[10:02] <CIA-61> console-setup: cjwatson * r88 ubuntu/ (Keyboard/ckbcomp debian/changelog):
[10:02] <CIA-61> console-setup: Set plain or shifted PrintScreen key to generate VoidSymbol rather than
[10:02] <CIA-61> console-setup: Ctrl-backslash (LP: #279973).
[10:11] <CIA-61> console-setup: cjwatson * r89 ubuntu/debian/changelog: releasing version 1.28ubuntu4
[12:29] <cjwatson> kirkland: should bug 120375 still have hardy tasks open, given that it seems to have been addressed in bug 290885? (I'm trying to clean up the 8.04.2 bug list a bit)
[14:49] <kirkland> cjwatson: hi there, welcome back...  120375 should be done.  i'm closing it now.
[14:49] <cjwatson> cool, thanks
[14:50] <kirkland> cjwatson: the server guide documentation for hardy should be fixed though, as one needs to run grub-install manually on /dev/md0 to get the bootloader installed on both disks
[14:50] <cjwatson> server guide rather than installation guide?
[14:51] <cjwatson> I assume in the installer it Just Works
[14:51] <kirkland> cjwatson: correct
[14:51] <kirkland> cjwatson: only an issue on upgrades
[14:51] <cjwatson> ok, I've never touched the server guide myself :)
[14:51] <kirkland> cjwatson: i'll take care of it
[14:52] <kirkland> cjwatson: you might enjoy this ...  http://1.bp.blogspot.com/_-mej0A6dVeU/SUXo_OoBNwI/AAAAAAAAALk/OgleC83i6_Q/s1600-h/ubuntu.png
[14:53] <kirkland> cjwatson: i created a new package, screen-profiles, with a couple of advanced screenrc profiles, and a "select-screen-profile" binary
[14:53] <cjwatson> yow, bright :)
[14:53] <kirkland> cjwatson: full blog post at http://blog.dustinkirkland.com/2008/12/ubuntu-server-includes-window-manager.html
[14:53] <kirkland> cjwatson: if you've got any neat "screen hacks" you're willing to share, ping me at some point about them :-)
[14:54] <cjwatson> the only interesting thing in mine is:
[14:54] <cjwatson> bind d eval 'exec keychain --quiet --clear' detach
[14:54] <cjwatson> bind ^d eval 'exec keychain --quiet --clear' detach
[14:54] <kirkland> oh, that's a good one
[14:54] <cjwatson> i.e. clear cached ssh/gpg tokens on detach
[14:54] <kirkland> yeah, totally
[14:55] <cjwatson> keychain is pretty slow to start up though - I'd prefer a saner reimplementation if anyone had one
[15:11] <kirkland> cjwatson: it's a good idea, i'll add it to my list
[15:11] <kirkland> cjwatson: comments are flowing into that blog with some really good ideas
[20:36] <Ahmuck> hi.  i'm interested in discussing some installer enhancements for the user
[20:36] <Ahmuck> if anybody is interested in listening
[20:49] <evand> cjwatson: What are your thoughts on maximizing ubiquity by default, and on the Install Ubuntu option fullscreening it?  My reasoning is that in some places you benefit from more space but it isn't immediately apparent, such as the partitioning page and timezone map especially.
[20:49] <evand> I mentioned this a few times at UDS and people seemed to be positive towards the idea.
[20:50] <Ahmuck> mazimizing ?
[20:51] <Ahmuck> k, i'll repeat some of my questions here.  a unified installer for both the alternate and the gui based installs would be nice.  currently under the alternate, you enter info, it does it's thingy, and then you do more info entry.  the gui install only asks for info once and then completes the install
[20:52] <evand> can you elaborate on what you mean by unified?
[20:52] <Ahmuck> secondly, is there a way to get the install to check the repository for updates and install the updates rather than the cd while installing?  this would save having to do it later.  or at least as an option
[20:52] <Ahmuck> evand: the same
[20:52] <Ahmuck> so the install process is the same
[20:53] <Ahmuck> iirc, the alternate asks for x info in the beginning and y info towards the end
[20:54] <evand> Ahmuck: what information is it asking you after installing packages?
[20:55] <Ahmuck> hrm, i'd have to go through an install and screenshot it, but i'd be willing if i thought i could get the order changed
[20:55] <Ahmuck> or at least have someone look at it
[20:55] <evand> I'm just curious.  My understanding was that it's been a long long time since you had a stage two to the alternate CD
[20:55] <Ahmuck> er, consider looking at it?  let me get some screenshots togather
[20:56] <Ahmuck> evand: long time?  as in 8.04.1 ?
[20:56] <Ahmuck> hrm, let me do a test in a vm and see
[20:56] <evand> 8.04.1 is already out
[21:48] <cjwatson> evand: early on, Mark explicitly asked for it not to be maximised so that people would be encouraged to use facilities of the live CD as well. I think it would be fine to full-screen it in only-ubiquity mode, though
[21:49] <cjwatson> Ahmuck: yes, it does ask for some things later, and we have been gradually working on moving them back. However for a variety of reasons this will *not* consist of a "unified installer" (at least not significantly more than it does right now; they do share quite a bit of code)
[21:49] <evand> ok, noted
[21:50] <cjwatson> Ahmuck: checking the repository for updates is one of those things where we can't win; some people want it one way and some the other. Could you file a bug on https://launchpad.net/ubuntu/+source/debian-installer/+filebug to have it switchable, though? I think that would be a good idea.
