[00:04] I'll have a look [00:25] shall securityfs be skept in umountfs? [00:49] umountfs does not work well, in particular it kills the /host when unmounting /boot [00:50] that's because /boot is bindmounted and its device is the hostedevice [00:51] when unmounted, even if the mountpoint is passed umount as opposed to the device, the umount -f -d arguments seem to be lethal [00:53] cjwatson, evand, comments? [00:54] is passed TO umount [05:15] cjwatson, hai [05:26] cjwatson, i want to talk about the live cum install cd [05:47] cjwatson, i want to talk about the live cum install cd === ceeka1 is now known as seekay [09:27] debian-installer: cjwatson * r871 ubuntu/ (10 files in 4 dirs): * Move to 2.6.24-5 kernels. [09:44] evand, cjwatson did you see my messages yesterday? [09:44] in short we still have issues with umountfs because of the umount flags used [09:46] the proposed solution is to keep track of both mountpoints and devices in top part of /proc/mounts (above /) [09:47] then skip the mountpoints that match, and for the devices that match use umount without -f flag [09:47] -f -d [09:54] xivulon: please send me a patch using diff -u and I'll take a look. [09:57] ok [10:05] evand do we need to skip securityfs? [10:05] like we do for procfs|linprocfs|devfs|sysfs|usbfs|usbdevfs|devpts? [10:06] xivulon: re console-setup/layoutcode, en_GB is flat wrong. gb is correct. [10:06] the names correspond to files in /usr/share/X11/xkb/symbols/ [10:22] hmm cjwatson I had that both in /proc/cmdline and in preseed but did not see that having any effect in installer/syslog and certainly the layout was us [10:22] locale instead worked as expected [10:24] when I do protected_mounts=$(sed -n '0,/^\/[^ ]* \/ /p' /proc/mounts) the eol is treated differently according to the shell used [10:26] how do I force eol in there? [10:27] xivulon: if it's not working then that sounds like a bug; changing the preseeding would at best be a workaround [10:29] My first thought was that my preseed was wrong [10:41] sh q: how do I keep end of lines in x=$(cat /proc/mounts) ? [10:45] quote it properly: x="$(cat /proc/mounts)" [10:45] the trailing newline at the very end will be stripped (so add it back in as necessary) but not the others [10:46] in general in shell, all $ expansions should be surrounded by "" unless you know better [10:46] (there are some specific cases where they shouldn't be, but those are the exception rather than the rule) [10:47] cjwatson I am trying to do: [10:47] x="$(cat /proc/mounts)";echo $x | grep "^/dev" [10:48] echo "$x" [10:48] the quoting rule applies everywhere :) [10:48] ah [10:52] evand, cjwatson see if the following will do: http://paste.ubuntu-nl.org/53419/ [10:52] have not tasted it [10:52] diff http://paste.ubuntu-nl.org/53420/ [10:54] ops diff is other way around [10:55] http://paste.ubuntu-nl.org/53421/ [10:59] teake 3: http://paste.ubuntu-nl.org/53422/ [11:02] did not test it though and cannot test until tonight, one issue I can thing of is the umount order is altered [11:15] take 4: http://paste.ubuntu-nl.org/53426/ [11:16] this should preserve the umount order [11:16] again, I did not taste it, cjwatson, evand, have a look [11:43] evand, cjwatson, another issue I noticed is that the suspend button is correctly hidden in loopinstallations, but the hibernate button is still there [15:07] ubiquity: evand * r2417 ubiquity/ (bin/ubiquity-dm debian/changelog): * Add usability support to 'only-ubiquity'. [15:10] xivulon: if you decide you want to add accessibility support to Wubi, it's as simple on the Ubiquity side as adding access= and a proper code (see casper-bottom/30accessibility) after the -- on the kernel cmdline. [15:21] evand yes absolutely [15:22] I'd say that after we have this working okish (=update-grub+umounhost+no-hibernation+layoutcode-preseeding) I will look into that [15:23] I only need to know what parameters ubiquity takes in that respect and how to fetch the info within windows to decide whether to use those parameters [15:24] umountfs* [15:29] it's v1, v2, v3, m1, m2 correct? Now I only have to find how to map registry keys to that, anyone has any clue? [15:31] http://msdn2.microsoft.com/en-us/accessibility/bb879984.aspx [15:33] in the link above see HCI profile [15:34] They have more categories then v/m though. [15:35] mpt what would you suggest about detecting windows accessibility settings and mapping them to ubuntu's? [15:37] I'd go through apps registered as mandatory in Ease of Access Center, and look at their HCI profile: [15:37] mild vision->v2, sever vision->v3, mild_dexterety->m1, severe_dexterety->m2 [15:37] speech/hearing/cognitive are ignored [15:38] xivulon, I'd need to sit down with screenshots of both to give any useful advice on that [15:40] can you maybe draft a page on how to do said mapping? I know next to nothing on the subject based my simple proposal only on 2 minutes googling [15:48] xivulon, ok, report a bug and assign it to me for the design, and I'll reassign it to you when I've attached it [15:52] under what project? [15:53] I mean it can be either wubi or a branch off migration assistant [15:53] evand what's your take on that? [15:55] why would this be part of migration-assistant? [15:55] if people install off CD, it would be nice to find out existing accessibility settings [15:56] not sure whether that is provided already [15:58] I assume they have the initial boot options in that case though [15:58] I'll that in wubi then [15:58] do^ [15:59] It's part of the initial menu. It occurs to me that I never implemented the migration-assistant defaults seeding. [15:59] I'm not sure if it would be wise to add accessibility detection to m-a as that fits in with language and timezone selection, but does not have a part of the ubiquity UI like they do. [16:00] I still think it would be great to have in wubi though. [16:00] be it automatically detected or another drop down box. [16:06] mpt, evand bug #185954 [16:06] Launchpad bug 185954 in wubi "Detect accessibility settings" [Medium,Confirmed] https://launchpad.net/bugs/185954 [21:08] my network card is not supported by the modules included in the debian-installer initrd... is there an easy way to unpack the initrd + add in modules + repack? [21:15] in particular i can't seen to mount the initrd as a loopback device and peek inside... complains about you must specify a filesystem type [21:16] is the initrd contained in mini.iso in fact a gzipp'ed ext2 filesystem or something different? [21:24] well gzip -dc initrd.gz | cpio -id seems to do the trick [21:52] test....