[00:12] I'm trying to install a minimal install of ubuntu with wubi using the instructions here: http://ubuntuforums.org/showpost.php?p=3274621&postcount=2 but I get a full install with gnome, unity, etc. instead of a tty login === cjwatson_ is now known as cjwatson === JanC_ is now known as JanC === mpt_ is now known as mpt [11:51] bug 959251 noone experirncing it in real hardware ? [11:51] Launchpad bug 959251 in ubiquity "** CRITICAL **: unable to create '/root/.cache/dconf'" [Undecided,New] https://launchpad.net/bugs/959251 [12:04] Is there a preseed option so that d-i would not use swap if it found a swap partition? [12:53] * cjwatson starts writing a test for bug 960047 [12:53] Launchpad bug 960047 in ubiquity "ubiquity crashed with UnboundLocalError in set_indicator_keymaps(): local variable 'engine' referenced before assignment" [High,Triaged] https://launchpad.net/bugs/960047 [12:54] glad you noted that - I was just about to pounce on it :) [12:54] tests for the win [12:54] * ev grabs a tea [12:55] test_set_indicator_keymaps_english (test_misc.MiscTests) ... Segmentation fault (core dumped) [12:56] wtf. maybe I need to upgrade first [13:03] bug 960039 might be worth a look [13:03] Launchpad bug 960039 in ubiquity "ubiquity greeter : Label of the network indicator changes to the keyboard layout code" [Undecided,New] https://launchpad.net/bugs/960039 [13:04] o/ [13:08] OK, does this segfault for anyone else? python -c 'from gi.repository import GdkX11; display = GdkX11.x11_get_default_xdisplay()' [13:08] still upgrading stuff, but I've upgraded most of the libraries ... [13:09] cjwatson, i does on an up to date precise amd64 [13:09] jibel: it segfaults? [13:09] cjwatson, yes [13:09] yay PyGI [13:09] bah, so how come the tests passed in stgraber's upload? [13:10] yay constructive ;-) [13:10] :) [13:10] segfaults for me as well, and I'm not nearly up to date [13:11] worked fine in the 2.9.30/i386 build [13:11] segv here too, you're not alone. [13:13] * infinity goes to hunt some caffeine before diving into things. [13:14] * ogra_ twiddles thumbs in front of the obligatory "waiting 60 seconds for network devices" message [13:14] cjwatson: that should segfault when you're not in a gtk mainloop (from previous tests) [13:15] oh. (FSVO "should") [13:15] I encountered this in the middle of the test suite, though [13:16] well, apparently there are cases in gi world where stuff are kind of expected to segfault ;) "like from gi.repository import Xkl; test = Xkl.Engine()" because you should instead us Xkl.Engine.get_instance() :) [13:19] 'tests/run test_misc' -> test_set_indicator_keymaps_english (test_misc.MiscTests) ... Segmentation fault (core dumped) [13:20] ogra_: it's not obligatory at all, you should only see that message if you have a broken network config? [13:20] ah, yes, it's a test isolation bug - it works if you run the whole test suite [13:20] (that's at least enough for me to get past this and fix what I actually wanted to work on) [13:21] slangasek, well, its obligatory if oem-config didnt run yet :) but i also end up with no /e/n/i after a successfull non networked install (not even loopback in there) ... the latter is on my list for things to look at after (if i ever *sigh*) found out whats wrong with automatic preseeding installs in oem-config [13:23] hmm, let me look at that ifupdown hook - is it possible we don't properly handle the case that there are *no* networks configured in /e/n/i? [13:23] intrestingly the log doesnt reveal much exciting stuff even if i run oem-config --automatic --debug [13:24] slangasek, i would rather blame oem-config to not run all modules ubiquity would usually [13:24] ogra_: no, just reviewed /etc/network/if-up.d/upstart, and it was written to handle the case when there are no interfaces configured in /e/n/i - as long as 'ifup lo' succeeds [13:25] ogra_: there *was* a problem with this error message showing up on first boot, which was tracked down to a resolvconf bug; but that should be fixed now [13:25] hmm, but then i shouldnt get the message [13:25] i.e., the resolvconf hook was exiting non-zero and preventing the upstart hook from ever firing [13:25] also shouldnt we drop lo from /e/n/i for completeness ? [13:26] no [13:26] that's where it's supposed to be configured :) [13:26] but upstart already handles it [13:26] after getting this message, can you grab me the output of 'sudo ifquery --list' and 'ls -l /run/network'? [13:26] upstart handles it to protect you from a misconfigured /e/n/i [13:27] it's actually possible that you might want to have additional configuration of lo, beyond the default [13:27] well, thats fine to add to /e/n/i then :) [13:28] slangasek: okay, net's back up, but mumble is still ill ;) [13:28] i just think (since years already) that its superfluous to have it there for the default [13:28] slangasek: resolvconf hook preventing the upstart hook was a while ago right? (want to make sure my revert of ifupdown's run-parts arguments didn't fail for some reason without me knowing ;)) [13:28] * ogra_ curses the HUD getting in his way all the time [13:29] bug 959251 not reproducible in vm, only in real hardware [13:29] Launchpad bug 959251 in ubiquity "** CRITICAL **: unable to create '/root/.cache/dconf'" [Undecided,New] https://launchpad.net/bugs/959251 [13:29] ogra_: sigh [13:29] feels like someone wants to teach me to take my hands off the kbd if i dont type [13:29] stgraber: yes, that was supposed to be fixed in 1.63ubuntu8 [13:31] could somebody look at 959251, please? [13:31] https://code.launchpad.net/~cjwatson/ubiquity/indicator-keymaps-960047/+merge/98407 [13:32] cjwatson: I'm currently working on bug 960096 but I'm happy to help someone with 959251 (as I'm the author of the gsettings code) or will have a look at it myself later today [13:32] Launchpad bug 960096 in ubiquity "Live session started with wrong layout" [Undecided,In progress] https://launchpad.net/bugs/960096 [13:34] bug 960036 and bug 960039 are new keyboard indicator bugs, although I don't know if they're regressions or just things that never worked [13:34] Launchpad bug 960036 in ubiquity "ubiquity greeter : changing language changes order of the indicators" [Undecided,New] https://launchpad.net/bugs/960036 [13:34] Launchpad bug 960039 in ubiquity "ubiquity greeter : Label of the network indicator changes to the keyboard layout code" [Undecided,New] https://launchpad.net/bugs/960039 [13:36] cjwatson: engine.lock_group(0) is meant to be called in all cases, not only when we reset the gsettings key [13:36] ogra_: are you using unity 2d or 3d? [13:36] cjwatson: engine.lock_group(0) is supposed to tell the indicator to select the first entry in the list [13:36] stgraber: hm, ok, so we should actually assign something to engine even if not nodes, then? [13:37] barry, usually 3d, for the hangout stuff i switched to an x86 machine though and that runs 3d ... its all massively confusing (cant find my terminals with alt-tab, hud gets in my way etc) [13:37] cjwatson: doh, should have seen that one... yes, that bit of code should be moved out of the if [13:38] ogra_: s/usually 3d/usually 2d/ ? [13:38] heh [13:38] usually 2d indeed [13:39] ogra_: maybe you could add a comment to bug 947468. i seem to be unconvincing [13:39] Launchpad bug 947468 in unity-2d "Alt summoning of heads-up-display interferes with Emacs" [Undecided,Confirmed] https://launchpad.net/bugs/947468 [13:39] barry, will do [13:39] thanks! trying to rally the troops :) [13:41] Any chance of getting support for the Broadcom 5720 QP NIC in the 10.04 LTS pxeboot images? [13:41] a flood of servers are coming to market bearing these [13:42] cjwatson: if you can move the display and engine lines from within the "if nodes" to just before it and then move back lock_group to outside the else, the MP looks good for merging [13:42] Teduardo, before or after april ? [13:42] Do you mean are the servers coming before or after april, or do you mean that the version of Ubuntu? [13:43] the servers indeed [13:43] I have some from Dell already [13:43] sitting here [13:43] 12.04 LTS releases in april [13:43] oh, I was referring to the 10.04 [13:43] and i dont think there were any new spins of 10.04 planned [13:44] Teduardo: #ubuntu-kernel would be better, in any event - the installer mostly isn't directly responsible