[08:59] zope configs are hard and weird [09:12] What's up? [09:13] trying to work out if you can inherit zope.conf [09:13] doesn't look like you can, so might need to do a bit of shuffling. [09:14] But also trying to work out if I need the settings that I'm trying to create (the ones) at all [09:16] Oh, those things, right [09:27] https://code.launchpad.net/~cjwatson/launchpad-buildd/+git/launchpad-buildd/+merge/383320 (Ensure that regex patterns with \-escapes are raw strings) [09:38] cjwatson: MP 383320 looks good to me at first glance, I'm tempted to clone and run unit tests, would u recommend, or would that turn into a time sync ? [09:41] *sink [09:49] Feel free [09:49] Though it's not very necessary since Jenkins will do that on merge [14:01] https://code.launchpad.net/~cjwatson/launchpad-buildd/+git/launchpad-buildd/+merge/383341 (Adjust X-LXD-mode header construction for Python 3) [14:05] Also I have a couple of backlogged code-gardening MPs that it would be good if people could look at: https://code.launchpad.net/~cjwatson/launchpad/+git/launchpad/+merge/382875 and https://code.launchpad.net/~cjwatson/launchpad/+git/launchpad/+merge/383046 [14:05] I'll take a look [14:14] (The latter has a very bulky follow-up MP, but I was planning to self-approve that if/when the prereq lands since it doesn't make sense to have people spend time on >4000 lines of more or less mechanical changes) [14:14] woo, this version works without having zope.app.server installed at all. Although tracelog then stops working, so we might need to have a replacement for that [14:19] tracelog is occasionally handy though I don't know how vital it is. If I had to trade tracelog for statsd (or similar) I'd certainly take statsd [14:20] might be possible to reintroduce it, atm the easiest way to smoke test stuff that's relying on ZServer functionality is to just uninstall it, but it's a dep for tracelog [14:20] Maybe. We don't necessarily have to emulate everything [14:20] so tracelog stops working at the same time [14:21] It's also probably a pretty simple piece of middleware [14:22] The page performance report in lp:lp-dev-utils does use the trace files [14:25] (which is visible at the bottom of http://lpqateam.canonical.com/) [14:25] yeah, I think it's safe to re-enable it eventually [14:25] well, less 'eventually' more, almost immeadiately, once I've fixed servers.py to stop importing and using zserver components [14:25] er [14:26] Right, possibly just reimplemented as simple middleware [14:26] I don't think I knew that site existed - how might I have discovered that site, because it's neat [14:26] Good question, not sure. I knew because the deployment report used to be there [14:28] How/where is this site generated/hosted/etc.? [14:28] I'd search wikis but my ISP is having a wobbly day [14:28] It's on carob, old-school ~lpqateam/public_html [14:28] Fab, thanks [14:29] Some is hand-rolled HTML, some is generated from bits of lp-dev-utils and similar, I think it's very much symbiosisware [14:29] But handy [14:29] Hm, not sure it's just you - I couldn't reach wiki.c.c! [14:30] It may well not be just me, I just knew of https://aastatus.net/34840 and there wasn't much point investigating further when DNS was timing out [14:31] Ah [14:31] ISP-buddy-five o/ [14:32] * SpecialK|Canon prods bits [15:16] https://code.launchpad.net/~cjwatson/launchpad-buildd/+git/launchpad-buildd/+merge/383344 (Treat build logs as binary files) [15:27] cjwatson: +1 [15:28] Thanks. Getting there! [15:29] About 500 lines of actual porting left and then some packaging stuff on top [15:30] Nice [15:37] pappacena: Your most recent comment on https://code.launchpad.net/~pappacena/launchpad/+git/launchpad/+merge/382779 mentions "with comments", but I don't see them - did you forget to save them? [15:39] uhm. I think there was only one comment, about the `if exact_match: return`, in the previously pushed revision. [15:40] Right, I was wondering about whether just removing the 'return rs.one()' was quite correct [15:40] This is the comment: https://www.irccloud.com/pastebin/zoiCdnDe/ [15:40] But just trying to think through it ... [15:41] Strange... the diff comments were actually not saved when I added the other comment. I just saved it now [15:41] And yes, I think you're right [15:44] r=me, just one test suggestion but go ahead after that [15:44] Ah, right! Indeed, a good suggestion. I'll do that, and land the MP [15:50] well, that was a bunch of test explosions... [16:00] https://code.launchpad.net/~cjwatson/launchpad-buildd/+git/launchpad-buildd/+merge/383345 (Treat build output files as binary files) [16:09] * pappacena reviewing [16:21] https://code.launchpad.net/~cjwatson/launchpad-buildd/+git/launchpad-buildd/+merge/383349 (Treat intltool-related files as binary files) [16:23] Total: 47 tests, 0 failures, 2 errors, 0 skipped in 3.310 seconds. (from test_servers) [16:23] that... went better than I expected [16:44] pappacena: Couple of tiny things on https://code.launchpad.net/~pappacena/turnip/+git/turnip/+merge/380451 (we should really get going on isort or equivalent ...) but otherwise good to land [16:53] * tomwardill launches a full test run [16:54] How long do they take for you? [16:58] not sure, not measured one for a while [16:58] although it just segfaulted... [16:58] (test:760044): Gtk-CRITICAL **: gtk_settings_get_for_screen: assertion 'GDK_IS_SCREEN (scree [16:58] n)' failed [17:02] ah, forgot to run xvfb [18:31] cjwatson, any known issue on Jenkins / Turnip test suite? Jenkins is failing some tests running git commands, and I cannot reproduce it locally. [19:28] pappacena: I don't recognise that one, and it's in code you changed, so hmm [19:31] pappacena: Ah, I see, look for the "Please tell me who you are" errors. The test container doesn't have a git identity configured [19:31] Yes, it is. I'll try to debug it. Sad that those tests passes locally (both on python2 and python3) [19:31] ahhm [19:32] Strange that it started failing now... [19:32] pappacena: Because your change switches to using git tag rather than pygit2 to create the tag [19:33] So rules for the git command line apply [19:34] Ah, yes! Right! I forgot about this change [19:37] pappacena: I think perhaps it'd be simplest to have RepoFactory.add_tag add '-c', 'author.name={}'.format(self.author.name), '-c', 'author.email={}'.format(self.author.email), '-c', 'committer.name={}'.format(self.committer.name), '-c', 'committer.email={}'.format(self.committer.email) [19:37] Since it's just that one command at the moment, and could be generalised later if we start using the git command line for more things in the test suite [19:38] I was checking the code, and got to the same conclusion. I'll just put those extra parameters in another method, to make it reusable when needed... [19:38] Sounds good, thanks [19:39] Thanks for the help! [19:39] It's a shame that the test runner there doesn't gather the stdout/stderr with the ERROR: output [19:40] Since then it would've been a lot more obvious [19:40] I suppose that's what we get for just using unittest discover though [19:40] Let me see if I can do something about it too. [19:41] Don't feel obliged, but it would be a good opportunity :) [20:22] cjwatson: quick review (for tomorrow?) https://code.launchpad.net/~pappacena/turnip/+git/turnip/+merge/380451 [20:22] Added some logging when the command fails... [20:26] pappacena: Looks OK if it works, go ahead. (Though I normally prefer self.addDetail for this sort of thing - compare turnip.pack.tests.test_functional)