## page was renamed from ReviewerMeeting20100120 = ReviewerMeeting20100609 = == summary == * jelmer graduates! * deryck points out the noise of 'make lint'. curtis 'volunteers' to replace pylint with a home-grown solution that rocks. * bigjools schools everyone on using the slave db. recommends lib/canonical/launchpad/doc/db-policy.txt as required reading. * discussion about moving code out of canonical/launchpad to complete the apocalypse. vote is overwhelmingly for making this a goal. == logs == === ameu === {{{ [15:00] #startmeeting [15:00] Meeting started at 09:00. The chair is bac. [15:00] Commands Available: [TOPIC], [IDEA], [ACTION], [AGREED], [LINK], [VOTE] [15:00] hi, welcome to the launchpad reviewers meeting. who is here? [15:00] apres moi le deluge [15:00] me [15:00] bac: can I add to the agenda that jelmer graduates? [15:00] me! [15:01] me [15:01] me [15:01] me [15:01] me [15:01] me [15:01] me [15:01] me [15:01] noodles775: i think that will fit nicely in the mentoring update! [15:02] EdwinGrubbs: ping [15:02] me [15:02] me [15:03] me [15:03] me [15:03] well that looks quorum-ish, so let's start [15:03] [topic] agenda [15:03] New Topic: agenda [15:03] * Roll call [15:03] * Agenda [15:03] * Outstanding actions [15:03] * Mentoring update [15:03] * New topics [15:03] * make lint [deryck] [15:03] * Using slave stores to speed up your code [bigjools] [15:03] * ec2 fix update. [15:03] me [15:03] [topic] outstanding actions [15:03] New Topic: outstanding actions [15:04] firstly, i'm lax and didn't even take up abentley's offer of assistance. will do so this week. :( [15:04] * bac and abentley to define new doctest policy regarding what is "testable documentation". [15:04] * sinzui to talk to QA about our QA tracking problem and create a proposal on the mailing list [15:04] sinzui: i didn't see this. was it sent? [15:05] That was co-opted by flacoste who was to have a talk with Urshina [15:05] I think we need to remove this task. [15:05] No I think we need to redefine it. I think we still have a problem [15:05] sinzui: ok. i'll remove it and follow up with francis [15:06] * Bjornt to set a policy on what can live in lib/lp, lib/services, and lib/coop [15:06] not done [15:06] BjornT: ok, roll to next week [15:06] * gary_poster and ursinha to email the list re: orphan branches policy and solution [15:06] because as RM I had two lists of items reporting QA state, and I do not think either was current at the time [15:06] bac: I discussed that with interested parties instead. We ill send out email once work is done. [15:06] we will [15:07] gary_poster: excellent. i'll consider it done then for this meeting [15:07] cool thanks [15:07] new items [15:07] [topic] mentoring update [15:07] New Topic: mentoring update [15:07] noodles775, good news? [15:08] Yup... Jelmer is (and has been for a while) very ready to graduate :) [15:08] Congrats jelmer_ :) [15:08] noodles775: thanks! :-) [15:08] noodles775: ... and thanks for mentoring, it's much appreciated. [15:08] well done jelmer_ [15:08] excellent jelmer_. thanks for your work noodles775 [15:08] congrats to jelmer_ ! [15:09] yay :-) [15:09] jelmer_: please look at the schedule and pick a day that needs EU help [15:09] [topic] * make lint [deryck] [15:09] New Topic: * make lint [deryck] [15:09] bac: ok [15:10] ok, so make lint has become increasingly noisy for me.... [15:10] so much so that it is not so useful anymore. [15:11] deryck: please remove pylint [15:11] 1) can we fix it? I hear sinzui has ideas. and 2) are people reviewing for lint anymore? [15:12] mwhudson, others suggested this, and I don't mind doing that. But I heard sinzui might have a better script altogether from what we have now. [15:12] 1) replace pyflakes with pep8 in a risky shell hack [15:13] 2) extract formater.py from my gdp project to be a standalone linter. It is 100% python, much faster than the shell tools, and it is more accurate [15:14] sinzui: how much effort is 2? [15:14] I have not put looked at it. probably doable in 6 hours [15:14] If others are fine with those suggestions, I'm happy to take on a branch to do this, if sinzui will review it. :-) [15:14] We need to add pep8 to contrib in the tree [15:14] for those using vim, please use the pyflakes plugin, it's