On 23/08/12 23:21, Brion Vibber wrote: > On Thu, Aug 23, 2012 at 2:02 PM, Evan Priestley > <epriest...@phacility.com>wrote: > >> We solve this in Phabricator by using BEGIN (depth 0) or SAVEPOINT (depth >> 1+) when incrementing the counter, ROLLBACK TO SAVEPOINT (depth 1+) or >> ROLLBACK (depth 0) when decrementing it after a failure, and nothing (depth >> 1) or COMMIT (depth 0) when decrementing it after a success. Our experience >> with transaction stacks has generally been good (no real surprises, doesn't >> feel magical, significantly reduces the complexity of transactional code), >> although we don't support anything but MySQL. > > Oooh, nice! Hadn't come across SAVEPOINT before. > > http://dev.mysql.com/doc/refman/5.0/en/savepoint.html > > -- brion
I proposed this same thing on the mailing list two years ago: http://thread.gmane.org/gmane.science.linguistics.wikipedia.technical/49500 Yes, I completely support changing the transactions to be nested using savepoints. They even are supported (or were) on all our db backends. However, I was told that it "might make us hold a lot of locks for much too long". So with fear to cause magical db overload, nothing was changed. :( _______________________________________________ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
