[12:04] <doko> lamont: the openoffice.org build did fail again on i386, same reason
[12:05] <doko> same with openoffice.org2
[12:39] <jbailey> Cool, it's into the testsuite this time, so disabling the mudflap thing was hopefully enough.
[01:11] <doko> infinity: ^^^ I don't know what lamont did before requeing the OOo* builds, but it didn't help
[01:57] <lamont> doko: I did a dpkg --configure on the annoyed package, and removed all the crap.
[01:58] <lamont> doko: which is to say, there's a pretty b0rked package in there.
[02:03] <infinity> lamont, doko : That dpkg segv is known, but some preliminary debgging has gotten knowhere, except "jee, looks like a horribly smashed stack, yay".
[02:04] <infinity> I'm hoping that drastically rearranging the mesa packages in the next few days will "work around" the dpkg bug, but I won't hold my breath either.
[02:06] <doko> it's just that during the past months OOo* packages only build on Mondays at new moon, when the tide is high and a new planet in our solar system is discovered :-(
[02:07] <lamont> :-)
[02:10] <doko> as more as you try to depend on external build dependencies, as more it becomes broken. obviously we don't have a different understanding of "broken" ;)
[02:17] <lamont> heh
[02:35] <lamont> and hppa climbs to within 14% of #1. :-)
[02:35] <lamont> aka #5 :-(
[02:56] <jbailey> Hmm.  g++-4.0 -m64 tests appear to be sucking wind.
[02:57] <jbailey> # of expected passes            215
[02:57] <jbailey> # of unexpected failures        1710
[02:57] <jbailey> # of unexpected successes       1
[02:57] <jbailey> # of expected failures          5
[02:57] <jbailey> # of unsupported tests          9
[02:57] <jbailey> I'm curious about the unexpected success. =)
[02:57] <lamont> that was testing to make sure something failed, maybe?
[02:57] <jbailey> No, dear.
[02:58] <lamont> hehe
[02:59] <jbailey> lamont: The glibc fix is working right for you then?
[02:59] <lamont> beautifully
[02:59] <jbailey> Cool.
[02:59] <jbailey> I'll need to hear from doko before I know what to do with this compile, assuming it produces packages correctly.
[03:00] <jbailey> I have this vague suspicion that it didn't actually produce the binaries or something.
[03:00] <jbailey> And sometime around then I need to get initramfs-tools made the default.
[03:01] <jbailey> lamont: If I do the upload of the kernel package, and a noop upload of the linux-source based on the new kernel package, will you take care of getting the changes into you repo?
[03:06] <lamont> ??
[03:09] <infinity> It is.
[03:09] <infinity> jbailey : There's no such thing as a "noop upload of linux-source", all the linux-image binaries are generated by it.
[03:10] <jbailey> lamont: Umm.  Whatever bits contains make-kpkg
[03:10] <jbailey> lamont: Which is the bit that needs changing.
[03:10] <infinity> Oh, "kernel-package"
[03:11] <jbailey> Right.
[03:11] <infinity> But still, no such thing as a noop upload of linux-source.  Unless you mean "no source change upload", with an non-op of rebuilding every kenrel image. :)
[03:11] <jbailey> no-op from my point of view.
[03:11] <jbailey> ;P
[03:11] <infinity> Oh.  No-effort, then. :)
[03:11] <lamont> ah, right.
[03:11] <infinity> I'll accept that.
[03:12] <lamont> so you upload kernel-package with the change, then fix the build-deps and upload the kernel
[03:12] <jbailey> Yup
[03:12] <infinity> (Though, you'll want to put in at least minimal effort to bump the kernel-package build-dep, if you don't want stuff to blow up)
[03:12] <jbailey> But I want to make sure that I don't make it hard to keep your baz repo happy for the kernel stuff.
[03:12] <lamont> ah, ok.  I get it...
[03:13] <lamont> how soon you want to do this?
[03:13] <infinity> Merging in a 1 line change plus a changelog entry shouldn't be too tough. ;)
[03:13] <lamont> ah, doesn't much matter...
[03:13] <jbailey> lamont: Tongiht, ideall.
[03:13] <jbailey> +y
[03:13] <lamont> jbailey: you hooked up for baz hackery?
[03:13] <lamont> or is that what you're asking me to do?
[03:13] <jbailey> infinity: I think they name the branches after the version numbers.
[03:13] <jbailey> lamont: I'm asking you to do that.  baz and I don't get along.
[03:13] <lamont> ok.
[03:13] <lamont> either I'll do it or I'll dump it on fabbione. :-)
[03:14] <infinity> jbailey : You're not alone.  baz and I have had some pretty heated arguments.
