[00:03:34] <maxandersen> bstansberry: stable syntax AND semantics ? :)
[00:04:21] <bstansberry> that's what subsystems are supposed to provide
[00:04:43] <bstansberry> if they expose something in their domain mgmt api, it's meant to be stable across major versions
[00:04:51] <bstansberry> sorry, within
[00:04:59] <bstansberry> a major release
[00:05:09] <bstansberry> and should be stable longer than that
[00:09:17] * stuartdouglas just wasted 20 minutes trying to get the seam resteasy example running on as7, only to realise that it is broken anyway
[00:11:37] *** maxandersen has quit IRC
[00:13:00] *** maxandersen has joined #jboss-as7
[00:13:01] *** ChanServ sets mode: +v maxandersen
[00:14:58] *** pferraro has quit IRC
[00:17:36] <jbossbot> git [jboss-as] push master 5b36578.. bstansberry at jboss dot com Fill in DESCRIBE handlers for datasources and resource adapters
[00:17:36] <jbossbot> git [jboss-as] push master 5fca53c.. bstansberry at jboss dot com Fix typo
[00:17:37] <jbossbot> git [jboss-as] push master URL: http://github.com/jbossas/jboss-as/compare/7108ef5...5fca53c
[00:22:28] <maxandersen> rawblem: stuartdouglas: just fyi, I just build as7 clean out of the box….nice(!)
[00:22:49] <stuartdouglas> cool
[00:23:02] <stuartdouglas> I just got a CDI based JSF app to run on AS7
[00:24:16] *** bstansberry is now known as bstans_afk
[00:33:04] *** maxandersen has quit IRC
[00:33:32] <stuartdouglas> dmlloyd: Do you have a minute to look at the changes at https://github.com/stuartwdouglas/jboss-as/commit/b3269c86e057e374447604716b3926521bb542e1
[00:33:33] <jbossbot> git [jboss-as] b3269c8.. Stuart Douglas web components
[00:34:00] <stuartdouglas> It gets web components working, and that branch also has CDI based injectors
[00:34:28] <stuartdouglas> one of the changes is that the proxies now implement a DestroyableComponent interface
[00:35:03] <stuartdouglas> which means that the component classes need the ee module to be accessible, which is not ideal
[00:35:29] <dmlloyd> well the thing is that you don't explicitly destroy the proxy really
[00:36:00] <stuartdouglas> I do need to be able to call pre destroy methods on wen components
[00:36:18] <dmlloyd> yeah but that should happen via other means
[00:36:18] <stuartdouglas> and in the SPI all I get is the proxy for the instance to destory
[00:36:29] <dmlloyd> there is not a 1:1 relation between proxy and component instance
[00:36:33] <stuartdouglas> so unless I maintain an identity map
[00:36:42] <stuartdouglas> it uses the interceptor functionality
[00:36:59] <stuartdouglas> so there is an interceptor that is only applied to the destroy method
[00:37:09] <stuartdouglas> that destroys the component (not the proxy)
[00:38:20] *** maxandersen has joined #jboss-as7
[00:38:21] *** ChanServ sets mode: +v maxandersen
[00:38:30] <dmlloyd> I don't see why you have to do it via an invocation on the proxy
[00:38:49] <dmlloyd> bearing in mind that a servlet won't use a proxy anyway
[00:39:02] <dmlloyd> asusuming that's waht you're using it for
[00:39:30] <maxandersen> damn internets...
[00:41:54] *** jbossbot has quit IRC
[00:45:35] *** frainone has quit IRC
[00:47:41] <stuartdouglas> hmm
[00:47:41] <maxandersen> dmlloyd: bstans_afk: Nihility: is AS7 in a state where it makes sense to try deploy apps that works on AS5/6 to it and expect it to somewhat work ? (sans the missing -ds.xml support
[00:47:41] <maxandersen> or is it to early yet?
[00:47:41] <dmlloyd> stuartdouglas: the reason you can't use or have a proxy for non-@ManagedBean servlets is that servlet (and other web component) classes can be final and have final members, which precludes proxies anyway
[00:47:41] * maxandersen wondering how many from the tooling team I should let attack AS7 ;)
[00:47:41] <stuartdouglas> yea, I just realised that
[00:47:41] <stuartdouglas> Ok, I need to change this a little bit
[00:47:41] *** rmaucher has quit IRC
[00:47:41] <dmlloyd> stuartdouglas: if it is a @ManagedBean with no name, it does not change anything, period - if it is a named @ManagedBean then it binds a proxy into JNDI but the proxy is useless as it will create the servlet class as a managed bean (and the @ManagedBean rules apply to that object instance)
[00:47:41] <stuartdouglas> that is the behaviour that we need but, if the servlet is a managed bean
[00:47:41] *** bobmcw has quit IRC
[00:47:49] <maxandersen> dmlloyd: is it expected/intended as7 hard errors on classpath manifest entries that it can't find ? (not sure why it can't find it but thats another issue ;)
[00:47:59] *** jbossbot has joined #jboss-as7
[00:48:00] *** ChanServ sets mode: +v jbossbot
[00:48:26] <dmlloyd> maxandersen, I think that's the current behavior.  Not sure if one can call that "expected" or not though
[00:48:33] <dmlloyd> it's really debatable
[00:48:59] <maxandersen> dmlloyd: yeah since i at least found three projects now where ithappens to me ;)
[00:49:23] <maxandersen> dmlloyd: and sure its "bad" and i think shuold be warned about …but hard errroring is harsh ;)
[00:49:38] <maxandersen> (though I like the option of having as7 being strict)
[00:49:51] <dmlloyd> it's not a question of strictness
[00:50:12] <dmlloyd> it's a question of, how do you have consistent behavior when part of your class path just appears or disappears while things are running
[00:50:31] <dmlloyd> right now if you have a.jar and b.jar which depend on one another, if you remove a.jar, b.jar stops
[00:50:45] <dmlloyd> if you replace a.jar with a new a.jar, b.jar wires up with it and resumes everything
[00:51:24] <maxandersen> dmlloyd: hmm…and thats all tied up via "plain old notsoanal manifest.mf declarations" ? :)
[00:51:46] <dmlloyd> that's just one way to have inter-JAR dependencies
[00:51:56] <dmlloyd> there's also extension-list and friends, which behaves similarly
[00:51:59] <dmlloyd> they're not optional
[00:52:12] <dmlloyd> class-path is tricky because in a JDK setting it is of course optional by nature
[00:52:22] <maxandersen> dmlloyd: sure those are explicit ones…where manifest.mf historically are more lenient (for good and bad)
[00:52:26] <dmlloyd> but we opted to make class-path mean modular dependency, which is better in a lot of ways
[00:52:39] <dmlloyd> but it necessarily means that it's tougher to change
[00:52:58] <dmlloyd> we have to keep things wired up consistently
[00:54:36] <dmlloyd> stuartdouglas: you left an incomplete thought there: <stuartdouglas> that is the behaviour that we need but, if the servlet is a managed bean
[00:54:38] <dmlloyd> ... then what?
[00:55:20] * maxandersen sees many nights having to debug users manifest.mf's in the future...
[00:55:51] <stuartdouglas> no, it was just a badly formed sentence, if the servlet has @ManagedBean then we need to create it from the jndi binding
[00:56:07] <stuartdouglas> or rather, we need an actual managed bean instance from somewhere
[00:56:07] *** jamezp has quit IRC
[00:56:17] <stuartdouglas> does not have to the JNDI binding
[00:56:21] <dmlloyd> for the purpose of supporting @AroundInvoke you mean?
[00:56:44] <dmlloyd> I can see that I suppose
[00:57:02] <stuartdouglas> yea
[00:57:03] <dmlloyd> note that we have to support @PostConstruct/@PreDestroy regardless of whether @ManagedBean is present
[00:57:07] <stuartdouglas> and other interceptors
[00:57:09] <stuartdouglas> I know
[00:57:15] <stuartdouglas> I am changing it now
[00:57:22] <maxandersen> dmlloyd: if I say that classpath resolution from war to ear looks broken in as7 would that be expected/knownbug ?
[00:57:36] <stuartdouglas> maxandersen: In what way?
[00:57:42] <dmlloyd> maxandersen: it depends on how you define "broken" and your expectations :)
[00:57:56] <dmlloyd> we really need a clear doc on how class-path works
[00:58:19] *** jamezp has joined #jboss-as7
[00:58:37] <maxandersen> I have a jboss-seam.jar listed in classpath of my war manifest.mf which is in web-inf/libjboss-seam.jar (which is resolved by Eclipse and previous AS's)
[00:59:20] <maxandersen> this is probably redundant … but unfortunately how eclipse classpaths are defined...
[00:59:55] <stuartdouglas> does give you an error about the class path entry not referencing a valid jar?