awesome: http://www.vim.org/scripts/script.php?script_id=2441 [15:15] sinzui: do you want to take that task, in your abundance of time? [15:15] bigjools: and the one for emacs rocks too! [15:15] I can look at it in 7 hours [15:15] sinzui: thanks. [15:16] bigjools, bac -- I imagine this is how make lint became useless is we all have individual linters, but having a useful make lint is nice when reviewing. [15:16] [action] sinzui to work on extracting a new linter [15:16] ACTION received: sinzui to work on extracting a new linter [15:16] it needs to be accurate though, the noise we get now on the webservice declarations is incredible [15:16] deryck: i agree. when people include the lint output in their MPs it makes me happy. [15:16] deryck, also the lp_review_body plugin will include it in the review body. [15:17] bigjools, my linter uses pyflakes, pep8 and builtin libs to provide s consistent report, I even designed the report to use GUI and command line output (makes testing easy) [15:17] nice [15:17] bigjools: but those webservice declarations are actually things we can fix to silence [15:17] [topic] * Using slave stores to speed up your code [bigjools] [15:17] New Topic: * Using slave stores to speed up your code [bigjools] [15:18] hai [15:18] thanks for the discussion on lint, ya'll. :-) [15:18] so with the snafu on db performance this week and in particular the depwait-scanner, I converted it to use the slave store as much as possible [15:18] deryck: a real southerner knows how to spell y'all! [15:18] dang. [15:18] you y'all is the plural [15:18] anyway [15:19] real southerners can't spell [15:19] I used the information in lib/canonical/launchpad/doc/db-policy.txt [15:19] which is a veritable mine of useful info on slave databases [15:19] I encourage everyone to read it [15:19] adeuring is about to convert bug search to use slaves. [15:19] the depwait scanner runs quite a bit quicker using the slaves [15:20] Not to mention potential future scaling. [15:20] I'm also perusing the code for anything that hard-codes MAIN_STORE which is utterly gross [15:20] I suggest we all do the same [15:20] bigjools: MAIN_STORE or MASTER_FLAVOR? [15:20] and maybe, just maybe LP will be that bit quicker [15:20] Because MAIN_STORE usually makes sense to me. [15:21] I am using the adapter instead [15:21] IMasterStore(object) [15:21] the getUtility is gross [15:21] (or even IStore(class)) [15:21] that doctest doesn't even mention it [15:21] the doctest also shows a cool with: syntax [15:22] anyway, read it guys [15:22] EOT [15:22] thanks bigjools [15:22] welcome [15:22] [topic] * ec2 fix update. [15:22] New Topic: * ec2 fix update. [15:23] so good news from mars on ec2. windmill tests are re-enabled. thanks mars. [15:23] bac, and bad news [15:23] there's a bit of a snafu if you use a symlinked ec2 to test legacy branches, as I do. It will make for some very quick, erroneous successes. [15:23] oh sorry [15:24] about to send out a warning to the list: you MUST merge devel into any branch you plan to run through ec2test for the next few hours [15:24] This mucks up db-devel ec2 runs - sorry, it will be fixed soon. [15:25] More details will be in the list mail I will send shortly. [15:25] thanks mars [15:25] [topic] move code out of c/l [thumper] [15:25] New Topic: move code out of c/l [thumper] [15:25] haha [15:26] in the asiapac meeting last week thumper brought up his desire to start getting more stuff out out canonical/launchpad [15:26] our big code re-org was started in march 2009...so it's been a long while that we've had stuff festering there [15:27] thoughts? [15:27] Devil's advocate (gary_poster prepares for rotten tomatoes): [15:27] Right now it seems like we have core bits like webapp in c/l [15:28] and everything else elsewhere [15:28] I fond things pretty easily [15:28] um, find [15:28] Is the cost of continuing the code reorg going to bring an appropriate level of benefit? [15:29] * gary_poster goes and looks at c/l as he should have before opening his mouth [15:29] gary_poster, oh. I have a branch that removes most of the glob imports and had to remove hacks to webapp imports [15:29] Do we need a lib/lp/webapp? [15:29] lib/lp/core maybe [15:30] lp/services/webapp [15:30] gary_poster, I