of common lines. The min-similarity-lines setting should be respected again as it was before pylint 2.7.0. pylint 2.7.0 astroid 2.5 Python 3.9.1 (default, Feb 1 2021, 20:41:56) [Clang 12.0.0 (clang-1200.0.32.29)] Pylint 2.7.0 seems to ignore the min-similarity-lines setting <!-- Hi there! Thank you for discovering and submitting an issue. Before you submit this, make sure that the issue doesn't already exist or if it is not closed. Is your issue fixed on the preview release?: pip install pylint astroid --pre -U --two Python source files that share 8 common lines min-similarity-lines=40 in the pylint config pylint 2.7.0 on the source files Before pylint 2.7.0, the min-similarity-lines setting was honored and caused shorter similar lines to be accepted. Starting with pylint 2.7.0, the min-similarity-lines setting seems to be ignored and the common lines are always reported as an issue R0801, even when the min-similarity-lines setting is significantly larger than the number of common lines. The min-similarity-lines setting should be respected again as it was before pylint 2.7.0. pylint 2.7.0 astroid 2.5 Python 3.9.1 (default, Feb 1 2021, 20:41:56) [Clang 12.0.0 (clang-1200.0.32.29)]"}
{"_id":"q-en-pylint-dev__pylint-4339-8558ae36fe4c16d5d1f6025d337c1b4663c9e4c374478a57d7d06f0c4c770471","text":"Support output to a file to support tox on CI servers Currently , which means when running Pylint from within there seems to be no way to output the results to a single (for example HTML) file that a CI server can store for display. used to be close to what is needed, but it , and also created a different file for each module which is difficult to navigate. Here is the that relates to this. this feature for the same reason:"}
{"_id":"q-en-pylint-dev__pylint-4398-b845328da0613136a6d4a60a555f643f62b64320bcbb4ab06e5301ea8a0601ff","text":"Add --fail-on option to always return error code if specific issues (or issue types) are found We're using pylint in CI with two primary goals: there are no error-category issues - this can be achieved by looking at the exit code the overall linting score doesn't deteriorate too much - this can be achieved by using However if is used, and the score is above that, then it passes even if there are error-category issues detected. Essentially it's acting as a \"only throw fail (non-zero) codes if under this\", instead of a \"fail if under this, otherwise change nothing\". Two possible solutions I can think of here are: a configuration option to prevent from consuming other exit statuses. I.e. a way to say \"if score is under N, return error code X, regardless of all else, otherwise change nothing\". a new option like which means that if there are any code (error-category) issues, or any issues, then fail, otherwise change nothing."}
{"_id":"q-en-pylint-dev__pylint-4421-66c795ac7e43c3b07b3179d378af78564136a6d2859d5aaa81005775d1a6a8da","text":"Pylint 2.8.2 broke pylint-quotes <!-- Hi there! Thank you for discovering and submitting an issue. Before you submit this, make sure that the issue doesn't already exist or if it is not closed. Is your issue fixed on the preview release?: pip install pylint astroid --pre -U --See See in attached Running this configuration on a file containing: results in an exception: should not throw exceptions, and instead generate report complaining about the use of double quotes () and illegal variable name . Result of output: Additional dependencies: seems to have just changed versioning scheme: broke this line:"}
{"_id":"q-en-pylint-dev__pylint-4492-e582b6b5a4cbbde2ad04af227197b19cf2ef7d8e5aced958459e840c96b87b4d","text":"Allow parseable output in parallel of human-readable output Currently, when selecting a format like JSON using the argument, it will replace the \"normal\" output. It would be nice to have it generate an output-file instead so the normal human-readable output is still there. This makes sens in a CI environment like travis or gitlab where you can see the stdout and stderr of each job but where you also want to parse the output. Allowing multiple output formats would solve this. Maybe with a new CLI argument? Or by adding a separator to the current value? For example: pylint -f # Would redirect the JSON data to and still write the normal report to stdout pylint 1.8.1, astroid 1.6.0 Python 3.5.2 (default, Nov 23 2017, 16:37:01) [GCC 5.4.0 ]"}
{"_id":"q-en-pylint-dev__pylint-4516-c8a020df3c02c2bbf1c2bc0044270e16e6ecfea6850ef59b2d3a9c7768b47539","text":"Ignore clause not ignoring directories This is a different issue to . a directory and within that a directory . files and . Put syntax errors in both. , run . Pylint does not ignore , producing the message Pylint ignores the file . ignore-patterns does not skip non-top-level directories. <!-- Hi there! Thank you for discovering and submitting an issue. Before you submit this, make sure that the issue doesn't already exist or if it is not closed. Is your issue fixed on the preview release?: pip install pylint astroid --pre -U --create a (where c.py will generate a pylint message, so that we get output) (along with the appropriate \\init\\.py files) Run pylint: pylint --ignore-patterns=b Run pylint: pylint --ignore-patterns=a c.py is skipped for ignore-patterns=a, but not for ignore-patterns=b c.py should be skipped for both pylint 2.1.1 astroid 2.1.0-dev Python 3.6.3 (v3.6.3:, Oct 3 2017, 17:26:49) [MSC v.1900 32 bit (Intel)]"}
{"_id":"q-en-pylint-dev__pylint-4551-b63a8b7c210412348a4164ac29632fdc924e693149119728f9177380fa183e2b","text":"Use Python type hints for UML generation It seems that pyreverse does not read python type hints (as defined by ), and this does not help when you use as a default value : Output of pyreverse : ! I would like to see something like : in the output. pylint- 1.6.5, astroid 1.4.9 Python 3.6.0 (default, Dec 23 2016, 11:57:41) [MSC v.1900 64 bit (AMD64)]"}
{"_id":"q-en-pylint-dev__pylint-4604-0f1b640fad4216fade82ff1b8fc26e25e455a84a610848002bfd0eae9898869e","text":"unused-import false positive for a module used in a type comment should not be emitted. Result of output: This is a follow up to ."}
{"_id":"q-en-pylint-dev__pylint-4661-aa24ad865a5ce1c360e90ebb93e011bf4550153dcfa4d66216bb92cc654170c9","text":"Make pylint XDG Base Directory Specification compliant I have this really annoying directory in my home folder. From what I can tell (I don't do C or C++), this directory is storing data. The problem with this is, quite simply, that data storage has a designated spot. The folder. This is a part of the . A system that designates the folders for specific things like cached files (), configuration files (), and data files (), among other things. The point is to keep user home directories clean and the user sane. This should be pretty easy to implement. Simply change the variables\/constants for where these files are made and stored to the appropriate directory. Simple as that, even for a large codebase (if it was done right)."}
{"_id":"q-en-pylint-dev__pylint-4669-40476797b88dac37a40407fe7c46155ffeafd30dd1d137c7bac265ff541389cd","text":"Similarities checker with \"ignore-signatures\" option enabled ignores functions with docstring-only bodies follow-up. Given multiple files: : : Result of : error shouldn't be reported in this case. Note: if functions have bodies everything works correctly. You could add to the example functions above and no error would be reported. Result of output:"}
{"_id":"q-en-pylint-dev__pylint-4703-34e0723e61837b07362144e82893071afd4a1b97fb96c776135bb3d5c53d22ee","text":"Imports within TYPECHECKING can induce \"cyclic-import\" Suppose you have two modules: a.py and b.py. imports , but needs in order to do type annotations, then this is usually done by importing within a block. Unfortunately, this causes pylint to report . Possibly related to"}
{"_id":"q-en-pylint-dev__pylint-4812-fc72d47f83234e1f5d6bf03ac4bfba423bc8e65132c4a75d08cb455746f2e335","text":"Config File Does not Expand Environment Variables If a config file is provided that has an environment variable, such as \"%AppData%\", it fails. This can be fixed by changing from to"}
{"_id":"q-en-pylint-dev__pylint-4858-cbc7c30a6fbdf2f008b66dbe65dd6e4c98d2f593a7d69191ab5b17b3fe0619db","text":"Ignore class methods signatures with similarities \"ignore-signatures\" option enabled Initial already closed issue . One more related closed question with examples . Similarities \"ignore-signatures\" option currently only ignores functions' signatures. Class methods won't be ignored with the current implementation which would be very useful to avoid false-negative errors for classes. With similarities \"ignore-signatures\" option enabled, class methods' signatures are ignored the same way as functions'."}
{"_id":"q-en-pylint-dev__pylint-4970-3970fe5d08a0e01e4c487872d2d69f75d2949f19ac9c42fbba51c36793ccf536","text":"Setting to should stop pylint from checking duplicate code Setting to in the rcfile doesn't disable checking for duplicate code, it instead treats every line of code as duplicate and raises many errors. Setting to should disable the duplicate code check. It works that way in many other linters (like flake8). Setting a numerical value in flake8 to (e.g. ) disables that check. requests being able to disable , but it is still open"}
{"_id":"q-en-pylint-dev__pylint-5175-8de5052b0654462b39886f2bccd4fbdaf7fa93c78568458ab149bdc49fc0de94","text":"args is considered as missing in documentation Hi, I met an issue with the following code: ` Return this warning: W9015: \"args\" missing in parameter documentation (missing-param-doc) No warning must be returned as there is a documentation concerning args pylint 2.5.3 astroid 2.4.2 Python 3.7.2 (default, Feb 25 2019, 14:07:05) [GCC 4.8.5 (Red Hat 4.8.5-36)]"}
{"_id":"q-en-pylint-dev__pylint-5231-ada31dcfb525d49e493915e651d1c5c35def9836ab76a3df478fec2b556fcdfe","text":"False positive missing-param-doc on numpy style when omitted Would not have expected either or checks to have failed. VS Code 1.61.2 Windows 10 Pro 21H1 No response False positive missing-param-doc on numpy style when omitted Would not have expected either or checks to have failed. VS Code 1.61.2 Windows 10 Pro 21H1 No response"}
{"_id":"q-en-pylint-dev__pylint-5446-a549550711d8adbce75af07e7442c921983bb646fd9edad864e45e1b478850d8","text":"The duplicate-code (R0801) can't be disabled Originally reported by: Anonymous It's seems like it's not possible to disable the duplicate code check on portions of a file. Looking at the source, I can see why as it's not a trivial thing to do (if you want to maintain the same scope semantics as other #pylint:enable\/disable comments. This would be nice to have though (or I guess I could just cleanup my duplicate code). Bitbucket:"}
{"_id":"q-en-pylint-dev__pylint-5595-411b3c8942ae1654bfd5d0db8ad3d2c8ebf81abca85b2efefc8a16810f54fe4e","text":"The negative score is not very beginner friendly For the past months I've been under the impression that the negative score is not very friendly for beginners. It somewhat demotivating to run over your project just to get a as a score, which would make look like a super drastic tool, that's super nitpicky about everything. I think we could do a better job here at welcoming everyone to using . Instead of having a score, we could check if other tools are doing something different and see what works best in general. The negative score is not very beginner friendly For the past months I've been under the impression that the negative score is not very friendly for beginners. It somewhat demotivating to run over your project just to get a as a score, which would make look like a super drastic tool, that's super nitpicky about everything. I think we could do a better job here at welcoming everyone to using . Instead of having a score, we could check if other tools are doing something different and see what works best in general."}
{"_id":"q-en-pylint-dev__pylint-5613-c6923f104a9c24a78d5767f4b95a1d9519d3f22be34491320940f56d89ba5813","text":"Please provide a better entry point to pyreverse Currently, I have to do: I would like to do something like:"}
{"_id":"q-en-pylint-dev__pylint-5730-df0ea51781d4ba6c284f160e3b03c6dd2334bb5776440c3fee32e9270c15a2a4","text":"[crash] doc params extension: --accept-no-param-doc: conflicting option string(s) There is a crash when using the doc_params extension. No crash."}
{"_id":"q-en-pylint-dev__pylint-5839-e8ad0de77661e4974b1e5633ff08eda74f892230018f64f19ab4d6d66a725fab","text":"Pylint needs to forbid to re-use the msgid or symbol of old deleted msgid\/symbol Right now it's possible to reuse old msgid that were removed, for example, everything from the python 3 checker (the problem was detected with `), or mixed-indentation, bad-whitespace, bad-continuation... maybe other that I don't remember that we deleted. We have a mechanism in place for renamed message with old_name, but nothing for removed message. Related to and This is going to cause occasional bug, but more than that confusion and inconsistencies when searching for the msgid online. Impossible to use an old message id or symbol."}
{"_id":"q-en-pylint-dev__pylint-5859-24843078f090c7ab4ab41df7b3e4639b5d1e6a234efb9d3347927f06fa3b73b8","text":"\"--notes\" option ignores note tags that are entirely punctuation If a note tag specified with the option is entirely punctuation, pylint won't report a fixme warning (W0511). will return a fixme warning (W0511) for the first line, but not the second. macOS 11.6.1 No response"}
{"_id":"q-en-pylint-dev__pylint-5951-1b873ef442b4a767229e248bd64d1362ef57f26ed1d45a783e1492f91936591f","text":"DOC: pyreverse supported format not recognized missing puml, plantuml, mmd, etc Some supported formats are not listed in the error messaging: Here, I have made the mistake of prepending a to . The error output should list (and and and ) as . Without the preceding , the command runs as expected. All supported formats are listed when user makes a typo in the format argument OpenSUSE Tumbleweed"}
{"_id":"q-en-pylint-dev__pylint-6059-d4c0ed2ab503db61b1890b60b48f0fa864de9b42a2cd4c2f181cbd7381e91240","text":"Is required As noted by ,  is not currently covered. If this required then we should add a unit test, otherwise we can remove this method. N\/A No response No response"}
{"_id":"q-en-pylint-dev__pylint-6196-f710fac5302fb5cced8408d400b107262465ab74a6728698b6a8d39f6c2ff79a","text":"Use, misuse and confusion of decorator The work on showed that specifically enabling just a single message is sometimes not working as expected. This is due to either missing message declaration in a checkers decorator. Ideally I would like to implement a new, optional checker aimed at Pylint core development and custom checker development. This checker would focus on correct message declaration and handling. I already started working on it, but noticed that there is especially quite some problems with the decorator. I totally understand that, because while being extensively used across all checker classes, there is not a single word on its purpose and effect in the documentation. Its docstring is also not really helpful, you have to dig deeper into the codebase to understand what it does. Before continuing my work on the new checker, I want to address and clarify some questions about the intended use of. After that this issue can be closed with a PR that extends the documentation, and if necessary some code changes. A brief explanation on what does: It helps the decide whether a callback for the current node type should be called or not. This is done by adding the list of messages passed in as a attribute on method object. When a checker is to the attribute: if the attribute does not exist, this method will be to the list of callbacks if it exists and at least one message of the list is enabled, this method will be to the list of callbacks otherwise, it will not be to the list of callbacks and will therefore not be executed Essentially this means: main (and only) intent of the decorator is to improve performance the does not have any effect to apply the decorator has no negative functional consequences, it only affects performance (which, of course, is always an issue) the decorator but forgetting to add messages that are emitted from this callback can lead to false negatives: if none of the messages actually included in the decorator are enabled, it will not be called the decorator to a callback that not only emits messages but also does other things can have nasty side effects (false positives, false negatives, crashes) What I want to gain a consensus on: we want to keep the name ? I don't think it really conveys the effect it has on the decorated method. shall we regard as best practice for this decorator? Should every callback be decorated and the list of messages must always be correct and consistent, or should we rather advise to use it only in cases where we expect a significant speedup when this callback is omitted? we want to keep the restriction that this decorator only has an effect on the \"top level callbacks\"? Seeing that a lot of those callbacks can emit quite a lot of messages and the logic for the checks is often split up into several subfunctions, it would be more flexible and also more robust (as it is easier to keep track of the messages that need to be included in the decorator) if one could apply it on the subfunctions rather than on the top level callback. Looking over the code base there are quite a lot of places where the decorator is not used correctly: used with simply passing in all messages (which has the same effect as just omitting it): used on methods that are not a callback (which has no effect) Used, but not including all messages possibly emitted by the callback No response_"}
{"_id":"q-en-pylint-dev__pylint-6357-2fdec0cc05bb24e03001f7fdff2237997013137673f3d634add875143b6cdf80","text":"Crash When parsing the following file: &lt;!-- If sharing the code is not an option, please state so, but providing only the stacktrace would still be helpful. --pylint crashed with a and with the following stacktrace: If I remove the comment, things work as expected. If I call without , things work as expected. If I downgrade (and ) to astroid-2.9.3 pylint-2.12.2, things work as expected. No response no astroid error ;-) Fedora 35 astroid==2.11.2 dill==0.3.4 flake8==4.0.1 isort==4.3.21 lazy-object-proxy==1.7.1 mccabe==0.6.1 platformdirs==2.5.1 pycodestyle==2.8.0 pyflakes==2.4.0 pylint==2.13.5 toml==0.10.2 tomli==2.0.1 wrapt==1.13.3"}
{"_id":"q-en-pylint-dev__pylint-6358-e17f3b033c9b88b2303714189e8c3b51bfe6a2e617d9384428b6ea7b7374b89e","text":"option ignored The option is ignored in 2.14. Place two files that both read like this (ideally in their own dir so you can just run that dir): Proper output in 2.12: Caused in No response No response"}
{"_id":"q-en-pylint-dev__pylint-6386-1021410d76ba079918d3bea3302f182c43862e5636cdd6594264c75c5b6da6bd","text":"Argument expected for short verbose option The short option of the option expects an argument. Also, the help message for the option suggests a value should be provided. The long option works ok & doesn't expect an argument: Similar behaviour to the long option."}
{"_id":"q-en-pylint-dev__pylint-6412-f062c2430267377c99c547ade975cb19c6f067ca111c69e6309bef1ce7c1b908","text":"Support isinstance checks through ABCMeta for checkers and reporters Currently the PyLinter categorizes checkers through the class attribute. This is not very standard, and is one more barrier for others to write plugins. I propose Changing the interfaces to have a metaclass Change PyLinter to do isinstance checks to categorize checkers\/reporters in addition to keeping the old checks"}
{"_id":"q-en-pylint-dev__pylint-6506-29da278b8298e3847a8c40919c3ac975f485def50a323762d923c332d9b39ba4","text":"Traceback printed for unrecognized option A traceback is printed when an unrecognized option is passed to pylint. No response The top part of the current output is handy: The traceback I don't think is expected & not user-friendly. A usage tip, for example: No response No response"}
{"_id":"q-en-pylint-dev__pylint-6517-9f897f442236c7627e2b2701ab3ecbed22478a3e8ee681fa28adae98bcb88e57","text":"Pylint runs unexpectedly pass if in pylintrc Runs unexpectedly pass in 2.14 if a pylintrc file has . (Default pylintrc files have . 's own config was fixed in to comment it out, but this might bite existing projects.) No response No response"}
{"_id":"q-en-pylint-dev__pylint-6526-b2f19880c34fe26994a59fa3fbf00f0a65182ef76bb5cb478472a962a4b62eda","text":"How does pylint decide where it should put stat files? I am from the VS Code python extension team, working on building an extension for VS Code using pylint. We have pre-release version out. An issue was reported on the extension that it creates pylint stat files in the project directory. We are not specifically passing anything to change the stat file location (see issue here ). Generally, the stat files go into pre-set directory like %USERPROFILE%\/AppData\/Local\/pylint on windows. What can cause these to be written to the current working directory. We use pass the source code into pylint via stdin, and the only two arguments we use are , . Any clue to debug this would be helpful. I could not find fall back location for the stat files and when would the fallback location be used. We use to run pylint, and we send source to pylint via stdin. Relevant code here:"}
{"_id":"q-en-pylint-dev__pylint-6528-05f7036c654a89f6d13937a37899838805b94c0ea096d25aec6f6e846a2b5035","text":"Pylint does not respect ignores in mode Pylint does not respect the , , or setting when running in recursive mode. This contradicts the documentation and seriously compromises the usefulness of recursive mode. No response All of these commands give the same output: should be ignored by all of the above commands, because it is in an ignored directory (even the first command with no ignore setting should skip it, since the default value of is . For reference, the docs for the various ignore settings from : No response No response"}
{"_id":"q-en-pylint-dev__pylint-6556-6cd993a50cc8f5a8589f8ed2f7c6a77cc72eb7bcfbdfbc4fdfc7bc79305dfd8d","text":"Can't disable bad-option-value code on a computer with a somewhat new pylint (2.4.3 in my example). Get a warning like that I want to ignore, as I'm writing code compatible with python2 and python3. said warning with . a \"Bad option value\" when other people run their pylint version (example: 2.3.1; and by people, sometimes I mean docker instances ran from Jenkins that I would rather not rebuild or that depend on other people and I can't modify) to disable said error with a global is ignored is ignored To be able to write code that works on several versions of pylint and not having to make sure every computer in the company and every docker container has the same pylint version."}
{"_id":"q-en-pylint-dev__pylint-6820-3bda6ba574f744059091f431105afbda95536c93960358a092ff3bab051593db","text":"Using in CLI silently fail to load the plugin See Warning that load-plugin is not the right argument."}
{"_id":"q-en-pylint-dev__pylint-6903-57e53cd64576dbfde182136bdadbd0d95e137e396f913a60edf6697129e6dfe3","text":"Running pylint in Kubernetes Pod with --jobs=0 fails I run pylint in multiple parallel stages with Jenkins at a Kubernets agent with . The newly introduced function is called to determine the number of cpus to use and returns 0 in this case. This leads to a crash of pylint because the multiprocessing needs a value 0. I checked the function and found out the following values from the files that are read in above mentioned function: This leads to the calculation then in line which is cast to an and therefore 0 then. No response I expect pylint to not crash if the number of available cpu is misscalculated in this special case. The calculated number should never be 0. A possible solution would be to append a at the end of this line. I'm not sure if the same can happen for the calculation in line though, as I don't know the exact backgrounds of that files. Ubuntu 20.04 Kubernetes Version: v1.18.6 Python 3.9.12 No response"}
{"_id":"q-en-pylint-dev__pylint-6937-2528d5f0fc8b584d976e326e48e511cb01daa250128a2ea0355f5f65f8260769","text":"--disable ignored when --errors-only specified a python file () with a single line import notfoundpackage --disable=E0401 --errors-only test.py_ reports *** Module test E0401: Unable to import 'notfoundpackage' (import-error) Your code has been rated at 10.00\/10 (previous run: 10.00\/10, +0.00) 11.5"}
{"_id":"q-en-pylint-dev__pylint-7080-3cc52d4ec3fd3a9ae172bb30bb0e5fa07b93e017b22ebd5396fb8252111d3be0","text":"ignores When running recursively, it seems in my settings in is completely ignored src\\gen* should not be checked Windows 10.0. No response"}
{"_id":"q-en-pylint-dev__pylint-7114-da6534d899c41b971b21f3c2c6508ceed0316111460bd2144015a4720dfaca28","text":"Linting fails if module contains module of the same name Given multiple files: Which are all empty, running fails: However, if I rename , succeeds: Alternatively, I can also , but that shouldn't be necessary anymore. Running if is present fails while searching for an file. Running if is present should succeed. Result of output: This also has some side-effects in module resolution. For example, if I create another file : With the content: Running will run fine, but will fail. Not just for module a, but for module r as well. Again, if I rename to , will work perfectly."}
{"_id":"q-en-pylint-dev__pylint-7228-1479ffec9a9c401881fb867399738369ef143e368ae3b48a45448c50a4bfef34","text":"rxg include '\\p{Han}' will throw error config rxg in pylintrc with \\p{Han} will throw err .pylintrc: not throw error macOS 11.6.7"}
{"_id":"q-en-pylint-dev__pylint-7277-43c64bd6c75dcc6e6a3cf110d5e5ad5a1257e8d25f0dae1912f89053304ac550","text":"removes first item from when running from . This is the line where the first item from is removed. I think there should be a check to ensure that the first item is , or before removing. No response Check if , or before removing the first item from No response No response"}
{"_id":"q-en-pylint-dev__pylint-7993-6879fdb0b5744c32d23f5658eb302038b4c3d74189b87fbae47d02b81724c7a6","text":"Using custom braces in message template does not work Have any list of errors: On pylint 1.7 w\/ python3.6 - I am able to use this as my message template However, on Python3.9 with Pylint 2.12.2, I get the following: Is this intentional or a bug? No response Expect the dictionary to print out with as the key. No response No response"}
{"_id":"q-en-pylint-dev__pylint-8124-251f7481c4ca60b1174ee293fd97f48140814557049bb7e032e38aa7b2b8311b","text":"false positive 'useless-import-alias' error for mypy-compatible explicit re-exports Suppose a package has the following layout: Since the submodules here implement public APIs, imports and re-exports them, expecting users to import them from the public, top-level package, e.g. . Since the implementations of and are complex, they are split into and for better maintainability and separation of concerns. So looks like this: The reason for the aliases here is to be explicit that these imports are for the purpose of re-export (without having to resort to defining , which is error-prone). Without the aliases, popular linters such as will raise an \"implicit re-export\" error ( -- part of ). However, pylint does not currently understand this usage, and raises \"useless-import-alias\" errors. Example real-world code triggering pylint false positive errors: No response No \"useless-import-alias\" errors should be flagged. No response No response_"}
{"_id":"q-en-pylint-dev__pylint-8169-f0bcf9f7acfd6f33459f32c36f31d9c816fbf5e25d44ad075658be3419d06d0e","text":"False positive when importing from ` option Simply importing exceptions from the  library is giving this error. Here's an example of how we import them: Pycharm can find the exception classes just fine. I know they exist. It could have something to do with how the library is using , but I don't know too much about how that works to draw that conclusion. Also, note that we're using version 1.95.1 of . We use it in some critical paths, so we can't update it to the latest version quite yet. The configuration written below is what I've tried, but it seems based on googling that that doesn't stop all errors from being ignored regarding those modules. So I'm still getting the issue. No error to be reported Intel based 2019 Mac Book Pro. Mac OS 13.1 (Ventura). Fish shell. ccxt==1.95.1"}
{"_id":"q-en-pylint-dev__pylint-8281-212c2368123f26e625af30d4c081732eb27fbcd48a95716b6862f63162189554","text":"Support globbing patterns for source-roots It's quite inconvenient having to specify every source root for complex multi-package projects like For complex multi-package projects it would be nice to be able to specify source roots as instead of listing every one of them. IMHO, it's better to go with globbing patterns rather than with regexp patterns since those give better support for path-specific matching. No response"}
{"_id":"q-en-pylint-dev__pylint-8312-31cc534602ca87403112dc27cda903164b85905234a1436451b8718888cb845e","text":"Support globbing pattern for input specification As mentioned in and , lets consider supporting cc Add globbing support right in this line No response_"}
{"_id":"q-en-pylint-dev__pylint-8683-0860fcd1eff9f0d7fe8363ce6fb400113313d8a33987e71ce102615ec61a9613","text":"fail\/warn on using parallel execution with custom plugins According to documentation: Actually, it is possible but silently broken. then Pylint should inform a user about it in such cases. As for now, I could run: without any warning or error. Unfortunately, linting results are not the same as a single process linting, but Pylint silently pass. So, results are not predictable. Proposal: emit a warning or better explicitly fail on using parallel execution with custom Pylint plugins, because people usually don't read the documentation while things works."}
{"_id":"q-en-pylint-dev__pylint-8757-3b5ae5ad02bd219ebdfbbd47d4c61f25e4ed3219cb64af90588c313670a8a701","text":"Different output with --jobs=1 and --jobs=2 Originally reported by: Robert Spier (BitBucket: ) Parallelism (--jobs) changes the output of pylint. It's not just the order of the tests, --jobs=2 outputs lines of output compared to only 21 for --jobs=1. pylint 1.3.1 reports no lint errors. PWD\/pylintrc app\/codein app\/melange app\/soc app\/summerofcode tests 2&1 | head \\ Module codein.callback W: 17, 0: import missing (no-absolute-import) W: 18, 0: import missing (no-absolute-import) W: 19, 0: import missing (no-absolute-import) W: 20, 0: import missing (no-absolute-import) \\ Module W: 17, 0: import missing (no-absolute-import) W: 18, 0: import missing (no-absolute-import) W: 20, 0: import missing (no-absolute-import) W: 21, 0: import missing (no-absolute-import) PWD\/pylintrc app\/codein app\/melange app\/soc app\/summerofcode tests 2&1 | head \\ Module main E: 46, 2: print statement used (print-statement) E: 47, 2: print statement used (print-statement) E: 48, 2: print statement used (print-statement) E: 49, 2: print statement used (print-statement) E: 50, 2: print statement used (print-statement) \\ Module tests.test_utils E:658, 8: print statement used (print-statement) E:662,10: print statement used (print-statement) E:667, 8: print statement used (print-statement) This is with the current head of the pylint repository. (1889: tip) The source code being linted is Bitbucket:"}
{"_id":"q-en-pylint-dev__pylint-8799-935c566344f900bfa49e3e7ecbd2e81e46f8a191cc2f01e88d783b746d8ed91f","text":"Short circuit if all checks disabled Running \"pylint --disable=all\" takes more than 3s! Running pylint without \"disable=all\" on a little project (150-lines telegram bot) takes more than 8s. It is non-usable. No response It is unacceptable that even on an empty file pylint runs for at least 3 seconds. I use the VS Code extension in my project, which for a small example with a 150-line project reacts to changes in 8 (!) seconds. This is literally impossible to use. Windows 11 No response"}
{"_id":"q-en-pylint-dev__pylint-8819-42556da31caf8f03623011eb94b080ba47f50956444c65c515c3caeee8def5a3","text":"enable\/disable options in rcfile should not depend on the order with which they are specified Hello, i'm running ; i kinda understand why behaves differently than (where the first command enables everything and then disable , while the second command the disable option is overwritten by ) but i dont think it should be the same in the rcfile: the section is (by default) before the section, so if i want to have the same effect of command1 i need to swap the sections around. on the cli i can specify multiple enable\/disable options, but that's not allowed in the rcfile, so the current result is extremely counter-intuitive; and rcfile with is clear what result the user wants: to enable all checks except for some, and i shouldnt need to move config options around. can you please get that fixed? Thanks!"}
{"_id":"q-en-pylint-dev__pylint-8898-1f9f0413c4ab800354dfe7c8820057b42ba3c248caac18b8011d635cfaf2ff0b","text":"bad-names-rgxs mangles regular expressions with commas Since pylint splits on commas in this option, instead of taking a list of strings, if there are any commas in the regular expression, the result is mangled before being parsed. The config below demonstrates this clearly by causing pylint to crash immediately. I would expect any valid regular expression to be expressible in this option. If not directly, adding some way to escape commas so that this issue can be worked around. Pop! OS 22.04 No response"}
{"_id":"q-en-pylint-dev__pylint-8929-5ed9b7dc1f4d761a161d3265a7746bbcf2234ae57239283a4df7dde5b3159291","text":"Exporting to JSON does not honor score option <!-- Hi there! Thank you for discovering and submitting an issue. Before you submit this, make sure that the issue doesn't already exist or if it is not closed. Is your issue fixed on the preview release?: pip install pylint astroid --pre -U --pylint on some random Python file or module: As you can see this outputs some warnings\/scoring: run the same command but with to export it to JSON: The output doesn't contain the scores now anymore: execute it with again but also supply the option: The score is not outputted when exporting to JSON, not even when is activated. The score is to the JSON, at least when is activated."}
{"_id":"q-en-pytest-dev__pytest-10051-63fac7d3e594c691d3e841405b8b2cb2f89813abf9bce4dbd9107cf62044e5da","text":"caplog.getrecords and conflict gets decoupled from actual caplog records when is called. As a result, after is called, is frozen: it does not get cleared, nor does it get new records. During test set up it is as , but the latter gets in , which diverges the two objects. Arch Linux, Python 3.9.10:"}
{"_id":"q-en-pytest-dev__pytest-10081-b63faa87bd25c5f83a0d8e069d0fbdbfa704ceb983e9571c0aee162005812c5e","text":"unittest.TestCase.tearDown executed for classes marked with when running --pdb <!-- Thanks for submitting an issue! Quick check-list while reporting bugs: --- [x] a detailed description of the bug or problem you are having [x] output of from the virtual environment you are using [x] pytest and operating system versions [x] minimal example if possible Running will run the of classes that are decorated with on the class level. Identical to , but with the on the class level rather than on the function level. Minimal test (adapted from ), : Some versions (full below): Test is properly skipped normally: but when running with , the teardown seems executed: Full versions:"}
{"_id":"q-en-pytest-dev__pytest-10115-557ee9e0410c1c2e35a8b8b70040ef2cb51fc5ca039110697ea69c97804e92cd","text":"Get rid of atomicwrites (unmaintained) PyPI has started enforcing 2-factor-auth for maintainers of various popular packages: For context, here is the mail I got: project. Unfortunately, this has caused the maintainer of to go on what I can only describe as a rampage, first deleting the project from PyPI (and then finding out it's not possible to restore it): then simply declare the project as unmaintained outright: No matter what the outcome of this will be, IMHO, given those actions I do not feel comfortable with trusting this dependency for something as popular as pytest. The library itself , and we only use it on Windows. It's MIT-licensed. Should we just copy the parts we need into pytest instead?"}
{"_id":"q-en-pytest-dev__pytest-10343-795c1189fb857f83096453f8a19e38836a35b358becf77426d0eb76a2ca55c7f","text":"Old-style hookimpl warning has no location information The old-style hookimpl deprecation warning from has no way to see where the culprit is implemented. I'm now getting: with no easy way to figure out what the problem is. I have 12 plugins installed, all of which might have a , and I'd rather not have to find out manually which one is the culprit. The error message should show either the plugin that's coming from, or at least the Python file it's in. Old-style hookimpl warning has no location information The old-style hookimpl deprecation warning from has no way to see where the culprit is implemented. I'm now getting: with no easy way to figure out what the problem is. I have 12 plugins installed, all of which might have a , and I'd rather not have to find out manually which one is the culprit. The error message should show either the plugin that's coming from, or at least the Python file it's in."}
{"_id":"q-en-pytest-dev__pytest-10356-ec0b99a62e4a78d85c134b4e8c2d58a66ddcfec2197d42f014c019c22efeea44","text":"Consider MRO when obtaining marks for classes When using pytest markers in two baseclasses and , inheriting from both of those baseclasses will lose the markers of one of those classes. This behavior is present in pytest 3-6, and I think it may as well have been intended. I am still filing it as a bug because I am not sure if this edge case was ever explicitly considered. If it is widely understood that all markers are part of a single attribute, I guess you could say that this is just expected behavior as per MRO. However, I'd argue that it would be more intuitive to attempt to merge marker values into one, possibly deduplicating marker names by MRO. I'd expect and to be markers for , but this only actually is the case with this metaclass. Please note that the repro case is Python 2\/3 compatible excluding how metaclasses are to (this needs to be taken care of to repro this issue on pytest 6) Consider MRO when obtaining marks for classes When using pytest markers in two baseclasses and , inheriting from both of those baseclasses will lose the markers of one of those classes. This behavior is present in pytest 3-6, and I think it may as well have been intended. I am still filing it as a bug because I am not sure if this edge case was ever explicitly considered. If it is widely understood that all markers are part of a single attribute, I guess you could say that this is just expected behavior as per MRO. However, I'd argue that it would be more intuitive to attempt to merge marker values into one, possibly deduplicating marker names by MRO. I'd expect and to be markers for , but this only actually is the case with this metaclass. Please note that the repro case is Python 2\/3 compatible excluding how metaclasses are to (this needs to be taken care of to repro this issue on pytest 6) Fix missing marks when inheritance from multiple classes <!-- Thanks for submitting a PR, your contribution is really appreciated! Here is a quick checklist that should be present in PRs. [] Include documentation when adding new features. [ ] Include new tests or update existing tests when applicable. [X] Allow maintainers to push and squash when merging my commits. Please uncheck this if you prefer to squash the commits yourself. If this change fixes an issue, please: for more information. Unless your change is trivial or a small documentation fix (e.g., a typo or reword of a small section) please: for details. Write sentences in the past or present tense, examples: Improved verbose diff output with sequences. Terminal summary statistics now use multiple colors. Also make sure to end the sentence with a . [x] Add yourself to in alphabetical order. --"}
{"_id":"q-en-pytest-dev__pytest-10371-dbf7349d77a7922ffc0022bda0a7f32be3e033b5ac516968e6426f0020dbe510","text":"RFE: allow to selectively disable loggers from command-line A common debugging strategy is to study the logs. But sometimes the program issues excessive logging messages, necessitating the selective disabling of babbler loggers. This SO captures the crux & solution of this Request For Enhancement: Although the proposed SO solution of a new option works perfectly fine, it is annoying to have to patch every new project, plus, it does not support auto-completion e.g. on bash. Would it make sense to include such capability into core code? In any case, I'm suggesting the new option to be called, to fit with the existing option names starting with `."}
{"_id":"q-en-pytest-dev__pytest-10442-347046ff723ca24e2f050a8b7fc4c4329e130d3f19a2f8a4ddb24698426110b1","text":"Keep temporary directory for failing tests only Currently pytest will keep the last 3 root temporary directories for each testing session, for all tests. This proposal is to change this behavior to only keep the directories for failed tests, instead of for all tests. This would save significant disk space for test suites which manipulate large amounts of data in the temporary directories. The behavior of keeping the last 3 root temporary directories would be kept unchanged. From: EDIT: after some discussion, the full proposal is defined in"}
{"_id":"q-en-pytest-dev__pytest-10482-c1d998d27aff9f809d94ea29acb559488c73c63e3e38b10ce279dea0dddac2aa","text":"Short test summary doesn't show the test name when skipping from a fixture I'm using Pytest 7.0.1on Ubuntu 18.04 with Python 3.6.9. Consider a test: and a fixture defined in that will skip a test based on some conditional check. Then running the test with pytest shows something like: The summary shows that some test was skipped but there's no indication which test was skipped. Instead, it should show the test name rather than the location in the fixture where the was called from. If there are multiple tests that are skipped from various locations, matching a test with its skip condition becomes impossible. There are some similar issues in , , which may be related."}
{"_id":"q-en-pytest-dev__pytest-10552-f26f46a067cb27f56b9a18479a9f7345b993b4b0f243f5aafb1368dfcf87ffd5","text":"Inconsistent support for staticmethod\/classmethod Pytest discovery & running of staticmethods\/classmethods is inconsistent. Here's an example: The most worrysome cases are and , which are not discovered by pytest. I think that there should at least be a warning or error to alert the user that their test code cannot be run. <details<summaryHere's the full output from running : <\/summary<\/detailspython v3.9.15, pytest v7.2.0, ubuntu 20.04 <details<summaryOutput of : <\/summary<\/details"}
{"_id":"q-en-pytest-dev__pytest-10624-308e8e0e2056acb486043fccb141bf5edd34e6410e4581d86f042e65c5ff8ce0","text":"when is dict containing zero value results in ZeroDivisionError Pytest behaves differently when comparing dictionaries containing zero values compared to lists containing same values pytest==7.2.0 Ubuntu 22.04"}
{"_id":"q-en-pytest-dev__pytest-10758-371d720a95f87809ccef288d17c929b3ffede0b67c2b21e6d19b64afec204d18","text":"Walrus operator causes different behavior in PyTest. I am currently testing the following function but find that the function passes the test in the normal Python terminal but fails in a PyTest run. The walrus operator is relatively new and not many people use it. I think that there may be an inconsistency in the execution environment."}
{"_id":"q-en-pytest-dev__pytest-10893-0b04847ce93bf93ad5a3f665ab0153cfd1b88abc2cb1e8954b47e45dd8f4a713","text":"Use of causes on Python 3.12 alpha The current Python 3.12 alpha has made a change to , deprecating the parameter and replacing it with an parameter. Something in Pytest's temp path fixtures calls an function which calls with the parameter. When warnings are treated as errors, this makes pytest fail any test using temp paths. Add to cpython 3.12 (alpha) has an API to test results (python\/cpython). This would not be an issue, except it was designed to trigger a warning if the test result doesn't have such a method (hence e.g. python\/cpython). This means when using pytest as runner for unittest tests and running an error is triggered as pytest's test result (which seems to be ) does not support this protocol. Now obviously this should be non-blocking, as hopefully nobody is running CI which blocks on testing 3.12, but there you go, I figure an early forewarning can't hurt."}
{"_id":"q-en-pytest-dev__pytest-10988-de8ea1255ac9932b9ba52a03fd666e1150a4ee02df3ada46fe44b1ab1a6c25dd","text":"Pytest trying to check if custom argument is a file crashes due to filename being too long I have a custom flag defined in , and when I try to assign it to a value that is too long pytest crashes before ever getting to my code. This reproduces even if the flag isn't defined, and even if the current working directory is . Failing example: If I reduce the length of the flag, I get the expected behavior for my project, and this different and expected error from my pytest MVP: I did a little digging into my version of pytest (7.0.0) to make sure I wasn't doing something wrong, but it looks like there is a blind call to with a path constructed from the argument in : It seems to me like there should be a try or something here, since in cases like mine the argument may not be a file at all, and that can cause OS level errors. Operating System: Ubuntu 20.04 LTS [x] a detailed description of the bug or problem you are having [x] output of from the virtual environment you are using [x] pytest and operating system versions [x] minimal example if possible Pytest trying to check if custom argument is a file crashes due to filename being too long I have a custom flag defined in , and when I try to assign it to a value that is too long pytest crashes before ever getting to my code. This reproduces even if the flag isn't defined, and even if the current working directory is . Failing example: If I reduce the length of the flag, I get the expected behavior for my project, and this different and expected error from my pytest MVP: I did a little digging into my version of pytest (7.0.0) to make sure I wasn't doing something wrong, but it looks like there is a blind call to with a path constructed from the argument in : It seems to me like there should be a try or something here, since in cases like mine the argument may not be a file at all, and that can cause OS level errors. Operating System: Ubuntu 20.04 LTS [x] a detailed description of the bug or problem you are having [x] output of from the virtual environment you are using [x] pytest and operating system versions [x] minimal example if possible"}
{"_id":"q-en-pytest-dev__pytest-11041-fbb2b9fe10de92b93c926cc1e0f7266d9cc02195b4a8adf8eccfff6649c0c102","text":"UnboundLocalError: cannot access local variable 'x' where it is not associated with a value There seems to be a regression in pytest version when a walrus operator is used in an assert line. Code: Fails the test with error: in pytest version , whereas with pytest version it passes successfully. My Python version is . Looks like it has to do with PR ."}
{"_id":"q-en-pytest-dev__pytest-11044-1aa3709cea7343bf3d85bc4ba8a786a26aea09652c80ee560bbb48636c94d25f","text":"Issue warning\/error if 'testpaths' does not match any files\/folders We should issue a warning (or even an error?) if does not match any files or folders. I think an error is reasonable, even if it might break some incorrectly-configured suite out there. Originally posted by in This is not really a bug, but an intended (albeit questionable) behavior: The values of are actually globs, so globbing for in the root yields nothing. Given it finds nothing, pytest will behave as if called from the command-line without any parameters, which makes it search recursively from the current directory looking for to collect. If you create the directory, then pytest will correctly search in that directory only. I agree those 2 facts are surprising: fact that is a glob. This is but easy to overlook, probably we should add a glob to the example there. silently not finding anything, and then proceeding as usual. I don't think we can do anything more for 1, but for 2 seems like we should at least emit a warning if is defined but does not match anything."}
{"_id":"q-en-pytest-dev__pytest-11047-73697cfca2e11232a7428b6d74cc557e0c30d83234c16b7d147eeb3e42b43d43","text":"support sub-second granularity\/precision in (strftime ) tl;dr pytest processing strftime specifier from arguments would allow me to accurately merge log messages from disparate sub-systems Tests I run have pytest log messages that print at the second granularity for the datetimestamp, e.g. . At the same time, other log file messages not generated by pytest print sub-second datetimestamps, e.g. . When reviewing the various logs, there are many message from other system components that are printing many log messages per second. Because pytest log messages are lacking sub-second precision, I am unable to align pytest log messages within other system log messages. For example, the system-under-test generates a log file like: and the driver of tests, pytest, generates a log file like: The pytest log messages cannot be precisely ordered among the other log messages that occurred during the datetime second , there were many things that occurred in the other system components within that second. Given the following pytest code To add sub-second granularity, it seems sensible to add within the but then I see the confusing output of pytest logging is ignoring the strftime specifier! I want pytest log messages to print sub-second granularity, e.g. process strftime within settings. <!-- A clear and concise description of what you want to happen. --Support strftime specifier in the various settings for date-format, e.g. , , . <!-- Provide examples of real-world use cases that this would enable and how it solves the problem described above. --In my complex testing system, this means all log messages would be printed to millisecond precision. This allows engineers investigating issues to more accurately merge disparate testing system logs by their natural ordering mechanism of a datetimestamp. <!-- Have you tried to workaround the problem using a pytest plugin or other tools? Or a different approach to solving this issue? Please elaborate here. --I can set the format to include . However, it's a little confusing to have to manipulate log datetimestamps by two different mechanisms, and . On the command-line run: This prints datetimestamps with millisecond precision <br \/It is more intuitive for pytest to process the Python strftime specifier within all options."}
{"_id":"q-en-pytest-dev__pytest-11125-405750a89099c574240364010d7bbc9a893519be4307cee64d2c2ba2ba4c506c","text":"Pytest 7.3.2 changes in behaviour regarding and In , we have two test suites - the unit tests at and the integration test suite at . Both and are listed in testpaths- We then run either or using \/. Each \/ dir contains a file, which adds some options using . One option that is common to both test suites is . Before 7.3.2, this setup seemed to work, we could run both unit tests and integration tests without issue. But on 7.3.2 (perhaps since ?) we get the following error: Is this an issue in our configuration, or a bug? Should we no longer use testpaths to list all the test suites? <details<summarypip list output<\/summary<\/detailsXref"}
{"_id":"q-en-pytest-dev__pytest-11143-9e133965f714bdd34fe727d44fab4dabd99b69ed1ddd2078cea74a8a133e8e9f","text":"Rewrite fails when first expression of file is a number and mistaken as docstring <!-- Thanks for submitting an issue! Quick check-list while reporting bugs: --- [x] a detailed description of the bug or problem you are having [x] output of from the virtual environment you are using [x] pytest and operating system versions [x] minimal example if possible use report an error it seems a error in collecting testcase"}
{"_id":"q-en-pytest-dev__pytest-11148-6f7b7e4b569e0e62b9b3e2cc07039d419b4689406f24b9e1d70dbe47b3ef4130","text":"Module imported twice under import-mode=importlib In pmxbot\/pmxbot I'm attempting to switch pmxbot off of pkgresources style namespace packaging to PEP 420 namespace packages. To do so, I've needed to switch to for the and re-organize the tests to avoid import errors on the tests. Yet even after working around these issues, the tests are failing when the effect of doesn't seem to have had any effect. Investigating deeper, I see that initializer is executed and performs its actions (setting a class variable ), but when that happens, there are two different versions of present, one in and another found in : I haven't yet made a minimal reproducer, but I wanted to first capture this condition."}
{"_id":"q-en-pytest-dev__pytest-11160-c94fb3b2858fbff7964f45ca6cd164561f668222475d7bae04ffb99d479cde9f","text":"() improperly matches warning When trying to pop a specific warning from a WarningsRecorder instance, the wrong warning is returned. I believe the issue is that pop uses issubclass I believe the correct comparison should be: Here is a minimum working example that triggers the buggy behavior: The test output is pytest 7.2.1 on archlinux. virtual environment is a clean conda environment with only python and pytest (and their dependencies installed from conda-forge). If this is indeed a bug, I'm happy to open a PR with my proposed solution. () improperly matches warning When trying to pop a specific warning from a WarningsRecorder instance, the wrong warning is returned. I believe the issue is that pop uses issubclass I believe the correct comparison should be: Here is a minimum working example that triggers the buggy behavior: The test output is pytest 7.2.1 on archlinux. virtual environment is a clean conda environment with only python and pytest (and their dependencies installed from conda-forge). If this is indeed a bug, I'm happy to open a PR with my proposed solution."}
{"_id":"q-en-pytest-dev__pytest-11178-8046c3a71b465d541a93425008741a558cbe892267e58d52770adb65e9a9e0a1","text":"fails with When using to test float and one of the objects in the statement contain I see the following TypeError: TypeError: unsupported operand type(s) for -: 'float' and 'NoneType'. macOS 12.6.3 python 3.10.4 pytest 7.2.1"}
{"_id":"q-en-pytest-dev__pytest-11217-abd0246f3fb61ad20b0fd3ffd73528625b24215957bddbf02716051a8904ba00","text":"fixtures: show test as skip location if skipped from an xunit setup function PR made it so that if a fixture calls , the skip location is shown as the test function, not the fixture. But it excluded xunit setup fixtures from this. I suspect this was done to make a pre-existing test pass, however I think that the same reason for fixtures applies to xunit fixtures just as well, so we shouldn't exclude it. Would also remove a string-hack that was used to implement this exclusion..."}
{"_id":"q-en-pytest-dev__pytest-5103-64d5b74cbb2e27f79825c4005253afbf375565f2461121825415cdd954bad29c","text":"Unroll the iterable for all\/any calls to get better reports Sometime I need to assert some predicate on all of an iterable, and for that the builtin functions \/ are great - but the failure messages aren't useful at all! For example - the same test written in three ways: A generator expression A list comprehension A for loop The only one that gives a meaningful report is the for loop - but it's way more wordy, and asserts don't translate to a for loop nicely (I'll have to write a or a helper function - yuck) I propose the assertion re-writer \"unrolls\" the iterator to the third form, and then uses the already existing reports. [x] Include a detailed description of the bug or suggestion [x] of the virtual environment you are using [x] pytest and operating system versions [x] Minimal example if possible"}
{"_id":"q-en-pytest-dev__pytest-5205-a38f17bf02ca0e371d2ea9efee4903694df134ee1e184d537900b9558edd166c","text":"Invalid XML schema for <propertiestags in JUnit reports The problem: JUnit breaks when it reads an XML generated by pytest if plugins make use of . This behavior happens with newer versions of hypothesis (). In fact, as already mentioned in , is adding inside which seems to be wrong (it should be inside ). See: . It happens with all junit families. Reproducing: I was trying to create a PR to fix this, but when I saw code I realized that what is needed to do could have more implications that I though. I think that nobody uses this feature with JUnit (as it breaks) and removing that is something to think about."}
{"_id":"q-en-pytest-dev__pytest-5221-5f83b66fbca032031e2e2dc2f9684c242a18f483caf2b59f5fddfdd1d45b86de","text":"Display fixture scope with It would be useful to show fixture scopes with ; currently the only way to learn the scope of a fixture is look at the docs (when that is documented) or at the source code."}
{"_id":"q-en-pytest-dev__pytest-5227-56c0d0f28c80841d7b0aa71144c147fbbbc1b6bda6ae15b7d7c4ebe49313836f","text":"Improve default logging format Currently it is: I think (module name) would be very useful here, instead of just the base filename. (It might also be good to have the relative path there (maybe at the end), but it is usually still very long (but e.g. could be substituted therein)) Currently it would look like this: Using instead:"}
{"_id":"q-en-pytest-dev__pytest-5254-d0244e52c787584a58043f86eb1773569fa72febb5992b3a3072cbccbd3ee25c","text":"does not correctly hide fixtures of the same name (it misses its dependencies) From This works: the parameter in the test correctly hides the fixture so the unique pytest node has id (instead of there being two nodes because of the fixture). However if the fixture that is hidden by the parameter depends on another fixture, that other fixture is mistakenly kept in the fixtures closure, even if it is not needed anymore. Therefore the test fails:"}
{"_id":"q-en-pytest-dev__pytest-5262-626b1b2f300bf70250af392aff25e08c9b118a1ddb42ea2165342c6067ef29c0","text":"pytest.capture.EncodedFile mode should not include (binary) <!-- Thanks for submitting an issue! Here's a quick checklist for what to provide: --- [x] a detailed description of the bug or suggestion Exception when youtube-dl logs to pytest captured output. Youtube-dl looks for in to decide whether to writes or . incorrectly advertises , the mode of the underlying stream. Its method raises an exception when passed . [x] output of from the virtual environment you are using [x] pytest and operating system versions [x] minimal example if possible"}
{"_id":"q-en-pytest-dev__pytest-5281-539414c2318bb9bc871b98d9774cf7e9cec652d3a10fb2f24fd474b67753de65","text":"Invalid unicode in with Jython With: Jython 2.7.1 and pytest 4.5.0, I'm having the error below: The error happens because has the following unicode literal: . I'm not sure about the proper fix... in that function seems to be doing a and later does , so, I'm not even sure if we're dealing with or there on Python 2 because it's mixed (my guess is bytes, but I'm not really sure). If it's bytes, doing it an actual literal instead of a literal -- i.e.: remove the -- does make Jython happy (would that be an appropriate solution?)"}
{"_id":"q-en-pytest-dev__pytest-5356-ba560a6d585e38a35e399a467dad3cd45944d1b5be73484309096559feeb1588","text":"Version 4.6.0 skips tests without apparent reason Since version 4.6.0 pytest skips tests without apparent reason: https:\/\/travis- [x] pytest and operating system versions: pytest 4.6.0 on Ubuntu Xenial @ Travis CI [x] minimal example if possible: link above I can't understand why it happens, so I had to immediately rollback to 4.5.0 and fix this version in dev dependencies."}
{"_id":"q-en-pytest-dev__pytest-5404-73511f2a43a3cba6f8ee34e6d9e3d81aaf9806a7b67f9ff67bfdec2ec3950859","text":"KeyError: 'wrapped' with from xxx import yyy [X] Include a detailed description of the bug or suggestion [ ] of the virtual environment you are using [X] pytest and operating system versions [X] Minimal example if possible I implemented and for automated testing. When I run the , I get a when I perform a . In the example I included, I do not get the KeyError if I comment out the line: ... from sty import fg ... No virtual environment used. Python: 3.6.5 & 3.7.1 PyTest: 4.4.0 OS: MacOS Mojave 10.14.4 I created a very simple . If I clone the repo and call either or in the root directory, I get the error. My apologies if this bug is obvious or I am doing something wrong."}
{"_id":"q-en-pytest-dev__pytest-5413-6f74777a70565c6d94eadb4135acf8192af55ed57b4bb66a27815b7b65c52e90","text":"str() on the context variable doesn't behave same as normal exception catch Pytest 4.6.2, macOS 10.14.5 prints But prints In order to get the full error message, one must do , which is documented, but this is a different interaction. Any chance the behavior could be changed to eliminate this gotcha? Pip list gives"}
{"_id":"q-en-pytest-dev__pytest-5495-32438f13f7f079614b9b896b1a6296a780a5b78d90b2d910155e47b1477282b8","text":"Confusing assertion rewriting message with byte strings The comparison with assertion rewriting for byte strings is confusing: 52 is the ASCII ordinal of \"4\" here. It became clear to me when using another example: Not sure what should\/could be done here."}
{"_id":"q-en-pytest-dev__pytest-5550-42a08161d59d3a4094f2802fb6b2daad8590bb296cef96b0606dedaf638258fc","text":"xunit2 format does not contain a root <testsuitestag junitfamily=xunit2 option still generates an old xml format tests report. junitfamily=xunit2 creates the .xml file in the legacy formatOS: MAC OS Mojave == 10.14.5 Sample output from the tests by running:"}
{"_id":"q-en-pytest-dev__pytest-5559-b1ea60fd2ca86821f00d9ad15ffd4e6a3b1df1e0207ebb9fd8c009c4e2a66325","text":"pytest stepwise doesn't work with xfail strict failures recommended pytest-runner in setuprequires means packages fail to install often The recommendation to add to means that all users of that package end up with an unnecessary pytest-runner package installed. This is bad because it bypasses pip hashes and"}
{"_id":"q-en-pytest-dev__pytest-5631-3b46150624518cc8e3410484ebbcea5a5eb9be25691104ec4e8be93265795a86","text":"ValueError when collecting tests that patch an array <!-- Thanks for submitting an issue! Here's a quick checklist for what to provide: --I'm trying to run pytest with a test file that contains patch where \"new\" is an array, for example: from import patch ([-5.5, 3.0])) ... This works fine with pytest 3.1.3, but when using pytest 3.6.0 the following error is received upon collection: Seems like a bug, that was introduced by the following fix: when using like: ([-5.5, 3.0])), p.new is an array and the check: \"p.new in sentinels\" returns an array of booleans instead of a boolean which causes the ValueError."}
{"_id":"q-en-pytest-dev__pytest-5692-c28c45c2651b9dafe1ce583e87818b2712a8603d2211a473008e20d88caa4841","text":"Hostname and timestamp properties in generated JUnit XML reports Pytest enables generating JUnit XML reports of the tests. However, there are some properties missing, specifically and from the XML element. Is there an option to include them? Example of a pytest XML report: Example of a junit XML report:"}
{"_id":"q-en-pytest-dev__pytest-5787-ea66b93810f3aa51f49e2fc01c950a40fc1c2d1d8fcec3a0887825c82540b1b9","text":"exception serialization should include chained exceptions given some simple tests: when run without xdist it displays whole exception trace nicely : but when run with xdist (), it just displays the last one: my setup:"}
{"_id":"q-en-pytest-dev__pytest-5808-d96e16d25fd9e54f3f9cb4d8d06cf1aafc372cd029193a994af8b662ce35386e","text":"Lexer \"python3\" in --pastebin feature causes HTTP errors The option currently submits the output of to using : For some , this will raise a \"HTTP Error 400: Bad Request\". As an example: with the attached . This is the underlying cause for the problems mentioned in . The call goes through fine if is changed from to . This would seem like the right thing to do in any case: the console output of a run that is being uploaded is not Python code, but arbitrary text."}
{"_id":"q-en-pytest-dev__pytest-5809-d96e16d25fd9e54f3f9cb4d8d06cf1aafc372cd029193a994af8b662ce35386e","text":"Lexer \"python3\" in --pastebin feature causes HTTP errors The option currently submits the output of to using : For some , this will raise a \"HTTP Error 400: Bad Request\". As an example: with the attached . This is the underlying cause for the problems mentioned in . The call goes through fine if is changed from to . This would seem like the right thing to do in any case: the console output of a run that is being uploaded is not Python code, but arbitrary text."}
{"_id":"q-en-pytest-dev__pytest-5840-95a2bb1da7e571b53082e40f21ede25e701857ecfd736b3cbf2d7cf97f5100e8","text":"2 ImportError while loading conftest (windows import folder casing issues) 1 works fine. after upgrade to 5.1.2, the path was converted to lower case"}
{"_id":"q-en-pytest-dev__pytest-5980-19fabf81638364bec4904302c5947e925eee7523367bbb5f8f66751851322fba","text":"Provide an alternative to --result-log After discussion in , mentions he would like to provide a replacement to in the core before removing it (). This potentially is an easy contribution given that we have the plugin already which can be used as a starting point. I would like for us to discuss how that \"log file\" will look like in this issue. I understand the rationale is to provide a line-based log file, which can be parsed using standard tools. I have used a log file in the past where each line was a JSON object, something like: pytest would then write each line to the file during , like does now. I suppose we also want to add an option to replay the tests in a log file, so users can reproduce a previous run that was saved to a log? you also mentioned that would not be an appropriate replacement, can you elaborate on why?"}
{"_id":"q-en-pytest-dev__pytest-6116-1e14e967272f73bfb4077e47cd9fdacb4fde9336eb872ed927b29b0cd76283c4","text":"pytest --collect-only needs a one char shortcut command I find myself needing to run very often and that cli argument is a very long to type one. I do think that it would be great to allocate a character for it, not sure which one yet. Please use up\/down thumbs to vote if you would find it useful or not and eventually proposing which char should be used. Clearly this is a change very easy to implement but first I want to see if others would find it useful or not. pytest --collect-only needs a one char shortcut command I find myself needing to run very often and that cli argument is a very long to type one. I do think that it would be great to allocate a character for it, not sure which one yet. Please use up\/down thumbs to vote if you would find it useful or not and eventually proposing which char should be used. Clearly this is a change very easy to implement but first I want to see if others would find it useful or not."}
{"_id":"q-en-pytest-dev__pytest-6186-b4247aff070678e5847601dd0571b4ca66ca60076d0a87ca055edae7b6f4bcb7","text":"Show deprecation warning if junitfamily is not set Show a deprecation warning if the user has not configured ()"}
{"_id":"q-en-pytest-dev__pytest-6197-e3c21587f4b84e96f573be00e0d94cf13987f789830d7504957a83a4cdb4a27e","text":"Regression in 5.2.3: pytest tries to collect random files This was caught by our build server this morning. It seems that pytest 5.2.3 tries to import any file under the current directory. (We have some package that is only used on windows and cannot be imported on linux.) Here is a minimal example using tox that reproduces the problem (I'm running on Debian 10 with Python 3.7.3):"}
{"_id":"q-en-pytest-dev__pytest-6202-2b2613f25ce2059e60d1abb77772890b9e65a7b5947d14bf055c97f4f9922255","text":"'.[' replaced with '[' in the headline shown of the test report The replaced with in the headline shown with long report output. The same problem also causing the vscode-python test discovery error. I trace back the source code. The headline comes from line 148. comes from line 437 The headline comes from line 306 This line of code causes the problem. We should replace it with . After changing this, run , pass all the tests and resolve this issue. But I can't find this line of code for what purpose."}
{"_id":"q-en-pytest-dev__pytest-6214-c696b03a4df092f82ae8f86a6b945ddfe4db200f85855a29a88026208cc8575e","text":"--setup-plan and --setup-only seem to claim different things I have the following example: When running with , I get the following output (indicating extra teardown of fixture1 and 2): When running with , the SETUP and TEARDOWN markers occur where my actual setup and teardown code executes: --setup-plan and --setup-only seem to claim different things I have the following example: When running with , I get the following output (indicating extra teardown of fixture1 and 2): When running with , the SETUP and TEARDOWN markers occur where my actual setup and teardown code executes:"}
{"_id":"q-en-pytest-dev__pytest-6283-c56a9cda7ad2720a2eaeee1a436b6015943dba8df4e426165a04dce59bd455eb","text":"Calls to () in skipif causes all logging calls to be duplicated to stderr Using pytest 5.2.2 \/ 5.3.0, when a function called from a uses a logging function, the captured log calls are duplicated to stderr. Minimal working example: Results in the following. Notice \"Schmift\" is printed both to \"Captured stderr call\" and \"Captured log call\". Removing the logging call from results in the expected behaviour, \"Schmift\" is not duplicated to stderr:"}
{"_id":"q-en-pytest-dev__pytest-6323-118fce84fbb711362522b538e94acf7ba6f28758573d8de460c3ced63cd51f10","text":"disallow absolute and non-normalized paths for mktemp followup to this is an potential issue and attack vector, absolute paths are no tmpdir and escaping paths aren't either, just normalizing would also break the world so we should only ever accept normalized relative paths for it"}
