[00:58] time to extract the logging oops code I think [00:58] poolie: you wanted to talk qbr ? [01:21] hi there [01:21] sure, if you have some time we could chat [01:21] otp at the momemnt but won't be long [01:22] cool [01:22] I'm on sky.net, or my landline, ring whenever [01:32] Morning! [01:32] hola [01:34] wgrant: so in the qas mailbox, there's mail for when new bugs get created but no mail for comments or privacy changes etc. am i doing something wrong? [01:34] nigelb: hello [01:34] wallyworld_: You've run send-bug-notifications.py? [01:34] wgrant: no. i though the cron job was set up on qas to do that [01:34] but no? [01:34] I don't think so. [01:35] i need to ask to get that script run then? [01:35] Yes [01:35] thanks [01:35] The general rule is that if something would be convenient and easy, qastaging has to have it done manually or it's impossible. [01:35] hah [01:35] with the new script server for [qa]staging [01:35] we'll get more things running [01:35] still need more db grunt for some tasks though [01:36] wallyworld_: you should never remove bad-commit-xxxx tags [01:36] wallyworld_: they control the bad-revision-range marking for the qa-tagger. [01:37] wallyworld_: *if* we were to rollback prod, not having it would be bad [01:37] oh, sorry [01:37] Hrm, no stub yet. [01:37] (think of it this way - that old revision is still bad, right ?) [01:37] yes [01:39] How does one get a run a garbo job running? [01:39] nigelb: there's cron scripts [01:39] garbo-hourly.sh etc [01:39] .py [01:39] so, that script has to get checked in? [01:40] sorry, [01:40] the cron scripts run garbo-hourly.py [01:40] which you can run yourself [01:40] on your dev system [01:40] I want it run on production :) [01:40] it happens automatically [01:40] just land the change, qa it, deploy. [01:41] ah [01:41] okay [01:41] for qa you'll need a paid dev help, to get losas to run it and then check the db [01:42] It still gets checked in even if its a one time thing right? [01:42] yes [01:42] code land qa deploy wait delete land qa done [01:43] wow [01:43] fixing tech debt is *hard* [01:43] I need to do 6 landings for this bug. [01:43] 3 in devel and 3 in db-devel. [01:44] 3 in db-devel ? [01:44] Yeah [01:44] so if we can avoid introducing tech debt in the first place...... [01:44] hheh [01:44] one for the new column, one to drop the old columns. Whats the third db-devel one ? [01:45] indices I guess. [01:45] ah, no [01:45] two. [01:45] I counted the garbo jobs in db-devel last time. [01:45] bug 5283 [01:45] <_mup_> Bug #5283: "Home page" vs. "Description" is misleading

< https://launchpad.net/bugs/5283 > [01:46] so in devel you need: [01:46] 1- add new column, change appservers to *write* to that column as well as the old columns, and introduct garbo job. [01:46] lifeless, hi, how about now? [01:46] 2- change appservers to read from the new column and drop the garbo job [01:46] 2 landings AFAICT [01:46] poolie: sure [01:47] lifeless: cool! that's better. I was counting each step as a differnt landing :) [01:48] OTP now [04:21] stub is awake but being productive off IRC. HA. [04:38] The bug privacy APIs are a mess :( [05:09] huwshimi: O HAI [05:10] StevenK: Hello! [05:10] huwshimi: I pointed you at https://code.launchpad.net/~stevenk/launchpad/include-ppa-url/+merge/76499 last week, but I missed your answer. Can we have a quick mumble about it? [05:11] StevenK: Oh right, sure [05:14] wgrant: Can you join mumble for one tick? [05:14] lifeless: How long since you've talked to Curtis about disclosure? [05:14] lifeless: eg. what we discussed last week about the mess around special-casing security? [05:29] several weeks [05:30] if you want me in on a stand up or party to a discussion - just let me know [05:30] Oh, JS, please die on a fire [05:30] lifeless: There are branches flying around now to sort of finish the partial implementation of that, so it probably needs discussion RSN. [05:30] StevenK: Eventually, everything will be in JS :p [05:31] Continue is not the right button in Firebug, what's the right one to step to the next statement? [05:32] lifeless: Doing anything except special-casing security is currently believed to be out of scope :( [05:33] Because stakeholders have not asked for it. [05:33] But I don't think we should break our privacy support just because stakeholders don't ask for it to not be broken :/ [05:34] StevenK: Next? [05:35] err [05:35] "Step Into" [05:35] BUt you may want to add more breakpoints [05:35] I generally breakpoint like 6 to 7 lines [05:38] wgrant: so, arrange a 3-way with you me and curtis minimally [05:38] wallyworld_: The event from clicking the ok button doesn't fire :-( http://pastebin.ubuntu.com/697061/ [05:42] wgrant: The bug is bug #859386. Would you mind triaging it appropriately and editing it so that it makes sense? [05:42] <_mup_> Bug #859386: Merged teams with ppas are still linked to from +archivesubscriptions < https://launchpad.net/bugs/859386 > [05:45] huwshimi: Thanks. I've duped it against but #684112 [05:45] <_mup_> Bug #684112: +archivesubscriptions shows links to merged users. <404>

