[12:06] (kiko/#launchpad) sabdfl, ping?
=== BradB [~bradb@modemcable010.194-131-66.mc.videotron.ca]  has joined #launchpad
[12:24] <BradB> ubuntu!
[12:25] (kiko/#launchpad) one for all all for one!
[12:27] <BradB> debian was my first linux (because it was only $35; redhat was like $110). it's good to be home.
=== BradB [~bradb@modemcable010.194-131-66.mc.videotron.ca]  has joined #launchpad
=== BradB is now known as BradB|out
[01:21] (Kinnison/#launchpad) night all
=== stub [~stub@dsl-246.248.240.220.dsl.comindico.com.au]  has joined #launchpad
[02:09] <BradB|out> stub: i followed your instructions verbatim on lp@ for restoring a dogfood db, and it ended up blowing away my hard drive in Panther.
=== BradB|out is now known as BradB
[02:11] <stub> Thats funny - I don't remember putting a mkfs in the instructions anywhere.
[02:11] <stub> You had a good chunk of free disk space?
[02:12] <BradB> the process went from doing weird things like spitting out messages "Invalid character \n" and "Invalid character \N", to running the full test suite more than once, to apparently rm -rf /'ing
[02:14] <BradB> the good news is that now i'm just running Ubuntu, the bad news is that, heh, shit.
[02:17] <BradB> stub: have you tried restoring the 11-18 dump? i'd be curious...
[02:18] <BradB> brb
=== BradB [~bradb@modemcable202.193-131-66.mc.videotron.ca]  has joined #launchpad
[02:36] <stub> That sounds impossible given the instructions just used three comands, none of which have anything to do with launchpad. If the test suite got invoked, you either had a psql in your path that wasn't postgresql or you had some pretty major file system corruption.  Its the sort of thing I can only imagine happening it you had the equivalent of 'ln /usr/bin/bash /usr/bin/psql. I've restored from that particular dump twice now using the 
[02:38] <BradB> it blew me away too. literally.
[02:39] <stub> Mmm... and took any audit trails along with it, so we will never know. 
[02:39] <stub> You have just inspired me to finally go and buy that external harddrive for backups.
[02:40] (daf/#launchpad) I could use one of those
[02:40] <BradB> me too
[02:40] <BradB> i didn't lose anything that can't be recovered, except time...OTOH, i make it back with tla in linux, heh
[02:43] <carlos> nigth
[03:18] (daf/#launchpad) is there a standard way of finding the user's home directory in Python?
[03:20] (kiko/#launchpad) hmm
[03:20] (kiko/#launchpad) there is os.path.expanduser
[03:27] <BradB> eh, why isn't there a keychain package?
[03:31] (daf/#launchpad) ?
[03:31] (daf/#launchpad) kiko: great, thanks
=== BradB [~bradb@modemcable202.193-131-66.mc.videotron.ca]  has joined #launchpad
[03:37] (daf/#launchpad) BradB: keychain package?
[03:38] <BradB> in ubuntu
[03:38] (daf/#launchpad) like the debian-keyring package in Debian?
[03:39] <BradB> i dunno, maybe
[03:39] <BradB> i just wanted ubuntu to help me out a bit here, trying to avoid typing my password for pubkey auth'ing
[03:41] (daf/#launchpad) pubkey authing what?
[03:42] <BradB> daf: during ssh auth
[03:43] (daf/#launchpad) isn't that what ssh-agent does?
[03:44] <BradB> keychain is used with, not in place of, ssh-agent
[03:45] (daf/#launchpad) ah, I see
[03:46] (daf/#launchpad) it must surely be in universe
[03:46] <BradB> ahaha, there we go...i'll have to take a look through there
[03:47] <BradB> but, eh, not worth it.
[03:47] <BradB> daf: what are you using instead of keychain?
[03:48] (daf/#launchpad) I just use ssh-agent
[03:48] <BradB> daf: don't you have to type your key's p/w every time you open a shell then?
[03:48] (daf/#launchpad) no
[03:48] <BradB> er, well, every time you login
[03:48] (daf/#launchpad) yes
[03:48] (daf/#launchpad) it's run as part of the X session
[03:48] (daf/#launchpad) when I log in
[03:48] <BradB> i was hoping to avoid that part
[03:49] (daf/#launchpad) which is maybe once a week
[03:49] (daf/#launchpad) on average
[03:49] (daf/#launchpad) I guess having to run ssh-add when I run is marginally annoying
=== BradB [~bradb@modemcable202.193-131-66.mc.videotron.ca]  has joined #launchpad
=== dilys [daf@muse.19inch.net]  has joined #launchpad
[07:07] (daf/#launchpad) dilys: welcome back
[09:43] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: many fixes for the PO import script (patch-822)
[09:54] (sabdfl/#launchpad) mornign all
=== Kinnison [~dsilvers@haddenham.pepperfish.net]  has joined #launchpad
[10:23] <Kinnison> Morning
[10:23] [nickolas(~nickolas@163puntacana92.codetel.net.do)]  hi! are u busy?
