[01:23] <stgraber> cjwatson: ubiquity failed to build on all the architectures because of the grub test
[01:24] <stgraber> cjwatson: on x86 because grub isn't in the build chroot, on the other architectures because of that + missing debconf keys for grub settings
[01:40] <CIA-32> ubiquity: stgraber * r5315 ubiquity/tests/test_ubi_partman.py: Disable all the grub tests for now
[01:41] <stgraber> temporary workaround as I'm not sure exactly how we want to fix these. I guess we'd have to make some of the tests arch-dependent and either mock whatever grub call we do or have the package b-d on grub
[01:41] <CIA-32> ubiquity: stgraber * r5316 ubiquity/debian/changelog: releasing version 2.9.33
[10:57] <infinity> cjwatson: Looks like the new additions to nic-firmware bloated netboot beyond FLOPPY_SIZE again.  I need some sleep, so I'll hand it off to you to poke with a stick.
[11:00] <infinity> cjwatson: Is there any reason we don't check the size of the stuff we want to jam on the floppy and just make the floppy the right size (rounded to the next sector)?
[11:00] <cjwatson> mostly just as a guard against unreasonable growth I think
[11:00] <infinity> Fair enough.
[11:00] <infinity> Well, I leave it to you to sort out how much bloat Tim added. :P
[11:01] <infinity> It looks to be about 600k uncompressed.  No idea how well compression shrinks it.
[11:02] <cjwatson> only trivially
[11:02] <cjwatson> I knew about the bnx2x addition, anyway - I think it's appropriate
[11:02] <infinity> Oh yes, I'm not complaining about it.
[11:03] <infinity> I even saw the conversation yesterday.
[11:03] <infinity> And just didn't even think about needing to bump FLOPPY_SIZE until it failed to build on me just now. :P
[11:03] <infinity> And now it's 5am, so I leave it in your capable hands.
[11:03]  * cjwatson nods
[11:04] <cjwatson> I even have both amd64 and i386 sbuilds locally now.
[11:04] <infinity> I assume nic-fw.udeb lands in all arches, though?
[11:05] <cjwatson> They might have different limits.  I'll wait for them to succeed or fail.
[11:05] <infinity> Yeah, ARM seems to have just finished.
[11:05] <infinity> I intentionally scored PPC down to avoid a race with linux-meta, feel free to abuse it in the other direction to force a failure.
[11:06] <cjwatson> It can wait a bit.  I have a queue of stuff to do anyway.
[11:06] <infinity> Oh.  Well, in that case, I might do the FLOPPY_SIZE bit when I wake up. ;)
[11:07] <infinity> I know where and how to fix it, if you have other shit to be doing anyway. :P
[11:07] <cjwatson> Oh, but linux-meta has built, so whatever, I'll score it back up now.
[11:07] <infinity> Built, but possibly not published..
[11:08] <infinity> Yeah, it's being published as we speak.
[11:09] <cjwatson> That doesn't race with d-i in any meaningful way anyway.
[11:09] <infinity> Check.  I can never be sure who cares about what and why.
[11:10] <infinity> I just have a mental checklist that has a vague and probably entirely unecessary order.
[11:10] <cjwatson> d-i just needs the kernel itself published.
[11:11] <infinity> Yeah.  I guess seeds/meta/d-i can all happen in parallel, and it's just that they all need to be done before one attempts images.
[11:27] <cjwatson> stgraber: I'm looking at those failures now.  Thanks for putting a temporary patch in place.
[11:28] <infinity> cjwatson: And yeah, PPC failed with ENOSPC too.
[11:31] <cjwatson> Right.
[12:06] <CIA-32> debian-installer: cjwatson * r1656 ubuntu/ (4 files in 4 dirs):
[12:06] <CIA-32> debian-installer: Bump amd64, i386, and powerpc netboot image sizes by a megabyte to
[12:06] <CIA-32> debian-installer: accommodate nic-firmware growth.
[12:12] <CIA-32> debian-installer: cjwatson * r1657 ubuntu/debian/changelog: releasing version 20101020ubuntu124
[13:04]  * cjwatson runs the test suite under strace to catch bugs where it's using stuff like dmidecode and udevadm
[13:05] <cjwatson> xvfb-run strace -f -etrace=execve -s 1024 -o test.trace tests/run --no-xvfb --no-build; grep execve test.trace | egrep -v 'locale.*charmap|debconf-copydb|debconf-communicate|dpkg.*print-architecture|dpkg-architecture|tests/run|gcc.*dumpmachine'
[13:16] <CIA-32> ubiquity: cjwatson * r5317 trunk/ (4 files in 2 dirs):
[13:16] <CIA-32> ubiquity: Fix a few cases where the test suite was insufficiently isolated from
[13:16] <CIA-32> ubiquity: the host system, running dmidecode and udevadm.
[13:25] <CIA-32> ubiquity: cjwatson * r5318 trunk/ (debian/changelog tests/test_ubi_partman.py):
[13:25] <CIA-32> ubiquity: Restore 'from ubiquity import gtkwidgets' to
[13:25] <CIA-32> ubiquity: test_ubi_partman.TestPageGtk.setUp; importing custom widgets fails
[13:25] <CIA-32> ubiquity: without this.
