issue_owner_repo
listlengths
2
2
issue_body
stringlengths
0
261k
issue_title
stringlengths
1
925
issue_comments_url
stringlengths
56
81
issue_comments_count
int64
0
2.5k
issue_created_at
stringlengths
20
20
issue_updated_at
stringlengths
20
20
issue_html_url
stringlengths
37
62
issue_github_id
int64
387k
2.46B
issue_number
int64
1
127k
[ "MonetDB", "MonetDB" ]
Date: 2005-10-04 05:44:51 +0200 From: Wouter Alink &lt;<alink>&gt; To: Jan Rittinger &lt;<jan.rittinger>&gt; Version: -- development Last updated: 2005-11-25 01:13:41 +0100 ## Comment 4032 Date: 2005-10-04 17:44:51 +0200 From: Wouter Alink &lt;<alink>&gt; I thought we should at least have a decent amount of bugs for coming thursday.... ;) <b>2</b>, <b>3</b>, 5.000000, 23.000000, 23.000000, 23.000000 The above is the answer, what is the query? <a><b>2</b><b>3</b></a>/(child::b|child::c), sum(<a><b>2</b><b>3</b></a>/b), sum(<a><b>2</b><b>3</b></a>/(b|c)), sum(<a><b>2</b><b>3</b></a>/(child::b|child::c)), sum(<a><b>2</b><b>3</b></a>/(self::b|child::b)) which i guess is incorrect... Note that the casting bug (the results should be denoted as integers instead of decimals) has already been reported before, there is at least a test for it on the Testweb. ## Comment 4033 Date: 2005-10-04 17:46:23 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 by the way, sum(<a><b>2</b><b>3</b></a>/(b)) also produces 23.000000 ## Comment 4034 Date: 2005-11-10 00:13:15 +0100 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 BugDay 2005-11-09, vzzzbx: ADDED TEST / SUCCESS p.s. the correct result should be <b>2</b><b>3</b>, 5, 5, 5, 5 ## Comment 4035 Date: 2005-11-10 14:08:31 +0100 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 We tracked your problem down to a bug (actually a misunderstanding) in the implementation of fn:data(). fn:data() is implicitly called due to the application of fn:sum(). Pathfinder tries to rewrite fn:data() calls into possibly more efficient implementations, which succeeded for your first example. The other examples lead to slightly different Core trees. Rewriting failed and you hit the buggy implementation of fn:data(). I've just committed a patch that makes rewriting slightly more intelligent, which "accidentally" fixes your example. Jan is currently working on the fix for the actual problem, fn:data(). ## Comment 4036 Date: 2005-11-25 01:39:47 +0100 From: @peterboncz Logged In: YES user_id=591107 Jan, was this fixed by your fn:data checkin of today? ## Comment 4037 Date: 2005-11-25 13:13:41 +0100 From: Jan Rittinger &lt;<jan.rittinger>&gt; Logged In: YES user_id=993208 yes -- at least it works with my version now :) ## Comment 4038 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1313008 at http://sourceforge.net/support/tracker.php?aid=1313008
XQuery: sum + (a|b)
https://api.github.com/repos/MonetDB/MonetDB/issues/710/comments
0
2020-11-30T08:50:10Z
2024-06-27T11:06:30Z
https://github.com/MonetDB/MonetDB/issues/710
753,282,715
710
[ "MonetDB", "MonetDB" ]
Date: 2005-10-04 05:06:55 +0200 From: Wouter Alink &lt;<alink>&gt; To: XQuery devs &lt;<bugs-xquery>&gt; Version: -- development Last updated: 2007-11-30 04:20:05 +0100 ## Comment 4025 Date: 2005-10-04 17:06:55 +0200 From: Wouter Alink &lt;<alink>&gt; The attached query (i could not produce a smaller example, it might be related to an earlier reported bug (if-for-let-bug)) returns the following error message (CVS head of pathfinder). ERROR: interpret_assignment: unknown variable 'ipik001' In the generated MIL of the XQuery the ipik001 is referenced only for destruction (ipik001 := nil), but never initialized. If this line is removed from the MIL-script, MonetDB complains about some other non-initialized variables. I guess the scopes are not handled correctly. The source-XML-doc is in my home-folder /home/alink/test-data/bug_ipik.xml I would appreciate it not being spread around too much. (we'll have a happy bug-day ;) ## Comment 4026 Date: 2005-11-10 00:05:06 +0100 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 BugDay_2005-11-09, vzzzbx: NO TEST ADDED note; i'm (at the moment) not able to reproduce the bug with a smaller document/query. when i come across a smaller example, i'll add it. If anyone knows what the problem, he/she is welcome to provide a test-script. (p.s. the query+doc provided do generate the bug but cannot be used in the testweb) ## Comment 4027 Date: 2005-11-25 00:56:04 +0100 From: @peterboncz Logged In: YES user_id=591107 I tested the query on the music.xml document as I lack your result.xml. On my document, the error did not occur anymore. could you re-check this with the latest version? ## Comment 4028 Date: 2006-11-25 04:20:02 +0100 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; ---- Original comment by: sf-robot@users.sourceforge.net ---- Logged In: YES user_id=1312539 Originator: NO This Tracker item was closed automatically by the system. It was previously set to a Pending status, and the original submitter did not respond within 365 days (the time period specified by the administrator of this Tracker). ## Comment 4029 Date: 2006-11-29 10:09:23 +0100 From: @drstmane Logged In: YES user_id=572415 Originator: NO over-ruling "sf-robot": re-set to "Pending" since adding a proper test has neither been done nor finally been discarded, yet. User (submitter) & developer (assignee), please check, again. ## Comment 4030 Date: 2007-11-30 04:20:05 +0100 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; ---- Original comment by: sf-robot@users.sourceforge.net ---- Logged In: YES user_id=1312539 Originator: NO This Tracker item was closed automatically by the system. It was previously set to a Pending status, and the original submitter did not respond within 365 days (the time period specified by the administrator of this Tracker). ## Comment 4031 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1312972 at http://sourceforge.net/support/tracker.php?aid=1312972 The original assignee of this bug does not have an account here. Reassigning to the default assignee for the component, bugs-xquery@monetdb.org. Previous assignee was nobody@users.sourceforge.net.
XQuery: unknown variable 'ipik001'
https://api.github.com/repos/MonetDB/MonetDB/issues/709/comments
0
2020-11-30T08:50:07Z
2024-06-28T13:34:11Z
https://github.com/MonetDB/MonetDB/issues/709
753,282,685
709
[ "MonetDB", "MonetDB" ]
Date: 2005-10-04 11:28:43 +0200 From: Wouter Alink &lt;<alink>&gt; To: Jan Rittinger &lt;<jan.rittinger>&gt; Version: -- development Last updated: 2005-11-14 08:10:20 +0100 ## Comment 4017 Date: 2005-10-04 11:28:43 +0200 From: Wouter Alink &lt;<alink>&gt; from what i could find on the internet, i assume that distinct-values should accept integer values, and also a sequence of mixed types. let $i := (1,1,2) return distinct-values($i) type error: no variant of function fn:distinct-values accepts the given argument type(s): integer+ ---------------------- let $i := (1,"aap",1) return distinct-values($i) type error: no variant of function fn:distinct-values accepts the given argument type(s): (string | integer)+ ## Comment 4018 Date: 2005-11-09 23:57:59 +0100 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 BugDay_2005-11-09, vzzzbx: TEST ADDED / SUCCESS ## Comment 4019 Date: 2005-11-10 11:44:51 +0100 From: Jan Rittinger &lt;<jan.rittinger>&gt; Logged In: YES user_id=993208 this was an old one -- during playing around in august 2004 the type of distinct-values got mixed up. It was string* and is with my next checkin atomic* again. Please close the bug report after my commit. ## Comment 4020 Date: 2005-11-14 18:29:33 +0100 From: @drstmane Logged In: YES user_id=572415 cannot yet close this bug: query: let $i := (1,1,2) return distinct-values($i) IMHO correctly returns: 1 2 But query: $i := (1,"aap",2.5) return distinct-values($i) results in: 1 2.500000 "aap" while I'd expect: 1 "aap" 2.500000 ... or am I wrong ... ? (cf., http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_BugDay_2005-11-09_0.9.3/distinct_integers.SF-1312716.out.00.html) ## Comment 4021 Date: 2005-11-14 19:59:30 +0100 From: @Teggy Logged In: YES user_id=642632 Hi Stefan, you are "wrong". The result of distinct-values() is considered a set and the order of items in the resulting sequence is implementation-dependent: http://www.w3.org/TR/xpath-functions/func-distinct-values --Teggy ## Comment 4022 Date: 2005-11-14 20:10:20 +0100 From: @drstmane Logged In: YES user_id=572415 Thanks, Teggy! I approved the stable output and close this bug report. Stefan. ## Comment 4023 Date: 2005-11-15 09:43:37 +0100 From: @drstmane Logged In: YES user_id=572415 Thanks, Teggy! I approved the stable output and close this bug report. Stefan. ## Comment 4024 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1312716 at http://sourceforge.net/support/tracker.php?aid=1312716
XQuery: distinct-values doesn't accept integers
https://api.github.com/repos/MonetDB/MonetDB/issues/708/comments
0
2020-11-30T08:50:04Z
2024-06-27T11:06:28Z
https://github.com/MonetDB/MonetDB/issues/708
753,282,650
708
[ "MonetDB", "MonetDB" ]
Date: 2005-10-03 08:02:57 +0200 From: @mlkersten To: @njnes Version: -- development Last updated: 2007-02-02 09:31:53 +0100 ## Comment 4008 Date: 2005-10-03 20:02:57 +0200 From: @mlkersten Problem reported below indicates we need better protection against overflowing log directories and at least ensure that database directory and log directory are in sink. The recovery process should be safe against removal of any accidental removal of directories. The long awaited for Mguardian should include a scheme to deal with overflow in this area, e.g. sent email to the dba when the space for logs become tight. --------------------------- It seems MonetDB has problem with handling of huge tables (larger than amount of memory available). I tried to load large amount of data into a table using COPY command. I loaded same data (6 millions records, 700+MBytes) few times. Loading was successful two times, on third time server crushed. After crush I could not start the server again, it output messages !WARNING: GDKlockHome: ignoring empty or invalid .gdk_lock. !FATAL: BBPinit: cannot properly process bat/BACKUP/. and exited. Removing of the database data directory did not help much - server output messages !ERROR:mvc_init: unable to create system tables !WARNING: BBPincref: range error 410 Couple minutes later it output !ERROR: BATSIGcrash: Mserver internal error (Segmentation fault), please restart . !ERROR: (One potential cause could be that your disk might be full...) Well, insufficient disk space could be a cause, before server crashed I had approximately 2GB free. But I have cleaned up 10 GB and problem persists. How can I troubleshoot and fix the server ? What are memory / disk space requirements when bulk loading data ? I tested with 32-bit system, 512 MB RAM + 1GB swap, Ubuntu Linux, MonetDB-4.9.3, nightly build dated September 30. -- Best regards, Andrei Martsinchyk ## Comment 4009 Date: 2005-10-03 20:29:05 +0200 From: @grobian Logged In: YES user_id=963970 the same problem was reported today by a few students who more or less did the same: loading a lot of data into MonetDB. ## Comment 4010 Date: 2005-10-04 13:31:01 +0200 From: @grobian Logged In: YES user_id=963970 from Andrei: It was due to insufficient disk space. After freeing up 10 GB I successfully loaded more than 2GB of data. Server crush if there is lack of disk space is an issue, but not critical one. I guess we need some better/more indicative error messaging. ## Comment 4011 Date: 2005-10-05 22:08:20 +0200 From: @njnes Logged In: YES user_id=43556 Protecting against users removing important data is impossible. The bugs we should fix are 1) don't crash on loading (to much) data, but report the problem and gracefully drop the inserts. When we startup without enough diskspace we should give more informative error messages. ## Comment 4012 Date: 2005-12-21 11:28:32 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay: stmane: NO TEST / POSTPONED: I'm not sure, whether it is feasible to provide a proper test script ... ## Comment 4013 Date: 2006-06-06 21:03:16 +0200 From: @grobian Logged In: YES user_id=963970 how about adding support for some "informative" error messages to start with? ## Comment 4014 Date: 2007-01-17 19:44:00 +0100 From: @romulogoncalves Logged In: YES user_id=1498628 Originator: NO Why this bug is not closed? Regards, Romulo ## Comment 4015 Date: 2007-02-02 09:31:53 +0100 From: @njnes Logged In: YES user_id=43556 Originator: NO Testing against out of space may cause other parts of the system to become instable (OS etc). So the output of such a test won't be stable. So we close this bug without a test! ## Comment 4016 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1312130 at http://sourceforge.net/support/tracker.php?aid=1312130
SQL:segfault upon recovery/restart
https://api.github.com/repos/MonetDB/MonetDB/issues/707/comments
0
2020-11-30T08:50:00Z
2024-06-27T11:06:27Z
https://github.com/MonetDB/MonetDB/issues/707
753,282,607
707
[ "MonetDB", "MonetDB" ]
Date: 2005-09-28 08:01:59 +0200 From: Christian Gruen &lt;<christiangruen>&gt; To: Fabian Groffen &lt;<fabian>&gt; Version: -- development Last updated: 2005-10-04 01:36:24 +0200 ## Comment 4002 Date: 2005-09-28 20:01:59 +0200 From: Christian Gruen &lt;<christiangruen>&gt; I successfully installed MonetDB on a Darwin 8.0 64bit-machine, using the --enable-bi1970-01-01 01:01:04 +0100 flag, whereas Pathfinder stops because of warnings, which are treated as errors. As soon as I remove the -Werror flags, I can't properly load the pathfinder module: _________________________________ -> FINK FILES ADDED (for Bison) export PATH=/sw/bin/:$PATH -> BUILD OPTION ADDED (otherwise wrong architecture for libxml2 is reported) > ./configure --prefix=/Users/test/monet --enable- bi1970-01-01 01:01:04 +0100 --build=powerpc64 > make warning being treated as errors: pf_support.glue.c: In function 'CMDpflock_unpack19420754': pf_support.glue.c:1217: warning: cast to pointer from integer of different size ... -> "-Werror" flags removed from runtime/ Makefile, line 294/295: > make > make install > Mserver --dbinit="module(pathfinder); pfstart();" Monet Database Server V4.8.2 Copyright (c) 1993-2005, CWI. All rights reserved. Compiled for powerpc-apple-darwin8.2.0/64bit with 64bit OIDs; dynamically linked. Visit http://monetdb.cwi.nl/ for further information. !ERROR: TBL_openmod: DL_open: library not found. !ERROR: TBL_loadmod: module load _pf_support cancelled. !ERROR: moduleClient: module(_pf_support) load error. !ERROR: TBL_openmod: DL_open: library not found. !ERROR: TBL_loadmod: module load _pathfinder cancelled. !ERROR: moduleClient: module(_pathfinder) load error. !ERROR: interpret: no matching MIL operator to 'pflock()'. MonetDB> loaded(); -> Pathfinder module is listed MonetDB> shred_doc("/Users/test/test/books.xml", "books.xml"); !ERROR: interpret: no matching MIL operator to 'lock_set(void)'. !MAYBE YOU MEAN: ! lock_set(lock) : void _________________________________ Regards, Christian G. ## Comment 4003 Date: 2005-09-29 09:12:27 +0200 From: @grobian Logged In: YES user_id=963970 About the build option: OSX Tiger provides dev-libs/libxml2-2.6.16 which appears to be pretty up-to-date to me. Fink builds 32-bits binaries/libraries, because they usually compile with gcc3 and also want to maintain compatibility with pre-G5 machines. Apple on the other hand usually builds 'fat' binaries/libraries that are 64-bits capable. It might be worth trying to just use the system provided libxml2 library. Because a compilation error on pf_support, I expect it normal that the pf_support library (dylib) cannot be loaded. This might be related to the powerpc64 build env. ## Comment 4004 Date: 2005-09-29 12:45:02 +0200 From: @grobian Logged In: YES user_id=963970 I tried configuring xquery on 64-bits using Apple provided tools and a bunch of extra files to compile MonetDB/XQuery on top of a stock install of OSX Tiger used by Gentoo on OSX. The extra file can be downloaded from: http://distfiles.gentoo.org/monetdb-pathfinder-bison-sources-0.8.2.tar.bz2 Actually I altered the ebuild to use 64-bits. It seems that the libxml2 provided by Apple doesn't do 64-bits, as configure bails out in config.log: configure:31139: checking for xmlParseDocument in -lxml2 configure:31169: gcc -m64 -o conftest -O2 -std=c99 -Wall -W -O6 -fomit-frame- pointer -finline-functions -I/usr/include/libxml2 -L/usr/lib -lxml2 -lz - lpthread -liconv -lm conftest.c -lxml2 >&5 ld64 warning: in /usr/lib/libxml2.dylib, file does not contain requested architecture ld64 warning: in /usr/lib/libz.dylib, file does not contain requested architecture ld64 warning: in /usr/lib/libiconv.dylib, file does not contain requested architecture ld64 warning: in /usr/lib/libxml2.dylib, file does not contain requested architecture ld64 warning: in /usr/lib/libgcc.a, missing required architecture ppc64 in fat file can't resolve symbols: _xmlParseDocument, referenced from: _main in ccU3iGQA.o ld64 failed: symbol(s) not found collect2: ld returned 1 exit status So the question now is what fink's library supplies (maybe `file` helps?) I compiled libxml2 as ppc64 library to see if it would work out. (env CFLAGS='-arch ppc64 -m64' ./configure --disable-dependency-tracking --with-python=no) Building a fat library somehow didn't work :/ However, I ended up with a 64-bit libxml2: % file /var/tmp/libxml2/lib/libxml2.2.6.20.dylib /var/tmp/libxml2/lib/libxml2.2.6.20.dylib: Mach-O 64-bit dynamically linked shared library ppc64 Then configure happily went to the end when doing appending "--enable-bi1970-01-01 01:01:04 +0100 --with-libxml2=/var/tmp/libxml2/" to my configure call. Make and make install also succeeded. Starting however revealed that it will take a lot more compiling to eventually get it going, because: % bin/Mserver -c etc/MonetDB.conf !WARNING: GDKlockHome: created directory /var/tmp/monetdb/var/lib/MonetDB/dbfarm/default/ !WARNING: GDKlockHome: ignoring empty or invalid .gdk_lock. !WARNING: BBPdir: initializing BBP. Monet Database Server V4.8.2 Copyright (c) 1993-2005, CWI. All rights reserved. Compiled for powerpc-apple-darwin8/64bit with 64bit OIDs; dynamically linked. Visit http://monetdb.cwi.nl/ for further information. MonetDB>module(pathfinder); !ERROR: DL_open: dlopen(/var/tmp/monetdb/lib/MonetDB/lib_pf_support.dylib,10) -> Error: dlopen(/var/tmp/monetdb/lib/MonetDB/lib_pf_support.dylib, 10): Symbol not found: _gzread !ERROR: Referenced from: /var/tmp/libxml2/lib/libxml2.2.dylib !ERROR: Expected in: flat namespace !ERROR: . !ERROR: TBL_openmod: DL_open: dlopen(/var/tmp/monetdb/lib/MonetDB/lib_pf_support.dylib, 10): Symbol not found: _gzread !ERROR: Referenced from: /var/tmp/libxml2/lib/libxml2.2.dylib !ERROR: Expected in: flat namespace !ERROR: . !ERROR: TBL_loadmod: module load _pf_support cancelled. !ERROR: moduleClient: module(_pf_support) load error. !ERROR: DL_open: dlopen(/var/tmp/monetdb/lib/MonetDB/lib_pathfinder.dylib,10) -> Error: dlopen(/var/tmp/monetdb/lib/MonetDB/lib_pathfinder.dylib, 10): Symbol not found: _gzread !ERROR: Referenced from: /var/tmp/libxml2/lib/libxml2.2.dylib !ERROR: Expected in: flat namespace !ERROR: . !ERROR: TBL_openmod: DL_open: dlopen(/var/tmp/monetdb/lib/MonetDB/lib_pathfinder.dylib, 10): Symbol not found: _gzread !ERROR: Referenced from: /var/tmp/libxml2/lib/libxml2.2.dylib !ERROR: Expected in: flat namespace !ERROR: . !ERROR: TBL_loadmod: module load _pathfinder cancelled. !ERROR: moduleClient: module(_pathfinder) load error. !ERROR: interpret: no matching MIL operator to 'pflock()'. MonetDB>pf_start(); !ERROR: interpret: no matching MIL operator to 'pf_start()'. MonetDB> zlib isn't 64-bits either, and I guess a few other libraries which are used (libiconv) as well, so it would basically require a full system install using 64-bits. In the long run this should eventually be possible using Portage on OSX. For now I think it's too much of a hassle to actually get it going. ## Comment 4005 Date: 2005-10-04 13:36:23 +0200 From: @grobian Logged In: YES user_id=963970 Marking CLOSED/CANTFIX. (there is no CANTFIX, won't fix sounded the closest :/ ) ## Comment 4006 Date: 2005-11-09 23:50:46 +0100 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 BugDay_2005-11-09, vzzzbx: NO TEST ADDED / COMPILATION ISSUE ## Comment 4007 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1307181 at http://sourceforge.net/support/tracker.php?aid=1307181
Compilation problems on G5 / Darwin 8.0
https://api.github.com/repos/MonetDB/MonetDB/issues/706/comments
0
2020-11-30T08:49:57Z
2024-06-28T13:34:10Z
https://github.com/MonetDB/MonetDB/issues/706
753,282,572
706
[ "MonetDB", "MonetDB" ]
Date: 2005-09-27 09:34:16 +0200 From: roeland &lt;<rweve>&gt; To: @njnes Version: -- development Last updated: 2005-10-09 01:06:49 +0200 ## Comment 3992 Date: 2005-09-27 21:34:16 +0200 From: roeland &lt;<rweve>&gt; When I was working on a benchmarking program and testing it, I did get the following exception: Exception in thread "main" java.sql.SQLException: Server sent unparsable update count: [ 4294966896 ] To re-create the error, I developed a simple program, called bug.java (see attachment). This is not nicely coded, but it does the trick. The program first initializes by creating a simple table, called tweety. If it already exists, it deletes all records in the table. Next, it inserts 25 records in this table. When the initialization is done, the program goes into a loop. Every time in the loop, it executes a count, insert and delete. This loop will be repeated 200 times. The program repeats itself 3 times. The first 2 times, the program does what it supposed to do, but the 3rd time it prints the exception (hopefully on your machines as well). I tested the program on Windows XP, with Monet 4.8.2 (msi installer). I'm a bit puzzled about this problem; I hope someone can help me out. The last part of the logfile of the last loop (I can only attach one file?): TX 1127848808984: write block: 67 bytes TX 1127848808984: sCREATE TABLE tweety (someid INT NOT NULL, somechar VARCHAR(255));; TX 1127848808984: zero block (flush) RX 1127848809015: zero block (flush) RX 1127848809015: new block: 45 bytes RX 1127848809015: read chunk: 45 bytes, left: 0 bytes RX 1127848809015: !CREATE TABLE: name tweety already in use TX 1127848809015: write block: 21 bytes TX 1127848809015: sDELETE FROM tweety;; TX 1127848809015: zero block (flush) RX 1127848809015: zero block (flush) RX 1127848809015: new block: 37 bytes RX 1127848809015: read chunk: 37 bytes, left: 0 bytes RX 1127848809015: - 4 querytype [ 4294966896 ] If more information is needed, please let me know. Roeland ## Comment 3993 Date: 2005-09-28 09:21:15 +0200 From: @grobian Logged In: YES user_id=963970 This problem seems to be twofold: 1) JDBC makes the incorrect assumption update counts are integers, and 2) it appears that the SQL backend goes crazy somehow, and it looks as if the transaction management fails to properly bring BATs in a consistent state. consider 1) as unfixeable: the specs define an integer update count, then 2) is left for Niels. I could not reproduce your bug exactly, but after running the example a few times and manually deleting, I got this, which indicates a bug to me: monetdb-> select * from tweety; +------------+----------+ | someid | somechar | +============+==========+ | 1 | test1 | | 2 | test2 | | 3 | test2 | | 4 | test2 | | 5 | test2 | | 1 | test1 | | 2 | test2 | | 3 | test2 | | 4 | test2 | | 5 | test2 | | 1 | test1 | | 2 | test2 | | 3 | test2 | | 4 | test2 | | 5 | test2 | | 1 | test1 | | 2 | test2 | | 3 | test2 | | 4 | test2 | | 5 | test2 | | 1 | test1 | | 2 | test2 | | 3 | test2 | | 4 | test2 | | 5 | test2 | +------------+----------+ 25 rows monetdb-> delete from tweety; -3375 affected rows ## Comment 3994 Date: 2005-09-28 09:55:54 +0200 From: @grobian Logged In: YES user_id=963970 I altered the bug.java slightly to make the testing phase somewhat longer. Altered the subject to reflect the bug problem somewhat better. ## Comment 3995 Date: 2005-09-28 13:04:43 +0200 From: @grobian Logged In: YES user_id=963970 the example code also reveals another (probably related) issue: a select count(*) gives 0 as answer for a table where just about 25 items have been inserted. ## Comment 3996 Date: 2005-10-09 01:06:49 +0200 From: @njnes Logged In: YES user_id=43556 When deleting all values of a table we clear the tables, a few cases were missed however. This is now fixed. ## Comment 3997 Date: 2005-11-09 23:49:40 +0100 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 BugDay_2005-11-09, vzzzbx: RE-ASSIGNED TO Fabian Fabian, do you still have the test-script you ran to reproduce the bug? ## Comment 3998 Date: 2005-11-10 09:28:06 +0100 From: @grobian Logged In: YES user_id=963970 vzzzbx: see attached files. ## Comment 3999 Date: 2005-11-10 11:26:16 +0100 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 that's java... do you perhaps have a non-java example? otherwise i would propose to leave this bug for now without a test-script. ## Comment 4000 Date: 2005-11-10 11:30:39 +0100 From: @grobian Logged In: YES user_id=963970 BugDay_2005-11-09, meltdown: NO TEST / TOO LARGE / POSTPONED ## Comment 4001 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1306171 at http://sourceforge.net/support/tracker.php?aid=1306171
SQL: update count makes no sense when deleting
https://api.github.com/repos/MonetDB/MonetDB/issues/705/comments
0
2020-11-30T08:49:54Z
2024-06-27T11:06:25Z
https://github.com/MonetDB/MonetDB/issues/705
753,282,539
705
[ "MonetDB", "MonetDB" ]
Date: 2005-09-27 02:23:43 +0200 From: roeland &lt;<rweve>&gt; To: MonetDB4 devs &lt;<bugs-monetdb4>&gt; Version: -- development Last updated: 2005-09-27 07:27:42 +0200 ## Comment 3982 Date: 2005-09-27 14:23:43 +0200 From: roeland &lt;<rweve>&gt; We have created a Java program that opens a connection to the Mserver using: [code] Class.forName("nl.cwi.monetdb.jdbc.MonetDriver"); Connection con = DriverManager.getConnection("jdbc:monetdb://localhost/database?debug=true", "monetdb", "monetdb"); Statement st = con.createStatement(); ... [/code] after that, it executes a couple of queries to the server for, let's say, 10 minutes. After those 10 minutes, the connection is closed with: [code] con.close(); [/code] If we start this program 2 times, everything runs fine. But if we want to start this program 3 times, the 3rd program is probably queued in Mserver, because it waits to connect to the server. If you kill one of the 2 programs running, the 3rd program immediately starts a connection to the server and executes the queries. Operating System: Windows XP (MonetDB-Installer-4.8.2.msi) with JDBC connection Java version: 1.5.0_04 Roeland and Michiel ## Comment 3983 Date: 2005-09-27 17:40:08 +0200 From: @grobian Logged In: YES user_id=963970 I can confirm this on linux, using MapiClient as well. A strace shows that the client blocks on receiving the challenge string. I cannot really figure out why the server doesn't send it. ## Comment 3984 Date: 2005-09-27 17:47:52 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 maybe too obvious, but did you try to change: mapi_clien1970-01-01 01:00:02 +0100 to a higher number in /etc/MonetDB.conf? ## Comment 3985 Date: 2005-09-27 18:12:02 +0200 From: roeland &lt;<rweve>&gt; Logged In: YES user_id=1352508 Well, apparently not many people are aware of this configuration option (we were told it was a bug). I deleted the comment before the variable and changed 2 in a higher number, restarted the server and it works fine now. ## Comment 3986 Date: 2005-09-27 18:17:33 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 I completely agree with you Roeland. I already proposed to add a link to the MonetDB.conf in the start-menu in the Windows-version (like Apache does), but the argument against this link at the time was that nobody complaint about it before.... i guess the time has come... ;) ## Comment 3987 Date: 2005-09-27 19:27:42 +0200 From: @grobian Logged In: YES user_id=963970 since when is that option there?!?!? Seems to me this rediculously low default should be increased and at least the server should actively reject the connections instead of keeping them in the TCP queue. ## Comment 3988 Date: 2005-09-27 20:12:50 +0200 From: @mlkersten Logged In: YES user_id=490798 I am not sure increasing the default is safe. Concurrency conflicts were the reason (as far as I can remember) to reduce it. For the time being is may be invalid, until a repeatable test is obtained to isolate any concurrency conflict. The reject/wait should indeed be an option of the Mapi library ## Comment 3989 Date: 2005-09-27 20:31:07 +0200 From: roeland &lt;<rweve>&gt; Logged In: YES user_id=1352508 The default setting (2) is a little bit annoying when working with Aqua Data Server (ADS) as well. Starting ADS starts 1 connection and a Query Analyzer (ctrl-q) window starts another one. If you want a 2nd Query Analyzer, ADS doesn't respond any more (actually it is waiting for Mserver, but that isn't visible). ## Comment 3990 Date: 2005-11-09 23:40:45 +0100 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 BugDay_2005-11-09, vzzzbx: NO TEST ADDED NEEDS FOLLOW UP: a test-script needs to be written that opens n client-sessions, checks if Mserver serves all of them, and then closes the sessions. As far as I know, such a test is not yet present. Please correct me if i'm wrong. ## Comment 3991 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1305797 at http://sourceforge.net/support/tracker.php?aid=1305797 The original assignee of this bug does not have an account here. Reassigning to the default assignee for the component, bugs-monetdb4@monetdb.org. Previous assignee was nobody@users.sourceforge.net.
Only max. of 2 open connections to Mserver possible
https://api.github.com/repos/MonetDB/MonetDB/issues/704/comments
0
2020-11-30T08:49:51Z
2024-06-28T13:34:09Z
https://github.com/MonetDB/MonetDB/issues/704
753,282,491
704
[ "MonetDB", "MonetDB" ]
Date: 2005-09-27 02:00:44 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @sjoerdmullender Version: -- development Last updated: 2005-11-15 11:54:36 +0100 ## Comment 3975 Date: 2005-09-27 14:00:44 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; OS: Windows 2000 Compiler: MS VC++ Toolkit 2003 Parser tools: UnxUtils embeddedclient.obj : error LNK2019: unresolved external symbol __imp__mapi_start_talking referenced in function _embedded_mil embeddedclient.obj : error LNK2019: unresolved external symbol __imp__pthread_create referenced in function _embedded_mil embeddedclient.obj : error LNK2019: unresolved external symbol __imp__mapi_embedded_init referenced in function _embedded_mil embeddedclient.obj : error LNK2019: unresolved external symbol __imp__monet_exec referenced in function _start_mil_server embeddedclient.obj : error LNK2019: unresolved external symbol __imp__THRnew referenced in function _start_mil_server embeddedclient.obj : error LNK2019: unresolved external symbol __imp__MT_getpid referenced in function _start_mil_server embeddedclient.obj : error LNK2019: unresolved external symbol _static_module_init referenced in function _start_mil_server embeddedclient.obj : error LNK2019: unresolved external symbol __imp__mo_free_options referenced in function _start_mil_server embeddedclient.obj : error LNK2019: unresolved external symbol __imp__mo_system_config referenced in function _start_mil_server embeddedclient.obj : error LNK2019: unresolved external symbol __imp__mo_builtin_settings referenced in function _start_mil_server embeddedclient.obj : error LNK2019: unresolved external symbol __imp__monet_init referenced in function _Monet_init embeddedclient.obj : error LNK2001: unresolved external symbol __imp__monet_daemon embeddedclient.obj : error LNK2001: unresolved external symbol __imp__GDKembedded embeddedclient.obj : error LNK2019: unresolved external symbol __imp__GDKgetenv_isyes referenced in function _Monet_init embeddedclient.obj : error LNK2001: unresolved external symbol __imp__GDK_vm_minsize embeddedclient.obj : error LNK2001: unresolved external symbol __imp__GDK_mem_bigsize embeddedclient.obj : error LNK2001: unresolved external symbol __imp__GDKdebug embeddedclient.obj : error LNK2019: unresolved external symbol __imp__GDKgetenv referenced in function _Monet_init embeddedclient.obj : error LNK2019: unresolved external symbol __imp__GDKsetenv referenced in function _Monet_init embeddedclient.obj : error LNK2019: unresolved external symbol __imp__GDKinit referenced in function _Monet_init embeddedclient.obj : error LNK2019: unresolved external symbol __imp__mo_substitute referenced in function _Monet_init embeddedclient.obj : error LNK2019: unresolved external symbol __imp__GDKfree referenced in function _Monet_init embeddedclient.obj : error LNK2019: unresolved external symbol __imp__mo_find_option referenced in function _Monet_init embeddedclient.obj : error LNK2019: unresolved external symbol __imp__GDKmalloc referenced in function _Monet_init NMAKE : fatal error U1077: 'cl' : return code '0x2' Stop. ## Comment 3976 Date: 2005-09-29 12:36:25 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 By trial and error, I just managed it to compile. Here are the changes I made: 1) Added missing LIBS 2) Changed embeddedclient.mx such that embedded_mil() is exported and embeddedclient.lib is created ## Comment 3977 Date: 2005-11-09 23:05:38 +0100 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 BugDay_2005-11-09, vzzzbx: NO TEST ADDED / COMPILATION ISSUE ## Comment 3978 Date: 2005-11-15 10:49:21 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 Do you still have difficulty compiling or can this report be closed? I can compile MonetDB and SQL without problems using the Microsoft tools (using up-to-date buildtools obviously). ## Comment 3979 Date: 2005-11-15 11:44:40 +0100 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 MonetDB isok, but building SQL fails: cd "winsetup" ... LINK : fatal error LNK1181: cannot open input file 'libodbc32.lib' All libs have this 'lib' prefix in Makefile.msc!? ## Comment 3980 Date: 2005-11-15 11:54:35 +0100 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 Aah, it works with up-to-date buildtools! I close the bug. ## Comment 3981 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1305779 at http://sourceforge.net/support/tracker.php?aid=1305779
libembeddedmil.dll: missing libs
https://api.github.com/repos/MonetDB/MonetDB/issues/703/comments
0
2020-11-30T08:49:47Z
2024-06-28T13:34:09Z
https://github.com/MonetDB/MonetDB/issues/703
753,282,435
703
[ "MonetDB", "MonetDB" ]
Date: 2005-09-25 07:03:12 +0200 From: @grobian To: @sjoerdmullender Version: -- development Last updated: 2006-01-15 02:48:29 +0100 ## Comment 3968 Date: 2005-09-25 19:03:12 +0200 From: @grobian /bin/bash ../../libtool --mode=compile gcc -m64 -DHAVE_CONFIG_H -I. -I../../../src/gdk -I../.. -I../../../src/gdk -I../common -DLIBGDK -g -O2 -std=c99 -D__EXTENSIONS__ -Wall -W -fno-strict-aliasing -DHWCOUNTERS -DHW_SunOS -DHW_sun4u -Werror-implicit-function-declaration -Werror -Wno-unused-function -Wno-unused-label -Wno-uninitialized -D_REENTRANT -c -o libbat_la-gdk_atoms.lo `test -f 'gdk_atoms.c' || echo '../../../src/gdk/'`gdk_atoms.c gcc -m64 -DHAVE_CONFIG_H -I. -I../../../src/gdk -I../.. -I../../../src/gdk -I../common -DLIBGDK -g -O2 -std=c99 -D__EXTENSIONS__ -Wall -W -fno-strict-aliasing -DHWCOUNTERS -DHW_SunOS -DHW_sun4u -Werror-implicit-function-declaration -Werror -Wno-unused-function -Wno-unused-label -Wno-uninitialized -D_REENTRANT -c gdk_atoms.c -fPIC -DPIC -o .libs/libbat_la-gdk_atoms.o ../../../src/gdk/gdk_atoms.mx: In function `fltFromStr': ../../../src/gdk/gdk_atoms.mx:1498: error: `__builtin_infinity' undeclared (first use in this function) ../../../src/gdk/gdk_atoms.mx:1498: error: (Each undeclared identifier is reported only once ../../../src/gdk/gdk_atoms.mx:1498: error: for each function it appears in.) ../../../src/gdk/gdk_atoms.mx:1501: error: invalid operands to binary == gmake[4]: *** [libbat_la-gdk_atoms.lo] Error 1 gmake[4]: Leaving directory `/nfs/poseidon/export/scratch2/Temp/monet/MonetDB/build-solaris/src/gdk' gmake[3]: *** [all] Error 2 gmake[3]: Leaving directory `/nfs/poseidon/export/scratch2/Temp/monet/MonetDB/build-solaris/src/gdk' gmake[2]: *** [all-recursive] Error 1 gmake[2]: Leaving directory `/nfs/poseidon/export/scratch2/Temp/monet/MonetDB/build-solaris/src' gmake[1]: *** [all-recursive] Error 1 gmake[1]: Leaving directory `/nfs/poseidon/export/scratch2/Temp/monet/MonetDB/build-solaris' gmake: *** [all] Error 2 % gcc --version gcc (GCC) 3.4.2 Copyright (C) 2004 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. math.h doesn't seem to define any NAN values for me. ## Comment 3969 Date: 2005-09-25 23:37:06 +0200 From: @drstmane Logged In: YES user_id=572415 Without access to a Solaris 10 machine, I cannot help much, here --- except: - search through all your system- & compiler-include files, where "NAN" is defined as "__builtin_infinity" (or something else) and under which circumstance/conditions - search through all your system- & compiler-include files, where "__builtin_infinity" is declared, and under which circumstance/conditions - check, whether the respective include files are (transitively) included by src/gdk/gdk_atoms.mx, and whether the respetive circumstance/conditions du hold at time of inclusion ## Comment 3970 Date: 2005-09-26 14:01:06 +0200 From: @grobian Logged In: YES user_id=963970 % find /usr/include -name "*.h" | xargs grep __builtin_infinity /usr/include/iso/math_c99.h:define INFINITY __builtin_infinity The only thing I can find on this is this: http://gcc.gnu.org/ml/gcc-patches/2005-05/msg02123.html (which doesn't sound too much as a quick solution) ## Comment 3971 Date: 2005-09-26 14:04:18 +0200 From: @grobian Logged In: YES user_id=963970 how do I easily tell GCC to look for it's includes here? /usr/sfw/lib/gcc/sparc-sun-solaris2.10/3.4.3/include/ instead of /usr/include/ ## Comment 3972 Date: 2005-11-09 23:04:39 +0100 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 BugDay_2005-11-09, vzzzbx: NO TEST ADDED / COMPILATION ISSUE ## Comment 3973 Date: 2006-01-15 14:48:29 +0100 From: @grobian Logged In: YES user_id=963970 Solaris support has been dropped till there's actually a demand for it. ## Comment 3974 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1304078 at http://sourceforge.net/support/tracker.php?aid=1304078
MonetDB: __builtin_infitiny on Solaris 10
https://api.github.com/repos/MonetDB/MonetDB/issues/702/comments
0
2020-11-30T08:49:44Z
2024-06-28T13:34:52Z
https://github.com/MonetDB/MonetDB/issues/702
753,282,404
702
[ "MonetDB", "MonetDB" ]
Date: 2005-09-25 04:57:06 +0200 From: @grobian To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: -- development Last updated: 2005-09-26 01:06:55 +0200 ## Comment 3959 Date: 2005-09-25 16:57:06 +0200 From: @grobian [helios:monet/MonetDB/build-solaris] % gmake make all-recursive /bin/bash: make: command not found gmake: *** [all] Error 127 [helios:monet/MonetDB/build-solaris] % which make make: Command not found. [helios:monet/MonetDB/build-solaris] % which gmake /opt/sfw/bin/gmake autoconf/automake/configure doesn't like gmake, or am I missing an obvious flag here? ## Comment 3960 Date: 2005-09-25 17:12:01 +0200 From: @grobian Logged In: YES user_id=963970 using some script-fu I replaced make with gmake and am so far still compiling ## Comment 3961 Date: 2005-09-25 17:31:04 +0200 From: @drstmane Logged In: YES user_id=572415 I never expeirenced similar problem on our Solaris 8 box. Couls you please check with the following simple Makefile, whether your gmake correctly sets $(MAKE) to "gmake"? 17:25:45 manegold@corona:/tmp $ cat Makefile all: echo $(MAKE) 17:25:48 manegold@corona:/tmp $ make echo make make 17:25:53 manegold@corona:/tmp $ gmake echo gmake gmake 17:25:56 manegold@corona:/tmp $ ## Comment 3962 Date: 2005-09-25 19:02:27 +0200 From: @grobian Logged In: YES user_id=963970 [helios:/var/tmp] % cat Makefile all: echo $(MAKE) [helios:/var/tmp] % gmake echo gmake gmake ## Comment 3963 Date: 2005-09-25 19:06:00 +0200 From: @grobian Logged In: YES user_id=963970 maybe I should add that the bootstrap script had some problems but appeared to continue. It complained about some syntax options for grep iirc that were too GNU-ish to be accepted ## Comment 3964 Date: 2005-09-25 23:39:30 +0200 From: @drstmane Logged In: YES user_id=572415 I don't think that the bootstrap problems are related to the make problems, but I cannot be sure unless I see the detailed bootstrap output... Further, maybe "gmake MAKE=gmake ..." helps... ## Comment 3965 Date: 2005-09-26 13:06:55 +0200 From: @grobian Logged In: YES user_id=963970 adding Solaris "make" to the path seemed to have cured whatever tool generated MAKE=make, probably because it realised it would never work. gmake now actually starts compiling without any script-fu. ## Comment 3966 Date: 2005-11-09 23:03:52 +0100 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 BugDay_2005-11-09, vzzzbx: NO TEST ADDED / COMPILATION ISSUE ## Comment 3967 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1303997 at http://sourceforge.net/support/tracker.php?aid=1303997
Build: unable to use gmake on Solaris 10
https://api.github.com/repos/MonetDB/MonetDB/issues/701/comments
0
2020-11-30T08:49:41Z
2024-06-28T13:34:52Z
https://github.com/MonetDB/MonetDB/issues/701
753,282,371
701
[ "MonetDB", "MonetDB" ]
Date: 2005-09-23 10:46:32 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; To: Peter Boncz &lt;<boncz>&gt; Version: -- development Last updated: 2005-10-19 01:00:20 +0200 ## Comment 3956 Date: 2005-09-23 10:46:32 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; just try the query '1 to 10' ## Comment 3957 Date: 2005-11-09 23:01:39 +0100 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 BugDay_2005-11-09, vzzzbx: TEST ADDED / SUCCESS ## Comment 3958 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1300366 at http://sourceforge.net/support/tracker.php?aid=1300366
constant stuff does not work in op:to
https://api.github.com/repos/MonetDB/MonetDB/issues/700/comments
0
2020-11-30T08:49:38Z
2024-06-28T13:34:51Z
https://github.com/MonetDB/MonetDB/issues/700
753,282,343
700
[ "MonetDB", "MonetDB" ]
Date: 2005-09-22 02:57:45 +0200 From: Jens Teubner &lt;<teubner>&gt; To: GDK devs &lt;<bugs-common>&gt; Version: -- development Last updated: 2006-01-24 08:32:42 +0100 ## Comment 3948 Date: 2005-09-22 14:57:45 +0200 From: Jens Teubner &lt;<teubner>&gt; I just got this reported from Konstanz. I cannot confirm myself, as I don't have bison 2.0 here. System is Darwin. Compilation fails with: gcc -m32 -DHAVE_CONFIG_H -I. -I. -I.. -I. -g -O2 -std=c99 -Wall - W -Werror-implicit-function-declaration -Werror -Wno-unused- function -Wno-unused-label -Wno-uninitialized -c gram.tab.c /usr/share/bison.simple: In function 'yyparse': /usr/share/bison.simple:361: error: implicit declaration of function 'free' make[3]: *** [gram.tab.o] Error 1 yyparse() is entirely under control of bison. To me this looks like bison generates a call to free() without including stdlib.h first, which would make this a bison bug. Adding an include <stdlib.h> into gram.y apparently fixed the problem. As I said, I can neither confirm nor invalidate the problem, for the lack of bison 2.0. ## Comment 3949 Date: 2005-09-22 18:04:44 +0200 From: @grobian Logged In: YES user_id=963970 /usr/share/bison.simple this is a system installed bison. So question is, is this really bison 2, or just bison 1.31 or something which is shipped by default on OSX. (Assuming the OSX version of Darwin is meant here) emerge monetdb should also work on OSX machines (though I doubt whether that results in the desired situation here) ## Comment 3950 Date: 2005-09-22 18:15:37 +0200 From: @Teggy Logged In: YES user_id=642632 The bison shipped with OS X (GNU bison v1.28) is incapable of dealing with token position information and thus not a viable choice in any case. I am using GNU bison v1.35 on my OS X box and cannot reproduce the problem reported by Jens (or by the Konstanz folks, for that matter). Why not ask Konstanz to use that version of bison instead? (Available via fink.) ## Comment 3951 Date: 2005-09-23 11:44:10 +0200 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 Teggy, the Konstanz people got their stuff compiled, don't worry. I mainly submitted that bug report to have the problem "on file". If somebody really feels like it, he/she can look into this. Otherwise we are at least aware of a possible issue if somebody runs into that problem. Jens ## Comment 3952 Date: 2005-09-25 23:47:13 +0200 From: @drstmane Logged In: YES user_id=572415 On Fedora Core 4 (64-bit), MonetDB, Pathfinder & Burg compile without problems with GNU bison 2.0 . Hence, it doesn't seem to be a GNU bison 2.0 problem/bug in general, but (only?) on Darwin, or at least some (which?) version of Darwin... ## Comment 3953 Date: 2005-12-19 16:59:25 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / COMPILATION ## Comment 3954 Date: 2006-01-24 20:32:41 +0100 From: @grobian Logged In: YES user_id=963970 Since bison isn't necessary any more to build MonetDB/XQuery, I think this bug is obsolete now. Please reopen if you think otherwise. For me the problem doesn't occur anymore on Darwin8. ## Comment 3955 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1298707 at http://sourceforge.net/support/tracker.php?aid=1298707 The original assignee of this bug does not have an account here. Reassigning to the default assignee for the component, bugs-common@monetdb.org. Previous assignee was nobody@users.sourceforge.net.
