=== ddaa [n=ddaa@nor75-18-82-241-238-155.fbx.proxad.net] has joined #launchpad-meeting === mpool [n=mbp@ozlabs.org] has joined #launchpad-meeting [11:01] ddaa: hi, here? [11:01] Hello [11:02] https://launchpad.canonical.com/BazaarMeetingAgenda [11:04] hm, so? [11:06] Meeting in 54 minutes [11:06] really? [11:06] Well, like every week... [11:06] oh, i used to have another one at 7 [11:06] ok, see you then === lifeless [n=robertc@ppp245-86.static.internode.on.net] has joined #launchpad-meeting [12:00] Yay [12:00] MEETING STARTS [12:01] This is the launchpad-bazaar meeting, so there are more branches for use bazaar monkeys to hang about [12:01] == Agenda == [12:01] Next meeting 2006-09-11, 10:00 UTC. [12:01] mpool will be on leave. [12:01] Say bzzzt if that does not fit. [12:01] * roll call [12:01] * production status [12:01] * invalid branches [12:01] * smart server [12:01] * advertising [12:01] * vcs-import knits [12:01] * release finder [12:01] * Python import [12:01] ddaa: both next week and the following week [12:01] * bzr lp:// [12:01] * 1.0 targets [12:01] * critical bugs [12:01] * pending sysadmin tasks [12:01] * proritizing important tasks [12:01] * any other business [12:02] == Roll call == [12:02] Spiv is on leave until September 10th. [12:02] length of meeting is? [12:02] (in minutes) [12:02] 45 mins [12:02] ok [12:02] I'm here. [12:02] i'm here [12:02] my fingers are here [12:02] good enough [12:02] == Prodution status == [12:02] Early last week, abnormally long delay for mirroring sftp branches were reported. It was caused by a routing issue on escudero that caused time-outs for all vcs-imports branches. [12:02] Once escudero was restored, normal service resumed. [12:02] A discussion is in progress (CC Launchpad mailing list) on how to rearchiture the branch puller to minimize latency in a future-proof way. [12:03] I'm having trouble understanding lifeless' last email, but that's off-topic for this meeting. [12:04] == Invalid branches == [12:04] Last week actions: [12:04] * ddaa to bother jelmer about the bzr-svn bug [12:04] * ddaa and lifeless to discuss bzrlib data validation. Not done. [12:04] Following up on last week. I ack bzr should prevent bad data from getting in, but since shit happens, the supermirror should be able to recover without needing administrator intervention. [12:04] I'm not too sure about how to do that. Does somebody has a plan and want to start the discussion? [12:04] I do not think I got it across to jelmer last week. [12:04] lifeless and I did not have that discussion. [12:05] jamesh: lifeless: any clue about "supermirror recovery from invalid data"? [12:05] ddaa: like I said, I dont think there is a discussion to have [12:05] is it a problem that is happening right now? [12:05] I thought you had changed that as per the discussion we had *in last weeks meeting* [12:06] ddaa: can the puller overwrite the invalid branch? [12:06] jamesh: I do not expect so, that's the issue [12:06] ACTION: ddaa to email to explain what he sees as a problem, since apparently it's not obvious [12:06] if the problem is not occuring right now... [12:06] SteveA: it's occuring right now [12:07] let's move on [12:07] ddaa: ideally we'd handle it in the same way as a branch format change: blow away the mirror and start again. The question is whether we can differentiate between invalid data and bzrlib bugs [12:07] how much is it occuring? [12:07] SteveA: all samba branches produced by bzr-svn [12:07] (i.e. do we get a specific exception or random failures inside bzrlib?) [12:07] potentially random failures, although this specific problem causes a BzrError. [12:08] == Smart server == [12:08] Last week spiv said: lifeless & mpool & I will be meeting in person tomorrow to continue work on the smart server. lifeless has figured out some good goals w.r.t. supermirror integration and HTTP integration that's the next step. [12:08] mpool: status? [12:08] we did meet, [12:09] lifeless: what was the last status you had from spiv before he left? [12:09] he was part way through splitting those branches [12:09] I hope/imagine he has pushed his current set to chinstrap/sodium [12:10] i hope we can merge at least part of it for 0.11, and have andrew merge the corresponding bits to the supermirror when he gets back [12:10] i'm not sure if this is realistiic [12:11] hes back with a week to spare for 0.11 [12:11] some at least will land, I'm sure of it [12:12] Do you think it's still on track for deployment before october 8th? [12:12] ddaa: not as much slack as i would like, but possible [12:12] (I think that's the Launchpad 1.0 time) [12:12] Full smart server deployment, no [12:12] lifeless: as opposed to what? [12:12] we're aiming for the RPC transport deployed into both