[00:04] good evening! [03:29] So I have a package I'm trying to build; and it's on a Bazaar branch. [03:29] JackyAlcine: so build it :) [03:30] My partner told me to use bzr-builddeb to build the package, but it keeps failing at dh_builddeb [03:30] JackyAlcine: pristine-tar to recover the orig tarball, and it should be happy [03:30] JackyAlcine: what's the error [03:30] something about signing. [03:30] JackyAlcine: pastebin, plox :) [03:31] JackyAlcine: you might have to use -k [03:31] http://pastebin.com/THzmhF6G [03:31] if the name on your key is not byte-for-byte identical with the changelog name [03:31] JackyAlcine: yeah, use that -k flag, or dch touch the changelog [03:32] JackyAlcine: dch touching would be much cleaner. downgrade it if you have to :) [03:32] (0.01-1~bzr1~jacky1) or something [03:32] if it actually matters, which I doubt :) [03:33] not saying your package does not matter, just saying it's not going to fail out people's boxes without them knowing [03:34] Alrighty. [03:34] JackyAlcine: hope you're well :) [03:35] I've been pulling my short hairs on my head trying to avoid BNF-style NLP for weeks now. lol [03:36] JackyAlcine: I told you, just use something written already [03:36] :) [03:36] I kinda am, I'm using LinkGrammar from the OpenCog proect/Abiword; but I'm trying to build my own just for experience purposes. [03:38] I figured that it'd save me time when I'd go for a PhD in Linguistics, if I ever plan on it. lol [03:39] JackyAlcine: you're still in high school, aye? [03:39] Yup, leaving in June (finally) lol [03:39] JackyAlcine: talk to me when you graduate college, then tell me it's a good idea :) [03:39] Lol alrighty then [04:24] Yeah, this is getting silly; I really can't build this package. [04:24] Even if I've 'dch --team' . [04:27] http://pastebin.com/VDBpCMrr if anyone's interested. [04:28] so it built, but wasn't signed [04:29] O.o [04:29] do you have a gpg key with that name & email address? [04:30] I do, but I changed my name to include the acute mark over the e, that'd mess things up wouldn't it? [04:30] it would [04:30] you could use the suggestion from earlier - using the -k option to pass in the key id, rather than matching on name [04:31] something like 'bzr-buildpackage -- -kKEYID' [04:35] * JackyAlcine shoots hands to the sky! [04:36] There is a God (and he's not Stallman!) [04:36] it's working then? [04:39] yup :D [04:40] Freaking love you guys, lol :D [04:40] great [04:41] Going to start from scratch to test it out and write this down so I'd remember. [04:41] It's that "-- -kKEYID" thing I need; how come that worked, but a typical "-" didnt? [04:42] because it's being passed as an option through to debuild rather than being an option to bzr-buildpackage itself [04:42] * JackyAlcine makes a note. [04:42] Oh, I think I can see that in the output as well. [04:42] JackyAlcine: the stuff after -- is passed to debuild [04:42] Thanks again ajmitch and paultag [04:43] no problem [04:53] debuild or dpkg-buildpackage? [04:53] * ScottK would have guessed the latter. [04:53] debuild, I think [04:54] from what I recall, it's what calls debsign [04:54] dpkg-buildpackage does that as well. [04:54] What dpkg-buildpackage doesn't do is call lintian if installed. [04:56] ScottK, it does in my instance. [04:56] right, it looks like the 'Now signing changes and any dsc files" is from debuild, before it calls debsign. it doesn't really matter in the end which one is getting called [04:58] No. It doesn't matter. [04:58] Just a point of curiosity. [04:58] just another bit of duplication [08:00] good morning [08:06] tumbleweed: Ahoy! [08:12] tumbleweed: want to start setting up challenges and appropriate classroom sessions? :) [08:12] tumbleweed: Better than we start early in the cycle than later === yofel_ is now known as yofel [10:53] Hmm, I think I should get an approval for a SRU for the tworld 64 bit arch patch … === akher0n is now known as akheron === Quintasan_ is now known as Quintasan === jtechidna is now known as JontheEchidna === moonflux_ is now known as moonflux === mdomsch__ is now known as mdomsch [16:27] Ok, I'm looking