21:44 deekayen: boombatower: 20 mins?! where are the config docs on achieving that?
21:45 boombatower: deekayen, that's what we need to develop
21:45 deekayen: even the dual quad core machine I setup on monday took 37 mins
21:45 boombatower: key is multi process and floating the db in memory
21:45 boombatower: hmm
21:45 boombatower: deekayen, with multi-process?
21:45 deekayen: 8
21:46 boombatower: wt
21:46 deekayen: i had nnewton and killes looking at it... they kind of gave up
21:46 boombatower: deekayen, the one running on t.d.o takes 5 min
21:46 boombatower: hm
21:47 deekayen: that's DamZ's?
21:47 boombatower: believe so
21:48 DamZ: deekayen: that's on MyISAM
21:48 DamZ: deekayen: InnoDB cannot acheive that sort of speed
21:49 DamZ: deekayen: I said a few times already that we don't need to test *every patch* on *every platform*
21:49 boombatower: DamZ, so what do you recommed...just running commits agains all
21:49 DamZ: deekayen: testing every patch on MyISAM and only testing a clean HEAD on the other platforms is vastly sufficient
21:49 boombatower: that could be problematic
21:50 boombatower: issue is we can break core...but no one will care...or bot will just disable itself
21:50 boombatower: as 2.x will do
21:50 DamZ: boombatower: I recommend testing every patch on a very optimized MyISAM environment, and running a clean HEAD once or twice a day (or on every commit)
21:50 DamZ: boombatower: it is *not* critical to break head on PostgreSQL or SQLite for example, but it is critical that we *know* it is broken
21:50 DamZ: (and how far it is broken)
21:51 boombatower: hmm
21:51 boombatower: not sure how to write a workflow around that
21:51 deekayen: currently reenabling broken clients on alternate platforms would be quite unmaintainable
21:51 deekayen: that'd have to be modified
21:52 DamZ: frankly the cost of testing every patch on every platform is *huge*
21:52 DamZ: it's simply wasting resources we don't have
21:52 DamZ: (and killing polar bears in the process)
21:52 boombatower: hmm
21:52 boombatower: we will need to redesign workflow
21:52 boombatower: which will set things back
21:53 deekayen: i thought we were just talking about db platforms, but multiplying db platforms X operating systems is just... um not possible as I see it
21:53 boombatower: i agree
21:54 boombatower: and not very useful
21:54 deekayen: but we should have a linux and a windows at least doing myisam
21:54 deekayen: on every patch
21:54 deekayen: no?
21:54 Druplicon: maybe?
21:54 boombatower: if we do windows we need to rework a few parts of pifr
21:54 boombatower: it uses a number of linux commands
21:54 chrisshattuck has left IRC ()
21:55 deekayen: ok, maybe not windows then
21:55 boombatower: it can be down the road if we want
21:56 boombatower: just not emediately
21:56 deekayen: so, a linux myisam client tests every patch, and the alternate db platforms test checkins and maybe open a critical issue for failures?
21:56 DamZ: deekayen: sounds sane
21:56 DamZ: boombatower: does that sound possible?
21:57 boombatower: sounds great...now I need to write it :)