/srv/irclogs.ubuntu.com/2009/03/13/#ubuntu-installer.txt

shtylmanevand: thanks for the heads up...I will take a look...ran it several times without issue...:/00:53
Scixcjwatson, where can I find pitti?06:42
superm1Scix, usually in #ubuntu-devel I see him, but i'd check /whois06:43
Scixsuperm1, tanks :) I'll try that06:43
=== mpt_ is now known as mpt
StevenKevand: Did you see mterry's oem-config bug?11:40
evand#?11:40
StevenKevand: Digging it up11:41
StevenKevand: #34175811:42
evandah, very cool.  I'll review and merge it in a little bit (just need to get through some other things that have been piled on this morning)11:45
evandStevenK: thanks for pointing me at that11:45
StevenKevand: No worries.11:45
CIA-3ubiquity: cjwatson * r3098 ubiquity/ (debian/changelog scripts/install.py):13:03
CIA-3ubiquity: broken_packages is fairly slow due to having to iterate over the whole13:03
CIA-3ubiquity: cache. Speed it up a bit by stopping when the number of broken packages13:03
CIA-3ubiquity: found reaches cache._depcache.BrokenCount; this improves blacklist13:03
CIA-3ubiquity: calculation time for the previously-mentioned DVD English-only install13:03
CIA-3ubiquity: from 3:30 to 2:30 (see LP #335596).13:03
ubottuLaunchpad bug 335596 in ubiquity "DVD install takes forever compared to the CD install during the step that it's calculating packages to remove, causing pain for OEM" [High,Triaged] https://launchpad.net/bugs/33559613:03
cjwatsonI have a suspicion that if we topologically-sort packages to be removed, it might go faster13:09
cjwatsonthe path where we have to identify broken packages and remove them too is slower13:09
evandout of curiosity, what method are you using to profile this?13:23
cjwatsonevand: time(1) ;-)13:25
evandheh13:26
cjwatsonI did try python-profiler but it threw a fit at something in python-apt13:26
cjwatsonso I've been sprinkling prints around to get a feel for what's taking time13:26
cjwatsonold-fashioned approach ...13:26
evandindeed13:30
cjwatsonbeginning to run out of ideas for making it very much faster though13:31
CIA-3ubiquity: evand * r3099 ubiquity/ (debian/changelog scripts/install.py):13:47
CIA-3ubiquity: Fix a bit of code that wasn't updated to reflect other changes in13:47
CIA-3ubiquity: remove_extras.13:47
evandI *think* that's right.13:48
cjwatsonevand: yes, looks right, sorry about that13:50
cjwatsonI got a bit confused because blacklisting and removal handle that differently13:50
cjwatsonI can't seem to get the topological sort approach to work, but on reflection I think it is going to be of reduced benefit anyway because there are so many dependency cycles in language packs13:53
evandno worries13:53
evandmm, ouch13:53
cjwatsoni.e. you end up going down the broken-packages slow path for all the language-pack-foo/language-pack-foo-base pairs anyway13:53
cjwatsonI wonder if it would be worth special-casing those ...13:54
cjwatsonlooks pretty fiddly though13:54
cjwatsonget_remove_list only ever tries one at a time at the moment13:54
evandhrm, still might be worth special casing if we run out of options.13:56
cjwatsonI think I'll wait and see whether the stuff I've done so far is enough13:56
CIA-3debian-installer: cjwatson * r1062 ubuntu/ (build/config/armel/orion5x/netboot.cfg debian/changelog):14:07
CIA-3debian-installer: Disable armel/orion5x/netboot/mv2120 flavour for now, since the kernel14:07
CIA-3debian-installer: is slightly too big.14:07
CIA-3debian-installer: cjwatson * r1063 ubuntu/debian/changelog: releasing version 20081029ubuntu2514:10
CIA-3oem-config: evand * r630 oem-config/ (debian/changelog lib/frontend/gtk_ui.py):16:36
CIA-3oem-config: Fix 'resize if too big for screen' logic (LP: #341758). Thanks Michael16:36
CIA-3oem-config: Terry!16:36
=== robbiew1 is now known as robbiew
CIA-3wubi: evand * r94 trunk/ (debian/changelog src/wubi/backends/win32/backend.py): (log message trimmed)18:07
CIA-3wubi: * os.environ is a dictionary, not a callable method.18:07
CIA-3wubi: * Only uncompress files in the boot directory.18:07
CIA-3wubi: * Call the correct bootloader installation method for each version of18:07
CIA-3wubi:  Windows.18:07
CIA-3wubi: * Use registry.{get,set}_value not the nonexistent18:07
CIA-3wubi:  registry.{get,set}_key.18:07
evandSo much easier to hack on.18:08
CIA-3ubiquity: cjwatson * r3100 ubiquity/ (3 files in 3 dirs):18:42
CIA-3ubiquity: * GTK frontend:18:42
CIA-3ubiquity:  - Restore set_window_hints method for use by windows other than the main18:42
CIA-3ubiquity:  one (it was still called in the Glade file), just in case we're using18:42
CIA-3ubiquity:  a window manager that pays attention to this. In these cases18:42
CIA-3ubiquity:  maximisation doesn't really make sense so we no longer permit that.18:42
CIA-3ubiquity: cjwatson * r3101 ubiquity/debian/changelog: Evan's r3099 fixes LP: #34231918:47
CIA-3oem-config: cjwatson * r631 trunk/ (3 files in 3 dirs):18:50
CIA-3oem-config: * GTK frontend:18:50
CIA-3oem-config:  - Full-screen the window when --only is used - i.e. more or less always,18:50
CIA-3oem-config:  except for testing. The extra screen real estate makes the timezone18:50
CIA-3oem-config:  map, among other widgets, easier to use. (Ported from Evan Dandrea's18:50
CIA-3oem-config:  similar change to Ubiquity; the KDE frontend needed no change as in18:50
cjwatsonevand: ^- does oem-config r631 look OK to you?18:50
CIA-3oem-config:  oem-config it's always full-screen.)18:50
CIA-3oem-config: cjwatson * r632 trunk/ (debian/changelog gui/glade/oem-config.glade):18:54
CIA-3oem-config: Use stock preferences-system icon (thanks, Michael Terry;18:54
CIA-3oem-config: LP: #341770).18:54
* evand checks18:58
evandyup, looks good to me19:00
CIA-3wubi: evand * r95 trunk/ (debian/changelog src/wubi/backends/win32/backend.py):19:16
CIA-3wubi: * Only uncompress files in the boot directory (LP: #341640). Thanks Agostino19:16
CIA-3wubi:  Russo!19:16
CIA-3wubi: * subprocess.Popen takes a list of string arguments.19:16
evandcloser.  I've at least got a bootloader entry now.19:16

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!