=== wallyworld changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: wallyworld | Firefighting: - | Critical bugtasks: 4*10 [00:03] lifeless: Around? [00:08] yup [00:08] lifeless: I'd like to discuss bugsummary-ng at some point. [00:09] Making it compatible with disclosure, and also making it performant. [00:09] And fixable. [00:10] voice? [00:10] Anything else would be likely to result in the heat death of the universe. [00:12] skynet? [00:12] Give me a sec to get it running, reinstalled on Monday. [00:12] 1 [00:12] fuuuu [01:16] wallyworld: O hai [01:16] yello [01:17] wallyworld: I've put two new photos in the river directory taken this morning which show the roughly 'normal' river level [01:18] remind me of the dir name? [01:18] wallyworld: http://wedontsleep.org/~steven/river/ [01:18] wallyworld: The two 03-09 photos are the normal leve [01:18] big difference [01:18] *level [01:19] Oh yeah [01:20] did you go for a swim? [01:20] Haha, no [01:21] It was quite impressive given how fast the water was flowing when it was up that high [01:22] indeed [01:22] IntegrityError: duplicate key value violates unique constraint "accesspolicy__product__type__key" [01:22] :-( [01:27] StevenK: Only in test_bug_mirror_legacy_access? [01:27] Possibly in wallyworld's tests as well, I guess. [01:31] It was about sixteen tests [01:31] I've just merged devel and am running product tests manually to see if wallyworld has made life harder for me. [01:32] who me? [01:32] lp.bugs.tests.test_bug_mirror_access_triggers.TestBugMirrorAccessTriggers.test_productseries_task is one failure [01:32] That's one of mine. [01:32] Easy to fix :) [01:33] Ah yes [01:33] * StevenK hugs alt-q in zsh [01:39] * StevenK fixes wgrant's test with +1/-21 [01:42] lp.registry.tests.test_accesspolicy.TestAccessPolicySource.test_create [01:42] lp.registry.tests.test_accesspolicy.TestAccessPolicySource.test_find [01:42] Those two :-( [01:44] Hmmm, they're going to be difficult now. [01:44] You may want to fix them to manually create a Product [01:44] Rather than use ProductSet via the factory [01:44] Ah [01:44] Or just use PROPRIETARY instead [01:52] * StevenK kicks TestAccessPolicySource.test_findByPillar hard [01:53] I can't use a tuple to describe that types that already exist, since they're returned as APs [01:54] And I don't want to just call findByPillar since that defeats the entire point of the test. [02:08] hi bigjools [02:10] StevenK: Why can't you use a tuple? [02:10] StevenK: I use tuples to match them throughout that test file. [02:12] wgrant: It didn't work. I'll keep digging after lunch. [02:16] It's not magic. It works. [02:27] so I whinged to telecom yesterday [02:27] my adsl is being line checked overnight [02:27] guess what hasn't happened today [02:29] It seemed to be being a bit bad this morning. [02:29] But maybe it was just Skype. [02:29] Rather stuttery. [02:31] yah [02:31] but wrkoing [02:32] True [03:03] * wgrant narrows eyes [03:03] [r=deryck][bug=922741] Remove ec2 scripts from launchpad tree [03:03] added: lib/lp/codehosting/scripts/tests/test_upgrade_all_branches.py [03:06] wallyworld_: Feel like reviewing https://code.launchpad.net/~wgrant/launchpad/bug-950502/+merge/96701? [03:18] stub: good mornink [03:18] morning [03:21] wgrant: hah, dirty patch? [03:21] stub: whenever you are caffeinated etc [03:21] Now is fine. Skype? [03:21] yupyup [03:24] lifeless: Hm? [03:24] wgrant: the ec2 + upgrade all thing [03:25] lifeless: The ec2 removal was a prereq for the upgrade all thing. [03:25] how so ? [03:25] But they were landed together with a rather misleading message. [03:26] ec2 depended on system bzrlib plugins [03:26] upgradeall reworks how we load plugins, which ends up shadowing the system ones. [03:29] * wgrant hacks SSO to pieces. [03:37] wgrant: http://pastebin.ubuntu.com/875545/ [03:38] StevenK: Or just change makeAccessPolicy to use a non-default type. [03:38] And you're neutering the other tests somewhat, as you're no longer testing various types. [03:39] Well, sure, but the APs *are* there [03:39] So if I can use them, it's a good idea [03:39] The fist hunk is also wrong. [03:39] find() returns a list, not a policy [03:41] Right, that fixed test_updateDistroSharee [03:41] And the other one is pretty obvious. [03:42] Iterate over getUtility and return a tuple [03:44] wgrant: i was out to lunch with bigjools. i'll look now [03:45] Is there something going on there we should know about? :) [03:45] Haha [03:45] bigjools was saying that he was still thinking about wallyworld_ after getting