for hardware support [13:44] stgraber: right - reload the MP? [13:44] Sorry, I meant the kernel in the installer... splitting hairs now =) [13:44] Teduardo: same kernel [13:45] * barry -> reboot [13:45] Teduardo: and the kernel packaging is responsible for selecting which modules to deliver to the installer, as well [13:45] cjwatson: looks good. Marked as approved on LP [13:45] right but if they update the kernel in the distribution, someone would then have to go back and port the new one to the installer images [13:45] Teduardo: which is a trivial thing to do, but the kernel would have to be sorted out first [13:45] honestly, not much point asking us before the kernel is changed [13:46] grrr, why do i get a completely corrupted syslog from time to time in my installer tests [13:46] ubiquity: cjwatson * r5280 trunk/ (debian/changelog tests/test_misc.py ubiquity/misc.py): [13:46] ubiquity: Fix misc.set_indicator_keymaps to handle ll_CC forms of lang, and also [13:46] ubiquity: to avoid crashing on unknown languages in general (LP: #960047). [13:47] ubiquity: cjwatson * r5281 trunk/ubiquity/misc.py: call engine.lock_group(0) in all cases, per stgraber [13:50] cjwatson: Did you have any changes staged/shelved for powerpc->powerpc-smp? If not, I'll do that all today. [14:02] 23699 Jan 1 00:24:53 localhost localechooser: info: debian-installer/language preseeded to 'en' (seen: true) [14:02] 23702 Jan 1 00:24:53 localhost localechooser: info: Preseeded language ignored: unknown language code [14:02] cjwatson, ^^^ whats a proper preseed value for that ? [14:03] ogra_: 'en'? [14:03] infinity: I do, but just http://paste.ubuntu.com/892213/, feel free to fiddle from there [14:03] (sees that actually nothing is wrong with oem-config --automatic if i get the "seen: true" in teh log the question is actually not asked [14:04] infinity, but thats what the pressed file has :P [14:04] cjwatson: Alright, so still need to touch lots of other bits. I'll get on that. [14:04] d-i debian-installer/language string en [14:04] maybe 'd-i debian-installer/locale string en_US.UTF-8' or something instead [14:05] not sure offhand why en would be failing, would require some digging and I ought to look at this grub2 regression [14:05] yeah, i didnt mean to pull you into my problems :) [14:05] njin: ok, reproduced your bug, trying to figure out what's happening there exactly... [14:05] great stgraber ! [14:06] ogra_: /usr/share/localechooser/SUPPORTED might be a good place to start. [14:06] 'en' really ought to work - it's the default for that template [14:06] awesome, thanks ! [14:07] cjwatson, well, i'm starting from a seemingly broken preseed file there might be something else intervening, i will start from scratch [14:07] ogra_: (or so the code implies) [14:07] But yeah, anything from languagelist should work too, which includes all the 2-char codes. [14:20] ev: http://paste.ubuntu.com/892235/ + mkdir /tmp/my-bus + dbus-daemon --config-file=system.conf --nofork [14:20] ev: that seems to work here [14:20] stgraber: cheers! [14:21] ev: that pastebin is just /etc/dbus-1/system.conf with the bus address changed and changed to run as you instead of messsagebus [14:33] njin: I "think" I have a fix. Hopefully this one will be the right one... this function is getting really big for something supposedly simple [14:35] do you need me to test something ? [14:36] stgraber:^^ [14:36] * bdmurray gets some more coffee [14:43] njin: not just yet, I'm fixing some tests, building a new usb stick with the fix and if it all works here, I'll push it to trunk, release a new ubiquity and we'll have a respin in a few hours for you to test [14:54] great [14:57] screen -r [14:58] DAMN ! [14:58] * ogra_ really starts to dislike unity 3D [14:58] this is not the screen you are looking for. [15:02] cjwatson: could you perhaps sanity check this comment: https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/933433/comments/8 [15:02] Launchpad bug 933433 in ubiquity "Kubuntu manual install crashed during bootloader configuration on an XFS partition" [Medium,Triaged] [15:05] barry: it would be worth comparing your syslog with the reporter's [15:06] cjwatson: it's exactly the same until you get to the traceback [15:06] cjwatson: i.e. there is no traceback any more [15:06] barry: ok, good [15:06] barry: sounds sane then; the reporter's on 10.04, after all [15:06] cjwatson: yep. cool, i'll move on to bug 959724. thanks [15:06] Launchpad bug 959724 in ubiquity "Limit boot loader installation target" [Medium,Triaged] https://launchpad.net/bugs/959724 [15:09] * barry -> tea [15:12] speaking of compcache - perhaps someone could look at bug 933712 [15:12] Launchpad bug 933712 in casper "casper configures system to use 50% ram for compressed swap post installation" [Low,New] https://launchpad.net/bugs/933712 [15:13] ogra_: want to take a look? [15:13] cjwatson, at ? (zram?) [15:13] 933712 [15:14] ah, sure [15:14] https://code.launchpad.net/~jamesodhunt/ubuntu/precise/ubiquity/bug-944191/+merge/98430 [15:14] * jodh reboots (USB woes...) [15:17] I'm not sure precisely what compcache-enable does, but if you just have zram swap(s) setup, it doesn't "steal" your RAM in any negative sense. [15:18] Although, now that I realise that's there in initramfs-tools, maybe I should fold zram-config into it. :P [15:21] jodh: could you push that to lp:~jamesodhunt/ubiquity/bug-944191 instead, and propose it for merging into lp:ubiquity rather than lp:ubuntu/ubiquity? we don't use lp:ubuntu/ubiquity [15:22] lp:ubuntu/ubiquity is an auto-import so attempts to merge actual branches into it tend to produce giant diffs :) [15:24] stgraber: still hating gtk today? I think bdmurray could use some help with bug #959486; my guess based on what he says he's seen is that the code is only setting the gtk labels after the window has already been displayed, so you only see them on a full window refresh, but I'm not sure where in the code that's happening [15:24] Launchpad bug 959486 in ubiquity "install alongside doesn't show which partition is which" [Medium,Triaged] https://launchpad.net/bugs/959486 [15:24] I think its in gtkwidgets.py [15:25] and I also noticed that the icons aren't appearing. There are supposed to be icons for the operating systems right? [15:25] slangasek: I'm currently hating gnome, trying to figure out why the keyboard layout stuff still doesn't work when going from greeter => live session... [15:25] icons> dunno [15:26] ev: ^^ this is your code for the partition widget, right? [15:26] bdmurray: I'm not too familiar with that part of the code so I'd have to poke at it for a moment, maybe ev knows though [15:26] likely my code [15:26] * ev digs [15:27] there is a FIXME line regarding rendering dots over the frame but I tried not painting the background and that didn't seem to help [15:28] eah, that's my code [15:28] so I'd suggest looking at the draw function (do_draw in PyGI-land) [15:28] see if that's firing [15:29] err actually [15:29] brb [15:32] sorry, elmo grabbed me [15:33] bdmurray: so I'd try to narrow this down by stubbing out the paint_background functions and see if that brings back the text at least [15:33] if that doesn't help, see if commenting out our custom draw function helps [15:33] ahem, with corrected merge branch - https://code.launchpad.net/~jamesodhunt/ubuntu/precise/ubiquity/bug-944191/+merge/98434 [15:34] infinity, i was thinking about that [15:34] jodh: uhm, shouldn't that be against lp:ubiquity? [15:34] ev: okay thanks [15:35] ev: yes, something odd happened there: https://code.launchpad.net/~jamesodhunt/ubiquity/bug-944191/+merge/98435 [15:35] ogra_: The (slightly better) logic in zram-config could certainly be rolled into compcache-enable. [15:35] infinity, yeah [15:35] ogra_: And then we just need to turn it on in places where we want it. [15:36] well, the prob with the bug seems to be that we need to parse the cmdline to make it only run in casper envs [15:36] i.e. only if boot=casper [15:37] but then we break it for everyone who *wants* to use it [15:37] (like ac100 users) [15:38] i dont really see any solution to the bug apart from "dont install casper on normal systems" [15:38] (which was the answer to the last iteration of the same bug we had years ago iirc) [15:40] ogra_: Well, we can enable it conditionally based on a