{"_id":"q-en-vscode-6ebdcd901316ad872ba60e73ed01ade96ab92f3fa4abdc60a5ce1425555412f7","text":"In version 0.10.1 it is no longer possible to write to the root of a drive (which worked up to 0.9.1). When trying to save to e.g. , the following error message appears: Tested on Windows 7 Ultimate and Windows 10 Professional, US language, German region settings.\nare you sure you can save there, can you try with nodepad?\nYes, I can write a text file with notepad. And I also tried other drives. What I found out in the meantime is that this happens only when creating a new file. Opening, editing and saving an existing file works correctly.\nso you do Ctrl+N, type something and then try to save it to r: from the windows save dialog?\nExactly. The error message appears trying to use Save, Save As or (when creating multiple files), Save All.\nis R: some drive created with subst or a real physical drive? I assume it is not the root drive where windows is installed?\nR: is a RAM disk - so I suspected this to be the problem at first. That's why I checked on other, \"real\" drives (except the system drive C:), and I can reproduce the behavior.\nWhat I'm not sure about is why VSCode is trying to run at all. When I do this from a command line (), I get an \"Access is denied\" there as well - which doesn't surprise me.\nIt actually tries to create the root path :) If you run VS Code as non-admin, it won't even let you try to save to but if you run VS Code as admin, it will let you try but fails on creating . !\nThanks, will look into it!\nThis looks like a bad regression in (). It relies on the EEXIST error but when reaching the root drive, it ends in EPERM. I think you touched this function last.\nAs of now, 0.10.8-insider build run as an administrator lets you save to D:<filenameAs far as I can see, this is truly closed."}
{"_id":"q-en-vscode-882802ede8b9d03547478d251c59456fc1b22e9626f6c5e4eefe285345462f71","text":"I can confirm this happens for me as well. I am using Mac OS X. The operating system's display scaling is off (i.e. 1:1). I have been experimenting and it seems as though the bug is that the right click menu is displayed according to the cursor position, but this calculation is not adjusted to take into account zoom. The reason I say this is because if I right click near the top left of the editor, the cursor is quite close to the top left of the menu. But if I right click near the bottom right of the editor, the cursor is much further away from the top left of the menu. The misalignment is proportional to the distance from the top left of the editor."}
{"_id":"q-en-vscode-52e531bee5bd90cb77e3e09dfbe5b5b8773a85b3542a7f9f61f72ece54d6f034","text":"This can happen when code is already running and you start it again from the command line: The process that is started is killed very early after sending a IPC message to the running instance and that running instance is then trying to restore/focus the last active window which often fails. Related: As a workaround if you have access to the windows registry: find change its value to (decimal) restart windows\nPushed a workaround.\nReopening, using the () workaround causes issues as it will undock windows it seems.\nThis happens EVERY time for me - on all 3 machines I use it on, Windows 7 and Windows 10.\nAs a workaround if you have access to the windows registry: find change its value to (decimal) restart windows\nJust tried the suggested workaround and it did not work. Using Windows 10 Home\nI've been hit with this bug suddenly after having style execution from WinSCP work flawlessly for months. On top of that, the suggested workaround doesn't work as is nowhere to be found in my registry. Reading this it looks like the issue was fixed but has there been a regression?\nSeems to have re-emerged for me. On first launch of VSCode it will focus on top of other windows, but on subsequent launches it stays in the background and flashes in the taskbar. I searched the registry but can't find any reference to . I'm on Win10 Build\nMust have made the rookie mistake of not starting the search in the registry root earlier. It turns out there are several hits and the first one in seems to have worked.\nI started running into this issue today. Started when I would open a new window from my current window and the new window would not appear on screen, only in the taskbar. Upon restarting the PC, no windows now appear. Tried the workaround but with no luck. Windows10 Build\nUpdate: I made it to work by right clicking the task bar editor thumbnail of VSC and clicking on move. This made my cursor move to the edge of the desktop. I then dragged the corner of the VSC window to bring it to the center. It looks like VSC window was off the desktop boundary for some reason I have made registry changes to hkeylocalmachine and hkeycurrentuser as mentioned but it still doesn't work. Any other suggestions? Reinstalling doesn't fix the issue either.\nThis is happening to me also since the previous 2 releases. Changing the value of did not help"}
{"_id":"q-en-vscode-ffc1d310b22f02bd796cbc1ca8ae0fd0fc7365ab1cf1dd4f24534509b25aa91e","text":"Unsure how I got into this state but changing different files doesn't fix it !\nJoao maybe you have an idea.\nSure, drag a file from the desktop over the explorer and press <kbdEscape</kbd.\nlooks suspicious in though changing that to call seems to break within-tree-dnd (while it fixes this issue).\nThis code is treacherous since it's full of hacks to make it work across all browsers Monaco used to support."}
{"_id":"q-en-vscode-ba9a6069d992286c78c63b7f608ebd449ed24b890769c11fb123420b5b60da82","text":"Unlike the VS Code installer for Windows, the VS Code uninstaller for Windows is not signed. You can fix this by using the directive in your Inno Setup script. See For dual signing it's easiest to use Inno Setup 5.5.8 instead of 5.5.6 as you currently use: 5.5.8 allows you to have two SignTool directives, one for the sha-1 signing and another for the sha-256 signing.\nTo keep the number of issues in our inbox on a manageable level, we're closing issues that have been on the backlog for a long time but haven't gained traction: We look at the number of votes the issue has received and the number of duplicates issues filed. Thanks for your understanding. Happy coding!"}
{"_id":"q-en-vscode-8b4176da39d3c93c5cc40cca14c1c61bbe3e94c36ce1953d39474a4a514c729b","text":":+1:\n+1\nWhy is this essential feature still not a thing?\nbecause it relies on which is non-trivial and it needs to be prioritized against other work.\n+1\n+1\n+1 or adding a heading list view(TOC view) on the right side, it's really convenient for editing a large markdown file.\nPlease refer to the org-mode tab feature to implement, it's powerful.\n+1\nduplicate of\nReopening to track the markdown specific work of\nI noticed that this is already implemented in the latest insider build, thanks a lot. One more request, would you please preserve the last blank line of a folded region? I mean becomes instead of One reason is that currently\nCan anyone update the status on this?\nit's coming to the next stable release."}
{"_id":"q-en-vscode-4a8c7d1d3d78ee4197bf9cfd5437af67f6ccde3bb0c20d861766158c78491f7b","text":"VSCode Version: 0.10.11 OS Version: Win10 Steps to Reproduce: this key binding: a code file (C#, JS, TS) \"publ\" followed by the space key Expected: The text should be (with a space at the end) Actual: The text is , no autocompletion occurred. Note: \"acceptSelectedSuggestion\" without the caret still works\nLooking thru the default bindings, I noticed this binding at the very bottom of the file: Pressing does not autocomplete in TypeScript either.\nI can confirm it doesn't work on TypeScript due to the context key not being set.\nThere is two problems: commit introduced a regression in that it turns around the boolean condition (missing ) ended up with having the command two times when working on the acceptOnEnter config setting\nDespite the fixes above this still happens. The reason seems to be that uses a timeout. That gives the triggerOnCharacter command a chance to run (and clear the previous suggest model) such that accept doesn't work\nthe timeout and like breakage happened in\nOh, the missing ! Ugh, sorry about that. The was just a leftover. I made a change that enables me to simply remove it. :+1: can you see whether the now works?\nworks"}
{"_id":"q-en-vscode-0b9cfdec6b503a498db47ddc80deec18a5d01396e1a8a70c9431829295703722","text":"VSCode Version: 0.10.11 OS Version: Windows 10 Steps to Reproduce: click on an unknown file type, like or Open with... VS Code and check always use this app to open files of this type The file gets the app icon of VS Code, not the document icon: !"}
{"_id":"q-en-vscode-c01ed4d4800630291b3c41463cad1943b56a15ead9499faac1b1b905453427d0","text":"Issue Id: <-bdb0-aec8-eb4a-</b<br /<br /Versions <br /- <br /- <br /Stack <br /TypeError: Cannot read property 'dispose' of undefined<br /<br /<br /<br /at Array.forEach (native)<br /<br /at Array.forEach (native)<br /<br /<br /<br /<br /\nWoopsie.\nSorry for the confusion, this one is still open, the other one is fixed."}
{"_id":"q-en-vscode-da77aa779aa07c9421f0eb3c7db1cc2691d27232fc9e483d1cb7b34acd45019c","text":"VSCode Version:0.10.13 alpha OS Version:Windows 10 Steps to Reproduce: an variable to your ex. ${env.PATH} \"command\": \"echo\", \"isShellCommand\": true, \"args\": [\"${env.PATH}\"], the task. This will expand correctly on Linux and OS X but not Windows since the expansion is case sensitive and PATH on Windows is \"Path\". Env Var expansion shouldn't be case sensitive so that the same task can run on all platform. This bug was requested by in vscode-docs issue\nthis is done by the resolver service. IMO env variables on all OSes are case insensitive. Should we address this.\nLet's keep it open and if some user is passionate he can tackle it\nOn linux and macOS environment variables are case sensitive, so \"Path\" and \"PATH\" are different variables on linux and macOS (but not on Windows). So your statement from above \"IMO env variables on all OSes are case insensitive.\" is not correct, right? so you are asking for making env var lookup case insensitive on Windows only, correct?\nI thought they are always case insensitive. So it should be case insensitive then on Windows only\nI'm actually about done with a change for this if you want to take a look.\nthanks, I will review the PR.\ndid you want me to squash it or anything (fixed a broken test because I forgot the existing code returned an empty string for missing environment variables)? Different projects have different requirements and I didn't see anything about squashing in your contrib docs.\nno need to squash, I will add the comments to the PR directly if something is needed. Thanks"}
{"_id":"q-en-vscode-b35ceacb553472d21854b18296ff072aa7dc512449d0f7006f1b0e32d803d110","text":"VSCode Version: 1.0.0 OS Version: OSX 10.11.3 I noticed that we can set to a keybinding. Which creates a nice input field in the sidebar to create the directory at the current selected position. ! But there is no possibility to do this for a file. Yes you can bind but it does not do the same thing. It just creates a new and when pressing <kbdcmd</kbd+ <kbds</kbdyou have to choose the destination in a picker. So, is it possible to add something for a new file as well? It would really improve my speed because I don't have to touch my mouse anymore to create a new named file in a specific directory.\nsounds like a good example for a PR?\nYes I looked into it but to be honest, I have to clue where to start :). So if you can point me in the right direction, I would be very happy to do one :).\nI would start by fulltext search for the action id () and that brings you here: From there you can get a feeling how this global action works and maybe add the new one. You will also see the one on top for creating an untitled file (which is not what you want). When you find the references of that action you will see where it is being registered to the system:\nAlright, awesome. Thank you very much Will see what I can do :)."}
{"_id":"q-en-vscode-02bb14a89c4a8c2e27780a7745ba3dfe033743a418665e760ef394ede1895c88","text":"VSCode Version: 1.0 OS Version: Ubuntu 16.04 Ubuntu 16.04 (Xenial Xerus) no longer carries , this causes font to be rendered with if Linux desktop user installed MS core fonts (more than likely). Given Ubuntu still carries open source fonts such as or , which are better for LCD than , consider: Adding those to the font list Remove from font list - let the OS decide\nI see uses of this font in editor, git and debug land. Starting with editor.\nI bumped into this today, after updating to insiders 1.2.0. I'm on Ubuntu 16.10. For now, workaround by manually overriding the font family: Either way, I see no sense in giving \"Droid Sans\" as a first option, even on Linux. If the user has the MS fonts installed, they should be used.\nI also noticed this. I installed \"fonts-noto\" package from the repositories because some people say Noto font family is similar to Droid.\nI solved this by simply installing the font from\nthis is somewhat related to which has similar issues for the UI font. The real fix here is to lean on Chromium to pull in the system monospace font from the OS if it's available (a system-ui font was proposed in ). Is there any reason just is not used? Also why is Courier New there at all?\nfor no good reason except version-to-version stability.\nI only just became aware of this but it's a pretty big issue imo. If I remove my on Ubuntu 16.10 it falls back to Courier New which really affects readability because of its weight and the fact it's a serif font. If this was what I saw as my first impression I'd be kind of turned off by the default font. Hack: ! Courier New: ! Sure it can always be changed but we can't rely on people knowing or bothering to do that. Also on this PC seems to fall back on Courier New.\nMy honest suggestion is to ship VSCode with a cool font such as Hack or Inconsolata and have one of those be the default. IMHO too many folks are left with ugly monospace fonts.\nI just read your comment but I'll post this anyway since we agree :stuckouttongue: have you considered bundling a standard monospace font that looks good on all platforms with the distributable? The developer of reached out to me a couple of months ago and asked if we'd be interested in doing that, gzipped the 4 font variants are only ~700kb.\nYeah, I've mentioned this to and a couple of times already. Not sure about the legal implications, but IMHO that would be a smart move for us.\nHere's the expression of interest I mentioned from the developer of Hack\nIs there a reason why couldn't be that alias for the system monospace font?\nThis now also affects Debian Stretch. After upgrading the font is displayed because the fonts-droid is not in the repositories anymore. Debian maintainers switching from to . I have followed the recommendations and have installed and replaced with in which in my opinion looks very good. Setting the default to on Debian results in a very ugly result. I changed the default font in Gnome Tweak Tool and in fontconfig to . as you can see it is not taken into account: ! Until a font is bundled with VSCode, could we add in the for Linux, and perhaps add as a suggested package in the Debian package? is in the section in Debian repositories. Unfortunately it is in the section in the Ubuntu repositories, thus users must activate it before being able to install the package, this is why I suggest to as a package instead of a recommends/depends.\nIsn't this a Debian bug then? AFAIK you can't change the font in gnome-tweak-tool. It will only set a dconf value, not the alias which is configured via fontconfig.\nNote that I have also defined as the font in fontconfig by defining an alias in: As you can see NotoMono is returned as the maching font for . And I confirm that the monospace font is now rendered using on other applications running on my system, except Visual Studio Code where it seems to default to . I don't think there is a bug in Debian because other application seem to behave correctly. I agree that the correct and clean solution would be to define as the only font listed in . Although: In my opinion on many systems the default monospace font is ugly. If we really want to use as a default then we need to find out why the font is rendered as on some systems when is defined as another font.\nCould be a Chromium/Electron bug then. Can you try in Chromium/Chrome? IMHO that's a bug of those systems and will never be fixed, if applications work around it.\nI have interesting results. I have crafted the following html page: And got the following results. ! I noticed that adding quotes around the monospace word would yield different results. I have changed to in VSCode and it is now rendered correctly: !\nTry Ubuntu Mono. Monaco is good one but personally, I like Ubuntu Mono. !\nHi, everybody! Replace 'Droid sans mono' by 'Ubuntu mono' looks good to me. I didn't download any font pack.\nthere was another issue a while ago discouraging the use of Ubuntu Mono as then it's picked up as the default on non-Ubuntu Linux distros . Ideally we would pick up whatever the display manager configures as the default monospace font, but monospace often doesn't give us that."}
{"_id":"q-en-vscode-3435ff3d8667cca4cd33c3f132027c776988d88995bc11b4d6d6490215bfc1d0","text":"VSCode Version: 1.2.1 OS Version: Windows 10.0. Steps to Reproduce: editing python document typing in a comment a contraction such as isn't that the editor creates a closing apostrophe that is inappropriate in this context.\nNot sure if this is inside the extension or core. But I have seen issues with wrong auto closing often in selfhost too.\nWell... this has been there since day 1. Does anyone know how JavaScript solves this, since it's solved there?"}
{"_id":"q-en-vscode-f0a3bd61efe349fbf08fb5c02fe106c350bbee2cdf98a99b8706a5a4482f0990","text":"VSCode Version: Version 1.3.0-insider (1.3.0-insider) 2016-06-30T05:01:53.185Z OS Version: OSX 10.11.5 (15F34) Steps to Reproduce: a new document, type 'ab'. the editor caret between the two characters. . In any standard OSX text area, this results in the text 'ba'. In VS Code, the result is 'bab', which I cannot imagine justifying as desirable. Note that there is a separate behavior when the caret is at end-of-line: in the EOL case, the previous two characters are transposed. VS Code handles this correctly; it is only the \"in between\" case that is broken (as far as I can tell). Any fix should make sure to test both cases.\nI had a look in the source to see if I could fix this since it's driving me nuts. :) I think it may be an Electron issue since I don't see where this editor command is defined. I'm new to the source code so I may be missing it, but I don't see where a command is defined for ctrl-t.\nThis seems to have gotten worse recently; occasionally ctrl-t will cause multiple lines of text below the cursor to get duplicated into the text buffer.\ndo you have any idea if the likely source of this bug is in the vscode or electron code bases? I also went hunting around in vscode and could not find anything. I tried downloading an electron tutorial app but could not find a text field in which to test the problem. Any guidance here would be much appreciated.\nFor what it's worth the Atom app does not exhibit the broken vscode behavior. However, Atom's implementation has slightly different behavior than 'traditional' bash and native Cocoa controls. The traditional behavior is to transpose characters before and after the cursor AND move the cursor to after the transposed pair, or if there is no character after the cursor (end-of-line case) then transpose the preceding two characters. Atom transposes before and after characters around the cursor but does not move them; for the end-of-line case it does no transposition but instead moves the cursor back by one character.\nI would assume that the solution would be to manually reimplement it, like VSCode already does for ^a ^e ^b ^f ^n ^p This extension does just that. It'd be nice if the functionality of this extension were bundled, since it's basic functionality that works on every other macOS textbox. The Sublime Text selection transpose is just bonus that's really nice.\nVerified that when the cursor is in between 2 letters, they get transposed. But when cursor is at eol, previously last 2 letters were getting transposed. That doesn't happen anymore. ?\nI tried ctrl+t in TextEdit and implemented the behaviour I observed there. Is the behaviour different than what TextEdit does?\nI was basing my comment off of the OP's\nctrl+t in the editor now transposes characters like the default mac text editor. IMHO that is a good thing.\nsadly it only matches half of the behavior; there are two cases. When I reported this bug I noted that the EOL case worked correctly but the in-between case did not. Now as of insiders build it appears that the situation has been reversed. Could we please reopen the issue? Thank you.\nI just noticed that the start-of-line case is also incorrect: TextEdit transposes the newline before the caret with the character after the caret while VSCode simply advances the caret.\nI cannot reproduce this issue. Using VS Code 1.12.1 and default key bindings. Move to have this issue closed?\nThis can probably be closed; I believe Transpose has been rewritten since then for Sublime compatibility?\nVSCode does not match TextEdit behavior in the following cases: (where n is newline and ^ is the insertion caret). In all of these cases VSCode does nothing where TextEdit does a transpose operation.\nThis can be closed. I can't reproduce this issue too. (version: 1.13.1)\nI can still reproduce the cases I describe above on OSX; VSCode Version 1.14.0-insider . I think that in order to close this issue somebody needs to explicitly decide that they want VSCode to ship with default behavior that does not match that of OSX and Cocoa.\nI can reproduce, too. VS Code's and commands both do something different with each other and both different from the macOS/emacs standard."}
{"_id":"q-en-vscode-c556e6bd24c589a108bb68d14ba153a41f4f60737bf96d7a2d124cb7eb6c560f","text":"For: scoped keybinding service scoped progress service For the title area and the editor itself. This requires a common parent (or the editor-container itself) to be reparented instead of the editor moving around. Enables more scenarios around menu and command contributions for contexts.\nI merged this to master: title container, editor container, progress bar widget and the instantiation service are now within a new container that gets reparented when a) you move an editor around or b) you close editor group LEFT when LEFT | RIGHT exist and RIGHT becomes LEFT as well as for when you have 3 editors open and you close the left or center group. Is there a way to verify this works for the scenario you disabled late before we finished June milestone? I remember you reverted a commit and I am not sure if this can be now reverted back again. !\nI can verify\nLGTM - will push the revert revert commit\nVerified via"}
{"_id":"q-en-vscode-fe55011eebfac313ba802ee92e08c432e56d5c1d303a114666f759cd4b4da52a","text":"VSCode Version: 1.3.0-Insider OS Version: MacOSX 10.11.5 Steps to Reproduce: document with long lines (longer then screen width to the right and move the cursor to accordingly -Cursor is invisible\nUnless I'm misunderstanding the issue, this is expected behaviour. How would you expect the editor to behave? This happens on every editor. If you'd like your long line to wrap, you can control that by setting the configuration to .\nNo, this has nothing to do with wrapping long lines. My cursor is literally invisible: !\nInteresting, but I still can't reproduce... Can you provide a screencast or similar?\nWhere is your focus? The cursor blinks/is rendered only when the editor is focused\nI tried reproducing it with new files, but I'm unable to. This happens with a few xsl files at work. I'll try to see if I can get a 100% reproducible file. The focus is in the active editor. I'm only moving the cursor to the right. Screencast: ! EDIT: The cursor is only visible for as long as it doesn't surpass the last character that was visible in the initial view region of the editor\nThank you for the gif! I've never seen that before. Do you see any errors in the console () ? I've quickly tried with a file with these contents but could not reproduce: Other things that might be relevant: does this reproduce when running code without extensions (close it completely and launch with ), did you customize the font and/or picked a different theme?\nI found a way to reliably reproduce this issue. The missing piece was, that it only happens when you have split the main editor. I can't reproduce it with just one editor visible. Launching with doesn't help. I did try with the default user settings and theme as well, but without success. I managed to nail it down to the following steps to reproduce this issue: vscode random file (doesn't matter) the editor with (this is important) random code/text with long lines in one of the 2 editor panels\nForgot to mention: unfortunately there are no errors in the devtools console.\nI'm also seeing this issue. VS Code 1.3.0 commit Windows 10.0. Edit: Oops, didn't notice that a fix was already committed yesterday.\ncan you try my solution in ?\nYes, but disabling extensions doesn't help. This issue only occurs when you have more than a single editor view. See:\nWe decided to fix this by removing the property from the tabs container. The visual impact is that if you use a theme that introduces a custom background color, it will only be picked up for the active tab. E.g. for our themes \"red\" and \"abyss\": ! ! To verify: be on windows have 2 groups open at least have enough tabs open to cause scrolling activate a tab that is at the end of the tab list in group one find a long line and arrow right verify the cursor is not lost\ncan I ask you to verify this on Windows using this 1.3.1 stable build: Thanks!\nI verified this."}
{"_id":"q-en-vscode-85d3d660931fae8f2ff8e625b0ade66357978a55177de2bf305459a62331bffb","text":"There is now a better Intellisense plugins for Vscode PHP: and . In it's known issues it says \"There may be duplicate suggestions for functions or variables. This is because VS Code has it's own (very limited) code-completion support for PHP, and there is currently no way to turn this off.\" This indeed is annoying, the vscode suggestions are too broad and are just noise compared to accurate entries by crane.\nIt should be possible to uninstall the VS Code php extension.\nfyi\nI think crane still uses vscode's syntax highlighter for PHP. So if one were to remove VSCode's internal PHP extension, would it also remove the syntax highlighter? P.S. I think simple options would suffice, e.g. \"\", so it could be turned off.\nYes - but we can also think about splitting our extension into one part that does highlighting and bracket config and another that does completions. That would allow for more freedom.\nThis is now annoying the testing of proper language server in here : Quote: P.S. I think the option would be a simplest thing to implement for now.\nRelated\nAny update on this?\nI have been holding back the PHP IntelliSense release with intelligent completion for over a month now because it is useless if VS Code just spams the completion list with countless functions that do not make sense in the context. When will we get the option to disable or uninstall the built-in PHP completion?\nI agree it makes sense to provide a temporary option to disable this. Just submitted a PR here:\nAlso just checking - at this point in time, is there anything else about the built-in support that is interfering with the experience? (hover, signature help, etc.?)\nthanks for looking into this. Signature help is being implemented right now, iirc there is no built-in hover. It would be great if all of them could be disabled (everything except syntax highlighting).\nI pushed the PR and changes the setting name for the code completion to \"\". If disabled, hover and signature help are also disabled.\nI don't think the setting achieves the desired result. Setting to I no longer get completion suggestions even from PHP Intellisense extension. If have to use to trigger the completion suggestions list.\nI don't think this is related, this only removes the built-in completion provider.\nSo something else which controls the display of suggestions list from extensions like php intellisense is broken?\nPart of the January build. Add to your settings to disable all completions from the built-in PHP extension."}
{"_id":"q-en-vscode-c105775a2570d6090959c87952b7e1283a3ff7307fe64d41ab814bb894b19971","text":"! H1 and H2 are not highlighted. The preview is fine. Tested on VS Code 1.3.1 on OS X 10.11.6\nI wanna give this one a try. Here is what I have achieved so far. !\nCouple of updates on this: to highlight both lines at the same time, the way I was trying above doesn't look practical. It automatically highlights the current line assuming a match of the begin, until it finds a non matching end/while (Basically the line highlights, and if the next line isn't a '-{3,}' then the syntax goes back to normal mode. alt heading 2 syntax (just the -{3,} part) is conflicting with the rule meta.separator.markdown, which takes precedence and hence the issue. Would it be a better option to fix this conflict instead?\nIt doesn't work with 1.9.0 !\nI believe this is as designed. Getting the TM grammar to highlight both the lines is tricky. If you refer to my previous comment, the first point: This would be confusing for the user. In simple terms, it means that as soon as a user starts typing, the line highlights, and goes back to normal syntax if the next line is neither of the alternate H1/H2 (---/===) text. What do you think?\nIs it possible to let the line white and if a line starts with --- oder ===, then this one and the previous one switch to blue? That is the same behavior like in other editors."}
{"_id":"q-en-vscode-2c5bc9958ae84a062d1aa138508650e263528b1db9c3a4ff9c3d0c6d50afc5ad","text":"VSCode Version: 1.4.0 OS Version: Windows 10 Steps to Reproduce: the \"editor.cursorBlinking\" setting to either \"smooth,\" \"phase,\" or \"expand.\" cursor either with mouse or arrow keys. The cursor briefly vanishes (as if it's starting the \"blink\" animation,) then immediately starts the defined animation.\nRelevant animations in PR welcome\nJust merged in the PR from . Thank you !"}
{"_id":"q-en-vscode-378099c03434bc24c0d8b49628bd698b26eabc3c1d312e136026f3386e4f3b3c","text":"VSCode Version: Code - Insiders 1.5.0-insider (, 2016-08-09T16:02:06.523Z) OS Version: Darwin x64 14.5.0"}
{"_id":"q-en-vscode-86c3195eb4c0bbe54150bdbb712db75eb14315ee1914fbe60643c04c41e6b5e1","text":"In order to create custom keybindings to toggle focus between the editor and something such as the terminal panel, a command to focus the active/most recently used editor group needs to be . I believe this is the best that can be accomplished currently: Focuses the first group, meaning if the second or third editor groups were focused this would feel wrong: Or this which focused the most recently used editor group but opens the last editor/file in the group: Some context on request:\nisn't that as easy as providing a global command that just puts focus into the active editor? E.g. ? PR welcome if you wanna give it a try.\nyes, I expect it to be pretty easy.\ngot a better name than ?\nworks for me"}
{"_id":"q-en-vscode-e9408c706edff1967d402e6f66d63a51593144cb468903052c0b2e47b67cf1db","text":"VSCode Version:1.6 insiders OS Version: doesn't matter All settings related to css, sass, less, html, json and markdown. This might be because these come from extensions that are built in?\nI've gone through the builtin extensions and opened new issues for each one that is missing setting localization. Please met me know if I missed any"}
{"_id":"q-en-vscode-11cd8e5dab524da3a34278b974837bc9ac63bb2b1320a5060c00289b4e7e223f","text":"VSCode Version: 1.6.0 OS Version: OSX 10.11.6 *.ini files comments-line starts with the SEMI-COLON. but VSCode has insert (or remove) the HASHMARK.\nPull request pushed, thanks a lot !"}
{"_id":"q-en-vscode-e3ea470a912b4739aaf0edae74bc7bdc1582b53b75877f6b84d8a2657d1ef62b","text":"VSCode Version: Code 1.7.2 (, 2016-11-21T22:14:18.217Z) OS Version: Linux x64 4.8.0-26-generic Focus on debug: ! Focus on extensions: ! Can the bright one be on debug as well?\nFor me all viewlets show the same feedback once focussed: ! Please clarify the issue you see.\nOh I mean via tab focus.\n! still they all look the same to me?\nI see what you mean, when you focus into the current active viewlet, the orange bar to the left shows up stronger. I am not sure who this, I think might be the one?\nThis issue has been closed automatically because it needs more information and has not had recent activity. Please refer to our for filing issues. Thank you for your contributions.\nClosing as designed, no plans to change it currently."}
{"_id":"q-en-vscode-6f017997c6ec8bc3d875a52b662d9148b10d9825b1f80a7038b1595d6ddf4435","text":"VSCode Version: 1.7.2 OS Version: Windows Server 2008/2012 Steps to Reproduce: No shortcut (underline) for flie right click context menu. ! Such underline shortcut is very useful to fast open the file with VSCode with keyboard.\nAre there any recommendations/specifications as to what the shortcut should be for this?\nI think the installer is not in the repo ;)"}
{"_id":"q-en-vscode-0cf061748e5835ab1556e310fe51e88fdc21ddb6c1542032fb00faea2680272f","text":"Steps to Reproduce: panel, maximizes the panel out, the panel is no longer maximized but still much larger than usual panel, maximizes the panel\nFixed by"}
{"_id":"q-en-vscode-856fd74899e6881c52e70ca501ab46ac082e26df5208d8bb11d9879dd957b8d9","text":"Test item VSCode Version: Code - Insiders 1.9.0-insider (, 2017-01-24T10:29:13.076Z) OS Version: Darwin x64 15.6.0 Extensions: None Steps to Reproduce: with: after in some size : : Same goes for single line expansions, although there it is not as bad because it is just one arrow-left key to correct.\nAnother CSS example is . Pressing when done should land the caret at the end of the line.\nThis is due to the way snippets get expanded. Take the below example of a snippet When you use the above in any file and after you enter the placeholder followed by ENTER, the cursor remains where the entry was made and does not go to end of line Is there any way to get the cursor to the end of the snippet after completion?\nYes, is the final cursor position which default to the end of the snippet. The sample above it is used to place the cursor elsewhere. Changing above to should do it.\n! So the being the final cursor position is a standard behavior for snippets in general or is this VS Code specific rule? The reason I ask is, in order to fix this bug if I make an update to then I wonder if I will be breaking any other editor that uses emmet If it is vscode specific, then I'll make the fix on our end\nThat's a good question. It is default Textmate behaviour but I don't know if emmet is also Textmate...\nemmet does borrow from text mate\nVery cool!"}
{"_id":"q-en-vscode-0f3d8e0aa1b9d6024cf8e30b4cc73b88e5559c2006477a5775cbeabb8b0bca5f","text":"This must have broken recently: !\nThe content is actually there in the DOM: !\nJan candidate?"}
{"_id":"q-en-vscode-e4b994a4d5d012e5c0befc42154e413c25267be862d7e272868a99555f7b7591","text":"<!-- Do you have a question? Please ask it on --- VSCode Version: 1.9.1 OS Version: Win 7/64 Home Premium SP 1 Please introduce a setting (ideally changeable via API) to allow an output window (panel) to NOT word-wrap. Reason: I've modified the TODO Parser extension () to write its output in tabular form, which is more concise length-wise. However, the unavoidable word wrapping in the output window usually messes up the formatting. In such cases, it would be preferable for an extension to be able to open an output window that DIDN'T word wrap, and allowed the user to scroll horizontally if required. This suggestion was mentioned as an aside at\nThis should be possible using language specific settings \"[Log]\": { \"editor.minimap.enabled\": false }\nI think this issue was closed in error. It is not about the minimap, and changing minimap settings does not affect the output window's wrapping. This doesn't make any difference either: \"[Log]\":{\"editor.wordWrap\": \"off\"} Here's an image of the output window as it currently behaves with a single long line of text. I'd like the option to have this displayed as one line with scrollbar. !\nThanks for reopening. :)"}
{"_id":"q-en-vscode-2ba0b13b8a069df349ddfa6f58e695f7e90c68c3564f7194ce1c3d0f99a3291f","text":"From For a file such as: We currently do not recognize the file as python even though the internal python extension's contribution matches. The reason is that the entire first line has to match: This seems inconsistent with how TextMate uses , and some of our grammars use patterns that do not work well with this restriction, such as : which fails to match:\nand Adding you since I had trouble tracking down who implemented the logic in originally. Do you know why we currently require the entire line to match? Can we do like the xml grammar and add to the end of some of the contributions?\nsorry I cannot answer this question, the code was a long time ago (I think by ) and since then has not been touched. I am fine changing this so that we align with TextMate's way of doing things. Happy to review a PR if you are interested."}
{"_id":"q-en-vscode-4efe546184406279c762ab890cb6894f3268018248efc5ad976a0b2a7cc36078","text":"When I put a fenced code area in the middle of a bulleted list, I notice that list entries following the fenced code area lose their formatting, like so: ! Is this intentional behavior? Am I formatting my markdown wrong? Is there any way that I can have formatting in both the fenced code block and the subsequent list items?\nThe current formatting is somewhat off: the second list is treated as code block. ! This is expected and is in keeping with the common mark spec. The correct formatting would be to indent fenced code block as well: javascript for(var i = 0; i < 5; i++){ (\"Hello\"); } ! Our syntax coloring is off here though, with the indented code block no longer being highlighted as js\nOkay. I didn't read the markdown specification closely enough to understand that the code block should be highlighted as well. I had tried indenting the code block, but when the indented code block lost syntax highlighting, I thought I was doing something incorrect. Should I update the title of this issue to indicate that the real problem is that indented code blocks lose formatting?\nThanks I've just gone ahead and updated the title of the issue to: \"Markdown Fenced Code Block in Indented List Not Properly Highlighted\". If you do not feel that the new title accurately captures the bug here, feel free to update the title again\nNew title looks good. Thanks!\nI made an extension to fix this Surely it would cause side effects but it is worth it for me at the moment."}
{"_id":"q-en-vscode-50aa8faff7bc7876dec14a6e056a690afc97835e9a4778d7813b49d227af3acc","text":"VSCode Version: 1.10.2 (code1.10.2-) OS Version: Ubuntu 16.04 LTS Steps to Reproduce: home directory w/ on AFS. . the following error: Code starts normally if moving the directory to a local filesystem and symlinking it in. I would reckon that it's probably not portable to put the kind of thing that is in the homedir, as it may very well be mounted in fun ways or on network filesystems. This behaviour is new since the previous release I ran, which I believe was .\nSquinting at the it seems like sockets and other runtime files shall be created in , which on my system is down in .\nNot sure what to do here. Best I can do is to recommend not using AFS. Does it support socket files?\nI don't have any choice in AFS for my home directory where the default XDG config directory is located. Sure I could point the config elsewhere, but then I'd lose access to the same configuration across machines which is the whole point of a shared homedir. AFS doesn't support named pipes in the filesystem intentionally as it's a distributed filesystem and the semantics for how they would work across the world is hard to pin down. As far as I interpret the XDG specification, you should not use the config directory for runtime state. They explicitly mention the runtime dir for that. I don't know where in the stack this is done, but it's quite wrong to use the possibly shared homedir for machine-local state. supports things like 's named pipes\nGot it, thanks.\nthe same thing for the socket files in /tmp/ like /tmp/vscode--a2f2-4ba7-8e8c- they should go in $XDGRUNTIMEDIR (if available, if not /tmp is ok). the benefit of $XDGRUNTIMEDIR is it goes away when the user logs out.\nThe same issue appears with sshfs mounted home directory\nWe were force to undo this change since it breaks many assumptions at startup, when this variable may or may not be set. Unfortunately, the correct workaround for these cases is to call code with .\nHi! Now 1.17. stopped working on AFS Not sure if the workaround you gave makes sense, as the $XDGRUNTIMEDIR directory gets deleted every time the user restarts. I would be happy if you could fix that. Thanks!\nDoes the workaround work for you at all?\nThanks for the fast reply. It does work for one session, but settings for example get deleted when I reboot.\nWhy is that? I don't really know AFS.\nMe neither, but what I think right now: The XDGRUNTIMEDIR variable gives something like on my machine, which apparently gets deleted when rebooting or the XDGRUNTIMEDIR changes. I can’t say right now as I am not at my computer. If you call code using your workaround, all user data gets stored in that directory and is inaccessible after reboot. But the problem ia not the user data, it is the .sock file that has trouble with AFS. So it is just this file that needs to be stored somewhere else in a non AFS directory.\nCurrently the files are within user data dir. So... in your system should they live somewhere else? ?\nYeah, for example. I don’t know what the .sock file does, but if it is not a problem that it is removed after reboot that might be an option. But isn’t that what the pull request did? Just storing the .sock file somewhere else...\nYes, did exactly that, but based on an env variable, which is very problematic in many cases. So, I propose to introduce a cli option to set the socket files location on Linux. So you'd need to launch it as: Would that work for you?\nCool, I would expect that to work for me. Thanks! Let me know if you want me to test.\nproblematic in what cases? the XDG recommendations are pretty clear and easy to follow. You have a DATA dir, CONFIG dir and RUNTIME dir. There are environment variables to specify the directories or there are defaults when the env vars are not there. What doesn't work in vscode with that setup?\nLet's talk Linux only. It makes a big difference in terms of process environment if you open an application from within a shell or using a UI gesture. The shell usually has all you favourite environments, often set up in files or similar (think , ). The process (usually some Gnome or KDE process) which spawns Code using the UI gesture doesn't have this environment because it never went through a shell, never really parsed and interpreted . So, users complain they can't use Node or Go within Code; the real problem is that Code's env doesn't match what their shell has. So: on startup, and only if Code wasn't launched in a shell, Code will the user's shell environment as best as it can. This is possibly a lengthy process (have you ever waited 1-2 seconds until you see that bash prompt?) so we can't have that as part of Code's critical startup sequence, otherwise we risk being slow to start as well. We do it in parallel while Code actually starts up and spawns a window. Eventually, when that env is know, it is promptly applied to all Code's processes. There's a race condition here. You see, we need to know the exact location of the IPC sockets at startup, before any window opens. If is (re)defined in the user's , we can't rely on it to compute the location of the sockets. We must know that value independently of the env loading code path. For the past few months we were relying on reading to determine the sockets paths. For some, this works well. For others, Code breaks in very unrelated ways: if is (re)defined in , there was a race condition between reading that value from and computing the sockets paths. It would often happen that the process which served the socket pointed one way while all other processes pointed the other way. We simply can't have that. One alternative for a fix is to ignore the value coming in from the user's . Do you think this would make sense?\nXDGRUNTIMEDIR is typically set by pamsystemd, even before their Desktop Envrionment or login shell is started. If people redefine that in .bashrc (for interactive shells only) a lot of things will break for those shells. just an example of the sockets I have there (i.e. software that would break if I changed tne env var): So in conclusion I have not heard of anyone redefining XDGRUNTIME_DIR for interactive shells only, but if they do that, that's the users mistake to break things intentionally.\nAlright, then let's do that: we'll simply ignore it when reading it after startup."}
{"_id":"q-en-vscode-1908d480043e9cb5ecea3fe01b24c8eac437de93a441279ac01fae540b910758","text":"I love the peek window. It's pretty much the best thing ever. I would love it ever so slightly more if VSCode would auto-focus it and insert the cursor where it belongs. e.g. if I clicked \"peek definition\", the cursor should be on the definition of what I peeked. This idea seems so obvious I can't believe it's not part of the default implementation, but I know you guys have your reasons for doing things. If there is such a reason, could this at least be made a setting? Thanks guys!\nFiar request. The reasoning for focusing the tree of the left that you might wanna quickly skim through all references/definitions/etc to get fast overview. That's also the reason for changing the peek editor when the focus changes in the list. We could add a command that allows to toggle between the selected reference/def and the peek view and then remember where you have been last\nAh, I understand. Speaking to my own use case, I use peek definition at least 98% of the time that I use the peek window, so at least 98% of the time there's only one file in the tree, and highlighting the tree doesn't make much sense. (I think it's a bit of a UX problem with the \"peek all references\" - it feels a little too much information in too constrained a space, so I normally do a global search instead - but that's an alternate request.) Your toggle command sounds great to me. I would also love if there was a keyboard shortcut to focus back on the original text document.\nmaybe the focus should be on the peek editor directly if there is only one entry in the list, that would be very helpful.\nIs there any progress being made on this?\nNot entirely sure whether it's appropriate to add this here, but I'd also suggest two other things for the peek window that're in the same vein: the \"Keep peek editors open even when double clicking their content or when hitting escape\" into two. The two aren't really equivalent, especially since double-clicking selects entire words and I'd for example like to be able to that, but still be able to close the window with Esc. 'd suggest changing the tab order of the peek window for the editor to follow the list. That way, when the peek window is opened you can select the appropriate file and Tab into it instead of Shift+Tabbing.\nAny push on this? this is a must have. I seriously don't know how people have gone without this.\nThis is an open source project, this issue is marked as 'help wanted', so how about you?\nI'll give it a try, looking for a way to automatically focus on peek window when there is only one entry in the list, according to suggestion. Do you guys think it's a good solution?\nI'd say a command to move focus between them and then a setting to define is focused by default is the better solution. That makes the behaviour predictable, e.g I often press Enter when having just one reference I wouldn't want a new line then\nfyi - this is the place (and sample) for adding commands that use the peek view widget:\nBesides, we still need a setting to define which component will receive focus by default when Peek Window is shown.\nSorry. I'm so far away from having time to learn how to contribute. I'd like to, but I'm still in college working on several other things at the moment. Also, wouldn't even know where to start learning how to build vscode extensions and stuff.\nHi, I'm a heavy user of peek editor. But I accidentally minimize the peek editor window , which means when I click peek definition or declaration or references, there will be just file along with its path that include the definition will display in the first row , and the editor window (which should be the second row) will not. I have tried many ways to get it back to view, but nothing works. And I search all the settings related to peek, but none of them are about how adjust the size of peek editor. Have you ever encountered the same issue? Thanks for your time to view my question.\nfyi - I have change the setting to:"}
{"_id":"q-en-vscode-48d9afdf11be971fd39ecbde2723054296b95bfbca2bb7d77b75e7887815728b","text":"Testing Insiders dated 28th March, 2017 Happens in linux and windows <img width=\"333\" alt=\"screen shot 2017-03-28 at 11 42 13 am\" src=\"\""}
{"_id":"q-en-vscode-991e90b2e596e1e53e29e7da917470ce206b44c6d86bc1a5e0b839bfd4b39463","text":"Testing in this setup Get this Code lense with '0 References' in ! Clicking the references link just shows an empty peek, it is not clear what is being search for. !\nNow I understand what is going on: ! The reference in the template isn't found by the Angular Language Service extension.\nDo you suggest we don't show the peek view from zero references code lens? Or should the peek view have more info on what would normally be displayed?\nsince the empty peek view is not helpful I would not show it and '0 References' should not be shown as a link on hover."}
{"_id":"q-en-vscode-dbee68ce1879ccd528722f6f3abf1ab48ecdf750920da16fa6579dc1da88ec00","text":"! Note the extra bar the right of the blue block quote bar\nI hope my fix will work for you. :smile:"}
{"_id":"q-en-vscode-c6320774dbe910593879c4777ff2f42d8609006e7cb5f86705dd4bf07c49bf78","text":"VSCode Version: Code - Insiders 1.12.0-insider (, 2017-04-13T06:10:58.128Z) OS Version: Windows_NT ia32 6.1.7601 Extensions: the listing exceeds the lower minimum of browsers' URL characters limit Steps to Reproduce: any file and try to horizontal scroll either via mouse wheel or shift scroll.\nHello i'm still experiencing this issue , I've noticed a separate issue was closed for Ubuntu users, but still noticing this with Windows.\nThis seems to be fixed in the latest release.\ncorrection, this issue still exists.\n/cc I'm seeing this issue again as well, on my Ubuntu work machine and on Windows: VSCode Version: Code - Insiders 1.12.0-insider (, 2017-04-21T06:12:00.370Z) OS Version: Windows_NT ia32 10.0. Extensions: ;\nSome more info: shift+mouse wheel does not work, however when I use the trackpad and two finger swipe horizontally it works fine. Unplugging and plugging in the mouse again didn't fix it.\nHmm the trackpad usually is for forward and back commands. I can confirm this is the same case for me. I'm wondering if electron has switched the forward and back commands for the horizontal scrolling. Because in my mouse settings if I set the horizontal scroll will to page forward and backward it works perfectly. But I shouldn't have to do that.\nI tried to reproduce and failed with my setup (Windows 10 and Windows 7 on macOS running in Parallels). Here is what I try: shift-scrolling in the editor wheel-left/right in the editor ! Are you doing something different? Does it reproduce when you run with blank settings?\nI tried to reload vscode insiders without any settings, and i'm still having the same experience.\nOne other note, I have mac os sierra at my home workstation and everything works perfectly there. It seems to be Windows specific. I'm running Windows 7\nHere are errors and warnings upon restarting insiders without typing anything. I wonder if this has anything to do with it.\nI assume those warnings would go away when you run code . I would not assume however that this would help you for this issue, but worth a try anyway.\nI wonder if we could do a little experiment. Can you: download run it from the window that opens make it small enough that horizontal scrollbars appear see if you can scroll ! If so, this would point to a general problem with our UI framework (electron) and we should report it there.\nseems to be this line that's throwing an error which may be causing the scroll to not work.\nI would think that is unrelated, but please follow my steps to find out more.\nI downloaded () on the Linux machine that I can reproduce it on and I can scroll just fine. ! Note that after this gif I updated to latest and it still doesn't work on:\nSorry I didn't see your message until just now. Yes it scrolls fine. On Mon, Apr 24, 2017, 10:31 AM Daniel Imms <:\nneed your help here for more debugging. First thing that comes to my mind is to add a breakpoint into  to see if this is even triggered when scrolling horizontally?\nit's hit, but is set, not as I would expect. ! is set though: ! Are we meant to handle manually?\nany chance you could debug stable to verify that e.deltaX is set? Is there any other property that has an increasing value? /cc if you have any clues on this Looking for chrome issues I stumbled upon:\nFixed via\nThank you guys, very much appreciated. Have realized over the last few days just how much I depend on horizontal scrolling.\nAs i'm not familiar with the insiders release pattern, any idea when we can expect this to come down the pipeline ?\nshould be in tomorrow's Insiders build\nThank you Daniel Imms <:"}
{"_id":"q-en-vscode-ce4e5b8f9cd25ce756a0b428accf250d73e0791bea6c9148240815939c97b389","text":"VSCode Version: Code - Insiders 1.13.0-insider (, 2017-05-26T05:14:23.944Z) OS Version: Windows_NT ia32 10.0. Currently you can multi-select items in the Git files view and if you right-click, the & menu items will only open the file/change of the focused item, with the stage/unstage command work with the whole selection. It would be great to have these 2 command act upon the whole selection."}
{"_id":"q-en-vscode-b7408bab4ebacba21e679ec4b5086831f855cbfa77f23e4e344426d3cfe4ca39","text":"<!-- Do you have a question? Please ask it on --- VSCode Version: Version 1.12.1 (1.12.1) OS Version: Windows 64 bits This issue is not a bug but just a suggestion/idea to improve CodeLens performance for TypeScript when and are activated. In this case tsserver command which can take times and response can be big is called twice: one call of navtree for providing references CodeLens. one call of navtree for providing implementation CodeLens. I tell me if CodeLens provider could manage a shared context to store some data that other code lens provider could use it. For instance in this case, we could have just one call of navtree: one call of navtree for providing references CodeLens and store the result in a shared context. the implementation CodeLens provider search navtree from the shared context."}
{"_id":"q-en-vscode-090837ec918ae15fe9f40cec11d7f43f88bf19c2f78c9714c7ea5dcfb2836273","text":"Currently blocked by:\nBranch is at\nThe arg is causing the issue on Windows. This: Calls which passes to .\nIt may not work because out-build doesn't exist. I can repro when doing:\nAfter spending quite a bit of time looking into this, I'm pretty stumped on why this is failing. any help would be great as I'm not sure where to go from here. Here's what I've found so far: I can reproduce using the following: Unit tests on Windows fail at random times Passing shows the window and all tests complete, with KeyboardMapper tests failing which I frequently when running unit tests on Windows. ! Listening to the in addition to printing test names seems to stop at random spots, telling me that this isn't related to a single test. Here are the end of some logs from test runs:The test runner pauses for 2-5 seconds after a the last output and then exits.\nI just tried to reproduce on Windows, for me the tests are looking all OK. Do you see the test failures on the command line as output?\nthe output stops before any failures example output: etc.\ndoes this reproduce with mocha tests (and the exact same version that electron 1.7.4 uses) too?\nSeems to happen when using node 7.9.0 as well which ships with 1.7.4. Currently the build machine is set up to use 7.10.0 (set in ).\nOk so it seems electron unrelated then. Can you explain this, are you saying the keyboard mapper test is causing this?\nNo, the keyboard mapper tests on Windows just have always seemed to fail for me a lot on most of my Windows machines.\nCreated to track upstream. It appears to be happening only when using because calls stop making it to the main process after a short period.\nPossibly related:\nUpdate: It appears is an issue with my home desktop only and is different to what we're seeing on the VSTS build server. So I started to add some logs and run more builds on VSTS to see what is going on and it appears like the tests are exiting with but is being set to 255. I made a change to the test script to change error code 255 to 0 to get around this issue since the tests were passing anyway After doing this I did a full build and it succeeded, however it does not launch. running with gives this:\nThat's pointing at , perhaps it's related to the snapshot stuff?\nCommenting out the \"Create loader snapshot\" step from fixes the exception on launch Also integration tests are not running on VSTS for Windows, this may have caught the snapshot issue earlier.\nWhen updating electron you must also bump the version of . Unsure if there is electron migration guide that should have been updated...\nAdded to the guide\nDiscussed with we will shift this to August.\ncould you look into the exit code issue when you get a chance? We're seeing VSTS giving an exit code of 255 even though appears to be getting called. The change for that is here:\nWhen I log in to the build machine and try to run , the unit tests don't even run at all. All I see is two processes stuck. Do you see this too? ! I also can't find the build where you had a 255 exit code.\ninterestingly I just noticed that Electron marks 1.7.5 as \"latest release\" while 1.7.4 is still a pre-release. We should consider being on the latest official release if possible imho. ! !\nwell I've never been able to log into the machine, I an echo %errorlevel% and it was printing 255 after the tests. It looks like VSTS has discarded the build as it wasn't marked as retain. If the Windows builds succeed after this line is removed then all is well:\nUpdate: I have Electron 1.7.5 on the branch , I'm running into build issues on VSTS/macOS though which appear to be related to using the new file watcher in the build? Not sure why an updated Electron would make this error occur yet though.\nWe got that error in the official smoke tests and Andre said in Slack that the Mac build machine was messed up. Try it again.\nWill do"}
{"_id":"q-en-vscode-b4b7c725c5e66391794bce240bc4fb64c590350868c0b311d5a58730b7c7c744","text":"Loosely related to ! We can use the 'files to include' box to do a search in an arbitrary folder, but it has to be an absolute path. It should also work when using a ~ to indicate the user's home directory."}
{"_id":"q-en-vscode-b1ae98dbd5707c4efe8636f689809e43e2065c1fbad42030c30acfbd3ca93f63","text":"VSCode Version: Code 1.17.1 (, 2017-10-10T14:24:50.851Z) OS Version: Windows_NT x64 10.0. Current default keybindings on Windows: has two key bindings, which seems unnecessary. On Linux the default key bindings are (IIRC): Which makes a lot more sense IMHO. Also using this default on Windows would also align Windows/Linux key bindings and maybe even avoid complications like . (same for of course)\nWe will not change this. See"}
{"_id":"q-en-vscode-77127d08f8089c9d3c79db34ed80cb542095641cf9c27b25aed1bd8a5124f9fb","text":"Generalization of Steps to Reproduce: a markdown file: the js code block, type Bug inserted. Root cause* Markdown defines and as auto closing pairs. Js does not. It appears that we always use the document language's autoclosing pairs instead of the autoclosing pairs of the embedded language at the current cursor position\nGood catch, auto close should check language around cursor instead of model.\nAnother case: Based on JS's auto closing rules, we would expect completion to"}
{"_id":"q-en-vscode-141fd196dcfd7de7969d7fa034def603a260f1c4d735deb9a01c5ea561d831d8","text":"VSCode Version: 1.18.0 / 1.19.0-insider OS Version: Windows 10 Steps to Reproduce: a Project with Workspace (Typescript) one external method inside other method in whatever class. Alt+F12 to open \"Peek Definition\" the method opened in peek definition .. bug reproduced.. file opened in peek definition now be opened in workspace. Even after save the file on peek definition, the file opened on workspace continue without save. <!-- Launch with to check. --Reproduces without extensions: No\nCan you please clairify what the problem is. Here's what I see: ! Is the error that file from the peek definition ( in my gif) still has the unsaved indicator in the workspace, even after you save it?\nYes, even after I save the file (). Pointer with focus inside peek definition, and I press and after to save and close peek definition the file continue with unsaved indicator. This behavior did not happen, even after I change file inside peek definition the file changed would not open.\nSee my gif. Note that the focus is on the peek definition and when I press the saved file is not , but . I believe this behavior is wrong. !\nAs designed, we open editors that are dirty to prevent data loss.\nBut this is nothing intuitive See my image, not would be interesting working like this? When change file, show the little ball in peek definition. ! When click to or , then show the confirmation: !\nand what should happen when you close the tab?\nIn this case if close the would could execute one stack of confirmation of files changed starting by\nAnd if close vscode, then would be open the file on tab with changes before close. Like this no data is lost.\nI do not like that behaviour, I prefer that one modified editor maps to one opened tab, anything else makes it very hard for the user to track dirty files.\nok.\nI've been in Visual Studio to see how it behaves and actually vscode is behaving the same way it does. However, in visual studio when we save the file in focus inside peek definition the file opened in the tab is saved. I think vscode should behave the same way. In this case, I reiterate that saving the and not file when I am with the cursor in peek definition when pressing is a bug. ! Now, see this behavior on vscode on press !\n?\nYeah thats a good one, lets take this issue to reflect that.\nhas there been any updates on this issue? I think this behaviour is rather reasonable and intuitive, though I'm kinda surprised this issue hasn't seen more activity (I'd expect more people to want this). I had someone ask about this very thing on my extension: and this could really help us:\nI've found that adding the following to accomplishes nearly the same thing: Works for me since I don't tend to keep unsaved files open often. Might be useful for CSS Peek.\nHi I agree with guys, that is really annoying, I'm wondering why they still didn't fix this problem ... !!! why Ctrl+S work in peek preview ?!? its fine in Visual Studio\n! I pushed a change to enable this by default if the embedded editor is focused. Pressing or will now save both the main editor and the embedded editor. Feedback welcome, it will be in tomorrows insider build:"}
{"_id":"q-en-vscode-8cda57361c9957773e46b3316399af93209fa47ac8b9644bc6711b94b3d622c0","text":"Default order in keybindings (default keybindings and intelliSense) is: ==When you copy from keybindings view and paste: =="}
{"_id":"q-en-vscode-13478de58a6c8fb1f750bdcebaa50ab3c12047686fee4fe5c5f64a0c9ce3c162","text":"Given , I should be able to Run In parent directory prompt, enter Get popup\nHasn't this issue been solved before ? If it hasn't I'll try to fix it.\nActually it seems the current solution can only handle in the config setting... but not in the input box!\nhello I could take a look at the issue if no one is already assigned. It would be my first one. Regards, Pascal.\nHi let me know if you're working on the issue. I think I've found a solution.\ngo for it !"}
{"_id":"q-en-vscode-be8d20d75b822a1678ad4681b1f0dca0b13b55f4880597b85db2b52a9c5cb054","text":"I've had a couple of people report debug adapter crashes recently. One of them I'm unable to reproduce which means I need to get more information from the user. Currently, when a debug adapter crashes, the error information that appears in the console is very vague: ! The actual exception that was thrown is not shown. Since I can't repro this, the only ways I can think of to debug are: try/catch around all debugger methods that handles errors and passes them back nicely (and remember to do this for all future methods) the user to run my project from source and debug the debug adapter to catch the error (whilst not the hardest thing in the world, it's a bit involved for an end user) Is it possible to get better information from the error back into the console without having a debugger attached to debug adapter?\nI don't know if the text \"ERR\" in the console was actually supposed to be my error; but it isn't. My error was \"AAAAAARGH\".\nthis request makes sense. Not sure though if I am getting your error when your process crashes. Can you be so kind and set a breakpoint make your server crash and check if your message is coming through as a or as something else. Thanks\nDo you mean the text? I also don't get that now when trying to repro, only the \"terminated unexpectedly\" message. If you still want me to try that breakpoint, can you explain how to add a breakpoint there. I did try breaking in my adapter and navigating the callstack (that file wasn't in there anywhere) and also checking the filenames of the files that were (they showed ).\nI mean let's try to see if your error \"AAAAARGH\" surfaces on our side. You should be able to just run regular vscode, F1 open developer tools, wait some time and put a breakpoint by openning a . Also try to put a breakpoint and check if the error stream spits up something.\nThe breakpoint is not hit, which is perfectly normal since the exception occurs before the socket connection is created. You should be able to reproduce by adding require('unknown_module') to the top of any debugger extension .js file\nOk, I got breakpoints in both and . First wasn't called and the second was called but is and is . So seems like the message isn't there. For testing, all I've done is put inside my debug adapter. Though now I think about it; the error isn't occurring in code that's been executed directly by Code - it's during the response from my debugger. In my debug adapter I spawn a process and I add a handler for . When something comes over and my callback runs, that's where the crash is occurring. I don't know if this changes things (for example, I don't know if this is on another thread?).\nThat one (in ) didn't seem to be hit either.\nActually, I think I can repro this really easily by just putting in the constructor of the ? Also gets .\nThis issue has the needs-more-info tag though I think it's easily reproducible with my last comment. Let me know if that's not the case!\nI have removed tha tag and plan to look into this issue this week. Thanks\nInvestigated and the issue here is that I am not getting any exception data from the spawned process. I simply get the code while all the other error chanels are blank. I am not sure what would be the proper solution here. Suggestions are very welcome. Also are all your debug adapters launched using node? If yes then it might be that our  launch is somehow eating up the error if it happens on startup. If this also happens for adapters which are not realying on node to be launched then is not to blame.\nplease try to disable the launch for node () and see whether this makes a difference. DAs should work independent from . But make sure that the node version used is roughly the same as Electron's version (i.e. 7.9.0).\nYes, using stdfork is the cause here. If I do not use it, I nicely get the error popping up in the extensions output channel. My weekend is starting if someone is feeling smart code review and let me know if you find something fishy about stderr not being wired through.\nThen you might want to involve/assign\nI am going on a vacation tomorrow evening, and this is like this since forever so pushing to April to investigate with Alex what is going on behind the curatans here."}
{"_id":"q-en-vscode-cccce9d325a20a265ed8a9b2acb7e4bf7e316f2026f3fb2cdf101dbb7200d538","text":"From: Steps to Reproduce: VSCode Version: 1.20.1 OS Version: macOS High Sierra 10.13.3 Steps to Reproduce: markdown file markdown to first line in preview or double click on the first line Does this issue occur when all extensions are disabled?: Yes"}
{"_id":"q-en-vscode-c54e45a660304bd526f1647c48d8854844187bf8a999eeee4fcbc6d59abea98a","text":"I'm using vscode for editing commit messages and merging: .gitconfig After editing the commit and closing the commit editing / merge tab is it possible focus back the integrated terminal? I had the same feature in my vim\n(Experimental duplicate detection) Thanks for submitting this issue. Please also check if it is already covered by an existing one, like: <!-- score: 0.561 --<!-- potentialduplicatescomment --\nI'm not sure there is a nice way to do this, any ideas?\nI am actually also hitting this quite often... not sure how to best fix this. We do have code in place to delete the , so we are aware of the editor closing that was triggered via a invocation. But bottom line is that we would need to know that the operation was triggered from the integrated terminal to begin with...\nthere's this: They may not be 100% reliable though as an external terminal could have been launched by the terminal (could check ). Maybe the off chance of one of those happening is worth forcing focus? Another way you could do this is track the PID it was launched from on the CLI/main size and see if it's a child process of the terminal. Seems like a lot of work for such a small thing, we also only track the terminal process tree on Windows currently.\nit is a bit of a hack but the terminal could also listen to the editor closing and checking if the file closed was part of  and then just put focus back to the terminal if: the terminal is open the terminal just had focus (?)\nit's not so easy to know that the terminal launched the code CLI, it may be down the track when we track the process tree more though. When an editor is closed that was launched via --wait, can we track the and and focus the terminal of the same renderer if they match?\nto clarify what I mean - integrated terminal - terminal which is integrated to the editor for example from vscode terminal: git add . git commit -a <vscode opens commit message editorI'm editing a commit -I'm pressing save and close the tab Integrated terminal should be focused back I apologize accidentally pressed the wrong button\nyeah but this has to happen from the terminal layer, cannot happen from the workbench to avoid layer breakage. Are you saying that these environment variables are passed to the instance of VS Code that is starting from the integrated terminal? We would probably need to wire these in through lots of places on the main side to eventually reach the renderer.\nwhen code is launched via the CLI it inherits the environment of the shell which contains these vars. Ok, let's go with listening to editor close and checking  as you suggest. That is definitely nicer than wiring the values through. The listening should probably be done in PRs welcome\nIs it alright to watch for waitMarkerFile deletion, since workbench will delete the marker file to signal when editing is done? See . Also, see the proposed fix . Using my development build, I'm not able to force vscode to use its window for resolving merge conflicts. Instead a new window is opened. My git config is identical to what aliev posted above. Any suggestions?\ntry to use that should force to reuse the last active instance. I think there are some things missing from your change: only do this listening if the marker file was opened from the integrated terminal handle the opening of the marker file within the running instance, in that case the request will come in via  and the configuration object that you are using will not reflect that, I think it would need a new event or something similar that gets emitted when a marker file is opened in the running instance\nI think this is not as straight forward for a pull request so I went ahead to implement this via I took your idea about watching for the marker file to get deleted, thanks for that\non to the next one :wink:"}
{"_id":"q-en-vscode-b5bbc265004055abf6adbc76ed04ef843b69f0bcef84f6a28b8a72338f500878","text":"In light themes the status bar text changes from white to black, but progress spinners do not so they're hard to see: \nThis icon is probably not themable yet. E.g. for the status bar feedback smiley we use trick to allow to change the color of it (). Depending on the outcome of we might move this into the notification area though.\nfyi, the fix is to use a font for the icon (in this case our octicons). I found the sync icon fitting because we already use it for SCM sync: ! Alternatively we could use this animation: !"}
{"_id":"q-en-vscode-be991a13075d578aaa880f49df75ff3c78e890ed7f2a53443d114d873f9bae9a","text":"Issue Type: <bFeature Request</bVSCode uses different conventions for the problem counters in the problems view and in the status bar. The problems view displays: 1 2 ... 999 1K 2K ... 10K+ and the tooltip shows the actual number. The status bar displays: for errors/warnings: 1 2 ... 99+ and the tooltip doesn't show the actual number. for infos: 1 2 ... (always the actual number), the tooltip doesn't show the actual number. <img width=\"118\" alt=\"problems-view\" src=\"https://user-\"<img width=\"108\" alt=\"status-bar\" src=\"https://user-\"VS Code version: Code 1.21.0 (, 2018-03-07T10:54:22.798Z) OS version: Darwin x64 17.4.0 <!-- generated by issue reporter --\nassigning to you. I personally would adopt the status bar to the problems view. Everything over 99 is not very informative anyways."}
{"_id":"q-en-vscode-5aba52e0c15c7ea7c03ebc92e617d0f27964b97d4b87d0c0eb6d115b64482f20","text":"<!-- Do you have a question? Please ask it on --<!-- Use Help Report Issue to prefill these. --- VSCode Version: 1.21.0 OS Version: macOS 10.13.1 Google Japanese Input it a new VSCode instance on keyboard, then hit enter, and do it again a few times : the suggestions should never overlap w/ text. Actual: the suggestions overlap w/ text. ! Does this issue occur when all extensions are disabled?: Yes This is not an issue with the default Japanese IME that comes w/ macOS. This setting seems to . (You'll lose some functionalities)  \nSame Here\nsame here\nSame Here\nThe suggestion box that shows up in this case is not the one for suggestWidget. This seems like a native control which I dont believe VS Code controls cc Is this a regression? Did this work as expected in earlier versions of VS Code and stopped working recently?\nI think it is probably caused by the position adjustment of the textarea, Google IME tries to put the candidate list at the proper position, but I don't have a stable reproduce yet.\nSorry, I am a new user to VSCode so not sure if this is a regression.\nI can repro it consistently. !\nNot just VSCode but Adobe Brackets seems to have the same issue as well - and for that, apparently, the issue is on CodeMirror: If it's an Electron issue not sure if upgrading Electron to 2.x or 3.x fixes it. This was not an issue on Atom (even when Atom was using Electron 1.x).\nI also searched through Twitter, and people have been reporting this since April 2015. Definitely not a regression.\nSame Here ! !\nIs there anything I can do to help? Also saw this reported on Teratail, a Japanese Stackoverflow clone, about a week ago.\n※ (temporary)(workaround)Google (…) <img width=\"637\" alt=\"2018-09-26 11 51 35\" src=\"https://user-\"\nTranslating what said: Temporary workaround: go to Google Japanese Input preferences, <img width=\"271\" alt=\"screen shot 2018-09-26 at 6 58 30 pm\" src=\"https://user-\"go to (\"suggest\") and uncheck (disable suggestion from input history) (disable live conversion) Then click button at the bottom. Then it no longer overlaps.\nthanks for your help. This issue was caused by how we implement our editor (virtualize everything but has a tiny textarea around the virtual cursor) and Google IME doesn't work the same as the OS builtin IME. I don't have a good idea how to hint Google IME about the position but your solution works perfectly for this issue. For anyone who is using English as system language, the preference with the workaround will look like below: <img width=\"637\" alt=\"image\" src=\"https://user-\"\nthanks for the follow-up! Just for completeness: Google maintains \"a subset of Google Japanese Input in the form of source code under OSS license without any warranty nor user support\" at (), though they're .\nSame here\nsame here\nsame here OS: windows 10 Language: Chinese IME: Sogou Pinyin, Baidu Pinyin !"}
{"_id":"q-en-vscode-e91cd8f0d5638b4ee90aa4837f818791bae96cd7c94092baff214dcb10c0119b","text":"VSCode Version: 1.21.0 OS Version: Windows 10 Redstone 4 Steps to Reproduce: create a configuration type: { \"type\": \"node\", \"request\": \"launch\", \"name\": \"Stuff\", \"program\": \"${workspaceFolder}Stuff\" }, Save correct behavior in drop down where you can select that item. the exact same item (change only the program path, leave name the same) as a second configuration. { \"type\": \"node\", \"request\": \"launch\", \"name\": \"Stuff\", \"program\": \"${workspaceFolder}Stuff2\" }, your new item missing in drop down because you used same name. same name, by changing \"Stuff\" to \"Stuff2\" { \"type\": \"node\", \"request\": \"launch\", \"name\": \"Stuff2\", \"program\": \"${workspaceFolder}Stuff2\" }, your new config shows up but you can not select it. (Hover over does not highlight item, clicking does not select item) VSCode correct behavior. Does this issue occur when all extensions are disabled?: Yes\nworking on this. fairly certain it's related to terminal disposal and issue drop-down options get updated before the instance count is changed. non- atomic from select box view. It would be very helpful for your review once I have something finalized , I think the order of execution is very critical here.\nthanks for looking into this. I can review the PR once it is ready. We can verify if that is the same issue by opening F1 developer tools and checking the debug console for potential errors.\nI was able to partially reproduce your issue. While I did not have a problem with two debug configurations with the same name, I did see the hover issue. It it appears that you can still select the item with the keyboard so it's actually probably a misplaced disable class within the CSS. It is also still possibly related to another index problem with a twist on the results. StayTuned. !\nOne thing I left out that I should have mentioned is I also have configurations spanning mutiple workspaces. This issue was in a single workspace, however I have multiple configs which it merges together.\ncc: I can reproduce this in several ways. It appears to be an odd \"sticky\" class label within the list. I have a fix , I would consider the work around. I will commit tomorrow with comments to at least avoid the problem for now. do you use insiders to check after I merge?\nI do not currently, but I can, I'll check it after the merge and let you know.\nMerged into master. Should be a next insiders. I am not closing this issue as I still think this is a workaround."}
{"_id":"q-en-vscode-0ce6a9961f2234f121681aa327298f85a524f2b161d2fddee2384e5b930f3631","text":"Refs: I like that the relatedInformation message has a nice different verbatim font and that makes it differ well enough, I suggest to not use a different background as imho the effect does not look that nice and we do not do it anywhere else to my knowledge. Also in the F8 we do not use a special background !\nYeah, it's the default rendering we have for markdown. Not sure I have change that without making people sad...\nAlso, the bullets do not appear in the Problems pane or when F8. Why have them here?"}
{"_id":"q-en-vscode-446f87812b5e4ae507073a0f2479fd81c8eefa281fc9ccb0c30e88288ce3f0fd","text":"When creating a breakpoint in an active debug session and this breakpoint cannot be verified, the resulting icon is almost invisible: ! The problem is that the icon's stroke is too thin. In previous versions the icon had a thicker stroke: ! I suggest that we restore the previous icon.\nAgree that it is a bit too light. I do not just want to take the previous icon since the color and style are different. Thus it would be great if could tackle this Also fyi\nNot quite as strong as in the 'before' above, but stronger. Marking verified. !"}
{"_id":"q-en-vscode-6bc4594c08f2b7e28d1785dc673732980b6e70fe853cc7ce84bebd51fd72b3be","text":"Issue Type: <bBug</bNavigate to any reccomended extension with a screen reader. The reader will read the extension name, followed by \"[object Object]\", rather than showing that the extension has been reccomended and why. VS Code version: Code 1.24.0 (, 2018-06-06T17:35:40.560Z) OS version: WindowsNT x64 10.0. <details<summarySystem Info</summary </detailsExtensions: none <!-- generated by issue reporter --"}
{"_id":"q-en-vscode-4188c4b60de17d64a30258a655fb9e05dff6f9827da903d35eece120f35f6b0b","text":"I've attempted to find any other requests for this but didn't seem to see any. At the moment the CVS context menu shows: ! Only \"Pop latest stash\" and \"Pop stash...\" are visible. Is it possible to change the default stash behaviour to \"apply\" rather than \"pop\"? Additionally, is it possible to add two lines under \"Pop stash...\" for the same actions, just using apply: Apply latest stash Apply stash... I just lost a few hours and considerable amount of hair when a popped stash was deleted. I then had to sift through doing on each . Having \"Apply stash...\" easily accessible might be nicer.\n(Experimental duplicate detection) Thanks for submitting this issue. Please also check if it is already covered by an existing one, like: <!-- score: 0.676 --<!-- potentialduplicatescomment --\nI suggest to simply add and commands.\nis anyone working on this, if not, I'll like to pick it up as a\nlooks like it might already be done in\nyeah, just confirmed, Thanks. If I may ask, is there any rule around tagging an issue that is being worked on as it would be multiple efforts if I happen to have started it without anyone's knowledge. I've been scouting for an issue with no assignee to work on :-( Thanks for the catch tho!"}
{"_id":"q-en-vscode-679b17bb544f44bee43a59955f869891894df3c84701b15a8ee6fa108661ebf2","text":"Issue Type: <bBug</bI haven't been able to reliably reproduce this, but it happens decently often. To repo: -click on a line in an editor (seems to happen more often if it is not the active line) of the time the context menu will open, but sometimes it just flashes in and disappears My guess is that it feels like some sort of conflict between showing a hover vs the context menu. And I'm not sure if GitLens plays a roll here or not (though it never happens with the native menus). Another observation, it seems to be more likely to happen right after you've switched to a editor as well (either that or just becomes less likely after the menu has been shown already or something). VS Code version: Code - Insiders 1.26.0-insider (, 2018-07-18T05:19:07.488Z) OS version: WindowsNT x64 10.0. <details<summarySystem Info</summary </details<!-- generated by issue reporter --\nI think vscodebot is drunk This should be tagged and likely assigned to\nit just loves to classify things as terminal :sob:\nLooks like this happens when the contributor codelens info gets to the editor which may trigger scroll which dismisses context menus. /cc\npossibly related\nI saw this issue while testing today.\nwas the behavior unique to the latest insiders or could you reproduce with stable\nI tested only on latest insiders did not try on stable\n/confirm still repro on 1.30.0 stable version\n/confirm\nSorry, I still this one from time to time. Not fixed, but also no steps. can we add logging when programmatically(?) closing a menu, esp. withing 1sec of opening\ndo you know if yours is the case that a menu item is triggered immediately (because the menu was drawn under the cursor and mouseup triggers the item) or that it is actually dismissed?\nThat - unless the command is a no op. I think this happens when I open the context menu but not immediately interact with it. Can it be that while the menu is open and context-key events comes in, causing the menu (native. macOS) to hide?\noh, native case, alright, i'll dig into it\nThis is now reproducing with the custom context menus as well. The scrollTop change does not appear to work anymore(fully?). Something to note is that in VS Code, with the custom titlebar enabled or using a mac, it is not possible to scroll the editor with the conext menu open. Thus, dismissing due to a scroll event is now wrong (I think). Dismissal due to scrolling seems like it was for the benefit of monaco more than vs code. See Do you think we could switch to event? This way, I believe, monaco's behavior should be preserved and vs code would not have this event triggered because of the blocking div when a context menu is open.\n:+1: That could work well."}
