[04:36] <holstein> cjwatson: i did get around to trying the install with the ubuntustudio live DVD.. just one hitch post-install that im sure someone knows about somewhere.. https://bugs.launchpad.net/ubuntu/+source/lightdm/+bug/922424
[04:36] <ubot2`> Launchpad bug 922424 in lightdm "lightdm won't start I have console(s) only" [Undecided,Confirmed]
[04:37] <holstein> anyways.. thanks again... sleeping..
[15:49] <CIA-11> ubiquity: cjwatson * r5132 trunk/tests/ (test_misc.py test_osextras.py): Fix some test isolation bugs.
[16:32] <jibel> bug 922640
[16:32] <ubot2`> Launchpad bug 922640 in ubiquity "Precise Pangolin Installer crashed on Sony Vaio VGN-SR39VN" [Undecided,New] https://launchpad.net/bugs/922640
[16:32] <jibel> I haven't verified it yet, but it seems related to the change with resolvconf
[16:36] <cjwatson> stgraber: another one for you then :)
[16:36] <cjwatson> actually never mind, I can just fix that
[16:37] <stgraber> it's going to be loooong day ;)
[16:38]  * stgraber is getting pinged in at least 5 different channels for resolvconf related issues :)
[16:46] <CIA-11> ubiquity: cjwatson * r5133 trunk/ (debian/changelog scripts/plugininstall.py): Cope with /etc/resolv.conf being a symlink (LP: #922640).
[16:47] <ogra_> cjwatson, out of curiosity, why cant d-i produce a netinst initrd from a PPA build ?
[16:49] <ogra_> that doesnt involve any debian-cd stuff, just needs udebs available, or am i wrong here ?
[16:52] <cjwatson> ogra_: assuming that Launchpad will publish the custom upload to a PPA!
[16:52] <ogra_> well, indeed, that ssumes that both, d-i and kernel are in that PPA and the binaries end up in that very same PPA
[16:52] <cjwatson> missing my point
[16:53] <cjwatson> custom uploads are a very specialised upload type
[16:53] <cjwatson> afaik they have only ever been performed to the primary archive
[16:53] <cjwatson> they require lots of specialised code in Launchpad to handle them
[16:53] <cjwatson> this is the *output* from the d-i build
[16:53] <cjwatson> ogra_: looking through the Launchpad code, I can't actually find anything that prevents it right now, so it's possible I'm wrong; but you're well into untested territory, I think
[16:53] <ogra_> hmm, k
[16:54] <cjwatson> and in any case I don't like the idea of branching the d-i package, it's hard enough to maintain sanely as it is
[16:54] <ogra_> yeah, thoug *if* we have to branch, the out of archive solution seemed saner to me
[16:54] <ogra_> if LP gets in our way thats indeed bad
[16:54]  * GrueMaster never thought of d-i as sane in the firstplace.
[16:55] <cjwatson> :-P
[16:55]  * ogra_ thinks infinity was also in favour of thje PPA idea though i dont know if he knows the LP specifics you mention
[16:55] <cjwatson> I guess you can try it and see if it works.  You should try it before making any plans that assume it works.
[16:55] <ogra_> indeed, though its NCommander's project and he seems to want to go another route
[16:56] <ogra_> with a second d-i in universe or so
[16:58] <cjwatson> I wouldn't accept that into the archive if I were the archive admin on duty
[16:58] <cjwatson> he hasn't talked to me about it
[16:59] <cjwatson> I really don't think it's appropriate to fork d-i in universe.  It buys absolutely nothing, since LP would still publish the output to main.
[16:59] <cjwatson> If we're going to have it in the primary archive, we might as well just have the existing debian-installer package build it out of universe.
[16:59] <ogra_> right, if thats the case we wont gain a thing
[16:59] <cjwatson> It is the case; I have checked the code.
[16:59] <ogra_> k
[16:59] <cjwatson>         self.targetdir = os.path.join(
[16:59] <cjwatson>             archive_root, 'dists', distroseries, 'main',
[16:59] <cjwatson>             '%s-%s' % (build_type, self.arch))
[16:59] <ogra_> yep
[17:00] <cjwatson> And there's no overrides facility that could be used to change that.  You'd have to build an entire new edifice of stuff.
[17:00] <ogra_> yeah, understood
[17:00] <infinity> Yeah, the custom upload processor stuff was never designed, per se.
[17:00] <ogra_> sounds like there is no sane solution at all
[17:00] <infinity> Just kinda slapped in.
[17:02] <cjwatson> It was inserted with a drill.
