[02:47] <CIA-87> usb-creator: superm1 * r324 usb-creator/debian/changelog: add bug number
[09:13] <ara> ev, good morning
[09:53] <ev> hi ara
[09:53] <ev> good morning
[09:53] <ara> ev, morning!
[09:53] <ara> ev, I pinged you because I wanted to ask you about your automated testing of the installer
[09:53] <ara> ev, how is that going?
[09:54] <ev> waiting on IS to give me space in the Millbank datacenter
[09:55] <ev> ara: RT 40560 and RT 40559
[09:56] <ara> ev, and in your local server? does it work already?
[09:57] <ev> as far as what I've tested previously.  I still need to do a bit of work on the PXE side of things.
[09:57] <ev> but yeah, once a machine starts booting the modified Ubuntu image, it's golden
[10:04] <ara> ev, great
[10:04] <ev> I've been a bit tied down with bug fixing at the moment, but I hope to finish it off this weekend / next.  Hopefully by then IS will have space ready.
[10:05] <ara> ev, are you running it in your local server on a regular basis? (and reporting back?)
[10:05] <ev> next> next week, not next weekend
[10:05] <ev> no, though I probably should start
[10:32] <ev> ugh
[10:33] <ev> apparently I didn't think through this timezone stuff enough
[10:33] <ev> a latitude longitude pair is not enough, as a location might be physically closer to one timezone than another, but still be in the further away one.
[10:34] <ev> fairly obvious, but I completely missed it :-/
[11:03] <ev> so I need to either layer another free service on top ( http://services.gisgraphy.com/geoloc/geolocsearch?lat=26.566460&lng=-80.053370&radius=&placetype=city&format=XML&from=1&to=10 ), or construct a more detailed map that color codes all the polygons, not just the timezone bands. This: http://efele.net/maps/tz/world/ has the latter, but is the wrong map projection.
[11:49] <ev> oh thank god, someone's done the work: http://blog.makerlab.org/2009/05/olson-code-timezones-geekery/
[11:49] <ev> I can probably build something better off the efele.net stuff, but this will be a good start
[11:58] <cjwatson> "I'd benchmark it if there was any conceivable contender for a faster solution" :-)
[12:06] <ev> heh
[13:07] <JeffP-BR> Hello All. I'm in trouble installing Ubuntu Server from a pen drive in a machine without cdrom drive. I've wrote the ISO image of Ubuntu Server 10.10_beta on the USB drive and the boot process is ok but the ubuntu installer stay asking for a CD rom drive. Does anybody can help me to solve this issue?
[13:08] <ev> put cdrom-detect/try-usb=true on the kernel command line.  usb-creator does this for you.
[13:09] <JeffP-BR> Hello ev. Thanks. I'll try this.
[14:49] <ev> okay, we have a solution for the hostname thing
[14:50] <ev> I've spoken with Ivanka and we're going to put it back in with some minor changes
[14:52] <ev> the autogenerated hostname will be username-model
[14:52] <cjwatson> model?
[14:53] <ev> where model is grabbed from dmidecode, if set, otherwise it's  username-desktop/laptop or username-ubuntu
[14:54] <ev> I have no preference on username-chassis vs username-distribution
[14:54] <ev> sorry, spoken with Ivanka and mpt to be clear
[15:06] <persia> Just as a side note, I have more than one product where the dmidecode System Information/Product Name is "To Be Filled By O.E.M.".  I have other systems where dmidecode is not available.
[15:07] <highvoltage> hi! are there issues on i386 with ubiquity atm? on Edubuntu, it stops at the timezone selection (but continues to install) and doesn't show the ubiquity-slideshow
[15:07] <highvoltage> it works fine on amd64 though
[15:08] <superm1> highvoltage, this is with current dailies?
[15:09] <highvoltage> superm1: yep
[15:10] <highvoltage> I'll try with an ubuntu daily next just to make sure that it's not edubuntu specific
[15:10] <superm1> highvoltage, can you check if both your amd64 and i386 have the new ubiquity (2.3.17) or if one of them failed to build and it's using the older one?  that's quite peculiar
[15:10] <highvoltage> they're both on 2.3.17
[15:11] <superm1> OK then when the i386 gets to the point that it can't proceed any more, open a terminal (ctrl-alt-t will do it) and run ubuntu-bug ubiquity
[15:11] <highvoltage> ah, the timezone window eventually loaded, it was just slow on i386, weird
[15:11] <superm1> oh, are you perhaps on a network that can't get to the internet?