possible problem with burg and bison 2.0
https://api.github.com/repos/MonetDB/MonetDB/issues/699/comments
0
2020-11-30T08:49:35Z
2024-06-27T11:06:18Z
https://github.com/MonetDB/MonetDB/issues/699
753,282,309
699
[ "MonetDB", "MonetDB" ]
Date: 2005-09-21 05:15:33 +0200 From: Wouter Alink &lt;<alink>&gt; To: XQuery devs &lt;<bugs-xquery>&gt; Version: -- development Last updated: 2005-11-25 01:03:20 +0100 ## Comment 3943 Date: 2005-09-21 17:15:33 +0200 From: Wouter Alink &lt;<alink>&gt; Given the query: let $d := doc("bug.xml") return if (1 = 0) then () else let $f := $d//file return element { "result" } { element { "matches" } { count($f) }, element { "results" } { for $i in 1 to 10 return element { "result" } { count($f[$i]/*) } } } And the document: <file><file>a</file><file>b</file></file> Results in: ERROR: CTrefine: both BATs must have the same cardinality and their heads must form a 1-1 match. The query cannot be reduced very much. Leaving out the 'if' or the 'element { "matches" }', or the '/*' will make the query behave correctly. ## Comment 3944 Date: 2005-09-23 10:13:27 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; Logged In: YES user_id=993208 In addition to the mil optimizer which prunes necessary code (see example below), the join recognition ignored that an if-then-else expression also introduces a new scope... The last checkin fixes the latter problem (and this bug). example: var v_item001 := v_item000; var v_kind001; prunes the kind code v_item001 := v_item001.insert(nil,12@0); v_item000.fetch(0); returns 1 tuple v_kind000.fetch(0); returns 0 tuples ## Comment 3945 Date: 2005-11-25 01:03:20 +0100 From: @peterboncz Logged In: YES user_id=591107 this works for me now. closing.. ## Comment 3946 Date: 2005-12-19 16:57:42 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: ALREADY IN TESTWEB pathfinder/tests/BugsViaSourgeforce/Tests/ID.1297836-1.xq ## Comment 3947 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1297836 at http://sourceforge.net/support/tracker.php?aid=1297836 The original assignee of this bug does not have an account here. Reassigning to the default assignee for the component, bugs-xquery@monetdb.org. Previous assignee was nobody@users.sourceforge.net.
XQuery: if+let+for+element
https://api.github.com/repos/MonetDB/MonetDB/issues/698/comments
0
2020-11-30T08:49:32Z
2024-06-27T11:06:17Z
https://github.com/MonetDB/MonetDB/issues/698
753,282,251
698
[ "MonetDB", "MonetDB" ]
Date: 2005-09-21 11:29:12 +0200 From: Jens Teubner &lt;<teubner>&gt; To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: unspecified Last updated: 2005-09-21 12:47:01 +0200 ## Comment 3939 Date: 2005-09-21 11:29:12 +0200 From: Jens Teubner &lt;<teubner>&gt; "[an error occurred while processing this directive]" http://monetdb.cwi.nl/Research/Articles/index.html ## Comment 3940 Date: 2005-09-21 12:47:01 +0200 From: @drstmane Logged In: YES user_id=572415 thanks for lettin us know! It should be fixed, again. ## Comment 3941 Date: 2005-12-19 16:56:11 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / DOCUMENTATION ## Comment 3942 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1297454 at http://sourceforge.net/support/tracker.php?aid=1297454
website: research articles page broken
https://api.github.com/repos/MonetDB/MonetDB/issues/697/comments
0
2020-11-30T08:49:29Z
2024-06-27T11:06:16Z
https://github.com/MonetDB/MonetDB/issues/697
753,282,222
697
[ "MonetDB", "MonetDB" ]
Date: 2005-09-20 03:31:57 +0200 From: @grobian To: Romulo Goncalves &lt;<goncalve>&gt; Version: -- development Last updated: 2007-03-28 07:34:32 +0200 ## Comment 3927 Date: 2005-09-20 15:31:57 +0200 From: @grobian monetdb-> create table a (id int primary key); Operation successful monetdb-> create table b (id int, foreign key (id) references a(id) on delete cascade); Operation successful monetdb-> insert into a values (1); 1 affected row monetdb-> insert into b values(1); 1 affected row monetdb-> delete from a; 1 affected row monetdb-> select * from b; +------------+ | id | +============+ | 1 | +------------+ 1 row monetdb-> although the ON DELETE CASCADE clause is accepted, it is not implemented. We better not accept this, or best just implement it. ## Comment 3928 Date: 2005-12-19 16:55:28 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / FAILURE sql/src/test/BugDay_2005-12-19_2.9.3/Tests/delete_cascade.SF-1296395.sql ## Comment 3929 Date: 2006-01-15 14:46:50 +0100 From: @grobian Logged In: YES user_id=963970 this needs major work, postponing for now. ## Comment 3930 Date: 2006-01-22 21:26:44 +0100 From: @drstmane Logged In: YES user_id=572415 postponed until after the release, I suppose? ## Comment 3931 Date: 2006-01-22 21:58:36 +0100 From: @grobian Logged In: YES user_id=963970 yes ## Comment 3932 Date: 2006-06-06 21:00:30 +0200 From: @grobian Logged In: YES user_id=963970 Niels, shall we remove the syntactical support for this from the parser? (Like is done for WITH CHECK OPTION?) ## Comment 3933 Date: 2007-01-18 11:03:25 +0100 From: @romulogoncalves Logged In: YES user_id=1498628 Originator: NO In my opinion it should be implemented. It will take some time, but it is a nice feature for our system. What do you think? Regards, Romulo ## Comment 3934 Date: 2007-01-24 11:35:54 +0100 From: @romulogoncalves Logged In: YES user_id=1498628 Originator: NO well, if you now try to execute a DELETE over a primary key with foreign keys you get an error message. See this example. The tables and constraints: create table t1(id int, name varchar(1024)); alter table t1 add constraint id_p primary key(id); create table t2(id_f int, age int, foreign key(id_f) references t1(id)); alter table t2 add constraint age_p primary key(age); The test: sql>delete from t1 where id = 0; MAPI = monetdb@localhost:50000 QUERY = delete from t1 where id = 0; ERROR = !DELETE FROM: cannot delete from table 't1' there are foreign keys depending on it, please use DELETE CASCADED! sql>drop table t2; sql>delete from t1 where id = 0; [ 1 ] sql> The same it will implemented for updates over primary keys. My question is: Is this enough to close the bug or the implementation of DELETE/UPDATE CASCADED is needed? Regards, Romulo ## Comment 3935 Date: 2007-01-24 11:41:01 +0100 From: @grobian Logged In: YES user_id=963970 Originator: YES > My question is: > Is this enough to close the bug or the implementation of DELETE/UPDATE > CASCADED is needed? No, this is not enough, because as far as I understand, "ON DELETE CASCADE" is still accepted, but not acted upon. ## Comment 3936 Date: 2007-03-28 17:09:15 +0200 From: @romulogoncalves Logged In: YES user_id=1498628 Originator: NO This bug, can be closed. The example given by this bug is covered by the test: sql/src/test/BugDay_2005-12-19_2.9.3/Tests/delete_cascade.SF-1296395.sql Which now is green. :) Regards, Romulo ## Comment 3937 Date: 2007-03-28 19:34:32 +0200 From: @grobian Logged In: YES user_id=963970 Originator: YES Closing, kudos to Romulo. ## Comment 3938 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1296395 at http://sourceforge.net/support/tracker.php?aid=1296395
SQL: ON DELETE/UPDATE CASCADE accepted but not executed
https://api.github.com/repos/MonetDB/MonetDB/issues/696/comments
0
2020-11-30T08:49:26Z
2024-06-27T11:06:15Z
https://github.com/MonetDB/MonetDB/issues/696
753,282,188
696
[ "MonetDB", "MonetDB" ]
Date: 2005-09-20 03:27:55 +0200 From: @grobian To: @njnes Version: -- development Last updated: 2005-10-23 12:09:26 +0200 ## Comment 3924 Date: 2005-09-20 15:27:55 +0200 From: @grobian monetdb-> create table t (id int); Operation successful monetdb-> insert into t values (1); 1 affected row monetdb-> insert into t values (select max(id) + 1 from t); Error: syntax error, unexpected SELECT in: "insert into t values (select" monetdb-> insert into t values ((select max(id) + 1 from t)); 1 affected row I think the extra ( ) here should be superfluous and it should be allowed to go without them. ## Comment 3925 Date: 2005-12-19 16:49:19 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS sql/src/test/BugDay_2005-12-19_2.9.3/Tests/subquery_syntax.SF-1296390.sql ## Comment 3926 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1296390 at http://sourceforge.net/support/tracker.php?aid=1296390
SQL: relaxing subqueries
https://api.github.com/repos/MonetDB/MonetDB/issues/695/comments
0
2020-11-30T08:49:23Z
2024-06-27T11:06:14Z
https://github.com/MonetDB/MonetDB/issues/695
753,282,154
695
[ "MonetDB", "MonetDB" ]
Date: 2005-09-19 11:08:09 +0200 From: Wouter Alink &lt;<alink>&gt; To: XQuery devs &lt;<bugs-xquery>&gt; Version: -- development Last updated: 2005-10-06 11:00:16 +0200 ## Comment 3912 Date: 2005-09-19 23:08:09 +0200 From: Wouter Alink &lt;<alink>&gt; from what i understand of; http://www.w3.org/TR/xmlschema-2/int it seems that the xs:integer type should support integers of at least 18 digits to meet the XQuery-specs as it is derived from xs:decimal. however, it seems that currently integers above the 2G- border return an error, even if monetDB is compiled in everything-64-bits-mode (at least MonetDB says: Compiled for x86_64-redhat-linux-gnu/64bit with 64bit OIDs; dynamically linked.) ## Comment 3913 Date: 2005-09-19 23:21:52 +0200 From: @drstmane Logged In: YES user_id=572415 In MonetDB, "int" are 32-bit (signed, i.e., [-((2^31)-1),(2^31)-1], or at most 9 digits) and "lng" are 64-bit (signed, i.e., [-((2^63)-1),(2^63)-1], or at most 18 digits. This is independed of whether MonetDB is compiled with 32 or 64 bit. Currently, xs:integer seems to be implemented as "int", and is hence limited to 9 digits. To allow for 18 digits, xs:integer would have to be implemented as "lng". ## Comment 3914 Date: 2005-09-19 23:27:30 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 that's already what i guessed, but i wanted to make sure i didn't overlook a special feature of monetdb... ;) i guess this means the bug-report is valid as the implementation doesn't met the specs. ## Comment 3915 Date: 2005-09-20 10:21:57 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 i think i have my mserver working with xs:integers as lng's now... have to wait for mtest to finish, if all is ok, i will check it in ## Comment 3916 Date: 2005-09-20 17:58:02 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 attached is the diff-script for the fix: the xs:integer type is now represented as 'lng' in MIL (used to be 'int') according to the testweb everything is fine, but i doubt if the tests are sufficient from what i understand of the milprint_summer code is that sometimes the 'num'-field of a 'PFcnode_t' is (ab)used for storing 'fid's ('c->sem.num = fid;'). these fid's are actually oid's, disguised as int's. this will not be very safe, but could somebody (preferably an expert, Jan/Stefan??) have a look at the attached code? ## Comment 3917 Date: 2005-09-20 19:15:30 +0200 From: @grobian Logged In: YES user_id=963970 wouter, nice diff! Too bad sf doesn't allow inline viewing, but ok. I was just wondering whether long long int would be portable, and whether %lld is also portable. I'm not sure on this, so ignore if invalid, but aren't there macros defined in MonetDB somewhere to deal with the right "long" on each and every exotic platform we test on? ## Comment 3918 Date: 2005-09-20 19:27:56 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 Stefan suggested something like "LNGFMT", but don't know in which library it is. Any clues? The diff at least fixes some casting from signed to unsigned. ## Comment 3919 Date: 2005-09-20 19:29:56 +0200 From: @drstmane Logged In: YES user_id=572415 Fabian, you're right, %lld is for sure not portable to (among others) native Windows compilers and the same holds most probably for long long int. However, pathfinder does not compile with native Windows compilers (at least not with the Microsoft compiler); on Windows, we're bound to using MinGW, and that is GNU. Nevertheless, we should indeed consider using the macros "lng" and "LNGFMT" as in MonetDB; probabaly, they'll have to be re-defined for pathfinder since the compiler is supposed to compile stand-alone without (too many) MonetDB dependencies... ## Comment 3920 Date: 2005-10-06 22:33:42 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 BugDay_2005-10-06: Done. CLAIMED BY vzzzbx ALREADY IN TESTWEB test-script "tests/XQuery/Tests/type_integer.xq" was already added ## Comment 3921 Date: 2005-10-06 23:00:15 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 fixed in CVS current-branch. due to the rather large amount of changes to the code, i checked it in as a "feature" in the current- branche ;) ## Comment 3922 Date: 2005-12-19 16:46:15 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: ALREADY IN TESTWEB pathfinder/tests/XQuery/Tests/type_integer.xq ## Comment 3923 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1295786 at http://sourceforge.net/support/tracker.php?aid=1295786 The original assignee of this bug does not have an account here. Reassigning to the default assignee for the component, bugs-xquery@monetdb.org. Previous assignee was nobody@users.sourceforge.net.
XQuery: ceiling xs:integer
https://api.github.com/repos/MonetDB/MonetDB/issues/694/comments
0
2020-11-30T08:49:20Z
2024-06-27T11:06:13Z
https://github.com/MonetDB/MonetDB/issues/694
753,282,122
694
[ "MonetDB", "MonetDB" ]
Date: 2005-09-19 04:47:51 +0200 From: @mlkersten To: @njnes Version: -- development Last updated: 2005-09-19 06:53:38 +0200 ## Comment 3908 Date: 2005-09-19 16:47:51 +0200 From: @mlkersten It is possible to generate serious crashes (Segfault) using your ability to drop tables from the catalog. It recovers, as far as I can detect upon restart. ## Comment 3909 Date: 2005-09-19 18:53:38 +0200 From: @grobian Logged In: YES user_id=963970 I assume you mean system tables here. But I bet you login as monetdb:monetdb, which makes you the equivalent of root on your linux box, hence I deem your bug to be invalid. Please reopen if I did not understand your bug correctly. ## Comment 3910 Date: 2005-12-19 16:42:46 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST ## Comment 3911 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1295482 at http://sourceforge.net/support/tracker.php?aid=1295482
SQL: self protection too minimalistic
https://api.github.com/repos/MonetDB/MonetDB/issues/693/comments
0
2020-11-30T08:49:17Z
2024-06-28T13:34:50Z
https://github.com/MonetDB/MonetDB/issues/693
753,282,087
693
[ "MonetDB", "MonetDB" ]
Date: 2005-09-19 10:41:45 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; To: Peter Boncz &lt;<boncz>&gt; Version: -- development Last updated: 2006-01-25 09:13:57 +0100 ## Comment 3900 Date: 2005-09-19 10:41:45 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; The help of MapiClient says about the option '-s': -s stmt | --statement=stmt /* run single statement */ It never came into my mind that this could be used to change the serialization type of my result if I use it in combination with '-l xquery'. As I started with '-s foo' I got some nice text which can be interpreted as sax events. However other options (dm, xml-seq, xml, sax) are not mentioned anywhere I looked (website & Mapi help). Can this please be mentioned at a prominent place and are there perhaps even more options? -- e.g. 'bar' seems to be the same as sax again. Therefore I propose to disable non-existing serialization modes. ## Comment 3901 Date: 2005-09-19 10:54:35 +0200 From: @drstmane Logged In: YES user_id=572415 In fact, this "-sxml" is kind of a (quick...) hack (sorry, Peter ;-)). No being a MAPI expert myself, I resently discussed it with Niels, and he said that the "challange" feature (and an extra option to MapiClient) would actually be the correct way to specify the serialization mode and other configurations for MonetDB/XQuery. I (temporarly?) assign this to Peter, as he intorduced the "-sxml" hack; of course, any MAPI expert it more than welcome to enlighten me or Peter how to solve this problem properly. ## Comment 3902 Date: 2005-09-19 16:29:12 +0200 From: @grobian Logged In: YES user_id=963970 I think niels meant that there is a 'language' field in the string sent by the client. This field can be exploited/used to tell the server in a nice way what kind out output it should produce. This would then involve a -l (--language) switch instead. ## Comment 3903 Date: 2005-09-19 16:44:52 +0200 From: @drstmane Logged In: YES user_id=572415 "-l xquery" / "--language=xquery" is already used (and documented) for XQuery. The point here is, that in addition to setting the language to XQuery, we also can set the "output style" (xml, dm, ...) and switch on/off other options like debugging, optimization, timing for the XQuery compiler, mis-using MapiClient's "-s"/"--statement=" option. This hack should be replaced by a proper implementation using the "challange-string" feature of the MAPI protocol, i.e., instead of prefixing the XQuery by a line that is derived from the mis-used "-s" option, we should send a challange string that contains not only things like "mode=blocked", but also things like "debug=<dbg-level>", "optimization=<opt-level>", "timing={on/off}" ... (Provided my humble knowledge of the MAPI protocol is not completely bogus...) ## Comment 3904 Date: 2005-09-19 19:04:27 +0200 From: @grobian Logged In: YES user_id=963970 The clean solution you propose is already forseen in MCL. I don't think the current solution is that bad (because it's under the hood, and the user doesn't see this happening, you can also change the output per query, although I don't think it is used now), the real problem is the interface through MapiClient I guess. I could imagine either one of these two hacks: 1% MapiClient --language=xquery:xml 2% MapiCliient -lxquery -o xml at 1 the language can be followed by a colon with the output mode, if not specified, the default is used, where at 2 a separate flag is used which can only be used with language xquery. I don't really have a preference, so I think it will be what's easier to implement in MapiClient. ## Comment 3905 Date: 2005-12-19 16:41:36 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / DOCUMENTATION Either the (mis)use of the -s option for XQuery should be better document, or an alternative solution should be implemented. In either case, no test. ## Comment 3906 Date: 2006-01-25 21:13:57 +0100 From: @peterboncz Logged In: YES user_id=591107 it is documented as of now.. ## Comment 3907 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1295145 at http://sourceforge.net/support/tracker.php?aid=1295145
documentation of option '-s' missing
https://api.github.com/repos/MonetDB/MonetDB/issues/692/comments
0
2020-11-30T08:49:14Z
2024-06-27T11:06:11Z
https://github.com/MonetDB/MonetDB/issues/692
753,282,047
692
[ "MonetDB", "MonetDB" ]
Date: 2005-09-18 08:28:04 +0200 From: @drstmane To: Peter Boncz &lt;<boncz>&gt; Version: -- development Last updated: 2005-09-19 04:05:53 +0200 ## Comment 3895 Date: 2005-09-18 20:28:04 +0200 From: @drstmane After changes made between Thursday, Sep 15 2005, 0:02 CEST, and Friday, Sep 16 2005, 0:02 CEST, tests tests/XQuery/Tests/let_for[23].xq do not generate/show "warning: variable reuse: at (0,0-0,0): $a" any more. I'm not sure, whether this was intended, or is a bug... ## Comment 3896 Date: 2005-09-19 10:49:15 +0200 From: @drstmane Logged In: YES user_id=572415 For the bookkeeping (might be helpful to fix the bug): The warning is still generated by the (stand-alone) compiler, but seems to ge swallowed in the Mserver-embedded compiler after Peter's following changes: compiler/compile.c: 1.41 -> 1.43 compiler/mil/milprint_summer.c: 1.103 -> 1.104 compiler/include/milprint_summer.h: 1.7 -> 1.8 " - fixed a bug introduced yesterday in mil_opt - we now report out-of-memory errors in milprint_summer code generation - hm. let's make it compile. " Sorry, but I have no time, now, to analyse the details and fix the problem ... ## Comment 3897 Date: 2005-09-19 16:05:53 +0200 From: @drstmane Logged In: YES user_id=572415 Fixed by Peter. ## Comment 3898 Date: 2005-12-19 16:38:28 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / REDUNDANT ## Comment 3899 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1294712 at http://sourceforge.net/support/tracker.php?aid=1294712
XQuery: tests tests/XQuery/Tests/let_for[23].xq fail
https://api.github.com/repos/MonetDB/MonetDB/issues/691/comments
0
2020-11-30T08:49:11Z
2024-06-27T11:06:10Z
https://github.com/MonetDB/MonetDB/issues/691
753,282,011
691
[ "MonetDB", "MonetDB" ]
Date: 2005-09-16 09:34:47 +0200 From: @grobian To: GDK devs &lt;<bugs-common>&gt; Version: -- development Last updated: 2006-06-06 08:58:48 +0200 ## Comment 3883 Date: 2005-09-16 21:34:47 +0200 From: @grobian This applies to the MonetDB 4.8.2 en SQL 2.8.2 and XQuery 0.8.2 tar distributions. (no group for 4.8.2 is available yet) the following files pollute the user's filesystem when installing the full family suite of MonetDB: testing stuff that the user didn't ask for IMHO: /usr/bin/Mapprove.py -> Mtest.py /usr/bin/Mdiff /usr/bin/Mfilter.py /usr/bin/MkillUsers /usr/bin/Mlog /usr/bin/Mprofile.py -> Mtest.py /usr/bin/Mtest.py /usr/bin/Mtimeout /usr/bin/prof.py /usr/lib/MonetDB/Tests /usr/lib/MonetDB/Tests/sample0 /usr/lib/MonetDB/Tests/sample1 /usr/lib/MonetDB/Tests/sample2 /usr/lib/MonetDB/Tests/sample3 /usr/lib/MonetDB/Tests/sample4 /usr/lib/MonetDB/Tests/smack00 /usr/lib/MonetDB/Tests/smack01 /usr/lib/sql/Tests /usr/lib/sql/Tests/odbcsample1 /usr/share/MonetDB/Mprofile-commands.lst /usr/share/MonetDB/Tests /usr/share/MonetDB/Tests/sample.class /usr/share/sql/Tests /usr/share/sql/Tests/Test_Cautocommit.class /usr/share/sql/Tests/Test_Csavepoints.class /usr/share/sql/Tests/Test_PStimedate.class /usr/share/sql/Tests/Test_PStimezone.class /usr/share/sql/Tests/Test_Rbooleans.class /usr/share/sql/Tests/Test_Rmetadata.class /usr/share/sql/Tests/Test_Rpositioning.class /usr/share/sql/Tests/Test_Rtimedate.class /usr/share/sql/Tests/Test_Sbatching.class compilation tools that should not/never be installed: /usr/bin/burg /usr/bin/epsffit (note: this file conflicts with app-text/psutils !!!!) /usr/bin/prefixMxFile /usr/lib/autogen /usr/lib/autogen/am.py /usr/lib/autogen/autogen.py /usr/lib/autogen/codegen.py /usr/lib/autogen/msc.py /usr/lib/autogen/var.py an application which I am not sure on what it is doing in the distribution: /usr/bin/calibrator /usr/include/MonetDB/calibrator /usr/include/MonetDB/calibrator/calib.m /usr/lib/MonetDB/lib_calib.la /usr/lib/MonetDB/lib_calib.so -> lib_calib.so.0.0.0 /usr/lib/MonetDB/lib_calib.so.0 -> lib_calib.so.0.0.0 /usr/lib/MonetDB/lib_calib.so.0.0.0 is this a useful .h file to have?: /usr/include/MonetDB/monet/yytree.h is this an installer file, or a configuration file? /usr/include/sql/odbcinst.h misc: /usr/share/MonetDB/conf /usr/share/MonetDB/conf/monet.m4 /usr/share/MonetDB/general.mil ? /usr/share/MonetDB/lady.gif /usr/share/MonetDB/quit.mil ## Comment 3884 Date: 2005-09-16 23:08:12 +0200 From: @grobian Logged In: YES user_id=963970 I forgot these files which were eliminated in my install: remove Mx and mel, they are only needed for compilation rm -f ${D}/usr/bin/Mx ${D}/usr/bin/mel ${D}/usr/bin/idxmx ## Comment 3885 Date: 2005-12-19 16:29:34 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: NO TEST / PACKAGING ## Comment 3886 Date: 2006-01-24 20:26:29 +0100 From: @njnes Logged In: YES user_id=43556 The buildtools aren't distributed ie can we close this bug. ## Comment 3887 Date: 2006-01-24 20:29:36 +0100 From: @grobian Logged In: YES user_id=963970 No, it still installs: /usr/bin/epsffit and that one is the major reason this bug was created, because that file collides with psutils. ## Comment 3888 Date: 2006-01-24 21:06:04 +0100 From: @drstmane Logged In: YES user_id=572415 epsffit is in buildtools, hence not in any distribution. nevertheless, epsffit and other tools like idxmx are IMHO legacy and could be removed. ## Comment 3889 Date: 2006-01-24 21:14:02 +0100 From: @grobian Logged In: YES user_id=963970 That is quite interesting. % grep epsffit $INSTALL_DIR/filelist.* /ufs/fabian/scratch/monetdb/stable/program-x86_64/filelist.buildtools.20060122180501:ufs/fabian/scratch/monetdb/stable/program-x86_64/bin/epsffit You seem to be right. Now I'm wondering where the file comes from... ## Comment 3890 Date: 2006-01-25 09:59:52 +0100 From: @grobian Logged In: YES user_id=963970 ok, rm -f doesn't return a non-zero exit code if files are missing... that explains a lot. The update list of questions/problems then is for 4.10 and beyond: testing stuff that the user didn't ask for IMHO: /usr/bin/Mapprove.py -> Mtest.py /usr/bin/Mdiff /usr/bin/Mfilter.py /usr/bin/MkillUsers /usr/bin/Mlog /usr/bin/Mprofile.py -> Mtest.py /usr/bin/Mtest.py /usr/bin/Mtimeout /usr/bin/prof.py /usr/lib/MonetDB/Tests /usr/lib/MonetDB/Tests/sample0 /usr/lib/MonetDB/Tests/sample1 /usr/lib/MonetDB/Tests/sample2 /usr/lib/MonetDB/Tests/sample3 /usr/lib/MonetDB/Tests/sample4 /usr/lib/MonetDB/Tests/smack00 /usr/lib/MonetDB/Tests/smack01 /usr/lib/sql/Tests /usr/lib/sql/Tests/odbcsample1 /usr/share/MonetDB/Mprofile-commands.lst /usr/share/MonetDB/Tests /usr/share/MonetDB/Tests/sample.class /usr/share/sql/Tests /usr/share/sql/Tests/Test_Cautocommit.class /usr/share/sql/Tests/Test_Csavepoints.class /usr/share/sql/Tests/Test_PStimedate.class /usr/share/sql/Tests/Test_PStimezone.class /usr/share/sql/Tests/Test_Rbooleans.class /usr/share/sql/Tests/Test_Rmetadata.class /usr/share/sql/Tests/Test_Rpositioning.class /usr/share/sql/Tests/Test_Rtimedate.class /usr/share/sql/Tests/Test_Sbatching.class n application which I am not sure on what it is doing in the distribution: /usr/bin/calibrator /usr/include/MonetDB/calibrator /usr/include/MonetDB/calibrator/calib.m /usr/lib/MonetDB/lib_calib.la /usr/lib/MonetDB/lib_calib.so -> lib_calib.so.0.0.0 /usr/lib/MonetDB/lib_calib.so.0 -> lib_calib.so.0.0.0 /usr/lib/MonetDB/lib_calib.so.0.0.0 is this a useful .h file to have?: /usr/include/MonetDB/monet/yytree.h is this an installer file, or a configuration file? /usr/include/sql/odbcinst.h misc: /usr/share/MonetDB/conf /usr/share/MonetDB/conf/monet.m4 /usr/share/MonetDB/general.mil ? /usr/share/MonetDB/lady.gif /usr/share/MonetDB/quit.mil ## Comment 3891 Date: 2006-03-09 12:37:57 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 I removed epsffit from the buildtools. It's available in other packages so there is no need to include it in our source tree. Does this close the bug? ## Comment 3892 Date: 2006-03-09 14:30:44 +0100 From: @grobian Logged In: YES user_id=963970 I don't think so. There are a lot other files that I wonder of whether they should be in or not (like the yytree.h file) ## Comment 3893 Date: 2006-06-06 20:58:48 +0200 From: @grobian Logged In: YES user_id=963970 Ok. Most of the stuff is cleaned up now, I'll punt whatever I don't like in the ebuild myself. ## Comment 3894 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1293196 at http://sourceforge.net/support/tracker.php?aid=1293196 The original assignee of this bug does not have an account here. Reassigning to the default assignee for the component, bugs-common@monetdb.org. Previous assignee was nobody@users.sourceforge.net.