the hosting and mirror environments [12:12] I'm not clear on what is "full deployment" as opposed to "core support" [12:13] the smart server will be a launchpad-code-free addition on top of that as it goes into bzr [12:14] lifeless: so you mean that the smart server may be used for the publishing side, but the hosting side will be post 1.0? [12:14] no [12:14] ddaa: i think we mean [12:14] all the launchpad-specific code will be done by then (dog willing) [12:14] but it need more pulls from bzrlib to be really exciting [12:15] what is the risk level of those "extra pulls", is that just some tuning that needs to go through review and merge, or is there some tricky outstanding issues? [12:16] ddaa: where are you driving with these questions? === lifeless up the wall [12:16] sorry, thats *what* [12:16] heh [12:16] trying to get a hang on the progress and Lauchpad 1.0 target delivery, since it seems I'm the one reporting on that [12:16] and it's all very fuzzy to me [12:17] but maybe it's a pointless excercice since it will be done when it's done and there is not anything we can do to make it happen faster [12:17] ok, so let's say it's likely we will have approximately alpha-level smart server support fielded by that date [12:18] the additional work in bzr will be to speed it up by doing higher-level operations [12:18] i'm keen to be running a smart-server on the supermirror soon, even if it provides no benefit over regular sftp [12:18] it is not risky in an architectural sense, i am confident it will go according to plan [12:18] so that we will know about how it works practically [12:18] SteveA: right, that's what we're aiming for first [12:18] right [12:18] and also to shake out any lp integration issues [12:18] SteveA: thats exactly what we've mapped out [12:18] and we can then add extra commands/calls or whatever you call it [12:18] i think we should get spiv on that as soon as we can when he returngs [12:19] mpool: hes already on it [12:19] later to incrementally improve it [12:19] cool [12:19] so can we call that closed? [12:19] SteveA: yes exactly [12:19] Yup. I feel enlightened. [12:19] the 1.0 goal should be to have the smart server in operation [12:19] how about we talk about something that has not changed in the last week [12:19] not that it actually buys users anything immediately# [12:19] == Advertising == [12:19] Last meeting actions: [12:19] * spiv: blog about similarities between SVN and bzr checkouts, in relation to Launchpad. [12:19] * ddaa: when rolled out, to blog about branch UI improvements. [12:19] * unassigned: when rolled out, blog about --create-prefix not being needed anymore. [12:19] spiv on vacation [12:20] ddaa and unassigned, waiting on rollout [12:20] moving on [12:20] ddaa: i think we need in the medium term to do something more than blog to communicate about this [12:20] The blog thing appears to be working well enough for new. [12:20] for *now* [12:20] how do you assess that? [12:21] but how do people find these scattered howtos 6 months from now? [12:21] mpool: I throw some chicken bones and read in them [12:21] It'd be good to get them in a central location (or at least linked from a central location) [12:21] any post should be put on bazaar-vcs.org/Welcome [12:21] in the news area [12:21] they should be reachable from launchpad in an organized way [12:21] yeah, we need something more structured, maybe like the Malone/Rosetta docs, and link from that. [12:21] lifeless: somethin g kg like that [12:21] help.launchpad.net maybe? [12:21] we don't need to sort that out now [12:22] indeeed it may be good to seek UI opinions from Uman or mpt or similar folks about where to put it [12:22] action for someone? [12:22] yes [12:22] SteveA: agree? who should do it? [12:23] I'd like you or ddaa to put it on help.launchpad.net [12:23] mh... since mpool is working on critical things like the smart server, it might be a better idea to use my time for that [12:24] ddaa: that would be great [12:24] ACTION: ddaa to start compiling blog data on help.launchpad.net [12:24] to start with, it's just a matter of making a page, linking it to the front page [12:24] SteveA: happy to do no more than that, then :) [12:24] ddaa: actually, to *really* start with, just write a very small spec for how this is documented, and seek comments on hat [12:24] I'd go the other way [12:25] SteveA++ [12:25] have a single place where we collect the documentation [12:25] SteveA: implement first, comments later? [12:25] that is publicly accessible [12:25] and then look for opportunities to better link it and put it on the bzr website or whatever [12:25] but we need to have the resource before we can use it [12:25] I like the idea of producing raw material as blog posts and integrate