[12:38] <lamont> doko_: btw, sid's buildd is totally clusterfied atm, too.
[12:38] <lamont> same toolchain issue.
[12:39] <jbailey> doko_: lib64g2c0_3.4.4-6ubuntu4_i386.deb
[12:39] <jbailey> doko_: (Insert utf-8 thumbs-up icon here)
[12:40] <lamont> jbailey: any hints on the _GLOBAL_OFFSET_TABLE_ borkage issue?>
[12:40] <lamont> (hppa, although istr fabbione bitching about something similar on sparc...)
[12:40] <jbailey> lamont: Got something I can read about it somewhere?
[12:40] <jbailey> HAving a broken GOT would really suck. =)
[12:40] <lamont> uh, yeah.
[12:41] <lamont> for i in build/chroot-unstable/usr/lib/*.so.*; do x=$([ -f $i ]  && objdump -T $i | grep _GLOBAL); [ -n "$x" ]  && echo $i $x; done | wc
[12:41] <lamont>      25     200    2600
[12:41] <lamont> that's a virgin-ish buildd chroot
[12:41] <lamont> just load current sid on an hppa box - then look at libz
[12:41] <lamont> or current breezy.
[12:41] <lamont> both are rather b0rked
[12:41] <lamont> build/chroot-unstable/usr/lib/libz.so.1.2.3 000234a0 g DO *ABS* 00000000 Base _GLOBAL_OFFSET_TABLE_
[12:42] <lamont> jbailey: oh, and the best part is that it segv's fakeroot on 90%+ of builds.
[12:44] <jbailey> Any idea when this started?
[12:46] <doko_> lamont: breezy as well?
[12:47] <lamont> doko_: same issue
[12:47] <lamont> the fakeroot b0rkage is timed with gcc-2.3.5 going into the chroot
[12:48] <lamont> zlib is a wonderful test case, since it takes < 2 minutes to build on an a500
[12:48] <doko_> lamont: and we didn't notify this earlier?
[12:50] <lamont> doko_: it takes a while to notice the fatality of it all, since one object with it exported isn't fatal
[12:51] <doko_> :-(
[12:52] <lamont> been trying to get attention on it for a few days now, but then it was just breezy that was broken, and noone cares about breezy/hppa.. :-(
[12:52] <lamont> but now that sid
[12:52] <lamont> 's broke....
[12:54] <jbailey> lamont: I need to hunt bdale down so that I can get my access on j5k restored.
[12:56] <lamont> bdale was mumbling about a hectic day today, tomorrow travelling, and thursday picking up the reigns on debian's  buildd sigining for me (ia64/hppa)
[12:59] <jbailey> Hmm.  thu-mon I'm travelling so I guess I won't sync up with him until next week.
[01:00] <jbailey> Thu I'll be online subject to the grace of wifi connections at the chicago airport.
[01:12] <lamont> ---Mutt: =buildd/universe [Msgs:1692/6648 New:6644 Post:3 341M] ---(subject/date)------------------------(1%)---
[01:12] <lamont> there's something wrong with that picture
[01:13] <jbailey> That is 341MB for the mbox, not for the message, right?
[01:13] <lamont> right
[01:13] <lamont> that's the index page
[01:46] <jbailey> amd64 test glibc build going.
[01:46] <jbailey> infinity: If this works, I have some fun for you. =)
[05:42] <infinity> jbailey : pong.
[05:43] <jbailey> infinity: Hey
[05:43] <jbailey> I thougt I was ready for the upgrade dance, but then I discovered that my glibc build had frozen (the testsuite doens't like my chroot)
[05:43] <jbailey> And then I reran it without -nc.
[05:43] <infinity> Sweet.
[05:43] <jbailey> So soonish, I'll have finished the testing.
[05:43] <jbailey> Unless you want to start now with the pieces that we know about either way. =)
[07:31] <infinity> I'll wait for you to be sure it's all good to go, then you can hand me the whole mess.
[07:33] <jbailey> infinity: 'k
[07:33] <jbailey> infinity: Found a slight bug in the final amd64 install (native)
[08:53] <doko> infinity: the OOo2 build did fail, because it couldn't find the archives?
[08:57] <infinity> doko : On i386?
[08:57] <doko> yes
[08:57] <infinity> That's a transient error that always results in an auto-give-back.
[08:57] <doko> so it's currently building?
