[01:48] <kirkland> cjwatson: mathiaz and I have both experienced something strange with today's server iso's
[01:49] <kirkland> cjwatson: they seem to slow down (almost hang, but they do recover and finish)
[01:49] <kirkland> cjwatson: usually around configuring the language pack
[01:49] <kirkland> cjwatson: i'll get you more details on the precise point at which the install seems to "stall"
[01:49] <kirkland> cjwatson: i just wanted to mention this now, since mathiaz said he noticed the same thing
[02:15] <cjwatson> I'll need logs
[02:16] <cjwatson> syslog will be sufficient to show me what's happening at what times
[09:08] <evand> kirkland: nope, I've updated bug 349440 accordingly
[09:14] <evand> cjwatson: I'll see what I can do about that bug after I finish this test install using the latest daily-live.
[12:21] <davmor2> cjwatson: have you seen bug 361627 :(
[12:21] <cjwatson> no
[12:22] <cjwatson> well, I have now. Give me a minute
[12:23] <cjwatson> it's a rather specialised case, but I'll see if we can get it fixed
[12:29] <CIA-4> user-setup: cjwatson * r175 ubuntu/ (debian/changelog user-setup-ask):
[12:29] <CIA-4> user-setup: Install ecryptfs-utils if it's already in use by users on an existing
[12:29] <CIA-4> user-setup: /home (LP: #361627).
[12:31] <CIA-4> user-setup: cjwatson * r176 ubuntu/debian/changelog: releasing version 1.23ubuntu19
[12:42] <evand> I have most of the code in place for ubiquity understanding the use_biggest_free option, I'm just looking for the best way to extract the partition that's going to be used for use_biggest_free from partman.
[12:43] <evand> my current thought is to pull it out of /var/lib/partman/snoop
[12:49]  * cjwatson looks
[12:49] <cjwatson> I've sort of been hoping to drop use of the snoop file soon since the choices are now more automatically parseable than they used to be, but I don't object if that's the easiest way ... let's see
[12:53] <cjwatson> use snoop if it's easier to work into the code as it stands, but I think you can also fish it out of choices-c for whichever automatically_partition question you get
[12:53] <cjwatson> choices-c should have an entry something like 50biggest_free__________/var/lib/partman/devices/=dev=sda//32-1048576
[12:54] <cjwatson> and I think that's basically just what's in the snoop file as well (except that the lines in snoop also have <tab><description of partman-auto/text/use_biggest_free> at the end)
[12:55] <cjwatson> indeed, menu_options should have the contents of snoop pre-parsed for you
[12:55] <davmor2> evand: I'll re-test in about 10 minutes
[12:57] <cjwatson> you can use self.find_script(menu_options, 'biggest_free') which will give you a list of (currently) either zero or one elements; if the list is non-empty, its element will be a three-element tuple of which the second element should be something like '/var/lib/partman/devices/=dev=sda//32-1048576'
[12:57] <davmor2> evand: I still think that the oem map should be the new one it looks wrong.  But then I guess the end user would never know
[12:58] <cjwatson> the split_devpart method will take that string apart into device and partition ids
[12:59] <CIA-4> ubiquity: evand * r3216 ubiquity/ (debian/changelog ubiquity/frontend/kde_ui.py):
[12:59] <CIA-4> ubiquity: Fix a crash when in OEM mode. Child widgets are members of the the
[12:59] <CIA-4> ubiquity: self.userinterface list.
[13:00] <evand> cjwatson: that's the exact code I was already toying with :)
[13:03] <evand> cjwatson: (re davmor2 ^) are you still planning on reviewing the patch from shtylman for Jaunty, or have you decided to shelve it to Karmic?
[13:05] <cjwatson> which patch?
[13:05] <cjwatson> oh, oem-config?
[13:05] <cjwatson> I guess I can, I have a local branch with it mostly merged anyway, just need to update to your latest changes
[13:07] <evand> ok
[13:11] <CIA-4> oem-config: cjwatson * r657 trunk/ (47 files in 4 dirs): merge lp:~shtylman/oem-config/timezone_map_gtk, with some additional modifications by me
[13:13] <cjwatson> I have *not* tested that yet
[13:18] <CIA-4> ubiquity: evand * r3217 ubiquity/debian/changelog: Add LP bug reference.
[13:31] <CIA-4> ubiquity: cjwatson * r3218 ubiquity/ubiquity/frontend/kde_ui.py: typos
[13:32] <cjwatson> I'm really not happy with the Kubuntu autopartitioning bar code. It looks quite broken in a number of places :-(
[13:34] <cjwatson> and it definitely isn't working if there are no partitions to resize
[13:34] <cjwatson> I suppose I might be able to bodge it together somehow ...
[13:39] <davmor2> cjwatson: Should ubiquity be able to resize a ext4 partition?
[13:45] <cjwatson> yes
[13:46] <davmor2> cjwatson: I'll get the logs
[13:51] <davmor2> cjwatson: http://www.davmor2.co.uk/partman do you need syslog too
[13:52] <cjwatson> davmor2: what precisely is the problem here?
