February 2013 - developers - lists.mariadb.org

Re: [Maria-developers] [Maria-discuss] Files under /etc/my.cnf.d/ read by default
by Honza Horak 05 Feb '13

05 Feb '13

This should have been posted better to maria-developers, probably, so re-sending.. On 02/05/2013 01:55 PM, Honza Horak wrote: > Hi guys, > > while we were packaging MariaDB into Fedora (finally) we used the > upstream's cnf files structure, which means /etc/my.cnf, that includes > all files from directory /etc/my.cnf.d/* using statement !includedir. > > That works fine when user installs (not updates) the MariaDB packages > without having them installed before. The problem is if user has changed > /etc/my.cnf before updating -- then /etc/my.cnf doesn't get updated > (!includedir won't be added) and we end with files under /etc/my.cnf.d/* > but nothing what includes them. That is quite confusing for users that > may change something in /etc/my.cnf.d/client.cnf but nothing happens. > > Since files under /etc/my.cnf.d/* are empty by default it shouldn't do > any harm if all files corresponding with /etc/my.cnf.d/*.cnf template > get loaded even without !includedir in /etc/my.cnf. Actually I believe > that is what users expect to happen when they see files under > /etc/my.cnf.d.. > > So, that's basically what I'm proposing -- read files under > /etc/my.cnf.d/* by default, without a need to use !includedir. What do > you think? > > Regards, > Honza > > _______________________________________________ > Mailing list: https://launchpad.net/~maria-discuss > Post to : maria-discuss(a)lists.launchpad.net > Unsubscribe : https://launchpad.net/~maria-discuss > More help : https://help.launchpad.net/ListHelp

1 0

[Maria-developers] InnoDB Compression
by Gordan Bobic 04 Feb '13

04 Feb '13

Hi, I've just tried testing InnoDB page compression, and I'm sure I must be doing something wrong because the disk space usage in my database directory is the same before and after the compression. To compress the tables I am using: ALTER TABLE $table ENGINE=InnoDB ROW_FORMAT=COMPRESSED KEY_BLOCK_SIZE=$size; with $size I tried being 8KB and 4KB, and in each case "du -h" on the DB directory is the same as before the compression. innodb_file_per_table is set. SHOW CREATE TABLE afterwards shows the ROW_FORMAT and KEY_BLOCK_SIZE to be as per the ALTER TABLE statement. 1) Am I doing it wrong? 2) Is the InnoDB compression feature available in MariaDB? 3) Is the compression on by default transparently? 4) Is there any other explanation for lack of effect? As a cross-check, I tried taking my biggest .ibd file and compressing it with "lzop -1" and that compressed it from 80MB down to 22MB, which doesn't sound right if the data in it was already compressed in any way. Has anybody got any possible explanations? TIA Gordan

2 2

Re: [Maria-developers] [Commits] Rev 3657: MDEV-4127 : Export additional symbols when building RPM
by Sergei Golubchik 04 Feb '13

04 Feb '13

Hi, Vladislav! May be you'd rather rename symbols with #define, just as redhat does? On Feb 03, Vladislav Vaintroub wrote: > At file:///H:/bzr/5.5/ > > ------------------------------------------------------------ > revno: 3657 > revision-id: wlad(a)montyprogram.com-20130203164516-ex2ftmyca5b507g3 > parent: elenst(a)ubuntu11.home-20130202225357-za9bglcf2ngv36xj > committer: Vladislav Vaintroub <wlad(a)montyprogram.com> > branch nick: 5.5 > timestamp: Sun 2013-02-03 17:45:16 +0100 > message: > MDEV-4127 : Export additional symbols when building RPM, > to enable both recompiling mysqli or odbc from sources in addition > to drop-in replacement functionality. > > The case in question is compiling mysqli from sources, that needs > client_errors via ER() macro. > > Previously, we exported it as mysql_client_errors (compatibly to > Fedora's style symbol renaming, see MDEV-3842). > However, if MariaDB header files are used when compiling mysqli, > client_errors needs to be exported with its original name. Regards, Sergei

2 2

[Maria-developers] Coverity scans of MariaDB source
by Christian Convey 04 Feb '13

04 Feb '13

Coverity has a deal called Coverity Scan, where they offer free static analyses of open-source programs (http://scan.coverity.com/) I've used Coverity Prevent at work, and I was pretty happy with the analyses it performed. Is it an intentional choice that MariaDB isn't making use of Coverity Scan? I'm interested in starting to contribute to MariaDB, and I'm wondering if it would be useful to the project for me to pursue this as a first task. - Christian

2 2

[Maria-developers] CMake vs. autotools on Linux?
by Christian Convey 04 Feb '13

04 Feb '13

Hi guys, I noticed that MariaDB is built (indirectly) with CMake on Windows, and with autotools on Linux. Why are both systems used? I would have assumed that because CMake is pretty good at generating Makefiles on Linux and OS X, there would be little reason to continue using autotools. Thanks, Christian

2 3

Re: [Maria-developers] Summary of the MDEV-3882 (.deb upgrade from mysql to mariadb fails due to apt-get considering it a downgrade)
by Kristian Nielsen 02 Feb '13

02 Feb '13

[Moving the discussion to maria-developers@, please use the public mailing lists for discussions like this] Alexey Botchkov <holyfoot(a)askmonty.org> writes: > No matter what dependencies or new packages i created inside MariaDB, > Debian's first idea was to try to install the 'later' version of > libmysqlclient18. At the moment > that was the Oracle's one. Yes, we could change our version's number to be > 'later' than Oracle's, but that would turn same problems on MySQL, and Not really. It's not "Oracle's", it is the native Debian/Ubuntu package. So if users want MariaDB, they can add our repository. If they want the official Debian/Ubuntu package, they can just not add our repository. So it makes sense that when users explicitly add the MariaDB repository, they get the MariaDB version of all the packages automatically. > would help us only temporarely - to the moment Oracle releases even > later' version. We can use epoch to ensure that our version will always be bigger than the official Debian/Ubuntu version of MySQL: http://www.debian.org/doc/debian-policy/ch-controlfields.html#s-f-Version Something like 1:5.5.29-mariadb1~wheezy. This will always be prefered by apt-get over any official Debian package. > I didn't find anything that could affect apt-get and guys on #Debian > didn't recommend anything. I believe it works to specify desired version explicitly, as mentioned in the bug report as workaround: sudo apt-get install mariadb-server libmysqlclient18=5.5.29-mariadb1~wheezy > So for now i'd do about this: > - change nothing in the existing mariadb-produced debian packages. > - recommend to use 'aptitude' installing the mariadb-server on Debian. > So how do you like my proposal? Any other ideas? I think it is better if we use the epoch to make MariaDB upgrades work automatically with both apt-get and aptitude, like they used to. Did you check how Percona handles the similar problem? - Kristian.

2 1