[00:12] <wgrant> wallyworld_: My branch APA branch has landed.
[00:17] <james_w> lifeless, get well soon
[00:24] <wallyworld_> wgrant:  thanks, will take a look
[00:50] <wallyworld_> wgrant: i just pointed chromium at https://launchpad.net/launchpad/+filebug and it works fine
[00:51] <wallyworld_> 18.0.1025.168
[00:54] <wgrant> wallyworld_: Ahh, indeed.
[00:54] <wgrant> the problem in fact affects both.
[00:55] <wgrant> wallyworld_: You have to get to the second step using non-AJAX
[00:55] <wallyworld_> wgrant: what project were you filing a bug against?
[00:55]  * wallyworld_ tries that
[00:55] <wgrant> That is, enter a summary and click Continue on the first step before the JS changes it to an AJAX Next.
[00:55] <wgrant> You have to be very quick
[00:55] <wallyworld_> hmmmm
[00:56] <wgrant> That then POSTs to +filebug to return the second step without AJAX
[00:56] <wgrant> Which presumably lacks some of the JS or something.
[00:56] <wgrant> "choices is undefined"
[00:56] <wallyworld_> corner case :-)
[00:57] <wgrant> Yeah.
[00:57] <wallyworld_> so i guess i should try disabling the button untill all ajax is loaded or something
[00:57] <wgrant> No.
[00:57] <wgrant> The JS on the second stage should just not crash :)
[00:57] <wgrant> Although it would be nice to solve the slowness on the first stage too.
[00:58] <wallyworld_> i'll see what can be done
[00:58] <wallyworld_> the trouble is doing the post before the ajax is loaded sort of violates the expected behaviour
[00:59] <wallyworld_> ie it was never designed to handle that use case
[01:00] <StevenK> Bleh, why does branch deletion always toss me to the "Please try again" page
[01:00] <wallyworld_> timeout?
[01:00] <wgrant> StevenK: The timeout is too low to delete all 80k branchrevisions
[01:01] <StevenK> DROP TABLE branchrevision;
[01:01] <wgrant> wallyworld_: The rest of the JS on the second step works fine...
[01:02] <wallyworld_> i need to look into it - the whole bug form setup is "complicated"
[01:02] <rick_h_> wallyworld_: wgrant which bug is this? (curious)
[01:03] <wallyworld_> bug 1008543
[01:03] <_mup_> Bug #1008543: ChoiceSources on +filebug are empty in Chromium <disclosure> <javascript> <regression> <ui> <Launchpad itself:In Progress by wallyworld> < https://launchpad.net/bugs/1008543 >
[01:03] <rick_h_> oh sucky, hadn't seen that one
[01:04] <wgrant> It's actually nothing to do with Chromium. It just happened that I tested it slightly differently in Chromium and Firefox.
[01:04] <wgrant> Was apparently slightly slower in Firefox.
[01:04] <rick_h_> gotcha, well that's a good thing at least, consistant
[01:04] <wallyworld_> yep
[01:05] <rick_h_> is this combo loader related? did it work without it?
[01:05] <wgrant> Unrelated
[01:05] <wgrant> This is code that was new last week
[01:05] <rick_h_> k, cool
[01:05] <wallyworld_> rick_h_: we need to support devs running combo loader with un-minified js ie add filter: 'raw', to the config just for devs
[01:06] <rick_h_> wallyworld_: yea, agree. I've been doing it manually, but would be nice to have it in config
[01:06] <rick_h_> no real reason not to run ith with always raw for devs
[01:06] <wallyworld_> rick_h_: we do need devs to run minified if required
[01:06] <wallyworld_> for testing etc
[01:06] <rick_h_> yea, but really QA would blow up if minifying messed anything up
[01:06] <wallyworld_> but we can add a property to the view which used a feature flag
[01:07] <wallyworld_> liek we do for yui version
[01:09] <StevenK> I think we should just set raw for devmode
[01:10] <wgrant> devmode is evil
[01:10] <wgrant> use feature flag
[01:13] <lifeless> james_w: thanks
[04:10] <StevenK> Hmmm. Can't reproduce bug 919252 on dev :-(
[04:10] <_mup_> Bug #919252: blueprints: can't reset Assignee etc to 'None' via popup <disclosure> <person-picker> <regression> <specifications> <Launchpad itself:Triaged> < https://launchpad.net/bugs/919252 >
[04:21] <wgrant> StevenK: Privs?