distribution: clashing and foreign files are installed
https://api.github.com/repos/MonetDB/MonetDB/issues/690/comments
0
2020-11-30T08:49:08Z
2024-06-28T13:34:50Z
https://github.com/MonetDB/MonetDB/issues/690
753,281,965
690
[ "MonetDB", "MonetDB" ]
Date: 2005-09-16 11:39:22 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: -- development Last updated: 2006-01-15 02:16:48 +0100 ## Comment 3869 Date: 2005-09-16 11:39:22 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; This was my first generated algebra query... First I thought some errors are not handled in the staircasejoin, but after removing a 'col_name' call (in line 74) it worked... That somehow seems *very* strange ## Comment 3870 Date: 2005-09-16 14:08:33 +0200 From: @drstmane Logged In: YES user_id=572415 Jan, I just ran your query against the latest Mserver (v4.9.3, CVS head) and it works fine for me (see attached output). Hence, could you please provide some more information about the crash, e.g., run Mserver in a debugger to locate the code where the crash occurs, and maybe why (deref. of null-pointer, stackoverflow, etc.)? ## Comment 3871 Date: 2005-09-17 23:30:59 +0200 From: @drstmane Logged In: YES user_id=572415 The crash occurs in "mallinfo", but apparently only on Jan's SuSE 9.3 (32bit) ... ??? Further investigation pending... ## Comment 3872 Date: 2005-09-19 13:03:47 +0200 From: @drstmane Logged In: YES user_id=572415 The same crash seems to occur with our nightly testing on Fedora Core 4 (32-bit) and Gentoo 1.6.13; cf. http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.32.32.d-Fedora4/tests_BugsViaSourgeforce/ID.1292727.col_name-segfault.run.out.00.html http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.32.32.d-Gentoo1.6.13/tests_BugsViaSourgeforce/ID.1292727.col_name-segfault.run.out.00.html It might hence be related to 32-bit and/or (GNU-)compiler-version and/or [g]libc-version ... ## Comment 3873 Date: 2005-11-06 20:58:25 +0100 From: @drstmane Logged In: YES user_id=572415 Jan, your MIL script does not seem to work any more with the latest version of pathfinder (CVS head), at least it now produces an empty result. Could you generate a new MIL(?) test script that reproduces the bug for you? Stefan ## Comment 3874 Date: 2005-11-07 09:30:38 +0100 From: Jan Rittinger &lt;<jan.rittinger>&gt; Logged In: YES user_id=993208 completed your attempt to apply the recent changes. It now crashes again and produces a result without col_name. ## Comment 3875 Date: 2005-11-20 18:25:15 +0100 From: @drstmane Logged In: YES user_id=572415 On none of out test machines, I can reproduce the problem; see http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests03/GNU.64.64.d-Fedora4/tests_BugsViaSourgeforce/ID.1292727.col_name-segfault.shred.out.00.html http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_BugsViaSourgeforce/ID.1292727.col_name-segfault.shred.out.00.html Jan, does the crash still occur for you, and if so, on which platform (OS +version, compiler +version, MonetDB version, pathfinder version) and under which circumstances (configure flags, within/outside Mtest, ...)? ## Comment 3876 Date: 2005-11-24 15:37:30 +0100 From: @drstmane Logged In: YES user_id=572415 As Jan just pointed out, the problems also occurs with nightly testing (again) on Fedora Core 4 (32-bit), Gentoo 1.6.13 (32-bit), and SuSE 9.3 (32-bit), cf.: http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.32.32.d-Fedora4/tests_BugsViaSourgeforce/ID.1292727.col_name-segfault.run.out.00.html http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.32.32.d-Gentoo1.6.13/tests_BugsViaSourgeforce/ID.1292727.col_name-segfault.run.out.00.html http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.32.32.d-SuSE9.3/tests_BugsViaSourgeforce/ID.1292727.col_name-segfault.run.out.00.html ## Comment 3877 Date: 2005-12-19 16:29:03 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: ALREADY IN TESTWEB: pathfinder/tests/BugsViaSourgeforce/Tests/ID.1292727.col_name-segfault.* ## Comment 3878 Date: 2005-12-30 16:15:58 +0100 From: @drstmane Logged In: YES user_id=572415 The same bug occurs (only) on the same platforms with all XMark tests when using the algebra translation; cf. e.g. http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.32.32.d-Fedora4/tests_XMark_alg/q01.out.00.html ## Comment 3879 Date: 2006-01-14 13:17:04 +0100 From: @drstmane Logged In: YES user_id=572415 The below mentioned XMark/alg tests have been "fixed" by avoising the use of "col_name()". However, the test script for http://sourceforge.net/support/tracker.php?aid=1314982 "kill(0) causes SEGFAULT" https://sourceforge.net/tracker/index.php?func=detail&aid=1314982&group_id=56967&atid=482468 now segvault on (only) the same platforms as this one; hence, they might be related... ## Comment 3880 Date: 2006-01-14 13:35:18 +0100 From: @drstmane Logged In: YES user_id=572415 The same platforms segfault also with the test of http://sourceforge.net/support/tracker.php?aid=921173 "extremely long and complex query causes SEGFAULT" https://sourceforge.net/tracker/index.php?func=detail&aid=921173&group_id=56967&atid=482468 ## Comment 3881 Date: 2006-01-15 14:16:48 +0100 From: @drstmane Logged In: YES user_id=572415 Closed as the bug is covered by this new report 1406591 "several tests cause segfault on the same subset of platforms" https://sourceforge.net/tracker/index.php?func=detail&aid=1406591&group_id=56967&atid=482468 ## Comment 3882 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1292727 at http://sourceforge.net/support/tracker.php?aid=1292727 This bug was marked DUPLICATE in the database it was moved from. Changing resolution to "MOVED"
Mserver segfault because of 'col_name'
https://api.github.com/repos/MonetDB/MonetDB/issues/689/comments
0
2020-11-30T08:49:05Z
2024-06-28T13:34:49Z
https://github.com/MonetDB/MonetDB/issues/689
753,281,928
689
[ "MonetDB", "MonetDB" ]
Date: 2005-09-16 11:20:34 +0200 From: Wouter Alink &lt;<alink>&gt; To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: -- development Last updated: 2005-09-18 11:20:43 +0200 ## Comment 3859 Date: 2005-09-16 11:20:34 +0200 From: Wouter Alink &lt;<alink>&gt; Shredding the document /ufs/alink/test-data/trecvid_bug.xml and answering the query for $d in $doc("trecvid_bug.xml") return element { "result" } { $d//VideoSegment } returns the following error: ERROR: XML Generation: idxTV, BAT[tmp]: 12228 not within [0,12227] what i discovered was that the transient WS was corrupted: PRE_SIZE didn't match PRE_LEVEL at some point. The xml-doc itself is valid xml and also firing queries NOT asking for VideoSegment-elements return correct results. Unfortunately the content of the xml-doc cannot be freely distributed, but those interested can contact me for a copy. ## Comment 3860 Date: 2005-09-16 13:51:00 +0200 From: @drstmane Logged In: YES user_id=572415 Wouter, three questions: 1) Do you refer to the CVS Head version of pathfinder (as indecated in this bug report) or to the stable 0.8.2 version resp. the XQuery_0-8 CVS branch (as you told me yesterday)? 2) Did you make sure, that it is not your code (-changes) that do corrupt the transient WS? 3) Did you try to run your Mserver with property checking enabled (i.e., Mserver --debug=10 ...)? ## Comment 3861 Date: 2005-09-16 15:23:51 +0200 From: @drstmane Logged In: YES user_id=572415 Here's a simple example: 15:22:15 manegold@corona:~/_/scratch0/Monet/Development/Current/MonetDB $ echo '<x><a><a/><a/></a></x>//a' | pf | Mserver --set monet_prompt='' Monet Database Server V4.9.3 Copyright (c) 1993-2005, CWI. All rights reserved. Compiled for x86_64-redhat-linux-gnu/64bit with 64bit OIDs; dynamically linked. Visit http://monetdb.cwi.nl/ for further information. <?xml version="1.0" encoding="utf-8"?> <XQueryResult> <a><a/><a/></a><a/><a/> </XQueryResult> 15:22:39 manegold@corona:~/_/scratch0/Monet/Development/Current/MonetDB $ echo 'element{"y"}{<x><a><a/><a/></a></x>//a}' | pf | Mserver --set monet_prompt='' Monet Database Server V4.9.3 Copyright (c) 1993-2005, CWI. All rights reserved. Compiled for x86_64-redhat-linux-gnu/64bit with 64bit OIDs; dynamically linked. Visit http://monetdb.cwi.nl/ for further information. <?xml version="1.0" encoding="utf-8"?> <XQueryResult> <y><a><a/><a/></a><a/>!ERROR: XML Generation: idxTV, BAT[tmp]: 9 not within [0,8] !ERROR: xquery_print_result_main: operation failed. 15:22:41 manegold@corona:~/_/scratch0/Monet/Development/Current/MonetDB $ ## Comment 3862 Date: 2005-09-16 18:28:12 +0200 From: @drstmane Logged In: YES user_id=572415 .. as Jan (R) already discovered, this is indeed caused a bug in the loop_lifted descendant (of self) step implementation --- it ook me a while to understand why, and what's actually going wrong --- by now, I have a vague idea, what's wrong, but I won't have time to fix it today ... ## Comment 3863 Date: 2005-09-18 19:53:45 +0200 From: @drstmane Logged In: YES user_id=572415 Fixed in CVS: loop-lifted descendant step did not work properly in some cases of overlapping subtrees in different iterations. Wouter, could you please check, whether it also works for you, now, and if so close this bug? Thanks! ## Comment 3864 Date: 2005-09-18 21:14:53 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 i'm sorry i have to disappoint you, but the fix didn't solve the bug (at least not on my system). I updated ll_staircasejoin.mx and recompiled, and the small test- script 'element { "y" } { ... }' now works ok, but the larger example (actually all four of the trecvid docs) still fail with the same error-message. Stefan, could you check this behaviour by trying the initial query of the bug report for $d in $doc("trecvid_bug.xml") return element { "result" } { $d//VideoSegment } on the trecvid data (/ufs/alink/test-data/trecvid_bug.xml)? ## Comment 3865 Date: 2005-09-18 23:08:02 +0200 From: @drstmane Logged In: YES user_id=572415 Second attempt: Now, Wouter's query doesn't produce any error anymore. However, I did not verify the (quite large) result by hand ... ## Comment 3866 Date: 2005-09-18 23:20:43 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 none of the 4 trecvid-documents produce errors anymore, but the content is not validated (a little difficult for the large documents). Stefan, do you think your second small example suffices assuming correct behaviour? i close the bug. ## Comment 3867 Date: 2005-12-19 16:28:29 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: ALREADY IN TESTWEB: pathfinder/tests/BugsViaSourgeforce/Tests/ID.1292711-1.* ## Comment 3868 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1292711 at http://sourceforge.net/support/tracker.php?aid=1292711
XQuery: incorrect WS
https://api.github.com/repos/MonetDB/MonetDB/issues/688/comments
0
2020-11-30T08:49:02Z
2024-06-27T11:06:07Z
https://github.com/MonetDB/MonetDB/issues/688
753,281,900
688
[ "MonetDB", "MonetDB" ]
Date: 2005-09-16 11:11:06 +0200 From: @drstmane To: @mlkersten Version: -- development Last updated: 2005-09-17 11:26:43 +0200 ## Comment 3855 Date: 2005-09-16 11:11:06 +0200 From: @drstmane !ERROR: interpret: no matching MIL operator to 'antijoin(BAT[int,int], BAT[int,int])'. cf. e.g. http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.64.64.d-Fedora4/tests_suite_00/test_36.out.00.html ## Comment 3856 Date: 2005-09-17 23:26:43 +0200 From: @drstmane Logged In: YES user_id=572415 Fixed: Martin added the missing MIL antijoin command. ## Comment 3857 Date: 2005-12-19 16:27:18 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: ALREADY IN TESTWEB ## Comment 3858 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1292701 at http://sourceforge.net/support/tracker.php?aid=1292701
MonetDB: new test tests/suite_00/Tests/test_36.milS fails
https://api.github.com/repos/MonetDB/MonetDB/issues/687/comments
0
2020-11-30T08:48:58Z
2024-06-27T11:06:06Z
https://github.com/MonetDB/MonetDB/issues/687
753,281,862
687
[ "MonetDB", "MonetDB" ]
Date: 2005-09-13 01:20:03 +0200 From: @drstmane To: Jan Flokstra &lt;<jflokstra>&gt; Version: -- development Last updated: 2005-09-13 04:40:00 +0200 ## Comment 3851 Date: 2005-09-13 13:20:03 +0200 From: @drstmane In case an error occurs while parsing an XML document, the error handler ("print_error()") that is registered via "xmlSetGenericErrorFunc()" in "untime/shredder.mx" seems to be called incorrectly, i.e., not with a variable argument list in one call, but with an extra call per argument!?? (see trace below for details) In case the argument (i.e., the error message) contains a format string, the respective arguments hence seem to be missing in the same call, causing at least "strange" error messages (arguments are filled in by random data from the stack), or even a crash. My knowledge of libxml2 seems to be too limited to analyse and potentially fix this bug in reasonable time. I'm actually not sure, whether the problem is libxml2 inherent, caused by a wrong definition or implementation of the error handler ("print_error()"), caused by an incorrect/insufficient registration of this error handler with libxml2, or caused by an incorrect/insufficient initialization/usage of libxml2's XML parser. The problem occurs at least with libxml2 2.6.20 on Fedora Core 4 (64-bit) and libxml2 2.6.17 on SuSE 9.3 (32-bit). Jan (F.), assuming that your libxml2 experience is more than mine (which is none), would you have time to have a look at this problem? Of course, any other libxml2 expert's opinion is also more than welcome! ======== 12:48:20 manegold@corona:~/_/scratch0/Monet/Development/Current/pathfinder $ l /tmp/foo.xml ls: /tmp/foo.xml: No such file or directory 12:48:21 manegold@corona:~/_/scratch0/Monet/Development/Current/pathfinder $ echo 'shred_doc("/tmp/foo.xml","foo.xml");' | Mserver --dbinit='module(pathfinder);' --set monet_prompt='' Monet Database Server V4.9.3 Copyright (c) 1993-2005, CWI. All rights reserved. Compiled for x86_64-redhat-linux-gnu/64bit with 64bit OIDs; dynamically linked. Visit http://monetdb.cwi.nl/ for further information. !ERROR: I/O !ERROR: warning : !ERROR: !ERROR: shred:mainSHREDDER: xmlCreateFileParserCtxt("/tmp/foo.xml") failed !ERROR: CMDshred2bats: operation failed. 12:48:30 manegold@corona:~/_/scratch0/Monet/Development/Current/pathfinder $ gdb --args Mserver --dbinit='module(pathfinder);' GNU gdb Red Hat Linux (6.3.0.0-1.21rh) Copyright 2004 Free Software Foundation, Inc. GDB is free software, covered by the GNU General Public License, and you are welcome to change it and/or distribute copies of it under certain conditions. Type "show copying" to see the conditions. There is absolutely no warranty for GDB. Type "show warranty" for details. This GDB was configured as "x86_64-redhat-linux-gnu"...Using host libthread_db library "/lib64/libthread_db.so.1". (gdb) break print_error Function "print_error" not defined. Make breakpoint pending on future shared library load? (y or [n]) y Breakpoint 1 (print_error) pending. (gdb) r Starting program: /net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Development/Current/.GNU.64.64.d..PREFIX./bin/Mserver --dbinit=module\(pathfinder\)\;[Thread debugging using libthread_db enabled] [New Thread 46912515620032 (LWP 6809)] [New Thread 1082132832 (LWP 6845)] Monet Database Server V4.9.3 Copyright (c) 1993-2005, CWI. All rights reserved. Compiled for x86_64-redhat-linux-gnu/64bit with 64bit OIDs; dynamically linked. Visit http://monetdb.cwi.nl/ for further information. Breakpoint 2 at 0x2aaab5076ae0: file /ufs/manegold/_/scratch0/Monet/Development/Current/pathfinder/runtime/shredder.mx, line 1380. Pending breakpoint "print_error" resolved MonetDB>shred_doc("/tmp/foo.xml","foo.xml"); [Switching to Thread 46912515620032 (LWP 6809)] Breakpoint 2, print_error (ctx=0x0, msg=0x3ff07deb4d "I/O ") at /ufs/manegold/_/scratch0/Monet/Development/Current/pathfinder/runtime/shredder.mx:1380 1380 { (gdb) cont Continuing. !ERROR: I/O Breakpoint 2, print_error (ctx=0x0, msg=0x3ff07debce "warning : ") at /ufs/manegold/_/scratch0/Monet/Development/Current/pathfinder/runtime/shredder.mx:1380 1380 { (gdb) cont Continuing. !ERROR: warning : Breakpoint 2, print_error (ctx=0x0, msg=0x3ff07e8bee "%s") at /ufs/manegold/_/scratch0/Monet/Development/Current/pathfinder/runtime/shredder.mx:1380 1380 { (gdb) cont Continuing. !ERROR: !ERROR: shred:mainSHREDDER: xmlCreateFileParserCtxt("/tmp/foo.xml") failed !ERROR: CMDshred2bats: operation failed. MonetDB> Program exited normally. (gdb) 12:49:20 manegold@corona:~/_/scratch0/Monet/Development/Current/pathfinder $ 12:49:21 manegold@corona:~/_/scratch0/Monet/Development/Current/pathfinder $ ## Comment 3852 Date: 2005-09-13 16:40:00 +0200 From: Jan Flokstra &lt;<jflokstra>&gt; Logged In: YES user_id=1054297 Retrieved original libxml2-error.c from net for correct example. Problems were: - we should use vfprintf()/vsnprintf() for correct argument handling. - function may be called more than once for 1 error message so we cannot use GDKerror() here. ## Comment 3853 Date: 2005-12-19 16:35:08 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: NO TEST / REDUNDANT: shred_doc error messages are covered by test pathfinder/tests/BugDay_2005-12-19_0.9.3/Tests/shred_doc_with_space.SF-1211372.* ## Comment 3854 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1289651 at http://sourceforge.net/support/tracker.php?aid=1289651
XQuery: error handler incorrectly called from libxml2??
https://api.github.com/repos/MonetDB/MonetDB/issues/686/comments
0
2020-11-30T08:48:55Z
2024-06-27T11:06:05Z
https://github.com/MonetDB/MonetDB/issues/686
753,281,826
686
[ "MonetDB", "MonetDB" ]
Date: 2005-09-09 03:10:39 +0200 From: @drstmane To: Fabian Groffen &lt;<fabian>&gt; Version: -- development Last updated: 2005-09-13 01:05:00 +0200 ## Comment 3842 Date: 2005-09-09 15:10:39 +0200 From: @drstmane After Peter's "noroot" changes in runtime/serialize.mx (revision 1.32), test runtime/Tests/text1.milS ======== module(pathfinder); xquery("dm", "<a>aap</a>/text()").printf(); xquery("dm", "<a>2</a>/text()").printf(); xquery("dm-mapi", "<a>aap</a>/text()").printf(); xquery("dm-mapi", "<a>2</a>/text()").printf(); xquery("xml-seq-mapi", "<a>aap</a>/text()").printf(); xquery("xml-seq-mapi", "<a>2</a>/text()").printf(); xquery("xml-mapi", "<a>aap</a>/text()").printf(); xquery("xml-mapi", "<a>2</a>/text()").printf(); ======== shows the following differences compared with the "stable" output: ======== --- ../.GNU.64.64.d..PREFIX./mTests.pathfinder.43/runtime/text1.stable.out 2005-08-21 13:17:37.000000000 +0200 +++ ../.GNU.64.64.d..PREFIX./mTests.pathfinder.43/runtime/text1.test.out 2005-09-09 14:38:47.000000000 +0200 @@ -30,7 +30,8 @@ 60 length 1 tuplecount 0 id -[ "text {\"aap\"}" ] +> +=[ "aap" ] xquery("xml-seq-mapi", "<a>2</a>/text()").printf(); - 3 querytype @@ -40,7 +41,8 @@ 60 length 1 tuplecount 1 id -[ "text {\"2\"}" ] +> +=[ "2" ] xquery("xml-mapi", "<a>aap</a>/text()").printf(); =<?xml version="1.0" encoding="utf-8"?> ======== I have to admit, I'm not sure, whether this is a bug, whether the original "stable" output is not correct, or whether the output changes was intended. Peter, could you (or someone else) please have a look at this? Thanks in advance! ## Comment 3843 Date: 2005-09-09 15:21:54 +0200 From: @grobian Logged In: YES user_id=963970 > @@ -30,7 +30,8 @@ > 60 length > 1 tuplecount > 0 id > -[ "text {\"aap\"}" ] > +> > +=[ "aap" ] This is wrong in any case (the '>' and '=' certainly don't belong there, hence XML:DB is broken. I remember from a discussion with Wouter that the output as it was (text {"aap"}) was the desired one. > xquery("xml-seq-mapi", "<a>2</a>/text()").printf(); > - > 3 querytype > @@ -40,7 +41,8 @@ > 60 length > 1 tuplecount > 1 id > -[ "text {\"2\"}" ] > +> > +=[ "2" ] again the '>' and '=' don't belong here and break XML:DB. This bug might be fixable in serialize.mx. I guess they are caused by Peter's change/decision to dismiss the modes somehow. I don't know since when this bug appeared. It is obvious to me that there is a mode missing somehow because the '=' indicates it is in another mode when printing the results. ## Comment 3844 Date: 2005-09-09 15:52:13 +0200 From: @drstmane Logged In: YES user_id=572415 The bug occurs since Peter's "noroot" changes in runtime/serialize.mx (revision 1.32) on Sunday Sep 04 2005, and might be releated to his "noroot" changes in compiler/compile.c (revision 1.40), compiler/include/pathfinder.h (revision 1.34), & compiler/mil/milprint_summer.c (revision 1.91) on Friday Sep 02 2005. ## Comment 3845 Date: 2005-09-13 09:45:15 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 the following fails as well: xquery: "aap", 1+1 using: MapiClient -lx -sxml gives: <XQueryResult> "aap" 2 </XQueryResult> In this example I would think that the double quotes should not be there (the answer is not even valid XML). I guess it has something to do with the same checkin. ## Comment 3846 Date: 2005-09-13 10:28:06 +0200 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 Wouter, I cannot see any problem in your example. The output is of course valid XML. There is no need to escape quotes in text content. You only need to escape them if you want to have quotes within quotes (within attribute values). The quotes appear in the serialization to denote that the output is actually a string. Note that the query "foo" is NOT the same as text { "foo" } ! The former returns a string, while the latter returns a text node. This is why the two are serialized differently. Of course, there is the ever-lasting discussion on how serialization should be performed, and you can argue for the one or the other solution. The output for your example appears to be consistent and correct for me, though. ## Comment 3847 Date: 2005-09-13 11:02:20 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 I apologize for claiming it to be non-xml, but on the XQuery website the following example is given: ---- <b>{1, "2", "3"}</b> The constructed element node has one child, a text node containing the value "1 2 3". ---- And because we do add a <XQueryResult/>-element around the answer, i would expect, correct me if i'm wrong, that the string-node would become a text-node. Of course this behaviour would only happen in case the 'xml'-serialization method is used. ## Comment 3848 Date: 2005-09-13 13:05:00 +0200 From: @grobian Logged In: YES user_id=963970 Fixed in CVS. Please continue (restart) discussion on the representation of multiple resource sets in a separate bug. ## Comment 3849 Date: 2005-12-19 16:22:56 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: ALREADY IN TESTWEB ## Comment 3850 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1286005 at http://sourceforge.net/support/tracker.php?aid=1286005
XQuery: test runtime/Tests/text1.milS failes
https://api.github.com/repos/MonetDB/MonetDB/issues/685/comments
0
2020-11-30T08:48:52Z
2024-06-27T11:06:04Z
https://github.com/MonetDB/MonetDB/issues/685
753,281,797
685
[ "MonetDB", "MonetDB" ]
Date: 2005-09-09 02:30:26 +0200 From: Jens Teubner &lt;<teubner>&gt; To: Peter Boncz &lt;<boncz>&gt; Version: -- development Last updated: 2005-11-20 07:01:56 +0100 ## Comment 3837 Date: 2005-09-09 14:30:26 +0200 From: Jens Teubner &lt;<teubner>&gt; Since Peter's parser modifications, one of the parser test cases fails. I actually have a slight preference to use the XQuery module syntax to declare functions for the SOAP stuff. (No full implementation of the module functionality, but just use it to declare your SOAP functions.) That may, by the way, would also allow you to omit the query part in the MIL code altogether and just print the PROCs that you need, no need to post-process anything. As (fortunately) your modifications to the parser are really small and should not have any negative effect on the remaining parsing rules, I could also live with the current solution/hack. Least thing to do then, however, is to fix up the test cases. ## Comment 3838 Date: 2005-11-20 18:30:27 +0100 From: @drstmane Logged In: YES user_id=572415 According to the TestWeb, this test is working fine, again. Peter, Jens, could you please check/confirm this, and in case you agree, mark this report "fixed" and close it. Thanks, Stefan ## Comment 3839 Date: 2005-11-20 19:01:56 +0100 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 SOAP stuff has been implemented the "proper" way in the meantime, and Pathfinder now supports modules to some degree. Peter's parser hack is no longer neccessary since then, and the problem is fixed again. ## Comment 3840 Date: 2005-12-19 16:22:38 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: ALREADY IN TESTWEB ## Comment 3841 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1285957 at http://sourceforge.net/support/tracker.php?aid=1285957
XQuery: test compiler/parser/bad-006.xq fails
https://api.github.com/repos/MonetDB/MonetDB/issues/684/comments
0
2020-11-30T08:48:50Z
2024-06-27T11:06:03Z
https://github.com/MonetDB/MonetDB/issues/684
753,281,756
684
[ "MonetDB", "MonetDB" ]
Date: 2005-09-08 01:35:36 +0200 From: mihalych &lt;<mihalych_>&gt; To: @njnes Version: -- development Last updated: 2005-11-26 09:00:38 +0100 ## Comment 3833 Date: 2005-09-08 13:35:36 +0200 From: mihalych &lt;<mihalych_>&gt; monetdb-> create table t1 (id1 int, val1 varchar(255)); Operation successful monetdb-> create table t2 (id2 int, val2 varchar(255)); Operation successful monetdb-> insert into t1 values (1,'1'); 1 affected row monetdb-> insert into t2 values (1,'2'); 1 affected row monetdb-> update t1 set val1 = (select val2 from t2 where id1 = id2) where id1 in (select id2 from t2); Error: ERROR: interpret: unknown variable 's23'. ERROR: interpret_params: join(param 2): evaluation error. ## Comment 3834 Date: 2005-11-26 09:00:38 +0100 From: @njnes Logged In: YES user_id=43556 The update code now handles this case, ie the update statement is now a lot more powerfull (not allways set = 'value'). Test script is added to src/test/BugTracker/Tests/update_with_correlated_subselect.SF-1284791.sql ## Comment 3835 Date: 2005-12-19 16:22:15 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: ALREADY IN TESTWEB: sql/src/test/BugTracker/Tests/update_with_correlated_subselect.SF-1284791.* ## Comment 3836 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1284791 at http://sourceforge.net/support/tracker.php?aid=1284791
SQL: Update with correlated subselect
https://api.github.com/repos/MonetDB/MonetDB/issues/683/comments
0
2020-11-30T08:48:47Z
2024-06-27T11:06:02Z
https://github.com/MonetDB/MonetDB/issues/683
753,281,733
683
[ "MonetDB", "MonetDB" ]
Date: 2005-09-07 05:21:11 +0200 From: Wouter Alink &lt;<alink>&gt; To: XQuery devs &lt;<bugs-xquery>&gt; Version: -- development Last updated: 2005-11-25 12:51:58 +0100 ## Comment 3826 Date: 2005-09-07 17:21:11 +0200 From: Wouter Alink &lt;<alink>&gt; (100 div 7) mod 5 gives the following error: ERROR: interpret: no matching MIL operator to '[%](BAT [void,dbl], BAT[void,dbl])'. My interpretation of the XQuery specs would be that we need a new MIL operator: [%](BAT[void,dbl], BAT[void,dbl]) because the following is being said on the w3c website: (by the way, the definition doesn't mention if it is allowed for A and B to have different types, but i guess not, as the last sentence identifies 'four' different operators, am i right?) ----------------- A div B A of type numeric B of type numeric op:numeric-divide(A, B) result of type: numeric; but xs:decimal if both operands are xs:integer A mod B A of type numeric B of type numeric op:numeric-mod(A, B) result of type: numeric [Definition: When referring to a type, the term numeric denotes the types xs:integer, xs:decimal, xs:float, and xs:double.] An operator whose operands and result are designated as numeric might be thought of as representing four operators, one for each of the numeric types. ## Comment 3827 Date: 2005-09-16 14:41:49 +0200 From: @drstmane Logged In: YES user_id=572415 Wouter, to be honest, I don't know how '%' (mod) is/should be defined on double (i.e., floating point numbers, or even non-integers in general). What's the result of 5.1 mod 4.2? 0.9? (Then 5.1 div 4.2 = 1.0 while 5.1/4.2 = 1.214285714?) What do the XQuery specs say about this? ## Comment 3828 Date: 2005-09-16 15:53:27 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 http://www.w3.org/TR/xquery-operators/ Section 6.2.6 gives a detailed description of the mod-operator. ## Comment 3829 Date: 2005-09-19 09:49:26 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; Logged In: YES user_id=993208 the current version of 'mod' just uses the '%' operator in MIL. Because it is probably rather hard to find a sensitive solution in MIL, I would call it an un unsupported feature... (& it should be mentioned in the KnownIssues file then) ## Comment 3830 Date: 2005-11-25 00:51:58 +0100 From: @peterboncz Logged In: YES user_id=591107 should work now ## Comment 3831 Date: 2005-12-19 16:21:40 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: TEST ADDED / SUCCESS: pathfinder/tests/BugDay_2005-12-19_0.9.3/Tests/div_mod.SF-1284049.* ## Comment 3832 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1284049 at http://sourceforge.net/support/tracker.php?aid=1284049 The original assignee of this bug does not have an account here. Reassigning to the default assignee for the component, bugs-xquery@monetdb.org. Previous assignee was nobody@users.sourceforge.net.
XQuery: (100 div 7) mod 5
https://api.github.com/repos/MonetDB/MonetDB/issues/682/comments
0
2020-11-30T08:48:43Z
2024-06-27T11:06:01Z
https://github.com/MonetDB/MonetDB/issues/682
753,281,691
682
[ "MonetDB", "MonetDB" ]
Date: 2005-09-07 01:33:27 +0200 From: @MarcinZukowski To: GDK devs &lt;<bugs-common>&gt; Version: -- development Last updated: 2005-09-17 11:29:22 +0200 ## Comment 3821 Date: 2005-09-07 13:33:27 +0200 From: @MarcinZukowski This script behaves incorrectly: var b:=new(oid,oid); b.insert(0@0,0@0); b.insert(1@0,1@0); b.insert(10@0,10@0); b.info().find("hsorted").print(); [ "65" ] ok b.info().find("tsorted").print(); [ "65" ] ok var m:=b.merge(); m.info().find("hsorted").print(); [ "65" ] ok m.info().find("tsorted").print(); [ "65" ] ok b.insert(5@0,0@0); b.info().find("hsorted").print(); [ "0" ] ok b.info().find("tsorted").print(); [ "0" ] ok var m:=b.merge(); m.info().find("hsorted").print(); [ "65" ] NOT OK! m.info().find("tsorted").print(); [ "65" ] NOT OK! var s:=m.split(); m.info().find("hsorted").print(); [ "65" ] NOT OK! m.info().find("tsorted").print(); [ "65" ] NOT OK! BATs returned by merge() are incorrectly marked sorted, causing the same data after split() to be also marked sorted. With -d10 Monet naturally complains, and that breaks amdb testweb :) ## Comment 3822 Date: 2005-09-07 13:57:18 +0200 From: @MarcinZukowski Logged In: YES user_id=607094 BTW, merge() in general doesn't have to work with 64-bit oids... ## Comment 3823 Date: 2005-09-17 23:29:22 +0200 From: @drstmane Logged In: YES user_id=572415 Fixed in CVS by Marcin (and me). ## Comment 3824 Date: 2005-12-19 16:11:55 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: TEST ADDED / SUCCESS: MonetDB/tests/BugDay_2005-12-19_4.9.3/Tests/merge_properties.SF-1283842.* ## Comment 3825 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1283842 at http://sourceforge.net/support/tracker.php?aid=1283842 The original assignee of this bug does not have an account here. Reassigning to the default assignee for the component, bugs-common@monetdb.org. Previous assignee was nobody@users.sourceforge.net.
merge() incorrectly marks BATs as sorted
https://api.github.com/repos/MonetDB/MonetDB/issues/681/comments
0
2020-11-30T08:48:40Z
2024-06-27T11:06:00Z
https://github.com/MonetDB/MonetDB/issues/681
753,281,657
681
[ "MonetDB", "MonetDB" ]
Date: 2005-09-05 11:25:39 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-10-05 10:42:12 +0200 ## Comment 3813 Date: 2005-09-05 11:25:39 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Test case: sql>select 2 * interval '1' second; [ 2 ] works, but sql>select interval '1' second * 2; !ERROR: second_interval_sht: operation failed. doesn't work. ## Comment 3814 Date: 2005-09-05 11:42:41 +0200 From: @grobian Logged In: YES user_id=963970 according to the big SQL book this should be: SELECT INTERVAL SECOND(1) * 2; (which should yield in an INTERVAL) ## Comment 3815 Date: 2005-09-05 12:31:50 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 Well, I can only use unofficial sources, e.g. drafts or: <http://savage.net.au/SQL/sql-2003-2.bnf.htmlinterval% 20value%20expression> This would indicate, that there's nothing special about <interval term 2>, i.e. my <interval literal> would be allowed. Out of curiosity, what's the 'big SQL book'? ## Comment 3816 Date: 2005-09-05 14:22:52 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 If you search at http://print.google.com for "you can also multiply or divide an interval", you'll find Melton/Simon: SQL: 1999, p. 158: INTERVAL ‘2' DAY * 3 -- results in an interval of 6 days ## Comment 3817 Date: 2005-09-05 23:24:25 +0200 From: @njnes Logged In: YES user_id=43556 added missing *,+,/,- functions on intervals ## Comment 3818 Date: 2005-09-06 09:46:43 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 Currently, interval <op> interval no longer works: sql>select interval '1' second + interval '1' second; !ERROR: interpret: no matching MIL operator to '+(lng, oid)'. ## Comment 3819 Date: 2005-12-19 16:28:52 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS sql/src/test/BugDay_2005-12-19_2.9.3/Tests/interval_times_int.SF-1281996.sql ## Comment 3820 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1281996 at http://sourceforge.net/support/tracker.php?aid=1281996
SQL: select interval '1' second * 2
https://api.github.com/repos/MonetDB/MonetDB/issues/680/comments
0
2020-11-30T08:48:37Z
2024-06-27T11:05:59Z
https://github.com/MonetDB/MonetDB/issues/680
753,281,621
680
[ "MonetDB", "MonetDB" ]
Date: 2005-09-03 11:44:11 +0200 From: Jens Teubner &lt;<teubner>&gt; To: Jens Teubner &lt;<teubner>&gt; Version: -- development Last updated: 2005-09-12 01:23:09 +0200 ## Comment 3809 Date: 2005-09-03 11:44:11 +0200 From: Jens Teubner &lt;<teubner>&gt; The let $a := doc("foo.xml") cast as xs:anyNode return $a/bar returns a type error within Jan's "simplifier" in milprint_summer.c: type error: can't cast type 'untypedAtomic?' to type 'node' The original Core tree is apparently type correct, while Jan's simplifier introduces some type error. ## Comment 3810 Date: 2005-09-12 13:23:09 +0200 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 As Jan noticed correctly, this query is actually illegal. `cast as' only allows casting to atomic types (no node types). I've just committed a fix for the typechecker, which now ensures that a cast's target type is always a subtype of atomic?. ## Comment 3811 Date: 2005-12-19 16:26:24 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS pathfinder/tests/BugDay_2005-12-19_0.9.3/Tests/type_error_in_cast.SF-1281176.xq ## Comment 3812 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1281176 at http://sourceforge.net/support/tracker.php?aid=1281176
Core simplification in milprint_summer.c leads to type error
https://api.github.com/repos/MonetDB/MonetDB/issues/679/comments
0
2020-11-30T08:48:34Z
2024-06-27T11:05:58Z
https://github.com/MonetDB/MonetDB/issues/679
753,281,587
679
[ "MonetDB", "MonetDB" ]
Date: 2005-09-02 06:25:14 +0200 From: @drstmane To: @sjoerdmullender Version: -- development Last updated: 2005-09-05 12:49:40 +0200 ## Comment 3805 Date: 2005-09-02 18:25:14 +0200 From: @drstmane When using a multiplex operation (e.g., "[+]") with void-headed synchronized/aligned BATs that have var-sized tail values, the dense (non-nil) void-heads are (IMHO unnecessarily) materialized to oid. The reason seems to be, that "array_opt" in struct "monet_sig_t" is set to FALSE for var-sized atoms (cf. function "TBL_multiplex_resolve" in "src/monet/monet_tbl.mx": line 2307: "sig->array_opt = sig->merge_opt = 0;", line 2436" "sig->array_opt = !ATOMvarsized(tpe);"), and hence, the "array" implementation in "src/monet/monet_multiplex.mx" (functions "interpret_{special,multiplex}_array") is NOT used. Could someone please check, whether this is indeed necessary? There seems to be code in "interpret_{special,multiplex}_array" that can handle var-sized atoms (cf. marco "FIRST_PARAM_FIXEDSIZE" in "src/monet/monet_multiplex.mx"). Here's a small experiment that depicts the problem: ======== $ cat /tmp/m.mil -------- var i := new(void,int).seqbase(0@0).insert(nil,1).access(BAT_READ); var s := [str](i); print(i,s); print(i.[+](1)); print(i.[+](i)); print(s.[+]("1")); print(s.[+](s)); quit(); ======== $ cat /tmp/m.mil | Mserver --trace --set monet_prompt="`echo -e '\n-> '`" -------- Monet Database Server V4.8.3 Copyright (c) 1993-2005, CWI. All rights reserved. Compiled for x86_64-redhat-linux-gnu/64bit with 64bit OIDs; dynamically linked. Visit http://monetdb.cwi.nl/ for further information. -> var i := new(void,int).seqbase(0@0).insert(nil,1).access(BAT_READ); -> var s := [str](i); -> print(i,s); ------------------------- h t t name void int str type ------------------------- [ 0@0, 1, "1" ] -> print(i.[+](1)); interpret_multiplex interpret_multiplex: min_size == REALLY_REALLY_BIG => tpe=BAThtype(b)=6 (b->htype=0) interpret_multiplex: sig->pack_fcn=0x2aaaaada76c7, sig->fast_fcn=0x2aaaaadb404b, sig->special=0, sig->array_opt=1, all_synced=0x730fd8, sig->merge_opt=0, all_sorted=0 interpret_multiplex: sig->fast_fcn != NULL && sig->array_opt && all_synced => interpret_multiplex_array interpret_multiplex_array ----------------- h t name void int type ----------------- [ 0@0, 2 ] -> print(i.[+](i)); interpret_multiplex interpret_multiplex: min_size == REALLY_REALLY_BIG => tpe=BAThtype(b)=6 (b->htype=0) interpret_multiplex: sig->pack_fcn=0x2aaaaada76c7, sig->fast_fcn=0x2aaaaadb404b, sig->special=0, sig->array_opt=1, all_synced=0x730fd8, sig->merge_opt=1, all_sorted=0 interpret_multiplex: sig->fast_fcn != NULL && sig->array_opt && all_synced => interpret_multiplex_array interpret_multiplex_array ----------------- h t name void int type ----------------- [ 0@0, 2 ] -> print(s.[+]("1")); interpret_multiplex interpret_multiplex: min_size == REALLY_REALLY_BIG => tpe=BAThtype(b)=6 (b->htype=0) interpret_multiplex: sig->pack_fcn=0x2aaaab10e443, sig->fast_fcn=0x2aaaab10ff79, sig->special=0, sig->array_opt=0, all_synced=0x7ae2a8, sig->merge_opt=0, all_sorted=0 interpret_multiplex: fall-through => 'durchfall' implementation => interpret_multiplex_fcn interpret_multiplex_fcn interpret_multiplex_fcn: argv[i].vtype == TYPE_bat && argv[i].len != int_nil => real_type = b->htype = 0 interpret_multiplex_fcn: desc.nbats == 1 && real_type == TYPE_void => desc.id.vtype = real_type = 0 ----------------- h t name void str type ----------------- [ 0@0, "11" ] -> print(s.[+](s)); interpret_multiplex interpret_multiplex: min_size == REALLY_REALLY_BIG => tpe=BAThtype(b)=6 (b->htype=0) interpret_multiplex: sig->pack_fcn=0x2aaaab10e443, sig->fast_fcn=0x2aaaab10ff79, sig->special=0, sig->array_opt=0, all_synced=0x7ae2a8, sig->merge_opt=0, all_sorted=0 interpret_multiplex: fall-through => 'durchfall' implementation => interpret_multiplex_fcn interpret_multiplex_fcn interpret_multiplex_fcn: argv[i].vtype == TYPE_bat && argv[i].len != int_nil => real_type = b->htype = 0 interpret_multiplex_fcn: argv[i].vtype == TYPE_bat && argv[i].len != int_nil => real_type = b->htype = 0 interpret_multiplex_fcn: desc.nbats == 2, real_type == 0 => desc.id.vtype = tpe = 6 ----------------- h t name oid str type ----------------- [ 0@0, "11" ] -> quit(); ======== $ cvs diff -u src/monet/monet_multiplex.mx -------- Index: src/monet/monet_multiplex.mx =================================================================== RCS file: /cvsroot/monetdb/MonetDB/src/monet/monet_multiplex.mx,v retrieving revision 1.41.2.2 diff -u -u -r1.41.2.2 monet_multiplex.mx --- src/monet/monet_multiplex.mx 23 Aug 2005 17:20:47 -0000 1.41.2.2 +++ src/monet/monet_multiplex.mx 2 Sep 2005 16:09:22 -0000 @@ -201,6 +201,8 @@ size_t min_size = REALLY_REALLY_BIG; char name[128]; + THRprintf(GDKout, "interpret_multiplex\n"); + strcpy(name, s); if (sig->assignop) { BBPfix(argv[1].val.bval); @@ -221,6 +223,7 @@ if (min_size == REALLY_REALLY_BIG) { all_synced = b; tpe = BAThtype(b); + THRprintf(GDKout, "interpret_multiplex: min_size == REALLY_REALLY_BIG => tpe=BAThtype(b)=%d (b->htype=%d)\n", tpe, b->htype); if (sig->assignop && b->batRestricted) { GDKerror("interpret_multiplex: %s: param 1 must be writeable.\n", name); return -2; @@ -276,25 +279,34 @@ @- execution @c + THRprintf(GDKout, "interpret_multiplex: " + "sig->pack_fcn="PTRFMT", sig->fast_fcn="PTRFMT", sig->special=%d, sig->array_opt=%d, all_synced="PTRFMT", sig->merge_opt=%d, all_sorted=%d\n", + PTRFMTCAST sig->pack_fcn, PTRFMTCAST sig->fast_fcn, (int)sig->special, (int)sig->array_opt, PTRFMTCAST all_synced, (int)sig->merge_opt, (int)all_sorted); if ((GDKfcn) sig->pack_fcn == sig->fast_fcn) { /* TOK_PROC */ + THRprintf(GDKout, "interpret_multiplex: (GDKfcn) sig->pack_fcn == sig->fast_fcn => interpret_multiplex_fcn\n"); return interpret_multiplex_fcn(name, argc, argv, res, sig->assignop, stk, min_size, tpe, sig->ret_tpe, (GDKfcn) t); } else if (sig->special) { if (sig->array_opt && all_synced) { + THRprintf(GDKout, "interpret_multiplex: sig->special && sig->array_opt && all_synced => interpret_special_array\n"); return interpret_special_array(name, argc, argv, res, sig->fast_fcn, min_size, sig->ret_tpe, sig->assignop, sig->special); } if (sig->merge_opt && all_sorted) { + THRprintf(GDKout, "interpret_multiplex: sig->special && sig->merge_opt && all_sorted => interpret_special_merge\n"); return interpret_special_merge(name, argc, argv, res, sig->fast_fcn, min_size, sig->ret_tpe, sig->assignop, sig->special); } } else if (sig->fast_fcn != NULL) { if (sig->array_opt && all_synced) { + THRprintf(GDKout, "interpret_multiplex: sig->fast_fcn != NULL && sig->array_opt && all_synced => interpret_multiplex_array\n"); return interpret_multiplex_array(name, argc, argv, res, sig->fast_fcn, min_size, sig->ret_tpe, sig->assignop, FALSE); } if (sig->merge_opt && all_sorted) { + THRprintf(GDKout, "interpret_multiplex: sig->fast_fcn != NULL && sig->merge_opt && all_sorted => interpret_multiplex_merge\n"); return interpret_multiplex_merge(name, argc, argv, res, sig->fast_fcn, min_size, sig->ret_tpe, sig->assignop, FALSE); } } { /* fall-through => 'durchfall' implementation */ + THRprintf(GDKout, "interpret_multiplex: fall-through => 'durchfall' implementation => interpret_multiplex_fcn\n"); return interpret_multiplex_fcn(name, argc, argv, res, sig->assignop, -1, min_size, tpe, sig->ret_tpe, sig->pack_fcn); } } @@ -416,6 +428,8 @@ exec_t desc; RowFcn wrap; + THRprintf(GDKout, "interpret_multiplex_fcn\n"); + memset(&desc, 0, sizeof(exec_t)); desc.name = s; desc.fcn = imp; @@ -434,6 +448,7 @@ if (b == NULL) return 0; real_type = b->htype; + THRprintf(GDKout, "interpret_multiplex_fcn: argv[i].vtype == TYPE_bat && argv[i].len != int_nil => real_type = b->htype = %d\n", real_type); desc.piv[desc.nbats] = b; desc.c[++desc.nbats] = desc.argv + i; } else if (argv[i].vtype == TYPE_bat) { @@ -450,8 +465,10 @@ desc.c[i + 1]->len = ATOMsize(desc.c[i + 1]->vtype); } if (desc.nbats == 1 && real_type == TYPE_void) { + THRprintf(GDKout, "interpret_multiplex_fcn: desc.nbats == 1 && real_type == TYPE_void => desc.id.vtype = real_type = %d\n", real_type); desc.id.vtype = real_type; } else { + THRprintf(GDKout, "interpret_multiplex_fcn: desc.nbats == %d, real_type == %d => desc.id.vtype = tpe = %d\n", desc.nbats, real_type, tpe); desc.id.vtype = tpe; } desc.id.len = ATOMsize(tpe); @@ -785,6 +802,8 @@ bulkerror_t bulkerror; BUN dst; + THRprintf(GDKout, "interpret_@1_array\n"); + /* keep compilers happy */ ah = 0; (void) ah; hdff = 0; (void) hdff; @@ -1047,6 +1066,8 @@ bulkerror_t bulkerror; BAT* retb; + THRprintf(GDKout, "interpret_@1_merge\n"); + @:mergeinit(1)@ @:mergeinit(2)@ if (argc==4) ======== ## Comment 3806 Date: 2005-09-05 12:49:40 +0200 From: @sjoerdmullender Logged In: YES user_id=43607 Fixed in a different way. In monet_multiplex, check whether all input BATs are synchronized (this was already done), and whether there is a dense input BAT (any one will do). If that is the case, the result BAT can be VOID-headed. ## Comment 3807 Date: 2005-12-19 16:21:01 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS MonetDB/tests/BugDay_2005-12-19_4.9.3/Tests/void_multiplex.SF-1280802.milS ## Comment 3808 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1280802 at http://sourceforge.net/support/tracker.php?aid=1280802
multiplex fails to maintain void-heads with var-sized tails
https://api.github.com/repos/MonetDB/MonetDB/issues/678/comments
0
2020-11-30T08:48:31Z
2024-06-27T11:05:57Z
https://github.com/MonetDB/MonetDB/issues/678
753,281,548
678
[ "MonetDB", "MonetDB" ]
Date: 2005-09-02 04:18:54 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-10-05 10:55:28 +0200 ## Comment 3801 Date: 2005-09-02 16:18:54 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Test case: sql>select cast( 1 as interval year ); !ERROR: interpret: no matching MIL operator to 'month_interval(sht, int, int)'. !MAYBE YOU MEAN: ! month_interval(str, int, int) : int sql>select cast( 1 as interval day ); !ERROR: interpret: no matching MIL operator to 'second_interval(sht, int, int)'. !MAYBE YOU MEAN: ! second_interval(str, int, int) : lng ## Comment 3802 Date: 2005-09-05 23:23:25 +0200 From: @njnes Logged In: YES user_id=43556 added missing conversion functions ## Comment 3803 Date: 2005-12-19 16:15:53 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS sql/src/test/BugDay_2005-12-19_2.9.3/Tests/cast_interval.SF.1280682.sql ## Comment 3804 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1280682 at http://sourceforge.net/support/tracker.php?aid=1280682
SQL: select cast( 1 as interval ...