[15:12] <highvoltage> it has internet
[15:12] <superm1> or needs a proxy to do so at least
[15:14] <highvoltage> doesn't need a proxy either
[15:19] <superm1> not anything that stands out to me for what's causing it then
[15:23] <highvoltage> superm1: turns out I was the problem. I expected a timezone to already be selected when the timezone dialog loaded and that the forward button would be selectable (I think it was like that in previous versions of ubiquity)
[15:23] <highvoltage> sorry :)
[15:23] <superm1> well a timezone should have already been selected..
[15:23] <superm1> it uses a geoip service to try to choose one for you automagically when you have internet access
[15:25] <highvoltage> I guess something goes wrong there then. it's in a virtual machine that's nat'd, but it has internet access without requiring a proxy being set up
[15:30] <davmor2> ev: I'm taking it the translations team are still working on the text in the slides?
[15:31] <ev> they will hopefully be when it lands
[15:31] <ev> it's currently blocked on Dylan finishing it
[15:33] <superm1> highvoltage, i just compared 2.3.16 to 2.3.17 and it looks fine to me, my timezone was picked in both.. if you keep running into it, please run ubiquity in debug mode and file a bug
[15:33] <Riddell> ev: are you still expecting to install libdvdcss from ubiquity?
[15:34] <ev> Riddell: yeah, it will require an exception, but I planned on doing that this weekend
[15:34] <superm1> how?  it's not in any repos?
[15:34] <Riddell> ev: any UI to that or will it just magically work with KDE UI?
[15:34] <ev> superm1: /usr/share/doc/libdvdread4/install-css.sh
[15:34] <ev> Riddell: magic
[15:35] <Riddell> cor
[15:36] <superm1> ev, that just frightens me a bit, wget'ing from a non archive site that hasn't been verified
[15:37] <ev> superm1: well, that's why it's behind a checkbox.  It's the user's choice to do so.
[15:42] <highvoltage> superm1: ok, will do
[15:56] <ev> thoughts on this: http://paste.ubuntu.com/491013/
[15:56] <ev> ?
[16:20] <ScottK> Riddell and ev: default hostname on ubiquity-kde 2.3.17 is blank (KDE U/I has always kept it).
[16:38] <ScottK> Riddell and ev: with ebiquity 2.3.17, the install options I get are "Install them side by side...", "Erase and use entire disk", and "Specify partitions manually (advanced)".  Are those all the ones I'm supposed to have?
[16:39] <ev> well, there's also use largest free space, but I haven't implemented that yet
[16:39] <ev> we removed it from the gtk frontend
[16:40] <ScottK> OK.
[16:41] <ScottK> Also the hostname was pre-filled on the full disk install, just not side by side.
[16:41] <ev> eh?
[16:41] <ScottK> We've kept the hostname in the -kde dialogue all along and when I tried a side by side install just now it was blank.
[16:42] <ev> ah, I'll fix that in a bit as I'm adding back the gtk hostname as we speak
[16:42] <ev> it's not dependant on what partitioning method you choose though
[16:42] <ScottK> Excellent.
[16:42] <ScottK> OK.
[17:10] <ScottK> ev: It's officially kind of odd then.  I get hostname pre-filled in on a full disk install, but not on a sidebyside.
[17:14] <ev> ...weird
[17:14] <ev> slightly unsettling.  I wonder if debconf is getting trashed.
[17:15] <ev> ScottK: can you please provide ubiquity -d logs for both
[17:15]  * ev meeting
[17:15]  * ScottK will try.
[18:08] <ev> the /var/log/installer/debug log will be the most important
[18:14] <ScottK> OK.  The install is finishing now.
[18:14] <ScottK> (the one where it wasn't pre-populated)
[18:33] <scott__> ev: Bug #634279 (This is ScottK)
[18:33] <ubot2> Launchpad bug 634279 in ubiquity (Ubuntu) "Hostname not pre-populated on ubiquity-kde side by side installs (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/634279
[19:00] <ScottK> ev: I figured out what I was doing different.  If I stick with the automatic username (first name from "Your name"), then no hostname.  As soon as I change the username from the pre-picked one, then the hostname fills right away.
[19:01]  * ScottK adds that to the bug.
[22:31] <CIA-87> ubiquity: evand * r4306 trunk/ (3 files in 3 dirs):
[22:31] <CIA-87> ubiquity: Translate latitude and longitude from the Geonames service to its
[22:31] <CIA-87> ubiquity: respective Olsen timezone database entry.
[22:42] <ev> ScottK: thanks
[23:19] <ev> cjwatson: jockey debconf> in the event that I can't crack this chestnut by final freeze, how crackful would it be to simply dpkg-divert the man-db trigger?