[01:25] <CIA-32> user-setup: cjwatson * r264 ubuntu/ (debian/changelog user-setup-apply): Stop copying /etc/crypttab if OVERRIDE_ALREADY_ENCRYPTED_SWAP is set; ubiquity will take care of this.
[01:28] <CIA-32> user-setup: cjwatson * r265 ubuntu/ (debian/changelog user-setup-apply):
[01:28] <CIA-32> user-setup: Revert the change from 1.42ubuntu2. We'll handle this in ubiquity
[01:28] <CIA-32> user-setup: instead; in d-i, it seems to be too difficult to get right without the
[01:28] <CIA-32> user-setup: aid of script libraries from cryptsetup.
[01:48] <CIA-32> user-setup: cjwatson * r266 ubuntu/debian/changelog: releasing version 1.42ubuntu3
[02:04] <CIA-32> ubiquity: cjwatson * r5386 trunk/ (5 files in 4 dirs):
[02:04] <CIA-32> ubiquity: If encrypting the user's home directory, wipe and enable encrypted swap
[02:04] <CIA-32> ubiquity: before starting the slideshow, because running the slideshow can use up
[02:04] <CIA-32> ubiquity: a lot of memory so we won't be able to disable swap once that's in
[02:04] <CIA-32> ubiquity: progress (LP: #979350).
[02:13] <CIA-32> ubiquity: cjwatson * r5387 trunk/ (d-i/manifest debian/changelog):
[02:13] <CIA-32> ubiquity: Automatic update of included source packages: apt-setup 1:0.55ubuntu3,
[02:13] <CIA-32> ubiquity: partman-base 153ubuntu4, user-setup 1.42ubuntu3.
[02:27] <CIA-32> ubiquity: cjwatson * r5388 trunk/ (debian/changelog scripts/plugininstall.py):
[02:27] <CIA-32> ubiquity: If /etc/network/interfaces doesn't exist when configuring the network in
[02:27] <CIA-32> ubiquity: oem-config, write out a skeleton one (LP: #985305).
[02:39] <CIA-32> ubiquity: cjwatson * r5389 trunk/debian/changelog: releasing version 2.10.14
[03:11] <infinity> cjwatson: Say, short of cargo-culting the LP task generation scripts and reducing them for a minimal use-case, do you have a simple recipe for generating Packages files with task sanity?
[03:12] <infinity> cjwatson: I assume I just need to grab the overrides and reference them magically in my apt-ftparchive config.
[03:12] <infinity> Oh, but I'm not using an apt.conf.  So, I'd need to fix that. :P
=== mpt_ is now known as mpt
[09:11] <FourDollars> cjwatson: I have some new found about https://bugs.launchpad.net/ubuntu/+source/partman-efi/+bug/972122 .
[09:11] <ubot2> Launchpad bug 972122 in partman-efi "Ubiquity pops out a warning message from partman-partitioning on UEFI BIOS when there is a EFI system partition existed." [Medium,Fix released]
[09:12] <FourDollars> cjwatson: For https://bugs.launchpad.net/ubuntu/+source/partman-efi/+bug/972122, it does fixed manually partitioning.
[09:12] <FourDollars> cjwatson: But how about if we provide the preseed.cfg to let it do partitioning automatically.
[09:13] <FourDollars> cjwatson: It will trigger /bin/autopartition, but the previous case will not.
[09:15] <FourDollars> cjwatson: If we provide a preseed.cfg to let ubiquity do partitioning automatically on a disk with EFI system partition already existed, that will trigger /bin/autopartition and make installation fail.
[09:16] <FourDollars> cjwatson: That is what I infer.
