[08:44] <jibel> stgraber, the segfault is not caused by libc6. the crash yesterday was caused by /var/lock being a file instead of a directory for some reason. Just bad coincidence. restarting testing from scratch :/
[09:16] <CIA-32> debian-installer: cjwatson * r1646 ubuntu/ (build/config/armel/omap4.cfg debian/changelog): Move armel/omap4 and armhf/omap4 to 3.2.0-1408 kernels.
[09:19] <CIA-32> debian-installer: cjwatson * r1647 ubuntu/ (29 files in 2 dirs): Update help text translations from Launchpad.
[09:23] <CIA-32> debian-installer: cjwatson * r1648 ubuntu/debian/changelog: releasing version 20101020ubuntu118
[10:53]  * cjwatson starts working on various hairy GPT-related bugs
[11:07] <ogra_> cjwatson, do you have any objection if i add a few lines to stgraber's screensaver disabling patch that also supresses suspend on lid close ?
[11:07] <ogra_> (just one more gsettings key to unset/set)
[11:09] <cjwatson> I don't think so
[11:10] <cjwatson> maybe add a new method for it
[11:10] <cjwatson> (there's already {disable,enable}_{volume_manager,screensaver})
[11:24]  * ogra_ will look into it, seems to be pretty fatal if you suspend while doing partitioning/resize operations
[11:31] <cjwatson> yeah, makes sense
[11:32] <cjwatson> though I wouldn't say fatal necessarily, but certainly risky
[11:32] <cjwatson> (presumably traded against the risk of hardware damage from not suspending on lid close ...)
[12:37] <CIA-32> partman-efi: cjwatson * r663 ubuntu/ (74 files in 4 dirs):
[12:37] <CIA-32> partman-efi: On x86 architectures, create EFI system partitions using FAT32 rather
[12:37] <CIA-32> partman-efi: than FAT16, and require newly-created ones to have a minimum size of
[12:37] <CIA-32> partman-efi: 34091008 bytes, experimentally verified as the minimum libparted will
[12:37] <CIA-32> partman-efi: accept (LP: #811485).
[12:38] <CIA-32> partman-auto: cjwatson * r625 ubuntu/ (4 files in 2 dirs): Increase minimum size of EFI System Partitions to 100MB (LP: #811485).
[12:39] <CIA-32> partman-auto: cjwatson * r626 ubuntu/debian/changelog: releasing version 93ubuntu21
[12:58] <CIA-32> partman-efi: cjwatson * r664 ubuntu/ (commit.d/format_efi debian/changelog):
[12:58] <CIA-32> partman-efi: Never format EFI system partitions that already contain a filesystem
[12:58] <CIA-32> partman-efi: (LP: #769669).
[13:47] <CIA-32> partman-efi: cjwatson * r665 ubuntu/choose_method/efi/do_option: don't even claim we're going to format EFI system partitions
[14:22] <ogra_> stgraber, cjwatson... http://paste.ubuntu.com/883285/ does that look sane (i picked the sledgehammer and disabled power mgmt completely instead of disabling 20 different PM actions which would have gotten really complex)
[14:23] <stgraber> ogra_: you probably should call disable_powermgr from somewhere
[14:23] <ogra_> lol, indeed
[14:24] <stgraber> but yeah, looks good :)
[14:26] <ogra_> http://paste.ubuntu.com/883298/ and fixed :)
[14:26]  * ogra_ wonders if his CIA setup is still correct
[14:26] <stgraber> ogra_: I think it's technically gnome-settings-daemon power management plugin (I think they killed g-p-m with gnome3)
[14:27] <ogra_> you mean for the changelog ?
[14:27] <stgraber> yeah
[14:27]  * ogra_ changes 
[14:30] <CIA-32> partman-auto: cjwatson * r627 ubuntu/ (debian/changelog lib/auto-shared.sh):
[14:30] <CIA-32> partman-auto: Add a hack to stop EFI System Partitions showing up as to-be-formatted
[14:30] <CIA-32> partman-auto: in the confirm-changes screen when there's an existing filesystem that
[14:30] <CIA-32> partman-auto: would cause partman-efi to skip them.