https://api.github.com/repos/MonetDB/MonetDB/issues/677/comments
0
2020-11-30T08:48:27Z
2024-06-27T11:05:56Z
https://github.com/MonetDB/MonetDB/issues/677
753,281,508
677
[ "MonetDB", "MonetDB" ]
Date: 2005-09-02 01:01:13 +0200 From: @peterboncz To: Jens Teubner &lt;<teubner>&gt; Version: -- development Last updated: 2005-09-08 10:32:12 +0200 ## Comment 3791 Date: 2005-09-02 13:01:13 +0200 From: @peterboncz I admit that I may be in need of a course on XQuery but I do not understand why the below query: (: q4 'SELECT alb.Title, ass.Title FROM Assets as ass, Albums as alb WHERE ass.AlbumId=alb.AlbumId LIMIT 2' :) let $assets := fn:subsequence(fn:doc("dataset.xml")//Album, 1, 2) cast as xs:anyNode* let $albums := $assets/parent::* for $album in $albums for $asset in $album/Asset return <Result><Album> { $album/Title } </Album><Track> { $asset/Title } </Track></Result> on dataset: <Music> <Artist> <ArtistId>1</ArtistId> <Name>artist1</Name> <Album> <AlbumId>1</AlbumId> <Title>album1</Title> <Asset> <GlobalId>1</GlobalId> <Title>asset1</Title> <TrackNr>1</TrackNr> </Asset> <Asset> <GlobalId>2</GlobalId> <Title>asset2</Title> <TrackNr>2</TrackNr> </Asset> <Asset> <GlobalId>3</GlobalId> <Title>asset3</Title> <TrackNr>3</TrackNr> </Asset> <Asset> <GlobalId>4</GlobalId> <Title>asset4</Title> <TrackNr>4</TrackNr> </Asset> </Album> </Artist> <Artist> <ArtistId>2</ArtistId> <Name>artist2</Name> <Album> <AlbumId>2</AlbumId> <Title>album2</Title> <Asset> <GlobalId>5</GlobalId> <Title>asset5</Title> <TrackNr>1</TrackNr> </Asset> <Asset> <GlobalId>6</GlobalId> <Title>asset6</Title> <TrackNr>2</TrackNr> </Asset> </Album> </Artist> </Music> produces the error: $ pf -M 04.xq syntax error, unexpected "let $" on line 3 (next token is `let $') parse error: XQuery parsing failed halted in ../../../compiler/parser/parser.y (PFparse), line 2239 ## Comment 3792 Date: 2005-09-02 13:09:26 +0200 From: @drstmane Logged In: YES user_id=572415 Hm, having even less XQuery knowledge, I added a "return" after the first "let ..." statement, and get: ======== illegal reference to context node: at (4,27-4,32): ``.'' is unbound halted in /net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/pathfinder/compiler/core/fs.brg (reduce), line 2796 ======== ... no idea, whether this helps ... ## Comment 3793 Date: 2005-09-02 13:23:34 +0200 From: @drstmane Logged In: YES user_id=572415 Apparently, the assignement to Jens got lost due to concurrent changes by Peter & me --- SF != ACID |-( ## Comment 3794 Date: 2005-09-02 14:05:10 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 from what i know casting can only be applied to atomic types. you should use 'treat as' for non-atomic types. so it should read: fn:subsequence(...) treat as anyNode* ## Comment 3795 Date: 2005-09-02 14:20:26 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 of course the question is whether anyNode is an atomic type or not... (i'm starting to having doubts) ## Comment 3796 Date: 2005-09-03 11:42:34 +0200 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 Pathfinder behaves correctly in Peter's query. XQuery's `cast as' only allows `SingleType's. That is, only QNames with optionally a question mark, no star or plus occurrence operator. Pathfinder will parse the * as a multiplication sign and the following keyword as a location step. The location step, however, is relative, without a legally bound context set, thus the error message. However, while playing around with your example, I noticed another ("real") bug, probably located in Jan's "simplifier" in milprint_summer.c. I will file a new bug report for that one. ## Comment 3797 Date: 2005-09-03 16:16:51 +0200 From: @peterboncz Logged In: YES user_id=591107 Thanks Jens, ok, I see. in the below, I iterate over the items and cast individually, however this fails. It does work with 'treat as', I noted. I re-opened this bug to raise another issue. The reason I was casting was that fn:subsequence returns item* when invoked on node*. This is contrary to what is mandated by the formal semantics. http://www.w3.org/TR/xquery-semantics/sec_fn_subsequence Maybe I did not correctly specify the signature in semantics/xquery_fo.c Is there soimething like type-variables (a la any::1 in MIL) that can specify that the input type is equal to the oputput type? this is not only an issue with subsequence, but with many more operators, as described in the formal semantics doc. Peter let $items := fn:subsequence(fn:doc("dataset.xml")//Asset, 1, 100) let $assets := for $item in $items return $item cast as xs:anyNode let $albums := $assets/parent::* for $album in $albums for $asset in $album/Asset where $asset/TrackNr = 1 return <Result><Album> { $album/Title } </Album><Track> { $asset/Title } </Track></Result> ## Comment 3798 Date: 2005-09-08 10:32:12 +0200 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 Specific typing rule is now implemented in CVS HEAD. ## Comment 3799 Date: 2005-12-19 16:09:47 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS pathfinder/tests/BugDay_2005-12-19_0.9.3/Tests/cast_as_anyNode.SF-1280519.xq I'm assuming the result (a syntax error) is correct. ## Comment 3800 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1280519 at http://sourceforge.net/support/tracker.php?aid=1280519
cast as anyNode*
https://api.github.com/repos/MonetDB/MonetDB/issues/676/comments
0
2020-11-30T08:48:24Z
2024-06-28T13:34:48Z
https://github.com/MonetDB/MonetDB/issues/676
753,281,476
676
[ "MonetDB", "MonetDB" ]
Date: 2005-09-02 11:22:01 +0200 From: @yzchang To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: -- development Last updated: 2005-09-05 08:57:59 +0200 ## Comment 3787 Date: 2005-09-02 11:22:01 +0200 From: @yzchang This link doesn't work (http://monetdb.cwi.nl/TechDocs/Core/gdk/index.html): MonetDB Kernel API and data structures. [an error occurred while processing this directive] ## Comment 3788 Date: 2005-09-02 13:00:19 +0200 From: @drstmane Logged In: YES user_id=572415 Indeed. Thanks for reporting it. The problem is, that texi2html was nolonger available on the new machines, and hence the respective documents were not created during the nightly testing's "make docs". I installed texi2html on my machine, and will update the website as soon as next night's testing has finished. ## Comment 3789 Date: 2005-12-19 16:00:18 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / DOCUMENTATION ## Comment 3790 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1280452 at http://sourceforge.net/support/tracker.php?aid=1280452
Kernel guide not avialable
https://api.github.com/repos/MonetDB/MonetDB/issues/675/comments
0
2020-11-30T08:48:22Z
2024-06-27T11:05:54Z
https://github.com/MonetDB/MonetDB/issues/675
753,281,442
675
[ "MonetDB", "MonetDB" ]
Date: 2005-09-01 03:26:36 +0200 From: @drstmane To: Peter Boncz &lt;<boncz>&gt; Version: -- development Last updated: 2005-09-02 11:18:03 +0200 ## Comment 3783 Date: 2005-09-01 15:26:36 +0200 From: @drstmane [Unless compiled with debugging explicitely enabled (`configure --enable-debug`) and hence optimization disabled (i.e., `gcc -g [-O0]` iso. the default `gcc -g -O2`),] the latest version of compiler/mil/milprint_summer.c (revision 1.89) fails to compile with: ======== make[3]: Entering directory `/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..BUILD./PATHFINDER/compiler/mil' /bin/sh ../../libtool --tag=CC --mode=compile gcc -m64 -DHAVE_CONFIG_H -I. -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/pathfinder/compiler/mil -I../.. -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/pathfinder/compiler/mil -I../include -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/pathfinder/compiler/mil/../include -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB/common -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB/gdk -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB/monet -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB/plain -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB/contrib -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB/mapi -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB/C -DLIBMIL -g -O2 -std=c99 -Wall -W -Werror-implicit-function-declaration -Werror -Wno-unused-function -Wno-unused-label -c -o libmil_la-milprint_summer.lo `test -f 'milprint_summer.c' || echo '/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/pathfinder/compiler/mil/'`milprint_summer.c gcc -m64 -DHAVE_CONFIG_H -I. -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/pathfinder/compiler/mil -I../.. -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/pathfinder/compiler/mil -I../include -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/pathfinder/compiler/mil/../include -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB/common -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB/gdk -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB/monet -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB/plain -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB/contrib -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB/mapi -I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..PREFIX./include/MonetDB/C -DLIBMIL -g -O2 -std=c99 -Wall -W -Werror-implicit-function-declaration -Werror -Wno-unused-function -Wno-unused-label -c /net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/pathfinder/compiler/mil/milprint_summer.c -fPIC -DPIC -o .libs/libmil_la-milprint_summer.o cc1: warnings being treated as errors /net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/pathfinder/compiler/mil/milprint_summer.c: In function 'translate2MIL': /net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/pathfinder/compiler/mil/milprint_summer.c:7750: warning: 'rc' is used uninitialized in this function make[3]: *** [libmil_la-milprint_summer.lo] Error 1 make[3]: Leaving directory `/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..BUILD./PATHFINDER/compiler/mil' make[2]: *** [install] Error 2 make[2]: Leaving directory `/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..BUILD./PATHFINDER/compiler/mil' make[1]: *** [install-recursive] Error 1 make[1]: Leaving directory `/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d..BUILD./PATHFINDER/compiler' make: *** [install-recursive] Error 1 ======== ## Comment 3784 Date: 2005-09-02 11:17:50 +0200 From: @drstmane Logged In: YES user_id=572415 Fixed by Peter. ## Comment 3785 Date: 2005-12-19 15:59:45 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / COMPILATION ## Comment 3786 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1279145 at http://sourceforge.net/support/tracker.php?aid=1279145
Pathfinder (milprint_summer.c) fails to compile
https://api.github.com/repos/MonetDB/MonetDB/issues/674/comments
0
2020-11-30T08:48:19Z
2024-06-27T11:05:53Z
https://github.com/MonetDB/MonetDB/issues/674
753,281,411
674
[ "MonetDB", "MonetDB" ]
Date: 2005-08-30 10:16:41 +0200 From: Wouter Alink &lt;<alink>&gt; To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: -- development Last updated: 2005-09-05 08:06:37 +0200 ## Comment 3778 Date: 2005-08-30 22:16:41 +0200 From: Wouter Alink &lt;<alink>&gt; The nightly testing results seem not accessible... clicking on any 'o' or 'X' returns the following error: Not Found The requested URL /projects/monetdb/Stable/pathfinder/.ntv.64.64.d-Fedora3/mTests0/compiler/core/tests/fs/.good-004.out.html was not found on this server. Apache/2.0.51 (Fedora) Server at crux.ins.cwi.nl Port 80 ## Comment 3779 Date: 2005-08-30 22:34:40 +0200 From: @drstmane Logged In: YES user_id=572415 Wouter, this is actually more a "feature" than a "bug": As stated on (e.g.) http://monetdb.cwi.nl/Development/TestWeb/Stable/pathfinder/index.html , the "Mtest.py -r -d0" test runs (i.e., with GDKdebug=0) are only executed during weekends (i.e., on Saturday & Sunday); only the "make check" test runs (i.e., with GDKdebug=10) are run every night (also during weekends). The reason is that executing both test runs simply takes too long (at least on some old & slow machines in our test-farm)... Hence, http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests03/ntv.64.64.d-Fedora3/compiler_core_tests_fs/good-004.out.00.html is not available today (Tuesday), while http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/ntv.64.64.d-Fedora3/compiler_core_tests_fs/good-004.out.00.html is available. Actually, later this week, I might have a look, whether it can patch our "TestTool" scripts to keep the last weekend's results for the "Mtest.py -r -d0" test runs during the whole week... ## Comment 3780 Date: 2005-09-02 13:01:36 +0200 From: @drstmane Logged In: YES user_id=572415 I (hopefully) fixed our TestTools. Next week will show, whether it works as expected/desired. ## Comment 3781 Date: 2005-12-19 15:58:59 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / DOCUMENTATION ## Comment 3782 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1276925 at http://sourceforge.net/support/tracker.php?aid=1276925
TestWeb: details not available
https://api.github.com/repos/MonetDB/MonetDB/issues/673/comments
0
2020-11-30T08:48:16Z
2024-06-27T11:05:52Z
https://github.com/MonetDB/MonetDB/issues/673
753,281,370
673
[ "MonetDB", "MonetDB" ]
Date: 2005-08-28 04:04:16 +0200 From: Ethan Aubin &lt;<eaubin>&gt; To: Peter Boncz &lt;<boncz>&gt; Version: -- development Last updated: 2005-08-30 01:05:31 +0200 ## Comment 3773 Date: 2005-08-28 16:04:16 +0200 From: Ethan Aubin &lt;<eaubin>&gt; I've a query which fails printing: GDKmmap(333250560) fail => BBPtrim(enter) usage[mem=29092948,vm=246087680] mallinfo.arena = 31477760 BBPTRIM_ENTER: memsize=29092948,vmsize=246087680 BBPTRIM: memtarget=0 vmtarget=1073741824 TRIMSCAN: mem=0 vm=1, start=1, limit=1 TRIMSCAN: 38170 0=doc_sema (0) TRIMSCAN: 38268 1=attr_own1 (0) TRIMSCAN: 38269 2=attr_qn1 (0) TRIMSCAN: 38270 3=attr_prop1 (0) .... (rest of failure is in attached file). I've narrowed the problem in the query to: declare function local:check-function($f as element(c:function)) as element(c:function)* { let $target := <c:target id="{ $f/c:name/text() }" category="Component"/> return if (some $performs-target in $db/c:relation/c:performs/c:target satisfies $target = $performs-target) then () else $f }; <badfunctions>{ for $f in $db/c:element/c:function return local:check-function($f) }</badfunctions> I've attached my config.log file. Is there any other information I could provide would be useful to the developers? - EA ## Comment 3774 Date: 2005-08-30 01:05:13 +0200 From: @peterboncz Logged In: YES user_id=591107 Hi, What happens is that MonetDB simply gets out of resources. The GDKmmmap() tells that this call for more memory fails on a request for 333MB of RAM. It also tells that the memsize is set to about 30MB and it is at that point already using 246 MB of virtual memory. This kind of thing is a limitation of the execution model the makes MonetDB fast (when it works). Other database systems in these situations justs get awfully slow; MonetDB throws in the towel. What I take from this is that you are running a 32-bit operating system with not too much memory installed. If you have a 64-bits system and a lot of RAM, MonetDB can take you much further. But I actually doubt that this is your problem.... In any case, I suspect that you would not have been content with a very slow running query either. So, the best solution is to modify your query in such a way that it gets to benefit from the strengths of database tehcnology in general. Basically, you ar submitting a query in a java/C programming style. The optimizer cannot spot the join pattern and it resorts to generating a so-called cartesian product. This leads to O(N*N) complexity, were a O(N) or O(Nlog(N) solution is possible. so.. maybe something along these lines works better: fn:distinct-values( for $f in $db/c:element/c:function for $t in $db/c:relation/c:performs/c:target [@category="Component"] where $t@id = $f/c:namr/text() return $f) this query uses a nested for-loop united by an simple equality condition in the where clause. MonetDB can detect this particular pattern as a join. Maybe it won't work if you split the nested loop in two parts, the inner being a user defined function. But I actually think it should work as well. ## Comment 3775 Date: 2005-08-30 11:22:38 +0200 From: @drstmane Logged In: YES user_id=572415 Ethan, did you try Peter's rephrased query? Does it work for you? How much main-memory does your system have, and how big is your XML document (filesize of the .xml file, and in case you know, number of c:target and c:function nodes). Thanks in advance, Stefan ## Comment 3776 Date: 2005-12-19 15:58:18 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / NOT A BUG ## Comment 3777 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1275140 at http://sourceforge.net/support/tracker.php?aid=1275140
GDKmmap fails executing query
https://api.github.com/repos/MonetDB/MonetDB/issues/672/comments
0
2020-11-30T08:48:13Z
2024-06-28T13:34:48Z
https://github.com/MonetDB/MonetDB/issues/672
753,281,349
672
[ "MonetDB", "MonetDB" ]
Date: 2005-08-28 02:51:52 +0200 From: Ethan Aubin &lt;<eaubin>&gt; To: Jens Teubner &lt;<teubner>&gt; Version: -- development Last updated: 2005-08-31 03:06:21 +0200 ## Comment 3768 Date: 2005-08-28 02:51:52 +0200 From: Ethan Aubin &lt;<eaubin>&gt; I've a query which fails with: type error: no variant of function pf:merge-adjacent-text-nodes accepts the given argument type(s): item* though there is no mention of merge-adjacent-text-nodes. E.g. declare namespace local = "urn:local"; declare variable $db := doc("big.xml")/database; declare function local:relations($ignore) { for $r in $db/relation return $r }; <html>{ local:relations($db) }</html> Any ideas what going on? - EA ## Comment 3769 Date: 2005-08-28 10:24:49 +0200 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 Hi Ethan, functions whose QName start with a `pf' are internal functions of our compiler (`pf' is for `Pathfinder', and the hash mark keeps you from explicitly calling these functions in the surface language). These functions (and their names) should never be visible to you, sorry for that confusion. I will also look into this as soon as I find the time. What might help you as a workaround (though I haven't tried), is to assign types to your function and its parameter. ## Comment 3770 Date: 2005-08-31 15:06:21 +0200 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 Fixed in CVS (stable branch). The specific typing rule for item-sequence-to-node-sequence() did not correctly deal with the type `item'. The function in Ethan's example lacks a type declaration and is thus treated as `item*' during static typing. Ethan, you may either checkout the fix from CVS (stable branch, tag XQuery_0-8), or add a type declaration to your function declaration as a workaround. Thanks again for the bug report. ## Comment 3771 Date: 2005-12-19 15:55:29 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS pathfinder/tests/BugDay_2005-12-19_0.9.3/Tests/type_error_in_function_resolution.SF-1274903.xq ## Comment 3772 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1274903 at http://sourceforge.net/support/tracker.php?aid=1274903
no variant of function #pf:merge-adjacent-text-nodes accepts
https://api.github.com/repos/MonetDB/MonetDB/issues/671/comments
0
2020-11-30T08:48:10Z
2024-06-27T11:05:50Z
https://github.com/MonetDB/MonetDB/issues/671
753,281,310
671
[ "MonetDB", "MonetDB" ]
Date: 2005-08-27 11:19:37 +0200 From: Ethan Aubin &lt;<eaubin>&gt; To: GDK devs &lt;<bugs-common>&gt; Version: -- development Last updated: 2005-08-29 05:05:39 +0200 ## Comment 3759 Date: 2005-08-27 23:19:37 +0200 From: Ethan Aubin &lt;<eaubin>&gt; Hi, Type declarations on functions don't seem to be working the way I expect: E.g. Say I'm using schema and data from the usecase STRONG 1.9.4.2 Q2 at http://www.w3.org/TR/xquery-use-cases/strong. I.e. the schema is: <?xml version="1.0" encoding="UTF-8"?> <xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema" elementFormDefault="qualified" targetNamespace="http://www.example.com/zips" xmlns:zips="http://www.example.com/zips"> <xs:element name="zips"> <xs:complexType> <xs:sequence> <xs:element maxOccurs="unbounded" ref="zips:row"/> </xs:sequence> </xs:complexType> </xs:element> <xs:element name="row"> <xs:complexType> <xs:sequence> <xs:element ref="zips:city"/> <xs:element ref="zips:state"/> <xs:element ref="zips:zip"/> </xs:sequence> </xs:complexType> </xs:element> <xs:element name="city" type="xs:string"/> <xs:element name="state" type="xs:string"/> <xs:element name="zip" type="xs:integer"/> </xs:schema> and the data is: <zips xmlns="http://www.example.com/zips"> <row> <city>Old Town</city> <state>PA</state> <zip>95819</zip> </row> <row> <city>Durham</city> <state>NC</state> <zip>27701</zip> </row> <row> <city>Durham</city> <state>NC</state> <zip>27703</zip> </row> </zips> And I want to get a list of the city nodes. galax lets me use the following xquery: declare namespace local = "urn:local"; import schema namespace z = "http://www.example.com/zips" at "zips.xsd"; declare function local:cities($r as element(z:zips)) as element(z:city)* { $r/z:row/z:city }; <cities> { local:cities(doc("zips.xml")/z:zips) } </cities> to get: <cities xmlns:z="http://www.example.com/zips" xmlns:local="urn:local"> <city xmlns="http://www.example.com/zips">Old Town</city> <city xmlns="http://www.example.com/zips">Durham</city> <city xmlns="http://www.example.com/zips">Durham</city> </cities> but monetdb fails on this same query printing: type error: no variant of function local:cities accepts the given argument type(s): element z:zips { item* }* Is a problem with the location hint, or the schema import? Thanks - EA ## Comment 3760 Date: 2005-08-28 10:19:05 +0200 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 Thanks again for your report, Ethan. I have to admit that our schema import feature is not really well maintained (nor thoroughly tested). I cannot make any promises for the upcoming week (We're in Trondheim on VLDB to present our engine). But I will try to look into this as soon as possible. ## Comment 3761 Date: 2005-08-28 13:44:00 +0200 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 Sorry, Ethan, but the error message is expected behavior. Your problem has nothing to do with schema import. Your import schema statment mainly delared the namespace z, the query itself did not reference any types from the loaded XML schema file. The problem is that the return type of the path expression `doc("zips.xml")/ z:zips' is element z:zips { item* }*. Note the star at the end, the expression returns zero or more elements named z:zips. Your function, however, only accepts arguments of type element z:zips { item* } (that is, exactly one element with name z:zips). Your query is thus not type-safe, and the Pathfinder compiler complains. The reason why you can run the query in Galax is probably that Galax (at least by default) does not implement XQuery's static type feature. Pathfinder implements this feature, which makes it appear very picky about types. I think Galax provides static typing via some additional command line switch (though as far as I remember that was known to be broken). There are two possible fixes for your query: (a) Tell Pathfinder that the path expression really returns exaclty one element. The `exactly-one()' function does that: local:cities (exactly-one (doc ("zips.xml")/z:zips)) (b) Let your function accept sequences of elements: declare function local:cities($r as element(z:zips)*) as element(z:city)* (note the star after the function argument) I will re-read the specs in Trondheim. I'm pretty sure though, that Pathfinder behaves correctly here. (Can somebody close that bug? For some reason I cannot.) ## Comment 3762 Date: 2005-08-28 15:10:35 +0200 From: Ethan Aubin &lt;<eaubin>&gt; Logged In: YES user_id=522388 Yikes, how did I miss that? Your explaination makes perfect sense. By default Galax does catch _some_ type errors but this gives me less confidence in it. Is there another implementation you use to sanity check your results? (I'd be happy to run my scripts on a different implementation or two, so I only send actual bug reports :-) Also maybe the error message could print the expected type along with the used type? The "variant of" part also confused me, my knee jerk reaction was that there was some ad-hoc function overloading going on somewhere. I'm guessing its a reminder of the implicit conversions on atomic types... Is there any talk of distributing pathfinder with a testsuite or the xquery usecases? ## Comment 3763 Date: 2005-08-28 15:36:46 +0200 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 Hi Ethan, don't hesitate to bother us with bug reports. If it turns out to be a "feature", don't worry. As for now, your ratio is still not too bad: two out of three are actual bugs :-). Your question for other implementations is a good one. I don't investigate on other implementations on a regular basis, and typically I don't have the time to really look into the details. Particularly the typing issues in XQuery are rather tricky, and it is not surprising that implementations have their problems with it. (Pathfinder does have its problems as well, let's see if you find them... :-).) One implementation that I like pretty much on the first view with respect to typing was the online demo at http://www. cogneticsystems.com/xquery/xquery.html. But I haven't really tested that one either. You're right that XQuery does not allow function overloading on the surface language (well, you may redefine a function with same name, but different number of arguments). Several built-in functions, however, are overloaded. Our typechecker does implements some overloading functionality to deal with those. (You are still not allowed to overload functions on the surface level, though.) So that is where the "variant of" comes from. But I admit that our error messages have some room for improvement... I'm not really sure what you mean with your question on a test suite. For development purposes we actually have a HUGE set of test queries that we run every night through automated scripts. The XPathMark test suite, for instance, is part of that. If somebody really finds a lot of time, we might as well add the use cases to our test setup. ## Comment 3764 Date: 2005-08-28 16:26:43 +0200 From: Ethan Aubin &lt;<eaubin>&gt; Logged In: YES user_id=522388 Re: The testsuite. I like a small test suite on my machine, so that I can check that tools on my box did nothing unexpected during the build. Its also the best way to see what behavior can be expected to work right out of the box. Are the nightly test scripts available online? Re: The usecases. They might be nice just to check cross-implementation consistency. E.g. have the nightly test queries been ran by a different xquery processor? > (well, you may redefine a function with same name, but > different number of arguments). Wow, I didn't know that. I learn something new everyday. Cheers - EA ## Comment 3765 Date: 2005-08-29 17:05:39 +0200 From: @mlkersten Logged In: YES user_id=490798 On behalf of Jens it is closed ## Comment 3766 Date: 2005-12-19 16:00:50 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: NO TEST / INVALID ## Comment 3767 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1274819 at http://sourceforge.net/support/tracker.php?aid=1274819 The original assignee of this bug does not have an account here. Reassigning to the default assignee for the component, bugs-common@monetdb.org. Previous assignee was nobody@users.sourceforge.net.
"type error: no variant of function ... accepts the the giv"
https://api.github.com/repos/MonetDB/MonetDB/issues/670/comments
0
2020-11-30T08:48:07Z
2024-06-28T13:34:47Z
https://github.com/MonetDB/MonetDB/issues/670
753,281,287
670
[ "MonetDB", "MonetDB" ]
Date: 2005-08-27 06:01:30 +0200 From: Ethan Aubin &lt;<eaubin>&gt; To: GDK devs &lt;<bugs-common>&gt; Version: -- development Last updated: 2005-08-27 09:19:57 +0200 ## Comment 3748 Date: 2005-08-27 18:01:30 +0200 From: Ethan Aubin &lt;<eaubin>&gt; Hi, The books1.xq example at http://monetdb.cwi.nl/XQuery/Demo/Books/index.html fails when run printing bad usage of XML namespaces: at (1,1-8,1): unknown namespace in qualified function name local:toc FWIW, galax seems parse and type the file correctly. Also the URL for books.xml is different in books2.xq and the html displayed for books2.xq (missing the files/ directory). ## Comment 3749 Date: 2005-08-27 18:45:28 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 Hello Ethan, Thank you for your bug-report. I tried the books1.xq, but could not reproduce the error you describe. Could you please provide us with some more information (Operating system/Version of MonetDB/XQuery?) If you use MonetDB/XQuery 0.8.0, would you please upgrade to MonetDB/XQuery 0.8.2 as many bugs have been fixed since 0.8.0. Thanks in advance, Wouter ## Comment 3750 Date: 2005-08-27 19:18:48 +0200 From: Ethan Aubin &lt;<eaubin>&gt; Logged In: YES user_id=522388 The README, NEWS, ChangeLog file are empty, but the directory and config.status file seems to indicate pathfinder 0.8.2 is being used with MonetDB version 4.8.2. This is on a debian linux box w/ kernel 2.6.11 If I add declare namespace local = "urn:local"; to the start of the file, then is works like a champ. BTW, Should add a bug report requesting a --version command line switch? ## Comment 3751 Date: 2005-08-27 21:05:59 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 ah, i see why i couldn't reproduce your bug. you meant books2.xq (somehow i thought the error was in books1.xq) I fixed the query books2.xq (and book.xml) on the website (removed the "local:" several times in the query, and disconnected the dtd from the xml-file). Note: it will probably take a few days before the website will show the new files. The "--version" flag can be translated in "cat VERSION", but maybe you're right and is it a missing feature. But this is not my area... (Sjoerd?) Thanks again for your bug report and for using MonetDB/XQuery. If you have any other comments or suggestions, we would be keen to hear from you. p.s. could somebody please mark this bug fixed (i don't have those privileges) ## Comment 3752 Date: 2005-08-27 21:19:57 +0200 From: @mlkersten Logged In: YES user_id=490798 Closed upon request ## Comment 3753 Date: 2005-08-27 21:43:51 +0200 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 Sorry guys, you closed that bug too quickly. Ethan actually found a bug in the compiler. The webpage was correct (we should have tested those queries before publishing, though). Our namespace checker did not predefine the namespaces `local' and `xdt'. I have committed a fix to the CVS stable branch. Ethan, thanks for the report (and sorry for the confusion). You will have to either checkout my fixed version from CVS (stable branch, tag XQuery_0- 8), or wait for the next nightly CVS snaphot. As a workaround, you can either remove the `local' prefixes in your query, or define the namespace directly in your query (via `declare namespace'). The --version option is actually a great suggestion. We will think about how to elegantly incorporate something like that in our release cycle. ## Comment 3754 Date: 2005-08-27 22:56:52 +0200 From: Ethan Aubin &lt;<eaubin>&gt; Logged In: YES user_id=522388 I'm very impressed by how responsive you've all been! I'm an xquery novice and found monet when frustrated with the speed of galax. It would take 10 minutes for a simple queries on a small 21meg db, which monetdb runs in a matter of seconds! Its blindingly fast! The XQuery front end does seem to be a bit quirky wrt syntax and typing. I'll post a new bug report about a typing problem in a few minutes. BTW, I'm also looking forward to when xquery modules are implemented. ## Comment 3755 Date: 2005-08-27 23:39:28 +0200 From: @grobian Logged In: YES user_id=963970 out of curiosity, what 'modules' are you looking for? ## Comment 3756 Date: 2005-08-28 00:11:36 +0200 From: Ethan Aubin &lt;<eaubin>&gt; Logged In: YES user_id=522388 MM: Just plain old XQuery modules. E.g. mymod.xq: module namespace my = "urn:some-functions"; declare function my:one() { 1 }; main.xq: import module namespace my = "urn:some-functions" at "mymod.xq"; my:one() + my:one() ## Comment 3757 Date: 2005-12-19 15:56:27 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: ALREADY IN TESTWEB Web site examples are tested in pathfinder/tests/WebSite/Tests/ ## Comment 3758 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1274649 at http://sourceforge.net/support/tracker.php?aid=1274649 The original assignee of this bug does not have an account here. Reassigning to the default assignee for the component, bugs-common@monetdb.org. Previous assignee was nobody@users.sourceforge.net.
functions and XML namespaces
https://api.github.com/repos/MonetDB/MonetDB/issues/669/comments
0
2020-11-30T08:48:04Z
2024-06-27T11:05:48Z
https://github.com/MonetDB/MonetDB/issues/669
753,281,255
669
[ "MonetDB", "MonetDB" ]
Date: 2005-08-26 04:49:56 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-09-01 09:58:47 +0200 ## Comment 3745 Date: 2005-08-26 16:49:56 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Test case: sql>select extract( year from interval '1' year); [ -1 ] sql>select extract( day from interval '1' day ); [ 22 ] sql>select extract( day from interval '1 01' day to hour); [ 31 ] sql>select extract( hour from interval '1 01' day to hour); ERROR = !types sec_interval(64,0) (lng) and time(0,0) (daytime) are not equal I have no idea what's going on. ## Comment 3746 Date: 2005-12-19 15:50:40 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: ALREADY IN TESTWEB: sql/src/test/bugs/Tests/extract_interval_bug-sf-1274088.* ## Comment 3747 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1274088 at http://sourceforge.net/support/tracker.php?aid=1274088
SQL: select extract( ... from interval ...
https://api.github.com/repos/MonetDB/MonetDB/issues/668/comments
0
2020-11-30T08:48:01Z
2024-06-27T11:05:47Z
https://github.com/MonetDB/MonetDB/issues/668
753,281,224
668
[ "MonetDB", "MonetDB" ]
Date: 2005-08-26 04:44:07 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-09-01 09:59:05 +0200 ## Comment 3741 Date: 2005-08-26 16:44:07 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Test case: sql>select cast('1-2' as interval year to month ); [ 1 ] I'd expect 14. Looks like cast() takes only leading digits (up to the hyphen) into account. ## Comment 3742 Date: 2005-08-30 14:16:04 +0200 From: @njnes Logged In: YES user_id=43556 intervals now keep their qualifiers (like year to month) information in the types digit and scale (ugh) this is needed for dynamic type conversions (from string), which is added to the convert code generation ## Comment 3743 Date: 2005-12-19 15:50:08 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: ALREADY IN TESTWEB: sql/src/test/bugs/Tests/interval_convert_bugs-sf-1274077-1274085.* ## Comment 3744 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1274085 at http://sourceforge.net/support/tracker.php?aid=1274085
SQL: select cast('1-2' as interval year to month )
https://api.github.com/repos/MonetDB/MonetDB/issues/667/comments
0
2020-11-30T08:47:58Z
2024-06-27T11:05:46Z
https://github.com/MonetDB/MonetDB/issues/667
753,281,191
667
[ "MonetDB", "MonetDB" ]
Date: 2005-08-26 04:35:02 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-09-01 09:59:18 +0200 ## Comment 3737 Date: 2005-08-26 16:35:02 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Test case: sql>select interval '1' year = interval '12' second; [ true ] Sure, the underlying atomic types compare equal, but the intervals are incomparable. From the SQL standard: Year-month intervals are comparable only with other year-month intervals. Day-time intervals are comparable only with other day-time intervals. ## Comment 3738 Date: 2005-08-30 14:14:59 +0200 From: @njnes Logged In: YES user_id=43556 conversions between interval are nolonger supported ## Comment 3739 Date: 2005-12-19 15:49:27 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: ALREADY IN TESTWEB: sql/src/test/bugs/Tests/interval_convert_bugs-sf-1274077-1274085.* ## Comment 3740 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1274077 at http://sourceforge.net/support/tracker.php?aid=1274077
SQL: select interval '1' year = interval '12' second
https://api.github.com/repos/MonetDB/MonetDB/issues/666/comments
0
2020-11-30T08:47:56Z
2024-06-27T11:05:46Z
https://github.com/MonetDB/MonetDB/issues/666
753,281,170
666
[ "MonetDB", "MonetDB" ]
Date: 2005-08-23 04:41:23 +0200 From: @MarcinZukowski To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: -- development Last updated: 2005-08-23 08:23:45 +0200 ## Comment 3732 Date: 2005-08-23 16:41:23 +0200 From: @MarcinZukowski It seems pump operations do not set the 'hkey' properly in all the situations, here's the script: var x:=new(oid,int); x.insert(20@0,88); x.insert(21@0,89); hkey=1 var z:=[int](x.slice(0,1)); slice with unique keys, hkey set {min}(z).info().find("hkey").print(); [ "1" ] x.insert(21@0,0); hkey=0 var z:=[int](x.slice(0,1)); slice with unique keys, hkey not set {min}(z).info().find("hkey").print(); [ "0" ] var z:=[int](x.slice(0,2)); slice with duplicates, hkey not set {min}(z).info().find("hkey").print(); [ "1" ] I think pump "by definition" should set hkey to true so there seems to be a bug. ## Comment 3733 Date: 2005-08-23 19:22:29 +0200 From: @drstmane Logged In: YES user_id=572415 Fixed in CVS; see below for details. Marcin, could you please verify, whether your bug is indeed fixed, and then close this bug report? Thanks, Stefan The pump implementation (interpret_setaggr) correctly set the key property for the head of the result. However, if the result was the same size as the input (i.e., the input'\''s head was actually key/unique), the code is "clever" and calls ALIGNsetH to propagate the properties of the input'\''s head to the result'\''s head; in case the input'\''s head was not marked key (though it was), ALIGNsetH overwrote the previously correctly set key property of the result'\''s head. I fixed this, be setting the key property of the result'\''s head only after the ALIGNsetH call. ## Comment 3734 Date: 2005-08-23 20:23:45 +0200 From: @MarcinZukowski Logged In: YES user_id=607094 After both changes it seems to work again. Thank you, Stefan! ## Comment 3735 Date: 2005-12-19 15:48:17 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: TEST ADDED / SUCCESS: MonetDB/tests/BugDay_2005-12-19_4.9.3/Tests/pump_properties.SF-1267155.* ## Comment 3736 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1267155 at http://sourceforge.net/support/tracker.php?aid=1267155
Pump properties
https://api.github.com/repos/MonetDB/MonetDB/issues/665/comments
0
2020-11-30T08:47:53Z
2024-06-27T11:05:44Z
https://github.com/MonetDB/MonetDB/issues/665
753,281,135
665
[ "MonetDB", "MonetDB" ]
Date: 2005-08-23 03:25:52 +0200 From: @sjoerdmullender To: GDK devs &lt;<bugs-common>&gt; Version: -- development Last updated: 2007-02-25 06:14:39 +0100 ## Comment 3723 Date: 2005-08-23 15:25:52 +0200 From: @sjoerdmullender In quite a few places in the code (certainly MonetDB, possibly others) the symbol BUFSIZ is used to allocate memory. The problem with this is that BUFSIZ may be too small on certain systems for the intended purposes. For example: On Windows, BUFSIZ is defined as 512, and this is too small for a decent-sized monet_mod_path value (actually encountered when trying to run Mserver from the build directory using RunMserver.bat). We should go look at all occurences of BUFSIZ to see if they are ok if the value is relatively small. The problem mentioned could perhaps be solved if mo_substitute were to allocate memory in which it returns the substituted value instead of substituting in-place. ## Comment 3724 Date: 2005-12-19 15:39:56 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: NO TEST / CODING TECHNIQUE ## Comment 3725 Date: 2006-01-15 15:00:40 +0100 From: @grobian Logged In: YES user_id=963970 any news on this? ## Comment 3726 Date: 2006-01-15 15:53:34 +0100 From: @mlkersten Logged In: YES user_id=490798 An assessment without actions yet. upgrade.mx : OK str.mx: OK pcrc.mx/prce_compile_wrap, pcre_select, pcre_uselect, pcre_replace: possible buffer overflow problems on pcre_compile calls logger.mx/logger_open: PROBLEM, BUFSIZ->PATHLENGTH logger.mx/logger_new: idem logger.mx/logger_exit: idem logger.mx/logger_clean: idem ascii_io.mx: OK gdk_desc.mx: Dangerous, double check gdk_posix.mx: OK gdk_utils.mx:OK ## Comment 3727 Date: 2006-01-20 11:11:05 +0100 From: @mlkersten Logged In: YES user_id=490798 gdk_desc.mx: not a real issue as long as atom type names are less then 512 logger.mx: not serious and protected against overflow only logger.new might cause a problem at some point. pcre.mx: doesn't seem a big problem, unless pcre has very long built in error messages. Most likely not. ## Comment 3728 Date: 2006-01-20 11:29:15 +0100 From: @mlkersten Logged In: YES user_id=490798 gdk_desc.mx: not a real issue as long as atom type names are less then 512 logger.mx: not serious and protected against overflow only logger.new might cause a problem at some point. pcre.mx: doesn't seem a big problem, unless pcre has very long built in error messages. Most likely not. ## Comment 3729 Date: 2006-06-06 20:56:35 +0200 From: @grobian Logged In: YES user_id=963970 Does this thing need some more search/replace work, or can we close it? ## Comment 3730 Date: 2007-02-25 18:14:39 +0100 From: @mlkersten Logged In: YES user_id=490798 Originator: NO Visited once more before closing: Stable, modules, plain uses it in asii-io, str, pcre, upgrade. No serious threads Stable, monetdb and gdk all gone. Stable, modules/mal, no harm Stable, optimizer, one additional protection added Stable, clients, seem protected ## Comment 3731 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1267093 at http://sourceforge.net/support/tracker.php?aid=1267093 The original assignee of this bug does not have an account here. Reassigning to the default assignee for the component, bugs-common@monetdb.org. Previous assignee was nobody@users.sourceforge.net.
