[12:09] hmm i can't find the archive [12:09] ops [12:10] back again [12:11] hey doko [12:12] ftbfs on ppc [12:12] fuck [12:12] where are we? xorg? [12:13] doko: yes. xorg [12:13] LD_LIBRARY_PATH=../../../exports/lib XLOCALEDIR=../../../exports/lib/locale ../../../exports/bin/xcursorgen [12:13] +X_cursor.cfg X_cursor [12:13] libpng error: Call to NULL read function [12:13] PNG error while reading X_cursor-16.png! [12:13] xorg [12:13] ENOCLUE [12:13] ^5 [12:14] I don't think a give-back will help [12:14] otoh, why are we building the cursor here? isn't that arch-indep? [12:14] lamont: all the buildd's do have the new built-essential? [12:14] doko: yes [12:15] it's a library or something for the server [12:15] just waiting on xorg to finish up in the next 20-30 min, then we can play with libraries [12:15] ok, then I'll go through the libs in main and upload libs, which don't depend on glu-something [12:15] doko: please wait [12:15] let's get Xorg to build first [12:15] doko: given that we build pretty much all of main in < 6 hours or so, it's not like there's a big hurry.../ [12:16] and we can upload all the libs in one go [12:16] doko: the Xorg deps and build-deps aren't stable yet [12:16] so something might need to change [12:17] ok, will wait [12:17] + after X there is still one x package that needs upload [12:17] once that one is in, you are clear to go [12:22] fabbione: so about the ppc xorg ftbfs.... [12:22] thoughts? [12:23] lamont: no.. not really [12:24] i am checking the buildlog now [12:25] i386 and amd64 == uploaded [12:25] ia64 is about 5/14's done [12:26] it looks like an error in libpng to me [12:28] which leads to the question of what to do to fix it... [12:28] lamont: try to kick back [12:28] given back [12:28] it's not an X problem imho [12:28] the error comes from libpng attempting to read that file [12:28] and the file is valid [12:28] is the abuse in x-common such that we could stall on uploading 1.0? [12:29] lamont: do you feel more confortable to wait to upload x-common 1.0? [12:29] it's just question of removing a Pre-Depends [12:29] well, if we did that, then ia64 could catch up... [12:29] sure it can wait [12:29] i don't see a big problem with that [12:29] sparc will catch up too [12:30] but again.. once xorg-common -11 is in, the pre-depends can disappear :) [12:30] right. and that's the next cron.daily run [12:30] sure.. i didn't plan to upload x-common until at least the 3 major arches have all the binaries in place [12:31] doko: you planning to upload a fixed gcc-4.0 (is ftbfs) soon? [12:31] even if i would pay gold to go to sleep now :) [12:32] lamont: no [12:32] have to find out, why it fails [12:33] daniels: you awake yet? [12:35] ah, right [12:35] lamont: ? [12:35] doko: while you're in there, could you disable the testsuite on hppa? [12:35] ahah [12:35] fabbione: was aimed at doko and gcc-4.0 [12:35] ehhe yeah === warthylog [~warthylog@port49.ds1-van.adsl.cybercity.dk] has joined #ubuntu-toolchain === Topic for #ubuntu-toolchain: GNU Compiler Collection, Glibc, Binutils, Linux-kernel-headers | GLIBC Todo: hppa, sparc NPTL, i386 biarch === Topic (#ubuntu-toolchain): set by jbailey at Wed May 11 01:20:59 2005 === fabbione [~fabbione@port49.ds1-van.adsl.cybercity.dk] has joined #ubuntu-toolchain [12:53] fuck [12:54] server died again for no reasons [12:54] i had the time to hear all the disks spinning down nicely [12:54] and puff... it was frozen [12:57] overtemp? [12:58] fabbione: on the bright side, with ccache, it only takes 10 minutes for powerpc to reproduce the failure [12:58] lamont: nope.. it would send the alarm [12:59] hmmm [12:59] what version of libpng gets installed in the different chroots? [12:59] is it the same? [12:59] was the same buildd, same chroot [12:59] libpng12-0 [01:00] i mean across the arches :) [01:00] should be the same... /me checks [01:01] libpng12-0_1.2.8rel-1 on all 4 [01:01] OTOH, ppc is the only big-endian arch in the pile [01:01] yeah [01:01] Bah, /me fixes the typo and starts again. [01:01] short gym trip, eh>? [01:01] so is sparc... but it will take time to get there [01:02] yeah. gcc-4.0 build running now [01:02] lamont: I felt guilt, so I started the build on amd64 and ia64 as well, by the time I got them running, ppc had