[06:21] <micahg> slangasek: BTW, aptitude seems to fail miserably at multiarch deps
[06:22] <slangasek> micahg: oh?
[06:22] <micahg> yeah
[06:24] <micahg> oh, maybe not, I'm confused now...well, I have the old e-d-s-common installed
[06:25] <micahg> before I installed all the i386 libs for flashplugin, it used to be able to resolve the dependencies, now it wants to remove a lot of stuff
[06:27] <slangasek> hmm
[06:27] <slangasek> how are you invoking aptitude? interactively, or commandline?
[06:28] <micahg> interactive
[06:44] <slangasek> well, I certainly get a lot of error output whenever I try to change the state of a package
[06:57] <StevenK> slangasek: http://wedontsleep.org/~steven/libao_1.1.0-2.debdiff
[06:59] <slangasek> StevenK: LGTM; mark libao4 Multi-Arch: same, and send it to Debian?
[06:59] <StevenK> slangasek: But what about the -dev changes?
[06:59] <slangasek> what about them?
[07:00] <StevenK> I had to move where the .a files are and so on, is that a problem?
[07:00] <StevenK> I'm also a little concern about the plugins
[07:00] <slangasek> nope; already allowed for by policy, and the standard way to go about it - it does mean that it may break reverse-build-depends, though, so you shouldn't upload to oneiric without an FFe
[07:01] <slangasek> what worries you about the plugins?
[07:01] <StevenK> slangasek: That the path changed
[07:01] <slangasek> well, I was assuming you had tested that the library still finds its plugins ;)
[07:02] <StevenK> slangasek: I haven't yet, I wanted to check that you thought the diff was sensible first
[07:02] <slangasek> yep
[07:03] <StevenK> slangasek: Only libao4 gets Multi-Arch: , nothing else does?
[07:04] <slangasek> well, libao-common should be made Multi-Arch: foreign (and probably also Architecture: all)
[07:04] <slangasek> libao-dbg can probably be marked M-A: same
[07:05] <StevenK> Yes, it's strange that -common isn't.
[07:05] <slangasek> and libao-dev, it depends on whether there are architecture-dependent headers in there
[07:05] <StevenK> It contains a conffile and a manual page only for pitys sake
[07:06] <slangasek> looks like the headers are arch-neutral, so you can mark libao-dev M-A: same as well
[07:06] <slangasek> that will make cross-building nicer :)
[07:07] <slangasek> btw, as long as you're in here you might as well drop the .la files from the package
[07:07] <slangasek> nothing in the archive uses them, and if something did, that something would be broken by them moving directory
[07:07] <StevenK> Even though it installs usr/include/ao/ao.h and so on?
[07:07] <slangasek> yep - as long as the files are the same across all archs, they get refcounted
[07:08] <StevenK> Ahh, but dpkg won't complain like it usually does when two packages contains the same file
[07:11] <slangasek> right, multiarch handles file overlapping differently
[07:11] <slangasek> since /usr/share/doc/*/changelog.Debian.gz exists in every package, for example
[07:12] <StevenK> slangasek: Good to know :-)
[07:12] <StevenK> Now, how to test plugin loading ...
[10:48] <irgendwer4711> hi, I have another problem with Bug #193507
[10:49] <irgendwer4711> I got this error with  BLK_DEV_INITRD enabled
[10:52] <salvatore> hello everybody
[10:52] <salvatore> anybody can help me
[10:52] <salvatore> with lamp configuration?
[10:52] <irgendwer4711> no developer active here now
[10:52] <irgendwer4711> I am waiting too
[10:52] <Hobbsee> support is in #ubuntu
[10:53] <salvatore> ok thank u!
[10:53] <irgendwer4711> for kernel problems too?
[10:53] <salvatore> no
[10:53] <irgendwer4711> okay
[10:53] <irgendwer4711> then can you help me
[10:53] <salvatore> i am tryng to install
[10:53] <salvatore> mysql on dropbox
[10:54] <salvatore> but i got a problem with phpmyadmin
[10:54]  * Hobbsee has no idea on that kernel issue
[10:54] <irgendwer4711> damn...
[10:54] <salvatore> no sorry
[11:20] <irgendwer4711> any kernel professional here?
[11:23] <penguin42> irgendwer4711: There is a #ubuntu-kernel I think specifically for kernel issues - but always ask the question rather than asking if anyone is here!
[11:24] <irgendwer4711> I asked already ;-)
[11:24] <irgendwer4711> I have another problem with Bug #193507 ; this problem occures with BLK_DEV_INITRD enabled too
[11:25] <penguin42> hmm ancient bug
[11:26] <irgendwer4711> I am using LTS ubuntu
[11:26] <penguin42> irgendwer4711: Lucid or Hardy ?
[11:26] <irgendwer4711> hardy
[11:27] <penguin42> irgendwer4711: OK, so you're rebuilding your kernel just without BLK_DEV_INITRD and you get the same error?
[11:27] <irgendwer4711> no with BLK_DEV_INITRD
[11:28] <penguin42> ok, so what are you changing and what exactly is the error?
[11:28] <irgendwer4711> (.init.text+0x82e): undefined reference to `early_populate_rootfs'
[11:29] <penguin42> and which version source are you using exactly?
