=== yofel_ is now known as yofel === dk__ is now known as DKcross === yofel_ is now known as yofel [19:58] o/ [19:58] huhu sistpoty :D [19:59] hi sebner [20:01] so welcome everyone [20:01] who's around for the practical fix ftbfs session? [20:01] * slytherin raises hand [20:01] I am until the baby starts crying :P [20:02] hey sistpoty [20:03] ok, so while the last session tried to give a walkthrough, let's get our hands dirty in this one? [20:03] I assume you know what FTBFS, BTS and PTS mean? [20:03] anyone who doesn't? [20:03] I don't know about BTS and PTS, bug tracking system, package tracking system ? [20:04] exactly [20:04] BTS == Bug Tracking System indeed [20:04] PTS -- IDK [20:04] so our worklist is still http://people.ubuntuwire.org/~wgrant/rebuild-ftbfs-test/test-rebuild-20090909-karmic.html [20:05] every package there has a link to BTS and PTS, where a fix might already be in a) a newer version or b) sitting in a bug at debian [20:05] of course looking at launchpad bugs page for a package also is a good idea [20:05] afaik, the list is updated every 2 hours, and uploaded packages are marked as superseeded [20:06] so everyone grab a package please, and tell which one you'll start working on [20:06] hggdh: PTS = Package Tracking System = http://packages.qa.debian.org/common/index.html [20:06] I am getting checkinstall [20:07] * sistpoty randomly grabs nstx [20:07] I have already filed three sync request, but they did not get processed. Should I bug archive admins? [20:08] * fabrice_sp grabs contacts [20:08] slytherin: depends how long these were in the queue... if these weren't handled during the last batch of syncs, then asking might make sense [20:08] they have been there for last two days [20:09] what is the frequency of the batch of sync? [20:09] slytherin: iirc there wasn't a batch yet, so patience :) [20:09] fabrice_sp: no idea to be honest. I assume it depends on the workload of archive admins [20:09] ok :-) [20:09] Ok. Iwill keep filing bugs. [20:09] so as everyone got a package now, start fixing it :) [20:10] * RainCT takes aegis [20:10] if there's anything you have no clue how to fix it, ask :) [20:11] ideally with a pasted log of the build error (the lines of the error suffice) and snippet of the code pasted [20:11] so then we can all look and try to find out how to solve the issue [20:11] and if anyone is done and can't upload himself, there should be many sponsors here :) [20:12] syspoty, please see http://pastebin.ubuntu.com/289481/ (nice, i386). What is the error here? Missing build-depends for jikes? [20:13] systpoty ^ [20:13] hggdh: looks more to me that jikes-classpath -> jikes cannot be satisfied [20:14] hggdh: are you on i386? [20:14] no, amd64 [20:14] hggdh: oh [20:14] hggdh: otherwise you could have tried to manually install build-dependencies [20:14] sistpoty: hggdh: jikes is removed form archives. Grab the version of nice from Debian and see if it builds. Should be a simple sync. [20:15] hggdh: of course you could use a linux32 chroot (/me tries to recall the command to create one) [20:15] slytherin: from unstable? [20:15] * fabrice_sp feels lucky: contacts only misses a build-dependency :-) [20:16] hggdh: yes [20:16] k [20:17] installwatch fails to build due a scandir64 header mismatch [20:18] joaopinto: need help or can you fix it? [20:18] jikes on Debian unstable is only for m68k [20:19] hggdh: I meant grab version of nice from Debian. [20:19] I guess i just need to copy the new function definition from dirent.h [20:19] heh. Sorry [20:20] hggdh: I'm quite clueless about java packages, but maybe there's a different -classpath to build-depend on? [20:21] sistpoty: I already gave him solution. [20:21] :) [20:22] for problems with java packages, ping me [20:22] slytherin, sistpoty: unstable nice does not (build) depends on jike anymore. [20:23] hggdh: make sure it builds in karmic chroot and then use requestsync to file sync bug, attach the build log, Iwill mark the bug confirmed. [20:23] roger [20:24] -chat not in use [20:24] ? [20:24] * slytherin is checking 'checkstyle' FTBFS [20:25] maco: the session was announced only 2 hours, is there a conflicting schedule? [20:25] 2 hours ago even [20:25] sistpoty: no i was confused to see > 1 person talking in here and nobody talking in -chat [20:26] ah, well, it's an interactive session :) [20:26] sistpoty is a champion organising meetings :-D [20:26] haha [20:27] why is elisa source still in archives? [20:28] slytherin: packages.ubuntu.com shows python-elisa being build from it [20:29] that's weird, I thought everything was build from moovida now. [20:30] wow we have three source packages starting with name