2015-04-01 02:36:30 <_mjones> I thought there was an option to 'abuild' for it not to clean up after itself, but I can't find that. 2015-04-01 02:47:50 mjones - look @ /etc/abuild.conf 2015-04-01 02:48:57 change to CLEANUP="deps" (so the pkg & src remain) 2015-04-01 02:49:03 <_mjones> ah 2015-04-01 02:50:07 <_mjones> # what to clean up after a successful build CLEANUP="srcdir pkgdir deps" # what to cleanup after a failed build ERROR_CLEANUP="deps" 2015-04-01 02:50:13 <_mjones> Forgot I had seen that. tnx! 2015-04-01 02:51:46 np 2015-04-01 03:13:01 <_mjones> >>> openjdk7: Entering /home/mjones/aports/main/gcc >>> gcc: Checking sanity of /home/mjones/aports/main/gcc/APKBUILD... >>> gcc: Analyzing dependencies... >>> ERROR: gcc: Conflicting package(s) installed: libiconv-dev gettext-dev >>> ERROR: gcc: abuildindex failed >>> ERROR: openjdk7: all failed 2015-04-01 03:13:10 <_mjones> -_- 2015-04-01 04:12:17 morning 2015-04-01 04:13:28 I'm updating the Go package, and upon building, when abuild traces for deps, abuild gives and error 2015-04-01 04:13:30 >>> ERROR: go*: libc.so.6: path not found 2015-04-01 04:13:32 so:libc.musl-x86_64.so.1 2015-04-01 04:14:00 it actually finalize the package and everything is working fine 2015-04-01 04:14:10 should this be ignored or ? 2015-04-01 04:16:00 <_mjones> libc.so.6 is standard Linux-speak for glibc 2015-04-01 04:16:52 <_mjones> I would think that if you symlinked libc.so to libc.so.6 then things would work, but.. 2015-04-01 04:17:03 <_mjones> it's like there's a manual-linking step in there. 2015-04-01 04:17:48 <_mjones> updating 1.4 to something? 2015-04-01 04:18:29 <_mjones> I bet there's a test suite in the build. 2015-04-01 04:18:51 <_mjones> no, you said in the deps. 2015-04-01 04:29:19 _mjones: updating to 1.4.2 2015-04-01 04:30:35 binary works fine though so it doesn't look like a big issue 2015-04-01 04:30:40 just not clean 2015-04-01 04:31:26 <_mjones> error in build or every execute? 2015-04-01 04:47:50 error in dep tracing of abuild 2015-04-01 04:50:12 <_mjones> It sounds like it should be patched out of the build, but is harmless. 2015-04-01 05:57:57 _mjones: re gcc and libiconv: https://bugs.gentoo.org/show_bug.cgi?id=104237 2015-04-01 05:58:46 basically, we have not had libiconv support 2015-04-01 05:59:06 autotools (and gcc configure script) has been stupid 2015-04-01 05:59:39 and if it finds gettext/libiconv, it would use it, regardless what you give of configure options 2015-04-01 06:00:12 so if you happen to have gettext/iconv headers installed at gcc compile time, the built gcc will use it 2015-04-01 06:00:24 which means that you cannot get rid of it afterwards 2015-04-01 06:00:40 at least you could not back then 2015-04-01 06:01:11 so as saftety, i added a conflict of gettext-dev and libiconv-dev to make sure those are not installed when gcc is built 2015-04-01 06:01:24 <_mjones> @ncopa: I hadn't at all considered the possibility that this meant I needed to _remove_ libiconv and gettext even though it says 'conflicting'. 2015-04-01 06:02:30 i suppose we could do the check differently 2015-04-01 06:04:18 if apk info --installed gettext-dev; then error "building gcc with gettext-dev installed is normally not a good idea. aborting"; return 1; done 2015-04-01 06:04:23 <_mjones> thing is, I was building openjdk. Not sure how gcc got pulled in. gcc is already installed. 2015-04-01 06:04:50 gcc was tried to get built? 2015-04-01 06:04:57 how did you try build openjdk? 2015-04-01 06:05:04 abuild -R? 2015-04-01 06:06:15 coredumb: libc.so.6 error means that something was precompiled against glibc 2015-04-01 06:06:19 and not built from sources 2015-04-01 06:09:49 <_mjones> oh, -R! 2015-04-01 06:10:27 <_mjones> sigh. 2015-04-01 06:11:47 <_mjones> @ncopa: do you have the ceph APKBUILD we talked about the other day? I want to rebase on yours and see if I might finish it. 2015-04-01 06:12:49 as diff: http://sprunge.us/IdRT 2015-04-01 06:13:02 curl http://sprunge.us/IdRT | git apply 2015-04-01 06:13:16 will store it in wip/cehp 2015-04-01 06:14:46 <_mjones> @ncopa: danke 2015-04-01 06:18:17 ncopa: mmmh don't know if there's anything precompiled in the Go sources ... 2015-04-01 06:19:09 <_mjones> almost seems like a manual linking step in the build... 2015-04-01 06:20:52 how do i find the file ? 2015-04-01 06:21:21 coredumb scanelf -Rn pkg | grep libc.so.6 2015-04-01 06:22:36 ok 2015-04-01 06:27:09 k there's some precompiled debug stuff 2015-04-01 06:29:04 actually what's the common action 2015-04-01 06:29:16 i guess they hsould be removed from the package ? 2015-04-01 06:29:35 <_mjones> seems rather unusual. I'd look for some doco first. 2015-04-01 06:31:57 coredumb: may I look at your diff? 2015-04-01 06:33:34 ncopa: sure gimme a second 2015-04-01 06:33:53 rebuilding with removing those precompiled testdata dirs :) 2015-04-01 06:35:32 yep better no more error :) 2015-04-01 06:36:34 <_mjones> mantis bugtracker. considered and rejected? 2015-04-01 06:36:52 i kinda dont like mantis 2015-04-01 06:36:57 it's ugly, no idea if it was considered 2015-04-01 06:37:12 we looked at it back then 2015-04-01 06:38:51 ncopa: http://pastebin.com/c1Kn68gS 2015-04-01 06:39:20 ./make.bash -a || return 1 2015-04-01 06:39:23 what does -a do? 2015-04-01 06:40:41 --all 2015-04-01 06:40:46 won't stay 2015-04-01 06:40:56 was testing why i didn't have any builtins 2015-04-01 06:41:07 ok so it shouldnt be there 2015-04-01 06:41:13 looks good 2015-04-01 06:41:29 those copies of src dir 2015-04-01 06:41:30 yeah needs a bit of review/cleanup 2015-04-01 06:41:34 do we still need them 2015-04-01 06:41:44 the link to the upstream bug 2015-04-01 06:41:45 yeah else no builtin :/ 2015-04-01 06:42:17 upstream bug says its resolved 2015-04-01 06:42:29 took me some time to understand why go list all was only showing cmd/xxxx 2015-04-01 06:42:40 ok 2015-04-01 06:42:52 yeah have to check if this doesn't need to be put somewhere else 2015-04-01 06:43:06 also they don't seem to provide vim and bash completion anymore 2015-04-01 06:43:08 # When this is resolved we can split out the source to a 2015-04-01 06:43:08 # go-doc sub package. 2015-04-01 06:43:14 i noticed that 2015-04-01 06:43:18 i'm kinda ok with that 2015-04-01 06:43:20 yep 2015-04-01 06:43:43 wanted to make it work with the least modifications first and see after ^^ 2015-04-01 06:44:59 i compared the tree with the Go binary distribution, and as all builtins are in the src/ directory 2015-04-01 06:45:18 i'm not so sure we can just drop a go-doc sub package like that 2015-04-01 06:45:30 how do i check? 2015-04-01 06:47:03 'go list all' 2015-04-01 06:47:04 i suppose 2015-04-01 06:52:57 golang seems to be a pretty nice language 2015-04-01 06:55:12 coredumb: the elf stuff is test data 2015-04-01 06:55:17 i dont know if we need it 2015-04-01 06:55:36 but basically, it is valid files 2015-04-01 06:58:56 <_mjones> out of space? "1.4G ceph-0.87.1" 2015-04-01 07:09:20 >>> go*: Package size: 122.9 MB 2015-04-01 07:09:24 not fun... 2015-04-01 07:09:41 yeah, as expected 2015-04-01 07:10:01 on the other hand, you get standalone binaries, so they doesn't depend on go after build 2015-04-01 07:10:22 but they are linked statically? 2015-04-01 07:10:59 there are a bunch of .a files 2015-04-01 07:11:59 looks like rss is broken on b.a.o :) 2015-04-01 07:12:02 :( even 2015-04-01 07:12:37 (morning all) 2015-04-01 07:13:05 i wonder if we should split the modules: http://sprunge.us/ABgC 2015-04-01 07:13:10 go- 2015-04-01 07:13:24 but it will probably be painful to track dependencies 2015-04-01 07:13:39 ncopa: yes it's huge :( 2015-04-01 07:14:03 coredumb: what about this: http://sprunge.us/OZUb 2015-04-01 07:14:17 basically same thing as your 2015-04-01 07:14:27 i kept the libc.so.6 binaries 2015-04-01 07:14:35 they will not stop the builders 2015-04-01 07:14:58 ok if they can stay no problem 2015-04-01 07:15:10 and technically, we need teach abuild to ignore them 2015-04-01 07:15:21 i think i just push that for now 2015-04-01 07:15:31 we can think of splitting it later 2015-04-01 07:15:38 ok cool 2015-04-01 07:15:44 ncopa: yes, statically 2015-04-01 07:16:05 i actually needed to update to build a consul package :) 2015-04-01 07:16:13 ncopa: gcc provides gccgo for external linking, but they're falling behind upstream 2015-04-01 07:17:57 >>> syncthing*: Package size: 24.9 MB 2015-04-01 07:18:02 bloaty 2015-04-01 07:18:05 :) 2015-04-01 07:18:05 but ok 2015-04-01 07:18:15 you got to love go 2015-04-01 07:18:23 yeah it can get huge 2015-04-01 07:19:11 they should write an office suite in go :) 2015-04-01 07:19:16 ha 2015-04-01 07:19:20 1G files 2015-04-01 07:19:29 well, c++ can be pretty big too 2015-04-01 07:24:25 consul binary is "only" 13MB 2015-04-01 07:24:27 :) 2015-04-01 07:24:45 i start to like more and more single binary deployments ^^ 2015-04-01 07:28:21 Apologies, i hadn't been following the conversation on redmine. I guess broken rss is a known issue :) 2015-04-01 07:29:06 consul is also written in go? 2015-04-01 07:29:17 barthalion: yep 2015-04-01 07:29:25 awesome piece of software 2015-04-01 07:29:29 indeed 2015-04-01 07:29:31 do you want a bug report for good measure? 2015-04-01 07:29:41 i should send the package to @aports to day 2015-04-01 07:29:47 today* 2015-04-01 07:37:44 ScrumpyJack: I'll look at bugs.a.o atom feed right away 2015-04-01 07:37:47 no need to file a bug 2015-04-01 07:40:08 ActionView::Template::Error (No route matches {:action=>"index", :controller=>"welcome", :format=>"atom"}): 2015-04-01 07:41:42 i wonder if that is due to my re-route of home page? 2015-04-01 07:42:15 could be 2015-04-01 07:42:21 i didnt touch routes.rb 2015-04-01 07:43:28 looks like we have to rotate that log 2015-04-01 07:43:40 that too 2015-04-01 07:43:41 yes 2015-04-01 07:44:04 or just receate the container with apk 2015-04-01 07:44:10 and then do that 2015-04-01 07:45:00 yup 2015-04-01 07:45:09 resetting routes.rb fixed the atom 2015-04-01 07:46:13 probably easier to redirect from ngnix 2015-04-01 07:46:57 heh 2015-04-01 07:47:03 yeah 2015-04-01 07:47:25 the less you touch ruby, the better it works. 2015-04-01 07:47:32 :) 2015-04-01 07:47:35 yes 2015-04-01 07:47:42 clandmeter: do you look at it? 2015-04-01 07:47:52 at what? 2015-04-01 07:48:13 redirect home to http://bugs.alpinelinux.org/projects/alpine/issues 2015-04-01 07:48:21 ok 2015-04-01 07:48:36 i wonder if that doesnt break anything else 2015-04-01 07:48:42 but only one way to find out 2015-04-01 07:52:15 stateless: py-crypt issue with fortify-headers: http://paste.ubuntu.com/10716923/ 2015-04-01 07:52:24 i bet its the same #define inline issue 2015-04-01 07:57:25 ncopa: didnt you mention you increased workers? 2015-04-01 07:57:46 i think i only changed config in unicorn 2015-04-01 07:58:14 yes, thats where you specify them 2015-04-01 07:58:23 but i only see one 2015-04-01 07:58:28 i guess its fine 2015-04-01 08:13:01 is there already some kind of guide how to package node stuff on alpine? stuff that can be installed with npm, or which uses npm for installing the dependencies? 2015-04-01 08:15:29 the best way of packaging node stuff is not packaging it 2015-04-01 08:16:19 mosez: there are no guides for node 2015-04-01 08:16:21 barthalion: mh, but maybe somebody else wants to have packages for etherpad or ethercalc as well ;) 2015-04-01 08:16:54 i have no experience with node or npm so i have no idea what would be a sane way to package those 2015-04-01 08:17:27 maybe i will do it manually in the meantime for my docker container 2015-04-01 08:30:26 ncopa: \o/ 2015-04-01 08:33:35 afaik npm being a package manager that require a repository running on node with some npm packages, packaging node stuff in apk would depend on npm packages which need to be packaged with node himself... :D 2015-04-01 08:41:26 npm ist comparable with gem from ruby :) 2015-04-01 08:47:58 i would like to clean up the lua packages before v3.2 freeze 2015-04-01 09:27:51 saw posgtgresql-bdr packaged this morning 2015-04-01 09:27:55 very nice 2015-04-01 09:27:59 have to test this :) 2015-04-01 09:48:07 samba-4.2 wasnon trivial 2015-04-01 09:48:15 has some new nss-wrapper thingy 2015-04-01 11:18:44 ncopa: yeah perhaps same issue 2015-04-01 11:19:05 but it is a bit weird that it cannot find the original symbol 2015-04-01 11:19:26 which would happen if wchar.h was not included first for some reason 2015-04-01 11:29:30 seems like its not same issue 2015-04-01 11:29:39 i coudnt find any #define inline 2015-04-01 11:30:25 yeah it seems weird 2015-04-01 11:30:51 can you add #warning in original wchar.h? 2015-04-01 11:33:58 http://sprunge.us/CBBF 2015-04-01 11:34:05 #warning ORIGINAL wchar.h 2015-04-01 11:34:56 hm so it does get included 2015-04-01 11:34:58 cool 2015-04-01 11:35:07 mbsnrtowcs 2015-04-01 11:35:10 where comes that from? 2015-04-01 11:35:20 wchar.h 2015-04-01 11:35:36 aha 2015-04-01 11:35:43 it is within a feature test macro block 2015-04-01 11:35:46 #if defined(_POSIX_SOURCE) || defined(_POSIX_C_SOURCE) \ || defined(_XOPEN_SOURCE) || defined(_GNU_SOURCE) || defined(_BSD_SOURCE) 2015-04-01 11:35:53 in musl's wchar.h 2015-04-01 11:35:59 so setting _GNU_SOURCE is needed 2015-04-01 11:36:22 or should fortify headers also do the #if defined? 2015-04-01 11:36:30 it probably does 2015-04-01 11:36:32 let me check 2015-04-01 11:36:37 maybe I missed it 2015-04-01 11:36:41 I typically copy the same tests as musl does 2015-04-01 11:36:54 yeah fortify doesn't do this 2015-04-01 11:37:08 it should do it though 2015-04-01 11:37:18 I will give you a small patch for fortify 2015-04-01 11:37:19 one sec 2015-04-01 11:38:41 does our docker package works also as registry? 2015-04-01 11:38:50 or we need docker-registry for that ? 2015-04-01 11:41:00 there is no registry in upstream tarball 2015-04-01 11:41:05 it's different project IIRC 2015-04-01 11:41:14 ok 2015-04-01 11:41:27 you can run docker-registry in docker 2015-04-01 11:41:32 I think it's recommended way 2015-04-01 11:41:46 yes, you can pull samalba/docker-registry 2015-04-01 11:41:55 ncopa: http://sprunge.us/RiJf 2015-04-01 11:42:28 now you probably still need to add the correct feature-test macro 2015-04-01 11:42:34 to get the overrides 2015-04-01 11:42:57 (if _GNU_SOURCE had already been set this would have silently gone unnoticed) 2015-04-01 11:43:04 if it works I will commit and maybe release 0.4? 2015-04-01 11:43:30 I should also add checks for wcpcpy and wcpncpy at some point 2015-04-01 11:44:16 be back lunch :) 2015-04-01 11:54:46 sweet dup.pw/$githash works 2015-04-01 11:58:08 stateless: that fixed it 2015-04-01 11:58:19 i'm gonna push that 2015-04-01 11:58:20 thanks! 2015-04-01 12:18:10 what's the best place to put contrib dir in py-$program package? 2015-04-01 12:18:24 if py-$program has useful scripts 2015-04-01 12:19:45 py-$program does not have his own dir, but installs on /usr/lib/python2.7/site-packages/$package 2015-04-01 12:21:50 what about $program-contrib? 2015-04-01 12:22:11 On which path? 2015-04-01 12:22:12 you will still need a directory :) 2015-04-01 12:22:19 /usr/share/pkgname? 2015-04-01 12:22:30 /usr/bin? 2015-04-01 12:22:51 what's the cleanest in your opinion? 2015-04-01 12:23:07 if you want them in your path, then thats the way. 2015-04-01 12:23:25 i think /usr/share/$pkg/contrib would work 2015-04-01 12:24:36 I'm packaging docker-registry, but it installs on /usr/lib/python2.7/site-packages 2015-04-01 12:24:53 in the package there are several scripts that might be useful 2015-04-01 12:25:15 so i think i'm going add them as subpkg contrib on /usr/share/$package7contrib 2015-04-01 12:25:19 */contrib 2015-04-01 12:26:56 it should be in $PATH then 2015-04-01 12:28:02 but honestly, I don't see anything interesting here https://github.com/docker/docker-registry/tree/master/contrib 2015-04-01 12:28:25 so /usr/share/$pkg will be good 2015-04-01 12:29:04 ncopa: should I tag a new fortify release? 2015-04-01 12:29:37 ah 2015-04-01 12:29:40 u pushed the patch for 0.3 2015-04-01 12:29:42 ok 2015-04-01 12:30:01 well will wait then until we have more fixes for 0.4 and we can remove the patches when we bump the fortify-headers port 2015-04-01 12:37:43 stateless: would be nice to have a new release soonish 2015-04-01 12:37:46 within a week or so 2015-04-01 12:37:58 will start build servers for v3.2 release 2015-04-01 12:38:13 would be nice to freeze fortify sources by then 2015-04-01 12:39:51 ok :) 2015-04-01 12:40:00 I will wait a few days in case we want to throw more fixes into 0.4 2015-04-01 12:40:13 sounds good 2015-04-01 12:40:17 so I can tag it on Monday or Tuesday 2015-04-01 12:40:20 :) 2015-04-01 12:40:30 i'll try have the arm builder up by then 2015-04-01 12:40:36 cool :) 2015-04-01 12:40:36 arm is the slow one 2015-04-01 12:40:52 i have a couple new wandboards waiting to be set up 2015-04-01 13:01:56 barthalion, it's script directory, not contrib. I want to name the subpkg "contrib" 2015-04-01 13:02:49 but i'm open to other ideas about how to call it 2015-04-01 13:08:05 ah, ok, my fault 2015-04-01 13:08:22 then I think it should go to $PATH 2015-04-01 13:09:07 ok, so /usr/bin/*.py 2015-04-01 13:16:55 dup.pw? 2015-04-01 13:17:07 yup 2015-04-01 13:17:17 i bought that 2015-04-01 13:17:37 and did a simple redirect in nginx 2015-04-01 13:18:00 i bought it because it was so short 2015-04-01 13:18:10 thought it could be useful 2015-04-01 13:18:51 I see 2015-04-01 13:19:04 right, it takes an hash 2015-04-01 13:21:46 was also thinking of setting up a vpn for infra 2015-04-01 13:21:54 we have some sites with alpine servers 2015-04-01 13:22:09 would be cool to let them talk over vpn with private addr 2015-04-01 13:22:32 then the idea was to use dup.pw as internal domain name 2015-04-01 13:23:32 http://dup.pw/v3.1.3 2015-04-01 13:24:18 might be i filter it on hash in future 2015-04-01 13:47:24 ncopa, freeswitch-x86 fails to build 2015-04-01 13:47:34 how do you provide sources for a package which must contain a .git directory for sake of compilation while their versioned release tarballs don't include ? 2015-04-01 13:47:46 rnalrd: i am working on that 2015-04-01 13:47:52 good tnx 2015-04-01 13:48:27 coredumb: ? 2015-04-01 13:48:35 use the release tarball? 2015-04-01 13:49:06 or the release tarball does not build due to missing .git dir? 2015-04-01 13:49:08 but the compilation fails as it requires the source dir to be a git repo and the release tarbal oviously isn't 2015-04-01 13:49:18 exactly 2015-04-01 13:49:42 in prepare, mkdir .git ? 2015-04-01 13:49:48 or patch it 2015-04-01 13:49:56 i must try if it's sufficient :D 2015-04-01 13:51:45 rather patch it 2015-04-01 13:53:22 i wonder what to do with nss_wrapper and samba 4.2 2015-04-01 14:24:52 ncopa: i have one addition to the s6 APKBUILD https://github.com/stef/aports/commit/c515c69af27ff09b681c442f0083ad1d85b6a4e9 2015-04-01 14:25:44 is the skalib dependency really needed? 2015-04-01 14:26:01 shouldnt the abi version detection handle that? 2015-04-01 14:27:30 i dunno, i guess it detects the lib as a dep, but does it know the minimum version of the lib that's needed? 2015-04-01 14:28:36 i also have a draft for "Setting up S6 in Alpinelinux 2015-04-01 14:28:39 " 2015-04-01 14:28:42 but it's very preliminary 2015-04-01 14:30:12 http://www.ctrlc.hu/~stef/s6.txt 2015-04-01 14:32:09 nice 2015-04-01 14:32:49 error: testing/s6/APKBUILD: patch does not apply 2015-04-01 14:33:39 again? ;-) 2015-04-01 14:34:05 ohmg 2015-04-01 14:34:12 didnt we have some kind of competition about aports commits? 2015-04-01 14:34:28 i think stef just got on top of that list 2015-04-01 14:34:28 better not, i'd dominate 2015-04-01 14:34:58 next one, you'll be converting all openrc scripts to s6 :p 2015-04-01 14:35:31 rpu3uO8PEVZZ: maybe you could create a branch named for-upstream or similar 2015-04-01 14:35:41 i guess i i have to go back to ubuntu, i suck 2015-04-01 14:35:59 which you make sure is always rebased against origin/master 2015-04-01 14:36:55 ubuntu... you dont suck that much... might be arch linux is enough :) 2015-04-01 14:37:54 if library versioning is done properly we shouldnt need to have the versioned dependencies for depend=... 2015-04-01 14:38:01 we might need it for makedepends though 2015-04-01 14:38:18 aah. ok, i understand. 2015-04-01 14:39:09 but generally, we consider the repo as one entity, so as long as the repo is in a syncronized state, it should work 2015-04-01 14:39:58 that is also why we only bump pkgrel for rebuilding against lib with new ABI 2015-04-01 14:41:17 clandmeter: btw i already started on some s6 scripts, you saw my https://github.com/stef/s6-services :) 2015-04-01 14:41:28 rpu3uO8PEVZZ: care to rebase the patch for me? 2015-04-01 14:41:41 yes, i'll just delete the complete repo 2015-04-01 14:41:47 and clone yours 2015-04-01 14:41:51 and change that one file 2015-04-01 14:41:53 make a local copy first 2015-04-01 14:41:56 local backup 2015-04-01 14:41:59 sure 2015-04-01 14:42:14 i have already 5 of those previous borked repos 2015-04-01 14:43:15 is s6 intended to replace openrc? 2015-04-01 14:43:19 by default? 2015-04-01 14:43:22 s/borked/out-of-sync/ 2015-04-01 14:43:34 stateless: no current plans for that 2015-04-01 14:43:37 ok 2015-04-01 14:43:42 i was also thinking of runit 2015-04-01 14:44:19 I'd probably prefer either of those compared to openrc (never studied openrc too closely) 2015-04-01 14:44:24 but it doesn't matter much 2015-04-01 14:44:29 (for me at least) 2015-04-01 14:44:37 was thinking of some runit script wrapper app that sources /etc/conf.d/ 2015-04-01 14:44:52 i am not 100% happy with openrc either 2015-04-01 14:44:58 its not perfect 2015-04-01 14:45:02 but works okish for now 2015-04-01 14:45:52 i have s6 running on my laptop on alpine, for a few months now. 2015-04-01 14:45:57 s6 has some really nice ideas, but i find it a bit "weird" 2015-04-01 14:46:18 indeed a bit weird, but i like it 2015-04-01 14:46:28 how is it to remove openrc from alpine without causing further pain when upgrading? 2015-04-01 14:46:37 i have not removed openrc 2015-04-01 14:46:45 openrc still handles the hw init stuff 2015-04-01 14:46:47 might be painful 2015-04-01 14:46:52 ah right I see 2015-04-01 14:46:58 i haven't gotten myself to port that to execline 2015-04-01 14:47:23 so i start s6 from local.d/s6.start. and have all services disabled in openrc 2015-04-01 14:47:23 what's so weird in s6? just curious because I have no experience with it. 2015-04-01 14:47:38 the scripting is a bit weird for example, in execline 2015-04-01 14:47:47 some is very similar to runit 2015-04-01 14:47:51 and not weird at all 2015-04-01 14:48:00 its djb kind of weird 2015-04-01 14:48:01 http://skarnet.org/software/execline/ 2015-04-01 14:48:13 haha. ok. gotit 2015-04-01 14:48:28 its just different 2015-04-01 14:48:55 but i like the goals and many of the ideas 2015-04-01 14:48:58 i'm not yet sure if going execline makes a difference 2015-04-01 14:49:40 but i'm playing with adding this minijail based seccomp sandboxing to all the s6 services, we'll see how that turns out. 2015-04-01 14:50:14 well, I like simplicity of systemd service files. 2015-04-01 14:50:15 huh. less chatting, more rebasing i guess. sorry 2015-04-01 14:50:17 inb4 shitstorm 2015-04-01 14:50:27 i do want a daemon supervisor 2015-04-01 14:52:46 also kinda weird but i guess workable is this: http://skarnet.org/software/s6/notifywhenup.html 2015-04-01 14:52:58 gosh packaging Go app is major PITA 2015-04-01 14:57:18 so i do not bump pkgrel for s6, right? 2015-04-01 14:57:40 and i only take the versions in depends_dev 2015-04-01 14:57:48 and i update the checksums 2015-04-01 15:05:52 how big is the alpine linux mirror? 2015-04-01 15:06:04 thinking of setting up a box to mirror it 2015-04-01 15:06:25 stateless: ~160GB iirc 2015-04-01 15:06:27 ok 2015-04-01 15:07:21 you don't need to mirror it all 2015-04-01 15:07:26 ditch unsupported releases 2015-04-01 15:08:24 it's around 45GiB without isos and EOL'ed versions 2015-04-01 15:08:33 ok :) 2015-04-01 15:09:12 ask nangel for domain if you don't want to use your own, and ping me to add your mirror to alpine-mirrors package 2015-04-01 15:09:44 sure :) will do 2015-04-01 15:17:03 is abuild automatically stripping binaries or should i specify it in package() ? 2015-04-01 15:19:58 ncopa: will this apply? https://github.com/stef/aports/commit/9957fefc6156171bd50a746cd5e7bf59535cc553 2015-04-01 15:49:28 <_mjones> coredumb: not that you mention it, I think nothing is stripped. 2015-04-01 15:50:45 <_mjones> oh, we have zramctl installed already? I didn't even look. 2015-04-01 15:50:58 <_mjones> Some things are stripped, some now. Checking APKBUILDs. 2015-04-01 15:52:34 <_mjones> install -s strips 2015-04-01 15:56:40 <_mjones> so abuild does it by default, then. 2015-04-01 15:56:49 <_mjones> options="!strip" 2015-04-01 15:57:46 i would like to clean up the lua packages before v3.2 freeze 2015-04-01 15:57:57 ncopa: what were you referring to ^^^ ? 2015-04-01 15:58:01 maybe I can help? 2015-04-01 15:59:35 _mjones: ok 2015-04-01 18:52:41 couldn't find anything in the wiki, is there a way to tell apk to not override a config file in the APKBUILD ? 2015-04-01 18:53:31 <_mjones> which config file? 2015-04-01 18:54:34 _mjones: any 2015-04-01 18:55:23 <_mjones> I don't understand then. 2015-04-01 18:56:11 example 2015-04-01 18:56:42 you don't want your /etc/nginx/nginx.conf to be overwritten when doing an apk upgrade 2015-04-01 18:56:45 right ? 2015-04-01 18:56:57 how's thta handled by apk ? 2015-04-01 18:57:01 <_mjones> oh 2015-04-01 18:57:20 <_mjones> it makes a pkg.conf.apknew, at least under some circumstances 2015-04-01 18:57:28 <_mjones> not sure if it checks hash of default one or not. 2015-04-01 18:57:37 mmmh 2015-04-01 18:57:49 <_mjones> example, my mkinitfs.conf got a mkinitfs.conf.apknew 2015-04-01 18:59:02 yeah but there's nothing in the APKBUILD telling it to do so 2015-04-01 18:59:14 so how's that decided ? 2015-04-01 19:00:03 <_mjones> not sure. 2015-04-01 19:00:26 <_mjones> makes me think it's not hashing and detecting the default config, though. 2015-04-01 20:33:05 ncopa: go to bed 2015-04-01 20:33:21 ncopa: packages have no legs and can wait 2015-04-01 20:57:57 http://pastebin.com/WLrjsSgh shouldn't consul-dev be created ? 2015-04-01 21:04:50 no - in the sub pkg function you need to create it 2015-04-01 21:04:58 mkdir -p "$subpkgdir"/usr 2015-04-01 21:05:03 for example 2015-04-01 21:06:40 BitL0G1c: well i shouldn't have a -dev subpkg on this one ... 2015-04-01 21:07:13 oh ok i get it my bad 2015-04-01 21:07:37 shouldn't have set subpackages= 2015-04-01 21:08:10 my mistake too - yes for a dev pkg it ordinarily s/be done automatically - the above is only for custom sub pkg 2015-04-01 21:15:38 >>> ERROR: consul*: Packages must not put anything under /srv, /usr/local or /opt 2015-04-01 21:15:49 rofl i like /srv :P 2015-04-01 21:16:10 <_mjones> same here. It's for everything I serve. 2015-04-01 21:16:19 <_mjones> It's LSB and iirc, posix maybe. 2015-04-01 21:16:54 <_mjones> Those are the three places you put local or degregated-app stuff. 2015-04-01 21:21:53 is there any logic behind refusing /srv ? 2015-04-01 21:25:17 <_mjones> if you're refusing the other two, then the method to the madness is consistent. 2015-04-01 21:25:47 <_mjones> I use /srv for webroots, vm roots, container roots, and NFS roots 2015-04-01 21:25:52 I run lxc in /srv/ by changing the data path in its /etc/ 2015-04-01 21:26:09 <_mjones> Yes, much better than default. 2015-04-01 21:26:58 <_mjones> Right now my alpine VMs are in /srv/alpine-dev, /srv/alpine3, and /srv/cephpine. 2015-04-01 21:27:55 so have to put it in /var 2015-04-01 21:28:00 bah 2015-04-01 21:31:44 lxc.lxcpath = /srv/lxc (set in /etc/lxc/default.conf) 2015-04-01 21:32:55 for consul the data directory is configurable too https://www.digitalocean.com/community/tutorials/how-to-configure-consul-in-a-production-environment-on-ubuntu-14-04 2015-04-01 21:41:25 BitL0G1c: indeed it is 2015-04-01 21:41:50 but setting it in a sane place by default from the package is important 2015-04-01 21:43:04 set some sensible defaults in /etc/conf.d/consul ? 2015-04-01 21:43:47 then /var/consul 2015-04-01 21:43:51 ^^ 2015-04-02 06:41:08 barthalion: would you want to try my consul package ? 2015-04-02 06:52:30 morning 2015-04-02 06:52:56 coredumb: I wish I have time for that 2015-04-02 06:53:00 crazy week, sorry 2015-04-02 06:53:04 hello 2015-04-02 06:53:30 coredumb: what do you want to try? 2015-04-02 06:54:15 clandmeter: opinions from a consul user about the sanity of defaults in the package 2015-04-02 06:54:37 did you already add it to aports? 2015-04-02 06:54:40 https://www.mauras.ch/securing-consul.html < basically having these defaults in the package 2015-04-02 06:55:01 clandmeter: nope not yet i'm finalizing it right now, testing the service script 2015-04-02 07:08:49 bah...it should be moved from testing :/ 2015-04-02 12:09:12 what is main/apk ? 2015-04-02 12:09:17 :) 2015-04-02 12:11:13 whoops 2015-04-02 12:12:19 :) 2015-04-02 22:38:10 <_mjones> A user has requested mainline nginx in repos in addition to -stable. Other packages are pretty cutting-edge, it seems like nginx stable is a more conservative choice than alpine usually makes. 2015-04-02 22:39:43 <_mjones> knite: <_mjones> A user has requested mainline nginx in repos in addition to -stable. Other packages are pretty cutting-edge, it seems like nginx stable is a more conservative choice than alpine usually makes. 2015-04-02 22:51:38 mjones - I could put the nginx-naxsi package to be more current 2015-04-02 22:52:02 i was about to update it soon once naxsi gets passed it's RC 2015-04-02 23:38:35 seems nice lib http://www.and.org/ustr/ 2015-04-02 23:39:58 <_mjones> naxsi can't be bundled separately? 2015-04-02 23:47:22 naxsi is a web application firewall for nginx - it's a nginx module 2015-04-02 23:48:10 <_mjones> I know, but the way you said that implied nginx-naxsi would package nginx. 2015-04-02 23:48:19 i'll update nginx & naxsi tomorrow - with nginx as a newer version 2015-04-02 23:48:36 as i was about to update it anyway as naxsi has a new release 2015-04-02 23:48:51 <_mjones> ok 2015-04-02 23:49:16 it's in testing but has worked ok for me 2015-04-02 23:51:55 in this version all the server strings are anonymised 2015-04-03 00:04:38 _mjones - will be sending a patch soon for the nginx-naxsi in testing updated to 1.7.11 - it builds 2015-04-03 10:29:07 where he requested it? 2015-04-03 10:29:28 I'd prefer us to stay with stable releases 2015-04-03 10:55:16 any lua pkg to handle rss ? 2015-04-03 10:57:59 vkrishn: https://github.com/slact/lua-feedparser ? 2015-04-03 10:59:47 would be nice if were a module to turbolua 2015-04-03 10:59:59 also there is , http://www.turbolua.org/modules turboredis 2015-04-03 11:00:15 redis v3.x seems nice! 2015-04-03 11:01:02 cluster features 2015-04-03 11:02:11 we have redis 3 in aports 2015-04-03 11:02:25 yes 2015-04-03 11:02:35 i was thinking of switching to redis for pkgs 2015-04-03 11:02:46 just for fun 2015-04-03 11:03:12 but i dont have that much experiance with nosql. 2015-04-03 11:03:18 sounds good 2015-04-03 11:03:43 for filelist it could be nice 2015-04-03 11:03:47 I docs are good 2015-04-03 11:04:13 it can probably remove lots of dup data 2015-04-03 11:04:20 and run much faster 2015-04-03 11:04:23 dup date ? 2015-04-03 11:04:34 yes 2015-04-03 11:04:36 for instance 2015-04-03 11:04:38 between archs 2015-04-03 11:04:56 /usr/bin/sh /usr/bin/ls 2015-04-03 11:05:16 im not sure yet. but i think it can. 2015-04-03 11:05:20 i.e symlink to busybox 2015-04-03 11:05:38 wrong example :) i mean regular files 2015-04-03 11:07:12 redis deduplicates data? 2015-04-03 11:07:45 I don't think so 2015-04-03 11:07:55 but would surely run faster than anything 2015-04-03 11:08:15 so a builder would trigger the pkg list, that would overwrite redis key->value 2015-04-03 11:08:26 yes 2015-04-03 11:08:29 no need to rebuild full db 2015-04-03 11:08:41 but there should also be logic to remove datra 2015-04-03 11:08:43 data 2015-04-03 11:09:19 meaning it has to be more tightly integrated with mqtt/git hooks 2015-04-03 11:09:42 yeah, or compare with aports dir 2015-04-03 11:11:25 would be nice, maybe also msg to msg.a.o/pkgs/update/ 2015-04-03 11:12:05 if somebody has an idea how to organize data in redis, im all ears :) 2015-04-03 11:12:19 i guess it will be repo:arch:pkgname value 2015-04-03 11:12:26 use hierarchy like mqtt msgs 2015-04-03 11:12:35 for keys 2015-04-03 11:12:39 clandmeter: yeah, looks sane 2015-04-03 11:12:59 but dedub would be nice 2015-04-03 11:13:22 most data in arch is dup 2015-04-03 11:13:40 s/dedub/dedup :) 2015-04-03 11:13:50 almost weekend. 2015-04-03 11:15:37 RAM is cheap :p 2015-04-03 11:15:47 really ! 2015-04-03 11:16:03 it was 4years back 2015-04-03 11:16:16 but now seems they are controlling again 2015-04-03 11:16:51 nobody taking any action, there was one mass action by companies, iirc 2015-04-03 11:25:09 filelist for edge x86 is ~27MB 2015-04-03 11:29:05 would be easy to call rss parser(lua-feedparser) from withing turbolua 2015-04-03 11:44:22 barthalion: filelist is around 150 for all iirc 2015-04-03 11:44:55 vkrishn: i added lua-tar to aports 2015-04-03 11:45:02 so now i can read apk contents with lua 2015-04-03 11:45:43 vkrishn: which rss you want to parsE? 2015-04-03 11:49:01 was thinking to use turbolua+rss to replace liferea 2015-04-03 11:49:27 other option is tt-rss(php) 2015-04-03 11:50:42 tt-rss(php), iirc , did not support sqlite 2015-04-03 11:53:40 sir, you are entirely correct 2015-04-03 11:57:23 read apk content with lua, why needed ? 2015-04-03 12:04:37 there used to be on redis<->mosquitto daemon, just cannot find it :( 2015-04-03 12:16:41 vkrishn: so i can get file list in lua and insert it into redis? 2015-04-03 13:55:13 anyone visited http://www.truecrypt.org/ recently 2015-04-03 13:56:13 http://t.co/NAhnJTMCgO 2015-04-03 13:57:41 rng issue 2015-04-03 13:58:06 found this, https://ciphershed.org/ 2015-04-03 13:59:55 unavailable, ftp://ftp.archlinux.org/other/tc/truecrypt-7.1.tar.gz 2015-04-03 14:15:46 7.1a 2015-04-03 14:15:51 not 7.1 2015-04-03 14:16:24 also we don't provide source mirrors via ftp anymore 2015-04-03 14:16:43 https://sources.archlinux.org/other/tc/ 2015-04-06 01:15:24 heads up - our hosting provider just informed us that wiki,mail,dl-4 and dl-3 may experience a 45 minute power outage in the next hour 2015-04-06 06:16:25 mail.a.o didnt come up properly 2015-04-06 06:16:40 postgresql didnt start 2015-04-06 07:19:39 I am going to push a new font for alpine site: http://wwwtest.alpinelinux.org/ 2015-04-06 07:19:42 what do you think? 2015-04-06 07:20:00 i also increased the letter spacing in the slogan 2015-04-06 07:43:07 ncopa: looks good for me. 2015-04-06 09:08:31 ncopa: http://dl.2f30.org/releases/fortify-headers-0.4.tar.gz 2015-04-06 09:08:34 tagged 0.4 :) 2015-04-06 09:15:35 super! 2015-04-06 09:15:36 thanks! 2015-04-06 09:18:03 np :) 2015-04-07 07:25:33 fabled: i need apk update to exit with error if all mirrors failed 2015-04-07 07:25:49 testcase: 2015-04-07 07:25:52 ncdev-edge-x86_64:~# apk update --repository $url/edge/main --repositories-file 2015-04-07 07:25:52 fetch http://liskamm.alpinelinux.uk/edge/main/edge/main/x86_64/APKINDEX.tar.gz 2015-04-07 07:25:52 ERROR: http://liskamm.alpinelinux.uk/edge/main/edge/main: IO ERROR 2015-04-07 07:25:52 WARNING: Ignoring APKINDEX.f72fcf56.tar.gz: No such file or directory 2015-04-07 07:25:53 OK: 139 distinct packages available 2015-04-07 07:25:55 ncdev-edge-x86_64:~# echo $? 