{"_id":"q-en-pytest-dev__pytest-6368-b15531fbb9eaab93d0fffd7aa0559347afa66d3be28351e8399527bbdee94607","text":"Module re-write doesn't work with non dist-info based installations More context behind this issue is available at: TL;DR: The function doesn't detect modules that can be rewritten, if they are installed by a method which doesn't adopt the dist-info format. An easy way to reproduce the problem is to: in one environment and in another and compare the output for: Because of this problem, rpm maintainers are unable to run the tests for pytest-mock, since they rely on which creates egg-info directories."}
{"_id":"q-en-pytest-dev__pytest-6680-ffdd4162aece9ade8bd1a6035797a4c2e66fe4911a9c120a37d9e4f98091bb24","text":"Improve deprecation docs for Node.fromparent In the \"Node Construction changed to Node.fromparent\" section in the deprecation docs, we definitely need to add: [x] An example of the warning that users will see (so they can find the session on google). [x] The warning should point to the deprecation docs. [x] Show a \"before -after\" example. [x] ensure fromparent will not support config\/session"}
{"_id":"q-en-pytest-dev__pytest-7046-aeec9e2c45a475c6af13732047a0ebcd63e12eb9ebc3e83f28a23caa784ad517","text":"Expressions match against folder structure above pytest root Say I have a test file If I want to match only , this works as expected However if I want to execute only using similar means: both tests are collected because is in the parent path even though it might not have anything to do with the test environment. Is this expected behaviour?"}
{"_id":"q-en-pytest-dev__pytest-7122-86ae247cacd0760fa8e61a9a6931d972f554c18c7edee8e4e7e2d4b5e260318a","text":"-k mishandles numbers Using . It seems that pytest cannot handle keyword selection with numbers, like . Considering the following tests: Selecting with works: But selecting with doesn't, as I get all tests: If I make it a string though, using , then it works again: I see there are some notes about selecting based on keywords here but it was not clear if it applied to this case: So, is this a bug? Thanks!"}
{"_id":"q-en-pytest-dev__pytest-7151-af64bacde8acd76c6084eec7879a4f2eb75c6738d84e854d9ea365dfd5704194","text":"unittest.TestCase cleanup functions not invoked on test failure stdlib unittest style cleanup functions registered with are not invoked when a test fails. It appears this issue was introduced in pytest version 5.4.0, examples below for version 5.4.1. Ubuntu 18.04.3 LTS Python 3.6.8 pytest 5.4.1"}
{"_id":"q-en-pytest-dev__pytest-7158-0112529d6f9a3ed92fd8299912d80146e3b7d1a6f58466b352b53106647beaeb","text":"Wrong relative path in skip report when tested file is upper than invocation directory The problem is if tested file is upper than invocation directory . It occurs in skip report. Example: path of tested file is: and is called in location: . Instead it should be."}
{"_id":"q-en-pytest-dev__pytest-7186-0ccae9829a492428db22ac2ea7655196905c53b5368b19db4a9514611a119c00","text":"pytest.doctest.issetuppy raises with LC_ALL=C and UTF-8 chars in runs into an exception when collecting tests from a repository where contains UTF-8 characters, but the locale is set to . Minimal example: : To reproduce: Tested on CentOS 7.7.1908:"}
{"_id":"q-en-pytest-dev__pytest-7205-0c70d4ba487a39359332abdf602df5b3203258e0a9ddca7514d1152e6fe55537","text":"BytesWarning when using --setup-show with bytes parameter With Python 3.8.2, pytest 5.4.1 (or latest master; stacktraces are from there) and this file: when running (note the to turn on ByteWarning and treat it as error), I get: Shouldn't that be using or something rather than (implicitly) ?"}
{"_id":"q-en-pytest-dev__pytest-7220-72d472d024d219648e5b296ee3310a58c9edb199fdb4587f75ce79508b64db62","text":"Wrong path to test file when directory changed in fixture Files are shown as relative to new directory when working directory is changed in a fixture. This makes it impossible to jump to the error as the editor is unaware of the directory change. The displayed directory should stay relative to the original directory. The full output is:"}
{"_id":"q-en-pytest-dev__pytest-7231-5a310350154fa8860e244d04a589efaba559e30b523147388320a7b59fac490b","text":"warn when logging fails Now an expectation will be thrown and written on the output... but no warning is thrown... so the test run can't be marked as failed."}
{"_id":"q-en-pytest-dev__pytest-7236-ed0bf9b8acbaf5c859c117ef2b6293789d2df655f80ec38751247d030e6c1416","text":"unittest.TestCase.tearDown executed on skipped tests when running --pdb With this minimal test: test is properly skipped: but when running with , the teardown seems executed: I would have expected the test to be skipped, even with . With , test was also skipped with , so this seem something that have changes between 5.4.2 and 5.4.1. (I would have loved to, but I don't have time to send a PR these days)"}
{"_id":"q-en-pytest-dev__pytest-7283-ed0bf9b8acbaf5c859c117ef2b6293789d2df655f80ec38751247d030e6c1416","text":"unittest.TestCase.tearDown executed on skipped tests when running --pdb With this minimal test: test is properly skipped: but when running with , the teardown seems executed: I would have expected the test to be skipped, even with . With , test was also skipped with , so this seem something that have changes between 5.4.2 and 5.4.1. (I would have loved to, but I don't have time to send a PR these days)"}
{"_id":"q-en-pytest-dev__pytest-7314-fedac4e59b0c8d695ab11a144de3bba000e0eb804155dbf34fc388e07d96f950","text":"pytest~=4.6: While implementing a test for I've hit this: Looking at the repo, it seems to have been fixed on master by as a part of But it still persists in the branch. The fix should be trivial: just add a fallback variable value before if-blocks. No need to backport that whole PR (unless somebody thinks that it should be done, of course). Ref: [4.6.x] Add a fallback for the term report letter When plugins return report objects in an unconventional state, may skip entering if-block branches that declare the variable. It would lead to UnboundLocalError: local variable 'letter' referenced before assignment being raised. This change fixes this by setting the initial value of the before the if-block cascade so that it always has a value. [ ] Include documentation when adding new features. [ ] Include new tests or update existing tests when applicable. [ ] Allow maintainers to push and squash when merging my commits. Please uncheck this if you prefer to squash the commits yourself. Unless your change is trivial or a small documentation fix (e.g., a typo or reword of a small section) please: for details. Write sentences in the past or present tense, examples: Improved verbose diff output with sequences. Terminal summary statistics now use multiple colors. Also make sure to end the sentence with a . [ ] Add yourself to in alphabetical order."}
{"_id":"q-en-pytest-dev__pytest-7324-54d4f4634754518527382661c3ed1569e6e769b2c21a2bdf166aea24fb449e46","text":"Pytest crashes the interpreter on debug build for 3.8+ Short reproducer Related issue for improvement of this behavior:"}
{"_id":"q-en-pytest-dev__pytest-7352-dd8459771ce312ed2bbfb987807536787b5766b14669a675aba597beaa142412","text":"FileNotFoundError: [Errno 2] No such file or directory: '\/tmp\/pytest-of-jenkins\/pytest-' Same issue as in . I think there are still some edge cases where this is not handled correctly. I am running a series of concurrent pytest processes and was able to reproduce the issue. Here is the stack trace: A potential mitigation to this problem might be to generate the numbered dir with a random suffix rather than a sequential one, unless some code depends on this logic:"}
{"_id":"q-en-pytest-dev__pytest-7373-94ae5bd6ff4b626a903a1b7c83e16c9403d8b7dc60498874f95c54aaf77aee76","text":"Incorrect caching of skipif\/xfail string condition evaluation Version: pytest 5.4.3, current master pytest caches the evaluation of the string in e.g. . The caching key is only the string itself (see in ). However, the evaluation also depends on the item's globals, so the caching can lead to incorrect results. Example: Running . Expected: is skipped, is not skipped. Actual: both are skipped. I think the most appropriate fix is to simply remove the caching, which I don't think is necessary really, and inline into ."}
{"_id":"q-en-pytest-dev__pytest-7432-d42cc4642a5e42ab82afbfeb8d3dab61c84f1b21d1bc5ab27eb775f9306513ce","text":"skipping: --runxfail breaks location reporting pytest versions: 5.4.x, current master When \/ marks are used to skip a test, for example the expected skip location reported should point to the item itself, and this is indeed what happens when running with : However, adding breaks this: The is only about xfail and should not affect this at all. Hint: the bug is in , the hook."}
{"_id":"q-en-pytest-dev__pytest-7466-2fdef014cbf3215afa52c6eb3fe3a6f5db665214b5333d7899f3c4759ccf8e10","text":"Support NOCOLOR to disable colored output is a variable. The general behavior is: A command line application emits colour when attached to a TTY, but no colour otherwise e.g. when attached to a pipe. stops the application from emitting colour even when attached to a TTY My extension to the standard is to add which forces the application to emit colour escape sequences even when not outputting to a TTY e.g. if we want to pipe into a log file with colour. Originally posted by in"}
{"_id":"q-en-pytest-dev__pytest-7468-3fce2a87faddc8ccab5ee4dc4d2b634c311d26a310eb7d0a4a74b9e865eedef9","text":"--log-file should create subdirectories If you provide a path with a subdirectory, which does not exist, it crashes with. For example execute produces something like this: Maybe someone want to add something like this However, there is the possibility to say that someone is not supposed to pass a directory path there. Originally posted by in"}
{"_id":"q-en-pytest-dev__pytest-7481-7482ef59c149b57b6f63f4ed5803a91279392b3b0ec973477f2ab96596a28db9","text":"pytest 6 (?): UnicodeEncodeError in terminalwriter on Windows Just got this on GitHub Actions with the pytest 6 rc: This is the test using pytest-bdd in qutebrowser: I can't reproduce this consistently - if I'm reading the traceback right, the test failed (probably flaky?) and then pytest failed while printing the traceback... Note that I also have pytest-instafail installed and it shows up in the stacktrace. Not sure if this is actually related to pytest 6, but I've never seen it happen before."}
{"_id":"q-en-pytest-dev__pytest-7490-6bec1679a5738110f22e11e887f1c743401f7737e06ec981d740659a3a7662c8","text":"Pytest 6: Dynamically adding xfail marker in test no longer ignores failure <!-- Thanks for submitting an issue! Here's a quick checklist for what to provide: --With pytest 5.x, we can dynamically add an xfail to a test object using (see example below). In 5.x this treated the failing test like a a test marked statically with an . With 6.0.0rc0 it raises. <details<\/details- [ ] pytest and operating system versions Pytest 6.0.1rc0 and MacOS 10.14.5 With 5.4.3 With 6.0.0rc0"}
{"_id":"q-en-pytest-dev__pytest-7499-777634deea9ce572703d336b47985b79de042f05889746286f71bd9876418a2b","text":"suggestion: improve raises confusing message when matching [x] a detailed description of the bug or suggestion I would like to check if strings have parenthesis in them when match is not None, so that people have that lightbulb moment, rather than head-scratching when it presents two identical strings as not matching. [ ] output of from the virtual environment you are using I believe this is not applicable right now. I'm happy to provide one if you believe other modules or libraries are causing this behavior. [x] pytest and operating system versions Windows 10 64-bit (I know), all updates installed Pytest 5.4.3 [x] minimal example if possible edited 22:31 (same day) as the example given was the fixed code including escapes and r prefix I believe a more helpful error message"}
{"_id":"q-en-pytest-dev__pytest-7500-0c1350d651eb22f89a86219bb0cbc5b8952603196317577dc6dc97b3e2d7322a","text":"Cleaning up temporary directories occasionally raises PermissionError On Windows, I'm finding that pytest occasionally raises an exception starting with while cleaning up its temporary directories. Below is an example of the output of a test session in which the exception arises. The test file contains only the function shown in the output. A necessary condition for the exception is that pytest's base temporary directory already contains at least three temporary directories to cause pytest to try to clean up at least one directory. Also, the exception occurred more often when the computer was under load. ============================= test session starts ============================= platform win32 -- Python 3.7.7, pytest-5.4.3, py-1.9.0, pluggy-0.13.1 rootdir: C:\\Users\\Desktop\\pytest-garbage collected 1 item F [100%] ================================== FAILURES =================================== ___ testtemp ___ tmppathfactory = TempPathFactory(givenbasetemp=None, trace=<pluggy.tracing.TagTracerSub object at 0x0000026E365FECC8, basetemp=None) def testtemp(tmppathfactory): for in range(1000): ....\\Programs\\Miniconda3-64\\envs\\pytest-garbage\\lib\\site-packages\\pytest\\ in mktemp basename = self.ensurerelativetobasetemp(basename) ....\\Programs\\Miniconda3-64\\envs\\pytest-garbage\\lib\\site-packages\\pytest\\ in ensurerelativetobasetemp if (self.getbasetemp() \/ basename).resolve().parent != self.getbasetemp(): ....\\Programs\\Miniconda3-64\\envs\\pytest-garbage\\lib\\site-packages\\pytest\\ in getbasetemp prefix=\"pytest-\", root=rootdir, keep=3, locktimeout=LOCKTIMEOUT ....\\Programs\\Miniconda3-64\\envs\\pytest-garbage\\lib\\site-packages\\pytest\\ in makenumbereddirwithcleanup considerlockdeadifcreatedbefore=considerlockdeadifcreatedbefore, ....\\Programs\\Miniconda3-64\\envs\\pytest-garbage\\lib\\site-packages\\pytest\\ in cleanupnumbereddir trycleanup(path, considerlockdeadifcreatedbefore) ....\\Programs\\Miniconda3-64\\envs\\pytest-garbage\\lib\\site-packages\\pytest\\ in trycleanup if ensuredeletable(path, considerlockdeadifcreatedbefore): ....\\Programs\\Miniconda3-64\\envs\\pytest-garbage\\lib\\site-packages\\pytest\\ in ensuredeletable if not (): ....\\Programs\\Miniconda3-64\\envs\\pytest-garbage\\lib\\ in exists () self = WindowsPath('') def stat(self): \"\"\" Return the result of the stat() system call on this path, like () does. \"\"\" E PermissionError: [WinError 5] Access is denied: 'C:\\Users\\\\AppData\\Local\\Temp\\pytest-of-\\garbage--fd35-4f5b-b6c5-\\.lock' ....\\Programs\\Miniconda3-64\\envs\\pytest-garbage\\lib\\ PermissionError =========================== short test summary info =========================== FAILED - PermissionError: [WinError 5] Access is deni... ============================== 1 failed in 0.83s ============================== It seems that sometimes the operating system continued to actually delete the files and directories inside an old directory even after the function (below) completed the call in its first statement to . Then, the second statement found that lingering directory, which renamed to the form . While was attempting again to delete its contents, the operating system finished actually deleting them, and the exception occurred. def cleanupnumbereddir( root: Path, prefix: str, keep: int, considerlockdeadifcreatedbefore: float ) -None: \"\"\"cleanup for lock driven numbered directories\"\"\" for path in cleanupcandidates(root, prefix, keep): trycleanup(path, considerlockdeadifcreatedbefore) for path in (\"garbage-\"): trycleanup(path, considerlockdeadifcreatedbefore) I tested simply reversing the two statements, so that pytest cleans old directories before numbered directories, and that appeared to prevent the exception in my testing. The operating system is Windows 10.0. Build , the file system is NTFS on a solid-state drive, I'm using a conda environment, and produces: Package Version ------------------ ------------------- atomicwrites 1.4.0 attrs 19.3.0 certifi 2020.6.20 colorama 0.4.3 importlib-metadata 1.7.0 more-itertools 8.4.0 packaging 20.4 pip 20.1.1 pluggy 0.13.1 py 1.9.0 pyparsing 2.4.7 pytest 5.4.3 setuptools 47.3.1.post20200622 six 1.15.0 wcwidth 0.2.5 wheel 0.34.2 wincertstore 0.2 zipp 3.1.0 I also encountered the same exception using pytest version 6.0.0rc1, although the session output differs because pytest* defers the clean-up until exit: ============================= test session starts ============================= platform win32 -- Python 3.7.7, pytest-6.0.0rc1, py-1.9.0, pluggy-0.13.1 rootdir: C:\\Users\\Desktop\\pytest-garbage collected 1 item . [100%] ============================== 1 passed in 2.67s ============================== Error in atexit.runexitfuncs: Traceback (most recent call last): File \"c:\\users\\programs\\miniconda3-64\\envs\\pytest-garbage\\lib\\\", line 1356, in exists () File \"c:\\users\\programs\\miniconda3-64\\envs\\pytest-garbage\\lib\\\", line 1178, in stat return (self) PermissionError: [WinError 5] Access is denied: 'C:\\Users\\\\AppData\\Local\\Temp\\pytest-of-\\garbage--f05a-46d7-bd84-\\.lock' Is swapping the statements within a good way to resolve this issue?"}
{"_id":"q-en-pytest-dev__pytest-7521-435aaf2d8723de95bd2f47ab52a8b96617c49d9a09a552f4779f1b93d78e469e","text":"pytest 6.0.0rc1: capfd.readouterr() converts \\r to \\n I am testing pytest 6.0.0rc1 with Fedora packages. This is the first failure I get, from borgbackup 1.1.13. I've distilled a reproducer: pytest 5: pytest 6: This is Fedora 32 with Python 3.8 (the original failure in borgbackup is Fedora 33 with Python 3.9). I could have not found anything about this change in the changelog nor at hence I assume this is a regression. I've labeled it as such, but feel free to change that."}
{"_id":"q-en-pytest-dev__pytest-7535-94e6a30b42c7fe0181796f50ea12f38689dacd21461b6bf1abca6028a2a10de2","text":"pytest 6: Traceback in contains repr of The tests fail with pytest 6: This is because of (, \"code: remove last usage of \") - it removed the , but the format string uses , so now we get the repr of the object instead of the repr of a string. I believe this should continue to use so the output is the same in all cases. cc"}
{"_id":"q-en-pytest-dev__pytest-7571-5e146db95630e7abf84b0fd6d80e2f3d5110d240d23d22f66029bf1f6c5642a2","text":"caplog fixture doesn't restore log level after test From the documentation at , \"The log levels set are restored automatically at the end of the test\". It used to work, but looks broken in new 6.0 release. Minimal example to reproduce: It prints \"0\" for pytest<6, \"42\" after."}
{"_id":"q-en-pytest-dev__pytest-7637-6182810cfca65e4db632db5a8a729126407439d11754c065b9320d9a8f842bfa","text":"Reintroduce warnings postponed in 6.0 A few warnings were introduced near the 6.0 release, so we can't comply with the \"2 versions minimum with warnings\", so for 6.0 these warnings were suppressed in We should reintroduce them in 6.1."}
{"_id":"q-en-pytest-dev__pytest-7648-a25f7a559117544f9fa69533a827f388f8c75c0a53d9267679b7c90d1a256a37","text":"Since pytest 6.0.0, pylint complains about unimplemented abstract methods in custom collector Hi, we have implemented a custom pytest collector that takes testcases from YAML files. Starting with pytest 6.0.0, pylint reports the following about that collector: The collector has worked fine for a long time, and it still works fine. The line pylint reports this on is the class definition of a collector class that is based on pytest.File: The whole source code is here: Versions: Python 3.8.0 (default, Oct 15 2019, 17:49:23). This happens on all Python 3.x versions we used pytest 6.0.1 with (3.5, 3.6, 3.7, 3.8), on macOS and Ubuntu. pylint 2.4.4 astroid 2.3.3 platform linux -- Python 3.8.0, pytest-6.0.1, py-1.9.0, pluggy-0.13.1 plugins: cov-2.10.0, yagot-0.5.0, requests-mock-1.8.0 Please let me know in case a collector has to implement these methods."}
{"_id":"q-en-pytest-dev__pytest-7673-4658ba455912def092d72631fcfc3e447999c0c8a8d08661673df7fe2ff0adc3","text":"logging: handler level restored incorrectly if caplog.setlevel is called more than once pytest version: 6.0.1 The fix in (backported to 6.0.1) has a bug where it does a \"set\" instead of \"setdefault\" to the . So if there are multiple calls to , the level will be restored to that of the one-before-last call, instead of the value before the test. Will submit a fix for this shortly."}
{"_id":"q-en-pytest-dev__pytest-7749-8e37a6de51d8f608a3180fb361582b24318985ece72e2ecb5d002c2437a5b253","text":"Crash with NameError through decorator during collection pytest crashes when trying to collect the following file: The failure representation works with : I've started writing a test, but it fails in both modes like the first case above:"}
{"_id":"q-en-pytest-dev__pytest-7939-e52d65ed897ad0190cbc2952c2b20709c32a1b2637ebdc4cce74a20e005b59ae","text":"[Feature] Allow a --sw-skip shorthand cli arg like --sw itself permits The stepwise plugin exposes a shorthand option for the stepwise itself, however it requires a longer arg only for skip, I think these should be consistent and should offer shorthand versions for both. Expected:"}
{"_id":"q-en-pytest-dev__pytest-7982-cf639d5779d20dc023081522d06364d5727d9263f038b340d09da06f564ad2e1","text":"Symlinked directories not collected since pytest 6.1.0 When there is a symlink to a directory in a test directory, is is just skipped over, but it should be followed and collected as usual. This regressed in (included in 6.1.0). For some reason I a in there, I don't remember why, but it does not match the previous behavior and should be removed. PR for this is coming up."}
{"_id":"q-en-pytest-dev__pytest-7985-f11003ed5da8754b3142f0a508124b8fc5ebccc3c16afd6b908ede19bcf3da9b","text":"Deprecate I don't see the point in removing it in a release just to reintroduce it again, that just makes things more confusing for everyone. Originally posted by in"}
{"_id":"q-en-pytest-dev__pytest-8022-79a1f2d4123235cce0404fdf6bfa53b9eb2d35d521049173f5cad9aae5ebc654","text":"Doctest collection only returns single test for <!-- Thanks for submitting an issue! Quick check-list while reporting bugs: --will only collect a single doctest because of this: Introduced a while back by here: See failing tests: Failing doctest collection When the module is an the doctest collection only picks up 1 doctest."}
{"_id":"q-en-pytest-dev__pytest-8033-b01207d9220383a315533a4de0cb0f9b03ec186aa4814405aa849dbc9558c085","text":"Class cleanups in Python 3.8+ are not called Did not see mentioned anywhere in this repo, which is the method that should be called. See Repo: With unittest: With Pytest:"}
{"_id":"q-en-pytest-dev__pytest-8055-7f8c1ec9504e69edd99fce95ea226dac3c33101a1b879d1e49e6b1cef84b5bdb","text":"set sys.unraisablehook (py38) Python 3.8 has a new hook: sys.unraisablehook Pytest should set this to be able to associate unraisable exceptions with tests"}
{"_id":"q-en-pytest-dev__pytest-8124-8fe8f2d916fcd8ea145aadc7d41868e245ca5bc310d05054414926ae01c2ebde","text":"Allow contibuting additional global variables for skipif\/xfail [ ] Include documentation when adding new features. [x] Include new tests or update existing tests when applicable. [X] Allow maintainers to push and squash when merging my commits. Please uncheck this if you prefer to squash the commits yourself. for details."}
{"_id":"q-en-pytest-dev__pytest-8250-830ba741a93e0abc192c833802910f38b9840ca698ff9df3a1a7c46c3491d868","text":"Problem with faulthandler when used with Twisted Logger and \"pytest --capture=no\" Dear developers, thanks a stack for conceiving and maintaining this excellent package. I never expected to file an issue or submit a patch here, but here we go. On behalf of , we are currently in the progress of finally upgrading to Python 3. Kotori is based on Twisted and uses for testing. Within , we tried hard to apply some magic to make everything work together on all occasions with respect to appropriately configuring to our needs. We are on macOS 10.13.6. The background on this is that the Twisted Logger intercepts the logging by providing a file-like wrapper around the stream, which is obviously not an actual stream. In this case, when running with , 's fails when trying to restore the stream through at [1] as this will actually return , in turn signaling an invalid file descriptor [2]. It will a) raise the exception outlined below and b) won't stop the process on teardown. [1] [2] This problem is mitigated by . With kind regards, Andreas."}
{"_id":"q-en-pytest-dev__pytest-8365-2be095ba39a8326ee98ec461c227b01eeb14eed802c665c233b51b5239f48c2b","text":"tmpdir creation fails when the username contains illegal characters for directory names , and rely on for determining the directory. I found that the user name returned by may return characters that are not allowed for directory names. This may lead to errors while creating the temporary directory. The situation in which I reproduced this issue was while being logged in through an ssh connection into my Windows 10 x64 Enterprise version (1909) using an OpenSSHforWindows7.7p1 server. In this configuration the command returns my domain username e.g. instead of as when logged in regularly using a local session. When trying to create a temp directory in pytest through e.g. this fails with the following error message: I could also reproduce this without the complicated ssh\/windows setup with pytest 6.2.2 using the following commands from a : Thanks for having a look at this!"}
{"_id":"q-en-pytest-dev__pytest-8399-88d10a4a64722a53576670a4c7f09f53a2c937eb1958f1eb6bab9694f665776d","text":"Starting v6.2.0, unittest setUpClass fixtures are no longer \"private\" <!-- Thanks for submitting an issue! Quick check-list while reporting bugs: --Minimal example: The expected (and previously implemented behavior) is that this fixture's name would start with an underscore, and would therefore only get printed if the additional flag was used. As it stands, I don't see a way to hide such generated fixtures which will not have a docstring. This breaks a code-quality CI script that makes sure we don't have undocumented pytest fixtures (and the code-base has many legacy tests that use unittest, and that will not get upgraded)."}
{"_id":"q-en-pytest-dev__pytest-8422-5766b092c29ecc8583802bf9ec069951485ed7d900f717111feba436c3aff5fd","text":"approx: Better handle Decimal in sequences and mappings <!-- Thanks for suggesting a feature! Quick check-list while suggesting features: --Right now handles Decimal comparisons gracefully, thanks to We can do this: also knows how to handle sequences and mappings: doesn't handle Decimal within sequences and mappings: Both of these tests fail with <!-- A clear and concise description of what you want to happen. --I would like these kind of tests to be passing :) A linked PR should be following shortly :eyes: <!-- Provide examples of real-world use cases that this would enable and how it solves the problem described above. --<!-- Have you tried to workaround the problem using a pytest plugin or other tools? Or a different approach to solving this issue? Please elaborate here. --<!-- Add any other context, links, etc. about the feature here. --"}
{"_id":"q-en-pytest-dev__pytest-8428-b4c556dd50b6edc23684eef5b124f9cf46ccc5e58b066d636fc9aa780d33c3d6","text":"Generate an error when a mark is applied to a fixture Follow up from . We should generate an error if a is applied to a fixture. There is a warning in about this problem which should be updated once this issue is dealt with."}
{"_id":"q-en-pytest-dev__pytest-8447-a3971c705e200bd5913f1278e457aaf0180d701c859ee5cf66c60802b89cfa2b","text":"Unexpected keyword argument 'path' from plugins While troubleshooting , I stumbled onto a new error, a that occurs when using pytest-black against the current main HEAD (), easily reproducible with an empty test file and pip-run: Same problem happens with pytest-checkdocs:"}
{"_id":"q-en-pytest-dev__pytest-8516-b147e1e36e0a0a21f83ff7317ad5de4ce90b1951875acf8c169f9e6dce450d7d","text":"Minor temporary directory security issue in pytest versions before 6.2.3 A minor temporary directory security issue was found in pytest versions before 6.2.3. This issue is fixed in pytest 6.2.3. pytest used to create directories under with world-readable permissions. This means that any user in the system was able to read information written by tests in temporary directories (such as those created by the fixture). Now the directories are created with private permissions. pytest used to silenty use a pre-existing ` directory, even if owned by another user. This means another user could pre-create such a directory and gain control of another user's temporary directory. Now such a condition results in an error."}
{"_id":"q-en-pytest-dev__pytest-8641-fba3e38881b6e3ebd5db96091a3147d00a94af3578e805f852aefa48f805c1ae","text":"No color output when specifying log format string with precision-formatted levelname <!-- Thanks for submitting an issue! Quick check-list while reporting bugs: --Pytest fails to output colorfully with the following log format string in This is due to  fails to match the format string due to the presence of precision bit."}
{"_id":"q-en-pytest-dev__pytest-8861-4d8893aeb718870822bc3513fdf78492be392671244911af84a30f0278901dac","text":"Internal error when adding a skip mark to a doctest inside a contextmanager To reproduce: This is a simplified issue from a real use case in the scikit-learn repo. We sometimes want to skip doctests, for example when matplotlib (an optional dependency) is not installed. If there is be a better way to do it with pytest than using , let me know. <details<summaryconda list output<\/summary<\/details"}
{"_id":"q-en-pytest-dev__pytest-8906-2b643b43918bb48271d6f7ea9adfb61a290ffb01e65fc234b37191ef7264286c","text":"Improve handling of skip for module level This is potentially about updating docs, updating error messages or introducing a new API. Consider the following scenario: is using Python 3,8 syntax: It should not be tested under Python 3.6 and 3.7. This is a proper way to skip the test in Python older than 3.8: My actual test involves parameterize and a 3.8 only class, so skipping the test itself is not sufficient because the 3.8 class was used in the parameterization. A naive user will try to initially skip the module like: This issues this error: Using outside of a test is not allowed. To decorate a test function, use the or decorators instead, and to skip a module use `pytestmark = {skip,skipif}. The proposed solution , does not work in my case: pytest continues to process the file and fail when it hits the 3.8 syntax (when running with an older version of Python). The correct solution, to use skip as a function is actively discouraged by the error message. This area feels a bit unpolished. A few ideas to improve: skip with in the error message. this seems in conflict with the spirit of the message. an alternative API to skip a module to make things easier: , which can call ."}
{"_id":"q-en-pytest-dev__pytest-8952-da6d9a8735d5a2aeb493830cb69aa0f6a99b0a34eaaae5324833b5c9f94b8d2a","text":"Enhance warning assertion capabilities while writing some other bits and pieces, I had a use case for checking the omitted, has a that doesn't quite offer yet the information is already available in there, I suspect there is a good reason why we don't have so I propose some additional capabilities on to handle warnings in isolation. With the full dict comparison may get a bit intrusive as far as warning capture is concerned. something simple like: Thoughts?"}
{"_id":"q-en-pytest-dev__pytest-8987-ef583578602882abed7ca457a2bba72803bd7dfc67171a0a18f0d18bfc8d92a4","text":"pytest -k doesn't work with \"\\\"? <div type='discussions-op-text'<supOriginally posted by nguydavi August 7, 2021<\/supHey! I've been trying to use passing the name I got by parametrizing my test. For example, Note the error message I tried escaping the but that didn't work, the only way I can make it work is to remove the backslashes completely, Is just not supported by ? Or am I missing something ? Thanks! EDIT: A possible test case Which is then referred by as . Essentially a new line character in the string brings the issue (or any escaped character probably)<\/div"}
{"_id":"q-en-pytest-dev__pytest-9064-af0aa5d99fbca58ae1ac31dcbc63149294bc9540b0933853f55e63e6c8489de0","text":"has no effect without [x] a detailed description of the bug or problem you are having The flag will \"exit on test failure and continue from last failing test\". The flag is supposed to \"ignore the first failing test but stop on the next failing test\". However, it does nothing unless used in conjunction with . The combo requirement is not mentioned in the help text, seems redundant, and is surprising behavior. I recommend that should act the same as . [x] output of from the virtual environment you are using [x] pytest and operating system versions MacOS, Python 3.9.5, pytest-6.2.4 [x] minimal example if possible Example three failures: has no effect: works as expected, stopping after first failure: only works with to stop after second failure: I believe the above behavior, the combo of the flags, should work like that even if only is used."}
{"_id":"q-en-pytest-dev__pytest-9066-9c003bd7d17641bad303219b11321f7382324568e4467f1cfcd77414bfd17f6c","text":"Show full qualified name on direct Node construction warning In , running pytest with many plugins installed gives this error: And is not clear which plugin is the culprit, I had to look at the source code of to figure it out. We might consider at least show the full qualified name of the offending class in that message, so users would see , which is a nudge in the right direction. Originally posted by in"}
{"_id":"q-en-pytest-dev__pytest-9133-9e9a8be8a7efb76a761741c7eeb32ed09052e00df9e979b04b9d4d4ddfa9f970","text":"Add a parameter to <!-- Thanks for suggesting a feature! Quick check-list while suggesting features: --<!-- What are you trying to do, that you are unable to achieve with pytest as it currently stands? --I'd like to be able to use to check deselected count. <!-- A clear and concise description of what you want to happen. --Add a parameter to <!-- Provide examples of real-world use cases that this would enable and how it solves the problem described above. --Plugins that use to change the and add change the deselected items need to be tested. Using to check the deselected count would be helpful. <!-- Have you tried to workaround the problem using a pytest plugin or other tools? Or a different approach to solving this issue? Please elaborate here. --Use instead of . returns a dictionary that includes , if there are any. However, if we have a series of tests, some that care about deselected, and some that don't, then we may have some tests using and some using , which is slightly annoying. <!-- Add any other context, links, etc. about the feature here. --"}
{"_id":"q-en-pytest-dev__pytest-9249-fc72136be4c2d7ef7d1ea0a08e00b67f264f7b35fdb11b03dfa87b9a0f8de418","text":"test ids with s cannot be selected with By default pytest 6.2.2 parametrize does user arguments to generate IDs, but some of these ids cannot be used with option because you endup with errors like when trying to do so. The solution for this bug is to assure that auto-generated IDs are sanitized so they can be used with -k option. Example: As you can see the first parameter includes a slash, and for good reasons. It is far from practical to have to add custom \"ids\" for all of these, as you can have LOTS of them. There is another annoyance related to the -k selecting for parameterized tests, is the fact that square braces have special meanings for some shells and in order to use it you must remember to quote the strings. It would be much easier if the display and selecting of parametrized tests would use only shell-safe format, so we can easily copy\/paste a failed test in run it. For example I think that using colon would be safe and arguably even easier to read: ."}
{"_id":"q-en-pytest-dev__pytest-9279-a3971c705e200bd5913f1278e457aaf0180d701c859ee5cf66c60802b89cfa2b","text":"Unexpected keyword argument 'path' from plugins While troubleshooting , I stumbled onto a new error, a that occurs when using pytest-black against the current main HEAD (), easily reproducible with an empty test file and pip-run: Same problem happens with pytest-checkdocs:"}
{"_id":"q-en-pytest-dev__pytest-9359-f8af5e29020d2ca64cfebcaa3a751091e332b19037381afe9bcd02609c138882","text":"Error message prints extra code line when using assert in python3.9 <!-- Thanks for submitting an issue! Quick check-list while reporting bugs: --- [x] a detailed description of the bug or problem you are having [x] output of from the virtual environment you are using [x] pytest and operating system versions [ ] minimal example if possible I have a test like this: The test \"testrightstatement\" fails at the first assertion,but print extra code (the \"t\" decorator) in error details, like this: And the same thing did not happen when using python3.7.10 Is there some problems when calculate the statement lineno? pytest 6.2.5 Windows 10 Seems to happen in python 3.9,not 3.7"}
{"_id":"q-en-pytest-dev__pytest-9475-f2548344cf21504f5046a0b56cc9c0c78fe1c22bc974eaeba9517c92b1831b20","text":"[prerelease] addition to is backwards-incompatible a new parameter to , cc However, this actually is an incompatible change: Doing e.g. followed by worked fine before, but now fails because the now included does not equal . This breaks pytest-bdd: - I could swear I also saw another project in fail after fixing the initial issue it had, but then Christmas and stuff came along and now I don't remember which one it was, and of course can't find it anymore. A (quite) reveals that more projects might be affected by this (excludes to avoid matches in copies of pytest's source code). Some examples I could dig up (but haven't verified): (maybe) I think the change in itself makes sense, but at the same time fixes like are a bit cumbersome. Two questions: What should we do about this for 7.0? (even if the answer just is \"live with it and document it as backwards-incompatible in the changelog) What (if anything) should we do about this so that it doesn't happen again for future releases? I guess not much we can do, as long as we want to assume 0 for outcomes which have not been given..."}
{"_id":"q-en-pytest-dev__pytest-9624-7f8f29b6f7d0846804d8927332bc33c284b48c82ba8fdacde959bfbd37554850","text":"INTERNALERRORTypeError: init() missing 1 required positional argument: 'testmethod' [ ] a detailed description of the bug or problem you are having [ ] output of from the virtual environment you are using [ ] pytest and operating system versions pytest-7.0.0 CentOS Stream 8 [ ] minimal example if possible Issue happened only once, as far as I know. I will update this issue if it happens multiple times and if I manage to reproduce it outside of GitHub Actions CI."}
{"_id":"q-en-pytest-dev__pytest-9646-7fc19aae0c791cc4103516bfdb001da0c941152690453817cd999f4928f84349","text":"Pytest 7 not ignoring warnings as instructed on <!-- Thanks for submitting an issue! Quick check-list while reporting bugs: --- [x] a detailed description of the bug or problem you are having [x] output of from the virtual environment you are using [x] pytest and operating system versions [x] minimal example if possible Hello, with the latest version of Pytest a series of new issues has started to pop-up to warn users (and plugin maintainers) about future changes on how Pytest will work. This is a great work done by Pytest maintainers, and pushes the community towards a better future. However, after informing plugin maintainers about the upcoming changes, I would like to silence these warnings so I can focus in fixing the errors detected on the tests (otherwise the output is too big and difficult to scroll through to find useful information). To solve this, I naturally tried to modify my file with a filter for these warnings, however Pytest is still printing some of them even after I the warning filters. The following script can be used to reproduce the problem: Relevant output: (The warnings seem to be shown only once per worker per plugin, but considering you have 32 workers with and 4 plugins that haven't adapted to the changes yet, the output can be very verbose and difficult to navigate) Pytest should not print the warnings: -"}
{"_id":"q-en-pytest-dev__pytest-9681-0068dfaf84fb6a2c22a26cb47a792d78fe553184544801eb783fb13d8469765d","text":"0 regression: Existence of messes up package discovery with importlib + pythonpath mode (ModuleNotFoundError: No module named 'tests.<test-packages') Consider the following demo project: (existence of this file breaks package discovery) pytest (version 7.0.0) errors with: What's particularly surprising: The example works, when renaming the to e.g. . I.e., it seems that the existence of the implies that it is no longer possible to import from . Imports like this used to work with pre 7 pytest versions in mode despite the existence of a ."}
{"_id":"q-en-pytest-dev__pytest-9709-93721099b8ca21c242557a973c9abe56689ec16cf9ba1ecd8f8784abb6d99337","text":"Current implementation of cannot be used to compare sets The current implementation of yields incorrect results when used to compare sets. Consider the following self-explanatory code comparing two (equal as by ) sets: Although the two sets are obviously the same, the last equality check using approx is failing. A quick view into the implementation of makes it obvious why this is the case: In , only is used (in ), but since sets are unordered, so is the resulting iterator. This means, for sets such an implementation cannot work. What makes things worse is the confusion that seems to exist here between the different abstract base classes: In the method, clearly is mentioned, explicitly. However, a set is not a sequence type, but only a collection type (because of the missing order). It is, however, iterable and since this is the only thing that is actually checked in the implementation, the code seems to work for sets, where, in fact, it does not. As a first step, I would suggest one could keep the current implementation, but explicitly check for sequence types (i.e. classes having a method) and delete all mentions of in the code as well as on the documentation page and make it crystal clear that there is only an implementation for sequence types. But what would be way better would, of course, be an implementation for arbitrary container comparisons. Tested with pytest version 7.0.1."}
{"_id":"q-en-pytest-dev__pytest-9780-0d074597c66f8c183f30cd79ef4bf31ffce1bd85c89e822018a43cdf75433f36","text":"Version 7.1.0 seems like not picking up Over SciPy we have started to see the message in the title in our Azure Pipelines which started picking up 7.1.0 instead of 7.0.1 lastn ight. From the deprecation and removal list we couldn't see which item might have caused it. One speculation we might generate is that our mark is not registered in our file explicitly and maybe registrations are now obligatory. Here is a complete SciPy test suite example with all installation steps provided. And our file lives here Thank you for all the efforts."}
{"_id":"q-en-pytest-dev__pytest-9798-e81ce73d90e31b049c9da4f0c6adf6d0e6120cc36f3694cb47268af43388d644","text":"ModuleNotFoundError for Numpy when fails <!-- Thanks for submitting an issue! Quick check-list while reporting bugs: --When using to compare lists in a test, a is raised for Numpy whenever the test fails. If the test succeeds, there is no such error. It appears that does not require Numpy to run, but the error message requires Numpy to display the details. Since does not require Numpy to function correctly, it should not require Numpy to display the error. macOS Big Sur Version 11.6.2 No which makes the whole error message confusing and leads you to believe it failed because Numpy is not installed instead of the fact it was an assertion error."}
{"_id":"q-en-pytest-dev__pytest-9911-e991626adebeb2774ca650a0eec76745fc628ead133daf8859b64eaef32ec299","text":"Improved error when (empty tuple) is passed to or When using as a context manager, you can pass an exception type or tuple of exception types; it's then an error if no exception is raised or if the exception raised is not an instance of the expected type(s). The same logic applies to , which has a near-identical API. The problem is that if you pass the empty tuple , this will always result in an error: even if an exception is raised, it can't be an instance of ! I think we should explicitly check tuple inputs, and raise a more helpful error message if they're empty. For example: \"Passing is an error, because it's impossible to raise an exception which is not an instance of any type. Raising exceptions is already understood as failing the test, so you don't need any special code to say 'this should never raise an exception'.\" (for bonus points, should have the same message, with instead of ) The same logic, and same error message, applies to the argument to . \"Passing is an error, because it's impossible to emit a warning which is not an instance of any type. To assert that no warnings are emitted, use <whatever we come up with for"}
{"_id":"q-en-pytest-dev__pytest-9956-e4773c65903206ca10ddd14d206559dd49ead93c1878449ac556c3653f50bbc6","text":"Test functions that return non-None should raise a warning\/error Consider this test: Did you spot the error? The second parametrization has a typo, is 9, not 8. But this test actually doesn't test anything, because it returns the assertion rather than asserting it. This is a common enough mistake, and it wouldn't normally be a problem except that it can silently cause false positives in test suites. I propose that test functions that return anything except None fail with a message that cues users that they probably meant to assert rather than return. This feature could be disabled (or enabled, if there are backwards-compatibility issues) via a config flag if necessary."}
{"_id":"q-en-scikit-learn__scikit-learn-10198-024b3b6f41472c6f7dace629f8f60753fabacbfd649405e7527dfa2bed01ff35","text":"add getfeaturenames to CategoricalEncoder We should add a ` to the new CategoricalEncoder, as discussed . I think it would be good to be consistent with the PolynomialFeature which allows passing in original feature names to map them to new feature names. Also see ."}
{"_id":"q-en-scikit-learn__scikit-learn-10297-20be491fc69e64c42e8f6c3d9a49e7663b9eceef391ee89460ef5fbe5d03844e","text":"linearmodel.RidgeClassifierCV's Parameter storecvvalues issue Parameter storecvvalues error on sklearn.linearmodel.RidgeClassifierCV import numpy as np from sklearn import linearmodel as lm #test database n = 100 x = (n, 30) y = (size = n) rr = lm.RidgeClassifierCV(alphas = (0.1, 1000, 0.1), normalize = True, storecvvalues = True).fit(x, y) Expected to get the usual ridge regression model output, keeping the cross validation predictions as attribute. TypeError: init() got an unexpected keyword argument 'storecvvalues' lm.RidgeClassifierCV actually has no parameter storecvvalues, even though some attributes depends on it. Windows-10-10.0.-SP0 Python 3.6.3 (default, Oct 15 2017, 03:27:45) [MSC v.1900 64 bit (AMD64)] NumPy 1.13.3 SciPy 0.19.1 Scikit-Learn 0.19.1 Add storecvvalues boolean flag support to RidgeClassifierCV Add storecvvalues support to RidgeClassifierCV - documentation claims that usage of this flag is possible: While actually usage of this flag gives"}
{"_id":"q-en-scikit-learn__scikit-learn-10377-4caa643409acfa797580304fbf638dc5bacbe5448fa4d78d953a9d3c685bb652","text":"BUG Inconsistent f1score behavior when combining label indicator input with labels attribute When using label indicator inputs for ypred and ytrue, metrics.f1score calculates the macro average over all label-specific f-scores whenever the labels parameter includes column index 0. It should only average over the label-specific scores indicated by the labels parameter, as it does when 0 is not present in the labels parameter. <!-- Instructions For Filing a Bug: --<!-- Please paste or specifically describe the actual output or traceback. --Windows-7-6.1.7601-SP1 Python 3.5.3 (default, May 15 2017, 10:43:23) [MSC v.1900 64 bit (AMD64)] NumPy 1.13.1 SciPy 0.19.0 Scikit-Learn 0.19.0 <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-10382-f9bfb32255652b27fd7ff1c8d900012b199a34d35f7a2164f29902371716204c","text":"TfidfVectorizer dtype argument ignored TfidfVectorizer's fit\/fittransform output is always np.float64 instead of the specified dtype FitFailedWarning raised by cross validation could do with better message Current message says \"Classifier fit failed\" but the estimator is not necessarily a classifier. (Perhaps that's too pedantic of me...) is not technically the best way to display an error message. We could either use and include the whole traceback; or we can use to print it properly (though I think this is the same as ). Another thing we can consider, now that provides structured results to and , is to store the full traceback in some array of ."}
{"_id":"q-en-scikit-learn__scikit-learn-10397-afe787c17253c35842a889cf15e56b333b89d8f26f51037c9f851b1c5188b7a7","text":"integers in RidgeCV alpha making one of the alphas a float fixes the problem. This should be handled internally. Python3.6 integers in RidgeCV alpha making one of the alphas a float fixes the problem. This should be handled internally. Python3.6"}
{"_id":"q-en-scikit-learn__scikit-learn-10427-c55516c573fc7a2c44484e3543186fb87a97e68ebff16880f22554d31b4cc7a9","text":"loadsampleimages uses deprecated imread DeprecationWarning: is deprecated! is deprecated in SciPy 1.0.0, and will be removed in 1.2.0."}
{"_id":"q-en-scikit-learn__scikit-learn-10428-140c7e99f95dde8d91df144e2ec16c75925c0ffcb7c1f901750e23486b0948f3","text":"Add common test to ensure all(predict(X[mask]) == predict(X)[mask]) I don't think we currently test that estimator predictions\/transformations are invariant whether performed in batch or on subsets of a dataset. For some fitted estimator , data and any boolean mask of length , we need: where is any of {, , , , }. Testing that predictions for individual samples match the predictions across the dataset might be sufficient. This should be to common tests at Indeed, reports that this is broken for one-vs-one classification. :'("}
{"_id":"q-en-scikit-learn__scikit-learn-10452-c5258ae37ff48c86a385f9ce9e048558a8d507301283dd030e2316f08196c9f7","text":"Polynomial Features for sparse data I'm not sure if that came up before but PolynomialFeatures doesn't support sparse data, which is not great. Should be easy but I haven't checked ;)"}
{"_id":"q-en-scikit-learn__scikit-learn-10459-e15d5bd1474205d24c232d206a183f6bdad431012de7000c554d95793113e97b","text":"[RFC] Dissociate NaN and Inf when considering forceallfinite in checkarray Due to changes proposed in , it seems that as currently a main limitation. will force both and to be rejected. If preprocessing methods (whenever this is possible) should let pass , this argument is not enough permissive. Before to implement anything, I think it could be good to have some feedback on the way to go. I see the following solutions: could still accept a bool to preserve the behaviour. Additionally, it could accept an to filter only . proposes to have an additional argument . was worried that it makes to complex. a private function (similarly to removing the numpy version checking) in the which can be shared between the preprocessing methods. They are the solutions that I have in mind for the moment but anything else is welcomed. I would be grateful for any insight."}
{"_id":"q-en-scikit-learn__scikit-learn-10483-e05cb344604de10b7b0d29d0389b61f5a5e5c7108863b38ea5c8449873eedb7d","text":"Move imputation out of preprocessing While we're considering additional imputers, I've wondered whether preprocessing is the right place for it. Yes, it is a preprocessing step before other learning, but it often makes use of other supervised and unsupervised learners and hence is a learning task of its own. And preprocessing is getting a bit cramped. We could also do as with other models and have imputers appear in modules on the basis of how they work rather than function: could appear in neighbors for instance. could appear where..? And the basic in dummy? probably not. In practice I think it is more useful for users to , akin to our clusterers and decomposition, and unlike our predictors and outlier detectors that are grouped by algorithm."}
{"_id":"q-en-scikit-learn__scikit-learn-10495-ec39267b8bfe6217f6a91f752b02666bd62030d04d814dab24e9ce4eecf55956","text":"checkarray(X, dtype='numeric') should fail if X has strings Currently, dtype='numeric' is defined as \"dtype is preserved unless is object\". This seems overly lenient and strange behaviour, as in where shows that works without error and produces an array of strings! This behaviour is not tested and it's hard to believe that it is useful and intended. Perhaps we need a deprecation cycle, but I think dtype='numeric' should raise an error, or attempt to coerce, if the data does not actually have a numeric, real-valued dtype. checkarray(X, dtype='numeric') should fail if X has strings Currently, dtype='numeric' is defined as \"dtype is preserved unless is object\". This seems overly lenient and strange behaviour, as in where shows that works without error and produces an array of strings! This behaviour is not tested and it's hard to believe that it is useful and intended. Perhaps we need a deprecation cycle, but I think dtype='numeric' should raise an error, or attempt to coerce, if the data does not actually have a numeric, real-valued dtype."}
{"_id":"q-en-scikit-learn__scikit-learn-10508-8d2c338664e04ec741d225c10b2b9b77a9387020e92b41285d385dfcb06cca25","text":"LabelEncoder transform fails for empty lists (for certain inputs) Python 3.6.3, scikit_learn 0.19.1 Depending on which datatypes were used to fit the LabelEncoder, transforming empty lists works or not. Expected behavior would be that empty arrays are returned in both cases."}
{"_id":"q-en-scikit-learn__scikit-learn-10558-2be9341038c1fde7cd08c3d6cf4387b4dde7fe73794cd1b007a6312c25614f09","text":"Deprecate Imputer with axis=1 After having tried to deal with a few issues related to extending behaviour, I believe we should be removing the parameter from . It seems a strange feature to support in a machine learning context, except perhaps where the features represent something like a time series. It is not stateful and can be performed with a FunctionTransformer. (We could even provide a function, if we felt it necessary, which would roughly be defined as .) It complicates the implementation, which already has a bunch of weird edge-cases (to handle sparse data with missing indicated by 0 which is an inefficient use of a sparse data structure; and to handle non-NaN missingness indicators), unnecessarily. It is often nonsensical to extend further features to the case. Do others agree? [MRG+1] Deprecate ` argument Deprecated the argument on the class."}
{"_id":"q-en-scikit-learn__scikit-learn-10577-49d12326e174cc5a42076056a38a722592dbce1befc17578c5bb15472dc3b91e","text":"Allow nsamples to be a tuple in makeblobs I'd like makeblobs to accept lists or tuples for nsamples to generate imbalanced classes. Could be used here for example: http:\/\/scikit-"}
{"_id":"q-en-scikit-learn__scikit-learn-10581-9a4ad38a0146e24f9cc4e6884543ce0e004963b99dfd458680736fbb190eb23a","text":"ElasticNet overwrites X even with copyX=True The function of an , called with , overwrites X, even when : ElasticNet overwrites X even with copyX=True The function of an , called with , overwrites X, even when : [MRG] FIX ElasticNet overwrites X even with copy_X=True Made changes as suggested by please review and suggest changes"}
{"_id":"q-en-scikit-learn__scikit-learn-10687-f2f72ddee6de97d41f20545650df9507341306195816528108f0402d6206aadb","text":"Shape of wrong for linearmodel.Lasso when using <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --Shape of wrong for linearmodel.Lasso when using Example: the second snippet should not raise, but it does. The first snippet is ok. I pasted it as a reference Linux-3.2.0-4-amd64-x8664-with-debian-7.11 ('Python', '2.7.3 (default, Mar 13 2014, 11:03:55) \\n[GCC 4.7.2]') ('NumPy', '1.13.3') ('SciPy', '0.19.1') ('Scikit-Learn', '0.18.2') <!-- Thanks for contributing! --[MRG] Shape of wrong for linearmodel.Lasso when using <!-- Thanks for contributing a pull request! Please ensure you have taken a look at the contribution guidelines: --"}
{"_id":"q-en-scikit-learn__scikit-learn-10777-c7879b785de43fef92691e7512670e4ed5ca49dae0ed612b6e4dcadfa11733d2","text":"no error on CountVectorizer(ngramrange=(2, 1)) I think if ngramrange[0] is greater than ngram_range[1] we should throw an error. Not sure what the current behavior is."}
{"_id":"q-en-scikit-learn__scikit-learn-10803-c241c90c6ece3bf4d646ab6cb8a0bb1d1a882c5c739cc92d823d13545f816d70","text":"weighted KDE Not sure this is the correct place, but I would very much appreciate the ability to pass a weight for each sample in kde density estimation. There exits a adapted version of : weighted KDE Not sure this is the correct place, but I would very much appreciate the ability to pass a weight for each sample in kde density estimation. There exits a adapted version of :"}
{"_id":"q-en-scikit-learn__scikit-learn-10844-525dddfc556a98091980fcfd1546eb9edc00c692d968544c45e55be13379388b","text":"fowlkesmallowsscore returns RuntimeWarning when variables get too big <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --<!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--sklearn\\metrics\\cluster\\ return tk \/ (pk qk) if tk != 0. else 0. This line produces RuntimeWarning: overflow encountered in intscalars when (pk qk) is bigger than 2*32, thus bypassing the int32 limit. Any code when pk and qk gets too big. <!-- Example: If the code is too long, feel free to put it in a public gist and link it in the issue: --<!-- Example: No error is thrown. Please paste or describe the expected results.--Be able to calculate tk \/ (pk qk) and return a float. <!-- Please paste or specifically describe the actual output or traceback. --it returns 'nan' instead. I propose to use (tk \/ pk) * (tk \/ qk) instead, which gives same result and ensuring not bypassing int32 <!-- Please run the following snippet and paste the output below. import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"NumPy\", numpy.version) import scipy; print(\"SciPy\", scipy.version) import sklearn; print(\"Scikit-Learn\", sklearn.version) --18.1 <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-10870-0aca51903d562372e02e335a5a81e82ae64c8dd3f2430d3fbf266fc12a5916ed","text":"In Gaussian mixtures, when ninit 1, the lowerbound is not always the max In Gaussian mixtures, when is set to any value greater than 1, the is not the max lower bound across all initializations, but just the lower bound of the last initialization. The bug can be fixed by adding the following line just before in : The test that should have caught this bug is in , but it just does a single test, so it had a 50% chance of missing the issue. It should be updated to try many random states. No error. In Gaussian mixtures, when ninit 1, the lowerbound is not always the max In Gaussian mixtures, when is set to any value greater than 1, the is not the max lower bound across all initializations, but just the lower bound of the last initialization. The bug can be fixed by adding the following line just before in : The test that should have caught this bug is in , but it just does a single test, so it had a 50% chance of missing the issue. It should be updated to try many random states. No error."}
{"_id":"q-en-scikit-learn__scikit-learn-10881-dbbfe717a34400a4fae396365652d2992fa974aa4d358ac4e12d33ace4faae55","text":"No warning when LogisticRegression does not converge <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --<!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--I've run LogisticRegressionCV on the Wisconsin Breast Cancer data, and the output of was 100 for all but 1 of the variables. The default of 100 iterations was probably not sufficient in this case. Should there not be some kind of warning? I have done some tests and ~3000 iterations was probably a better choice for maxiter... <!-- Example: If the code is too long, feel free to put it in a public gist and link it in the issue: --<!-- Example: No error is thrown. Please paste or describe the expected results.--Some kind of error to be shown. E.g: \"result did not converge, try increasing the maximum number of iterations (maxiter)\" <!-- Please run the following snippet and paste the output below. import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"NumPy\", numpy.version) import scipy; print(\"SciPy\", scipy.version) import sklearn; print(\"Scikit-Learn\", sklearn.version) --import platform; print(platform.platform()) Darwin-16.7.0-x8664-i386-64bit import sys; print(\"Python\", sys.version) ('Python', '2.7.14 (default, Oct 5 2017, 02:28:52) \\n[GCC 4.2.1 Compatible Clang 4.0.1 (tags\/RELEASE_401\/final)]') import numpy; print(\"NumPy\", numpy.version) ('NumPy', '1.13.3') import scipy; print(\"SciPy\", scipy.version) ('SciPy', '1.0.0') import sklearn; print(\"Scikit-Learn\", sklearn.version) ('Scikit-Learn', '0.19.1') <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-10899-6ee6c7545bd211b1ba36c1368b6d71d5e1189cc9cfdaf2ecebeb073af39aba79","text":"Setting idf is impossible <!-- Instructions For Filing a Bug: --Rather than a bug i guess that would go as a sort of \"enhancement proposition\" ? I'm currently trying to persist a by basically saving its parameters in a mongoDB database and then rebuilding it alike. This technique works for but simply blocks for as there is no way to set . Is there any actual architectural reason why setting this attributes raise an error ? if yes, do you have an idea for a workaround ? I obviously want to avoid keeping the matrix on which it has been fitted as it would completely mess up the architecture (i believe that the saving\/loading process should be separated from the whole treatment\/learning process and trying to keep both would mean having to propagate a dirty return) functioning example on CountVectorizer This might not seem very impressive, but dictionnaries can be stored inside of mongoDB databases, which means that you can basically restore the or at least an identical copy of it by simply storing and the output of . Now the incriminated piece of code I would expect that at least one would work. However, both return an error. In the first case, it seems logical, as there is no idf\/idf_ parameter In the second case, i suppose that encapsulation forbids the direct setting I think that being able to reproduce a fitted object (even if it is only for non-classifier objects) without having to recompute it at each launch would benefit a lot of applications. I'm currently developping a RestAPI that has to do heavy computations on data before feeding it to the vectorizer, having to relearn the whole model with each computing is very slow, and means i have to currently wait up to half an hour for modifications that are sometimes about 1 line of code. Windows-10-10.0. ('Python', '2.7.11 (default, Feb 16 2016, 09:58:36) [MSC v.1500 64 bit (AMD64)]') ('NumPy', '1.11.0') ('SciPy', '0.17.1') ('Scikit-Learn', '0.17.1') <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-10908-ed65098c93e4ad4407c5f18da4c03db30d81cae802e729147393d6cce73b0756","text":"CountVectorizer's getfeaturenames raise not NotFittedError when the vocabulary parameter is provided If you initialize a and try to perform a transformation without training you will get a exception. On the other hand if you provide the at the initialization of the vectorizer you could transform a corpus without a prior training, right? The 's calls method which sets the instance variable. In the same manner I believe that the method should not raise if the vocabulary parameter is provided but the vectorizer has not been trained."}
{"_id":"q-en-scikit-learn__scikit-learn-10913-d7b2f599418be45db88cf7bf6b954bf0fc141616916447dc70b405b22be772c0","text":"MultiLabelBinarizer breaks when seeing unseen there be an option to handle this instead? <!-- Instructions For Filing a Bug: --I am not sure if it's intended for MultiLabelBinarizer to fit and transform only seen data or not. However, there are many times that it is not possible\/not in our interest to know all of the classes that we're fitting at training time. For convenience, I am wondering if there should be another parameter that allows us to ignore the unseen classes by just setting them to 0? Example: Result: array([[1, 0, 0], [0, 1, 0], [0, 0, 0]]) (the current version 0.19.0 would say ) I can open a PR for this if this is a desired behavior. Others also have similar issue:"}
{"_id":"q-en-scikit-learn__scikit-learn-10949-126adc08adde0b340f00a419eb09993ff81de1b3c9dc9533a7d1e70fec0f614f","text":"warnondtype with DataFrame No warning is thrown Linux-4.4.0-116-generic-x8664-with-debian-stretch-sid Python 3.6.3 (default, Nov 3 2017, 19:19:16) [GCC 7.2.0] NumPy 1.13.1 SciPy 0.19.1 Scikit-Learn 0.20.dev0 Pandas 0.21.0 warnondtype with DataFrame No warning is thrown Linux-4.4.0-116-generic-x8664-with-debian-stretch-sid Python 3.6.3 (default, Nov 3 2017, 19:19:16) [GCC 7.2.0] NumPy 1.13.1 SciPy 0.19.1 Scikit-Learn 0.20.dev0 Pandas 0.21.0"}
{"_id":"q-en-scikit-learn__scikit-learn-10982-4b2211686d7373f792295c3565f8ce15d27d74ce1f2a2af33c7a4cb46a972187","text":"[RandomizedSearchCV] Do not enforce that niter is less than or equal to size of search space Instantiating with greater than the size of (i.e. the product of the length of each distribution\/array in the grid) will fail with an exception at . This is a bit annoying for me because I have an app where I'm letting the user specify the number of iterations to run from the command line, also I've been fiddling around with the param grid so keeps changing. I don't want to have to work out the exact grid size when it goes below, say, 50; if I specify that should be interpreted as an upper bound on the number of iterations. Would it be possible to add an option (off by default) to the constructor specifying whether to throw in such cases? e.g. By passing (the option would default to )"}
{"_id":"q-en-scikit-learn__scikit-learn-10986-04c0e2030ab780bfeb5de7da99049e9c695f155ff5a60c0292659feb971ed6cc","text":"Warm start bug when fitting a LogisticRegression model on binary outcomes with . <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --Bug when fitting a LogisticRegression model on binary outcomes with multiclass='multinomial' when using warm start. Note that it is similar to the issue here i.e. only using a 1D object on binary outcomes even when using as opposed to a 2D object. <!-- Example: If the code is too long, feel free to put it in a public gist and link it in the issue: --from sklearn.linearmodel import LogisticRegression import sklearn.metrics import numpy as np # Set up a logistic regression object lr = LogisticRegression(C=, multiclass='multinomial', solver='sag', tol=0.0001, warmstart=True, verbose=0) # Set independent variable values Z = ([ [ 0. , 0. ], [ 1., 0. ], [ 1., -0.], [-0., -0.], [ 1., 1.], [-0., -1.], [-1., -0.], [ 2. , -0.], [-0., -1.], [ 2., -0.], [-1., 2.], [-1., 2. ], [ 2., -1.], [ 0., -0.], [ 1., 0.], [ 1., -0.], [ 0., 1.], [-0., -0.], [-1., -0.], [-1. , 0.]]) # Set dependant variable values Y = ([1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 1, 0, 0, 1, 1], dtype=np.int32) # First fit model normally (Z, Y) p = lr.predictproba(Z) print(sklearn.metrics.logloss(Y, p)) # ... print(lr.intercept) print() # Now fit model after a warm start (Z, Y) p = lr.predictproba(Z) print(sklearn.metrics.logloss(Y, p)) # ... print(lr.intercept) print() The predictions should be the same as the model converged the first time it was run. The predictions are different. In fact the more times you re-run the fit the worse it gets. This is actually the only reason I was able to catch the bug. It is caused by the line here w0[:, [1]] = coef As is , but is , this causes the value to be broadcast into the . This some sort of singularity issue when training resulting in worse performance. Note that had it not done exactly this i.e. was simply initialised by some random values, this bug would be very hard to catch because of course each time the model would converge just not as fast as one would hope when warm starting. The fix I believe is very easy, just need to swap the previous line to if nclasses == 1: w0[0, [1]] = -coef # Be careful to get these the right way around w0[1, [1]] = coef else: w0[:, [1]] = coef <!-- Please run the following snippet and paste the output below. import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"NumPy\", numpy.version) import scipy; print(\"SciPy\", scipy.version) import sklearn; print(\"Scikit-Learn\", sklearn.version) --Linux-4.13.0-37-generic-x86_64-with-Ubuntu-16.04-xenial Python 3.5.2 (default, Nov 23 2017, 16:37:01) NumPy 1.14.2 SciPy 1.0.0 Scikit-Learn 0.20.dev0 (built from latest master) <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-11040-9068dde187d0e23eb1932e3567572565871922a0fc56f9054f71529eb29d5783","text":"Missing parameter validation in Neighbors estimator for float n_neighbors This should be caught earlier and a more helpful error message should be raised (or we could be lenient and cast to integer, but I think a better error might be better). We need to make sure that also works."}
{"_id":"q-en-scikit-learn__scikit-learn-11042-1ea45737f97cd73147c80d270212085f3bfc35da3ea298f3bbf9be1cc70666a7","text":"OneHotEncoder does not output scipy sparse matrix of given dtype OneHotEncoder ignores the specified dtype in the construction of the sparse array when mixed input data are passed, i.e with both categorical and real data type Platform: Linux-4.13.0-38-generic-x86_64-with-debian-stretch-sid Python: 3.6.3 (default, Oct 13 2017, 12:02:49) [GCC 7.2.0] NumPy: NumPy SciPy: SciPy 1.0.1 Scikit-Learn: Scikit-Learn 0.19.1"}