[21:51] <chorn> I'm having an issue with kernelparms being ignored as bootcd/installer options.  Is that a question for here or in #ubuntu-kernel ?
[21:51] <cjwatson> chorn: depends which parameters
[21:51] <chorn> rootdelay
[21:51] <chorn> as long as i have rootdelay=360 or so it should work fine
[21:52] <chorn> but it's not having any effect
[21:53] <cjwatson> chorn: so, firstly, I assume you aren't talking about the server/alternate installer, whose initrd pays no attention to rootdelay
[21:53] <cjwatson> chorn: apart from that, rootdelay= appears to be parsed by initramfs-tools, using this code:
[21:54] <cjwatson>         rootdelay=*)
[21:54] <cjwatson>                 ROOTDELAY="${x#rootdelay=}"
[21:54] <cjwatson>                 case ${ROOTDELAY} in
[21:54] <cjwatson>                 *[![:digit:].]*)
[21:54] <cjwatson>                         ROOTDELAY=
[21:54] <cjwatson>                         ;;
[21:54] <cjwatson>                 esac
[21:54] <cjwatson>                 ;;
[21:54] <chorn> right, this is a stock 8.10 desktop
[21:54] <cjwatson> now, that code seems to do the right thing when I run it by hand
[21:55] <Ahmuck> recently someone asked me to try fedora, and the fedora installs very quickly.  it appears to "copy" the live cd over.  is there a way to speed up the ubuntu installer?
[21:55] <cjwatson> chorn: I would suggest booting with 'debug=console rootdelay=360' to see if the extra trace output clarifies anything
[21:55] <cjwatson> Ahmuck: that's exactly what the Ubuntu live CD installer does
[21:55] <chorn> something isn't quite right with it, i had a dapper server install working on it w/ just rootdelay as a tweak
[21:55] <cjwatson> Ahmuck: we can't do that with the alternate installer because that would remove precisely the flexibility that is the *point* of the alternate installer
[21:56] <Ahmuck> cjwatson: i haven't run test times yet
[21:56] <chorn> I have not tried debug=console!
[21:56] <Ahmuck> well, in the live cd, i was comparing install times
[21:56] <cjwatson> I haven't heard anyone else object to the speed of the Ubuntu live CD's installer, I must say
[21:57] <cjwatson> there can sometimes be problems if our kernel puts the CD drive into the wrong mode or something
[21:57] <cjwatson> chorn: make sure to put debug= before rootdelay= or else it won't debug the parsing of rootdelay= itself ...
[21:57] <Ahmuck> cjwatson: thx for the reply on the unified installation.  i assume the code is too different to allow for a more unified installer.  i was really looking at it from a user's perspective, and having them both ask the same questions at the same point in the install
[21:57] <chorn> done, waiting for busybox
[22:00] <cjwatson> Ahmuck: yeah, again we can't make them identical without removing some of the flexibility that the alternate installer needs to deliver
[22:00] <Ahmuck> ah, ok
[22:09] <cjwatson> there are some cases where we've avoided questions after copying packages (which, in the alternate installer, would go after the base system is installed or after bulk package installation - basically things that require software to be installed in the target system before they can even decide which questions to ask!) in the live CD installer simply by removing configurability
[22:10] <cjwatson> there are a lot of trade-offs involved ...
[22:10] <cjwatson> I'd like to pull the questions involved in configuring apt earlier, though
[22:10] <cjwatson> that's feasible, it's just that nobody's done the work yet
[22:11] <cjwatson> if you compare with warty you'll notice a fairly big difference in how early questions are asked ;-)
[22:17] <chorn> hrm
[22:18] <chorn> ROOTDELAY='300'
[22:18] <chorn> but it isn't actually doing it
[22:18] <chorn> i can put in rootdelay=2000 and it's the same
[22:18] <cjwatson> it should get passed as a timeout to usplash_write
[22:18] <cjwatson> and it should affect the loop timeout
[22:19] <cjwatson> try without the splash argument so that you can see better
[22:19] <chorn> i remove quiet and splash
[22:19] <cjwatson> you aren't using boot=<anything>?
[22:19] <chorn> nope
[22:19] <cjwatson> I'm looking at /usr/share/initramfs-tools/scripts/local here
[22:19] <cjwatson> you might want to do the same
[22:19]  * cjwatson goes to change baby
[22:19] <chorn> k!
[22:35] <chorn> odd, dmesg | more in busybox doesn't do anything
[22:38] <chorn> thanks for the help, i'm going to leave this until tomorrow
[22:40] <cjwatson> wow, my daughter is such a drama queen :)
[22:40] <cjwatson> right
[22:40] <cjwatson> you can try just 'debug' and then look in /tmp/initramfs.debug if you prefer that to a console dump
[22:41] <cjwatson> (don't ask me, I didn't design the parameters ...)
[22:41] <cjwatson> in 9.04 I moved it to /dev/.initramfs/initramfs.debug so that it can be available after boot if the system gets that far
[22:50] <chorn> I see "Begin: Mounting root filesystem... ..."
[22:50] <chorn> but i don't find that string in my current server's initramfs-tools
[22:51] <chorn> ah, it's in init
[22:53] <chorn> and i'm guessing /scripts/casper is only for the installer
[22:53] <chorn> my server only has casper-premount
[23:10] <cjwatson> oh, maybe scripts/casper doesn't implement rootdelay
[23:10] <cjwatson> yeah, apparently not
[23:29] <chorn> hrm
[23:29] <chorn> is there any reason it shouldn't?