/srv/irclogs.ubuntu.com/2009/04/03/#ubuntu-installer.txt

CIA-28ubiquity: cjwatson * r3174 ubiquity/ (debian/changelog scripts/install.py):01:35
CIA-28ubiquity: Sync with netcfg 1.44: Define IPv6 ::1 address as "localhost" (thanks,01:35
CIA-28ubiquity: Robb Topolski; LP: #301430).01:35
CIA-28base-installer: cjwatson * r361 ubuntu/ (debian/changelog library.sh): Silently skip non-existent devices in /proc/swaps (LP: #290947).01:45
CIA-28ubiquity: cjwatson * r3175 ubiquity/ (debian/changelog scripts/install.py): Silently skip non-existent devices in /proc/swaps (LP: #290947).01:47
xivulonsteve, in the metalink for the beta there are some entries with preference = 110 (>> bittorrent)02:43
CIA-28wubi: Agostino Russo * r108 trunk/ (18 files in 8 dirs):03:30
CIA-28wubi: * Disabled ISO backup because download resume is not fully supported03:30
CIA-28wubi:  at the moment03:30
CIA-28wubi: * Isolist.ini now points to the final release URLs03:30
CIA-28wubi: * Allow the uninstaller to delete the directory containing itself by03:30
CIA-28wubi:  running the python code from a different process/location (LP:03:30
CIA-28wubi:  #341605)03:30
davmor2cjwatson: I like the new colour scheme on the partitioning section of ubiquity09:36
cjwatsonthank evand :)09:42
CIA-28os-prober: cjwatson * r232 ubuntu/debian/ (changelog rules):09:43
CIA-28os-prober: Install Mac OS X probe on i386/amd64 too, although it will be marked as09:43
CIA-28os-prober: unsupported in grub-installer as AFAIK it can only be booted directly09:43
CIA-28os-prober: from EFI (LP: #353639).09:43
davmor2I would do but he seems to be away :(09:45
CIA-28os-prober: cjwatson * r233 ubuntu/ (debian/changelog os-probes/mounted/x86/30utility): Detect Acronis Secure Zone (thanks, Alexey Fisher; LP: #354334).09:47
cjwatsondavmor2: yeah, off this week09:47
CIA-28os-prober: cjwatson * r234 ubuntu/debian/changelog: releasing version 1.29ubuntu209:51
ogracjwatson, ubiquity still inhibits screensaver and dpms, right ?10:52
cjwatsonshould do10:52
ogra(or did that change at some point)10:52
ograok10:52
ograbecause i see screen blanking on the babbage board ... must be a kernel setting then10:52
cjwatsonsee poke_screensaver in ubiquity/frontend/gtk_ui.py10:52
cjwatsonit's on a timer, so I suppose the timer could be wrong; ideally we ought to switch to --inhibit10:53
ograi belive you, no need to look, i'll check the kernel config10:53
cjwatsonscreen blanking might not be through the screensaver though?10:53
ograthats what i think10:53
ograi think there is some additional backlight control stuff from freescale10:53
ogragrin ... there is a kernel setting called CONFIG_ARCH_MXC_CANONICAL=y10:55
ograaha ... CONFIG_BACKLIGHT_MXC=Y and CONFIG_BACKLIGHT_MXC_MC13892=y ... MC18392 is the framebuffer driver if i'm not wrong10:56
cjwatsonsuperm1: hmm, I tried out that langpack filtering code with 120 language-support-* names, and the filtering is instantaneous here10:59
cjwatsonsuperm1: do you think you could try it out with something like this instead of the list comprehension?11:02
cjwatsonnew_to_install = []11:02
cjwatsonfor lp in to_install:11:02
cjwatson    syslog.syslog('Testing language pack: %s' % lp)11:02
cjwatson    if self.get_cache_pkg(cache, lp) is not None:11:02
cjwatson        syslog.syslog('Language pack found')11:02
cjwatson        new_to_install.append(lp)11:02
cjwatsonto_install = new_to_install11:02
cjwatsonsuperm1: then we'd get more detailed timestamps11:02
xivulonhi davmor211:59
davmor2Hello xivulon11:59
xivuloncould you please test r108? http://people.ubuntu.com/~evand/wubi/jaunty/wubi-r108.exe12:00
davmor2yes was going to12:00
davmor2xivulon: this is the one that should allow removal correct?12:00
xivuloncorrect12:00
davmor2cool :)12:00
xivulonOther than 341605 it might also address: #341607 and #34752912:01
xivulonCould you please also re-test 207137?12:03
xivuloncjwatson I have a recursive dir deletion code I would like you to review12:04
xivulon, that has been in for some time, but because of a bug that code path was only rarely hit12:04
