[01:11] tumbleweed: what tests did fail when you were packagng eventlet? I've just tried 0.9.14, and all pass (or are skipped) on both natty and sid pbuilders. [03:40] can anyone please explain me this error http://paste.ubuntu.com/566833/ [03:41] c2tarun : the problem is probably the line above that [03:41] micahg: http://paste.ubuntu.com/566834/ [03:43] micahg: here are the lines above that error ^^ [03:44] c2tarun: nope [03:44] micahg: you mean more lines? [03:44] c2tarun: yeah, until there's a line with error in it [03:45] That comes when the linker can't find the things to link against. Might be libraries not installed, or not referenced by the linker. Might be API changes. Might be bugs. [03:45] micahg: ya i found the line with error http://paste.ubuntu.com/566835/ [03:46] Most common cause is the order of arguments to the linker: you need to add anything linked against before the thing linking against. This is new behaviour with the recent toolchain, so it's not surprising if some older buildsystems fail. [03:47] persia: are you explaining my error? [03:47] c2tarun: you might want to move either -lssl or -lcrypto before the bacula libs [03:49] micahg: in rules file right? [03:50] c2tarun: depends on the package, usually in configure of some sort [03:51] micahg: actually I am not able to understand the error. Can you please explain a bit. [03:52] c2tarun: you might want to read this: https://wiki.ubuntu.com/NattyNarwhal/ToolchainTransition, looks like an --as-needed issue [03:52] micahg: ok, sure :) i'll read it. thanks [04:17] micahg: in the wiki page you gave me they used dpkg -S to search which package contains the particular library. I tried in manual pages but I think there is no information about -S can you please tell why we use -S [04:25] c2tarun: dpkg -S in man dpkg is right above OPTIONS (dpkg-query portion), `man dpkg-query` gives you the normal paragraph stuff for -S [04:28] I am getting an error while building binary package, the error is due to some libraries are not directly linked. But I dont know which libraries. How can i find? I tried "dpkg -S TLSv1_method" but got an error that method not found. [04:35] ping [04:52] c2tarun: If you are looking for a specific file, apt-file might be of help. [05:01] arand: here is the error I am getting http://paste.ubuntu.com/566845/ [05:03] arand: as we can see that the library libbac.so is trying to use method TLSv1_method. So I have to link the library containing TL* method. but I dont know which lib contains it and where do I put the name of that library in order to link it. [05:05] c2tarun: it's probably ssl or crypto [05:05] c2tarun: dpkg -S is when you're looking for a library that's not on the linker line, in this case, I believe you have the library, just not in the correct order [05:06] micahg: ok, so from where can i access the order ? I mean which file is responsible for this? [05:23] c2tarun: usually configure [05:25] Although it can vary wildly based on the buildsystem: it's often good to examine the logs and see what command has the order, and then backtrack through the variable substitutions. [05:33] persia: where can I find the log files, you just mentioned? ^^ [05:34] Your build logs? Depends how you build it. There's usually a .build file that is the log or a symlink to the log. [05:34] persia: I build it using pbuilder-dist natty build *.dsc. [05:34] last_operation.log in pbuider [05:34] err, pbuilder-dist [05:36] micahg: last_operation.log just tells me that error occured while compiling btape.c but didn't tell from where the command is issued, I mean from which file [05:37] whereas source.build file is kind of a log for creating source package, but not binary package. [05:39] c2tarun, That file will contain some command that had an error. This command will have been generated by whatever build system the source uses. [05:40] In most cases, the actual command is the result of variable substitution, so you have to backtrack through the various build files from the one that generated the command that had the error to the one that controls how that appears. [05:40] At some point, you'll get back to the files that are available in the source package, and you can modify the appropriate file. [05:42] persia: here is the beginning of error http://paste.ubuntu.com/566849/ I am not getting how to backtrack. [05:43] Look through the buildsystem files to find out what is calling libtool. [05:43] Then look through the buildsystem files to find out what is controlling how libtool is called. [05:43] I am sorry but what do you mean by buildsystem files? [05:43] Then continue until you get to the place you need to change. [05:44] makefiles, cmakefiles, setup.py