[01:00:38] <maxandersen> let me give it to you…two sec (btw. this is richfaces photoalbum app so its rahter complex..but I felt encouraged after deploying hudson to it ;)
[01:01:20] <stuartdouglas> I probably should make it not be an error if the class path entry can't be found
[01:02:09] <maxandersen> http://pastebin.com/jvGef4Yx
[01:02:23] <maxandersen> note: I actually *think* this one should be found...
[01:03:39] <maxandersen> want me to give you the jar to see the mess?
[01:03:55] <maxandersen> (full mess) i mean.
[01:04:06] <stuartdouglas> yea, sure
[01:04:40] *** pferraro has joined #jboss-as7
[01:04:47] *** ChanServ sets mode: +v pferraro
[01:06:37] <maxandersen> okey..uploading…takes a while…big one ;)
[01:20:20] <dmlloyd> we could almost certainly make class-path refs optional *within* an EAR
[01:20:26] <dmlloyd> that's very low risk
[01:20:42] <dmlloyd> it's just top-level class-paths that we'd be more strict about I think
[01:21:10] <stuartdouglas> if it references another deployment then it will cause the deployment to halt till it become available though
[01:21:42] <dmlloyd> right that's what I mean, if we know it's within an EAR then we also know that if something is "missing" it's never going to suddenly appear
[01:21:49] <dmlloyd> so we could just write it off
[01:22:00] <dmlloyd> it's just things that reference other TLDs
[01:22:35] <dmlloyd> stuartdouglas: back on the topic of managed beans though, if you use the org.jboss.as.naming.JndiInjectable#returnInjectedValue() method that should be where our destroy methods would be called
[01:22:56] <dmlloyd> (for cases where the user looks up a MB programmatically, we never call destroy, which is probably fine)
[01:23:05] <Nihility> maxandersen found an issue where if it's empty it fails that we should totally support
[01:23:11] <dmlloyd> yeah agreed
[01:23:27] <dmlloyd> I already pointed stuartdouglas at that issue, which is when I noticed that he's not in JIRA :|
[01:23:31] <stuartdouglas> that should be fixed in that branch I posted earlier
[01:23:38] <Nihility> I agree on missing entries in side the deployment context
[01:24:09] <stuartdouglas> https://github.com/stuartwdouglas/jboss-as/tree/JBAS-8920
[01:24:11] <jbossbot> jira [JBAS-8920] Deployment fails with empty classpath manifest [Open (Unresolved) Bug, Major, David Lloyd] https://issues.jboss.org/browse/JBAS-8920
[01:24:15] * maxandersen still wonders if it would the right time to start pushing larger parts of the tooling team on as7 with existing apps to test….or if its still too early?
[01:24:18] <stuartdouglas> may need to be rebased now
[01:24:32] <dmlloyd> maxandersen: testing never hurt anyway
[01:24:34] <dmlloyd> anyone*
[01:24:48] <dmlloyd> that's more useful to us than having them sit on their thumbs
[01:24:51] <Nihility> maxandersen: It might be a waste of your time until after beta1 at
[01:24:53] <Nihility> Least
[01:24:55] *** rawblem has quit IRC
[01:25:08] <maxandersen> dmlloyd: sure - but also dont want to DOS you with issues ;)
[01:25:18] <Nihility> as an example we don't have ejbs fully working yet
[01:25:21] <maxandersen> Nihility: when is that?
[01:25:30] <dmlloyd> week or two
[01:25:33] <dmlloyd> or three :)
[01:25:39] <maxandersen> dmlloyd: :) so no alpha2?
[01:26:03] <Nihility> it's supposed to be the 15th but we are cutting it very close
[01:26:03] <maxandersen> stuartdouglas: sorry - my upload speed is for some reason horrible tonight
[01:26:18] <stuartdouglas> once I finish my web component changes pretty much all non-ejb non-jpa CDI apps should work
[01:26:46] <Nihility> maxandersen: But yeah I think we should get a list of yur issues together
[01:26:57] <stuartdouglas> dmlloyd: The issue I have is getting hold of the jndiInjectable from the managed bean proxy instance
[01:26:59] <Nihility> And then go through them assign priorities etc
[01:27:31] <stuartdouglas> for web components were there is no proxy, I can sort of do a map of Class -> methods to call
[01:27:31] <Nihility> after beta1 though, in the meantime you can create jiras for what you find or start posts
[01:27:56] <stuartdouglas> but destroying managed beans is more complex, as you also need to call lifecycle methods
[01:28:01] <maxandersen> Nihility: k….jiras would be easiest for me…ok for you ?
[01:28:08] <stuartdouglas> which was where my DestroyableComponent came into it
[01:28:14] <Nihility> Yeah that's perfect
[01:28:30] <maxandersen> ok ;)
[01:28:41] * maxandersen creates multicolor log output jira ;)
[01:28:46] <maxandersen> jk
[01:38:50] *** pferraro has quit IRC
[01:43:44] *** asaldhan1 has left #jboss-as7
[01:49:16] <maxandersen> dmlloyd: i recall seeing a doc/wiki about as7 logging stuff…google aint helping me. got a pointer?
[01:51:22] *** stalep has quit IRC
[01:51:41] <dmlloyd> your memory deceives you
[01:51:51] <dmlloyd> there are no AS7 logging docs that I'm aware of
[01:52:24] <dmlloyd> stuartdouglas: you should never need to get ahold of the JI from the proxy though
[01:52:31] <dmlloyd> only the reverse
[01:52:32] <maxandersen> dmlloyd: ok ;(
[01:52:39] <dmlloyd> whoever holds the JI is in charge of the lifecycle
[01:55:07] <maxandersen> stuartdouglas: i just realized what was wrong with my war deployment….let me check if I just shot my self in the foot..
[01:56:00] * dmlloyd afk
[01:56:02] <stuartdouglas> dmlloyd: but the tomcat InstanceManager api just has a destory(instance) method
[01:56:23] <stuartdouglas> I need to associate that instance with a ComponentInstance (or whatever) somehoe
[01:56:33] <dmlloyd> you need only to pass it back into JI
[01:56:40] * dmlloyd really afk now (story time)
[01:58:37] <stuartdouglas> dmlloyd: Great in theory, but returnInjectedValue is not implemented :-)
[02:04:05] <maxandersen> stuartdouglas: do you know how to enable/disable logging in as7 ? :)
[02:04:09] <maxandersen> as7 is using 80% cpu on deploying my photoalbum.ear ;) …..
[02:04:15] <maxandersen> for 4 minutes now
[02:04:45] <stuartdouglas> you can configure it in standalone.xml
[02:04:52] <maxandersen> stuartdouglas: dmlloyd: btw. the reason for the manitfest error earler was that I mistakenly had deployed the war "oustide" the ear.
[02:05:07] <maxandersen> stuartdouglas: so i requires a restart?
[02:05:36] *** jamezp1 has joined #jboss-as7
[02:06:05] *** jamezp1 has left #jboss-as7
[02:06:30] <stuartdouglas> I think so, I am not really an expert on this, there may be some way to change it through the deployment api
[02:06:50] <maxandersen> k
[02:07:06] *** jamezp has quit IRC
[02:07:20] * maxandersen thinks he triggered a while(true) loop in as7...
[02:07:30] *** stuartdouglas_ has quit IRC
[02:07:38] <stuartdouglas> hmm
[02:08:01] <stuartdouglas> did you upload the ear?
[02:08:18] *** mbg|away has quit IRC
[02:08:19] <stuartdouglas> for me to check out
[02:08:21] *** stuartdouglas_ has joined #jboss-as7
[02:08:57] *** mbg has joined #jboss-as7
[02:08:57] *** ChanServ sets mode: +v mbg
[02:09:13] <maxandersen> stuartdouglas: well i was uplading the war until i realized that bug ;) let me start ear upload instead.
[02:12:07] <maxandersen> stuartdouglas: in about 20 minutes http://dl.dropbox.com/u/558690/photoalbum-ear.ear should contain the faulty ear.
[02:12:19] <maxandersen> i might have fallen a sleep before that.
[02:13:06] <stuartdouglas> ok
[02:13:25] <stuartdouglas> incidentally have you ever used yjp?
[02:13:28] <maxandersen> this ear deploys out of box on EAP 5 (AS 5.1)
[02:13:36] <maxandersen> yjp ..the profiler ?
[02:13:40] <stuartdouglas> It has a really neat feature where it can just attach to a profile a running vm
[02:13:51] <Nihility> whew
[02:13:54] <stuartdouglas> very useful for finding infinite loops
[02:13:55] <Nihility> just got little one to bed
[02:14:30] <maxandersen> stuartdouglas: ah like visualvm kinda thinkg ? i guess I could try connect that to the as 7 that is now burning 90% cpu the last 10 minutes
[02:14:30] <Nihility> dmlloyd: what was the shining example of a good Description generator?
