2016-04-01 06:26:14 quick question, if a package that is already in main/ suddenly requires 2 new aports to be created, will these aports go into testing first or directly into main ? since package wont compile without new libs 2016-04-01 06:26:25 suqdiq: those normally go directly to main 2016-04-01 06:28:39 re python3 2016-04-01 06:28:43 i want that 2016-04-01 06:28:55 but nobody has had the time to work on it :-/ 2016-04-01 06:31:05 ncopa: https://bugs.alpinelinux.org/issues/5357 hope this is enough 2016-04-01 06:31:20 coredumb its enough 2016-04-01 06:31:31 one question, you say it needs to be rebuilt with dovecot 2016-04-01 06:31:47 does it mean if minor version updates, the plugin needs to be rebuilt too? 2016-04-01 06:32:46 for example, if you upgrade from dovecot-2.2.22 to 2.2.23, do we need to rebuild the plugin too? 2016-04-01 06:33:51 ncopa: I'm not sure about it 2016-04-01 06:33:54 http://wiki2.dovecot.org/Pigeonhole/Installation#Compiling 2016-04-01 06:34:02 that's what I understand from there 2016-04-01 06:35:02 "IMPORTANT: You need to recompile Pigeonhole when you upgrade Dovecot to a new version, because otherwise the Sieve interpreter plugin will fail to load with a version error. " 2016-04-01 06:35:07 does it check the full version 2016-04-01 06:35:14 or only the ABI version 2016-04-01 06:36:10 that I don't know 2016-04-01 06:36:27 i am ok with backporting it anyway (i think) but it is good to know that we need to rebuild it in case we sec fix dovecot 2016-04-01 06:36:28 but I think if said like that => full version 2016-04-01 06:36:50 ncopa: yeah that's why I noted it in the bug 2016-04-01 06:36:51 thats what it sounds like yes 2016-04-01 06:37:04 i'm glad you mentioned it 2016-04-01 06:37:07 need to check that 2016-04-01 06:37:44 ok 2016-04-01 06:37:56 is there some quick way to test if the plugin works? 2016-04-01 06:38:19 apk add dovecot dovecot-pigeonhole-plugin && /etc/init.d/dovecot start 2016-04-01 06:38:20 ? 2016-04-01 06:38:31 and grep the logs? 2016-04-01 06:38:50 actually on EL the plugin package has the exact same version as the dovecot package 2016-04-01 06:39:07 so I guess they rebuild the plugin with dovecot everytime 2016-04-01 06:39:18 the current dovecot-2.2.22 was built after dovecot-pigeonhole-plugin 2016-04-01 06:39:30 sounds like they build it together with the dovecot package yes 2016-04-01 06:39:39 I have not tested the community package to be honest 2016-04-01 06:39:55 but it means that current plugin in edge is broken 2016-04-01 06:40:03 It may very well be 2016-04-01 06:40:11 could you please test it? in a vm, in docker or lxc or whatever? 2016-04-01 06:40:24 i dont want backport anything that is not tested 2016-04-01 06:40:25 yes let me clone a VM 2016-04-01 06:41:03 you actually need to enable the plugin in LDA and LMTP protocols to have it working 2016-04-01 06:41:18 so just installing it is not sufficient to test 2016-04-01 06:42:12 let me see if I can do something efficient that won't take me half a day to test it 2016-04-01 06:43:28 just enable the plugin would be engouhg 2016-04-01 06:43:40 hello, i made a patch that disables stack size checking in qt5-declarative when using musl. can i submit that without using git send-email ? 2016-04-01 06:44:21 mangelis: you can: apk add tpaste && git format-patch -1 --stdout | tpaste 2016-04-01 06:44:23 and paste the url here 2016-04-01 06:44:29 or do a github pull request 2016-04-01 06:44:45 https://github.com/alpinelinux/aports/pulls 2016-04-01 06:45:47 ncopa: not working 2016-04-01 06:46:09 http://pastebin.com/yhkSADPc 2016-04-01 06:46:18 first doesn't depend on dovecot 2016-04-01 06:46:30 and well fails at version as expected 2016-04-01 06:47:07 at least error is crystal clear ^^ 2016-04-01 06:49:47 let me compile the latest version and make a patch 2016-04-01 06:50:16 oh 2016-04-01 06:50:34 seems like dovecot abi changes with minor versions 2016-04-01 06:50:36 not good 2016-04-01 06:50:57 one way to solve it would be to build it from the dovecot apkbuild 2016-04-01 06:51:07 ncopa: yep 2016-04-01 06:51:16 i think that makes most sense 2016-04-01 06:51:19 I think other distros do that as well 2016-04-01 06:51:26 ok 2016-04-01 06:51:30 lets do that then 2016-04-01 06:51:40 just check so we dont introduce circular deps 2016-04-01 06:52:15 http://pkgs.fedoraproject.org/cgit/rpms/dovecot.git/tree/dovecot.spec < source8 2016-04-01 06:55:37 ncopa: you want me to do it ? 2016-04-01 06:57:43 i will not be able to do it today 2016-04-01 06:57:50 so yes, it would be nice if you could 2016-04-01 06:59:39 should I only patch testing dovecot ? 2016-04-01 07:00:08 patch main/dovecot 2016-04-01 07:00:12 in edge 2016-04-01 07:01:04 yeah that's what I meant sorry :D 2016-04-01 07:02:21 ncopa, http://tpaste.us/36vn 2016-04-01 07:03:13 mangelis: can you explain a bit why this is needed? 2016-04-01 07:03:45 why do we need to disable a stack check? 2016-04-01 07:03:55 do we run out of stack space? 2016-04-01 07:04:17 musl has alot smaller default stack size for pthreads 2016-04-01 07:04:20 its 80k 2016-04-01 07:04:23 while glibc uses 8MB 2016-04-01 07:04:37 because qt tries to check stack size using glibc-specific extension in pthreads 2016-04-01 07:04:47 ok 2016-04-01 07:04:53 and musl implements it differently 2016-04-01 07:04:55 could you add that to the commit message 2016-04-01 07:05:29 "qt tries to check stack size using glibc-specific extension in pthreads" 2016-04-01 07:05:34 good enough commit message 2016-04-01 07:05:44 or write that at the top in the patch file 2016-04-01 07:06:11 okay 2016-04-01 07:09:53 would also be nice if you could submit it upstream 2016-04-01 07:10:15 oh, you also need bump pkgrel 2016-04-01 07:10:23 http://tpaste.us/GVed is this ok? 2016-04-01 07:10:25 increase pkgrel with 1 2016-04-01 07:10:43 good enough 2016-04-01 07:11:00 i can bump pkgrel for you 2016-04-01 07:11:03 thanks 2016-04-01 07:12:22 by upstream you mean doing similar patch to the master branch of aports? 2016-04-01 07:12:59 hello 2016-04-01 07:13:38 do you have any idea about this virt-manager error: Error: g-invoke-error-quark: Could not locate gvir_init: Error relocating /usr/lib/libvirt-glib-1.0.so.0: libintl_fprintf: symbol not found (1) 2016-04-01 07:13:51 libintl is installed, indeed 2016-04-01 07:14:15 mangelis: i mean qt upstream. send a bug/patch to the qt project 2016-04-01 07:14:39 ah, right 2016-04-01 07:15:16 although i think musl should perhaps fix their implementation of the glibc extension ;) 2016-04-01 07:15:17 I guess that libintl needs to be rebuild with libintl_fprintf symbol, but I'm not sure 2016-04-01 07:15:38 mangelis: then you file a bug request or patch to musl ;) 2016-04-01 07:15:48 i dont know if the extension makes sens to implement 2016-04-01 07:16:14 Mo0O: i have no idea sorry 2016-04-01 07:16:35 np, I'll continue to investigate on my free time :) 2016-04-01 07:16:38 thx ncopa 2016-04-01 07:17:06 mangelis: you could ask in #musl if they want or if they are planning to implement the extension 2016-04-01 07:18:10 mangelis: the patch was against 3.3-stable? 2016-04-01 07:18:15 yes 2016-04-01 07:20:34 i'll push this: http://tpaste.us/GmN7 2016-04-01 07:20:36 thanks! 2016-04-01 07:21:35 :) 2016-04-01 07:27:33 ncopa: re py3k, what is the main work you refer to? creating the actual packages or to decide how to organize them? 2016-04-01 07:29:45 Y0Gi: both 2016-04-01 07:30:02 first we need figure out what the options are 2016-04-01 07:30:10 then we need decide how to move forward 2016-04-01 07:30:13 then we do it 2016-04-01 07:30:56 alright. there has been some discussion here so far for a start. 2016-04-01 07:31:28 first question probably is if py2 support should be kept. and I see reason enough to do so. 2016-04-01 07:32:04 i think so too 2016-04-01 07:32:09 scares me to kill it 2016-04-01 07:32:35 i need to go out for a while 2016-04-01 07:32:35 bbl 2016-04-01 07:33:00 bye 2016-04-01 07:55:45 ncopa: sorry, I've found the origin of this error, it's was because I was mixing stable/main with edge/community, my bad 2016-04-01 08:29:24 ncopa: what's the order of processing of abuild ? it builds then install then processes all sub packages right ? 2016-04-01 09:21:45 it looks like there is a problem with php7-gd in testing 2016-04-01 09:22:09 i get a "so:libwebp.so.5 (missing)" 2016-04-01 09:22:29 and libwebp only has a libwebp.so.6 2016-04-01 09:44:40 how should one handle a subpackage that itself contains man and dev files ? 2016-04-01 12:04:43 hey :) 2016-04-01 12:04:54 has anyone else this problem on edge? 2016-04-01 12:04:56 ERROR: unsatisfiable constraints: 2016-04-01 12:04:56 libvpx-1.4.0-r1: 2016-04-01 12:04:56 conflicts: libvpx-1.5.0-r0 2016-04-01 12:04:56 satisfies: gst-plugins-good1-1.6.3-r0[so:libvpx.so.2] firefox-45.0.1-r0[so:libvpx.so.2] 2016-04-01 12:04:56 libvpx-1.5.0-r0: 2016-04-01 12:04:56 conflicts: libvpx-1.4.0-r1 2016-04-01 12:04:56 satisfies: ffmpeg-libs-3.0.1-r0[so:libvpx.so.3] 2016-04-01 12:42:59 coredumb: usually we only have one -dev and -doc subpackage per package so a subpackage doesn't have its own -doc and -dev subpackages 2016-04-01 12:43:46 leo-unglaub: yes, I solved it by purging ffmpeg and libvpx, upgrading and reinstalling them afterwards 2016-04-01 12:48:23 nmeum: yeah "usually" 2016-04-01 12:48:45 this one integrates a subpackage that would normally be a package on its own 2016-04-01 12:49:00 it's a plugin * 2016-04-01 12:49:13 it has it's own doc and includes then 2016-04-01 12:49:23 means the subpackage will have them 2016-04-01 12:54:27 leo-unglaub: i just pushed the fix for webkitgtk (i think) 2016-04-01 12:54:36 i pushed a big update yesterday 2016-04-01 12:55:03 libvpx, libwebp, live-media, x256 and ffmpeg 2016-04-01 12:55:08 all abi breakages 2016-04-01 12:56:05 ncopa: ^ about pigeonhole plugin having its own man/doc/includes 2016-04-01 12:56:13 @ncopa what about https://github.com/alpinelinux/aports/pull/42, is there anything to do before merging? 2016-04-01 12:56:33 coredumb: i think we can just ship them with dovecot-doc 2016-04-01 12:56:59 and dovecot-dev 2016-04-01 12:57:01 ncopa: ok easier for me then :) 2016-04-01 12:57:03 to make things simple 2016-04-01 12:57:29 was planning to ship them with dovecot-pigeonhole 2016-04-01 12:57:51 jirutka: i am ready for that pull req, just been out today 2016-04-01 13:26:24 ncopa: your fix does not work 2016-04-01 13:26:54 fix for webkit? 2016-04-01 13:27:12 i think the problem is that you have something that holds back the upgrade 2016-04-01 13:27:49 $ sudo apk upgrade -U -a --latest | tpaste 2016-04-01 13:27:49 http://tpaste.us/ARL9 2016-04-01 13:27:58 thats what i have on my workstation 2016-04-01 13:29:30 i have something installed that depends on libvpx or libwebp that is not rebuilt 2016-04-01 13:29:36 and i think that was webkitgtk 2016-04-01 13:30:12 and its not done building yet 2016-04-01 13:35:45 hmmm 2016-04-01 13:35:49 i dont have webkit installed 2016-04-01 13:35:58 at least i hope not *g* 2016-04-01 13:36:28 ok 2016-04-01 13:36:40 leo-unglaub: can you please do: apk upgrade -U -a --latest | tpaste 2016-04-01 13:37:26 and apk version | tpaste 2016-04-01 13:37:32 http://tpaste.us/A9on 2016-04-01 13:37:58 http://tpaste.us/Gnxx 2016-04-01 13:38:43 hum 2016-04-01 13:38:46 no firefox there 2016-04-01 13:39:24 ok 2016-04-01 13:39:27 thats why 2016-04-01 13:40:08 commit c90b3a5e00e512a51e15c504f1a79673e1b77ebd 2016-04-01 13:40:09 Author: Natanael Copa 2016-04-01 13:40:09 Date: Thu Mar 31 10:11:10 2016 +0000 2016-04-01 13:40:09 testing/firefox: rebuild against libvpx-1.5.0 2016-04-01 13:40:13 thats not yet built 2016-04-01 14:04:16 andyshinn: i think we have a problem with the cached 404 2016-04-01 14:04:28 do you think we can reduce cache time for 404 errors? 2016-04-01 14:04:32 to 1 minute or similar? 2016-04-01 14:05:05 or maybe even 10 sec if possible 2016-04-01 14:10:15 or even 1 2016-04-01 14:33:29 nsd 4.1.9 is available 2016-04-01 14:36:08 will push that now 2016-04-01 14:57:19 I set TTL for 404 to 10 seconds, it is still not working? 2016-04-01 14:58:37 ncopa: awesome, thanks! 2016-04-01 15:00:21 andyshinn: still not working 2016-04-01 15:00:41 can you set it to 1? 2016-04-01 15:00:59 i think the backend can handle 1 request per second :) 2016-04-01 15:01:50 andyshinn: the good thing is that i can reproduce it 2016-04-01 15:26:02 Strange. Which URL? I'll check now. Though I had working yesterday 2016-04-01 15:30:10 so is it something like this that happens: APKINDEX gets uploaded before package finished uploading, use gets latest APKINDEX, then tries to grab the package that doesn't exist yet, the 404 gets cached? 