2015-04-07 07:25:57 0 2015-04-07 07:44:28 ncopa, ok, file a ticket please 2015-04-07 07:51:16 #4040 2015-04-07 07:51:23 http://bugs.alpinelinux.org/issues/4040 2015-04-07 07:51:49 but when i think of it, i think apk update can return number of update errors 2015-04-07 08:00:38 fabled: suggested fix: http://sprunge.us/bYae 2015-04-07 08:01:20 it will not return any error for apk add -U ... but i think thats ok 2015-04-07 08:01:56 that way you can use apk update to verify status of mirrors 2015-04-07 08:23:02 fabled: proposed patch for #4019 http://sprunge.us/TVAE 2015-04-07 08:41:24 ncopa, both look mostly ok, i think i'd let update print some end message even with errors 2015-04-07 08:42:45 i just want get rid of the "OK:" 2015-04-07 08:43:27 so it does not give the false impression that things are ok 2015-04-07 08:44:07 also, the commit.c fix should not be pushed to any 2.5.x release as it changes the behaviour 2015-04-07 08:44:14 i think it requires 2.6 2015-04-07 08:44:41 how about same as in commit: http://git.alpinelinux.org/cgit/apk-tools/tree/src/commit.c#n312 2015-04-07 08:45:30 yeah 2015-04-07 08:51:12 fabled: like this: http://sprunge.us/WZVa 2015-04-07 08:51:39 sure 2015-04-07 08:51:45 feel free to push that, and the other fix 2015-04-07 08:52:07 i'll look into finishing xattr after lunch 2015-04-07 08:52:22 hopefully being able to tag a new apk-tools for edge tomorrow or so 2015-04-07 08:52:48 ok. i was plannign to set up the arm builders here this afternoon 2015-04-07 08:53:36 ok. ping me if you need any help with that. though, the tarball for wandboard should work for you. 2015-04-07 08:56:13 to we have packages for mysql or mariadb galera clusters? 2015-04-07 08:58:08 s/to/do/ 2015-04-07 11:37:59 ncopa: Can you remind me what ARM board do you use for building? 2015-04-07 11:39:13 scadu: wandboard 2015-04-07 11:39:24 ncopa: thanks. 2015-04-07 11:39:39 wandboard quad to be exact 2015-04-07 14:39:20 hu? problems with pkgs.alpinelinux.org? 2015-04-07 14:41:28 the server seems to be down 2015-04-07 14:44:20 is http://nl.alpinelinux.org down too? 2015-04-07 14:44:37 and alpinelinux.org too 2015-04-07 14:45:57 seems like it 2015-04-07 14:49:07 clandmeter: whats up with alpinelinux.org? 2015-04-07 14:49:47 * clandmeter has quit (Ping timeout: 246 seconds) 2015-04-07 14:49:59 sounds like internet is broken 2015-04-07 14:54:32 NL infra isp is currently down, which means www,nl master mirror, bugs,forum, and probably more 2015-04-07 14:57:08 pkgs as well :) 2015-04-07 15:01:10 damn, i need to take another mirror 2015-04-07 15:02:13 http://mirrors.gigenet.com/alpinelinux/edge/ :) 2015-04-07 15:05:42 android tethering ftw :) 2015-04-07 18:18:14 ACTION hides behinds a large bush of commits 2015-04-08 03:54:11 <_mjones> 4.2GB free space is not sufficient to compile openjdk7, even when all outside dependencies are already installed. Inconceivable! 2015-04-08 07:22:03 morning 2015-04-08 07:32:38 will AL v3.2 have lua v5.1 or just luajit ? 2015-04-08 07:47:10 found another simple rss parser, http://www.codemadness.nl/blog/sfeed-simple-feed-parser.html 2015-04-08 08:06:51 i have a plan how to migrate git.a.o to gitolite 2015-04-08 08:08:55 I'll miss having push access to everything :p 2015-04-08 08:13:19 http://www.gitphp.org/ ;) 2015-04-08 08:14:15 why do thety reimplement this ugly ui in php? 8) 2015-04-08 08:15:21 hmm.. GitLab acquires Gitorious 2015-04-08 08:16:19 mosez: you had some git software written in go 2015-04-08 08:19:31 ncopa: not me... i'm just using it :) 2015-04-08 08:19:37 ncopa: and it got a nice ui :) 2015-04-08 08:21:26 and thoughts on having go/ subfolder in git aports repo 2015-04-08 08:35:52 ncopa: those dup.pw's dont work for other repos 2015-04-08 08:42:58 clandmeter: true 2015-04-08 08:42:59 hum 2015-04-08 08:43:19 i suppose we need embed the repo in there 2015-04-08 09:57:00 cool 2015-04-08 09:57:14 i mounted nfs on my wandbard 2015-04-08 09:57:22 and created a lxc container on nfs 2015-04-08 11:07:14 main/lua-feedparser , having problem parsing format at, http://pear.php.net/feeds/latest.rss 2015-04-08 11:08:31 any chance of getting this, http://bugs.alpinelinux.org/issues/1774 2015-04-08 11:21:53 thanks 2015-04-08 11:22:55 http://www.keplerproject.org/ , this project seems not active 2015-04-08 11:23:29 we have 5 pkg 2015-04-08 11:24:27 moved to github 2015-04-08 11:37:55 http://www.keplerproject.org/lualdap/ ! 2015-04-08 11:38:38 looking at luaforge 2015-04-08 11:41:50 seems more viable, https://github.com/zx2c4/lualdap 2015-04-08 11:42:02 abandoned in 2007, and revived in 2015 2015-04-08 13:11:11 apk 2.6.0_rc1 is badly broken 2015-04-08 13:11:14 $ sudo apk upgrade -U -a 2015-04-08 13:11:14 Illegal instruction 2015-04-08 13:19:48 uh, blame it on my gitrepo :) 2015-04-08 13:24:02 hum 2015-04-08 13:24:05 its only local 2015-04-08 13:24:06 phew 2015-04-08 13:26:52 :) 2015-04-08 13:31:14 heh 2015-04-08 13:31:19 it looks like its fortify 2015-04-08 13:31:26 that triggered it 2015-04-08 13:31:48 Program terminated with signal 4, Illegal instruction. 2015-04-08 13:31:48 #0 memcpy (n=4392, src=0x284c6799ec8, dest=0x284c6798ff0) 2015-04-08 13:31:48 at /usr/include/fortify/string.h:32 2015-04-08 13:31:48 32 __builtin_trap(); 2015-04-08 13:36:17 smells like offby one bug 2015-04-08 13:36:46 ha :) 2015-04-08 13:36:59 either off by one or overlapping src/dst 2015-04-08 13:37:06 yup 2015-04-08 13:37:18 at least it works :P 2015-04-08 13:37:30 this is cool 2015-04-08 13:37:32 really cool 2015-04-08 13:38:28 hm the difference between the addresses is smaller than 'n' 2015-04-08 13:38:30 :) 2015-04-08 13:46:39 nice 2015-04-08 13:47:05 I am really surprised that fortified firefox can actually start :P 2015-04-08 13:47:27 http://git.alpinelinux.org/cgit/apk-tools/tree/src/io.c#n227 2015-04-08 13:47:32 there it happens 2015-04-08 13:47:45 but i have no idea why 2015-04-08 13:48:33 is left.ptr overlapping with buffer? 2015-04-08 13:48:56 and is left.len correct? 2015-04-08 13:49:27 DEBUG: left.len=4392, sizeof(isbs->buffer)=8192 2015-04-08 13:49:27 Illegal instruction 2015-04-08 13:49:48 i'm gonna check the diff of left.ptr 2015-04-08 13:49:53 it might be overlapping 2015-04-08 13:50:00 ok 2015-04-08 13:51:48 note that __builtin_object_size() will calculate sizeof(isbs->buffer) + sizeof(size_t), so there's still a possibility for false negatives (allowing the overflow through) 2015-04-08 13:52:09 that's because it calculates the size of the outter object start at isbs->buffer 2015-04-08 13:52:30 there is a way to force it to calculate the size of the inner object but I think there are cases where this breaks non-UB programs 2015-04-08 13:52:39 this is not related to your crash though, just a reminder 2015-04-08 13:54:39 so struct foo { int a; char b[16]; int c; } foo; printf("%zu\n", __builtin_object_size(foo.b, 0)); will print 20 not 16 2015-04-08 13:55:25 ncopa, stateless: yes, left.len points into buffer 2015-04-08 13:55:35 left.ptr? 2015-04-08 13:55:41 might be safer to use memmove. 2015-04-08 13:55:48 cool :) 2015-04-08 13:56:10 also you have a function pointer called read() :p 2015-04-08 13:56:16 yes 2015-04-08 13:56:29 (would break the macro-only version of fortify) 2015-04-08 13:56:46 backtrace: https://gist.github.com/ncopa/c5647401353e2e61f239 2015-04-08 13:57:15 yeah it is an overlap problem 2015-04-08 13:57:46 memmove() will work as long as the length is ok 2015-04-08 13:58:00 yes,length is guaranteed ok 2015-04-08 13:58:07 its slower though? 2015-04-08 13:58:11 yeah. to be compliant it needs to be memmove 2015-04-08 13:58:38 i think there's no big difference 2015-04-08 13:58:51 it is probably not the dominant factor but I am not familiar with the code 2015-04-08 14:01:14 DEBUG: left.len=4392, sizeof(isbs->buffer)=8192, left.ptr-buffer=3800 2015-04-08 14:01:31 there u go :) 2015-04-08 14:02:03 and memmove fixed it 2015-04-08 14:02:04 cool 2015-04-08 14:02:46 yeah. i remember coding that 2015-04-08 14:03:13 dst is guaranteed to be less than src 2015-04-08 14:03:21 so i thought memcpy was ok. it was long time ago. 2015-04-08 14:03:37 but "no overlap" is not guaranteed always 2015-04-08 14:04:28 it is never ok if they overlap 2015-04-08 14:04:41 but it happens to work sometimes depending on how memcpy is implemented (forward/reverse copying etc.) 2015-04-08 14:04:57 fortify proved to be useful 2015-04-08 14:05:00 very nice 2015-04-08 14:05:25 openbsd memcpy also has this check for overlapping src/dst but they are considering to remove it after the 5.7 release 2015-04-08 14:06:25 yes 2015-04-08 14:06:50 i think it'll work in most implementations 2015-04-08 14:06:56 but it's not allowed by the standard 2015-04-08 14:07:00 so it's application bug 2015-04-08 14:07:38 seems musl ends up calling memcpy anyway 2015-04-08 14:07:45 :) 2015-04-08 14:07:46 if (s+n <= d || d+n <= s) return memcpy(d, s, n); 2015-04-08 14:07:53 in memmove 2015-04-08 14:08:10 yeah but this is internal knowledge :) 2015-04-08 14:08:19 yes 2015-04-08 14:08:25 glibc has similar iirc 2015-04-08 14:08:53 at the time my coding was more based on that than the standards 2015-04-08 14:09:29 so yeah. nice catch. it's conformance issue; though in practice it was working ok. 2015-04-08 14:09:40 and as you saw, i pushed the fix out already :) 2015-04-08 14:09:52 i am happy 2015-04-08 14:10:15 nothing was broken (in practice) but code quality improves 2015-04-08 14:10:37 we could get surprises if we ported apk to some crazy platform 2015-04-08 14:10:50 yeah 2015-04-08 14:10:57 fabled: I am thinking of adding also out of bound read checks to some functions 2015-04-08 14:11:12 it is probably worth it although there'll be more branches 2015-04-08 14:11:37 so for memcpy() there would be an additional check to see if we reading beyond src 2015-04-08 14:12:01 s/we reading/we are reading/ 2015-04-08 14:12:19 glibc fortify doesn't do that 2015-04-08 14:34:05 are there other imporant things we need in abuild before i tag a release? 2015-04-08 15:45:17 I have read some backlog talking about a consul package in preparation. count me in as avid tester ;) 2015-04-08 16:18:20 Hi guys. I made a change to Python3 package yesterday, so it moved pyconfig.h from Python3-dev to Python3, but today after new package arrived I've noticed an issue in /usr/lib/libpython3.4m.so.1.0, when I run `python3` the following error occures: 2015-04-08 16:18:20 Error relocating /usr/lib/libpython3.4m.so.1.0: __flt_rounds: symbol not found 2015-04-08 16:18:20 Does anybody know how to fix this? 2015-04-08 16:20:50 `strings /usr/lib/libpython3.4m.so.1.0 | grep __flt_rounds` on the -r0 package shows nothing, but there is indeed that __flt_rounds symbol in the new -r1 package library 2015-04-08 16:25:13 (I changed my nick from Guest16747. I'm still waiting for an answer) 2015-04-08 16:28:12 if it's c++ code then you need a patch from musl git 2015-04-08 16:29:10 hm it should be in musl 1.1.8 release 2015-04-08 16:29:21 http://git.musl-libc.org/cgit/musl/commit/?id=11d1e2e2ded07673411ba872c1e3d0096dc79439 2015-04-08 16:29:25 frol__: ^ 2015-04-08 16:41:44 frol__: checking 2015-04-08 16:42:06 works for me 2015-04-08 16:47:12 hmmm, where do I get musl 1.1.8 on Alpine? 2015-04-08 16:47:18 fetch http://dl-4.alpinelinux.org/alpine/v3.1/main/x86_64/APKINDEX.tar.gz 2015-04-08 16:47:18 fetch http://nl.alpinelinux.org/alpine/edge/testing/x86_64/APKINDEX.tar.gz 2015-04-08 16:47:18 v3.1.3-16-gc9a8f0f [http://dl-4.alpinelinux.org/alpine/v3.1/main] 2015-04-08 16:47:18 v150306-704-g0515300 [http://nl.alpinelinux.org/alpine/edge/testing] 2015-04-08 16:47:18 OK: 7360 distinct packages available 2015-04-08 16:47:22 musl-1.1.5-r3 2015-04-08 16:47:24 musl-1.1.5-r3 2015-04-08 16:47:27 musl-utils-1.1.5-r3 2015-04-08 16:47:29 musl-dbg-1.1.5-r3 2015-04-08 16:47:31 musl-dev-1.1.5-r3 2015-04-08 16:47:35 fetch http://dl-4.alpinelinux.org/alpine/v3.1/main/x86_64/APKINDEX.tar.gz 2015-04-08 16:47:38 fetch http://nl.alpinelinux.org/alpine/edge/testing/x86_64/APKINDEX.tar.gz 2015-04-08 16:47:40 v3.1.3-16-gc9a8f0f [http://dl-4.alpinelinux.org/alpine/v3.1/main] 2015-04-08 16:47:42 v150306-704-g0515300 [http://nl.alpinelinux.org/alpine/edge/testing] 2015-04-08 16:47:44 OK: 7360 distinct packages available 2015-04-08 16:47:49 musl-1.1.5-r3 2015-04-08 16:47:51 musl-1.1.5-r3 2015-04-08 16:47:53 musl-utils-1.1.5-r3 2015-04-08 16:47:55 musl-dbg-1.1.5-r3 2015-04-08 16:47:57 musl-dev-1.1.5-r3 2015-04-08 16:51:15 That is odd. On pkgs.alpinelinix.org there is musl 1.1.8 in main, but I see only 1.1.5 from my Alpine Docker image 2015-04-08 16:52:45 Ok, I've found it in edge/main. Checking... 2015-04-08 16:53:00 frol__, yes, in edge/main 2015-04-08 16:54:01 Now it works! Thank you guys! 2015-04-08 17:31:42 pastebins so hard, wow. 2015-04-08 17:34:50 diversity also means there's someone else at the other end of the bell-curve 2015-04-08 17:35:24 ACTION waves at the otherside of a git-repo bellcurve 2015-04-08 18:00:23 ncopa, newsbeuter gives error while starting 2015-04-08 18:00:25 Starting newsbeuter 2.9... 2015-04-08 18:00:25 Loading configuration...done. 2015-04-08 18:00:25 Opening cache...done. 2015-04-08 18:00:25 Loading URLs from /dev/shm/urls...done. 2015-04-08 18:00:25 Loading articles from cache...done. 2015-04-08 18:00:28 STFL Parser Error near ''. 2015-04-08 18:00:30 Aborted 2015-04-08 18:01:02 I have a strace file too 2015-04-08 18:02:29 writev(2, [{"", 0}, {"STFL Parser Error near '", 24}], 2STFL Parser Error near ') = 24 2015-04-08 18:02:29 writev(2, [{"", 0}, {"'.\r\n", 4}], 2'. 2015-04-08 18:02:29 ) = 4 2015-04-08 18:02:47 stfl has a requirement, "STFL is a wide-character library it will only work on systems which 2015-04-08 18:02:47 do have wide-character support in the system libraries" 2015-04-08 18:03:46 gtg 2015-04-08 18:11:08 Could you please test `python3 -c 'import socket'`? 2015-04-08 18:11:22 I'm getting "ImportError: No module named '_socket'" 2015-04-08 18:17:55 <_mjones> vkrishn: I'm sure I saw wchar functions supported on musl. Do you have a link to that stfl requirement? 2015-04-08 18:18:11 he's gone 2015-04-08 18:18:23 <_mjones> oops. 2015-04-08 18:19:18 <_mjones> a runtime error, too. widechar is deprecated in light of utf-8 unless I'm mistaken. 2015-04-08 18:20:13 frol__: correct, it's broken 2015-04-08 18:20:59 :( 2015-04-08 18:24:17 and _socket-something.so is indeed not in /usr/lib/python3.4/lib-dynload/ 2015-04-08 18:24:21 I don't see any problem with package itself. My guess is that ./configure script failed to find any socket.h and dropped it from making, but it worked before! 2015-04-08 18:24:40 yeah, that's my guess too 2015-04-08 18:24:48 I'll take a look now, I think 2015-04-08 18:25:09 Thank you 2015-04-08 18:30:59 It would be great if you also update Python 3.4.2 to Python 3.4.3 in the package. 2015-04-08 18:31:10 sure, will do 2015-04-08 18:31:18 let's start from that then 2015-04-08 18:40:58 please drop me a line once you have any update 2015-04-08 18:41:29 I'm testing 3.4.3 build atm 2015-04-08 18:48:38 /usr/lib/gcc/x86_64-alpine-linux-musl/4.9.2/../../../../x86_64-alpine-linux-musl/bin/ld: cannot find -lsocket 2015-04-08 18:48:42 I wonder if it's expected 2015-04-08 18:49:48 gotta go, I'll work more on tomorrow 2015-04-08 18:49:59 python3's APKBUILD wouldn't mind more love 2015-04-08 18:50:07 like running test suite, and so on 2015-04-08 18:50:25 <_mjones> a few others like that too. 2015-04-08 18:50:40 <_mjones> iojs test suite doesn't pass 100% but there are no regressions so... 2015-04-08 19:38:43 There are some errors while Python making, which directly point that _socket and ossaudiodev modules failed to build: 2015-04-08 19:38:47 building '_socket' extension 2015-04-08 19:38:47 gcc -fPIC -Wno-unused-result -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -Werror=declaration-after-statement -I./Include -I. -IInclude -I/tmp/Python-3.4.3/Include -I/t 2015-04-08 19:38:47 mp/Python-3.4.3 -c /tmp/Python-3.4.3/Modules/socketmodule.c -o build/temp.linux-x86_64-3.4/tmp/Python-3.4.3/Modules/socketmodule.o 2015-04-08 19:38:47 /tmp/Python-3.4.3/Modules/socketmodule.c: In function 'makesockaddr': 2015-04-08 19:38:47 /tmp/Python-3.4.3/Modules/socketmodule.c:1175:14: error: dereferencing pointer to incomplete type 2015-04-08 19:38:49 if (a->can_ifindex) { 2015-04-08 19:38:51 ^ 2015-04-08 19:38:53 /tmp/Python-3.4.3/Modules/socketmodule.c:1176:32: error: dereferencing pointer to incomplete type 2015-04-08 19:38:56 ifr.ifr_ifindex = a->can_ifindex; 2015-04-08 19:38:58 ^ 2015-04-08 19:39:00 /tmp/Python-3.4.3/Modules/socketmodule.c:1183:38: error: dereferencing pointer to incomplete type 2015-04-08 19:39:02 a->can_family); 2015-04-08 19:39:04 ^ 2015-04-08 19:39:07 /tmp/Python-3.4.3/Modules/socketmodule.c: In function 'getsockaddrlen': 2015-04-08 19:39:09 /tmp/Python-3.4.3/Modules/socketmodule.c:1802:28: error: invalid application of 'sizeof' to incomplete type 'struct sockaddr_can' 2015-04-08 19:39:11 *len_ret = sizeof (struct sockaddr_can); 2015-04-08 19:39:16 building 'ossaudiodev' extension 2015-04-08 19:39:18 gcc -fPIC -Wno-unused-result -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -Werror=declaration-after-statement -I./Include -I. -IInclude -I/tmp/Python-3.4.3/Include -I/t 2015-04-08 19:39:20 mp/Python-3.4.3 -c /tmp/Python-3.4.3/Modules/ossaudiodev.c -o build/temp.linux-x86_64-3.4/tmp/Python-3.4.3/Modules/ossaudiodev.o 2015-04-08 19:39:22 In file included from /tmp/Python-3.4.3/Modules/ossaudiodev.c:33:0: 2015-04-08 19:39:24 /usr/include/sys/soundcard.h:1:29: fatal error: linux/soundcard.h: No such file or directory 2015-04-08 19:39:26 #include 2015-04-08 19:40:36 please, use pastebin or whatever. 2015-04-08 19:40:53 I'm sorry, will do it now 2015-04-08 19:41:17 cool. 2015-04-08 19:43:40 http://pastebin.com/GMTs1RAv 2015-04-08 19:45:58 <_mjones> linux/soundcard.h indeed exists on Alpine. Have you installed package "linux-headers"? 2015-04-08 19:47:52 My point was about _socket, but yes after I installed linux-headers ossaudiodev module got compiled 2015-04-08 19:52:25 <_mjones> frol__: I suspect "-Wall -Wstrict-prototypes" is causing that. 2015-04-08 19:52:27 The last post in the thread here describes something very similar: https://forums.gentoo.org/viewtopic-p-6498222.html?sid=4516014b0d6778720f84688886824757 2015-04-08 19:55:20 <_mjones> Similar. Maybe musl header. But listed as applying to old kernels -- not applicable -- and "Confirmed as fixed at least in Python 3.3.5." 2015-04-08 19:56:00 I have applied the patch and it works! 2015-04-08 19:57:54 <_mjones> are you fixing the APKBUILD for a submission? 2015-04-08 19:58:49 <_mjones> ACTION wonders how many of these patches are being submitted upstream. Not necessarily accepted, but submitted. 2015-04-08 19:59:16 I'm not very familiar with APKBUILD, so could you make necessary changes if I provide the patch? 2015-04-08 20:02:04 http://pastebin.com/KLHnAXbc 2015-04-08 20:04:37 report this to python upstream 2015-04-08 20:05:11 Hmm, it seems that this was in upstream - http://bugs.python.org/issue20065 2015-04-08 20:06:07 It is marked as fixed and patch is similar, so it will take some time to investigate when and why it was reverted or something 2015-04-08 20:08:23 I will be able to continue working on this only next week or so :( 2015-04-08 20:08:50 Have a nice day, guys. Bye 2015-04-08 20:23:58 Frosh: stateless ↑ 2015-04-08 20:24:04 wut, not Frosh, sorr 2015-04-08 20:24:06 -y 2015-04-08 20:24:22 still, stateless ↑; may it be fortify issue? 2015-04-08 20:30:17 <_mjones> probably not reverted. Probably a bad macros. Like #ifdef __GLIBC__ instead of #ifdef __linux__ 2015-04-08 21:04:28 ok, I guess I'll just apply his patch 2015-04-08 21:05:10 missed that message somehow 2015-04-08 21:21:31 got that text from stfl-0.24.tar.gz/stfl-0.24/README 2015-04-08 21:22:00 forgot to mention, update stfl pls 2015-04-08 21:23:26 _mjones, I read the logs for any msgs left :) 2015-04-08 21:27:10 gtg, watch some tv (zero hour) 2015-04-08 21:28:31 <_mjones> vkrishn: dalias wanted to know how that error happened. However, I'm not so sure it's a musl error. 2015-04-09 00:09:47 http://bugs.alpinelinux.org/issues/3847 - this can be closed - 3.1.3 is not affected 2015-04-09 06:04:43 paxmark? 2015-04-09 06:06:26 barthalion, from main/paxmark, it's similar to paxmark.sh (symlink shipped) that gentoo uses 2015-04-09 06:06:33 it basically does paxctl + xattr setting at this time 2015-04-09 06:06:53 we can drop paxctl after release cycle 2015-04-09 06:09:09 fabled: ah 2015-04-09 06:09:17 great to hear that 2015-04-09 06:10:58 apk is xattr enabled now 2015-04-09 06:11:17 so i'm just converting all aports to use paxmark, and we change kernel config to xattr pax markings 2015-04-09 06:19:34 morning 2015-04-09 06:19:42 re stfl and widechar 2015-04-09 06:20:07 looks like you need define _XOPEN_SOURCE_EXTENDED to get the widechar includes in ncurses 2015-04-09 06:21:05 firefox failed to build? 2015-04-09 06:21:38 main/nginx-lua/APKBUILD 2015-04-09 06:21:42 collect2: error: ld returned 1 exit status 2015-04-09 06:21:53 ./../../media/webrtc/signaling/signaling_sipcc/ccsip_spi_utils.o: In function `recv': 2015-04-09 06:21:53 /usr/include/fortify/sys/socket.h:20: multiple definition of `write' 2015-04-09 06:21:53 ../../../media/webrtc/signaling/signaling_sipcc/cc_device_feature.o:/usr/include/fortify/sys/socket.h:20: first defined here 2015-04-09 06:21:57 huh? firefox? 2015-04-09 06:22:02 sounds like fortify issue 2015-04-09 06:22:34 ah 2015-04-09 06:22:39 main/firefox 2015-04-09 06:22:57 check git log of testing/firefox 2015-04-09 06:23:07 i had to add a patch for fortify iirc 2015-04-09 06:23:22 also, while at it check which is the latest ESR relelease for ff 2015-04-09 06:23:37 i suspect there are a new ESR release 2015-04-09 06:23:40 so upgrade while at it 2015-04-09 06:23:58 fix-fortify-inline.patch ? 2015-04-09 06:24:05 probably yes 2015-04-09 06:24:35 yes 2015-04-09 06:24:37 that one 2015-04-09 06:25:44 hum 2015-04-09 06:25:47 libiconv 2015-04-09 06:25:59 we use the iconv from musl now? 2015-04-09 06:26:22 headers and lib? so we should not linuk with -liconv? 2015-04-09 06:29:58 ncopa, ok i have ff commit pending test build 2015-04-09 06:30:36 fabled: should we add -liconv nowdays? 2015-04-09 06:31:10 ncopa, mmm... if we are using musl iconv, no 2015-04-09 06:31:22 if we want to link against the external iconv library, then yes 2015-04-09 06:31:29 in most cases musl iconv should beok 2015-04-09 06:31:58 my concern is that libiconv-dev is in makedepends 2015-04-09 06:32:12 will it use musl headers or libiconv headers? 2015-04-09 06:32:27 i suppose if it uses libiconv headers we should also use -liconv 2015-04-09 06:32:38 if we use musl headers we should *not* use -liconv 2015-04-09 06:32:45 checking 2015-04-09 06:32:48 ah 2015-04-09 06:32:58 libiconv is dummy transitional pkg in musl 2015-04-09 06:33:03 ok 2015-04-09 06:33:08 having libiconv-dev dependency just ships empty libiconv.a 2015-04-09 06:33:12 so -liconv should be safe? 2015-04-09 06:33:14 ok 2015-04-09 06:33:16 goody 2015-04-09 06:33:17 it's safe 2015-04-09 06:33:26 better to just remove libiconv-dev dep, and -liconv addition 2015-04-09 06:33:34 but we can remove 2015-04-09 06:33:36 ok 2015-04-09 06:33:48 we want to kill libiconv from aports 2015-04-09 06:33:58 maybe we should remove libiconv package and force build breakage? 2015-04-09 06:34:01 so we can clean that up 2015-04-09 06:34:06 might be an idea yes 2015-04-09 06:34:41 maybe sed s/libiconv-dev// 2015-04-09 06:34:47 yeah 2015-04-09 06:35:05 fabled: do you want me to test ff build? 2015-04-09 06:35:13 i'm building that already 2015-04-09 06:35:16 ok 2015-04-09 06:35:21 should finish soon 2015-04-09 06:36:42 ncdev-edge-x86_64:~/aports/main$ grep -- -liconv */*.patch */APKBUILD | sprunge 2015-04-09 06:36:42 http://sprunge.us/URCb 2015-04-09 06:36:59 ncdev-edge-x86_64:~/aports/main$ grep libiconv-dev */APKBUILD | sprunge 2015-04-09 06:36:59 http://sprunge.us/ICfa 2015-04-09 06:37:25 i need to update openjdk7 for paxmark 2015-04-09 06:37:28 i can fix that while at it 2015-04-09 06:37:44 goody 2015-04-09 06:38:25 should be simple clean up in most places 2015-04-09 06:38:28 i think we dont need rebuild 2015-04-09 06:38:29 ye 2015-04-09 06:39:35 i think we just start with purging libiconv-dev and fix things as they break 2015-04-09 06:41:42 ncopa, sure 2015-04-09 06:41:52 openjdk7 fix is simple. i have changes pending test. 2015-04-09 06:41:55 will push after ff 2015-04-09 06:42:17 it's nice. i get to delete three patches :) 2015-04-09 06:47:52 i think I'll send an email to alpine-devel 2015-04-09 06:48:10 mention pax xattr changes too? 2015-04-09 06:48:17 separate email maybe 2015-04-09 06:48:36 maybe twit about it too 2015-04-09 06:51:33 huh? 2015-04-09 06:52:06 oh, maybe it was from previous build 2015-04-09 06:52:37 coffee 2015-04-09 07:40:23 <_mjones> I'll show you 2015-04-09 07:54:45 fabled: qemu issue: http://sprunge.us/ORaV 2015-04-09 08:00:36 <_mjones> unrelated fyi: a user reports that gnu sed and tar fail some 'make check' tests, and I confirm that. fyi, gnu tar 1.28 failures from abuild: http://ix.io/hL2 2015-04-09 08:04:01 _mjones: could you figure out exactly what fails and why? 2015-04-09 08:04:28 my guess is that gnu sed/tar make invalid assumptions from the libc 2015-04-09 08:04:55 <_mjones> uh, no. I'm afraid I'm already far behind on submitting the few diffs I have, and the ones on which I'm working. 2015-04-09 08:05:06 <_mjones> @ncopa: that's my guess as well. 2015-04-09 08:05:55 <_mjones> I wil mark them down to look at them when the others are submitted, though. 2015-04-09 08:06:59 <_mjones> I was trying to build openjdk7 but it's massive. I see no version bump despite fabled dropping the patches. 2015-04-09 08:07:32 <_mjones> Looks like icedtea 2.6.0 is out. I ran out of virtual disk space trying to build it. 2015-04-09 08:12:14 icedtea-web? 2015-04-09 08:12:52 no, icedtea for openjdk 2015-04-09 08:14:10 <_mjones> yes. 2015-04-09 08:16:04 i'd like to fix openjdk for arm too 2015-04-09 08:36:42 I have found some minutes to investigate Python3 package further. There are two ways I found to fix "ImportError: No module named '_socket'". The correct way is to have linux-headers before ./configure step. Should I add 'linux-headers' to 'makedepends' in the package or could you test it on your build server before I make a pull-request? 2015-04-09 08:38:03 you can add it to make depends 2015-04-09 08:38:12 makedepends* 2015-04-09 08:38:21 ok 2015-04-09 08:38:32 and in commit message you explain that it is needed for ImportError: No module named '_socket 2015-04-09 08:39:00 prefix the first line with /python3: 2015-04-09 08:39:17 I'm doublechecking this on my server now and will add a pull-request in 10-20 minutes 2015-04-09 08:39:58 you can also do: git format-patch -1 --stdout | sprunge 2015-04-09 08:40:08 and paste the url here 2015-04-09 08:40:16 and i'll check it 2015-04-09 08:40:59 interesting: out of memory 2015-04-09 08:41:16 <_mjones> @ncopa: on what? 2015-04-09 08:41:24 http://bld2.alpinelinux.org/buildlogs/build-edge-x86_64/main/firefox/firefox-31.6.0-r0.log 2015-04-09 08:41:29 firefox build 2015-04-09 08:41:37 frol_: let me finish something at work and I'll push it 2015-04-09 08:42:22 <_mjones> @ncopa: haha. 640M wasn't enough to build something the other day so the VM got 4G, which I can afford. 2015-04-09 08:42:49 http://sprunge.us/QDXd 2015-04-09 08:42:55 48G should work 2015-04-09 08:43:22 it could be a pax problem 2015-04-09 08:43:37 or a mem leak 2015-04-09 08:46:37 <_mjones> must be a dual processor xeon board to have 48G. 2015-04-09 08:47:00 <_mjones> Or a single-processor lga 2011 I guess. 2015-04-09 08:47:14 cpuinfo: http://sprunge.us/CbeW 2015-04-09 08:47:21 _mjones: dont make him show off his toybox 2015-04-09 08:47:31 :) 2015-04-09 08:47:36 too late 2015-04-09 08:47:39 :) 2015-04-09 08:47:40 frol_: I'm ready to push, so ping me if it works for you 2015-04-09 08:48:06 <_mjones> I've been shopping and working on hardware this week, so it's on my mind. 2015-04-09 08:48:11 frol_: I have no keys to my playground container on work laptop to test it properly 2015-04-09 08:49:09 barhalion: I'm clonning the repo to make 'git format-patch -1 --stdout | sprunge' 2015-04-09 08:49:35 Adding 'linux-headers' works for me 2015-04-09 08:49:35 hum 2015-04-09 08:49:52 firefox xpcshel segfaults 2015-04-09 08:50:19 frol_: ok, pushing it then 2015-04-09 08:51:55 aha 2015-04-09 08:52:00 ff needs paxmark 2015-04-09 08:52:36 it already has it 2015-04-09 08:53:12 fabled: i have a feeling that paxmark is broke 2015-04-09 08:53:42 hmm 2015-04-09 08:53:44 xpcshell segfaults when building firefox 2015-04-09 08:53:52 it's possible it needs some additional flag 2015-04-09 08:55:16 oh 2015-04-09 08:55:16 duh 2015-04-09 08:55:20 remove -c 2015-04-09 08:55:23 i fix it 2015-04-09 08:55:45 do i need pkgbump? 2015-04-09 08:55:49 no 2015-04-09 08:56:02 it never built 2015-04-09 08:56:34 barthalion: http://pastebin.com/mjc86yrP 2015-04-09 08:56:46 paxmark checks only argv[1] for flags 2015-04-09 08:56:57 and -c is just ignored for compat 2015-04-09 08:58:04 maybe make it fail with error? 2015-04-09 08:59:02 ncopa, it should have returned error 2015-04-09 09:02:10 i suppose paxctl returns success if called without filenames 2015-04-09 09:05:15 ncopa: barthalion: could you test it: http://pastebin.com/mjc86yrP ? 2015-04-09 09:06:02 frol_: looks like barthalion already pushed it: http://dup.pw/94bbd1c8 2015-04-09 09:06:20 Wow, great 2015-04-09 09:06:49 frol_: thanks for reporting it and for digging up the solution 2015-04-09 09:07:42 Thank you all for Alpine Linux distro! 2015-04-09 09:09:02 fabled: either we need to make paxmark swallow -c as $1 or we need to make it error when its there 2015-04-09 09:09:05 ncopa, all of main except gcc is now in paxmark 2015-04-09 09:09:15 --- openjdk/make/Defs-internal.gmk.orig 2014-04-17 16:35:03.218100233 +0000 2015-04-09 09:09:15 +++ openjdk/make/Defs-internal.gmk 2014-04-17 16:35:18.085038258 +0000 2015-04-09 09:09:15 @@ -308,9 +308,9 @@ 2015-04-09 09:09:15 PAX_COMMAND = /usr/sbin/paxmark.sh 2015-04-09 09:09:15 endif 2015-04-09 09:09:16 ifeq ($(ARCH), i586) 2015-04-09 09:09:18 - PAX_COMMAND_ARGS="-msp" 2015-04-09 09:09:19 so doing that and testing after wards 2015-04-09 09:09:20 + PAX_COMMAND_ARGS="-c -msp" 2015-04-09 09:09:22 else 2015-04-09 09:09:24 - PAX_COMMAND_ARGS="-m" 2015-04-09 09:09:26 + PAX_COMMAND_ARGS="-c -m" 2015-04-09 09:09:28 endif 2015-04-09 09:09:30 2015-04-09 09:09:30 ncopa, that patch is removed 2015-04-09 09:09:32 # Common make arguments (supplied to all component builds) 2015-04-09 09:10:08 --with-pax=paxmark 2015-04-09 09:10:08 i'm off for lunch 2015-04-09 09:10:12 will fix carnage aftewards 2015-04-09 09:10:22 i believe it runs paxmark -c ... 2015-04-09 09:10:25 ncopa, no 2015-04-09 09:10:29 the patch add -c was removed 2015-04-09 09:10:41 i had to add -c when openjdk built with --with-pax=paxctl 2015-04-09 09:10:55 back in 30-40mins 2015-04-09 09:11:07 ok 2015-04-09 09:30:19 \o/ a libressl contender https://github.com/Microsoft/openssl 2015-04-09 10:05:28 Does firefox failure block python3@testing package build as well? 2015-04-09 10:28:22 are builders active ? 2015-04-09 10:28:27 vkrishn, no 2015-04-09 10:28:34 i'm investigating the current breakage 2015-04-09 10:28:47 it seems python is broken 2015-04-09 10:39:27 i wonder what to do with symlinks 2015-04-09 10:39:40 xattrs can be attached to the dereferenced file, or the symlink itself 2015-04-09 10:39:54 where as the elf marking, just modifies always the underlying file 2015-04-09 10:58:36 can there be a ping or any means from dev.a.o to bld[\d].a.o or viceversa to know its on, and send msg.a.o when not 2015-04-09 10:59:29 i think ncopa was planning some sort of web page to show the status real time 2015-04-09 10:59:37 may every 10mins, like build/build-edge-x86 status: not active 2015-04-09 10:59:45 they are back on, and building stuff now 2015-04-09 10:59:54 ok 2015-04-09 11:00:37 currently is says " build/build-edge-x86 idle" when active but not doing anything 2015-04-09 11:01:37 I am guessing that webpage will use msg.a.o 2015-04-09 11:06:31 vkrishn: i have some sort of plan 2015-04-09 11:06:55 generally, i want simplify the current git hook 2015-04-09 11:07:11 to only generate mqtt message 2015-04-09 11:07:38 and i want teach buildrepo to submit mqtt messages 2015-04-09 11:08:01 i am baning my head into lua atm 2015-04-09 11:08:22 one annoyance is that there are no nice arg parser 2015-04-09 11:08:53 second is that there are no nice way to run external programs safely 2015-04-09 11:11:04 like wanting to avoid mqtt_exec ? 2015-04-09 11:11:22 once i have buildrepo to send proper mqtt messages we can fix the rest 2015-04-09 11:11:30 ok 2015-04-09 11:11:43 first step is have a strategy or plan on how messages are sent 2015-04-09 11:11:46 and what messages 2015-04-09 11:11:49 and on what topics 2015-04-09 11:12:21 could you pls take note to keep hierarchy of msgs similar to filesystems 2015-04-09 11:12:31 yes 2015-04-09 11:12:49 but how do we map things to the hiearchy? 2015-04-09 11:12:59 git events 2015-04-09 11:13:02 for example 2015-04-09 11:13:23 we need publish each git commit (?) 2015-04-09 11:13:31 at least if we want each commit into irc 2015-04-09 11:13:36 first I need to know a rough diagram, like origin or msgs and final consumption 2015-04-09 11:13:59 then we want events on each git push 2015-04-09 11:14:18 the builders does not need to trigger new build for every git commit, but for every git push 2015-04-09 11:14:31 then we want message for new tags 2015-04-09 11:14:57 i know github can emit mqtt messages 2015-04-09 11:15:00 we could look what they do 2015-04-09 11:15:29 we could have separate topic for new branches too 2015-04-09 11:15:41 I issue is if github is down so is the builders(likely) 2015-04-09 11:16:08 no 2015-04-09 11:16:11 we dont know that 2015-04-09 11:16:29 then we want builder status messages 2015-04-09 11:16:38 builder starts build package 2015-04-09 11:17:16 builder status is: building , upload to mirror, idle 2015-04-09 11:17:27 or something like that 2015-04-09 11:17:38 we want different topic for build errors 2015-04-09 11:18:01 should post what builder, what package broke 2015-04-09 11:18:06 maybe even with url to log 2015-04-09 11:18:34 then we have mirror messages 2015-04-09 11:18:38 rsync/mirror messages 2015-04-09 11:18:58 new packages uploaded to mirror/repo/arch 2015-04-09 11:19:19 possibly error messages if upload failed 2015-04-09 11:20:47 a solution could be a dump/terminate msgs to /proc like filesystem 2015-04-09 11:21:07 then sync that filesystem within systems consuming it 2015-04-09 11:21:48 that is not an issue yet 2015-04-09 11:22:10 i think a plan for what mqtt topics and what message contains 2015-04-09 11:22:13 is first step 2015-04-09 11:22:48 avoid call of exec from direct msg read 2015-04-09 11:22:58 but use the terminated msgs 2015-04-09 11:23:51 direct msgs read/exec is prone to qos/net/persistant/stale... issues 2015-04-09 11:24:11 and timestamps to msgs 2015-04-09 11:24:18 add* 2015-04-09 11:24:48 mqtt has qos features 2015-04-09 11:24:50 stale msgs would have same timestamps 2015-04-09 11:25:02 mosquitto has 2015-04-09 11:25:05 upto 2 2015-04-09 11:25:31 but really 2015-04-09 11:25:55 first problem is what messages do we publish where (what topoic) 2015-04-09 11:27:13 / // 2015-04-09 11:27:37 / // .... 2015-04-09 11:28:17 / /|/ 2015-04-09 11:28:34 example 2015-04-09 11:28:57 edge broken currently? 2015-04-09 11:29:24 i think we got the carnage fixed. edge-x86_64 finished just. and edge-x86 is building still. 2015-04-09 11:29:44 what was wrong? 2015-04-09 11:29:49 build/build-edge-x86_64 2015-04-09::16:56:17/git:update 2015-04-09 11:29:57 recent xattr changes 2015-04-09 11:29:58 class: git, git event type: commit, git repo: aports, git branch: master 2015-04-09 11:30:37 build/build-edge-x86_64 2015-04-09::16:56:17/git:update is wrong 2015-04-09 11:30:51 ERROR: unsatisfiable constraints: 2015-04-09 11:30:51 libiconv-dev (missing): 2015-04-09 11:30:51 required by: gettext-dev-0.19.4-r0[libiconv-dev] glib-dev-2.44.0-r0[libiconv-dev] 2015-04-09 11:31:04 clandmeter: apk upgrade? 