| cjwatson | stgraber: so FYI, it looks like I may well be unavailable for a while; ubiquity currently fails to pass its test suite because something in (I think) GTK has caused the UI size to expand such that it fails the netbook screen size tests | 00:38 |
|---|---|---|
| cjwatson | I started bisecting through GTK but was called away | 00:39 |
| cjwatson | I fear that may well be the best way to find the problem | 00:39 |
| cjwatson | unless the test is actually too strict in which case it's OK to weaken it; it should be possible to determine at least from bzr history what actual screen size it was aiming for and verify whether it would fit on that | 00:40 |
| cjwatson | I'll be around on and off, maybe, but am at least off work tomorrow | 00:40 |
| GrueMaster | stgraber: Netboot install is fixed, thanks (just tested on one of my pandas). Time for mass-reimaging. | 00:45 |
| GrueMaster | ubuntu | 00:45 |
| GrueMaster | bah. | 00:45 |
| stgraber | cjwatson: ok, thanks for the update. I'll also take care of the release meeting as I'm attending it anyway (will reply to the ML too) | 00:46 |
| stgraber | GrueMaster: cool, good to hear | 00:46 |
| cjwatson | star, thanks | 00:46 |
| cjwatson | (feel free to strongarm somebody else into dealing with the ubiquity failure, e.g. desktop or whatever, just braindumping quickly here) | 00:46 |
| cjwatson | it's quickest to run 'tests/build' and then 'tests/run' or 'xvfb-run tests/run' (possibly with a test_gtkui parameter) depending on whether you want to see the output or not | 00:48 |
| cjwatson | er I mean see bits of UI that are shown | 00:48 |
| cjwatson | you can use something like http://paste.ubuntu.com/833381/ to make it show each of the pages in succession and wait long enough for you to see them, which is handy for debugging this kind of thing | 00:49 |
| stgraber | cjwatson: yeah, from what bdmurray said earlier it looks like we might have another resolvconf issue in ubiquity I'd need to investigate anyway, so I'll set sometime aside to do some ubiquity stuff I guess (Feature Freeze is getting dangerously close though) | 00:49 |
| cjwatson | I know :-/ | 00:49 |
| stgraber | oh yeah, the delayed test is a good idea to see exactly what's wrong. I guess I'll run that with old and new gtk to see if something obvious appears (sounds too easy ...) | 00:50 |
| cjwatson | most of what I haven't already done for FF is toast at this point | 00:50 |
| cjwatson | I did that and there was nothing that jumped out, but I couldn't get my local build with old gtk to use the right theming | 00:50 |
| cjwatson | so maybe worth either trying with a packaged build or figuring out what's needed to get the theming in place | 00:51 |
| stgraber | k, I'll dig a bit. IIRC the requirement is 1024x600 (standard netbook resolution) and I have hardware with that resolution + wireless and webcam, so should be easy to verify on hardware and update the test if needed | 00:51 |
| cjwatson | if we have to do an upload for some other reason then I guess it would be OK to temporarily disable the test, but we'd probably want some kind of release-critical bug in that case | 00:52 |
| cjwatson | I'm not convinced the test is very robust right now, so it could of course just be a test bug which the new GTK triggers | 00:53 |
| cjwatson | there was a CSS padding/border handling change in the new GTK which it's possible could have tickled something; maybe also other changes | 00:53 |
| cjwatson | anyway - bed, hopefully tomorrow will be less tiring | 00:53 |
| cjwatson | night all | 00:54 |
| antarus | PRECISE!!!! | 00:54 |
| * antarus stabs | 00:54 | |
| === CIA-2 is now known as 45PAAROFP | ||
| === 45PAAROFP is now known as CIA-90 | ||
| === amichair is now known as amichair_ | ||
| === _Josh_ is now known as jacalvo | ||
| cr3 | does in-target in the preseed late_command rely on /etc/resolv.conf or /target/etc/resolv.conf to resolve domain names for commands needing to access the internet? | 14:29 |
| === shadeslayer_ is now known as shadeslayer | ||
| cjwatson | cr3: in-target doesn't do any name resolution itself. That's typically the responsibility of libc, which will do whatever's normal in the filesystem namespace it finds itself in - /target, in the case you ask about. | 15:55 |
| cr3 | cjwatson: ok, for the first part about in-target doing name resolution, that's probably bad wording on my part :) as for the second part, aren't there a bunch of files missing in /target like proc/, dev/ and sys/ which might be needed somehow? | 15:58 |
| cr3 | s/files/directories/ | 15:59 |
| cjwatson | No, they're bind-mounted from the installer environment for the duration of in-target's execution. | 16:05 |
| cjwatson | Those would in any case probably not be needed for name resolution, but are needed for many other things. | 16:06 |
| cr3 | cjwatson: thanks for the detailed explanation and sorry for the lag of this response :) | 17:12 |
| CIA-90 | ubiquity: stgraber * r5173 ubiquity/ (debian/changelog ubiquity/frontend/gtk_ui.py): Merge branch from Brian Murray fixing LP: #901381 (apport not starting when ubiquity crashes due to non-working inotify on overlayfs) | 20:05 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!