[00:22] mwhudson, wgrant: unfortunately I'm already past 5000 revisions :-/ [00:23] jelmer: I just checked IRC logs, and it was between r3000 and r4000. [00:24] wgrant: Thanks [00:24] So I wonder what's different between Michaels' setup and mine [00:25] jelmer: Did you import 5000 revs, or up to r5000? [00:25] ah, good point [00:25] I imported 5000 revisions [00:25] r1008 [00:25] so there's still "hope" [00:26] I'm not sure which Michael meant. I guess there's one easy way that will probably find out. [01:37] mwhudson: Hmm, it definitely seems to slow down over time [01:37] mwhudson: but no failures yet. [01:41] mwhudson, reproduced [01:42] Yay. [01:43] strangely enough I only seem to be able to reproduce it on fetches from remote hosts [01:43] and it's related to encodings in commit messages/authors [01:50] Aha. [02:05] The issue only crops up when we try to recreate the original git object from the data we have in Bazaar [02:06] We need to create the original git object again to use it as the delta base, but of course that's only necessary during a pull (during a clone or 'full' pull we already have all base texts available in git packs) [02:08] Ah, yes. So is the bzr revision bad, or just not being recreated properly? [02:23] it's just not being created properly - it strips off characters that are invalid for the expected encoding at the moment [02:23] so we lose a few characters [02:24] and then of course the sha1 of the newly created commit is different [02:26] Ah. [02:50] anybody around to review the patch to unbreak the merge from stable into db-devel? [13:16] mwhudson: I got to 18k revisions without problems [18:59] mwhudson, hi [19:04] jelmer: another hour perhaps [19:04] lifeless, Isn't he in NZ while you're in AU ? [19:04] yes [19:04] lifeless, or are you just up at a crazy hour? [19:04] I woke a little...early today [19:05] ah :-) [19:15] jelmer: hello [19:16] mwhudson: I just sent you an email. We should make sure to get a new version of bzr-git onto lp. [19:17] jelmer: subject? [19:17] jelmer: i guess it's r727 we want, right? [19:18] mwhudson: there's bzr-git in the subject, don't recall what it is exactly [19:18] mwhudson, yeah [19:18] jelmer: it doesn't look like i received it :/ [19:19] * jelmer flushes his mail queue [19:20] mwhudson, should be sent now [19:20] jelmer: it's interesting that bzr-git spends nearly 30% of it's time shoving things into sqlite [19:20] jelmer: i got your mail now :-) [19:22] jelmer: if a branch is suffering from this problem the branch is broken and needs to be re-imported from scratch, right? [19:23] mwhudson: yes [19:23] jelmer: btw, new testrepository is sexy :) and packaged in debian [in NEW] [19:23] lifeless, ah, awesome. I need to give it a try [19:24] lifeless, I saw your addition of .testr.conf to bzr [19:24] jelmer: you could write a .testr.conf for lp :) [19:24] I might [19:25] \o/ [19:31] jelmer: out of curiosity, what do you think a sensible revision pull limit is for lp? [19:31] it's at 1000 now, which i think might be a bit low [19:31] i actually wonder if a time-based limit might almost make more sense [19:32] import revisions for an hour, then stop [19:32] or something [19:32] mwhudson: as long as you reschedule imports I don't think the limit is really important [19:32] jelmer: ok [19:33] anywhere between 1k and 10k seems reasonable [19:33] testing the kernel locally sure used up quite a lot of my bandwidth :-) [19:34] we should be getting some new code import slaves soon-ish btw [19:34] they will be much much better speced that the machines we currently have [19:34] *than [19:34] ouch, you have a limit on your bandwidth? [19:34] only a practical one [19:35] but it takes a lot longer here to get the kernel.org repo than it does the data centre machines, that's all i meant [19:37] ah [19:40] mwhudson: oh, we'll need a newer dulwich too [19:42] jelmer: oh [19:49] jelmer: you haven't rebased dulwich again have you? [19:51] mwhudson, no [19:51] hm [19:51] must have had old local branches then [19:56] oh great, someone broke buildbot? [19:57] say its not so [19:57] looks like r9055 [19:58] or maybe it's a merge integration thing [19:59] oh i see there's already list discussion about it [19:59] (it is an integration thing) [20:09] jelmer: fwiw the git import succeeded locally without the new dulwich [20:09] s/git/gedit/ [20:49] morning [20:50] mwhudson: when do you want to chat? [20:50] thumper: in a few minutes? [20:50] sure, just call [21:16] thumper: https://staging.launchpad.net/successful-updates.txt [21:43] jelmer: wow, that bzr-svn symlink thing is breaking so many imports [21:43] jelmer: is the fix landed already? [22:06] Argh. [22:06] ? [22:06] * wgrant rechristens Soyuz "who could possibly think that was a good idea?" [22:07] jelmer: as incentive for doing a .testr.conf [22:07] https://code.edge.launchpad.net/~lifeless/tribunal/testrepository/+merge/20321 [22:19] mwhudson: https://bugs.edge.launchpad.net/launchpad-code/+bug/297398 [22:19] Bug #297398: support password/passphrase authentication for bazaar [22:20] (bug #529710 was the cause of my horror) [22:52] * thumper taps his fingers waiting for the staging email to arrive in the imap box [22:52] mwhudson: how's the qa on the import changes going? [22:53] thumper: well [22:53] as in, it all seems to work [22:53] i'm not sure there's much more i can do without losa help [22:56] * thumper now waits for staging incoming email processor to kick in at noon [23:03] damn it [23:03] * thumper needs a losa too [23:03] mwhudson: I'm walking up to collect Maia, afk for a bit [23:04] thumper: ok [23:24] * mwhudson lunches