2015-04-09 11:31:12 did that already 2015-04-09 11:32:01 ah 2015-04-09 11:32:11 i know 2015-04-09 11:32:13 my fault sorry 2015-04-09 11:33:17 i'm on it 2015-04-09 11:33:23 i see it 2015-04-09 11:33:26 you didnt bump it 2015-04-09 11:33:36 correct 2015-04-09 11:33:47 you are really on top of the list now 2015-04-09 11:33:50 i should have bumped it whenever it was in depends_dev 2015-04-09 11:34:02 you passed barthalion now :) 2015-04-09 11:34:12 your hate list? :) 2015-04-09 11:34:30 the list of who forgot to bump and would write a script to detect it list. 2015-04-09 11:34:49 I even list list to much 2015-04-09 11:34:59 s/to/too 2015-04-09 11:34:59 lol 2015-04-09 11:35:00 :) 2015-04-09 11:35:15 hey, when was the last time I forgot to bump something? :P 2015-04-09 11:35:15 clandmeter: youre on top of the list-list :) 2015-04-09 11:35:41 Python3 package works like a charm now! Thanks 2015-04-09 11:35:51 good! 2015-04-09 11:36:04 frol_: you should thank yourself, you fixed it :) 2015-04-09 11:36:11 ACTION would like to write one correct english sentence for ones. 2015-04-09 11:36:49 lol 2015-04-09 11:37:27 whopps 2015-04-09 11:37:30 i had a queue 2015-04-09 11:37:49 lets see what i broke now 2015-04-09 11:37:57 lol 2015-04-09 11:40:20 \o/ 2015-04-09 11:42:12 algitbot: prio gettext! 2015-04-09 11:46:06 something is slow 2015-04-09 11:46:14 build boxes are slow 2015-04-09 11:46:29 probably tired of all the work 2015-04-09 11:46:50 lol 2015-04-09 11:46:59 enjoying the weather probably 2015-04-09 11:47:28 wishing they could enjoy the weather i suppose 2015-04-09 11:48:02 we should never have allowed the weather report http traffic to the build servers! 2015-04-09 11:48:27 thats why the server room does not have windows 2015-04-09 11:48:38 lol 2015-04-09 11:48:52 not even in kvm? :) 2015-04-09 11:51:14 hmm.. what happened to earlier updates... not picking up stfl !/ 2015-04-09 11:52:00 vkrishn: build servers are banging their heads into broken aports 2015-04-09 11:52:24 ok 2015-04-09 12:03:10 ncopa, vlc failed? 2015-04-09 12:03:22 yup 2015-04-09 12:03:28 fix already pushed 2015-04-09 12:03:37 main/sdl: rebuild to fix -dev's depend 2015-04-09 12:05:22 ncopa: why is it so slow? 2015-04-09 12:05:56 who? me? 2015-04-09 12:06:34 :) 2015-04-09 12:06:38 the builder 2015-04-09 12:06:47 i think its a bug 2015-04-09 12:06:51 it chokes 2015-04-09 12:06:55 and exits 2015-04-09 12:06:59 ttdev is also building 2015-04-09 12:07:09 oh, just finished 2015-04-09 12:07:57 if a fix is pushed while running a build that fails, it will not pick up the fix til next run 2015-04-09 12:08:54 and i thought that i'd fix testing while at it 2015-04-09 12:10:46 ok ill build gettext local for now. 2015-04-09 12:31:31 clandmeter: i think you'll have your gettext now... 2015-04-09 12:41:54 ncopa: thanks :) 2015-04-09 18:54:26 and there is, https://www.gnu.org/software/gnuzilla/ for firefox esr alt 2015-04-09 18:54:55 same issues/error with newsbeuter 2015-04-09 18:55:28 snownews seems nice but could not get to add some feed urls 2015-04-09 18:57:06 such a shame that Icecat is developing by one person, probably. 2015-04-09 18:57:32 yep. 2015-04-09 19:45:13 new GnuTLS release: Version 3.4.0. 2015-04-09 19:49:37 ;), looks heavy for AL 3.2 2015-04-09 19:50:25 and, QEMU 2.3.0-rc2 Released 2015-04-09 20:23:37 vkrishn, did you get my message that newsbeuter fails because it indirectly uses '//TRANSLIT' with iconv 2015-04-09 20:32:07 not strace, but last few changes were done in aports , maybe that fixes it 2015-04-09 20:36:02 not in strace* 2015-04-09 21:13:03 what would be alt for, iconv_t cd = ::iconv_open((tocode + "//TRANSLIT").c_str(), fromcode.c_str()); 2015-04-09 21:13:20 maybe, 2015-04-09 21:13:21 #if !(__linux) && !defined(__GLIBC__) && !defined(__APPLE__) && !defined(__OpenBSD__) 2015-04-09 21:13:33 or something 2015-04-09 21:16:28 musl has no alternative i think 2015-04-09 21:17:29 first stfl should have some saner error handling such that stfl_ipool_create returns 0 when the charset is not supported in iconv. 2015-04-09 21:18:32 more better would be upgrade snownews to handle newer rss specs 2015-04-09 21:18:34 then newsbeuter could easily fall back to not using //TRANSLIT when not supported 2015-04-09 21:19:01 if tocode is utf8 then translit is not needed, in musl everything can be represented as unicode codepoints and that can be encoded as utf8 2015-04-09 21:19:49 the only other tocode supported is probably ascii in which case you can use your own heuristics after encoding the input into utf8 2015-04-09 21:20:07 (ascii transliteration can only be done heuristically) 2015-04-09 21:21:30 it's used for nl_langinfo(CODESET) to WCHAR_T 2015-04-09 21:22:51 https://github.com/akrennmair/newsbeuter/blob/master/src/stflpp.cpp 2015-04-09 21:23:10 i see, wchar_t is supported too and there translit does not matter 2015-04-09 21:23:25 http://svn.clifford.at/stfl/trunk/iconv.c 2015-04-09 21:24:00 ascii and utf16/ucs2 are the supported cases where translit matters 2015-04-09 21:26:06 nl_langinfo(CODESET) is always utf-8 now 2015-04-09 21:26:20 on musl... 2015-04-09 21:26:27 so skipping the translit for musl is safe 2015-04-09 21:27:14 maybe musl could recognize UTF-8//TRANSLIT as utf8 2015-04-09 21:32:52 so it should work , If I just remove/comment any //TRANSLIT refs 2015-04-09 21:33:35 for musl that's ok 2015-04-09 21:38:01 ok, thanks, would give a try and maybe generate a patch 2015-04-09 21:40:54 php 7 should be fun, the benchmarks looks nice! 2015-04-09 22:23:52 yabaa, dabaa, doo 2015-04-09 22:23:57 it works :) 2015-04-09 22:24:58 though the font spacing/column formatting seems not ok 2015-04-09 23:00:31 http://sprunge.us/RjVR <- patch for stfl, make symlink 2015-04-09 23:01:55 http://sprunge.us/DBHe <- patch for newsbeuter, translit fixes 2015-04-09 23:03:50 oop, resending, cgit subject not ok 2015-04-09 23:14:28 http://sprunge.us/gITS <- patch for stfl, make symlink (NEW) 2015-04-09 23:14:43 http://sprunge.us/KSEG <- patch for newsbeuter, translit fixes (NEW) 2015-04-09 23:27:53 also works, tested basic usage, if possible move to main/ pls 2015-04-10 04:03:29 <_mjones> find . -name config.sub | while read f; do if ! ./$f armv6-alpine-linux-muslgnueabihf 2>/dev/null; then 2015-04-10 05:42:03 hmm 2015-04-10 07:17:06 ncopa, tested newsbeuter basic usage, if possible move to main/ pls after applying //TRANSLIT patch 2015-04-10 07:21:54 translit patch? 2015-04-10 07:22:26 http://sprunge.us/gITS <- patch for stfl, make symlink (NEW) 2015-04-10 07:22:31 http://sprunge.us/KSEG <- patch for newsbeuter, translit fixes (NEW) 2015-04-10 07:23:58 doughdemon and nsz guided and suggested the changes 2015-04-10 07:30:11 looks good 2015-04-10 07:30:14 i can apply those 2015-04-10 07:30:27 can you report translit issue upstream? 2015-04-10 07:30:30 and send patch upstream 2015-04-10 07:31:23 its musl related, maybe a if...endif wrap can be done upstream 2015-04-10 07:31:29 ok would do 2015-04-10 07:31:59 not send patch but report, issue in AL 2015-04-10 07:33:58 #if (__linux) && !defined(__GLIBC__) 2015-04-10 07:35:17 or, musl could recognize UTF-8//TRANSLIT as utf8, till it implements it 2015-04-10 07:48:19 done, https://github.com/akrennmair/newsbeuter/issues/174 2015-04-10 07:58:30 should I resend patch using, #if (__linux) && !defined(__GLIBC__) ? 2015-04-10 08:00:15 hmmhuh 2015-04-10 08:00:24 kernel headers likely 2015-04-10 08:09:13 >>> ERROR: qemu-img*: Please create a subpackage for qemu-system-tricore 2015-04-10 08:09:28 fabled: ↑ 2015-04-10 08:10:56 ncopa: do you happen to keep your outdated packages script on git? 2015-04-10 08:37:27 barthalion: i think not 2015-04-10 08:37:32 probably because its ugly 2015-04-10 08:37:54 i think i want a better solution for that problem 2015-04-10 08:38:06 but its related to other general abuild/APKBUILD problems 2015-04-10 08:38:25 and i think the real solution is replace APKBUILD format 2015-04-10 08:38:35 but its a big thing 2015-04-10 08:38:38 I would like to take a look though, you won't beat my dirty hacks 2015-04-10 08:38:47 trust me 2015-04-10 08:39:01 do you have your dirty hacks in git? 2015-04-10 08:40:13 i have it in git: http://git.alpinelinux.org/cgit/ncopa/upstream-monitor/ 2015-04-10 08:40:19 I just parse pacman -Ss output and compare matching pkgnames to APKINDEX 2015-04-10 08:40:28 so nothing interesting to show 2015-04-10 08:40:30 thanks 2015-04-10 08:40:40 my local git checkout is dirty though 2015-04-10 08:40:59 $ git diff | sprunge 2015-04-10 08:40:59 http://sprunge.us/MSXe 2015-04-10 08:41:17 i think we should have some hook in APKBUILD itself 2015-04-10 08:48:10 hum 2015-04-10 08:48:27 it smells that lua-posix broke the API (again) 2015-04-10 08:48:33 http://sprunge.us/eXRH 2015-04-10 08:48:54 I wonder if there is any sense in keeping flashcache 2015-04-10 08:49:03 i wonder that too 2015-04-10 08:49:14 you can achieve the same thing thing bcache or lvm iirc 2015-04-10 08:49:23 btw, i am about to add another 3rd party kernel driver 2015-04-10 08:49:37 nslink, devicemaster 2015-04-10 08:50:01 i think i might even build a separate iso for it 2015-04-10 08:50:32 oh 2015-04-10 08:50:45 might not be needed due to the update kernel script thingy 2015-04-10 08:51:42 ncopa: any plans to upgrade kernel to later version? 2015-04-10 08:53:50 csync2 New upstream release 2.0 2015-04-10 08:56:50 I downloaded libreoffice git bundles, then converted them to squashfs and wrote to dvd 2015-04-10 08:56:54 now I can update the dvd by aufs ;) 2015-04-10 08:58:30 I figured out the question to dev who want to create their own distro, make the first pkg libreoffice :-)) 2015-04-10 08:59:06 its volumnous, covers almost all usable pkgs of a distro 2015-04-10 08:59:23 i want to have new kernel to support overlayfs. 2015-04-10 09:00:22 clandmeter: we want that too 2015-04-10 09:00:29 problem is maintenance and grsecurity 2015-04-10 09:01:28 ncopa: what about vanilla? 2015-04-10 09:01:34 you want to keep it in sync with grsec? 2015-04-10 09:01:40 i'd prefer that yes 2015-04-10 09:01:44 however 2015-04-10 09:01:58 I think we can go with upstream with vanilla 2015-04-10 09:02:00 i wonder if we should have a separate linux-lts_grsec kernel 2015-04-10 09:02:27 we coudl upgrade vanilla kernel to newet upstream 2015-04-10 09:02:29 however 2015-04-10 09:02:44 there is some value keep it in sync wiht grsec 2015-04-10 09:02:47 cant we just have lts in stable? 2015-04-10 09:03:20 if there are issue with grsec kernel, then we can ask: try vanila kernel 2015-04-10 09:03:23 and can compare 2015-04-10 09:03:26 true 2015-04-10 09:03:33 we lose that if they are not in sync 2015-04-10 09:03:36 reminds me 2015-04-10 09:03:46 the latest config changes needs be adjusted in vanilla kernel too 2015-04-10 09:04:06 me and fabled are interested in 3.18 kernel 2015-04-10 09:04:20 but we need find out if we can support it 2015-04-10 09:04:35 the 3.10 has not been updated for long time 2015-04-10 09:04:47 due to lack of resources 2015-04-10 09:05:05 which parts of 3.18 are of your interst? 2015-04-10 09:05:12 overlayfs :) 2015-04-10 09:05:47 yeah 2015-04-10 09:06:49 for which part will you use overlayfs? 2015-04-10 09:06:55 lxc? 2015-04-10 09:08:09 I use aufs for chroot 2015-04-10 09:15:57 with overlayfs, would be interesting to create a master bootable src dvd for AL basic-dev iso 2015-04-10 09:16:19 just insert, boot, and start compiling ;) 2015-04-10 09:36:19 clandmeter, you around? 2015-04-10 09:36:57 sure 2015-04-10 09:37:18 what do you think about package kodi ? 2015-04-10 09:37:29 funny you ask 2015-04-10 09:37:39 i tried yesterday to bump xbmc 2015-04-10 09:37:49 but got issues with fority i think 2015-04-10 09:40:07 also ffmpeg can be a pita 2015-04-10 09:41:30 lol 2015-04-10 09:41:34 that's interesting 2015-04-10 09:41:43 since yesterday /me also worked on it 2015-04-10 09:41:47 and i got stucket do ffmpeg 2015-04-10 09:41:55 kodi tries to compile his own ffmpeg 2015-04-10 09:42:13 there's a patch that enable external ffempg, but does not apply to kodi 2015-04-10 09:42:36 guess who most likely wrote that patch… 2015-04-10 09:42:47 also, in configure you can choose to enable ffmpeg as shared, auto or force 2015-04-10 09:42:49 barthalian :) 2015-04-10 09:43:01 barthalion, so you can redo it ;) 2015-04-10 09:43:29 since i was sick at home, i complied shairplay to have airplay mirroring 2015-04-10 09:43:38 there is a reason xbmc has internal ffmpeg 2015-04-10 09:43:39 but ive not yet pushed it 2015-04-10 09:43:46 err kodi 2015-04-10 09:43:53 inernal ffmpeg does not build? 2015-04-10 09:44:00 clandmeter, exactly. ncopa, no 2015-04-10 09:45:41 yes, upstream siliness 2015-04-10 09:45:59 maybe projects that depends on ffmpeg have that issue 2015-04-10 09:46:11 s/maybe/many 2015-04-10 09:46:26 if they insist on using own outdated ffmpeg, they are either stupid or doing something ugly in the code 2015-04-10 09:46:39 barthalion, +1 2015-04-10 09:46:40 especially after 2.0 release 2015-04-10 09:46:44 probably both 2015-04-10 09:49:32 http://forum.kodi.tv/showthread.php?tid=117188 2015-04-10 09:49:35 Even if it is old 2015-04-10 09:49:46 "Note that XBMC has patches to ffmpeg which you may not have on the external version." 2015-04-10 09:49:55 this is a not of one of the developers 2015-04-10 09:49:58 FernetMenta 2015-04-10 09:50:04 which is a liquor, inItaly... 2015-04-10 09:50:22 this might explain why they do this things...maybe are drung 2015-04-10 09:50:25 *drunk 2015-04-10 09:51:23 clandmeter: the ca-certificates issue 2015-04-10 09:51:33 how do i reproduce it? 2015-04-10 09:51:52 if in install ca-certificates in a new root 2015-04-10 09:51:58 where should the certs be? 2015-04-10 09:53:09 just run th esciprt? 2015-04-10 09:53:39 /etc/ssl/certs 2015-04-10 09:53:56 found it yes 2015-04-10 09:54:04 sudo /usr/sbin/update-ca-certificates --fresh |sprunge 2015-04-10 09:54:14 you can remove sprunge :) 2015-04-10 09:54:37 ncopa: seems edge still has issues with libiconv 2015-04-10 09:55:08 ok i have fix for update-ca-certs 2015-04-10 09:55:51 -posix.chmod(tmpfile, 0644) 2015-04-10 09:55:51 +posix.chmod(tmpfile, "rw-r--r--") 2015-04-10 09:55:59 that looks correct right? 2015-04-10 09:56:41 looks like original mode was broke 2015-04-10 10:18:14 hmm 2015-04-10 10:18:20 xattrs do not seem to be fully working 2015-04-10 10:24:24 hmm.. looks like firefox and java are not right 2015-04-10 13:39:18 "AL Drive" seems a nice idea as a product launch in next CES ! 2015-04-10 13:40:30 any thoughts ! or any VC 2015-04-10 13:49:17 hum 2015-04-10 13:49:34 the update-kernel script does not work as expected on my wandboard 2015-04-10 13:50:07 humpf 2015-04-10 13:50:11 and apk segfaults 2015-04-10 13:50:12 sigh 2015-04-10 14:48:03 ACTION is having problems with apk 2015-04-10 14:48:32 created an updated freeswitch package, scp'd it to another box, and tried to install with 'apk add -f --allow-untrusted' 2015-04-10 14:48:40 nothing happens, just says OK 2015-04-10 14:48:54 been some recent changes 2015-04-10 14:49:00 yesterday, it was worse 2015-04-10 14:49:17 edge box was giving blank lines instead of proper output 2015-04-10 14:49:39 it segfaults for me on arm 2015-04-10 14:50:24 ok, so consider this a heads up to check the '-f --allow-untrusted' functionality while you're in there :) 2015-04-10 15:30:00 I have read is some backlog that a consul package is on his way. count me in as avid tester (even better with consul-template) 2015-04-10 19:47:13 ncopa: i'm a bit swamped these days i'll try to find some time to test mkinitfs next week 2015-04-12 00:26:34 Newer pkgs: 2015-04-12 00:26:34 CLooG 0.18.3 2015-04-12 00:26:34 http://repo.or.cz/w/isl.git - isl-0.14.1 2015-04-12 00:26:34 mpc1 - mpc-1.0.3.tar.gz 2015-04-12 04:52:17 more thoughts on mqtt msgs format, 2015-04-12 04:52:19 / / /<[where|action|data|msg]> 2015-04-12 15:08:42 hi, I'd like to make a image for a Cubie Board 2 ( A20 (sun7i) ), where can I find some docs that could guide me ? 2015-04-12 15:11:17 Cubie BOard is like a RaspberryPi 2015-04-12 15:13:15 its chip (SoC) is a Dual-Core Cortex-A7 ARM CPU http://linux-sunxi.org/A20 2015-04-13 03:03:10 what shell runs abuild? busybox's ash? (probably a silly question, i'm just so used to bashisms at this point that it might be awkward to write a complex APKBUILD for me) 2015-04-13 04:24:46 ttilley: correct 2015-04-13 06:03:37 morning 2015-04-13 06:03:57 fabled: apk segfault on armhf: https://gist.github.com/ncopa/368034eaba8dd8ce49e9 2015-04-13 06:05:10 i wonder if it is related the recent memmove change 2015-04-13 06:05:46 no 2015-04-13 06:06:07 it's something else. i look at it in a moment. 2015-04-13 06:06:39 i mean, it could be something broken in musl, arm 2015-04-13 06:43:55 ncopa, ^^^ fixes your errors 2015-04-13 06:44:17 and i need to fix dup.pw 2015-04-13 07:30:58 morning 2015-04-13 07:56:45 ncopa: you about? 2015-04-13 07:57:36 ScrumpyJack, morning. he was here earlier, but is probably now away for awhile. 2015-04-13 08:07:07 isp down again... 2015-04-13 08:07:31 nice when they eat their own dogg food. 2015-04-13 08:07:40 impossible to reach their support desk 2015-04-13 08:07:55 you mean your isp? 2015-04-13 08:08:09 (alpinelinux site unreachable for me) 2015-04-13 08:08:24 (so i guess you mean hosting provider?) 2015-04-13 08:08:50 yes 2015-04-13 08:08:55 our infra is down now aswel 2015-04-13 08:09:15 what is the infra? 2015-04-13 08:09:29 build box etc? 2015-04-13 08:09:29 nl.a.o 2015-04-13 08:09:36 bugs 2015-04-13 08:09:39 pkgs 2015-04-13 08:09:42 rsync 2015-04-13 08:09:47 ok 2015-04-13 08:09:53 www 2015-04-13 08:09:58 msg.a.o 2015-04-13 08:10:05 perhaps we should mirror all that 2015-04-13 08:10:43 high availablity is not on top of our list atm :) 2015-04-13 08:11:08 if this isp continius like this, ill replace it. 2015-04-13 08:25:22 still got availablility problems? 2015-04-13 08:25:27 ah, read it 2015-04-13 08:26:27 just use a coreos cluster... *scnr* 2015-04-13 10:14:21 ok 2015-04-13 10:14:27 finally getting a look at new u-boot and rpi2 2015-04-13 10:14:41 seems that new u-boot should be mostly usable on all arm targets out-of-box 2015-04-13 10:14:59 trying to use that. and make the u-boot image also support rpi + rpi2 2015-04-13 10:15:19 rpi still needs separate kernel; but it should be possible to have one kernel supporting both with separate dtbs 2015-04-13 10:49:33 fabled: whats the adv of using uboot for rpi? 2015-04-13 10:49:51 hopefully we can have single image for all models 2015-04-13 10:50:19 rpi will auto load uboot? 2015-04-13 10:50:29 but seems the tree does no support building support for rpi+rpi2 enabled kernel :/ 2015-04-13 10:50:38 rpi boot sequence would be then: 2015-04-13 10:50:44 videocore -> uboot -> kernel 2015-04-13 10:50:57 uboot would load kernel+initramfs+dtb 2015-04-13 10:51:04 so uboot is kernel.img? 2015-04-13 10:51:05 dtb being the only difference depending on model 2015-04-13 10:51:07 yes 2015-04-13 10:51:10 that would be the basic idea 2015-04-13 10:52:08 so you add symlinks for each device to uboot (or copy it)? 2015-04-13 10:52:25 the boot media would have uboot specific for each model 2015-04-13 10:52:43 and there would need to be a script, or some manual step to select which image is loaded 2015-04-13 10:53:06 for rpi you get two uboots: rpi (model 1) and rpi_2 (for model b 2) 2015-04-13 10:53:24 separate uboots for different arm boards 2015-04-13 10:53:35 but the u-boot image will auto detect which dtb to use 2015-04-13 10:53:54 so for e.g. rpi1 model B and B+ would be using the same u-boot, which loads corresponding dtb automatically 2015-04-13 11:06:43 hmmh... seems rpi2 currently would need it's own kernel config 2015-04-13 11:11:09 ugh 2015-04-13 11:13:31 yeah. it's a mess still. seems. 2015-04-13 13:08:33 where did armhf linux-grsec.apk go? 2015-04-13 13:08:44 looks like its missing: http://nl.alpinelinux.org/alpine/edge/main/armhf/ 2015-04-13 13:09:28 build-edge-armhf:~/aports/main/linux-grsec$ ls ~/packages/main/armhf/linux-* | s 2015-04-13 13:09:28 prunge 2015-04-13 13:09:28 http://sprunge.us/ZZOH 2015-04-13 13:11:55 build failed? 2015-04-13 13:12:59 gmp failed earlier 2015-04-13 13:13:26 ah 2015-04-13 13:13:34 likely temporary network issues at the build attempt time 2015-04-13 13:13:46 i think it was set to continue on build failure 2015-04-13 13:13:50 and skip pre\vious fails 2015-04-13 13:13:58 yes 2015-04-13 13:14:15 i changed that now 2015-04-13 13:14:15 i'm on it 2015-04-13 13:14:20 ok 2015-04-13 13:14:27 arm has still build fails 2015-04-13 13:14:32 please don't make it abort 2015-04-13 13:14:37 ok 2015-04-13 13:14:51 i just did, i'll restore it then 2015-04-13 13:15:29 restored 2015-04-13 13:16:32 maybe thats why update-kernel didnt update to newer kernel for me 2015-04-13 13:16:44 the newer was not yet built 2015-04-13 13:16:49 quite likely yes 2015-04-13 17:24:56 ACTION is still having problems running 'apk add -f --allow-untrusted' on packages that I build 2015-04-13 17:25:10 apk doesn't do any installation, but packages show up in world 2015-04-13 17:25:32 running edge and building on a (different) edge box 2015-04-13 17:25:47 fabled, ncopa: ^^^ 2015-04-13 18:30:53 tdtrask: did you try to specify full path? i remember that as an issue before. 2015-04-13 18:34:52 I'll try now 2015-04-13 18:35:19 have to spin the machine back up 2015-04-13 18:39:03 clandmeter: no difference :( 2015-04-13 18:41:58 tdtrask: weird 2015-04-13 18:45:32 another issue for ncopa: 2015-04-13 18:45:32 ACF doesn't install proper packages on edge now that you modified the lua packages to install only the correct version 2015-04-13 18:45:59 haserl is set up to Lua 5.2 by default, but default Lua is 5.1 2015-04-13 18:46:00 so, we end up using Lua 5.2 (built into haserl) and only have Lua 5.1 packages installed 2015-04-13 18:48:48 'apk add lua5.2' fixes the issue 2015-04-13 18:49:01 can we change the default Lua version to 5.2 for alpine 3.2? 2015-04-13 19:17:50 ACTION created #4064, #4065, and #4066 so ncopa can track the issues 2015-04-14 09:08:51 clandmeter, fyi: http://sprunge.us/iNJi 2015-04-14 09:08:55 this is kodi APKBUILD 2015-04-14 09:09:32 now i'm stuckt to the error: cannot find _G_va_list 2015-04-14 09:09:39 which is a glibc function 2015-04-14 09:09:47 this need a patch 2015-04-14 09:11:54 and i'm not able to fix it.. 2015-04-14 09:17:02 fcolista: 2015-04-14 09:17:04 https://github.com/xbmc/xbmc/blob/master/xbmc/cores/DllLoader/exports/wrapper.c#L461 2015-04-14 09:17:05 #if _FORTIFY_SOURCE > 1 2015-04-14 09:17:05 change that to: 2015-04-14 09:17:05 #if defined(__GLIBC__) && _FORTIFY_SOURCE > 1 2015-04-14 09:17:37 gr8! 2015-04-14 09:19:23 ncopa: that looks like some insane code in wrapper.c 2015-04-14 09:19:28 I don't want to know :P 2015-04-14 09:19:37 same... :) 2015-04-14 09:21:09 ACTION has the patch, going to apply it 2015-04-14 09:21:39 ffmpeg issue can be solved by adding x264-dev and enabling --enable-ffmpeg=shared 2015-04-14 09:24:05 nice 2015-04-14 09:26:36 upstream has a flag to enable shared ffmpeg? 2015-04-14 09:26:41 sounds too god to be real 2015-04-14 09:26:46 yes 2015-04-14 09:26:52 good even 2015-04-14 09:26:55 finally, i would say... 2015-04-14 09:27:08 I bet they don't support such builds anyway 2015-04-14 09:27:23 but even if you enable it, if he doesn't try x264 he tries to compile ffmpeg 2015-04-14 09:27:32 i've found the solution on AUR post... 2015-04-14 09:32:17 ok. Another error: htsq.h:12:23: fatal error: sys/queue.h: No such file or directory 2015-04-14 09:32:18 I suppose is included in bsd-compact-headers. 2015-04-14 09:36:09 >>> linux-grsec*: Package size: 126.4 MB 2015-04-14 09:36:11 its huge 2015-04-14 09:36:17 i'd like to split it somehow 2015-04-14 09:36:27 localhost:~# uname -a 2015-04-14 09:36:27 Linux localhost 3.18.11-0-grsec #1-Alpine SMP Tue Apr 14 09:16:03 GMT 2015 x86_64 Linux 2015-04-14 09:51:57 Ok, this is the error i'm hitting now: http://pastebin.com/ia1XWYyC 2015-04-14 09:55:21 looks that this kind of declaration "struct FILEWRAP : public FILE" is not right 2015-04-14 09:55:43 fcolista, correct, FILE is opaque, and cannot be inherited from 2015-04-14 09:55:47 well, my build env does not like 2015-04-14 09:55:53 it works in glibc, but not musl 2015-04-14 09:56:08 I had this suspect :) 2015-04-14 09:56:15 glibc exposes FILE; but it might mean things break when glibc updates it 2015-04-14 09:56:26 so this need another patch to make kodi complile 2015-04-14 09:57:54 yes 2015-04-14 11:29:54 I'm working on awall for archlinux (i'm quite tired with shorewall). awall's makefile uses "install -d" command to create /usr/sbin and /var/run directory. With busybox install it works if dirs exist, while with coreutil's install fails if dirs exist 2015-04-14 11:31:35 umh. Both are actually symlinks 2015-04-14 11:31:42 that's why it fails then 2015-04-14 12:11:42 nsz: binutils 2.25-r1 should have sysroot as requested 2015-04-14 12:11:43 thanks 2015-04-14 12:40:16 Hi guys. It's me again. Today I want to have mono on Alpine Linux. Is that possible? 2015-04-14 12:40:26 nice 2015-04-14 12:40:55 frol_: I don't see why it wouldn't be 2015-04-14 12:41:05 I think buildroot had some minor patches for it 2015-04-14 12:49:42 Is there a shortcut to create an apk package from .mk? http://git.buildroot.net/buildroot/tree/package/mono/mono.mk 2015-04-14 12:57:26 other than rewriting it into APKBUILD format, no 2015-04-14 13:26:49 ok i think i will push the 3.18 kernel ++ now 2015-04-14 13:32:42 I have another unrelated question. Does anybody know about `fcloseall` in glibc? It seems there is no such function in musl libc. (undefined reference error is raised by gcc if I try to compile a file with this function) Do you have any suggestions about how to solve this? 2015-04-14 13:36:30 frol_: This function is a GNU extension. 2015-04-14 13:56:19 AL does not get this grsecurity-3.1-3.18.11-201504051405-alpine.patch from upstream ? 2015-04-14 13:58:29 no :) 2015-04-14 13:58:47 vkrishn: do you mind if we purge flashcache kernel module? 2015-04-14 13:59:44 hmm... ok, but did I request it ;) 2015-04-14 14:00:04 so the question is: do you still use it 2015-04-14 14:00:16 i actually made it compile at least 2015-04-14 14:00:21 but i dont know if it works 2015-04-14 14:00:52 no, I think I wanted to test bcache, dm-cache.. btrfs.. zfs 2015-04-14 14:01:50 bcache works fine 2015-04-14 14:01:54 ok 2015-04-14 14:02:04 at least in Arch 2015-04-14 14:02:09 set it up yesterday 2015-04-14 14:02:43 :) 2015-04-14 14:07:29 ncopa, phalcon v2 would hit the shelves on apr 17 ;), when is dday for v3.2RC tag ? 2015-04-14 14:08:11 also newsbeuter works kinda ok, can it be moved to mains pls ? 2015-04-14 14:08:28 define "kinda" 2015-04-14 14:08:49 on console the text columns are not aligned 2015-04-14 14:09:05 I think its a bug in musl/iconv 2015-04-14 14:09:06 i was hoping to get arm builder up this week 2015-04-14 14:09:39 then we need some day to fix all build problems 2015-04-14 14:09:42 some days 2015-04-14 14:10:00 so rc will be out next week, best case 2015-04-14 14:10:07 ok 2015-04-14 14:10:07 more likely the week after that 2015-04-14 14:21:27 http://bstring.sourceforge.net/features.html , could be old page but nice 2015-04-14 14:34:03 +CONFIG_OVERLAY_FS=m :-)) 2015-04-14 14:35:38 # CONFIG_LUSTRE_FS is not set 2015-04-14 14:36:09 was reading about it http://opensfs.org/ 2015-04-14 15:18:05 which .patch format does abuild accept ? i have this error http://sprunge.us/MfSS 2015-04-14 15:32:54 ok this format worked diff -urpN oscam-0_svn20150414-orig/module-dvbapi.c oscam-0_svn20150414/module-dvbapi.c > ../patch1.patch 2015-04-14 18:39:33 # supervisord -h 2015-04-14 18:39:33 Error relocating /usr/lib/libpython2.7.so.1.0: __flt_rounds: symbol not found 2015-04-14 18:39:40 is this a general problem? 2015-04-14 18:39:59 or just docker related? 2015-04-14 18:40:51 supervisord works for me 2015-04-14 18:40:54 im on edge though 2015-04-14 18:41:02 me too 2015-04-14 18:41:12 i tried to start it within docker 2015-04-14 18:41:36 update musl 2015-04-14 18:41:56 __flt_rounds is fixed in 1.1.8 2015-04-14 18:42:48 thx 2015-04-14 19:16:18 linux 3.18 \o/ 2015-04-14 19:17:24 /o\ 2015-04-14 19:17:39 -o- 2015-04-14 19:18:02 –o– 2015-04-14 19:18:53 anyone tried kde on alpine? 2015-04-15 01:32:45 ncopa: i don't particularly favor mailing lists as a means of patch submission, so here's llvm 3.6.0 with clang, compiler-rt, and additional fixes: http://sprunge.us/QSBA 2015-04-15 01:34:02 i broke the patches into as many smaller pieces as i could 2015-04-15 01:35:56 instead of one large "llvm-musl-alpine.patch", it has split out: adding the code for triples, the text of the triples for matching, alpine as a distro, some defaults when compiling on alpine (-z relro, --hash-style=gnu), defaulting to compiler-rt as a runtime, using musl dynamic linker paths, etc 2015-04-15 01:36:15 all as individually reviewable patches 2015-04-15 01:37:15 16 patches in total 2015-04-15 01:39:04 i would have added support for clang-tools-extra but essentially all the tests for clang-tidy fail. no idea why. 2015-04-15 01:51:32 ...really wish you guys had a bugzilla or something 2015-04-15 01:53:43 http://bugs.alpinelinux.org/ ? 2015-04-15 01:54:39 ha 2015-04-15 01:54:54 the wiki says to send everything to the mailing list and never mentions a bugzilla 2015-04-15 02:00:14 mailing list is probably preferred 2015-04-15 04:18:06 if glusterfs, ceph, lustre not making to v3.2, http://sheepdog.github.io/sheepdog/ seems easier, pls have a look 2015-04-15 06:11:19 hmm 2015-04-15 06:11:27 asterisk fails on arm due to missing cap_* symbols 2015-04-15 06:11:29 sounds link issue 2015-04-15 07:29:29 ttilley: why not just upgrdea llvm in main? 2015-04-15 07:59:29 ncopa: did you talk to fabled about apk license field? 2015-04-15 08:00:00 i think i just edited abuild 2015-04-15 08:00:22 because i think it is a bug in abuild 2015-04-15 08:00:34 apk does not treat license as a list (array) 2015-04-15 08:00:42 it is simply a text string 2015-04-15 08:01:13 i think the origina idea of treating it as array was that we could share the license files 2015-04-15 08:01:20 so we dont need N copies of GPL license 2015-04-15 08:03:03 yeah i thought it was a list from most apkbuilds 2015-04-15 08:03:10 but now i see qt5 and it has a sting 2015-04-15 08:03:12 string 2015-04-15 08:24:31 clandmeter, kodi's packaging is stuck because of this error: http://pastebin.com/Gg6H1jus. 2015-04-15 08:24:39 This need a patch that i'm not able to do. 2015-04-15 08:24:41 FYI 2015-04-15 08:28:54 fcolista: do you have an apkbuild i can try? 2015-04-15 08:29:03 sure 2015-04-15 08:29:05 one sec 2015-04-15 08:29:25 http://sprunge.us/MdbI 2015-04-15 08:29:54 i've added shairplay package that is not yet committed in the official aports 2015-04-15 08:29:58 so you can remove it 2015-04-15 08:30:31 look at the patch, also 2015-04-15 08:30:42 http://sprunge.us/jgRg 2015-04-15 08:30:50 this is fix-wrapper-c.patch 2015-04-15 08:31:03 the others are from main/xbmc package 2015-04-15 08:38:10 fcolista: why -j2? 2015-04-15 08:38:29 i have two cpus, feelfree to remove it 2015-04-15 08:38:47 you dont set them in abuild.conf? 2015-04-15 08:39:26 it was on another buildenv i have @home and i worked on it while i was sick, i didn't do fine tuning 2015-04-15 08:40:07 ok ill see if i can get any further with some help. 2015-04-15 08:41:54 xbmc feels like a very old building with lots of plasterwork located somewhere in venice. 2015-04-15 08:50:46 fcolista: im getting a different error with ffmpeg 2015-04-15 08:51:05 fcolista: http://sprunge.us/MAcg 2015-04-15 08:51:58 I didn't stumble on that error before 2015-04-15 08:52:07 are you using my APKBUILD ? 2015-04-15 08:53:40 yeah 2015-04-15 08:54:04 i apk updated, maybe its a new ffmpeg in aports 2015-04-15 08:54:16 umh, could be 2015-04-15 08:54:23 check on the git log 2015-04-15 08:55:10 nah it still errors 2015-04-15 08:57:33 I'm building it now after update && upgrade 2015-04-15 09:09:59 ok i get to the same error you got 2015-04-15 10:46:43 ncopa: i'm preparing the musl gcc patchset for upstream reveiw and noticed that alpine has additional patches on top of gregor's ones 2015-04-15 10:47:39 i see patches for ada, gcj and a few other that may be relevant 2015-04-15 10:50:18 i think i'll try to upstream gregor's patches first, but if you know anything that's important and should be in the initial version of the upstream 'musl gcc support' then please inform me 2015-04-15 10:57:12 fcolista: \o/ 2015-04-15 11:05:28 i dont understand the changes in this patch: 2015-04-15 11:05:29 http://git.alpinelinux.org/cgit/aports/plain/main/gcc/ada-musl.patch 2015-04-15 11:05:55 is there some issue with cpu_set_t* function argument? 2015-04-15 11:09:39 i dont know 2015-04-15 11:09:45 it was fabled who did that 2015-04-15 11:10:06 i wonder if he took that from gentoo for pie stuff 2015-04-15 11:10:52 it is not impossible it was due to uclibc also 2015-04-15 11:11:02 clandmeter, you got kodi compile? 2015-04-15 11:11:10 yes 2015-04-15 11:11:15 \0/ 2015-04-15 11:11:16 cleaning up now 2015-04-15 11:11:17 How? 2015-04-15 11:11:30 with google and help from fabled :) 2015-04-15 11:11:37 eheh :) 2015-04-15 11:11:56 do you mind if i commit it? or you prefer? 2015-04-15 11:12:09 clandmeter: how does that FILE fix look? 2015-04-15 11:12:12 i'm curious 2015-04-15 11:12:20 just remove it 2015-04-15 11:12:27 :D 2015-04-15 11:12:30 :) 2015-04-15 11:12:53 falbed said it wasnt needed 2015-04-15 11:13:09 oh 2015-04-15 11:13:12 interesting 2015-04-15 11:14:24 hum 2015-04-15 11:14:30 still no arm kernel 2015-04-15 11:15:41 whoops 2015-04-15 11:15:59 the build was actually running... 2015-04-15 11:16:21 fcolista: think we need to upstream some of the patches 2015-04-15 11:16:45 they are related to musl, tough 2015-04-15 11:42:26 fcolista: ^ 2015-04-15 11:47:44 nsz, yes, including in the ada header did not work out as it had wrong feature tests; and adding proper more feature test had namespace conflicts with others 2015-04-15 11:47:51 so i ended up using void* 2015-04-15 11:48:04 IIRC 2015-04-15 11:49:43 clandmeter, \o/ 2015-04-15 11:53:42 ncopa, yeah, arm is still building kernel 2015-04-15 11:54:08 the 14:15 upload msg was from mesa/gtk+ rebuild 2015-04-15 13:02:44 i will merge linux-grsec and linux-virtgrsec APKBUILDs 2015-04-15 13:03:03 simplifies maintenance 2015-04-15 13:05:20 weren't we talking about it like 3 months ago? 2015-04-15 13:07:54 barthalion, yes, there just wasn't as strong incentive as now 2015-04-15 13:07:55 :) 2015-04-15 13:08:07 once that's done, i'll copy it for generating rpi2 kernel 2015-04-15 13:08:23 seems rpi upstream does not support build that would support rpi1+rpi2 2015-04-15 14:50:19 any volunteers to add _luaversion to main/lua-xctrl and replace the source url to point to github? 2015-04-15 14:50:45 https://github.com/yetanothergeek/xctrl/issues/3 2015-04-15 14:50:52 he was kind enough to create release tags 2015-04-15 14:51:13 and upgrade to latest release tag ofc 2015-04-15 14:55:09 ncopa, jsut doing it 2015-04-15 14:56:08 thanks 2015-04-15 14:57:25 _luaversion is the variable that allows to have both 5.1 and 5.2 right? 2015-04-15 14:59:02 yes 2015-04-15 14:59:04 this works (but not added _luaversion yet) http://sprunge.us/ZeXW 2015-04-15 14:59:07 ok 2015-04-15 14:59:12 look at: git log lua-* 2015-04-15 14:59:34 git log -p main/lua-* 2015-04-15 14:59:48 k 2015-04-15 15:00:13 the "add lua 5.3 and use install_if" 2015-04-15 15:05:33 humpf 2015-04-15 15:11:52 ncopa, http://sprunge.us/BBTK 2015-04-15 15:12:10 lua-xctrl build for 5.1,5.2 and 5.3 lua 2015-04-15 15:12:15 version 2015-04-15 15:13:09 i gotta go 2015-04-15 15:13:12 looks ok 2015-04-15 15:13:21 i will be away thurs and friday 2015-04-15 15:13:31 fcolista: if it buidls and works, push 2015-04-15 15:13:33 thanks 2015-04-15 15:25:15 "Executing gtk-update-icon-cache-2.24.25-r0.trigger" - does this update all icon caches ? 2015-04-15 15:25:40 sounds so 2015-04-15 15:26:15 ok 2015-04-15 15:26:52 (I'm updating spacefm) 2015-04-15 15:48:27 relocation error in gtk+3.0 in edge http://hastebin.com/tesefavohi.vbs 2015-04-15 15:49:02 gtk+3.0@edge (3.16.2-r0) 2015-04-15 15:50:37 how are relocation errors solved ? I have a similar error in xpra http://xpra.org/trac/ticket/820 2015-04-15 16:04:05 nm -D shows the "missing" gtk symbols 2015-04-15 16:08:52 as "U" & no hex address - GDK functions (the xpra errors) are also "U" - so these issues are probably related 2015-04-15 17:49:54 BitL0G1c: rebuild, probably 2015-04-15 17:52:14 any devs around to give input on http://bugs.alpinelinux.org/issues/4068 ? 2015-04-15 17:52:57 ttilley: can I take a look tomorrow? 2015-04-15 17:53:29 barthalion: certainly. i'm a random internet stranger, not a cop. ;) 2015-04-15 17:54:31 would there be any interest in an llvm pkg that builds itself in several stages so as to remove the dependency on gcc/libgcc_s/libstdc++? 