think that mailnotification.py needs to be refactored into the separate applications so that it can be got rid of altogether. [15:30] Yeah, webapp doesn't describe a lot of its contents well. [15:30] But that might be a separate issue, thinking about it. [15:31] gary_poster, everyone. do not talk about moving these until you have also made plans to deal with shipit [15:31] * sinzui cannot consider asking for a review of super import fix without first decoupling shipiy [15:31] * deryck cries "death to mailnotification.py!" [15:31] part of thumper's annoyance was seeing *new* stuff appearing in c/l. [15:31] is mailnotification.py really a reason to get rid of c/l? [15:32] 1. do not land in c/1, 2. if you must change work with something in it, move it first [15:32] bigjools, +1 on core [15:32] gary_poster, lib/lp is a reason to get rid of lib/canonical/launchpad [15:32] I don't like that. [15:33] gary_poster: which that? [15:33] what sinzui proposed [15:33] abentley, but what's the benefit? [15:33] gary_poster, no, that file should be dismantled anyway. regardless of what happens to lib/canonical [15:33] right now, this is what I see: [15:33] (agreed, deryck) [15:33] gary_poster, consistency. Not having two names for something. [15:34] lp has lots of bits that are specific to the core apps [15:34] gary_poster, I would have been fine with keeping canonical/launchpad, but since we have come this far, we should finish the renaming. [15:34] l/c has stuff that...foundations touches. [15:35] abentley: fair enough, I see your perspective. I don't buy that the cost is worthwhile. I don't buy that sinzui's approach alone is going to get us anywhere except more separation for what foundations touches [15:36] But this is a vote sort of thing [15:36] gary_poster: indeed [15:36] So, I said my piece :-) [15:36] gary_poster, I'm +1 on refactoring. There are no less than six places that the test infrastructure is set up - refactoring at least straightens stuff out: if you find something, you will try to move it to a better place [15:36] gary_poster, it does not get us far because most engineers are not comfortable refactoring the code, and I believe they fear that they do not have permission or time do dtrt [15:37] [vote] We should move towards dismantling canonical/launchpad into more appropriate spots within lib/lp. [15:37] Please vote on: We should move towards dismantling canonical/launchpad into more appropriate spots within lib/lp.. [15:37] Public votes can be registered by saying +1/-1/+0 in the channel, private votes by messaging the channel followed by +1/-1/+0 to MootBot [15:37] E.g. /msg MootBot +1 #launchpad-meeting [15:37] +1 [15:37] +1 received from abentley. 1 for, 0 against. 0 have abstained. Count is now 1 [15:37] +1 [15:37] +1 received from mars. 2 for, 0 against. 0 have abstained. Count is now 2 [15:37] -1 [15:37] -1 received from gary_poster. 2 for, 1 against. 0 have abstained. Count is now 1 [15:37] +1 [15:37] +1 received from adeuring. 3 for, 1 against. 0 have abstained. Count is now 2 [15:37] +1 [15:37] +1 received from gmb. 4 for, 1 against. 0 have abstained. Count is now 3 [15:37] +1 [15:37] +1 received from bigjools. 5 for, 1 against. 0 have abstained. Count is now 4 [15:37] +1 [15:37] +1 received from bac. 6 for, 1 against. 0 have abstained. Count is now 5 [15:37] The easy work was done. Most of the remaining code does not have an obvious place. We often need to know the tree names to do a rename, ans we do not know them for these service-like code [15:38] +1 [15:38] +1 received from henninge. 7 for, 1 against. 0 have abstained. Count is now 6 [15:38] That's probably enough, isn't it :-) [15:38] +1 [15:38] +1 received from sinzui. 8 for, 1 against. 0 have abstained. Count is now 7 [15:38] voting ends in 10 seconds [15:38] +1 [15:38] +1 received from jelmer_. 9 for, 1 against. 0 have abstained. Count is now 8 [15:38] [endvote] [15:38] Final result is 9 for, 1 against. 