[06:05] <fabbione> morning
[06:41] <desrt> word.
[07:06] <fabbione> hey desrt 
[07:06] <fabbione> happy with your new headers?
[07:06] <fabbione> desrt: btw i did look at that driver you requested
[07:07] <fabbione> i might included it in the next kernel
[07:07] <fabbione> (was too close to release for the previous)
[07:07] <fabbione> but you need to get somebody to pkg the utilities
[07:07] <desrt> fabbione; oh.  i didn't know you received those bugs :)
[07:07] <desrt> and yes.  the headers are working flawlessly.  thank you very much :)
[07:08] <fabbione> desrt: i get them automatically :)
[07:08] <desrt> i love bugzilla :)
[07:08] <fabbione> i don't
[07:09] <fabbione> it tells me that i suck at maintaining the kernel :P
[07:09] <desrt> oh.  why not?
[07:09] <desrt> no.  it tells you that maintaining a kernel is a hard job
[07:09] <fabbione> nah just kidding ;)
[07:09] <desrt> btw... the first time i ever talked to you was at jdub's recommendation... almost 6 months ago
[07:10] <desrt> i don't know if you remember why that was
[07:11] <fabbione> hmmm i can't really remember...
[07:11] <desrt> gamin.
[07:11] <fabbione> probably.. gamin sucks :P
[07:11] <desrt> and the reason i brought it up: still some cases where it stops working :/
[07:11] <fabbione> ahhh yeah
[07:11] <fabbione> now i reacall
[07:11] <desrt> my desktop stopped updating again yesterday
[07:12] <fabbione> there is probably nothing to update :)
[07:12] <fabbione> it was sunday
[07:12] <desrt> no.. i mean, i was saving files with firefox
[07:12] <desrt> and i had to ctrl+r the desktop to get them to appear
[07:13] <fabbione> what kernel?
[07:13] <desrt> 2.6.12-whatever
[07:13] <desrt> 3?
[07:13] <fabbione> because the latest gamin needs the latest kernel
[07:13] <desrt> -4
[07:13] <desrt> oh hmmm
[07:13] <fabbione> inotify did change from something to ioctl...
[07:13] <fabbione> or whatever...
[07:13] <desrt> maybe i installed a new gamin but had an old kernel
[07:13] <fabbione> and they need to be synced
[07:14] <desrt> and have only recently rebooted
[07:14] <fabbione> mostlikely
[07:14] <desrt> ok
[07:14] <desrt> hm
[07:14] <desrt> actually, it's broken right now
[07:14] <desrt> Linux moonpix 2.6.12-4-686-smp #1 SMP Fri Jul 22 12:56:02 UTC 2005 i686 GNU/Linux
[07:15] <desrt> i just touched a file on my desktop... nautilus doesn't show it until manual refresh
[07:15] <fabbione> can you check if dmesg has comething?
[07:15] <fabbione> something...
[07:15] <desrt> what am i looking for?
[07:15] <fabbione> dunno
[07:15] <fabbione> anything ...
[07:15] <desrt> [4294670.897000]  inotify syscall
[07:15] <desrt> this is all, really
[07:15] <fabbione> ok i will need to check gamin source and our inotify patch
[07:15] <fabbione> perhaps the syscalls aren't synced properly
[07:16] <fabbione> i am pretty sure i have an idea of what's broken...
[07:18] <desrt> btw: restart nautilus -> works fine again
[07:24] <fabbione> hmm
[07:24] <fabbione> i wonder if it is a nautilus issue
[07:24] <fabbione> if the syscalls were not synced, it wouldn't work at all
[07:24] <fabbione> desrt: there is a tutorial on the gnome/gain website
[07:24] <fabbione> that explains how to debug gamin
[07:25] <infinity> (Like when running a breezy kernel on hoary, for instance)
[07:25] <infinity> God, my lack of updating desktop is irritating.
[07:25] <desrt> infinity; you too?
[07:25] <fabbione> iirc sending a kill -USR2 to the gamin process will enable and dump debugging info in /tmp
[07:25] <desrt> oh.  that's useful.
[07:25] <fabbione> infinity: yes.. you cannot run breezy kernels on hoary
[07:25] <infinity> desrt : Yes, but I know why mine doesn't work. :)
[07:25] <desrt> infinity; :)
[07:25] <fabbione> desrt: but please check the website for the correct signals
[07:26] <desrt> mine looks just like the same old bug that was a problem around hoary-prerelease time
[07:26] <fabbione> and be sure to issue the command once again to stop it :)
[07:26] <infinity> fabbione : Sure I can, I just have a few broken features.  On the other hand, my network works and my machine doesn't crash.  The machine's going breezy once I make X less suck.