[04:22] <StevenK> wgrant: I'm doing as owner, but I have no idea of the Linaro set up.
[04:29] <StevenK> wgrant: And he says he can do it via +people
[04:30] <wgrant> StevenK: It's possible that it's been accidentally fixed, I supposed. Have you tried both Firefox and Chromium?
[04:30] <StevenK> I don't have Chromium to try
[04:30] <StevenK> It works in Firefox at least
[04:36] <wgrant> sudo apt-get install chromium-browser
[04:36] <StevenK> wgrant: *sadface*
[04:36] <wgrant> You can't expect to work on a webapp without it.
[04:38] <StevenK> wgrant: What about IE, Opera and Konquerer? :-)
[04:39] <wgrant> Covering both WebKit and Gecko successfully makes it a lot more likely that Trident, Presto and KHTML will work.
[04:39] <wgrant> It's still nice to test in the other three
[04:39] <wgrant> But testing in two is far far better than testing in one.
[04:39] <wgrant> Particularly since they're the dominant browsers
[04:40] <StevenK> wgrant: I thought Trident was only 9? Wasn't the engine in 8 something different?
[04:41] <wgrant> StevenK: Nope, since IE4.
[04:41] <wgrant> Except for Mac OS versions
[04:48] <StevenK> Chromium has no Help->About
[04:49] <StevenK> I guess Google considers indentification too hard
[04:49] <StevenK> wgrant: Works for me on dev with Chromium too
[04:50] <wgrant> StevenK: Huh, it doesn't have a classical menubar at all.
[04:50] <wgrant> They must have added it for globalmenus.
[04:50] <wgrant> Click the in-window menu icon, then About Chromium.
[04:55] <StevenK> wgrant: I'm suspecting it's already fixed, but I don't have acess as a driver to stuff on prod
[04:56] <StevenK> wallyworld_: How is information_type ChoiceSource going?
[05:38] <wallyworld_> StevenK: it's done, i have to finished the tests. but the tests are breaking due to animation issues so i need to sort that out
[05:40] <wallyworld_> StevenK: the code (minus tests) if you are interested, https://code.launchpad.net/~wallyworld/launchpad/infotype-widget-1007984/+merge/108665
[07:49] <adeuring> good morning
[08:15] <wallyworld_> StevenK: i've finished updating the tests
[08:15] <wallyworld_> https://code.launchpad.net/~wallyworld/launchpad/infotype-widget-1007984/+merge/108665
[08:19] <StevenK> wallyworld_: Looks good.
[08:19] <wallyworld_> coolio. seems to work nicely
[08:22] <StevenK> wallyworld_: Approved.
[08:23] <StevenK> stub: https://code.launchpad.net/~stevenk/launchpad/spph-packageupload-index/+merge/108515 would love a review.
[08:25] <stub> StevenK: Your partial index WHERE clause is backwards
[08:25] <StevenK> Oh, IS NOT NULL
[08:25] <StevenK> Bleh
[09:25] <lifeless> stub: I'm not here today, sick - I may be a few days getting over this
[09:26] <lifeless> stub: the storm and sockets thing; I think my main concern is group abnormal stuff into magical storm handling codepaths
[09:26] <stub> lifeless: I think I have food poisoning, or some sort of stomach bug.
[09:26] <lifeless> stub: like IntegrityError we know the db is live; SocketError - thats kindof a WTF moment.
[09:26] <lifeless> stub: ugh.
[09:27] <lifeless> stub: my sympathies; I have horrid stomach cramps at the moment - side effect of the virus
[09:27] <lifeless> stub: anyhoo, just wanted to put that out there - I'm not totally hostile to storm capturing the error, if those risks are assessed.
[09:27] <lifeless> stub: probably we need two distinct bugs though:
[09:27] <lifeless>  - one for handling the error
[09:27] <lifeless>  - one for -why- is it occuring
[09:28] <stub> lifeless: The goal of the Storm code is to reconnect when the db dies or is uncontactable or whatever. By design, this means it will hide failures if we neglect to log them.