feisty. I guess they should be removed now. [20:32] ok, checkinstall builds, but I have changed directly the source [20:32] time to do a runtime test [20:32] Patch 15fix-kfreebsd.diff does not remove cleanly (refresh it or enforce with -f) [20:32] grrrrrrrrrrrrr [20:33] slytherin: I do not have a chroot available right now, so I submitted nice to my PPA [20:33] I should have debclean first [20:34] hggdh: if it is too much trouble for you I will check and file sync bug [20:35] slytherin: no prob. I will wait for the PPA build to complete, and then open a sync req [20:35] ok [20:36] dammit. forgot to specify where the dput was going to :-(. Ah well, it will be refused, anyway [20:37] joaopinto: pbuilder is probably a cleaner way to test the build [20:37] contacts uploaded [20:37] * fabrice_sp_ is looking for another sexy pacakge [20:37] right now I am puzzled with the existing patchs versus my patch [20:38] sistpoty: http://paste.ubuntu.com/289506/ does that make sense? [20:39] maco, I prefer chroots :P [20:40] RainCT: is needle/haystack fumbled upon later? if not, it'd be better to declare haystack/needle as unsigned const char* [20:40] * fabrice_sp_ is looking at comedilib [20:40] joaopinto: oh ok i thought you were doing just on the system without a chroot at all (though pbuilder is just an automated chroot, i think) [20:40] arch, quilt, I do I add a new patch with quilt ? [20:40] brb, 20 minutes, baby bath [20:41] joaopinto, you mean how? [20:41] ok [20:41] checkstyle has too many build-dep, download takes lot of time. moving on to other packages. [20:41] RainCT: if the contents are changed later, it makes sense. (though the slightly more readable way is needle=const_cast(haystack_start);) [20:41] (applying for c++) [20:42] sistpoty: does haystack++ qualify as "fumbled upon"? [20:43] RainCT: no... *haytack=... would imply that [20:43] *haystack = [20:43] even though fumbled upon is a vague term, I admit :P === fabrice_sp_ is now known as fabrice_sp [20:47] sistpoty: well, the problem is it complains because of "memchr(haystack, *needle, haystack_len)", which I see at http://www.cplusplus.com/reference/clibrary/cstring/memchr/ is overloaded for both "const void*" and "void*" [20:48] so dunno why it complains, maybe because it's doing "return memchr(..)" and the function is defined as returning "void *"? [20:49] or because of the second arugment being const? [20:50] RainCT: can you paste the entire function? [20:50] fabrice_sp, yes, sorry, I meant, how [20:50] sync requested for libgnu-regexp-java [20:51] joaopinto, quilt new to create the patch [20:51] grr, a glibc change and quilt was not a goood choice :P [20:51] quilt add zfile> to add a file to the patch [20:51] lol [20:51] seems to be the later [20:51] ah no, it failed :P [20:52] fabrice_sp, do i need to care in which stage of the build I am ? like debclean first ? [20:53] grr, I should fix cowbuilder on my laptop, compiling on the netbook is awful :P [20:53] joaopinto, better have a clean env before patching [20:53] sistpoty: http://paste.debian.net/48656 [20:54] well, a proper fix for this wold require a GLIBC version check, but let it just build [20:55] sync requested for libgnujmi-java [20:55] RainCT: ah, it's an extern "C" function. That's a very interesting construct, because it sees the c++ definition [20:56] RainCT: so the fix is in line 32 of your past [20:56] RainCT: return (void *)memchr(haystack, *needle, haystack_len); [20:57] I need to read a quilt howto [20:57] sistpoty: ah! ok, thanks [20:58] joaopinto: https://wiki.ubuntu.com/PackagingGuide/PatchSystems#quilt%20%28example%20package:%20xterm%29 :) [20:58] joaopinto: in short: export QUILT_PATCHES=debian/patches [20:58] joaopinto: quilt push -a [20:58] joaopinto: quilt new [20:58] joaopinto: quilt shell (then edit all files of interest, exit subshell) [20:58] joaopinto: quilt refresh [20:58] joaopinto: quilt pop -a [20:59] ok, let me add that to my notes [20:59] tks [20:59] Oh, nice. So that annoying "quilt add .. # and if you forget it you're screwed" isn't needed?? and I didn't know until now???? :P [21:01] RainCT: /me overheard it in -motu recently ;) [21:01] I've just discovered quilt shell :-/ [21:02] hmmm, comedilib is not a so nice choice (problems with regenerating configure script :-/ ) [21:02] janito@janito-desktop:/tmp/quilt-mU0V5M$ quilt refresh [21:02] Nothing in patch 20glibc2.10-scanw.diff [21:03] is this the expected output from refresh ? [21:03] no [21:03] joaopinto: you need to exit the shell first then do refresh [21:03] oh, ops, forgot to exit :P [21:04] filed sync request for libpicocontainer-java [21:05] slytherin: I believe you're a robot with