BUFSIZ considered harmful
https://api.github.com/repos/MonetDB/MonetDB/issues/664/comments
0
2020-11-30T08:47:50Z
2024-06-27T11:05:43Z
https://github.com/MonetDB/MonetDB/issues/664
753,281,098
664
[ "MonetDB", "MonetDB" ]
Date: 2005-08-23 01:33:06 +0200 From: @drstmane To: Peter Boncz &lt;<boncz>&gt; Version: -- development Last updated: 2005-12-25 11:29:36 +0100 ## Comment 3711 Date: 2005-08-23 13:33:06 +0200 From: @drstmane When executing XQuery queries via the "xquery(str,str)" MIL command, error messages that occur during the evaliadtion of the query get lost/swallowed. When executing XQuery queries via "MapiClient -lxquery" the respective error messaged do appeatr as expected. Most probably, this is caused by the fact that CMDxquery() in runtime/pathfinder.mx redirects GDKout (and thus GDKerr) to a buffer to capture the output of exec_xquery(). However, me knowledge of MonetDB/MIL error message handling is apparently not sufficient to solve this problem... ======== $ Mserver --dbinit='module(pathfinder); mapi_start(); pfstart();' -------- Monet Database Server V4.9.3 Copyright (c) 1993-2005, CWI. All rights reserved. Compiled for x86_64-redhat-linux-gnu/64bit with 64bit OIDs; dynamically linked. Visit http://monetdb.cwi.nl/ for further information. MonetDB>xquery("xml","doc(\"abcde.xml\")").printf(); -------- > Missing: > !ERROR: shredder.mx:fileSize: cannot stat `abcdef.xml': No such file or directory > !ERROR: CMDshred2bats: operation failed. -------- !ERROR: CMDxquery: operation failed. !ERROR: interpret_params: printf(param 1): evaluation error. MonetDB> ======== $ echo 'doc("abcdef.xml")' | MapiClient -lx -------- ERROR: shredder.mx:fileSize: cannot stat `abcdef.xml': No such file or directory $ ======== ## Comment 3712 Date: 2005-08-23 13:49:35 +0200 From: @yzchang Logged In: YES user_id=341633 Hi Stefan, I thought this was lost by the changes in monet.mx? Didn't it help? ## Comment 3713 Date: 2005-08-23 13:53:27 +0200 From: @drstmane Logged In: YES user_id=572415 Nope. The fixes in monet.mx only solved the problem that the error state got lost completely. Now, at least the last/top-level error message "!ERROR: CMDxquery: operation failed." appears, again; but the actual reason for this failure (i.e., "!ERROR: shredder.mx:fileSize: cannot stat `abcdef.xml': No such file or directory") is still lost... ## Comment 3714 Date: 2005-11-18 19:14:20 +0100 From: @drstmane Logged In: YES user_id=572415 added test scripts as tests/BugsViaSourgeforce/Tests/ID.1267018.swallowed_error_messages-[12].* ## Comment 3715 Date: 2005-11-24 23:59:07 +0100 From: @peterboncz Logged In: YES user_id=591107 this is fixed now. ## Comment 3716 Date: 2005-12-19 15:39:01 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: ALREADY IN TESTWEB: pathfinder/tests/BugsViaSourgeforce/Tests/ID.1267018.swallowed_error_messages* ## Comment 3717 Date: 2005-12-21 00:21:34 +0100 From: @drstmane Logged In: YES user_id=572415 re-opened as the (proper) error messages seem to get swallowed, again: http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_BugsViaSourgeforce/ID.1267018.swallowed_error_messages-2.out.00.html ## Comment 3718 Date: 2005-12-23 18:20:43 +0100 From: @peterboncz Logged In: YES user_id=591107 has been fixed ## Comment 3719 Date: 2005-12-23 18:35:10 +0100 From: @drstmane Logged In: YES user_id=572415 sorry for being so (too?) "picky", but IMHO, the fix has "disappeared" again: in http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.32.32.d-Fedora4/tests_BugsViaSourgeforce/ID.1267018.swallowed_error_messages-2.out.00.html I would expext the original error message !ERROR: I/O warning : failed to load external entity "abcde.xml" but all that appears is !ERROR: interpret: no matching MIL operator to 'printf(void)'. checking monet's testing logs, is seems as if the bug was fixed on Nov 24 or Nov 25, but reappeared due to chenges made on Dec 02 ... this should help to locate which changes made the bug re-appear ... (I'm currently not online enough to check this myself, sorry) A fix is not "urgend", but I re-open the report, again, to keep *us* reminded. ## Comment 3720 Date: 2005-12-23 18:36:21 +0100 From: @drstmane Logged In: YES user_id=572415 oops, sorry, small typo: the changes that made the bug re-appear must have been on Dec 01 ... ## Comment 3721 Date: 2005-12-25 11:29:36 +0100 From: @drstmane Logged In: YES user_id=572415 (re-)fixed by Peter. Thank you very much for the instant action! ## Comment 3722 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1267018 at http://sourceforge.net/support/tracker.php?aid=1267018
XQuery: xquery() swallows error message
https://api.github.com/repos/MonetDB/MonetDB/issues/663/comments
0
2020-11-30T08:47:46Z
2024-06-27T11:05:42Z
https://github.com/MonetDB/MonetDB/issues/663
753,281,058
663
[ "MonetDB", "MonetDB" ]
Date: 2005-08-22 08:13:30 +0200 From: @drstmane To: @yzchang Version: -- development Last updated: 2005-08-23 09:48:50 +0200 ## Comment 3707 Date: 2005-08-22 08:13:30 +0200 From: @drstmane On our SunOS 5.8 (Solaris 8) machine "apps", pathfinder fails to compile: ======== [...] gcc -m64 -DHAVE_CONFIG_H -I. -I/var/tmp/monet.GNU.64.64.d.24482/pathfinder/runtime -I.. -I/var/tmp/monet.GNU.64.64.d.24482/pathfinder/runtime -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB/common -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB/gdk -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB/monet -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB/plain -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB/contrib -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB/mapi -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB/C -I/ufs/acoi/local/SunOS/64/include/libxml2 -I../compiler/include -I/var/tmp/monet.GNU.64.64.d.24482/pathfinder/runtime/../compiler/include -DLIBPATHFINDER -O2 -std=c99 -D__EXTENSIONS__ -Wall -W -O2 -fomit-frame-pointer -finline-functions -DHWCOUNTERS -DHW_SunOS -DHW_sun4u -Werror-implicit-function-declaration -Werror -Wno-unused-function -Wno-unused-label -Wno-uninitialized -DEMBEDDED -DNO_CGI -DNO_AUTH -c pathfinder.c -fPIC -DPIC -o .libs/lib_pathfinder_la-pathfinder.o /bin/bash ../libtool --mode=compile gcc -m64 -DHAVE_CONFIG_H -I. -I/var/tmp/monet.GNU.64.64.d.24482/pathfinder/runtime -I.. -I/var/tmp/monet.GNU.64.64.d.24482/pathfinder/runtime -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB/common -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB/gdk -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB/monet -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB/plain -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB/contrib -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB/mapi -I/var/tmp/monet.GNU.64.64.d.24482/MonetDB/.GNU.64.64.d-SunOS5.8/include/MonetDB/C -I/ufs/acoi/local/SunOS/64/include/libxml2 -I../compiler/include -I/var/tmp/monet.GNU.64.64.d.24482/pathfinder/runtime/../compiler/include -DLIBPF_SUPPORT -O2 -std=c99 -D__EXTENSIONS__ -Wall -W -O2 -fomit-frame-pointer -finline-functions -DHWCOUNTERS -DHW_SunOS -DHW_sun4u -Werror-implicit-function-declaration -Werror -Wno-unused-function -Wno-unused-label -Wno-uninitialized -DEMBEDDED -DNO_CGI -DNO_AUTH -c -o lib_pf_support_la-pf_support.glue.lo `test -f 'pf_support.glue.c' || echo '/var/tmp/monet.GNU.64.64.d.24482/pathfinder/runtime/'`pf_support.glue.c /var/tmp/monet.GNU.64.64.d.24482/pathfinder/runtime/pathfinder.mx: In function `setup_connection': /var/tmp/monet.GNU.64.64.d.24482/pathfinder/runtime/pathfinder.mx:1133: syntax error before '.' token /var/tmp/monet.GNU.64.64.d.24482/pathfinder/runtime/pathfinder.mx:1139: `S_un' undeclared (first use in this function) /var/tmp/monet.GNU.64.64.d.24482/pathfinder/runtime/pathfinder.mx:1139: (Each undeclared identifier is reported only once /var/tmp/monet.GNU.64.64.d.24482/pathfinder/runtime/pathfinder.mx:1139: for each function it appears in.) /var/tmp/monet.GNU.64.64.d.24482/pathfinder/runtime/pathfinder.mx:1142: `err' undeclared (first use in this function) /var/tmp/monet.GNU.64.64.d.24482/pathfinder/runtime/pathfinder.mx:1158: `sock' undeclared (first use in this function) gmake[3]: *** [lib_pathfinder_la-pathfinder.lo] Error 1 [...] ======== See http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.GNU.64.64.d-SunOS5.8/make.out for details. ## Comment 3708 Date: 2005-08-23 09:48:50 +0200 From: @sjoerdmullender Logged In: YES user_id=43607 Solaris has define s_addr _S_un._S_addr, so s_addr cannot be used as variable name. ## Comment 3709 Date: 2005-12-19 15:38:21 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: NO TEST / COMPILATION ## Comment 3710 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1265933 at http://sourceforge.net/support/tracker.php?aid=1265933
pathfinder fails to compile on SunOS/Solaris
https://api.github.com/repos/MonetDB/MonetDB/issues/662/comments
0
2020-11-30T08:47:43Z
2024-06-27T11:05:41Z
https://github.com/MonetDB/MonetDB/issues/662
753,281,031
662
[ "MonetDB", "MonetDB" ]
Date: 2005-08-21 12:08:45 +0200 From: @Teggy To: Jan Rittinger &lt;<jan.rittinger>&gt; Version: -- development Last updated: 2005-09-12 03:34:57 +0200 ## Comment 3703 Date: 2005-08-21 12:08:45 +0200 From: @Teggy The creation of element node content seems to exhibit a problem. Works OK: <a> { (1,2,3) } </a> --> <a>1 2 3</a> <a> { for $x in (1,2,3) return $x } </a> --> <a>1 2 3</a> Fails: <a> { for $x in (1,2,3) where $x gt 0 return $x } </a> --> <a/> (: empty element :) This seems odd to me (note that the 'where' predicate lets all bindings of $x pass). ## Comment 3704 Date: 2005-09-12 15:34:57 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; Logged In: YES user_id=993208 The translation recognized a relational join. Its internal core representation was of type 'none' which was recognized as subtype of type 'node*'. Fixed both return type of relational join and subtype relationship of type 'none'. ## Comment 3705 Date: 2005-12-19 15:37:58 +0100 From: @drstmane Logged In: YES user_id=572415 BugDay_2005-12-19, stmane: ALREADY IN TESTWEB: pathfinder/tests/BugsViaSourgeforce/Tests/ID.1265413.element_creation_and_content* ## Comment 3706 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1265413 at http://sourceforge.net/support/tracker.php?aid=1265413
Element creation and element content
https://api.github.com/repos/MonetDB/MonetDB/issues/661/comments
0
2020-11-30T08:47:41Z
2024-06-27T11:05:40Z
https://github.com/MonetDB/MonetDB/issues/661
753,281,003
661
[ "MonetDB", "MonetDB" ]
Date: 2005-08-19 04:18:30 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-08-30 02:03:42 +0200 ## Comment 3699 Date: 2005-08-19 16:18:30 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Test case: sql>select interval '-1-2' year to month; [ -10 ] I'd expect -14, i.e. the first minus sign is a monadic operator for the whole literal. ## Comment 3700 Date: 2005-08-22 16:29:56 +0200 From: @njnes Logged In: YES user_id=43556 Parsing of the minus sign is now done globably instead of only in the first part of the interval string. ## Comment 3701 Date: 2005-12-19 15:41:24 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: ALREADY IN TESTWEB sql/src/test/bugs/interval_parse_bug-sf-1263967.sql ## Comment 3702 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1263967 at http://sourceforge.net/support/tracker.php?aid=1263967
SQL: select interval '-1-2' year to month
https://api.github.com/repos/MonetDB/MonetDB/issues/660/comments
0
2020-11-30T08:47:38Z
2024-06-27T11:05:40Z
https://github.com/MonetDB/MonetDB/issues/660
753,280,977
660
[ "MonetDB", "MonetDB" ]
Date: 2005-08-19 03:14:33 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-08-30 02:04:07 +0200 ## Comment 3695 Date: 2005-08-19 15:14:33 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Test case: sql>select date '2005-08-19' + interval '1' day; [ 2005-08-20 ] works, but sql>select interval '1' day + date '2005-08-19'; doesn't work: ERROR = !binary operator sql_add(date,date) unknown Fortunately, the standard allowed these elements to be commutative under addition: <http://savage.net.au/SQL/sql-2003- 2.bnf.htmldatetime value expression> ## Comment 3696 Date: 2005-08-22 16:31:14 +0200 From: @njnes Logged In: YES user_id=43556 Commutative functions are resolved both ways if needed now. ## Comment 3697 Date: 2005-12-19 15:40:48 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: ALREADY IN TESTWEB sql/src/test/bugs/interval_data_commutative_add_bug-sf-1263928.sql ## Comment 3698 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1263928 at http://sourceforge.net/support/tracker.php?aid=1263928
SQL: select interval ... + date ...
https://api.github.com/repos/MonetDB/MonetDB/issues/659/comments
0
2020-11-30T08:47:35Z
2024-06-27T11:05:39Z
https://github.com/MonetDB/MonetDB/issues/659
753,280,938
659
[ "MonetDB", "MonetDB" ]
Date: 2005-08-19 03:09:49 +0200 From: Jens Teubner &lt;<teubner>&gt; To: @yzchang Version: -- development Last updated: 2005-08-19 04:41:16 +0200 ## Comment 3690 Date: 2005-08-19 15:09:49 +0200 From: Jens Teubner &lt;<teubner>&gt; Compiling the Pathfinder runtime gives me pathfinder.mx: In function `CMDxquery': pathfinder.mx:963: warning: 'dst' might be used uninitialized in this function Indeed, the return value of CMDxquery is undefined if exec_xquery() fails: /* run the query, but collect all output in a buffer */ THRsetdata(0,s); ret = exec_xquery(mode, xquery); THRsetdata(0,o); stream_close(s); if (ret == GDK_SUCCEED) { dst = GDKmalloc(b->pos+1); memcpy(dst, b->buf, b->pos); dst[b->pos] = 0; } buffer_destroy(b); stream_destroy(s); *res = dst; return ret; (Btw., indentation seems to be messed up in this file.) ## Comment 3691 Date: 2005-08-19 16:41:16 +0200 From: @yzchang Logged In: YES user_id=341633 sorry, my compiler is not strict enough:( ## Comment 3692 Date: 2005-08-22 08:27:11 +0200 From: @yzchang Logged In: YES user_id=341633 sorry, my compiler is not strict enough:( ## Comment 3693 Date: 2005-12-19 15:39:54 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / COMPILATION ## Comment 3694 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1263926 at http://sourceforge.net/support/tracker.php?aid=1263926
compilation fails in pathfinder.mx
https://api.github.com/repos/MonetDB/MonetDB/issues/658/comments
0
2020-11-30T08:47:32Z
2024-06-27T11:05:38Z
https://github.com/MonetDB/MonetDB/issues/658
753,280,903
658
[ "MonetDB", "MonetDB" ]
Date: 2005-08-19 03:03:39 +0200 From: @sjoerdmullender To: Peter Boncz &lt;<boncz>&gt; Version: -- development Last updated: 2005-11-25 02:26:47 +0100 ## Comment 3686 Date: 2005-08-19 15:03:39 +0200 From: @sjoerdmullender The third argument of BATfetchjoin (estimate) is declared as size_t. However inside it compares the value with oid_nil. An oid is not necessarily the same as size_t, so this could possibly go wrong. We should check this. The same holds for the other functions that take an estimate. ## Comment 3687 Date: 2005-11-25 02:26:47 +0100 From: @peterboncz Logged In: YES user_id=591107 well, I think that size_t is a datatype with a range always bigger than oid, so I don't think this is a problem. ## Comment 3688 Date: 2005-12-19 15:39:10 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST ## Comment 3689 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1263921 at http://sourceforge.net/support/tracker.php?aid=1263921
BATfetchjoin takes size_t arg for oid
https://api.github.com/repos/MonetDB/MonetDB/issues/657/comments
0
2020-11-30T08:47:29Z
2024-06-28T13:34:46Z
https://github.com/MonetDB/MonetDB/issues/657
753,280,864
657
[ "MonetDB", "MonetDB" ]
Date: 2005-08-19 10:48:17 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-11-26 06:06:17 +0100 ## Comment 3682 Date: 2005-08-19 10:48:17 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Test case: create table t( c time ); insert into t values( time '12:34:56'); select c from t where extract( hour from c ) = 12 and extract( minute from c ) = 34; gives an empty result set: 3 querytype t table_name c name time type 8 length 0 tuplecount 12 id ## Comment 3683 Date: 2005-11-26 18:06:17 +0100 From: @njnes Logged In: YES user_id=43556 The optimization (reducing selects) found the base columns equal but incorrectly concluded the functions didn't alter the column. Added test script src/test/BugTracker/Tests/select_extract.SF-1263796.sql ## Comment 3684 Date: 2005-12-19 15:38:03 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: ALREADY IN TESTWEB sql/src/test/BugTracker/Tests/select_extract.SF-1263796.sql ## Comment 3685 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1263796 at http://sourceforge.net/support/tracker.php?aid=1263796
SQL: select ... where extract( ...) = ... and extract( ...
https://api.github.com/repos/MonetDB/MonetDB/issues/656/comments
0
2020-11-30T08:47:26Z
2024-06-27T11:05:35Z
https://github.com/MonetDB/MonetDB/issues/656
753,280,827
656
[ "MonetDB", "MonetDB" ]
Date: 2005-08-18 05:31:18 +0200 From: @yzchang To: @sjoerdmullender Version: -- development Last updated: 2006-01-16 02:39:48 +0100 ## Comment 3678 Date: 2005-08-18 17:31:18 +0200 From: @yzchang In stream.mx: the implementation of socket_read returns a '0' to indicate no data have been read. This is not necessarily an error, but the implementation still sets errnr to READ_ERROR: s->errnr = READ_ERROR; I think this line code should be removed. ## Comment 3679 Date: 2005-12-19 15:36:35 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST Probably the change should be made and we should see what fails then. ## Comment 3680 Date: 2006-01-16 14:39:48 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 Checked in as revision 1.121 of src/common/stream.mx. ## Comment 3681 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1263146 at http://sourceforge.net/support/tracker.php?aid=1263146
errnr set without occurence of an error
https://api.github.com/repos/MonetDB/MonetDB/issues/655/comments
0
2020-11-30T08:47:23Z
2024-06-27T11:05:35Z
https://github.com/MonetDB/MonetDB/issues/655
753,280,793
655
[ "MonetDB", "MonetDB" ]
Date: 2005-08-18 03:55:26 +0200 From: @yzchang To: GDK devs &lt;<bugs-common>&gt; Version: -- development Last updated: 2005-08-18 10:39:33 +0200 ## Comment 3674 Date: 2005-08-18 15:55:26 +0200 From: @yzchang In stream.mx: the implementation of socket_read returns a '0' to indicate no data have been read. This is not necessarily an error, but the implementation still sets errnr to READ_ERROR: s->errnr = READ_ERROR; I think this line code should be removed. ## Comment 3675 Date: 2005-08-18 22:39:33 +0200 From: @sjoerdmullender Logged In: YES user_id=43607 Duplicate of http://sourceforge.net/support/tracker.php?aid=1263146. ## Comment 3676 Date: 2005-12-19 15:34:04 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / REDUNDANT ## Comment 3677 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1263042 at http://sourceforge.net/support/tracker.php?aid=1263042 The original assignee of this bug does not have an account here. Reassigning to the default assignee for the component, bugs-common@monetdb.org. Previous assignee was nobody@users.sourceforge.net. This bug was marked DUPLICATE in the database it was moved from. Changing resolution to "MOVED"
errnr set without occurence of an error
https://api.github.com/repos/MonetDB/MonetDB/issues/654/comments
0
2020-11-30T08:47:20Z
2024-06-28T13:34:46Z
https://github.com/MonetDB/MonetDB/issues/654
753,280,755
654
[ "MonetDB", "MonetDB" ]
Date: 2005-08-17 09:53:55 +0200 From: @sjoerdmullender To: Jan Flokstra &lt;<jflokstra>&gt; Version: -- development Last updated: 2005-09-07 02:49:45 +0200 ## Comment 3666 Date: 2005-08-17 09:53:55 +0200 From: @sjoerdmullender Shred a document (e.g. <a><b>c</b><d><e/><f/></d><g a="42"/></a>) and then look at the produced bats (with e.g. print(bat("pre_size0"),bat("pre_level0").[+](0),bat("pre_prop0"),bat("pre_kind0").[+](0)); ). Do this on a 64-bit machine. The result show an interesting value in the prop column: MonetDB>print(bat("pre_size0"),bat("pre_level0").[+](0),bat("pre_prop0"),bat("pre_kind0").[+](0)); --------------------------------------------------------- h t t t t name void int int oid int type --------------------------------------------------------- [ 0@0, 10, -1, 18446744071562067968@0, 4 ] [ 1@0, 8, 0, 0@0, 0 ] [ 2@0, 2, 1, 1@0, 0 ] ... Looking more closely at the value: 18446744071562067968 == 0xffffffff80000000. Some investigation revealed: The value comes from a (nat) value nil which is passed as an argument to direct_oid2bat (in shredder.mx:emitnode). The corresponding argument (the fourth) of direct_oid2bat is an oid. So here there is an automatic type conversion from nat to oid. The type nat is defined as int near the top of shredder.mx. There is separate code for 32 and 64 bit systems, but they are currently identical. Note that the value nat (int) nil is represented as 0x80000000 and the type conversion causes a sign extension to 64 bits, resulting in the observed value. I experimented with chaning the type definition of nat to if (SIZEOF_OID==4) /* 32 bits systems */ typedef int nat; define NIL ((nat) INT_MIN) define NATFMT "u" else typedef unsigned long long nat; define NIL ((nat) LLONG_MIN) ifdef HAVE___INT64 define NATFMT "I64u" else define NATFMT "llu" endif However this gave too many errors in the tests. Note that the if test is also different from the original. The original has if (SIZEOF_VOID_P==4) whereas I have if (SIZEOF_OID==4) The comment (elided here) says that the type of a nat has to correspond to some other type in a bat, but it is totally unclear to me which type that should be, and so I wasn't able to check that. So two bugs really: - nat -> oid conversion is not correct, especially the value of nil is not converted to oid_nil. - the comments near the typedef of nat are incomplete. ## Comment 3667 Date: 2005-08-17 11:09:46 +0200 From: Jan Flokstra &lt;<jflokstra>&gt; Logged In: YES user_id=1054297 Repaired the bug by changing the function direct_oid2bat(..., oid v) to direct_natoid2bat(.... nat_v). I now check if this nat_v == NIL and if so I insert 'oid_nil' in the BAT. This should solve the problem. I cannot really test it myself because we have only a working 64bit monet with 32 bit oid's. When this all works I think I will remove ALL 32/64 bit ifdefs because in the current implementation it should work without these. They were inherited from the old loader by Torsten. ## Comment 3668 Date: 2005-08-17 11:42:22 +0200 From: @drstmane Logged In: YES user_id=572415 Maybe this is also the right time to re-think the usage of "nat" at all. Wouldn't is be more natural in a MonetDB context to stick to the proper MonetDB data types (chr, sht, int, lng, oid)? ... and btw, there's a simple way to get a working 64bit MonetDB with 64bit oid: just compile it ;-) (I'm "juggling" with more than 4(!) different MonetDB versions on a daily basis...) (if you need access to a 64-bit machine, just let me know!) ## Comment 3669 Date: 2005-08-17 12:26:18 +0200 From: Jan Flokstra &lt;<jflokstra>&gt; Logged In: YES user_id=1054297 Last time I tried the 64bit-arch, 64bit oid version it crashed. But I tried again on our new 'teehuis' and now it worked. The fix works, there's a correct [0@0, nil] tuple in bat("pre_prop0") and printing the doc again also succeeds. It complains BAT[ATTR_PROP] is not dense but this is another problem I think. I will have a look at this too. I will also have a look at the eventual conversion of nat to oid. ## Comment 3670 Date: 2005-09-05 08:10:07 +0200 From: @drstmane Logged In: YES user_id=572415 Any news on this one? ## Comment 3671 Date: 2005-09-07 14:49:44 +0200 From: Jan Flokstra &lt;<jflokstra>&gt; Logged In: YES user_id=1054297 Fixed and closed. Removed 32/64bit arch ifdefs and removed confusing comments. Furthermore made provisions for the conversion of nat->oid but did not use this solution by default because a lot of nat's are counters and not oid's. Maybe in the future we should retype them into posint's. ## Comment 3672 Date: 2005-12-19 15:32:53 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS pathfinder/tests/BugDay_2005-12-19_0.9.3/Tests/nat_oid_conv.SF-1261765.milS ## Comment 3673 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1261765 at http://sourceforge.net/support/tracker.php?aid=1261765
problematic nat -> oid conversion on 64 bit system
https://api.github.com/repos/MonetDB/MonetDB/issues/653/comments
0
2020-11-30T08:47:17Z
2024-06-27T11:05:33Z
https://github.com/MonetDB/MonetDB/issues/653
753,280,720
653
[ "MonetDB", "MonetDB" ]
Date: 2005-08-16 12:59:06 +0200 From: Jens Teubner &lt;<teubner>&gt; To: @sjoerdmullender Version: -- development Last updated: 2006-01-24 08:52:40 +0100 ## Comment 3651 Date: 2005-08-16 12:59:06 +0200 From: Jens Teubner &lt;<teubner>&gt; Building Pathfinder currently requires a recent version of bison to be available on the build system (>=1.33, I think). Some platforms (e.g., Mac OS X) don't provide this by default. I think the idea in the autotools is that users are not required to have flex/bison installed on their system at all. Tarballs should ship with pre-built .c files, processed with flex/bison before making the tarball. Flex/bison is only required if the corresponding .l/.y files change, i.e., on developers' machines. I tested simply setting EXTRA_DIST in the respective Makefile.ag files. This results in the .c files being shipped in the tarball, and Pathfinder compiles fine even without flex/bison. On `make clean', however, the .c files are deleted, and a subsequent build without flex/bison will fail. My understanding would be that .c files built from flex/bison sources should be shipped with distribution tarballs, and they should not be deleted by `make clean'. This way, Pathfinder would easily compile without flex/bison. The ./configure script, by the way, tests for the availability of flex/ bison. If a tool is not found, the corresponding variable will be unset. This is (intended) default behaviour of the autotools (as flex/bison are not really requirements anyway). However, it leads to misleading error messages when building Pathfinder (The `d' is the empty variable for the bison executable, plus the `-d' option for bison.): Making all in burg lockfile -r 2 waiting d ../../dl/pathfinder-0.8.2/burg/gram.y || { rm -f waiting ; exit 1 ; } /bin/sh: line 1: d: command not found AFAIK, Fabian stumbled on the same problem, when preparing ebuilds for OS X. ## Comment 3652 Date: 2005-08-16 14:25:54 +0200 From: @grobian Logged In: YES user_id=963970 I "solved" this in the Gentoo ebuild by unpacking an additional tarball with the generated bison sources. This works great on OSX, but isn't the nicest solution of course. The additional tarball is uploaded on the Gentoo mirrors and only used for OSX users on Gentoo for now. It would be great if the sources supplied these files by default, as well as a "make clean-bison-only" or alike command for users that like to generate the sources themselves... ## Comment 3653 Date: 2005-11-21 16:25:23 +0100 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 Seems like it is even worse than I had written it in August. Apparently, MonetDB's ./configure does *not* check for the presence of bison/yacc. If nothing is found, $YACC is simply set to 'yacc', resulting in yacc -d mel.yy || { rm -f waiting ; exit 1 ; } /bin/sh: line 1: yacc: command not found make[3]: *** [mel.tab.h] Error 1 I think we should find a means to compile MonetDB (and Pathfinder) without lex/flex/yacc/bison installed on the system. Most software packages I know install fine without them. If this is not possible, a proper ./configure error is a must, IMHO. Btw. I left "Group" at "Pathfinder CVS". However, this affects *all* MonetDB packages, CVS and stable. ## Comment 3654 Date: 2005-11-21 16:39:32 +0100 From: @drstmane Logged In: YES user_id=572415 I completely agree with you, Jens. I still "dream" of a source distribution (tarball and/or src.rpm) where all preprocessing has been done, and hence, neither lex/flex/yacc/bison not the buildtools are required to compile from that source distribution. Though this might not be achieved easily, I propose to have this done (or found evidence that it is not possible) before the next release (i.e., no release until all bug reports including this one are properly closed) --- details about the upcoming release plans will be discussed here at CWI tomorrow or on Wednesday... ## Comment 3655 Date: 2005-11-21 17:56:34 +0100 From: @grobian Logged In: YES user_id=963970 I can redo the trick I did for Gentoo (in fact I will have to ;) ) and make the sources available on our sourceforge stuff here, then let Gentoo mirror from there. The "additional sources" are now solely hosted on the Gentoo mirrors. ## Comment 3656 Date: 2005-11-21 18:27:06 +0100 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 Thanks Fabian for your offer. However, I guess you are realizing yourself that such "hacks" are not the right way how we should address this problem, and you'd have to redo that hack on every release/ installation. Users outside the MonetDB developer community might run into the same problem (in fact, chances that non- developers don't have the respective tools installed are even higher). I will find ways to get MonetDB installed here, don't worry. But I would still prefer a proper solution in order to make the installation hassle-free for potential MonetDB users. Of course, I'd be willing to help. My knowledge on the MonetDB build process is close to zero, but I could help, e. g., with testing here in Munich. Anyway, thanks for caring so much about me :-) ## Comment 3657 Date: 2005-11-21 19:12:02 +0100 From: @grobian Logged In: YES user_id=963970 Let me put it like this: From where I'm sitting what you propose it far from what MonetDB should support. The problem of bison here (that I indeed solved with a very dirty hack on OSX) is directly related to the OS/distro that is being installed on. A regular Linux system won't have any problems with satisfying the dependencies for XQuery or any other MonetDB product. The problem is in the rather crappy stock install Apple delivers to the user. Solaris might be the same problem. If bison was not the problem, then the autotools would have been. I deal with many many packages, and to be honest, I don't know of any upstream that includes generated bison/yacc files. The most they do is simply make their stuff backwards compatible whenever this is easily possible (often done with python for True/False I think). There is a good reason for not including such generated bison files, since a newer version of bison might be able to do smarter things, or fix bugs. While diving into this and actually trying to get the most ancient archs compiling without any help is a quite generous thing to do, I believe it draws a far to big burden on the MonetDB project (probably Stefan...). For architectures that are problematic, we should just have nightly builds available. IMHO, MonetDB would be anything but the first project to just require some decent recent versions of the used tools. As practically for OSX: Fink and DarwinPorts are able to supply more uptodate versions of the required tools. Gentoo can build MonetDB using the little cheat I made, but in the future it will also not need this for OSX, because at this very moment I have access to a prefixed install that allows me to install those new tools, like Fink does. For any other arch such as Solaris: in the end Gentoo for Solaris will work there too. Binary installations are for (simple) users, source installations are for the advanced users (that should be able to deal with the problem here). Feel free to disagree ;) ## Comment 3658 Date: 2005-11-21 22:54:37 +0100 From: @drstmane Logged In: YES user_id=572415 Well, of course there are various solutions. The minimal one is to properly document that MonetDB requires (among others) at least bison 1.35 to compile from the (CVS) sources, and to properly check for this in configure, and abort compilation in case no proper bison is found. Then it's the task of who ever tries to compile MonetDB to ensure that his/her machine meats the requirements. For the CVS sources, there is no other option. On our MacOS X testing machine, I simply installed bison-1.35 in my home --- works even simpler than installing MonetDB: configure; make; make install ;-) The ebuild (for MacOS X, only?) should then also depend on a bison >=1.35 ebuild --- if there is none, yet, we could consider making one --- cannot be more difficult than making an ebuild for MonetDB, right? For the source distributions (tarball, src.rpm), we should try to make them compile without lex/flex/bison/yacc, and also without the buildtools (Mx, mel, burg); this would also simplify documentation, packaging and cross-compiling. The major hurdel I sees right now is that we cannot have .mx.in files any more; if/where possible, we should to turn them into pure .mx files, i.e., eliminate the modifications done by configure; where/if this doesn't work, we have to think of another solution (turning .mx.in into .in.mx is obviously no solution). Actually, thinking about this, the cleanest solution for all the (platform independent?) pre-processing (buildtools, lex/flex/bison/yacc) would be to do it during the bootstrap; then, the "clean" source distributions are for free ... ## Comment 3659 Date: 2005-11-22 09:35:21 +0100 From: @grobian Logged In: YES user_id=963970 > On our MacOS X testing machine, I simply installed > bison-1.35 in my home --- works even simpler than installing > MonetDB: configure; make; make install ;-) > The ebuild (for MacOS X, only?) should then also depend on a > bison >=1.35 ebuild --- if there is none, yet, we could > consider making one --- cannot be more difficult than making > an ebuild for MonetDB, right? Forget about the ebuild. Dependencies as you describe are handled like that. I only cheated for OSX because it otherwise would block on bison, which cannot be installed since the OS already provides this. This problem is being solved, so the only relevant part is the source installations. I argue that for source installations, you could simply let configure die when it does not find what it wants to have. Your "clean" source installation idea is pretty much cool since you don't need the buildtools for it, but I doubt whether you can get it running. What I remember from last time we tried to do it together is that we stumbled upon so many problems that it wasn't fun any more (hence we were not able to ship updated sources to the students). ## Comment 3660 Date: 2005-11-22 09:49:46 +0100 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 > I argue that for source installations, you could simply > let configure die when it does not find what it wants to > have. Your "clean" source installation idea is pretty > much cool since you don't need the buildtools for it, but > I doubt whether you can get it running. I don't have any clue about Mx. But for lex/yacc, I read in the automake manual: ``The intermediate files generated by `yacc' (or `lex') will be included in any distribution that is made. That way the user doesn't need to have `yacc' or `lex'.'' So in some way, this functionality appears to be readily provided by automake. At least for Pathfinder, we don't have any Mx generated lex/yacc files. So I think I am not "dreaming" when I ask for a "clean" distribution. If I understand correctly, lex/yacc need not be called during bootstrapping. Instead, the `dist' make target will take care of building the required .c files and put them into the source tarball. I will look into this right now... ## Comment 3661 Date: 2005-11-22 11:41:28 +0100 From: Jens Teubner &lt;<teubner>&gt; Logged In: YES user_id=731390 Ok, I've now looked into the Pathfinder sources, where it turned out to be rather simple to remove the dependency on lex/yacc for users. For now, I have edited Makefile.am files manually; I'm an absolute novice in Python and autogen, where a proper fix needed to be made. The only files affected are scanner.l and parser.y in compiler/parser. So first I have cleaned up the Makefile.am. I have replaced the *_SOURCES lines with a simple libparser_la_SOURCES = scanner.l parser.y abssyn.c and eliminated *all* explicitly listed dependencies. I've also removed the BUILT_SOURCES, MOSTLYCLEANFILES, and EXTRA_DIST variables (though we might still want to fill the latter with Makefile.ag and Makefile.msc). Then I stumbled over a minor problem: The lexer depends on the file parser.tab.h, generated by yacc/bison. The automake way to handle this is to add the respective header file to BUILT_SOURCES: BUILT_SOURCES = parser.h Apparently, automake prefers `parser.h' over `parser.tab.h', so I renamed it accordingly in scanner.l. Pathfinder currently does have a parser.h in its include directory. To avoid confusion and problems with include paths, I simply removed include/parser.h and added its declarations to abssyn.h (and adapted include commands in source files accordingly). After running autoheader, automake, and autoconf, `make dist' will add the generated sources to the distribution tarball, and everything compiles fine without lex/yacc. I did not have the `make clean' problem mentioned in the original bug report. (I did, however, discover a different problem with .m files in the distribution tarball on the SourceForge page. If ./configure is called within the source directory, a `make distclean' apparently deletes a bit too much. A subsequent ./configure ; make fails with a missing pathfinder.m.) So apart from the necessary modification of autogen, does anything else hinder us from removing the dependency on lex/yacc? ## Comment 3662 Date: 2005-12-19 15:16:37 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / COMPILATION I'd say this is actually an RFE. not a bug. ## Comment 3663 Date: 2006-01-22 19:39:35 +0100 From: @drstmane Logged In: YES user_id=572415 While we managed to get all other buildtools out of the source distribution, there was/is no time left to do (and double-check!) the same for yacc/bison/[f]lex before this release. Postponed until after the release. ## Comment 3664 Date: 2006-01-24 20:52:40 +0100 From: @grobian Logged In: YES user_id=963970 Stefan did a better job than he thought himself. It just works and looks as if this idea is implemented. Closing bug. ## Comment 3665 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1260796 at http://sourceforge.net/support/tracker.php?aid=1260796
Building without bison >=1.33
https://api.github.com/repos/MonetDB/MonetDB/issues/652/comments
0
2020-11-30T08:47:14Z
2024-06-27T11:05:31Z
https://github.com/MonetDB/MonetDB/issues/652
753,280,690
652
[ "MonetDB", "MonetDB" ]
Date: 2005-08-15 10:04:06 +0200 From: @grobian To: @sjoerdmullender Version: -- development Last updated: 2006-02-07 11:57:26 +0100 ## Comment 3646 Date: 2005-08-15 10:04:06 +0200 From: @grobian On Gentoo, during the nightly testing, the following assertion is raised for several tests: Mserver: dl-open.c:438: dl_open_worker: Assertion `imap->l_type == lt_loaded' failed. ## Comment 3647 Date: 2005-12-19 15:13:29 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / REDUNDANT ## Comment 3648 Date: 2006-01-15 16:23:47 +0100 From: @drstmane Logged In: YES user_id=572415 last night, the bug occured with `make check` (aka `Mtest.py -d10`) only for one test: http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.32.32.d-Gentoo1.6.13/tests_XQuery/fun7.err.00.html with `Mtest.py -d0`, it occured last night for 4 tests: http://monetdb.cwi.nl/testing/projects/monetdb/Current/sql/.mTests03/GNU.32.32.d-Gentoo1.6.13/src_benchmarks_arno/create.err.00.html http://monetdb.cwi.nl/testing/projects/monetdb/Current/sql/.mTests03/GNU.32.32.d-Gentoo1.6.13/src_test_BugDay_2005-10-06_2.9.3/rollback_and_autocommit.SF-926453.err.00.html http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests03/GNU.32.32.d-Gentoo1.6.13/tests_BugsViaSourgeforce/ID.1206985.err.00.html http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests03/GNU.32.32.d-Gentoo1.6.13/tests_XQuery/instance1.err.00.html I suppose, though, the list of test for which this bug happens is quite random... ## Comment 3649 Date: 2006-02-07 23:57:26 +0100 From: @njnes Logged In: YES user_id=43556 This has to do with the random crashes bug. So closing this one. Because of the 'exit without cleanup' this is also gone. ## Comment 3650 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1259668 at http://sourceforge.net/support/tracker.php?aid=1259668 This bug was marked DUPLICATE in the database it was moved from. Changing resolution to "MOVED"
Testing: Gentoo testing sometimes asserts in Mserver
https://api.github.com/repos/MonetDB/MonetDB/issues/651/comments
0
2020-11-30T08:47:10Z
2024-06-28T13:34:45Z
https://github.com/MonetDB/MonetDB/issues/651
753,280,645
651
[ "MonetDB", "MonetDB" ]
Date: 2005-08-12 12:44:28 +0200 From: @Teggy To: Jan Rittinger &lt;<jan.rittinger>&gt; Version: -- development Last updated: 2005-09-12 03:36:56 +0200 ## Comment 3642 Date: 2005-08-12 12:44:28 +0200 From: @Teggy Currently, this query fails at runtime (while it shoud not): let $a := <a> <b c="foo"/> <d/> </a> return $a/(d|b/@c) ERROR: cannot handle attributes in function pf:distinct-doc-order. The problem seems to be that pf:distinct-doc-order is unable to order the 'c' attribute node with respect to the 'd' element node (the XPath child axis step is required to return its result in document order). According to the XQuery spec, attribute nodes immediately follow their owner elements in document order (although the order of attributes sharing the same owner is undefined). Clearly, attribute node 'c' should precede element node 'd' here. Can/should we modify pf:distinct-doc-order such that attributes are ordered according to the document order rank of their owner elements? Nice weekend all, --Teggy ## Comment 3643 Date: 2005-09-12 15:36:56 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; Logged In: YES user_id=993208 Added code which copes with attributes exactly as described in the bug report (order on iter|frag|pre|attr). ## Comment 3644 Date: 2005-12-19 14:52:15 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: ALREADY IN TESTWEB pathfinder/tests/BugsViaSourgeforce/Tests/ID.1257463.distinct-doc-order.xq ## Comment 3645 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1257463 at http://sourceforge.net/support/tracker.php?aid=1257463
Document order and attribute nodes
https://api.github.com/repos/MonetDB/MonetDB/issues/650/comments
0
2020-11-30T08:47:07Z
2024-06-27T11:05:29Z
https://github.com/MonetDB/MonetDB/issues/650
753,280,601
650
[ "MonetDB", "MonetDB" ]