[09:28] <lifeless> [and the folk that are stressing, (OEM), will only care about the second one]
[09:28] <stub> I think we log them though.
[09:28] <stub> lifeless: yup
[09:29] <lifeless> stub: so, if you need sick leave, JFDI, drop uhm elliot francis and I a mail (elliot cause I'm sick ATM, francis cause of this librarian issue so he knows its on hold while you recover)
[09:30] <lifeless> I'm going to crawl back off into my little hole and think healing thoughts.
[09:30] <StevenK> lifeless: Or stomach amuptation?
[09:30] <stub> lifeless: IIRC what happens is that Storm catches all the possible crazy i-can't-talk-to-the-database exceptions and reraises them as DatabaseDisconectionErrors, flagging the connection for reconnection. We log the DatabaseDisconnectionErrors. Trick is to make sure we are logging what we should be logging. I think this is handled by storm manipulating the psycopg2 base classes so the disconnection error we catch is actually a psycopg2 exception.
[09:30] <lifeless> StevenK: did I mention the razor blade throat with laryngitis ?
[09:31] <StevenK> lifeless: :-(
[09:31] <stub> lifeless: you sound worse than me. I'll probably take sick leave once I've investigated this landscape replication issue.
[09:31] <lifeless> stub: I leave it in your capable hands; my main concern is about the 2-bug separation, not one bug w/2 tasks
[09:31] <stub> yup
[09:31] <lifeless> stub: thanks!
[09:31] <lifeless> Lynne and Cynthia had this, seems to be a 5-day thing
[09:31] <stub> erm... u1 replication issue
[09:32] <lifeless> day 2 of it for me (yes, sick on a pub holiday... win)
[09:33] <StevenK> stub: That index has landed as r15360 on devel.
[10:44] <wgrant> StevenK: That index looks backwards
[10:44] <wgrant> It's packageupload IS NOT NULL, when we want to search for ones where it IS NULL.
[11:02] <stub> did he go an land the backwards version?
[11:43] <StevenK> wgrant: stub said it was backwards :-(
[11:49] <wgrant> StevenK: stub probably mistook it for the PackageUpload.changesfile != None
[11:55] <StevenK> And ENOSTUB
[11:55] <StevenK> Let me revert
[11:59] <wgrant> StevenK: Can't revert
[11:59] <wgrant> StevenK: Wasn't it applied live?
[12:00] <wgrant> (I don't know if it was, but it probably was)
[12:00] <StevenK> And we have no UK webops to check
[12:00] <wgrant> jjo is around now
[13:09] <deryck> Morning, everyone.
[13:44] <wgrant> rick_h_: Thanks.
[13:45] <wgrant> rick_h_: A scan of 22000ish rows is much quicker than a scan of 22000ish rows then a join through 5 multi-million row tables :)
[13:46] <rick_h_> wgrant: yea, just got beat into my head that LIKE was evil too much over the years
[13:47] <sinzui> Sweet. I just fixed a post-3-ui-cleanup bug. It only took 2.5 years
[13:47] <rick_h_> sad that regex was slow, but oh well
[13:47] <rick_h_> sinzui: ummm woooo!
[13:47] <wgrant> rick_h_: The regex is still pretty fast, but 3x slower so meh.
[13:48] <wgrant> rick_h_: (do note though that postgres 9.1 can use trigram indices to satisfy LIKE '%foo%' efficiently)
[13:48] <wgrant> It's just not worth it here.
[13:48] <rick_h_> yea, something tells me dropping an 8 table join is probably a good plan regardless of the new plan lol
[15:25] <jcsackett> sinzui: i'm looking at bug 1007208. do we want to change the character limits for all pickers, or just the person picker?
[15:25] <_mup_> Bug #1007208: Searching for 'ev' in the 'add a member' picker fails <disclosure> <person-picker> <Launchpad itself:Triaged> < https://launchpad.net/bugs/1007208 >
[15:27] <sinzui> jcsackett, excellent question
[15:28] <sinzui> jcsackett, The crux of the issue is that Lp Ids can be 3 characters. Lp has a nasty habit of creating Lp-ids from user email addresses, which are often 2-characters.
[15:29] <sinzui> jcsackett, Thus person pickers/searches must support 2-characters.