{"_id":"q-en-scikit-learn__scikit-learn-11043-f20110836c9c59989264aa669da3a8a84ae4084bf88fc287b898d1d14aa66b6c","text":"FunctionTransformer should not convert DataFrames to arrays by default I would expect a common use of FunctionTransformer is to apply some function to a Pandas DataFrame, ideally using its own methods or accessors. As noted in , it can be easy for users to miss that they need to set validate=False to pass through a DataFrame without converting it to a NumPy array. I think it would be more user-friendly to have by default, which would pass through DataFrames to the function, but otherwise convert its input to a 2d array. For strict backwards compatibility, the default should be changed through a deprecation cycle, warning whenever using the default validation means a DataFrame is currently converted to an array. Do others agree?"}
{"_id":"q-en-scikit-learn__scikit-learn-11151-fd56b30de8742fa05292b24407ff5bdddb13799824612e18787ab7386245fa10","text":"catch more warnings in common tests Right now the output of travis is too large to render, partially because a lot of deprecation warnings and partially because of warnings from the common tests. The common tests should catch all the deprecation warnings and probably also convergence warnings or numerical warnings (we decrease the number of iterations for faster testing)."}
{"_id":"q-en-scikit-learn__scikit-learn-11160-6283019ba3ab8de0836c997a58fba49fee428598733613f85ec70795ad8adce8","text":"output options? Is it possible to add output options to http:\/\/scikit- It would be really useful to have a output or output. Right now it outputs as a string that must be printed but it's difficult to use the results. I can make a quick helper script if that could be useful? [MRG] Classification report Dict-of-Dicts output <!-- Thanks for contributing a pull request! Please ensure you have taken a look at the contribution guidelines: --<!-- Example: -- This PR adds an option of returning the classification report in the form of a Dictionary of Dictionaries. Will add tests for the code, if the code is approved. <!-- Please be aware that we are a loose team of volunteers so patience is necessary; assistance handling other issues is very welcome. We value all user contributions, no matter how minor they are. If we are slow to review, either the pull request needs some benchmarking, tinkering, convincing, etc. or more likely the reviewers are simply busy. In either case, we ask for your understanding during the review process. For more information, see our FAQ on this topic: http:\/\/scikit- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-11235-27d9ac34c79d37f9733deb49ea3408bc1572a6b3d738f2046dbbc2240a7bf5f8","text":"Consistency issue in StandardScaler There is an issue of consistency with with and between the sparse and dense case. it make sense to support this case. It will return the identity matrix which is not the use case for the . If we wish a transformer to do so, one should use the I assume. we consider this behaviour normal, we need to: In the dense case, force to be after each iteration of . In the sparse case, compute the non-NaNs values and update which is not computed. It leads currently to an error if calling twice (i.e. will fail). IMO, we should make a checking at raise an error. WDYT?"}
{"_id":"q-en-scikit-learn__scikit-learn-11243-c230540e648c15cf0ee7e3989be667005ff62d4c547266d08ddf56d817eca226","text":"minmaxscale does not ignore NaNs The class ignore NaNs. Its counterpart function does not. The needs to add the option . implement the tests and this fix. However, it should be done in another proper PR."}
{"_id":"q-en-scikit-learn__scikit-learn-11264-3f1083d43da48276e0037daf651270e7f052338c7b2f77eea0a0a2fdad000557","text":"randomizedsvd is slow for dokmatrix and lilmatrix (and its object-oriented interface, ) is extremely slow for certain types of sparse matrix. Either all sparse matrices should be processed in roughly the same amount of time, or a warning should be printed. silently takes up to 50x longer than necessary. Windows-10-10.0.-SP0 Python 3.6.0 (default, Dec 23 2016, 11:57:41) [MSC v.1900 64 bit (AMD64)] NumPy 1.14.4 SciPy 1.1.0 Scikit-Learn 0.19.1 Also tested on: Linux-4.16.11-1-ARCH-x8664-with-arch-Arch-Linux Python 3.6.5 (default, May 11 2018, 04:00:52) [GCC 8.1.0] NumPy 1.14.5 SciPy 1.1.0 Scikit-Learn 0.19.1"}
{"_id":"q-en-scikit-learn__scikit-learn-11281-816369f924af347ddd59133c68c3a945c73eb0cd79fc3a29a861de75efa48312","text":"Should mixture models have a clusterer-compatible interface Mixture models are currently a bit different. They are basically clusterers, except they are probabilistic, and are applied to inductive problems unlike many clusterers. But they are unlike clusterers in API: they have an parameter, with identical purpose to they do not store the of the training data they do not have a method And they are almost entirely documented separately. Should we make the MMs more like clusterers?"}
{"_id":"q-en-scikit-learn__scikit-learn-11310-8394db30a4ce8eda604c3a58d4cadd12cab745d1ad7201bc3faa795aaa920fe0","text":"Retrieving time to refit the estimator in BaseSearchCV Basically, I'm trying to figure out how much time it takes to refit the best model on the full data after doing grid\/random search. What I can so far do is retrieve the time it takes to fit and score each model: In case I run this on a single core, I could time the whole search procedure and subtract the time it took to fit the single folds during hyperparameter optimization. Nevertheless, this isn't possible any more when setting . Thus, it would be great to have an attribute which is simply the time it took to refit the best model. Usecase: for we want to support uploading the results of hyperparameter optimization, including the time it takes to do the hyperparameter optimization."}
{"_id":"q-en-scikit-learn__scikit-learn-11333-a9bf20c6f6694cf5dc63801fd56a99dca5eb0ce4cfccaff4cfb3d290ece8a34c","text":"BaseCompostion.setparams broken where there are no estimators raises an error when the composition has no estimators. This is a marginal case, but it might be interesting to support alongside ."}
{"_id":"q-en-scikit-learn__scikit-learn-11346-4f58eeceaeace0a4068417f829e2c26756e73f1b7ef4f233cb4998510c0fa140","text":"ValueError: assignment destination is read-only, when paralleling with njobs 1 When I run with njobs 1, there is a chance to raise exception . The code is shown as follow: The bigger is, the higher chance get. When is small (lower than 2000, I verified), everything works fine. Once is bigger than 2000, there is a chance to get the exception. When is bigger than 5000, it is 100% raised. My version infor: OS: OS X 10.11.1 python: Python 2.7.10 |Anaconda 2.2.0 numpy: 1.10.1 sklearn: 0.17 The full error information is shown as follow"}
{"_id":"q-en-scikit-learn__scikit-learn-11391-6c8a468dbe2c2fc5239688cb8d4c538482dbbb66234bb0d316652eb40b9dd1e1","text":"Cryptic error in imputers due to missing checking in getmask By working on the it seems that there is a missing checking between and dtype: In short, is raised by equal in case of numeric and string mixed dtype. We should put a check in which is shared across classes."}
{"_id":"q-en-scikit-learn__scikit-learn-11496-29d92463555fc306ae5d6993b9c1ddc3d7543c9e3d778befb4c00e5206172909","text":"BUG: SimpleImputer gives wrong result on sparse matrix with explicit zeros The current implementation of the can't deal with zeros stored explicitly in sparse matrix. Even when stored explicitly, we'd expect that all zeros are treating equally, right ? See for example the code below: Whereas the expected result would be"}
{"_id":"q-en-scikit-learn__scikit-learn-11574-cf6b263acb72ff1c07eeea396d03f893c88dc533f07c3873f5c77e16e242f818","text":"IsolationForest contamination deprecation in init not in fit need to move the deprecation and fix the tests."}
{"_id":"q-en-scikit-learn__scikit-learn-11578-d86d8d4ee081ba57936e60f5d22b9d80dc31cdcbbf80552c5cade969b1ab5a6e","text":"For probabilistic scorers, LogisticRegressionCV(multiclass='multinomial') uses OvR to calculate scores Description: For scorers such as that use to get probability estimates out of a classifier, the predictions used to generate the scores for do not seem to be the same predictions as those generated by the method of . The former uses a single logistic function and normalises (one-v-rest approach), whereas the latter uses the softmax function (multinomial approach). This appears to be because the instance supplied to the scoring function at line 955 of within the helper function , () , is initialised, () , without a multiclass argument, and so takes the default, which is . It seems like altering L922 to read so that the instance supplied to the scoring function at line 955 inherits the option specified in would be a fix, but I am not a coder and would appreciate some expert insight! Likewise, I do not know whether this issue exists for other classifiers\/regressors, as I have only worked with Logistic Regression. Minimal example: Versions: Linux-4.4.0-72-generic-x8664-with-Ubuntu-14.04-trusty Python 2.7.6 NumPy 1.12.0 SciPy 0.18.1 Scikit-learn 0.18.1 [WIP] fixed bug in logregscoringpath <!-- Thanks for contributing a pull request! Please ensure you have taken a look at the contribution guidelines: --<!-- Example: -- In logregscoringpath method, constructor of LogisticRegression accepted only fitintercept as argument, which caused the bug explained in the issue above. As suggested, adding multiclass as argument when creating logistic regression object, solves the problem for multiclass case. After that, it seems like other similar parameters must be passed as arguments to logistic regression constructor. Also, changed interceptscaling default value to float Tested on the code provided in the issue by with various arguments on linearmodel.logistic.logregscoringpath and linearmodel.LogisticRegression, seems ok. Probably needs more testing. <!-- Please be aware that we are a loose team of volunteers so patience is necessary; assistance handling other issues is very welcome. We value all user contributions, no matter how minor they are. If we are slow to review, either the pull request needs some benchmarking, tinkering, convincing, etc. or more likely the reviewers are simply busy. In either case, we ask for your understanding during the review process. For more information, see our FAQ on this topic: http:\/\/scikit- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-11596-05e57c5dd6962e705846f516b1a97d75cc60bfed2c091a63c3800c65087380a7","text":"Add sklearn.showversions() similar to pandas.showversions (with numpy blas binding info) Some numeric issues are related to the specific blas that numpy is using. I'm wondering if it makes sense to add the relevant ` invocations to the template to make it easier for people to report."}
{"_id":"q-en-scikit-learn__scikit-learn-12258-edf2cc7a780928d70858e8990fdcf89640a800543eeaec4d95ace327b79878ab","text":"silhouettesamples gives incorrect result from precomputed distance matrix with diagonal entries silhouettesamples gives incorrect result from precomputed distance matrix with diagonal entries. When using silhouettesamples and metric='precomputed', if the input distance matrix has non-zero values along the diagonal then the silhouette scores are incorrect. Suggested Solution Before calculating the scores the diagonal entries of a precomputed distance matrix should be set to zero. Example: [0., 0.952, 0., 0., 0., 0.] [0., 0.552, 0., 0., 0., 0.] Darwin-17.7.0-x8664-i386-64bit Python 3.6.4 (default, Jan 16 2018, 12:04:33) [GCC 4.2.1 Compatible Clang 4.0.1 (tags\/RELEASE_401\/final)] NumPy 1.15.1 SciPy 1.1.0 Scikit-Learn 0.20.0"}
{"_id":"q-en-scikit-learn__scikit-learn-12421-46f576f4f8993d11dd8a897da929dd22812fd4322370f5173739c993e92024b9","text":"OPTICS: self.coredistances inconsistent with documentation&R implementation In the doc, we state that `, but it's not the case. Result from scikit-learn: Result from R:"}
{"_id":"q-en-scikit-learn__scikit-learn-12443-133711e88d511e26846eb2e652606afd554221092ef5a42a2f8507d887aea61f","text":"OneHotEncoder throws unhelpful error messages when tranform called prior to fit <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --<!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--OneHotEncoder throws an instead of a when tranform is called prior to fit if is called prior to being fit an is thrown if includes arrays of of unicode type <!-- Example: If the code is too long, feel free to put it in a public gist and link it in the issue: --<!-- Example: No error is thrown. Please paste or describe the expected results.--<!-- Please paste or specifically describe the actual output or traceback. --<!-- Please run the following snippet and paste the output below. For scikit-learn = 0.20: import sklearn; sklearn.showversions() For scikit-learn < 0.20: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"NumPy\", numpy.version) import scipy; print(\"SciPy\", scipy.version) import sklearn; print(\"Scikit-Learn\", sklearn.version) --<!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-12462-22effba65c751f1bf512a769738b4d12e8dc86bb59381af22d145f67145f0bd8","text":"SkLearn method generating error with Dask DataFrames When using Dask Dataframes with SkLearn, I used to be able to just ask SkLearn for the score of any given algorithm. It would spit out a nice answer and I'd move on. After updating to the newest versions, all metrics that compute based on (ytrue, ypredicted) are failing. I've tested , , , and Work-around shown below, but it's not ideal because it requires me to cast from Dask Arrays to numpy arrays which won't work if the data is huge. I've asked Dask about it here: and they've said it's an issue with the SkLearn check, and that they won't be addressing it. It seems like it should be not super complicated to add a that says \"if shape doesn't return a tuple revert to pretending shape didn't exist\". If others think that sounds correct, I can attempt a pull-request, but I don't want to attempt to solve it on my own only to find out others don't deem that an acceptable solutions. Trace, MWE, versions, and workaround all in-line. MWE: Output of error: Problem occurs after upgrading as follows: Before bug: Update from conda, then bug starts: Work around:"}
{"_id":"q-en-scikit-learn__scikit-learn-12471-8c5e1ca1d0bd18c668074e46a0dcd0e9eef0ef756130e4ed6a47914546f0e47b","text":"OneHotEncoder ignore unknown error when categories are strings This bug is very specific, but it happens when you set OneHotEncoder to ignore unknown entries. and your labels are strings. The memory of the arrays is not handled safely and it can lead to a ValueError Basically, when you call the transform method it will sets all the unknown strings on your array to OneHotEncoder.categories[i][0] which is the first category alphabetically sorted given for fit If this OneHotEncoder.categories[i][0] is a long string, and the array that you want to transform has small strings, then it is impossible to fit the whole OneHotEncoder.categories[i][0] into the entries of the array we want to transform. So OneHotEncoder.categories[i][0] is truncated and this raise the ValueError. Here we should get an sparse matrix 2x4 false everywhere except at (1,1) the '22' that is known System: python: 2.7.12 (default, Dec 4 2017, 14:50:18) [GCC 5.4.0 ] machine: Linux-4.4.0-138-generic-x8664-with-Ubuntu-16.04-xenial executable: \/usr\/bin\/python BLAS: macros: HAVECBLAS=None cblaslibs: openblas, openblas libdirs: \/usr\/lib Python deps: Cython: 0.25.2 scipy: 0.18.1 setuptools: 36.7.0 pip: 9.0.1 numpy: 1.15.2 pandas: 0.19.1 sklearn: 0.21.dev0 I already implemented a fix for this issue, where I check the size of the elements in the array before, and I cast them into objects if necessary."}
{"_id":"q-en-scikit-learn__scikit-learn-12486-590950253436eec53711953fc6b8876351673d253658258d20a2b8002c6a9655","text":"ck estimator is classifier & numclasses=2 in <!-- Thanks for contributing a pull request! Please ensure you have taken a look at the contribution guidelines: --<!-- Example: See also . Please use keywords (e.g., Fixes) to create link to the issues or pull requests you resolved, so that they will automatically be closed when your pull request is merged. See --We are fixing this issue: We a test in the file that raises a ValueError if the user is either trying to use a non classifier model for a classification problem, or is using a dataset with only one class. <!-- Please be aware that we are a loose team of volunteers so patience is necessary; assistance handling other issues is very welcome. We value all user contributions, no matter how minor they are. If we are slow to review, either the pull request needs some benchmarking, tinkering, convincing, etc. or more likely the reviewers are simply busy. In either case, we ask for your understanding during the review process. For more information, see our FAQ on this topic: http:\/\/scikit- Thanks for contributing! -- BUG: Using GridSearchCV with scoring='rocauc' and GMM as classifier gives IndexError When performing grid search using GridSearchCV using ootb scoring method 'rocauc' and ootb GMM classifier from sklearn.mixture.GMM I get an index error. Code to reproduce: Sorry if the format is incorrect. First time I am posting. ERROR: File \"*\/python2.7\/site-\", line 175, in call ypred = y_pred[:, 1] IndexError: index 1 is out of bounds for axis 1 with size 1"}
{"_id":"q-en-scikit-learn__scikit-learn-12583-876aacb67dda97b457308777e3b89db43c2762b7096757d993d0aaf55365840f","text":"addindicator switch in imputers For whatever imputers we have, but especially , we should have an parameter, which simply stacks a transform onto the output of the imputer's ."}
{"_id":"q-en-scikit-learn__scikit-learn-12585-bf5d84233fa8da7dc425cff91e4d7753956b6f46ca0fa3bee59059b33cbf0e41","text":"clone fails for parameters that are estimator types fails when one or more instance parameters are estimator types (i.e. not instances, but classes). I know this is a somewhat unusual use case, but I'm working on a project that provides wrappers for sklearn estimators () and I'd like to store the wrapped estimators as their classes - not their instances - as a parameter inside of a wrapper that behaves like an estimator itself. from sklearn.preprocessing import StandardScaler from import clone clone(StandardScaler(withmean=StandardScaler)) No error. Change , line 51 to: elif not hasattr(estimator, 'getparams') or isinstance(estimator, type): I'm not sure whether this might break stuff in other places, however. I'd happily submit a PR if this change is desired. sklearn: 0.20.0"}
{"_id":"q-en-scikit-learn__scikit-learn-12625-903eb42900cdce8d082adbcdb364f179348698216b1f81c2ece0b1a111748c2b","text":"TypeError: \"iteration over a 0-d array\" when trying to a pandas.Series <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --<!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--When trying to call on a instance, an error is thrown with scikit-learn version 0.20.0. Version 0.19.1. works just fine. The states that the input to can be \"array-like\", and  should fulfill this requirement since it is a \"one-dimensional ndarray\". <!-- Example: If the code is too long, feel free to put it in a public gist and link it in the issue: --<!-- Example: No error is thrown. Please paste or describe the expected results.--This should be the output (as it is in version 0.19.1): A workaround is replacing with , which also yields this output. <!-- Please paste or specifically describe the actual output or traceback. --<!-- Please run the following snippet and paste the output below. For scikit-learn = 0.20: import sklearn; sklearn.showversions() For scikit-learn < 0.20: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"NumPy\", numpy.version) import scipy; print(\"SciPy\", scipy.version) import sklearn; print(\"Scikit-Learn\", sklearn.version) --<!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-12626-00c47c858c1c33ecc120c875c28701823044f7cfd97455c0041c1a288df4d7cf","text":"makecolumntransformer has different order of arguments than ColumnTransformer I'm not sure if we discussed this or did this on purpose, but I find this very confusing. `. I guess it's too late to change this? Though we did say it's experimental and subject to change."}
{"_id":"q-en-scikit-learn__scikit-learn-12656-36d245a417a7343cc4d8604a1b417c701698246a54d410d5c182dc8fdbcf9714","text":"Fix Issue <!-- Thanks for contributing a pull request! Please ensure you have taken a look at the contribution guidelines: --<!-- Example: See also . Please use keywords (e.g., Fixes) to create link to the issues or pull requests you resolved, so that they will automatically be closed when your pull request is merged. See -- Removed the labels parameter in hammingloss function. Removed all instances of labels within the method. Since hammingloss is used only with mulitlabel input, changed len(labels) to [1]. First time contributing to open source! Please let me know what I can do to improve! <!-- Please be aware that we are a loose team of volunteers so patience is necessary; assistance handling other issues is very welcome. We value all user contributions, no matter how minor they are. If we are slow to review, either the pull request needs some benchmarking, tinkering, convincing, etc. or more likely the reviewers are simply busy. In either case, we ask for your understanding during the review process. For more information, see our FAQ on this topic: http:\/\/scikit- Thanks for contributing! --[MRG] Handling parameter labels removal from hammingloss <!-- Thanks for contributing a pull request! Please ensure you have taken a look at the contribution guidelines: --<!-- Fix Issue Please use keywords (e.g., Fixes) to create link to the issues or pull requests you resolved, so that they will automatically be closed when your pull request is merged. See --<!-- Please be aware that we are a loose team of volunteers so patience is necessary; assistance handling other issues is very welcome. We value all user contributions, no matter how minor they are. If we are slow to review, either the pull request needs some benchmarking, tinkering, convincing, etc. or more likely the reviewers are simply busy. In either case, we ask for your understanding during the review process. For more information, see our FAQ on this topic: http:\/\/scikit- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-12704-a324c7b66133f28b32a653d18d7c314e712b973127f85a4f376bddd0e1543edb","text":"regression in ColumnTransformer in in 0.20.1 with columns=pd.Index This came up in one of my teaching notebooks actually (and might be in my book as well). This is very natural because columns are of type pd.Index, and so if you take some subset of columns from ` you'll now run into this error. So... 0.20.2?"}
{"_id":"q-en-scikit-learn__scikit-learn-12733-f514e5afcfdb8cebe2f45c19efe7408607792294f2f832c54ea7f167a02bb737","text":"traintestsplit excepting negative integers and floats The following minimal example doesn't fail: Is it a bug or indented feature? According to the docs neither of the above should make sense. See for reference for example."}
{"_id":"q-en-scikit-learn__scikit-learn-12758-8c4feaedc3bb5d499248674db2e25c545399ab3fc1dd9d845d412b34a8b2db5d","text":"Showing micro-average in classification report is confusing This is a follow up on . I don't think it makes sense to include the micro-average for multi-class classification. The three columns will always show the same value, all of which being the same as accuracy. I find that confusing. If you want to show this (I don't see why you'd want to show the same number three times), I would at least make it clear in the report that it's accuracy. IncrementalPCA fails if data size % batch size < ncomponents throws The error occurs because the last batch generated by may be smaller than . I reduced the iris data to 101 instances, so the last batch has only a single data instance, which is less than the number of components. As far as I see, none of the current unit tests run into this. ( could, if the code that raises the exception would compare with - which it should, but doesn't). Skipping the last batch if it is to small, that is, changing to fixes the problem. please confirm that this solution seems OK before I go preparing the PR and tests. No error is thrown."}
{"_id":"q-en-scikit-learn__scikit-learn-12760-3bcc5d53071d1384447b5d1c6fe1a192ab5f341be312e2ec27c80402864e3e4c","text":"Davies Bouldin measure: division by zero I'm facing a problem with the davies bouldin measure. This is the warning that I get: .local\/lib\/python3.7\/site- RuntimeWarning: divide by zero encountered in truedivide score = (intradists[:, None] + intradists) \/ centroiddistances This is the implementation in sklearn: I found on stack overflow: With this implementation I don't get any warnings, but the results differ: Davies Bouldin measure: division by zero I'm facing a problem with the davies bouldin measure. This is the warning that I get: .local\/lib\/python3.7\/site- RuntimeWarning: divide by zero encountered in truedivide score = (intradists[:, None] + intradists) \/ centroiddistances This is the implementation in sklearn: I found on stack overflow: With this implementation I don't get any warnings, but the results differ:"}
{"_id":"q-en-scikit-learn__scikit-learn-12827-3e21883b9721cd96dd11047f45195763fe7ffef2208b57c274f2c27c49269881","text":"DOC: add details to QuantileTransformer documentation I think the documentation of should say how it is implemented. There is even a about it and we could take some elements of the answer. Besides I was thinking that to map to a uniform distribution, the implementation was just computing the empirical cdf of the columns but it does not seem to be the case."}
{"_id":"q-en-scikit-learn__scikit-learn-12834-2e959068a6cc4b69e3e5056453c7b86f87796f13c6253d145d49609ef19ae838","text":"fails for multioutput ensemble models with non-numeric DVs <!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--Multioutput forest models assume that the dependent variables are numeric. Passing string DVs returns the following error: I'm going to take a stab at submitting a fix today, but I wanted to file an issue to document the problem in case I'm not able to finish a fix. I wrote a test based on which currently fails: No error is thrown, can run for all ensemble multioutput models <!-- Example: No error is thrown. Please paste or describe the expected results.--<!-- Please paste or specifically describe the actual output or traceback. --I replicated this error using the current master branch of sklearn (0.21.dev0). <!-- Please run the following snippet and paste the output below. For scikit-learn = 0.20: import sklearn; sklearn.showversions() For scikit-learn < 0.20: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"NumPy\", numpy.version) import scipy; print(\"SciPy\", scipy.version) import sklearn; print(\"Scikit-Learn\", sklearn.version) --<!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-12860-9f7fe94a0217b7023ca761181b7ef4fa8e2ea5ccd1838fb979c6defc124ccb62","text":"Suggestion: Add support for unpenalized logistic regression provides unpenalized OLS, and , which supports , also supports . But if you want plain old unpenalized logistic regression, you have to fake it by setting in to a large number, or use from instead."}
{"_id":"q-en-scikit-learn__scikit-learn-12938-7891a056eda2a9271b9a2239f1fbda61f3bf85668e61505c7d4f90486308d479","text":"AttributeError: 'PrettyPrinter' object has no attribute 'indentat_name' There's a failing example in , and here's a piece of code causing it: With the stack trace:"}
{"_id":"q-en-scikit-learn__scikit-learn-12961-a58227d70629a40e7db4a7d547fe1eb7530179df87353b4db75714cd8f597f2b","text":"ValueError not enough values to unpack (expected 5, got 0) I'm using with in sklearn v 0.20.1. MacOS Mojave. The issue traces to these two lines below, with Setting did not work, still hitting the same issue. From the trackback, I suspect that the line below could be returning None. This doesn't always happen, but only sometimes. My param search was able to run for a few rounds, and then hit this issue. The problem could be that lightgbm did not return valid results (have not verified), but either way, this error could be handled better. Does anyone have more insight into the potential cause? Thanks."}
{"_id":"q-en-scikit-learn__scikit-learn-12973-6c8312452a9c2a27f78220fc418a98957c590f6524255e9a03791ab3a71c705c","text":"LassoLarsIC: unintuitive copyX behaviour Hi, I would like to report what seems to be a bug in the treatment of the parameter of the class. Because it's a simple bug, it's much easier to see in the code directly than in the execution, so I am not posting steps to reproduce it. As you can see here, LassoLarsIC accepts a copyX parameter. However, it also takes a copyX parameter a few lines below, in the definition of . Now there are two values (potentially contradicting each other) for copyX and each one is used once. Therefore can have a mixed behaviour. Even worse, this can be completely invisible to the user, since copyX has a default value of True. Let's assume that I'd like it to be False, and have set it to False in the initialization, . I then call and my choice will be silently overwritten. Ideally I think that copy_X should be removed as an argument in . No other estimator seems to have a duplication in class parameters and fit arguments (I've checked more than ten in the linear models module). However, this would break existing code. Therefore I propose that takes a default value of and only overwrites the existing value if the user has explicitly passed it as an argument to . I will submit a PR to that effect."}
{"_id":"q-en-scikit-learn__scikit-learn-12989-04d8d52bccbce102d6f71b0648ad0ba05340f9fcaf2505e16403b680c97a1bdd","text":"and have inconsistent default init <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --<!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--and have inconsistent default init. has while has . See As suggested, we could change the default in with a deprecation process. <!-- Example: If the code is too long, feel free to put it in a public gist and link it in the issue: --<!-- Example: No error is thrown. Please paste or describe the expected results.--<!-- Please paste or specifically describe the actual output or traceback. --<!-- Please run the following snippet and paste the output below. For scikit-learn = 0.20: import sklearn; sklearn.showversions() For scikit-learn < 0.20: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"NumPy\", numpy.version) import scipy; print(\"SciPy\", scipy.version) import sklearn; print(\"Scikit-Learn\", sklearn.version) --<!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-13010-1a86caef19c697a8922d6d52d777af730658218b89064a7f064bb23ac93add24","text":"StandardScaler fit overflows on float16 When using StandardScaler on a large float16 numpy array the mean and std calculation overflows. I can convert the array to a larger precision but when working with a larger dataset the memory saved by using float16 on smaller numbers kind of matter. The error is mostly on numpy. Adding the dtype on the mean\/std calculation does it but I'm not sure if that how people here would like to do it. The normalized array"}
{"_id":"q-en-scikit-learn__scikit-learn-13017-5db2374690a325374334093375865e7d9966688db15689a8ff01fe52568fa66c","text":"ColumnTransformer behavior for negative column indexes <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --The behavior of when negative integers are passed as column indexes is not clear. The first transformer should either raise an error or give the same result as the second transformer"}
{"_id":"q-en-scikit-learn__scikit-learn-13046-dae3efca16863dddc50319d77e5e0d0f3e72bc9dce294a5d21ed16f7b0216bf3","text":"MissingIndicator failed with non-numeric inputs fails with string and object type numpy arrays"}
{"_id":"q-en-scikit-learn__scikit-learn-13087-9ffecdc28f866aa716fb95fb8f349d72efb93277bad79c247a5182ae2b971cc2","text":"Feature request: support for arbitrary bin spacing in calibration.calibrationcurve I was using , and it currently accepts an parameter to specify the number of bins to evenly partition the probability space between 0 and 1. However, I am using this in combination with a gradient boosting model in which the probabilities are very uncalibrated, and most of the predictions are close to 0. When I use the calibrated classifier, the result is very noisy because there are many data points in some bins and few, if any, in others (see example below). In the code below, I made a work-around to do what I want and show a plot of my output (in semilog space because of the skewed distribution). I haven't contributed to a large open-source project before, but if there's agreement this would be a useful feature, I would be happy to try to draft up a PR. Notice the large disparity in results between the different bins chosen. For this reason, I think the user should be able to choose the bin edges, as in numpy's or matplotlib's functions. !"}
{"_id":"q-en-scikit-learn__scikit-learn-13124-dfe0314ba08a3c33d21ea749de298cd4f095918e3e6f3cf71020970f5e7687e7","text":"sklearn.modelselection.StratifiedKFold either shuffling is wrong or documentation is misleading <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --Regarding the shuffle parameter, the documentation states: \"Whether to shuffle each stratification of the data before splitting into batches\". However, instead of shuffling samples within each stratum, the order of batches is shuffled. As you can see in the output below, 1 is always paired with 11, 2 with 12, 3 with 13, etc. regardless whether shuffle parameter is True or False. When shuffle=True, the batches are always the same for any randomstate, but appear in a different order. When cross-validation is performed, the results from each batch are summed and then divided by the number of batches. Changing the order of batches does not change the result. The way shuffle works now is completely useless from cross-validation perspective. import numpy as np from sklearn.modelselection import StratifiedKFold RANDOMSEED = 1 samplesperclass = 10 X = np.linspace(0, samplesperclass2-1, samplesperclass 2) y = np.concatenate(((samplesperclass), (samplesperclass)), axis=0) print(X, '\\n', y, '\\n') print('\\nshuffle = False\\n') kfold = StratifiedKFold(nsplits=10, shuffle=False, randomstate=RANDOMSEED) result = 0 for foldn, (trainidx, testidx) in enumerate((X, y)): print(trainidx, '\\n', testidx) print('\\nshuffle = True, Random seed =', RANDOMSEED, '\\n') kfold = StratifiedKFold(nsplits=10, shuffle=True, randomstate=RANDOMSEED) result = 0 for foldn, (trainidx, testidx) in enumerate((X, y)): print(trainidx, '\\n', testidx) RANDOMSEED += 1 print('\\nshuffle = True, Random seed =', RANDOMSEED, '\\n') kfold = StratifiedKFold(nsplits=10, shuffle=False, randomstate=RANDOMSEED) result = 0 for foldn, (trainidx, testidx) in enumerate((X, y)): print(trainidx, '\\n', testidx) <!-- Example: No error is thrown. Please paste or describe the expected results.--I expect batches to be different when Shuffle is turned on for different randomstate seeds. But they are the same <!-- Please paste or specifically describe the actual output or traceback. --[ 0. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 19.] [1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.] shuffle = False [ 1 2 3 4 5 6 7 8 9 11 12 13 14 15 16 17 18 19] [ 0 10] [ 0 2 3 4 5 6 7 8 9 10 12 13 14 15 16 17 18 19] [ 1 11] [ 0 1 3 4 5 6 7 8 9 10 11 13 14 15 16 17 18 19] [ 2 12] [ 0 1 2 4 5 6 7 8 9 10 11 12 14 15 16 17 18 19] [ 3 13] [ 0 1 2 3 5 6 7 8 9 10 11 12 13 15 16 17 18 19] [ 4 14] [ 0 1 2 3 4 6 7 8 9 10 11 12 13 14 16 17 18 19] [ 5 15] [ 0 1 2 3 4 5 7 8 9 10 11 12 13 14 15 17 18 19] [ 6 16] [ 0 1 2 3 4 5 6 8 9 10 11 12 13 14 15 16 18 19] [ 7 17] [ 0 1 2 3 4 5 6 7 9 10 11 12 13 14 15 16 17 19] [ 8 18] [ 0 1 2 3 4 5 6 7 8 10 11 12 13 14 15 16 17 18] [ 9 19] shuffle = True, Random seed = 1 [ 0 1 3 4 5 6 7 8 9 10 11 13 14 15 16 17 18 19] [ 2 12] [ 0 1 2 3 4 5 6 7 8 10 11 12 13 14 15 16 17 18] [ 9 19] [ 0 1 2 3 4 5 7 8 9 10 11 12 13 14 15 17 18 19] [ 6 16] [ 0 1 2 3 5 6 7 8 9 10 11 12 13 15 16 17 18 19] [ 4 14] [ 1 2 3 4 5 6 7 8 9 11 12 13 14 15 16 17 18 19] [ 0 10] [ 0 1 2 4 5 6 7 8 9 10 11 12 14 15 16 17 18 19] [ 3 13] [ 0 2 3 4 5 6 7 8 9 10 12 13 14 15 16 17 18 19] [ 1 11] [ 0 1 2 3 4 5 6 8 9 10 11 12 13 14 15 16 18 19] [ 7 17] [ 0 1 2 3 4 5 6 7 9 10 11 12 13 14 15 16 17 19] [ 8 18] [ 0 1 2 3 4 6 7 8 9 10 11 12 13 14 16 17 18 19] [ 5 15] shuffle = True, Random seed = 2 [ 1 2 3 4 5 6 7 8 9 11 12 13 14 15 16 17 18 19] [ 0 10] [ 0 2 3 4 5 6 7 8 9 10 12 13 14 15 16 17 18 19] [ 1 11] [ 0 1 3 4 5 6 7 8 9 10 11 13 14 15 16 17 18 19] [ 2 12] [ 0 1 2 4 5 6 7 8 9 10 11 12 14 15 16 17 18 19] [ 3 13] [ 0 1 2 3 5 6 7 8 9 10 11 12 13 15 16 17 18 19] [ 4 14] [ 0 1 2 3 4 6 7 8 9 10 11 12 13 14 16 17 18 19] [ 5 15] [ 0 1 2 3 4 5 7 8 9 10 11 12 13 14 15 17 18 19] [ 6 16] [ 0 1 2 3 4 5 6 8 9 10 11 12 13 14 15 16 18 19] [ 7 17] [ 0 1 2 3 4 5 6 7 9 10 11 12 13 14 15 16 17 19] [ 8 18] [ 0 1 2 3 4 5 6 7 8 10 11 12 13 14 15 16 17 18] [ 9 19] System: python: 3.7.2 (default, Jan 13 2019, 12:50:01) [Clang 10.0.0 (clang-1000.11.45.5)] executable: \/usr\/local\/opt\/python\/bin\/python3.7 machine: Darwin-18.2.0-x8664-i386-64bit BLAS: macros: NOATLASINFO=3, HAVECBLAS=None libdirs: cblaslibs: cblas Python deps: pip: 18.1 setuptools: 40.6.3 sklearn: 0.20.2 numpy: 1.15.2 scipy: 1.2.0 Cython: None pandas: 0.23.4 <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-13135-5cce6af9fbee808605290e39bc48dac195724ef0246d288ea02d37ca210923e6","text":"KBinsDiscretizer: kmeans fails due to unsorted binedges with fails in certain situations, due to centers and consequently binedges being unsorted, which is fatal for np.digitize. A very simple way to reproduce this is to set nbins in the existing testnonuniformstrategies from to a higher value (here 5 instead of 3). In this simple example it seems like an edge case to set nbins to almost the number of data points. However I've seen this happen in productive situations with very reasonable number of bins of order log2(number of unique values of X). No error is thrown. <!-- Thanks for contributing! --KBinsDiscretizer: kmeans fails due to unsorted binedges with fails in certain situations, due to centers and consequently binedges being unsorted, which is fatal for np.digitize. A very simple way to reproduce this is to set nbins in the existing testnonuniformstrategies from to a higher value (here 5 instead of 3). In this simple example it seems like an edge case to set nbins to almost the number of data points. However I've seen this happen in productive situations with very reasonable number of bins of order log2(number of unique values of X). No error is thrown. <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-13142-a1fc730aa2e26862787be555254ae863a59926dca0943337689aa378cbae30e2","text":"GaussianMixture predict and fitpredict disagree when ninit1 When is specified in GaussianMixture, the results of fitpredict(X) and predict(X) are often different. The unit test doesn't catch this because it does not set . No exceptions."}
{"_id":"q-en-scikit-learn__scikit-learn-13143-87ecf055eb03b4e34bae81aa70ee0ed6e14a3411d0757c9bcdbca1877154424a","text":"precisionscore shows incorrect value precisionscore shows incorrect value ([[0,0,1],[0,1,0],[0,0,1]]) B=A precisionscore(A,B, average=None) array([ 0., 1., 1.]) array([ 1., 1., 1.]) array([ 0., 1., 1.]) import platform; print(platform.platform()) Darwin-14.5.0-x8664-i386-64bit import sys; print(\"Python\", sys.version) ('Python', '2.7.10 (default, Jul 14 2015, 19:46:27) \\n[GCC 4.2.1 Compatible Apple LLVM 6.0 (clang-600.0.39)]') import numpy; print(\"NumPy\", numpy.version) ('NumPy', '1.13.3') import scipy; print(\"SciPy\", scipy.version) ('SciPy', '1.0.0') import sklearn; print(\"Scikit-Learn\", sklearn.version) ('Scikit-Learn', '0.18.1')"}
{"_id":"q-en-scikit-learn__scikit-learn-13157-8ab2f72371c5e798d109a261db3bdd501b4b405f100ea2ae0b48eb80c3479ee9","text":"Different r2score multioutput default in r2score and base.RegressorMixin We've changed multioutput default in r2score to \"uniformaverage\" in 0.19, but in base.RegressorMixin, we still use ` (). Also see the strange things below:"}
{"_id":"q-en-scikit-learn__scikit-learn-13165-3f039653dea62a2ff2d175f40652f435ac0a57f88673b2090751c5d766e7f12d","text":"Ensure monotonic bin edges for KBinsDiscretizer strategy quantile The percentiles returned from np.percentile are monotonic up to possible numeric instabilities. Monotonicity is enforced by applying a simple maximum on subsequent values to deal with this case and increase robustness. The additional line is a no-op in almost all cases. This is unfortunate, but since there is essentially no performance impact, I guess robustness is worth the effort."}
{"_id":"q-en-scikit-learn__scikit-learn-13221-4823c2ec27e21eadd178bc6bd2e0248d86ed74e51c8733edcd8ee3551b6d4b0d","text":"gamma='scale' in SVC I believe that setting in is not meeting its intended purpose of being invariant to the scale of . Currently, is set to . However, I believe it should be . Rationale: if you scale by 10 you need to scale by 1\/100, not 1\/10, to achieve the same results. See the definition of the RBF kernel : the \"units\" of are 1\/x^2, not 1\/x. I also tested this empirically: scaling by 10 and scaling by 1\/100 gives the same result as the original, whereas scaling by 10 and scaling by 1\/10 gives a different result. Here is some code: Note that will become the default setting for in version 0.22. Related: ,"}
{"_id":"q-en-scikit-learn__scikit-learn-13241-577a9c2a3c71ac10260e461fd7839ddf7682e270ee672944a0c85cc08078fef2","text":"Differences among the results of KernelPCA with rbf kernel Hi there, I met with a problem: When I run KernelPCA for dimension reduction for the same datasets, the results are different in signs. Just to reduce the dimension to 7 with rbf kernel: pca = KernelPCA(ncomponents=7, kernel='rbf', copyX=False, njobs=-1) pca.fittransform(X) The same result. The results are the same except for their signs:( [[-0. -0. -0. 0. -0. -0. 0.]] [[ 0. 0. 0. -0. -0. -0. -0.]] [[-0. -0. 0. 0. 0. 0. 0.]] 18.1"}
{"_id":"q-en-scikit-learn__scikit-learn-13253-af8cebe3f3560a95bd580922f16cea1653f7a31cefcf1d5271465bf6f21d8c95","text":"ENH: support DataFrames in OneHot\/OrdinalEncoder without converting to array Left-over to do from Idea is to support DataFrames without converting to a contiguous array. This conversion is not needed, as the transformer encodes the input column by column anyway, so it would be rather easy to preserve the datatypes per column. This would avoid converting a potentially mixed-dtype DataFrame (eg ints and object strings) to a full object array. This can introduces a slight change in behaviour (it can change the of the in certain edge cases, eg when you had a mixture of float and int columns). (Note that is not yet necessarily means to have special handling for certain pandas dtypes such as categorical dtype, see , in an initial step, we could still do a on each column \/ coerce each column to a numpy array)."}
{"_id":"q-en-scikit-learn__scikit-learn-13280-559789aa73792fd5f804f5ec9d0b5afc9ddfddc588f7a769e50047671b05cf86","text":"partialfit does not account for unobserved target values when fitting priors to data My understanding is that priors should be fitted to the data using observed target frequencies and a variant of to avoid assigning 0 probability to targets not yet observed. It seems the implementation of does not account for unobserved targets at the time of the first training batch when computing priors. RuntimeWarning: divide by zero encountered in log self.classlogprior = ((self.classcount) - scikit-learn version: 0.20.2 This behavior is not very intuitive to me. It seems requires for the right reason, but doesn't actually offset target frequencies to account for unobserved targets."}
{"_id":"q-en-scikit-learn__scikit-learn-13283-ec20c1011e0617ae69b0e297d6e66ca1fb0b3c976b59808348d0092d0703d213","text":"Isolation forest - decisionfunction & averagepathlength method are memory inefficient Isolation forest consumes too much memory due to memory ineffecient implementation of anomoly score calculation. Due to this the parallelization with njobs is also impacted as anomoly score cannot be calculated in parallel for each tree. Run a simple Isolation forest with nestimators as 10 and as 50 respectively. On memory profiling, it can be seen that each building of tree is not taking much memory but in the end a lot of memory is consumed as a for loop is iteration over all trees and calculating the anomoly score of all trees together and then averaging it. - line 267-281 ` Due to this, in case of more no. of estimators(1000), the memory consumed is quite high. Possible Solution: The above for loop should only do the averaging of anomoly score from each estimator instead of calculation. The logic of isoforest anomoly score calculation can be moved to base estimator class so it is done for each tree( i guess file-similar to other method available after fitting) The memory consumption is profound as we increase no. of estimators. The fit method calls decision function & average anomoly score which are taking quite a lot memory. the memory spike is too high in the very end, that is in finall call to method. <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-13313-b66dcb193fa805722bb30413739306ec6e9985ba4ce8f0562d54b4bfd431b5c7","text":"checkclassweightbalancedclassifiers is never run?! checkclassweightbalancedclassifiers(name, Classifier, Xtrain, ytrain, Same for `"}
{"_id":"q-en-scikit-learn__scikit-learn-13328-28d8defe9620605041dcd8f1b834bc2e9f93e2273ced3ca1d7cc5390b767738e","text":"TypeError when supplying a boolean X to HuberRegressor fit when fitting with boolean predictors. No error is thrown when of is (second line of code in the snipped above, ) Boolean array is expected to be converted to by as it is done by, say . is thrown: Latest versions of everything as far as I am aware: <!-- Thanks for contributing! --<!-- NP! --"}
{"_id":"q-en-scikit-learn__scikit-learn-13333-bc5bbeeb78fd9a20a2e14123286a9baf3198e77e60ab0acc1b3044fac48967d6","text":"DOC Improve doc of nquantiles in QuantileTransformer The uses numpy.percentile(Xtrain, .) as the estimator of the quantile function of the training data. To know this function perfectly we just need to take . Then it is just a linear interpolation (which is done in the code afterwards). Therefore I don't think we should be able to choose and we should prevent users from thinking that the higher the better the transformation. As mentioned by IRL it is however true that it can be relevant to choose when is very large. I suggest to add more information on the impact of in the doc which currently reads: For example using 100 times more landmarks result in the same transformation Interestingly if you do not choose correctly, the linear interpolation done afterwards does not correspond to the numpy.percentile(Xtrain, .) estimator. This is not \"wrong\" as these are only estimators of the true quantile function\/cdf but I think it is confusing and would be better to stick with the original estimator. For instance, the following raises an AssertionError."}
{"_id":"q-en-scikit-learn__scikit-learn-13363-11ea2763cf603229d35a28206fbb6183b583c9e1b775e7cd378187cc3a4f5fc9","text":"returnintercept==True in ridgeregression raises an exception <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --<!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--<!-- Example: No error is thrown. Please paste or describe the expected results.--(the values can differ, but at least no exception should be raised) <!-- Please paste or specifically describe the actual output or traceback. --<!-- Please run the following snippet and paste the output below. For scikit-learn = 0.20: import sklearn; sklearn.showversions() For scikit-learn < 0.20: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"NumPy\", numpy.version) import scipy; print(\"SciPy\", scipy.version) import sklearn; print(\"Scikit-Learn\", sklearn.version) --<!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-13368-9ca62f446136785479b9e79598ab8516d5587de3fb4231e7b8311d80216c0e5c","text":"crossvalpredict returns bad prediction when evaluated on a dataset with very few samples returns bad prediction when evaluated on a dataset with very few samples on 1 class, causing class being ignored on some CV splits. Verified on the scikit latest dev version."}
{"_id":"q-en-scikit-learn__scikit-learn-13436-aca5254569b57162794adb0329b7ec01da85abc575f95828bfbedbad3601094d","text":"Confusing pretty print repr for nested Pipeline Taking the examples from the docs (https:\/\/scikit-) that involves some nested pipelines in columntransformer in pipeline The repr that you get for this pipeline: which I found very confusing: the outer pipeline seems to have only 1 step (the 'preprocessor', as the 'classifier' disappeared in the ). It's probably certainly not easy to get a good repr in all cases, and for sure the old behaviour was even worse (it would show the first 'imputer' step of the pipeline inside the column transformer as if it was the second step of the outer pipeline ..). But just opening this issue as a data point for possible improvements. Without knowing how the current repr is determined: ideally I would expect that, if the full repr is too long, we first try to trim it step per step of the outer pipeline, so that the structure of that outer pipeline is still visible. But that is easier to write than to code .. :) cc"}
{"_id":"q-en-scikit-learn__scikit-learn-13447-8577cf1e736bea655ab31d41575c0e8fa4f3df86016aee24e0d047048be31a4a","text":"labelrankingaverageprecisionscore: sampleweighting isn't applied to items with zero true labels labelrankingaverageprecisionscore offers a sampleweighting argument to allow nonuniform contribution of individual samples to the reported metric. Separately, individual samples whose labels are the same for all classes (all true or all false) are treated as a special case (precision == 1, ). However, this special case bypasses the application of sampleweight (). So, in the case where there is both non-default sampleweighting and samples with, for instance, zero labels, the reported metric is wrong. <!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--See example in Average of AP of first and second samples = 0.625 Sum of AP of all three samples, divided by sum of weighting vector = 2.25\/2 = 1.125 System: python: 3.6.7 (default, Oct 22 2018, 11:32:17) [GCC 8.2.0] executable: \/usr\/bin\/python3 machine: Linux-4.14.79+-x8664-with-Ubuntu-18.04-bionic BLAS: macros: SCIPYMKLH=None, HAVECBLAS=None libdirs: \/usr\/local\/lib cblaslibs: mklrt, pthread Python deps: pip: 19.0.3 setuptools: 40.8.0 sklearn: 0.20.3 numpy: 1.14.6 scipy: 1.1.0 Cython: 0.29.6 pandas: 0.22.0 <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-13454-fbe17b50d1a738951deb526d344e85ff4f41480866dd16e566920750326ce741","text":"Confusing error message in OrdinalEncoder when passing single list of categories Small example: The error message is still using the old , which makes it very confusing. (another issue is that we might be able to actually detect this case)"}
{"_id":"q-en-scikit-learn__scikit-learn-13467-4758e74c1db78ed86f375ec20644f56bb3a12d31fbfa46bf040f339e8bc249e1","text":"Implement RMSE (root-mean-square error) metric and scorer RMSE seems to be a popular metric but now one has to calculate it through `. Wiki page:"}
{"_id":"q-en-scikit-learn__scikit-learn-13472-86746692c4dc2a991c0dabf8556847ee3f9275537e11f1e792eba6f70229ed0d","text":"GradientBoostingRegressor initial estimator does not play together with Pipeline Using a pipeline as the initial estimator of GradientBoostingRegressor doesn't work due to incompatible signatures. The signature of is which cannot be called with three positional arguments as above. So I guess line 1421 in should read instead and the issue is solved. Right?"}
{"_id":"q-en-scikit-learn__scikit-learn-13485-0e2f83caa7f20d880d047ac5fc5473b2c0e8ee81c933a98664ef6e9b3804a3bd","text":"Be more tolerant in checkarray for CalibratedClassifierCV For our package , the function is very convenient for Weakly Supervised Learners, as it can make PairsClassifier estimators return a probability for a pair of points to be labeled as similar or dissimilar, when those return a decision function. However, we currently cannot use it because our inputs can be 3D (example: ), and uses with default parameters that does not allow 3D inputs. However, other meta-estimators like do not call , so we can use them easily in metric-learn. Is the in really useful or could we do without it ? If we could, I'd be happy to provide a PR to do so"}
{"_id":"q-en-scikit-learn__scikit-learn-13496-10af1cdb8e062da8ffb4487d8de312b7f99cfb8c86cf152582de304dc184c95f","text":"Expose warmstart in Isolation forest It seems to me that supports incremental addition of new trees with the parameter of its parent class, . Even though this parameter is not exposed in , it gets inherited from and one can use it by changing it to after initialization. To make it work, you have to also increment on every iteration. It took me a while to notice that it actually works, and I had to inspect the source code of both and . Also, it looks to me that the behavior is in-line with that is behind e.g. . To make it more easier to use, I'd suggest to: expose in , default ; document it in the same way as it is documented for , i.e. say: add a test to make sure it works properly; possibly also mention in the \"IsolationForest example\" documentation entry;"}
{"_id":"q-en-scikit-learn__scikit-learn-13497-585ca4c614e13ed612b81fbfd0aee228586c62e5acf7c54052c19e82071305a7","text":"Comparing string to array in estimatemi In ` but discrete features can be an array of indices or a boolean mask. This will error in future versions of numpy. Also this means we never test this function with discrete features != 'auto', it seems?"}
{"_id":"q-en-scikit-learn__scikit-learn-13536-2a9cb6b724ba3f7c6c8b428429a673753c10f2bb9356ff6cb81d8cfd7970fd00","text":"improve error message when passing sampleweight to Pipeline Many estimators take a parameter named . does not, since it wants its parameters to be prefixed by the step name with a delimiter: This error message is not friendly enough. It should explicitly describe the correct format for passing to a step in a Pipeline."}
{"_id":"q-en-scikit-learn__scikit-learn-13549-551f0fc9455d0d035946831f7237ee8045ae1222fc9313a1989c051ed342b2d9","text":"Stratified subsampler utility? I have some data and that I want to subsample (i.e. only keep a subset of the samples) in a stratified way. Using something like is almost what I need. But that will error if: I happen to want exactly samples () I want something close to which is not enough to have a stratified test or train set () Basically what I need is a option to , that's probably the most appropriate place to introduce this."}
{"_id":"q-en-scikit-learn__scikit-learn-13554-27c46735d15953fff31de3d0ce161652659ce9f316619259f9721944b4745c43","text":"Numerical precision of euclideandistances with float32 <!-- Instructions For Filing a Bug: --I noticed that sklearn.metrics.pairwise.pairwisedistances function agrees with when using np.float64 arrays, but disagrees when using np.float32 arrays. See the code snippet below. I expect that the results from sklearn.metrics.pairwise.pairwisedistances would agree with for both 64-bit and 32-bit. In other words, I expect the following output: The code snippet above produces the following output for me: [WIP] Stable and fast float32 implementation of euclideandistances Superseds PR These commits implement a block-wise casting to float64 and uses the older code to compute the euclidean distance matrix on the blocks. This is done useing only a fixed amount of additional (temporary) memory. This code implements several optimizations: since the distance matrix is symmetric when , copy the blocks of the upper triangle to the lower triangle; compute the optimal block size that would use most of the allowed additional memory; cast blocks of to float64; precompute blocks of if not given so it gets reused through the iterations over ; swap and when is given, but not . Note that all the optimizations listed here have proven useful in a benchmark. The hardcoded amount of additional memory of 10MB is also derived from a benchmark. As a side bonus, this implementation should also support float16 out of the box, should scikit-learn support it at some point. Add a test for numeric precision (see ) Surprisingly bad precision, isn't it? Note that the traditional computation sqrt(sum((x-y)*2)) gets the results exact. <!-- Thanks for contributing a pull request! Please ensure you have taken a look at the contribution guidelines: --<!-- Example: See also . Please use keywords (e.g., Fixes) to create link to the issues or pull requests you resolved, so that they will automatically be closed when your pull request is merged. See --<!-- Please be aware that we are a loose team of volunteers so patience is necessary; assistance handling other issues is very welcome. We value all user contributions, no matter how minor they are. If we are slow to review, either the pull request needs some benchmarking, tinkering, convincing, etc. or more likely the reviewers are simply busy. In either case, we ask for your understanding during the review process. For more information, see our FAQ on this topic: http:\/\/scikit- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-13618-98ea708806500758e6b151e035af58f24b324b14356e8d1c0c3a88c2483409f1","text":"Allow setting of initial hyperparameters of BayesianRidge <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --<!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--I propose to be able to set initial values ​​of the hyperparameters (alpha, lambda) in BayesianRidge().fit. I tried to fit a sinusoidal curve with polynomials by Bayesian ridge regression, but the default initial values did not work well (left figure). So, I corrected the fit method of the BayesianRidge class so that I could set the initial value, and the regression worked well (right figure). ! <!-- Example: If the code is too long, feel free to put it in a public gist and link it in the issue: --The code I wrote is Here. <!-- Example: No error is thrown. Please paste or describe the expected results.--The right figure <!-- Please paste or specifically describe the actual output or traceback. --The left figure <!-- Please run the following snippet and paste the output below. For scikit-learn = 0.20: import sklearn; sklearn.showversions() For scikit-learn < 0.20: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"NumPy\", numpy.version) import scipy; print(\"SciPy\", scipy.version) import sklearn; print(\"Scikit-Learn\", sklearn.version) --System: python: 3.6.6 (default, Mar 8 2019, 18:24:30) [GCC 7.3.0] executable: \/home\/*\/.pyenv\/versions\/3.6.6\/bin\/python machine: Linux-4.15.0-47-generic-x8664-with-debian-buster-sid BLAS: macros: NOATLASINFO=1, HAVECBLAS=None libdirs: \/usr\/lib\/x8664-linux-gnu cblaslibs: cblas Python deps: pip: 19.0.3 setuptools: 41.0.0 sklearn: 0.20.3 numpy: 1.16.2 scipy: 1.2.1 Cython: 0.29.6 pandas: 0.24.2 <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-13620-cbdddff4ccfa372a4c83cd63908eb9a6a0f57ad5c40add42d21c17705945e4bf","text":"Bug in Gradient Boosting: Feature Importances do not sum to 1 I found conditions when Feature Importance values do not add up to 1 in ensemble tree methods, like Gradient Boosting Trees or AdaBoost Trees. This error occurs once the ensemble reaches a large number of estimators. The exact conditions depend variously. For example, the error shows up sooner with a smaller amount of training samples. Or, if the depth of the tree is large. When this error appears, the predicted value seems to have converged. But it’s unclear if the error is causing the predicted value not to change with more estimators. In fact, the feature importance sum goes lower and lower with more estimators thereafter. Consequently, it's questionable if the tree ensemble code is functioning as expected. Here's sample code to reproduce this: Output: In fact, if we examine the tree at each staged prediction, we'll see that the feature importance goes to 0 after we hit a certain number of estimators. (For the code above, it's 712.) Here's code to describe what I mean: Output: I wonder if we’re hitting some floating point calculation error. BTW, I've posted this issue on the mailing list . There aren't a lot of discussion, but others seem to think there's a bug here too. Hope we can get this fixed or clarified. Thank you! -Doug Windows-7;'Python', '2.7.9 ;'NumPy', '1.9.2';'SciPy', '0.15.1';'Scikit-Learn', '0.16.1'"}
{"_id":"q-en-scikit-learn__scikit-learn-13628-9871b2d300d6bf4e31bd0a4a13747b6e442aea235032edc0ae57f173680c96f1","text":"brierscoreloss error Hello. I think I found a mistake in brierscoreloss. When you have a target = 1 and a prediction = 1 too, brierscoreloss should be 0 (best result), but it gives 1. Why is it happening? Because checkbinaryprobabilisticpredictions gets target with only one class and convert it to 0. And metric calculates for target = 0 and prediction = 1. The same problem for target = 1 and prediction = 0. brierscoreloss is 0 (the best result), but it should be 1. Examples: Approx = [0, 0, 0, 0] Target = [1, 1, 1, 1] Weight = [1, 1, 1, 1] brierscoreloss(Target, Approx, sampleweight=Weight) result is 0 Approx = [1, 1, 1, 1] Target = [1, 1, 1, 1] Weight = [1, 1, 1, 1] brierscoreloss(Target, Approx, sampleweight=Weight) result is 1 Maybe we should fix it? Thank you."}
{"_id":"q-en-scikit-learn__scikit-learn-13641-4188ef2df79ee706a95743356389c9e2c36e47ac5587cd6cae02ceff2d7b3d75","text":"CountVectorizer with custom analyzer ignores input argument Example: Same for . Not sure if this should be fixed or just documented; I don't like changing the behavior of the vectorizers yet again..."}
{"_id":"q-en-scikit-learn__scikit-learn-13704-f9b5ad5a64cbadc3a143d078f47a59ceed2da18622d6216f14b333012068dcad","text":"VarianceThreshold doesn't remove feature with zero variance When calling VarianceThreshold().fittransform() on certain inputs, it fails to remove a column that has only one unique value. The Variance threshold should produce This issue arose when I was using VarianceThreshold on a real dataset (of which this is a subset). It appears to work correctly in other situations (for instance I can't reproduce this behaviour if I replace the first column with 1's). System python: 3.5.6 (default, Aug 26 2018, 16:30:03) [GCC 4.2.1 Compatible Clang 4.0.1 (tags\/RELEASE401\/final)] machine: Darwin-18.2.0-x8664-i386-64bit executable: \/anaconda3\/envs\/tensorflow\/bin\/python3 BLAS macros: NOATLASINFO=3, HAVECBLAS=None libdirs: cblaslibs: cblas Python deps setuptools: 40.2.0 numpy: 1.15.4 sklearn: 0.20.0 Cython: None scipy: 1.1.0 pandas: 0.24.0 pip: 19.0.1"}
{"_id":"q-en-scikit-learn__scikit-learn-13726-997e095b6eea1f77672d73af642f52078c0131ed15df024011881c613458dd68","text":"ncomponents kwarg missing in SpectralClustering The kwarg defined in the function allow the user to choose how many eigenvalues\/eigenvectors should be used in the classification. However this kwarg cannot be accessed\/modified when using class, and is set to default ( = ). Since the class calls the spectralclustering function, I guess the user should be able to access this kwarg ?"}
{"_id":"q-en-scikit-learn__scikit-learn-13779-d4ecdf652b2495cc9464785601157c4095bd8c622d6d4e2fe030179aed0e789d","text":"Voting estimator will fail at fit if weights are passed and an estimator is None Because we don't check for an estimator to be in support, is failing`."}
{"_id":"q-en-scikit-learn__scikit-learn-13780-f3316bf446464fd5cf9ad85cfd7d925277e3e7d1c7a54f776ba79c7e4de40d50","text":"Handle 'drop' together with None to drop estimator in VotingClassifier\/VotingRegressor As mentioned in the following , the and should accept to drop an estimator from the ensemble is the same way that is doing now."}
{"_id":"q-en-scikit-learn__scikit-learn-13828-3c42542f8383ba444d8f5d0114fd15f1fc61a218854d7f33f375771ace237b71","text":"sklearn.cluster.AffinityPropagation doesn't support sparse affinity matrix <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --sklearn.cluster.AffinityPropagation doesn't support sparse affinity matrix. <!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--A similar question is at . It focuses on default affinity. no error raised since it works for dense matrix. Traceback (most recent call last): File \"<stdin\", line 1, in <moduleFile \"D:\\Miniconda\\lib\\site-packages\\sklearn\\cluster\\\", line 381, in fit , verbose=self.verbose, returnniter=True) File \"D:\\Miniconda\\lib\\site-packages\\sklearn\\cluster\\\", line 115, in affinitypropagation preference = (S) File \"D:\\Miniconda\\lib\\site-packages\\numpy\\lib\\\", line 3336, in median overwriteinput=overwriteinput) File \"D:\\Miniconda\\lib\\site-packages\\numpy\\lib\\\", line 3250, in ureduce r = func(a, kwargs) File \"D:\\Miniconda\\lib\\site-packages\\numpy\\lib\\\", line 3395, in median return mean(part[indexer], axis=axis, out=out) File \"D:\\Miniconda\\lib\\site-packages\\numpy\\core\\\", line 2920, in mean out=out, kwargs) File \"D:\\Miniconda\\lib\\site-packages\\numpy\\core\\\", line 85, in mean ret = (ret \/ rcount) ValueError: setting an array element with a sequence. System: python: 3.6.7 (default, Oct 28 2018, 19:44:12) [MSC v.1915 64 bit (AMD64)] executable: D:\\Miniconda\\ machine: Windows-7-6.1.7601-SP1 BLAS: macros: SCIPYMKLH=None, HAVECBLAS=None libdirs: D:\/Miniconda\\Library\\lib cblaslibs: mkl_rt Python deps: pip: 18.1 setuptools: 40.6.2 sklearn: 0.20.1 numpy: 1.15.4 scipy: 1.1.0 Cython: None pandas: 0.23.4 <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-13864-0a45ad13edf09841967cf836b0f055f4aab90e8292e8a84dc9368ef5cbe0c168","text":"AttributeError thrown when calling metrics.pairwisedistances with binary metrics and Y is None thrown when calling with binary metrics if is . No error. Should return a of shape containing the pairwise distances. This worked correctly in sklearn version 0.20.3. I think the problem was introduced in there is now a code path where has its checked without any prior check as to whether is ."}
{"_id":"q-en-scikit-learn__scikit-learn-13877-1efa9755dcf7bdd890456e362906b804162689de4816499badc78544905a5cea","text":"pairwisedistances returns zeros for metric cosine when executed in parallel returns a list of zeros when calculating with equal to -1 or greater than 2. Using calculates the expected results. Using the metric returns non-zero results, but the values seem to be integers instead of floats. I executed with different values for and . The outputs were as follows:"}
{"_id":"q-en-scikit-learn__scikit-learn-13910-72fc34f8480d023da860d2cc1da597367ea6850608cb49a93ff732843711e74e","text":"Untreated overflow (?) for float32 in euclideandistances new in sklearn 21.1 I am using euclidean distances in a project and after updating, the result is wrong for just one of several datasets. When comparing it to one can see that in version 21.1 it behaves substantially different to 20.3. The matrix is an ndarray with size (100,) with float32. The data are in this zip Can be found when using sklearn 20.3, both behave identical. When using version 21.1 has many 0 entries and some unreasonably high entries Sklearn 21 System: python: 3.6.7 (default, Oct 22 2018, 11:32:17) [GCC 8.2.0] executable: \/home\/lenz\/PycharmProjects\/pyrolmm\/venvsklearn21\/bin\/python3 machine: Linux-4.15.0-50-generic-x8664-with-Ubuntu-18.04-bionic BLAS: macros: HAVECBLAS=None, NOATLASINFO=-1 libdirs: \/usr\/lib\/x8664-linux-gnu cblaslibs: cblas Python deps: pip: 9.0.1 setuptools: 39.0.1 sklearn: 0.21.1 numpy: 1.16.3 scipy: 1.3.0 Cython: None pandas: None For sklearn 20.3 the versions are: System: python: 3.6.7 (default, Oct 22 2018, 11:32:17) [GCC 8.2.0] executable: \/home\/lenz\/PycharmProjects\/pyrolmm\/venvsklearn20\/bin\/python3 machine: Linux-4.15.0-50-generic-x8664-with-Ubuntu-18.04-bionic BLAS: macros: HAVECBLAS=None, NOATLASINFO=-1 libdirs: \/usr\/lib\/x8664-linux-gnu cblas_libs: cblas Python deps: pip: 9.0.1 setuptools: 39.0.1 sklearn: 0.20.3 numpy: 1.16.3 scipy: 1.3.0 Cython: None pandas: None <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-13933-fe8c7fc8701c5f834cc96bbab9ac46fee7c39940060497c1eb956d653d288438","text":"GBDTs should bin train and validation data separately? In the new GBDTs we bin the data before calling (for early-stopping). That means that the validation set is also used to find the bin thresholds (it is of course not used to find the split points!). I feel like the \"data leak\" is very minimal, but it seems more correct to bin Xtrain and Xval separately. WDYT?"}
{"_id":"q-en-scikit-learn__scikit-learn-13960-55656da2887a68f6d5aa36a5b379383acd2e4df84074e08f7a4bfc2781216d5a","text":"IncrementalPCA should accept sparse input <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --is by design suited to application to sparse data in a way that most PCA classes are not. However, it is not written to accept this by default. No error should be thrown. I'd be happy to submit this as a PR if it's desirable. <details<\/details<!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-13974-fcf5ad942a01371b15ffb613e6249c2e95cac1ab0502158c62b064716551a7c9","text":"Design of addindicator in SimpleImputer may fail when running cross validation <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --<!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--The design of depends on missing values exist in the training data. This will break cross validation. <!-- Example: If the code is too long, feel free to put it in a public gist and link it in the issue: --<!-- Example: No error is thrown. Please paste or describe the expected results.--No error is thrown. <!-- Please paste or specifically describe the actual output or traceback. --The original design was adopted because, if the training data has no missing value, there will be a column with all zeros. This type of error will appear when we try to do grid search over the parameter. One way to work around this is to split the data in such a way that missing values are available (for each column that has a missing value) in both the training set and test set. <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-14024-ca83521153002ef6c435fe4e460ebd83d51e8208c15e6223c0ce4e3e1eaf13db","text":"Zero division error in HistGradientBoosting NaN This dataset is a bit weird in that it has 100 classes with 16 samples each. The default parameter don't work very well but we should fail more gacefully. cc"}