[10:10] <ogra_> cjwatson, is there any reason why d-i doesnt dump its version somewhere into syslog ? its hard to say what version a user used if he used the "current" link on the download page
[10:13]  * ogra_ doesnt get why the system in Bug #985520  doesnt boot 
[10:13] <ubot2> Launchpad bug 985520 in debian-installer "pandaboard ES does not boot after using armhf netinstall" [Undecided,New] https://launchpad.net/bugs/985520
[10:22] <cjwatson> infinity: I think you can grab extra override files out of ubuntu/indices/; I don't have a recipe
[10:22] <cjwatson> ogra_: the version of d-i itself (i.e. the build system) is often not desperately helpful
[10:23] <ogra_> well, i would like to know which version the user downloaded
[10:23] <cjwatson> there are lots of more relevant versions dumped into syslog, normally
[10:23] <cjwatson> argh I wish FourDollars would just answer the questions I asked
[10:25] <stgraber> cjwatson: we just got bug 985526 reported on the tracker but without anything useful in the logs that I could find. I'll do a test install with the same media to check
[10:25] <ubot2> Launchpad bug 985526 in ubiquity "exit with error if encryption is selected" [Undecided,New] https://launchpad.net/bugs/985526
[10:27] <cjwatson> stgraber: erk
[10:29] <jibel> I tried to reproduce and didn't get the error.
[10:29] <jibel> but there is now a long delay between the moment the user clicks on 'continue' in user setup and the moment the slideshow starts
[10:31] <stgraber> jibel: yeah, that part is expected, it's cjwatson's fix to the OOM problem
[10:31] <stgraber> jibel: it's basically disabling your current swap, wiping it, encrypting it and re-enabling it before starting the slideshow
[10:32] <jibel> stgraber, I know, I'm just pointing that there is no feedback, and from a user view it's like if nothing happened when he clicked on continue
[10:32] <cjwatson> Yes, there is a long delay with no feedback.  I'm afraid you'll have to cope with that.
[10:33] <cjwatson> There was no safe way that I could see to add progress reporting there.
[10:33] <cjwatson> Or I can revert the whole thing back such that it fails for lots of people if you'd prefer (but I wouldn't prefer that).
[10:35] <cjwatson> Also, it was 4am by the time I finished this as it was.  Refinements weren't happening.
[10:37] <stgraber> it's installing fine here, also booting into ubiquity-dm, also installing in Italian
[10:38] <cjwatson> hm, so I wonder what went wrong for Fabio
[10:39] <cjwatson> some weird race?  but why a crash with no error output ...
[10:40] <cjwatson> oh
[10:40] <cjwatson> Apr 19 09:50:26 ubuntu ubiquity: cp: cannot stat `/target/etc/fstab'
[10:40] <cjwatson> Apr 19 09:50:26 ubuntu ubiquity: : No such file or directory
[10:40] <stgraber> gah, you were faster again, I was just about to copy/paste that one ;)
[10:41] <stgraber> wondering if it's related to:
[10:41] <stgraber> Apr 19 09:49:43 ubuntu ubiquity: umount: /mnt/tmpmount: device is busy.
[10:41] <stgraber> Apr 19 09:49:43 ubuntu ubiquity:         (In some cases useful info about processes that use
[10:41] <stgraber> Apr 19 09:49:43 ubuntu ubiquity:          the device is found by lsof(8) or fuser(1))
[10:41] <stgraber> Apr 19 09:49:43 ubuntu ubiquity: rmdir: failed to remove `/mnt/tmpmount'
[10:41] <stgraber> Apr 19 09:49:43 ubuntu ubiquity: : Device or resource busy
[10:41] <cjwatson> maybe, but I think not, I think there's a race between partman_commit and usersetup
[10:42] <stgraber> hmm, no, doesn't look like it's. The umount failure is likely on sda3 and he's installing on sdb
[10:43] <cjwatson> unless that failure caused partman_commit to bail out entirely
[10:43]  * cjwatson hates clear_partitions
[10:44] <cjwatson> that's bug 946663
[10:44] <ubot2> Launchpad bug 946663 in ubiquity "Installer stuck at "Removing conflicting operating system files..."" [High,Triaged] https://launchpad.net/bugs/946663
[10:46] <cjwatson> stgraber: it could well be that failure
[10:46] <cjwatson> I can add a workaround
[10:48] <stgraber> I'm wondering what's keeping a fd open in /mnt/tmpmount... did we ever got someone to run lsof right after that?