failed. [01:02] i wonder if libpng is broken on ppc [01:02] fabbione: It is. === lamont hugs jbailey [01:02] fabbione: sb had me test that yesterday [01:02] is fix known? [01:02] because it's blocking the transition [01:02] Ugh, really? [01:02] xorg is ftbfs/ppc [01:03] fabbione: you saw that xorg/amd64 was rejected, yes? [01:03] don't tell me that there is known fix and it had not been uploaded... [01:03] lamont: meh.. no why? [01:03] Gimme a sec to fire these things up. [01:03] Rejected: xprt_6.8.2-11_amd64.deb: old version (1:0.1.0.alpha1-10) in breezy >= new version (6.8.2-11) targeted at breezy. [01:03] so that's 2/3 [01:04] fabbione: Nope, I was just asked to test it. I was hacking on other stuff so didn't explore it any further. === lamont hopes we get things settled in the next 90 minutes, so he can go to his fire dept training [01:05] doko: ew... patch fuzz in gcc-4.0... :-) [01:05] hmmm [01:05] that should have happened on i386 too [01:05] xprt is an arch: any [01:05] and why on heart daniels still ship that shit [01:05] ok let's make a summary [01:06] libpng is fucked on ppc [01:06] that is a blocker for xorg [01:06] xorg needs xprt love [01:06] and another upload [01:06] sounds about right. [01:06] otherwise we cannot go for libs [01:06] or ppc will lag eons behind [01:07] and dropping ppc from breezy is _not_ an option... [01:07] should we reupload libpng compiled with 3.3 on powerpc? [01:07] doko: is that known to fix it? [01:07] doko: are we sure that is the right thing to do? [01:08] no [01:08] than no [01:08] fabbione: you have ppc at your place? [01:08] somebody needs to investigate the real issue [01:08] lamont: no [01:08] i have been begging for one for a long time === lamont does, but not really... would take 24 hours or more to test-build xorg [01:08] i keep getting i386trash [01:08] lamont: davis? [01:09] davis is an option.. [01:09] elmo about? [01:09] elmo can upgrade a chroot and that's it [01:09] exactly [01:09] jbailey, lamont: there is nothing more i can do for today [01:09] i am crashing [01:09] lamont: ? [01:09] g'night fabbione [01:10] daniels should be awake soon [01:10] good night [01:10] can you freshen the breezy chroot on davis? [01:10] elmo^^ [01:10] elmo: with the new xorg build-dep ;) [01:10] then I'm going to try building libpng with gcc-3.3, and have you install that in the chroot. [01:10] and then we'll see if that fixes the xorg build problem [01:10] if there is something, i will have my mobile phone with me [01:10] and then I'll cry [01:11] lamont: no.. just wake up the kid :) [01:11] heh [01:11] i am off [01:11] good night guys [01:11] lamont: I have to go in 10 minutes, is there anything you need me to do nowish? [01:11] g'n Fabio! [01:11] jbailey: not unless you can upload a fix for libpng. :-( [01:12] jbailey: any hints on what it was? or you just verified it was NFG? [01:12] Just verified it was nfg. [01:12] that is, braindump would be cool. [01:12] and done. sigh. [01:12] way quick braindump [01:12] lamont: doing [01:22] elmo: and libpng3 build-deps, of course. [01:23] nothing to install? [01:23] and done [01:23] probbaly nothing to install [01:29] ARGH [01:29] Rejected: xprt_6.8.2-11_ia64.deb: old version (1:0.1.0.alpha1-10) in breezy >= new version (6.8.2-11) targeted at breezy. [01:30] yeah - we discussed that a bit a while back. [01:30] it's on the list of things to deal with. [01:31] but libpng is a bigger blocker... [01:31] unless you wanna make the old xprt just go away... :-) But that would be, well, wrong. [01:32] elmo: in davis:breezy-chroot: dpkg -i ~lamont/breezy/libpng12*.deb please [01:32] eh, what are they? [01:33] they're the same source, built with gcc-3.3 [01:33] installed [01:36] dpkg-checkbuilddeps: Unmet build dependencies: xtrans-dev [01:36] installed [01:37] so if the libpng3 build fixes things, we'll know in about 50 minutes... if it doesn't, then I'll actually have a tree to start debugging on. [01:40] meanwhile, sort break time [01:48] lamont: what do you sort? [01:48] short [01:49] :-) [01:52] lamont: ok, I found the 4.0 bootstrap failure. mind, if I upload a fixed one? [01:52] doko: why would lamont mind? unlike Debian, he has 3 buildds per arch on a GB LAN :P [01:53] not that I'm bitter every time vore grinds to a halt after