xivulonhttp://bazaar.launchpad.net/%7Eubuntu-installer/wubi/trunk/annotate/head%3A/src/pylauncher/deletedir.c12:05
xivuloncalled by pylauncher.c12:05
xivulonIt seems ok to me but it would be very annoying if something goes wrong there12:05
cjwatsonxivulon: would it be possible to call python code instead?12:06
xivulondavmor2, would you mind adding comments to the above bugs?12:06
cjwatsonsince there's shutil.rmtree12:06
davmor2xivulon: np's12:06
xivuloncjwatson, that is the code that actually launches python itself, and python is contained in the dir that has to be deleted, which is something you cannot do in windows12:07
cjwatsonI don't think I'm qualified to review that function, unfortunately; I don't know the Win32 file handling API at all12:07
cjwatsonthe top-level logic looks fine but there could be all sorts of lurking gotchas in there that I'd never spot12:07
xivulonyou have far better overall understanding than I do anyway12:08
cjwatsonis there any possibility that it would need to cope with symlinks?12:08
xivulonnot in the wubi dir to be deleted, I have no symlinks there12:09
xivulonjust plain files and dirs12:09
cjwatsondo you need to handle errors from line 59 before going on to attempt to delete the file?12:09
cjwatsonif you intend to ignore errors there, it would be better to just not set result and try to delete the file anyway, I'd have thought12:10
cjwatsonit appears that at the moment if setting attributes fails but deleting the file succeeds, then the function will fail, which seems unnecessary12:10
cjwatsonI don't know if that is possible12:10
cjwatsonoh, it looks like you leak memory from file_path every time around the inner loop12:11
xivulonIn our case all files should have normal attributes anyway, ah good catch12:12
cjwatsonyou need to freestr that each time round12:12
cjwatsonand probably then remove the freestr outside the loop (line 74)12:13
cjwatsonalso you leak pattern if file_handle == INVALID_HANDLE_VALUE12:13
cjwatsonI think that's everything I can see12:13
xivulonwill patch that tonight, please have also a look at pylauncher.c, that is where actually things could be problematic (in terms of wrong delete_dir argument)12:13
cjwatsonliteral strings should have type const char[] not char[]12:14
cjwatson(assuming that doesn't cause trouble when passing to other functions, in which case you might need casts)12:15
cjwatsondo you need to handle errors from delete_directory in main?12:15
xivulonI wouldn't think so, there isn't much I can do if that fails, they will just have a dungling temp directory12:16
cjwatsonI'd use sprintf instead of all of those strcats personally, but that's a style point12:16
cjwatsonbest hope that targetdir is not very long ...12:17
cjwatsonare you sure you don't want to make cmd bigger, calculated based on MAX_PATH?12:17
cjwatsonand message should clearly be allocated 14 bytes bigger than cmd :-)12:17
xivulonC:\Documents and Settings\arusso\Local Settings\Temp12:17
cjwatsonpersonally I'd allocate cmd and message dynamically12:18
cjwatsonreallocing if necessary around the argument loop12:18
cjwatsondoes CreateProcess not have any way to avoid having to parse the command out of a string?12:18
xivulonnot that I am aware of, I would love to poass an array of strings12:19
cjwatsonmm, subprocess.py does similar stuff12:21
cjwatsondo you need to take care to quote strings in the command line the way it does?12:21
xivulonI assume that it would be safer to quote strings12:21
xivulonI did that in header.c might have to do that again in pylauncher.c12:22
xivulonIn fact I think that is pretty much required12:22
davmor2xivulon: it got 90's-ified blue block going across the bar rather than the plush green throbber12:25
xivulondavmor2, when you test r108, please do so _without_ admin privileges, you should be prompted once by UAC12:25
xivuloncan you provide me with a screenshot?12:26
davmor2xivulon: I'm just running as is from normal user12:26
xivulonwith no admin rights...12:26
xivuloncjwatson: thx, will incorporate your comments tonight/tomorrow12:27