for suggestions to set up a classroom session and an associated challenges, maybe have a buglist of things that need fixing. Does anyone have a suggestion or want to join me in setting it up? [16:56] nigelb: what's the target group? and how many experience do they have? [16:56] geser: The target group is attracting new developers. So the experience would be low to none. [16:57] hmm [16:57] try looking at some FTBFS [16:57] that early in the development cycle, there are still several easy ones [16:58] The plan is to have a classroom session and list a few bugs so that people can tackle of that type and offer a few folks to ask for help [16:59] this challenge thing has very much changed from what I thought the original idea was [17:01] nigelb: of course someone would have to look at the FTBFS and list the easy one suitable for your class [17:01] Laney: wait, what was the original idea? I might have misunderstood. [17:02] geser: Yeah, and I'll need to find a few people readily available to help in the channel. [17:02] when is that session? [17:02] my idea was that developers post up lists of 'challenges' or tasks that they have going on (e.g. a haskell transition) and then people can come along and work with that person to get it done [17:03] Laney: That's there too :) [17:03] one of the main points was the individual connection that develops [17:03] geser: Its open, there are no dates committed. I can't do this alone. I'm just helping do the logistics bit of it. I'm trying to find folks to help with the packaging bit. [17:04] Laney: What I'm trying to do was to get the classroom session to promote specific tasks. Like, I'd really like you to talk about the Haskell transition and if you have the time help people through fixing the packages [17:05] I think it might be too transient [17:19] Laney: transient? [17:19] you have to be at the session otherwise you miss the boat [17:19] and symetrically the teacher has to figure out how to format their challenge as a lesson [17:20] I don't know if it meshes [17:20] Laney: I understand, I'm trying to figure out how to reformat so we can have a better experience for a new contributor. [17:20] I'm open to doing something else. [17:22] If you want to do 'bitesize' 'ftbfs' 'rcbugs' -style ones then it might be best to write a short introduction on some website (what you'd do in the classroom session) and then have this channel as a place to come for help [17:22] 'for the next two weeks we will try to get this list of 50 bugs down to 0' [17:22] any idea why my debuild -S -sa is faling http://paste.ubuntu.com/612344/ [17:23] Laney: okay, I'm game for that. So now, I just need to find folks to help write that initial article and post it on the planet [17:23] :-) [17:23] Technoviking: you need to run update-maintainer [17:24] but... what are you changing in smuxi? [17:24] updated a file to fix a problem using twitter [17:25] https://bugs.launchpad.net/ubuntu/+source/smuxi/+bug/787645 [17:25] Ubuntu bug 787645 in smuxi (Ubuntu) "Error connecting to Twitter in smuxi" [Undecided,New] [17:25] found a fix upstream [17:28] I happen to be part of the maintainer team for that package in Debian (and so is the upstream maintainer), so if you plan on asking for this fix to be included then it's best to talk to us first :-) [17:29] Laney: what should I do? [17:30] speak to meebey in #smuxi [17:37] Laney: Clearly, its been a while since I worked on packaging. I can't even find the docs for FTBFS [17:37] :( [17:37] I don't know if there is a single reference for that [17:38] FTBFS is so incredibly broad [17:39] I don't see how you could possibly write a single reference for it [17:39] FTBFS is to build systems as bugs are to programs [17:40] admittedly, there are patterns, but they vary over time [17:43] Oh, right. I generally pop in on IRC and ask for help. Now I remember. [17:43] cjwatson: I think the idea was to pre-select bugs that should be relatively easy to fix [17:43] so you could write some documentation to target those [17:45] Laney: What are the difficulty levels for the haskell transition for a newcomer? [17:46] the one