- developers-list - monetdb.org

Re: [Monetdb-developers] MonetDB: default - Top-level bootstrap/configure/make.
by Stefan Manegold 07 Jan '11

07 Jan '11

Dear all, since I have not yet managed to update the TestTools to cope with Sjoerd's changes, i.e., the new (single) top-level bootstrap/configure/make, I will suspend "Current" ("default") testing as of tonight, until I manage to update (or reimplement) the TestTools accordingly. Please accept my aplologies for any inconveniences that this might cause. Stefan On Fri, Jan 07, 2011 at 01:41:44PM +0100, Sjoerd Mullender wrote: > Changeset: 09ddaf9782c1 for MonetDB > URL: http://dev.… [View More]monetdb.org/hg/MonetDB?cmd=changeset;node=09ddaf9782c1 > Modified Files: [1219] > Branch: default > Log Message: > > Top-level bootstrap/configure/make. > > This huge changeset is to turn the various "sort-of" independent > subpackages of MonetDB into one coherent package. The principle idea > is that we now have a single top-level configure and make instead of > having to run configure and make many times to build everything. > > >From now on, the way to build MonetDB is to run the command > "./bootstrap" in the top-level directory, and then to run > ".../configure" and "make" in the build directory (which can be the > same as the source, but doesn't have to be). On Windows, the way to > build is to run "nmake" in the NT subdirectory or to run "nmake /f > ..\NT\Makefile" in an empty directory underneath the top level. > > The subpackages affected are: MonetDB, clients, MonetDB4, MonetDB5, > sql, pathfinder, geom, testing, java, and python. Not affected is > template (although it likely won't currently work) since it is meant > as a template for external (to the MonetDB suite) extensions. > > Note that the python subpackage is included in the list: there is now > support to run the Python setup.py program with autogen. > > The top-level configure has a number of options to enable and disable > the building of subpackages: > --enable-monetdb4 enable support for MonetDB4 (default=auto) > --enable-monetdb5 enable support for MonetDB5 (default=auto) > --enable-sql enable support for MonetDB/SQL (default=auto) > --enable-geom enable support for geom module (default=auto) > --enable-pathfinder enable support for MonetDB/XQuery (default=auto) > --enable-pftijah enable support for TIJAH (default=auto) > --enable-probxml enable support for PROBXML (default=auto) > --enable-odbc compile the MonetDB ODBC driver (default=auto) > --enable-testing enable support for testing (default=auto) > --enable-console enables direct console on the server (involves > security risks) > --enable-java (try to) build Java components > --enable-jdbc build the MonetDB JDBC driver > --enable-xrpcwrapper build the MonetDB XRPC wrapper > --enable-merocontrol build the Merovingian control driver > > The options to specify the presence and location of programs and > libraries that are used to build the subpackages were of course also > integrated into the top-level configure, as were the following > options: > --with-default-backend=BACKEND > select a default back-end > (milprint_summer/algebra/sql) for Pathfinder > (default=algebra) > --with-password-backend=HASHALG > password hash algorithm, one of MD5, SHA1, > RIPEMD160, SHA224, SHA256, SHA384, SHA512, defaults > to SHA512 > > Note that the Pathfinder configuration without MonetDB4 is still possible: > .../configure --disable-monetdb4 --disable-monetdb5 --enable-pathfinder ... > > Other changes include the following: > - One top-level MonetDB.spec file so that all RPMs can be built in one > go; > - One top-level debian subdirectory with Debian/Ubuntu configuration; > - Removed all *-config and *-config.bat files: there use is superseded > by the single top-level configure and various *.pc (pkgconfig) files; > - All paths in Makefile.ag files to other subpackages are now relative; > - .mx files in the mel package have been unpacked to their constituent > .cc (etc.) files: adding back support for C++ inside autogen was too > hard otherwise; > - Tar balls don't include generated files (apart from configure): the > build process, also for tar ball distributions, includes building > and using Mx, but also now requires bison/yacc and flex/lex (there > have been problems in the past with our distributing generated bison > output); > - mel is only built when MonetDB4 is enabled, burg is only built when > pathfinder is enabled; > - Some symbols in burg were replaced to avoid name clashes on Windows; > - Some files (monetdb.ico, banner.bmp, website.html, license.rtf) were > moved to buildtools/conf; > - Use the currently approved way (by autoconf) of checking for alloca > and alloca.h; > - Packaged vault.sql into vault.mx since autogen/Mx get confused > otherwise; > - Set the version number globally to 11.0.0 (we need a single version > number for binary packages, and it must be higher than the highest > old version number). > > -- | Stefan.Manegold @ CWI.nl | DB Architectures (INS1) | | http://CWI.nl/~manegold/ | Science Park 123 (L321) | | Tel.: +31 (0)20 592-4212 | 1098 XG Amsterdam (NL) | [View Less]

