[12:01] You didn't mark it resolved. I'll mark it resolved now. [12:01] BradB: didn'[t know it existed. [12:01] :) [12:01] heh [12:01] Malone bug #82 fixed for product The Launchpad: Possible initZopeless transaction issues [12:01] https://dogfood.ubuntu.com/malone/bugs/82 [12:06] lifeless: btw, when do we get Zope 3 upgraded? [12:10] BradB: dunno. I'm waiting on the distro team right now to get svn support back. [12:12] ok [12:24] kiko: after i add a bug, do i want to be returned to the bug listing, or do i want to land on the page to edit the bug i've just added? i'm not sure if the common case is to report a bug and add a bunch of extra info about it, or rather to report a bug and get on with your day while other people (e.g. maintainers) figure out if there's a CVE ref#, ext url's, etc. to add [12:24] there are both use cases [12:25] for me, i think it's been that normally i just report a bug and get on with my day, rather than start adding a bunch of other crap right away [12:25] and, in fact, i've done the occassional rapid-fire bug reporting [12:25] it often happens both ways [12:25] sometimes I'll forget [12:25] and then add a comment [12:27] sometimes I don't want to add an initial comment that is too long [12:27] sometimes I just want to file 10 friggin bugs and go watch a movie [12:28] you need to cater for both use cases (and others, of course) [12:29] I think I know what should be done: either the user should be taken to the last page of bug results after they report a bug (so they can see the one they just added and click and edit it more) or the bug listing sorting should be sane and go newest to oldest, which would achieve the same goal but in a different way. [12:30] I'm betting that the common case is /not/ to do a bunch of editing on a bug after it's been added, but I guess time will tell. [12:44] you can cater both if you present the buglist with a short message box at the top with a link to the bug [12:44] [ Bug _XXXX_ filed ] [12:44] this would also fix one of the 9x bugs I reported [12:45] but if the bug listing is shown newest to oldest, they already have a link to their bug right at the top of the listing. is it worth creating another link above that? [12:48] I followed up last week to that bug you reported, btw. [12:48] The user already does get an email [12:53] BradB, dude, the bug # filed needs to be *prominent* on the page [12:53] not in the list [12:53] posting a bug number and pasting it somewhere is *the* use case to cater for :) [12:54] kiko: Where does Bugzilla put it? I don't recall offhand. [12:54] Malone puts it in the subject line of the email the user receives. [12:54] a bit fat box saying Bug XXX filed [12:54] email is nice [12:54] but you want it in the web UI, definitely [12:55] ok [12:57] BradB, I'll give you a good usability test of malone in barcelona, you'll owe me ice cream later. [12:59] heh [01:00] i'm a make-this-usable-for-human-beings nazi as it is, but more feedback from others is always good, particularly bugzilla maintainers :) [01:04] I'm a human being too [01:04] perhaps the most human of all bugzilla maintainers :) [01:06] hah === stub [~stub@dsl-246.248.240.220.dsl.comindico.com.au] has joined #launchpad === Kinnison turns into a pumpkin. G'night [01:16] pompkin [03:49] Merge to rocketfuel@canonical.com/launchpad--devel--0: Full text indexing, database schema patches (patch-909) [03:54] hur fucking ray [03:55] Now to work out wtf the pagetest I had to disable works fine locally but not on chinstrap :-( [04:27] New Malone bug #106: "Disabled Malone pagetest", submitted by Stuart Bishop [04:27] https://dogfood.ubuntu.com/malone/bugs/106 [04:45] Merge to rocketfuel@canonical.com/launchpad--devel--0: Turn on Celso's doctest patch for dogfooding (patch-910) === stub [~stub@dsl-246.248.240.220.dsl.comindico.com.au] has joined #launchpad === stub [~stub@dsl-246.248.240.220.dsl.comindico.com.au] has joined #launchpad === stub [~stub@dsl-246.248.240.220.dsl.comindico.com.au] has joined #launchpad === lifeless_ [~robertc@dsl-253.4.240.220.rns01-kent-syd.dsl.comindico.com.au] has joined #launchpad === lifeless [~robertc@dsl-253.4.240.220.rns01-kent-syd.dsl.comindico.com.au] has joined #launchpad === stub [~stub@dsl-246.248.240.220.dsl.comindico.com.au] has joined #launchpad [07:57] lifeless: Sender not authorised to commit to package rocketfuel@canonical.com/sqlobject--test--0.6 [07:57] lifeless: (when I try to get pqm to merge changes from me) [09:22] New