[12:01] Thanks [12:01] or mawson [12:03] it's amazing how far having tabs along the top of the page and portlets can make it look like you rewrote plone === ddaa gets the page [12:05] ;) [12:05] BradB: do you use vim on OS X? [12:05] Where's the "Plone Powered" link? ;-P [12:05] limi: I can, but I'm mainly an emacs guy. [12:06] ddaa: heh heh [12:06] https://launchpad.ubuntu.com/doap/projects/canonical/malone [12:06] two tabs [12:06] selected [12:06] I guess because both "projects" and "malone" are inthe url === SteveA --> bed [12:09] SteveA: weird. i wondered what the logic behind that was. [12:24] spiv_: woo, transactional zopeless! === BradB is now known as BradB|away [12:32] courtesy of taxi.py [12:32] Now... if only I could get this import to go that far before hanging... [12:46] ddaa: happy taxi's ? [12:46] Not there yet :-( [12:46] I ran an import not long ago and it just hung... [12:46] And I have been carried away trying to debug the hang. [12:46] strace -p pid - bet you a beer its waiting on a futex [12:46] That's vague... [12:46] if its waiting on a futex, its the Queue [12:46] I'm tracing the damn thing with twisted spewage. [12:46] night! [12:52] lifeless: regarding the hangs, I often have them after "launched local CVS server" [12:52] Just not when I want them to happen of course. [12:53] Is there any pyarch around there? [12:53] after the local CVS server launches, there is an rlog call, which can take non-trivial time. [12:54] there isn't any pyarch at that point. [12:55] That's hang situation. Not impatience. No network traffic, no cpu usage, no i/o frenzy. [12:55] ok, thats a concern then. [12:55] what does strace show you ? [12:55] Not stracing... [12:55] (when I see that, its often very low cpu, but present) [12:55] spewing, but it's working okay atm. [12:56] ok, next time it happens, attach a strace to the slave before you kill it [12:56] Well... I do have some background activity... so maybe I'm missing it... but that would suprise me very very much. [12:56] Okay. [12:57] If it's hung, I s'pose strace will show nothing, right? [01:00] strace will show the call that hung [01:00] Mh. Okay. [01:01] if you see multiple processes as the slave (old style threads), attach to them all, one will be the slave event loop, and not helpful. [01:01] Merge to rocketfuel@canonical.com/launchpad--devel--0: translationsPortlet implemented and inserted into SourcePackage page. Closes bug #2139 (patch-702) [01:01] lifeless: yeah... I have many such threads most of the time... [01:02] Bug 2139 resolved: Rosetta link (portlet?) for Soyuz Source Package index [01:02] https://bugzilla.warthogs.hbd.com/bugzilla/show_bug.cgi?id=2139 === ddaa is irresistibly draws to a ridiculously expensive and overpowered thinkpad t42 [01:17] buy a hoverbook [01:17] goes well with the massage-keyboard [01:17] maybe tla will finally not be dog-slow if he gets a Powerbook ;) [01:17] he has a powerbook at the moment [01:18] a iBook actually... and so does bob2 [01:18] ...running Linux ;) [01:18] we just happen to be using a real operating system :-P [01:18] limi: of course. [01:18] a real operating system that doesn't support the wifi card in the Powerbook :P [01:18] btw, you guys cannot recognize great technology. Touch keyboards are the future. [01:19] Well... maybe... they are maybe a bit awkyard with legacy layouts... [01:22] limi: my ibook does not have wifi. That's the main problem. [01:22] I was cheap when i bought it. [01:22] And get your fruit company to be a bit more tight assed when it comes to hardware specs. [01:23] *a bit less [01:55] heh [01:55] yeah [02:54] kiko: you're a vim user, right? [02:55] limi, indeed I am. [02:55] trying to make the switch ;) [02:55] is there an autocomplete HTML mode/plugin available? [02:55] found this: http://vim.sourceforge.net/scripts/script.php?script_id=301 [02:56] and put it in my .vim/ftplugins [02:56] but nothing happens :] [02:56] <-super-newbie [02:57] kiko: and you are the kind of person that requires nick highlight too, right? ;) [02:57] yeah, because 1000 people want to talk to me all the time in 50 differnet channels :) [02:58] hehe [02:58] I can sympathize ;) [02:58] kiko: anyway, what do you use for HTML editing in vim? [02:58] hmmm [02:59] nothing much, to be honest. [02:59] kiko: autocomplete and tag matching is what I need - if it supports folding, that's a bonus :) [02:59] ok [02:59] that script looks pretty cool [02:59] kiko: but is there a special magic to activate an ftplugin? [03:00] placing it in your /ftplugin dir should work, I suspect [03:01] kiko: it's in ~/.vim/ftplugin/ now, but I see no love - I suspect you need to put it into XML mode or something? [03:01] hmmm [03:01] there seems to be something in html.vim there [03:02] I can take a look a bit later and test it myself, will let you know of the results [03:03] hm, seems to be a file extension thing [03:03] works if I have an .html file [03:03] thanks :) [03:05] yes [03:05] you can :set filetype=html if you want to force that [03:05] and you can set hooks in your .vimrc that do that for you [03:06] ok [03:07] kiko: and the extension -> filetype mapping is done where? need to define .pt as XML :) [03:07] you can just set it manually in your .vimrc [03:07] au! BufRead,BufNewFile *.ptl set filetype=python [03:07] au! BufRead,BufNewFile *.pt set filetype=html [03:07] aha [03:07] thanks :) [03:09] there we go :) [03:09] cool === limi feels like a vim pro already ;) [03:20] Merge to thelove@canonical.com/dists--bazaar--1.3: now we have the name, propogate it throughout (patch-1) [03:26] Merge to thelove@canonical.com/dists--bazaar--1.3: s/baz/bazaar in file names (patch-2) [03:38] Merge to thelove@canonical.com/dists--bazaar--0: propogate name change (patch-1) [03:49] thelove? [03:57] thelove beby [03:57] thelove [03:57] https://www.warthogs.hbd.com/Bazaar === kiko raises eyebrows [03:58] nice! [04:01] lifeless, I can't say how much I think you rock for starting this. [04:01] thank you. Mark gave me the chance and the prod. [04:02] can we get an inventory of the patches applied to it so far? [04:02] there aren't any yet., [04:02] I mean, a list with a summary of descriptions [04:03] still bootstrapping the name change. [04:09] I see. I imagine we'll keep one, or would we be using tla inventories to browse them? [04:10] I'm not sure what you are looking for specifically. [04:10] if you mean a changelog - what features are added, the changelog for it should be sufficient. [04:11] we will be maintaining a separate list of UI forward-compatible-breaking changes. [04:11] that's that I'm thinking, yes. [04:12] have a look at the draft web page, its linked from the bazaar page. [04:12] I saw it, yeah. [04:12] theres a section on it just for that. === kiko is now known as kiko-afk === BradB|away is now known as BradB === BradB is now known as BradB|zzz === carlos [~carlos@69.Red-80-33-181.pooles.rima-tde.net] has joined #launchpad [11:06] morning [11:17] Hi carlos [11:20] Right; which database should we be using on mawson? [11:36] elmo: ping? [11:37] What's a futex and why would gzip be waiting on one? [11:38] a future mutex? [11:39] fast mutex, that's a linux kernel thing. [11:39] "fut alors" [11:39] Right; 'cos apt-ftparchive gets stuck [11:40] and this is irritating because it's why my test-run didn't finish [11:41] http://ds9a.nl/futex-manpages/futex2.html [11:42] The sys_futex system call provides a method for a program to wait for a value at a given address to change, and a method to wake up anyone waiting on a particular address. While the addresses for the same memory in separate processes may not be identical, the kernel maps them interally so the same memory mapped in different locations will correspond for sys_futex calls. Futexes are typically used to implement the contended case of a lock in shared [11:49] So can I stop it using them since it's clearly broken? [11:51] I'd rather suspect a userspace bug. [11:52] LD_ASSUME_KERNEL=2.4 ..... [11:52] seems to let it get on with things and run [11:53] evil... but good to know you can do that... [11:53] aye === Kinnison will have to try and debug whatever is causing gzip to hang [11:53] but for now at least; my test run might complete [11:56] apt-ftparchive is doing its job :-) === lulu [~lu@host217-37-231-28.in-addr.btopenworld.com] has joined #launchpad [12:08] Kinnison: upgrade your gzip [12:08] it's #1854 [12:08] elmo: zhongshan [12:09] meh [12:09] :-) [12:09] Done. 8000MB in 13533 archives. Took 15m48s [12:09] w00 w00 w00 [12:09] fixed [12:09] spankyou === Kinnison blows the caches away and tries again [12:10] elmo: zhongshan still has that odd "apt-extracttemplates /var/cache/apt/archives/libc6-dev_2.3.2.ds1-13ubuntu2_i386.deb" running [12:11] well; s/running/in the process table/ [12:11] yeah, it doesn't matter [12:12] do you know if rsync --link-dest also detects renames? === Kinnison takes this workrave break and goes to get dressed [12:24] elmo: Could you look at zhongshan:~dsilvers/apt.conf and tell me if it looks sane to you? It built zhongshan:~dsilvers/dists/... :-) [12:58] Kinnison: launchpad_dogfood [12:58] stub: gotcha [12:59] elmo: did you lock the wiki yet? [01:00] yes, just managed to. moin was being.. uncooperative [01:02] ok. I've just been chatting to enrico [01:02] he's going to announce a plan to the docs mailing list [01:04] Merge to rocketfuel@canonical.com/launchpad--devel--0: File list publishing for lucille (patch-703) [01:29] elmo: I need to get a little food. After that, I'm going to set up cacheing in plone, following vika's instructions. [01:29] elmo: There are some changes need to the apache config. I'll mail those to you in a bit. [01:30] ok, cool [01:32] but, I suggest not to do them until I've done the plone set-up [01:32] the cool thing is, I don't think I even have to restart plone === lifeless [~robertc@dsl-78.1.240.220.rns01-kent-syd.dsl.comindico.com.au] has joined #launchpad [01:49] here [01:49] I got the result of the taxi test. [01:49] good/bad/indifferent ? [01:49] psycopg.OperationalError, FATAL: database "launchpad_ftest" does not exist [01:49] spiv_: around? [01:50] Yeah. [01:50] ^^^^^^^6 [01:50] It's using the default value from canonical.lp.dbname. [01:50] I need to doublecheck some stuff, but at first I find that suprising since job loading is done from the database. [01:50] ddaa: taxi runs in the other process [01:50] ddaa: Ah... but job loading happens in the master. [01:50] completely decoupled [01:50] Including decoupled configuration of things like databases :) [01:50] spiv_: other threads do not use the same value of canonical.lp.dbname? [01:51] ddaa: threads have nothing to do with this ? [01:51] ddaa: What else accesses the db from the slave? I thought the slave got its directions from the master? [01:52] lifeless: I do not know what has to do with what, but if the only db name used is read from the canonical.lp module, and that module is loaded only once, I cannot see how that can happen. [01:52] spiv_: nothing else in the slave talks to the db [01:53] ddaa: is the slave running from the same checkout as the master ? [01:54] lifeless: I do not see how it could be differently, and anyway I do not see how that would be relevant. The database name is set in master.cfg immediately after first importing launchpad.db. [01:54] ddaa: *thats in the master* [01:54] Ha... [01:54] *taxi is in the slave* [01:54] Okay. [01:54] get much sleep last night ? :) [01:54] Then how should I configure that in the slave? [01:55] Not much... I stayed up waaay to late looking for a french place to buy the specific lappy I'd like. [01:55] I'm not sure I'd get the full benefit of the garantee if I buy it from some other country... [01:56] Soooooo... spiv_? What's the place to set the dbname for the slave? [01:57] ddaa: canonical.lp.dname ;) [01:57] I'd very much prefer something that does not involve hardcoding settings... [01:58] maybe that data could be passed down the wire... e.g. in the Job? [01:58] Yeah, it probably could be. [01:59] So, if I just put an additional attribute in the job loaded from the master, it's going to be accessible by the slave? The job is pickled or something then passed on the wire? [02:00] Or make a sitecustomize.py that overrides it ;) [02:00] ddaa: no, don't pass it to the Job [02:01] in production that would be a PITA. [02:01] a sitecustomize.py is more appropriate. [02:01] spiv_: Please explain. I want all configuration to stay in the botmaster directory. You suggest putting an optional module