home. [03:46] Maybe I should ring Belinda. [03:46] she was watching [03:47] Only watching? :-P [03:51] rubbing [03:51] * StevenK puts himself up as OCR, so he can review his lunch. [03:51] wgrant: i think you should add the new flag to flags.py in your branch [03:54] wallyworld_: Mmm, I deliberately chose not to, but I guess I might as well. [03:55] wgrant: well, if there's a good reason not to [03:56] I'm lazy and it's a short-lived flag? :) [03:57] wgrant: Did you want me to not remove InformationType.USERDATA from test_create and test_find ? [03:57] StevenK: That was there to test that it actually filtered properly by type, not just by pillar. [03:58] I can add UNEMBARGOEDSECURITY instead if you wish [03:58] wgrant: for clarity,and given it's a few lines of cut and paste, you should add it i think [04:03] wallyworld_: Done [04:03] thanks :-) [04:03] lifeless: Do you know if Orange plans to remove the old bug listings eventually? [04:07] wgrant: r=me. thanks [04:07] wallyworld_: Thank you. [04:12] wgrant: I believe its on deryck's plate, as is removing the in-browser results cache [04:12] wgrant: remind me tuessday if you want me to ping him about it [04:13] * StevenK stabs buildbot a few times with a rusty knife. [04:13] lifeless: It's fairly large and annoying tech-debt, so it'd be nice if it went away. [04:13] Because having a CI system that's not INSANE is too hard. [04:13] wgrant: remind me tuessday if you want me to ping him about it [04:13] k [04:13] Yay, xmlrpc-private 99% down from 1.7s to 0.88s in two weeks [04:13] Next stop 0.3s. [04:14] If I can't deploy this DB patch tonight due to buildbot losing it's mind, I'm going to be very very cranky. [04:17] wgrant: what's the min and median for xmlrpc-private? if you have that? [04:18] mwhudson: I don't think we have those. But there's mean. [04:18] https://devpad.canonical.com/~lpqateam/ppr/lpnet/latest-daily-categories.html and https://devpad.canonical.com/~lpqateam/ppr/lpnet/latest-daily-pageids.html [04:19] Interesting [04:19] That says the 99% was 0.74 [04:19] But lpstats says 0.88 [04:19] Maybe lpstats lags more than a day behind. [04:19] Oh damn it [04:19] (mean 0.14) [04:20] mean of 0.14, not bad [04:20] Updating devel has made bin/ec2 disappear [04:20] StevenK: bzr branch lp:lp-dev-utils somewhere [04:20] somewhere/ec2 land [04:20] Well, duh [04:20] StevenK: it would have been more fun if it lost rf-get [04:20] mwhudson: CodeImportSchedulerAPI has a 99% of 0.68 [04:21] mwhudson: Which is a worry [04:21] Considering what it does. [04:21] No, rf-get is copied to /usr/local/bin by rf-setup [04:21] So you can't lose it on purpose [04:21] AH [04:21] wgrant: well, it used to time out a lot ... [04:21] mwhudson: Sure, but there's no justification for it to take more than 20ms. [04:21] or was that the other bit [04:22] wgrant: heh well, can you get any xml-rpc method to take less than 100ms? [04:22] There's 60-70ms of overhead [04:22] Which I'm going to tackle soon [04:22] my impression (from a few years ago) was that the maximum performance from xml-rpc methods was not very good [04:22] But <100ms is doable at present, yes. [04:22] ah ok [04:23] In fact, Aaron broke it [04:23] bin/ec2 still exists [04:23] StevenK: It doesn't still exist. [04:23] Only in your existing branch. [04:23] Because buildout doesn't delete things. [04:23] 70ms of overhead is still pretty rubbish of course, i hope you can make inroads there [04:23] mwhudson: That's the plan. [04:23] Right [04:23] ironically the twisted-based authserver had a way better base line [04:24] mwhudson: AuthServerAPIView is 0.32/0.05 [04:24] So 50ms is doable. [04:24] Which is interesting. [04:24] But the queries in there should be roughly 1-3ms. [04:24] So there's still a fair bit of overhead. [04:24] wgrant: i wonder if codeimportschedulerapi rollbacks relatively frequently [04:25] I really want to throw Zope out a window and use Pyramid instead. === jtv1 is now known as jtv [04:25] That would be a fun branch to review. [04:25] mwhudson: It's possible. [04:25] mwhudson: But given the way it works, I wouldn't expect it to conflict with the scanner much. [04:25] And not much else should be locking rows for long very often. [04:26] wgrant: well [04:26] wgrant: there are 4 machines that call it, right? [04:26] Surely not. [04:26] Hmm. [04:26] */1 cronjob, I guess... [04:26] plausible [04:26] right [04:27] wgrant: make clean && make and bin/ec2 still exists [04:27] it could also be missing an index i suppose [04:27] StevenK: Does make clean actually remove stuff from