[15:30] <jcsackett> sinzui: right, i totally agree with the ">1" constraint for person pickers. i assume the bug picker doesn't do any checking regarding length, b/c that would be silly. that leaves...branch pickers?
[15:30]  * jcsackett slaps head.
[15:30] <sinzui> jcsackett, I think we can say projects and distros are less of a problem because users name them,
[15:30] <jcsackett> sinzui: yeah. and as i just realized, branch pickers aren't an issue b/c of branch naming patterns.
[15:30] <jcsackett> so this probably is just a person picker issue.
[15:31] <sinzui> jcsackett, branch pickers are buggered by design. We might make them workable in a month when we optimise the slow queries to not blindly search all Lp
[15:31]  * sinzui is looking at projects
[15:33] <sinzui> jcsackett, we 204 2-character project names, and no complaints. Lets no touch projects
[15:34] <jcsackett> sinzui: sounds good to me. i'll leave them alone.
[15:39] <cjwatson> sinzui: of course the case at hand was one where the user deliberately changed his username to a two-character one, rather than it being automatically created from an e-mail address :-)
[15:42] <sinzui> cjwatson, I think that is fine. Email addresses and IRC nicks can be two characters, and Lp allows them. It was ridiculous to require users to enter 3 characters to search.
[17:01] <bdmurray> I've noticed that the list of subscribers is in reverse alphabetical order in the subscribers portlet and I've a fix for this
[17:02] <bdmurray> however it seems like this might have been intentional
[17:02] <bdmurray>              // With multiple subscribers being added to the same section,
[17:02] <bdmurray>             // the last one is listed first.
[17:03] <bdmurray> so before carrying on I wanted to see if it is intentional and if so find out why
[17:05] <bdmurray> oh well there is a bug about it - 912137 so it must be wrong
[18:36] <timrc> When I review commits in Launchpad I sometimes see 'Timothy R. Chavez <timothy.chavez@canonical.com?' and so no hyperlink is produced for my name... is there something I'm doing wrong?
[18:36] <beuno> timrc,
[18:36] <beuno> <timothy.chavez@canonical.com?
[18:36] <beuno> wrong closing
[18:37] <beuno> >
[18:37] <beuno> need to fix bzr whoami
[18:37] <timrc> right, but where do I specify that... I'm using bzr launchpad-login timrchavez
[18:37] <timrc> hm ok
[18:38] <beuno> timrc, can't change it retroactively, btw
[18:38] <timrc> yeah, bummer
[18:38] <maxb> timrc: 'bzr whoami'
[18:38] <timrc> you sort of can with the last commit by uncommitting though
[18:38] <beuno> yeah
[18:38] <timrc> thanks for helping me with a bzr prob in the #launchpad-dev channel
[18:39] <beuno> it's 30 cents extra in here!
[18:44] <timrc> ;)
[22:04] <sinzui> wallyworld_, looks like mumble cannot connect properly
[22:06] <sinzui> wallyworld_, wgrant. I am having networking issues with mumble
[22:06] <wallyworld_> sinzui: should we try skype?
[22:07] <sinzui> I am killing UbuntuOne
[22:09] <sinzui> wallyworld_, your voice is 3 octaves lower. Did you balls drop?
[22:09] <wgrant> sinzui: Sounds like you just need a little bit more upstream bandwidth.
[22:09] <wallyworld_> lol
[22:10] <sinzui> My wife is having trouble too. I think our provider switched us to 1st gear
[22:26] <sinzui> StevenK, I think as a driver you can mess with this https://blueprints.launchpad.net/gdp/+spec/gdplaunchpad
[22:58] <wgrant> Hm
[22:59] <wgrant> Why is icon-sprites.png in the tree now...
[23:00] <StevenK> I just saw that too
[23:00] <wgrant> sinzui changed the filename and ignore rule, and I guess someone blindly added the unknown file.
[23:00]  * wgrant fixes.
[23:06] <wgrant> Except that buildbot-poller apparently hates me.
[23:16] <StevenK> wgrant: Has your fixed landed now?
[23:16] <wgrant> StevenK: No. i suspect that buildbot-poller will remain confused until devel finishes.
[23:17] <StevenK> Bleh
[23:24] <StevenK> wgrant: devel just finished.
[23:36] <StevenK> wgrant: Can haz fix landed?