2015-04-15 17:55:34 this is something to be discussed with guys who are ops here 2015-04-15 17:55:41 and they are all gone atm, probably 2015-04-15 17:56:14 with those patches (and a couple more for libcxx/libcxxabi, but just for build process) there's no reason why clang can't be self-hosting on musl/alpine 2015-04-15 17:56:15 ncopa is away till next week, and the rest but kaniini lives in CET 2015-04-15 17:56:32 eek 2015-04-15 17:57:09 you're from the other side of the ocean, am I correct? :) 2015-04-15 17:57:22 hi 2015-04-15 17:57:27 ttilley: i am interested in such a thing 2015-04-15 17:58:07 however, for most projects i have shifted over to freebsd because it is easier for my team to hack 2015-04-15 18:01:45 kaniini: libcxx/libcxxabi have a strange circular dependency with each other and clang that make it awkward to set up, and then clang needs to be rebuilt against the result using a previous build of clang. -_- 2015-04-15 18:02:57 if libcxxabi isn't already installed, libcxx needs to be built against libstdc++ from gcc OR libcxxrt... 2015-04-15 18:03:02 which brings up another good point 2015-04-15 18:03:09 freebsd doesn't use libcxxabi at all 2015-04-15 18:03:15 they use libcxxrt 2015-04-15 18:03:48 i'm not sure what the potential advantages are of using that in place of libcxxabi, personally 2015-04-15 18:17:31 ttilley, self hosting clang is of interest. and we might eventually make it default tool chain. 2015-04-15 18:18:11 fabled: the linux kernel still can't be compiled using clang, unfortunately 2015-04-15 18:18:55 yeah. that's related to the 'eventually' part :) 2015-04-15 18:19:14 patches keep coming in to replace gcc-isms with equivalent code... just as fast as new gcc-isms are added 2015-04-15 18:19:15 heh 2015-04-15 18:19:46 at least the linux guys aren't as llvm/clang hostile as the glibc folks 2015-04-15 18:20:26 the glibc guys have said, essentially, that they are a GNU project with a GNU compiler and they couldn't care less 2015-04-15 18:20:48 lol 2015-04-15 18:20:51 yeah 2015-04-15 18:21:34 i really like how llvm handles cross compiling 2015-04-15 18:22:01 but it's still not as optimizing, and there's the kernel compiling issue 2015-04-15 18:27:45 it looks like libcxxrt is about 1/5 the size of libcxxabi and is better supported on bsd/linux... and it supports x86/x86_64/arm, which i assume is all alpine cares about. 2015-04-15 18:34:19 nice. yes, sounds good fornow. we might add other arches later on, but for the immediate future that's our set. 2015-04-15 20:13:14 barthalion: gtk+3.0-3.16.2-r1 still has relocation errors http://hastebin.com/iwaholivew.avrasm 2015-04-15 20:27:36 something else needs rebuild then 2015-04-15 20:27:46 but I go to bed for now 2015-04-15 20:37:05 ok np 2015-04-15 23:44:21 if version numbers match, which repo does apk select? the first listed in the config? 2015-04-16 04:30:08 ttilley: the newest one iirc, unless you specify which version you want 2015-04-16 11:08:50 fabled: on edge i get: ERROR: Failed to set xattrs on usr/lib/firefox-37.0.1/firefox.apk-new: Not supported 2015-04-16 11:09:11 which kernel? 2015-04-16 11:10:16 3.14.34-0-grsec 2015-04-16 11:10:24 its the latest edge iso kernel 2015-04-16 11:16:24 clandmeter, yeah. you'd need newer kernel 2015-04-16 11:16:45 i wonder if setting xattr should be made soft failure 2015-04-16 11:22:58 fabled: so there is no iso currently that supports it? 2015-04-16 11:23:06 i also tried 3.1 vanilla iso 2015-04-16 11:23:20 i think no 2015-04-16 12:15:07 fabled: i just updated to 3.18 vanilla 2015-04-16 12:15:12 still getting the same error 2015-04-16 12:15:58 mmm 2015-04-16 12:16:01 hard-disk install? 2015-04-16 12:16:10 usb 2015-04-16 12:16:20 tmpfs 2015-04-16 12:16:45 hmm.. grsec patches xattr stuff a bit. on vanilla kernel / might need mount option user_xattr 2015-04-16 12:19:16 tmpfs: No value for mount option 'user_xattr' 2015-04-16 12:25:26 hmm 2015-04-16 12:25:41 clandmeter, maybe vanilla tmpfs does not allow xattrsthen i'll check 2015-04-16 12:27:24 alpine-iso is also broken 2015-04-16 12:27:37 so i cannot generate a newwer grsec iso 2015-04-16 12:28:00 clandmeter, yes, vanilla tmpfs does not allow user.* xattrs 2015-04-16 12:28:07 and pax flags is user.pax.flags 2015-04-16 12:28:10 hmmh 2015-04-16 12:29:22 i dont care that much about ff, but i need python. 2015-04-16 12:30:26 we do both markings still 2015-04-16 12:30:31 so even with the error, it should work 2015-04-16 12:30:50 though if its fatal error, the pkg might extracted be only partially 2015-04-16 12:44:40 damn, this aint my day 2015-04-16 12:46:59 fabled: i get grsec kernel now running, but python gets killed right after start. 2015-04-16 12:47:07 can that be grsec related? 2015-04-16 12:47:48 or does your xattr stuff break binary? 2015-04-16 12:49:57 http://sprunge.us/JXST 2015-04-16 13:25:05 clandmeter, well 2015-04-16 13:25:14 there's two options 2015-04-16 13:25:32 either the pax flags are wrong. but usually it would go further 2015-04-16 13:25:43 other option is that pt_elf and xattr paxflags mismatch 2015-04-16 13:26:37 in python case i think that's the issue 2015-04-16 13:26:39 pax.flags="em" 2015-04-16 13:26:46 but paxctl says mx 2015-04-16 13:26:48 i wonder why 2015-04-16 13:28:00 fabled: i am using grsec now, so apk doesnt error 2015-04-16 13:28:41 it can apply the flags, but it leaves my python unusable 2015-04-16 13:29:00 yes, because the flags are in addition in the binary itself 2015-04-16 13:29:03 and there's a mismatch 2015-04-16 13:31:42 it's the e 2015-04-16 13:31:49 clandmeter, just a moment 2015-04-16 13:32:07 i'm on it 2015-04-16 13:50:41 hmm 2015-04-16 13:51:07 still gets killed. 2015-04-16 13:52:29 huh 2015-04-16 13:52:48 duh 2015-04-16 13:57:31 clandmeter, works now? 2015-04-16 13:57:52 \o/ 2015-04-16 13:58:10 i think python was pretty much the only affectedp ackage 2015-04-16 13:58:15 it was related to using -z 2015-04-16 14:00:59 thx for the fast fix :) 2015-04-16 14:01:30 np. it was my bad. 2015-04-16 14:01:44 it also got only just unmasked yesterday, when we enabled both pax markings in kernel 2015-04-16 14:01:51 it worked with the first xattr kernel i tried 2015-04-16 14:04:00 im also getting an error about mouting /media/usb 2015-04-16 14:04:06 havent seen that error before 2015-04-16 14:07:25 what error? 2015-04-16 14:17:14 early in boot 2015-04-16 15:05:12 Natanael "I will be away on thursday and friday" Copa 2015-04-16 15:05:15 :p 2015-04-16 15:07:00 barthalion, i pushed his queue on his behalf :) 2015-04-16 15:07:52 ah 2015-04-16 15:15:10 barthalion: look at the committer field, you dont do that all the time? 2015-04-16 15:22:20 ba dum tss 2015-04-16 16:04:14 huh 2015-04-16 16:05:37 i remember seeing that once before. it seems to be a random self-check failure in libre's test suite 2015-04-16 16:10:33 clandmeter: yeah, somehow I always forget 2015-04-16 18:18:40 http://hastebin.com/enutunujoq.vbs 2015-04-16 18:18:49 any idea ^^ ? 2015-04-16 18:19:31 http://lists.alpinelinux.org/alpine-devel/1371.html ? 2015-04-16 18:20:06 alacerda, so far we patched out any execinfo usage 2015-04-16 18:20:16 hum... 2015-04-16 18:21:07 fabled, like this: http://git.alpinelinux.org/cgit/aports/commit/?id=2ba8ccba1648f11da1cbc7a4ccb1c87d92aead46 ? 2015-04-16 18:27:11 depnds on the binary 2015-04-16 18:29:06 alacerda: you can write something like: 2015-04-16 18:29:09 ++if defined (__GLIBC__) 2015-04-16 18:29:24 +#include 2015-04-16 18:29:50 ++#endif 2015-04-16 19:13:25 AmatCoder, thanks... trying 2015-04-16 19:31:37 the #if defined statement worked but in the code there is: 2015-04-16 19:31:37 call_stack_size = backtrace (call_stack, G_N_ELEMENTS (call_stack)); 2015-04-16 19:31:48 and backtrace is from execinfo 2015-04-16 19:31:56 how do we deal with this? 2015-04-16 19:32:27 it says: 2015-04-16 19:32:36 /home/alacerda/aports/testing/mate-power-manager/src/mate-power-manager-1.9.90/src/egg-debug.c:88: undefined reference to `backtrace' 2015-04-16 19:52:13 alacerda: probably you also could ifdef'ed the body of 'egg_debug_backtrace' function 2015-04-16 19:53:00 it just print some traceback to console 2015-04-16 19:53:26 hum.. ok, gonna try 2015-04-16 20:01:44 alacerda: kill it with fire 2015-04-16 20:05:26 barthalion, ? 2015-04-16 20:08:29 backtrace() 2015-04-17 00:00:55 barthalion: i updated the llvm patch list. i slipped up and included one that didn't belong yet. i've updated bug #4068 2015-04-17 06:37:35 G'day. 2015-04-17 06:49:43 hrrm 2015-04-17 06:49:53 alpinelinux.org/packages seems b0rken. 2015-04-17 06:52:48 http://pkgs.alpinelinux.org/packages 2015-04-17 06:57:43 oh 2015-04-17 06:57:44 thank you 2015-04-17 06:57:49 I was merely reporting it. 2015-04-17 06:57:50 :) 2015-04-17 07:11:42 i'm trying to make a rpi image that would work on any rpi 2015-04-17 07:12:03 using device tree, but for some reason it just breaks when enabling DT on thekernel image 2015-04-17 07:48:49 plans anybody to upgrade nodejs? 8) 2015-04-17 07:55:02 it wasn 2015-04-17 07:55:04 argh 2015-04-17 07:55:07 it wasn't building 2015-04-17 07:55:12 segfault during compilation 2015-04-17 07:55:29 oom? 2015-04-17 07:55:44 on our insanely good buildbots? 2015-04-17 07:56:39 you never know the quality of upstream code, take a look at kodi ;) 2015-04-17 07:56:52 oh, I'm aware of kodi quality 2015-04-17 07:56:57 no surprise there 2015-04-17 07:57:12 I remember that gcc was upgraded short time before, so I blame it and move on, as couldn't really do anything without fabled's or ncopa's brain 2015-04-17 07:57:51 i remember ncopa had an issue with oom recently. dont know if it was upstream or our issue though. 2015-04-17 08:18:45 now a friend of mine is building io.js within an alpine docker container -.- 2015-04-17 08:19:30 iojs is in testing 2015-04-17 08:24:28 iojs is not an option as he said a moment ago. one of his deps is not iojs aware 2015-04-17 08:24:40 so he needs to get a more current nodejs version compiled 2015-04-17 08:43:03 fcolista: this latest commit should fix kodi on alpine. kudos to fabled! 2015-04-17 08:43:20 if it works as expected, we can move it to main and remove xbmc. 2015-04-17 08:43:23 gr8! 2015-04-17 08:43:26 \o/ 2015-04-17 08:43:28 thx fabled 2015-04-17 10:20:37 \o/ 2015-04-17 10:20:39 rpi2 booted 2015-04-17 10:22:52 ok 2015-04-17 10:23:03 and i know how to make hybrid image that works on all rpi models 2015-04-17 10:23:06 nice 2015-04-17 10:23:14 need to just update build scripts now 2015-04-17 10:23:21 i'll be off for weekend 2015-04-17 10:23:34 will post edge rpi+rpi2 images on Mon or TUe 2015-04-17 13:04:32 interesting error in compiling nodejs 0.12.2 : 2015-04-17 13:04:43 #include No such file or directory 2015-04-17 13:04:53 :-O 2015-04-17 13:08:34 HAVE_IFADDRS_H is set to 1 by default. I think linker figures out by himself if this is right or not 2015-04-17 13:08:45 musl do have linux/if_packet.h 2015-04-17 13:39:05 I am in need of this: http://lists.alpinelinux.org/alpine-aports/0350.html is it possible to apply ? 2015-04-17 13:43:14 alacerda: sprunge it pls 2015-04-17 13:44:08 fcolista: i think you need to add linux-headers? 2015-04-17 13:45:22 clandmeter, already done 2015-04-17 13:46:47 you still get that error? 2015-04-17 13:49:50 yes 2015-04-17 13:58:07 clandmeter, http://sprunge.us/CbNI 2015-04-17 13:58:08 thanks 2015-04-17 14:02:01 fcolista: mine errors on uv_stat_t 2015-04-17 14:02:39 you have --shared-libuv on configure? 2015-04-17 14:02:40 ./configure --prefix=/usr \ 2015-04-17 14:02:40 --shared-zlib \ 2015-04-17 14:02:40 --shared-openssl || return 1 2015-04-17 14:02:45 this is my ./configure 2015-04-17 14:08:41 fcolista: non shared is building fine 2015-04-17 14:12:00 it compiles his own libuv 2015-04-17 14:13:16 fcolista: i think it needs a new version of libuv 2015-04-17 14:30:17 wow. libuv in main is pretty old 2015-04-17 14:30:48 aaaaah 2015-04-17 14:30:49 src/unix/linux-core.c:55:30: fatal error: linux/if_packet.h: No such file or directory 2015-04-17 14:30:52 again :) 2015-04-17 14:31:02 linux-headers 2015-04-17 14:31:13 yup 2015-04-17 14:31:25 it's building 2015-04-17 14:31:45 it needs cleanup 2015-04-17 14:31:49 i have it local 2015-04-17 14:32:00 it has autoconf now 2015-04-17 14:32:14 yes 2015-04-17 14:32:24 i've an updated APKBUILD for that 2015-04-17 14:32:49 nodejs still has issues with cares 2015-04-17 14:32:52 c-ares 2015-04-17 14:34:51 i guess you need to build it with non shared c-ares 2015-04-17 14:35:25 libuv is only needed for nodejs 2015-04-17 14:35:31 so i guess you can bump it 2015-04-17 14:35:41 is actually what arch does 2015-04-17 14:35:54 arch builds all non shared 2015-04-17 14:36:16 fedora is keeping it behind 2015-04-17 14:36:36 https://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/net-libs/nodejs/nodejs-0.12.2-r1.ebuild?view=markup 2015-04-17 14:37:48 alacerda: its added to aports, if you didnt see it yet. 2015-04-17 14:38:00 clandmeter, thanks.. I am testing it 2015-04-17 14:38:18 gentoo build all shared 2015-04-17 14:38:21 nice work 2015-04-17 14:38:36 fcolista: yes except c-ares 2015-04-17 14:38:46 yeah 2015-04-17 14:38:54 would it work for us the same? 2015-04-17 14:38:59 just leave c-ares out 2015-04-17 14:39:00 ACTION never used nodejs 2015-04-17 14:39:24 probably needs an older version of c-ares 2015-04-17 14:40:07 ok besides nodejs, no packages use libuv-dev 2015-04-17 14:40:16 so no need to bump pkgrel for other packages 2015-04-17 14:42:19 correct 2015-04-17 14:43:13 c-ares-dev is already at the last version 2015-04-17 14:43:27 it's 2013 also 2015-04-17 14:46:01 I saw a patch related to the error 2015-04-17 14:46:20 could be they have a snapshot in their source tree 2015-04-17 14:46:46 yeah 2015-04-17 14:51:58 clandmeter, i'm going to push libuv and nodejs 2015-04-17 14:52:12 sure 2015-04-17 14:55:41 fcolista: do you use alpine on your desktop now? 2015-04-17 14:57:48 not yet clandmeter 2015-04-17 14:57:55 i'm using arch 2015-04-17 14:58:27 but i'd like to move to alpine and dockerizing app not available for alpine 2015-04-17 14:58:37 that's my short-term goal 2015-04-17 14:58:54 which apps you are missing? 2015-04-17 14:59:57 still not have tought about that 2015-04-17 15:00:04 gotta check what i've installed :) 2015-04-17 15:00:25 the onlyu thing missing is chromium form e 2015-04-17 15:00:35 and a good keyboard 2015-04-17 15:01:30 kodi seems to be working fine. think ill move it to main this weekend 2015-04-17 15:01:40 just watched some tv :) 2015-04-17 15:04:01 I was trying to get xpra working for running X apps in containers - but I need a little help with a relocation error in GDK 2015-04-17 16:02:41 I hope to be able to try and test consul on AL soon ^^ 2015-04-17 19:02:49 clandmeter seems iso found in your link have some bug with partitioning an disk http://abload.de/img/vm_install01j7uow.png 2015-04-17 19:17:37 Jean-Scotch: i've pushed the package to @aports 2015-04-17 19:17:45 let's wait for it to come to testing :) 2015-04-17 19:20:32 nah actually i sent it to @devel 2015-04-17 19:20:35 but well 2015-04-17 19:20:37 ^^ 2015-04-17 21:13:11 clandmeter it happens on SCSI and SATA disk inside vmware workstation 11.1.0 2015-04-17 21:19:15 coredumb: great. thanks. I will lurk ;) 2015-04-17 22:09:14 python 2.7.9-r5 is missing pip, which should be present by default according to the Python docs. 2015-04-18 05:35:13 Jean-Scotch: if you can't wait i can put you the package online from my build as well :) 2015-04-18 07:28:53 coredumb: it should be nice ^^ I'have great expectation about consul and really want to give it a try and now my servers are all alpine boxes ;) 2015-04-18 07:30:20 I will start with about 100 machines (10 physical and 90 virtual and containers) 2015-04-18 07:39:41 100 machines nice one :) 2015-04-18 07:40:06 Jean-Scotch: consul is just amazing :) 2015-04-18 07:40:15 let me put you the package online 2015-04-18 07:40:45 great ! 2015-04-18 07:40:54 Jean-Scotch xen (90 virtual)? 2015-04-18 07:41:00 kvm 2015-04-18 07:41:23 and lxc 2015-04-18 07:42:59 Jean-Scotch: you use a virtualization platform/manager or plain KVM ? 2015-04-18 07:43:53 plain KVM. I launch my KVM with "screen -m -d -S KVM-name qemu ..." 2015-04-18 07:44:10 Jean-Scotch: ok 2015-04-18 07:44:16 Jean-Scotch: http://repos.mauras.ch/alpinelinux/x86_64/ there you go 2015-04-18 07:44:33 there's a bit more than consul though 2015-04-18 07:44:47 I see ^^ 2015-04-18 07:45:55 fwiw i implement details from https://www.mauras.ch/securing-consul.html in the package 2015-04-18 07:47:38 ok. i already read that page ;) 2015-04-18 07:48:16 will spin a bunch of VM to give it a try first ;) 2015-04-18 07:49:02 let me know if you find issues ^^ 2015-04-18 07:49:08 sure 2015-04-18 08:20:32 coredumb: have you send these patches to alpine-aports? 2015-04-18 08:22:20 ok, you did 2015-04-18 08:22:26 no idea why they aren't applied 2015-04-18 08:22:29 will take care of it 2015-04-18 09:17:41 coredumb: I will report at http://wiki.alpinelinux.org/wiki/User_talk:Jch/consul 2015-04-18 09:39:32 clandmeter: what do you think about setting up patchwork? 2015-04-18 10:14:31 coredumb: https://paste.xinu.at/lEpSL/ 2015-04-18 10:17:16 coredumb: left other comments on mailing list 2015-04-18 10:38:30 barthalion: i sent them on -devel instead of -aports i think 2015-04-18 10:38:31 my bad 2015-04-18 10:41:29 barthalion: the fun of Go compilation 2015-04-18 10:41:35 i'll see what i can do 2015-04-18 10:43:08 barthalion: it did actually compile right ? 2015-04-18 10:43:22 yes, because you didn't use '|| return 1' 2015-04-18 10:43:30 actually this error is not really one 2015-04-18 10:43:34 it created a package, I haven't tested it 2015-04-18 10:43:37 it needs a second pass -_- 2015-04-18 10:43:44 this is stupid 2015-04-18 10:43:47 then you should do '|| true' on lines that are ok to fail 2015-04-18 10:43:54 ok 2015-04-18 10:44:02 will add that 2015-04-18 10:44:03 (with a comment why) 2015-04-18 10:44:08 and return 1 on the rest 2015-04-18 10:44:20 I believe we don't want the compilation itself failing 2015-04-18 10:44:26 i added a comment on the following line iirc 2015-04-18 10:44:54 hm, right 2015-04-18 10:45:01 "# Doesn't make sense but seems it can't get everything in one pass" 2015-04-18 10:45:16 still, isn't clear enough to me without explicit or and true 2015-04-18 10:45:42 there were some trailing whitespace as well, you might want to check git diff before commiting 2015-04-18 10:45:50 yep 2015-04-18 10:45:53 will fix that 2015-04-18 10:49:23 thanks 2015-04-18 11:18:48 will wait for -r1 ;) 2015-04-18 11:40:22 Jean-Scotch: won't change anything ^^ 2015-04-18 13:30:54 barthalion: sent 2015-04-18 13:31:46 actually i was wondering if i should make godeps output to /dev/null as well but well thought it could still be useful to have the output 2015-04-18 15:21:51 coredumb: ugh, godep takes so much time 2015-04-18 15:34:49 clandmeter i filled a bug report http://bugs.alpinelinux.org/issues/4072 2015-04-18 15:54:03 coredumb: I can't really push a package that 2/3 of time eats waiting for godeps timeout 2015-04-18 15:56:52 does it timeout for git checkout? 2015-04-18 15:57:51 yeah, looks like that 2015-04-18 15:58:06 why not create a snapshot? 2015-04-18 15:58:27 examples of that? packaging go stuff makes me cringe 2015-04-18 15:58:58 i think you can run go fetch --switch-without-building 2015-04-18 15:59:16 then tar it up and send it to dev.a.o 2015-04-18 15:59:26 hm 2015-04-18 15:59:54 i think go fetch should never be run in abuild -r 2015-04-18 16:00:50 i think i created a go apkbuild recently which does just that. dunno if i ever pushed it. 2015-04-18 16:02:04 I don't know yet how it fits godeps command 2015-04-18 16:02:19 but yeah, git fetch isn't the best idea 2015-04-18 16:02:28 s/godeps/godep/g 2015-04-18 16:04:33 go get -v -u -d 2015-04-18 16:04:42 thats the command i ment 2015-04-18 16:06:21 this apkbuild is not functioning, but i think the snapshot part is ok. http://sprunge.us/hHbK 2015-04-18 16:07:02 thanks, gonna take a look tomorrow 2015-04-18 17:30:36 barthalion: you got timeouts ? 2015-04-18 17:31:29 mmmh yeah we should prefetch everything then 2015-04-18 17:32:22 i'll check the best way to do so 2015-04-18 17:32:35 gonna be huge source ... 2015-04-18 19:34:13 clandmeter: I've seen go get -d not download enough deps 2015-04-18 19:34:39 seems it sometimes need two or more runs to get everything ... 2015-04-18 19:34:51 issue with too much levels of deps ? 2015-04-18 19:35:11 even godeps on consul building fails to get two deps 2015-04-18 19:35:17 godep* 2015-04-18 20:28:59 libevent@edge should probably be moved to main libevent 2.0.22 was released jan 2014 - TOR is broken with libevent in main (2.0.21) - Error relocating /usr/bin/tor: evutil_secure_rng_set_urandom_device_file: symbol not found 2015-04-19 13:40:55 Hi can someone please do something with 2 YEARS OLD!!!!! CUPS wiki article (its almost impossible to make it work right), half of article is obsolete (for example command lppasswd no longer working anymore).... thanks ... 2015-04-19 13:43:24 lol 2015-04-19 13:43:53 "somebody do something" 2015-04-19 13:43:57 obviously not him 2015-04-19 13:46:19 because i cant, i not now how :) 2015-04-19 13:46:36 if i do, it will be done long ago 2015-04-19 13:47:21 *know 2015-04-19 13:47:39 sorry for grammar i am from slovakia 2015-04-19 13:49:50 barthalion: so i asking others... 2015-04-19 13:51:29 I don't think setting up cups on Alpine requires quantum physics phd 2015-04-19 13:51:39 most likely any tutorial will do, like from archwiki 2015-04-19 13:52:09 basically apk add cups && service cups start, or whatever is package and service named 2015-04-19 13:52:44 you may want to add your user to the "lp" group if you don't have permissions to print 2015-04-19 13:52:53 i try it already ... do you think i not trying everything ??? i trying it for 1 week and i now how because i set it up on ubuntu under 15 minutes 2015-04-19 13:53:10 ubuntu does a lot of magic behind the scenes 2015-04-19 13:53:20 so it's bad indicator of knowledge 2015-04-19 13:53:23 you cant even install printer something at device detection is corrupted 2015-04-19 13:53:52 i install it on ubntu under cli (i dont have gui in server) 2015-04-19 13:54:03 ubuntu server is still ubuntu, you know 2015-04-19 13:54:17 bring your logs to bugs.alpinelinux.org 2015-04-19 13:54:33 but i addmit i have less knowelage, so because of this i asking someone to do something with article 2015-04-19 13:56:04 whoever it is, I hope he'll just remove it 2015-04-19 13:56:39 my printer was very happy to be shared by alpine thinclient around home network, no matter how bad it sounds, 2 releases ago 2015-04-19 13:58:32 i love alpine because of low system resource consumption too.. (what else i can put on machine with 384mb ram :D ) 2015-04-19 14:06:36 (openwrt) 2015-04-19 14:08:23 384MB is low memory setup? 2015-04-19 14:15:46 lawl 2015-04-19 14:16:15 I thought that 128MB is plenty. 2015-04-19 14:18:16 Alpine worked flawlessly on thinclient with <64MB of RAM. 2015-04-19 14:18:57 my beard is so thick 2015-04-19 14:19:01 unix beard* 2015-04-19 14:20:15 your what? :f 2015-04-19 14:31:24 did you see my message yesterday about putting the libevent in edge into main ? (so TOR is not broken) - or do I need to raise a bug ? 2015-04-19 15:21:48 there is no libevent in testing to move it to main 2015-04-19 15:31:39 was in edge 2015-04-19 15:32:51 I had to 'apk add libevent@edge' to make TOR work - & the latest libevent release was January 2014 2015-04-19 16:09:47 Hello, I'm new to alpine linux (and IRC) and I like what I see, but I would really like to have nmtui (networkmanager ncurses gui) as an apk; what's my best option, is there a developer willing to take this on, or am I basically on my own? Networkmanager ist available in edge/testing, but it's an older version which does not include nmtui. 2015-04-19 16:11:31 I can try updating nm - I'm sending a bunch of patches today - what is the current version that has ncurses ? 2015-04-19 16:11:50 1.0 2015-04-19 16:14:11 That is, the 0.9.8.10-rc1 version that is available has ncurses, but not nmtui; It "only" has nm-cli. 2015-04-19 16:15:16 That should have bin nmcli not nm-cli. 2015-04-19 16:16:01 1.0 build running 2015-04-19 16:16:35 Wow, thanks! 2015-04-19 16:23:41 it will probably be tomorrow before one of the guys applies the patch - in the meantime these instructions are good for making wireless work http://linuxcommando.blogspot.nl/2013/10/how-to-connect-to-wpawpa2-wifi-network.html 2015-04-19 16:24:15 + the comment from "Emerson Prado" a bit further down 2015-04-19 16:27:00 Thanks, I'm quite familiar with networking and getting it working is not the problem; I actually need nmtui in order to be able to use alpine as the basis of a project of mine. 2015-04-19 16:28:04 Sorry I'm so slow to answer, but I'm quite dyslexic - not good for IRC :-). 2015-04-19 16:28:20 no problem 2015-04-19 16:32:55 If the patch is applied, does that mean it will show up in the repros with an "apk search"? 2015-04-19 16:36:25 yes - in a day or so 2015-04-19 16:38:43 OK, then I'll check back in a day or so. Thanks for your help and time. 2015-04-20 06:12:24 morning 2015-04-20 06:19:46 hi all. barthalion, re: patchwork, i created this page 2015-04-20 06:19:47 http://wiki.alpinelinux.org/wiki/Patchwork 2015-04-20 06:20:01 it might be a start point 2015-04-20 06:20:44 cool 2015-04-20 06:20:56 but I guess it needs ncopa approval :p 2015-04-20 06:21:39 ncopa: tldr we could use patchwork to make sure we don't miss any patch on mailing lists 2015-04-20 06:25:20 +1 barthalion 2015-04-20 06:28:47 sounds good to me 2015-04-20 06:28:53 i have no experience with it 2015-04-20 06:29:07 but if it is a tool that helps us not miss patches, then i'm all for it 2015-04-20 06:31:08 it gives good overview what's the status of patches 2015-04-20 06:31:11 like: https://patchwork.archlinux.org/project/pacman/list/ 2015-04-20 06:31:40 although I have no experience in setting it up either, so we need to employ fcolista and clandmeter for that :) 2015-04-20 06:32:04 i also never had experience with that, but i love to experiment :) 2015-04-20 06:34:00 clandmeter: could you please create a container for fcolista to do patchwork? 2015-04-20 06:59:05 sure 2015-04-20 06:59:14 does it need a db? 2015-04-20 07:01:14 fcolista: your instructions mention pgsql 2015-04-20 07:01:28 mysql can be used as well 2015-04-20 07:01:31 yes 2015-04-20 07:01:36 itworks with both 2015-04-20 07:01:50 dunno what kind of backend we use 2015-04-20 07:02:06 currently i only have mysqldb 2015-04-20 07:02:48 patchwork doesnt do any versioning? 2015-04-20 07:03:16 seems not 2015-04-20 07:04:39 do we care to pkg it? 2015-04-20 07:05:43 yes, sure. 2015-04-20 07:06:14 maybe also an email to upstream developer to add versioning would be nice :) 2015-04-20 07:06:33 at the looks of it, i dont think they care. 2015-04-20 07:06:44 yeah, i got the same feeling 2015-04-20 07:07:02 you can create dated snapshots. 2015-04-20 07:07:34 yes 2015-04-20 07:08:34 to easier maintain infra, its nice to have the things we use in aports. 2015-04-20 07:09:11 I was looking at gogs, but im not sure what to think about it. 2015-04-20 07:11:09 the founder is chinese, so it must be full of mallware :p 2015-04-20 07:12:48 ACTION hides because most of his colleagues are from that region. 2015-04-20 07:17:56 testing new u-boot on wandboard, so arm builder will be offline for a while 2015-04-20 09:26:53 http://sprunge.us/hCjU <-- main/php-phalcon upgrade v2.0, update pls 2015-04-20 12:29:33 humpf, i messed up my build env while rebuilding for new boost :-/ 2015-04-20 12:50:56 bitcoin fails to build due to templates magic 2015-04-20 13:05:41 bitcoin build error with boost 1.58: http://sprunge.us/PQDK 2015-04-20 13:05:47 i have not a clue what is wrong 2015-04-20 13:07:32 looks like "breaking change" under "Variant" here: http://www.boost.org/users/history/version_1_58_0.html 2015-04-20 13:09:35 so -DBOOST_VARIANT_USE_RELAXED_GET_BY_DEFAULT should fix it 2015-04-20 13:09:44 assuming I'm correct 2015-04-20 13:09:51 ncopa: ↑ 2015-04-20 13:10:26 i read the boost release notes 2015-04-20 13:10:30 and came to same conclusion 2015-04-20 13:11:44 yup 2015-04-20 13:11:48 that fixed it 2015-04-20 13:14:56 i didnt check though 2015-04-20 13:15:08 i suppose patching bitoin itself would be better 2015-04-20 13:15:17 might a 'const' could be removed or similar 2015-04-20 13:16:00 15:11:44 <@ncopa> that fixed it 2015-04-20 13:16:01 15:14:51 <@ncopa> i didnt check though 2015-04-20 13:16:03 wat 2015-04-20 13:16:09 i mean 2015-04-20 13:16:09 I see contradiction there :D 2015-04-20 13:16:23 i did add -DBOOST_RELAX_A_BIT 2015-04-20 13:16:43 but what i didnt check was to try fix the type in bitcoin code 2015-04-20 13:16:59 i think it does stricter type checking 2015-04-20 13:17:05 ah, got it 2015-04-20 13:22:13 coredump: about consul tests, I'm stuck because I cannot access to the HTTP API. Everything looks denied I do not find how to open it a bit... The DNS API is working fine in conjunction of dnsmasq (I couldn't make it work with dnscache). 2015-04-20 13:24:28 coredumb: DNS API is alreay a winning feature ;) 2015-04-20 13:25:37 but to many missing features (like introspection) 2015-04-20 13:30:23 coredumb: see the log of my tests at http://wiki.alpinelinux.org/wiki/User_talk:Jch/consul 2015-04-20 13:51:15 bitcoin lol 2015-04-20 13:52:12 kaniini: bitcoin and tox, brother, this is da future 2015-04-20 13:52:21 serious netsec 2015-04-20 13:53:15 did tox ever fix their dubious security primitives 2015-04-20 14:13:24 kaniini: $ pkgconf --exists hogweed; echo $? 2015-04-20 14:13:24 1 2015-04-20 14:13:31 $ sprunge < /usr/lib/pkgconfig/hogweed.pc 2015-04-20 14:13:31 http://sprunge.us/DeTe 2015-04-20 14:13:51 kaniini: i wonder if pkgconf gets confused by the # in there 2015-04-20 14:14:06 it shouldn't. 2015-04-20 14:14:22 oh, i see 2015-04-20 14:14:26 yes, probably. 2015-04-20 14:14:36 yes 2015-04-20 14:14:42 removing those fixed it 2015-04-20 14:14:50 its nettle.pc from neettle-3.1 2015-04-20 14:15:53 kaniini@Circe ~/d/pkgconf> ./pkgconf tests/lib1/comments-in-fields.pc --libs 2015-04-20 14:15:54 -lhogweed # -lgmp 2015-04-20 14:15:59 oh dear ;) 2015-04-20 14:23:14 hi all. There's a way to switch from x86 to x86_64 without reinstalling? 2015-04-20 14:23:23 from disk-install 2015-04-20 14:24:09 fchome: you need 64 bit kernel for that 2015-04-20 14:24:23 so you need boot up 64 bit kernel first 2015-04-20 14:24:32 yeah 2015-04-20 14:24:45 i'm wondering to start from a usb 2015-04-20 14:24:46 otherwise, you could n theory change /etc/apk/arch 2015-04-20 14:25:17 and just apk info -q | xargs apk fix 2015-04-20 14:25:19 or similar 2015-04-20 14:25:26 as first step i changed arch and did for i in $(cat /etc/apk/world); do apk fix $\;done 2015-04-20 14:25:37 sorry, $i 2015-04-20 14:25:52 you need run 64 bit kernel first 2015-04-20 14:26:02 right 2015-04-20 14:26:12 you could probably boot from usb 2015-04-20 14:26:20 mount your root some place 2015-04-20 14:26:40 and then do it with apk --root ... 2015-04-20 14:27:40 Jean-Scotch: you followed my guide on the default blacklist ? 2015-04-20 14:27:46 thx ncopa 2015-04-20 14:28:01 for the KV ACLs i mean 2015-04-20 14:50:44 hum.... building a go binary within a docker container sucks :( 2015-04-20 14:51:08 the size of the container doesnt get reduced again after removing all the build tools :( 2015-04-20 14:51:41 looks like i have to use some build system that injects the binary into the plain container :( 2015-04-20 14:53:00 ncopa: should be fixed by 0.9.9. 2015-04-20 14:54:14 thanks! 2015-04-20 14:54:17 that was quick :) 2015-04-20 15:08:21 was kodi building stopped? 2015-04-20 15:08:29 I can't find it in testing 2015-04-20 15:11:35 kodi was moved to main iirc 2015-04-20 15:12:35 is not in main also 2015-04-20 15:13:21 fchome: http://git.alpinelinux.org/cgit/aports/commit/?id=c2b1d1d1c9563a464540ed7cfac49f42cca3ceb8 2015-04-20 15:14:44 scadu seems that this package is not builded 2015-04-20 15:14:53 since is not available 2015-04-20 15:15:05 http://pkgs.alpinelinux.org/package/main/x86/kodi 2015-04-20 15:15:07 ba dum tss 2015-04-20 15:15:22 did you do apk update? 2015-04-20 15:15:39 yes. Seems that in nl.alpinelinux.org is available 2015-04-20 15:16:19 while in dl-3 not 2015-04-20 16:36:54 new images 2015-04-20 16:36:56 uboot: http://dev.alpinelinux.org/~tteras/alpine-uboot-150420-armhf.tar.gz 2015-04-20 16:37:20 should work on wandboard, cubie/beagle is also compiled, but completely untested. 2015-04-20 16:37:42 and now uploading universal rpi image that works on all rpi models (a,a+,b,b+,b2) 2015-04-20 16:44:08 and the rpi image is at: http://dev.alpinelinux.org/~tteras/alpine-rpi-150420-armhf.rpi.tar.gz 2015-04-20 16:44:47 ttyl 2015-04-20 17:04:35 coredumb: found! incompatibility between setup-proxy and consul... 2015-04-20 17:07:01 not really consul but the way curl goes to localhost 2015-04-20 17:26:24 mosez: you need to do everything in one layer to reduce the space 2015-04-20 17:26:32 mosez: each RUN == new layer 2015-04-20 17:34:25 yeah, fixed it that way 2015-04-20 18:19:37 fabian_a thanks for the alpine-rpi-150420-armhf.rpi.tar.gz 2015-04-20 18:23:22 Jean-Scotch: :) 2015-04-21 08:13:57 Taskwarrior 2.4.3 Released 2015-04-21 08:18:09 any thoughts on paste.a.o to complement patchwork.a.o 2015-04-21 08:39:43 vkrishn: whats the use of a self hosted paste service? 2015-04-21 08:41:24 we all use sprunge anyway 2015-04-21 08:42:04 yep 2015-04-21 08:42:16 and if the data is private, dont paste it :) 2015-04-21 08:42:54 sprunge is setup in like 10 minutes i think. but still. 2015-04-21 08:48:42 barthalion: you tried gogs before right? you were having issues? 2015-04-21 08:52:24 there are race conditions sometimes 2015-04-21 08:52:31 I think they are reported on github 2015-04-21 08:54:32 it seems it sucks with flat repo's like alpine has 2015-04-21 08:54:37 takes ages to load them. 2015-04-21 09:15:22 can someone with more knowelage check this http://bugs.alpinelinux.org/issues/4072#change-11932 (it breaks setup-apline) 2015-04-21 09:18:48 crow: its edge, its suppose to break :) 2015-04-21 09:20:32 clandmeter heh yes but edge will be also mergaed to stable so it will then be broken thee to right? 2015-04-21 09:20:55 repo pin stable and add it from stable, as a short time solution. 2015-04-21 09:21:32 i can try that, thanks 2015-04-21 09:28:34 1. expiry (can be set to 6months) 2015-04-21 09:28:34 2. Javascript/Hot cross/linking from withing patchwork.a.o 2015-04-21 09:30:03 3. searchable 2015-04-21 09:30:13 4. tagable 2015-04-21 09:31:33 its not about being private, AL is public repo/data 2015-04-21 09:31:58 it exclusive al paste 2015-04-21 09:33:16 I still don't get it 2015-04-21 09:33:29 if you need to put something searchable and tagable, do so on wiki 2015-04-21 09:34:00 yes, but is scrap data 2015-04-21 09:34:11 so we have sprunge 2015-04-21 09:34:13 ix 2015-04-21 09:34:24 and any other thing that you can use instead 2015-04-21 09:35:42 hmm... 2015-04-21 09:36:18 I don't see value in it 2015-04-21 09:36:36 other than increases burden of infra maintainers 2015-04-21 09:37:09 yes, thats the (-) point 2015-04-21 09:46:34 vkrishn: nobody is stopping you to set one up and maintain it. 2015-04-21 09:48:03 :) 2015-04-21 09:49:21 some features that can be enabled in rsyslog, http://lists.alpinelinux.org/alpine-devel/4466.html 2015-04-21 10:44:31 ncopa: llvm on x86 fails to build since yesterday. disabled for now. 2015-04-21 11:14:08 ok 2015-04-21 11:14:14 thanks 2015-04-21 11:14:48 its testing/llvm not main/llvm 2015-04-21 11:15:45 clandmeter: ^^^ 2015-04-21 12:13:08 is it possible that nslookup does not resolve correctly? i'm using skydns and with the simple nslookup it fails quite often. if i install bind-tools first it works fine. 2015-04-21 12:29:14 musl needs these defines uclibc have added for compatibility with apps that support systemd builds http://cgit.openembedded.org/openembedded-core/commit/?id=7562021eef5b7585122c92db8b686808ebe7d85e - networkmanager will not build with IPTOS_CLASS_CS6 undefined - is there a way to patch /usr/include/netinet/ip.h in an APKBUILD as a temporary fix ? 