[14:32] <ogra_> bah, sigh i need to pul a new version of the ubiquity tree, seeems mine is to outdated
[14:33] <CIA-32> partman-efi: cjwatson * r666 ubuntu/debian/changelog: releasing version 24ubuntu3
[14:36] <CIA-32> partman-auto: cjwatson * r628 ubuntu/debian/changelog: releasing version 93ubuntu22
[16:02] <ogra_> bdmurray, cjwatson, why is bug 948163 casper now ? (teh fix is committed in ubiquity already)
[16:02] <ubot2`> Launchpad bug 948163 in casper "System suspends upon closing the lid while installing/partitioning using a live CD" [Wishlist,New] https://launchpad.net/bugs/948163
[16:03] <cjwatson> I didn't make that change
[16:03] <cjwatson> all I did was assign it to you since you were working on it
[16:03] <ogra_> ah, yeah- you just assigned it
[16:03]  * ogra_ changes it back before setting fix comitted
[16:03] <cjwatson> that said the reporter does explicitly ask for suspend to be disabled while running gparted on the live CD too
[16:04] <ogra_> hmm
[16:05]  * ogra_ missed that comment, you are right... but that should go into gparted, we shouldnt completely turn off PM in the live session 
[16:05] <ogra_> since i know some poeple tend to use that as actual desktop session
[16:05] <cjwatson> inclined to agree
[16:06] <stgraber> +1
[16:07] <ogra_> opened a task for it
[16:44] <stgraber> ev: around?
[16:45] <stgraber> ev: I just saw jibel_ report bug 955232
[16:45] <ubot2`> Launchpad bug 955232 in ubiquity "Switch 'prepare' and 'wifi' steps" [Undecided,New] https://launchpad.net/bugs/955232
[16:45] <stgraber> ev: I'm guessing it's been done "by design" but indeed doesn't make a lot of sense to have wifi after prepare, should we just switch them (assuming we can get a UIFe) or was there a good reason to have them in that order?
[16:46] <stgraber> mpt: ^
[16:48] <mpt> too ... many ... interruptions ...
[16:51] <mpt> stgraber, in the spec <https://docs.google.com/a/canonical.com/Doc?docid=0AU5sFuLRpCpBZGZra2pqY2pfMTAxZ25rcnBnNXY#_Preparing_to_install_Ubuntu_5_5289094087202102> I don't see any explanation of why "Preparing" was before "Connect to the Internet"
[16:51] <mpt> stgraber, oh, I think I know why. If you choose not to install updates, then there's little point in setting up an Internet connection then.
[16:52] <mpt> But there's nothing to say that the step *shouldn't* show up if you unchecked the checkbox
[16:53] <mpt> ok, I was mistaken
[16:53] <mpt> I just asked ev
[16:54] <mpt> He explained that if you have Broadcom wireless, the Broadcom driver needs to be downloaded and installed before you can set up a wi-fi connection
[16:54] <mpt> and that's part of the "Install this third-party software" checkbox
[16:54] <mpt> which is part of the "Preparing" step
[16:55] <mpt> which, therefore, goes before the "Connect to the Internet" step.
[16:55] <stgraber> right but then the user can't also install the updates
[16:55] <stgraber> because the update checkbox is disabled until you're connected to the internet
[16:56] <stgraber> and so requires the user to do: prepare -> wireless -> back to prepare -> tick the checkbox -> wireless -> continue
[16:56] <mpt> Sorry, I was mistaken
[16:56] <mpt> The driver is already on the CD, but you need to check the checkbox to install it and then use it
[16:57] <mpt> so you don't need to download it
[16:57] <mpt> but that doesn't solve the problem you describe
[16:58] <stgraber> I guess ubi-wireless should be moved before ubi-prepare and if it detects it needs a driver, then start jockey or something to install it
[16:59] <mpt> so one solution would be for "Download updates while installing" to be sensitive if *either* (you have an Internet connection) *or* (you have Broadcom *and* you checked "install this third-party software")
[17:00] <stgraber> well, broadcom + third-party software != internet connectivity
[17:00] <stgraber> so ubiquity may well fail later on if we go with that assumption
[17:01] <superm1> stgraber: but what if they don't want non-free drivers?