there and have buildbot try to load it? [02:02] Mh... actually, it should be in the slave's directory... [02:02] echo "import canonical.lp; canonical.lp.dbname='blah'" > sitecustomize.py; PYTHONPATH=. twistd etcetc... [02:03] I see. I'd load that information from Job.py? [02:03] In runJob, maybe? [02:03] (When starting up, i.e. before running a program, python will try to import sitecustomize, so if you have it somewhere on your PYTHONPATH...) [02:03] Ha. Thanks, I was not aware of that. [02:04] It's a bit of hack to do it that way. [02:04] I think perhaps mixing configuration and code like we currently are with canonical.lp.dbname is not ideal. [02:04] It will be time to change it later if there is a need for other sitecustomize... [02:05] A spiv_: sure, canonical.lp could be modified to try to load e.g. lp_config and get hostname and dbname from there if it's available in the pythonpath. [02:06] Right. [02:06] spiv_: should not be very difficult to do... [02:06] But the details should be debated on the launchpad list, I think. [02:06] I suspect SteveA and others will have useful insights here. [02:07] Okay. I'll write a message to launchpad@ right now. [02:07] Another option would be to read them from environment variables. === carlos [~carlos@69.Red-80-33-181.pooles.rima-tde.net] has joined #launchpad [02:08] And I'll do the sitecustomize.py thing in the meantime. [02:08] [env] yeah, I'll suggest that too. [02:08] But I think loading a file is simpler. [02:08] file please [02:09] Harder to get wrong. === spiv_ nods [02:13] can we use ZConfigure for everything, and have one standard launchpad.conf file ? [02:13] and not lots of files for different purposes === ddaa fears the Z ;-) [02:15] SteveA: That'd be great if every upstream package we used (e.g. Twisted, Buildbot) already used ZConfigure :) [02:15] But unfortunately that's not the case. Buildbot's config stuff is a bit of a mess, too. [02:16] why can't *we* use ZConfigure? [02:16] we're not using any of these tools in the raw. Can't we write our own script to start buildbot using our own config? [02:17] Buildbot's a nasty case, config-wise. It really is a bit of mess that I wouldn't want to touch without a couple of days to properly fix it. [02:18] (Maybe it's been fixed in upstream SVN, though..?) [02:18] Are there any good docs on ZConfigure, btw? [02:18] yes [02:18] wonderful docs [02:18] I'm perfectly happy with customised lp.__init__ === SteveA -> food [02:19] and if we spend more than 10 minutes fixing it, we've already blown the cost of customising two installs 5 times over [02:19] lifeless: Which is my point stated much more directly :) === lifeless looks at the scrollback. [02:20] 2149-2220 [02:20] thats 30 minutes === lifeless declares this discussion over. don't fix it. [02:20] :) [02:21] SteveA: Sorry, I was fishing for a hint as to their location, rather than their existence :) [02:28] SteveA: are we having a meeting; or can I go to the post office; bank; etc. ? === debonzi [~debonzi@200.158.100.251] has joined #launchpad [02:50] ddaa: The launchpad_ftest database won't exist atm. unless you use a test harness that sets it up and tears it down, such as canonical.launchpad.ftests.harness [02:51] stub: you are referring to my post on launchpad@? [02:51] I'm working on a branch making this saner atm === stub is referring to irc a few minutes ago [02:51] On mawson are we changing lib/canonical/lp/__init__.py to say launchpad_dogfood instead? [02:52] I do not want to use launchpad_ftest [02:52] I want to use a persistent database for development purpose. aka launchpad_dev. [02:52] oh... sorry. I have test suites on my mind :-) [02:53] Yes - canonical.lp.__init__.py should not have launchpad_ftests in there. [02:53] It's a sane default to have. [02:53] esp. if the db does not exist! === Kinnison updates that file then [02:54] Hmm... maybe. The branch I'm working on currently sets it to whatever is specified in the LP_DBNAME environment variable, or launchpad_dev if it isn't set. I'm currently bogged down in repairing unit tests though === BradB|zzz is now known as BradB [02:54] Actually, a sane default would probably be something that cause a failure if not overriden... but that's still another issue. [02:54] I would have changed that if it were referencing launchpad_dev, like the Makefile and ZCML were. :) === ddaa tries hard to keep only one copy of every tree to minimize maintenance. [02:55] BradB: if we're making $anotherdomain.