Malone bug #107: "Unable to view bug 70", submitted by Andrew Bennetts [09:22] https://dogfood.ubuntu.com/malone/bugs/107 [09:33] spiv: fixing now [09:34] spiv: done [09:41] Merge to rocketfuel@canonical.com/sqlobject--test--0.6: Add a limited SELECT DISTINCT to SQLObject (patch-5) [09:44] lifeless: Ta [09:46] Ooh... might fix that __del__ crap that is bugging us. Tomorrow ;) [09:51] yay [09:55] New Malone bug #108: "Duplicate bugs in bug listing", submitted by Andrew Bennetts [09:55] https://dogfood.ubuntu.com/malone/bugs/108 [10:12] hmm, what's the easiest way to display a date in a nicely formatted way in a page template? [10:12] the default stringification seems to be along the lines of "2004-11-15 16:04:49.157656" [10:14] Steve made a helper and announced it on launchpad@ just recently - tal:mydate/fmt:i_forget [10:15] Where i_forget is defined in the FormatAPI class in tales.py [10:15] oh [10:16] that's just what I want [10:16] well, I think minute-resolution would be marginally preferable to second-resolution, but that's minor [10:18] I think it is still open for discussion [10:19] well, I think the API is probably stable [10:19] the exact formats it uses can probably changed in the future === silbs [~sbsm0084@host217-37-231-28.in-addr.btopenworld.com] has joined #launchpad [10:24] hi Jane! === lulu [~lu@host217-37-231-28.in-addr.btopenworld.com] has joined #launchpad [10:45] Morning [10:48] stub: ping? [10:49] Kinnison: pong [10:49] daf: You're up unreasonably early. What's up? [10:49] stub: When's the next dogfood update? [10:49] Tomorrow [10:49] Kinnison: s/early/late/ [10:49] About 16 hours time [10:50] I will do more frequent updates for beer though [10:50] stub: I see === SteveA [~steve@adsl-213-190-44-43.takas.lt] has joined #launchpad [10:56] hi spiv [10:57] hi daf [10:57] mornin' steve [10:57] (hi everyone else! but, I particularly want to talk with spiv and daf) [10:57] hi Steve [10:57] Hi Steve :) [10:58] spiv: how's the work on shipit going? [10:58] daf: are you on a morning or an evening at the moment? [10:58] daf, stub, i don't think stevea's date formatter work is ready for prime time, it didn't work for me on datetimecol's [10:58] sabdfl: hmm, it worked beautifully for me [10:59] SteveA: on an evening [10:59] SteveA: did you commit a fix or two to that? [10:59] sabdfl: datetimecols return mxDateTime objects, which would not be hard to cope with [10:59] I haven't committed a fix to make it handle "None"/NULL yet [10:59] ah, right, that was the issue [10:59] ah, ok, I don't think I have a code path that tickles that [11:00] I need to do a bit of bazzing, then I'll commit that fix [11:00] SteveA: we missed you yesterday [11:00] daf: traveling back from england. took a bit longer than I expected. [11:01] SteveA: oh, problems with flights? [11:03] more with trains. [11:03] can we talk about rosetta a bit? [11:04] certainly [11:05] Merge to rocketfuel@canonical.com/launchpad--devel--0: fix PO file autovification, 'Last Change' feature (patch-911) [11:05] SteveA: btw, I fixed that initzopeless thing [11:05] can I upload a pot file to a temporary place on the system, for it to be imported a bit later? [11:05] lifeless: what did you do? [11:05] make it "reentrant" ? [11:05] no, joined the new transaction to the connection data adapter [11:06] ok [11:06] thanks [11:06] SteveA: I need to workrave -- can we talk in ten minutes or so? [11:06] yes [11:06] spiv: still around? === lulu [~lu@host217-37-231-28.in-addr.btopenworld.com] has joined #launchpad [11:22] SteveA: it's nearly working -- we stumbled on some some unicode-related problems yesterday [11:23] what are the problems? [11:23] and, have you worked out how to make a test-case for them? [11:24] the PO parser was written to only parse string objects [11:24] and it broke when it was passed unicode objects [11:24] we have identified the causes of the problems [11:24] I have asked Carlos to write a test case [11:24] so... it should be passed a string object [11:25] it shouldn't be passed a unicode object, because that doesn't make a whole lot of sense [11:26] hmm, I was wondering about that [11:26] let me try to explain [11:27] we have two kinds of data: string data and bytes data [11:27] when string data comes in, we always convert it into unicode objects [11:27] when bytes data comes in, we always leave it as bytes [11:28] ok so far? [11:28] a jpeg is bytes [11:28] yes [11:28] text entered into a form is "text" so, strings to become