[00:45] <maxb> https://dev.launchpad.net/DetailedCoverLetterTemplate is quite daunting
[00:45] <maxb> does it really all apply for a 3-line change?
[00:46] <maxb> e.g. http://bazaar.launchpad.net/~maxb/launchpad/py2.5-buildmailman/revision/9329
[00:55] <mthaddon> maxb: I imagine (but am definitely not really that qualified) that you'd want to include all those section, but if they're not relevant, very briefly explain why
[01:26] <jml> maxb, I almost never follow the template & I've yet to get in trouble :)
[01:26] <maxb> :-)
[01:26] <jml> maxb, I find it helpful to think of the headings as a checklist
[01:27] <jml> maxb, so that when I'm writing a cover letter, I think "Have I explained the implementation? Have I mentioned how to do QA?" etc.
[01:28] <jml> flacoste, do you know how to submit a branch to launchpadlib?
[01:28]  * jml keeps getting rejected by PQM
[01:29] <wgrant> jml: Merge to trunk!
[01:29] <wgrant> It's not PQM-owned.
[01:30] <jml> wgrant, ahh, ok
[01:30] <jml> my local trunk branch is branched from pqm
[01:30] <jml> hmm
[01:30]  * jml corrects
[01:30] <wgrant> jml: Is there a DEV_SERVICE_ROOT yet?
[01:30] <jml> wgrant, no, how about I sneak one in with this change
[01:31] <wgrant> jml: Thanks.
[01:35] <jml> hmm
[01:35] <jml> we should set lp:launchpad to append_revisions_only
[01:35] <wgrant> lp:launchpadlib, you mean?
[01:35] <jml> no :)
[01:35] <jml> although that too
[01:36] <wgrant> But lp:launchpad isn't pushable to.
[01:36] <jml> I almost typed 'bzr push lp:launchpad' from my launchpadlib branch.
[01:36] <wgrant> So why does it matter?
[01:36] <jml> wgrant, it is by me :)
[01:36] <wgrant> Oh.
[01:36] <wgrant> Of course.
[01:36] <jml> I don't want permissions to do so, but I have them.
[01:36] <wgrant> Wasn't that meant to go away?
[01:37] <jml> I couldn't persuade thumper to do so.
[01:40] <jml> well, not in the short term
[11:18] <maxb> When a test reports failure, without any error messages whatsoever, how do you debug?
[11:26] <wgrant> maxb: Pastebin?
[11:27] <maxb> preparing
[11:29] <maxb> http://paste.ubuntu.com/265476/
[11:30] <wgrant> maxb: Odd. But my guess is that you don't have the right python-sqlite.
[11:30] <maxb> hmm
[11:30] <wgrant> Or sqlite2 or whatever it is.
[11:31] <maxb> I seem to have both
[11:31] <wgrant> That's a pretty wild guess, though.
[11:31] <wgrant> Does it say anything more if you run just that test with lots of '-v's?
[11:40] <maxb> no :-(
[11:41] <wgrant> maxb: Does -D drop you into anything useful?
[11:44] <maxb>     self.assertFalse(broken_builder.builderok)
[11:44] <maxb> hm
[11:46] <maxb> eww
[11:46] <maxb> we really ought to disallow assertions without messages, or make them show a traceback
[11:48] <maxb> It's throwing a twisted.trial.unittest.FailTest and that's disappearing into the voids
[11:48] <maxb> *void
[11:49] <wgrant> Ah, yuck.
[11:50] <wgrant> Ah, that's from the top failure which I had no idea about.
[11:50] <wgrant> In what environment is this failing?
[11:51] <maxb>  karmic/py2.4 + a few innocuous patches which I can't believe are related
[11:51] <maxb> but I'll revert them anyway and try again
[11:51] <wgrant> Hmm
[11:51]  * wgrant tests.
[11:51] <maxb> on lp:launchpad/stable
[11:54] <wgrant> maxb: Um, it fails here too, but doesn't contribute to the failure counts.
[11:55] <maxb> !
[11:55] <wgrant> How odd.
[12:03] <wgrant> maxb: Ah. Sampledata change.
[12:03] <maxb> ooh
[12:03] <maxb> right
[12:04] <wgrant> maxb: But not in the direction that I would expect...
[12:04] <maxb> but....... doesn't make check reload it all every time?
[12:04] <wgrant> Yes.
[12:04] <wgrant> There's something strange here.
[12:04] <wgrant> The test should fail.
[12:04] <wgrant> And I suspect buildbot logs will say it does.
[12:05] <wgrant> But it's not counting in the totals.
[12:05] <wgrant> This is very very bad, or we are very confusing.
[12:05] <wgrant> (the change was to make 'bob' in the ftest sampledata OK, so this test should obviously faily)
[12:05] <wgrant> But if it's in stable, the test never failed.
[12:05] <wgrant> But we know it did.
[12:06] <maxb> unpossible! :-)
[12:06] <wgrant> And:
[12:06] <wgrant> Failure in test testScanRescuesJobFromBrokenBuilder (lp.buildmaster.tests.test_manager.TestBuilddManagerScan)
[12:06] <wgrant>  testScanUpdatesBuildingJobs (lp.buildmaster.tests.test_manager.TestBuilddManagerScan) (1.593 s)
[12:06] <wgrant> That makes me very scared.
[12:06] <wgrant>   Ran 4 tests with 0 failures and 0 errors in 9.342 seconds.
[12:08]  * wgrant retries the test with old and new clean sample data.
[12:29]  * maxb kicks off a testrun of devel/jaunty/py2.4 and goes out shopping
[12:29] <wgrant> security.py is not being friendly.
[12:31] <maxb> and after shopping, I shall get stuck into writing merge cover letters
[12:32] <wgrant> Fun fun.
[12:51] <wgrant> maxb: So, that test has been failing. I thought to check an ec2test log from a couple of days ago, and it's shown there (the only failure).
[12:51] <wgrant> But the test suite completes successfully.
[12:51] <maxb> riiigh
[12:51] <maxb> t
[12:51]  * wgrant files a bug.
[12:51] <wgrant> Two bugs
[12:52] <maxb> well, that sucks, and I have more weird failures too
[12:52] <maxb> For example a couple of launchpadlib-1.5.1's doctests are exploding because I'm getting bytestrings and the doctest expects unicode strings
[12:52] <wgrant> Same.
[14:00] <james_w> maxb: I landed a fix for that a few weeks ago I thought
[14:05] <james_w> ah, that was wadllib
[14:05] <james_w> http://bazaar.launchpad.net/~lazr-developers/wadllib/trunk/revision/11
[14:05] <james_w> might help though
[22:30] <maxb> Data point: running make check on an AAO takes 9¼ hours
[23:19] <wgrant> maxb: *Ouch.*