2016-04-01 15:33:26 Upgrading critical system libraries and apk-tools: 2016-04-01 15:33:26 (1/1) Upgrading firefox (45.0.1-r0 -> 45.0.1-r1) 2016-04-01 15:33:27 ERROR: firefox-45.0.1-r1: package mentioned in index not found (try 'apk update') 2016-04-01 15:34:30 http://dl-cdn.alpinelinux.org/alpine/edge/testing/x86_64/ 2016-04-01 15:34:41 shows firefox 45.0.1-r0.apk 2016-04-01 15:34:48 rsync.alpinelinux.org has -r1 2016-04-01 15:40:05 strange, i think index is being cached, but the actual -r1 package is downloading for me 2016-04-01 15:41:48 now it came 2016-04-01 18:02:29 imagemagick is broken :( 2016-04-01 18:02:53 ERROR: unsatisfiable constraints: 2016-04-01 18:02:53 so:libwebp.so.5 (missing): 2016-04-01 18:02:53 required by: 2016-04-01 18:02:54 imagemagick-6.9.3.7-r0[so:libwebp.so.5] 2016-04-01 18:03:12 ncopa: is that a known issue on edge? 2016-04-01 18:03:52 something is holding back the libwebp upgrade 2016-04-01 18:04:06 you probably have somethign that depends on so:libwebp.so.4 2016-04-01 18:04:57 but this issue started today 2016-04-01 18:05:23 i pushed the abi breakage yesterday 2016-04-01 18:05:50 i pushed upgrade of libwebp libvpx ffmpeg x265 which all has abi breakage 2016-04-01 18:06:05 i think i have rebuilt all that should be rebuilt 2016-04-01 18:06:12 but i probably missed something 2016-04-01 18:06:34 try `apk upgrade -U -a --latest` 2016-04-01 18:08:31 andyshinn: can you please rebuild the alpine-3.3 image? I just pushed a fix for musl to solve https://github.com/docker-library/ruby/issues/77 2016-04-01 18:08:45 upgraded but still can't install the required packages :( 2016-04-01 18:09:00 something is depending on the older version 2016-04-01 18:09:05 something needs to be rebuilt 2016-04-01 18:09:12 may be a bit tricky to figure out what 2016-04-01 18:09:39 `apk version` should show what wants to upgrade 2016-04-01 18:10:07 try figure what you have installed that depends on one of the mentioned libraries that is not listed there 2016-04-01 18:10:16 i tried apk -U add git mariadb-dev nodejs tzdata imagemagick where it fails 2016-04-01 18:10:27 might also help with `apk upgrade -U -a` 2016-04-01 18:10:55 https://gist.github.com/tboerger/db5575cec5dcee24a0a609ef2f346c70 and this is my world file 2016-04-01 18:11:44 hum 2016-04-01 18:11:46 dont know 2016-04-01 18:11:50 i have to go 2016-04-01 18:11:53 sorry 2016-04-01 18:12:09 :( 2016-04-01 18:12:23 i'll open PR soon for updated base musl 2016-04-01 18:13:34 thanks! have a nice weekend everyone! 2016-04-01 18:18:43 mosez: i can reproduce 2016-04-01 18:18:44 hm 2016-04-01 18:21:05 ok 2016-04-01 18:21:13 i think it only needs a rebuild 2016-04-01 18:24:14 mosez: i pushed update of imagemagick. please try in 5 mins 2016-04-01 18:24:57 cool, thanks! 2016-04-01 18:25:18 ncopa: have a nice weekend! 2016-04-02 02:29:35 I have a version 2 isync patch to send - my mail is down at the moment - will trying sending again in the morning 2016-04-02 06:24:38 zeromq v4.1.4 <-- bump pls 2016-04-02 09:45:25 BitL0G1c: you can post your patch here, I'll try to apply it today 2016-04-02 09:45:49 on the phone now 2016-04-02 16:21:05 where / what does packet_types.h exist as in Alpine/musl? 2016-04-03 15:44:15 hey :) 2016-04-03 16:11:11 hi 2016-04-03 18:14:07 hey everyone! I'm trying to build a new package but I'm struggling with "abuild checksum" step. I keep getting this error: curl: (22) The requested URL returned error: 404 Not Found. 2016-04-03 18:14:35 I triple checked my "source" url and it works just fine for me. Why is it not working via abuild? 2016-04-03 18:15:05 Could you show us your APKBUILD? 2016-04-03 18:16:15 sure! here: http://pastebin.com/KRYWSY3X 2016-04-03 18:17:48 the output of my echo statement is : https://github.com/mkoppanen/php-imagick/archive/phpseven.tar.gz 2016-04-03 18:18:45 And it returns 404… 2016-04-03 18:18:46 so? 2016-04-03 18:18:51 curl: (22) The requested URL returned error: 404 Not Found 2016-04-03 18:18:51 >>> php7-imagick: Updating the md5sums in APKBUILD... 2016-04-03 18:18:51 md5sum: can't open 'phpseven.tar.gz': No such file or directory 2016-04-03 18:18:53 >>> ERROR: php7-imagick: md5sum failed 2016-04-03 18:18:55 ERROR: Service 'php-memcached' failed to build: The command '/bin/sh -c cd php7-imagick && abuild checksum' returned a non-zero code: 1 2016-04-03 18:19:02 try to open https://github.com/mkoppanen/php-imagick/archive/phpseven.tar.gz in browser, it returns 404 2016-04-03 18:19:02 ^^ that's the output 2016-04-03 18:19:30 oh.... I feel dumb 2016-04-03 18:19:52 also there’s no version number in the url 2016-04-03 18:19:55 i had a bad URL 2016-04-03 18:20:14 my issue was that it was mkoppanen/imagick instead of ../php-imagick 2016-04-03 18:20:20 duh! :D thank you! 2016-04-03 18:20:24 np 2016-04-03 18:21:05 btw `$_builddir` should be named `$builddir` (it’s not mentioned on wiki yet) 2016-04-03 18:21:50 oh! thank you! 2016-04-03 18:22:08 this is my first apk package submission... so I'm a newbie 2016-04-03 18:22:26 yay! my checksum just passed 2016-04-03 18:23:42 I think that `--prefix` should be `"$pkgdir"/usr` 2016-04-03 18:24:09 otherwise it will probably install it directly into / that’s not what you want 2016-04-03 18:24:38 eh, sorry, I’ve overlooked something… 2016-04-03 18:24:42 it should be okay as it is 2016-04-03 18:24:51 :D 2016-04-03 18:24:59 thanks! 2016-04-03 18:25:09 gonna test before submitting 2016-04-03 18:39:14 hey @jirutka, I got the build to work but now I'm getting: make: *** No rule to make target 'install'. Stop. 2016-04-03 18:39:17 any clues? 2016-04-03 18:39:53 your makefile doesn’t have target named install…? 2016-04-03 18:41:39 I didn't write the make file... you see when I run these commands in vanilla shell, everything works... but it's only through the 'abuild -r' that it fails 2016-04-03 18:42:38 hm, maybe try to switch arguments order: make install INSTALL_ROOT="$pkgdir/" 2016-04-03 18:43:03 are you sure that you’re running it in the right directory? 2016-04-03 18:46:54 yes because make compiles everything successfully... It's only after the make is done that it complains ... 2016-04-03 18:47:00 maybe I should just remove the install part 2016-04-04 00:57:45 ncopa: http://patchwork.alpinelinux.org/patch/1843/ 2016-04-04 00:58:09 Couldn't finish it friday I had too leave early for the week end :) 2016-04-04 01:28:50 ncopa: https://github.com/gperftools/gperftools/issues/693 - I know you were playing with this, if I were to somehow statically compile this? Has it been usable for you? 2016-04-04 07:19:05 ncopa, any reasons why fping was not upgraded to 3.13? It was just missing or there are other reasons? 2016-04-04 07:19:32 http://sprunge.us/TMDO 2016-04-04 08:44:43 fcolista: i pushed fping 3.13 now 2016-04-04 09:26:08 hey :) 2016-04-04 09:52:31 hi 2016-04-04 10:54:10 is valery here? 2016-04-04 10:54:56 I don't think so 2016-04-04 12:20:08 we should probably make 'lua' use lua 5.3 2016-04-04 12:32:46 okay with me 2016-04-04 12:53:35 getting dns to run is easy, but getting dns to run in a way it is secure is hard ... 2016-04-04 12:54:04 getting "hidden primary" to run in nsd is hard 2016-04-04 12:55:55 people use DNS in incredibly misguided way though 2016-04-04 12:56:55 a simple, straightforward use of DNS is pretty easy to get to run reliably, but as soon as a company is more than 1000 employees, they feel obligated to stack so many hacks on top of DNS 2016-04-04 12:57:16 that it becomes a complete mess 2016-04-04 12:58:08 (that, and sysadmins generally don't even know the difference between a DNS server and a DNS cache, due to a historical design mistake) 2016-04-04 13:01:35 yeah, in general sysadmins are so fucking missinformed about dns 2016-04-04 13:02:19 one of our customers always blames me when there company internet connetion is bad because i run the dns servers and he still thinks traffic runs over the dns server *g* 2016-04-04 13:02:46 actually those are ex customers ... because one night i explained dns to him an called him a fucking idiot ... 2016-04-04 13:02:51 but that was worth it 2016-04-04 13:03:58 you should have explained you're not even working at the company anymore, and that what you say doesn't necessarily express the views of the company 2016-04-04 13:04:15 but that it's probable that the company also thinks he's a fucking idiot 2016-04-04 13:04:30 (for style points) 2016-04-04 13:06:25 dns should probably be at least two servers, probably three, not all of which need to be running on the same machine 2016-04-04 13:07:08 ammunta: running multiple dns servers on the same machine? that is not very clever in my mind 2016-04-04 13:07:11 ammunta: first thing should be clarifying what "dns" means: name serving or resolution 2016-04-04 13:07:49 ISPs provide resolution service, companies usually serve names, and that's *not* the same thing at all 2016-04-04 13:08:10 but people say "dns servers" in both cases, which generally leads to massive confusion 2016-04-04 13:11:06 well, lets just say there are a lot of people out there who have no idea how dns works and how to set it up properly 2016-04-04 13:11:28 my luck that i am one of the onces who knows how to do it correctly :) 2016-04-04 13:11:51 and thanks to alpine its also very stable and secure 2016-04-04 13:12:52 I wouldn't go so far as to say "secure". DNS isn't secure anyway, unless you invest a lot of energy and convenience to make it so. 2016-04-04 13:14:19 skarnet: i am a big fan of the dns system introduced by gnu-net 2016-04-04 13:14:43 the idea is great and the shadow records are awesome ... the only problem ... the main developer is an idiot .. 2016-04-04 13:15:21 >includes the GNU Name System 2016-04-04 13:15:24 stopped reading there 2016-04-04 13:15:34 skarnet: and those should be separate services, because they do different things 2016-04-04 13:15:55 skarnet: you should read on 2016-04-04 13:16:04 the theory behind it is solid 2016-04-04 13:16:11 just the execution sucks horrible 2016-04-04 13:16:25 ammunta: exactly. Putting them on the same port was a design failure, and problems with DNS can be traced back to Mockapetris and the original RFCs. 2016-04-04 13:17:29 leo-unglaub: "solid theory, terrible execution" is basically included in any project whose name starts with "GNU" 2016-04-04 13:17:44 skarnet: are you sure the first part is included 2016-04-04 13:17:57 most of the time 2016-04-04 13:18:30 (theory doesn't mean software design) 2016-04-04 13:19:13 I wasn't thinking software per se either 2016-04-04 13:23:06 re: security, I can't think of a single threat model that dns doesn't fail amazingly hard 2016-04-04 13:23:24 (nontrivial ones, I mean) 2016-04-04 13:26:27 dns is the sort of protocol you design assuming you're on a network where all the operators are trusted 2016-04-04 13:27:33 probably because it actually was 2016-04-04 13:28:13 yes 2016-04-04 13:28:18 the things we get stuck with 2016-04-04 13:30:49 it takes a good dash of misanthropy to make something that isn't broken 2016-04-04 13:32:22 huh? 2016-04-04 13:33:22 I'd say it takes a certain willingness to disregard history and do the hard work yourself, but that doesn't necessarily pair with misanthropy 2016-04-04 16:16:00 bump zeromq pls, v4.1.4 2016-04-04 17:36:31 could someone please to an nslookup on leo-unglaub.net and post the result 2016-04-04 17:36:33 thanks :) 2016-04-04 17:37:05 SRVFAIL NX DOMAIN 2016-04-04 17:37:27 thx :) 2016-04-04 17:53:00 look ma...everyone are here today! 2016-04-04 17:53:28 hi oneinsect 2016-04-04 17:54:08 ncopa: rpi 3 is lovely only because of alpine 2016-04-04 17:54:11 i am enjoying it 2016-04-04 17:54:15 with wireless ap mode 2016-04-04 17:54:20 cool! 2016-04-04 17:54:41 i couldn't figure out how to get alpine on my rpi 2016-04-04 17:54:48 i wish you supported more boards like orange pi and more recently nanopi m1 2016-04-04 17:54:49 but I was trying to build it myself 2016-04-04 17:55:02 ncopa: 2016-04-04 17:56:30 ncopa: look here 2016-04-04 17:56:31 http://wiki.friendlyarm.com/wiki/index.php/NanoPi_M1 2016-04-04 17:57:06 @ncopa: I have problem with JRuby (I’m writing abuild for it) / openjdk8… it’s trying to load libcrypt.so and I have no clue why… do you have some advice? 2016-04-04 17:59:02 sounds like it either is using something precompiled against glibc 2016-04-04 17:59:38 or is dlopen 2016-04-04 17:59:44 but what? openjdk or jruby? I’ve compiled libjffi myself and it shouldn’t contain any other JNI 2016-04-04 17:59:45 and assume glibc 2016-04-04 17:59:53 no idea 2016-04-04 18:07:57 OMG, they must be kidding me! JRuby folks bundled some stupid .so files directly into JARs! 2016-04-04 18:08:18 not jruby.jar, but some deps 2016-04-04 18:08:22 great, really great… 2016-04-04 18:18:31 someone should tell these BLOB distributors that not everyone runs shitty glibc >_< 2016-04-04 19:48:32 What is even worse than autohells? autohells invoked from maven plugin! 2016-04-05 05:37:03 ncopa: https://gist.github.com/Ttech/d0047b4c7b38e929798ee95d753d00c1 not sure if you want 2016-04-05 05:56:15 currently trying to figure out how to deal with: source 0.14.4.tar.gz needs to be renamed to avoid possible collisions since its triggered before i can modify the source filename 2016-04-05 05:58:41 oh never mind, i see :: 2016-04-05 06:26:24 but general thought was they call it libcaf, but its part of actor-framework, some packages call it by libactorframework and some call it by libcaf, not sure how to do that in abuild 2016-04-05 06:35:00 Hello everyone - Question. Does the package list for apk only receive updated packages when alpine linux releases a new version? 2016-04-05 06:35:54 I'm trying to get the newest Nginx on my docker container, but alpine:latest only installs 1.8.1 so I can't use http2 2016-04-05 06:43:52 Seany_: we normally don't add new features to stable branches 2016-04-05 06:44:09 but you can use latest and greatest from alpine:edge, which is rolling 2016-04-05 06:44:25 re nginx, i think upstream is pushing an official alpine image 2016-04-05 06:45:42 Seany_: i think you can `docker run -it nginx:alpine` 2016-04-05 07:20:02 okay so just for future reference if a package is updated on apk I should just use edge to download it 2016-04-05 09:19:31 hey :) 2016-04-05 09:19:58 how is everyone on this fine fine day? 2016-04-05 09:24:47 im good thanks. how are you leo-unglaub? 