[02:29] <FourDollars> cyphermox: https://code.launchpad.net/~ubuntu-installer/ubiquity/trusty-proposed is out to date. How can I provide the patch for ubiquity?
[02:29] <FourDollars> s/to/of/
[03:57] <cyphermox> FourDollars: I'm updating it, should just take a minute
[03:58] <FourDollars> cyphermox: Thx
[15:22] <Laney> cyphermox: let's chat installer (hi!)
[15:24] <cyphermox> Laney: hey!
[15:25] <Laney> :)
[15:25] <Laney> so look at this bit of code: https://bazaar.launchpad.net/~ubuntu-installer/ubiquity/trunk/view/head:/scripts/plugininstall.py#L1304
[15:25] <Laney> and also at a manifest-remove file in an iso (casper/)
[15:26] <Laney> some of them are suffixed with :arch and they fail to be considered by this code meaning that too many packages are removed in some circumstances
[15:26] <Laney> I guess we could fix this at either end - which would you prefer?
[15:27] <Laney> whoops I linked you to the wrong bit: https://bazaar.launchpad.net/~ubuntu-installer/ubiquity/trunk/view/head:/scripts/plugininstall.py#L1411
[15:29] <cyphermox> oh, fcitx...
[15:29] <Laney> ya
[15:29] <Laney> but really a ubiquity bug
[15:29] <cyphermox> sure
[15:29] <Laney> so you can do something like split(':')[0]
[15:30] <cyphermox> ubiquity should understand this, yeah
[15:30] <Laney> ok cool, two minutes
[15:30] <cyphermox> I'm wondering if there is a case where you might want to remove just one arch instance of a package.
[15:33] <Laney> is that theoretical currently?
[15:34] <Laney> :)
[15:38] <cyphermox> yes, but it's probably not hard to teach ubiquity to treat this properly
[15:38] <cyphermox> cf. https://bazaar.launchpad.net/~ubuntu-installer/ubiquity/trunk/view/head:/ubiquity/install_misc.py#L594
[15:39] <cyphermox> that said, if you don't want to touch that right now, could you file a bug so I can get to it somewhere in the future? :)
[15:45] <Laney> I will do, thanks
[15:47] <Laney> also... that werewolf is really scary
[16:08] <cyphermox> yeah :D
[16:15] <cyphermox> slangasek: infinity: I think I have the check for "PReP on RAID" good enough, just need to update the template for the error message. it will block going any further than partitioning if the partition isn't on a non-md device.
[18:03] <slangasek> cyphermox: mm.  it shouldn't even give PReP as an *option* for unsuitable partitions; but there needs to be a finish.d check to make sure that the partition table as a whole is valid and has at least one PReP partition
[18:14] <cyphermox> That's exactly what I did
[20:40] <infinity> cyphermox: When you say "non-md", you mean "non-md, with the exception of raid1", I guess?
[20:41] <infinity> cyphermox: Although, that assumes the raid1 superblock wouldn't blow up a PReP partition, I suppose that's worth testing before claiming it works. :P
[20:41] <cyphermox> no, I mean non-md. looks like things blow up anyway at least in grub-installer, because grub-install still finds those to be md devices and decides it can't find md0,2 or whatever.
[20:42] <cyphermox> I wasn't quite able to fix that just yet, so I'll have it disabled for now when it's RAID; until I can fix grub-installer and grub to handle things properly
[20:42] <cyphermox> maybe it's just missing some module, even, but they did seem to be there.
[20:43] <infinity> cyphermox: Huh.  Is that unique to the prep-bootdev thing, or grub-installer in general hates on partitioned full-disk md?
[20:43] <cyphermox> I already hacked up grub-installer some to try to install to the underlying devices, but I didn't quite get it right.
[20:43] <cyphermox> unique to non-BIOS x86.
[20:43] <cyphermox> I mean
[20:43] <infinity> non-BIOS-x86? :)
[20:43] <cyphermox> it only works there, it won't work on EFI, and unsure whether it works properly for PReP
[20:44] <infinity> Right.  Kay.
[20:44] <infinity> Definitely worth a revisit, but not right before a release.
[20:44] <cyphermox> TL;DR: it needs more testing
[20:44] <cyphermox> I think not allowing install of PReP on a RAID will do the trick for now until we can get the install just right
[20:46] <cyphermox> my code for grub-installer is most likely wrong but already on my PPA so I can iterate on the wrongness.