your ftbfs-fixing speed :P [21:06] I think I am done. Filed quite a few sync requests which were easy to verify. Others are difficult because of the download required (and it is 1:30 am here) [21:06] I will continue tomorrow [21:07] thanks a lot slytherin! [21:07] you are welcome. [21:13] Ok I'm looking at the pygobject ftbfs, because it looked quite simple to fix, but the script missing is confusing me [21:13] I'm pretty sure I just need to bump the versions of automake and autoconf it's looking for [21:14] the error "The important program mcopidl was not found!" should be fixed by adding --without-arts to configure script, no? [21:14] fabrice_sp: afict, yes [21:14] hggdh: are you done with nice build? [21:14] funkyHat: does it run autotools during build? or just the configure? [21:15] funkyHat: if the latter, I assume you could try to regenerate the autotools as a patch [21:15] sistpoty, if it does not work, what else can I try ? (log: http://pastebin.ubuntu.com/289531/) [21:15] * funkyHat feels stupid. How would I do that? [21:16] funkyHat: run autoreconf -i in the top-src-directory (assuming you have build-depends installed) [21:17] fabrice_sp: have you tried to run configure --help manually? [21:17] fabrice_sp: maybe it's --disable-arts [21:17] ohhh, right [21:17] fabrice_sp: others than that, I guess the solution is found in configure.ac (or configure.in whichever exists) [21:18] ok. Will check both. [21:18] thansk ;-) [21:18] thanks for trying to fix it :) [21:18] I have done quilt push -a first, this is right isn't it? [21:19] funkyHat: and have QUILT_PATCHES set? then yes [21:20] Yes I set that too :) [21:20] funkyHat: then quilt shell [21:20] funkyHat: autoreconf -i [21:20] funkyHat: cross fingers :P [21:21] ah, quilt shell, I didn't do that -.- [21:21] can I do quilt add . ? [21:21] funkyHat: no idea really, have discovered how quilt works myself only recently. maybe someone else knows? [21:22] <- back in 5 minutes [21:22] I'll start again, not like I've done anything complicated [21:22] hggdh: I am done with nice build. I will file sync request if you are not doing it. [21:22] funkyHat: . might work too or you use quilt add * :) [21:23] sebner: ah, yeah * may have worked, too late! [21:23] heh [21:24] oh, I just did quilt shell before creating a new patch. bah! [21:28] hggdh: done. filed sync request for nice. [21:33] W: checkinstall source: package-uses-deprecated-debhelper-compat-version 4 [21:33] should I fix this ? [21:33] or i should simply introduce furhter differences to the Debian package ? [21:34] erm, avoid introducing [21:34] Ok, the patch quilt has generated is 50k lines long... so I'm missing something :D [21:34] joaopinto: at feature freeze, try to keep the changes to a minimum, so don't fix lintian warnings [21:35] ok, added a 2 lines patch , it builds, next, changelog ? [21:35] joaopinto: yep [21:35] funkyHat: that's quite usual for running autotools [21:35] checkinstall (1.6.1-8ubuntu1) karmic; urgency=low [21:35] looks good ? [21:35] yes [21:36] but add some content :) [21:36] http://ubuntu.pastebin.com/m6816d943 is the list of files that quilt adds when I exit the shell [21:37] * fabrice_sp uploaded cyclades-serial-client [21:38] funkyHat: you can drop config.h.in~ (backup file) [21:38] funkyHat: but all others are expected [21:38] funkyHat: eventually you can drop the files in autom4te.cache (not too sure though) [21:39] Is there a simple way to do that with quilt? [21:39] Oh, remove :) [21:39] sistpoty, http://pastebin.com/m666b289f <- looks good ? [21:40] joaopinto: maybe describe what it does? (as in change foobar to fix FTBFS)? [21:40] ok [21:41] grrr, brb [21:41] anybody have an idea about "channel.c:3437: error: format not a string literal and no format arguments" [21:41] fabrice_sp: generally yes, but not without the context [21:41] line is "sendto_local_ops_flag(UMODE_SERVNOTICE, get_str(STR_SPLIT_MODE_OFF));" [21:41] sorry [21:41] I'll pastebin the source [21:42] fabrice_sp: which package is this? source of the function might not give what I'm looking for (definition of UMODE_SERVNOTICE, definition of senddto_local_ops_flag) [21:43] dancer-ircd [21:43] * sistpoty looks [21:45] it seems to be because of the get_str call [21:46] sistpoty: so is the massive patch going to cause problems (assuming, of course what I've done actually fixes it)? [21:47] funkyHat: nope, it's just rerunning autotools ;) [21:47] Ok :), just seems a bit weird putting such a huge patch in debian/patches [21:48] fabrice_sp: that's just done really, really ugly from what I've seen :( [21:48] :-/ [21:48] fabrice_sp: in src/numeric.c, you can see the string definitions for get_str [21:48] Could I run autoreconf -i in debian/rules instead? [21:49] fabrice_sp: if these don't have any format character in them, you can use sendto_local_ops_flag(UMODE_SERVNOTICE, "%s", get_str(STR_SPLIT_MODE_OFF)); [21:49] fabrice_sp: which in this case is the right thing, but I'm sure there are other occurances as well [21:49] funkyHat: then you'd need to build-depend on automake [21:49] sistpoty, ok. I'll check the content of the string [21:50] fabrice_sp: for this case, it should work... just seen it ;) [21:50] cool :-D [21:50] Well, I'll wait and see if this even compiles to begin with :) [21:58] sistpoty, Added patches/20glibc2.10-scandir required to build with glibc2.10 [21:58] \o/ [21:58] the line on changelog [21:59] erm, my apt-cacher-ng is broken [22:01] This is the new build log, after running autoreconf -i http://pastebin.com/f462d33dd -- still pretty broken [22:01] well, it will build, what's the next step ? [22:04] What's actually causing the build to fail? [22:04] joaopinto: upload it (or paste the debdiff somewhere, then I'll upload it) [22:06] where can I upload it to ? [22:06] * fabrice_sp is tired. Will continue tomorrow morning [22:06] bye [22:07] joaopinto: if you're a motu straight to the archive [22:07] ok, I am not, next option :P [22:07] joaopinto: otherwise either paste the debdiff somewhere or attach it to a bug ;) [22:07] ok, let me check the initial page, is there an LP bug listed there ? [22:08] uff, debdiff, I need to check how to generate that [22:09] ok, that one is easy [22:09] hum, something is wrong, the patch is not shown on the debdiff, only the changelog change [22:10] brb [22:13] funkyHat: defsgen.py is listed twice to installe somewhere (question is where though) [22:14] Oh, it was just that line that was the problem, for some reason I thought that didn't look important enough [22:14] it's twice in codegen/Makefile.{am,im} - which one of those generates the other? [22:14] *in [22:15] Ah, it tells me :) [22:15] funkyHat: Makefile.am generates .in if autotools are run. Usually Makefile.in is shipped by upstream and used as in the builds (which then generates the Makefile by the configure run) [22:15] funkyHat: so if there's no build-dependency on automake you'll need to patch both [22:17] So I should correct Makefile.am and then re-run autoreconf -i, and then update my patch in quilt [22:19] funkyHat: probably simpler is to directly patch Makefile.am/Makefile.in, but whatever you prefer [22:23] True :) [22:23] Was good practise using quilt though ;) [22:30] aha! It built [22:31] Now, is there anything obvious I can do to test it before I submit my debdiff? [22:31] (pygobject) [22:33] funkyHat: also submit it to debian or upstream ;) [22:33] funkyHat: but others than that, no [22:36] Should I just submit the quilt patch I created? [22:36] Don't know if they'll like it, as it's rather huge [22:37] funkyHat: ideally just the diff of Makefile.am (and eventually a Makefile.in) [22:37] funkyHat: the rest should be recreatable by running autotools, right? [22:38] Right :) [22:40] sistpoty, any ideas why the new patch would not show up on debdiff ? [22:42] joaopinto: not too sure actually... did you eventually overwrite the old source package (calling debuild prior to adding a changelog entry?) [22:42] hum, that could be [22:43] let me get the original diff [22:45] sistpoty, it was that, next time i need to add the changelog entry first to be safe [22:45] joaopinto: yes, that's what I'm usually doing as first step ;) [22:46] (even if I just write "fix things" in it to make sure to get a new version) [22:46] is there a bug associated with the FTBFS event ? [22:46] I mean at LP [22:47] joaopinto: not for universe [22:47] :( [22:47] where should I attach the debdiff to ? [22:48] joaopinto: unless someone filed one by hand of course [22:48] joaopinto: file a new one ;) [22:48] ok [22:48] Where can I find debian FTBFS? I don't see this bug reported at b.d.o [22:49] funkyHat: there isn't really a list. however installing a file twice is certainly a bug [22:49] erm, let me do a runtime test first [22:50] sistpoty: right, so I'll just file a bug for that the [22:50] *then [22:50] yes [22:52] sistpoty: ah, debian are on a more recent version of the package than we are... [22:52] I should have checked that, shouldn't I? [22:52] funkyHat: yes, always check that first ;) [22:53] wheee [22:53] Right [22:55] * sistpoty files a sync request for nufw [22:56] ok, as it's getting more and more quiet in here, let's move to #ubuntu-motu, shall we? [22:56] * funkyHat moves [22:57] btw.: all the questions you have asked here are very much ontopic for #ubuntu-motu as well :)