Date: 2005-08-11 05:19:05 +0200 From: @yzchang To: @sjoerdmullender Version: -- development Last updated: 2005-08-12 04:41:47 +0200 ## Comment 3638 Date: 2005-08-11 17:19:05 +0200 From: @yzchang Failed to compile the last version of shredder.mx. Got the following error during "make": gcc32 -m64 -DHAVE_CONFIG_H -I. -I../../runtime -I.. -I../../runtime -I/ufs/zhang/install/monetdb/include/MonetDB -I/ufs/zhang/install/monetdb/include/MonetDB/common -I/ufs/zhang/install/monetdb/include/MonetDB/gdk -I/ufs/zhang/install/monetdb/include/MonetDB/monet -I/ufs/zhang/install/monetdb/include/MonetDB/plain -I/ufs/zhang/install/monetdb/include/MonetDB/contrib -I/ufs/zhang/install/monetdb/include/MonetDB/mapi -I/ufs/zhang/install/monetdb/include/MonetDB/C -I/usr/include/libxml2 -I../compiler/include -I../../runtime/../compiler/include -DLIBPF_SUPPORT -std=c99 -Wall -W -g -Werror-implicit-function-declaration -Werror -Wno-unused-function -Wno-unused-label -DEMBEDDED -DNO_CGI -DNO_AUTH -c shredder.c -fPIC -DPIC -o .libs/lib_pf_support_la-shredder.o ../../runtime/shredder.mx: In function `pf_compress': ../../runtime/shredder.mx:414: implicit declaration of function `BATins_kunique' cc1: warnings being treated as errors ../../runtime/shredder.mx:414: warning: assignment makes pointer from integer without a cast make[2]: *** [lib_pf_support_la-shredder.lo] Error 1 make[2]: Leaving directory `/ufs/zhang/src/pathfinder/build/runtime' make[1]: *** [install] Error 2 make[1]: Leaving directory `/ufs/zhang/src/pathfinder/build/runtime' make: *** [install-recursive] Error 1 ## Comment 3639 Date: 2005-08-11 17:36:53 +0200 From: @sjoerdmullender Logged In: YES user_id=43607 Did you update MonetDB as well? ## Comment 3640 Date: 2005-12-19 14:48:26 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / COMPILATION ## Comment 3641 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1256794 at http://sourceforge.net/support/tracker.php?aid=1256794
Failed to compile shredder.mx
https://api.github.com/repos/MonetDB/MonetDB/issues/649/comments
0
2020-11-30T08:47:05Z
2024-06-27T11:05:28Z
https://github.com/MonetDB/MonetDB/issues/649
753,280,563
649
[ "MonetDB", "MonetDB" ]
Date: 2005-08-11 11:08:10 +0200 From: @drstmane To: @sjoerdmullender Version: -- development Last updated: 2005-08-11 11:49:30 +0200 ## Comment 3634 Date: 2005-08-11 11:08:10 +0200 From: @drstmane /bin/sh ../../../libtool --mode=compile gcc -m32 -DHAVE_CONFIG_H -I. -I/var/tmp/monet.GNU_32-d.12794/MonetDB/src/modules/plain -I../../.. -I/var/tmp/monet.GNU_32-d.12794/MonetDB/src/modules/plain -I../../common -I../../gdk -I../../monet -I/Users/monet/soft/pcre-5.0/include -DLIBLOGGER -O2 -std=c99 -Wall -W -O6 -fomit-frame-pointer -finline-functions -Werror-implicit-function-declaration -Werror -Wno-unused-function -Wno-unused-label -Wno-uninitialized -D_REENTRANT -c -o lib_logger_la-logger.lo `test -f 'logger.c' || echo '/var/tmp/monet.GNU_32-d.12794/MonetDB/src/modules/plain/'`logger.c gcc -m32 -DHAVE_CONFIG_H -I. -I/var/tmp/monet.GNU_32-d.12794/MonetDB/src/modules/plain -I../../.. -I/var/tmp/monet.GNU_32-d.12794/MonetDB/src/modules/plain -I../../common -I../../gdk -I../../monet -I/Users/monet/soft/pcre-5.0/include -DLIBLOGGER -O2 -std=c99 -Wall -W -O6 -fomit-frame-pointer -finline-functions -Werror-implicit-function-declaration -Werror -Wno-unused-function -Wno-unused-label -Wno-uninitialized -D_REENTRANT -c logger.c -fno-common -DPIC -o .libs/lib_logger_la-logger.o cc1: warnings being treated as errors /var/tmp/monet.GNU_32-d.12794/MonetDB/src/modules/plain/logger.mx: In function 'log_read_format': /var/tmp/monet.GNU_32-d.12794/MonetDB/src/modules/plain/logger.mx:215: warning: passing argument 2 of 'stream_readInt' from incompatible pointer type make[5]: *** [lib_logger_la-logger.lo] Error 1 make[4]: *** [all] Error 2 See also http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.GNU_32-d/Darwin8.2.0/make.out . ## Comment 3635 Date: 2005-08-11 11:49:30 +0200 From: @sjoerdmullender Logged In: YES user_id=43607 Fixed in CVS by adding a cast. I'll close this bug report, even though I didn't actually test anything. :-) If the bug is not actually fixed, please reopen. ## Comment 3636 Date: 2005-12-19 14:47:54 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / COMPILATION ## Comment 3637 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1256535 at http://sourceforge.net/support/tracker.php?aid=1256535
module "logger" fails to compiler on Darwin
https://api.github.com/repos/MonetDB/MonetDB/issues/648/comments
0
2020-11-30T08:47:02Z
2024-06-27T11:05:27Z
https://github.com/MonetDB/MonetDB/issues/648
753,280,536
648
[ "MonetDB", "MonetDB" ]
Date: 2005-08-09 04:53:47 +0200 From: Wouter Alink &lt;<alink>&gt; To: @sjoerdmullender Version: -- development Last updated: 2005-08-10 01:22:17 +0200 ## Comment 3630 Date: 2005-08-09 16:53:47 +0200 From: Wouter Alink &lt;<alink>&gt; typing help("hot") gives the result shown below. should hot() return void or return a bat? --------------- PROC: hot(BAT[any::1,any::2]) : void no text available COMMAND: hot(str) : void MODULE: bat COMPILED: by adm on Thu Jul 28 10:15:18 2005 Makes a BAT very hot for the BBP. The chance of being chosen for swapout is small, afterwards. Returns the BAT itself. ## Comment 3631 Date: 2005-08-10 13:22:17 +0200 From: @drstmane Logged In: YES user_id=572415 fixed in CVS: neither hot() nor cold() return anything; fixed their help messages accordingly. ## Comment 3632 Date: 2005-12-19 14:47:12 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / DOCUMENTATION ## Comment 3633 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1255017 at http://sourceforge.net/support/tracker.php?aid=1255017
Documentation: hot()
https://api.github.com/repos/MonetDB/MonetDB/issues/647/comments
0
2020-11-30T08:46:59Z
2024-06-27T11:05:26Z
https://github.com/MonetDB/MonetDB/issues/647
753,280,507
647
[ "MonetDB", "MonetDB" ]
Date: 2005-08-08 11:44:19 +0200 From: @grobian To: @njnes Version: -- development Last updated: 2006-01-19 08:11:30 +0100 ## Comment 3622 Date: 2005-08-08 11:44:19 +0200 From: @grobian When running the server as follows: % Mserver Monet Database Server V4.8.3 Copyright (c) 1993-2005, CWI. All rights reserved. Compiled for x86_64-redhat-linux-gnu/64bit with 64bit OIDs; dynamically linked. Visit http://monetdb.cwi.nl/ for further information. MonetDB>module(sql_server,mapi,monettime); MonetDB>mapi_start(); Then starting a MapiClient: % MapiClient -lmil Monet Database Server V4.8.3 Copyright (c) 1993-2005, CWI. All rights reserved. Compiled for x86_64-redhat-linux-gnu/64bit with 64bit OIDs; dynamically linked. Visit http://monetdb.cwi.nl/ for further information. mil>quit(); MAPI = anonymous@localhost:50000 ACTION= read_line QUERY = quit(); ERROR = Connection terminated The server appears to have crashed: MonetDB>Segmentation fault while, when doing it as follows: % Mserver Monet Database Server V4.8.3 Copyright (c) 1993-2005, CWI. All rights reserved. Compiled for x86_64-redhat-linux-gnu/64bit with 64bit OIDs; dynamically linked. Visit http://monetdb.cwi.nl/ for further information. MonetDB>module(sql_server); MonetDB>module(mapi); MonetDB>module(monettime); MonetDB>mapi_start(); the server keeps on running. Perhaps when starting all modules in one module() call the refcounting of the modules goes wrong. ## Comment 3623 Date: 2005-08-08 13:33:54 +0200 From: @arjenpdevries Logged In: YES user_id=572978 I think it is the order in which modules are loaded that is the problem; must be related to the code generated by MEL for unloading modules. We recently encountered something similar, and the workaround was to trace the dependency graph of the modules, and make sure all are loaded before using MapiClient. So, maybe this does not lead to segmentation fault: module(streams,ascii_io); module(mapi); module(sqlserver); (sorry for not having submitted the bug report before - mea culpa) ## Comment 3624 Date: 2005-08-08 18:16:29 +0200 From: @arjenpdevries Logged In: YES user_id=572978 I think it is the order in which modules are loaded that is the problem; must be related to the code generated by MEL for unloading modules. We recently encountered something similar, and the workaround was to trace the dependency graph of the modules, and make sure all are loaded before using MapiClient. So, maybe this does not lead to segmentation fault: module(streams,ascii_io); module(mapi); module(sqlserver); (sorry for not having submitted the bug report before - mea culpa) ## Comment 3625 Date: 2005-12-19 14:45:56 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / FAILURE sql/src/test/BugDay_2005-12-19_2.9.3/Tests/multi-module_crash.SF-1254041.py This bug is very likely the same as http://sourceforge.net/support/tracker.php?aid=1385152. ## Comment 3626 Date: 2006-01-15 03:06:24 +0100 From: @drstmane Logged In: YES user_id=572415 Most probably also related to 1400200 "Mserver: module-order does matter" https://sourceforge.net/tracker/index.php?func=detail&aid=1400200&group_id=56967&atid=482468 and/or 1385247: "Mserver assertion failed with module(sql_server,mapi);" http://sourceforge.net/tracker/index.php?func=detail&aid=1385247&group_id=56967&atid=482468 ## Comment 3627 Date: 2006-01-18 19:38:03 +0100 From: @drstmane Logged In: YES user_id=572415 Fixed by Niels' " too many bugs related to module(x,y,z) so we simply remove this feature ie use module(x);module(y);module(z); instead. ". To be closed tomorrow, once confirmed by the test web; cf. http://monetdb.cwi.nl/testing/projects/monetdb/Current/sql/.mTests103/GNU.64.64.d-Fedora4/src_test_BugDay_2005-12-19_2.9.3/multi-module_crash.SF-1254041.out.00.html http://monetdb.cwi.nl/testing/projects/monetdb/Current/sql/.mTests103/GNU.64.64.d-Fedora4/src_test_BugDay_2005-12-19_2.9.3/multi-module_crash.SF-1254041.err.00.html ## Comment 3628 Date: 2006-01-19 08:11:30 +0100 From: @drstmane Logged In: YES user_id=572415 TestWeb confirmed the fix. Closed. ## Comment 3629 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1254041 at http://sourceforge.net/support/tracker.php?aid=1254041
server crashes when multiple modules are supplied
https://api.github.com/repos/MonetDB/MonetDB/issues/646/comments
0
2020-11-30T08:46:56Z
2024-06-27T11:05:25Z
https://github.com/MonetDB/MonetDB/issues/646
753,280,472
646
[ "MonetDB", "MonetDB" ]
Date: 2005-08-03 04:34:49 +0200 From: @drstmane To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: -- development Last updated: 2005-08-03 07:28:51 +0200 ## Comment 3618 Date: 2005-08-03 16:34:49 +0200 From: @drstmane Pathfinder fails to compile, as runtime/shttpd.c fails to compile: ======== /bin/sh ../libtool --tag=CC --mode=compile gcc -m64 -DHAVE_CONFIG_H -I. -I/var/tmp/monet.GNU_64-d.21003/pathfinder/runtime -I.. -I/var/tmp/monet.GNU_64-d.21003/pathfinder/runtime -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB/common -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB/gdk -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB/monet -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB/plain -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB/contrib -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB/mapi -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB/C -I/usr/include/libxml2 -I../compiler/include -I/var/tmp/monet.GNU_64-d.21003/pathfinder/runtime/../compiler/include -DLIBPF_HTTPD -O2 -std=c99 -Wall -W -O6 -fomit-frame-pointer -finline-functions -falign-loops=4 -falign-jumps=4 -falign-functions=4 -fexpensive-optimizations -funroll-loops -frerun-cse-after-loop -frerun-loop-opt -Werror-implicit-function-declaration -Werror -Wno-unused-function -Wno-unused-label -DEMBEDDED -DNO_CGI -DNO_AUTH -c -o lib_pf_httpd_la-shttpd.lo `test -f 'shttpd.c' || echo '/var/tmp/monet.GNU_64-d.21003/pathfinder/runtime/'`shttpd.c gcc -m64 -DHAVE_CONFIG_H -I. -I/var/tmp/monet.GNU_64-d.21003/pathfinder/runtime -I.. -I/var/tmp/monet.GNU_64-d.21003/pathfinder/runtime -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB/common -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB/gdk -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB/monet -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB/plain -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB/contrib -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB/mapi -I/var/tmp/monet.GNU_64-d.21003/MonetDB/.GNU_64-d/Fedora4/include/MonetDB/C -I/usr/include/libxml2 -I../compiler/include -I/var/tmp/monet.GNU_64-d.21003/pathfinder/runtime/../compiler/include -DLIBPF_HTTPD -O2 -std=c99 -Wall -W -O6 -fomit-frame-pointer -finline-functions -falign-loops=4 -falign-jumps=4 -falign-functions=4 -fexpensive-optimizations -funroll-loops -frerun-cse-after-loop -frerun-loop-opt -Werror-implicit-function-declaration -Werror -Wno-unused-function -Wno-unused-label -DEMBEDDED -DNO_CGI -DNO_AUTH -c /var/tmp/monet.GNU_64-d.21003/pathfinder/runtime/shttpd.c -fPIC -DPIC -o .libs/lib_pf_httpd_la-shttpd.o /var/tmp/monet.GNU_64-d.21003/pathfinder/runtime/shttpd.c: In function â<80><98>get_dirâ<80><99>: /var/tmp/monet.GNU_64-d.21003/pathfinder/runtime/shttpd.c:1335: error: â<80><98>S_IFDIRâ<80><99> undeclared (first use in this function) /var/tmp/monet.GNU_64-d.21003/pathfinder/runtime/shttpd.c:1335: error: (Each undeclared identifier is reported only once /var/tmp/monet.GNU_64-d.21003/pathfinder/runtime/shttpd.c:1335: error: for each function it appears in.) /var/tmp/monet.GNU_64-d.21003/pathfinder/runtime/shttpd.c: In function â<80><98>decideâ<80><99>: /var/tmp/monet.GNU_64-d.21003/pathfinder/runtime/shttpd.c:2524: error: â<80><98>S_IFDIRâ<80><99> undeclared (first use in this function) make[3]: *** [lib_pf_httpd_la-shttpd.lo] Error 1 ======== See also http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.GNU_64-d/Fedora4/make.out . ## Comment 3619 Date: 2005-08-03 19:28:51 +0200 From: @drstmane Logged In: YES user_id=572415 fixed in CVS: To ensure portability, please make sure that "pf_config.h" is the first include file in each .c file, either directly or via "pathfinder.h" ! ## Comment 3620 Date: 2005-12-19 13:56:22 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / COMPILATION ## Comment 3621 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1251087 at http://sourceforge.net/support/tracker.php?aid=1251087
Pathfinder fails to compile
https://api.github.com/repos/MonetDB/MonetDB/issues/645/comments
0
2020-11-30T08:46:53Z
2024-06-27T11:05:24Z
https://github.com/MonetDB/MonetDB/issues/645
753,280,434
645
[ "MonetDB", "MonetDB" ]
Date: 2005-08-02 10:56:17 +0200 From: Wouter Alink &lt;<alink>&gt; To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: -- development Last updated: 2005-08-02 07:29:49 +0200 ## Comment 3611 Date: 2005-08-02 10:56:17 +0200 From: Wouter Alink &lt;<alink>&gt; When shredding a large document, the result was: mil>shred_doc("/export/scratch0/alink/xmark/xmark_550MB.burk.xml","xmark_550MB.burk"); Shredded XML doc("xmark_550MB.burk"), total time after commit=-2073.-482s After some digging, i guess that the time() function is limited to denote intervals of +-4000 seconds. ## Comment 3612 Date: 2005-08-02 10:59:42 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 by the way, could somebody tell me what the time "-2073.-482" should have been? So i don't have to re-shred the document to get the timing results? ## Comment 3613 Date: 2005-08-02 13:21:28 +0200 From: @sjoerdmullender Logged In: YES user_id=43607 (unsigned int) ((-2073*1000 - 482) * 1000) us = 2221.485s And of course the limits are more like +-2148 seconds. ## Comment 3614 Date: 2005-08-02 14:41:01 +0200 From: @drstmane Logged In: YES user_id=572415 the problem is, that MIL "usec()" returns only an int, while C "GDKusec()" provides a lng... could be fixed easily (2x "int" -> "lng" in src/modules/plain/alarm.mx); though I'd like to run (all) tests first, to see, whether anyone relies on usec() returning int ... ## Comment 3615 Date: 2005-08-02 19:29:49 +0200 From: @drstmane Logged In: YES user_id=572415 fixed in CVS: The problem is actually not "time()" (which returns milliseconds as int, and hence allows to cover 2147483 secs ~= 24 days), but "usec()", which returned microseconds as int, and hence allowed to cover only 2147 secs ~= 35 minutes. Since the underlying GDKusec provides the milliseconds as lng, there is no reason why "usec()" should not return a lng as well. ## Comment 3616 Date: 2005-12-19 13:55:53 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST Testing is fairly simple: sleep for more than 2150 seconds and then print the value of usec() in the same way as shred_doc does it, however it would take prohibitively long (2150 seconds, i.e. more than 35 minutes). ## Comment 3617 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1250067 at http://sourceforge.net/support/tracker.php?aid=1250067
MIL: time() limited to < (+-4000s)
https://api.github.com/repos/MonetDB/MonetDB/issues/644/comments
0
2020-11-30T08:46:50Z
2024-06-27T11:05:23Z
https://github.com/MonetDB/MonetDB/issues/644
753,280,393
644
[ "MonetDB", "MonetDB" ]
Date: 2005-07-29 02:50:49 +0200 From: Wouter Alink &lt;<alink>&gt; To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: -- development Last updated: 2005-08-03 08:36:52 +0200 ## Comment 3606 Date: 2005-07-29 14:50:49 +0200 From: Wouter Alink &lt;<alink>&gt; Currently the 'key(true)' command is ignored when the head-column is void. However, sometimes this results in strange behaviour (see example below). According to Peter the 'key'-settings should be saved (for the times when void->oid) MonetDB>var c:=bat(void,int).seqbase(0@0).insert(0@0,20); MonetDB>c.insert(2@0,20); MonetDB>c.key(true); MonetDB>c.insert(2@0,15); MonetDB>c.print(); ----------------- h t name oid int type ----------------- [ 0@0, 20 ] [ 2@0, 15 ] MonetDB>var c:=bat(void,int).seqbase(0@0).insert(0@0,20); MonetDB>c.key(true); MonetDB>c.insert(2@0,20); MonetDB>c.insert(2@0,15); MonetDB>c.print(); ----------------- h t name oid int type ----------------- [ 0@0, 20 ] [ 2@0, 20 ] [ 2@0, 15 ] MonetDB> ## Comment 3607 Date: 2005-08-03 08:36:52 +0200 From: @drstmane Logged In: YES user_id=572415 fixed in CVS. ## Comment 3608 Date: 2005-08-03 19:01:47 +0200 From: @drstmane Logged In: YES user_id=572415 fixed in CVS. ## Comment 3609 Date: 2005-12-19 13:48:59 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: ALREADY IN TESTWEB MonetDB/tests/BugsViaSourgeforce/Tests/ID.1247428.milS ## Comment 3610 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1247428 at http://sourceforge.net/support/tracker.php?aid=1247428
MIL: key(true) behaviour
https://api.github.com/repos/MonetDB/MonetDB/issues/643/comments
0
2020-11-30T08:46:47Z
2024-06-27T11:05:22Z
https://github.com/MonetDB/MonetDB/issues/643
753,280,356
643
[ "MonetDB", "MonetDB" ]
Date: 2005-07-28 10:55:04 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-07-29 11:34:39 +0200 ## Comment 3602 Date: 2005-07-28 10:55:04 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Test case: sql>create table test( i int default next value for non_ex_seq); 6 querytype sql>insert into test; 4 querytype [ 1 ] sql>select * from test; 3 querytype test table_name i name int type 10 length 1 tuplecount 2 id [ 0 ] sql>insert into test; Dunno if it should be allowed to create the table when non_ex_seq doesn't exist. (Similar, if it will be dropped afterwards - would require to check dependencies between database objects.) But I suppose that the insert succeeds unintentionally. (The second insert hangs.) ## Comment 3603 Date: 2005-07-29 11:34:39 +0200 From: @njnes Logged In: YES user_id=43556 'next value for' now correctly checks for the existance of the sequence number. This also makes sure no tables are created without sequence number. Unfortunatly dropping sequence number cannot check (at the moment) for existing tables using it. But the insert after the drop should correctly fail. ## Comment 3604 Date: 2005-12-19 13:43:45 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS sql/src/test/BugDay_2005-12-19_2.9.3/Tests/default_next_value_for_non_ex_seq.SF.1246631.sql ## Comment 3605 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1246631 at http://sourceforge.net/support/tracker.php?aid=1246631
SQL: ... default next value for non_ex_seq
https://api.github.com/repos/MonetDB/MonetDB/issues/642/comments
0
2020-11-30T08:46:43Z
2024-06-27T11:05:21Z
https://github.com/MonetDB/MonetDB/issues/642
753,280,323
642
[ "MonetDB", "MonetDB" ]
Date: 2005-07-25 11:15:52 +0200 From: Wouter Alink &lt;<alink>&gt; To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: -- development Last updated: 2005-07-26 07:42:33 +0200 ## Comment 3595 Date: 2005-07-25 11:15:52 +0200 From: Wouter Alink &lt;<alink>&gt; Using PF 0.8 + MDB 4.8 (latest CVS) on FC4, 64-bit When Mserver is started using: module(pathfinder); mapi_start(); pfstart(); the attached script will result in a Mserver crash (mostly after 2 times). ## Comment 3596 Date: 2005-07-25 12:28:02 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 the solution to this bug is rather simple... what i stated above was incorrect, i used the following initialization: module(pathfinder); module(mapi); mapi_start(); pfstart(); when leaving out module(mapi); the server doesn't crash anymore. ## Comment 3597 Date: 2005-07-25 12:51:52 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 just a little bit to early... after retrying it after lunch the bug reappeared even when module(mapi); was left out. putting module(pathfinder); module(mapi); mapi_start(); and pfstart(); in different order also didn't make any difference. ## Comment 3598 Date: 2005-07-25 14:24:52 +0200 From: @drstmane Logged In: YES user_id=572415 Some quick tests seem to indicate, that this is also related to Fedora Core 4's gcc 4.0.0-8 on x86_64 and possibly burg, since the segfault seem to occur in burg code ... When compiling pathfinder only (not MonetDB; though I suppose that it should also if also MonetDB is compiled with gcc32) with Fedora Core 4's gcc 3.2.3-47.fc4 ('gcc32') everything seems to work fine...!?? ## Comment 3599 Date: 2005-07-26 07:42:33 +0200 From: @drstmane Logged In: YES user_id=572415 Testing confirmed that this seems to be a compiler/64bit problem that only occurs with Fedora Core 4's gcc 4.0.0-8 on x86_64 and Intel's icc 8.1 on IA64; for Fedora Core 4, I recommend to compile pathfinder with gcc 3.2.3 (configure with "--with-gcc=gcc32") instead of with the default gcc 4.0.0. ## Comment 3600 Date: 2005-12-19 13:38:17 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: NO TEST / COMPILATION Even though there is a test script attached, I didn't add it to the test suite since the problem is purely a compilation problem and the resolution was "Invalid". ## Comment 3601 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1244326 at http://sourceforge.net/support/tracker.php?aid=1244326
XQuery: crash after repeatedly querying
https://api.github.com/repos/MonetDB/MonetDB/issues/641/comments
0
2020-11-30T08:46:41Z
2024-06-28T13:34:44Z
https://github.com/MonetDB/MonetDB/issues/641
753,280,297
641
[ "MonetDB", "MonetDB" ]
Date: 2005-07-21 11:38:00 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-08-17 10:00:10 +0200 ## Comment 3591 Date: 2005-07-21 11:38:00 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Running select * from env where 1 in ( select 1 from env ); shows MAPI = monetdb@localhost:45123 QUERY = select * from env where 1 in ( select 1 from env ); ERROR = !ERROR: interpret: no matching MIL operator to 'reverse(sht)'. !MAYBE YOU MEAN: ! reverse(BAT[any::1,any::2]) : BAT [any::2,any::1] ## Comment 3592 Date: 2005-07-21 22:05:22 +0200 From: @njnes Logged In: YES user_id=43556 Added the missing implementation ## Comment 3593 Date: 2005-12-19 13:25:25 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS sql/src/test/BugDay_2005-12-19_2.9.3/Tests/select_from_env.SF-1240701.1242164.sql ## Comment 3594 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1242164 at http://sourceforge.net/support/tracker.php?aid=1242164
SQL: ... where 1 in ( ...
https://api.github.com/repos/MonetDB/MonetDB/issues/640/comments
0
2020-11-30T08:46:37Z
2024-06-27T11:05:19Z
https://github.com/MonetDB/MonetDB/issues/640
753,280,264
640
[ "MonetDB", "MonetDB" ]
Date: 2005-07-19 10:14:47 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-11-28 11:26:48 +0100 ## Comment 3586 Date: 2005-07-19 10:14:47 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Test case: sql>select * from env where name = ( select 'prefix' from env ); 3 querytype env, env table_name name, value name varchar, varchar type 6, 17 length 32, tuplecount 14, id ... I'd expect a 'cardinality violation' if the scalar subquery returns more than one row. ## Comment 3587 Date: 2005-11-28 23:26:48 +0100 From: @njnes Logged In: YES user_id=43556 Added missing dynamic check for cardinality (ie single value (possibly one per outer reference)) Added test script src/test/BugTracker/Tests/cardinality_violation.SF-1240701.sql ## Comment 3588 Date: 2005-12-19 13:20:16 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS sql/src/test/BugDay_2005-12-19_2.9.3/Tests/select_from_env.SF-1240701.sql ## Comment 3589 Date: 2005-12-19 13:26:03 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: moved test to select_from_env.SF-1240701.1242164.sql ## Comment 3590 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1240701 at http://sourceforge.net/support/tracker.php?aid=1240701
SQL: SELECT ... where name = ( select ... )
https://api.github.com/repos/MonetDB/MonetDB/issues/639/comments
0
2020-11-30T08:46:35Z
2024-06-27T11:05:18Z
https://github.com/MonetDB/MonetDB/issues/639
753,280,232
639
[ "MonetDB", "MonetDB" ]
Date: 2005-07-18 02:22:28 +0200 From: @yzchang To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: -- development Last updated: 2005-07-26 07:43:18 +0200 ## Comment 3580 Date: 2005-07-18 14:22:28 +0200 From: @yzchang Hello, Execution of several example x-queries causes Mserver to crash (segmentation fault). I recompiled MonetDB/XQuery from scratch for my 64-bits system, but it didn't help. Regards, Jennie =================== error info given by gdb: (gdb) run Starting program: /ufs/zhang/install/monetdb/bin/Mserver [Thread debugging using libthread_db enabled] [New Thread 46912516581280 (LWP 31444)] [New Thread 1075841376 (LWP 31447)] Monet Database Server V4.9.0 Copyright (c) 1993-2005, CWI. All rights reserved. Compiled for x86_64-redhat-linux-gnu/64bit with 64bit OIDs; dynamically linked. Visit http://monetdb.cwi.nl/ for further information. MonetDB>module(pathfinder); MonetDB>pfstart(); [New Thread 1077942624 (LWP 31448)] MonetDB>[New Thread 1080043872 (LWP 31449)] [New Thread 1082145120 (LWP 31450)] MonetDB>mapi_start(); [New Thread 1084246368 (LWP 31451)] MonetDB>[New Thread 1086347616 (LWP 31452)] [New Thread 1088448864 (LWP 31453)] MonetDB> Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 1080043872 (LWP 31449)] 0x00002aaab5a03509 in reduce (p=0x2aaab5d3b630, goalnt=57) at ../../../compiler/core/fs.brg:675 675 rule = PFfs_rule (STATE_LABEL (p), goalnt); (gdb) where 0 0x00002aaab5a03509 in reduce (p=0x2aaab5d3b630, goalnt=57) at ../../../compiler/core/fs.brg:675 1 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b630, goalnt=56) at ../../../compiler/core/fs.brg:729 2 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b630, goalnt=55) at ../../../compiler/core/fs.brg:729 3 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b630, goalnt=54) at ../../../compiler/core/fs.brg:729 4 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b630, goalnt=53) at ../../../compiler/core/fs.brg:729 5 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b630, goalnt=52) at ../../../compiler/core/fs.brg:729 6 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b630, goalnt=51) at ../../../compiler/core/fs.brg:729 7 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b630, goalnt=50) at ../../../compiler/core/fs.brg:729 8 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b630, goalnt=49) at ../../../compiler/core/fs.brg:729 9 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b630, goalnt=48) at ../../../compiler/core/fs.brg:729 10 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b630, goalnt=47) at ../../../compiler/core/fs.brg:729 11 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b630, goalnt=46) at ../../../compiler/core/fs.brg:729 12 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b630, goalnt=45) at ../../../compiler/core/fs.brg:729 13 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b630, goalnt=26) at ../../../compiler/core/fs.brg:729 14 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b630, goalnt=25) at ../../../compiler/core/fs.brg:729 15 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b630, goalnt=87) at ../../../compiler/core/fs.brg:729 16 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b6f0, goalnt=87) at ../../../compiler/core/fs.brg:729 17 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=80) at ../../../compiler/core/fs.brg:729 18 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=79) at ../../../compiler/core/fs.brg:729 19 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=65) at ../../../compiler/core/fs.brg:729 20 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=63) at ../../../compiler/core/fs.brg:729 21 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=62) at ../../../compiler/core/fs.brg:729 22 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=60) at ../../../compiler/core/fs.brg:729 23 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=58) at ../../../compiler/core/fs.brg:729 24 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=57) at ../../../compiler/core/fs.brg:729 25 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=56) at ../../../compiler/core/fs.brg:729 26 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=55) at ../../../compiler/core/fs.brg:729 27 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=54) at ../../../compiler/core/fs.brg:729 28 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=53) at ../../../compiler/core/fs.brg:729 29 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=52) at ../../../compiler/core/fs.brg:729 30 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=51) at ../../../compiler/core/fs.brg:729 31 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=50) at ../../../compiler/core/fs.brg:729 32 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=49) at ../../../compiler/core/fs.brg:729 33 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=48) at ../../../compiler/core/fs.brg:729 34 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=47) at ../../../compiler/core/fs.brg:729 35 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=46) at ../../../compiler/core/fs.brg:729 36 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=45) at ../../../compiler/core/fs.brg:729 37 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=26) at ../../../compiler/core/fs.brg:729 38 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=25) at ../../../compiler/core/fs.brg:729 39 0x00002aaab5a03654 in reduce (p=0x2aaab5d3b7d0, goalnt=87) at ../../../compiler/core/fs.brg:729 40 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c170, goalnt=87) at ../../../compiler/core/fs.brg:729 41 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c1d0, goalnt=87) at ../../../compiler/core/fs.brg:729 42 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c2b0, goalnt=80) at ../../../compiler/core/fs.brg:729 43 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c2b0, goalnt=79) at ../../../compiler/core/fs.brg:729 44 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c2b0, goalnt=65) at ../../../compiler/core/fs.brg:729 45 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c2b0, goalnt=63) at ../../../compiler/core/fs.brg:729 46 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c2b0, goalnt=62) at ../../../compiler/core/fs.brg:729 47 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=61) at ../../../compiler/core/fs.brg:729 48 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=60) at ../../../compiler/core/fs.brg:729 49 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=58) at ../../../compiler/core/fs.brg:729 50 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=57) at ../../../compiler/core/fs.brg:729 51 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=56) at ../../../compiler/core/fs.brg:729 52 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=55) at ../../../compiler/core/fs.brg:729 53 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=54) at ../../../compiler/core/fs.brg:729 54 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=53) at ../../../compiler/core/fs.brg:729 55 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=52) at ../../../compiler/core/fs.brg:729 56 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=51) at ../../../compiler/core/fs.brg:729 57 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=50) at ../../../compiler/core/fs.brg:729 58 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=49) at ../../../compiler/core/fs.brg:729 59 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=48) at ../../../compiler/core/fs.brg:729 60 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=47) at ../../../compiler/core/fs.brg:729 ---Type <return> to continue, or q <return> to quit--- 61 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=46) at ../../../compiler/core/fs.brg:729 62 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=45) at ../../../compiler/core/fs.brg:729 63 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=26) at ../../../compiler/core/fs.brg:729 64 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=25) at ../../../compiler/core/fs.brg:729 65 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=24) at ../../../compiler/core/fs.brg:729 66 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=73) at ../../../compiler/core/fs.brg:729 67 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=65) at ../../../compiler/core/fs.brg:729 68 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=63) at ../../../compiler/core/fs.brg:729 69 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c6f0, goalnt=62) at ../../../compiler/core/fs.brg:729 70 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=61) at ../../../compiler/core/fs.brg:729 71 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=60) at ../../../compiler/core/fs.brg:729 72 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=58) at ../../../compiler/core/fs.brg:729 73 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=57) at ../../../compiler/core/fs.brg:729 74 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=56) at ../../../compiler/core/fs.brg:729 75 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=55) at ../../../compiler/core/fs.brg:729 76 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=54) at ../../../compiler/core/fs.brg:729 77 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=53) at ../../../compiler/core/fs.brg:729 78 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=52) at ../../../compiler/core/fs.brg:729 79 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=51) at ../../../compiler/core/fs.brg:729 80 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=50) at ../../../compiler/core/fs.brg:729 81 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=49) at ../../../compiler/core/fs.brg:729 82 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=48) at ../../../compiler/core/fs.brg:729 83 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=47) at ../../../compiler/core/fs.brg:729 84 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=46) at ../../../compiler/core/fs.brg:729 85 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=45) at ../../../compiler/core/fs.brg:729 86 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=26) at ../../../compiler/core/fs.brg:729 87 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=25) at ../../../compiler/core/fs.brg:729 88 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c630, goalnt=24) at ../../../compiler/core/fs.brg:729 89 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c7b0, goalnt=25) at ../../../compiler/core/fs.brg:729 90 0x00002aaab5a03654 in reduce (p=0x2aaab5d3c7b0, goalnt=87) at ../../../compiler/core/fs.brg:729 91 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d150, goalnt=87) at ../../../compiler/core/fs.brg:729 92 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d1b0, goalnt=87) at ../../../compiler/core/fs.brg:729 93 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=80) at ../../../compiler/core/fs.brg:729 94 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=79) at ../../../compiler/core/fs.brg:729 95 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=65) at ../../../compiler/core/fs.brg:729 96 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=63) at ../../../compiler/core/fs.brg:729 97 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=62) at ../../../compiler/core/fs.brg:729 98 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=60) at ../../../compiler/core/fs.brg:729 99 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=58) at ../../../compiler/core/fs.brg:729 100 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=57) at ../../../compiler/core/fs.brg:729 101 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=56) at ../../../compiler/core/fs.brg:729 102 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=55) at ../../../compiler/core/fs.brg:729 103 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=54) at ../../../compiler/core/fs.brg:729 104 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=53) at ../../../compiler/core/fs.brg:729 105 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=52) at ../../../compiler/core/fs.brg:729 106 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=51) at ../../../compiler/core/fs.brg:729 107 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=50) at ../../../compiler/core/fs.brg:729 108 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=49) at ../../../compiler/core/fs.brg:729 109 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=48) at ../../../compiler/core/fs.brg:729 110 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=47) at ../../../compiler/core/fs.brg:729 111 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=46) at ../../../compiler/core/fs.brg:729 112 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=45) at ../../../compiler/core/fs.brg:729 113 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=26) at ../../../compiler/core/fs.brg:729 114 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=25) at ../../../compiler/core/fs.brg:729 115 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d290, goalnt=24) at ../../../compiler/core/fs.brg:729 116 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d350, goalnt=25) at ../../../compiler/core/fs.brg:729 117 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d350, goalnt=24) at ../../../compiler/core/fs.brg:729 118 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d350, goalnt=23) at ../../../compiler/core/fs.brg:729 119 0x00002aaab5a03654 in reduce (p=0x2aaab5d3d3b0, goalnt=1) at ../../../compiler/core/fs.brg:729 120 0x00002aaab5a10126 in PFfs (r=0x2aaab5d3d3b0) at ../../../compiler/core/fs.brg:3310 121 0x00002aaab599c822 in pf_compile_MonetDB (xquery=0x1c68f10 "<example> { \n <doc>\n <test>Hello World</test>\n </doc>/test/text() \n} </example>\n \n", mode=0x1c68f08 "dm-mapi", ---Type <return> to continue, or q <return> to quit--- prologue=0x40601cb8, query=0x40601cb0, epilogue=0x40601ca8) at ../../compiler/compile.c:562 122 0x00002aaab599ad18 in CMDxquery (mode=0x1c68f08 "dm-mapi", xquery=0x1c68f10 "<example> { \n <doc>\n <test>Hello World</test>\n </doc>/test/text() \n} </example>\n \n") at ../../runtime/pathfinder.mx:797 123 0x00002aaab599b74a in CMDxquery_server (stk=3, lt=0x1bf30c8, res=0x40602180) at ../../runtime/pathfinder.mx:979 124 0x00002aaaaaabe4a0 in interpret (stk=3, lt=0x1bf30c8, res=0x40602180) at ../../../src/monet/monet_interpreter.mx:845 125 0x00002aaaaaadb3f2 in handleRequest (t=0x2aaaabbe8e30, q=0x1bf60e8, res=0x40602180) at ../../../src/monet/monet_queue.mx:533 126 0x00002aaaaaadb843 in doRequest (t=0x2aaaabbe8e30, preference=0x0) at ../../../src/monet/monet_queue.mx:559 127 0x00002aaaaab2fcc0 in monetInterpreter (status=0x403fdfa4) at ../../../src/monet/monet_process.mx:111 128 0x000000351d50697c in start_thread () from /lib64/libpthread.so.0 129 0x000000351c8c9c2e in clone () from /lib64/libc.so.6 130 0x0000000000000000 in ?? () ## Comment 3581 Date: 2005-07-21 12:44:44 +0200 From: @drstmane Logged In: YES user_id=572415 This problem seems to occur only with Fedora Core 4's gcc 4.0.0-8 on x86_64; hence, I suspect that this is a compiler problem/bug. Actually, it seemt to be worth, the less(!) optimization is usd, i.e., when configured with "--enable-debug", more tests fail than when configured with "--enable-optimize" ... !?? ## Comment 3582 Date: 2005-07-23 12:04:56 +0200 From: @drstmane Logged In: YES user_id=572415 The observed segfaults in the pathfinder compiler (mainly occuring in brg code) indeed seem to be caused by (a bug in?) Fedora Core 4's gcc 4.0.0-8 (on x86_64) --- when compiled with gcc 3.2.3 (configured with "--with-gcc=gcc32") on the same Fedora Core 4 machine, the pathfinder compiler and hence all MonetDB/XQuery tests work fine. However, similar (if not identical) segfaults seem to occur on IA-64 (Itanium) with Intel's icc 8.1-26 ... !?? ... to be continued ... ## Comment 3583 Date: 2005-07-26 07:43:18 +0200 From: @drstmane Logged In: YES user_id=572415 Testing confirmed that this seems to be a compiler/64bit problem that only occurs with Fedora Core 4's gcc 4.0.0-8 on x86_64 and Intel's icc 8.1 on IA64; for Fedora Core 4, I recommend to compile pathfinder with gcc 3.2.3 (configure with "--with-gcc=gcc32") instead of with the default gcc 4.0.0. ## Comment 3584 Date: 2005-12-19 13:15:59 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 NBugDay_2005-12-19, sjoerd: O TEST / COMPILATION ## Comment 3585 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1240211 at http://sourceforge.net/support/tracker.php?aid=1240211
Segmentation fault at Mserver by example XQeury queries
https://api.github.com/repos/MonetDB/MonetDB/issues/638/comments
0
2020-11-30T08:46:31Z
2024-06-28T13:34:43Z
https://github.com/MonetDB/MonetDB/issues/638
753,280,194
638
[ "MonetDB", "MonetDB" ]
Date: 2005-07-18 12:34:01 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-07-19 09:54:57 +0200 ## Comment 3576 Date: 2005-07-18 12:34:01 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; The following results in a server crash: sql>create table x( a int, b varchar(9)); 6 querytype sql>delete from x where 1 = 1; MAPI = monetdb@localhost:45123 ACTION= read_line QUERY = delete from x where 1 = 1; ERROR = Connection terminated ## Comment 3577 Date: 2005-07-18 15:41:03 +0200 From: @njnes Logged In: YES user_id=43556 delete now works with predicates which do not use the table. ## Comment 3578 Date: 2005-12-19 13:14:04 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS sql/src/test/BugDay_2005-12-19_2.9.3/Tests/delete_where_1_equals_1.SF-1240162.sql ## Comment 3579 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1240162 at http://sourceforge.net/support/tracker.php?aid=1240162
SQL: DELETE ... where 1 = 1
https://api.github.com/repos/MonetDB/MonetDB/issues/637/comments
0
2020-11-30T08:46:28Z
2024-06-27T11:05:16Z
https://github.com/MonetDB/MonetDB/issues/637
753,280,155
637
[ "MonetDB", "MonetDB" ]
Date: 2005-07-18 10:41:00 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-07-19 09:54:36 +0200 ## Comment 3572 Date: 2005-07-18 10:41:00 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; The following results in a server crash: sql>prepare select * from env where name in (?); MAPI = monetdb@localhost:45123 ACTION= read_line QUERY = prepare select * from env where name in (?); ERROR = Connection terminated ## Comment 3573 Date: 2005-07-18 11:13:57 +0200 From: @njnes Logged In: YES user_id=43556 moved set_type_param into check_types. This should fix the remaining prepare - type check related crashes. ## Comment 3574 Date: 2005-12-19 13:04:32 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS sql/src/tests/BugDay_2005-12-19_2.9.3/Tests/prepare-where.SF-1238867.1238959.1238965.1240124.sql ## Comment 3575 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1240124 at http://sourceforge.net/support/tracker.php?aid=1240124
SQL: PREPARE select ... where ... in (?)
