[00:03] sinzui: there are foundational bugs about trailing whitespace [00:03] sinzui: all I want is a ui=sinzui :) [00:03] sinzui: I'll look into the whitespace issue [00:04] sinzui: it appears that the multiline editor adds some carriage returns at the end for some reason [00:04] yet to be determined [00:07] mwhudson: got any comments on the pics above? [00:11] thumper: The UI is r=me. You may want to verify the Comments IField is strippedText* and then you want to verify the setter does the cleaning instead of the widget. I had to do something similar in other Ajaxed forms [00:11] sinzui: ok, ta [00:14] thumper: looks nice [00:14] * mwhudson afk for lunch [00:16] thumper: the StrippedTextLine is flawed. StrippedTextWidget is doing the work. The comment's IField type must work like c.l.field.URIField [02:07] anyone know our windmill xpath expressions? [03:04] mwhudson: you around for a pre-impl call? [03:04] thumper: sure [08:03] Gnargh @ people who don't check security declarations before they export objects. [08:13] going to ile a bug? [08:14] Just done so now. [08:14] bug #458833 [08:18] good morning [09:15] Morning! [09:23] bigjools: Morning... Want to have a look at bug #458833? [09:52] wgrant: have you tried to rocketfuel-branch today? [09:53] bigjools: No, I don't normally use it. [09:53] * wgrant tries [09:53] ok [09:53] I got an error when it builds the wadl - "ImportError: No module named apt_pkg" [09:53] awesome [09:54] Ah. [09:54] I know why that is. [09:54] missing __init_ ? [09:54] There's a new python-apt in primary. [09:54] Which supersedes the one in launchpad/ppa which supports python2.4. [09:54] GAH [09:54] Again. [09:54] ffs [09:55] maxb usually updates his, which then gets copied into launchpad/ppa, but it's easy enough for anybody to do it. [09:55] For now, you can just downgrade. [09:55] yarp [09:55] rockstar, bzr-removable [10:35] allenap: I've a question about utilities/ec2 if you've the time. [10:35] gmb: Sure. [10:37] allenap: What Python should ec2 be using? It breaks with 2.4 because of the PQM issue and it breaks with the system python because of something to do with paramiko (don't have a traceback to hand at the moment). [10:37] allenap: It does work with python 2.5 but I figured we were hoping to skip that and go straight to 2.6 with the rest of LP. [10:38] gmb: It'll work with either 2.5 or 2.6. [10:39] allenap: Oh, okay. My paramiko error must've been caused by something else. Cool, thanks. [10:39] gmb: The utilities/ec2 script is a bit broken right now. See "Problems with utilities/ec2 test -s "..." on Jaunty (and possibly on Karmic too)?" on launchpad-dev@. [10:40] allenap: Yeah, test -s is what broke for me. Using the version in an old branch of stable worked for me though. [10:40] Is a separate 2.6 branch going to be maintained for 6 months? [10:40] ISTM that neither trunk can be moved to 2.6 before Lucid. [10:43] wgrant: Because 2.6 isn't on hardy? [10:43] allenap: Right. [10:44] wgrant: there is A Plan [10:44] which involves a ppa [10:44] wgrant: I don't think a little thing like that will stop them :) [10:44] mwhudson: Urgh. [11:02] Morning, all [11:04] bigjools: hmm..... but how did you get that newer python-apt installed?! the requirements of launchpad-developer-dependencies should have forced you to stick with the old version until the PPA was updated! [11:04] * bigjools shrugs [11:05] My updating just now has presented me it as a held-back package (which is what prompts me to go and update it [11:05] weird === henninge_ is now known as henninge [11:49] deryck: hey, you might want to look at my lazr-js branch too [11:51] thumper, ok, will do. do you do any work on this patching problem? Or just fixes to get your stuff working? [11:51] deryck: the lazr-js branch is all css styling to use a class rather than an id [11:52] deryck: mwhudson has looked at it [11:52] deryck: but i was wondering about replacing rather than adding css styles [11:52] deryck: in my branch multiline edit branch I add the lazr-multiline-edit class to the bugs page too [11:52] deryck: in preparation for the lazr-js fix [11:53] thumper, ok, cool. thanks for making this better. :) [11:53] deryck: np, I'm off now, it is almost midnight [11:56] cool, enjoy the weekend. === matsubara-afk is now known as matsubara === mrevell is now known as mrevell-lunch [13:13] maxb: did you figure out why your PPA package wants to get upgraded to karmic's? [13:14] It doesn't, for me :-) [13:14] maxb: I downgraded and it wants to upgrade again for me, is there any debugging I can do for you? [13:15] What wants to upgrade it? apt-get ? aptitude? update-manager? [13:15] tried the first 2 and