[00:08] jml: morning! [00:08] thumper: https://bugs.edge.launchpad.net/launchpad-code/+bug/430354 [00:08] Bug #430354: BazaarBrachStore needs to use push --overwrite [00:08] mwhudson: ta [00:08] jml: the bzr-git tests are failing in that not-really-failing way :( [00:09] :( [00:09] kfogel, hi :) [00:11] jml: do you know how i can find out what the failure actually is? [00:12] mwhudson, what are the circumstances? [00:12] mwhudson, I'm afraid I know next to nothing about the problem. [00:13] jml: https://pastebin.canonical.com/22151/ [00:13] oh _those_ [00:15] mwhudson, I'll need a coffee to help you properly, but I think I've put most of the relevant info in a bug report [00:15] jml: can you remember which bug report ? [00:16] mwhudson, I'm looking for it now. [00:17] mwhudson, http://twistedmatrix.com/trac/ticket/4003 [00:18] there's one in Launchpad... [00:18] i've found the launchpad one [00:18] mwhudson, what is it? [00:18] https://bugs.edge.launchpad.net/launchpad-foundations/+bug/425113 [00:18] Bug #425113: Some tests can fail without detection [00:18] it wasn't triaged (!) [00:20] launchpad overall is rather poor at triage [00:20] man wtf [00:21] oh [00:22] mwhudson, it's a nasty problem. [00:22] mwhudson, I can try to address it if you'd like, but I'd really also like to finish off the permissions branch I've started. [00:22] jml: i'll dig for a moment myself [00:23] mwhudson, ok. let me know if you want context or to bounce ideas. [00:29] mwhudson: how do you invoke ec2test now it is a module in the tree? [00:30] thumper: ./utilites/ec2test.py is still tehre [00:30] mwhudson: ok [00:30] mwhudson: I noticed the move, but not the addition [00:31] mwhudson: :( [00:31] thumper: :( ? [00:31] mwhudson: my "hack" of using an alias for ec2test doesn't work [00:31] now it needs a path [00:33] * thumper changes the symlink to an alias [00:38] thumper: ping [00:40] jml: a fairly small hack to zope.exception gets things working again [00:41] mwhudson, oh nice. [00:41] mwhudson, diff? [00:41] jml: well what's in eggs isn't in a bzr branch of course :/ [00:41] jml: but it's "f.f_locals" -> getattr(f, 'f_locals', {}) [00:42] *nod* [00:42] barry: hi [00:42] mwhudson, I still suggest twisted should change... [00:42] thumper: hi will you be around for a while? i need to run out before the stores close [00:42] barry: yeah, although I might head out for coffee and lunch soon [00:42] thumper: no worries. it's gonna be 45m-1h anyway i'm guessing [00:43] thumper: will ping you when i get back [00:43] ok [00:44] jml: yes, but in the interests of not having undetected failures in our test suite... [00:44] mwhudson, oh yes. [00:44] mwhudson, I'm very happy to have urgent bandaids applied [00:48] jml: maybe this is a cleaner/easier band-aid: [00:48] from twisted.python.failure import _Frame [00:48] _Frame.f_locals = property(lambda self: {}) [00:48] mwhudson, yeah, I think so. [01:10] mwhudson, is there anything you'd like me to do wrt this? [01:11] jml: review this branch in a bit, i guess [01:15] mwhudson, ok, thanks. [01:19] maxb, your unittest branch is up for review [01:20] ok [01:21] or was that a question? [01:21] I know it's up for review, I marked the MP as such [01:22] karmic appears safe again, btw, on amd64 and i386, at least if you're using archive or gb.archive [01:28] oh, phew, the failures in the bzr-git tests are really shallow [01:31] maxb, sorry, yes it was a question. [01:31] maxb, I'll review it :) [01:32] oh, in fact I have [01:32] hurrah [01:32] You reviewed it without realising? [01:32] :-) [01:34] jml: "peculiarly" as in "why repeat an operation that your base class is already doing?" [01:34] maxb, it's been a very distracted morning [01:35] maxb, iirc, it's there for Python 2.3 support [01:35] but I can't recall exactly. [01:35] * jml does a science thing [01:35] fair enough (and that'll be 2.4 support) [01:35] maxb, Twisted supports python 2.3 :) [01:36] ok, 2.4 and earlier [01:36] It's just a different way of providing such compatibility to which other TestCase subclasses have taken, and less intrinsically clear that it's there for compatibility [01:37] hmm, I need to write more branches, that's my last one :-) [01:38] right. [01:40] "WARNING: The following packages cannot be authenticated!" [01:40] the list includes libc6.. wtf [01:46] sadness is submitting a merge proposal at ??:?5:?? o