=== Kinnison stares irritatedly at piles of SQL
=== debonzi [~debonzi@200.158.100.251]  has joined #launchpad
[11:39] <Kinnison> hi debonzi 
[11:43] <Kinnison> So does 'make check' run the ftests? And if so; how come it doesn't tell me how many it ran or how well they went?
[11:50] <debonzi> Kinnison, 
=== lifeless [~robertc@dsl-151.5.240.220.rns01-kent-syd.dsl.comindico.com.au]  has joined #launchpad
[12:02] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: Setting NDIFF_REPORT as Default PageTest Failure Report Format (patch-823)
=== cprov [~cprov@200.158.100.251]  has joined #launchpad
=== debonzi [~debonzi@200.158.100.251]  has joined #launchpad
[12:41] (sabdfl/#launchpad) what's with the clausetables?
[12:41] (sabdfl/#launchpad) do they improve performance?
[12:56] (sabdfl/#launchpad) cprov: i see you've added a lot of clauseTables, do they improve performance?
[12:56] <cprov> sabdfl: no AFAIK but postgres avoids "add_missed_From_tables" by default
[12:56] <cprov> sabdfl: sorry, I was unclear, I just added them because postgres doesn't add them by default 
[12:57] (sabdfl/#launchpad) how does the add_missed_from_tables thing work?
=== carlos [~carlos@69.Red-80-33-181.pooles.rima-tde.net]  has joined #launchpad
[01:00] <cprov> sabdfl: postgres.conf , if true postgres will add all missed from tables automatically in every query, default is false
[01:00] (sabdfl/#launchpad) what happens if it doesnt have the missed tables?
[01:01] (sabdfl/#launchpad) and where is that useful dogfood dump backup being stashed now on mawson?
[01:02] <carlos> hi
[01:03] <cprov> sabdfl: the query will fail if you forget some clause table, just it  
[01:03] (sabdfl/#launchpad) ok.... haven't seen that yet
[01:04] (sabdfl/#launchpad) hiya carlos
[01:04] (sabdfl/#launchpad) buenos dias?
[01:04] (sabdfl/#launchpad) have to get into the spirit for es-conf
[01:05] <carlos> sabdfl: yep
[01:05] <carlos> :-)
[01:06] <carlos> sabdfl: you should ask daf his spanish book :-P
=== lifeless_ [~robertc@dsl-151.5.240.220.rns01-kent-syd.dsl.comindico.com.au]  has joined #launchpad
=== kiko [~kiko@200-206-134-238.async.com.br]  has joined #launchpad
[02:33] (elmo/#launchpad) SteveA: ?
[02:49] (SteveA/#launchpad) hi elmo
[02:52] (elmo/#launchpad) SteveA: I can't get the uploader to run in an upate-to-date copy of launchpad; I sent you a traceback - any chance you could have a look at it? 
[02:53] (SteveA/#launchpad) looks like a failure to run "make" in the launchpad/sourcecode/zope directory
[02:54] (elmo/#launchpad) blink
[02:54] (SteveA/#launchpad) although, to be fair, it is that dependency that I've agreed with the other zope3 maintainers shouldn't be in the ftp server
[02:55] (elmo/#launchpad) I thought I'd done that in this tree - I remember it running make check for eons in london.  but apparently not.  thanks, I'll try that
[02:55] (SteveA/#launchpad) k
[02:55] (SteveA/#launchpad) one change I need to make to the ftp code is to isolate the callbacks with a try:except:
[02:56] (SteveA/#launchpad) I'd suggest leaving the filesystem.py and server.py modules intact except for bugfixes, and treat the main() of server as running a test server
[02:56] (SteveA/#launchpad) then write an uploader.py to actually do what you need it to do
[02:57] (elmo/#launchpad) yeah - I'm getting permission denied on uploads, so I'll start there ;)
[02:58] (SteveA/#launchpad) how did you invoke it
[02:58] (SteveA/#launchpad) ?
[02:58] (SteveA/#launchpad) a couple of notes
[02:58] (SteveA/#launchpad) you need to create the directory that is the "root" for uploads
[02:58] (SteveA/#launchpad) inside the root will be one directory for each client's main connection
[02:59] (elmo/#launchpad) as in your example and in my mail
[02:59] (SteveA/#launchpad) I'm not sure what the error is if that directory doesn't exist or isn't writeable
[02:59] (elmo/#launchpad) I precreated the dir
[02:59] (elmo/#launchpad) anyway, sorry, I can/will debug it - I just didn't want to do that when zope was involved :)
[03:02] (SteveA/#launchpad) not got jabber running?
[03:15] (elmo/#launchpad) woops, have now
[03:31] (elmo/#launchpad) ah, hmm, I see what you mean about the try:except:'s
[03:38] (elmo/#launchpad) ./lib/ThreadedAsync
[03:38] (elmo/#launchpad) ./sourcecode/zope/src/ThreadedAsync
[03:38] (elmo/#launchpad) #
[03:38] (elmo/#launchpad) what's up with that?  they seem the same?
[03:50] (SteveA/#launchpad) yeah
[03:50] (SteveA/#launchpad) the things in lib are often symlinks to things in sourcecode
[03:55] <carlos> sabdfl: ping
[03:58] <carlos> who is importing the projects/products into launchpad's dogfood database?
