[05:10] <mjg59> doko: binutils_2.16.1-2ubuntu7_i386.deb  works, binutils_2.16.1cvs20051109-1ubuntu1_i386.deb doesn't
=== chmj [n=chmj@196.44.1.98]  has joined #ubuntu-toolchain
=== doko_ [n=doko@dslb-088-073-096-073.pools.arcor-ip.net]  has joined #ubuntu-toolchain
[08:42] <fabbione> hey doko_ 
[09:49] <doko> hi fabbione
[11:01] <fabbione> doko: hey...
[11:02] <fabbione> doko: do you think it is possible to switch default gcc build option for sparc?
[11:03] <doko> fabbione: sure, should be possible
[11:03] <fabbione> doko: --with-cpu=v7 -> --with-cpu=v8
[11:03] <fabbione> when do you think that could be done?
[11:05] <fabbione> or sort of.. when do you plan to upload the next version of gcc?
[11:05] <doko> as long as I can add hppa specific fixes, so that lamont doesn't glare ...
[11:06] <doko> today
[11:06] <fabbione> ok great!
[11:06] <fabbione> that's fine
=== jbailey [n=jbailey@modemcable139.249-203-24.mc.videotron.ca]  has joined #ubuntu-toolchain
[02:30] <lamont> doko: LP is building dapper/hppa now, so I don't care so much... 'twould be nice to let it catch up sometime soon, though... :-)
[02:31] <jbailey> lamont: Meh.  You have 3 buildds, don't you? =)
[02:31] <doko> yes, one for gcc-4.0, gcj-4.0, gcj-4.1 ;-P
[02:32] <lamont> jbailey: not sure how many of them are online yet
[02:34] <lamont> jbailey: I see 2 so far (primero, kohnen)
[02:34] <lamont> and yes, primero is building gcj-4.0
[02:34] <lamont> er, 4.1
[02:35] <jbailey> Well, it'll be nice to maybe see the 'pending' list in Soyuz actually get down to zero. =)
[02:44] <infinity> jbailey: That won't happen until backports is sorted, unfortunately.
[02:44] <infinity> (Which I should get on, I guess)
[02:45] <infinity> jbailey: Also, didn't you say you had some breezy-updates uploads pending?
[02:45] <jbailey> infinity: I did.  Colin asked me to resubmit them, and I haven't.
[02:45] <infinity> jbailey: I could be on crack, but I'd swear someone else set up breezy-updates chroots wen I wasn't looking, cause I see some successful builds in that pocket.
[02:45] <jbailey> I also have glibc timezone updates for our three live ones.
[02:45] <jbailey> Yeah, breezy-updates is working now.
[02:46] <infinity> glibc timezone updates kinda need to happen NOW.
[02:46] <infinity> We're dangerously close to that being a problem here in Melbourne.
[02:46] <jbailey> Oh right.  Aussie's and their screwed up timezone bits.
[02:46] <infinity> Like, the bug hits in a week or something.
[02:46] <jbailey> Okay.  I'll try to send through warty, hoary and breezy sometime today.
[02:47] <jbailey> BEWARE THE IDES OF MARCH
[02:47] <infinity> (Don't feel too bad, I only saw the update come in on Windows Update a week or so ago...)
[02:49] <infinity> jbailey: Ping Kamion and I both when you've uploaded them, so he can approve them immediately, and I can babysit and make sure all three pockets actually work (I suspect only breezy-updates does, and even that I'm suspicious of)
[02:49] <jbailey> I've used packages from breezy-updates. =)
[02:49] <infinity> Yeah, I'm more suspicious of the chroots than of the publishing. :)
[02:49] <infinity> It does seem to more or less work right.
[02:50] <jbailey> https://launchpad.net/+builds/+build/174707
[02:51] <mjg59> doko: Are there any binutils packages earlier than the November CVS?
[02:51] <jbailey> mjg59: Probably breezy-final
[02:52] <jbailey> There was only what, 3 weeks in there?
[02:52] <infinity> Which was non-CVS.
[02:52] <infinity> jbailey: Lots more than 3 weeks upstream.. :/
[02:52] <jbailey> Yeah, point.