[07:26] <desrt> fabbione; produces fabulous amounts of output? :)
[07:26] <fabbione> desrt: quite :)
[07:27] <fabbione> infinity: yeah well.. broken features for you are show stoppers for users
[07:27] <fabbione> infinity: you know that ;)
[07:27] <infinity> ;)
[07:27] <fabbione> fantastic..
[07:27] <desrt> fyi, http://www.gnome.org/~veillard/gamin/debug.html
[07:27] <fabbione> fresh breezy install.. X manages to hang my keyboard because fixed fonts are missing....
[07:28] <fabbione> LOVELY
[07:28] <desrt> fixed fonts aren't missing
[07:28] <desrt> mkfontdir is missing
[07:28] <desrt> X can't generate the fontdir cache... therefore can't find any fonts
[07:28] <infinity> It's not "missing", it's just gone for a bit of a jaunt around the block.
[07:28] <desrt> well... i'll be very happy when it gets back from its jaunt :)
[07:28] <infinity> People get so upset about the smallest binaries...
[07:29] <desrt> mkfontdir, xset, xkb*, xmodmap, ...
[07:29] <infinity> desrt : It's all coming back in good time.
[07:29] <desrt> i copied all of the stuff i needed (including fontdir caches) in from my laptop :)
[07:29] <infinity> SOme of the xbk stuff already came back in xkbutils.
[07:30] <fabbione> now.. where is daniels when i need to test my new dildo set?
[07:30] <desrt> fabbione; those new USB-powered dildos?
[07:30] <infinity> I'd be nice to him, if I were you.
[07:31] <infinity> We still need to assing the "make sure X upgrades smoothly from hoary" task to someone.
[07:31] <fabbione> desrt: yes.. with IRDA remote control support ;)
[07:31] <fabbione> infinity: remember that i did Xfree86 -> Xorg
[07:31] <fabbione> infinity: i know what we are talking about
[07:31] <fabbione> and Xorg -> Xorg is simpler
[07:31] <infinity> (Daniel's got the "make it modular and make it work" task, and I've got the "make sure the bloody thing builds in a deterministic and bootstrappable fashion" task...)
[07:32] <fabbione> infinity: btw.. xkbdutils didn't enter the archive yet?
[07:32] <infinity> Yeah, upgrading is the simplest of the bunch, but it's still going to nee a mess of testing, and soon.
[07:32] <infinity> Oh, is it still NEW?  Or did I get the name wrong?
[07:32] <fabbione> no idea.. you tell me :)
[07:33] <infinity> Or... It's FTBFS, and I've been too busy to notice.
[07:33] <fabbione> i got the new libxaw.. that's good
[07:34] <fabbione> what pkg has mkfontdir now?
[07:36] <fabbione> daniels: ?
[07:36] <fabbione> modprobe -k dildo
[07:36] <fabbione> USB: Detected sodomotron on port 1:1.1
[07:37] <fabbione> USB: enabling /dev/sodomotron device for immediate access
[07:42] <desrt> one day... one day i will upload.
[07:46] <fabbione> hmmm i can't remeber how i did fix the KBD problem...
[07:48] <fabbione> AH RIGHT
[07:48] <fabbione> xkbdutils :)
[07:49] <fabbione> dpkg-deb: building package `xkbutils' in `./xkbutils_7.0-2_i386.deb'.
[07:49] <daniels> fabbione: mkfontdir doesn't really exist right now
[07:49] <daniels> i'll fix that tonight
[07:49] <daniels> not at home at the moment
[07:50] <daniels> and yeah, I'll be making sure upgrades work fine, but right now they're not even close
[07:50] <fabbione> daniels: eheh dude... you know how much i love to tease you
[07:50] <daniels> xkbutils got source ACCEPTED, but still needs its binaries NEWed
[07:50] <fabbione> but i need to get a working breezy on one machine
[07:50] <fabbione> and i am checking all the gotchas
[07:50] <daniels> yeah, fair enough
[07:50] <daniels> it's still a work in progress :) but gotta run now
[07:51] <daniels> later
[07:51] <fabbione> (EE) Couldn't load XKB keymap, falling back to pre-XKB keymap
[07:51] <fabbione> (EE) Couldn't load XKB keymap, falling back to pre-XKB keymap
[07:51] <fabbione> daniels: help me only one sec
[07:51] <fabbione> i fixed it yesterday.. but i can't remember how
[07:51] <fabbione> do you recall it?
[07:51] <desrt> fabbione; you can just copy the fonts.cache-1 files from another box
[07:51] <fabbione> desrt: i already fixed the fonts problem
[07:51] <infinity> daniels : Binaries kinda had to build first.  *cough*
[07:51] <desrt> :)
[07:51] <infinity> daniels : I just uploaded a new libxaw that xkbutils is now build-dep on.
[07:53] <desrt> you guys never dpkg-buildpackage, right?  you just upload the source package to the build farm....