{"_id":"q-en-scikit-learn__scikit-learn-14053-77994308b6531843845a243f8996d4e5a4bf790f42ec7ad332ee98179acfd4c4","text":"IndexError: list index out of range in exporttext when the tree only has one feature <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --returns when there is single feature. <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-14067-3c4bb17cf6ac2caf0fc39380853c1e3f535d3107f9552c81f4c8df4cf3fc89d6","text":"ARD Regressor accuracy degrades when upgrading Scipy 1.2.1 -1.3.0 Hi, bit of a tricky one, I'm hoping someone will have some time and\/or suggestions for further investigation! There seems to be an often-occurring worsening of performance (i.e. accuracy, although run-time increases too!) from the ARD regressor when upgrading from Scipy 1.2.1 -1.3.0. On a very simple dataset (see code snippets below) where a near-perfect fit should be achievable, typical error seems to degrade from order 1E-5 to 1E-2. Notably, convergence iterations seem to increase also from ~a few (~5) to around 50-200 iterations. Here's the headline plot, plotting absolute co-efficient error when fit across 1000 datasets generated with different random seeds: ! Note how with Scipy==1.2.1, errors are largely constrained to <0.01, while with Scipy==1.3.0 they range up to 0.05 (and in a few rare cases the algorithm produces garbage results, see later). I guess this could be (probably is?) a Scipy rather than Sklearn issue, but probably the only way to confirm \/ isolate that would be to start here. It's also possible that this worsening of behaviour is a weirdness of my particular toy example, but the difference in behaviour seems large and unexpected enough to warrant further investigation, I'd hope! OK, so here's a short snippet on just a single seed if you're curious to try this yourself. I'm generating three vectors of normally distributed values, 250 samples. Then the target is just a perfect copy of one of those vectors (index=1). We measure the accuracy of the fit by simply checking how close that coefficient is to 1.0 (the other coefficients always shrink to 0., as you'd hope): Scipy 1.2.1: Scipy 1.3.0 It could be that there's some oddity of the random data from a single seed, so I set up some short scripts to first generate a static collection of 1000 of the datasets as seen above, then collate the results from both versions of scipy. The snippets are as follows: Make data: Test sklearn: Plot results: A little investigating summary statistics of those datasets in notebook gives the following points of comparison:"}
{"_id":"q-en-scikit-learn__scikit-learn-14087-9e81667a69c13331de50639119353c5963b597429ea2a56a77a309765237ebea","text":"IndexError thrown with LogisticRegressionCV and refit=False The following error is thrown when trying to estimate a regularization parameter via cross-validation, without refitting. No error is thrown."}
{"_id":"q-en-scikit-learn__scikit-learn-14092-7d348000d91fdf13e028aa399c664a01d7d985ce88cc00a242f7328f8f404a96","text":"NCA fails in GridSearch due to too strict parameter checks NCA checks its parameters to have a specific type, which can easily fail in a GridSearch due to how param grid is made. Here is an example: The issue is that for : 1 is not a float, and for : np.int64 is not int Before proposing a fix for this specific situation, I'd like to have your general opinion about parameter checking. I like this idea of common parameter checking tool introduced with the NCA PR. What do you think about extending it across the code-base (or at least for new or recent estimators) ? Currently parameter checking is not always done or often partially done, and is quite redundant. For instance, here is the input validation of lda: most params aren't checked and for those who are there's a lot of duplicated code. A propose to be upgrade the new tool to be able to check open\/closed intervals (currently only closed) and list membership. The api would be something like that: where validoptions would be a dict of . e.g for the param of , it can be either a float or a string in a list, which would give Sometimes a parameter can only be positive or within a given interval, e.g. of must be between 0 and 1, which would give positivity of e.g. would be ."}
{"_id":"q-en-scikit-learn__scikit-learn-14114-e5d78d9719f8a16f5d4d459a88617315a51572e528eb1c7ef487865d7cfe5374","text":"AdaBoost's \"SAMME\" algorithm uses 'predict' while fitting and 'predictproba' while predicting probas Subj. This seems to me to be a wrong approach, moreover this drives to such mistakes: <preAdaBoostClassifier(algorithm=\"SAMME\", baseestimator=SVC()).fit(trainX, trainY).predictproba(testX) NotImplementedError Traceback (most recent call last) <ipython-input-108-1d666912dadain <module() ----1 AdaBoostClassifier(algorithm=\"SAMME\", baseestimator=SVC()).fit(trainX, trainY).predictproba(testX) \/Library\/Python\/2.7\/site- in predictproba(self, X) 716 proba = sum(estimator.predictproba(X) w 717 for estimator, w in zip(self.estimators, --718 self.estimatorweights)) 719 720 proba \/= () \/Library\/Python\/2.7\/site- in <genexpr((estimator, w)) 715 else: # self.algorithm == \"SAMME\" 716 proba = sum(estimator.predictproba(X) w --717 for estimator, w in zip(self.estimators, 718 self.estimatorweights)) 719 \/Library\/Python\/2.7\/site- in predictproba(self, X) 493 if not self.probability: 494 raise NotImplementedError( --495 \"probability estimates must be enabled to use this method\") 496 497 if not in ('csvc', 'nusvc'): NotImplementedError: probability estimates must be enabled to use this method <\/pre"}
{"_id":"q-en-scikit-learn__scikit-learn-14125-889a8e20227e233eb00af12e9adf49024f0382e01be767825be6fe182e217eda","text":"[MRG] Fix 'SparseSeries deprecated: scipy-dev failing on travis' <!-- Thanks for contributing a pull request! Please ensure you have taken a look at the contribution guidelines: -- Issue: SparseSeries deprecated: scipy-dev failing on travis <!-- Example: See also . Please use keywords (e.g., Fixes) to create link to the issues or pull requests you resolved, so that they will automatically be closed when your pull request is merged. See --Use a Series with sparse values instead instead of . <!-- Please be aware that we are a loose team of volunteers so patience is necessary; assistance handling other issues is very welcome. We value all user contributions, no matter how minor they are. If we are slow to review, either the pull request needs some benchmarking, tinkering, convincing, etc. or more likely the reviewers are simply busy. In either case, we ask for your understanding during the review process. For more information, see our FAQ on this topic: http:\/\/scikit- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-14141-775b58d9bee62cc1198d69fca17a30dc76ef5b7c6140b76bcd5daa24d58cd014","text":"Add joblib in showversions joblib should be to the dependencies listed in showversions or to the issue template when sklearn version is 0.20."}
{"_id":"q-en-scikit-learn__scikit-learn-14237-8051a1dcea1056f5a8a4475114a2efa8e1202a1dc8e519d251ba06c52ec119fd","text":"Named col indexing fails with ColumnTransformer remainder on changing DataFrame column ordering I am using ColumnTransformer to prepare (impute etc.) heterogenous data. I use a DataFrame to have more control on the different (types of) columns by their name. I had some really cryptic problems when downstream transformers complained of data of the wrong type, while the ColumnTransformer should have divided them up properly. I found out that ColumnTransformer silently passes the wrong columns along as when specifying columns by name, using the option, and using DataFrames where column ordering can differ between and In this case, the wrong columns are passed on to the downstream transformers, as the example demonstrates: When using ColumnTransformer with a DataFrame and specifying columns by name, should reference the same columns when fitting and when transforming (['b'] in above example), regardless of the column positions in the data during fitting and transforming. appears to, during fitting, remember remaining named DataFrame columns by their numeric index (not by their names), which (silently) leads to the wrong columns being handled downstream if the transformed DataFrame's column ordering differs from that of the fitted DataFrame. Position in module where the index is determined: My current workaround is to not use the option but specify all columns by name explicitly. System: python: 3.7.3 (default, Mar 30 2019, 03:44:34) [Clang 9.1.0 (clang-902.0.39.2)] executable: machine: Darwin-17.7.0-x8664-i386-64bit BLAS: macros: NOATLASINFO=3, HAVECBLAS=None libdirs: cblaslibs: cblas Python deps: pip: 19.1.1 setuptools: 41.0.1 sklearn: 0.21.2 numpy: 1.16.4 scipy: 1.3.0 Cython: None pandas: 0.24.2"}
{"_id":"q-en-scikit-learn__scikit-learn-14309-0a81cf1af6452c2d16416371b18c5913033b05f12ff6971a34c811317885fe1e","text":"plotpartialdependence() fails when used on DecisionTreeRegressor <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --fails when using a as the estimator. The problem appears to be related to the presence of a attribute (with a value of ) on the estimator, despite it being a regressor and not a classifier. Deleting the attribute from the estimator allows to successfully run. <!-- Example: If the code is too long, feel free to put it in a public gist and link it in the issue: --No error is thrown. <!-- Example: No error is thrown. Please paste or describe the expected results.--<!-- Please paste or specifically describe the actual output or traceback. --A is thrown: <!-- Please run the following snippet and paste the output below. For scikit-learn = 0.20: import sklearn; sklearn.show_versions() For scikit-learn < 0.20: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"NumPy\", numpy.version) import scipy; print(\"SciPy\", scipy.version) import sklearn; print(\"Scikit-Learn\", sklearn.version) --<!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-14430-9b313bfc7c572de7e577ed66a1616f4f38de8b2e17c748933a78f6ecccf7168f","text":"Pickling Tokenizers fails due to use of lambdas Cannot pickle a using the builtin python module, likely due to the use of lambdas in Example: Program runs without error Traceback: Instead of the builtin , use , which can capture the expression. <!-- Please run the following snippet and paste the output below. For scikit-learn = 0.20: import sklearn; sklearn.showversions() For scikit-learn < 0.20: import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"NumPy\", numpy.version) import scipy; print(\"SciPy\", scipy.version) import sklearn; print(\"Scikit-Learn\", sklearn.version) --Version information: I think this is similar to issues: (looking at the stackoverflow thread at , it suggests using which also seems to work for the toy example) Naively, I would make one of the two changes below, but I am not familiar with the scikit-learn codebase, so they might not be appropriate: the FAQ to direct people to other serialization libraries (perhaps I missed this recommendation?), e.g. at or the use of the lambdas in the vectorizer and replace them with locally def'd functions. I suspect that this solution is flawed because it doesn't account for other uses of lambdas elsewhere in the codebase, and the only complete solution would be to stop using lambdas, but these are a useful language feature."}
{"_id":"q-en-scikit-learn__scikit-learn-14450-ae9a879a8c4c0fee982d2159d19cfbb1d8f98cf17c732d2af0260c46cb72fd4d","text":"PLS reports \"array must not contain nan\" if a feature is constant Originally reported at by Reproduce with: The obscure error message is due to the presence of a column containing only 0."}
{"_id":"q-en-scikit-learn__scikit-learn-14458-ac02d72b7526b862e55549e0a5bbc7e9f9010736a827fdb1dd345d1f4916280b","text":"Need for conversion with SAG Running the following code snippet will lead to an error: I am wondering if we are not to strict here, and we should include a since that will expect a C-contiguous numpy array."}
{"_id":"q-en-scikit-learn__scikit-learn-14464-b18ba857292928707ed57fa18f4578595b5aecfc73833da52485c8633d96acee","text":"Cloning custom transform replaces values in init dictionary <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --<!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--Let us say we have a custom transform that has some arguments. When the is instantiated, these arguments are set in the init. When we clone (as happens in , for example), the arguments get copied successfully. However, if the arguments are sent to a structure such as a dictionary, the clone replaces them with None. In cases where None does not cause errors, this creates a silent error, as the cloned version of will run, producing different results from its original version (which is how I ran into this problem in the first place). Fully replicable example follows. <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-14496-083ff9cbfe1a96b446e322c1e86c97a159ebb18d002af10fa8ec44e5d41ff415","text":"[BUG] Optics float minsamples NN instantiation None yet. When passing minsamples as a float to optics l439 & 440 execute to bring it into integer ranges, but don't convert to int: When instantiating the NearestNeighbours class with a float it raises due to the float (l448). Error message: Fix: the int(...) is for backwards compatibbility to Python 2 where with T Number, while Python3 <!-- Please be aware that we are a loose team of volunteers so patience is necessary; assistance handling other issues is very welcome. We value all user contributions, no matter how minor they are. If we are slow to review, either the pull request needs some benchmarking, tinkering, convincing, etc. or more likely the reviewers are simply busy. In either case, we ask for your understanding during the review process. For more information, see our FAQ on this topic: http:\/\/scikit- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-14520-8c08a213711e9751d8d80ab8457b1cca39d6ce48551eccc32eaed4fa18ab227e","text":"Copy param ignored in TfidfVectorizer I was playing with vectorizers and I found this: However that parameter is not used later in the method. Here is used: Is there anything I am missing?"}
{"_id":"q-en-scikit-learn__scikit-learn-14544-bc1d22f5808c240b45899caf0bc03885d885c1f7ed5ad67df76f26e805774da1","text":"RFC ColumnTransformer input validation and requirements There have been some issues around ColumnTransformer input requirements that I think we might want to discuss more explicitly. Examples are an actual bug when changing columns: and how to define number of input features . Related is also the idea of checking for column name consistency: Mainly I'm asking whether it makes sense to have the same requirements for ColumnTransformer as for other estimators. ColumnTransformer is the only estimator that addresses columns by name, and so reordering columns doesn't impact the model and actually everything works fine. Should we still aim for checking for reordering of columns? Right now, , which makes sense, since it only cares about the columns it is using. Should we allow that going forward as long as remainder is not used? (if remainder is used, the result of would have a different shape and downstream estimators would break, so I think we shouldn't support it in this case). In either case, what's ` or the number of columns actually used? Does it even make sense to define it if we allow adding other columns that are ignored?"}
{"_id":"q-en-scikit-learn__scikit-learn-14591-90d1e8de2ecd1a11f307ca9905014ecc1de1efc282bf5b055c7036a26ffca54d","text":"LassoCV always sets precompute to False before fitting the chosen alpha value I'm using a very large data-set. After fitting 100 x 3-fold cross-validated LASSOs at lightning speed (a few seconds total), LassoCV stalls at the final hurdle: fitting a LASSO with the chosen alpha value to the whole data-set (waiting over half an hour - it should only take approximately 50% longer than a single fold...). After a lot of head-scratching I found the reason why. In 's () just before calling the final () (line 1223 in Python2.7\/sklearn0.19.0), there is the rather inconspicuous line So even if you've specified precompute as True when calling LassoCV, it is ignored. Why is this? It's making the computation impractically slow (should it even be this slow without precompute?) - literally just commenting the line out makes the fit instantaneous. Am I missing something here mathematically - I can't see anything wrong with using a precomputed Gram matrix for the final fit when it was used for all of the cross-validation fits? The implementation seems to imply it should be used for performance whenever numsamples num_features. Why hard set it to False?"}
{"_id":"q-en-scikit-learn__scikit-learn-14629-76057838f36caa750675ae80519d5f8531c10c14f29d3d72cf6b451ae20a0281","text":"AttributeError with crossvalpredict(method='predictproba') when using MultiOuputClassifier I believe there is a bug when using with a . I think the problem is in the use of here: To obtain the attribute of a , you need instead. If core team members have any idea of how to address this, I am happy to submit a patch. Array with prediction probabilities. System: python: 3.6.8 (default, Feb 21 2019, 18:30:04) [MSC v.1916 64 bit (AMD64)] executable: C:\\Users\\nak142\\Miniconda3\\envs\\myo\\ machine: Windows-10-10.0.-SP0 BLAS: macros: libdirs: cblas_libs: cblas Python deps: pip: 19.1.1 setuptools: 41.0.1 sklearn: 0.21.2 numpy: 1.16.4 scipy: 1.2.1 Cython: 0.29.12 pandas: 0.24.2"}
{"_id":"q-en-scikit-learn__scikit-learn-14704-2a1ab24630a33e5e680e1a48636405cc07742c97743b7c94a704e58a287ebd8c","text":"StratifiedKFold makes fold-sizes very unequal I found this when trying to write tests for . Right now, `: We could achieve something like but our rounding doesn't let us :-\/"}
{"_id":"q-en-scikit-learn__scikit-learn-14706-22739ebdcb9b98695bab5454687d4a0da80a42d1afca71ce620f9afabdc743e6","text":"kernelapproximation.Nystroem does not support precomputed kernel The documentation says that precomputed kernels are supported in Nystroem, but in reality it does not seem to be the case: https:\/\/scikit- Example code: Leads to"}
{"_id":"q-en-scikit-learn__scikit-learn-14710-0d074fc7138221486edb98782ebd850f9ba034d44fbfc3b4ced38f61516f97aa","text":"HistGradientBoostingClassifier does not work with string target when early stopping turned on <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --<!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--The scorer used under the hood during early stopping is provided with being integer while are original classes (i.e. string). We need to encode each time that we want to compute the score. <!-- Example: If the code is too long, feel free to put it in a public gist and link it in the issue: --No error is thrown <!-- Please paste or specifically describe the actual output or traceback. --Maybe one solution would be to do:"}
{"_id":"q-en-scikit-learn__scikit-learn-14732-70198089dc2239107645a704889334ac18ba84fbe14889f0aec1b6ba0a80bf68","text":"medianabsoluteerror multioutput Multioutput is not currently supported in . Is this a design choice or has it just not been implemented yet? In case of the latter, I am happy to submit a PR."}
{"_id":"q-en-scikit-learn__scikit-learn-14764-d521fcab4467aa9c0015d27a78c4efc65259e9251f64ec6aca3ec8515fc451ed","text":"datasets :: makeclassification() weights parameter should be a sequence (not just a list). If there is a pertinent reason that must be a list, while all other iterable parameters are arrays, then it should be mentioned in the docstring. Otherwise, the docstring should be amended as in , e.g. \"weights : list of floats or None (default=None)\" -\"weights : sequence of floats or None (default=None)\", along with amended lines 165 and 171 (see Corrections). Should not fail: as valid as a list: #wimlds [MRG] Added an example to the …or class <!-- Thanks for contributing a pull request! Please ensure you have taken a look at the contribution guidelines: --<!-- Contributes to . --I an example to the () <!-- Please be aware that we are a loose team of volunteers so patience is necessary; assistance handling other issues is very welcome. We value all user contributions, no matter how minor they are. If we are slow to review, either the pull request needs some benchmarking, tinkering, convincing, etc. or more likely the reviewers are simply busy. In either case, we ask for your understanding during the review process. For more information, see our FAQ on this topic: http:\/\/scikit- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-14806-a54ee8039cb9585f245793100253d8f4b8caf1a1549b2ed78255f4773d21bd15","text":"IterativeImputer behaviour on missing nan's in fit data Why is this behaviour forced: Features with missing values during transform which did not have any missing values during fit will be imputed with the initial imputation method only. This means by default it will return the mean of that feature. I would prefer just fit one iteration of the chosen estimator and use that fitted estimator to impute missing values. Actual behaviour: Example - The second feature missing --mean imputation Example adjusted - Second feature has values --iterative imputation with estimator Maybe line 679 to 683 should be optional with a parameter like force-iterimpute."}
{"_id":"q-en-scikit-learn__scikit-learn-14869-c7533df8ac491ae7c2eb8e77bb47162a835fa853ec66f4b0517caee460970e2e","text":"HGBC with categoricalcrossentropy fails silently on binary classification gives: And works fine. should either generalize or raise an error on binary classification. Ping"}
{"_id":"q-en-scikit-learn__scikit-learn-14890-65e0e404015cec8b9867346c2d6641af18528137969892e11c3fb992bb0e14c4","text":"Fitting TransformedTargetRegressor with sampleweight in Pipeline Can't fit a using . May be link to ? Example: ! Fitting with <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-14894-04f68579d285a67652a2f0953821b7b730a10afefbb17b1f46faa32148e30e3b","text":"ZeroDivisionError in sparsefit for SVM with empty supportvectors When using sparse data, in the case where the supportvectors attribute is be empty, fitsparse gives a ZeroDivisionError No error is thrown and"}
{"_id":"q-en-scikit-learn__scikit-learn-14898-8e5ec504598a92368fa1c2229fc55e53f5047efd51f3c9c2d9de001c92ea6074","text":"Documentation section 3.3.1.1 has incorrect description of brierscoreloss In the documentation, section 3.3.1.1. \"Common cases: predefined values\" includes the remark As far as I can tell, this is true for all of the listed metrics, except the . In the case of , a lower loss value is better. This is because measures the mean-square difference between a predicted probability and a categorical outcome; the Brier score is minimized at 0.0 because all summands are either or when the model is making perfect predictions. On the other hand, the Brier score is maximized at 1.0 when all predictions are opposite the correct label, as all summands are either or . Therefore, the definition of the is not consistent with the quotation from section 3.3.1.1. I suggest making 2 changes to relieve this confusion. a function which simply negates the value of ; this is a direct analogy to what is done in the case of . A better model has a lower value of log-loss (categorical cross-entropy loss), therefore a larger value of the negative log-loss implies a better model. Naturally, the same is true for Brier score, where it is also the case that a better model is assigned a lower loss. reference to from section 3.3.1.1. Brier score is useful in lots of ways; however, because it does not have the property that a larger value implies a better model, it seems confusing to mention it in the context of section 3.3.1.1. References to can be replaced with , which has the property that better models have large values, just like accuracy, ROC AUC and the rest of the listed metrics."}
{"_id":"q-en-scikit-learn__scikit-learn-14908-1355daad0d5e2c3a2a4d42820f8a0168d6676a156bb8894afb6feeac67ec4db8","text":"Error could be improved with DummyClassifier constant strategy when constant value not in training data Error: We could add in the error message what constant value was provided (in this case ) and what the possible values are. This could be something like this (improvement more than welcome): Context: this was seen during the EuroScipy tutorial. The adult census dataset classes has a space in it at the beginning and the provided value did not have the space. Putting what the provided value was and what were the possible values would have helped the user fixing the problem."}
{"_id":"q-en-scikit-learn__scikit-learn-14983-f8c317685c9f07f6cc62e9d6e2e1524ad7d771dec51b529b30c7e4ae3f156798","text":"RepeatedKFold and RepeatedStratifiedKFold do not show correct repr string and do not show correct \\repr\\ string."}
{"_id":"q-en-scikit-learn__scikit-learn-14999-344601e9d11807831b7d051edf5b3e3092b64928eb4888c6f9350b8bb17fa971","text":"data leak in GBDT due to warm start (This is about the non-histogram-based version of GBDTs) X is split into train and validation data with . As noted, in a warm starting context, this will produce a leak if If is a instance: some samples that were used for training in a previous might be used for validation now."}
{"_id":"q-en-scikit-learn__scikit-learn-15028-dbe139bf2039801e02af74761fba98c5c9714beda3b2e96baf70c8b294df4d8e","text":"Deprecate classes attribute in DecisionTreeRegressor This partially relates to issue Currently, if you fit a decision tree regressor, and call the attribute , it will return none. This attribute does not appear on the doc string and shouldn't. This was surfaced from an issue related to mismatch attributes ( ) Reviewed the and worked with on different options including using a decorator on a property but it triggered the deprecation message when calling fit which was bad. In this PR, the was changed to in the parent. And a test was to the"}
{"_id":"q-en-scikit-learn__scikit-learn-15086-d00e5eafc6316c24e67793531169ade85dda077c7dbed55b325d34c194429fc2","text":"MultiTaskLassoCV with fitintercept=True returns wrong results There is something wrong with and binary features. It always returns the same mse for all the alphas and hence chooses a huge regularization zeroing out all coefficients. The same holds for too. However, this doesn't happen with . Moreover it doesn't happen if I set , or if I generate random normal features. I am working on anaconda, windows system, with python 3.7.1 and with scikit-learn v0.21.3, numpy v1.16.2. Consider the following code: It returns On the other hand, if I generate normal features X, then things are good: which returns: Also weirdly if I set , then things are good even with binary features: which returns"}
{"_id":"q-en-scikit-learn__scikit-learn-15094-b3a7a7ee6374aca14dd94f28ef5e756cb23a43064d4f222bd783af9dbc0e1133","text":"MaxAbsScaler Upcasts Pandas to float64 <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --I am working with the Column transformer, and for memory issues, am trying to produce a float32 sparse matrix. Unfortunately, regardless of pandas input type, the output is always float64. I've identified one of the Pipeline scalers, MaxAbsScaler, as being the culprit. Other preprocessing functions, such as OneHotEncoder, have an optional argument. This argument does not exist in MaxAbsScaler (among others). It appears that the upcasting happens when is executed. Is it possible to specify a dtype? Or is there a commonly accepted practice to do so from the Column Transformer? Thank you! Example: Darwin-18.7.0-x8664-i386-64bit Python 3.6.7 (default, Jul 2 2019, 02:07:37) [GCC 4.2.1 Compatible Clang 4.0.1 (tags\/RELEASE_401\/final)] NumPy 1.17.1 SciPy 1.3.1 Scikit-Learn 0.20.3 Pandas 0.25.1"}
{"_id":"q-en-scikit-learn__scikit-learn-15096-cbae8baf29aaceb680a6bb50af227d1477f414a41c19d445cf7663040d904a1e","text":"GridSearchCV saves all fitted estimator in cvresults['params'] when params are estimators I use GridSearchCV to optimize the hyperparameters of a pipeline. I set the param grid by inputing transformers or estimators at different steps of the pipeline, following the Pipeline documentation: I couldn't figure why dumping cvresults would take so much memory on disk. It happens that cvresults['params'] and all cvresults['param*'] objects contains fitted estimators, as much as there are points on my grid. This bug should happen only when njobs = 1 (which is my usecase). I don't think this is intended (else the arguments and attributes refit and best\\estimator wouldn't be used). My guess is that during the grid search, those estimator's aren't cloned before use (which could be a problem if using the same grid search several times, because estimators passed in the next grid would be fitted...)."}
{"_id":"q-en-scikit-learn__scikit-learn-15100-9e3186e18cff6ec26213a4fa9e169a847052c15d95f3743f0297b42bb19969db","text":"stripaccentsunicode fails to strip accents from strings that are already in NFKD form <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --<!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--The feature of and related does not work as expected when it processes strings that contain accents, if those strings are already in NFKD form. and should both be normalized to the same string, . is not changed, because does nothing if the string is already in NFKD form."}
{"_id":"q-en-scikit-learn__scikit-learn-15119-becae7a912d19d2b725cc41fbbd7a0f7a4c9ebe3ba0a2516514fdf736813d75c","text":"Inconsistent fit + transform and fittransform for FeatureUnion Is there a reason why the method signature accepts but neither nor do? It seems to go against the pattern that is the same as calling ? I see there's been discussion on supporting but it's not clear if this is resolved. My case is I'm trying to migrage code I wrote a while back where I used a Pipeline and each of my transformers adds columns to a dataframe, to a FeatureUnion where each transform only returns the new columns. One of my transforms takes a third data set in addition to X and y which is used as the transform. I guess as a workaround I'll make it a param of the transform rather than a fitarg."}
{"_id":"q-en-scikit-learn__scikit-learn-15120-78e360ae1083f57a8a6eecde5617346f851f59d5fbd0f7d7ba0f1590c29f8d39","text":"meanshift and MeanShift don't have the same API I'm trying to make call (related to ) but has a parameter and uses the default, so I cannot preserve backward compatibility without adding to . Should I just do that?"}
{"_id":"q-en-scikit-learn__scikit-learn-15138-e9b01bd13e075388330325bc81d49a2312262feb262665f1bc9e0de66d4d3a4e","text":"Stacking: add an option to use the original dataset when training finalestimator I think it will be readonable to add an option to use the original dataset when training finalestimator. This seems reasonable and has proved to be useful in some Kaggle competitions. Reference: implementation from mlxtend usefeaturesinsecondary : bool (default: False) If True, the meta-classifier will be trained both on the predictions of the original classifiers and the original dataset. If False, the meta-classifier will be trained only on the predictions of the original classifiers."}
{"_id":"q-en-scikit-learn__scikit-learn-15393-027989f0d809ace596458ec3c5556f2d99996d8844072d1e721be52b822cc56c","text":"imputationorder \"ascending\" and \"descending\" are inverted in IterativeImputer ImputerTriplets in fitted imputationsequence lists are appended according to imputationorder, but order is inverted"}
{"_id":"q-en-scikit-learn__scikit-learn-15495-f57106f30ee7b31e0c9f4cedd0898977d7589be4f831ef0e0681f7d945bda121","text":"Use checksampleweight to consistently validate sampleweight We recently introduced which returns a validated array. We should use it consistently throughout the code base, instead of relying on custom and adhoc checks like or (which are now handled by ). Here's a list of the estimators\/functions that could make use of it (mostly in or ): [x] CalibratedClassifierCV [x] DBSCAN [x] DummyClassifier [x] DummyRegressor [x] BaseBagging [x] BaseForest [x] BaseGradientBoosting [x] IsotonicRegression [x] KernelRidge [x] GaussianNB [x] BaseDiscreteNB [x] KernelDensity [x] BaseDecisionTree (I left-out the linearmodel module because it seems more involved there) Could be a decent sprint issue ? To know where a given class is defined, use e.g."}
{"_id":"q-en-scikit-learn__scikit-learn-15512-5f1e3303fbfbf26c571dd53c7a54c46fe86655960bdfa826f3054a901e165cce","text":"Return values of non converged affinity propagation clustering The affinity propagation Documentation states: \"When the algorithm does not converge, it returns an empty array as clustercenterindices and -1 as label for each training sample.\" Example: I would expect that the clustering here (which does not converge) prints first an empty List and then [-1,-1,-1], however, I get [2] as cluster center and [0,0,0] as cluster labels. The only way I currently know if the clustering fails is if I use the verbose option, however that is very unhandy. A hacky solution is to check if maxiter == niter but it could have converged exactly 15 iterations before maxiter (although unlikely). I am not sure if this is intended behavior and the documentation is wrong? For my use-case within a bigger script, I would prefer to get back -1 values or have a property to check if it has converged, as otherwise, a user might not be aware that the clustering never converged. System: python: 3.6.7 (default, Nov 21 2018, 02:32:25) [GCC 4.8.2 (Red Hat 4.8.2-15)] executable: \/home\/jenniferh\/Programs\/anaconda3\/envs\/TFRDKit119\/bin\/python machine: Linux-4.15.0-52-generic-x8664-with-debian-stretch-sid BLAS: macros: SCIPYMKLH=None, HAVECBLAS=None libdirs: \/home\/jenniferh\/Programs\/anaconda3\/envs\/TFRDKit119\/lib cblaslibs: mkl_rt, pthread Python deps: pip: 18.1 setuptools: 40.6.3 sklearn: 0.20.3 numpy: 1.15.4 scipy: 1.2.0 Cython: 0.29.2 pandas: 0.23.4"}
{"_id":"q-en-scikit-learn__scikit-learn-15524-f0f9df2dd9a15d925c7a8ae5148b870743bd884679f6fc27c7915bd26c5cf6bc","text":"Nested Cross Validation for precomputed KNN A nested cross validation prediction using a knn with precomputed metric raised an error Should return the predictions made by the model as the following code produces: sklearn 0.20.2"}
{"_id":"q-en-scikit-learn__scikit-learn-15535-3ec3cca36b44b4269592dcbca2dd3d6c61132bcaa49c9a802dba8f9e7bd863e1","text":"regression in input validation of clustering metrics ValueError: could not convert string to float: 'b' while works with a warning? this worked in 0.21.1 without a warning (as I think it should) Edit by I removed the in the second code snippet."}
{"_id":"q-en-scikit-learn__scikit-learn-15625-0d2978d081966cb56c8b5779526d71a369b9fd15eabb1e56bde1682ed31e6190","text":"ENH: add normalize parameter to metrics.classification.confusionmatrix Allows to get a normalized confusion matrix directly from the function call. I use frequently and find the need to always normalize the matrix manually maybe unnecessary. I am aware of the fact that other functions like already have this exact functionality implemented, so probably the lack of the parameter is intentional and I'm missing the why. But in case its not intentional you might find this contribution useful :)."}
{"_id":"q-en-scikit-learn__scikit-learn-19664-840df73cf51e7bc13ac803e594e4ff9c69083ebd13b5e5a6158e9a4ff8cc1a7d","text":"LabelPropagation raises TypeError: A sparse matrix was passed LabelPropagation (and LabelSpreading) error out for sparse matrices. Sparse case should work as does the dense one. Changing in line 224 to seems to fix the problem for me (BTW: a similar check accepting sparse matrices is done in BaseLabelPropagations predictproba at line 189). This fix also heals LabelSpreading. FIX LabelPropagation handling of sparce matrices Label propagation and spreading allow to classify using sparse data according to documentation. Tests only covered the dense case. Newly coverage for sparse matrices allows to reproduce the problem in . The proposed fix in works for the extended tests. Supporting scipy's dokmatrix produces the UserWarning \"Can't check dok sparse matrix for nan or inf.\". So this format seems to be unsuitable? fails for sparse matrices"}
{"_id":"q-en-scikit-learn__scikit-learn-23099-f85722921306686c63235d0e9192c6a4ab4c196c61b62fd08ae9caf3ba472913","text":"GPR enforce before calling In , sampling in the prior before calling via will assume that is made of a single target. However, this is not necessarily the case. Therefore, the shape of the output of before and after is different. In order to solve this inconsistency, we need to introduce a new parameter . Before calling this parameter should be explicitly set by the user. After , we can use the information of the target seen during without explicitly setting the parameter."}
{"_id":"q-en-scikit-learn__scikit-learn-24677-101438b439c74899d3e86f5c4677d203bae3afd8b9588b5114c0f0eebbb59d59","text":"[MRG] Fixes sklearn.metrics.silhouette_samples for sparse matrices <!-- Thanks for contributing a pull request! Please ensure you have taken a look at the contribution guidelines: -- The changes update the reduce function used for computing the intra-cluster and inter-cluster distances. The current version is failing at, a) the pre-computed check for sparse matrices while getting the diagonal elements b) when trying to index a sparse matrix to pass weights to np.bincount function <!-- Please be aware that we are a loose team of volunteers so patience is necessary; assistance handling other issues is very welcome. We value all user contributions, no matter how minor they are. If we are slow to review, either the pull request needs some benchmarking, tinkering, convincing, etc. or more likely the reviewers are simply busy. In either case, we ask for your understanding during the review process. For more information, see our FAQ on this topic: http:\/\/scikit- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-24769-6e48a2218771b354eea06769a55b330fd2c5a6683166e30a1d3e666b12c86424","text":"Add meanaverageprecision Mean average precision (mAP) is a standard multi-class extension of average precision using OVR: (informationretrieval)#Meanaverage_precision Recently I prefer AP over AUC so I think it would be cool to add this. Maybe is interested? probably knows how to do this ;)"}
{"_id":"q-en-scikit-learn__scikit-learn-25102-3b8d05274993cd497a4b79cf366accfa87260389dc6357886f30327d3875d57b","text":"Preserving dtypes for DataFrame output by transformers that do not modify the input values It would be nice to optionally preserve the dtypes of the input using pandas output for transformers . Dtypes can contain information relevant for later steps of the analyses. E.g. if I include pd.categorical columns to represent ordinal data and then select features using a sklearn transformer the columns will loose their categorical dtype. This means I loose important information for later analyses steps. This is not only relevant for the categorical dtypes, but could expand to others dtypes (existing, future and custom). Furthermore, this would allow to sequentially use ColumnTransformer while preserving the dtypes (maybe related to ). Currently, this behavior is not given as one can see with this code snippet (minimal example for illustration purposes): Output (using sklearn version '1.2.dev0'): The ouput shows that both the and are converted to in the dataframe output. Maybe one could adjust the to also allow to preserve the dtypes. This would mean one changes the to add: another argument to . If not None the outputted dataframe uses to set the . The of the could be provided to by if the dtypes is set to preserve in the config. One could adjust specific transformers for which this might be relevant. Such a solution would need more work and does not seem to be inline with the simplicity that pandas output provides to the user for every transformer. is also interested in this feature."}
{"_id":"q-en-scikit-learn__scikit-learn-25232-91528ec4ec57449f72a5150b464fb8fce96356cefce396999a483589c6bafe0c","text":"IterativeImputer has no parameter \"fillvalue\" In the first imputation round of , an initial value needs to be set for the missing values. From its : I have set the initial strategy to . However, I want to define this constant myself. So, as I look at the parameters for I find : When strategy == “constant”, fillvalue is used to replace all occurrences of missingvalues. If left to the default, fillvalue will be 0 when imputing numerical data and “missingvalue” for strings or object data types. Based on this information, one would assume that also has the parameter , but it does not. The parameter needs to be to for when is set to . If this parameter is , please also allow as , for optimal compatibility with decision tree-based estimators. No response No response"}
{"_id":"q-en-scikit-learn__scikit-learn-25299-382413a457abb4987076913b74b79eb6a92423015716b3b3a0e367df1e65989f","text":"BUG logloss renormalizes the predictions renormalizes internally such that it sums to 1. This way, a really bad model, the predictions of which do not sum to 1, gets a better loss then it actually has. Result: Result:"}
{"_id":"q-en-scikit-learn__scikit-learn-25308-a1cf8a019c1750ec723c267de96f414192aea8b588e48c6d4a15310dee88df9a","text":"Make error message uniform when calling before While working , we found out that we are not consistent with the error type and message when calling before . From Please link the PRs you'll make with this issue so that I can update this list for all of us to know which estimators have been worked on: [x] AdditiveChi2Sampler() [x] Binarizer() [x] MaxAbsScaler() [x] MinMaxScaler() [x] Normalizer() [x] OrdinalEncoder() [x] PowerTransformer() [x] QuantileTransformer() [x] RobustScaler() [x] StackingClassifier() [x] StackingRegressor() [x] StandardScaler() [x] TfidfTransformer() [x] VotingClassifier() [x] VotingRegressor() [x] GaussianRandomProjection() [x] GenericUnivariateSelect() [x] IterativeImputer() [x] RFE() [x] RFECV() [x] SelectFdr() [x] SelectFpr() [x] SelectFromModel() [x] SelectFwe() [x] SelectKBest() [x] SelectPercentile() [x] SequentialFeatureSelector() [x] SparseRandomProjection() [x] VarianceThreshold() [x] KNNImputer() [x] SimpleImputer() [x] SplineTransformer() [x] DictVectorizer() [x] KBinsDiscretizer() [x] MissingIndicator() [x] IsotonicRegression() <details<\/detailsI assume that the most adequate error should be a asking to fit the estimator. WDYT?"}
{"_id":"q-en-scikit-learn__scikit-learn-25370-65c5849621044a7ebacc3f3d0e2e10b82c1caa58f92088e5acdc33292fcdff24","text":"sklearn.setconfig(transformoutput=\"pandas\") breaks TSNE embeddings TSNE doesn't work when the I tracked down this bug in the sklearn codebase. The issue is here: What's happening is that returns a Pandas array under API, with the columns being named \"pca0\" and \"pca1\". So when is called, we get an IndexError, because you'd have to index with in this situation. Possible fix could be changing line 996 to this: which I am happy to make a PR to do unless somebody has a cleaner way. Cheers! No error is thrown, a 2-dimensional pandas array is returned"}
{"_id":"q-en-scikit-learn__scikit-learn-25443-280196cde6c2940f03cf91cb427cf0ceacd8c66746b1af4bc84fe46d2010acb4","text":"With MLPClassifer, when warmstart is True or coeffs are provided, fit doesn’t respect maxiters With MLPClassifer, when warmstart is True or coeffs are provided, fit doesn’t respect maxiters. The reason for this is, when fitting, max iteration check is equality (==) against When warmstart is true or coeffs are provided, initialize is not called; this method resets niter to 0. Based on this implementation, there is doubt as to the meaning of maxiter. Consider, if maxiter is 1 and fit terminates due to reaching maximum iterations, subsequent fittings with warmstart true will never terminate due to reaching maximum iterations. This is bug. An alternate interpretation is maxiter represents the maximum iterations per fit call. In this case, the implementation is also wrong. The later interpretation seems more reasonable. Iteration 1, loss = 0. ConvergenceWarning: Stochastic Optimizer: Maximum iterations reached and the optimization hasn't converged yet. Iteration 2, loss = 0. ConvergenceWarning: Stochastic Optimizer: Maximum iterations reached and the optimization hasn't converged yet. Iteration 3, loss = 0. ConvergenceWarning: Stochastic Optimizer: Maximum iterations reached and the optimization hasn't converged yet. Iteration 1, loss = 0. ConvergenceWarning: Stochastic Optimizer: Maximum iterations reached and the optimization hasn't converged yet. Iteration 2, loss = 0. Iteration 3, loss = 0. Windows-7-6.1.7601-SP1 Python 3.6.0 (v3.6.0:, Dec 23 2016, 08:06:12) [MSC v.1900 64 bit (AMD64)] NumPy 1.12.0 SciPy 0.18.1 Scikit-Learn 0.18.1"}
{"_id":"q-en-scikit-learn__scikit-learn-25500-5a7a389e62b9bbd0c7cf8cf42289e95f42c698f96242aa82960d8b586d275776","text":"CalibratedClassifierCV doesn't work with CalibratedClassifierCV with isotonic regression doesn't work when we previously set . The IsotonicRegression seems to return a dataframe, which is a problem for in where it tries to put the dataframe in a numpy array row . It should not crash."}
{"_id":"q-en-scikit-learn__scikit-learn-25570-a61e7a0f35f4b7911c9c543ed707ec728a7fe180e3c66b0eb26b324d49df56e1","text":"ColumnTransformer with pandas output can't handle transformers with no features Hi, ColumnTransformer doesn't deal well with transformers that apply to 0 features (categoricalfeatures in the example below) when using \"pandas\" as output. It seems steps with 0 features are not fitted, hence don't appear in ( l.856) and hence break the input to the function (l.859). Here is some code to reproduce the error. If you remove .set_output(transform=\"pandas\") on the line before last, all works fine. If you remove the (\"categorical\", ...) step, it works fine too. The step with no features should be ignored. Here is the error message:"}
{"_id":"q-en-scikit-learn__scikit-learn-25601-5f333b43ae1b0b3847942a2f3b46675f28e5d1300b2054612da6467fdfd2f786","text":"RandomForestClassifier classweight\/maxsamples interaction can lead to ungraceful and nondescriptive failure The acceptable values for are . One possible option for is . However, for values of near zero and , the model fails with an unhelpful error related to having an empty array as an indexer. This is only likely to come up in a grid search, as it seems unlikely that someone would deliberately force a subsample of zero items. However, it might be useful to have a minimum of one sample, or it might be useful to fail gracefully with a more descriptive error. EITHER: No error is thrown OR"}
{"_id":"q-en-scikit-learn__scikit-learn-25638-deac0aa5fd511e7f43bac9ede7acab236a7be25f6827ac6a96cabb9dc606c1eb","text":"Support nullable pandas dtypes in I would like to be able to pass the nullable pandas dtypes (\"Int64\", \"Float64\", \"boolean\") into sklearn's function. Because the dtypes become dtype when converted to numpy arrays we get : Repro with sklearn 1.2.1 We should get the same behavior as when , , and dtypes are used, which is no error: Our current workaround is to convert the data to numpy arrays with the corresponding dtype that works prior to passing it into . No response_"}
{"_id":"q-en-scikit-learn__scikit-learn-25672-ccbcb7bc244bd9d992c9ad1c30992c45bebfb17314cadd4a283cca0a25835168","text":"NDCG score doesn't work with binary relevance and a list of 1 element See this code example: It works correctly when the number of elements is bigger than 1: score and Computing does not make sense for single predictions. Throw an error if is a list of length 1 for NDCG and DCG. Test that this throws the appropriate error by running: <!-- Please be aware that we are a loose team of volunteers so patience is necessary; assistance handling other issues is very welcome. We value all user contributions, no matter how minor they are. If we are slow to review, either the pull request needs some benchmarking, tinkering, convincing, etc. or more likely the reviewers are simply busy. In either case, we ask for your understanding during the review process. For more information, see our FAQ on this topic: http:\/\/scikit- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-25694-52577ab81f3368518ecab3102acbe62ab55cbc02d1ba84f9800100090b0ae078","text":"MLPRegressor.partialfit produces an error when earlystopping is True WIth , when using , works fine, but partial fit produces the following error: I think this is related to this change: If early stopping is not supported for partial fit, it should handle this gracefully. If this is a bug - it should be fixed."}
{"_id":"q-en-scikit-learn__scikit-learn-25733-04b4e584787247d5b1fbc4ee7df360d35dcd8527f1b4ac0cfb5e5e47fbd64203","text":"FutureWarning is not issued for deprecated class FutureWarning is not issued when using as a baseclass but its is not called in the subclass, here: As the comment suggest in your decorator, overriding ` in the class decorator indeed solves this issue. I'm happy to open a PR with the fix. Also, relatedly, I wonder whether you would be interested in using a generic deprecation package instead. Basically we have the same functionality in astropy (I feel it's actually has more features e.g. this works there out of the box, it helps with arg renames\/removals, etc.), there is also a deprecated decorator in matplotlib, and also a very basic one in numpy. I feel that having one for the wider ecosystem would be beneficial instead of the current system where we all roll our own. At the numfocus summit I recall some interest from the mpl side, so I'm happy to get the ball rolling in this quoter if it's a thumb up from multiple projects. Darwin-17.7.0-x8664-i386-64bit Python 3.7.5 (default, Nov 1 2019, 02:16:38) [Clang 10.0.0 (clang-1000.11.45.5)] NumPy 1.19.0.dev0+ SciPy 1.4.1 Scikit-Learn 0.23.dev0 Quick fix for class deprecation decorator This is a quick and dirty fix for my use case, but looking at the astropy decorator I suspect there may be some corner cases that still doesn't work."}
{"_id":"q-en-scikit-learn__scikit-learn-25744-ed757f85495c767eaa6ab8377f05a1a6ad3513fd66c8914eb40791afcb603a0c","text":"Setting minsamplessplit=1 in DecisionTreeClassifier does not raise exception If is set to 1, an exception should be raised according to the paramter's constraints: However, accepts without complaining. With scikit-survival 1.0, this raises an exception as expected: I suspect that this has to do with the Intervals of the constraints overlapping. satisfies the constraint, whereas the constraint should have precedence. No exception is raised."}
{"_id":"q-en-scikit-learn__scikit-learn-25747-d668dc7534a05e3d8eba6e6564b2ce2316c33122764db98912b0a7a5cc1359b6","text":"FeatureUnion not working when aggregating data and pandas transform output selected I would like to use transform output and use a custom transformer in a feature union which aggregates data. When I'm using this combination I got an error. When I use default output it works fine. No error is thrown when using transform output."}
{"_id":"q-en-scikit-learn__scikit-learn-25774-9899cc3ab159cc9d588f7a494a9c5f220ee28e0419c7fa9ed19a45cb6c16aa70","text":"FIX ignore nan values in partial dependence computation This PR implements the default behaviour suggested in that is ignoring values in both numerical and categorical features. Up to now, there is a bug since the computation of the percentile is impacted by the values for numerical features. In addition, introducing in the grid will potentially introduce a bias in the partial dependence computation depending on how missing values are handled in the model. Therefore, it is safe to remove them. To be consistent, then it is also safe to not include it as a category in the categorical features. In the future, we can think of adding an option to define the expected behaviour. I expect this PR to fail because we use instead of that does not use the same interpolation to compute the quantiles."}
{"_id":"q-en-scikit-learn__scikit-learn-25805-aff086285f572ddc9e1bfc340c751a3c0f67bcaa555be19b044f97eff27d8e2e","text":"CalibratedClassifierCV fails on lgbm fitparams Hi, I'm trying to use CalibratedClassifierCV to calibrate the probabilities from a LGBM model. The issue is that when I try CalibratedClassifierCV with evalset, I get an error ValueError: Found input variables with inconsistent numbers of samples: [, 1] which is caused by checkconsistentlength function in The input to eval set is [Xvalid,Yvalid] where both Xvalid,Yvalid are arrays with different shape. Since this format is a requirement for LGBM eval set , I am not sure how will I make the checkconsistentlength pass in my scenario. Full code is given below: = (, 152) = (,) = (, 152) = (,)"}
{"_id":"q-en-scikit-learn__scikit-learn-25931-da388d6f0e3fc64f59aed8f635d0b14b5bc5fff6a43102a1ced69da7ceda0d62","text":"X does not have valid feature names, but IsolationForest was fitted with feature names If you fit an using a it generates a warning This only seems to occur if you supply a non-default value (i.e. not \"auto\") for the parameter. This warning is unexpected as a) X does have valid feature names and b) it is being raised by the method but in general is supposed to indicate that predict has been called with ie. an ndarray but the model was fitted using a dataframe. The reason is most likely when you pass contamination != \"auto\" the estimator essentially calls predict on the training data in order to determine the parameters: Does not raise \"X does not have valid feature names, but IsolationForest was fitted with feature names\" raises \"X does not have valid feature names, but IsolationForest was fitted with feature names\""}
{"_id":"q-en-scikit-learn__scikit-learn-25973-5a0542d4f648578dd03b649eea25d0008208e2c18436315759a0a5f67a6342db","text":"Unable to pass splits to SequentialFeatureSelector This runs fine with e.g. , but according to the documentation, it should also be able to take an iterable of splits. However, passing splits from the cross validator fails Im fairly certain I have done similar things in the past to other classes in scikit-learn requiring a parameter. If somebody could confirm wether this is a bug, or I'm doing something wrong, that would great. Sorry if this is unneeded noise in the feed. Expected to run without errors"}
{"_id":"q-en-scikit-learn__scikit-learn-26194-2bbaa113f38b8fd50a5781e157d954f1b482358b6168802a3bb1bd46b7275a05","text":"Thresholds can exceed 1 in while providing probability estimate While working on , I found out that something was odd with that returns a threshold greater than 1. A non-regression test (that could be part of ) could be as follow: The reason is due to the following: Basically, this is to add a point for and . However, the rule does not make sense in the case is a probability estimate. I am not sure what would be the best fix here. A potential workaround would be to check in which case we should clip to not be above 1."}
{"_id":"q-en-scikit-learn__scikit-learn-26242-70c2515c25f6add856aae6c2d73d9ccbbc6b2239e68f3c72e1ef7f677a903684","text":"AdaBoost: deprecation of \"baseestimator\" does not handle \"baseestimator=None\" setting properly Scikit-learn 1.2 deprecated 's in favour of (see ). Because there are also validators in place, old code that explicitly defined stopped working. A solution that fixes the deprecation is to add a possible to a list allowed values in ; I will do that in a PR. No error is thrown."}
{"_id":"q-en-scikit-learn__scikit-learn-26289-cf2f093a79a637bc84842963e7d75fd4c775bc1e731c585c720a73df0658a931","text":"failing when featurenames supplied folks, I'm not sure why this works for but not for Can anyone help?"}
{"_id":"q-en-scikit-learn__scikit-learn-26318-7d7e305834129946b47a734597bb5fabdc8a792e4c660bb82eb040570b8869ef","text":"Incorrect documentation for behavior on BaseForest-derived classes The documentation states: This is also true for all classes that derive from , such as and . However, the does not reflect this behavior. When , it does not fit a new forest and instead just recomputes the OOB score if applicable. There are two potential fixes: the documentation to state: the actual behavior of this method to fit a new forest in the case where to reflect the existing documentation."}
{"_id":"q-en-scikit-learn__scikit-learn-26323-523ac1f78e7b1398ce9b5921fce881041ff4c5b1394e79f30cf69751e23afee3","text":"ignores the if it's an estimator When using on a , it sets the output to its sub-transformers but it ignores the transformer defined in . This issue causes the following to fail when gathering the results: Thus not gathering the final result correctly."}
{"_id":"q-en-scikit-learn__scikit-learn-26400-52f9737c13597654cc1d33040cd68a59cbcd3b28cf405eb513ebe8c8555a8fb5","text":"PowerTransformer fails with unhelpful stack trace with all-nan feature and method='box-cox' throws a difficult-to-debug error if x contains an all-nan column. Either no error is thrown and the all-nan column is preserved, or a descriptive error is thrown indicating that there is an unfittable column"}
{"_id":"q-en-scikit-learn__scikit-learn-26644-63f5e70b8c38eaa239ef69915b296977c5a100707dbf7f8e8a38b56f99fc784e","text":"partialdependence should respect sample weights Currently, the inspect.partialdependence funtions calculate arithmetic averages over predictions. For models fitted with sample weights, this is between suboptimal and wrong. Add new argument \"sampleweight = None\". If vector of right length, replace arithmetic average of predictions by weighted averages. Note that this does not affect the calculation of ICE curves, just the aggregate. No response No response_"}
{"_id":"q-en-scikit-learn__scikit-learn-3840-9952eb1a18b56d7f00d3e9aaf378644d8162af85212ddd7dfeba69454ff08761","text":"partial AUC I suggest adding partial AUC to the metrics. this would compute the area under the curve up to a specified FPR (in the case of the ROC curve). this measure is important for comparing classifiers in cases where FPR is much more important than TPR. The partial AUC should also allow applying the McClish correction. see here: http:\/\/cran.r-"}
{"_id":"q-en-scikit-learn__scikit-learn-7760-a5b6dce84a6d9cfd9a885ad9b0741b04d70a4ff7ec19383f656110d46f76f3e5","text":"Stronger common tests for setting init params? \/ checkestimator In a solution was proposed that did something like The common tests let this pass, though that should wreck havoc on and . I haven't looked into it but I think the tests should fail on this."}
{"_id":"q-en-scikit-learn__scikit-learn-8554-a97f3c564e3d43ac497346c440a8a0b6fad84752f4458a03a4920bdac3f1b068","text":"Isomap and LocallyLinearEmbedding do not accept sparse matrix input (contrary to documentation) The mentions that should support sparse matrix. The error comes from the 5 of from . If documentation is correct should be called with  I can submit a PR. also accepts sparse matrix according to documentation on and methods. Given that also uses the arpack solver, I guess that it also should accept sparse matrices. Check of checkarray calls in the manifold subfolder For reference, my backtrace"}
{"_id":"q-en-scikit-learn__scikit-learn-9288-b410de1afbeb4a2e6320a89a93d5b30609d1ea6dce61aed1642c83ecf2b2fc8d","text":"KMeans gives slightly different result for njobs=1 vs. njobs 1 <!-- If your issue is a usage question, submit it here instead: StackOverflow with the scikit-learn tag: Mailing List: For more information, see User Questions: http:\/\/scikit- --<!-- Instructions For Filing a Bug: --<!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--I noticed that gives a slightly different result depending on if or . <!-- Example: If the code is too long, feel free to put it in a public gist and link it in the issue: --Below is the code I used to run the same clustering on a varying number of jobs. <!-- Example: No error is thrown. Please paste or describe the expected results.--Should expect the the clustering result (e.g. the inertia) to be the same regardless of how many jobs are run in parallel. <!-- Please paste or specifically describe the actual output or traceback. --The case has a (slightly) different inertia than the parallel cases. <!-- Please run the following snippet and paste the output below. import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"NumPy\", numpy.version) import scipy; print(\"SciPy\", scipy.version) import sklearn; print(\"Scikit-Learn\", sklearn.version) --Darwin-16.7.0-x86_64-i386-64bit Python 3.6.1 (default, May 11 2017, 13:04:09) [GCC 4.2.1 Compatible Apple LLVM 6.0 (clang-600.0.57)] NumPy 1.13.1 SciPy 0.19.1 Scikit-Learn 0.20.dev0 <!-- Thanks for contributing! --"}
{"_id":"q-en-scikit-learn__scikit-learn-9304-68e80feb6d91213dfb2da48e4819db2481a72b7b178171e2b342795c4dc6cd2d","text":"Bug: the predict method of Pipeline object does not use the exact predict method of final step estimator I am trying to use Pipeline with a customized final step estimator. This final estimator predict method can output std when using returnstd=True. But the predict method of Pipeline does not allow returnstd option, gives error on scikit- Line 54. In the user guide:user guide http:\/\/scikit- says the following, but the predict method in Pipeline is not the same as that in final estimator \"\"\" Calling fit on the pipeline is the same as calling fit on each estimator in turn, transform the input and pass it on to the next step. The pipeline has all the methods that the last estimator in the pipeline has, i.e. if the last estimator is a classifier, the Pipeline can be used as a classifier. If the last estimator is a transformer, again, so is the pipeline. \"\"\""}
{"_id":"q-en-scikit-learn__scikit-learn-9775-2ed966f780c54de95c7cb2002e0a1a8bcbf41c37b1e0041582dc2d6f3f364127","text":"sklearn.manifold.t_sne.trustworthiness should allow custom metric boolean parameter should be replaced by more standard ."}
{"_id":"q-en-scikit-learn__scikit-learn-9939-edd8aceb779ece6f0e4f779308d06687098d9901fd3ccb2ca5debd5261e2f536","text":"Incorrect predictions when fitting a LogisticRegression model on binary outcomes with . Incorrect predictions when fitting a LogisticRegression model on binary outcomes with . <!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluateevery 0--<!-- Example: If the code is too long, feel free to put it in a public gist and link it in the issue: --If we compare against R or using , the log loss (which is approximately proportional to the objective function as the regularisation is set to be negligible through the choice of ) is incorrect. We expect the log loss to be roughly The actual log loss when using is . See the stack exchange question for more information. The issue it seems is caused in In the case even if we cannot reduce to a 1D case by throwing away one of the vectors of coefficients (as we can in normal binary logistic regression). This is essentially a difference between softmax (redundancy allowed) and logistic regression. This can be fixed by commenting out the lines 762 and 763. I am apprehensive however that this may cause some other unknown issues which is why I am positing as a bug. <!-- Please run the following snippet and paste the output below. import platform; print(platform.platform()) import sys; print(\"Python\", sys.version) import numpy; print(\"NumPy\", numpy.version) import scipy; print(\"SciPy\", scipy.version) import sklearn; print(\"Scikit-Learn\", sklearn.version) --Linux-4.10.0-33-generic-x8664-with-Ubuntu-16.04-xenial Python 3.5.2 (default, Nov 17 2016, 17:05:23) NumPy 1.13.1 SciPy 0.19.1 Scikit-Learn 0.19.0 <!-- Thanks for contributing! --"}
{"_id":"q-en-sphinx-doc__sphinx-10021-bb1740d73d23162de5b20247b2ac4b80e2fc039d95a15939af63a2970324e6b0","text":"autodocunqualifiedtypehints does not work well with autodoctypehints=\"description\" autodocunqualifiedtypehints does not work well with autodoctypehints=\"description\". Typehints in info-field-list are also changed to the short style when autodocunqualifedtypehints enabled. N\/A No response Mac 10.1 HEAD of 4.x autodoc No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-10048-4637aa850ee7fdd4e067c7eb72c4dc3e75c6e444e2d97273c1cb49797dacb084","text":"Change \"Permalink to this headline\" to use \"heading\" instead Sphinx-generated tooltips for headings use \"Headline\" instead of \"Heading\". Picking out of Build any Sphinx documentation, containing one or more headings (eg: ). When hovering over the Sphinx-injected \"headerlink\", notice that it says \"Headline\" instead of \"Heading\". Instead of: Permalink to this headline The tooltip should be: Permalink to this heading <img width=\"235\" alt=\"Screenshot 2022-01-01 at 17 20 09\" src=\"https:\/\/user-\"N\/A N\/A No response No response The relevant chunk of code: This will also need some update to the translated strings, which means that this will likely need to be a Sphinx 5.0+ change?"}
{"_id":"q-en-sphinx-doc__sphinx-10097-15f250636d4d341d0d4f9ebc6ce4267d4dcc78f5c70b2328c6a469924266a59a","text":"Latex: long option name overflows in Index Let's consider something like: Leads to the following with . ! Build snippet. No response Build the snippet No response Linux 8 0 No response No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-10137-6a45edf00fb4df77517b9ccf45e62124dd1e87e648d2b37a4b4e4b1481a6fb79","text":"Allow to bail out extlink replacement suggestion Feature via Consider the following ext link: and the following text: ` Can we somehow bailout out the check here, or perhaps the suggestion should only apply if there's no in the extlink, what do you think? cc"}
{"_id":"q-en-sphinx-doc__sphinx-10191-ff957b9494f6e98352d3cd68b551459dcc2168ada8d5bfe7711755a1f96c9e02","text":"Alternating multiply referred footnotes produce a ? in pdf output In some circumstances footnote mark is rendered as and there is no hyperlink file : then . Footnotes are rendered correctly See above code ! Mac 7 (CPython) 0 and current 4.x (v4.5.0+\/) No response No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-10320-306a50add322b7b61748c401d45e24e698379c7b670feca403497d993e493d63","text":"Overridden, overloaded class docstring return type rendered as None Some overloaded class definitions show ` as a return type, when there shouldn't be any return type. This seems to happen when the overloaded functions are overridden in the final docstring. A class without this problem is also provided for comparison. Exactly the same as ! ! Ubuntu 3 0 '', '', 'sphinxrtdtheme' No response Related Sphinx issue: This was originally reported in"}
{"_id":"q-en-sphinx-doc__sphinx-10321-cf082be255fae242cd9ed7b07324339c1854e3906ce5ae935dd8f0bf52843331","text":"autodocpreservedefaults=True does not work for mixture of keyword only arguments with\/without defaults If I understand correctly, the following is a valid signature of a class method: When documenting this with autodoc and , generates a with for the arguments. This later raises an exception in since the dunder method now returns instead of a string. Basically what happens is that ast generates a value in the of the since the first keyword argument is required, but simply ignores that argument because the is empty. This leaves the in the to be picked up when the keyword argument with default value is processed -- instead of the actual default. This can't be resolved by the call which therefore simply returns , which ends up as the of the . Imo this could simply be resolved by ing the corresponding from the if a parameter with empty is encountered. Create a module with contents and auto-document while setting in your Make sure sphinx tries to document all parameters, (since it's a method, they will be documented when the autodoc directive has , if you try the same with a module level method you need to document the parameters) The correct default value should be documented. The Warning Message also is pretty worthless (probably the value should not be formatted with a simple but instead with a ?) Any Tested with versions 3.8 0"}
{"_id":"q-en-sphinx-doc__sphinx-10323-dc4680e880c4f5ab86829b445a41572c00ff1e80adea272d6216641f79d9819f","text":"Use of literalinclude prepend results in incorrect indent formatting for code eamples Cannot determine a mechanism to use literalinclude directive with or to match code example indentation, as leading whitespace is removed. Example of including xml snippet, that should be prefixed with . File: File : Produces the following valid xml, which is indented poorly: I cannot think of good warning free way to indent to match the included code example. Expect leading white space to be preserved in output: No response Mac 10 0 ['', ''] No response Using creatively almost provides a workaround: Produces a warning, which fails the build with build policy. Use of could be a good solution, if was applied only to the literalinclude and not to the and content."}
{"_id":"q-en-sphinx-doc__sphinx-10325-2304225d2bad3df725fba4aa73861ba07969b3b524b78c9af1a698a86109ceb2","text":"inherited-members should support more than one class Is your feature request related to a problem? Please describe. I have two situations: A class inherits from multiple other classes. I want to document members from some of the base classes but ignore some of the base classes A module contains several class definitions that inherit from different classes that should all be ignored (e.g., classes that inherit from list or set or tuple). I want to ignore members from list, set, and tuple while documenting all other inherited members in classes in the module. Describe the solution you'd like The :inherited-members: option to automodule should accept a list of classes. If any of these classes are encountered as base classes when instantiating autoclass documentation, they should be ignored. Describe alternatives you've considered The alternative is to not use automodule, but instead manually enumerate several autoclass blocks for a module. This only addresses the second bullet in the problem description and not the first. It is also tedious for modules containing many class definitions."}