1 0

[Monetdb-developers] changes to building MonetDB
by Sjoerd Mullender 05 Jan '11

05 Jan '11

This is an important announcement for everyone who wants to build MonetDB from sources (tar balls and Mercurial). This announcement is only relevant to the "default" branch. Pretty soon I will commit and push a big change to the way MonetDB is built. This message is a heads-up for that change and contains information on how to build MonetDB in the (near) future. Currently, in order to build the MonetDB suite, you need to do the following: * build and install buildtools (only if building … [View More]from Mercurial sources); * build and install MonetDB; * build and install clients; * build and install MonetDB4; * build and install MonetDB5; * build and install sql; * build and install pathfinder; * build and install geom; * build and install java; * build and install python; * build and install testing. Some of the above are optional, and some variation in the order is possible. In all cases (except python), building and installing consists of running: * ./bootstrap * .../configure ... * make * make install This whole process is going to change. Very soon you will need to run the commands * ./bootstrap (not for tar balls) * .../configure ... * make * make install only once in the top-level directory of the MonetDB suite (well, ./bootstrap has to run there, the other commands can run either there or in a separate build directory (recommended)). For now, there are no changes to the way the Python module is built. The advantage of this change is clear: you only need to run these commands once instead of 9 times. This means that both bootstrap and configure will be faster, even though the single instance needs to do a bit more work than any single instance in the old situation (but there were 9 of those). Another advantage is that things like * make dist (create tar ball) * make rpm (create binary RPM) * make srpm (create source RPM) also need to be run only once for the complete suite. We no longer have a separate Super Source Tar Ball which was just a tar ball containing all the separate source tar balls. Note that in the old situation, there were a whole bunch of configuration options in the Pathfinder configure. These options have been moved to the top-level configure. It used to be possible to only configure and build Pathfinder. That is now supported by using options on the top-level configure disabling building of MonetDB4 and MonetDB5 subpackages: .../configure --disable-monetdb4 --disable-monetdb5 [pathfinder-specific config options] Attached is the check-in message that I plan to use (modulo last minute changes). -- Sjoerd Mullender [View Less]

1 0

[Monetdb-developers] Announcement: New Jun2010-SP1 bugfix release of MonetDB suite
by Sjoerd Mullender 21 Dec '10

21 Dec '10

The MonetDB team at CWI/MonetDB BV is pleased to announce the Oct2010-SP1 bug fix release of the MonetDB suite of programs. The most important and visible change is a security fix in the SQL COPY INTO command. This is detailed in the release notes. More information (including release notes) on this release is available at <http://monetdb.cwi.nl/Development/Releases/Oct2010/>. The download location is <http://dev.monetdb.org/downloads/>. -- Sjoerd Mullender

1 0

[Monetdb-developers] updating the default branch.
by Romulo Goncalves 25 Nov '10

25 Nov '10

Hello, I just updated my default branch and I restarted my mserver, i.e., it will use a old dbfarm. I get the following complain: !MALException:user.main[2497]:address of pattern vault.prelude missing I hope you will not ask me to reload my databases :) Romulo