kernel cmdline, and add that to the ac100 cmdline. [15:40] compcache=noreallyimeanit [15:40] infinity, thats what happens already ... well, the COMPCACHE:SIZE var is shipped in casper and only if thats present compcache is actually used ... [15:41] and we have a kernel cmdline var to *dis* able it [15:41] Yeah. [15:42] i dont really like to fiddle around with kernel cmdlines more than we need [15:42] Dirty hack, check for [ -d casper-bottom ] && ! boot=casper? [15:42] ie: if casper is installed, but not in use? [15:44] Still, seems like it's just breaking one weird corner case (you have casper locally for some odd reason, *and* want compcache) to prevent another (you have casper installed locally when you shouldn't). [15:45] yeah, that sounds pretty sane [15:46] otoh i would just close the bug :P [15:46] its not a bad thing to have 50% zram [15:46] yay, I've got NetworkManager running on a different system bus (provided I stop the existing one first) [15:46] cheers stgraber [15:46] and i dont see why it would be a bad user experience to suddenly have more ram :) [15:47] infinity: if you want compcache explicitly, though, you can always configure it in initramfs.size [15:47] initramfs.conf [15:47] or another conf.d script [15:51] cjwatson, still, i would just close that bug, a) using zram has no actual drawbacks, b) casper is usually not installed ... if infinity wants to fold his zram handling into the casper scripts i think thats fine but independent [15:54] [50786.728259] init: whoopsie main process (1783) killed by ABRT signal [15:54] I see that in dmesg and not in syslog fwiw [15:56] ev: https://bugs.launchpad.net/ubuntu/+source/whoopsie-daisy/+bug/960326 [15:56] Launchpad bug 960326 in whoopsie-daisy "whoopsie-daisy SIGSEGV but no crash file generated" [Undecided,New] [15:57] * cjwatson posts hopefully enough analysis to bug 855871 to deconfuse the reporter [15:57] Launchpad bug 855871 in partman-partitioning "Grub install fails after manual xfs partitioning" [Medium,Triaged] https://launchpad.net/bugs/855871 [15:57] I'm about to have a go at testing http://paste.ubuntu.com/892356/ for that [15:59] but first, COFFEE [16:02] cjwatson, i would appreciate a look at bug 924018 i added a syslog and debconf seems to even though it marks questions as "seen" still ask them [16:02] Launchpad bug 924018 in ubiquity "Preseeding doesn't work with oem-config" [Medium,Confirmed] https://launchpad.net/bugs/924018 [16:02] and i dont see anything in the ubiquity code that would prevent the automatic mode [16:09] ack, will require some meditation [16:10] hm, 'debug-oem-config' would've been useful [16:10] i just added --debug to the oem-config-wrapper cmdline [16:10] can you get that? (I realise it takes a while, sorry ...) [16:10] since i hacked it anyway [16:10] oh, but there's no debconf debug output in your log [16:10] maybe it's in /var/log/oem-config.log? [16:10] oh, one sec [16:11] LOL [16:11] yeah, and i was looking at the wrong log all the time [16:11] sorry, let me try myself first then before i waste your time [16:12] (thopugh i can attach it anyway) [16:12] sure [16:12] done [16:14] ogra_: is this the debconf frontend? [16:14] yeah, it is [16:15] yep [16:15] ARGH ! [16:15] the HUD just started lobO while i typed yep [16:15] *libO [16:15] --automatic has little effect on the debconf frontend - it only does anything much useful with frontends that filter the debconf protocol [16:16] apprently the p from yep got typed into the HUD window [16:16] so that is, I think, a red herring [16:16] hmm [16:16] but why is it asking the preseeded questions then [16:16] are they actually getting preseeded by anything? [16:17] there is a preseed.cfg that gets applied by jasper on first boot, yes [16:17] for example, there's no sign that clock-setup/utc is preseeded here, but that's in the preseed file [16:17] and i also applied it manually inveween to make it pick up changes [16:17] (using debconf-set-selections /presseed.cfg) [16:18] *preseed.cfg [16:18] it definitely is preseeded, i can see it with debconf-get-selections [16:18] oh, wait, clock-setup/utc is preseeded, you're right, odd [16:18] * cjwatson goes and