{"_id":"q-en-sphinx-doc__sphinx-10353-8eb04417131a996b1a05dc31b075798136499c7665ceeacd4a948bd1894fae2d","text":"Reference targets not found when using When using the above directive in my for sphinx, I suddenly get the following output: Previously I had , but when changing this behaviour suddenly it cannot find these names\/targets. NOTE: I use the following command to generate my docs, hence being able to see all warnings as errors:- I expected the docs to build correctly with the resolved types and reference links. N\/A I think ! Linux - Debian 10 6 0 , , , , sphinxcontribtrio, resourcelinks No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-10427-599216cd07c06172e8823577c21559714e469403a5a814e0ccaa0a5c30e2585c","text":"doesn't work on class methods Default values for class methods are rendered as their , even with in Extract the and run . Note that the default value for the argument of the regular and static methods render correctly as , but the default value for the class method renders as its repr. All three methods' argument's default values should render as Screenshot of the rendered attached example demonstrating incorrect rendering: ! Arch Linux 10.0 Tested on 4.5 and current 5.x branch (commit ) No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-10435-d37929c8787753180f2d476d5a54843f2feb05f19304aee88c6b4508deb8494a","text":"LaTeX: new Inline code highlighting from adds whitespace at start and end in pdf output The enhancement activates syntax highlighting for the Docutiles role. For LaTeX output, a space character is inserted at start and end of the inline code. Example a priori, mark-up should be: But I have no no strong opinion if good or bad. See screenshots. in and . No response extracted from with current: ! if space characters removed from file produced by LaTeX writer: ! For comparison prior to merge: ! Mac 9 No response No response Relates"}
{"_id":"q-en-sphinx-doc__sphinx-10449-6511c55d18e7218ce56c412a6a0de716cb6d63fd2bb3392eb4d4c731cf30f535","text":"causes autoclass to put a return type Using the option causes Sphinx's to include the class's \"return type\" for code such as this: <details<summaryOld repro, the repository no longer exists<\/summary<\/detailsa folder. that folder create files: : : : a virtual environment and install Sphinx 4.4 in it. into the docs folder and build the documentation with a command (in activated virtual environment): in the browser and see the issue. I expected there to be no return type listed for the class. Here's a link to generated docs: https:\/\/sphinx-issue- Windows 10, Ubuntu 18.04 7, 3.8, 3.9 0 No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-10451-175bd7c2b3ade2b9781863262693382c5488dbb0b256639a501a1c3c3c8fce52","text":"Fix duplicated args and kwargs with autodoctypehints Fix duplicated args and kwargs with autodoctypehints Bugfix Consider this when using the autodoc extension and the setting . WIth sphinx 4.2.0, the current output is where the args and *kwargs are duplicated and incomplete. The expected output is"}
{"_id":"q-en-sphinx-doc__sphinx-10457-f077303def68a952342dd54f9bee9b9979a095cf0e3c33034b83cc6cf50fd526","text":"fails to remove more than one meta-field from a fieldlist If a field list contains more then one \"meta\"-field, the generated documentation shows all but the first meta-field. It is a classical coding bug: the function stops removing meta fields after the first meta-field. Generated documentation contains a box with the text . The generated documentation does not contain any traces of the meta-fields. No response Linux 6 and 4.5.x No response The function filtermetafields was introduced by commit , which I'll provide a pull request with a test and a fix. fails to remove more than one meta-field from a fieldlist If a field list contains more then one \"meta\"-field, the generated documentation shows all but the first meta-field. It is a classical coding bug: the function stops removing meta fields after the first meta-field. Generated documentation contains a box with the text . The generated documentation does not contain any traces of the meta-fields. No response Linux 6 and 4.5.x No response The function filtermetafields was introduced by commit , which I'll provide a pull request with a test and a fix."}
{"_id":"q-en-sphinx-doc__sphinx-10466-3f90234a563571b9187d734651537d4d7d42fffa986e448da8eea5a5babd2ef3","text":"Message.locations duplicate unnecessary When running there are times the list of locations is duplicated unnecessarily, example: or as shown in this screen viewing of the 'pot' file result: <img width=\"1552\" alt=\"Screenshot 2022-01-15 at 20 41 41\" src=\"https:\/\/user-\"After debugging a little, the problem appeared to be in the file: in the 'init' method. My simple solution is this: Note, this solution will probably needed to be in the and in the as well. Follow instructions on this page which comprises of sections for installing dependencies, download sources. then load the file: into an editor and search for and you will see repeated locations appear there. The message id is: There should only be ONE instance of and there are NO duplications of other locations. No response MacOS Catalina 10.15.7 9 1 No response No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-10481-f7c0c4ab978450f5df673cebc2b96f7a63dca231f81b809f2e5610e6aca2f1f0","text":"If a project defines \"language = None\" in , treat it like \"en\" Hello, I started working on integrating Sphinx 5 to Fedora to ensure distribution packages work smoothly when the final is out. I ran across is a side effect of the change inspired by . If a project has already \"language = None\" defined in their (which, it seems, used to be an issue before commit), the new behavior will cause the documentation build to error out. The projects created after the mentioned commit seem not to be affected. In a sample of ~40 packages, 2 have run across this issue. A naive check using shows that for a half a million indexed GitHub projects there is around 6k which have the string in their (I removed the commented strings from the equation). For older projects using Sphinx, this change will be disruptive and will require the same commit in the same place for each and every one of them. The exact error: Describe the solution you'd like When Sphinx encounters NoneType for language, it could set the language to English and log the fact for the user (possibly asking them to make adjustments to ) instead of erroring. It's not that different than the current behavior in 5.0.0b1. When if I run and set no language, the variable is not present at all in , although in the background my project is processed as English. Describe alternatives you've considered Aforementioned manual change for each affected project, which I'm afraid of."}
{"_id":"q-en-sphinx-doc__sphinx-10492-1184cf13ea9b6d1f82640cbd95b30067ec76df85d96e6457bff0a01636434b06","text":"Typo in new language warning See the following snippet: Both and are used incorrectly. Noticed in Read the Docs: Mention \"language\" without typos."}
{"_id":"q-en-sphinx-doc__sphinx-10504-15756f94d9e635da25f067b6d75f65dca345c51603913d33c272f44dfb7591c3","text":"IndexError: list index out of range make[1]: [Documentation\/Makefile:96: htmldocs] Error 2 make: [Makefile:1678: htmldocs] Error 2 ==ERROR: A failure occurred in build(). Aborting... archlinux Linux RYZEN 5.18.0-arch1-1 x8664 , , , , , , No response"}
{"_id":"q-en-sphinx-doc__sphinx-10551-469578a0095cbbb0cbb70194e25371649211242dc6ac05d9262bee846a18d4d0","text":"Spurious space in default parameter values that are negative numbers in HTML output. For several projects, I've noticed a problem in the HTML output of functions that have a parameter with a default value that is a negative number. In the rendered HTML, there is a spurious space between the minus sign and the first digit. A typical example is being rendered as . This issue was originally raised with . Here are links to examples in several projects: SciPy: see . see . NumPy: Pandas: Matplotlib: I wasn't able to find an existing issue for this. I don't know which versions of Sphinx were used in all those projects, so I don't know if the problem still exists in the latest version of Sphinx. Also, it looks like those projects all use the , so it is possible that the problem is the theme and not Sphinx itself. See the links. No response See the links No response See the above comments. Probably varied; see the links. Maybe several; see the links. No response No response No response_"}
{"_id":"q-en-sphinx-doc__sphinx-10614-c3cde410fbbe24ef9839c4c55c6669126096e9e85c76fa617ec3deac0d7f8618","text":"inheritance-diagram 404 links with SVG I have created some SVG inheritance diagrams using the plugin. If the inheritance diagram is created in a file that is not in the root directory, the links lead to a 404 page. This issue does not happen in the default (png?) mode. This issue is similar to and however this is reproduced with only first party extensions. Here is a small demo that can be used to reproduce the issue. 1) Extract the folder from the zip 2) run 3) run (I believe this is the command pycharm is running) 4) Open the website to view (I am doing this through pycharm on firefox) 5) Navigate to see that the links work. 6) Navigate to see that the links do not work. My understanding of this bug is that the links in the SVG file are relative to the SVG file (because it is embedded using the object tag) however the rest of the link is written as if it was relative to the file the SVG is embedded on. Here are the correct links to the files Below are some examples of the links generated in the SVG file. They are formatted with the link the file was embedded on followed by the actual link text in the SVG file and then the path that firefox expands that to (the link when clicked on) I would expect that the links would go to the correct page when clicked on and not to a 404 page. No response Windows 1 2 , , No response No response_"}
{"_id":"q-en-sphinx-doc__sphinx-10673-d98ded249a6426198d16222e60027190016bbfb5b3f5c9390b17eb9d8cf6917d","text":"toctree contains reference to nonexisting document 'genindex', 'modindex', 'search' Is your feature request related to a problem? Please describe. A lot of users try to add the following links to the toctree: like this: See: And probably more. However when doing this we get: Describe the solution you'd like The following directive should be possible and do not rise errors: .. toctree:: :maxdepth: 1 :caption: Indices and tables genindex modindex search ``"}
{"_id":"q-en-sphinx-doc__sphinx-10757-92c78c07b1a2a8438a7127cc7acc967f6e37c56f9936b04339086e3fe180d113","text":"linkcheck should check URLs of raw directives Is your feature request related to a problem? Please describe. When using a with the option, the builder does not check this URL. Describe the solution you'd like I would expect the URLs of directives to be checked by ."}
{"_id":"q-en-sphinx-doc__sphinx-10819-63a816373718df17e821fdc408caf9ef994c02a145e3ddcebdd6dfd2010fc1a7","text":"Use the index directive as a source for search Is your feature request related to a problem? Please describe. My problem is the search engine is not good at finding terms that are indexed, for example: should find should find should find and a few others should find should find ... Describe the solution you'd like I think using the global index as a source for the search engine is a good way to enhance this and allow people to manually boost a search result by using the bang of the index directive. (). I can try to implement it, but I'm still not sure this is a good idea. Generated Index can point to anchors, I'm not sure the current searchindex can hold them in its current state."}
{"_id":"q-en-sphinx-doc__sphinx-11109-09579ea4296b6a22bd7f582ea513aeb5e46ec8eb0ed483cbc21d9cfb135d615a","text":"should improve formatting of Union, Optional, Literal, constants This is a sub-issue of split off here. When displayed normally, , , and add a lot of noise to the type signature and obscure the important information. Instead, it is much cleaner to display them using the PEP 604 () syntax: --Additionally, for it is cleaner to strip the text \"Literal\" and just display the literal value with normal Python syntax highlighting: -This is implemented in the tensorstore documentation via an ast transformation: This should be supported in Sphinx via a config option. The other improvement, of using syntax highlighting for constants, should also be integrated."}
{"_id":"q-en-sphinx-doc__sphinx-11192-638cffefda2cca1d97b1c0f73f7332032316140865363b2ff64be20ffa976a02","text":"Smaller\/wrong search index with sphinx-build -j auto Using the latest version and building with then the searchindex is considerably larger and searches work as expected. Some quick testing showed that Sphinx 6.0.0 did not have this problem while Sphinx 6.1.0 onwards does. This fails: sphinx-build -j auto -n -b html -d build\/doctrees -q -E . build\/html This works: sphinx-build -n -b html -d build\/doctrees -q -E . build\/html No response Sphinx (almost empty!) and other files are at"}
{"_id":"q-en-sphinx-doc__sphinx-11266-a5f53341c71c64db3107263f0c3f3c8b2c73e7e5e735efcf2cd195c90ca53583","text":"LaTeX: missing space before colon after \"Voir aussi\" for seealso directive in French Here is a screenshot ! No response This was caused by which fixed . See Sorry, my bad. LaTeX: missing space before colon after \"Voir aussi\" for seealso directive in French Here is a screenshot ! No response This was caused by which fixed . See Sorry, my bad."}
{"_id":"q-en-sphinx-doc__sphinx-11311-2ee8c298f9ec911d9f76d192d5dd0766315894d04070914b1dfa0f6312016570","text":"[FEATURE] Avoid losing information in . In light to , the current implementation of does the following assuming that the builder's current translator class is not a subclass of : If one or more node classes specified by (and derived from ) are not handled by the translator class, all are transformed into nodes instead (with the same attributes). If in addition the node is not handled by the visitor class, it is transformed into an node instead as well. This implementation choice suffers from the following disadvantages: Whenever a node inherits from , the class must be to the list. A custom translator not inheriting from should not be penalized if it only implements and handle it internally. Currently, they can emulate that behaviour by implementing and checking whether the XML attributes are those that we would find on nodes. As such, I suggest the following features: Use the hook together with a meta-keyword (defaults to ) to automatically add a class derived from to the list. That way, users may still use as a regular base class without affecting the nodes expected by Sphinx itself. A custom translator class orthogonal to must satisfy one of the following exclusive conditions: explicitly supports all nodes described in by providing methods. If a node inherits from but is not in , then must explicitly specify the corresponding visitor method. That way, nodes expected from Sphinx (i.e., those in ) are correctly dispatched and the remaining nodes are simply kept as is and handled as any other nodes. specifies a generic fallback method and possibly zero or more visitor methods for handling specific nodes. The nodes are then dispatched to their most precise type. does not have an interface for nodes and should therefore use a fallback. However, if there are other post-transforms coming after the fallback transform, it would be good to keep some information on the type the node originally had before converting them into nodes. For instance, we could add as an extra XML attribute of the transformed inline node. For , since the latter does not inherit from , I suggest that the fallback implementation simply add as an XML attribute in order to have a similar behaviour."}
{"_id":"q-en-sphinx-doc__sphinx-11312-2a6d6a6e6a16406366c693bb7677d963f63008eb1543cd0e9a96093020e68fa8","text":"util.inspect.objectdescription: does not emit reliable ordering for a set nested within another collection Differences appear in some v5.3.0 generated object descriptions for v1.8.1, as demonstrated by . Arguably it could make sense for code authors to intentionally write elements in their code files in a way that does not correspond to their computed sort order -- as a means to communicate with human readers about abstract ideas that aren't relevant to computers at runtime, for example. However, the current behaviour does result in non-reproducible documentation output. In particular, the ordering of a class attribute with a value that contains a set-within-a-tuple seems unreliable across differing builds: ... is emitted variously as ... ... or ... cc who has been . It is not yet clear to me exactly what circumstances cause the ordering of elements to vary - and it's OK not to proceed until that's figured out (maybe not a blocker, but it would be nice to have confidence about the cause). From searching around on previous issues while writing up this bugreport: I wonder if this could be an edge-case for \/ follow-up to . Although these build log links are somewhat ephemeral, the system environment details for two builds that produce differing output are visible at: https:\/\/tests.reproducible- https:\/\/tests.reproducible- No response_"}
{"_id":"q-en-sphinx-doc__sphinx-11316-c4c7e3708e4a123088e09ac3b383727e3e40e6a4eee3638437a67172d7d7e929","text":"Napoleon causes warning about missing end-string when encountering a reference on the first line of a docstring in a dataclass When using , a docstring of a dataclass attribute will cause an warning for a reference that's placed on the first line of this docstring. The reference itself is processed correctly. This causes the warning This one works just fine. Notice the linebreak in the docstring No response"}
{"_id":"q-en-sphinx-doc__sphinx-11445-640d4b1369dc40c323e3017ff12a34d066e7f804ad3820a100d06fafa57f252d","text":"Using rstprolog removes top level headings containing a domain directive If is set, then any documents that contain a domain directive as the first heading (eg ) do not render the heading correctly or include the heading in the toctree. In the example below, if the heading of were instead of then the heading displays correctly. Similarly, if you do not set then the heading will display correctly. This appears to have been broken for some time because I can reproduce it in v4.0.0 of Sphinx : : No response_"}
{"_id":"q-en-sphinx-doc__sphinx-11489-5f1d6c960975f7ec51086ed8cd64d831808153f95aa47b8e13d79ef9624f9046","text":"Allow disabling linkcheck anchor checks for specific URLs Is your feature request related to a problem? Please describe. It appears GitHub has made the Markdown renderer\/file viewer require JavaScript which breaks linkcheck anchor checks. Describe the solution you'd like A config which disables the linkcheck anchors check based on a regex of the entire URL would allow for cases like these to be handled while still validating whether the page itself exists and keep anchor checks enabled for others. Describe alternatives you've considered A clear and concise description of any alternative solutions or features you've considered. Additional context This is what looks like without JavaScript enabled: <img width=\"1007\" alt=\"Capture d’écran 2023-07-07 à 17 00 57\" src=\"\"- Related issue:"}
{"_id":"q-en-sphinx-doc__sphinx-11502-138113db482bb6898de30feea9994a276c5426df99d485538d38399b885534ea","text":"Keep 'translated' node attribute Is your feature request related to a problem? Please describe. In my internationalized documentation, I am adding markers to untranslated or partially translated pages, to warn the user that they can see English content and nudge them to help translating (e.g., like this: !). To do this, I'm essentially duplicating part of the transform. This feels clumsy because the transform already knows which nodes are translated and which aren't. In fact, it sets an attribute on the translated ones. However, this attribute is considered internal, so it deletes it at the end: Describe the solution you'd like I'd like to know if it would be acceptable to just delete the two lines of code above in order to let extensions know whether a node has been translated. Describe alternatives you've considered Adding the functionality for \"untranslated\" markers to Sphinx itself."}
{"_id":"q-en-sphinx-doc__sphinx-11503-5598fb582599ed0b926fbec123b6b61f7032975a73fc32a288d85535e23848ad","text":"linkcheck builder: begin using requests.Session functionality during linkchecking Is your feature request related to a problem? Please describe. At the moment, the builder performs individual (or similar HTTP request method) operations during linkchecking, without any explicit connection or session pooling. This may be inefficient, because it seems likely that for many use cases, linkchecking will make multiple requests to the same host (because documentation references are likely to have host-locality). Describe the solution you'd like Confirmation that connection pooling is not currently in use would be a good starting point; in other words: we should confirm that linkchecking of multiple URLs on a single host results in multiple TCP connections. Ideally this should be written as a test case. If we can confirm that the problem exists, then we may be able to use some of the from the library that's already in use here to enable connection pooling. Describe alternatives you've considered None so far, although open to suggestions (and improvements on the definition of this feature request). Additional context See note \/ suggestion about use of context managers:"}
{"_id":"q-en-sphinx-doc__sphinx-11510-d94d9a2eb26f4de279e9131a9732c8f4f4830c71d342ac60e9b9dca715d2b66c","text":"source-read event does not modify include'd files source In , we use a custom extension to do some search and replace in literal blocks, see We discovered ( and ) that this does not work on all files and some are left out of this mechanism. Such is the case for include'd files. I could reproduce on Sphinx 5.0.2. something-to- my- will contain: Note that the dumping docname and source[0] shows that the function actually gets called for something-to- file and its content is correctly replaced in source[0], it just does not make it to the final HTML file for some reason. should contain: No response Linux 10 2 Custom extension using source-read event No response No response source-read event does not modify include'd files source In , we use a custom extension to do some search and replace in literal blocks, see We discovered ( and ) that this does not work on all files and some are left out of this mechanism. Such is the case for include'd files. I could reproduce on Sphinx 5.0.2. something-to- my- will contain: Note that the dumping docname and source[0] shows that the function actually gets called for something-to- file and its content is correctly replaced in source[0], it just does not make it to the final HTML file for some reason. should contain: No response Linux 10 2 Custom extension using source-read event No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-11544-48ee7b2ede0013ae0a82d9a3b5586e13da3c2acdac54c24bfa21d02be507c5b0","text":"linkcheck failing after Sphinx 7.1.0 release Starting with , my package(s) started reporting failures due to \"Anchor not found\", e.g., Reverting to Sphinx 7.0.1 fixes the issue. reveals the issue started with , this PR: $ git clone $ cd photutils $ tox -e linkcheck No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-11550-7a5bc0e64ae41298b487841f0ea86f2733f2ea4b5a76c885035d9f404ca123d1","text":"autodoc preserve defaults leads to exception on multiline lambda In , I've stumbled into an issue where the docs builds are failing (). After some , I learned that the issue seems to be related to the use of with and the use of where the lambda is on a different line from the . Weirdly, removing the carriage return after suppresses the error. Also, converting to a traditional decorator or replacing the lambda with a simple function also suppresses the error:"}
{"_id":"q-en-sphinx-doc__sphinx-7234-647c491eced8ad1c9be23d70b9a8ada2f603e993e3385a269b826c883f0fc745","text":"Support for functions It would be nice if there was some mechanism to automagically pick up the overloads to a function and list them together... e.g."}
{"_id":"q-en-sphinx-doc__sphinx-7268-0961ff3d95d8df82a5c585c981b6d070111536e3f56f7005d389e6fd52cdd0b3","text":"autodoc: Load automatically After typehints enough matured, it should be loaded automatically from autodoc extension. refs:"}
{"_id":"q-en-sphinx-doc__sphinx-7305-06b7c60608ab924cc7c1c8fd30292171fa0bcffd157b4a97bf6f753183cb8a8e","text":"\"Could not parse arglist\" with operator pow Describe the bug With the rst I get the errors Expected behavior A clear and concise description of what you expected to happen. Environment info OS: Ubuntu 18.04 Python version: 3.6.9 Sphinx version: branches 3.x and master, but not 2.x Sphinx extensions: none Extra tools: none"}
{"_id":"q-en-sphinx-doc__sphinx-7350-4b0e2c5467eb3bbc6a356c0c5a95dde8b58464ec704e53bfad5aaa93e0f4cf94","text":"Napoleon's Attributes directive ignores :noindex: option. Description of the bug Sphinxcontrib-napoleon's directive appears to ignore the option. The following reST code produces an index that includes the directives found in but leaves out all other directives: Expected behavior The above example should produce an empty document index. Environment info I am using the Sphinx packages that are provided by Ubuntu 18.04 and installed Napoleon with pip3 afterwards: The file is from https:\/\/sphinxcontrib- I used to configure my directory, edited to include and set the Python path, then typed ."}
{"_id":"q-en-sphinx-doc__sphinx-7351-ab749c8bd294727422dea952afc35a414fe4a3b071e59291bcc434b8c63293e0","text":"Files with same name but different extensions leads to unexpected behaviour Describe the bug If there are multiple files with the same name but different file extensions; sphinx will silently choose only one to parse: To Reproduce Steps to reproduce the behavior: Given I have an extension installed to parse with: This will actually include in the document, not Expected behavior A clear and concise description of what you expected to happen. Ideally you would have a config option to specify the order of preference for file extensions, or if not set, a warning would be logged. Your project Link to your sphinx project, or attach zipped small project sample. First noted in: Screenshots If applicable, add screenshots to help explain your problem. Environment info OS: Mac Python version: 3.7.6 Sphinx version: 2.4.4 Sphinx extensions: Additional context Add any other context about the problem here. [e.g. URL or Ticket]"}
{"_id":"q-en-sphinx-doc__sphinx-7356-0e467b10db6c8898073790d7167fa52b3ac68b33de3fbc721a75dfdf1af1ebc9","text":"Breaking change to Python domain IDs Describe the bug Previously, anchors for Python functions were using underscores, changed this to dashes. To Reproduce Document some Python function whose name contains underscores: Expected behavior This used to create a fragment identifier , but since this creates . Your project This breaks links to python functions when used with : Apart from that all links (containing underscores) from external sites to Python API docs created by Sphinx (which I guess are a lot) will break!"}
{"_id":"q-en-sphinx-doc__sphinx-7374-0e467b10db6c8898073790d7167fa52b3ac68b33de3fbc721a75dfdf1af1ebc9","text":"Breaking change to Python domain IDs Describe the bug Previously, anchors for Python functions were using underscores, changed this to dashes. To Reproduce Document some Python function whose name contains underscores: Expected behavior This used to create a fragment identifier , but since this creates . Your project This breaks links to python functions when used with : Apart from that all links (containing underscores) from external sites to Python API docs created by Sphinx (which I guess are a lot) will break!"}
{"_id":"q-en-sphinx-doc__sphinx-7395-d0a5ace3a2ffb0451b991ea8d8bf5ec5569d1f3e9d94398d890bf618366adf7b","text":"The index directive with \"builtin\" type generates different index entry than the function directive and generate similar but different index entries. The former generates: The latter generates: Differences: are or not after the name of the function. are or not around \"built-in function\". takes one or two lines. \"max\" is a link or not. Link: This issue was reported on the Python bug tracker: Would be nice to unify and merge index entries generated by \"function\" and \"index\"."}
{"_id":"q-en-sphinx-doc__sphinx-7440-af98f10024d27b796380101a726b3b55bd9f428289133435da29454066c9281d","text":"glossary duplicate term with a different case Describe the bug To Reproduce Steps to reproduce the behavior: Expected behavior MySQL != mysql term right ? Your project Environment info OS: Unix Python version: 3.6 Sphinx version: 3.0.0 Additional context Did occur some hours ago, maybe you just released the version https:\/\/travis-"}
{"_id":"q-en-sphinx-doc__sphinx-7454-fb915dd38f96a1bffa3305ade42b00bb0b06138c02c56fac8dff24447ba16cc0","text":"Inconsistent handling of None by Describe the bug With , a function that returns generates a clickable link to . With , the in the signature is not clickable. To Reproduce Steps to reproduce the behavior: On running the above reproducer, note that the last two lines are: This contains a link from to the docs, but not one from to the docs. If you uncomment the line in the reproducer script and rerun it, you'll instead see: Expected behavior That in a type hint links to the documentation for the singleton regardless of whether 'description' or 'signature' mode is used. Environment info OS: Linux 4.4.0 Python version: 3.8.1 Sphinx version: 3.1.0.dev20200408 Sphinx extensions: , Additional context I installed a version of Sphinx that contains the fix for using:"}
{"_id":"q-en-sphinx-doc__sphinx-7462-0fba22e8fab0d4c351fce047d444c147966d9300ea808889af5d36e3e6fc3492","text":"for empty tuple type annotation Describe the bug Following notation for empty tuple from like I get To Reproduce Steps to reproduce the behavior: contents of snippet to module and set it to be explorable by sphinx. dependencies, in my : docs. Expected behavior Docs are built and there is with valid type annotations. Your project Environment info OS: Windows 10, but also reproduces on . Python version: 3.8.0 Sphinx version: 3.0.1 Sphinx extensions:"}
{"_id":"q-en-sphinx-doc__sphinx-7501-af98f10024d27b796380101a726b3b55bd9f428289133435da29454066c9281d","text":"glossary duplicate term with a different case Describe the bug To Reproduce Steps to reproduce the behavior: Expected behavior MySQL != mysql term right ? Your project Environment info OS: Unix Python version: 3.6 Sphinx version: 3.0.0 Additional context Did occur some hours ago, maybe you just released the version https:\/\/travis-"}
{"_id":"q-en-sphinx-doc__sphinx-7557-496768d4d316286d9ce0c00412aaab9a09c50e4f98f2268b96828943306010a3","text":"Decorated inherited method has no documentation Describe the bug If an inherited method has a decorator, it does not inherit the parent's doc. To Reproduce Expected behavior Both methods should be documented Environment info OS: Linux (Arch) Python version: 3.6 Sphinx version: 2.1.2 Sphinx extensions: [] Additional context Related to but calls so it should work."}
{"_id":"q-en-sphinx-doc__sphinx-7578-3d2147691f6317ef336bd6a1fbd715ff9582eacb44413bbd68eb4c3d91a61e1a","text":"BUG: Autosummary links broken in master As of commit (according to ) sphinx-gallery builds of a are failing where references to objects are no longer usable: I wondered if it was due to , but we explicitly have so I don't think so. I'll keep working on boiling this down to a more minimal example. But if anyone wants to try in the meantime, something like this should work:"}
{"_id":"q-en-sphinx-doc__sphinx-7593-681212f25b11fec2e4316c9211788e6d693e2ea42980cf443d6756dcfc5d8538","text":"Convert :kbd: to nested <kbdHTML elements Is your feature request related to a problem? Please describe. describes \"a sequence of keystrokes\". Sphinx converts it to a single  element: becomes Describe the solution you'd like Potentially sphinx could parse the value and convert it to individual (or nested) elements, which is what as a way to represent key combinations. Something like: Describe alternatives you've considered Alternatively the sphinx user could rewrite their markup: Related"}
{"_id":"q-en-sphinx-doc__sphinx-7597-1b23047b7500ae8af8d3dc5a87703ced44496ee4a893e81c0acd9de9a06484fd","text":"py domain: Change a type annotation for variables to a hyperlink Is your feature request related to a problem? Please describe. py domain: Change a type annotation for variables to a hyperlink Describe the solution you'd like option was to python directives since 2.x. But it has been represented as mere text. It must be useful if it is converted to a hyperlink to the type definition. Describe alternatives you've considered No Additional context No"}
{"_id":"q-en-sphinx-doc__sphinx-7615-4f43563304c477f2604f811b28bbdcb9aadb871b23c109a89f3da11d36020063","text":"Sphinx, unlike Docutils, incorrectly renders consecutive backslashes Describe the bug Sphinx incorrectly renders four or more consecutive backslashes. In pure Docutils, they are renderer properly according with RST spec. To Reproduce The following snippet demonstrantes buggy rendering. Expected behavior Two backslashes should be rendered as . Three still as . Four and five as . Six as and so on. This is how it works in Docutils. Screenshots ! Environment info OS: Linux Python version: 3.6 Sphinx version: 3.0.2 Sphinx extensions: none"}
{"_id":"q-en-sphinx-doc__sphinx-7686-a912f5565dc08c6eeccbc167c7dccfa7ce8647b90685df5bbe9b141e24fd96f8","text":"autosummary: The members variable for module template contains imported members Describe the bug autosummary: The members variable for module template contains imported members even if autosummaryimportedmembers is False. To Reproduce As a result, I got following output: Expected behavior The template variable should not contain imported members when is False. Your project No Screenshots No Environment info OS: Mac Python version: 3.8.2 Sphinx version: 3.1.0dev Sphinx extensions: Extra tools: No Additional context No"}
{"_id":"q-en-sphinx-doc__sphinx-7738-d84a05772d1c2a85514a745f69a8e72012efb27d509427d1e1077c045e08de2b","text":"overescaped trailing underscore on attribute with napoleon Describe the bug Attribute name shows up as in the html (visible backslash) with napoleon. To Reproduce Steps to reproduce the behavior: empty contains run add to extensions in add to PYTHONPATH=. make clean html open in web browser and see the ugly backslash. Expected behavior No backslash, a similar output to what I get for (the type shows up differently as well, but that's not the point here) Older versions like 2.4.3 look ok to me. Environment info OS: Linux debian testing Python version: 3.8.3 Sphinx version: 3.0.4 Sphinx extensions: , Extra tools:"}
{"_id":"q-en-sphinx-doc__sphinx-7748-3dfc40706f96c96c72343d01f1267f05bb7057ac7bf373d1197da2a694877b99","text":"autodocdocstringsignature with overloaded methods When using swig to wrap C++ classes for python, if they have overloaded methods, I believe the convention is to place the signatures for each of the overloaded C++ methods at the start of the docstring. Currently, can only pick up the first one. It would be nice to be able to pick up all of them."}
{"_id":"q-en-sphinx-doc__sphinx-7757-6588d8fa144640902df589a008167fa57631ea7dd9e6f70d9aa2f9e4e9d5b168","text":"The default value for positional only argument has vanished Describe the bug The default value for positional only argument has vanished To Reproduce Build following document: Result: <img width=\"148\" alt=\" 2020-05-30 23 43 01\" src=\"https:\/\/user-\"Expected behavior The default value is shown. Your project No. Environment info OS: Mac Python version: 3.8.2 Sphinx version: 3.1.0dev Sphinx extensions: No Extra tools: No Additional context No"}
{"_id":"q-en-sphinx-doc__sphinx-7760-d7dfb9f752a0a45adba88076f0395d42c409e93c58e8024cdbb59438838e2fb3","text":"coverage ext - config to print missing coverage as warning Is your feature request related to a problem? Please describe. I run CI tests with tox and want to test my docs also. I created the following tox env but the problem is that the env does not fail on missing coverage. Describe the solution you'd like My idea is to simply add another config variable like for the extension to not only print the missing coverage to files but also as a warning to stdout (or stderr?) like the builder on broken links. Describe alternatives you've considered I considered writing a script which checks if there is content in the file and returns the appropriate exit code. The script could then be called by after the builder ran. EDIT: Added config name."}
{"_id":"q-en-sphinx-doc__sphinx-7762-9aa42419329515c6a38682fb250ad261dfcf0db6a93ae49ca09ac711354dbd60","text":"Add support for custom HTTP headers @ linkcheck Is your feature request related to a problem? Please describe. Currently, HTTP header is hardcoded: And when I hit servers that require custom headers, the only option is to add those URLs to the ignore list which is what I'd like to avoid. Describe the solution you'd like Make HTTP headers configurable. Describe alternatives you've considered Adding the affected URL to Additional context We have a GitHub Actions badge in README which then gets embedded into Sphinx docs. Running used to work but now it doesn't. After some debugging I discovered that if the HTTP query doesn't have HTTP header, it works. But the header that Sphinx injects causes GitHub's server to respond with . Interestingly, if you open this URL in a browser, it works: Google Chrome sends the following header: ."}
{"_id":"q-en-sphinx-doc__sphinx-7814-fc743d5aadf0e9f524799f55bd4910b3e79830416ed99bf87e67d59c1169f965","text":"Warnings raised on variable and attribute type annotations Describe the bug autodoc signature for non-builtin types raises warning and thus fails nitpicking: To Reproduce Steps to reproduce the behavior: Create a file with the following content: Use sphinx-apidoc to generate an rst file, while enabling autodoc and intersphinx: Make sure the in the Sphinx contains Run with loud warnings and nitpicking: . You will get an error message Expected behavior I'd expect Sphinx to resolve the type annotation and possibly link both classes. Environment info OS: Linux Python version: 3.8.3 Sphinx version: 3.1.0 Sphinx extensions: , Additional context I think the issue stems from the change in which appears to try to lookup the entire type annotation as a single class. Using instead of solves this particular issue: However, it doesn't seem to work with custom classes. Take this snippet for example: This causes the following warning:"}
{"_id":"q-en-sphinx-doc__sphinx-7831-9b90f4cff36413de7ca93e27eeb3208ea6a4e7614050edc766308bc03b44c7e3","text":"regression in 3.1.0 for methods which are s and have a type comment Describe the bug methods decorated with trigger errors similar to this when built with sphinx 3.1.0 (this worked in 3.0.4): To Reproduce Steps to reproduce the behavior: Expected behavior This should work! Your project Screenshots N\/A Environment info OS: ubuntu 20.04 Python version: 3.8.2 Sphinx version: 3.1.0 Sphinx extensions: sphinx-rtd-theme, sphinx-prompt Extra tools: N\/A Additional context N\/A From a bisection, this is the commit which introduced the regression: CC"}
{"_id":"q-en-sphinx-doc__sphinx-7854-1eab53fa7dff4bc74fd86e66489a95a0bcd8543a1fa150dde2b32a1b8a233825","text":"Support for parameterized GNU style attributes on C++ code. Hi folks. My C++ codebase uses GNU attributes for code like Unfortunately, it looks like Sphinx doesn't support them. I'm running Sphinx 3.1.1, though this functionality doesn't appear to have changed in 4. I tried to get clever with the custom attribute support you offer, but can't seem to get that to work either. Is there a right way to do this? I'd honestly be fine having the attributes stripped entirely for doc generation if there isn't another option. Even though I'm bumping up against a sharp edge, I really appreciate Sphinx. Thanks so much for making a useful tool."}
{"_id":"q-en-sphinx-doc__sphinx-7859-3bbfdb343b95589aa90e8aac3caf04695e262ab9792bd96a2629e98ca6438d20","text":"autodoctypehints = 'descriptions' throws KeyError: 'module' Describe the bug autodoctypehints = 'description' does not work, results in To Reproduce Environment info OS: Linux Python version: 3.5.6 Sphinx version: 3.0.4 Sphinx extensions: , Additional context Full error message: Log file"}
{"_id":"q-en-sphinx-doc__sphinx-7889-f72cd0585fcabf741f7e94087af604ca16b3b5f83648adc4c8a58aa69d57e8fc","text":"Autodoc extension's mock file throws TypeError for generic-typed classes. Describe the bug When building the docs for a generically-typed class, a TypeError is thrown as Autodoc's attempts to concatenate a to a . See the attached log: To Reproduce Observe the docs command fails with a TypeError. Expected behavior Docs can still be built when generics are involved. Your project Environment info OS: Mac 10.15.5 (19F101) Python version: 3.7.7 Sphinx version: 3.1.1 Sphinx extensions: , , , , Additional context This might just be me not knowing how to make Sphinx-friendly generic typing, if that's the case please let me know!"}
{"_id":"q-en-sphinx-doc__sphinx-7906-7ca144ea1330b30d1602eeec7bf4a2166a444d4a1a2b75b050305d2e1f42c82b","text":":noindex: prevents cross-referencing If a flag is to a directive, it can't be cross-referenced, and no permalink to it is generated. The following ReST: generates the following HTML: I would expect only to result in no index entry, not to prevent cross-referencing or permalinking. The HTML generated for the two class directives should be the same, i.e. the HTML for the Unindexed class should be OS: Linux Mint 19.1 (based on Ubuntu 18.04) Python version: 3.8.1 Sphinx version: 3.0.0 (HEAD) but also occurs with Sphinx 2.x, 1.x"}
{"_id":"q-en-sphinx-doc__sphinx-7910-08957ae310766c8c54d8e1b11f4b8cde9e4acc2999782ab7691b1e64e46566e5","text":"Decorated init doesn't show up in docs Subject: Decorated init won't be documented. I'm working on I have , so will be documented. But if I decorate the method, it will not show up in docs. I decorate it with , so the decorated object still has the same . I've found that the bug is due to this commit: I've printed the relevant variables in that function: And the root cause of the issue is in this line of code: Because now is not the method itself, but a decorated method, its does not contain the class anymore. This makes sphinx think it's not a method, i.e. . OS: <Unix\/Linux\/Mac\/Win\/other with version: ArchLinux Python version: 3.6 Sphinx version:1.6.5"}
{"_id":"q-en-sphinx-doc__sphinx-7923-8b6dc1f7e9bce2e3708e3d087dcbd581dd8c90e930b45cec5d055ca26c8fa27c","text":"Bad refs in pot files, when using rst_epilog To Reproduce produces with bad string numbers and \"\\<generated\\\" refs:"}
{"_id":"q-en-sphinx-doc__sphinx-7930-ec85969b1777d4fd870679568db0af0cac7b259b33c7e167397e5958328fdac7","text":"Regression: autodoc Dataclass variables reference target not found Describe the bug When I use and with my code which includes a dataclass with a variable of a custom type, I get a warning. To Reproduce Open the attached project . Install Sphinx. Run . Expected behavior I expect there to be no warning, or a clear message saying how I can avoid this warning. Your project Environment info macOS latest Python 3.7.7 Sphinx 3.1.2 (reproducible also with 3.1.0 but not 3.0.4)"}
{"_id":"q-en-sphinx-doc__sphinx-7961-fe9430a21d4fbe5f1fa71ead26ee932cde4e6f750ff5083322dad412e779b327","text":"Support mathjax 3.0 , and has a number of improvements - most notably on the speed of loading (something like an 80% decrease in load time). It would be great to get this into the Sphinx MathJax extension. The suggest that it will not be trivial to upgrade, but it may be worth it given the performance and modularity improvements. What do folks think?"}
{"_id":"q-en-sphinx-doc__sphinx-7975-f95591d7785d75801a6cdc56337c7dda3d5a8bfade69689d629444e3186a39e9","text":"Two sections called Symbols in index When using index entries with the following leading characters: £, and ← I get two sections called Symbols in the HTML output, the first containing all @ entries before ”normal” words and the second containing £ and ← entries after the ”normal” words. Both have the same anchor in HTML so the links at the top of the index page contain two Symbols links, one before the letters and one after, but both lead to the first section."}
{"_id":"q-en-sphinx-doc__sphinx-7985-5b766785ffdc2cedf65bccf94d643f477bb9d7dfbea296db75c39f265cf4a7f8","text":"linkcheck could also check local (internal) links Subject: linkcheck currently doesn't check local (internal) links, but this would be useful. <!-- Important: This is a list of issues for Sphinx, not a forum. If you'd like to post a question, please move to sphinx-users group. !forum\/sphinx-users Thanks, --See above. Create a template project with sphinx-quickstart, put the following in Run Also a check for the local link. N\/A OS: Arch Linux Python version: 3.6 Sphinx version: 1.7.6"}
{"_id":"q-en-sphinx-doc__sphinx-8020-a32fffded567cb2d2fc609b00fea578f385bcdd2b8a689e0af62bf49820b0061","text":"2 introduces WARNING: py:class reference target not found: Callable[[...], Awaitable[None]] Describe the bug My project, including the above function definition, builds cleanly with Sphinx 3.1.1 but with 3.1.2 I get the following warning. Note that the resulting parameter documentation does properly hyperlink and but does not get a link. https:\/\/qtrio-- To Reproduce Steps to reproduce the behavior: Expected behavior The error doesn't happen in either case and , , and all get properly hyperlinked. Your project Screenshots ! Environment info OS: Linux - Ubuntu 20.04 Python version: 3.8.3 Sphinx version: 3.1.2 Sphinx extensions: sphinxautodoctypehints sphinxqtdocumentation sphinxcontrib_trio Additional context"}
{"_id":"q-en-sphinx-doc__sphinx-8026-58f2db67b6bfbea0aa68c1382986997d19d6bad3b22ea3aa9ec2159dc2eac42c","text":"More customization options for figurelanguagefilename for localized images Is your feature request related to a problem? Please describe. I'm working on the localization of the , and we use a to hold the massive PO files that we use for Weblate and Sphinx. I'm now working on image localization () and I find that even though  seems to provide some customization option, it's impossible to host the localized images in the separate repository. Indeed, both the and substitution tokens resolve to absolute paths from the host system (e.g. ), and since I can't do post-processing of the subsituted string, I can't strip the base path to attempt using something like . Describe the solution you'd like I'd like the addition of one or more new path substitution tokens that can be used to customize the path to localized images more freely. For example, for this structure: and referencing , I could imagine two useful substitution tokens: Alternatively (and possible as a better solution), and could be changed to actually be relative to the Sphinx project's root folder, i.e. the in my example above, i.e.: (While the documentation currently states that these would be and , which is wrong in my tests on Linux with Sphinx 1.8.5). I don't specifically need access to the file-relative path in my use case, but I imagine that some projects may benefit from it. Describe alternatives you've considered I don't see any alternative option that would enable the workflow I want to use with localized images not being either in the original image's folder, or in one of its subfolders. Edit: We came up with ~two~ three hacks that we could use to workaround the currently limited substitution tokens offered by : Passing a class that redefines with our custom post-processing: Symlinking all localized images from the out-of-tree location to the in-tree location where Sphinx wants it: Overriding since apparently Python lets us do that, so we can do the post-processing we want: I did not test with latest Sphinx as it's not properly supported on ReadTheDocs yet, but the relevant code seems to be unchanged from 1.8.5: Additional context"}
{"_id":"q-en-sphinx-doc__sphinx-8028-207212608f9f54cbdc5bf365aeaab379433fe14e1c4e8e6ac8acc95be937367e","text":"Instance attributes are omitted with autosummary This issue is a followup of . Sorry, I forgot to open it. Is your feature request related to a problem? Please describe. Instance attributes are omitted when using the autosummary, as they cannot be imported. Consider the following class via will document the attribute, but via won't. Describe the solution you'd like Instance attributes should be documented, too Describe alternatives you've considered it did work before , i.e. for (although the documentation was omitted). Additional context files to reproduce the issue (including page build) Screenshot of !"}
{"_id":"q-en-sphinx-doc__sphinx-8035-25d7c4e4854d095a81b7f377df436f2d2d37ce219e7f6bad1210477ac0cf3399","text":"Support defining specific for autodoc Is your feature request related to a problem? Please describe. Currently, if I'm using autodoc, the option does not allow specification of which private members to document. The current behavior is to document all private members, but what if I would only like to document 1 or 2? Describe the solution you'd like For to take arguments, similarly to how currently works Describe alternatives you've considered The current best way to do this is to explicitly list each class in a module and use Some prior discussion:"}
{"_id":"q-en-sphinx-doc__sphinx-8037-bb531e16dbfe2673c0bd21b40570164fd1cb9b635aa78adfea04b4b02d089064","text":"Error in template parameter list. Describe the bug I appear to have some valid C++14 syntax which causes Sphinx's C++ parser to choke. Here is the full error trace: To Reproduce Steps to reproduce the behavior: Expected behavior Build should succeed"}
{"_id":"q-en-sphinx-doc__sphinx-8056-36fb6dc690143e1b78edb620fd7d3d7c0f019922d774d9f5a930b2b73d64db64","text":"Render error when combining multiple input parameters in docstring Describe the bug & Reproduce My team is writing a function in Python, which contains 3 inputs that are similar, so we want to put them in the same line in the docstring. As described in 4. Parameters in , this is possible if you write something like this: However, this produces: <img width=\"406\" alt=\"\" src=\"https:\/\/user-\"Even worse, when \"optional\", the rendered HTML stays the same as the screenshot above, so there is no way to tell whether it is optional: Expected behavior Something like x1, x2 (arraylike, optional_) - Input arrays, description of x1, x2. Environment info OS: macOS 10.15.5 (19F101) Python version: 3.7.7 Sphinx version: 3.0.3. Extra tools: browser: Firefox 79.0a1 or Safari 13.1.1 Sphinx extensions:"}
{"_id":"q-en-sphinx-doc__sphinx-8058-6498783eef88ec7415d9f451b0e6d737574b280f2663c46078810ca2de67e4ee","text":"Allow more compaction in gettext builder The OKFN people manually merge all generated message catalogs into a single file for upload. {{{gettext_compact}}} should support that use case too. Bitbucket: Originally reported by: Robert Lehmann Originally created at: 2011-10-11T16:43:20.007"}
{"_id":"q-en-sphinx-doc__sphinx-8075-1c1afca4480bb4503071546e22e70be1737bd37433a24606685a848895361851","text":"References to figures without captions: errors in both HTML and LaTeX Describe the bug Using figures without captions causes errors in both HTML (though these are properly reported when source is processed) and in LaTeX (they are not reported until LaTeX says there were undefined references). This was the test document, compiled with sphinx 2.2.2 from pypi; was to , the project was otherwise generated with sphinx-build with no other changes. It is attached here: and these are the results: warning: WARNING: undefined label: fig-sample-nocaption (if the link has no caption the label must precede a section header) is HTML output (the error highlighted corresponds to the warning mentioned above): ! is LaTeX (pdflatex) output: ! Expected behavior I expect to produce valid LaTeX input without undefined references; figures to be referencable in LaTeX (this could be an optional setting perhaps causing uncaptioned figured to produce only \"Figure 4.1.\" caption); about figure not being captioned to be more meaningful -- I understand that non-numbered figure cannot be referenced via :ref: (as the label will not resolve to any text) but the warning is not pointing to how to fix the issue. Environment info OS: Ubuntu 18.04 LTS Python version: 3.6.8 Sphinx version: 2.2.2 Sphinx extensions: none Extra tools: pdflatex TeXLive"}
{"_id":"q-en-sphinx-doc__sphinx-8095-a6ddb2f06239ebd65f38672c50b72493059753a87ac8fefd72c5d4317fb236ea","text":"Warning: Inline literal start-string without end-string in Numpy style Parameters section Describe the bug The following docstring generates a warning on the line of the timeout parameter. Removing the quote around cause the warning to go away. To Reproduce Steps to reproduce the behavior: Expected behavior I do not expect to see a warning there and was not seeing any before 3.2 Your project The project is build under the Documentation build action. Environment info OS: Mac Os and Linux Python version: 3.8.2 and 3.8.5 Sphinx version: 3.2.0 Sphinx extensions: \"\", \"\",\"\", \"\", \"\", \"\", \"\""}
{"_id":"q-en-sphinx-doc__sphinx-8120-26ed2b3a50a6aa329c186e38d77b66dba97e0ff3ef18b2b743ed4ba4600cb3e0","text":"locale\/ translation ignored Describe the bug I read [1] as it should be possible to add a file ) and through that change translations or add additional translation to &lt;language. When I add. [1] [2] To Reproduce Steps to reproduce the behavior: Notice that . Expected behavior The caption label for the figure (again for the sake of testing). Your project Screenshots ! Environment info OS: Arch Linux Python version: 3.7.3 Sphinx version: 2.1.2 Sphinx extensions: none Extra tools: none"}
{"_id":"q-en-sphinx-doc__sphinx-8125-52bb63c3bb426be814436e858ff8502097ec28198393cf1100783a73c12f0ffa","text":"All identifiers should be passed to autodoc-skip-member, even if not in all Is your feature request related to a problem? Please describe. I have a Python project for which I've generated docs. It did well. Almost everything was doc'ed. I knew about the default of excluding anything starting with '_' so I created an autodoc-skip-member function that would include those. It works. I see a bunch of private methods and functions included. But...I have a module, let's call it XYZ that has both private and public functions. Sphinx is only doc'ing the public functions of XYZ, but not its private functions. I have other modules that have private functions, and it's doc'ing those .I a print() to the autodoc-skip-member function and it doesn't even print out the of those private functions so it seems it's not even seeing it. I figured it out. The module had certain identifiers listed in , and was not passing anything else in that module to autodoc-skip-member. Describe the solution you'd like Sphinx should pass through all members, not just the ones in . Maybe add something to the object that is passed to autodoc-skip-members that indicates it would have been excluded by the designator, or some such. Describe alternatives you've considered I simply removed the from the file."}
{"_id":"q-en-sphinx-doc__sphinx-8202-83a98518b4a4b226e81c88676cc82ecd49d989f995448c05b8737dabe756386f","text":"Problems with doc comments using Generic class and extending it Describe the bug This is a very specific bug (I think), when we have a class with doc comment for an attribute and I extend it. If the child class has the method (using typed parameters) and I try to reassign the attribute from class I got the following error: To Reproduce Steps to reproduce the behavior: Code example: Python script rst If you try to modify something in the script maybe the error will not appear, it's a very specific corner case! Expected behavior The warning should not appear! Your project Environment info OS: Linux 5.7.9-1-MANJARO Python version: 3.8.3 Sphinx version: 3.1.1 Sphinx extensions: related issue"}
{"_id":"q-en-sphinx-doc__sphinx-8264-1cbfb0064e500cc63d2ee8ba6d5da24c8dc933c3427f46361f11b7a7b1367cab","text":"TypeError: 'type' object is not iterable Hi All, this is the first time I try to create a documentation using Sphinx. venv is a virutal environment created with miniconda. Using the Miniconda3 prompt, I activated the environment and tried to create the documentation. As suggested in the error, please find below the error log. Thanks a lot for your help! PS: for privacy, I hide my absolute path calling it PATH."}
{"_id":"q-en-sphinx-doc__sphinx-8265-07e46c627e96313e75be72bc8e79f9da6ffd14b6ef22b252b27d55ab4dfa841a","text":"docstring default arg is broken Describe the bug docstring default arg is broken in html. Python class method is rendered as To Reproduce Steps to reproduce the behavior (Dockerfile): Expected behavior A clear and concise description of what you expected to happen. Python class method is rendered as Your project Link to your sphinx project, or attach zipped small project sample. Screenshots If applicable, add screenshots to help explain your problem. ! Environment info OS: [e.g. Unix\/Linux\/Mac\/Win\/other with version] Linux Python version: [e.g. 3.7.1] 3.7 Sphinx version: [e.g. 1.8.2] sphinx-build 3.1.1 Sphinx extensions: [e.g. , recommonmark] Extra tools: [e.g. Browser, tex or something else] None Additional context Add any other context about the problem here. [e.g. URL or Ticket] None"}
{"_id":"q-en-sphinx-doc__sphinx-8269-4e1cd7a962e1fe739eab1683ed49fb5966a21ad86a1fff57b63ddb0f624e0113","text":"Linkcheck should report HTTP errors instead of Anchor not found Describe the bug The command always reports that it was unable to find the anchor when  is , even when the server replied with an error status code (e.g. 404, 500). While it is indeed unable to find the anchor, the real issue is that the server encountered an error. To Reproduce Expected behavior Actual Expected output Same as when . Environment info OS: Linux 5.8.12.a-1-hardened Python version: 3.8.5 Sphinx version: 3.2.1"}
{"_id":"q-en-sphinx-doc__sphinx-8273-bc30a981a17dfc9c1ffd67383889a2348025eb446c483f318603828200f4ab8c","text":"Generate man page section directories Current man page generation does not conform to search functionality Currently, all generated man pages are placed in to a single-level directory: . Unfortunately, this cannot be used in combination with the unix environment variable. The program explicitly looks for man pages in section directories (such as , etc.). Describe the solution you'd like It would be great if sphinx would automatically create the section directories (e.g., , , etc.) and place each generated man page within appropriate section. Describe alternatives you've considered This problem can be over come within our project’s build system, ensuring the built man pages are installed in a correct location, but it would be nice if the build directory had the proper layout. I’m happy to take a crack at implementing a fix, though this change in behavior may break some people who expect everything to appear in a directory."}
{"_id":"q-en-sphinx-doc__sphinx-8278-6f54e813ac44239c5223b7c087b498ad15b27d915d4a5bfec526edcf836710b1","text":"Sphinx changes number system from hexadecimal to decimal for function default arguments Describe the bug When documenting a function, which requires a default argument specified in hexadecimal: Sphinx 3.2.1 will render HTML documentation where the hexadecimal value is in a different number system, than what was specified by the user (decimal), as seen in the following output from the above input: ! Expected behavior The expected behaviour would be to present the default argument in the same radix\/number system as the user typed in, thus in this case the documentation would also display the value in hexadecimal. In the above example of with colors, it is unintuitive what the color might refer to, but it's easy read RGB colors in hexadecimal format - e.g. 0xffffff is white, thus which number system is used can make a huge difference. The same for example goes for e.g. specifying access rights for files () Screenshots ! Environment info OS: Mac OS X 10.14.6 Python version: 3.8.5 Sphinx version: 3.2.1 Sphinx extensions: None Extra tools: None"}
{"_id":"q-en-sphinx-doc__sphinx-8282-717e7efc1585f0803d8806fc19f8b132cfc43fd1c0f78663550c0e5077716a39","text":"autodoctypehints does not effect to overloaded callables Describe the bug autodoctypehints does not effect to overloaded callables. To Reproduce Expected behavior All typehints for overloaded callables are obeyed setting. Your project No Screenshots No Environment info OS: Mac Python version: 3.8.2 Sphinx version: 3.1.0dev Sphinx extensions: Extra tools: No Additional context No"}
{"_id":"q-en-sphinx-doc__sphinx-8284-bc30a981a17dfc9c1ffd67383889a2348025eb446c483f318603828200f4ab8c","text":"Generate man page section directories Current man page generation does not conform to search functionality Currently, all generated man pages are placed in to a single-level directory: . Unfortunately, this cannot be used in combination with the unix environment variable. The program explicitly looks for man pages in section directories (such as , etc.). Describe the solution you'd like It would be great if sphinx would automatically create the section directories (e.g., , , etc.) and place each generated man page within appropriate section. Describe alternatives you've considered This problem can be over come within our project’s build system, ensuring the built man pages are installed in a correct location, but it would be nice if the build directory had the proper layout. I’m happy to take a crack at implementing a fix, though this change in behavior may break some people who expect everything to appear in a directory."}
{"_id":"q-en-sphinx-doc__sphinx-8291-828deabf5b160a922867695660ee8243fd6100c4f59cf572ac560ebc34e4b9ed","text":"[Napoleon] Retrieve type of attributes from type hints when using google-style Following , I an section my classes docstring. As I have annotations (because I use ), I did not add types in the docstring., as per the style guide. When generating the documentation with from , with , the attributes are not typed. I tried using the flag of , but this resulted in duplicated attributes: one with the explanation, the other with the type-hint. While it is possible to not use type hints with and replace by , this is not an option for e.g. . I also tried , which silenced sphinx's warnings but still resulted in two definition of the attribute in the documentation. It would be nice if (or ) extracted the type hint of the class, or merged the attributes when using . That would avoid duplicated types definitions in either the code or the doc. Currently, either the code has an annotation and a type in the docstring, or the generated documentation has two entry for each attribute. This might be related to and . Additional info sphinx version"}
{"_id":"q-en-sphinx-doc__sphinx-8362-5150371f7887ed609e57aece978f176776c11f64695bcdcb8d6ec7d408629eaa","text":"autodoc uses a wrapper's signature for a decorated class Describe the bug uses a wrapper's signature for a decorated class. To Reproduce Branch: Steps to reproduce the behavior: In , add:. In , add: Then, run: This outputs: Expected behavior A clear and concise description of what you expected to happen. Your project Link to your sphinx project, or attach zipped small project sample. Screenshots If applicable, add screenshots to help explain your problem. Environment info OS: [e.g. Unix\/Linux\/Mac\/Win\/other with version] Python version: [e.g. 3.7.1] Sphinx version: [e.g. 1.8.2] Sphinx extensions: [e.g. , recommonmark] Extra tools: [e.g. Browser, tex or something else] Additional context Add any other context about the problem here. [e.g. URL or Ticket]"}
{"_id":"q-en-sphinx-doc__sphinx-8435-f0e86c36b7ea4e4b957c9ffd8a156af561fb9bce1c394c98ea8a1e0d19586708","text":"autodoctypealiases does not effect to variables and attributes Describe the bug autodoctypealiases does not effect to variables and attributes To Reproduce Expected behavior should be applied to and . Your project N\/A Screenshots N\/A Environment info OS: Mac Python version: 3.9.0 Sphinx version: HEAD of 3.x branch Sphinx extensions: Extra tools: Nothing Additional context N\/A"}
{"_id":"q-en-sphinx-doc__sphinx-8459-d62824a1f96ad55b05c479d44ad8bdc60d535897b0d2905575454257c2a0111b","text":"autodoctypealiases doesn't work when autodoctypehints is set to \"description\" Describe the bug autodoctypealiases doesn't work when autodoctypehints is set to \"description\". To Reproduce I get, Then if I remove I get, Expected behavior instead of in both cases. Environment info OS: Mac Catalina 10.15.7 Python version: 3.7.9 Sphinx version: 3.3.1 Sphinx extensions: , ,"}
{"_id":"q-en-sphinx-doc__sphinx-8474-53ac4aef807573f09f85e090a10dda87710b0dc624feb89c37ae93a9bf639698","text":"v3.3 upgrade started generating \"WARNING: no number is assigned for table\" warnings We've updated to Sphinx 3.3 in our documentation, and suddenly the following warning started popping up in our builds when we build either or .: I looked through the changelog but it didn't seem like there was anything related to that was changed, but perhaps I missed something? Could anyone point me to a change in the numref logic so I can figure out where these warnings are coming from?"}
{"_id":"q-en-sphinx-doc__sphinx-8475-55d290b930e6936d301ccdaaa125db399661e387dc59eaa16adec57e1c35e74d","text":"Extend linkchecker GET fallback logic to handle Too Many Redirects Subject: linkcheck - fallback to GET requests when HEAD requests returns Too Many Redirects Bugfix Some websites will enter infinite redirect loops with HEAD requests. In this case, the GET fallback is ignored as the exception is of type and the link is reported as broken. This extends the except clause to retry with a GET request for such scenarios. Classifying this as a bug fix as URLs like used to pass the linkchecking prior to Sphinx 3.2.0 but are now failing as HEAD requests have been enforced (). \/cc"}
{"_id":"q-en-sphinx-doc__sphinx-8481-b80fce57b50b8d752e7ed71cb2a34ff14576436961fe8de5352d0186d189a762","text":"autoattribute could not create document for slots attributes correctly Describe the bug autoattribute could not create document for slots attributes correctly To Reproduce The build succeeded. But docstring is not shown. Expected behavior Document should be generated successfully like when I used . Your project N\/A Screenshots N\/A Environment info OS: Mac Python version: 3.9.0 Sphinx version: HEAD of 3.x Sphinx extensions: Extra tools: Nothing Additional context Nothing"}
{"_id":"q-en-sphinx-doc__sphinx-8506-a2944b75531a87d539352828eefff982124b1addde1357c8b3ea39a140f94654","text":"Sphinx 3.2 complains about option:: syntax that earlier versions accepted Sphinx 3.2 complains about use of the option:: directive that earlier versions accepted without complaint. The QEMU documentation includes this: as part of the documentation of the command line options of one of its programs. Earlier versions of Sphinx were fine with this, but Sphinx 3.2 complains: Sphinx ideally shouldn't change in ways that break the building of documentation that worked in older versions, because this makes it unworkably difficult to have documentation that builds with whatever the Linux distro's sphinx-build is. The error message suggests that Sphinx has a very restrictive idea of what option syntax is; it would be better if it just accepted any string, because not all programs and OSes have option syntax that matches the limited list the error message indicates."}
{"_id":"q-en-sphinx-doc__sphinx-8509-0d851bbbcf4f7f7f11c88bf5d1b61d08c7e15f8d0809c6a99cd981cd9a570dd8","text":"Use uplatex for default latexengine for Japanese docs Is your feature request related to a problem? Please describe. Use uplatex for default latexengine for Japanese docs. Describe the solution you'd like Since v2.3, Sphinx supports uplatex as an alternative of latexengine for Japanese docs (refs: , ). uplatex is able to build a document without conversion character encoding internally. It allows using unicode characters in documents. Additionally, uplatex is compatible with platex (current default latexengine for Japanese docs). Describe alternatives you've considered Nothing. Additional context Nothing."}
{"_id":"q-en-sphinx-doc__sphinx-8539-2e47016cd5ee928366941ca6d585ed4f32fa7cd17cefcca7494300dde832de25","text":"autodoctypehints='description' does not combine well with autoclasscontent='class' With this configuration: Type hints from the method are reflected in descriptions (given explicitly in the docstrings), and are also generated for the class itself. To Reproduce Expected behavior No \"Parameters\" section should be to the class docs if already present for the docs simply because of the type hints. Your project Sample attached. Environment info OS: Ubuntu 18.04 Python version: 3.7, 3.8, 3.9 Sphinx version: 2.4.4 autodoctypehints=\"description\" doesn't use init type hints Describe the bug Type hints attached to the method are not used when , but are used when . To Reproduce Create with these contents: Create with these contents: Generate documentation into an directory: Expected behavior The parameter of the constructor should have a type annotation, like the parameter of does. When , the signature does include the type annotation. Adding causes the type hint to be used, but: believe the type hint should be used even for like it is if , and causes a annotation to be , which is not helpful for a constructor and which doesn't match the behavior of (there's no in that case). Environment info OS: Linux Python version: 3.8.5 Sphinx version: 3.2.1 Sphinx extensions: and Additional context This appears to be intentional behavior as it was the fix for , but I believe it is incorrect because it is inconsistent with how signature type hints are handled."}