2016-04-05 09:25:18 fine, and you? 2016-04-05 09:25:20 glad to hear it :) i am fine as well! 2016-04-05 09:26:00 nmeum: what do you think about being backwards compat? http://tpaste.us/GdQy 2016-04-05 09:44:13 ncopa: looks good to me 2016-04-05 11:42:34 whats the best way to commit a update for a aport that is already there? i need a bitlbee-dev sub package with the header files 2016-04-05 11:42:40 should i email the maintainer? 2016-04-05 11:42:46 i wrote a patch for it 2016-04-05 11:47:12 git format-patch --stdout -1 | tpaste 2016-04-05 11:50:49 http://tpaste.us/24Y4 2016-04-05 12:37:00 xsteadfastx: thanks 2016-04-05 12:37:19 :) 2016-04-05 12:42:11 should i just add the patch? or contact the maintainer? 2016-04-05 12:50:22 xsteadfastx, best practice is to send to alpine-aports mailing list and cc the maintainer 2016-04-05 12:50:45 alpine-aports goes automatically to patchworks.alpinelinux.org and will be processed (in due time) by the core alpine staff 2016-04-05 12:57:33 fabled: did it :) thank you 2016-04-05 12:57:56 xsteadfastx: http://git.alpinelinux.org/cgit/aports/commit/?id=a24ccd7e959ce745e58b5e0c4dcdb90ea864dbc2 2016-04-05 12:58:06 i applied it using the tpaste above 2016-04-05 12:58:40 ncopa: ok... sorry now i have it in patchworks too 2016-04-05 12:59:21 patchwork should notice it's applied and automatically mark it as 'applied' 2016-04-05 12:59:58 ah ok... perfect :) thanks 2016-04-05 13:00:11 now i can submit my package for the bitlbee plugin :) 2016-04-05 13:09:12 @ncopa: could you please merge https://github.com/alpinelinux/aports/pull/42, or disable Travis for aports? it fails on PRs now because of missing .travis.yml 2016-04-05 13:13:29 oh yes 2016-04-05 13:13:32 sorry 2016-04-05 13:15:08 And I’ve finally fixed problems with JRuby, so I’m gonna push jruby abuilds and friends today. 2016-04-05 13:22:52 huh, spam on GitHub? o.O https://github.com/alpinelinux/aports/pull/42#issuecomment-205797892 2016-04-05 13:24:09 testing/apache-ant (https://github.com/alpinelinux/aports/pull/43) and main/musl (https://github.com/alpinelinux/aports/pull/44) are green :) 2016-04-05 13:54:20 holy fuck .. i feel disy ... just migrated 1200 zonefiles fro a shitty webinterface without an api into my system ... 2016-04-05 14:37:12 wow, why are we using travis now? 2016-04-05 14:37:30 nmeum, to filter github pull requests 2016-04-05 14:38:59 what's the advantage of this? we don't filter patchwork requests either, don't we? 2016-04-05 14:39:08 nmeum: see https://travis-ci.org/alpinelinux/aports/builds/120888672 2016-04-05 14:39:35 it tries to build the package so you can easily filter out broken PRs 2016-04-05 14:40:47 click on > Building package main/musl in the log, you can see all the build log 2016-04-05 14:40:50 ncopa: what were your thoughts on the libcaf vs actor-framework-lib? 2016-04-05 14:41:05 but so far github pull requests have never been the „standard way” of submitting patches 2016-04-05 14:41:19 and what? 2016-04-05 14:41:40 nmeum, we don't enforce github pull requests. but do accept them. 2016-04-05 14:42:24 it’s the supported way and thanks to CI it’s easier to check if the patches are okay 2016-04-05 14:45:47 I am not against using a CI system per se I just dislike the fact that we rely on a travis instance hosted by someone else and that we are only using it for github pull requests and not for the patches submitted over the ML 2016-04-05 14:46:21 well, I think that you’re welcome to setup CI for patches send via email or whatever… ;) 2016-04-05 14:49:16 I set Travis for GitHub pull requests; it’s not perfect, but it works very well and it works now 2016-04-05 14:52:37 nmeum: as usually, 1) someone had time to do the job 2) travis support seems more trivial than setting up CI for mailing list 2016-04-05 14:53:07 we still have to test changes manually for patches from irc and aports-devel, so at least we have one burden less 2016-04-05 14:55:22 as I said: I am fine with it: I was mostly just suprised that the repos suddendly contained a .travis.yml 2016-04-05 14:56:32 okay :) 2016-04-05 16:12:23 nmeum: i am not too happy with having a specific service config in the repo 2016-04-05 16:12:39 but even if its ugly, it does not really hurt 2016-04-05 16:12:46 and it will hopefully do good 2016-04-05 16:54:09 ncopa: sorry to come up with this topic one more time, but: Python 3. do you see a feasible way to allow the introduction of library packages into Alpine soon-ish or will Alpine be unable to provide them (and, thus, ease the upgrade to Python 3 for both users and developers) for the foreseeable future? 2016-04-05 18:48:15 I would suggest to move this discussion about python3 to the mailing list. What do you think? 2016-04-05 20:31:43 Y0Gi: you are right. we someone to kick us in the rear part to get python3 fixed 2016-04-05 22:09:45 ncopa: could you please consider accepting %1719 and %1718? 2016-04-05 22:52:53 ncopa: still here? 2016-04-06 05:50:25 ncopa, seems qt5-qtbase-dev should depend on all the sub packages 2016-04-06 05:50:29 now i'm getting failure like: 2016-04-06 05:50:42 CMake Error at /usr/lib/cmake/Qt5Sql/Qt5SqlConfig.cmake:27 (message): 2016-04-06 05:50:42 The imported target "Qt5::Sql" references the file 2016-04-06 05:50:42 "/usr/lib/qt5/plugins/sqldrivers/libqsqlodbc.so" 2016-04-06 05:50:42 but this file does not exist. Possible reasons include: 2016-04-06 05:50:43 ... 2016-04-06 05:50:51 $ apk info -W /usr/lib/cmake/Qt5Sql/Qt5Sql_QODBCDriverPlugin.cmake 2016-04-06 05:50:51 /usr/lib/cmake/Qt5Sql/Qt5Sql_QODBCDriverPlugin.cmake is owned by qt5-qtbase-dev-5.6.0-r1 2016-04-06 05:50:56 .. 2016-04-06 05:51:08 but the -odbc package is not pulled in when depending on the -dev 2016-04-06 06:47:45 why if abuild find this file: /usr/lib/charset.alias package() fails? 2016-04-06 07:08:19 fc1979, it's incorrectly generated by gnulib 2016-04-06 07:08:31 for glibc it's not generated, but it detects glibc only 2016-04-06 07:08:49 if all packages would ship that file the package manager would not like it when same file comes from many packages 2016-04-06 07:09:00 with different content, and it cannot merge it 2016-04-06 07:09:11 so it was deemed simples to workaround the gnulib bug just be deleting that file manually 2016-04-06 07:09:16 so i should remove it? 2016-04-06 07:09:22 ok 2016-04-06 07:10:12 yes, that's what all other packages do too 2016-04-06 07:10:18 you can grep aports tree for charset.alias 2016-04-06 07:11:11 sounds good. Thx fabled 2016-04-06 07:48:38 morning 2016-04-06 07:51:40 fabled: 2016-04-06 07:52:10 what is the current U-Boot release? is it based on upstream? 2016-04-06 07:54:13 all python modules seems to be built for python2, would that be right? 2016-04-06 09:23:39 hey :) 2016-04-06 09:50:21 fabled, looks that i've a conclict of a struct definition here: 2016-04-06 09:50:21 In file included from /usr/include/net/ethernet.h:10:0, 2016-04-06 09:50:22 from iouyap.c:38: 2016-04-06 09:50:22 /usr/include/netinet/if_ether.h:96:8: error: redefinition of 'struct ethhdr' 2016-04-06 09:50:22 struct ethhdr { 2016-04-06 09:50:22 ^ 2016-04-06 09:50:24 In file included from /usr/include/linux/if_tun.h:20:0, 2016-04-06 09:50:26 from iouyap.c:35: 2016-04-06 09:50:28 /usr/include/linux/if_ether.h:139:8: note: originally defined here 2016-04-06 09:50:30 struct ethhdr { 2016-04-06 09:51:02 i remember this was solved somehow with a patch 2016-04-06 09:51:20 aha 2016-04-06 09:51:21 ok 2016-04-06 09:51:44 sipp/musl-fixes.patch 2016-04-06 09:53:02 no 2016-04-06 09:55:05 Do not include netinet/if_ether.h for musl to prevent struct ethhdr 2016-04-06 09:55:28 ok 2016-04-06 09:57:09 it's the eternal uapi problem :/ 2016-04-06 09:57:17 #if HAVE_NETINET_IF_ETHER_H to #if HAVE_NETINET_IF_ETHER_H && !HAVE_LINUX_IF_TUN_H 2016-04-06 09:57:40 i'm going to include the if_ether only if ^ 2016-04-06 10:02:29 skarnet; 2016-04-06 10:02:30 http://sprunge.us/ZNVV 2016-04-06 10:03:05 "when in doubt, include sys/types.h" 2016-04-06 10:03:13 :D 2016-04-06 10:03:29 do you think would be good to send it to upstream ? 2016-04-06 10:03:33 yes 2016-04-06 10:03:36 To make it compatible with musl? 2016-04-06 10:03:37 Ok 2016-04-06 10:03:56 I guess the point is to make sure libc definitions are included before kernel headers are called? 2016-04-06 10:04:40 So it's matter of order? 2016-04-06 10:04:45 mostly, yes 2016-04-06 10:04:46 but 2016-04-06 10:05:00 I understand the iouyap.c changes, but why do you have to include sys/types.h in the other files? 2016-04-06 10:05:09 does omitting it cause an error? 2016-04-06 10:05:12 yes 2016-04-06 10:05:24 u_long, u_char and other are not found 2016-04-06 10:05:32 ah, BSD types 2016-04-06 10:05:43 should define _BSD_SOURCE somewhere 2016-04-06 10:06:22 there's no anywhere..so i should add that define and include sys/types.h 2016-04-06 10:06:37 or add -D_BSD_SOURCE to the compilation options 2016-04-06 10:06:45 and yes, including sys/types.h never hurts 2016-04-06 10:07:04 it's been a solution to 80% of portability problems for years :D 2016-04-06 11:33:40 ncopa: have you gotten the link to the mirror document 2016-04-06 12:01:49 ncopa: do we still want to replace main/php with testing/php7? If so: We should do that soon before more people start submitting php7-* packages 2016-04-06 12:02:52 we have issue about that too 2016-04-06 12:03:07 http://bugs.alpinelinux.org/issues/5367 2016-04-06 12:04:34 oh good to know 2016-04-06 12:09:23 please dont asume everyone who is using php wants to use php7 2016-04-06 12:09:42 because php7 is the first php release in a loooooong time that really breaks stuff 2016-04-06 12:09:57 and people would have a rude awakening when half there website would not run anymore 2016-04-06 12:10:38 for #5367 i think the best solution is a simple alias in his sh config 2016-04-06 12:13:03 but maintaining two php version and php packages for both version (e.g php-redis, php7-redis) sucks imho. Arch Linux also simply upgraded their php package without providing a php 5 package, don't know if they ran into any major problems though 2016-04-06 12:13:42 i think that was our will too - upgrade the existing php packate to 7.x 2016-04-06 12:13:52 but some php users were hesitant apparently 2016-04-06 12:17:50 sadly i use php a lot and i can say upgrading from php5 to php7 is not as easy as prople think 2016-04-06 12:17:57 a lot of scripts have to be updated 2016-04-06 12:18:03 of course no problem for me 2016-04-06 12:18:19 but when i keep in mind how much old crap is still running as ph pscripts ... 2016-04-06 12:18:26 that could cause a lot of problems 2016-04-06 12:18:34 even more than a openssl -> libressl switch 2016-04-06 12:19:54 I think it would be benefical if (old) PHP crap will stop running… 2016-04-06 12:20:09 hehe, indeed 2016-04-06 12:20:17 but thats a very hard cut 2016-04-06 12:20:56 Alpine doesn’t support shitty things like glibc, so why it should support PHP? ;) :P 2016-04-06 12:21:43 leo-unglaub: problems might always arise when upgrading software but never changing a running system or never upgrading software isn't really an option, is it? People will have to upgrade to php7 sooner or later anyway and this change won't probably effect most users until the release of 3.4 2016-04-06 12:27:00 well, thats an argument. on the other site php5 is still supported 2016-04-06 13:15:05 i'd like to support both php5 and php7 2016-04-06 13:15:14 its an interesting case 2016-04-06 13:15:30 we have same issue with python3 2016-04-06 13:15:58 I don’t think it’s the same situation 2016-04-06 13:16:17 maybe not the same, but similar at least 2016-04-06 13:16:54 there are some factors we need to consider 2016-04-06 13:16:59 and Lua 5.1/2/3 2016-04-06 13:17:00 in the case of Python, we’ll probably end up with foo as metapackage with subpackages py-foo and py3-foo 2016-04-06 13:17:17 and lua too yes 2016-04-06 13:17:57 - we want encourage people to upgrade their code (eg we want encourage use of php7 instead of php5) 2016-04-06 13:17:58 eh, py-foo, py2-foo and py3-foo 2016-04-06 13:18:37 what do you think? is there any option other than making subpackages for py2 and py3? 