[00:00] <cjwatson> I'm not sure
[00:00] <infinity> Well, the fix in ubiquity is fairly obvious, so I'll test that.
[00:01] <cjwatson> d-i might get away with it on grounds of the relevant bit of c-s only being called rather late, or maybe during base-installer while update-initramfs is diverted
[00:01] <cjwatson> (sorry, buried in Qt localisation stuff at the moment)
[00:01] <infinity> Yeah, s'all good.
[00:01] <infinity> You gave me enough of a hint to deal with ubiquity here.  I might poke d-i later.
[00:02] <cjwatson> moving the diversion should be safe if it passes a basic install attempt, and should make ubiquity faster, so
[00:02]  * infinity nods.
[00:02] <infinity> I'm sure you remember livecd-rootfs.  I'm a big fan of divert early, divert often. :P
[00:02] <cjwatson> by quite a bit, on many machines
[00:02] <cjwatson> uh-huh
[01:07] <CIA-45> ubiquity: cjwatson * r5007 trunk/ (bin/oem-config-wrapper debian/changelog):
[01:07] <CIA-45> ubiquity: Always start oem-config with LANG=en_US.UTF-8; this doesn't affect
[01:07] <CIA-45> ubiquity: translations since we use debconf for those, but forces Qt to use UTF-8
[01:07] <CIA-45> ubiquity: encoding for text fields (LP: #837042).
[01:08] <cjwatson> infinity: if you get that fixed tonight and fancy uploading, be my guest
[01:09] <infinity> cjwatson: Yeah, I plan on it.  Dinner happened. ;)
[01:09]  * cjwatson nods
[01:10] <cjwatson> Sleep is about to happen here
[01:10] <infinity> I should say so.
[01:10] <infinity> Enjoy.
[01:10] <infinity> And happy weekending.
[01:15] <cjwatson> You too
[05:26] <infinity> cjwatson: So, I just spent countless iterations rewriting and debugging various bits of ubiquity to use chroot_setup more consistently.  And yet, none of it fixed my oem-config issues.  Because I'm just smart enough to have missed:
[05:26] <infinity>     if target == '/':
[05:26] <infinity>         return
[05:27] <infinity> cjwatson: So, I suspect the real solution here is just to comment out the update-initramfs call in console-setup-apply.  I'm not entire sure what it's doing there anyway.  (Testing that 1-character fix now)
[06:45] <infinity> cjwatson: And the real (really, honest) problem actually seems to be that flash-kernel-installer just plain isn't being run.  And that has me stumped.
[07:12] <infinity> How.  Did.  This.  Ever.  Work?
[07:21] <CIA-45> ubiquity: adconrad * r5008 trunk/.bzrignore: Add d-i/sources.list.local to .bzrignore
[08:51] <cjwatson> infinity: err ... please don't, it's needed in general because c-s configuration is copied into the initramfs
[08:53] <cjwatson> infinity: I would be orders of magnitude more comfortable with a fix local to ubiquity than one that mangles c-s directly
[08:59] <infinity> cjwatson: c-s-apply is local to ubiquity.
[08:59] <infinity> cjwatson: But that's not actually the problem here anyway.
[08:59] <infinity> cjwatson: The problem is that flash-kernel-installer just plain isn't being run.
[09:00] <infinity> So, I figured this might relate:
[09:00] <infinity>                  elif (arch == 'armel' and
[09:00] <infinity> -                      subarch in ('dove', 'imx51', 'iop32x', 'ixp4xx', 'orion5x', 'omap')):
[09:00] <infinity> +                      subarch in ('dove', 'imx51', 'iop32x', 'ixp4xx', 'orion5x', 'omap', 'omap4')):
[09:00] <infinity>                      from ubiquity.components import flash_kernel
[09:00] <infinity> But hey, it still doesn't get called.  And now I'm stupidly confused as to (A) how this ever worked, and (B) why I can't make it work now. :/
[09:04] <cjwatson> local to ubiquity> oh, ok
[09:05] <cjwatson> is archdetect returning the right answer?
[09:07] <cjwatson> and if it is, um, have you straced or otherwise definitely confirmed that it's not being called as opposed to just not doing anything?
[09:13] <infinity> cjwatson: It's the only thing in components/ that doesn't have a .pyc to go with the .py, which seems fairly damning.
[09:13] <infinity> cjwatson: And archdetect is returning armel/omap4, which seems right.
[09:14] <cjwatson> something isn't trying to run it as non-root for some reason?
[09:15] <cjwatson> I would definitely be attaching strace -f to plugininstall once it starts at this point, though
[09:36] <infinity> Yeah, I'll try to dig deeper in a bit.  Currently unbricking a mobile phone.  It's been one of those nights.
[11:07] <infinity> Well, phone unbricked.  That's progress, right?
[19:56] <CIA-45> ubiquity: cjwatson * r5009 trunk/debian/changelog: releasing version 2.8.0
[21:03] <CIA-45> ubiquity: cjwatson * r5010 trunk/ (d-i/update-control debian/changelog debian/control): Build-depend on gir1.2-soup-2.4 for timezone tests.
[21:52] <CIA-45> ubiquity: cjwatson * r5011 trunk/debian/ (74 files in 2 dirs): Update translations from Launchpad.
[22:07] <CIA-45> ubiquity: cjwatson * r5012 trunk/debian/ubiquity.templates-imported: fix attributions
[22:17] <CIA-45> ubiquity: cjwatson * r5013 trunk/debian/ (39 files in 2 dirs):
[22:17] <CIA-45> ubiquity: Update imported translations from gtk+3.0 3.2.0-0ubuntu1 and gnome-panel
[22:17] <CIA-45> ubiquity: 1:3.2.0-0ubuntu1.
[22:25] <CIA-45> ubiquity: cjwatson * r5014 trunk/debian/changelog: releasing version 2.8.1