{"_id":"q-en-vscode-ddbbb149a85f5931969814c2101a980d42a0e6fe921d3349670f37bf335cd94b","text":"Issue Type: <bBug</bI have a file named F1000-v117.au3. When copying/pasting the file, it creates a file named F1001-v117.au3. This is not the expected result. F1000 is the name of the file and not a version number that should be incremented. Changing the regex within the incrementFileName function fixes the issue. Basically, make sure the first section of the file name is numeric only. to VS Code version: Code 1.25.1 (, 2018-07-11T15:43:11.471Z) OS version: WindowsNT ia32 6.1.7601 <details<summarySystem Info</summary </details<!-- generated by issue reporter --\nPR for this is welcome It would be great if the PR contains a unit test fyi\nI'll take a look tomorrow.\nis the expected result right?\nActually it seems like other cases are broken now. Have a file , copy and paste it The new file is , in 1.26 it was\nThanks for finding that Due to that I had to revert the original fix. So now copying gives However this issue is now not fixed, but it was like that in previous stable so nothing urgent -removing mielstone let me know if you want to fix your fix and provide a working PR. Thanks\nHey guys, I took a crack at fixing this regex have some questions on how the naming / number schema works and I want to make sure I have an understanding of the naming schema. We have our cases here: suffix: prefix: single number: no number: ends with number: mixed (new case / why the above PR broke): prefixed with number: The issue is that mixing letters and numbers triggers the prefix regex which incorrectly increments the number. But all the different variations of prefix / suffix and single number / mixed with letters make this particularly tricky. It seems like edits need to be made to existing regexes to ensure that they are matching digit only groups. However, the above PR regex edits and the edits I've made consistently break other regexes. Any ideas or considerations here?\nThis issue is being closed to keep the number of issues in our inbox on a manageable level, we are closing issues that are not going to be addressed in the foreseeable future: We look at the number of votes the issue has received and the number of duplicate issues filed. More details . If you disagree and feel that this issue is crucial: We are happy to listen and to reconsider. If you wonder what we are up to, please see our and guidelines. Thanks for your understanding and happy coding!"}
{"_id":"q-en-vscode-a5bdb67b921b3824757a750d96c9ac5fad05aa0b8cc9672f79a0b0b0c7a898bc","text":"Testing When using the '...' menu to add sort filter options, the option becomes disabled in the menu, preventing it from being twice from that UI. I think it would be nice if the suggestions behaved similarly, not suggesting filters that are already present in the input. ! VS Code version: Code - Insiders 1.26.0-insider (, 2018-07-31T05:18:06.861Z) OS version: Windows_NT x64 10.0. <!-- generated by issue reporter --"}
{"_id":"q-en-vscode-4fbf9f8af5234b012e0d00f9d30b60e99813121876e1e7d5f605adb6c3c10e3a","text":"Tested in Windows with Insiders for 8/1 Add a couple of spaces in the extension search box Move focus to any of the open editors Come back to the extension search box and enter a letter !"}
{"_id":"q-en-vscode-2f55c1df44185debd9e2b64ac1445b6fa2a6bea2598d79efa80979ff98b399b7","text":": Check whether it's ok to map to : Add 'null' or \"None\" item to enum : =- : ?? I don't see this in vscode's core settings Also, . Example:\nThese are the settings explicitly that I could find with a simple search. Of these, the HTML ones are the only concern. For most of them we should probably just get rid of accepting null, as it doesn't make a ton of sense to have just mean either default or some secondary default. But has very different behavior between and so we'll have to decide on something to do there.\n[ ] : fix requires API discussion over adding an additional schema parameter [x] : fixes [x] : fixed [ ] : Probably wont fix? Would require some new components and UI/UX discussion. [x] : fixes\ncan you add some steps for how to verify?\nHere are some settings with compound types - Test that they work as you'd expect. I can't remember which extension has the nullable enum but you can test it with a contribution like\nseems to work for the first one, but not second. e.g. here I cannot type \"null\": ! I can only type numbers (I am on Windows).\nFor the null|number type, an empty input box will correspond to in the json"}
{"_id":"q-en-vscode-5f0d060a51c9a7f36b12b64852616e25d6a4402538e727a692f0a7c54203c3dd","text":"<!-- Please search existing issues to avoid creating duplicates. --<!-- Also please test using the latest insiders build to make sure your issue has not already been fixed: --<!-- Use Help Report Issue to prefill these. --- VSCode Version: master commit OS Version: macOS 10.13.6 Steps to Reproduce: clone vscode ouput: how to fix: update typescript version from to will fix it."}
{"_id":"q-en-vscode-ebaa529fe9bcf3151e99c2309f8970f0b58ac785f7c7d4f6c57c4d403d8aa011","text":"debugging the loaded scripts view open debugging, notice exception in the console can not reproduce with Stable so it might be that my listener refactoring caused this If you pause on exception you will see that caused it.\ndid you have a chance to look into this? Or would you like me to investigate? I suspect this is breaking the loaded scripts view and is the root cause for\nno, please investigate since this and was working before, and I haven't touched this code for two weeks.\nend session events are not arriving in As a consequence the tree for the session is not properly removed.\nI see something differnetly, the end sessions are arriving however when the last session end arrives view starts refreshing the tree and while that is in progress the view gets disposed which throws an exception. Looking into a fix...\nI have created a PR which address this issue"}
{"_id":"q-en-vscode-9343e72810f307a66cd89a3625321d37ad1765877425f4fa4befc3f491c99365","text":"<!-- Please search existing issues to avoid creating duplicates. --<!-- Also please test using the latest insiders build to make sure your issue has not already been fixed: --<!-- Use Help Report Issue to prefill these. --- VSCode Version: 1.28.0 (1.28.0) OS Version: macOS High Sierra 10.13.6 (17G65) Steps to Reproduce: a markdown file preview to the side ! <!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes/No Yes\n(Experimental duplicate detection) Thanks for submitting this issue. Please also check if it is already covered by an existing one, like: <!-- score: 0.721 --- <!-- score: 0.717 --- <!-- score: 0.672 --- <!-- score: 0.651 --- <!-- score: 0.645 --<!-- potentialduplicatescomment --"}
{"_id":"q-en-vscode-64333ee7f1bdd36682e2c5b288ae7af650fa82224dc65b00749b1219120ea760","text":"Voiceover does not read new option details because of describedby issue. May require the new ARIA combo box model.\nAny problem with me creating new tags for NVDA and voiceover like there is already a tag for jaws? Accessibility-NVDA Accessibility-Voiceover I will also volunteer to do triage on accessibility issues if I can identify them."}
{"_id":"q-en-vscode-9bca18a74a200d52532c2b17e589ae5ceea522c6fb37cf6204a737d46b108534","text":"It's clear that the usual set of icons is a minimal case, and that you want the default set to stay minimal. But I think the option to differentiate between start with or without debugging is a feature that every developer could use. Maybe an option in the config to switch the current command to would already be enough. Yes, a plugin could add an extra button. But I don't think this belongs within a plugin, as it's an extension of the existing button. I'm willing to do the work required. But before that: what are your preferences?\nthis is a fair feature request. I would keep it open and wait if it generastes more user traction. For us to add it to the touch bar we would also need a new icon which would be easily distinguihsable from the current start icon (green play). I would also check what icon Visual Studio is using. The icon design is the actual work, adding is just adding a one liner here\nVisual Studio uses a same filled play icon for the default start action. It differentiates with an outline version of that same icon for starting without debugging. Seems reasonable?\nMakes somewhat sense. Can you design that icon froum our start icon? fyi\nWait, shouldn't it be the one from ? Anyhow, I made a mockup based on the old designs found in Visual Studio: ! The stripes indicate that it's faster than full debug. While the hollow arrow indicates that it's not as complete as debugging. SVG: Alternatively it could look like this: ! But, in my opinion, it does not convey the faster execution as well.\nNot sure if the three lines fit the rest of our icons, I'd go with just the outline. Thanks for creating this!\nI like how you explained the icons and I agree with your point of view. However let's listen to and for consistency just use the hollow icon. can you please provide the PR with that icon and we can merge it in. Since I do not have a mac touchbar I do not know, but does it support multiple themes? Do we also need to design the icon for the dark theme?\nI'll check! I don't think it has multiple themes though. And what about the location? I think it's the white icon from the folder that's being displayed on my TouchBar. It's white and has a different aspect ratio. Should I redesign that one as well?\nYou are correct, here is the location So I would place it just next to that one. And yes the icon should follow the style of those icons, so an empty white triangle seems fitting. Thanks a lot!\nCool! One more thing: That icon is not an equilateral triangle. That's correct? Because I could fix both icons at the same time. The current options: <img width=\"1067\" alt=\"screenshot 2018-10-19 at 12 04 50\" src=\"https://user-\"What's your preference,\nI would not change the current shape. Though I leave this decision up to\nAbout the theming: Apple states that you should supply themed images (black on transparent), but I don't see these in the VS Code codebase. So I guess Electron has a specific implementation to translate the png icons to TouchBar-safe icons. The png resolution does not seem to match up either, even though it does look good on the TouchBar.\nThanks for looking into this. I am not sure how to change the theme for the touchbar - if the theme changes and our icons look out of place than I would be open for a PR which does the theming of our touch bar icons\nPinging on the design decision.\nSorry about that, this somehow fall off my radar. Assingin to and November for discussion\nI'll start a PR in the meantime.\nVerification steps: a macbook that has a touchbar or use a file that you can run (i.e. python file with ) the <img width=\"80\" alt=\"image\" src=\"https://user-\"button to start without debugging should run the file, similarly as going from"}
{"_id":"q-en-vscode-530af879a203a7a2bb8f4ba2cfb71ac661e0c033a71401ae0bac76b9ee1b3c17","text":"Issue Type: <bBug</bWhen working with tagged template literals in JS and JSX (styled-components is my use case), the backticks used to auto-close when typed, but that recently changed. The big issue though is that if you go into a multi-line literal, and then try to type the closing backtick, vscode inserts two backticks, leaving you with three. For example if I type: I would expect to have the closing tick automatically inserted and the cursor in between them. Instead it just stops there. If I then go on to enter some css and then manually enter the closing backtick, I get: I then have to manually delete the third backtick, as if I delete the second, the whole pair is deleted because of the auto-close rules. I have verified this is happening on Mac and Windows 10, with extensions enabled and disabled, and in the Insiders build. VS Code version: Code 1.28.1 (, 2018-10-11T18:07:48.477Z) OS version: Darwin x64 18.0.0 <details<summarySystem Info</summary </details<details<summaryExtensions (20)</summaryExtensionVersion ------ ActiveFileInStatusBar1.0.3 better-comments2.0.2 bracket-pair-colorizer1.0.60 vscode-eslint1.6.1 es7-react-js-snippets1.8.7 prettier-vscode1.6.1 dotnet-test-explorer0.5.4 vscode-styled-components0.0.22 mssql1.4.0 python2018.9.0 csharp1.16.2 vsliveshare0.3.790 indent-rainbow7.2.4 vscode-docker0.3.1 seti-icons0.1.3 code-settings-sync3.1.2 mdx0.1.0 vscode-status-bar-format-toggle1.4.0 vim0.16.10 quokka-vscode1.0.154 (3 theme extensions excluded) </details<!-- generated by issue reporter --\n(Experimental duplicate detection) Thanks for submitting this issue. Please also check if it is already covered by an existing one, like: <!-- score: 0.498 --<!-- potentialduplicatescomment --\nIt seems to me the editor only adds backtick if it's legal for a template string to be in that place. However your example is only legal if styled components are allowed. Could you give an example with a valid js syntax? If it only happens with styled component syntax, could be it's a bug (or a lack of feature) in the styled component extension.\nIt works with just plain ES6 syntax as well. This example: Is valid ES6, no React/styled-components, and the same issue occurs when trying to type the multi-line literal. I should note that manually closing the literal on the same line does work okay, although it doesn't auto-close. It is only on multi-line template literals that the auto-close fires on the last tick.\nI'm seeing this bug also in python, where backticks have no special meaning . Sometimes I use them to denote arguments in a user string: But when I type the second backtick, vscode inserts two, resulting in: Strangely, this only happens if there is before inserting the backtick. If I type insert in any other part of the string, the doubling doesn't happen. should I open a different issue with this problem, to be tagged as a bug? If so, does it belong here or to ? Thanks !\nCan this be for this iteration?\nAs a workaround, set: Opened to track possible solution\nThanks for that! Anything workaroundish for the cursor coming inside to the next line with an indent on press Enter? !\nFWIW, I found disabling the Babel JavaScript extension put a stop to doubling up on backticks when closing a pair. (However as I rely on that extension, disabling it isn't really an option.)"}
{"_id":"q-en-vscode-c653c59262b463882d35a1ac1fb4c715b2623da10a614a48f15b3290dfa579e1","text":"Issue Type: <bBug</bWith at least one folder loaded in Code: to Code Preferences Settings on either \"User Settings\" or \"Workspace Settings\" Notice tab spacing is jumpy when changing active selection. It should NOT be as jumpy. I believe it's caused by the text being bolded on active selection. <iDesired</i! <iActual</i! VS Code version: Code 1.28.2 (, 2018-10-17T00:18:43.347Z) OS version: Darwin x64 18.0.0 <details<summarySystem Info</summary </details<!-- generated by issue reporter --\nHere's the\nI'm not sure about this change, I think the bold is necessary to make the selected tab noticeable. Some themes hide the underline on the selected item, and in some themes, the color difference is not really noticeable. So I would worry that this would make it impossible to tell which tab is selected. I also don't really have an issue with the slight shifting, I've never noticed it before now.\nWe should try to remove this shift as it's not really smooth. In our other tabs (panels, settings json) we don't use alternate font-weights for active/inactive, so I think we should keep them consistent and the same. Alternatively, if we really want to make the states look different we could adjust the opacity (just like the other tabs) as long as they pass the color contrast ratio: !\nI agree that it's not smooth. It's also not consistent with the other tabs in the product. There is a hack that can remove the shifting (discussed with this ), however I think that might be overdoing it to achieve this effect. I would add the opacity change as well to help reinforce the two states.\nShould I just borrow the inactive color from the panel tabs, at least for the builtin themes? Or if I should use opacity, can you tell me what color/opacity to use? Looking through some popular ones, I don't think this change would be a problem for that many other themes really... Also for folder settings, the folder name and string \"Folder Settings\" are currently distinguished by size and color, so we might still need 3 colors.\nActually the current panel tab colors in Monokai don't meet the contrast requirement: ! !\nWe only need to meet the color contrast on the themes we own, which is light/dark/hc. Here's the contrast on the inactive text: Here's the colors and opacity for the active/inactive panels: For the folder settings, you can change the opacity to (which passes) and will end up looking like this: !\nNot using the panel foreground color because the background is not always the same, but I think it looks good as is after the PR."}
{"_id":"q-en-vscode-eadf9821ff1500b61f69fec3b81f386e9e9ea8f3d5d08ccc0b98e6882a24de74","text":"Issue Type: <bBug</b VS Code version: Code 1.28.2 (, 2018-10-17T00:23:51.859Z) OS version: Windows_NT x64 10.0. <!-- generated by issue reporter --\n(Experimental duplicate detection) Thanks for submitting this issue. Please also check if it is already covered by an existing one, like: <!-- score: 0.504 --- <!-- score: 0.472 --- <!-- score: 0.457 --<!-- potentialduplicatescomment --"}
{"_id":"q-en-vscode-88112fa717e085e3f4af1c4dbce093bc06bf6e21c4b6d3ada541a9f4041c9118","text":"Issue Type: <bBug</bIn Lua, when comment contains \"function\", like: -- blah blah function then type Enter, the next line will auto indent for one tab, but it shoud not auto indent. VS Code version: Code - OSS 1.27.2 (, 2018-09-14T00:41:55.375Z) OS version: Linux x64 4.18.16-arch1-1-ARCH <details<summarySystem Info</summary </details<details<summaryExtensions (5)</summaryExtensionVersion ------ vscode-language-pack-zh-hans1.27.2 python2018.10.1 cpptools0.20.1 Go0.7.0 markdown-all-in-one1.7.0 </details<!-- generated by issue reporter --\nIntroduced by"}
{"_id":"q-en-vscode-8843114f61e192a590725213f99d3470c2139136b4b95c1fba2a3f38731d489a","text":"Issue Type: <bBug</bthe settings editor a setting that has a dropdown so that the options are displayed editors The dropdown remains open: ! Not sure if this is an issue with the select itself, or the settings editor. VS Code version: Code - Insiders 1.30.0-insider (, 2018-12-03T06:12:08.874Z) OS version: Darwin x64 18.0.0 <!-- generated by issue reporter --\nNice, for me it even jumps to the corner of the screen for some reason after switching to the editor. ! I think we will need some way to hide the list when the settings editor hides? Is there a way to force it to hide from outside currently?\nHa, I was afraid this would occur. The _sticky flag that Jackson for debugging was changed to true in a recent PR. I was going to ask if this was intentional. My Insiders was just old enough to not have this. was this left as 'true' by mistake when you're doing some mouse work for list? I will verify if this is the case and fix unless requested otherwise...\nGood catch, it looks like a mistake, I'll revert it so it can get into tomorrow's build. Let me know if that's not the case\nActually looks like you are on it go ahead if you want to create a PR.\nFixed with,\n- you have to merge, I'm not allowed :-("}
{"_id":"q-en-vscode-1ce663a2704c10d35f403a99939762b75a525274e20af06444013c5500a3aa55","text":"Issue Type: <bBug</bUsing similar contribution in extension: Command palette is doing it correctly: Keybinginds is only showing . VS Code version: Code 1.29.1 (, 2018-11-15T19:13:36.375Z) OS version: Windows_NT x64 10.0. <!-- generated by issue reporter --\n(Experimental duplicate detection) Thanks for submitting this issue. Please also check if it is already covered by an existing one, like: <!-- score: 0.447 --<!-- potentialduplicatescomment --\nI'm going to work on this."}
{"_id":"q-en-vscode-7dc4f5829fc538ae6677e12fcabf0f0a74c3c4456f76a8f711c3d11037215e2a","text":"Currently when one types a term in the search-bar of the settings-UI and no matching settings are found then a Clear Search option appears which deletes the term in the search-bar if clicked. However, this option does not appear when matching settings are found. If one wants now to remove the term from the search-bar one has to manually select the typed term and press the delete-button. It would be more convenient if there always would be a 'Delete Search'-Button. This could be a little X symbol on the right of the search-bar. Currently when the Clear Search option is clicked the search-term is deleted but the search-bar loses focus. It would be more convenient if the search-bar would get the focus such that one could immediately start typing the new search-term.\nThis should be a case for all the inputs around the editor (search, replace etc.), to have a little to quickly erase the input's content. !\nI disagree, the user is going to be typing anyway, they can press backspace. In the search view at least you can still use the \"clear search results\" button to clear all the inputs whether results are present or not (in Insiders only)\nThis I agree with, I will keep the issue open to do this.\nMay I ask you why there is only a clear-search-option when no matched settings are found? Why can't we have it also when results are found? Feels kind of half-baked to me. Either there is an option available which enables to clear every search or there is no such option.\nI don't want to clutter the UI by adding buttons for something that doesn't really need a button.\nWhy then not remove the clear-search-option completely? What purpose does it give in the case when no settings are found in contrast to when settings are found?\nI guess that when no settings are found, you definitely want to clear the filter because there's nothing else to do, but when you are searching, you either found the setting you want or you want to tweak your search query to search for something different. When the page is full of settings, adding a button/link to clear the query would be UI noise.\nI find myself often in one of the following situations when some settings for my typed search-term are found: Wrong settings are found because I didn't know the correct search-term and now I want to delete the complete term and type a completely knew one with the hope that the new term will show me the correct results. I typed the correct term and found the correct settings. After changing the settings I need to type a completely knew term because I want to search for the next settings which I want to change. I typed the correct term and found the correct settings. After changing the settings I want to come back to the default settings-view (where all settings are show as if you just opened the settings-UI). For this however I need first to select the entire search-term (double/triple click the term or select the term manually with the mouse) and then I need two press the del-button and afterwards move my hand back to the mouse. Instead one could simply click on a tiny del-icon right within the search-bar and instantly start typing the new term or even keep the hand on the mouse and keep searching trough the default-search-view. In fact I find the case when no results are found the most useless for a clear-search-button because when no settings are found I instantly can see that. In such a case my hands are usually still on the keyboard what means that it is easier to delete/correct the term with the keyboard instead of reaching to the mouse, clicking on the button and then again moving back to the keyboard to type the new-search-term.\nActually such a clear-search-button already exists in the keyboard-shortcuts-search-bar. You can see it in the screenshot on the far right within the search-bar. !\nGot me there. I guess would take a PR that adds the same button to the settings editor.\nI've been working on this issue.. Currently I have the search clear button working in the settings editor, I just need to tidy up placement as there is a search counter that appears where the action bar icons normally appear. I was thinking to just place the search counter that appears, slightly to the left beside where the clear button will be. What would you suggest for this?"}
{"_id":"q-en-vscode-b9fb021e388e3076c8d40a6a96e49e6f838540358d2c72bb36396fb4104ea890","text":"Issue Type: <bFeature Request</bThis new feature is over-eager and triggers on unsaved files that are not part of the commit, which I don't think makes sense. If I have staged file and am committing only file , I should not receive a warning that is unsaved. Ideally it will only consider staged files unless I am committing all staged and unstaged. VS Code version: Code 1.30.2 (, 2019-01-07T22:54:13.295Z) OS version: Windows_NT x64 10.0. <!-- generated by issue reporter --\nHi, can I work on this one?\nSure!\nI'm not in favor of changing this. Or if it is changed, perhaps give finer grain control over when it prompts. I always want a prompt because I'm a bit lazy and usually let VSCode add the files to staging for me when I commit.\nI see the value in your use case, but I don't think it makes sense in this feature. If you committed with a file that was untracked but saved this warning wouldn't help you. I think it would make more sense as a separate feature that warns on commit if there are untracked files regardless of save state.\nExperienced this today. I had to loop on this experience 3 times before I finally groked what was occurring and my choices in the dialog. I believe there are two issues here: rules triggering the dialog dialog itself In my scenario today, this was triggered by me having an untracked file named open in VSCode. was previously saved to disk. It continues to be untracked. I had made some more changes to this file, but not yet saved those to disk. The key concern here for me is that this is an untracked file. I do not want to have any dialogs given to me by VSCode for untracked files when I attempt to commit. This is the dialog I was presented: I kept clicking cancel. Because I equated with . I wanted to answer to the question . On my 3rd time, I read again the dialog and every choice. I finally clicked and got the effect I desired. This dialog is poorly designed. The question proposed is a yes/no question \"Would you like to save it before committing?\". Unfortunately, the three buttons are not or . Instead they are answers to a question not asked...something like \"What would you like to do now?\". Naturally, that's a poor question and instead, this dialog needs to be reworked.\nI've rephrased it to . Changed the to support for your use case.\n, The text you proposed doesn't address the situation. Referencing the post I made, the file in question was unstaged. Therefore the text you propose doesn't address the unstaged files that are triggering this dialog box. You could change the text to but that's messy. Having two dialog boxes (one for unsaved staged, one for unsaved unstaged) is also messy. And one dialog box for each file is equally messy. All of this suggests to me that the feature itself needs rethinking, not just text in the dialog box. Personally (as a focus group of one), I don't want the feature. But someone somewhere though this feature was cool. So exploring with that group can perhaps surface a better mechanism to address their needs.\nI see what you mean now. I simply removed the qualification for now. Thanks for the ping. I don't really think this small feature warrants such a big investment as a user study group, though.\nOne more small thing, the warning says But if the file has saved changes and also some unsaved changes, then the file will be included, it's just the unsaved changes that won't be included. This seems like it would be a common case so maybe it could say \"will be incomplete in the commit\" or \"is unsaved and those changes will not be included\" or \"has unsaved changed that won't be included in the commit\" something.\nChanged to ."}
{"_id":"q-en-vscode-d58e2d1735199ddcd5474620043339a7fdf087137bc5051b72cd91a1742f8ce8","text":"Issue Type: <bBug</bRepro steps: settings to for a setting screen really large :bug: Content does not center with the header <img width=\"1282\" alt=\"image\" src=\"https://user-\"VS Code version: Code - Insiders 1.31.0-insider (, 2019-01-23T06:13:04.232Z) OS version: Darwin x64 18.2.0 <details<summarySystem Info</summary </details<details<summaryExtensions (6)</summaryExtensionVersion ------ vscode-browser-preview0.2.5 vscode-svgviewer1.4.7 prettier-vscode1.7.3 vscode-typescript-tslint-plugin0.3.0 debugger-for-chrome4.11.1 code-settings-sync3.2.4 (5 theme extensions excluded) </details<!-- generated by issue reporter --"}
{"_id":"q-en-vscode-b5aaf96a1140e9f1ee95a199cffa4e79f06a00fc086642e5be5225eef65c7b62","text":"VSCode Version: 1.31.0 OS Version: windows 10 Steps to Reproduce: the Help menu and select the release notes. any menu again click, right click, or select operation on this content page will not automatically close the open menu. <!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes\n(Experimental duplicate detection) Thanks for submitting this issue. Please also check if it is already covered by an existing one, like: <!-- score: 0.502 --- <!-- score: 0.45 --<!-- potentialduplicatescomment --\nDo not repeat with the above questions\nSame root cause as I believe\nActually will keep open to make sure this is resolved by the fix"}
{"_id":"q-en-vscode-7af342caba10683e34814bd3ea27fd8020976b46e36844702028f7a9103da1bb","text":"Broken due to refactorings of the workbench CSS. Fix is simple (CSS only change)."}
{"_id":"q-en-vscode-ec26a30ada59fc03a00eacb12a6b5b8b06d337800dfed7f5eabb42d5a608ab1e","text":"I do not see a duplicate of my issue: ! So I'm in the clear. <!-- Use Help Report Issue to prefill these. --- VSCode Version: 1.32.0 OS Version: Windows 10 1809 Stable Steps to Reproduce: Visual Studio Code, click on the \"Search\" sidebar menu item to bring up the huge Search menu. ! for a common word used quite a lot. Like \"a\", \"let\", \"var\", \"i\", and so forth. the Search displays the total results count, dismiss a few of them. total results count should decrement by 1 per dismissal, or indicate both the original total results count and the remaining non-dismissed total results count side by side when dismissing the search results. For example: ! Does this issue occur when all extensions are disabled?: Yes\nFor me, it does decrement. You don't see that? You are clicking the 'x' icon on a result row, right?\nIt doesn't. !\nOk, so I was about to get it to decrement, but that's not done in real-time. ! Both of the GIFs are done in 1.32.1.\nThat's right, I thought you were talking about the total shown at the top (\"x results in y files\") in your original post.\nThis also happens when a new match is or deleted in an existing file match. !"}
{"_id":"q-en-vscode-4d777b8ef35876ef9b60b422fd8787885b220a4aefebc5fc31e042ac83434d3d","text":"Issue Type: <bBug</bSee also - but note that people say they Debug console behaves normally Debug console scrolls to top after each evaluated line this code as breakpoints at \"Uncaught exceptions\" in VS Code hitting SystemExit exception, run in Debug console. a bunch of times (1-5 times) in the Debug console. that at one point, Debug console scrolls to top to bottom of Debug console output in Debug console. that you don't see due to Debug console scrolling near top again. VS Code version: Code 1.32.1 (, 2019-03-08T00:51:49.278Z) OS version: WindowsNT x64 10.0. <details<summarySystem Info</summary </details<details<summaryExtensions (5)</summaryExtensionVersion ------ latex-workshop6.2.0 svn-scm1.47.1 python2019.2.5558 multi-command1.3.0 code-settings-sync3.2.7 </details<!-- generated by issue reporter --\n(Experimental duplicate detection) Thanks for submitting this issue. Please also check if it is already covered by an existing one, like: <!-- score: 0.628 --- <!-- score: 0.584 --- <!-- score: 0.578 --- <!-- score: 0.544 --- <!-- score: 0.502 --<!-- potentialduplicatescomment --\nYes, looks like is a duplicate of this one, although does not have specific reproduction steps. This one does. Also, it looks like they investigated 1.32.0-insider, while I am still seeing this in 1.32.1. Also, I can repro this in the latest insider build with only the Python extension and no settings changed whatsoever (see below). Finally, I must admit finding repro steps was a bit difficult. So I am not fully convinced this issue is fixed, as implied by .\nI can not reproduce this. Can you attach a gif / video showing this can you repro this using python?\nSure, here's a GIF: !\nI can not reproduce with node. please try reproducing with python if you F1 developer tools put a breakpoint what do you see as scrollTop, scrollHeight and renderHeight at the exact moment when the debug console scrolls unexpectedly?\nI encountered exactly the same issue as\nthen you can also try to answer the quesionts i asked . Get your hands dirty and do some debugging. Thanks\nBreakpoint is hit twice per manual . After the first time manually entering , with the breakpoint hit at line 85: scrollTop = 0, scrollHeight = 19, renderHeight = 214 scrollTop = 0, scrollHeight = 480, renderHeight = 214 After the second time manually entering : scrollTop = 266, scrollHeight = 480, renderHeight = 214 scrollTop = 266, scrollHeight = 80, renderHeight = 214 (+) After (+), debug console is scrolled to top. I noticed that during (+), the debug console is empty, but not during any of the other breakpoint hits. I double-checked: is not a typo of , the above numbers are correct.\nYes, I too can replicate this issue on my Mac\nOk, thanks for help everybody. I can repro on my mac. Investigating...\nAfter investigating and discussion with we figured out it is a dup of\nVersion: 1.34.0-insider Commit: Date: 2019-04-19T19:59:19.092Z Electron: 3.1.8 Chrome: 66.0.3359.181 10.2.0 V8: 6.6.346.32 OS: Darwin x64 18.2.0 This issue is not fixed in latest Insider and bugs me every day: !\nI'm facing this since few months now. Any clue on how to avoid this issue? Please do not keep on closing this issue as it is real.\nCheck the insider release and see if it's fixed for you (it was for me). It seems the maintainers have chosen to leave this broken for everyone until their next major release next month which is unfortunate.\nI'm on Insiders, it's not fixed for me.. Version: 1.34.0-insider Commit: Date: 2019-04-19T19:59:19.092Z Electron: 3.1.8 Chrome: 66.0.3359.181 10.2.0 V8: 6.6.346.32 OS: Darwin x64 18.2.0\nI have the same issue. Version: 1.33.1 Commit: Date: 2019-04-11T08:20:22.771Z Electron: 3.1.6 Chrome: 66.0.3359.181 10.2.0 V8: 6.6.346.32 OS: Linux x64 4.15.0-34-generic snap\nI'm having this issue even on the latest Insiders version (1.34.20-insider) See gif here :\nSame issue here. Version: 1.33.1 (system setup) Commit: Date: 2019-04-11T08:27:09.080Z Electron: 3.1.6 Chrome: 66.0.3359.181 10.2.0 V8: 6.6.346.32 OS: Windows_NT ia32 10.0.\nplease do not close this issue. I am experiencing same problem: VSCode version: 1.33.1 System Software Overview: System Version: macOS 10.14.4 (18E226) Kernel Version: Darwin 18.5.0 Boot Volume: Mac OS Boot Mode: Normal Computer Name: iMac001 Secure Virtual Memory: Enabled System Integrity Protection: Enabled Time since boot: 7 days 6 minutes\nsorry that this is a pain. I am looking into it but can not reproduce. Can you provide exact steps please?\nHi OP here. I can still repro the bug in official 1.34.0, so this is certainly not fixed. However, my repro steps from the original description do not work any longer as running actually ends the process that is debugged. I managed to repro the same issue with . So these are my current repro steps: this code as in VS Code hitting the breakpoint, run in Debug console. a bunch of times (1-5 times) in the Debug console. that at one point, Debug console scrolls up quite a bit to bottom of Debug console output in Debug console. that you don't see due to Debug console scrolling near top again. Here's a gif as well. (The gif misses step 6, but I think it gets the main issue across.) !\nThanks for the steps, I will look into this this or next week.\nwe have pushed a fix with the help of Please try it with VSCode insiders starting from Tuesday which will have the fix. Thank you\nI have just checked and it seems to me the issue is gone, thank you! (Of course, I will be more confident once an actual release has the fix as I haven't re-setup insiders the exact same way I use my release version - let's hope for the best!)\nThanks. Adding verified label. Let us know how it goes once you get the stable update.\nStable looks good to me, too, thanks!"}
{"_id":"q-en-vscode-589de69925f42ed39d15416ea7c335760f121b55e29de7965e31df2419bded82","text":"Hi, I am a bit new, and I have a simple code example on VS Code v1.32.3. It says I have an error, ok fine..... but then when I chose the \"Quick Fix\" option there is never a Quick Fix for this or any other thing in the past I tried to do a quick fix on. Whats the point of having a Quick Fix feature, if it doesn't do anything ? Im a bit confused. . This is the code I tried to do a quick fix on (its an extremely simple code): . ! When I try to fix, I get this: . !\nI also get this issue for Python but seems to apply to any language. I have the Python extension as well as Intellicode. Still a current issue with 1.33 (Insiders) with the latest update.\nQuick Fix... action is shown even when there are no quick fixes (code actions) available. May be it can hidden then.\nOptimally, an actual corrective action would be taken (or at least suggested) based on the feedback provided by the language extensions (Python and Intellicode for my use-case).\nSame here, I always get the \"No code actions available\" when I try to Quick Fix my code, no matter the language used, and even if I have no extension installed. ! ! I do not really know why it is not working, but I can't get to remember since when this bug is occurring to me ( I remember last summer it was okay, and I think during December too while I was working on an Angular Project )\nOur C/C++ extensions does have lightbulb code actions for the error squiggle, but the quick fix still says \"no code action\" available -- is that a bug? Let us know if our extension isn't doing something correctly. I didn't understand the docs at . ! !\nCan you please provide simple steps to repro this?\nIt occurs when you have a .c or .cpp file with text . We report the squiggle message via textDocumentpublishDiagnostics and the code action from textDocumentcodeAction. Do we need to do something special to associate the code action with the squiggle's QuickFix?\nQuick Fix action asks for all code actions in that diagnostic range and of kind . What kind are your quick fix code actions?\nOkay thanks a lot for informing us -- the \"CodeActionKind\" type is completely new to us. Our code action just returns a Command and not the additional info like the CodeActionKind. I filed a work item for us at . We have a lot of \"catching up\" to do with the LSP updates.\nI have had this issue for weeks with C#, and I finally figured out that it was just the menu. Basic stuff where it used to suggest \"using System;\" no longer showed up. After reading this issue, I tried the command and the keyboard shortcut works but not the menu. This image shows what happens when clicking the \"Quick fix...\" menu: <img width=\"387\" alt=\"screenshot from menu\" src=\"https://user-\"And this is what shows when pressing : <img width=\"414\" alt=\"screenshot from keyboard\" src=\"https://user-\"This looks similar to the screenshots from and .\nSeriously kills productivity, I am reinstalling PyCharm\nLooks like C# extension is also not using the code action kind for their code actions. Probably we have to show all code actions for the given range.\nI see the same issue with JSHint/JavaScript. ! !\nThe same thing (No code actions available) with PowerShell.\nSame issue with TypeScript in .tsx file. ! !\nThis has been nearly two months! any fixes for this ? I'm having this issue in the latest version as well(Python)!\nThis feature becomes very important when you're upgrading code from a legacy version (such as Python 2) to the updated version (Python 3). I'm sure there are other languages have similar stipulations such as converting C to C++ or C# or vice versa (or even C# to C++ and vice versa).\nI also have this issue, and this is very important when create the new project, but the works for me!\nI also had this issue and only learned today from some searching that (on Linux) brings up the quick fix list of actual working suggestions EDIT: As shows below, that's the same issue I find when hovering using C#\nThis doesn't seem to be a problem for C# in my case atm (I'm using Unity/Mono). I exclusively use though.\nI get the same with Powershell. ! ! However selecting it and press ctrl+. gave me the suggestion !\nOn a Mac, the Quick Fix shortcut appears to be ⌘. (Command + Period). It would be super helpful if we could get the quick fix to appear on mouse hover, though!\nHello there, I have the same problem since today. Quick fix was working a week ago but now clicking on \"quick fix\" and then clicking on suggestion do nothing. But if I use ctrl+; to show suggestion it work.\nJust to clarify what was fixed by : we now only show the button in the hover if we know for sure that that quick fixes are available. A few extensions provide quick fix style code action but do not explicitly mark them as quick fixes. Those code actions will show up in the lightbulb menu but not if you click in errors hovers (because we only want to show code actions that we know for sure will fix something). Extensions must mark their quick fix code actions as  to have them show up in the problem hover menu too\ncan you verify that the python extension is good in this regard?\nI took a look at my installed extensions and noticed the C# extension was complaining that it needed to be reloaded. Hmm, so check your extensions list first. This is what I was presented after clicking reload. Notice the version I'm running as of today's date stamp after I reloaded the C# extension. I hope this helps some of you. ! ! ! Be sure to click on the instance first, then use ctrl+. for windows ! You should now see the using statement, for example like on line 1 !"}
{"_id":"q-en-vscode-e9246eac3587555dbd876ce8bd09a868f83fb7338c8d642293c42a06df7eba9d","text":"<!-- Please search existing issues to avoid creating duplicates. --<!-- Also please test using the latest insiders build to make sure your issue has not already been fixed: --<!-- Use Help Report Issue to prefill these. --- VSCode Version: 1.33.1 OS Version: Windows_NT x64 10.0. Steps to Reproduce: a new user snippet file for the project. The one I encounter the issue with is: the snippet and attempt to press TAB after writing the first variable cursor does not move to the next stop, instead a tab character is The desired behavior would be for the cursor to move to the next tab stop. <!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes\nSeems to work just fine for me... Any special keybinding or setting? !\nThe action you took is not the problem I encounter and does indeed work the same way for me. Could you try to edit \"SystemName\", the first tab stop, then press TAB to go to \"Required Components\"? vsCode just inserts a tab character after whatever I wrote instead of going to the next tab stop.\nOk, thanks for clarifying. For some reason snippet mode stops when making that edit... Not sure why\nI have encountered similar issue, however, I could add a single blank space in the snippet which \"fixed\" the tab navigation. Also, my example is far simpler. See\nThis is a regression from I will revert that change and re-open the corresponding issue (which is much more rare than these issues)\nThank you for fixing this, looking forward to the next release."}
{"_id":"q-en-vscode-6968627360c93657a682bf1a6c0779503f12830189cd50d5e05ea38072a07cb5","text":"VSCode Version: 1.33.1 OS Version: WindowsNT x64 10.0. Steps to Reproduce: folder (Ctrl+K F) a search on specified path through Ctrl+Shift+F ! With literally any folder on my system opened search works: ! I'm not sure if this is intended behavior, as the message suggests that it shouldn't work if I don't have that particular_ folder currently opened anyways, and yet it does if I open anything. Does this issue occur when all extensions are disabled?: Yes\nI was just about to file an issue to find out if searching an arbitrary folder / folder tree was even possible since the note says only the currently opened folder will be searched. This is one of the things Sublime has that I was hoping VS Code had. I never tried opening any ol' folder and giving it a run.\n, can i take a look at this ?\nShould fix the message 'You have not yet opened a folder. Only open files are currently searched - Open Folder' makes it sound like my searches will fail with no folder open, but as long as I plug a path in it works like the issue specifies. Maybe add something along the lines of 'unless a path is specified'\nI could say, \"You have not opened or specified a folder, ...\" I'm not sure if that's too wordy, I don't think a warning message has to explain how the whole thing works, what do you think?\nMake sure you reopen issues when they are \"verification-found\"\nThis would work. How would it work once I specified a folder but haven't hit send would the message go away due to having a valid path?\nNo, I think it should just be updated once when the search actually runs\nThat seems like it would work, it would at least imply that the action being performed would work.\nNow it makes more sense !"}
{"_id":"q-en-vscode-4acb251473a1944dec0e0065730d2e6566f8fe612337d84e9d77c70bbe9d5d42","text":"Issue Type: <bBug</bHi. I am assigning trade.occId = occId; i.e. to a property (in this case it contains 4 spaces. In the JavaScript hover view however, the property on 'trade' reads (only one space) [btw, I copied both these values right out of the IDE view]. I've just been chasing ghosts for some time because of this discrepancy and not expecting the IDE altering assigned values when viewing objects. Update ---Alright, sorry. I am not sure how to submit the first value on GitHub with the 4 spaces between GS and 1 preserved, even when using 'code' quotes (same as the VSC issue...). So the first value is \"GD....190517P00155000\" when replacing spaces with dots... Thanks for VS Code otherwise :-) VS Code version: Code 1.33.1 (, 2019-04-11T08:27:14.102Z) OS version: WindowsNT x64 10.0. <details<summarySystem Info</summary </details<details<summaryExtensions (4)</summaryExtensionVersion ------ vscode-scss0.6.2 debugger-for-chrome4.11.3 vscode-scss-formatter1.4.1 vscodeintellicode1.1.5 </details<!-- generated by issue reporter --\nCan you please provide a screenshot of what you see? Also try disabling all extensions. I am not sure if such an hover is provided by out of the box typescript/javascript extension.\nHi. Please see below. I disabled all extensions besides Debugger for Chrome, which I need to set a break-point to be able to inspect values. The issue is in most obvious in the field 'occId', which in fact has 4 spaces after 'GD' and not just one as displayed. Thanks for following up. [image: ] !\nAh I see.. its an hover while debugging.\nYou did not attach an image. So please attach an image and some reproducable steps so we can investigate more. Thanks\nIt did, (please check below the quote). Anyway, your question made me come up with a better/simpler example that should make things clear. Here you go: ! Btw, Gruetzi to all you guys in Zuerich. I'm originally from Basel :)\nThanks for the image. Cheers back to Basel :)\nVSCode insiders will have the fix for this on monday. Try it out and let us know if it is ok for you. Though I verified and the fix from looks good\nThanks for the work and update. Unfortunately I can't switch from VSCode stable, so I'll take your word for it until this hits release :-)"}
{"_id":"q-en-vscode-6139cd524bfe3015d0e83c2123fd64e55a2a2eb4c06086b8a3fae2500e22c9a6","text":"Steps to reproduce: clone open a terminal Debug viewlet Launch client Observe that building never finishes Now switch the terminal to the npm: watch task Observe: building finished. Looks like the terminal is not delivering any output as long as the view is not visible.\nare there any differences on when onLineData occurs in the terminal with conpty?\ndiscovered a new issue"}
{"_id":"q-en-vscode-5e2372ce728040aa42f3cca7980ec324f48f4df14c0ffb9bb9f3e75a9dfbab18","text":"Steps: right click into explorer root node you see native and custom menu ! Looks like the new tree is not preventing default and bubbling of event like the old tree did?"}
{"_id":"q-en-vscode-0bb2417eb6bd18945b4db3a7dc8a821a1a41117cbd2a2f13f274a60df207ee2e","text":"Issue Type: <bBug</bWe are developing an extension that implements a source control provider. It successfully adds s to its section. However the standard themes display the group captions in UPPERCASE by styling them with For our use case it's important to show the group captions in their actual case. AFAIK there isn't a way for our extension to make that happen. VS Code version: Code 1.34.0 (, 2019-05-15T21:59:37.030Z) OS version: Windows_NT x64 10.0. <!-- generated by issue reporter --\nYeah... let's then change the casing style and just uppercase them on the Git extension directly.\n, Removing the styling , is that it or I should add it somewhere ? Because you said\nYeah we want to keep the git extension section as , etc.\n, Could not find the exact detail of what to remove. Removing the style will remove it in all places , so that is not correct. I dont understand what group captions actually mean ( CC : , if you could help ) I found there are these ResourceGroup () for which the style is applied. Please help me out with this on the correct approach.\nI think proposes that you do remove that style, but that you also change the code to explicitly uppercase the localized captions of the the three items being created there. With those changes Git users won't see any change of behaviour, but other SCM extensions can set mixed-case captions and have them display to the end user.\n, Sorry got confused with the group captions , Sent a PR to fix this !\nI'd love to see your PR get merged, but for some reason I don't understand it apparently failed in the Windows part of Azure Pipelines. can you help get this moving?\nThe windows build error might be some random errors , Happened in one more PR ( ) I guess that is not an issue and should confirm that.\nIs there some way you can resubmit your PR to the build pipeline? Maybe it will pass the second time. Or if it doesn't perhaps there will be more info in the log."}
{"_id":"q-en-vscode-64148e872d2bb5f8cd341a83f57edab816a058cdb6c4a518b0192e8d70300555","text":"VSCode Version: 1.35.0 OS Version: Windows 10 Steps to Reproduce: With 64-bit VS Code installed, run run VSCodeSetup-ia32- with /SUPPRESSMSGBOXES /VERYSILENT You'll get a dialog that says \"please uninstall the 64-version before attempting to install the 32-bit version.”\nHello. I can take this issue if you want, I have reproduced and identified the bug. I will open a PR when ready :+1:\nVerification: 32 bit to install 64 bit, msg box should come up to install 64 bit from the command line with /VERYSILENT, msg box should not come up\nor still displays the message box.\nIt appears that latest insider release was built using this commit: Commit fixing this is was hours later: Is it possible that insiders version doesn't include the fix yet ?\nOh yes. Normally I catch myself on this, but It's still early in the morning I guess. Thank you for pointing that out , I'll test it once insiders contains the fix."}
{"_id":"q-en-vscode-82b12bc7b10cd2ceb7c08b184832e12ac5b908a468b6a1858a1b6f631f8d5aba","text":"I'd like a setting to optionally prevent deprecated things from showing up in intellisense suggestions. For example, I'd like a way to prevent from being suggested here, so that I don't use it despite a non-deprecated version being available. ! My linter can catch it, but it'd be even nicer never to see the deprecated option. There are several documentation systems that use the tag: JSDoc: TSDoc: JavaDoc: Yard (for Ruby): Other systems use different, similar tags. I'm not sure if this is an issue for the VSCode core, or for something language-specific.\nSorry if this is a duplicate or is already possible. There are enough issues here and questions/answers on Stack Overflow that it's hard to be confident. This issue is related:\nNot showing deprecated suggestions can be implemented by individual language extensions today (not striking out but tracks that). Adding just to see this is something we want better VS Code api support for in some way Otherwise, we can track this as a feature request for js/s\nI'd say this is a dupe of and it would be us adding some API flag and then allows users to configure how the suggest UX is handling that flag\n/duplicate of\nThanks for creating this issue! We figured it's covering the same as another one we already have. Thus, we closed this one as a duplicate. You can search for existing issues . See also our guidelines. Happy Coding!\nOkay, that makes sense. What does \"js/s\" mean?\nis it possible to add a setting like , so I can disable it to prevent deprecated from showing up in suggestions list?\nHey any update on that?\nVerify that the setting exists and that it filters deprecated elements from IntelliSense"}
{"_id":"q-en-vscode-0c9c22b87fe2337301c863c8066e21515a10609a415a183ec1ce3dc671003846","text":"Steps to Reproduce: a couple of branches on the branch picker =:bug: the branches are no longer sorted by commit date This works fine in stable, so seems like a regression. INSIDERS ! STABLE ! (my git version is )\nStable: Insiders: Yup, there's a missing."}
{"_id":"q-en-vscode-b9358f74d1dc445c28527efa4d7eb2a7e3c5f1970797693e790a6b017dd6a89b","text":"Problem: A popup opens, displaying: \"The 'Remote - WSL' extension is recommended for opening a terminal in WSL.\" It has a 'Dont Show Again' and an 'Install' option. Clicking Don't Show Again has no effect. Expected Behaviour: Once I click Don't Show Again, this popup doesn't appear every single time I open the WSL terminal in VS Code. VSCode Version: 1.38.1 (user setup) OS Version: Windows_NT x64 10.0. Steps to Reproduce: WSL Ubuntu installed on Windows. the WSL integrated terminal on VS Code. Does this issue occur when all extensions are disabled?: No.\n(Experimental duplicate detection) Thanks for submitting this issue. Please also check if it is already covered by an existing one, like: <!-- score: 0.614 --- <!-- score: 0.559 --- <!-- score: 0.549 --- <!-- score: 0.54 --- <!-- score: 0.52 --<!-- potentialduplicatescomment --\nI wanna take a shot at this!\nDoes this happen on different workspaces or on the same workspace? Edit: I have checked and this only happens when you open different workspaces (Different project folders with VSCode) I personally think that this should not be the case and should be global so I will make a pull request for it.\nSorry for the late reply. I see you've already pushed a fix for this, so thanks a lot :)\nShould I close the issue now?"}
{"_id":"q-en-vscode-4116c9419f661b23ee3e6e016f7b47bee837be805116a76f7f4c8aed32f862d1","text":"Steps to Reproduce: web in Safari on compact menu = it moves !\nI diagnosed this and see 2 issues: the menu items are not removed from the DOM when menu is not showing () you seem to be using instead of to hide the menu items, is there any particular reason? using the issue is not as bad anymore (I only see one initial jump) I suggest we use / (our helper), but I am not sure if you rely on the behavior of visibility?\nI mentioned the reasoning in the linked issue, but in the compact case we can remove them from the DOM safely so patching that should help here."}
{"_id":"q-en-vscode-6374a496026838802bee60dd9759185721c32a7db393b85961966878ba3e26e9","text":"Issue Type: <bBug</ba repo with a large amount of changed files in one folder and use the tree view <kbd+</kbdnext to the folder to stage the changes or <kbd-</kbdto unstage them. : ! This was supposedly fixed in for the discard command but either the other commands or the tree view was forgotten as a potential source of problems. Clicking the button next to \"changes\" works fine: ! VS Code version: Code 1.40.0 (, 2019-11-06T17:02:13.381Z) OS version: WindowsNT x64 6.1.7601 <details<summarySystem Info</summary </details<details<summaryExtensions (16)</summaryExtensionVersion ------ vscode-eslint1.9.1 gitlens10.1.2 toggle-excluded-files1.7.0 EditorConfig0.14.2 RunOnSave0.2.0 prettier-vscode3.3.5 vscode-test-explorer2.14.6 rest-client0.22.2 vscode-docker0.8.2 vscode-language-pack-de1.40.2 vscode-jest3.0.2 project-snippets0.5.0 java0.52.0 vscode-coverage-gutters2.4.2 vscode-icons9.5.0 vscode-todo-highlight1.0.4 </details<!-- generated by issue reporter --\nYup, the same fix as needs to be applied to the stage action.\n+1 Same issue here in the latest vs code and latest git Windows 10 home: v10.0. VS Code: 1.44.2 Git version: 2.26.2.windows.1"}
{"_id":"q-en-vscode-df898a494ac2c62996c345c43c9024626cdf7644ef022c33212472338ca6da7a","text":"I use VSCode in three different places; my home computer, my work computer, and as a portable version I carry on a drive when I need to use a machine that doesn't have it. I use an extension to keep my settings synced up between editors, and the only issue I've encountered so far is that the git path doesn't match between any of them. On my home machine I have it installed to C of course, work likes to be funny and install it on A, and for the one on my drive I have a relative path set so that no matter what letter my drive gets, that VSCode can always find git. I already attempted to use an array myself just to see if it'd work: But VSCode reads it as one entire value. What I'd like is for it to recognize it as an array and then try each path in order until it finds Git or runs out of paths.\nIt is better to create an issue in extension repo with requesting the settings ignore feature.\nHy\nThis is a good feature request, please feel free to submit a PR.\nAh, thanks. I must have just seen the section title and completely glossed over it when I was reading through the docs.\nI could help out with this feature-request, if you would like. At a glance, it looks like it would require a modification to the \"createModel()\" function in the \"\" file (line 34). We could get the \"pathHint\" as a string array and then possibly modify the logic inside of the \"findGit()\" function, or, if possible, iterate through the \"pathHint\" array until a valid \"\" value is returned. EDIT: Sorry, didn't realize a PR was already submitted...\nRaised PR!\nVerification The setting now accepts a list of values. The git extension must go through them one by one until it finds a usable git executable."}
{"_id":"q-en-vscode-6df23f16d44f7775deb3d73b09233f292f56642303f9b82c81746c0f3b335274","text":"Snippets that use the SELECTION variable should be grouped into a new \"Surround With...\"-command, e.g support selection with try-catch or if-else. We should also consider not showing such snippets in IntelliSense because the selection is always where users are typing.\nclosing this in favour of\nAfter and given that a dedicated command is more visible we should add this\nhi, Thanks for your adding support for outside code and could please add some identification for those snippet support and/or when users are using command e.g.: when using insert snippet, users never know which snippet is outside-code supported there is the ui: ! and the snippet code from C# official extension ! for that reason, I have to add some tip/text for the name of snippet when I develop the snippet extension. after that, user could know that which is outside code supported. ! I could do it for myself and my extension, I there are many snippets that have no tip or identification Please add some identification for those outside code supported in UI when users using insert snippet. that will be useful and helpful for user to distinguish\n! maybe some icon for text is good\nBTW, there is another issues related to this. please\nhi, should I file a new issue?\nTo verify open a typescript files select some code select F1 Surround with Snippet try-catch ensure the selected code is surrounded by the snippet\nVerification looked good with both TS and CS files."}
{"_id":"q-en-vscode-7958450a329d220fb184fadefc5e01efc9b3d561836a59f6ae99a7fe90543724","text":"<!-- Please search existing issues to avoid creating duplicates. --I noticed in the settings-editor that there are now duplicate clear buttons when the user searches but finds no result. <!-- Describe the feature you'd like. --I propose getting rid of the \"Clear Search\" text in the bottom left corner as there is now a button in the search bar that achieves the same thing but with less clutter (in my opinion). This change would make the settings-editor more consistent with the keyboard shortcuts tab as well. !\n(Experimental duplicate detection) Thanks for submitting this issue. Please also check if it is already covered by an existing one, like: <!-- score: 0.464 --<!-- potentialduplicatescomment --\nYeah we can probably just get rid of the text link\nWhat about the \"Clear Filters\" link? Should there be an extra Clear Filters button that appears in the search bar itself, or let the link stay?\nWould you mind sending me a picture of the \"Clear Filters\" link ? Where exactly is the link ? I will look into this.\n! When you use one of the @ filters it appears.\nFor now I think we can let the 'Clear Filters' link stay. This issue was originally meant to get rid of the duplicates. In another PR I could add a button to the search bar that adds this , and then we could get rid of that link though, although other tabs don't have that sort of button . What do you think ?\nYeah the \"Clear Filters\" link is all about discoverability and I'd rather leave it"}
{"_id":"q-en-vscode-9059b295b72941b51e13ec7ca0658879eb15eecc4a9aa9ad444d272edccf0500","text":"A lot of extensions have configuration code. It is extremely annoying that since some version ago we are unable to copy anything from that pane, mouse or keyboard. Steps to reproduce: Go to extension description Try to copy something Video:\nSee I'm surprised that you can't copy using the keyboard. What key sequence are you using? What command(s) does your Keyboard Shortcuts page show that bound to?\nIm using the usual cmd+c which works in other apps. And cmd+c also works in the editor mode, terminal mode - so i think its not my key bindings customization. I recorded video in presenter mode (displaying what i press) showing copying something else + not working for ext description:"}
{"_id":"q-en-vscode-fc0dce5092f55ea4c88baac7dcda853e8f1a3df4ec8a05d977010a7ed10dd401","text":"VSCode Version: Code 1.41.0 (, 2019-12-11T18:37:42.077Z) OS Version: Windows_NT x64 10.0. Steps to Reproduce: a Markdown file on a network shared folder (ex.: ). that network shared folder to a drive letter in Windows (ex.: ). the Markdown file in VS Code using the mapped drive letter (ex.: ) and open its Side-by-Side Preview (). The preview should be correctly formatted. the same Markdown file in VS Code, but using the file's UNC path (ex.: ) and open its Side-by-Side Preview. The preview is incorrectly formatted. Does this issue occur when all extensions are disabled?: Yes Correct format (file opened through mapped drive letter): ! Wrong format (file opened through UNC path): !\nI would like to work on this if possible.\nThere was a typo in the code was causing this. I have tested on Windows 10 with a file on a UNC path with the fix above and it is working fine."}
{"_id":"q-en-vscode-be704571296d2202d0d0cdbe456ec50ac0be6b073d62c4c61f79c6cf47f0d0ec","text":"If you return in a completion item, the property is when calling through . Setup: Test code: Output is: Full repro project is here:\nIt looks like the conversion is missing. The doc states that additionalTextEdits should not overlap with the main edit range or each other, should that be checked or is that done when the edits are carried out? Adding this: ! works, unless the range intersections need to be checked.\nNo checks required - this is already the way back from the UX and check should have been carried out already"}
{"_id":"q-en-vscode-acb78ccfec4362df4565adf511a8a4eed9f781f3d7c08a2f0abed9ca399894bb","text":"<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --Currently, when a debugger is executed (ie by hitting F5), a terminal is opened and the debugger is executed as a command. This command can get pretty long as it also changes to the working directory and sets some environment variables. Running the Python debugger in a simple project with only one file runs a command which is 243 characters long. This, however, means that these commands which are not meant to be run by the user but are instead run automatically are still included in the user's shell history file by default. I propose changing the command generation function to include a space at the very beginning - on many shells across various distros, is set to not add commands to history if they start with a space character. This also gives some control to the user whether they'd like for the debugger commands to be stored in history or not.\nIf you were wondering why someone might want this, I like that the history is there. All the time I go to my command line and use up arrow enter to rerun the last auto generated command. I've also gone through the history a bunch to rerun previous auto generated commands too.\nAye, this is why I opted for simply adding a space instead of something more aggressive such as - the shell is still in control whether to preserve history or not with . EDIT: Also, in my shell, commands which were run with a space can still be reached by pressing the up key in the active session, but I'm not sure this is the case in other shells.\nThis works for the integrated terminal, but I noticed the external terminal doesn't get the space on OSX !\nOn MacOS the leading space character seems to get dropped in the Applescript that talks to the Terminal application. So the feature at hand only works for the integrated terminal for now. I've created a for the external terminal. Since has verified that the fix works for the integrated terminal, I'm adding the \"verified\" label."}
{"_id":"q-en-vscode-5dd48cae206265bd572674144076f87c30090036659fcb881007129e6b1b6f7e","text":"Right now when a breakpoint is hit, the function in is called, and it in turn calls . The logic of is that it shows the wanted element at the very bottom of the scroll view. What this means is that in the Call Stack view, we'll only see the thread that hit the breakpoint and the first frame at the very bottom -- Ideally I think we want to show this at the top, so that we can see the whole stack trace, not just the first frame.\ntry it out in VS Code insiders from Friday and let us know if it behaves good for you so we add a verified label here. Thanks\ndid you get a chance to give this a spin?\nsorry for the late reply. I tested it on the latest Insiders Build and it works as expected -- on breakpoint, the stack view shows the current thread in the middle if there are multiple threads in the program and the behavior is not changed compared to the previous version if there are fewer threads.\nAlso another question for you I can see that you guys (i.e. the VSCode team at Microsoft) are working heavily on Debugging UX improvements and our teams at Facebook could also suggest a few improvements we could benefit from. What is the process on picking the improvements to work on? Is it only done via feature requests? If so, how do you pick and prioritize the requests?\nThanks for coming back, adding verified label. For suggestions on improvments please create new feature requests. Feel free to ping me on them. Thanks"}
{"_id":"q-en-vscode-9fa374cf1a71b9e6b0fcabc8f905543d73d1ef0432a3c237e32ebf812402334e","text":"Could be related to the Untitled file title changes. The tests hang here: ! /cc\nThe same on macOS."}
{"_id":"q-en-vscode-0cb5ab01cb10f3b4726d4b682e0186836b8e8487954d1af1799f08b7d04c2562","text":"<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --Add an method to the interface. The current options are: trying to stat the and seeing if it fails — we have no way to tell why it failed; reading a parent directory — we may not have read access to the parent directory; using node's — only works for the local filesystem.\nI see s are spec'd to thow on when the file doesn't exist, but does not spec such errors.\nWe have no plans for adding and instead should be used (this goes along the deprecated to NodeJS's -deprecation). Still, I do agree that we should spec the error\nWould be a passable solution?\nRe-open for API proposal/finalization. Proposal is to add a field to the -class\nshould I rename the issue to better reflect the proposal?\nReactions during the API meeting were positive, so I am confident that we can finalize this during the current milestone. One feedback item was to always have a value for , e.g. let its default value be something like or\nI did a quick search of the codebase, and it looks like is usually used for such cases in the various error code enums.\nThis is now in and will be stable API with the next relase. thanks for the help"}
{"_id":"q-en-vscode-14423c4d6e33b74eb3bbf95b9e588242528d14c7f9f01cea32f6419c7be72315","text":"Refs: A text document can be readonly if it's underlying file system provider is readonly. I am not sure how a custom editor implementor can figure this out and provide the correct experience.\nCould we add a property to ? Does VS Code internally already know if a given textDocument is read only?\nWe basically ask the file system provider for the resource of the document for being readonly or not. That should be doable."}
{"_id":"q-en-vscode-76175482a85d8556fd1378b338e7781c311daf23395037b59bba97d01af57571","text":"Stable above, insiders below: !\nDoesn't seem to happen on macOS and the regression happened in the last couple of days. have panel/view-related changes happened that could have caused this?\nAffects tasks/custom terminals pretty badly: <img width=\"146\" alt=\"Screen Shot 2020-02-28 at 6 42 45 AM\" src=\"https://user-\"<img width=\"226\" alt=\"Screen Shot 2020-02-28 at 6 43 15 AM\" src=\"https://user-\"\nLikely my fault from drag and drop change"}
{"_id":"q-en-vscode-07eb11dc6f9713b38a38cedcc9ba87b0eff424678cf33b0aa490654808b72ac8","text":"Following up on microsoft/TypeScript, is now a valid completion directive. It should be recommended over in the editor walkthrough that mentions comment directives.\n(Experimental duplicate detection) Thanks for submitting this issue. Please also check if it is already covered by an existing one, like: <!-- score: 0.668 --<!-- potentialduplicatescomment --"}
{"_id":"q-en-vscode-16c5838b6363a2d329d995eede075963d41f0db9e266fd9b6ecc146ad0744d5d","text":"<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Also please test using the latest insiders build to make sure your issue has not already been fixed: --<!-- Use Help Report Issue to prefill these. --- VSCode Version: 1.43.0 OS Version: n/a I have the set to in my settings: Given a Does not produce any warnings as expected. However in : The linter says\nThis has the same root cause as : script are owned by the html extension instead of by the JavaScript extension. Fixing that is difficult However we could have a small fix that just makes html understand the setting. Marking this as help-wanted\nHello I'd like to try my hands on fixing this issue. Can you give some pointers as to which files are good places to start?\nThis change will need to be made in the HTML extension Here's where we configure TypeScript for the html extension: You should: Pass the value for in with the initial configuration Make sure that the language service updates its value for when the VS Code setting changes\nIs there another way to get the setting value other than vscode.workspace.getConfiguration()? The 'vscode' module cannot be found when I try to import it into javascriptMode. I think 'vscode' would have to be as a dependency in in order to import 'vscode'.\nRight now I'm passing in a default 'false' for experimentalDecorators in the initial configuration and passing the settings into getJavascriptMode.doValidation to update compilerOptions and calling ts.createLanguageService(host) every time doValidation is invoked. It has the expected behavior, but is there a better way to update the compiler options without creating the service every time it runs validation?\nIs called when the file is changed? If it is, try updating the object that it returns\nI am reproducing this error now with the following code: I realize this is known, but I will add the mention with respect to . Here is the relevant docs page, if a person wanted to make a unit test using If anyone finds this in the future, if is set in , it should suppress this error. Additionally, both and files, in the root of the project, take precedence over that setting."}
{"_id":"q-en-vscode-093d5103722067bb107af7b0891466050beec43b213a9ebf4a4bb250140f28ca","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Also please test using the latest insiders build to make sure your issue has not already been fixed: --<!-- Use Help Report Issue to prefill these. --- VSCode Version: 1.44.2 OS Version: Ubuntu 19.10 Steps to Reproduce: navigate editors through QuickOpen \"navigate next\" and \"navigate previous\" work as \"navigate next after next\" and \"navigate previous of previous\". So simply said it skip next/previous editor. Here are keybindings I use ` works ok, but and don't <!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes\nPR: The issue is that we are listening to keys within the quick input control to do certain things but we do not prevent bubbling. This means our keybinding service will also receive these keys and check for actions to perform. In the old quick open widget we always prevented bubbling for that reason."}
{"_id":"q-en-vscode-aaada1df970c9be2e3a7044b5fddf8a25365fde88684cb1d7a42f4d819f48d87","text":"<!-- Do Not Delete This! bugreporttemplate --VSCode does a decent job highlighting many different shell script languages. But it doesn't seem to be aware that pdksh, dash, mksh, and posh, especially configuration files like and shebanged extensionless scripts, are POSIXy shell scripts. They just highlight as Plain Text files instead. Same for , that's a Tcsh shell script.\nPR welcome!"}
{"_id":"q-en-vscode-0fe708a240a64c1e376ae00e8e3e678584f0e07c8d1275b42ffe37ff9574cf72","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Also please test using the latest insiders build to make sure your issue has not already been fixed: --<!-- Use Help Report Issue to prefill these. --- VSCode Version: 1.44.2 OS Version: macOS Steps to Reproduce: an extension providing in-editor comments via the comments API an editor containing comments the terminal at the bottom a command which causes the comments to be updated. For our comment provider, we automatically detect comments associated with your source control commit. So, for example, use the command line to checkout a different commit. comments will change, causing some of the previous comments to be hidden. When this happens, the editor is focused. This causes you to lose focus on the terminal. The end result UX is that you're typing commands in the terminal, and your focus gets stolen to an open editor for no visible reason, making it annoying if you wanted to keep running more commands. Notably, this depends on the way the comments API is used. I believe many comment providers do not respond to source control, and require clicking a button to activate/deactivate, so it's likely that's why this hasn't been reported before. For us, this is really annoying, and I could see this hitting other providers in the future. <!-- Launch with to check. --Does this issue occur when all extensions are disabled?: No, because you need an extension which contributes via comments API\n(Experimental duplicate detection) Thanks for submitting this issue. Please also check if it is already covered by an existing one, like: <!-- score: 0.562 --- <!-- score: 0.513 --- <!-- score: 0.501 --- <!-- score: 0.493 --- <!-- score: 0.492 --<!-- potentialduplicatescomment --"}
{"_id":"q-en-vscode-f542a8a8aa4fa8597c48fdf075aa183f5e438099357e8b42e4df39ce89d517ed","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --I'd like to suggest an improvement to the Monokai Dimmed theme. Monokai Dimmed doesn't have colors set, resulting in 'ugly looking' fancy terminal features, such as Powerline. For a nice looking improvement, the terminal colors can be taken from the Monokai theme: Before: ! After: ! I intend to open a Pull request shortly that realizes this suggestion.\nThis looks much better, thanks for the PR! ! !\nThanks for accepting the PR! As a side note, I quickly looked at the other themes, and many of them haven't set any terminal colors, and even the ones that have them set usually haven't aligned ANSI black with the background, leading to weird-looking 'fancy shell features': ! ! Then again, there are documented ways how to adjust the color palettes for a particular VSCode installation or workspace, and this is quite likely a bit of an edge case anyway. Thanks for all the good work you guys have been doing!\nIs there an option to disable this? or if there's not what would be the values one would need to set on terminal.X to revert what the commit changed? I prefer the non washed terminal and noticed this changed on last update. Edit: For anyone looking to use default terminal colors, add this to"}
{"_id":"q-en-vscode-b681f4089c30b3b40688768b958ae96e2d1147bdecc0fb9ac4f8117c9cd8d768","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Also please test using the latest insiders build to make sure your issue has not already been fixed: --<!-- Use Help Report Issue to prefill these. --- VS Code version: Code - Insiders 1.46.0-insider (, 2020-06-05T12:27:56.662Z) OS version: Darwin x64 18.7.0 Steps to Reproduce: all network connections. VSCode. \"Developer: Startup Performance\" from the Command Palette. Expected result: Startup Performance measurements are displayed. Actual result: Startup Performance never finishes \"Loading...\". Error logged in Dev Tools Console. ! <!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes\nFailing requests are from the extension service, not sure if this is intended during the startup metrics collection. Deferring to\nI do not know how the start up perf are recorded and I assume these network requests failures are unrelated to it. Any insights on this?\nI don't believe this is specific to the performance editor but must be rooted somewhere lower. The perf editor shows while it the eventually lifecycle phase and extensions being registered. So, I believe the culprit must be somewhere there\nAssigning to and for lifecycle phase and extensions being registered steps respectively\nRunning with and checking logs, the is reached: There are 2 more things the editor awaits on: Maybe Jo could check on the former and Alex on the latter.\nThe former is again installed extension registered and lifecycle phase awaiting\nthis call is never returning: But is needed for this barrier to open:\nGood digging. Thanks Adding to decide if the update service should eventually error or return false when asked for latest without internet access\nI think throwing would make more sense for a service call."}
{"_id":"q-en-vscode-ff83a646f55e9d93b65e72163fd4193e67f760834ba43523c5299c906c97d7b0","text":"Issue Type: <bBug</bPlotting a large plot (e.g seaborn pairplot) with many columns via jupyter interactive mode and then viewing in plots viewer is very slow to view/zoom/pan VS Code version: Code 1.45.1 (, 2020-05-14T08:27:35.169Z) OS version: WindowsNT x64 10.0. <details<summarySystem Info</summary </details<details<summaryExtensions (11)</summaryExtensionVersion ------ vscode-django0.20.0 vscode-markdownlint0.36.1 python-extension-pack1.6.0 gc-excelviewer2.1.35 r1.4.2 vsc-python-indent1.10.1 MagicPython1.1.0 anaconda-extension-pack1.0.1 python2020.6. vscode-yaml0.8.0 vscodeintellicode1.2.8 </details<!-- generated by issue reporter --\nThis sounds like a problem caused by the extension providing the Jupyter Notebooks. Please file the issue there.\nHi same problem here, are there any workarounds in place for this?\nI've moved to"}
{"_id":"q-en-vscode-2b0fc6c1f59da43b10493abb74a2af5ba3015f1471a624d95dacc38400a304e1","text":"Noticed how the process explorer is bundling and still. Can we not use Codicons for this? //cc\nOh this seems to be used in the issue reporter, not process explorer. But not sure how to actually see it...\nare we actually using these icons? I'm not seeing it: !\nit looks like these are only used when VSCode is connected to a remote to show the remote processes as well. I didn't even know we had that, pretty cool: ! Anyhow, my request still seems valid.\nPR up at !"}
{"_id":"q-en-vscode-f5d9638b55fe5ef0c9e31866be2d86242ad0faeed82ccd9ed3be581dae6e0b49","text":"Issue Type: <bBug</bSemantic token colorization fails when a token type ID matching any object.prototype property name is provided. Some examples are: , and . is the most problematic name since it is likely to be used as a token type ID. Any string should be able to be used as a token type ID without causing the colorization to fail. Because tries to call on what was thought to be a but was actually a , which throws a TypeError and causes the colorization to fail. The easiest way is probably to use a instead of an object for . Another solution might be to instantiate using instead of an object literal in order to avoid the object.prototype. VS Code version: Code 1.47.2 (, 2020-07-15T18:22:15.161Z) OS version: Linux x64 4.19.0-9-amd64 <!-- generated by issue reporter --"}
{"_id":"q-en-vscode-1baad8997c221271bbc64af733ad3ff6ff3339d9dc8b4a3c72c28822d0c53c29","text":"latest VS Code insiders, turn on Voice Over on mac P -notice how the screen reader does not read any of the items in the quick pick This works just fine on Linux and Windows. Also works fine for stable. So it seems like an chrome update in electron broke this I can also repro running in . In Firefox and Safari all is good. I can not reproduce in Chrome cannary, so there seems to be a fix in Chrome latest. Ideally we would find the fix in Chrome and backport it to our Electron. fyi And for Chrome smartness, maybe one of your collegues fixed something like this (should be related to )\nI can add this to my list (which is starting to grow I'm afraid). BUT, you don't need smartness; you need the most excellent bisect-build tool! :smile: Docs here: Once you have the regression range, you just sit there as pre-built binaries are presented for you to try and mark as good or bad. Best. Tool. Ever. Lemme know if you have any questions or need help.\nThanks a lot. Let's see if already has this set up, otherwise I can also look into it.\nYup I can do the bisect shortly.\nBisect result for when regression was introduced: Potential candidate: https://chromium-\nLooking for fix now.\nUpstream fix: https://chromium- Crbug:\nImpressive. Thanks a lot!"}
{"_id":"q-en-vscode-98e37356f693fec921eadc275429918723824afb5e9488b64664d59684b6c5e5","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Also please test using the latest insiders build to make sure your issue has not already been fixed: --<!-- Use Help Report Issue to prefill these. --- VSCode Version: 1.47.3 OS Version: Darwin x64 18.7.0 Steps to Reproduce: completeFunctionCall new tyepscript file, ype cursor at <kbdTAB</kbdExpected: Actual: Does this issue occur when all extensions are disabled?: Yes/No"}
{"_id":"q-en-vscode-a75ff7981614290a9907f6848e24cf828d9d6ba1bb97aa761a63a7469215cd0a","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Also please test using the latest insiders build to make sure your issue has not already been fixed: --So, this issue is here for a long time in vscode, i've just noticed another issue with it. Basically, vsCode shows incorrect icons, see on the screensot: ! You see that it shows react icon for babel js and simple js icon for react js. I actually know that it may be better to use .jsx files for react project, but is there any way to fix this? and also Is there a way to automaticaly recognize whether the file is for react or simple javascript(by detecting import/require export/module.exports statements or smth else)?? This is the main question for the issue. <!-- Use Help Report Issue to prefill these. --- VSCode Version: 1.48.1(Latest) OS Version: Windows 10 2004 Steps to Reproduce: on the language selector in right bottom corner. in Javascript. <!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes, and it also shows more incorrect variant. !\nMaybe it is the issue with the icon pack extensions that I use, so if needed i'll provide my extensions list.\nI don't see this in 1.48.1 ! Can you try without custom extensions? (close all windows and run code --disable-extensions`)\ni've already mentioned in my post that i tried without extensions and provided a screenshot. It shows react icon for both js and react js. UPDATE: to be honest, I've just tried again with all windows closed and nothing changed. same result\nI think you have the following setting configured. Is that correct?\nYep, you are right. Should i remove that setting entirely or change it? Also the main question was: can VSCode automatically detect whether js file is for react or vanilla js? UPDATE: I removed that setting and now it shows js icon for reactjs ! However, if I select react js the file icon becomes correct) !\nIt's a bug in the language selection dialog. For getting the langauge icon, we use the icon of the first file extension. In your case that file extension has be reassigned to a different language. I create a PR.\nGreat, good luck and notify me if some actions from my side will be needed)\nI pushed a fix. You can already try it out in our insiders build or in the next stable release.\nI am not sure this is fixed, what are the steps? I configured my settings as outlined. I see this: !\nThe screenshot looks correct. In the language selector, the entry has the correct JS icon. Verification steps: Seti theme add to settings:open a js file, verify it has the react icon open the language mode quick pick, filter for verify that JavaScript shows the JS icon and the JavaScript React the react icon\ngreat, i'll check it in nearest stable update)\nOk, but the main reason I started this issue is if vscode can automatically detect reactjs files and set the propper icon. So please, tell me if I need to start a new issue or if there is a solution for this \"problem\"\nYes, a new issue is needed. That said, it's a difficult request as language detection has to be very efficient as there can be hundreds of files visible in a view. Analyzing the content and even parsing it or resolving imports would be very expensive. Fee free to file the issue anyways, but I don't think we can do it."}
