[11:41] <jetienne> q. when somebody with good knowledge of unionfs ? if i modify a file, it copies the whole file and then do the modficiation ? or does it store only the modification ?
[11:53] <cjwatson> jetienne: my understanding is the former
[11:54] <jetienne> cjwatson: no issue if i modify the first byte of a 5gbyte file ?
[11:56] <cjwatson> then you'll suddenly find 5GB more data in the overlay part of your filesystem :-)
[11:57] <jetienne> cjwatson: cool :) thanks
[11:57] <cjwatson> unionfs is not bad at dealing with partial changes to entire filesystems efficiently, but it is not really designed to do the same for individual files
[12:20] <ev> I'm just waiting for 2.6.32-12 to hit the archive, then I'll push out a new d-i with the kernel version bump and console-setup fix.
[12:22] <StevenK> ev: I didn't NEW it since it hadn't built everywhere and accepting kernels is a large pain.
[12:23] <ev> no worries
[12:38] <ev> cjwatson: is it okay with you if I add an item to the sprint agenda for the two of us to sit down with michaelforrest and discuss the installer design spec, or would you prefer to handle that via email / comments in the document?
[12:42] <cjwatson> ev: agenda fine
[12:42] <ev> cool, thanks
[12:42] <cjwatson> though perhaps consider me optional just in case
[12:42] <ev> sure thing
[12:54] <exlt> 20081029ubuntu79 no longer functions as a workaround to console-setup kbd grok.. :(  appears the 2.6.32-10 udebs are no longer on mirrors
[12:54] <exlt> ev: thanks for working on the fixup - I see the cia notes on it
[12:55] <exlt> and cjwatson :)
[13:55] <dmarkey> is it possible to give a http url for preseed/file with the cdrom
[13:59] <cjwatson> no, that's the purpose of preseed/url
[15:07] <dmarkey> cjwatson: thanks
[15:24] <ev> superm1: apologies for taking so long to get to it, but I'm reviewing your branch now.
[15:24] <superm1> ev, cool thanks, not a big problem
[19:56] <cr3> my usual preseed now seems to prompt the user for "configure the keyboard", when this use to be answered automatically
[20:05] <cody-somerville> cr3, d-i or ubiquity?
[20:06] <cr3> cody-somerville: d-i
[20:07] <cr3> cody-somerville: but if there's a change I should know about in ubiquity too, I'm interested. I could test it in a moment
[20:07] <cody-somerville> cr3, Sounds like you're missing "auto" from your boot arguments.
[20:16] <ev> cr3: it's a known bug
[20:16] <ev> it's been adddressed, I've just been waiting on kernel builds so I can push out a new d-i
[20:17] <ev> which I'll do now, actually
[20:17] <cr3> ev: thank!
[20:22] <CIA-3> debian-installer: evand * r1232 debian-installer/debian/changelog: No-change rebuild to pick up new components.
[20:24] <ev> (wasn't NEW'ed yet, that will have to wait until tomorrow)
[20:27] <CIA-3> debian-installer: evand * r1233 debian-installer/debian/changelog: releasing version 20081029ubuntu81
[20:28] <ev> cr3: tomorrow's CDs should be fixed ^
[20:28] <cr3> ev: you rock, thanks!
[20:28] <ev> sure thing
[20:28] <cr3> ev: do you know if the bug also affects ubiquity? I'll be trying an automated install in a moment
[20:29] <ev> I don't believe it does
[20:35] <ev> superm1: looks good.  Am I okay to merge this (I've made a few small typo corrections)?
[22:21] <superm1> ev, yeah it's in good shape to merge
[22:21] <superm1> there's some pages that need UI work with it still, but those will be worked out later
[22:22] <superm1> i think there should only be one merge conflict currently and it should be clear what to do with it
[22:25] <superm1> although you need to make sure that https://code.edge.launchpad.net/~mythbuntu/debian-cd/fix-autologin/+merge/18059 is merged before the next ubiquity release
[22:33] <ev> hooray, znc playback actually worked for once
[22:35] <CIA-3> ubiquity: evand * r3718 ubiquity/ (29 files in 9 dirs): Merge with Mario's Mythbuntu plugin branch.
[22:36] <superm1> ev, cool thanks.  curious, what's znc provide over bip?
[22:36] <ev> for me, a iPhone push notifications plugin that doesn't work because I was silly enough to pick a two character IRC nick
[22:37] <superm1> haha
[22:37] <superm1> ah it looks like it's more featureful than bip for sure as i'm reading their pages
[22:37] <superm1> a webadmin ui?  bip has a flat text file that is pia to configure unless you copy it from kirkland` to start
[22:40] <ev> lol
[22:41] <ev> this also lets you configure interactively from the shell, or create a minimal configuration and use "saveconfig" IRC messages to *admin.
[22:42] <CIA-3> tzsetup: evand * r509 tzsetup/ (debian/changelog tzsetup):
[22:42] <CIA-3> tzsetup: Fix the sed statement around the geoip detection to account for the
[22:42] <CIA-3> tzsetup: XML being on a single line.
[22:42] <ev> ugh, a million times fail ^
[22:45] <superm1> i'll have to add it to my list of stuff to play with at some point
[22:47] <CIA-3> tzsetup: evand * r510 tzsetup/debian/changelog: releasing version 1:0.26ubuntu3
[22:53] <ev> hrm, quitting ubiquity seems to just run the component over again now
[22:58] <superm1> hm, can't say i tried that... was the previous handler dependent on looking for self.current_page == None?
[23:01] <superm1> i think that's what it was.  so the alternative would be to reinstate that break on current_page = None, and instead don't set it to none during progress_loop
[23:03] <ev> indeed, though it needs to return from there
[23:03] <ev> as breaking will show the reboot dialog
[23:13] <CIA-3> ubiquity: evand * r3719 ubiquity/ubiquity/frontend/gtk_ui.py: Properly exit the plugin loop when the quit button is pressed.
[23:13] <ev> I really need to do something about ubiquity/keyboard_names.py, I suspect that's seriously slowing us down.
[23:14] <ev> what with it being gigantic
[23:29] <cjwatson> ooh, of course, translations would have expanded it rather a bit
[23:30] <cjwatson> must do interesting things to memory use too :)
[23:32] <ev> indeed, that's what I mean
[23:32] <ev> apologies for the confusing wording
[23:33] <ev> it's over a megabyte of python data structure goodness
[23:33] <ev> clearly that's abuse :)
[23:37] <cjwatson> it sort of made sense to reformat the original perl as python to start with, but yeah, maybe a quick-to-parse text file or set of files would be better now
[23:40] <ev> yeah, definitely.  I've noted this so I don't forget.
[23:41] <ev> damn, monodevelop-python's as you type syntax checking is so far thoroughly disappointing