peers at debconf [16:19] looking at the ubiquity plugins i see a lot of hardcoded places where "seen" is set to false ... not sure that relates anyhow if the debconf frontend is used [16:20] * ogra_ quickly picks up some fresh coffee [16:20] the reason the seen flag has no effect is that we explicitly set "Reshow: true" in the debconf configuration file [16:20] bin/ubiquity:start_debconf [16:20] now there was a reason I did that :-) let me do some archaeology [16:21] dates back to before the oem-config mrge [16:23] well, it doesnt seem to affect ubiquity --automatic [16:24] and i guess using oem-config that way is a very special usecase [16:24] (though important for QA to do automated install tests) [16:25] not saying it shouldn't be fixed somehow, I just want to remember the exact reason so as not to break other things [16:26] right, i was meaning to say, i wouldnt mid if it was special cased even with a special switch :) [16:26] *mind [16:26] that's one plausible option, yes [16:26] heh, original commit that added Reshow was "Add a debconf_ui frontend which can be used in text mode on a server." [16:26] LOL ! [16:27] anyway, I think it was basically because those questions had already been shown by d-i so if we wanted them to be actually asked then we needed to tell debconf to reshow [16:27] obviously doesn't make sense in automatic mode [16:27] right [16:27] so, let's see [16:28] cjwatson: So, platform seeds, live-build, find-live-filesystem, and debian-installer. Can you think of where else the powerpc-smp change needs to happen? [16:29] infinity: live-build or livecd-rootfs depending on where you want to do it [16:29] cjwatson: live-build was in the list already and done. :) [16:29] infinity: debian-cd/tools/boot/precise/boot-powerpc and possibly elsewhere [16:29] Oh, debian-cd, right. I only touched cdimage. [16:29] * infinity looks. [16:30] maybe data/precise/yaboot/ [16:30] or maybe not, it just has "powerpc64" [16:30] Yeah, that's subarch, not kernel flavour. [16:30] Cause we're wildly inconsistent there. [16:31] traditionally [16:31] yeah [16:31] and there's casper_subarch=powerpc [16:31] so ok, change casper_subarch, and leave the d-i subarch as powerpc [16:31] then debian-cd should cope from there [16:32] Yeah, that's what I'm getting from my grep too. Cheers. [16:33] ogra_: try http://paste.ubuntu.com/892423/ ? [16:33] something along those lines anyway ... [16:33] * ogra_ hacks that in [16:34] * infinity can't help but think he's missing something... [16:34] cjwatson: What's the name of the d-i component with all the kernel lookup tables? [16:34] infinity: base-installer [16:35] * infinity recalls having to touch that for some ARM kernels recently... [16:35] Right. [16:35] yeah, you probably need to fix that [16:40] ev: well I've managed to narrow it down to something in do_draw of StylizedFrame [16:43] cjwatson, still getting questions for user-setup and tasksel but it improved a lot and it could well be my tinkering over the whole day that messed up the debconf db [16:43] okay [16:43] i'll try with a clean image but that will take a while [16:44] I think I can blame bug 960096 on lightdm [16:44] Launchpad bug 960096 in ubiquity "Live session started with wrong layout" [Undecided,In progress] https://launchpad.net/bugs/960096 [16:44] (yeah, took me 3 hours to get to that conclusion) [16:44] cjwatson: What would you prefer for base-installer? Minimal 1-line diff ( http://paste.ubuntu.com/892440/ ), or simplyfying that section to remove all the SMP business? [16:44] now poking mterry to confirm and hopefully fix it [16:45] infinity: something that won't make me go insane when merging in future - I don't mind too much [16:45] infinity: remember to update the tests thouogh [16:46] cjwatson: Yeah, I think the tests may alread pass with powerpc-smp, but double-checking. [16:46] ogra_: OK - and that could even just be an inadequate preseed file [16:46] ogra_: sounds like a good improvement anyway [16:46] right [16:46] it definitely is ! [16:46] thanks so much ... i thought i would go insane here :) [16:47] GrueMaster, ^^^ so we got a huge step forward [16:47] ogra_: happy to look at a