$tld, can we use one of the existing plones, or do we need a third instance? [02:55] I love "aba change-version", it's really a basic building block for my methods. [02:56] ddaa: raw switch [02:56] Will switch ;-) as soon as I have cleared up the "asap" stuff. [02:56] ddaa: I think I'll try that (dbname=None unless overridden in lp/__init__.py) and see what happens [02:57] stub: => launchpad@ === SteveA returns [02:57] Kinnison: we'll be having a meeting. If you want to go to the I guess you can report first [02:58] SteveA: Okay; cool === Kinnison is on the phone to the insurance broker right now; so I'll say the moment I can type lots [02:58] elmo: I'd make another instance. [02:58] ddaa: Hehe... you chose exactly the same env variable I did (or did that patch get to rocketfuel?) :-) [02:58] stub: maybe that's just because it's the obvious name? [02:58] BradB: meh. how would I make another instance? I guess I can just copy most of it, but what about Data.fs ? [02:59] bin/mkzopeinstance.py [02:59] spiv_: is normal that the default database for zopeless is dbname = "launchpad_ftest" ? [03:00] elmo: Is this intended to be the same site just with a different name? [03:01] BradB: no, different site [03:01] hello. welcome to this week's launchpad meeting. [03:01] Yeah, so bin/mkzopeinstance.py [03:01] all present, please say "aye!" [03:01] aye! [03:01] aye [03:01] eye [03:02] eieieio [03:02] (old macdonald had a sheep, e i e i o) ? [03:02] and on that sheep there was a farm, eieio [03:02] carlos: ? [03:02] daf sends apologies [03:03] yodeleitooooo [03:03] spiv_: ? [03:03] aye [03:03] to start with, a couple of announcements. [03:04] I'm off on vacation from tomorrow, back on 8th [03:04] Mark is away at a conference in Germany, and then in South Africa, back on 8th [03:04] #1 & #2 gone... [03:04] Stuart has agreed to be the head launchpad dude until then. [03:05] Let's agree on the time and date of next week's launchpad meeting now, [03:05] SteveA: yes [03:05] I'm here [03:05] hi carlos [03:05] How about Wednesday, 1230 UTC? [03:06] sounds good [03:06] sure [03:06] ok [03:06] launchpad-brazil ? [03:06] no problem for me [03:06] stub: ? [03:07] lifeless: infoImporter and infoUpdater changes sent to pqm. [03:07] I can do that, although earlier is good if it doesn't put brazil out. [03:07] cprov is comming in some minutes but I think is not a problem to him to [03:07] ddaa: great [03:07] debonzi: 1200 ? [03:08] would 1200 be okay instead of 1230? [03:08] SteveA, ok too.. [03:08] lifeless: the infoUpdater worksforme, but I have not actually tested that it did not change jobs that are not in the "unassigned" product. But that's what the code should do. [03:08] ok, let's so 1200 [03:08] ok, warning heard [03:08] lets not confuse the lunchbox guys by interrupting too much. [03:08] tell me if you want additional "interactive python" support. [03:09] next, summary of the last meeting [03:09] SteveA: perhaps it's too early for daf [03:09] it's on my todo list -- and I still haven't to-done :-/ [03:09] What's the right place to talk then? [03:09] carlos: daf isn't here. so, he doesn't get a say :-p [03:09] ok ;-) === kiko [~kiko@200-206-134-238.async.com.br] has joined #launchpad [03:09] hi Ki [03:09] hi kiko [03:09] hello SteveA [03:10] ddaa: private chat to me I guess [03:10] SteveA: 1200 is okay by me [03:10] Kinnison: are you okay to talk yet? [03:10] yep [03:10] ok. before we start, the theme of this meeting is "dogfood" [03:11] Right [03:11] mmmm [03:11] To that end; my primary concern is getting the librarian going on mawson [03:11] can it also be "soyuz is protected in an SEP field"? :) [03:11] I'm about 20 minutes off completing that :-) [03:11] go for it, Kinnison [03:12] Okay; so things I know about for dogfooding on mawson... [03:12] librarian is needed -- I'm working on that now [03:12] gina -- I've got a patch