< https://launchpad.net/bugs/684112 > [05:45] bug #684112 [05:45] <_mup_> Bug #684112: +archivesubscriptions shows links to merged users. <404>

< https://launchpad.net/bugs/684112 > [05:46] wgrant: oh right, thanks :) [05:46] hi all, could i get a review of https://code.launchpad.net/~mbp/launchpad/858618-affecting-me/+merge/76880 (and a few other small ones) [05:49] huwshimi, what do you think of my screenshot on bug 858618 - i presume it's unobjectionable for ui but thought i'd ask [05:49] <_mup_> Bug #858618: hard to find bugs that affect you in a project < https://launchpad.net/bugs/858618 > [05:54] poolie: I think it fits nicely with the other content there. Such a lovely little feature :) [05:54] yeah [05:54] copy & paste ftw! [05:55] huwshimi: hey [05:55] I was askign to chat with you [05:55] nigelb: Hey there [05:55] i hope also to soon do /~/+affectingbugs but i didn't quite get to it [05:55] the implementations of bugs/~mbp and bugs/bzr are alarmingly different [05:55] poolie: Yeah that would be nice too [05:55] to my eyes anyhow [05:56] huwshimi: How would one go about overriding the default CSS on one particular page? (Without killing kittens :P) [05:56] o/ nigelb [05:56] Morning poolie! [05:57] nigelb: I'm not sure exactly what you mean (well, I understand the kittens bit). [05:58] huwshimi: It fix for bug 806660 is to split the form into two tables. This adds some extra margin. [05:58] <_mup_> Bug #806660: "Add a new address" in e-mail settings does the wrong thing when pressing Enter < https://launchpad.net/bugs/806660 > [05:58] The margin is part of the default CSS, which I want to override and set to 0. [05:59] s/It fix/The fix/g [06:03] poolie: if thats going in the side portlet [06:03] poolie: you'll want to test the performance carefully, or feature flag it [06:03] poolie: the portlet has some inefficiences in it [06:06] yeah it's doing one more query [06:10] nigelb: Do you have the code pushed to LP yet? [06:10] lifeless, what would be a reasonable test? [06:11] poolie: performance of that query on the most-affected person in Ubuntu [06:11] if its under say 100ms its probably ok for now [06:12] huwshimi: No, its in a branch. [06:12] let me see if I pushed the branch [06:13] lifeless, and, sorry how can i most easily find out the query time? [06:15] huwshimi: this is what I've done so far. https://code.launchpad.net/~nigelbabu/launchpad/registry-email-add-806660/+merge/75676 [06:15] nigelb: Ah cool thanks [06:15] I have the manual style override in my local branch, but I didn't push it. [06:18] nigelb: So overrides normally happen by either applying an existing class (you'll have to hunt through and see if there are any generic classes that do what you want. Failing that try and write a generic class for the situation. Failing that write a more specific class. [06:19] poolie: generate the query locally, then get a losa to run it on prod [06:19] nigelb: If you're writing new classes make sure they end up somewhere logical in the style sheet (i.e. not the end of the file) [06:19] poolie: for magic ocnstants, use a few separate queries to establish them (e.g. the person most affected, the distro id etc) [06:20] ok [06:20] i can get at the staging db [06:20] so that should give me the right constants? but not necessarily representative timing? [06:21] nigelb: More specifically that means applying a second class to the table (I'm guessing that's where you're currently applying the styles at the moment). [06:21] nigelb: Have a hunt around in the stylesheet (do you know where that is?) for the .form class and see if anyone has done something similar. [06:24] poolie: right [06:32] huwshimi: cool, I'll look [06:32] (At work right now and away from my hacking laptop) [06:33] nigelb: Great, let me know if you get stuck [06:36] wallyworld_: this may be more of what we discussed before… I filed bug 859369. I've also got a multi-thousand-line lint fixup branch going into testing. [06:36] <_mup_> Bug #859369: Hideous lint in choiceedit.js < https://launchpad.net/bugs/859369 > === jtv1 is now known as jtv [06:37] multi-thousand line. wow. [06:42] could anyone guess why, running tests locally, things that try to upload to the librarian fail with [06:42] UploadFailed: Server said: 500 Internal server error [06:42] lifeless, re bug search timing [06:42] many of those portlets are going to be retrieving similar kinds of data - counts across the bugs in a particular pillar [06:43] i suppose they could possibly be bummed together into a smaller number of queries [06:43] at least for, for example, the aggregation by status or by importance [06:57] poolie: yes, a couple of passes of such grouping have already been done [07:00] * wgrant hates the product/distribution split [07:02] good morning [07:08] o/ adeuring [07:08] hi poolie === jpds_ is now known as jpds [07:49] hi danilos [07:49] could you read https://code.launchpad.net/~mbp/launchpad/855150-no-potemplate-mail/+merge/76878 for me [08:14] can someone confirm that canonical.librarian.tests.test_db_outage.TestLibrarianDBOutage.test_outage was broken on the 19th [08:17] poolie, sure [08:21] poolie, btw, note that you are also fixing bug 353648 with this [08:21] <_mup_> Bug #353648: Template import success notifications shouldn't be sent to package uploaders < https://launchpad.net/bugs/353648 > [08:21] oops :) [08:22] poolie, well, if your bug was only about this (you did raise other problems), I would have marked it as a duplicate of that bug and extended it to not be just about package uploaders but project maintainers as well :) so, you are actually fixing that other bug only :) [08:23] poolie, anyway, looks good, r=me [08:24] poolie, fwiw, there is web UI that shows if the template has been imported already, it might be hard to find though, but that's a different problem [08:27] awesome [08:28] yes, talking it over with robert, we thought that if it's not visible (enough) that can be a separate bug [08:28] so we could probably just do a short blog post saying the mails are going to cease [08:28] i don't think any other prep is needed [08:29] poolie, yeah, agreed, I'd even say that it's most important to let our CHR people know about it if people come asking (i.e. "no success emails anymore, go to your project/+imports page to see the status of the upload") [08:34] stub: hey, still around ? [08:35] lifeless: yer [08:35] would you be up for a brief call - want to touch base on some rt's with you [08:36] lifeless: k [08:36] lifeless: lets see if skype works... [08:38] lifeless: skype didn't survive the upgrade [08:38] stub: There's a multi-arch workaround for that. [08:38] jtv: choiceedit.js came across directly from lazr-js [08:39] https://bugs.launchpad.net/ubuntu/+source/skype/+bug/830440 [08:39] <_mup_> Bug #830440: skype: error while loading shared libraries: libXss.so.1: cannot open shared object file: No such file or directory

< https://launchpad.net/bugs/830440 > [08:39] stub: ^^ [08:40] wallyworld_: aigh. [08:40] ? [08:41] wallyworld_: I saw your name in “bzr annotate” and so hoped you'd be familiar with the code. [08:41] There's some fuss about which of JavaScript's delightful comparison operators to use, and I didn't want to touch that for fear of introducing subtle breakage. [08:42] I think we should probably just ban ==... [08:42] jtv: sadly no. when i brought everything across, i shaved a yak or two to get it all to play nicely, that;s all [08:42] JS type coercion: Australia says no. [08:43] jtv: i've had the same issues. once, i converted == to === and some bugs appeared [08:43] wgrant, stop the floats! [08:43] wallyworld_: well, at least that validates my response. [08:43] so i had to use Y.Lang.isValue() instead [08:43] jamesh: so the exception name qualification bug in python-oops : are you going to patch it, or its just a nice to have? [08:43] wallyworld_: I did that in one place. Very C-ish loop. [08:43] poolie: Heh. [08:43] ah C, the good old days :-) [08:44] lifeless: was planning on putting together a patch. Just haven't done so yet. [08:44] jamesh: no worries [08:44] Any reviewers in the house? https://code.launchpad.net/~jtv/launchpad/post-857155/+merge/76837 [08:44] stub, are you OCR'ing today? [08:44] I pine for the C [08:44] jamesh: let me know if the libraries still need work to play nice [08:44] jtv: No. My brain isn't working today. [08:45] stub: insert coffee. [08:45] jtv: Done all that. I've just got the lurgy that has been doing the rounds. [08:45] And no more Monday reviewer for Europe, I see. Maybe I should change slots to cover that hole. [08:45] stub: feel better. [08:45] lifeless: sure. [08:45] bigjools: C is good, but C++ is better :-) [08:45] Monday is light. [08:46] -lol- http://www.boganipsum.com/ [08:46] wallyworld_: naturally! better than most things [08:46] lifeless: I think that did the rounds while you were away :P [08:46] lifeless: But yes, hilarious. [08:46] wgrant: probably yeah :) [08:46] markov FTW [08:49] stub, lifeless, are the query changes in https://code.launchpad.net/~mbp/launchpad/855150-mail-disabled/+merge/76876 dangerous at all? [08:49] stub: are you installing that workaround, or should I call a landline? [08:49] stub: had you landed the patch that you approved earlier today or should I ask the OCR? [08:49] err, db patch [08:50] lifeless: turtles all the way down. Finished reporting the bug on software centre (well... confirming it affecting me) [08:50] stub: lovely! [08:50] lifeless: It could take a while. Mobile better. [08:50] poolie: 61 + def test_get_recipients_team_with_disabled_account(self): [08:50] 62 + """Mail is not sent to teams containing people with a non-active account. [08:50] 63 + [08:51] poolie: surely 'mail is not sent to people with non active accounts' ? [08:51] poolie: or are you intending to disable all notifications if one person in the team-owning team is inactive? [08:51] (if so, your patch doesn't achieve that ...) [08:53] wgrant: pretty boring run so far… the version we have in Sources.gz is always the oldest version we have in the database. [08:53] stub: kk, will be a minute, nappy change time [08:54] ... and perhaps a hose. [08:54] lifeless, the docstring is wrong, the intention of the test is meant to be 'mail is not sent to team members who have inactive accounts' [08:54] wallyworld_: it's not better than cricket though [08:55] wgrant: then again, the actual domination algorithm hasn't changed, so arguably we don't need to continue the current run. We could just alter Sources.gz for a version we've already dominated, and re-run. [08:55] poolie: so the change is fine I think [08:55] poolie: but it's incomplete [08:55] poolie: this covers the case of 'when contacting a team who do we wamil' [08:56] poolie: emm, I have a memory of there being another case for members of (rather than walking the owner tree) [08:56] poolie: you might like a separate test where the disabled person is a member of the team rather than the owner [08:57] bigjools: not much is better than cricket :-D [08:57] poolie: (oh, and to change that test to have the owner not be in the team [the default is the owner is a member]) [08:57] poolie: HTH [08:59] so you're saying, two tests: [08:59] 1- disabled person is a member (not the owner) [08:59] 2- disabled person is the owner (not a member) [09:00] arguably there should be one for indirect membership [09:00] for curiousity was doing this by tweaking the storm query a clean way? [09:04] wgrant: stats snapshot shows gina making pretty good progress actually… https://pastebin.canonical.com/53350/ [09:10] poolie: I think so yes [09:11] poolie: I may be wrong / misremembering [09:12] lifeless: have working skype [09:12] oh, good timing [09:13] stub: have you signed in? [09:14] hmm, so how do I sensibly check if a branch is private from within the Launchpad code ? accessing .private raises Unauthorized [09:21] Anyone willing to review? https://code.launchpad.net/~jtv/launchpad/post-857155/+merge/76837 === mpt_ is now known as mpt [09:39] lifeless: would you be up for a review? [09:40] jtv: not just now sorry; otp, then nabbing tom, then familty [09:40] OK [09:40] Anyone else? [09:40] Reviewer needed! [09:41] Reviewer needed, since Friday, for critical bug. No volunteers? It doesn't have to be the same people every time! https://code.launchpad.net/~jtv/launchpad/post-857155/+merge/76837 [09:41] poolie, you're a reviewer, right? [09:42] allenap, maybe you can review it for me? Q/A is already well underway, so no very obvious huge blunders. [09:43] jtv with trainer wheels [09:43] Ah [09:43] jtv: Okay. [09:43] thanks! [09:43] wgrant: I think gina's about done with sid. [09:43] Yup. Stats time. [09:44] jtv: And it's only running over sid, so that's good. [09:44] Yes. [09:44] I didn't see any point in running it on the other series. [09:45] Neither. [09:45] And we don't have a source archive for them. [09:45] Stats before & after: [09:45] https://pastebin.canonical.com/53347/ [09:45] https://pastebin.canonical.com/53353/ [09:45] jtv it's not obviously wrong but i really have very little understanding what this is doing [09:46] poolie: Gavin's having a closer look, thanks. [09:50] wgrant: stats look fairly sane to me, prima facie. Only sid seems affected, about the same numbers of packages got deleted and stayed published as before, and we get about 1 DSDJ per deleted package. [09:50] jtv: It doesn't seem to have done anything obviously entirely insane, which is about as much as I expected to glean from this test run. [09:51] jtv: Approved, with a caveat about