[00:10] I've cleaned everything but the eggs cache === beuno-afk is now known as beuno [00:17] gah, what? [00:18] devel works, my python2.5 branch does not [00:18] But there's nothing there that goes anywhere near lazr anything [00:19] Getting distribution for 'lazr.uri==1.0.1'. [00:19] Installing lazr.uri 1.0.1 [00:19] Caused installation of a distribution: [00:19] lazr.uri 1.0 [00:19] with a different version. [00:19] Got None. [00:19] While: [00:19] Installing. [00:19] Getting section filetemplates. [00:19] Initializing part filetemplates. [00:19] Error: There is a version conflict. [00:19] We already have: lazr.uri 1.0 [00:19] What on earth is buildout attempting to tell me? [00:25] oohkay, purging python-lazr-uri .deb made it work... [00:26] isn't buildout supposed to be insulating you from system python versions? [00:34] /home/maxb/launchpad/lp-branches/python2.5/lib/canonical/config/__init__.py:19: UserWarning: Module lazr was already imported from None, but /home/maxb/launchpad/lp-branches/python2.5/lib is being added to sys.path [00:34] !? [00:45] maxb: there was a fix to lazr.config today I believe [00:50] * maxb observes that the zope sourcedep still isn't completely obsolete, if the symlinks are to be believed [00:57] OK after cleaning absolutely everything, now I can't "make" devel eitehr [00:57] either [00:57] :( [00:57] my make is failing too in mailman [00:58] heh, at least you got that far [00:58] mwhudson: hey, build engineer [00:59] thumper: as of monday! :) [00:59] *next* monday [00:59] :) [00:59] does your make work? [00:59] dunno [00:59] running pull now [01:00] i just read something about make in mailman failing from last week, francis said he'd fixed it though [01:01] mwhudson, hello [01:02] thumper, I was just looking at https://bugs.edge.launchpad.net/launchpad-code/+bug/418290 [01:02] Bug #418290: Person's code facet has unclear section linking to the teams the person is a member of [01:02] jml: yeah [01:02] jml: hello [01:02] File "/home/tim/src/lp/devel/eggs/zope.proxy-3.5.0-py2.4-linux-i686.egg/zope/proxy/_zope_proxy_proxy.py", line 6, in __bootstrap__ [01:02] imp.load_dynamic(__name__,__file__) [01:02] ImportError: /home/tim/src/lp/devel/eggs/zope.proxy-3.5.0-py2.4-linux-i686.egg/zope/proxy/_zope_proxy_proxy.so: cannot open shared object file: No such file or directory [01:02] thumper, can we simply fix the links to link to the code listing? [01:03] jml: there was a "fix" recently that changed the fmt:link to always go to the root site [01:03] jml: we may be able to get it to go to the code site [01:03] jml: I've not yet investigated [01:03] jml: I think it is just team/fmt:link/code [01:03] (or branches) [01:04] So, what is buildout *supposed* to do if I have lazr.uri 1.0 installed as a .deb but versions.cfg says 1.0.1 ? [01:04] it is a bit of a mixup [01:04] thumper, yeah, I think the 'fix' is a good idea [01:04] maxb: look in your download-cache [01:04] thumper, but it's different from forbidding links to particular facet pages :) [01:05] thumper, and it seems that linking to the team branch listings actually is desirable [01:05] mwhudson: Hmm. it appears to be throwing its hands up in horror and bleating incomprehensible error messages instead [01:05] it is there [01:05] maxb: i'm not completely surprised [01:05] right. [01:05] jml: the bug was that normally a link to a person or team should take you to the overivew [01:05] jml: but in this case we should link to listings, and probably give it a better title too [01:05] s/title/heading [01:06] thumper, I was about to file a bug about the heading, actually :) [01:06] jml: file away and assign to me [01:06] thumper, will do. [01:08] grrr [01:08] how can I code if I can't run the tests [01:08] a rhetorical question [01:09] * jml thinks about the new project page [01:14] leonardr, hello? [01:16] leonardr, I have three simple