0 abstained. Total: 8 [15:38] lol [15:38] :-) fair enough [15:39] I saw pagetitles.py was updated for shipit yesterday. we do not support page titles, but shipit and some pages are still using the dead code [15:39] easy for every who's *not* doing the work to +1 that :) [15:39] the results won't be official until the new zealand and colorado precincts report [15:39] everyone* [15:39] Who is fixing shipit? [15:39] heh [15:40] bigjools: yeah, signaling our intent is easy. doing the work is harder. [15:40] (well, I'm not going to be keen on scheduling foundations to do much of that to be honest, except as part of other refactoring, as sinzui said.) [15:40] gary_poster: i think that is appropriate. [15:40] cool [15:40] [topic] peanut gallery [15:40] New Topic: peanut gallery [15:40] I will post an email about names. The migrater script can move most of the chunks of code if we have names for them. [15:41] any other topics in the remaining 4 minutes? [15:41] actually, I was wondering if anybody had any good suggestions for my regular review day [15:41] no one was around this morning when I needed a review. :-) [15:41] I'm on thursdays together with noodles775 at the moment but according to the schedule we already have one person for each weekday in europe [15:42] jelmer_: it looks up to you. every day has EU coverage [15:42] Friday - it's the best day; Just ask sinzui [15:42] unless pqm is closing that day ;) [15:42] shhh [15:42] heh [15:42] :-) [15:43] jelmer_: I think you could support jtv because he is not really in EU timezone [15:43] Isn't allenap in the proper EU slot? [15:43] good suggestion, henninge [15:43] I give out RCs before PQM closes. I think you do want to ask me for the review in that case [15:43] jtv: allenap is on rotation [15:43] facepalm [15:43] of course [15:43] thanks for coming everyone and for the good discussion [15:44] #endmeeting [15:44] Meeting finished at 09:44. }}} === asiapac === {{{ [22:40] Let's start the NZ-CO-NC reviewers meeting [22:40] * rockstar [22:41] you don't say? [22:41] so we had a lively meeting this morning [22:41] no one actually got to any of the outstanding actions. i think i should remind people on mondays but i never get around to it... [22:42] deryck whinged about 'make lint' spewing way too much junk to be useful. [22:43] sinzui has a plan to replace pyflakes with a tool he's written. claims it is much more accurate and faster. [22:43] not sure when he'll get to it. [22:43] bac, I think pyflakes is okay, it's pylint that's annoying. [22:43] bac, a few times I've been close to just changing the Makefile to kill the pylint call and asking forgiveness later. [22:43] it is possible i got those two confused [22:44] anyway the plan is for sinzui to swap out the crap one with something better. [22:44] I love pyflakes, but pylint is more an exercise in adding ignores than anything else. [22:44] and then we all buy him beer [22:44] bac, many of us run pyflakes in $EDITOR and it works pretty quickly. [22:45] rockstar: right. i do in emacs. bigjools pointed out the vim plug in. [22:45] (and no flame war erupted) [22:45] I run pyflakes, pep8 in my editor + real xml parsing the checks entities [22:45] sinzui, aren't you using gedit? [22:45] bigjools also brought up the huge gains he got by converting some code to use the slave store. [22:45] yes [22:45] * rockstar shudders [22:45] I wrote the plugin [22:46] he suggested lib/canonical/launchpad/doc/db-policy.txt should be required reading for all of us. [22:47] mars talked a bit about his fixes to ec2 and mentioned some short-term pain wrt db-devel [22:47] but the good news is it looks like he's got a handle on getting it to be more reliable and that makes me very happy. [22:48] and then i brought up thumper's concern from last time about moving stuff out of canonical/launchpad [22:48] we had a good discussion where gary questioned whether it was worth the effort. [22:49] everyone else thought it was worthwhile. but it will remain as a techdebt issue to be handled as we have time. [22:49] I think it is worth the effort :) [22:49] yeah, i figured it would get one or two votes here [22:50] oh and the other good news was that jelmer has graduated as a reviewer. [22:51] did everyone see flacoste's email suggesting a 'text' review type for mrevell to look at textual changes? sounds like a good plan to me. [22:52] that's about it from the AMEU meeting. [22:52] either of you have new topics? [22:53] going once [22:53] no [22:53] ok. well i guess that's it then. [22:54] thanks bac [22:54] i'll see you later. }}}