- developers-list - monetdb.org

[Monetdb-developers] TTT goes MonetDB-Bug-Day
by Stefan Manegold 06 Oct '05

06 Oct '05

[ I deliberately send this email to various related mailing lists; ] [ in case you receive this email more than once, it might be time ] [ to teach your email client to eliminate multiple copies of the ] [ very same email... ] Dear all, (admittedly on short notice --- sorry for any related inconvenience) you're all "invited" to join a community effort to significantly improve the the MonetDB quality control system: Let's (try to) make "Mtest.py-compliant" test scripts from all bug reports that have been filed on SF since July 2002, and thus ensure that these bugs do not re-occur, respectively ensure that we (via the "TestWeb"; cf. http://monetdb.cwi.nl/Development/TestWeb/) immediately notice whenever code changes do trigger an "old" bug. Summary: ======== When: Thursday, Oct 6, 2005, "all day" Where: Mainly at CWI, but you're also welcome to join "virtually" working from home or any other place in the world; for those who want to turn the Bug-Day into a Bug-"Party", I reserved our meeting room (C0,01) from 12:30 - 17:00 --- you need to bring your own "equipment" though ... Who: Primarily everyone who is in touch with MonetDB, and/or one of its companions/front ends (SQL, XQuery, etc.), has rights to check-in to the MonetDB CVS repository on SourceForge, and is reasonably familiar with "Mtest.py" (cf. http://monetdb.cwi.nl/monet/src/testing/README) But I will also (try to) provide "Testing HowTo" by Thursday, so others are welcome to join; further, if you cannot check-in to the MonetDB CVS repository yourself, a colleague of your's (or one of the core developers) might do this for you, once you prepared the complete test. What: I will prepare a list of all bug that have to test script, yet, and assign them round-robin to all participants (probably, I will try to classify them in say "MIL", "SQL", "XQuery", and "others", and you're welcome to choose your preferred class, given availability). More detailed instructions will follow by Thursday. Background: =========== Next to providing high performance, maintaining correctness and stability of the code base is the major goal of MonetDB and its companions. First of all, the bug tracking system on SourceForge (http://sourceforge.net/tracker/?group_id=56967&atid=482468) is used intensively to report any problems related to MonetDB --- since MonetDB moved to SourceForge in July 2002, almost 700 bug reports have been filed, most of which have been solved. Secondly, our automatic Nightly Multi-Platform Regression Test System (aka "TestWeb"; cf. http://monetdb.cwi.nl/Development/TestWeb/) helps us to ensure that bugs that sneak in with modified or newly added code are detected (and fixed) instantly. Obviously, the test system only work for that features/code that are indeed covered by the available test scripts... Given these two tools, the bug tracker and the test system, we can easily build a system that prevent the (unnoticed) re-occurrence of bug that have been reported and fixed before: We "simply" have to add a proper test script for each bug report. Unfortunately, so far there are test script only for a fraction of the ~700 bug reports. Given that many bugs are reported by users that are not allowed to add test script to the CVS repository, it is not convenient (not even possible) to require that each bug report includes an "Mtest.py-compliant" (cf. http://monetdb.cwi.nl/monet/src/testing/README) test script. Nevertheless, most bug reports come with a query, a script, or a detailed description how to reproduce the respective bug; hence, it should be rather simple to turn these queries / scripts / description into proper "Mtest.py-compliant" test scripts, and thus significantly improve our automatic quality control system. Since this task is quite a challenge (in terms of volume, not intellectual complexity ;-)), the idea popped-up to make it a "community effort", as sketched in the "Summary" above. The advantage of this coordinated community effort should be, that all of us dedicate our time to this concurrently, share the same "context" and hence can support each other efficiently when/where necessary. In advance, thank you very much for your cooperation! Regards, Stefan -- | Dr. Stefan Manegold | mailto:Stefan.Manegold@cwi.nl | | CWI, P.O.Box 94079 | http://www.cwi.nl/~manegold/ | | 1090 GB Amsterdam | Tel.: +31 (20) 592-4212 | | The Netherlands | Fax : +31 (20) 592-4312 |