questions for you, if you have a moment. [01:32] is spm back today? [01:32] aye [01:34] spm, welcome back :) [01:34] spm, feeling better? [01:35] jml: yeah muchly. The joy of children and bringing home colds from school.... [01:35] heh [01:36] spm, my todo list says I have a couple of things to ask you about [01:36] spm, one of them being bazaar.edge.launchpad.net [01:41] maxb: I blew away the contents of the /eggs dir, and make clean build worked ok [01:53] * mwhudson lunches [01:58] jml: since i forgot to quit irc, you may ask your questions :) [01:58] leonardr, heh, thanks :) [01:59] leonardr, first, what should I call the constant for the production API service root. I've picked LPNET_SERVICE_ROOT for now, but STABLE_ and PROD_ might also fit. [01:59] leonardr, second, what should the URL be? I've got https://api.launchpad.net/beta/ now, but the 'beta' troubles me a little. [02:00] jml: i would go with PRODUCTION or STABLE [02:00] lunch time [02:00] leonardr, third, when I authorize an app for the production service, I am taken to an edge web page (probably because of the beta user redirect). Is this a bug? [02:00] look at the other constants and do what they do. i believe they also have /beta/ [02:01] leonardr, they also have beta. [02:01] leonardr, so I should follow what they do? [02:01] yes, the constants are going to behave like "HEAD" or "trunk" in a version control system [02:01] they will point to the most current version of the web service [02:01] which right now is /beta/ [02:01] * jml prefers 'PRODUCTION' to 'STABLE', since the latter might have API stability connotations. [02:01] sounds good [02:02] #3 is a bug -- you're in the launchpad beta testers team, so your requests to lpnet redirect you to beta [02:02] flacoste filed this bug a couple weeks back [02:02] s/beta/edge/ [02:02] cool. [02:02] leonardr, thanks for that. I'll land the patch with the new constant then. [02:03] great [02:04] Is the beta team restriction in place on the lpnet API vhost? [02:04] wgrant: yes, that's the bug [02:05] leonardr: Not the same bug... I mean, api.edge.launchpad.net is restricted to beta testers. Is api.launchpad.net? [02:05] ah [02:05] i don't know. are you sure that restriction is still in place? i thought we got rid of it [02:05] I don't know. [02:05] but if it's still in place on edge, it should still be in place on lpnet [02:05] It is certainly in place on edge. [02:06] I saw somebody complaining about it on identi.ca earlier in the week. [02:06] maybe that's another thing we should change [02:06] Restricting it on edge might be reasonable. [02:07] It's not restricted on edge. I used the api long before I joined the beta-testers team [02:08] Hm. [02:08] Then why was it crashing for a non-beta-tester, I wonder.. [02:10] Huh, you're right. It's not restricted on edge (or at least wasn't 1.5 months ago). === kiko is now known as kiko-zzz [02:17] huh. So the testsuite no longer runs unit tests first? [02:20] It did for me yesterday, but that was without the new zope.testing. [02:30] oh right. [02:30] new zope.testing. [02:31] * jml has work to do! [02:34] jml: What are you doing? [02:35] wgrant, enabling parallelization of the Launchpad test suite [02:35] Ooh. Excellent. [02:35] but also, paging the new zope.testing version into my brain [02:35] so I can work on making sure it's stdlib unittest compatible [02:35] (mostly so that lifeless's unit testing work can be applied to it) [02:38] subunit? [02:47] wgrant, subunit and testresources [02:47] and ideally, much of the work in bzrlib.tests and bzr-ec2test, but let's not get ahead of ourselves [02:59] Argh. 844 line diff :( [03:01] wgrant, most reviewers can be bribed to review diffs over the allowed size [03:02] particularly with code that is an improvement [03:03] Almost half of it is new tests. [03:05] thumper, beuno: Is https://bugs.edge.launchpad.net/launchpad-code/+bug/236442 