[00:01] <cjwatson> in that case I'm confused
[00:01] <persia> It's rather that the location pull-down defaults to the "nearby" places for a given language, and so one fiddles with the map for a bit when one isn't there.  I thought r2836 miht have instead defaulted to the open world map.
[00:01] <cjwatson> (the change above fixes half of the problems with an incorrect keyboard default in ubiquity, the other half being fixed in casper)
[00:02] <persia> Oh.  Completely unrelated then.
[00:02] <cjwatson> oh, no, I haven't touched the map lately, although I do agree it's problematic
[00:02] <persia> I'm not sure it's not generally correct to default to a nearby location, but it's perhaps a little tricky when the guess isn't close.
[00:03] <cjwatson> English, Spanish, and Portuguese are the hardest cases. The defaults are much better for just about anything else.
[00:05] <persia> I would have expected French to be in that list as well
[00:06] <cjwatson> oh, Canada. Yes. The population distribution is *much* more heavily skewed towards the default of France there, though.
[00:06] <cjwatson> actually Portuguese isn't hard because pt and pt_BR are treated as different languages.
[00:07] <cjwatson> there are lots of American-continent Spanish speakers though.
[00:07] <persia> Actually, I was thinking of West Africa more than Canada.  I'd expect Canada to have the same ease as for Brasil.
[00:08] <cjwatson> nobody seems to have created any locales for West African French speakers
[00:08] <cjwatson> fr_CA isn't a different language to quite the same extent as pt_BR, and we don't treat it as such
[00:08] <cjwatson> there's a slightly different vocabulary of course, but pt_BR is totally divergent
[00:08] <persia> It's colonial French, and it's the same timezone anyway.  Most people in West Africa are at least trilingual.
[00:09] <cjwatson> oh, sure, not arguing, it just isn't reflected in the locale data and I don't want to get into maintaining that :)
[00:09] <cjwatson> realistically my job is a heck of a lot easier if I can pretend that locale data reflects reality and punt to them when it doesn't ;-)
[00:10] <persia> heh.  You must not be that much in favour of the move to have only a single translation for all Arabic then.
[00:11] <cjwatson> as far as the installer's concerned, there already is
[00:11] <cjwatson> we only have ar.po, not ar_EVERYTHING
[00:11] <persia> Wasn't that recent?  I thought that the various Arabic translation teams only organised to a single .po file in the last year or so.
[00:12] <persia> Or is the single file installer-specific?
[00:12] <cjwatson> the latter
[00:12] <cjwatson> installer translations are mostly done in Debian
[00:12] <cjwatson> for ubiquity, we only import the translations that match what Debian has, since otherwise we get a crappy mixture
[00:13] <persia> Yeah.  Translations are tricky, and it's best to have a single authoritative source (whatever happens to be right for a given project).
[02:34] <CIA-50> partman-ext3: cjwatson * r740 ubuntu/debian/changelog: releasing version 52ubuntu2
[02:35] <CIA-50> partman-reiserfs: cjwatson * r805 ubuntu/debian/changelog: releasing version 41ubuntu2
[02:38] <CIA-50> yaboot-installer: cjwatson * r262 ubuntu/debian/changelog: releasing version 1.1.12ubuntu2
[02:48] <CIA-50> ubiquity: cjwatson * r2837 ubiquity/bin/ubiquity-dm: make sure to only start gconfd for GTK-based frontends
[02:50] <CIA-50> ubiquity: cjwatson * r2838 ubiquity/debian/changelog: releasing version 1.9.18
[04:52] <CIA-50> partman-base: TheMuso * r105 ubuntu/ (debian/changelog lib/base.sh lib/commit.sh):
[04:52] <CIA-50> partman-base: * Merge some updated dmraid partitioning changes, thanks to Frans Pop
[04:52] <CIA-50> partman-base:  <fjp@debian.org>.
[04:52] <CIA-50> partman-base:  - base.sh: new function is_multipath_dev.
[04:52] <CIA-50> partman-base:  - commit.sh: display correct dmraid partition info when confirming
[04:52] <CIA-50> partman-base:  changes.
[04:53] <CIA-50> partman-base: TheMuso * r106 ubuntu/debian/changelog: releasing version 121ubuntu6
[14:22] <CIA-50> ubiquity: cjwatson * r2839 ubiquity/ (82 files in 3 dirs):
[14:22] <CIA-50> ubiquity: Make "Before:" and "After:" strings (displayed next to partition bars)
[14:22] <CIA-50> ubiquity: translatable.