[11:29] <irgendwer4711> 2.6.24.6 I think
[11:30] <penguin42> can you check ? That fix in that bug went in 2.6.25-10.16
[11:31] <irgendwer4711> wich file
[11:31] <irgendwer4711> dpkg says "2.6.24.29.31"
[11:32] <penguin42> irgendwer4711: sounds promising
[11:32] <irgendwer4711> ok
[11:32] <penguin42> can you pastebin your config?
[11:33] <irgendwer4711> I try a diff to last working conf
[11:33] <penguin42> yeh
[11:36] <irgendwer4711> hm how to print only diffrent with diff ^^
[11:37] <penguin42> diff oldfile newfile
[11:37] <irgendwer4711> I mean with wide format
[11:38] <penguin42> I think that's sdiff - but can't remember how to get it to just show changes
[11:38] <irgendwer4711> done
[11:39] <irgendwer4711> http://pastebin.com/2KYn6DR3
[11:40] <irgendwer4711> I tried to activate some energy savings options und disabled unused drivers
[11:41] <penguin42> irgendwer4711: Without having a dig in the source I'd suggest putting the CONFIG_ACPI_CUSTOM_DSDT_INITRD back the way it was?
[11:41] <irgendwer4711> but I dont have a dsdt patch file
[11:42]  * penguin42 doesn't know much about DSDT
[11:42] <irgendwer4711> there is a file in initrd that patches the acpi tables
[11:43] <irgendwer4711> but not for mine acpi
[11:43] <irgendwer4711> but I can enable this an try
[11:47] <penguin42> irgendwer4711: Yeh looking at the Ubuntu diff the early_populate_rootfs is ifdef'd on CONFIG_ACPI_CUSTOM_DSDT_INITRD but the ifdef in init/main.c is ifdef'd on CONFIG_BLK_DEV_INITRD
[11:48] <penguin42> irgendwer4711: If you've not got the custom dsdt initrd just comment out the call in init/main.c somewhere around line 650
[11:48] <irgendwer4711> I do a new kernel make now
[11:49] <irgendwer4711> a missing dsdt patch file is not a problem, just a short message in dmesg
[11:49] <penguin42> irgendwer4711: But yeh it does look like it's using the wrong ifdef - I don't know the history of the older version where bug 193507 was fixed whether the ifdef changed from there
[11:51] <irgendwer4711> wait to end of kernel make
[11:56] <irgendwer4711> okay its donw
[11:56] <irgendwer4711> no error now
[11:56] <penguin42> good
[11:56] <irgendwer4711> thank you
[11:56] <irgendwer4711> bye
[11:57] <penguin42> bah, I was going to tell him to open a new bug
[13:00] <natschil> Hello. To prevent all packages from modifying /etc/example, do I do "dpkg-divert /etc/example --rename /dev/null" or is there some more elegant way?
[13:21] <natschil> Alternatively, how can I prevent any package from writing to /etc/example
[13:28] <sgnb> natschil: create a dummy package with a random name containing this file?
[13:29] <natschil> what about dpkg-divert --divert /etc/example --rename /tmp/some_random_file_name ?
[14:23] <cjwatson> natschil: 'chattr +i /etc/example' would cause any attempt to write to that file to file
[14:23] <cjwatson> er, to fail
[14:24] <cjwatson> personally I would be hesitant about trying to use dpkg-divert for that because at least historically it has had odd interactions with dpkg conffiles and it doesn't do anything useful for maintainer-script-managed configuration files
[14:31] <cjwatson> ScottK: I may be able to help with k3b in the not too distant future, if somebody else can test the result; I'll queue it up
[15:07] <natschil> cjwatson: let's say im installing something that wants to change /etc/example. Will that still fail?
[15:10] <cjwatson> natschil: I'm not sure how to respond except by repeating what I said above
[15:11] <natschil> cjwatson: ok thanks. I think I will use dpkg-divert instead and hope for the best :/
[15:11] <cjwatson> well, I can repeat the fact that I do not recommend that
[15:11] <cjwatson> chattr +i will cause *any* attempt to write to that file to fail
[15:12] <cjwatson> dpkg-divert will arrange for a very limited subset of writes to be redirected elsewhere, which doesn't cover all the things packages do with files in /etc, and which I suspect will have crazy problems even if it does sometimes work
[15:12] <cjwatson> the only advantage of dpkg-divert here is that it would arrange for the package to install successfully anyway
[15:13] <cjwatson> (maybe)
[15:17] <ScottK> cjwatson: I should be able to test it.
[15:17] <ScottK> Help would be much appreciated.
[15:19] <cjwatson> most of them aren't too hard once you get the patterns.  xvidcap is stubbornly segfaulting right now ...
[15:27] <directhex> Laney found the mono commit which breaks on some (not all) i386 buildds
[17:04] <SpamapS> lamont: [ -x /usr/bin/dh_apport ] maybe ?
[17:05] <SpamapS> lamont: for the build-dep we can do an | to make it optional, right?
[17:53] <broder> cjwatson: for what it's worth, i used diversions heavily when i was at mit, and i think we whined enough that we were able to get the bugs with diversions+conffiles ironed out
[17:57] <cjwatson> broder: ah, that's something at least
[18:45] <cjwatson> ScottK: give http://paste.ubuntu.com/671830/ a try?
[18:45] <cjwatson> (please try not to laugh at my rusty C++)
[20:21] <cjwatson> hmm, I wonder if my xvidcap crash is actually a libav bug
[20:24] <cjwatson> looks like its MAP_ANON detection fell over in 0.7 or so
[20:25]  * cjwatson tries to construct a test that isn't "build xvidcap with this dodgy patch"