[09:17] <CIA-31> ubiquity: cjwatson * r4914 trunk/bin/ (oem-config-remove-gtk ubiquity ubiquity-dm): remove trailing whitespace
[09:22] <CIA-31> ubiquity: cjwatson * r4915 trunk/ (bin/ubiquity-dm debian/changelog): /apps/metacity/general/num_workspaces should be an int, not a boolean.
[11:15] <CIA-31> ubiquity: evand * r4916 trunk/ (debian/changelog ubiquity/plugins/ubi-language.py):
[11:15] <CIA-31> ubiquity: Don't try to connect signals on the release notes label in the oem-
[11:15] <CIA-31> ubiquity: config version of the language page, as it doesn't exist there.
[14:03] <ev> been trying to fix the iconview wrapping issue today
[14:04] <ev> while building a test harness for ubi-language at the same time
[14:04] <ev> but the former is proving quite annoying
[14:04] <cjwatson> I'll be attacking bug 837681 once I get a current image sucked down
[14:04] <ubot2`> Launchpad bug 837681 in ubiquity "Automatic partitioning corrupts GUID partition table (GPT)" [Critical,New] https://launchpad.net/bugs/837681
[14:06] <cjwatson> I fully expect it to do my head in
[14:13] <stgraber> ev: can you look at http://paste.ubuntu.com/678921/ as a fix for bug 770320? I actually don't see any country in the current list having that problem. (Ethiopia is gone in Oneiric apparently)
[14:13] <ubot2`> Launchpad bug 770320 in ubiquity "Sorting non-English list: French installer sorts accented names at end of alphabet" [Low,Triaged] https://launchpad.net/bugs/770320
[14:14] <stgraber> I noticed that you used another locale aware sorting method for the initial language step, though I couldn't find a way to easily make it work for the keyboard
[14:14] <stgraber> also, I'm not sure if using os.getenv('LANG','en_US.UTF-8') will get me the current locale (as in the one selected in ubi-language.py) or if I should use some other way of retrieving it
[14:16] <ev> I'll have a look after this iconview stuff
[14:16] <cjwatson> debian-installer/locale
[14:16] <cjwatson> or debian-installer/language if you only want the language code
[14:27] <stgraber> Just noticed that this step already has a get_locale() function that seems to do the right thing. http://paste.ubuntu.com/678928/
[15:05] <ev> finally: http://paste.ubuntu.com/678954/ now to just figure out how to test for the text wrapping
[15:11] <CIA-31> ubiquity: evand * r4917 trunk/ (debian/changelog ubiquity/plugins/ubi-migrationassistant.py):
[15:11] <CIA-31> ubiquity: Construct a GtkTreeCellDataFunc with the correct number of arguments
[15:11] <CIA-31> ubiquity: in the migration assistant plugin (LP: #829987).
[15:12] <jibel> ev, on the timezone screen, does ubiquity tries to download something from the net ?
[15:13] <ev> yes
[15:13] <ev> lots, actually
[15:13] <ev> it hits geoip.ubuntu.com to find your current location if it can
[15:13] <ev> it also his geoname-lookup.ubuntu.com every time you type in the box (after sufficient time has past)
[15:14] <jibel> ok, that would explain why it hangs there sometimes. Could the call be non-blocking ?
[15:15] <jibel> when the connection is firewalled and these addresses are not authorized, ubiquity completely freezes
[15:44] <ev> jibel: please file a bug for that
[15:44] <ev> if you could provide a test case as well, that would be ace
[15:44] <skaet> cjwatson,  been able to poke at 837681 yet?   or is image still downloading?