[04:01] (elmo/#launchpad) SteveA: a real symlink, or you mean, virtually?
[04:01] (elmo/#launchpad) oh it is a real symlink.  jesus I am so not having a good day
[04:10] (SteveA/#launchpad) elmo: take 10 mins for a nice cup of tea perhaps?
[04:19] (sabdfl/#launchpad) elmo: don't forget to breathe ;-)
[04:30] (SteveA/#launchpad) daf, carlos: nice work on making the po import script clearer
[04:30] <carlos> sabdfl: ping?
[04:30] <carlos> SteveA: it was all done by daf, but thanks
[04:31] <carlos> sabdfl: I don't understand our current set of projects/products in launchpad_dogfood
[04:34] (sabdfl/#launchpad) carlos: i think those mostly came in from a run of nicole
[04:34] <carlos> sabdfl: for instance, we have gedit as product and project
[04:34] (sabdfl/#launchpad) which looks for sourceforge / freshmeat projects with the same name as packages
[04:35] (sabdfl/#launchpad) it's automated, so not very smart
[04:35] <carlos> hmm
[04:35] (sabdfl/#launchpad) it's just a way for us to get a quick start on a lot of data
[04:35] <carlos> then is it ok If create a gnome project and add there gnome-applets and gnome-panel?
[04:35] (sabdfl/#launchpad) we'll start to build the real production project/product db properly in Jan
[04:35] (sabdfl/#launchpad) yes, if they are not already there
[04:35] (sabdfl/#launchpad) please do!
[04:36] <carlos> no, they are not
[04:36] <carlos> ok, thanks
[04:36] (sabdfl/#launchpad) can you be sure to link them to the relevant package names too?
[04:36] <carlos> ok
[04:43] <carlos> hmm
[04:43] <carlos> we have launchpad down at the moment...
[04:44] (SteveA/#launchpad) elmo: I'm about to submit a merge request to pqm to add a try:except: with logging around the callbacks.
[04:45] (elmo/#launchpad) SteveA: sooper, thanks
[05:04] (elmo/#launchpad) this is going to sound silly, but is baz safe to use for launchpad?  I only ask, 'cos IIRC kinnison and/or sabdfl weren't using it on Wednesday
[05:08] (SteveA/#launchpad) elmo: the advice from the arch team has been to use baz
[05:09] <Kinnison> elmo: I'm only not using baz because I'm used to tla now and can't afford the mental-shear to switch
[05:09] <carlos> do we have .deb packages already?
[05:10] (elmo/#launchpad) I think they're linked off bazaar.canonical.com
[05:12] <carlos> ok
[05:12] <carlos> thanks
[05:28] <carlos> SteveA: I'm importing now the .pot/.po files into a test database at mawson and as soon as I see it worked I will do it into the dogfood database
[05:29] <carlos> wow, mawson is really fast (my laptop takes some minutes, mawson took just some seconds...)
[05:30] (sabdfl/#launchpad) elmo: i haven't wanted to try baz till it was at 1.0, but if the arch guru's recommend it i'll give it a shot
[05:33] (sabdfl/#launchpad) hm... jblack not around? ddaa? arch team awol?
[05:33] (sabdfl/#launchpad) elmo: maybe you can help, the deb download page for bazaar is formatted in a way that makes it rather hard to know which link to click on :-)
=== BradB [~bradb@modemcable202.193-131-66.mc.videotron.ca]  has joined #launchpad
[05:41] (elmo/#launchpad) hmm, sucks yeah, I'll fix the apache config
[05:45] (sabdfl/#launchpad) so... anybody know where on mawson to find the regular dogfood db dump?
[05:45] (elmo/#launchpad) locate(1) sez /srv/launchpad.ubuntu.com/database-backups/
[05:46] <carlos> anyone knows how to execute launchpad?
[05:46] <carlos> dogfood one is down and rosetta alpha is also down
[05:46] (elmo/#launchpad) as of, yesterday, rosetta alpha was still running as daf
[05:46] (elmo/#launchpad) dunno if that's been fixed yet tho
[05:47] <carlos> elmo: I'm working on kill that instance
[05:47] <carlos> but it's supposed to be running until I finish importing some data
[05:47] <carlos> elmo: where are you leaving today?
[05:48] <carlos> we will need your help to do the virtual domain change
[05:48] <carlos>  /s/where/when/
[05:48] (elmo/#launchpad) carlos: I'm at home - so I'm not going anywhere
[05:48] <carlos> but I suppose you will go to sleep :-P
[05:49] (elmo/#launchpad) sleep's usually optional if necessary
[05:50] (sabdfl/#launchpad) WALLS! look out for the fast-moving WALLS!
[05:50] (sabdfl/#launchpad) elmo: mawson thinks it's rosetta?
[05:50] <carlos> X-)
[05:51] (elmo/#launchpad) sabdfl: it's a bit schizo still - I'll finish off removing vestiges of the old name after I've done the uploader
[05:52] (sabdfl/#launchpad) ah
[05:52] (sabdfl/#launchpad) that would be a nice goal for hoary
[05:52] (sabdfl/#launchpad) "make it so the machine name is specified in one-and-only-one place"
[05:53] (sabdfl/#launchpad) (sane place)
[05:53] <kiko> hmmm 
[05:53] <kiko> how do we deal with /etc/hosts?