[18:02] <nebuchadnezzar> hello
[18:03] <nebuchadnezzar> is this the good place to ask about partman recipes ?
[18:04] <cjwatson> nebuchadnezzar: yes
[18:05] <nebuchadnezzar> Ok, I have some problem defining a recipe for hardy: http://pastebin.com/m5dc29b58
[18:05] <nebuchadnezzar> this works but: I have 2 extended LVM partition instead of one primary sda4 LVM
[18:06] <nebuchadnezzar> and the volume groupe name is the host name, not testvg :-/
[18:06] <nebuchadnezzar> any hints ?
[18:06] <cjwatson> you're using Debian documentation rather than Ubuntu documentation
[18:06] <cjwatson> don't do that :-)
[18:06] <cjwatson> the documentation you're reading describes a feature of LVM preseeding that is very new and isn't present in Ubuntu yet (not even intrepid, let alone hardy)
[18:07] <nebuchadnezzar> erf
[18:07] <cjwatson> specifically, in_vg and lv_name aren't supported yet
[18:09] <cjwatson> however, it doesn't look like you're doing anything complicated enough to require the new syntax, unless you really care about the LV names
[18:10] <cjwatson> what you should do is the following:
[18:10] <cjwatson>  * drop all vg_name{ ... }, in_lv{ ... }, and method{ lvm } instructions
[18:10] <cjwatson>  * make sure all the partitions you want to be on LVM have $lvmok{ }
[18:11] <cjwatson>  * preseed 'd-i partman-auto-lvm/new_vg_name string testvg' to set the VG name
[18:11] <cjwatson> and that should be enough
[18:12] <cjwatson> oh, and one other thing
[18:12] <nebuchadnezzar> ok
[18:12] <cjwatson>  * entirely drop the stanza that defines the LVM wrapper partition (the one that just has method{ lvm } and vg_name{ testvg })
[18:13] <cjwatson> the installer will create one of those automatically for you
[18:13] <nebuchadnezzar> ok
[18:13] <nebuchadnezzar> I think it's the reason why I have 2 extended partitions
[18:13] <cjwatson> I would also advise you to drop $primary{ } from everything except the /boot partition; nothing else needs to be primary, and it's better to have fewer primaries
[18:14] <cjwatson> yes
[18:15] <nebuchadnezzar> well, logicall partition are harder to find with gpart after a crash ;-)
[18:15] <cjwatson> really? gosh, that's lame :-)
[18:15] <cjwatson> (of gpart)
[18:15] <nebuchadnezzar> when the partition table goes away
[18:16] <nebuchadnezzar> that the reason I always have /boot / and swap on primary, it's easyear when recovering to have /etc easylly accessible ;-)
[18:16] <nebuchadnezzar> ok, thanks for you advices cjwatson
[18:17] <nebuchadnezzar> I pass a long time trying to guess what the problem
[18:18] <cjwatson> even with this gpart problem, there's no need to recover swap that way; two primary partitions leaves you a lot more flexible than three
[18:19] <cjwatson> but really, gpart ought to be able to recover logical partitions too; the only thing that's different about logical partitions is that their partition table lives in a non-fixed position on the disk, and the whole point of gpart is for when you can't rely on the partition table :-)
[18:20] <cjwatson> just means you might have to create an extended partition table in order to put them back
[18:20] <cjwatson> sorry to harp on about it, but the four-primary-partitions limit drives me nuts, because I've had to write code that deals with it and it's a pain in the neck
[18:20] <nebuchadnezzar> extended partition are chained, so if you miss one you can have some problems
[18:20] <cjwatson> so I like to do my bit for advocacy
[18:21] <cjwatson> if you only have four partitions to start with, there's no reason to have a chain of extended partition tables
[18:21] <nebuchadnezzar> cjwatson: yes, I think having only LVM could be cool
[18:21] <cjwatson> I believe we support /-on-LVM
[18:22] <nebuchadnezzar> yes yes yes
[18:22] <cjwatson> anyway; you might find my general advice to use Ubuntu documentation rather than Debian documentation useful in other areas too. Partitioning isn't the only place where we're a little different for one reason or another
[18:23] <nebuchadnezzar> yes, but googleing partman recipe give near only debian references
[18:23] <cjwatson> https://help.ubuntu.com/8.04/installation-guide/i386/ - the only problem I know of there is that that where it says partman-auto/purge_lvm_from_device you need to use partman-lvm/device_remove_lvm instead
[18:24] <cjwatson> googling for "partman recipe Ubuntu" gives some but I agree they aren't as well-linked and the proper link is on the second page. I'll have to do something about that
[18:24] <nebuchadnezzar> ok, I found only this recipe reference: http://d-i.alioth.debian.org/svn/debian-installer/installer/doc/devel/partman-auto-recipe.txt
[18:24] <nebuchadnezzar> which is, I think, far from being complete to understand the partman process
[18:25] <cjwatson> that's the detailed recipe reference from Debian
[18:25] <cjwatson> and for unstable at that, not etch
[18:30] <nebuchadnezzar> ok, it works better now, thanks, but the LVM partition is still a logicall one, not a big issue in fact ;-)
[18:30] <cjwatson> yeah, it will be. It seems unlikely that you will be able to use gpart to recover an LV ;-)
[18:31] <cjwatson> (and, seriously, backups if it's that much of a problem.)