{"_id":"q-en-sphinx-doc__sphinx-8548-67351145178a612a4a00e52888cc802aeb00fb8ef26416212ff9e2551840b668","text":"autodoc inherited-members won't work for inherited attributes (data members). autodoc searches for a cached docstring using (namespace, attrname) as search-key, but doesn't check for baseclass-namespace. Bitbucket: Originally reported by: Anonymous Originally created at: 2011-08-02T17:05:58.754"}
{"_id":"q-en-sphinx-doc__sphinx-8551-4f31ae1ae9be8f9912d8c93b2f5a27896eef289a40a699bb82186188d2f4def0","text":":type: and :rtype: gives false ambiguous class lookup warnings Describe the bug The implicit xrefs created by the info fields seems to do lookup differently than explicit xref roles. For unqualified names it seems like they search for the name in every (sub)module instead of in the current module and then parent modules. To Reproduce gives the warnings which refer to the 4 unqualified type names. The . Expected behavior No warnings, and the two mentioned types should resolve to. Environment info Sphinx version: tested both with v3.3 and with master"}
{"_id":"q-en-sphinx-doc__sphinx-8552-80f70b681b417ee591315036bcc45b1a7016da85979553dcd3dc3d00de268ba7","text":"Napoleon: Support NumpyDoc \"Receives\" docstring section Support the \"Receives\" section of NumpyDoc guidelines; it is related to Yields, which is already supported. Receives Explanation of parameters passed to a generator’s .send() method, formatted as for Parameters, above. Since, like for Yields and Returns, a single object is always passed to the method, this may describe either the single parameter, or positional arguments passed as a tuple. If a docstring includes Receives it must also include Yields."}
{"_id":"q-en-sphinx-doc__sphinx-8579-4270a7ed042e3eedddd0f9b4b914545ae028a4d13b7092216fd0081ff93af875","text":"Linkcheck crashes in 3.4.0 Describe the bug When running linkcheck in Weblate docs, it crashes with: To Reproduce Steps to reproduce the behavior: Expected behavior No crash :-) Your project Screenshots CI failure: Environment info OS: Linux Python version: 3.8.6 Sphinx version: 3.4.0 Sphinx extensions: several, but should not be relevant here Extra tools: none involved Additional context Add any other context about the problem here. [e.g. URL or Ticket]"}
{"_id":"q-en-sphinx-doc__sphinx-8593-f1cffa1265c6c243bf5a3238b1343cbfd874551122c5b99f1ee76730c2757afc","text":"autodoc: does not effect to variables Describe the bug autodoc: does not effect to variables. To Reproduce I expect is shown on the built document, but not shown. Expected behavior should be shown on the built document. Your project No Screenshots No Environment info OS: Mac Python version: 3.9.1 Sphinx version: HEAD of 3.x Sphinx extensions: Extra tools: No Additional context No"}
{"_id":"q-en-sphinx-doc__sphinx-8595-825e8fd986f27894028890ab0b930efc336eca4cf98e7b8e7f35c526eace1aa0","text":"autodoc: empty all attribute is ignored Describe the bug autodoc: empty attribute is ignored To Reproduce All foo, bar, and baz are shown. Expected behavior No entries should be shown because is empty. Your project No Screenshots No Environment info OS: Mac Python version: 3.9.1 Sphinx version: HEAD of 3.x Sphinx extensions: Extra tools: No Additional context No"}
{"_id":"q-en-sphinx-doc__sphinx-8611-17361fcedd9154253802bc4b0aa5460d9b3f052ce1737279bbc151acc2147978","text":"Attribute undocumented error for an inherited attribute Describe the bug Sphinx complains about missing documentation for an attribute of an inherited class ( is an attribute of , inherits from ). This showed up with Sphinx 3.4.0 and continues with 3.4.1. I thought this might have been the same issue as , but it has not resolved. To Reproduce Steps to reproduce the behavior: <details<summaryFull session<\/summary<\/detailsExpected behavior I expected no warnings\/errors since the attribute is only inherited. Your project Screenshots n\/a Environment info OS: Ubuntu Linux Python version: CPython 3.8 and 3.9 (others untested) Sphinx version: 3.4.0 and 3.4.1, was ok for <3.4 Sphinx extensions: \"\", \"\", \"\", \"\", \"\", \"sphinxautodoctypehints\", \"sphinxqtdocumentation\", \"sphinxcontribtrio\", Extra tools: n\/a Additional context n\/a"}
{"_id":"q-en-sphinx-doc__sphinx-8620-4ac922aec089d46594d18cdec0e9d47852869dd1c029f46fea19ca429bab96b9","text":"kbd role generates HTML that's difficult\/impossible to style for compound-keystrokes Describe the bug The role doesn't mark container elements with a CSS class, so they can be styled differently that child elements representing actual keystrokes. To Reproduce For the below input the following output is generated: Now we have a problem here: there is no way to differentiate between the parent container element and the child element with CSS selectors! If we provide a CSS style to draw a border around a element, so single keystrokes are displayed correctly, then in the second example we'd get a double border (borders around the child elements, and another border around the parent element). If you want to have borders around single keystrokes, and only around the child elements in compound keystrokes, as shown on the screenshot below, the only way is to differentiate the container and child elements with a CSS class. ! Expected behavior Single keystrokes are fine as they are, no change needed: For compound keystrokes, the container element should be marked with a CSS class (e.g. ) so it can be styled differently than the child elements: Environment info OS: Windows Python version: 3.9.1 Sphinx version: 3.4.0 Sphinx extensions: - Extra tools: -"}
{"_id":"q-en-sphinx-doc__sphinx-8621-840438073a7dfe43200e1ebd152aadcdaf4ee2584f2fbe1496ca3d86579f9f41","text":"kbd role produces incorrect HTML when compound-key separators (-, + or ^) are used as keystrokes Describe the bug The role produces incorrect HTML when: 1) defining standalone keystrokes that use any of the compound-key separators (, and ) 2) defining compound keystrokes where one or more keystrokes use any of the compound-key separators (, and ) To Reproduce For the below three keyboard definitions: The following three incorrect output is generated: (1) is treated as a separator with two \"blank\" keystrokes around it. (2) is treated as a separator with two \"blank\" keystrokes around it. (3) is treated as a separator within a compound-keystroke, with two \"blank\" keystrokes around it. Expected behavior For single keystrokes that use , or, just a single element should be created. For compound-keystrokes, the algorithm should differentiate between , and characters appearing in separator vs keystroke positions (currently, it's very simplistic, it just treats all these characters as separators using a simple regexp). Screenshot ! Environment info OS: Windows Python version: 3.9.1 Sphinx version: 3.4.0 Sphinx extensions: - Extra tools: -"}
{"_id":"q-en-sphinx-doc__sphinx-8627-739a5546dbf1eaa3b1dec17653fe958c3f10049a9262b3badd16401a6696ab02","text":"autodoc isn't able to resolve struct.Struct type annotations Describe the bug If is declared in any type annotations, I get To Reproduce Simple Simple Command line: Expected behavior If you comment out the 2 functions that have type annotations, you'll see that resolves fine and shows up in the resulting documentation. I'd expect that would also resolve correctly. Your project n\/a Screenshots n\/a Environment info OS: Ubuntu 18.04, 20.04 Python version: 3.8.2 Sphinx version: 3.2.1 Sphinx extensions: '', '', '', '', '' Extra tools: Additional context [e.g. URL or Ticket]"}
{"_id":"q-en-sphinx-doc__sphinx-8633-0d7c025e1d6f3a7b0826978ef505e4cb9582ea38528985bd138ec1df75d807c8","text":"building psf\/requests documentation fails with AttributeError: 'LookupDict' object has no attribute 'name' Describe the bug Since sphinx 3.4.1 building documentation fails with the following: Attached the full log: The issue is not reproducible using sphinx 3.3.1. To Reproduce In addiction to build requests' documentation I managed to reproduce the issue in the interactive python console using the following: I'm able to reproduce the issue in a clean virtualenv with only requests and sphinx installed. Thanks!"}
{"_id":"q-en-sphinx-doc__sphinx-8638-a17fe71da52e1d4befa3b65a7acde2014f2e29b2c401b9449e4874e9e71c3923","text":"Instance variables link to other variables of the same name in the project Describe the bug Assume autodoc is used via apidoc. In theory other combinations of autodoc (or maybe even without it) can cause this to occur, but this would be the most common occurrence. If a global variable (or really, any kind of variable, just that this would be the most common occurrence) exists, and inside a class you decide to document a variable of the same name, the document of the instance variable will link to the other occurence of a variable under the same name. This can even occur across subpackages and even across other classes of those subpackages (although this appears to occur less often and seemingly...randomly? This only occurs sometimes (presumably due to the referencing heuristic?)). This is a problem, because, for example, could be and usually is completely unrelated to . Furthermore, (instance variable) could be completely unrelated to (global variable). Of course this latter example is far less likely, but the autolinking of these two together, is strange. To Reproduce Steps to reproduce the behavior: then open 127.0.0.1:8008 in a browser Expected behavior That the class variable documentation not be linked to any other. It is unreasonable to expect these to be in any way related whatsoever. If they happen to be, the user can decide to document it as such with a simple reference to the other variable, such as \"see :const:\\\". There is no reason that a variable on some class of some database-oriented subpackage autolink to the variable on some class of some config-related subpackage (this is what occurred in my codebase, which is private at least while in development. I cannot provide anything except a heavily censored screenshot, as I do not know of a way to trick the referencing heuristic to cause a link to occur in an demo repo). Your project Screenshots Not really applicable because this is example independent but here you go anyway: ! Environment info OS: Ubuntu 14.04.5 (probably irrelevant) Python version: 2.7.6 (probably irrelevant) Sphinx version: 1.8.3 Sphinx extensions: autodoc, intersphinx, and other (probably irrelevant) extensions (todo, viewcode, githubpages in the demo repo, among others in the private repo) Extra tools: Any Browser, sphinx-apidoc"}
{"_id":"q-en-sphinx-doc__sphinx-8658-b0fefc0363d955c16c97108eb4e35877aa22d847768ad1b82823cf8c62eb037c","text":"Napoleon: more custom docstring section styles Although the option help renders custom docstring section, the style is inconsistent with the rest of the doc. For example, I have a custom docstring section . I would like it to be displayed as or docstring section. However, option rendesr in a different style shown in the following picture. ! It will be really helpful if we can customize the custom sections a bit more. The following setting has a similar effect, but it renders the Parameters name instead of the custom name. I would like to do something like the following so that my Custom section has the same style as the Parameter section, and it still keeps my custom name: or The following link includes more details about the solutions: Others people have expressed a similar desire (see sphinx-contrib\/napoleon) If you are interested, I would like to provide a PR for this. Thanks!"}
{"_id":"q-en-sphinx-doc__sphinx-8674-1bfb8ff72d6cb00bb5e2f88dbc92c309d7a24b39e1cac27cd0b572eb963b665a","text":"Enable passing options to pygments Hi, Related to issue 175 (and discussed on #python-docs), it would be nice to have a way of passing options to pygments. This is essential for small snippets of PHP (the :startsinline: option) - there are probably other cases too. Default options might go in , with sourcecode block options allowed. Thanks Bitbucket: Originally reported by: Paul Biggar Originally created at: 2009-06-23T19:02:39.208"}
{"_id":"q-en-sphinx-doc__sphinx-8679-b46340cef5fb1c046b91567f088f8b49e85d3fcbf8de9f3a38e26f7dee56d3ee","text":"reference pointing at a \"rubric\" directive containing a part breaks Describe the bug The reference text is broken when trying to display a reference pointing at a \"rubric\" directive containing a part that is. However, a title with a quoted text is rightfully displayed by a reference Will result in: ! To Reproduce Just use a directive with a quoted text in it, as shown below: from Expected behavior Text displayed by the reference should be displayed entirely Your project Environment info OS: alpine linux Python version: 3.7 Sphinx version: v3.4.1"}
{"_id":"q-en-sphinx-doc__sphinx-8684-398e77e40f5c5028e57298d3555947baf14677e382d8a495804f52739e72369c","text":"ROI: extend dedent for code-block and literalinclude Hi, I think it would be nice to let behave like the corresponding Python one: It may just that if nod edent value is provided, then a full dedent is performed."}
{"_id":"q-en-sphinx-doc__sphinx-8697-1cae9bb1e5210ddc7668a3c19d7448e43ced61d6d04fc8b8cfa9e4cb156b101d","text":"The reference to the same file is interpreted as a duplicate Tested on Windows 7 64bit machine, Python 2.7.3, Sphinx 1.1.3 If there is reference in various ways to the same resource, it is interpreted as different file with duplicate name. If there is a files structure as follows: And we have the following code in Then during the build of html output we will have three files (while only one is expected): The same issue is with figure directive. In attachment there is a simple Sphinx project just to illustrate the issue. IMO the problem is because all paths in Sphinx code are not normalized (() function is missing). Bitbucket: Originally reported by: Originally created at: 2013-02-18T14:47:34.934"}
{"_id":"q-en-sphinx-doc__sphinx-8707-be4b4db5128b8887e19f28279e8d44777d9e6772409aa18590673e87cf4d1d50","text":"viewcode does not work when Describe the bug viewcode does not work when To Reproduce Expected behavior Always enabled for HTML builds (except singlehtml and epub) Your project No Screenshots No Environment info OS: Mac Python version: 3.9.1 Sphinx version: HEAD of 3.x Sphinx extensions: Extra tools: No Additional context No"}
{"_id":"q-en-sphinx-doc__sphinx-8713-0338a0b1d60ef42299fa8f1f92c1fa7b3e8740cc54e138c22eb4fb6a4063e663","text":"napoleonuseparam should also affect \"other parameters\" section Subject: napoleonuseparam should also affect \"other parameters\" section Currently, napoleon always renders the Other parameters section as if napoleonuseparam was False, see source whereas it would make sense that this section should follow the same formatting rules as the Parameters section. Note the difference in rendering. See above. Alternatively another separate config value could be introduced, but that seems a bit overkill. N\/A OS: Linux Python version: 3.7 Sphinx version: 1.8.1"}
{"_id":"q-en-sphinx-doc__sphinx-8719-4b3b9aa5cb17c88a8c5938b83af607903f50c5ccaed374d31ce1097444825c3b","text":":kbd: role with Caps Lock rendered incorrectly Describe the bug Caps Lock` produces: Expected behavior Environment info OS: Debian unstable Python version: 3.9.1 Sphinx version:"}
{"_id":"q-en-sphinx-doc__sphinx-8721-56d29fc8fede81030ff2b174d120c930c1ef3c9aeec56ec0db99cdd17a6ac2ef","text":"viewcode creates pages for epub even if on Describe the bug viewcode creates pages for epub even if on To Reproduce Expected behavior module pages should not be created for epub by default. Your project No Screenshots No Environment info OS: Mac Python version: 3.9.1 Sphinx version: HEAD of 3.x Sphinx extensions: Extra tools: No Additional context No"}
{"_id":"q-en-sphinx-doc__sphinx-8729-1b853cf0d417b701226d371c600473748800aa9a6388e342f44fd91b86203484","text":"sphinx-apidoc on namespaces with only subpackages\/namespaces Describe the bug --implicit-namespaces allows us to use namespaces in our project tree. At the current deployed implementation (3.4.3) it makes a difference if i have a submodule in my modulepath (root python namespace) or only subpackages\/namespaces. Running with a submodule (some python file) it will choose the modulepath as the one and only module otherwise, without a submodule, it will use the subpackages\/namespaces as individual python modules. The difference in the output is, that the of the modulepath is not created and does contain all subpackages\/namespaces as submodules and not only the root module. I would recommend to introduce an option to not search recursively for modules, or disable the recursive search if implicit-namespaces are activated. To Reproduce Steps to reproduce the behavior: Expected behavior Expecting same result in apidoc_out2, with only one module inside Instead is missing and is different. Your project Environment info OS: Win Python version: 3.8.0 Sphinx version: 3.4.3"}
{"_id":"q-en-sphinx-doc__sphinx-8731-8e7c1ae8041ecd5a289a870602255629e0a190418c88f6c5f856aad6da44c813","text":"Return type for Yield attribute not properly hyperlinked in apidocs Describe the bug Typically when one defines a return value in a method they will use the following syntax: When doing so the return type, in this case, is hyperlinked to the api docs for the appropriate data type (ie: using interphinx). However, when writing doc strings for generators using the same \/ similar syntax provided by the Yield attribute, the data type associated with the generator is not appropriately decorated. Taking the previous example code snippet and making a small modification as show below is sufficient to demonstrate the change in behavior: The HTML markup produced by this code snippet simply format the data type using custom font formatting based on the theme, but it fails to associate \/ generate the hyperlink for the API docs for the data type as is done by the Return attribute. NOTE: I am using the apidocs Sphinx extension to generate the API docs for my projects in case that has any impact on this behavior. To Reproduce Steps to reproduce the behavior: Expected behavior The data type should be decorated \/ marked up the same way it appears when using the Return attribute when using the Yield attribute Your project Screenshots Example output generated on ReadTheDocs: Environment info OS: MacOS Python version: 3.8.0 Sphinx version: 3.1.2 Sphinx extensions: , , Extra tools: any web browser (safari, chrome, firefox, IE) Additional context (I thought this might be a bug with the napoleon extension so I reported it here first, but they recommended I report the issue here instead)"}
{"_id":"q-en-sphinx-doc__sphinx-8771-bad54fd9607782020a36592bdcdce77cf81baea4dea77c08e5bed4ec99be2e6f","text":"Keeping original signatures for functions\/methods I'm using Sphinx's autodoc feature to document my API. Example: The generated documentation now shows the following signature: It would be nice if there was a way to tell Sphinx to keep the name of the constant value, i.e. At the moment the only alternative is to write all signature by hand again. Bitbucket: Originally reported by: Sebastian Rahlf Originally created at: 2011-08-31T11:29:50.797"}
{"_id":"q-en-sphinx-doc__sphinx-8801-cd035194d84e3587bee4a736b717498f79acb91142ac16fa33b40cb2d54a9ae7","text":"autodoc: The annotation only member in superclass is treated as \"undocumented\" Describe the bug autodoc: The annotation only member in superclass is treated as \"undocumented\". To Reproduce is not documented. It will be shown if I give option to the autoclass directive call. It seems the attribute is treated as undocumented. Expected behavior It should be shown. Your project No Screenshots No Environment info OS: Mac Python version: 3.9.1 Sphinx version: HEAD of 3.x Sphinx extensions: Extra tools: No Additional context No"}
{"_id":"q-en-sphinx-doc__sphinx-8863-354c267ad2285a8758d32d6f588bf7132fd54c83cf4e547f3624c76969a263d7","text":"[LaTex] code-block printed out of margin this is not handled correctly currently: long hex strings ! code: ` link to rst file: Originally posted by in"}
{"_id":"q-en-sphinx-doc__sphinx-8951-9fb90abbb317836fe3610567f676f0ca992e8451555afcad377bfd7485e84ce6","text":"Support C++20's spaceship operator C++20 has the new spaceship operator for three way comparisons which can also be (not sure if the latter matters to Sphinx). I hope this is a problem with Sphinx at all and not with the Breathe extension I'm also using. For example, this reST code in Sphinx currently leads to the following warning\/error"}
{"_id":"q-en-sphinx-doc__sphinx-8969-43e4720c7033120fa68e47e6018c51b3262d7392fd02b2db5c7da13c5d1efb64","text":"Absolute\/\"source-relative\" imports for csv-table :file: Describe the bug To be honest, I am not sure if this can be considered a bug, or if it is rather a feature request. Sorry about that. When using the directive, the use of with absolute paths are really absolute, unlike with (eg) the directive, where absolute paths are treated relative to the source directory (herein called \"source-relative\"). I do understand that there is a difference in the 2 cases, because with the path is not specified in . Yet, I do not see a possibility to mimic this behavior in the directive. To Reproduce A file in : Relative imports: are treated, as expected, relative to the file: Absolute imports: are treated, opposed to my expectations, like real absolute paths: and not like relative-to-source paths. Expected behavior I would expect this to work like absolute paths in the (eg) directive. But as stated in the beginning, probably I am wrong with my expectation, and this should be a feature request to add an option to use \"source-relative\" paths with the directive. Environment info OS: Win Python version: 3.8.5 Sphinx version: 3.2.1"}
{"_id":"q-en-sphinx-doc__sphinx-9015-d9f5ab42b647729c13a7b02d1d0c242897dcc8c1ec323dd473dc3449c3769086","text":"Property gets linked instead of external type in class with inline type annotations Describe the bug The project is using autodoc with inline Python type annotations. Given the following class (snipped to the relevant parts): The class and method get rendered : ! In the class, the gets rendered as and is not linked. In the method, the gets rendered as and links to the property. Expected behavior I expect it to show as . A is also OK I guess (if it's meant to refer to a \"traceback object\"), but in that case it should link to or instead of the local property. To Reproduce I can try to create a minimal reproduction if the above is not sufficient or gets outdated. Your project Environment info OS: Linux Python version: 3.9.2 Sphinx version: 3.5.2 Sphinx extensions:"}
{"_id":"q-en-sphinx-doc__sphinx-9104-fd536a80cee912825ad550212ec3140ef86added606923cca2c2b9b98a6299c8","text":"WARNING: py:class reference target not found: .. Describe the bug I think this might be referring to the ellipsis but I'm not sure at present. Note that I've already tracked the other warnings (outcomes and event type) to other issues. This time I did remember to try latest 3.x (). It showed the same issue. To Reproduce Steps to reproduce the behavior: <details<summarySteps with output including complete traceback<\/summary<\/detailsExpected behavior I'm not completely sure but at least a reference to a specific thing in my code. If it is referring to the ellipsis then perhaps it just shouldn't have an issue at all. I'm guessing it should just be a plain in the output. It could link to but that doesn't really describe it's meaning in typing and the generic type for which it is a parameter would already be linked. Your project Environment info OS: Ubuntu Linux 20.04 Python version: 3.8.3 Sphinx version: 3.2.0 Sphinx extensions: sphinxqtdocumentation sphinxcontribtrio"}
{"_id":"q-en-sphinx-doc__sphinx-9128-460658f0707893f3452c0eb42842a05584c128e3389812341dcf35f15311819c","text":"autodoc: duplication warning on documenting aliased object Describe the bug autodoc: duplication warning on documenting aliased object To Reproduce Expected behavior No warning Your project N\/A Screenshots N\/A Environment info OS: Mac Python version: 3.9.4 Sphinx version: HEAD of 4.0.x Sphinx extensions: Extra tools: No Additional context No"}
{"_id":"q-en-sphinx-doc__sphinx-9171-458022b06d9f7b07ab5bbbf37453a3dc105926f6e96bf04c6ebdc20036dfedcb","text":"Omitting constructor signature from class header using I'm a bit surprised to see that (in the Python domain) will create a heading like which seems quite unintuitive to me, as it mixes two concepts: the declaration of \"class Foo\", with a constructor call \"Foo(args)\". How can I suppress the (automatic) addition of the constructor signature to the generated heading ? I actually want to document the constructor with an additional nested directive, and I also have an overloaded method, making the above all the more confusing. Am I missing or misunderstanding something ?"}
{"_id":"q-en-sphinx-doc__sphinx-9180-e7ed110ed54d95978466fdeed4b4d9c60b2c35f64f1262457c7bed267873d830","text":"Special member is not documented with automodule Describe the bug If I have a module which has a special attribute (starting with a double underscore) it will not be documented by automodule even if it's specified in :special-members:. According to my research it's because it's not identified as an attribute. If a non-special attribute is used, everything works fine. To Reproduce To execute the build I use Expected behavior The attribute should be documented. Environment info OS: Linux Python version: 3.7.1 Sphinx version: 3.5.4"}
{"_id":"q-en-sphinx-doc__sphinx-9207-453f436fab5c5b89e76939b3bd0149ca05f0c61a859a8741619c249f81f6edef","text":"Sphinx 4.0 finds more than one target for cross-reference Hello everyone, Describe the bug After upgrading to 4.0.0 we have discovered that existing code is raising a warning with . In 3.5.4 we did not see this warning To Reproduce I have set up a minimal reproduction repository here: Steps to reproduce the behavior: Output of the sphinx-build command: Expected behavior no warning appears Your project Environment info OS: Linux Python version: 3.9.4 Sphinx version: 4.0.0 Sphinx extensions: Additional context I have bisected the issue to be introduced by Thanks verry much"}
{"_id":"q-en-sphinx-doc__sphinx-9229-b0e298bcfb634fcea297e4e80c963c8c0be9351b25d863580804dce7afbef7f7","text":"Inconsistent behaviour with type alias documentation (not overwriting all the default messages, just some) Describe the bug Hello, I have 3 muiltiline docstrings for type aliases (using the next-line documentation syntax). For 1 one them the docstring is correctly shown in the rendered HTML, but for 2 of them, the docstrings are ignored and the only thing shown is the text. I suppose this is related to , but I might be doing something wrong here (so if you could point me out in the correct direction that would be very good). To Reproduce The following is a reduced example of something happening in : a directory with : I run: the path adjustment in to the generated , and to the toctree in . opening in the browser should show the reported inconsistency. Expected behavior The docs should show the contents in the docstrings for all the type aliases instead of the the default text. Your project Screenshots ! Environment info OS: Win10 WSL: Python version: 3.6.9 Sphinx version: 3.1.2 Sphinx extensions: Additional context Possibly related to"}
{"_id":"q-en-sphinx-doc__sphinx-9230-9390548fb49e38024a125bdcc76435e18e3a5527c6882396aabe445400da4aca","text":"Doc rendering is incorrect when :param has datatype dict(str,str) Describe the bug I have a parameter defined under docstring of a method as:- :param dict(str, str) opcmeta: (optional) Which is being incorrectly rendered in the generated docs as:- str) opcmeta (dict(str,) –(optional) To Reproduce Create any method with the docstring containg the above param Expected behavior The param should be rendered in the generated docs as:- opc_meta (dict(str,str)) – (optional) Your project Screenshots <img width=\"612\" alt=\"Screen Shot 2021-05-12 at 12 30 50 PM\" src=\"https:\/\/user-\"<img width=\"681\" alt=\"Screen Shot 2021-05-12 at 12 32 25 PM\" src=\"https:\/\/user-\"Environment info OS: Mac Python version: 3.9.0 Sphinx version: 4.0.1 Sphinx extensions: [\"\", \"\", \"\", \"autodocsumm\"] Extra tools: Browser Firefox. Additional context N\/A"}
{"_id":"q-en-sphinx-doc__sphinx-9231-bbcc167ec428b714c1ccb73e1b3789dad6e270b0a9a5c93d6c5dc44a0aeeda07","text":"manmakesectiondirectory should not be enabled by default Enabling by default in breaks projects relying on the previous behavior. This is a serious problem for Linux distributions that will end up with misplaced and unusable man pages. Please consider keeping it disabled by default; the benefit of being able to use MANPATH in the output directory does not justify this kind of breakage. I also noticed that the current implementation generates paths like instead of . Only the latter can be used with MANPATH which appears to be the main motivation behind . Examples of breakage I've seen so far (and we've only had sphinx 4.0.x in Arch Linux for three days): does not expect the section subdirectory and results in man pages for built-in shell commands being installed to instead of and also fails to filter out , and which are meant to be installed to . copies the output directory to resulting in paths like (note the additional directory)."}
{"_id":"q-en-sphinx-doc__sphinx-9233-f5742a07aa81ceb45dfe3e7eb4c8b0d47ae70aa6a61a76a910227de8200cb83c","text":"New hook to customize base list I would like to change the formatting of the base list for classes. Specifially I would like to provide information about parameterized types (e.g. ]). See agronholm\/sphinx-autodoc-typehints for how I want to use it. For that I need a new hook\/event similar to the existing . I propose the signature . The first five arguments are exactly the same as for the existing events. The last one is the list of formatted strings generated in for the bases. It can be modified to change what is output. Alternatively, to provide even more freedom, the hook can return a string. This string is then inserted instead of the \"Bases: ...\" line. I can provide an implementation in a pull request, if you like the idea."}
{"_id":"q-en-sphinx-doc__sphinx-9234-14c91cd38a72b80385b556d41739dd033288bdecf3e797af074126ba1c2eeb7a","text":"Link checker should be able to prohibit unknown redirects Is your feature request related to a problem? Please describe. A lot of links become stale or move. Good websites will provide redirects to the correct new location or return an HTTP error code. Bad websites will redirect to an unrelated page or the root of the website. Preventing all redirects does not allow links to URLs like which redirects to It needs to be possible to allow these redirects but disallow others. Describe the solution you'd like It should be possible to prohibit unknown redirects by listing all of the allowed redirects as pairs of URLs. Describe alternatives you've considered Post-process by removing filenames and line numbers then sorting it and comparing it with known good output. Additional context A link to (which used to work) now redirects to Linkcheck allows this but the original link is not valid and needs to be updated to the article's new URL of Linkcheck should be able to report an error for this redirect."}
{"_id":"q-en-sphinx-doc__sphinx-9246-9eae30e2f75b9e639a2e38b1a6b440305db0a7e3bb3e839432b888d9f6dcd545","text":"Unknown node: : : : Run: OS: Linux Python version: 3.9.4 Sphinx version: 4.0.1 Sphinx extensions: , , This is similar to the issue that was fixed in . I think, that I've tracked the issue to the plugin. This plugin connects itself to events like this however, their implementation doesn't have the snippet, which eventually leads to the above error. I could submit this as a bug report to the author of , but I wanted to make sure, that this behaviour is intended. It seems to me like a potential antipattern. For example, the above snippet (with minor variations) is already duplicated internally in - - And any plugin that connects to the event must also now add this snippet (which is complicated by the fact that doesn't seem to be available in ). Do you think that maybe the value should be resolved when the event is created in ?"}
{"_id":"q-en-sphinx-doc__sphinx-9258-8fa5e840592e038343241d0f76ffeb1cd62be373cbc08917c7dafe6ff4f24f64","text":"[RFE] Support union types specification using | (vertical bar\/pipe) Please add a support for specifying multiple types acceptable for a parameter\/attribute\/variable. Use case: Imagine that there is a function that accepts both and . The docstring would look like: Such a syntax is already supported by e.g. ."}
{"_id":"q-en-sphinx-doc__sphinx-9260-df2065b9581260d1c2620ad040cf08e2c4146a21d31c9d5fd026512b83319d17","text":"Linkchecker croaks on specific anchors of GitHub-rendered reStructuredText documents Dear Sphinx developers, first things first: Thanks a stack for your paramount work on Sphinx. You already saved many souls of people writing technical documentation and probably also beyond this audience. We just observed a minor woe with Sphinx' linkchecker we wanted to share with you. We really like that the linkchecker is able to check anchors within HTML documents as contributed by on behalf of . With kind regards, Andreas. Describe the bug We had the link [1] in our documentation, and, maybe after upgrading to more recent versions of Sphinx, the linkchecker suddenly started croaking on that. After changing it to [2], it worked again. When inspecting the source code of the respective HTML page, you can clearly see that the anchor name defined by is technically correct. However, it apparently has worked before by referencing . So, we are wondering if something changed on GitHub's reStructuredText renderer or even Browsers interpreting the HTML link anchors differently. When invoking those links [1,2] in the Browser, actually both work, including navigation to the appropriate place within the page. Funny, hm? [1] [2] Expected behavior Technically, from the perspective we know our way around HTML, the behavior is probably the right thing and correct. However, as we can see, something might have been changed on the HTML standard that Browsers are capable of interpreting different styles of defining link anchors. So, it might be worth to revisit this space and maybe improve the linkchecker implementation on those aspects. Environment info OS: Linux Python version: 3.9.2 Sphinx version: 3.5.2 Firefox: 86.0"}
{"_id":"q-en-sphinx-doc__sphinx-9261-0aad80c3c27c3a8df632dac29823e6bb95823a0b3aa5bb595de94d96a9f43f82","text":"Building autodocs for derived classes produces invalid warnings Describe the bug Generating API docs using the autodoc extensions, when a base class has doc strings in the constructor as well as in the class definition, and the autoclasscontent feature is set to \"both\", produces superfluous warnings when parsing derived classes with overloaded constructors with no doc strings on them. To Reproduce a base class with a constructor, and provide both a class doc string and a constructor doc string, something like this: a derived class that has an overloaded constructor, with no doc string, something like this: the auto content option in Sphinx to combine the class doc string and constructor doc strings together (ie: in the script) to generate the docs using the apidoc extension (ie: when using , just run ) Expected behavior Expected behavior: the API docs for both the base and derived classes should be generated without warnings. Actual behavior: docs for the base class generate correctly, but docs for the derived class produce the following warning \"docstring of sample.MyDerived: WARNING: Unexpected indentation.\" Environment info OS: MacOS 11.3.1 (reproducible on several other Linux \/ Mac systems) Python version: 3.6.8 (reproducible on several newer versions as well) Sphinx version: 4.0.1 (reproducible on versions as far back as v2.2.0) Sphinx extensions: , Extra tools: (optional)"}
{"_id":"q-en-sphinx-doc__sphinx-9281-ef9b990d0a134a5d1556f05f20362a02e445504d1b99eac529e951869ae49bc4","text":"Python Enum values (used to show default values in function signatures) are rendered ugly. Python Enum values (used to show default values in function signatures) are rendered ugly. To Reproduce I made a minimal example to show the issue: Expected behavior I would hope the signature rendered as: uglyenumfunc(e: uglyenum.MyEnum = MyEnum.ValueA) → None Unfortunately, it renders as: uglyenumfunc(e: ugly_enum.MyEnum = <MyEnum.ValueA: 10) → None Environment info Python version: 3.9.5 Sphinx version: 4.0.2 Sphinx extensions: autodoc"}
{"_id":"q-en-sphinx-doc__sphinx-9289-f48c8fb3377aa98aeb78d2016c801fa965ea5a2b342f6d228edcc6b80b529eca","text":"pythonuseunqualifiedtypenames does not work on function descriptions Describe the bug When combining the new configuration with , the type names are not shortened even though links are created. To Reproduce Here's a minimal project using Sphinx 4.0.2 on Python 3.9: This produces documentation looking like this: ! Expected behavior I'd expect the link to be shortened to like it is when ."}
{"_id":"q-en-sphinx-doc__sphinx-9309-72b2d0e4e2a0412f1fd8b6bc43aec7ab5599538fae0ff5156383921ba5732fad","text":"Linkcheck Reports Broken Link when Remote Server Closes on HEAD Request Describe the bug Running on a document that contains an external link to a website may report the link is broken when a web browser may successfully open the link. Specifically, if the website closes its connection when receiving the request method, then will receive a exception, which bypasses the logic that would otherwise have it make an request. A specific example of a website exhibiting this behaviour is To Reproduce Steps to reproduce the behaviour: Observe linkcheck reporting a broken link: Open in your browser Click \"a link to the US Patent Website\" Observe the link opening and rendering normally Expected behavior If a link is valid, and the website is returning valid content for an request, should not report the link as . Internally, in , if a call to raises a exception, it should attempt a just like it does with and . Your project Environment info OS: (but this does not appear to be OS-dependent) Python version: and Sphinx version: and Sphinx extensions: none Extra tools: any common web browser Additional context Using , we can see that this particular website closes connections when receiving : Observe that this returns the expected HTML content Observe that this fails to return any content from the server"}
{"_id":"q-en-sphinx-doc__sphinx-9320-54cd5ff2f0c8e3099f98135ad9d2b931a54f71b04e775ae7601bafec655314d6","text":"with existing doesn't exit easily Describe the bug I've attached a screenshot in the screenshots section which I think explains the bug better. I'm running in a folder with a already existing. It says \"Please enter a new root path name (or just Enter to exit)\". However, upon pressing 'Enter' it returns an error message \"Please enter a valid path name\". To Reproduce Steps to reproduce the behavior: Expected behavior After pressing Enter, sphinx-quickstart exits. Your project n\/a Screenshots ! I press Enter for the first prompt. Environment info OS: Ubuntu 20.04 Python version: Python 3.8.5 Sphinx version: sphinx-build 3.2.1 Sphinx extensions: none Extra tools: none Additional context I had a quick search but couldn't find any similar existing issues. Sorry if this is a duplicate."}
{"_id":"q-en-sphinx-doc__sphinx-9350-042a6262ea69cb472fd9637fefb4cea1115663ec0e265de631ea53dcef24bff5","text":"Man page using :samp: with braces - font doesn't reset There are issues with the man page rendering when using :samp: The font change doesn't reset after the closure of the samp but runs onto subsequent lines until certain 'resetting' elements are encountered, such as a bullet, heading and probably others I haven't tested for. :samp: is ok. HTML and Latex (pdf) formats are also unaffected by this issue. MWE included with screen shot. Text that should be plain. A bullet that should be plain. And back to normal. :samp: More text that should be plain. A bullet that should be plain. And back to normal. :samp: All ok on this line. !"}
{"_id":"q-en-sphinx-doc__sphinx-9367-626fbc93dc23198cd9d47ef543c88d5e55539275d338090b80f58d52ddd20709","text":"1-element tuple rendered incorrectly Describe the bug This is a followup to which has been addressed in . However the special case of a 1-element tuple is still not handled correctly. is rendered as , but should keep the trailing comma. To Reproduce Add a testcase at"}
{"_id":"q-en-sphinx-doc__sphinx-9386-9ba409880b108edc7a4f9042ec4f306add460dd6455bdc5fc295556e693a66b0","text":"Autodoc still using typehint types in properties Describe the bug Despite being set to , properties have type hints rendered in the output. To Reproduce Steps to reproduce the behavior. Download the zip below and go to then run Expected behavior No type hints in the output. Your project Screenshots ! Environment info OS: Windows 10 Python version: 3.9.2 Sphinx version: 4.0.2 Sphinx extensions: Extra tools: N\/A Additional context Issue discovered in a real project (edit: I removed it using CSS)."}
{"_id":"q-en-sphinx-doc__sphinx-9459-31e9d25af8dbb63230a11b2320571c7376d0c8ce51e9e7f8cdb05de44ee0ce52","text":"Add an option to not use intersphinx references as a fallback That intersphinx references are used as fallbacks if a reference is not available in a project recently caused us a problem. We have a site () which is the latest build of . We use intersphinx as follows, so we can link to the latest documentation in some sections: Our docs included a label () so this reference was available on . This label was removed on a branch, along with one of the two references to this label. In this case we would expect that building the branch would fail, as there is a reference to a label which doesn't exist. Unexpectedly, the branch built successfully, because the label was found in as a fall back when it was not found locally. The branch was merged, and then later, when changed (because was built again) builds stopped working because there was a reference to a non-existent label. It would be good to have an option to not fall back, maybe something like ."}
{"_id":"q-en-sphinx-doc__sphinx-9461-32d0e328e4be967ae766d6273014e152857e9da31643826001739ffdaf7e2daf","text":"Methods decorated with and do not get documented. EDIT: The problem seems to be that returns , thus sphinx can just lookup . However, returns the type of the returned object, since essentially, a ends up behaving like a class attribute. So Sphinx doesn't really have a chance to extract the docstring. EDIT 2: Seems like this will get fixed in python 3.10, cf. I will try to test this with the beta release. That is, since python version 3.9 we can write code like However, sphinx does not seem to document any such methods (regular decorated methods get documented just fine.) The following methods were erroneously not documented: - - - Methods that are decorated with both and should be documented appropriately. No response Ubuntu 20.04.2 LTS 6 3 , No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-9464-d9d0e5efd71a449247084de49f946ea3eb20ed28924ca17a1821f25b7d62350c","text":"PEP 585 type hints not rendered correctly If you use a PEP 585 generic as an annotation i.e. , autodoc renders the annotation as rather than , this behaviour differs from using which renders as expected. Fixing this is quite simple as far as I can tell, just needs to check if the annotation has and if it does, return An annotation of to be rendered as ! Mac Python 3.9.3 1 autodoc No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-9467-042699b9b4269573b2d5ebb25015fe8cfe426b232251f8e4ce77cfa4216eb0a7","text":"github linkcheck anchor change in 4.1.0 break some usage Given a link like: in a github doc, with release 4.1.0 this will fail with linkcheck, while previously it worked. It passes. No response Linux 5 0 No response No response The relevant change happened in Failing test logs:"}
{"_id":"q-en-sphinx-doc__sphinx-9547-e3e8ef784f530a10daff86fafc22f90bbead3b775e6fb2b0f944f0062a176b36","text":"Missing support for GNU extension types like or I'm currently working on the transition of the GNU C compiler (GCC) manuals and I noticed there are unsupported C extensions like: where I see the following parsing error: Right now, there's some special casing for e.g. 'unsigned' type: One possible fix is adding the mentioned C extension handling for the following types: or I can see a domain parser can become public via an API entry point: What do you think?"}
{"_id":"q-en-sphinx-doc__sphinx-9591-1456deda96cf56e99dd8717fba7701f953a6df91c9c06f14be27ed48983a38ca","text":"Cross-references don't work in property's type annotations A documented type in property's type annotation does not get cross-referenced: I expected the documented type in property's type annotation to be cross-referenced. Here's a link to the generated docs: https:\/\/sphinx-issue- Windows 10, Ubuntu 18.04 7, 3.8, 3.9 2 No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-9602-73b09a45bd93f81136984b20e94bf89e2d1ca04d5eee2dd4857f433960926889","text":"Nitpick flags Literal annotation values as missing py:class When a value is present in a type annotation as , sphinx will treat the value as a . With nitpick enabled, values like end up failing, because is not a class. This is a problem for builds which want to use to catch doc errors. Setup a simple function which uses Literal, then attempt to autodoc it. e.g. I've pushed an example to . Just run with available to see the failing build. (or whatever literal value) should be present in the type annotation but should not trigger the nitpick warning. No response Linux 8, 3.9 2 autodoc No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-9654-e3399d60435535ce47b914acf4275574190615b0afc8b7b750ed821b3af15e81","text":"Incorrect base class detection I have a chain of subclasses, like A -B -C -D. When I document class D, it lists the base class as A instead of C. I would expect the base class to be the direct super class. No response macOS 10.15.7, Linux (whatever RtD uses) 11 1, 4.0.2 No response I don't think this is the same as because I can reproduce it with 4.0.1 too. Could be related to . For a class like , even though the super class is , the base class is listed as ."}
{"_id":"q-en-sphinx-doc__sphinx-9658-d8653d6405ff07eed702bce51c9f067ab758f946ef71427acbb2823cb8198a95","text":"Inherited classes not correctly documented when mocked We're experiencing an issue when documenting classes that inherit mocked classes. However, classes which inherit other classes from our own package are ok. This issue appears to be dependent on the version: : Everything is OK. : Classes that inherit mocked classes are not documented. (see ). This is fixed in . : The previously missing classes are now documented, but there is a problem with the \"Bases\" section in the docs. Example: In the docs for in this readthedocs build https:\/\/seldon--, the base class is listed as \"Bases: \" instead of \"Bases: \". The \"Bases\" section should report not . i.e. see https:\/\/seldon-- ! ! Ubuntu 18.04 (used by readthedocs\/build:6.0) 11 [\"\", \"\", \"\", \"\", \"\", \"\", \"\", \"\", \"\", \"sphinxautodoctypehints\", \"\", \"nbsphinx\", \"nbsphinxlink\", \"mystparser\"] No response demo PR: readthedocs demo build: https:\/\/seldon--"}
{"_id":"q-en-sphinx-doc__sphinx-9665-8ead2b3b9486d7f688650a6901148e712e4350d0faeb6f64212f42d25bf3f94a","text":"python bases in packages - follow up on Is your feature request related to a problem? Please describe. Same problem as described in Describe the solution you'd like sphinx checks if the base class is documented somewhere. If yes, it inserts the reference to the corresponding page automatically. Describe alternatives you've considered As originally proposed in , allow to return a list of strings that will be appended to Additional context was marked as but that PR unfortunately doesn't solve the problem: While I now can return a different base class, I still can't control the text of the hyperlink for that class. With being allowed to return strings, one could just return something like telegram.TelegramObject` PS: maybe we can just reopen and continue discussion there …"}
{"_id":"q-en-sphinx-doc__sphinx-9673-029c5a974363288b8b9502e011fe15fd4849dc619342afba974689d63d8544a4","text":"autodoctypehintsdescriptiontarget not working with Napoleon I was trying to use the config option combined with the Napoleon plugin (using Google style). The return types were missing from the resulting documentation. Just generate the documentation using Napoleon and the config options: Generate the documentation of a function with the following docstring: As the return is specified, the return type should be present in the documentation, either as a rtype section or as part of the return description. ! Win 8 0 '', '', '', '', '', No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-9698-18d31524ca55bd17c4209eec783bcd72e80b3003cff09ed17b92e0d3afa1acde","text":"An index entry with parens was registered for directive with option An index entry with parens was registered for directive with option. It should not have parens. An index entry for the property should not have parens. N\/A <img width=\"528\" alt=\" 2021-10-03 13 00 53\" src=\"https:\/\/user-\"Mac 6 HEAD of 4.x No response No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-9711-fe2c3e43a897690a88f3e893cfcccb69bd2cc1a5758907381dab49c6087cf8bb","text":"needsextensions checks versions using strings The check is handy for verifying minimum extension versions, but it only checks versions in a 'string-like' manner. This means any version 9 is not allowed for any check of something 1. That is, treated as string '0.6' '0.10', but treated as versions '0.6' < '0.10'. Since Sphinx does the former, some extension versions may not be allowed when they should be. This passes just fine, because the requirements pin sphinx-gallery to 0.9. But if you then update to the current 0.10 release: results in a failure due to a \"not new enough\" version: sphinx-gallery 0.10.0 should be accepted if 0.6 is the minimum specified. No response Fedora 6 2 No response No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-9797-ba8fca66fbcd3ee45e61b55b8950eb117d6ea57275cb77fcef92a3439fad29e6","text":"Overridden classmethod documentation not inherited with autodoc The documentation for a classmethod in a parent class is not inherited by the method re-defined (without documentation) in a child class Module : Sphinx configuration : Documentation index : Build: Overridden classmethod to inherit documentation from parent class ! Ubuntu 20.04 8 0 returns the expected value"}
{"_id":"q-en-sphinx-doc__sphinx-9798-53ceacfe18b620e4e2325e776cf7ba4644bcdedc0ec919a834bd51d6ec67fc6a","text":"Nitpick flags Literal annotation values as missing py:class (with type hints in description) This is basically the same issue as , which was fixed in . However, I still get this issue when using . Comment out the line in and it is successful, as shown below (and removing the build artifacts to start fresh). No error, the build should succeed. See attachment in \"How to Reproduce\" section N\/A - output is shown in \"How to Reproduce\" section Linux and Windows 9 0 , N\/A This re-produces for me on both Linux and Windows. I think the source of it issue is probably from in since this function would otherwise be skipped if the type hints are left in the signature. But I haven't yet been able to track it all the way down to the error."}
{"_id":"q-en-sphinx-doc__sphinx-9799-f1cb94533c542a2a1ce40f39925e7645d8dac87d48b87218d0ad4cebff477a5e","text":"Re-opening : hexadecimal default arguments are changed to decimal I am experiencing the exact same problem as described in : hexadecimal default arguments are changed to decimal. Autodoc the following function: HTML result looks like this Hexadecimal defaults should not be converted to decimal, or at least there should be an option to enforce this behaviour. I'm afraid this is private No response Linux Ubuntu 20.04 10 0 autodoc, intersphinx, napoleon Chromium 94 No response"}
{"_id":"q-en-sphinx-doc__sphinx-9828-5aaad5a4277ac5f45b1559764195ac8c91232fc4b1344d93187fdecda88784ae","text":"Support for fuzzy translations Problem Entries in po files that are currently marked as \"fuzzy\" are not used in \"mo\" files. The original source language is used instead of the translated language. Fuzzy translations are translations that in general need to be reviewed by a translator. For example all machine translations could be marked as fuzzy. It would be desirable to let the user decide whether to include these fuzzy translations in the generated documentation. Proposed solution Add a new configuration parameter \"usefuzzytranslations\". Change in line 72 of the file writemo(filemo, po) to writemo(filemo, po, usefuzzytranslations) Writing mo files with fuzzy translations is supported by the babel package, as can be seen in the documentation:"}
{"_id":"q-en-sphinx-doc__sphinx-9829-ad0dbe19a58592222c56b8158d388411743b7b39f92953dfbfeb64e86c585ed3","text":"Add a way to defer loading of MathJax Is your feature request related to a problem? Please describe. It is quite tricky to configure MathJax to work with Sphinx currently. Sphinx loads MathJax asynchronously since and While this was fine for MathJax 2, because of the special kind of blocks mentioned in , it doesn't work well with MathJax 3. Indeed, in MathJax 3, MathJax expect a config block to be present before MathJax is loaded. Sphinx 4 parameter: This assumes that the is a simple dictionary, which isn't sufficient: that configuration should be able to contain functions, for example. The only possibility at the moment is to add a separate script file containing a MathJax configuration and to load it with. Describe the solution you'd like There are three possibilities: Allow arbitrary strings for mathjax3config, and in that case don't JSON-serialize them. Change to when loading MathJax. Make it possible for users to change to themselves. At the moment this isn't possible because the flags is unconditionally : The latter two are preferable because they would allow individual pages to use different MathJax config by using a block to override the default MathJax configuration on a given page (the script in that defer` option). CC , the author of . Thanks!"}
{"_id":"q-en-sphinx-doc__sphinx-9902-3633ee9f880ce8bf5a0b27df0e7ca1654fc3a40efc6899bd08c6650fb485f16d","text":".. py:data:: :type: option does not correctly link to type supplied when using the .. py:data:: directive there is the :type: option to specify the type for the data being stored in the module level attribute. It does not work within the context of the module. What I mean by that is I have to key out the entire module path and the class for the type instead of importing the type into the module and only having to type in the type class name. say I have the following in the module \"mylibrary.module1\" The output I get when building the docs is this. and \"Unit\" is not a clickable link. However if I have the following in the module \"mylibrary.module1\" I get this for an output and \"mylibrary.module2.Unit\" is a clickable link that brings up the documentation for the Unit class to be able to do the following and have the output be and \"Unit\" be a clickable link that leads to the class that is specified in :type: No response Windows 8 2 No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-9982-1b163d77566f31b3761669363e65a049359e737c976171ae3ac85843dd859ba3","text":"Different build warnings are colored differently Some types of warnings seem to be different colors than others. An image warning is light red, a transition warning is dark red. This behavior exists regardless of being set. Use the branch specified. I purposely broke an image's path to showcase the differently-colored warning so it's not on main or develop. Based on , I'd expect all warnings to have the same color, and to not match the color of errors. ! macOS 10.15.7 (Catalina) 7 Sphinx v4.1.2 No response No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-9987-4ff74a90d84bc2b18a55a298425b3f4b22ffb3046f274c55be3f0850263f2910","text":"Using position-only parameter syntax in prevents docstrings for attributes from being parsed I'm currently using in most of my functions, including , however this prevents inline, before, and after doc strings from being parsed. Once is open, you will see that only has the property documented. Both classes should have the property documented <img width=\"636\" alt=\"image\" src=\"https:\/\/user-\"macOS 11.6.1 10 1 , No response No response"}
{"_id":"q-en-sphinx-doc__sphinx-9999-8064c53cfb6e3e09ddfa7c03931d3cf17a9a051c72c16028ac7454b1bab3d731","text":"Latex: terms are not separated by a newline I use simple indentations for terms and their explanation: Example: Which results in the following HTML output: Alabaster: ! RTD theme: ! While xelatex output does not contain a new line: ! Build the snippet. No response Build the snippet No response Linux 8 0 No response No response No response Latex: terms are not separated by a newline I use simple indentations for terms and their explanation: Example: Which results in the following HTML output: Alabaster: ! RTD theme: ! While xelatex output does not contain a new line: ! Build the snippet. No response Build the snippet No response Linux 8 0 No response No response No response LaTeX: glossary terms with common definition are rendered with too much vertical whitespace as in title and No response see code snippet ! Mac 0 2 Latex: terms are not separated by a newline I use simple indentations for terms and their explanation: Example: Which results in the following HTML output: Alabaster: ! RTD theme: ! While xelatex output does not contain a new line: ! Build the snippet. No response Build the snippet No response Linux 8 0 No response No response No response"}
{"_id":"q-en-sympy__sympy-11384-28c1629f3903487915d6bee1b6c6dce2f3766f8c44db964ab60011b1a3ac49fa","text":"fps should print as a formal power series When I first used , I didn't realize it really was a formal power series as it claims to be, because it prints like a normal series (same as ) But if you look at the string form, you see That is, it really does represent it as the formula (albiet, not simplified). It out to print it like this, so you can see that that's what it's working with. Side question: if you enter something it can't compute, it just returns the function Is that intentional? It seems like it ought to raise an exception in that case."}
{"_id":"q-en-sympy__sympy-11400-dbf208a47af037517e7c1223611915e921e11ee05c37182fedbb4ce2f953e0ea","text":"ccode(sinc(x)) doesn't work I don't think has , but it could print"}
{"_id":"q-en-sympy__sympy-11438-fc2d3608678184d86e6c52c86b0a870ad5b00c5d08fc5d9dc2a6add0e288c0e6","text":"diophantine: misclassification A check should be made that all powers are the same (not only that they are even)."}
{"_id":"q-en-sympy__sympy-11618-0c4d947e7fc5e048fccc05b95de41d4df3d319a7781299d436535979172a1376","text":"distance calculation wrong The 3rd dimension is being ignored when the Points are zipped together to calculate the distance so is being computed instead of ."}
{"_id":"q-en-sympy__sympy-11787-2574833c1382bae54d0d0b1b6c92257c7eff305024758ad2450bd3417c111ec4","text":"nsolve wrong result Here's a plot of the function in question. As you can see, the root should be around 0.7. !"}
{"_id":"q-en-sympy__sympy-11794-fe08cce32ea59279c1ae954058e08a433c4b9ca6d6d48907f52db125a56e4158","text":"ASCII printing for Singularity Function. Implementation of ASCII printing for Singularity Functions is needed."}
{"_id":"q-en-sympy__sympy-11796-f4f2f05b44a443770daff463258341bfc0bb133ce802240f3e46bba8413616ed","text":"Where oo belongs? (Concept) Hi again, well, i'm little confuse of the conditions to take or not in some sets: First the means the interval is created excluding and , and interval interpret it in that way, but now: Here is a little conflict, in first place Interval show don't take but now it is there? in some way is fine to have a way to can represent the from Interval. Now from this point we have some points: How they will interpret the limit concept? basically two options, limit is: or ? This point is very important, because define the behavior for sets, and affects directly like this issue: so, for now only to match the math in all sets we can say the limit is calculated via now, what is the effect of this in Sympy?, first this enable the limit concept in every unbounded set, for now i found this two issues: for example, actually we have this: this return should be something like... ? because in the limit we don't know what is the proportion of and , so we can't calc it. actually this concept is applied in some way like: Now the next question, will represent the infinite, as a integer, real or what? i know this question don't have sense, but let me try explain it: if the oo represent the infinite without form, it can exist in S.Naturals, and S.Reals, but if you represent the infinite like the interval between it, where is the limit of x to infinite while always , in other way where A go to and B do to , but it need always will respect this condition so between and can exist any type of numbers, so can't exist in because can contains a real number for example, but the extension of that concept says can't exist in any set, because always will exist a bigger set, in sympy you have an approximation of it, is , but don't will be true completely, because, why is it the limit set?, can represent two things, the limit of the actually human knowledge (or applied to Sympy), or the 'master' set, thinking its like the perfection of the sets knowledge. Obvs, to make some sense in the actual system the option is interpret without limit or form, and take the second interpretation of (if you take the first. can't exist in any place). If you disagree you always can discuss and change the behavior. Objetives of this issue: Get a clear definitions in Sympy of: Infinite Limit UniversalSet Then, clear the behavior of this concepts in Sympy, and to finish, set the behavior in Sympy. Thx. Cya."}
{"_id":"q-en-sympy__sympy-11818-181a04bfa4fbf17ebc02141f9b485ac172dacfbf464f05a587bfe8293a439bf0","text":"Union(FiniteSet(oo), S.Complexes) gives S.Complexes (should remain unevaluated) Hi, well searching i found this: i don't know with this where belongs, is part of Complexes or not? Thx. Cya."}
{"_id":"q-en-sympy__sympy-11822-28906436d3005974ea35aa65d499a36f3fd683010a76b8afffd3ea409892336f","text":"Exception when printing Symbol('') It has something to do with the unicode pretty printer. works."}
{"_id":"q-en-sympy__sympy-11831-a1e5b1513aa138b26f41e9c65b393bd80090c2ac4cffa413c01a4feb2d5729fc","text":"set intersection gives TypeError: object of type 'Naturals0' has no len() This is from Optimistically marking this as easy to fix (I could be wrong)."}
{"_id":"q-en-sympy__sympy-11862-57be8f68a7a2567055f3f288c87c72f679cf1d61999a3ea6fa55cd2de13aac03","text":"lambdify precision loss with module=mpmath from high-precision Floats Floats with more than 16 digits are converted to double precision somewhere. Consider: Note has a 64-digit accurate Float in it (and the value is correct). But lambdifying and is not same: The help string for shows no loss: I haven't figured out how to look at the actual code yet, but somewhere something is being converted to double-precision (which might be fine for module=numpy but should not happen here)."}
{"_id":"q-en-sympy__sympy-11870-fdbc74e7245fb78a9556c429b9699db2ce6d626bfcf4293ad29eb004687d493e","text":"simplifying exponential -trig identities Ideally, this would yield . Is there a way to do this? As a corollary, it would be awesome if could yield . Thank you for your consideration!"}
{"_id":"q-en-sympy__sympy-11897-4857f9e81a5f8173bbb300decc8dd6c259c194d7000dca647aea0f0d28202c0d","text":"LaTeX printer inconsistent with pretty printer The LaTeX printer should always give the same output as the pretty printer, unless better output is possible from LaTeX. In some cases it is inconsistent. For instance: (I also don't think the assumptions should affect printing)."}
{"_id":"q-en-sympy__sympy-11989-d1887aa49888d17dc103692b0fbfec27583ed7e075491a45a480174b9d390cdb","text":"diophantine doesn't find solutions for 2x2+y2-16 Diophantine returns the empty set but and are solutions. I suspect there is some issue with the solving of the elliptic case, as for example also doesn't return all solutions: it returns and but fails to find and"}
{"_id":"q-en-sympy__sympy-12096-6d2ec1ce32d8cfbc33d12e31bde752859fc3b0e74adf23b434e9309763f7bc04","text":"evalf does not call imp recursively Example from The code for this is in . It isn't calling evalf recursively on the return of ."}
{"_id":"q-en-sympy__sympy-12144-cdf6c5aaea59d250a88ef723acd34c3e2912129e4494e35ff452c3c886560fb6","text":"Dummy fails when is re-evaluated in S and srepr Hi all! Well, i found this little problem, Dummy is a nice function but in the way is implemented exist this: Basically, if you works with Dummy, and you get the expression with srepr or similar, when you eval it every Dummy will be interpreted as a new Dummy, so it fails, in the example you can see we can't eval the second expression because the 2 Dummy expression are interpreted as differents, other ex: So thinking a solution can be use a hash or similar function, like: Always will exist the hash collision, but at least to cover all possible can be sympy check the existents Dummy expressions and check the hash never by equal to other, the probability of the hash collision, to be equal, in the same context, with the same name, and in the same expression, i think is very low. Maybe instead use a hash can be a random expression, with chars like , that can help to avoid hash collision. Thx. Cya."}
{"_id":"q-en-sympy__sympy-12171-d5ab3cc86af804b4b40e214deccc592c89ae46c53df5f04ac8e1f813971c661d","text":"matematica code printer does not handle floats and derivatives correctly In its current state the mathematica code printer does not handle Derivative(func(vars), deriver) e.g. Derivative(f(t), t) yields Derivative(f(t), t) instead of D[f[t],t] Also floats with exponents are not handled correctly e.g. 1.0e-4 is not converted to 1.0^-4 This has an easy fix by adding the following lines to MCodePrinter: def printDerivative(self, expr): return \"D[%s]\" % (self.stringify(, \", \")) def printFloat(self, expr): res =str(expr) return res.replace('e','^')"}
{"_id":"q-en-sympy__sympy-12183-6734b0b0767c830cf35e3ec93befa58d6cdc749ffef933e5583eaa5c1f4fad35","text":"Empty intersection should be UniversalSet It should be . See (settheory)#Nullary_intersection"}
{"_id":"q-en-sympy__sympy-12194-d261e189faf4530b4898bf660fbbe9d8c59d36aacc368ba8c96a9c7937c21d49","text":"List of prime factors including multiplicity is missing The vast majority of the time when using prime factorization in algorithms I'm interested in a list of prime factors. E.g. for 24 it's . However sympy (to my knowledge) only provides which returns a dictionary. I always end up writing a helper function: factorlist = lambda fac: sum(([p] * fac[p] for p in sorted(fac)), []) It would be really nice if factorizations in this format were directly provided by sympy, either by an option to (e.g. ) or under a new function name entirely."}
{"_id":"q-en-sympy__sympy-12214-03081143eeeb8ff814ffb9d3c743ebb0773032869785aba38a73eafdc92050aa","text":"b splines of degree 1? Looking at the spline_basis function: This seems to break while the following is fine: I can tweak in extra knots, i.e. placing them at (0.1,0.2, 10.1,10.2 in the example) but this is still not really giving the desired basis matrix. Or am I just missing something very obvious? I always generated knots for bspline basis matrix by taking the list of knots and:"}
{"_id":"q-en-sympy__sympy-12227-16f3f4dd49c59129a0e558e2be3d7562aa454614cec34f574f223ab0cf67af10","text":"Float constructor should allow to set binary precision I discussed this in some other issues, but I don't think I've opened an issue for it. The Float constructor only allows you to set the decimal precision. It then converts this to binary precision and saves that in . The problem here is that there are fewer decimal precisions than binary precisions. For example: So for instance, if you had a float that used 54 binary digits, it would be difficult to create a Float from it. I'm not even sure if it's possible, since in the current constructor, the argument overrides the fourth argument of a tuple input. One issue here is that everywhere else in the Sympy and mpmath code, refers to binary precision and refers to decimal precision. But currently, the second argument to is called , but refers to the decimal precision. So ideally, it should be where would override the default value of , and calling both (even consistently), like would be an error. Since the keyword argument name has to change, it would be a backwards compatibility break for anyone who calls Float like . Ideally this would require a deprecation cycle, but it's not possible to detect if someone is calling meaning the old way (20 decimal digits) or the new way (20 binary digits). The good news is that and call this argument , so there's no need to change things there."}
{"_id":"q-en-sympy__sympy-12270-32308435b73f07332364fbecc61a4680fc5ac036851eb92e61f5d843db7be659","text":"cannot extract_multiplicatively(-2) from (-2x - 2y) I think this might be a bug. similarly: Assuming it is indeed a bug, here is why it happens: Look in where: which is then made into a non-evaluated , from which cannot be multiplicatively extracted; for example: do you think this is bug? (see your commit : not clear whether we still need this special case anyway)"}