[05:55] (elmo/#launchpad) hmm, pychecker doesn't work with the zope code
[05:55] (elmo/#launchpad) do you guys not use it?
=== BradB never does
[05:56] <BradB> getting too crazy about code quality too early isn't worth it (moderation, everything in, etc. but still, the most important thing is having good tests proving that your stuff works))
[05:57] <BradB> elmo: particularly because we have this rule with launchpad development now that says "if somebody makes a change and it breaks your app, it's *your* fault" :)
[05:59] <BradB> SteveA: is your pqm merge looping?
[06:00] (elmo/#launchpad) yes
[06:00] <BradB> darn
[06:00] (elmo/#launchpad) killed.
[06:00] <BradB> indeed
[06:01] <BradB> here goes mine...
=== Kinnison stares at a diff in "what the fuck is that then" mode
[06:03] <carlos> is postgresql thread friendly?
=== kiko is now known as kiko-fud
[06:04] <carlos> I mean, if I execute several process importing data into postgres will it use the other mawson's processors?
[06:05] <kiko-fud> do you mean frontend or backend threads?
[06:06] (elmo/#launchpad) carlos: you're likely to be IO bound, not CPU bound anyway
[06:06] (sabdfl/#launchpad) BradB: where did you find those instructions for restoring a dogfood dump?
[06:06] <carlos> elmo: ok
[06:06] <carlos> kiko-fud: backend
[06:06] (elmo/#launchpad) oh dear god - pychecker has some how broken my tree
[06:07] (sabdfl/#launchpad) elmo: correctly, i hope
[06:08] (elmo/#launchpad) no, as in the uploader, even Steve's original unmodified thing no longer imports. my brain is going to implode.
[06:08] <BradB> sabdfl: in stub's email to lp@. my suspicision is that there were chars in the dump file that OS X didn't like, and then all hell broke loose (it spit out tons of "Invalid character \N" and "Invalid character \n" messages, then somehow start running the test suite, several times over, then onto rm -rf /'ing)
[06:08] (sabdfl/#launchpad) exciting
[06:09] <BradB> elmo: who knows how to find out what the last pqm merge was on chinstrap?
[06:09] <BradB> mine started, but i have no idea what ended up happening to it.
[06:10] (sabdfl/#launchpad) oh god
[06:10] (sabdfl/#launchpad) that's definitely not going to work
[06:11] (elmo/#launchpad) bradb: star-merge: unable to merge unrelated trees.
[06:11] (sabdfl/#launchpad) piping to less gives a ton of binary nonsense
[06:11] <BradB> elmo: wee, that sounds fun.
[06:11] (elmo/#launchpad) BradB: is what's at the end of the log for your pqm request
[06:11] (elmo/#launchpad) I assume you'll get mail about it eventually
[06:12] (elmo/#launchpad) arch command star-merge ('brad.bollenbach@canonical.com/launchpad--devel--0',) failed (512): tla exited with error code 2* build reference tree for brad
[06:12] (elmo/#launchpad) .bollenbach@canonical.com/launchpad--devel--0--patch-135
[06:12] (elmo/#launchpad) that too, at the top
[06:12] (elmo/#launchpad) everything else is noise
[06:14] (sabdfl/#launchpad) BradB: those instructions must be bogus
[06:15] <BradB> sabdfl: did you try following them? i'd be curious to know if they worked for you.
[06:15] (sabdfl/#launchpad) no, they don't
[06:15] <BradB> i did it verbatim, i.e. exact same day and everything.
[06:15] (sabdfl/#launchpad) the bunzip2 produces a semi-binary file
[06:16] <BradB> :/
[06:16] (sabdfl/#launchpad) there is SQL in there, but not as you know it
[06:16] (sabdfl/#launchpad) it's all mixed with binary junk
[06:16] <BradB> sounds like my hunch was correct perhaps :/
[06:20] (sabdfl/#launchpad) ok, this output should be sent to pg_restore
[06:20] (sabdfl/#launchpad) but it has problems
[06:20] (sabdfl/#launchpad) and i suspect stubb will need to do some more work
[06:47] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: added batching/paging to bug listing (patch-824)
[06:53] <BradB> sabdfl: i learned that lesson the hard way. on the upside, it gave me an excuse to install ubuntu. :P
=== BradB gets food before continuing on the Super Widget
=== BradB is now known as BradB|lunch
[07:04] (daf/#launchpad) carlos: hi
[07:04] <carlos> daf: hi
[07:05] (daf/#launchpad) how's things?
[07:05] <carlos> I had some problems with the scripts (will fix with the planned refactoring) but now I'm importing the first set of files into a temporal database
[07:05] <carlos> it takes too long
[07:05] <carlos> it's too long
[07:06] (daf/#launchpad) hmm, really?
[07:06] <carlos> yep
[07:06] <carlos> the pot was really fast
[07:06] (daf/#launchpad) ok, if one or two files work in the temporary database, then I think it's ok to do the real thing
[07:07] (daf/#launchpad) ah, but the PO files are slow?