much closer to being fixed now that we have spark lines? [03:05] Bug #236442: Show how active I've been [03:06] Well, you don't have sparklines any more... [03:06] jml: I killed them [03:07] jml: I intend to bring them back, but they need fixing [03:07] thumper, that seems perfectly fair to me. [03:07] thumper, I think that the bug I mentioned doesn't benefit from sparklines so much as actual graphs. [03:07] yeah [03:09] * thumper primal screams [03:09] mwhudson, just re-discovered https://bugs.edge.launchpad.net/launchpad-code/+bug/85326 [03:09] Bug #85326: Codehosting server should initiate a pull attempt

[03:09] thumper, wassup? [03:10] the query for person active reviews is an arse [03:10] we need two different base collections [03:10] and different queries [03:10] unioned [03:10] jml: "hmm" [03:15] mwhudson, it's apparently my oldest open bug on launchpad-code [03:18] jml: that's quite a surprise [03:21] mwhudson, yes. [03:29] bloop [03:32] thumper, when you've got time, I'd like to talk about Monty Taylor's post to launchpad-users [03:33] yeah, I'm actually trying to get work done right now :( [03:33] thumper, understood. [04:00] jml: i think, probably, 85326 should be wontfix/subsumed by our eventual message queue stuff [04:00] jml: in the fullness of time, the puller might not even run on the same machine as the ssh server [04:00] mwhudson, indeed. [04:00] mwhudson, but it could also be done before-hand. [04:01] jml: you mean as a temporary improvement? [04:01] mwhudson, yeah [04:01] well, yes it could [04:01] mwhudson, I don't think it would hinder any of the message queue work [04:01] and it wouldn't be too difficult a change, I think [04:02] jml: there's quite a lot of stuff in vaguely similar areas i'd rather do first, i think [04:02] (combine puller and scanner, maybe pre-exec-ing bzr lp-server processes) [04:02] maxb: I think you want http://lpdebs.canonical.com/{dapper,jaunty}/ [04:02] mwhudson, I'd combine the puller and scanner first for sure. [04:03] maxb: That has 0.11 to 0.35. [04:04] mwhudson, I think I'd probably trigger a pull before working on pre-exec, since I think waiting for a pull is a worse kind of waiting than waiting for a bzr lp-serve process to spawn [04:04] (also because I know how to trigger a pull :)) [04:04] jml: i'm interested to see how my/our puller rewrite works in practice [04:04] probably the memory issue trumps both of those [04:04] mwhudson, me too! [04:05] mwhudson, I guess it's waiting for a prod rollout? [04:05] it will reduce the latency before a pull starts to 10s in some circumstances [04:05] right [04:06] I have to confess I'd forgotten about that patch. [04:08] maxb: By 'jaunty' I of course mean 'hardy' === ursula_ is now known as Ursinha [05:40] mwhudson: (mwhudsondoyle?) ref bug 411250. Ha! it seemed to go away on it's own - perhaps thumper can enlighten? it "felt" like a -ve cache issue.... [05:40] Bug #411250: rewriter seems to negative cache for a long time [05:41] spm: there's enough moving parts for things to be able to get pretty confusing [05:41] spm: but unless it happens again, i intend to basically not think about this again :) [05:41] heh. and fair enough too. [05:42] fwiw, nothing quite like nagios generated sms's on a regular basis to encourage bug-reporting. :-D [05:52] I need a better title for: 'Reviews requested I can do' [05:53] because I can now review anything [05:53] what it really means is: [05:53] a team I'm a member of has been asked to review this proposal [05:53] but that's kinda wordy [06:02] hmm [06:02] I'm about to wander off and go and make dinner [06:02] coming back on line when the kids are in bed === thumper is now known as thumper-cooking [06:03] thumper-cooking: The key word in the existing title is 'requested'. === thumper-cooking is now known as thumper [06:03] stub: I have a different