they both want to [13:18] This makes no sense [13:18] aptitude says: python-apt [0.7.13.2ubuntu3launchpad~maxb1 -> 0.7.13.2ubuntu4] [13:18] After upgrading to vanilla karmic, the dependencies of launchpad-developer-dependencies would be violated [13:18] allenap, when you are around would you check and make sure that bug 451158 is fixed in the version of launchpad i just released? [13:18] Bug #451158: AttributeError: 'Credentials' object has no attribute 'authorizeSession' [13:18] bigjools: you do have lp-dev-deps installed? [13:19] * bigjools facepams [13:19] facepalms, too [13:19] leonardr: Okay. [13:19] guess what! [13:19] the karmic upgrade must have removed it :( === mrevell-lunch is now known as mrevell [13:21] * maxb disappears for lunch [13:21] maxb: and now it works, thanks [13:44] leonardr: I get this now: http://pastebin.ubuntu.com/299769/ [13:44] aaargh [13:44] leonardr: I got the Python console by: bzr branch trunk boot && cd boot && virtualenv --no-site-packages . && bin/python bootstrap.py && bin/buildout && bin/py [13:49] i tested get_token_and_login but not login_with [13:52] i think i need to create a fake launchpad so i can write tests for these methods [13:52] leonardr: get_token_and_login() works fine. [13:52] leonardr: Something for a quiet afternoon perhaps ;) [13:52] yeah, i think it's just called something other than 'credentials' now [14:05] allenap, it looks like the problem is that the credentials object is no longer set as 'credentials' on the service root [14:07] leonardr: Is is somewhere else, or is this a pita to fix? [14:07] it's a bit of a pita [14:08] i'll probably end up doing another lazr.restfulclient release that sets it [14:09] i bet it would currently be available as ._browser.authorizer, but it makes more sense just to set it again [14:26] allenap: would you try your code in conjunction with https://code.edge.launchpad.net/~leonardr/lazr.restfulclient/set-credentials ? [14:26] and also review that branch? (i'm writing mp now) [14:27] leonardr: Okay. [14:28] Okay, can someone help my poor, tired brain: Why would bug 438985 occur in a web environment but not in the harness or in a test? [14:28] Bug #438985: Trying to make myself as bug supervisor of my project oopses [14:28] Oh, wait. [14:28] I know why. [14:28] It's a DB object thing isn't it? [14:29] Okay, so, my next question: What's the best way for me to ensure that I've got two differnt objects that refer to the same row in the database? [14:29] allenap: Do you know? ISTR you having similar problems a while back (I could, of course, be making this up). [14:31] gmb: Why do you want to ensure that you have two different objects for the same row? [14:31] allenap: So that I can write a regression test for the above bug. [14:31] gmb: (a is not b and a.id == b.id) [14:31] allenap: Right, that's what I need. [14:31] Is there a simple way to do it that you know of? [14:32] gmb: Oh, and check the table. [14:32] gmb: I don't really understand. If you have two objects for the same row from the same store that's a bug in Storm isn't it? [14:32] allenap: Okay, I'm confused... I don't need to check that they're different, I need them to actually *be* different. How do I create two different object referring to the same DB row? [14:33] allenap: Well, it might be, but the problem is that we're comparing two Person objects using 'is' instead of '==', and it breaks in the web environment but not in tests or harness (because the object stays the same between queries) [14:33] allenap: Fixing the bug is easy. Writing the test to make sure it doesn't happen again isn't. [14:33] gmb: Ah, can't you just instantiate them and then set their properties? Or, if you don't need them to actually work, you could use a Snapshot. [14:34] allenap: Ah, interesting idea. Yes, I could cheat like that, I guess :). [14:34] \0/ for hacks [14:34] allenap: Thanks, I'll try that. [14:35] gmb: If the test is just to prevent someone from using "is" again, then it sounds like it's not worth it. [14:35] allenap: Would you be happy for that to land, as a reviewer? [14:35] barry: stand-up [14:36] It's a bit of a bugger of a thing to detect, so I'm inclined to at least try to get the test to fail. [14:39] gmb: I guess I don't really know how this bug plays out in the code, so I'll keep quiet from here on :) [14:39] allenap: Fair enough. Suffice it to say it's a PITA. [14:42] sinzui: just you and me? :) [14:48] leonardr: That took me a while ("now, how do I actually use an unreleased library here?") but in the end, it works fine. [14:48] allenap: sorry, i could have helped oyu with that [14:50] leonardr: I found HackingLazrLibraries which