{"_id":"q-en-vscode-0da08f55b732405ac7e8871c511ed9d7c41584aee7bb86f8539b5c615da58f68","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --Since you can have multiple s in a project, it would be great if vscode would be able to detect them and apply the right Language Mode (Groovy). The feature is essentially the same how vscode handles s."}
{"_id":"q-en-vscode-a8d74787efa2d6cc9fc5a90c8e4b86bc537481820f7d7b32b3e0509a909e2191","text":"Issue Type: <bBug</bsomething like Gentoo Linux VSCode in terminal get something like Possible fix: Don't hardcode the path to these :^) VS Code version: Code 1.49.1 (, 2020-09-16T23:23:33.049Z) OS version: Linux x64 5.8.8-gentoo <details<summarySystem Info</summary </details<details<summaryExtensions (6)</summaryExtensionVersion ------ gitlens10.2.2 cpptools1.0.0 vsliveshare1.0.2740 LiveServer5.6.1 vscodeintellicode1.2.10 vscode-icons11.0.0 (1 theme extensions excluded) </details<!-- generated by issue reporter --"}
{"_id":"q-en-vscode-50499b690972edff261751eb71f3097b2e4431ed77b8cb491ad3f9cfc3ae5a65","text":"When executing search from extension api 2 checkboxes are not exposed: Preserve Case Use Exclude Settings and Ignore Files !\nHey , Can you please explain a little more about the issue?\nIt's possible to trigger a search from extension by using this api: And it has a good amount of options: But 2 of them are missing:\nThank got the idea. I will raise a PR to fix the issue ASAP :)\nI have raised the . If you get time please review :)\ngenuinely didn't notice the existing PR, apologies\nThanks :). No problem\nIt great now that it's possible to control it but why does it called ? Would it possible to have the same name as in label:\nUpdated to as per suggestion and for consistency with search editor post . Also these to the keybindings intellisense registry."}
{"_id":"q-en-vscode-42b167f28eb8e223a93d1679f564f583d034f6d7676953136408ad30b2f028fa","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Also please test using the latest insiders build to make sure your issue has not already been fixed: --<!-- Use Help Report Issue to prefill these. --Version: 1.50.0-insider Commit: Date: 2020-09-18T05:28:58.063Z Browser: Mozilla/5.0 (Linux; Android 10; Surface Duo) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/77.0.3865.116 Mobile Safari/537.36 EdgA/45.09.62.5078 Steps to Reproduce: Codespaces in a browser on Android (for me, I used Surface Duo) in the editor and hit key In the editor pane you expect it to delete the last character Instead, it does nothing. Here the input: ! NOTE: This appears to be a Monaco bug as the samples on the Monaco website have the same problem. In the terminal you expect it to delete the last character on the prompt I typed \"asdf\" and then backspace rather than doing nothing or deleting anything... It adds an extra \"asd\" to my prompt (so I now have \"asdfasd\"). Then I hit backspace again... It adds \"as\" to my prompt (so I now have \"asdfasdas\")... Backspace again adds an \"a\" (so now \"asdfasdasa\".... Backspace again and this time it does nothing.... Backspace one more time and it actually starts working. Here's the input next to codespaces: ! I've seen very similar behavior in Azure Cloud Shell in the Azure Mobile App so I have a feeling the terminal bug is in Feel free to reach out if you want me to give something a try! <!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes/No\nThis is probably because you're using an on screen keyboard. It's probably a tricky problem for the terminal which is why hasn't been solved yet.\nI believe it's tricky :/ one clarification, I also tried using a bluetooth keyboard and got the same result.\nForgot that we already had an old issue in Monaco-editor and\nUPDATE: I switched to GBoard from SwiftKey to see if that has any change (SwiftKey is the default installed keyboard because Microsoft :) ) and GBoard actually WORKS... both in the terminal and in the editor... The only thing that didn't work in my minimal testing was backspace on an empty line did not delete the line.\nWow that's great to hear, that there may be a temporary working solution with alternative keyboards. However I wanted to add that for me, connecting a physical keyboard via USB, with an adapter, was working perfectly (since the virtual keyboard doesn't appear and, I guess, is not used at all).\nThe editor should now work fine via\nnice! When might I be able to give this a spin?\nWith the next insiders or right now if you can run from master on your device. I hope things will work with a real device, I have been using an Android emulator to implement this...\nI no longer have this phone to try this out :(\nThis could have also improved with"}
{"_id":"q-en-vscode-c41b65cc01c8282b2e68a711645b4bd10026b7124072da9470f861b2c96b1277","text":"1) install the Remote-SSH and the Codespaces extension 2) Open the remote explorer 3) Select SSH Targets 3) Grab the SSH Targets view and drag it onto the activity bar -the drop down still shows both entries but without any view -the dropped view only shows SSH Targets (as expected) and so there is no way to get to Codespaces anymore 4) Use the SSH Targets view and move it back into the Remote Explorer -two views, one of them a zombie\nLooks like somehow the remote explorer got enable for moving views again..."}
{"_id":"q-en-vscode-e46fc494f3ae10cb4ac21ff92415e41ee112b1407107d4bd3a05846632a8f81d","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --The UI allows us to click-set a breakpoint and then do \"Edit Breakpoint\" and set Expression, Hit Count or Log Message at any point before or during a debug session. However, not all debug adapters support these features. As soon as the debug adapter runs and capabilities are received on initialization, it should be clear if these will actually be supported. Before the debug session starts, I can hover over the breakpoint and see its details. ! Once debug session starts, breakpoint gets an exclamation point in the middle of it. At first, I thought that meant that the breakpoint was special - i.e. had conditions, but now I am realizing that it means the breakpoint has a problem. If I stop on entry and hover over the breakpoint or if I hover over it while the program is running, I do see \"Breakpoints of this type are not supported by the debugger\". I can no longer see all of the conditions on hover that I could see before the session started. The error message does not make it clear which of the features are not supported. That would be quite helpful as the debuggers might support different subsets. ! If the debug session hits the breakpoint, then none of the hovers are available anymore for any breakpoint, hit or not. At first I thought this was the only available behavior (see first version of this bug), not realizing that I could get more info if hovering in other conditions. It would be great if the same available information was displayed no matter when you hover over a breakpoint. ! It would also be great if the unsupported options could be disabled in the UI as unavailable while the debug adapter is running.\nWhat you are saying makes sense, however before a user starts debugging there is no debug adapter so we can not disalow setting log points function breakpoint etc even if they will not be supported - we do not know. Yes we could disable some functionality while the debug adapter is running. For that I have decided to leave this issue open and assing to the backlog so we potentially tackle this in the future.\nTo clarify my original request: I am not asking for any modification before the DA runs. I understand that it is impossible and tried to make that clear in the first paragraph of my original post. I do think that it is important to make this clear after the DA runs. Disabling functionality for new breakpoints is the the last suggestion I made and would be a cherry on top. Please do take a look at my thoughts in the middle of my original report. Depending on where you are in the program, you might or might not get a hover error message \"Breakpoints of this type are not supported by the debugger\", but even when you do (not always), it doesn't list any of the details. Is it the hit count that is not supported? The expression? Both? There is no way to tell.\nAgree. We should improve this.\nI will fix this I don't want to do this, since the fact that one debug type is running does not mean that the user doesn't want other debuggers to be run as well."}
{"_id":"q-en-vscode-e173031d6d2511b04627bd20248cbfd243456b19f28fb40745c5d0dcc4fe7935","text":"Connect a mac-windows SSH remote Open terminal See this popup ! So fails here because it checks \".PATH\" on the environment when this is actually \"Path\" on windows. has special behavior where it resolves property accesses in a case-insensitive way on Windows. We changed many accesses from using to a constructed environment object, and need the same behavior there.\nis this a mac-specific issue or are there steps to repro/verify on Windows?\nCan't be verified on windows. Thanks"}
{"_id":"q-en-vscode-5c979b6cc339bc59d83c1d85bdd3307a3d803319863ccee3a3e60194e28d8603","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Also please test using the latest insiders build to make sure your issue has not already been fixed: --<!-- Use Help Report Issue to prefill these. --- VSCode Version: 1.50.1 OS Version: Steps to Reproduce: vscode/extensions/theme-tomorrow-night- that wrong value for themeLabel is given (\"Quiet Light\" instead of \"Tomorrow Night Blue\") <!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes Incorrect value \"Quiet Light\" commited in"}
{"_id":"q-en-vscode-22b1fe8b6fa694f0ef8b6c24f7aadc32d5dde64811c0198968f905753a6954e6","text":"Testing Unfiltered notebook noob feedback: ! I was expecting a tooltip to tell me what it meant.\nAgreed that we should add tooltip for it.\nhow would you interpret it if there was a tooltip that said \"Execution Order\"?\nwhen you run all cells it goes in that order? I was expecting that to be sequential from the top?\nIt's the order that it was executed it, not the order it will execute in. Hopefully a user would figure that out if they run the cell and see it change.\noh, more explicit would be better then imo I also question why this is useful but I'm guessing this is a common notebooks feature elsewhere?\nYeah it's what people expect from Jupyter and can help you understand the state of the kernel\nThanks that needs to be localized though. btw if you are going to merge your change in, you can just commit directly, no PR needed"}
{"_id":"q-en-vscode-8d69aac923a3c201ea7ba81b36302d1ba5e45bd11c3e212febe4802e9d887fc5","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Also please test using the latest insiders build to make sure your issue has not already been fixed: --FYI Related to ? Steps to Reproduce: Set up a keybinding for with any Trigger the keybinding, search editor opens but none of the are used Same keybinding works in Stable Demo using a workspace file: Same result: (not involving OneDrive) ! <!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes Version: 1.53.0-insider (user setup) Commit: Date: 2021-01-29T21:49:22.151Z Electron: 11.2.1 Chrome: 87.0.4280.141 12.18.3 V8: 8.7.220.31-electron.0 OS: Windows_NT x64 10.0. Log Window) with enabled: [2021-01-31 19:44:09.185] [renderer1] [trace] CommandService#executeCommand [2021-01-31 19:44:15.502] [renderer1] [trace] KeybindingService#dispatch Ctrl+Shift+G [2021-01-31 19:44:15.502] [renderer1] [trace] CommandService#executeCommand\nThanks, I can reproduce.\nTo verify: Add a keybinding like: Execute it, expect to see the context line input properly filled out.\nWorks for me (W10). Used this keybinding just to test more args (see below for a comment in the args): I suppose an issue should be filed on not working if it was intended to - it does properly insert the full path, ala but fails silently and button does nothing.\nthanks, as far as I'm aware directly putting that absolute path into the entry would fail as well, correct? So not a new issue.\nYes, it fails inserting it directly into the arg (escaped properly) and in the view in the Panel input - although in Search view it does show the error highlight around the query box and hover error message. I will not file an issue on this."}
{"_id":"q-en-vscode-7968e33d67cf90c3b0a845d794b37065dc83f9fff6d95e41d5b262899d1d337b","text":"There is no drop hover feedback in the Explorer, for example, when in Light+ theme.\nBroken by:"}
{"_id":"q-en-vscode-2a830dc48e0e433e37dce5fc523ee8480d2dd047f3fd8031abd22578cdd0f8e9","text":"I've seen this happen a few times now, and this last time it seemed to happen right after I closed a VS Code window. After that, I got the following notification in all my other open VS Code windows. ! Here is the Shared Process output\nGoing to look into whether we can wait for the process input to stop before actually killing the process for windows as another workaround for This process can take a bit of time now if needed since it's off the window being shutdown which has limited time to react.\nIt looks like we are using the same method as before where we indicate to dispose immediate. Perhaps it's different before because the window going down would more aggressively shut down its child processes?\nwe can consider bringing this in for the recovery build if we start seeing reports.\nNo, a few seconds after I open a second VSCode instance, this error message appears in both instances. It also happens if I open a second terminal in one instance\nan extension may be closing the terminal, or it might be getting relaunched if an extension changes the environment (like the git extension will do when opening a new workspace). Does it happen if you open a new window and then wait 10 seconds before opening a terminal, such that extensions can activate?\nI had the problem even with uninstalled extensions. But the good old \"did you try turning it off an on again\" helped. After a windows reboot, everything works fine\nTo clarify the scope of this fix since is still seeing lingering issues in insiders for this that we aren't able to nail down. Verified here means that the mechanism to delay killing the process on Windows for up to 5 seconds works as expected.\nDon't know if this helps, but I've noticed that it consistently happens if I open a terminal in 2nd level (is that the vscode term for running vscode \"inside\" vscode?), after closing that 2nd level all terminals are stuck.\nVSCode 1.55.0, the issue is still there. Single instance of VSCode looks fine, but starting another instance fires the warning almost immediately, breaking terminals in both instances.\nIt did not happened with git console but happens a lot with powershell but git has this other problem .\nI've been getting this recently, though consistently throughout my day, while doing remote ssh sessions through the vscode terminal window. I open a new terminal and then use wsl2, and within wsl2 doing the ssh sessions. I experience this in both of the following VSCode versions: Version: 1.55.1 (user setup) Commit: Date: 2021-04-07T15:06:02.360Z Electron: 11.3.0 Chrome: 87.0.4280.141 12.18.3 V8: 8.7.220.31-electron.0 OS: WindowsNT x64 10.0. Version: 1.56.0-insider (user setup) Commit: Date: 2021-04-09T09:50:06.816Z Electron: 11.4.2 Chrome: 87.0.4280.141 12.18.3 V8: 8.7.220.31-electron.0 OS: WindowsNT x64 10.0.\nfor me, even single instanse of vscode show this message, if i want to restart \"watch\" task\nUpdate:\nStarted getting this error in VSCode 1.55.2 after upgrading to Chrome OS v90; appears to cause the whole VM to seize. Persists after full powerwash & fresh VM/Code install, no exts enabled. :( UPDATE: FWIW, I have not experienced the issue on Chrome OS Version 90.0.4430.100 since updating to Code v1.56.0\nI'm getting this error all the time since some months. If I start an npm script in VS Code this error occurs. So I had to switch to the git bash terminal (Windows). Hope this gets fixed... Still happens with 1.56.0\nI am receiving this same error myself. I only have a few extensions installed and I've been extremely stable until I installed the PowerShell Extension. At that point I immediately receive the error that \"The connection to the terminal's pty host process is unresponsive\", as well as two others: \"the language service could not be started\", and \"Loading the PowerShell extension is taking longer than expected. If you're using privilege enforcement software, this can affect start up performance\" I'm pretty new to using VS and programming in general so I'm not sure what to do about this.\nthis should be fixed in the build as of today, coming to stable in a month or so.\nHey forgive me if this is a dumb question but the Insiders build is kind of like a \"beta\" build? So there's the potential for it to have issues in other areas? Just weighing the pros and cons of getting the extension working vs. potentially having issues in other areas.\nyes it is, you may get some instability but also access to new fixes/features. But this issue is fixed (finally), it just won't be available in stable until early next month.\nsounds like you have the accessibility mode on, click the \"screen reader optimized\" button in the status bar.\nI get this all the time even in the Insiders build and with no extensions. It started for me right after updating to macOS Big Sur v11.4\nthis is the only know case where this can happen on macOS\nI'm getting this issue all the time at Windows10 (so I guess it's not just related to macOS) I usually keep open plenty of VS code projects (usually 2-5 at the same time). TBH it just got so unbearable, I launch external terminals manually whenever it's possible. (So basically every time, except when I'm using a debugger) My system specs are Any feedback/fix suggestions are greatly appreciated\nthe new release which fixes this should be out next week."}
{"_id":"q-en-vscode-93c5e4df7e774ff1c494b588c78cae0035430b1d1d8fa6f45e21830139afb07c","text":"We should filter out the terminal if this is undefined, otherwise bad bugs like could happen where we restore a terminal into the wrong window: The chances of this happening are pretty slim so not doing this in Feb.\nFalling back to 0 as it has been seems to work well, if we need to handle this on reconnect we will have a persistentProcessId of 0 which isn't valid anyway."}
{"_id":"q-en-vscode-9b7a71ab2b67098caecc4d9a8d524d77026f21c1a5c4c0e12266c018146b1a01","text":"Tracks bundling TS 4.2.3 with VS Code. This includes a\nShould be in recovery and in main now"}
{"_id":"q-en-vscode-e3f7456f5b3e4dd4e435c41af2797f904d83a8d9d696db8bfd7f5dc251f2456e","text":"Depends on This is common for process explorers on Windows at least, might also make sense to bring this to Linux.\nThis should be doable now that has landed via"}
{"_id":"q-en-vscode-effa1c67cce3d01c489f2b3b2f501ba58a180dbab6f50347315d9aec4056f834","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --I'm always using Dark+ Theme at VSCode (regardless of macOS light/dark appearance), but when set to , title bar style is mismatching with VSCode theme (if system-wide appearance is light). looks like this: <img width=\"1045\" alt=\"image\" src=\"https://user-\"Electron has API and API that can overrides system appearance per app/window, so I injected to (yeah, its unsupported, i know) and looks pretty great. <img width=\"1045\" alt=\"image\" src=\"https://user-\"However, I think code injection is not supported, possibly corrupting the installed VS Code, and disabling it with every VS Code update. so I'll suggest new settings that overrides macOS appearance based on VSCode theme. Note: I think this new settings may conflicts with and .\nThis is pretty cool, I was not aware we can change the title style to dark style. is this API robust to use? Do you see issues with it? I would be willing to look at + test a PR in this area if the change is small. But I wonder if a setting would be needed as well?\nHm, one issue is that theme settings can change per workspace but from the main side we have no way of reading the workspace settings...\nI tested this and it seems to work really well on macOS at least. Still, I wonder if this needs to be a separate setting that the user can control it and wether we should not do this automatically. do you have thoughts on this? I know you have theming related code and settings that automatically switch the theme in VSCode between dark and light depending on the OS setting. Here we would introduce a setting for the user to \"hard-code\" the appearance of the application to be either dark or light. I would think this setting can only be application scoped and not per workspace though because we need to resolve this from the main side. Demo: !\nIt looks good! But it means we no longer get events when the OS change schemes, right? BTW when the themeservice sets a new theme, the main process gets an event:\nI think the idea is that you must use whenever you are making a decision about which theme is being used. Not sure we do that already. explains the impact quite well. Yeah but that is probably too late for a smooth startup. Imagine a workspace configures the theme to be light but the default setting is dark, we would probably still set dark first only to update to light then.\nIf the setting is true then and, to get information about the OS scheme. But from reading the spec and what also mentioned, I believe once we set , will become and remain true and we no longer get to know what the OS scheme is and when it changes. Can you verify if that's like that? That would be a problem as we could no longer support . It's the same as for the initial window background. The ThemeMainService listens to theme changes and remembers the background of the last used theme. Yes, that can result in flickering when themes are configured in the workspace settings.\nHm, I wonder if we should actually deprecate and have a new setting with these choices: : check OS theme : pick a color scheme that matches VSCode theme (dark or light) : hardcode : hardcode This would solve the issue of conflicting settings.\nIs that new setting just for the appearance of the window? How can it replace the old setting(s)? !\nYeah you are right, both are probably too different to merge into one setting because \"old\" setting is for changing the theme in VSCode based on OS configuration and the \"new\" setting is for changing the OS configuration (for the window) based on the VSCode theme.\nThinking more about this, this should work: If is true, have . The window appearance as well as the workbench color theme will be in sync as the workbench color theme follows the OS appearance. If is false, observe the current color theme and set or so the window appearance matches the color theme This could be done in the themeMainService. If needed there can also be an extra setting to control whether changing the window appearance based on color theme is wanted. What needs investigation is how this affects the detection of high contrast\nI am worried that this is a somewhat breaking change for everyone that just runs with default settings and might be confusing for users. If macOS is running in light mode (default), it is probably weird that VSCode suddenly appears dark, both in the window title as well as context menus. So I think either a new setting or an update to the existing setting is still needed.\nWith the new setting you can now explicitly opt into mode even when the OS is set to . You can give our preview releases a try from:"}
{"_id":"q-en-vscode-52f4ec18455f068b8bff5079e472f4c48c408941e0be6aac6f20e098c3c91b14","text":"It doesn't look right compared to all the other placeholders: ! It's more noticeable when using editor font (font gets smaller): VS Code version: Code - Insiders 1.56.0-insider (, 2021-04-16T11:25:09.662Z) OS version: Windows_NT x64 10.0. <!-- generated by issue reporter --"}
{"_id":"q-en-vscode-605beb61f629a1600c1e4b10f898a0bc64bd9ef6d1f43b529841864d931cc20a","text":"See I think the best way for this would be for VS Code core to rename the terminal if gets a new request with a different name -- opened an issue for that Or, at least, try to use a terminal of the same name if one exists (which would be easier and probably more feasible since the is not writable.) Code pointer:\nAlso fyi\nI'm going to take a swing at this. Here is my current understanding of the problem and I would appreciate any feedback or comments. The issue - Launching debug sessions in integrated terminal will open duplicate terminals. Solution - Code can attempt to find a terminal instance that has the same title as the launch request and is not currently in use. If located, it can reuse that instance, otherwise it can use the existing logic to create a new instance. I believe this should also address the concerns in as Code will stop spawning new terminal instances for repeat debug sessions. Please let me know if I have any misunderstandings on the issue.\nSounds good to me\nI'm looking over the current code for checking out an integrated terminal and came across the minUseDelay for determining which of the terminals is available for use. It was a fix for . As I understand it, when compound node tasks were run, they would all try to grab the same terminal instance so only the first one to run would actually get the terminal, the others would fail to run. By introducing the minUseDelay, it insured that all of the launch tasks would get a new terminal instance. Once microsoft/vscode-js-debug is merged, each Node launch task will most likely have a unique name and therefore a dedicated terminal instance (ideally) which would mean we could always return the first terminal without a child process that has the same title and get rid of minUseDelay. Right?\nObviously if someone had two launch configurations with the same name (why?) this would fall apart.\nWe could even enforce name-matching and reopen the terminal if nothing's running in it and the name mismatches"}
{"_id":"q-en-vscode-a8b8940820b17900ffb990553f8531a61b6e3361e17cee73d502a4dc81a49288","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.56.2 OS Version: macos big sur I am unable to find any record of someone else having this problem! Can someone point me to how I can start to debug this issue? Steps to Reproduce: a file with some text in it: using a relative path, the file opened is blank: opened using an absolute path, the text is correctly rendered: One hint is that when i edit the file (after opening with a relative path), vscode displays this error: ! Presumably, the code CLI is not using the cwd to open the relative path, and is assuming is a file in the filesystem root. I know this is super weird and am very eager to debug! Has anyone seen anything like this before? I have already tried disabling extensions, looking into logfiles etc.\n/duplicate\ndo you by chance have a environment defined?\nAha! I do! How did that get set?! I've an unset to my shell profile but wondering how it is that this variable is being set in the first place.\nI would like to understand how this variable can end up in your terminal environment. We do set it from application code here: But this cannot end up in the launching terminal imho... If you have any clues how you got there, please let me know. Also: I assume you are not using the integrated terminal right? //cc for terminal experts, maybe you have an idea how something we write from JavaScript can end up in the terminal environment.\nThe terminal environment code is a little complex, but except for cases like being set for extension terminals or getting contributed via an extension, this variable is meant to be sanitized in the terminal's process:\nyeah I was also thinking that any prefixed environment variable would not make it to the integrated terminal, but would be good to ensure we have not regressed. However, here I actually think, people use the external (OS) terminal and still see set, which I cannot explain how that can happen from a process launched by the terminal.\nCan confirm! This issue concerned using an external terminal to launch a vscode window!\nis it at all reproducible for you somehow?\nI see we only sanitize the external terminal env on Windows.\nI think so! If I remove the unset from my zsh profile then the issue arises again, so it's reproducible in that respect. Is that what you mean?\nhow do you open terminals for using them? via a command VSCode provides or via Finder/Dock?\nUsually the Finder/Dock variety (specifically, entering the letter \"t\" into spotlight and hitting enter without looking).\nI cannot explain this issue then. If you are always opening terminals from the dock, none of the environment variables that VSCode defines should be there. Here is what I think can happen (correct me if you do something different): you open VSCode (either dock or terminal, doesn't matter) at one point you open a external terminal from VSCode (through some command or action) you type in that new terminal it fails to recognise the current working directory if now you open a terminal using dock or finder and try the same () it works\nThat all sounds right (I do sometimes open terminals from vscode!), except the issue arises even when the terminal is opened externally (via FInder). My immediate issue has been patched (via unset in zsh profile) and I'm alright to drop the concern if this is likely to be something extremely specific to my setup!\nI cannot explain that. A terminal that is opened via finder can only contain environment variables as configured by startup scripts ( et al) to my knowledge. But happy to learn how this can happen otherwise...\nOk cool! I'm running some grep -r to see if anyone else might be sneakily setting this variable. Thanks for all your help!\nI started seeing this issue after upgrading to MacOS big sur. Even when vscode is not running, the $VSCODE_ env variables are set when I open a new terminal from the dock\nI think I figured out the cause of this for me at least. I have zsh set up to auto load tmux. so when a integrated terminal in vscode is launched it creates a tmux session and the env from that somehow leaks into other tmux sessions. I just disabled tmux auto loading and the problem disappeared.\nDo you by any chance use tmux inside the integrated terminal in vscode ?\nI do not!\nmaybe this is a red herring, in my testing when I run the command to open an external terminal, I do not see any of VSCodes environment variables appearing. Though it is true that we do not call the method in: Meanwhile I pushed to make sure we never on macOS/Linux, so this issue should not be present anymore.\nare the verification steps in the original issue still valid/could you give some?\nYes the verification steps in the original issue are still valid one second. writing better instructions\nto verify, close vscode open an external terminal and modify an environment variable in a repo directory starting with for example. open vscode and launch an external terminal from that same repo confirm that the environment variable you set is not present"}
{"_id":"q-en-vscode-a569df0bda1ea0cdb26a1f2cb21c48db41ef7d10dfe234e88195493b056cc8e2","text":"Steps to Reproduce: Workspace and trust it code-insiders from a folder that is not trusted. -Shows the dialog code-insiders from a folder that is not trusted. -No warning dialog the window code-insiders from a folder that is not trusted. Expected: No warning dialog should be shown Actual: A warning dialog is being shown\nPushed a fix for this to main. Fix is low risk so we can take it as a candidate."}
{"_id":"q-en-vscode-9a73d7280fe4ccac23ca7908c76c014ce25480dbe1592d7528e20bdca8a1aff6","text":"I recently came across CSS rule. According to the : I do not think we want the panning and zooming, but the removal of the 300ms delay sounds great to get a more snappy UI when using touch for e.g. action bars. Maybe worth a try?\nI wonder if would even be the right thing, but not sure it removes the 300ms delay:\nInteresting, I see we use this property already, e.g. in the list:\nWorth experimenting with. do you remember why you that css property for the list? For now assigning to June, might push out due to lack of time.\nDid not have time, pushing out to August, since in July I am on vacation\nProbably for the 300ms removal.\nAdding bug, so we get this verified. Steps: ipad. sure the touch behaves as expected in the editor and some views touch triggers context menu\nfeels pretty snappy to me!"}
{"_id":"q-en-vscode-672bf5634a40e1e4cdebd2085e2ad8ffabc76f6f1fcbc4fb5ba79dee65c695f9","text":"Issue Type: <bBug</bI currently store some arbitrary JSON-serializable object using and updating keys. At the top level, my configuration variable is an array, and each element is a dictionary that contains some keys. One of those keys map to a value like this: This sets fine, and I can look in my workspace's to see all keys and values there as expected. However, when I try to get the value from the workspace configuration again, the empty string key and its value are missing, meaning I unable to retrieve the array with in it. So it seems VSCode filters out keys which are blank strings when reading workspace configurations, although it will set/save them. This is currently blocking a project I'm working on and should be a trivial fix. VS Code version: Code 1.57.1 (, 2021-06-17T13:28:07.755Z) OS version: Windows_NT x64 10.0. Restricted Mode: No <!-- generated by issue reporter --\nI am sorry that I do not understand the usage of the API here and what is being done. Can you please provide a sample code snippet to reproduce the issue?\nHere's a round trip example of me saving an object to the workspace configuration, and retrieving it. It shows all data is stored appropriately, but data is lost when reading it back if a blank string key exists. The message will be shown. If you comment out/remove the key-value pair in , it will instead say . So VSCode is losing data when obtaining data from the workspace configuration if you store an object which has a blank key somewhere (at any level of depth).\nThere was a fairly quick initial response here, but since then a good amount of time has elapsed. Has there been any progress updates or planning related to this issue? Can you confirm you now understand the issue given the above proof-of-concept? It should be very clear.\nThanks for the info. It is in my list of issues to investigate, I will try it out and get back to you."}
{"_id":"q-en-vscode-46bd5baad834beb477f219a49c8dd3b3512817ce20dec9c3ea997264c1fb6f3f","text":"And we could use in browser.\nSorry, just confirm, Would this will be released in July (the next) version?"}
{"_id":"q-en-vscode-ba06946dd93cccbeb23d4491506edf110892444687c08b0f17bb32c7bddbf712","text":"Since it seems pretty simple to stop the server now, It seems like using the task as a way to run a persistent server is sort of unnecessary. Settings should have a way to toggle whether they want the task to run or not.\nVerification steps: this setting (or set to ) ! the command palette or other means to open an external preview, when doing this, the server logging task should not run automatically. setting to , the task should automatically run with the external preview launch (default behavior)."}
{"_id":"q-en-vscode-0828f566422d8e8c18cbc730616877952fc7c3fc18227b8bc60b9f7252fe0e41","text":"Issue Type: <bBug</bThe user defined keybiding editorScroll command with arg property \"' to': 'up'\" repeats slower than \"'to':'down'\". The key combination doesn't seems to have influence. [edit] I tested it further, and found what triggers this behavior is the conjunction of this keybinding with the setting smooth scrolling : \"editor.smoothScrolling\": true, Windows and Linux versions have the same behavior. VS Code version: Code 1.58.2 (, 2021-07-14T22:30:16.440Z) OS version: Linux x64 5.4.0-77-generic Restricted Mode: No <details<summarySystem Info</summary </details<details<summaryExtensions (5)</summaryExtensionVersion ------ Bookmarks13.1.0 vscode-language-babel0.0.31 cpptools1.5.1 LiveServer5.6.1 ninja0.0.1 </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\n(Experimental duplicate detection) Thanks for submitting this issue. Please also check if it is already covered by an existing one, like: <!-- score: 0.458 --<!-- potentialduplicatescomment --\nI could not reproduce. I tried with the following keybindings: The scroll speed appears similar to me for both keybindings.\nI tested it further, and found what triggers this behavior is the conjunction of this keybinding with the setting smooth scrolling : Windows and Linux versions have the same behavior.\nI can reproduce a slowdown. On my macOS computer, scrolling down went 114 lines down in 10s, while scrolling up went 100 lines up in 10s.\nNot sure if my case is special. My issue was closed as duplicate of this one. Windows 10 here. On two completely different and unrelated machines but with the same VSCode user configuration. Scrolling down a document of 114 lines: 4.5 seconds Scrolling up again: 13 seconds This is a major slowdown by the factor 3!"}
{"_id":"q-en-vscode-1e6496627fbe96f6e1662550d0cb454ad68e5d1c513a6b14e1e9d3c86a4a515e","text":"Testing disassembly view on screen reader on an element, notice the following being read out \"Can set breakpoint, Instruction address: 0x000000ea, Instruction: of\" We need to imrpove this aria label: most important piece of information needs to go first, so that blind users can skip the boring repetitive part. So start with the \"0x000000ea address, \" it more concise, I think \"can set breakpoint\" is not needed here do not understand the last part \"intruction: of\", this needs to be fixed I think\nwill fix this.\nIs there a way to update aria label via accessibility provider? Looks like it's set only when the element is rendered? I'm wondering if we can add/remove all texts about breakpoint and stack frame. If not, we should probably remove texts related to breakpoints. Also, it's a great suggestion that we should reduce the amount of duplicates. So I'll change it to . As for , it's likely the disassembly instruction you are looking at is . For now, each line in disassembly view looks like: Where there may not be InstructionBytes. For example, Will have aria label (currently) .\nno, it is not possible to dynamically update the aria label. This sounds good, I would just cut it short to be"}
{"_id":"q-en-vscode-ce69298da8e7d2d89aae8081747e5dce2117e8e0b34961c30643617a8d716124","text":"Testing the steps in the test plan item to create an interactive window I get a notification saying there are extensions on the marketplace that can handle . I feel like this is unnecessary noise\nWhich extension is it recommending?\nThe marketplace query is ext:interactive when I click on Search Marketplace !\nI forgot that we had this but we automatically query show this when there are extensions in the marketplace that match the query . I'm curious about how this works, it seems overly sensitive since these extensions don't use that file extension, seems that it just matches based on keywords? I think the only way to avoid it here is to change the extension that the interactive window uses, which is probably fine.\nDiscussed with and even if we picked an extension that isn't yet in use, we can't prevent extensions from introducing collisions later. What about the following (essentially, parse the notebook URI from the cell URI, and check if it's an untitled://, file:// or vscode-remote:// URI based on the notebook URI instead)?\nWhat will the interactive window be? It won't pass that check?\nIs the motive here is to not show any recommendations for this resource?\nYeah we don't want to show the '.interactive' file recommendation because it's not actually a file on disk. The interactive notebook schema is 'interactive'.\nThen we shall be filtering out in our recommendations logic as you recommended above."}
{"_id":"q-en-vscode-2b729f3b406e8a5e6e56d49143c944703337fff12b5efdb219afabd20037e107","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- Version: 1.59.0-insider Commit: Date: 2021-07-31T00:42:32.346Z Electron: 13.1.7 Chrome: 91.0.4472.124 14.16.0 V8: 9.1.269.36-electron.0 OS: Linux x64 5.13.5 Steps to Reproduce: a project directory that is hooked up with tasks via v2. The file must define one of the tasks as being the default. you have bound to a shortcut (mine is ctrl+shift+b) the key combination to trigger the build picker combobox to choose a task appears instead of the default task being executed as used to happen before.\nI'm not able to reproduce this issue. Do you have a repository I can test with? I tried with: If there is more than one default build task, then we have always shown the picker. If there is only one default build task, then we run it.\nI'm experiencing this with internal company code which cannot be shared. I can share the file. It should work with any c# based project.\nHas the schema for changed at all in the past month?\nI've replicated this on other similar projects, even those that have just one task defined.\nThanks for the additional details and for reporting this issue!"}
{"_id":"q-en-vscode-6101d79b42073c0ae7f05f77bf90df5704efabb2e191b580fd649a60d2770cf2","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --It would be ideal if the Testing user interface would allow to sort/filter based on duration. Use case: I have a large number of tests in my project and they're slow, I want to sort by duration to identify long running tests. !\n(Experimental duplicate detection) Thanks for submitting this issue. Please also check if it is already covered by an existing one, like: <!-- score: 0.595 --- <!-- score: 0.493 --<!-- potentialduplicatescomment --\nI want to work on this shall I give it a try ?\nCan I get a little help here ? I think I am not able to find the UI in the app. Is this screen capture same as above ? (Left panel) ! I made two tests locally, but I am not seeing the time it took to run the tests ?\nYou'll need to use to make the Test Explorer use the native experience.\nHey admin, i have just started my open source contribution journey. Can you assign this task to me? so that i can start working on it.\nFeel free to submit a PR\nHey Conor, can I work on this issue\n, I found that following changes will be needed adding a class to adding to export const enum TestExplorerViewSorting { ByLocation = 'location', ByStatus = 'status', } inside the file\nThe TestExplorerView is where it happens. You can do a \"find all references\" on and that'll give you a good indication of places to update\nThere is such a line for sort ByLocation , but not there for ByStatus , why is it so\nWhat changes should be made to this file , please help me out\nHave a pull request for this:"}
{"_id":"q-en-vscode-98822df831fbf91ab40e9596692cdd70a1ebde22a43043b0d3de6b9ee5d0979e","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes/No unable to start container without extensions enabled <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: Version: 1.60.0-insider (user setup) Commit: Date: 2021-08-16T05:21:19.431Z Electron: 13.1.8 Chrome: 91.0.4472.164 14.16.0 V8: 9.1.269.39-electron.0 OS: WindowsNT x64 10.0. OS Version: Steps to Reproduce: VSCode Insiders Edition on Windows Insider gives error every time when logging in to GitHub and other sites 'writing login information to the keychain failed with error 'not enough memory resources are available to process this command'. Also, have set up ssh, but still requires web authentication for extensions every time I open VSCode. It gives the error, but then continues to login after the error and does not save login information. after opening VSCode insiders login to github error and does not save login information tested on current VSCode and doesn't require login every time and no issues with memory (working) Version: 1.59.0 (system setup) Commit: Date: 2021-08-04T23:13:12.822Z Electron: 13.1.7 Chrome: 91.0.4472.124 14.16.0 V8: 9.1.269.36-electron.0 OS: WindowsNT x64 10.0. Regular Edition does not give the error and I do not have to login to github everytime the app starts.\n(Experimental duplicate detection) Thanks for submitting this issue. Please also check if it is already covered by an existing one, like: <!-- score: 0.528 --- <!-- score: 0.528 --- <!-- score: 0.452 --<!-- potentialduplicatescomment --\n! did not fix the issue Snapshot of the error message: !\n!\nthere was an issue with the last insiders build. Can you update to the latest one and let us know if you're still seeing this?\nThank you. I installed the update, but unfortunately, it is still doing it. I did a full memory diagnostic and it did not find any issues with memory, and there is plenty of it available. Version: 1.60.0-insider (user setup) Commit: Date: 2021-08-19T05:15:46.693Z Electron: 13.1.8 Chrome: 91.0.4472.164 14.16.0 V8: 9.1.269.39-electron.0 OS: Windows_NT x64 10.0. ! !\n!\nJust updated again. Issue remains. Thank you. Version: 1.60.0-insider (user setup) Commit: Date: 2021-08-20T17:51:50.560Z Electron: 13.1.8 Chrome: 91.0.4472.164 14.16.0 V8: 9.1.269.39-electron.0 OS: Windows_NT x64 10.0. !\nThis is likely an issue with the library we use for storing secrets called but it's impossible for me to reproduce... so it'll be hard for me to make progress here.\n+1 I am now getting this issue on my laptop Where can I find logs or what information do you need that would assist in resolving the issue?\nHey did you manage to fix this? Same problem..\nThird that Same issue also...\nI am also getting the same issue as well, both trying to sign in with my Microsoft Account for Settings Sync and with GitHub for Settings Sync, GitLens, or LiveShare. Visual Studio 2022 is also having problems connecting to GitHub, and since the new connection successful screen doesn't display the token information, I cannot even connect it manually.\ndo you see this only when using a Remote SSH/Remote Container connection? Your original video that it was a Remote connection. To the others in the thread, do you also experience this when using Remote?\nI see it when using the Accounts -Sign In commands in the bottom left. Whether I use it to sign in to use Settings Sync or LiveShare (with either Microsoft Account or GitHub), or using GitHub for GitLens. I now get the token information with the instructions: Copy the token. Switch back to VS code. Click Signing in to in the status bar. Paste the token and hit enter.5. But, there is no \"Signing in to \" in the status bar, just the error message noted.\nAre you using a Remote? Like Remote SSH, Remote WSL, Remote Containers?\nno, just launching VS Code and attempting to sign in. Seems to be a problem with saving the token on the redirect.\nI solved this by changing my pagefile settings by doing: System Properties -Advanced -Performance -Advanced -Virtual Memory -Change -uncheck 'Automatically manage paging file size for all drives' -check 'System managed size' -OK -Reboot edit: I'm not sure why the pagefile would affect it though. I have 128GB of RAM.\nThank you , that seems to have fixed it for me as well.\nThanks a lot it worked for me\nSadly, this did not work for me and I share the doubt about how a page file setting can affect this. I even tried it with the paging file completely disabled. Any other ideas?\nif you have nodejs installed, you can run this simple one-liner to see if the problem is how your keychain is setup: or if you're on Insiders: This is using the following open source package: What you should get is something like this: If that succeeds, then the problem is possibly in vscode. If that does complain, then we'll need a fix in keytar itself.\nI can try this tomorrow.\nHi Tyler, I understand that you can't reproduce the problem, making it hard for you to investigate. I'm happy to try to help. I am on Windows 11 if that makes a difference. Also, on a completely separate Windows 10 desktop right next to me I do not have the problem. I ran the test you suggested and I got back the expected response for github authentication. However, my difficulties are with microsoft authentication. So I changed the command to look for . I got back 100 account/password pairs. 98 of them had in the account field, for example: (Some digits changed in case they are sensitive) The account fields for the other two were and The password fields look quite different from the one retrurned by github. Something like: Probably the most important thing is that only the password returned with the account looks exactly like the one returned by github, sort of like SSH keys. The formats are the same, the digits are just different. So, if I had to guess it looks like the caller of this function is getting back a lot more data than expected and/or needs to filter it to get only the account.\nthanks for giving that a try. The content of is different than GitHub. It has a different secret storage schema (that I could get into ... but lets call it as it is \"legacy reasons\") Anyway, the fact that the script I suggested worked fine, but VS Code fails leads me to believe that this might have something to do with Electron and how is loaded into that. Let me chat with on this to see if he has anything thoughts\nI have a different problem. When I dumped my keychain, I have 123 entries in it so \"not enough resources is probably not surprising. I wonder whether these accumulate over time and then the problem arises. Usually, I am asked to re-authenticate once a couple of weeks so it would take a while before they accumulate. I will remove them now so I can use my shit.\nFor my normal VSCode (which I stopped using since we last tacked this problem), i have the following targets: So probably this is how it looks \"in the making\". Should I clean that up as well or leave it for testing?\nFWIW, I've worked around the problem by deleting all the credentials of the form: where \"xxx\" at the end is some random-looking decimal number. After restarting and reauthenticating, several subsequent restarts of VSCode did not require re-authentication and the number of credentials of the above form holds steady.\nYes, me too. I am guessing it will slowly grow as tokens expire. The library that gave does not have functionality to delete, so that is probably what is causing this.\nI found the same thing; in my case it was a whole bunch of credentials placed there by Adobe Creative Cloud that were taking up all the space. Once I completely cleared out the Windows Credential Store, I was able to get both VS Code and Visual Studio to correctly connect to GitHub again. Get Outlook for iOS From: Vladislav Kosev Tuesday, July 12, 2022 10:37:40 AM To: microsoft/vscode Pierre Plourde Comment Re: [microsoft/vscode] error logging in 'writing login information to the keychain failed with error 'not enough memory resources are available to process this command' () Yes, me too. I am guessing it will slowly grow as tokens expire. The library that does not have functionality to delete, so that is probably what is causing this. — Reply to this email directly, view it on GitHub, or unsubscribe You are receiving this because you commented.Message ID:\nThat process for keychain setup worked for me, I got the result expected. So what does that mean for vscode? A bug? Every time I restart VSCode or even reload the window, I have to login again for CoPilot and Settings Sync - I use the Github login. Using MacOS Monterrey.\nI had the same issue - too many credentials (Windows 11: Start Credential Manager Windows Credentials). Mine included an obscene number of vscode entries, as noted above, so I deleted them all with a powershell line I cobbled together through Google: . It still threw the error so I hand-pruned most of the remaining entries and it took. Good luck! (FWIW, Win 11, 8GB RAM + 24GB Page File) !\nI think this one is more likely the cause. Because after I deleted all the credentials in the credential manager, it still automatically logs me out. There is no out of resource error for me. It may be the syncing issue leads to duplicated credential creation.\nFor the folks that have cleared the large list of credentials in their credential manager... can you please check to see if that number is increasing again? My suspicion is that some of you got into a bad state around the time that I was making heavy changes in this space... but now that that has settled, vscode's code isn't polluting the cred store anymore... This is coming from my lack of reproduction of the issue even though I've tried numerous ways to do it. With that said, users could still get into this bad state though, if they use other applications that use the keychain (like someone mentioned with Adobe products). For this case, we should at the very least include docs in the troubleshooting keychain issues here:\nHi Tyler, I have not had any failures along the lines of the original problem (needing to login each time vscode is started) since I cleared the credentials back in July. Today, I see only about 5 hits from the command you posted. And I think I saw about the same number when I checked at some point between July and now. I tend to agree with your suspicion.\n\"Target\" is localized. Need to chcp 437 first. I now clear my credentials and will try to reproduce.\nThanks I'll update my regex\nupdated mine and I'm updating your quote so that no one gets confused\nBased on confirmation, I will optimistically close this issue once the doc PR goes in... however, if you still see the problem of VS Code's credentials growing an unreasonable amount, please don't hesitate to reach out.\nOne other thing I tried is a minimal Electron repro where I created 1000 secrets to see if I could reproduce it... I could not. For those that wanna assist in getting to the true root of the problem, I've documented a potential repro here: If you can reproduce the error mentioned in this issue using my Gist, please let me know so we can let the Electron team know. In any case, I'll still close this once goes in.\nI have confirmed that once there are too many credentials, the Windows call fails with . This issue doesn't have to do with Electron or Node.\nGiven this, the best we can do is document that workaround which is done now. Thanks for the investigation\nHi , may I ask how to list the credential? And what's the upper limit? I don't think my box have \"too many credentials\". But it still hit this issue. Thanks\n, isn't it possible to remove the expired credentials automatically? It doesn't seem normal that you must delete credentials by yourself. There is not even a command that does it, you need excel/PowerShell to do it.\nWe're still blocked by the issue. May anyone please suggest a workaround? Thanks.\nsee to get the list of credentials.\nThanks, . This is my observation: ! I notice tons of credential created by VS code. To me, the bug seems: vs code incorrectly wrote tons of credentials to Windows, which causes the \"not enough memory resources\" issue. This is not a Windows issue but a vscode issue. Any thoughts? Thanks CC (, who seems the PM to me)\nI think that the working assumption is that there was a temporary problem in VS code that generated the entries in the Credential Manager. See the explanation in this in this same issue thread. You might try deleting those entries in your Credential Manager and make sure you have the latest revision of VS Code. I did exactly that back in July and it worked. I see that at least some of your entries are in or before that timeframe.\nThanks. . I ran Then the issue is gone. I disagree with this statement: I have confirmed that once there are too many credentials, the Windows call CredWrite fails with ERRORNOTENOUGHMEMORY. This issue doesn't have to do with Electron or Node. Again, to me, the bug seems: vs code incorrectly wrote tons of credentials to Windows, which causes the \"not enough memory resources\" issue. This is not a Windows issue but a vscode issue.\nIt's a Windows issue because it could be caused by other programs too. See Adding a command to clear credentials sounds like a separate feature request to me. Edit: Actually, if a bunch of credentials are still being incorrectly written even with VS Code version 1.71, that's a VS Code bug.\nHi , are you saying it's by design that VS code creates tons of the credentials? IMO, once vscode creates ton of credentials, it's by design that Windows OS prompts OOM. I read, in different way: both adobe/vscode should not create tons of credentials. It's their bugs to create tons of credentials.\nI cleared my credentials back in July and haven't seen any evidence that \"VS code creates tons of credentials\". Right now, I have 5 that refer to \"\" user IDs and I suspect that has to do with my Microsoft account history. Tyler said that was a long story and has to do with legacy/older systems. I think I tried to explain before that Tyler suspected that there was some bad code in one of the releases of VS Code that caused the creation of these credentials. This would have probably been in the timeframe of the first half of this year. But apparently, that problem is no longer in current builds of VS Code. So you may be right that it is a VS Code bug, but it appears to be one that is fixed, at least according to my observations.\nThanks, . I think the issue still exists. 2 days ago, I have deleted all extra vscode credential using, cmdkey /list ForEach-Object { /delete $_.Matches.Groups[1].Value } But today, login failed again. Then I observed that vscode creates a bunch of credential again, which does not sounds expected to me. ! , do you have any idea? Thanks\nThis is why I asked if anyone was seeing the large quantities of vscode secrets after clearing out their secrets. And why I used the phrase \"optimistically close\" . You've provided new information with that last message and you've confirmed that there's some scenario where this still happens. With that said can you please supply: if you're using a Microsoft account or work or school account (if multiple, tell me about all of them) how many windows of vscode you typically have open give me the full text of one of these keys (don't worry it's only public data)\nThanks if you're using a Microsoft account or work or school account (if multiple, tell me about all of them) -Only hotmail account. etc, how many windows of vscode you typically have open -3 instance. But I use a python script to open them give me the full text of one of these keys (don't worry it's only public data) -Where to find these text?\nHey just jumping on this as per the comment in . I have just performed the credential manager clean out as per the command line in and my multiple VSCode windows are now all signed in, and persist across Exit's and re-open. There were 178 entries for the specific pattern on my machine. Edit: With things now working, there are 22 entries. I will keep an eye on things, and will post back here later on. For initial context: Windows Account 5 windows open at all times; 4 are folders/workspaces, and 1 is general files and unsaved instances\nI've started experiencing the issue that appears to be mentioned in (but has now been merged into this ticket), but it appears that VSCode is correctly(?) reading my credentials from Seahorse (on Ubuntu) but still asks me to sign in every time I want to start or join a Live Share session. I've tried clearing out all the relevant credentials from Seahorse, and that (as expected) asks me to reauthenticate with my Microsoft account, but then continues to ask me to sign in whenever I start a new session. From Output Microsoft Authentication, everything seems to look fine to me - it finds two(?) stored sessions and seems to refresh the token fine... :shrug: Details about VS Code and my OS: In addition: It's a Microsoft work account () Usually only one or two, depending on the project. However, Live Share appears to open one window initially (that I have to click 'sign in' on) then opens a separate window (which I also have to sign in on) From Seahorse, I assume it's this bit you need? If you need to know anything else, don't hesitate to ask :+1:\nJust chiming in here. I just used the hint with deleting all the credentials for vscode. I had like 100+ of them. After that I signed in to settings sync using my Microsoft account. No more memory error on save. And after restarting vscode it signed me in automatically again.\nI have the hardest time reproducing this issue... for those of you that do see it, can you do me a favor and: the Credential Manager open on the side what you normally do (open multiple windows, just use VS Code) to see if you can notice when the number of credentials increases and let me know if there's a particular action that causes it I'll continue to try reproducing this...\nThank for continuing to try. Current it seems to not happen anymore or it happens very infrequently. In credential manager I have: 1 github token 8 tokens - 4 from yesterday and 4 from 30.08.2022. Not sure why 4. 5 numbered from 0 to 3 + one keylist.\nthose \"8 tokens\" are in the form right? Can you share them here? And if you don't want to can you at least tell me what's different about them? My guess is just a guid at the end...\nYes, there they are. Interestingly enough They were 4 pairs Two of the pairs are modified today (so they are somehow active, because the date changed from yesterday). There they are (dates are in front)\nJust a little context on the pairs... Cred Man on windows has a limit to the number of characters you can store so we chunk them. The are the 2nd part of the other one. can you tell me how many things you've logged into with your Microsoft account? Based on 2 items being updated on Nov 4, I am going to guess...2... LiveShare & Settings Sync maybe? The ones from Aug 30 are odd... around that time, I was not working heavily in credentials land so that leads me to believe that you saw the bug that other folks in this thread are mentioning... the slow slow climb in the number of secrets. It's unclear to me why that happened... I ask myself questions: Were these credentials being used before that date and one day they weren't read properly so new credentials were created? This is possible. The library we use can be a bit dicey... and Windows certainly isn't perfect. before the \"sign in every time behavior\". Do you recall having to randomly sign in to Microsoft every once in a while? If this is the case, I need a better clean up mechanism... if possible. Were these credentials created accidentally as a duplicate... like could a 2nd window of VS Code think it was responsible for storing the credential and thus it stored a duplicate secret which was then overwritten by the updating of the keylist. Worth double checking... I thought only one window handles the creation of secrets, but maybe I'm missing something. If this is the case, the keylist might be getting mangled with 2 instances of vscode writing to the same place Btw, all of the non-keylist ones can probably be deleted (I'm assuming the date modified on those are a while ago). Those are from the old structure of secrets where all secrets were stored in one (chunked) credential.\nYes, these two. It's not that this doesn't happen, but it is diffucult to remember so far back.\nHi folks it has been a bit... I wanted to let you all know that I made a change that I think will be helpful here. It's available in the VS Code Insiders Build: What I'm hoping you see is no more gradually increasing number of credentials in CredMan. Please give it a try and let me know how it goes. I know it may take days/weeks/months to fully verify that this change made a difference... but it'll be worth it!\nWish I found this issue earlier. I just counted 188 entries in my Credential Manager for . I've been seeing this problem for the last few weeks (months?), where vscode constantly signs me out of my settings sync. Finally looked into it this morning and found the, \"Not enough memory resources\" output message, which led me to this issue. I deleted all the authenticator tokens from CredMan, and it seems to have resolved the issue (for the last 24 hours, at least). I use VSCode Insiders full-time and update it on a near-daily basis. Some sort of toast message, triggered by the the \"not enough memory\" error, might be nice to alert the user to the issue and instruct them to clear out their ms-auth tokens.\nSpoke too soon. After a few days of daily Insiders updates and whatnot, I've continually been prompted to sign into Microsoft to sync settings, and each time, I end up with a dozen or so new tokens into CredMan. They appear to just grow unbounded. Is that expected? Am I going to have to periodically run the command to clean these up? I assume I'll eventually hit the, \"Not enough memory\" error at some point if I just let the list continue to grow?\nThanks for the update. We'll be doing more drastic work in this space this month and next and my hope is that this will be resolved then.\nI don't know if this is useful or not, but I've only seen a Microsoft Authentication problem when running multiple instances of VS Code. By this I mean: Start VS Code normally, do work... Open another instance with the \"Workspaces: Duplicate As Workspace in New Window\" command Do more work Shut down VS Code for the day with File-Exit, and not by closing each window individually with X in the upper right. Next day, start VS Code just by clicking the icon in my taskbar. Both instance windows show up in the same state I left them in the day before Work some more I haven't seen a problem for months because I WAS NOT doing the above workflow. When I opened another instance, I closed them separately, which means I only get one window on the next startup of VS Code. But I did do the above workflow last night and today and about 4-5 hours after starting it this morning the Microsoft authentication failed with messages: I have no idea what's causing this, but it seems that problems related to this happen more often with multiple instances running. There were a few people commenting on this issue that they were running multiple instances. I also don't know if this is happening because VS Code was started with 2 instances, like I did this morning, or if it is just sufficient to have more than one instance where the additional instances were started with that Duplicate command after the first instance was created. I tend to suspect the former because I never saw the problem when starting a VS Code session with just one initial instance. I saw it today after starting a session with 2 initial instances/windows. Again, I have no idea, but it feels like the two instances of VS Code are banging on the authenticator at the same time or are running some non-reentrant code at the same time on the way to the authenticator. HTH\nI cleared out the credentials, but in less than a day I was back to more than 100 entries. The problem also started showing up on two machines that didn't have it, immediately after the latest update. (As in, 100+ credentials created starting in the minutes after VS Code restarted for the upgrade.) Notably, in light of the above comment, my setup does involve multiple restored windows - and the upgrade restart does restore them.\nThanks all. Sorry to hear this issue has not been fixed yet and sad that I've not been able to repro this. I will say that we have some plans in the works for dealing with secret storage in light of 's archival. More on that later.\nI'm also running into this, my Windows Credential manager is full of lines like this: ! See and since yesterday it is full again, so I'd be happy to help debug this if you can point me to the right logging commands.\nNot sure related or not, but the latest version would cause consuming high CPU, especially when VS Code thinks I need to re-log in my account.\nEven after log-in, CPU usage would not restore. Need to close vscode all over\nI think I have finally found the issue and have a PR out for this. Thank you all for your patience and providing good information. To help me, you can get the insider build here: My change will be in the insider build that will go out either Thursday or Friday.\nSo far so good, I installed the insider build, played with some different git repo folders, and I have only 2 entries in the Credential Manager: ! The second one seems to be fore github copilot.\nMy change hasn't gone in yet\nI haven't seen this problem for about a week on Insiders build, after several attempts to remove all credentials manually or with script. Problem re-emerged few times, but works as expected after another run of the script. Hope update would fix it for everyone and it won't re-emerge again. P.S. I haven't use a lot of workspaces lately (as mentioned in PR), will try to get into bad state before update once again.\nLooking at your fix, I might be more aggressive about deleting all the chunks even if one of those deletes fails, for example, could 2 vscode instances be trying to delete the same chunks at the same time? Do you know what the user scenario is that results in very large passwords that need chunking? That might help me test it better...\nThis isn't possible because deletes are now handled sequentially thanks to that PR. Microsoft Auth via Settings Sync. The limit in CredMan is a measly 2500 characters. A refresh token that gets stored in Cred Man is usually way larger than that. That refresh token also gets updated every reload of a window (every time Microsoft Auth extension is activated) hence the issue with multiple windows writing at the same time when a VS Code update happens.\nYour suspected behavior written up in the PR definitely tracks with my usage patterns and observations. I'm looking forward to the fix! Should we clear out the hundreds of tokens inside CredMan ahead of your fix going live? Or will they get cleared/handled as part of your revised operations? Thanks!\nyou'll want to clear out CredMan.\nI just updated to 1.76.1. Now multiple vscode windows cannot sign in sync. A sign-in of one window will kick off others'.\nIt's likely the same problem since this issue was addressed in 1.77 which is only in insiders and will be released in a few weeks.\nI was able to reproduce on my PC with then before update and it \"spammed\" credentials. After cleaning up credentials, updating, cleaning up again - 12+ days without issues. I think I was asked once (it might been for GHPR extension), but there is definitely no \"spam\" in credentials so far. However, it took noticeable time to repro it in so hard to tell for sure\nThanks for your report This one...... will be hard to verify.... and I think what you did is as close to a proper verification as we will get. I will mark this as verified and we will see what the 1.77 stable release gives us in a couple weeks. If anyone sees this issue in 1.77-insiders until then, please don't hesitate to reach out.\nThis issue doesn't look the same as , but I'll wait for 1.77 and reopen my bug report if the issue is not fixed.\nI have the exact same issue as and I agree that it doesn't appear the same as . I will also wait for the new fix and see if it fixes my problem.\nNow I am asked regularly to choose account so sync my settings from and when the list appears the Microsoft account appears as \"(signed in)\". I cleared the stored authentications a week ago (before the 1.77 update).\nLooks like folks still see this in 1.77 at least partially. I need a new issue because this one is just way to busy... so we'll keep going in\nthis is useful\nJust want to report that on Windows, the keychain could get into a bad state due to this issue, after which even upgrading to the newer version of vscode doesn't help'. One would need to take the actions suggested in , as mentioned above, which solved the issue for me.\nit is actually worse now. I am being asked to select the account to sync settings from repeatedly upon launching VS Code. The Microsoft accont I am using appears as \"(Signed in)\", so the token does not appear to be lost, just disacossiated. I can't see anything within the logs.\nIs it worse or did your Windows credentials just get full? That happens to me from time-to-time and it's annoying. posted above provided a recipe for deleting all of the Windows credentials. This fixes it for a while, then I have to do it again.\nfolks, re-posting what I'm saying above:"}
{"_id":"q-en-vscode-44b79b26e5e1893eda3539bcf71dfa540d7effb0b0ec7951a70564e88993033d","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --With , changes in a diff editor: [![enter image description here][1]][1] With , changes in a diff editor: [![enter image description here][2]][2] With , changes in a normal (non-diff) editor: [![enter image description here][3]][3] How to ignore changes in leading/trailing whitespace in a non-diff editor as well? If such a setting doesn't exist, please consider adding one. [1]: [2]: [3]:\nI've removed the label, because this issue is about the hard-coded value . If that line of code would be commented out, I think everything would work as desired (the quick diff would respect the setting).\nMaybe I'm misunderstanding, but this doesn't seem to work? ! In case it isn't clear in the picture, I just whitespace\nI reproduced as well.\nTurns out this isn't true.\nSorry about that!\nNo worries, I know you were just hinting.\nVerification: There's a new setting to control this."}
{"_id":"q-en-vscode-763d9ef87493f1075a7570f415cbd6e6964578f3399133d8bada9b48495621d0","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.60.0 OS Version: MacOS 10.14.3 Steps to Reproduce: & paste into markdown file <kbd⌘</kbdand click the link .ext)\nThis is my first go at launching into the vscode code, but hopefully this information may be useful. Perhaps we could update the parsing of the URI schemes in general to strip out the angle brackets when editing markdown? See my findings below for why I'm suggesting this, but I'm also unsure of what problems and side effects doing that may cause. Main points of interest While the Command + Click doesn't follow the web link when you're a Markdown file, it does follow the link in the of the Markdown file. So I think markdown syntax is actually ok In debugging the project a little bit, I found that the link you try to follow that has angle brackets may be trying to or something similar (instead of open a web browser) In the function , I think the scheme of http(s) isn't recognized because of the < brackets in the link portion. See around line 18, where it tries to recognize the scheme, but doesn't recognize it when it tries calling In the file you can add a few new schemes, and then you're able to follow the link just fine. Though, this is probably not the final way the issue should be fixed"}
{"_id":"q-en-vscode-6fc52db764e6d2ead81a9f5f1099574a1b00e6363eba231d7a75782e582cb105","text":"Issue Type: <bBug</bthe settings UI. for a setting (e.g. \"inlay\"). the settings tab to a new editor group. Note that the search you entered previously has been cleared. https://user- VS Code version: Code 1.60.0 (, 2021-09-01T10:54:53.442Z) OS version: Darwin x64 20.6.0 Restricted Mode: No <details<summarySystem Info</summary </details<details<summaryExtensions (26)</summaryExtensionVersion ------ vscode-css-formatter1.0.1 vscode-nginx-conf0.1.3 reorder-lines-by-length1.0.2 docs-view0.0.9 pgformatter1.17.0 vscode-eslint2.1.25 gitlens11.6.0 vscode-diff1.4.2 shell-format7.1.0 go0.27.2 postgresql0.0.2 rainbow-csv1.9.1 render-crlf1.5.21 template-string-converter0.5.3 python2021.9. vscode-pylance2021.9.2 jupyter2021.8. jupyter-keymap1.0.0 debugger-for-chrome4.13.0 nbsp-vscode0.0.1 vscode-commons0.0.6 vscode-yaml0.22.0 final-newline0.3.2 vscode-hexdump1.8.1 change-case1.0.0 markdown-all-in-one3.4.0 </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nA function called in is being called after the drop, which is correct. But, the function in the same file name isn't being called when the user does step 3 above. Another repro is to use the \"move editor to X group\" commands. Therefore, the fix doesn't have to do directly with dnd."}
{"_id":"q-en-vscode-0c8e6226fe0da12c475adf537a6e6b84675f21660d8743599cf9e80133baec6f","text":"Testing !\nThe binary is indeed available in the server folder but the libc library it is linked with does not exist on the image.\nRipgrep should also build for rust target in to support alpine images so that it is uses the musl c library.\nI can add that, thanks\nRipgrep couldn't build under that target. My understanding is… when building with musl, it uses jemalloc, jemalloc uses backgroundthreadsruntime_support, and that doesn’t support our target. If we can figure out how to un set that flag on jemalloc that might work around it but i don’t know what consequences that would have\nfrom it is not sure whether the bug is in jemalloc config or cargo, but the immediate working solution is to build the crate with I will see if that works.\nMay I know if there is some workaround for this bug for users? If not, I do not think its good to ship Alpine ARM 64 builds with broken experience.\nIt will require users to install package that will include the required libraries for the executable built using glibc.\nNot with your PR, right And thanks for the PR!\nYup is not needed with the musl binary produced from the PR.\nI didn't put up a PR to adjust to download for alpine. Can you take care of it ? Thanks!"}
{"_id":"q-en-vscode-cc1beac198b88c0364a9ff9edbcdd7bb2a3e376b0b2c26c4f56475106a3e5aeb","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes/No <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: Insiders OS Version: Ubuntu Steps to Reproduce: New machine Install VS Code insiders Open an existing notebook text/latext output is rendered by built-in renderer Install jupyter based on VSCodes recommendation Re-load VSCode text/latext output is still rendered by built-in renderer Note: This workflow is not going to work as we know of a lot of text/latext that doesnt' work with the built-in renderer, & novice users won't know that they need to change the renderer to Jupyter Notebook renderer. I've submitted a PR to remove the text/latex renderer from VS Code\nI believe this is where we select the renderer (we currently always take the first one): I agree that automatically deprioitizing built-in renderers makes sense\nStarting from a clean user-data-dir, I follow the steps, and it picks the built-in one by default. ! If I pick the extension renderer, it seems to remember that selection.\nHow do I get the latex mime types to show in that picker? I only see one: !\nTo verify: sure you have the jupyter extension installed and are in an ipynb notebook a normal python cell, run: the output is rendered nicely <img width=\"515\" alt=\"Screen Shot 2021-11-01 at 9 20 35 AM\" src=\"https://user-\""}
{"_id":"q-en-vscode-57609f3613f189aeebda9b780b629b1f85d78fac5e067398762eee13e76e010b","text":"Issue Type: <bBug</bI have two files in my project in different locations. and . The 'search in files' popup always shows first, event when searching for the latter: ! please include case-sensitivity in the sort function for this. VS Code version: Code 1.60.2 (, 2021-09-22T12:00:31.514Z) OS version: WindowsNT x64 10.0. Restricted Mode: No Remote OS version: Linux x64 5.10.16.3-microsoft-standard-WSL2 Remote OS version: Linux x64 5.10.16.3-microsoft-standard-WSL2 Remote OS version: Linux x64 5.10.16.3-microsoft-standard-WSL2 Remote OS version: Linux x64 5.10.16.3-microsoft-standard-WSL2 <details<summarySystem Info</summary </details<details<summaryExtensions (5)</summaryExtensionVersion ------ jupyter-keymap1.0.0 remote-ssh0.65.8 remote-ssh-edit0.65.8 remote-wsl0.58.2 hexeditor1.8.2 </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --"}
{"_id":"q-en-vscode-a844cbd07fd10e0f000c4bf76e80e6432e8ad3bd0339298a9a15830bd2e37cc6","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --There is a new color which can be used when we split an editor in group. The layout can be horizontal or vertical. Please support different border color settings for vertical and horizontal layouts. For example in the following vertical layout I don't need/want any border. It is easy to distinguish each editor. ! But it is hard to distinguish editors in horizontal layout without a border. !\nSounds reasonable, open for help for a meaningful small PR.\nVerification: configure colors for the new theme colors and and verify they apply properly as advertised depending on the split in group layout"}
{"_id":"q-en-vscode-5a2a894cc2c05c628917481081aa3477e2158146c9d719f3595c7ef68196e67d","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --In Jupyter Notebooks, finding witrh regex works but replacing with capturing groups doesn't. I'd like if it does. I tried as in the usual editor, but the replacement treats as literal. I had to work around by copying the code section to the usual editor, and then find and replace. This seems like an inconsistent UX, and personally quite inconvinient.\nShould I have opened this at ? If so, please close this so I can move on.\nThanks again for the help, it makes the find experience in notebook closer to the normal text editor one.\ndid you create a test plan item for this? I just the verification needed to it."}
{"_id":"q-en-vscode-1e94ea7691f1c728070a1e8dca70c057dfe04c905621e6a989e70bac837c938f","text":"Pull out most xterm handling from into its own class that encapsulates addons, latest write/parse, _enableWebglRenderer/other renderer stuff, customkeyhandler?, etc. The goal is to make much smaller and enable easy unit testing of previously core behavior by not having so many service dependencies. Expose as public ITerminalInstance to avoid creating passthrough functions like scrollToTop Expose as public but only to TerminalInstance (ie. don't expose on XtermTerminal's interface) will be mostly a glue class (process manager <-xterm), not intended to be unit testable but by integration tests"}
{"_id":"q-en-vscode-70a65e8e1111c3cd6356b28096f7c1a49a48de4a0e1bd06235847f83a984b53a","text":"There are too many issues with local echo and can really degrade the experience right now for local workspaces. We should add an on/off/auto setting where auto is disabled in local workspaces, so opt-in ssh in local terminal support is there.\nwhat steps would we take to verify? Looks like there's a new setting?\nI tested the setting"}
{"_id":"q-en-vscode-d84c893624e31d8d87abde12e26b66b50c77db8b4bbb7262416d8e325ee1c976","text":"Issue Type: <bBug</bworkspace file Create the other folder. it in Visual Studio Code on Windows  New Terminal, select 'scripts' error: The terminal process failed to launch: Starting directory (cwd) \"D:sourcesappscriptsD:sourcesappscripts\" does not exist. VS Code version: Code - Insiders 1.62.0-insider (, 2021-10-29T05:16:23.014Z) OS version: WindowsNT x64 10.0. Restricted Mode: No <details<summarySystem Info</summary </details<!-- generated by issue reporter --\nCan you share your file and also do a trace log while creating a terminal?\nThanks, I don't see any logs indicating an attempt was made to create a terminal. Can you pls try again and capture the logs associated with that?\nshould i wait before logs are written?\nYou should try to create the terminal (reproduce the bug) and then copy the logs which should contain information about the process\nsorry, I missed important setting in repro. I've created reproducible sample. Unpack it and open workspace file. And then try to open any terminal Seems like error is not stored in logs. I see [2021-10-29 17:40:03.486] [renderer1] [trace] RequestService#request [2021-10-29 17:40:03.525] [renderer1] [trace] {\"name\":\"EntryNotFound (FileSystemError)\"} [2021-10-29 17:40:03.525] [renderer1] [trace] {\"name\":\"EntryNotFound (FileSystemError)\"} [2021-10-29 17:40:04.960] [renderer1] [trace] CommandService#executeCommand [2021-10-29 17:40:04.960] [renderer1] [trace] CommandService#executeCommand _workbench.pickWorkspaceFolder [2021-10-29 17:40:06.367] [renderer1] [trace] terminalInstance#ctor (instanceId: 1) {\"cwd\":{\"{workspaceFolder}\",\"scheme\":\"file\"}} [2021-10-29 17:40:06.572] [renderer1] [debug] Terminal process exit (instanceId: 1) with code undefined [2021-10-29 17:40:06.572] [renderer1] [debug] Terminal process exit (instanceId: 1) state 2 [2021-10-29 17:40:06.572] [renderer1] [trace] terminalInstance#dispose (instanceId: 1)\nI've updated bug description with missing part\nI have reproduced a similar issue when setting\nThe issue seems to be caused by commit I have replied here\nIn the latest insider build this bug it still reproduced with the exact same steps. However folder is different. Before fix it concatenated it this way for folder \"scripts\" \"D:sourcesappscriptsD:sourcesappscripts\" After fix you can see the results below on screenshots: This is shown when I try to open 'scripts' folder (from steps in bug) ! This is shown when I try to open 'other' folder (from steps in bug) ! As I can see, concatenation of relative paths became broken. Should we reopen it?\ngood point, thanks for raising this. Didn't check the version info hard enough in your report\nI want to add some words of appreciation of VS Code. I'm using it for 4 years already and incredibly happy about it. And thinking about lots of efforts in the background that you and other team members are putting to make this masterpiece editor, I feel envy that I'm not moving the entire world of development forward like you guys..."}