question for you [06:04] thumper-cooking: The existing title seems fine (although I think 'Requested reviews I can do' is better) [06:04] stub: we have constantly issues with the security.cfg [06:04] stub: I want to have some group defined for "bzr-identity" that has the select rights on all the right tables [06:04] stub: and have that group used in our other db users [06:04] stub: is there an example of how to set this up? [06:05] Check out garbo, garbo-hourly and garbo-daily - it will show you how inheritance works in that file. [06:05] stub: ok, ta [06:05] * thumper really leaves to cook now === thumper is now known as thumper-cooking [07:04] well i've now read all the mail that had arrived by the time i got back on tuesday morning [07:04] but not all the mail that's arrived since then :) === Ursinha is now known as Ursinha-afk [07:15] mwhudson, heh [07:28] Morning [07:39] hi [08:11] wgrant: oh great, that gets me all but 0.7 through 0.10 [08:12] yep. [08:17] lifeless: Care to cast your mind back to 2006 and speculate where launchpad-dependencies (0.8) dapper might have been kept? [08:17] in dapper [08:17] [yes, it was actually in the distro at one point] [08:17] 0.6 is the latest in the distro. [08:17] I located 0.11 to 0.35 this morning. [08:17] And >= 0.36 are in the PPA. [08:18] sorry, nothing springs to mind then [08:19] It also looks like 0.2 and 0.3 only ever lived in pre-Soyuz Dapper. [08:19] ah well, 4 missing versions is a lot better than 30ish [08:20] maxb, can you give me a file name that I could search for? [08:20] launchpad-dependencies_0.7.dsc, 0.8, 0.9, 0.10 [08:21] no joy. [08:21] It's not critical, there's info in debian/changelog [08:22] cool. [08:22] danilos, good morning [08:23] maxb: I will grab a recent Karmic live CD in the next couple of days and see if I can reproduce your failures in a really clean environment. === jtv1 is now known as jtv [08:40] meh, I have a couple of new failures, even on 2.4 [08:40] https://dev.launchpad.net/LaunchpadOnKarmic [08:42] maxb: How's the 2.5 run looking? [08:42] see the marker :-) [08:42] maxb: Ah, I see. [08:43] maxb, so few? [08:43] 2 consecutive test runs take longer than overnight :-( [08:43] jml: hmm? which? [08:44] maxb, after , there looks to be very few tests that are actually failing [08:44] I'm pleasantly surprised :) [08:44] ah, well after the marker, those are still the results from before zbuildout landed [08:45] But presumably zbuildout should make it *better*. [08:45] ohh, I see. [08:45] good morning [08:46] Oh, and the page doesn't take account of the fact I had to cheekily "pycentral pkgremove python-lazr-uri" to make the buildout work at all === thumper-cooking is now known as thumper [08:48] has the lazr packaging mess been sorted out yet? [08:48] (I guess your comment means "no", maxb) [08:48] buildout is the solution to everything *handwave* [08:49] Well, it certainly hadn't been sorted when I was attempting to get these test runs going before sleeping last night [08:49] * wgrant just up'ed devel and is about to try. [08:50] wgrant: dpkg -l python-lazr-uri ? [08:51] maxb: Not installed, thankfully. I must have an old launchpadlib. [08:51] test with stable [08:51] That is a strange package name. [08:51] lucky you! :-) [08:52] devel might well have test failures (who can say!) [08:56] jml: good point, well made :-) Though it's the buildout flat out failing to do what its raison d'etre is that's really annoying me. [09:00] * wgrant throws broken eggs at SHHH [09:02] Aha, it works. But zc.tracelog is being a bit loud. [09:08] Next meeting I intend to ask the peanut gallery to express opions on SHHH [09:08] and opinions [09:20] lifeless, push-pop-progress, right? [09:20] wgrant: I've just herded your structural subs branch off to ec2. I started a run last night but the instance crashed and burned some time this morning. [09:23] gmb: Thanks. [09:25] jml: yes [09:25] lifeless, done [09:30] g'night [09:30] Enjoy your evening :) [09:35] seeya jml, and congrats BTW [09:49] bigjools: on? [09:51] lifeless: staying alive? [09:52] oo oo oo oo [09:52] staying alive [09:52] staying alive [09:52] oo oo oo oo [09:52] :) === mpt__ is now known as mpt [10:02] morning [11:00] morning, all. [11:00] hi deryck ! [11:09] People seem to think that packaging is trivial and automatic... === henninge_ is now known as henninge [11:22] wgrant: it ought to be [11:22] Quickly will be a step in the right direction [11:23] bigjools: For simple apps, sure. [11:23] bigjools: But libraries, not so much. [11:24] .deb is way over-engineered for most stuff IMO [11:25] bigjools: what, in particular, do you think is over-engineered for packaging, say, a gnome game? [11:26] nothing in particular [11:26] There's nothing wrong with .deb for that. [11:26] The build system, perhaps. [11:26] bigjools: so what's "most stuff" then? [11:27] did you see jono's session at All Hands when he learned to package something? [11:28] yep [11:28] So you don't mean .deb. You mean the source package format. [11:28] probabl [11:28] y [11:28] and the number of people in the audience who argued over how to do something? [11:29] For setuptools or autotools applications, it's very easy with CDBS or dh7. [11:29] but then I don't have a great deal of experience packging stuff, I am probably talking out of my arse [11:29] Heh. [11:29] however, I think it could be easier [11:30] Before DH7 and CDBS were big, it was pretty awkward. [11:30] bigjools: yeah, so the process could be a lot more helpful - or *feel* a lot more intuitive with the right tools, ,but the biggest difficulty is that the problem-domain is complex [11:30] yep. [11:30] I think it's only as complex as you want to make it [11:30] bigjools: that was partly due to the nature of some of the people in the room :-) [11:30] james_w: no doubt :) [11:37] james_w: btw I owe you an apology, I was confusing you with someone else when I was talking about lpnet vs edge API usage, sorry :( [11:37] np [11:37] I'm just happy it's there :-) [11:37] yeah, I was surprised it wasn't from the start [11:44] bigjools: it doesn't count unless it's in an email (an apology that is :) [11:45] bigjools: btw, I was surprised it wasn't either, flacoste did it basically a day after we've decided to go with it, but it never went out [11:45] danilos: shaddap and do your job :) [11:47] * danilos crawls back into his corner [11:48] mars: btw, is conversions.html based on db-stable or something? if I run it over my local copy of devel, I get much better stats for translations :) [11:48] (also, conversions.html doesn't belong on devpad) [11:48] danilos: the report does not lie, you're slackers! [11:49] bigjools: heh, that much is true as well, but I've got the code from mars $HOME and tweaked it a bit so it displays good results for translations [11:49] lol [11:49] if project=translations: done=100% [11:50] wgrant: that was exactly what I was thinking while I was typing this, but Launchpad team is not used to having publicly accessible accounts [11:50] wgrant: I'll raise it with the team [11:50] bigjools: heh, exactly :) [11:51] geez, what is this zc.tracelog spam in my "make run" output now ... [11:51] bigjools: I wondered that... [11:51] bigjools: that's what wgrant loved yesterday, 9124 or 9224 or something :) [11:51] should be DEBUG not INFO [11:52] danilos: Note I also loathed it, though. [11:52] 'cause it's all eggs. [11:52] wgrant: heh, I know, I know [11:52] wgrant: I've got about the same feelings about eggs/buildout combination [12:12] adeuring: hey [12:12] hi danilos [12:12] adeuring: did you by any chance ask for some hwdb parsing runs on staging yesterday? [12:13] danilos: no. what's the problem? [12:13] adeuring: if you did, I got some logs in my email, but didn't get the logs I actually asked for :) [12:13] adeuring: ah, never mind then, it seems Chex ran the wrong script for me then [12:32] EdwinGrubbs, ping [12:48] beuno: ping [12:49] henninge, hi [12:50] beuno: did you notice this bug I assigned to you? bug 418610 [12:50] Bug #418610: New translate page needs new icons [12:50] henninge, I did not [12:51] beuno: do you think you can get something done with that before you're gone? [12:51] henninge, could you please attach a ascreenshot of where those icons would go? [12:51] beuno: sure === Ursinha-afk is now known as Ursinha [12:51] henninge, I will try [12:51] beuno: cool, thanks [13:03] wgrant: Your branch passed its tests and is now in the PQM queue. === matsubara-afk is now known as matsubara [13:04] gmb: Thanks. [13:04] np [13:05] * wgrant shall prepare the other two pieces tomorrow. === kiko-zzz is now known as kiko === abentley1 is now known as abentley === ursula_ is now known as Ursinha === abentley1 is now known as abentley === ahma_ is now known as ahma === salgado_ is now known as salgado [14:03] beuno: I attached the screen shot. [14:04] henninge, thanks === abentley1 is now known as abentley [14:37] barry, ping [14:41] deryck: pong [14:53] beuno: ping [14:54] barry, pong [14:54] beuno: hi, two things. first, can you come to the ameu reviewer's meeting in 6m? [14:55] beuno: second, i'd like to talk about what to do about link display on the page the link points to [14:57] reviewers, lurkers and beuno -> #launchpad-meeting in 3m [14:57] sure [15:00] barry, lets see about the second after the meeting [15:00] beuno: sounds good [15:00] reviewers, lurkers and beuno -> #launchpad-meeting [15:19] barry, sorry I wasn't very useful [15:19] was on a call :) [15:20] beuno: no, it was cool :) [15:21] kfogel: ping, not urgent [15:21] beuno: so, about menu links. currently, navmenus (and only those i think) hide links for the page that you are viewing. i don't like this :). i've heard that in breadcrumbs the link to the page your viewing is still shown, but it is dimmed and not clickable. i personally don't like that either (i'd rather the link be active but otherwise visually indicative). we're looking for some ui pronouncement on [15:21] and fix the tests [15:22] mrevell: pong placeholder -- on phone, but will ping when off [15:23] barry, could you tell me more about why you want them clickable? [15:24] beuno: i'm sure it's not a good reason, but i very often use it as a cheap reload. usually the active link is way closer to my cursor than that leeeettlleee teenie browser reload button way up there in the upper left [15:25] barry, I understand [15:25] the problem is, it may misslead you to think it will take you somewhere else [15:25] or "different" [15:25] beuno: do you think we could do something visual to indicate it's the page your one, but leave the link active? like grey it down a bunch or something? [15:26] barry, I'm thinking [15:26] I'm trying to weigh in the cost/benefit [15:26] "make barry happy" weighs quite a bit, but, OTOH, we have like 500k users... [15:27] beuno: :) okay, so here's the scenario... [15:28] beuno: i just clicked on "View all sprints" and i'm transported to /sprints/+all. now i kind of remember i'm on that page because i clicked it recently, but maybe someone pinged me in irc that they just added two new meetings. i need to reload the page. what are my options? (note this happens fairly often with merge proposals and bug comments) [15:29] that's a great scenario to address [15:29] now [15:29] barry, do you use gmail? [15:29] beuno: no [15:29] ok, what gmail does [15:29] as a web application [15:29] well, i have a gmail account, but... ;) [15:29] is that if something changes while your on the page [15:29] it tells you [15:29] or it just updates it [15:30] depending on the scenario [15:30] *that* is the proper solution [15:30]