one of your upload sprees or anything [01:55] elmo: yes, that' why I uploaded the last one built for sparc ... [01:56] doko: dude, you REALLY shouldn't do that [01:56] please don't upload binaries unless you're a buildd [01:57] oh. debian. well, still shouldn't [01:57] doko: upload away [01:57] heh, it doesn't even go to sarge. lamont: will do [01:57] doko: this has my hppa fix too? [01:57] heh, in Ubuntu, you CAN'T upload binaries unless you're a buildd [01:57] (or have access to their secret key) [01:57] I wish I could enforce that level of facism in Debian :( [01:58] lamont: fix? crappy workaround of disabling the testsuite a fix? ;-) [01:58] doko: yeah. that "fix". :-) [01:59] doko: I'm willing to bet that it's the long-standing bug in signals, but really can't say for sure [01:59] i'm here now [01:59] elmo: xprt isn't something we care about for the time being [01:59] if the rest of the upload goes through, that's a win [01:59] the rest of the upload so doesn't go through [02:00] if it did, I'd shoot myself [02:00] daniels: it's kinda this atomic thing, you see... === lamont pictures elmo-with-a-gun, fears for sesame street [02:00] elmo: the xtrans stuff isn't a bug, mmkay [02:01] elmo: this is why I've said several times that I despise xtrans and want it to die [02:01] daniels: one option is to just not deliver the xprt binary package... [02:01] although I expect that breaks some other packages down the line [02:02] elmo: ok [02:02] hi daniels, short night ... ;) [02:02] i'll prepare a new xorg [02:02] heh [02:02] doko: he was gone almost 8 hours... [02:03] daniels: we also need to figure out what's up with libpng/ppc [02:03] lamont: you're right, that's too much [02:03] and note that fabbione had to do some stuff to xorg to make it build with gcc-4.0.... [02:03] if you have a shared source archive, greate... if not, you need to fetch from the archive... [02:03] yeah, I slept in by an hour [02:03] doko: and I need ccache bindings for ada, so that the gcc-4.0 build can go fast. [02:04] ditto for gcj [02:05] lamont: hmm, these are not yet upstream? [02:05] night all [02:06] lamont: a lot of the gcc slowness is hte test suite .. [02:06] elmo: not when it's disabled. :-) [02:06] g'night elmo [02:06] night [02:06] ccache doesn't help in the testsuite [02:07] right\ === lamont is watching the ada build churn along [02:07] while waiting for the xorg build to fail agian [02:08] elmo: if you're still around, I'm wondering why xorg seems to have made it in on i386 - was there no xprt agony there? [02:10] was xprt ftbfs on i386? [02:10] nigt elmo [02:12] daniels: dunno [02:12] what source package? [02:12] xprt [02:13] xprt doesn't show up in wanna-build for i386 [02:13] or any arch, for that matter [02:13] hoary or breezy [02:14] um [02:14] so this might sound a little bit nasty [02:14] but is there any chance you could hack the changes file to exclude xprt and shepherd the rest of the upload through? [02:14] no [02:16] agh [02:16] (not just because all the .debs and the .changes file get deleted...() [02:17] the ubuntu archive remains unviolated by hand-edited changes files, and shall remain so on my watch. === infinity violates is when lamont isn't looking. [02:17] s/is/it/ [02:19] infinity: you don't have the key to violate it [02:20] infinity: looks like your buildd career did finish before it did begin ;-P [02:23] ;) [02:23] lamont: http://people.ubuntu.com/~lamont/buildLogs/j/java-gcj-compat/1.0.28-0.0ubuntu1/ ??? [02:26] doko: any chance something Provides: libgcj-dev? [02:27] lamont: what, you're telling me you're the paragon of cleanliness and archive sanctity now? :P [02:27] daniels: I always play by the specified rules. [02:27] that's what makes it so much fun. :-) [02:28] ehh, yes ... libgcjX-dev ... looks like I should remove it [02:31] daniels: my forte lies in analyzing complex systems and determining how to accomplish a specified task within the structure imposed by that system. [02:32] some call it hackish and kludgy, others also call it necessary. [02:32] s/it/it only/ [02:33] ubuntu is nice because we have a very minimal set of complexity in the system. mostly, it's just implemented correctly to begin with. hence we can keep it more pure and clean... [02:33] that, and one of the