[00:00] i want to keep `libbox0.so.2 libbox0-2 #MINVER#` like libpipeline1. [00:01] do i have to keep like `libbox0.so.2.0.0 libbox0-2 #MINVER#` and update everytime the SO VERSION? [00:03] kuldeep: does your binary package create a libbox0.so.2? [00:04] kuldeep: becuase libpipeline1 contains both: [00:04] /usr/lib/x86_64-linux-gnu/libpipeline.so.1 [00:04] /usr/lib/x86_64-linux-gnu/libpipeline.so.1.4.1 [00:04] which one a symlink to the other [00:05] afaik: file libbox0.so.2.0.0 and symbolic link libbox0.so -> libbox0.so.2.0.0 [00:05] libbox0.so is wrong, isn't it? [00:05] an unversioned lib? [00:06] nacc, https://paste.debian.net/903842/ [00:06] libbox0.so is a symbolic link to libbox0.so.2.0.0. is that a problem? [00:06] kuldeep: right so your symbols file and the contents of the package disagree [00:06] kuldeep: you told the symbols file to look at a library called libbox0.so.2 [00:06] but you don't ship any such library [00:07] so the build system recognizes that and switches to the library it did find [00:07] afaict that's what the log you got is saying [00:07] nacc, will it be happy (with the symbolic link?) "`libbox0.so libbox0-2 #MINVER#`" ? [00:08] [checking my trick] [00:08] libbox0.so belongs exclusively in libbox0-dev [00:09] libbox0-2 should ship libbox0.so.2 as a symlink to libbox0.so.2.0.0 [00:09] kuldeep: I don't know, maybe? but you have to refer to an actually shipped library either way; that seems like it should work, but it seems odd that you're not shipping libbox0.so.2 [00:09] you must ship the thing that's named in your SONAME or stuff won't work [00:09] kuldeep: so then, no, you should't do your trick, as it's not in the same package [00:09] ok [00:10] it is not correct to use libbox0.so in .symbols - that field is supposed to be the SONAME [00:10] cmake isnt generating the libbox0.so.2 and libbox0.2.0 have to look into that. [00:10] right, fix that. [00:11] cjwatson: right, duh, thanks for clearing that up :) [00:11] and re symbols containing 0.3.0+65, you're meant to edit that. [00:12] it'll only automatically fill in the minimum version if the symbols file didn't already have a version for that symbol. [00:12] kuldeep: right i think cjwatson's advice earlier was to use makeshlibs to give you a base symbols file that you then make correct in practice [00:12] it just uses the current version because that's the conservative choice and it can't know any better. [00:13] i think dh_makeshlibs even mentions that :) [00:22] So do Bad Things happen when I accidentally close a terminal when it's dputting to a PPA and it's on the source tarball? [00:23] cmake only allow me to set "SET_TARGET_PROPERTIES(box0 PROPERTIES SOVERSION ") but it only allow one value. [00:23] multiple call to SET_TARGET_PROPERTIES result in final only being generated. [00:24] so, either i can generated libbox0.so.2 or libbox0.so.2.0 or libbox0.2.0.0 + a symbolic link libbox0.so to the actual shared library [00:28] /usr/lib64/libpipeline.so -> libpipeline.so.1.4.1 | /usr/lib64/libpipeline.so.1 -> libpipeline.so.1.4.1 | /usr/lib64/libpipeline.so.1.4.1 [00:32] kuldeep: keep in mind that libpipeline.so lives ina different binary pacakge [00:41] nacc, it is mandatory to provide /usr/lib64/libpipeline.so -> libpipeline.so.1.4.1 ? === tsimonq2alt is now known as tsimonq2 [00:49] kuldeep: well, in the -dev package, probably? otherwise how would you compile against it? [00:49] *link* against it, rather [01:05] tsimonq2: no; either it finished anyway or the whole thing is discarded [01:05] kuldeep: /usr/lib64/ isn't a thing in Ubuntu [01:05] cjwatson: Thanks. [01:05] kuldeep: but yes, a libfoo.so symlink is required in the -dev package as nacc says, since that's what ld looks for [01:06] cjwatson, ok [01:06] I can't believe that cmake is actually as useless as you're suggesting though; I'd suggest you find a competent existing library package that's built using cmake [01:07] cjwatson, atleast it isnt doing the that by default. [01:08] finding libraries that uses cmake is hard. trying ... [01:08] im doing on https://anonscm.debian.org/cgit/users/ [01:10] grep-dctrl -FBuild-Depends,Build-Depends-Indep cmake -a -Pr ^lib -nsPackage /var/lib/apt/lists/gb.archive.ubuntu.com_ubuntu_dists_*_Sources | sort -u [01:10] (well, maybe not gb.archive in