[05:42] moin [05:46] lifeless: in transit? [05:47] mwhudson: den haag, GNU Hackers Meeting [05:47] lifeless: o right [05:48] lifeless: have you considered xapian for searching btw? [05:49] yes [05:49] gmane use it [05:49] but it has almost zero buzz on internet scale sites [05:49] ok [05:56] next-rollout instructions are on the main deployments page, right ? [06:17] mwhudson: https://code.edge.launchpad.net/~lifeless/launchpad/librarian/+merge/30880 [06:26] mwhudson: what happens when lazr encounters a new, unknown key? [06:26] lifeless: i'd have to experiment to be sure [06:27] k [06:27] * lifeless experiments [06:27] ah yes, it suicides [06:27] * lifeless doesn't understand this choking-consistency approach [06:28] mwhudson: think I could convince you to click on 'approve' twice ? === almaisan-away is now known as al-maisan === al-maisan is now known as almaisan-away === almaisan-away is now known as al-maisan [10:55] lifeless: hi, could I get your vote on https://code.edge.launchpad.net/~james-w/launchpad/devel/+merge/30887 I'm having trouble submitting it the manual way. [10:56] \o/ installing vm for playing around with LP! [10:56] nigelb: cool [10:56] I wonder if there are some low hanging bugs I could take a look at to figure out how to start. [10:56] james_w: we should make that easier! [10:57] lifeless: we should [10:57] nigelb: well, like with most projects thare a lots of bits around [10:57] nigelb: I suggest, pick up something you're interested in, and follow your nose / ask for help dig in etc [10:58] lifeless: hm I'll wait for things to be set up so I can play around [10:59] nigelb, fwiw, I found the easiest things to work on were changing text in templates [10:59] find an error in the UI, grep source code to find where it appears, fix, fix tests, go [10:59] ahhh, like the bug I reported and you fixed. [10:59] I should hunt around around for something like that then [10:59] nigelb, yeppers [11:06] poolie: whats the MP url? [11:06] https://code.edge.launchpad.net/~mbp/launchpad/flags/+merge/30581 [11:07] kicked off [11:08] thanks [11:08] what did you type? [11:08] ec2 test -b for me gives a weird misparsed url [11:08] robertc@lifeless-64:~/source/canonical/launchpad-repo/devel$ PYTHONPATH=lib python utilities/ec2 land https://code.edge.launchpad.net/~mbp/launchpad/flags/+merge/30581 [11:08] specifically two urls string joined together [11:09] when i tested it i got some failures in codehosting, and i'm pretty sure i didn't break them [11:09] how does that work with ec2test? [11:09] I think ec2test merges trunk [11:09] you have to keep merging trunk until other people's branches are fixed? [11:09] oh right, it does [11:09] in this case i want it to merge db-devel, but i guess it can get that from the mp [11:10] https://lpbuildbot.canonical.com/one_box_per_builder will show you if we're recently broken on trunk etc [11:13] poolie: http://ec2-204-236-213-119.compute-1.amazonaws.com/current_test.log should let you track it in tribunal [11:13] thanks, i'll have a poke at ec2 test [11:14] for the snags i hit [11:14] i just didn't want to leave thi stalled any longer [11:14] cool [11:14] I'm glad :) === al-maisan is now known as almaisan-away [11:17] lifeless, you should paste that email onto the blog [11:18] which one ? [11:20] poolie: ^ [11:20] blog.l.n. [11:22] poolie: which email! [11:22] architecture overview [11:22] ah [11:22] s//architecture reviewe process/ [11:22] hmmm, do you think users will be that interested ? [11:22] architecture overview review process methodology definition process [11:22] :-P [11:23] perhaps I'll put it on my blog [11:23] that should get the developer orientated types pretty effectively [11:24] up to you [11:24] i think it will be interesting to users on the edge of being developers [11:24] at the moment the lp blog is a bit blah and i'd like to see it built up into something more attractive [11:25] its kindof there to be there [11:25] if you know what I mena [11:25] mean [11:25] I think the shorter iteration cycle will help a lot [11:25] rather than splat per month [11:25] splat per day [11:27] istm you are wearing the TA hat for that post [11:27] rather than the rbtc hat [11:28] sure [11:28] I'll think on it. [11:28] I don't believe in role-blogs as a concept [11:28] people and roles are not partitionable like that [11:29] I like the blog for communicating with the LP community as a thing [11:29] for widespread