[00:00] what are you in, if not in the live CD environment? [00:00] I think I'm confused [00:00] ubiquity --automatic [00:00] no desktop [00:00] I simply started xterm and was playing off that [00:02] tz in ubiquity defaults to NY [00:02] ah but it otherwise it works!!! [00:03] so it must be the preseed file [00:03] I did not use it at all [00:04] tring live desktop + preseeding [00:05] reproduced, hacking [00:07] yes when I preseed things go wrong even when layoutcode & co are commented [00:13] I can change wubi code while you hack to comment out locale/layoutcode/variantcode and add them as boot parameters [00:13] if not 'UBIQUITY_AUTOMATIC' in os.environ: [00:13] self.db.fset('console-setup/layout', 'seen', 'false') [00:13] self.db.fset('console-setup/variant', 'seen', 'false') [00:13] evand: why was that again? [00:14] that's what's causing this ... [00:14] ago: please do that regardless [00:14] will do [00:14] ago: there's still a ubiquity bug to fix here, but we definitely won't be able to make it work with the preseed file approach to those parameters [00:15] if it works with boot params it's perfectly ok [00:23] hmm what do I do with rev 488 (updated metalink url)? shall I revert that? [00:23] cjwatson do you think this edit will go in rc or not? [00:23] which edit? [00:24] moving layoutcode from preseed to boot args [00:25] I don't know the up-to-the-minute state of CD builds, ask Steve [00:27] sorry i haven't followed entirely, but just saw the tail end of your conversation. as it stands, are keyboard layouts still not working through the preseed then? [00:27] specifically and only with automatic-ubiquity [00:27] ah i see [00:27] it's actually a change from gutsy which was part of the early automation work [00:28] need to figure out with Evan whether we still need it [00:28] well additionally for noninteractive mode though, since the console setup script in casper runs before the preseed (and the console setup script doesn't read from the preseed), they don't take effect immediately on that terminal [00:28] unless chosen and seeded on the command line [00:28] mario_limonciell: like I said, can't fix it (for hardy) for the terminal [00:28] sorry [00:29] oh it's not critical [00:29] not a big deal [00:29] as long as the resultant system ends up right [00:29] it's harder than that, even if you put it on the kernel command line, casper's console script can't change the keymap because usplash is running [00:29] usplash puts the console into KD_GRAPHICS mode and you can't change the keymap when it's in that state [00:29] it's all rather a pain [00:30] oh yuck. [00:30] I think it'd take effect once console-setup has run from within ubiquity, though [00:31] have to rectify my previous statement, booting into live cd (not automatic-ubiquity) works whether you use custom-installation or not [00:44] ok the above changes will be post-rc FYI [00:53] so this is really pretty horrible [00:54] ubiquity has to allow console-setup/layout to have its seen flag set to true because otherwise automation fails (the question gets asked) [00:54] but if console-setup/layout is seen then console-setup.config never puts the value from console-setup/layoutcode in place [00:56] I do want to know how come it's marked seen in the first place though [00:57] hmm! it's marked seen in the squashfs [00:58] I bet that isn't helping [01:10] ok, http://paste.ubuntu.com/7245/ is what I'm thinking of for the moment [01:30] evand pushed wubi rev 489 [01:30] didn't really test it, cannot keep my eyes open anymore, won't be in rc anyway [01:32] ubiquity: cjwatson * r2665 ubiquity/ (debian/changelog ubiquity/components/console_setup.py): [01:32] ubiquity: * Fix console-setup seen flag handling; console-setup/layout and [01:32] ubiquity: console-setup/variant need to be marked unseen even in automatic mode so [01:32] ubiquity: that they get their defaults set correctly, so use a different method to [01:32] ubiquity: avoid displaying the question in preseeded installations (LP: #188492). [01:32] ago: happy birthday [01:32] ;-) [01:32] ah that is the 20th of April [01:32] heh, timing [01:33] but thanks anyway [01:33] in advance for the present [01:33] I had no idea :) [01:35] heh [01:41] night and thanks again [02:54] dear lord, what happened in here [02:54] * evand reads scrollback [02:58] ah, looks reasonable. Thanks for taking care of that, cjwatson. [03:05] evand: your partman-base patch looks right. The grep over /proc/swaps is maybe a little general - perhaps 'grep "^$device[0-9]"'? - but over-generality isn't actually