unicode objects [11:28] you'd never try to represent a jpeg as a unicode object [11:29] in python, this gets rather confusing. this is a "wart" in python [11:29] why is it a wart? [11:29] because the "str" type is commonly used for both "encoded string" data and for "bytes data" [11:30] but, when it is used for "encoded string" data, it does not know what encoding was used [11:30] it is lossy [11:30] so, if you have a str object, is it string or is it bytes ? [11:30] text or data? [11:30] you can't know for certain [11:30] The same wart exists in Lua [11:30] you can't know. so you need to make sure your program treats "str as encoded string" specially [11:31] and always passes around the encoding with it [11:31] Kinnison: that's interesting. Are they going to fix it? [11:31] so, back to po files [11:31] SteveA: Given there's currently no encoding support in native Lua I doubt it [11:31] your po file is data [11:32] the parser wants data [11:32] and it gives you back, um... I can't remember actually. maybe you can tell me? [11:32] it gives you message objects [11:32] ok, so they include unicode strings [11:32] yes [11:32] okay [11:32] so, data in, unicode out [11:33] and, the po file tells you what encoding it is in [11:33] yes [11:33] so, that's easy [11:33] which presents an interesting challenge [11:33] with all the above in mind, what is the unicode problem? [11:33] since you have to start parsing the PO file before you know what encoding it's in [11:34] but, that's all the po parser's job [11:34] yes [11:34] and that's done [11:34] yes [11:34] so, tell me about the current unicode problem, bearing the above in mind [11:35] bearing the about in mind, I think the problem is that the database is returning Unicode objects for the raw PO file [11:35] s/about/above/ [11:36] ok, that sounds about right [11:36] what type is the column? [11:36] Merge to rocketfuel@canonical.com/launchpad--devel--0: Make fmt:date handle None properly. (patch-912) === daf checks [11:36] thanks dilys [11:36] text [11:36] which is wrong [11:36] okay [11:37] what should it be? BINARY i suppose [11:37] or bytes [11:37] I can't remember the name [11:37] so, you know what to do now. [11:39] spiv spiv spiv [11:39] calling spiv [11:39] come in, number spiv [11:41] daf: I guess it goes: 1. produce database change on your machine. 2. check it works. 3. update code (if needed) 4. ask for DBA patch on mailing list. [11:41] daf: did I miss anything? [11:43] SteveA: the fact that I'm thinking of going to bed? :) [11:44] before you go, please write a note to the launchpad mailing list about all this. that gives carlos and stub a head-start when they turn up today. [11:44] ah, ok [11:44] thanks [11:44] I was already composing an email to Carlos about it [11:44] should I send it to the mailing list instead? [11:46] well, sent to Carlos, CCing you [11:46] did you get the email I sent on Sunday night? [11:47] not yet. I still need to catch up on some mail. [11:47] the list is good, because [11:47] 1. the issue is one that may affect others working on these web apps [11:47] 2. the DBAs need to know about it [11:48] points taken === cprov [~cprov@200.158.100.251] has joined #launchpad [11:54] daf: can we have a meeting about rosetta either later today or tomorrow? [11:54] Morning cprov [11:54] Kinnison: morning [11:56] SteveA: yes [11:57] SteveA: I think later today it still possible [11:57] SteveA: but if it doesn't happen, let's plan for earlyish tomorrow [11:57] SteveA: launchpad meeting tomorrow yes? [11:58] daf: I want us to set a time, check with carlos, and then we all know what we're doing. [11:59] Kinnison: yep. I'll send a mail in a bit. === carlos [~carlos@69.Red-80-33-181.pooles.rima-tde.net] has joined #launchpad [12:00] hi carlos. daf and I are just talking about when the three of us can meet up to talk about rosetta. [12:00] hi [12:01] just tell me when it's ok for you [12:01] SteveA: any chance we can start a bit earlier than 12.30 (say 12.15) because I may have to leave at 13.15 to collect my partner (they moved his days in ely to tue/wed from mon/tue) [12:01] daf: around? [12:01] SteveA: would 2000 UTC tonight be too late for you? [12:02] carlos: yes [12:02] Kinnison: fine with me, and I'm sure fine with the .au folks. [12:02] BradB / daf: can we have a launchpad meeting tomorrow at 12:00 or 12:15 UTC? [12:02] daf: which problem are you talking about exactly? [12:02] daf: the one related to adding unicode support to the