https://api.github.com/repos/MonetDB/MonetDB/issues/636/comments
0
2020-11-30T08:46:25Z
2024-06-27T11:05:15Z
https://github.com/MonetDB/MonetDB/issues/636
753,280,124
636
[ "MonetDB", "MonetDB" ]
Date: 2005-07-15 04:54:21 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-07-18 11:16:34 +0200 ## Comment 3567 Date: 2005-07-15 16:54:21 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Sorry again for this esoteric test case! It results in a server crash: sql>prepare select ? from env; MAPI = monetdb@localhost:45123 ACTION= read_line QUERY = prepare select ? from env; ERROR = Connection terminated Dunno whether or not this should be allowed. ## Comment 3568 Date: 2005-07-15 17:10:54 +0200 From: @grobian Logged In: YES user_id=963970 no, it shouldn't crash no, it is not allowed, for the prepare cannot know what to expect for the ?, ie. type, size, etc. ## Comment 3569 Date: 2005-07-15 18:01:29 +0200 From: @njnes Logged In: YES user_id=43556 Result columns need to have a type, this is enforced now. ## Comment 3570 Date: 2005-12-19 13:04:30 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS sql/src/tests/BugDay_2005-12-19_2.9.3/Tests/prepare-where.SF-1238867.1238959.1238965.1240124.sql ## Comment 3571 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1238965 at http://sourceforge.net/support/tracker.php?aid=1238965
SQL: PREPARE select ? ...
https://api.github.com/repos/MonetDB/MonetDB/issues/635/comments
0
2020-11-30T08:46:22Z
2024-06-27T11:05:14Z
https://github.com/MonetDB/MonetDB/issues/635
753,280,077
635
[ "MonetDB", "MonetDB" ]
Date: 2005-07-15 04:41:26 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-07-18 11:15:20 +0200 ## Comment 3563 Date: 2005-07-15 16:41:26 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Maybe it's related to http://sourceforge.net/support/tracker.php?aid=1238867, but even more esoteric: sql>prepare select * from env where ? = ?; MAPI = monetdb@localhost:45123 QUERY = prepare select * from env where ? = ?; ERROR = !subquery result missing I guess there is an additional difficulty in type inference. ## Comment 3564 Date: 2005-07-15 18:00:36 +0200 From: @njnes Logged In: YES user_id=43556 We now enforce atleast one known type on expressions. ## Comment 3565 Date: 2005-12-19 13:04:26 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS sql/src/tests/BugDay_2005-12-19_2.9.3/Tests/prepare-where.SF-1238867.1238959.1238965.1240124.sql ## Comment 3566 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1238959 at http://sourceforge.net/support/tracker.php?aid=1238959
SQL: PREPARE ... where ? = ?
https://api.github.com/repos/MonetDB/MonetDB/issues/634/comments
0
2020-11-30T08:46:19Z
2024-06-27T11:05:13Z
https://github.com/MonetDB/MonetDB/issues/634
753,280,046
634
[ "MonetDB", "MonetDB" ]
Date: 2005-07-15 04:08:19 +0200 From: @swingbit To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: -- development Last updated: 2005-07-21 07:33:14 +0200 ## Comment 3556 Date: 2005-07-15 16:08:19 +0200 From: @swingbit casting bats makes them writable. var b:=new(oid,int).insert(0@0,10).insert(1@0, 10).access(BAT_READ); b.access().print(); [1] b.[dbl]().access().print(); [0] ## Comment 3557 Date: 2005-07-15 16:21:19 +0200 From: @swingbit Logged In: YES user_id=883451 Apparently the same happens more in general for multiplexed operations (not for all of them though). Is this a bug or a feature? ## Comment 3558 Date: 2005-07-21 12:47:52 +0200 From: @drstmane Logged In: YES user_id=572415 I'd say it's a BUG, since neither bat casts nor (general) multiplexes should return a writeable BAT (not even it the input BAT was writeable!). Do you by any chance have a more specific analysis, which casts/multiplexes do return a writeable BAT and which don't? ## Comment 3559 Date: 2005-07-21 17:57:11 +0200 From: @swingbit Logged In: YES user_id=883451 mhh... I cannot reproduce it on normal multiplexes... maybe it applies only to casts. But I confirm that it happens to all the casts. Notice that it does not happens, for istance, when casting between oid and 64bits types (for 64bits machines). ## Comment 3560 Date: 2005-07-21 19:33:14 +0200 From: @drstmane Logged In: YES user_id=572415 fixed in CVS (src/monet/monet_interpreter.mx): when interpret_batcast() returns a new BAT (instead of a view), make sure, the new BAT is marked BAT_READ; (in case of TOK_BATCAST, we do (can?)not use CATCHVAL to take care of this...) ## Comment 3561 Date: 2005-12-19 13:08:46 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS MonetDB/tests/BugDay_2005-12-19_4.9.3/Tests/bat_cast_makes_writable.SF-1238930.milS ## Comment 3562 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1238930 at http://sourceforge.net/support/tracker.php?aid=1238930
bat cast makes bats writable
https://api.github.com/repos/MonetDB/MonetDB/issues/633/comments
0
2020-11-30T08:46:16Z
2024-06-27T11:05:12Z
https://github.com/MonetDB/MonetDB/issues/633
753,280,026
633
[ "MonetDB", "MonetDB" ]
Date: 2005-07-15 02:26:42 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-07-18 11:15:01 +0200 ## Comment 3552 Date: 2005-07-15 14:26:42 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Sorry for this esoteric test case! It results in a server crash: sql>prepare select * from env where 1 = ?; MAPI = monetdb@localhost:45123 ACTION= read_line QUERY = prepare select * from env where 1 = ?; ERROR = Connection terminated ## Comment 3553 Date: 2005-07-15 15:12:03 +0200 From: @njnes Logged In: YES user_id=43556 The type checks for the simple predicate case were to late (the type_check code automatically updates the types of parameters (?)). ## Comment 3554 Date: 2005-12-19 13:04:17 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS sql/src/tests/BugDay_2005-12-19_2.9.3/Tests/prepare-where.SF-1238867.1238959.1238965.1240124.sql ## Comment 3555 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1238867 at http://sourceforge.net/support/tracker.php?aid=1238867
SQL: PREPARE ... where 1 = ?
https://api.github.com/repos/MonetDB/MonetDB/issues/632/comments
0
2020-11-30T08:46:14Z
2024-06-27T11:05:11Z
https://github.com/MonetDB/MonetDB/issues/632
753,280,007
632
[ "MonetDB", "MonetDB" ]
Date: 2005-07-14 07:43:51 +0200 From: Wouter Alink &lt;<alink>&gt; To: @njnes Version: -- development Last updated: 2005-07-21 05:03:19 +0200 ## Comment 3546 Date: 2005-07-14 19:43:51 +0200 From: Wouter Alink &lt;<alink>&gt; When shredding a document using MapiClient in a Cygwin environment (compiled with MinGW): MapiClient -lmil -q -s "shred_doc (\"c:\\\\cygwin\\\\home\\\\Wouter\\\\test\\\\temp.xml\",\"tem p.xml\");" MapiClient shreds the document successfully, BUT, the file (/home/Wouter/test/temp.xml) will be kept 'locked' by Mserver. I found this out, because I wanted to delete the file, but I first had to kill Mserver, before the lock on the file was released. (i'm using MonetDB 4.9.0 (latest CVS) + Pathfinder 0.9.0 (latest CVS)) ## Comment 3547 Date: 2005-07-15 14:13:55 +0200 From: @njnes Logged In: YES user_id=43556 added code to cleanup the xml2 parser context (I hope this will close the file). Could you test this? ## Comment 3548 Date: 2005-07-21 12:50:46 +0200 From: @drstmane Logged In: YES user_id=572415 Wouter, do Niels' changes solve the problem? ## Comment 3549 Date: 2005-07-21 17:03:19 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 bug is fixed, thanx! ## Comment 3550 Date: 2005-12-19 12:52:44 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 BugDay_2005-12-19, sjoerd: TEST ADDED / SUCCESS pathfinder/tests/BugDay_2005-12-19_0.9.3/Tests/file_locked_after_shredding.SF-1238352.* ## Comment 3551 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1238352 at http://sourceforge.net/support/tracker.php?aid=1238352
XQuery: shredder keeps lock on files
https://api.github.com/repos/MonetDB/MonetDB/issues/631/comments
0
2020-11-30T08:46:10Z
2024-06-27T11:05:10Z
https://github.com/MonetDB/MonetDB/issues/631
753,279,964
631
[ "MonetDB", "MonetDB" ]
Date: 2005-07-14 04:13:16 +0200 From: Wouter Alink &lt;<alink>&gt; To: @sjoerdmullender Version: -- development Last updated: 2005-07-14 05:45:07 +0200 ## Comment 3542 Date: 2005-07-14 16:13:16 +0200 From: Wouter Alink &lt;<alink>&gt; Excuse me for being new to compiling in a cygwin environment, but I'm not able to compile Pathfinder (from CVS) using the HowToStart-MinGW (see error message below, which occurs during 'make'). I tried to find what i did wrong, and came across: "The values for the --with-mel and --with-mx options should be the full (Cygwin) path to the respective binaries, which should be compiled for Cygwin (i.e. not using the MinGW tools)." And maybe this is the reason, and if so, how do I compile Mx and mel seperately? ------------------- ... ... ... Making all in runtime make[2]: Entering directory `/home/Wouter/src/pathfinder-0.8/MinGW/runtime' /home/Wouter/MinGW/bin/Mx.exe -x m ../../runtime/pf_support.mx ../../runtime/pf_support.mx: .\pf_support.m - not modified /home/Wouter/MinGW/bin/mel.exe -I../../runtime - I/home/Wouter/MinGW/include/MonetDB - I/home/Wouter/MinGW/include/MonetDB /common - I/home/Wouter/MinGW/include/MonetDB/gdk - I/home/Wouter/MinGW/include/MonetDB/monet - I/home/Wouter/MinGW/include /MonetDB/plain - I/home/Wouter/MinGW/include/MonetDB/contrib - I/home/Wouter/MinGW/include/MonetDB/mapi - I/home/Wouter/Min GW/include/MonetDB/C - I/home/Wouter/MinGW/include/libxml2 - I../compiler/include -I../../runtime/../compiler/include -pr oto pf_support.m > pf_support.proto.h C:\cygwin\home\Wouter\MinGW\bin\mel.exe: "../../runtim e/pf_support.mx", line 29, Module not found at "9" of monettime. C:\cygwin\home\Wouter\MinGW\bin\mel.exe: "../../runtim e/pf_support.mx", line 29, syntax error at "1" of ;. Found 2 syntax errors. make[2]: *** [pf_support.proto.h] Fout 1 make[2]: *** Bestand `pf_support.proto.h' wordt verwijderd make[2]: Leaving directory `/home/Wouter/src/pathfinder-0.8/MinGW/runtime' make[1]: *** [all-recursive] Fout 1 make[1]: Leaving directory `/home/Wouter/src/pathfinder-0.8/MinGW' make: *** [all] Fout 2 ## Comment 3543 Date: 2005-07-14 17:45:06 +0200 From: @sjoerdmullender Logged In: YES user_id=43607 Where is the bug? If you don't follow the instructions that are given, the error is not with the instructions. As to your question, you either: - compile MonetDB using Cygwin (in addition to compiling it using MinGW); - check out buildtools from MonetDB SourceForge and follow the instructions. ## Comment 3544 Date: 2005-12-19 17:10:10 +0100 From: @yzchang Logged In: YES user_id=341633 BugDay_2005-12-19, Jennie: NO TEST / COMPILATION ## Comment 3545 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1238228 at http://sourceforge.net/support/tracker.php?aid=1238228
compiling PF using MinGW
https://api.github.com/repos/MonetDB/MonetDB/issues/630/comments
0
2020-11-30T08:46:08Z
2024-06-28T13:34:43Z
https://github.com/MonetDB/MonetDB/issues/630
753,279,934
630
[ "MonetDB", "MonetDB" ]
Date: 2005-07-13 01:22:50 +0200 From: Wouter Alink &lt;<alink>&gt; To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: -- development Last updated: 2005-07-14 03:55:23 +0200 ## Comment 3537 Date: 2005-07-13 13:22:50 +0200 From: Wouter Alink &lt;<alink>&gt; When issuing the following query: doc("video.xml")//*/self::scene in a document which looks like: <video> <scene id="1"/> <scene id="2"/> <scene id="3"/> </video> the result is: WARNING: BBPdecref: range error 450 ------------- NOTE: i still use a version (0.8) of pathfinder from a few weeks ago, so if it is already fixed, please ignore this bugreport ## Comment 3538 Date: 2005-07-13 13:53:57 +0200 From: @drstmane Logged In: YES user_id=572415 [Sorry, couldn't resist...] Cetero Censeo ...: You can easily have this monitored automatically every night with the latest version by adding the respective test to the CVS repository ... then, you'll also know, when which bugs are fixed, and hence, when you might want to upgrade your version ... ;-)) (ps: "you" does not only refert to you, Wouter, but to everyone who reports bugs ... ;-)) ## Comment 3539 Date: 2005-07-14 15:55:23 +0200 From: @drstmane Logged In: YES user_id=572415 Fixed in CVS (XQuery_0-8 release branch): There is no reason for mvaljoin to require that the BATs in ws_item have dense heads. Thanks to Wouter for submitting the bug report and providing the test. Thanks to JanR for pointing me to the origin of this bug. ## Comment 3540 Date: 2005-12-19 18:53:46 +0100 From: @yzchang Logged In: YES user_id=341633 BugDay_2005-12-19, Jennie: TEST ADDED / SUCCESS src/pathfinder/tests/BugDay_2005-12-19_0.9.3/Tests/self_axis_not_supported.SF-1237444.milM ## Comment 3541 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1237444 at http://sourceforge.net/support/tracker.php?aid=1237444
XQuery: self-axis not supported?
https://api.github.com/repos/MonetDB/MonetDB/issues/629/comments
0
2020-11-30T08:46:05Z
2024-06-27T11:05:08Z
https://github.com/MonetDB/MonetDB/issues/629
753,279,902
629
[ "MonetDB", "MonetDB" ]
Date: 2005-07-12 03:42:55 +0200 From: thijs &lt;<thijswes>&gt; To: @njnes Version: -- development Last updated: 2007-12-21 04:20:03 +0100 ## Comment 3532 Date: 2005-07-12 15:42:55 +0200 From: thijs &lt;<thijswes>&gt; The PF documentation, ie. HowToStart-PF, states that the document cache keeps track of the "last-modified" time. This is supposed to be checked on each request. However, when a file is modified, pathfinder does not update it's cache. Suppose we have a document test.xml that looks like this: <?xml version="1.0"?> <A/> Then echo 'doc("file://test.xml")'|MapiClient -lx produces <A/> (and so it should). But if we then modify test.xml to read: <?xml version="1.0"?> <B/> and run the same command again echo 'doc("file://test.xml")'|MapiClient -lx the output is still: <A/> ## Comment 3533 Date: 2005-12-19 18:03:31 +0100 From: @yzchang Logged In: YES user_id=341633 BugDay_2005-12-19, Jennie: RELEASE Don't know how to write test script, sorry! ## Comment 3534 Date: 2006-12-20 04:20:04 +0100 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; ---- Original comment by: sf-robot@users.sourceforge.net ---- Logged In: YES user_id=1312539 Originator: NO This Tracker item was closed automatically by the system. It was previously set to a Pending status, and the original submitter did not respond within 365 days (the time period specified by the administrator of this Tracker). ## Comment 3535 Date: 2007-12-21 04:20:03 +0100 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; ---- Original comment by: sf-robot@users.sourceforge.net ---- Logged In: YES user_id=1312539 Originator: NO This Tracker item was closed automatically by the system. It was previously set to a Pending status, and the original submitter did not respond within 365 days (the time period specified by the administrator of this Tracker). ## Comment 3536 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1236711 at http://sourceforge.net/support/tracker.php?aid=1236711
\"last-modified\" time ignored in document cache
https://api.github.com/repos/MonetDB/MonetDB/issues/628/comments
0
2020-11-30T08:46:02Z
2024-06-27T11:05:07Z
https://github.com/MonetDB/MonetDB/issues/628
753,279,865
628
[ "MonetDB", "MonetDB" ]
Date: 2005-07-07 04:46:33 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-07-18 11:14:26 +0200 ## Comment 3526 Date: 2005-07-07 16:46:33 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; PREPARE doesn't work for queries with like predicates: sql>prepare select * from env where name like ?; MAPI = monetdb@localhost:45123 QUERY = prepare select * from env where name like ?; ERROR = !parse error in: "prepare select * from env where name like ?" sql>prepare select * from env where name like 'm%'; 10 querytype prepare, prepare, prepare table_name type, digits, scale name varchar, int, int type 0, 0, 0 length 0 tuplecount 0 id sql>execute 0 ('m%'); 3 querytype env, env table_name name, value name varchar, varchar type 14, 55 length 11, tuplecount 5, id [ "monet_pid", "456" ] ... sql>execute 0 (); MAPI = monetdb@localhost:45123 QUERY = execute 0 (); ERROR = !wrong number of arguments for prepared statement ## Comment 3527 Date: 2005-07-15 10:06:18 +0200 From: @njnes Logged In: YES user_id=43556 support for ? in like statements is now added. ## Comment 3528 Date: 2005-07-15 14:20:27 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 I see Resolution is 'Fixed'. Please don't forget the second test case: prepare select * from env where name like 'm%'; ## Comment 3529 Date: 2005-07-15 15:18:01 +0200 From: @njnes Logged In: YES user_id=43556 The wrong number of arguments bug is related to the 'm%' being treated as an literal which is put in an argument list. We do this to improve query reuse (query caching). Queries where only the literals differ are handled by the same mil proc. But for prepared statement we shouldn't do this literal replacement. This is now fixed in the parser. ## Comment 3530 Date: 2005-12-19 15:10:36 +0100 From: @yzchang Logged In: YES user_id=341633 BugDay_2005-12-19, Jennie: TEST ADDED / SUCCESS src/sql/src/test/BugDay_2005-12-19_2.9.3/Tests/prepare_doesnot_like_LIKE.SF-1234205.sql Note: the test script is added to show "prepare select * from env where name like ?" works (correctly). The ID of the prepared statement is hard coded in the test script, so the test can fail if the prepared statement gets another ID. A test script in application level should be added to dynamically retrieve the ID of the prepared statement and call 'execute' with the right ID. But this is out of the scoope of the BugDay. ## Comment 3531 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1234205 at http://sourceforge.net/support/tracker.php?aid=1234205
SQL: PREPARE doesn't like LIKE
https://api.github.com/repos/MonetDB/MonetDB/issues/627/comments
0
2020-11-30T08:45:58Z
2024-06-27T11:05:06Z
https://github.com/MonetDB/MonetDB/issues/627
753,279,835
627
[ "MonetDB", "MonetDB" ]
Date: 2005-07-04 03:22:03 +0200 From: rimluga &lt;<rimluga>&gt; To: @sjoerdmullender Version: -- development Last updated: 2005-07-04 04:39:12 +0200 ## Comment 3521 Date: 2005-07-04 15:22:03 +0200 From: rimluga &lt;<rimluga>&gt; Hi, am sorry to report this request for information here, I tried to email you, and I didn't succeed. I would like to know if Monet allows automatic key generation, Best, ## Comment 3522 Date: 2005-07-04 16:39:12 +0200 From: @sjoerdmullender Logged In: YES user_id=43607 MonetDB does not allow automatic key generation. I don't know how you tried to email us. You should be able to reach us through any of the following addresses: monet@cwi.nl monetdb-users@lists.sourceforge.net (MonetDB users list) monetdb-developers@lists.sourceforge.net (MonetDB developers) ## Comment 3523 Date: 2005-12-19 13:18:58 +0100 From: @yzchang Logged In: YES user_id=341633 BugDay_2005-12-19, Jennie: NO TEST / DOCUMENTATION ## Comment 3524 Date: 2005-12-19 13:23:30 +0100 From: @grobian Logged In: YES user_id=963970 FYI: The CVS version of MonetDB/SQL supports sequences, such that auto_increment, serial and identities can be used. ## Comment 3525 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1232208 at http://sourceforge.net/support/tracker.php?aid=1232208
key autogeneration
https://api.github.com/repos/MonetDB/MonetDB/issues/626/comments
0
2020-11-30T08:45:56Z
2024-06-28T13:34:42Z
https://github.com/MonetDB/MonetDB/issues/626
753,279,806
626
[ "MonetDB", "MonetDB" ]
Date: 2005-06-30 02:17:37 +0200 From: @swingbit To: GDK devs &lt;<bugs-common>&gt; Version: -- development Last updated: 2005-07-15 11:21:59 +0200 ## Comment 3518 Date: 2005-06-30 14:17:37 +0200 From: @swingbit bat casting from lng to oid fails with assertion on a 32 bits machine: var a:=new(void,lng).seqbase(0@0); a.insert(nil,0LL).insert(nil,1LL).insert(nil,2LL); var b:=[oid](a); Mserver: /net/db2/export/data1/roberto/MonetDB/src/MonetDB/src/monet/monet_interpreter.mx:1493: interpret_batcast: Assertion `b->dims.tailtype != 6 && tpe != 6' failed. Aborted Notice that the following does NOT fail. The only difference is that the bat 'a' is not sorted on its tail: var a:=new(void,lng).seqbase(0@0); a.insert(nil,0LL).insert(nil,2LL).insert(nil,1LL); var b:=[oid](a); ## Comment 3519 Date: 2005-12-19 14:34:59 +0100 From: @yzchang Logged In: YES user_id=341633 BugDay_2005-12-19, Jennie: TEST ADDED / SUCCESS src/MonetDB/tests/BugDay_2005-12-19_4.9.3/Tests/assertion_on_bat_cast.SF-1230320.milS ## Comment 3520 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1230320 at http://sourceforge.net/support/tracker.php?aid=1230320 The original assignee of this bug does not have an account here. Reassigning to the default assignee for the component, bugs-common@monetdb.org. Previous assignee was nobody@users.sourceforge.net.
Assertion on bat cast
https://api.github.com/repos/MonetDB/MonetDB/issues/625/comments
0
2020-11-30T08:45:53Z
2024-06-27T11:05:04Z
https://github.com/MonetDB/MonetDB/issues/625
753,279,769
625
[ "MonetDB", "MonetDB" ]
Date: 2005-06-29 03:07:29 +0200 From: @swingbit To: @sjoerdmullender Version: -- development Last updated: 2005-07-04 05:02:30 +0200 ## Comment 3514 Date: 2005-06-29 15:07:29 +0200 From: @swingbit .COMMAND count ( BAT[void,void] b, lng cnt ) : lng = CMDsetcount_bat; "Returns the current size (in number of elements) of a BAT." I guess this should be: "Sets the size (in number of elements) of a BAT." Also, this text -is- in bat.mx: The sole reason not to move it to the bat.mx module is that overall the kernel does not support bat[void,void] ## Comment 3515 Date: 2005-07-04 17:02:30 +0200 From: @sjoerdmullender Logged In: YES user_id=43607 Fixed. ## Comment 3516 Date: 2005-12-19 12:43:01 +0100 From: @yzchang Logged In: YES user_id=341633 BugDay_2005-12-19, Jennie: NO TEST / DOCUMENTATION ## Comment 3517 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1229685 at http://sourceforge.net/support/tracker.php?aid=1229685
Too quick copy and paste in bat.mx
https://api.github.com/repos/MonetDB/MonetDB/issues/624/comments
0
2020-11-30T08:45:50Z
2024-06-27T11:05:03Z
https://github.com/MonetDB/MonetDB/issues/624
753,279,743
624
[ "MonetDB", "MonetDB" ]
Date: 2005-06-29 10:46:03 +0200 From: @swingbit To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: -- development Last updated: 2005-06-29 01:50:33 +0200 ## Comment 3510 Date: 2005-06-29 10:46:03 +0200 From: @swingbit In the MIL documentation: http://monetdb.cwi.nl/TechDocs/FrontEnds/mil/index.html section "2.1.5 Unique OIDs", the following link is dead: http://monetdb.cwi.nl/TechDocs/Core/Modules/builtin/index.htmlmod_2_2_0 as the 'builtin' module has been replaced by 'kernel', I suppose. Also, the 'kernel' module documentation on the website gives the following: [an error occurred while processing this directive] ## Comment 3511 Date: 2005-06-29 13:50:33 +0200 From: @drstmane Logged In: YES user_id=572415 Fixed monetweb: (quite some time ago) module "kernel" had been renamed to "builtin"... ## Comment 3512 Date: 2005-12-19 12:37:11 +0100 From: @yzchang Logged In: YES user_id=341633 BugDay_2005-12-19, Jennnie : NO TEST / DOCUMENTATION ## Comment 3513 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1229552 at http://sourceforge.net/support/tracker.php?aid=1229552
dead link in MIL docs
https://api.github.com/repos/MonetDB/MonetDB/issues/623/comments
0
2020-11-30T08:45:47Z
2024-06-27T11:05:02Z
https://github.com/MonetDB/MonetDB/issues/623
753,279,720
623
[ "MonetDB", "MonetDB" ]
Date: 2005-06-28 09:03:51 +0200 From: @grobian To: @sjoerdmullender Version: -- development Last updated: 2005-07-24 06:25:40 +0200 ## Comment 3500 Date: 2005-06-28 21:03:51 +0200 From: @grobian When compiling monetdb with gcc 4.0 and readline 5.0 on OSX Tiger, the compilation dies: cc1: warnings being treated as errors ../../../../../src/mapi/clients/C/Mapi.mx: In function 'read_line': ../../../../../src/mapi/clients/C/Mapi.mx:2683: warning: format '%d' expects type 'int', but argument 2 has type 'ssize_t' ../../../../../src/mapi/clients/C/Mapi.mx: In function 'mapi_execute_internal': ../../../../../src/mapi/clients/C/Mapi.mx:3125: warning: format '%u' expects type 'unsigned int', but argument 2 has type 'size_t' ../../../../../src/mapi/clients/C/Mapi.mx: In function 'mapi_query_part': ../../../../../src/mapi/clients/C/Mapi.mx:3296: warning: format '%u' expects type 'unsigned int', but argument 2 has type 'size_t' make[7]: *** [libMapi_la-Mapi.lo] Error 1 make[6]: *** [all] Error 2 make[5]: *** [all-recursive] Error 1 make[4]: *** [all-recursive] Error 1 make[3]: *** [all] Error 2 make[2]: *** [all-recursive] Error 1 make[1]: *** [all-recursive] Error 1 make: *** [all] Error 2 [aphrodite:current/MonetDB/build] % The build is using 32 bits ## Comment 3501 Date: 2005-07-15 10:40:03 +0200 From: @njnes Logged In: YES user_id=43556 is this bug still open? Could you try again as the mapi code is changed since the bug was reported, ie line numbers aren't valid anymore. ## Comment 3502 Date: 2005-07-23 10:31:46 +0200 From: @drstmane Logged In: YES user_id=572415 Just for info: compilation works fine with gcc 4.0.0 and readline 5.0 on Fedora Core 4. ## Comment 3503 Date: 2005-07-23 20:37:07 +0200 From: @grobian Logged In: YES user_id=963970 Still the same errors: cc1: warnings being treated as errors ../../../../../src/mapi/clients/C/Mapi.mx: In function 'read_line': ../../../../../src/mapi/clients/C/Mapi.mx:2683: warning: format '%d' expects type 'int', but argument 2 has type 'ssize_t' ../../../../../src/mapi/clients/C/Mapi.mx: In function 'mapi_execute_internal': ../../../../../src/mapi/clients/C/Mapi.mx:3125: warning: format '%u' expects type 'unsigned int', but argument 2 has type 'size_t' ../../../../../src/mapi/clients/C/Mapi.mx: In function 'mapi_query_part': ../../../../../src/mapi/clients/C/Mapi.mx:3296: warning: format '%u' expects type 'unsigned int', but argument 2 has type 'size_t' make[7]: *** [libMapi_la-Mapi.lo] Error 1 See attached path for a fix for this issue. "%zu" and "%zd" should be used on Darwin according to the printf(3) man-page. With the attached patch readline works as expected ## Comment 3504 Date: 2005-07-24 01:17:35 +0200 From: @drstmane Logged In: YES user_id=572415 Your patch indicates, that your using revision 1.160 (Jun 5 2005) of src/mapi/clients/C/Mapi.mx; however, the problematic definitions of SSZFMT have been removed from src/mapi/clients/C/Mapi.mx as of revision 1.156.2.5 (Jun 7 2005, MonetDB_4-8 release branch) respectively revision 1.161 (Jun 12 2005, development trunk, aka. CVS HEAD). Hence, please update your source code, and try again. The up-to-date code contains a fixed definition of SSZFMT (now centralized in src/common/monet_utils.mx) that is compliant with your patch and should work fine in your case... ## Comment 3505 Date: 2005-07-24 10:16:50 +0200 From: @grobian Logged In: YES user_id=963970 thanks CVS for ignoring my -A and -C The latest news is to get over 10000 erors on gdk_scanselect, but maybe cvs has not updated the sources there as well... to be continued. ## Comment 3506 Date: 2005-07-24 13:28:11 +0200 From: @drstmane Logged In: YES user_id=572415 In case youre not sure, whether your current working directory has the latest source even after a 'cvs up -dPA', you might want to consider getting a fresh check-out ('cvs -d... co -P MonetDB') in a new location, and compare both tree, e.g., via 'diff -r [-xCVS]' ... Further, unless you already did this, you might want to redo all compilation from scratch, i.e., cd <sourcedir> ./de-bootstrap cvs up -dPA ./bootstrap rm -rf <builddir> mkdir <builddir> cd <builddir> <sourcedir>/configure ... make ... ## Comment 3507 Date: 2005-07-24 14:57:47 +0200 From: @grobian Logged In: YES user_id=963970 I've done this 'clean' compile and let it simmer when I was bicycling. No errors whatsoever found, so this bug report can be closed as invalid. Sorry for the hassle! ## Comment 3508 Date: 2005-12-19 11:04:06 +0100 From: @yzchang Logged In: YES user_id=341633 BugDay_2005-12-19, Jennie: NO TEST / COMPILATION ## Comment 3509 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1229245 at http://sourceforge.net/support/tracker.php?aid=1229245
Compilation error Darwin readline5 gcc4
https://api.github.com/repos/MonetDB/MonetDB/issues/622/comments
0
2020-11-30T08:45:44Z
2024-06-27T11:05:01Z
https://github.com/MonetDB/MonetDB/issues/622
753,279,694
622
[ "MonetDB", "MonetDB" ]
Date: 2005-06-28 04:02:58 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: Steffen Goeldner &lt;<sgoeldner>&gt; Version: -- development Last updated: 2007-12-21 04:20:04 +0100 ## Comment 3486 Date: 2005-06-28 16:02:58 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Running use DBI(); $dsn = 'dbi:ODBC:driver=MonetDB ODBC Driver'; $dbh = DBI->connect( $dsn ) or die $DBI::errstr; $tbl = 'perl_dbd_test'; $dbh->do("drop table $tbl"); $dbh->do("create table $tbl( A int, B varchar(50), C varchar(300), D date )"); $sth = $dbh->prepare("insert into $tbl (A, B, C, D) VALUES (?, ?, ?, ?)"); $sth->bind_param( 1, 1 ); $sth->bind_param( 2,'foo'); $sth->bind_param( 3,'me' x 120 ); $sth->bind_param( 4,'1998-05-13'); $sth->execute; shows DBD::ODBC::st execute failed: [MonetDB][ODBC Driver 1.0]!unexpected end of input (SQL-42000) ## Comment 3487 Date: 2005-06-28 21:05:02 +0200 From: @grobian Logged In: YES user_id=963970 does the server survive this? ## Comment 3488 Date: 2005-06-29 08:40:58 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 Yes, it survives. Is it possible to trace the input it receives? ## Comment 3489 Date: 2005-06-29 09:21:45 +0200 From: @grobian Logged In: YES user_id=963970 if you don't have strace available for your platform, I think tools like ethereal are your only solution if you don't want to hack sources. ## Comment 3490 Date: 2005-06-30 16:22:33 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 I added traces on the client-side (src/mapi/clients/C/Mapi.c): @@ -1800,7 +1800,9 @@ p = q + 1; q = strchr(p, PLACEHOLDER); } + printf("***: =%d %d ... %s\n", lim, strlen(hdl->query), hdl->query + k - 10); checkSpace(strlen(p)); + printf("***: =%d %d ... %s\n", lim, strlen(hdl->query), hdl->query + k - 10); strcpy(hdl->query + k, p); if (hdl->mid->trace) printf("param_store: result=%s\n", hdl->query); The output is: ***: =314 315 ... 998-05-13' ***: =572 314 ... 998-05-13 I.e., lim < strlen(hdl->query), which means a buffer overrun. Trying to make room for \' and \0 seems to work: @@ -1779,13 +1779,13 @@ buf[0] = *(char *) src; buf[1] = 0; val = mapi_quote(buf, 1); - checkSpace(strlen(val)); + checkSpace(strlen(val)+3); sprintf(hdl->query + k, "'%s'", val); free(val); break; case MAPI_VARCHAR: val = mapi_quote((char *) src, hdl- >params[i].sizeptr ? *hdl->params[i].sizeptr : -1); - checkSpace(strlen(val)); + checkSpace(strlen(val)+3); sprintf(hdl->query + k, "'%s'", val); free(val); break; @@ -1800,7 +1800,7 @@ p = q + 1; q = strchr(p, PLACEHOLDER); } - checkSpace(strlen(p)); + checkSpace(strlen(p)+1); strcpy(hdl->query + k, p); if (hdl->mid->trace) printf("param_store: result=%s\n", hdl->query); O.k.? ## Comment 3491 Date: 2005-06-30 17:22:41 +0200 From: @sjoerdmullender Logged In: YES user_id=43607 Looks good. Can you check this in in the stable branch? ## Comment 3492 Date: 2005-07-01 14:36:48 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 O.k., fixed: <http://cvs.sourceforge.net/viewcvs.py/monetdb/MonetDB/src/ mapi/clients/C/Mapi.mx?r1=1.156.2.7&r2=1.156.2.8> ## Comment 3493 Date: 2005-12-19 16:51:15 +0100 From: @yzchang Logged In: YES user_id=341633 BugDag_2005-12-19, Jennie: REASSIGN to Steffen or Sjoerd. Don't know how to make test script for Perl, sorry! ## Comment 3494 Date: 2005-12-19 18:12:22 +0100 From: @drstmane Logged In: YES user_id=572415 see also https://sourceforge.net/tracker/index.php?func=detail&aid=1225455&group_id=56967&atid=482468 ## Comment 3495 Date: 2006-01-15 15:14:21 +0100 From: @grobian Logged In: YES user_id=963970 is this bug pending because there is no test available? ## Comment 3496 Date: 2006-01-15 15:39:08 +0100 From: @drstmane Logged In: YES user_id=572415 right, all "pending" bugs still don't have a test script, although one could create one --- usualy requires "more effort" and/or new functionallity in Mtest.py and/or detailed checks whether to add a test script at all... ## Comment 3497 Date: 2006-12-20 04:20:04 +0100 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; ---- Original comment by: sf-robot@users.sourceforge.net ---- Logged In: YES user_id=1312539 Originator: NO This Tracker item was closed automatically by the system. It was previously set to a Pending status, and the original submitter did not respond within 365 days (the time period specified by the administrator of this Tracker). ## Comment 3498 Date: 2007-12-21 04:20:04 +0100 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; ---- Original comment by: sf-robot@users.sourceforge.net ---- Logged In: YES user_id=1312539 Originator: NO This Tracker item was closed automatically by the system. It was previously set to a Pending status, and the original submitter did not respond within 365 days (the time period specified by the administrator of this Tracker). ## Comment 3499 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1229018 at http://sourceforge.net/support/tracker.php?aid=1229018
!unexpected end of input
https://api.github.com/repos/MonetDB/MonetDB/issues/621/comments
0
2020-11-30T08:45:41Z
2024-06-27T11:05:00Z
https://github.com/MonetDB/MonetDB/issues/621
753,279,652
621
[ "MonetDB", "MonetDB" ]
Date: 2005-06-25 10:13:01 +0200 From: @grobian To: @sjoerdmullender Version: -- development Last updated: 2005-06-25 10:49:28 +0200 ## Comment 3480 Date: 2005-06-25 22:13:01 +0200 From: @grobian On behalf of: I don"t know whether MonetDB.conf in CWI\MonetDB\etc is used, but it contains the following lines: Where MonetDB is installed prefix=C:\Documents and Settings\sjoerd\My Documents\src\MonetDB\NT exec_prefix=C:\Documents and Settings\sjoerd\My Documents\src\MonetDB\NT Roger ## Comment 3481 Date: 2005-06-25 22:49:28 +0200 From: @sjoerdmullender Logged In: YES user_id=43607 This is known and worked around by using BAT scripts that override the prefix variable from the MonetDB.conf file. The BAT scripts are installed in the top-most folder in the Program Files folder whereas the EXE files which they call are "hidden" in a subfolder. The reason for doing it this way is that it is too much effort (because of the way MonetDB is built) to do it differently. Note that the location of the MonetDB.conf file is built-in and wrong, so this too is overridden by the BAT scripts. ## Comment 3482 Date: 2005-06-25 22:55:40 +0200 From: @grobian Logged In: YES user_id=963970 isn't it better to leave it out in the distribution? It looks quite confusing. ## Comment 3483 Date: 2005-06-25 23:01:46 +0200 From: @sjoerdmullender Logged In: YES user_id=43607 Perhaps, but then submit it as an RFE. The way it is now is not a bug. ## Comment 3484 Date: 2005-12-19 11:50:22 +0100 From: @grobian Logged In: YES user_id=963970 BugDay_2005-12-19, meltdown: NO TEST / CONFIGURATION ## Comment 3485 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1227493 at http://sourceforge.net/support/tracker.php?aid=1227493
Windows: config file contains developer's paths
https://api.github.com/repos/MonetDB/MonetDB/issues/620/comments
0
2020-11-30T08:45:38Z
2024-06-28T13:35:23Z
https://github.com/MonetDB/MonetDB/issues/620
753,279,620
620
[ "MonetDB", "MonetDB" ]