[02:14:56] <dmlloyd> stuartdouglas: yeah don't worry about that little detail for now :)
[02:15:01] <Nihility> ROFL
[02:15:29] <dmlloyd> Nihility: @ManagedBean's scanner for things which create components, and @Resource for things which inject into them\
[02:15:53] <maxandersen> daamn as7 is trashing memory ;)
[02:15:53] * dmlloyd is out for the night, will check in at various times tomorrow tho
[02:15:57] <stuartdouglas> dmlloyd it just moves the problem from the InstanceManager to IndiInjectable, at some point we need to go from ProxyInstance -> ComponentInstance
[02:16:00] <maxandersen> fun to watch
[02:16:31] <dmlloyd> stuartdouglas: yeah I'm like 70% I have that figured, we'll have to discuss later though
[02:16:52] <Nihility> maxandersen: what are you testing?
[02:16:56] <stuartdouglas> ok, well I will ignore it for now
[02:17:06] <maxandersen> Nihility: just deploying http://dl.dropbox.com/u/558690/photoalbum-ear.ear (might not be up yet)
[02:17:18] <maxandersen> its a 23 mb app that deploys fine on EAP 5
[02:17:27] <Nihility> is it a seam app?
[02:17:41] <maxandersen> Nihility: yes
[02:18:18] <maxandersen> stuartdouglas: hehe - attaching visualvm to as7 is not heathy..now getting classcircularityerrors in the as log about netbeans jars ;)
[02:18:36] <Nihility> try jconsole
[02:19:07] <stuartdouglas> yea, you need to specify the system class thing in jboss modules
[02:19:38] <Nihility> circulatory is interesting though
[02:19:44] <maxandersen> stuartdouglas: system class thing ?
[02:20:10] <stuartdouglas> for yjp it is -Djboss.modules.system.pkgs=com.yourkit
[02:20:38] <stuartdouglas> you probably need to do that for the visualvm classes
[02:20:57] <maxandersen> Nihility: jconsole works fine (boring ;)
[02:21:05] *** torben has quit IRC
[02:21:39] <maxandersen> stuartdouglas: ah got it….i guess it would be nice some common profiler classes would be put on system.pckgs bydefault to allow easy profiling...?
[02:22:04] <stuartdouglas> I just have that stuff in my JAVA_OPTS in bash_profile
[02:22:35] <Nihility> maxandersen: thats not a bad idea
[02:22:47] <Nihility> maxandersen: btw jprofiler already fixed their stuff to not require it
[02:23:27] <stuartdouglas> yea, you have to instrument the ClassLoaders as well so that they can always find your classes
[02:23:38] <maxandersen> ok ctrl+c on as7 now…just sitting there looking lazy...
[02:24:45] <maxandersen> yeah..I've killed as7….
[02:26:41] <Nihility> maxandersen: could you get a thread dump
[02:26:44] <Nihility> when its going nut
[02:26:45] <Nihility> s
[02:27:06] <maxandersen> jstack enough for ya ?
[02:27:16] <Nihility> yes that will do
[02:28:51] <maxandersen> Nihility: i also broke the filedeployment in this go… ;)
[02:29:10] <maxandersen> how do i undeploy something from as7 that is in standalone.xml but not in thedeployment flder anymore ?
[02:29:20] <maxandersen> only way edit standalone.xml?
[02:29:25] <maxandersen> only/simplest..
[02:29:56] <Nihility> yeah that would be the simplest
[02:33:57] <maxandersen> Nihility:  JBAS-8926
[02:33:59] <jbossbot> jira [JBAS-8926] AS7 goes in infinite loop/freezes when deploying photoalbum.ear (richfaces/jsf2/seam app) [Open (Unresolved) Bug, Major, Unassigned] https://issues.jboss.org/browse/JBAS-8926
[02:41:33] <maxandersen> Nihility: stuartdouglas: https://issues.jboss.org/browse/JBAS-8927
[02:41:35] <jbossbot> jira [JBAS-8927] jvisualvm on as7 with profile enabled results in ClassCircularityErrors - way to preconfigure AS7 allow it ? [Open (Unresolved) Enhancement, Major, Unassigned] https://issues.jboss.org/browse/JBAS-8927
[02:41:56] <maxandersen> and btw. http://dl.dropbox.com/u/558690/photoalbum-ear.ear is now live.
[02:42:07] * maxandersen need to sleep - see you next week.
[02:42:27] * stuartdouglas is downloading photoalbum now
[02:58:46] *** jbossbot has quit IRC
[03:06:28] *** stuartdouglas_ has quit IRC
[03:06:28] *** mbg has quit IRC
[03:06:28] *** emuckenhuber has quit IRC
[03:06:28] *** bstans_afk has quit IRC
[03:06:28] *** pgier has quit IRC
[03:06:29] *** stuartdouglas has quit IRC
[03:06:29] *** Nihility has quit IRC
[03:06:29] *** dmlloyd has quit IRC
[03:06:29] *** ChanServ has quit IRC
[03:06:30] *** jfclere has quit IRC
[03:06:30] *** smarlow has quit IRC
[03:06:30] *** nickarls has quit IRC
[03:06:31] *** maxandersen has quit IRC
[03:07:18] *** mbg has joined #jboss-as7
[03:07:18] *** stuartdouglas_ has joined #jboss-as7
[03:07:18] *** maxandersen has joined #jboss-as7
[03:07:18] *** stuartdouglas has joined #jboss-as7
[03:07:18] *** dmlloyd has joined #jboss-as7
[03:07:18] *** gibson.freenode.net sets mode: +vvvv mbg maxandersen stuartdouglas dmlloyd
[03:07:18] *** Nihility has joined #jboss-as7
[03:07:18] *** emuckenhuber has joined #jboss-as7
[03:07:18] *** jfclere has joined #jboss-as7
[03:07:18] *** bstans_afk has joined #jboss-as7
[03:07:18] *** pgier has joined #jboss-as7
[03:07:18] *** gibson.freenode.net sets mode: +vvvv Nihility emuckenhuber bstans_afk pgier
[03:07:18] *** smarlow has joined #jboss-as7
[03:07:18] *** nickarls has joined #jboss-as7
[03:07:18] *** ChanServ has joined #jboss-as7
[03:07:18] *** gibson.freenode.net sets mode: +vo smarlow ChanServ
[03:11:39] <Nihility> looks like a bug in jboss metadata
[03:11:40] <Nihility> hahaha
[03:11:42] <Nihility> tight loop
[03:13:17] <Nihility> org.jboss.metadata.parser.jboss.JBossAppMetaDataParser.parseLoaderRepository
[03:13:32] <stuartdouglas> I just started looking at that as well
[03:14:33] <Nihility> its missing an advance call
[03:14:38] <Nihility> thats why
[03:19:09] <Nihility> stuartdouglas: so i was going to start looking into filling out missing general EE annotations
[03:23:52] <stuartdouglas> which ones
[03:24:20] <Nihility> llike the @DataSourceDefition thing
[03:24:25] <Nihility> maybe @EJB
[03:24:37] <Nihility> just wanted to make sure im not overlapping with you
[03:25:00] <stuartdouglas> ok, I am playing around with the component stuff a bit, but it should not overlap
[03:25:07] <stuartdouglas> I am still looking at web components
[03:25:16] <stuartdouglas> but not that much cause it is Saturday :-)
[03:25:35] <Nihility> haha
[03:25:44] <Nihility> im going to work through the weekend
[03:27:03] <Nihility> so ill be around if you want to discuss anything
[03:27:11] <stuartdouglas> cool
[03:28:13] <stuartdouglas> so once I get web components sorted out what is next with regard to EJB's ?
[03:28:49] <Nihility> we need to have component descriptions for everything
[03:28:59] <Nihility> and then annotation processing to produce them
[03:29:22] <Nihility> dmlloyd and i were talking about doing annotations first
[03:29:30] <Nihility> then doing xml after beta1
[03:29:47] <stuartdouglas> that makes sense
[03:30:03] <Nihility> the ejb guys are working on descriptions to actually working ejb2
[03:30:07] <Nihility> ejbs
[03:30:18] <Nihility> but i bet they will need help
[03:30:33] <Nihility> Jaikiran has been sick
[03:31:22] <stuartdouglas> once the web component stuff is done I don't really have anything lined up
[03:31:32] <stuartdouglas> so I could help out there if needed
[03:32:53] <stuartdouglas> Is JPA going to be done for Beta1?
[03:36:45] <Nihility> yeah it needs to be
[03:36:57] <Nihility> smarlow is working on it
[03:37:12] <Nihility> but he might also need help
[03:56:16] *** pgier has quit IRC
[04:03:21] <smarlow> Nihility, stuartdouglas:  its getting closer, should probably merge in soon :)
[04:04:25] <smarlow> stuartdouglas:  is a merge pass of your web support coming in soon or will that wait until its all done?