{"_id":"q-en-vscode-000cdf97eca5021e8a9ce63803f3c843bb9c209e72b737628449f779f752015e","text":"Does this issue occur when all extensions are disabled?: Yes As described in the title and the clip below will say the rest, the canvas seems overwhelmed with the output and doesn't wrap it. https://user- The log: Because the lower trimmed part are wrapped, it thinks it don't have to show the scrollbar so we can't view the rest if the line is long enough. This is critical. ! If we put some colors, we will see the lines overflow at the edge: ! Smaller outputs are fine: ! <details<summarySystem info</summaryVersion: 1.62.0-insider Commit: Date: 2021-10-29T07:49:33.336Z Electron: 13.5.1 Chrome: 91.0.4472.164 14.16.0 V8: 9.1.269.39-electron.0 OS: Linux x64 5.11.0-38-generic </detailsI haven't got a chance to test on other OSs. This is my test notebook:\nthis is just a bug, but could probably be considered as part of the discussion about\nthanks for the feedback. What are your thoughts on the current behavior, where we supply a link at the top of the output that allows you to open the output in a separate editor tab? This behavior can indirectly essentially solve problems like this.\nIt's definitely a good design. The issue actually doesn't stop me from doing my job. My opinion is, as you said, it's just a bug - and I just want to let you know.\nI'm taking this with the assumption you're not currently looking at it."}
{"_id":"q-en-vscode-5c4589e2d6b3419eeadd38c13c65bc6f52cdfd82702990933ac68a0e531635e9","text":"VS Code version: 1.63.0-insider Jupyter Extension version : v2021.10. Python Extension version : v2021.12.-dev OS : Windows 11 Python and/or Anaconda version: 3.9.7 Type of virtual environment used : N/A Jupyter server running: Local Render correctly the Markdown text using the default renderer: ! This is the correct rendering for this: ! a new Jupyter notebook a new cell code with the following Python: Ctrl-Enter to render cell Via \"Choose Output Minetype\", selecting the \"non default\" renderer (i.e. the one with \"(vscode.markown-language-feature\") for Markdown: ! Gives the expected result: ! Also, I could not find a setting allowing me to make the functioning \"Output Mimetype\" the default... Thanks!\nThe actual repro attached.\nthis might be another example of the wrong renderer being chosen.\nI get the correct default renderer first time now. text/plain is the only other output option. !"}
{"_id":"q-en-vscode-3d60a9d953c78ec8055d511fda255f5c5205a58694b7182142043c3718e95b43","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --Hello, great teacher! In the long-term use of vscode, I found that the number of method parameter overloads has a dark foreground color, and my eyes will be tired if I see it for a long time. I ask the teacher to allow us to modify the foreground color to our own suitable foreground color. It will be great for people who have been coding for a long time! Hope this problem can get your support and help from the teacher to improve it in the next milestone. Thank you very much! !\nI would advise you to switch to light theme, is what I recommand. !\nThanks for your answer! But looking at the light theme for a long time will make the eyes more tired. These are one of the UI elements that are used very, very frequently. Improving it will be very friendly to most people. From the picture you provided, it can be seen that even in the light theme, it is also very dark, and it takes a very difficult time to recognize it.\nIf you want to improve the situation more perfect, it is recommended to add the following oval background color settings, which will be very beautiful and modern, and can bring people a very pleasant coding experience more in line with ergonomics. !\nThe contrast on the numbers needs to be increased. Here's the relevant css:\nI would recommend these themes below : official dark pro is coming (dark black) Choose the one which is comfortable for you after trying with all the three hope i was helpful\nThank you for your reply! I tried these topics a long time ago and can't solve this problem.\nHello, teacher! Is it convenient to modify the style of this css?\nHello, teacher! This problem is really important to me, because my eyesight is not very good, so I hope you can accept it and help solve it in the next milestone. After solving this problem, I will no longer report UI-related problems in the future. I really hope you can help solve it, thank you very much for your hard work and efforts!\nHello, teacher! This problem is really important to me, because my eyesight is not very good, so I hope you can accept it and help solve it in the next milestone. After solving this problem, I will no longer report UI-related problems in the future. I really hope you can help solve it, thank you very much for your hard work and efforts!\nyou can just modify the css just like said. go to , then search for , find next to it, and change it to .\nIf possible you can switch to the Cobalt theme. I am using it from long time and it's too good.\nthanks for your help! I tried the solution you gave, but a similar solution could not solve my problem.\nDear teacher! Can you solve this problem in the next milestone? If the teacher can solve it in the next milestone, the students really thank you very much! You can refer to the following style design: It is recommended to add the following oval background color settings, which will be very beautiful and modern, more ergonomic, and can bring a very pleasant coding experience. !\nI edited this issue. I appreciate any feedback on the PR\nHello, teacher! The new year has come, I wish the teachers a happy new year! ! ! In the new year, I hope that the teacher can help solve this problem in time, thank you very much! To solve this problem, there are no other UI related problems to be solved later.\nHello, teacher! VSCode is a great project that gives users the greatest autonomy from the very beginning, allowing users to personalize configurations according to their own habits, so it can fully surpass the long-born jetbrion series in just 5 years. The most important reason is that VSCode's open source thinking and allowing users to customize and configure VSCode development methods to the greatest extent can achieve today's achievements. I hope VSCode can continue to develop this way of thinking and development, focusing on every reasonable request of users and need. Thank you here! vscode contributors\nHello, teacher! VSCode is a great project that gives users the greatest autonomy from the very beginning, allowing users to personalize configurations according to their own habits, so it can fully surpass the long-born jetbrion series in just 5 years. The most important reason is that VSCode's open source thinking and allowing users to customize and configure VSCode development methods to the greatest extent can achieve today's achievements. I hope VSCode can continue to develop this way of thinking and development, focusing on every reasonable request of users and need. Thank you here! vscode contributors\nThank you very much! teacher!\nHey! I also think that text should be more readable. I created a PR to fix that:"}
{"_id":"q-en-vscode-85a1dec72fb85034e9a2d56f92ae1e83f360daf018329d099754acc62072e6a6","text":"The current API does not expose the same level of internal granularity as the user setting allows. This proposal is to expose the telemetry levels. The current and can still be used a simpler API and this one can be considered more advanced and granular. Extension authors cannot just observe the user setting as this will lead to doing the wrong thing.\nAnother possibility is that we could collect an array of acceptable levels if we're worried about the possibility that the levels will not continue to be progressive but a permissions system instead. I.e. collecting errors and crashes would have be an array . None wouldn't really make sense in this context and would probably go away in favor of an empty array.\nThis is implemented but will remain in the proposed state for awhile for testing\nClosing in favor of"}
{"_id":"q-en-vscode-52bbcbbfe3a6b0bc1eb4b7c4bbd0c84ea737577e3ca348b8a0ebc4e85cbf8a4c","text":"Testing When browsing themes while offline, I didn't get any message that I could not connect to the marketplace and/or could not retrieve themes, with the only visible feedback being that the theme search would instantly return with an empty quickpick. It would be nice if this information could be surfaced somehow.\nVerified: !"}
{"_id":"q-en-vscode-dca27573578db0b1e26638e626a661dbfea9c4aec912758aa725477f1174f48d","text":"Are all these settings still \"experimental\"? This seems like a mature feature !\nThis is a good point. I believe the wording is used to indicate the functionality has some quirks. We will see if thinks it's fine to remove it.\nYep let's do it, I think we just didn't remember to remove"}
{"_id":"q-en-vscode-50ff39af5d4399cf7b1aea0e3486b492974cbcb773b0eace1cd26cbfc5bdd80b","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes (disabling Emmet) <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: - 1.62.3 OS Version: - Ubuntu 21.10 Steps to Reproduce: a .js file which is denoted a React file by VSCode part of the coding implement state for a counter with the following simple line: getting as far as typing I wish to just exit the square brackets however Emmet takes the first position tab results in an HTML string appearing, see pictures Question: How do I move Emmet down the list of suggestions in Autocomplete for a React file, I am using other extensions like TabOut (to get out of the square brackets without any autocomplete) which would be more appropriate. ! ! ! !\nThis issue heavily affect the react users.\nVerification steps: a JSX file. . should not be an Emmet suggestion.\n/verified"}
{"_id":"q-en-vscode-7498b5edbb9713d00259afd0e46dd7e0b174aa17f243ec3c189d37d7b2085719","text":"<!-- Please search existing issues to avoid creating duplicates. --- VS Code version: 1.62.3 (Universal) Jupyter Extension version (available under the Extensions sidebar): v2021.10. Python Extension version (available under the Extensions sidebar): v2021.11. OS (Windows Linux distro) and version: Mac Python and/or Anaconda version: 3.7.12 Type of virtual environment used (N/A ...): conda Jupyter server running: Remote I expect code cells to begin executing immediately and (for simple code snippets) to finish executing immediately (e.g. print('hello')) Once the jupyter notebook starts to contain a non-trivial amount of stuff, I start to observe the following behaviors: (1) code cells that contain simple tasks (e.g. ) start taking several seconds or more to complete, (2) it may take several seconds or even minutes for vscode to even visually show that it will start executing a code cell, and (3) in severe cases, the whole vscode editor will become unresponsive and I will need to force shut it down. Below is code for a jupyter notebook that contains a minimal example. The problems start to appear once you've cycled through the code cells 2 or 3 times. I think one of the main contributing factors to this issue are just notebooks that have a lot of \"stuff\" in them. To demonstrate that, the last code block prints 50 scatter plots as pngs, which the native jupyter lab server over the browser handles just fine... but vscode seems to have issues. Moreover, in this simple example I have observed instances where it starts to take a while to save the notebook and the editor may even become unresponsive. Other things to note: I am running this jupyter notebook on an AWS EC2 instance with ~90GB memory 48 CPUs (using a AWS linux ami: ). I have not yet observed these same issues when I have the kernel running on my local mac machine. <details<summaryJupyter example </summary<p</p</detailsHere is a screenshot where it takes almost 3 seconds for a simple print statement. I've seen worse in some of my code related to real projects, though. This is just what I was able to reproduce with a minimal example. <img width=\"1593\" alt=\"Screen Shot 2021-11-22 at 4 48 27 PM\" src=\"https://user-\"\nThanks for filing this issue & I'm sorry you're running into this issue, please could you try the following: Check the performance withouth any data frames in the outputs You can run most of the same cells, but not display the outputs (e.g. don't run the cells with ) I'd like to check whether the perf is a result of large outputs. If that doesn't make any difference please could you: Install VS Code Insiders Go into your settings and search for and change the logging level for Jupyter extension to Reload VS Code Try the same thing & provide the logs from the output panel (use the command to get to the output panel)\nThanks . Not displaying large amounts of output does help. I experimented with the code a bit and here are my observations: (1) If I disable the cell with just the statement (but still outputting the figures), while it helps a little, I still see a significant slowdown. In particular, once I execute the code block that prints the figures (see below) and it is done, I try to execute the first code block again (containing the import statements), but there is a good 30-60 second delay until it will start to executing. (2) If I still create the figures but no longer display them (see below), this helps immensely with performance... there are basically no issues. (3) While still hiding the figures as in (2), but I go back to displaying and increasing the number of columns and rows of that are display to 200 (see below), then there is small decrease in performance. (4) If I do the same as (3), but take it even further by displaying 500 rows and columns of , then performance significantly degrades (see the screenshot). Once it was done displaying , it took ~19 seconds for a print statement, ~35 seconds to make the figures (normally takes a ~3-5 seconds), and it took over a minute to execute that last empty cell (the execution time was hidden once it was done). <img width=\"1585\" alt=\"Screen Shot 2021-11-23 at 10 14 06 AM\" src=\"https://user-\"\nThanks for the feedback. Please could you: Install VS Code Insiders Go into your settings and search for and change the logging level for Jupyter extension to Reload VS Code Run the command An editor will be opened with some text such as , please copy all of that & paste that into this issue Try simple print statements (without any data frames, and figures, lets keep it very simple) Please ensure the notebook is empty & does not contain any outputs only 1-2 cells with simple print statements. Provide the logs from the output panel (use the command to get to the output panel) Reload VS Code Repeat what you did earlier Provide the logs from the output panel (use the command to get to the output panel) Please note, I'll need two logs, one without any data frames, no plots, just print statements. The other is the one with dataframes and the like.\nHere is the text for : Even if vscode totally froze up while trying to run something in jupyter, I was still always able to ssh into the server no problem. Here is the Jupyter output from just the print statements: <details<summaryLog</summary</detailsHere is the Jupyter output where I print large dataframes and plots: <details<summaryLog</summary</detailsFew other things I've observed: (1) In the second log above, I see there are some lines like in it, but I was not always able to reproduce that output. (2) Here are some additional logs that I observed while it took 15 minutes to execute a cell with a simple print statement (but it was executed after cells with the dataframes and plots): <details<summaryLog from </summary</details<details<summaryLog from </summary</details\nThanks for the logs. Transferring to VS Code due to connectivity issues with remtoe SSH\nAlso having this issue with jupyter over ssh, in a notebook that has a decent number of plots.\nThis issue happened with me also (especially when i am insider a docker container inside the remote machine)\nI have the same issue, even when I freshly open a notebook with some outputs it is very slow and unresponsive. <img width=\"273\" alt=\"image\" src=\"https://user-\"I have a stable connection to the server: !\nSame problem, sometimes it works fine, sometimes it is unusable, stable and insider !\nAfter discussion in sync. Possibly add logging for every message that's only turned on if the user explicitly is told to (for debugging purposes).\nexperiencing similar issues when working on jupyter via ssh to an ubuntu server from an ubuntu pc\nALso had the issue where running a cell lags(or starts after some time) for jupyter notebooks when using remote SSH aws azure\nStill the same problem\nI also experience the same. I observed that: It happens with Jupyter notebooks, but not in the Interactive window. It does not happen if you connect to a Jupyter kernel that is not managed by VScode. It happens when your notebooks has many figures. I hope these may help diagnose the problem.\nSame here..\nAs an additional info to what enlightened us. Jupyter execution on remote ssh machine slows down (i.e., gets stuck) when notebook size is large (64 MB). VS code executing notebook cell normally for smaller sized (ie., 1MB ) notebook\nWould it be possible to share such a notebook? Is the size 64 MB with or without ouput in the notebook? If its a large file you could optionally email it to one of us (us = Microsoft maintainers of VS Code)\nsorry that is confidential work for the time being.\nExperiencing the same issue as well. Networking is not a problem since I connect to the notebook's machine on a local network\nI've also been having this issue lately although I think it only happens in certain circumstances. I can't share the notebook I'm working with at the moment as it's private (sorry!), but it does have many matplotlib figure outputs and dataframe outputs. The notebook itself is only 3.6 MB in size. I'm running the notebook on a Linux server with Python 3.9.7, specifically I'm connecting to the machine remotely with \"Connect to Host\" to edit and run code on the server remotely. Latest version of VS Code and all extensions, connecting from a machine running Windows 11. It seems connecting on the same network as the machine with my Linux laptop doesn't result in any issues, and I haven't noticed any problems with running the notebook from there. However, connecting from a different network when working from home causes the same notebook to be very laggy, including lagging for 30+ seconds on simple code cells containing e.g. just a print statement. I don't see how network speed/my connection to the server from home could be the issue (12ms ping, ~5 Mbps upload / ~50 Mbps download connections). Maybe there's something in the different OSes having different results, could also be unrelated though. I should try connecting with a Linux machine not on the same network and see if I can reproduce the results.\nSame problem here, just a correction it is not dependent on the figure number or commands difficulty, will take to execute 30 to 40s, and yes with a separate instance of Jupyter Lab server (not managed by VScode). Directly connecting to Jupyterlab over ssh port forwarding order of magnitude faster. Workflow: Windows to Linux remote SSH (cluster) Another instance in the cluster used to set up JupyterLab instance Change Jupyter server to remote of the previously created instance execution super slow (even delay between starting and finishing) ~ 30 second delay followed up by ~ 40 s execution Connecting directly from Windows tunnel to JupyterLab ~ few seconds execution Note: Both the Jupyter Lab instance and the initial remote ssh share access to the file system, just different hardware. Also after changing to remote, some shenanigans happen with the vscode, unable to open files, etc, had to restart everything to make it work again.\nI've had the same issue with VScode and Jupyter - notebooks becoming unresponsive after a few cells are used. The console shows continuous (every 20 seconds) reconnecting. with the level set to \"trace\". I can share a relevant file if it is helpful.\nI've had the same slow issue with VScode and Jupyter when connecting to a remote docker container via ssh, and I've also observed that keeps producing tremendous streams\nIf someone can share a notebook that triggers this issue (lots of outputs, graphs with lots of data, etc?) that would be helpful, thanks\nI noticed that if I switch the format from svg to png (by changing the setting ), the notebook would run smoother. My network: Roundtrip latency: 5.520ms Up: 2898.6 Mbps Down: 3007.5 Mbps Authority: ssh-remote+** Roundtrip latency: 21.320ms Up: 25.8 Mbps Down: 385.0 Mbps A test notebook (1KB zip file), need , and to run:\nHaving the same exact issue and it renders the jupyter unresponsive: ! And it's still running. A temporary solution I found is to click on \"Clear outputs for all cells\" and more often than not it works smooth again. Only works if the cell stops executing though...\nI think notebook size may be the reason.\nI resolved my issue by closing the JUPYTER: VARIABLES in the terminal below Maybe in my case was an issue with the output variables inside the jupyter section ..\nI resolved it by using a jupyter server running on the remote machine which was not started by VSC. I'll use that until this bug is fixed unfortunately\nlife saviour! \"Clear outputs for all cells\" worked smooth as butter (:\nHappy to help. I saw in one of the other tickets this bug has a synergy with notebook size, and since clearing all outputs reduces notebook size, this might be a temporary fix until the size increases eventually and we need to do it again. I would love to see this fixed as this bug is the biggest hindrance to me day-to-day vscode experience.\nSame here, this issue was driving me crazy. That's the solution indeed, I had a lot of local memory usage in my code, not sure about what was the actual issue. EDIT: it seems I spoke too early. When a cell takes very long to compute, vscode seems to \"disconnect\" from the kernel, and any subsequenc cell evaluation waits forever without running.\nI pushed a change which should mostly fix this slowness with large outputs. I think there are more optimizations we could do. The issue comes when a save happens, and this blocks the remote extension host process for some period of time. Then a symptom of that might be that an execution appears to hang or take a long time. Would appreciate anyone trying this out in tomorrow's vscode insiders build. If you're seeing an issue related to the variables view, that wouldn't be fixed here, however I recently pushed a different fix that may help with that, and you could try it out in insiders + the jupyter pre-release extension.\nI have been experiencing these slowdowns (specifically on an AWS server) and I don't use the variables view. I definitely found that clearing cell outputs eliminates the hangs. I'm confused how this is caused by saving though, because I just found that the notebook I was working on wasn't being saved. I was doing some work tethered to my phone as I rode a bus into remote mountains. Certainly noticed that the bad connectivity meant even more waiting for things to execute than usual. Unfortunately, when I got back to stronger internet vscode was complaining it couldn't connect and was hanging when it tried to save. I figured it was fine since I saw 's comment and naively interpreted that as meaning I should feel comfortable that my notebook was getting saved each time it hung. Nope. When I finally had to reload window to get things back to working I found that all the code I wrote on the bus was no longer in the notebook (no worries, I had copied it out just in case, so no loss). I'm trying the 2023-01-13T07:49:05.588Z insiders build right now (is the \"tomorrow's\" build that should have the changes?). I think it is still too soon for me to be able to offer an opinion on how much of an improvement this fix might be. It is hard because the problem is erratic: sometimes cells execute immediately and sometimes it takes 20 seconds or longer.\nYeah, that version has the latest fix, you might have hit a different issue. And the issue might still have been kicked off by the notebook attempting to save. I'll leave this issue open to track namely, we still send notebook data between the processes more often than we technically have to, and I think that can be improved. Sorry about that, glad you had it saved, the intent is that the unsaved changes would have been persisted on the frontend.\nI can confirm that this change is a big improvement for my use case. If those additional optimizations ever get prioritized that would be amazing and appreciated, but I'm already grateful for this improvement. Thank you!\nOn second thought, closing this issue, and moving other work to\nI have tried the insiders version on a podman container accessed on a machine over ssh (so I do vscode remote via ssh to the machine and then I use remote containers to access the container). It does not open a terminal and seems to be spinning forever with the following error in the logs: 2023-01-27 10:40:00.790 [info] [LocalProcess0][resolveAuthority(attached-container,27)][1013ms] waiting... 2023-01-27 10:40:00.818 [error] Shell server failed: Error: stream ended with:4 but wanted:9 at u (c:Usersvince.vscode-insidersextensionsms-vscode-distextension) at c:Usersvince.vscode-insidersextensionsms-vscode-distextension at s (c:Usersvince.vscode-insidersextensionsms-vscode-distextension) at Socket.<anonymous(c:Usersvince.vscode-insidersextensionsms-vscode-distextension) at (node:events:538:35) at endReadableNT (node:internal/streams/readable:1345:12) at process.processTicksAndRejections (node:internal/process/task_queues:83:21) May this be related to the recent changes or shall I seek for another bug report and eventually open a new one?\nNot related to this change- you can open an issue against the Remote Containers extension, I'm not sure how to interpet that error\nHello, I have installed latest Insider's VS Code, maybe it's a bit better now, but in general issue is still there. Very slooow Jupyter Notebooks on large content. I'm using jupyter/r-notebook docker image to run on AWS EC2 instance and have 2 options to acces it: from browser on 8888 port to web version of Jupyter notebooks. It works lightning fast and responsive. over SSH to EC2 instance and then \"Attach to running container\" So if I compare those 2 options there is a huge difference in UI when notebook has some data and charts. For me it feels like VS Code continuously sends data to server. Even dropdown for intellisence takes few seconds to load. It's so sad that I'm unable to use VS Code Jupyter Notebooks for my work. But if I create a brand new empty notebook in VS Code - everything works like a charm!\nWe have experimental saving logic for Remote SSH, it would be great if you can give this a try and see if it improves the performance on large notebooks Install latest VS Code Insiders Remote SSH into your remote machine Add to your Remote/User settings Reload the window Then test the scenarios which used to slow or block the network Thanks in advance!\nThe improvements are noticeable, thank you for your great work! However, the latency is still there and appears to be significant from time to time. I love this ssh-notebook function, and I believe there must be more folks like me. Hope your team can prioritize this issue. Appreciate!\nHi, I just checked VS Code Insiders 1.81.0 version with \"Remote Save\" feature enabled in settings. I have generated in a loop 30 charts in first cell and after this I run code in a new cell. It still hangs for a while after generating charts for the first run. If I run second cell code again, it executes immediately as expected. So we still have an issue when a lot of data (or visual data) was generated recently. I checked the same code in web UI of Jupyter - and it works fast. Anton. From: Peng Lyu Friday, June 30, 2023 12:33 AM To: microsoft/vscode bo44arov Mention Re: [microsoft/vscode] Jupyter over remote ssh sometimes becomes slow and/or unresponsive (Issue ) have experimental saving logic for Remote SSH, it would be great if you can give this a try and see if it improves the performance on large notebooks Install latest VS Code Insiders Remote SSH into your remote machine Add \"notebook.experimental.remoteSave\": true to your Remote/User settings Reload the window Then test the scenarios which used to slow or block the network Thanks in advance! Reply to this email directly, view it on GitHub, or unsubscribe You are receiving this because you were mentioned.Message ID:"}
{"_id":"q-en-vscode-f184aa94d3216157356cc3068b4cbba3aab323a8bf775f22983f73cb6c74c3fb","text":"Tracks shipping the TypeScript recovery build: 4.5.3:\nVersion: 1.63.0 Commit: Date: 2021-12-07T05:15:48.091Z Electron: 13.5.2 Chromium: 91.0.4472.164 14.16.0 V8: 9.1.269.39-electron.0 OS: Darwin x64 19.6.0 !\ntypescript log: !\nthis problem appeared after the last update, until friday everything was working fine\nis there any temporary solution until fix will be shipped?\nthis problem to be related to libraries in my work, temporary solution until fix, is locked echarts version to\ndid someone was able to work around it? it is not working with the latest typescript as well... should we downgrade our vscode version?\nDowngrading to 1.62 fixed it for me. You can get it from: Also disable automatic updates before downgrading, otherwise it will push you back to the faulty release.\nClosing for verification.\nthanks\nI had the same issue. Thanks for the fix. :)\nTHANK YOU SO MUCH, this intellisense drive me crazy since last Friday!\nJust tried 1.63.2 and it's still failing with the same message. I submitted a new issue with the stack trace. Contact me if you want more info Version: 1.63.2 Commit: Date: 2021-12-15T09:37:28.172Z Electron: 13.5.2 Chromium: 91.0.4472.164 14.16.0 V8: 9.1.269.39-electron.0 OS: Darwin x64 20.5.0 TypeError: Cannot read property 'flags' of undefined at getPropertyOfObjectType (/Applications/Visual Studio ) at getPropertyOfType (/Applications/Visual Studio ) at resolveESModuleSymbol (/Applications/Visual Studio ) at getTargetOfNamespaceImport (/Applications/Visual Studio ) at getTargetOfAliasDeclaration (/Applications/Visual Studio ) at resolveAlias (/Applications/Visual Studio ) at getSymbol (/Applications/Visual Studio ) at resolveNameHelper (/Applications/Visual Studio ) at resolveName (/Applications/Visual Studio ) at resolveEntityName (/Applications/Visual Studio ) at resolveEntityName (/Applications/Visual Studio ) at getSymbolOfPartOfRightHandSideOfImportEquals (/Applications/Visual Studio ) at getTargetOfImportEqualsDeclaration (/Applications/Visual Studio ) at getTargetOfAliasDeclaration (/Applications/Visual Studio ) at resolveAlias (/Applications/Visual Studio ) at resolveSymbol (/Applications/Visual Studio ) at mergeSymbol (/Applications/Visual Studio ) at /Applications/Visual Studio at Map.forEach (<anonymous) at mergeSymbolTable (/Applications/Visual Studio ) at mergeSymbol (/Applications/Visual Studio ) at /Applications/Visual Studio at Map.forEach (<anonymous) at mergeSymbolTable (/Applications/Visual Studio ) at mergeModuleAugmentation (/Applications/Visual Studio ) at initializeTypeChecker (/Applications/Visual Studio ) at Object.createTypeChecker (/Applications/Visual Studio ) at Object.getTypeChecker (/Applications/Visual Studio ) at synchronizeHostData (/Applications/Visual Studio ) at Object.getProgram (/Applications/Visual Studio ) at ConfiguredProject.Project.updateGraphWorker (/Applications/Visual Studio ) at ConfiguredProject.Project.updateGraph (/Applications/Visual Studio ) at ConfiguredProject.updateGraph (/Applications/Visual Studio ) at ProjectService.createLoadAndUpdateConfiguredProject (/Applications/Visual Studio ) at ProjectService.assignProjectToOpenedScriptInfo (/Applications/Visual Studio ) at /Applications/Visual Studio at Object.flatMap (/Applications/Visual Studio ) at ProjectService.applyChangesInOpenFiles (/Applications/Visual Studio ) at <computed(/Applications/Visual Studio ) at /Applications/Visual Studio at IOSession.Session.executeWithRequestId (/Applications/Visual Studio ) at IOSession.Session.executeCommand (/Applications/Visual Studio ) at IOSession.Session.onMessage (/Applications/Visual Studio ) at Interface.<anonymous(/Applications/Visual Studio ) at () at Interface.onLine () at Interface.normalWrite () at () at () at addChunk () at readableAddChunk () at () at Pipe.onStreamRead ()\nI'm seeing the same thing on as well. Downgrading to worked for me as well. Version: 1.63.2 Commit: Date: 2021-12-15T09:37:28.172Z Electron: 13.5.2 Chromium: 91.0.4472.164 14.16.0 V8: 9.1.269.39-electron.0 OS: Darwin x64 20.6.0 <details</summary[2021-12-21 15:25:54.627] [exthost] [info] Extension host with pid started [2021-12-21 15:25:54.627] [exthost] [info] Skipping acquiring lock for Support/Code/User/workspaceStorage/. [2021-12-21 15:25:54.742] [exthost] [info] ExtensionService#doActivateExtension vscode.microsoft-authentication, startup: false, activationEvent: 'onAuthenticationRequest:microsoft' [2021-12-21 15:25:54.751] [exthost] [info] ExtensionService#doActivateExtension vscode.configuration-editing, startup: false, activationEvent: 'onLanguage:json' [2021-12-21 15:25:54.754] [exthost] [info] ExtensionService#doActivateExtension vscode.extension-editing, startup: false, activationEvent: 'onLanguage:json' [2021-12-21 15:25:54.759] [exthost] [info] ExtensionService#doActivateExtension , startup: false, activationEvent: 'onLanguage:json' [2021-12-21 15:25:54.775] [exthost] [info] ExtensionService#doActivateExtension , startup: false, activationEvent: 'onLanguage:json' [2021-12-21 15:25:54.782] [exthost] [info] ExtensionService#doActivateExtension vscode.typescript-language-features, startup: false, activationEvent: 'onLanguage:typescript' [2021-12-21 15:25:54.795] [exthost] [info] ExtensionService#doActivateExtension VisualStudioExptTeam.vscodeintellicode, startup: false, activationEvent: 'onLanguage:typescript' [2021-12-21 15:25:54.829] [exthost] [info] ExtensionService#doActivateExtension , startup: true, activationEvent: '' [2021-12-21 15:25:54.847] [exthost] [info] ExtensionService#doActivateExtension , startup: true, activationEvent: '', root cause: [2021-12-21 15:25:54.849] [exthost] [info] ExtensionService#doActivateExtension , startup: true, activationEvent: '' [2021-12-21 15:25:54.897] [exthost] [info] ExtensionService#doActivateExtension ms-vscode-, startup: true, activationEvent: '' [2021-12-21 15:25:55.232] [exthost] [info] ExtensionService#doActivateExtension , startup: true, activationEvent: '' [2021-12-21 15:25:55.246] [exthost] [info] ExtensionService#doActivateExtension , startup: true, activationEvent: '' [2021-12-21 15:25:55.479] [exthost] [info] ExtensionService#doActivateExtension , startup: false, activationEvent: 'onAuthenticationRequest:github' [2021-12-21 15:25:55.548] [exthost] [info] Eager extensions activated [2021-12-21 15:25:56.091] [exthost] [info] ExtensionService#doActivateExtension , startup: false, activationEvent: 'onStartupFinished' [2021-12-21 15:25:56.096] [exthost] [info] ExtensionService#doActivateExtension , startup: false, activationEvent: 'onStartupFinished' [2021-12-21 15:25:56.098] [exthost] [info] ExtensionService#doActivateExtension , startup: false, activationEvent: 'onStartupFinished' [2021-12-21 15:25:56.108] [exthost] [info] ExtensionService#doActivateExtension eamodio.gitlens, startup: false, activationEvent: 'onStartupFinished' [2021-12-21 15:53:52.346] [exthost] [info] extension host terminating: renderer closed the socket [2021-12-21 15:53:52.367] [exthost] [info] Extension host with pid exiting with code 0 [2021-12-21 15:53:55.138] [exthost] [info] Extension host with pid started [2021-12-21 15:53:55.138] [exthost] [info] Skipping acquiring lock for Support/Code/User/workspaceStorage/. [2021-12-21 15:53:55.198] [exthost] [info] ExtensionService#doActivateExtension vscode.microsoft-authentication, startup: false, activationEvent: 'onAuthenticationRequest:microsoft' [2021-12-21 15:53:55.204] [exthost] [info] ExtensionService#doActivateExtension vscode.configuration-editing, startup: false, activationEvent: 'onLanguage:json' [2021-12-21 15:53:55.209] [exthost] [info] ExtensionService#doActivateExtension vscode.extension-editing, startup: false, activationEvent: 'onLanguage:json' [2021-12-21 15:53:55.214] [exthost] [info] ExtensionService#doActivateExtension , startup: false, activationEvent: 'onLanguage:json' [2021-12-21 15:53:55.229] [exthost] [info] ExtensionService#doActivateExtension , startup: false, activationEvent: 'onLanguage:json' [2021-12-21 15:53:55.236] [exthost] [info] ExtensionService#doActivateExtension , startup: true, activationEvent: '' [2021-12-21 15:53:55.238] [exthost] [info] ExtensionService#doActivateExtension , startup: true, activationEvent: '', root cause: [2021-12-21 15:53:55.241] [exthost] [info] ExtensionService#doActivateExtension , startup: true, activationEvent: '' [2021-12-21 15:53:55.287] [exthost] [info] ExtensionService#doActivateExtension ms-vscode-, startup: true, activationEvent: '' [2021-12-21 15:53:55.334] [exthost] [info] ExtensionService#doActivateExtension , startup: true, activationEvent: '' [2021-12-21 15:53:55.352] [exthost] [info] ExtensionService#doActivateExtension , startup: true, activationEvent: '' [2021-12-21 15:53:55.431] [exthost] [info] ExtensionService#doActivateExtension , startup: false, activationEvent: 'onAuthenticationRequest:github' [2021-12-21 15:53:55.598] [exthost] [info] Eager extensions activated [2021-12-21 15:53:55.601] [exthost] [info] ExtensionService#doActivateExtension , startup: false, activationEvent: 'onStartupFinished' [2021-12-21 15:53:55.607] [exthost] [info] ExtensionService#doActivateExtension , startup: false, activationEvent: 'onStartupFinished' [2021-12-21 15:53:55.608] [exthost] [info] ExtensionService#doActivateExtension , startup: false, activationEvent: 'onStartupFinished' [2021-12-21 15:53:55.615] [exthost] [info] ExtensionService#doActivateExtension eamodio.gitlens, startup: false, activationEvent: 'onStartupFinished' [2021-12-21 15:53:55.958] [exthost] [info] ExtensionService#doActivateExtension vscode.typescript-language-features, startup: false, activationEvent: 'onLanguage:jsonc' [2021-12-21 15:53:56.730] [exthost] [info] ExtensionService#doActivateExtension VisualStudioExptTeam.vscodeintellicode, startup: false, activationEvent: 'onLanguage:typescript' [2021-12-21 15:58:22.661] [exthost] [error] [vscode.typescript-language-features] provider FAILED [2021-12-21 15:58:22.662] [exthost] [error] Error: Could not create TS service. Service state:{\"type\":0} at w.service (/Applications/Visual Studio ) at w.executeImpl (/Applications/Visual Studio ) at w.execute (/Applications/Visual Studio ) at u.ensureConfigurationOptions (/Applications/Visual Studio ) at u.ensureConfigurationForDocument (/Applications/Visual Studio ) at /Applications/Visual Studio at b.interruptGetErr (/Applications/Visual Studio ) at w.interruptGetErr (/Applications/Visual Studio ) at l.provideHover (/Applications/Visual Studio ) at o.provideHover (/Applications/Visual Studio ) at /Applications/Visual Studio at We.withAdapter (/Applications/Visual Studio ) at We.$provideHover (/Applications/Visual Studio ) at t.doInvokeHandler (/Applications/Visual Studio ) at t.invokeHandler (/Applications/Visual Studio ) at t.receiveRequest (/Applications/Visual Studio ) at t.receiveOneMessage (/Applications/Visual Studio ) at /Applications/Visual Studio at u.fire (/Applications/Visual Studio ) at r.fire (/Applications/Visual Studio ) at /Applications/Visual Studio at u.fire (/Applications/Visual Studio ) at r.fire (/Applications/Visual Studio ) at n.receiveMessage (/Applications/Visual Studio ) at /Applications/Visual Studio at u.fire (/Applications/Visual Studio ) at w.acceptChunk (/Applications/Visual Studio ) at /Applications/Visual Studio at Socket.T (/Applications/Visual Studio ) at () at addChunk () at readableAddChunk () at () at Pipe.onStreamRead () [2021-12-21 16:03:34.858] [exthost] [warning] TextEditor is closed/disposed </details\nSame here for 1.63.2, downgraded until this is solved....\nhaving the same issue here, how do you guys downgrade?\nSame happened to me. Had to downgrade to 1.62 (October 2021 update) for it to work.\nwhy is this issue closed? this is still not resolved...\nFor me, this can be fixed by in Version: 1.63.2 (Universal) Commit: Date: 2021-12-15T09:37:28.172Z (3 wks ago) Electron: 13.5.2 Chromium: 91.0.4472.164 14.16.0 V8: 9.1.269.39-electron.0 OS: Darwin arm64 21.2.0\nI was running node 12. Upgrading node versions also fixes this bug. Edit: Spoke too soon. I saw tool tip working for some locally defined types and thought it was working. It still can't resolve types outside the local file and pops up the crash warning after 5 minutes.\nit works for 2 minutes every time i reopen vs code until it crashes again. can u provide what version of ts u have installed in your projects? because for me this solution didn't help. i also downgraded the vs code version and it work for like half an hour and then again the same error\nSure. I hope this can be helpful for you.\nI think this is being fixed by While waiting for new vscode to have new typescript, installing ms- seems to help. (Got the hint on )."}
{"_id":"q-en-vscode-fe67dfdfb8f0d5a6312a6e5454d23a4d8ef8c993d36d6ffdbd1d6820ba937be9","text":"I was trying to copying a new setting in the problems hover and it was extremely painful to do so: https://user-\nThis is editor hover and hence assigning to\nFWIW this works fine for me on Windows."}
{"_id":"q-en-vscode-83e95ddb6dc2ed0ca1936410e6a7f1018024373262e65c7538f7f04aae780c2b","text":"Issue Type: <bBug</bI wanted to be able to copy the call stack through a keyboard shortcut so I created a binding for but found that it doesn't do anything. I tried different combinations of clauses and ensured that my keystroke was not being captured by something else but still no luck. Invoking the command directly in my extension like so also did not work and the thenable resolved to . The comment and post imply this should work outside of the callstack context so I've filed this as a bug as opposed to a feature request for a keyboard shortcut to copy the call stack. VS Code version: Code 1.63.2 (Universal) (, 2021-12-15T09:37:28.172Z) OS version: Darwin x64 21.2.0 Restricted Mode: No <details<summarySystem Info</summary </details<details<summaryExtensions (15)</summaryExtensionVersion ------ vscode-bazel0.5.0 arm1.5.2 vscode-diff1.4.2 fig0.10.1 misc-additions1.0.2 lldb-vscode0.1.0 jupyter2021.11. remote-ssh0.68.0 remote-ssh-edit0.68.0 vscode-remote-extensionpack0.21.0 cpptools1.7.1 errorlens3.4.1 vim1.21.10 clang-format1.9.0 material-theme3.13.6 </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nThe comment in code applies to the commands in general but that particular command can't be called globally currently. It copies a callstack from one thread and there can be multiple callstacks available so one has to selected to copy. Do you want the keybinding to work on the focused thread, or does that not help?\nYes, defaulting to the focused thread was my expectation."}
{"_id":"q-en-vscode-02e41e74f19a81966a043f89bdd7f7f3956a9644a99d35f90bc7c5cfdc692407","text":"We should not sync this Should this be per terminal instance? No We should persist similar to recent commands as well Should persisted entries work across different shells? Probably We should invalidate entries similar to commands Let's support removing entries, using a icon that shows only on active and hovered item Let's not use the x for failed, we can improve visualizing errors later on"}
{"_id":"q-en-vscode-9753b9e3ac8acd8fd440da504925001a77fbe39b50f69e9367f093b2c7db00cc","text":"Testing !\nWe'll need to repro on mac and look at the sequences to verify whether this is an issue or we're going the right thing.\nI think this is working as designed since it's just padding on the top, however we could check the location of the terminal view and if it's on either side, top align the terminal instead of bottom aligning it."}
{"_id":"q-en-vscode-73e570c236d61dcb0ce7978973ceb81627fbdcc6f1a26aabf9943e5ac16f5e2f","text":"Testing have no terminals, no panel visible reload vs code :bug: the open detected link, open last link commands are in command palette"}
{"_id":"q-en-vscode-6dfc01be3e4f9930d3178bc8d2fabaf13cb0e8d02c1f4713d668bd152213521b","text":"I have a profile for PowerShell using the and once I enable the the prompt is no longer loading. With it enabled I see this prompt: ! When I disable I get my expected prompt: ! My prompt function within my PowerShell profile:\nthis is happening because line is throwing because there is no history in a newly created session\nchampion , should be able to handle this early next week\ni could do a pr\nalso, here you should use and check if the CWD is a filesystem path. like:\nthat would be awesome! I'm fairly new to pwsh so any comments on the rest of the script would be appreciated too\nUnder what cases would the file system not be available?\nthe provider could be something other than filesystem, like the registry, certificate or environment you can see them if you run\njust tried , neat\nWe probably want to keep the CWD even on non-filesystem providers actually, just give it a different scheme or something in that case to let VS Code differentiate it from a regular path. That way we would still get the cwd folder showing up in the tab: !\nalso, is it only supposed to support powershell 7 and not win PS?\nYep, no windows powershell\ndoes vscode also supports osc 9;9 for CWD Tracking like win term?\nsome discussion on the topic in , my plan is to move us over to using our own custom sequences to avoid shell integration getting confused and give us freedom to change what them as we please.\nlet me know if I'm missing something though, I don't think running another terminal's shell integration script is particularly useful. Instead, you would use the arg injection or set it up manually (docs aren't written yet).\nI see you have the sequences we support in your profile, these would break when we action as we'd probably move to use\nyeah, I will probably remove them after the chages in and get into a release\nGood to hear, let us know if you hit any other problems as I have a pretty simple pwsh setup\nCan confirm latest build resolves this now on Inisiders: !"}
{"_id":"q-en-vscode-88aba33baf3a8e2390fe17e628a46c12f5200a4f0df4b45007e94524f217c74f","text":"I am doing Remote Development, and have a few Source Control Repositories. In the Source Control pane, I have all the repositories listed in the Source Control section. One of the repositories has a file that I want to discard. It is listed under \"Untracked Changes\", under the repository it is in. If I click the \"Discard Changes\" button on the line with the file, I am prompted for confirmation, as I expect. But, if I click the same button on the line above the file -- the \"Discard All Untracked Changes\" button on the \"Untracked Changes\" header line under the repository I want -- then I am first shown a \"Choose a repository\" dropdown that lists all the repositories. I don't understand why this choice is needed, since I clicked the \"Discard All Untracked Changes\" button on the Untracked Changes for a specific repository -- why must I select this repository again? If I choose the repository, I am still shown the same (or similar) confirmation dialog as I was before, so the selection does not seem to be needed for confirmation. Clicking the plus (\"+\") button (i.e., Stage All Untracked Changes) on the Untracked Changes line seems to have the same issue.\nThank you!"}
{"_id":"q-en-vscode-e9c4772c4b9ccd64683ac6284cfab524a7aba21065155e8c7f9335b38939d85b","text":"Issue Type: <bBug</bWhile doing some experimentation with Bootstrap and the corresponding scss files, I encountered an issue while changing hexcolor codes. I was able to reproduce it on a different Windows machine (Windows 11 in that case) as well as on macOS (Monterey). Steps to reproduce: Open a large scss file. In my case, I used from the Bootstrap framework. Start changing a color by typing a different hex value. The issue only happened when changing at least two of the six digits and sometimes only after making a couple of changes. It did not happen when using the color picker. will become unresponsive eventually. For example the color pickers do not show up anymore, saving a file takes multiple seconds and the git integration stops working entirely. When the builtin SCSS extension is disabled, everything works fine. Therefore I would assume that this extension is causing the issue. VS Code version: Code 1.63.2 (, 2021-12-15T09:40:02.816Z) OS version: WindowsNT x64 10.0. Restricted Mode: No <details<summarySystem Info</summary </detailsExtensions: none<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nI can reproduce open a scss file with on the end of line 15, add a character . Note that diagnostics appear, Remove the character, diagnostics disappear (all good) on the end of line 15, add a character . Nothing happens. Hover over a symbol, hover computes forever: Extension host is blocked remove the letter, reload the window disable the emmet extension repeat the steps, now all is working When profiling, the emmet extension is busy in The bug got introduced in 1.53\nThere are 3 conditions that must be met to trigger this issue: must return one of , , , , or (so that in returns true). File must be longer than 1000 lines (so that is called as noted above). The first must begin with The culprit is from When trying to skip over blocks, this function attempts to peak back 1 character when it encounters to check for a (1)()` is then called to move the position back to where it was at the start of the function, but in this case moves it back to position 1:2. This causes the loop that calls to loop infinitely, as the exit condition (position being to , in this case the start of the file) never occurs. Pull Request Resolves this by checking if the character being \"peaked\" at is a and breaking out of before is called. An alternative solution would be to not call if is within 3 characters of (as there is not enough space remaining to contain .\nThanks for the explanation I'll comment on the PR."}
{"_id":"q-en-vscode-bad459e80a375c0175f28d02c2a46ca371e129dfa30d1230c42e53966388b4bc","text":"Version: 1.65.0-insider (user setup) Commit: Date: 2022-02-11T05:19:13.554Z Electron: 16.0.8 Chromium: 96.0.4664.110 16.9.1 V8: 9.6.180.21-electron.0 OS: Windows_NT x64 10.0. Steps to Reproduce: an extension (such as the C/C++ extension) that uses platform-specific VSIX's in the marketplace. a Remote WSL workspace. Install that extension into the remote. the extension UI for the instance installed remotely, try to select \"Install Another Version...\" or \"Install Another Prerelease Version...\" This results in an error: ! (There is a similar error in latest non-insider version, 1.64.2)\nCan you please enable the log level to trace and provide following logs while performing this action? Log (Shared) Log (Window) Log (Remote)\nWhen this repro's, I see the following in (with and is set to . I do not see any such setting as ) : is similar: was not logged to for this repro.\nYou can use following command to set log level to trace\nLog (Shared): Log (Windows): Log (Remote Server): Log (Remote Extension Host): Please re-open. Please don't block waiting for me to provide information. This bug is extremely easy to reproduce.\nPlease re-open.\nTo verify: Open a remote window (make sure local and remote are on different Operating Systems) Install C++ extension Install another version of C++ extension (right click on the extension and you will see this action) Make sure another version is installed."}
{"_id":"q-en-vscode-c786d4380e1e093b5ed832bae33c3098eb1e3f06af522bdc15e587e6c146dae0","text":"Testing Linux some commands in your terminal 1 Terminal Clear Command History 1 Terminal run recent commands the commands are still there, nothing got cleared !\nThis is by design as clearing history only clears the previous session commands. We don't discard the current session as they're all in the active terminal marked by decorations. It does however clear the active terminal's current session history from the previous session history (which hides by default as they would duplicate the content).\nI find this very counterintuitive... As no other clear command in VS Code behaves that way.\nI cleared recently opened and the active windows still show up, this is essentially the same thing the terminal history is doing: !\nOn macOS, i run , and when i run again, i can see all the commands, including those from the previous session. <img width=\"802\" alt=\"Screen Shot 2022-02-24 at 11 07 24 AM\" src=\"https://user-\"\nI agree that it's a little confusing. Perhaps we should say instead?\nI now get the clear command history and a clear previous session command history command. Neither command seems to be clearing the command history. I'm on Windows with pwsh. !\nthe change you made was to the quickaccess code, this is nothing to do with the terminal so we need to revert that\ncan you help with verification steps for this one. Since I do not see the \"Clear Command History\" in quick pick. Does this only appear on Linux?\nthe title of the command is\nthanks. After running that commend I still see the recent commands. So the command history is not cleared. Is this expected?\nI still see previous session history commands given to the terminal after running the command.\nis the group called or is it or something similar? Only the first is expected to be cleared. This worked fine for me just now."}
{"_id":"q-en-vscode-5f6b0cd27423f2bb77929bc47380f132924b9f3cd6486e1673ae55f81f562bb7","text":"Can we add some additional padding to the top of the TOC, to match the container on the right? !\nGood idea. Assigning to March, though, because when I padding earlier, the end of the list got cut off, and I couldn't find a quick workaround.\nThis one's tough. A few attempts: Increasing the padding and doing nothing else results in the last entry being cut off. Increasing the padding and manually making the lists render at different heights results in the last entry showing, but the scrollbar not touching the bottom when the user scrolls to the bottom of the list. Creating a new element to stick above the scrollable list results in the shadow being pushed down. We want to keep the shadow at the top edge of the container. Adding a fake settings group with no settings to try and get the list to add a spacer element does nothing, because a settings group needs settings to show up in the ToC.\nhow did we get the padding to work before? Maybe we can ask the list widget to provide a padding param to adjust the height?\nBefore, we weren't using a splitview. The shadow was lower, and there was no sash. If we add padding elements above the splitview views right now, we will have the sash and the shadow intersect again in a + shape rather than a T shape.\nDo you want the padding inside the list view, or outside of it? If you want it inside the listview, you can add it to an element and include this in the size returned from . If you just put css padding inside the list by itself, that will mess the list up\nI would like the padding to be inside the listview. Adding the padding to the element results in a different focus border shape for that element.\nWhat if you use a margin-top on the first element instead of padding?\nI believe in this case, either the whole list gets pushed down and truncated at the end, or if I adjust the list height manually, the scrollbar fails to scroll all the way to the end when the list is also at the end. I also reopened an older version of Code and noticed that the list itself didn't actually have padding. There was padding above it, and the shadow was lower, as a result. !\nI wouldn't mind it if the list had that top margin, I still think the line connecting to the user/workspace heading gives the impression that the top is also resizable (like editor grids) so staying away from that as much as possible would be preferred: <img width=\"855\" alt=\"CleanShot 2022-03-14 at 14 03 38 src=\"https://user-\"<img width=\"851\" alt=\"CleanShot 2022-03-14 at 14 02 34 src=\"https://user-\"\nVerification steps: Settings editor should look closer to In other words, there should be a gap between the vertical splitview sash and the horizontal tabs/scope widget underline. one scrolls down on either side, horizontal shadows should appear below the horizontal underline.\nVerified at described, not sure it looks better now than before though because the shadow no longer aligns with the horizontal bar above."}
{"_id":"q-en-vscode-f13ae7622482da9c0b29e32d1a1e54ccc1bc904adf011d03d00196b1d9830ece","text":"Testing :bug: success terminal command decorations are still rendered !\nonly reproes sometimes - regression related to the change that was reverted in . We should figure out why zsh sends the sequence for the viewport so that can be re-.\nI bet I caused this in , need after a splice...\nUpstream fix merged, updating now...\nI'm having the same error on macOS.<img width=\"1346\" alt=\"Screen Shot 2022-02-24 at 9 44 26 AM\" src=\"https://user-\"\nStill seeing this: ! Version: 1.65.0-insider (user setup) Commit: Date: 2022-02-25T05:17:33.544Z Electron: 13.5.2 Chromium: 91.0.4472.164 14.16.0 V8: 9.1.269.39-electron.0 OS: Windows_NT x64 10.0. Repro steps new pwsh terminal :bug: see success decorations\nif you run , you should see them all cleared from the terminal. running in the terminal only clears the viewport.\nare you scrolling? a gif might be helpful because I cannot reproduce\n!\nYou might also be seeing\nOh--I thought running in the terminal was supposed to clear the decorations too. Seems a little weird that the command decorations would stick around--I would disable the decorations if that is the behavior moving forward since the dots become quite distracting: ! Terminal: Clear does clear out the decorations, btw.\nI don't think we can do anything about not working as expected in pwsh/windows. can we?\nconpty sends , should we be clearing markers whenever erase in display clears any line? Just not sure about when conpty reprints the screen, could delete all existing markers. If that doesn't work we could maybe check if the command was and handle it specially? Might be complicated though.\nI think it would be easy to handle by checking if the command entered was clear or similar (could even do this per shell type since the command name may differ). More complicated and potentially destructive to clear the markers whenever erase in display is called imo. We can test it though.\nReopening to cover cases like this specially"}
{"_id":"q-en-vscode-31da136399c7100f2c3fb5f0df3bedaec2728b76a5fd39ac2d8796db9a4dce06","text":"Testing We should consider increasing the line height for headers in order to provide spacing between lines. !\nPossibly related: and\ncc\nIt is also perhaps reasonable to suggest that the top margin is too large on the first header so you end up with more space at the top of the markdown block than at the bottom and it is not possible to balance if you want a header at the top. It might be worth considering some styles that remove the margin-top from headers when inside a ?"}
{"_id":"q-en-vscode-42d827d5f09e9a6ab3d21d947a089428bcc257b8aac955f651e0b6ee8b3dd002","text":"Issue Type: <bBug</bRunning tasks with dependencies in a multi-root workspace executes the wrong dependency tasks, if there are multiple tasks defined under the same label but in different root folders. VSCode will just pick the first occurance in the order of the root folders defined in the workspace settings. To reproduce the issue: a simple multi-root workspace, e.g. each root folder, add a with two tasks. One task depends on the other. Example tasks of : and have the same tasks configured except that they echo their respective folder name instead. task of either or . VS Code version: Code 1.65.1 (, 2022-03-08T02:06:27.846Z) OS version: WindowsNT x64 10.0. Restricted Mode: No <details<summarySystem Info</summary </detailsExtensions: none<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nExperiencing same issue on: VS Code version: 1.65.1 OS version: macOS 12.2.1 BuildVersion:21D62\nTo verify: or create a multi root workspace with 2 roots, FolderOne and FolderTwo (or names of your choice). each of your roots add a .vscode folder. each of your .vscode folders add a file with the following contents: \"PrimaryTask\" in FolderOne (command \"Tasks: Run Task\" and choose the task with the label \"PrimaryTask\" that says \"FolderOne\" next to it). Verify that both PrimaryTask and DependsTask echo FolderOne's path. \"PrimaryTask\" in Folder Two (command \"Tasks: Run Task\" and choose the task with the label \"PrimaryTask\" that says \"FolderTwo\" next to it). Verify that both PrimaryTask and DependsTask echo FolderTwo's path."}
{"_id":"q-en-vscode-9bd378496c491096d3a9a565c003ea0a271c284af8b4af76cbc6f6be1e8de56c","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.65.0 OS Version: Windows 11 Enterprise Steps to Reproduce: a remote window, create an untitled file and paste these contents. ! this untitled file, a save prompt (modal) will appear. Click Save button. You will see something like this: ! suggested file name is $schema: and only azuremlschemas.azureedg is highlighted and not the whole file name which was suggested This works fine on a local window as the suggested file name is azuremlschemas.azureedg: as shown below: !\nFor untitled files, the first line is used as file name. I am not sure why there would be any difference between the simple and native dialog but I actually think the simple one behaves correctly here, though there seems to be mentioned highlighting issue.\nI think our main concern is the highlight, for the name we would ideally like to have custom inputs from the extension but that is a completely different discussion.\nAll the simple file dialog knows is that there is a default save URI given to it. The highlight is done by using the basename of that default URI. I took a look at where the default save URI comes from, and it looks like suffers from the same issue of using the basename that simple file dialog does: The best fix seems to be to fix the default save URI. PR incoming."}
{"_id":"q-en-vscode-a8957896718758e5e9b2a63e792a4344051237f7e0126f29b57cd79cf6e3a074","text":"This tracks the issue discovered in See for this discussion as well. We originally thought that issue had the same root causes as this one, but investigation revealed that was not the case\nI've debugged this in an OSS build. It's a difficult consistently hitting the original issue from but I do some really odd behavior around view state being saved and restored Here's what I was doing: active PR view the views window back to active PR view once it is loaded After step 2, I see the following data being written for the views in : ! However when the state is then restored by , I see the following data being loaded (look at the sizes specifically): ! Will try to track down if some other code is writing this incorrect size data or if the reload is somehow restoring it incorrectly on reload\nAdding and to see if you have any ideas about what's going on here. I've been debugging through the storage calls (at least on the renderer side) and we seem be writing the values as expected but on reload we still get back the incorrect values\nReopening this as I'm still seeing this issue on\nAfter debugging further, here's what is happening: Basics: The GitHub PR view contains 3 views (from top to bottom): , , On reload with a PR, all 3 of these views start as visible. The PR extension then hides the view by updating a context key Bug: window view is created first After finishes, we have the sizes: we add After finishes, we have the sizes: we add The sizes are now now call for the various views (in a loop) Here however, we have very large saved view size for (1012) This means that when we do for the welcome view, the other two views to become quite small. In my testing after finishes here, the sizes of the three views are: then remove the view (I believe this is triggered by the PR extension changing the view's context key) causes the last view () to fill in the now empty space below it complete this time, the sizes of the two remaining views are: So in summary: the view has a very large saved view size (presumably from initial load) and this large size effects the sizes of the other two views even though the welcome view is immediately hidden However I don't know what the proper fix for this. Should we try saving off the fact that the welcome view has been conditionally hidden for example? Should we try saving a more accurate size for the welcome view? I think I've made about as much progress as I reasonable can in this space without any real understand of the view code. Can you please take this issue from here?\nThank you for looking into this and tracking it down .\nSure - thanks for update\nTo be clear, is this the bug we are planning to fix? (After refresh the bottom view is taking more space) https://user-\nYes, that's the one.\nI can reproduce this behaviour with following sample extension <details<summaryExtension</summary</detailshttps://user- analysis is correct In the beginning first 2 views are visible and are sized equal. After 5 seconds, third view comes in and sizes are still distributed. After 2 seconds, third view is made hidden by setting the context. At this point, last view is removed and its space is allocated to the view (second view) before last view. First view takes 33% and Second view takes 67% of the total size. Instead the removed size shall be equally distributed to all visible views. Used DistributeSizing option while removing the pane and this seems to fix it. https://user- CC - I do not have much experience in using split views and do not know what are the side affects of using option. So please take a look and update/enhance the fix if needed.\nGreat analysis! the docs say: ! Which means: When using in , all other views will be resized to distribute the total splitview size. Without that option, the regular layout algorithm will happen... usually that means that only the adjacent view will stretch.\nI am not sure if the following doc says the same or I am wrong: As per the doc, I was expecting that the removed view space is distributed equally across other views. But it is not and it is doing as you mentioned. This also explains why the views sizes are getting reset to equal sizes after a view is removed (when DistributeSizing option was used).\nThanks for calling that out! I'm improved the docs there.\nGiven that it is not doing what I am expecting - removed view space is distributed equally across other views. Is there an other option to achieve this?\nNot today, no.\nIt means the above fix has the side affect that views sizes get reset when a view is removed - which I think is ok. CC\nFirst off, thank you all for your investigations. however, I'm not sure this is a viable solution. Any time a view is removed this will resize all views which can feel quite wrong. See the gif below where we hide the outline view. The only other expanded view is quite small and becomes now half of my screen. The same can happen with drag and drop of views as well. !\nI agree the new behaviour is not ideal but it is acceptable when compared to existing. Probably we can improvise it by not to distribute space when the view is collapsed? BTW I also observe the same in editor groups. https://user-\nI've approved the PR. I think the change we have made is a band-aid that helps, but doesn't cut to the root of the problem. I thought through various view/hide scenarios and in each case, I had a different feeling about what I expected to happen with the sizing of the views. IMO, our previous behavior was more correct from the pane view side as an intentional \"hide\" operation on a view should not redistribute every view in the container. However, the problem here is that there is a problematic extra view that is not controlled by the user. In this case, we are just trying to get sane sizing. If we invest more time in this, I think the fix lives at a higher layer."}
{"_id":"q-en-vscode-583dd539b5b11af766dba3b77584ce93a54ffb3713b7ed127e1341d28f8d26d8","text":"<!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.66.0-insider OS Version: MacOS 12.3 In iterm2: <img width=\"810\" alt=\"image\" src=\"https://user-\"In VSCode terminal when enable shellIntegration: ! If I disable shellIntegration, everything goes ok: !\nrelated to\nI believe this is already fixed, can you see if you can reproduce in the ? /\"needs-more-info\"\nNot quite yet... this is why I personally have shell integration disabled btw. !\nThis is working for me ! Here's what I did ! from\nLooks like suggestion text does show up, but it uses a different history file. So it will suggest entries you have made with shell integration enabled, but not those made in normal terminals.\nshould it be using the same history file?\nis this fine now because we revert to the original? We definitely want to share the zsh history file.\nit's using the history from within VS Code only - when should it get refreshed and also suggest things from the broad shell history? tried reloading the window <img width=\"464\" alt=\"Screen Shot 2022-06-30 at 11 37 34 AM\" src=\"https://user-\"\nalso does not pull from history of terminals that did not have shell integration enabled\nit depends when zsh flushes history to , not sure if it's when the shell closes or a command is run.\nhmm history is correct for terminals without shell integration, but not in ones w shell integration - even when I reload the window and create a new one <img width=\"971\" alt=\"Screen Shot 2022-06-30 at 12 37 19 PM\" src=\"https://user-\"\nwhat's set to?\n<img width=\"978\" alt=\"Screen Shot 2022-06-30 at 12 45 40 PM\" src=\"https://user-\"\nI'm guessing we need to fix up as well to point at ?\nwould you think my error is related to the above discussion? (Cc <img width=\"847\" alt=\"image\" src=\"https://user-\"After a my prompt looks normal again: <img width=\"846\" alt=\"image\" src=\"https://user-\"I believe this usually happens when vscode is updated ...? EDIT: I just found a pre-existing issue in the mcfly repo .\nwe expect this to be fixed in the upcoming version, see Sorry about the hassle, please let us know if it's not fixed in 1.71"}
{"_id":"q-en-vscode-2c52c38fd994de9932033ebc2ac82d52c3af3e7f45eeaa87ffa1417a9bcd9d03","text":"Currently says that the coverage folder is generated at the root of the directory. But this isn't currently the case, the coverage folder is generated under the directory instead."}
{"_id":"q-en-vscode-2374e186b3ab9ce34fe972756c9bac0b56d9a930dd144ec68c1b69f6fbebeac8","text":"Enable triggering go to definition on a markdown reference link to jump to the link definition Go to def should jump to the definition at the end of the file\nRelated\nCreated a PR fixing this\nWorks well, I was a bit confused to see this tooltip with 'Execute command' though, is there a way to make that more context-specific? !"}
{"_id":"q-en-vscode-7f1209f9618d72eb9e6931e3d535ebf3903fe4af3ec5cd928a06f0d39c29af9a","text":"<!-- Do Not Delete This! bugreporttemplate --Version: 1.66.0 (system setup) Commit: Date: 2022-03-30T05:50:14.623Z Electron: 17.2.0 Chromium: 98.0.4758.109 16.13.0 V8: 9.8.177.11-electron.0 OS: Windows_NT x64 10.0. New \"striped\" gutter decorations were in . These decorations do not respect the existing SCM: Diff Decorations Gutter Width setting: Changing the above setting has no effect on the new decorations.\nI think this was an oversight and have a PR that can help address this. Thanks!"}
{"_id":"q-en-vscode-9e67743279d5bafec61bedb6c9e65a9b308f957f004bd9e789967cab09a6f8fe","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.66 OS Version: ArchLinux Shell integration doesn't work with oh-my-zsh and powerlevel10k. The indicator is only shown for the first line and the disappear. Another problem is the warning from powerlevel10k about instant prompt (see attached screenshot) Steps to Reproduce: zsh oh-my-zsh and powerlevel10k theme vscode and start new zsh terminal to run some commands Screenshot: !\nHaving the same issue.\nRefs Out of curiosity, P10K explicitly disables the integration (as it does not work anyway and yet makes the prompt slower). So, if someone tries to debug this situation, make sure to comment out the following line before testing: P10k is extensively used (has almost 30k stars), so it would be very nice if VS Code can manage to make the Shell Integration work with it.\nWe'd like to support it, the summary of the discussion/situation around this was: We're using custom sequences to control everything and extend the shell integration. Since we launch the process we can guarantee we're in VS Code (except for edge cases). VS Code's shell integration will eventually be enabled by default, it works by wrapping your PS1, PROMPTCOMMAND and eventually . In order to have the most impact, we aim to support most_ shells so more users get it by default, with a manual setup option for complex shells and sub shells. This will come in the form of essentially running the script manually in your rc file and disabling the automatic injection. Powerlevel10k is complex, it handles shell integration used by other terminals manually. I don't blame for not being keen on our solution if it means he needs to implement a VS Code-specific solution. After a bunch of back and forth I recommended we all take a step back, disable it explicitly for now and let us stabilize more so we don't rush into something we will regret. We will probably at some point document the sequences (it's still getting stabilized and subject to change) or maybe support the more common sequences which would let it just work without additional changes. My main concern with supporting multiple shell integration protocols is them conflicting with each other, but that's probably the right thing to do and we could have a switch which disables the other protocol.\nHaven't tested this much but this seems promising after some trivial changes: ! Not sure yet why we're not getting the cwd sequence, I'm forcing shell integration on via\nAnother problem is there doesn't seem to be a nice way to extract right prompt or continuations so this ends up happening: <img width=\"486\" alt=\"Screen Shot 2022-06-10 at 1 41 39 pm\" src=\"https://user-\"We could make some guesses by examining the content of the cells but that's not ideal. It would be a shame if we lost the fairly reliable extraction of the command since a lot of the interesting future functionality will be built on that like extension APIs and autocomplete.\nIn general in zsh it's impossible to extract the command from the data written to the terminal. The command isn't always visible. For example, try entering a command that's taller than the height of the terminal (use ESC Enter or Alt-Enter to insert a line break). Or try running and entering a command that's longer than the width of the terminal. If a terminal needs to know the command being executed, it'll have to provide an extension so that zsh can supply the command explicitly.\nI'm willing to implement a VS Code-specific solution but I would need you to provide a spec of VS Code terminal extensions and promise backward compatibility.\nThe feature that exposes the commands is not particularly discoverable but it works pretty great for the general case in all shells at the moment, I know there are edge cases when it will fall apart though. For pwsh the column and line are not reliable because of conpty so we do pass through the command line there and it works well: ! Great, I wouldn't want you to without us committing to it . We're still not at the stage where reliability will become more important so we're just passively thinking about this problem atm. I'll let you know if we'd like you to do that. For now we can just disable the parts that need the command which will still give us the markers/command navigation. I have a couple of questions that would save me a bunch of time if you know: I see the window icon/title is coming through in and , is there a trick to getting the cwd to be sent via something like , or ? Is the right way to test against?\nSounds good. Powerlevel10k doesn't set terminal title and doesn't report the current directory. These things aren't related to prompt and can be done separately. Powerlevel10k provides only prompt. The best way is to set . No need to export it. is normally set by the native integration of iTerm2. It doesn't work with Powerlevel10k, so Powerlevel10k undoes that and enables its own integration.\nI put a PR out which should make the upcoming release With it, shell integration should work with powerlevel10k if it emits the more common standard OSC 133 A, B, C, D sequences. This won't give all the features but it will turn on command decorations with exit status, command navigation (ctrl/cmd+up/down) and the overview ruler. if you could emit the OSC sequences for VS Code that would be great, checking $TERM_PROGRAM = vscode should be a reliable way of doing that. I'll let you know if we end up committing and documenting the sequences we use, in which case a command line sequence could light up more features.\nAwesome! Have you tried it? Powerlevel10k emits OSC 133 if you set . It's a good idea to verify that it works with , too. Powerlevel10k emits OSC 133 if users explicitly request it with , or if they activate the built-in shell integration in their terminals. Powerlevel10k detects the latter, undoes the damage if necessary, and turns on its own OSC 133 implementation. Currently this is implemented for iTerm2 and Kitty and I could implement the same logic for vscode: If is set when powerlevel10k is sourced, unset it and enable OSC 133. Does this sound reasonable?\nI installed the stable build of powerlevel10k and never seems to work, turns on OSC 133 though. when the one is set works fine. That's perfect . We have started using that variable to prevent recursing or calling the scripts multiple times, ie. so it works when it's injected by VS Code and installed manually:"}
{"_id":"q-en-vscode-79e6d23efc88e535558f833e5906fae6a1273f419d6f969551118b9f82295007","text":"When I press Ctrl+Shift+P to open the command panel and type \"\", it shows \"No matching commands\". So I can't execute this command by Ctrl+Shift+P. I can only bind a shortcut key to this command and then execute it via the shortcut key. I think this is a bug that the command panel search is incomplete.\nPlease edit this issue title to become:\nindicates it may already be there and you just need to know what the command text is.\nOK, but according to my observation, there are other commands that cannot be found in the panel. For example, \"\".\nThe issue I raised is different from him. His issue is that the shortcut key does not work after binding. My issue is that I can't find the command in the command panel. In fact, some other commands cannot be found, indicating that the search of the command panel is incomplete.\nSome commands never appear in the Command Palette because they define their clause as\nI open the page where I set the keyboard shortcut and see that their when clause is - not false.\nSorry, I wasn't clear. This is what I am referring to.\nI read your link and I probably understand. Please set the when clause of \"\" to \"inDebugMode\" instead of false. This is more in line with our needs.\nLooks like that command is missing from here: is also missing.\n/confirm\nI will submit a PR\nThanks!"}
{"_id":"q-en-vscode-c021869ca42fce9796f88996a4b544e2e67f3ee3f1c26d5ccced711f34b9377c","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.66 OS Version: Windows 11 Steps to Reproduce: diff view to revert the output changes :bug: after reverting the diff view still says there are deletions/additions output changes using the diff view :bug: stage the file and reopen the diff view, all the changes I reverted are back"}
{"_id":"q-en-vscode-3be58d6325f90bbeb2c0404b05041e953264c8b18c3504bfc462ea0c4eaa5eec","text":"Ref: I would expect the comment editor to behave just like a regular editor window, but in my case it is not. I don't hear screen reader feedback as I move through words and lines with the keyboard. Also, when I press a routing cursor key on my braille display, it does not move the cursor to the desired character. Tested with NVDA and JAWS. Invoke the action to leave a comment in the current line Write some text Use ctrl+left and ctrl+right to move through words Use up and down arrow keys to move through lines If you have a braille display with routing keys, press a routing cursor key to move to a specific character in the text"}
{"_id":"q-en-vscode-a66819ebd8bca345298a150f1953f886bd9b5a6249bb1d3b584fd9a645af6014","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: No <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.66.1, 1.66.2 OS Version: Ubuntu 20.04.4 So the issue I'm having is as an extension developer (). I've had users report an issue that I have tracked back down to vscode - I can confirm the issue on the above versions, but don't know the full extent on other os versions. () The error being received is: \"Input is not valid base64-encoded data.\" I've tracked this down to getting the secret from secret storage (). I use the SecretStorage api to securely store credentials used by my extension. The api docs don't seem to indicate a change - but something appears to have changed (whether by design and the docs need to change, or by introducing a bug). The issue does not appear to affect vscode 1.66.0 - reverting to that version allows it to work. I do not know whether it affects whether it affects secrets - but can confirm on secrets. Steps to Reproduce (with referenced extension - so far): In vscode 1.66.0 - with extension, create a db connection with pwd to store creds (without pwd won't use store creds so it won't use the SecretStorage) Upgrade vscode to view the connection explorer and it will error when attempting to pull the secret from storage.\ncan you help me triage this. Not sure who owns it.\nwas the above steps tested with snap build or Debian build ?\nSnap\nThe problem is that encryptionService was not working with snaps before as seen in , with we made some changes to our native dependency that now enabled the encryption service for snaps as well. So previously unencrypted data are thrown as error. Migration: 1) At the encryption service layer, encrypt before decrypting for snaps 2) At the extension level, store the secrets again with before retrieving it (not sure if this is possible unless the extension could retrieve the stored data outside the service)\nOk, that may explain the issue, and I can confirm this is not broken on Windows. Is there a way in the extension to detect installs via snap?\nI have a fix^^ out for this. It will be included in 1.67 which should be released early next month. If you want to workaround this, first catch the SecretStorage Error and do one of two things: Ask the user for the thing you stored in SecretStorage again. Easiest to implement... inconvenient for the user. Temporarily (just for 1.66) include in your project and read the secret directly using the service will be: (exactly) and then the account is whatever your key is. Once you get that secret value, store it back using the API to prevent future errors. Very inconvenient for you, but the user doesn't have to do anything. Personally, I recommend the first one, because the second one can be very tricky... and you probably generally wanna be robust if getting the secret fails (like if the user locks their keyring for some reason... keyrings are very tricky).\nVerification (on Linux): install stable Snap v1.66.2 Log in to GitHub for settings sync or extension Close vscode Update to latest snap (v1.67.0) When you launch VS Code, you should still be logged in to GitHub\nthose steps didn't work for me, I was signed after installing that snap <img width=\"979\" alt=\"image\" src=\"https://user-\"\nupdated to a newer stable build liked above. Can you share whats in GitHub Authentication Output pane?\ndoesn't seem to work for Will need to investigate further but don't think this needs to be a candidate.\nMaybe this revealed other issue - but I can confirm with 1.67.0, my extension appears to be working again.\nand now I am seeing (in 1.67.1) a new error \"Cannot get password\"\nCan confirm the issue from Github Authentication log fails with the error\nOk I have a fix for this. Unfortunately due to the nature of this fix, I was unable to migrate secrets from older versions forward... but at least going forward, we shouldn't see any issues where secrets can't be decrypted after an update.\nVerification is weird but here it is: On Linux remove all VS Code stables snaps or debs Now install the latest unreleased main stable candidate snap (Like on the builds page) - Open it, log in to anything (GH/MSFT for settings sync for example) Close vscode Now install the stable build that is built on my branch () (all this branch does is revs the version to 1.69) Open it, and verify that your login persisted across versions\nOk, just got the update to 1.68 - this is still an issue. I'm not familiar with your merge process, but it looks like the PR that was linked here for excluding from the snap package was closed without merging. Is this still coming? It's almost been 2 months since my extension has been working properly for Snap users.\nmaybe my message: wasn't clear. You will see this issue in the 1.68 update. However, going forward, you shouldn't see it. 1.69 should not have this problem.\nThanks, so right now the password hasn't gotten migrated, and right now we are getting the , if I understand correctly in 1.69, this issue should disappear - will it still attempt to decrypt the password and then error, or will it just starting working. If it's going to error, what should I tell the users of my extension? I would assume that the password would have to be re-entered and then it would get stored in the new way and start working.\nAh that's a good point... I did not clear the password after the fact. I should do this. That password is essentially dead at this point so it should be cleaned up. Yes that's correct. If you overwrite the password using the SecretStorage API you should be good to go. I think this is generally a good practice to handle when SecretStorage doesn't return an expected result because there are a lot of layers all the way down to the OS that need to work."}