was a good thing to read anyway. [14:51] allenap: can i remind you to vote on the mp? [14:52] leonardr: Just done. [14:57] barry: look at lib/canonical/__init__.py [14:57] barry: I think you can filter all Set, SHA, and MD5 from there. [14:59] sinzui: ah, very interesting. thanks for the link. one problem is that this won't filter anything for the twisted applications that run in a subprocess. but i think i have hacked tachandler to do that. [14:59] barry: I think you can. There used to be hacks in the tac files for error loging [15:02] EdwinGrubbs: what is the status of bug 436507 [15:02] Bug #436507: Distribution series source package page has no links to distribution source package page [15:02] sinzui: I didn't work on it. It just needs a couple of minor changes before it can be put in review. [15:03] okay [15:03] EdwinGrubbs: the timeout bug is more import [15:04] sinzui: if I don't work on the timeout, I might get karmic retribution [15:05] correct [15:17] jml: ping [15:17] barry, I'm on a call right now -- how can I help? [15:18] jml: just ping me when you're free, i'd like some suggestions on how to handle a tricky twisted problem [15:18] barry, ooh, fun. [15:18] barry, will do. [15:18] jml: thanks! yes, "fun" :) === ursula is now known as Ursinha [15:30] hi beuno [15:32] hi bac [15:33] leonardr, lazr.restfulclient's setup.py says it uses zope.interface, but zope.interface isn't used in the code [15:33] jml: it's used by lazr.restful, which is a testing dependency of lazr.restfulclient [15:33] you might be able to remove it from setup.py but it will still be downloaded and used [15:34] leonardr, ahh ok. [15:34] leonardr, I'm just trying to figure out what extra dependencies bzr would need if it started to use launchpadlib [15:38] henninge: r=me for your branch from this morning [15:54] adeuring: thank you [15:54] henninge: welcome :) [16:16] jml: reping (if you're free) [16:18] barry, like a bird on a wire [16:18] barry, what's up? [16:18] jml: maybe -> #launchpad-sprint? === salgado is now known as salgado-lunch [16:25] gmb: did you make any progress on bug 438985? [16:25] Bug #438985: Trying to make myself as bug supervisor of my project oopses [16:26] BjornT_: I've got a fix playing in ec2 now. [16:27] gmb: i mainly wanted to know if you figured out what the problem was [16:28] adeuring: when approving merge proposals, it would be great if you could set the review type to "code". The new "ec2 land" command looks for that to find the reviewer for the the branch. Danke. [16:28] BjornT_: wgrant already did that: we were comparing two Person objects with "is" instead of "==" in the check to see whether one could subscribe the other. So even if you were subscribing yourself the comparison returned False, hence the OOPS. [16:28] henninge: thanks for the head-up. I wasn't aware to this [16:29] BjornT_: The problem I had was writing a suitable regression test, but allenap helped me figure that out. [16:29] adeuring: np [16:29] adeuring: I wasn't either until it just refused to work for me ... ;-) [16:31] gmb: do you know why you had different objects there? what does the test look like? [16:31] BjornT_: No, I don't. I'll paste a diff of the test for you, hang on... [16:32] gmb: ok. did you try passing in a security wrapped object to the method? that should do it; i'm quite sure that is want is happening. [16:33] BjornT_: Actually, no I didn't. The test looks like this: http://pastebin.ubuntu.com/299839/. [16:34] BjornT_: The security-wrapped object would probably be a better test, actually. What do you think? [16:35] BjornT_: Actually, I wonder, looking at it now - is there a chance that that test could pass when it shouldn't? It failed for me every time without the fix, but perhaps I'm being overly optimistic about it. [16:36] gmb: makePerson() is buggy. it strips the security proxy from the created person [16:36] Ah. [16:37] Eurgh. [16:38] gmb: there is no rationale for stripping it (except for modifying some attributes), so i would try fixing it. if that fails, you should add an XXX, explaining why you get the created person from the db again. [16:38] abentley: ping [16:38] BjornT_: Okay, will do. Thanks. [16:38] barry: pong [16:39] abentley: hi. i'm having a problem with 'bzr shelve somefile'. it asks me whether i want to shelve it, i say yes, but then nothing gets put on the shelve and the change in the working tree doesn't get backed out. any idea what's going on? [16:39] abentley: bzr shelve --all works fine though (but it's too coarse for what i need to do now) [16:40] barry: And does "bzr shelve" work if you shelve only somefile? [16:40] abentley: no [16:40] barry: do you