[02:56] <doko> mjg59: no have to build these myself
[02:57] <mjg59> doko: Ok
[02:57] <infinity> mjg59: http://snapshot.debian.net/archive/2005/09/06/debian/pool/main/b/binutils/binutils-dev_2.16.1cvs20050902-1_i386.deb
[02:57] <infinity> mjg59: Looks like snapshot's back up. :)
[02:58] <infinity> mjg59: That may help give you a small window to search in.
[02:58] <mjg59> Hurrah
[02:58] <infinity> Err, s/-dev//
[02:58] <infinity> http://snapshot.debian.net/archive/2005/09/06/debian/pool/main/b/binutils/binutils_2.16.1cvs20050902-1_i386.deb
[02:58] <infinity> Better.
[02:59] <jbailey> mjg59: Binary searching for the EFI build breakage?
[02:59] <infinity> mjg59: You've rebuilt the binutils 2.16.1 package with our current toolchain to make sure that it's not some scary "GCC started miscompiling binutils recently" bug, right?
[02:59] <mjg59> infinity: Yes
[02:59] <infinity> (Though, given your initial findings in the binary diff, I suspect it's binutils alone at fault)
[02:59] <mjg59> jbailey: Yup
[03:06] <mjg59> 20050902 works, 20051109 doesn't
[03:29] <mjg59> doko: 20051001 works
[03:30] <mjg59> Given a CVS checkout from a certain date, can I just update it to a later date?
[03:30] <mjg59> (Or, indeed, revert it to a previous date?)
[03:30] <infinity> Assuming it has the CVS dirs in it, sure.
[03:31] <mjg59> How? update -D?
[03:31] <infinity> -D or -d, I don't recall.
[03:31] <infinity> -D
[03:32] <infinity> cvs up -D 2005-10-13
[03:32] <infinity> Or so.
[03:32] <mjg59> Ok
[03:57] <mjg59> Breakage was between the 22nd and 27th of October
[03:57] <infinity> That's getting much more narrow.
[03:57] <mjg59> Yes
[03:58] <mjg59> I'm doing the 25th now, then I'll just diff the trees
[04:01] <mjg59> 25th is broken
[04:09] <mjg59> 24th works
[04:11] <jbailey> \o/
[04:11] <jbailey> Now to figure out if it's opcodes, bfs, binutils, gas...
[04:15] <mjg59> Hm. 2773 lines of diff.
[04:16] <mjg59> Pile of changes in bfd
[04:16] <mjg59> And also in gas
[04:16] <mjg59> And in ld
[04:16] <mjg59> Ho-hum
[04:19] <infinity> 2773 lines isn't so bad, in binutils land.
[04:24] <mjg59> Ok, it's either the ld or bfd changes
[04:25] <mjg59> Which takes us down to 1312 lines
[04:26] <mjg59> I blame bfd.
[04:26] <mjg59> Since the ld changes are just down to changes in bfd
[04:27] <mjg59> doko: It's either H.J. Lu's changes on 2005-10-23 or one of the changes on 2005-10-24
[04:28] <mjg59> doko: Does that narrow things down enough for you?
[04:37] <doko> mjg59: yes, thanks! still having OOo fun at the moment ...
[04:56] <mjg59> doko: No problem
[11:17] <mjg59> doko: Ok, it's H.J. Lu's fix for PR/1487
[11:18] <mjg59> The one from 2005-10-23
[11:20] <doko> mjg59: looks like you couldn't rest ;-)
[11:20] <mjg59> Heh
[11:28] <jbailey> CANT SLEEP.   GENETICALLY ALTERED FIREFLYS (or whatever it is you grow) WILL EAT ME
[11:29] <mjg59> Fruitflies
[11:29] <jbailey> mjg59: One of these days I'd like to take you for a drink and get you to tell me how one does gene alterations.
[11:29] <mjg59> Haha
[11:39] <mjg59> doko: The diff is at http://www.codon.org.uk/~mjg59/tmp/binutils.diff
[11:39] <mjg59> If that's reversed, it's happy
[11:41] <doko> ok, submitting an upstream bug report