[01:25] xnox: debconf is certainly not optional for anything remotely related to the installer; it's a core facility [01:25] However, debconf-set is not part of debconf :-) [01:27] ubiquity: cjwatson * r5485 trunk/debian/ (changelog ubiquity.install-any): Install debconf-set from debian-installer-utils, useful for preseeding. [07:52] cjwatson: aha =) [08:49] mpt: https://wiki.ubuntu.com/Ubiquity/AdvancedPartitioningSchemes [08:49] mpt: https://picasaweb.google.com/105922848292507689403/Beginning?authuser=0&feat=directlink [08:49] mpt: https://picasaweb.google.com/105922848292507689403/SingleDiskStandardInstall?authuser=0&feat=directlink [08:50] please note this is just the first slideshow: the beginning steps & typical full-disk install [08:51] the screenshots do not always show every single click/tab, usually the default highlight is: "save option" (no, cancel, back, etc), simply the top option, rarely something useful (e.g. default to english keyboard). [09:38] xnox, great, thanks. Will you be around in 3 hours? What time zone are you in? [09:38] mpt: I am UK based. I will be around in 3 hours. [09:39] mpt: I have a meeting at 15:00 UK time (GMT+1 right now?!) [09:39] ok [09:40] mpt: I still need to do slideshows for the other cases. Were the first two ok? [09:40] as in would you still prefer video? [12:38] xnox, I don't see Raid or LVM in either of those galleries [12:38] xnox, yes, slideshows would be fine [12:39] mpt: I have added 'guided lvm' & crypt slideshows to the https://wiki.ubuntu.com/Ubiquity/AdvancedPartitioningSchemes [12:39] * mpt wonders wtf "relidaving" means in "Relidaving gnupg..." [12:39] RAID is still to do [12:40] mpt: link to all galleries: https://picasaweb.google.com/105922848292507689403 [12:42] thanks [12:51] xnox, have you met (or talked to) Dustin Kirkland? [12:52] mpt: I know who he is. I have seen him at the UDS. I believe he is not aware who I am. [13:09] xnox, I'm still reading this stuff, but meanwhile, bug 791885 and bug 817507 look like they might be duplicates. [13:09] Launchpad bug 791885 in ubiquity "Ubiquity crash with RAID 0 array - grub could not install" [Undecided,Confirmed] https://launchpad.net/bugs/791885 [13:10] Launchpad bug 817507 in ubiquity "Grub-install fails on the RAID0 array" [Undecided,Confirmed] https://launchpad.net/bugs/817507 [13:10] ok thanks. [13:10] * xnox is considering to rename wiki page to Ubiquity/ThisStuff =) [13:11] "will not have swap or will use a swapfile" [13:12] cjwatson, iirc there's been a long-standing to-do to switch from using a swap partition to a swap file by default. Is that right? [13:13] mpt: yes, but it's not massive priority. The questions about hibernate/suspend are still not fully resolved, afaik. [13:15] Indeed, it's something we'd eventually like to do to simplify partitioning; hopefully this LTS cycle but it's not scheduled for 12.10 [13:15] Given previous experience I'd not want to do it by default without explicit support time allocated from the kernel team [13:16] cjwatson: alternative & server d-i cd's are build from debian-installer package, ubiquity or somewhere else? [13:18] They contain objects from the debian-installer source package, but the ISO images themselves are built by a branch of debian-cd [13:25] xnox, do you have handy a screenshot of the current advanced partitioning step, so I can design something that doesn't require the new overall design? [13:32] cjwatson: ok. [13:33] mpt: you mean ubiquity's advanced partitioning? [13:33] xnox, yes [13:33] * xnox had to reboot due to X.org crashing. sorry for the delay. [13:35] mpt one moment [13:42] mpt: https://picasaweb.google.com/105922848292507689403/Ubiquity?authuser=0&feat=directlink [13:43] there are a few [13:43] eugh, there's something gone strange with that "Install this third-party software" checkbox [13:44] mpt: I think you need to allow 'download updates' before that one becomes active [13:44] no [13:44] you don't have to [13:44] it's simply 'gone ugly in a VM' [13:46] xnox, looks like a few of those dialogs have coped badly with being ported to GTK3 [13:46] "Create a new partition" in particular [13:46] and "Create a new empty partition on this device?" [13:46] +table [13:46] mpt: =/ this is precise release image, not quantal's gtk3 [13:46] in a VM [13:46] odd [13:47] I will try bare-metal boot and take screenshots again to check the quality [13:47] xnox, no, Ubiquity in 12.04 uses GTK3 [13:47] oh ok. [13:48] * xnox was thinking about python3 (sorry) which is being ported in quantal [13:48] * xnox it must be friday [13:50] xnox, which varieties of Raid does the alternate installer currently allow? [13:50] mpt: I remember that cjwatson was saying something along the lines, that it should be possible to ctrl+click on devices and then "combine" them into RAID/LVM or convert it into encrypted volume. [13:50] Choices: RAID0, RAID1, RAID5, RAID6, RAID10 [13:50] mpt: let me check, I think 0, 1, 5, 6. Plus it allows nesting raids to do 1+0 and 0+1. [13:52] My notes say "RAID 0 isn't really RAID at all, but a close alternative to LVM (bug 43453), though it is possible to run one on top of the