[18:31] <cjwatson> or UPS, or ...
[18:32] <nebuchadnezzar> yes I know, I just dislike logical partition in fact ;-)
[18:32] <nebuchadnezzar> thanks a lot
[18:34] <cjwatson> I've just made http://people.ubuntu.com/~cjwatson/d-i/doc/devel/partman-auto-recipe.txt available, which is the Ubuntu equivalent of the URL you mentioned above
[18:34] <cjwatson> (note that it is for intrepid but I don't believe there's a significant difference at this point)
[18:39] <nebuchadnezzar> ok, I'll put my example on a web page with a good name to be indexed by search engine :-)
[18:42] <nebuchadnezzar> one question I found no answer, is it possible to concatenate 2 recipes ? for example a base recipe, for /boot / and the LVM, an extended conditionnal recpie ? I have different "flavor" I can choose when booting the CD
[18:42] <nebuchadnezzar> all could use the base recipe and it's own extended one
[18:42] <nebuchadnezzar> not sure to make my point
[18:49] <cjwatson> nebuchadnezzar: you can generate a recipe using a preseed/early_command script, write it to a file, and then use partman-auto/expert_recipe_file to point to it
[18:49] <nebuchadnezzar> ok thanks
[18:50] <nebuchadnezzar> I'll play with all of this for now
[18:50] <nebuchadnezzar> see you
[20:36] <Zelut> is there a way to specify a mirror location for the netinstall within PXE?
[20:36] <Zelut> or do I need to point it to a ks/preseed with that defined?
[20:46] <cjwatson> Zelut: mirror/country=manual mirror/http/hostname=archive.ubuntu.com mirror/http/directory=/ubuntu
[20:49] <Zelut> cjwatson: you're the man
[21:14] <hardwire> anybody used wubi to boot an initrd and kernel prepared for remote rescue?
[21:17] <hardwire> and does it just use ntldr?
[23:13] <bdmurray> evand: somebody has commented on bug 38442 regarding alpha 6 and the new partioner dialog
[23:14] <cjwatson> I'll ask him to file a new bug
[23:17] <kirkland> cjwatson: working on bug bdmurray just filed, http://people.ubuntu.com/~brian/tmp/cryptmount-message.png
[23:17] <kirkland> cjwatson: the log_*_msg is easy to solve in cryptroot, needs a ". /scripts/functions"
[23:18] <kirkland> cjwatson: the other one, /sbin/udevsettle ....
[23:18] <kirkland> cjwatson: i'm looking at initramfs-tools and I see in TODO:  udevsettle timeouts handling
[23:18] <cjwatson> you want Keybuk for that
[23:18] <kirkland> cjwatson: is adding /sbin/udevsettle to initramfs complicated?
[23:18] <kirkland> cjwatson: k
[23:19] <cjwatson> I have never quite got my head around the exact correct situations to use udevsettle; all I know is that it isn't obvious and I tend to guess wrong
[23:19] <kirkland> cjwatson: and simply silencing that /sbin/udevsettle in 2>/dev/null is not correctly, probably?
[23:19] <cjwatson> absolutely not!
[23:19] <cjwatson> never throw away errors
[23:19] <cjwatson> (well, not usually, anyway)
[23:20] <cjwatson> err, are you sure you don't just need to change it to /sbin/udevadm settle?
[23:20] <cjwatson> udev<lots> got replaced by a single binary
[23:21] <cjwatson> udevsettle is just transitional to that, and it may be that it's not in the initramfs any more because the initramfs ought to be easy to get everything up to date
[23:39] <bdmurray> cjwatson: great, thanks!