[14:23] cjwatson: how was the original timezone map done? Can that be updated, cause a few countries have changed "default" timezones. [14:35] xnox: I'm not entirely sure - ask ev [14:39] we're chatting about it now [17:02] If one hits quit in the first page of the installer (ubiquity), is the designed behavior to start a lilve session or to go back to the do you want to try or install screen? [17:02] live session iirc [17:03] Thanks. [17:03] Then it's working ... [19:36] If an install is apparently halted at system installation 67% and Ubiquity is eating ~all the CPU, any immediate suggestions on how to figure out where I'm stuck (this is an encrypted LVM install)? [19:38] xnox: ^^^ suggestions? [19:45] ScottK: Nothing interesting in syslog? [19:45] * ScottK looks [19:47] It says something about media change. [19:47] The USB is there and mounted though. [19:48] There's also a depmod error a little bit further up about failed to symbolic link [19:48] Linux-image postinst failed. [19:50] postinsts failing sounds bad. [19:50] The installer should probably notice such things and error out, instead of appearing to hang, mind you. :P [19:50] infinity: http://paste.ubuntu.com/1413155/ [19:50] But that's probably the problem. [19:51] I did encrypted LVM on this system on Monday and it worked. [19:51] * ScottK wonders did something change or is it gremlins? [19:51] Oh, that may not be a failed postinst, just a verbose one. [19:52] Ah. Right. [19:52] It's certainly the media change event there that screwed you. [19:52] But why or how that happened, I dunno. [19:52] OK. [19:52] Maybe I bumped the usb stick. [19:52] Unless your CD is actually incomplete. [19:53] I've done several installs with it already. [19:53] (That's the same message you get with d-i on multiple-CD installs... Which we don't do) [19:53] Kay, if it's worked before, you may have just had some cosmic rays. [19:54] Or /cdrom/ genuinely has nothing in it due to some other random bug umounting or mounting over it. [19:54] If you remember how you got there, I'd give it a whirl to reproduce. [19:54] I'm trying again. [19:54] I did check that the USB was present and mounted. [19:55] So perhaps I bumped it or something and then Ubiquity failed to notice when it reappeared? [19:55] Possibly. Dunno. A lot of maybes at this point. [19:55] Ones I'll fail to care about if a second try works. [19:55] It may be high time for us to rename /cdrom/ in casper/d-i/etc to avoid looking like we live in the past. [20:06] I'm glad I've done this install enough times that knowing the language I'm installing in isn't important. [20:20] Heh. [20:20] I still get messed up sometimes when I test software in right-to-left languages. [20:31] Officially gremlins I guess. It's past the point it hung last time. [21:37] Here's a bit of a conundrum ... [21:37] When installing with lvm/encryption, passphrase entry is done before you select your keyboard. [21:38] So if it turns out that your keyboard selection is wrong and you fix it, then the keyboard layout you install with is different than the one you set your passphrase with. [21:38] Which can make some passphrases hard to type ... [21:46] That seems a bit backward. Shouldn't language/keyboard be the first things asked? [21:46] xnox: You still around? [21:47] We've been round in circles with design on this for years. [21:47] A quick IRC discussion won't solve it. [21:48] Kay.. [21:48] Bringing partitioning forward was a genuine win in many ways (it's one of the things people often praise when they talk to me about the installer in person), but it does raise this awkward case [21:48] The best compromise we've found was to add a keyboard indicator [21:48] I don't recall what the KDE frontend has, though [21:49] Yeah, but the keyboard indicator does you no good if you don't realise this particular problem will be a problem. [21:49] Especially since passphrases are blind, so you may not know you're typing it "wrong". [21:50] Yeah, I guess it may be worth providing some kind of hint specially for the crypto case [21:51] We can't just pull keyboard back though - that involves pulling location back as well (for sane defaults) and at that point we've lost most of the gain from parallel questions