that material in a central place. [12:26] but, hey, we have more critical stuff to do now [12:26] and jamesh made a good point about how people find the articles N months from now [12:26] Moving on? [12:26] another option is having a "using bzr with launchpad" section on the bzr website [12:26] done [12:27] == vcs-imports knits == [12:27] It appears that the outstanding bzr patches have been merged to bzr.dev, what's needed now is a bzr update on rocketfuel. [12:27] lifeless: can you sync/cherrypick rocketfuel's bzr? [12:27] The last outstanding patch is launchpad/import-batch-progress. Should the BatchProgress code be submitted to bzr upstream, or should it stay in Launchpad? [12:27] https://devpad.canonical.com/~jamesh/pending-reviews/david/launchpad/importd-batch-progress/full-diff [12:27] we should just update to 0.10 [12:27] which has been released [12:27] lifeless: by all means [12:27] that may or may not include all the patches [12:27] I think it has 1 of them [12:27] its also much faster than the current bzr we're running [12:28] can you cherrypick the missing bits as well? [12:28] might even protect us from Jelmer's branches :) [12:28] performance is not our big issue ATM [12:28] you have already applied the fix to our branch right ? [12:29] lifeless: I cannot commit to rocketfuel's bzr [12:29] ok [12:29] I rolled out a custom branch to importd when deploying the knits conversion code [12:29] I suggest just updating to stock bzr 0.10 and keeping your extra fix whereever it is now [12:29] the second fix will be in 0.11 [12:29] and AFAIK we've done all the knit conversions right [12:29] ? [12:30] lifeless: right, but it looks like there's a new upgrade to do soon. [12:30] ddaa: ??? [12:30] about root id [12:30] 0.11 at the earliest [12:30] okay, then it should be okay to temporarily regress on format converision support [12:31] and as that will come in via bzr, I dont see that theres any relevance for these fixes to 'our bzrlib' - them being in the bzr that comes is is all thats needed [12:31] the relevance is avoiding deploying custom bzr code to importd [12:31] but I guess it's no longer necessary, as you point out [12:31] What about the BatchProgress thing? [12:32] what thing ? [12:32] it's the Progress class that gives heartbeat output for buildbot [12:32] Bah, let's move on. [12:32] I dont know anything about that, other than you wrote it [12:32] ACTION: ddaa sort out what to do with BatchProgress with lifeless and mpool [12:33] (we're 2/3 the way through the meeting timeslot, btw) [12:33] == Release finder == [12:33] Last meeting action: [12:33] * jamesh: report on Dyson full test run. [12:33] yeah, we're running late :( [12:33] less discussion, more yes/no. DeferToList [12:34] meta: the thing to do here is either to agree to trip certain agenda points from the meeting, or agree with attendees for an extended meeting. [12:34] It is definitely working better than before, but we got a new exception in the HTTP walker class when trying to read docbook.org indexes [12:34] s/trip/strip/ [12:35] SteveA: I can just speed up some status report items [12:35] Looks like a simple mis-use of BeautifulSoup, so hopefully we'll have a better result next week [12:35] okay, so still in the works, progress being made [12:35] (I should be able to add tests for this failure pretty easily too) [12:35] == Python import == [12:35] https://launchpad.net/products/launchpad-bazaar/+bug/56360 [12:35] Was busy catching up, merging stuff and having design discussions via email. No progress on that. Need to decide on prioritization later this meeting. [12:35] ya, I'm nagging on the release finder regularly ;) [12:35] no need to discuss python import right now, later/post-meeting... [12:36] didn't we have prioritisation discussions for it last week ? [12:36] jamesh: note that I updated our BeautifulSoup in mainline launchpad [12:36] == bzr lp:// == [12:36] Last meeting action: [12:36] * mpool: integrate comments to BranchIndirection [12:36] ...to the latest release [12:36] lifeless: not as such, as far as I recall [12:36] SteveA: yeah. The way the code was using beautiful soup seemed to bypass the encoding detection, leading to problems with the non-ASCII page data [12:37] I think that spec has not changed since last week. mpool, can you spend some time on that? (yes/no) [12:37] jamesh: latest B.S. is stricter about unicode output [12:37] ddaa: mpool, mark and I talked a little about this today [12:38] mpool said he'd update the spec [12:38] okay, see you next week about BranchIndirection [12:38] SteveA: the p-r-f code is doing "soup = BeautifulSoup(); soup.feed(data)", which fails. Doing "soup = BeautifulSoup(data)" works. [12:38] == 1.0 targets == [12:38] [stripping some stuff that