debconf log from a fresh image if it's not fixed / immediately obvious [16:47] ok [16:48] Awesome! [16:48] do you have a new preseed for me to try or does this require code changes? [16:48] cjwatson: Looks like the SMP magic comes from Debian, so I'll just go with the 1-line diff, much easier to merge. [16:48] GrueMaster, we need to add automatic-ubioquity to the cmdline and yes, there are some minor code changes [16:48] infinity: OK [16:49] ogra_: ok. Will wait for your signal to give it a whirl. [16:49] GrueMaster, i want to let the running install finish and will then start over with a clean image to see if we can get rid of the last bits [16:49] YAY ! [16:49] * ogra_ just logged in with ubuntu/ubuntu [16:50] so even though it asked me the user-sertup question about fullname and login, it applied the preseeded passwd (which it didnt ask for) [16:51] blast, I need an amd64 sbuild instance [16:51] hooray for mk-sbuild [17:08] mk-sbuild is awesome [17:08] * barry -> lunch [17:09] out for lunch too, hopefully this afternoon will be more productive for ubiquity ;) at least I'm pretty familiar with gnome and lightdm's keyboard handling... [17:09] https://code.launchpad.net/~cjwatson/partman-partitioning/gpt-boot-partition-855871/+merge/98460 - review/thoughts welcome [17:10] dinner [17:20] base-installer: adconrad * r971 ubuntu/ (4 files in 3 dirs): Switch to powerpc-smp kernels [17:41] * infinity notices that he's hungry, and goes to do something about that. [17:44] ARGH ! [17:44] ogra_: ? [17:44] * ogra_ curses unity once more ... so it popped up the SD card nautilus window but the globalmenu still had the close buttons for the terminal i was working in [17:45] which made me close my work window with all its history [17:45] Global menu follows focus. [17:45] Isn't it a feature that autofs-windows don't steal focus? [17:45] focus follows mouse here :P [17:45] and it seems thats still buggy [17:45] You use ffm with global menus? [17:46] That sounds horrifying. [17:46] (though i heard a lot promises it would be fixed in 12.04) [17:46] i use FFM since warty [17:46] Sure, I always used to use FFM. [17:46] But I can't see how it would ever play nicely with global menus. [17:46] and this install is from hoary ... just updated every release [17:46] Unless you have it on a really long delay, and move FAST. :P [17:47] which got me unity 3D ... but since i havent used any x86 desktop in over a year i never noticed how bad it is [17:48] anyway, to move to the close button i had to move across the desktop and thus across the fullscreen nautilus window [17:48] which i would have expected to switch the menu and window buttons to nautilus indeed [17:49] anyway, thanks god at least the terminal preserved its history [17:49] with ffm and compiz+unity, I would expect that mousing up to the top bar would cause compiz to recursively generate menus for the bar itself ;) [17:49] but I guess unity is smarter than that, at least [17:50] I used ffm with global menus for a while and it didn't bother me desperately because (a) I tend to use the keyboard to get at menus and (b) if I have multiple windows on the screen then they tend to be terminals and I don't care about their menus anyway [17:50] well, i'm using unity 3D for the first time in my life this week .... i must admit i'm shocked [17:50] but these days I have the global menu turned off [17:50] i mean, i played with it before ... but never actually used it [17:50] 2D doesnt have all these issues [17:50] 2D just steals your keybindings every other week [17:51] Heh. [17:51] well, i rarely use much more than alt-tab [17:51] not that I'm bitter [17:51] which doesnt seem to work as i need it at all here atm [17:51] i never end up in the window i expect to end up in [17:51] Well, to be fair, 2D's keybindings just mimick the 3D defaults. The problem is that they're all hardcoded. :/ [17:52] Sure would love it if some of these Linux programmers remembered what config files were for. [17:52] bah, and there is another prob with oem-config --automatic ... [17:52] the upstart job somehow should respect the kernel cmdline setting :P [17:53] infinity: yes, on 3D they're configurable