[07:53] <infinity> dpkg-buildpackage -S :)
[07:53] <fabbione> i do
[07:53] <infinity> But yes, I do binary builds before I upload.
[07:53] <infinity> I just don't upload the binaries.
[07:54] <infinity> I do "dpkg-buildpackage -uc -us -S ; dpkg-buildpackage -uc -us -b"... If the latter succeeds, I sign and upload the results of the former.
[07:58] <desrt> is the ftp archive signing machine-generated?
[07:58] <fabbione> desrt: ?
[07:59] <desrt> the binary .deb's are signed by "ftpmaster@ubuntu.com"
[07:59] <fabbione> the .deb are not signed
[07:59] <infinity> The uploads are.
[08:00] <desrt> but aren't the actual .deb's md5'd and then the list is signed?
[08:00] <desrt> like, the actual binary debs
[08:00] <fabbione> desrt: nope...
[08:00] <fabbione> there is no need to
[08:00] <infinity> Yes.  The binary .changes are signed.
[08:00] <fabbione> desrt: only the sources are signed
[08:00] <infinity> fabbione : The binaries are too, dude.
[08:01] <desrt> then what's the point of apt-key and why does apt-get sometimes warn you about unauthenticated packages?
[08:01] <fabbione> infinity: yeah but in the .changes
[08:01] <infinity> fabbione : Yes, that's what he was talking about.
[08:01] <fabbione> there is no signatures of the md5 inside the deb
[08:01] <desrt> right.. i think it's in the Releases file
[08:01] <fabbione> desrt: it's a chain
[08:01] <fabbione> basically you verify that the Release file has a good signature
[08:02] <fabbione> that means that the Packages.gz can be trusted via md5 check on top of the Release file
[08:02] <infinity> desrt : That is the Releases file, which verifies the integrity of the Packages file, which verfies the integrity of the debs, which were verified by a signed .changes when they were uploaded.
[08:02] <infinity> And yes, Releases.gpg is autosigned.
[08:02] <desrt> infinity; gotcha.  thanks :)
[08:02] <infinity> Release.gpg, even.
[08:02] <fabbione> if the Contents of Packages is verified, it means that the md5 contained in it for the binary debs can be trusted
[08:03] <desrt> makes sense
[08:03] <desrt> i imagine that we can't be too far away, though, from someone being able to feasibly calculate MD5^-1(x)
[08:04] <infinity> MD5 can be broken, but can you break it AND deliver data that does something useful (or malicious)?
[08:04] <infinity> That's a much bigger challenge.
[08:04] <desrt> well... you can write a bad package
[08:04] <infinity> Collisions can happen.  Interesting collisions, I'm not so sure about.
[08:05] <desrt> then add some useless padding bytes into the end of it to make the md5 match
[08:05] <infinity> Harder than you'd think.
[08:05] <infinity> Especially since the file size also has to match.
[08:05] <desrt> difficult... but md5 is getting weaker by the day
[08:06] <fabbione> desrt: that's why the Release contains also teh SHA1
[08:06] <infinity> I suspect a picture of my cat probably has the same MD5 as libc6.deb, but I'm not sure how I can use this to take over the world.
[08:06] <desrt> that i didn't know.
[08:06] <fabbione> and mathing both md5 adn sha1 at the same time is impossible
[08:06] <desrt> that makes it very much more difficult
[08:06] <infinity> fabbione : It's the sha1 of the Packages file, we don't carry the sha1 of the debs.
[08:06] <desrt> plus.. honestly
[08:06] <infinity> fabbione : So you could still poision individual debs.  But really, I think the argument is moot.
[08:07] <desrt> half the time it says "packages can't be authenticated" i just hit "yes, continue" anyway
[08:07] <desrt> so who cares :)
[08:07] <infinity> If I have access to poision you in such a sophisticated way, I can probably do something muhc simpler to you.
[08:08] <fabbione> desrt: echo 'APT::Get::AllowUnauthenticated "1";' > /etc/apt/apt.conf
[08:08] <fabbione> and you will forget even to press enter ;)
[08:09] <desrt> nah... i'd at least like it to tell me :)
[08:10] <infinity> It still tells you.
[08:10] <infinity> It just doesn't pause for confirmation.
[08:10] <infinity> "Foo bar baz are unauthenticated, but installing anyway, cause the override is on <zooom!>"
[04:02] <jbailey> lamont: *poke*
[04:03] <jbailey> lamont: I still need you to let me know if I can book you soonish for another biarch setup, or if infinity is capable of doing it.
[04:03] <jbailey> lamont: doko is threatening to beat me...
[04:03] <jbailey> And as interesting as that sounds, Angie is the jealous type.
[07:14] <fabbione> doko: gcc-snapshot is running the testsuite here... on sparc
[08:06] <doko> fabbione: it's not biarch, that's the reason it works
[08:12] <fabbione> ah
[08:12] <fabbione> ok