jibelstgraber, bug 800561 regressed on latest daily.09:51
ubot2`Launchpad bug 800561 in libxklavier "No way to add other keymap than english on Live CD" [Wishlist,Confirmed] https://launchpad.net/bugs/80056109:51
=== amichair is now known as amichair_
* ogra_ bangs his head against the oem-config colorchange 12:28
ogra_its so weird, if i run it manually i get proper colors from oem-config-remove ...12:37
ogra_hmm, i wonder if we actually run the debconf version of oem-config-remove or if we get into a weird gtk/aptdaemon fallback mode12:43
stgraberjibel: hmm, how did it regress exactly?13:26
stgraberjibel: currently you should only get the keyboard indicator when in greeter mode13:26
stgraberjibel: as all the other modes went through gfxboot and we don't want to loose the selection the user made back then13:26
ogra_hmpf ... so setting DEBIAN_FRONTEND=text in oem-config-remove makes the install hang completely after it shows 50% on serial13:27
* stgraber grabs current daily to check exactly what's happening with the keyboard layout13:31
jibelstgraber, for example, select french in the greeter, the indicator lists french layout (fr, ca, ...) but no other layout can be added, 'us' for example, the '+' button is disabled.13:46
stgraberjibel: ah right, that's normal13:51
stgraberjibel: you need to remove some before you can add some more13:51
stgraberjibel: X can only support 4 layouts at a time13:51
stgraberjibel: the change I pushed yesterday was to implement that limit properly13:52
stgraberjibel: before that you'd see all of them but only the first 4 could be selected13:52
stgraberwhich was even more confusing13:52
stgraberso I now added some kind of logic to try and guess a good list of 4 entries for each language, we'll have to live with that for now until we can somehow get statistical data on layout usage to instead use the top-4 layouts for a given language13:53
* jibel re-reads the bug report13:53
stgraberso currently the expected (from my point of view) behaviour is to only see the indicator in greeter mode, selecting a language will give you a maximum of 4 layouts and if you need some more, you'll need to remove one of them in the preferences and add another one13:55
stgraberI agree that the dialog isn't that good at explaining the limitation though, might be worth a bug for the desktop team to improve that a bit :)13:55
jibelstgraber, ok that's bad UI of gnome-control-center13:55
stgraberyeah, that's an old X limitation that gnome chose to respect (they probably could have done their own thing and not use the alternate layout xatoms) but didn't make clear to the user in their UI...13:57
jibelyep, I refreshed my memory with comment #11. UI improvement but low prio IMO.14:02
stgraberjibel: can you file a bug against gnome-control-center for the UI improvement? I don't feel like reusing the current bug report as we've been using for multiple bugs already ;)14:05
GrueMasterstgraber: Bug 567200 is still valid if the slideshow shows openoffice/libreoffice.14:59
ubot2`Launchpad bug 567200 in ubiquity-slideshow-ubuntu "On Netbook Edition installer shows apps that are not installed" [Undecided,Invalid] https://launchpad.net/bugs/56720014:59
stgraberogra_: ^ :)15:00
ogra_we dont even have had the "netbook edition" in several releases15:01
ogra_and we use the same slideshow (if at all) that x86 uses15:01
ogra_i'm 100% sure its gone15:02
stgraberright, I'll make it clear that the slideshow is gone so there isn't anything to fix :)15:04
* ogra_ roars in direction of oem-config 15:06
GrueMasterI didn't know slideshow was gone (I know the netbook edition is).15:08
ogra_GrueMaster, even if it isnt, we use the same one x86 uses15:09
ogra_there is no separate package for it or anything ... (in netbook there was iirc)15:09
GrueMasterNo there wasn't.  But if we are using the same slides as x86, and it contains images of LibreOffice, then it is wrong (hence my point).15:10
ogra_why, you can always install LibO15:12
GrueMasterI thought the slideshow was supposed to reflect apps on the image?15:15
GrueMasterTo be honest, I rarely pay attention to the slideshow anymore.  Too busy multitasking.15:17
GrueMasterAnd I don't even see it in the current manifest, so nevermind.15:17
GrueMasterIt is in the x86 images though.15:18
GrueMaster(maybe that should be a bug?)15:19
pgranerstgraber, do you have any idea of how to use the $resumemethod{} from a preseed to preserve an existing partition?15:39
stgraberpgraner: I only had to do it once, in this preseed http://www.stgraber.org/download/preseeds/rl-desktop-home.seed15:46