[08:57] <infinity> It's building right now on rothera, and seems t obe doing fine.
[08:58] <doko> wow, finally :-)
[08:58] <infinity> Yeah, I'm keeping an eye on it.  I'd kinda like to see it build. :)
[08:59] <jbailey> feh, stat spits out numbers in hex.
[08:59] <jbailey> I need them in decimal.
[09:00] <jbailey> Heya doko.  I found one tweak that I needed to do in the amd64 native build.  Testing the i386 biarch build and then I hand it to Adam. =)
[09:00] <jbailey> In about 20 minutes thanks to ccache. =)
[09:01] <jbailey> I think, anyway.
[09:01] <jbailey> Time is an illusion.  Lunchtime doubly so.
[09:01] <doko> glibc biarch build?
[09:01] <jbailey> doko: right.
[09:03] <infinity> ccache and I are going to get married and have babies.
[09:03] <jbailey> infinity: Is that so you don't have to remember how?
[09:04] <infinity> See, if you could just take the high road for once and not go for the nerd interpretation... :)
[09:06] <jbailey> Mm, I suppose
[09:07] <jbailey> I wonder if I could detect a succesful ccache hit somehow and force make -j 500 after that?
[09:29] <infinity> watch ccache -s, see if the hits are going up, if so, clean and restart with -j 500?.. I dunno.. Ugly to automate it, easy to do by hand.
[09:39] <jbailey> Woot, looks good.
[09:39] <jbailey> sending her up.
[09:46] <infinity> Where to?
[09:46] <infinity> And do you have nice instructions on what order this stuff needs to be bootstrapped in?
[09:49] <jbailey> infinity: The idea is that you fill the chroot with the pieces from all three, and then you should be able to just build it all.
[09:50] <jbailey> It's going into chinstrap:~jbailey/i386-amd64
[09:50] <infinity> doko : I assmue the OOo2 build failure on amd64 is expected (or, at least, understood)?
[09:51] <infinity> jbailey : No particular order I should need to build them in?
[09:51] <infinity> jbailey : Since they're all ready to go, I guess that makes sense.
[09:52] <jbailey> Right. =)
[09:52] <jbailey> All the build-deps will be satisfied already, you're just masturb^Wrebuilding them.
[09:54] <infinity> And the source packages are signed?
[09:54] <infinity> So once I've done the initial bootstrap, I can upload the sources and have them rebuild again?
[09:55] <infinity> (Not like I couldn't sign them myself, but I want all this stuff signed by you anyway, so I have a scapegoat if you root us...)
[09:56] <jbailey> Dude, if I root you, I'd be sure you never kne.
[09:56] <jbailey> +w
[09:56] <jbailey> It's not like a kernel root, where you might replace it with one you built yourself.
[09:56] <jbailey> Only insane people build their own glibc.
[09:57] <infinity> I always build my own custom glibc, yo, cause that's the l33t thing to do.
[09:57] <infinity> I turn on the sooper sekrit options.
[09:57] <jbailey> Dude.
[09:58] <jbailey> lamont told me that -pipe is a known optimisation that Ubuntu is using.
[09:58] <jbailey> Make sure to use *that* on your build.
[09:58] <infinity> I use -fno-slow-code, it really helps.
[10:00] <infinity> Does this need to be bootstrapped on all 4 arches, or is it just one or two that need by-hand love?
[10:00] <doko> infinity: yep, the build is missing a bit ... (exit 63)
[10:00] <jbailey> infinity: i386 only.
[10:00] <jbailey> infinity: amd64 migiht be tomorrow. =)
[10:00] <infinity> Alright, cool.
[10:01] <infinity> rsync?
[10:01] <infinity> paranoid about scp cutting out?
[10:02] <jbailey> rsync is apparently faster.
[10:03] <jbailey> I've read a number of rants about scp having buffering issues that somehow rsync over ssh doesn't have.
[10:04] <infinity> Interesting.
[10:04] <infinity> I have such shit upstream, and shit bandwidth/latency to the DC that I really wouldn't care or notice.
[10:06] <jbailey> I usually don't care, except that this takes so bloody long as it is, I can't bear to make it longer.
[10:07] <infinity> Heh.
[10:14] <infinity> jbailey : Where's the glibc?
[10:14] <jbailey> infinity: Dude, gcc-3.4 ihasn't finished going up yet.
[10:14] <jbailey> karlheg: Hey!