[03:14] <jbailey> Sound slovely. =)
[03:15] <jbailey> infinity: I suggest hanging out on #bzr and encouraging mpool to ignore any heritage they consider taking from baz. =)
[03:15] <infinity> I figure the path of least resistance there is to re-brand svn as bzr and let it loose on the world.
[03:15] <jbailey> infinity: I think they've found it helpful to have me occasionally say "What?  Why would I want to know what the hell a library is before I do a checkout?"
[03:15] <infinity> While I may like that, however, I suspect it may irk others.
[03:16] <jbailey> Rebranding git as bzr would be a better match.
[03:16] <jbailey> At the moment they serve about the same goals.
[03:16] <desrt> 'git' is a much cooler name, though
[03:16] <jbailey> git/cogito
[03:17] <jbailey> desrt: I've assumed the pronounciartion of 'bzr' is 'buzzer', which is kinda fun, too.
[03:17] <desrt> jbailey; you pronouce it like "cathedral" backwards :)
[03:17] <lamont> jbailey: I've heard it pronounced bazzer
[03:17] <jbailey> I like the Linus quote on the OLS t-shirts this year, something to the effect of "I'm an egotistical bastard.  I name all my projects after myself. First Linux, now git"
[03:18] <lamont> lol
[03:19] <jbailey> Ooo Ooo!  debhelper time!
[03:39] <jbailey> BOOYAH
[03:39] <jbailey> I have a biarch gcc-4.0 now, too
[03:39] <jbailey> And no amd64-libs on my system
[03:41] <infinity> \o/
[03:43] <jbailey> infinity: What hours are you working these days?
[03:43] <jbailey> infinity: If LaMont's away, I'm goign to need you for the bootstrap fun.
[03:43] <desrt> jbailey; dude.  infinity is a busy guy
[03:44] <jbailey> desrt: YEs, and I'm about to make him busier...
[03:44] <infinity> In theory, 10-6 AEST (2400 - 0800 UTC, I think)
[03:45] <infinity> In practice, I'm in and out all evening, unless I have something else to do.
[03:45] <infinity> desrt : I appreciate the 3rd party atempt to dodge work, but jbailey's asking e to do something suqarely in my job description (for once, yay), so I can't very well cop out with "I'm budy". :)
[03:45] <infinity> s/budy/busy/
[03:46] <infinity> Although, how anyone could want help from someone whose typing is this horrible, I don't know.
[03:46] <infinity> Maybe I need to turn off all the filesharing applications on my girlfriend's computer, so I'm not typing 12 words ahead of my visual feedback.
[03:47] <jbailey> infinity: 'kay.  Once I get doko's final okay on the gcc-4.0 failures, I have the binaries that you need to install in the chroots.
[03:47] <jbailey> infinity: Doing the rebuilds shouldn't take any of your attention (Just one each, about 3 hours per gcc, 2 hours for glibc)
[03:48] <jbailey> infinity: Then drop those in the buildd chroot and feed it the source packages that I'll have signed for you.
[03:48] <infinity> jbailey : <nod>.. I understand the concept. :)
[03:48] <jbailey> infinity: Cool, it's as much to confirm the steps for me as anything. =)
[03:48] <infinity> jbailey : Just be a dear and sign the binaries you expect me to bootstrap with as well.
[03:48] <jbailey> infinity: Sure.  Signed changes file good enough?
[03:48] <infinity> Good enough for me, yeah.
[03:49] <lamont> better be... :-)
[03:49] <lamont> infinity: thanks man
[03:49] <infinity> Unless you're a l33t MD5 hax0r.
[03:49] <jbailey> infinity: Will you be fussed if the glibc you build with is slightly different that the one you're building?
[03:49] <jbailey> infinity: My current build doesn't have lamont's hppa patch in it yet.
[03:49] <infinity> Doesn't hurt my feelings any.
[03:50] <jbailey> And the gcc-3.4 that I'll be building suggests libc6-dev-amd64 instead of libc6-amd64
[03:50] <jbailey> So basically stupid little changes like that.
[03:50] <jbailey> But that I'd rather not piss away 8 hours on rebuilding for if I can avoid it.
[03:50] <infinity> I'll get over it.  So long as the final binaries produces at the end of the bootstrap are correct.
[03:50] <jbailey> That's the promise.
[03:50] <infinity> That's the whole point, really.  The intermediate shit is expected to be goofy anyway.
[03:51] <lamont> jbailey: and technically, it's two builds of each....  That is, what gets uploaded is built using only debs built in the data center
[03:51] <infinity> He doesn't need to care about that last step, mind you. :)
[03:51] <lamont> infinity: true
[03:52] <jbailey> Right. ;)
[03:52] <infinity> Well, he does if the new gcc/glibc combo somehow fails to build itself, but that shouldn't happen.