cjwatsoncool12:27
xivulondavmor2, I am going over the remaining UI glitches this w/e please file a bug for the bar12:29
davmor2xivulon: http://www.davmor2.co.uk/vistabar.jpg12:31
xivulonby the way I bought a macook :)12:32
xivulonI think we can have a mubi at some stage, since grub2 efi is coming along quite nicely (thanks to bean123)12:33
xivulonmacbook12:33
xivulonalthough because of lack of journal support in hfs , I doubt it will be possible to do loopinstallations12:34
xivulonon the plus side, osx supports online resizing, so it would be possible to do a bootcamp type of app12:35
davmor2cool12:37
xivulondavmor2 I know the issue for the vista bar12:41
davmor2xivulon: it hasn't whinged about the cd not being there.  But after dinner I'll add wubi to a cd and try that for sure12:41
superm1cjwatson, sure.  will do12:42
cjwatsonthanks12:42
xivulonthe windows resource declaring that themes have to be used is associated to pylauncher, which in turns runs python.exe, but python.exe does not have that resource12:42
CIA-28ubiquity: cjwatson * r3176 ubiquity/ (5 files in 4 dirs): Use small print rather than italics in "Who are you?" (LP: #287626).12:44
* cjwatson establishes some runes for booting a live CD in kvm with an arbitrary screen resolution (or at least something that the relevant X driver supports)13:12
cjwatsonboot without splash and with break=casper-bottom, then: sed -i '/Modes.*"/s/".*/"800x600"/' /root/usr/bin/dexconf13:13
cjwatsonand exit 0 to continue booting13:13
CIA-28oem-config: cjwatson * r646 trunk/ (debian/changelog lib/timezone_map.py):13:57
CIA-28oem-config: * GTK frontend:13:57
CIA-28oem-config:  - Use the full size of the underlying images for the timezone map if the13:57
CIA-28oem-config:  screen is big enough.13:57
CIA-28oem-config: cjwatson * r647 trunk/ (debian/changelog gui/glade/step_timezone.glade): Fix spacing on timezone page (LP: #336745).13:59
davmor2xivulon: ah good your still around.  I'm just going for a removal now14:19
davmor2xivulon: windows app removal gave a error and removed ubuntu from the app list but all the folders are still in place so not quite sure what happened there14:24
superm1cjwatson, unfortunately those partman-base changes did hurt us as I'm finally testing the daily today with new ubiquity14:27
superm1the partman screen comes up with the warning14:28
cjwatsonsuperm1: nngh. --debug output please?14:32
superm1simultaneously (on a diff box w/ older ubiquity) I tested that patch again that you proposed a few days ago before adding the debug info from earlier this morning.  it worked this time, so I am going to run it once more to ensure the previous data point is a read herring caused by multiple ubiquity runs w/o a reboot14:32
cjwatsonthat'd be a relief, that's an RC bug :)14:33
superm1s/read/red/ that is14:33
superm1sure give me a bit, i'm letting that install finish off anyhow to see if grub2 stuff works out with it14:34
superm1cjwatson, so the second run with that patch was successful too, so I'd say pull that in and ignore my old data point w/ the multiple ubiquity runs14:55
cjwatsonsuperm1: cool, that's good news, thanks15:07
CIA-28ubiquity: cjwatson * r3177 ubiquity/ (debian/changelog scripts/install.py):15:19
CIA-28ubiquity: Revert r2984 from ubiquity 1.11.5, and instead filter out language packs15:19
CIA-28ubiquity: that don't exist in the live filesystem's apt cache; this allows us to15:19
CIA-28ubiquity: accurately produce incomplete language support notifications again.15:19
CIA-28ubiquity: Thanks to Mario Limonciello for testing (LP: #337748).15:19
CIA-28ubiquity: cjwatson * r3178 ubiquity/ (debian/changelog ubiquity/frontend/mythbuntu_ui.py):15:27
CIA-28ubiquity: * Mythbuntu frontend:15:27
CIA-28ubiquity:  - Don't crash if --no-migration-assistant was used (LP: #354368).15:27
superm1cjwatson, bug 354534 is the results of grub2 testing, i'll do one more to file the partman-base change separately however15:44
ubottuError: Could not parse data returned by Launchpad: timed out (https://launchpad.net/bugs/354534/+text)15:45