files, Makefile.PL files, perhaps shell scripts, autotools files, whatever. It entirely depends on how your source generates binaries. [05:44] persia: ok === almaisan-away is now known as al-maisan [07:27] good morning [07:31] dholbach: good morning :) [07:32] hi c2tarun [07:32] dholbach: hi :) [07:57] should we be building with -Werror? [08:00] that will cause issues with all upstream packages that don't take their warnings seriously [08:00] well, I'm looking at blobwars, and upstream doesn't see to like to deal with the return value of fread [08:01] s/see/seem [08:03] tumbleweed: would it be better for me to just add code to catch the return value of frad? [08:03] *fread [08:06] tumbleweed: weirdly enough, it doesn't fail in Debian [08:07] a warning added in gcc 4.5? /me is still pulling source... [08:11] tumbleweed: fails for me on maverick as well (should be gcc-4.4) [08:11] tumbleweed: I can pastebin the error if you like [08:13] micahg: naah, I'll just give up on this and play on a box in EU instead [08:14] tumbleweed: there are only 4 instances where the return value isn't processed [08:17] micahg: yeah it does look buggy [08:17] tumbleweed: so, what should I do? [08:17] patch it and submit a patch upstream, I guess [08:19] tumbleweed: does this look ok, or should I actually process the return value? http://paste.ubuntu.com/566871/ [08:19] micahg: catching a return value and not doing anything with it (i.e. hiding the bug) is worse than not catching it [08:21] tumbleweed: ok, there's one working example in the code, I"ll try to follow that [08:25] micahg: Many of the things that can go wrong with file handling are very unlikely to happen in normal circumstances, so just aborting can be a sane option. [08:29] tumbleweed: ok, I think I got it [08:33] Better to retry or something, rather than abort. A user typically can't tell the difference between an abort and the segfault that would likely otherwise occur from the failed operation. [08:33] tumbleweed: ok, here's the error now, I seem to be using the variable, but it's not happy http://pastebin.ubuntu.com/566877/ [08:34] Mind you, you only want to retry once or twice, or just provide some error message to the user, and go for manual retry, to avoid getting into a non-terminating loop. [08:34] err, I think I see the issue :-/ [08:35] this is one problem after another [08:35] * micahg wonders why Debian's -Werror doesn't catch this, nor upstreams [08:42] * micahg is glad Firefox isn't built like this [08:47] first fread, the fseek, now fgets [08:54] micahg: does Debian use -D_FORTIFY_SOURCE=2 by default? [08:56] geser: idk [08:57] AFAIK Debian doesn't use it and I'm not sure that's the reason for your error but most likely [08:57] geser: we seem to have some type of option warn_unused_result [08:58] Good to have that option. I think those are the cause of a majority of our segfaults. [08:58] * micahg will continue fixing this another night [08:58] (well, maybe not a majority: use-before-set is very popular as well) === al-maisan is now known as almaisan-away === warp10- is now known as warp10 [15:09] Where do I see build logs, I got informed that flightgear gives troubles in natty? [15:10] I might be blind but I don't find it on launchpad.net/ubuntu/+source/flightgear … [15:11] Rhonda: linked from the builds https://launchpad.net/ubuntu/+source/flightgear/1.9.1-1.1 [15:17] Rhonda: is it for the flightgear package currently in the archive or for a possible sync? I know that sebner was looking at building flightgear 2.0.0-3 and had issues with "ld --no-add-needed" [15:18] dholbach: around? [15:18] ari-tczew, yes, but a bit busy - how can I help? [15:19] dholbach: sponsors overview has been hanged [15:19] I guess I know the reason for it [15:20] dholbach: btw. could you request to Launchpad remove ~ubuntu-main-sponsors? it's not needed anymore. [15:20] can somebody else please do this? [15:20] I'm really quite busy [15:20] and regarding the sponsoring overview, I hope this will fix it: https://code.launchpad.net/~dholbach/ubuntu-sponsoring/lp-usage/+merge/49648 [15:21] tumbleweed: ^^ [15:22] dholbach: do you mean someone else request remove ~ubuntu-main-sponsors? [15:22] yes [15:22] you have my ok [15:22] dholbach: OK will do it. [15:22] thanks [15:23] Rhonda: if it was about flightgear 2.0.0-3 see http://pastebin.com/VHvQmEDk for the build failure (solutions seems to be to move "-lsgio" after "-lOpenThreads", at least it worked when I tried it manually inside my pbuilder) [15:24] geser: Yes, sebner did ping me to look at it. [15:27] geser: I'm looking into building the unpatched package and work through his patches after that to