bin/? [04:27] steven@liquified:~/launchpad/lp-branches/devel% ls -lh bin/ec2 [04:27] -rwxr-xr-x 1 steven steven 616 2012-03-09 15:23 bin/ec2 [04:27] steven@liquified:~/launchpad/lp-branches/devel% make clean >/dev/null [04:27] steven@liquified:~/launchpad/lp-branches/devel% ls -lh bin/ec2 [04:27] ls: cannot access bin/ec2: No such file or directory [04:28] mwhudson: I guess that's actually not implausible, given how small the table is. [04:28] ah right, codeimportjob is kept small [04:28] i guess there is some query you can run to find out if it's being seqscanned [04:28] wgrant: So, yes, it does. [04:29] oy oy oy sqlobject methods [04:35] (wifi fail) [04:36] Sure sure [04:36] StevenK: deploying a db patch on friday - keen :) [04:36] StevenK: Lies [04:37] mwhudson: the min and median are on the PPR [04:38] Ah, median is, but I can't see min... [04:38] lifeless: oh right, i see median now, don't see min though [04:38] connection reset from SSO. Hmmm. [04:38] wgrant: It does! [04:38] StevenK: I'd say your devel is out of date. [04:38] No, Aaron didn't fix setup.py [04:39] Possibly not, but I did. [04:39] steven@liquified:~/launchpad/lp-branches/devel% bzr revno [04:39] 14926 [04:39] Revision: 14927 [04:39] Commit Message: [r=wgrant][no-qa] Remove ec2 leftovers. [04:40] I bet you fixed while I was putting together a branch to do the same thing, you cheater. [04:40] Busted. [04:41] However, lp-dev-utils/ec2 is broken [04:41] sudo apt-get install python-tz python-boto [04:41] bzr branch lp:bzr-pqm ~/.bzrlib/plugins/pqm [04:41] s/bzrlib/bazaar/ [04:42] Nice to know it just works out of the box [04:42] [04:42] At least it's not SSO or Launchpad :) [04:43] * StevenK removes his local branch that destroys the ec2 leftovers and glares at wgrant. [04:44] :) [04:44] Like I'm going to let you get the LOC credit. [04:44] I was about to commit and push! [04:45] oh look [04:45] *that* was ADSL [04:45] fail [04:46] * StevenK brands lifeless' connection as Abominable Digitial Subscriber Line. [04:47] mwhudson: https://devpad.canonical.com/~lpqateam/ppr/lpnet/latest-daily-categories.html [05:07] lifeless: fyi, I put out a new release of pygpgme last night. The main new feature is Python 3 compatibility, which I doubt you need yet but thought you might want to know [05:19] jamesh: thanks for the headsup [05:19] wgrant: LOC for what? [05:19] lifeless: while the package should still support older Python versions, the test suite probably requires 2.6 now though [05:35] wgrant: LOC for what? [05:37] lifeless: I beat StevenK to deleting some ec2 remains. [05:38] wgrant: ah,c ool [05:38] I thought that was address to me, for some reason [05:38] SSO's dependency system might be even worse than Launchpad's, I think. [05:38] !cite [05:40] It's very good at branching the same stuff from LP every time you ask it to update deps. [05:40] Hah === almaisan-away is now known as al-maisan === wallyworld_ changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: - | Firefighting: - | Critical bugtasks: 4*10 [07:27] morning [07:54] lifeless: What do you think about pages like https://launchpad.net/ubuntu/+ppas? The slow stats can all reasonably be publicly cached for a day, or we could just remove them. [08:01] hey poolie, did you want to talk about that bug? [08:13] wgrant: so, I think the page should be just a search widget, without that ridiculous 'show me empty ppas' [08:13] wgrant: the stats aren't (IMNSHO) interesting, nor are the lateset uploads to any ppa [08:14] activity -might- be interesting, but probably best as a ranking component for search results [08:14] the supported series list probably wants to be something like [08:15] 'PPAs support . The current list of Ubuntu versions that can be targeted in a PPA build is ' [08:51] lifeless: Latest uploads might be vaguely interesting, and are quick to calculate. [08:54] gd mrning [08:57] wgrant: 'might' isn't a good reason for clutter [09:31] lifeless: Lies. [10:02] You don't need a reason for clutter. === adeuring changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: adeuring | Firefighting: - | Critical bugtasks: 4*10 === al-maisan is now known as almaisan-away === matsubara-afk is now known as matsubara === danhg_ is now known as danhg [13:19] salgado: mrevell any idea on when the roll out will now happen for the blueprints [13:20] czajkowski, The migration of Ubuntu work items will take place, most likely, after the 12.04 release, at a time to suit the Ubuntu Engineering team and wider Ubuntu community. So, not sure of an exact date. [13:20] will it happen before UDS ? [13:21] as bleprint creation will start once 12.04 is released for next release and for uds sessions [13:22] czajkowski, I think that's a decision for the Ubuntu community. We'll fit with whatever schedule works for them. It's possible that the work items feature could go live before we migrate the existing Ubuntu blueprints to the new format. [13:24] nods ok [13:24] thanks === almaisan-away is now known as al-maisan [14:04] Morning, all. [14:07] deryck: morning get your machine all up and running [14:08] czajkowski, hey. unfortunately, still busted launchpad-developer-dependencies on precise for me. [14:08] will be following up to my email shortly. [14:10] :( [14:20] sinzui: with the sprite issue, if you goto a bug page eg https://bugs.launchpad.net/launchpad/+bug/950562 you can see examples in the bug title and security portal [14:20] <_mup_> Bug #950562: Sharing service needs to support 'Some' < https://launchpad.net/bugs/950562 > [14:21] i'll revert the css though, but i think there's an underlying css issue we have [14:23] wallyworld_, I see gecko has issues, webkit does not! [14:23] ffs [14:24] wallyworld_, We will look at a proper fix next week. I agree this needs fixing and I have some ideas [14:24] ideally we need to alter our css to address these x-browser issues ? [14:26] we often need to add a reset rule to before we apply our rule, or we look for a proprietary selector/property that compensates for the mis-aligned background [14:27] ok [14:28] maybe one day there quirks will go away [14:28] these [14:29] wallyworld_: i think a nontrivial part of the web economy thrives on the quirks. don't bet on them ever vanishing. what you call a pain in the ass, someone else calls "competitive advantage". :-P [14:29] lol [14:29] sadly true [14:30] yup. it's only funny cause it's heartbreaking. :-P [14:30] yeah, i wanted the sharing page to look real nice like [14:30] and truncated sprites messed it up :-( [14:30] the sprites things is so irritating. [14:31] Looks like firefox does not believe line-height is inclusive with the space allocated to the padding. this looks like a violation of the box model [14:31] sinzui: able to chat at say 10? [14:31] stupid firefox [14:31] yes, but it need to get more coffee. This cup killed a passing hnat [14:31] gnat [14:32] wallyworld_, keep you sprite hack! [14:33] really? [14:33] I agree it fixes firefox. It does change the webkit layout. subtly, but I think you need to look hard to see the spacing change. [14:34] i could make it 2 or 3 pixels instead of 4 [14:34] just enough to fix the issue [14:34] * sinzui checks [14:34] i tried 3 and i think it worked ok [14:35] keep 4 [14:36] will do. i'll fix the other issues in the morning and send off the ec2. [14:36] I I think the actual space should be 6 (18 line-height - 12 font-size = 6px) [14:36] to [14:36] ok, so change it to 6 [14:36] * sinzui checks [14:40] wallyworld_, yes, use 6px. webkit has already allocated that same space as a part of the box allocated to the padding [14:41] ok, will do. thanks for the help. i'm pleased this will fix other pages as well as sharing [14:41] wallyworld_, The edit icon next to a header moved slightly in webkit (different line-height rules), but there is no change in normal budy text [14:41] This fixes the bug page and the project front page [14:42] PS, it is after midnight for you. It is Saturday. have you been drinking? [14:42] cool. i'll also address the other anchor stuff you mentioned [14:42] no, sadly not. [14:42] i checked email to see if you had +1ed my mp [14:43] cause i want to get everything landed [14:43] I am doing that now. I am commenting about my reverse decision [14:43] sadly, we can't turn on the fflag since edits will mess up stuff [14:44] and the sharing page is kinda useless if you can't edit stuff [14:44] I think pointing people to staging and qastaging is good enough for the curious [14:44] sounds ok to me [14:44] deryck: in the dark recesses of my mind, i think i had to manually install postgresql-8.4-debversion to allow the lp database packages to be installed [14:45] but that issue may have been fixed since then [14:45] that's the only thing i can contribute to your problem [14:54] wallyworld_, yeah, I did try that. Just tried again now and get: postgresql-8.4-debversion : Depends: postgresql-8.4 but it is not installable [14:54] wallyworld_, and hey, btw. :) How's everything going? :) [14:54] deryck, you need to pull the packages from oneiric [14:55] They Lp is running on obsolete PG [14:55] sinzui, ah. and how do I do that? Sorry to