[05:19] Well well well. I got accessibility working from wubi, with a little command-line hackery. Need to try it without hackery first, to make sure it isn't a flook.! :) [06:10] TheMuso: you managed to fix the command line getting truncated at 256 bytes bug? [06:11] evand: Not quite. I switched the access=* around with the -- [06:11] The -- means nothing, so as long as access=* comes before that, we should be fine., [06:11] My understanding was that anything after the -- was carried over to the installed system. [06:11] Just confirmed it without touching a thing. Gets truncated at access= [06:11] indeed, I haven't found a fix for that yet. [06:12] Hrm right. [06:12] Its probably an internal kernel thing I'm guessing. [06:12] I believe it's grub4dos, as the problem does not exist outside of Wubi. [06:13] Alternatively, have you considered putting a lot of the stuff thats currently on the command line, into a file, instead of on the command-line? Stuff like locale, keyboard, etc. [06:13] Right. [06:13] indeed grub4dos has the kernel command line statically set to 256 bytes, but we patched it to match regular grub. The problem still exists however. [06:13] Yeah you said as much. [06:13] TheMuso: I imagine those options are carried over from when Wubi used the alternate CD [06:14] But that would be a nice temporary workaround. [06:14] evand: Right, because thats another option, and I think wubi should detect what CD is being used, and act accordingly/// [06:14] I don't believe it even uses the alternate CD anymore. [06:14] Ah ok. [06:15] Makes sense. [06:16] I'll mention the idea of moving the locale and keyboard into the preseed file to Ago tomorrow (er, later today :) ) [06:17] Ok, because I think its our best bet, as they really don't need to be on the command-line, IMO. [06:17] indeed, though I'd still like to fix that bug as it's eventually going to bite us again, I imagine. [06:18] Likely enough, yes. [07:32] cjwatson, could you tell me what part of d-i is causing this stub to come up? http://imagebin.org/14534 [07:32] we don't need language support in our alternate disk (and languages are not in our seeds), so it'd be more ideal for that to not show [08:34] superm1: that's pkgsel; preseed pkgsel/install-language-support to true or false (I assume false) to get rid of it [08:36] 256 bytes> odd, the kernel has #define COMMAND_LINE_SIZE 2048 [09:20] cjwatson: is there something missing that prevents using relatime by default? didn't we decide that it would be good to have?-) [09:26] mostly, a technical difficulty inside partman with regard to picking any default mount options [09:27] might be able to do something though [09:31] cjwatson: partman-base? [09:32] no, a mix of other partman-* [09:32] p.s. please don't reassign stuff to plain 'partman' - it doesn't exist [09:33] I'll have a look at it [09:33] yeah, there was some bug recently.. [09:34] I know, I've got it here [09:34] ah so it was exactly this one :) [09:34] duh [09:34] * tjaalton handles too many bugs.. [09:36] anyway, thanks for having a stab at it [10:15] user-setup: cjwatson * r99 ubuntu/ (debian/changelog functions.sh): [10:15] user-setup: * Make is_system_user always return false if OVERRIDE_SYSTEM_USER is set [10:15] user-setup: (LP: #153310). [10:19] user-setup: cjwatson * r100 ubuntu/debian/changelog: releasing version 1.16ubuntu5 [10:25] oem-config: cjwatson * r424 oem-config/ (debian/changelog scripts/user-setup-wrapper): [10:25] oem-config: * Allow user creation even if a non-system user already exists (thanks, [10:25] oem-config: Torsten Spindler; LP: #153310). [10:56] * cjwatson blinks [10:56] 'EXT3-fs: Unrecognized mount option "relatime" or missing value' [10:56] I thought we had all the support we needed? [10:57] oh, maybe mount is supposed to translate it [11:06] what this the max strlen for the kernel commandline options? [11:10] seems to be 2048, but I still get a truncated string at 256, evand had a patch for grub4dos [11:11] will need to recheck my testing rig [11:13] as far as the kernel itself goes I believe it is 2048; there are several places along the way it could be truncated though [13:15] partman-partitioning: cjwatson * r666 ubuntu/ (debian/changelog free_space/new/do_option): [13:15] partman-partitioning: * When creating a new filesystem, set the default mount options to the [13:15] partman-partitioning: contents of /lib/partman/mountoptions/${fs}_defaults if it exists [13:15] partman-partitioning: (LP: #160450). [13:16] partman-partitioning: cjwatson * r667 ubuntu/debian/changelog: releasing version 54ubuntu3 [13:32] debian-installer: cjwatson * r892 ubuntu/debian/changelog: * No-change rebuild to pick up new components (busybox 1:1.1.3-5ubuntu12). [13:34] partman-basicfilesystems: cjwatson * r556 ubuntu/ (mountoptions/ext2_defaults debian/changelog): * Default to mounting ext2 with relatime (LP: #160450). [14:06] the kernel itself is 4096 on x86, iirc. [14:07] partman-basicfilesystems: cjwatson * r557 ubuntu/debian/ (changelog partman-basicfilesystems.templates): * Add Choices-C entries for mount options. [14:11] do you rekon it is better to try to fix the truncation or do a workaround to shorten the cmdline? [14:12] I mean, of course the first option is better, but not sure what sort of work is involved (wrt "several places"). [14:13] I think we should go with the workaround anyway (as discussed in PM, we're not using the alternate CD), but continue to try and find the source of the bug. [14:14] locale and layout can be preseeded in the preseed file on the desktop CD. The accessibility stuff needs to remain on the kernel cmdline though. [14:14] As it's part of casper. [14:16] I'd need "initrd" preseeding, in the sense that the settings are read in casper-premount off hard disk [14:17] Do not think that any question is really needed before casper-premount [14:17] might even be possible to pass accessibility stuff in there [14:18] Also in lupin, we might want shorter aliases for "iso-scan/filename" and "debian-installer/custom-installation" [14:18] I'd suggest "isoscan" and "c-i" [14:20] The accessibility stuff has to be on the kernel cmdline, aiui. I don't think shorter aliases are necessary. Moving the locale and layout preseeded questions to a file should be sufficient. [14:21] In my case locale/layout/variant are also in preseed, the preseed file is copied to initrd / during casper-premount and the options are imported in casper-bottom (iirc) [14:22] ithat basically should be +/- equivalent to initrd preseeding [14:22] casper could process the accessibility stuff when reading the preseed file [14:22] yes, that is something I was thinking [14:23] putting the locale in the preseed file doesn't actually work right now, I think, but could be made to [14:23] as mentioned the preseed file is first seen in casper-premount but processed in casper-bottom [14:24] I would definitely recommend finding the source of the truncation; that causes all kinds of irritating problems [14:24] I'm confused, are you saying we modify casper to do such a thing? Because the accessibility stuff isn't a debconf question. Also, I thought the locale in the preseed file worked as we're using the desktop CD and the preseed file doesn't get read until very late anyway. [14:24] syslinux should allow at least 512 chars [14:24] evand: oh, good point [14:24] indeed, I shared that thought last night. [14:24] (not in debconf) [14:25] I know it will bite us eventually :) [14:25] I could grep out the relevant values from the preseed or another file and set things separately in pre-mount [14:26] As it stands we're only two characters shy of fitting the whole thing, I don't think drastic measures are needed. Stripping out the debconf preseeds should be more than sufficient. [14:26] cjwatson locale preseeding does not work, what about keyboard layout/variant? [14:26] do I need commandline too for those [14:27] evand I think we need a bit more than 2 chars, particularly if we want accessibility in there [14:28] we now also have "noprompt" in there [14:28] ah, I forgot about that. Still, that's 11 bytes. The debconf questions still cover that. [14:29] the removal of them, that is. [14:30] evand can you have a quick wubi run and let me know what things should be removed from commandline and what added to preseed? [14:30] will do [14:30] if you press esc at boot you also have options such as vebose mode ans safe graphic mode that may add a few bytes [14:31] relevant files are /ubuntu/install/boot/grub/menu.lst and /ubuntu/install/custom-installation/preseed.cfg [14:32] tjaalton: my suspicion is that it's grub4dos' fault. syslinux works fine, but wubi doesn't make use of it. [14:32] ah [14:36] cjwatson: so you suggested that I was correct about the accessibility stuff, but is there a problem with preseeding locale in a file for ubiquity that I'm not seeing? [14:37] evand: ah, right [14:38] evand: locale is configured in 14locales; preseed file is read in 24preseed and doesn't duplicate the work [14:38] it will preseed it successfully for ubiquity, probably, but not for the running live system [14:39] ah, hrm. [14:47] will send an email to bean123/tinybit [14:48] hmm, partman-auto/purge_lvm_from_device doesn't seem to work anymore, what's the replacement? [15:13] The longest option is 298 bytes. Removing console-setup/layoutcode and console-setup/variant code brings us down to 243. [15:20] will upload the changes tonight [15:22] thanks [15:26] cjwatson, thanks. would you mind merging that from the debian-cd branch at http://bazaar.launchpad.net/~mythbuntu/debian-cd/mythbuntu-debiancd/ [15:26] then? [15:30] which reminds me, I need to dig for those instructions now that I have my IRC server back up and running. [15:30] superm1: done [15:30] thanks [15:31] i might have a few more task related questions, but we'll see how the next daily pans out now [16:00] cjwatson: what are your thoughts on having a wubi download link on ubuntu.com? [16:01] personally I like the idea [16:01] as mentioned at UDS, I'd be glad to donate/redirect wubi-installer.org domain [16:01] as does xivulon :) [16:01] ...and xivulon wife (most importantly)... [16:02] heh [16:11] oem-config: evand * r425 oem-config/ (debian/changelog lib/frontend/gtk_ui.py lib/zoommap.py): * Port zoommap fixes from ubiquity. [16:18] ubiquity: evand * r2537 ubiquity/ (d-i/manifest debian/changelog debian/control): [16:18] ubiquity: * Automatic update of included source packages: choose-mirror [16:18] ubiquity: 2.19ubuntu2, console-setup 1.21ubuntu4, localechooser 1.42ubuntu2, [16:18] ubiquity: partman-base 114ubuntu3, partman-basicmethods 38ubuntu1, partman-jfs [16:18] ubiquity: 25, partman-partitioning 54ubuntu3, partman-target 54ubuntu1, user- [16:18] ubiquity: setup 1.16ubuntu5. [16:42] ubiquity: evand * r2538 ubiquity/debian/changelog: releasing version 1.7.15 [16:44] ubiquity: evand * r2539 ubiquity/ (configure configure.ac): Bump to 1.7.16 [16:45] cjwatson: any objection to me putting out a new oem-config? [16:45] go ahead [16:48] thanks, will do [16:49] oem-config: evand * r426 oem-config/ (d-i/manifest debian/changelog): [16:49] oem-config: * Automatic update of included source packages: console-setup [16:49] oem-config: 1.21ubuntu4, user-setup 1.16ubuntu5. [17:03] dpkg-reconfigure xserver-xorg does not seem to prompt for the video driver anymore, is this a known change? [17:04] #ubuntu-x [17:04] that sort of thing is entirely up to the package's maintainer scripts [17:06] thanks, I thought it might've been a debconf thing because specifying -plow didn't change anything. actually, specifying -pfoo had the same effect and didn't give any errors [17:09] it's not a debconf problem [17:09] -plow doesn't invent questions that the maintainer scripts don't ask. :-) [17:11] oem-config: evand * r427 oem-config/ (configure configure.ac): Bump to 1.32 [17:11] oem-config: evand * r428 oem-config/ (debian/changelog gui/glade/step_timezone.glade): [17:11] oem-config: * Use a GtkEventBox rather than a GtkScrolledWindow for the tzmap to [17:11] oem-config: match ubiquity (LP: #198179) [17:12] cjwatson: did you miss the question about a link to wubi on the ubuntu.com site, or is it something that you need to give some thought? [17:26] I don't feel strongly; newz2000 would be the one to ask [17:30] indeed, had already planned on it, just wanted to run it by you first [17:30] ok, thanks [17:31] xivulon: if you could start a conversation with Matthew Nuzum for what you'd like to have done and CC me on it, I'd very much appreciate it. [17:31] will do [17:32] thanks [20:10] debian-installer: cjwatson * r893 ubuntu/debian/changelog: releasing version 20070308ubuntu32 [20:14] I'd be happy to help change the accessibility code to read from the preceed file if that makes things any easier... [20:16] partman-basicfilesystems: cjwatson * r558 ubuntu/debian/changelog: releasing version 56ubuntu3 [20:20] partman-ext3: cjwatson * r731 ubuntu/debian/ (changelog partman-ext3.templates): * Add Choices-C entries for mount options. [20:20] partman-ext3: cjwatson * r732 ubuntu/ (mountoptions/ext3_defaults debian/changelog): * Default to mounting ext3 with relatime (LP: #160450). [20:20] partman-ext3: cjwatson * r733 ubuntu/debian/changelog: releasing version 49ubuntu2 [20:21] partman-jfs: cjwatson * r719 ubuntu/debian/ (changelog partman-jfs.templates): * Add Choices-C entries for mount options. [20:21] partman-jfs: cjwatson * r720 ubuntu/ (mountoptions/jfs_defaults debian/changelog): * Default to mounting jfs with relatime (LP: #160450). [20:21] partman-jfs: cjwatson * r721 ubuntu/debian/control: Ubuntu maintainer address [20:22] partman-jfs: cjwatson * r722 ubuntu/debian/changelog: releasing version 25ubuntu1 [20:22] partman-reiserfs: cjwatson * r798 ubuntu/debian/ (changelog partman-reiserfs.templates): * Add Choices-C entries for mount options. [20:22] partman-reiserfs: cjwatson * r799 ubuntu/ (mountoptions/reiserfs_defaults debian/changelog): * Default to mounting reiserfs with relatime (LP: #160450). [20:22] partman-reiserfs: cjwatson * r800 ubuntu/debian/control: Ubuntu maintainer address [20:23] partman-reiserfs: cjwatson * r801 ubuntu/debian/changelog: releasing version 39ubuntu1 [20:23] partman-xfs: cjwatson * r751 ubuntu/debian/ (changelog partman-xfs.templates): * Add Choices-C entries for mount options. [20:23] partman-xfs: cjwatson * r752 ubuntu/ (mountoptions/xfs_defaults debian/changelog): * Default to mounting xfs with relatime (LP: #160450). [20:24] partman-xfs: cjwatson * r753 ubuntu/debian/control: Ubuntu maintainer address [20:25] partman-xfs: cjwatson * r754 ubuntu/debian/changelog: releasing version 35ubuntu1 [20:30] I'm having trouble customizing the live system so it has both the correct time and the correct timezone. [20:31] Simply fixing the /etc/localtime link within filesystem.squashfs and deleting the casper-bottom/02timezone script (and rebuilding the ramdisk) results in the correct timezone, but the wrong actual time -- until I run `ntpdate-debian -u' from the running live system. [20:37] TheMuso: Thanks, but I don't think it will be necessary and I'd like to avoid making any large changes like that as part of the workaround. [20:38] evand: Fair enough, I'm of the same thought as well, but if there is no other options... [20:45] TheMuso: indeed :) [20:46] evand: Had a chance to do any dmraid testing? [20:47] TheMuso: thought you might ask. That's my plan for all of this evening leading up to the meeting. I only just got settled in and unpacked yesterday. [20:48] evand: Fair enough, I'll be around if you need me. [20:48] great [20:50] As for getting it into hardy however, thats another story. The support folks don't feel that its something that they want to support, especially since it may give Ubuntu more of a bad rep t han anything. The example given was someone using fakeraid who has/gets a support contract with Canonical... [20:54] Correct me if I'm wrong, but surely the OS just sees a normal disk when using fakeraid? [20:55] twb: No thats not the case at all. [20:56] twb: Fakeraid is made up with a BIOS that writes special metadata to the hard disks, and in Windows' case, drivers that read that metadata and transparently handle the RAID functionality. [20:56] if that were the case, it would be easy :-/ [20:56] cjwatson: Exactly. [20:58] from everything that's been said, I'm coming round to deferring it [20:58] which is a shame, but we *did* start on it kind of late, and didn't promise anything [21:00] Yeah its a shame, as I do have a slight vested interest in it also. I am tracking upstream progress however, and will keep everybody posted, and will see what we can do for intrepid. [21:01] Oh well. Fakeraid is pretty silly when you had md. [21:02] twb: Yes, but just about every consumer/prosumer/low-end server board comes with it these days, and most people don't know any; better. [21:02] from the point of view of specifying a system, yes; from the point of view of having a fakeraid system already and trying to put Ubuntu on it ... [21:03] And yes, there is what cjwatson said regarding dual booting/installing Ubuntu onto an existing fakeraid. [21:03] I'll find the master bug about dmraid support and pass it over to you (remind me if I don't); it could do with an "official"-looking update on status [21:03] So ubiquity supports md and lvm now? [21:03] cjwatson: Ok. [21:07] evand: Nevertheless, it would be interesting to see how those fakeraid BIOSs you have behave, as well as any Windows tools that may be available for them. [21:07] As we could at least mention something about different controllers. [21:11] twb: no [21:12] but nor was it planned for it to support dmraid in this cycle [21:12] (the discussion above was just for dmraid support at all, and in d-i) [21:12] Fair enough. [21:12] it's on my list for the 8.10 UDS [21:12] TheMuso: indeed, I still plan on investigating this. [21:12] UDS? [21:13] Ubuntu Developer Summit [21:13] LVM and RAID in ubiquity is? [21:13] dmraid is [21:13] ah, indeed [21:15] Well, I "fixed" my clock grief issue by running ntpdate-debian -s -u at the end of /etc/rc2.d. That means something in rcS.d or rc2.d is doing something dumb w.r.t. the clock, and (bizarrely) ntpd -g isn't fixing it. [21:24] cjwatson: I'd be happy to help/draft a spec for dmraid if need be. [21:26] that would be good