[13:52] <cjwatson> yes, I need syslog
[13:54] <davmor2> cjwatson: On ubuntu I did a manual install where I used ext4 for basically the whole drive bar swap.   In Kubuntu I've gone to resize the drive and it just says it can't and goes to manual mode
[13:55] <davmor2> http://www.davmor2.co.uk/syslog
[13:55] <cjwatson> Apr 15 12:37:29 ubuntu ubiquity: Please run 'e2fsck -f /dev/sda1' first.
[13:56] <cjwatson> looks like it wasn't cleanly unmounted. perhaps you just turned off the vm?
[13:59] <davmor2> cjwatson: it's hw I rebooted
[14:03] <cjwatson> not sure there's much I can do, resize2fs doesn't like it ... maybe it crashed before unmounting it properly
[14:04] <davmor2> cjwatson:  I've run the e3fsck and started ubiquity again and it's gone through this time.  I'll try it again after the tests are complete to makes sure it's not an issue
[14:05] <davmor2> e2 even
[14:10] <cjwatson> oh my
[14:10] <cjwatson>             if '(%s)' % k.strip('=dev=') in txt:
[14:11] <cjwatson> that SO doesn't do what the author expected
[14:12] <cjwatson> (try it with '=dev=vda' ...)
[14:13] <cjwatson> actually, how is this working at all for anyone? '=dev=sda' will be turned into 'sa'
[14:13] <evand> yikes, that's definitely my doing
[14:14] <soren> >>> '=dev=sda'.strip('=dev=')
[14:14] <soren> 'sda'
[14:14] <cjwatson> hmm
[14:14] <soren> It only really breaks for virtio devices.
[14:14] <cjwatson> oh, it'll only strip contiguous characters
[14:14] <soren> Or people with enough devices to have an "sde"
[14:14] <cjwatson> no, that won't break either
[14:14] <cjwatson> it'll stop at the 's'
[14:15] <cjwatson> oh, but it's strip not lstrip
[14:15] <cjwatson> so yes, you're right
[14:15] <soren> Only from the start and end of the string until it reaches a not-to-be-stripped character.
[14:15] <cjwatson> I ran into this because I decided to start using virtio devices for routine testing
[14:15]  * soren tips his hat at cjwatson for that
[14:16] <soren> There more testing I can get of virtio devices, the better :)
[14:19] <CIA-4> ubiquity: cjwatson * r3219 ubiquity/ (3 files in 2 dirs):
[14:19] <CIA-4> ubiquity: Fix stripping of '=dev=' from device names when creating partition bars
[14:19] <CIA-4> ubiquity: to avoid reducing e.g. '=dev=vda' to 'a'.
[14:28] <davmor2> cjwatson, evand: https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/361674 is interesting.  It only seems to effect french installs.
[14:30] <cjwatson> ok, I take back my criticism of the Kubuntu autopartitioning bar code; it seems to be mostly working now that I've fixed that '=dev='-stripping bug
[14:48] <cjwatson> davmor2: will look, though I have two other things in front of that in the queue
[14:49] <davmor2> cjwatson: ara has found another couple too
[14:49] <davmor2> in d-i
[14:51] <cjwatson> ok, I'm reading #ubuntu-testing backlog
[14:52] <davmor2> cjwatson: I'm going to see if I can confirm them as I'm doing 64bit installs on alternate now
[14:52] <CIA-4> ubiquity: cjwatson * r3220 ubiquity/ (2 files in 2 dirs):
[14:52] <CIA-4> ubiquity: * KDE frontend:
[14:52] <CIA-4> ubiquity:  - Start the resize slider at the maximum possible size for the partition
[14:52] <CIA-4> ubiquity:  being resized; this is in line with the GTK frontend, and fixes the
[14:52] <CIA-4> ubiquity:  situation where the previous default (half-way between the minimum
[14:52] <CIA-4> ubiquity:  size and the current size) is out of range.
[14:57] <cjwatson> davmor2: I don't think ara's bug need confirmation or further information
[14:57] <cjwatson> bugs
[14:57] <cjwatson> the second is probably a dup of an old bug, but I need to wait for my browser to finish rebooting
[15:09] <CIA-4> ubiquity: cjwatson * r3221 ubiquity/debian/changelog: typo
[15:55] <shtylman> cjwatson: anything I should look at (relating to the partition bars)
[15:58] <cjwatson> 14:30 <cjwatson> ok, I take back my criticism of the Kubuntu autopartitioning bar code; it seems to be mostly working now that I've fixed that '=dev='-stripping bug
[15:59] <cjwatson> I think it was just failing because /dev/vda wasn't handled properly, and that's fixed
[15:59] <cjwatson> I made a correction to its handling of the resize bar's initial position
[16:01] <davmor2> shtylman: there is a suggestion for the colour on bug 348461
[16:02] <shtylman> davmor2: thanks...will try to put that in today :)
[16:08] <davmor2> cjwatson: I can confirm using max on auto-resize throws up an error about grub do you want me to add the log to ara's bug?