cjwatsonthanks15:46
superm1thanks for the mythbuntu fix above too15:46
cjwatsonnp15:48
cjwatson/cdrom/scripts/chroot.sh: line 32: /target/var/log/installer/chroot.sh.log: No such file or directory16:09
cjwatsonsuperm1: I assume that's from your preseeding? if so heads-up ...16:09
superm1cjwatson, it's because it's not really ran in a factory server, but a faked env I believe16:12
cjwatsonah16:12
cjwatsonthis is odd, looks like all the progress templates are missing ...16:12
cjwatson14:27 <superm1> cjwatson, unfortunately those partman-base changes did hurt us as I'm finally testing the daily today with new ubiquity16:16
cjwatson14:28 <superm1> the partman screen comes up with the warning16:16
cjwatsonsuperm1: but it's a warning at the top of the screen rather than a dialog, right?16:16
superm1cjwatson, yes16:16
cjwatsonsuperm1: I'm not clear on why that's bad - you genuinely can't create, resize, or delete partitions on that device, surely16:16
cjwatsonand that seems like something it's appropriate to warn about16:16
superm1I still need to keep ubiquity/partman-skip-unmount boolean true otherwise there is a popup warning dialog16:16
superm1Oh I was certainly able to make a partition on the device16:17
superm1I picked the free space and made an ext3 partition to install to16:17
cjwatsonbut the kernel will not be able to reread the partition table unless /cdrom is unmounted16:17
cjwatsonfrom /dev/sda216:17
superm1which is what I've always thought too, but it's worked out fine16:18
superm1I never really could identify why it worked16:18
cjwatsonmaybe it works provided that it's at the end16:18
cjwatsonI really don't want to imply that it's supported, though16:18
cjwatsonbecause I'll have to deal with the bugs when it breaks16:18
robbiewcjwatson: we're up16:18
superm1cjwatson, yeah, i'm fine with a preseed like before to just support this case16:18
superm1but i think the warning is good for the general case16:18
cjwatsonsuperm1: bet it wouldn't work if you attempted to resize /dev/sda2, for example16:20
superm1cjwatson, yeah I think resizing a live filesystem itself is asking for trouble, but just adding new partitions appears to be fine16:21
superm1especially since /dev/sda2 was vfat16:21
cjwatsoncan I reject all bugs from your users about breakage? ;-)16:21
superm1haha, well I doubt they would get to that point of filing bugs if they got a system broken from this16:22
cjwatsonalthough, if it's just in the factory process, then it doesn't seem important to tweak the UI16:22
cjwatsonI can do something about the continuing need for ubiquity/partman-skip-unmount16:22
superm1yeah it's just factory process.  the recovery media imitates factory process as best as possible, but just asks a yes/no do you want to recover question16:23
cjwatsonand then noninteractive?16:23
superm1well we're switching to automatic-ubiquity for 9.04 since vesa appears to finally be a feasible fallback16:23
superm1if it'd be useful,  I can post an ISO of recovery media generated from today's daily CD and you can take a look at how things work16:24
cjwatsonnah, it's OK, ISO downloads pretty slow for me as previously advertised :)16:25
superm1ok, i've added bug 354573 to keep track of the partman-base breakage. let me know if you need anything else for it16:38
ubottuLaunchpad bug 354573 in partman-base "Can no longer perform factory installs due to changes in partman-base" [Undecided,New] https://launchpad.net/bugs/35457316:38
cjwatsonsuperm1: ah, looking into the kernel in a bit more detail, it looks as if you can modify partitions precisely if they're after any mounted ones on the disk16:38
cjwatsonsuperm1: "can no longer perform factory installs" is not the same as "there's a warning on-screen, but it works anyway"16:39
cjwatsonwhich is it? :)16:39
superm1haha, well i'll put automated back in there :)16:39
cjwatsonbut it would work with ubiquity/partman-skip-unmount, wouldn't it?16:40
cjwatsonsuperm1: BTW, in general I need the syslog too, but I'm assuming that it's the same log as in 35453416:41