2015-04-21 15:51:53 hi 2015-04-21 19:54:09 ncopa or clandmeter, please... could you appply: http://sprunge.us/YWeM 2015-04-21 19:54:10 ? 2015-04-21 20:30:28 algitbot: did you bump? 2015-04-21 20:32:10 i mean alacerda :) 2015-04-21 21:09:37 does go pkgs not require "go-" prefix ? 2015-04-21 22:48:22 clandmeter, sorry i didnt 2015-04-21 22:48:25 =/ 2015-04-21 23:11:31 ACTION spanks alacerda  2015-04-21 23:15:23 ACTION learns the lesson 2015-04-21 23:23:19 vkrishn: i dont see any reason why. 2015-04-22 00:23:16 ah, yes, that would be havoc, as it would imply doing same with haskell, erlang, scala, go, rust.. etc 2015-04-22 05:57:35 what's up guys 2015-04-22 05:59:58 morning 2015-04-22 06:00:40 oh wow cool 2015-04-22 06:00:44 heh 2015-04-22 06:00:50 sorry just been dead on irc 2015-04-22 06:01:16 so are you a dev on alpine or you just run it? 2015-04-22 06:01:27 (i just found out about it like a week ago ) 2015-04-22 06:01:41 i am a dev and i run it :) 2015-04-22 06:02:08 you can pretty much check who commits to aports to see who's dev :p 2015-04-22 06:02:24 oh wow 2015-04-22 06:02:28 http://git.alpinelinux.org/cgit/aports/stats/ 2015-04-22 06:02:30 sorry , didn't mean to be rude 2015-04-22 06:02:37 long day 2015-04-22 06:02:54 i didnt take it as rude 2015-04-22 06:03:20 my computer's actually been wonky lately ... i actually rented a vps to build linux from scratch overnight 2015-04-22 06:03:30 a 32 core like 64gig ram vs 2015-04-22 06:03:31 lol 2015-04-22 06:03:35 vps* 2015-04-22 06:03:51 so i'm actually really excited 2015-04-22 06:04:15 about the opportunity that i feel is in place, in terms confluence of circumstances 2015-04-22 06:04:22 it's like the perfect storm for innovation in linux 2015-04-22 06:04:51 cool :) 2015-04-22 06:04:59 thats even better than our build server(s) 2015-04-22 06:05:08 anyway i think alpine linux is the example of what should be runnig in the datacenters 2015-04-22 06:05:12 we only have 24 cores and 48GB ram.. 2015-04-22 06:05:15 haha 2015-04-22 06:05:21 well i'm exaggerating a *little* 2015-04-22 06:05:31 but otoh... we have two of them :) 2015-04-22 06:05:32 oh 2015-04-22 06:05:35 i am not ;) 2015-04-22 06:06:38 lemme see.. okay so i picked up a kvm vps , 12 cpu , 32gigs ram, 400gb ssd for about 50c an hour 2015-04-22 06:06:45 at vultr 2015-04-22 06:06:47 https://www.vultr.com/pricing/ 2015-04-22 06:07:07 and then i grabbed one of their other 'disk' instances to run as an nfs mount 2015-04-22 06:07:14 i'm compiling linux from scratch 2015-04-22 06:07:23 and trying to see if i can get nix to run in it 2015-04-22 06:07:54 to see what nix would be acceptable for 2015-04-22 06:08:16 i love the idea of a functional programming language to define packages and linux configs 2015-04-22 06:08:29 but the nixos i downloaded in a vm seemed like a memory hog 2015-04-22 06:09:08 so anyway, yeah lfs is running overnight. . . . the other thing i want to try is see just how bad it's gonna be to extract systemd out of the buildz 2015-04-22 06:09:33 whether it's going to be possible to support gnome3 easily or whether that's gonna be impossible eventually 2015-04-22 06:09:57 oh and i also want to try out replacing openrc with that fakes systemd shim 2015-04-22 06:10:10 you might get it working partially with uslessd 2015-04-22 06:10:13 yeah! 2015-04-22 06:10:15 that's what its called 2015-04-22 06:10:16 uselessd 2015-04-22 06:10:20 was trying to remember the name 2015-04-22 06:10:27 i knew it was someting funny 2015-04-22 06:10:35 and the guys faq was hillariuos 2015-04-22 06:10:37 hahah 2015-04-22 06:11:07 so i was like , man replacing systemd with uselessd is gonna be a great exercise even if it doesnt work 2015-04-22 06:11:30 but yeah, alpine i want to run as my 100% datacenter linux 2015-04-22 06:11:54 so i'm sort of feeling out how far i can go, what i have re-use that's already written on github 2015-04-22 06:12:00 or in packages 2015-04-22 06:12:25 as i'm sure you are aware there is an explosion in these 'hypervisor' OSs 2015-04-22 06:12:30 cause of docker 2015-04-22 06:13:16 but alpine can already do like twice the stuff of these super-popular tech companes here 2015-04-22 06:13:26 like whatever they are called.. joyent 2015-04-22 06:14:18 theyve got an 'apline'd ' version of open solaris that essentally isn't all that diffferent from alpine 2015-04-22 06:15:05 but i think it's amusing that alpine sitll has way more features than ths cutting edge 'cloud' OS 2015-04-22 06:15:08 at least for now 2015-04-22 06:15:34 if kvm , zfs, and a patched dtrace works in alpine 2015-04-22 06:16:05 you have surpassed san francisco's 'cloud OS hipsters' without trying 2015-04-22 06:16:13 haha 2015-04-22 06:16:46 but i suppose they probably have a reason 2015-04-22 06:16:49 for picking solaris 2015-04-22 06:16:53 maybe the license 2015-04-22 06:17:40 what inspired you to create this distro ncopa 2015-04-22 06:17:49 or work on it 2015-04-22 06:17:57 just intellectual curiousity? 2015-04-22 06:19:04 ..... 2015-04-22 06:19:05 ... 2015-04-22 06:20:32 I believe there was an article on wiki or somewhere how it started 2015-04-22 06:20:43 oh and one more question . hopefulyl u see this before u call it a night... Is there any 'risk' of running musl and ulibc with gnu c side by side ? like is that risky from clobbering shared library symbols 2015-04-22 06:21:11 do you have to do one or the other (muscl/ulib OR glibc ) or can u run em both without much hassle 2015-04-22 06:21:23 and still gcc stuff regularly 2015-04-22 06:21:32 without having to specify the flags manually 2015-04-22 06:21:38 by default? there is no way to use glibc 2015-04-22 06:21:40 anyone else 2015-04-22 06:21:42 answerthat 2015-04-22 06:21:51 yeah 2015-04-22 06:21:58 i used a run from-ram distro called leaf bering or something 2015-04-22 06:22:03 I guess you can build it, and so on, but I don't think it's gonna work 2015-04-22 06:22:05 many years ago 2015-04-22 06:22:10 naw i get it 2015-04-22 06:22:18 i dont like gnu libc 2015-04-22 06:22:21 i really hate it actually 2015-04-22 06:22:38 i found uclibc cool, and found that gentoo could build uclibc systems 2015-04-22 06:22:46 but i'm just thinking in pragmatic terms if i need something. . . the larger it is the more i might haev to patch it, rght? 2015-04-22 06:22:47 the only way I see is to install glibc somewhere else than /lib and /usr/lib 2015-04-22 06:23:00 and play with LD_PRELOAD for glibc-built applications 2015-04-22 06:23:07 but for lowend machines it was not good to compile it all, so i ended up building a distro with uclibc 2015-04-22 06:23:23 lets say it's a big appplication called like .. oh i dunno.. something thats not a drive 2015-04-22 06:23:27 say a daemon.. but a big one 2015-04-22 06:23:32 that has lots of syscalls 2015-04-22 06:23:36 no kernel stuff tho 2015-04-22 06:23:49 eh syscalls = call to kernel? 2015-04-22 06:23:49 won't that need a bit of customization to link against ulibc/musl 2015-04-22 06:23:54 well u know what i mean 2015-04-22 06:23:56 hhaha 2015-04-22 06:24:02 "not FUSE" 2015-04-22 06:24:02 it's not about syscalls but importable code 2015-04-22 06:24:03 basically 2015-04-22 06:24:04 haha 2015-04-22 06:24:17 musl doesn't forgive it 2015-04-22 06:24:17 just read(),, write(), and a dozen others 2015-04-22 06:24:25 we switched to musl 2015-04-22 06:24:28 then maybe some links out to a few libc symbols 2015-04-22 06:24:33 and we run huge things 2015-04-22 06:24:34 like in shared libs 2015-04-22 06:24:48 is it a binary only or you build it from source? 2015-04-22 06:24:53 if you build it from source it should be safe 2015-04-22 06:24:54 no no 2015-04-22 06:24:57 building from source 100% 2015-04-22 06:25:15 stuff we run with musl libc: 2015-04-22 06:25:16 for now anyway.. its not an actual problem i have yet 2015-04-22 06:25:16 firefox 2015-04-22 06:25:17 then you're good to go 2015-04-22 06:25:21 its more i'm trying to understand 2015-04-22 06:25:21 libreoffice 2015-04-22 06:25:31 because i don't know this stuff as well as you guys do haha 2015-04-22 06:25:42 qemu, xen (hypervisor etc) 2015-04-22 06:25:43 i just started rolling my own linux recently 2015-04-22 06:25:51 but i do alot of virtauliation stuff yeah 2015-04-22 06:26:07 we did have an issue with xen 2015-04-22 06:26:14 with hvmloader 2015-04-22 06:26:25 oh wow 2015-04-22 06:26:25 due to they compile 32bit with -m32 2015-04-22 06:26:28 okay ncopa 2015-04-22 06:26:32 lets take like firefox 2015-04-22 06:26:39 or libreoffice 2015-04-22 06:26:46 i run them on my desktop 2015-04-22 06:26:47 how much source code modification did u do 2015-04-22 06:26:52 to get em to copile 2015-04-22 06:26:57 or nothing 2015-04-22 06:27:06 http://git.alpinelinux.org/cgit/aports/tree/main/firefox 2015-04-22 06:27:13 thanks 2015-04-22 06:27:16 you have the patches there 2015-04-22 06:27:20 okay 2015-04-22 06:27:22 some are lifted from fedora 2015-04-22 06:27:24 thank you this is a huge help to me 2015-04-22 06:27:38 i was so confused 2015-04-22 06:27:52 libreoffice apparently only needs 2 patches 2015-04-22 06:27:57 http://git.alpinelinux.org/cgit/aports/tree/main/libreoffice 2015-04-22 06:28:04 okay 2015-04-22 06:28:18 normally the problem is that source code is non-portable 2015-04-22 06:28:29 or that source code make wrong assumptions 2015-04-22 06:28:42 they often seem to think that linux == glibc 2015-04-22 06:29:10 yeah i see 2015-04-22 06:29:15 so you probably have less issues 2015-04-22 06:29:20 so we have various patches that does nothing else than replace #if defined(__linux__) with #if defined(__GLIBC__) 2015-04-22 06:29:26 with stuff that already compiles on both bsd and nix, right? 2015-04-22 06:29:32 correct 2015-04-22 06:29:37 okay .. hmm 2015-04-22 06:29:39 fascinating 2015-04-22 06:29:40 sometimes you find stupid stuff as 2015-04-22 06:29:45 #if osx 2015-04-22 06:29:50 #else if windows 2015-04-22 06:29:59 #else if 2015-04-22 06:29:59 well that explainz why i failed so hard at bootstrapping lfs onto alpine 2015-04-22 06:30:00 ... 2015-04-22 06:30:02 haha 2015-04-22 06:30:22 i think i saw gcc-dev and mixed it up with glibc 2015-04-22 06:30:28 i now see there is no glibc package at all 2015-04-22 06:30:32 that's really cool 2015-04-22 06:30:35 correct 2015-04-22 06:30:37 even more impressive 2015-04-22 06:30:38 wow 2015-04-22 06:30:42 we use musl libc 2015-04-22 06:30:48 yeah i saw that part 2015-04-22 06:30:53 for everything 2015-04-22 06:31:01 bootstrapping gcc was non-trivial 2015-04-22 06:31:13 but not impossible :) 2015-04-22 06:31:16 but u know . liek i'm not a huge fan of some of that gnu code . . . some is okay 2015-04-22 06:31:20 some is sprawling 2015-04-22 06:31:27 so that's rad 2015-04-22 06:31:29 then you are in the right place :) 2015-04-22 06:31:34 actually 2015-04-22 06:31:35 no wonder the thing runs so far 2015-04-22 06:31:37 fast* 2015-04-22 06:31:48 one of the things that inspired me was bsd 2015-04-22 06:31:52 so then really ths is actually an ideal server OS 2015-04-22 06:31:53 openbsd, freebsd 2015-04-22 06:31:58 exactly what i was about to say 2015-04-22 06:32:02 you have the flexiblity of linux 2015-04-22 06:32:04 but i wanted to use linux kernel 2015-04-22 06:32:07 with the sprawl and the security hopes 2015-04-22 06:32:11 due to harware support etc 2015-04-22 06:32:13 holes* 2015-04-22 06:32:23 oh we patch the kernel 2015-04-22 06:32:26 grsecurity 2015-04-22 06:32:29 yeah that's really cool 2015-04-22 06:32:41 yup saw that.. that's part of the reason i was like i gotta download this 2015-04-22 06:32:46 humm 2015-04-22 06:33:52 that also explains the conspicuous absense of gtk and gnome 2015-04-22 06:33:54 haha 2015-04-22 06:34:04 absence of gtk? 2015-04-22 06:34:11 well 2015-04-22 06:34:12 we have gtk. but it's not mandatory install. 2015-04-22 06:34:13 im searchig now 2015-04-22 06:34:16 i dont seeit 2015-04-22 06:34:21 gtk+2.0, gtk+3.0 2015-04-22 06:34:29 oh i need wildcards duh 2015-04-22 06:34:30 sorry 2015-04-22 06:34:40 there is no gnome 3 atm, but I believe it's getting more possible, albeit slowly 2015-04-22 06:34:41 just looking at the window managers u guys ported 2015-04-22 06:35:05 yeah 2015-04-22 06:35:10 well i hate gnome3 personally 2015-04-22 06:35:15 but i know alot of ppl love it 2015-04-22 06:35:18 for their desktops 2015-04-22 06:35:35 well, I don't love it, it's just convenient 2015-04-22 06:35:36 i use xfce (qt based i think) or mate (gtk2 based thinkk) 2015-04-22 06:35:59 i'm really curious to try like nextstep or of these other new nongtk window managers 2015-04-22 06:36:13 like 'window maker / fluxbox: 15 years later' 2015-04-22 06:36:16 heh 2015-04-22 06:36:40 do you guys run alpine on the desktop? 2015-04-22 06:36:54 i do 2015-04-22 06:36:56 i use xfce 2015-04-22 06:37:15 i use it on my work desktop, and on my laptop 2015-04-22 06:37:19 okay .. rad 2015-04-22 06:37:29 does it support the usual hardware 2015-04-22 06:37:35 wel let me rephrase that question 2015-04-22 06:37:55 things that you might miss is thunderbird 2015-04-22 06:38:03 i saw the firmware bundled in the minimal disk for wireles... do you guys find your hardware support for laptops is pretty decent? 2015-04-22 06:38:03 skype 2015-04-22 06:38:13 iwlwifi firmware 2015-04-22 06:38:22 i think its possible to hack something to work 2015-04-22 06:38:26 for most harware 2015-04-22 06:38:28 stuff like that is usually built against kernel, not particular libc 2015-04-22 06:38:31 yeah 2015-04-22 06:38:33 well u know 2015-04-22 06:38:41 if mainline linux support it 2015-04-22 06:38:42 broadcom wireless drivers and so on 2015-04-22 06:38:45 then no problem 2015-04-22 06:38:48 well deban would never ship my wireless card 2015-04-22 06:38:52 but i think it worked with alpine 2015-04-22 06:39:06 myabe because we ship firmware 2015-04-22 06:39:07 i had to copy it onto a usb drive 2015-04-22 06:39:09 i think yeah 2015-04-22 06:39:12 okay rad 2015-04-22 06:39:35 so ppl can netinstall from laptops easily 2015-04-22 06:40:00 wow this is rad 2015-04-22 06:40:13 i really hope this distro continues to grow 2015-04-22 06:40:22 in light of all the chaos in linux-land 2015-04-22 06:40:49 it's a very innovative yet stabilizing set of design choices 2015-04-22 06:41:42 liek netbsd it probaly can run on my laptop and a toaster 2015-04-22 06:41:49 heheh 2015-04-22 06:42:53 well thanks for answering my questions you guys 2015-04-22 06:43:01 ncopa, barthalion, et 2015-04-22 06:43:13 np 2015-04-22 06:43:19 i dont want to keep you longer but you guys rock 2015-04-22 06:43:44 learned a bunch of new stuff in 20 mins that wuld have taken me a day 2015-04-22 06:44:37 so anyway i'll be afk but ill defintely make a pull request if i come up with anything useful for you guys 2015-04-22 06:44:48 would be great 2015-04-22 06:44:52 yeah 2015-04-22 06:44:56 so one lastthing 2015-04-22 06:45:00 feel free to use patches from our aports tree 2015-04-22 06:45:01 i dont see the zfs on linux port 2015-04-22 06:45:08 is someone already almost done with that 2015-04-22 06:45:10 i think we have it in edge/testing 2015-04-22 06:45:12 okay 2015-04-22 06:45:23 cool i'll do somethng else 2015-04-22 06:45:28 might be there is one last piece missing for the 3.18 kernel 2015-04-22 06:45:39 maybe i'll implement userspace crypto for chacha as fuse 2015-04-22 06:45:40 or something 2015-04-22 06:46:11 (chacha + poly are brand new ciphers, super awesome cause they are fast as hell stream ciphers with autentication aka HMAC built in all the time) 2015-04-22 06:46:37 anyway f i have something cool i'll send it to you guys first 2015-04-22 06:46:57 ncopa: what's missing 2015-04-22 06:46:59 jsut zfs? 2015-04-22 06:47:03 yes 2015-04-22 06:47:09 spl built 2015-04-22 06:47:13 really 3.18 supports zfs? 2015-04-22 06:47:14 no way 2015-04-22 06:47:15 zfs did not 2015-04-22 06:47:16 ? 2015-04-22 06:47:18 oh oh 2015-04-22 06:47:20 okay 2015-04-22 06:47:22 i think it does not 2015-04-22 06:47:23 i was surprise 2015-04-22 06:47:31 i fixed some parts 2015-04-22 06:47:38 so i think its pretty close 2015-04-22 06:47:40 yeah i just looked at 3.18 i dindt see much that stood out in the release notes 2015-04-22 06:47:46 looked at it over the weekend 2015-04-22 06:47:47 might even be trivial to make it build 2015-04-22 06:48:00 i think i pushed the patches to edge/testing/zfs-grsec 2015-04-22 06:48:06 something about synaptics touchpads 2015-04-22 06:48:07 i rememer 2015-04-22 06:48:16 oh and btrfs 2015-04-22 06:48:18 that was the big one 2015-04-22 06:48:24 btrfs should work 2015-04-22 06:48:27 btrfs fixes are 3.16->3.18 2015-04-22 06:48:40 cool 2015-04-22 06:48:47 i havent tested it with 3.18 kernel though 2015-04-22 06:49:00 well i'll be afk a bit gotta get this build started while i'm paying $24/day haha 2015-04-22 06:49:02 seeya 2015-04-22 06:49:07 wb ;) 2015-04-22 06:49:07 afk 2015-04-22 06:50:02 whoa whoa 2015-04-22 06:50:04 i'm back 2015-04-22 06:50:07 double take 2015-04-22 06:50:12 you guys dontt have dm-crypt 2015-04-22 06:50:13 ? 2015-04-22 06:50:35 can u do block-encrypted boot devices or is that on hold at the moment 2015-04-22 06:50:48 i see some patches 2015-04-22 06:51:18 and closed bugs 2015-04-22 06:51:19 http://bugs.alpinelinux.org/issues/3599 2015-04-22 06:51:45 anyway hit me up if you guys need someone to fix crypto stuff... i'm pretty good at that sort of thing 2015-04-22 06:52:11 looks like u got it ncopa but i'll be around the irc channel if there are crypto bugs again 2015-04-22 06:52:11 afk 2015-04-22 07:09:39 issues in the current edge packagings: http://dev.alpinelinux.org/~tteras/index-errors.png 2015-04-22 07:11:53 how come there is a problem between and boost and its library? 2015-04-22 07:12:11 ah, it's a circular dep 2015-04-22 07:12:12 got it 2015-04-22 07:12:59 I don't get seafile-common though 2015-04-22 07:16:37 it depends on it self 2015-04-22 07:16:43 either directly, or via some provides 2015-04-22 07:16:49 should be direct 2015-04-22 07:16:58 as depends via provides are annotated 2015-04-22 07:17:01 like for boost 2015-04-22 07:17:27 http://dev.alpinelinux.org/~tteras/index-errors-main.png for errors in main repo only 2015-04-22 07:18:00 I see 2015-04-22 07:18:20 some are conflict deps. like musl-* conflicts with uclibc-* 2015-04-22 07:43:15 fabled: the red parts are non existing deps? 2015-04-22 07:43:26 or circular 2015-04-22 07:43:44 clandmeter, boxesin red are non-existant; dependency arrows in red are circular 2015-04-22 07:44:06 red-dotted boxes that is 2015-04-22 07:44:13 ok 2015-04-22 07:45:24 nice 2015-04-22 07:45:26 very nice 2015-04-22 07:45:35 that first obex thingy 2015-04-22 07:45:43 commit 853f2871416a49e7dbe31d0bca29cbbcc9114878 2015-04-22 07:45:43 Author: Natanael Copa 2015-04-22 07:45:43 Date: Tue Oct 29 13:42:16 2013 +0000 2015-04-22 07:45:43 main/obexd: remove. merged into bluez 5 2015-04-22 07:45:43 2015-04-22 07:45:43 http://www.bluez.org/release-of-bluez-5-0/ 2015-04-22 07:45:51 hard to spot otherwise 2015-04-22 07:47:30 oh man 2015-04-22 07:47:37 i hate boost haha 2015-04-22 07:47:50 ubuntu always installed like 50 versions of it to get my stuff to compile 2015-04-22 07:48:18 heh 2015-04-22 07:49:52 ne wayz stuff like that (libs, deps, and the need for parallel namespaces) is why i was studying nix 2015-04-22 07:50:11 and the other similar package managment tools 2015-04-22 07:50:51 i think debian has been the most successful, but what it cant do is install two shared libs side by side because of naming conflicts 2015-04-22 07:51:04 also transactions arent atomic 2015-04-22 07:51:14 so state can become inconsitent 2015-04-22 07:52:21 so i think what you guys just mentioned about libboost , dev environments ("containers"), package managers... If that can be solved in an open source way 2015-04-22 07:52:22 man oh an 2015-04-22 07:52:31 that would be rad 2015-04-22 07:54:58 like a packaging standard so good, and so useful 2015-04-22 07:55:25 that it becomes universal by consent rather than by choice 2015-04-22 07:57:36 'i've been on a package management kick since i started building linux-from-scratch... found two great articles both 'for' and 'against' the use of a stateful, immutable, declarative , and human readable technique for package management 2015-04-22 07:58:13 the article asks broadly, as in "why do we have apt, pip, windows installer, eclipse plugins, gimp plugins, etc etc" 2015-04-22 07:58:16 .. 2015-04-22 07:59:03 but this article makes the point that it's not just state, dependencies, and namespaces that matter 2015-04-22 07:59:42 and it apparently is in favor of something like git 2015-04-22 07:59:45 http://blog.ezyang.com/2014/08/the-fundamental-problem-of-programming-language-package-management/ 2015-04-22 07:59:48 ... 2015-04-22 08:01:28 this second artcle is even more critical of making new package managers, but it states how we really have a large problem... domain specific languages (like config files in /etc) or whatever... and then even our cloud orchestration stuff like puppet or vagrant has their own packages 2015-04-22 08:01:29 ahah 2015-04-22 08:01:31 http://www.standalone-sysadmin.com/blog/2014/03/just-what-we-need-another-package-manager/ 2015-04-22 08:01:34 man 2015-04-22 08:02:17 so much potential for innovation right now in the linux/bsd crossover space 2015-04-22 09:50:49 what's the default policy re: backports? 2015-04-22 09:50:56 it's about #4067 2015-04-22 09:53:30 fcolista: I think it's fine to backport it to 3.1 2015-04-22 09:53:42 fcolista, you may backport any 0.12.x 2015-04-22 09:53:49 but not 0.13 or greater 2015-04-22 09:54:22 granted that you don't need to change config for a 0.12.x update 2015-04-22 09:54:33 i've not a 3.1 build env, only edge. 2015-04-22 09:54:54 feel free to backport it :) 2015-04-22 09:57:30 should we start naming hooks/functions mariadb or still mysql? 2015-04-22 10:43:47 hooks/functions? 2015-04-22 10:43:59 will anyone be sad if i purge mplayer? 2015-04-22 10:44:19 build failure due to fortify i think 2015-04-22 10:44:29 http://sprunge.us/EfeR 2015-04-22 10:44:46 there are probably a define inline somewhere in there 2015-04-22 10:46:13 nobody uses it anyway, I guess 2015-04-22 10:46:25 there is nothing mplayer does that mpv can't 2015-04-22 10:46:35 also smplayer can make use of mpv since some time, so 2015-04-22 13:49:28 ncopa: I know I had offered to help with lua- packages, but I haven't had time 2015-04-22 13:49:32 just apologizing 2015-04-22 13:53:10 np 2015-04-22 13:53:19 please: http://sprunge.us/aJXT 2015-04-22 13:53:32 :) 2015-04-22 13:53:35 http://sprunge.us/DjiV 2015-04-22 13:53:36 :) 2015-04-22 14:02:04 reboot arm builder 2015-04-22 14:02:09 it'll be off briefly 2015-04-22 14:14:39 ok 2015-04-22 14:14:41 looks ok 2015-04-22 14:14:47 new arm edge builds: 2015-04-22 14:14:49 http://dev.alpinelinux.org/~tteras/alpine-rpi-150422-armhf.rpi.tar.gz 2015-04-22 14:14:55 ncopa: please add him to the ml 2015-04-22 14:15:01 http://dev.alpinelinux.org/~tteras/alpine-uboot-150422-armhf.tar.gz 2015-04-22 14:15:18 rpi image boots on rpi1+rpi2 2015-04-22 14:15:53 uboot boots on wandboard, and potentially also cubie/beagle 2015-04-22 14:16:01 ncopa: patches@patchwork.alpinelinux.org 2015-04-22 14:16:17 fabled: nice 2015-04-22 14:17:01 rcpt to: 2015-04-22 14:17:01 250 2.1.5 Ok 2015-04-22 14:17:02 ok 2015-04-22 14:17:05 i will add it 2015-04-22 14:17:06 thanks! 2015-04-22 14:17:33 i absolutly dont know how this thing works 2015-04-22 14:17:36 :D 2015-04-22 14:17:46 umh 2015-04-22 14:17:56 kudos to fcolista, he did most of it. 2015-04-22 14:17:58 patch does not appears in patchwork 2015-04-22 14:18:11 which patch? 2015-04-22 14:18:20 did you sent an email? 2015-04-22 14:18:30 http://patchwork.alpinelinux.org/project/aports/list/ shows one patch 2015-04-22 14:18:31 yes an ampty one 2015-04-22 14:18:45 do I need to resend the patches I sent to aports ? 2015-04-22 14:18:48 it should only accept mails from ML 2015-04-22 14:18:53 BitL0G1c: no 2015-04-22 14:18:58 we will try import them 2015-04-22 14:19:07 clandmeter, yeah you're right. 2015-04-22 14:19:13 we == ncopa ;-) 2015-04-22 14:19:16 So id you sent directly an email it should not accept it 2015-04-22 14:19:48 i was just checking if mail routing was correct 2015-04-22 14:23:47 i injected first email 2015-04-22 14:23:48 nice 2015-04-22 14:24:55 nice 2015-04-22 14:25:06 fcolista: we also need to have aports local 2015-04-22 14:25:18 fetch commits and auto handle patches 2015-04-22 14:25:55 and because you are our offical patchwork admin... ;-) 2015-04-22 14:26:24 http://patchwork.alpinelinux.org/patch/2/ 2015-04-22 14:26:25 nice :) 2015-04-22 14:26:59 i think i need an hand to setup this workflow 2015-04-22 14:28:16 did the cron part work? 2015-04-22 14:28:27 no 2015-04-22 14:39:22 hum 2015-04-22 14:39:24 clandmeter 2015-04-22 14:39:30 i now get: 450 4.1.2 : Recipient address rejected: Domain not found 2015-04-22 14:39:53 hmm 2015-04-22 14:40:10 i didnt change anything 2015-04-22 14:40:16 fcolista: ? 2015-04-22 14:40:32 didn't touch 2015-04-22 14:40:42 wird 2015-04-22 14:40:45 weird 2015-04-22 14:41:02 ncopa, from who are you receiving this error? 2015-04-22 14:41:22 ncopa-desktop:~$ nc patchwork.alpinelinux.org 25 2015-04-22 14:41:22 220 bugs.alpinelinux.org ESMTP Postfix 2015-04-22 14:41:22 helo tanael.org 2015-04-22 14:41:22 250 bugs.alpinelinux.org 2015-04-22 14:41:22 mail from: 2015-04-22 14:41:23 250 2.1.0 Ok 2015-04-22 14:41:25 rcpt to: 2015-04-22 14:41:27 450 4.1.2 : Recipient address rejected: Domain not found 2015-04-22 14:41:34 its bugs.a.o 2015-04-22 14:41:36 must be my faulth 2015-04-22 14:41:37 last email that have reached patchwork.alpinelinux.org is 2015-04-22 14:41:39 Apr 22 14:23:07 alpine-patchwork mail.info postfix/local[1119]: 99BF560CEF817: to=, relay=local, delay=0.28, delays=0.05/0/0/0.23, dsn=2.0.0, status=sent (delivered to command: /usr/share/webapps/patchwork/apps/patchwork/bin/parsemail.sh) 2015-04-22 14:41:56 it's another postfix then.. 2015-04-22 14:42:05 yes its bugs.a.o 2015-04-22 14:42:12 clandmeter, are you playing with transport maybe ? 2015-04-22 14:42:50 patchwork.alpinelinux.org smtp:[172.16.4.10] 2015-04-22 14:43:28 dns setting? 2015-04-22 14:43:35 transport_maps = hash:/etc/postfix/transport 2015-04-22 14:43:35 relay_domains = $transport_maps 2015-04-22 14:43:47 did you edit recepient restrictions? 2015-04-22 14:43:52 no 2015-04-22 14:43:58 it worked earlier 2015-04-22 14:44:01 i didnt change anthing since your last email 2015-04-22 14:44:06 :) 2015-04-22 14:44:08 me too 2015-04-22 14:44:10 rcpt to: 2015-04-22 14:44:10 250 2.1.5 Ok 2015-04-22 14:44:16 that was 16.16 2015-04-22 14:44:17 hum 2015-04-22 14:44:53 Recipient address rejected: Domain not found; 2015-04-22 14:44:56 why? 2015-04-22 14:45:02 its in relay_domains 2015-04-22 14:45:08 dns 2015-04-22 14:45:18 can you lookup the dns from bugs.a.o? 2015-04-22 14:45:50 what does dns matter here? 2015-04-22 14:45:57 its not sending to mx 2015-04-22 14:46:04 oh 2015-04-22 14:46:06 wait 2015-04-22 14:46:15 myhostname = patchwork.alpinelinux.org 2015-04-22 14:46:23 #mydomain = patchwork.alpinelinux.org 2015-04-22 14:47:34 ncopa: try now 2015-04-22 14:48:38 450 4.1.2 : Recipient address rejected: Domain not found 2015-04-22 14:48:53 ncopa: its probably my local dns 2015-04-22 14:55:05 ncopa: try again? 2015-04-22 14:55:11 not sure it caches stuff 2015-04-22 14:55:48 250 2.1.5 Ok 2015-04-22 14:56:09 ok 2015-04-22 14:56:11 freaking dns looping 2015-04-22 14:56:19 i think i imported the patches now 2015-04-22 14:57:21 w00t 2015-04-22 14:57:26 algitbot: thx 2015-04-22 15:04:45 how do we use patchwork? 2015-04-22 15:04:48 i registered 2015-04-22 15:04:56 the confirmation url pointed to example.com 2015-04-22 15:07:03 ncopa: we are learning while doing :) 2015-04-22 15:07:09 := 2015-04-22 15:07:12 :) 2015-04-22 15:07:12 ok 2015-04-22 15:07:15 im going home 2015-04-22 15:07:17 see u 2015-04-22 15:07:20 cu 2015-04-22 15:26:59 is there a web interface to view gentoo runscripts ? (or do I need a gentoo vm) 2015-04-22 15:30:20 what do you mean? 2015-04-22 15:40:34 clandmeter - I'm just trying to view a /etc/init.d/xxxxx for a package in gentoo 2015-04-22 15:40:51 I can only find an ebuild 2015-04-22 15:48:30 should be in their cvs 2015-04-22 15:48:49 which pkgs? 2015-04-22 15:50:40 libteam 2015-04-22 15:55:16 so if it's not in here there isn't one https://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/net-misc/libteam/ 2015-04-22 15:55:29 yeah no init there 2015-04-22 16:27:45 great job with patchwork :) 2015-04-22 16:31:11 BitL0G1c: gpo.zugaina.org 2015-04-22 16:33:22 barthalion - many thx - will bookmark it :) 2015-04-22 16:36:09 hrm 2015-04-22 16:36:18 what changes the status of patch on patchwork? 2015-04-22 16:46:57 barthalion: git commit 2015-04-22 16:48:01 isn't there also "on review" thing? 2015-04-22 16:51:21 sure 2015-04-22 16:51:37 atleast i tink there is 2015-04-22 16:52:27 now it just looksup commits and set to accepted if found 2015-04-22 16:54:03 http://sprunge.us/BcZI 2015-04-22 16:54:05 IMHO status should be changed if someone replied to the message 2015-04-22 16:54:08 those are the possible states 2015-04-22 16:54:29 i think thats automatic? 2015-04-22 16:55:03 ah, vino has been merged before I even sent my message 2015-04-22 16:55:07 might be, sorry for noise 2015-04-22 16:55:46 let's check again 2015-04-22 17:00:40 http://patchwork.alpinelinux.org/patch/1/ 2015-04-22 17:00:53 got my message, didn't change the status 2015-04-22 17:14:18 barthalion: ok so what is the rule? 2015-04-22 17:14:33 the reply needs to be a project manager? 2015-04-22 17:16:33 I cant find much info regarding patchwork 2015-04-22 17:16:41 trial and error setup 2015-04-22 17:39:35 I don't know a thing either 2015-04-22 18:02:25 i guess if just somebody replies "wow nice patch" isnt enough to set a diff status 2015-04-22 19:53:19 am i the only one having icu-libs conflicts on edge ? 2015-04-22 19:54:28 http://dev.alpinelinux.org/~tteras/index-errors.png 2015-04-22 20:01:23 barthalion: btw haven't had time to check consul deps packing 2015-04-22 20:01:38 will try to find time to fix this this week 2015-04-22 20:20:25 np 2015-04-22 20:21:12 don't tell that to guy who promises something new every week and actually does it month later :p 2015-04-22 20:33:23 ncopa: how's fortify? any more issues discovered? :) 2015-04-22 20:45:35 clandmeter right now i tried the stable repo to pin and install sfdisk but it fails to do a partition still: http://sprunge.us/HMaG 2015-04-22 20:49:18 crow: can you sprunge world for me? 2015-04-22 20:50:58 sfdisk is part of util-linux 2015-04-22 20:51:06 could it be you used cfdisk? 2015-04-22 20:51:32 well i added sfdisk@stable 2015-04-22 20:51:51 what do you mean sprunge world? 2015-04-22 20:51:55 sprunge < /etc/apk/world 2015-04-22 20:52:23 http://sprunge.us/KFcX 2015-04-22 20:52:49 you have sfdisk from edge 2015-04-22 20:52:53 or atleast not pinned 2015-04-22 20:54:13 well i pined stable repo, and hen apk add -U sfdisk@stable 2015-04-22 20:54:14 Installing sfdisk@stable (2.25.2-r1) 2015-04-22 20:55:52 ok 2015-04-22 20:56:06 what is in your world now? 2015-04-22 20:56:37 and what does sfdisk -v tell you 2015-04-22 20:57:56 that is my world currently (i booted your iso) pined stable repo and did add sfdisk@stable, then setup-disk 2015-04-22 20:58:02 as seen in my paste 2015-04-22 20:58:17 sfdisk from util-linux 2.26.1 2015-04-22 20:58:28 so i need to add util-linux ? 2015-04-22 20:59:16 crow: no 2015-04-22 20:59:34 Installing sfdisk@stable (2.25.2-r1) 2015-04-22 20:59:43 it shows you are installing 2.25 2015-04-22 20:59:49 why it shows you have 2.26? 2015-04-22 20:59:57 yes but sfdisk -v shows this: sfdisk from util-linux 2.26.1 2015-04-22 21:00:20 show me your repository file 2015-04-22 21:00:51 and can you remove cfdisk before installing from stable? 2015-04-22 21:00:52 clandmeter i did, here again before this i did setup-alpine and on hdd part i did none. http://sprunge.us/HMaG 2015-04-22 21:02:19 maybe setup-disks adds it from edge again? 2015-04-22 21:05:00 crow: vim +444 /sbin/setup-disk 2015-04-22 21:05:15 it adds sfdisk 2015-04-22 21:05:25 maybe you can remove it and see if that helps 2015-04-22 21:06:19 ok 2015-04-22 21:10:53 clandmeter that was it, in that way it works, (removing on line 444 the sfdisk from apk add.. http://sprunge.us/ZETG 2015-04-22 21:18:54 weird 2015-04-22 21:19:11 pinned packages shouldnt get replaced 2015-04-22 21:29:33 coredumb: I relly enjoy consul. Hope for envconsul and consul-template ;) 2015-04-22 21:35:36 It solves a lot of problems I had convoluted solution for... 2015-04-22 22:04:28 Jean-Scotch: hehe :P 2015-04-22 22:04:40 it's a very nice piece of software ^^ 2015-04-23 00:03:20 How can I fix "mkdir: can't create directory '/usr/share/ntopng': Permission denied" when building a package with abuild 2015-04-23 00:17:33 How can I fix "mkdir: can't create directory '/usr/share/ntopng': Permission denied" when building a package with abuild 2015-04-23 03:26:08 Hello everybody 2015-04-23 03:26:42 gruntings 2015-04-23 03:28:17 I need a little help with an APKBUILD which generates an unfunctional apk 2015-04-23 03:28:39 I reported a bug 2015-04-23 03:28:46 http://bugs.alpinelinux.org/issues/3751 2015-04-23 03:29:27 But I want to solve it temporarly 2015-04-23 03:30:15 If I compile from source manualy is working 2015-04-23 03:31:40 But if I use APKBUILD everything goes with no error in compiling 2015-04-23 03:32:55 But the resulted binary has problems with it's libraries 2015-04-23 03:35:36 When I compile it manualy fwknop-libs are instaled in /usr/local/lib (I should run as root when I make install) 2015-04-23 03:36:46 Can anybody give me a clue in having a working applucation? 2015-04-23 06:36:01 morning 2015-04-23 06:36:14 Jean-Scotch: re #4015 2015-04-23 06:36:21 how does your boot line look like? 2015-04-23 06:36:33 alpine_dev=nfs ? 2015-04-23 06:37:00 ncopa: think so. I'm checking... 2015-04-23 06:37:42 APPEND ip=dhcp alpine_dev=nfs:192.168.1.149:/srv/boot/alpine modloop=http://192.168.1.189/modloop-grsec nomodeset quiet apkovl=http://192.168.1.189/localhost.apkovl.tar.gz 2015-04-23 06:38:07 UUID=*|LABEL=*|/dev/*|nfs);; 2015-04-23 06:38:11 will not work then 2015-04-23 06:38:21 should be: UUID=*|LABEL=*|/dev/*|nfs:*);; 2015-04-23 06:39:01 the logic for searching fstab also looks broke 2015-04-23 06:40:49 It's not the best logic, I agree, but it's working in real world ;) 2015-04-23 06:41:46 I'm in the process of switching all my boxes to PXEboot based on AL ;) 2015-04-23 06:43:42 My solution is only a dirty hack to solve locally my problem... I do not think it's a distro grade solution ^^ 2015-04-23 06:44:57 I'm polishing it every day a bit. I hope to have a working full cloud solution based on AL and _very_ lightweight for this summer... 2015-04-23 06:52:45 I'm doing it bottom-up, one software piece at the time and only when absolutely needed. Some ash scripting as glue is all needed. 2015-04-23 06:53:26 hum 2015-04-23 06:53:36 i think you are right ...|nfs);; is correct 2015-04-23 06:53:56 your patch looks good 2015-04-23 06:54:12 its the other surrounding code that looks bad 2015-04-23 06:54:30 It took me a while to have it run properly as intended for my use case... 2015-04-23 06:54:58 what is broken though 2015-04-23 06:55:07 And it's easy to hack it manually to bootstrap the first apkovl... 2015-04-23 06:55:08 is if you have UUID=... in /etc/fstab 2015-04-23 06:55:31 and not having the alpine_dev=UUID=... on boot cmdline 2015-04-23 06:55:47 i think your "hack" is correct 2015-04-23 06:56:13 ;) 2015-04-23 06:59:22 + if [ "$fs" = "$search_dev" ]; then 2015-04-23 06:59:22 + echo "$mnt" 2015-04-23 06:59:22 + return 2015-04-23 06:59:22 + fi 2015-04-23 06:59:26 except that hunk though 2015-04-23 06:59:27 hum 2015-04-23 07:00:03 Jean-Scotch: can you paste me the contents of /proc/mounts 2015-04-23 07:00:12 whith the nfs 2015-04-23 07:00:18 grep nfs /proc/mounts 2015-04-23 07:01:57 ncopa: I just spin a new one and: 2015-04-23 07:02:12 localhost:~# mount|grep nfs 2015-04-23 07:02:44 192.168.1.149:/srv/boot/alpine on /media/alpine type nfs (ro,relatime,vers=3,rsize=131072,wsize=131072,namlen=255,hard,nolock,proto=tcp,timeo=600,retrans=2,sec=sys,mountaddr=192.168.1.149,mountvers=3,mountproto=tcp,local_lock=all,addr=192.168.1.149) 2015-04-23 07:03:44 similar for grep nfs /proc/mounts 2015-04-23 07:06:01 usually I do "umount /dev/alpine" at this stage 2015-04-23 07:06:19 sorry "umount /media/alpine" 2015-04-23 07:11:51 so dev = 192.168.1.149:/srv/boot/alpine in /proc/mounts 2015-04-23 07:13:31 Jean-Scotch: I think this should work: http://sprunge.us/fNPR 2015-04-23 07:19:39 ncopa: then shouldn't resolve_dev() be aware of nfs too? or will readlink takes care of that part? 2015-04-23 07:22:14 don't we need to have the value "/media/alpine" as the response from find_mnt()?? 