2016-04-06 13:18:47 - we want try avoid unecessary breakage for upgrades 2016-04-06 13:19:03 - we want avoid too much maintenance work for us 2016-04-06 13:20:02 the point 2 also involves make upgrades smooth 2016-04-06 13:21:04 for minimal impact on upgrades, we can keep the php name (and python) and introduce python3 and php7 (like we have now) 2016-04-06 13:21:22 however, that will not encourage people to upgrade 2016-04-06 13:21:31 there’s no problem with package for interpreter itself, but for libraries… 2016-04-06 13:21:38 including libs 2016-04-06 13:21:50 so keep py-* and introduce py3-* 2016-04-06 13:22:07 that will have minimal impact on upgrades 2016-04-06 13:22:22 we did something similar for qt4 -> qt5 2016-04-06 13:22:25 well, but what if I don’t care about specific slot? we need also some metapackage 2016-04-06 13:22:50 right, the problem is that if you just want php/python you'll get the old 2016-04-06 13:23:05 so it does not encourage people to upgrade 2016-04-06 13:23:17 and we may get stuck with the old longer than wanted 2016-04-06 13:23:33 so I was thinking about making py-* just a metapackage and introducing py2-* for actual Python 2 package that will `provides=py-*` 2016-04-06 13:23:51 similar to what we do with lua i suppose 2016-04-06 13:23:59 i kinda like what we do with lua 2016-04-06 13:24:15 but its more work on us 2016-04-06 13:24:33 to be honest I still haven’t look how are lua packages done 2016-04-06 13:24:35 ACTION too, but small install base and small number of packages 2016-04-06 13:24:45 makes Lua easier 2016-04-06 13:24:59 is it one APKBUILD with meta package and subpackages for each slot? 2016-04-06 13:25:08 the other side of the scale, we could simply drop the old python2/php5 and only provide python3/php7 2016-04-06 13:25:16 that would "encourage upgrade" 2016-04-06 13:25:25 but will break things for sure 2016-04-06 13:25:32 so i dont think that is realistic 2016-04-06 13:25:44 and "encourage rage" :) 2016-04-06 13:25:48 :D 2016-04-06 13:25:50 yeah 2016-04-06 13:25:55 this would be great, but then we will need to port some legacy python2-only packages :( 2016-04-06 13:26:10 i dont think we want do that 2016-04-06 13:26:18 i kinda like the way arch does it 2016-04-06 13:26:33 I like the way Gentoo does it… don’t know how it’s in Arch 2016-04-06 13:26:50 php upgrades to 7.x and introduce php5 2016-04-06 13:26:54 gentoo has slots 2016-04-06 13:27:00 Gentoo makes it easy even for package mantainers 2016-04-06 13:28:01 i think what i want do do is 2016-04-06 13:28:09 rename php to php5 2016-04-06 13:28:17 and keep php7 2016-04-06 13:28:26 so we have php5 and php7 2016-04-06 13:28:35 and meta package for "php" 2016-04-06 13:28:40 which pulls in the latest 2016-04-06 13:28:59 and both php5 and php7 will `provides=php`? 2016-04-06 13:29:14 not necessarily 2016-04-06 13:29:41 if you have something that depends on php 2016-04-06 13:29:48 you'll have to chose 2016-04-06 13:30:01 does it work with any php? if so then depends=php 2016-04-06 13:30:09 and you will effectively depend on newest 2016-04-06 13:30:28 if it requires specific php, then you'll have to depends=php5 2016-04-06 13:30:53 do we have many php-* packages? 2016-04-06 13:31:02 are there potensially many php-* packages? 2016-04-06 13:31:09 i know we have a ton of py-* 2016-04-06 13:31:14 what if you have php5 installed becauouse you need it for some legacy crap and wants to install some package that depends on php? then you will end up with both php5 and php7 installed even when you don’t need them both 2016-04-06 13:31:42 correct 2016-04-06 13:31:52 there’s actually none php-* package https://pkgs.alpinelinux.org/packages?name=php-*&repo=all&arch=x86_64&maintainer=all 2016-04-06 13:32:39 is there potensial to be many? 2016-04-06 13:32:47 i have seen a bunch of php7-* patches coming in 2016-04-06 13:33:07 I hope not… but just because I don’t accept PHP :) 2016-04-06 13:33:11 :) 2016-04-06 13:33:24 the same problem we have with openjdk… some old packages depends on openjdk7-jre 2016-04-06 13:33:32 we can easily solve this using provides=openjdk 2016-04-06 13:33:46 yes, but i am a bit sceptic 2016-04-06 13:34:10 fabled: what do you think about adding provides=openjdk to both openjdk7 and openjdk8? 2016-04-06 13:34:12 actually I’m planning to do that later 2016-04-06 13:34:25 ncopa, i guess that would be ok 2016-04-06 13:34:29 and adding provides=python to both python2 and python3? 2016-04-06 13:34:41 and provides=php to both php5 and php7? 2016-04-06 13:34:46 it’s true that openjdk7 is not supported anymore, but what we’ll be prepared for openjdk9 2016-04-06 13:35:23 fabled do we add provides=lua to lua5.[123] too? 2016-04-06 13:35:25 it’s easy with openjdk, the problem with python is that packages are installed into a different location 2016-04-06 13:35:46 jirutka: re lua 2016-04-06 13:35:50 apk does not allow two lua=1.2 stuff installed 2016-04-06 13:35:57 so it would need to be versionless provides 2016-04-06 13:35:59 right 2016-04-06 13:36:00 those can coexist 2016-04-06 13:36:54 for lua modules we do: 2016-04-06 13:37:06 lua-foo is empty 2016-04-06 13:37:12 meta package 2016-04-06 13:37:31 lua5.1-foo has install_if="lua-foo lua5.1" 2016-04-06 13:37:47 which means that lua5.1-foo gets pulled in if you have both lua5.1 and lua-foo installed 2016-04-06 13:38:02 if you apk del lua5.1 then will lua5.1-foo also get deleted 2016-04-06 13:38:10 it works pretty well i think 2016-04-06 13:38:28 it sounds good :) 2016-04-06 13:38:39 except that things get messy when you also need support luajit 2016-04-06 13:39:11 this means that you can use an explcit lua version: lua5.1-foo 2016-04-06 13:39:30 or you can use lua-foo and get all the lua version you currently have installed 2016-04-06 13:40:14 the only downside i can see with it is that we provide full compat for older stuff that we maybe don't really need 2016-04-06 13:40:27 some packages dont need the lua5.1 support 2016-04-06 13:40:31 but get it anyway 2016-04-06 13:40:59 and the APKBUILDs are a bit complicated 2016-04-06 13:41:24 i think we may wnat do something similar with python 2016-04-06 13:41:32 with php i dont know 2016-04-06 13:41:39 with php we probably want do it differently 2016-04-06 13:42:24 i mean 2016-04-06 13:42:43 we probably dont want support php5 and php7 from same 3rd party php-* module 2016-04-06 13:42:53 i need to go out for a while 2016-04-06 13:42:55 see u 2016-04-06 13:46:15 hello 2016-04-06 13:46:22 any one here 2016-04-06 13:59:59 anyone still alive 2016-04-06 14:00:26 any aliens attacked our bases? all so silent? 2016-04-06 14:00:28 :-D 2016-04-06 14:04:43 oneinsect: pro-tip: ask a specific, preferably relevant, and non-meta question :) 2016-04-06 14:04:44 hi, we’re alive ;) 2016-04-06 14:06:15 can anyone guide me on howto use an armbian kernel and uboot but alpine userspace on Orange Pi One? 2016-04-06 14:42:39 fabled: 2016-04-06 14:42:42 ncopa: 2016-04-06 14:42:49 the problem keep persisting 2016-04-06 14:42:58 gst-plugins-ugly1 2016-04-06 14:43:04 is not available in edge 2016-04-06 14:43:12 http://dl-2.alpinelinux.org/alpine/edge/main/armhf/ 2016-04-06 14:43:16 strange 2016-04-06 14:43:20 why it keeps happening 2016-04-06 14:53:38 ncopa, do we have an example of an APKBUILD python package valdi for both 2 and 3 ? 2016-04-06 14:53:39 *valid 2016-04-06 14:53:56 not yet, as I know 2016-04-06 14:55:16 so i need to create another APKBUILD for py3-$package 2016-04-06 14:55:56 please wait until ncopa decide how to handle it 2016-04-06 14:56:07 yes sure.. 2016-04-06 14:56:18 the idea is do it in the same way as lua packages 2016-04-06 14:56:35 yes 2016-04-06 14:56:40 one APKBUILD with subpackages for py2 and py3 2016-04-06 14:56:45 i dislike the idea of duplicate things 2016-04-06 14:56:56 _pyversions="2 3" 2016-04-06 16:10:51 fcolista: i dont like duplicate either 2016-04-06 16:11:04 so _pyversions="2 3" sounds like the way to go 2016-04-06 16:13:19 does APK have something like eclass in Portage? 2016-04-06 16:18:36 ncopa: 2016-04-06 16:44:19 jirutka: no, we have default functions that could probably do something similar 2016-04-06 16:44:31 but we wanted to avoid it in past because all the magic involved 2016-04-06 17:08:21 ncopa: any thoughts of releasing Curl from clutches of Al's toolchain, so new features can be tested and used ? 2016-04-06 18:04:59 I am trying to install perl module: Number-Format-1.75.tar.gz and it fails the tests for t/locale.t 2016-04-06 18:08:09 http://paste.fedoraproject.org/350547/14599660/ 2016-04-06 18:09:40 http://paste.fedoraproject.org/350548/45996617/ <- t/locale.t 2016-04-06 18:28:14 since locale is glibc's, the test should be skipped ? 2016-04-06 18:53:53 https://github.com/Entware/openwrt-packages/commit/cecee995b0ed69132b230fa605b99a3cafa1e5b7 <- 2016-04-06 18:56:57 perl-5.22.1-r0 -> $ perl -V:d_setlocale -> d_setlocale='define'; 2016-04-06 19:00:43 why are there four copies of the same patch on the ML 2016-04-06 19:13:30 Which one? 2016-04-06 19:44:15 friends 2016-04-06 19:44:19 anyone knows how to run 2016-04-06 19:44:25 alpine linux in Intel Edison 2016-04-06 19:44:30 which is a x86 2016-04-06 19:44:32 board 2016-04-06 19:45:01 and what’s the problem? 2016-04-06 19:45:50 it requires a custom uboot to work 2016-04-06 19:46:00 and requiring flashing the mmc 2016-04-06 19:46:07 wonder howto go about it 2016-04-06 19:46:36 https://github.com/01org/edison-u-boot 2016-04-06 23:09:55 ncopa: you can configure Travis to build only if .travis.yml is present, then it’ll not fail on e.g. 3.x-stable branches (https://travis-ci.org/alpinelinux/aports/settings) 2016-04-06 23:16:45 there also also few commits pushed to master that failed to build on travis… have they passed on the build server? 2016-04-06 23:21:41 hm, the alpine mirror used on travis is synced every hour, but at 7:24 AM one build failed because of missing rabbitmq-c that was build at 5:22 AM (according to pkgs.a.o)… but I’m not sure if these times are in the same timezone 2016-04-06 23:23:24 what mirror is most up-to-date and on a fast network? I’ve tried dl2 once but it was very slow so I’ve switched to my mirror 2016-04-07 10:49:17 does Valery hang out here? 2016-04-07 11:49:29 hey :( 2016-04-07 11:49:37 wrong smilie ... i ment :) 2016-04-07 11:50:07 ;-) hey 2016-04-07 12:13:10 fabled: 2016-04-07 12:13:13 ncopa: 2016-04-07 12:13:45 generally we are advised to format the sdcard in fat32 and toggle the boot flag and copy the rpi alpine distro files and it boots 2016-04-07 12:13:47 why do i need to type a password for useradd from https://pkgs.alpinelinux.org/package/testing/x86/shadow? 2016-04-07 12:13:54 even if i'm root 2016-04-07 12:14:24 or even on groupadd oO 2016-04-07 12:14:59 is this process specific to rpi or can any u-boot/script.bin made to boot that way on any board? 2016-04-07 12:15:10 cd .. 2016-04-07 12:15:46 i have starte to write ansible playbooks for my server and it hangs because it's waiting for a password in the background :( 2016-04-07 12:18:30 mosez: why not use addgroup / adduser 2016-04-07 12:20:31 xen_roger: http://docs.ansible.com/ansible/user_module.html#requirements-on-host-that-executes-module 2016-04-07 12:21:36 beside that groupadd and useradd provide more options :) 2016-04-07 12:24:29 according with this: 2016-04-07 12:24:29 https://www.python.org/dev/peps/pep-0453/#recommendations-for-downstream-distributors 2016-04-07 12:24:44 looks that we need to redefine out py-* builds by using pip 2016-04-07 12:24:51 *our 2016-04-07 12:25:16 probably, an helper like apkbuild-cpan for py it's needed 2016-04-07 12:25:40 mosez: tested shadow (4.2.1-r3) - groupadd foobar - no password promt. is ansible running with root permission ? 2016-04-07 12:27:00 useradd also works without password request 2016-04-07 12:27:30 xen_roger: http://imgur.com/QwHaO0N as root 2016-04-07 12:27:31 mosez: i have it on my list to fix useradd/groupadd: http://lists.busybox.net/pipermail/busybox/2016-February/083907.html 2016-04-07 12:30:41 mosez, ncopa: http://pix.toile-libre.org/upload/original/1460032178.png 2016-04-07 12:31:07 ncopa: btw: I don't know if it is just me but since you removed ping from bbsuid (there by removing the suid bit) it no longer works without root rights 2016-04-07 12:31:24 ah 2016-04-07 12:31:28 yes 2016-04-07 12:31:34 is that intented? 2016-04-07 12:31:36 *intended 2016-04-07 12:32:40 yes, you need to set /proc/sys/net/ipv4/ping_group_range 2016-04-07 12:32:57 http://man7.org/linux/man-pages/man7/icmp.7.html 2016-04-07 12:33:08 search for ping_group_range 2016-04-07 12:33:12 mosez: user and group are created without password requests ... http://pix.toile-libre.org/upload/original/1460032302.png 2016-04-07 12:33:44 ncopa: ah, thanks 2016-04-07 12:34:07 it means we can do ping without suid root 2016-04-07 12:34:18 yeah I get that I just didn't know about that sysctl 2016-04-07 12:34:18 which is great :) 2016-04-07 12:34:28 indeed 2016-04-07 12:34:29 we need to doc it 2016-04-07 12:34:31 that's strange oO 2016-04-07 12:35:07 uid=0(root) gid=0(root) groups=0(root),0(root),1(bin),2(daemon),3(sys),4(adm),6(disk),10(wheel),11(floppy),20(dialout),26(tape),27(video) 2016-04-07 12:35:13 i'm definitly root 2016-04-07 12:35:18 oneinsect, that's rpi specific. rpi has custom bootloader. we could add u-boot to rpi boot sequence but i'm not sure how finalized u-boot's support for rpi is 2016-04-07 12:35:56 mosez: check `which useradd` and verify that it is not the busybox symlink 2016-04-07 12:36:20 262437 -rwxr-xr-x 1 root root 119420 Sep 7 2015 /usr/sbin/useradd 2016-04-07 12:36:27 ncopa: maybe we should simply change its default value? 2016-04-07 12:41:52 mosez: https://nopaste.me/view/raw/b0df8e99 could you verify these steps ? 2016-04-07 12:42:25 fabled: one quick question ....can i use kernel (uImage) and script.bin and somehow link the alpine modloop userspace image file to boot alpine in orange pi 2016-04-07 12:43:20 can i fat32 format remain for the complete sdcard since uImage and script.bin are anyway vfat 2016-04-07 12:43:34 possible? with just a bootflag? as we do for rpi? 2016-04-07 12:50:40 modloop is kernel specific. 2016-04-07 12:57:57 xen_roger: https://nopaste.me/view/raw/abfbba76 2016-04-07 12:59:37 lol, this must be related to the scaleway alpine image 2016-04-07 12:59:46 my docker container also doesnt require any passworf 2016-04-07 12:59:54 s/passworf/password/ 2016-04-07 13:00:01 that was my next guess ... 2016-04-07 13:00:18 https://github.com/scaleway/image-alpine this seems to be the image they provide 2016-04-07 13:00:23 maybe your installation / images tainted 2016-04-07 13:00:53 so... just build your own ;) always a good idea 2016-04-07 13:06:17 yeah always a good idea to maintain own images -.