[04:05:03] *** bobmcw has joined #jboss-as7
[04:05:04] *** ChanServ sets mode: +v bobmcw
[04:11:06] <stuartdouglas> smarlow: my web support should be ready to merge tomorrow
[04:12:01] <smarlow> stuartdouglas:  great, maybe I could hack with a managed bean instead but I'm hoping to see how far I can get with injecting a PersistenceContext into a servlet
[04:26:30] *** pferraro has joined #jboss-as7
[04:26:30] *** ChanServ sets mode: +v pferraro
[04:36:54] <Nihility> stuartdouglas: did your patch get merged
[04:37:37] <stuartdouglas> Nihility: The class-path one?
[04:37:47] <Nihility> no the web component one
[04:37:57] <stuartdouglas> no, I had to change it a bit
[04:38:09] <stuartdouglas> It should be ready to merge tomorrow
[04:38:14] <Nihility> ah ok
[04:38:32] <stuartdouglas> It was sort of treating all web components as managed beans
[04:38:45] <stuartdouglas> which does not work because they might not be proxiable
[04:39:10] <Nihility> ah yes
[04:39:14] <Nihility> final methods
[04:42:42] <stuartdouglas> I do have this one that needs to be merged: https://github.com/stuartwdouglas/jboss-as/tree/JBAS-8920
[04:47:41] <Nihility> oh i just noticed jbossbot is gone
[04:47:49] <Nihility> dmlloyd: not sure if you are aware
[04:48:16] <Nihility> so how did you do the web stuff
[04:48:24] <Nihility> with the mixture of annotation processing in metadata
[04:48:29] <Nihility> and in container
[04:49:57] <Nihility> stuartdouglas: its pushed
[05:17:56] *** mbg has quit IRC
[05:44:10] *** mbg has joined #jboss-as7
[05:44:11] *** ChanServ sets mode: +v mbg
[05:54:23] *** bstans_afk has quit IRC
[05:59:03] <stuartdouglas> Nihility: I pushed back the component parsing to post module, and then just pulled the metadata out of WarMetaData
[06:24:01] *** pferraro has quit IRC
[06:26:04] *** smarlow has quit IRC
[07:25:46] *** mbg is now known as mbg|away
[07:41:51] *** mbg|away is now known as mbg
[07:58:46] *** rawblem has joined #jboss-as7
[07:58:46] *** rawblem has joined #jboss-as7
[07:58:46] *** ChanServ sets mode: +v rawblem
[08:28:25] *** rawblem has quit IRC
[08:41:47] *** rawblem has joined #jboss-as7
[08:41:47] *** rawblem has joined #jboss-as7
[08:41:47] *** ChanServ sets mode: +v rawblem
[08:50:39] *** stuartdouglas has quit IRC
[08:50:40] *** stuartdouglas_ has quit IRC
[08:51:11] *** stuartdouglas has joined #jboss-as7
[08:51:12] *** ChanServ sets mode: +v stuartdouglas
[08:53:43] *** stuartdouglas_ has joined #jboss-as7
[08:53:44] *** stuartdouglas_ has joined #jboss-as7
[08:53:44] *** ChanServ sets mode: +v stuartdouglas_
[08:54:47] *** stuartdouglas__ has joined #jboss-as7
[08:54:47] *** stuartdouglas_ has quit IRC
[08:54:49] *** stuartdouglas__ has quit IRC
[08:54:49] *** stuartdouglas__ has joined #jboss-as7
[08:54:49] *** ChanServ sets mode: +v stuartdouglas__
[08:56:01] *** stuartdouglas has quit IRC
[08:56:01] *** stuartdouglas__ is now known as stuartdouglas
[08:57:45] *** rawblem has quit IRC
[09:11:18] *** rawblem has joined #jboss-as7
[09:11:19] *** rawblem has joined #jboss-as7
[09:11:19] *** ChanServ sets mode: +v rawblem
[10:51:36] *** maxandersen has quit IRC
[11:18:18] *** torben has joined #jboss-as7
[11:19:56] *** torben has quit IRC
[11:19:56] *** torben has joined #jboss-as7
[11:39:22] *** rawblem has quit IRC
[11:56:24] *** frainone has joined #jboss-as7
[11:56:24] *** ChanServ sets mode: +v frainone
[12:14:03] *** maxandersen has joined #jboss-as7
[12:14:04] *** ChanServ sets mode: +v maxandersen
[12:38:41] <maxandersen> emuckenhuber: ping
[12:38:54] <emuckenhuber> maxandersen: pong
[12:39:19] <maxandersen> trying to build/run AS from eclipse - which class should I run ?
[12:42:25] <maxandersen> emuckenhuber: ^
[12:43:03] <emuckenhuber> hmm, how do you want to run AS from eclipse? you can look at standalone.sh
[12:43:37] <emuckenhuber> you basically tell jboss-modules which Main / module to run
[12:44:17] <maxandersen> you dont run as from eclipse to do debugging  ? or you maybe just launch it directly and attach debugger or ?
[12:45:02] <emuckenhuber> i always use this remote debugger thingy - there is a property in standalone.conf
[12:45:18] <maxandersen> well i probably need to figure out why m2eclipse won't build any of the projects ….seems it cant find any of the source dependencies
[12:47:29] <maxandersen> have you seen/fixed something like this from m2eclipse/as build ?
[12:47:50] <maxandersen> 3/5/11 12:39:02 PM CET: Build errors for jboss-as-messaging-7.0.0.Alpha2-SNAPSHOT; org.apache.maven.lifecycle.LifecycleExecutionException: Failed to execute goal on project jboss-as-messaging: Could not resolve dependencies for project org.jboss.as:jboss-as-messaging:jar:7.0.0.Alpha2-SNAPSHOT: Failed to collect dependencies for [org.jboss.msc:jboss-msc:jar:1.0.0.Beta6 (compile), org.jboss.as:jboss-as-server:jar:7.0.0.Alpha2-SNAPSHOT (compile
[12:47:50] <maxandersen> emuckenhuber: ^
[12:49:42] <emuckenhuber> hmm it works for me - i can remember seeing some LifecycleExecutionExceptions, but they did not really affect anything
[12:55:04] *** maxandersen has quit IRC
[12:59:40] *** maxandersen has joined #jboss-as7
[12:59:41] *** ChanServ sets mode: +v maxandersen
[13:00:27] *** emuckenhuber has quit IRC
[13:03:54] *** emuckenhuber has joined #jboss-as7
[13:03:54] *** ChanServ sets mode: +v emuckenhuber
[13:05:05] <maxandersen> emuckenhuber: might not help i'm using mvn on a teethered Edge connection...
[13:26:53] <maxandersen> emuckenhuber: im trying to use the cli … kinda hard ;) anywhere to get examples of how/what operations that can give info ?
[13:28:45] <emuckenhuber> maxandersen: well there are some global operations like: "read-resource", "read-resource-description", "read-operation-names"
[13:29:24] <maxandersen> so /connect, read-resource should work ?
[13:29:44] <maxandersen> (which it doesn't for me ;)
[13:29:45] <emuckenhuber> maxandersen: yes so /connect and then ":read-resource"
[13:29:58] <maxandersen> ah :
[13:30:39] <emuckenhuber> maxandersen: so you can also do "subsystem=web:read-resource"
[13:30:48] <emuckenhuber> or just do :read-operation-names
[13:31:03] <emuckenhuber> then you get a list of available operations
[13:31:21] <emuckenhuber> and ":" basically executes the operation on the current node
[13:31:42] <maxandersen> damn I really have to twist my fingers to remember put / in front of connect/help etc. ;)
[13:32:29] <maxandersen> …and doesnt help I crash the cli everytime I tab into most things….if I could just get the darn eclipse to compile so I could fix these issues while I find them ;)
[13:33:19] <emuckenhuber> hehe, well you can try to use IDEA ;)
[13:33:52] <maxandersen> where would one read about server status kind of things ? i.e. memory, threads, servergroup count etc. ?
[13:37:23] <emuckenhuber> hmm, yeah we did not really add things like that yet
[13:38:00] <maxandersen> mkay
[13:39:00] <maxandersen> emuckenhuber: so its all only just config data ?
[13:39:44] <emuckenhuber> yes
[13:41:45] <emuckenhuber> well we a few metrics on the web connector - so you could e.g. do subsystem=web,connector=http:read-attribute(name=bytesReceived)
[13:45:05] * maxandersen rebuilding as7 so cli won't throw me out just because i code complete ;)
[13:45:35] <maxandersen> emuckenhuber: about metrics, ok…btw. are you just remembering these strings by heart or are they "readable" somewhere ?
