[12:32] <tjaalton> I'm trying to debug why the live-cd does not create a xorg.conf, which should be done in xserver-xorg.preinst
[12:33] <tjaalton> is it because ': > "$XORGCONFIG"' does not work in the environment?
[12:34] <tjaalton> I don't know if it works or not :)
[12:38] <cjwatson> the preinst thing is limited to run on fresh installs only which isn't the case when casper reconfigures it
[12:39] <cjwatson> I don't know what you guys intend to do in the reconfiguration case
[12:39] <tjaalton> oh right
[12:40] <tjaalton> well, if the file is missing it assumes that's the way the user wants
[12:55] <cjwatson> stupid ISP
[12:55] <cjwatson> 12:50 <cjwatson> casper doesn't delete it
[12:55] <cjwatson> 12:51 <cjwatson> ah, but livecd-rootfs does
[12:55] <cjwatson> 12:51 <cjwatson> feel free to suggest a better approach
[12:56] <cjwatson> tjaalton: I missed anything you said after "well, if the file is missing it assumes that's the way the user wants"
[12:56] <cjwatson> ultimately, we want to force a reconfiguration from scratch on the user's system
[12:56] <tjaalton> heh, I didn't add anything after that
[12:57] <tjaalton> but I'll think about it
[13:19] <thom> cjwatson: can i usefully preseed hw-detect/select_modules to load additional modules? or is there a better way?
[13:23] <cjwatson> thom: no, it's not preseedable, and the only thing you could do by preseeding it anyway would be to reduce the set of modules that's loaded (given it defaults to everything in its multiselect list)
[13:25] <thom> hrm
[13:25] <cjwatson> thom: easiest way would probably be to do 'register-module <modulename>' in a preseed/early_command script
[13:26] <cjwatson> though, hmm, I wonder if that's early enough
[13:27] <cjwatson> no, that just causes it to be loaded in the installed system
[13:28] <thom> the problem is that i'm trying to use partman-auto-raid, and dm-mod isn't getting loaded, so devicemapper goes *boom*; i just wanna load dm-mod and friends
[13:28] <cjwatson> meh, surely that's a p-a-r bug
[13:29] <thom> surely it is
[13:29] <thom> but i need a quick workaround :)
[13:29] <thom> i'll verify that's the fix and file a bug after
[13:29] <cjwatson> d-i preseed/early_command string modprobe dm-mod
[13:29] <cjwatson> ?
[13:30] <thom> tried that; it's too early
[13:30] <cjwatson> mm
[13:30] <thom> early_command happens before md-modules gets downloaded
[13:33] <cjwatson> d-i preseed/early_command string (echo '#! /bin/sh'; echo 'modprobe dm-mod') > /lib/partman/init.d/00dm_mod_hack; chmod +x /lib/partman/init.d/00dm_mod_hack
[13:33] <cjwatson> ?
[13:33] <thom> i'll give it a go
[13:33] <cjwatson> I don't *think* it needs further quoting
[13:40] <thom> what it does need sadly is an mkdir -p first; rebooting :)
[13:42] <cjwatson> surprising, since partman should be unpacked by that point
[13:42] <cjwatson> oh, no, it shouldn't
[13:43] <cjwatson> sorry, got the order wrong again :)
[13:43] <thom> no problem mate
[13:43] <thom> i owe you much beer for all the help at some point anyway
[13:46] <thom> also, i'd forgotten how awful wsvn is
[16:34] <evand> cjwatson: Does this look acceptable to you? http://pastebin.ubuntu.com/2487/
[16:36] <cjwatson> evand: looks pretty good, though I think you still need to read all the output from the PARTITIONS command even if you aren't using it
[16:37] <evand> oh, fair enough.  I'll fix and then commit.
[16:37] <evand> thanks
[16:37] <cjwatson> the idiom I generally follow is to assign the stuff you need to another variable and then just continue round the loop, e.g. root_size="$size"
[16:37] <evand> ok
[16:38] <cjwatson> have you considered summing the sizes of /, /usr, /var or something like that as an alternative heuristic?
[16:38] <cjwatson> not that I want to bikeshed, just wondering ...
[16:38] <cjwatson> perhaps a refinement would be to make it a warning and let people continue if they have filesystems other than /
[16:39] <evand> I hadn't, but that could be tricky as they could have a ridiculously small / and a large /var
[16:39] <evand> ok, that sounds reasonable
[16:39] <cjwatson> yeah, it's not going to be perfect
[16:39] <cjwatson> don't let me bikeshed you into not doing this though - I'd far rather have your proposal than nothing at all
[16:40] <evand> I'll give it some more thought
[16:40] <evand> but yes, this can be an iterative process.