harmful here [03:07] evand: if you've tested that, go ahead and commit/upload at your leisure [03:07] I've marked the bug release-critical [03:07] oh, in case you somehow end up at /dev/sdaa? [03:08] yeah, something like that [03:08] hmm, except RAID devices get p0 etc. on the end [03:08] so perhaps it's better to ignore me and take the hit of occasionally swapoffing something unnecessarily [03:09] most people will never notice [03:09] heh, fair enough. [03:09] indeed [03:09] ok, will test one last time to be extra sure and then upload [03:10] not for RC now, I guess [03:11] we could try, but slangasek knows where to find our addresses :) [03:11] ubiquity: cjwatson * r2666 ubiquity/ (debian/changelog partman/check.d/03partition_too_small): [03:11] ubiquity: * Fix partition_too_small check only to warn about partitions not marked [03:11] ubiquity: as "do not use" and marked for use with a filesystem. [03:11] ^-- I haven't actually found a bug report identifiably due to that yet, but I think it makes sense - double-check? [03:14] what's the difference between acting_filesystem and use_filesystem? [03:14] err, I always have to look it up [03:15] one moment [03:19] ok, use_filesystem means that the method requires the partition to have a file system attached to it (as opposed to swap or don't use or whatever) [03:19] acting_filesystem says which filesystem should be used [03:20] use_filesystem has an interaction with preseeded partitioning methods [03:20] i.e. it's actually set in partman-auto recipes [03:20] I think in this case testing presence of acting_filesystem and use_filesystem would be equivalent [03:22] ok, I just noticed that we're using both, use in finish.d and acting in check.d. [03:22] though actually, I'm not entirely sure of that. I think it's possible to have use_filesystem but not having selected an actual filesystem to use [03:22] err nevermind [03:23] I'm looking at two completely different bits of code. [03:23] feels sort of like an error condition but I'd have to stare at it some more [03:23] ok [03:24] 218391 is impressive [03:24] /dev/sda11 on /target type ext3 (rw,relatime,errors=remount-ro) [03:24] /dev/sda10 on /target type ext3 (rw) [03:24] /dev/sda7 on /target type reiserfs (rw) [03:24] /dev/sda8 on /target type ext3 (rw) [03:24] /dev/sda9 on /target type ext3 (rw) [03:25] I notice also: [03:25] /usr/lib/ubiquity/ubiquity/zoommap.py:333: DeprecationWarning: integer argument expected, got float [03:25] self.big_pixbuf = self.pixbuf.scale_simple(w * 4.5, h * 4.5, gtk.gdk.INTERP_BILINEAR) [03:25] maybe int() needed there [03:26] I really wish something in the partman log mentioned mountpoints [03:27] heh [03:30] ubiquity: cjwatson * r2667 ubiquity/ (debian/changelog partman/check.d/03partition_too_small): also make sure that partition_too_small skips free space 'partitions' [03:32] I think I'll investigate 218391 tomorrow; need sleep now [03:32] yeah, I can imagine. [03:32] cheers [03:32] the partman-base change works by the way, uploading now. [03:33] great, thanks [03:34] evand: did somebody notify ubuntu-doc of the KDE resize widget changes (bug 215131)? [03:34] Launchpad bug 215131 in ubiquity "resize widget changes for clarity" [Undecided,New] https://launchpad.net/bugs/215131 [03:35] partman-base: evand * r86 partman-base.ubuntu/ (debian/changelog lib/base.sh): [03:35] partman-base: * Disable swap on all the swap partitions for the device being [03:35] partman-base: changed, rather than just the ones that will exist after [03:35] partman-base: partitioning (LP: #218394). [03:37] hope I was right to close 215131 [03:39] I think so [03:40] partman-base: evand * r87 ubuntu/debian/changelog: releasing version 114ubuntu5 [06:53] cjwatson, can you remind me where to edit to change what gets put on the live-cd desktop? In particular, interested in the 'install now' option. [07:10] nevermind, found it I believe [09:54] <_ruben> is there a complete list of 'd-i directives' and similar which can be used as boot parameter during install and/or in a preseed file? [09:55] <_ruben> including stuff like specifying a custom (local) install mirror [10:38] _ruben: the installation guide is the best source for that [10:40] <_ruben> cjwatson: hmm.. couldnt find a very detailed list there, guess i'll have to go look a bit more closely [10:44] preseeding appendix. it's got lots of stuff [10:47] <_ruben> hmm .. seems the 7.04 info is more extensive than 7.10 [10:50] you mean on help.ubuntu.com? [10:51] there have been some delays in getting the installation guide published there, so