[17:03] <stgraber> I think it'd be sane to:
[17:03] <stgraber>  - show ubi-wireless before ubi-prepare
[17:03] <stgraber>  - if third-party driver is required => prompt and run jockey, if that was needed, automatically tick the checkbox in ubi-prepare (optional)
[17:03] <stgraber>  - if not required, don't do anything (obviously)
[17:03] <mpt> stgraber, the reverse is also true -- it's no use showing the wi-fi details before you've chosen to install the software that would make them work.
[17:03] <stgraber>  - then run ubi-prepare as usual after ubi-wireless
[17:04] <stgraber> mpt: depends, someone may want ot have wifi for the install but not post-install
[17:04] <stgraber> mpt: in that case, showing ubi-wireless and prompting to run jockey, yet not ticking the checkbox in ubi-prepare would achieve that
[17:05] <mpt> stgraber, we choose not to offer that option (wi-fi for install but not post-install), the ratio of usefulness to confusion isn't high enough
[17:05] <mpt> If you don't want the wi-fi post-install you can always delete that network
[17:06] <mpt> The wi-fi setup in the installer is used for three things
[17:06] <mpt> (1) downloading updates, if you chose to do that
[17:06] <mpt> (2) detecting your location for your time zone
[17:07] <mpt> (3) using the Internet post-install.
[17:07] <stgraber> right and we only achieve 3) currently
[17:07] <stgraber> both the others depend on ubi-prepare detecting internet connectivity
[17:07] <mpt> right
[17:08] <cjwatson> also language packs
[17:10] <stgraber> I guess we could duplicate the check and the checkbox to ubi-wireless but that sounds rather hackish...
[17:11] <mpt> (This is not the only catch-22 in the installer: there's a similar problem where typing a location requires having a roughly correct keyboard layout, but we use the location to guess your keyboard layout.)
[17:17] <stgraber> at least you can click the map ;)
[17:17] <stgraber> mpt: so should we defer that bug to 12.10?
[17:24] <mpt> stgraber, short answer, yes. Long answer I posted in 955232.
[19:32] <CIA-32> ubiquity: stgraber * r5256 ubiquity/ (debian/changelog ubiquity/plugins/ubi-partman.py): Switch from ListStore.append to ListStore.insert_with_valuesv to workaround gtk's non-atomic appends
[19:33] <stgraber> ^ doing some stress testing of the manual partitioner for that one, but seems to work fine so far
[19:34] <stgraber> ok, looks like it needs a bit more patching, it's not the only part of the code failing for that bug...
[20:01] <CIA-32> ubiquity: stgraber * r5257 ubiquity/ubiquity/plugins/ubi-partman.py: Port all remaining uses of .append() to .insert_with_valuesv() but they will all appear as empty strings now because of bug https://bugzilla.gnome.org/show_bug.cgi?id=672065
[20:01] <stgraber> workaround one gtk/pygi bug and find another...
[20:01] <ubot2`> Gnome bug 672065 in introspection "string GValues in flat arrays are always empty" [Normal,New]
[20:13] <CIA-32> ubiquity: stgraber * r5258 ubiquity/debian/changelog: Add warning to changelog so anyone who needs to upload ubiquity before Gnome:672065 is fixed, reverts commit 5256 and 5257
[20:14] <stgraber> jibel_: any luck tracking down bug 950282?
[20:14] <ubot2`> Launchpad bug 950282 in oem-priority/precise "Installation failing with pop-up "The installer encountered an unrecoverable error and will now reboot."" [Critical,Confirmed] https://launchpad.net/bugs/950282
[20:29] <jibel> stgraber, I can't reproduce with today's build. I also upgraded everything from b1 to today and no segfault. So I don't know and I don't like self-fixing bugs.
[20:35] <stgraber> jibel: yeah, I don't like that much either... can you post a comment in the bug report asking for the initial reporter to see if he can maybe still reproduce with today's build?