[13:45:51] *** smarlow has joined #jboss-as7
[13:45:52] *** ChanServ sets mode: +v smarlow
[13:46:45] <emuckenhuber> maxandersen: they should be part of the resource-description
[13:47:08] <emuckenhuber> and it should be possible to include them doing: subsystem=web,connector=http:read-resource(include-runtime=true)
[13:47:17] <emuckenhuber> but that does not seem to work for whatever reason
[13:50:56] <maxandersen> mkay :)
[13:52:10] <smarlow> dmlloyd, stuartdouglas:  could someone answer the following question about https://github.com/scottmarlow/jboss-as/blob/master/jpa/src/main/java/org/jboss/as/jpa/processor/JPAAnnotationParseProcessor.java that I based on ResourceInjectionAnnotationParsingProcessor.
[13:56:47] <smarlow> dmlloyd, stuartdouglas:  Question:  I would like to bind an ObjectFactory to be used for each of the binding steps.  I see that we have a few different flavors of BindingSourceDescription but since I cannot step into the above module yet, I'm kind of flying blind on learning/hacking the ObjectFactory (or maybe service?) into the above.
[13:59:28] <smarlow> dmlloyd, stuartdouglas:  My ObjectFactory (or equivalent), will want to support injecting into hibernate session/sessionFactory as well.  So whatever the custom solution is, I'll want to pass in parameters that I have in hand (thank you for getting all of parameters that I need into my hands with the recent EE effort :-).
[14:02:51] * smarlow hopefully, I won't be as blind, once I can run some of my test cases :-)
[14:50:12] <maxandersen> emuckenhuber: if im getting these errors in an neverending stream when running domain.sh what did I do wrong ?
[14:50:13] <maxandersen> 4:48:31,261 INFO  [org.jboss.as.process.Host Controller.status] (reaper for Host Controller) Process 'Host Controller' finished with an exit status of 1
[14:50:14] <maxandersen> 14:48:31,261 INFO  [org.jboss.as.process.Host Controller.status] (reaper for Host Controller) Starting process 'Host Controller'
[14:50:48] <maxandersen> ah….dont pass parameters to domain.sh  :)
[14:50:51] <maxandersen> then that happens
[14:51:56] <emuckenhuber> hmm, yeah that happens when the HC fails to start we try to restart
[14:52:16] <maxandersen> what I did was ./domain.sh —help
[14:52:23] <maxandersen> ./domain.sh works.
[14:54:02] <maxandersen> bummer…cli tab completion doesn't work when running against domain.sh started server ;(
[15:02:26] *** mbg has joined #jboss-as7
[15:02:26] *** ChanServ sets mode: +v mbg
[15:05:05] <maxandersen> emuckenhuber: is http access described on http://community.jboss.org/wiki/AS7DetypedManagementAPI available ? im just getting blank pages when trying to access them?
[15:05:57] <emuckenhuber> maxandersen: http://localhost:9990/domain-api/?json.pretty=true&recursive=true
[15:07:13] <maxandersen> ah gotcha
[15:07:29] <maxandersen> hmm - does your browser also want to save the result instead of just showing it raw ?
[15:07:32] <maxandersen> emuckenhuber: ^
[15:08:01] <emuckenhuber> hmm, no that works fine for me
[15:09:00] <maxandersen> weird..
[15:10:50] <maxandersen> emuckenhuber: well curl works - and thats actually much simpler than CLI ;)
[15:13:04] <emuckenhuber> hehe, true - i'm still waiting on being able to deploy using curl though :)
[15:14:18] <maxandersen> how do I reach info about  a server via this api? cant find tehm ;)
[15:14:55] *** bstansberry has joined #jboss-as7
[15:14:56] *** ChanServ sets mode: +v bstansberry
[15:15:07] <maxandersen> emuckenhuber: ^
[15:15:28] <emuckenhuber> in the domain mode - when you do a recursive query, there should be a section called hosts
[15:15:29] *** maxandersen has quit IRC
[15:21:20] *** mbg is now known as mbg|away
[15:22:10] *** maxandersen has joined #jboss-as7
[15:22:10] *** ChanServ sets mode: +v maxandersen
[15:22:25] <maxandersen> emuckenhuber: damn internets!
[15:22:50] <maxandersen> emuckenhuber: is the http url scheme described somewhere ?i.e. how address attributes/operations etc?
[15:22:54] <emuckenhuber> maxandersen: so when you run in the domain mode - there should be a "hosts" section, which contains the running servers
[15:23:19] <emuckenhuber> maxandersen: ?operation=attribute&name=attributeName
[15:23:30] <emuckenhuber> i think...
[15:26:16] <maxandersen> host:read-resource (in cli) doesnt do anything but dump root content again..
[15:27:22] <emuckenhuber>  host=undefined:read-resource
[15:28:32] <maxandersen> eh…okey…i thought "undefined" meant "not there"
[15:28:46] <maxandersen> didn't seem to need that when using standalone?
[15:29:56] <emuckenhuber> well the name "undefined" might be wrong... details..
[15:30:04] <maxandersen> emuckenhuber: ^ …same for http access…
[15:30:22] <emuckenhuber> yes, that's how the addressing works you always need key=value
[15:30:28] <maxandersen> emuckenhuber: just wondering what that undefined means in this context?
[15:30:43] <emuckenhuber> the host name "undefined" is most likely wrong
[15:31:30] <emuckenhuber> or not defined
[15:33:31] <maxandersen> but could there be other names below host ? I would assume it would be host/hostname/ instead of host/undefined/hostname as it seem to be now...
[15:34:08] <maxandersen> host/undefined/server/sername actually..
[15:35:00] <emuckenhuber> maxandersen: you can define a hostname llike: http://pastebin.com/Dnfvmp4T
[15:35:12] <emuckenhuber> it should be be host/hostName/server/serverName
[15:35:44] <maxandersen> ah yes sorry…forget that extra level of indirection.
[15:35:48] <maxandersen> forgot i mean.
[15:36:11] <maxandersen> ok - so the domain controller xml just dont have one for the default host...
[15:36:18] <maxandersen> or rather preconfigured host I assume
[15:46:52] <maxandersen> emuckenhuber: what does "null" mean in the json result for i.e. system-properties/management ? just that there is no attributes/children below it ? only operations ?
[15:50:44] <emuckenhuber> "null" usually means it has no value
[15:55:15] <maxandersen> k
[16:03:39] *** pferraro has joined #jboss-as7
[16:03:40] *** ChanServ sets mode: +v pferraro
[16:10:45] *** bobmcw has quit IRC
[16:19:57] <maxandersen> emuckenhuber: read-resource list all attributes, right ? or is it listing "children" ?
[16:20:53] <emuckenhuber> maxandersen: read-resource includes all configuration attributes yes
[16:21:18] <maxandersen> emuckenhuber: so when I have this:
[16:21:19] <maxandersen> [~] subsystem=deployment-scanner,scanner=default:read-resource
[16:21:19] <maxandersen> {
[16:21:19] <maxandersen>     "outcome" => "success",
[16:21:19] <maxandersen>     "result" => {
[16:21:19] <maxandersen>         "path" => "deployments",
[16:21:19] <maxandersen>         "scan-enabled" => true,
[16:21:19] <maxandersen>         "scan-interval" => 5000,
[16:21:20] <maxandersen>         "relative-to" => "jboss.server.base.dir"
[16:21:20] <maxandersen>     },
[16:21:21] <maxandersen>     "compensating-operation" => undefined
[16:21:21] <maxandersen> }
[16:21:39] <maxandersen> why does this fail:
[16:21:41] <maxandersen> ~] subsystem=deployment-scanner,scanner=default:read-attribute(name="path")
[16:21:41] <maxandersen> {
[16:21:41] <maxandersen>     "outcome" => "failed",
[16:21:41] <maxandersen>     "failure-description" => "No known attribute called path"
[16:21:41] <maxandersen> }
[16:22:04] <maxandersen> read-resource says there is an attribute named path…but read-attribute says there aren't (at least thats how I read it)
[16:24:08] <maxandersen> should I pastebin it instead?
[16:24:12] <emuckenhuber> maxandersen: looks like a bug
[16:24:48] <maxandersen> should I jira it instead ? :) ….I guess this could also just be a bug in the cli..?
[16:26:07] <emuckenhuber> if i use the http API i see a NPE, so it does not seem to be specific to the cli
[16:28:31] *** bobmcw has joined #jboss-as7
[16:28:32] *** ChanServ sets mode: +v bobmcw
[16:28:49] <maxandersen> emuckenhuber: ok…so I jira it?
[16:29:19] <emuckenhuber> maxandersen: hmm, yeah
[16:31:44] <maxandersen> emuckenhuber: whats the url you used from http?