- 2016-04-07 13:08:48 not to maintain ... but as you see it is different from what the official images does ... so build it yourself and find out what´s the issue and feel free to report if you think it should be fixed. 2016-04-07 13:11:05 i can't even find anything evil on this damn official image 2016-04-07 13:11:48 and i need a bunch of custom stuff to make it running on the scaleway datacenter as you can see on the github repo 2016-04-07 13:17:17 mosez: I checked the repo out, build that image, and there is also no password promt while creating group or user via shadow froim testing 2016-04-07 13:19:43 than i have no idea where this crap comes from 2016-04-07 13:20:02 did you try to build that image your own ? 2016-04-07 13:20:15 no 2016-04-07 13:35:28 damn.... i'm not allowed to start a second arm machine -.- 2016-04-07 13:35:59 hopefully the x86_64 machine got the same issue 2016-04-07 14:52:19 fabled: 2016-04-07 14:52:28 then how do i use the generic arm image 2016-04-07 14:52:43 i now have a kernel (uImage) and script.bin etc 2016-04-07 14:52:48 specific for that board 2016-04-07 15:10:36 guys we can compile the latest bleeding edge kernel in alpine? 2016-04-07 15:10:40 say 4.6 RC2 2016-04-07 15:10:42 is possible? 2016-04-07 15:11:03 why it may not be possible? 2016-04-07 15:34:12 just trying to find instructions to do it 2016-04-07 15:34:49 do you need hardened kernel or just plain vanilla? 2016-04-07 15:36:33 https://github.com/scaleway/image-alpine/issues/38 opened an issue for the scaleway alpine image now -.- 2016-04-07 15:36:55 plain vanilla is fine 2016-04-07 15:37:00 i dont need hardened 2016-04-07 15:37:11 then you can just download sources and compile it 2016-04-07 15:37:26 make menuconfig, make && make install 2016-04-07 15:37:34 alpine? 2016-04-07 15:37:40 it doesn’t matter 2016-04-07 15:37:51 we’re talking about kernel 2016-04-07 15:38:05 well how do it to boot with alpine userspace? 2016-04-07 15:38:32 like with any other? I really don’t understand… 2016-04-07 15:38:42 hmmm 2016-04-07 17:15:27 fabled: 2016-04-07 17:38:25 you should really stop pinging people around and ask your questions instead 2016-04-07 17:38:39 if anyone is available and knows, they will respond 2016-04-07 17:48:18 okie sure 2016-04-08 08:36:58 can somebody explain why my ca-certificates file is empty? 2016-04-08 08:37:06 i'm even not able to clone from git :( 2016-04-08 08:37:37 ca-certificates file? 2016-04-08 08:37:53 /etc/ssl/certs/ca-certificates.crt 2016-04-08 08:38:09 fatal: unable to access 'https://github.com/andsens/homeshick.git/': error setting certificate verify locations: 2016-04-08 08:38:13 CAfile: /etc/ssl/certs/ca-certificates.crt 2016-04-08 08:38:13 CApath: none 2016-04-08 08:38:31 do you have the ca-certificates package installed? 2016-04-08 08:38:40 try apk fix ca-certificates 2016-04-08 08:39:39 # apk fix ca-certificates 2016-04-08 08:39:39 (1/1) [APK unavailable, skipped] Reinstalling ca-certificates (20160104-r2) 2016-04-08 08:39:51 still empty 2016-04-08 08:39:58 it was not fixed 2016-04-08 08:40:15 it was skipped because its unavailable 2016-04-08 08:41:00 the ca-certificates has a trigger script that will run `update-ca-certificates` 2016-04-08 08:41:14 which should generate the /etc/ssl/certs/ca-certificates.crt 2016-04-08 08:41:54 i dont know how you have set up your repos or if you did apk upgrade 2016-04-08 08:42:04 but why is it unavailable? i have updated my index which points to http://nl.alpinelinux.org/alpine/edge/main 2016-04-08 08:42:12 i dont know 2016-04-08 08:42:30 what do you have in your /etc/apk/repositories? 2016-04-08 08:42:35 only edge/main? 2016-04-08 08:42:45 did you apk upgrade -U? 2016-04-08 08:42:49 http://nl.alpinelinux.org/alpine/edge/main 2016-04-08 08:42:49 http://nl.alpinelinux.org/alpine/edge/community 2016-04-08 08:42:49 @testing http://nl.alpinelinux.org/alpine/edge/testing 2016-04-08 08:42:49 maybe apk upgrade -U -a will help 2016-04-08 08:43:02 i think you can remove that last line 2016-04-08 08:43:07 eh 2016-04-08 08:43:07 no 2016-04-08 08:43:16 keep it i think 2016-04-08 08:43:27 what happens if you do apk upgrade -U -a --simulate 2016-04-08 08:43:33 Executing ca-certificates-20160104-r2.trigger 2016-04-08 08:43:36 but still empty 2016-04-08 08:44:20 damn arm machine -.- never had those issues within my docker containers 2016-04-08 08:44:37 and if you do: update-ca-certificates --fresh 2016-04-08 08:45:12 a bunch of "Warning! Cannot copy to bundle: /usr/share/ca-certificates/mozilla/ACCVRAIZ1.crt" lines 2016-04-08 08:47:38 if (!copyfile(fullpath, tmpfile_fd)) 2016-04-08 08:47:38 fprintf(stderr, "Warning! Cannot copy to bundle: %s\n", fullpath); 2016-04-08 08:48:07 i bet its send file that fails 2016-04-08 08:48:13 sendfile 2016-04-08 08:48:16 can you strace it? 2016-04-08 08:48:53 apk add tpaste && update-ca-certificates 2>&1 | tpaste 2016-04-08 08:49:13 hum 2016-04-08 08:49:14 no 2016-04-08 08:49:25 its not sendfile 2016-04-08 08:49:34 oh 2016-04-08 08:49:39 it could be 2016-04-08 08:49:57 its on scaleway server right? 2016-04-08 08:50:28 sorry 2016-04-08 08:50:33 i need the strace 2016-04-08 08:50:40 apk add tpaste && update-ca-certificates 2>&1 | strace | tpaste 2016-04-08 08:50:45 sendfile64(3, 5, [0], 2772) = -1 EINVAL (Invalid argument) 2016-04-08 08:51:02 right 2016-04-08 08:51:08 thats what i'm interested in 2016-04-08 08:51:10 i will paste a full trace if you need it 2016-04-08 08:51:13 no 2016-04-08 08:51:19 i think thats enough 2016-04-08 08:51:36 i wonder if its related their kernel config 2016-04-08 08:51:40 whats the uname -a 2016-04-08 08:52:07 Linux heimdall 3.2.34-30 #17 SMP Mon Apr 13 15:53:45 UTC 2015 armv7l Linux 2016-04-08 08:52:34 and yes, it's a scaleway server 2016-04-08 08:52:34 morning, and happy Friday! 2016-04-08 08:52:52 and i already opened an issue related to password prompt on groupadd/useradd 2016-04-08 08:53:30 unrelated, but i'm struggling to start a scaleway C1 this morning 2016-04-08 08:55:22 scrumpyjack: looks like there are some issues on c1 :) 2016-04-08 08:55:58 i managed to spin one up, i have an IP and storage, but it won't start 2016-04-08 08:56:12 scrumpyjack: start it 2 or 3 times, than it works :D 2016-04-08 08:56:15 they must be out of threads 2016-04-08 08:56:26 yeah, been trying all morngin 2016-04-08 08:56:31 morning even 2016-04-08 08:56:52 oh... yesterday i tried to start it 2 or 3 times, than i had been booted correctly. 2016-04-08 08:57:02 mosez: we could take the discussion to #scaleway 2016-04-08 08:57:03 it* 2016-04-08 08:57:10 ncopa: sure 2016-04-08 08:58:00 ncopa: on irc.dedibox.fr? :) 2016-04-08 08:58:19 Connecting to irc.dedibox.fr (62.210.156.11) port 6667... 2016-04-08 08:58:28 seems so yes :) 2016-04-08 10:15:08 cound someone take a look at patchwork? there are 63 patches in the queue 2016-04-08 10:16:12 woohoo 2016-04-08 10:16:22 and also to GitHub please :) 2016-04-08 10:18:42 and make me sandwich 2016-04-08 10:19:07 np, but how to deliver it to you? 2016-04-08 10:19:07 ScrumpyJack: I can look at them but I fear it won't really help :P 2016-04-08 10:19:37 heh. it's Friday, everyone's a comedian :) 2016-04-08 10:20:03 oh no I'm a comedian all week long 2016-04-08 10:20:12 lul 2016-04-08 10:37:45 kut 2016-04-08 10:37:50 oops 2016-04-08 11:01:24 nmeum: i think i need to revert ca661f064fe894af1b2f36466953d8932c770f9b 2016-04-08 11:02:44 libreoffice fails to build 2016-04-08 11:03:52 checking for MDDS... no 2016-04-08 11:03:52 configure: error: Package requirements (mdds >= 0.12.0) were not met: 2016-04-08 11:04:13 wb mosez :) 2016-04-08 11:04:23 ty :) 2016-04-08 11:05:13 apk update; apk upgrade; update-ca-certificates --fresh still prints a lot of Cannot copy to bundle 2016-04-08 11:05:49 which ca-certificates version did you get? 2016-04-08 11:06:08 still ca-certificates-20160104-r2 2016-04-08 11:06:34 you need -r3 2016-04-08 11:06:56 you can use dl-cdn.alpinelinux.org as mirror 2016-04-08 11:06:58 http://nl.alpinelinux.org/alpine/edge/main/armhf/ no -r3 2016-04-08 11:07:04 sure 2016-04-08 11:08:27 there is also only -r2 :) 2016-04-08 11:08:29 http://dl-cdn.alpinelinux.org/alpine/edge/main/armhf/ 2016-04-08 11:09:32 oh 2016-04-08 11:09:42 arm builder is not done building mesa 2016-04-08 11:10:35 mosez: sorry, we need to wait for the arm builder to catch up 2016-04-08 11:10:48 sure 2016-04-08 11:10:53 irssi is running :) 2016-04-08 11:11:04 got to fix my caddy now... don't know why it fails to start 2016-04-08 11:24:25 ah, damn... let's encrypt fails because of the missing certs -.- 2016-04-08 12:29:35 ncopa: there’s another failed commit on master, testing/hvtools – did it passed on the build server? 2016-04-08 12:29:35 https://travis-ci.org/alpinelinux/aports/builds/121475426 2016-04-08 12:32:28 hm, I see now, it succeeded on the build server http://build.alpinelinux.org/buildlogs/build-edge-x86_64/testing/hvtools/hvtools-4.4.6-r0.log 2016-04-08 12:32:28 what’s difference between these environments? 2016-04-08 12:33:09 ncopa: hm…strange. How do other distros handle this? Not being able to upgrade mdds due to libreoffice kind of sucks 2016-04-08 12:34:24 I’ve just tried to build hvtools on my machine, inside LXC container, and it failed in the same way as on Travis… how’s that possible that it haven’t failed on the build server? 2016-04-08 12:34:52 jirutka: what error do you get? 2016-04-08 12:35:00 you may need to apk upgrade -U -a 2016-04-08 12:35:11 ncopa: archlinux seems to has a fix for it, maybe we want to use that too instead? https://projects.archlinux.org/svntogit/packages.git/tree/trunk/PKGBUILD?h=packages/libreoffice-fresh 2016-04-08 12:35:16 ncopa: see https://travis-ci.org/alpinelinux/aports/builds/121475426 2016-04-08 12:35:24 hum... still waiting for the ca-certificates update :( 2016-04-08 12:35:32 and on my machine http://haste.fit.cvut.cz/oyuqera.log 2016-04-08 12:35:41 ncopa: sorry wrong link https://projects.archlinux.org/svntogit/packages.git/tree/trunk/PKGBUILD?h=packages/libreoffice-fresh#n180 2016-04-08 12:37:02 from what branch is built the build server – testing or stable v3.3? 2016-04-08 12:38:08 edge (git master) 2016-04-08 12:38:18 that error comes due to old linux headers 2016-04-08 12:38:24 eh, pardon, I meant edge or stable 2016-04-08 12:38:34 Travis runs on edge too 2016-04-08 12:38:58 hmm… what version of linux headers does it need? 2016-04-08 12:39:01 you need to build linux-headers, then upgrade the linux-headers and then finally build hvtools 2016-04-08 12:39:08 hvtools needs linux-headers-4.4.x 2016-04-08 12:39:44 ah, so it’s caused by delay in synchronization of mirrors 2016-04-08 12:39:52 could be 2016-04-08 12:39:56 what mirror do you use? 2016-04-08 12:40:10 mine https://github.com/alpinelinux/aports/blob/master/.travis/repositories 2016-04-08 12:40:18 I’ve tried dl-2, but it was quite slow 2016-04-08 12:40:34 i'd recommend dl-cdn.a.o 2016-04-08 12:40:49 but 2016-04-08 12:40:51 fit.cvut.cz mirror is synced each 15 minutes 2016-04-08 12:41:09 what if you in same pull request have a dependency 2016-04-08 12:41:16 that is needed 2016-04-08 12:41:34 lets say i pushed linux-headers update and hvtools in same git push 2016-04-08 12:41:47 to build hvtools you need the updated linux-headers 2016-04-08 12:41:51 if it’s in the same PR and the same branch (e.g. testing, stable), then is should works 2016-04-08 12:42:19 why does not hvtools work then? 2016-04-08 12:42:21 if it’s in different branch, then it doesn’t, because builddirs… https://github.com/alpinelinux/aports/blob/master/.travis/build-pkgs#L27 2016-04-08 12:43:01 I don’t know how to sort deps across different repos 2016-04-08 12:43:06 * branch → repo 2016-04-08 12:43:32 (1/2) Installing linux-headers (4.1.12-r0) 2016-04-08 12:43:58 yes, it picks wrong 2016-04-08 12:44:04 hm 2016-04-08 12:44:10 I think that newer linux-headers wasn’t available in the mirror when it was built 2016-04-08 12:44:31 linux-headers 4.4.6 2016-04-08 08:32:30 2016-04-08 12:44:48 probably yes 2016-04-08 12:44:50 hvtools 2016-04-07 16:40:49 2016-04-08 12:45:03 uh… 2016-04-08 12:45:17 ok, I should look at commit instead 2016-04-08 12:45:26 :) 2016-04-08 12:46:09 nmeum: i dont know if its worth to patch libreoffice and the other pkg that uses mdds 2016-04-08 12:46:14 well, you’ve pushed hvtools *before* updated linux-headers 2016-04-08 12:47:12 ncopa: are there any other packages which don't work with >= 0.1.2? If it is just libreoffice it might be worth it 2016-04-08 12:47:13 pardon… 2016-04-08 12:47:33 it was in the correct order, I’m blind 2016-04-08 12:47:59 I’m gonna change mirror to dl-cdn.a.o 2016-04-08 12:53:19 ncopa: there it is https://github.com/alpinelinux/aports/pull/50 2016-04-08 12:55:59 ncopa: i have to go home soon. do you know the queue length until it gets build? 