[10:48] <cjwatson> yes, it's too racy to catch
[10:48] <CIA-32> partman-target: cjwatson * r929 ubuntu/ (debian/changelog finish.d/clear_partitions):
[10:48] <CIA-32> partman-target: Exit 0 at the end of finish.d/clear_partitions, so that partman-commit
[10:48] <CIA-32> partman-target: will carry on even if the final rmdir fails (which is a bug in itself,
[10:48] <CIA-32> partman-target: but shouldn't blow away the whole install; LP: #985526).
[10:49] <cjwatson> almost tempted to add a stupid umount/sleep/umount/sleep/... loop hack
[10:49] <cjwatson> but WTF
[10:49] <stgraber> didn't we move to "umount -l" at some point?
[10:49] <stgraber> I remember it being mentioned during the installer sprint at least
[10:50] <cjwatson> in some places
[10:50] <cjwatson> we can try it I guess; I'm just uncomfortable with having no idea what the true problem is
[10:51] <stgraber> I think -l would be best for 12.04 and in 12.10 we probably should make that a umount || lsof -n && umount -l or something
[10:51] <stgraber> so hopefully we can catch what's keeping an open fd
[10:54] <CIA-32> partman-target: cjwatson * r930 ubuntu/ (debian/changelog finish.d/clear_partitions):
[10:54] <CIA-32> partman-target: Use 'umount -l' in finish.d/clear_partitions, to work around an as yet
[10:54] <CIA-32> partman-target: unidentified race that keeps the filesystem busy (LP: #946663).
[10:54]  * ogra_ votes for -l too ... better than a loop
[10:54] <ogra_> but also a hardcoded lsof call that dumps into the log ;)
[10:56] <CIA-32> partman-target: cjwatson * r931 ubuntu/debian/changelog: releasing version 77ubuntu2
[11:10] <cjwatson> gema: bug 985491 analysed, but do you think we can encourage more productive behaviour from translation QA here?
[11:10] <ubot2> Launchpad bug 985491 in ubiquity "No locale-dependent substitution of ${TIME}; install_progress_text has no translation support" [Medium,Triaged] https://launchpad.net/bugs/985491
[11:12] <gema> cjwatson: victor started in our team this week, kate insisted that she wanted chinese testing as soon as possible, sorry about the bug not being split into different issues
[11:13] <gema> cjwatson: right now, for precise, what do you want me to do, do you want me to ask him to split it?
[11:14] <gema> cjwatson: reading your bug, I will help him split it
[11:14] <gema> your comment, I mean
[11:16] <cjwatson> yep, the installer bugs at least aren't RC anyway
[11:16] <cjwatson> that's no problem, I just want to instil good practices early for new starters as it'll save time later :)
[11:16] <cjwatson> we have too many grab-bag "translation issues" bugs that never get closed because they're multiple problems
[11:18] <gema> cjwatson: ack
[11:20]  * cjwatson tries to reproduce bug 985368
[11:21] <ubot2> Launchpad bug 985368 in ubiquity "ubiquity crashed with AttributeError in on_link_clicked(): 'NoneType' object has no attribute 'split'" [Undecided,New] https://launchpad.net/bugs/985368
[11:30] <CIA-32> ubiquity: cjwatson * r5390 trunk/ (debian/changelog ubiquity/plugins/ubi-language.py):
[11:30] <CIA-32> ubiquity: Fix crash when attempting to update the installer and a language isn't
[11:30] <CIA-32> ubiquity: selected for some reason (LP: #985368).
[11:38] <stgraber> cjwatson: I'm really not too happy about my last gsettings change (running in dbus-launch), we end up with ~10-15 dbus-daemon + dbus-launch running at the end of the install
[11:38] <stgraber> each of them with their own dconf-writer running, even though that "works" (AFAICT), I feel like I need to clean that up a little
[11:39] <cjwatson> where was that change?