be so dumb about this. [14:55] jcsackett, I am now ready [14:55] sorry for the dealy [14:56] I download them from Lp. I think you need for. [14:56] * sinzui looks [14:57] sinzui: no delay at all. i asked about 10, it's now 10. :-) [14:58] sinzui: the licience review of a project which had a made up licience I take it you denied it ? [14:58] sinzui, ah, I get you now. I can chase down the packages myself now. Thanks. [14:59] deryck, See the binary packages listed on https://launchpad.net/ubuntu/oneiric/+source/postgresql-8.4 [14:59] We want postgresql-plpython-8.4 [14:59] postgresql-client-8.4 [14:59] postgresql-contrib-8.4 [14:59] postgresql-doc-8.4 [14:59] postgresql-plpython-8.4 [14:59] sinzui: hangout, mumble, skype? [15:00] hangout [15:00] sinzui: cool. starting one now. [15:02] sinzui, many thanks! [15:06] Aren't all those packages in the PPA? [15:07] oh.. thinking of the lucid backports [15:50] sinzui: you vanished! :-P [15:50] I did. [15:50] Phone rang [15:50] dig. [15:51] i think we were done though, so all is well. thanks. :-) [15:51] fab === al-maisan is now known as almaisan-away === salgado is now known as salgado-lunch === salgado-lunch is now known as salgado === matsubara is now known as matsubara-lunch === deryck is now known as deryck[lunch] === matsubara-lunch is now known as matsubara === danhg_ is now known as danhg === deryck[lunch] is now known as deryck [18:29] deryck, btw, re your emails about LP on precise, you can develop LP with lxc on precise, and then you are using lucid packages. can be nice. that's what I do. You don't pollute your main system that much either--getting rid of LP would mean getting rid of the LP directories in your home dir and getting rid of the associated LXC instance(s). We can help if you'd like. There's a wiki page about doing it manually, [18:29] and a script to give some automation. I'm guessing you are past the point that this would have been helpful, but I still wanted to mention it [18:32] gary_poster, your remark implies you do not know about .rocketfuel-env.sh http://pastebin.ubuntu.com/876388/ [18:33] I have never had my home directory polluted [18:33] gary_poster, ah, that actually might be nice. I'm *more-or-less* working. :) But if I could get into the more category than the less, might be worth it. :) [18:33] gary_poster, can you point me at the wiki page? [18:34] sinzui, I was vaguely familiar with that, but I don't mind having stuff in my home directory. The pollution I mind is having everything else installed on my system (postgres being taken over, tons of packages installed, and so on) [18:34] but to each his own in this regard I'd say :-) [18:34] and thank you [18:34] deryck, https://dev.launchpad.net/Running/LXC [18:34] deryck, we use automated scripts now, so we haven't touched this in a while, but I think it will still be good [18:35] gary_poster, awesome. I'll have a look now. [18:35] cool [18:35] does that have wgrants and wallyworld_'s updates? the instructions did not work for precise last week [18:35] * sinzui looks [18:36] well I has Ian's and I think that means he got it running [18:40] deryck, keep in mind Lp will be running Precises in production in a few months. Lp engineers we need enough engineers running precise to be sure we can switch over without dedicating a team to make it happen in the chain [18:40] sinzui, I was actually just thinking about that. [18:41] sinzui, deryck, you can also create a precise container [18:41] lxc is still nice [18:42] gary_poster, yes. I agree that keeping out ridiculous Lp deps in a box is best. [18:47] deryck, You are doing a fresh install aren't you. My upgrade to precise was easy...given that I have learned to pin packages to ensure dist-upgrade does not remove my income. [18:48] sinzui, not really a fresh install. My laptop came preinstalled with Oneiric, then I got LP running, then I upgraded to Precise.... [18:48] sinzui, however, I didn't pin packages, so it removed all of lp, which is what got me in this shape. [18:48] well, all of lp deps. [18:49] deryck, right update removed all PPA from dep checks to ensure the update goes well. You have to say no to all removes. You can them pin the one you need see issues with on reboot [18:50] I have pgsql and rabbitmq pinned === danhg_ is now known as danhg [20:04] * sinzui high fives StevenK for demonstrating that Lp can give any project a complimentary 30 day commercial subscription === cody-somerville_ is now known as cody-somerville === matsubara is now known as matsubara-afk === salgado is now known as salgado-afk