clock [01:49] yeah. [01:49] Somebody should do Something about that. [01:50] What's special about ??:?5:?? o clock ? [01:52] maxb, the diff on MP pages is generated by a cron job that runs at ??:05:?? [01:53] rather than by, say, some daemon that's listening for events [01:55] jml: thanks for the review [01:55] mwhudson, np [02:35] thumper: ping [02:36] barry: pong [02:36] thumper: is this a good time to talk? [02:36] barry: lets talk titles :) [02:36] yes [02:36] thumper: cool. skype? [02:36] yep [02:36] firing it up [02:37] barry, hey [02:37] jml: hi [02:37] barry, I know you're talking to thumper right now, but do you know what's going on with the '+filebug' thing appearing in page titles? [02:37] actually what I mean to say is... look at the bug I'm about to link to :) [02:38] jml: salgado was supposed to be landing a branch that allowed views to fix this. don't know if it landed yet [02:38] barry, but individual views must fix this? [02:39] jml: i haven't looked at his branch, but i believe this is the case. there was some talk about labels getting in the act but dunno what the state of that is [02:40] barry, ok, thanks. [02:40] barry, it's a little sad that the default is quite so wrong [02:42] jml: agreed [02:53] barry, is there something we can do to improve the default? [02:54] (also, would it be more convenient for me to raise this on the list?) [02:54] jml: i think it would be better to raise it on the list. i'm not sure what the state of salgado's branch is, or exactly what it changes [02:54] barry, ok, thanks. === abentley1 is now known as abentley === barry` is now known as barry [03:38] * jml away for a bit [03:52] Erm. [03:52] Why do I see tests for the new braindead titles? [04:04] wgrant: because we are out of time to do page conversions. It is faster for me to convert pages and send people to add the need breadcrumb adapter afterward [04:04] wgrant: salgado has a branch that fixes most of the bad breadcrumbs, that will also fix the titles [04:14] sinzui: Ah, thanks. [04:14] wgrant: I think I can send every team member to fix breadcrumbs after tomorrow [04:16] launchpad-project 3.0 is timing out: OOPS-1355EB146 [04:18] Ah, there we go. [04:19] At least 1397 queries issued in 15.77 seconds [04:19] Ah. [04:27] phew [04:27] wow [04:27] which project? [04:28] launchpad-project? [04:28] thumper: launchpad-project? [04:28] s/?/./ [04:32] * thumper wonders where all the queries are :) [04:34] 1397? that's a little high [04:35] thumper: Doesn't the OOPS tell you? [04:35] wgrant: well, yes [04:36] wgrant: well, yes and no [04:36] it'll tell you what they are, but not where they are called from [04:36] thumper: Ah. [04:36] I haven't looked extensively at the OOPS format. [04:38] SELECT BugTag.bug, BugTag.id, BugTag.tag FROM BugTag WHERE BugTag.bug = %s ORDER BY BugTag.tag repeated 700 times [04:38] Ah, of course. [04:38] SELECT OfficialBugTag.distribution, OfficialBugTag.id, OfficialBugTag.product, OfficialBugTag.tag FROM OfficialBugTag WHERE OfficialBugTag.product = %s ORDER BY tag 394 times [04:38] I'm sure it is fast with no data [04:38] :) [04:38] Indeed. [04:39] * wgrant files a bug. [04:39] Is that malone or launchpad-registry? [04:40] wgrant: is that the 3.0 milestone page? [04:40] sinzui: Yes. [04:40] rock [04:40] That is my excuse to remove the tags [04:40] That does not follow. [04:40] Ah. [04:41] I think it makes the list hard to read, and I do think they help me understand the bugs I am seeing [04:41] sinzui: Did you mean "I do not think they help me[...]"? [04:42] wgrant: yes. [04:42] sinzui: They should be useful, but are not in their current location. [04:42] sinzui: And why do I have an empty sidebar on that page? [04:42] right [04:42] It takes up some large horizontal fraction of my screen, and is empty. [04:43] wgrant: You cannot subscribe to bugs on that page? [04:43] * sinzui liked the page without the side bar [04:43] sinzui: I can't. That seems like a bug. [04:43] It is [04:43] Oh. [04:43] It's the project, not product, milestone page. [04:43] So it's not a real thing to be subscribed to. [04:44] right... [04:45] I can make the side discretional in the template. [04:45] sinzui: But that leaves a largely useless sidebar on other milestones. [04:46] Is there nowhere better to put the link? [04:47] I can use a horizontal list after the description. It breaks style