[11:40] <stgraber> I think the root of the gsettings issue is that dconf-writer doesn't get killed properly and so doesn't dump its changes to disk, instead I think we should try to spawn it from ubiquity-dm (maybe inside a dbus-launch call if that's what it needs) and kill that when ubiquity-dm exits
[11:40] <stgraber> cjwatson: r5356 in ubiquity/gsettings.py
[11:40] <stgraber> essentially a workaround for bug 960096
[11:41] <ubot2> Launchpad bug 960096 in libxklavier "Live session started with wrong layout" [Medium,Confirmed] https://launchpad.net/bugs/960096
[11:42] <cjwatson> do you think it's RC?  I agree it's not particularly elegant, but ...
[11:43] <cjwatson> maybe a branch staged for Q
[11:43] <stgraber> yeah, I haven't seen any report of it blowing up, though I can't see how 10 dconf-writer running and writing to the same binary DB could be a good thing
[11:44] <CIA-32> ubiquity: cjwatson * r5391 trunk/ (d-i/manifest debian/changelog): Automatic update of included source packages: partman-target 77ubuntu2.
[11:48] <CIA-32> ubiquity: cjwatson * r5392 trunk/debian/changelog: releasing version 2.10.15
[12:05] <stgraber> right, change done, tested and merge proposal sent. Now to look at something else.
[12:44] <gema> cjwatson: continuing the saga of chinese translation problems: bug 985524, bug 985598, bug 985605, bug 985614 and bug 985616
[12:44] <ubot2> Launchpad bug 985524 in debian-installer "Wrong translation during the installation process of Ubuntu 12.04 precise-alternate-amd64.iso with choosing Simple Chinese as the installation language" [Undecided,Confirmed] https://launchpad.net/bugs/985524
[12:44] <ubot2> Launchpad bug 985598 in debian-installer "the Chinese font size seems not the same" [Undecided,New] https://launchpad.net/bugs/985598
[12:44] <ubot2> Launchpad bug 985605 in debian-installer "The keyboard config page should be translated into Chinese" [Undecided,New] https://launchpad.net/bugs/985605
[12:44] <ubot2> Launchpad bug 985614 in debian-installer "Something wrong in the keyboard layout selection page" [Undecided,New] https://launchpad.net/bugs/985614
[12:44] <ubot2> Launchpad bug 985616 in debian-installer "sometimes, there is no color at the lower right corner of the screen during the installation process" [Undecided,New] https://launchpad.net/bugs/985616
[12:45] <gema> cjwatson: hope that is more manageable
[12:46] <cjwatson> it's certainly better as multiple bugs, yes, thanks
[12:46] <gema> np
[12:47] <cjwatson> Although somebody should tell Victor that not everyone speaks Chinese and thus if he's going to say that a translation is wrong he also needs to say what the correct translation would be :-)
[12:47] <gema> haha, good point
[12:50] <gema> cjwatson: I was wondering myself how are you guys going to search for text that is in a picture, not having a chinese keyboard
[12:51] <cjwatson> with difficulty ...
[12:51] <gema> I will ask him to type the wrong and the right translation clearly, so that it makes it easier
[12:52] <cjwatson> well, if he's going to be contributing correct translations, Launchpad Translations is better at that than bugs ...