Date: 2005-06-25 08:29:07 +0200 From: @MarcinZukowski To: @mlkersten Version: -- development Last updated: 2005-06-25 08:39:41 +0200 ## Comment 3474 Date: 2005-06-25 20:29:07 +0200 From: @MarcinZukowski Try this: var int := 1 ; printf("int=%d\n",int); it prints: int=5 I think the type names should be reserved keywords. ## Comment 3475 Date: 2005-06-25 20:38:52 +0200 From: @grobian Logged In: YES user_id=963970 This is already 'fixed' in the 'next release'. Monet Database Server V4.99.19 Copyright (c) 1993-2005, CWI. All rights reserved. Compiled for i686-redhat-linux-gnu/32bit; dynamically linked. config:/ufs/fabian/scratch/monet5/program/etc/MonetDB5.conf dbfarm:/ufs/fabian/scratch/monet5/program/var/MonetDB5/dbfarm dbname:demo Visit http://monetdb.cwi.nl/ for further information. >int := 1; >io.printf("int=%d\n", int); int=1 > ## Comment 3476 Date: 2005-06-25 20:39:40 +0200 From: @mlkersten Logged In: YES user_id=490798 This a remnant of a design decision taken in MIL and is used to get the underlying type index. It won;t be resolved before MonetDB version 5, where a clear distinction is made between variable identifiers and type identifiers. ## Comment 3477 Date: 2005-06-25 23:14:50 +0200 From: @mlkersten Logged In: YES user_id=490798 This a remnant of a design decision taken in MIL and is used to get the underlying type index. It won;t be resolved before MonetDB version 5, where a clear distinction is made between variable identifiers and type identifiers. ## Comment 3478 Date: 2005-12-19 11:48:58 +0100 From: @grobian Logged In: YES user_id=963970 BugDay_2005-12-19, meltdown: TEST ADDED / WONT FIX MonetDB/tests/BugDay_2005-12-19_4.9.3/Tests/type_names_are_not_keywords.SF-1227464.* Note, this test fails, hence should probably not be enabled ## Comment 3479 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1227464 at http://sourceforge.net/support/tracker.php?aid=1227464
type names are not keywords
https://api.github.com/repos/MonetDB/MonetDB/issues/619/comments
0
2020-11-30T08:45:35Z
2024-06-28T13:35:22Z
https://github.com/MonetDB/MonetDB/issues/619
753,279,580
619
[ "MonetDB", "MonetDB" ]
Date: 2005-06-22 01:42:26 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @sjoerdmullender Version: -- development Last updated: 2006-06-06 08:51:31 +0200 ## Comment 3456 Date: 2005-06-22 13:42:26 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Running use DBI(); my $dsn = 'dbi:ODBC:driver=MonetDB ODBC Driver'; my $dbh = DBI->connect( $dsn ) or die $DBI::errstr; $dbh->{RaiseError} = 1; $dbh->{PrintError} = 0; my $sql = 'select * from env where name = ?'; my $sth = $dbh->prepare( $sql ); $sth->bind_param( 1, undef ); NULL $sth->execute; shows DBD::ODBC::st execute failed: [MonetDB][ODBC Driver 1.0] !cannot have parameters (?) in normal queries, use PREPARE; (SQL-42000) Looks like Mapi is unable to bind a NULL; mapi_param_store(MapiHdl hdl) ... if (hdl->params[i].inparam == 0) { hdl->query[k++] = PLACEHOLDER; ... ## Comment 3457 Date: 2005-06-23 13:59:31 +0200 From: @grobian Logged In: YES user_id=963970 ok. I suppose you somehow add type information to the NULL value you send out. If you do so, however, you get a crash anyway: exec 1(cast(null as varchar(1024))); MonetDB>Segmentation fault work in progress ;) ## Comment 3458 Date: 2005-06-23 15:42:54 +0200 From: @grobian Logged In: YES user_id=963970 I think this is now fixed in CVS current: % MapiClient -ls sql>prepare select * from env where name = ?; 10 querytype prepare, prepare, prepare table_name type, digits, scale name varchar, int, int type 0, 0, 0 length 1 tuplecount 1 id [ "varchar", 1024, 0 ] sql>exec 1(null); 3 querytype env, env table_name name, value name varchar, varchar type 0, 0 length 0, tuplecount 1, id sql> ## Comment 3459 Date: 2005-06-23 17:49:21 +0200 From: @grobian Logged In: YES user_id=963970 This bug doesn't deal with prepare/exec functionality of the backend. As such, by coincidence a bug with the same effect on the output was fixed in the backend. The status of this bug is unaffected. ## Comment 3460 Date: 2005-06-24 17:06:30 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 IMO, mapi_param_store should substitute the PLACEHOLDER with NULL/nil, see the attached patch. ## Comment 3461 Date: 2005-07-04 14:18:49 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 In the light of http://sourceforge.net/support/tracker.php?aid=1229018, I suggest a revised patch: --- src/mapi/clients/C/Mapi.mx 4 Jul 2005 10:59:22 -0000 1.156.2.10 +++ src/mapi/clients/C/Mapi.mx 4 Jul 2005 12:10:46 -0000 @@ -2519,8 +2519,8 @@ hdl->query[k] = 0; if (hdl->params[i].inparam == 0) { - hdl->query[k++] = PLACEHOLDER; - hdl->query[k] = 0; + checkSpace(5); + strcpy(hdl->query + k, hdl->mid->languageId == LANG_SQL ? "NULL" : "nil"); } else { void *src = hdl->params[i].inparam; /* abbrev */ @@ -2604,9 +2604,8 @@ strcpy(hdl->query + k, src); break; } - k += strlen(hdl->query + k); } - + k += strlen(hdl->query + k); i++; p = q + 1; q = strchr(p, PLACEHOLDER); ## Comment 3462 Date: 2005-07-04 15:39:19 +0200 From: @sjoerdmullender Logged In: YES user_id=43607 The problem is that currently ODBC (through the good offices of Mapi) does the parameter stuff itself instead of having the server do the work. As to your patch, does that really work? If I try this query: select * from tables where query = NULL; I get an error from the server, whereas when I try this query: select * from tables where query = cast(NULL as varchar(1)); I get the expected result (the 1 can be another number also). It seems to me that your patch produces the first of these two queries (when you try select * from tables where query = ? with parameter value NULL). What should happen is that the NULL is translated into the correct cast(NULL as <TYPE>) depending on the type of the parameter. This is really something that ODBC should do, not Mapi, but currently the translation from ? to value happens in Mapi. (Actually, this is something that the server should do, but we're not there yet. That requires a change to the protocol. We're working on that, but it's slow going with all the other stuff that needs doing.) ## Comment 3463 Date: 2005-07-05 09:21:15 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 It works (and should work, IMHO) for INSERT/UPDATE: insert into x values( 1, null ); update x set s = null where i = 2; I choose the SELECT example as the smallest test case that demonstrates the current behavior (where NULLs are not bound). Of course, I don't expect that 'WHERE s = NULL' works (at least in a portable way), even if the parser would accept it. DBI devotes quite a large section on this topic: <http://search.cpan.org/~timb/DBI- 1.48/DBI.pmPlaceholders_and_Bind_Values> ## Comment 3464 Date: 2005-07-05 09:41:48 +0200 From: @sjoerdmullender Logged In: YES user_id=43607 Last night I came up with the idea that the ODBC driver could use the server to figure out information about the parameters. The server now understands the PREPARE statement (you just put "prepare" in front of a valid SQL statement-with-parameters) and it returns a table with information about the parameters. Most importantly, it returns the expected type. Even if we then do the filling in of the parameters ourselves (in ODBC or Mapi), we at least know the correct type and can convert any values to that type. But then again, this may be overkill in light of the text you referred to. So perhaps for now we should just accept your fix. ## Comment 3465 Date: 2005-07-15 11:05:14 +0200 From: @njnes Logged In: YES user_id=43556 patch applied. Could you test the result? ## Comment 3466 Date: 2005-07-15 13:52:04 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 parse error in: "select * from env where name = ;" You omitted one line: @@ -2521,7 +2521,6 @@ if (hdl->params[i].inparam == 0) { checkSpace(5); strcpy(hdl->query + k, hdl- >mid->languageId == LANG_SQL ? "NULL" : "nil"); - hdl->query[k] = 0; } else { void *src = hdl->params [i].inparam; /* abbrev */ ## Comment 3467 Date: 2005-12-19 11:41:33 +0100 From: @grobian Logged In: YES user_id=963970 BugDay_2005-12-19, meltdown: NO TEST / Perl-DBD-ODBC -> NO CLUE Is it possible to have an ODBC thing running without having to register it somewhere? Is there a HAVE_PERL or HAVE_DBD thing for TestWeb? Maybe a new ODBC C sample should be written to cover this. ## Comment 3468 Date: 2005-12-19 18:12:31 +0100 From: @drstmane Logged In: YES user_id=572415 see also https://sourceforge.net/tracker/?func=detail&atid=482468&aid=1229018&group_id=56967 ## Comment 3469 Date: 2006-01-15 15:14:40 +0100 From: @grobian Logged In: YES user_id=963970 Is this bug fixed in the meanwhile? ## Comment 3470 Date: 2006-02-14 10:41:21 +0100 From: @grobian Logged In: YES user_id=963970 ping/bump ## Comment 3471 Date: 2006-02-14 11:52:42 +0100 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 Works for me: my $sql = q(insert into env values('x',?)); my $sth = $dbh->prepare( $sql ); $sth->bind_param( 1, undef ); NULL $sth->execute; $sql = 'select * from env where value = ?'; $sth = $dbh->prepare( $sql ); $sth->bind_param( 1, undef ); NULL $sth->execute; while ( my $r = $sth->fetch ) { print @$r, "\n"; } gives: x ## Comment 3472 Date: 2006-06-06 20:51:31 +0200 From: @grobian Logged In: YES user_id=963970 bug is fixed. No idea on how to add a test as we now use server side prepares. ## Comment 3473 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1225455 at http://sourceforge.net/support/tracker.php?aid=1225455
Unable to bind NULL
https://api.github.com/repos/MonetDB/MonetDB/issues/618/comments
0
2020-11-30T08:45:31Z
2024-06-27T11:04:56Z
https://github.com/MonetDB/MonetDB/issues/618
753,279,541
618
[ "MonetDB", "MonetDB" ]
Date: 2005-06-16 10:15:57 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; To: @njnes Version: -- development Last updated: 2007-05-29 02:05:54 +0200 ## Comment 3450 Date: 2005-06-16 22:15:57 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; I shredded a 1GB XML document. If I shut down Mserver and restart it the committed, non-empty BATs of the 1 GB document are set to 'diffs' in the dirty field. (suffixes: 0 = 11MB, 1 = 110MB, 2 = 3 = 1 GB) 21:49:16 rittinge@sn-1:~/pathfinder/Linux$ echo 'ls();' | Mserver Monet Database Server V4.9.0 Copyright (c) 1993-2005, CWI. All rights reserved. Compiled for x86_64-redhat-linux-gnu/64bit with 32bit OIDs; dynamically linked. Visit http://monetdb.cwi.nl/ for further information. MonetDB>----------------------------------------------------------------- -------------------------------------------------------- name htype ttype count heat dirty status kind refcnt lrefcnt name str str str lng int str str str int int type --------------------------------------------------------------------------------- ---------------------------------------- [ "attr_own0", "void", "oid", 38266, 0, "clean", "disk", "pers", 0, 1 ] [ "attr_own1", "void", "oid", 381870, 0, "clean", "disk", "pers", 0, 1 ] [ "attr_own2", "void", "oid", 3829768, 0, "diffs", "disk", "pers", 0, 1 ] [ "attr_own3", "void", "oid", 3829768, 0, "diffs", "disk", "pers", 0, 1 ] [ "attr_prop0", "void", "oid", 38266, 0, "clean", "disk", "pers", 0, 1 ] [ "attr_prop1", "void", "oid", 381870, 0, "clean", "disk", "pers", 0, 1 ] [ "attr_prop2", "void", "oid", 3829768, 0, "diffs", "disk", "pers", 0, 1 ] [ "attr_prop3", "void", "oid", 3829768, 0, "diffs", "disk", "pers", 0, 1 ] [ "attr_qn0", "void", "oid", 38266, 0, "clean", "disk", "pers", 0, 1 ] [ "attr_qn1", "void", "oid", 381870, 0, "clean", "disk", "pers", 0, 1 ] [ "attr_qn2", "void", "oid", 3829768, 0, "diffs", "disk", "pers", 0, 1 ] [ "attr_qn3", "void", "oid", 3829768, 0, "diffs", "disk", "pers", 0, 1 ] [ "doc_height", "oid", "int", 4, 0, "clean", "disk", "pers", 0, 1 ] [ "doc_location", "oid", "str", 4, 0, "clean", "disk", "pers", 0, 1 ] [ "doc_name", "oid", "str", 4, 0, "clean", "disk", "pers", 0, 1 ] [ "doc_size", "oid", "lng", 4, 0, "clean", "disk", "pers", 0, 1 ] [ "doc_timestamp", "oid", "null", 4, 0, "clean", "disk", "pers", 0, 1 ] [ "id_pre0", "str", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "id_pre1", "str", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "id_pre2", "str", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "id_pre3", "str", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "idref_pre0", "str", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "idref_pre1", "str", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "idref_pre2", "str", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "idref_pre3", "str", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "kind_pre_00", "void", "oid", 167864, 0, "clean", "disk", "pers", 0, 1 ] [ "kind_pre_01", "void", "oid", 1666310, 0, "clean", "disk", "pers", 0, 1 ] [ "kind_pre_02", "void", "oid", 16703210, 0, "diffs", "disk", "pers", 0, 1 ] [ "kind_pre_03", "void", "oid", 16703210, 0, "diffs", "disk", "pers", 0, 1 ] [ "kind_pre_10", "void", "oid", 304816, 0, "clean", "disk", "pers", 0, 1 ] [ "kind_pre_11", "void", "oid", 3024328, 0, "clean", "disk", "pers", 0, 1 ] [ "kind_pre_12", "void", "oid", 30312003, 0, "diffs", "disk", "pers", 0, 1 ] [ "kind_pre_13", "void", "oid", 30312003, 0, "diffs", "disk", "pers", 0, 1 ] [ "kind_pre_20", "void", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "kind_pre_21", "void", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "kind_pre_22", "void", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "kind_pre_23", "void", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "kind_pre_30", "void", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "kind_pre_31", "void", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "kind_pre_32", "void", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "kind_pre_33", "void", "oid", 0, 0, "clean", "disk", "pers", 0, 1 ] [ "kind_pre_40", "void", "oid", 1, 0, "clean", "disk", "pers", 0, 1 ] [ "kind_pre_41", "void", "oid", 1, 0, "clean", "disk", "pers", 0, 1 ] [ "kind_pre_42", "void", "oid", 1, 0, "diffs", "disk", "pers", 0, 1 ] [ "kind_pre_43", "void", "oid", 1, 0, "diffs", "disk", "pers", 0, 1 ] ... ## Comment 3451 Date: 2005-07-15 13:58:50 +0200 From: @njnes Logged In: YES user_id=43556 A bbp.dir file will now only save 'persistent' information. So the incorrect BBPSWAPPED isn't used on the next run, ie making all bats correctly 'on disk' again. ## Comment 3452 Date: 2005-12-19 11:32:23 +0100 From: @grobian Logged In: YES user_id=963970 BugDay_2005-12-19, meltdown: NO TEST / NO CLUE Please someone else with a clue on how to test this properly, do so. ## Comment 3453 Date: 2006-12-20 04:20:04 +0100 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; ---- Original comment by: sf-robot@users.sourceforge.net ---- Logged In: YES user_id=1312539 Originator: NO This Tracker item was closed automatically by the system. It was previously set to a Pending status, and the original submitter did not respond within 365 days (the time period specified by the administrator of this Tracker). ## Comment 3454 Date: 2007-05-29 14:05:54 +0200 From: @njnes Logged In: YES user_id=43556 Originator: NO many tests with ls() output exist, ie tests exist, closing now. ## Comment 3455 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1222165 at http://sourceforge.net/support/tracker.php?aid=1222165
property dirty is set to \'diffs\' directly after startup
https://api.github.com/repos/MonetDB/MonetDB/issues/617/comments
0
2020-11-30T08:45:28Z
2024-06-27T11:04:55Z
https://github.com/MonetDB/MonetDB/issues/617
753,279,513
617
[ "MonetDB", "MonetDB" ]
Date: 2005-06-16 06:02:49 +0200 From: Wouter Alink &lt;<alink>&gt; To: XQuery devs &lt;<bugs-xquery>&gt; Version: -- development Last updated: 2005-07-21 02:47:17 +0200 ## Comment 3444 Date: 2005-06-16 18:02:49 +0200 From: Wouter Alink &lt;<alink>&gt; I haven't seen this one before: WARNING: BATpropcheck: BAT tmp_732 was incorrectly marked keyed! It was generated by using the following query: element { "result" } { for $q in doc("result.xml")//folder except doc("result.xml")//folder//folder where $q//*[contains(.,"aap")] return $q/@name } when it is rewritten as: element { "result" } { let $a := doc("result.xml") for $q in $a//folder except $a//folder//folder where $q//*[contains(.,"aap")] return $q/@name } it returns: ERROR: err:XQDY0025: attribute names are not unique in constructed element 'result'. which is probably correct. ## Comment 3445 Date: 2005-06-16 22:59:11 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; Logged In: YES user_id=993208 The document result.xml could be useful to find the bug :) ## Comment 3446 Date: 2005-07-21 12:52:11 +0200 From: @drstmane Logged In: YES user_id=572415 I totally agree with Jan ... ;-) ## Comment 3447 Date: 2005-07-21 14:47:17 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 i tried the queries again on the test-set and it seemed like the error has disappeared. maybe some other bug-fix fixed this bug? i close it for now (i could not provide the test-document, because of privacy...) ## Comment 3448 Date: 2005-12-19 11:29:38 +0100 From: @grobian Logged In: YES user_id=963970 BugDay_2005-12-19, meltdown: NO TEST / INVALID ## Comment 3449 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1222021 at http://sourceforge.net/support/tracker.php?aid=1222021 The original assignee of this bug does not have an account here. Reassigning to the default assignee for the component, bugs-xquery@monetdb.org. Previous assignee was nobody@users.sourceforge.net.
XQuery: duplicate attributes
https://api.github.com/repos/MonetDB/MonetDB/issues/616/comments
0
2020-11-30T08:45:25Z
2024-06-28T13:35:22Z
https://github.com/MonetDB/MonetDB/issues/616
753,279,485
616
[ "MonetDB", "MonetDB" ]
Date: 2005-06-16 04:51:48 +0200 From: Wouter Alink &lt;<alink>&gt; To: Jan Rittinger &lt;<jan.rittinger>&gt; Version: -- development Last updated: 2006-01-18 01:27:21 +0100 ## Comment 3434 Date: 2005-06-16 16:51:48 +0200 From: Wouter Alink &lt;<alink>&gt; element { " " } { "test" } returns: < >test</ > ## Comment 3435 Date: 2005-06-16 17:06:29 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 after a short tour over the keyboard: element { "/:~`',.;[]{}|<>+_-=!@$%^*()&"&apos;><" } { "test" } ## Comment 3436 Date: 2005-06-16 17:40:24 +0200 From: @grobian Logged In: YES user_id=963970 you forgot the most interesting output: </:~`',.;[]{}|<>+_-=!@$%^*()&"'><>test<//:~`',.;[]{}|<>+_-=!@$%^*()&"'><> :) ## Comment 3437 Date: 2005-06-16 21:41:06 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; Logged In: YES user_id=993208 As there is no validity check for tagnames there is nothing surprising here. If someone is willing to add a check, the function castQName in line 2466 of milprint_summer.c is the right place :) ## Comment 3438 Date: 2005-11-25 02:06:10 +0100 From: @peterboncz Logged In: YES user_id=591107 fixed; implemented invalid_qname() test ## Comment 3439 Date: 2005-12-19 11:28:14 +0100 From: @grobian Logged In: YES user_id=963970 BugDay_2005-12-19, meltdown: TEST ADDED / NO CLUE xquery/tests/BugDay_2005-12-19_0.9.3/Tests/tag_name_validation.SF-1221984.* Using MapiClient by hand, the added test results in: % MapiClient -lxq tag_name_validation.SF-1221984.xq ERROR: err:XPTY0004: illegal qname ' ' using mtest this error doesn't show up anywhere. I have no idea whether the above error is correct, given the bug report, I do suspect so... Someone needs to check the output and check in the correct stable output. ## Comment 3440 Date: 2006-01-14 21:15:52 +0100 From: @drstmane Logged In: YES user_id=572415 Question for the XQuery experts: Is(n't?) the error message the correct/expected output for this query? If so (my layman's guess), we can approve the current output as stable, and close this report. If not, what is the correct/expected output, and who's going to fix it? see also http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.32.32.d-Fedora4/tests_BugDay_2005-12-19_0.9.3/tag_name_validation.SF-1221984.err.00.html http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.32.32.d-Fedora4/tests_BugDay_2005-12-19_0.9.3/tag_name_validation.SF-1221984.out.00.html ## Comment 3441 Date: 2006-01-18 09:47:11 +0100 From: @drstmane Logged In: YES user_id=572415 Sorry, I provided the wrong URLs. The correct ones are http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_BugDay_2005-12-19_0.9.3/tag_name_validation.SF-1221984.out.00.html http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_BugDay_2005-12-19_0.9.3/tag_name_validation.SF-1221984.err.00.html ... any verdict on the correctness of the currect output ? ## Comment 3442 Date: 2006-01-18 13:27:21 +0100 From: Jan Rittinger &lt;<jan.rittinger>&gt; Logged In: YES user_id=993208 approved output ## Comment 3443 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1221984 at http://sourceforge.net/support/tracker.php?aid=1221984
XQuery: tag name validation
https://api.github.com/repos/MonetDB/MonetDB/issues/615/comments
0
2020-11-30T08:45:22Z
2024-06-27T11:04:53Z
https://github.com/MonetDB/MonetDB/issues/615
753,279,455
615
[ "MonetDB", "MonetDB" ]
Date: 2005-06-15 05:59:20 +0200 From: Wouter Alink &lt;<alink>&gt; To: Jan Rittinger &lt;<jan.rittinger>&gt; Version: -- development Last updated: 2006-01-15 11:45:51 +0100 ## Comment 3423 Date: 2005-06-15 17:59:20 +0200 From: Wouter Alink &lt;<alink>&gt; Using Pathfinder 0.8 CVS ERROR: merged_union: tail of first BAT must be sorted. let $a := (<a/>,<a/>) for $i in 0 to 3 let $b := zero-or-one($a[$i]) return element { name($b) } { $b/@* } ## Comment 3424 Date: 2005-06-16 13:58:17 +0200 From: @drstmane Logged In: YES user_id=572415 Wouter, this is not an merged_union problem, but a property problem "somewhere" else in MonetDB. I can only check it once I'm back in A'dam last week of June. For now, please try to run Mserver with property checking enabled; i.e., `Mserver --debug=10`. Stefan. ## Comment 3425 Date: 2005-06-16 16:50:58 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 If I use the option: --debug=10 instead of the error, the following result is returned: </>, <a/>, <a/>, </> this is the correct answer, although it is invalid. i am not sure what the problem is, but i guess that the element-constructor should not accept the empty string (or a string with spaces) as the xml-tag-name. ## Comment 3426 Date: 2005-06-16 23:02:39 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; Logged In: YES user_id=993208 it is now fixed by adding more chk_order calls, and still throws no error message because of empty element name This bug could have been avoided if order properties are exploited even more :) x: sorted|sorted y: sorted|sorted x.join(y): sorted|sorted ## Comment 3427 Date: 2005-07-21 13:06:45 +0200 From: @drstmane Logged In: YES user_id=572415 Even in this case, correct property propagation is a bit more compilcated than it might seem: x: sorted|sorted y: sorted|sorted ==> x.join(y): sorted|sorted only holds if (1) the respective join algorithm scans one of the BATs (say x) sequentially, looking up the matches in the other BAT (say y); and (2) with duplicate values in the join column/attribute of the looked-up BAT (y), results are produced in the original order; and (3) no join-attribute value occurs multiple times in both BATs (x & y). In our case (MonetDB), (1) hold for all join algorithms (AFAIK), (2) holds for mreg-join, but not for hash-join, and (3) could only be approximated by "(3') (at least) one of the two join columns is unique (key)"... ... I'll check, whether our join algorithms do exploit (3') ... ## Comment 3428 Date: 2005-11-25 00:23:29 +0100 From: @peterboncz Logged In: YES user_id=591107 it seems this is fixed. Wouter, can you confirm this? ## Comment 3429 Date: 2005-12-19 11:10:18 +0100 From: @grobian Logged In: YES user_id=963970 BugDay_2005-12-19, meltdown: TEST ADDED / NO CLUE pathfinder/tests/BugDay_2005-12-19_0.9.3/Tests/zero-or-one_merged-union.SF-1221336.* Wouter, can you check the output of this test? It returns nothing for me. ## Comment 3430 Date: 2006-01-14 21:36:14 +0100 From: @drstmane Logged In: YES user_id=572415 The original version of the query results in a ERROR: err:XPTY0004: illegal qname '' which is IMHO correct, since in the first iteration ($i==0), $b=zero-or-one($a[0]) is empty (XQuery indices start at 1, not at 0 ??), hence, name($b) is empty, which is indeed not a legal qname. I added a second test query that iterates only over the existing items in $a (1 to 2) let $a := (<a/>,<a/>) for $i in 1 to 2 let $b := zero-or-one($a[$i]) return element { name($b) } { $b/@* } and that one nicely (and correctly, right?) returns <?xml version="1.0" encoding="utf-8"?> <XQueryResult> <a/><a/> </XQueryResult> Do the XQuery expert agree?? If so, we should approve http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.32.32.d-Fedora4/tests_BugDay_2005-12-19_0.9.3/zero-or-one_merged-union.SF-1221336.out.00.html http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.32.32.d-Fedora4/tests_BugDay_2005-12-19_0.9.3/zero-or-one_merged-union.SF-1221336.err.00.html and close this report... ## Comment 3431 Date: 2006-01-15 11:22:47 +0100 From: @Teggy Logged In: YES user_id=642632 Yes, the output looks OK to me. (In an XQuery sequence $s of the form (x,y,z), item x is $s[1], not $s[0].) --Teggy ## Comment 3432 Date: 2006-01-15 11:45:51 +0100 From: @drstmane Logged In: YES user_id=572415 The exquery expert does agree; hence, we can close this one. ## Comment 3433 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1221336 at http://sourceforge.net/support/tracker.php?aid=1221336
XQuery: merged_union
https://api.github.com/repos/MonetDB/MonetDB/issues/614/comments
0
2020-11-30T08:45:19Z
2024-06-27T11:04:52Z
https://github.com/MonetDB/MonetDB/issues/614
753,279,416
614
[ "MonetDB", "MonetDB" ]
Date: 2005-06-15 03:00:16 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @njnes Version: -- development Last updated: 2005-06-24 02:50:23 +0200 ## Comment 3419 Date: 2005-06-15 15:00:16 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; Running create table t(i int); prepare insert into t(i) values (?); results in a server crash. ## Comment 3420 Date: 2005-06-24 14:50:23 +0200 From: @grobian Logged In: YES user_id=963970 On CVS HEAD this currently works fine: % MapiClient -ls sql> create table t(i int); 6 querytype sql> prepare insert into t(i) values (?); 10 querytype prepare, prepare, prepare table_name type, digits, scale name varchar, int, int type 0, 0, 0 length 1 tuplecount 0 id [ "int", 9, 0 ] sql>exec 0(1); 4 querytype [ 1 ] sql> So I assume this has been fixed in the mean time. ## Comment 3421 Date: 2005-12-19 10:47:06 +0100 From: @grobian Logged In: YES user_id=963970 BugDay_2005-12-19, meltdown: ALREADY IN TESTWEB amongst others: sql/src/test/BugTracker/Tests/large_prepare.SF-1363729.* ## Comment 3422 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1221151 at http://sourceforge.net/support/tracker.php?aid=1221151
Crash: prepare insert into ...
https://api.github.com/repos/MonetDB/MonetDB/issues/613/comments
0
2020-11-30T08:45:16Z
2024-06-27T11:04:51Z
https://github.com/MonetDB/MonetDB/issues/613
753,279,379
613
[ "MonetDB", "MonetDB" ]
Date: 2005-06-15 01:44:02 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; To: Stefan Manegold &lt;<Stefan.Manegold>&gt; Version: -- development Last updated: 2005-07-04 10:59:08 +0200 ## Comment 3407 Date: 2005-06-15 13:44:02 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; When I tried to load a 1GB document (xmlgen -f10) with shred_doc() it does not finish. I tested it on the 64-bit machine at CWI (since I first thought it could be a ram problem) and on the 32-bit suse machine in KN. In both cases Mserver starts with 100% CPU utilization and falls back to 1% after a minute. After half an hour it then wrote something to disk (and kept idling?). Could bug [ 1207707 ] 'XQuery: shredding attributes takes time' be the reason? ## Comment 3408 Date: 2005-06-15 13:50:36 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; Logged In: YES user_id=993208 it just finished: Shredded XML doc("auction10MB.xml"), total time after commit=0.859s Shredded XML doc("auction100MB.xml"), total time after commit=8.039s Shredded XML doc("auction1GB.xml"), total time after commit=1506.999s ... it just finished on the faster 64-bit machine ## Comment 3409 Date: 2005-06-15 13:50:48 +0200 From: Wouter Alink &lt;<alink>&gt; Logged In: YES user_id=621590 I don't think the "shredding attributes"-bug is related, but maybe the "bus error"-bug, because the "Bus Error" message was only raised after several minutes (of which CPU-utilization is only 100% in the first few seconds) ## Comment 3410 Date: 2005-06-16 11:48:02 +0200 From: Jan Flokstra &lt;<jflokstra>&gt; Logged In: YES user_id=1054297 I had a quick look at the problem and I had the same symptons. I am not really an expert in virtual memory but I first thought the system was swapping. But when I ran vmstat it showed no pageswaps in/or out. I only saw a modest amount of bi (blocks in) beween 400 and 700. Furthermore vmstat said that of my 1G memory 250M was active and 750M was inactive!!!! In the coming days I will have a deeper look at the problem but I think the problem could be caused by Monet memory management and especially the way we (I?) use it! ## Comment 3411 Date: 2005-06-16 13:55:05 +0200 From: @drstmane Logged In: YES user_id=572415 Jan^2, could you please change the memory settings of MonetDB and try again: Before calling shred_doc, call the following on a 1 GB machine: vm_minsize ( 500000000LL); mem_bigsize( 500000000LL); mem_maxsize(1000000000LL); and on a 8GB machine: vm_minsize(4000000000LL); mem_bigsize(4000000000LL); mem_maxsize(8000000000LL); Please let us know, whether this helps. Stefan. ## Comment 3412 Date: 2005-06-16 14:29:10 +0200 From: Jan Flokstra &lt;<jflokstra>&gt; Logged In: YES user_id=1054297 THIS WORKS! For me this does the trick. It even starts swapping after a while. My figures are on a 1G machine: auction100M=6s, auction1G=260s. Hope it also works for sqrt(Jan^2). What should we do/conclude concerning this bug? A message in the manual about working with very large data sets?? ## Comment 3413 Date: 2005-06-16 16:56:54 +0200 From: @drstmane Logged In: YES user_id=572415 Yes, for now, a hint in the docu/HowToStart might be good. Eventually, we should make Mserver set/use the respective values by default: vm_minsize & mem_bigsize: ~50% of the physically available memory; mem_maxsize: ~100% of the physically available memory. ... I should have done this before the release, but it apparently got lost ... ## Comment 3414 Date: 2005-06-16 16:58:24 +0200 From: @drstmane Logged In: YES user_id=572415 ps: We would actually have to verify, first, whether these settings also work fine for other OSs than Linux... ## Comment 3415 Date: 2005-06-16 22:09:17 +0200 From: Jan Rittinger &lt;<jan.rittinger>&gt; Logged In: YES user_id=993208 Stefan, your comment helped: Shredded XML doc("auction1GB.xml2"), total time after commit=88.563s But there is another open point. The same problem happens for querying the 1GB document and we can't ask users to set this property. Somebody who just uses our system will conclude that the system is NOT able to evaluate larger documents and therefore also does NOT scale :( The default loading of this values therefore is important (in my opinion). ## Comment 3416 Date: 2005-07-04 10:59:08 +0200 From: @drstmane Logged In: YES user_id=572415 fixed in CVS (MonetDB_4-8 release branch): by default, Mserver now sets mem_maxsize to 95% of the physically available amount of main-memory, and vm_minsize and mem_bigsize to 1/2 of mem_maxsize. ## Comment 3417 Date: 2005-12-19 10:42:15 +0100 From: @grobian Logged In: YES user_id=963970 BugDay_2005-12-19, meltdown: NO TEST / TOO LARGE ## Comment 3418 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1221107 at http://sourceforge.net/support/tracker.php?aid=1221107
XQuery: shred_doc does take to long for 1GB document
https://api.github.com/repos/MonetDB/MonetDB/issues/612/comments
0
2020-11-30T08:45:13Z
2024-06-27T11:04:50Z
https://github.com/MonetDB/MonetDB/issues/612
753,279,340
612
[ "MonetDB", "MonetDB" ]
Date: 2005-06-10 05:14:28 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; To: @sjoerdmullender Version: -- development Last updated: 2006-01-20 01:26:36 +0100 ## Comment 3398 Date: 2005-06-10 17:14:28 +0200 From: Steffen Goeldner &lt;<sgoeldner>&gt; OS: Windows 2000 Compiler: MS VC++ Toolkit 2003 Parser tools: UnxUtils Start the server: bin\Mserver.exe -c etc\MonetDB.conf --dbinit="module (mapi); mapi_start();" --set gdk_debug=128 Start a client: bin\MapiClient.exe -c etc\MonetDB.conf -s quit(); The server exits with: ... MAPIlisten:client accepted 704 ------------------------------------------------------------------ [ 0, 2, running, 0, "listen(get_mapi_port (),5," ] [ 1, 1, running, 0, "unknown(0) " ] [ 2, 0, running, 0, "unknown(1) " ] MIget = unknown(1) setchannel Mserver mapi read Mserver mapi write ------------------------------------------------------------------ req | tid | status | wakeup | action ------------------------------------------------------------------ [ 0, 2, running, 0, "listen(get_mapi_port (),5," ] [ 1, 1, running, 0, "unknown(0) " ] [ 3, 0, 1, 0, "unknown(1) " ] [ 4, 0, running, 3, "quit() " ] MIget = quit() setchannel Mserver mapi read Mserver mapi write Assertion failed: s->access == ST_WRITE, file .\..\..\..\src\common\stream.mx, line 414 ## Comment 3399 Date: 2005-06-24 14:54:53 +0200 From: @grobian Logged In: YES user_id=963970 I can not reproduce this problem on Linux using todats current branch ## Comment 3400 Date: 2005-07-05 10:04:12 +0200 From: @sjoerdmullender Logged In: YES user_id=43607 I can reproduce this. Are you sure you have assertions enabled? The problem is a bit tough to fix, though. When the client exits, the server thread is destroyed, and with it the streams to and from the client. But after this happens (call to freeClient() in closeClient()), a request is still sent to the thread to kill itself (putkillRequest()). This latter call tries to print to the stream that was just closed if PARDEBUG is set (debugmask=128). ## Comment 3401 Date: 2005-12-19 10:41:33 +0100 From: @grobian Logged In: YES user_id=963970 BugDay_2005-12-19, meltdown: TEST ADDED / NO CLUE MonetDB/tests/BugDay_2005-12-19_4.9.3/Tests/stream_assert_quit.SF-1218291.* What's the status of this bug? ## Comment 3402 Date: 2006-01-14 21:48:32 +0100 From: @drstmane Logged In: YES user_id=572415 Any news? The respective test seems to work fine on all our testing platforms: http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.64.64.d-Fedora4/tests_BugDay_2005-12-19_4.9.3/stream_assert_quit.SF-1218291.out.00.html http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.64.64.d-Fedora4/tests_BugDay_2005-12-19_4.9.3/stream_assert_quit.SF-1218291.err.00.html ## Comment 3403 Date: 2006-01-18 09:49:16 +0100 From: @drstmane Logged In: YES user_id=572415 Steffen, does this bug still occur with you? ## Comment 3404 Date: 2006-01-18 16:47:06 +0100 From: Steffen Goeldner &lt;<sgoeldner>&gt; Logged In: YES user_id=1174087 Yes: Assertion failed: s->access == ST_WRITE, file .\..\..\..\src\common\stream.mx, line 421 ## Comment 3405 Date: 2006-01-20 13:26:36 +0100 From: @sjoerdmullender Logged In: YES user_id=43607 After Stefan's change to write 0xDB in freed memory, this test now crashes. The fix is to not attempt to print the queue after the thread was killed. Fixed in CVS. ## Comment 3406 Date: 2010-05-04 09:32:09 +0200 From: Pseudo user for Sourceforge import &lt;<sfimport>&gt; This bug was previously known as tracker item 1218291 at http://sourceforge.net/support/tracker.php?aid=1218291
Assertion failed: s->access == ST_WRITE: stream.mx, line 414
https://api.github.com/repos/MonetDB/MonetDB/issues/611/comments
0
2020-11-30T08:45:10Z
2024-06-27T11:04:49Z
https://github.com/MonetDB/MonetDB/issues/611
753,279,317
611