[07:07] <carlos> yep
[07:07] <carlos> also, the statistics update method seems like it's broken 
[07:07] <carlos> because it always give you 0
[07:07] (daf/#launchpad) ok, I guess we should try doing more profiling work
[07:07] (daf/#launchpad) hmm
[07:07] <carlos> but that's not a problem
[07:07] <carlos> we could fix it later
[07:08] <carlos> daf: I want to add some code to allow imports that don't have plural form information
[07:08] <carlos> and they don't have any plural form msgset
[07:08] <carlos> like gnome-panel, lots of files are being rejected because we don't have the plural form information
[07:08] (daf/#launchpad) that would be fine
[07:09] (daf/#launchpad) that's the way things should be
[07:09] <carlos> and rosetta and dogfood server are down 
[07:09] (daf/#launchpad) if there are no plural msgstrs, no plural form information should be necessary
[07:09] (daf/#launchpad) they're down?
[07:09] <carlos> yep
[07:10] <carlos> I asked for the virtual info change already so we don't depend on admins
[07:10] <carlos> as the .po files are being imported already (you should have it in your inbox)
[07:11] (daf/#launchpad) yep, just read it
[07:11] (daf/#launchpad) looks good
[07:12] <carlos> daf: so, I'm going to import real data into dogfood
[07:12] <carlos> is that ok?
[07:12] <carlos> :-)
[07:12] (daf/#launchpad) yeah, if the test import seemed to work
[07:13] (daf/#launchpad) any idea on why the PO import might be so slow?
[07:14] <carlos> no idea, I need to do more profile
[07:14] <carlos> did we solved the problem that made an update with every field change?
[07:15] (daf/#launchpad) no
[07:15] (daf/#launchpad) well, not sure
[07:15] (daf/#launchpad) we can try using the same techniques as we did before for analysing the performance
[07:16] (daf/#launchpad) I suggest we get a Postgres log for an import of a 2-msgid PO file
[07:16] <carlos> ok
[07:16] <carlos> let me start the import
[07:16] (daf/#launchpad) and examine the queries and how long they're taking
[07:16] (daf/#launchpad) sure, go for it
[07:16] <carlos> and I get on it
[07:17] <carlos> importing
[07:23] <carlos> ok, profile mode on
[07:23] <carlos> :-P
[07:24] (daf/#launchpad) on your own machine?
[07:25] <carlos> yes
[07:26] <carlos> it's easier for me to work that way
[07:26] <carlos> it's slower, but easier
[07:26] <carlos> and mawson is already with a high load
[07:26] <carlos> importing .po files
[07:26] (daf/#launchpad) right, and it would slow down the import even more
[07:26] (sabdfl/#launchpad) so guys, can imports only be done behind the scenes via a script?
[07:26] (sabdfl/#launchpad) how big is a really big po file?
[07:27] (daf/#launchpad) the biggest in Ubuntu are aroudn 4500 messages
[07:27] (sabdfl/#launchpad) so not huge
[07:27] (daf/#launchpad) well, that is huge
[07:27] (daf/#launchpad) comparatively speaking
[07:27] (daf/#launchpad) it might be a week of work for a translator working full time
[07:28] (sabdfl/#launchpad) not huge in storage terms
[07:29] (sabdfl/#launchpad) seems we could create a table to store "uploaded PO files"
[07:29] (sabdfl/#launchpad) i'm trying to think of a way to get people putting PO files into rosetta themselves
[07:29] <carlos> hmmm
[07:29] <carlos> sabdfl: but only as a temporal storage, right?
[07:29] (sabdfl/#launchpad) carlos: exactly
[07:30] (sabdfl/#launchpad) so it's a link off the project / product pages
[07:30] (sabdfl/#launchpad) "start translating this product"
[07:30] (SteveA/#launchpad) how big is it in mk ?
[07:30] (SteveA/#launchpad) how big is it in kb ?
[07:31] (SteveA/#launchpad) I mean, how big is that 4500 message file in kb?
[07:31] <carlos> that's more or lees the web upload feature we have planned to implement but adding a moderation queue
[07:32] (SteveA/#launchpad) we could implement a queue that puts a raw po file in the zodb, and then have a separate app server / script to process that into the database behind the scenes
[07:32] <carlos> SteveA: I think it's less than 1 MB
[07:32] (SteveA/#launchpad) so, there are no technical problems there
[07:32] <carlos> but it could reach the 1MB easily
[07:33] <carlos> SteveA: that procedure sounds good
[07:34] (SteveA/#launchpad) and, it certainly fits the "this can be nuked and it isn't a disaster" model of using the zodb 
[07:34] (SteveA/#launchpad) we'd change from MemoryStorage to FileStorage, though.  that's no big deal.
[07:36] (sabdfl/#launchpad) i think we could do a "sanity check" on the file, and store it, then Daf could import it behind the scenes, and mail the user that it's ready to go
=== salgado [~salgado@200-206-134-238.async.com.br]  has joined #launchpad
[07:38] (sabdfl/#launchpad) the main thing would be to keep it really simple, but get it up asap
[07:39] (daf/#launchpad) SteveA: well, the size might vary a lot, but one 4000-message template I have here is nearly 400kb
[07:39] (daf/#launchpad) storing PO files on the server sounds to me like something the librarian might do for us
[07:40] (SteveA/#launchpad) there's the issue of storing the file, and the issue of maintaining the queue.  the queue could be as simple as daf's mailbox for now.