that's on now is probably not suitable [17:46] in general though they are pretty easy [17:46] no-change rebuilds [17:46] I'd be concerned that you don't learn anything much from no-change rebuilds [17:47] that is true [17:47] well, beyond the first few I suppose [17:47] get dev environment set up, figure out which ones do do, build, check deps are correct, upload [17:48] it's admittely not very difficult or engaging [17:48] yeah, but you'd learn all of that and more from something that's actually a challenge [17:49] * cjwatson <- not a big fan of giving people purely mechanical tasks - better if they're interesting [17:49] I'm just going to plunge in and pick a failure and see if I can fix it. [17:50] s/it/something. [17:57] ok, so I picked https://launchpadlibrarian.net/71420729/buildlog_ubuntu-oneiric-i386.telepathy-python_0.15.19-2.1_FAILEDTOBUILD.txt.gz because I'm fairly familiar with python. [17:57] nigelb: you might want to look at https://launchpadlibrarian.net/70903387/buildlog_ubuntu-oneiric-i386.redis_2%3A2.2.5-1_FAILEDTOBUILD.txt.gz [17:57] nigelb: that one should be fixed, it was a bug in cdbs [17:58] geser: The one I pasted earlier? [17:58] nigelb: the telepathy-python one [17:58] ah :) [17:58] ok, looking at redis [18:00] nigelb: if you want you can ask for a give-back of telepathy-python in #ubuntu-devel (it builds in my oneiric pbuilder) [18:00] geser: yeah, I was about to ask if can do that :) [18:00] can do it [18:00] *I can [18:00] :) [18:06] geser: Is what I'm meant to fix the 'undefined reference to log' ? [18:06] where is a good howto for quilt? [18:08] nigelb: that's the error (I can also give you a hint: the change to "ld --as-needed" caused it; if you need more help let me know) [18:08] geser: okay :) [18:08] also 'man log' [18:09] and I think it's only fair to point to http://wiki.debian.org/ToolChain/DSOLinking, since that went out in a -devel-announce mail IIRC [18:10] there was also https://wiki.ubuntu.com/NattyNarwhal/ToolchainTransition [18:10] ah, the linking isn't right. [18:11] urgh, one of the examples in https://wiki.ubuntu.com/NattyNarwhal/ToolchainTransition is explicitly backwards [18:11] ok, I believe I need to look into the make file [18:11] Technoviking: http://pkg-perl.alioth.debian.org/howto/quilt.html [18:12] iulian: thanks [18:12] cjwatson: sorry, I won't recommend it anymore [18:12] no, it's fine to recommend it, I'll fix it [18:13] * micahg has been meaning to review that page for a while [18:14] https://wiki.ubuntu.com/NattyNarwhal/ToolchainTransition?action=diff&rev2=17&rev1=16 [18:15] should be better now [18:16] cjwatson: I thought that you have to specify a library before you use it [18:18] cjwatson: should your 2nd change in that diff really be a question? (ending in '?') [18:19] ah, I see, the first example shows adding the missing library, the second the proper ordering [18:21] micahg: there were two ld changes during natty development: one was "ld --add-needed" (missing libraries) and the second one was "ld --as-needed" (proper ordering) [18:22] geser: right, I thought the example was for the second when it was for the first [18:26] geser: it was a mistake, the next diff fixes it [18:26] micahg: precisely the opposite. You *must* specify a library *after* any other objects/libraries that use it. [18:27] The examples are confusingly ordered; I didn't have time to fix that [18:27] cjwatson: so library resolution is right to left? [18:28] I think of it differently such that it is left to right [18:28] imagine the linker walking from left to right along the link lilne [18:28] *line [18:28] at each step, it uses the object/library to resolve any previously unresolved symbols, and gathers a list of symbols that are still unresolved [18:29] if a library doesn't resolve any symbols, it discards it [18:29] (this is option-dependent, but I'm describing behaviour in the strictest case) [18:29] ELF linking is more complicated than that, but I find that model sufficient to work with when fixing this kind of bug [18:30] oh, that makes sense if it discards when it doesn't do anything, I didn't know that [18:30] that's