[12:52] <cjwatson> if I'm lucky I can recognise it from context
[12:52] <gema> yes, the problem is that we are raising these for precise
[12:52] <gema> so kate needs to have visibility of what's going on
[12:53] <cjwatson> nothing I've seen yet is RC for precise, honestly
[12:53] <gema> no, still half an hour for my meeting with her
[12:53] <gema> it's her call
[12:53] <gema> he asked for the testing
[12:53] <cjwatson> if we were going to be doing a serious translation push for Chinese, it should have been weeks ago
[12:53] <gema> she
[12:53] <gema> agreed
[12:53] <cjwatson> right, I think she wants testing of the Chinese images, but more like "does it actually work at all sanely"
[12:53] <gema> that, and many other things
[12:54] <cjwatson> I wasn't aware she was after 100% translation
[12:54] <cjwatson> (say)
[12:54] <gema> me neither, until yesterday when she was pushing for this, so I give her the bugs and she can decide, not my call
[12:55] <gema> maybe she thinks some are more visible and hence worth fixing than others
[12:55] <stgraber> I've done quite a few Chinese installs recently (it's one of the usual tests for Edubuntu) and quite a few d-i dialogs aren't fully translated, same for the post-install experience
[12:55] <gema> we'll be better positioned for Q, with victor in the team, we didn't have anyone that could read chinese before
[12:55] <stgraber> but nothing that needed a bug there as it was just missing translations that the translators need to work on
[12:56] <gema> nothing that you could read, stgraber , I am guessing
[12:56] <gema> if something is wrongly translated, depending on what it actually says, it may need fixing
[12:56] <stgraber> a weird one though is the d-i dialog telling you that your current language isn't fully translated was entirely in english ;)
[12:56] <gema> haha
[12:57] <gema> at least they can guess from that message the meaning x)
[12:57] <stgraber> gema: right, I'm only interested in things that can't be translated and that's what I'm after in these test installs. I don't care about the translation being wrong as I can't do anything about it anyway
[12:57] <gema> stgraber: ack
[12:57] <gema> stgraber: that's where we'll try to help
[12:57] <stgraber> last I checked I didn't even have the required LP rights to fix a french translation
[12:58] <cjwatson> I do, but I can only do so when it's utterly blatant and trivial to fix
[12:58] <cjwatson> I'm certainly not getting into correcting wording in 60 languages
[12:58] <gema> cjwatson: what language do you translate?
[12:58] <cjwatson> none
[12:58] <gema> ah, ok
[12:59] <cjwatson> I'm only fluent in English although I read several other languages
[12:59] <cjwatson> (with varying degrees of difficulty)
[12:59] <gema> ok
[12:59] <cjwatson> but in some cases it's clear that a translation is wrong regardless of that
[12:59] <gema> yep
[12:59] <stgraber> so it sounds like what we need for Chinese is a native speaker who's good at translation to go through the whole desktop and update the translations on LP, but too late for 12.04 as we need new langpacks + installer uploads, so would have to wait for .1
[12:59] <cjwatson> for instance when there are HTML tags embedded in the translated string and somebody translated the tag names
[13:00] <gema> cjwatson: that's a good one
[13:00] <gema> I think we should ask for duolingo's help for spanish and german
[13:00] <CIA-32> console-setup: cjwatson * r434 ubuntu/debian/ (42 files in 2 dirs):
[13:00] <CIA-32> console-setup: Update Ubuntu-specific translations from Launchpad (fixes most of LP
[13:00] <CIA-32> console-setup: #985605).
[13:00] <cjwatson> In most cases for the installer, the most productive way to do this is to translate d-i upstrea
[13:00] <cjwatson> m
[13:01] <cjwatson> The exceptions are strings that are ubiquity-specific and a handful of other Ubuntu-specific strings
[13:01] <cjwatson> there are reasonably stock instructions in the translator bits of the wiki somewhere nowadays, I think
[13:01] <gema> stgraber: how do we know someone is good at translating chinese?
[13:02] <gema> stgraber: I mean I trust victor knows chinese, but being good at translating takes more than just being native in the language
[13:02] <stgraber> gema: one thing would be to have him join the Chinese translator mailing-list, read the team guidelines (if they have some), ... translation teams are usually good at peer review
[13:02] <gema> stgraber: I will ask skaet how she wants to proceed
[13:03] <gema> stgraber: ack
[13:03] <stgraber> gema: in an ideal world, he should only report bugs against d-i/ubiquity/anything when the translation is present in LP but not visible in the UI. For the other cases, it'd be easier to just contribute new translations.
[13:04] <cjwatson> +1
[13:04] <cjwatson> we do *not* want to be forced into the role of arbitrating between disagreeing translators ...
