Wednesday, 2010-07-21

[2010/07/21 00:13:23] @ Log started by gepetto
[2010/07/21 00:13:23] @ jmccune joined channel #puppet-dev
[2010/07/21 00:31:27] @ Quit: jmccune: Ping timeout: 258 seconds
[2010/07/21 01:35:24] @ Quit: tmz: *.net *.split
[2010/07/21 01:36:48] @ tmz joined channel #puppet-dev
[2010/07/21 07:20:41] @ rcrowley joined channel #puppet-dev
[2010/07/21 07:27:25] @ notbrien joined channel #puppet-dev
[2010/07/21 09:38:25] @ jmccune joined channel #puppet-dev
[2010/07/21 09:48:50] @ plathrop-away is now known as plathrop
[2010/07/21 10:10:34] @ lak joined channel #puppet-dev
[2010/07/21 10:24:14] @ Quit: lak: Quit: lak
[2010/07/21 11:20:56] @ lak joined channel #puppet-dev
[2010/07/21 11:22:43] <masterzen> lak: if you or Markus have time, would you mind checking my last comment on #4302?
[2010/07/21 11:22:44] <gepetto> masterzen: lak: #4302 is http://projects.puppetlabs.com/issues/show/4302 "Puppet - Bug #4302: compile performance on 2.6 - Puppet Labs"
[2010/07/21 11:28:58] <lak> masterzen: looks like you nailed it - comment made
[2010/07/21 11:35:11] <masterzen> lak: it's way better with stale? commented, but not yet as fast as 0.25.5 :(
[2010/07/21 11:35:31] <masterzen> lak: 7.3s compared to 5.28s
[2010/07/21 11:35:58] <lak> hrm
[2010/07/21 11:36:22] <lak> i expect it's similar kinds of issues
[2010/07/21 11:36:25] <lak> that's really annoying :/
[2010/07/21 11:36:47] <lak> i was really expecting 2.6 to be a lot faster, but i guess i did some stupid stuff that overwhelmed the faster stuff
[2010/07/21 11:36:48] <masterzen> it might also because I log tons of benchmark output
[2010/07/21 11:37:39] <masterzen> lak: removing my extra logging, I get 6.4s compared to 5.28s
[2010/07/21 11:37:51] <masterzen> lak: so we're now close
[2010/07/21 11:38:13] <lak> we really need to add general instrumentation to the system, somehow
[2010/07/21 11:38:13] <duritong> cool
[2010/07/21 11:38:30] <masterzen> lak: yes, that would be really cool and useful
[2010/07/21 11:38:44] <lak> it's tough to do in a performance-neutral way, though
[2010/07/21 11:39:08] <lak> i haven't looked at dtrace+ruby in a long time, but that's probably the route to go - make sure our code paths work well with dtrace's expectations
[2010/07/21 11:40:47] <masterzen> yes, that would be really a good idea. Introducing dtrace probes in the codebase...
[2010/07/21 11:41:05] <masterzen> I'll have a look to this as a side project this week-end :)
[2010/07/21 11:44:11] <lak> :)
[2010/07/21 11:44:21] <markus> Another thought: incident counting can be useful as a guide and is performance neutral in that the tests don't interfere with what you are trying to measure.
[2010/07/21 11:44:47] <markus> If you expect something to be computed once per run and it's hit 4362 times....
[2010/07/21 12:07:42] <blkperl> markus: anyone coming to oscon GsoC BoF? :)
[2010/07/21 12:07:59] <lak> blkperl: i'm trying to convince someone :)
[2010/07/21 12:08:10] <markus> Not sure. I'm not going to be in PDX this week
[2010/07/21 12:08:18] <blkperl> lak: cool :)
[2010/07/21 12:08:22] <markus> James maybe?
[2010/07/21 12:08:24] <blkperl> markus: ahhh :(
[2010/07/21 12:08:42] <markus> Matt was also mentioned, as he's been working with William as a peer.
[2010/07/21 12:08:50] <blkperl> ya send Matt :)
[2010/07/21 14:03:06] <masterzen> lak: sorry to bother you again, what's the planned ETA for 2.6.1?
[2010/07/21 14:29:15] @ plathrop is now known as plathrop-away
[2010/07/21 14:54:33] @ Quit: lak: Quit: lak
[2010/07/21 15:11:48] @ Quit: londo: Remote host closed the connection
[2010/07/21 15:41:38] @ Quit: notbrien: Quit: notbrien
[2010/07/21 15:43:31] @ lak joined channel #puppet-dev
[2010/07/21 15:44:58] <lak> masterzen: i'm not so sure; that's more of a markus question, i think
[2010/07/21 16:42:54] @ londo joined channel #puppet-dev
[2010/07/21 16:42:56] <jamesturnbull> masterzen: soon?
[2010/07/21 16:43:15] <jamesturnbull> masterzen: I suspect we've got 5 issues or so that we should release plus some queue tickets
[2010/07/21 16:45:21] @ Quit: lak: Quit: lak
[2010/07/21 18:54:47] @ Quit: jmccune: Ping timeout: 264 seconds
[2010/07/21 18:57:23] <rcrowley> In 0.25, there was Puppet::Resource::Reference which was useful with Puppet::Resource::Catalog#create_resource. What is the proper substitute now that 2.6 has removed Puppet::Resource::Reference?
[2010/07/21 19:11:35] <rcrowley> Nevermind, this commit explained it: http://github.com/reductivelabs/puppet/commit/7089446697ad550c22012bc2b5572030727d67e1
[2010/07/21 20:00:30] @ Quit: rcrowley: Quit: rcrowley
[2010/07/21 20:13:30] @ rcrowley joined channel #puppet-dev
[2010/07/21 20:26:37] @ Quit: rcrowley: *.net *.split
[2010/07/21 20:26:38] @ Quit: tmz: *.net *.split
[2010/07/21 20:31:55] @ rcrowley joined channel #puppet-dev
[2010/07/21 20:31:55] @ tmz joined channel #puppet-dev
[2010/07/21 21:24:26] @ Quit: rcrowley: *.net *.split
[2010/07/21 21:24:27] @ Quit: tmz: *.net *.split
[2010/07/21 21:29:48] @ rcrowley joined channel #puppet-dev
[2010/07/21 21:29:48] @ tmz joined channel #puppet-dev
[2010/07/21 23:15:11] <sejo> is there a reason why puppet is now versioned 2.6.0 and not 0.26.0 ?
[2010/07/21 23:42:07] @ Quit: stahnma: Read error: Connection reset by peer
[2010/07/21 23:42:21] @ stahnma joined channel #puppet-dev
[2010/07/21 23:42:21] @ Quit: stahnma: Changing host
[2010/07/21 23:42:21] @ stahnma joined channel #puppet-dev

Generated by irclog2html.py 2.6 by Marius Gedminas - find it at mg.pov.lt!