2016-04-08 12:56:07 http://build.alpinelinux.org/ 2016-04-08 12:56:19 its chewing libvirt now 2016-04-08 12:56:34 i dont know how long time it will take 2016-04-08 12:56:40 but it may be another hour 2016-04-08 12:56:48 we can test it next week 2016-04-08 12:56:52 thanks 2016-04-08 12:57:06 btw https://github.com/alpinelinux/aports/pull/49 fails because it depends on previous PRs, I’ll force travis to rebuild it when the dependent PRs is merged to ensure that it’s correct 2016-04-08 12:57:38 jirutka: do you think you could merge the pr's? 2016-04-08 12:57:48 so its one pr with multiple commits 2016-04-08 12:57:52 than i will reboot back for now 2016-04-08 12:58:02 then i can git pull in one shot 2016-04-08 12:58:04 I’m thinking about extending travis scripts to parse PR message and when it finds something like ”depends on #xx, #yy”, then it’ll also merge these in the build 2016-04-08 12:58:19 mosez: ok. sorry and thanks 2016-04-08 12:58:30 np 2016-04-08 12:58:35 ncopa: do you mean jruby and these two dependent PRs or all of them? 2016-04-08 12:58:36 thanks for your help :) 2016-04-08 12:59:32 not sure how they depend on each other 2016-04-08 12:59:59 but in general, if you have group of commits, then its easer for me to pull 2016-04-08 13:00:08 if they are in same pr 2016-04-08 13:00:16 i do one pull and get it all 2016-04-08 13:00:23 instead of doing 5 different pulls 2016-04-08 13:00:27 which takes more time 2016-04-08 13:00:29 jruby (#49) depends on java-jffi (#46) and java-jansi-native (#45)… then there are java-snappy and java-lz4 which are independent of jruby 2016-04-08 13:00:55 but they are all java 2016-04-08 13:01:01 so i'd just merge them all 2016-04-08 13:01:09 it’s against general practice to send small PRs, but okay, I have no problem with it 2016-04-08 13:01:17 so I’ll send new PR and close these 2016-04-08 13:01:35 many small commits is good 2016-04-08 13:01:45 btw these aports are not just plain Java, all of them have native extensions, that’s why I’ve made aports 2016-04-08 13:02:01 ok 2016-04-08 13:02:57 btw native extensions in Java are really a nightmare… native binaries bundled in JARs, ugh 2016-04-08 13:03:11 i heard you mention that 2016-04-08 13:03:13 and “of course” everyone expects that Linux = glibc 2016-04-08 13:03:50 sounds like they don't understand how open source works 2016-04-08 13:04:45 JNI (java native interface) was designed intentionally to be horrible, to discourage people using it… I belive that it was good decision… the problem is that some people don’t get it and still use it… 2016-04-08 13:05:54 Intentionally designing bad interfaces to discourage people from using them is a terrible idea. Because people *will* use them anyway. 2016-04-08 13:08:51 actually, they did not so much 2016-04-08 13:09:09 it’s very rare in Java world 2016-04-08 13:09:14 and it has really great benefits 2016-04-08 13:09:27 C extensions are always pain in the ass 2016-04-08 13:21:06 ncopa: here we go https://github.com/alpinelinux/aports/pull/51 (all green https://travis-ci.org/alpinelinux/aports/builds/121695697) 2016-04-08 13:22:14 and I should probably bring attention to https://github.com/jirutka/alpine-aports/blob/java-aports/testing/jruby/jruby.post-install, it’s quite hack-ish… 2016-04-08 13:22:47 but I guess that it’s better to do it in post-install than add this symlink to the package itself 2016-04-08 13:23:03 we have a libc-compat package 2016-04-08 13:23:41 I know… but libc-compat in the current stable branch doesn’t install symlink for crypt.so 2016-04-08 13:24:15 file a bug for it on bugs.a.o and i'll cherry pick that fix for 3.3-stable 2016-04-08 13:24:16 I’ve added 4 days ago in symlink in https://github.com/alpinelinux/aports/pull/44 2016-04-08 13:24:43 * I’ve added this symlink 4 days ago in… 2016-04-08 13:25:03 i think its a bug so we can cherry-pick it to 3.3-stable 2016-04-08 13:25:18 okay, so should I open bug? 2016-04-08 13:25:29 yes please 2016-04-08 13:25:44 thanks! 2016-04-08 13:26:09 i think i will reboot the builders on monday 2016-04-08 13:26:13 just fyi 2016-04-08 13:29:16 ncopa: you have any idea on how to fatload alpine initramfs with moodloop, kernel? 2016-04-08 13:29:43 i have got upto bootloader using u-boot from mainline 2016-04-08 13:29:50 fatload? 2016-04-08 13:30:09 ncopa: https://bugs.alpinelinux.org/issues/5385 2016-04-08 13:32:25 fatload mmc 0 0x48000000 boot/vmlinuz 2016-04-08 13:33:16 is there a way to tell u-boot to load vmlinuz or initramfs 2016-04-08 13:33:35 any ideas from a fat partition? 2016-04-08 13:34:03 i am just a few steps away for making nanopi m1 work with alpine 2016-04-08 13:34:09 oneinsect: i have no idea :-( 2016-04-08 13:34:26 hmmm 2016-04-08 13:34:45 jirutka: thanks! 2016-04-08 13:35:11 how is rpi able to load without a rootfs file system? does it first load the kernel vmlinuz and then initramfs? 2016-04-08 13:35:14 is that how it works? 2016-04-08 13:37:05 oneinsect: i dont know 2016-04-08 13:37:21 no problem 2016-04-08 13:37:23 sorry for bugging 2016-04-08 13:37:31 speaking of that, i have a proj to make a macbook pro dual boot 2016-04-08 13:37:43 but thats with uefit 2016-04-08 13:37:50 ncopa: uh, you’ve already merged jruby, I wanted to remove the post-install hack for crypt.so and set `libc6-compat>=1.1.12-r5` instead 2016-04-08 13:38:20 i merged it because it was so convenient :) 2016-04-08 13:38:39 so you want to keep this post-install hack there? 2016-04-08 13:38:44 nah 2016-04-08 13:38:50 we post fix it 2016-04-08 13:39:09 should I send a patch? 2016-04-08 13:39:16 yes please 2016-04-08 13:39:50 its a trivial fix i assume, you can probably just: git format-patch --stdout 1 | tpaste 2016-04-08 13:44:13 ncopa: okay, here: http://haste.fit.cvut.cz/raw/hikimaq.patch 2016-04-08 13:45:28 nice 2016-04-08 13:45:34 i bump pkgrel for you too 2016-04-08 13:45:37 thanks! 2016-04-08 13:46:25 :) 2016-04-08 14:10:39 oneinsect: http://wiki.alpinelinux.org/wiki/Alpine_on_ARM -> Booting Linux kernel 2016-04-08 14:31:49 xming: 2016-04-08 14:32:04 0 to 8kb is kept free and 8 to 1024 includes 2016-04-08 14:32:09 SPL Loader 2016-04-08 14:32:11 and u-boot 2016-04-08 14:32:44 just trying to see next how to make u-boot pickup alpine kernel 2016-04-08 14:36:12 i am willing to write extensive documentation ...and post this and make alpine run on as many boards as I can ... 2016-04-08 14:50:53 "uboot uses extlinux.conf file to locate the kernel/initramfs/... just like syslinx, you need to put that file on the partitions which is marked bootable in the /boot directory" 2016-04-08 14:51:11 it's under "Booting Linux kernel" 2016-04-08 14:57:40 but xming: will mainline u-boot support extlinux.conf? 2016-04-08 14:58:16 mainline u-boot supports extlinux.conf as per the doc I wrote 2016-04-08 14:58:36 great 2016-04-08 14:58:46 what about tar ball support patches? 2016-04-08 14:58:49 do we need them? 2016-04-08 14:59:51 kernel is fine but initrd in your wiki points to /boot/initramfs-3.17.0-0.rc4.git2.1.fc22.armv7hl.img 2016-04-08 14:59:58 an img? 2016-04-08 15:01:14 the alpine-uboot-3.3.3-armhf has apks directory 2016-04-08 15:02:26 this is what i did dd if=u-boot-sunxi-with-spl.bin of=/dev/sdc bs=1024 seek=8 2016-04-08 15:02:27 next 2016-04-08 15:02:38 mkfs.fat /dev/sdc1 2016-04-08 15:03:08 copied the apks dir to /dev/sdc1 2016-04-08 15:04:15 I haven't tried initramfs yet 2016-04-08 15:04:27 I was told that tarball support was upstreamed 2016-04-08 15:07:12 any ideas if it made through the v2016.03 release of u-boot released 3 weeks ago? 2016-04-08 15:10:05 I guess so, just try :D 2016-04-08 15:10:35 okie i am good break myself tonite 2016-04-08 15:10:40 on this 2016-04-08 15:10:50 i will buzz if need any help 2016-04-08 15:10:58 gonna* 2016-04-08 15:26:08 xming: one quick question 2016-04-08 15:26:43 my SPL/uboot is vfat partition and my second partition is also vfat 2016-04-08 15:27:09 so extlinux.conf should reside in the second partition under the boot directory 2016-04-08 15:28:36 extlinux.conf should be in /boot on a patition marked as bootable (DOS style) 2016-04-08 15:32:54 well there are two partitions 2016-04-08 15:32:59 the first one or the second one? 2016-04-08 15:33:08 the one which you marked bootable 2016-04-08 15:33:17 okie 2016-04-08 15:33:21 i am guessing the second one 2016-04-08 15:33:28 check with fdisk 2016-04-08 15:33:57 i am right i think coz i dded the first partition with u-boot with spl 2016-04-08 15:35:03 it doesn't matter, you just need a partition with bootable flag set 2016-04-08 15:35:24 okie 2016-04-08 15:36:07 so it could be one big partition 2016-04-08 15:36:12 great 2016-04-08 15:36:15 thanks let me try 2016-04-08 15:37:06 jut like on x86 2016-04-08 15:37:23 BTW spl and uboot doesn't need to be on a partition 2016-04-08 15:37:28 why not? 2016-04-08 15:37:43 just start your first partition at block 2048 2016-04-08 15:38:06 for allwinner everyone seems to suggest this 2016-04-08 15:38:07 SPL/uboot 2016-04-08 15:38:16 dd if=/dev/zero of=/dev/sdX bs=1M count=8 2016-04-08 15:38:28 see no patition 2016-04-08 15:38:43 it's not of=/dev/sdXY 2016-04-08 15:38:50 but of=/dev/sdX 2016-04-08 15:39:41 yes rather dd if=u-boot-sunxi-with-spl.bin of=/dev/sdc bs=1024 seek=8 2016-04-08 15:40:46 okie another nasty issue 2016-04-08 15:41:08 allwinner h3 is mainline almost fully in kernel 4.6 R2 2016-04-08 15:41:34 is there any wiki howto to compile alpine using latest bleeding edge 2016-04-08 15:53:26 use the vanilla aport 2016-04-08 15:54:10 http://git.alpinelinux.org/cgit/aports/tree/main/linux-vanilla/APKBUILD 2016-04-08 15:56:37 hmm 2016-04-08 15:57:02 you won't have grsec 2016-04-08 16:01:35 Hi I have a strange situation. When I check with ''grep -rn root /etc/password', it outputs: "1:root:x:0:0:root:/root:/bin/ash" which categorically states that root has ash as SHELL, but when I check with 'echo $SHELL', it outputs '/bin/bash'. Very strange in an alpine micro lxc container running in a debian jessie host. 2016-04-08 16:01:58 What could be the reason for such strange behaviour? 2016-04-08 16:03:49 exec bash in root's .profile or any of the included files? 2016-04-08 16:04:04 the login shell isn't necessarily the one you end up with when logging in. 2016-04-08 16:04:51 the login shell as defined in /etc/passwd is only the shell that /bin/login executes, the one that interprets your login files. 2016-04-08 16:05:04 From there on, anything goes. 2016-04-08 16:06:46 skarnet: there is no specific .bashrc or .bashrc_profile at all. 2016-04-08 16:08:30 .profile ? .shrc ? /etc/profile ? 2016-04-08 16:08:50 .bash_profile would only be executed if bash is the login shell, which it's not. 2016-04-08 16:13:46 hi, I just installed, pinned kernel to edge, abuilded zfs... but it is faulty... is 2016-04-08 16:14:00 anyone using zfs? 2016-04-08 16:15:23 zenny1: again, how are you entering the container 2016-04-08 16:16:05 what command did you use, because this sound's a lot like didn't login, just started sh with old environment 2016-04-08 16:23:52 xming: for testing grsec wont be necessary 2016-04-08 16:34:31 Why does the rpi image by default allocate 256mb gpu memory? Does not seem to have any hardware accelerated gpu features (lack of opengl, not even firefox is gpu accelerated for css3) 2016-04-08 16:36:11 you can always change it 2016-04-08 16:36:16 check config.txt 2016-04-08 16:36:22 and you will understand why 2016-04-08 16:36:37 hardware accelerated gpu is coming in mainline for rpi as far i know 2016-04-08 16:37:51 whoa 2016-04-08 16:37:56 shame I don't use my pi for anything graphical 2016-04-08 18:28:41 is there a way to force subpackages to rebuild? 2016-04-08 18:33:17 skarnet: there is nothing inside the /root that relates with any shell except .ash_history 2016-04-08 18:33:46 ammunta: 'lxc-attach -n containername' is how I am entering into the container. 2016-04-08 18:36:04 However I can 'export SHELL=/bin/ash' and do whatever I want, but not immediately after login. 2016-04-08 20:52:57 What was the question? 2016-04-08 21:27:09 Just for info: I wrote wiki article about running JRuby on Alpine Linux and issues with JNI (aka native extensions). https://github.com/jruby/jruby/wiki/JRuby-on-Alpine-Linux 2016-04-09 04:44:02 friends why is CONFIG_ARCH_SUNXI not set in config-vanilla.armhf 2016-04-09 09:27:43 xming: i am here 2016-04-09 09:29:31 quickly is uInitrd same as our initramfs 2016-04-09 09:29:44 i am able to boot upto kernel 2016-04-09 09:29:59 but its not picking up from the vfat folder 2016-04-09 09:30:03 from root 2016-04-09 09:30:05 the apks 2016-04-09 09:30:37 Target filesystem doesn't have requested /sbin/init... 2016-04-09 09:30:40 Running /scripts/local-bottom ... done. 2016-04-09 09:30:44 Running /scripts/init-bottom ... mount: No such file or directory 2016-04-09 09:30:47 No init found. Try passing init= bootarg. 2016-04-09 09:31:15 i am trying to boot a mailine kernel with its own initrd with alpine userspace 2016-04-09 09:31:42 if i setenv bootargs root=/apks/armhf rootwait panic=10 ....it says 2016-04-09 09:31:42 Rebooting automatically due to panic= boot argument... 2016-04-09 09:31:51 any ideas? 2016-04-09 09:32:04 the boot flaged partition is fat 2016-04-09 09:32:24 where it contains /apks /boot and other folders 2016-04-09 09:32:50 boot folder contains linux vmlinuz-4.6.0-rc1-sunxi etc 2016-04-09 09:46:32 I am sorry, I always have trouble to decipher your descriptions 2016-04-09 09:46:42 are you using your own initramfs? 2016-04-09 09:48:15 well okie 2016-04-09 09:48:20 let me begin 2016-04-09 09:48:53 i have got my own dtb sun8i-h3-orangepi-pc.dtb, my own kernel vmlinuz-4.6.0-rc1-sunxi and my own uInitrd 2016-04-09 09:49:21 now what arguments should i pass so as to make them look for alpine userspace files 2016-04-09 09:49:35 i tried setenv bootargs modules=loop,squashfs,sd-mod,usb-storage 2016-04-09 09:49:48 but it seems to give an error in finding them 2016-04-09 09:51:25 pass the right root= and initrd= in your extlinux.conf 2016-04-09 09:52:08 No init found. Try passing init= bootarg. 