4 11

[Monetdb-developers] Announcement: New Oct2010 feature release of MonetDB suite
by Sjoerd Mullender 15 Nov '10

15 Nov '10

It is not October anymore, but nevertheless, the MonetDB team at CWI/MonetDB BV is pleased to announce the Oct2010 feature release of the MonetDB suite of programs. More information (including release notes) on this release is available at <http://monetdb.cwi.nl/Development/Releases/Oct2010/>. The download location is <http://dev.monetdb.org/downloads/>. -- Sjoerd Mullender

1 0

Re: [Monetdb-developers] MonetDB: Oct2010 - approved output
by Stefan Manegold 30 Oct '10

30 Oct '10

Niels, is it by accedent or by design that with mitosis & mergetable, mat.slice() is not longer used with LIMIT ? If by design, we'd need to change (or even drop) test mat.slice_limit1.Bug-2645, that fails single-threaded since you below checkin; cf., http://monetdb.cwi.nl/testing/projects/monetdb/Candidate/sql/.mTests103/GNU… If by accident, we'd need to investigate whether this is indeed OK ... Stefan On Wed, Sep 15, 2010 at 03:18:54PM +0200, Niels Nes wrote: > Changeset: … [View More]

1 0

[Monetdb-developers] help troubleshoot: cannot connect to MonetDB
by Tripathi, Jaya 25 Oct '10

25 Oct '10

C:\Proximity\bin>gui.bat localhost:30000 0 [main] WARN kdl.prox.util.Util - * log config file not found (using defaults) : C:\Proximity\bin\prox.lcf' Exception in thread "main" Unable to connect: Connection refused: connect at kdl.prox.monet.BlockedModeStream.<init>(BlockedModeStream.java:162) at kdl.prox.monet.Connection.open(Connection.java:401) at kdl.prox.db.DB.open(DB.java:935) at kdl.prox.db.DB.open(DB.java:914) at kdl.prox.app.GUI2App.main(GUI2App.java:238)

1 0

[Monetdb-developers] MonetDB deadlocks on clients, now what?
by Stefan de Konink 19 Oct '10

19 Oct '10

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA512 Lately MonetDB starts to deadlock on the clients. Bugs were reported, typically I have connected ~8 clients connected at the same time. http://bugs.monetdb.org/show_bug.cgi?id=2670 Do other people observe this? Is there a good way to debug where it comes from, or show how it hangs? Stefan -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.16 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ … [View More]

1 0

[Monetdb-developers] test missing on pkg-config in clients configure
by Jan Rittinger 10 Oct '10

10 Oct '10

Hi all, I got the following error message due to a missing pkg-config installation—Perhaps there could be a better error message: /home/rittinge/MonetDB.hg/clients/configure: line 2894: syntax error near unexpected token `monetdbtools,' /home/rittinge/MonetDB.hg/clients/configure: line 2894: `PKG_CHECK_MODULES(monetdbtools, monetdb-stream >= $req_monetdb_ver' -- Jan Rittinger Lehrstuhl Datenbanken und Informationssysteme Wilhelm-Schickard-Institut für Informatik Eberhard-Karls-Universitä… [View More]

2 5

[Monetdb-developers] write error
by Mark Bucciarelli 07 Oct '10

07 Oct '10

Hi, I'm getting a "write error" when running a fairly large and complex query (five tables and subselect in where). I run $ mclient -l sql -d <db> -u <usr> < t.sql password: ... % 6, 26, 1, 103, 229, 5 # length write error $ echo $? 1 I've run the exact query on SQLite, so I'm pretty sure the SQL is good. This is Feb2010. I'm thinking about upgrading, but I'm close to a release so I'd prefer not to. Does this description remind anyone of a commit to … [View More]

3 2