[13:04] <cjwatson> that's the role of a translation team
[13:04] <stgraber> gema: I'm fine with these few extra bugs for now, but if we start getting similar bugs for every language and software in Ubuntu, we won't be able to manage the load
[13:04] <cjwatson> stgraber: ... or when text is visible in the UI but not available for translation in LP
[13:05] <gema> ok, maybe victor, jibel and myself should all be part of a translator team for chinese, french and spanish and fix things there
[13:05] <gema> going forward
[13:05] <stgraber> cjwatson: right
[13:05] <cjwatson> or when it isn't possible to construct a correct translation due to a software defect
[13:05] <cjwatson> gema: if you're interested in complete translations for those languages, yes, you should
[13:05] <gema> I am going to add that as a topic to discuss at UDS
[13:05] <cjwatson> talk to dpm; he can give you pointers
[13:05] <gema> ok
[13:06] <cjwatson> hm, there's something wrong with my translation update scripts
[13:06] <cjwatson> that console-setup translation should actually be complete ...
[13:09] <cjwatson> ah, that's better, they work better with the right options
[13:10] <CIA-32> console-setup: cjwatson * r435 ubuntu/debian/ (37 files in 2 dirs): invoke my translation update script with the right options, fixing LP: #985605 properly
[13:10] <jibel> gema, there are people whose interest is translation, it is not mine. And given my skills in non-english languages, I am not sure it is a good idea that I contribute to translations :)
[13:11] <jibel> s/non-english/non-french/
[13:13] <antarus> your english looks fine so far ;p
[13:13] <antarus> I wonder if there are any ubuntu developers in Munich...
[13:15] <ogra_> we used to have a map on planet.ubuntu.com where you could look .... but that seems to have vanished
[13:16] <antarus> oh man
[13:17] <antarus> I wonder if we could get a corp blog on planet
[13:17]  * antarus fears the paperwork involved
[13:24] <gema> sorry, I broke my irssi
[13:25] <gema> some days I wish I didn't break everything I touch
[13:26] <stgraber> aren't you paid to break everything you touch?
[13:26] <gema> stgraber: yes, but I need at least a working browser to report the problems ;)
[13:26] <CIA-32> debian-installer: cjwatson * r1679 ubuntu/ (build/Makefile debian/changelog):
[13:26] <CIA-32> debian-installer: Add all characters in /usr/share/console-setup-mini/kbdnames.gz to the
[13:26] <CIA-32> debian-installer: reduced font (LP: #985614).
[13:26] <antarus> gema: launchpad has a commandline api ;p
[13:27] <gema> antarus: that's a new one, I will  look into that
[13:27] <gema> antarus: but I also break launchpad from time to time
[13:27] <antarus> don't get me wrong, I haven't used it
[13:27] <antarus> it has too many dependencies that I have to check in :/
[13:28] <gema> and my car's radio breaks too easily too
[13:29] <gema> need to figure out how to raise bugs with VW
[13:37] <cjwatson> gema: https://help.launchpad.net/API/launchpadlib
[13:37] <cjwatson> absolutely invaluable, well worth learning
[13:38] <stgraber> +1
[14:03] <didrocks> hey
[14:03] <mpt> hi didrocks
[14:03] <mpt> So, about OneConf
[14:03] <mpt> This would just restore installed software, not (a) config files or (b) home folder, right?
[14:04] <didrocks> right, just installed software
[14:04] <didrocks> I see two use case: cloning an install with the same package set
[14:04] <didrocks> or reinstalling, and so "taking over" the old computer identifier
[14:05] <gema> cjwatson: added to my todo
[14:06] <mpt> didrocks, so the precondition is that you've chosen "Sync Between Computers" on the old computer?
[14:07] <mpt> and that the sync has finished
[14:07] <didrocks> mpt: indeed, otherwise, you have no set available
[14:07] <mpt> ok
[14:07] <didrocks> mpt: otherwise, it will just be a "subscribe" thing. Not sure we want that
[14:10] <mpt> didrocks, ok, so a first draft of the intro text in the installer: "If you've synced a computer before in Ubuntu Software Center, you can reinstall the same software on this one that you had on that one."