[16:09] <CIA-4> ubiquity: cjwatson * r3222 ubiquity/ (debian/changelog ubiquity/components/timezone.py):
[16:09] <CIA-4> ubiquity: Preseed tzsetup/selected up-front rather than filtering it and then
[16:09] <CIA-4> ubiquity: preseeding it; filtering it causes the backup state to be reset, which
[16:09] <CIA-4> ubiquity: causes an infinite loop when backing up from the timezone page when the
[16:09] <CIA-4> ubiquity: default country only has one timezone (LP: #361674).
[16:09] <cjwatson> davmor2: no thanks, I'm sure it's just that the default size needs to be increased
[16:10] <davmor2> cjwatson: now are you sure before I reboot?
[16:10] <cjwatson> yes, thanks
[16:10] <CIA-4> oem-config: cjwatson * r658 trunk/ (debian/changelog lib/components/timezone.py):
[16:10] <CIA-4> oem-config: Preseed tzsetup/selected up-front rather than filtering it and then
[16:10] <CIA-4> oem-config: preseeding it; filtering it causes the backup state to be reset, which
[16:10] <CIA-4> oem-config: causes an infinite loop when backing up from the timezone page when the
[16:10] <CIA-4> oem-config: default country only has one timezone (LP: #361674).
[16:10] <davmor2> cool
[16:40] <lool> How is it guaranteed that some items are translated in Ubuntu install mediums?
[16:40] <lool> Say, ubiquity is translated in French, how could I make sure other software is?
[16:40] <lool> Like language-selector
[16:41] <cjwatson> you can't, in general, since it requires language packs
[16:41] <lool> But ubiquity manages to
[16:41] <cjwatson> although I think language-selector is handled outside language packs
[16:41] <cjwatson> ubiquity does not use language packs, for good reason
[16:41] <cjwatson> but that doesn't mean we can do that in general
[16:41] <lool> Is it to make sure that the installer will be translated that ubiquity doesn't use langpacks?
[16:42] <cjwatson> language-selector uses a similar approach to ubiquity; it's excluded from translation stripping (in pkgbinarymangler) and we update the translations in there by hand every so often
[16:42] <cjwatson> it's because the installer displays text before it has a chance to install language packs ...
[16:42] <cjwatson> so if we didn't exclude the installer from language packs, we'd have a chicken-and-egg problem
[16:42] <lool> Exactly; in my case, after install, I was told that language data was incomplete
[16:42] <lool> With half of the strings in English
[16:43] <lool> And I'd like to blame it on $foo
[16:43] <cjwatson> can I see the syslog?
[16:43] <cjwatson> sometimes that happens because networking was unavailable; sometimes it's a bug
[16:44] <lool> networking wasn't available
[16:44] <lool> Because I did the install fullscreen, outside of a live session
[16:44] <lool> So didn't get the NM icon
[16:44] <cjwatson> right, can't do much about that right now although there is a bug open
[16:45] <lool> As long as it's a known limitation, I'm ok
[16:45] <lool> I had in mind that stuff like NM should be excluded as well
[16:45] <lool> Because you use it to grab langpacks
[16:45] <lool> But it quickly opens the door to a lot of things to have translated on the CD   :-/
[16:45] <cjwatson> the amount of NM text you actually need to bring up networking isn't very much really
[16:46] <cjwatson> and we can't ship translations for the whole desktop into all languages, sadly
[16:46] <cjwatson> I think there's a bug open, anyway. But at any rate it is known
[16:47] <lool> cjwatson: Yes, I think we agree strongly; it would be feasible, but it's hard to draw the final line on what we really need to include and I know we can't include all
[16:47] <lool> I guess you don't care about the syslog anymore
[16:49] <cjwatson> lool: nope, thanks
[17:37] <CIA-4> partman-auto: cjwatson * r288 ubuntu/ (2 files in 2 dirs):
[17:37] <CIA-4> partman-auto: Add another 300MB to the amount of space we require to be left by
[17:37] <CIA-4> partman-auto: automatic resizing, to account for (a) growth in installation size, (b)
[17:37] <CIA-4> partman-auto: variable language pack sizes, and (c) increased memory sizes resulting
[17:37] <CIA-4> partman-auto: in upward pressure on swap partition sizes (LP: #361707).
[18:00] <mdz> I just noticed that ubiquity prompts for the OEM ID before the keyboard is configured.  is this a problem?
[18:04] <cjwatson> mdz: yes, albeit a minor one since I expect most OEMs can cope with a US keyboard (or whatever the default for the language they selected at the boot loader is). Bug 290421, which I notice you filed ;-)
[18:05] <mdz> cjwatson: ah, the memories
[18:05] <cjwatson> it should probably go on the user setup page, but there was all that tempting free space on the language page ...
[18:05] <mdz> way back in intrepid RC
[18:06] <cjwatson> we've had trouble with the user setup page; its size has been one of the things impeding support for small screens
[18:06] <cjwatson> evand stuck a scrollbar in as a nasty workaround
[18:06]  * cjwatson -> dinner
[21:21] <lfaraone> Shouldn't usb-creator not display partitions that are marked with volume.ignore?