{"_id":"q-en-vscode-61980b33ce2f39df07530889b2ed7c9970a45a4f13bda1df1700e9374318240e","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.66.2 OS Version: Ubuntu Studio Xfce4 Linux x64 5.13.0-19-lowlatensy Steps to Reproduce: to next line: VS Code from the Applications Menu or double click the associated file error message: ! Upd: If I comment this line VS Code runs without error.\nDo you see any errors in you for User and Workspace?\nno, I don't see any error in my .\nWhat are you trying to do with that line?\nthis line changes the title of my terminal window to the currently running command, so that I can see in the taskbar which command currently running in each terminal window. but it shouldn't matter to . and a few versions ago it was. I have been using this line since very beginning of using Linux. and didn't throw this error until recently.\nI that line to my and I don't see that error when VS Code launches, we source a \"dev environment\" by running your default shell. When this fails we show the error message that your shell returned. You have some issue in your shell init files that's causing that error, you need to investigate on your end as it's a problem with your environment, not with VS Code. See here for what we do before showing this error:\nI'd disagree with this being an environment problem and I'm not sure if should be closed. Unable to resolve your shell environment: Unexpected number in JSON at position 35 I'm seeing the exact same problem as the OP in 1.68.0 (can't say when it started, skipped a few versions) and it's triggering over a custom git prompt in bash on Ubuntu that I've been using fine in all sorts of environments since forever. Happy to provide more info to debug this if that helps and if you'd provide some guidance what to do. Also: if there is an issue in a user's shell configuration, wouldn't it be useful to make the error message a bit more user friendly? As a suggestion, it'd be good if it explained that and maybe render the actual command VS Code tried to run and that failed? Or some more information in VS Code logs?\nThis is a better repro for me from To repro in code-oss, add that to .bashrc and run (or debug Main Process target) after commenting out this line:\nThe exception happens here: '\" + JSON.stringify() + \"\"'\u0007{\"key\":\"value\",...}' This is what rawStripped is: <img width=\"622\" alt=\"Screenshot 2022-12-02 at 10 19 27 am\" src=\"https://user-\"Here's what it is when it works:\nAnd the raw when it doesn't work: <img width=\"1141\" alt=\"Screenshot 2022-12-02 at 10 23 34 am\" src=\"https://user-\"The 2 marks are in the sequence, we should be matching on the last 2 marks, not the first 2."}
{"_id":"q-en-vscode-a7071cbdb722acbd380ecd81657fc7e879141078efe319c80bfed5c86aac5979","text":"The issue is that when I resize the page to see it along with other content, I can see some text over the scroll page area. I think it wasn't supposed to work like this, that's why i'm reporting it here. ! ! .\nThe scrollbar is intentionally transparent. We have a color that is used only when the minimap is enabled. We could potentially use it always if set to allow defining opaque scrollbars.\nOh. Thank you for explaining :) In this case, sorry for bothering! Best, Alvaro\nTo verify, configure the color and that will be applied and will make the overview ruler opaque. e.g."}
{"_id":"q-en-vscode-cee813e73a4b3b435e6060d9e76bc4552ee0f3dbbed4913c5e3c4baa1429c6b3","text":"Import following settings profile Timeline and Outline views are not updated. Refreshing the window updates the views. Expected that Timeline and Outline views shall be moved together into sidebar as a separate viewlet\nReopening this and making this a dup of because this is not completely fixed.\n/duplicate"}
{"_id":"q-en-vscode-324bc9310456c0ec356b89d1a6c28e49d4d96adfe6ecbd074215b4a4e777d239","text":"Issue Type: <bBug</bany command line run or pipe another process that runs indefinitely to that a window does not open and the command does not return. The file should open and update as new data is piped to the code cli command. This works on the Mac when you run but does not work on Windows. This is not a 1.67.0 regression VS Code version: Code 1.67.0 (, 2022-05-04T12:06:02.889Z) OS version: Windows_NT x64 10.0. Restricted Mode: No <!-- generated by issue reporter --\nThis is probably related to the file watcher not recognizing changes to the file on Windows, have to debug why that is the case but might be a limitation of the single file watcher.\nThe underlying reason seems to be that writing to a file via stream API where the file handle is not closed does not trigger file events on Windows at least. Maybe as a workaround we have to use the less efficient method that opens a file handle each time and closes it.\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!\nI am still experiencing this issue on the version of insiders below. Version: 1.69.0-insider (user setup) Commit: Date: 2022-06-24T06:32:08.615Z (13 hrs ago) Electron: 18.3.4 Chromium: 100.0.4896.160 16.13.2 V8: 10.0.139.17-electron.0 OS: Windows_NT x64 10.0.\nHere is how to verify: ensure you are on latest insiders from a terminal on Windows run verify output is updated\nI tried on the latest insiders with that command and vscode-insiders still does not not open for me. If I run it hangs and never opens vscode-insiders If I run without the -t flag, it will open after the command completes. Version: 1.69.0-insider (user setup) Commit: Date: 2022-06-27T07:59:47.622Z (10 hrs ago) Electron: 18.3.5 Chromium: 100.0.4896.160 16.13.2 V8: 10.0.139.17-electron.0 OS: Windows_NT x64 10.0.\nThis works for me on Windows 11 with both Powershell and , but may need further investigation"}
{"_id":"q-en-vscode-beef0b444286bffe392a21679eab53e28f8098c14872a7753555b226844fccf7","text":"Issue Type: <bBug</bthe editor language to C++ to to /paste the Sample code below the editor window so that it is narrower than the line of code provided that the horizontal bracket underline stops at the end of the first line, when the (soft-wrapped) line break occurs. to that the horizontal bracket underline now continues onto the wrapped second line. VS Code version: Code 1.67.0 (, 2022-05-04T12:08:42.978Z) OS version: Linux x64 5.17.4-200.fc35.x8664 Restricted Mode: No <details<summarySystem Info</summary </detailsExtensions: none <!-- generated by issue reporter --\n! !\nUpdate: Note that this only occurs if the wrapped line starts in the left-most column. Indented, wrapped lines are not affected, regardless of setting.\nThanks for reporting this! I think it shouldn't be too hard to fix. If you like, we'd be happy for a PR that fixes this!\nHi, I found the snippet that produces the bug: The if statement controls if the first character of the wrapped line is in column 0. If so, then it ignores the guides for that line. By simply removing the if statement the result is the following: ! It seems that this solves the bug. Given my limited knowledge of the codebase, before opening a pull request, it would be better to clarify the purpose of the if statement as it is not fully clear to me. Should it suppress possible vertical bracket guides in case of wrapped lines? Was it to address a specific bug?\nI don't know anymore. Since unfortunately I didn't link that comment against any issue nor git blame helps here, lets just remove it and see if someone complains.\nI think I've solved it. You the comment in (\"Implement line scopes.\"), which was a month before the implementation of horizontal guides (in ). So, at the time, it made sense, but horizontal guides invalidated the logic the check was based on.\n(Oh, and nice find, Thanks for finding and fixing this.)"}
{"_id":"q-en-vscode-7cf00f1c2a8c807a6045c22974684ccb4f830e3b8df72539d110c1f6f808ed16","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --<!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --In Visual Studio Code 1.65 and earlier, icons by extensions to the debug toolbar were dimmed when disabled: ! But in Visual Studio Code 1.67.0, icons by extensions to the debug toolbar are no longer dimmed when disabled: ! I am working on an extension which adds icons to the debug toolbar, using the . The icons are disabled while the program is running, and enabled when it is stopped. A typical icon definition looks like this: with a corresponding definition in the : It seems that Visual Studio Code 1.67 no longer provides a way for extensions to vary the colour of toolbar icons when they are disabled. Looking at the Visual Studio Code source, it seems that the built-in toolbar icons are created like this: and then styled like this: but I cannot see any way for extensions to access this functionality.\nI see this, hey this is from . Seems like this will work for codicons but not svgs. Any idea how to deal with that? We could add a new class to svg icons so we can keep the opacity rule, if that helps. This probably applies to all toolbars, besides the debug one.\nIn the short term, restoring the opacity rule for SVG icons makes the most sense. But in the medium term, it would be nice if extensions had a way to access codicon-like behaviour, perhaps via the . At the moment there is no way for extensions to vary the colour for icons defined via this contribution point, but built-in icons use in their SVG, with the colour being specified via CSS.\nThanks for catching. Raised a PR to revert back to a slightly updated opacity rule for the moment. can you clarify what you're hoping to achieve re: codicon-like behavior? Are you hoping to use theme tokens in your contributed SVGs?\nLet's take this as a candidate. Could you also prepare a PR against the release branch?\nThanks for the super-quick fix! Look forward to trying it out in 1.67.2 (I assume). Concretely, what I would like to achieve is to be able to put icons in the debug toolbar that look like this when the program is stopped: ! and like this when the program is running (where grayed-out icons represent commands that are temporarily disabled): ! In these screenshots, six of the commands are using bulit-in icons (Continue/Pause, Step Over, Step Into, Step Out, Restart and Stop), and the other eleven commands are using icons defined by the extension. I would like the two sets of icons to match in style if possible. If I can choose colours from the user's theme, that would be perfect, but I don't mind exactly how the functionality is provided: you know better than me what implementation would make the most sense."}
{"_id":"q-en-vscode-34b64e22be19ba50ab3ad497d63f023bf572ea9c56125ffa1ac70253d1817f37","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --<!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --Same issue as , the tag is being to the repository again. Breaking automated tools like Renovate Bot thinking that there is a new version to update. <img width=\"934\" alt=\"image\" src=\"https://user-\"\nIs there a way a GitHub action could help identify who pushed the tag and delete the tag automatically? /cc\nWe can id the tag and ping the creator, we don't give our Actions git write access so can't go and delete the tag automatically.\nDeleted tag manually, action to ping the pusher upon pushing it back to the repo."}
{"_id":"q-en-vscode-8869e5cc569a1f1422fd52bf7b643fca4d388156c8a654f288621422d9421c45","text":"Issue Type: <bBug</brun in the terminal where is the path to an existing directory see VSCode open the workspace as expected run in the terminal where is the path to an existing directory focus changes back to VSCode, but showing the window for project instead of project Here is a movie of the experience: ! In previous versions of VSCode, when opening a new project with , the new window would show up above any existing windows. This is what I expected to happen. VS Code version: Code 1.67.1 (, 2022-05-06T12:37:14.619Z) OS version: Darwin arm64 20.6.0 Restricted Mode: No <details<summarySystem Info</summary </details<details<summaryExtensions (17)</summaryExtensionVersion ------ vscode-tailwindcss0.8.3 vscode-eslint2.2.2 copilot1.19.5865 go0.33.0 terraform2.22.0 vscode-env0.1.0 jsonnet-language1.0.0 vscode-docker1.22.0 python2022.6.2 vscode-pylance2022.5.1 remote-containers0.234.0 remote-ssh0.80.0 remote-ssh-edit0.80.0 excalidraw-editor2.0.16 code-spell-checker2.1.11 svelte-vscode105.16.0 cython0.1.0 </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nLooks similar to and .\nSame issue. On Mac OS, VsCode 1.67 How to reproduce: shortcut to open new window <kbdCMD</kbd+ <kbdSHIFT</kbd+ <kbdN</kbdshortcut to open new file/folder <kbdCMD</kbd+ <kbdO</kbdThe selected file/folder is now opened in the previously focused vscode window instead of the newly opened one. To solve the problem, you have to think to click in the newly opened window. When you do so, you see the focus takes effect, proving the window was not actually focused (EDIT: What is confusing is that you can see the cursor blinking inside of the editor of the new window, even though it is not focused) I can tell this is a relatively new issue (1 or 2 months max), because I'm doing the procedure above almost every day and it has been pissing me off since then (sorry, but truly annoying).\nSounds a bit like or\nDo you have configured?\nThis is a regression from\nThanks for the fix!"}
{"_id":"q-en-vscode-78bc4bfb2ae99b582128a6c87faff209bdd91ffbb4a048dc4c82f0231097578d","text":"https://user- Steps to Reproduce: the panel, see the comments tab the comments tab, reload the window :bug: comments panel keeps re-appearing Version: 1.68.0-insider (Universal) Commit: Date: 2022-05-13T17:25:40.854Z Electron: 17.4.3 Chromium: 98.0.4758.141 16.13.0 V8: 9.8.177.13-electron.0 OS: Darwin arm64 21.4.0\ndoes the file you have open have a comment in it? We show the comments view the first time you open a file that contains comments (once per reload).\nno comments, regardless of what workspace I am in it always appears (even in an empty workspace) <img width=\"475\" alt=\"CleanShot 2022-05-31 at 08 10 17 src=\"https://user-\"\nI'm not sure what's going on here, maybe you know. The Comments view is actually registered with , so I'm surprised it's even possible to hide the view. Is there anything else I need to do to prevent it from hiding or, even better, to respect the hide after reload?\nThe action takes is hiding the View Container so that's why it is hideable. There isn't anything that you can set on the View Container to prevent the hiding. I am also at a loss. can you \"Export Settings as a Profile\"? the profile should have the stored information about your views and view containers\nhmm I don't see anything alarming in there, but I can now reproduce it ... but I don't get it in an empty workspace\nis there any logging that can be in views to help narrow this down? I don't think there's anything else I can do from comments.\nafter lots of debugging, I think I have a fairly simple fix in the PR above."}
{"_id":"q-en-vscode-7f64a7c38aa83e5c484a8497d98377a88dab3ac5dd622e8098065a8dfe7dafc2","text":"This is a continuation of A unit test shouldn't take longer than 50ms. An easy way to \"speed up\" setTimeout and friends is -utility To reproduce these numbers run open the file and remove the first non-json lines run[x] 51ms, MarkdownRenderer Code block renderer asyncRenderCallback should not be invoked if result is immediately disposed [x] 104ms, MarkdownRenderer Code block renderer asyncRenderCallback should not be invoked if dispose is called before code block is rendered [x] 55ms, ExtHostAuthentication Can get multiple sessions (with different scopes) in one extension [x] 55ms, ExtHostAuthentication Can get multiple sessions (from different providers) in one extension [x] 60ms, ExtHostAuthentication clearSessionPreference - true [x] 72ms, ExtHostAuthentication silently getting session should return a session (if any) regardless of preference - [x] 93ms, Editor Controller issue : Greater than 1000+ multi cursor typing replacement text appears inverted, lines begin to drop off selection [x] 51ms, Notebook Find Update find matches basics 3 [x] 53ms, Notebook Find Update find matches basics [x] 51ms, Notebook Find Update find matches,\nThanks - for fixing tests in UserDataAutoSyncService\nMade IndexedDB tests faster but not fast enough to finish within less than 50ms. Because the bottle neck is IndexedDB. Any write transaction in IndexedDB is taking ~20ms and couple of tests need at least 2-3 write transactions. Hence checking off IndexedDB tests Before ! After !\nI think tests like that which still take a long time should be renamed to\nOkay. Only couple of tests are taking around 80ms. But since the cut off is 50ms, I see I do not have option other than changing them to integration tests. So changed it.\nAny reason I was assigned back?\nSorry, my bad\nfyi - list starts now at and I have re-assigned some folks\nIt seems you overwrote the checkboxes\nThese are new checkboxes, everything slower than 125ms - I copied the list from the collapsed section\nSorry that I am confused. Following tests are fixed before and they were checked off. tests were fixed before and tests are moved to integration tests . You can see the updated results when I run locally: tests are not taking more than 125ms and tests do not exist. <img width=\"1134\" alt=\"image\" src=\"https://user-\"I am not sure if I am missing anything here Anyway, I have made tests much faster\nIts also weird that I do not see any collapsed sections in my browser <img width=\"976\" alt=\"image\" src=\"https://user-\"\nI read through and couldn't find any of my tests. I think this is a case of \"wrong Alex\".\nA bunch of tests that were using fake timers, like the one, got faster after\nSome observations after having looked into mine: in some of my slow tests I tend to do many assertions in a single test as opposed to doing 1 assertion per kind, as such I now split these tests up into smaller chunks but this will obviously not reduce overall execution times in some tests I was able to squeeze out a bit more by doing instead of awaiting promises one by one I attribute slowness in my tests mainly to the fact that: I access the disk (because I actually want to test APIs or native behaviour) I access a native library such as SQLite I was trying out the fake timers thing, but obviously that does not help a lot for disk access or native libraries. Bottom line: if we want unit tests to finish in range then I think we need to pull out all unit tests that test disk or native libraries into a new suite that is either integration tests or can be controlled by some flags. I am not a big fan of integration tests mainly for the reason that I cannot run them as nice as I can with our test runner support (I brought this up in our standup to Thoughts?\nOne thing we noticed is that quite some tests use the . Those will surely benefit from an InMem-provider <img width=\"944\" alt=\"Screenshot 2022-06-09 at 15 21 59\" src=\"https://user-\"\nThis would also be less of a problem if we were able to run tests in parallel, as slow tests are usually not computationally-blocking slow tests. Mocha does have a flag that runs all tests in parallel, which may cause breakages. Vitest/jest allows explicitly marking tests as parallel-capable via . Maybe we could do something similar.\nI second this. I agree we need to test the , but that's a single test suite in itself. Every other component should be unit tested with a which doesn't use any IO. If it's required that they do, I argue that is the very definition of integration test.\nfyi - I reran the data extraction and updated the main comment here\nWorking on a couple of things with more adoption of figured out that we run a for every to ensure data is flushed to disk which makes writing slower and figured out a way to disable this for most tests except the ones that want to test writing specifically adopted scheme for some of my tests and even found a bug in our in-memory fs-provider Before: After:\nMaybe after my PR landed, the list could be regenerated.\nMay be its better to switch to use fake timers by default for all tests?\nMine will be addressed by moving to integration via I found new ones in notebook land, adding I have asked to support our selfhost test extension showing test decorators in files because one papercut I have is that I can no longer use the selfhost test extension in those test files.\nwe should action these as part of this issue: <img width=\"837\" alt=\"image\" src=\"https://user-\"\nExcellent في الجمعة، ١٢ يناير ٢٠٢٤، ٣:٢٧ ص Matt Bierner"}
{"_id":"q-en-vscode-c642dc99fff1f3b78d54e9e17107c788345c11bfbf39d6d5a129be839d432257","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes/No <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.67.2 OS Version: Windows 10 TL/DR: When saving untitled documents that have name with extension, the proposed file name is not the same and can be completely different: -(I do not have a file with that name ANYWHERE on my PC. I have somewhere though) -Steps to Reproduce: Issue 1: extension command that executes the extension and activate the command to get a new untitled file tab to save the file. The proposed filename is despite the original Untitled file name already ending with which is a valid extension for YAML files Issue 2: extension command that executes the extension and activate the command to get a new untitled file tab to save the file. The proposed filename is , which is unexpected and very surprising. I do not have a file with that name anywhere on my PC. I have somewhere though. Note that the VSCode- extension is here despite VSCode detecting the file format as YAML based on contents. I expect that when saving the file VSCode proposes my given filename and extension to the user. I expect that when saving the file VSCode does not propose completely unrelated file names like .\nThanks for creating this issue! It looks like you may be using an old version of VS Code, the latest stable release is 1.67.2. Please try upgrading to the latest version and checking whether this issue remains. Happy Coding!\nHere is how to open an untitled file with associated file path: As for being able to suggest an extension, here is the related issue:\nThere are certain cases where the proposed solution might not work. The solution requires an absolute file path, but it's not always available. For example, with VSCode Web the workspace might not contain any directories.\nI've tested your suggestion and it does not seem to work. The following code still proposes the file name with double extension: does not work either. This bug really breaks the CustomEditor extensions. The saved file is no longer associated with the custom editor due to the mangled name/extension. More so, the Custom editor immediately closes upon saving the file. So the flow becomes: /New File.../ Pipeline on building a pipeline in the custom editor. -VScode prompts for file name suggesting . The user clicks OK immediately closes the custom editor and reopens the file using the text editor (since the file name no longer matches the extension)\nMoving to custom editors team for this case.\nI think part of this issue is triggered by language detection. I have a hypothesis that the extra is /triggered by the YAML language detector. When the file contains binary data the issue does not seem to be triggered. (I've tested on the modified PawDraw extension modified to propose the file name)\nI've just verified the hypothesis. I've modified the CatScratch editor to use the extension and \"New Cat Scratch Document\" command based on PawDraw. The issue started betting triggered (proposed file name). So the issue is triggered when the document contains valid YAML data (with includes empty file), but does not trigger when the document contains invalid YAML (e.g. binary data).\nThis also might be related to the \"Duplicate editor tab\" bug I've previously opened: I've recorded the video that shows both bugs: https://user- See how with first click on \"Scratch\" another tab opens (with text editor). When saving the custom editor tab, the tab closes (due to the bug we're currently discussing: ).\nRepro branch:\nme because Ark-kun mentioned language detection... and I own the ML language detection but this has nothing to do with that from what I can tell... it also is easily reproducible without custom editors as the original issue describes. I this to an barebones extension: With the behavior I get when I save the file is: ! Now if I change it to I get: ! I have no idea what is suppose to be. Trying a different file type (which is a language we ship a grammar for) I notice that this does in fact work as expected: ! So with this said, my suspicion is that something is up with YAML... I noticed I had the Python extension installed so I disabled it and well well well: ! So one that needs to go to the Python extension: \"don't clobber the YAML file association\". Looks like there are several instances of that in the extension: As for why becomes I'm not totally sure... but a discovery I made was with this list: in the built-in YAML extension (which just provides syntax highlighting). is at the top of the list. If I move to the top of this list, suddenly, this works: ! So two is \"all possible extensions should be taken into consideration\". I think you own this? What do you think?\nnice analysis The issue here is that our logic for figuring out a file name for untitled files is pretty weak: We basically just append the first extension for the language, but we can be smarter: preserve extension when it matches (same for languages that only have file names) actually replace the extension when it does not match so that we do not have 2 extensions piling up I think this fix addresses both issues but would be good to verify then. PR:\nso then after this change, what the Python extension is doing is correct to associate YAML with a new file extension?\nYes I do not think that is an issue.\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!"}
{"_id":"q-en-vscode-31efd30f99ad2604c0e19bebb3dedff404e18f15536d38bdf39575363a27d900","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --The vscode shell integration currently needs to be turned off or disabled for anyone with complex prompt commands, or any trap DEBUG. Please consider supporting some kind of chaining model so we can work with the shell integration as needed. Even better - use the defacto package that has already been written. VSCode is the 3rd script that wants to mess with prompts and traps - but it's the first to completely ignore Please consider shipping this along side and sourcing/using it. If that is unacceptable, consider at least detecting this module has been loaded as documented in the repo - and using it: fwiw: iterm2 has very very similar functionality to the vscode shell integration, but I've been able to work with it. Although they embedded the script directly, they kept the compatible function names and mechanisms and interop fine.\nI believe we have already fixed this: Adding to make sure\nWell, line 100 breaks my prompt. Adding the line as the 1st line of vscpreexec shows why. I think perhaps there needs to be 1 more state 'not in prompt' that causes preexec to bail until all promptcommand processing is done. Alternatively, if this is all to send vsccommandoutputstart - it seems like using PS0 on bash = 4.4 would be a lot cleaner?\nHi This was on 1.67.2 with dropped in place.\nWe've made some big changes in the past few days around debug trap and PS1 handling in bash, I believe the issues here are fixed. We may at some point hook into similar to but for now that will just increase the number of cases we would need to test for (contributions very welcome).\nActually I'll keep this open with help wanted to track integrating with bash-preexec.\nOK. Pulling the - Things look better, but not quite right. The problems I see (only when DEBUG trap is being used): Still seems to be not preserving $? for promptcommand Seems to execute the trap twice. repro with setup instructions below: Do 'cd' into some directory. The command '..' should then pop one directory up, but it will try to process twice and pop twice. Instead of 'success dots' in the gutter, all commands seem to think there are fail exit status. Here are my dotfiles: and a branch to help debug: The debug branch just build an array of what commands were that triggered the trap and print them at the start and end of building the prompt. It does show the trap being called twice when shell integration is enabled - and once otherwise. The repro environment/steps: Repros in OSX and WSL Ubuntu 22.04 for me. It uses dotbot. In a scratch user directory you can run ./install to link everything into ~/. Pass to ./install to avoid installing extra tooling it finds missing (pygments/tmux/coreutils)There is a script that get's sourced . This is the root of all evil. It's on the hacky side. The function is small and is (I believe) the only relevant part. I should stop using it, but it's sooooo_ ingrained into my muscle memory. To verify: Removing from makes everything work great with none of the above mentioned problems. One hack, that seems irrelevant is that uses debug trap directly, but also detects and uses the pre-exec infrastructure if it has been loaded (to not break iTerm2's shell integration). I appreciate the time and effort going into this feature. It's becoming a shockingly advanced embedded shell environment.\nthanks again, planning on tackling this in June\nI use just merged in a fix for this. It would be great if someone could help verify in tomorrow's insiders build as I'm not normally a user of bash-preexec.\nStill some problems around . Very simple repro: clean environment. My bash and vscode versions are: use this .bashrc: For me (In WSL, repros in plain OSX): ! Without ! If extdebug works, then this would be the next simplest test (I think). Same clean environment. The following\nAre there verif steps for this?\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!\nWith the more complicated scenario, the decorations don't seem to show blue and red anymore. ! I realize that in the screenshot I forgot to add , but that didn't make a difference.\nwhat is your exact debug trap when that happens?"}
{"_id":"q-en-vscode-e758746b0ab0e9def4b32680796ea3abd2f5a815f0e7c25dcce4245c55c09c91","text":"I am the maintainer of VS Code basic latex language support. The language defines / as a bracket pair. This leads to wrongly highlight <img width=\"415\" alt=\"Capture d’écran 2022-05-26 à 08 07 09\" src=\"https://user-\"I believe this is related to fixed it in by surrounding brackets using letters only with . However, in LaTeX every command starts with and the fix does not apply. This issue occurs on Version: 1.67.2 and also on Version: 1.68.0-insider (Universal) Commit: Date: 2022-05-25T05:16:10.689Z Electron: 17.4.4 Chromium: 98.0.4758.141 16.13.0 V8: 9.8.177.13-electron.0 OS: Darwin x64 21.5.0 This problem was originally reported at\nshould be extended to check if the bracket starts/ends with a word. Only at those borders, should be . Should be easy to implement. I'm open for a PR here ;)\nhey I am a beginner in open source contributions . Could you please guide me how too start contributing\nI'd start here:\nHi I'm just starting out with open source contributions and just finished reading the how to guide you linked in this thread. I'm having some trouble understanding exactly what the issue is, could you please explain it a bit more? Is rangledot supposed to be highlighted as a whole?\nThis issue is actually already handled by PR , which will be merged for the July 2022 release.\nVerification steps: Install the LaTeX extension Verify that and is matched, but not with in\nI would personally highly go against this fix As it would disallow word based brackets inside word strings with no alternative (other than to reinvent the bracket highlighter again) I have no example of any language being affected But I would question the arbitration decesion to limit such function Instead, I think LaTeX should fix their bug (which was present before coloured brackets) by just disallowing/allowing certain word based bracket scopes via the and sections in !\nLet's discuss this in a separate issue once this actually is a problem!"}
{"_id":"q-en-vscode-ef8154453070b2aacba8b8375a158b788ec33cfe3463a87f8d0a49fa7ebab08b","text":"Testing Version: 1.68.0-insider (user setup) Commit: Date: 2022-05-31T05:17:04.479Z Run the command and select Would be nice to get a message in the log\nLooks like the the message is not shown as it is of type and is less than than. Would be good to always show it, regardless of the current and the new level. Same problem with the initial message when the output is started."}
{"_id":"q-en-vscode-b5c564decfc313db137ec05b24986e18a3e714cb7e1c616acbc087c5d6ddfb6a","text":"Testing Things I would change: make it clear in the picker that selecting a language from the lower list will install it have a button in the picker that installs and syncs it. don't reveal the extension viewlet don't show that dialog <img width=\"444\" alt=\"capture\" src=\"https://user-\"\nis there a way to not show this dialog when installing an extension using the Extension management service? Here's the specific line:\nHow would you make this clearer? I figured since we had a section called that will always be there on Desktop it's implied that must not be installed. What will the default behavior be then when you accept a quick pick item? That it will install but not sync? That's not what the default is for the button in the extension pane so that seems a bit odd to me. This was to give a progress indication. We do the same thing for the Theme picker. I would like to hide this dialog. I didn't see it locally because Settings Sync isn't available in but I asked Sandeep above.\nYes, you can pass machine scoped flag to avoid this. Passing for this flag will not sync the extension and will sync it.\nI thought the main reason for the dialog was to inform the user that something is installed. I can easily live with only the picker Since the dialog offered the option, I thought it is important. I can live with the default without a choice. I found it disturbing. I could live without progress or some window progress."}
{"_id":"q-en-vscode-42826047b7ae7b80323556308179af06e5b830577349da0957d5316b73aeb417","text":"! I see this sometimes, but it often gets fixed if I try again, or close and reopen the editor, etc. Right now it keeps happening and it's just impossible to click the hover. I don't know how this API works but I assume this is on vscode's side. Happy to set breakpoints and debug\nI couldn't reproduce this so I'm not sure what's the root problem. However, I'm saving here a patch which simulates a hover very similar to the one in the gif: <img width=\"1141\" alt=\"image\" src=\"https://user-\"diffn- editor.focusNotebookCell(newCell, newFocusMode, { focusEditorLine: 1 });n+ await editor.focusNotebookCell(newCell, newFocusMode, { focusEditorLine: 1 });n\nI tried for quite a while but couldn't repro. I think there might be some settings that cause the issue reproduce. Does this reproduce when you comment out all your user settings?\nYeah, it reproes with all settings commented out\nThanks to via we could find a repro. Reproduces consistently when hovering at the end of the line: It turns out it has to do with the range the hover provider returns. If the hover provider returns a range that spans one line above the hover request, the hover will render one line above. If the hover provider returns a range that spans two lines above the hover request, the hover will render two lines above, etc. Here's a simplified patch: https://user-\nTo verify, use gitlens and hover over the end of the line at\nWorks great but raises the question, why does gitlens return a hover with a multiline range? And I get a highlight in this region, but it blinks and disappears on line 919, and on other lines, no highlight. !\nPR to improve the highlight range at\nI'm still seeing the original issue at"}
{"_id":"q-en-vscode-4c0479c9d7434cfaebb5d7d0782df895642d1796aa3384c992c69176af06808a","text":"While verifying I noticed that if I click on the warning I'm not taken to the size limit setting which I would've expected to be taken to. ! Version: 1.68.0-insider (Universal) Commit: Date: 2022-06-02T05:16:30.515Z Electron: 17.4.6 Chromium: 98.0.4758.141 16.13.0 V8: 9.8.177.13-electron.0 OS: Darwin x64 21.5.0 Jupyter Extension version (available under the Extensions sidebar): v2022.5. Python Extension version (available under the Extensions sidebar): v2022.7.\nThis link is specified in core code.\nThe issue doesn't seem to be fixed in the latest insiders. Reopening\nThe issue is now fixed for me in the latest Insiders, at least on Windows 11."}
{"_id":"q-en-vscode-078c45aed97b30fe83ae0e2eafa9a8983e0859875a1f655860d252f1aca69e3b","text":"Open the color theme picker and select the first option () Observe that it takes around 5 seconds for the results to show. This used to be much snappier"}
{"_id":"q-en-vscode-3da4824da0c3578a688fd1f65379f3b7e62c02f592c1a075d6b12f6520c304b7","text":"!\nThink I saw this today\nthat was when I here, which causes other issues so i'm removing\nworking on it\nrepro is to create some decorations via commands and then delete the terminal"}
{"_id":"q-en-vscode-e3fff8047014f42862f0172aee983e457d67f18906e696a36011352cf4ad9ffe","text":"!\nThis happens because command executed happens but not command finished\nAnd that's because of this:\nnice, the fix works https://user-"}
{"_id":"q-en-vscode-7d124e98e487f29580d47eb05c2c5972581bbfe8fc2cf36de68d29f31f361679","text":"If a password is not decrypted properly, then it should be removed from the secret storage because it's essentially dead. Originally posted by in\nFriendly ping! Looks like this issue requires some further steps to be verified. Please provide us with the steps necessary to verify this issue.\nVerification steps: grab and install in VS Code Insiders Run command Follow the notifications ping me if you have any questions or issues."}
{"_id":"q-en-vscode-8066f2c2c828f8347a0e218a4d2a26f2ada629e05394009beb091362395e520d","text":"Issue Type: <bBug</bI do not see the error or success icons for the gutter icon on my shell integration enabled terminals; instead it is always the empty circle. The integration appears to otherwise be working (the Go to Recent Directory and Run Recent Command commands are working as expected). I presume that somehow, the integration is not picking up the result of the operation being run. For reference, I am using but otherwise don't believe I have anything special in my Bash configuration that would affect this. VS Code version: Code 1.68.0 (Universal) (, 2022-06-08T11:44:00.250Z) OS version: Darwin x64 21.5.0 Restricted Mode: No <details<summarySystem Info</summary </details<details<summaryExtensions (76)</summaryExtensionVersion ------ Bookmarks13.3.0 project-manager12.6.0 vscode-sqlite0.14.1 atlascode2.10.12 github-markdown-preview0.3.0 jsdoc-markdown-highlighting0.0.1 markdown-checkbox0.3.2 markdown-emoji0.2.1 markdown-footnotes0.0.7 markdown-mermaid1.14.0 markdown-preview-github-styles1.0.1 markdown-yaml-preamble0.1.0 gitignore0.7.0 vscode-markdownlint0.47.0 vscode-eslint2.2.2 xml2.5.1 gitlens12.0.7 EditorConfig0.16.4 graphviz-preview1.5.0 flow-for-vscode2.1.0 vscode-mysql0.4.1 copilot1.28.6134 remotehub0.34.0 vscode-pull-request-github0.44.0 go0.33.1 todo-tree0.0.215 triggertaskonsave0.2.17 vscode-guid1.4.20 path-autocomplete1.19.1 tablegenerator1.0.4 plantuml2.17.2 svg1.4.18 docthis0.7.1 vscode-position1.1.2 yarn-lock-syntax0.1.3 Kotlin1.7.1 vscode-language-babel0.0.36 pull-request-quantifier0.1.2 extension-manifest-editor0.1.5 csharp1.25.0 dotnet-interactive-vscode1.0. network-edge-devtools1.0.0 data-workspace-vscode0.2.0 mssql1.14.2 sql-bindings-vscode0.1.0 sql-database-projects-vscode0.16.2 vscode-postgresql0.3.0 python2022.8.0 vscode-pylance2022.6.10 jupyter2022.5. jupyter-keymap1.0.0 jupyter-renderers1.0.8 hexeditor1.9.6 powershell2022.5.1 remote-repositories0.12.0 vscode-js-profile-flame1.0.2 vsliveshare1.0.5615 vsliveshare-pack0.4.0 resourcemonitor1.0.7 vscode-commons0.0.6 vscode-yaml1.8.0 sass-indented1.5.1 gitconfig2.0.1 mdx0.1.0 codevalet-vscode1.5.0 vcl0.1.2 msbuild-project-tools0.4.3 jest-snapshot-language-support1.1.1 intellicode-api-usage-examples0.1.1 vscodeintellicode1.2.21 vscode-icons11.12.0 quokka-vscode1.0.478 wallaby-vscode1.0.338 jinja0.0.8 ReactSnippets2.4.0 vscode-proto30.5.5 (2 theme extensions excluded) </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nIt's working fine for me (I have the icon customized): <img width=\"447\" alt=\"Screen Shot 2022-06-10 at 11 29 56 am\" src=\"https://user-\"This is with a fresh install of stable bash-git-prompt Version: 1.69.0-insider Commit: Date: 2022-06-10T05:16:48.648Z Electron: 17.4.7 Chromium: 98.0.4758.141 16.13.0 V8: 9.8.177.13-electron.0 OS: Darwin x64 21.5.0 Could you see if it happens on the ?\nI am installing the Insiders build now. I will let you know.\nHmmm, doesn't seem to work. <img width=\"232\" alt=\"image\" src=\"https://user-\"\nHow does it detect the exit code? Perhaps I have something installed that is interfering, either in my bash config or elsewhere?\nYeah it's meant to change depending on the exit code of the program. Some info that would help us investigate: Set log level to debug via the command palette, then in Help Developer Tools console it will show some shell integration related logs when you run a command Do you set PROMPT_COMMAND anywhere? Do you use bash-preexec?\nI'll get this going and let you know. Ah yes, besides using , I have this in my bash config: I'll try disabling that and see what happens. It would be a shame to lose it but perhaps there's a better way of implementing that so that it's compatible, if that turns out to be the issue. Not that I am aware of, no.\nOK, so I disabled my PROMPT_COMMAND change and tried a few things in a new terminal that weren't working before. Now, when I use it is still an empty circle, but when I do , I get the green success lozenge!! So I'm guessing that was the main issue; though I do wonder if I should be seeing that green circle for and such too (that said, this is the non-Insiders build - I'll try Insiders in a second). I'll add the log level and see what I get.\nI can confirm that the Insiders build works as advertised now, so it was my history manipulation code that was getting in the way. Thank you for helping out. I'll close this issue as it appears to be an issue on my side, not yours. Thank you for helping troubleshoot.\nLet's keep this open to track the issue so we handle it better.\nwould you like me to gather the logging info as requested, since this is staying open?\nI think we understand the problem enough to reproduce it, it's just a case we need to detect and handle better.\nThis is happening because we actually depend on to differentiate regular commands that succeeded or failed vs ones exited via or running an empty line. Specifically this call: I'll disable shell integration if includes as that specifically is the reason it was failing.\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!\nTo verify: In set Enable the shell integration Open a terminal, it should not have shell integration enabled\n/verified I see that the integration is not enabled as the lozenges are missing."}
{"_id":"q-en-vscode-33c3047cf74d705137b0b3b93c6a252fbe838169574b1670335fe6b1ae4d90f5","text":"I'm always returning to projects and see the terminal messages filled with the same repeated message and would like to clean them up and not be shown as I see no value in them for myself. <img width=\"630\" alt=\"CleanShot 2022-06-14 at 09 10 55 src=\"https://user-\"<img width=\"622\" src=\"https://user-\"\nThis bugs me too, tracks only showing it once in this case by avoiding saving new parts if there is no terminal input, similar to how it only saves and restored in the first place if the terminal has been interacted with. I don't think we should be hiding the message all together as it's misleading what is part of this shell's session, command decorations would be broken for the ones that aren't part of this session (look like a bug) and it would show up like this which is a bit weird: !\nWould it be possible to have an option to hide the previous session altogether and just show a new session? That would be a simple approach that would work for me personally, similar to <img width=\"566\" alt=\"CleanShot 2022-06-14 at 09 40 18 src=\"https://user-\"\nyou can just create a new terminal altogether by setting to false\nthank you! I was searching for a setting and didn't ever find that one, kept looking for \"terminal history\" and \"terminal clear\"\nthe name is definitely not discoverable. I wonder if we should change it\nor can we add any meta data or keywords to settings that will be picked up by search? cc\nI guess?\nOh that doesn't work, do we have a way to do search keywords?\nyou actually want , the setting will disable process reconnect across window reloads."}
{"_id":"q-en-vscode-3fbc07060d23bca7386b3760312a315b46947b2450960f45cf52d9da5e849f16","text":"Issue Type: <bPerformance Issue</bTerminals doesn't seem to free all GPU memory when closing them, making alt-tabbing from another application to vscode slower over time, there are other tickets around vscode freezing when alt-tabbing back to it and this seems to be one of the reasons as Windows needs to load/unload 1GB+ of GPU memory when alt-tabbing occurs. One way of reproducing this: 20 integrated terminals the terminals again (now only welcome screen open) Windows task manager -Performance tab something covers vscode completely (i.e. task manager is maximized) the GPU memory usage is 500MB+ lower than if vscode is visible (i.e. task manager is not maximized so it doesn't cover vscode completely, forcing Windows to keep vscode GPU memory active). scales linearly if you open more terminals, i.e. 40 terminals keeps 1GB+ of GPU memory even when they are closed I can reproduce it on the insiders build as well, so Electron 18 doesn't seem to solve this issue. vscode visible (2 GB GPU memory usage at the bottom of the screenshot): ! vscode still open but behind other windows (1 GB GPU memory usage at the bottom of the screenshot): ! VS Code version: Code 1.68.1 (, 2022-06-14T12:48:58.283Z) OS version: WindowsNT x64 10.0. Restricted Mode: No Remote OS version: Linux x64 5.10.102.1-microsoft-standard-WSL2 <details<summarySystem Info</summary </details<!-- generated by issue reporter --\nI tried to repro with stable, but I wasn't able to. What are the steps for verification?\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!\nHello, I'm waiting for these to cleanup the terminals properly: But as I am the author of those PRs I don't think additional verification is needed from me once they get merged.\nfyi I think I noticed a few bugs about alt-tabbing being slow"}
{"_id":"q-en-vscode-1efde4bbe6fcb92a698e8ff18c2681f694e77b42cb2bb998462ac4941457e590","text":"Previously the terminal could be resized fairly quickly, to see what's going on and then hide it again (with 2 lines still visible). With the new update however, it can only be resized by one line per keystroke. So in order to use the terminal adequately, there are 10 - 20 keystrokes required, which is unacceptable (in my opinion). It would be great, if you could specify \"args\" for the command which control how many lines the terminal is resized by. Similar like the cursorMove command: Thanks for consideration, Floyd Haremsa <!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. -→\nI'm not sure what you mean. changing the height of the panel? https://user-\nYes, changing the height of the panel, but with keyboard shortcuts. Previously it worked pretty well (changing the height by lets say 10 lines, I don't quite remember). But now you can only resize by one line at a time (per hitting the keyboard shortcut), like this: https://user- and that is just way too slow and inefficient. This way I'm forced to resize it with the mouse (which is also too slow) or having it open all the time, which is bad on small screen sizes. Sorry for not having it clarified enough earlier.\nwe think this is related to the increase view size change - thoughts ?\nThis is indeed caused by The new behavior is actually the intended behavior, but the terminal is in full control if it wants to have a number-of-lines-based resize setting.\nI am also affected by this and it is crazy to work mouseless, resizing terminal is not viable with keyboard like this. Any work around?\nhow many rows would you expect it to change?\n3-4 would be nice\nI'd say about 7 - 10 lines at a time. But if I can control it my self, that's even better.\nAlso being affected by this. The terminal now resizes by 1 line at a time. Previously it was 9 lines (3072×1920 on 16\" with terminal.integrated.fontSize = 13) Something in the range of 4-8 lines might be a decent compromise considering the original request that led to the change ()\nI think the best approach would be to just have a default of let's say 4 lines, but also a setting that controls this, and/or as I mentioned, arguments to the command for being able to map multiple resize settings to different shortcuts:\nI'd rather not add a setting for this as it's pretty obscure and would see hardly any usage. An additional complication here is I'd like to eventually remove the terminal resize commands since we now have the increase/decrease view size commands which weren't around when the terminal resize pane commands got . Seems like 4 rows should work for most people.\nCan confirm this is fixed in Insiders. Thanks for the quick response"}
{"_id":"q-en-vscode-f21de36281c22ea994c021ac4420b1913fbbe778b2f8a633c94e771e7bf5f74e","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --Executing \"Surround with snippet\" from the command palette while there's no surround-able snippet, results in an empty pick list: ! Shouldn't it be like showing an information message saying there's no snippet to pick, or just not prompting the user to pick anything?\nPrompt might be too annoying but quick pick itself might have a way to show a lightweight in-place message\nDo you mean something like this (using )? !\nJust submitted the PR. Could you please check it?"}
{"_id":"q-en-vscode-3ac5bd3e0270eea5c8c4912855d19c9282951c07e0da010fca13bc2aea89e6f3","text":"Ah, sorry for the breakage... I believe asking for both is the best option for now but I can also add a \"migration\" path into the implementation, e.g use when seeing Originally posted by in"}
{"_id":"q-en-vscode-2737eec760b0b7a6027312bf62125d6c090e9d28c369dcdfa9a5cb1514d80073","text":"I really like how I can see the word-level change per line now in the merge editor, but shouldn't the colors be identical to the diff editor? !\nI agree the new color needs to be tweaked. Would you also use the green background?\nI think making this look as close as a diff editor if possible would be nice because I am so used at looking at diffs with their color coding. //cc"}
{"_id":"q-en-vscode-08b811e2f91d471c4b9d26d1b942097cb31d85075a745aaabc9dabbfc03f643f","text":"Testing !\nThat's actually a description of what clicking the button will do. This was as designed but maybe has a better idea?\nSame feedback from in\nThe \"...\" in \"Show Command Actions...\" tells me it's clickable. Just removing the \"...\" would help, but it still looks like something I want to click on."}
{"_id":"q-en-vscode-bc4f9dc84da65b1fd80881bceea5d129533f96321a80eb6fe69e0d4e2465878d","text":"Testing !\nI just noticed that is not localized and is not meant for display. That means that \"Label\" was not a good name for that function, but we can either just remove the state or find the code that drives the state label in the callstack.\nActually, it seems that only displays , , or , which is probably not super useful to the user anyways. It seems like the info like is only available on each thread, and a debug session can have multiple threads. Perhaps we should just remove the state for now."}
{"_id":"q-en-vscode-e64df50b5ab286eb1476838e7deaddc3fee3d7d8347f5f52c8cb4d688cfc6ec6","text":"Testing Current: ! Proposal: !\nSpoke with and I have submitted to add the file icon association to seti-ui repository."}
{"_id":"q-en-vscode-e5911d02868ece5c995920a7c27acdcb7b8b4583737b313d006ffd79e3765fcc","text":"Testing When I use cmd+up/down to navigate between POIs in task output, the blue box renders for a second on the item I have selected, but then disappears. I didn't understand what was happening to the focus border.\nthat is by design.\nThis has come up before, I think you had an idea to improve this?\nI had the same experience in where I was on a slow machine. I think I had suggested just using a line on the side (like our activity bar) that remains until your focus changes: <img src=\"https://user-\"\nTo the left of the command decorations? ! I like how visible the outline is when jumping between commands, so we could keep that as is and just also show the persistent line? !\nyea I'd be open to trying that out, I think using both might be a bit overkill but just having the single line can also be hard to notice."}
{"_id":"q-en-vscode-9544a45002d6b97826769c2bbb906523e36bb1c384fc28d280713449e93d97dc","text":"Testing Clone a git repository Open git repository in VS Code desktop Make a change to one or more files Invoke the command You are presented with an empty picker <img width=\"736\" alt=\"image\" src=\"https://user-\""}
{"_id":"q-en-vscode-7dc9a0950d5fab20471bbce534fd6e56b220177d739ca438688e00e0db0c89ef","text":"Issue Type: <bBug</bI opened theme extension and click enable / disable fast. I see \"Set Theme\" button when extension is disabled and no \"Set Theme\" when it is enabled. VS Code version: Code 1.68.1 (, 2022-06-14T12:48:37.209Z) OS version: Darwin arm64 21.5.0 Restricted Mode: No Remote OS version: Linux arm64 4.9.253-tegra <!-- generated by issue reporter --\n<img width=\"582\" alt=\"Screenshot 2022-06-28 at 11 08 49\" src=\"https://user-\"<img width=\"660\" alt=\"Screenshot 2022-06-28 at 11 08 53\" src=\"https://user-\"\nI can repro this. First assigning to for extension details page.\nCool, thanks"}
{"_id":"q-en-vscode-7510f22f6f009a2e68f64a5c949ffc8ddf81d6f0a38d1090b2b63d54f4e79a66","text":"! cmd+f, do a search clear the text in the find input it says \"0 of -1\"\nwas this insiders or code - oss? If the latter can you make sure is updated?\nneed to see if this reproduces in OSS\ndoesn't reproduce in the demo... I think I updated before the PR checks had finished maybe so that fix wasn't included?\nI tested with my branch of symlinked so that's probably why I wasn't seeing it\nyep happens in OSS so the fix is to update in vscode"}
{"_id":"q-en-vscode-1b8f8b94c60765b841218bc288d30108a91484734f4edf011a8a147b2679d590","text":"Issue Type: <bBug</bTesting I have Code Spell Checker () extension provides diagnostics without error code, and rust-analyzer extension provides diagnostics with error code. When all diagnositcs don't have code, it displays correctly. When some diagnostics have code and some don't, scroll over the Problems value, some code changes randomly. https://user- VS Code version: Code - Insiders 1.69.0-insider (, 2022-06-29T05:20:28.364Z) OS version: WindowsNT x64 10.0. Restricted Mode: No Remote OS version: Linux x64 5.10.102.1-microsoft-standard-WSL2 Remote OS version: Linux x64 5.10.102.1-microsoft-standard-WSL2 <details<summarySystem Info</summary </details<details<summaryExtensions (12)</summaryExtensionVersion ------ Bookmarks13.3.0 remotehub0.35. remote-containers0.241.1 remote-ssh0.83. remote-ssh-edit0.80.0 remote-wsl0.66.3 remote-repositories0.12.0 material-icon-theme4.18.1 material-theme3.15.2 toml0.6.0 rust-analyzer0.4.1110 code-spell-checker2.2.5 </details<!-- generated by issue reporter --"}
{"_id":"q-en-vscode-0aa8edfa0d820f28bc0e6066a7d6ac65ae44b299d64040de7c8efc74006aaec4","text":"a folder/workspace that contains a git repository a topic branch changes to main and the topic branch that would result in a merge conflict main into the topic branch to get a conflict Expected: commit action button is disabled Actual: commit action button is enabled\nThe button is now disabled while merging is in progress, but the button and the dropdown both take on a background color when clicked, which was confusing to me: !\nAlso, while the commit action button is disabled, the commit button above the input box is not: !\nThat is expected. At the moment the enablement is only toggled for the commit action button and not the commit action in the toolbar.\nThis is currently being tracked as"}
{"_id":"q-en-vscode-61ff6dac4ced87eec5ce45cbd8dc5bcc252fda909bf51af861d0acc08680fc6a","text":"See the grey row that moves up when it scrolls !\nI looked into this a bit, but is this something that is within the tree code? Nothing sticks out to be surrounding us registering the drag and drop handlers when it comes to a focus, or hover rendering\nThe grey row is the CSS state which stays enabled throughout a native DnD, but stuck in the same position the event fired. It's arguably a browser bug. But we can eat it as a list bug. I see no other way than to disable the hover decoration while the list is being dragged from. Having both and labels is a bit confusing. Those are mutually exclusive.\nOh I always used polish to indicate something that is a small change (at least to the user) and then it's a bug because it shouldn't happen . Idk, I guess I'll stick with just bug"}
{"_id":"q-en-vscode-de3dcfa1b0acd3c4f2430204eeda50fa55e39f41052818af3cd1ed2f62ca83de","text":"Lately (since a few days) I see the explorer show an expand all icon even though some folders are expanded (in a multi root workspace): <img width=\"300\" alt=\"image\" src=\"https://user-\"This must be a recent regression, not sure if in explorer or tree widget.\nSee\nThis breaks my muscle memory, I always collapse the explorer to start a new task and the new behaviour breaks this. I really see no value of this new behaviour? It is not even expanding all folders, only the ones that can be resolved? What is the use case?\nAs a suggestion: only show the expand button when really all nodes are collapsed, but keep the collapse button when any node is expanded.\nI'm with this is quite different behavior in a multi-root workspace, I have one workspace with 5 root folders and another with 10+, prior to this change clicking the expand/collapse button in explorer would collapse all the folders in each root folder, clicking again would then collapse all the root folders. What I get now is the following: If all root folders are collapsed, clicking the expand/collapse button does nothing If one root folder is open the expand/collapse button expands and collapses folders within that root folder More than one root folder open the same as before for all of the open root folders Perhaps we could have an option to opt-in to this new behavior, I can't think of a single reason why I would want to expand every folder in a project.\nFeature got removed."}
{"_id":"q-en-vscode-8eded13f75c437dacf7a8399638c6ed9117ff70fd7b2a1ba3d35c5b4cb0334ae","text":"to start the default build ! This happened right after something strange, which might be the underlying cause. I had the build running already, the one with two terminals open. But I still pressed and noticed another two terminals showed up with the same build tasks! I closed them all. Then I pressed and ended up where the screenshot above shows. cc\nI reloaded, then pressed before the workbench was done loading and got this: ! I waited a bit more, pressed it again and it worked. This is very likely a separate issue, so feel free to treat it as such.\nmight be a bug with the new reconnect?\nYeah I think I saw this once\nThat's exactly right. Steps: selfhost setup Shift B, you now have 2 build commands running VS Code VS Code, you still have 2 build commands running Shift B, You now have 4 build commands running.\nC the old ones The new ones appear loading, even though they have finished compilation: ! Worse: the window horizontally, and eventually you'll get lucky and the spinning icons dissappear: !\nit's not a new bug that you can launch tasks that are already running\nIf you press you should get the default build tasks, which are long running. If you press it again, they shouldn't be launched again and they aren't. This is good and expected. The bug is: restart Code (tasks now get restored), press and you'll get the same build tasks running again. This shouldn't happen: !\nsame root cause as , which is not new\nThis is my selfhost setup. I am 100% sure this is new.\nthe two bugs that you've identified here though are: the error about a terminal group command not found was already aware that statuses do not work after reload\nin your old setup, run ctrl shift b, ctrl shift b again. you'll get 4\nin stable !\nI think you are correct it's a different issue. I see what's going on now.\nmight still be the same root cause, but i'll try to investigate for a bit\nThanks\ncould you pls verify this is fixed for you in insider's keeping in mind these two that I have a PR out to fix as well?\nSorry for reopening but I could still repro in today's build () very likely due to the . !\nThis is happening because this conditional is entered on ctrl+shift+b after reload: Looks like isn't restored on reconnect: !\nThe task has the wrong id? !\nis persisted as a guid, not the expected string? !\nIt's the custom task with the same label that gets the uuid:\nends up returning true, that's why _id ends up as a uuid: The task's type is undefined after reconnection. It's mean to be , this is what it's meant to be: !\nNote that this only reproduce for compound tasks.\nThat was a journey and I'm ready to throw in the towel for now. I believe the issue is either compound task state is not restored properly after reconnect here: And/or the recently used tasks cached value does not contain all recently used tasks as I only see the compound and Core - Build, no Ext - Build.\nany thoughts would be appreciated. I spent ~ 3 hours on this and Daniel did as well.\nI will take some time tomorrow to dig into it.\nSome notes from debugging: When comparing tasks that come out of the recently used task list, they must be compared by their recently used key as this is the only stable id value (map key and task id can change). Instead of having a separate code path for reconnect, which is what is causing some state to be lost (like the ), if possible, it would be better to try the reconnect in the same place that we currently create a terminal. This would hook up all the needed state again and re-attach problem matchers. I've made an example PR here that starts to address the above points. Note that this is relatively untested, I just checked whether it fixes the problem described in this issue. Other notes: Using the recent tasks from the task service for re-connecting is problematic since users can use to entirely disable recent tasks. Instead, you will probably need a separate LRU that functions exactly like but is only used for reconnection. Tangentially related: you also need to nudge/rerun problem matchers on terminal lines that have already been written so that they can find their start pattern and the task doesn't have a spinner forever."}
{"_id":"q-en-vscode-26859056dabbbcf3048cc393fa4d1ecb6fcd9a98d15e7335da95afcfdd8707a5","text":"Steps: some changes on the \"continue working on...\" command and open in desktop :bug: VS Code opens but nothing loads https://user- Desktop Version: 1.70.0-insider (Universal) Commit: Date: 2022-07-26T05:17:37.262Z Electron: 18.3.5 Chromium: 100.0.4896.160 16.13.2 V8: 10.0.139.17-electron.0 OS: Darwin arm64 21.6.0 Web Version: 1.70.0-insider Commit: Date: 2022-07-26T05:17:04.193Z Browser: Mozilla/5.0 (Macintosh; Intel Mac OS X ) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/103.0.5060.114 Safari/537.36\nThis action should be disabled in since we cannot access the local filesystem from the browser.\nI'm still able to run the command from insiders vscode dev in Firefox. My Desktop Insiders then opens the repo using remote repositories but on the main branch instead of the branch I was working on. I'm not sure if this is the expected behaviour\nI verified that this now works as expected https://user-"}
{"_id":"q-en-vscode-df8edff1299a2a4b7f575a524a2e59ba94dbedd1d0e3db7d758171895d790e92","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.69.2 OS Version: MacOS BigSur and Monterey Steps to Reproduce: Open a terminal editor. a file from Finder into the terminal. This tries to open the file (unlike when the terminal is in the panel, which will paste the path).\nI found that if I prevent the following code from executing, it works. So it seems like prevents the following code to handle pasting the file path. I did find that if you hold Shift before starting to drag the file, it will paste due to this code: I'm guessing there's some way to get that to realize it's a terminal editor and return as well.\nI submitted PR for this.\nFile is still opened\ndid you hold shift? That's the method of passing the file through to the editor\nworks correctly with shift held"}
{"_id":"q-en-vscode-a6965518cea8d6f3cfeca54a175bf7f88f2adef49c7d9af1c16cfc5cbbf9cfd0","text":"! The settings to select an codicon for a decoration are not playing well with our product icon theming approach. We want that a product icon theme can style all icons of our UI with a new icons. Each icon that we show should be defined with a component specific name E.g. , , ... That way the theme can come up with a custom icon for each icon location The base icon names (, , ...) should not be used directly, only as defaults, otherwise it gets complicated for a theme to change an icon as it affects many places, unknown to the theme. What might look good with our codicon based product icon theme, might look wrong with a theme that uses a different visual for I suggest to remove the settings as we don't do that at other places. Instead we should refer to product icon themes as the way to change icons.\nRequested the aliases in\nWe do allow switching icons for terminals in this way which I think is fine as they're very temporal: ! I see what you're saying here though, let's remove the setting and leave it up to the icon theme . We have this setting now so it solves that part of what these icon settings allowed:\nI was on vacation so I couldn't jump in here. It was not necessary to request new codicons or codicon aliases unless a new icon shape was needed, e.g. for a concept we didn't have yet. There still seems to be a confusion of Codicons vs product icon icons. The Codicon library is a set of default icons that are part of the Codicon font. We use these as default icons in VS Code. See the spec of Codicon: codicon aliases should be component agnostic (no 'terminal', 'debug'). The componnet specific ids go in the product icon registry. These are the ids that wil be used in a product icon theme when styling an icon. Product icon ids are component specific and are defined in the component itself, e.g. as done here: This defines a new product icon , and uses the codicon 'add' icon as the default icon. When using these icons in code, must be used (or in labels by the id ). We never want to reference in components, unless for the default in . Some years ago, all our components directly referenced codicons and their ids. When adding icon theming support we this indirection and I and migrated most components, including terminal. But given the huge number of icon references that work is still not complete and there are still plenty of references to Codicon. That's why a codicon and a product icon () still have the same API and we don't enforce this with the compiler. Also there are some codicon aliases that have component specific names. We can't clean up the aliases in order to not break existing code in extensions. Eventually no one should use there anymore. please do not reference Codicon directly anymore unless in registerIcon please follow the pattern as seen in terminalIcons please don't add component specific Codicon aliases FYI\nFYI I was following what seemed to be the thing to do as demonstrated by the icons. Could we remove those and similar examples and/or add a comment with this guidance?\nGood point, we should clean these up. The codicons have been around from old times before we did the product icons. There are product icon registration under the same name , so we can remove the codicons that are just aliases, and give the others a different name, to avoid confusion."}
{"_id":"q-en-vscode-305a76ee017e45361593bba65b84bb469e453188773201e12fd8cde6260dc810","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: V1.70.0 OS Version: Ubuntu 20.04 Steps to Reproduce: to latest V1.70.0 version of VSCode starship prompt for the bash shell. I have my starship prompt configured to update if the previous command execution failed for unknown reasons. After the latest update, every command execution seems to fail only in VSCode's terminal. Below is the expected behaviour, in a bash terminal: The screenshot of the same with colors: ! Now in the VSCode's terminal (running bash): (It shows the error condition even in a brand new instance of the terminal) A screenshot of the same with colors: ! This behaviour of always erroring out even for a successful command execution started today after VSCode auto updated to V1.70.0 Below is the information I got from menu: Version: 1.70.0 Commit: Date: 2022-08-04T04:38:48.541Z Electron: 18.3.5 Chromium: 100.0.4896.160 16.13.2 V8: 10.0.139.17-electron.0 OS: Linux x64 5.15.0-43-generic snap\nDid you notice\n, thanks for the prompt response! Yes, by disabling shell-integration, starship prompt started behaving as expected. Shall I close this issue or consider this a bug with shell-integration ?\nI don't see why there was a need to enable Shell Integration by default, spent the last 2 hours trying to figure out what was wrong\nThis probably has nothing to do with starship (I have the same issue without starship). This is just a specific case of and should probably be closed as a duplicate"}
{"_id":"q-en-vscode-789d7e08cf1a7a110c5b69826763b4fecacb2bbec0e7bf59689c81e4cf1faf57","text":"I noticed this in smoke test runner logs leading me to think that the type of the exit code is not a but an now?"}
{"_id":"q-en-vscode-1a9d12d0e8f8fab241a07f1083b5b9dac81ce05fc440cef732d7f1fabb18deff","text":"Type: <bBug</bnotebook an image from the workspace into a markdown cell Bug this inserts the full uri, such as It should insert the relative path instead VS Code version: Code - Insiders 1.71.0-insider (Universal) (, 2022-08-10T05:18:38.020Z) OS version: Darwin x64 21.6.0 Modes: Sandboxed: Yes <details<summarySystem Info</summary </details<!-- generated by issue reporter --"}
{"_id":"q-en-vscode-ab663924d707203fb15f7f8e9e0c3c7a1e9bce42bb0b2c30677c403cf79f0bc4","text":"Testing As mentioned in the title it should be in this modal: <img width=\"405\" alt=\"image\" src=\"https://user-\"\nThanks"}
{"_id":"q-en-vscode-1f40c5aff5f9bdc441e154cc96afaba3e209d322d9f0b0e67cdd4f03e077048c","text":"Testing When my cursor isn't inside of a conflict, this command does nothing. Should it only be visible in the quickpick when it does something? Or should there be an error message when you try to run it but it doesn't do anything? Same with the \"Toggle Current Conflict from Right\". <img width=\"549\" alt=\"image\" src=\"https://user-\"\nI fear if we hide it when there is nothing to toggle, users won't discover that there is a (keybindable) command for this functionality. How did you find this command? Was your cursor in a conflicting region?\nIt wasn't in a conflicting region at the time\nLet's show an error message if this command does not do anything.\nIf I run the toggle command in an invalid region, I get a notification, which is good. But, if I then toggle a valid region after, the notification still stays on the screen for a while. Can the notification be hidden when the user toggles a valid region?\nCan notifications be removed? I thought that is not possible. Also, I think this use-case is quite rare. The notification is just to teach that you should be in a conflict."}
{"_id":"q-en-vscode-816c581136f77ec9fc9ab96d22ccaf1aa6791ee5bed4891d7c1ae02a3ec4b2e4","text":"<img width=\"603\" alt=\"image\" src=\"https://user-\"\n!"}
{"_id":"q-en-vscode-8ef0af1988b358ee884fe5ac72cf076b4c177ddcc543044c2da87427a8265afc","text":"Refs: I do not understand this condition: How does that make sure a method exists? The only check for this that is valid currently is , the is just a detail of the file system and not a capability.\nThis looks like an actual typo to me or missing code review, not sure the consequences, as it could lead to errors when wrongly asserting, so maybe moving to August for investigation.\nTLDR I misunderstood how the declared capabilities relate to the actual declared methods and the meaning of . For context, this change came in through and the intention of this change was to support the scenario where an has declared but not : since is optional in the interface and the capability requires both and to be implemented: I made this change because when implementing a with but not , this would result in throwing here: I see how it's possible that a which has check does not have a method implemented. Additionally I see that we have other examples of how to handle the requirement of a readonly filesystem which implements both and , e.g.: I am reverting Apologies for the incorrect code change and thank you for pointing this out."}
{"_id":"q-en-vscode-1a66e46917b58668938b28319b2b318f0b249526d38a5283f63098c6d98754ae","text":"Latest insider. <img width=\"433\" alt=\"image\" src=\"https://user-\"We should make sure the delete action is clearly visible."}
{"_id":"q-en-vscode-784fd7ea4629417c32266f19022d68407c3212502ad127c318d9d11c1ed7615f","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes/No <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.71.0 Steps to Reproduce: a codespace a merge conflict the conflict file missin the button <img width=\"1461\" alt=\"image\" src=\"https://user-\"missin the button <img width=\"1535\" alt=\"image\" src=\"https://user-\"This is context keys, current file scheme is <img width=\"673\" alt=\"image\" src=\"https://user-\"<img width=\"703\" alt=\"image\" src=\"https://user-\"but the scheme is <img width=\"316\" alt=\"image\" src=\"https://user-\"so the button is missing because is <img width=\"1148\" alt=\"image\" src=\"https://user-\"\nI can confirm that \"Open in Merge Editor\" & \"Accept Merge\" buttons are missing when VS Code is connected to WSL via Remote - WSL extension.\nThanks for the proper investigation!\nLet's make this a 1.70.2 candidate\nFixed in main but keeping this open to cherry-pick things onto release/1.71"}
{"_id":"q-en-vscode-34ab045df847c030601a6843351febf3088f38752ebc59a481286b248e43a3e5","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes/No <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: Stable & Insiders OS Version: Any Steps to Reproduce: Jupyter extension the following cell against a Python kernel The expected output is fairly simple: ( can appear some where in the middle) However VS Codes output is incorrect We are missing the line breaks (looks like this PR seems to have broken that) (using blockinline-block` fixes this). We have duplicate entries (as you can see below, we have multiple times) i think this has to do with the fact that we're trying to re-use the previous output and append the new stuff into the previous output element Note: This can be repro'ed in stable as well, but if i were to remove the code that tries to append to the previous element, then it looks as expected. Not sure when this part of the code broken though. Didn't go through the change log to determine this, you might have a better idea.\nnote that this is from the test, so that could probably be modified to check for the fix\nIt looks like this was briefly solved (at least for me as I reported in ) but the problem is now reappearing.\nI agree, I can replicate the same bug again, hence reopening this issue. Looks like subsequent change regressed the fix.\nseems to be a worse regression now (only in insiders) and repros without needing to do anything with background threads (unless I just have some bad setting): !\nI'm seeing something similar with the JavaScript/tslab notebook I use for teaching async and promises. Previously, the output was inconsistent across runs with sometimes missing or duplicated output. Now all the expected output is present but with line breaks appearing only after some async code (e.g. ) is executed. <img width=\"661\" alt=\"image\" src=\"https://user-\"\nI reproduced output shown by above with missing new lines in code-insiders at ~14:00 UTC +1 (CET). In a code-insiders session that has been running last 24 hours. Restarting code-insiders approximately an hour later (15.00 UTC+1), the output is now shown with correct line breaks, see screenshot below. Version: 1.75.0-insider (user setup) Commit: Date: 2023-01-09T05:29:27.679Z Electron: 19.1.8 Chromium: 102.0.5005.167 16.14.2 V8: 10.2.154.15-electron.0 OS: Windows_NT x64 10.0. Sandboxed: Yes Output after code-insiders restart: !\nconfirmed that this no longer seems to reproduce on the latest.\nAdding label as confirmed above."}
{"_id":"q-en-vscode-ccf0e61816584991171c0737c8fd51533c18097c72a4bfbef39690c3399d4eaf","text":"shell integration enabled a zshell terminal just exits instantly a zsh terminal now works. However, some shell integration markers are still present, and my .zshrc didn't load -- no theme and no aliases are available <img width=\"359\" alt=\"image\" src=\"https://user-\"vs in iterm: <img width=\"246\" alt=\"image\" src=\"https://user-\"\nI have seen issues where this is happening, but haven't been able to reproduce yet myself\na recent change which could've caused this is\ndoes your set your on the machine that you're seeing this on? what's your in the external terminal?\nis unset when I'm using zshell in iterm\nThanks for creating this issue! We figured it's covering the same as another one we already have. Thus, we closed this one as a duplicate. You can search for . See also our . Happy Coding!\nThis issue closed? This problem still seems to be occurring. Version: 1.72.0-insider (Universal) Commit: zsh 5.9 (x86_64-apple-darwin21.3.0) and I don't set ZDOTDIR variable\nThis issue still exists https://user-\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!"}
{"_id":"q-en-vscode-517294a24c6d6a5beb9001b479073027d481666667efc6ab0e8cc440e6f9aacf","text":"This throws in the monaco editor. To see the exception: clone as a sibling to vscode run run run open http://127.0.0.1:8080/monaco-editor/test/manual/?editor=src observe the error: <img width=\"1323\" alt=\"image\" src=\"https://user-\"\nI get this so I think I need some verification steps:"}
{"_id":"q-en-vscode-6fc34413c940a175667bf2eea47b04d021c299d7afcae21333d6cda76ba73af9","text":"Type: <bBug</b! VS Code version: Code - Insiders 1.72.0-insider (, 2022-09-23T05:17:37.024Z) OS version: Windows_NT x64 10.0. Modes: Sandboxed: Yes /assign <!-- generated by issue reporter --\n/assign\nCompletely agree on this. Tree ident guide highlights are pretty awful throughout the hole app, except the editor itself because of the color lines.\nGreat find!"}
{"_id":"q-en-vscode-22fcbbe4d88cb0d97c6725ab6d9bf0238cc5a5a55ebe87fee0ace85fc4e00cc5","text":"Type: <bBug</bLight+ on Insiders ! Light+ on Stable ! Dark+ on Insiders: ! Dark+ on Stable: ! VS Code version: Code - Insiders 1.72.0-insider (, 2022-09-23T15:06:00.185Z) OS version: Windows_NT x64 10.0. Modes: Sandboxed: Yes <!-- generated by issue reporter --\nregression from"}
{"_id":"q-en-vscode-9a878a5589f6cfec7579a3a4b3e72d40210b1c8b0b9048b91563916fcede5078","text":"Testing on run continue on clone locally I see a browser-native dialog to open desktop I see a custom message that just reads \"We tried opening Visual Studio Code - Insiders on your computer.\" which is confusing Unsure if we have to show the message or not but its wording needs to be improved. I mean tried and succeed in opening desktop so don't confusing me with this message. If we cannot prevent it I'd suggest to rephrase it and assume that it worked and give me a close option. Today, there this only covers the \"didn't work cases\" despite success https://user-\nWe cannot tell from the browser whether opening the protocol url succeeded, and the current wording for this dialog is based on a similar dialog from Office which also unconditionally appears in the browser: Maybe renaming 'Continue here' to 'Cancel' or 'Close' would help? Some other ideas for wording--when you launch a Zoom meeting from a browser in the desktop client, it says !\nI can accept that but the wording should be positive/happy. Like: \"This session now continues in VS Code Desktop. You can close this window, all your changes have been carried along. Try again if VS Code Desktop didn't start yet.\""}
{"_id":"q-en-vscode-b4352a3ecfa04a6f29d4cecd4c0ed58a021d959a37572784196fa71e1fc15672","text":"Testing It would seem to make sense to start at the first unhandled conflict (currently starts at the top of the file).\nThis would also help a first-time user understand what they are looking at. Without a conflict being highlighted, it is easy to get confused over all the filled and unfilled rectangles: <img width=\"2060\" alt=\"image\" src=\"https://user-\"With the first conflict selected this is less the case because the 3 unfilled rectangles stand out: <img width=\"2061\" alt=\"image\" src=\"https://user-\"\nBut the first conflict is revealed automatically, right?"}
{"_id":"q-en-vscode-371fb3974a7fe07cd861b55be9c0bd9286bb799ef80091b8c6dae0353f131309","text":"Testing According to , we can have git checkout a specific branch as part of the cloning command. That way we can avoid slowness as well as a double progress indicator.\nVerification From your browser, enter . This should clone the repo and checkout the branch , and there should only be one progress notification for the clone progress, there shouldn't be a second progress notification for the checkout\nI just get messages like Version: 1.73.0-insider (user setup) Commit: Date: 2022-10-26T09:31:30.519Z Edit: It's missing url=. Will try again.\nThanks"}
{"_id":"q-en-vscode-018c91068ac8836c4fe02a141b94b0f2e04b980388314de069a77932ac054609","text":"Here with the long path a \"...\" would help: <img width=\"673\" alt=\"image\" src=\"https://user-\""}
{"_id":"q-en-vscode-46edf9977c73151358daae7ea75752c048b9c83aa1ebb05bcb41d3ab2498c011","text":"Just debugging in Insiders, it looks like the opposite: the free port matcher doesn't match anything unless the output is more than its configured 20 lines <img width=\"499\" alt=\"image\" src=\"https://user-\"If I add then it works. THat should be enough for me to verify the UI side of this... Originally posted by in\nremove this completely\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!"}
{"_id":"q-en-vscode-a38f22c80796a1ba9c2a99777a2ba235466f8fc177fe0f90d5dfd9761e51aa3a","text":"Type: <bBug</ba new profile from empty click on activity bar \"Move Primary Side Bar Right\" ! VS Code version: Code - Insiders 1.72.0-insider (, 2022-09-29T08:07:23.017Z) OS version: WindowsNT x64 10.0. Modes: Sandboxed: Yes Remote OS version: Linux x64 5.15.62.1-microsoft-standard-WSL2 Remote OS version: Linux x64 5.15.62.1-microsoft-standard-WSL2 <details<summarySystem Info</summary </detailsExtensions: none <!-- generated by issue reporter --"}
{"_id":"q-en-vscode-538819e20959a9c2ccb6e35f8ceb98b1eae2be5ce2df96ddb1ba5b6d55fa4d39","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --The debug button is really small. Can it be big? The drop down menu as well as the other debug buttons could be on a toolbar under the debug title and this could be much bigger and easier.  Or the play button could be bigger button like the 'commit' button: <img width=\"422\" alt=\"Screen Shot 2022-10-10 at 8 00 21 PM\" src=\"https://user-\"\nThe clickable area of that button is 16x20. Most menu buttons are 22x22. I haven't had an issue with this but could it be a ux or accessibility issue? Here's what it would look like if it were 22px wide <img width=\"312\" alt=\"image\" src=\"https://user-\"The big button is an interesting idea but I'm not a big fan of that button. I find it really aggressive and loud and I disable it in scm.\nMaking it 22px makes sense. I think it should be consistent with the other button sizes across the workbench.\nthanks a lot :) it has been the fastest fix on history :))\nDid slightly less than the full 22px - due to the border and the background color, I think that feels too wide. But this should help."}
{"_id":"q-en-vscode-62811875f4277937aca9549355a427463a769d83da8f5446e5f7d1fc39fc8f1f","text":"A remote code execution vulnerability exists in VS Code 1.71 and earlier versions for malicious notebooks. These notebooks could use command uris to execute arbitrary commands, including potentially dangerous commands The fix is available starting with VS Code 1.71.1. The mitigates this attack by performing input validation on the URL pointing to the repository to be cloned. Do not open notebooks from untrusted sources. The patch for this can be found at An advisory for this can be found at MSRC details for this can be found at\nthe docs and advisories say it's fixed in 1.71.1, but it's only fixed in 1.72.1"}