what --as-needed changes [18:30] "--as-needed causes a DT_NEEDED tag to only be emitted for a library that satisfies an undefined symbol reference from a regular object file or, if the library is not found in the DT_NEEDED lists of other libraries linked up to that point, an undefined symbol reference from another dynamic library." [18:31] I thought --as-needed meant you have to specify a library before it's used (I guess I didn't read that description well enough) [18:31] cjwatson: thanks for clarifying that for me [18:31] the previous version of the wiki page did rather reinforce that idea ... [18:31] np [18:32] this is actually traditional Unix linking - I remember dealing with this when writing cross-Unix-platform build systems ten years ago [18:32] (aside from AIX, which is basically entirely backwards) [18:33] so anything that's been cross-platform for ages will get it right, but stuff that's got used to the GNU linker's historical permissiveness often gets it wrong [18:33] much like bashisms, really [19:15] geser: I got the hint that I need to fix the make file ordering, but I can't figure out where to start looking in the make file. [19:18] I *think* I need to pass -lm for building vm.c, I' [19:18] Is that the right direction? [19:22] nigelb: if you look closely, -lm is already there [19:22] nigelb: (-lm is only needed on the "link line", i.e. the one that combines a bunch of .o files into an executable; it's not needed when compiling vm.c into vm.o) [19:23] cjwatson: oh. [19:23] nigelb: the discussion above in this channel should help explain, though [19:23] * nigelb looks [19:25] bdrung, tumbleweed: it looks like you guys should do a devscripts backport in the u-d-t daily ppa [19:25] a bunch of tools broke on natty, at least, when you switched to the devscripts.logger module [19:30] cjwatson: heh, the discussion above just horribly confused me on right to left and left to right : [19:30] :) [19:37] cjwatson: I'm guessing the linking of -lm and the other libraries should move further to the right and the .o files should go before them? [19:41] jkjkjkjkjk~. [19:47] nigelb: yes [19:47] Does this look right? http://pastebin.ca/2068596 [19:47] I couldn't figure out a more easier way to get the libraries to move right. [19:51] nigelb: you got the right idea, you might need to move $(CCLINK) to the end of the call in line 23 and 30 (after those .o and .a files in that command as they might need symbols from the linked libs too) [19:51] but you will notice it when you try to build the package [19:51] okay :) [20:08] broder: yes. i will do that once we devscripts is uploaded [20:09] broder: as you probably see, we are moving a bunch of scripts from u-d-t to devscripts [20:10] broder: you can build devscripts from git head for testing [20:11] I'm using pbuilder-dist on lucid to build for oneiric, how do I fix that? [20:12] fix what? [20:12] oh, it doesn't recognize oneiric. [20:12] Warning: Unknown distribution «oneiric». Do you want to continue [y/N]? n [20:12] * nigelb checks spelling [20:13] lrwxrwxrwx 1 root root 5 2011-05-24 19:32 /usr/share/debootstrap/scripts/oneiric -> gutsy [20:13] nigelb: dch? [20:14] i think its the same as this bug but s/natty/oneiric/ https://bugs.launchpad.net/ubuntu/+source/debootstrap/+bug/673038 [20:14] Ubuntu bug 673038 in debootstrap (Ubuntu Lucid) "SRU: Add (Ubuntu) natty as a symlink to gutsy." [Undecided,Fix released] [20:16] maco: so just do a symblink or get the latest debootstrap? [20:16] nigelb: symlink is a fine workaround [20:16] thats all we'd do to it for the SRU anyway === yofel_ is now known as yofel === stalcup is now known as vorian [21:23] geser: That package built successfully :) [21:23] MP if you would be interested in sponsoring it - https://code.launchpad.net/~nigelbabu/ubuntu/oneiric/redis/fix-ftbfs/+merge/62204 [21:25] * nigelb heads to bed === ximion1 is now known as ximion === ximion is now known as ximion1 === ximion1 is now known as ximion [22:57] tumbleweed: poke for "[stefanor] Ask derivatives front desk for distro-info naming help" - this blocks the python-distro-info package creation