and may confuse user, but *I* think it makes the page usable. [04:47] huh. I killed rocketfuel-get at the wrong point and managed to lock lp:~launchpad/lp-source-dependencies/trunk. Surprised pulling that branch requires a lock! [04:47] sinzui: Possibly put it at the bottom of the Activities portlet, with the bug stats? [04:48] wgrant: Edit and delete do not belong there. [04:48] sinzui: It's not editing in the traditional sense. [04:49] agreed, certainly not when the milestone has a release [04:50] I can make something that looks like the side portlets and float it to the right. [04:50] https://edge.launchpad.net/~wgrant/+archive/ppa/+packages does something not too bad. [04:51] Although I'm sure it breaks the guidelines. [04:51] It might be better if it had the same gray background. [04:52] It does, but a side portlet with one item is dumb [04:52] Yea. I will try that after I sleep [04:52] Great. [04:57] sinzui: what's the status of the page_title breadcrumb? [04:58] salgdao is doing test fixes to the branch. I think I can land tomorrow [04:58] thumper: we have made it hard on him since we keep landing pages [04:59] * thumper nods [04:59] how come no one told me I needed to create a named IBreadcrumb adapter for "code"? [04:59] I spent ages chasing why I wasn't seeing breadcrumbs [05:01] thumper: I'll ask salgado to explain what is happening [05:02] sinzui: I think it'll help, but I understand it now [05:02] barry and salgado had a few meeting about this. I do not think their work plays well together [05:02] sinzui: what would help is knowing how to add additional breadcrumbs [05:02] sinzui: like "Active reviews for XYZ" [05:03] I can create custom IHierarchy adapters (like we have now) [05:03] I'm wondering if there is a simpler way [05:03] thumper, interesting problem. since crumb is a traversed object, I think you need to pop an object onto the stack, or ... [05:04] it is a view, not an object [05:04] I need to have a fake object that has a breadcrumb relating to the page [05:04] thumper: views are an object, that is why the view name is in the crumbs [05:04] yeah, yeah [05:04] ... [05:05] I'm trying to fix my last few pages first === poolie1 is now known as poolie [05:05] * sinzui stumbles to bed [05:05] I want to have a view that is registered against two interfaces [05:05] the problem is that one object implements both interfaces [05:05] :( [05:06] * thumper wonders how helpful the #zope channel is [05:07] * sidnei raises an eyebrow [05:07] thumper: you might get better help on #plone [05:08] sidnei: really? [05:08] thumper: surely [05:08] sidnei: ta [05:09] thumper: i don't see where the problem is. can you give an example of what you're trying to do? [05:09] BjornT: normally zope complains if there are two directives that can apply to one context object [05:10] BjornT: I have a view that I want for the majority of implements of IHasMergeProposals [05:10] back [05:10] BjornT: but I want a special one for IPerson [05:10] BjornT: and a person implements IHasMergeProposals [05:12] thumper: hmm, ok. i guess zope doesn't complain, but you don't know which of the views will get used for Person? [05:13] I think it does complain [05:13] * thumper checks [05:14] thumper: one solution would be to make IPerson inherit from IHasMergeProposal, except that registry shouldn't depend on code... [05:14] But Registry depends on *everything*... is that wrong? [05:14] BjornT: it does right now [05:14] wgrant, it's a work in progress. [05:15] You might be able to do it by registering an adapter for Person -> view rather than IPerson -> view [05:15] jml: How will that work? Adapting things to ICodeProduct or whatever? [05:15] wgrant, well, we already do a great deal of that in lp.code [05:15] wgrant, see IBranchTarget, etc. [05:15] True. [05:15] wgrant: yes. in an ideal world, everything can depend on registry, but registry shouldn't depend on code, bugs, soyuz, etc. [05:15] BjornT: however our API decisions violate that [05:15] BjornT: But that seems impractical unless a huge amount of adaption is involved. [05:15] wgrant, how it will work with the REST APIs is another, and much more interesting, question. [05:16] jml: True. [05:16] BjornT: as we mix everything in [05:16] wgrant, "impractical"? [05:16] wgrant, dumping huge amounts of responsibility on Person and Product is equally impractical. [05:16] BjornT: if there is interface inheritance, does it take the most derived? [05:17] jml: That's not impractical at all. It makes sense, unless you start splitting the apps up. [05:17] thumper: yes. i wonder, why do you need breadcrumbs for IHasMergeProposals? what does that view look like? [05:18] BjornT: the breadcrumbs that beuno wanted was to show "Active reviews for XYZ" when looking at a bmp [05:18] I'm not yet convinced [05:18] but at least now I know more what I need to do to fix the breadcrumbs [05:18] BjornT: I'm actually looking at the +activereviews registration [05:18] BjornT: if I can register +activereviews against IHasMergeProposals, but specialise for IPerson [05:19] BjornT: then that is a good thing [05:19] wgrant, I disagree. [05:19] jml: Why does it not make sense to have the methods of a Person on IPerson? [05:19] wgrant, it violates the principle of single responsibility [05:20] jml: A distribution has lots of binary packages. Distribution.searchBinaryPackages should not exist, simply because Soyuz owns BinaryPackageRelease? [05:20] wgrant, and they are not, intrinsically, methods of a person [05:21] thumper: You can avoid importing code into registry if you want. Create IPersonHasMergeProposals in code and attach that marker interface to Person in code's zcml. [05:21] wgrant, I'm not completely happy with the way we've decided to arrange our tree [05:21] and soyuz / registry seems a particularly thorny case [05:21] It is. [05:22] Some structural things ended up in Registry. [05:22] But others didn't. [05:22] And they're all very tightly bound to the rest of Soyuz. [05:22] stub: but it adds a method that a person must implement [05:22] stub: it uses the mixin idiom we use [05:22] wgrant, so I would love to hear of better ways to split up the tree [05:22] oic. [05:22] although [05:23] ... [05:23] jml: Run away! [05:23] we could just adapt... [05:23] Indeed [05:23] we have the technology [05:23] thumper: i think you can only do that, if IPerson inherits from IHasMergeProposals. looking at it, it looks like it does do that already. we have to think about how to handle this case with our new code structure [05:23] stub: but right now, with the API, to give a person the abiltiy to getBranches we need a mixin [05:24] BjornT: what are the planned changes? [05:25] thumper: IPerson is in registry, so it can no longer inherit from application-specific interfaces, like IHasMergeposals, IHasSpecifications, and so on [05:25] How is lazr.restful going to work? [05:26] BjornT: I'd be happy for the change if we can work out the API [05:27] it seems kinda kludgy right now [05:27] thumper: yeah. it's a nice idea in theory, but we don't yet know how to actually do it practically [05:27] :) [05:28] thinking is easy [05:28] doing is hard [05:29] meh. [05:29] thumper, if you don't know what to do, you haven't finished thinking yet. [05:30] :) [05:30] true [05:30] keep thinking jml [05:30] (but a good way to push thinking along is to do something!) [05:51] sounds like you want an Extension concept [05:53] :( [05:53] build failed [05:53] no space left on device [05:53] mwhudson: just rekick? [05:53] thumper: yeah :( [05:56] * mwhudson is reorganizing ec2tests command line parsing code [05:56] * mwhudson is going to the pub later [05:57] this is merely a happy coincidence [06:00] mwhudson, :D [06:05] which database are the tests using? [06:06] launchpad_ftest [06:06] which iirc doesn't actually exists apart from when a test is running [06:06] it's copied from launchpad_ftest_template [06:07] thanks. [06:07] I'm trying to figure out why this apparently correct code isn't working. [06:10] jml: soyuz hates you? [06:11] yes. [06:11] one day, I hope to make it afraid of me. [06:13] sometimes zope is very cool [06:14] thumper, what's it done now? [06:15] jml: registering views against IHasMergeProposals :) [06:15] jml: what zope is good at [06:15] thumper, ahh yeah. [06:15] jml: that with branch collection adaptation FTW [06:16] heh :) [06:16] thumper, jkakar has been sending me interesting emails about use of a similar pattern in landscape [06:16] I really must get around to actually reading them and maybe even replying. [06:19] * thumper has just fixed (deleted) the last remaining old style template for code [06:19] * thumper EODs for now [06:20] thumper, yay [06:20] thumper, have a good evening [06:35] the