rules is 'it shall remain unviolated' :-) [02:33] lamont : If that's your forte, you should have been a tax accountant, not a programmer. :) === lamont _does_ do his own taxes... [02:34] heh [02:34] infinity: note that said forte also applies very well in computer security, forensic analysis, and so forth [02:34] our family calls it "rules lawyering" [02:35] Oh, I realise it has many uses, it just happens that good tax accountants make a lot more than those other professions. :) [02:35] although, I must admit, life is more interesting when you have to explain both the letter of the law and the intent of the law to your kids, so as to avoid hearing "but dad, you said ..." [02:35] (I consulted to one who worked exclusively with clients who had salaries of over a million a year... His income was significantly higher) [02:35] heh [02:36] infinity: fwiw, I apply it pretty much to all aspects of my existance [02:36] lamont: please build 4.0-5, not -4 on your slooow hppa box [02:36] debian? [02:37] breeeezzzzyyy [02:37] so 0ubuntu5. check [02:37] actually, I think I'm going to let 0ubuntu2hppa1 finish building first. :-) [02:37] yes, skip 0ubuntu4 [02:38] ew. [02:38] hmm, I'm going to sleep, doing more harm then good [02:38] doko: given hoary/hppa's pseudo-existance [02:38] I really can't build any of the cxxlibs before you upload them, yes? [02:38] given breezy/hppa with g++4.0 [02:39] build them using g++-3.4, hack around gcc-defaults [02:40] actually, it just means adding build-essential and gcc-defaults to the no_auto_build list [02:40] since they haven't built yet. === lamont screams, uploads his libpng3 hack [02:41] dpkg-genchanges: binary-only upload - not including any source code [02:41] dpkg-buildpackage: binary only upload (no source included) [02:41] lamont: I don't understand what you gain [02:41] doko: fix gcc-{3.4,4.0} so that it builds libpng3 correctly [02:42] doko: gain with not building build-essential, gcc-def? it means that I don't have to hack around gcc-defaults... [02:42] then once you upload the libs, then it's no issue again [02:43] yes, but you end up with the wrong C++ ABI [02:44] so you verified, that libpng3 is miscompiled with gcc-3.4 _and_ 4.0 on powerpc? [02:44] libpng doesn't Depend: C++ [02:45] seems to be wrong on gcc-4.0. didn't check 3.4 [02:45] I didn't write this [02:45] ahh, ok [02:45] daniels: could you destill a testcase from the xorg build? [02:45] um, possibly [02:46] Depends: libc6 (>= 2.3.4-1), zlib1g (>= 1:1.2.1) [02:46] Depends: libpng12-0 (= 1.2.8rel-1ubuntu1), zlib1g-dev [02:47] lamont: if you do have time, could you recheck with a libpng3 compiled with -O2 and/or -O1? [02:49] sure [02:51] doko: except that I was testing it by using the package installed in the chroot... and I can't update that... [02:51] actually, it's built without -O [02:51] see debian/patches/makefile.patch [02:51] ? [02:52] nm [02:52] gcc -Wall -D_REENTRANT -g -O3 -c -o pngset.o pngset.c [02:52] ah. that part comes from debian/rules [02:53] -O3?? [02:53] daniels: yeah. crack [02:53] DEB_BUILD_OPTIONS overwrites this ... [02:53] doko: yes [02:54] looking at the archives I see issues with -funroll-loops, which is enabled by default in 4.0 at -O3. So I assume -O2 will help here ... [02:54] hrm... I suppose LD_LIBRARY_PATH could help force things to the rebuilt package. [02:54] (breezy-chroot)daniels@davis:~/redglass $ xcursorgen X_cursor.cfg X_cursor [02:54] (breezy-chroot)daniels@davis:~/redglass $ dpkg -l libpng\* [02:54] [...] [02:54] ii libpng12-0 1.2.8rel-1ubuntu1 PNG library - runtime [02:54] that's the new version. [02:55] which is built with gcc-3.3 [02:55] note the lack of segfault [02:55] you have ppc there? [02:55] note the 'daniels@davis' ;) [02:55] ah, that's in the same chroot where I just verified my fix. [02:55] oh, so you installed it? [02:55] note that -1ubuntu1 source will be in the archive in about 8 minutes [02:55] no, elmo did. before he left. === lamont has no root on the porting boxes [02:56] ok, going to bed now, 3am ... [02:56] good night [02:56] g'night doko [02:56] daniels: once we have xorg/xprt love, then we can upload doko's libraries [02:56] doko: before you leave... [02:56] yes? [02:56] where do all the libs live, so we can upload them once X is