[10:14] <jbailey> karlheg: I just pushed initramfs-tools 0.16 up.
[10:14] <infinity> jbailey : Hrm, I didn't see any rsync tempfiles...
[10:15] <infinity> jbailey : Still don't.  So I kinda assumed you'd stopped uploading.
[10:15] <jbailey> karlheg: I still haven't integrated everything from you, I needed to push this stuff first.
[10:15] <jbailey> Hmm
[10:15] <jbailey> That ssh session appears to be hung
[10:15] <infinity> jbailey : And, you have stopped.  At least, there's no rsync processes on chinstrap.
[10:15] <jbailey> In fact, Angie machine appears to be hang.
[10:16] <jbailey> No, just the network
[10:18] <jbailey> infinity: See?  Another good reason to use rsync.
[10:18] <infinity> Heh.
[10:18] <infinity> I fall back torsync when my first scp fails. :)
[10:18] <infinity> (which isn't often)
[10:18] <jbailey> Have network again.
[10:18] <jbailey> So it just sort of went insane.  Lovely.
[10:19] <jbailey> I now have a ping running as well.
[10:21] <infinity> Dude, why are you sending the orig?
[10:21] <infinity> No wonder you complain about how long it takes. :)
[10:21] <jbailey> I thought I'd be polite and not make you fetch it.
[10:22] <jbailey> I'll remove them from the next ones.
[10:22] <infinity> Well, that's sweet of you, but still. ;)
[10:22] <jbailey> I was mostly complaining before because the session was hung, I think.
[10:22] <infinity> Yeah, I wouldn't complain if I had your upstream.
[10:22] <infinity> Looks to be, what?  1Mbit or so?
[10:23] <infinity> Mine's 256 kbps.
[10:23] <jbailey> A little less ;)
[10:23] <jbailey> Yeah, sucks to be you.
[10:23] <infinity> Thpt.
[10:23] <jbailey> But we were the first kids on the block with 1200bps modems too, so nyaah
[10:23] <infinity> Pfft, I revel in my oldskoolery with the 110 baud accoustic coupler in the closet.
[10:24] <jbailey> You still have yours?
[10:24] <infinity> Yeah.
[10:24] <infinity> Tossed most of the ones in between.
[10:24] <jbailey> Our 50bps and 110bps modems are barely at the edge of my earliest memories.
[10:24] <jbailey> I do remember the 300 clearly.
[10:24] <infinity> The coupler is somewhat of a novelty, so I kept it.
[10:25] <jbailey> "For novelty use only"? 
[10:25] <infinity> My 300 was my first self-contained modem.  FANY.
[10:25] <infinity> No phone required.
[10:25] <infinity> s/FANY/FANCY/
[10:25] <jbailey> Our 300 would auto-answer (That's how we ran the bbs), but we had to use the rotary to dial in.
[10:25] <infinity> Then the 2400bps Hayes "smart" modem.  Still not sure what was so smart about it.
[10:25] <jbailey> The AT command set.
[10:26] <infinity> Yeah, brilliant.
[10:26] <jbailey> Hayes sold two lines.
[10:26] <jbailey> And then the US Robotics 9600 SysOp deal.
[10:26] <jbailey> Everyone openning BBSs for a month so they could get a cheap modem.
[10:26] <infinity> I skipped 9600, but got a USR 14.4/16.8 Courier DST.
[10:26] <infinity> I was so l33t.
[10:27] <jbailey> Ah, I took a detour with the telebit trailblazer.
[10:28] <infinity> And then we upgraded the BBS to 12 lines, all with ZyXEL 19.2's, just cause we got a volume deal on them.
[10:28] <jbailey> Upside: Unlike hayes, it got the speeds that it claimed.   Downside: Unlike hayes, noone else had one.
[10:28] <infinity> No one else had a 19.2, but like we cared, we were COOL.
[10:28] <jbailey> Creepy.  What software?
[10:28] <infinity> TBBS.
[10:29] <jbailey> Ah, I did rbbs-pc and then..
[10:29] <jbailey> maximizer?
[10:29] <jbailey> I can't remember what it was called.
[10:29] <infinity> Wow... TBBS.org... Neat... I had no idea anyone still used it.
[10:30] <infinity> Old fans die hard, I guess.
[10:30] <jbailey> infinity: Get better ago.
[10:30] <jbailey> amo
[10:30] <jbailey> Because you KNOW the g and the m are close on the keyboard.