2016-04-09 09:52:22 is that before or after pivot root? 2016-04-09 09:53:35 after it boots up the kernel with initrd....it looks for ....and says Target filesystem doesn't have requested /sbin/init... 2016-04-09 09:53:52 paste the extlinux.conf 2016-04-09 09:54:02 I am guessing in the dark 2016-04-09 09:54:07 yes one moment 2016-04-09 09:54:53 LABEL Linux Mailine 4.6RC2 2016-04-09 09:54:56 LINUX boot/vmlinuz-4.6.0-rc1-sunxi 2016-04-09 09:54:59 INITRD boot/uinitrd 2016-04-09 09:55:03 FDT boot/sun8i-h3-orangepi-pc.dtb 2016-04-09 09:55:07 APPEND BOOT_IMAGE=/boot/vmlinuz-4.6.0-rc1-sunxi modules=loop,squashfs,sd-mod,usb-storage console=${console} 2016-04-09 09:55:37 that the extlinux.conf file 2016-04-09 09:57:36 can you paste the boot log? Not here, use tpaste/pastebin/... 2016-04-09 09:58:18 yes a moment 2016-04-09 10:01:42 http://pastebin.com/XTb3WH0b 2016-04-09 10:02:18 just to note i am manually loading the kernel etc 2016-04-09 10:04:23 **Unable to read file /boot/extlinux/boot/uinitrd 2016-04-09 10:04:23 for failure retrieving initrd 2016-04-09 10:04:36 you need to make symlink in /boot 2016-04-09 10:06:51 okie 2016-04-09 10:07:02 i have edited extlinux.conf 2016-04-09 10:07:21 added LINUX /boot/vmlinuz-4.6.0-rc1-sunxi 2016-04-09 10:07:29 note the forward slash 2016-04-09 10:07:47 it then says **Unable to read file /boot/uinitrd for failure retrieving initrd SCRIPT FAILED: continuing... 2016-04-09 10:07:54 even though the file is there 2016-04-09 10:08:07 do i still have to make a symlink? 2016-04-09 10:09:28 the point is i am manually loading them...and it boots fine...what could be the right root=? 2016-04-09 10:21:40 yes you need to (cd /boot && ln -s . boot) 2016-04-09 10:22:18 what do you mean by "manually loading"? 2016-04-09 10:23:16 setenv machid 1029 2016-04-09 10:23:20 load mmc 0:1 0x43000000 boot/dtbs/sun8i-h3-orangepi-pc.dtb 2016-04-09 10:23:24 load mmc 0:1 0x42000000 boot/vmlinuz-4.6.0-rc1-sunxi 2016-04-09 10:23:28 load mmc 0:1 0x45000000 boot/uInitrd 2016-04-09 10:23:31 bootz 0x42000000 0x45000000 0x43000000 2016-04-09 10:24:03 if i say manually load load mmc 0:1 0x45000000 boot/initramfs-grsec 2016-04-09 10:24:07 it says invalid format 2016-04-09 10:25:33 this must be your board/Allwinner specific way, sorry I don't know anything about that 2016-04-09 10:25:40 hmmm 2016-04-09 10:26:48 even on rk boards, it's possible to use mainline uboot/kernel to start linux without any vendor specific stuffs 2016-04-09 10:27:10 no i mean for me i can start linux 2016-04-09 10:27:21 its not able find the init from initramfs 2016-04-09 10:28:42 which fs are you using for /boot? 2016-04-09 10:28:47 in sense how do get initramfs-grsec to load... 2016-04-09 10:28:49 fat 2016-04-09 10:28:55 2048 onwards its fat 2016-04-09 10:28:59 with boot flag on 2016-04-09 10:29:21 oh fat 2016-04-09 10:29:25 yes 2016-04-09 10:29:27 correct 2016-04-09 10:29:29 not ext4 2016-04-09 10:29:52 coz i believe once initramfs-grsec starts and takes over alpine loads to ram...it doesnt need touch the file system 2016-04-09 10:30:00 isnt it? 2016-04-09 10:31:17 I think so 2016-04-09 10:31:28 Wrong Ramdisk Image Format Ramdisk image is corrupt or invalid...that is what u-boot says ...when i ask it to load initramfs 2016-04-09 10:31:35 initramfs-grsec 2016-04-09 10:31:57 i have got my own dtb sun8i-h3-orangepi-pc.dtb, my own kernel vmlinuz-4.6.0-rc1-sunxi and my own uInitrd 2016-04-09 10:32:13 and now you are using alpine's initramfs? confusing 2016-04-09 10:32:29 well uinitrd works ... yes i tried replacing it with alpine's initramfs 2016-04-09 10:32:44 so that i get a alpine userspace 2016-04-09 10:32:54 with my own kernel+dtb 2016-04-09 10:33:08 on a fat formated sdcard 2016-04-09 10:34:59 okie may be the format of alpine's initramfs is not what u-boot is expecting 2016-04-09 10:35:17 someone suggested 2016-04-09 10:35:23 mkimage -n 'Ramdisk Image' -A arm -O linux -T ramdisk -C gzip -d initramfs.cpio.gz initramfs.uImage 2016-04-09 10:37:03 initramfs-grsec is gzipped cpio 2016-04-09 10:39:00 hmm 2016-04-09 10:39:06 wonder why u-boot is not taking it 2016-04-09 10:47:00 can someone please confirm if indeed initramfs-grsec is an ARM Linux RAMDisk Image (gzip compressed) 2016-04-09 14:21:12 how would I go about having a package that pulls from git? specifically I want to pull specific commits from two repos from github and then extract them. however the extraction ends up leaving the directories as program-hash.tar.gz. 2016-04-09 14:27:47 that's, uh, not supported 2016-04-09 14:44:27 that does pose a problem then :( 2016-04-09 14:45:33 all that is really needed is the ability to use tars extract to specific directory flag 2016-04-09 14:47:00 Adran: well, you can download repository as a tar for any commit, e.g. https://github.com/alpinelinux/aports/archive/09bb3ef33a4eea4ed140b39103179372b9d9a07a.tar.gz 2016-04-09 14:47:17 jirutka: thats what i am doing 2016-04-09 14:47:23 so what’s the problem? 2016-04-09 14:47:30 however unless you call 09bb3... the pkgver 2016-04-09 14:47:38 you won't be able to extract it 2016-04-09 14:48:13 pkgver doesn’t need to be the same as hash 2016-04-09 14:48:34 just bumping the package would be more difficult, because you need to find new hash 2016-04-09 14:48:56 part of my problem is this package decided that instead of using submodules they wanted to use separate repos 2016-04-09 14:49:03 so I have to download two commits as archives 2016-04-09 14:49:14 this is also no problem 2016-04-09 14:49:17 just inconvenient 2016-04-09 14:49:38 alright, I might be missing something, what variable would I use to know which directory to go into? 2016-04-09 14:50:24 APKBUILD is almost just a shell script, there’s very little magic involved 2016-04-09 14:50:48 what software you would like to package? 2016-04-09 14:51:33 i've finished packaging bro, going to upload it, it works fine. but they prefer to use git for all their newer development so I'm trying to make a git version as well. 2016-04-09 14:52:05 i created _gitcommit_bro and _gitcommit_plugins variables to identify the two directories 2016-04-09 14:52:14 it’ll be probably easier if I write you an example 2016-04-09 14:52:26 maybe. i can save you a bit of time though 2016-04-09 14:52:33 wait a moment, this software is actually named “bro”? XD 2016-04-09 14:52:46 yes, ids 2016-04-09 14:52:52 ah, okay :) 2016-04-09 14:53:26 _gitrel_bro="b5f1fb33fac78d305653feb799bd936a16083e3b" _gitrel_plugins="ab61be0c4f128c976f72dfa5a09a87cd842f387a" 2016-04-09 14:53:27 source="$pkgname-bro.tar.gz::https://github.com/bro/bro/archive/$_gitrel_bro.tar.gz $pkgname-plugins.tar.gz::https://github.com/bro/bro-plugins/archive/$_gitrel_plugins.tar.gz" 2016-04-09 14:53:36 is what I'm doing currently so at least I can target specific commits 2016-04-09 14:53:40 If the software is called "bro", that's already a giant red flag 2016-04-09 14:53:59 skarnet: heh, surprisingly its very good software 2016-04-09 14:54:15 uh, wait few minutes, I’ve totally forgotten to start few servers after planned power outtage… 2016-04-09 14:55:35 jirutka: here's the last "standard" release one I have finished https://gist.github.com/Ttech/35ba2fe54362faf171989b86af60c0fb 2016-04-09 14:55:41 skarnet: I ... think it predates that use of "bro" 2016-04-09 14:56:16 f*cking nagios, it should remind me… 2016-04-09 14:56:51 COPYING says 1995-2015 2016-04-09 14:57:10 so that may explain how it could be plausibly nonterrible 2016-04-09 14:57:16 no comment on actual quality 2016-04-09 14:57:33 well 2016-04-09 14:57:36 if you look at my gist 2016-04-09 14:57:39 its quality is kinda sad 2016-04-09 14:57:53 it uses every glibc / gnu thing it could possibly find 2016-04-09 14:58:10 and it works on Alpine? 2016-04-09 14:58:17 even memfrob() ? 2016-04-09 14:58:20 jirutka: yeah 2016-04-09 14:58:31 skarnet: lol well now you're just being mean. :D 2016-04-09 14:58:45 jirutka: Took me a week, I'm bad at this, but it works nwo 2016-04-09 14:58:47 now* 2016-04-09 14:58:50 what’s memfrob()? 2016-04-09 14:59:14 look it up :) 2016-04-09 14:59:51 ah, our ICT disabled all notifications globally… omfg… they’re really idiots 2016-04-09 15:00:04 jirutka: what 2016-04-09 15:00:11 now you don't have to worry about those annoying emails 2016-04-09 15:00:15 they've helped you! 2016-04-09 15:00:17 XD 2016-04-09 15:01:12 also here are all the packages I've built so far https://gist.github.com/Ttech/d0047b4c7b38e929798ee95d753d00c1 2016-04-09 15:01:13 :( 2016-04-09 15:01:21 if anyone is brave enough to check for quality 2016-04-09 15:02:58 oh forgot to add an init script for bro... oops 2016-04-09 15:05:05 > but they prefer to use git for all their newer development 2016-04-09 15:05:05 do they use git tags for releases? 2016-04-09 15:09:08 well, it seems that they releases bro-plugins independetly to bro, so it’d be maybe better to divide them into separate packages 2016-04-09 15:09:31 jirutka: Thats hard though, I did that in my bro package, but it requires bro to build the plugins (so I did subpackages) 2016-04-09 15:09:37 like bro and bro-plugins, where bro-plugins would be just “virtual” package with submodules for each separate plugin 2016-04-09 15:09:37 bro source* 2016-04-09 15:09:57 whole source code, not just header files atc? 2016-04-09 15:10:00 etc 2016-04-09 15:10:02 yeah. 2016-04-09 15:10:06 uh, okay 2016-04-09 15:10:09 the configure option is specifically --build-dir= 2016-04-09 15:10:10 :( 2016-04-09 15:10:45 and do they use git tags for releases or they don’t gonna bother with proper releasing? 2016-04-09 15:10:56 they haven't been bothering with proper releases 2016-04-09 15:11:02 aha 2016-04-09 15:11:26 so I was going to try to just try the git way to have something that has newer features but very broken. :\ 2016-04-09 15:12:15 git is good for CI and flexibility but you can't do without numbered releases, for reproducible builds :/ 2016-04-09 15:13:08 you can tag releases in git 2016-04-09 15:13:14 that’s not problem 2016-04-09 15:13:26 yes, but you need to do the tagged releases :) 2016-04-09 15:13:31 problem is that they don’t release this software at all… 2016-04-09 15:13:36 commits are "tagged" 2016-04-09 15:13:57 skarnet: they've been working on a development version that is incompatible with the last release... 2016-04-09 15:14:03 Their solution is "just use git" 2016-04-09 15:14:27 Branches mother!@#$ers, do you speak it? 2016-04-09 15:14:34 I’m confused… 2016-04-09 15:16:10 when I look at https://github.com/bro/bro/releases, then the last release is 2.4.1 from Sep 6, 2015… is this really the last release? if not, will it be or they don’t do releases anymore and tell users to use just the current HEAD? 2016-04-09 15:17:54 I don't think its official policy not to do releases, I just think they're really really developing. 2016-04-09 15:18:23 okay 2016-04-09 15:18:30 Don't want to make you spend too much time on this 2016-04-09 15:18:41 Just was hoping someone knew how to do git apkbuilds 2016-04-09 15:18:52 give me few minutes 2016-04-09 15:21:00 do you know what will be the next release – 2.5 or 3.0 (in terms of semver)? 2016-04-09 15:38:53 are there somwhere described rules for Alpine’s pkgver? 2016-04-09 15:41:52 ah, here http://git.alpinelinux.org/cgit/apk-tools/tree/src/version.c …uses Gentoo format, good 2016-04-09 15:57:47 Adran: this should work, but I don’t have some required dependencies, so couldn’t verify it; http://haste.fit.cvut.cz/qujaqoz.sh 2016-04-09 16:00:08 it seems that Alpine’s version scheme doesn’t allow non-numeric characters except predefined suffixes, so we can’t embed git SHA into pkgver :( so instead I used 2.5_git and you should update pkgrev when updating package to newer revision… it’s just one of the possible options, I don’t know if the less evil or not 2016-04-09 16:00:39 afk 2016-04-09 17:16:50 fabled: ncopa: 2016-04-09 17:16:55 i am able to run alpine 2016-04-09 17:16:58 on orange pi 2016-04-09 17:17:03 and nanopi m1 2016-04-09 17:17:13 the quadcore sub $10 board 2016-04-09 17:17:24 can we have an alpine release for these also? 2016-04-09 17:20:53 one final issue remains though 2016-04-09 17:20:57 Mounting modloop ...mount: mounting /dev/loop0 on /.modloop failed: Invalid argument 2016-04-09 17:21:08 i need to generate a modloop file for the kernel 2016-04-09 17:21:11 any ideas how 2016-04-09 18:09:08 okie generating a new squashfs file 2016-04-09 18:30:16 oneinsect, does it require separete kernel? or just boot image? 2016-04-09 18:30:31 could you write to mailing list how to do it? 2016-04-09 18:30:56 fabled: its too simple...infact its mainline now 2016-04-09 18:31:02 cool 2016-04-09 18:31:08 if you time can you spare me 5 minutes 2016-04-09 18:31:13 if it's mainline i'm sure we can do it 2016-04-09 18:31:15 i wanted to discuss this with you 2016-04-09 18:31:16 i'm off soon 2016-04-09 18:31:19 please email 2016-04-09 18:31:29 okie i will email to mailing list then? 2016-04-09 18:31:29 i'm traveling tomorrow 2016-04-09 18:31:34 yes, please email to list 2016-04-09 18:31:50 i'm off the netherlands tomorrow for all of next week 2016-04-09 18:32:00 this is serious...we need to get an image out for orange pis, nanopis, what not pis....this will bring lots of people 2016-04-09 18:32:01 to alpine 2016-04-09 18:32:06 yes 2016-04-09 18:32:15 i will do that in sometime now 2016-04-09 18:32:22 perfect 2016-04-09 18:32:29 will you be able to compile 4.6 RC2 2016-04-09 18:32:31 if needed? 2016-04-09 18:32:34 fabled: you’re Nathan Angelacos, right? 