[07:40] (sabdfl/#launchpad) i dont want to get complicated
[07:40] (daf/#launchpad) yeah, let's do the simplest thing that works
[07:40] (sabdfl/#launchpad) daf's mailbox hasn't got us a lot of traction thus far ;-)
[07:41] (SteveA/#launchpad) storing the file could be as simple as sticking it on the filesystem in a particular directory under a name that gets mailed to daf
[07:41] (SteveA/#launchpad) or even, just mail daf the po file
[07:41] (SteveA/#launchpad) mailing daf the whole file is easiest for now.
[07:41] (SteveA/#launchpad) 30 minutes' work
[07:41] (SteveA/#launchpad) to write a mail thing that has an attachment
[07:41] (SteveA/#launchpad) and that includes reading the example in the python standard lib
[07:41] (sabdfl/#launchpad) except we want to be able to show in the web form that there is a translation there waiting to be processed
[07:42] (SteveA/#launchpad) then use an annotation to store the metadata
[07:42] (sabdfl/#launchpad) but yes, mail would be a start that could be done... tonight ;-)
[07:42] (SteveA/#launchpad) I need to check with stub that keeping a Data.fs around on the dogfood server is okay, and how the config file should look for that
[07:44] <carlos> does any one have rights to execute launchpad's dogfood?
[07:44] <carlos> it would help me a lot to test that the po imports are being done correctly...
[07:45] (elmo/#launchpad) carlos: dude, you do
[07:45] (elmo/#launchpad) it runs as 'launchpad'
[07:45] (elmo/#launchpad) you can become the launchpad user now
[07:45] <carlos> elmo: ok, let me ask again....
[07:46] <carlos> does any one have rights and the knowledge to execute launchpad's dogfood?
[07:46] <carlos> :-)
[07:46] (elmo/#launchpad) heh
[07:46] <carlos> I don't have any idea about how it's executed
[07:46] <carlos> don't think a screen + make run is the usual solution...
[07:50] (daf/#launchpad) I think it might be as simple as running "make run" somewhere in /src
[07:50] (daf/#launchpad) using screen gets around the fact that launchpad is not daemonised yet
[07:55] <carlos> ok, dogfood server working again
[07:55] <carlos> aarrgg
[07:55] <carlos> and I forgot to import the .pot file first
[07:56] <carlos> :-(
[07:57] (daf/#launchpad) oops
[07:58] <carlos> daf: isn't dogfood server automatically updated from rocketfuel?
[08:03] (daf/#launchpad) I don't know, I'm afraid
[08:03] <carlos> https://launchpad.ubuntu.com/rosetta/projects/gnome/gnome-panel/main-2.8/translate
[08:03] <carlos> ok, the pot was imported
[08:04] (daf/#launchpad) ah, good
=== BradB|lunch is now known as BradB
[08:29] (sabdfl/#launchpad) carlos: no, it isnt auto-updated
[08:29] (sabdfl/#launchpad) stub tends it lovingly
[08:29] (sabdfl/#launchpad) night all
[08:30] <carlos> sabdfl: ok, night
[08:37] <carlos> daf: there is a problem with the virtual hosting
[08:37] <carlos> of rosetta.shuttleworthfoundation.org
[08:38] <carlos> SteveA, daf: around?
[08:38] <carlos> we get links to: https://rosetta.shuttleworthfoundation.org/rosetta/prefs
[08:38] <carlos> when it should be https://rosetta.shuttleworthfoundation.org/prefs
[08:42] (daf/#launchpad) carlos: hmm
[08:42] (daf/#launchpad) might be a matter of moving the /+/
[08:42] (daf/#launchpad) or /++/, or whatever it is
[08:42] (daf/#launchpad) the virtual hosting URLs confuse me
[08:43] <carlos> daf: well, me too, if that makes you feel better :-)
[08:45] (SteveA/#launchpad) hello carlos
[08:45] <carlos> SteveA: hi
[08:45] (SteveA/#launchpad) what does the vh config look like now?
[08:45] (SteveA/#launchpad) is it as in the email?
[08:46] <carlos> ProxyPass / http://localhost:9020/rosetta/++vh++https:rosetta.shuttleworthfoundation.org:443/++/
[08:47] (SteveA/#launchpad) that looks like it should work as you want
[08:47] (SteveA/#launchpad) oh
[08:47] (SteveA/#launchpad) it is not a virtual hosting problem
[08:47] (SteveA/#launchpad) it is just that the links are wrong
[08:48] (daf/#launchpad) ah
[08:48] (SteveA/#launchpad) the breadcrumbs are wrong too
[08:48] <carlos> absolute links instead of relative ones?
