[10:51] * gmb lunches [12:01] bac, fwiw, I am in the middle of QA for your bug fix [12:01] bac, successfully got epiphany-browser (universe) packages imported for oneiric, need to check that they still don't get uploaded for natty even with XS-Ubuntu-Langpack: yes set [12:10] danilos: i just read through the backlog on #launchpad-dev. thanks for helping out. [12:10] danilos: you still trying to ensure natty fails? [12:10] bac, np, yeah [12:17] bac, btw, that price you quote for driving 450 miles, is it for the petrol only, or tolls and such as well? because it looks huge (like less than 10 mi/gallon), and I wonder if we can see that tank you're driving :P [12:17] danilos: that is the allowed reimbursement as set by the Federal IRS. [12:17] bac, ah, ok [12:17] it is supposed to be inclusive of all costs for driving a private vehicle: petrol, wear and tear, insurance [12:18] bac, right, understood [12:18] most people find it very generous, especially if you have an efficient car [12:18] bac, I was just wondering if you really have a car that can eat that much gas :) [12:18] i do, but would not be driving it down! :) [12:18] s/gas/petrol/ [12:18] bac, heh, fair enough [12:18] it is for taking the dog to the park [12:19] and, yes, i'd be happy to get 10mpg out of it [12:21] heh, nice :) [12:21] bac, what car is it? we don't get to see many cars like that in here [12:22] danilos: http://www.flickr.com/photos/squishy/2289984572/ [12:25] bac, that's a defender? [12:26] no, pre-defender. a 1977 Series III. they are very closely related but this one has a teeny 4 cylinder [12:28] bac, ah, interesting [12:29] bac benji danilos gmb call in 1 or 2 [12:29] Righto [13:05] bac, hi, bigjools is stuck in something else right now, and I was unable to QA the natty package translations non-extraction (though ideally, we'd also QA that it keeps working for pre-oneiric main packages and such) [13:09] bac, I am unsure how much I'd be able to help further, but basically, we need https://dogfood.launchpad.net/ubuntu/+source/epiphany-browser/2.30.6-1ubuntu6/+build/2492859 to start building (that's where we are stuck at now, and until bigjools can help, there's nothing we can do), when it completes, there should either not be .translations.tar.gz tarball at https://dogfood.launchpad.net/ubuntu/oneiric/+queue for epiphany-browser, or [13:09] if it's there, when it's accepted (only bigjools can do that) and processed, no files should up on https://translations.dogfood.launchpad.net/ubuntu/natty/+source/epiphany-browser/+imports [13:09] bac, if you feel lucky, you may also decide that what QA I did is sufficient, because the new functionality to let stuff through indeed works [13:10] benji, hey, you mentioned how you'd like to talk to me :) [13:11] danilos: I just pushed lp:~benji/launchpad/bug-734765 for you to look at [13:12] danilos: and here are the important SQL queries executed when rendering https://translations.launchpad.dev/ubuntu/hoary/+templates on a clean dev instance: http://paste.ubuntu.com/653099/ [13:16] benji, doesn't look bad at all, so only new queries are for fetching other-side POTemplate records [13:17] danilos: exactly, but if there are 1300 templates, then that means (potentially) 1300 new queries to render the ubuntu page [13:20] benji, yeah, so you could still pre-fetch those with another left join in the similar way thus turning it all into a single query [13:21] benji, you'd need a ClassAlias to use POTemplate in the same query twice, but that's not a big deal [13:21] well, that's where I fell down, I can't figure out how to do that, because I've already used... [13:21] ClassAlias is new to me, looking that up [13:22] ooh! that looks perfect [13:22] hope is restored ;) [13:22] benji, you might also decide to inner join SourcePackageName while at it thus really making the entire page content fetched with a single query :) [13:23] danilos: ok, thanks for your help. i'll work with julian to get it moving [13:23] bac, you are welcome, I am sorry I didn't get it completely done [13:23] yeah, I've been thinking about that; once I make it no worse than the original I'll see if I can make it a bit better without too much more effort [13:23] danilos: np, you got it rolling since i didn't know which packages were appropriate [13:30] gary_poster: Is there any procedure for creating a lazr.restful release other than JFDI? [13:30] gmb, sadly yes :-) lemme see if I can find it [13:30] Boo [13:32] I think there is something specific to lazr.restful, but it's what I've found so far... [13:32] https://dev.launchpad.net/ReleaseChecklist [13:32] still lookig [13:33] gary_poster: is this what you're looking for? https://dev.launchpad.net/HackingLazrLibraries [13:34] specifivally https://dev.launchpad.net/HackingLazrLibraries#Releases [13:34] benji, yes! thanks [13:34] gmb ^^^ [13:34] gary_poster, benji Thanks. [13:34] yw [13:34] np [13:37] gary_poster, benji So, this seems to count as a feature change. Should it therefore go to 0.19? [13:37] s/it/the version number [13:38] gmb, yes [13:38] Ok [13:38] version numbers are cheap (at least if you buy them from me) [13:39] well...I can match his price [13:39] I'll beat whatever benji's offering! [13:39] heh [13:39] * danilos shuts up [13:39] :-) [13:48] Well, I'll go do something else whilst that builds, since I've turned off site-packages... [14:03] gary_poster: JSON cache card unblocked and in review. \o/ [14:03] bac, yay! [14:07] danilos, small branch if you are willing and able: https://code.launchpad.net/~gary/launchpad/bug791492/+merge/69469 [14:21] AAAAAAAAAAAH [14:21] Failing tests in trunk. [14:21] Nothing to do with me so far as I can tell... [14:21] * gmb digs [14:21] :-( [14:22] Hmm. [14:22] 2 tests fail when run all in one bundle, pass in isolation. [14:22] gmb, if you used a system Python, that may be biting you. some lazr packages could not be easilu upgraded to the buildout that handles system pythons cleanly (and that code is being ripped out in trunk because I've not been willing to maintain it) [14:23] gary_poster: Ah. I just used bin/test... I thought that used bin/py... [14:23] gmb, it does, effectively, but bin/py is just a cover around the underlying python [14:24] OIC. [14:24] with the new buildout, that cover is careful like virtualenv is [14:24] gary_poster: Right. So, these tests pass when run on their own (or together, just without everything else) [14:25] Not sure if that's likely to be a manifestation of what you described above. [14:25] gmb, oh, no, that sounds like unclean tests then :-/ [14:25] Right. [14:26] The tests are both doctests under example/base/tests/ [14:28] gary_poster: They both fail in the same way (to start with, anyway): http://pastebin.ubuntu.com/653141/ [14:29] Oh, hang on. [14:29] I might be havening a stupid here... [14:29] :-) [14:31] gary_poster, looking (sorry, dropped out for some food) [14:31] how dare you! [14:34] gary_poster, sorry, it won't happen again [14:35] That's what I like to see: abject subservience. Oh yeah. [14:36] gary_poster, anyway, your branch looks good, I just don't like the hard-coded 5 for the mailman xmlrpc timeout, but if you feel that's ok (and shouldn't be a config option or some such), go for it :) [14:36] danilos, eh, good point. I don't even know for sure that mailman has access to the config... [14:37] I can dig around for that I guess [14:37] If I can see mailman being configured in some way or other I can try to follow the existing pattern [14:37] gary_poster, excellent, thanks, it's otherwise r=me [14:37] heh, mostly I see configuration via monkeypatches, of which this is one :-) [14:37] cool thanks [14:37] thanks for the lint fixes as well [14:39] cool [14:52] danilos, for your info and mild amusement, there is config approach and it is *really* gross (because of the "Mailman was not designed to be a library" thing) :-) . I'm running with it. [14:52] gary_poster, sounds "comforting" ;) [14:52] heh [15:09] gary_poster: So, after much faffing I can confirm the same failures in lazr.restful 0.18.1 (the current version). I can't seem to find the root cause, though. [15:10] It doesn't seem to be a test cleanliness problem [15:10] :-( [15:10] gmb, nor a clean python problem? [15:10] gary_poster: I don't think so, but I don't know what that would look like (I haven't got a virtualenv working properly yet; working on that now) [15:11] gmb, virtualenv won't help. Well, maybe it would. Heck, I've forgotten. :-/ [15:11] Argh. [15:11] A clean-built Python from python.org is definitively clean. [15:11] that said, benji or I should probably help you [15:12] I'd like to finish up the review-response I'm working on [15:12] then I can try to dupe and dig with you [15:12] unless benji wants to distract himself :-) [15:12] gary_poster, that would be perfect, thanks :) [15:12] (My next step was to ask for someone to try and dupe this) [15:13] lunch is distracting me perfectly well at the moment, but after that I can help [15:15] :-) [15:15] k [15:38] * danilos -> out [15:42] gmb: anything I can help with? [15:42] benji: Well, if you can grab the tip of lazr.restful trunk, build it, run bin/test and see if it fails, that would be a great start [15:43] it would... is something interfearing with that? ferral marsupials maybe? [15:44] benji: Sadly, no. That would at least be fuzzy in a fun way. This is just fuzzy in a I-can't-figure-out-why-it-breaks-yet way. [15:44] Confirmation that it's not just me would be nice :) [15:52] ah poop [15:53] I had an interruption when CHR was supposed to start [15:53] then I conveniently forgot it [15:53] I'll do a bit now and a bit after lunch [15:54] gmb: since this is a fresh OS install, my lazr.restful buildout is still chugging along. I'll let you know how the tests fare when it gets done. [15:54] benji: Cool. It took ages for me too, so that wasn't a surprise :) [15:55] gmb: if you have at .buildout/default.cfg file that looks something like this: http://paste.ubuntu.com/653190/ then it will be faster next time [15:55] gmb: tests done: all passed [15:55] benji: Right. It's my environment then. [15:56] That's a relief. [15:56] gary_poster: So, with benji's help, I'm satisified this is an environmental weirdness. [15:56] yay! [15:56] And so will press on... [15:56] I used Python 2.7.1, hopefully the results don't vary with Python version. [15:56] bleh :-( [15:57] * gary_poster decides they do not :-) [15:57] heh [15:57] Sssh. [15:57] The tests pass as long as you run them in the right way. [15:57] The first three words are where the emphasis lies. [15:57] rerunning with 2.6.5 just to see [15:58] yow, 2.6.5 doesn't look good [15:59] Damn. [15:59] benji: Is it doctests in examples/base that are failing? [16:00] Unicode error in field.txt looks like the root cause, there may be others. [16:00] ?!?!? [16:00] Haven't seen that one. [16:00] no error in examples/base [16:00] hostedfile.txt and representation-cache.txt fail for me. [16:02] yep, the lazr.restful trunk is definately broken under 2.6.5 [16:03] Okay. [16:03] So, not just me then. [16:06] benji: Oh, I see the field.txt error now. I was missing it in all the other errors. [16:20] benji, gary_poster: So, um... What's the best way to proceed from here. I don't see any error in the tests themselves, and whilst it's likely because I'm running out of brainjuice I'm unsure of the best way to fix the problem. [16:21] That said, 2.6.5 is the system python on Lucid, so we could really do with not having this issue... [16:22] the unicode error looks like a bug in Python to me -- which would explain why 2.7.1 doesn't fail in that way. The NameError in hostedfile.txt looks really wierd, but I haven't looked closely at it either. [16:24] benji: Yeah, that one's been confusing me, but I put that down to not knowing much about lazr.restful. [16:24] suprisingly enough, lazr.restful still needs variables to be defined before they're used [16:25] benji: Well, yes. But given that the test passes when run in isolation, I figured it wasn't that simple. [16:25] really? that's an interesting clue [16:26] bin/test -ct hostedfile.txt -t representation-cache.txt [16:26] passes. [16:28] Ah. [16:28] benji: `bin/test -ct example` fails. [16:28] Running the tests directly (and individually) passes. [16:28] Oooh! [16:29] Ah [16:29] benji: So, I missed out field.txt. [16:29] If you include that, the whole thing falls apart. [16:29] So this is likely an isolation problem on top of that unicode error, no? [16:29] well, at least running them indepenently works, that means that they don't depend on each other (but instead interfere with each other) [16:29] * benji needs an IRC client with spell check. [16:30] yep, isolation indeed [16:37] gary_poster: Fun and games at lazr.restful central, then ^^ [16:37] How do you want me to proceed here? [16:45] * gary_poster was dealing with minor child emergency, sorry [16:45] * gary_poster reads traceback, gmb [16:46] gary_poster: child > our fussy library problems; no worries. [16:46] :-) [16:47] ok, just thinking about guiding principles here... [16:48] you could argue that we shouldn't make an official release that is known broken [16:48] you coudl counter argue that it works in 2.7 [16:48] but I'm inclined to say that the principle wins [16:48] if we take that as a given, [16:48] then we are left with two options [16:49] make a local release within LP [16:49] and figure out the tests to make a real release. [16:49] s/and/or/ [16:50] The second would be safer [16:50] and preferrable [16:50] preferable even [16:51] but the first option is not inconceivable. gmb, how about this. I expect your EoD is within an hour or so, yes? Send me the branch and I'll give it a whirl [16:51] this afternoon [16:52] I finished my branch and was going to go bug hunting [16:52] gary_poster: It's actually in lp:lazr.restful. AFAICT it's been broken since at least r189 (my commits are 190 and 191). [16:52] If I don't get it working this afternoon, then your tomorrow morning make a local Launchpad release [16:52] and move on [16:52] gary_poster: That sounds fine. [16:53] ok cool gmb [16:53] My EoD is actually in ten minutes :) [16:53] heh ok [16:53] gary_poster: r189 being broken suggests that we made a release that was broken - 0.18.1 [16:53] (I suspect that we only caught this because my dev machine is still running Lucid) [16:53] if it was not already broken before that, yeah [16:53] Otherwise I would never have spotted the problem. [16:53] ah! [16:53] interesting [16:54] ok [16:54] gary_poster: The field.txt failure seems to cause the other two. [16:54] * gary_poster is going to have lunch [16:54] bin/test -ct field.txt -t hostedfile.txt -t representation-cache.txt fails horribly. Dropping field.txt removes the problem. [16:55] gary_poster: Okay. I'm EoDing shortly but will be on IRC for a while yet, so feel free to ping me if you've any questions. [17:00] cool [17:00] thanks [17:46] my schedule: [17:46] in 15 minutes [17:46] return from lunch [17:46] then do CHR [17:46] for an hour [17:47] then team lead call for an hour [17:47] then I have half an hour before my EoD [17:47] IOW, it strikes me I don't have a lot of lazr.restful time. But I'll try to squeeze it in... [17:54] danilos: i know you're not here, but https://dogfood.launchpad.net/ubuntu/+source/epiphany-browser/2.30.6-1ubuntu6/+build/2492859 completed successfully [17:54] and nothing showed up at https://translations.dogfood.launchpad.net/ubuntu/natty/+source/epiphany-browser/+imports [17:55] so i think this means the test is working as desired! i'm marking it qa-ok. thanks a ton for your help [20:02] what did you mean, benji? [20:02] there was only one open question and you addressed it [20:04] bac: the bug import? That was indeed a question, but I got to it from the "bug import questions" link, I expect that there are other open questions [20:04] oh [20:04] yeah, I see lots of open questions [20:04] i see you're right [20:04] i have all of the links as bookmarks in a tab and then open them all at once [20:04] i just saw it was a Question [20:05] i'll get to the others [20:23] benji: fwiw, that question about plural forms in Interlingua could be addressed here: https://translations.launchpad.net/+languages/ia [20:24] thanks; is that something we can put in the FAQ thing so noobs like me can benefit? [20:50] benji: i reckon. i didn't go right to it but had to search around a bit. [20:50] benji: i suspect it comes up so rarely it isn't that important [20:50] other than esperanto and interlingua they aren't inventing that new languages that often [20:51] I was thinking more for the use of the CHRer; we need a sane way of figuring out what to do with all the disperate requests we get. [20:51] benji: i agree. i'd just guess that plural forms is pretty well covered now for most languages [20:51] i've never done one before... [21:32] gmb, pushed fix, sent you an email with explanation [21:32] night all [21:45] danilos: there are several open questions related to translations that i cannot answer. could you try to handle them during your next CHR? [21:45] https://answers.launchpad.net/launchpad-project/+questions?field.search_text=&field.sort=RELEVANCY&field.sort-empty-marker=1&field.actions.search=Search&field.language=en&field.language-empty-marker=1&field.status=OPEN&field.status-empty-marker=1