{"_id":"q-en-vscode-2e06a7c2af194b86c3bd48836ef2239987a5b420dc5f1afcaf988d5e4e347dd5","text":"Since long we automatically profile the extension host when it becomes unresponsive. In that cases we send a signal via telemetry and in bad cases we ask users to file issues. During that we ask them to attach the corresponding profile but we have mixed results with that. So, in addition to the existing flow we want to log a richer telemetry event that uploads the heaviest sample trace from the profiling"}
{"_id":"q-en-vscode-476b9925b0343f91a998ae790050cbc0fb4262d7e6cd2abafa6f477b677cf1ce","text":"When the /gifPlease command is used the issue owner sometimes replies to the GH notification email and attaches their files. The attachment gets lost. Please update the instructions to say that files must be submitted via the GH website."}
{"_id":"q-en-vscode-62516b452b24fae731f075de23c177ebcc958c044ec93f754eb7da2b92c1b3db","text":"Just upgraded to macos 12.3, try to ENABLE System Integrity Protection if you have disabled it before: Originally posted by in\n[(42)] opendir /Users/xxx/Library/Application Support/Code/Crashpad/attachments/-107c-4a3e-8990-: No such file or directory (2) objc[2187]: Class WebSwapCGLLayer is implemented in both /System/Library/Frameworks/WebKit.framework/Versions/A/Frameworks/WebCore.framework/Versions/A/Frameworks/libANGLE- (0x7ffa4f13a8f0) and /Applications/Visual Studio (0x115684ce0). One of the two will be used. Which one is undefined. [(171)] EXCCRASH should not contain exception 12 [(70)] exception EXCGUARD invalid in EXCCRASH [main 2022-11-01T07:19:07.195Z] CodeWindow: renderer process gone (reason: killed, code: 9)\nLocal OS Version: macOS 13.0 (22A380)\nIf the issue persists with latest stable , can you provide the crash logs following the steps at , thanks!\nmy version is 1.74\nIn your case, is SIP enabled or disabled ? Does it crash when you run with ?\n! The same mistake. Are you running normally on the latest Mac 13 system\nI think is not specifically a vscode problem. Im getting this for every Chromium <108 based electron app like: Logitech G-Hub I replaced the with the latest it working since but please don't do that it requires SIP disabled and this is not a solution not even a workaround. !\nThis \"not even a workaround\" works here as well. Thank you for sharing! SIP was reduced beforehand due to OCLP, so I don't know about it. But definitely not recommended reducing/disabling it! For anyone curios:\nIf you are using OCLP with reduced SIP you just need to add ipccontrolport_options=0 to the nvram boot-args.\nUpstream issue\nsudo nvram boot-args=\"-v csenforcementdisable=1 amfigetoutofmyway=1 keepsyms=1 intcoprocunrestricted=1 amfiunrestricttaskforpid=1 amfiallowanysignature=1 csdebug=1 ipccontrolport_options=0\" exec this command,program is ok\nare there steps for this?\nthis requires booting macOS with custom arguments specifically SIP and AMFI disabled, better for authors to verify this issue. can you help verify if this issue is addressed with our latest insiders without having to add to the boot args, thanks!\nAre there any conclusions? I also have this problem\nthis will be fixed in our next stable release\nToday it keeps failing. !\nplease create a new issue with crash dumps based on the steps here , thanks!"}
{"_id":"q-en-vscode-f0d717c8fb99480a165b548ae5e2a45921d709f45fe2684e66f8495ae5988691","text":"When running VS Code (or the Monaco editor) all theme colors are exposed as CSS variables. That means components don’t need to use JavaScript to read and set colors but can simply rely on CSS. When a theme changes, the variable values automatically update. Also, with CSS variables you have less code and everything in one place. The (long) list below is all users of . Those that only set/update colors are candidates for adopting CSS variables The names of the CSS variables can be \"derived\" with the following logic the theme color id and replace dots with slashes For instance the color id is exposed as CSS variables . Find a sample adoption in this commit: [ ] [ ] [x] [x] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [x] [x] [x] [x] [x] [x] [x] [x] [x] [x]\nHi, is this issue still up for grabs? I would like to be assigned to one of the bullets.\nYep! Just feel free to grab whichever one you want and open a PR for it :)\nHi! I would like to grab\nHi How did you linked your PR in every line of the description in this issue? I am not able to edit the description and I opened a PR for 12 hours ago. I guess edited the description to add your PR but not mine?\nMaybe I mentioned this issue in the pr?\nYes, I was just wondering if there was an automated way of doing this, but I dont think so. Thank you!\nI wrote a small helper that generates dynamic CSS rules but with the css variables filled in: You can use this for simple cases. Hopefully, it prevents some typos. The code is at:\nTerminal is done, the following are intentional exceptions: We need to pull the color into TS to pass to as it uses it in canvas calls. We need to apply 0.5 opacity to the resolved color for the background, using won't work as it will fade the text.\nTunnel view, comments, and tree view are done.\nwhat is the recommendation when more than the color is part of a ? To give an example: Here, we conditionally make an element if we have a . We cannot just blindly move that over to CSS because we then lose the check for the color.\nI don't know. Has to stay in code....\ncan I listen to a change for a color key via theme service to conditionally toggle a class based on a color being defined or not?\nYes, and\nIf anyone else is still working on this, you can get completions with this handy extension\nWhat is the recommendation if I have to hardcode a codicon css, for eg., I am deriving the codicon class programatically as follows If I have to move this to CSS then I have to hardcode the codicon class as follows Is it ok to hardcode codicons like this?\nLooking at where verifiedPublisherThemeIcon is used, it's always in a dom node that also has the class . Better avoid the codicon class name, but use\nInteresting issue where a migration to CSS variable caused a bug (). It looks like some visual behaviour is defined by the OS/runtime and not by default user agent CSS to fallback to. So if your CSS variable ends up as , you get some default (?) and the result is that the OS styling is disabled. Relevant SO for my case:\nI looked over mine again and they are all there still intentionally: I don't think we can get the same effect of in CSS using the variable. Same as above. This is needed as the color is handed off to\nplease check those occurrences anyways so that we can remove them from the todo list\nFor , I couldn't find references to . For , we use the RGB values from the foreground var and we use a custom alpha. Since this only applies to the text, I don't think that there's a CSS-only equivalent. We could create a new var for this or just leave this as is; I'm not sure what's best in this case.\nyou're setting there so you can probably use a plain style, unless the element also has a background color as that would also apply there.\nThe element also has links, which have a normal alpha, so then the link opacity is affected.\nyou could have another rule with higher specificity that re-applies the link color?\nFor you could define a new (terminal hover specific) color. The default value of the color can express that it is a function of To access a color value use\nAs discussed in standup, tabs have quite an involved CSS ruling that involves not only colors but also other properties. I would like to keep it this way via and not invent various class names to get a similar effect to make maintenance easier there.\nClosing this as I believe we are done here"}
{"_id":"q-en-vscode-4c9ff479c62adbd9ab836e62209b1c6784c0dabbb48a7c055b56d80c4cc856f7","text":"We very likely do not need to ship all of this in the web bits: This was already once done before in , and when dependencies got renamed this was once again forgotten about."}
{"_id":"q-en-vscode-af25546298936ca263839f82facc7a3bdaf61021f927d83acddbbc7b6b24f6df","text":"+F to open the editor find widget. Focus should be on the find input box. case-sensitive search. to the case-sensitive search button. :bug: Keyboard focus is on that element, but it's not clear that keyboard focus is on that element. !\nworks fine on my machine with Light+ and Dark+ theme, what theme are you using? ! !\nThanks! I guess the problem depends on the theme. The problem seems to occur on at least Quiet Light, Solarized Light, and Monokai.\nCan confirm the the issue in those themes on my windows machine. I never sent a pull request to vscode but that seems easy to solve, the question is what highlight color would fit best each theme, i might take a look later unless there are some rules about how to set some of the theme colors... the colors effecting this seem to be \"focusBorder\" and \"inputOption.activeBorder\" I would love to offer some new colors later today/tomorrow and pull request them\nI propose giving a bit of contrast to the \"focusBorder\" I think that focus border should have a bit of contrast from background as this is marking elements the user needs to spot quickly, even in themes with less contrast light quiet light i think this is best to increase contrast in those areas, this is only showing for elements put in focus and wont change the feeling of the theme by much. Here are the new colors i propose and will commit and pull request later if those look acceptable For Quiet Light: ! For Solarized Light: ! For Monokai: !"}
{"_id":"q-en-vscode-eb1db42d074a4d7da2094d7d48755579124d6e8784dbfe4a32310c3f3cdd25b4","text":"open vscode for web on macOS ctrl+click into a tab that is not active context menu opens but also the tab opens fyi this is similar to and requires the same fix. Interestingly on Electron this does not seem to be an issue, maybe produces context menu event in Electron but not in Web?"}
{"_id":"q-en-vscode-c46eda2c4ef7c6626ec01f599dcdea61c6a8aec8bbfc238e1d2fb0bf15baaf99","text":"When rendering Markdown preview, Unicode newline characters U+2028 LINE SEPARATOR and U+2029 PARAGRAPH SEPARATOR are replaced with an empty string at the source preprocessing stage. This behavior was introduced in to but it breaks the document in multiple ways: double-click word selection does not work as expected, screen readers may announce the text incorrectly, line wrapping and hyphenation does not work correctly, markdown-it plugins that rely on or may work incorrectly, even linkify fails: and so on. Since then it was fixed at the markdown-it side, and is no longer needed: It would be better to revert this change, and keep the Markdown source as is before feeding it to the parser. Steps to reproduce: a Markdown document like this: preview: Expected result: LS and PS are displayed as is, links end at Unicode newlines. Actual result: no Unicode newline characters displayed, links also consume the immediately following word. <!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: OS Version: WindowsNT x64 10.0.\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!"}
{"_id":"q-en-vscode-9acf84e53eeb7e8fc8421877b9a5c1154d7a5c8564b64d5ce38ad59f7cedca4d","text":"Does this issue occur when all extensions are disabled?: No VS Code Version: 1.73.1 OS Version: Windows_NT x64 10.0. Steps to Reproduce: ! to the extension bar. to hover some items. Some one may not show the default scroll bar, but the problem is exists. ! I have tried to fix it. It seems caused by the visual bug from the issue . The scroll bar will show when the content is too long. This solution is great, but it causes another little bug in the screenshot. I think it's unwise to remove the tag, so I tried another way, just add to the tag : ! I'll make a PR after this issue has submitted.\nI can't reproduce this srollbar for all extensions in the list: !\nThis bug is not 100% recurring. Some extensions' descriptions may just exceed the calculated.\nWhat's ur testing extension id?\nThese extensions have the problem on my machine.\n! !\nMaybe it's related to high-resolution screen. I'm sure it's an accidental problem, but it still exists. If you have the conditions, try adjusting the screen resolution scale to 150%.\nthere should be some bugs when doing sub pixel sampling whether it's related to windows DPI settings. can you reproduce this regularly without changing DPI?\nCaused by can you please have a look at this?"}
{"_id":"q-en-vscode-40cf4adf6fbd65b84c858df9dc05990cde3d067759d87a362f391b04402f8c35","text":"Ref Right now I see: as a menu entry and command name \"Get Started\" as the tab name when looking at the welcome page \"Show welcome page on startup\" and \"Reset welcome page walkthrough progress\" \"Getting Started with VS Code\" as a walkthrough name There's a mix of \"Welcome\" and \"Get Started\" usage that makes it hard to differentiate between concepts. We could either revert back to \"Welcome\" for the welcome page itself or rename it to differentiate from the first walkthrough. Thoughts <img width=\"424\" alt=\"CleanShot 2022-11-21 at 10 19 04 src=\"https://user-\"<img width=\"200\" alt=\"CleanShot 2022-11-21 at 10 19 12 src=\"https://user-\"<img width=\"516\" alt=\"CleanShot 2022-11-21 at 10 19 20 src=\"https://user-\"<img width=\"273\" alt=\"CleanShot 2022-11-21 at 10 19 23 src=\"https://user-\"<img width=\"625\" alt=\"CleanShot 2022-11-21 at 10 19 56 src=\"https://user-\"\nfor historical context on naming here\nYeap..I believe changing the label in the tab to \"Welcome\" instead of \"Get Started\" is a good option - It should say Welcome on the tab here instead of Getting Started: ! When the user navigates away from the welcome page, the tab name should change to the walkthrough's name: It should say Learn the Fundamentals on the tab here instead of Getting Started: ! And this breadcrumb should say Welcome: !\n\"Welcome\" was the old page and we \"Get Started\" as new name during development. \"Get Started\" did fit better for the walkthrough style, was a more action-style verb and used in many other apps. Not sure if more thoughts went into it, as the mockups that started in were before my time and already used \"Get Started\". I haven't many users ever get into Get Started or Welcome as command, so renaming would be low risk.\nIs there a user issue mentioning the confusion? Asking to understand if we also need to account for user keyword search behavior in the command pallet when looking for walkthroughs or the Getting Started page after its been closed. Currently, most walkthroughs use the \"Get Started\" terminology - Searching for \"Get...\" in the command palette surfaces the walkthroughs. I wonder if search terms are influenced by the page on which users first saw the walkthroughs and will the naming change have implications on the keywords to find walkthroughs?\nNo specific issue that I know of, I'm just spotting the issue of the strange overloading myself. might know if we have any hard data on people searching for the command. But I'd also assume it's quite unlikely compared to going to the \"Help\" menu, for example."}
{"_id":"q-en-vscode-6b94d0b9945fe09937f6d10a463f20d6a8b420bc32eb40282cd31db76bb8e061","text":"I suspect this comes from\nYou're right, this is: Right now, the module assumes that it will be called in the following order: It is possible that if a is skipped by the browser that we don't mark and that would cause an error at\nGood morning I'm having trouble logging into the app. I get the following error. Uncaught SyntaxError: Failed to execute 'measure' on 'Performance': The mark '<REDACTED: user-file-path' does not exist. I hope you can help me with the solution. thank you."}
{"_id":"q-en-vscode-9bf64aa338b731d7a9477f6adc6de17cec973bec4fcc5549b86a472c7b2e5395","text":"I've seen this a lot over the past month and it seems like a major issue for notebooks, please consider investigating further for November ! ! ! ! While running integration tests today (23rd November): ! Version: 1.74.0-insider (user setup) Commit: Date: 2022-11-23T05:25:25.108Z Electron: 19.1.3 Chromium: 102.0.5005.167 16.14.2 V8: 10.2.154.15-electron.0 OS: Windows_NT x64 10.0. Sandboxed: Yes\nit's likely related to the position performance optimization this month.\nCan you try debugging. I can never reproduce this The best place to start debugging is: Let me know if you want to pair on this later this afternoon too\nWas finally able to repo this but not reliably. It seems to rely on a new being created, but this doesn't always happen. The most reliable repo for me: with a notebook and text editor in a single editor group the notebook editor down to create a new group below the text editor grab the notebook editor and drag it right to move the bottom editor to the right I think the root cause is that the notebook widget is passed the wrong position. I'm going to work around this by making the notebook editor ignore the position it is given, however this will also undo the performance fixes we made this iteration for notebook layout I don't fully understand the grid layout code but it seems that is being set incorrectly in cases where a new BranchNode/SplitView is created. Here's what I see happening after step 3 of the repo above: setter is called calls which creates a new the constructor, a new Split View is created is called on it. At this point, of the split view is: the time layout is called on the notebook editor widget, is computed as zero. This results in the wrong layout. Looking at the code, we can see that is passed in as the absolute offset in step 4: <img width=\"1104\" alt=\"Screenshot 2022-11-30 at 12 48 54 PM\" src=\"https://user-\"If I then slightly resize the window, here's what happens: call on the split View for the new group now is: we layout the notebook editor, is now correctly computed as\nCreated new issue: , thanks for the steps"}
{"_id":"q-en-vscode-038269ffa764df136ca53cdc8935004833a0903c61fa1d33992a78b38875c41a","text":"Very often I loose the yellow markers in the editor for references, see here for example: ! Here I find references of  and quickly dismiss the results of the one file before clicking on the other. Both files were opened before.\nDidn't reproduce for me. Please update with repro steps\nsteps be in a multi root workspace with and close all editors open click on  of (at the end of the file) run (I have it bound to command) expand by clicking it once in the references view click the first result remove the 2 references by clicking on the X starting from the first click on the last result from no more highlights in the editor at all Full video: !\nWhen removing the -entry (all together or all by removing all its children) I am not (always) getting the selection change event.\nanything special for reproing this? I haven't been able to hit it.\nSelection change event should be fixed now.\nThanks\nfyi noticing a few behavioural differences now, which maybe we tolerate but users might not: clicking on a file name now opens the first reference, previously I believe it was not navigating but just expanding/collapsing (see video 1) triggering find references seems to reveal a reference in another file in some cases (see video 2) ! !"}
{"_id":"q-en-vscode-f7b1067c2e876c94dd4b466b8029ccfdf41498e00f1dea612d2a2dbfcf948fe9","text":"A bug in our tracker for dirty models, somehow never reported but obvious in an issue I noticed with Copilot. The end result is that you see a dirty indicator e.g. in the explorer but no editor that is opening and associated with it. Steps:\nTried to look at the steps to repro but seems the link to the copilot issue is broken or at least I am not able to access it.\nI'm still getting this with copilot nightly in insiders <img width=\"652\" alt=\"image\" src=\"https://user-\"\nPlease note that my change does not fix the Copilot issue. The steps to verify are unclear: it is expected when you see an untitled editor opening. You must test with auto save on.\nSetting back verified label based on the screenshot."}
{"_id":"q-en-vscode-f4584483eeacc9b6b578d477980cde5539e707f5312e45cfc98ca6e4f06a758f","text":"Running out of sources I see this:\nThis means that the contribution isn't getting evaluated before is called. Not too familiar with how / when is handled, but we call when the is created\nI think perhaps moving that to the top of the constructor (before registering the command handlers) will fix this\nThis happens on window open\nand the quick fix does work for that extension !\nSeeing this in error telemetry for Insiders from 12/23:\nThis is the 6th error by machines in telemetry https://vscode-\nI cannot reproduce this in Insider's. did you notice this after doing anything in particular?\nI got this error when using remote dev container <details<summaryWindow</summary</details\nFixed in\nActually still found it in my devtools today:\nI can't reproduce, but I can simulate the race by adding to 's promise in\nHI, im having the same problem here 2023-02-16 17:10:14.453 [error] Error: No registered selector for ID: ms- at R.registerQuickFixProvider (vscode-file://vscode-) at r.$registerQuickFixProvider (vscode-file://vscode-) at m.N (vscode-file://vscode-) at m.M (vscode-file://vscode-) at m.H (vscode-file://vscode-) at m.G (vscode-file://vscode-) at vscode-file://vscode- at g.invoke (vscode-file://vscode-) at n.deliver (vscode-file://vscode-) at a.fire (vscode-file://vscode-) at r.fire (vscode-file://vscode-) at Z.onmessage (vscode-file://vscode-) I got vs code 1.71.1 running on W11 21H2 How i am supposed to fix this, or it's not a problem and vs code is working normally? Thank you!\nUpdate your VS code version, where this is fixed\nSorry, i typed incorrectly mi vs code version. It is the latest one 1.75.1, i reinstalled it today.\nI can still reproduce in latest stabel 1.75.1 https://user-"}
{"_id":"q-en-vscode-8031a9c7f59199959bf982fa509f67fd97b8c3bb4f552f0f1f4ef7c6e2a6094f","text":"Type: <bBug</bis enabled. Ensure is set to , or (any option that will restore the last open window after quiting and re-opening VSCode). a terminal in the editor area, e.g. using the command \"Terminal: Create New Terminal in Editor Area\". Type something into the terminal e.g. . VSCode via \"Developer: Reload Window\". Note that the terminal tab in the editor area is still there (restored) after the window is reloaded. This is the expected behavior. VSCode then reopen the application. Note that the terminal tab is not present (is not restored) when the window is restored (and since there is no terminal tab, there is also nowhere to restore the session to ) . This is unexpected behavior as previously this terminal tab would be restored into the editor area, with its session history, even after quitting and reopening VSCode. Another terminal persistence bug on 1.74.0, also related to killing terminals: VS Code version: Code 1.74.0 (, 2022-12-05T16:37:54.581Z) OS version: Darwin arm64 22.2.0 Modes: Sandboxed: No <details<summarySystem Info</summary </details<details<summaryExtensions (16)</summaryExtensionVersion ------ vscode-neovim0.0.94 calva2.0.321 file-browser0.2.11 portal0.35.0 viml0.1.7 prettier-vscode9.10.3 vscode-htmlhint1.0.4 vscode-sshfs1.25.0 ssh-config-syntax-highlighter0.0.4 sqltools0.26.0 sqltools-driver-pg0.4.0 errorlens3.6.0 learn-vim0.0.28 vspacecode0.10.10 whichkey0.11.3 markdown-all-in-one3.5.0 </details<!-- generated by issue reporter --\nI cannot reproduce that in Insider's https://user-\nhave you typed in the terminal before reloading? if not, that's by design\nPardon me I was unclear in the original writeup. The problem occurs in step 4 and is now reflected in edits above. Thank you! Regarding the direct question: yes, I have typed in the terminal before reloading.\n/duplicate\nReviving editor-area terminals was working great in 1.73.x. This was marked as a duplicate of a feature request. Is it not a regression?\nCreated a PR with a possible fix\nVSCode v1.74.2 occurs the same problem, does the fix takes effect or I need to config some settings?\nThe fix will be available in 1.75 release or you can try it in insiders"}
{"_id":"q-en-vscode-6d84dc58a2b8df395473e9e32704cf00cdfc56b9b6568b8b878ad851f26bb04b","text":"a change to the file and save, do not commit the diff view from the SCM view up and down in the notebook diff view :cheese: see cell metadata area skittering on screen (the recording below does not capture how frequently the skittering actually happens) https://user-\nJust ran into this again: https://user-\nI opened for that, I think we can actually just reopen this one\nThis issue should be gone now since we fixed the horizontal bar issue."}
{"_id":"q-en-vscode-62d89015bf60cffdf93990e3fc6c7ef9d8b3e3c8ae95eaff68e2f1c8ce27f656","text":"re The automatic perf monitoring has shown a high number of calls to . This means the call/function itself is slow or it is called at a very high rate (so that the profiler always sees it running) The most common stacktrace via which this is reported comes through .\n/duplicate\nThanks for creating this issue! We figured it's covering the same as another one we already have. Thus, we closed this one as a duplicate. You can search for . See also our . Happy Coding!\nI don't think this is a dupe because the callstack comes in via which I don't see in the other trace\nBut I think the root cause is same for both that is slow as mentioned in the summary of the other issue. Hence I marked it as duplicate.\nPfff - that would mean it is catastrophically slow. My suspicion is that it is just slow but that comes in combination with being over eagerly called? Maybe some event leakage or some over eager calling through some other means. This symptom might disappear when has become fast but I think we need to understand why it ends up on a hot code path in the first place\nWill investigate why this is being eagerly called too."}
{"_id":"q-en-vscode-30c7bfa18b69781cbccf501a7f39c6bcb5e4feff64389e6cf58ce70538775bb7","text":"I think these used to be blue: <img width=\"297\" alt=\"image\" src=\"https://user-\"vs <img width=\"193\" alt=\"image\" src=\"https://user-\"\nUsing the \"Default Light+\" theme this is working (on main). So it is either already fixed ( or theme dependency (unlikely) but what theme is this? <img width=\"1004\" alt=\"Screenshot 2022-12-21 at 17 59 20\" src=\"https://user-\"\nRan into the same issue with my theme. After a recent vscode update the coloring of the icons is gone. I'm using the theme.\nI also cannot reproduce from source, but it still reproduces in latest Insiders using :\nToday's insiders also doesn't have colors for me, so this might be a packaging issue or something specific to my setup. Are you able to reproduce with Insiders? <img width=\"661\" alt=\"image\" src=\"https://user-\"\nYeah, I can repro in insiders with an empty profile\nThis also happens in other places that show symbol icons and is only happening because of how things are ordered in the packaged CSS. what's the procedure to fix this? <img width=\"1250\" alt=\"Screenshot 2022-12-22 at 17 32 26\" src=\"https://user-\"\nWhile the CSS rules in a single file will remain in order when being loaded or packaged, there are unfortunately no guarantees to the CSS rules order of multiple files. That is because CSS files can't express dependencies on other CSS files, so they will be loaded or packaged in any order. The exact order depends on the internals of the AMD loader/bundler, but it isn't guaranteed to be consistent, as it can depend on fs read times, etc. I think it is a coincidence that this works every time when running from source. The correct way to fix this is to use to ensure the desired rule wins regardless of the css file order. I speculate that this only worked before (when CSS was generated from JS) because the dynamic CSS was injected in a tag towards the bottom of the element."}
{"_id":"q-en-vscode-71a4659611f5facad9b4f428b89ae760c3367b042e344c2b670b7c9a0fb0914d","text":"Type: <bBug</bAt some point recently the \"view\" -\"testing\" menu stopped opening the testing view. The menu item is present but clicking it has no observable effect. Additionally, the option is no longer available in the activity bar on the left. Note that it's not simply hidden. It's also missing from the context menu to show / hide items as well. This was working up until a few days ago. What does work is the command pallette. I can see from the logs that the extension is loaded, and by typing \"testing\" into the command pallette, I can still open the view, and from there on it functions normally. VS Code version: Code 1.74.2 (Universal) (, 2022-12-20T10:26:09.430Z) OS version: Darwin arm64 22.2.0 Modes: Sandboxed: No <details<summarySystem Info</summary </details<details<summaryExtensions (46)</summaryExtensionVersion ------ better-comments3.0.2 aws-toolkit-vscode1.60.0 better-tests0.3.1 atlascode2.10.12 emmet-jss1.0.0 path-intellisense2.8.4 dart-code3.56.0 flutter3.56.0 vscode-markdownlint0.48.1 vscode-eslint2.2.6 githistory0.6.19 xml2.5.1 gitlens13.2.0 prettier-vscode9.10.3 vscode-pull-request-github0.56.0 go0.37.0 rest-client0.25.1 git-graph1.30.0 dotenv1.0.1 djlint2022.10.1 vscode-docker1.23.3 isort2022.8.0 python2022.20.1 vscode-pylance2022.12.20 jupyter2022.11. jupyter-keymap1.0.0 jupyter-renderers1.0.12 vscode-jupyter-cell-tags0.1.6 vscode-jupyter-slideshow0.1.5 remote-containers0.266.1 cmake-tools1.12.27 cpptools1.13.8 cpptools-extension-pack1.3.0 vsliveshare1.0.5788 awesome-flutter-snippets4.0.1 printcode3.0.0 material-icon-theme4.23.0 gotestexplorer0.1.13 vscode-yaml1.10.1 LiveServer5.7.9 pdf1.2.2 cmake0.0.17 alphabetical-sorter2.0.1 vscode-maven0.40.2 JavaScriptSnippets1.8.0 markdown-all-in-one3.5.0 (1 theme extensions excluded) </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nWhich extension is providing this? Maybe it isn't activating successfully. The Extension Host channel of the Output tab may reveal something.\nTechnically, this isn't an extension. It's the built in testing support. I think it's this feature:\n/gifPlease\nThanks for reporting this issue! Unfortunately, it's hard for us to understand what issue you're seeing. Please help us out by providing a screen recording showing exactly what isn't working as expected. While we can work with most standard formats, files are preferred as they are displayed inline on GitHub. You may find helpful as a browser-based gif recording tool. If the issue depends on keyboard input, you can help us by enabling screencast mode for the recording ( in the command palette). Lastly, please attach this file via the GitHub web interface as emailed responses will strip files out from the issue. Happy coding!\nHi, I'm helping Iris look into the issue. Here's a video of what we're seeing. \"testing\" should show in the activity bar and in the context menu at the start of the recording but isn't listed. But the functionality that it should point at is still accessible via the command pallet, shown later in the video. !\nThat is because you moved the Testing view into the secondary sidebar. It therefore does not appear in the context menu on the primary sidebar. You can reset its location by dragging the beaker icon back into the primary sidebar, or using the command.\nfyi if we want to make view location stuff more clear here\nthere seems to be a bug still with the Testing main menu entry. It should open the view wherever it is. Repro: testing container to secondary side bar secondary side bar View Testing =no response\nOh, right, I was just going off the last screencap. That's strange, but should be fixable..."}
{"_id":"q-en-vscode-345d98ab135792333af4e1f8de8a8b850c77ecee6985e66ed3d23254a6818666","text":"start the GH auth flow from window1 focus window2. This step is not necessary if you use multiple screens and the browser is on the same screen as window2. But because I couldn't easily record both screens I manually focus here. continue the GH auth flow in the browser the flow returns to window2 instead of the originating window I grabbed the URL and it has a windowId in it: But perhaps the windowId gets lost somewhere along the way or perhaps there is a bug on the main process. Anyways, here is the recording: https://user-\nAlex and I talked about this and based on what I'm seeing, I don't think it's an auth issue and seems to be a Uri Handler issue. If it was an auth bug, I would expect the to be missing from the URL but that doesn't seem to be the case here.\nlogging revealed something that I then was able to verify in code: the change in changed the kind. When previously a was passed, now it is a . So the code here needs to be adjusted: I think the code could get some type love and not use for that matter. The full value I get logging in the router is:\nHm, I am actually not a 100% sure if that commit caused the regression because I also see you changed the receiving end, but the problem is still the of and being passed in when previously it was just the\nGreat catch, that's exactly right! I did not adopt the serialization format in the router. I'm open to suggestions, though I'm not sure there is anything that can be done here. The whole point of these implementations is to (de)serialize live objects into opaque JSON things. You can add a bunch of interfaces that lock things in, but in the end we still need the generic and methods, so I'm not sure we'd gain anything.\nBasically what  does by avoiding having to code out the underlying raw methods of the IPC flow and have just a typed interface for both sender and receiver. Yes, in the end you cannot know for sure what data is sent because we cannot enforce it, but (or something similar) at least allows to only have 1 place for types and when changing something, it would show an error on both ends.\nHow would you adopt that in the router?\nSo I've never been able to repro this bug... are there some steps that really focus on this fix?"}
{"_id":"q-en-vscode-cb9f3958114e8195979a8cffd55348ed4a3cd0a950e422946878bc61f44b744c","text":"Yesterday I ran VS Code Insiders from the Downloads folder, and I used . And today, after updating to latest Insiders I had to log in again. Here's what I noticed in one of the logs: I often also use which downloads various versions of VS Code Insiders and uses a custom user data dir. Often I end up being signed out . cc\nI have seen this error before and I had a time when I had to log in all the time, but not for a while anymore.\nNow again I hit this almost daily that I have to re-login.\nAfter 1on1, learned that this is caused simply by launching VS Code from a different path. A suggested fix is to not destroy existing credentials on startup when we fail to decrypt but only do so when we actually need to write new credentials to the store.\nWhat are some verif steps? I'm a little confused on what you mean by \"launching VS Code from a different path\".\nverification: Launch VS Code insiders, make sure you're logged in to something Close VS Code Insiders Open a terminal (not in VS Code Insiders) and run (This will spin up old insiders builds): Note that you are not logged in to anything (expected) in the terminal say or doesn't matter Now once is done open your VS Code Insiders You should still be logged in to your stuff"}
{"_id":"q-en-vscode-c5120303f0c53c7926e9daf4fc49ff69da91b58fdeb6a921d7a17b61dc6bcf0e","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.74.2 OS Version: macOS 12.6.2 Steps to Reproduce: an extension and specify a relative path to an in the in png format, and specify repository as . (Our repository is private) get a warning , despite the icon showing correctly in the marketplace and the VS Code extension manager (We also have images in our that we hosted externally. Our repository is private after all). Extension:\nI get the same warning ! despite my fields being defined as Source: Interestingly the icon shows up in the marketplace just fine. !\nI see the icon is served from the CDN now and we can remove the warning about the relative icon path. What about badges in the and images in the README markdown, are these also served from CDN?\nThose are all served from the raw GitHub routes, this is why vsce needs to know what repository hosts the sources.\nBut the icon isn't, not anymore at least... Our repo is private, the icon works just fine and is uploaded along with the vsix and hosted in some CDN. Images in the readme are indeed served directly so will require this check or an absolute URLs.\nAh! This is just the workbench linter. I also forgot that we actually show the icons served directly from the MP, not GH. Sorry about the confusion, thanks for following up This warning is unnecessary by now, let's remove it."}
{"_id":"q-en-vscode-b52a35be654ee984b25c723d8faf0a9b439599f9e505394bb9abc54877168d16","text":"This looks like we have telemetry loggers without appenders. Can we add a strict check or in the API impl?"}
{"_id":"q-en-vscode-c3f4e5eb8a3fada703f93da77eadc55e4cb2a94a7e4f513d136aab0ed898828c","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --When I use command in VSCode to disable the extension of the current folder/workspace it doesn't work. Other attempts: also invalid when adding options. Expected Result: Disable the specified extension. !\nTo extend it a little bit: These commands do work, when the workspace/folder is not yet open! (with disabled extensions) Even we did not found any documentation about it...\nRight, this works only when the window is not opened. I do not think we send the CLI args to the opened window. Also this option is to disable the extension for the current session only. This will be reset when user restarts VS Code without this option. I will update the help doc.\nThank you for your reply. I understand that the disabled will be reset when VS Code is restarted, but I still hope that the command will take effect in the current window, such as the following. Is it possible to solve it, or other alternatives?\nI am not sure if it is possible to update the args for the current window. CC\nOnly when you trigger a reload:\nThank you. So for the user, there's no way to solve this problem, right?\nYou cannot change the args of the current window as a user.\nI would like to add, that this feature is not documented online somewhere right? I could not find anything about it... I just stumbled over it. Can an extension do this? Well it would reload the screen anyway, which is what we would like to prevent...\nYeah not even internally you can change without a window reload at the moment unless a specific component supports this of course, which depends on the use case.\nQuestion: The user can simply disable a extension per Click, even defined if globally or per workspace, is there no such API an extension could call doing this? Its related to this comment: And we have currently 2 approaches (extensions) trying to improve the situation: Disable extensions Unwanted recommendations see for screenshots\nwould know\nThere is no API for this - May I know why an extension has to do this? Related, Have you tried the new profiles feature that is in preview? - This can be enabled using setting where you can create different profiles (settings, extensions, .... ) for different scenarios/use cases and switch seamlessly.\nHi, Yes, this is required in Vue, You can view the details here: I tried to use the profiles feature, but it seems that it cannot configure “Built-in Extensions”, and it is not very convenient to configure it once for each project, especially in teamwork.\nYou can configure enablement state of built in extensions in profiles. May I know if you tried this and if does not work and file an issue, because it is supposed to work. You do not need to configure for each project. You configure a profile once and open the workspace using this profile.\nI just tried it, the profiles feature can't meet this requirement, first of all, the exported configuration file is compressed, it's hard to read and modify., After formatting it, I found that it can't disable the extension. When I switch between different profiles, it's more like uninstalling those extensions that are not selected, but built in extensions cannot be uninstalled. I tried using a new profile, then disabled a built in extensions in VS Code and exported it, and when I imported it, the extension wasn't disabled.\nThis is not hand written and machine generated. Yes, but built in extensions can be disabled and profile carry this information. Can you please share the profile?\nOk, here is the exported config file after disabling built-in extensions. Extension id: vscode.typescript-language-features VS Code: 1.75.0-insider\nThanks for the profile. I could repro the bug. Filed a separate issue tracking the bug so that it can be verified too.\nThank you for your patience and reply, I wish you a happy new year."}
{"_id":"q-en-vscode-687d9ff7d600799d8f53ce7c7a7658fbeb47d964a80f7ddd083ff313bb1ab9eb","text":"These 2 methods are called when any setting changes: I would have expected an early return if no related setting changed."}
{"_id":"q-en-vscode-4ea5c34863901f3b94be7b83f8c032d8f2fae554ff66b17af3b20a2c4253bb43","text":"Found this in devtools, I played a bit with settings: <img width=\"1904\" alt=\"image\" src=\"https://user-\"\nLet me know if you can repro and find out the issue is deep in the list/tree.\nGot you repro steps! the workbench window rather small. into 2x2 editor layout, so you get even smaller editors <kbdCmd ,</kbdto open the Settings Editor in one of the editor groups for <kbdCmd W</kbdto close the Settings Editor That exception shows up!\nI just closed two other duplicates of this. There are repro steps since June. Let's see if we can drive this to a fix. Let me know if you believe it's deep down in the list/tree."}
{"_id":"q-en-vscode-698b51d79cbc44615a14d9c6fff6f5521bec62b8263a668671a0dd7fdb24e3d2","text":"(See attached recording) I have enclose it in new use cmd+left to go to start of line ghost suggestions get accepted https://user-"}
{"_id":"q-en-vscode-adef4b9d92f382cc9e8fa50efc3973be477a71a6cc990765ad631bc84854e108","text":"Testing We shouldn't have tooltips here: !\nwhat do you think? Currently, all toolbar actions have tooltips.\nYeah, unsure. All toolbars/actionbars have it but most have an icon and not the same text as the tooltip\nBoth options are fine for me. Whatever you prefer. I am slightly leaning towards no title. So we do not have hover in a hover (feels IDE-ish and heavy)."}
{"_id":"q-en-vscode-42a315430675af4ca2df6b788cc163ce4d7dc49f5acc09123ed7e41d6e444a05","text":"There's an unnecessary whitespace in \"i18n/vscode-language-pack-\". The whitespace after \"command:\" makes the link dead. !\nyou can add a comment to this string like so: to ensure that the command never has a problem like this. Notice that the example string has, in the Locked area, a at the start of it. This will make sure there is no whitespace between them."}
{"_id":"q-en-vscode-5bdea3efa4428851a3a44be59a9a499a15e6442c8444a64836d31811afcc9328","text":"a TS file Bug The repo index downloads quickly but TS initialization takes a long time From a profile, I noticed that some calls from the TS Server take between 200 and 400ms: ! I also noticed that we seem to be making network requests which invoke for a bunch of repos with names that look like file paths. ! Following this further, I noticed the request on JS/TS sends over two copies of every file: Note that these paths are transformed into URIs that look like: The first one is the correct path AFAIK? Not sure where the second one is coming from, but it seems to be the request that generates the network requests Does remote hub have two file system providers, one for and one for ? If it does, should JS/TS ignore the open documents?\nYes, there are two file system providers, and you can just deal with the URIs.\nI consistently see two requests for TS server, is that expected? <img width=\"966\" alt=\"image\" src=\"https://user-\"\nYes, we start two instance of the server: one fast one and one for project wide support. Hopefully with caching though this does not actually result in two network requests"}
{"_id":"q-en-vscode-7da4053d661503a8c50574bad8523997bb318d95727db15468c9df24d2a249c8","text":"Tracks picking up the Ts 4.9.5 recovery build Includes:\nTo verify, open a TS file and check the TS version in the language status item"}
{"_id":"q-en-vscode-bf5af087a3367c5b04adfec117c7fae9c78fc8eb9c381d1b01b8f2990cebf755","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --I am not happy with ctrl-shift-f. My steps: hit ctrl-shift-f enter the search term see several search results. want to go to the first match : There is no easy way to get to the first search result. After that I want to use up/down keys to go through the search results. The focus should not be in the source code. Screenshot of state of Step 5: ! Goal: Up to now I use the mouse and click on the first search result. If I use the mouse, I am at this state: ! After thinking about this some days, I think the best solution would be to map the ENTER to \"Go to first search result\". But not like F4, since F4 moves the cursor to the file, and not to the search result list. The ENTER should give the first item in the search result the focus. Then I can navigate to the matching result, and the next ENTER will bring me to the source code line. This would be great for all people who can't or don't want to use a mouse. Related:\nHi again! I'm just a bit confused on this part: For me, F4 moves to the first match, not the first file. Is this not what you're seeing? !\nI think right now this might be the best you can do: is a better command to consider than what you were using. The problem is that <kbdEnter</kbdis used to re-run the search if you make any changes to the inputs. But the above keybinding does focus the first result without moving the cursor, and then <kbdEnter</kbddoes move the cursor.\nthank you very much. If I add they keybinding via \"Open Keyboard Shortcuts (JSON)\", it works. But not always. After alt+enter I would like to use the up/down arrow keys and see the preview in the files. The strange thing: sometimes up/down shows me the places in the files (and the current entry in the result list is highlighted in blue color), and sometimes up/down does not show me the preview in the files. We are on the right track, but not there yet. Thank you very much for your kind support.\nFor me, F4 moves to the first match. If I press up/down keys, the cursor moves in the file. But I would like to move up/down in the search result list and see the preview on the right side (in the window of the source code editor).\nHmm, that's interesting, I'm also seeing that the files aren't previewing after using . will look into this more.\nI think I can get reproducible steps to the state where the file doesn't open when you have it highlighted in the tree view a long search it is searching, run using the up/down arrows to browse the search results, the file preview isn't triggering :bug: tree node is focused, but not selected :bug: This bug might be related to This is weird in that it only happens sometimes."}
{"_id":"q-en-vscode-51cdc6d9dc17fec6c0fb35b5ad2b990577888776a741af4d24282fbee18f23a8","text":"Not sure why this is happening, probably just need to restart, but I notice the notification is different depending on whether I press F5 to continue or press the play button: F5: <img width=\"1320\" alt=\"image\" src=\"https://user-\"Debug play button: <img width=\"359\" alt=\"image\" src=\"https://user-\"\nI think is from js-debug?\nLogs would be helpful, this indicates V8 responded with some kind of error (which I haven't seen before...) Alternatively, are there steps you take to reproduce this issue that you can share? /jsDebugLogs\nPlease collect trace logs using the following instructions:\nI don't care about the error, I think it was some process in a bad state. The issue is one notification is a warning and one is an error <img width=\"185\" alt=\"image\" src=\"https://user-\"!"}
{"_id":"q-en-vscode-5b9831e8a84097601b3bfee2ae52a88e9d73bf63d2b3327d34e625b8c50970f5","text":"Steps: from command line run it stalls We maybe missing a web worker related directive in our HTML files? //cc\nYeah, looks like needs to allow-list\nMay candidate? But I am unsure when/where/how the CSP are into effect. where do you see this? I have checked CSP allows blob as worker-src codespaces CSP allows blob as worker-src code-tunnels blob also allowed blob also allowed ????\nI saw it with our web bits from our server, so basically: download server run it access the web UI I am not sure this would be a candidate for today, but maybe recovery and only if we see an actual issue from it. So far this is only an issue for perf testing, but we do not perf test our server, only atm."}
{"_id":"q-en-vscode-27c400d41f1a040194a51836679beec5910f5c36829c7a7b81c30b03eb821a4e","text":"When developing extension the warning in : is wrong because it doesn't check if the extension is supported on older VSCode versions:"}
{"_id":"q-en-vscode-95b85d92e8ec3ad3bee76df8bd04a0a26936164d55ef3d788599c5fce8bb9573","text":"Install Java extension pack Open VS Code repo Cmd+Shift+B Java extension prompts error and the task detection never finishes\nWe have to wait for extensions to be registered before reading the tasks because of task contributions from extensions. I don't think we can change this. thoughts?\nWe could have a timeout per task provider. If a task provider doesn't provide tasks for 3 minutes after we ask for tasks then it's likely it will never provide tasks. That would let us show the tasks for all the other extensions except the slow one.\nI just hit this again , in a VS Code instance that had been running for quite some time. Are there logs that we can provide next time this happens? I enabled trace logging and looked in the Tasks output channel but did not see anything other than a warning about shell tasks being unavailable.\nI will try to get to this in March, no logs are needed as we understand how this could happen\nOriginal repro steps still work\nSteps still work sometimes - saw this when I reproduced !\nAlright, onto something - the consistent repro for me is when there's a multi-root workspace - with each repo contributing a default build task. the build task see quick pick with a bunch of tasks, not just the default build tasks When that's the case, yields none because none of the tasks have a property set yet and that is required for this predicate to be true So enters here and by that time, the group is defined ! and since there are multiple, this gets called\nThen it enters and concatenates non-defaults to the defaults. I don't think we want to do that. !\nthe fix is to only get the defaults and show those. we don't care about the others here. !\nRegression introduced by , but there's also the timing component as indicated above - the group isn't set when we expect it to be. Perhaps in the single repo case, the hasn't yet been set by the time , the second attempt, occurs\nI think we should be awaiting before running the build command\nApparently this still happens\nok in OSS where I'm reproducing, it goes in here like I thought ! then there are no workspace tasks found ! so it goes here !\nonly the are used here !\nso the issue is the aren't resolved in this case. it's not timing dependent - I can repro long after the window has opened. !\nI think here should have a size 0 !\ninteresting - I wonder if this is the issue !\nworked: ! failed: !\nso I think what's happening is if I trigger the build task before folder open, the promise is set but the tasks for that folder aren't parsed. then we return that promise indefinitely. I think checking if the is is in order here and in that case, we want to update the workspace tasks instead of returning the promise\nyep - I believe that works. ! So the repro is to run the build task before the folder opens.\nJust reproduced twice. When I did, both times it said this:\nstill happens in the build I created with the latest fix. note that i don't see extensions getting activated in the status bar here at all I also don't see the workspace folder undefined message anymore, but still see that the tasks are unavailable in the current environment.\nthe moment I change the file, it starts to work. IE when this is called\nWhile it's possible (extremely rare) that a user will still hit this if they run the build task before a particular execution context has been set, this fix will prevent that from reproducing. To verify, open VS code (typically a restart) and run the build task. If you do see the quickpick instead of the default build task getting, run, then the next time you run the build task, you should not see it. The above should be hard to reproduce though - and only possible if you time it just right. Before, this would happen more frequently and was reproducible until window reload."}
{"_id":"q-en-vscode-8566f9394d2178e179b7e6c05f1d924543c3ae5e66bea94066d6781baf845570","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes/No <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: NA OS Version: NA Steps to Reproduce: The github education team sees about calls from Visual-studio-code in an hour for api/user. This is quite unusual and since we are rate limiting anyway, it is not an immediate issue. From initial discussions with I understand that the call is made only for telemetry and has no impact on the user. So, raising this bug, if we can find a better way to do telemetry without having to rate limit. This here shows some more details.\n, can you please help me understand what is the likelihood of this work getting prioritised and by when can we expect to resolve this issue.\nI wasn't able to get to it this week during our allocated debt time so I will try again to make time early April. I'll let you know if anything changes."}
{"_id":"q-en-vscode-5ce533ff9752f9a00550f5f3d04c1013c97187cb585a751ee3bb3d806bd6a9ee","text":"I try to add a keybinding with ${selectedText}. but it will resloved as a normal string. Steps to Reproduce: ! Version: 1.75.0 (user setup) Commit: OS: Windows_NT x64 10.0. Sandboxed: No\nwould you happen to know why the variable isn't resolving to a value here? Is there something that the command must contain to resolve variables?\nThe decision for whether variables should be resolved in a command needs to be made on a per command bases. If you want, you can run your commands arguments through the variable resolve when your command is called. For example, the terminal uses the variable resolver here:\nFrom It would be great if more variable resolution was achieved in keybindings - I think it has been requested many times.\nverif steps: a keybinding for : that the keybinding correctly sets to the current path."}
{"_id":"q-en-vscode-b18d2d3240141ca9993fc2991eca615d1a96724570d1ae117be9defb3466f624","text":"This is a weird statement that seem to do nothing. I am creating this bug to accompany a PR to fix it. If there is an actual reason I am also happy to add a comment for it.\nThanks for the fix"}
{"_id":"q-en-vscode-b7a6f190723617cc52a776f4494c7adefe5086063a2a0a5bb5acb9eda23fc038","text":"an object that has many long property values There is no scrollbar the prototype There is a scrollbar !\nProbably only with the horizontalscrolling setting enabled? I know that's a bit tricky\nWhat's the setting for that? I don't have any scroll setting that appears debug related enabled. Just sticky scroll for the editor.\nI think you have set, right?\nNo I do not."}
{"_id":"q-en-vscode-d46f20ea20fc07fa480fa40763a9d623b03bcb46d13fc3937129e5d7da328844","text":"Testing When I was on the section of the walkthrough I clicked on \"open remote repository\" and it had the dropdown. I noticed that I couldn't just click outside the dropdown (back on the welcome page) to make the dropdown disappear and instead had to use the \"esc\" button. From my experience with other dropdowns if you click away from the dropdown it disappears and think that behavior would make sense here.\ndoes the quick pick come from ? If yes, who should I direct this issue to?\nYes that's where the quickpick comes from. owns the desktop git extension.\ncould you please help with this? Thanks.\nClicking outside of the dropdown still doesn't escape. !\nIt turns out that the \"Remote Repositories\" extension does not leverage the picker from the git-base extension so the fix that I made is not relevant. To verify this, please download the latest pre-release version of the \"Remote Repositories\" extension."}
{"_id":"q-en-vscode-4a92da6f6fbcc14a87a421fba052e367e296612a9400b1b320d2b4fdd0413734","text":"With &#x60;\"editor.wordWrap\": \"on\"&#x60;, the color hints (inline color decorators) may be double-displayed near the wrapped view edges. &#x2D;--- Version: 1.75.1 Commit: User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:109.0) Gecko/ Firefox/111.0 Embedder: <!-- generated by web issue reporter --\nCan you share a screenshot?\n<img src=\"https://user-\" alt=\"\" /\nHi could you post here the example from the image?\nUpdate\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!\n/verified <img src=\"\" alt=\"\" /Basically alright. But the early wrapping may be optimized. (see near line 6 \"#\", \"#\")"}
{"_id":"q-en-vscode-58446ca3c7da8f78eb068cb097cc01ca4464b81eac021940a2ab4f7a80229d87","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.76.0 OS Version: Windows 10 The most recent VS Code update broke bracket colorization for PHP files, when you close a PHP tag then open it again before using a closing bracket. Steps to Reproduce: a PHP file a PHP open tag , open a bracket with , use an end tag , step 2, but with a closed bracket show as red, and the coloured line that joins the brackets doesn't show. Note: The red brackets are normally only used for errors, so it makes it look like there's errors in the file Screenshot: !\nSame problem here on Unbutu! Could you fix it please?\nSame error on ubuntu Ubuntu 22.04.2 LTS\nAlso happens if you just have the opening tag but no closing tag at the end: !\nHello, thank you to open this topic. I have the same issue while writing PHP code on VSC. It would be great if could be fixed. Have a nice day !!\nSame issue here on Mac OS 13.2.1\nSame here\nFor anyone who needs this fix before the next update: Add this to your settings JSON (fix via PR )\nI have a similar issue using math equations in markdown, the image on the left is version 1.76 and the one on the right is 1.75.1, both of them with the same settings (by sharing them). The fix suggested by did not resolve my problem. Should I open a new issue since it is a different language and context? !\nPossibly related on Stack Overflow:\nRelated: :\nSame issue with me\nYou should scope it to language, otherwise the bracket setting applys to all languages\nThe update and modified settings as suggested by didn't solve the issue I mentioned in . Might be related to . Below is the today's insiders version w/o any extensions with the same issue despite even scoping the brackets as suggested by above. !\nomh same here lol.."}
{"_id":"q-en-vscode-56f0c166bab049bc4f9a72a7166073a99d6e5fd04fcddbef9664ba3ed3c92984","text":"On Windows: clone open workspace look at the workspace settings: all files are associated with schema . Note that this is a relative path to the workspace file open both files and check that code completion gives back property . You can also click on the JSON language selector ( next to JSON in the status bar) to see the schema that is associated :bug: No code code completion or schema association\n76 candidate PR:"}
{"_id":"q-en-vscode-01c587bca3095c23b81024db49d761632c43e7669b8a8c6fc92d9e6ce7a06889","text":"Type: <bBug</bContext changes causes images to flicker and, worse, quick picks to close. Run Open walkthrough Open step or Hit the button Quick pick closes right after clicking and image flashes. VS Code version: Code - Insiders 1.77.0-insider (, 2023-03-08T05:22:25.379Z) OS version: Darwin x64 22.3.0 Modes: Sandboxed: Yes <!-- generated by issue reporter --\nSeems as the context changes, which changes focus and closes the quick picks.\nI can't repo this on OSS. But used the developer tools on insiders to debug it. fires the context changed event too many times which causes us to reload the media I removed rebuilding step if the step is already open which should fix the issue for the walkthroughs. But there is likely a larger issue to be fixed with the being fired too many times. You can turn off settings sync and try the steps described above - no repro."}
{"_id":"q-en-vscode-b00414a562d82f438dc7cfa0a997678a914d8ebdda70351fd5f346dbca6230cb","text":"Version : 1.76.1 Commit : Date : 2023-03-08T16:32:15.546Z Navigateur : Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/102.0.4985.0 Safari/537.36 Edg/102.0.1235.1 with the extension installed (stable) switch the display language to French, reload VS Code switch the display language to English, reload VS Code open the remote menu, see that there's still a French menu entry in there ! Open the extension view ! I also see it the other way around: !\nAhhh so it looks like this is returning remote-repositories extension: with the old german manifest. What’s odd to me is that remote-repositories is being returned by and ... but even still, I don't know much about this code, but I would expect that the cache would cache the manifest in its raw form with the placeholders and then allow the runtime to translate the manifest (which could be optimized to cache both raw and translated form and only re-compute when language changes)\nI'm giving this one to you since it seems like a issue from the caching changes.\nIt seems we are translating the manifest always even if it is cached. But, I observed that some how (I do not know, may be in the old code), cache is having the already translated manifest. Therefore, while translating nothing gets translated. We do update cache every time but it seems if the version has not changed, we reuse the entry from the cache. There is no reason we have to reuse the entry from the cache. So fixed it by recomputing the cache entry always while updating the cache."}
{"_id":"q-en-vscode-971b94ebe46a657b3d52c92d6efede52ce5c675662c52a1016d6d952deba2ddc","text":"Type: <bBug</bWhen working in a CSS document, if I use the query, a pop-up occurs stating: I have not noticed any negative impact besides the pop-up obscuring terminal menu. When looking at the output, the log states: VS Code version: Code 1.76.1 (, 2023-03-08T16:32:00.131Z) OS version: WindowsNT x64 10.0. Modes: Sandboxed: No <details<summarySystem Info</summary </details<details<summaryExtensions (11)</summaryExtensionVersion ------ better-comments3.0.2 prettier-vscode9.10.4 gml-support1.7.4 vscode-language-babel0.0.37 git-graph1.30.0 powershell2023.2.1 vsliveshare1.0.5834 vscode-css-variables-autocomplete0.3.1 LiveServer5.7.9 code-spell-checker2.19.0 vscode-icons12.2.0 </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nType: Bug same probleme, working in a CSS document, using query, a pop-up occurs stating: VS Code Version: 1.76.1 (Universal) (, 2023-03-08T16:32:00.131Z) OS version: MacOs Monterey 12.6.3 (21G419) Electron: 19.1.11 Chromium: 102.0.5005.196 16.14.2 V8: 10.2.154.26-electron.0 OS: Darwin x64 21.6.0 Sandboxed: No\nA similar error seem to be happening when writing in a .scss file. Writing or pasting makes the error popup appear, but a does not. To be a little clearer, here's a list of patterns with that can be pasted in a .scss file to trigger the error : - ( there's a space here at the end) - - (note: does not trigger the error.)\nType: Bug same problem here, working in a CSS document, using query, a pop-up and related terminal occurs stating: Error: selectionRange must be contained in fullRange at Function.validate (/Users/james/Desktop/Visual Studio ) at new t (/Users/james/Desktop/Visual Studio ) at V (/Users/james/Desktop/Visual Studio ) at c (/Users/james/Desktop/Visual Studio ) at (/Users/james/Desktop/Visual Studio ) at Object.asDocumentSymbols (/Users/james/Desktop/Visual Studio ) at i (/Users/james/Desktop/Visual Studio ) at n.provideDocumentSymbols (/Users/james/Desktop/Visual Studio ) <img width=\"1231\" alt=\"스크린샷 2023-03-26 오전 10 32 42\" src=\"https://user-\"\nMismo problema, esperar alguna actualización para que lo corrijan? ! !\nI'm having the same issue.\nI'm confused. It says this was fixed but I'm still having the same issue. Nothing i've clicked on in this page is telling me how to fix it and use\nI will be in the next insider build\nSame issue\nSame issue\nAs mentioned, the fix is in 1.78. If you still see this, please add the build you are on and a small code sample to reproduce."}
{"_id":"q-en-vscode-3b4ad927a6827cd02fc0e42b94fd3d65f3da53920f29b0b1b316217a88dd3f2d","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.76.1 OS Version: macOS 13.2.1 (22D68) an that uses a schema other than or . For example, : text renders as a hyperlink (dotted underline), but isn't cmd/ctrl+clickable like other like OSC 8 hyperlinks with the / scheme. Other OSC 8-capable terminals (like ) will open the link correctly. Here's a video demonstrating the issue with several other URL schemes: https://user- As you can see, only and URL schemes result in clickable links. If I remember correctly, this regressed in the 1.74 -1.75 update, which leads me to believe it might be related the introduced in that version. This functionality can be super useful for tools like , which let you add hyperlinks to line numbers in diffs. When this functionality worked in VS Code 1.74, those links would be clickable: !\nSince everything supported this, I relied on it, damn. Would be amazing to have this back. Only non-heuristic way to get clickable file links in the terminal that handle all paths (e.g. paths with spaces)\nWe currently do not allow non-http(s) links due to allowNonHttpProtocols here: This may just be a case of adding that property to , but we need to make sure it won't introduce a security issue.\nHm, running code in a terminal generally means one trusts it. After all at this point it has access to a lot of things. Displaying data from untrusted sources without stripping ANSI escapes is more a problem of the tool displaying the data. I think when a terminal link is hovered, VS Code should just show a little link target floaty thing that shows where the link points to. Other terminal emulators do that. If you’re being paranoid, show a little when a link label looks like an URI but the corresponding link points to another URI. But that’s a situation people have in the web all the time, so they should know it can happen.\nsure, but you can some file and it will evaluate sequences which is a common attack vector. The user is typically unaware that could result in execution and/or malicious links being inserted. The hover is probably sufficient, I would want to double check with someone on the team who has done some general URI handling when looking at the PR.\nI understand the reasoning for disabling this, as VS Code focuses pretty strongly on security given its popularity (workspace trust, confirmations when opening links, well-defined extension API), but would it be possible to allow non-HTTP hyperlinks in the terminal via a setting? Something like ? It seems like implementing this would be a simple pass-through to 's property on (which you previously linked to). The setting could even have a security warning in its description. It'd be really nice to have this functionality back.\nWho needs to sign off on a decision here? This is affecting me pretty much daily (since my testing framework uses this and I simply can’t make VS Code behave)\nIf would be nice to support at least links, so that one could open files within VS Code, by clicking on a link in a terminal.\n+1, our framework uses OSC 8 links as one of the exclusive features vs competitors. Missing support in VS Code makes it irrelevant and degrades our productivity.\nEasy testing below, only the first 2 work: Note that we wouldn't be able to support any arbitrary link without additional work as there are security concerns.\nbash should work\nyeah the file needs to exist for it to work\nare you connected to the window via the Remote - WSL extension? If not this is as expected currently.\ncreated to track your 1. For 2 right now proper WSL link support only works within Remote - WSL windows, so making this work well in local windows would be a new feature request."}
{"_id":"q-en-vscode-60624f1531c94c0b5ae70f5c74e5c73b62cf8ea6fad5be24d4b7cb43ce71ed1a","text":"Using the test-provider-sample, adding newlines on a line marked as a test duplicates the test icon: ! It sorts itself out on the next update (at least in the sample, I'm still investigating to figure out if there's another issue, or that's in my extension). But it still looks bad - especially if a language is debouncing edits and/or discovery may be slow.. You can make the issue more pronounced (as I did for the GIF) by updating the test-provider-sample to have a delay in its updates:\nI was able to repro the full issue I'm seeing in Dart (where the icons never disappear) using the sample too: ! My changes to the sample are here: I changed it so that tests can span multiple lines (since the original code split on lines then searched each line) and use the full range as the for the test. Now when pressing in the test, the original issue (VS Code duplicate the icons) still occurs, but also after the updates, since the range computed by the provider doesn't change from the range VS Code currently has (at least that's my guess), there's no update to the gutter icons so they no longer get cleaned up.\nThat's correct, we let the editor do its default decoration tracking until an update is received from the extension. for testing decorations, I already set , but is there a similar option to avoid extending the range over lines when edited? Or, alternatively, a way to only apply the glyphMarginClassName to the first line of the range. Reference:\nYou can create two decorations per test. One which has a range, a hover message, etc. And another one which renders the glyph margin with a collapsed range (perhaps the start position of the range). If you use for the second one, given it begins as a collapsed range, it will never grow, it will always remain a collapsed range."}
{"_id":"q-en-vscode-5e54b9d551bb259cad31b9c475d6381b67c37299c3b5acd08ccf05910a673a6d","text":"Ubuntu 22.10, Gnome 43.1, fresh VM - The errors are not shown the second time I call and also VS Code seems to start and work I saw , but that should be in by now."}
{"_id":"q-en-vscode-3b87a4f2d976654cb06aa0292a97c9c1d805225c305f036e660348ce1acc64e6","text":"a notebook and run a cell that will continuously update over time to drag the scrollbar while output is still being written :bug: scrollbar is not held and returns to the bottom\nStreaming output items are all compressed into a single item in (?), so we would need to stop doing that unless needed (for editing previous lines) Also, the current method only allows passing a single output item, but the output renderer will need to be able to render the entire output as it currently does, or just append the new data. Current: Proposed method that would only be used for streaming outputs and if the chosen renderer contains the method: Another alternative would be adding info to the OutputItem itself, containing only the appended string if appending should happen."}
{"_id":"q-en-vscode-3ca9a79cf5fc6420ff7493d42ab8678e3676bcddac1b2a3e11db8c25600fa4ba","text":"Testing When there's too much output, this link uses to explain what it does: <img width=\"76\" alt=\"image\" src=\"https://user-\"<img width=\"205\" alt=\"image\" src=\"https://user-\"This isn't accessible, you should set such that it's read by a screen reader."}
{"_id":"q-en-vscode-e4572014a0855abb10d0906466cbf39d298bc7966e83dc59b55e706bf27fdb56","text":"Testing a Markdown file with headers up to H3, enable Sticky Scroll and scroll up and down in the editor. Stick Scroll a header leve and press . Try to scroll up and down with the arrow keys. Unclear where the cursor focus is. Try typing, the editor doesn't have focus Focus should should probably go to the editor on the selected header. That is the behavior with breadcrumb navigation\nThis is the general behavior of Sticky Scroll and also reproduces if you select a level via the mouse\nVerified on OSS"}
{"_id":"q-en-vscode-3b3d4d0582ee1cf59b12eb9ce83eda35d0e4e407d0693f2ad7c3441ff2456606","text":"Currently when you use navigate accessible buffer you get a quick pick with the options, I expected changing these to scroll to the position without accepting it, and escape to revert to the original scroll position.\nAs part of this, providing ~1/3 of the viewport of context above would also be good\nWe'd need to think about how we to communicate the preview information to a screen reader user, as a pre\"view\" is usually visual\nI don't think making this available to the screen reader makes sense, it would just complicate navigating the list. It is useful for sighted users though.\nOh I understand now - you mean on change of the active quick pick item, reveal the line and on enter, focus that line (as it already does)\nYep\nTo verify, in a terminal with shell integration, run some commands. to enter the accessible buffer. to . Without pressing , use the arrow keys to go between the options. It should scroll to the command if it is outside of the viewport.\nWorks but I opened"}
{"_id":"q-en-vscode-eac9af73b7a05327a0c2052daf6958a64e8ce2947557e5c622139f2576021c59","text":"Using here, I cannot enter the accessible buffer https://user-\nThe issue is that here, the default tab behavior is to focus the terminal tabs element edit: hard to repro, that's not the cause\nok this happens when is not applied to . See here that the buffer editor does get focused here, but you cannot see it https://user-\nRepro: run some commands - - - Try to wasn't applied to so the accessible buffer isn't visible\nhttps://user-"}
{"_id":"q-en-vscode-13297bdaf2806fc5bb27ce5bf0a0d1721bb2b34f44248de4ea87647ca7a8f6b0","text":"<img width=\"980\" alt=\"image\" src=\"https://user-\"\nI suggest to move the docs into our wiki. PR to remove these files:"}
{"_id":"q-en-vscode-4d171807d049adad47c7c3098c979c5853a85f2c581164f16b802d95b615517c","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.78.0 OS Version: Windows 10 (x64) Steps to Reproduce: \"Centered Layout\". the editor by independently moving its sashes (holding down Alt). and restart VS Code. The position used to be remembered in previous versions, but starting with version 1.76, the position gets reset on every start. Only the width is being remembered. It's still broken in the current insider build of VS Code 1.78.0.\nDo we already know, which commit causes this regression?\nI looked through the commits for 1.76.0 and I suspect it was .\nYeah, it is solved for me."}
{"_id":"q-en-vscode-75fdbf3f00d275265244ae88f0460080e6fd45e9da5d60ba7448d7f29bbc5349","text":"The Angular language service extension spawns a child process using and listens for the and events. The current version works as expected. Without any changes to the project, simply updating to results in the forked child process acting differently (never erroring and never closing). I'm using Mac 13.3.1 but we also had a report that the forked process isn't working in WindowsNT x64 10.0. either. There was a similar issue (though not entirely the same) a while back: Apologies for the rather involved reproduction... Here's how I'm able to reproduce and compare the behavior between the versions: our the repository with vscode insider version 1.78.0-insiders out the branch (we're actually removing this in the next version so it doesn't exist on ) dependencies with debugging, put a debugger statement before and one in the . The project is built with bazel so breakpoints would have to be put in the compiled JS files. the build script with the extension development host (Dev Client + Server run configuration) In the extension host, open the and then open the template file -- debugger should stop on the fork and if you continue, you will see that the debugger stops in the stderr data listener but never in the or . The output of the stderr also indicates that it is listening for a debugger and that it gets attached. The fork used which should have prevented this. The process also never produces any results or completes. You can have the fork retry by running the Angular: Restart Angular Language server command in the extension host. I've also verified that the spawn args are the same between versions and both include \"--ms-enable-electron-run-as-node\".\nIssue was caused by the allocator changes to processes which now schedule delayed tasks to purge memory cache ;l=1183 and that revealed a bug in the causing the process to hang.\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!"}
{"_id":"q-en-vscode-3f02a8d4d644cb94b30b1b4d3fad6018c279d21152e104320c2aa1b8ca30442f","text":"<img width=\"1440\" alt=\"image\" src=\"https://user-\"path: Versions:\nI have one question wish can get help: how vscode terminal detect the path cwd? why can the path correctlly go to the file with a releative path? <img width=\"1440\" alt=\"image\" src=\"https://user-\"I guess it may use the But seems I can get the cwd in TerminalLinkProvider now: <img width=\"1440\" alt=\"image\" src=\"https://user-\"\nWe check a few things to determine which to use\nvite become more and more popular recent days, I think it would be convinient to support the path with query string: !\nWhen I click, it opens the right file, but it underlines the first query argument when I hover, is that an issue? <img width=\"539\" alt=\"image\" src=\"\"\nnot much of an issue but it would look more polished by just underlining the whole thing"}
{"_id":"q-en-vscode-3dbabea543e8651e338d0486866ed15069d8db40ef3d9b6b4cd689ffedd716b5","text":"Does this issue occur when all extensions are disabled?: Yes VS Code Version: 1.77.3 OS Version: Ubuntu 20.04.5 Steps to Reproduce: (incorrect category, should be themes) outputs: suggestion: Get warning:\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!"}
{"_id":"q-en-vscode-eda9d7ff9d923b4b4da2286edfa193f00b53a40fc9d5236fbb55dbd5c8b897a9","text":"Type: <bBug</bHow to reproduce the issue: and Run the Extension the , and then the pane. Use the dropdown and select the Dynamic Configuration Provider . With the checked out branch, it will show the quick pick options for Option 1 - 3, but will not show a quick pick for the resolved configurations. This seems to work for , but not Expected: ! Actual: ! VS Code version: Code 1.77.3 (, 2023-04-12T09:16:02.548Z) OS version: WindowsNT x64 10.0. Modes: Sandboxed: No <details<summarySystem Info</summary </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nThanks for creating this issue! It looks like you may be using an old version of VS Code, the latest stable release is 1.78.0. Please try upgrading to the latest version and checking whether this issue remains. Happy Coding!\nI've updated to: and it still reproduces.\nfor prioritization help as per email thread\nSent a PR for this. But what is the inner quickpick? I'm a little unsure about the approach, I don't think we expect a provider to include some UI. Your extension doesn't own the UI in this flow- you are just implementing a provider method which will be invoked at different times, expected to return a list of configs. It might make sense in one situation to show the quick pick, but we could make changes in vscode to call your provider at a different time where it would be unexpected to have a quick pick pop up. And, any extension can implement a config provider for your debug type. So for example, multiple extensions could all be showing their own UI in this scenario. However, from memory, I think the Python debug extension might already have a similar multi-step picker UI for a scenario like this."}
{"_id":"q-en-vscode-da01fd14c40c7438de703b059e04749bdf4768abbd9588e60f3fccc9441357b7","text":"https://user- problem extension:\nI have the same problem in tooltips of TreeItem. I thought it was associated with my extension, but apparently not. This behavior does not happen with the GitLens extension.\nThis issue only happen to stausBarItem at left side\n!\nthis issue should be mark new release label\ncan you help look into this issue?\nWasn't this always like this? I just tested and it's not hoverable in 1.78.0 either. You can hold alt to hover it though which is a common feature of hovers (outside the editor)\n1.78 is the current VSCode version - the one where it broke. 1.77 and below works properly.\nDoh, checking an older one\nLooks like the hover only sticks if there are textual links in it. Example is the hover on the next to the language selector. I didn't change anything in the hover area recently.\nthat's how it used to work, it doesn't anymore. and I made some changes around this recently, it was probably us.\nthis is by design if the content does not contain any links. Hold alt to \"lock\" any hover\nThis is happening because you're using an anchor tag, a change was made recently to only force hovering when the hover contained \"](\". I'll add an anchor tag check.\nCurious, because in these tooltips I never used links (html or md) and it worked. Recently this behavior started This tooltip has always been used by my users to copy information, by simply selecting and CTRL + C\npreviously the logic would disable hide on hover if it was any markdown string which is why it was like that. If the user wants to copy information they can do so by . I know it's not the most discoverable but it's intuitive once you know about it. It's a harder problem than you might expect because if it didn't behave like this we would get the opposite type of issue where hovers are really annoying because they don't dismiss. So we don't hide hovers on mouse over only when there's a high chance of interaction."}
{"_id":"q-en-vscode-8484309218af39b1694908dcc44550ac1a5a07289281afd7d93512385d817144","text":"Type: <bBug</bCC: I noticed that I can't navigate the accessible buffer using ctrl left arrow or ctrl+right arrow. When using this combination orca treats the whole line as if it were a single word. Except for misunderstandings, this happens because the blank spaces present in the accessible buffer are being replaced by the character 0xa0. I ran the command echo \"this is test\" in the VSCode terminal and ran the same command in the gnome terminal. Browsing character by character with orca I noticed that in the gnome terminal space appears as 0x20 and in the accessible buffer it appears as 0xa0. VS Code version: Code - Insiders 1.79.0-insider (, 2023-05-04T22:12:16.054Z) OS version: Linux x64 6.2.14-300.fc38.x8664 Modes: Sandboxed: Yes Remote OS version: Linux x64 4.14.146-93.123.amzn1.x8664 <details<summarySystem Info</summary </details<details<summaryExtensions (44)</summaryExtensionVersion ------ doxdocgen1.4.0 vscode-markdownlint0.50.0 vscode-eslint2.4.0 EditorConfig0.16.4 copilot1.86.82 rest-client0.25.1 better-cpp-syntax1.17.2 docomment1.0.0 azure-pipelines1.208.0 vscode-docker1.25.0 csharp1.25.7 vscode-dotnet-runtime1.6.0 vscode-kubernetes-tools1.3.11 remote-containers0.293.0 remote-ssh0.101. remote-ssh-edit0.84.0 azure-account0.11.4 azurecli0.5.0 cmake-tools1.14.31 cpptools1.15.4 cpptools-extension-pack1.3.0 makefile-tools0.7.0 powershell2023.5.0 remote-explorer0.5. vsliveshare1.0.5864 oracledevtools21.5.0 fabric8-analytics0.3.6 java1.19. vscode-xml0.25. vscode-yaml1.12.2 sonarlint-vscode3.17.0 cmake0.0.17 intellicode-api-usage-examples0.2.7 vscodeintellicode1.2.30 vscode-boot-dev-pack0.2.1 vscode-spring-boot1.46.0 vscode-java-debug0.50. vscode-java-dependency0.22. vscode-java-pack0.25. vscode-java-test0.38. vscode-maven0.41. vscode-spring-boot-dashboard0.13.0 vscode-spring-initializr0.11.2 gitblame10.2.0 (1 theme extensions excluded) </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nWe replace these spaces intentionally, but I'm not sure why . thoughts?\nAs it is, I can't use orca to select words and I can't navigate by words either.\nWe replace it with a non-breaking space such that white space appears the same as it would in the terminal. I wasn't aware it has this impact on screen readers (maybe it's just orca?) but we could workaround it by replacing (2 spaces) with . That should keep tables, ls, etc. aligned while fixing the problem at hand. It seems like an Orca issue though as I'm pretty sure a non-breaking space should be getting treated as a space.\nI also think this should be corrected on the Orca side because the same problem also exists when I'm editing a file. I can't navigate through words in a text when they are separated by the 0x0 character.\" Reported this issue on the Orca mailing list.\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!\nThis doesn't appear to be fixed: ! ! It is changing multiple lines to include a single space character as expected, but see the first line where single spaces are changed. I'm questioning the fix we came up with though, monaco replaces spaces anyway in the editor: ! I think we should revert this and have orca fix it, instead of trying to workaround their problem which could end up causing other issues for us?\nI ended up reporting this problem to the orca developers because the problem also exists when I'm editing a file that has words separated by the character 0xa0. I believe the problem will be dealt with on the orca side.\nIf I understand what reported correctly, I don't think this is an Orca issue. When Orca is in focus mode, it does not control or consume input events. In other words, Ctrl+Right/Left arrow will do whatever they would do without Orca running. And from some testing, when the words are separated by 0xa0, VS Code (or Chromium) is treating the string as a single word. If, in those same places, I switch Orca to browse mode, Orca IS controlling/consuming the input events, and it already does treat the string as individual words. But if I'm missing something, please let me know what I should be fixing in Orca. Thanks!\nHere are the steps I am using to reproduce the problem which I believe is reporting: sure Orca is NOT running. a new document. \"This\" Ctrl+Shift+U \"00a0\" followed by return to insert a non-breaking space character \"is\" steps 4 and 5. \"a\" steps 4 and 5. \"test.\" Home to move to the beginning of the line. Ctrl+Right to move to the next word. Expected results: The caret would move just past the \"s\" of \"This\". Actual results: The caret immediately jumps to the \".\" at the end of the sentence.\nthanks for the follow up, you're right. we should not be replacing it with non-breaking spaces as monaco handles their presentation. When you paste in the characters like describes (I did it by running 'u00a0' in console and copying that) you can see they're treated specially, we also have a lint rule to prevent this in vscode: !"}