happy? [02:57] chinstrap:~doko/cxxsrc, but without kdelibs please [02:57] no kdelibs. check [02:57] lamont: yep, I'm testing the xprt-less build [02:57] daniels: once you get an xorg that is xprt-happy, let me know. (then we can see about finishing debugging the libpng thing...) [02:58] doko: in any case, xcursorgen, and davis:~daniels/redglass/X_cursor*, is your testcase [02:58] lamont: will do === warthylog [~warthylog@port49.ds1-van.adsl.cybercity.dk] has joined #ubuntu-toolchain === Topic for #ubuntu-toolchain: GNU Compiler Collection, Glibc, Binutils, Linux-kernel-headers | GLIBC Todo: hppa, sparc NPTL, i386 biarch === Topic (#ubuntu-toolchain): set by jbailey at Wed May 11 01:20:59 2005 === #ubuntu-toolchain [freenode-info] why register and identify? your IRC nick is how people know you. http://freenode.net/faq.shtml#nicksetup [06:54] (fabbione/#ubuntu-toolchain) at what time did i go offline? [06:55] (jbailey/#ubuntu-toolchain) daniels: if gcc-3.3 fixed it, then it's probably altivec fuckage. I can try to take a look at that tomorrow. [06:57] And in the meantime, Fabio showing up for his morning work is my hint that it's bed time. [06:57] g'n all! [06:57] hey jb [06:57] jbailey: klibc is FTBFS on sparc :) [06:57] goot night [06:57] night jbailey [06:57] daniels: want to try a -14 that has fabbione's fixes from -11?????>????"?? [06:57] fabbione: 3h17m before you joined [06:57] fabbione: Forward me a build log? It's working everywhere else. =( === lamont goes to bed [06:58] lamont: the what now? [06:58] all the bugs that fabbione fixed so that it actually builds with gcc-4.0 were reintroduced in your upload. [06:58] please fix them. [06:58] oh, biteage [06:58] * Add patch 991_ubuntu_gcc_flags.diff as a temporary workaround to a borked [06:58] gcc-3.4 check. [06:58] oh fuck.. daniels you didn't merge from the archive? [06:58] also, please test your upload in a gcc-4.0 chroot, k? [06:59] fabbione: no, I didn't know you made any changes to xorg [06:59] lamont: this is with gcc4 [06:59] fabbione: I want to work magic of some sort that combines all the build daemon activity, including ports into an rss feed that I can just look at and show even things that would be on ports.ubuntu.com. [06:59] fabbione: is it just #991? [06:59] daniels: and build-deps [06:59] ok [06:59] daniels: go back about 6 hours in your scrollback... [07:00] daniels: we also need to figure out what's up with libpng/ppc [07:00] May 17 18:03:12 lamont: you're right, that's too much [07:00] May 17 18:03:14 and note that fabbione had to do some stuff to xorg to make it build with gcc-4.0.... [07:00] May 17 18:03:39 if you have a shared source archive, greate... if not, you need to fetch from the archive... [07:00] although maybe that was pre-coffee [07:00] anyway, past my bedtime [07:00] good night lamont [07:01] night lamont [07:01] fabbione: -13 is ftbfs, libpng should be fixed, just need a gcc-4.0 friendly -14, and xorg should actually make it into the archive. [07:01] lamont: ok thanks === lamont will be back in about 7-8 hours, modulo getting kids out the door to school [07:01] i won't be around that long [07:01] not after 19 hours of yesterday [07:02] fabbione: sure [07:02] heh [07:02] at any rate, if things need buildd love, I expect you'll see elmo before me., [07:04] indeed, i'll get it sorted out somehow === infinity [~adconrad@loki.0c3.net] has joined #ubuntu-toolchain [07:08] i need to do a mirror sync to get the new packages, so I'll go afk for about an hour while my internet connection is useless [07:11] daniels: my archive is synced [07:11] do you want me to merge for -14? [07:11] it will take me the same amount of time as for you to wait [07:15] it's ok, i'll do more testing in chroots and stuff === doko [~doko___@dsl-084-059-032-191.arcor-ip.net] has joined #ubuntu-toolchain [07:26] hmmm something is already pulling in x-common [07:26] oh right [07:26] xrender [07:30] yep [07:30] the /usr stuff should pull it in [07:30] yeah it does [07:30] else it'll be stuck in /usr/X11R6 [07:30] but it's ok [07:30] the buildd likes it [07:31] do xft and xcursor have the pre-depends? [07:31] no, but they depends on xrender [07:31] that pre-depends on x-common [07:31] so that's enough to pull in * in the proper wat [07:31] way [07:31]