2015-04-23 07:25:04 ncopa-desktop:~$ readlink -f 192.168.8.1:/media/alpine 2015-04-23 07:25:04 ncopa-desktop:~$ 2015-04-23 07:26:14 resolve_dev will return empty if nfs 2015-04-23 07:26:34 for i in "$search_dev" "$search_real_dev"; do 2015-04-23 07:26:34 [ -z "$i" ] && continue 2015-04-23 07:26:34 for j in "$dev" "$real_dev"; do 2015-04-23 07:26:34 if [ "$i" = "$j" ]; then 2015-04-23 07:26:34 [ -z "$j" ] && continue 2015-04-23 07:26:34 echo "$mnt" 2015-04-23 07:26:36 return 2015-04-23 07:26:38 fi 2015-04-23 07:26:40 done 2015-04-23 07:26:42 done 2015-04-23 07:26:58 search_dev=192.168.1.149:/srv/boot/alpine 2015-04-23 07:27:15 and dev=192.168.1.149:/srv/boot/alpine 2015-04-23 07:27:45 both search_real_dev and real_dev will be "" 2015-04-23 07:27:46 empty 2015-04-23 07:27:59 ok 2015-04-23 07:28:15 so the loop there will find it 2015-04-23 07:29:57 seems good 2015-04-23 07:34:06 Jean-Scotch: i think you might be able to set rc_need=networking or rc_use=networking in /etc/conf.d/modloop 2015-04-23 07:34:23 to keep networking up til after modloop is stopped 2015-04-23 07:34:35 ha! cool! 2015-04-23 07:49:33 seems setup-alpine in edge does not ask anymore for the hostname 2015-04-23 07:58:25 huh? 2015-04-23 08:00:43 :) 2015-04-23 08:03:29 ncopa used iso from clandmeter http://dev.alpinelinux.org/~clandmeter , start the installation with setup-apline , you will not be prompted to enter hostname 2015-04-23 08:04:34 ncopa just for the info: i used @stable repo pining but seemt setup-disk script adds and overwrites the installed sfdisk@stable: http://sprunge.us/ZETG 2015-04-23 08:06:45 ncopa: setup-disk is indeed broken because of changes to sfdisk. you cannot specify bytes anymore. and also noticed it didnt ask about hostname, though i didnt check whats wrong. 2015-04-23 08:08:47 This version provides a completely new sfdisk(8) command; the new version is based on libfdisk. If your use cases depend on sfdisk(8), then it is strongly 2015-04-23 08:08:50 recommended to be careful and re-test your scripts. The new version supports 2015-04-23 08:08:53 MBR and GPT disk labels (SGI and SUN are also supported but not well tested). 2015-04-23 08:08:56 The new version no longer supports some obscure MBR-specific command-line 2015-04-23 08:08:58 options nor legacy CHS addressing. 2015-04-23 08:12:13 clandmeter thanks for the confirmation :) 2015-04-23 09:02:35 vim does something interestingwith fortify source 2015-04-23 09:02:37 -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=1 2015-04-23 09:15:41 Does anyone have an idea how can I temporarly solve the issues from this bug? http://bugs.alpinelinux.org/issues/3751 Thanks 2015-04-23 09:36:34 giuliano77: looks like we have fwknop in testing 2015-04-23 09:37:45 Yes, it worked in version 2.6.2 2015-04-23 09:38:13 "I compiled the apk from source using Stuart's APKBUILD and the error is the same. Then I entered the source folder, compiled the application manually and it's indeed working but the libraries are installed in /usr/local/lib not in /usr/lib." 2015-04-23 09:38:29 built here libraries goes into right place, /usr/lib 2015-04-23 09:38:30 But is not working anymore 2015-04-23 09:38:48 how can i set up a simple test case? 2015-04-23 09:39:09 Just build it 2015-04-23 09:39:17 i did 2015-04-23 09:39:49 ok 2015-04-23 09:40:05 fwknop: fko_spa_data_final: Error 95 - Invalid key length 2015-04-23 09:40:25 And run the command for generate keys and save a stanza 2015-04-23 09:40:45 It will triger the error 2015-04-23 09:41:01 If build manually 2015-04-23 09:41:18 Everything works fine 2015-04-23 09:43:06 # If using gnupg2 see http://www.gnupg.org/faq/GnuPG-FAQ.html#how-can-i-use-gnupg-in-an-automated-environment 2015-04-23 09:43:06 # & sign the client public key on the server with a normal key which has a password 2015-04-23 09:43:06 # disable GPG_DECRYPT_PW & enable the following variable: 2015-04-23 09:43:06 #GPG_ALLOW_NO_PW: Y; 2015-04-23 09:43:19 But just when run make install as root 2015-04-23 09:43:32 ho do you build it when you build it manually? 2015-04-23 09:44:10 .configure, make and make install 2015-04-23 09:44:25 so you dont have --with-gpgme 2015-04-23 09:44:38 nor --enable-afl-fuzzing 2015-04-23 09:44:39 No 2015-04-23 09:44:45 No 2015-04-23 09:46:40 http://sprunge.us/SPVa 2015-04-23 09:46:51 that happens if i remove gpgme 2015-04-23 09:47:38 aha 2015-04-23 09:47:47 its not gpgme 2015-04-23 09:48:14 Afl fuzzing? 2015-04-23 09:48:17 its the --enable-afl-fuzzing 2015-04-23 09:48:43 yup 2015-04-23 09:48:44 So should I disable it in APKBUILD 2015-04-23 09:49:24 what does it do? 2015-04-23 09:49:30 Are major security disatvantages? 2015-04-23 09:49:37 why was it added in first place? 2015-04-23 09:50:03 I dob't know 2015-04-23 09:50:31 Preventig fuzzing attacs 2015-04-23 09:50:42 Probbably 2015-04-23 09:51:18 [server] Added support for the 'American Fuzzy Lop' (AFL) fuzzer from Michal Zalewski. This requires that fwknop is compiled with the '--enable-afl-fuzzing' argument to the configure script as this allows encryption/digest short circuiting in a manner necessary for AFL to function properly. The benefit of this strategy is that AFL can fuzz the SPA packet decoding routines implemented by libfko. See the test/afl/ directory for some automation around AFL 2015-04-23 09:51:18 fuzzing. 2015-04-23 09:51:29 I will talk with mr. Rash 2015-04-23 09:51:41 introduced with 2.6.4 2015-04-23 09:51:48 i can disable it for now 2015-04-23 09:52:31 introduced with http://dup.pw/79ec102ae1 2015-04-23 09:53:05 BitL0G1c: it was you patch wasnt it? 2015-04-23 09:53:06 Ok, I will see what is about 2015-04-23 09:53:16 i can disable it meanwhile 2015-04-23 09:53:30 Thanks 2015-04-23 12:36:50 will http://www.gnustep.org/index.html be useful ? 2015-04-23 12:59:29 main/task update to 2.4.3 pls 2015-04-23 13:02:41 wow, nice nginx released v1.8.0, with new features 2015-04-23 14:31:49 Frosh, tested and working: http://wiki.alpinelinux.org/wiki/MATE 2015-04-23 14:31:50 ;) 2015-04-23 17:00:34 clandmeter - ok - will resubmit the albatross-themes patch with the missing 1.73 patch included - will be sending the networkmanager patches first 2015-04-23 17:53:18 alacerda, sorry been busy with the new team 2015-04-23 18:08:04 clandmeter - the patches for spice-gtk virt-viewer spice-vdagent sent on 20th have not been imported into patchwork yet 2015-04-23 18:18:01 whats the best way to get linux-dev from the current iso? 2015-04-23 18:18:16 hi 2015-04-23 18:23:48 shafire, did you call me yesterday?: 2015-04-23 18:28:53 yeah, because of a package, but I forgot already why 2015-04-23 18:29:20 shafire - http://nl.alpinelinux.org/alpine/v3.1/main/x86_64/ 2015-04-23 18:29:53 BitL0G1c: mh, is the kernel for 3.1 all the same? 2015-04-23 18:31:06 i think it just changed to 3.18 recently 2015-04-23 18:31:39 thats the problem 2015-04-23 18:31:50 I use alpine from usb 2015-04-23 18:31:54 3.14 http://nl.alpinelinux.org/alpine/v3.1/main/x86_64/dahdi-linux-grsec-dev-3.14.39-r0.apk 2015-04-23 18:32:04 but I need to build a kernel module before updating 2015-04-23 18:32:12 oops http://nl.alpinelinux.org/alpine/v3.1/main/x86_64/linux-grsec-dev-3.14.39-r0.apk 2015-04-23 18:34:20 3.1.1, 3.1.2 and 3.1.3 uses all 3.14.39? 2015-04-23 18:34:26 I don't think so :S 2015-04-23 18:37:37 my usb has linux-grsec-3.14.36-r0.apk 2015-04-23 18:38:46 Could I just download 3.14.36 from kernel, build the module and load it with linux-grsec-3.14.36-r0.apk? 2015-04-23 18:38:53 from kernel.org* 2015-04-23 18:41:04 for only a module yes 2015-04-23 18:42:36 so, the module does not need a grsec patched kernel? 2015-04-23 18:42:45 if it doesn't when i had to recreate a kernel - it was @ http://git.alpinelinux.org/cgit/aports/plain/main/linux-grsec/grsecurity-2.9.1-3.10.28-unofficial.patch?h=2.7-stable 2015-04-23 18:43:21 if you are only building a single module - you may be lucky 2015-04-23 18:43:39 hope this works 2015-04-23 18:43:48 back after dinner.... - fingers crossed 2015-04-23 20:07:25 its so bad that I can not get access to the linux-*-dev.apk of the current iso :S 2015-04-23 20:07:37 Hi guys, is there any issues with dl-4 repository mirror? 2015-04-23 20:07:50 shafire you can switch to edge and get newer kernel 2015-04-23 20:08:09 crow: I can not and I don't want 2015-04-23 20:08:12 I use alpine from usb 2015-04-23 20:08:33 you mean run from ram? 2015-04-23 20:09:11 there is also option to pin edge repo and then apk -U linux@edge 2015-04-23 20:09:29 yes, I run from ram 2015-04-23 20:09:30 but if you don't want... 2015-04-23 20:10:51 whats the difference between linux-virtgrsec and linux-grsec 2015-04-23 20:10:51 ? 2015-04-23 20:10:55 you could also download edge iso and do setup-bootable on that same usb 2015-04-23 20:12:33 no :D 2015-04-23 20:12:53 I want to use lsb 2015-04-23 20:13:04 lsb? 2015-04-23 20:13:15 lbu, sry 2015-04-23 20:13:31 edge should use that same lbu config 2015-04-23 20:13:55 ah, now I understand 2015-04-23 20:14:20 but thats also not what I want... 2015-04-23 20:21:37 shafire http://git.alpinelinux.org/cgit/aports/commit/?h=3.1-stable&id=87c3cfeb898610ff07cc1613045a9c195a180b1b - you can recreate the 3.14.36 from that info 2015-04-23 20:26:31 download this & build it http://git.alpinelinux.org/cgit/aports/snapshot/aports-87c3cfeb898610ff07cc1613045a9c195a180b1b.tar.bz2 2015-04-23 20:27:20 all the files you need are in main/linux-grsec 2015-04-23 20:27:45 ok 2015-04-23 20:27:47 I will try 2015-04-23 20:28:24 just create an Alpine LXC container - this is all you need for a build environment 2015-04-23 20:30:46 http://git.alpinelinux.org/cgit/aports/tag/?id=v3.1.3 2015-04-23 20:30:49 I see 2015-04-23 20:32:14 my 3.1.3 usb has linux-grsec-3.14.36-r0 2015-04-23 20:36:23 ok 2015-04-23 20:37:28 check what version linux-grsec you have in the apk folder on your usb 2015-04-23 21:14:10 the usb stick is using grsec in usb root or not? 2015-04-23 21:36:09 the linux-grsec in your apk folder on your usb stick will tell you the kernel it is using 2015-04-23 21:37:56 yes in the root 2015-04-23 21:38:23 /apks/x86_64 or x86 2015-04-23 21:41:38 if you are actually in the running usb stick now - /media/sda1/apks 2015-04-23 22:05:49 BitL0G1c: okay thanks 2015-04-23 22:11:46 if you have linux-grsec-3.14.36-r0.apk as i suspect - download http://git.alpinelinux.org/cgit/aports/snapshot/aports-87c3cfeb898610ff07cc1613045a9c195a180b1b.tar.bz2 - extract main/linux-grsec - install 'alpine-sdk' & then in the same directory as the APKBUILD run 'abuild -r' 2015-04-23 22:12:59 the built packages will be in ~/packages/main/x*/ 2015-04-24 06:03:40 hi 2015-04-24 07:41:32 I tried to build the flashcache-grsec package 2015-04-24 07:41:37 but I get symlink errors 2015-04-24 07:43:06 when I build the grsec-kernel and build the flashcache module, compiling worked 2015-04-24 07:43:17 but insmod gives also symlink errors back 2015-04-24 07:43:35 shafire: did you install the linux-grsec-dev package? 2015-04-24 07:43:41 yes 2015-04-24 07:44:00 lxc-start 1429860883.653 ERROR lxc_conf - conf.c:mount_entry:1711 - No such file or directory - failed to mount 'shm' on '/usr/lib/lxc/rootfs/dev/shm' 2015-04-24 07:44:11 did anyone remember what the fix for that was? 2015-04-24 07:46:30 flashcache: Unknown symbol dm_put_device (error 0) 2015-04-24 07:46:58 Did I forget to install a dev package? 2015-04-24 07:48:50 how do you build it? 2015-04-24 07:48:54 do you use the apkbuild? 2015-04-24 07:49:43 I checked git aports 3.1.3 out 2015-04-24 07:50:06 I built the linux-grsec package via abuild -r 2015-04-24 07:50:17 I installed the new linux-grsec-dev package 2015-04-24 07:50:50 I wanted to build flashcache-grsec for the new linux-grsec, but I get these symbol errors 2015-04-24 08:49:23 2015-04-24 09:55:19 please git pull git://git.alpinelinux.org/fab/aports 2015-04-24 09:55:26 there are some updates 2015-04-24 09:58:32 clandmeter: I took a look at patchwork code 2015-04-24 09:58:46 apparently you need to send a message with custom header to change patch state 2015-04-24 09:58:54 patch.state = get_state(mail.get('X-Patchwork-State', '').strip()) 2015-04-24 09:59:14 I believe we can patch it, as this function checks whether the mail is a patch or a comment 2015-04-24 09:59:32 so patchwork will set it to "under review" if any respond to patch has been sent 2015-04-24 10:04:08 barthalion: does it matter who replies? 2015-04-24 10:04:26 no, only the header matters 2015-04-24 10:05:22 do we want that? 2015-04-24 10:05:47 I think we want to have it under review if anyone replied 2015-04-24 10:06:25 or at least from a.o domain + other commiters that doesn't use it 2015-04-24 10:06:54 I dislike the idea of having to send a custom header to change the state 2015-04-24 10:14:29 ncopa: stop breaking stuff and go to lunch :p 2015-04-24 10:18:06 barthalion: we could "force" ppl to use a.o domain if they have commit access 2015-04-24 10:19:00 pwclient is handy 2015-04-24 10:19:26 i fixed automatic aproving 2015-04-24 10:19:40 so the list gets cleaned automtically now 2015-04-24 10:19:41 why do you want to force me :( 2015-04-24 10:20:15 I don't want to think how to configure my e-mail client to use a.o alias as sender 2015-04-24 10:21:59 it was just a suggestion, im open to ideas 2015-04-24 10:23:04 you can also have a static list of commiters 2015-04-24 10:23:17 I think anyone can do a code review, not only "we" as in "people with commit access" 2015-04-24 10:23:20 hi 2015-04-24 10:23:28 the same way you can be a developer without it 2015-04-24 10:25:53 ok, so what is your suggestion? 2015-04-24 10:30:18 just set it as "under review" after any reply 2015-04-24 10:30:53 I don't think I've seen someone replying "wow nice patch" yet 2015-04-24 10:34:06 i did ;-) 2015-04-24 10:34:14 ok 2015-04-24 10:34:21 i think you are right 2015-04-24 10:34:34 will you provide a patch? 2015-04-24 12:24:19 Hi guys, could you please help me with issue while building a package? 2015-04-24 12:24:19 $ abuild build 2015-04-24 12:24:19 checking build system type... Invalid configuration `x86_64-alpine-linux-musl': machine `x86_64-alpine-linux' not recognized 2015-04-24 12:24:19 What have I missed? (abuild -r works just fine, but it removes dependencies at the end) 2015-04-24 12:27:31 frol_ abuild unpack && abuild prepare && abuild build 2015-04-24 12:27:49 or just disable deps removal in abuild.conf 2015-04-24 12:28:33 Cheers! I missed prepare stage 2015-04-24 12:28:53 Thank you a lot 2015-04-24 12:29:32 np 2015-04-24 12:30:10 Does anybody know if krb5 (Kerberos) was disabled for some good reason in freeradius3 package? 2015-04-24 12:38:54 barthalion: will you provide a patch? or is it possible to enable it somehow? 2015-04-24 12:39:28 I'll patch it 2015-04-24 12:43:27 frol_: cause nobody ever had use for it? 2015-04-24 12:44:13 It seems that my patch is ready! :) 2015-04-24 12:44:32 Let me try to contribute properly 2015-04-24 12:45:08 looks like we area also few minors behind 2015-04-24 12:45:22 barthalion, http://sprunge.us/cfWX 2015-04-24 12:45:27 if this is ok, i can push 2015-04-24 12:45:47 nevermind 2015-04-24 12:45:51 i forgot subpkg 2015-04-24 12:47:41 I don't think you wanted to send it to me 2015-04-24 12:47:54 more to frol_ I guess 2015-04-24 12:49:04 no actually were you :) 2015-04-24 12:49:05 I'll patch it 2015-04-24 12:49:34 :) 2015-04-24 12:49:39 its about patchwork 2015-04-24 12:49:44 he will patch it 2015-04-24 12:50:02 ah...i tought he was going to patch freerad with kerberos 2015-04-24 12:50:03 ok 2015-04-24 12:50:03 fcolista: read up a bit 2015-04-24 12:50:29 seems frol_ alraedy made a patch and want to send it to ml (my guess) 2015-04-24 12:51:04 yes, didn't read the backlog 2015-04-24 12:51:40 maybe you can pwclient git-am # ;-) 2015-04-24 12:51:49 exactly :D 2015-04-24 12:52:16 pwclient pulls patches from patchwork? 2015-04-24 12:52:18 sounds nice 2015-04-24 12:52:37 yes :) 2015-04-24 12:52:42 i like pwclient a lot 2015-04-24 12:55:25 seems our lives got much easier now 2015-04-24 12:58:36 I've just sent the patch! 2015-04-24 12:59:10 My first patch via git send-email :) 2015-04-24 13:01:48 frol_: missing pkgdesc in subpkg 2015-04-24 13:03:00 subpkg function is wrong 2015-04-24 13:03:11 $pkgname-krb5 2015-04-24 13:03:18 this is subpkg 2015-04-24 13:03:19 fcolista: I've just noticed that :( 2015-04-24 13:03:23 function is called ldap() 2015-04-24 13:03:40 I will fix both 2015-04-24 13:03:46 yes correct 2015-04-24 13:04:17 clandmeter, I'm gonna set the state "rejected" :) 2015-04-24 13:04:36 seems like other subpkgs also missing pkgdesc 2015-04-24 13:04:50 i think its better if they all have it 2015-04-24 13:04:55 +1 2015-04-24 13:04:57 I will add those pkgdesc too 2015-04-24 13:05:41 clandmeter, what happens if i set "rejected" to pw? 2015-04-24 13:05:46 I'm gonna try 2015-04-24 13:06:18 it wll change status to rejected. thats it. 2015-04-24 13:06:47 I mean: is the dev informed about that? 2015-04-24 13:06:57 frol in this case? 2015-04-24 13:07:23 looks like that 2015-04-24 13:07:25 the only logic behind it is, when New or Under Review, it will be listed with default filters 2015-04-24 13:07:29 in patchwork 2015-04-24 13:07:42 yes 2015-04-24 13:07:49 but there is a 10 min delay 2015-04-24 13:07:57 for emails 2015-04-24 13:08:18 i think pw will merge notifictions within 10 minutes 2015-04-24 13:08:33 so it doesnt spam too much 2015-04-24 13:10:26 would be nice if we could reject it with a reason. and patchwork will send mail to ml with reason. 2015-04-24 13:10:54 yeah. I was looking for a comment box for that purpose. 2015-04-24 13:12:49 hi 2015-04-24 13:12:56 hi 2015-04-24 13:13:12 how can I rebuild the APKINDEX? 2015-04-24 13:13:15 with apk or abuild? 2015-04-24 13:13:33 How about now? 2015-04-24 13:14:26 shafire: apk index 2015-04-24 13:15:14 ERROR: Will not write binary index to console without --force 2015-04-24 13:16:09 apk index --help 2015-04-24 13:16:22 you need to redirect it 2015-04-24 13:16:36 I see 2015-04-24 13:19:29 shafire: dont forget to specify which apk's you want to index 2015-04-24 13:20:47 fcolista: clandmeter: does my patch look good now? 2015-04-24 13:21:40 why do I get WARNING: Ignoring /home/dpr/packages/main/x86_64/APKINDEX.tar.gz: BAD signature :S 2015-04-24 13:22:00 index is singed 2015-04-24 13:22:06 signed 2015-04-24 13:22:19 frol_, looks fine, but some cosmetic changes are needed IMHO. 2015-04-24 13:22:21 you need to put your pubkey in etc/apk/keys 2015-04-24 13:22:34 pubkey of abuild 2015-04-24 13:22:41 fcolista: let me fix those tabs/spaces 2015-04-24 13:23:35 frol_, you got it ;) 2015-04-24 13:23:49 pkgdesc tab 2015-04-24 13:24:36 Also first letter in pkgdesc can be in Caps (LDAP is all in caps, perl is all in lowercase) 2015-04-24 13:24:44 but this is IMHO 2015-04-24 13:25:02 i would set "Perl", "Python", "Client" 2015-04-24 13:25:16 and "Kerberos...bla bla bla" as well 2015-04-24 13:25:41 but, again, it's just mho 2015-04-24 13:26:56 I copied that from Ubuntu... but now as you have said it, I will fix this to be better 2015-04-24 13:31:19 fcolista: one more 2015-04-24 13:34:45 nice weekend! 2015-04-24 13:36:17 clandmeter: the key is there, but still giving me the message :S 2015-04-24 13:39:24 http://techdeem.com/gigabyte-first-arm-based-server-released/ 2015-04-24 13:41:02 vkrishn: does it have already a price? 2015-04-24 13:44:57 WARNING: Ignoring /home/dpr/packages/main/x86_64/APKINDEX.tar.gz: UNTRUSTED signature 2015-04-24 13:45:06 is it possible that it do not accept my key for main? 2015-04-24 13:46:39 vkrishn: I'm curious why does it have CPU cooler at all... and does it is limited to <4GB of RAM? 2015-04-24 13:46:57 I could not find, still trying 2015-04-24 13:47:04 vkrishn: I'm curious why does it have CPU cooler at all... and does it is limited to <4GB of RAM? 2015-04-24 13:48:14 clandmeter: could you please review my last patch? http://patchwork.alpinelinux.org/patch/59/ 2015-04-24 13:48:17 1.7GHz Quad-Core processor 2015-04-24 13:48:17 frol_, thanks. tabs in prepare() are too much 2015-04-24 13:48:33 128Gb ram 2015-04-24 13:48:47 the indentation you had before was ok 2015-04-24 13:48:51 no need to change it 2015-04-24 13:49:58 fcolista: I've just decided that mixing spaces and tabs is generaly bad, so I fixed all of them 2015-04-24 13:51:18 fcolista: should I keep spaces only in prepare() ? 2015-04-24 13:52:39 frol_, you can keep the original indentation of prepare() 2015-04-24 13:52:49 dunno if ncopa has different idea tough 2015-04-24 13:53:13 ok, I will send a new patch soon :) 2015-04-24 13:53:29 can apk/abuild delete a package that I build and rebuild the index? 2015-04-24 13:55:25 apk index -o APKINDEX.tar.gz 2015-04-24 13:55:28 this is wrong or? 2015-04-24 13:55:48 fcolista: done 2015-04-24 13:57:19 good job frol_ 2015-04-24 13:59:10 better spec, http://www.cnx-software.com/2015/03/27/gigabyte-mp30-ar0-is-an-arm-server-motherboard-powered-by-applied-micro-x-gene-1-soc/ 2015-04-24 15:00:16 clandmeter: how do you build a new index? apk index do not do what I want, I think 2015-04-24 15:00:32 clandmeter: I want to build the APKINDEX.tar.gz 2015-04-24 17:18:18 hi 2015-04-24 17:18:28 someone already tried to compile openjdk8? 2015-04-24 17:57:46 alpine3:~/lib/modules/3.14.36-0-grsec/extra/flashcache# insmod flashcache.ko insmod: can't insert 'flashcache.ko': unknown symbol in module, or unknown parameter alpine3:~/lib/modules/3.14.36-0-grsec/extra/flashcache# dmesg | tail [ 823.925481] flashcache: Unknown symbol dm_io_client_create (err 0) 2015-04-24 17:57:48 shit :( 2015-04-24 18:19:25 how do you build a iso for a specific version? 2015-04-24 19:02:23 I think, I will switch from flashcache to lvm-cache 2015-04-24 19:02:28 then I have no problems anymore 2015-04-24 19:54:31 please, merge my patchwork http://patchwork.alpinelinux.org/patch/60/ 2015-04-25 09:20:02 http://mirror.yandex.ru/ <- no alpine anymore? 2015-04-25 09:20:16 http://mirror.yandex.ru/mirrors/alpine/ 2015-04-25 09:20:17 ah ok 2015-04-25 10:44:22 Hello BitL0G1c, could you give me an update on nm/nmtui; thanks. 2015-04-25 11:50:10 Hello BitL0G1c, could you give me an update on nm/nmtui; thanks. - I'm not familiar with IRC; seeing that you've joined, I've repeated my request/question. 2015-04-25 12:09:24 Hi trebor_ : it was a bit more work than I thought but you can see the status of the patches at http://patchwork.alpinelinux.org 2015-04-25 12:10:15 I tested the daemon starting / stopping & the nmtui ncurses interface worked 2015-04-25 12:10:58 there is a bug in the nmtui code - if you have not configured any connections & then try to edit one it will crash 2015-04-25 12:11:26 setting up new connections & then editing them is fine 2015-04-25 12:11:27 Hi BitL0G1c, thanks, I'll have a look at it. 2015-04-25 12:14:01 it now supports bonding with libteam which should have slightly lower overhead than traditional interface bonding 2015-04-25 12:15:59 Any chance of reducing dependencies to support only the ncurses stuff? - Not that I've looked at the 1.0 apk, but the 0.9.8.10-rc1 had quite a dependencies. 2015-04-25 12:25:34 I'm interested in using it in an tmpfs-initrd where a layman can setup networking before kexecing to the "real" system. I've been scripting a custom environment with debootstrap, but I've become fed up with systemd, hence I looked for an alternative to debian/ubuntu and found alpine; I really like what I've seen so far. 2015-04-25 12:29:35 you can setup a build environment in LXC & rebuild it how you wish. It has more dependencies - building without systemd was a bit of a challenge as it seems to depend on it 2015-04-25 12:31:35 Can "http://patchwork.alpinelinux.org" be used in "/etc/apk/repositories"? 2015-04-25 12:34:45 I'm in a qemu right now. I'll have to read the docs on how to setup a build environment; as I said, I'm very new to alpine linux. 2015-04-25 12:40:24 you an setup an alpine lxc container on any linux - this works best for me as a dev environment 2015-04-25 12:49:34 I haven't really played with lxc yet, I'll have to give it a try. Unfortunately, I'm not really a developer. I'm quite fluent at shell scripting, but not much else. All but the most trivial patching needed to get stuff compiled is a stopper for me; not to mention patching needed to get stuff to compile with musl. 2015-04-25 13:07:44 I've been looking at your patches. It seems, that it really was more than quite a bit of work. I really appreciate your help/work, thanks! 2015-04-25 14:11:15 wait, what 2015-04-25 14:11:19 yandex is mirroring us? 2015-04-25 14:11:27 cool 2015-04-25 14:11:41 (y) 2015-04-25 14:14:26 why does lbu want to save directories? 2015-04-25 14:14:40 A etc/ssl A etc/ssl/misc A etc/ssl/private 2015-04-25 14:14:56 lbu ls shows: etc/ssh/ etc/ssh/sshd_config etc/ssh/ssh_host_dsa_key 2015-04-25 14:20:08 maybe a bug? 2015-04-25 14:20:13 lbu ci did not save these folders 2015-04-25 14:20:15 :S 2015-04-25 14:54:20 Hi, I'm trying to create a package for mono, but I'm failing directly by following the wiki http://pastebin.com/HKUBHABs 2015-04-25 14:57:50 we don't support gliderlabs image 2015-04-25 14:58:02 talk to andyshinn or on #gliderlabs 2015-04-25 14:58:21 however, looks like you just should use apk add -U alpine-sdk instead 2015-04-25 14:58:32 weoooo: ↑ 2015-04-25 14:58:40 I see thanks. It's more of a basic question. 2015-04-25 14:58:46 About alpine than the image of it. 2015-04-25 14:58:54 Did you see my comment on the issue tracker about mono? 2015-04-25 14:59:14 weoooo: have you been the commentor in bugs? 2015-04-25 14:59:16 most likely not, I rarely read content there 2015-04-25 14:59:16 for mono 2015-04-25 14:59:31 Yes. 2015-04-25 14:59:48 If you need any assistance, I'd happily help. 2015-04-25 14:59:58 are you interested in mono? 2015-04-25 15:00:01 Yes. 2015-04-25 15:00:08 llvm? 2015-04-25 15:00:11 No. 2015-04-25 15:00:18 just normal mono 2015-04-25 15:00:20 ok 2015-04-25 15:00:22 Only-defaults, what's supported and nothing else. 2015-04-25 15:00:35 I'm trying to create a really stable prod env 2015-04-25 15:00:46 And according to the mono folks LLVM is far gone, but not there yet. 2015-04-25 15:00:50 do you program in c# or f#? 2015-04-25 15:00:52 F# 2015-04-25 15:01:07 I like functional programmers :) 2015-04-25 15:01:15 Alrighty ;) 2015-04-25 15:02:13 I'm building https://github.com/SuaveIO/ 2015-04-25 15:02:29 It's a tiny, async, F# web server that runs cross-platform. 2015-04-25 15:02:45 With applicative-functor style syntax, inspired by Happstack and Snap FFW 2015-04-25 15:02:47 FW 2015-04-25 15:03:39 are you ademar or haf? 2015-04-25 15:04:13 haf 2015-04-25 15:04:30 but I've registered that name sometime on irc and I can't seem to find the password ;) 2015-04-25 15:07:02 ok 2015-04-25 15:07:07 I created that ticket a year ago 2015-04-25 15:07:12 switched then to haskell 2015-04-25 15:07:28 I will look what I can do for mono 2015-04-25 15:07:50 Ok, that would be great! 2015-04-25 15:12:03 i think i tried mono, but it has issues 2015-04-25 15:12:27 @clandmeter: in what sense? 2015-04-25 15:12:30 weoooo, mono still needed ? http://www.phoronix.com/scan.php?page=news_item&px=FreeBSD-CoreCLR-Build 2015-04-25 15:14:21 vkrishn: well, while it builds on FreeBSD, it can't print Hello World to the console yet, so it's still early days. 2015-04-25 15:14:40 ah. ok 2015-04-25 15:14:49 Mono is likely going to split along the same lines as .Net - everything it turning into nugets 2015-04-25 15:15:08 and then you download what you need on top of your CLR - mono or CoreCLR doesn't matter much then 2015-04-25 15:15:29 imagine, like java, but without the funky 'please click that you agree the license arguments' ;) 2015-04-25 15:16:13 weoooo: error: redefinition of 'struct _fpstate' 2015-04-25 15:16:39 :) 2015-04-25 15:17:31 clandmeter: is that a compile error from the stock mono repo? 2015-04-25 15:17:57 You could try the latest stable release https://github.com/mono/mono/tree/mono-3.12.1 2015-04-25 15:18:15 I do not see any problems for "agree the license" 2015-04-25 15:18:20 /aports/testing/mono/src/mono-3.12.1 2015-04-25 15:18:47 shafire: was just kidding :) 2015-04-25 15:18:56 it's annoyed me when I deployed an ELK stack 2015-04-25 15:19:58 but mono has f# :) 2015-04-25 15:20:20 but jvm has clojure, scala, frege, ... 2015-04-25 15:21:20 Yes. To be honest there are many great languages out there. 2015-04-25 15:22:23 I say as I used to say when my devs asked: as long as you can support the operations of your language and it fits with our micro-service architecture and you can communicate sensibly with our services somehow (messages/RPC) go ahead. 2015-04-25 15:23:36 Getting side-tracked. Is there anything I can do to help with mono? Obviously I'm a newbie to this OS, so it won't be fast, but I might be able to answer your questions. 2015-04-25 15:25:07 found some more interesting apps, https://github.com/jpirko/qemu-rocker, http://www.grpc.io/ 2015-04-25 15:27:41 hi 2015-04-25 15:27:47 I'm trying to install libxft-dev on alpine linux 2015-04-25 15:27:58 but it fails http://pastebin.com/raw.php?i=Zt3KUwqu 2015-04-25 15:28:08 does anyone know what I could do? 2015-04-25 15:36:56 dnaaa: have you installed infinality? 2015-04-25 15:37:34 yes 2015-04-25 15:37:44 freetype-infinality 2015-04-25 15:38:34 i think you need to use regular freetype if you want to build something 2015-04-25 15:38:44 that's sad 2015-04-25 15:38:58 to go back to bad font rendering just to build something 2015-04-25 15:39:15 isn't there something like libxft-infinality-dev? 2015-04-25 15:39:27 why you need better font rendering on the console? 2015-04-25 15:39:48 i mean, you do developement on your desktop? 2015-04-25 15:40:00 I use alpine as general desktop 2015-04-25 15:40:16 I browse the web, watch movies, and all that here on this alpine desktop 2015-04-25 15:40:20 I don't do development 2015-04-25 15:40:43 why you need libxft-dev? 2015-04-25 15:41:00 to build st 2015-04-25 15:41:33 use a lxc container, that would be better. 2015-04-25 15:41:37 st.c:29:25: fatal error: X11/Xft/Xft.h: No such file or directory 2015-04-25 15:41:52 lxc container? how do I do that? 2015-04-25 15:42:03 apk add lxc 2015-04-25 15:42:23 it installed bash! I don't want that 2015-04-25 15:42:36 bash is horrible mess 2015-04-25 15:42:44 I want to remain with ash 2015-04-25 15:42:49 lxc-create -n devbox -t alpine 2015-04-25 15:42:51 you will 2015-04-25 15:42:55 ok 2015-04-25 15:43:09 bash is only used for some lxc scripts 2015-04-25 15:43:21 which we couldnt unbashism 2015-04-25 15:43:32 can I delete everything after st is built? 2015-04-25 15:43:38 lxc-create: utils.c: get_template_path: 1582 No such file or directory - bad template: alpine 2015-04-25 15:43:39 lxc-create: lxccontainer.c: do_lxcapi_create: 1344 bad template: alpine 2015-04-25 15:43:44 lxc-create: lxc_create.c: main: 274 Error creating container devbox 2015-04-25 15:43:53 apk add lxc-templates 2015-04-25 15:44:14 devbox is the name of your container 2015-04-25 15:44:21 you can name it whatever you like 2015-04-25 15:44:34 done 2015-04-25 15:44:45 cd /etc/init.d/ 2015-04-25 15:44:56 ln -s lxc lxc.devbox 2015-04-25 15:45:11 done 2015-04-25 15:45:20 that that init script 2015-04-25 15:45:21 will I be able to undo all this stuff? I just want to compile st 2015-04-25 15:45:23 start 2015-04-25 15:45:28 sure 2015-04-25 15:45:39 your container is in /var/lib/lxc 2015-04-25 15:46:06 lxc-console -n devbox 2015-04-25 15:46:26 clandmeter: where is the path you mentioned? the testing folder? 2015-04-25 15:46:30 btw, its all on our wiki 2015-04-25 15:46:42 weoooo: i patched it 2015-04-25 15:46:45 but fails again 2015-04-25 15:46:54 clandmeter: asks for login and pass 2015-04-25 15:47:38 clandmeter: but are you on a branch? 2015-04-25 15:47:57 cd aports/testing && cd mo 2015-04-25 15:48:02 becomes cd monkey/ 2015-04-25 15:48:07 and that's not right 2015-04-25 15:48:28 weoooo: https://github.com/mono/mono/blob/master/libgc/os_dep.c 2015-04-25 15:48:44 that part is not musl friendly 2015-04-25 15:49:03 devbox login: 2015-04-25 15:49:15 dnaaa: http://wiki.alpinelinux.org/wiki/LXC 2015-04-25 15:49:30 I can't find anything about login and password on that page 2015-04-25 15:49:42 root has no password 2015-04-25 15:49:52 ok 2015-04-25 15:49:59 and you also need to bridge your interface 2015-04-25 15:50:02 if you need networking 2015-04-25 15:50:20 that info is on the wiki 2015-04-25 15:50:32 can I copy my st folder from my alpine to devbox? 2015-04-25 15:50:43 then I compile it on devbox? 2015-04-25 15:50:52 and how to install on normal desktop? 2015-04-25 15:50:57 clandmeter: that's out of my depth - I don't do C, normally. If I start an issue in the mono issue track, could you work with them to make it work with musl? 2015-04-25 15:52:08 file-mmap-posix.c: In function 'open_file_map': 2015-04-25 15:52:09 file-mmap-posix.c:274:79: error: 'DEFFILEMODE' undeclared (first use in this function) fd = open (c_path, file_mode_to_unix (mode) | access_mode_to_unix (access), DEFAULT_FILEMODE); 2015-04-25 15:52:22 probably glibc specific 2015-04-25 15:54:27 clandmeter: please, the wiki doesn't cover many stuff 2015-04-25 15:54:35 I just want to build st :( 2015-04-25 15:54:55 i gave you 2 options 2015-04-25 15:55:03 i dont know what more to do 2015-04-25 15:55:20 what 2 options? 2015-04-25 15:55:29 I'm inside devbox now 2015-04-25 15:55:41 dont use infinality and build it on your desktop 2015-04-25 15:55:42 how do I build st? The st source is on the host 2015-04-25 15:55:46 or use lxc container 2015-04-25 15:55:53 ok, I'm using lxc 2015-04-25 15:55:58 I'm logged in devbox 2015-04-25 15:56:07 make sure you have networking 2015-04-25 15:56:59 first do the networking part 2015-04-25 15:56:59 I have the st source on my host, do I still need netowrking to transfer it to devbox? 2015-04-25 15:57:11 I'm afraid of doing too much stuff and not being able to undo it later 2015-04-25 15:57:23 I'll probably regret this 2015-04-25 15:57:25 but if needed I can set the networking, I think 2015-04-25 15:57:27 why not apk add st? 2015-04-25 15:57:38 :) 2015-04-25 15:57:40 barthalion: because I want different config 2015-04-25 15:57:58 mkay, good enough 2015-04-25 15:58:09 dnaaa: i told you, your container is in /var/lib/lxc/rootfs 2015-04-25 15:58:11 err 2015-04-25 15:58:14 clandmeter: Found some patches for musl for mono https://github.com/maximeh/buildroot/tree/master/package/mono - but it doesn't seem to talk about mmap. However, that item seems to have been dealt with before here, towards the end: http://git.alpinelinux.org/cgit/aports/plain/testing/vdr/musl-compat.patch 2015-04-25 15:58:19 dnaaa: i told you, your container is in /var/lib/lxc/containername/rootfs 2015-04-25 15:59:00 clandmeter: sudo cp -r ./st /var/lib/lxc/devbox/ didn't work 2015-04-25 15:59:23 if I ls on devbox it shows nothing 2015-04-25 15:59:45 check the folder first 2015-04-25 15:59:49 its called rootfs 2015-04-25 15:59:58 inside that is your container 2015-04-25 16:00:19 dont just copy it to lxc/devbox 2015-04-25 16:00:31 not even directly in rootfs 2015-04-25 16:00:38 rootfs/root probably 2015-04-25 16:01:09 $ sudo ls /var/lib/lxc/devbox/ 2015-04-25 16:01:10 config rootfs 2015-04-25 16:01:12 it's there 2015-04-25 16:01:26 but I can't cp to it or cd into it, so I don't know what to do 2015-04-25 16:01:57 become root? 2015-04-25 16:02:50 worked, nice 2015-04-25 16:04:58 trying to set networking now 2015-04-25 16:05:31 did the "Prepare network on host" the wiki told 2015-04-25 16:06:06 you should first craete network 2015-04-25 16:06:10 then the container 2015-04-25 16:06:22 alse it wont add network specfiics to its config 2015-04-25 16:10:10 oh, 3.2 builders are up 2015-04-25 16:10:11 cool 2015-04-25 16:12:05 What are your recommended setup for building packages on OS X? Docker only takes me so far... 2015-04-25 16:12:26 use virtual machine 2015-04-25 16:12:30 docker isn't native either 2015-04-25 16:13:46 abuild removes unneded packages after succesful package build 2015-04-25 16:13:59 and can be configured to do it also after failure 2015-04-25 16:14:08 so it's not like you need to recreate vm every time 2015-04-25 16:14:16 too hard for me, I wish alpine was easier. building st was just 'make install' for me in the past :( I'll have to keep the default config (which is really bad) 2015-04-25 16:14:21 but thanks for trying to help 2015-04-25 16:14:27 clandmeter: where is the testing package of mono? Something I can bootstrap from 2015-04-25 16:15:03 weoooo: on my devbox 2015-04-25 16:15:07 its not in aports 2015-04-25 16:15:51 hmmm... What do you recommend I do? Should I wait for you to do something, or should I try some myself first? What is the process? 