14:11:15 <galderz> #startmeeting
14:11:15 <jbott> Meeting started Tue May 31 14:11:15 2011 UTC.  The chair is galderz. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:11:15 <jbott> Useful Commands: #action #agreed #help #info #idea #link #topic.
14:11:43 <galderz> ok, who should start? vblagoje, wanna go?
14:11:49 <vblagoje> ok galderz
14:12:07 <vblagoje> i think i can safely declare that ISPN-83 is not reproducible
14:12:21 <vblagoje> i have one minor cosmetic fix for it
14:12:24 <vblagoje> nothing major
14:12:45 <vblagoje> michal and I were not able to reproduce errors he has seen and originally reported
14:12:52 <galderz> vblagoje: ah right
14:13:03 <vblagoje> other than that i have started looking at my other task
14:13:11 <galderz> so, we can now remove FLUSH without probs?
14:13:36 <vblagoje> which is ISPN-1118
14:13:56 <vblagoje> yeah i think so galderz
14:14:32 <vblagoje> which is a move of our wiki content to Confluence
14:14:54 <vblagoje> does anyone know anything about that? I pinged MArk but no response yet
14:15:10 <galderz> vblagoje: cool - no, i dunno much about it
14:15:21 <galderz> mark was actually just today in neuchatel office
14:15:30 <galderz> he's just left, but i'd suggest you ping him again
14:15:44 <vblagoje> ah ok, I'll post to jboss core?
14:15:56 <vblagoje> others have to do the same move don't they?
14:16:27 <vblagoje> the confluence move that is
14:16:37 <galderz> i think so
14:16:45 <vblagoje> anyway, that's it from me more or less
14:17:01 <galderz> ok vblagoje, thx
14:17:23 <galderz> i'll go now
14:17:51 <sannegrinovero> #topic galderz
14:17:58 <galderz> so, finally last week sorted out https://issues.jboss.org/browse/JBPAPP-6113 which had been a PITA to track
14:18:13 <galderz> the problem is that we're logging exceptions incorrectly in one place
14:19:16 <galderz> and the problem is that the socket read timeout was leading to transport objects being returned fine
14:19:28 <galderz> so, afterwards data might arrive and other clients would read it
14:19:33 <galderz> anyway, that's now fixed
14:20:05 <vblagoje> cool
14:20:12 <galderz> apart from that a lot of work on smaller jiras including: adaptive marshalling, the klockwork report
14:20:17 <galderz> testsuite failures
14:20:40 <galderz> and Friday I was going through a month's worth of community posts which i had not responded to
14:20:51 <vblagoje> i see a lot of NPEs related to TT, you guys?
14:20:56 <galderz> and some interesting stuff was discovered such as the hazercast stuff
14:21:12 <galderz> vblagoje: what TT?
14:21:27 <vblagoje> TransactionTable sorry
14:22:01 <galderz> vblagoje: i've no idea, you'll have to be more specific please
14:22:12 <sannegrinovero> yes I see some as well in the unit tests, but somehow they don't make the tests fail.
14:22:15 <vblagoje> ok one sec
14:22:39 <galderz> sannegrinovero: vblagoje: it depends on whether they're expected or not :)
14:22:49 <galderz> anyway,
14:23:12 <galderz> also the cloudbees env is looking pretty good, so i'll be documenting that this week and announcing it to the dev list
14:23:13 <sannegrinovero> galderz, well IMO an NPE is a programmer error. should never happen, and get a more meaningfull error.
14:23:39 <galderz> sannegrinovero: it depends where it comes from, but once again, raise it up
14:24:30 <galderz> this week is a short one for me
14:24:45 <galderz> sannegrinovero: dberindei: vblagoje: as said last week - my plan is to release CR4 tomorrow
14:24:52 <galderz> how's stuff looking for you?
14:24:57 <sannegrinovero> ok I'll say it again : the testsuite doesn't look right for me :P. Seriously, I'm saying that I don't pass the tests since a year, I thought this was common knowledge now.
14:25:36 <vblagoje> yeah galderz, see for yourself, these NPEs are definetely not wanted
14:25:39 <sannegrinovero> good for me, I just saw that Query is having two test errors, I'll try fixing that before the release, but it should not be anything serious to block it, in case I can't get to it.