[15:45] <cjwatson> I'm working on constructing a test case
[15:45] <cjwatson> since it involves a double installation and faking the architecture, and since compiz hates me, and since I've had meetings, it's taking a little while
[15:46] <jibel> ev, bug 837217
[15:46] <ubot2`> Launchpad bug 837217 in ubiquity "Time Zone selector hangs and ubiquity freezes if network connection is firewalled" [Medium,New] https://launchpad.net/bugs/837217
[15:47] <skaet> cjwatson,  gotcha.   Just trying to assess scope of impact and whether we can get away with release noting it, or its going to be another late night. :/
[15:49] <cjwatson> I don't know
[15:49] <ev> thanks jibel
[16:03] <jibel> on a netboot install what the task 'Ubuntu desktop USB' is supposed to install ?
[16:03] <jibel> it installs something, but for sure it is not a desktop.
[16:03] <cjwatson> it should be equivalent to what's on the USB images (well, currently called DVD)
[16:04] <cjwatson> bug on tasksel please?  I suspect it's only installing the extra packages and not ubuntu-desktop as well
[16:10] <bdmurray> ev: wrt bug 837287 should there be a panel or not?
[16:10] <ubot2`> Launchpad bug 837287 in ubiquity "live cd panel / top menu is cropped" [High,Fix released] https://launchpad.net/bugs/837287
[16:13] <ev> bdmurray: there should be a panel in ubiquity 2.7.24
[16:14] <superm1> ev, seems that there is still a problem in 2.7.24 :(
[16:14] <ev> rubbish
[16:14] <superm1> it's loading those indicators now finally but it doesn't feel like it's important enough to draw em'
[16:15] <bdmurray> so 837287 shouldn't really be fixed then
[16:16] <superm1> no not yet
[16:17] <CIA-31> ubiquity: evand * r4918 trunk/ (debian/changelog ubiquity/plugins/ubi-language.py): Don't let the text wrap for the language names in oem-config.
[16:17] <ev> I have a test case for that in progress
[16:39] <bdmurray> I noticed the "OEM Configuration (temporary user)" seemed longer than lightdm(?) could handle when doing an oem install the other day
[16:41] <cjwatson> lightdm should be fixed then, I'm pretty sure there are real people with names that long
[16:42] <cjwatson> I saw several such just reviewing UDS sponsorship applications today
[16:42] <cjwatson> (particularly those from Spanish-speaking countries)
[17:28] <cjwatson> ev: if you have ubiquity stuff to upload for b1, don't wait for 837681, since I think it's probably not in ubiquity
[17:31] <skaet> ev, would be good to get the migration-assistant bug fix included in any respins.
[17:51]  * cjwatson -> dinner
[17:52] <cjwatson> ev: please let me know when you're around and if you intend to do more ubiquity work tonight; otherwise I intend to upload ubiquity from trunk after dinner
[17:55] <stgraber> cjwatson: I'll push my fix for bug 770320 then
[17:55] <ubot2`> Launchpad bug 770320 in ubiquity "Sorting non-English list: French installer sorts accented names at end of alphabet" [Low,Triaged] https://launchpad.net/bugs/770320
[17:55] <stgraber> (very far from critical, just wanted something easy to do this morning ;))
[18:37] <stgraber> cjwatson, ev: Actually, I have a question for one of you two before I push the fix for bug 770320. I'm a bit lost in ubi-language, trying to figure out if it's supposed to call locale-gen or not.
[18:37] <ubot2`> Launchpad bug 770320 in ubiquity "Sorting non-English list: French installer sorts accented names at end of alphabet" [Low,Triaged] https://launchpad.net/bugs/770320
[18:38] <stgraber> My current fix uses locale.setlocale(locale.LC_ALL, '') which works fine as long as the locale exists, which isn't always the case
[18:38] <stgraber> casper does a locale-gen for the locale selected in gfxboot so that covers a lot of cases
[18:38] <stgraber> but when you then select another locale in ubiquity, I'd expect it to also call locale-gen, which localechooser is apparently supposed to do but in my case doesn't seem to happen
[18:39] <stgraber> (and so ubiquity explodes as I'm trying to do a setlocale() on a non-existing locale)
[18:40] <stgraber> so I'm wondering if it's a bug that locale-gen isn't called when exitting ubi-language or if that code is only supposed to be called much later (and so will make other plugins depending on the locale fail)
[19:22] <cjwatson> stgraber: no, it's not supposed to be called, it takes too long and uses too much memory; the installer in general doesn't need a generated locale
[19:22] <cjwatson> e.g.