it isn't up yet; use the installation-guide-i386 (etc.) package in the archive [10:52] <_ruben> ah, yes [10:52] <_ruben> help.ubuntu.com is what i was looking at [10:53] <_ruben> hmm ... gutsy's install guide is versioned 20070319ubuntu2 .. sounds rather 'feisty' [10:55] the bit at the start is just the version of the Debian guide that it's based on [11:00] casper: cjwatson * r504 casper/ (debian/changelog scripts/casper-bottom/10adduser): [11:00] casper: * Ensure that the live CD user's Desktop directory is owned by them, not [11:00] casper: by root (LP: #218576). [11:01] <_ruben> cjwatson: ah [12:55] do we have any Migration-Assistant devs here? [12:55] davmor2: evand is the guy [12:55] (US/Eastern time) [12:56] damn [13:21] davmor2: well, US eastern should soon be awake :) [14:03] hi [14:03] evand: hello I got issues with M_A [14:04] evand: Should M-A on 64bit transfer details for 32bit XP? [14:07] it should, yes [14:08] evand: it isn't even registering that there is another OS and yet 32bit has work flawlessly I just checked [14:10] evand: I reinstall xp and re-run the test what logs do you need or is it just running the install with --debug? [14:14] run ubiquity in --debug mode and attach /var/log/syslog and /var/log/installer/debug to a new bug [14:14] thanks! [14:16] evand: Np's I'll ping you with the number once it's up :) [14:17] thanks! [14:40] evand: daft question I got the live cd booted up now. Is it the log files of the installed system or the running live one? [14:42] the live one, though if the install completes successfully the same log files can be grabed from /var/log/installer/{syslog,debug} as it copies them to the target system at the end of the install. [14:44] evand: yes it installs fine just M-A not working. [14:44] ok [14:48] evand: installing. about 15 minutes-ish [14:49] ok [14:52] evand not sure if you followed yesterday discussion, but basically wubi 490 is up [14:53] to address layoutcode issues, I also added automation_failure_command [14:55] should go in after rc [15:01] wubi is displaying no backdrop on Kubuntu-KDE4 but is strangely showing a ubiquity window behind the installing system window :) [15:04] xivulon: ok [15:05] davmor2: curious, can you file a bug for that with a screenshot? [15:06] xivulon: I'm not sure automation_failure_command is a good idea in the case of wubi. A failure in automation does not necessarily constitute a unrecoverable failure. [15:06] but that's of course your perogative. [15:07] prerogative* [15:08] evand: no to screenshot I'll see if kubuntu does the same. It's before the system loads [15:09] ok [15:11] evand ok will comment it out, it's only one line in the preseed template [15:12] feel free to do it yourself data\preseed.lupin [15:12] ok [15:12] wubi downloader won't run until metalinks are in place of course [15:13] indeed [15:15] evand: there's no debug log in /var/log/installer/ there's casper.log, initial-status.gz, partman, syslog and version. I've already uploaded syslog. [15:16] hrm, and you ran the installer with --debug? [15:16] evand: yes there is a debug in /var/log/ [15:18] no, that's for the kernel [15:18] hrm [15:19] could I trouble you to run the installer again and grab the debug log before it reboots? I'll have to see if this wasn't just a fluke and the debug log stopped getting copied out. [15:20] evand: where is it likely to be on the livecd I'll run it again until it gets to the confirmation piont that should be enough shouldn't it? [15:20] /var/log/installer/debug [15:20] it should show up as soon as you start ubiquity with --debug [15:21] bug 218673 if you want to view the syslog while you wait :) [15:21] Launchpad bug 218673 in ubuntu "64bit Migration-Assistant can't see XP" [Undecided,New] https://launchpad.net/bugs/218673 [15:32] evand debug should be up now [15:39] evand: is that okay for you need to do more tests? [15:49] should be good for the initial look over [15:49] thanks [15:49] I'll let you know if I figure out what's causing this or need more information [15:50] evand: it looks like it's listed just not showing in the menu if you need a screenshot of what I get I'll do you one after rc is out when I got more time okay [15:51] ok [15:59] davmor2: so the only difference between your two installs was the architecture on the installer CD? [15:59] evand: yes 32bit work 63bit fails [16:00] and just to clarify, you used imaging software or something similar to put both install attempts at the same starting point. That is, both systems had the same or roughly the same filesystem from the