[01:12] cjwatson: Reading your updates to bug #538536 : is this in a state where I should proceed with install, or do you want me to keep the system uninstalled for a couple more weeks? [01:12] Launchpad bug 538536 in parted "automatic partitioning broken on Intel Macs" [High,Fix released] https://launchpad.net/bugs/538536 [01:12] persia: I think it's worth you going ahead with a test [01:15] OK :) If it succeeds, I'd like to leave the system installed : do you feel confident enough with your simulation that this is a safe path? [01:15] (otherwise I'll do a test and leave it uninstalled) [01:18] I think that should be safe, assuming it works at all [01:19] partitioning doesn't usually tend to leave subtle hidden traps around for later [01:19] if there are traps, they're big and pointy [01:21] heh. Yeah. I was more concerned about subtle stuff in the simulation. I'll run a test as soon as my cdimage mirror has 2.2-1ubuntu3 in the server image. [01:22] what I'm not certain about, and can't easily be, is whether the result will boot [01:22] we're using grub-pc rather than any of the -efi variants - and if you use one of the -efi variants, AFAIK you need to do somewhat different partitioning [01:23] so I suspect that the installer would need to be explicitly prepared for that somehow [01:23] Is the use of an -efi variant expected to be required for this sort of hardware? [01:23] grub-pc should work, but you might need something EFI-ish to chain to it, like refit [01:23] or you might just be able to hold down Option at boot and have it work [01:24] the exact details have probably changed a bit since I was last deeply involved with Apple kit [01:24] Hrm. I was planning to leave this system entirely headless. I guess I'll plan on keeping it not installed, unless it just works flawlessly. [01:24] anyway, I've been meaning to go to bed for about 40 minutes, so I shall go :) [01:25] night [01:25] Night :) [10:07] usb-creator: evand * r277 usb-creator/ (bin/usb-creator-helper debian/changelog): Don't write usb-creator-helper's log to /root (LP: #461064). [11:28] partman-base: cjwatson * r192 ubuntu/parted_server.c: abstract partition creation constraints [11:59] cjwatson: The 20100315 Server image was unable to handle manual LVM, but successfully completed with guided LVM. As you predicted, the result didn't boot. Shall I add the logs to a new bug, or to the old one? [12:01] your old one is fine [12:02] I probably won't get further for beta-1 at this point [12:03] No worries. I'm fairly sure I can get the system to be a basic sbuild host with what I have an grub-efi. Due to other circumstances, I'm not going to be able to set it up properly until right around Beta 2 release anyway, so I should be able to do another end-to-end run for that. [12:04] And because of this, running more installs next week (but not this coming weekend) shouldn't interrupt much. [13:16] ubiquity: cjwatson * r3931 ubiquity/ (debian/changelog ubiquity/components/ubi-language.py): [13:16] ubiquity: * KDE frontend: [13:16] ubiquity: - Honour ubiquity/show_alpha_warning. [13:43] ubiquity: cjwatson * r3932 ubiquity/ (debian/changelog ubiquity/components/ubi-partman.py): [13:43] ubiquity: * Noninteractive frontend: [13:43] ubiquity: - Fix up plugin problems in the partman component (LP: #539493). [13:44] ev: I'm worried about the fact that we're seeing tracebacks from the noninteractive frontend from people who never asked for it. How would you feel about http://paste.ubuntu.com/396177/ ? [13:47] cjwatson: I think that's fine, but maybe check with Mario as I believe he added the original code [13:47] hmm, not sure we have time to wait for Mario to get up [13:47] if memory serves it was to work the problem now solved by your failsafe X work [13:47] I'd just upload it then, given the above [13:48] superm1: ^- can you review the above post-hoc? I think that noninteractive should only be provided if you explicitly ask for it [13:48] ubiquity: cjwatson * r3933 ubiquity/debian/ (changelog ubiquity.ubiquity.upstart): [13:48] ubiquity: Never start the noninteractive frontend unless it was explicitly [13:48] ubiquity: requested. [13:49] cjwatson, the worry was that if a monitor isn't plugged in in the factory, X doesn't start, so automatic doesn't work [13:50] superm1: the fact that noninteractive is being started on innocent users' machines TERRIFIES ME [13:50] yes, I know it generally doesn't do much because it isn't preseeded, but even so [13:51] cjwatson, understood, perhaps only if automatic-ubiquity was passed do the noninteractive fallback then, if it's just only or maybe-ubiquity mode don't have a fallback? [13:51] and we get all sorts of weird crash reports as a result of this. We need something better [13:51] it would be OK in automatic mode [13:52] wubi runs in automatic mode [13:52] but I suppose that's still okay [13:52] yeah, I think so, it's preseeded what it wants to do [13:54] ok, thanks, done [13:54] ubiquity: cjwatson * r3934 ubiquity/debian/ (changelog ubiquity.ubiquity.upstart): Only fall through to the noninteractive frontend in automatic mode. [13:58] ubiquity: cjwatson * r3935 ubiquity/debian/changelog: better bug number [14:01] *blink* bug 539353 has an RTF file attached [14:01] Launchpad bug 539353 in ubiquity "Install crashed" [Undecided,New] https://launchpad.net/bugs/539353 [14:01] that's a blast from the past [14:02] I wonder what possesses people to do something like