{"_id":"q-en-vscode-f919b29fb2775ba7d15adc41d09661d67f54f9fdf2fcc26c01fbff8c05254c8d","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --<!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.79.0-insider OS Version: macOS 12.6.3 I am working on a file explorer tree view extension and doesn't work properly if is 2 or 3. When expand is 3 only the first level 3 folder expands. I have checked and doesn't get called for the rest of the level 3 folders. If I call again with the rest of the level 3 folders expand. When I use with 2 or 3 on a large folder (hundreds of items) the promise never fulfills even though the folder expands (at least up to level 2). I don't get any error in any of the two cases.\ncan you share a minimal example that reproduces this issue?\nSure thing, here you have it:\nThank you for the repro repo!\nNo problem, thanks for the fix!"}
{"_id":"q-en-vscode-6777bd56018c6b1c704b818c0e0cc72d6a0c94fe6e2b41d3fe80d26c6e4f5258","text":"Callstack is similar to the one mentioned in but the issue remains First failure Second failure\nThe logs read very normal, we literally hit the 2min timeout. Since these tests do double the work (start VS Code 2 times), I will see if increasing the timeout helps."}
{"_id":"q-en-vscode-d38e364bad78804158c3b64cc601b3ca727f4a44161815db0053037c1c8cecda","text":"I'm seeing some flickering here, just Write a multi-line commit message cmd+enter to commit !"}
{"_id":"q-en-vscode-80631e566796868ec5ed3326102358e191527a40bffb38a71b191ecdb36b2600","text":"I see this in the console log when I reload vscode insiders window: <img width=\"1463\" alt=\"image\" src=\"\"vscode insiders with the vscode welcome page open (I can't repro without the welcome page) window and you'll see the errors in the console logs"}
{"_id":"q-en-vscode-d88bf4fb0d44d4448c608e8be2f7fdace98f8975a38bfa28f62430d0ffb76e56","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: Steps to Reproduce: an invalid decoration in warning in ext host logs that says instead of the extension name:\nThanks for creating this issue! It looks like you may be using an old version of VS Code, the latest stable release is 1.78.2. Please try upgrading to the latest version and checking whether this issue remains. Happy Coding!"}
{"_id":"q-en-vscode-a38b19deda3e45bdc3e93aaacbe001c14ba1a32ef610ea266223f0f4ef1db43e","text":"Testing ! !\nThis is because of the same problem that caused\nappears to be closed now, does that mean this has been fixed?\nNo, this issue is about fish acting atypical in that it applies the environment before is applied. This is because of how the automatic injection works in fish via an environment variable. Still not sure how we can handle this.\nDiscussed offline, maybe we could use the first pre exec hook in fish to run shell integration, instead of .\nNice !"}
{"_id":"q-en-vscode-631dc7746d8d5439ebefdb2af7c80ec591b8dd5bb7c22a212c22ed2a3a91ddcd","text":"On the notebook toolbar we have multiple sections of actions, and the first two are core sections: insert and execution. The Insert section is very stable and it always contains two actions: insert code and insert markdown. However the execution section has actions from both VS Code core and Jupyter extension, and with a mix use of visibility and enablement, it is not user friendly and often leads to mis-clicks. Initial state <img width=\"811\" alt=\"image\" src=\"\"Start execution <img width=\"884\" alt=\"image\" src=\"\"Execution fails <img width=\"903\" alt=\"image\" src=\"\"Execution succeeds <img width=\"810\" alt=\"image\" src=\"\"Initial state <img width=\"657\" alt=\"image\" src=\"\"Start execution <img width=\"867\" alt=\"image\" src=\"\"Execution fails <img width=\"728\" alt=\"image\" src=\"\"Execution succeeds <img width=\"625\" alt=\"image\" src=\"\"As shown above, currently only is using enablement while all others are using visibility, which can lead to change of positions for the actions and often when users want to interrupt or restart the kernel, their position changes and users click on wrong actions. cc\nWhat are users misclicking on? From my memory, I remember Go To vs Restart. I think that \"Restart\" has a confirm dialog, are there other destructive actions that should also have a confirm dialog? Or, should we use enablement and have all actions always visible?\nFor Jupyter kernel, I ran into Go To vs Restart and sometimes Restart vs next action on the toolbar (since Go To disappears), and for non Jupyter, it's Interrupt vs Clear all outputs. It might be hard to enforce that but we could probably at least ensure the ones that are always visible won't change their position when the execution state changes. For example, if Go To only shows at the end in this group, and Run All is always replaced by \"Interrupt\" or \"Stop\", then it's less likely to mis-click."}
{"_id":"q-en-vscode-757c9b3fab198426c67152ee7859d130409477b17ea537438071bbabd195c86e","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.77.3 OS Version: Windows 11 Steps to Reproduce: a git repository a file with brackets in the name, e.g. the file in the git tab and add it to the file Desired behavior: The filename is to the such that the file is ignored Observed behavior: The file name is to but it is not ignored. This is because the syntax of does not interpret brackets literally. To ignore a filename that includes brackets requires escaping brackets with backslashes. gitignore after step 3: for desired effect should be: Proposed fix: Pre-process file names to to escape special characters.\nThanks for creating this issue! It looks like you may be using an old version of VS Code, the latest stable release is 1.79.0. Please try upgrading to the latest version and checking whether this issue remains. Happy Coding!"}
{"_id":"q-en-vscode-3b66a6d4b7a050b6d270f2301fccca7308d673448ff087b21cfd21bc6e8b44ba","text":"Looks like we have broad support for . Static initialisers are the slight exception as they are only in Safari since ~2months. We might want to special case them, but otherwise we should be good to go:\nOn native privates: using them will be faster than weak-map-privates but slower than normal properties (much, much slower for Safari):\nSome, perf number for some new API native privates - , faster than WeakMap privates but slower than conventional privates slower than []-access needs which need broader adoption because of semantic change.\nUpstream issues for private property performance v8 - Webkit -"}
{"_id":"q-en-vscode-ac3a183dc38a385a0a214972bddf66df1d865c96fd65cad1edc6fa728eb5e3af","text":"I had a \"Testing\" profile with various remote extensions installed. I deleted it (while being in it) and then recreated it empty. I still found all remote menu contributions even though the profile should have been empty. Maybe switching profiles should ask to reload the window when extensions require so? Steps to Reproduce: fresh user data dir, fresh extensions dir create an empty profile install SSH extension notice how the remote explorer view appears delete the profile notice how the remote explorer view is still there, same as the status bar entry for remote\nAnother case from is interesting: I do not think using a profile is a good idea for when extensions require a window reload to be disabled. So you need to then probably ask the user to reload the window or do it automatically.\nAFAIK every extension requires a window reload after being disabled (except for package-json only extensions). We cannot unload code from extensions and cannot enforce them not doing anything anymore. There we always reload the window or at least the extension host\nWhile switching profiles we restart extension host. You say this is not enough to unload extensions? Also while trusting/untrusting a workspace.\nNo, only those that are activate with code loaded. Here are steps: fresh user data dir, fresh extensions dir create an empty profile install SSH extension notice how the remote explorer view appears delete the profile notice how the remote explorer view is still there, same as the status bar entry for remote\nI see this can be reproduced with above steps. But Remote Explorer view disappears with switching profile. So I think this is a bug (either in views or profiles) when profile is deleted. But status bar entries always appear."}
{"_id":"q-en-vscode-7aaf5acbdbfb06ce0ed63b49d59cbb04e6e042684c26bfae5cd37af225a7d04c","text":"Testing I am currently one commit behind the remote branch and I am able to successfully create a link without seeing any dialog. The way the test plan item is written it implies that in this scenario, I should see a dialog. Is that the case or did I misunderstood the test plan item?\nTesting with while being a few commits behind, I get the following dialog: <img width=\"264\" alt=\"image\" src=\"\"When I Cancel, I don't get a link. When I Pull & Copy I get a link that is seemingly unaffected by the pull. Why then require the pull? (I tested with incoming commits affecting the file I copy a link from.)\nThe link that is copied might point to a range that is invalidated by incoming commits, so we notify in order to avoid you copying a link that points to a different area in the file. But perhaps we should also offer a Copy Anyway option here."}
{"_id":"q-en-vscode-9d8ec5c39f74de49f0873135e1cb13aa008aa4890f4fee7bf2ec874744fe2115","text":"Repro: OSS trace logs window ! I don't have copilot installs in OSS so I wouldn't expect anything to happen here. Not sure what the 2 chars are.\nProbably"}
{"_id":"q-en-vscode-920dc197476b50d7a4820e424e045783ddaca18fc16a361d877bbcb5cd1519f9","text":"Issue filed by Hi, I would like to ask about a problem related to the usage of WorkspaceConfiguration.Update() We received an issue This happens when the C# extension starts, we are migrating some old configurations to its new name when using a non-default profile. (This doesn't repro when use default profile) See the code here: I tested locally and verified the issue in feedback happens when executing during the extension activation. What I did is created a non-default profile, and put one of our deprecated options in the global profile. <img width=\"985\" alt=\"image\" src=\"\"Then the error occurs when removing the old configuration see this ! The old configuration is removed in this case. Is this our misuse of the API or is this a bug in vscode?\nProfile / config - assigning first to Sandeep this is from the C# extension, so if you have additional questions feel free to ask or let me know if I can help in any way.\nIt would be great if we can have a summary of steps to reproduce\ncan you provide steps please? Thank you\nMy repro step: Install the C# extension in the preview channel. <img width=\"665\" alt=\"image\" src=\"\"- Use this sample solution, open it. Create a non-default profile. <img width=\"515\" alt=\"image\" src=\"\"- Open the profile in json editor. Add these to the JSON. Close VSCode and restart it. You can see one configuration is changed. <img width=\"602\" alt=\"image\" src=\"\"C# extension is not activated. And the error log is shown in the extension host output window.\ncan you verify that this fix indeed fixes the problem you reported. Thank you!\nVerified it is solved in 1.81.0-insider"}
{"_id":"q-en-vscode-e7c3488db33cd4f2559a0ff57586139cb16a2d8b9fa72e27dd4073a09b4619b3","text":"cmd+F to open find widget Close it with escape cmd+up to navigate to previous command Nothing happens, I have to click or run a command to reset focus in the terminal\nI can't reproduce this, if you still can are there any more details on the repro case? The keyboard troubleshooting command output would be helpful if only you can repro.\nSorry it's working now, agreed that it was probably a context key thing so I'll check that next time\nI can repro this when the serch input have has text selected, PR"}
{"_id":"q-en-vscode-ca19f33561c00a2e1b68385f6cf37b044b66404d5a68481fd4fec1efc590d873","text":"\"Please do not close this bug. This bug should only be closed by TCS, C+AI Accessibility or the Divisional Driver\". #A11ySev2; #A11yTCS; #A11yMAS; #DesktopApp; #Visual Studio Code Client; #WCAG1.3.1; #DesktopApp; #BM-VisualStudioCodeClient-Win32-Aug2022; #NVDA; #Win32; #Win11; #A11yeDAD; #ScreenReader; #FTP; #NVDA; #AINotInScope; #Closed; Upon using the left or right arrow key to navigate the characters in the Terminal command, NVDA announces a blank instead of correctly providing the character feedback. As the arrow keys are pressed, NVDA should announce the characters in the command being navigated. Application Name: Visual Studio Code Version: 1.80.0-insider (user setup) V8: 10.8.168.25-electron.0 Windows 11 22H2 OS build .1000 Screen Reader: NVDA (2023.1) Terminal in Visual studio Code. on NVDA. terminal, type any command. left/ right arrow key to move NVDA focus within the written command. NVDA behavior The incorrect feedback provided by NVDA/JAWS affects the accessibility and usability of the Terminal application for users who rely on screen readers. This behavior hinders their ability to navigate and interact with command-line interfaces effectively. \" - Identify accessibility bugs before check-in and make bug fixing faster and easier.\"\nnot sure if we're tracking this but we're talked about the fix I had in mind for this which was to synchronize ' textarea content with the help of shell integration.\nThis also happened to me when I was trying to edit a recent command in the integrated terminal (weirdly, arrowing on the newly typed command doesn't seem to be an issue). This makes integrated terminal nearly unuseable with the long command (in which typo is very much possible). I have also created the issue here:\nVerified the issue in insiders 1.82.0 The issue is still repro. Please tell me in which version the issue is fixed. ! #Regressed:08-21-23; #PendingInput;\nThis works as shown here ! That is, until the terminal loses and then regains focus. That's a separate issue IMO and thus can be handled as a different bug tracked here\nI verified the issue in below version of Visual Code Insiders. Version: 1.82.0-insider (user setup) Issue is still reproing. Please tell me in which version of Visual code Insiders the issue is fixed.\nwhen reporting insiders versions you need to go to Help About and copy out all the version details. In particular the commit and date are needed to identify the insiders version.\nThis was verified by on both windows and mac. is having issues with it so we're going to meet about that.\nI can reproduce the issue using insider version 1.82.0-insider (system setup), commit . I'm not sure why, but it appears that the issue has been resolved when using VS code inside WSL2 with the bash shell. However, when using VS code on the Windows host with PowerShell as the shell, I still encounter problems when navigating and editing both newly inputted commands and the previous commands recalled from the recent command history.\nand I met and he showed it not working on windows. I'm guessing this is a conpty issue\nwhich version of windows are you on ?\nI use windows 11 Pro Version 22H2 (OS Build .2215)\nI'm also able to repro on the latest Insiders on Windows using NVDA.\nThis is working on Windows for me. (10.0. Build ) which version do you have?\nIn System About I see version .2134 Using NVDA on Windows, and with screenreader mode enabled: a pwsh integrated terminal type left and right arrow keys announced text is \"empty\" whenever those arrow keys are pressed.\nBelow are the Environment Details of the latest Visual Studio Insider installed in my system: Version: 1.82.0-insider (user setup) Commit: Date: 2023-08-30T05:33:47.972Z Electron: 25.7.0 ElectronBuildId: Chromium: 114.0.5735.289 18.15.0 V8: 11.4.183.29-electron.0 OS: Windows_NT x64 10.0.\nYeah I can reproduce now. Seems to have regressed for Windows\nis undefined for Windows\nit may also in the wrong position on Windows, if I remember it's typically at 0. This will get fixed eventually in conpty.\nwhat's odd to me is that it was working for me before\nI'm not sure when it would be undefined, I guess it may be inconsistent? Command executed does not ensure command started had occurred:\nWe need info on the command line before command execution and for windows, we don't have that currently.\nAs with a lot of things on Windows ptys you'll need to add some heuristics here. It might work fine to just use . You could also try add some basic prompt detection for standard\nTo verify, enable . Then follow the steps here.\nI did as you said. But still the issue is still reproducing. Please help ! Environment details: Version: 1.82.0-insider (user setup) Commit: Date: 2023-08-31T05:33:59.698Z Electron: 25.7.0 ElectronBuildId: Chromium: 114.0.5735.289 18.15.0 V8: 11.4.183.29-electron.0 OS: Windows_NT x64 10.0. NVDA version: 2023.1\nyou need to update you're using an old version\nVerified the issue in below environment. Version: 1.82.0-insider (user setup) Commit: Date: 2023-09-01T05:33:39.919Z Electron: 25.7.0 ElectronBuildId: Chromium: 114.0.5735.289 18.15.0 V8: 11.4.183.29-electron.0 OS: Windows_NT x64 10.0. Issue is fixed. #Closed; #Regressed:09-04-23;"}
{"_id":"q-en-vscode-1f4db0c297eb6f730f1442074e961a16255f670055eaa9b320b6b54d5b8c090d","text":"some terminals and type in them the window or restart VS Code they don't persist Started happening in OSS on Wednesday July 5th\nis undefined !\nit gets set to !\nthat's the return value (see )\nyes, true. but the call returns so it's getting reset somewhere. not via !\nI'm on to the cause... The process is getting killed after detaching on reload, so reconnect breaks: Revive after restart is working for me.\nAh found it: Investigating...\nParsed state is null so reconnect stops here:"}
{"_id":"q-en-vscode-2d9cedb09ce5c3fe70aff15cc84e55f684ed982e453ecdfad41ffc7fadf59120","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --For example when I open a project in v, I expand the folder on the file tree. When I refresh the whole page, the default behavior now is that the folders are collapsed, which means that I need to expand the folders one by one again the last time I opened them. !\nIf the expansion state of the file tree folder can remember the last expansion like tabs, similar to the gif below, I think this is a better effect. !\nI have fixed this problem and will mention a pr later.\nThe reason for this problem is that there are multiple calls to the function during the rendering of the file tree. The state will be read from the cache for the first time, but before the execution is completed for the first time, the will be executed again. At this time, the tree already has a value. However, because the file tree data is not ready (because the data on the Web side is asynchronous), the state returned by the function is empty, resulting in the final file tree returning to the default collapsed state."}
{"_id":"q-en-vscode-94f5c23a56e08fa6bcf46ebf09d84a23b3854a3b4e6e55b2a615055440588789","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes/No <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: : 1.81.0-insider : : 2023-07-12T11:13:11.418Z : Mozilla/5.0 (Macintosh; Intel Mac OS X ) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/114.0.0.0 Safari/537.36 OS Version: Mac os 12.2.1 (21D62) Steps to Reproduce: Configuration Display Language to . devtools after page auto refresh !\neg. does not exist, but exists.\nthanks for working on this. The true fix for this issue actually needs to be in 's source code. I've made that fix and this should work for you now. Let me know if you still see any trouble."}
{"_id":"q-en-vscode-b440d3021dc29fe87a529d9284aab009aea16d735a447645f4a5aa00cda0a6c7","text":"Type: <bBug</bI get nightly updates on VSCode insiders and I get the crash on whenever I restart the app for updating it. VS Code version: Code - Insiders 1.81.0-insider (, 2023-07-13T05:33:32.822Z) OS version: WindowsNT x64 10.0. Modes: <details<summarySystem Info</summary </details<details<summaryExtensions (21)</summaryExtensionVersion ------ vscode-chromium-vector-icons1.0.2 togglehs0.1.2 gitlens2023.7.1305 copilot1.96.259 copilot-chat0.5. mojom-language0.0.1 vscode-webidl0.0.1 vscode-clangd0.1.24 cpplint0.1.3 autopep82023.7. python2023.13. vscode-pylance2023.7.21 remote-ssh0.103. remote-ssh-edit0.86.0 cpptools1.16.3 remote-explorer0.5. gnls0.1.4 java1.21. code-spell-checker2.20.5 change-case1.0.0 vscode-proto30.5.5 </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nAny details on the crash? Any error messages? Any logs?\nI've attached the Crash Dump. Note: I used 7-zip to compress to 7z and then to zip (to make it smaller that GitHub allowed size and format).\nThe same thing has been happening to me.\nThis has been happening to me too.\nThanks guys, I've confirmed the issue!\nGot a stack trace:\nCaused by , namely:"}
{"_id":"q-en-vscode-0aa6c0623ce45e993474f2f7cc47455f8e08ae7c7c39882dbee5892dc94666b1","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: No <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.80.1 OS Version: Windows 10 The C++ extension would like to redirect our users to our walkthrough if they pick an option in one of our quick picks: \"help me install a compiler\". To achieve this, we've been trying to invoke the command with a link to the correct step in the walkthrough similar to how python does it. The problem I'm seeing is that if we run this command before the Welcome page is loaded, it does not navigate to the walkthrough at all. It just displays the Welcome page and the following error appears in the Developer Console: . If we run the command after the Welcome page loads (or just run the command twice in a row), it will work. To work around it, I was going to try to detect whether the Weclome page or my walkthrough loaded after the first command attempt, but it doesn't look like the Welcome page's tab shows up in the list of so I'm not sure how to detect it. The safest thing for now seems to be to run the command twice every time, but it would be nice if it could be fixed to work on the first attempt.\nI can't get this to work, even after opening the welcome window first - it only ever opens the generic welcome page for me. I tried with the c++ walkthrough steps and one in jupyter ext Am I doing something wrong?\nFixed by\nThis still doesn't work properly. The welcome page opens on the correct page, but then immediately switches back to the main welcome page. !\nI'm unable to repro Here are my repro steps: a sample extension a command in the sample extension to open a walkthrough coming from an extension using and installed test extension and invoked the above command. Other things to try: Invoke command before opening welcome page and after.\nrepro's when debugging the extension with the command at least (I didn't try packaging into a visx) a command to the helloworld-sample extension in the and registered during activation install 5 to run, run the command :bug: similar to though it also sort of tries to shift away from the main walkthrough page, then it works correctly the second time !"}
{"_id":"q-en-vscode-3c9c0ec2e5082398d0f537f325f2b0e1b6e5b110759bddc24dad11686730e0c1","text":"From version 1.80 onwards, we've been getting CI failures when testing our vscode extension with this error message: \"Error while encrypting the text provided to safeStorage.encryptString. Encryption is not available.\" I suspect this is related to the move off keytar described in the release notes of 1.80 and here For the purposes of testing our extension, we pass additional launchArgs to VSCode, notably . As I understand it, this instructs VSCode to use some sort of in memory keytar so that the secrets are not actually persisted to disk? This is the desired behaviour for us so that no secrets persist after the the test suite concludes. This has been working well until now, and based on a quick search for on GitHub, it seems like a popular approach when testing VSCode extensions. Notably this only seems to happen in our linux CI. Mac and Windows continue to work without a problem (as also demonstrated below) <!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: , Same issue happens on 1.80.1 OS Version: Ubuntu 22.04.2 Steps to Reproduce: I've created a minimal repro here: The line that ends up resulting in the error: GitHub Actions Build showing the error: PR demonstrating no errors happened on :\nWhen encryption isn't available, we use an in-memory store instead so I think the current behavior should give you the desired behavior, no? Or does it actually throw instead and cause the extension to not function?\nYeah exactly, it just throws right now (on linux). Same code works without issue on 1.79\nI think this is because you see a dialog in the product to ask you to do something. I need to make that dialog non-blocking. In the meantime, you can do this: And configure basic text encryption which is fine because the container will be destroyed after.\nThanks! I to the launchArgs, but seem to be getting the same error:\nI think maybe you want not . I just tried on my Ubuntu machine and worked.\nFYI I have opened to track allowing an in memory secret storage. If we get you unblocked with I'll dupe this to that.\nthanks for continuing to look into this! I applied your suggestion ( without the quotes) but looks like on GitHub Actions (ubuntu) it still results in the same issue:\nit should be instead of in , without that will not be called and causes the above error to be thrown.\nSetting seems to do the trick for me – thank you!\nFYI, I will change it to be to align with what Electron actually accepts."}
{"_id":"q-en-vscode-6df5d077d244d18da8ffaad02ec81fe820c1a9168b49cce5e2d934dbf5a546b3","text":"all chat audio cues enabled, run something in the inline chat that takes more than 4 seconds to come back that the chat response received cue doesn't play - in the console, there's an error about how a user gesture is required. this is expected to a different tab, leaving the chat widget still open to the tab with the chat widget visible hear the chat response received cue going between tabs and notice the cue gets played every time\nso this happens even when it doesn't take that long !\nthis is because is called when the editor tab changes\nthat happens here !\nlooks like we need to move this to before is deleted so we can call it like\nI have a different approach that will work. Ideally we would fix this at the root - so pls feel free to do so as you're more familiar with the state code.\nThis still repros for me with the steps from\nSo switching tabs plays an audio cue for you ?\nThis does not repro for me with the steps in this issue or the steps in the other one"}
{"_id":"q-en-vscode-628657b2da39467893c306bb0c35ccb808da225fa971962dcd660589102139cc","text":"When a user opens a file extension that does not have a specific recommendation tied to it, we show the general notification that points users to the marketplace. This has the following flaws: Most popular file types are already covered by specific extensions, so what gets left in this experience is leftovers For most of the extensions there is no good extension to handle this experience. Thus opening the Extension page does not provide value This is adding un-needed pings to the Marketplace service Also we had a bug that this feature was not available in stable and only 3 users found it is not there Would love for us to get some telemetry on this actions to see if this is actually bringing value to users. If not, we should either: Remove this flow completely Have a pre-defined set of file extensions for which we show this. Checking with tags seems to be very faulty. In my example below there is not an extension that helps with PowerPoint files. !\nOnly of these notifications end up with users clicking on Search Marketplace ! When it is actually clicked here are the top extensions ! Thus I suggest that we: Remove this dialog Add explicit extension recommendation for and . Since those seem to be missing. what do you think?\nSounds good to me\nGreat. Ok, you remove the dialog when you find time, I will add the recommendations for csv and pdf. Or if you go on vacation, I can also remove the dialog. I am working in August. Thanks\nFeel free to remove it in August, otherwise I can also do it in September. Here it is\nTo verify: a file with some strange extension. For example -make sure no notification shows up a .py file or .go file. Make sure you get an extension recommendation for specific python / go extensions"}
{"_id":"q-en-vscode-116ba27b0883784b846939a6d3ed8ef4704569b3e423f4c0ec5952ef7bb8d00a","text":"Notice the \"Define Keybinding\" button: !\nhow did you get this to happen? I cannot repro\nThis happens when the keybinding editor is in the background\nyeah, that's what I tried. !\nmight have been fixed via that styling pr where I a high z-index?\nCan you check the dom? z-index shouldn't be the solution to this problem\ni'm not able to reproduce so don't know if checking the dom would help\nThis issue has been closed automatically because it needs more information and has not had recent activity. See also our guidelines. Happy Coding!\n! I'm able to reproduce it consistently. Open keyboard shortcuts (JSON) While the editor has focus, press alt+F1 Notice that the dialog has the \"define keybinding\" button Also, I can only use alt+f1 once. After I close the dialog I have to reload VS Code. And I suggest to add the accessibility help command to the F1 command palette. I have a different keybinding on Alt+F1 and for testing this feature, it would be very helpful if you don't have to lookup/redefine a keybinding.\nStill cannot reproduce but will keep trying it's in the command palette <img width=\"671\" alt=\"Screenshot 2023-08-10 at 11 13 34 AM\" src=\"\"that's a different bug, thanks.\nOh sorry, I didn't see it! I can also investigate the context key issue if it doesn't reproduce for you. ~Actually, that is very weird, now I cannot reproduce anymore. But I was able to reproduce on two different devices (though with synced settings).~ It seems like it happens when something other than the editor has focus: ! When I now open accessibility help using the command palette, I get this: !\nOh, like the panel?\nYes"}
{"_id":"q-en-vscode-25cddc7c782fbe70fe2775969842ebab0eaf509398ab83f8d9e22f187b59086d","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes/No <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: Insiders OS Version: macOS Steps to Reproduce: a notebook it a memory snapshot cell text models are not released <img width=\"1648\" alt=\"image\" src=\"\""}
{"_id":"q-en-vscode-ccfb993c1fdeaeaff83680294cef25d1fa0a0d7553706bdea509c7932115eb1a","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: code-cli insiders 1.81.0 (commit ) OS Version: Windows 10 21H2 .3086 I believe is only partially fixed, even in code-cli 1.81.0 (commit ). The window does not appear when only or code- (the cli) is running, however, as soon as a tunnel is connected to, a window appears (as code- is run). If the user closes the window, the tunnel connection is dropped. I believe the fix is to pass CREATENOWINDOW to creationflags in spawnserverprocess as seen here: However, without the proprietary \"mixin\" , it seems my builds of code cli are not able to properly run code-server, failing with [2023-06-28 19:09:25] info [rpc.3] Starting server... [2023-06-28 19:09:25] debug [rpc.3] Starting server with command... Command { std: \"C:Usersredacted.vscode-insiderscliserversStable-serverbincode-\" \"--connection-token=redacted\" \"--accept-server-license-terms\" \"--start-server\" \"--enable-remote-auto-shutdown\" \"--socket-path=.pipecode-oss--ab53-485a-bc7d-\", killon_drop: false } [2023-06-28 19:09:25] trace [rpc.3] [codeserver.13] The syntax of the command is incorrect. [2023-06-28 19:09:26] debug [tunnels::connections::ws] received liveness pong To reproduce the issue, on a windows machine: an arbitrary tunnel to authorize the machine: code- tunnel --name test the tunnel with the code at the tunnel service with code- service tunnel install that no is visable, and code- is running in the background connect to the tunnel again a appear on the desktop, close the and observe the tunnel disconnect on\nThanks for the issue. Please feel free to put in the PR, I can edit it if needed so that you get the credit for the fix :)\nConfirming this fix is working in . On tunnel connect a does not appear."}
{"_id":"q-en-vscode-0b2d63aa32b4b3f059c38b48e597c82b43dbe5ee57e6ff281455f3dd3bd1239f","text":"Tracking issue to ensure that we can run a build with all tests passing, mainly smoke tests. Previously tracked as"}
{"_id":"q-en-vscode-5769d128ddde591ff70480ec63cd7af26f1d0a13b5dc2e2933f315bfcfa86875","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: Latest OS Version: MacOS Latest Steps to Reproduce: I have already disabled the following but I still have an empty top bar taking space in fullscreen mode, whereas before this change I could have removed the topbar completely. Before <img width=\"1439\" alt=\"Screenshot 2023-08-03 at 10 13 57\" src=\"\"After <img width=\"1439\" alt=\"Screenshot 2023-08-03 at 10 13 32\" src=\"\"Similar issue:\nWorkaround: It seems to be fixed once I exit/enter the fullscreen mode again.\nTo clarify--this is the issue you're running into? That issue is just about being able to toggle it in the first place, which does work as expected today. This behavior seems scoped to fullscreen mode specifically.\nI also have this problem. I need that top bar completely removed, but I can turn off stuff and it still stays up. Is this really forced on us to not be able to remove?\nactually that worked, had to go outside of fullscreen and now in fullscreen it correctly disappears. Cheers ;)\nYeah, I don't think this is intentional but just wanted to confirm the issue."}
{"_id":"q-en-vscode-64284108d9cb2e3db6dabbb32f2356365067744caa8ab6bd094976e4d3aed7d6","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes Executing current (1) cell with Shift-Enter jumps/scrolls to the end of the output section of next (2) cell. This causes problems if the next cell has an output already that is very long (eg. graphs). You have to scroll back a lot unnecessarily. The screen will jumpt to the end of the output of the next (2) cell. It should ideally jump to the code part of the next (2) cell, showing the code on the middle of the screen, allowing to see the end of the output of the just executed (1) cell. Current behaviour is a bit illogical and also you have to scroll back to see the results of the cell you just ran. See Video: <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --Version: 1.81.0 (system setup) Commit: Date: 2023-08-02T12:37:13.485Z Electron: 22.3.18 ElectronBuildId: Chromium: 108.0.5359.215 16.17.1 V8: 10.8.168.25-electron.0 OS: Windows_NT x64 10.0. Steps to Reproduce: Have one cell with any code a cell below that outputs lots of output, more than a screen worth of results the cell (1) with shift-Enter\nduplicate of\nre-opening since this behavior will still exist for and could be improved\nI'm unable to get the next cell to scroll completely Here's my notebook document along with the output <details<summaryipynb file</summary</detailsScroll setting has been set to When shift+executing the second cell (which has 40 lines), I expect the 3rd cell to be displayed towards the middel or at least stay at the bottom, so that I can see the output from the second cell (as thats' what i just ran) I would expect there to be enough scrolling to display the next cell as well ass the output of the previous cell I.e. display the next cell towards the middle As you can see from the screen recording, the previous cell is still running (blue progress bar on the top) However the 3rd cell is taking up all of the realestate Actually you can replicate this even when running the first cell (shift+ enter). There no way to see the previous output except change the scroll setting to I would expect there to be no scrolling at all, as the first line in the next cell (in fact full input cell) is visible\nplease feel free to create a new issue and close this if its unrelated.\n- I'm not sure I follow, but you mentioned a screen recording. Can you create a new issue with that? This issue is also for the default\nverification: in settings (default) two cells with the code both cells so they have output the viewport so that one output takes up the whole view by itself the first cell with shift+enter that the input of the second cell (which should now be focused) is fully visible"}
{"_id":"q-en-vscode-aa2c64d80a38bb0fd0720447e5ead3e15298d59f650c0aa1c63997db3fad5281","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: N/A (remote extensions are needed) <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.81.0 (Universal) and 1.82.0-insider (Universal) OS Version: MacOS 13.3.1 (22E261) [Ventura] Steps to Reproduce: a repository with a devcontainer from GitHub Codespaces (either in browser VS Code, or desktop) at the Source Control tab - it says \"Scanning folder for git repositories…\" I have seen this issues before months back, and it was fixed in Insiders and subsequently the release version - but it is back again and very frustrating. I have a minimal repo here - you will need to create a Codespace to see the issue (or choose a different devcontainer workflow). The issue could be to do with the docker image used, I think last time there was some suggestion of a problem with Alpine (or Alpine containers), and I believe the Jekyll container is based on Alpine. too.\nit looks like has resurfaced. I am able to reproduce with the latest Insiders release.\nVery likely candidate , will investigate tomorrow.\nSteps for verification: Reopen in a container Verify server launches and integrated terminal works fine\nI've checked in the latest Insiders release and the same bug seems to still be present when using GitHub Codespaces (both in web version of VS Code insiders, and desktop version of VS Code insiders when connecting to the Codespace) Version: 1.83.0-insider (Universal) Commit: Date: 2023-09-13T05:34:48.836Z Electron: 25.8.0 ElectronBuildId: Chromium: 114.0.5735.289 18.15.0 V8: 11.4.183.29-electron.0 OS: Darwin x64 22.4.0\nthe alpine servers run fine with the provided container configuration, however the git scm is hanging for a different reason can you please look into it\nI have created a new Codespace on the linked repository and I am still able to reproduce the problem. As far as I can tell the extension how is completely busted as running \"Show Running Extensions\" is empty. In the Window output log I see the following: If I look at the Git extension I see the following: !\nAh missed the logs, is based on whereas our server alpine modules are built on . We don't plan to revert the alpine version as is reaching EOL I am closing this as designed, please update the jekyll container to use"}
{"_id":"q-en-vscode-34b8e75c3a12ed63ed5cbf6a3587ef08951ad61175d7fd8fdfc2663eebb11cee","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --Currently, it says \"No forwarded ports. Forward a port to access your running services locally.\". But when using the Ports tab locally (no remote), this doesn't have the same meaning because I can already access the port locally. The suggestion is to change the text in the local port forwarding case to show something more meaningful. Just as an example, something like \"No forwarded ports. Forward a port to securely access your locally running services over the Internet.\""}
{"_id":"q-en-vscode-01ee4f029461a1c38c07ba2f418beb83be0917d7afc94ce0c03a229ca553fcd2","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.81.0 OS Version: Windows_NT x64 10.0. Steps to Reproduce: In , set the content to then start VS Code with shell integration enabled and use Windows PowerShell. This bug will occur: The use of at: Needs to be guarded for Windows PowerShell (where it doesn't exist). /cc\nSharing from my profile:"}
{"_id":"q-en-vscode-7c7b9886e8a0d7d9a8ebbd022b50f80fdea42f03c94ddb3433038b27c6cd0f6d","text":"Type: <bBug</bOpen terminal. Start typing. Note the noticeable delay in the text rendering on screen even though the cursor advances. In fact some timess the text never renders. Makes the terminal pretty unusable. VS Code version: Code - Insiders 1.82.0-insider (, 2023-08-10T05:33:20.243Z) OS version: WindowsNT x64 10.0. Modes: <details<summarySystem Info</summary </details<details<summaryExtensions (19)</summaryExtensionVersion ------ togglehs0.1.2 vscode-eslint2.4.2 githistory0.6.20 gitlens2023.8.1005 dotnet-test-explorer0.7.8 copilot1.101.325 copilot-chat0.7. vscode-mojom0.1.4 mojom-language0.0.1 vscode-clangd0.1.24 csharp2.0.328 vscode-dotnet-runtime1.6.0 python2023.15. vscode-pylance2023.8.21 remote-wsl0.81.0 cpptools1.17.1 powershell2023.6.0 vscode-jest5.2.3 clang-format1.9.0 </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\ndoes changing help?\nYes. It apparently was set to auto previously. If it turn it to \"off\" the issue goes away.\nrepro's for me with set to or . does not repro for or gif: !\nI can't reproduce this unfortunately\nI still have a consistent repro if you want me to try anything or collect diagnostics.\nFindings from looking at it together: It repro'd in Code - OSS It did not repro in xterm demo It did not repro in Code - OSS with symlinked webgl addon It repro'd after running scripts/update-xterm\nI'm going to manually look through all the code that went in around rendering ~1-2 weeks ago to see if I can spot something.\nMaybe it's related to Chromium.\nUpstream bug filed at , thanks to\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!\nVerified on Linux\nverified on windows"}
{"_id":"q-en-vscode-07633b17c72c251edb8a9fbe4372bff87a9ef95920177671a60f8ae970820707","text":"a screen reader enabled, run some commands enter arrow left and right arrow keys to inspect the previously run command you cannot"}
{"_id":"q-en-vscode-3a92ad595f536f78a5bb3c8459c7ff221be34fc83ee25a6564ef714a46724ea4","text":"Toggling the command center works, but the title bar height is stuck until the window is resized. I also noticed some issues where I had to click twice to open the command center once enabled. The command center input then disappears while the command center quick pick opens.\nMaybe an electron update regression"}
{"_id":"q-en-vscode-4a3056e60a8e26874b844a68be4375f6f4925faab028683bbd03b230496e5932","text":"Some small tweaks are needed for the new -type it should be an interface, not a type. it lacks a jsdoc comment it's not clear why can be optional?\nWill correct it right away It's not something we expect all consumers to consider, as many folks may not care about applying env variables to a particular workspace. In future we may introduce other parameters, and consumers who don't care about it, can simply omit them: A consumer who only cares about shell: cc/"}
{"_id":"q-en-vscode-23a3e1342b753dcf7dbdfb32e66df6c959585e6733054659e0014d92316e74fa","text":"the notification content is sufficiently long, insufficient height is allocated for the action buttons to grow, consequently they appear cramped and lack border radius a notification is focused the focus border radius is truncated !\nThe border radius issue on the buttons still occurs if the VS Code window is resized !\nI cannot reproduce when resizing the window, are you sure you tested on latest? Here is the sample code I run from a command to trigger this: !\nYeah, still repros in today's insiders <img width=\"328\" alt=\"image\" src=\"\"\nCan you distill better steps for me to reproduce exactly into this state, ideally even running out of sources?\nWhat I do is open the VS Code repo in VS Code Insiders without having first actioned the notifications above, you can probably trigger the notifications by starting with fresh user data and extensions dir. Then resize the VS Code window so that the notification text starts to wrap. Eventually the button radius gets cut off when I do this.\nthough I don't see the button's border getting cut off but I do see the whole notification border is getting cut off at the corner ! steps to reproduce prerequisite -have 2 profiles with recommended extension not installed to switch the profiles will show the above dialog click on it so that notification get's focus and see the border at the corner is not rounded properly\nbut for this to reproduce you need to make the window small enough that the notification itself gets smaller in width right? In that case, there is a slight chance that the text of the notification wraps to the next line and then pushing the buttons down. Notifications do not have any logic to grow larger to make room for the wrapped text because we assume that they are small enough that a user would not cause them to shrink. You would need to make the window size smaller than for this to happen (notifications are wide). I still fee the original issue is addressed, meaning that the notification when opening applies the correct size.\n! !\nYes this is true, I will close and mark verified"}
{"_id":"q-en-vscode-c69ca03e5ea5ab88f6d79fd635a5e55ca20dfd572be898332b00bd0bb3c0a5ef","text":"Testing open enable search for click on a result in observe the selection is set on the input code and not on the output\nSeems that navigating to the result also causes it to flicker (~13 seconds in to the sceencap)."}
{"_id":"q-en-vscode-a3be996e1d2f8dfcc17e17d9e8b38411c7c2dad5007483214ca1ef7ddf5984f5","text":"Crash keys to help narrow down the source of crash\nis the first build with above crash keys annotated, result Wrapper class is not the one from the list, need to add the missing ones , , , and\nCrash has not been triggered with updating to Electron 27, closing optimistically.\nI am not sure how to verify this but I will monitor again for new integration test crashes if there are any crashes attached.\nWe have a new instance of this crash with Electron 28, reopening for debugging.\nHad a new failure with Electron 29 as well, refs Sadly I missed porting the gin annotations to 29 branch\nAnother run\nAnother crash\nSeems like the crash rate of this particular one has increased again, adding back the crash keys again.\nAdded crash keys and triggered test builds to debug further. Work can be tracked in\nI have now tracked down the source of crash, but before I get down into the final details lets walk through the nature of the crash and understand what is happening. The crash started appearing a year ago in our CI mainly with integration tests across all 3 platforms inconsistently. Sometimes the crash would disappear after runtime updates but recently it has been pretty consistent in our Windows CI. Although I was unsuccessful to repro it locally but given its frequent presence in CI I decided to rely on which are meant for cases like this. On shutdown of the main process, the runtime eventually has to gracefully cleanup the environment and its associated tasks from the event loop which in rare cases lead to UAF in certain listeners. The initial reports point to the resource cleanup in Newer reports point to the resource cleanup in The runtime exposes JS api that are backed by C++ implementations through V8. is the base class that allows exposing C++ objects as corresponding V8 objects. For example to expose a following object in Javascript, This is how we would do the same in C++ and expose it through V8 The work to construct an object instance from this C++ class will eventually be performed by . works in a similar fashion but it is limited to holding the C++ function pointers that needs to be invoked from V8. In the above example, these pointers would be and . Now going back to the implementation of , you will see that a parameter called is passed in to the function. This is the pointer to the static member from the above example which can store metadata about the object, in this case it about the type of owner of the object in C++. The primary use case of this pointer in is to act as a cache key to store the object templates. The second point of interest in this implementation is the call to which acts the primary resource management hook between the C++ heap and Javascript heap. This works pretty much the same as how . We request a callback to be invoked when the corresponding object is garbage collected. You can refer to for why there are two separate phases of callback registration in the C++ version. class 1) It could be deleted when the V8 isolate gets disposed ;l=39-45 2) Or, it could be deleted when the second weak callback fires The order of destruction between which is held by and guarantees that the destruction entry points don't race against each other. I will leave the analysis of this destruction sequence to the interested readers. The initial reports which don't appear now could be attributed to consuming the fix from https://chromium- and I can only speculate about this as there is not much we can understand from the old reports. class 1) It could be destroyed when the second weak callback fires ;l=39 But there is another scenario of destruction happening during shutdown that is not captured. This is where crash keys come in handy. The keys I are in 1) Extended the to hold the name of the C++ class that is impacted 2) Extended the with a new boolean that will be set to when is called. This will help in cases where either of the weak callbacks are called after the class is destroyed. 3) I also want to see the other way, so new boolean member to which will be set to when the second weak callback is called. We already have the member that tracks whether the first weak callback was called. Now in we can have condition that basically checks which can confidently pinpoint that someelse destroyed this class between the first and second callbacks. After a couple of retries in the CI, I was able to trigger the crash As expected I had a bunch of dumps from and one for the actual crash. This is how they looked, see the size of the crash dumps to know which one had the crash keys, <img width=\"772\" alt=\"Screenshot 2024-09-26 at 12 54 17\" src=\"\"Crash key values Run DumpWithoutCrashing stack Actual crash stack So we had a class in Electron that worked in theory similar to which destroys the C++ object on the disposal of isolates in concern that the finalization callbacks might have not run But the only issue is that function runs at the destruction of the microtasks runner which can race with the pending task flushing from the event loop on destruction of as seen from the stack traces. 1) Convert into a that actually watches destruction of 2) Clear weakness for the wrapper inside so that we can do the following in second weak callback If you have made it this far, hope you enjoyed this debugging session! Now this disjoint between the C++ heap and Javascript heap leading to these uaf issues is not new. The V8 team had already worked on addressing it for the Blink rendering engine via the project called and it is also available as a standalone library for embedding in any C++ project that uses V8 Through this library, the trace based garbage collector can now view both the C++ heap and Javascript heap together to help make lifetime management simpler. IIUC, could adopt cppgc when the heap setup mechanisms are moved from blink into V8 as part of We can then explore the adoption in Electron."}
{"_id":"q-en-vscode-510f017fa6dea14c91b8d25bf8dc40a60168ebc11ef8a24b4d751ef9b6e2ae86","text":"Probably should print an error message instead of panicking, and maybe do that before printing the URL to connect to.\nIf I use port 8000 in another process, then run , it still says but it's not actually available on that port. I would expect it to pick 8001 or something.\nMake sure you're binding to the same interface. serve-web always binds to 127.0.0.1. Depending on your OS and network, that may not conflict with a service listening on if localhost is ipv6."}
{"_id":"q-en-vscode-20205698d489a734bfb9c22b683dab480d6ff3ead83328d887203e1ab5509493","text":"There's no separator line between the two components. !\nSeeing the seperator: <img width=\"69\" alt=\"image\" src=\"\""}
{"_id":"q-en-vscode-cc74cd66306356b36c1d44661605c8c08bc993b0d24749e1f159a7501b06e796","text":"a notebook where two cells take up more than the scrollable region between the two cells, note how far down the region scrolls the panel off/on between the two cells again the scroll placement is different because returns a different value. !"}
{"_id":"q-en-vscode-26b51cc88b908ddcabeaa7858e5a5659183cfda597efc87e020cc82d40bda563","text":"Type: <bBug</bCC a new terminal and hit enter in a new line and hit enter (you must create at least two lines of code to test b/c the single line does not have this issue) UpArrow LeftArrow multiple times to test if you can hear \"echo world\" character by character. Notes: For NVDA and JAWS on Windows, it keeps saying \"d\" which is the last character. VS Code version: Code - Insiders 1.83.0-insider (, 2023-09-19T08:56:35.775Z) OS version: WindowsNT x64 10.0. Modes: <details<summarySystem Info</summary </details<details<summaryExtensions (91)</summaryExtensionVersion ------ android-dev-ext1.3.2 aiprm-lang0.0.2 Bookmarks13.4.1 openscad1.2.1 spellright3.0.118 zoterolatex0.4.1 mermaid-markdown-syntax-highlighting1.5.2 doxdocgen1.4.0 vscode-markdownlint0.51.0 vscode-eslint2.4.2 vscode-quick-select0.2.9 vscode-deno3.22.0 gitlens14.3.0 EditorConfig0.16.4 prettier-vscode10.1.0 vscode-google-translate1.4.13 codespaces1.15.3 copilot1.112.422 copilot-chat0.8. remotehub0.60.0 vscode-github-actions0.26.2 vscode-pull-request-github0.72.0 overleaf-workshop0.1.5 cslpreview0.2.2 easy-snippet0.6.3 path-autocomplete1.25.0 latex-workshop9.14.0 lilypond-syntax0.1.1 scheme0.2.0 better-cpp-syntax1.17.2 google-search0.0.1 vscode-lua-format1.3.8 lilypond-formatter0.2.3 lilypond-pdf-preview0.2.8 lilypond-snippets0.1.1 vslilypond1.7.3 zotero0.1.10 git-graph1.30.0 vscode-docker1.26.0 black-formatter2023.4.1 flake82023.6.0 isort2023.11. python2023.16.0 vscode-pylance2023.9.10 jupyter2023.8. jupyter-keymap1.1.2 jupyter-renderers1.0.17 vscode-jupyter-cell-tags0.1.8 vscode-jupyter-slideshow0.1.5 remote-containers0.311.0 remote-ssh0.106.4 remote-ssh-edit0.86.0 remote-wsl0.81.3 vscode-remote-extensionpack0.24.0 azure-repos0.36.0 cmake-tools1.15.31 cpptools1.17.5 cpptools-extension-pack1.3.0 js-debug-nightly2023.9.1317 powershell2023.6.0 remote-repositories0.38.1 vscode-github-issue-notebooks0.0.129 vscode-selfhost-test-provider0.3.18 vscode-serial-monitor0.10.0 vsliveshare1.0.5883 autodocstring0.6.1 pandocciter0.10.3 shiny-python0.1.4 shinyuieditor0.4.3 quarto1.98.0 r-debugger0.5.4 java1.22.1 vscode-xml0.26.1 r2.8.1 multi-command1.6.0 vscode-deepl1.0.6 abc-music0.4.0 lua3.7.0 latex-utilities0.4.10 cmake0.0.17 errorlens3.13.0 intellicode-api-usage-examples0.2.8 vscodeintellicode1.2.30 vscode-arduino0.6.0 vscode-java-debug0.54.0 vscode-java-dependency0.23.1 vscode-java-pack0.25.14 vscode-java-test0.39.1 vscode-maven0.42.0 markdown-all-in-one3.5.1 grammarly0.22.1 (1 theme extensions excluded) </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nYeah, this is acting very buggy on windows\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!\nrepro steps as listed should work. Lmk if you have questions\nNavigating left and right is fine, but after typing a command, NVDA reads out a misspelled version of the command multiple times !\nreopening because marked verification-found per the process.\nIt's a lot better today. Closing and marking as verified. Edit: nvm, but will file a new issue for the bug.\ncan still repro. He's going to send me the output logs. The command decorations are off here. !\nthe command start line is off here, which is causing this issue. Do you know why / how that happens?\nDoes not reproduce for me on Windows.\nthis can happen sometimes on Windows. What I'd recommend is add some logic when a command starts that does a quick scan of nearby lines to re-adjust for the common PS/cmd prompt case if the current line does not look like a prompt. This is a workaround until shell integration sequences are passed through correctly by conpty.\nfeature as we'll want to call this out in the terminal section of the release notes"}
{"_id":"q-en-vscode-274997f8bd5de0c53b59e452f123d07a64d370471eba0be41443ddbec6c6eb9f","text":"I'm on a mac. setting it to be works though\nit started to work the moment I had at least one comment in the editor\ndid the original keybinding work after you comment? Even before adding a comment I'm seeing the command work.\nthat keybinding still doesn't work for me (maybe my keyboard or something)...? when I use a different keybinding, it does try to run the command, but I'm now seeing this error. !\nI have confirmed that the default keyboard shortcut works on mac: . There's probably something up with your keybindings that's causing this. You could try the \"Developer: Toggle Keyboard Shortcuts Troubleshooting\" command and see what it outputs. I'm fixing this error though:"}
{"_id":"q-en-vscode-90ebc04d37c53f27bc2968785527ea15d91647f65dd5ac7492ffb85632a20c20","text":"Testing Followed the steps to turn on the debug toolbar with commandCenter options. However, when the debugging started, clicking the area around the directory name did not trigger 'search files'. (the blank space at the right side of 'vscode-python' in terms of the attached screenshot. I was able to trigger 'search files' only when pressing specifically on the directory name, in this case, 'vscode-python. ++ Nice feature by the way! <img width=\"983\" alt=\"Screenshot 2023-09-26 at 10 42 11 AM\" src=\"\"<img width=\"512\" alt=\"Screenshot 2023-09-26 at 11 02 46 AM\" src=\"\""}
{"_id":"q-en-vscode-6f46db60608292e9463028a7120ebeffb029b5048851b6c638cb079a774d58c9","text":"Testing I was expecting quick fixes for the whole line to be included, ranked by distance to the cursor, up to a certain maximum !\nYep! That was one thing I was experimenting with, and atm it does just find the closest one. Played around with allowing all quickfixes as well, but was unsure what would feel better. Will likely work on a version where we show all potential quickfixes in the line."}
{"_id":"q-en-vscode-cafe40620ace9e08faf77e85028a8e9dbe68393ffcb2bc679ca1838058026276","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes/No <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: OS Version: Steps to Reproduce: on a file from the \"All Changes\" node that the diff editor does not have the commit for the right hand side !"}
{"_id":"q-en-vscode-9ebabdf583487d56f1722ef8cf8feadd25fcd9e36d186b3de524501ea65d1018","text":"<!-- Do Not Delete This! bugreporttemplate --- VS Code Version: 1.83 Insiders OS Version: Windows 11 The friendly Settings editor name for should be \"Editor Code Action Widget: Include Nearby Quick Fixes\" not \"Quickfixes\". Description should also be \"...nearest Quick Fix within a line...\" to make other mentions of Quick Fix. !\nFor example: !"}
{"_id":"q-en-vscode-77a03d3a5b1d1cf3cfa64ddc6224b6d7c8b7c3c0302ae63925b2a066129033e7","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes/No <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: OS Version: Steps to Reproduce: up a keybinding to launch quick text search a diff editor some text quick text search Nothing in input: ! Expected: !\nHi , I guess the PR has fixed this already. I tried reproducing it locally with the changes of my current PR, and it seems to be working fine. Attaching the screen recording for your reference.\nYou need to open a diff editor. For example: open a folder that is a git repo Make a change in any file Go to the source control viewlet and click on the changed file: <img width=\"1679\" alt=\"Screenshot 2023-10-08 at 2 57 57 PM\" src=\"\"Select text in the right editor The bug appears.\nGot it, will try to correct it\nHi , I understand that you must be busy. Can you take a look any time you are available? I need some help in figuring out where exactly the diff editor is interacting with search bindings. For example, I am not able to understand how hitting cmd+f calls the search function from the diff, or where it is getting called from. Similarly for the quickTextSearch, can you point me in the right direction?\nI was looking for code pointers and ended up finding the bug anyways:"}
{"_id":"q-en-vscode-3719c789f525efbbd73143a892b037ae53682d56cc2f30a1bafc110f49da2acf","text":"I tested on Win11 & Win10. It works as long as you start a new workbook and don't save the jupyter notebook. Once you save the notebook, your pastes of images from the clipboard will no longer be treated as inline attachments. Originally posted by in\nPossibly getting overridden by a markdown file paste provider?\nwere you able to replicate my findings?\nReviewed this thoroughly, and found the following. All cases are performed with a screenshot copied to the clipboard, and pasting into markdown notebook cells using ctrl+v. Using Windows 11 and current insiders. With setting unsaved notebook (command pallete -new jupyuter notebook -) results in broken markdown image link pasted, broken rendering (see screenshots attached) ! ! saved notebook creates image file on disk, creates markdown image link, renders correctly. With setting unsaved notebook (command pallete -new jupyuter notebook -) results in correct inline attachment image link pasted, renders correctly saved notebook results in correct inline attachment image link pasted, renders correctly\nWith the above investigation, I believe this is likely an issue with markdown paste providers, assigning to take a look at this. to clarify, are you expecting the images to be created on-disk upon paste, and referenced via an image markdown link? Or are you looking for the experience, with images stored in the notebook itself. The latter can be achieved and still works properly with the setting\nI prefer the later, where the images are stored in the notebook, this makes it easier to share with others in these scenarios. This used to be working and then stopped working in the build I noted in the original question refenced, but I just tested in the latest build and it is back to working as expected, so must have been a bug that has now been fixed: The version I just tested with is updated now and it is working as expected with the Version: 1.83.1 (system setup) Commit: Date: 2023-10-10T23:48:05.904Z Electron: 25.8.4 ElectronBuildId: Chromium: 114.0.5735.289 18.15.0 V8: 11.4.183.29-electron.0 OS: Windows_NT x64 10.0.\nWorks fine for me too with the default settings. Will disable pasting into untitled notebooks with the setting off"}
{"_id":"q-en-vscode-5d622880b2ac1b3748c1115ca67b710ba5da7b25867897b0ae009795ef109eac","text":"I was working with a notebook, switched tabs, and tried to switch back, now the editor is blank and my progress with the notebook is lost. this seems to repro every time.\nstack is irrelevant, the issue can repro without any errors"}
{"_id":"q-en-vscode-64332eaf79976e8d6fb85b92f7d1c28daa3b48f1ece0e04b99e72d14ce97ac79","text":"Type: <bBug</bPowerShell in editor area NOT type anything in the terminal input alt+F2 to open accessible view for the terminal ctrl+shift+o to open symbol list DownArrow key and you will hear: \"No editor symbols\" ESC key Note: pressing ESC key also closes the Accessible Buffer View. VS Code version: Code - Insiders 1.84.0-insider (, 2023-10-06T05:35:21.061Z) OS version: WindowsNT x64 10.0. Modes: <details<summarySystem Info</summary </details<details<summaryExtensions (92)</summaryExtensionVersion ------ android-dev-ext1.3.2 aiprm-lang0.0.2 Bookmarks13.4.1 openscad1.2.1 spellright3.0.118 zoterolatex0.4.1 mermaid-markdown-syntax-highlighting1.5.2 doxdocgen1.4.0 vscode-markdownlint0.52.0 vscode-eslint2.4.2 vscode-quick-select0.2.9 vscode-deno3.25.1 gitlens14.3.0 EditorConfig0.16.4 prettier-vscode10.1.0 vscode-google-translate1.4.13 codespaces1.15.4 copilot1.123.470 copilot-chat0.9. remotehub0.62.0 vscode-github-actions0.26.2 vscode-pull-request-github0.74.0 overleaf-workshop0.3.0 cslpreview0.2.2 easy-snippet0.6.3 path-autocomplete1.25.0 latex-workshop9.14.1 lilypond-syntax0.1.1 scheme0.2.0 better-cpp-syntax1.17.2 vscode-bard0.0.10 google-search0.0.1 vscode-lua-format1.3.8 lilypond-formatter0.2.3 lilypond-pdf-preview0.2.8 lilypond-snippets0.1.1 vslilypond1.7.3 zotero0.1.10 git-graph1.30.0 vscode-docker1.26.1 black-formatter2023.4.1 flake82023.8.0 isort2023.11. python2023.18.0 vscode-pylance2023.10.10 jupyter2023.9. jupyter-keymap1.1.2 jupyter-renderers1.0.17 vscode-jupyter-cell-tags0.1.8 vscode-jupyter-slideshow0.1.5 remote-containers0.317.0 remote-ssh0.106.5 remote-ssh-edit0.86.0 remote-wsl0.81.7 vscode-remote-extensionpack0.24.0 azure-repos0.38.0 cmake-tools1.15.31 cpptools1.17.5 cpptools-extension-pack1.3.0 js-debug-nightly2023.10.417 powershell2023.9.4 remote-repositories0.40.0 vscode-github-issue-notebooks0.0.130 vscode-selfhost-test-provider0.3.18 vscode-serial-monitor0.10.0 vsliveshare1.0.5883 autodocstring0.6.1 pandocciter0.10.3 shiny-python0.1.4 shinyuieditor0.5.0 quarto1.101.0 r-debugger0.5.4 java1.23.0 vscode-xml0.26.1 r2.8.1 multi-command1.6.0 vscode-deepl1.0.6 abc-music0.4.0 lua3.7.0 latex-utilities0.4.10 cmake0.0.17 errorlens3.14.0 intellicode-api-usage-examples0.2.8 vscodeintellicode1.2.30 vscode-arduino0.6.0 vscode-java-debug0.54.0 vscode-java-dependency0.23.1 vscode-java-pack0.25.14 vscode-java-test0.40.0 vscode-maven0.42.0 markdown-all-in-one3.5.1 grammarly0.22.1 (1 theme extensions excluded) </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nThe editor's go to symbol is getting invoked in this case. We should prevent that from happening via a context key.\nnote that this action is disabled when there are not symbols for the current accessible view. good bug find though . The fix will make it such that no quick pick opens."}
{"_id":"q-en-vscode-106f6398624c7d7a75a10a760a057daf65ba1062bc428ee840f7a92c0e493983","text":"I keep pressing the X on this message and it still keeps coming back. Every single time I open a window. I have told it to go away almost 50 times now. I am being driven towards insanity. <img width=\"826\" alt=\"image\" src=\"\"May this bug report stand in evidence that I have gotten the message. Thank you. Version: 1.83.0 (Universal) Commit: Date: 2023-10-03T16:13:15.449Z (3 days ago) Electron: 25.8.4 ElectronBuildId: Chromium: 114.0.5735.289 18.15.0 V8: 11.4.183.29-electron.0 OS: Darwin x64 18.7.0\nWill just remove the banner.\nshould we mark this as author verification required? I'm assuming those testing internally would have updated their MacOS\nhow do I remove the banner in windows 8?\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!\nis the issue fixed for you on the latest ?\nFeel free to mark as verified via code review."}
{"_id":"q-en-vscode-f37c5fadce182dabe94af5722d703b8b1ec06c6704eb679f02706c262c3a9816","text":"Type: <bBug</bMake some notification and press ctrl+K, ctrl+shift+N and try to clear the notification list. Screen readers are expected to hear \"Cleared\"; however, it is not currently supported. VS Code version: Code - Insiders 1.84.0-insider (, 2023-10-09T06:11:57.204Z) OS version: WindowsNT x64 10.0. Modes: <details<summarySystem Info</summary </details<details<summaryExtensions (92)</summaryExtensionVersion ------ android-dev-ext1.3.2 aiprm-lang0.0.2 Bookmarks13.4.2 openscad1.2.2 spellright3.0.118 zoterolatex0.4.1 mermaid-markdown-syntax-highlighting1.5.2 doxdocgen1.4.0 vscode-markdownlint0.52.0 vscode-eslint2.4.2 vscode-quick-select0.2.9 vscode-deno3.25.1 gitlens14.3.0 EditorConfig0.16.4 prettier-vscode10.1.0 vscode-google-translate1.4.13 codespaces1.15.4 copilot1.123.477 copilot-chat0.9. remotehub0.62.0 vscode-github-actions0.26.2 vscode-pull-request-github0.74.0 overleaf-workshop0.3.0 cslpreview0.2.2 easy-snippet0.6.3 path-autocomplete1.25.0 latex-workshop9.14.1 lilypond-syntax0.1.1 scheme0.2.0 better-cpp-syntax1.17.2 vscode-bard0.0.10 google-search0.0.1 vscode-lua-format1.3.8 lilypond-formatter0.2.3 lilypond-pdf-preview0.2.8 lilypond-snippets0.1.1 vslilypond1.7.3 zotero0.1.10 git-graph1.30.0 vscode-docker1.26.1 black-formatter2023.4.1 flake82023.8.0 isort2023.11. python2023.18.0 vscode-pylance2023.10.10 jupyter2023.9. jupyter-keymap1.1.2 jupyter-renderers1.0.17 vscode-jupyter-cell-tags0.1.8 vscode-jupyter-slideshow0.1.5 remote-containers0.317.0 remote-ssh0.106.5 remote-ssh-edit0.86.0 remote-wsl0.81.7 vscode-remote-extensionpack0.24.0 azure-repos0.38.0 cmake-tools1.15.31 cpptools1.17.5 cpptools-extension-pack1.3.0 js-debug-nightly2023.10.917 powershell2023.9.4 remote-repositories0.40.0 vscode-github-issue-notebooks0.0.130 vscode-selfhost-test-provider0.3.18 vscode-serial-monitor0.10.0 vsliveshare1.0.5883 autodocstring0.6.1 pandocciter0.10.3 shiny-python0.1.4 shinyuieditor0.5.0 quarto1.102.0 r-debugger0.5.4 java1.23.0 vscode-xml0.26.1 r2.8.2 multi-command1.6.0 vscode-deepl1.0.6 abc-music0.4.0 lua3.7.0 latex-utilities0.4.10 cmake0.0.17 errorlens3.14.0 intellicode-api-usage-examples0.2.8 vscodeintellicode1.2.30 vscode-arduino0.6.0 vscode-java-debug0.54.0 vscode-java-dependency0.23.1 vscode-java-pack0.25.14 vscode-java-test0.40.0 vscode-maven0.42.0 markdown-all-in-one3.5.1 grammarly0.22.1 (1 theme extensions excluded) </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --"}
{"_id":"q-en-vscode-075b6d2f1083feb8876901413f2ed51f0d3a58445fd4370a489250f8b43a1f79","text":"After updating I see this: ! After some investigation with we found that it's cause by this line in : Which calls into: Which calls in to which accesses state. This is only initialized later in the method though. Rough repro based on memory: 1 (Windows): Update client to latest 1: Set activity bar to top 1: Sync 2 (macOS, old client): Sync (pulls in activity bar top setting?) 2: Update to latest 2: Open client\nThanks for the fix. I got report of this bug from some users and I am asking for developer tools console log (thanks"}
{"_id":"q-en-vscode-fe6cabaf5a568ddffefc8f96416ff06082e284dfd759ab2263d3f6cc2b3bd9eb","text":"Testing Triggered by trying : !\nCan you clarify what better message you would have expected?\nProbably something telling me how I messed the URL up. Is their an allowlist or some way to know what the valid authorities are so you can tell me how I specified the wrong one? For instance, I specified an HTTP URL that I wanted to open, so being told that some port isn't opened suggested to me that some unspecified port wasn't opened on the GitHub side, not that I have a malformed URL.\nSuch a URL basically requires you to have an extension installed that can handle the remote authority. There is only very few: SSH, DevContainer, WSL and GH and GH Codespaces I think."}
{"_id":"q-en-vscode-1e30af62d56215aea92f0682e427cfc8ecc187e934f1b3ed58aa88f22717e429","text":"Expected ! Actual !\nRegressed in\nThe markdown preview in the chat is also broken for Dark Modern but is fixed by your change: !\nAnd the markdown preview in the settings editor: !\nI looked into this some more and have the following findings block rendering is broken for Dark Modern, Light Modern, Dark HC and Light HC What changed: we a new theme token, , adopted it in and , and updated the values for to look good with a different background color But we didn't adopt token for blocks in Markdown previews, which is why is broken for Markdown preview Some options Adopt for blocks in Markdown preview -- this amounts to a rather drastic styling change for Markdown previews ! Revert Revert the change to and drop the usage of so that Markdown preview, chat, and settings UI are in a reasonable state for all themes, which I attempted in\nAlso affects the Extensions Details page since it is a Markdown preview !\nVerification: for Light/Dark Modern and Light/Dark HC, make sure that blocks are properly rendered in Markdown preview, settings UI and chat"}
{"_id":"q-en-vscode-6a247f019b80c58fc70c94f62aab3181dbffc8abc70fdb82a47a2345bf73557a","text":"VS Code version: Code - Insiders 1.84.0-insider (, 2023-10-27T10:41:08.734Z) OS version: WindowsNT x64 10.0. Try applying these variables using the environment collection API: Even though screenshot says they're applied as expected: ! equal sign is stripped from the end when printing the value: ! <details<summaryOriginal bug report </summary<pfrom dotenv import loaddotenv; import os with open(\".env\", \"r\") as f: f.readlines() (\"TESTKEY\") (\"TESTKEY2\") (\"TESTKEY3\") (\"TESTKEY_4\") envFile: \"<placeholder\" venvPath: \"<placeholder\" venvFolders: \"<placeholder\" languageServer: \"Pylance\" testing • unittestEnabled: true experiments • optInto: [\"pythonTerminalEnvVarActivation\"] </p</details\n! Showing the environment variables have the '=' after I unset Python: Env File ! Showing the environment variables do not have '=' sign if loaded from the Python extension\nThanks for the bug report! We investigate issues in order based on priority and severity, which includes the impact it has on your ability to use the extension to do productive work, and the number of people affected. If other users come forward and leave a comment demonstrating they are seeing/reproducing the problem then we will raise this issue's priority. Thanks for understanding and the patience!\nThank you. I'm adding some more information if it can be useful in finding the problem. I cleared the python extension cache after re-setting the \"Env File\" setting, because I noticed it was loading a cached value. ! Note that when reading here in the trace log, the environment variable KEY2 has the equal sign at the end, and it's in a different color. The equal sign is also in the output of \".vscode-insiders/extensions/ms-\" As far as I can tell the parsing code at is correct, and captures all the equal signs in the key. Hence the sign appears in all of the logs in the output tab, it's just not there in the terminal and/or python if the .env file is loaded by the extension: ! Last, I don't know if it's related, I'm getting the following debug messages from the Pylance extension:\nThanks, I'm able to reproduce it. Looks like an issue with environment collection API in VS Code itself, hence transferring.\nThis makes sense as we currently split on and so those characters won't work currently when applied via shell integration: For pwsh:\nI have this issue too. I noticed if I run my backend server (which takes a secret key from the .env file) in my system terminal it behaves as expected, but if I run it in the vscode terminal it truncates the key after the equals sign in the key. This broke my backend! Please fix this. Thanks.\nI face this issue too. Waiting for fixing. Thanks As a potential workaround, you can run the command directly from the terminal after every .env file changes for each terminal window\nI am having the same issue when the project is setup for a C++ python extension with leak sanitizers. This requires to the sanitizer library and disable leak detection via . Unfortunately, in the terminal only remains which yields to a unusable terminal due to See\nI am also experiencing this issue with a Django Secret key containing an '=' sign and would appreciate a fix. I can work around it, but it's not ideal since this runs counter to the expected behavior and how the command behaves. Thank you.\nThis so frustrating. I had my code working perfectly I was testing the whole day and then suddenly it broke. I spent hours figuring it out. I managed to debug it and found that the == signs in the .env variables were stripped out. I was so confused why did it work earlier and then I realised that a new version of VS Code was installed today. Did it break my code?Not 100% sure . Anyway VS Code is awesome. But the lesson of the day : stable environment . Don't update it constantly.\nI'm facing the same issue. Here are some potential clues based on my project: it only occurs in the init constructor for a class. The class is located in a .py that is in a different folder than the .env. The error occurs when the code is invoked from an api endpoint but works fine otherwise. I tried making adjustments by setting the .env path explicitly but no luck. Please help.\nAdding my use case: Azure services often have a connection string that's used when coding some integration and said strings usually have the '=' in them, For me, the problem can be mitigated if I leave the python.envFile config value empty, but it is a bit annoying that this \"automatic\" feature is somewhat hidden\nCan confirm that I am also seeing this behavior in the exact same situation. VSCode Terminal - drops \"=\" ! Plain Terminal - returns correct !\nVerification steps: Create an extension that does: Debug the extension Activate it Open a powershell terminal Run , ensure that it prints Repeat the same in a bash terminal but run\nThis still had an issue for fish shell which should be fixed in To verify, repeat steps above but for fish.\nverified bash/fish/zsh"}
{"_id":"q-en-vscode-c368aa2852c200ad8a40766ea7bf158d43867aba479738955355bce7539fc0cd","text":"There is a strange issue in vs code when i split the window, and click on any code or syntax , it will highlight similar code/syntax but the highlig ht is persistent, it wont go away, it works fine in the first window but when it gets split to the right it doesnt work if i click anywhere else in the window, in the left side the currentCrew is highlighted now, it just wont clear. sys info Version: 1.84.2 (Universal) Commit: Date: 2023-11-09T10:52:33.687Z (1 wk ago) Electron: 25.9.2 ElectronBuildId: Chromium: 114.0.5735.289 18.15.0 V8: 11.4.183.29-electron.0 OS: Darwin x64 21.6.0 <img width=\"1680\" alt=\"Screen Shot 2023-11-22 at 9 42 52 AM\" src=\"\"\ncould you help with triaging this issue? Thanks!"}
{"_id":"q-en-vscode-59aa14e9f71d10756be9cab16b57180bb361c21b831ca2d9b17c105723343fac","text":"So that you know that you are debugging. //cc\nI'm not sure on this, personally I would find it kind of annoying to have the toolbar on all windows. Maybe it's a setting, or can be dragged between windows? (Is there reusable cross-window drag logic yet?)\nTo clarify: not visible on all windows at the same time, but visible on the window that has focus maybe? Otherwise, it will be quite hard to step through a file using mouse in floating windows?\nI think having it appear in the active window (when in floating mode) is good.\nWe solved this issue for quick open where we also want to appear it at the window that has focus. For that, there is which will always be the right container to put global UI elements and is derived from the currently active window. There is a event that can be used to reparent the UI element live.\nThanks!!!\nVerification: Start debugging, and drag an editor out to an aux window Make sure the debug toolbar appears on whichever window has focus The debug toolbar should reappear in the same place per-window as you focus it"}
{"_id":"q-en-vscode-b5d549b45084f35f08d3e6dbe947e6763da2a85359a45ea9537ca604aee81203","text":"Type: <bBug</bCC When terminal content reaches the F2 (Accessible View) buffer size, users may want to scroll up to check the previous content. However, the command does not take effect in the Accessible View buffer, making it difficult for screen reader users to check the pushed content. VS Code version: Code - Insiders 1.85.0-insider (, 2023-11-24T15:21:39.004Z) OS version: WindowsNT x64 10.0. Modes: Remote OS version: Linux x64 4.18.0-477.27.1.el88.x8664 <details<summarySystem Info</summary </details<details<summaryExtensions (105)</summaryExtensionVersion ------ aiprm-lang0.0.2 Bookmarks13.4.2 spellright3.0.118 mermaid-markdown-syntax-highlighting1.6.0 vscode-firefox-debug2.9.10 codespaces1.16.2 remotehub0.62.0 overleaf-workshop0.4.4 easy-snippet0.7.2 scheme0.2.0 better-cpp-syntax1.17.2 lilypond-snippets0.1.1 vscode-azurevirtualmachines0.6.5 vscode-edge-devtools2.1.3 jupyter-keymap1.1.2 remote-containers0.325.0 remote-ssh0.108. remote-ssh-edit0.86.0 remote-wsl0.81.8 vscode-remote-extensionpack0.24.0 azure-repos0.38.0 remote-explorer0.5. remote-repositories0.40.0 vscode-serial-monitor0.10.0 vscode-speech0.3. multi-command1.6.0 windows-terminal0.7.0 errorlens3.15.0 Bookmarks13.4.2 openscad1.2.2 zoterolatex0.4.1 mermaid-markdown-syntax-highlighting1.6.0 doxdocgen1.4.0 vscode-markdownlint0.52.0 vscode-eslint2.4.2 vscode-quick-select0.2.9 vscode-deno3.28.0 gitlens14.5.1 EditorConfig0.16.4 prettier-vscode10.1.0 shell-format7.2.5 vscode-google-translate1.4.13 codespaces1.16.2 copilot1.138.568 copilot-chat0.11. remotehub0.62.0 vscode-github-actions0.26.2 vscode-pull-request-github0.77. cslpreview0.2.2 path-autocomplete1.25.0 latex-workshop9.15.0 lilypond-syntax0.1.1 scheme0.2.0 better-cpp-syntax1.17.2 vscode-lua-format1.3.8 lilypond-formatter0.2.3 lilypond-pdf-preview0.2.8 lilypond-snippets0.1.1 vslilypond1.7.3 git-graph1.30.0 vscode-docker1.28.0 black-formatter2023.6.0 flake82023.10.0 isort2023.10.1 python2023.20.0 vscode-pylance2023.11.10 jupyter2023.11. jupyter-keymap1.1.2 jupyter-renderers1.0.17 vscode-jupyter-cell-tags0.1.8 vscode-jupyter-slideshow0.1.5 azure-repos0.38.0 cmake-tools1.16.31 cpptools1.18.5 cpptools-extension-pack1.3.0 js-debug-nightly2023.11.1317 remote-repositories0.40.0 vscode-github-issue-notebooks0.0.130 vscode-selfhost-test-provider0.3.21 vscode-serial-monitor0.10.0 vsliveshare1.0.5892 autodocstring0.6.1 pandocciter0.10.3 shiny-python0.1.5 shinyuieditor0.5.0 quarto1.107.0 r-debugger0.5.4 java1.24.0 vscode-xml0.26.1 r2.8.2 multi-command1.6.0 vscode-deepl1.0.7 lua3.7.3 latex-utilities0.4.10 cmake0.0.17 errorlens3.15.0 intellicode-api-usage-examples0.2.8 vscodeintellicode1.2.30 vscode-java-debug0.55.0 vscode-java-dependency0.23.2 vscode-java-pack0.25.15 vscode-java-test0.40.1 vscode-maven0.43.0 markdown-all-in-one3.5.1 grammarly0.24.0 (2 theme extensions excluded) </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nAlso should work for\nTo verify: Run a few commands in the terminal to open the accessible view / (mac) to scroll to the top and bottom of the terminal"}
{"_id":"q-en-vscode-70bd67df8ab1b0563849f268cf379a8ff02389add269cd3b3bef0efe86ae0401","text":"Follow up on Currently we wrap the kb in brackets: <img width=\"193\" alt=\"image\" src=\"\"But we use parentheses in other places where the keybinding is shown, eg hover over problems or terminal pane label, or other icons. <img width=\"199\" alt=\"image\" src=\"\"\nHi thank you for posting this issue. Indeed when I chose the square brackets I was not sure what type of brackets to use. In the editor toolbar menu for example, we appear to be using square brackets. However now that I am looking at the keybindings it appears to me we may be using square brackets for keybinding chords only. I'll ask in the UX channel about this. <img width=\"423\" alt=\"Screenshot 2023-11-28 at 17 26 10\" src=\"\""}
{"_id":"q-en-vscode-41a1d36383f6298cd92dc87644b2db63146cbd6e944a3f5df2d785e65d4968b9","text":"From testing Steps to Reproduce: multiple terminals multiple terminals in the terminal view to change their color to red the terminals still select, select again time run instead Bug Only a single terminal at a time is reset"}
{"_id":"q-en-vscode-b9d6fe1cfe33b55353be443a5bb17197a9d17a15cb0b22eb73f0d286b9cf392c","text":"open the log in the output view press Ctrl + I. :bug: Inline chat opens in the output view. It's not possible to enter a question. Should not open in the output view.\nTried this and it now does nothing when I press ctrl + I. This is the expected behavior now right?"}
{"_id":"q-en-vscode-91642069782ca42e5039fd0aeb2570bf8e134274c20fde5dfa1b644434c423db","text":"Does this issue occur when all extensions are disabled?: Yes Version: 1.84.2 (user setup) Commit: Date: 2023-11-09T10:51:52.184Z Electron: 25.9.2 ElectronBuildId: Chromium: 114.0.5735.289 18.15.0 V8: 11.4.183.29-electron.0 OS: WindowsNT x64 10.0. Steps to Reproduce: ! also causes to fail. Related issues: Workarround: call\nI cannot reproduce, so you just move views around a bunch until the explorer disappears? Do you have any extension?\nYes, Make Folders appear independently and move Explorer into Folders. I expect view back to Explorer include Folders instead of Explorer eventually disappearing. Explorer disappearing can cause other problems, including but not limited to fail. fail. As I said at the beginning: I can even reproduce it in a completely fresh installation of the macOS version of vscode without any extensions installed. !\nOkay so I can see that refresh explorer doesn't work in this case, that's for sure. I cannot get folders to completely disappear\nIn fact, what disappeared was the label . After the move, Its position is replaced by . The folder tree itself is not affected. As shown in the picture, the MOVIES folder contents can still be browsed, but its label(or button anyway) in activity bar is from to . The view hierarchy relationship changes from: Explorer Folders Open editors timeline outline to: Folders Open editors timeline outline The ability to browse folders itself is not affected, but those commands bound to or target to will be invalid. This is also the root cause of .\nIt's going to be confusion but \"Explorer\" is actually the container while \"Folders\" is what we call the view. So when you move Folders out of the explorer it creates its own container called \"Folders\" and this is working as intended. is a different root cause, but there's definitely a bug here around refresh and other explorer commands. Thanks for the find!\nTo verify: the folder view (the file tree structure) somewhere else separate from the explorer view container refresh explorer You should stay on the folder view and not be moved to a different view container. You should also see the refresh properly occur."}
{"_id":"q-en-vscode-74a78d040152a9825d31b9688ff8cbeb8ef1c364a457fe3895d5e9a793ab0189","text":"See\nI have not seen this before and I suspect an unhealthy CI, the log indicates a successful download to me: Code is here: I am using VS Code test module , so moving to\nalso indicates a broken stable download. Maybe this was just unlucky timing with us releasing a new version?\ncc in case this is CDN related.\nWe need to bump to 2.3.8:"}