needs no discussion] [12:38] bzr-roundtrip-svn: not for 1.0, but discussion is in progress on the bazaar-ng mailing list about how to make bzr-svn acceptable to importd. [12:38] mpool: SteveA wanted to know what you were thinking of the general direction of this discussion. [12:39] i have not updated the BranchIndirection spec [12:39] feeding a soup is not usually adding things to a soup, but rather consuming the soup [12:39] the roundtripping discussion seemed ok to me... [12:40] == Critical bugs == [12:40] [skipping that, simple status reporting, see the BazaarMeetingAgenda] [12:40] action: mbp: read up/tick off svn roundtripping discussion [12:40] FWIW sivs latest work is missing stuff done thursday and friday [12:40] mpool: that would be nice [12:41] == Sysadmin tasks == [12:41] [skipping that, see BazaarMeetingAgenda] [12:41] == Prioritizing important tasks == [12:41] There are several important new tasks competing for our time. Spiv is out being even more smart and serving than usual, so it's between jamesh and me. [12:41] * Removing Arch support. Low risk, relatively large (est. 16h ddaa). 1.0 target. [12:41] * Python import, currently bug 56360. High risk (may have other hidden blockers), unknown size (est. at least 6h, at most ???). [12:41] * Bug 31308 (native productseries branch) Low risk, medium sized (est. 8h). === SteveA wonders about having a "before you go on leave" wiki article... (1. mail people about outstanding work. 2. push all your branches...) [12:41] To that, I would like to add: bug 34540 (cannot delete a branch) which I think is the largest remaining barrier to adoption of the branch registry. Just allowing to disable branches, without implementing garbage collection should be relatively simple. [12:41] * Bug 34540 (cannot delete a branch). Low risk, medium sized (est. 8h). [12:41] Since all those tasks look very important and urgent to me, I suggest: [12:41] * ddaa: removing arch support and python import. [12:41] * jamesh: productseries branch and deleting a branch. [12:41] SteveA: what do you think? [12:42] Okay, that's one item I want a decision on, but can be discussed post meeting (lifeless and mpool not really required for that) [12:42] ddaa: I'm also working on automatic BugBranch link creation now, btw [12:42] I'd like to see the python import set off, cos I want to see whether it will break and how. [12:43] SteveA: IMO arch support removal takes priority as it's a 1.0 goal [12:43] removing arch support doesn't cost us anything to delay a while [12:43] "productseries branch" refers to switching to import_branch and user_branch, right? [12:43] jamesh: yes [12:43] ddaa: it doesn't buy us anything this week, but having a python import does [12:43] SteveA: okay [12:43] ddaa: removing arch support has been turned down by steve in the last 4-5 meetings I can recall [12:44] ddaa: I'd really like it to stop being an agenda item, its wasting cycles [12:44] ddaa proposed it as a task for him and tim p. to do together, which I rejected [12:44] I think ddaa should do it, but I think it is better to do python first [12:44] because it is estimated as less time [12:44] and it has more overall benefit, imo [12:45] after that i'd like to suggest working on a firefox import [12:45] if there is not already one [12:45] they're looking at moving from cvs [12:45] SteveA: maybe you should add "before you go on leave" to the policies wiki? [12:45] ok, piss-break before the review meeting [12:45] mozilla/firefox makes heavy use of CVSROOT/modules aliases [12:45] which is not really supported by importd [12:45] mozilla will melt your brain [12:46] anyway, that's offtopic for that discussion. [12:46] SteveA: jamesh: I think native-branch-for-series and deleting-branch are more important than bug-branch stuff [12:46] are they are important blockers in the existing system [12:47] I'd like very much to see progress on those, but I do not think I will have the time soon. [12:47] ddaa: let's talk about your priorities etc. later today [12:47] Fine. [12:47] I've only done some speccing for the bugbranch stuff, so I can easily move over to the productseries branch stuff [12:47] == Any other business? == [12:48] One minute to say "hey!". [12:48] hey! [12:48] two things [12:48] I wouldn't mind some comments on https://launchpad.canonical.com/AutomaticBugBranchLinks after I flesh it out a bit more [12:48] firstly, everyone in this meeting should read the 32/Bazaar spec [12:48] https://wiki.canonical.com/32/Bazaar [12:49] ddaa, thanks for your commens [12:49] if you have not, please do so before next week [12:49] jamesh: post to the LP mailing list CC key people when you've done the fleshing out [12:49] ddaa: jamesh, mpool and lifeless and I had a meeting earlier today