14:26:28 <vblagoje> grep failureCompletingTransaction in test suite
14:27:10 <galderz> vblagoje: raise a JIRA or something, specify which tests...eetc
14:27:11 <galderz> ?
14:27:26 <vblagoje> ok deal
14:28:01 <vblagoje> but we should not go ahead with this CR if we have such a NPE problem that needs to be investigated, wdyt?
14:28:46 <galderz> vblagoje: sure
14:29:06 <galderz> but guys, when you see these things, create jiras asap
14:29:14 <galderz> the sooner we spot them the sooner we can fix them
14:29:26 <vblagoje> yeah makes sense
14:30:29 <vblagoje> I'll create a jira and you verify and send of to Mircea. Looks like his turf
14:30:30 <sannegrinovero> yes, I would to that, but the feedback from my tests doesn't seem to be much valuable since it's consistently broken for me, and it seems not for the other developers. So I don't trust the output here.
14:31:12 <galderz> sannegrinovero: hmmmm, in what way
14:31:34 <galderz> vblagoje: if it's something recent, maybe try to see if any recent jira in that code has had an impact?
14:31:41 <sannegrinovero> galderz, actually I reported this one: https://issues.jboss.org/browse/ISPN-1036
14:32:17 <vblagoje> yeah sanne that's the one!!!
14:33:00 <vblagoje> its been around for a few weeks
14:33:11 <galderz> vblagoje: sannegrinovero: that could well be related to the recovery and transactional changes mircea made
14:33:18 <galderz> though he said there that it was fixed
14:33:29 <galderz> sannegrinovero: you could just reopen the jira and say: it's not fixed :)
14:33:35 <sannegrinovero> ok vblagoje, so it seems it was not resolved. not sure if we should reopen that one or if you want to create a new one
14:33:38 <galderz> then mircea can have a look to it again
14:33:44 <sannegrinovero> deal
14:33:48 <galderz> sannegrinovero: i'm fine with either
14:33:55 <galderz> the point is letting mircea know
14:34:00 <vblagoje> i'll reopen
14:34:04 <galderz> pity he's not here right now
14:34:35 <galderz> normally sannegrinovero, to decide between reopening or not, i use this logic
14:34:48 <sannegrinovero> reopened
14:34:59 <galderz> if the issue has been fixed (some code has been written), and the version has been released already, then create a new jira
14:35:06 <galderz> otherwise, just reopen
14:35:29 <vblagoje> ok I'll create a new one
14:35:43 <vblagoje> it is slightly different actually
14:36:00 <galderz> ok
14:36:46 <sannegrinovero> vblagoje, I had it reopened already
14:37:06 <vblagoje> :-)
14:37:11 <vblagoje> ok what do we do then?
14:39:01 <sannegrinovero> proceed with the meeting
14:39:01 <sannegrinovero> galderz, you're done? shall I go?
14:39:03 <galderz> vblagoje: check the jira and if you think is diff, leave the new one open
14:39:21 <galderz> and let mircea now taking in account the aim to release CR4 tomorrow
14:39:29 <vblagoje> yeah I'll do that. this one is slightly different sanne
14:39:34 <sannegrinovero> sure
14:39:37 <galderz> i'm done myself - btw, public holiday on Thursday/Friday
14:39:54 <galderz> who's next? sannegrinovero? dberindei?
14:40:07 <sannegrinovero> I'll go
14:40:12 <sannegrinovero> #topic sanne
14:40:26 <galderz> welcome trustin
14:40:27 <galderz> :)
14:40:36 <trustin> hey
14:40:39 <trustin> good day
14:40:39 <sannegrinovero> so apart from smaller pull request reviews, I didn't code much
14:40:49 <sannegrinovero> hi trustin , just in time for the second half of the meeting
14:40:52 <galderz> sannegrinovero: is talking about last week's work on his side
14:40:56 <trustin> hehe
14:41:00 <trustin> cool
14:41:10 <galderz> indeed sannegrinovero, thx a million for putting up with my pull req!
14:41:23 <sannegrinovero> but besides coding I've spent again some time thinking about locking
14:41:53 <sannegrinovero> galderz, np, it was time, was lagging behind a bit
14:42:12 <sannegrinovero> and well you have all seen some considerations on the ML, still going on.