2015-04-25 16:16:11 the apkbuild is not that complex 2015-04-25 16:16:16 i can sprunge it if you like 2015-04-25 16:16:32 have you worked with abuild already? 2015-04-25 16:17:06 Not anything. 2015-04-25 16:17:32 hmm, now it fails to link 2015-04-25 16:19:20 weoooo: http://sprunge.us/SOQh 2015-04-25 16:22:35 clandmeter: alright, interesting, let me know when you're further 2015-04-25 16:39:44 Does someone know why it says 'Cannot find device "br0"', when stopping networking? 2015-04-25 16:44:13 Is this package broken? http://pkgs.alpinelinux.org/package/main/x86/bridge-utils 2015-04-25 16:44:18 No binaries inside 2015-04-25 16:45:03 you need to run the init script in it 2015-04-25 16:45:21 which will load the bridge module 2015-04-25 16:46:17 I downloaded the package and there is nothing inside 2015-04-25 16:46:25 $ tar xfvz bridge-utils-1.5-r2.apk x .SIGN.RSA.alpine-devel@lists.alpinelinux.org-4a6a0840.rsa.pub x .PKGINFO x .dummy 2015-04-25 16:46:39 we're so lightweight we're shipping empty packages, brother 2015-04-25 16:46:47 enjoy 2015-04-25 16:47:00 yolo 2015-04-25 16:47:05 (y) 2015-04-25 16:51:05 so is this intended or a bug? 2015-04-25 16:53:44 surely a bug 2015-04-25 16:55:06 ok, will create an issue 2015-04-25 16:55:59 #3090 2015-04-25 16:56:05 oups 2015-04-25 16:56:12 #3690 2015-04-25 16:56:15 barthalion: ^ 2015-04-25 16:56:33 algitbot wants to hide this issue :D 2015-04-25 16:57:37 2 packages - bridge & bridge utils - one of them has an init script that needs to be started 2015-04-25 16:59:03 $ tar xfvz bridge-1.5-r3.apk x .SIGN.RSA.alpine-devel@lists.alpinelinux.org-4a6a0840.rsa.pub x .PKGINFO x etc/ x etc/network/ x etc/network/if-pre-up.d/ x etc/network/if-pre-up.d/bridge x etc/network/if-post-down.d/ x etc/network/if-post-down.d/bridge 2015-04-25 16:59:12 BitL0G1c: these will start automatically, or? 2015-04-25 16:59:59 they would start with the network 2015-04-25 17:00:16 I just remember a script was needed 2015-04-25 17:00:33 I use openvswitch 2015-04-25 17:01:15 Is openvswitch a replacement for a bridge? 2015-04-25 17:01:36 doesnt alpien have default bridge support? 2015-04-25 17:02:53 shafire: well, kind of 2015-04-25 17:03:12 ovs allows you to achieve more with less work 2015-04-25 17:03:26 the bridge built into the kernel worked - it is just easer to use openvswitch with lxc containers 2015-04-25 17:03:30 so you can easily separate "clusters" of virtual machines 2015-04-25 17:03:35 can someone show me the network/interfaces for openvswitch? 2015-04-25 17:04:44 BitL0G1c: why is it easier? 2015-04-25 17:04:59 http://blog.scottlowe.org/2014/01/23/automatically-connecting-lxc-to-open-vswitch/ 2015-04-25 17:05:24 you just set in the container config the ovsup & down scripts 2015-04-25 17:07:05 in the container config I use: 2015-04-25 17:07:08 lxc.network.script.up = /etc/lxc/ovsup 2015-04-25 17:07:08 lxc.network.script.down = /etc/lxc/ovsdown 2015-04-25 17:07:26 ok, but it doesnt seem easier then my current brigde setup 2015-04-25 17:08:12 after you sell your soul to openstack there is no return… 2015-04-25 17:08:32 mono build :) 2015-04-25 17:08:42 linking issue solved? 2015-04-25 17:09:03 yes, just remove unit test 2015-04-25 17:09:09 nice 2015-04-25 17:09:23 clandmeter: can you build with dynamic llvm support? 2015-04-25 17:09:27 like to use it too 2015-04-25 17:09:38 im going to commit it as is 2015-04-25 17:09:42 ok 2015-04-25 17:09:48 then im going to take a both 2015-04-25 17:09:50 bath 2015-04-25 17:10:25 boat 2015-04-25 17:12:02 ok lets see later :) 2015-04-25 17:22:27 Hello. 2015-04-25 17:42:12 S_K: hi there 2015-04-25 17:42:23 hi 2015-04-25 17:42:35 I will try now openvswitch 2015-04-25 18:14:28 I have a small problem with abuild. 2015-04-25 18:16:15 define small 2015-04-25 18:17:29 I get: 2015-04-25 18:17:30 skuhne@devlok:/data/packages/daemon abuild build 2015-04-25 18:17:30 /usr/bin/abuild: line 2025: ./configure: not found 2015-04-25 18:17:30 >>> ERROR: daemon: build failed 2015-04-25 18:17:30 skuhne@devlok:/data/packages/daemon 2015-04-25 18:18:01 But, it is in source and when I make unpack I can build it. 2015-04-25 18:18:03 run just abuild 2015-04-25 18:18:20 well, because you're not supposed to specify function manually if you don't know you're doing 2015-04-25 18:18:51 skuhne@devlok:/data/packages/daemon abuild 2015-04-25 18:18:51 >>> daemon: Checking sanity of /data/packages/daemon/APKBUILD... 2015-04-25 18:18:51 >>> daemon: Analyzing dependencies... 2015-04-25 18:18:51 >>> daemon: Cleaning temporary build dirs... 2015-04-25 18:18:52 >>> daemon: Checking sha512sums... 2015-04-25 18:18:55 daemon-0.6.4.tar.gz: OK 2015-04-25 18:18:57 >>> daemon: Unpacking /var/cache/distfiles/daemon-0.6.4.tar.gz... 2015-04-25 18:18:59 /usr/bin/abuild: line 1: ./configure: not found 2015-04-25 18:19:00 please 2015-04-25 18:19:01 >>> ERROR: daemon: all failed 2015-04-25 18:19:03 skuhne@devlok:/data/packages/daemon 2015-04-25 18:19:05 use pastebin 2015-04-25 18:19:08 abuild unpack && abuild prepare && abuild build 2015-04-25 18:19:09 Ok. 2015-04-25 18:20:42 http://pastebin.com/G7tW3Fg7 2015-04-25 18:20:55 Is there a log file or something? 2015-04-25 18:21:11 log file of what? there is no ./configure fail 2015-04-25 18:21:13 file* 2015-04-25 18:21:13 Or can I get abuld more verbose? 2015-04-25 18:21:20 is this APKBUILD from our git repository? 2015-04-25 18:21:54 I'm not sure with the git repo. 2015-04-25 18:22:02 I don't see it there, so probably not 2015-04-25 18:22:20 Oh, no it is my own. 2015-04-25 18:22:51 can you share it then? 2015-04-25 18:23:28 hard to say what's wrong if there is no APKBUILD to test 2015-04-25 18:24:10 mkay, ./configure runs conf/linux 2015-04-25 18:24:13 conf/linux wants perl 2015-04-25 18:24:16 http://pastebin.com/0b4xcY1y 2015-04-25 18:24:48 maybe configure is bash? 2015-04-25 18:24:57 It is installed. When I run "abuild unpack" I can build it, with a small patch. 2015-04-25 18:25:15 configure runs out of box. 2015-04-25 18:25:27 no, it's sh 2015-04-25 18:25:33 S_K: $_builddir is empty 2015-04-25 18:25:37 here you go 2015-04-25 18:25:41 It looks like abuild is in the wrong dir or something. 2015-04-25 18:25:56 well, it is in the directory you told him to be 2015-04-25 18:26:07 Ah, there for I was asking for a log file. 2015-04-25 18:26:56 But, abuild unpack It it self, so it should know it, was my idea. 2015-04-25 18:26:57 abuild logs everyting to console 2015-04-25 18:27:00 no need for a logfile 2015-04-25 18:27:08 So, that is all ... 2015-04-25 18:27:28 we don't do any magic in abuild 2015-04-25 18:27:37 packager can download many tarballs at once in APKBUILD 2015-04-25 18:27:51 you can't be sure in which one he wants to start 2015-04-25 18:28:48 How to set $_builddri? 2015-04-25 18:29:05 line 17 2015-04-25 18:29:13 _builddir="$srcdir/$pkgname-$pkgver" 2015-04-25 18:29:21 Ahhhh ... 2015-04-25 18:30:02 newapkbuild would have done that for you. 2015-04-25 18:30:16 ACTION never used it 2015-04-25 18:30:23 I'm still alive though 2015-04-25 18:30:35 i never not used it :) 2015-04-25 18:30:36 I've used it ... 2015-04-25 18:30:50 but not good enough. 2015-04-25 18:30:57 yes, but did you specify the src? 2015-04-25 18:31:39 barthalion: did you find time for that patch? 2015-04-25 18:33:45 I'll push it tomorrow 2015-04-25 18:33:50 fighting with nasty server 2015-04-25 18:34:37 So, patch works too. 2015-04-25 18:34:50 clandmeter: you know, coming from arch, where we rebuild everything manually, I can at least set $_builddir by hand here :) 2015-04-25 18:47:27 barthalion: push to where? 2015-04-25 18:48:05 aports 2015-04-25 18:48:21 to patchwork itself? 2015-04-25 18:48:28 yeah 2015-04-25 18:48:32 do you want to test it before? 2015-04-25 18:48:44 to testing/patchwork, that's what I mean 2015-04-25 18:48:53 patchwork is in main now 2015-04-25 18:49:01 will you make it configurable? 2015-04-25 18:49:19 uh-uh 2015-04-25 18:49:23 ok, I can try to 2015-04-25 18:50:29 i mean, if somebody else ever wants it installed, it will have your option added :) 2015-04-25 18:51:18 hmm, i dont have grsec kernel, and i cannot access builders dmesg. ill disable mono for now, cause i think its grsec related. 2015-04-25 19:41:26 clandmeter: did you manage to build mono? :) 2015-04-25 20:28:18 clandmeter: I just realised v4.0 has musl support http://www.mono-project.com/docs/about-mono/releases/4.0.0/#musl-libc 2015-04-25 20:45:06 The C++ parts compile for me with your patch file; how stable would you think they are? 2015-04-25 20:55:16 barthalion: Thanks for your help. It builds up. 2015-04-25 21:09:56 weoooo: it build, but i think it fails with grsec enabled 2015-04-25 21:10:06 Is that default? 2015-04-25 21:10:25 is what default? 2015-04-25 21:10:29 Because there are lots of logs when building: 'mono is not a valid ELF executable'. 2015-04-25 21:10:32 grsec. 2015-04-25 21:10:51 alpine default kernel is grsec 2015-04-25 21:11:14 but i use vanilla kernel on my box 2015-04-25 21:15:30 I see 2015-04-25 21:17:22 Mono contains a JIT that modifies code on the fly, is that why grsec is triggered? 2015-04-25 21:18:52 clandmeter: how do I install the libgdiplus-dev package that you created today? 2015-04-25 21:19:08 I know I can do abuild -R, but I want to keep the generated files to inspect them 2015-04-25 21:19:34 yes, its probalby jit 2015-04-25 21:20:02 apk add libgdiplus-dev 2015-04-25 21:20:34 ok, needed to enable testing, found the uri in /etc/apk/repositories 2015-04-25 21:20:47 how does openjdk handle its jit then? 2015-04-25 21:22:34 paxctl 2015-04-25 21:24:45 or paxmark 2015-04-25 21:55:22 Can someone help me with grsec and java? 2015-04-25 21:57:55 http://en.wikibooks.org/wiki/Grsecurity/Application-specific_Settings#Java 2015-04-25 21:58:11 I've built mono now, and it's a binary; how do I test it? abuild package generated a pkg/ folder with the binaries and libs. Now what? 2015-04-25 21:59:19 S_K: 'apk add paxctl' & 'paxctl -czm /path/to/binary' 2015-04-25 21:59:45 i think you need paxmark 2015-04-25 22:00:19 weoooo: there should be a packages folder in your home 2015-04-25 22:00:39 its a setting in abuild.conf 2015-04-25 22:01:03 I can run mono successfully now 2015-04-25 22:01:14 BitL0G1c: Thanks a lot!!! 2015-04-25 22:01:28 install paxmark - it pulls in paxctl 2015-04-25 22:01:47 I've installed paxctl. 2015-04-25 22:01:52 yes, it uses paxctl 2015-04-25 22:01:58 What does paxmark? 2015-04-25 22:02:00 but there is some logic in paxmark 2015-04-25 22:02:07 i didnt readup on it 2015-04-25 22:04:23 S_K: most likely the JIT sets the x-bit on memory it just wrote to. And having executable user-modifiable memory is a recipe for disaster normally. So I suppose mprotect is the flag paxmark uses to keep track of whether a process wishing to have executable and writable memory should be allowed to. 2015-04-25 22:05:51 clandmeter: there's nothing in ~/packages 2015-04-25 22:06:31 there's a 'testing' folder, but no packages 2015-04-25 22:06:42 I need to do 'abuild -R', right? 2015-04-25 22:08:10 Trying... 2015-04-25 22:08:11 abuild -r 2015-04-25 22:08:17 not capital R 2015-04-25 22:09:22 BitL0G1c: thank you 2015-04-25 22:12:51 ;-) np - I think R rebuild everything including the dependencies 2015-04-25 22:16:28 ouch 2015-04-25 22:17:20 -R Recursively build and install missing dependencies (using sudo) 2015-04-25 22:18:29 the nicest feature is it automatically purges the dependencies 2015-04-25 23:09:09 I like openvswitch 2015-04-26 07:04:11 good morning 2015-04-26 10:21:01 virt-manager gives me following error: 'module' object has no attribute 'VIR_DOMAIN_BLOCKED' 2015-04-26 10:21:20 I see in the logs that someone mentioned it last month 2015-04-26 10:21:26 any updates? 2015-04-26 20:07:11 there is at least with stable vanilla 3.1.3 install problem with lm_sensors startup. It will failg as it looking for the MODULE_0 inside /etc/conf.d/lm_sensors . Which in our case is the HWMON_MODULES. Here startup without chaning /etc/conf.d/lm_sensors http://sprunge.us/gIeU , and here after changing http://sprunge.us/ZhcC . This seems like to be a reported bug in gentoo to https://bugs.gentoo.org/show_bug.cgi?id=480018 . Should i submit an bug re 2015-04-26 20:15:34 then however lm_sensors removed this MODULE_0 legacy sysconfig syntax 2 years ago http://lm-sensors.org/changeset/6115/lm-sensors/trunk 2015-04-27 07:04:39 please, merge my patchwork http://patchwork.alpinelinux.org/patch/60/ 2015-04-27 09:22:18 fabled: i think musl-dev should depend on linux-headers 2015-04-27 09:22:35 and bsd ]:> 2015-04-27 09:22:37 ACTION runs 2015-04-27 09:22:43 there are some of the headers shipped by musl that does #include 2015-04-27 09:23:02 ACTION trows a bsd after barthalion  2015-04-27 09:23:40 i think the bsd headers thingy is correct 2015-04-27 09:23:47 linux-headers is not i think 2015-04-27 09:24:23 but agree, if musl includes stuff, it makes sense 2015-04-27 09:52:18 stateless: fortify error: http://sprunge.us/EaZA 2015-04-27 09:52:37 oh 2015-04-27 09:52:47 looks like it is pulling in gnulib 2015-04-27 09:52:48 bah 2015-04-27 10:02:50 ncopa should i report this as alpine bug also? 2015-04-27 10:02:51 (22:07:05) (crow) there is at least with stable vanilla 3.1.3 install problem with lm_sensors startup. It will failg as it looking for the MODULE_0 inside /etc/conf.d/lm_sensors . Which in our case is the HWMON_MODULES. Here startup without chaning /etc/conf.d/lm_sensors http://sprunge.us/gIeU , and here after changing http://sprunge.us/ZhcC . This seems like to be a reported bug in gentoo to http 2015-04-27 10:02:53 s://bugs.gentoo.org/show_bug.cgi?id=480018 . Should i submit an bug report to? 2015-04-27 10:02:55 (22:15:29) (crow) then however lm_sensors removed this MODULE_0 legacy sysconfig syntax 2 years ago http://lm-sensors.org/changeset/6115/lm-sensors/trunk 2015-04-27 10:03:55 crow: yes, i think so 2015-04-27 10:03:58 set targe t3.1.4 2015-04-27 10:04:05 sure 2015-04-27 10:04:07 set target to 3.1.4 2015-04-27 10:06:20 ncopa regarding https://bugs.alpinelinux.org/issues/3107 i used vanilla iso 3.1.3 and have had same issue about Out of memory. So it should not be PAX releted? 2015-04-27 10:11:39 ncopa lm_sesnors bug report https://bugs.alpinelinux.org/issues/4103 2015-04-27 10:11:44 #4103 2015-04-27 10:12:12 should not algitbot not replay with url? 2015-04-27 11:00:54 algitbot: wake up! 2015-04-27 11:01:01 algitbot: hi 2015-04-27 11:01:07 #4103 2015-04-27 11:01:12 hum 2015-04-27 11:01:51 we should talk to the developer of that bot 2015-04-27 11:02:41 did it come with pickup and return? 2015-04-27 11:04:25 yeah we should probably fire the developer of algitbot :) 2015-04-27 11:08:35 ncopa: did you fix the fortify error? 2015-04-27 11:08:58 well the gnulib collision or whatever it is 2015-04-27 11:09:28 stateless: yes. i simply removed gnulib from the equation 2015-04-27 11:09:44 seems like the only reason it was there was to provide snprintf 2015-04-27 12:29:08 fabled: what do you think about adding linux-headers as depend to musl-dev? 2015-04-27 12:29:34 due to things like /usr/include/sys/vt.h 2015-04-27 12:29:50 dunno 2015-04-27 12:29:58 i have no strong feeling either way 2015-04-27 12:30:03 it's a compatibility link 2015-04-27 12:30:15 for some old usage of sys/vt.h that was later moved to linux/ 2015-04-27 12:31:49 there are more of those 2015-04-27 12:31:51 ./usr/include/sys/soundcard.h:#include 2015-04-27 12:31:51 ./usr/include/sys/kd.h:#include 2015-04-27 12:31:51 ./usr/include/sys/vt.h:#include 2015-04-27 12:32:26 i suppose we could either remove those files (and patch the sources who uses them to use linux/*.h) 2015-04-27 12:32:46 or we could add linux-header as depends for musl-dev 2015-04-27 12:33:33 or we could ship those 3 in linux-compat-headers 2015-04-27 12:36:55 ncopa, mmm 2015-04-27 12:37:07 split them to linux-compat-headers inside musl and make that depend on linux-headers? 2015-04-27 12:37:15 yes 2015-04-27 12:37:35 i feel really tempted to just add linux-headers back as makedepends 2015-04-27 12:37:40 depends* 2015-04-27 12:38:31 barthalion: we use libfetch hosted from arch linux. what do you recommend we use as source url for libfetch? 2015-04-27 12:38:43 ftp://ftp.archlinux.org does not work 2015-04-27 12:49:34 new encoder, http://x265.org/ 2015-04-27 12:51:48 kinda nice for handling zram,swap..etc https://github.com/rpodgorny/systemd-swap 2015-04-27 12:55:50 ncopa, https://sources.archlinux.org/other/libfetch/ 2015-04-27 13:00:09 vkrishn: thanks! 2015-04-27 13:14:41 ncopa: it has been moved, let me find the link 2015-04-27 13:14:56 ncopa: https://sources.archlinux.org/ 2015-04-27 13:15:36 it got it. thanks1 2015-04-27 13:16:09 but I think we should mirror it on our own 2015-04-27 13:16:21 we as in Arch don't have it in repositories anymore 2015-04-27 13:16:36 so it's still there only because nobody cares about cleaning this stuff up 2015-04-27 13:16:40 does arch linux use libfetch at all? 2015-04-27 13:17:10 what they use instead of libfetch? 2015-04-27 13:18:07 pacman uses libcurl for a long time already 2015-04-27 13:18:12 ok 2015-04-27 13:18:16 we don't package libfetch anymore 2015-04-27 13:18:21 it has been dropped to AUR 2015-04-27 13:18:22 understand 2015-04-27 13:18:56 who's that played with proxychains? 2015-04-27 13:19:36 we don't have in alpine but there are failed building logs 2015-04-27 13:20:02 looks that pwclient does not work behind a proxy 2015-04-27 13:20:20 proxychains could be used as workaround 2015-04-27 13:42:55 i'm taking off arm builder for the evening (planned maintenance) 2015-04-27 13:43:25 i'll try make sure it gets busy whenever it is back :) 2015-04-27 13:43:31 :) 2015-04-27 13:59:43 clandmeter, you around? 2015-04-27 14:01:06 looks interesting, https://github.com/holmium/dnsforwarder 2015-04-27 14:02:37 How do I get this merged? https://github.com/haf/aports/tree/feature/mono-v4.0 2015-04-27 14:30:59 weeooo: i think mono 4 got merged already 2015-04-27 14:33:51 ncopa: Ah, yes, it has! Awesome 2015-04-27 14:34:10 i dont think it builds on x86 (32bit) though 2015-04-27 14:34:24 That's no problem for me though, I need x64 2015-04-27 14:38:19 jwhois still active here, https://github.com/jonasob/jwhois 2015-04-27 14:38:44 last release is 2005 2015-04-27 14:38:52 3.2.3 2015-04-27 14:44:03 yes, but master branch seems active 2015-04-27 14:45:02 with inspiring commit messages 2015-04-27 14:47:01 commit 27c4a0283548a0c3f2a0067e2270c9196a7f4ca7 v4.0 2015-04-27 14:49:43 ok, so it seems it has problems building... is there a better log available? 2015-04-27 14:52:15 weeooo: http://bld1.alpinelinux.org/buildlogs/build-edge-x86/testing/mono/ 2015-04-27 15:24:49 ncopa: I'm gonna see if I can get the mono people involved 2015-04-27 15:25:01 but it seems to be an architecture-specific musl-specific error 2015-04-27 15:25:13 With './.libs/libmonosgen-2.0.so: undefined reference to `__sync_fetch_and_add_8'' 2015-04-27 15:25:54 weeooo: i thought it was gcc related? 2015-04-27 15:26:20 Because of https://groups.google.com/forum/#!topic/codership-team/AzKmEJz7vTQ ? 2015-04-27 15:26:26 from the first results i got from google 2015-04-27 15:27:24 yeah, same here 2015-04-27 15:27:28 weeooo: i didnt readup that much on the error 2015-04-27 15:27:33 so i could be wrong 2015-04-27 15:27:33 true 2015-04-27 15:27:37 me neither 2015-04-27 15:27:43 but you are right 2015-04-27 15:27:48 mono ppl to the resque 2015-04-27 15:27:56 but it seems to be gcc-linux specific as you say -- but I think it can still be related to the architecture 2015-04-27 15:28:09 yes it is 2015-04-27 15:28:33 will you report it upstream? 2015-04-27 15:28:48 i will disable x86 for now 2015-04-27 15:29:25 they made it musl compat, so i have good hopes they will fix this too. 2015-04-27 15:36:14 Yes, I created https://bugzilla.xamarin.com/post_bug.cgi 2015-04-27 15:36:29 *https://bugzilla.xamarin.com/show_bug.cgi?id=29427 2015-04-27 15:41:17 probably better to attache the build log 2015-04-27 15:41:33 not sure how long the logs stay alive on builder 2015-04-27 15:48:05 said and done 2015-04-27 15:49:15 thx! good work 2015-04-27 15:54:57 fcolista: patchwork has https support 2015-04-27 19:54:47 https://github.com/scadu/aports/commits/master bump pkgver for phpmyadmin, thunar, mopidy and syncthing. 2015-04-28 01:33:44 barthalion, http://git.alpinelinux.org/cgit/alacerda/setup-box/tree/json 2015-04-28 01:34:22 preparing setup-box to be as flexible as possible. 2015-04-28 01:35:14 the user can create his own .json file to customize setup-box 2015-04-28 07:22:54 ncopa, re: redmine and issue tracker, have you ever take a look at roundup? Is the one used by python community. 2015-04-28 07:22:55 http://roundup.sourceforge.net/ 2015-04-28 07:23:39 just thiniking louder, since i don't know specifically all the business requirements for us 2015-04-28 07:26:25 can you close issues from git commits? 2015-04-28 07:27:05 can you have issue relations (duplicate issue #, depends on issue # ...) 2015-04-28 07:27:34 can you have a single issue that needs to be fixed in multiple git branches 2015-04-28 07:51:24 coredumb: do you have an idea about when the consul package will be pushed to edge/testing? Is there still a lot of work to do before that point. May I be of some help? 2015-04-28 08:51:25 Jean-Scotch: it requires first to make offline packages of deps 2015-04-28 09:45:38 coredumb: im looking at your apkbuild for consul 2015-04-28 09:46:45 it's fine, it just should use some tarball instead of using go to fetch deps 2015-04-28 09:46:51 because two libraries timeout 2015-04-28 09:46:58 that's why I didn't want to push it as is 2015-04-28 09:47:19 yep 2015-04-28 09:47:57 fetcing source is fine 2015-04-28 09:51:07 timeouts aren't fine 2015-04-28 09:51:23 they dont timeout for me 2015-04-28 09:51:46 or maybe im doing it wrong :) 2015-04-28 09:52:09 well, they did both on my laptop and lxc container on your srv 2015-04-28 09:52:09 yeah they don't for me as well 2015-04-28 09:52:19 but if it's just me, go ahead and push it 2015-04-28 09:52:31 im getting a build error 2015-04-28 09:52:52 hm, it built fine back then, except downloading 2015-04-28 09:55:02 what a mess 2015-04-28 09:55:12 clandmeter: apk error or just 2 libs that can't be built ? 2015-04-28 09:55:27 seems godeps can't get them in one pass -_- 2015-04-28 09:55:39 this could have been output to /dev/null 2015-04-28 09:55:59 hence the need to do the go get after the godep part 2015-04-28 09:56:04 i can see now why you fetches that json file 2015-04-28 09:56:14 fetched 2015-04-28 09:56:25 clandmeter: yes you need the godep file else it won't build 2015-04-28 09:56:48 fetching all deps in a tarball would require one to set fetched repos to correct commit 2015-04-28 09:56:54 kind of pita :( 2015-04-28 09:57:46 i can't understand why those go deps can't built in one pass -_- 2015-04-28 09:58:05 does it happen everywhere, btw? 2015-04-28 09:58:19 happens as well on a go get xxxxx/consul btw 2015-04-28 09:58:26 sounds like yes 2015-04-28 09:59:08 yeah sounds like "we're strangely hitting to much deps and dunno what to do" 2015-04-28 11:10:43 coredumb: i think i got a working apkbuild. but its kinda nasty 2015-04-28 11:17:31 coredumb: i have removed the two faulty repo's from godep file. according to https://groups.google.com/forum/#!topic/consul-tool/HV_3-BcDUl8 2015-04-28 11:32:18 i read but i still dont get it, what for is the consul? any real scenario? 2015-04-28 12:03:21 ncdev-edge-x86_64:~/aports/main$ grep sbin/runscript */*.initd | wc -l 2015-04-28 12:03:21 259 2015-04-28 12:03:30 we need rename runscript to openrc-run 2015-04-28 12:04:20 openrc upgrade pushed 2015-04-28 12:04:26 watch out for breakages... 2015-04-28 12:23:01 ncopa, did we have an issue with ca-certificates? 2015-04-28 12:23:16 i'm asking because of that: 2015-04-28 12:23:35 buildlab-edge:~# update-ca-certificates 2015-04-28 12:23:35 WARNING: ca-certificates.crt does not contain exactly one certificate or CRL: skipping 2015-04-28 12:24:08 it's edge 2015-04-28 12:25:31 dunno 2015-04-28 12:33:17 looks that we don't have StarSSL CA in ca-certificates.crt 2015-04-28 12:33:17 http://sprunge.us/NJVG 2015-04-28 12:33:29 (13:29:52) algitbot: aports:master | Carlo Landmeter | testing/consul: new aport | http://dup.pw/c05fd028 Great! I will go one step further with it ;) and will of course update my w.a.o page about it 2015-04-28 12:34:44 *startSSL 2015-04-28 12:37:32 Jean-Scotch: ok please. i dont know if it actually works. 2015-04-28 12:37:43 patches are welcome 2015-04-28 12:38:21 I already use the previous apk prepared by coredumb in my lab 2015-04-28 12:38:44 but an apk in repo was needed to integrate into my PXE setup ;) 2015-04-28 13:03:01 clandmeter, i think #64 is missing bump pkgrel 2015-04-28 13:03:14 http://patchwork.alpinelinux.org/patch/64/ 2015-04-28 13:03:16 this one 2015-04-28 13:03:31 yup 2015-04-28 13:03:44 i've applied the patch and now pgsql is building 2015-04-28 13:03:45 you can do it if you want. 2015-04-28 13:03:48 yeah 2015-04-28 13:03:52 at the end i'll do it 2015-04-28 13:09:26 ACTION slaps fcolista with pwclient 2015-04-28 13:12:17 or git am at least… 2015-04-28 13:23:23 hum 2015-04-28 13:23:54 i will push abuild to force init.d script to have sbin/openrc-run instead of runscript 2015-04-28 13:24:52 fcolista: i dont think i want do that 2015-04-28 13:25:20 i dont what ship headers and -dev stuff with libpq 2015-04-28 13:25:46 ncopa, yes. talked with clandmeter about that 2015-04-28 13:25:51 that's why commit was reverte 2015-04-28 13:25:55 *reverted 2015-04-28 13:25:59 thanks clandmeter :) 2015-04-28 13:26:02 it was just an experiment with pwclient 2015-04-28 13:26:15 and thats why i've been slapped :) 2015-04-28 13:26:32 clandmeter: thanks for correctly slapping fcolista :) 2015-04-28 13:26:37 :) 2015-04-28 13:26:44 well, maybe it was also a bit my mistake 2015-04-28 13:26:58 :) 2015-04-28 13:27:00 thanks for slapping yourself then :) 2015-04-28 13:27:56 we all have to learn sometimes :) 2015-04-28 13:28:19 i think we need to setup a policy how to submit patches 2015-04-28 13:28:31 yes 2015-04-28 13:28:32 that libpq is a good example of a bad patch 2015-04-28 13:28:47 even if those hearders are ok, i would like to know why. 2015-04-28 13:28:54 there is a ticket for libpq 2015-04-28 13:28:59 someone asked for libpq-dev 2015-04-28 13:29:31 not happy with pulling in full postgresql-dev when only libpq headers and lib are needed 2015-04-28 13:29:36 its a policy thing 2015-04-28 13:30:00 #4109 2015-04-28 13:30:06 algitbot: hi 2015-04-28 13:30:11 i think we also need to be carefull of github sources 2015-04-28 13:30:16 yes 2015-04-28 13:30:17 algitbot: lovely as always 2015-04-28 13:30:21 lots of apkbujilds have v$pkgver 2015-04-28 13:30:27 http://bugs.alpinelinux.org/issues/4109 2015-04-28 13:30:40 v$pkgver is not bad if tarball is renamed 2015-04-28 13:30:42 (if that 2015-04-28 13:30:45 github sources should use :: 2015-04-28 13:30:46 did those enter aports? 2015-04-28 13:30:48 yes 2015-04-28 13:30:51 and yes 2015-04-28 13:31:04 i think we should have abuild to fail 2015-04-28 13:31:09 yep 2015-04-28 13:31:27 it could result i really starnge issues :) 2015-04-28 13:40:30 clandmeter, barthalion, any suggestion for good error message when you dont rename v$pkgver.tar.gz source tarball? 2015-04-28 13:41:17 unindentifiable tarball name or something like that? 2015-04-28 13:41:53 ncopa: ↑ 2015-04-28 13:42:19 im'm thinkgin: "source $name needs to be renamed" 2015-04-28 13:42:46 well, it doesn't say why 2015-04-28 13:43:51 possible source collision detected 2015-04-28 13:44:15 "... to avoid collision with other packages with same version" 2015-04-28 13:44:49 3 second to self destruct 1..2.. 2015-04-28 13:44:53 :D 2015-04-28 13:45:12 ha 2015-04-28 13:45:38 we already have coffie apk, why not bomber abuild :) 2015-04-28 13:45:45 "unindentifiable source name: $source needs to be renamed to avoid possible collision" 2015-04-28 13:45:58 we could have it run the apk del --russian-roulette easter egg 2015-04-28 13:46:10 :) 2015-04-28 13:46:35 and we let fcolista try it first 2015-04-28 13:46:44 :D 2015-04-28 13:47:21 case ${i##*/} in 2015-04-28 13:47:22 v$pkgrel.tar.*|$pkgrel.tar.*) 2015-04-28 13:47:22 die "source ${i##*/} needs to be renamed 2015-04-28 13:47:22 to avoid possible collisions" 2015-04-28 13:47:22 ;; 2015-04-28 13:47:23 esac 2015-04-28 13:48:55 http://sprunge.us/bHfi 2015-04-28 13:50:17 ncopa: pkgrel? 2015-04-28 13:50:24 no 2015-04-28 13:50:41 tag is normally v$pkgver 2015-04-28 13:50:48 pkgrel is alpine specific 2015-04-28 13:51:05 bah 2015-04-28 13:51:07 thanks 2015-04-28 13:51:08 now i see 2015-04-28 13:51:14 :) 2015-04-28 13:51:29 ACTION slaps ncopa with pkgrel 2015-04-28 13:52:38 its a good thing to think out load. 2015-04-28 13:53:22 http://tinyurl.com/mg7xxs5 2015-04-28 13:54:06 yup 2015-04-28 13:54:12 good thing with review too 2015-04-28 13:54:18 it's a good thing to be the one slapping, not slapped 2015-04-28 13:54:27 :) 2015-04-28 14:17:40 there were no other pending abuild changes needed? 2015-04-28 14:21:45 clandmeter: why we have samba36? 2015-04-28 14:24:52 hmm 2015-04-28 14:24:57 let me think 2015-04-28 14:25:13 i think because of the size 2015-04-28 14:25:29 it has openwrt patches 2015-04-28 14:25:46 regular samba is huge 2015-04-28 14:26:31 but i cannot recall the use case here. 2015-04-28 14:26:46 could be it was rpi related (or alix) 2015-04-28 14:26:55 yeah, ithink it was alix 2015-04-28 14:30:13 you wanted full samba file server? 2015-04-28 14:30:19 or only needed samba client 2015-04-28 14:30:30 samba server 2015-04-28 14:30:38 i know cifs-utils 2015-04-28 14:35:34 heads up 2015-04-28 14:35:38 huge push 2015-04-28 14:36:00 build/build-edge-x86 1/205 1796/2001 main/smartmontools 6.3-r1 2015-04-28 14:36:21 whoops 2015-04-28 14:57:12 https://github.com/scadu/aports bump pkgver for phpmyadmin, thunar, mopidy and syncthing. I'd be grateful if you could pull. 2015-04-28 15:15:13 cant we just add them if build fails? :) 2015-04-28 15:30:36 ncopa in your list pust are the dup.pw link and after page opens i see this: Bad object id: 25a56f81 2015-04-28 15:31:25 at least these: (16:12:10) (algitbot) abuild:master | Natanael Copa | abuild: require that .initd files are sbin/openrc-run | http://dup.pw/553762ca 2015-04-28 15:33:27 .pw only works for aports commits 2015-04-28 16:27:44 Hello BitL0G1c, I've installed nm (thanks) and started dbus and nm, but nm isn't showing any interfaces in nmtui; any ideas? 2015-04-28 17:09:10 trebor - you have to add them 2015-04-28 17:15:12 BitL0GIc - Shouldn't nm automatically connect eth0 via dhcp (when available) if it's not explicitly configured in /etc/network/interfaces? 2015-04-28 17:18:37 BitL0GIc - Even if I add it via nmtui-edit, it doesn't show up nmtui-connect? 2015-04-28 17:48:02 BitL0G1c - "nmcli d" is also showing no devices. 2015-04-28 17:59:32 I don't use network manager - have you configured dhcpd ? (it will be looking for that) 2015-04-28 18:00:13 trebor: did you commit those patches? 2015-04-28 18:00:27 err contrib the patches 2015-04-28 18:04:50 BitL0G1c - dhcp is working, does "nmcli d" show any devices for you? 2015-04-28 18:05:44 hold on 2015-04-28 18:06:38 clandmeter - No, that's beyond my skill level. 2015-04-28 18:09:33 nmcli d shows eth0 & lo as unmanaged 2015-04-28 18:09:43 when run as root 2015-04-28 18:10:11 the post-install script told you to add yourself to the plugdev group to modify interfaces as a user 2015-04-28 18:10:26 i think that is what you are missing 2015-04-28 18:11:02 'addgroup -S plugdev' 2015-04-28 18:11:12 "nmcli d" shows no devices for me - I'm running as root. 2015-04-28 18:11:28 'addgroup username plugdev' 2015-04-28 18:11:50 run 'setup-interfaces' 2015-04-28 18:12:15 & possibly 'rc-service networking start' 2015-04-28 18:12:32 & 'rc-update add networking default' 2015-04-28 18:15:54 I've already got networking in the rc-scripts; it also starts if explicitly configured in "/etc/network/interfaces". But still, "nmcli d" shows no devices. 2015-04-28 18:26:00 fabled: around? 2015-04-28 18:26:22 clandmeter, not really 2015-04-28 18:26:27 :) 2015-04-28 18:26:39 im missing ipt_ULOG.h 2015-04-28 18:27:37 its a netfilter kernel header from what i can find 2015-04-28 18:27:59 yes. i think it's deprecated 2015-04-28 18:28:38 yes. ipt_ULOG is obsolete 2015-04-28 18:28:42 replaced by ipt_LOG 2015-04-28 18:28:59 the header is probably not compatible. so it likely needs code changes too 2015-04-28 18:29:58 oh, ok then ill disable pmacct 2015-04-28 18:31:57 thx 2015-04-28 19:15:41 clandmeter you have time for this: >>> ERROR: oscam: oscam.initd is not an openrc #!/sbin/runscript 2015-04-28 19:17:02 crow: are you mixing repo's? 2015-04-28 19:17:45 oh thats on your build env 2015-04-28 19:18:04 apk update && apk upgrade 2015-04-28 19:19:19 also known as apk upgrade -U for lazy 2015-04-28 19:20:46 well yea i mixed something, i am on vanilla 3.1.3 and i pulled packages. 2015-04-28 19:21:01 git clone git://dev.alpinelinux.org/aports 2015-04-28 19:22:15 which repos do you use? 2015-04-28 19:23:35 clandmeter well i am at 3.1.3 vanilla stable http://liskamm.alpinelinux.uk/v3.1/main 2015-04-28 19:24:51 you are building edge packages on stable repo's 2015-04-28 19:24:54 thats not supported 2015-04-28 19:26:12 how can i pull "stable" packages? 2015-04-28 19:26:34 what does that mean? 2015-04-28 19:27:33 if you want to build edge pacakges, then your build system should be edge (edge repos) 2015-04-28 19:27:37 well i would like to build stable packages on stable repo 2015-04-28 19:27:55 checkout the stable branch in aports 2015-04-28 19:28:01 but that wont have the latest oscam 2015-04-28 19:28:14 of course you can modify that yourself 2015-04-28 19:28:31 git brancha -a shows you our branches 2015-04-28 19:28:40 cheap keyboard 2015-04-28 19:28:50 (always blame the hardware) 2015-04-28 19:29:30 git branch -a 2015-04-28 19:29:42 git checkout branch to switch 2015-04-28 19:30:04 where barnch is the name of the branch 2015-04-28 19:31:29 clandmeter np about version i can change that myself 2015-04-28 19:31:51 you can also cherry-pick what you need 2015-04-28 19:32:09 git checkout -b stable-crow origin/version 2015-04-28 19:32:27 note sha1 sums from cgit or 'git log master' 2015-04-28 19:32:35 and git cherry-pick sha1 2015-04-28 19:33:10 clandmeter: I assume that there's no point in posting url with my aports clone and asking for merge, right? I mean we have patchwork now and so on. 2015-04-28 19:33:32 And my request was ignored. 2015-04-28 19:33:43 scadu: it was not ignmored by me 2015-04-28 19:33:55 im trying to fix the mess ncopa made 2015-04-28 19:34:06 clandmeter: Oh I see. 2015-04-28 19:34:33 It's kinda surprising. 2015-04-28 19:34:34 it's interesting how clandmeter always ends up fixing mess we left… 2015-04-28 19:34:44 ncopa and mess. 2015-04-28 19:35:42 scadu: i think we first want to fix current state of aports before we start commining 2015-04-28 19:35:45 commiting 2015-04-28 19:36:02 thats why patchwork is nice, so we wont forget. 