Time is UTC+2 (CET) [16:46:16] bah git is not happy https://integration.wikimedia.org/ci/job/pywikibot-core-tox-nose34-trusty/2/console :D [16:48:29] git reset not found? [16:52:16] jenkins is stupid [17:11:04] tried again with GIT_TRACE=1 https://integration.wikimedia.org/ci/job/pywikibot-core-tox-nose34-trusty/10/console [17:13:09] 'git-submodule' 'foreach' '--recursive' 'git reset' '--hard' [17:13:14] and yeah [17:13:17] "git reset" does not exist [17:14:51] that stupid plugin is broken beyond repair [17:33:05] xzise: the jenkins git plugin is broken beyond repair :-( [17:33:29] I have tried using the Jgit implementation (pure java, does not use the git command) and it fails as well https://integration.wikimedia.org/ci/job/pywikibot-core-tox-nose34-trusty/12/console [17:34:43] Hmmm is that a problem with trusty? [17:34:49] hop with Jenkins [17:34:52] Or what does that plugin do? [17:35:08] when resetting submodules it screw up because it uses: [17:35:13] git submodule foreach "git reset" [17:35:17] note the quotes [17:35:27] so it attempts to execute literally "git reset" which is not found [17:35:30] should be: git reset [17:35:33] i.e. two arguments [17:35:44] the Jgit implementation simply does not work [17:37:17] But why it's happening to nose34 and not to any other test [17:37:28] yeah that is a good question :d [17:37:52] I guess https://integration.wikimedia.org/ci/job/pywikibot-core-tox-nose uses submodules as well isn't it ? [17:39:32] ah [17:39:41] looking on https://integration.wikimedia.org/ci/job/pywikibot-core-tox-nose/1023/console [17:39:48] it does not process submodules at all [17:40:03] there are some copy floating around but they are from August 8th [17:41:45] And git submodule foreach "git reset" does work for me [17:42:41] the pywikibot-core-tox-nose job does not process / update them apparently :-( [17:51:20] git reset --ALL-the-things [17:52:06] I give up for today sorry :( [17:52:12] hashar: git-reset should exist as command, I think [17:52:16] yeah [17:52:25] but gitreset does not [17:52:25] except it doesn't on my computer :/ [17:52:26] :D [17:52:33] though we could create a symbolic link [17:52:40] no, I meant like git-receive-pack which can also be invoked with git receive-pack [17:53:19] $ git submodule foreach "'echo foo'" [17:53:19] Entering 'AJAXPoll' [17:53:25] /usr/local/Cellar/git/2.1.0/libexec/git-core/git-submodule: line 553: echo foo: command not found [17:53:47] though running the job on Precise works fine https://integration.wikimedia.org/ci/job/pywikibot-core-tox-nose34-trusty/17/console [17:53:56] it has git version 1.7.9.5 [17:54:16] while Trusty has git version 1.9.1 [17:56:20] hashar: maybe use single instead of double quotes? [17:56:26] don't think it should matter, bit still [17:56:32] or even no quotes at all.. [17:56:51] might be a change of behavior in how git handles args passed to submodule foreach [17:57:42] yeah [17:57:48] that is a new feature of git 1.9.0 :-] [17:58:15] it is the first paragraph of the release notes https://raw.githubusercontent.com/git/git/master/Documentation/RelNotes/1.9.0.txt [17:58:18] \O/ [17:59:16] the Jenkins git client we use is outdated and I refrained from updating it cause it tends to crash everything [17:59:19] will try tomorrow [18:00:24] valhallasw`cloud hands hashar wine and cheese [18:00:38] yeah that is a very scary plugin :D [18:00:44] off for some call then back home (hashar rushing to conf call then dad rush hours)