superm1well I did this w/o grub2, so it's a tad different.  i'll add it16:41
superm1so I had ubiquity/partman-skip-unmount in the seed (attached the seed too)16:41
superm1warning still comes up with it there.16:42
superm1if that's taken out, then there is a ubiquity dialog box that pops up talking about /cdrom being mounted16:42
cjwatsonthe installation_medium_mounted warning should have no effect on automation, so we can take that out of the equation16:42
cjwatsonI'd like to identify the dialog though16:43
superm1installation_medium_mounted is what's showing up on the partitioner page. w/o ubiquity/partman-skip-unmount, ubiquity/partman-failed-unmount is also shown (and the install can't proceed)16:59
superm1perhaps, the reason why the automation is bogging out for the partitioner page - the recovery partition does not take up "most of the disk", so that check for it doing so wouldn't pass17:01
=== Nicke_ is now known as Nicke
cjwatsonsuperm1: that check only governs the warning, which shouldn't affect automation17:28
superm1oh17:28
cjwatsonwell, it also governs whether the device shows up in partman at all, but in that respect the current behaviour is in your favour17:29
=== nxvl_ is now known as nxvl
superm1cjwatson, oh, i was just reading through changelogs a little, and then saw that partman-auto 84ubuntu2's stood out. "Exclude devices containing the installation medium from automatic partitioning (LP: #347916)".  That sounds like the situation considering we've got a recipe that uses automatic21:40
superm1cjwatson, yup, I just tried reverting the diff from 84ubuntu1 to 84ubuntu2 and --automatic works again21:45
kirklandcjwatson: hey, for intrepid, in the alternate installer, we have some debconf text that tells the user to record their randomly generated passphrase if they setup an encrypted private22:04
kirklandcjwatson: we seem to have lost that page in the alt/server installer for jaunty22:05
kirklandcjwatson: could we get that page back?22:05
cjwatsonsuperm1: sigh, I wish your requirements weren't so different from everyone else's :-/22:18
cjwatsonsuperm1: I think we'll just have to reintroduce partman/filter_mounted22:18
cjwatsonkirkland: um, you removed that22:18
kirklandcjwatson: okay, i'll go take a look22:18
cjwatsonkirkland: was it not obsolete with encrypted-home?22:19
kirklandcjwatson: the first half is obsolete22:19
kirklandcjwatson: "do you want to pick a password, or generate one"22:19
kirklandcjwatson: we always generate one22:19
kirklandcjwatson: the next screen though is needed22:19
kirklandcjwatson: "your generated password is BLAH, write this down!"22:20
cjwatsonput a branch up and I'll review it22:20
cjwatsonaim for minimal string changes (ideally a strict subset of what was there before, if possible), please22:20
kirklandcjwatson: k, will do22:20
kirklandcjwatson: i think it will be identical22:20
kirklandcjwatson: what's the specific package?  usersetup?22:20
cjwatsonuser-setup22:21
kirklandcjwatson: thanks.22:21
cjwatsonlp:~ubuntu-core-dev/user-setup/ubuntu22:21
CIA-28partman-base: cjwatson * r150 ubuntu/ (debian/changelog init.d/parted):23:28
CIA-28partman-base: Support partman/filter_mounted preseeding again, for some special cases23:28
CIA-28partman-base: that know they can get away with modifying partitions after the one23:28
CIA-28partman-base: containing the installation medium (LP: #354573).23:28
cjwatsonsuperm1: I'm reverting your description back to the original; partman-auto uses information that partman-base writes out, and so simply restoring partman/filter_mounted preseeding is both more generally applicable and more elegant23:31
cjwatsonwell, more or less the original23:31
CIA-28partman-base: cjwatson * r151 ubuntu/debian/changelog: releasing version 129ubuntu623:34
CIA-28ubiquity: cjwatson * r3179 ubiquity/ (d-i/manifest debian/changelog): Automatic update of included source packages: partman-base 129ubuntu6.23:39
CIA-28ubiquity: cjwatson * r3180 ubiquity/debian/changelog: releasing version 1.12.323:45

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!