stuff, its great; its like [11:29] lp-dev list subset of lp-users list subset of actual lp-users subset of folk interested in lp [11:29] and the blog is pretty far reaching [11:30] (I know they aren't strict subsets, its an analogy) [11:32] ah thanks for the buildbot link [11:32] interesting to see the build of their bzr integration there [11:32] :) [11:41] lifeless, did that vm fail? the url doesn't work [11:41] ec2test probably only opened my ip [11:41] its up to lp.archiveuploader.tests.test_uploadprocessor.TestUploadProcessor.testOrderFilenames [11:42] jml! :) [11:43] jml: your diff is empty [11:43] lifeless, *your* diff is empty [11:43] jml: did thoust push ? [11:43] lifeless, I did! [11:44] should be good now. [11:44] jml: up for reviewing my librarian branch ? [11:44] lifeless, no, sorry. [11:44] jml: adds oops logging to it [11:44] ah well :( [11:44] lifeless, maybe once I've emptied my physical in-tray [11:45] jml: that would be nice; I would like to kick it off within 22 hours or so [11:45] lifeless, ok. I'll make a note. === NCommander is now known as NC|Alaska [13:32] poolie: fyi https://bugs.edge.launchpad.net/malone/+bug/607960 [13:32] <_mup_> Bug #607960: timeouts on Distribution:+bugs doing searches [21:39] 'morning mwhudson [21:40] mwhudson: do you have any experience with threads in C Python? [21:42] jelmer: only a little [22:10] morning [22:11] I'm heading off for a dentist appt shortly [22:17] mwhudson: I'm trying to use pythread.h for a C worker thread without having to rely on pthreads, but I can find much documentation about the API. [22:21] jelmer: i guess the headers and the source are probably the most reliable documentation anyway :/ [23:15] wb [23:15] what week is it ? [23:19] I suspect week 2. [23:19] I think week 3 [23:19] Or at least I hope so. [23:19] rollout was on the 6th of july [23:19] The Epic was week 0. [23:19] ah, ok - week 2 then indeed [23:19] * wgrant consults the calendar. [23:20] https://dev.launchpad.net/Releases/2010Calendar says week 2. === jelmer changed the topic of #launchpad-dev to: Launchpad Development Channel | Week 2 of 10.08 | PQM is open | https://dev.launchpad.net/ | Get the code: https://dev.launchpad.net/Getting | On-call review in irc://irc.freenode.net/#launchpad-reviews | Use http://paste.ubuntu.com/ for pastes === lifeless changed the topic of #launchpad-dev to: Launchpad Development Channel | Week 2 of 10.08 | firefighting: buildbot slaves are down | PQM is open | https://dev.launchpad.net/ | Get the code: https://dev.launchpad.net/Getting | On-call review in irc://irc.freenode.net/#launchpad-reviews | Use http://paste.ubuntu.com/ for pastes [23:21] lifeless: re https://code.edge.launchpad.net/~lifeless/launchpad/malone/+merge/30904, it seems mostly reasonably [23:22] *reasonable [23:22] lifeless: have you talked to the bugs developers about this? [23:22] no [23:22] why? [23:22] I mean - the tests and code are pretty clear. [23:22] just wondering really [23:22] I think we're a bit too siloed at the moment, really. [23:22] but maybe thats just me. [23:22] yeah probably [23:23] * mwhudson attempts to articulate more clearly [23:23] mwhudson: while you're around [23:23] perhaps you could review by librarian oops branch [23:23] lifeless: have you thought about ways to detect if this change will reduce the usefulness of the dup search? [23:23] mwhudson: yes, and - I'm sure it will, to a point. [23:24] mwhudson: the bugs devs that aren't on leave have all expressed their desire for and rather than or, FWIW [23:24] for terms and tags and so on [23:24] ok cool [23:24] I think though, that as far as dup detection goes, we'll find that we end up with a cluster of dups that indirect to a sane master [23:25] so we'll get a bit more chaff up front to seed it, and then it will be better than it is today. [23:25] fingers crossed. [23:25] lifeless: i guess what would have made me really happy to review this would be some evidence of pre-implementation discussion in the mp [23:25] mwhudson: I intend to bring up pre-impl as a misfeature of process [23:25] it was instituted to fix a bug we've now fixed other ways. [23:26] lifeless: maybe [23:26] I think its great, and important, to socialise changes, but pre-impl doesn't do that. [23:26] i don't mean pre-imp in a formal sense i guess [23:26] this particular change has been discussed in detail with poolie as a teddy bear [23:26] i just mean 'someone other than