[01:00] <BenC> desrt: not that I know of
[01:00] <BenC> desrt: still around?
[04:26] <desrt> BenC; pong
[04:26] <desrt> BenC; please re-enable it :)
[04:30] <lamont> BenC: it might be from whatever the DC machines wanted... might be good to ping elmo on the HUGETLB* stuff
[04:31] <lamont> at which point he'll say one of (a) HELL YES, (b) HELL NO, or (c) whatever
[05:24] <fabbione> morning
[05:24] <fabbione> desrt: TLB is sort of a problem
[05:25] <desrt> how so?
[05:25] <fabbione> because it adds an extra layer to the mm iirc
[05:25] <fabbione> that slow dows things a bit
[05:25] <fabbione> + it was never turned on
[05:25] <desrt> ah
[05:25] <desrt> it was probably just on in my custom kernel that i was running
[05:26] <desrt> when you guys started shipping 64bit i switched
[05:26] <fabbione> desrt: we can discuss these goodies for dapper
[05:28] <fabbione> http://bugzilla.ubuntu.com/show_bug.cgi?id=16469 <-
[05:29] <fabbione> didn't we fix this one?
[05:30] <desrt> fabbione; i was actually rather hoping for HUGETLB in breezy
[05:31] <desrt> at least in powerpc64-smp :)
[05:31] <fabbione> desrt: too late. it involves an ABI change mostlikey
[05:31] <desrt> arggghh
[05:31] <fabbione> and we are truely deep freeze now
[05:32] <desrt> i don't believe in freezes :)
[05:35] <fabbione> we do :)
[05:35] <fabbione> you lose :P
[05:46] <desrt> does anyone even use ppc64? :)
[05:47] <desrt> how do i find out if ABI would be broken?
[05:48] <fabbione> desrt: yeah quite a lot of people
[05:48] <fabbione> desrt: there is an ABI checker in our source code
[05:48] <fabbione> s/source/build/
[05:53] <desrt> k.  i'll have a look at it at work
[05:53] <desrt> basically, we need need need hugepages
[05:54] <desrt> and running a custom kernel in ubuntu isn't the most comfortable exercise
[05:54] <fabbione> desrt: for dapper for dapper for dapper
[05:54] <desrt> particularly in breezy :)
[05:54] <desrt> k.  our research is occuring now :)
[05:54] <fabbione> desrt: why not?
[05:54] <desrt> fabbione; because you have to recompile it every time the kernel is updated
[05:54] <fabbione> it's easy to build custom kernels
[05:54] <fabbione> nope...
[05:54] <fabbione> that's why you make a custom
[05:54] <desrt> i'm all about getting my stuff into upstream :)
[05:55] <desrt> right... but i mean... if you want the fixes in the new kernel
[05:55] <desrt> then you have to rebuild your config against the new source
[05:56] <fabbione> well of course
[05:56] <fabbione> given that you need the new fixes
[05:56] <fabbione> the only reason why i would compile a custom kernel is if i need special features or remove unused features
[05:57] <desrt> well
[05:57] <desrt> with ubuntu pretty much everything is a module
[05:57] <desrt> which is nice
[05:57] <fabbione> but... 
[05:57] <fabbione> for ex..
[05:57] <fabbione> a security fix comes up on ppp
[05:57] <desrt> i'd disable inotify
[05:57] <desrt> it sucks :)
[05:57] <fabbione> and you don't use and build ppp
[05:57] <fabbione> it's pointless to update
[05:57] <desrt> right.
[05:57] <fabbione> i agree..
[05:57] <desrt> i was joking
[05:57] <desrt> *ahem*
[05:57] <fabbione> so your upgrade path is different
[05:58] <fabbione> idiotify sucks
[05:58] <fabbione> brb
[05:58] <desrt> hahahah
[05:58] <desrt> that's a great name!
[06:08] <fabbione> re
[06:10] <desrt> do you have an idea of how to do better than inotify?
[06:10] <desrt> if you do i'd seriously like to hear it
[06:14] <fabbione> i think the kernel code is ok
[06:14] <fabbione> it's userland that truely sucks
[06:15] <desrt> hmm
[06:15] <desrt> interesting
[06:15] <desrt> in what ways do you feel this to be the case?
[06:16] <fabbione> because i had to debug gamin more than once
[06:16] <desrt> (hah.  i'm being such a psych)
[06:16] <fabbione> so i read both the codes
[06:16] <desrt> ahh
[06:16] <fabbione> and i can tell you that gamin is really bad code
[06:16] <fabbione> or at least used to be
[06:16] <desrt> ok.  ya.. the gamin backend for inotify is a bit scary
[06:16] <desrt> my problems with inotify, incidentally, are based on the kernel :)
[06:17] <desrt> it's the best interface that we could possibly have
[06:17] <desrt> but it's still quite bad
[06:17] <desrt> alas.  bed now.
[06:17] <desrt> cheerio.
[06:18] <fabbione> good night
[10:32] <\sh> gentlemen, i updated #16539
[02:24] <zul> hey
[02:39] <fabbione> yo
[03:37] <BenC> good morning
[03:40] <fabbione> hey Ben
[10:13] <lamont> when is -19 uploading?
[10:19] <zul> when benc gets to it? :)
[10:19] <zul> im heading home folks talk to you later
[10:24] <lamont> mjg59: HP OB900... clues on what I need to do to make sound work in breezy (or hoary, for that matter...)
[10:46] <BenC> lamont: you need anything last minute?
[10:46] <lamont> BenC: just waiting for it
[10:47] <lamont> I mean, today is _almost_ over in London.....
[10:48] <lamont> (not that I'm impatient or anything...)
[10:48] <BenC> hehe, ok
[10:48] <BenC> was just about to do the upload
[10:49] <BenC> waiting to see if I got some response for a couple of bug reports first
[10:54] <BenC> ok, upload in ~10 minutes
[11:10] <fabbione> BenC: rocking
[11:23] <jbailey> mjg59: Should the kernel stop asking us to load vesafb now, then?  Or does it still need to load it in case the user desn't have usplash
[11:23] <jbailey> mmm.
[11:23] <jbailey> Not that it would matter at that point, given that fbcon isn't also loaded.