doko | slangasek, do you remember which one? | 00:00 |
---|---|---|
slangasek | doko: the flag? not offhand | 00:00 |
slangasek | doko: google suggests 'expect stderr' restriction | 00:01 |
doko | hmm, it alreadys has: Restrictions: allow-stderr | 00:01 |
bdmurray | Riddell: oh, its mostly arm64 except for perlkde | 00:01 |
Riddell | bdmurray: so probably fine to release all except perlkde then | 00:02 |
slangasek | doko: right, https://bugs.launchpad.net/ubuntu/+source/autopkgtest/+bug/1197005 says that's the right syntax; I don't know then | 00:02 |
ubot2` | Launchpad bug 1197005 in autopkgtest (Ubuntu) "Add "expect stderr" restriction" [Medium,Fix released] | 00:02 |
doko | slangasek, ok, I'll check with piotr tomorrow. care to unblock it? the test did always fail | 00:03 |
xnox | doko: there could be other things, e.g. crash files generated (at least i'm under the impression that causes adt tests to fail, i can be wrong) | 00:04 |
doko | but then, who did overwrite it in the past? | 00:04 |
xnox | doko: i don't think it ever migrated from trusty-proposed to trusty. | 00:05 |
xnox | doko: the previous one was copied from saucy i think. | 00:05 |
slangasek | no | 00:05 |
slangasek | there's a previous 'force-badtest' hint from infinity | 00:05 |
xnox | doko: and tests did not pass. | 00:06 |
xnox | doko: if [ -x /usr/bin/python2.6 ]; then\ | 00:06 |
xnox | test -f debian/python-foo/usr/lib/python2.6/dist-packages/foo/__init__.p;\ | 00:06 |
xnox | test ! -f debian/python-foo/usr/lib/python2.6/dist-packages/foo/spam.py;\ | 00:06 |
xnox | fi | 00:06 |
xnox | grep -qe "Depends: .*python\(:any\)\? (<<" debian/python-foo/DEBIAN/control | 00:06 |
xnox | [ "`readlink debian/python-foo/usr/lib/python2.7/dist-packages/foo/absolute_link_to_tmp`" = "/tmp" ] | 00:06 |
xnox | make[1]: *** [check] Error 1 | 00:06 |
xnox | make[1]: Leaving directory `/tmp/adt-run.Vm3DtR/dsc0t-dh-python-testtmp/adttmp/tests/t201' | 00:06 |
xnox | make: *** [test201] Error 2 | 00:06 |
xnox | https://jenkins.qa.ubuntu.com/view/Trusty/view/AutoPkgTest/job/trusty-adt-dh-python/ARCH=amd64,label=adt/8/artifact/results/dsc0t-dh-python-stdout/*view*/ | 00:06 |
infinity | That's the same test that failed on the previous version and doko assured me it was fine and it would be fixed. :P | 00:07 |
slangasek | doko: but I don't like kicking the can down the road indefinitely. If there's not a pressing need for this fix in trusty, it should stay in -proposed and someone should fix the bug | 00:07 |
xnox | that's for 1.20131021-1ubuntu3 | 00:07 |
xnox | hm. me tries to find actual results in jenkins | 00:07 |
slangasek | ugh pbuilder | 00:08 |
slangasek | this is why I don't have autopkgtest installed on my laptop. :P | 00:08 |
xnox | doko: are you asking unblock for ubuntu3 or ubuntu4 ? cause britney hasn't asked to test ubuntu4 yet. | 00:08 |
xnox | slangasek: i use scripts from lp:auto-package-testing which don't have side effects of installing garbage on my machines. | 00:09 |
doko | xnox, 4 | 00:10 |
xnox | slangasek: those cleanly download ubuntu cloud image and do all the dirty work in a snapshot of thereof. and that's the same way adt tests are being executed in the lab. | 00:10 |
doko | infinity, did I? can't remember | 00:10 |
infinity | doko: Yeah. :) | 00:10 |
doko | slangasek, I do want this for the test rebuild | 00:11 |
infinity | doko: I think I let it in cause you said the version I unblocked was critical for some reason or other, but you also claimed the test would get fixed (perhaps by upstream). | 00:11 |
infinity | Unless you're upstream. | 00:11 |
infinity | Then, it was by you. :P | 00:11 |
xnox | doko: then you don't get to claim all tests have passed, in the ADT environment, cause ADT has not run yet for ubuntu4 yet ;-) | 00:11 |
doko | no, python helpers are a thing I'm staying away from. too many hostile people out there ... | 00:12 |
slangasek | xnox: needing to grab a cloud image when I have perfectly good, pbuilder-free schroots here also seems onerous ;) | 00:12 |
doko | hmm, I need to learn how to run autopkgtests locally ... | 00:12 |
xnox | slangasek: adt tests are expected to be able to break test-beds =/ so, lxc or kvm, no chroots here when running those. | 00:13 |