me things this is a good idea' [23:27] *thinks [23:27] various bugs folk on hte list [23:27] a prelude-to-it with francis [23:27] :P [23:27] lifeless: ok, that's great [23:27] lifeless: less back and forth if you'd said this in the mp though [23:27] mwhudson: whats at the heart of wanting to know that someone other than you and I think its a good idea? [23:28] mwhudson: I'm open to changing, but the bzr team doesn't have this friction, so the first thing I want to do is understand what it solves. [23:28] lifeless: the thing here is that there's an element of a design decision here [23:28] 'is this going to affect dup search quality' [23:29] if it was just making tests more precise or whatever i wouldn't be quibbling [23:29] mwhudson: go on [23:29] or if it was in an area i'm more comfortable with [23:31] lifeless: so it makes me more comfortable to hear there's been discussion about it [23:31] that's all [23:32] mwhudson: ok; I worry there is something fishy at the bottom of this : a sort of fear of fixing things outside-ones-area, or something. [23:33] Is that fishy at all? [23:34] It's not *good*, but it's the way the team is set up. [23:34] wgrant: yes, we all need to own the stack we work with - top to bottom [23:34] lifeless: where is your librarian oops branch? [23:35] https://code.edge.launchpad.net/~lifeless/launchpad/librarian/+merge/30880 [23:35] wgrant: its only slightly setup that way [23:36] wgrant: it would be sad if unpaid coontributors felt safer and more able to hack on any part of the system than paid contributors. (Not saying mwhudson feels that way: we're exploring a fear *I* have after all) [23:38] lifeless: approved both branches [23:38] thanks [23:38] lifeless: I'm happy hacking anywhere, but I always check with a member of the relevant team. That team controls the code, I feel, so everything should go through them. [23:39] lifeless: to go over my feelings a bit, i don't think it's area-related [23:39] wgrant: this has kindof accreted [23:39] lifeless: it's just that imho some changes require discussion [23:39] and some don't [23:39] discussion in the direction sense [23:39] wgrant: there are unmaintained areas which are kindof a side effect of partitioning. I dunno, its a theory. [23:40] this one did, and i wanted to make sure such discussion had happened [23:40] lifeless: Are they really a side-effect of partitioning? [23:40] I thought they were quite deliberately completely abandoned and left to rot. [23:40] And by extension make Launchpad look bad. [23:40] wgrant: har har har [23:40] Well, maybe that last bit wasn't deliberate. [23:40] mwhudson: mmm. This is perhaps a beer talk. [23:41] mwhudson: In particular though, I don't want us to confuse 'high risk, needs mitigation' with 'changing an old decision' [23:42] mwhudson: for the search, there is a single if block to totally reinstate the old (fails with timmeout 99% of time) behaviour [23:42] mwhudson: so the change has the ability to be rolled back trivially [23:42] lifeless: Fails with timeout 99% of the time *in some contexts*. [23:42] It's not completely broken. [23:42] wgrant: thats true, its only 50% on launchpad itself. [23:42] and 40% or so on bzr [23:42] lifeless: i probably made this into a bigger deal than necessary by explaining myself poorly [23:42] lifeless: it's monday morning :-) [23:42] Hm. I rarely see it time out on LP projects. [23:43] mwhudson: :) sunday am now [23:43] wgrant: turn off edge redirect. [23:43] lifeless: Ah. [23:43] wgrant: you're already running the first iteration of lifeless-tuning on this. [23:43] mwhudson: sorry, I mean, monday am, sunday late evening. [23:44] mwhudson: its not a big deal, its just a background task. [23:44] lifeless: cool [23:44] mwhudson: j'adore bzr culture, and I want to bring it into LP [23:44] one of those things is to unblock *even on risky things* by finding ways to reduce the impact / reduce the time to recover [23:44] lifeless: well ok, but i think the launchpad culture of talking about things before doing them is actually not a bad thing [23:44] mwhudson: pros and cons [23:45] mwhudson: I adore getting solid constraints and requirements and set based design [23:45] mwhudson: for things of that scope [23:45] mwhudson: for iterating, I think a simpler model works well. Anyhow, as I say - background task. [23:46] * lifeless twiddles waiting for spm :P [23:54] lifeless: When are you flying? [23:54] 1800 [23:56] Ah.