parser? [12:02] daf: that would be fine with me. We must be on time, though. [12:03] daf: it's ok for me also [12:03] carlos: you should read the chat daf and I had about this. [12:03] ok === carlos opens the irclogs [12:04] SteveA: no objections [12:04] it was just 30 mins or so ago [12:04] carlos: I'm happy to try and explain further if the situation isn't clear [12:05] SteveA: (in regards to tomorrow's meeting) [12:05] SteveA: and yes, I will set an alarm for tonight's [12:05] daf: it's not yet in the log, will wait [12:06] hmm [12:06] let's let daf get some sleep. carlos and I can chat about it here. [12:06] perhaps the log bot isn't here [12:06] it is important that both of you understand the issue. [12:06] daf: it has a delay [12:06] oh [12:06] carlos: I can mail you the logs if needed. [12:06] don't worry [12:06] daf: great, thanks. [12:06] I can wait [12:06] daf: go to sleep! [12:06] :-P [12:07] carlos: well, I'd like you to get up to speed on this quickly, so we can fix the problem quickly [12:07] so we have a meeting tonight at 20:00UTC and tomorrow at usual time? [12:07] Kinnison: let's say the meeting will be at 12:00 UTC. If brad needs to be a bit late, then that's okay. [12:07] carlos: a bit earlier tomorrow [12:07] SteveA: okay; cool. Thanks dude. [12:07] daf: good night! [12:08] SteveA: it can be later if I'm able to be first to report and we don't get bogged down in TAL or whatever it was last week which confused the fook out of me [12:08] carlos: yes [12:08] daf: don't worry, Steve will help me with it. Makes sense what you said in your mail [12:08] carlos: do you have jabber running? [12:08] ok, buenas noches [12:08] SteveA: yes [12:08] daf: buenas noches [12:08] what's the id ? [12:08] carlos@pemas.net === kiko [~kiko@200-206-134-238.async.com.br] has joined #launchpad [12:09] hmm, anything went wrong, I don't see you after accepting the request :-? [12:10] the request is one-way. but, you can make the same request in the opposite direction. [12:10] sometimes the apps get confused. maybe you need to restart your jabber client? [12:14] SteveA: usualy, the first request adds also the contact into my list [12:15] SteveA, daf: Will look into it today, I agree with the proposed change [12:16] ok, great [12:16] ping me if you have any doubts [12:16] or any questions ;-) [12:17] :-) [12:17] the only question is about the sqlobject to use [12:17] if it's a binary field, could I use a StringCol? [12:18] (the sqlobject _type_ to use) [12:18] I don't know what sqlobject offers for bytes data [12:18] maybe BradB or spiv knows? [12:18] ok [12:18] or we can look it up [12:19] will search the documentation [12:20] http://sqlobject.org/docs/FAQ.html#binary-values [12:20] it suggests you store it base64 encoded [12:21] which would work okay, and avoid changing the database again [12:21] ok === Kinnison grrs as he realises 'make check' isn't working [12:21] it is sat idle [12:22] :-( [12:23] that's happened to me before. but, I changed the check_on_merge script to use threaded IO, which fixed it [12:25] Hmm, has this been undone? [12:26] nope [12:26] still there [12:26] can you see what is blocking? [12:26] ctrl+c and stare at the TB? [12:26] 843 ? S 0:00 postgres: dsilvers launchpad_ftest [local] idle in transaction [12:26] 867 ? S 0:00 postgres: dsilvers launchpad_ftest [local] idle [12:26] those are the two postgres processes I can see [12:26] I just did ^C and it hasn't stopped yet === Kinnison straces it and ponders [12:27] dsilvers@petitemort:~/dev-canonical/launchpad/launchpad/database/schema$ strace -p 32751 [12:27] Process 32751 attached - interrupt to quit [12:27] futex(0x8b0bd28, FUTEX_WAIT, 0, NULL [12:27] this smells like a futex issue [12:28] I shall dist-upgrade and try again [12:28] right [12:28] you on hoary or warty or something else? [12:28] warty [12:30] nothing's upgraded for me for ages on warty. [12:30] I wonder if my sources.list is correct [12:30] Do you not have security in your list? [12:30] SteveA: you need the security and updates line [12:31] deb http://archive.ubuntu.com/ubuntu warty main restricted universe multiverse [12:31] deb-src http://archive.ubuntu.com/ubuntu warty main restricted universe multiverse [12:31] deb http://archive.ubuntu.com/ubuntu warty-security main restricted [12:31] deb-src http://archive.ubuntu.com/ubuntu warty-security main restricted [12:32] hmm, and I don't have the updates line.. (for things like the calendar :-P) [12:33] good morning [12:33] how is