[13:30] <CIA-32> ubiquity: cjwatson * r5319 trunk/ (7 files in 4 dirs): bump to 2.10
[14:31] <CIA-32> ubiquity: cjwatson * r5320 trunk/ (debian/changelog tests/test_migrationassistant.py): Avoid touching D-Bus in migrationassistant test.
[14:51] <CIA-32> ubiquity: cjwatson * r5321 trunk/ (3 files in 2 dirs):
[14:51] <CIA-32> ubiquity: Stop leaking locale changes from gtkui and migrationassistant tests into
[14:51] <CIA-32> ubiquity: the rest of the test suite.
[14:52] <CIA-32> ubiquity: cjwatson * r5322 trunk/ (debian/changelog tests/run tests/test_ubi_partman.py):
[14:52] <CIA-32> ubiquity: * Disable tests that require grub-installer templates only on non-x86
[14:52] <CIA-32> ubiquity:  architectures.
[14:52] <CIA-32> ubiquity: * Re-enable test_maybe_update_install on x86, and add the missing
[14:52] <CIA-32> ubiquity:  mock.patch calls that caused it to fail in clean build environments.
[14:53] <cjwatson> right, that's the tests sorted out, phew.  anything else I should include for 2.10.0?
[14:53] <cjwatson> maybe a translation refresh
[14:54] <barry> cjwatson: hmm.  if you're itching to make a new release, go for it.  i was going to spend a bit of time today looking at a few more bugs
[14:54] <barry> bug 792652
[14:54] <ubot2`> Launchpad bug 792652 in ubiquity "ubiquity crashed with ValueError in command(): I/O operation on closed file" [High,Confirmed] https://launchpad.net/bugs/792652
[14:54] <bdmurray> I'm getting an error trying to run tests - xvfb-run: error: Xvfb failed to start
[14:55] <cjwatson> bdmurray: not been seeing that one
[14:55] <barry> also, did i see stgraber disabled some of the grub installation tests yesterday because of failures on i386 buildds?  is that something i can/should look at?
[14:55] <cjwatson> barry: I want to make sure this is in before beta2 freeze
[14:55] <cjwatson> barry: oh, I just fixed that, I'm afraid
[14:55] <cjwatson> literally a minute before you joined
[14:56] <barry> cjwatson: \o/  i'll look at the diffs to see the fix.  as for the other bugs, i don't think they're critical, so don't wait for me
[14:57] <cjwatson> it was just lots of fiddly stuff, nothing groundbreaking
[14:57] <cjwatson> missing mocks, missing test skipping conditions, that kind of thing
[14:57] <barry> i see from the changelog you figured out why that import of gtkwidgets was there :)
[15:10] <bdmurray> I have a working test now, but I'm not clear on how to reduce the duplicate code in it http://paste.ubuntu.com/895189/
[15:11] <stgraber> bdmurray: I wouldn't worry about that little code duplication
[15:11] <stgraber> bdmurray: why is the first "@mock.patch('sys.stderr')" commented?
[15:12] <bdmurray> stgraber: I just forgot to remove it
[15:17] <stgraber> jibel: did you see any regression or pretty bad bugs in the installer today?
[15:17] <stgraber> I was surprised at how little installer bugmail I received overnight :)
[15:19] <cjwatson> barry: yeah, belatedly
[15:19] <cjwatson> bdmurray: something like:
[15:19] <cjwatson> def setUp(self):
[15:19] <cjwatson>    # stuff that's already there
[15:19] <jibel> stgraber, no regression so far. I essentially tested multi-disk setup, install with an existing windows partition and upgrades from ubiquity and found nothing interesting this morning.
[15:20] <cjwatson>     mock_stderr = mock.patch('sys.stderr')
[15:20] <stgraber> jibel: that's good to hear ;)
[15:20] <cjwatson>     patched_stderr = mock_stderr.start()
[15:20] <cjwatson>     self.addCleanup(mock_stderr.stop)
[15:20] <jibel> stgraber, just 1 minor detail, the kb indicator has no label at all now. This is confusing because ibus uses the same icon.
[15:20] <cjwatson>     patched_stderr.write = self.write_side_effect
[15:21] <cjwatson> def write_side_effect(self, *args, **kwargs):
[15:21] <cjwatson>     ...
[15:21] <stgraber> jibel: yeah, that was unfortunately the fix for the other indicator bug (where the label would be mixed with nm's or bluetooth's icon)
[15:21] <stgraber> jibel: I think the bug here is that ibus doesn't have a different icon... we have the same issue with some of the accessibility options
[15:22] <stgraber> (actually it'd make sense for ibus to use the same keyboard menu ...)
[15:44] <CIA-32> ubiquity: cjwatson * r5323 trunk/ (151 files in 3 dirs): Update translations from Launchpad.
[16:20] <CIA-32> ubiquity: cjwatson * r5324 ubiquity/debian/changelog: releasing version 2.10.0
[16:30] <bdmurray> has anybody looked at bug 909179?
[16:30] <ubot2`> Launchpad bug 909179 in ubiquity "[Precise] Installation freezes after "Who are you?" (Ubiquity hangs in ubi-webcam.py)" [Undecided,Confirmed] https://launchpad.net/bugs/909179