about getting jamesh to do some work on the bug-branch stuff and lp://// stuff. [12:49] secondly, following on from what you just mentioned [12:49] there are a few bzr-lp features we need to focus on [12:49] - launchpad-url [12:50] aka branch indirection [12:50] - redirecting from a branch page to branch content [12:50] - native branches for products, as you just mentioned [12:50] and bug-branch connections [12:50] many of these require both bzr and lp changes [12:50] i think jamesh owns bug-branch connections now, [12:51] how about the others? any comments on who's looking after them, or relative priority? [12:51] or whether they're for lp 1.0? [12:51] my pet peeve ATM is branch-delete [12:51] I really think it's an adoption blocker [12:52] ddaa: i want to learn why you think that, but let's talk later today about it [12:52] branch-delete may be in less demand when branch move/rename is available [12:52] not having any main branch for bzr is a peeve for me [12:52] it is often a case of "I put the branch in the wrong place and now want it deleted" [12:52] jamesh: certainly [12:52] ddaa, it sounded like you worked out with mark the changes that would be necessary to fix that, is that true? [12:53] mpool: there was an email discussion between sabdfl and jamesh while I was on vacation [12:53] mpool: ddaa proposed I look at this bug (see "productseries branch" earlier) [12:53] sabdfl essentially approved the direction we are taking, without reservation, after a few explanations from jamesh [12:53] ok, so it's all good? not blocked? [12:54] blocked on getting somebody to actually do it :) [12:54] If jamesh can do it this week, I'm cool with that. Delete can wait another week or two. [12:55] since it's not a technical prereq for anything else [12:55] ok, well, it's not for me to say who will do it [12:55] that's between you, stevea, jamesh & others [12:55] i just want to make sure it's flowing along [12:56] branch indirection is blocked on me fixing the spec [12:56] for example. [12:56] Okay, I'll put it on the top of my "jamesh please" list :) [12:56] okay [12:56] ok, how about redirecting from pages to brnahces? [12:56] meta: should have agreed an extension to the meeting. [12:56] there are bugs open, i don't know their numbers. === ddaa flogs himself [12:57] setting the productseries branch is a prereq for a bunch of other stuff, so I'll try and get it sorted this week [12:57] (e.g. the branch indirection spec) [12:57] jamesh: that's true; doing it soon would be great [12:57] good point [12:57] mpool: I sugges that you revive the discussion that on the mailing list, so we can take the time to dig up the bugs etc. [12:57] ok [12:57] ok, that's all then [12:58] MEETING CLOSED [12:58] thanks david [12:58] Thank you all for your attention. [12:58] Sorry for being late. [12:58] after the review meeting, i'll be going for some lunch === ddaa will going for lunch now [12:58] and when I get back, I'd like to talk with you about these tasks [12:58] unless somebody wants to talk to me about something right now === ddaa -> lunch [02:48] SteveA: ping [03:29] hi ddaa [03:35] ddaa: ping [03:36] was workraving [03:37] SteveA: you wanted to talk about the tasks for the next weeks [03:38] first, I'd like to talk meta-meeting [03:38] ok [03:38] shall we have a voice call? [03:39] please no [03:39] porquoi? [03:39] ^u [03:39] I have some distraction at home, between the kittens and the gf who is having a one week sick leave [03:40] that's not really a good reason [03:40] well okay [03:40] let's have voice [03:40] skype okay? [03:41] should work [03:41] or ekiga if you prefer, works okay here [03:42] i've had not much success with ekiga [03:42] probably need to change my router [03:42] so, let's do skype today [03:43] i'm running skyep [03:43] setting up [03:43] but i don't seem to have you listed at all in my addressbook [03:44] david.allouche [03:45] I cannot seem to get online for some reason... [03:45] It seems I cannot get online. [03:47] well [03:47] i just tried adding you to my skype contacts [03:47] and I get an error [03:47] "david.allouche was added to your Contacts, even though the Skype Name was not found [03:47] The Skye Name does not exist or this person has not been online for a long time" [03:48] I got online like a couple of moths ago [03:48] but it does not work now [03:48] let's try ekiga [03:49] after a few months, they remove your account [03:50] i'm running ekiga now [03:50] one minute please [03:50] https://wiki.canonical.com/VOIP/Extensions [03:52] bingo, she got angry at me because I wanted to be alone for the call [04:22] oops [04:22] what happen? [04:22] someone set up us the bomb? [04:22] sound got scrambled, I reset the connection [04:23] ok, i just hung up