[03:52] <jbailey> lamont spent alot of time convincing me to not care about buildd time.
[03:52] <lamont> I just wanted him to understand that it was actually a small amount of work..
[03:52] <jbailey> Something or other about Ubuntu not having m68k whiners^Wmachines that we have to worry about bogging down.
[03:52] <infinity> buildd time in Ubuntu is pretty.. Plentiful.
[03:53] <lamont> yeah.  cycles we got
[03:53] <infinity> It's not the "lack of m68k" factor, so much as the "overkill redundancy" factor.
[03:53] <jbailey> True.
[03:53] <lamont> OTOH, if you don't have my glibc fix, we'll have to talk...
[03:53] <lamont> :-[)
[03:53] <infinity> We can rebuild the archive in n otime flat on a single buildd of any of the 3 release arches... So what do we do?.. Have 3 buildds for each.  Yay.
[03:53] <jbailey> But fundamentally when I upload, I usually think about "Is this change enough to merrit bogging down an m68k machine for a day or two?"
[03:54] <jbailey> (In Debian, obviously)
[03:54] <jbailey> Or an arm machine.
[03:54] <infinity> jbailey : I wouldn't worry about the m68k case in Debian either, if I were you.
[03:54] <infinity> jbailey : If we're backlogged, your 5 uploads in a day will go unnoticed anyway, and if we're caught up, we're caught up, so what do you care? :)
[03:55] <jbailey> True. =)
[03:55] <jbailey> Does the buildd still kill in-progress builds when w-b notifies that a newer version is going?
[03:55] <infinity> If the buildd gets the mail, sure.
[03:56] <infinity> (Well, it doesn't kill the build, it just purges the build directory, which causes it to fall the fuck over miserably and confuse people reading the logs)
[03:56] <infinity> But, in essence, it works.
[03:56] <jbailey> Ahahaha
[03:56] <jbailey> sweet
[03:56] <jbailey> The buildd software is such cruft.
[03:57] <infinity> Removing the build tree out from under the build is the most reliable way to make it fail in a hurry.
[03:57] <infinity> It's just confusing to understand why it died, if you're not used to it.
[03:58] <lamont> mind you, I don't think hppa/ia64 do the nuke-the-build-tree trick
[03:58] <jbailey> Oh sure.  And a tire-iron is the fastest way to get a cyclist off his bicycle.
[03:58] <jbailey> But still...
[03:58] <infinity> No, a new driver in Germany is.
[03:58] <infinity> Which is a joke that my adopted countrymen here iN Australia may find in poor taste...
[03:59] <infinity> (Did you guys get that news story over there?)
[03:59] <jbailey> I do remember a friend of mine remind me that the fastest way to a man's heart is through the breastplate.
[03:59] <jbailey> Followed by someone else pointing out the through the base of the throat might be faster because you're pretty guarnateed to get it right in one hit.
[03:59] <jbailey> infinity: I don't think so...
[04:00] <jbailey> Ahahaha
[04:00] <infinity> jbailey : Oh, a few weeks ago, the entire Australian women's cycling team was run over in Germany by someone who had been licensed for only a few months.
[04:00] <jbailey> Google images gives this hit first for "Tire Iron":
[04:00] <jbailey> http://www.dot.wisconsin.gov/safety/motorist/behaviors/aggressive/
[04:00] <jbailey> infinity: Oh ouch. =(
[04:01] <infinity> http://news.bbc.co.uk/sport1/hi/other_sports/cycling/4695385.stm
[04:02] <jbailey> infinity: Oh ouch. =(
[04:02] <infinity> That's okay, I didn't know any of them, and besides, it's against my religion to offer condolences.
[04:03] <jbailey> Oh, I know why I missed that.
[04:03] <jbailey> I spent the 20th sitting in the senate watching the same-sex marriage debate.
[04:04] <infinity> Oh, I came up with the solution to the same-sex marriage debate yersterday evening, in a fit of absolute brilliance.
[04:04] <jbailey> Oh?
[04:05] <jbailey> I'm happy enough with the solution that Canada came up with.
[04:05] <infinity> I was sitting here, filing my taxes, and noted that several government agencies, from two different governments, consider Zofia and I to be husband and wife, despite having never been issued a marriage certficate, realised that in modern western society, the "legal marriage certificate" has become completely meaningless (from a legal standpoint), and decided the fastest way to end to debate is to stop issuing them altogether, to any couples.
[04:05] <jbailey> That was one option that was debated.