{"_id":"q-en-sympy__sympy-12286-cdf6c5aaea59d250a88ef723acd34c3e2912129e4494e35ff452c3c886560fb6","text":"Dummy fails when is re-evaluated in S and srepr Hi all! Well, i found this little problem, Dummy is a nice function but in the way is implemented exist this: Basically, if you works with Dummy, and you get the expression with srepr or similar, when you eval it every Dummy will be interpreted as a new Dummy, so it fails, in the example you can see we can't eval the second expression because the 2 Dummy expression are interpreted as differents, other ex: So thinking a solution can be use a hash or similar function, like: Always will exist the hash collision, but at least to cover all possible can be sympy check the existents Dummy expressions and check the hash never by equal to other, the probability of the hash collision, to be equal, in the same context, with the same name, and in the same expression, i think is very low. Maybe instead use a hash can be a random expression, with chars like , that can help to avoid hash collision. Thx. Cya."}
{"_id":"q-en-sympy__sympy-12301-06d1b5a6684d3742aa5471c781d5f7aab36a6449035011a7f6cac368fb894941","text":"Test failure in Travis I was able to reproduce this locally on a 64-bit system running Ubuntu 16.04. How to reproduce:"}
{"_id":"q-en-sympy__sympy-12307-945ae7227bc7d6135bdc3c73041cfe19fbcaefbd370d1bff84740e17f55ffc81","text":"Codegen: sign function in Fortran The Fortran code generated by Sympy for the sign function is not a valid Fortran syntax. With Sympy 1.0 and Python 3.6: (The same behavior is obtained with .) The syntax of the sign function in Fortran is given . I guess it would be preferable to generate something like or depending of the context. I'm not sure how to get the same behavior than Sympy for , but at least it would be compilable."}
{"_id":"q-en-sympy__sympy-12419-175baeab5f75226fb988ee97f9b34bdb005f4ab7e2eab89fea9e07dc978f1a7b","text":"Sum of the elements of an identity matrix is zero I think this is a bug. I created a matrix by M.T * M under an assumption that M is orthogonal. SymPy successfully recognized that the result is an identity matrix. I tested its identity-ness by element-wise, queries, and sum of the diagonal elements and received expected results. However, when I attempt to evaluate the total sum of the elements the result was 0 while 'n' is expected."}
{"_id":"q-en-sympy__sympy-12428-8be8969dce4f6464b4396389d43732a1e1323889c7f7c05b49d44ba67baa69f1","text":"DiagonalMatrix[i, j] -0 Similar to , DiagonalMatrix references to position (i,j) give 0:"}
{"_id":"q-en-sympy__sympy-12454-15ad0f3f2a441786e2926f745ce23b6c5dd7cd7443697528ec27b311fe58d2c0","text":"isupper() raises IndexError for tall matrices The function Matrix.isupper raises an IndexError for a 4x2 matrix of zeros. The code for isupper() is For a 4x2 matrix, isupper iterates over the indices: The attempt to index the (3,2) entry appears to be the source of the error."}
{"_id":"q-en-sympy__sympy-12481-4a9c91555a196bc0df21b092ec3fa60166df639d6d15c9e58ef124d3e2529fe8","text":"constructor fails with non-disjoint cycles Calling raises a instead of constructing the identity permutation. If the cycles passed in are non-disjoint, they should be applied in left-to-right order and the resulting permutation should be returned. This should be easy to compute. I don't see a reason why non-disjoint cycles should be forbidden."}
{"_id":"q-en-sympy__sympy-12529-12f62d4f7254a4611b1f597000869827bd88ca60139b44faa63324b220fb9f9a","text":"Totient of any non-integer number According to the Totient function definition on , the totient of non-integer numbers is not there. But in sympy: the value is returned, instead of an error."}
{"_id":"q-en-sympy__sympy-12812-e534fb3834d4d5372b713274ea73f55055345c78dd789756a96a3d3d12050c00","text":"diffgeom does not support multivector fields (0,m) well enough A limitation of the diffgeom package is that it does not provide support for fields higher than (0,m). TensorProduct, WedgeProduct, and others all assume inputs of differential form-fields, while mixed multivector fields make sense in some contexts (such as Poisson geometry). I'll work on implementing (n,m)-fields in the current code, where it makes sense."}
{"_id":"q-en-sympy__sympy-12881-64948709eaa0d08185299fcac5b634efc39657e71e375a9d64256302c99b8649","text":"Poly(x,x,y).free_symbols -{x, y} instead of just {x} No free symbols of generators that don't appear in the expression of the polynomial should appear in the set of free symbols."}
{"_id":"q-en-sympy__sympy-13018-d5f0f39701d25347f6d93622a047addaffabc7a28505758d888e79723370b313","text":"Wrong\/surprising result from noncommutative The following behavior in Sympy seems surprising: I would have expected this produces . The issue appears to be that uses fractional powers in the substitution. This then raises questions on what noncommutative symbols in Sympy actually mean mathematically. At least in my use cases for them, certainly does not imply . I would suggest something like the following:"}
{"_id":"q-en-sympy__sympy-13031-f7e41fc1e44f86ed31d544026000c9928c456e175d38fdd1d5417f86634acd7d","text":"Behavior of Matrix hstack and vstack changed in sympy 1.1 In sympy 1.0: returns Now, same in sympy 1.1: returns (0, 3) whereas: returns (1, 6)"}
{"_id":"q-en-sympy__sympy-13043-ee56fc6284a1d8342038d8a3c92e390b55fcfb9a9f3415613bdc23061bc57224","text":"decompose() function in intpoly returns a list of arbitrary order The decompose() function, with separate=True, returns , which is ordered arbitrarily. What is this used for? It should be sorted somehow, or returning a set (in which case, why not just use the returned dictionary and have the caller take the values). This is causing test failures for me after some changes to the core. CC"}
{"_id":"q-en-sympy__sympy-13146-4b5af696a76540c74d27cfb3f613d438dddf4d9648cc3e1406ed5f419682fcb1","text":"Exponent doesn't fully simplify Say I have code like this: The output is How do I simplify it to 0?"}
{"_id":"q-en-sympy__sympy-13173-bcd4550d0a1dcec1900affc04ff25642fa9a982a1e25df2cbae778730a328e25","text":"degree(multivariate) -degree of first generator When giving the degree function a multivariate expression, the default degree returned is for the first generator which is chosen canonically but arbitrarily. When the degree method of Poly is called, this is not so bad because one would\/should know what the generators are. For the case of using the function, however, I believe it allows ambiguity to pass silently by not requiring the generator to be specified. Here, the return value depends on the symbols chosen"}
{"_id":"q-en-sympy__sympy-13177-19c32f90d7bead39dbd1779ddacf8a4acd07f55e7eb7db81be9870d89e89a503","text":"Mod(x2, x) is not (always) 0 When the base is not an integer, is not 0. The base is not tested to be an integer in Mod's eval logic: so but"}
{"_id":"q-en-sympy__sympy-13185-36f2f607326f694dfbe176ad0c645c3bdadff93a1bf1201a3f55fa3db2920954","text":"cse() has strange behaviour for MatrixSymbol indexing Example: This gives the following output: and are just copies of the matrices and , respectively."}
{"_id":"q-en-sympy__sympy-13198-01a9023c5bc9c0b4baa7afbc935a830cd92ad36952908c89e1b887090fc9797d","text":"Problem factoring trivial polynomial The factoring of w is incorrect - look at the order of x it is 10^-8 not 10^-4. Bug in simplify ? Hi, I'm running into issues where simplify comes up with incorrect results. sympy verion 1.0, python version 3.52, ubuntu 16.04. I've also verified the same with the Sympy online console (). The problem comes when trying to simplify a mass matrix for a simple 3-DOF robot: If I use the matrix above directly, I get correct results: which should be: If on the other hand, I simplify first I get an incorrect result: I was wondering if this is a bug with simplify, or am I doing something wrong? Thanks."}
{"_id":"q-en-sympy__sympy-13264-d5b7e3abc1aaad4eae7cb2df48aa3af00b62e6c426aac4370a3cb6d7624d3947","text":"simplify doesn't do the last simplification step Note: I'm using Python 3.5.1 and sympy 0.7.6.1. Full script is attached in the .zip file: I've been using Sympy to compute Laplacian in different geometrical coordinates system. This usually creates extremelly long expressions and simplify usually does a great job. However, I've come against one particular case where simplify does most of the job but doesn't perform the ultimate simplification step. I've reduced the problem set getting rid of all of the Derivative terms. The initial expression is still huge (675 ops!): ! A first call to sucessfully reduces it to : ! Which is great (23 ops), but somehow frustrating as we can factorize it a little more. If I apply again, then I get the expression I wanted (7 ops): ! If simplify worked perfectly, shouldn't yield a different result from From my understanding of how simplify works, it successively tries different simplification strategies and compares them. In that case, the order of the different strategies matters. In my particular case, the strategy than eventually did the final factorisation step may have been called to early. I fixed that calling simplify again. We could fix it applying an optional recursive mode at the end of the simplify function: I can make the pull request if that helps."}
{"_id":"q-en-sympy__sympy-13265-39ea86f878e98524c38a20897c5c373046d4777e4e6443c078f1de39c2fa9671","text":"Simplification fails to recognize sin expressed as exponentials The expression for has not been recognized, while expressions for and have."}
{"_id":"q-en-sympy__sympy-13286-90fad2585a177762e95b018c4e9fa1c9ceb5bf28fc381509bc39bf1a983ad2d4","text":"periodicity(Abs(sin(x)),x) return 2pi periodicity(Abs(sin(x)),x) returns 2pi instead of pi periodicity(x 2, x) give recursion error and It fixes issue it will stop any relational Expression from entering into infinite recursion and return None It improves the periodicity of absolute trigonometric function issue"}
{"_id":"q-en-sympy__sympy-13301-f4368fa287aca2748a13913c8f0c1c0528f06634aced7c6157a7ee23396500de","text":"AccumBounds needs recreatable str\/srepr The str and srepr should be recreatable. The pretty printer should be the only thing that prints the pretty form."}
{"_id":"q-en-sympy__sympy-13309-6a1ee55db632a34c272d5420be04b545be8e7d3d0a1553fc11a76daab6be06f1","text":"implement Piecewise rewriting for Min and Max This was mention in . I am just puttng this here as a separate issue."}
{"_id":"q-en-sympy__sympy-13346-963a518ca9611aed0f94e1862ea733b8f83e8bc5da9590cccaa4099a0106c806","text":"Test failures on non-NumPy environment Following tests fail on where NumPy is not installed: Temporarily fixed by : Following doctest fails:"}
{"_id":"q-en-sympy__sympy-13361-c65d3fa4f8d7c92f848c1e812aede8b8c0f35ff1a4fa5d6fbe737b5a8835d3f1","text":"evalf with subs argument incorrectly evaluates expressions with floor returns instead of 20. (Python 3.6.1, SymPy 1.1.1). It seems that incorrect attribute is somehow attached to the result of , which is why 20 is then rounded to the nearest power of 2. Considering that works correctly, perhaps a fix would be to make pass its subs argument to first, i.e., act the same as Based on a"}
{"_id":"q-en-sympy__sympy-13369-502327462beffb86b01d00e36d78fa2dac538f7d399096b4e49b430cfba6ab11","text":"Eigenvalues of a 3 by 3 symbolic matrix are not found by .eigenvals throws Since the characteristic polynomial is cubic, it can be solved explicitly: returns three roots of the characteristic polynomial. I do not understand why these are not found by which returns (hence the failure of method). Declaring x as a real symbol does not help. Based on Stack Overflow post"}
{"_id":"q-en-sympy__sympy-13372-97d4ad3e506be8f6ab7e08adf71425030c14df35e14df982a3b3dcb49e137f1d","text":"UnboundLocalError in evalf I found this after changing the order of Mul args in Based on the code, I think the elif clauses that define reprec and imprec should have an . That appears to fix it, although I didn't try to debug to see why the arg order is mattering here."}
{"_id":"q-en-sympy__sympy-13437-26947036a3b44780b45420dbc0dde5bf4f1aebd3927674b98bd7634a216dcfa1","text":"bell(n).limit(n, oo) should be oo rather than bell(oo) should take the value infinity, but the current output is . As the Bell numbers represent the number of partitions of a set, it seems natural that should be able to be evaluated rather than be returned unevaluated. This issue is also in line with the recent fixes to the corresponding limit for the Fibonacci numbers and Lucas numbers. I'm new to Sympy, so I'd appreciate the opportunity to fix this bug myself if that's alright."}
{"_id":"q-en-sympy__sympy-13471-eada7f355ce2d4334463f71a146f06d378409a242f959102fc145057e0cf5885","text":"Python 2-3 pickle fails with float-containing expressions Dumping a pickled sympy expression containing a float in Python 2, then loading it in Python 3 generates an error. Here is a minimum working example, verified with sympy git commit (master at time of writing), Python 2.7 and Python 3.6: and the result:"}
{"_id":"q-en-sympy__sympy-13480-811a1bdfae27bb10277c6b06eba8e9e2ba2b556f2f2aa5084d8b498cec56dad9","text":".subs on coth(log(tan(x))) errors for certain integral values from sympy import * x = Symbol('x') e = coth(log(tan(x))) print(e.subs(x, 2)) ... File \"C:\\Users\\E\\Desktop\\sympy-master\\sympy\\functions\\elementary\\\", line 590, in eval if cotm is S.ComplexInfinity: NameError: name 'cotm' is not defined Fails for 2, 3, 5, 6, 8, 9, 11, 12, 13, 15, 18, ... etc."}
{"_id":"q-en-sympy__sympy-13551-38080d0745cf5a4b94476ff6b9fb74e707ace156bfa10bc535909ae84f70509d","text":"Product(n + 1 \/ 2k, [k, 0, n-1]) is incorrect from sympy import from import n,k p = Product(n + 1 \/ 2k, [k, 0, n-1]).doit() print(simplify(p)) 2(n(-n + 1)\/2) + nn print(p.subs(n,2)) 9\/2 This is incorrect- for example, the product for is . The correct expression involves the ."}
{"_id":"q-en-sympy__sympy-13574-3e60cf8f6ac778bb91a16f0ab17594097f902ae450c5bd290b01304ee98ccb4f","text":"randMatrix won't generatte symmetric sparse matrices When setting the percent parameter to anything else than 100, randMatrix fails to generate symmetric matrices. Consider the following examples: ' ' ' In [1]: import sympy In [2]: from sympy.matrices import randMatrix In [3]: randMatrix(3, symmetric=True, percent=1) Out[3]: Matrix([ [13, 61, 13], [59, 29, 59], [88, 13, 61]]) In [4]: randMatrix(3, symmetric=True, percent=50) Out[4]: Matrix([ [90, 60, 0], [ 0, 0, 0], [60, 59, 25]]) In [7]: randMatrix(3, symmetric=True, percent=99) Out[7]: Matrix([ [0, 0, 19], [0, 0, 0], [0, 0, 0]]) In [9]: randMatrix(3, symmetric=True, percent=0) Out[9]: Matrix([ [78, 78, 61], [68, 8, 61], [22, 68, 8]]) ' ' ' Also, the documentation says , but the behaviour is the opposite. Setting it to 100 (default) produces the expected behaviour. The problem happens with both the released version from pypi and git master."}
{"_id":"q-en-sympy__sympy-13581-eedf92d45cb9316929d9f679d4b557f85d14df76442934a7bcdc0af80c5cfca6","text":"Mod(Mod(x + 1, 2) + 1, 2) should simplify to Mod(x, 2) From Also, something like could be simplified. Recursively."}
{"_id":"q-en-sympy__sympy-13619-5788bbb397c89b07d5e37f878ad623ad6f15993efc2b2ca8a75a063b07b2a2f2","text":"Undefined functions with number arguments should have is_number be False Original issue for : Original author:"}
{"_id":"q-en-sympy__sympy-13624-9f6e1698b408fc63493c78702daf59308a0d3eab17b899e3374615bdc23a1c92","text":"Python code printer (pycode) should support Assignment There is a lookup on 'contract', either we should give it a default in the or we should make the code accessing use with a default."}
{"_id":"q-en-sympy__sympy-13647-9d8714313229af68922566b8d1941823fc8dac82eae8c51459658a40b6233abe","text":"Matrix.col_insert() no longer seems to work correctly. Example: The 3 x 3 identify matrix to the right of the columns of twos is shifted from the bottom three rows to the top three rows. Do you think this has to do with your matrix refactor?"}
{"_id":"q-en-sympy__sympy-13682-782f5d153a32b16be53f2d63b596219dd934b6d7834da56480b23d7b9a1de529","text":"Ordinal arithmetic It would be nice if SymPy would have provided ordinal arithmetic. That would require either definining a new object called , that will either inherit from or from (option 1), or new assumption(s) that will allow the current symbols to be ordinals. How can it work with assumptions? Roughly as follows: and arithmetic should be defined: addition, multiplication and exponentiation (all are not commutative in general), and some other useful methods, such as: relationals supporting ordinals cantor normal form prime ordinals (perhaps also with the assumption) limits of sequences of ordinals cardinals (related to ) some \"known\" ordinals such as omega, omega_1, ..., epsilon numbers, etc. you can find a python implementation of non-symbolic ordinal arithmetic. I can be up to the task of defining a symbolic variation, but I don't think I can implement it in SymPy."}
{"_id":"q-en-sympy__sympy-13744-8745621163bb85252820931ec9e667c6be775e01a22dbd2edbce3379c56ded93","text":"coset_table doctest failure (Fatal Python error: Cannot recover from stack overflow.) See https:\/\/travis- Here is the test header"}
{"_id":"q-en-sympy__sympy-13757-1497d767c3a6fab13198ad1b790478cb2f8350d354654573637f8001891b42ed","text":"Multiplying an expression by a Poly does not evaluate when the expression is on the left side of the multiplication Tested in Python 3.4 64-bit and 3.6 64-bit Version: 1.1.2.dev0"}
{"_id":"q-en-sympy__sympy-13761-e13b22fddec39d9094455952e5880c264056262316a713b0ca9c450bad623b81","text":"Cannot simplify x + csch(sinc(1)) from sympy import * x = Symbol('x') print(simplify(x + csch(sinc(1)))) ... File \"C:\\Users\\E\\Desktop\\sympy-master\\sympy\\simplify\\\", line 433, in f rv = fmap KeyError: sinc (I should have said: cannot apply the simplification function, since I'm not expecting any simplification to actually take place)."}
{"_id":"q-en-sympy__sympy-13768-a8cd9e9a06a18dcf17de8fff22d8d35d7c3ed8fa0b27c7c4a54f8d17520969a8","text":"fix the dimension mismatches when using (dot) the dimension mismatched when using A.dot(B) where A is matrix B is 1 x m or n x 1 matrix before fixing it, if we used B as m x n matrix where n or m != 1 it gives a strange answer, but after fixing it raises error if m or n not equal 1"}
{"_id":"q-en-sympy__sympy-13773-8b0cdab22ee9c983f319afa0937ccf53edf7dbce3666332b4477a5420a6d87dd","text":"@ (matmul) should fail if one argument is not a matrix Right now () just copies , but it should actually only work if the multiplication is actually a matrix multiplication. This is also how NumPy works"}
{"_id":"q-en-sympy__sympy-13798-593504aa53f9e1dabc299d2f7a6622b849d50329d9123cf59cffb7d256c689f2","text":"latex() and mulsymbol The pretty-printing function accepts a kwarg that must be one of four choices. I would like to be able to supply my own choice which is not in the list. Specifically, I want the multiplication symbol to be (i.e., a thin space). This is what I mean Thin spaces are used by sympy to separate differentials from integrands in integrals. Is there a reason why the user cannot supply the of their choosing? Or are the 4 choices a historical artifact? I'm willing to attempt making a PR to allow to be arbitrary (and backwards-compatible) if such a PR would be considered."}
{"_id":"q-en-sympy__sympy-13806-8bd562e3dd8a0270632e5f4b7d9a705e72349803a3fffa086b43736d92a38941","text":"No support for [{90^^\\circ }] I have latex [{90^^\\circ }], which means angle ninety degree, for example (cos(90 degree)) = 0, please add support for that? I appreate your reply."}
{"_id":"q-en-sympy__sympy-13808-a0a174dfc2900f3891cbd63d5aba900909acee507a0cface4599b145a6cec3d4","text":"integrate(1\/(2-cos(theta)),(theta,0,pi)) Sympy produces NaN. Actually for integrate(1\/(a-cos(theta)),(theta,0,pi)) for a 1 should be pi\/sqrt((a-1)*(a+1)). So, the right answer should be pi\/sqrt(3). However sympy seems to use the subtitution like t = tan(x\/2) which is infinite when x = pi. When I try integrate(1\/(2-cos(theta)),theta) , I get \"sqrt(3)I(-log(tan(x\/2) - sqrt(3)I\/3) + log(tan(x\/2) + sqrt(3)I\/3))\/3\". Simplify() or trigsimp() doesn't work. And I don't understand why imaginary number appears."}
{"_id":"q-en-sympy__sympy-13840-18a2f47b2560f11e31935acddc723a3a55b31af8d410c9bbe68aef65885d29dc","text":"Max & Min converting using SymPy Why many languages likes js and R cannot be converted from Max & Min? !"}
{"_id":"q-en-sympy__sympy-13852-e8d8ac3ad161c36fd1cc3f5591a59f53b07faf537e961d2e141b2afabdff71ac","text":"Add evaluation for polylog Original issue for : Original author: Why does the expansion of polylog(1, z) have exppolar(-Ipi)? I don't see a reason for exppolar here: To my understanding, and are exactly the same function for all purposes. They agree for <1 by their power series definition. Both are branched at 1 in the same way. The mpmath evaluation implements their branch cuts consistently: when z is real and greater than 1, the imaginary part of both functions is -pi. I tested the evaluation at thousands of random points, real and complex: both return the same values. SymPy also agrees they have the same derivative, which is z\/(1-z): But with the current implementation of , it would seem that expandfunc changes the derivative of the function: returns which doesn't simplify to 0. In general, I think that having exppolar in expressions like is just not meaningful. The additional information contained in \"polar\" is the winding number of some path about 0. Here, because of + 1, this ends up being the winding number about 1, which is irrelevant because log is not branched at 1."}
{"_id":"q-en-sympy__sympy-13865-596d8936068da4e4075dab7ed015a26727071d1beaa6c2e245f0e08ce1058c62","text":"ODE incorrectly classified as Bernoulli A bug reported on returns . This is clearly due to only excluding f(x); it should also exclude x and dx."}
{"_id":"q-en-sympy__sympy-13877-dc4da66f5b09b03c098854dde379cfc2e68bdbbc5da530decb7f6c57b5888f5c","text":"Matrix determinant raises Invalid NaN comparison with particular symbolic entries from sympy import from import a f = lambda n: det(Matrix([[i + aj for i in range(n)] for j in range(n)])) f(1) 0 f(2) -a f(3) 2a(a + 2) + 2a(2a + 1) - 3a(2a + 2) f(4) 0 f(5) nan f(6) Traceback (most recent call last): File \"<pyshell\", line 1, in <modulef(6) File \"<pyshell\", line 1, in <lambdaf = lambda n: det(Matrix([[i + aj for i in range(n)] for j in range(n)])) File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\matrices\\expressions\\\", line 53, in det return Determinant(matexpr).doit() File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\matrices\\expressions\\\", line 37, in doit return () File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\matrices\\\", line 270, in evaldeterminant return () File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\matrices\\\", line 416, in det return self.evaldetbareiss() File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\matrices\\\", line 213, in evaldetbareiss return cancel(bareiss(self)) File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\matrices\\\", line 211, in bareiss return signbareiss(( - 1, - 1, entry), pivotval) File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\matrices\\\", line 211, in bareiss return signbareiss(( - 1, - 1, entry), pivotval) File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\matrices\\\", line 211, in bareiss return signbareiss(( - 1, - 1, entry), pivotval) [Previous line repeated 1 more times] File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\matrices\\\", line 55, in new rows, cols, flatlist = cls.handlecreationinputs(args, kwargs) File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\matrices\\\", line 2041, in handlecreationinputs for j in range(cols)]) File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\matrices\\\", line 2041, in <listcompfor j in range(cols)]) File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\matrices\\\", line 208, in entry cancel(ret) File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\polys\\\", line 6423, in cancel f = factorterms(f, radical=True) File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\core\\\", line 1193, in factorterms return do(expr) File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\core\\\", line 1189, in do [do(a) for a in p.args]) File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\core\\\", line 1189, in <listcomp*[do(a) for a in p.args]) File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\core\\\", line 1171, in do if all(a.ascoeffMul()[0] < 0 for a in listargs): File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\core\\\", line 1171, in <genexprif all(a.ascoeffMul()[0] < 0 for a in listargs): File \"C:\\Users\\E\\AppData\\Local\\Programs\\Python\\Python36\\lib\\site-packages\\sympy\\core\\\", line 323, in lt raise TypeError(\"Invalid NaN comparison\") TypeError: Invalid NaN comparison Correct me if I'm wrong but isn't the Bareiss algorithm only valid for integer matrices, which cannot be assumed here?"}
{"_id":"q-en-sympy__sympy-13895-f09881c35d5071947111b2e9e5e26d7d848dff88f64ae624d1d73f9b89ce9c96","text":"(-x\/4 - S(1)\/12)x - 1 simplifies to an inequivalent expression from sympy import x = Symbol('x') e = (-x\/4 - S(1)\/12)x - 1 e (-x\/4 - 1\/12)x - 1 f = simplify(e) f 12(-x)(-12x + (-3x - 1)x) a = S(9)\/5 simplify(e.subs(x,a)) -1 - 3215(1\/5)2(2\/5)\/225 simplify(f.subs(x,a)) -1 - 32(-1)(4\/5)60(1\/5)\/225 N(e.subs(x,a)) -1. N(f.subs(x,a)) -0. - 0.189590423018741I"}
{"_id":"q-en-sympy__sympy-13903-b571fa52afba675f9dfd85f2ce9092083e38f9f0277aba29e78e1f9e8b9f9adf","text":"max & min i found most language cannot be converted into max & min like octave,Fortran and others (js and R have been fix , thx ;) )"}
{"_id":"q-en-sympy__sympy-13915-69006d2b10982ed4d50b7399aada67b18d2db3178135eb25744e9ebdc2eb9cd9","text":"Issue with a substitution that leads to an undefined expression If b is substituted by a, r is undefined. It is possible to calculate the limit But whenever a subexpression of r is undefined, r itself is undefined."}
{"_id":"q-en-sympy__sympy-13962-bd0fc90c2a55d8b5d28b9535f4d431b0b1f57079c5fa3be362c9a24a4ff1a4fc","text":"Printing should use short representation of quantities. There is a test that explicitly expects that printing does not use but : Is there a reason behind this? I find it quite user-unfriendly to look at instead of . It would be very easy to change here: But then, the above test would fail. Is anyone emotionally attached to the current verbose display of units and quantities? Use abbreviated form of quantities when printing Currently, the abbreviation used in the definition of quantities, e.g. in the definition of , is hardly used anywhere. For example: returns: This PR modifies printing of quantities to use the abbreviation if one was provided. Example: now returns: NOTE: I changed an existing test that explicitly expected the non-abbreviated name to be printed. I just do not see the point of such behaviour, but I am happy to be educated otherwise."}
{"_id":"q-en-sympy__sympy-13971-e9f0050460aefd9e7aa1198cec5b4a7f3b41a6b98de8c01234ce94138e70d5fb","text":"Display of SeqFormula() The Jupyter rendering of this command backslash-escapes the brackets producing: Copying this output to a markdown cell this does not render properly. Whereas: does render just fine. So - sequence output should not backslash-escape square brackets, or, should instead render?"}
{"_id":"q-en-sympy__sympy-13974-f49daab937c9dab3e2abb330f7c11ad0173ad4d541663d9b55d565cfe26aae56","text":"Evaluating powers of Powers of tensor product expressions are not possible to evaluate with either method nor the function. This is an example session showing the issue where and shows expected result for and respectively."}
{"_id":"q-en-sympy__sympy-13978-a8ba326202e056eb1f8366254fd19002e46939ffe13756ab85f8e682615585f7","text":"Generation of wrong octave code for imaginary number representation Hi, sympy generates code like which gives an error in Octave 4.0. Would it be better to substitute it with ?"}
{"_id":"q-en-sympy__sympy-13988-d71747c3f5114ce9fd2ab3e38830f26c023bd7dbad25a06ef7af069aaa56fc9b","text":"() should output a Sum() object Currently, () outputs an evaluated summation instead of an unevaluated expression: For large n this takes a long time to compute. It seems like this should output an unevaluated sum and if the user wants to expand the sum they'd call on the result. It may not be worth deprecating this behavior, but maybe we need to have a method."}
{"_id":"q-en-sympy__sympy-14024-2f755bfae4d45b54aff6dee79ff64a771d75f0fc584caa40140fa9b56775af17","text":"Inconsistency when simplifying (-a)x * a(-x), a a positive integer Compare: vs"}
{"_id":"q-en-sympy__sympy-14031-d06e2f3eedf66b108be953dcce14a4863ae80b9d02c2d3550f70110b19df633b","text":"Failed coercion of an expression with E and exp to a field element throws This is the same kind of an issue that dealt with, apparently there is more to be done."}
{"_id":"q-en-sympy__sympy-14038-5933a01d964e44b392a9832dab22b27a1d6b22c85b53af138e23317e945fe920","text":"product(1 - a2 \/ (n*pi)2, [n, 1, oo]) should not evaluate to 0 (if the product is evaluated the correct result is )"}
{"_id":"q-en-sympy__sympy-14070-8f4f938d7ef07747b1bd2edbae809fdc0251cf545461620cd8c515a235673d4d","text":"logcombine(log(3) - log(2)) does nothing Original issue for : Original author:"}
{"_id":"q-en-sympy__sympy-14082-9e7cb03c0f5151b17838ce4d0933c720fc3eda4210ded45154eef8293a7a4e43","text":"Integrate(1\/(x2 + y2), x) returns a wrong result The correct result is . It seems similar to ."}
{"_id":"q-en-sympy__sympy-14085-9123b72746d45ec2d39fee0df1feff71a7c12c247a297e455ee4bd3ec2fb6a1e","text":"sympify(u\"α\") does not work Original issue for : Original author: Original owner:"}
{"_id":"q-en-sympy__sympy-14104-ae12ab9eeac93afaf7b54f43794b0279129f9651ba251b10bb5f136e2c5fc655","text":"pprint(x*DiracDelta(x, 1)) gives TypeError: unorderable types: NoneType() int() I bisected it to commit : CC"}
{"_id":"q-en-sympy__sympy-14166-f93fa8cf30f91411217297a5ab584256f63f684b0aa3239189d05729df629222","text":"Typesetting of big-O symbol Currently typesetting of big-O symbol uses the ordinary 'O', we can use the typesetting as defined here ."}
{"_id":"q-en-sympy__sympy-14180-b68a3e37f00ae3668a01aefe96142c3a24d776fbaf076416bb5554ac84111d9d","text":"Converting to LaTeX was converted to which is in some fields is log10(10) or log2(10). There is '\\ln' in LaTeX and should be converted to it not to"}
{"_id":"q-en-sympy__sympy-14248-c4cdfc2ff76a2dcd4a10ddf41e90b30fe4dd7ef530a8009902aee29bcd404715","text":"The difference of MatrixSymbols prints as a sum with (-1) coefficient Internally, differences like a-b are represented as the sum of a with , but they are supposed to print like a-b. This does not happen with MatrixSymbols. I tried three printers: str, pretty, and latex: Output: Based on a"}
{"_id":"q-en-sympy__sympy-14296-09a174aa8fac18e1b1a4e3ff2dfe1b592191dd6d6f546d10947cf35990769e1a","text":"Sign of generator of an algebraic numberfield calls , but the implementation is defective as it does not change the minimal polynomial. I think this could be fixed in two ways: Add code to create the changed minimal polynomial. Ignore the sign and remove the code changing it. I am inclined to prefer the latter, simpler solution, but I would also like to hear other suggestions."}
{"_id":"q-en-sympy__sympy-14308-240b8f2d9b6dd41bdb6ad7cd53f65a1957075fb698d7d93f6596a952ce6ae5d8","text":"vectors break pretty printing Also, when it does print correctly, the baseline is wrong (it should be centered)."}
{"_id":"q-en-sympy__sympy-14317-4a266ab342226e4f79b6e2769eecc413df327bfb2aef9b3e5d5b0029d62e9161","text":"LaTeX printer does not use the same order of monomials as pretty and str When printing a Poly, the str and pretty printers use the logical order of monomials, from highest to lowest degrees. But latex printer does not."}
{"_id":"q-en-sympy__sympy-14333-caf3424a69e339412e268c638f7cd3a0bba3250d16163a93cefac34b616198f6","text":"Modular inverse for negative modulo and sign resolve SymPy assigns the same sign as . Mathematically, has range for , for . The former sign assignment is used in C\/C++\/Java built-in modulo operator as returns . The latter sign assignment is used in Python's built-in modulo operator as returns . SymPy does not find modular inverses for negative modulo (because of the check). Moreover, as checked from WA (uses the same sign as rule)"}
{"_id":"q-en-sympy__sympy-14396-de7451dfb75ed021f2ec478a3c9b9dca32f0f87a414d087351d55d90a256a1f2","text":"Poly(domain='RR[y,z]') doesn't work Also, the wording of error message could be improved"}
{"_id":"q-en-sympy__sympy-14564-01cd094863e210942b94b8d21a5b06820f94981122e1c6a45f828588a0c8735b","text":"ambiguous behavior of ConditionSet Does that mean the \"set of all x in S for which condition(x) is True\" or \"the set S if the condition(x) is True\"? Also, should there be a doit method or autoevaluation for something like this? Other fixes: should be S.EmptySet should be unchanged: the dummy is not there to help with the condition, only to test idientify where the element from the base_set should be tested in the condition."}
{"_id":"q-en-sympy__sympy-14575-52839642d6156bbafb810904c8847ed9d9eeddf739185cd773462d4dd70f7931","text":"Incorrect binomial documentation The for states: This is the usual definition of the binomial coefficient, but the implementation does not follow this. To be exact, returns , even for negative . See . For example: ` It shouldn't be hard to fix this either way (changing the documentation or the conditionals in ). Is there a preference as to which one should change?"}
{"_id":"q-en-sympy__sympy-14627-3889a2112147800e87c351b0e98618df4e31d09337ec2b630c1c79ebc5c34df7","text":"binomial(n,n) needs simplify to become 1 After , does not become 1. Sure. But even with appropriate assumptions, we need to simplify: was that intentional? Maybe its not important given that DTRT with ... Thoughts?"}
{"_id":"q-en-sympy__sympy-14699-4717f6ade59192c7b38d4a372018ed52af3353dcdb17b13f8fba1cd0f72564d1","text":"orientnew in sympy.physics.mechanics does not support indices Original issue for : Original author:"}
{"_id":"q-en-sympy__sympy-14774-7541baabcd95b87c31f17654d27b6e5a87dc13c33380e3a00831be965f30dc28","text":"Latex printer does not support full inverse trig function names for acsc and asec For example works as expected returning But gives instead of A fix seems to be to change line 743 of from to"}
{"_id":"q-en-sympy__sympy-14817-636fb95db85679664a7df69e3d75229226de0ed0f51b5bdda70f0eb1b9ba3485","text":"Error pretty printing MatAdd The code shouldn't be using sympify to handle string arguments from MatrixSymbol. I don't even understand what the code is doing. Why does it omit the when the first argument is negative? This seems to assume that the arguments of MatAdd have a certain form, and that they will always print a certain way if they are negative."}
{"_id":"q-en-sympy__sympy-14976-7eff69347dcbef152ad333ac9779f3c52f21e0b89ea25875c8a70e02ec05233d","text":"lambdify(modules='mpmath') doesn't wrap rationals This results in reduced precision results from , because the 232\/3 isn't evaluated at full precision. Originally reported at"}
{"_id":"q-en-sympy__sympy-15011-5513217ea9a0cfe8f5b6bbb382046ef6d89d01457fce70f34ceeb2cf9aee941f","text":"lambdify does not work with certain MatrixSymbol names even with dummify=True is happy with curly braces in a symbol name and with s, but not with both at the same time, even if is . Here is some basic code that gives the error. The following two lines of code work: The following two lines of code give a :"}
{"_id":"q-en-sympy__sympy-15017-99241cecc403fb90e163aa7365d328535416fe9af538899ab76f2e2b8a317c8d","text":"of rank-0 arrays returns 0 always returns zero for rank-0 arrays (scalars). I believe the correct value should be one, which is the number of elements of the iterator and the observed behaviour in numpy. In numpy we have the following: This was tested in sympy 1.2-rc1 running in Python 3.6.6 of rank-0 arrays returns 0 always returns zero for rank-0 arrays (scalars). I believe the correct value should be one, which is the number of elements of the iterator and the observed behaviour in numpy. In numpy we have the following: This was tested in sympy 1.2-rc1 running in Python 3.6.6"}
{"_id":"q-en-sympy__sympy-15085-6c8ab12aab75156b560190085b367c09e64623a9172fa79adbdb5206a866352b","text":"[regression] lambdify with Matrix: I'm trying to create a lambda function from a sympy expression that involves a dot product with a . Since at least sympy 1.2, this fails. MWE: Error message:"}
{"_id":"q-en-sympy__sympy-15151-3ff81212e3630b069d01dbd23bae48c62ae54c6840831c441003a648f923b298","text":"not pretty printing indexed(x1, i) not sure if this is expected behavior but i'm expecting x_{1,i} !"}
{"_id":"q-en-sympy__sympy-15225-49cd256514d6e64e428c2137fb67ad18e4f031036e98f262bf38f6b88ca91cb3","text":"xor boolmap equivalent to xnor? - Flaw in finger fingerprint results in The simplified functions fro f1 and f2 are correct, and clearly different, yet bool_map returned what it though to be a valid symbol mapping?"}
{"_id":"q-en-sympy__sympy-15231-7316f6b4b27088979ec724a1dd644bdd1474d9a0e0a6eea1deca2d4310d1552a","text":"autowrap fortran fails with expression containing Mod function twice Minimal example: Error: And here's the generated code. Clearly the problem is that Mod gets imported twice. Hopefully this is an easy fix but I don't know where to start."}
{"_id":"q-en-sympy__sympy-15232-6dedfe521aadd59d9aeda9bb827d84fb5129a2b9e09302956b20139071c7aa97","text":"factor() function issue for non-commutative objects In commit by from line 652 to line 660 in the file , there are some parts that I couldn't understand: I am trying to add a non-commutative class whose freesymbols are commutative. (e.g., operator with time dependence) In this case, even if the object itself is non-commutative, the factor() function gives the wrong result, because of the lines In my understanding, this line treats a non-commutative object as commutative if all its freesymbols are commutative. What is the purpose of this line?"}
{"_id":"q-en-sympy__sympy-15241-f1f32dac05439c6fff3681cebabda4445a3040c4b87ac60fbd4dae6457c940cb","text":"better canonicalization of variables of Derivative Better canonicalization of will be had if any symbols, appearing after functions, that are not in the free symbols of the function, appear before the functions: should equal ."}
{"_id":"q-en-sympy__sympy-15286-0edc0775dc8f1df47ca350a76b1b01508c44f83513beb76e7264fbbd46bfe004","text":"recognize elliptical integrals Original issue for : Original author: Added function for finding equation of Ellipse using slope as parameter and faster method for calculation of circumference of ellipse Added function for finding equation of Ellipse using slope as parameter. Added another method for calculation of circumference of ellipse. Added a new method called Pluralized the following methods -- This PR uses the approach to finding equation of ellipse using slope, length of semi minor axis and length of semi major axis as inputs given This could be an functionality to the equation finding method in class . Thanks to for providing the approach. Please take a look at this PR and suggest changes. I will be glad to implement them. Thanks. <!-- BEGIN RELEASE NOTES --geometry function for finding equation of Ellipse using slope as parameter <!-- END RELEASE NOTES --"}
{"_id":"q-en-sympy__sympy-15304-23a3e4e6db8f7490fbe54c6fe299824fbbe12951ce731603197a709bcb90c8d8","text":"Problem with Beam module I think there is a problem with the Beam module when using a force with an order bigger than 1. This is an example where R should be 1\/3 and remains constant when changing the length of the beam, but unfortunately it's not the case. I believe it has to do with the formation of the load equation, where the higher order compensation isn't been taken into consideration. !"}
{"_id":"q-en-sympy__sympy-15308-e3324b6b4b7dbe898acf45fd79614ff73faae2b39a929ccafcb67492c4ef90eb","text":"LaTeX printing for Matrix Expression The bad part is not only is Trace not recognized, but whatever printer is being used doesn't fallback to the LaTeX printer for the inner expression (it should be )."}
{"_id":"q-en-sympy__sympy-15320-58b0d6ccee102cc36de637161330deb4df3f6b1f5fe26150cc906316986dded7","text":"RootOf might ignore generator This is related to but a little different. I'm not sure of all the ramifications but just want to document this. Note how feeding RootOf an expression in terms of a new generator might ignore that generator if the form of the polynomial is the same"}
{"_id":"q-en-sympy__sympy-15345-be5774fb2fb6915ba2458c3819c007134086c5cf5b89d159d568b68292186864","text":"mathematica_code gives wrong output with Max If I run the code then I would expect the output which is valid Mathematica code but instead I get which is not valid Mathematica code."}
{"_id":"q-en-sympy__sympy-15346-6b47f2a4f5dabe5c18f786bd0ec77f642a3b5ad0a16c09c1edf1143b7f94223a","text":"can't simplify sin\/cos with Rational? latest cloned sympy, python 3 on windows firstly, cos, sin with symbols can be simplified; rational number can be simplified says but says"}
{"_id":"q-en-sympy__sympy-15349-f45e16ece3700b2ce30c090c52fd51d0b4ea088fb8600484aea90bfad715b125","text":"Incorrect result with Quaterniont.torotationmatrix() There appears to be an error in the output. The simplest example I created to illustrate the problem is as follows: One of the functions should be negative. What was the reference of the original equations?"}
{"_id":"q-en-sympy__sympy-15446-1c06dac4bc0573008ab1e72cc6ffb0aad58c8151330f2180a56eb7922b895322","text":"LaTeX printer omits necessary parentheses in matrix products such as x(-y) The product of x and -y, where x, y are MatrixSymbols, is printed as by the LaTeX printer: Source: on Stack Overflow."}
{"_id":"q-en-sympy__sympy-15523-7238737424ce2581e1b3a7cdf986801aaecbf6da643ef24297ec1b40baff2ce8","text":"Python 3.7 test failures Running tests from master under Python 3.7.1 (which isn't tested on Travis) I see some failures. The fastest ones to see are from but there is another in . I'm not sure what has changed in 3.7 to cause this. The error from is"}
{"_id":"q-en-sympy__sympy-15542-782d966575105bf0fc1420dab7131f753b061ea8dd9a805d5d97be82d480811b","text":"Should Point.distance(Line) return distance? In Geometry module, can be used to compute distance, but cannot. Should this be made symmetric?"}
{"_id":"q-en-sympy__sympy-15567-171f0063cb8a710b8ce39b00c2911cbc9c383ec51da066f887830b78cbde90e4","text":"SymPy's doesn't agree with the builtin divmod Both are mathematically correct according to the invariant in the docstring, , but we should be consistent with Python. In general in Python, the sign of mod should be the same as the sign of the second argument. Our is already correct, so it's just that needs to be corrected"}
{"_id":"q-en-sympy__sympy-15586-fcbc64850e5427ae49904b5e5ea4c0cf492ddb0004e44479936685f974e56fa9","text":"Can't get an inverted float matrix with sympy and numpy Trying to get an inverted matrix with sympy and numpy. lambify fails with the following message:"}
{"_id":"q-en-sympy__sympy-15596-8291091a1cb81a0238d194f36178a33657019da590f63ade20df752a42378073","text":"The behavior of degree(f, x) on rational functions I wanted to know if SymPy can compute degrees of rational functions, so I tried Perhaps the degree of a rational function is not implemented, but if so, should this fail instead?"}
{"_id":"q-en-sympy__sympy-15599-f9de2ae32bf5fe5503d83b82906d79a3f5759b4bfc7ca2f54d91ad580e14a326","text":"Mod(3i, 2) unchanged should reduce to (as reported in ) and will do so with a change something like this: Returns correct result to Mod(3i, 2). modified the to return correct answer to Mod(3i, 2). a test (All as suggested by ) Earlier sympify(3k%2) Mod(3k,2) Now sympify(3k%2) Mod(k,2) Release Notes <!-- BEGIN RELEASE NOTES --functions fixed a bug in mod a test <!-- END RELEASE NOTES --"}
{"_id":"q-en-sympy__sympy-15609-16a1aa9a2ff822039e515add1e2f977f79670fe8f8c5e08c3d5ae2355ac28fd3","text":"Indexed matrix-expression LaTeX printer is not compilable The LaTeX string produced by the last command is: LaTeX complains about a double subscript . This expression won't render in MathJax either."}
{"_id":"q-en-sympy__sympy-15635-19aa0d5b70126d025531cd0c8089f4def5db2ff955ad15d362bd2293df0496af","text":"Sets printing issues Some issues with str and srepr printing in sets. Some notes: The printer should always generate valid Python, which recreates the expression (but may require some variables to be defined). The printer should generate an expression that recreates the expression exactly, using only the names from (or other relevant imports for other submodules, but that isn't relevant for the sets). Fancy printing should be relegated to the pretty printers ( and ). Here are the issues I found: The former creates a list, not an interval. The latter isn't even valid Python. (and probably others) isn't a name that is imported from . It should print as . The printers should probably do the same. It's not valid Python. It should print as . Printing as is fine when gets merged. There are likely others. I didn't check too much. An audit of the printing in the sets module would be worthwhile."}
{"_id":"q-en-sympy__sympy-15678-563dda1f698f406d0827575e790c70d2d261adffd6751c70a3faae7baf13a20c","text":"Some issues with idiff idiff doesn't support Eq, and it also doesn't support f(x) instead of y. Both should be easy to correct."}
{"_id":"q-en-sympy__sympy-15685-4d834cf399253b91102015654b7a4f90026f503f16e6db7e3645867cbb96fe40","text":"Make .scalefactor private in the units module sympy version: 1.3 Python version: 3.6.6 Operating System: Win10 Dividing a Quantity with dimension voltage by a Quantity with dimension current yields ohm\/1000 when I expected ohm. In the SI system, 1 V\/ 1 A = 1 Ω. The problem is related to the kilogram workaround and the property . The default scalefactor for a Quantity is 1. The docstring for `scalefactor' states: But, the scale factor for ohm is 1000. ohm.scalefactor 1000 This value of 1000 conflicts with the definition. is a user facing property and should be consistent with the unit system definition, in this case the SI. The kilogram workaround should be an internal implementation factor and not exposed to the user."}
{"_id":"q-en-sympy__sympy-15809-c2c3e36d02b9425eaa63cc356b8cb304db2af83aa497f61cc6722da81f3d9687","text":"Zero-argument Min() and Max() Right now and with no arguments raise . It might be mathematically more convenient to have them return and , respectively. See for why these are valid answers mathematically."}
{"_id":"q-en-sympy__sympy-15875-a124b8fa8f4320af4d4cc78eb61f35fd58d01270df5b7ba015e805207c5354f7","text":"iszero is incorrect on complex integer should return if it cannot decide, but should never give the wrong answer. However: This is causing errors in determining the rank of a matrix. See issue Fixing iszero for complex numbers while Add References to other Issues or PRs Other comments: <!-- BEGIN RELEASE NOTES --- core Fix becoming on some expressions with . <!-- END RELEASE NOTES --"}
{"_id":"q-en-sympy__sympy-15933-1db82af44ff3b7b00c968aea14978f234f41066610961700aab940bcc7190bfc","text":"Wrong measurement for one qubit state. Hi, sympy developers. measure_all(qapply(Qubit('0'))) returns [(0, 1)] best, Vladimir."}
{"_id":"q-en-sympy__sympy-15970-26e72a78f72ae7bc56092c73c5e7350fb188054f22af9a6e5156d766fe79921f","text":"Use '\\ ' instead of '\\quad' for latex of lists, tuples, and dicts See Twitter discussion."}
{"_id":"q-en-sympy__sympy-15971-cc2e46ea59c4cc4199c551d733e44ddd853176303cf7359d1a19bb27d7e49627","text":"Add pretty printing functionality for lerchphi fuction <!-- Your title above should be a short description of what was changed. Do not include the issue number in the title. --<!-- If this pull request fixes an issue, write \"Fixes #NNNN\" in that exact format, e.g. \". See .--Added functionality to pretty print lerchphi function in Fixes the lerchphi part of . <!-- Write the release notes for this release below. See for more information on how to write release notes. If there is no release notes entry for this PR, write \"NO ENTRY\". The bot will check your release notes automatically to see if they are formatted correctly. --<!-- BEGIN RELEASE NOTES --NO ENTRY <!-- END RELEASE NOTES --"}
{"_id":"q-en-sympy__sympy-15976-4ba0833dc6498cc20d1f774ae3e2c77c66c627701c2883945ac42efeabd06590","text":"A symbol ending with a number is made invisible when printing with MathML A variable with a number, such as x1, is made invisible when printing in a MathML format. import sympy from import mathml x2, y, z = sympy.symbols('x2 y z') y = x2z+x23 f = open('', 'w') f.write('\\n') f.write('\\n') f.write('\\n') f.write('\\n') f.write((y, printer='presentation')+'\\n') f.write('\\n') f.write('\\n') f.write('\\n') f.close() Viewing the output in Safari 12.0.2: <img width=\"93\" alt=\"screen shot 2018-12-31 at 12 21 00 pm\" src=\"https:\/\/user-\"If 'x' is used instead of 'x2', it works as expected: x, y, z = sympy.symbols('x y z') y = xz+x3 <img width=\"78\" alt=\"screen shot 2018-12-31 at 12 26 24 pm\" src=\"https:\/\/user-\"BTW, I'm on a MacBook Pro, OS 10.14.2, Sympy 1.3, in Eclipse 2018-19, and Python 3.7."}
{"_id":"q-en-sympy__sympy-16003-73357edda94dc6fa4738e9aff50312a03a374a264ec2daee08eac95b8892404d","text":"MathML presentation printing of multiple derivatives messed up Currently, the MathML presentation printed version of the expression looks like: ! while a proper rending would be more along the lines of the LaTeX equivalent: ! Hence, the method should be improved, first and foremost to print all the derivative variables on a single line and to get the correct power in the numerator. It is also preferred if the actual function ends up on a separate line (not sure if there is some logic to tell when this should or should not happen). If possible, the logic to group adjacent identical terms can be applied, see the discussion and code in which gives an idea of how to implement it. [To be closed] Added printderivative2 methods from <!-- Your title above should be a short description of what was changed. Do not include the issue number in the title. --<!-- If this pull request fixes an issue, write \"Fixes #NNNN\" in that exact format, e.g. \". See . Please also write a comment on that issue linking back to this pull request once it is open. -- As the attached diff in was pretty large due to line endings, I extracted the interesting parts, the methods for LaTex, pretty and MathML printers. Not sure what to do with it. It looked quite promising in the original PR. Maybe one should have a switch to select between these two methods of printing? I have not checked the code more than modifying it to work with current Python and sympy version, at least from a \"no-static-warnings-in-Spyder\"-perspective. <!-- Write the release notes for this release below. See for more information on how to write release notes. The bot will check your release notes automatically to see if they are formatted correctly. --<!-- BEGIN RELEASE NOTES --NO ENTRY <!-- END RELEASE NOTES --MathML presentation printing of multiple derivatives messed up Currently, the MathML presentation printed version of the expression looks like: ! while a proper rending would be more along the lines of the LaTeX equivalent: ! Hence, the method should be improved, first and foremost to print all the derivative variables on a single line and to get the correct power in the numerator. It is also preferred if the actual function ends up on a separate line (not sure if there is some logic to tell when this should or should not happen). If possible, the logic to group adjacent identical terms can be applied, see the discussion and code in which gives an idea of how to implement it."}
{"_id":"q-en-sympy__sympy-16052-e2a04454816100a5d3d1996ad4e69411faee32ac91f855f6cb71c1c97c3a3270","text":"Fix Matrix Equality and MatrixExpr <!-- Your title above should be a short description of what was changed. Do not include the issue number in the title. -- <!-- If this pull request fixes an issue, write \"Fixes #NNNN\" in that exact format, e.g. \". See . Please also write a comment on that issue linking back to this pull request once it is open. --Test case for equality in matrices. <!-- Write the release notes for this release below. See for more information on how to write release notes. The bot will check your release notes automatically to see if they are formatted correctly. --<!-- BEGIN RELEASE NOTES --matrices test for equality in matrices <!-- END RELEASE NOTES --"}
{"_id":"q-en-sympy__sympy-16085-b99a02d0fbdb026b95283c6e17abb02202a3a85b3506f11c9c3e4744c738194d","text":"Make assumptions about indexed symbols I think this should be possible to do: Without assumptions, it's hard to do algebra with indexed symbols, e.g., get to simplify to . I understand that the class Indexed inherits from Expr, not from Symbol, and there are no assumptions on Expr. Still, we already have some quasi-assumptions on Indexed and IndexedBase in the form of class property. And the objects of class Idx, which also inherits from Expr, have them in the form as . Could parse and use the that it accepts, instead of trying to pass them to ? Optionally, maybe even could do that, so an assumption can be to a particular indexed object? By the way, any attempt to pass to is a potential bug because this method, inherited from , does not accept any keyword arguments."}
{"_id":"q-en-sympy__sympy-16106-6c5e604e5b2ca2f37d5b9e3b3269e9ca9d3c609ff05379e135afc8385b974ce7","text":"mathml printer for IndexedBase required Writing an object to MathML fails with a exception: : It also fails for more complex expressions where at least one element is Indexed."}
{"_id":"q-en-sympy__sympy-16221-7134f8c862e3d1deee4ff66174e555e403d1d9263cccc6c8b1608beed6c0154c","text":"Mathematica code: allow printing of matrices and arrays. Our printers for Wolfram Mathematica do not support matrices and arrays. We should add support for it."}
{"_id":"q-en-sympy__sympy-16281-bc392c824520dec54c26d876ab850484bcfb8dcaaef9787d6bfca7199b06bd07","text":"Product pretty print could be improved This is what the pretty printing for looks like: (if those don't look good in your browser copy paste them into the terminal) This could be improved: Why is there always an empty line at the bottom of the ∏? Keeping everything below the horizontal line is good, but the bottom looks asymmetric, and it makes the ∏ bigger than it needs to be. The ∏ is too fat IMO. It might look better if we extended the top bar. I'm unsure about this. Compare this That's still almost twice as wide as the equivalent Sum, but if you make it much skinnier it starts to look bad."}
{"_id":"q-en-sympy__sympy-16331-c969d5538fb898847580c6d49c65162af921be0b3f86fd3a8b8e668fbcea2eab","text":"Issue ufuncifying a two argument function with the Cython backend It works if the function just has one argument, or if you use a different backend. Issue ufuncifying a two argument function with the Cython backend It works if the function just has one argument, or if you use a different backend."}
{"_id":"q-en-sympy__sympy-16334-290f6639e0ddaeb5299915ffaeb7a15f37ae42a0900d61ae145ed6dd611b4bdd","text":"S(0)real(!=0) should be (0 or zoo) and hence non-positive. Consider the following code from master: Since (where a is real and non-zero) should always be (0 or ). Therefore, the result should have been ."}
{"_id":"q-en-sympy__sympy-16422-73a72d9872fedd6f0bb71d3bdf36c9e6577c886a44ff041cf0a52ea7e8b64721","text":"latex printing of BaseScalar It appears that the subscript and script are reversed here for the latex form of BaseScalar:"}
{"_id":"q-en-sympy__sympy-16437-f5fc3e0e20fe9de7de0527564d3df4617705c20e78f50ee44b59e4c81c71785c","text":"Float from string with underscores misplaces decimal point This is correct: None of these are: I think this is an upstream bug in mpmath:"}
{"_id":"q-en-sympy__sympy-16450-b992b8287115584f000e6a54f3c3ccc99824a63067fcf0d7543e0ef6223b1b7a","text":"Posify ignores is_finite assmptions Posify removes a finite assumption from a symbol: I think that posify should preserve the finiteness assumption. Possibly other assumptions should be preserved as well (integer, rational, prime, even, odd...)."}
{"_id":"q-en-sympy__sympy-16474-54b322b431b070e45d8f9d33e9e094a031de54ecd765273b77772a90da752685","text":"Add LaTeX and pretty printers for HadamardPower Furthermore, HadamardProduct may be extended to support the division symbol. [ ] Add latex printer [ ] Add mathml printer [ ] Add pretty printer"}
{"_id":"q-en-sympy__sympy-16503-9b017f71c28fcabd272c8547bcc57b3961954f3540dc23818ca3cc72a0437d4e","text":"Bad centering for Sum pretty print The and the should be aligned. I'm not sure if the should be lower of if the should be higher."}
{"_id":"q-en-sympy__sympy-16527-da2cd240ab37cb111401b124eadccb4e1e873a1d43e31e9d85308eecf69052d5","text":"collectconst() does cannot collect rationals does not seem to support the collecting of numeric rationals in SymPy 1.3. This is probably a duplicate of , but I've provided a description below. In [41]: var('a:d') In [42]: f = a + b + c \/ 2 + d \/ 2 In [43]: print(collect_const(f, Rational(1, 2), Numbers=True)) a + b + (c\/2 + d\/2) I'm expecting"}
{"_id":"q-en-sympy__sympy-16632-9ca6f1d0a5799a50b086be96ac8090b0621d03303cdc2dc7fe28719593ca4238","text":"parsing expressions with powers from sympy.parsing.sympyparser import ( parseexpr, standardtransformations, implicitmultiplicationapplication, ) and if i will do something with expr1 and expr2 i will get wrong results: print(expr1 - expr2) --6n - 3n*2n # Not Zero!!! i have tried simplify, powsimp (force=True too), powdenest, also i have tried create var('n', positive=True, integer=True) and set parameter localdict into parseexpr and use another transformations in parsing, but i could not get 0. Maybe somebody tell me how i can get zero after expr1 - expr2? Thank you"}
{"_id":"q-en-sympy__sympy-16637-e76dcd781a3950d30e9b60fae4d3a8bdc7d188421c532b82e9542d8b3b249c06","text":"Intersection between Plane and Segment3D returns points outside of the segment It seems to me that, as far as ), and so additional intersections that aren't actually within the segment are returned: The expected result is of course that returns an empty list in this case. An easy workaround for the time being is to just filter these out manually: But it would be nice if this could be fixed in sympy itself. Version info: Python 3.5.3 Sympy 1.2-522-g0b350a21a () Updated intersection method in class Plane Updated the isinstance(o, (LinearEntity, LinearEntity3D)) part of intersection method in Plane class so that the returned points lie in both the intersection objects. <!-- BEGIN RELEASE NOTES --geometry updated the intersection method of plane class so that intersection points lie in both the intersection objects if the other entity is LinearEntity. <!-- END RELEASE NOTES --"}
{"_id":"q-en-sympy__sympy-16766-0f83728ae74ffd2257254a5477660cf2c7004a43d4bcf943b82456ff23c76b02","text":"PythonCodePrinter doesn't support Indexed I use to generate some functions and save the code for further use. But the generated code for operation has some warnings which can be confirmed by following code; the output is We should add following method to :"}
{"_id":"q-en-sympy__sympy-16781-f4677f9031653a82cac8123fb35faae675a38fed6efc967ff8616093649e97f3","text":"dotprint doesn't use the correct order for x2 The dot diagram in the tutorial is wrong (). It shows but it should show since it represents , not . I can't figure out how to make dot give the vertices in the right order. Whatever the fix is, we should fix this in the dot printer as well as the tutorial."}
{"_id":"q-en-sympy__sympy-16792-b70863231baa52dd51a8065297a3cf59125b4ad64c8e368ef299e545198fb52d","text":"autowrap with cython backend fails when array arguments do not appear in wrapped expr When using the cython backend for autowrap, it appears that the code is not correctly generated when the function in question has array arguments that do not appear in the final expression. A minimal counterexample is: This should of course return but instead fails with: A little inspection reveals that this is because the corresponding C function is generated with an incorrect signature: ( should be , not in this case) I've found that this error won't occur so long as depends at least in part on each argument. For example this slight modification of the above counterexample works perfectly: This may seem like a silly issue (\"why even have as an argument if it doesn't appear in the expression you're trying to evaluate?\"). But of course in interfacing with external libraries (e.g. for numerical integration), one often needs functions to have a pre-defined signature regardless of whether a given argument contributes to the output. I think I've identified the problem in and will suggest a PR shortly."}
{"_id":"q-en-sympy__sympy-16862-1e14dc69ee6e8310575c94d785afd3fb4c8d911cc31c4ba6da4f66f52a8f9269","text":"Bug in ImageSet I think we should not use . Reason : It works fine when returns a non-empty set. But when it returns an empty set, an error is raised. As calling on an empty set raises an . This is most likely the reason why this is failing for and passing for Also, why are we type-casting the to a ? Ping Maybe you can help me out here."}
{"_id":"q-en-sympy__sympy-16886-a9891c1636430429f68e2bcc4fc798efd8970835c013eac2dc96e86b12a8d4a9","text":"Morse encoding for \"1\" is not correct The current Morse mapping in contains an incorrect mapping of The correct mapping is ."}
{"_id":"q-en-sympy__sympy-16906-3e40f73a3bead0bca5576503d4620f839acf3a9afddc6a977d07ab999740b0da","text":"Added OneMatrix str, pretty, and MathML presentation printing <!-- Your title above should be a short description of what was changed. Do not include the issue number in the title. --<!-- If this pull request fixes an issue, write \"Fixes #NNNN\" in that exact format, e.g. \". See . Please also write a comment on that issue linking back to this pull request once it is open. --Follow up to Added printing of to str, pretty and MathML presentation printers Also improved unicode pretty printing of ZeroMatrix and Identity to use unicode double struck characters. Before: <img width=\"373\" alt=\"beforezeroone\" src=\"https:\/\/user-\"After: <img width=\"249\" alt=\"afterzeroone\" src=\"https:\/\/user-\"(Not clear why the LaTeX renders as it does, it is a correct LaTeX expression...) <!-- Write the release notes for this release below. See for more information on how to write release notes. The bot will check your release notes automatically to see if they are formatted correctly. --<!-- BEGIN RELEASE NOTES --printing Added printing of OneMatrix to str, pretty, and MathML presentation printers. <!-- END RELEASE NOTES --"}
{"_id":"q-en-sympy__sympy-16943-38218575e05e5a8b7794799e83fb059a363afdd233abb38c79bbd23b4733c4a7","text":"doesn't always return a formal power series See the following: I assume the logic is that the value in question is already a polynomial so doesn't need expansion, but it doesn't support the same interface, and doesn't suggest that this is what is supposed to happen. fix issue"}
{"_id":"q-en-sympy__sympy-16988-3b797e76ba3d5245289939f011075d5136a5bedfe71c717ffe7d3aa9ae6e9731","text":"Intersection should remove duplicates The answer should be or remain unevaluated. The routine should give the same answer if duplicates are present; my initial guess is that duplicates should just be removed at the outset of instantiation. Ordering them will produce canonical processing."}
{"_id":"q-en-sympy__sympy-17010-3128aeaa1eae457436e159fd4ffb7071c44e21f111f054c97af9caef3dc02310","text":"Some uses of couldextractminussign can lead to infinite recursion The issue coming from was fixed, but there is another issue has come up in my pull request Several places in use couldextractminussign() in a way that can lead to infinite recursion. For example, in : The problem is that both and could be True, leading to infinite recursion. This happens in my branch for , but probably an example could be found for master too. We need a more robust way to canonicalize this. Ideally we'd want to remove the minus sign from the highest order term. Is there a fast way to do that?"}
{"_id":"q-en-sympy__sympy-17022-8442443fcaa2e26f5cbdd0fd9351cfa42be26a5904afd9e1af26b6d16584b415","text":"Lambdify misinterprets some matrix expressions Using lambdify on an expression containing an identity matrix gives us an unexpected result: Instead, the output should be , since we're adding an identity matrix to the array. Inspecting the globals and source code of shows us why we get the result: The code printer prints , which is currently being interpreted as a Python built-in complex number. The printer should support printing identity matrices, and signal an error for unsupported expressions that might be misinterpreted."}
{"_id":"q-en-sympy__sympy-17038-df129f2783061ab64f451171fbfd2ddaa01295a3ad007ff2051544044a21ada7","text":"isqrt gives incorrect results The function in gives incorrect results for some inputs. For example: or Versions: Python 3.7.3, SymPy 1.4, macOS 10.14.5 For small values of , the uses (aliased to ): The main problem is that the bound used for here is much too large, at almost . If (and it's quite a big if) we can assume that Python floats are IEEE 754 binary64 format and that supplies a correctly-rounded (using round-ties-to-even) square root function, then the largest bound that can safely be used here is , or . If we can assume IEEE 754 binary64 s but can't assume a correctly-rounded , then is still safe for smallish , where the definition of \"smallish\" depends on how accurate is. For example, if is known to be accurate to within 2 ulps, then it's possible to show that is safe for . (The is necessary here: wouldn't be safe even for tiny , since e.g. if the result of is off by a single ulp downwards, would produce instead of .) Whether is correctly rounded or not will depend on the platform: Python's just wraps the function from C's math library. On modern x64 hardware, one would expect and hope that C's gets mapped to the appropriate SSE2 instruction, in which case it'll be correctly rounded. But on ARM there may well not be a hardware sqrt instruction to map to, and a hand-implemented libm sqrt could easily be incorrectly rounded for some inputs. In the unlikely (but possible) case of non-IEEE 754 binary64 s, it's probably safer to avoid using at all. But this case is likely to be exceedingly rare, and doesn't seem worth worrying about in practice. I guess one option for fixing this while retaining performance for small would be to continue to use the code, but then to check the result is correct before returning it, falling back to the slow integer-only path if that check doesn't pass."}
{"_id":"q-en-sympy__sympy-17067-b682fd4385e0394a5c2083137d8d4d18363fd2e4683fb48e1fc6058e70c74f27","text":"Simplify returns incorrect result with trig functions %0A%23--%0Aexpr%20%3D%20(-sin(beta%2F2)sin(alpha%2F2%20-%20gamma%2F2)sin(alpha%2F2%20%2B%20gamma%2F2)%2F(2cos(beta%2F2)cos(alpha%2F2%20%2B%20gamma%2F2)2)%20%2B%20sin(beta%2F2)cos(alpha%2F2%20-%20gamma%2F2)%2F(2cos(beta%2F2)cos(alpha%2F2%20%2B%20gamma%2F2)))%2F(sin(alpha%2F2%20%2B%20gamma%2F2)2%2Fcos(alpha%2F2%20%2B%20gamma%2F2)2%20%2B%201)%20%2B%20(sin(alpha%2F2%20-%20gamma%2F2)sin(alpha%2F2%20%2B%20gamma%2F2)cos(beta%2F2)%2F(2sin(beta%2F2)cos(alpha%2F2%20-%20gamma%2F2)2)%20-%20cos(beta%2F2)cos(alpha%2F2%20%2B%20gamma%2F2)%2F(2sin(beta%2F2)cos(alpha%2F2%20-%20gamma%2F2)))%2F(sin(alpha%2F2%20-%20gamma%2F2)2%2Fcos(alpha%2F2%20-%20gamma%2F2)2%20%2B%201)%0A%23--%0Aprint(mathematicacode(expr))%0A%23--%0A%23%20Using%20Mathematica%20to%20Simplify%20that%20output%20results%20in%20-Cos%5Balpha%5DCot%5Bbeta%5D%0A%23--%0A%23%20That%20is%20also%20the%20result%20that%20one%20can%20get%20using%20basic%20trig%20identities%0A%23--%0Aexpr%0A%23--%0Asimplify(expr)%0A%23--%0A%23%20That%20is%20the%20incorrect%20result%0A%23--%0A]) shows an incorrect result when applying to a fairly large (but ultimately basic) expression involving lots of trig functions. I get the same result on version 1.4 on my own computer, and was getting this result from 1.3 before I updated today. EDIT: Note that Ethan reduced this to a much simpler expression below, so that's obviously a better MWE. I'm sorry that I haven't been able to cut it down to smaller size and still get an error; I have tried. The MWE is this: The output is [which could be further simplified to ]. It should be as verified by Mathematica (directly using the output of ), and by direct calculation using trig identities. This isn't just a matter of refusing to do something that may not always be true; this is really the wrong result. The expression looks big and ugly, but is actually pretty simple when you stare at it for a minute: <img width=\"802\" alt=\"Screen Shot 2019-06-20 at 2 11 30 PM\" src=\"https:\/\/user-\"Those denominators can be simplified immediately to of the respective arguments. Then you just cancel, group terms, and apply standard trig identities, and get the correct result. Sympy will actually correctly simplify each fraction individually, but then it refuses to further simplify the sum — but at least that sum would be correct, unlike what it does with this whole expression. (Of course, this happened in the middle of a much longer calculation, so I didn't know it was going wrong, and have spent the past few days wondering why I was getting inconsistent results when using this wrong answer.)"}
{"_id":"q-en-sympy__sympy-17115-5cb88b296211c748593a7e18fc4d800104d4803d8dcaf997c34e69ed084a4db5","text":"Piecewise doesn't works correctly <!-- The title above should be a short description of the issue. --Code Result It's not working on as well, see the screenshot below, ! [1] told that it is working fine on their systems. Please help me out, either we should fix it or please suggest an alternative approach. I suspect that the error is caused, due to in . May be at least should be allowed to pass through the following loop,"}
{"_id":"q-en-sympy__sympy-17139-02dab6936dfc56e396ed99fa0caacb82bb1ee95063d9f3e5a524ad3ca7f3881e","text":"simplify(cos(x)**I): Invalid comparison of complex I ()"}
{"_id":"q-en-sympy__sympy-17150-b9fe28f32548667c8b56ea59c141e899c586ce4c910940c657573b0591155af0","text":"Incorrect extraction of base powers in log class Evaluating produces , but it should produce . The issue seems to originate around line in during extraction of base powers, where is evaluated to but should evaluate to : if den.isInteger: return n + log(arg \/\/ den) \/ log(base) else: return n + log(arg \/ den) \/ log(base) I would suggest to fix the issue by removing the conditional and keeping the else branch (seems like a case of premature optimization). Alternatively, this also seems to fix the issue: if arg.isInteger and den.is_Integer: return n + log(arg \/\/ den) \/ log(base) else: return n + log(arg \/ den) \/ log(base) That said, seeing that this code was not changed recently, the issue may run deeper."}