that. [14:11] copy&paste into a word processor? [14:11] ubiquity: cjwatson * r3936 ubiquity/debian/changelog: releasing version 2.2.1 [14:13] persia: Yes. [14:14] soren: In many environments in the past, real text editors are hidden from the users. I believe this is still true in recent versions of Mac OS X. [14:15] So users develop habits that continue into other environments. [14:22] * soren admits to have had a brief period where he exclusively jotted down notes in LaTeX [14:46] Hi! [14:46] I have a zenity dialog I'd like to be displayed close to the end of the installation that will perform some additional configuration in the target system [14:47] I put it in /usr/lib/ubiquity/target-config, it's number is the last in the bunch that's there and it's excecutable, I'm not sure if that's the best place to put it but it seems so [14:48] ubiquity doesn't pop up the dialog though, it just seems to hang while doing the "Configuring target system..." stage, the script runs fine if executed seperately [14:49] any advice on how to get it working properly would be greatly appreciated! [15:57] usb-creator: evand * r278 usb-creator/ (5 files in 5 dirs): [15:57] usb-creator: * Provide a format confirmation dialog (LP: #443330). [15:57] usb-creator: * Error out of formatting if we're unable to unmount all of the [15:57] usb-creator: partitions (LP: #507420). [15:57] usb-creator: * Provide feedback via a spinning cursor and disabled format button [15:57] usb-creator: while formatting (LP: #457737). [15:57] usb-creator: * Only call gtk.main_quit if we're in a mainloop. [16:10] ubiquity: cjwatson * r3937 ubiquity/ (bin/ubiquity-dm debian/changelog): [16:10] ubiquity: Quit plymouth if it doesn't have an active VT, rather than trying to [16:10] ubiquity: reuse the VT it doesn't have, which results in X running on VT1. [16:12] ah, I was wondering why that was happening [16:13] interesting [16:14] the kde frontend is using nearly 100% cpu [16:30] ubiquity: cjwatson * r3938 ubiquity/debian/changelog: releasing version 2.2.2 [18:36] I have dug thru the last week's irc logs and ref'ed partitioning bugs - there is something wildly broken with partman recently.. been using a very simple partman-auto/expert_recipe for months almost identical to the example on https://help.ubuntu.com/9.10/installation-guide/i386/preseed-contents.html [18:37] errors out with cannot write partitions outside of device (or very similar) [18:38] removed all partitioning preseeding and picked default "guided - use entire disk" [18:38] can you please file a new bug with your logs (syslog and partman) attached? [18:38] partitions suggested are...interesting [18:38] logs are not exactly available since it cannot install [18:38] sure they are [18:39] go back to the main menu and select "Save debug logs" [18:39] ah, that's right - sorry forgot [18:39] or, switch to alt-f2 and run 'anna-install openssh-client-udeb' to get yourself scp [18:40] there are known problems, but I'll need to analyse your logs in order to determine whether what you're seeing is related to anything else [18:40] but even the suggested defaults of (completely blank 500G drive) http://12.am/tmp/lucid_partman-hmmm.png - that's quite unexpected [18:40] yes [18:40] but I would prefer to debug logs rather than screenshots. :-) [18:41] ok - could you suggest a package to file under? partman-(something) ;) [18:41] partman-base [18:41] perfect [18:41] thanks, cjwatson - hate to keep bugging you :) [18:42] that's ok, I'd rather know about problems [18:54] cjwatson: bug 539774 is probably the same thing. [18:54] Launchpad bug 539774 in debian-installer "Manual partitioning sizes are not honored at all" [Undecided,New] https://launchpad.net/bugs/539774 [18:55] (bug report has partman logs attached) [18:56] possibly. [18:56] elmo reported something similar-sounding too. [18:56] start of first partition at 2049 was what I saw yesterday, too [19:01] exlt: as in LBA 2049? [19:01] that's intentional [19:01] starting at 16GB+ is a somewhat different matter, and is not intentional [19:02] starting at 1MiB improves compatibility with various hardware and OSes [19:02] no - same as that bug has - cyl 2049 and 16.8G free before first part [19:04] oh, *cylinder* 2049 [19:04] helps to specify your units ;-) [19:05] even if they are crazy obsolete ones like cylinders ;-) [19:05] * cjwatson wonders if it's possible to query blkid's idea of topology from the command line [19:07] I could be wrong on cyl - whatever the "Start" value is in fdisk -l - cyl, sector.. [19:08] it is cylinders [19:09] perhaps that intentional start after the first 2048 cylinders is the bug [19:14] no, you have misstated what I said [19:14] I never said that there was an intentional start after 2048 cylinders [19:32] I've reproduced the weird partitioning alignment in kvm, so should be able to fix it [19:33] the size of the disk matters - my usual test size of 3GB exhibited 1MiB alignment, but 200GB exhibited 16GiB alignment [19:37] cjwatson: hrm, but why did it not align everything; e.g. why was 2GB I specified for swap created as gap *after* the 16GB swap space it created? [19:39] who knows, it's clearly screwed [19:45] cjwatson: #539822 has my logs attached [19:50] sbeattie: it could easily be a difference between primary and logical partitions or something - at this point I'm strongly suspecting a bug inside libparted [19:50] exlt: thanks, since I've reproduced it I should be OK from here [19:51] :)