[16:32:27] <emuckenhuber> http://localhost:9990/domain-api/subsystem/deployment-scanner/scanner/default?operation=attribute&name=path
[16:33:59] <maxandersen> brb
[16:38:58] *** maxandersen has quit IRC
[16:40:51] *** mbg|away is now known as mbg
[16:44:07] *** maxandersen has joined #jboss-as7
[16:44:08] *** ChanServ sets mode: +v maxandersen
[16:48:05] *** pferraro has quit IRC
[16:48:32] <maxandersen> emuckenhuber: https://issues.jboss.org/browse/JBAS-8928
[17:08:53] *** maxandersen has quit IRC
[17:22:36] <bstansberry> emuckenhuber: the problem is one of the branches in the read-attribute handler isn't creating an OperationResult
[17:24:00] <bstansberry> the one that handles the case where it's not an attribute it's a child type
[17:26:22] <bstansberry> this distinction between attributes and children is going to prove to be an issue
[17:26:24] *** torben has quit IRC
[17:29:50] <emuckenhuber> ah, yes i see
[17:30:31] * bstansberry ponders having the read-resource handler create intermediate levels to separate attributes and children
[17:31:48] <bstansberry> looking quickly at the read-resource handler it's doing a lot of manipulation; i.e. not just copying out the model.
[17:33:13] <bstansberry> so having it fix up the result wouldn't kiil things
[17:33:17] <bstansberry> kill
[17:36:09] *** pferraro has joined #jboss-as7
[17:36:09] *** ChanServ sets mode: +v pferraro
[17:36:17] <emuckenhuber> with intermediate levels you mean for recursive queries ?
[17:44:13] <emuckenhuber> btw. i tried to use the VFS zip inputStream thingy to calculate the hash for a exploded deployment, but that takes like a minute - so looks like we need to do that in a different way
[17:48:31] <Nihility> bstansberry: Wouod that lead to asymmetric modifications/reads
[17:49:57] <bstansberry> emuckenhuber: sorry -- use my nick; i'm coding so may not see responses otherwise
[17:52:00] <bstansberry> intermediate levels == {"attributes"=>{"scan-enabled" => true, "scan-interval" => 5000,"relative-to" => "jboss.server.base.dir"},"children=>{"path"=>"deployments"}...
[17:52:14] <bstansberry> let me try that again ;)
[17:52:37] <bstansberry> oh, no, ^^^ was ok
[17:53:20] <bstansberry> so a read-resource would add "attribute" and "children" levels to make clear which is which, even though the internal model doesn't store it that way
[17:53:27] <emuckenhuber> so, based on the resource description?
[17:53:59] <Nihility> How do you reference path after see that result though
[17:54:03] <bstansberry> a quick look at the handler now (*very* quick) I saw a lot using the registry already; i.e. it's using the registry to decide
[17:54:40] <bstansberry> Nihility: you mean in another operation?
[17:54:45] <Nihility> yes
[17:55:18] <emuckenhuber> yeah, a lot of the handlers do use the registry - except the read-resource(recursive=true)
[17:55:21] *** mbg is now known as mbg|away
[17:55:29] <bstansberry> subsystem=deployment-scanner,scanner=default:read-attribute(name=,path=deployments
[17:55:29] <Nihility> what is path in the internal model
[17:56:08] <bstansberry> oh internally it's just subsystem/deployment-scanner/scanner/default/path
[17:56:23] <bstansberry> but the end user doesn't care about the internal model
[17:56:52] <bstansberry> OT: this bug may be a poor example of what we're talking about
[17:57:10] <bstansberry> since I don't think "path" is actually a child!!!
[17:57:49] <bstansberry> checking now, but I bet it's an unregistered attribute
[17:57:50] <emuckenhuber> well the issue is that you have to register the attribute in order to use read-attribute - which is obviously not added ;)
[17:58:28] <bstansberry> right, so that needs fixign
[17:58:31] <bstansberry> fixing
[17:58:55] <bstansberry> but, let's assume it was a child, and max just didn't realize that
[17:59:12] <Nihility> If it was a child it would have had {
[17:59:32] <bstansberry> an attribute could too
[17:59:38] <bstansberry> an attribute can be complex
[17:59:54] <Nihility> Right am object attribtute
[18:00:33] <Nihility> my big worry is that what they read would look different then what they write
[18:00:51] <bstansberry> you can't write a resource
[18:01:17] <bstansberry> an "add" would look different though
[18:01:26] <Nihility> that's what I mean
[18:04:21] <bstansberry> if there are children, an add may look different anyway
[18:04:46] <Nihility> Btw I thats not the only attribute with this problem
[18:04:53] <bstansberry> but, still, it's a valid point
[18:04:57] <Nihility> I noticed this when I was playing with the http API
[18:05:18] <bstansberry> yeah, it may be very common
[18:05:39] <bstansberry> i'm thinking whether the handler should just deal with it
[18:06:54] <bstansberry> i.e. if it's not a registered child but it exists in the model, read it
[18:08:06] <Nihility> To you're earlier point, while they may look different, I imagine in most cases they would be very very similar. It would be strange if calling code had to flatten attributes and children every time
[18:08:56] <Nihility> Yes read attribute should just resolve any non addressable thing
[18:11:25] <Nihility> not that I am againt the idea
[18:11:43] <bstansberry> if it's addressable we can return a better error message
[18:12:14] <bstansberry> i wouldn't add an intermediate level without input from the heiko's
[18:12:21] <bstansberry> they are the sophisticated clients
[18:12:39] <Nihility> also note that the description tells you what's what
[18:12:55] <bstansberry> yeah, that's what i've been relying on
[18:13:27] <bstansberry> but then when you see what people do when they use the cli, it changes your perspective a bit :)
[18:14:06] <Nihility> actually to be honest
[18:14:18] <Nihility> A more intuitive approach would be one read call
[18:14:36] <bstansberry> just read-resource
[18:14:58] <bstansberry> i think we need more for things like reading a single metric
[18:15:00] <Nihility> and it just traverses the model exactly as you say
[18:15:35] <Nihility> really the only reason something is a resource is because it has operations
[18:15:41] <Nihility> That's about it
[18:16:04] <bstansberry> it's also a boundary for recursion
[18:16:24] <Nihility> ah yes that's a good point
[18:17:21] <bstansberry> but maybe this usage pattern isn't real
[18:17:32] <Nihility> yeah it's probably not
[18:17:38] <bstansberry> i.e. do a read-resource, see "path" then do read-attribute
[18:18:07] <bstansberry> a better error message describing that "path" is a child might be sufficient
[18:18:23] <Nihility> i think reading a single attribute or rather a single value makes sense
[18:18:31] <bstansberry> oh yeah, for sure
[18:18:35] <Nihility> as you say polling metrics
[18:18:47] <Nihility> Oh nm you were talking about something else
[18:19:02] <Nihility> yes a good error message is all you need
[18:19:13] <bstansberry> but doing it after you've just read the resource, that's more "I'm playing with the CLI" than a real workflow
[18:20:06] <bstansberry> emuckenhuber: can you fix this stuff? I'm in domain-ops mode this weekend
[18:20:14] <emuckenhuber> bstansberry: sure
[18:20:17] <Nihility> we could also have a flag (label resource)
[18:20:31] <bstansberry> ?
[18:21:20] <Nihility> like the read resource call could take an option that adds extra meta infofmraton
[18:22:00] <Nihility> imaresouece=true
[18:22:25] <bstansberry> imaresource then becomes a reserved word
[18:22:34] <Nihility> well the nesting is the only way to handle undefine children
[18:22:56] <Nihility> hmm
[18:24:12] <bstansberry> for now, let's keep it simple. emuckenhuber, I think the task is:
[18:24:24] <bstansberry> 1) Have that handler not return null
[18:25:30] <bstansberry> 2) If the attribute isn't registered, the handler should check if there is a child of that type registered. if yes, the error message should be helpful, say it's a child, use 'read-xxxx' to read etc
[18:26:04] <bstansberry> 3) if there's no child registered either, but the data exists in the model, assume it's a missing attribute registration and read it
[18:26:29] <bstansberry> 4) if there's no child registered and no data in the model, it's a user error
[18:27:05] <Nihility> the cli actually could be smarter it could either react to the error and translate it or it could read the descripton and add formatting like make resources bold
[18:27:17] <bstansberry> 5) ping Ike and see if having an intermediate level would actually be preferable for him -- if so we can consider that
[18:29:27] <bstansberry> yeah, the cli could potentially make it clearer
[18:29:55] <bstansberry> although we wouldn't want to mess with people trying to script the CLI
[19:20:05] *** pferraro has quit IRC
[19:26:20] <emuckenhuber> bstansberry: https://github.com/emuckenhuber/jboss-as/commit/180689192ccf166715eb6370a799bcb3cee4194b
[19:27:22] <bstansberry> } else if(context.getSubModel().hasDefined(attributeName)) {
[19:27:34] <bstansberry> that should be "has" not "hasDefined"
[19:28:12] <emuckenhuber> right
[19:29:54] <bstansberry> ah, i hadn't noticed that the 3rd branch in the old code would return null as well
[19:31:26] <emuckenhuber> so i squashed that into this commit: https://github.com/emuckenhuber/jboss-as/commit/66d65f2d351c8141b7c4dbee526f8d26094b869e
[19:31:57] <emuckenhuber> hmm, so is there something missing? maybe some more tests
[19:32:30] <bstansberry> you mean on this issue?