[08:48] (SteveA/#launchpad) well, they can't be relative ones
[08:49] (SteveA/#launchpad) I suggest a quick hack for now
[08:50] (SteveA/#launchpad) that is, add some code in the page template that is the rosetta main template
[08:50] (SteveA/#launchpad) on the rosetta layer
[08:50] (SteveA/#launchpad) that looks to see if there is "rosetta/++vh++" in the URL
[08:50] (SteveA/#launchpad) and if so, uses alternative links
[08:50] (SteveA/#launchpad) this is a hack until I have done the "context-specific urls" work
[08:51] <carlos> ok
[08:51] (SteveA/#launchpad) also, don't show stuff that you need to be logged in to see until you're logged in
[08:51] <carlos> daf: could you do it? or should I add it into my TODO list?
[08:51] (SteveA/#launchpad) I guess I should make the standard unauthorized page redirect to /+login?camefrom=URL
[08:52] (SteveA/#launchpad) but, I must go and watch a film now
[08:52] (daf/#launchpad) ok
[08:52] <carlos> SteveA: enjoy it
[08:52] (SteveA/#launchpad) thanks
[08:53] (daf/#launchpad) we should probably take the breadcrumbs off the main template
[08:53] (daf/#launchpad) that's one change I missed
[08:57] <carlos> daf: I'm going to leave also in about 30 minutes or so 
[08:57] <carlos> daf: I hope tonight the initial import is done
[08:58] <carlos> one thing I will try to fix this weekend is the statistics thing
[08:58] <carlos> because althought there are translations the pofile thinks it's 100% untranslated
[09:05] (daf/#launchpad) hmm
[09:05] (daf/#launchpad) one hack to fix the links would be to redirect /rosetta/rosetta/* to /rosetta/*
=== carlos don't have any idea about it
[09:07] <dilys> New Malone bug #55: "we should be able to import always pofiles without plural forms", submitted by Carlos Perell Marn
[09:07] <dilys> https://launchpad.ubuntu.com/malone/bugs/55
[09:08] <carlos> daf: dude, we need to optimize the import...
[09:08] <carlos> real    62m52.963s
[09:08] <carlos> user    2m39.403s
[09:08] <carlos> sys     0m9.266s
[09:08] <carlos> 8 po
[09:08] <carlos> that's in mawson
[09:08] (daf/#launchpad) ouch
[09:08] <carlos> about 512 messages each po
[09:09] <carlos> aaarrrggggggggg
[09:09] <carlos> man, I really suck
[09:10] <carlos> I think the import is not being done as it should...
=== carlos checks
[09:11] <carlos> ok, the import was done correctly
[09:13] <carlos> I cannot understand how is that a .pot import is so fast
[09:13] <carlos> but the .po is so slow...
[09:14] <carlos> the gnome-applet's .pot import took about 1 minute (I think less than a minute) and it has more than 1300 msgsets
[09:14] (SteveA/#launchpad) well, a pot file stands on its own, and is related to its product
[09:14] (SteveA/#launchpad) a po file is related to the pot file
[09:15] (SteveA/#launchpad) so importing a po has an effect on the pot data in terms of relationships
[09:15] (SteveA/#launchpad) (getting food before the film)
[09:15] (SteveA/#launchpad) (ridicule, groovy french film about institutionalized piss-taking)
[09:15] <carlos> :-P
[09:16] <carlos> SteveA: I know a .po file should take more time, but about 10 times more ?
[09:17] (SteveA/#launchpad) hmm
[09:17] <carlos> In about 1 hour we will have the data imported (finally)
[09:18] (SteveA/#launchpad) okay, that's great
[09:18] (SteveA/#launchpad) I wonder if we can profile po importing to see what takes all the time?
[09:18] <carlos> SteveA: yes, daf asked me that already
[09:18] <carlos> we did it already a month ago
[09:18] <carlos> we should do it again with latest code changes
[09:20] (SteveA/#launchpad) what was taking all the time a month ago?
[09:21] <carlos> duplicated queries, several updates for the same table instead of just one update with all changes and some other things I don't remember at the moment
[09:21] <carlos> let me check
[09:22] <carlos> https://bugzilla.warthogs.hbd.com/bugzilla/show_bug.cgi?id=1973
[09:26] <carlos> ok, I should leave now
[09:27] <carlos> if you need anything from me just send me an email, I will try to be online tomorrow to fix the statistics of the imported pos
=== kiko-fud is now known as kiko
=== lifeless_ is now known as lifeless
[10:41] <kiko> sabdfl, I finally have an account on malone
[10:42] <kiko> BradB, sabdfl: isn't "the complete bug list" something resembling 200,000 bugs on bugzilla.mozilla.org?!
[10:43] <kiko> BradB, Add -> Add new bug or Report new bug on the complete bug list page, btw.
[10:47] <BradB> kiko: no idea, but, as you can see, our "complete bug list" is filtered on what the user wants to see. it'll get better filtering once the super widget is in place.
[10:47] <kiko> hopefully
[10:48] <kiko> we have 250,000 bugs in bugzilla.mozilla.org.
[10:48] <BradB> so by default, you'll only see what you want to see, which is unfixed bugs (new or accepted) for stuff you maintain.
[10:48] <kiko> got my suggestion for the button label change?
[10:48] <BradB> yeah
[10:48] <kiko> the text for the front page should change, then
[10:48] <BradB> it will, yeah
[10:48] <kiko> it should say "View my bug list" or "View open bugs assigned to me" or "in packages I maintain".