[14:12] <didrocks> mpt: sounds to confirm the intend. Not sure if the "this one… that one" is heavy in english (it is in French), but otherwise sounds good
[14:12] <mpt> yes, needs polishin
[14:12] <mpt> g
[14:12] <didrocks> mpt: I think there should be a checkbox as well for "this computer replace the selected one" (in proper english ;))
[14:12] <mpt> didrocks, ah, so you don't have the old one hanging around in the list
[14:12] <didrocks> as it's overwriting, meaning it's the same installation or that we will never use the old one
[14:12] <didrocks> right
[14:13] <mpt> ok
[14:13] <didrocks> the most difficult part IMHO is the credential one
[14:13] <didrocks> not sure we want the ubuntu sso dialog to popup on top of ubiquity
[14:13] <cjwatson> in general installing a load of extra software in the context of the installer is going to create a bunch of bugs; we should consider reliability
[14:14] <cjwatson> perhaps stage things for installation after the first reboot or something
[14:15] <mpt> didrocks, ideally we'd embed it
[14:15] <didrocks> cjwatson: oh really? passing the list for additional installation to ubiquity is leading to issue?
[14:15] <mpt> but if we have to stage it for after first restart, then we'd either need to store the token somewhere, or ask for sign-in after the restart
[14:17] <stgraber> didrocks: we had a few problems with the upgrade method that's using apt-clone and does roughly the same thing (install a whole bunch of packages). If one of them fails to install, ubiquity tends to fail
[14:17] <didrocks> mpt: indeed or copying the token in the target
[14:17] <didrocks> stgraber: ah…
[14:18] <didrocks> yeah, that can an issue then
[14:18] <cjwatson> didrocks: we have too many bugs already about things like language-pack installation, apt-clone handling for the quasi-upgrade case, all sorts of stuff like that
[14:18] <cjwatson> the root problem really is that if the installation fails you tend to be stuffed and have to start again
[14:18] <stgraber> didrocks: it mostly happens at release time when the mirrors are overloaded
[14:18] <didrocks> hum, marking things to install isn't making things better than just starting software-center
[14:19] <cjwatson> workaroundable sometimes with "try again" type dialogs, but for this kind of complex thing it feels that pushing to post-install would be a lot less risky
[14:19] <stgraber> Ideally we could have ubiquity do the installs, have a try-again option if it fails and if that doesn't help, queue all the installs (langpacks, drivers, packages from oneconf, ...) for post-install
[14:19] <didrocks> mpt: I'm really unsure about what to do with those infos then
[14:19] <cjwatson> but then what if the install of those packages fails half-way through or something
[14:20] <stgraber> and then have a generic thing (update-manager or similar) pick that up and process any remaining install post-install
[14:20] <cjwatson> or a postinst fails and leaves the system in a busted state
[14:20] <cjwatson> all it would take would be one broken package in software-center
[14:21] <cjwatson> now admittedly this is a set of problems we have already, but sort of hidden off in a corner
[14:21] <stgraber> cjwatson: though there wouldn't be much risk running that post-install code at the end of ubiquity when we know the system is already bootable, right?
[14:21] <mpt> didrocks, ah, but if we delay asking for sign-in until the restart, then we'd have to delay asking for which computer you want to sync from until after the restart too ... otherwise we wouldn't have the list of computers to choose from.
[14:21] <didrocks> yeah, we shouldn't make it first
[14:22] <didrocks> mpt: the sign-in can be in ubiquity, then, we copy the credentials to the targeted system
[14:22] <stgraber> (as in, wouldn't be worse than running it form the user session after the first boot, result should be identical)
[14:22] <didrocks> s/first/worse
[14:22] <cjwatson> stgraber: maybe, yeah
[14:22] <cjwatson> I do hope this means the desktop team is going to work on the installer ;-)
[14:23] <didrocks> what you mean by "post-install" is the same that all the scripts we executes at the end,
[14:23] <didrocks> cjwatson: on that part, we can help some volonteers that will do the oneconf integration as their school project :)
[14:23] <mpt> A school project mucking around with an OS installer, what could possibly go wrong
[14:24]  * mpt ducks
[14:25] <didrocks> mpt: well, when I touched ubiquity for UNE and the post-intaller, I think the portion of code was definied enough to not screw everything. As most of the work is done by the oneconf backend which is already a separate process over dbus, the work done ubiquity side should be really small
[14:25] <cjwatson> gema: did you have that meeting with skaet?
