October 30, 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

[00:08:11] *** aslak has quit IRC
[00:33:32] *** sbryzak has quit IRC
[00:53:05] *** struberg has quit IRC
[01:18:26] *** stuartdouglas has joined #weld-dev
[02:39:01] *** alesj has quit IRC
[06:19:41] *** alex57711 has quit IRC
[06:21:04] *** alex5771 has joined #weld-dev
[09:09:31] *** struberg1 has joined #weld-dev
[09:45:26] *** struberg1 is now known as struberg
[09:46:04] <struberg> stuartdouglas ping
[09:46:12] <struberg> good evening!
[09:59:40] <stuartdouglas> hey
[09:59:58] <stuartdouglas> struberg: pong
[10:01:00] <struberg> stu did you get a chance to read my comment in CDI-139?
[10:01:02] <jbossbot> jira [CDI-139] Support for unmanaged instances [Open (Unresolved) Feature Request, Major, Unassigned] https://issues.jboss.org/browse/CDI-139
[10:01:40] <struberg> to store CreationalContext + instance ref inside Instance<T> for @Dependent instances which got created
[10:02:08] <struberg> and then on Bean#destroy of the Instance to release them again
[10:03:30] <struberg> or might it be that they have a longer lifetime?
[10:08:40] <stuartdouglas> there is still the leak
[10:08:47] <struberg> where?
[10:08:58] <struberg> I rather think it might sometimes be too aggressive
[10:09:04] <stuartdouglas> if it is injected into an app scoped bean
[10:09:21] <struberg> then all will get destroyed after the app closes
[10:09:25] <struberg> imo thats not a mem leak
[10:10:11] <struberg> currently they will _never_ get destroyed, even if the bean containing the Instance<T> is only @RequestScoped or even shorer
[10:10:30] <stuartdouglas> I think the way it is currently implemented in weld they are
[10:10:46] <struberg> when do you destroy it?
[10:11:04] <struberg> at least the @PreDestroy will not get called as far as I've seen in quick tests
[10:11:13] <stuartdouglas> or maybe not then
[10:11:25] <struberg> simple as that
[10:11:45] <struberg> for @Dependent beans you need to store the CreationalContext + Contextual + instance in InstanceImpl
[10:11:54] <struberg> for every created @Dependent bean
[10:12:31] <struberg> otherwise it's just as using BeanManager#getReference
[10:12:55] <struberg> if you use getReference to create @Dependent instances, they will also never get released properly
[10:14:33] <stuartdouglas> In the weld code it looks like beans created through get are added to the CreationalContext for the Instance
[10:14:46] <stuartdouglas> so they should get destroyed when the Instance is
[10:15:10] <struberg> oh yes that is clever
[10:15:24] <struberg> but you only add @Dependent beans to the creationalContext, right?
[10:15:30] <stuartdouglas> yea
[10:15:41] <stuartdouglas> non dependent beans just follow the normal scoping rules
[10:15:47] <struberg> yup
[10:15:59] <struberg> then this should work already in Weld
[10:16:17] <struberg> can you please write a quick test case in Weld, maybe I had an error in mine
[10:16:28] <struberg> and if it works well, then we just need to clarify the behaviour in the spec
[13:03:48] *** alesj has joined #weld-dev
[13:42:32] *** alesj has quit IRC
[17:50:21] *** struberg has quit IRC
[17:58:08] *** struberg has joined #weld-dev
[18:21:03] *** alex57711 has joined #weld-dev
[18:24:01] *** alex5771 has quit IRC
[19:10:49] *** alex5771 has joined #weld-dev
[19:13:53] *** alex57711 has quit IRC
[20:07:15] *** struberg has quit IRC
[23:15:38] *** struberg has joined #weld-dev
[23:24:25] *** struberg has quit IRC
[23:31:33] *** alesj has joined #weld-dev
[23:51:16] <alesj> stuartdouglas: yt?
[23:51:20] <stuartdouglas> yes
[23:51:37] <alesj> you're a Keynote expert, right? :-)
[23:51:41] <stuartdouglas> no
[23:51:46] <stuartdouglas> Never used it
[23:51:49] <alesj> damn
[23:52:10] <alesj> i cannot get my Theme to be available
[23:52:14] <alesj> saving it
[23:52:19] <alesj> but then cannot choose it
[23:54:43] <alesj> stuartdouglas: what's the easiest way to search for files; e.g. *.kth?
[23:54:58] <stuartdouglas> find . -name \*.kth
[23:55:09] <stuartdouglas> at least, thats the way I do it :-)

top