everybody? [12:33] what's "restricted" ? [12:33] hi kiko [12:33] restricted is a bit like "non-free" I believe [12:33] SteveA: binary drivers [12:33] oh, okay [12:34] kiko: I think that's multiverse [12:34] it's non-free software that's critical for machine support, i.e. gfx, wifi etc. drivers [12:34] elmo, can you do something for me ? [12:34] well, binary drivers are non-free as well :) [12:34] not sure I need any of that [12:34] kiko: good point [12:34] :-D [12:34] much easier to do this interactively ... [12:34] kiko: yes, but he's still right, multiverse is our equivalent of non-free [12:35] lifeless? [12:35] on auckland, in the arch archives, delete all the category directories for all the imports. [12:35] carlos is always right, no news there [12:35] i.e. === carlos leaves for an hour or so [12:36] in xorg@arch.ubuntu.com there is a top level dir of [12:36] \=meta-indo [12:36] bah [12:36] I hate public administration [12:36] mmmm meta-indo [12:36] \=meta-info [12:36] and a .archive-version file [12:36] and N other dirs, one for each pjrect in the archive. [12:36] I need to nuke those other dirs [12:37] and I need you to echo "Bazaar archive format 1 0" to all the .archive-version files.... but that step will happen later today, after we release. [12:37] stevea: I think you will also want deb http://archive.ubuntu.com/ubuntu warty-updates main restricted universe multiverse [12:38] lifeless: you mean the .listing file? [12:39] elmo: .listing is irrelevant [12:39] look at xfs@arch.ubuntu.com instead, will make more sense === Kinnison workraves [12:40] acl attr xfsdump xfsprogs should go byby. [12:40] oh, ok [12:40] =meta-info should stay [12:40] lifeless: umm, you sure about this? 'cos if I do it, I'm going to rm them... [12:40] yeah, I have copies. [12:41] auckland [12:41] is a public copy only. [12:41] just do -not- do it on galapagos. [12:42] find www/* -maxdepth 1 ! -name \=meta-info -type d -mindepth 1 | xargs rm -fr [12:42] ^-- running [12:43] yup, cool [12:44] spiv: do we have distinct support yet? === Kinnison [~dsilvers@haddenham.pepperfish.net] has joined #launchpad === Kinnison retries the make check [01:07] Okay, it failed :-( [01:07] ProgrammingError: ERROR: database "launchpad_ftest" is being accessed by other users [01:07] Considering this is a totally fresh boot, I find that hard to believe [01:08] Or rather, it has to be a bug in the test suite [01:08] Kinnison: are you using an SMP kernel? [01:08] No [01:09] oh, well, we had this problem on mawson, for a while, fwiw. it kind of went away on it's own [01:09] Feh :-( [01:09] thanks dude [01:09] I think at the time, I downgraded to a UP kernel, but we appear to be back on SMP now [01:09] There also appears to be proper failures [01:09] you might try running postgres without NPTL [01:09] There appear to be real ftest failures in my scrollbuffer [01:09] lifeless: done [01:10] lifeless: I assume I should do what it says to the stuff in 'deleteme' ? [01:13] yes [01:13] thanks [01:17] well, a re-run gives the same issues :-( [01:17] that has put a dampner on my ability to run make-check before pqm merging :-( [01:18] try recompiling everything? === Kinnison tries [01:19] Kinnison: the same futex issues? if so, try without NPTL, you should be able to just LD_ASSUME_KERNEL the postgres server [01:20] elmo: not the futex issue. no [01:20] elmo: the futex issue went away with the upgrade [01:21] oh, ok [01:21] oh? thats interesting. [01:21] we've had terrible futex issues with buildbot/importd [01:24] SteveA: a make clean && make check still fails [01:24] how odd [01:25] is it pagetests that fail? [01:25] yeah [01:25] it appears in part to be complaining about the full text search stuff [01:26] but I followed stub's instructions from his mail [01:46] For someone who knows more about these things than me: https://chinstrap.warthogs.hbd.com/~dsilvers/clean-and-check.log [01:47] ProgrammingError: ERROR: could not find tsearch config by locale [01:47] that one appears to be the most relevant [01:55] what do you have for this: SELECT * from pg_ts_cfg; [01:57] launchpad_dev=# SELECT * from pg_ts_cfg; [01:57] ts_name | prs_name | locale [01:57] -----------------+----------+-------------- [01:57] default | default | C [01:57] default_russian | default | ru_RU.KOI8-R [01:57] simple | default | [01:57] (3 rows) [01:57] and, what is your machine's locale? [01:57] locale -a perhaps [01:58] dsilvers@petitemort:~$ locale -a [01:58] C [01:58] POSIX [01:58]