[19:32:40] * bstansberry needs a 3rd AS 7 checkout
[19:32:54] <emuckenhuber> :)
[19:32:58] <emuckenhuber> i just meant this issue for nwo
[19:34:12] <bstansberry> I think this covers it
[19:41:01] <emuckenhuber> ok, cool - thanks for spotting the hasDefined
[19:42:12] <dmlloyd> smarlow: it really depends on whether you're creating a new instance each time, or returning a common instance, from your JNDI binding.
[19:43:09] <smarlow> dmlloyd:  returning a new instance each time should work. :)
[19:46:09] <bstansberry> emuckenhuber: pushed
[19:46:13] <bstansberry> thanks!
[19:55:54] * smarlow hibernate validator seems to need slf4j-log4j12 and that is in the banned dependencies list (noted as causing errors)
[20:02:07] <bstansberry> can someone have a look at https://github.com/bstansberry/jboss-as/commit/edbb17d9183bf46156a37b8026ade0b13e618f31 please ?
[20:08:35] <emuckenhuber> bstansberry: seems ok for me, certainly better than having a "undefined" host :)
[20:08:38] <emuckenhuber> or server
[20:09:11] <bstansberry> yeah. it's more like the Alpha1 behavior, which got lost
[20:09:31] <bstansberry> i'm changing it though so it only tries to determine the default if the parser didn't read a name
[20:09:51] <bstansberry> more robust; why potentially fail calculating something not needed?
[20:24:11] <bstansberry> the CLI doesn't seem to be able to cross-proxies
[20:24:58] <bstansberry> well, maybe that means nothing
[20:25:15] <bstansberry> i.e. i shouldn't expect it to work yet
[20:46:11] <Nihility> smarlow: there is an sl4j api that works with jboss logging
[20:46:41] <Nihility> smarlow: we should exclude the sl4f-log4j one and add the special one instead
[20:47:26] <smarlow> Nihility:  okay, I'll try that. thanks!
[20:51:36] <Nihility> smarlow: https://repository.jboss.org/nexus/content/repositories/releases/org/jboss/slf4j/slf4j-jboss-logging/
[20:53:51] <smarlow> I'll add a module definition for that and include it in the hibernate validator module
[20:53:56] <smarlow> thanks :-)
[20:59:44] <smarlow> cool, got a PersistenceProvider loaded (without the validator but I'll circle back and try that)
[21:03:38] *** jamezp has joined #jboss-as7
[21:24:17] <smarlow> Nihility:  looks like org.jboss.slf4j:slf4j-jboss-logging ^ is also in the banned dependency list
[21:29:28] <Nihility> oops
[21:29:32] <Nihility> i picked the wrong one
[21:29:34] <Nihility> slf4j-jboss-logmanager
[21:29:35] <Nihility> :)
[21:29:50] <smarlow> cool, that is easier to deal with :)
[21:31:23] <smarlow> yup, slf4j-jboss-logmanager is already defined as well.  I'll add to hibernate validator module definition and see what happens
[21:36:53] *** jamezp has quit IRC
[22:07:20] *** jamezp has joined #jboss-as7
[22:23:49] <Nihility> jamezp: looking for a small project to contribute :)
[22:23:52] <Nihility> ?
[22:24:07] <jamezp> Sure.
[22:24:55] *** pferraro has joined #jboss-as7
[22:24:55] *** ChanServ sets mode: +v pferraro
[22:25:16] <Nihility> you might not find it that interesting, so i dont blame you laughing in my face
[22:25:43] <jamezp> Doesn't matter to me at all :-)
[22:26:17] <Nihility> but we need a maven plugin to deploy stuff using the remote management api
[22:26:29] <Nihility> its probably a very small amount of work
[22:26:34] <Nihility> it just gets sidelined alot
[22:26:56] <jamezp> I'd be happy to help.
[22:27:05] <Nihility> awesome you rock
[22:27:19] <jamezp> Never done a maven plugin, so I'd get a chance to do some learning :-D
[22:28:13] <Nihility> i have no idea if there is a common convention for deploying to app servers with maven
[22:28:56] <jamezp> I'll take a look. I don't remember ever seeing one, but I haven't really looked that hard.
[22:29:09] *** mbg|away is now known as mbg
[22:29:18] <Nihility> yeah i imagine its just something you do as a custom goal
[22:29:39] <Nihility> and then the consuming project decides what lifecycle to map it to
[22:29:57] <jamezp> Right, that makes sense.
[22:31:15] <Nihility> btw maxandersen (jboss tools) had a cool feature idea for logging in domain mode
[22:31:27] <Nihility> that we give each server a different color!
[22:31:36] <Nihility> (color logs!)
[22:31:45] <jamezp> I was actually just looking into that :-)
[22:31:50] <Nihility> really?
[22:31:57] <Nihility> cool
[22:32:06] <jamezp> I saw something about it here on Friday I think. So I thought I would look into it.
[22:32:35] <Nihility> i go kind of crazy with color prompts and so on
[22:33:16] <jamezp> I could see in some instances it being useful.
[22:33:46] <Nihility> it would certainly help draw attention to things
[22:33:52] <Nihility> like if its a single server
[22:33:58] <Nihility> and displaying an error message in red
[22:34:00] <jamezp> I can also see it getting too crazy if you start coloring everything though :-)
[22:34:05] <Nihility> lol totally
[22:34:20] <Nihility> and hisorically the big problem with color is that not all terminals support it
[22:34:51] <jamezp> Yes, I spend large parts of my days in terminals.
[22:35:05] *** pferraro has quit IRC
[22:35:15] <Nihility> ah yes so you know the pain of logging in on a serial console and some jackasses' color settings making things unreadable
[22:35:18] <Nihility> !
[22:35:47] <jamezp> Hahaha
[22:36:31] <Nihility> did you ever enable git's color support?
[22:36:33] <Nihility> its very nice
[22:36:52] <jamezp> I haven't, but maybe I should. I could actually see that being useful.
[22:37:15] <Nihility> yeah like for example if a diff has a trailing whitespace character
[22:37:22] <Nihility> it marks that as a red block
[22:37:42] <jamezp> That's awesome. Definitely checking that out.
[22:37:59] <Nihility> i also use the git shell feature
[22:38:05] <Nihility> and it supports color as well
[22:38:16] <Nihility> which is more of a nicety
[22:38:28] <Nihility> when im in the dir the branch is a bold yellow color
[22:38:43] <Nihility> so hard to forget what i last checked out
[22:38:45] <jamezp> Exactly, which is probably how the logging should handle it. An optional nicety.
[22:40:02] *** pferraro has joined #jboss-as7
[22:40:02] *** ChanServ sets mode: +v pferraro
[22:46:04] <jamezp> Nihility: Any specifics on what exactly the Maven plugin needs to do? Just deploy an app?
[22:47:10] <Nihility> yeah supporting just basic standalone deployment initially
[22:47:21] <Nihility> domain is a bit more involved and might not make sense
[22:48:01] <Nihility> let me get the details for you
[22:48:13] <jamezp> Perfect, thanks!
[22:48:30] <stuartdouglas> should it also support starting/stopping the container, and running in embedded mode like the jetty plugin?
[22:48:55] <Nihility> yeah i think we should consider doing an embedded mode thing like jetty at some point
[22:49:04] <Nihility> not sure if that should be a separate plugin or not
[22:50:29] <Nihility> http://community.jboss.org/wiki/AS7DetypedManagementAPI
[22:50:44] <Nihility> http://community.jboss.org/wiki/FormatofaDetypedOperationRequest
[22:51:32] *** pferraro has quit IRC
[22:51:46] <stuartdouglas> I think there should just be one plugin
[22:52:47] <Nihility> stuartdouglas: morning btw!
[22:54:30] <Nihility> jamezp: the demos show an example of using the above, although you dont need to do the whole deployment plan thing
[22:54:59] <jamezp> Cool. I'll read through them.
[22:55:12] <Nihility> jamezp: you likely just need to call the simple replace-deployment, deploy, and perhaps undeploy
[22:55:31] <Nihility> let me find the class that implements those
[22:56:06] <stuartdouglas> Nihility: https://github.com/stuartwdouglas/jboss-as/tree/webComponents
[22:56:21] <stuartdouglas> I think the web component stuff is pretty much ready to go
[22:57:14] <Nihility> excellent!