[17:06] <evand> cjwatson: also, once this is implemented, what are your thoughts of getting rid of ubiquity/text/partition_help?  This should account for the space warning, and the partitioner already warns you if you don't specify a swap partition (though admittedly it doesn't specify a size).  I remember us discussing removing this at FOSSCamp, but I don't recall what the conclusion was.
[17:06] <evand> I recall "yes", but I want to be sure before I go ripping anything out.
[17:13] <cjwatson> evand: it should either be removed or replaced by something that pops up much more detailed help (e.g. partman-target/help if appropriate)
[17:13] <cjwatson> evand: so +1
[17:14] <evand> ok
[18:39] <mgunes> Hi, I'm triaging a bug and need to know whether the desktop CD boot menu strings are translated as part of the debian-installer package
[18:44] <CIA-4> ubiquity: evand * r2375 ubiquity/ (3 files in 2 dirs):
[18:44] <CIA-4> ubiquity: * Added partman/check.d/03root_too_small to warn the user when their root
[18:44] <CIA-4> ubiquity:  partition isn't large enough to contain /rofs plus a bit (LP: #48355).
[18:51] <evand> mgunes: gfxboot-theme-ubuntu
[18:54] <mgunes> evand, is the "Help" text displayed when hitting F1 in there? I've finally been able to get the PO tarball for debian-installer, and it seems to be in it
[18:59] <mgunes> OK, just figured out: basic menu strings are in gfxboot-theme-ubuntu, and "Help" is in debian-installer, thanks
[19:02] <evand> no problem
[19:09] <evand> cjwatson: hrm, perhaps I should add a boot_too_small check, given bug #69165 .  Would du -s /boot * (estimate number of kernel releases) be reasonable, or do you think I should just hardcode a full size in there?
[19:10] <evand> This seems slightly more difficult to get right.
[19:10] <evand> argh, ubotu
[19:10] <evand> https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/69165
[19:11] <evand> cjwatson_: let me know if your network connection caused you to miss my comments a moment ago.
[19:12] <evand> also, any objection to me dragging ubotu in here?
[19:27] <evand> https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/136665 still sounds like a GTK bug to me.  Plenty of things have can_focus set and tabbing works fine on them.  I'm tempted to implement seb's solution anyway though, as it works.
[19:28] <evand> cjwatson_: as you originally handled the bug, are you ok with that?
[20:39] <cjwatson_> evand: I suspect I missed your comments
[20:39] <cjwatson> evand: feel free to drag in ubotu
[20:40] <cjwatson> evand: sounds like can_focus means something slightly different on a combobox; hard to say. I think implementing Seb's solution is fine provided that it's still possible to tab *to* the combo box.
[20:41] <evand> it is, and will do on both
[20:41] <evand> 14:10:42 < evand> cjwatson: hrm, perhaps I should add a boot_too_small check, given bug #69165 .  Would du -s /boot * (estimate number of kernel releases) be reasonable, or do you think I should just hardcode a full size in there?
[20:41] <evand> 14:10:51 < evand> This seems slightly more difficult to get right.
[20:41] <evand> 14:11:04 < evand> argh, ubotu
[20:41] <evand> 14:11:09 *** Netsplit over, joins: mpt
[20:41] <evand> 14:11:13 < evand> https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/69165
[20:42] <cjwatson> hmmm. not obvious.
[20:42] <evand> indeed
[20:43] <cjwatson> so
[20:43] <cjwatson> for any partition that has been assigned, it should be at least as big as the corresponding subdirectory of /rof
[20:43] <cjwatson> /rofs
[20:43] <cjwatson> (minus anything that's covered by other partitions)
[20:44] <cjwatson> in the case of /boot, we know it needs to have room to grow a bit, so yeah, maybe multiply by three or something
[20:45] <evand> ok
[20:46] <evand> I suppose I should generalize all of this into a single question, so s/root/partition/g.
[20:47] <evand> that is, rename the existing check.d test.
[20:47] <cjwatson> yeah, that's what I think I was driving at
[20:48] <evand> ok, will do
[20:48] <cjwatson> wouldn't be possible for d-i, but should be possible for ubiquity
[20:48] <evand> thanks
[20:48] <cjwatson> or at least, would be prohibitively slow for d-i
[20:48] <evand> heh, d-i, slow?  Never.
[20:49] <cjwatson> there's reasonably slow and unreasonably slow :)
[20:49] <evand> haha, fair enough
[22:31] <CIA-4> ubiquity: evand * r2376 ubiquity/ (3 files in 2 dirs): * Modified scripts/install.py to handle ENOSPC.