relationships between soyuz objects are very complicated. [06:37] (+1, understatement) [06:38] I just hope that this actual behaviour I'm reaching for makes sense. [06:41] flymake-mode has changed my life [06:41] mwhudson, heh :) [06:42] mwhudson, any particular instance this time? [06:42] jml: just help when refactoring [06:42] jml: copy paste block of code into function [06:42] jml: add arguments to function until pink goes away [06:42] mwhudson, yeah. [06:42] that's a nice thing [06:43] although it always makes me wonder how much more automated it could get [06:43] brm [06:44] never really made my life any better when I used it === henninge is now known as henninge-bbl [07:04] * mwhudson EODifies [07:05] jml: want to review this branch chock full of fun tomorrow? [07:05] it's here for now bzr+ssh://bazaar.launchpad.net/~mwhudson/launchpad/ec2-entrypoints [07:08] mwhudson, yes. [07:08] mwhudson, I might even review it today... [07:09] jml: it needs docstrings, i know that much [07:14] heh heh [07:15] oh and copyright headers and __all__ and so on [07:20] *nod* [07:22] jml: What are you fighting with Soyuz about this time? [07:26] wgrant, permissions, still. [07:26] wgrant, but I think I've won this battle [07:33] jml: what was your feel on my py2.5-unittest-compatibility? You did "review approve" but not "merge approve" [07:35] maxb, I review approved. [07:35] maxb, but I confess I forgot that you can't land branches yourself :) [07:35] maxb, so I was expecting you to change the comment or not at your discretion and then land it :) [07:36] * maxb reminds jml that self-landing is not an option :-) [07:36] maxb, do you want to tweak the comment? if yes, please do so and poke me. if not, I'll land your patch now. [07:37] maxb: what is the thing you find odd? [07:37] maxb: or rather what makes it odd to you? [07:37] that it should redo logic that its base class's __init__ does too [07:38] ok, one edited comment coming right up [07:38] maxb, heh, thanks :) [07:38] I'd lost track of whether the "peculiarly" was in the BMP or the code itself [07:39] urgggle, unitteset._CmpToKey [07:39] that really should be in e.g. list.something [07:40] I'm doing too many tasks at once. Please give me a moment while I serialize a few of them. [07:40] jml: "chooses to write to it .... as part of its own method of arranging for pre-2.5 compatibility" ? [07:41] maxb, perfect. [07:41] I must admit to curiousity about why the property is needed [07:42] I'm guessing something is being naughty [07:43] pushed, many thanks for the attention during this branch's bumpy path to this point [07:43] maxb, np. I'll land that now. [07:46] good grief, ec2 demo works [07:46] zomg. [07:47] jml https://bugs.edge.launchpad.net/launchpad/+bug/430504 [07:47] (no action needed) [07:47] (which probably isn't true on trunk btw) [07:48] mwhudson, has your headless improving branch landed on stable yet? [07:49] jml: dunno about stable, it's in devel though [07:50] mwhudson, ok. I'll run from there, see how things work out. :) [07:52] what's ec2 demo? make run, in an ec2 instance? [07:54] Yep. [07:54] Woah. ec2test is split. [07:55] heck yes [07:55] mwhudson has skills. [07:56] also, ec2test --headless takes much less time. hurrah. [07:57] maxb, your branch is submitted. you'll get an email when the test run finishes. [07:57] "# Perl allows for inplace editing unlike sed." [07:57] sed -i? [07:58] * jml shrugs [07:58] thanks [07:58] bye [07:58] maxb, bye. [07:58] wgrant, you wouldn't happen to be rewriting it in Python, would you? [07:59] jml: Can't really do that without a public AMI! [07:59] And no. [08:00] oh. pity. [08:24] good morning [08:53] adeuring, hello [08:53] hi jml! [09:26] Guten morgen [09:31] g'night all :) [09:54] bigjools: Available for a few ddeb-related questions? [09:55] wgrant: yes, ask away, but I am pretty busy and have a call in 5m so I'll do what I can [09:56] bigjools: Great, thanks. [09:56] bigjools: process-death-row currently crashes on copy/debug archives, because they lack an archive_url. [09:56] I guess it's important for debug archives, but not copy archives as they're not published. [09:56] Should I define an archive_url for debug archives? [09:57] wgrant: yes, sounds sane [09:58] wgrant: https://bugs.edge.launchpad.net/soyuz/+bug/244145 [09:58] Bug #244145: process-death-row script chokes on rebuild archives [09:58]