[04:05] <desrt> jbailey; seriously though.  infinity has to fix X and the kernel :)
[04:06] <jbailey> The problem with it is that 'marriage' is an initernationally recognised term protected by various treaties.
[04:06] <jbailey> If we were to stop issuing marriage certificates, there are a number of legal inconveninences.
[04:06] <infinity> It even seems to have a positive side, that people who hook up, get drunk, and get married that night won't be considered "married" by the government, since you'd need defacto status (12 months or more, etc) to be considered actually committed to each other.
[04:06] <jbailey> It would be far better to issue one to everyone at birth.
[04:06] <jbailey> Relax the rules of marriage to be from 1 to N people.
[04:06] <jbailey> And just make everyone marries. =)
[04:07] <infinity> Hah, that's a curious option, too. :)
[04:07] <infinity> From the government standpoint, though, I do like the "only recognise people who pass the defacto test" concept.
[04:08] <desrt> what government is this?
[04:08] <infinity> My ficticious one.
[04:08] <infinity> That I will set up when I invade Tasmania.
[04:08] <desrt> so.... .nz?
[04:08] <ajmitch> infinity: on another note, is a test rebuild of the archive planned sometime before release?
[04:09] <infinity> ajmitch : It's ongoing.  elmo dumps and reloads the breezy-autotest wanna-build database every few days, and wee see how badly off everything is.
[04:09] <infinity> ajmitch : I've only been holding off on FTBFS bug filings, because I know a large number of those bugs are rooted in specific problems with packages that are being fixed.
[04:09] <ajmitch> alright
[04:10] <ajmitch> we'll probably need to get onto a few of those for universe
[04:10] <infinity> Once I /think/ things should be buildable, I'll start irritating the packages which clearly still aren't.
[04:10] <jbailey> infinity: I want lkh related bugs sooner rather than later!
[04:10] <jbailey> (Of course, lamont has been diligently finding most of those for me...)
[04:13] <jbailey> interdiff: gcc-4.0_4.0.1-4ubuntu2.diff.gz doesn't contain a patch
[04:13] <jbailey> hmm
[04:13] <jbailey> Ah, need -z
[04:14] <infinity> What have you got against debdiff? :)
[04:14] <jbailey> I want to send doko what I changed in the source package for review.
[04:15] <infinity> Yes, and?
[04:16] <infinity> 'debdiff 1.dsc 2.dsc' is a lot less typing.  Something I'm rather fond of.
[04:16] <jbailey> debdiff looks like it works on .deb files
[04:16] <jbailey> I did interdiff -z *diff*
[04:16] <infinity> Yeah, I do "debdiff *dsc". :)
[04:16] <infinity> Close enough, I guess, if your fingers are trained for the former.
[04:17] <lamont> jbailey: see http://people.ubuntu.com/~lamont/buildLogs/Test/Lists/breezy-autotest.all.$ARCH
[04:17] <lamont> iirc
[04:17] <jbailey> lamont: Nice, thanks.
[04:17] <lamont> of course, if something is failed there, and failed in buildLogs/Lists/breezy.all.$ARCH, then it could just be broken.
[04:18] <lamont> installed in breezy, and Failed/Building in breezy-autotest is generally bad.
[04:18] <lamont> (unless we're actually doing a build at that moment, which is rare)
[04:18] <jbailey> lamont: Grep from "Failed", yes?
[04:18] <jbailey> Cool, thanks.
[04:18] <lamont> grep Building
[04:19] <lamont> unless infinity has been marking them failed...
[04:19] <infinity> No, that's something I wanted to change.
[04:19] <lamont> Building --> really building, or failed-if-done
[04:19] <infinity> We should update the autosigner on sanae to auto-fail failed logs for -autotest, rather than leave them in building.
[04:19] <infinity> Since -autotest is supposed to be, y'know, automated.
[04:20] <lamont> hrm... good point..
[04:20] <lamont> you feel safe hacking that?
[04:20] <infinity> Yep, it's just awaiting a round tuit.
[04:20] <infinity> Which I should find after today.
[04:20] <lamont> well, if you haven't done it before I'm back on the 15th, I'll get to it..
[04:20] <infinity> (I have to attack a few BreezyGoals, before mdz and I meet this afternoon and he gives me hell for only doing half my job)
[04:21] <lamont> hehe
[04:21] <lamont> good idea
[04:29] <lamont> jbailey: that means fabbione love later tonight for the kernel, in all likelihood
[04:30] <jbailey> lamont: 'kay, thanks.
[04:30] <jbailey> lamont: Good sleeps, my friend.
[10:15] <doko> lamont, infinity: any news about the OOo* build failures / buildd b0rkage ?
[10:55] <doko> lamont, infinity: any news about the OOo* build failures / buildd b0rkage ?