2015-04-28 19:37:40 i do wonder if there is not easier way of tracking this linux header crap 2015-04-28 19:39:57 scadu: ill try to commit them now, cause this other things is going to take a while anyway. not sure when its in repo though 2015-04-28 19:40:09 I used to push changes to my github repo and then asking for merge on #alpine-devel. I'm not sure if sending patches to alpine-aports ml would be better. 2015-04-28 19:40:57 Probably ml because of patchwork. 2015-04-28 19:44:52 yes, its probably better now. 2015-04-28 19:45:05 but commiters still need to get used to it though 2015-04-28 19:49:17 scadu: i cannot merge cause of ncopa openrc commit 2015-04-28 19:49:25 syncthing 2015-04-28 19:56:13 I've just done git pull seing mess in my aports. 2015-04-28 19:56:22 ncopa was faster ;f 2015-04-28 19:57:43 scadu: maybe try to git send email them? 2015-04-28 19:58:00 not sure how much commits it were? 4? 2015-04-28 19:59:08 yes. 2015-04-28 19:59:20 syncthing, mopidy, thunar and phpmyadmin 2015-04-28 20:00:22 are you using mopidy? 2015-04-28 20:00:41 from time to time. 2015-04-28 20:00:54 BitL0G1c - Any other ideas to nm? I've noticed that my installation has no plugdev group; maybe I'm missing something? I am root though. 2015-04-28 20:02:44 are you sure you have networking started ? my nm sees my interfaces 2015-04-28 20:03:47 I'm connected over ssh. 2015-04-28 20:05:29 networkmanager is started ? 2015-04-28 20:07:55 BitL0G1c - Yes nm is started. 2015-04-28 20:08:11 ncopa \o/ 2015-04-28 20:08:31 my root is also a manager of plugdev 2015-04-28 20:09:11 I don't have a plugdev group. 2015-04-28 20:09:20 (07:10:51 PM) BitL0G1c: 'addgroup -S plugdev' 2015-04-28 20:09:20 (07:11:01 PM) trebor: "nmcli d" shows no devices for me - I'm running as root. 2015-04-28 20:09:20 (07:11:17 PM) BitL0G1c: 'addgroup username plugdev' 2015-04-28 20:14:07 BitL0G1c - I've added the group and user; no change. 2015-04-28 20:17:45 these are the services i have running http://hastebin.com/ohamoqocuh.1c 2015-04-28 20:19:05 trebor - you have dbus started ? & console-kit installed ? 2015-04-28 20:32:52 BitL0G1c - Dbus is started and consolekit installed. 2015-04-28 20:35:35 BitL0G1c - Why is nm not shown in your services? 2015-04-28 20:37:32 i started it manually 2015-04-28 20:38:01 root@kvm64 [~]# rc-service networkmanager status 2015-04-28 20:38:01 * status: started 2015-04-28 20:39:04 ah, ok. 2015-04-28 20:42:40 you did log off & on again after adding yourself to plugdev ? 2015-04-28 20:42:54 if 'groups' does not show plugdev 2015-04-28 20:43:02 you need to log off & on 2015-04-28 20:47:55 BitL0G1c - Yes, even restarted. I've also started nm manually with -d -n, but I can't detect anything. 2015-04-28 20:50:01 "nmcli g" STATUS shows disconnected. 2015-04-28 20:52:43 That was STATE, not STATUS. 2015-04-28 20:55:46 trebor - http://hastebin.com/palalohile.hs 2015-04-28 20:58:27 ok, "nmcli d" only shows the header "DEVICE TYPE STATE CONNECTION" and nothing else. 2015-04-28 21:10:18 BitL0G1c - Did you modify anything in "/etc/NetworkManager/"? 2015-04-28 21:25:54 I don't have anything there 2015-04-28 21:26:35 oops yes i do 2015-04-28 21:26:47 but i've not touched anything 2015-04-28 21:27:30 http://hastebin.com/ubarayesok.avrasm 2015-04-28 21:28:56 /etc/NetworkManager/NetworkManager.conf - I've commented out dhcp 2015-04-28 21:28:56 2015-04-28 21:30:12 with it set - i still show devices 2015-04-28 21:34:41 here are the modules i have loaded http://hastebin.com/uyiniyamih.hs - i normally need to 'modprobe ipv6' on a new install - I think you need to check /var/log/messages & 'dmesg' 2015-04-28 21:38:15 ok, I've got an error message in /var/log/messages: alpine daemon.info NetworkManager[1357]: loading system config file (/etc/NetworkManager/NetworkManager.conf) caused error: Key file does not h 2015-04-28 21:38:15 ave group 'ifupdown' 2015-04-28 21:40:21 this also seems wrong: alpine daemon.info NetworkManager[1357]: (-542897248) ... get_connections (managed=false): return empty list. 2015-04-28 23:24:38 As a C#/Windows developer, where is the best place to start getting into OpenSource software development, in particular with Alpine? 2015-04-28 23:42:52 have a look at http://wiki.alpinelinux.org/wiki/Developer_Documentation - an Alpine LXC container makes a very good build environment 2015-04-28 23:43:24 Thank you BitL0G1c 2015-04-28 23:43:35 is your machine Windows ? 2015-04-28 23:43:42 Win 8 at the moment 2015-04-28 23:43:58 I want to get into / contribute to opensource, I understand windows isn't the best platform for that. 2015-04-28 23:44:01 I am welcome to change :) 2015-04-28 23:44:25 on my gaming machine I run Debian inside vmware 2015-04-28 23:45:14 Alpine LXC containers in Debian for the build environment 2015-04-28 23:45:37 alpine KVM virtual machines for testing 2015-04-28 23:46:19 the advantage of using vmware over virtualbox is if your machine has virtualization extensions you can run vm's inside vm's 2015-04-28 23:47:15 I am familar with both and prefer VMware for sure. 2015-04-28 23:47:29 I'm getting "into" alpine as I have a friend messing with Docker, has peeked my interest. 2015-04-28 23:48:19 alpine lxc is very good too - a running system is 5 -10 meg 2015-04-28 23:48:47 holy 2015-04-28 23:49:52 I have a KVM Alpine that runs 4 alpine lxc guests - uses 52 meg ram 2015-04-28 23:50:45 Why so many? 2015-04-28 23:51:02 32 / 64 bit + musl / uclibc 2015-04-28 23:51:06 good call 2015-04-28 23:51:29 musl c is a learning curve 2015-04-28 23:53:36 it doesn't use the hacks glibc has picked up http://www.musl-libc.org/faq.html 2015-04-28 23:55:05 going from C# to C is going to be brutal 2015-04-28 23:57:08 "accelerated C++" is a good book to start with 2015-04-28 23:59:25 https://ebooks-it.org/020170353x-ebook.htm - it has exercises at the end of each chapter & is easy to pickup 2015-04-28 23:59:54 it's also possible to find most of the stack exchange recommended list http://stackoverflow.com/questions/388242/the-definitive-c-book-guide-and-list 2015-04-29 00:01:48 perfect 2015-04-29 06:03:41 Accelerated C++ is outdated 2015-04-29 06:03:49 it's from 2000 iirc 2015-04-29 06:03:56 we have C++14 2015-04-29 06:31:21 morning 2015-04-29 06:31:26 sorry for breaking things 2015-04-29 06:39:53 morning 2015-04-29 06:40:21 ncopa, i'd like to move proxychains-ng on main. Would be fine for you? I'm using it and is working. 2015-04-29 06:40:48 uh-oh 2015-04-29 06:40:54 sprunge is over quota :) 2015-04-29 06:43:38 if make install fails it will not error but build a broken package 2015-04-29 06:43:44 you need || return 1 2015-04-29 06:44:06 ok 2015-04-29 06:44:12 both in install and install-config 2015-04-29 06:44:14 ? 2015-04-29 06:44:17 - make DESTDIR="$pkgdir" install 2015-04-29 06:44:17 - make DESTDIR="$pkgdir" install-config 2015-04-29 06:44:17 + make DESTDIR="$pkgdir" install install-config || return 1 2015-04-29 06:44:18 i suppose yes 2015-04-29 06:44:30 oh nice 2015-04-29 06:45:08 do we have a best practice to follow re APKBUILD? 2015-04-29 06:45:21 not anything written 2015-04-29 06:45:23 but we need that 2015-04-29 06:45:33 i havent felt that motivated to do that 2015-04-29 06:45:39 because i'd like replace APKBUILD format 2015-04-29 06:45:47 start from scratch 2015-04-29 06:46:42 hum linux has an uapi now 2015-04-29 06:46:44 what kinf of format you want to use? APKBUILD is very similar to PKGBUILD and i find it more clear than what other distro does 2015-04-29 06:46:52 like ebuild e.g. 2015-04-29 06:46:57 yes 2015-04-29 06:47:14 one of the goals with apkbuild was to make it very transparent what is going on 2015-04-29 06:47:23 to avoid a full EAPI ala gentoo 2015-04-29 06:47:40 however, the price for that is more maintenance work 2015-04-29 06:48:10 lol, sprunge over quota 2015-04-29 06:48:15 i saw 2015-04-29 06:48:17 not first time 2015-04-29 06:48:23 who was that guy saying we don't need own pastebin? :P 2015-04-29 06:48:30 ACTION hides 2015-04-29 06:48:36 what it me? 2015-04-29 06:48:41 i think we might need one 2015-04-29 06:48:47 nope, it was me 2015-04-29 06:48:49 :) 2015-04-29 06:48:50 i'm looking for alternatives 2015-04-29 06:48:59 webpaste 2015-04-29 06:49:08 wgetpaste sorry 2015-04-29 06:49:20 i wonder if we should build or own sprunge like service 2015-04-29 06:49:31 well, why don't we set up just sprunge? it's open source 2015-04-29 06:49:38 aha 2015-04-29 06:49:39 +1 2015-04-29 06:49:43 where i find the sources? 2015-04-29 06:49:53 https://github.com/rupa/sprunge 2015-04-29 06:50:15 ix.io 2015-04-29 06:50:15 although it wants to use some google thing 2015-04-29 06:50:21 so ix.io may be better idea 2015-04-29 06:50:53 i am kinda ok with using google storage 2015-04-29 06:51:37 ix needs python 2015-04-29 06:51:49 so does sprunge 2015-04-29 06:52:21 yup. Not big change then 2015-04-29 06:55:08 http://ix.io/ibn 2015-04-29 06:56:15 http://ix.io/client 2015-04-29 06:56:22 is unversioned 2015-04-29 06:56:40 yes 2015-04-29 06:56:41 20121205 2015-04-29 06:56:45 this is the last one actually 2015-04-29 06:56:51 gradm, paxctld have new releases 2015-04-29 06:56:59 i putted 0.4 'cause arch does 2015-04-29 06:57:08 but we need to check the version somehow 2015-04-29 06:57:24 http://ix.io/ 2015-04-29 06:57:35 on "CHANGES" this is the last version 2015-04-29 06:57:41 20121205 2015-04-29 06:57:59 any idea how to clone gradm, http://cvsweb.grsecurity.net/?p=gradm.git;a=summary ? 2015-04-29 07:02:10 even better, like except text from mosquitto_pub 2015-04-29 07:05:15 hackish way to get ix.io version: 2015-04-29 07:05:15 wget ix.io -q -O - | grep -A2 CHANGES | tail -1 2015-04-29 07:17:29 fcolista: what do you think about this: 2015-04-29 07:17:32 ncdev-edge-x86_64:~/aports/main/sprunge$ git diff . | ix 2015-04-29 07:17:32 http://ix.io/ibr 2015-04-29 07:18:09 :) 2015-04-29 07:18:10 love it 2015-04-29 07:18:38 i think we should rename sprunge is some different name then 2015-04-29 07:18:47 *in 2015-04-29 07:19:03 i was not aware that also dpaste would work with sprunge package 2015-04-29 07:19:14 i had forgotten it too 2015-04-29 07:19:17 and its currently broke 2015-04-29 07:19:21 ncopaste 2015-04-29 07:19:29 or alpaste 2015-04-29 07:19:40 albin 2015-04-29 07:19:50 albin was cool :) 2015-04-29 07:19:50 no albin no 2015-04-29 07:19:55 :) 2015-04-29 07:20:01 lol 2015-04-29 07:20:03 why not? 2015-04-29 07:20:15 it's normal name here in Poland 2015-04-29 07:20:23 it's re pastebin 2015-04-29 07:20:25 though not very common nowadays 2015-04-29 07:21:02 https://www.namecheap.com/domains/registration/results.aspx?domain=albin.pw 2015-04-29 07:21:03 alpaste-ix alpaste-dpaste alpaste-sprunge 2015-04-29 07:21:44 somehow alpaste makes me think about spaghetti 2015-04-29 07:21:55 lol :) 2015-04-29 07:22:42 whats wrong with sprunge and google? 2015-04-29 07:23:26 nothing, other than sprunge exceeded free quota 2015-04-29 07:23:40 even sprunge.a.o? 2015-04-29 07:23:43 :) 2015-04-29 07:23:52 are you on a pastespree? 2015-04-29 07:24:20 seems it works now 2015-04-29 07:24:34 ncdev-edge-x86_64:~/aports/main/sprunge$ git diff . | dpaste 2015-04-29 07:24:34 https://dpaste.de/VGYk/raw 2015-04-29 07:25:03 django for pastebin seems like overkill 2015-04-29 07:25:39 ncopa: can you take a look at hiredis? 2015-04-29 07:26:04 ncopa: it seems they compile it with c99 2015-04-29 07:27:15 i guess we dont want to set compiler flags to c99 for all packages needing redis support? 2015-04-29 07:27:44 ncopa: how much renewal of pw domain costs? 2015-04-29 07:28:14 ncopa: https://github.com/redis/hiredis/issues/321 2015-04-29 07:30:40 fatal error: linux/netfilter_ipv4/ipt_ULOG.h: No such file or directory 2015-04-29 07:30:48 where did ipt_ULOG.h go? 2015-04-29 07:32:51 ncopa: depricated 2015-04-29 07:33:05 https://lwn.net/Articles/507794/ 2015-04-29 07:33:54 i already moved something to unmaintained because of that. 2015-04-29 07:36:03 i was wondering if we could add a temporary compat header in a seperate package? 2015-04-29 07:45:47 new sprunge is not yet available.. 2015-04-29 08:00:36 hmm... paxctl is still at same version, I read paxctld version 2015-04-29 08:19:17 seems nifty, https://github.com/jech/polipo 2015-04-29 08:19:44 vkrishn, paxctld will never be allowed to alpine 2015-04-29 08:19:50 we want to fix paxflags on package time 2015-04-29 08:19:58 it makes sense only on distributions that do not support pax 2015-04-29 08:20:15 ok 2015-04-29 08:21:51 like arch, we have paxd here for that 2015-04-29 08:22:31 we don't want paxd 2015-04-29 08:22:46 it make sense only when package manager upgrades your packages and resets the packages 2015-04-29 08:22:52 without pax flags 2015-04-29 08:23:04 in alpine we build everything paxmarked with both ELF and Xattr markings (since 3.2) 2015-04-29 08:23:18 apk audit will yell at you if you run paxd and it changes things under 2015-04-29 08:23:45 and seriously 2015-04-29 08:23:56 if you want security (since you run pax/grsec) 2015-04-29 08:24:05 and you have daemon that can alter binaries behind your back. that's insanity 2015-04-29 08:34:02 fabled: I know we don't, it makes no sense to use it here 2015-04-29 08:37:18 paxctl is used to change/alter those markings ? 2015-04-29 08:37:38 yes 2015-04-29 08:37:46 paxctl modifies the binary itself, by writing new ELF header 2015-04-29 08:38:06 paxctl-ng allows you to choose whether to write to the binary, or xattrs 2015-04-29 08:38:21 in alpine 2015-04-29 08:38:28 if I modify lots of them (in theory only), how to keep track of them ? 2015-04-29 08:38:28 the packaging script should use 'paxmark' which does both 2015-04-29 08:38:45 vkrishn, in alpine, APKBUILD should do it 2015-04-29 08:38:51 there's no need to modify them after package install 2015-04-29 08:39:09 the whole paxd stuff is for system whose packages are not built with paxflags 2015-04-29 08:39:25 ok 2015-04-29 08:39:52 since we want the packager to do the marking at package build time 2015-04-29 08:39:57 paxd is useless 2015-04-29 08:40:05 no. paxd is harmful, in alpine. 2015-04-29 08:40:10 :) 2015-04-29 08:44:27 fabled: stupid question, but is it possible to scan binaries which need markings? 2015-04-29 08:44:38 usually no 2015-04-29 08:44:57 marking is needed on binaries that try to do certain things pax forbids by default 2015-04-29 08:45:14 for instance, i packaged mono 2015-04-29 08:45:17 usual suspects are JIT/VM implementations 2015-04-29 08:45:23 it uses jit 2015-04-29 08:45:32 yes, it'll need marking likely 2015-04-29 08:45:48 so the only way is to try? 2015-04-29 08:46:11 on mono and similar VM only things; i would not mind marking all executables 2015-04-29 08:46:30 infact, i think that is what is done with openjdk 2015-04-29 08:50:42 fabled: ok 2015-04-29 08:50:58 fabled: so paxmark replaces paxctl-ng? 2015-04-29 08:51:48 yes 2015-04-29 08:51:55 use paxmark to mark binaries 2015-04-29 08:52:03 it'll create both, the ELF and the xattr, markings 2015-04-29 08:52:19 it currently uses paxctl for ELF, and setfattr for xattr 2015-04-29 08:52:36 if we want to change to paxctl-ng later on, we just update paxmark to use it 2015-04-29 08:53:38 i was reading that script, but couldnt make much sence of it. 2015-04-29 09:03:55 main/attr url is http://acl.bestbits.at/ , better would be http://savannah.nongnu.org/projects/attr 2015-04-29 09:12:27 ok, thats he hiredis error clandmeter talked about 2015-04-29 09:12:28 hum 2015-04-29 09:16:00 ncopa: yes, like http://git.alpinelinux.org/cgit/aports/commit/?id=dad9b58f3f17337c4ec54ef89d9ac49894fd0565 2015-04-29 09:16:03 do we want that? 2015-04-29 09:16:19 not really 2015-04-29 09:16:24 its a hiredis upstream bug 2015-04-29 09:17:14 we could also revert hiredis and wait for updtream 2015-04-29 09:19:30 ncopa: isnt it easier to just let abuild go over the build errors and generate a list of broken pkgs? 2015-04-29 09:19:40 we cna then devide the list 2015-04-29 09:19:53 yes 2015-04-29 09:19:56 you are right 2015-04-29 09:20:14 im going the same thing as you are doing. its a matter who is first.. 2015-04-29 09:20:37 im sure barthalion is happy to pitch in ;-) 2015-04-29 09:21:09 if you have the list, sure 2015-04-29 09:21:44 I don't like "check build-edge-*.log, find the package, fix it, repeat" procedure 2015-04-29 09:22:10 +1 2015-04-29 09:23:14 http://sprunge.us/UeZY 2015-04-29 09:23:24 thats one list 2015-04-29 09:24:05 thats from 3.2? 2015-04-29 09:24:09 yes 2015-04-29 09:24:10 also, I'm actually at work, so I can help once I get back home 2015-04-29 09:24:34 wednesday is not the best day for dealing with non-work related it stuff 2015-04-29 09:24:41 there are a bunch of other packages that does not even try build due to missing deps 2015-04-29 09:24:53 so once the v3.2 list is fixed, we will have a new list 2015-04-29 09:25:31 seems nice, https://kallithea-scm.org/repos/kallithea 2015-04-29 09:25:32 edge builders have ~20 pkgs i think 2015-04-29 09:25:55 we shoudl probably try prioritize edge builders 2015-04-29 09:35:15 v3.2 list splitted in 3: http://sprunge.us/jgOD 2015-04-29 09:36:25 it's all headers or any build failure? 2015-04-29 09:37:37 I take the third section 2015-04-29 09:37:49 second 2015-04-29 09:41:47 barthalion: no 2015-04-29 09:42:26 hiredis has done a pretty bad mistake 2015-04-29 09:42:41 fixing it properly will break new api 2015-04-29 10:00:43 build/build-edge-armhf builds firefox like in 1 sec :)) 2015-04-29 11:14:20 main built: 14 2015-04-29 11:14:20 main tried: 19 2015-04-29 11:14:20 main deleted: 993 2015-04-29 11:14:22 hum 2015-04-29 11:14:36 i wonder what i messed up now... 2015-04-29 11:15:14 oops ;) 2015-04-29 11:16:01 i am dangerous nowdays 2015-04-29 11:16:37 if you have ideas how to improve things - now is a bad time to present those :) 2015-04-29 11:17:25 lol 2015-04-29 11:17:39 are you refering to somebody in particular? 2015-04-29 11:18:23 not really :) 2015-04-29 11:18:35 i had great ideas discussion yesterday too 2015-04-29 11:18:40 how to proerly do -dev packages 2015-04-29 11:18:45 properly do -dev packages 2015-04-29 11:19:50 i wonder if i should rsync from a mirror 2015-04-29 11:19:59 regarding that 2015-04-29 11:20:01 sounds so 2015-04-29 11:20:18 unless you want to waste next week on more build failures like last time 2015-04-29 11:20:24 shouldnt we also exclude -lang from -dev? 2015-04-29 11:20:37 ACTION hides 2015-04-29 11:20:43 clandmeter: please. wait til post v3.2 2015-04-29 11:33:40 >>> ERROR: umurmur: source 0.2.16.tar.gz needs to be renamed to avoid possible collisions 2015-04-29 11:33:44 cool 2015-04-29 11:33:51 we have it catching things 2015-04-29 11:34:24 :D 2015-04-29 11:34:38 so something good happend today atleast 2015-04-29 11:34:59 really, alot good has happened 2015-04-29 11:35:11 the things that breaks are things that needs be fixed 2015-04-29 11:42:23 whee 2015-04-29 11:42:48 i think edge builders are ok again 2015-04-29 11:42:56 thanks a lot for helping me clean up my mess 2015-04-29 11:43:44 I'll continue with the first bulk: http://sprunge.us/jgOD 2015-04-29 11:44:10 owncloud-texteditor.tar.gz: FAILED 2015-04-29 11:47:36 barthalion: what happend to arch ftp? 2015-04-29 11:48:05 https://sources.archlinux.org/ 2015-04-29 11:48:32 clandmeter: ↑ 2015-04-29 11:49:27 argh 2015-04-29 11:49:47 /etc/init.d/sysctl is broken for me in edge. has this been spotted and mentioned? 2015-04-29 11:49:50 stateless: gnulib is causing pain for us 2015-04-29 11:50:10 ScrumpyJack: ? i thight thtat i fixed it? 2015-04-29 11:50:16 ScrumpyJack: which arch? 2015-04-29 11:50:26 please leave my packages chunk, I'll handle it this evening 2015-04-29 11:50:41 ncopa: i686 2015-04-29 11:51:26 what error do you get? 2015-04-29 11:51:36 barthalion: trhx 2015-04-29 11:51:49 shell tests broken or something 2015-04-29 11:51:52 [ 2015-04-29 11:52:05 i saw that but hum 2015-04-29 11:52:15 i have no idea how, and where 2015-04-29 11:52:59 ScrumpyJack: does the relevant line looks like: if [ -e "$i"..]; then 2015-04-29 11:53:02 or similar 2015-04-29 11:53:06 the .. before ] 2015-04-29 11:53:48 there are no ... 2015-04-29 11:54:00 ok 2015-04-29 11:54:01 hiredis works :D 2015-04-29 11:54:42 #! is /sbin/openrc-run 2015-04-29 11:54:56 clandmeter: https://github.com/redis/hiredis/issues/321 2015-04-29 11:55:27 openrc-run is correct 2015-04-29 11:56:01 sh is the one complaining about the tests 2015-04-29 11:56:09 gtg 2015-04-29 12:01:16 ncopa: http://sprunge.us/jaWO 2015-04-29 12:01:28 is that fortify? 2015-04-29 12:02:23 yes 2015-04-29 12:02:25 i think so 2015-04-29 12:02:28 mplayer has same issue 2015-04-29 12:02:53 most likely there are something that does: #define inline 2015-04-29 12:03:09 which turns "inline" into "" 2015-04-29 12:03:24 doing that will break fortify 2015-04-29 12:04:03 i have also seen a couple of gnulib functions that breaks fortify 2015-04-29 12:05:48 heh 2015-04-29 12:05:50 this is fun 2015-04-29 12:06:00 $ ls -l stdio.h 2015-04-29 12:06:01 -rw-r--r-- 1 ncopa ncopa 13046 Apr 29 12:05 stdio.h 2015-04-29 12:06:08 $ make 2015-04-29 12:06:09 ... 2015-04-29 12:06:18 make: *** [all] Error 2 2015-04-29 12:06:24 so i remove stdio.h 2015-04-29 12:06:40 and after running make again 2015-04-29 12:06:43 it comes back! 2015-04-29 12:06:55 that is so typical gnulib 2015-04-29 12:07:03 its so hard to get rid of it 2015-04-29 12:07:10 its so hard to fix it 2015-04-29 12:41:35 https://gist.github.com/zeha/7652abca3725daad811e list of removed pkgs from jessie, maybe useful while we are at next release 2015-04-29 13:23:44 till now my experiment with the new consul package are very pleasant ^^ thanks to all of you for providing it ;) With it I'm fullfilling a layer in the IaaS solution I'm building with AL only. Next step will be envconsul. I will try to package it following the consul APKBUILD 2015-04-29 13:38:06 Jean-Scotch: consul package is maybe not the best example of a package, but i dont think any go pkg is. 2015-04-29 13:38:40 yeah, go isn't really designed for good packaging 2015-04-29 13:39:02 and for sure not consul 2015-04-29 13:39:10 go seems to be packaging kinda all kinds of pkgs, maybe someone starts a alpinelinux-go 2015-04-29 13:39:22 i did some other pkg, which was kind of okish 2015-04-29 13:39:26 that pkgs only core linux+go pkgs 2015-04-29 14:03:48 barthalion: i did libsndfile and libnl3 from your list 2015-04-29 14:08:57 domo arigato 2015-04-29 14:24:27 BitL0G1c - I've tried a new install of alpine-3.1.3, set edge/main and edge/testing in "/etc/apk/repositories", did an "apk update; apk upgrade" installed networkmanager, and I still can't get nm to work? 2015-04-29 14:25:07 trebor - you may need udev - on my machine I have also done http://wiki.alpinelinux.org/wiki/XFCE_Setup 2015-04-29 14:25:59 nmcli n on, nmcli n off, nmcli n c also don't seem to do anything. 2015-04-29 14:26:20 & http://it-offshore.co.uk/linux/alpine-linux/30-alpine-linux-spice-kvm-desktop 2015-04-29 14:26:48 I did find an error in the logs: grsec: From 10.0.2.2: Segmentation fault occurred at 0000000000000008 in /usr/bin/nmcli[nmcli:1523] uid/euid:0 2015-04-29 14:28:09 ok - I'm on kernel 3.14.37-0-grsec 2015-04-29 14:29:39 ok, will try the infos in those links. I was planning on using X and virt-viewer anyway. 2015-04-29 14:30:21 should I not have done an "apk upgrade"? That bumped my kernel version. 2015-04-29 14:30:21 what does 'uname -r' show ? 2015-04-29 14:30:50 3.18.12-1-grsec 2015-04-29 14:31:04 ok - there's the difference 2015-04-29 14:32:27 Does that mean I should do a fresh install and only "apk add networkmanager@testing" or such, without upgrading? 2015-04-29 14:34:08 I have my repos set to 3.1 & testing / edge pinned 2015-04-29 14:35:15 http://hastebin.com/feliruyimo.rb 2015-04-29 14:35:20 Ok, I'll start another try. 2015-04-29 14:35:40 I'll see if i can get it to segfault 2015-04-29 14:36:31 I just did an update & the kernel is still 3.14 : (1/3) Upgrading linux-grsec (3.14.37-r0 -> 3.14.39-r0) 2015-04-29 14:36:31 Ok, I'll get back once I've tried my luck... 2015-04-29 14:37:55 if you pin repos you can upgrade & downgrade packages individually from edge 2015-04-29 14:38:23 to upgrade to 3.18 I 'apk add linux-grsec@edge' 2015-04-29 14:38:23 ' 2015-04-29 14:39:15 I've got "edge/main" "edge/testing" set; and did an "apk update;apk upgrade". 2015-04-29 14:42:00 I don't get any segfault on 3.18 2015-04-29 14:42:12 i think you just need udev 2015-04-29 14:43:32 http://hastebin.com/azacehicoj.xml - I'm also using the virtio network driver 2015-04-29 14:50:02 BitL0G1c - I installed udev yesterday after you sent me your "rc-update show", but that didn't help. I'll try it again in the new install. 2015-04-29 14:50:32 just try adding xfce as shown on the wiki 2015-04-29 15:02:28 ncopa: gst-ffmpeg-0.10 freerdp perl-mail-clamav 2015-04-29 15:02:36 those are the ones i have issues with 2015-04-29 15:08:39 ok. time to go home 2015-04-29 15:08:50 clandmeter: i'll have a look at those tonight, or tm 2015-04-29 15:08:52 thanks! 2015-04-29 15:22:54 BitL0G1c - It's working, without my having installed X/xfce! Now I'll try to find out why. Thanks for your help and patience. 2015-04-29 15:41:23 trebor - you were probably missing dbus / policykit / udev or something similar - if you work out what it was please let me know 2015-04-29 15:45:21 BitL0G1c - I'm trying to work it out. I'll let you know once I've found my mistake. 2015-04-29 15:49:39 ;-) gracias 2015-04-29 15:52:59 BitL0G1c - Have you looked at cjdns by chance? 2015-04-29 15:57:41 no - I use dnscrypt-proxy with unbound as a caching dns server 2015-04-29 15:58:06 install dnscrypt-setup & run 'dnscrypt-setup' 2015-04-29 16:00:21 cjdns looks great 2015-04-29 16:00:30 It's quite a bit more than secure dns; I just thought you might like it. Have you seen bup or sshuttle? 2015-04-29 16:02:51 bup is different thing than sshuttle, unless you mean different bup 2015-04-29 16:03:49 Totally different (same developer though), but I like both. 2015-04-29 16:03:57 oh, didn't know about that 2015-04-29 16:05:36 cjdn is totally unrelated to dns 2015-04-29 16:05:40 *cjdns 2015-04-29 16:15:24 yes - i was using tinc to connect machines over ipv6 - this is an alternative 2015-04-29 16:22:13 Does alpine have any official tools like debootstrap to get/make an even smaller custom install? 2015-04-29 16:27:15 trebor http://wiki.alpinelinux.org/wiki/How_to_make_a_custom_ISO_image 2015-04-29 16:28:38 crow - Ah, nice thanks. 2015-04-29 17:03:08 trebor: just use mini iso or apk-tools directly 2015-04-29 18:15:26 It seems alpine is missing a few essential ;-) packages: kexec-tools and libpam-script. 2015-04-29 19:47:48 a lot of packages from my chunk are false negatives 2015-04-29 19:48:01 ncopa, clandmeter ↑ 2015-04-29 19:55:21 wat 2015-04-29 19:56:25 it makes no sense, I just fixed the checksums 2015-04-29 19:56:50 anyway, 11 left from my part 2015-04-29 20:52:52 ncopa: dude 2015-04-29 20:52:54 ncopa: go to bed 2015-04-30 06:18:11 i am removing some of the old alpine releases from mirrors 2015-04-30 06:18:25 edge-uclibc, v2.0, v2.1 and v2.2 2015-04-30 06:18:30 nobody should use those 2015-04-30 06:18:56 and in emergency, we have them on ancient.alpinelinux.org 2015-04-30 07:01:06 ncopa: is icu broken on edge? 2015-04-30 07:01:26 ive seen multiple complaints about broken aports 2015-04-30 07:07:20 clandmeter: not that i know? 2015-04-30 07:07:37 3 users complain 2015-04-30 07:07:49 but could be its mixed repo's 2015-04-30 07:07:52 not sure 2015-04-30 07:17:27 there were some icu-dependent stuff on the fabled's image he posted few days ago 2015-04-30 07:37:56 uh? 2015-04-30 07:39:21 ah, "alpine" failed 2015-04-30 07:41:06 yeah, I don't see why, the checksum was correct when I pushed it 2015-04-30 07:46:07 barthalion: could be the source is corrupt on distfiles 2015-04-30 07:48:42 ncopa: can you remove source from builders distfiles for alpine (the email client) :) 2015-04-30 07:52:26 v3.2 or edge? or both 2015-04-30 07:52:35 edge for sure 2015-04-30 07:52:44 it should fix the build 2015-04-30 07:53:19 alpine-2.20.tar.xz? 2015-04-30 07:53:24 yeah 2015-04-30 07:53:36 done 2015-04-30 07:53:41 atleast we think thats it 2015-04-30 08:13:48 (y) 2015-04-30 08:19:46 clandmeter, gotta love pwclient 2015-04-30 08:21:08 we should expand it to alpine-devel btw 2015-04-30 08:21:43 why 2015-04-30 08:21:44 +1 2015-04-30 08:21:47 ? 2015-04-30 08:22:06 someone send patches to alpine-devel ? 2015-04-30 08:22:14 shouldnt ppl just post to aports? 2015-04-30 08:22:23 patches related to abuild? 2015-04-30 08:22:25 or apk-tools? 2015-04-30 08:22:30 these are send to alpine-devel 2015-04-30 08:23:11 current setup doesnt handle them. 2015-04-30 08:23:27 atleast not automatic approving 2015-04-30 08:23:53 hm, indeed 2015-04-30 08:24:00 no idea why I thought it's so simple 2015-04-30 08:24:10 probably because I didn't drink coffee yet 2015-04-30 08:24:12 if you do everyting manually, it is 2015-04-30 08:24:23 im not saying we shouldnt 2015-04-30 08:25:18 clandmeter, i'm still drafting the doc as FAQ format 2015-04-30 08:25:35 ok 2015-04-30 08:25:57 can you put something lready on wiki? 2015-04-30 08:25:58 after pwclient apply $PATCH_ID, we test that patch actually works 2015-04-30 08:26:26 once this is done, and is working, we pwclient git-am $PATCH_ID 2015-04-30 08:26:56 i think thats the best way, but im open for suggestions. 2015-04-30 08:28:02 it's fine. 2015-04-30 08:28:43 does pwclient commits ? 2015-04-30 08:28:54 stateless: i have figured out why libavutil does not build with fortify 2015-04-30 08:28:57 or after pwclient git-am i need to git commit ? 2015-04-30 08:29:01 and then git push? 2015-04-30 08:29:23 I do pwclient git-am to test if patch works 2015-04-30 08:29:33 I can always do git reset if it doesn't 2015-04-30 08:29:40 or fix it and amend the change 2015-04-30 08:29:45 then git push 2015-04-30 08:30:18 between git-am and git push, how do you commit? 2015-04-30 08:30:28 git commit? 2015-04-30 08:30:34 or pwclient supports commit? 2015-04-30 08:30:38 why would I commit if I don't change anything 2015-04-30 08:30:43 git-am runs git am 2015-04-30 08:30:51 git am "imports" the commit 2015-04-30 08:31:41 ok. So pwclient apply complicates the thing 2015-04-30 08:31:52 because actually modified the local git tree 2015-04-30 08:31:58 exactly 2015-04-30 08:32:11 and pwclient apply doesn't retain commit ownership 2015-04-30 08:32:12 so you need to git checkout 2015-04-30 08:32:20 and then pwclient git-am 2015-04-30 08:32:37 so you should only run 'pwclient apply' on separate branch 2015-04-30 08:32:45 but I don't see any advantage of doing so 2015-04-30 08:33:02 just do git-am, and if change is broken, run git reset --hard @~1 2015-04-30 08:36:19 fcolista: if you dont mind resettting your git tree, then a direct pwcleint git-am is ok 2015-04-30 08:36:45 yes, sounds a better option. 2015-04-30 08:37:02 well, it's not better option, it's just my workflow 2015-04-30 08:37:10 it's git, you do what you find convenient to you 2015-04-30 08:37:24 barthalion, i'm going to describe the workflow on wiki 2015-04-30 08:38:10 so, could be fine to have a consistent way (then, each one can do as he prefers) 2015-04-30 08:51:19 please correct me if i'm wrong. If i pwclient apply, then i need anyway to do pwclient git-am. 2015-04-30 08:51:45 'cause if i'm going to do git commit, it would result that i'm the one who is committing 2015-04-30 08:52:00 and not the original guy who sent the patch 2015-04-30 09:06:15 fcolista: apply just applies to patch and does not do anytning git related 2015-04-30 09:06:57 Correct. What's the next step then? 2015-04-30 09:07:02 pwclient git-am 2015-04-30 09:07:05 no 2015-04-30 09:07:07 i need to commit the change 2015-04-30 09:07:08 git checkout -- . 2015-04-30 09:07:12 oh yes 2015-04-30 09:07:15 and pwclient git-am 2015-04-30 09:07:17 right, sure. 2015-04-30 09:07:28 we want to keep authorship of the change 2015-04-30 09:07:33 cause local tree got modified by pwclient apply 2015-04-30 09:07:43 just took for granted. 2015-04-30 09:07:44 ok 2015-04-30 09:12:28 once patch is committed with git am, before pushing is it possible to git pull --rebase ? 2015-04-30 09:13:41 yes, as usually 2015-04-30 09:13:48 ok. 2015-04-30 09:14:06 you need to have your changes rebased before the push 2015-04-30 09:14:14 that's how we roll 2015-04-30 09:14:19 or get slapped 2015-04-30 09:15:56 dat algitbot abuse 2015-04-30 09:15:59 ACTION slaps vkrishn with a mosquitto msg 2015-04-30 09:16:09 he's not even sitting here 2015-04-30 09:16:16 he is watching 2015-04-30 09:17:05 I know 2015-04-30 09:17:10 no privacy at all 2015-04-30 09:23:06 clandmeter, i'm going to paste patchwork notes on wiki. It's very draft, so it can be improved in order to be a real wiki doc :) 2015-04-30 09:23:15 soudns good if i put it on w.a.o ? 2015-04-30 09:23:19 *sounds 2015-04-30 09:23:27 yes 2015-04-30 09:23:28 please 2015-04-30 09:24:02 its a wiki, its made to collaborate 2015-04-30 09:24:13 yeah, that's the idea :) 2015-04-30 09:44:38 url http://www.codenix.com/ in main/tolua++ may not be available 2015-04-30 09:46:26 vkrishn: do you have any idea where tolua++ is maintained now? 2015-04-30 09:46:33 seem like its abandoned 2015-04-30 09:47:02 clandmeter, ncopa, barthalion : http://wiki.alpinelinux.org/wiki/Patch_Workflow 2015-04-30 09:47:12 let's the doc begine 2015-04-30 09:47:14 *begin 2015-04-30 09:48:34 github has repo, but not ++, option https://github.com/LuaDist/tolua 2015-04-30 09:50:43 which is in testing/ 2015-04-30 09:51:24 i think conky need tolua++ 2015-04-30 09:52:48 but it can be build without it 2015-04-30 09:53:23 what reminds me to bump syslog-ng 2015-04-30 09:54:08 https://github.com/LuaDist/toluapp 2015-04-30 09:55:09 atleast the download link is available, though not active 2015-04-30 09:55:10 ah 2015-04-30 09:55:13 thats nice 2015-04-30 09:57:19 nice qemu 2.3 :-)) 2015-04-30 10:06:09 ncopa: please remove eventlog_0.2.12.tar.gz from distfiles 2015-04-30 10:06:22 (and retrigger the build, if it's possible without pkgrel bump) 2015-04-30 10:20:27 hmm... did not come to know what went poof with aports git ! 2015-04-30 10:23:34 there is a technique where you can branch the current master or particular commit(till where it is ok), then swap that branch with master 2015-04-30 10:24:01 git reset is always there 2015-04-30 10:25:39 and we all have various dated backups 2015-04-30 10:32:19 barthalion: done 2015-04-30 10:36:50 thx 2015-04-30 10:47:23 ncopa: you also removed edge-uclibc? 2015-04-30 11:07:48 clandmeter: yes 2015-04-30 11:23:36 sorry about mqtt msg 2015-04-30 11:24:20 pwclient seems nice tool 2015-04-30 17:14:03 sys install attempt of alpine linux edge and i get http://sprunge.us/ULNW 2015-04-30 17:15:31 i see there is a bug #4060 about it 2015-04-30 17:15:39 any movement? 2015-04-30 17:34:28 it runs ok with -L - I just installed with it on my patched version http://it-offshore.co.uk/downloads/setup-disk.new sfdisk still shows an option for --Linux too 2015-04-30 17:53:20 BitL0G1c: what's that url? 2015-04-30 17:55:30 http://it-offshore.co.uk/downloads/setup-disk.new - I used it today for a reinstall on a vps 2015-04-30 17:56:13 (with a custom lvm scheme) 2015-04-30 17:57:42 to setup luks or lvm I use http://it-offshore.co.uk/downloads/setup-lvm - you can wget them both 2015-04-30 18:45:59 any help as why main/libtool/APKBUILD says depends="bash libltdl" ? 2015-04-30 19:07:49 hey all – quick question on the PAX stuff if anyone knows much about it... 2015-04-30 19:08:20 I notice a bunch of changes have been made recently like this: http://git.alpinelinux.org/cgit/aports/commit/?id=47db907cdf7ac0669d2729f9aa146dcfdbf76a82 2015-04-30 19:09:28 Can anyone explain what paxmark is doing there that paxctl didn't previously? 2015-04-30 19:10:51 Because it seems to be resulting in binaries that are much larger than they were previously 2015-04-30 19:12:47 Alternatively, does anyone know if Timo Teräs is on Twitter (or here?) – he seems to be the one making most of the changes 2015-04-30 19:33:57 Ok, I'll email him 2015-04-30 20:25:15 desktop:~# rc-service udev status 2015-04-30 20:25:15 * rc-service: No such file or directory 2015-04-30 20:25:24 any idea why this appen? ^^ 2015-04-30 20:25:29 happen*