[10:48] <kiko> but okay, back to reporting my karmabug.
[10:48] <BradB> kiko: those are "quick searches"
[10:49] <BradB> that will be available on the listing page itself.
=== kiko clicks a link and sees this massive Select SourcePackage window
[10:49] <kiko> what is this "4" thing?
[10:49] <kiko> 4?
[10:49] <BradB> kiko: ask stub, he implemented it. :)
[10:49] <BradB> kiko: what i'm working on now is making a widget that doesn't suck.
[10:50] <BradB> using "Sources" in Zope 3, which are somewhat new.
[10:50] <kiko> I see
[10:50] <BradB> Sources, btw, are Z3's thing to deal with huge vocabs.
[10:50] <kiko> BradB, cprov needs some timpeters-fu on his difflib hack, I'm going to try and proxy some upstreamage
[10:51] <lifeless> kiko: know of a patchlib for python ? something that can parse patches ?
[10:51] <kiko> nope -- difflib goes the reverse direction, too. hmmmmm.
[10:51] <kiko> I know of one for Perl if that helps..
[10:52] <BradB> Perl rocks
[10:52] <kiko> lifeless, "parse" in what sense, btw?
=== kiko posts bug
[10:53] <kiko> waits waits wait
[10:53] <kiko> HOHO!
[10:54] <kiko> BradB, bug for you babe
[10:54] <dilys> New Malone bug #56: "Implement initial karma-gathering framework", submitted by Christian Robotton Reis
[10:54] <dilys> https://launchpad.ubuntu.com/malone/bugs/56
[10:54] <kiko> BradB, look at bug 56.
[10:54] <BradB> KARMA?!!!!
[10:54] <kiko> package 44bsd? :)
[10:54] <kiko> that's borks
[10:54] <BradB> kiko: dude, stub implemented that dirty search widget stuff.
[10:54] <kiko> not just search widget, dude, it misreported the package
[10:54] <BradB> i was a bit surprised when it just put ID's in the boxes, but oh well, it's going away anyway.
[10:55] <lifeless> kiko: parse in take (say) the output of diff -NRup olddir newdir and create an iterable of file-diffs, each of which is a collection of hunks
[10:55] <kiko> that's not my point -- the ID was translated into something else
=== BradB looks
[10:55] <kiko> lifeless, hmmmm. yep, beats me
[10:56] <kiko> dude, filling out the summary is a pain in the @#@!!
[10:56] <BradB> kiko: i know :/
[10:56] <kiko> it should *not* be a requisite for filing bugs, really
[10:56] <kiko> it can be an optional field to be used instead of the initial description, IMO.
[10:56] <BradB> kiko: i already fixed it, with patch-100, but sabdfl refused to let it happen. :/
[10:56] <BradB> kiko: he'll change his mind though, because *every* user (including me) has whined about it.
[11:02] <kiko> in bugzilla, the initial description is the bug summary
[11:02] <kiko> it would be trs cool to have an optional summary to substitute it
[11:02] <kiko> but most bugs wouldn't use it, because some bugs have short lifespans
[11:04] <kiko> BradB, I get the feeling also that unless we want do discourage ascii art in the description, the textarea should be 80 cols wide and server-wrapped.
[11:04] <kiko> it's very useful in bugzilla
[11:05] <dilys> New Malone bug #57: "Collect karma for bugs reported", submitted by Christian Robotton Reis
[11:05] <dilys> https://launchpad.ubuntu.com/malone/bugs/57
[11:07] <kiko> BradB, I shouldn't get an +add control if I am already ..
[11:07] <kiko> oh.
[11:07] <kiko> the +add control in the box is confusing 
[11:09] <kiko> how do I CC: another user?
=== BradB [~bradb@modemcable202.193-131-66.mc.videotron.ca]  has joined #launchpad
[11:13] <BradB> ugh, that's the second time gnome's hung on me today
[11:17] <kiko> <kiko> BradB, I shouldn't get an +add control if I am already ..
[11:17] <kiko> <kiko> oh.
[11:17] <kiko> <kiko> the +add control in the box is confusing 
[11:17] <kiko> <kiko> how do I CC: another user?
[11:17] <kiko> BradB, also, see bug 56
[11:17] <kiko> first my comment came out with some gnarly double-spacing
[11:17] <kiko> then 44bsd-rdist is confusing me
[11:18] <BradB> kiko: to Cc people, just click Subscribe. there's no way yet to do it while adding a bug, but i'm aware that that'll be required.
[11:18] <BradB> with problems with the widget, i think stub would be the one to address that.
[11:19] <BradB> kiko: that ugly <pre> crap was also done by stub.
[11:20] <BradB> on bug messages.
[11:25] <kiko> BradB, could the Person listing in the Add Person page (bad title) be ordered in some way?
[11:27] <BradB> kiko: yeah, but again, the person, sourcepackage, product widgets are going away, so anything else done to them is more-or-less wasted effort.
[11:28] <kiko> hmmm, going away.
[11:30] <BradB> to be replaced by something that's usable, yeah.
[11:30] <BradB> which will look and be used somewhat differently.
[11:30] <BradB> i.e. search and select
[11:31] <kiko> hmmm
[11:31] <kiko> okay.