testing. [09:51] Quite. Then again, the algorithm's substantially the same so it's important to know we didn't skip large numbers of dominations. [09:51] thanks allenap! [09:53] jtv, could you look at https://code.launchpad.net/~mbp/launchpad/858618-affecting-me/+merge/76880 for me? [09:53] OK [10:08] lifeless, tests added; they pass [10:09] cool [10:09] anything else? don't want to rush but would like to kick it off before i sign off [10:11] seems fine to me :) [10:18] poolie: done with your MP. Hope it answers your question; do let me know if it doesn't. [10:20] lifeless: testing your oops_twisted [10:20] it's blowing up [10:20] http://pastebin.ubuntu.com/697174/ [10:25] bigjools: no publisher configured [10:25] bigjools: file a bug though, with no publisher it shouldn't blow up [10:25] lifeless: should be [10:26] lifeless: here's the branch change: http://pastebin.ubuntu.com/697177/ [10:26] so the failure is in fallback_report [10:26] which takes the ids, the report and the event, and is called after the publishing deferred fires [10:27] lifeless: it's adding a FileLogObserver [10:27] bigjools: [10:27] + if options["oops-dir"]: [10:27] + repo = DateDirRepo(options["oops-dir"], options["oops-prefix"]) [10:27] + config.publishers.append(defer_publisher(repo)) [10:28] bigjools: so 2 things; one - there is a bug (please file it on python-oops-twisted, will fix for you tomorrow) [10:28] ok [10:28] and two, you need this line: [10:28] config.publishers.append(defer_publisher(repo.publish)) [10:29] the bug is that if there is no publisher, the fallback code crashes, rather than e.g. forwarding the original event [10:29] lifeless: what is the bug (for the title) ? [10:29] sorry, should have read first [10:29] 'if there is no publisher, the fallback code crashes' [10:29] :P [10:30] :) [10:30] was too busy typing to read you :) [10:31] no worries [10:31] lifeless: that change didn't help [10:31] A tachyon walks into a bar. The bartender says "We don't serve your kind here." The tachyon replies "You did tomorrow". HAHAHAHAHA [10:32] * bigjools sees tumbleweed [10:32] * bigjools hears a dog barking in the distance [10:32] bigjools: this is what I think will fix it (I'll do the test in the morning) [10:32] you lot have no sense of humour [10:32] http://pastebin.com/6qWMhzZn [10:33] bigjools: However, the root issue is that its not seeing the id in the report [10:33] wallyworld_: that's irony :) [10:33] :-) [10:34] bigjools: how are you testing this? [10:34] "Your paste has triggered our automatic SPAM detection filter" [10:34] nice [10:34] wallyworld_: we have a sense of humour, but we keep up with the news. Had the joke included a speeding neutrino… [10:35] * wallyworld_ thought they were the same thing [10:35] doh [10:35] i fail [10:35] lifeless: I am not starting the rabbit server before txlongpoll - I can't think of a decent unit test yet [10:35] bigjools: ok, and how are you starting txlongpoll ? [10:35] bin/twistd -n amqp-longpoll -l mylog -u guest -a guest -f 9999 [10:35] bigjools: I believe options['date-dir'] is not set. [10:35] wallyworld_: they may be the same thing for all I know, but I can't imagine anyone being surprised if a tachyon went faster than the speed of light. [10:35] AH [10:35] fuck [10:36] bigjools: which will cause you to have no publisher. [10:36] indeed [10:36] \o/ [10:36] * jtv vows, if ever invited, to do his round of the Top Gear test track in reverse in order to get the lowest-ever time. [10:36] http://pastebin.ubuntu.com/697182/ [10:37] bigjools: just like a bought one [10:37] shame about the poor backtrace [10:37] but thats out of our immediate control [10:37] lifeless: yeah, and the fact it writes them every second [10:37] (which is the reconnection interval) [10:37] but hey [10:37] bigjools: thats something we can throttle in oops-tools / oops-amqp [10:38] that'd be a nice enhancement [10:38] (e.g. throttle on the server side, not client - or we can do client as well, in future, but I'm more interested in server side smarts) [10:38] easier to deploy [10:39] lifeless: heh, the fallback output is a little whack [10:39] oh ? [10:39] http://pastebin.ubuntu.com/697183/ [10:39] wtf [10:39] bigjools: our big test run on dogfood is done; mind if I run some smaller ones? [10:39] jtv: sure [10:39] bigjools: bug for that please. [10:39] thx [10:39] heh [10:39] also, wtf. wtf. wtf. [10:41] lifeless: so I should be able to have a fallback without the publisher set, right? [10:42] that's just the first bug [10:42] yep [10:42] ok, that saves me some code :) [10:42] you can drop the patch I pastebinned into your local egg for now [10:42] right, ta [10:43] night night [10:43]