and 2D they're hardcoded, as a rule ... [17:53] * ogra_ has wonderful debug output from two runs now .... but nothing automatic at all :P [17:53] it does, but doesn't have any automatic handling [17:54] I suggest you call it automatic-oem-config rather than automatic-ubiquity, in line with debug-oem-config [17:54] (might not necessarily want the same for both) [17:54] cjwatson: They're only configurable in 3D because unity happens to be an enormous compiz plugin. I doubt it was a conscious decision to make them configurable, but rather just how you work in that framework. [17:54] infinity: true [17:55] ogra_: so http://paste.ubuntu.com/892536/ say? [17:56] anyone had a chance to look at my partman-partitioning change? [17:57] I think I missed the call for review. [17:57] 48 minutes ago? [17:57] * infinity looks. [17:57] 17:09 https://code.launchpad.net/~cjwatson/partman-partitioning/gpt-boot-partition-855871/+merge/98460 - review/thoughts welcome [17:58] I wonder how much "efi" code in Linux is actually just workaround for Intel Macs above and beyond normal efi handling. [17:58] cjwatson, heh, thats exactly what i have in my oem-config.conf here too ... but didnt start up the automatic mode it seems [17:59] well, not using automatic-oem-config but automatic-ubiquity here atm [17:59] i suspect the various levels of wrappers swallow it somewhere [18:00] * ogra_ gets new coffee and will then track that down [18:02] * jodh -> foooooood [18:03] cjwatson: Only two questions. [18:03] cjwatson: In the case where you throw your hands up in the air and exit 0, what if I didn't actually intend to boot from a non-GPT disk? [18:04] * bdmurray takes lunch [18:04] cjwatson: And should the debconf templates about BIOS/EFI boot partitions perhaps hint at an appropriate size for them? I'd have no clue what to do if I got that dialog. [18:05] barry, I added a test case to reproduce bug 792652 in a VM (vbox in my case) but I reproduce it more easily on a crappy laptop. [18:05] Launchpad bug 792652 in ubiquity "ubiquity crashed with ValueError in command(): I/O operation on closed file" [High,Incomplete] https://launchpad.net/bugs/792652 [18:05] infinity: Unfortunately, at the partitioning stage, we can't know. grub-installer will fail later, which I think is regrettably the best we can do in this case. [18:05] infinity: I can add size advice, sure. [18:06] barry, also, while reproducing it I got bug 629321 with similar steps but on first step of the installer. [18:06] Launchpad bug 629321 in ubiquity "ubiquity crashed with AttributeError in debconffilter_done()" [Undecided,Confirmed] https://launchpad.net/bugs/629321 [18:06] cjwatson: Kay. The answer to the former is unfortunate, but such is life. The shell looks otherwise sane to me. Add a bit telling me WTF goes in a boot partition (or just tell me how big it should be), and I'll be happy. [18:07] oh, as for what goes in the boot partition, the "EFI boot partition" and "Reserved BIOS boot area" terms there are things you can select in "Use as:" [18:07] jibel: cool. i want to spend a little more time on my current bug, then i'll look at these again. i think we have a general problem that it's pretty easy to click on buttons that get the front-end out of sync. [18:07] cjwatson: Well, you tell me, I guess ("boot loader code"), but that doesn't mean much to most people. :) [18:07] perhaps I need to spell that out? [18:07] jibel: but we'll probably have to fix them one-by-one [18:07] cjwatson: Spelling out the type/use-as a bit more might be nice. I felt it was obvious, but I understand all of this to start with. [18:08] cjwatson: But size is really the glaring omission. 512 bytes, several blocks, several MB, who knows? ;) [18:09] infinity: http://paste.ubuntu.com/892553/ maybe? [18:10] cjwatson: That looks much better, yeah. [18:10] cjwatson: Also, 35MB? What the heck goes in there? [18:10] cjwatson: (I assume the 1MB one is just a stage2 bootloader or something) [18:10] minimum size for fat32 ... [18:10] That's not true. Limitation of dosfstools? [18:11] parted [18:11] (You could format 1.44MB floppies with fat32, and even ntfs, if you felt the stupid desire) [18:11]