8 9

[Monetdb-developers] BUG in MonetDB V4.8.2 for windows
by Romulo Gonçalves 05 Oct '05

05 Oct '05

This email is to report a bug in MonetDB V4.8.2 for windows. My name is Rómulo Gonçalves, number 0536601. When I try to create a table with multiply primary key with the following code (SQL99) : CREATE TABLE URLS ( ID_URL int, ID_Division int, PRIMARY KEY (ID_URL, ID_Division) ); ALTER Table URLs ADD foreign key(ID_Division) references URL_Division(ID_Division); I receive the following error: "[MonetDB][ODBC Driver 1.0]! Could not find referenced primary key in table urls" Then I tried with the following code: CREATE TABLE URLS ( ID_URL int, ID_Division int, primary key(ID_URL), primary key(ID_Division) ); And no error happen when I create the table. Here is the bug. The MonetDB accept that we insert more than one key and then it accepts the first (in this case ID_URL) as a primary key. But there is another bug. I can not create relational tables because I need to have a multiply primary key, in this case one tuplo (ID_URL,ID-Division) and for this I need to have the possibility to do like the following example: CREATE TABLE URLS ( ID_URL int, ID_Division int, * PRIMARY KEY (ID_URL, ID_Division) -> MonetDB not accept this way to define multiple primary key.* ); Maybe the report is not a perfect report because is the first time that I report bugs of one System. I hope that you understand the two bugs of MonetDB that I found, if there is something that you do not understand, please contact me. Greets, Rómulo....

2 1

[Monetdb-developers] Help
by joseph_sciarra＠dofasco.ca 26 Sep '05

26 Sep '05

I am working on an ORACLE Database and the xqueries I am running take forever to run Is there anything I can Do?? Thanks Joe

2 1

[Monetdb-developers] JDBC: transaction/savepoint issue
by Andrei Martsinchyk 02 Sep '05

02 Sep '05

Hello, I am trying to use MonetDB as a backend for Java application I am developing. I have noticed that inserts the application does have no effect. Statements are executed without exceptions, but data rows do not get into table. I have investigated the problem and have figured out that problem is in transaction and savepoint handling. If transaction had savepoints, it could be silently rolled back when Connection.commit() is called. I have finished with short Java program that demonstrate the bug(?), I am attaching it to the message. Please, take a look. I tested with MonetDB-4.8.2+sql-2.8.2 (release, source tarball) and with MonetDB-4.8.3+sql-2.8.3 (nightly build, August 30). Platforms: 32-bit Linux RedHat 9, 32-bit Linux Ubuntu 5.04 -- Best regards, Andrei Martsinchyk mailto:andrei.martsinchyk@gmail.com

2 3

[Monetdb-developers] Directories not created by install
by Robi Sen 02 Sep '05

02 Sep '05

I just installed the windows version of MonetDB and tried to follow the Tutorial on Mknife but was unable to connect. Looking at my system neither of the files /infosets/MILinfoset.xml;/infosets/MILdemo.xml where created nor where the directories. Is this a know issue?

2 1

Re: [Monetdb-developers] Re: [ monetdb-Bugs-1274819 ] "type error: no variant of function ... accepts the the giv"
by Ethan Aubin 28 Aug '05

28 Aug '05

On Mon, Aug 29, 2005 at 12:03:29AM +0200, Ying Zhang wrote: > I'm very sure there is a "make check" for every MonetDB module, so does > pathfinder. I believe it! I just found the immense TestWeb page today. I was initially assuming that pathfinder was distributed with its tests like MonetDB. [Side node: IMO, the web page might be easier to navigate if the subnodes on the left hand side had the same content as the submenus on the top bar. E.g. Starting from http://monetdb.cwi.nl/XQuery/index.html and clicking the Development on the left hand side lists a submenu 'Download from CVS', while clicking Development on the top takes me to a page with 'Partners', 'Contributors', 'TestWeb', etc on the left menu. This was how I missed the TestWeb earlier] Your were right about Mtest not being in my path. I forgot to 'rehash' after installing MonetDB. Cheers - EA