[22:57:27] <stuartdouglas> that patch also add CDI injection into components
[22:58:20] <stuartdouglas> The only major bit that is missing is @PreDestroy methods
[22:59:09] <stuartdouglas> but dmlloyd said he had a plan for them, so I will see what he comes up with
[23:00:05] <Nihility> predestroy for servlets?
[23:00:09] <Nihility> or managedbeans
[23:00:26] <stuartdouglas> servlets
[23:00:39] <stuartdouglas> or managed beans used as sevlets
[23:01:05] <Nihility> jamezp: take a look at DeploymentUploadStreamAttachmentHandler (thats what implements the server piece to add a new deployment to the repository), and also DeploymentReplace, DeploymentAdd (which enable or replace a deployment in the repo)
[23:01:21] <stuartdouglas> basically the trick is going from an instance to a ComponentInstance to destroy the component
[23:01:26] *** pferraro has joined #jboss-as7
[23:01:26] *** ChanServ sets mode: +v pferraro
[23:01:42] <Nihility> ok so dmlloyd and i were talking about managedbeans
[23:01:55] <jamezp> Nihility: Perfect, I'll take a look. Thanks!
[23:01:57] <Nihility> and we agreed that predestroy is basically pointless
[23:02:01] <stuartdouglas> we could do it with a weak concurrent identity based HashMap, but that is kinda yuck
[23:02:04] <stuartdouglas> not always
[23:02:15] <Nihility> the only marginally performant way to do it
[23:02:18] <Nihility> is in a finalizer
[23:02:23] <stuartdouglas> e.g. if you inject them into something that does have a well defined lifecycle
[23:02:24] <Nihility> but that bogs down gc, and is slow
[23:02:47] <Nihility> and its somewhat dangerous
[23:02:55] <Nihility> yes basically the only safe way
[23:03:02] <Nihility> and reasonable fast way
[23:03:13] <Nihility> is to have whoever gets the instance
[23:03:20] <Nihility> decide when to call preDestroy
[23:03:26] <Nihility> so if a user looks it up in jndi
[23:03:30] <Nihility> they should just call it :)
[23:03:38] <Nihility> servlets on the other hand
[23:03:43] <Nihility> have a well defined lifecycle
[23:03:49] <stuartdouglas> That sounds reasonable
[23:04:26] <stuartdouglas> yea, the problem I have is that in the tomcat SPI it is just destroy(instance)
[23:04:47] <stuartdouglas> and I need some way of going from that instance to the corresponding ComponentInstance
[23:05:00] <Nihility> hmm
[23:05:06] <stuartdouglas> my first approach was to have proxies that implement DestryableComponent
[23:05:30] <stuartdouglas> however there is no guarantee that servlet classes are proxiable
[23:06:02] <Nihility> can you intercept preDestroy?
[23:06:16] <Nihility> i mean can the user intercept it
[23:06:18] <stuartdouglas> for servlets? no
[23:06:38] <Nihility> basicause imo
[23:06:41] <stuartdouglas> but it is not enough to just call the preDestroy method, you also need to release all the injected resources
[23:08:03] <Nihility> ok so really the web code needs to have a component instance for its destroy
[23:08:53] <Nihility> the only other hack i can think of is if you can go from servlet instace -> servlet context -> component instance
[23:09:02] <stuartdouglas> I could use a concurrent identity based hashmap to map instances to ComponentInstances, but that is kinda yuck
[23:09:32] <Nihility> yeah thats not ideal at all
[23:10:03] <Nihility> theres no reason though why that destroy method couldnt be changed to something else
[23:10:14] <Nihility> it could even be a generic type with an extra Object field
[23:11:09] <Nihility> so how do the injection deps clear up right now?
[23:11:40] <stuartdouglas> at the moment they don't
[23:12:02] <stuartdouglas> they will be gc'ed, but they won't have pre destroy called
[23:12:37] <stuartdouglas> sevlet classes can be final right?
[23:12:42] <Nihility> so basically anything that a servlet depends on
[23:12:46] <stuartdouglas> otherwise I could just subclass the the servlet
[23:13:07] <Nihility> will never be released
[23:13:47] <Nihility> yeah well theres nothing that says they can not be
[23:15:23] <stuartdouglas> The web interface that this uses is org.apache.tomcat.InstanceManager
[23:15:33] <stuartdouglas> which is used quite a bit in the web code
[23:15:45] <stuartdouglas> changing it will probably be non-trivial
[23:17:18] <bstansberry> jamezp, Nihility: on the logging in color thing...
[23:17:34] <bstansberry> what Max and I had talked about at Devoxx was the console log in domain mode
[23:17:50] <bstansberry> [Host Controller] 16:12:17,385 INFO  [org.jboss.as.host.controller.mgmt] (pool-1-thread-5) Server [Server:server-one] registered using connection [org.jboss.as.protocol.ConnectionImpl@19336006]
[23:17:51] <bstansberry> [Server:server-one] 16:12:17,443 INFO  [org.jboss.remoting] (MSC service thread 1-2) JBoss Remoting version 3.1.0.Beta2
[23:17:51] <bstansberry> [Server:server-two] 16:12:17,475 INFO  [org.jboss.msc] (main) JBoss MSC version 1.0.0.Beta6
[23:18:24] <bstansberry> [Host Controller] [Server:server-one] and [Server:server-two] would have different colors
[23:18:55] <jamezp> Ah yes, that makes sense actually.
[23:19:01] *** emuckenhuber has quit IRC
[23:19:19] <bstansberry> not even the whole line, just the start
[23:19:38] <bstansberry> that was the basic use case; i'm sure people can go nuts with color ;)
[23:20:13] <jamezp> And I'm sure they would :-) I see it all the time with terminal colors.
[23:20:20] <bstansberry> In that youtube demo I did there's a spot where I pause for 5 secs thinking something didn't work, just because I couldn't find the log entry :)
[23:32:00] <Nihility> stuartdouglas: the only other option is that it looks like tomcat uses a standardwrapper object to wrap all servlets
[23:32:41] <Nihility> StandardWrapper.unload() seems to be what calls destroy
[23:34:17] *** emuckenhuber has joined #jboss-as7
[23:34:17] *** emuckenhuber has quit IRC
[23:34:17] *** emuckenhuber has joined #jboss-as7
[23:34:17] *** ChanServ sets mode: +v emuckenhuber
[23:34:41] <jamezp> Nihility: Any specific package names and artifact name you want to use for the maven plugin?
[23:34:46] <dmlloyd> smarlow / Nihility don't worry slf4j is already in there, no additional deps needed
[23:35:03] <Nihility> hi dmlloyd!
[23:35:12] <Nihility> yeah we figured that out
[23:35:22] *** frainone has quit IRC
[23:36:32] <Nihility> stuartdouglas: some googling shows that quite a few people use final servlets
[23:36:35] <Nihility> http://google-web-toolkit.googlecode.com/svn/javadoc/1.6/com/google/gwt/user/server/rpc/RemoteServiceServlet.html#doPost(javax.servlet.http.HttpServletRequest,%20javax.servlet.http.HttpServletResponse)
[23:36:45] <stuartdouglas> yea, I don't think subclassing is an option
[23:37:58] *** pferraro has quit IRC
[23:42:41] <smarlow> dmlloyd, Nihility:  yeah, I might be wrong about the cause of the failure to load the hibernate validator.
[23:43:20] *** bstansberry is now known as bstans_afk
[23:44:03] <stuartdouglas> Nihility: other than the no PreDestroy thing, does the patch look ok?
[23:46:39] <Nihility> stuartdouglas: why is the WebEEcomponentdeployer empty?
[23:48:52] <stuartdouglas> oops
[23:50:46] <stuartdouglas> It's gone now
[23:51:09] *** mbg is now known as mbg|away
[23:54:14] <Nihility> man this is a big patch
[23:54:14] <Nihility> haha
[23:55:32] <stuartdouglas> its only 1300 lines :-)
[23:55:37] <Nihility> true
[23:55:41] <Nihility> it just looks huge
[23:55:52] <Nihility> so was that just a bogus class
[23:55:53] <Nihility> ?
[23:55:59] *** mbg|away is now known as mbg
[23:56:05] <stuartdouglas> yea, left over from my first one
[23:56:12] <stuartdouglas> originally I had two deployers
[23:56:35] <stuartdouglas> for the web components, one to create them, and then one to create the instantiators
[23:56:55] <stuartdouglas> now I just create the instantiators straight away and lazily load the required service
[23:57:54] <jamezp> Nihility: I've got to head out soon to pick up my wife from the airport. I'll get something going on the maven plugin, then see if I can't take on the logging colors.
[23:58:07] <Nihility> ok cool
[23:58:09] <Nihility> have a safe trip
[23:58:26] <jamezp> Will do thanks!
[23:59:42] *** bstans_afk is now known as bstansberry

   March 5, 2011  
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | >