[05:46] hey [05:46] anybody work with qt ? [07:34] david_0001: I don't, but I know zematynnad has been doing a bit of qt dev lately. He'll be around in 7-8hrs. [08:02] good morning all [09:43] noodles775: can I interest you in reviewing https://code.launchpad.net/~jml/pkgme-service/end-to-end-pkgme/+merge/83624 [09:45] jml: Indeed you can :) I'll just finish my changes for another review and start in 15mins. [09:45] * noodles775 claims it. [09:46] noodles775: thanks. Will probably also ask james_w to have a look over it when he turns up later today. [10:46] jml: `fab test` gives me ImportError: No module named twisted.internet - that should be in the setup.py right? [10:46] noodles775: yes. [10:47] noodles775: well, I guess. [10:47] it's a test dependency only [10:47] Why guess? [10:47] Ah, ok. [10:47] Right - then test_requirements.txt [10:47] Well, if that actually works. [10:47] s/test_requirements.txt/test-dependencies.txt/ but why wouldn't it work? [10:48] Ah, right, a pip install of twisted... yeah, let's see. [10:49] noodles775: Given the many hours I spent last week wrestling with this whole easy_install/pip/virtualenv set up, and given that we have hard requirements in test-dependencies, I'm sceptical of anything working just like that. [10:49] noodles775: maybe I'll be pleasantly surprised. [10:51] jml: I'm happy to just install it locally and we can update the README to say what packages need to be installed system-wide - if it doesn't work for you. [10:51] noodles775: cool. (incidentally, I didn't have to install locally. Wonder why.) We'll see. bootstrapping now. [10:58] jml: would `PKGME_LIB_DEPS_DB_PATH = 'all-symbols.db'` work (similar for PKGME_OUTPUT_DIRECTORY)? ie. can we assume they're relative to the settings.py file? (and add bzr ignores) [10:59] noodles775: well, they aren't going to be like that on production, probably [10:59] noodles775: but I guess those are better wrong values than the ones I have in there. [11:00] jml: sure, but on production we'll have a separate config. Yep. [11:07] noodles775: would those paths be relative to settings.py or relative to the root directory? [11:08] jml: hrm, not sure.. try it and see (probably the CWD - but you'll find out) [11:10] * noodles775 reads up on lifeless's fixtures package. [11:16] it's pretty straightforward. [11:16] Yep - nice. [11:18] How did you go with adding twisted? [11:19] I've probably not understood fixtures properly, but why are you using a fixture inside non-test code? (tasks.py) [11:20] noodles775: I didn't add Twisted. It just worked. [11:20] noodles775: Because fixtures aren't actually about testing. They are just most often useful there. [11:20] noodles775: I guess they are like souped-up context managers. [11:21] jml: RE fixtures - ok. Re twisted: Sure - but didn't it ust work because you *already* had it installed? [11:21] noodles775: well, on the system, obviously. [11:21] noodles775: tbh, I wasn't paying that much attention and there's been a lot of water under the bridge since then. Maybe I did install it in the virtualenv w/ pip. [11:22] * noodles775 tries adding it to the test dependencies. [11:22] it works then [11:23] Yep, `virtualenv/bin/pip install twisted` worked fine here.. great. [11:40] jml: with twisted in the virtualenv, I get six errors, all coming down to this: http://paste.ubuntu.com/755918/ [11:40] s/coming down to/like/ [11:41] noodles775: you're using Python 2.6 [11:42] noodles775: can we not use 2.7 in production? [11:52] jml: I guess we can - all our current deployment machines are lucid with py2.6, so that's what I use for my dev vm. [11:54] well, if it's going to cause a deployment hassle, let's not do that [11:54] easy enough to do try/finally [11:54] (and also to update 'fab bootstrap' to use python2.6) [11:55] Wow - I'd not realised python2.7 wasn't available in the lucid archive? [11:55] (python3 is) [11:56] noodles775: yeah [11:56] I want to go taunt barry now [11:56] heh [11:56] "It's nearly 2012 and I'm writing new code in Python 2.6" [11:56] XD [11:59] taunting done [11:59] I've also asked Canonical IS whether I can use Python 2.7 without causing extra deployment hassle [11:59] Cool. [12:06] * jml fab bootstraps again [12:16] noodles775: now works with 2.6. [12:17] Thanks jml, re-pulling. [12:40] noodles775: how's it going? [12:42] jml: sorry - looking at other bugs, but just re-bootstrapped and running tests.. (2F's) [12:42] ... and the test run seems to have paused? [12:42] one of the tests is long, but it has a two minute timeout [12:42] ah [12:43] it's wonderful how this virtualenv eliminates differences between environments, no? [12:43] jml: heh, does it not? the failures I'm seeing are missing packages from the virtualenv... [12:44] noodles775: but if I just bootstrapped and got no errors, and you bootstrapped and got failures due to missing packages, how is it actually helping? [12:44] jml: not sure why I don't have a proper result for the error, but http://paste.ubuntu.com/755959/ [12:44] jml: the bootstrap worked fine... actually, let me check the scroll. [12:45] jml: it would help if we'd ensured that all the dependencies were in the virtualenv, wouldn't it? I'm not sure what you mean, otherwise? [12:45] huh what? [12:46] so, if the purpose is to provide a reproducible, controlled environment and we are getting different environments running on different machines, then it's not doing its job very well [12:46] I mean, virtualenv can only (try to) make the virtualenv's the same on both computers - which I *assume* (perhaps stupidly) that it's doing... but earlier you had system packages installed that I didn't etc. [12:46] it's got --no-site-packages [12:47] anyway, the 0755 thing is an easy fix [12:47] Cool [12:48] Oh, I'd not realised you were using --no-site-packages, not sure what that means for the earlier missing twisted package. [12:48] probably that I installed it and then forgot that I installed it. [12:49] more interestingly, I don't know what it means for the failing 'want' script [12:49] you should have devportalbinary.binary in your PYTHONPATH [12:49] oh, I think I know [12:49] you need to run 'fab test' from within the virtualenv [12:49] * noodles775 does so [12:53] jml: cool, that gave me different results... pasting now, but can you update the README about `fab test`? [12:53] noodles775: already in the process of doing so [12:54] jml: sweet - here's the paste... /me installs debuild. http://paste.ubuntu.com/755970/ [12:55] noodles775: yeah, that's just debuild. I guess there's no way of specifying a non-Python dependency other than documentation. [12:55] Nope (devscripts) [12:57] I wish apps.u.c had URLs that excluded series. [12:59] it doesn't? [12:59] * noodles775 checks [13:00] oh, you can URL hack to get tem [13:00] them, rather [13:00] but not by browsing [13:00] Ah - right..., no I think you're right. [13:00] It redirects you if you remove it. [13:00] So you can refer to a package without the series with a url. [13:00] to natty :) [13:01] It depends on whether your browser's user-agent tells anything useful, but yes, not sure why the default is configured to be natty. [13:01] (using chromium it's correctly redirecting me to oneiric) [13:02] probably because it was natty when it was written, and because asking LP for the current stable release was too much work at the time [13:03] (not sure if there's another reliable way to determine the current release) [13:03] Yep. Anyway, it's not just devscripts, I'm now getting an error because dh isn't found... installing. [13:03] Is there a meta-package I should use instead? [13:04] * noodles775 retries with debhelper installed. [13:04] Don't know, sorry. There's no documentation for this in pkgme, which is where there should be such docs. [13:04] Whoohoo, just the 0755 error... sweet. [13:05] http://paste.ubuntu.com/755976/ [13:05] So it's just devscripts and debhelper (as far as we know, I didn't start with a clean VM) [13:05] jml: you're around still this afternoon right? [13:06] * noodles775 considers heading off for lunch [13:06] noodles775: yes. [13:11] noodles775: a quick scan doesn't show any mutual reverse dependencies of those two packages [13:15] * jml goes to lunch [13:15] which may run a little long, as I have to buy a gift for the Millbank secret santa thing. [13:24] actually really going now [13:53] morning [13:56] Morning james_w [13:57] How was your break? [13:57] good thanks [13:57] how are you? [13:59] Pretty well - not getting a lot of coding done today, but other useful stuff :) [14:24] hey james_w, good morning, glad to hear you had a nice break [14:24] hi dpm [14:45] james_w: https://code.launchpad.net/~jml/pkgme-service/end-to-end-pkgme/+merge/83624 would benefit from your review [14:45] sure thing [14:54] jml: is the import on 426 inline for a reason? [15:01] jml: also, when you'd asked about how to test against a running server, I forgot to mention using wsgiintercept... we use something like: [15:01] http://bazaar.launchpad.net/~rnr-developers/rnr-server/trunk/view/head:/src/reviewsapp/tests/helpers.py#L35 [15:01] * noodles775 looks for an example of where it's used. [15:02] noodles775: yes [15:02] noodles775: it is used on line 428 [15:02] But I guess that would be less of an integration test then... so it may not be useful for you anyway. [15:03] noodles775: it's in the function itself because I feel embarrassed about importing something called 'acceptance' at the top of a module. [15:03] heh, sure... [15:05] not sure how the WSGI intercept would help here [15:05] although, that'd be more re-usable as a fixture :) [15:06] Indeed - I'm keen to see how we can clean up a lot of our test helpers with fixtures. [15:08] for me, the great thing is the way they make it easy to provide debugging info [15:11] jml: does the DjangoSettings fixture also mean that I can't be running a dev server while tests run? [15:11] noodles775: it means you can't launch one, IIUC. [15:11] noodles775: if one is already running, won't make a difference [15:11] Great. [15:11] unless django does auto-reload on settings changes [15:12] in which case, then it will make a difference :) [15:12] I could try rewriting that fixture to use DJANGO_SETTINGS_MODULE [15:18] Would being able to overwrite settings on the command line help? [15:19] s/help/avoid the need for the DjangoSettings fixture? [15:19] noodles775: yes, that would help [15:19] noodles775: as long as celery got the message too [15:22] jml: you could start it with the same overrides couldn't you? Anyway, if it helps, [15:22] http://packages.python.org/django-configglue/walkthrough.html#command-line-integration [15:25] jml, james_w, I'm not sure if you've seen this question on the ML, I think you guys are the best people to answer it: https://lists.ubuntu.com/archives/ubuntu-app-devel/2011-November/000041.html [15:27] noodles775: aiui, that command line would change the global settings in much the same way as manually editing the file, wouldn't it? [15:27] dpm, I'll reply, thanks for the notice [15:27] cool, thanks james_w! [15:27] noodles775: i.e. it doesn't offer a way to change settings & run in the same command [15:28] jml: I think it does - the example is just confusing because it's calling the settings command... but you can call the runserver command in the same way? [15:28] I'll give it a try [15:32] Although the project's not yet using configglue, so it might be a bit of setup. [15:32] james_w: btw, we have a call scheduled now, I assume it's not on (but would be keen for some sort of voice chat today) [15:33] jml, I'm happy to talk now [15:33] james_w: cool. now is good for me. [15:33] james_w: I had a call scheduled in this slot anyway :) [15:35] jml: ah, and I missed your earlier comment - no, it's just overriding the setting for the process, it's not touching the file afaik. [15:37] jml, james_w: given that james_w's got twisted foo and is looking through the MP too, I might leave it there (eyes are starting to glaze over). [15:37] noodles775: np. thanks for your feedback so far [15:37] noodles775: I think apart from DjangoSettings I've incorporated all of it already. [15:38] Yes, all tests are passing now on my lucid instance :-) Great. [15:43] jml: heh, just saw your commet msg "Change the default settings to ones that are equally wrong..." - I assume you won't be using django_project/settings.py at all for a deploy, so i think they're now valid for a dev setup? [15:44] noodles775: if there's an all-symbols.db handy in the cwd. [15:45] If it helps, here's a typical config branch (for apps.ubuntu.com): https://bazaar.launchpad.net/~canonical-ca-hackers/isd-configs/uwc-config/files [15:45] jml: ah - I assumed that was somehow obtained as part of the bootstrap. [15:45] noodles775: no, maintained by a different service [16:35] james_w: oh, I also want to bend your ear about notmuch sometime [16:36] sure thing [20:04] dpm, hi David. [20:04] hi JoaoSantana [20:05] dmp, I'm the one who is preparing a ubuntu-defaults for Brazil, remember? === achuni_ is now known as achuni === EvilJackyAlcine is now known as [Jacky]