see wether it's related to that. :) [15:27] Actually natty currently is synced, not merged, so I wonder if a plain sync would work. [15:28] Rhonda: the pastebin is the error you'll get when you try to build it in natty [15:28] you should be able to get in Debian with binutils-gold too [15:28] dholbach: FYI https://answers.launchpad.net/launchpad/+question/145351 [15:29] thanks ari-tczew [15:29] geser: In his PPA he has a ubuntu15 version or such? [15:33] Rhonda: I tried the unmodified Debian package when he asked me for help about the FTBFS [15:33] Ah, alright. [15:33] Got anywhere? :) [15:34] Hmm, there is no bugreport in Debian about binutils-gold. Weren't they all tested and filed? [15:34] when I reordered the libs in the g++ (inside the pbuilder shell) and moved "-lsgio" after "-lOpenThreads" the linking worked [15:35] but I didn't look how to get this into the Makefile [15:53] hey! I was just looking at some universe packages that look like they haven't been rebuild for py2.7 - mind if I upload them or should I get in contact with someone who coordindates it? [16:00] mvo: haven't heard of anyone doing that, feel free to do it [16:00] maybe drop a mail to -motu if it's a lot [16:06] mvo: go for it, I think. barry was taking care of them last time I looked [16:25] 3clear [16:25] oopps [16:51] thanks Laney, tumbleweed [17:37] kklimonda: don't assign if you're want sponsorship [17:38] ari-tczew: I was asked to [17:38] erm, don't assign to myself [17:38] right [17:39] but I did unassign myself.. hmm [17:39] ok, it didn't work [17:40] ;) [18:33] YokoZar: What does "err:dosmem:DOSMEM_MapDosLayout Need full access to the first megabyte for DOS mode" mean I need to do? [18:47] ScottK: means you need to disable mmap_min_addr, I think. sudo dpkg-reconfigure wine1.2 should ask you about it. [18:48] kees: Thanks. [18:49] Gotta love documentation ... They are provided "as is". It is assumed that if you are interested in this, you can pretty well figure out what to do with it. [18:49] from a REAME.TXT [18:53] udienz: you subscribed ubuntu-sponsors to gkamus bug, did you got enough advocates? [18:56] I have an idea looking on udienz case on sponsors overview [18:56] ari-tczew, gkamus not advocates by anyone right now [18:56] when someone is looking for MOTU review @REVU, subscribe ubuntu-sponsors [18:57] geser, persia: what's the process of update policy? [18:58] (firstly discussing) [19:00] kees: That was exactly it (including dpkg-reconfigure wine1.2). Thanks again. [19:03] ScottK: cool; you're welcome. :) [19:03] (I even remembered to put it back when I was done) [19:12] ari-tczew: which policy? [19:13] geser: when someone is looking for MOTU review @REVU, subscribe ubuntu-sponsors [19:17] ari-tczew: I don't know if we have a process for it, but a first step would be to write a mail describing the change and rationale for it, and wait some time (1-2 weeks) for feedback. I guess mailing ubuntu-motu and ubuntu-devel would be good for asking for feedback. [19:18] geser: what do you think about this idea? [19:20] not sure myself yet, have to think about it and read your rationale, but my first questions are a) has ubuntu-sponsors the resources for it and b) is this the right path? === zul_ is now known as zul [19:21] I ask about b) because I'm not sure that REVU helps us in the long run or only attracts hit-and-run packages which MOTU has to maintain and update in the future (and MOTU lacks the resouces for it) [19:24] but this is good way to get new packages quickly [19:25] yes, but do those packagers stay to maintain them? [19:25] What are you trying to achieve? Patch pilots looking at new packages? [19:25] (that's just my opinion about REVU and shouldn't stop you) [19:27] Is natty currently badly broken? I am thinking about dist-upgrading now. [19:28] well, I think REVU is good for getting packages in shape, they we should throw them over to Debian once they're ready [19:28] but for that use case, the sponsoring queue is certainly not appropriate [19:29] what does it mean? Makefile.am:109: linker flags such as `-Wl,--no-as-needed' belong in `src_liblwmsg_la_LDFLAGS [19:29] Laney: not exactly patch pilots, just information to MOTUs about packages which needs review [19:30] ari-tczew: I'm guessing that was from some linker flags in LIBS / CFLAGS rather than LDFLAGS [19:30] that information is already available on REVU [19:30] if people wish to review new packages (it's a very differnet kind of review IMO) then they can look there [19:31] tumbleweed: I'm trying to fix ./common/.libs/liblwsmcommon.so: undefined reference to `lwmsg_handle_type_class' [19:31] and I guess -Wl,--no-as-needed would be helpful [19:33] ari-tczew: using -Wl,--no-as-needed should be IMHO only be a last effort [19:35] geser: I can't find any lib related to ftbfs message [19:37] which package is this? [19:37] likewise-open [19:37] if --no-as-needed works than the link order needs to get fixed [19:39] geser: as I wrote above: Makefile.am:109: linker flags such as `-Wl,--no-as-needed' belong in `src_liblwmsg_la_LDFLAGS [19:39] so it doesn't work [19:40] ari-tczew: that's not that it doesn't help you, it's that you were using it wrong :) [19:40] but yes as geser said, --no-as-needed isn't the preferred solution [19:52] when is feature freeze? [19:54] Feb 24th [19:54] ty [19:55] nhandler: soren: could one of you accept my email to ubuntu-motu? It got sent from a wrong account. [20:18] kklimonda: Done. [20:18] ScottK: kees is right, and you can blame him for that particular error ;) [20:18] kklimonda: I also approved one to ubuntu-server@ [20:19] Although I'm unsure of its future in wine1.3 as there is dosbox integration going on (albeit not a complete sort yet) [20:19] YokoZar: it might be nice to update that error to hint to the dpkg-reconfigure command [20:19] hmm, I suppose [20:23] Good idea. [20:24] soren: thanks === zul_ is now known as zul [21:13] if i edit a file outside of quilt, is there any way i can get quilt to catch those changes and add it to my topmost patch? [21:14] for those that don't dislike quilt ;) [21:55] bcurtiswx: "quilt add $file" before you start editing and later "quilt refresh" [21:55] geser, i meant after i edited the file.. did a bzr diff and added that to a patch. Thx though :) === Guest44544 is now known as mikeyeates [22:03] Hi all, I am having real trouble trying to build a deb with a bunch of php files in it. Are there any examples of building a deb without any compiling? I am just trying to copy the files in the dist tarball to /usr/lib/mypackage [22:03] mikeyeates: zend-framework [22:11] Thanks micahg Ill have a look at it [22:12] mikeyeates: you should copy them to the usual staging directory (debian/tmp or debian/mypackage), not /usr/lib/mypackage directly [22:12] in case you tried that [23:23] I tried copying the rules and Makefile files from zend-framework but I keep getting permission errors (the same type as when I was trying from scratch) [23:23] /usr/bin/make install-framework DESTDIR=/home/mike/speedy/releases/cogs-framework-0.9.1/debian/cogs-framework [23:23] make[1]: Entering directory `/home/mike/speedy/releases/cogs-framework-0.9.1' [23:23] install -d -o www-data -m 755 /home/mike/speedy/releases/cogs-framework-0.9.1/debian/cogs-framework/usr/share/php/cogs-framework/ [23:23] install: cannot change owner and permissions of `/home/mike/speedy/releases/cogs-framework-0.9.1/debian/cogs-framework/usr/share/php/cogs-framework/': Permission denied [23:23] make[1]: *** [install-framework] Error 1 [23:23] make[1]: Leaving directory `/home/mike/speedy/releases/cogs-framework-0.9.1' [23:23] make: *** [install] Error 2 [23:23] dpkg-buildpackage: error: fakeroot debian/rules binary gave error exit status 2 [23:23] This must be something simple, does anyone know what I am doing wrong? [23:24] mikeyeates: Who owns /home/mike/speedy/releases/cogs-framework-0.9.1/debian/cogs-framework/usr/share/php/cogs-framework/ ? [23:25] mike [23:25] all operations are done under my user [23:26] it is chmod 700 though, but I didnt make that directory [23:31] do you have a www-data user? [23:33] yep [23:33] huh [23:33] I thought it was all done under the fakeroot program which should mean it works? [23:34] If you use fakeroot debian/rules or debuild, anyway [23:34] I am using debuild [23:37] even sudo debian/rules binary fails in the same way [23:39] sudo install -d -o www-data -m 755 /home/mike/speedy/releases/cogs-framework-0.9.1/debian/cogs-framework/usr/share/php/cogs-framework/ [23:39] even that does not work [23:41] Try sudo strace -f install …etc [23:43] This looks like the relavent part [23:43] mkdir("cogs-framework/", 0700) = -1 EEXIST (File exists) [23:43] open("cogs-framework/", O_RDONLY|O_NOCTTY|O_NONBLOCK|O_DIRECTORY) = 3 [23:43] fstat(3, {st_mode=S_IFDIR|0700, st_size=4096, ...}) = 0 [23:43] fchown(3, 33, 4294967295) = -1 EACCES (Permission denied) [23:47] EACCESS and not EPERM, strange [23:47] anyway, clean up the whole mess if you can [23:48] I just had a thought... I am doing this in a virtual machine with the source directory ssh mounted from the desktop machine, maybe that is borking it? [23:48] ill try from the desktop [23:48] oh yeah, fuse would be denying permission to root [23:49] to prevent normal users from showing root strange stuff, like setuid files and devices [23:49] thanks for your help.. [23:49] :)