{"_id":"q-en-vscode-ef63eb4a9749663a4fdd3d7b67d67fceb9d141a4b8b60a7ba5488fdac5868219","text":"Type: <bBug</bI use the shell: When I start VS Code not from the terminal, I see a popup with the error: I know why this issue occurs - it's because VS Code is assuming bash semantics for launching the user's shell in order to solicit environment variables. I encountered the . In that report, I reached out to the xonsh project (xonsh/xonsh) to see if the shell might be able to support bash syntax for this purpose, but after some consideration, they rejected the idea, instead asking editors to honor the shell. Here I'm making the same request to VS Code, namely that if the shell is xonsh (\"xonsh\" appears in the user's shell path), to use a syntax specialized for xonsh to extract the user's environment, namely: If direct support for xonsh can't be , would you consider adding a configuration variable such that a user of xonsh could customize VS Code to honor xonsh as the default shell? VS Code version: Code 1.85.0 (Universal) (, 2023-12-06T18:18:04.614Z) OS version: Darwin arm64 23.1.0 Modes: <details<summarySystem Info</summary </details<details<summaryExtensions (7)</summaryExtensionVersion ------ copilot1.141.0 copilot-chat0.11.0 black-formatter2023.6.0 isort2023.10.1 python2023.22.0 vscode-pylance2023.11.10 sublime-keybindings4.0.10 </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nI acknowledge that starting VSCode from the CLI does work around the issue, but even then only for that launch. If I launch VS Code from the CLI, then it has to restart (such as due to an update), it once again is unable to resolve settings and the error appears.\nI believe this is or ?\n<!-- -96bd-47a8-a0e8- --This feature request is now a candidate for our backlog. The community has 60 days to the issue. If it receives 20 upvotes we will move it to our backlog. If not, we will close it. To learn more about how we handle feature requests, please see our . Happy Coding!\nThis is a backlog candidate but still PRs are welcome. Here's where explicit xonsh support would need to be :\nthis is being in with minimal testing, could you track any comments/issues with it while I'm out in January? We mainly just need author verification\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!\nVerified by installing and starting insiders from the terminal. I didn't see the error mentioned above. would be good if you'd confirm this.\nI can confirm, with 1.86.0-insider, the error is gone. I was unable to test on the original machine, but I've tested on a new, similarly configured machine (and confirmed that the default shell is configured for xonsh, same as reported above). I'm so delighted that it's fixed and that I was able to help. Thanks for the guidance and support. /verified"}
{"_id":"q-en-vscode-728bb67cf40e616a19deb36940fef3eaf425166f2375892c6a09a568cb76a866","text":"on a line towards the bottom of the view port trigger suggest notice how the details view is always bottom aligned it should be top aligned as long as it isn't taller than the list <img width=\"1057\" alt=\"image\" src=\"\""}
{"_id":"q-en-vscode-c0e57d9d95f95f0101a5d9627b7927765c782dbe1916722928a5b6bb3dac9c3a","text":"Type: <bBug</bCC In inline Copilot chat, \"Regenerate\" button is not readable by some screen readers, such as JAWS. This is because the button is labeled with title attribute. Please add aria-label so that JAWS can recognize the label. For better communication, I have recorded a gif: ! VS Code version: Code - Insiders 1.86.0-insider (, 2023-12-20T05:36:57.756Z) OS version: WindowsNT x64 10.0. Modes: <details<summarySystem Info</summary </details<details<summaryExtensions (116)</summaryExtensionVersion ------ vscode-openapi4.19.2 android-dev-ext1.3.2 aiprm-lang0.0.2 Bookmarks13.4.2 openscad1.2.2 spellright3.0.118 zoterolatex0.4.1 mermaid-markdown-syntax-highlighting1.6.0 doxdocgen1.4.0 vscode-markdownlint0.53.0 vscode-eslint2.4.2 vscode-quick-select0.2.9 vscode-deno3.30.0 gitlens14.6.1 EditorConfig0.16.4 prettier-vscode10.1.0 vscode-firefox-debug2.9.10 shell-format7.2.5 vscode-google-translate1.4.13 codespaces1.16.5 copilot1.143.601 copilot-chat0.12. remotehub0.62.0 vscode-github-actions0.26.2 vscode-pull-request-github0.79. cloudcode2.2.1 overleaf-workshop0.6.1 cslpreview0.2.2 easy-snippet0.7.5 path-autocomplete1.25.0 latex-workshop9.16.1 lilypond-syntax0.1.1 scheme0.2.0 better-cpp-syntax1.17.2 language-julia1.65.2 google-search0.0.1 vscode-lua-format1.3.8 lilypond-formatter0.2.3 lilypond-pdf-preview0.2.8 lilypond-snippets0.1.1 vslilypond1.7.3 zotero0.1.11 git-graph1.30.0 azure-dev0.8.0 vscode-azureappservice0.25.1 vscode-azurecontainerapps0.6.1 vscode-azurefunctions1.13.1 vscode-azureresourcegroups0.8.3 vscode-azurestaticwebapps0.12.2 vscode-azurestorage0.15.3 vscode-azurevirtualmachines0.6.5 vscode-cosmosdb0.20.0 vscode-docker1.28.0 vscode-edge-devtools2.1.3 black-formatter2023.7. flake82023.10.0 isort2023.11. python2023.23. vscode-pylance2023.12.101 jupyter2024.1. jupyter-keymap1.1.2 jupyter-renderers1.0.17 vscode-jupyter-cell-tags0.1.8 vscode-jupyter-slideshow0.1.5 remote-containers0.329.0 remote-ssh0.108. remote-ssh-edit0.86.0 remote-wsl0.81.8 vscode-remote-extensionpack0.25.0 azure-account0.11.6 azure-repos0.38.0 cmake-tools1.16.32 cpptools1.19.1 cpptools-extension-pack1.3.0 js-debug-nightly2023.12.1817 powershell2023.11.1 remote-explorer0.5. remote-repositories0.40.0 remote-server1.6. vscode-github-issue-notebooks0.0.130 vscode-node-azure-pack1.2.0 vscode-selfhost-test-provider0.3.22 vscode-serial-monitor0.11.0 vscode-speech0.3. vsliveshare1.0.5900 windows-ai-studio0.2.1 autodocstring0.6.1 pandocciter0.10.3 shiny-python0.1.5 shinyuieditor0.5.0 quarto1.109.0 r-debugger0.5.4 java1.25.1 vscode-xml0.26.1 vscode-yaml1.14.0 r2.8.2 multi-command1.6.0 AudioQ0.0.2 vscode-deepl1.0.10 abc-music0.4.0 lua3.7.3 latex-utilities0.4.10 cmake0.0.17 vscode-terminal-here0.2.4 windows-terminal0.7.0 errorlens3.16.0 intellicode-api-usage-examples0.2.8 vscodeintellicode1.2.30 vscode-arduino0.6. vscode-java-debug0.55.0 vscode-java-dependency0.23.3 vscode-java-pack0.25.15 vscode-java-test0.40.1 vscode-maven0.43.0 markdown-all-in-one3.5.1 grammarly0.25.0 (1 theme extensions excluded) </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nThank you for the issue JooYoung. I cannot access the gif because it seems to be a private link."}
{"_id":"q-en-vscode-b24a1cd7ee671dabe9c50e6761320f9ba5209a96a3036424f9191d75072abedd","text":"Tree Sticky Scroll uses caching to make sure that it does not have to rerender often. During the animation each frame/scroll has to be rerendered, which causes DOM elements to be removed and reconstructed. This could be improved by reusing the DOM elements and just adapting their heights."}
{"_id":"q-en-vscode-4136713e326f21e8e947048dea197ecf4c59bc48fcb10e6a001cf77ce7e7f3f0","text":"Type: <bBug</bWhen I commit from the git extension, I frequently get an error message that \"Git: Extention '' CANNOT use API proposal: telemetry\". When I look at \"Running Extensions\", git reports 32 uncaught errors. I do not know if the two error counts are related. This only recently started happening (last couple of weeks). VS Code version: Code 1.85.1 (, 2023-12-13T09:49:37.021Z) OS version: WindowsNT x64 10.0. Modes: Remote OS version: Linux x64 5.15.90.1-microsoft-standard-WSL2 <details<summarySystem Info</summary </details<details<summaryExtensions (13)</summaryExtensionVersion ------ remote-containers0.327.0 remote-wsl0.81.8 rewrap1.16.3 markdown-table-prettify3.6.0 vscode-markdownlint0.53.0 git-graph1.30.0 python2023.22.1 vscode-pylance2023.12.1 cmake-tools1.16.32 cpptools1.18.5 rewrap1.16.3 cmake0.0.17 markdown-all-in-one3.6.1 </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\n/gifPlease\nThanks for reporting this issue! Unfortunately, it's hard for us to understand what issue you're seeing. Please help us out by providing a screen recording showing exactly what isn't working as expected. While we can work with most standard formats, files are preferred as they are displayed inline on GitHub. You may find helpful as a browser-based gif recording tool. If the issue depends on keyboard input, you can help us by enabling screencast mode for the recording ( in the command palette). Lastly, please attach this file via the GitHub web interface as emailed responses will strip files out from the issue. Happy coding!\nThis issue has been closed automatically because it needs more information and has not had recent activity. See also our guidelines. Happy Coding!\nHere's an example of the ruling.\nThis bug is still active. I can't get rid of the problem. My version of VSCode is: Versión: 1.85.2 (user setup) Confirmar: Fecha: 2024-01-18T06:40:10.514Z Electron: 25.9.7 ElectronBuildId: Chromium: 114.0.5735.289 18.15.0 V8: 11.4.183.29-electron.0 SO: Windows_NT x64 10.0.\nare you able to reproduce this issue every time you try to commit? When the error occurs, could you open the developer console using the \"Developer: Toggle Developer Tools\" command and share all the errors/warnings in there? Are you able to reproduce this issue with all the extensions disabled? Thanks!\nYes, every time I commit I have this problem. ! ! I only have 3 extensions, I disable them and the same problem occurs. ! Sorry for my English. I don't know if you mean this. But it's what I see when I open Toogle developer tools.\nstill getting it too. I have dozens of git repositories across multiple versions of git (development containers with different Linux versions, Windows, etc.) Happens frequently, but not always, but does seem to happen consistently on some repositories and not others.\nthank you! After you reproduce the problem, could you also please send me the contents of ? Also, are you able to reproduce this issue with all extensions disabled? The easiest way to try this is to use the \"Help: Troubleshoot Issue...\" command from the command palette.\nIt seems that way. Thank you\nOk. First i delete output of git. ! After. I disable all extensions. ! After. I can one time more to make commit. ! New commit ! After commit ! Git output complet: 2024-02-01 10:24:20.195 [info] No remotes found in the git config file. 2024-02-01 10:24:20.244 [info] git config --get commit.template [54ms] 2024-02-01 10:24:20.244 [info] git for-each-ref --format %(refname) %(objectname) %(objectname) refs/tags [49ms] 2024-02-01 10:24:20.286 [info] git status -z -uall [39ms] 2024-02-01 10:26:19.989 [info] git add -A -- . [56ms] 2024-02-01 10:26:20.264 [info] git -c user.useConfigOnly=true commit --quiet --allow-empty-message --file - [269ms] 2024-02-01 10:26:20.308 [info] git config --get commit.template [39ms] 2024-02-01 10:26:20.317 [info] No remotes found in the git config file. 2024-02-01 10:26:20.364 [info] git config --get commit.template [49ms] 2024-02-01 10:26:20.372 [info] git for-each-ref --format %(refname) %(objectname) %(objectname) refs/tags [49ms] 2024-02-01 10:26:20.413 [info] git status -z -uall [38ms] 2024-02-01 10:26:20.480 [info] No remotes found in the git config file. 2024-02-01 10:26:20.528 [info] git config --get commit.template [51ms] 2024-02-01 10:26:20.533 [info] git for-each-ref --format %(refname) %(objectname) %(*objectname) refs/tags [54ms] 2024-02-01 10:26:20.579 [info] git status -z -uall [40ms] 2024-02-01 10:26:41.809 [info] git ls-files --stage -- C:UserssriccOneDrive - ICCADocumentosAngularSergieCodecargaPaginasAppsrcappcontacto [49ms] 2024-02-01 10:26:41.859 [info] git cat-file -s [43ms] 2024-02-01 10:26:41.908 [info] git show --textconv [42ms] I hope it's everything you've asked for.\nSOLVED For me, I found out it was because I was on a detached head on Git. For some reason, reattaching to master solved it.\nThank you! I have to admit that it's not clear to me what is happening here. Let me chat with my colleagues and see if I can come up with an action plan.\n+1 for gilad905's suggestion. I was able to easily repro the problem: new repo from command line. first commit. out main and commit from git source-control panel in vscode, works. out the new head (the second commit), detached. another change and try to commit it in vs code Expected: It works correctly Actual: The commit is created in the repository, but the \"telemetry\" error message shown above appears.\nThanks Fix for this will be in Monday's VS Code Insiders release. Apologies for the inconvenience!\nthanks\nI am not, and have not been using repos. Sorry this is not the only problem occuring.\nplease share your repro steps in case you are able to reproduce with the latest VS Code Insiders release. Thanks!"}
{"_id":"q-en-vscode-53aa6e053cf4f585569eaa73b6a331a2bc5764f6b212eed69843416e491f21e2","text":"viewing a very long list in the variables viewer will not show all entries. debug variables view handles this with nested groups: <img width=\"231\" alt=\"image\" src=\"\"\nverification steps: the notebook variable view with the cell the variable view in the debug sideview check that the children of the variable are accurate"}
{"_id":"q-en-vscode-44c964f114e58f1f923426358289d0b4aef18e008189909a9ae8dd6643258d69","text":"VS Code Version: 1.85.2 OS Version: Linux (OpenSUSE Tumbleweed) as well as Windows 10 If I create an automatic folderOpen task in a folder that I access through VS Code's Remote SSH extension, this task is sometimes run when I open my folder, but it often isn't. It's simple for me to reproduce this on both Windows and Linux, and also with the latest (as of today) Insiders build on Windows. It seems timing related. It also seems like it's less likely to happen if I keep a file open in the workspace when I close it, so that the same file is automatically opened when I open it the next time. I've trusted the workspace. I have put together a minimal containerized SSH server with a tiny workspace to reproduce this, see I start the container and then connect to it with VS Code over SSH. Please note that the main issue isn't with this container setup. I'm seeing this when connecting to normal remote SSH servers as well. The container is just an attempt to build a minimal reproducible setup, as well as to avoid publishing private details in this report. I have also some extra logs to VS Code's source code to try to debug this, see this commit: I unfortunately wasn't able to get the remote ssh extension working properly with my custom VS Code OSS build. From what I can tell, it seems that Microsoft actually doesn't want me to do that. But I was able to find the corresponding places in the minified file for my proper VS Code installation, and put my extra logs there instead. With those changes, the logs from a typical bad run (on Windows) look like this: The logs from a typical good run (on Windows) look like this: The main difference between these two runs is in what's assigned to in (), when () tries to fetch the folderOpen tasks (). In the good case, I get a config object containing my task. In the bad case, there is no such config object. This is also reflected in the log I to (), where the array is empty in the bad case. Please note that in the bad case, there is an update of the tasks fetched in the task service happening soon after VS Code tried to run the automatic tasks. You can see that is called with runSource=3 (i.e. ), and that the subsequent variable assignment actually contains my task. Unfortunately, the code to run the automatic tasks has already been triggered by this time, and isn't retriggered. In the good case, there is no such update of the fetched tasks based on reconfiguration. But the fetched tasks are also updated when the automatic task is run, which is why you can see similar traces but with . Steps to Reproduce: a container like the one in gist your .ssh/config file to access the containerized SSH server (optional) Patch your VS Code build (the minified if you can't use remote SSH with a custom build) to get more logs, like to the remote SSH workspace and trust it a couple of times and note that the automatic task isn't always started\nthanks for all of the info. Not sure when I'll get to this, feel free to submit a PR if you can find the fix.\nMany thanks for acknowledging this! I might try to make a fix, but it's a bit complicated to test it as it seems I'm not allowed to use the Remote SSH extension together with my custom VS Code OSS build. Then I need to continue hacking the minified workbench js file of my normal VS Code installation instead, like I did to add the extra logs. Is there any way around that, that you are allowed to share?\nwould know more about selfhosting the remote setup.\nI learned that it will indeed be hard to self host this/test it. I'll probably need to make a custom build to do so. Thanks for the investigation\nI tried something out here: It seems to work fine when I add the corresponding code to the minified file as well. Waiting for task reconfiguration for some extra time if the automatic task count is zero isn't the prettiest solution, of course, but please have a look if you have the time.\nThank you, your changes look good. I pulled most of them over and am creating a custom test build for verification.\nNice! You can probably skip the final two if statements in tryRunTasks, then, as you kept them in runWithPermission instead.\nwas able to test and verify your changes fix the issue! I have a branch ATM with the changes ready to go. Do you want to copy them over and create the PR so you get credit?\nGreat news! Sure, I could do that. Credit is not super important to me, but it's always good to know whom to blame... :⁠-⁠) I can give it a go later today. It was a while since I did that on GitHub, but I'll probably manage.\nAlso no pressure , just wanted to offer. I can instead create the PR myself and give you credit in our release note thank yous section. Just LMK which you'd prefer.\nI'll give it a go. It's a good exercise. Do you have any guidelines for commit messages and such?\nI think I managed to do it correctly! :smile:"}
{"_id":"q-en-vscode-3f079db05c50ab082c87c2d4d491e6a7e03e81ee6c374103991b06b599ca8fd5","text":"Type: <bBug</bAs I typed into the search input, nothing fancy: See screenshot ! VS Code version: Code - Insiders 1.87.0-insider (, 2024-02-01T09:55:21.613Z) OS version: Darwin x64 23.3.0 Modes: <details<summarySystem Info</summary </details<details<summaryExtensions (60)</summaryExtensionVersion ------ html-snippets0.2.1 tsl-problem-matcher0.6.2 docs-view0.1.0 emojisense0.10.0 github-markdown-preview0.3.0 markdown-checkbox0.4.0 markdown-emoji0.3.0 markdown-footnotes0.1.1 markdown-mermaid1.21.0 markdown-preview-github-styles2.0.3 markdown-yaml-preamble0.1.0 mermaid-markdown-syntax-highlighting1.6.0 esbuild-problem-matchers0.0.3 kusto0.4.4 prettier-vscode10.1.0 auto-close-tag0.5.15 auto-rename-tag0.1.10 codespaces1.16.9 copilot1.156.686 copilot-chat0.13. remotehub0.63. vscode-pull-request-github0.81. cortex-debug1.12.1 debug-tracker-vscode0.0.15 memory-view0.0.25 peripheral-viewer1.4.6 rtos-views0.0.7 vscode-azureresourcegroups0.8.3 vscode-dotnet-runtime2.0.1 debugpy2024.0.0 python2023.25. vscode-pylance2024.1.103 jupyter2024.1. jupyter-keymap1.1.2 jupyter-renderers1.0.17 remote-containers0.339.0 remote-ssh0.108. remote-ssh-edit0.86.0 remote-wsl0.86.0 anycode0.0.72 azure-account0.11.6 azure-repos0.38.0 remote-repositories0.40.0 remote-server1.6. vscode-github-issue-notebooks0.0.130 vscode-markdown-notebook0.0.26 vscode-selfhost-test-provider0.3.25 vscode-speech0.5. vscode-typescript-next5.4. prisma5.9.1 vscode-xml0.26.1 vscode-yaml1.14.0 kuskus-kusto-syntax-highlighting1.2.0 svg-preview2.8.3 rest-book6.2.1 pdf1.2.2 vscode-mdx1.7.0 volar1.8.27 vscode-typescript-vue-plugin1.8.27 markdown-all-in-one3.6.2 </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nI believe owns this area, assigning it to her.\nHmm, that usually comes with a faulty install. If you keep getting this, can you try re-installing?\nThis issue has been closed automatically because it needs more information and has not had recent activity. See also our guidelines. Happy Coding!\nThe STR needs a GH remote repo. Still broken for me after updating/re-instaling."}
{"_id":"q-en-vscode-f8343fd3d3d0e87f8d8566a558be3d7fc1579f6da4fe479bedaed80d2c403747","text":"open normal quick pick () type notice you want symbol search and select the cursor is random, within the word, and doesn't select the previous input anymore"}
{"_id":"q-en-vscode-5b36c18150e31017fbd21ad4935ab1b2363784e62797a3b0a4df8279fc8b3901","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.86.1 OS Version: Mac Sonoma 14.3 Steps to Reproduce: Note: the following is just one way to easily reproduce, but this really is an issue with the CSS variable that affects more than just the Jupyter extension (hence why the issue is opened in this repository). the Jupyter extension a cell, execute code that in the cell output, the green color is using the theme variable <img width=\"263\" alt=\"image\" src=\"\"the following in the terminal that the green ANSI styling is different in light themes: <img width=\"280\" alt=\"image\" src=\"\"<img width=\"427\" alt=\"image\" src=\"\"<img width=\"419\" alt=\"image\" src=\"\"Works as expected in dark themes: <img width=\"377\" alt=\"image\" src=\"\"<img width=\"290\" alt=\"image\" src=\"\"This is a problem because in light theme, the contrast ratio requirement is not met. This means that when the CSS variable is used there will be an accessibility bug. Thanks!\nThanks for creating this issue! It looks like you may be using an old version of VS Code, the latest stable release is 1.86.1. Please try upgrading to the latest version and checking whether this issue remains. Happy Coding!\nConfirmed reproing on 1.86.1 as well\nGithub Tags: #BM-VisualStudioCodeClient-Win32-Jan2024; #Visual Studio Code Client; #A11yMAS; #WCAG1.4.3; #AILimited; #A11yTCS; #ColorContrast; #A11ySev2;\nthe terminal has the minimum contrast ratio feature that fixes the contrast ratio, anywhere else that piggy backs on that color might not meet contrast.\nAssigned to April as it comes from s360\nso no matter what color we use, terminal will automatically adjust it? I can change the default color used to solve the contrast issue outside of terminal. Just not sure how to keep them aligned.\nyes the terminal will by default slide the color to be 4.5:1 (). It's fine if their luminosity is a little inconsistent imo, we just need to make sure the color meets contrast elsewhere.\nTweaked colors for light and light hc themes in !\nthanks a lot. I think we can close this bug as fixed now, right?\nBuild failing, will auto close when it merges"}
{"_id":"q-en-vscode-8033564c8060e963ad3c3f82c8a3f545d01c6076b5d02d42938e6cad72561d30","text":"Steps to Reproduce: a notebook and put in a cell and run it to establish a kernel for that notebook vscode and run the cell before the kernel is auto-selected :bug: notebook variable view will not initialize\nHI is this a candidate? This issue is assigned to the recovery 2 milestone, but it does not have a candidate label and there appears to only be one PR to main. We will discuss this issue in standup and decide if this should be merged to the release branch.\nAs mentioned in a private discussion, if we accept this candidate recovery release will be held on Monday which is when the next endgame starts. To avoid overlap of the two endgames, this fix should be included instead in the next release."}
{"_id":"q-en-vscode-866d816e9a65a3368bbb3f4f53292de95a1fba48de4abf201a60485b93ee8b46","text":"Type: <bBug</b! I think the log files are listed in the order the LogOutputChannels got created. Unless this is intentional / important I suggest we sort them alphabetically. I also propose adding the contributing extension's ID as the description element of the list item for LogOutputChannels created by extensions. VS Code version: Code - Insiders 1.87.0-insider (, 2024-02-13T05:47:27.450Z) OS version: Windows_NT x64 10.0. Modes: <!-- generated by issue reporter --\nChanged it to show in the same way like in the output channel dropdown !\nThis is a little confusing because a provider that shows in the Output view dropdown might not show in \"Developer: Open Log File...\". I assume this is because Output has all output channels and \"Open Log File\" only has log output channels. I specifically noticed this with ESLint.\nThats right. Not all output channels are logs."}
{"_id":"q-en-vscode-42a88106c0cf7196faef67422d4a529caef327c85ddb8980028f7af9e634fa73","text":"With the custom hover now making appearance everywhere in the workbench UI, I think we should try to make it a bit more compact. Previously I had reported as but I think is still too large and there is a lot of empty space as well. I suggest to find the sweet spot between native hover size and custom hover size. <img width=\"213\" alt=\"image\" src=\"\""}
{"_id":"q-en-vscode-03483ca8f53af53bfe132189f1397bd7a519e20b737f4bc77c2b0284e1cc01f4","text":"Testing <img alt=\"image\" src=\"\"For this, I put an invalid , and as expected, the field in the issue reporter is not set to a specific extension. However, I'd appreciate a warning log in the debug console that warns devs that they have an invalid"}
{"_id":"q-en-vscode-1f1e45108f39c95defa06bdb4acf7ef2a00ec205b5195361b825073ac7a08b02","text":"Testing I had the arg For and clicking \"Create on GitHub\" did nothing. I would expect an error to log in the console for this so that devs know what has happened.\nDid this log anything in the console? (ie, or something like that)\nI didn't see anything in the debug console.\nTechnically, it's a valid URI, but it doesn't resolve to a valid endpoint."}
{"_id":"q-en-vscode-858a2cf157fda1b98e8529de703ed833b4fb6c234b2944179d73d7732f8fc427","text":"Testing I'm not sure what the correct fix is but if the text was generated outside of the view, I would image the editor would scroll horizontally as text gets inserted. (same as typing in the editor). !\nYes, very good catch."}
{"_id":"q-en-vscode-6e09a56c2b7ba249c55d6cfca2b5a300a273800dde9935ae13383541f63ad7ba","text":"This issue is about allowing extensions to use language model that are contributed by other extensions. The API proposal is and its scope is to allow to use the chat feature of language models. Remaining finalisation todos [x] consolidate different LanguageModelChatXYZMessage-type into one type and a role-enum [x] move System message/role into its own proposal [x] have a select-function that's the only way of getting to a chat model [x] [ ] consider to have more sturcture on , e.g have a fake reply message with role like others (ollama, open ai, anthropic) I renamed to and I will continue to explore a stream of structured objects. That would be additive to and plugged in as its foundation. add something like add a StopReason-enum that's also used in LanguageModelChat\nDocs:\nWith there will be no more language model access object. Instead all requests are made directly via\nHi Is there any plans to add a , so I can implement my own calls to other models other than the Copilot provided models? Thanks!\nYeah, that exists. It is the which is still very rough and currently not yet scheduled for finalisation. Tho, feel free to give it a spin and report what's missing/wrong. For instance, locally I have a dummy extension that uses ollama to drive this API\nI assume you are referring to the API? So if I use , I can then use the id I give it with ?\nNo, I mean the proposal that I have mentioned\nFYI that I made the following changes to the proposal there is no more and generally no more referring of language model by string-identifier. there is that takes a selector and returns matching chat models or nothing () These changes allow for a better implementation of the providing-end (-proposal) and most importantly make it more obvious that models can (and will) change. The following pattern had become popular'ish , ....)`. This is brittle because extensions break as soon as the model id isn't known anymore. Instead models need to be selected, making it hopefully more clear that a desired model doesn't exist. The new selector-mechanism allows for new flexibility as well. A couple of samples To restate, the -function can return which means no model was found. That's because the selector is to strict (using ) or because no models are known (missing extensions). The shape of language model selectors is this Note that property order represents how specific it is properties are matched based on equality, when present it must match we will likely add some comparison selector for input and/or output tokens\nIsn't this still possible? We can call any LLM API in our handler and use the stream.markdown() to display the response, right?\n'y.provider.provideLanguageModelResponse is not a function'\nI'm noticing a difference in behavior of the API when running in Codespaces from the browser. When running against a Codespace with desktop Insiders, everything works as expected. When running against the same Codespace with browser Insiders, always returns , implying there are no chat models available.\nMake sure that you are on latest pre-release of the copilot-chat extension That's certainly possible when copilot-chat isn't installed in the codespace.\nIt is installed, and regular Copilot Chat works normally, but our extension can't get a language model.\nAnd it is latest prerelease?\nYep! Version: 1.90.0-insider Commit: Date: 2024-05-13T12:40:35.552Z Browser: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.0.0 Safari/537.36 Edg/124.0.0.0 ! ! Here's regular chat working: ! And our extension emitting to its output log, via this line of code: ! The same codespace but using desktop frontend instead of browser (working as expected): !\nReally confusing... In my sample remote-setup (which isn't codespaces) this is working just fine and I see so smoking guns for why this wouldn't work in codespaces. For now, I have some logging that should help us get to the bottom of this\nI see mention of moving the System prompt to a separate Proposal, can that be linked to? I am not finding it and we use it in our extension. Also is this section of the guide still accurate? It's not showing those commands as imports anymore and from my reading of the types it looks like it should be closer to:\nThis is place of all API proposals: Likely outdated\nWhere should I look for the additional logs?\nEnable trace logging, show the window-log and check for messages starting with . There should be some for registering and some for selection <img width=\"963\" alt=\"Screenshot 2024-05-15 at 08 15 30\" src=\"\"\nI did find something after running into a similar issues. Can you confirm that a later/second invocation yields models and/or that the -event fires and that after that chat models are selected?\nI found some log lines for selection but nothing for registration: We end up doing language model selection several times and each time it returns . I even tried doing the following and still there were no log lines about registration: all 3 extensions (Copilot Chat, Copilot, and our extension) browser window trace window logging extensions a query with regular Copilot (success) a query with our extension (failure) However, when using VSCode Desktop as the frontend, I do see log lines about language model registration:\nThanks for the update. That would really mean the language model API doesn't register. Can you check the chat log (Output GH Copilot Chat) for messages starting with\nI don't see any messages like that from either browser or desktop frontend on the Codespace. However, on the browser session there was this message which did not appear on the desktop session--might be spurious but I'm not sure:\nWe have finalized the Language Model and Chat Participant API Thus, you can now publish your extensions to the VS Marketplace. This API is finalized in VS Code Insiders, and will be finalized in VS Code Stable in July. Thus we suggest that you use an engines: ^1.90.0 property in your , and VS Code Stable will gracefully handle your extensions and ignore your usage of Language Model and Chat API until the API gets finalized in Stable in July. Full docs We have decided to only finalise the message. And message might be finalised in one of the future iterations. Main reason for this is general API consideration - as we are still learning how the prompt API would fit with some of the non-GPT models. Also we believe that the prompt API requires more responsible AI testing from our team. If your extension would like to use the prompt API - please let us know if you hit issues with using the prompt as a replacement. We are excited to see what you build\nI am still wondering if there will be a API so I can use other models inside of the chat API. Is this something coming in the future? For now, it can be shimmed by wrapping and handling other entries (perhaps I want to point to my local Ollama instance).\nStill a proposal and not yet planned for finalisation"}
{"_id":"q-en-vscode-c942b89f96c5eb6983028c2a8416d921a79fd7902a6596daa5946ac41eeee3b1","text":"Fyi\nVerification: install the pre-release version of speech extension: make sure is configured to (it should be by default) switch VS Code display language to another language that is supported by the speech extension such as german or portuguese start voice dictation in a text editor via \"Start Dictation in Editor\" verify you get prompted to install the related language pack for the display language you picked"}
{"_id":"q-en-vscode-eff9e01af36f1bb7a74ae55cd8976b3b06eca7bc3fb14d42045f5c2ff6837c84","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --As introduced in version 1.87.0 () drag&drop tabs now features a border indicator. It would be nice to have an option to disable this indicator. (Indicator also makes tabs width change and subsequent unwanted jump effect but that's perhaps a bug issue.) !\nIf you are not interested in seeing the indicator you can change it's colour to be transparent the following way: Regarding the jumping of the tabs, this is definitely a bug, however, I am not able to reproduce it so it seems like it might be related to some none default setting. Could you share the none default tab related settings you are using?\nOf course, I believe jumping of the tabs is related to:\nI've tried this but I do not see any jumping ! Possibly related to zooming, what are you using?\nI've debugged my a bit and it seems combination of these settings is the culprit:\nI have tried the combination of these two settings and it doesn't lead to any jumping in my case.\nThat's odd. I've tried my installation only with these two settings active (1.87.1 on latest macOS) and I can reproduce the bug. !\nI tested it on windows, will try it out on a mac next week.\nThis issue has been closed automatically because it needs more information and has not had recent activity. See also our guidelines. Happy Coding!\nI tried the drag and drop on a macos and was also not able to reproduce your bug. Will leave the issue closes for the moment.\nAre you sure you've tried the settings above? I was able to reproduce the bug on different MacBook as well as on Windows machine.\nI'm able to reproduce now. Looking into it.\nIn this specific scenario (actions left and shrink tab) we set an ::after element of width 5 for the shadow effect, however, we also use the ::before and ::after elements for drag and drop. Will try to fix next week. Could be a bit tricky"}
{"_id":"q-en-vscode-350e88c3f6f30c0ad1f11e2d578315fd9091fde2cf2a043283f8c26aa5db574c","text":"Steps to Reproduce: to a folder/workspace with multiple repositories the Source Control viewlet, and in the Source Control Repositories view select one repository to the Explorer viewer back to the Source Control viewlet and in the Source Control Repositories view pick another repository <img width=\"538\" alt=\"image\" src=\"\"the Refresh button renders the input with the correct height"}
{"_id":"q-en-vscode-0e7ece20330df7379e582a65f77f7df134c1517092b410376b4720d7b473c52b","text":"Refs: [x] anyOS [x] anyOS Complexity: 4 We have at long last finalized the test coverage API. Implementing an extension that produces data for the API from scratch is challenging, and requires domain-specific knowledge, so you might find playing with the selfhost test provider to be the most effective way to test it The new types in the API include: - - - Please: Do some testing that the API and UI functions as you would expect Read through the API and verify that its documentation makes sense to you See that UI interactions work nicely as you would expect (you can use any extension that produces coverage to test this part)"}
{"_id":"q-en-vscode-ea49dc195e9b82e08c5deef5f7465433ba0c1e00dd031a933a865d0f12533008","text":"Testing The should be a plain property and not a method\nI was thinking about this guideline when I switched it to a method as it's creating a new object every call: Can switch it back, is it worth a discussion in the API sync?\nI don't think it does create a new object but that's also not the point here. Your data is never sync with an async iterable but the point is the consumer should just have which is still an async iterable (and likely implemented as lazy getter)\nI might be doing something wrong as I don't think I've used an async iterable before, but doesn't this need to return a separate iterable per call in order to support multiple extensions? Currently the is a multiplexer for multiple iterators:\nOh, I missed the invocation inside .\nMaybe a better name for a method would be or ."}
{"_id":"q-en-vscode-92f03382bc6003026fd98955be6d9b7b92b74159f31120264a39343ea2817da0","text":"an extension with the following code: on the refactoring in the bulk edit pane opens the following multi diff editor: <img width=\"1093\" alt=\"Screenshot 2024-03-27 at 17 23 56\" src=\"\"above is not correct because it looks like the file is . In fact it is deleted."}
{"_id":"q-en-vscode-31fd02ef01bf2b96b504c5a7163a1ae71732024d79568c6a08260224582380da","text":"VS Code Version: Version: 1.87.2 OS Version: macOS 13.6.4 Steps to Reproduce: in the terminal happens in in , see that IWindowOptions is empty What should happen (works in iTerm2 and macOS Terminal): in the terminal should see a report printed to the screen The window size reporting functions are useful when connecting through tunnels that do not support TTY resize event forwarding. While many SSH and Telnet clients do support this, some tunnels do not (for example, a raw socket tunnel to a vt220-supporting application) and some programs may assume that the client supports this escape sequence. The information required to determine the terminal size in character is already available and is currently used to set the TTY's size. This function is fully implemented in , but not vscode. A small proof-of-concept where I use the debugger to enable this option appears to work as expected."}
{"_id":"q-en-vscode-17431b7cba3f0e1cad9f1c10a8f1bdd9df876bdd392c887450049500b57880a9","text":"Does this issue occur when all extensions are disabled?: I dont know VS Code Version: 1.88.0-insider OS Version: Windows 10 - 10.0. Build Steps to Reproduce: Start debugging python script. It creates a new \"debugging terminal\". This should work stop debugging debugging. If you look at the debugging terminal, you will see a input (generated by VS Code) then command for debugging the selected file. path of the is somehow truncated, hence cannot find the file. debugging is however starting: you will be prompted with the message: Timed out waiting for launcher to connect. As seen on the gif below: Path 1st time: (Correct path) Path 2nd time: (truncated path) !\nI'm getting similar behaviour when using Git Bash as default profile. First time debugging the Python Debug Console opens and the application runs without issue: the subsequent times I sometimes get: ... which does in fact executes the code, but then I also sometimes get: I don't know what the call to comes from (possible the end of the directory name, but in in the last instance is not called. This inevitable gives this error. ! Which then starts a new Python Debug Console. Similar to what is above. I don't see this behaviour when using PowerShell or Cmd. IMO There's some weird truncation happening. Version: 1.88.0 (user setup) Commit: Date: 2024-04-03T13:26:18.741Z Electron: 28.2.8 ElectronBuildId: Chromium: 120.0.6099.291 18.18.2 V8: 12.0.267.19-electron.0 OS: Windows_NT x64 10.0.\nThank you for adding details! This is the exact same problem that I experience: I didn't try with another terminal (Powershell or Cmd). I will see if that can fix my problem temporary using .\nAs mentionned by This issue is not here using . However, this is still adding extra command that could be the explanation behind it: ! The first run it is only using: On the second run, it : before the command. There is still the before this line though (2nd blank line)\nI am seeing the problem when I set my default terminal profile to any of the following and run my debug launch script twice. Each shell will fail on the second launch with the same mangled command line. bash command prompt powershell Work around: revert back to vs code 1.87.\nI am having a similar problem launching Python from VS.Code with or without debugging (both cmd and wt as terminal.external.windowsExec). If the path to the Python executable contains space as in default I end up with and an error message in the launched terminal window: Yet this works fine if I pick a Python interpreter from venv or somewhere else such that there is no space in path e.g. Curiously - the internal terminal and internal console work with either path\nit seems sending the ctrl+c in a separate is not sufficient -- any suggestions?\nIf you want to execute something again we have a helper on the render process side , this conditionally sends a ^C and may wait for a shell integration prompt. That might work?\nIt looks like that does the same thing with a 100ms delay. I'll adopt the delay as well, see if that fixes it\nplease could you confirm when this will be released? It doesn't seem to have gone out in the latest April release. I've tried the insiders version and can't debug anything so it's even worse.\nThis is available on Insiders. Please open another issue with the details for not being able to debug anything; that's probably not related to this fix...\nCan any of you verify that this is fixed in the newest version of VS Code Insiders? Thanks!\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!\nThe external terminal launch is working with space in pythonInterpreter path () using both and as Windows Terminal in current: 1.89.1 and insiders 1.90.0 /verified\nRunning on version 1.90.0 I can not longer observe the truncation. The execution command looks intact even after multiple runs. ! There's an extraneous whenever you start a run but I guess that's intentional.\nI am experiencing this issue using Python Debugger: Current File with Arguments. I am using ver VSCode ver 1.9"}
{"_id":"q-en-vscode-c03ea17f48f594c50c90c37e10264bb30ad39634d0cb4d76e6310f6d9fa6087d","text":"macOS, VS Code Insiders into explorer focus via keyboard hover does not show. There is no command to trigger the custom hover to show Like the activity bar, we have to make sure that the custom hover can be accessed in the tree/list only with the keyboard.\nWould you suggest adding a keybinding to toggle the visibility of the custom hover of the html element which has focus?\nYeah, ideally it is . The same shortcut that we use in the editor for showing the hover. So it is consistent and easy for user to understand that there is one shortcut which does the right thing everywhere.\nIsn't this a duplicate of ? Btw we seem to be having this support for extension provided trees at least:\nI think it's good to have this one separate to .\nSolving this for all trees/lists is pretty hard as the tree/list does not know anything about the underlying elements being rendered. For extension trees it is much easier as they all consist of a label which the creates and controls.\nOne option could be to utilize the to get the tooltip. Currently it already provides\nAfter further discussions we came to the conclusion that we would use the DOM tree to figure out if there is any custom hovers available and trigger the hover by dispatching a mouse event on the HTMLElement. This might seem rather hacky but it ensures the consumers of the tree only have to support hover once and lets the tree handle keyboard triggers. Handling focus is still tricky and a open problem. Until this focus problem is properly fixed the treeview will keep it's own implementation which directly uses the hover service and moves the focus into the hover. The treeview implementation of this has it's own flaws but will b kept for the meantime to not cause regressions."}
{"_id":"q-en-vscode-5141f063e579bf3b2463227831c301f81d11f901541ffbaed7b30c70b4f60042","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.88.1 (Latest) + GitHub Codespaces OS Version: All? (Tested on Linux + MacOS) Steps to Reproduce: a folder or workspace (create if necessary): all the content from the above file and save it the command palette 'Tasks: Run Task' '+ Configure a Task' 'Open file' any provided task template notification error: \"Unable to write to Folder Settings because no resource is provided.\" Expected behaviour: Empty is populated with the contents of the selected task template <img width=\"898\" alt=\"Screenshot 2024-04-18 at 16 50 01\" src=\"\""}
{"_id":"q-en-vscode-fae7c06e4f037da30fc28d84d4c74b47207c5f09133ea4e9ef82b92c76629fb7","text":"Under some circumstances, the Code Action button won't go away and will still run its assign command (incorrectly) Steps to Reproduce: an error with 'Quick Fix' available ! the cursor on the same line and press to bring up the quickfix ! to cancel ! the code action button ! the button again ! again ! ! at step 4. the button should disappear, but it does not I found this bug while adding Code Actions to my personal extension VS Code Version: 1.88.1 OS Version: Windows 11"}
{"_id":"q-en-vscode-b52e2eadefe9cf78bf7bfe1e394cbe07b6f6f07cbfa239fa5f4e10124c37d1a1","text":"Testing \"Note: Using this setting will disable Window: Auto Detect Color Scheme\" should finish with a period."}
{"_id":"q-en-vscode-5296f6ee896c6b70da9904f047cba1ea820611da44c0f7be868482885b54301d","text":"Testing open show outline and have sticky scroll enabled scroll the cell with \"# Header \" into view delete the cell :bug: the sticky scroll breaks, the error below is printed <img width=\"1482\" alt=\"Screenshot 2024-04-23 at 08 51 18\" src=\"\"\nI'm unable to repro this. Please can you share a screen recording of the steps. Perhaps recoding from opehning the notebook to selecting and deleting. Thanks.\nLooking at the stacktrace tells me this is coming here which means the model had been already\nfollowing the steps above (empty profile) always reproduce the error (console) for me. The screen cheese shows in sticky scroll and/or other paces of the outline tree"}
{"_id":"q-en-vscode-3e1684aa80fead34812870e5a12106b9dbf5ab7c5ccbf19523f203b8e8880dd7","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --<!-- Describe the feature you'd like. --I'm creator of [SWC][] and I want vscode to use the officla JSON schema introduced by for by default. I'm going to create a PR. [SWC]"}
{"_id":"q-en-vscode-0940464a0c55ecf1892cd70de1b137718b9ff9326188f4ea8a3497f436435de3","text":"If I open a Dart project, right-click on a folder in the explorer and then \"Run Tests\", I see: ! However if I open the test explorer tree first and then try again, everything works. It appears that \"Run Tests\" from explorer does not trigger the to allow initial test discovery. ( I presume this is a bug and not intended?)\neven after I've shown the explorer tree and triggered discovery, it seems like \"Run Tests\" now works on the \"tests\" folder, but not on an individual test file (even though I've confirmed the URI for that file is being set when creating the test tree item). Is there something else I need to do besides having the URI on the test file node?\nDoes this reproduce on Insiders for you? It should be fixed with\nlooks like that fix handles the second issue above (running the file now works), however the original issue above is still an issue - if I don't trigger discovery of the tests before using it, it reports no tests found.\nFix works for me, thanks! :) /verified"}
{"_id":"q-en-vscode-7925301601e5099541225091a4e0be5f44ec61029e59b32c4c07ff954ad3a417","text":"The command was augmented with the argument here in Because the does not include the , any attempts to mint a key will result in a 404 if the server-base-path is set. I think this is the reason why I am experiencing an issue that any sign-in attempt in the browser (e.g., Github Copilot, Settings Sync) do not persist between page refreshes. <!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.89.1 OS Version: x64 Steps to Reproduce: the browser, observe 404s to\nThanks for creating this issue! It looks like you may be using an old version of VS Code, the latest stable release is 1.89.1. Please try upgrading to the latest version and checking whether this issue remains. Happy Coding!\nassigned to you since you approved the original PR"}
{"_id":"q-en-vscode-abbda3a63cf7b1b279e4148e054af2e4112c245c056b7712fecbb5a7d4c5bd0f","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.89.0 OS Version: Linux x64 6.6.26-1-MANJARO Steps to Reproduce: VS Code. Developer Tools and watch Network requests. VS Code, open Settings. that Settings will not display any settings until the requests to are completed. reading the results of the queries, observe that this request is related to the absolutely despicable advertisement for GitHub Copilot_, to detect if it's already installed (or something) in order to show the ad in the Settings UI and that you are now even more incensed that something which should work immediately was taking 7 seconds or longer due to the fact that we are now living through a dystopia. Then quickly get over it and start considering a move to a different editor or a fork. Please choose one of the following suggestions: remove the GitHub Copilot advertisement from the Settings UI. waiting for network requests in order to show the Settings UI. this issue and ignore this unquestionably sane plea to help battle the enfecalization of this beloved product.\nThanks for creating this issue! It looks like you may be using an old version of VS Code, the latest stable release is 1.89.1. Please try upgrading to the latest version and checking whether this issue remains. Happy Coding!\nThe AUR hasn't been updated with 1.89.1 yet but looking at the resolved issues I don't see any related to this.\nI have a 1-second timeout per network request. In other words, the delays should now have an upper limit on their duration."}
{"_id":"q-en-vscode-880deb52b7d5fee355b8f63be92090cff621869dc30ffa8d651816ff0f470bb5","text":"You should not be assuming that the editor for which the action is being run for is the active editor. Make sure to correctly extract the context from the arguments passed to the script. The run method should be defined the following way: You can get the correct context with and resolve it with Make sure to adopt this for the following MultiDiffActions: - You can test that it works correctly with and 2 editor groups"}
{"_id":"q-en-vscode-b81f911b44467a2fdda27c0595bc32be43a44d2bb7ebfbc6bc488a3428e07e8b","text":"make an inline chat request that changes code select the re-generate action :bug: the previous changes aren't undone"}
{"_id":"q-en-vscode-dae8ca5db27394c8046c3fe61c650cee2f60209874d6e5b55866ad721ac8d383","text":"My extension provides document links by calling on with a implementation. It provides tooltip, because it always sets the field to the return values of . I confirmed it works (tooltip is shown) by actually running the extension. I wrote an integration test for it. It activates the extension and calls where is the URI of an opened document that our extension supports. It returns the expected objects except those lack the field. Am I missing anything, or is it a bug of the command that it doesn't set the field? Version: 1.89.1 Commit: User Agent: Mozilla/5.0 (X11; CrOS x86_64 .0.0) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.0.0 Safari/537.36 Embedder: server-distro <!-- generated by web issue reporter --"}
{"_id":"q-en-vscode-2c920815f98fff5cf9c894c4527fc15f470d85ced1bc7534e639585b01e72557","text":"I removed this to move over to using the agent, which required setting . Chatted with and I should add another agent to support the agent in both places, only making it the default for the terminal location. fyi\nVerified that is available in panel chat."}
{"_id":"q-en-vscode-c4efb7a0aebb038b35e23024228ffd438d06c539c0be209e08ffc1c979dd9991","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.90.1 OS Version: Rocky 8 Steps to Reproduce: with the option with a string containing the desired connection token. server starts up with the specified connection token. with the option with a path to a file with the desired connection token. server starts up with a randomly-generated UUID as the connection token. The exposes a which would reasonably be expected to work, especially since does work as expected and overrides generation of a random token. In particular, in calls to handle either generating a fresh token or using a provided one. This only handles and as far as I can tell there is no code in the option handling that would replace this value with the token read from the file . The option only appears to be used in different contexts like running the downloaded servers, and the value in the ServeWebArgs version appears to only ever be overwritten by the previously-mentioned token minting code. When executing the function, it would be much preferable to provide a specified connection token in a file, for reasons similar to those in to prevent the connection token from being accessible by other users. I do not believe this is a security vulnerability, since you have to explicitly opt in to providing a connection token in a potentially unsafe manner as-is, but being able to provide one in a file would be a substantial security improvement when using specified connection tokens. I believe that this code should read in the file at the path provided by the argument and use the contents as the token when provided. If this isn't intended to be used, I would think it should be removed from the options provided for the CLI command instead of being non-functional.\nHi Are you able to verify if this fix works? You can do so by getting the latest pre-release version of the remote-ssh extension and trying these same steps there. We want to make sure we get this bug verified, and the fix is correct, before we release this to stable. Thanks!\nI checked the code-insiders build I downloaded today (1.92.0). It does not appear to have a fix for this issue and the current insiders build has the same behavior as before. I also took a look at the that closed this issue and although it is titled as relevant to this issue, it contains a that appears like it would perhaps be relevant to the remote-ssh extension but does not have any relation to this issue with connection tokens and the CLI that I can tell. I think perhaps that PR/commit got mixed up from different issue than this one.\nReopening with verification found based on authors report\nhuh, not sure what happened to the branch in that PR. I'll resubmit the change."}
{"_id":"q-en-vscode-32178141a66fc224fd8927d8a4abe870c58d473fa0a37be7b622f6952b707825","text":"See\nI think it's related to this line not clearing the quick pick: It's meant to hit backspace:"}
{"_id":"q-en-vscode-8714f22987805347bfa9b643059a91f7106a4acd8b8877fd59ae233083cc1c05","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --Add support for --proxy-bypass-list in the file so that we can provide a default proxy bypass list to VS Code users. Setting NO_PROXY doesn't work as a solution because it would either have to be set as a user / system environment variable thus impacting all applications, or we would have to rely on our VS Code users to explicitly pass --proxy-bypass-list as an argument to code on the command line each time they launch. Having the option to set this in would provide us with a way to package this in our default enterprise installation of VS Code so users would get this setting by default, but just for VS Code.\nCould we just in . I was not aware that only a subset of our command line arguments can be configured there actually. fyi\nYes we can expose it Yeah the list of chromium commandline flags in quit large and always changes with version updates it would be a nightmare if we are to support all of them in . At this time we only support those on a case by case basis.\nThanks, makes sense! Should I assign this one to you? If you are busy, I am happy to provide a PR in July if you give me a code pointer?\nHere is an example of recent flag addition , feel free to send a PR. I can review it. Thanks!\nThanks for that great example. Here's the PR I verified that adding \"proxy-bypass-list\" does not produce errors in I did not verify if this list actually gets passed to Chrome. We can ask if we get this out to an insiders build.\nFYI: There is a PR in progress for adding a setting for the feature:\nThanks. But unlike that PR, I think this flag just needs to be passed to Chromium and it should get handled. i.e. I did not add any code in VS Code that actually handles the flag.\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!\nThis flag doesn't seem to work for me unfortunately. Using mitmproxy I still see the bypassed domains: !\nCan confirm, preparing a fix now.\nthanks for the quick fix\nverification steps: Setup mitmproxy Start the application with proxy server Check the logs to see requests from the application are intercepted, you should see some domains like or Open and set Restart the application with same proxy server Confirm that requests to these hosts are not seen in the intercepted logs anymore"}
{"_id":"q-en-vscode-ff3984358f7f7b9377b96bea7840769f485325836337b4c2285211c6d690e1bc","text":"Testing enable link underline open settings editor and scroll to \"Text Editor\" section in-text links are underlines but command links aren't unsure if this should be aligned or not? <img width=\"759\" alt=\"Screenshot 2024-06-25 at 09 50 40\" src=\"\""}
{"_id":"q-en-vscode-c1f46a818f7adf4149772cdc612170bc4f92552fa6e52b05741d38af65ecbe1f","text":"Testing Once again, I'm not sure what the norms are here, but would it be useful to have the actions in the status bar show in the accessibly view? <img width=\"607\" alt=\"image\" src=\"\"\nHi yes that would make sense. Good point!"}
{"_id":"q-en-vscode-1e4c917e6b1cb72aa6a698545d5c8c97a1c5a403ce683007a9828b93d63e0579","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes/No <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.91.0 (Universal) OS Version: Darwin arm64 23.5.0 Steps to Reproduce: new profile test to new profile in Devcontainer sample container is loaded without extensions. Opening in DevContainer with the default profile works correctly. Works correctly with the previous version of vscode 1.90.2\nSame For Me\n+1 extensions from are not enabled in the container when using a custom profile. Although the container logs does report that the extension was successfully installed, and I can see it is installed in the container, in . It's just not enabled when using a custom vscode profile."}
{"_id":"q-en-vscode-b552610e9d0ffad68fb25e765032ac2945931a4d872624740aaa81c704cbc137","text":"<!-- Do Not Delete This! featurerequesttemplate --<!-- Please read our Rules of Conduct: --<!-- Please search existing issues to avoid creating duplicates. --VSCode Version 1.91.0 introduces a new \"New Window with Profile menu item\". I see it's not currently available in the command palette, like the \"New Window\" command. <img width=\"603\" alt=\"image\" src=\"\"It would be nice to have this available in the command palette."}
{"_id":"q-en-vscode-ad5c3ba8cafe8f771672f22f9a727a8b7b59fbd015e01f4cc3cd128ff9609c71","text":"Type: <bBug</bscreen ready accessiblity mode to be on. Look for screen reader optimized at the bottom bar of VS Code. settings, search for \"accessibility volume\". Set it to 0. an action in a VS Code editor that creates a sound (which you haven't configured to be \"off\"/\"never\"), eg trigger an inline copilot suggestion. a sound (make sure your computer audio is on) even though you set volume to 0. VS Code version: Code 1.91.1 (, 2024-07-09T22:07:46.768Z) OS version: Darwin arm64 23.5.0 Modes: <details<summarySystem Info</summary </details<details<summaryExtensions (71)</summaryExtensionVersion ------ better-comments3.0.2 quitcontrol-vscode4.0.0 solarized-light-custom4.0.1 solidjs3.0.7 vscode-tailwindcss0.12.5 turbo-console-log2.10.4 ruff2024.30.0 regex0.4.0 doxdocgen1.4.0 vscode-eslint3.0.10 xml2.5.1 es7-react-js-snippets3.1.1 gitlens15.2.0 prettier-vscode10.4.0 figma-vscode-extension0.3.5 auto-rename-tag0.1.10 shell-format7.2.5 kotlin0.2.34 macros1.2.1 copilot1.213.0 copilot-chat0.17. vscode-pull-request-github0.92.0 go0.41.4 gti-vscode0.6.1 vscode-graphql-syntax1.3.6 better-cpp-syntax1.27.1 vscode-go-template0.2.1 cmake-language-support-vscode0.0.9 Kotlin1.7.1 git-graph1.30.0 vscode-docker1.29.1 vscode-dotnet-runtime2.0.9 black-formatter2024.3. debugpy2024.9. isort2023.10.1 python2024.10.0 vscode-pylance2024.7.1 jupyter2024.6.0 jupyter-keymap1.1.2 jupyter-renderers1.0.18 vscode-jupyter-cell-tags0.1.9 vscode-jupyter-slideshow0.1.6 remote-containers0.375.1 remote-ssh0.112.0 remote-ssh-edit0.86.0 remote-wsl0.88.2 vscode-remote-extensionpack0.25.0 cmake-tools1.18.43 cpptools1.20.5 cpptools-extension-pack1.3.0 remote-explorer0.4.3 remote-server1.5.2 test-adapter-converter0.1.9 vsliveshare1.0.5932 vsix-viewer1.0.5 vscode-hack2.20.0 quicktype23.0.170 webgl-glsl-editor1.3.8 code-spell-checker3.0.1 tauri-vscode0.2.6 blueprint3.0.0 best-dark-themes-pack1.0.2 best-light-themes-pack1.0.2 cmake0.0.17 vscode-lldb1.10.0 vscode-swiftformat1.6.10 change-case1.0.0 luahelper0.2.28 pretty-ts-errors0.5.4 markdown-all-in-one3.6.2 material-theme3.17.2 (28 theme extensions excluded) </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\nthanks for taking a look ! i believe this will resolve as well\nSpecifically the setting doesn't get saved to Even if you write it manually in , VSCode removes it on next start. Both and have this problem."}
{"_id":"q-en-vscode-b616a0f2df97ef1e6064e6f720cf23f9108d28e4a73aa3332072f7158cd2b425","text":"Postgres supports dollar quoted strings as an alternative to regular strings. This is useful when you need to include the character itself but don't want to escape it. The official documentation can be found . Unfortunately, the default SQL highlighter in VSCode does not seem to support this syntax. Steps to Reproduce to a new file the \"SQL\" language mode the following content: Expected should not start a new string. The content between and should be identified as a string and highlighted as such. Actual ! Additional context Version: 1.91.1 Commit: User Agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36 Embedder: Also tested in the desktop version with identical results.\nNote that there may be good reason the editor tries to highlight code in the dollar quoted block. Dollar quoted strings are often used to declare function bodies in postgres and there are therefore (perhaps most commonly) instances where the content inside the dollar signs would be better off using \"regular\" sql highlighting. However, the reported issue is very annoying as it breaks highlighting for the rest of the file. If implementing a full parser with support for these types of \"sometimes strings\" requires too much effort, a passable alternative would be to at least stop the highlighter from leaking outside the dollar quoted scope.\nTh grammar we have is for MS SQL and doesn't support features outside that. Duplicate of\nIn that case, I think you should rename the user-visible language mode to \"MSSQL\" rather than \"SQL\". I find just calling it \"SQL\" to be a deceptive marketing practice if you are not going to support other SQL dialects in it.\nIt is fair to say that MS SQL won't work for all SQL flavors. I will discuss renaming it with the team.\nBased on some quick digging, MS SQL uses Transact-SQL (T-SQL). The most appropriate thing to rename the built in SQL grammar to would be T-SQL. However, I'm not sure that T-SQL will mean something to most users.\nI'm going to wait and try renaming at the very beginning of our next milestone so that I can see if any insiders users are thrown off first.\nI think T-SQL is not well known enough and will be confusing. I'll try renaming to MS SQL.\n<!-- UNABLETOLOCATECOMMITMESSAGE repo --Issue marked as unreleased but unable to locate closing commit in repo history. If this was closed in a separate repo you can add the label directly, or comment .\n<!-- UNABLETOLOCATECOMMITMESSAGE --Issue marked as unreleased but unable to locate closing commit in issue timeline. You can manually reference a commit by commenting , or directly add the label if you know this has already been releaased\nTo verify: Open a file in VS Code Open the language picker from the status bar Verify that you don't see SQL, but instead see MS SQL"}
{"_id":"q-en-vscode-521c260803cc312a2a7aaa150958ee72dee42fc063f7b5e831258ba15f370cc8","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes/No, this is an extension-related issue that arises across extensions with walkthroughs <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.91.1 OS Version: Windows 11 Extension walkthroughs are not launching upon install. Expected behavior: walkthrough launches automatically upon install. Tested on C# Dev Kit extension and Python extension. Seemed to work in 1.90.0 but does not work in 1.91.1. Steps to Reproduce [Windows Sandbox]: a Windows Sandbox Edge and download VS Code (stable) from the VS Code website the C# Dev Kit extension. Note: Upon install the walkthrough is supposed to launch automatically. This is not happening, but I can manually open the walkthrough using these steps. command palette. command 'Welcome: Open Walkthrough' 'Get Started with C# Dev Kit' ! steps 3-7 with the Python extension. Same behavior. Steps to Reproduce [VS Code stable and VS Code insider]: VS Code C# Dev Kit (if you have it installed) VS Code C# Dev Kit extension. Note: Upon install the walkthrough is supposed to launch automatically. This is not happening, but I can manually open the walkthrough using these steps. command palette. command 'Welcome: Open Walkthrough' 'Get Started with C# Dev Kit' steps 2-7 with the Python extension. Same behavior.\nclosedWith\nusing the steps above, I am not seeing the walkthrough open when I reinstall the extension\nWas just talking with about this- if you've had it installed in the past, it won't work, but for me it works in a window with a clean user-data-dir"}
{"_id":"q-en-vscode-2eb667221ee8a0d756ef2f9c56e0a325670377905b726cc7e96d0d4f37992651","text":"Connect to my mac with Remote-SSH See these errors from the server in the log (Remote-SSH output channel) doesn't seem to cause a problem but it's noisy and looks like an issue\ncould you point me at which line this is? Our stack beautifier doesn't work on server code afaik\nI don't know which line it is, I only have that stack\nIt's this line: Somehow node-pty can return undefined for Doesn't seem like it should be possible considering or are the fallbacks for :\nI'm seeing this on my own machine so could do some debugging if you want to add logs or something"}
{"_id":"q-en-vscode-bac13877f68786cbb8ee5f74b4b0009a882116bbc1f1b435b79cc2668a936512","text":"Type: <bBug</bWelcome - Learn the Fyndamentals Safely browse and edit code Workspace Trust Link 404 - page not found VS Code version: Code 1.91.1 (, 2024-07-09T22:06:49.809Z) OS version: Windows_NT x64 10.0. Modes: Restricted <!-- generated by issue reporter --\ncould you please provide a screenshot of where this broken link is? Thanks!\nGot it - issue is in the walkthrough. Fix incoming. Thanks for signaling this\nIs this a candidate? The change has been merged to the main branch but not the release branch, and so the fix is not in the latest Insiders.\nIMO, it can go with the next one.\n<!-- UNABLETOLOCATECOMMITMESSAGE repo --Issue marked as unreleased but unable to locate closing commit in repo history. If this was closed in a separate repo you can add the label directly, or comment .\n<!-- UNABLETOLOCATECOMMITMESSAGE --Issue marked as unreleased but unable to locate closing commit in issue timeline. You can manually reference a commit by commenting , or directly add the label if you know this has already been releaased"}
{"_id":"q-en-vscode-d8470ef404e48e1c6da369389a73a3cf7c24cc6c1f33063d5d4671d5d151a027","text":"Type: <bBug</bIssue Reporter, e.g. from off Command Palette. the first dropdown to VS Code version: Code - Insiders 1.92.0-insider (, 2024-07-24T07:50:18.385Z) OS version: Windows_NT x64 10.0. Modes: <!-- generated by issue reporter --\ngood catch, thanks! looks like an issue with the refactor"}
{"_id":"q-en-vscode-164a9fda5043994f762540e0b0ef9f2f39bc9966650fdba2bdf432f7f2c5c0e2","text":"Type: <bBug</bIf you rename a folder to \"Cut Sheet\" for example, that folder and anything below it becomes a \"greyed\" out folder. If you have a file that is named \"Cut \" for example, this is also greyed out. Renaming any of these makes them a normal folder/item icon. I have confirmed that this is not a .gitignore issue, nor does it appear to be a theme issue. Please see the image below. ! VS Code version: Code 1.91.1 (, 2024-07-09T22:06:49.809Z) OS version: WindowsNT x64 10.0. Modes: <details<summarySystem Info</summary </details<details<summaryExtensions (81)</summaryExtensionVersion ------ amazon-q-vscode1.17.0 vscode-intelephense-client1.10.4 phpserver3.0.2 vscode-yarn-script0.0.1 path-intellisense2.9.0 codeium1.10.1 doxdocgen1.4.0 vscode-markdownlint0.55.0 composer-php-vscode1.49. intelli-php-vscode0.12. phptools-vscode1.49. profiler-php-vscode1.49. xml2.5.1 gitlens15.2.1 vscode-html-css2.0.10 EditorConfig0.16.4 prettier-vscode10.4.0 vscode-yarn2.1.0 copilot1.218.0 copilot-chat0.17.1 remotehub0.62.0 vscode-pull-request-github0.92.0 gc-excelviewer4.2.59 todo-tree0.0.226 vscode-test-explorer2.21.1 better-cpp-syntax1.27.1 vscode-nuget-package-manager1.1.6 cmake-language-support-vscode0.0.9 vscode-autohotkey-plus-plus5.0.5 rainbow-csv3.12.0 compare-folders0.24.3 vscode-docker1.29.1 csdevkit1.8.14 csharp2.39.29 vscode-dotnet-runtime2.1.1 vscodeintellicode-csharp2.1.11 debugpy2024.8.0 gather2022.3.2 isort2023.10.1 python2024.10.0 vscode-pylance2024.7.1 jupyter2024.6.0 jupyter-keymap1.1.2 jupyter-renderers1.0.18 vscode-jupyter-cell-tags0.1.9 vscode-jupyter-slideshow0.1.6 remote-containers0.379.0 remote-ssh0.112.0 remote-ssh-edit0.86.0 remote-wsl0.88.2 cmake-tools1.18.43 cpptools1.21.3 cpptools-extension-pack1.3.0 live-server0.4.14 powershell2024.2.2 remote-explorer0.4.3 remote-repositories0.40.0 remote-server1.5.2 test-adapter-converter0.1.9 vscode-speech0.10.0 vsliveshare1.0.5936 indent-rainbow8.3.1 word-count0.1.0 material-icon-theme5.8.0 LiveServer5.7.9 vscode-vba0.11.0 logos-vscode1.0.1 pdf1.2.2 cmake0.0.17 highlight-matching-tag0.11.0 intellicode-api-usage-examples0.2.8 vscodeintellicode1.3.1 vscode-gradle3.15.0 vscode-java-debug0.58.0 vscode-java-dependency0.24.0 vscode-java-pack0.27.0 vscode-maven0.44.0 vscode-todo-highlight1.0.5 php-debug1.35.0 php-pack1.0.3 php-intellisense1.3.3 (1 theme extensions excluded) </details<details<summaryA/B Experiments</summary</details<!-- generated by issue reporter --\n/needsMoreInfo Is it reproducible with all extensions disabled? You can try this with <kbdF1</kbdand Do you have mentioned in setting for user or workspace or do you have it in ignore file?\nThanks for creating this issue! We figured it's missing some basic information or in some other way doesn't follow our . Please take the time to review these and update the issue. Happy Coding!\nIt still occurs when I reload window with extensions disabled. And I do not have mentioned in my files.exclude in my user or workspace settings JSON. My workspace does not have a a files.exclude. My user settings has the following: \"files.exclude\": { \"/.classpath\": true, \"/.project\": true, \"/.settings\": true, \"/.factorypath\": true }, See attached image showing extensions disabled and both files. ! And here's a picture of my .gitignore file: !\nSame problem here !\nPR fixing it\nFix reverted by"}
{"_id":"q-en-vscode-ba6c9a8b0db4608fd91ca247f4069b4346512b01ff3dec64bff59fed3904609a","text":"a workspace: and drop into Expected Inserts Actual Inserts\nClosing as fixed in main and release branch"}
{"_id":"q-en-vscode-a634484411be104a25aa712baf83f055e9469d88575a6c5f286b7db7b3703f2e","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --Does this issue occur when all extensions are disabled?: Yes <!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: 1.92.0, commit OS Version: Windows 10 Enterprise, 22H2 build .4651 Error message: Steps to Reproduce: a new VS code window ! a new file and drag out to make another window ! \"Create New Terminal in Editor Area\" on that new window ! error will show up ! \"Try Again\" will reload the terminal, but it doesn't accept inputs, cannot type anything there ! Creating terminal in main window and drag it to side windows does work, but I have to do that with every window reload. Yesterday I could still use terminals in side windows like normal, this only occured after updating to 1.92 July update Thanks for reading :)\n! Facing same issue\ncoming from xterm here I think !\nI have the same problem on Fedora 40 Sway. After I click , it will create a non-interoperable terminal.\nI'm also having this issue, I thought it might be a problem with WSL but it started for me after updating to VSCode 1.92.0 !\nQUICK FIX!!! Just create a new In-window terminal and drag it to the new terminal window.\nWill this be hotfixed or should we downgrade?\nsorry this won't get a patch release as there's a few workarounds. To downgrade in the meantime you can grab 1.91.1 here , you'll also want to set\nHi. Didn't do the trick for me. Other hack: Ctrl+J (new term in main window) Ctrl + Shift + P Terminal: Move Into New Window Drag & Drop into wanted window. Cumbersom many many times a day."}
{"_id":"q-en-vscode-feece0bd27aec8ee4ce198270d0061a4392b07bc56d6d65be241e709f36b12fd","text":"Note that quick pick instances and their events are disposables. Noticed a few cases of missing:"}
{"_id":"q-en-vscode-de099ed552ec84b93c7781575c06b349411a6a2fdce9906a1255bf5621a7b672","text":"Type: <bBug</bThis issue was supposed to be fixed in , but we have recently received complaints from some users regarding this. I counted the first 50 issues with Vue extension installed in , and 25 users are using 1.8.27 ~ 2.0.10 (2.0.10 was released 4 months ago ), only 19 users are using the latest version (2.0.22 ~ 2.0.28). Numbers: 2.0.10: 13 (Issue , Issue , Issue , Issue , Issue , Issue , Issue , Issue , Issue , Issue , Issue , Issue , Issue ) 2.0.8: 6 (Issue , Issue , Issue , Issue , Issue , Issue ) 2.0.7: 2 (Issue , Issue ) 2.0.6: 2 (Issue , Issue ) 1.8.27: 2 (Issue , Issue ) 2.0.22~2.0.28: 19 Since some versions in 2.0.0~2.0.10 may have serious problems, this is probably the reason why the Vue extension has continued to receive negative reviews on the Marketplace and VSCode has continued to receive bug reports about the Vue extension for several months. I found a Windows computer with 2.0.7 installed and was able to reproduce the issue of not being able to update the version. The following is the content of . (Formatted for easier reading) And I can reproduce the problem of not being able to update by overwriting the above metadata to the file of another macbook. is installed, uninstall it and then close VSCode and reopen it. 2.0.7 from https://open- and install it to VSCode, VSCode can normally prompt that Vue extension needs to be updated. and modify the metadata of the last extension to: VSCode will find that the Vue extension cannot detect newer versions. VS Code version: Code 1.92.1 (, 2024-08-07T20:16:39.455Z) OS version: Darwin arm64 23.1.0 Modes: <!-- generated by issue reporter --\ncc\nI just checked our telemetry for the most popular combinations of VS Code version and Vue extension version and here's what I see. This proves that vue version and are indeed not getting properly updated. The root cause of this is the fact the vue extension, unpublished from the Marketplace and published again the same name (thus exposing a bug in the Marketplace) as explained here We still want to help transition your users, and I will go ahead and move this issue to the VS Code repo and Sandeep will investigate next week once he is back from vacation. Thank you !\nUpdate is not detected automatically. It is detected only when user is searching for the extension.\nSorry about this inconvenience caused. Fixed it - We will now detect the update automatically.\n<!-- AUTHORVERIFICATIONREQUEST --This bug has been fixed in the latest release of ! you can help us out by commenting if things are now working as expected. If things still don't seem right, please ensure you're on version of Insiders (today's or later - you can use in the command palette to check), and leave a comment letting us know what isn't working as expected. Happy Coding!\nI just tested it and confirmed it's fixed in the latest insiders version, thanks a lot!\nAdding label based on above comment"}
{"_id":"q-en-vscode-b8a7c59b87b8929da15186ec80ea99324bc2b3f327d05dd61aaa3c36e379e229","text":"For: Passing in an empty markdown file fixes this issue currently. [x] Add the ability to go back to the last focused view instead of the welcome page.\nRemoving the button on walkthroughs if opened from the cmd. This should prevent the walkthrough from going to the welcome page. The button is now only shown when a walkthrough is invoked from the page or another walkthrough."}
{"_id":"q-en-vscode-cfbceab2242ed81fd497126ab46ede55122f3264b3e9d7ed95b8055b62ff54ac","text":"Run the following cell Stage the notebook Make changes to the HTML Run again and view the diff Problems Clicking the will display a quick pick, however selecting anything is a noop !\n/cc"}
{"_id":"q-en-vscode-34e433cdb0777e364c7dfe233444798bc5adaf81af2d4eb398cb1a13feead24b","text":"Testing I had this code: But I still got results from a workspace folder: ! Full Extension:\nI happen to want to report the same problem When there is only one item in the exclude string, the exclusion is correctly applied. When the exclude string contains more than one item, the exclusion is not applied. !\nI think that your issue is a duplicate of\nSeems that this only happens when is not defined. Strange."}
{"_id":"q-en-vscode-76ad1a0a889a451cf0cb40810e9dfbfa42a3d14a5854cd475fc62bc429767786","text":"We'd like to have a safety net if we ever break Insiders users once more to the point of them not being able to update.\nImplementation is done. Requires"}
{"_id":"q-en-vscode-ecdf493dac87282705f1543b0ebcffe548f973dac74d141059d553cef0aa1847","text":"move editor into its own window trigger find the input box is cropped, the border isn't fully visible !\nI can reproduce and I am puzzled as to why this is. Opening for help wanted for a CSS wizard to step in if possible. Steps: open a text editor in the main window open the find widget and focus it, notice how the blue outline color goes all the way around the input box right click into the editor tab and \"Copy into New Window\" do the same and notice that the outline is cropped I already see that the wrapper around the is in the main window and in the auxiliary window ! ! CSS is defined from here: and\nI can fix this, please assign me this issue if it's still open\nwe cannot assign issues to non-members of the team, but feel free to open a PR. Thanks!\nsure, thanks for letting me know."}
{"_id":"q-en-vscode-eadf9d60b7e76d46cdc0b0f09dc0f21d47e3ef359001427c9912824f66358043","text":"Enable the setting which enabled the native edit context adoption (make sure to reload the editor) Toggle word wrap on Resize the editor, so one of the lines is wrapped Type on that line Part of the line is deleted/incorrectly edited\nLooks like this could be an issue with the coordinate system chosen too."}
{"_id":"q-en-vscode-d883d3419475839b9f14394da0dffb4fc396818f9a5cc95718c113143732a75b","text":"If the diff element is disposed while the diff element is being resolved/built, then we might end up leaking the text models. Here its possible that the element was disposed just after the first line, i.e. once the promise resolves the element has been disposed. Thus leaking the text models. We have similar code in other parts and handle this correctly."}
{"_id":"q-en-vscode-94de4d64a3717014b90f1705aa5c3c6a7ffe86149a37ee58fab1026e8ab4032b","text":"<!-- Do Not Delete This! bugreporttemplate --<!-- Please read our Rules of Conduct: --<!-- Read our guide about submitting issues: --<!-- Search existing issues to avoid creating duplicates. --<!-- Test using the latest Insiders build to see if your issue has already been fixed: --<!-- Instead of creating your report here, use 'Report Issue' from the 'Help' menu in VS Code to pre-fill useful information. --<!-- Launch with to check. --<!-- If you answered No above, use 'Help: Start Extension Bisect' from Command Palette to try to identify the cause. --<!-- Issues caused by an extension need to be reported directly to the extension publisher. The 'Help Report Issue' dialog can assist with this. --- VS Code Version: Version: 1.93.0-insider (user setup) Commit: Date: 2024-09-04T13:13:15.344Z Electron: 30.4.0 ElectronBuildId: Chromium: 124.0.6367.243 20.15.1 V8: 12.4.254.20-electron.0 OS: Windows_NT x64 10.0. OS Version: Windows 11 Insiders Steps to Reproduce: in the first paragraph doesn't appear to be a link, which I was surprised about: ! I hover over it, I see it's supposed to be a link, but clicking on it does nothing: ! The other docs links at the bottom of the playground appear correctly as links and do open when I click on them: !\nThis seems related to the bookmark issue we had in the release notes. The other bookmark links in the interactive playground don't work either.\nFixed the toc links. Need another fix for the documentation link too (which use command uris)\nThe Editor Playground is not showing at all anymore in the latest insiders. When selecting the Editor Playground menu item, a black editor comes up.\nhm working from oss builds. Insiders, it seems like the error is:\nActually unrelated. Seems like a duplicate of and fixed by"}
{"_id":"q-en-vscode-dcec8d1ffa625490e57ca817d9002157a83f6940e56acfe5118d05c527f174d2","text":"When using vscode through a , if I switch profiles and then open an existing project (ctrl+r, left click), the url then updates with some url parameters (). The window loaded with that payload doesn't work correctly. For example, code search only works on the already opened files, and not the entire folder. A workaround is to remove the payload from the url. Version: 1.93.1 Commit: User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X ) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/128.0.0.0 Safari/537.36 Embedder: <!-- generated by web issue reporter --"}
{"_id":"q-en-vscode-119af8bf46bbfb258cf2c86219fe740d869b4293de5d2fa62dd3c2806a6cd549","text":"terminal I -notice double microphone icons I do not have the voice extension installed. ! fyi\nCaused by:\nlooks the most relevant\nWhy does the terminal have its own execute menu?\nIs it possible to clean up the parallel execute menu and submit actions? I'm unclear whether I can just move the voice actions to or you need a new parallel menu for terminal customizations.\nI did the first one, seems to work, but please clean up the terminal-specific stuff if possible\nCan you be more specific?\nso I do have a lot of special code for terminal in using stuff like , which maybe now is obsolete? Did something change to consolidate inline chat in terminal?\nThis isn't actionable without more info. Terminal chat still has a separate widget implementation, it extends inline chat and shares a lot of code, but must be separate as it's not a monaco content widget for example.\nYeah I did a quick attempt to remove my terminal specific code but restored it, as it still seemed to be needed.\nThat was directed at the terminal widget stuff- it has its own execute menu and execute action, and I only looked briefly but it seemed like it was doing the same thing as the base chat submit action. TerminalChatWidget uses InlineChatWidget which uses the normal ChatWidget, I'd expect that the execute toolbar can be the same regardless of whether it shows up in an editor or a terminal.\ndid all that stuff so she might be able to comment better than me.\nWhen I last investigated this, , the chat widget's submit action was dependent on an editor, which we don't have in terminal chat, and was defined in a place where the was not accessible. I can look into doing this now though as it seems the code has changed since then."}
{"_id":"q-en-vscode-79e492788456964c493c72c5c1ae579311e7ffb8110d8efa6d45e0d5dc9a693c","text":"<!-- Please search existing issues to avoid creating duplicates. --- Version: 1.94.0-insider (user setup) Commit: OS: Windows_NT x64 10.0. Jupyter Extension version (available under the Extensions sidebar): v2024.9. (pre-release) Python Extension version (available under the Extensions sidebar): v2024.15. (pre-release) First part of the video: cursor in a markdown cell, somewhere in the visible area of the screen out of the cell to double click somewhere in the visible area of the screen to highlight text (either same line or not) Cell scrolls to the top so I can't see what I'm trying to highlight (the text is highlighted as intended though) Second part of the video: cursor in markdown cell, somewhere in the NOT visible area of the screen out of the cell to double click somewhere in the visible area of the screen to highlight text Cell scrolls to the previously placed cursor area (which is great if I'm single-clicking into the cell, but in this case, it feels glitchy to me)\nI have the same issue!"}