[10:27] <mpt> Interesting, no bugs have ever been reported about Ubiquity
[10:27] <mpt> https://launchpad.net/products/ubiquity/+bugs :-)
[10:31] <mpt> Perhaps bugs from /distros/ubuntu/+source/ubiquity will need to be registered there once popular derivatives are using Ubiquity
[12:15] <siretart> all ubiquity bugs are ubuntu specific. there are no ubiuity upstream bugs ;)
[12:23] <cjwatson> mpt: I have enough trouble managing them in one place without managing them in two ;-) the product/distro separation is often a pain
[12:24] <cjwatson> I'd love to have bugs (optionally) transparently gatewayed from one to the other
[12:24] <cjwatson> mpt: (also, I only bothered to create the ubiquity product relatively recently, IIRC)
[12:24] <cjwatson> when I needed a directory on the supermirror
[12:26] <cjwatson> ahh, that's better. The version of ubiquity in bzr now supports just 'ubiquity --debug' (and sudos or whatever for itself) rather than needing 'gksudo env UBIQUITY_DEBUG=1 ubiquity' or 'kdesu --nonewdcop env UBIQUITY_DEBUG=1 ubiquity' depending on the frontend
[12:27] <cjwatson> main push for that was so that I could also support 'ubiquity --new-partitioner'
[12:49] <mpt> cjwatson, yeah, the idea was to prevent upstream from getting pestered with bug reports about distro-specific customizations
[12:49] <mpt> Ubiquity (and Upstart, I guess) are unusual cases where Ubuntu == upstream
[12:51] <cjwatson> hence (optionally)
[12:52] <mpt> right
[01:43] <mpt> Reported "Handle a distribution being its own upstream for a package" <https://launchpad.net/bugs/76416>
[02:25] <cjwatson> mpt: thanks
[04:54] <mark> expert_recipe is a bitch
[05:15] <tepsipakki> whee
[05:15] <cjwatson> tepsipakki: how much free space is there on the disk?
[05:15] <cjwatson> it's possible it's a 64-bit arithmetic issue
[05:16] <tepsipakki> nope, it's a laptop, cfdisk says ~32GB
[05:16] <tepsipakki> partman-auto: Expert recipe too large (2200008150 > 32003)
[05:16] <tepsipakki> that first figure puzzles me
[05:17] <tepsipakki> calculating the miminal sizes gives me 9650
[05:17] <tepsipakki> s/miminal/minimal/
[05:18] <tepsipakki> oops
[05:18] <tepsipakki> argh
[05:19] <tepsipakki> maybe it was just a typo after all.. every entry should end with a dot, right?-)
[05:19] <tepsipakki> I've added specifiers for lvmok{ }, and misplaced one
[05:19] <cjwatson> ah, that could be it. it's unfortunately very fragile
[05:20] <mark> quite!
[05:20] <mark> has just cost me another hour to find out that I had a space where it didn't like it
[05:20] <cjwatson> the size is in megabytes, so I guess my lack of safe 64-bit arithmetic isn't a big problem
[05:21] <cjwatson> won't manifest until 4PB or so
[05:21] <tepsipakki> it's also unnecessarily duplicated at places
[05:21] <mark> is there a way to ehm, create a partition, format it with filesystem and mount it... *except* if that partition with the correct size already exists?
[05:21] <tepsipakki> method{ format } and format{ } etc
[05:21] <mark> then it should just keep the data on it
[05:21] <cjwatson> it's very much tied to the representation in /var/lib/partman/devices/
[05:22] <cjwatson> mark: I can't think of one. There are several things for which I'd like some kind of conditional syntax in recipes.
[05:22] <mark> it's a complicated problem
[05:22] <mark> especially with RAID and LVM and all that ;)
[05:24] <mark> autoinstallations are a bit scary for our database servers right now
[05:24] <tepsipakki> oh, and now that I'm here lets just say that I'll try to fix some of the bugs that are dear to me ;)
[05:24] <tepsipakki> in d-i
[05:25] <mark> what does edgy support with LVM, in terms of auto partitioning?
[05:27] <tepsipakki> hmm, I don't remember anymore.. didn't use edgy much but I recall it did support auto-lvm
[05:28] <cjwatson> tepsipakki: please do; http://wiki.ubuntu.com/InstallerDevelopment
[05:28] <cjwatson> I'm taking on more management responsibility in Canonical, so my time for installer work is going to be reduced (although not eliminated altogether)
[05:29] <tepsipakki> yeah, I've been looking at netcfg-mess :)
[05:30] <cjwatson> heh, I was just merging that
[05:30] <mark> congratulations I guess? ;)
[05:30] <cjwatson> what's bugging you about netcfg?
[05:30] <cjwatson> mark: maybe ;-)
[05:31] <cjwatson> I manage about half the distro team now
[05:31] <mark> nice
[05:31] <tepsipakki> cjwatson: well, the dhcp/fqdn-stuff
[05:31] <mark> yes! :)
[05:32] <mark> that sucks a bit
[05:32] <cjwatson> hmm, that's not bitten me much
[05:32] <cjwatson> but netcfg has been short on maintenance over the last year or so since joshk got busy with school
[05:32] <tepsipakki> I get "xxx.xxx.xxx.xxx host.domain.domain host.domain" in /etc/hosts, if dhcp gives fqdn as the hostname
[05:32] <cjwatson> ah
[05:33] <mark> my wishlist:
[05:33] <mark> - a way to preseed a "search" line for resolv.conf
[05:33] <mark> - a way to use the ip as requested by DHCP configured statically
[05:33] <mark> the latter might be a bit dangerous but makes sense in server environments
[05:34] <tepsipakki> mark: the latter can be done
[05:34] <mark> I am currently doing it
[05:34] <mark> in early_command
[05:34] <tepsipakki> ah
[05:34] <mark> preseeding the ip and then rerunning netcfg
[05:34] <tepsipakki> yes
[05:34] <cjwatson> a custom dhclient script might be able to do that
[05:34] <cjwatson> though it would be a lot more perverse than doing it directly in netcfg
[05:35] <mark> # Redo network configuration statically
[05:35] <mark> echo d-i netcfg/get_ipaddress string $(ifconfig | grep "inet addr" | cut -d ' ' -f 12 | sed 's/addr://' | grep -v 127\.0\.0\.1) > /tmp/st
[05:35] <mark> atic_net.cfg
[05:35] <mark> debconf-set-selections /tmp/static_net.cfg
[05:35] <mark> killall.sh; netcfg; true
[05:35] <mark> it's a bit dirty but it seems to work
[05:35] <cjwatson> yow
[05:35] <cjwatson> but yes, that ought to work ...
[05:36] <cjwatson> killall.sh is so badly named it isn't funny
[05:36] <mark> agreed :)