1 0

[Monetdb-developers] Re: [ monetdb-Bugs-1274819 ] "type error: no variant of function ... accepts the the giv"
by Jens Teubner 28 Aug '05

28 Aug '05

(It starts getting a communication thread now, no longer a bug track. So I'll reply via email, with CC to the monetdb-developers list.) > 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? Of course are those test scripts available. Just type `make check'. Be aware, though, that those test cases take their time. Also note that there are currently some tests in that do not work as expected. So you will not get a 100% pass. The `make check' run will print a summary after processing. You will also find a directory that contains the detailed test results marked up with HTML (directory `mTests'). Tests with unexpected outcome are highlighted in red; you will see the diffs between actual and expected outcome. Results of the nightly test runs, btw., are available on the MonetDB TestWeb website: http://monetdb.cwi.nl/Development/TestWeb/index.html. > 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? Most test cases are tested ``by hand'', sometimes by comparing against other implementations. At least in theory, all of them are cross-checked by hand. > > (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. :-) Regards from Munich Jens -- Jens Teubner Technische Universitaet Muenchen, Department of Informatics D-85748 Garching, Germany Tel: +49 89 289-17259 Fax: +49 89 289-17263 "Tea or coffee?" "Coffee, without cream, please." "It will be without milk, we have no cream."

2 1

[Monetdb-developers] new requirements for building MonetDB and associated projects
by Sjoerd Mullender 24 Aug '05

24 Aug '05

There is a new (well already quite a few weeks old) CVS module available at our SourceForge repository called buildtools. This module is now a prerequisite for compiling MonetDB and all other projects. Check out buildtools: export CVSROOT=... cvs checkout buildtools cd buildtools # read REAME file and follow instructions Then you can update your MonetDB and other projects and bootstrap and build those projects. The main difference is: - autogen.py must be found in your $PATH (autogen is now part of buildtools); - Mx must either be found in your $PATH, or you must specify the option --with-mx=.../Mx on the configure command line; - mel must either be found in your $PATH, or you must specify the option --with-mel=.../mel on the configure command line; - for Pathfinder, burg must either be found in your $PATH, or you must specify --with-burg=.../burg on the configure command line. -- Sjoerd Mullender

3 4

[Monetdb-developers] AMD64 version of MonetDB for Windows?
by Cardillo Raymond A Civ AFRL/IFEA 19 Aug '05

19 Aug '05

Has anyone out there taken the time to build an MSI for the AMD64 versions of Windows? The x64 version of Windows was released a few months ago and was available in Beta for a while, so I'm hoping someone has either done this already, or can provide some advice on how to build/compile for this platform. Anyone know of future plans to make this one of the standard platform builds? Windows x64 is only going to grow in popularity from here, especially amongst researchers and others who can benefit from 64 bits. Thanks for any info, Ray Cardillo

2 1

[Monetdb-developers] quotes around chr values
by Sjoerd Mullender 17 Aug '05

17 Aug '05

I have prepared a patch to put (single) quotes around chr values in tables as printed by the Mserver. E.g.: MonetDB>'a'.print(); [ 'a' ] Previously this resulted in MonetDB>'a'.print(); [ a ] In addition to adding quotes, my patch also escapes control characters: MonetDB>chr(10).print(); [ '\012' ] Previously: MonetDB>chr(10).print(); [ ] This also shows a very important reason for wanting to make this change. Another example which is possibly even worse is chr(0).print(). I ran the pathfinder and sql tests, and only in pathfinder there are 2 tests that are affected (I will appprove them when I check the patch in). This message is a heads up. If you think you might be affected (JDBC, Perl, others?), then maybe you will need to make changes once I check this in. The Mapi library already deals with single-quoted values. -- Sjoerd Mullender

1 0