other." [13:52] Launchpad bug 43453 in ubiquity "live cd partitioner doesn't understand lvm properly" [Medium,Confirmed] https://launchpad.net/bugs/43453 [13:52] Does that match your understanding? [13:53] It's not redundant, but it's also not best understood as an alternative to LVM [13:54] In what situations would you choose one or the other? [13:54] It's a method of aggregating multiple block devices into one by alternating blocks among the disks [13:54] You might use RAID0 if you just want a very low-performance-impact way to aggregate a bunch of disks together [13:55] So is LVM slower than RAID0? [13:55] Let me finish :) [13:55] sorry [13:56] What you get out of it is effectively a single partition whose size is the sum of all the components; but it still roughly behaves as a simple traditional partition [13:58] LVM is more a swiss army knife approach: you feed it a load of physical volumes (partitions) and it lets you create logical volumes on top without having to care very much about which disks they happen to live on; if for example you later need to replace a disk, you can say to LVM "please move all my data off this disk I'm about to remove" and then swap in a new one [13:58] I don't think there's a major performance difference, but RAID0 is very much simpler [13:58] xnox, no RAID 10? [13:58] RAID10 is there, I was quoting the code above [13:59] RAID 10 is same as RAID 1+0. FIrst you create RAID 1 from 2 disks (I) then another RAID1 from another 2 disks (II). Then assemble both I & II into RAID0. [13:59] (4 disks in total) [14:00] * xnox stacking like a matroshka doll ... [14:00] similarly you can do reverse, e.g. RAID 0+1 when you do it the other way around =) [14:01] http://en.wikipedia.org/wiki/RAID_0 has some example applications [14:02] I do think it's kind of niche, but mostly it's there because there's not much cognitive difference between four choices and five, and it is something people occasionally ask for so I don't think it's worth the rather minor optimisation of removing it and making those people either redesign their workflow or go and assemble arrays by hand [14:06] thanks cjwatson [14:12] xnox, if you want something to do right now, I suggest choosing which thing you want to add first: (a) LVM, (b) RAID 0, (c) other RAID types. Then add a button at the bottom right of the partition table, for now labelled (a) "LVM Setup…", (b) "RAID 0 Setup…", or (c) "RAID Setup…", that opens an intro dialog. [14:12] xnox, then on Wednesday I'll finish the design of the intro dialog. [14:12] (a) LUKS (crypt) [14:13] (b) LVM [14:13] I was leaving out LUKS until you've talked with Dustin :-) [14:13] because he's been working on full-disk encryption, and I don't know whether it's the same kind [14:14] no it's not the same [14:14] and it would be annoying if we ended up with two kinds of full-disk encryption, neither of which you could switch to after installation [14:16] I think it'd be crazy to do full-disk encryption other than via cryptsetup [14:17] (i.e. LUKS) [14:17] No idea why Dustin wants to reinvent that wheel in a slightly different shape [14:17] * xnox away on a call [14:25] any objections to a ubiquity upload? [14:25] None from me, was mostly holding off in case you had more to do with m-a [14:26] nope [14:26] sorting that now then [14:26] fixes one critical so we definitely should [14:26] ta [14:27] ubiquity: evand * r5486 ubiquity/ (d-i/manifest debian/changelog): [14:27] ubiquity: Automatic update of included source packages: base-installer [14:27] ubiquity: 1.122ubuntu8, flash-kernel 2.28ubuntu43. [15:04] ubiquity: evand * r5487 ubiquity/debian/changelog: releasing version 2.11.2 [15:43] cjwatson: is it possible there was a bug in /etc/default/grub handling? https://bugs.launchpad.net/ubuntu/+source/ubuntu-meta/+bug/995555/comments/3 [15:43] Launchpad bug 995555 in ubuntu-meta "package ubuntu-standard 1.267 failed to install/upgrade: ErrorMessage: dependency problems - leaving unconfigured" [Low,Fix committed] [15:43] cjwatson: for CMDLINE_LINUX=" quiet splash" e.g. with a space in the beginning? [15:43] Well, that isn't actually a bug [15:44] I mean, having the space there [15:44] Oh, but you mean if that's what the file is like on the way in? [15:45] Hi again :) What should I use in a preseed's partman/early_command on Ubiquity in lieu of debconf-set? <- this doesn't seem to work on Ubiquity, only d-i [15:46] I fixed that in ubiquity 2.11.2. You had quit IRC so I couldn't tell you at the time. [15:46] However, you can use this instead: [15:46] cjwatson: oh! damn time zone differences :/ [15:46] . /usr/share/debconf/confmodule; db_set partman-auto/disk /dev/sda [15:47] If that's part of a longer script, make sure that ". /usr/share/debconf/confmodule" is the very first thing in the script [15:47] cjwatson: yes, it's a small script, I'll do that [15:48] cjwatson: this solution is what we'll use probably, since we need it to work in older releases, so the 2.11.2 fix may not be available on those :9 [15:48] Sure [15:48] cjwatson: thanks :) as usual [15:48] bdmurray: Certainly there are multiple known bugs in /etc/default/grub handling anyway; it's a mess