xnox | doko: an sbuild hook would be nice, to install and run the tests at the end of the build, wouldn't it? | 00:13 |
xnox | or just use a schroot run to do them. | 00:13 |
slangasek | doko: sudo adt-run /path/to/dh-python_1.20131021-1ubuntu4.dsc --- adt-virt-schroot trusty | 00:13 |
doko | infinity, looks like pitti did try to fix it, but didn't succeed. I don't think that I did ask you in the past ... | 00:13 |
xnox | slangasek: oh, nice. | 00:14 |
doko | will try to address this tomorrow | 00:14 |
infinity | doko: Given my hint unblocking it, you must have asked me. | 00:14 |
doko | anyway, have to wait for an eglibc upload for the test rebuild | 00:14 |
slangasek | infinity: you only add hints when doko asks you? | 00:14 |
=== xnox is now known as doko_ | ||
doko_ | infinity: can you please unblock cmake? | 00:15 |
infinity | doko: eglibc being worked on right now. | 00:15 |
=== doko_ is now known as xnox | ||
infinity | xnox: By fixing libarchive? | 00:15 |
=== Ursinha-afk is now known as Ursinha | ||
xnox | infinity: yes please. I'm a bit clueless about it. extracted the actual test-suite output and the test case is sane. not sure if it's symptoms of the same eglibc issue or not =/ | 00:16 |
infinity | xnox: If it uses strstr, it could be. | 00:17 |
infinity | And pretty much everything uses strstr. | 00:17 |
infinity | So, we'll see. | 00:17 |
xnox | infinity: but seriously, can you migrate cmake before/for the archive rebuild ? cause it's only invalidated by dependency.... | 00:47 |
infinity | xnox: Uhm, no. | 00:47 |
xnox | Uhm, ok. | 00:47 |
infinity | xnox: "invalidated by dependency" means it DEPENDS on that other package. | 00:47 |
infinity | xnox: As in, it depends on that version of libarchive. | 00:48 |
infinity | I'm not sure why that would be, in this case, mind you. | 00:50 |
xnox | infinity: i don't see how though =/ hm, not a condition i've seen before from britney. | 00:50 |
infinity | Happens all the time during ABI transitions, etc. | 00:50 |
infinity | Oh, it might be because of ppc64el. cmake on ppc64el depends on libarchive13, which only exists in -proposed. | 00:51 |
infinity | We could temporarily break that, but I'd prefer not to. | 00:51 |
infinity | Or I could cheat, by moving the ppc64el libarchive into the bootstrap repo. | 00:52 |
xnox | yeah... cause there is cmake/ppc64el in trusty-release.... | 00:52 |
slangasek | doko: fwiw, on the failing test ([ "`readlink debian/python-foo/usr/lib/python2.7/dist-packages/foo/absolute_link_to_tmp`" = "/tmp" ]), I see the following: | 00:53 |
slangasek | debian/python-foo/usr/lib/python2.7/dist-packages/foo: | 00:53 |
slangasek | lrwxrwxrwx 1 root root 51 Dec 20 00:43 absolute_link_to_tmp -> ../../../../share/pyshared/foo/absolute_link_to_tmp | 00:53 |
slangasek | that is obviously not an absolute symlink, and the test should be fixed | 00:53 |
slangasek | (or dh-python itself fixed) | 00:53 |
infinity | Oh, it's already in the bootstrap repo, not as easy to cheat as I was thinking. | 00:54 |
infinity | xnox: Anyhow, let's see if my glibc magically fixes it shortly. | 00:54 |
infinity | Well, "shortly"... Once I've built it. | 00:54 |
xnox | infinity: ah britney doesn't consider bootstrap repo. makes sense. | 00:55 |
slangasek | quite | 00:55 |
xnox | and copying across libarchive13/pp64el binary without source, to trick britney into migrating cmake, is wrong on so many levels..... right, i'll go bother about something else, whist infinity cooks eglibc =) | 00:56 |
slangasek | we could equally force-migrate libarchive before it's fixed on i386 | 00:57 |
slangasek | but if it's done soon... why bother | 00:57 |
infinity | xnox: britney sort of does consider the bootstrap repo to avoid too much pain, actually. | 00:58 |
infinity | But I may have missed a corner case. | 00:58 |
infinity | And I don't much care in this case, we should just fix the bugs. | 00:58 |
infinity | xnox: Huh. So, I can't reproduce that libarchive test failure. | 01:20 |
xnox | infinity: on i386? | 01:21 |
xnox | infinity: i can, but then again i haven't rebooted my machine into fresh kernels in a while. | 01:22 |
infinity | xnox: On i386, yes. In the same chroot where I *can* reproduce the java vs strstr bug. So, it's not that. | 01:22 |
xnox | infinity: yeah, it did smell fishy, and not fully related to the eglibc bug. | 01:22 |