press "y" twice in the case where it doesn't work? [16:40] barry: I'm not sure why you would get that. [16:41] al-maisan: we just tested your permission change on testing and it works, thanks [16:41] james_w: it only asks me once, so i only hit 'y' once [16:41] barry: does it show a diff? [16:41] barry: That's very odd. [16:41] \o/ [16:41] james_w: it does! [16:41] abentley: yeah [16:41] barry: It should always confirm the overall shelve operation. [16:42] abentley: i never get that last confirmation [16:42] abentley: unless i use --all [16:42] Shelve? [yNfq?]y [16:42] Selected changes: [16:42] M lib/canonical/launchpad/daemons/tachandler.py [16:42] [16:42] then it exits [16:43] % bzr shelve --list [16:43] No shelved changes. [16:43] barry: Are you hitting lowercase y? [16:43] abentley: yep [16:43] Does it say "no changes to shelve" before it quits? [16:44] abentley: nope [16:44] oh wow [16:44] abentley, james_w: um, if i run it in a gnome-terminal instead of an emacs shell buffer, i get the confirmation message! [16:45] huh [16:45] and when i give the second 'y', it works [16:45] james_w: shall i file a bug on bzr? [16:45] barry: Please do. [16:45] abentley, james_w thanks! [16:46] barry: I'm not sure-- it could be caused by the way prompts are written/overwritten with \r. [16:46] abentley: very possibly so [16:46] barry: Or it could be the way we retrieve a single character. [16:48] barry: If you want to play with it, it's in bzrlib.shelf_ui.Shelver.prompt [16:48] abentley: thanks. my stack is deep right now so i'll just report the bug and move on ;) [16:49] barry: Understood. Wouldn't want you to get a recursion traceback. [16:49] abentley: MemoryError [16:51] abentley, james_w: https://bugs.edge.launchpad.net/bzr/+bug/459185 [16:51] Bug #459185: 'bzr shelve' with confirmation fails in an emacs shell buffer === matsubara is now known as matsubara-lunch === danilo__ is now known as danilos === salgado-lunch is now known as salgado === beuno is now known as beuno-lunch === gary_poster is now known as gary-lunch === dpm is now known as dpm-afk === matsubara-lunch is now known as matsubara === gary-lunch is now known as gary_poster === beuno-lunch is now known as beuno [18:48] abentley, could you give me hints on where to put pdb trace to find this hidden exception: http://pastebin.ubuntu.com/299933/ [18:49] rockstar: I could, but I've got a good guess at the cause: you haven't configured an oops directory. [18:51] abentley, you mean error_dir ? [18:53] rockstar: possibly [18:54] abentley, looks like it, but it doesn't appear to be set in any of the other jobs in configs/schema-lazr.conf [18:54] rockstar: True. [18:55] abentley, configs/testrunner/launchpad-lazr.conf ? [18:55] It looks like that's what I need. Thanks. [18:55] rockstar: The error isn't accessible to you because the test is running a script. Run the script yourself in order for pdb to work. [18:56] abentley, yeah, but it didn't fail when I just ran the script. [18:56] rockstar: Probably you didn't set the CONFIG environment. [18:57] abentley, maybe. Adding it to the testrunner configs made my test work. [18:57] rockstar: Great. [18:57] abentley, this job script stuff is quite nice once you get all the intricacies of it. [18:58] Thanks. It could probably do with some reworking, because I put it together gradually. [18:58] rockstar: And the oops reporter should scream if you try to configure it with None as the error_dir. [18:59] abentley, yeah, probably. [19:13] abentley, chat? [19:13] rockstar: sure. [19:15] rockstar: Where? [19:16] abentley, GTalk === salgado is now known as salgado-afk === matsubara is now known as matsubara-afk === MTeck-ricer is now known as MTecknology [22:23] is it normal for rocketfuel-get to complain about "Not a branch" twice (once for shipit, once for canonical-identity-provider)? [22:26] mzz: Yes. Those two components remain proprietary. [22:26] ah, ok. I didn't notice anything breaking, just making sure I wasn't accidentally running a script normal people don't use [22:27] That script will die if one of the essential components fails to download [22:57] so! I'm trying to fix bug 456764. Iiuc I need to teach lazr.restful how to convert a timedelta to json first. I figured I'd just do that separately, build an egg with the version bumped, put it in lp-sourcedeps, and tell launchpad to use it (versions.cfg?). But to do that I need a python 2.4 setuptools. [22:57] Bug #456764: please expose builder queue state through the api [22:57] questions: is this at all sane, and is there a python-setuptools for python 2.4 somewhere? [22:59] or hmm, I guess I could be lazy and just let ez_setup grab setuptools