14:42:19 <sannegrinovero> An important consideration which came up:
14:42:59 <sannegrinovero> is that all atomic operations (replace, putIfAbsent, etc..) are quite broken when run in a transaction, as the decision ISPN takes to branch in the either/or operation
14:43:18 <sannegrinovero> is affected by the repeatable read effect of the transaction
14:43:45 <sannegrinovero> and this has some consequences, for example I finally identified why ISPN-909 could happen
14:44:16 <sannegrinovero> basically some operations in the Lucene Directory use the atomic operations, and the conclusion right now is that you can not use a TM on the caches being used as index storage
14:45:17 <sannegrinovero> I'll have to spend some more time on the transactions and isolations in Infinispan, and come up with some tests proving this.
14:45:55 <sannegrinovero> So I'm still not sure if we *can* actually postpone lock acquisition to the prepareTX phase. We need to think more about this.
14:46:10 <trustin> brb toilet
14:47:01 <galderz> sannegrinovero: that's pretty important
14:47:16 <sannegrinovero> that's it for me. was PTO yesterday as all UK, and am pushing forward with OGM this week, I'm also preparing a keynote-demo again for the JBug here.
14:47:19 <galderz> what's your plan for this week?
14:47:55 <galderz> ok sannegrinovero, that sounds good
14:48:02 <vblagoje> i am following manik's two week plan summary
14:48:14 <sannegrinovero> as above, and possibly fix the two failing tests in Query
14:48:41 <vblagoje> he outlined b4 disappearing into mountains..
14:48:57 <galderz> yeah vblagoje, i was asking sannegrinovero in particular
14:49:13 <galderz> thx sannegrinovero
14:49:18 <vblagoje> yeah yeah cool
14:49:33 <sannegrinovero> oh, one more thing
14:50:10 <galderz> sounds like steve jobs… :)
14:50:11 <sannegrinovero> this locking stuff is pretty confusing and complex, not really easy to follow on the MLs. so we polished the wiki page a bit, but I'm looking forward to meet about this.
14:50:58 <galderz> sannegrinovero: i agree; there's a lot of things that need to be taken into account
14:51:57 <sannegrinovero> yes, in worst case maybe we can organize something "virtual", or meet just some of us like we did in London.
14:52:25 <sannegrinovero> let's see what sjacobs and manik think about it when manik is back..
14:52:28 <sannegrinovero> dberindei, or trustin your turn?
14:52:44 <trustin> my turn
14:52:51 <sannegrinovero> #topic trustin
14:52:56 <trustin> nothing much actually
14:53:11 <trustin> most time spent to integrate infinispan into JBoss AS 7
14:53:34 <trustin> There were some weird build problems but everything seems to be settled down now
14:54:06 <trustin> there are some issues related with class loading as I posted to the infinispan-dev, which might require the modification in the Infinispan itself
14:55:09 <trustin> What's left is to integrate the REST server, too.  It is somewhat different from the other ones such as Hot Rod server and memcached, but I hope it's simple to do.
14:55:13 <galderz> trustin: for the jboss marshalling stuff, best check with dmlloyd
14:55:35 <galderz> hopefully he's still reading the infinispan dev list :)
14:55:49 <trustin> galderz: yeah, but I guess it's our code that needs a change because we are specifying hard-coded class loader
14:56:14 <trustin> .. which assumes the river marshaller factory is in the same class loader of the marshaller spi
14:56:15 <galderz> trustin: yeah, i'm pretty sure we need a more pluggable way to discover the river marshaller
14:56:50 <trustin> so, I'll do the rest of the integration task this week
14:56:55 <galderz> depdending on whether we're in an OSGI/module env, or traditional J2EE
14:57:02 <galderz> J2EE/java
14:57:05 <trustin> yeah
14:57:23 <trustin> that is all at the moment.
14:57:57 <galderz> cool
14:58:10 <galderz> i think that's about it unless dberindei responds :)
14:58:15 <galderz> let's give him a few mins
15:00:00 <alesj> trustin, pferraro: when is web http session replication mechanism based on Ispan coming into AS7?
15:00:13 <alesj> similar to what we have in AS6
15:00:30 <pferraro> alesj: as soon as I get my tests to pass
15:00:49 <pferraro> alesj: I'm hoping today
15:01:03 <alesj> pferraro: are you using the same api abstractions as for AS6?
15:01:35 <galderz> right, i think that finishes the meeting guys
15:01:36 <pferraro> alesj: no - I've dropped the CacheContainerRegistry abstraction
15:01:51 <galderz> #endmeeting