floating for the build stuff; as soon as I'm told she can import against the current schema I'll polish it and commit it [03:12] Getting lucille to publish the contents of the dogfood db will take a little more work [03:13] I now have a completely published archive successfully on zhongshan [03:13] so I need to script the publishing process and then I'll be in a position to start the publishing on mawson [03:13] But that kinda wanders back into my expected lucille plan really [03:13] If anyone wants me to do anything else for mawson in the short-term; can you say *NOW* please? [03:14] Merge to rocketfuel@canonical.com/launchpad--devel--0: transactional infoImporter, tweak infoUpdater for production use (patch-704) [03:14] dilys: thanks babe. [03:15] SteveA: right; that's about it wrt. dogfood then [03:15] SteveA: in more general terms; my plans for lucille for the coming week are... [03:15] 1. Get the lucilleconfig columns populated and get the canonical.lucille.Publisher class generating apt.conf files for apt-ftparchive [03:16] 2. Design the version domination algorithm (mostly stealing from katie if elmo will let me) [03:16] Kinnison: stub probably wants to automate the code upgrade/db upgrade/data migration stuff on mawson. [03:16] 3. Get domination stuff. [03:16] BradB: for the most part; that shouldn't affect me much [03:17] oh and somewhere in there I need to split launchpad/{database,interfaces}/publishing.py into more manageable chunks and get my pending patch cleaned up to be integrated [03:17] But that shoudl be it. [03:17] phew. [03:17] splitting up publishing would be appreciated, cool [03:18] domination is what was called overrides? [03:18] SteveA: No === Kinnison will add 'domination' to the glossary [03:18] Domination is the process by which the new version of a package ousts the older version from the archive [03:18] This is related to it having been built cleanly on all architectures etc. === Kinnison needs to chat with elmo about this [03:19] really? [03:19] I never heard that term. and what do you mean by ousts? becomes the new "current"? [03:20] kiko: Yes; that's what I mean [03:20] I see. [03:20] kiko: I haven't used the term before; so you won't have heard it yet :-) [03:20] well, one thing that does happen is different versions of current for different architectures. does that fit in with your reasoning? [03:21] Yes [03:21] It's all to do with the Pending,Published,Superceded,PendingRemoval state machine in the publishing tables [03:21] I was just lead off by your phrase "on all architectures". [03:21] yeah. [03:22] But I fear we're getting off-topic here :-) [03:22] BradB: you've been forging ahead with the set-up on mawson. Can you talk a bit about that, what people need to do, and what is still to be done? [03:22] hm [03:22] I put a new code drop on mawson late yesterday [03:22] elmo needed to do something things for some people, and I think that's mostly (if not all) done. === cprov [~cprov@200.158.100.251] has joined #launchpad [03:23] stub (and whoever else) need to come up with the policy for how we're going to update the code and update the db schema and do data migration [03:23] some discussion was started on lp@ about this, but last i checked, no firm decision reached [03:23] sorry, I'm late [03:23] not sure if mark wants a hand in that [03:23] hi cprov [03:24] so, here's the problem: [03:24] data migration and db schema upgrades should be no problem - we went through the issues at oxford. I just need to write some fairly minor code. [03:24] we need to decide and get stuff working before mark gets back [03:24] oh, ok [03:24] indeed [03:24] so yeah, for us, the data on the dogfood server is "production" data (...goes to get link...) [03:24] hi, steve... did I miss something important ? (debonzi is doing a report ...) [03:25] https://launchpad.ubuntu.com/doap/projects/canonical [03:25] cprov: not really. main thing is "dogfood", and making sure we can get there soon [03:25] If you get a 403, you need to install the client cert. [03:25] Where do we get the cert from? [03:25] It was mailed to lp@ [03:25] from elmo, with love [03:25] SteveA: ok === Kinnison looks for it [03:26]