[10:30] <infinity> Fat fingers.
[10:34] <jbailey> Reminds me.
[10:35] <jbailey> In Toronto you can get doughnuts delivered.
[10:36] <infinity> !
[10:36] <infinity> Krispy Kreme?
[10:36] <infinity> If so, I'm moving.
[10:37] <infinity> doko : OOo2 on i386 looks successful.
[10:38] <infinity> doko : dpkg-shlibdeps really hates you, though.
[10:41] <jbailey> Yeah, KK
[10:45] <doko> infinity: great!
[10:50] <jbailey> infinity: W00h00
[10:50] <jbailey> It should be there.
[04:27] <doko> infinity: I'm unable to reproduce the failure of expect-tcl8.3 on i386. even the configure check on the buildd looks correct to me.
[05:16] <doko> speeding up grep on amd64 by a factor of 40 ...
[05:47] <jbailey> doko: Eh?
[05:47] <jbailey> ccache for grep?
[05:47] <jbailey> =)
[05:48] <infinity> doko : I can't reproduce it either, neat.
[05:56] <doko> infinity: so try again?
[05:56] <doko> jbailey: not exactly, but the debian patches seem to add something, what is not optimized by -O2 ...
[05:57] <doko> I did have to wait some minutes for grepping 600MB language data ...
[05:57] <infinity> doko : I am retrying them.  Then again, it wasn't that long ago that they failed, so I don't have high hopes.
[05:57] <doko> infinity: ok, and what to do, if they fail again?
[05:57] <infinity> Dig deeper, I guess.
[05:59] <doko> infinity: so, I don't have access to the buildd :-)
[05:59] <infinity> Yes, I know.  I'll look into it.  Right now, I'm working on some bootstrapping issues.
[05:59] <infinity> And it's also 2am. :/
[06:00] <doko> infinity: gcc/glibc?
[06:01] <jbailey> doko: It seems to have shown up with a symbol error on his build.  I've asked him to paste it here in case it's something obvoius to you.
[06:01] <jbailey> Dunno why it didn't show up here.
[06:02] <doko> nice ... :-/
[06:02] <infinity> http://people.ubuntu.com/~adconrad/
[06:03] <infinity> That was building against Jeff's gcc and glibc.
[06:03] <infinity> I could try again against the glibc I just built, if that may make a difference.
[06:03] <jbailey> It ought not to.  Those were all clean builds from yesterday using my own builds.
[06:03] <infinity> Well, yes, I was assuming it shouldn't make a difference, hence the not installing new packages between builds.
[06:04] <infinity> Was just going to build all three, install them all, then kick fresh rebuilds via wanna-build.
[06:04] <jbailey> Lemme do a test in my chroots with -lm
[06:04] <infinity> But.  Y'know.  Nothing's ever simple.
[06:05] <jbailey> Oh.
[06:05] <jbailey> *sigh*
[06:05] <jbailey> Symlink points to /lib.  I thought I had fixed that. =(
[06:06] <infinity> In glibc?
[06:06] <jbailey> Erm.
[06:06] <jbailey> Right
[06:06] <jbailey> But I've build gcc-3.4 with this glibc.
[06:06] <infinity> Toss me a patch, and I'll rebuild glibc.
[06:06] <jbailey> Oh, I know what happened.
[06:07] <jbailey> Hmm.
[06:07] <jbailey> I need to think about this for a sec.
[06:07] <jbailey> This was the fix for amd64 native.
[06:07] <jbailey> then I did a build test of i386-amd64 biarch
[06:07] <jbailey> Didn't rebuild gcc-3.4 after that.
[06:07] <jbailey> *sigh*
[06:07] <infinity> Wasn't that supposed to be halfway there for breezy?
[06:08] <infinity> I was kinda surprised to see lib/lib64 popping up all over, personally.
[06:08] <infinity> I thought we really, really didn't want that.
[06:08] <doko> infinity: yeah, that doesn't exist on m68k ;-P
[06:08] <infinity> doko : Thpt.
[06:09] <doko> infinity: it makes ia32-libs a bit lighter
[06:10] <doko> gets rid off amd64-libs
[06:11] <jbailey> infinity: The problem isn't we don't have a better mechanism for handling biarch atm.
[06:13] <jbailey> I really have to wonder if ia32-libs can *actually* still be built atm.
[06:43] <doko> yep, you now have a build dependency on amd64-libs :-)