[17:02] <infinity> ogra_: The "do it in a PPA" idea was predicated on the idea that the final build would be changed to make the by-hand tar.gz be a .deb someone could move around and publish elsewhere.  It was a thought I never bothered to finish because it was just us banging out ideas for interim solutions (where I still say the best solution for a kernel and d-i we don't want to support is just to build it locally)
[17:02] <ogra_> ah, ok
[17:03] <infinity> But we could also alter d-i to build against universe for specific targets, I suppose.  That could introduce other hilarity where, for instance, we currently demote udebs to universe when we don't want d-i to see them. :P
[17:05] <ogra_> oh, yeah, indeed ... fun ahead
[17:08] <ogra_> i wonder if we couldnt just push the kernel to main and demote it right after the last d-i upload
[17:09] <ogra_> (and likely mangle the d-i output for this kernel to not end in main at all)
[17:09] <stgraber> cjwatson: seems like d-i has a similar issue trying to copy /etc/resolv.conf (in 40netcfg based on the log I have here)
[17:10] <cjwatson> can I see the log?  /target/etc/ should be empty ...
[17:10] <cjwatson> not necessarily the case in ubiquity
[17:10] <cjwatson> ogra_: that's worse
[17:10] <cjwatson> better to have d-i be honest
[17:11] <cjwatson> ogra_: your plan would make it hard to do post-release uploads
[17:11] <ogra_> well, actual post release uploads of that kernel would likely be with a supported version
[17:11] <cjwatson> just please don't play silly archive games
[17:11] <ogra_> what michael tries here is to fill the gap until a supported kernel exists i think
[17:12] <ogra_> k
[17:12] <cjwatson> if d-i is using a kernel from universe, it should say so
[17:12] <ogra_> oh, i didnt mean to hide that fact indeed
[17:12] <ogra_> just to make the building easier
[17:14] <stgraber> cjwatson: oh, sounds like my bad then (and GrueMaster's). We both had mirror issues so at least I probably have a dirty /target, he might well have had that too
[17:16] <GrueMaster> wait, what was my bad???
[17:17] <ogra_> GrueMaster, everything !
[17:17] <cjwatson> stgraber: can I see the log anyway?
[17:17] <GrueMaster> I don't break things, I only find weak points in others work.
[17:20] <stgraber> cjwatson: http://paste.ubuntu.com/819048/
[17:20] <stgraber> cjwatson: my DNS server being stupid made that download fail.
[17:22] <cjwatson> ah, right.  we should be robust anyway.
[17:23] <cjwatson> but that code looks wrong in resolvconf land
[17:23] <cjwatson> it'll cause all new systems to start out with /etc/resolv.conf as a file and resolvconf will ignore them
[17:23] <cjwatson> can you suggest a replacement?
[17:28] <stgraber> cjwatson: what was done in LXC and live-build is basically copy the content of /etc/resolv.conf from the outside to the target during install/build time, then at the end restore the symlink (moving it initially to .orig and moving back .orig to resolv.conf at the end)
[17:29] <cjwatson> d-i needs to preserve manually configured nameservers
[17:29] <cjwatson> in static network configuration
[17:31] <stgraber> right, these should be converted to dns-search and dns-nameservers entry in /etc/network/interfaces
[17:32] <cjwatson> is that new with resolvconf?
[17:33] <stgraber> it's implemented through an ifupdown hook shipped by resolvconf but I don't think it's particularly new
[17:33] <stgraber> so if we know we have resolvconf on the target, we should be setting these, otherwise we want to copy /etc/resolv.conf as we used to
[17:45] <bdmurray> the launchpad.net/wubi/ page has an external downloads link which points to sourceforge which has a very old version of wubi
[17:46] <bdmurray> would linking to evan's home directory on people be better?
[17:47] <cjwatson> yeah
[17:47] <cjwatson> stgraber: hah, netcfg already writes dns-* as well :-)
[17:48] <cjwatson> so we can just not copy resolv.conf if it's a symlink
[17:48] <cjwatson> er if resolvconf is installed
[17:48] <stgraber> cjwatson: sounds good
[17:50] <CIA-11> netcfg: cjwatson * r1283 ubuntu/ (base-installer.d/40netcfg debian/changelog):
[17:50] <CIA-11> netcfg: Don't copy /etc/resolv.conf to target if resolvconf is installed. (We
[17:50] <CIA-11> netcfg: already write resolvconf configuration to /etc/network/interfaces.)