xnox | infinity: went to try wacking it again, but i guess it's you who have retriggered a rebuild ;-) | 01:23 |
infinity | Yeah. | 01:24 |
infinity | If it fails on the buildd, I'll start comparing package versions. | 01:24 |
infinity | This chroot is stale by a few days, since I haven't upgraded it since I started looking at the java thing. | 01:25 |
infinity | Nope, it passed on the buildds now. | 01:25 |
infinity | So, racy test, or problem solved elsewhere. | 01:25 |
=== Ursinha-afk is now known as Ursinha | ||
slangasek | doko_: this autopkgtest has been broken since the first commit in the dh-python branch, one wonders why Piotr has tests if he's not going to use them. :P I can't tell if it's the test or dh_python that should be fixed; I'm filing a bug in Debian and unblocking | 01:57 |
infinity | Grr. I do love it when people sync -f over my changes and then DON'T FIX THE RESUTLING FTBFS. | 02:00 |
infinity | resulting, too. | 02:00 |
StevenK | "It built locally, must be a buildd issue!" | 02:05 |
* StevenK hides. | 02:05 | |
slangasek | some kind of irony that I installed autopkgtest to verify a dh-python bug, and on removal it left behind .pyc files | 02:15 |
infinity | StevenK: Given that the changelog specifically stated the fix was powerpc-specific, "it built locally" would be a pretty lousy excude for sync -f. | 02:17 |
infinity | excuse, too. | 02:17 |
infinity | And the old patches applied cleanly too, so "I think upstream might have fixed it" would also not work. :P | 02:18 |
Laney | It's easy to miss sync FTBFSen given that they don't mail | 09:05 |
infinity | Laney: It's easy to miss the FTBFS, it's a bit harder to miss the patches you overwrote. ;) | 09:31 |
infinity | Laney: But we chastised him appropriately with "I can't test it" doesn't mean "drop the patches", it means "ask someone who can". | 09:31 |
Laney | Oh yeah, it's still naughty to do that, but usually you'd get shouted at by soyuz before anyone else notices. | 09:33 |
infinity | Agreed that people who copy packages should get FTBFS grump mail. | 09:34 |
infinity | wgrant: Why don't they? Bug, or actually hard to do for some reason? | 09:34 |
wgrant | infinity: Both. | 09:34 |
infinity | wgrant: Let me guess, we have no concept of creator on those, just the signer? | 09:35 |
wgrant | infinity: It needs to guess at the publication to use. | 09:35 |
infinity | wgrant: How so? Isn't the build record a bit more specific in its lineage than the source? | 09:36 |
wgrant | infinity: The build just lists an archive, suite, architecture and sourcepackagerelease. | 09:36 |
infinity | wgrant: I mean, if you FTBFS in a PPA, it's the person who uploaded/copied to that PPA, if you FTBFS in primary, it's obviously a build record that was created after a copy to primary, which you can only do once. You can only create a record for a source once in any archive. | 09:37 |
wgrant | Notifications have historically gone to SPR.creator because derp. | 09:37 |
wgrant | They need to go to SPPH.creator | 09:37 |
infinity | wgrant: Okay, but do we know who copied the SPR into the archive? | 09:37 |
wgrant | But for that we need to guess at which SPPH is relevant. | 09:37 |
wgrant | Which apparently doesn't always work. | 09:37 |
wgrant | There's a similar problem with attributing translations for copies, not quite sure what's going wrong. | 09:38 |
infinity | Would it be mean for me to assign the bug to Celso and make him fix it the day he starts? | 09:38 |
wgrant | Yes :) | 09:40 |
infinity | You're no fun. | 09:41 |
apw | that would be very [just do it] mean thing to [just] do [it] | 09:44 |
Laney | Cruel to be kind | 09:50 |
=== doko_ is now known as doko | ||
sergiusens | can someone please check on golang-gocheck and golang-go-flags please? | 13:46 |
sergiusens | from the new queue | 13:46 |
xnox | infinity: doko: demote python-pyste to universe (got denewed into main?!) & demote gccxml src+bin to universe. Ready as per component-mismatches. | 13:55 |
doko | xnox, done | 14:01 |
xnox | cheers! | 14:04 |
=== sil2100 is now known as KURWA | ||
=== KURWA is now known as sil2100 | ||
=== psivaa is now known as psivaa-holiday | ||
=== Ursinha-afk is now known as Ursinha |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!