[14:25] <mpt> cjwatson, the main objective is presenting the "reinstall this set of software" option at the same point as the other installation options, otherwise it's a lot less obvious. Where the rest of the process happens matters less.
[14:25] <cjwatson> mpt: *nod*
[14:26] <gema> cjwatson: yes, she want's someone else to fix the problems
[14:26] <gema> cjwatson: she's on the case
[14:26] <cjwatson> didrocks: of course that kind of arrangement can take a bit of fiddling to make work given that the installer is in a different root filesystem from where the backend would want to live
[14:29] <didrocks> cjwatson: indeed, especially in computing the diff
[14:35] <mpt> hm hm hm
[14:36] <mpt> didrocks, so will it be practical to calculate the amount of disk space required by the packages in the selected sync profile?
[14:37] <didrocks> mpt: hum, interesting, I should poke mvo about it, but yeah, should be possible
[14:37] <didrocks> mpt: also, that would mean that we install the application diff, right? (all packages manually installed) and removed what wasn't in the previous package set?
[14:38] <didrocks> (like if the previous package set had banshee, it will now remove rhythmbox and install banshee)
[14:39] <mpt> didrocks, I don't mind which way it's done. The reason I asked was so that (for example) we don't offer people the option to install Ubuntu alongside Windows, if the amount of software they had installed on Ubuntu means there isn't enough disk space to do that.
[14:42] <didrocks> mpt: yeah, makes sense, I'll ensure we will get that in
[14:42] <didrocks> (and good thinking!)
[14:55] <mpt> didrocks, what does OneConf know about each synced computer, other than (1) its hostname, (2) what software is installed, and (3) when it was last synced?
[14:56] <didrocks> mpt: the logo of it (meaning, a preview of their wallpaper in the "computer icon") but it's not uploaded yet to apps.ubuntu.com because of duplication concerns.
[14:57] <didrocks> mpt: on the (2), we have the full list of installed packages, knowing if it's marked as being "auto" installed or "manually" installed
[14:57] <didrocks> and that's about it, do you think it should know more,
[14:57] <mpt> didrocks, ooh, if we know its hostname, and you choose "[/] This computer will replace the selected one", we should default to the same hostname
[14:57] <didrocks> ?
[14:57] <didrocks> yeah, that will be pretty neat :)
[14:58] <didrocks> (if we need to know more in the future, adding fields is really trivial)
[14:58] <didrocks> without any API break
[14:58] <mpt> meeting, bbl
[15:05] <CIA-32> console-setup: cjwatson * r436 ubuntu/debian/changelog: releasing version 1.70ubuntu5
[15:12] <runeks> i'm trying to create a custom Ubuntu Precise live cd (with GTK3 compiled with Wayland support). for this I need to put a file in the .config folder that is in the home directory. how do I do this? I tried putting the file in /etc/xdg (because it seems to contain files that end up in ~/.config) but that didn't work.
[15:15] <stgraber> runeks: hmm, I'm a bit surprised you actually have to create something in .config that's not configurable system-wide. But #ubuntu-desktop would probably be better for that. If you want to have a file show up for every new accounts created on the system, use /etc/skel
[15:16] <runeks> stgraber: i've actually edited the .bashrc file in /etc/skel. but this ends up in the home folder. how do I make it end up in the .config folder under the home folder? simply make a .config folder in /etc/skel and put the file in there?
[15:17] <stgraber> runeks: yes
[15:18] <runeks> cool. will try that. thanks!
[15:33] <runeks> works like a charm. thanks stgraber!
[15:33] <stgraber> np
[17:37] <CIA-32> debian-installer: cjwatson * r1680 ubuntu/debian/changelog: releasing version 20101020ubuntu135
[22:37] <stgraber> bug 985919 is pretty weird... quickly checking if I can reproduce
[22:37] <ubot2> Launchpad bug 985919 in ubiquity "ubi-partman crash when network is disable" [Undecided,New] https://launchpad.net/bugs/985919
[22:52] <stgraber> right, can't reproduce with Ubuntu so probably something racy or really Lubuntu specific, will see if Julien can reproduce it reliably