your case) [01:10] should find you a decent list to poke at [01:11] cjwatson: Speaking of that, is there a technical reason why we don't have an httpredir.debian.org for Ubuntu? [01:12] not afaik [01:13] Then who do I talk to about asking that Canonical set one up? :) [01:15] RT, I'd presume ... [01:16] cjwatson: Oh, so do you think it would be worth bringing up for discussion or should I Just File The Goshdarn RT? :) [01:16] I have absolutely no idea [01:16] Ok, fair enough \o/ [01:16] it's nearly end of year and I'm just trying to finish off my project so I can stop :P [01:17] cjwatson: Fair enough, cheers, good luck, happy holidays. :) [01:17] * tsimonq2 takes note to not poke Colin until January so he can finish his project [01:24] https://anonscm.debian.org/cgit/users/abe/nmu/plplot.git/tree/debian PLplot dont have any symbol file. (uses cmake) [01:38] cjwatson, https://cmake.org/cmake/help/v3.0/command/install.html it says "lib.so -> lib.so.1" documented as im seeing [01:38] I think you'd be best off not addressing me about cmake issues [01:38] I don't personally like it and don't use it, so can't help you further there [01:39] cjwatson, i pinged you because of "I can't believe that cmake is actually as useless as you're suggesting though; I'd suggest you find a competent existing library package that's built using cmake" [01:39] so, atleast i found in the documentation :) [01:39] but if you mean what the symlink target should be, then it doesn't matter whether it's to lib.so.1 or to lib.so.1.X.Y [01:39] it just has to exist at all [01:40] anyway, please stop pinging me about cmake-related things, thanks :) [01:40] * tsimonq2 coughs at #debian-mentors on OFTC and #ubuntu-motu on freenode as well... [01:40] yep, much better venues for all this [01:40] basically none of this is specific to launchpad.net at this point [01:40] cjwatson, ok. (maybe i should stop giving you reason to curse cmake :p ) [01:42] Oh jeez, looks like someone's doing a mass rebuild... :P [01:42] cjwatson, nacc and dobey for helping me out throughout the way of learning, building my first debian package. i have been a constant source of ping's for you :D [01:44] cjwatson, nacc and dobey THANK YOU for helping me out [01:44] kuldeep: Although I haven't helped much, don't give up here! Go get help in those channels I listed. I'm not an expert but you just have to keep at it and you'll become good. ;) [01:45] (I'm not good but I'm better than I was 3 months ago...) [01:45] :) [01:47] for the past two days, i have been slicing cjwatson libpipeline package and result in a ping when share any link. im have saturated cjwatson. sorry :) [01:49] thanks you to tsimonq2 and wgrant (and anyone i missed) too. [01:51] kuldeep: np, gl! [02:03] :) === chihchun_afk is now known as chihchun === chihchun is now known as chihchun_afk [10:57] Finally, i got my deb file without error. :D [10:57] (my fault not cmake because i didnt set the SOVERSION and VERSION properties correctly) [10:57] so, finally i got the libbox0_0.3.0.tar.xz libbox0-2_0.3.0_amd64.deb libbox0-dev_0.3.0_amd64.deb /var/cache/pbuilder/build/ [15:35] In July a maxima package bug was fixed that still hasn't shown up in the xenial updates. No one is assigned. What is a reasonable course of action? [15:36] #ubuntu-devel is where you want to ask about ubuntu development [15:38] The reason I asked here is this response to the launchpad bug tracking: "A thing the maintainer perhaps could do would be to create a launchpad mirror of the sourceforge repo maxima's code is kept in - and then to initiate a nightly build ppa for maxima." https://bugs.launchpad.net/ubuntu/+source/maxima/+bug/1602941/comments/8 [15:38] Ubuntu bug 1602941 in maxima (Ubuntu Xenial) "Draw cannot be loaded in 16.04 (Repackage needed since gcl has changed?)" [Undecided,Confirmed] [15:47] yes, but that doesn't get it into xenial updates [15:47] if someone wants to create a daily build PPA, they can. but it doesn't have anything to do with fixing bugs in ubuntu itself [15:51] yeah, that comment is from somebody uninformed but trying to be helpful, rather than somebody who knows about how stable updates happen === maclin1 is now known as maclin === JanC_ is now known as Guest35600 === JanC__ is now known as JanC === pbek_ is now known as pbek