[17:51] <CIA-11> netcfg: cjwatson * r1284 ubuntu/debian/changelog: releasing version 1.68ubuntu8
[17:52] <stgraber> cjwatson: thanks!
[17:52] <antarus> cjwatson: you test that dhcpv6 timeout patch I sent?
[17:53] <cjwatson> antarus: oh, no, sorry, have been swamped - I've made a note to do it next week
[17:56] <antarus> ok no worries
[17:56] <antarus> I think the deadline for P was in mid-feb?
[17:56] <antarus> so we are trying to get all our patches upstreamed before then
[17:57] <cjwatson> for bug fixes the deadline is still ages away
[17:57] <cjwatson> feature freeze is soonish
[17:58] <antarus> ipv6 is a feature! :)
[17:58] <antarus> but good to know ;)
[17:58] <cjwatson> the installer already supports ipv6, this is a bug fix :)
[18:05] <CIA-11> ubiquity: cjwatson * r5134 trunk/gui/gtk/stepWebcam.ui: typo
[18:08] <bdmurray> cjwatson: as far as I can tell data/isolist.ini for wubi 245 doesn't seem to have release.ubuntu.com in it so is downloading precise images
[18:08] <bdmurray> I'm looking at http://bazaar.launchpad.net/~ubuntu-installer/wubi/trunk/revision/247 and bug 922330
[18:08] <ubot2`> Launchpad bug 922330 in wubi "Wubi.exe from 11.10 disc installs 12.04 snapshot." [High,Confirmed] https://launchpad.net/bugs/922330
[18:11] <bdmurray> this is easier to parse http://bazaar.launchpad.net/~ubuntu-installer/wubi/trunk/view/245/data/isolist.ini
[18:18] <CIA-11> ubiquity: cjwatson * r5135 trunk/ (debian/changelog src/webcam/webcam.c ubiquity/gtkwidgets.py):
[18:18] <CIA-11> ubiquity: Scale down the webcam page a bit so that it works on netbook screens
[18:18] <CIA-11> ubiquity: (LP: #869239).
[18:19] <CIA-11> ubiquity: cjwatson * r5136 trunk/ (debian/changelog tests/test_gtkui.py): Improve test_pages_fit_on_a_netbook so that it tests all pages.
[18:24] <CIA-11> ubiquity: cjwatson * r5137 trunk/src/webcam/webcam.c: shrink webcam page slightly further so that the tests really pass
[18:24] <cjwatson> bdmurray: sorry, I already have four more things I need to do before I can finish for the week and it's 6:25pm on Friday :-(
[18:24] <cjwatson> anyway I see releases.ubuntu.com in the URL you give ...
[18:26] <bdmurray> In the first url?  That's what was being changed in rev 247 the 245 rev link shows it using cdimage
[18:26] <bdmurray> Anyway I've assigned it to the team
[18:29] <CIA-28> ubiquity: cjwatson * r5138 trunk/ (d-i/manifest debian/changelog):
[18:29] <CIA-28> ubiquity: Automatic update of included source packages: choose-mirror 2.39ubuntu2,
[18:29] <CIA-28> ubiquity: debian-installer-utils 1.88ubuntu2, flash-kernel 2.28ubuntu37,
[18:29] <CIA-28> ubiquity: partman-base 153ubuntu1.
[18:29] <cjwatson> oh, in diskimage=
[18:29] <cjwatson> not sure what we can do about that now.
[18:29] <cjwatson> 247 was the panic last-minute respin wasn't it
[18:31] <CIA-28> ubiquity: cjwatson * r5139 trunk/tests/ (test_misc.py test_osextras.py): Make pyflakes pass again.
[18:47] <CIA-28> ubiquity: cjwatson * r5140 trunk/debian/changelog: releasing version 2.9.11
[19:40] <cjwatson> argh, ubiquity was uploaded outside revision control
[20:32] <CIA-28> ubiquity: cjwatson * r5141 trunk/debian/changelog: merge 2.9.11, released outside bzr
[20:56] <CIA-28> ubiquity: cjwatson * r5142 trunk/debian/changelog: releasing version 2.9.12
[23:10] <CIA-28> ubiquity: cjwatson * r5143 trunk/ (debian/changelog tests/test_osextras.py): Make test_osextras reliable across architectures (sort lists).
[23:18] <CIA-28> ubiquity: cjwatson * r5144 trunk/ (d-i/manifest debian/changelog): Automatic update of included source packages: netcfg 1.68ubuntu8.
[23:29] <CIA-28> ubiquity: cjwatson * r5145 trunk/debian/changelog: releasing version 2.9.13