stgraberpgraner: in that example, the /home partition won't be formated and will be mounted in the target system15:46
stgraberusing method{ keep }15:46
pgranerstgraber, thx let me see if I can get this working!15:48
stgraberI'm not really familiar with partman's internals but I wrote that preseed for a customer before I joined Canonical and AFAIK they're still using it and it didn't eat any of their /home yet15:48
pgranerstgraber, I have to preserve a vfat partition for arm since the bootloader and other goodies are there15:49
pgranerand it has to be the first on the disk15:49
ogra_pgraner, we used to have partman-uboot for that ... but NCommander never got to port it to panda afaik15:50
ogra_but thats exactly hoding the reciepe for keeping the first partiton intact15:51
pgranerogra_, yea I'm trying to keep from having to install to a USB stick as /dev/sda1 and then munging the kernel and initrd on the mmc, then undoing it all so it will pxe boot when I'm done testing15:51
ogra_pgraner, though i know that GrueMaster worked out proper preseeds (he is really good at that) for netboot installs on pandas15:51
ogra_that also work withough the partman-uboot package15:52
pgranerogra_, ack15:52
ogra_for the PXE stuff you can just dump a script into d-i's late_command that mangles your bootloader setup btw15:53
GrueMasterActually, the partman fat bug still exists.  It can probably be worked around in preseeding, but I have dedicated usb-sata disks for my test systems (far faster anyways).  I use the SD card for u-boot and kernel only (and yes, I can fully pxe boot at will).15:54
GrueMasterHere's my preseed:  http://paste.ubuntu.com/886494/ .  Note that it points to my local mirror, but that is trivial.  It also installs openjdk-6-jre-headless for jenkins to take control on reboot.16:00
stgraberjibel: your installer terminal not updating is overlayfs' fault18:13
stgraberjibel: I duped it to bug 882147 and added a comment for that other inotify use case18:13
ubot2`Launchpad bug 882147 in linux "overlayfs does not implement inotify interfaces correctly" [High,Triaged] https://launchpad.net/bugs/88214718:13
jibelstgraber, ack, that was my guess too.18:16
* jibel wonders about the usefulness of this terminal18:16
stgraberwell, I think it's mostly useful to show we're doing something18:17
stgraberso at the moment it's more than useless as it makes the installer look ineficient :)18:17
GrueMasterskaet: Do you have everything you need for the release manifest for armadaxp?  afaik, we are only doing a netboot armhf+armadaxp with 5yr support under ubuntu server (which is listed on the wiki).21:45
stgraberbdmurray: I just finished rewriting the lp-integration script, should be much easier to understand and change now21:47
stgraberbdmurray: I'm doing a last run, then will replace the old one in the branch21:47
stgraber(that's the launchpad integration script for the ISO tracker)21:47
darienMy company has been using an unattended Ubuntu 8.04 LTS ISO for a year (which I built); I've now built a 10.04 LTS ISO as well, and included some custom packages and updates into it; however, one package I added yesterday is causing a dependency error which, near as I can tell, shouldn't happen.21:50
darienSpecifically, there's no reason I can tell why it would try to install mysql-common, but in the installer syslogs it gives me 'percona-server-common: conflicts with mysql-common', even though I've never specified mysql-common to be installed and nothing that I can see depends on it21:50
darienso my question is, where can I dig more into what it's actually trying to install (and why) during the install-packages phase of the installation?21:51
darienmost of the actual installer documentation (from an end-user customization point of view) is out of date, or doesn't provide any detail into the actual inner workings of the installer, which I feel like I need to understand in order to debug this21:52
stgraberbdmurray: one change is that I'm no longer looking for comments from the qatracker account in the history as that wouldn't work with multiple tracker instances as we now have23:41
stgraberbdmurray: instead I'm posting the comment before tagging the bug, if multiple tags are associated with a tracker instance, the bot will post if any tag is missing23:42
stgraberbdmurray: so the current way of avoiding the bot messing with a bug is to add bot-stop-nagging and then do any change you want to the bug23:42

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!