start. [16:02] evand: Yes I use partimage to refresh xp each time. This gives me XP with bookmarks email all setup so M-A has something to import. [16:03] ok, thanks [16:07] evand: Also your right about login for messenger there are no details apart from a random encrypted file I found. I am guessing that is the thing that attaches to windows live and announces your system is online. [16:07] hrm, what's the path to the file? [16:07] as that's more than I've found [16:09] I'll see if I can find it again I think it's in a random sub folder assigned to live rather than messenger but I'll post you the path again after rc is out. [16:10] ok, thanks [16:27] evand: here is a photo of that kde4 install screen http://picasaweb.google.com/davmor2/UbuntuScreenshots/photo#5190235907148011634 [16:50] yikes [17:12] evand: obviously no shiney light though ;) [17:15] evand: it's the same with Kubuntu too [17:15] brb [17:15] so that's all you see? You never see an install window, or does it appear in front of that? === davmor2 is now known as davmor2_away === davmor2_away is now known as davmor2 [19:05] evand: ping [19:08] davmor2: pong [19:09] evand: is there no wubi on the dvd images? [19:09] davmor2: correct, we removed it as it takes far too long and uses too much space [19:09] davmor2: umenu however, should show up without the wubi option [19:09] otherwise that's a bug in umenu that needs to be reported [19:09] it takes too long because of processing all the language packages at the end? [19:09] or just the copy process [19:10] evand: no there is nothing booting in XP at all [19:10] because it tries to turn an entire DVD into an ISO [19:10] oh i see [19:10] davmor2: curious, there's no umenu.exe and autorun file on the root of the DVD? [19:11] I'll double check but I don't think so [19:12] evand: md5sum/cdromupgrade/README.diskdefines/ubuntu the rest are folders [19:13] ah, I see the problem [19:13] I am double checking stuff and I noticed that /host gets mounted with the following options: [19:13] rw,nosuid,nodev,noatime,relatime,user_id=0,group_id=0,allow_other [19:14] evand: What's the problem? [19:14] It is my understanding that it will make /host rw to everyone [19:14] isn't that a security concern? [19:14] davmor2: I accidentally made it so that umenu is only included on the cd builds if we're not building a dvd [19:14] fixing now [19:14] cjwatson, evand let me know what you think of the above [19:16] evand: so does that mean that we'll have winfoss on the dvd or just the info files? [19:17] xivulon: I'm not overly concerned about that, but perhaps cjwatson will disagree. [19:17] evand: when has wubi been removed from the DVD ? [19:17] davmor2: it'll have umenu and wubi [19:17] err just umenu [19:18] branch nick: cdimage.public [19:18] timestamp: Tue 2008-04-01 10:54:46 -0400 [19:18] message: Don't install Wubi to the DVD. It takes too long to make an ISO of the DVD and uses too much space. [19:18] stgraber: ^ [19:19] ok [19:20] evand: next time you do that kind of change (especially to DVD) can you please send me or the ubuntu-qa ML a mail so we know we need to update the testcases in the tracker ? [19:20] now it's too late to remove the Wubi testcase for final testing :( [19:23] ah, sorry about that. [22:23] evand: have you seen xivulon recently? [22:27] bdmurray: no. He's usually quick to reply to email though. [22:31] okay, do you know off hand where to find the Wubi-installer log? [22:32] %TEMP%\Wubi-revXXX.log [22:32] something like that [22:38] just in case I haven't said it before I hate vista [22:52] bdmurray: even Windows users do :) [23:18] bdmurray, re 204128, does wubi jam after that or can you resume somehow [23:19] xivulon: I've tried it 2x now. One time it said complete and ejected the disk then hung and Windows hung too. I forcibly rebooted and was in a bad state. [23:20] xivulon: The second time the I was able to forcibly quit Wubi, reboot and install. [23:23] it's a strange issue because from the log it appears that all my code gets executed. [23:23] is this only with CD or also with ISO? [23:24] xivulon: I've only tested with a CD. I have a log from rev487 too but it ends the same way [23:24] ...installation successful... [23:25] right [23:30] I am doing a build with some more logging [23:39] bdmurray, can you pls try http://wubi-installer.org/devel/minefield/Wubi-8.04-beta-rev491.exe [23:41] and this one: http://wubi-installer.org/devel/minefield/Wubi-8.04-beta-rev491-nsis234.exe [23:50] xivulon: okay, just download and execute those files? [23:50] yes with the cd in the tray