[19:22] <cjwatson>   * Set LANGUAGE as well as LANG when changing locale; this allows us to get
[19:22] <cjwatson>     correct debconf translations without needing to run locale-gen (closes:
[19:22] <cjwatson>     Malone #43071).
[19:22] <ubot2`> Launchpad bug 43071 in ubiquity "Ubiquity in Kubuntu gets killed by out-of-memory with 256 MB RAM" [High,Fix released] https://launchpad.net/bugs/43071
[19:23] <cjwatson> I'm afraid you can't use collation ordering from the locale
[19:23] <cjwatson> or at least not reliably; I suppose it's OK to use it if the locale is available anyway
[19:25] <stgraber> cjwatson: http://paste.ubuntu.com/679160/
[19:26] <stgraber> that's a lot of fallback to implement to get the best possible locale...
[19:26] <stgraber> (even if we could have run locale-gen in ubi-language, the _locale module seems to keep a cache of what's valid, so we'd still need to restart ubiquity ...)
[19:27] <cjwatson> yeah, that's a non-starter
[19:27] <cjwatson> (restarting ubiquity)
[19:28] <cjwatson> sorting isn't a critical function, I think it's OK to have as much or as little in the way of fallback as you think worthwhile
[19:28] <cjwatson> why choices.sort in one branch and choices=sorted() in the other?
[19:28] <stgraber> I'm just going to run ubi-console-setup one more time with a weird language just to make sure it works, but I think the current code should work fine (I'll probably move it to i18n later so we can use it at some other places)
[19:29] <cjwatson> (also, PEP-8 would be "choices = sorted(choices)")
[19:29] <stgraber> yeah, I just noticed that one, changed it to choices.sort() everywhere now
[19:31] <cjwatson> hm, another thing, the effect of this is to add an .encode('utf-8') to each choice - is that absolutely guaranteed safe?
[19:31] <cjwatson> strikes me as potentially worrisome
[19:31] <cjwatson> I wonder if it would be safer to encode in the sort key function
[19:32] <cjwatson> so choices.sort(key=lambda c: locale.strxfrm(c.encode('utf-8'))) # or similar
[19:40] <stgraber> cjwatson: ok, tested with yet another language, seems to work fine. Pushed to lp:ubiquity
[19:45] <stgraber> cjwatson: I'm now starting to look at beta2 bugs for ubiquity, so I'm not expecting anything else to include before the next upload.
[19:45] <cjwatson> ok
[19:45] <cjwatson> I'll go ahead and upload this lot the
[19:45] <cjwatson> n
[19:46] <cjwatson> still working on the suspected parted bug :-/
[19:52] <CIA-31> ubiquity: cjwatson * r4920 trunk/ (d-i/manifest debian/changelog): Automatic update of included source packages: flash-kernel 2.28ubuntu29.
[19:58] <CIA-31> ubiquity: cjwatson * r4921 trunk/debian/changelog: releasing version 2.7.25
[20:01] <superm1> cjwatson, re all GPT installs being busted or not, i've managed to do an EFI install that re-used an existing ESP at the start of the disk with the 08/31 daily
[20:04] <cjwatson> yeah, I'd got far enough to conclude it wasn't all of them
[20:04] <cjwatson> big gap between all and one though :)
[20:05] <cjwatson> it stinks of a bug in parted's built-in gptsync, and I'll need to do most of the work of tracking it down before I can make a decent guess at the impact
[20:05] <cjwatson> my first test was non-EFI with the bios_grub partition at the start of the disk, and that seemed to partition fine
=== kentb is now known as kentb-afk
[22:39] <stgraber> the Kubuntu manual partitioning issue is: http://paste.ubuntu.com/679259
[22:39] <stgraber> will poke at it in a bit