{"_id":"q-en-sympy__sympy-17173-05daa61c074839bab61b389bfadc6df02f2fc2bd7dd76f7ad5dbbc8acb756133","text":"Automatic evaluation of RisingFactorial(n, k) with n negative integer, k non-integer This could be evaluated automatically. Note that this causes problems when used in larger expressions, for example:"}
{"_id":"q-en-sympy__sympy-17176-b7075c2a354704d927aaf5f1db509017c8890acd61431674cd6a68c23bae7e84","text":"xor3 boolmap equivalent to xnr3 Extension of Results in: Also due to a flaw in the finger fingerprint routine: Results in an identical fingerprint: This is also broken for XOR4 and XNR4. I haven't tested for more inputs beyond 4"}
{"_id":"q-en-sympy__sympy-17188-94f6873f609a2b501630ab2fa05a5c06192a57179521bc581572a8ffc6851fd4","text":"Piecewise simplification (to avoid doit growth) Perhaps that situation should be recognized during instantiation. Or else might need to call first to avoid repeating what is known (as was reported ):"}
{"_id":"q-en-sympy__sympy-17194-3c5be4d2c6ba8a167a9d2dbbe0ed4e7fefe335c59d05da06d2a24d0ee60fb715","text":"printingMathML erroneous I believe I found some mathML printing errors, and noted below what I think it should be should be: should be: should be:"}
{"_id":"q-en-sympy__sympy-17223-999abf30b4a2f43e4786c63a68e46e983d216617e79023a94cdb8bf90ed6fb68","text":"Match doesn't respect matrix non-commutativity doesn't obey the commutativity of matrix expressions when matching scalars between two expressions. As a demonstration: should be , since the associated matrix multiplications are in a different order (and so not indentical)."}
{"_id":"q-en-sympy__sympy-17251-e7e284e1473820a8bfca3fae7bbb55baf13b26682bda4a5503bf5bdd77259167","text":"exp doesn't simplify based on its periodicity In current master, doesn't use its periodicity to automatically reduce its argument, not even for purely imaginary arguments:"}
{"_id":"q-en-sympy__sympy-17271-f4a2dd2266887a30077da5074dfeebc0b9c16971686be521cd24e784a88c830e","text":"frac(zoo) gives TypeError Not sure what should happen, but not this. I am trying to cover these lines in a test: Clearly, they are covered by calling since the comes from that line, but I do not really want an exception..."}
{"_id":"q-en-sympy__sympy-17273-88c9c3f28387d6564ac73fe17a269e03e9079f931b6723dc0b4712cdbbaa9a23","text":"Incorrect value for hbar in The numerical value given for hbar differs from the CODATA recommended one of: ."}
{"_id":"q-en-sympy__sympy-17288-b2c28ca9e729284362c9916183b24f36cc57e92638a30f25121ebf12beaccc15","text":"Latex printing issue on derivatives with superscripted symbols When I display the derivatives of second-or-more order with superscipted symbols, it gives wrong latex expression and fails to print correctly with IPython environment. For example, this gives instead of nice latex formula. I figure that this is because x^{}^{2} is wrong latex expression. It should be x^{2}, and thus I report it as an issue."}
{"_id":"q-en-sympy__sympy-17313-3b1a3df5ba7999b23cf0c597258d923bc1e357a5739fc2cdc57a9a198511fd73","text":"ceiling(pos) 0 should be true Also, shouldn't , and evaluate to True, too?"}
{"_id":"q-en-sympy__sympy-17318-d9ba92517a065321251aa1491359b04b227965ddebabc3caa9a7d048325be1c7","text":"sqrtdenest raises IndexError If an expression cannot be denested it should be returned unchanged. IndexError fixed for sqrtdenest. Now if the expression can't be denested, it will be returned unchanged. Old Result: New Result:"}
{"_id":"q-en-sympy__sympy-17340-a45915ebce23377faa6391f7e42d669a9f0cd74240e52aeab8dd37a3ec284355","text":"Permutation args are now subtypes of Basic Permutation has currently a subtype in its args. I changed that to"}
{"_id":"q-en-sympy__sympy-17630-e18df66ce9fb1712e6316eba41bedf0b8f2845d701a7eb4d072f12fa05495cbc","text":"Exception when multiplying BlockMatrix containing ZeroMatrix blocks When a block matrix with zero blocks is defined then block-multiplying it once seems to work fine: but block-multiplying twice throws an exception: This seems to be caused by the fact that the zeros in are not but : However, I don't understand SymPy internals well enough to find out why this happens. I use Python 3.7.4 and sympy 1.4 (installed with pip)."}
{"_id":"q-en-sympy__sympy-17653-5f3491a3bcd12c42a750ab8af2365658bbd164a237e9e261d10fa2ec02b45584","text":"new of IndexedBase upcasts subclass of Symbol to Symbol First, thanks for creating SymPy. We're using it in a code generation framework: We noticed that our tests are failling when using the current master of SymPy. To reproduce run: While should work fine. The reason for this failure seems to be in the call of the constructor of . We're a calling it with a subclass of () which is essentially a sympy Symbol with type information. Is there a reason why the out-commented line is necessary? It causes our TypeSymbol to lose its type information by re-constructing a new Symbol. When I out-comment this line everything works fine again. In new of IndexedBase upcasts subclass of Symbol to Symbol First, thanks for creating SymPy. We're using it in a code generation framework: We noticed that our tests are failling when using the current master of SymPy. To reproduce run: While should work fine. The reason for this failure seems to be in the call of the constructor of . We're a calling it with a subclass of () which is essentially a sympy Symbol with type information. Is there a reason why the out-commented line is necessary? It causes our TypeSymbol to lose its type information by re-constructing a new Symbol. When I out-comment this line everything works fine again. In"}
{"_id":"q-en-sympy__sympy-17655-8930bc345b6bcc5e1e62580b5d3607f7c332cf2ce2b3b344f1d11065e3b5c4e7","text":"Unexpected exception when multiplying geometry.Point and number This line works fine But when I write the same this way it raises an exception The expected behaviour is, that both lines give the same result"}
{"_id":"q-en-sympy__sympy-17696-b05cbb5172d83a9ad0d0653cb3b4fbb39c0ed0755c7b741dbd60cc6d878b43ee","text":"Refine with sign Consider the following code: All the returned expression are . However, at least for and , the results should be . This probably is due to the lack of capabilities for . A PR similar to should fix this behaviour. Related issues: and ."}
{"_id":"q-en-sympy__sympy-17720-9ffea696ff9cfc7b09864b911772524655618c8588dc8cce5271bded427020bb","text":"sqrt(sympify('')) gives wrong answer It is easily verified that this answer is not correct; notably:"}
{"_id":"q-en-sympy__sympy-17770-9d742587d2f5e2fe805a0c2f6a2c68b98897eaa4b55a71a2cccd63c8fb9c457d","text":"Corrected functions in subclasses of Corrected the and functions. These functions are still sort of incomplete (couldn't think of all cases, any improvements are welcome). They can directly be used to define properties like and for , , and ."}
{"_id":"q-en-sympy__sympy-17809-57dfb47ebd9819ff0864cd32d9f906ece302d5470fc1d54280bee0b40faef2d1","text":"Modified ispositive for cosh function returns if x is real. Also modified the function to return the function itself or its negative if the function is positive or negative. Example: abs(cosh(x)) should simplify to cosh(x) for real x Sympy 1.0 only simplifies in a limited set of cases:"}
{"_id":"q-en-sympy__sympy-17813-d362d368422ef85c0eb1981a0103c438aa19c2ea69e995e493319fb46549f256","text":"Intersection of ImageSet gives incorrect answer. After git bisecting by this has changed the output of ping -"}
{"_id":"q-en-sympy__sympy-17821-69e47903f32dedfec84cbd8fe22fe76aa70b580a2348edd313d2b50ddff6e923","text":"Catalan rewrite and doctests for latex equations First, implement . Also, something I've been thinking about for while: we have lots of LaTeX in our docs. In many cases we could generate those equations ourselves instead of typing them manually (I found errors while doing for example). This PR should demonstrate the idea. what do you think? Will this work? Its certainly nice for maintainance, although it is probably slightly less readable... (If we want to do this widely, the latex printer could probably be optimized for things like and when it uses instead of .) <!-- BEGIN RELEASE NOTES --core Catalan can be rewritten as a sum <!-- END RELEASE NOTES --"}
{"_id":"q-en-sympy__sympy-18030-4ea7807808cf3c2672aa0c79809d0c9bd0e8834ac29da54101c5a5c6d2a11ecc","text":"interpolate could provide value instead of nan Since the default x value for interpolation is the interpolation at 5 could just return 115 instead of nan."}
{"_id":"q-en-sympy__sympy-18057-81bf751bee61b4f555089299e38db0f4084d3146a6bf2a3c616d690065561897","text":"Sympy incorrectly attempts to eval reprs in its eq method Passing strings produced by unknown objects into eval is very bad. It is especially surprising for an equality check to trigger that kind of behavior. This should be fixed ASAP. Repro code: Results in: On the line: Where code is: Full trace: Related issue: an unknown object whose repr is will incorrectly compare as equal to a sympy symbol x:"}
{"_id":"q-en-sympy__sympy-18062-70c3aa3f567477369de84249d89ac01003143dfd837b1ec05348e9b80b107785","text":"imageset broken for complex numbers With current master: The (or any other value) is simply dropped."}
{"_id":"q-en-sympy__sympy-18087-f2fe8fb4ff754ffc99c83db97bd9dcc03c227182e487078796b12fdfa295ce9c","text":"Simplify of simple trig expression fails trigsimp in various versions, including 1.5, incorrectly simplifies cos(x)+sqrt(sin(x)**2) as though it were cos(x)+sin(x) for general complex x. (Oddly it gets this right if x is real.) Embarrassingly I found this by accident while writing sympy-based teaching material..."}
{"_id":"q-en-sympy__sympy-18109-496a910797b19aa0427c666051605fcadde5229f8c2a82746ba277ac111c8c7d","text":"Why pretty print of \"oo - oo\" has result \"nan\", if the evaluation is disabled?"}
{"_id":"q-en-sympy__sympy-18130-653601074dd3304a695afd7a5f4caed06d79f278c04ed951012770b85a784528","text":"ImageSet of n**2-1 returns EmptySet as intersection with Integers (diophantine bug)"}
{"_id":"q-en-sympy__sympy-18137-46c32ee701d4436641ced9733800669b325e42dc6094732887119a403e62d0d7","text":"Range(1).intersect(FiniteSet(n)) raises TypeError: cannot determine truth value of Relational"}
{"_id":"q-en-sympy__sympy-18168-57bef7825543c226ef57fe7ba58286a84ec4392d47be0890021350dc5b2f6118","text":"Are of 's properties valid? currently, there properties aren't initialized . if there properties are thought wheather Q(S.Rationals) is open or closed set in R (S.Reals), should return and . if use there properties as undefined, should return a exception ? Are of 's properties valid? currently, there properties aren't initialized . if there properties are thought wheather Q(S.Rationals) is open or closed set in R (S.Reals), should return and . if use there properties as undefined, should return a exception ?"}
{"_id":"q-en-sympy__sympy-18189-12b5673a44a4425ba43165d380dbb2082e7670fcd11edd16a03288bb011e1043","text":"diophantine: incomplete results depending on syms order with permute=True diophantine: incomplete results depending on syms order with permute=True"}
{"_id":"q-en-sympy__sympy-18191-7224bfc097612f03822c7671e2b2d0ee040c50421f63735ca351565c796a076b","text":"Recursion error: sqrt(1 \/ tan(1 + I)) There should be an additional check on this line:"}
{"_id":"q-en-sympy__sympy-18199-c9470a3feee1eae707a94a6fbcf06cc5b468aa85d0373321a86e3346f806062e","text":"nthrootmod function misses one root of x = 0 mod p. When in the equation x*n = a mod p , when a % p == 0. Then x = 0 mod p is also a root of this equation. But right now does not check for this condition. has a root . But it does not return it."}
{"_id":"q-en-sympy__sympy-18200-aa91a9303dcb43c7c5b22ef091d962dd5af8af8ee988465d2dfd9269fe495b09","text":"ImageSet(Lambda(n, n**2), S.Integers).intersect(S.Integers) raises AttributeError This is in the related intersection code. See also: , and"}
{"_id":"q-en-sympy__sympy-18211-fc2c0ae3b8cbfb0498b52bf88b64b28d1abd65bbf2b7e0f91d40bc178440ca87","text":"raises instead of returning The problem is Here raises but probably a should be returned by instead. The obvious result of here is Originally posted by in"}
{"_id":"q-en-sympy__sympy-18256-49c3a35341bdb9323fd83040d22cdb234fc47ba2500e8f21e064b8e5506c1c00","text":"Un-parenthesize superscripted symbol Let's think of these superscripted symbols, x^{i}, x^{} <img src=\"^{i},&space;x^{}\" title=\"x^{i}, x^{}\" \/Currently, SymPy parenthesizes these symbols when they are taken to power: \\left(x^{i}\\right)^{2}, \\left(x^{}\\right)^{2} <img src=\"\\left(x^{i}\\right)^{2},&space;\\left(x^{}\\right)^{2}\" title=\"\\left(x^{i}\\right)^{2}, \\left(x^{}\\right)^{2}\" \/However, LaTeX has its own way to represent these symbols without parentheses by nesting them: {x^{i}}^{2}, {x^{}}^{2} <img src=\"{x^{i}}^{2},&space;{x^{}}^{2}\" title=\"{x^{i}}^{2}, {x^{}}^{2}\" \/These are distinguised from 'powered superscription's, which are: x^{i^{2}}, x^{\\^{2}} <img src=\"^{i^{2}},&space;x^{^{2}}\" title=\"x^{i^{2}}, x^{^{2}}\" \/Then, shouldn't it be better to print them without parentheses? As far as I know, that is the conventional way to print power of superscripted symbols. In , you can see that nabla sign is superscripted then powered without any parenthesis: {\\nabla^{}}^{2} <img src=\"{\\nabla^{}}^{2}\" title=\"{\\nabla^{*}}^{2}\" \/"}
{"_id":"q-en-sympy__sympy-18351-c7486627d7e8936c16fd46b8f6a2a28259a1b2faa13a5f091e08ee619cac1bbb","text":"[WIP] Additional matrix support for NumPy printer <!-- Your title above should be a short description of what was changed. Do not include the issue number in the title. --<!-- If this pull request fixes an issue, write \"Fixes #NNNN\" in that exact format, e.g. \". See . Please also write a comment on that issue linking back to this pull request once it is open. -- Adds support for the following matrix expressions to the NumPy printer: [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] <!-- Write the release notes for this release below. See for more information on how to write release notes. The bot will check your release notes automatically to see if they are formatted correctly. --<!-- BEGIN RELEASE NOTES --printing numpy printer support for more matrix expressions <!-- END RELEASE NOTES --"}
{"_id":"q-en-sympy__sympy-18532-219254aeb5cba64b62c0a00f13e3afd2db9069ffc8e561d343f2ebd841b9f818","text":"() should return objects with no args instead of subclasses of Atom with no arguments returns subclasses of in . But the correct definition of a leaf node should be that it has no . This should be easy to fix, but one needs to check that this doesn't affect the performance."}
{"_id":"q-en-sympy__sympy-18587-6649a57270588d82438f939338f64365bc5da03e961287bb4fd5ba6139b8e783","text":"combinatorics.Permutation - exception not raised if wrong size is passed to constructor If I create object from list, which length is greater than argument, then the argument is ignored (and exception is not raised): Is there any reason for such behaviour? It seems to me that it would be better to raise an exception."}
{"_id":"q-en-sympy__sympy-18605-69227f604613f1187ebbe29c1ae478a2e099e76df57c7c71af67c8f3130cb390","text":"Idx object can accepts non-integer bounds It is my understanding that this code should not work: the lower and upper bounds should be integers. Instead it runs without any errors: Note however that: produces the expected result: The check for integer dimension should be done to both upper and lower bounds when a tuple is given as argument."}
{"_id":"q-en-sympy__sympy-18621-6a6783a87c3931df94fccddcf4fdc3425ef62dea5c74e816ded5482d5d0bf1d4","text":"BlockDiagMatrix with one element cannot be converted to regular Matrix Creating a BlockDiagMatrix with one Matrix element will raise if trying to convert it back to a regular Matrix: Instead having two elements will work as expected: This issue exists for sympy 1.5.1 but not for sympy 1.4"}
{"_id":"q-en-sympy__sympy-18630-bd68922ee82667db107e3af2fe37972b81afade3a99c36caab411f68a467178e","text":"AttributeError: integrate(1 \/ (1 + x4)(S(1)\/4), [x, 0, oo]) It is trying to compute the of a hypergeometric function:"}
{"_id":"q-en-sympy__sympy-18633-2ef136f2ff8a94c8753012c880b7aaf51ef9d9775701f6af15fc8c31c36bfe1e","text":"Random test failing in testtensorpartial_deriv It happened in I suspect the problem is the random value sometimes get zero and doesn't evaluate."}
{"_id":"q-en-sympy__sympy-18650-33e905db4922ee3fd478ed22dff0c47bebb54bf3127e635840e4c4497f3a2029","text":"sqrt(8)**Rational(2, 3) doesn't simplify The results should just be ."}
{"_id":"q-en-sympy__sympy-18667-583aa3235a2a8b9f6e0ee6fa9161e249568a2705ac1748100de4e49f6c748910","text":"Added new feature SchurNumber <!-- I have a new feature in the combinatorics module the Schurnumber --The Schur number S(k) is the largest integer n for which the interval [1,n] can be partitioned into k sum-free sets. I have also made the partition which can be proven by induction and I have test cases"}
{"_id":"q-en-sympy__sympy-18698-d7e3cbb9e0c58af59fafa6a8713eddc0bda28057d05f8380a69a3b755d37fd2e","text":"sqf and sqflist output is not consistant The example below is wrong in the sense that we should have (x*2 - 5_x + 6, 3) and not 2 factors of multiplicity 3. whereas below is correct --- one factor of multiplicity 2"}
{"_id":"q-en-sympy__sympy-18728-4b14e890b834f4cb27fcc979c6175a7bd25a89768c20cb780c9ab7a2ff11a8d9","text":"Fix iszero and ispositive for Pow <!-- Your title above should be a short description of what was changed. Do not include the issue number in the title. --<!-- If this pull request fixes an issue, write \"Fixes #NNNN\" in that exact format, e.g. \". See . Please also write a comment on that issue linking back to this pull request once it is open. -- Still a work in progress. This attempts to get working in all cases. This PR fixes so that it doesn't assume finiteness unless justified by assumptions. For example on master: This is incorrect because if a or b can be infinite then it is possible for to be zero e.g. if is infinite and b is negative (a)<1 and b is positive infinite (oo) (a)1 and b is negative infinite (-oo) With this PR we have meaning that could be zero or could not be zero. Also on master we have whereas this PR gives in that case since could be zero and therefore not be positive. This PR also since This comes from the fact that is properly considered nonzero if is declared as finite. At the time of writing we still have this problem with the PR though: I'm not sure that this PR should be merged as is. Many of the changes are adding flags that probably shouldn't be there. Those flags would not be necessary if the other assumptions implied finite=True. For example in some cases a symbol is declared as and with this PR I've had to add also. Probably it would be better to fix those cases by having in the assumptions system. Most of the other cases are for symbols declared as which could also be fixed. Now that I (hopefully) have the tests passing I still want to improve this PR by tidying up the implementation. Also it still doesn't work in all cases as the zoo\/n example above shows. Perhaps instead of merging this I could build new PRs on top. One could add . Then I could follow that with splitting real into extendedreal and real and having as well. That would be a big change though so maybe it would be better to merge this first. <!-- Write the release notes for this release below. See for more information on how to write release notes. The bot will check your release notes automatically to see if they are formatted correctly. --<!-- BEGIN RELEASE NOTES --core Fix iszero assumptions for and when or could be infinite. <!-- END RELEASE NOTES --"}
{"_id":"q-en-sympy__sympy-18763-e389e1a0f7972e5dd267ecaeaa56bf0de2ecce857b27085fc89bdbb4916b6fa8","text":"Incorrect parenthesizing of Subs Here is an example. LaTeX printing of this gives: ! It would be better to be parenthesized to: !"}
{"_id":"q-en-sympy__sympy-18765-f18d3b4ae8bc7e994b4bf0cb82db291e86981c90550c5c903d1838df34e86b62","text":"Introduce 'evaluate' keyword argument in MatAdd and MatMul . Here is an example: I believe it would be better to add option which canonicallizes the object, so that"}
{"_id":"q-en-sympy__sympy-18810-6b2863aa080b3678da45733be8c90117cc182e2c790893e26b7b648bee4d9694","text":"generate_derangements mishandles unsorted perm The following is incorrect: The routine is assuming that the is sorted (though this is not a requirement):"}
{"_id":"q-en-sympy__sympy-18835-9d0e0fd0652d3f44ac260965b5863f7425f87fb176e24bd1e65850976ca987c1","text":"uniq modifies list argument When you iterate over a dictionary or set and try to modify it while doing so you get an error from Python: It would be good to do the same thing from within because the output will silently be wrong if you modify a passed list: I think this would entail recording the size at the start and then checking the size and raising a similar RuntimeError if the size changes."}
{"_id":"q-en-sympy__sympy-18908-0567ac92a9ec724fa38a96c41c6b0a76418cc86f9a29f9a459590eac3d0cf0a0","text":"Add more SciPy functions to code printer Here is a list of special functions supported in SciPy: Many of them are not supported in the SciPyPrinter and should be ."}
{"_id":"q-en-sympy__sympy-18922-acb82e0366ef02a36eb9ccea844ab1dcad4fc3fa7d051be3eb999480b6881249","text":"Incorrect plot with constants I have attached the screenshot of the plot. The x-axis gets moved to y=3, instead of creating a horizontal line at y=3. !"}
{"_id":"q-en-sympy__sympy-18961-84396d1ca1676c71ae689983be363ae004c89d0c9492132f59eebe748e47a416","text":"give a argument Like , I think a argument could be to so the length would be padded with 0s if necessary to reach the given bit length: Tests, too, should be ."}
{"_id":"q-en-sympy__sympy-19007-dc444730f804bfb5c73496a13d5e380edb09f1e4c692bed4c622feb8ae4a17ae","text":"Wrong matrix element fetched from BlockMatrix Given this code: I get this output: is the wrong here. should not be simplified as that element may come from either or ."}
{"_id":"q-en-sympy__sympy-19016-afe98f6e862ecdedc4aa7594b8750d327cb70b106c4f17cd7c744ddd9241903f","text":"isfiniteset property not implemented for Range Currently, returns nothing, since isfiniteset is not implemented in class Range. I'd like to do that. I was thinking of something like this: Any suggestions\/views\/ideas are highly appreciated. I will submit a PR for the above changes soon. Also there are some other issues, like: and don't work for ranges in which one of the elements is a symbolic integer, i.e., Any ideas regarding fixing the same are highly appreciated, I'd really like to fix it."}
{"_id":"q-en-sympy__sympy-19040-442f405f3e9a79f57d65e1304fcf8c217daa957efc951664dfed39e682b2fc0b","text":"Factor with extension=True drops a factor of y-1 I guess this related (or a duplicate of?) This is from stackoverflow: Factor with extension=True drops a factor of y-1 <!-- Your title above should be a short description of what was changed. Do not include the issue number in the title. --Factor with extension=True drops a factor of y-1 <!-- If this pull request fixes an issue, write \"Fixes #NNNN\" in that exact format, e.g. \" (see for more information). Also, please write a comment on that issue linking back to this pull request once it is open. -- <!-- Write the release notes for this release below. See for more information on how to write release notes. The bot will check your release notes automatically to see if they are formatted correctly. --<!-- BEGIN RELEASE NOTES --NO ENTRY <!-- END RELEASE NOTES --"}
{"_id":"q-en-sympy__sympy-19091-23f245500921646d9f841beb524eaad0d55a04147785b23db9326a178b76e63e","text":"Tensor contractions are wrong This is essentially a generalization of . The problem in the current implementation is that contractions are handled before applications of the metric, which leads to incorrect results such as in . In : And thus, the issue is that is being called prior to (where the metric is applied). The reason that this ordering matters is because tensor contraction is itself the abstraction of applying the metric to the tensors that represent psuedo-riemannian manifolds. In essence, it means that we must have it that !; however, this isn't the case here. I've tried tampering with the code above, but by the way tensors have been designed, this bug is essentially unavoidable. As a consequence, the tensor module needs to be refactored in order to get accurate results. (Also, I couldn't help but notice that the last argument to isn't used). had mentioned that he had this sort of refactoring in the works, but based on his fork, progress seems to be slow. I think discussions should be in order for reorganizing how tensors actually represent their components in this module."}
{"_id":"q-en-sympy__sympy-19110-0e8fa7256a2700999771a4b8ba66d5c1b5e29c576d4e8cc21c94f12cb27e962c","text":"ZeroMatrix should not be falsey We have: I don't see any sense in having a ZeroMatrix instance evaluate to False. This happens because of the method defined for : The method is not needed now that Python 2 is not supported. The method is not needed because a should not evaluate to False in a boolean context. The linked lines of code should simply be removed."}
{"_id":"q-en-sympy__sympy-19182-53a08d4a919aa5488d833f454b4533cfaaaedde1a19322b9c4669013c89dc2c7","text":"test('') fails on master ? test('') seems to fail on master. I'm running Python3.5, sympy-dev, MacOS 10.15.3"}
{"_id":"q-en-sympy__sympy-19201-76bc437d4ead6d8b7944ce526df228d3c0d69bb1063743219be36810ec788bc4","text":"Improvement to printing symbolic matrix slices Printing of slices seems sub-optimal, so here are my proposed changes. Let me know if any of this is controversial. Assuming A[:, :] unchanged A[5:, 5:] A[::2, ::2] unchanged A[0, 0] A[0, 0] All of these would affect the str\/pretty\/latex printer. I see no reason to drop the '0' from the start of a slice, but not 'n' at the end, especially since we otherwise never hint at the dimensions of the matrix while printing. Also, brackets are missing, making slices of composites display in an unfortunate way: (A + B)[:2, :2] A + B[:2, :2]"}
{"_id":"q-en-sympy__sympy-19254-91396f736867a06e7786d4f67bfcbd07815d02fcca774d6ea4726ed09cf93e2e","text":"improvement The method can be significantly improved by using the Knuth-Cohen bound instead. After our research with Prof. Ag.Akritas we have implemented the Knuth-Cohen bound among others, and compare them among dozens of polynomials with different degree, density and coefficients range. Considering the results and the feedback from Mr.Kalevi Suominen, our proposal is that the mignottebound should be replaced by the knuth-cohen bound. Also, for mutli-variants polynomials should be replaced appropriately."}
{"_id":"q-en-sympy__sympy-19346-3b9afef630128a68414f6272cfbac93e911ca7b92ce6a289d7084881190b3f08","text":"srepr not printing dict and set properly prints the element in and correctly. However, prints the elements in and wrong. Is this behavior intended? If it isn't, fixing it will be an easy job."}
{"_id":"q-en-sympy__sympy-19487-8716df9a3dac76f3806d5115a2dcfe632552da262d9245d3a6a0c95e4b69af07","text":"Rewrite sign as abs In sympy the function is defined as for all complex non-zero . There should be a way to rewrite the sign in terms of e.g.: I'm not sure how the possibility of being zero should be handled currently we have Maybe should be as well. Otherwise maybe rewrite as Abs would have to be careful about the possibility of the arg being zero (that would make the rewrite fail in most cases)."}
{"_id":"q-en-sympy__sympy-19495-a2d421d78ef567ce424d0e0f2fb28969a36b889ef312020dc418d1b9c1e0ec80","text":"Strange\/wrong? behaviour of subs with ConditionSet \/ ImageSet I'm not sure what to think of the following: The first two outputs are completely as expected, but if I construct a similar ConditionSet with an ImageSet instead of a FiniteSet, a plain gives a strange result (). It's as if the bound variable of the ConditionSet were mistaken for a . Only after having typed the above, I found issue , so I'd like to add that a subs on the plain ImageSet is working as intended:"}
{"_id":"q-en-sympy__sympy-19601-eeb1d8bd9d6dbf3fe607dc684fb8eca0eb6509d48358d8090513fa45799dbe71","text":"sympy 1.6.1 ? Hi, is there any plan on releasing 1.6.1, just like we had 1.5.1 shortly after 1.5? , but the patches we've submitted have already been merged into SymPy master, and we're looking forward to jump to 1.6. Thanks!"}
{"_id":"q-en-sympy__sympy-19637-8e4a2f5995590badd658261ddf13dbca7e08f84a224a0ebbc48dd046f5bba46e","text":"kernS: 'kern' referenced before assignment from import kernS text = \"(2*x)\/(x-1)\" expr = kernS(text) \/\/ hit = kern in s \/\/ UnboundLocalError: local variable 'kern' referenced before assignment"}
{"_id":"q-en-sympy__sympy-19713-a6353340bdc96c01d61d7a2c9dcf4b3cad692fb40c4666062428fc07f2a9bd64","text":"GeneratorsError raised when creating element of fraction field of polynomial ring I see this construction is not possible while it can be constructed from an element of The same error raises for an element of So this can be relevant"}
{"_id":"q-en-sympy__sympy-19783-f756e546b06632ccb343be731f0b3bcb26d920703404dbba3e5400a9b7028765","text":"Dagger() * IdentityOperator() is not simplified As discussed on the mailing list the following does not work."}
{"_id":"q-en-sympy__sympy-19885-17b36d6304b0bfe48027187a7e7d1a3226d10f9f5f444d7f9d5aacbd54fdd18d","text":"Wrong answer when using \"linsolve\" to solve a system of linear equations ! I can get the correct answer when I use the function \"solve\". However, when I use the \"linsolve\", it returns a wrong answer to me. It is a very serious bug and I hope that your team can fix it. The following part is my code: (I also upload it with the name \"\".)"}
{"_id":"q-en-sympy__sympy-19954-3a3a7c86352f2ba0d0d9cca2e36560461947f7d67237b45862745b624296dc44","text":"sylowsubgroup() IndexError I use sympy 1.6.1, with numpy 1.18.5, scipy 1.4.1, under Python '3.8.5 (default, Aug 5 2020, 09:44:06) [MSC v.1916 64 bit (AMD64)]'. The code that I run as the following gives IndexError for sylowsubgroup(): from sympy.combinatorics import DihedralGroup, PermutationGroup, Permutation G = DihedralGroup(18) S2 = G.sylowsubgroup(p=2) Traceback (most recent call last): File \"<input\", line 7, in <moduleFile \"D:\\anaconda38\\envs\\default\\lib\\site-packages\\sympy\\combinatorics\\\", line 4370, in sylowsubgroup blocks = self.minimalblocks() File \"D:\\anaconda38\\envs\\default\\lib\\site-packages\\sympy\\combinatorics\\\", line 2207, in minimalblocks del numblocks[i], blocks[i] IndexError: list assignment index out of range The same error shows up as well when I set: G = DihedralGroup(2*25) S2 = G.sylowsubgroup(p=2)"}
{"_id":"q-en-sympy__sympy-20049-83b8e15d6b134487db3e8c32df7ff9e1c23c7214b9e9f24f0a40437f2e8220cb","text":"() should calculate the velocity if possible If you specify the orientation of two reference frames and then ask for the angular velocity between the two reference frames the angular velocity will be calculated. But if you try to do the same thing with velocities, this doesn't work. See below: The expected result of the should be: I think that this is possible. Maybe there is a reason it isn't implemented. But we should try to implement it because it is confusing why this works for orientations and not positions."}
{"_id":"q-en-sympy__sympy-20115-a6082a853ac490f74da147f4d61dce65c53d667c1cea4377838f60919978a6d4","text":"Lambdify does not identify frac() Hello, this is my first issue ever. The following code gives the following error If there exists a fix, please let me know. Replacing () with just frac() does not help either."}
{"_id":"q-en-sympy__sympy-20131-8a069584c6ae56e3a8a253c8b730b4b2608332630249bce2d8a3655c4895b067","text":"Warn the user when trees of points or trees of reference frames are not self consistent. has Point and ReferenceFrame. These can be positioned and oriented relative to objects of their same type, respectively. The user is expected to define relative positions and orientations in a consistent manner and the relationships among the objects should be tree graphs. It would be helpful to warn the user if they set positions and orientations that create incorrect graphs, for example adding a cyclic branch; the graphs should be acyclic. You can also create inconsistencies when it comes to calculating velocities and angular velocities, which is done automatically if possible. Here is a point example: The velocities of O in N are different depending on if you calculate based on P or Q. It is also impossible to choose between P or Q. Right now, P or Q will be chosen based on which comes first in . We should warn the user that this graph is inconsistent when trying to calculate the velocity of O in N. This same thing can happen with ReferenceFrame. I suspect there will be issues when users have kinematic loops and may want to define the loop of point positions that specify the loop. Currently, that is not supported. If you specify a loop through a succession of setpos() or locatenew() calls, you get an invalid point tree. Kinematic loops have to be dealt with by adding the algebraic equation and forming DAEs instead. These warnings need some careful thought. The first step would be to define precisely what graphs are consistent and non consistent, in terms of 's capabilities and design. Once that is defined, some methods to check for consistencies can be . There will be inconsistencies related purely to position and orientation as well as inconsistencies related to the automated calculation of velocities. There is discussion in this PR that is relevant:"}
{"_id":"q-en-sympy__sympy-20139-b3f368505385a11c74ad0a3bc9ef2d52cad6ea59da2574cf2e500db5038cc1fa","text":"Use Str instead of Symbol for name of MatrixSymbol <!-- Your title above should be a short description of what was changed. Do not include the issue number in the title. --<!-- If this pull request fixes an issue, write \"Fixes #NNNN\" in that exact format, e.g. \" (see for more information). Also, please write a comment on that issue linking back to this pull request once it is open. --<!-- Write the release notes for this release below. See for more information on how to write release notes. The bot will check your release notes automatically to see if they are formatted correctly. --<!-- BEGIN RELEASE NOTES --- matrices will store Str in its first argument. <!-- END RELEASE NOTES --"}
{"_id":"q-en-sympy__sympy-20154-eab5c064dac49626ac688cc726775d61cef91fda55903ac18f8f522a72c2e574","text":"partitions() reusing the output dictionaries The partitions() iterator in sympy.utilities.iterables reuses the output dictionaries. There is a caveat about it in the docstring. I'm wondering if it's really that important for it to do this. It shouldn't be that much of a performance loss to copy the dictionary before yielding it. This behavior is very confusing. It means that something as simple as list(partitions()) will give an apparently wrong result. And it can lead to much more subtle bugs if the partitions are used in a nontrivial way."}
{"_id":"q-en-sympy__sympy-20169-0c0853feb71245f44ec2721d5f486dbaf475fb8a96743fbc63c9e59b84ca2832","text":"Lens makers formula Currently lens maker formula is limited to thin lens. It should also work for plano- lenses(plano-concave and plano convex) and thick lens."}
{"_id":"q-en-sympy__sympy-20212-12426206e7a4ec67696302bd285051716af93f9a019420685e42456cda2fb89f","text":"0-oo produces 0, the documentation says it should produce zoo Using SymPy 1.5.1, evaluate produces . The documentation for the Pow class states that it should return , aka"}
{"_id":"q-en-sympy__sympy-20264-1b7c8eecca06ddc699573fb1ba1a6ce1b9c415b9029b921e62f710543f498e92","text":"RecursionError occured while convert sympy expression to latex when expression like Pow(1\/n,-1) Hello, Thanks for awesome library :) I find issue but I can't figure out why the following phenomena occur, so I write an issue. I use and i tried to latex function to change sympy expression to latex. But RecursionError occured while change Pow(Rational(1,n), evaluate=False) result to latex. It error occured only when (numerator = 1 and denominator=Positive integer bigger than 1) or (numerator = -1 and denominator=Negative integer lower than -1) . Below is code of i runned:"}
{"_id":"q-en-sympy__sympy-20322-0aecc0b4444937bce1d0bbc9deebbc476b2ae9825a59afe20c50f313edc0f76d","text":"Inconsistent behavior for sympify\/simplify with ceiling In sympy v1.5.1: In sympy v.1.6.2: Is there a way to ensure that the behavior is consistent, even though evaluate is equal to when parsing?"}
{"_id":"q-en-sympy__sympy-20428-432d07e9519f4ca13a27a63f2798c2a88a338472a11c0498c8617c2eeb0ffaef","text":"Result from cleardenoms() prints like zero poly but behaves wierdly (due to unstripped DMP) The was the immediate cause of the ZeroDivisionError in . Calling on a complicated constant poly that turns out to be zero: The result prints like the zero polynomial but behaves inconsistently: ( points out this is a bug in ), but other Poly methods don't handle very well. e.g. Also sometime in the last year Poly.primitive has been changed to slightly better handle this bad poly. but in earlier versions of SymPy: which was the cause of the ZeroDivisionError reported in . Looking at the underlying DMP, there is an unstripped leading 0 in the list representation of the Poly which should be"}
{"_id":"q-en-sympy__sympy-20442-0f414d5520151be2b2cb474a9513834fb675449018036f7af076aac52156d9b6","text":"convertto seems to combine orthogonal units Tested in sympy 1.4, not presently in a position to install 1.5+. Simple example. Consider . The convertto behavior is odd: I would expect the unchanged original expression back, an expression in terms of base units, or an error. It appears that convert_to can only readily handle conversions where the full unit expression is valid. Note that the following three related examples give sensible results:"}
{"_id":"q-en-sympy__sympy-20476-28be15c7f54100dc3264539ce3f9c7b53d301c61cc2fe1bbdafd472c6d12cefb","text":"simplify() returns wrong result when applied to a certain trigonometric expression The function from sympy returns a wrong result when applied to the following trigonometric expression: <img src=\"\\cos\\left(\\frac{\\beta}{2}\\right) \\sin\\left(\\frac{\\beta}{2}\\right) \\cos(\\beta) \\cos(\\gamma)^2 \\sin(\\gamma)\"The latter gets simplified to: <img src=\"\\frac{\\cos(2\\beta - 2\\gamma) - \\cos(2\\beta %2B 2\\gamma)}{16}\"instead of: <img src=\"\\cos(\\gamma)\\frac{\\cos(2\\beta - 2\\gamma) - \\cos(2\\beta %2B 2\\gamma)}{16}\"Here is a piece of code that reproduces the issue: Python version: 3.8 SymPy version: 1.6.2 simplify() returns wrong result when applied to a certain trigonometric expression The function from sympy returns a wrong result when applied to the following trigonometric expression: <img src=\"\\cos\\left(\\frac{\\beta}{2}\\right) \\sin\\left(\\frac{\\beta}{2}\\right) \\cos(\\beta) \\cos(\\gamma)^2 \\sin(\\gamma)\"The latter gets simplified to: <img src=\"\\frac{\\cos(2\\beta - 2\\gamma) - \\cos(2\\beta %2B 2\\gamma)}{16}\"instead of: <img src=\"\\cos(\\gamma)\\frac{\\cos(2\\beta - 2\\gamma) - \\cos(2\\beta %2B 2\\gamma)}{16}\"Here is a piece of code that reproduces the issue: Python version: 3.8 SymPy version: 1.6.2"}
{"_id":"q-en-sympy__sympy-20590-38663e20b1a7ed1b3d48fe6c39c32ba453208ce863703bfbb5c9ecdf13ec0557","text":"Symbol instances have dict since 1.7? In version 1.6.2 Symbol instances had no attribute This changes in 1.7 where now exists (and returns an empty dict) I may misinterpret this, but given the purpose of , I assume this is a bug, introduced because some parent class accidentally stopped defining ."}
{"_id":"q-en-sympy__sympy-20639-17e569e1d45734465e6af9a01ca4434555528e11693435d4557cd79db8f7c04d","text":"inaccurate rendering of pi**(1\/E) This claims to be version ; I just merged from the project master, so I hope this is current. I didn't notice this bug among others in LaTeX and str not fooled:"}
{"_id":"q-en-sympy__sympy-20801-a1b014841093cdc30e9c38995435d38a8dd9d36cd24c3b14f44b08122c02a860","text":"S(0.0) == S.false returns True This issue is related to those listed in . As shown by comparing to returns 2 different results depending on the order in which they are compared: Based on the results of comparison to : I assume we would want to return True as well?"}
{"_id":"q-en-sympy__sympy-20916-02c7019b33a28aa9698cecc1fb51dff488a2ee216ed76ad698c09b53a58424ee","text":"pprint unicode does not format subscripts on Greek letters Good: [ -t₀⋅w₀ -t₁⋅w₀ -t₂⋅w₀] Bad: [ -t₀⋅ω0 -t₁⋅ω0 -t₂⋅ω0]"}
{"_id":"q-en-sympy__sympy-21055-4aa273673660132e386a682141f0f61b958e84b00166bcd0465007b3a3a94015","text":"does not understand how to simplify complex arguments Just learned about the refine-function, which would come in handy frequently for me. But does not recognize that argument functions simplify for real numbers. I cann't find any open issues identifying this. Easy to fix, though."}
{"_id":"q-en-sympy__sympy-21171-bebe5b177901e45dc83396728e0582cad2449b7c3bc72ec75dcb2e98d5c1270c","text":"printSingularityFunction() got an unexpected keyword argument 'exp' On a Jupyter Notebook cell, type the following: Then: The following error appears:"}
{"_id":"q-en-sympy__sympy-21208-ab99cc0421cfd020534ef70206c3ee1f4686e06140c3e86cf20be7edacb60785","text":"Results diverge when use on a matrix or its elemetns create a one-element matrix A as below: when use on matrix A: when use on the single element of A: but if use method on A, the result is the same as above: is this a bug or supposed behavior of matrix calculus?"}
{"_id":"q-en-sympy__sympy-21259-1ea053bb061e9c5fe852759940e900881606beeb1694c99393d03b73e561f1d6","text":"Improve asrelational for Range Ranges don't have a rule in that file but they are complicated unless their steps are both the same and the modulus of the starting values relative to the step is the same. Note that the following is wrong (but is outside the scope of this PR): It should be Originally posted by in"}
{"_id":"q-en-sympy__sympy-21271-1407569fd5e1510bb0a9121663c38fe1db22d94dc05c06a4cfeab8a2ef6abf65","text":"Doctest failure in See discussion in . CC There was a doctest failure but the test was disabled in to unblock CI. The cause of the failure remains unfixed though. The failure was:"}
{"_id":"q-en-sympy__sympy-21313-7cf29d166daeedb5afa4f9535e6fff42d16990a69db8af17333233bfb797fef7","text":"don't canonicalize imageset based on Float While writing this about how to get something resembling a float-version for range to work, I tried to think about how I would do this in SymPy. Although Floats present their own difficulties, there is canonicalization being done with expressions that makes this even worse: That last result should represent an error. The reason it happens is because is replaced with : So while canonicalization is OK for Rationals, it should not be done to Floats. Working around this issue, here is a version of that might work for SymPy:"}
{"_id":"q-en-sympy__sympy-21379-af6e6c780791c40be276623df4e2250b2134b3fe0ca193b21e5deb64811bd4ba","text":"Unexpected when using simple for particular expressions I am seeing weird behavior with for particular expressions with hyperbolic sinusoids with piecewise arguments. When applying , I obtain an unexpected . For context, I was umbrella-applying a casting from int to float of all int atoms for a bunch of random expressions before using a tensorflow lambdify to avoid potential tensorflow type errors. You can pretend the expression below has a at the end, but below is the MWE that I could produce. See the expression below, and the conditions in which the exception arises. Sympy version: I am not really sure where the issue is, but I think it has something to do with the order of assumptions in this specific type of expression. Here is what I found- The error only (AFAIK) happens with or in place of , otherwise it succeeds The error goes away if removing the division by The error goes away if removing (but stays for most unary functions, , , etc.) The error only happens with real symbols for and ( does not have to be real) Not too sure how to debug this one."}
{"_id":"q-en-sympy__sympy-21432-377b20c6c0d3c4968bc0d93389d3cc74acc16604a0b7c34614ef755dbd08697a","text":"powdenest(sqrt(sin(x)2), force=True) does not work Since gives , I would expect to be ."}
{"_id":"q-en-sympy__sympy-21476-e5705a748bff2463a11bcd08a0427c7ec0fdd05143e7b67830acff45b894cd41","text":"diffgeom.CoordSystem.transform with indirect relation raises KeyError Below code is a minimal example for an indirect transform via a temporary intermediate coordinate system. raises a KeyError. MWE: This raises a KeyError. DEBUGGING:"}
{"_id":"q-en-sympy__sympy-21527-38d310f5d1742253e7c83524bd5a796ead415325d307c90ecd9b94cd895c55fb","text":"linsolve fails simple system of two equations Any thoughts on why these don't all return the same solution? Thanks!"}
{"_id":"q-en-sympy__sympy-21567-d007c5f3776a42a94c1d9d692bfa2acfcae6cb6a40c9f3dba612f356cb3d5a7c","text":"for , Although we have fixed , it should be extended further for . The behavior for was to expand it in denominator with , so it could be taken as a reference."}
{"_id":"q-en-sympy__sympy-21586-efe24f689ac17dbf26d418fb2188345c30b5cccea16f6eac2580e75ccff427a9","text":"parseexpr with localdict I have 2 expressions that differ only in E*x` and I want to parse it: Right, but if And Thank you!"}
{"_id":"q-en-sympy__sympy-21596-947beda509e89c7bea26226690ba436cf724f552b6f5d64a026f3bab069d82dc","text":"bug in is_subset(Reals) Solving issue has given rise to another bug. Now: This output is incorrect. Correct output is:"}
{"_id":"q-en-sympy__sympy-21612-298788ab33ee3944d606f673bfbfb200f3516ce1e14b0ce27cd567a4be212bf2","text":"Latex parsing of fractions yields wrong expression due to missing brackets Problematic latex expression: is parsed to: . Expected is: . The missing brackets in the denominator result in a wrong expression. 1.8 1.6.2 root python3 Python 3.6.9 (default, Jan 26 2021, 15:33:00) [GCC 8.4.0] on linux Type \"help\", \"copyright\", \"credits\" or \"license\" for more information. from import parselatex parselatex(\"\\frac{\\frac{a^3+b}{c}}{\\frac{1}{c^2}}\") ((a3 + b)\/c)\/1\/(c2)"}
{"_id":"q-en-sympy__sympy-21614-616bcbb2d9c9b65a57974016d679b6ce4c2dd8af79ad2294659f49fe28661a99","text":"Wrong Derivative kind attribute I'm playing around with the attribute. The following is correct: This one is wrong:"}
{"_id":"q-en-sympy__sympy-21627-e24b8c80e588cf3a928d93ca34467ffdd3c5e24956bc9ea112d5ae0dbae8b879","text":"Bug: maximum recusion depth error when checking is_zero of cosh expression The following code causes a error when checked if it is zero:"}
{"_id":"q-en-sympy__sympy-21769-f4a07331ffa74c4720c887b17e3a96bce4d0103cd591e051ac52fca2264bdff7","text":"Latex repr of CG with power does not render correctly There seem to be Latex rendering problems when a Clebsch-Gordan coefficient () is, for instance, squared: ! <details<summaryCode snippet<\/summary<\/detailsI ran this with Sympy v1.7.1 It could be that this is strictly speaking a Jupyter lab\/notebook problem, because the that returns is (I think) valid syntax: Still, a simple fix for would be to wrap the in braces: <details<summaryResult in Jupyter<\/summary! <details<summaryCode snippet<\/summary<\/details<\/detailsRelated issues: and"}
{"_id":"q-en-sympy__sympy-21806-ae885d43e7549d21609abc396c4f66ed5df4bcffc3ed2b233f83aee5d2e31c50","text":"Quaternion class has no overridden evalf method class has no overridden method."}
{"_id":"q-en-sympy__sympy-21847-50589a49800e064cf5c6977fe6bbb5717820b8c166dd097766dc56fe5cd59371","text":"itermonomials returns incorrect monomials when using mindegrees argument returns incorrect monomials when using optional argument For example, the following code introduces three symbolic variables and generates monomials with max and min degree of 3: The code returns , when it should also return monomials such as that also have total degree of 3. This behaviour is inconsistent with the documentation that states that The monomials are also missing when is increased above ."}
{"_id":"q-en-sympy__sympy-21849-aec35bd73715ca9d7d30adb1a9f0cc8d9f81119cf8ecc77a4f191fd840f639a3","text":"LambertW displaying in jupyter lab When using JupyterLab and trying to raise LambertW function to some power (for example multiplying with itself) then it raises error when trying to display it. Same thing happens if I use sp.Pow() method as well. For example sp.Pow(sp.LambertW(2), 2) causes the error as well."}
{"_id":"q-en-sympy__sympy-21930-ccb3733fd7489ee42d4069b2e6a33fad5ced3b88919c5ac5c2c0bf06bd20a6ae","text":"Issues with Latex printing output in second quantization module There are Latex rendering problems within the \"secondquant\" module, as it does not correctly interpret double superscripts containing the \"dagger\" command within Jupyter Notebook. Let's see a minimal example So, it doesn't render correctly, and that's because the double superscript . It should be correct by adding curly brackets"}
{"_id":"q-en-sympy__sympy-21932-2bd933f683fd0215acb058c57749fdaa12676ffe0da7765d198297b1d0c40d79","text":"imageset over Range does not work Simple example: It throws the exception while it tries to check where is the object."}
{"_id":"q-en-sympy__sympy-21952-c3002d918f8744790332180dd2692dccefc9d23c6338fd472d7107dee6ffc53d","text":"If n is even, n**2\/2 should also be even The following: should return , but it returns (of course, this is also an enhancement). That makes me think that perhaps symbolic integers should keep a more complex \"assumptions\" method, which generalizes \"iseven\" and \"isodd\" to instead contain a dictionary of primes that are known to divide that integer, mapping to minimum and maximum known multiplicity. I would like to think about it and post a proposition\/plan, but I am not sure what is the correct github way of doing this. Updated evalisodd to handle more complex inputs Changed the function evalisodd to handle integer inputs that have a denominator, such as The example expression x is recognized by SymPy as an integer, but can be decomposed into n,m,and 1\/2. My new function evaluates the oddness of each part and uses this to calculate the oddness of the entire integer. Addresses issue"}
{"_id":"q-en-sympy__sympy-22080-2c0f016ab900fb09920a6eb381ffdf98dfc4442e59a24dfb07b29ef216928673","text":"Mod function lambdify bug Description: When lambdifying any function of structure like sympy moves the multiplier into the first argument of Mod, like , WHEN we specify This is an example from Sympy online shell"}
{"_id":"q-en-sympy__sympy-22098-08f4bb75edca72531f623a1c72ca9b1b7bcff9acd60f12f6b802c81e3b9033bb","text":"parse_expr partially evaluates under sqrt with evaluate=False Python 3.8.5 (default, Jul 28 2020, 12:59:40) [GCC 9.3.0] on linux with sympy v1.7 I was expecting to get: does not seem to propagate correctly, since the used sympy functions support on their own:"}
{"_id":"q-en-sympy__sympy-22383-6a20e0561042b4f41d626b156eb4b074be7e7d9bf07df7e65e7a417031711ba7","text":"maint(ci): drop testing for Python 3.6 <!-- Your title above should be a short description of what was changed. Do not include the issue number in the title. --<!-- If this pull request fixes an issue, write \"Fixes #NNNN\" in that exact format, e.g. \" (see for more information). Also, please write a comment on that issue linking back to this pull request once it is open. -- describes adopting NEP 29 which would mean dropping support for Python 3.6 now. Drop testing in CI for Python 3.6 and PyPy 3.6 <!-- Write the release notes for this release below between the BEGIN and END statements. The basic format is a bulleted list with the name of the subpackage and the release note for this PR. For example: solvers Added a new solver for logarithmic equations. functions Fixed a bug with log of integers. or if no release note(s) should be included use: NO ENTRY See for more information on how to write release notes. The bot will check your release notes automatically to see if they are formatted correctly. --<!-- BEGIN RELEASE NOTES --NO ENTRY <!-- END RELEASE NOTES --"}
{"_id":"q-en-sympy__sympy-22456-e43b04daba75a62295daa039e40cefaf45f4febb01c5107272b86a7b00f93015","text":"Argument invariance of String Currently, the class does not support argument invariance like: , but instead uses the invariance . The former should hold for any subclass, which is."}
{"_id":"q-en-sympy__sympy-22706-96a9687d658b6bec90a1990a4ebc24eee4cc5c757d1d315d6d4ad7a7048be858","text":"IndexError in StrPrinter for UnevaluatedMul gives if coeffisneg(n[0]): IndexError: list index out of range"}
{"_id":"q-en-sympy__sympy-22714-94308bdf5eceb86dfc2b58e1207c689abd8c5ba75811a8ebab19c665028ab0ad","text":"simpify gives with evaluate(False) crashes unexpectedly with However, it works without . Both of following commands work"}
{"_id":"q-en-sympy__sympy-22740-a54293f4dce15552d9b09be139b4b220119f035879682b17b81690c8a1d6319f","text":"lambdifying Heaviside(Min(x, y)) not working When the original issue in , the following becomes a problem leads to Seems like there is a problem having\/generating select inside of select. leads to which causes a similar error gets rid of the in the Piecewise and works. Hence, it probably makes sense to try to detect the ITE in the Piecewise and if so, rewrite it as And and\/or simplify\/simplify_logic. Related to While writing this issue, I figured out the solution. Will keep this to not have to explain in the PR why I did what I did..."}
{"_id":"q-en-sympy__sympy-22840-36f2f607326f694dfbe176ad0c645c3bdadff93a1bf1201a3f55fa3db2920954","text":"cse() has strange behaviour for MatrixSymbol indexing Example: This gives the following output: and are just copies of the matrices and , respectively."}
{"_id":"q-en-sympy__sympy-22914-37af6d5c0ffdb61618a40b7616f2976316c153c94c389c2e5a47c853d4899c97","text":"PythonCodePrinter doesn't support Min and Max We can't generate python code for the sympy function Min and Max. For example: the output is: Similar to issue , we should add following methods to PythonCodePrinter:"}
{"_id":"q-en-sympy__sympy-22934-04880b773e28c253c808ef44c763cb1abf0818f41d6c057e0e954fe0fa0446c1","text":"x not in QQ.frac_field(1\/x) Again, probably it should."}
{"_id":"q-en-sympy__sympy-22969-ef2b0e3358053d1b0b5c6046092e66ad7018dc090b364277aa4e338a4b6abe73","text":"Gaussian Optics \/ Refractive index not considered Dear sympy maintainers, please correct me, if I am wrong, but I fear that the beam waist in class is not correctly computed. From the source: def w0(self): \"\"\" The beam waist (minimal radius). See Also w : the beam radius at intensity Examples from import BeamParameter p = BeamParameter(530e-9, 1, w=1e-3) p.w0 0. \"\"\" return sqrt() After transmission through a surface with the refractive index changing, the Rayleigh length zr would change, while wavelength stays the same. According to this implementation, w0 changes, which is not physical. If I might help to solve this, I would be happy to contribute. However, I have not a very good understanding of sympy, and this code is interfaced of course with sympy. Best regards, Lukas Gaussian Optics \/ Refractive index not considered Dear sympy maintainers, please correct me, if I am wrong, but I fear that the beam waist in class is not correctly computed. From the source: def w0(self): \"\"\" The beam waist (minimal radius). See Also w : the beam radius at intensity Examples from import BeamParameter p = BeamParameter(530e-9, 1, w=1e-3) p.w0 0. \"\"\" return sqrt() After transmission through a surface with the refractive index changing, the Rayleigh length zr would change, while wavelength stays the same. According to this implementation, w0 changes, which is not physical. If I might help to solve this, I would be happy to contribute. However, I have not a very good understanding of sympy, and this code is interfaced of course with sympy. Best regards, Lukas"}
{"_id":"q-en-sympy__sympy-23021-03594ce3487bda32bb74e8281bce51fd68fffecd4d312c54c8ddfec9d9296d5a","text":"decompogen doesn't like Min\/Max perhaps related to and Add Min\/Max support for decompogen <!-- Your title above should be a short description of what was changed. Do not include the issue number in the title. --<!-- If this pull request fixes an issue, write \"Fixes #NNNN\" in that exact format, e.g. \" (see for more information). Also, please write a comment on that issue linking back to this pull request once it is open. -- Before addition --- After addition --- <!-- Write the release notes for this release below. See for more information on how to write release notes. The bot will check your release notes automatically to see if they are formatted correctly. --<!-- BEGIN RELEASE NOTES --solvers Added \/ support for . <!-- END RELEASE NOTES --"}
{"_id":"q-en-sympy__sympy-23117-ba208072793714e79bc654e2a6089391a7d5ebbe93d2454d194cb20e66f1f6e8","text":"sympy.Array([]) fails, while sympy.Matrix([]) works SymPy 1.4 does not allow to construct empty Array (see code below). Is this the intended behavior?"}
{"_id":"q-en-sympy__sympy-23141-30bfc2ee4abc4081a7e5b224a3f299922f97e7602c83bef559db5211aaae0060","text":"polytopeintegrate and maxdegree Trying to build on the examples as ` works but fails: Is that the expected behavior? Why? Thanks!"}
{"_id":"q-en-sympy__sympy-23191-5b94c4a4f680fb03486f65355031ff4441172382ef9d40c85e1df8c9c2eebb15","text":"display bug while using pretty_print with object in the terminal The following code jumbles some of the outputs in the terminal, essentially by inserting the unit vector in the middle - Output:"}
{"_id":"q-en-sympy__sympy-23262-3162f7c4adfe3c47438ae4fb7fe372bf2ca8fbcce8daf9df43a1768001a35fac","text":"Python code printer not respecting tuple with one element Hi, Thanks for the recent updates in SymPy! I'm trying to update my code to use SymPy 1.10 but ran into an issue with the Python code printer. MWE: SymPy 1.9 and under outputs: But SymPy 1.10 gives Note the missing comma after that causes an integer to be returned instead of a tuple. For tuples with two or more elements, the generated code is correct: In SymPy 1.10 and under, outputs: This result is expected. Not sure if this is a regression. As this breaks my program which assumes the return type to always be a tuple, could you suggest a workaround from the code generation side? Thank you."}
{"_id":"q-en-sympy__sympy-23296-a5f058fe77cd2f985a02e0699e53f362f821a91117b10e711e0e9e48b0c54700","text":"Dimensionless number in exponent not recognized as dimensionless in collectfactoranddimension The following raises an error: returns: Using does not raise an error here: Added fix for wrong dimension calculations in case of some mathematical functions …al functions <!-- Your title above should be a short description of what was changed. Do not include the issue number in the title. --<!-- If this pull request fixes an issue, write \"Fixes #NNNN\" in that exact format, e.g. \" (see for more information). Also, please write a comment on that issue linking back to this pull request once it is open. -- Fixed the return value for the function collectfactorand_dimension(). Previous return value led to wrong dimensional evaluation for expressions containing predefined mathematical functions. <!-- Write the release notes for this release below. See for more information on how to write release notes. The bot will check your release notes automatically to see if they are formatted correctly. --<!-- BEGIN RELEASE NOTES -- Fixed dimensional evaluation for expressions containing predefined mathematical functions. <!-- END RELEASE NOTES --"}
{"_id":"q-en-sympy__sympy-23413-79e73a8da4eaee503cdbf6abf918f814bad4a748acbe73902a944f960a16fa17","text":"bug with HNF removing rows I expect to give but instead I get It seems to be falsely identifying my matrix as rank-deficient and removing the row when I try to achieve a row-style HNF using flips and transposes."}
{"_id":"q-en-sympy__sympy-23534-517d49553815766a496491e5e9f6a8be53792ff3e18e9dcd2dfe06883fb584dc","text":"Using symbols to create functions doesn't work if there is an extra layer of parentheses Sympy version == 1.10.1 Using to create symbol-like objects like instances of as shown in the creates objects of class instead of if there is an extra layer of parentheses. The extra layer of parentheses are necessary to deconstruct the output as separate tuples. Running the code: <class ''<class ''"}
{"_id":"q-en-sympy__sympy-23560-4b11dbf60bca563186489d781ba8d3c69779e8531817b80ccf94e6b61237580c","text":"idiff should use (or better) should not use full blown to implicitly solve for a derivative which will appear in a linear fashion. would be better to use or else simply use a low-level linear solver for the calculation. The following equation takes too long to solve for : The solution can be found as"}
{"_id":"q-en-sympy__sympy-23808-f4fa6bacabc248816af3d4b2c0cc5e709869509851f0073e375b91a6e9964e69","text":"recursion overflow even for for small As title, I am trying to run for and , but it throws a : I am not sure if this classifies as a bug, but it seems unintended especially since 10^6 is such a small number and has a \"easy\" factorisation (10^6 = 2^6 * 5^6)."}
{"_id":"q-en-sympy__sympy-23824-541016ff2b0ef4d47ecddff4eb36f0840fe8d9c14e26d7fcf1549d71a59871ce","text":"() incorrectly reverses order of leading uncontracted gamma matrices The kahanesimplify() function applies such as to simplify products of gamma matrices in which contracted matrices occur. Leading gamma matrices without contractions should be unaffected, but a bug causes such leading terms to be prepended in reverse order. The bug is illustrated by the following example: The result is Both and should simplify to , but the order of and is flipped in the second case due to the bug. I found the source of the bug and it is simple to fix. In the leading matrices are removed at the beginning of the function and then inserted at the start of the product at the end of the function, and the insertion loop is just backward. I'll generate a pull request for this shortly."}
{"_id":"q-en-sympy__sympy-23950-f71013bc7300c512b5b70ca527758e7904928f7b43b15e2855244f408d1e7ef0","text":"returns Contains This is wrong because Contains is not a set (it's a boolean). It results in failures in other places because it doesn't have asrelational (since it isn't a set). For instance, from"}
{"_id":"q-en-sympy__sympy-24066-0de29a25bb998e9bfa1aecc8a477d88ec916cd399cf8b83cd567de90270e0205","text":"SI.collectfactoranddimension() cannot properly detect that exponent is dimensionless How to reproduce:"}
{"_id":"q-en-sympy__sympy-24102-535ecffe253ad0d68353f1840a23da097e1930bffb55c04651b76b7e57e0c55e","text":"Cannot parse Greek characters (and possibly others) in parsemathematica The old Mathematica parser in the package was able to parse e.g. Greek characters. Hence the following example works fine: As of SymPy v. 1.11, the function is deprecated, and is replaced by . This function, however, seems unable to handle the simple example above: This appears to be due to a bug in , which is why I have opened this issue. Thanks in advance! Cannot parse Greek characters (and possibly others) in parsemathematica The old Mathematica parser in the package was able to parse e.g. Greek characters. Hence the following example works fine: As of SymPy v. 1.11, the function is deprecated, and is replaced by . This function, however, seems unable to handle the simple example above: This appears to be due to a bug in , which is why I have opened this issue. Thanks in advance!"}
{"_id":"q-en-sympy__sympy-24152-2d8cca3ba34512a354b8a73f52600baadc4e7de99799ed0b54324a8d611c2ffd","text":"Bug in expand of TensorProduct + Workaround + Fix The expansion of a TensorProduct object stops incomplete if summands in the tensor product factors have (scalar) factors, e.g. This is clearly not the expected behaviour. It also effects other functions that rely on .expand(tensorproduct=True), as e.g. qapply() . Repeat .expand(tensorproduct=True) as may times as there are tensor factors, resp. until the expanded term does no longer change. This is however only reasonable in interactive session and not in algorithms. .expand relies on the method TensorProduct.evalexpandtensorproduct(). The issue arises from an inprecise check in TensorProduct.evalexpandtensorproduct() whether a recursive call is required; it fails when the creation of a TensorProduct object returns commutative (scalar) factors up front: in that case the constructor returns a Mul(cfactors, TensorProduct(..)). I thus propose the following code fix in TensorProduct.evalexpandtensorproduct() in I have marked the four lines to be \/ modified: The fix splits of commutative (scalar) factors from the tp returned. The TensorProduct object will be the one nc factor in nc_part (see constructor), if any. Note that the constructor will return 0 if a tensor factor is 0, so there is no guarantee that tp contains a TensorProduct object (e.g. TensorProduct(U-U, U+V)."}
{"_id":"q-en-sympy__sympy-24213-168806f9595278bbd51fff57e7e98ae33169b487563af2a44c8d515130d76d33","text":"collectfactorand_dimension does not detect equivalent dimensions in addition Code to reproduce: Results in:"}
{"_id":"q-en-sympy__sympy-24325-7b929d243bc7703887d3c64dfc8ed8692753e4ec1ecb46bacd1a83ed62f96996","text":"Numerical error on conversion of coulomb to statcoulomb It should be fixed as : Numerical error on conversion of coulomb to statcoulomb It should be fixed as :"}
{"_id":"q-en-sympy__sympy-24370-c68d58a78c080dd652b70fa5115a1ae58d2de4474d358bff80d38c15fbd476df","text":"Floor division with sympy.Integer gives: Argument of Integer should be of numeric type, got floor(1024\/s0) gives oddly enough, it works if the lhs is a plain Python int. Floor division with sympy.Integer gives: Argument of Integer should be of numeric type, got floor(1024\/s0) gives oddly enough, it works if the lhs is a plain Python int."}
{"_id":"q-en-sympy__sympy-24443-968ad0deca36b074bf3d3cec64ceac65d3e626c278107bd139a0b85a9bdeb74d","text":"is broken on PermutationGroups The issue is in the internal function, where it handles the case of a : When is an inverted generator, the test fails. I think the whole thing can be greatly simplified."}
{"_id":"q-en-sympy__sympy-24455-6ab9e381c2cfc78e821dc504f4dab4c2f7c6512d55e7b2efdf9821a054d3498e","text":"incorrect on \"unnamed\" S3 If we construct a group that's isomorphic to S3, but don't use the function to form it, then is coming up . I think the property method is fine up until the end. It ends like this: Let's talk about what's happening here. First, let be the group, given by generators . For any prime dividing the order of the group, let be the subgroup generated by the th powers of the given generators. The for-loop is saying, \"If is cyclic, then for all dividing the order of the group, ,\" which I think is true. Fine. But the last three lines, after the for-loop, want to say the converse is also true, which it must not be, since appears to be a counterexample. The question is whether there's a good way to save this. In other words, there is always the last ditch, brute force method, where we just look through all the elements of the group for a generator: but it seems like we almost had a better method here, and it would be nice if we could patch it. I'm wondering if anyone whose group theory is not as rusty as mine might know a way. If is the subgroup of consisting of all products of th powers of elements of , it seems like the theorem we wanted might have been true with in place of ? I don't know. But it does seem like the discrepancy between and might be the issue here. In the case of presented as , is just which does have index (hence the failure), whereas and has index , and the test would not have failed."}
{"_id":"q-en-sympy__sympy-24539-690e95abd2f10389ed0e4a2d8cd5b28aa3dee7810df5f1b7af0e66b057c9c802","text":"not accepting symbols The method is supposed to let you set the symbols you want to use, but, as it stands, either you pass the wrong number of symbols, and get an error message, or you pass the right number of symbols, and it ignores them, using instead:"}
{"_id":"q-en-sympy__sympy-24562-4342843a4f642c156cbb270f0eb6a8fc6b0d99516e2d7fe37ca6ba8fb8f0cf13","text":"Rational calc value error python 3.11, sympy 1.11.1 when calc Rational('0.5', '100'), the value is 1\/; but Rational(0.5, 100) the value is 1\/200, this value is the true value, and the version of sympy 1.8 is normal"}
{"_id":"q-en-sympy__sympy-24638-9ee75664bb779ed37898675fbfac9ca013adc47f4e1a74484bb451725a2e8dd4","text":"MatMul(x.T, OneMatrix(k, 1)).diff(x) causes RecursionError The following code creates a RecursionError due to a missing array_derive registration for OneMatrix:"}
{"_id":"q-en-sympy__sympy-24661-f38fd46c47432c58e5f580ad9009818a95435ff9c85416519d0500270a489277","text":"The evaluate=False parameter to is ignored for relationals See also and This inequality evaluates even though is given: The result that should be returned is:"}
{"_id":"q-en-sympy__sympy-24723-701ad062af96e01a5a53065859f734ef00e70ede50244c378adfdb7f438dbe63","text":"Stats: Matrix Normal Distribution returns wrong probability density function As explained on the , the PDF of Matrix Normal Distribution is as follows: ! When initializing a matrix Normal Distribution in Sympy with simple parameters (as mentioned in the docs), the PDF you get is: Whereas the actual PDF should be: References: The difference is in the constant of the PDF and should be a simple issue to solve."}
{"_id":"q-en-sympy__sympy-24909-09341389964ce0ab67db02808fe472a04c9a72e1399a94a9200ba724ff8fa036","text":"Bug with milli prefix What happened: What I expected to happen: milliW should evaluate to milli watts \/ mW or more generally times some unit evaluates to the number 1. I have tried this with Watts and Volts, I'm not sure what other cases this happens. I'm using sympy version 1.11.1-1 on Arch Linux with Python 3.10.9. If you cannot reproduce I would be happy to be of any assitance."}