[00:14] <stgraber> slangasek: looks like that worked. The upgrade is still running but it's past the point where the crash would usually happen.
[00:14] <cjwatson> slangasek: ah, I was about to assign 1034806 to myself :)
[00:14] <cjwatson> slangasek: I think it probably wants to look like http://paste.ubuntu.com/1140237/ ?
[00:15] <slangasek> cjwatson: <cough> yes, I just realized I should probably assign it to myself because I've been working on it for the past half hour :P
[00:15] <cjwatson> which would apply to quantal, even though using python3 there dodges the issue
[00:15] <slangasek> yeah, that matches mine
[00:15] <slangasek> just working on the test case now
[00:15] <cjwatson> good, we followed the same chain of logic
[00:15] <cjwatson> a test case was what I hadn't got to yet
[00:16] <stgraber> I'd have thought that with all that LP work, you'd know be completely converted to TDD ;)
[00:16] <slangasek> also needs the iso-codes build-dep, which is in quantal already but not yet in precise
[00:17] <slangasek> stgraber: well when 80% of the work is tracking down the bug and the other 10% is split evenly between figuring out the fix for the bug and figuring out the test case... :)
[00:17] <cjwatson> stgraber: :-P  I make sure the test fails without the fix, but that doesn't always mean I write the test first ...
[00:18] <stgraber> ;)
[00:18] <cjwatson> (but it's true that the balance in LP is necessarily shifted more towards using the test suite, since it's so much less work to engineer a reproduction environment that way than by bothering to actually run a full instance)
[01:11] <slangasek> cjwatson: do you have time to review the mp I just sent for that buG?
[01:13] <cjwatson> urr, pretty confused by it being committed to lp:ubuntu/precise-proposed/aptdaemon and thus containing a bunch of other stuff
[01:13] <slangasek> hmm
[01:13] <slangasek> gar, not at all what I told it to do
[01:13] <slangasek> and I suppose uncommitting will simply anger the branch god
[01:14] <cjwatson> but looking at the diff
[01:15] <slangasek> oh, I see, I tried to create a branch under a 'precise-proposed' directory and UDD laughed at me
[01:15] <slangasek> sigh
[01:17] <slangasek> http://paste.ubuntu.com/1140316/
[01:17] <slangasek> that's consistent
[01:18] <cjwatson> .../precise/...
[01:18] <cjwatson> compare https://code.launchpad.net/~ubuntu-branches/ubuntu/precise/aptdaemon/precise-proposed/+merge/119222
[01:18] <cjwatson> i.e. the third bit is a series not a suite
[01:19] <cjwatson> anyway, approve review sent with a couple of minor notes
[01:19] <slangasek> right, so the fact that 'lp:ubuntu/precise-proposed/aptdaemon' works is entirely because someone wants to watch me suffer ;)
[01:20] <slangasek> thanks for the review
[01:20] <slangasek> -p ab> heh, that was my next commit
[01:22] <stgraber> slangasek: upgrade completed, no error and a working unity desktop post-reboot
[01:22] <slangasek> stgraber: huzzah
[01:23] <stgraber> hopefully we'll know if we broke anyhing else in the process when we get some jenkins run with -proposed on Monday ;)
[01:23] <slangasek> indeed
[01:44] <cjwatson> and accepted
[02:05] <slangasek> cjwatson: the regressions/ subdir doesn't seem to be processed at build time by the testsuite in quantal; is it expected to be?
[02:06] <slangasek> hmm, language-selector was marked for python3 but the dbus backend isn't actually compatible at all :/
[02:11] <slangasek> cjwatson: heh, and in fact test_lp768691 doesn't run under python3 at all (either the old or new test) because there's no such thing as 'unicode'
[02:16] <cjwatson> slangasek: I don't know
[02:16] <cjwatson> That would be a glatzor question
[02:16] <slangasek> fair'nuff
[02:46] <slangasek> ok, that's all the aptdaemon UnicodeDecode crashes that errors knows about, phew
[18:08] <slangasek> ev: so, that strange dip in all the crash graphs on 5/17... when did the retracer backlog get cleared?
[18:08] <ev> slangasek: I don't think it has to do with the retracers, as they're not involved for python crashes, which are also showing the dip
[18:09] <slangasek> ok
[18:09] <slangasek> interesting
[18:09] <ev> the retracers are a whole 'nother ball of pain for me, actually:
[18:10] <ev> ugh, damn open id
[18:10] <ev> the link I was going to paste isn't working
[18:10] <slangasek> it's also possible that many of these crashes hit users more than 3 times per week, so a week after upgrade they've stopped reporting
[18:10] <slangasek> but I wouldn't expect it to be quite so sharp, or quite so universal, in that case
[18:11] <ev> yeah
[18:11] <slangasek> anyway, just throwing out the thought... back to cleaning house now :)
[18:11] <ev> at least the lines aren't exactly the same
[18:11] <ev> it's not like it's just somehow sourcing the same data for all of them
[18:11] <ev> but there's something special about the 17th
[18:11] <ev> and I have no idea what it is yet :-/
[18:11] <ev> I looked for apport uploads around that time
[18:11] <ev> but there really isn't anything that lines up
[18:12] <ev> and I suspect we'd see more of a gradual drop off if that broke it