2016-04-09 18:32:46 jirutka, no. i'm Timo Ter*s 2016-04-09 18:33:02 jirutka, Nathan is usually nangel if he's around 2016-04-09 18:33:20 ah, okay 2016-04-09 18:34:23 oneinsect, we try to stick with kernel lts releases. we can see if we'll add linux-vanilla to testing that is the bleeding edge 2016-04-09 18:34:47 but let's go through this discussion on mailing list 2016-04-09 18:35:00 yes we may need bleeding edge as a seperate release 2016-04-09 18:35:02 sure 2016-04-09 20:15:30 anyone have any ideas on this: https://github.com/ariya/phantomjs/issues/14186 2016-04-10 05:46:42 does anyone use /etc/init.d/ipset ? 2016-04-10 07:05:21 BitL0G1c: no why? 2016-04-10 07:05:26 you need any help? 2016-04-10 07:21:23 it's ok - I think I worked out how to format /etc/ipset.d/* 2016-04-10 07:33:30 hmm 2016-04-10 11:56:36 german mirror is down http://mirror1.hs-esslingen.de 2016-04-10 14:19:12 its up for me? 2016-04-10 16:31:08 fabled: 2016-04-10 16:31:12 i am still working on 2016-04-10 16:31:13 http://wiki.alpinelinux.org/wiki/DIY_Fully_working_Alpine_Linux_for_Allwinner_and_Other_ARM_SOCs 2016-04-10 16:31:17 once i complete 2016-04-10 16:31:24 i put up for the mailing list 2016-04-11 04:29:09 ncopa: why is /opt/ disallowed from a package build? Especially when the package presumes that? (as I just realized) 2016-04-11 05:22:43 never mind, fixed the program, but /opt I thought was for specific bundled releases? 2016-04-11 05:32:07 and jemalloc appears not to work properly 2016-04-11 05:41:42 https://gist.github.com/Ttech/63966d6d00c340d71ab07fc127cfea27 2016-04-11 06:27:21 Adran: fhs reasons, although i think abuild is a bit stricter than fhs iirc 2016-04-11 06:28:55 also it occurs to me that as defined by fhs /opt is actually really stupid 2016-04-11 06:58:50 Adran: http://www.pathname.com/fhs/pub/fhs-2.3.html#OPTADDONAPPLICATIONSOFTWAREPACKAGES 2016-04-11 07:06:32 wow, -r3 is still not available :D 2016-04-11 07:06:59 good that it works with a more current kernel 2016-04-11 07:12:17 what is the equivalent of libmysqlclient-dev in alpine? I am having a trouble as stated https://discuss.erpnext.com/t/errors-with-bench-init/11676/14 2016-04-11 07:12:41 Erpnext developers say it is because of the lack of libmysqlclient-dev 2016-04-11 07:16:24 mosez: i think its available 2016-04-11 07:16:28 http://dl-cdn.alpinelinux.org/alpine/edge/main/armhf/ca-certificates-20160104-r3.apk 2016-04-11 07:20:31 http://dl-cdn.alpinelinux.org/alpine/edge/main/armhf/ still shows -r2 2016-04-11 08:00:11 can someone reproduce this issue when installing giblib (edge)? ERROR: giblib-1.2.4-r8: BAD signature 2016-04-11 08:03:20 not here, but what mirror is it from 2016-04-11 08:04:04 mosez: it's cached 2016-04-11 08:04:32 it also might depend on where you're viewing from 2016-04-11 08:05:58 stwa, no errors for me too 2016-04-11 08:06:38 i am using http://dl-5.alpinelinux.org/alpine/edge/main/x86_64 2016-04-11 08:08:08 I am using http://nl.alpinelinux.org/alpine/edge 2016-04-11 08:11:20 slukin: i switched to mirror1.hs-esslingen.de/pub/Mirrors/alpine/edge and it works… 2016-04-11 08:11:35 fine 2016-04-11 08:12:24 not really, i thought the mirrors should all have the same packages, right? 2016-04-11 08:12:51 yes, you are right 2016-04-11 08:13:35 i am trying to understand how such an error occurs 2016-04-11 08:17:15 network error maybe? 2016-04-11 08:17:27 did it just happen once, or did it repeatedly happen 2016-04-11 08:17:38 Or it could be an EVIL MALICIOUS MIRROR 2016-04-11 08:17:59 I could reproduce "BAD signature" error with http://dl-5.alpinelinux.org repo 2016-04-11 08:18:06 repeatedly, i don't think a network error changes package signatures 2016-04-11 08:18:33 An early EOF could 2016-04-11 08:19:02 skarnet: the archive looks okay 2016-04-11 08:21:24 I'm not reproducing on apk verify 2016-04-11 08:22:28 ammunta: i have downloaded the package from both repos, the contents are the same but the binaries differ 2016-04-11 08:23:07 bit rot 2016-04-11 08:27:21 ... hm 2016-04-11 08:27:52 they have different .PKGINFO and they're signed by different keys 2016-04-11 08:27:56 the contents also differ 2016-04-11 08:28:40 somehow they have different dependencies computed 2016-04-11 08:29:17 http://sprunge.us/IXdI is this supposed to happen 2016-04-11 08:52:14 w 35 2016-04-11 09:44:55 Adran: this should work, but I don’t have some required dependencies, so couldn’t verify it; http://haste.fit.cvut.cz/qujaqoz.sh 2016-04-11 09:44:55 it seems that Alpine’s version scheme doesn’t allow non-numeric characters except predefined suffixes, so we can’t embed git SHA into pkgver  so instead I used 2.5_git and you should update pkgrev when updating package to newer revision… it’s just one of the possible options, I don’t know if the less evil or not 2016-04-11 10:25:07 hey friends :) 2016-04-11 10:56:26 hmm, the first time in all this years that i have a problem with btrfs 2016-04-11 10:56:27 Stale file handle 2016-04-11 10:56:38 lets see how good the repair tools are :) 2016-04-11 10:56:54 hey friends, I need some help to play with erpnext in alpine linux, but having problem to figure out the libmysqlclient-dev package (mariadb-dev) didn't help. tried "apk add mysql-dev --update-cache --repository http://dl-3.alpinelinux.org/alpine/3.1/main/ --allow-untrusted" outputs "ERROR: http://dl-3.alpinelinux.org/alpine/3.1/main/: No such file or directory" error. 2016-04-11 10:56:54 but very interresting because i always had trouble with ext3/ext4 but never with btrfs 2016-04-11 11:07:27 ammunta: as i can see mirrors are just rsynced, so how can packages be built differently? 2016-04-11 11:13:50 stwa: packages are not different 2016-04-11 11:14:19 every mirror should have the exact copy of the package. it the package is different someone has modified the package on there own 2016-04-11 11:14:29 thats why you should always check the signatures on the package 2016-04-11 11:16:13 so APKBUILD files no longer have the prepare function? 2016-04-11 11:16:17 leo-unglaub: i had signature problems, that's why i ask… these packages do differ: http://dl-5.alpinelinux.org/alpine/edge/main/x86_64/giblib-1.2.4-r8.apk http://mirror1.hs-esslingen.de/pub/Mirrors/alpine/edge/main/x86_64/giblib-1.2.4-r8.apk 2016-04-11 11:16:19 leo-unglaub: don't rule out malice, but I'm pretty sure the discrepancy can be explained by some automation snafu. 2016-04-11 11:19:56 skarnet: hmmm 2016-04-11 11:20:01 it is possible 2016-04-11 11:20:34 but the difference is 0.3kb ... that is a lot for just some flipped bits 2016-04-11 11:21:09 an automation bug can end up making a lot more difference than a few flipped bits 2016-04-11 11:21:31 leo-unglaub: ammunta pointed out the dependencies: http://sprunge.us/IXdI 2016-04-11 11:21:33 I'd look at the mirroring scripts right in the eye 2016-04-11 11:21:35 well, the package structure itself is valid 2016-04-11 11:27:32 timestamp differ 2016-04-11 11:34:13 leo-unglaub: btrfs is on the top :D http://events.linuxfoundation.org/sites/events/files/slides/AFL%20filesystem%20fuzzing%2C%20Vault%202016.pdf 2016-04-11 11:51:03 xming: i hate you for that link ... now i cannot sleep anymore *g* knowing how fucked all our filesystems are *g* 2016-04-11 11:51:28 leo-unglaub: I use btrfs on my desktop too 2016-04-11 11:52:47 and I will never use xfs any where any more, once it ate the whole filesystem on a production server after a proper umount+shutdown, xfs tools didn't even recognize it as xfs any more 2016-04-11 11:53:46 so personally I just treat that paper the same as anything from horricle 2016-04-11 11:54:18 http://events.linuxfoundation.org/sites/events/files/slides/AFL%20filesystem%20fuzzing%2C%20Vault%202016.pdf 2016-04-11 11:54:28 %1 2016-04-11 11:54:32 %100 2016-04-11 11:54:44 i'm improving the bot 2016-04-11 11:55:15 ncopa: no more prepare() in APKBUILD, or do we add it manually when needed? 2016-04-11 11:55:29 depends on what you need it to do 2016-04-11 11:55:38 the comparison to ufs would be very interresting 2016-04-11 11:55:39 apply patches 2016-04-11 11:55:48 then its generally not needed 2016-04-11 11:55:53 but damn ... 5 s for BUG() in btrfs is depressing 2016-04-11 11:57:06 i quite liked the loop, and not having to think about prepare() 2016-04-11 11:57:17 wha's the prefered method now? 2016-04-11 11:57:40 you can still create prepare() if you want 2016-04-11 11:58:13 sure, but newabuild doesn't any more 2016-04-11 11:58:51 because any *.patch defined in $source will automatically be applied 2016-04-11 11:59:02 you may need set builddir though 2016-04-11 11:59:41 aah cool. yes i remember reading that conversation. can you put something in APKBUID file or elsewhere that exaplins the defaults? 2016-04-11 11:59:49 s/exaplins/explains 2016-04-11 12:00:33 i sent an email to mailing list....long back 2016-04-11 12:00:42 however i am not yet accepted 2016-04-11 12:00:46 anything wrong? 2016-04-11 12:01:15 5 s to BUG() is ... rather quick 2016-04-11 12:03:33 oneinsect: what was it, ports patch? 2016-04-11 12:03:45 alpine-devel 2016-04-11 12:03:56 oneinsect@gmail.com 2016-04-11 12:04:22 fabled: asked me to join and send out some email 2016-04-11 12:04:34 regarding alpine in orange pi arm boards 2016-04-11 12:06:33 oneinsect: can I have your opinion on orange pi? 2016-04-11 12:07:04 http://wiki.alpinelinux.org/wiki/DIY_Fully_working_Alpine_Linux_for_Allwinner_and_Other_ARM_SOCs 2016-04-11 12:07:18 its good if it can run alpine linux and now it does 2016-04-11 12:07:20 so yesss 2016-04-11 12:07:21 worth it 2016-04-11 12:07:39 go for orange pi pc - has dynamic voltage regulator 2016-04-11 12:07:43 keeps the product cool 2016-04-11 12:07:45 :-P 2016-04-11 12:08:10 i am so happy it runs alpine 2016-04-11 12:08:25 alpine is p sweet 2016-04-11 12:08:25 its good because it runs alpine. PERDIOD 2016-04-11 12:08:30 :-P 2016-04-11 12:08:33 loving apkbuild 2016-04-11 12:08:45 anyways, what's the procedure for submitting a pull request 2016-04-11 12:08:50 for a package 2016-04-11 12:09:29 ScrumpyJack: you wanna buy one? 2016-04-11 12:09:36 i purchased nanopi m1 2016-04-11 12:09:40 runs alpine as well 2016-04-11 12:09:44 and its good as well 2016-04-11 12:09:47 10$ 2016-04-11 12:10:11 i mean orange pi lite is like $9 and orange pi pc is like $15 2016-04-11 12:10:32 better than paying to macdonalds 2016-04-11 12:12:02 wonder why the mailing list 2016-04-11 12:12:05 is not accepting me 2016-04-11 12:12:09 may be it doesnt like me 2016-04-11 12:12:17 ncopa: 2016-04-11 12:12:20 i've been playing with AVR MCU, arduino, RPi, BBB. wondering what orange pi will add to the experiance 2016-04-11 12:12:33 diy h.264 encoder 2016-04-11 12:12:38 web radio 2016-04-11 12:12:48 at cheap cost 2016-04-11 12:13:00 it wont add much except for cost i believe 2016-04-11 12:30:13 grr, why doesn't my doc subpackage build 2016-04-11 15:55:30 anyone know if timo teras is in this channel 2016-04-11 16:01:30 mepholic: not now 2016-04-11 16:01:38 hm 2016-04-11 16:02:04 how does one submit a pull request for an apkbuild? 2016-04-11 16:02:12 for a package that already exists 2016-04-11 16:02:24 you want to submit a modification? 2016-04-11 16:02:41 you can do it on github, post your patch here or send to alpine-aports mailing list 2016-04-11 16:02:45 mepholic: edit the APKBUILD, increment pkgrev and send pull request 2016-04-11 16:04:18 the commit message should be something like: “main/foo: fix missing something” 2016-04-11 16:23:26 jirutka: surf builds but seg faults 2016-04-11 17:00:00 jirutka: perfect 2016-04-11 17:00:02 ty 2016-04-11 17:05:56 np 2016-04-11 21:01:41 is there any easy way to see what dependancies a package needs 2016-04-11 21:01:44 hm 2016-04-11 21:01:46 ldd i suppose 2016-04-11 21:04:05 usually reading config 2016-04-11 21:04:07 or readme 2016-04-11 21:24:52 mepholic: apk info -R or if you didn't install the package yet just use apk add -s 2016-04-11 21:37:02 Adran: O 2016-04-11 21:37:05 er 2016-04-11 21:37:10 I'm writing an apkbuild 2016-04-11 21:37:40 right 2016-04-11 21:37:59 when I do those, I usually try to compile the package manually 2016-04-11 21:38:01 see where it fails 2016-04-11 21:38:18 once I've done that, now I know mostly what the package requires and what it takes to get it working 2016-04-11 21:39:58 https://paste.ee/p/w5Wx0#5LDXIfM8aoW4uKubkWvwR2cwdelUQygz 2016-04-11 21:40:03 Adran: I see 2016-04-11 21:40:07 hm 2016-04-11 21:40:12 it's just interesting 2016-04-11 21:40:22 that the rest of the package/subpackages 2016-04-11 21:40:26 looks like it worked 2016-04-11 21:40:28 don't seem to have any external deps 2016-04-11 21:40:42 or your system already had them previously intalled 2016-04-11 21:40:49 line 106-111 2016-04-11 21:41:12 yup 2016-04-11 21:41:23 are the deps I added manually to the apkbuild based on the output of ldd of the .so that's built for freeswitch-perlesl 2016-04-11 21:41:36 is that proper? 2016-04-11 21:42:36 thing is, it didn't bomb out for me on compilation, ended up adding so many freaking packages to this thing during the build process, by the time I was done, I didn't really keep track of what was actually needed 2016-04-12 02:57:11 hi, does anyone know what license https://github.com/alpinelinux/aports/blob/master/main/libc-dev/sys-cdefs.h is under? same as musl? 2016-04-12 07:59:12 good morning :) 2016-04-12 08:02:49 morning 2016-04-12 08:06:45 i bought another ssd yesturday and put it in my laptop. then i installed fedora on the same machine ... its around 20% slower than alpine 2016-04-12 08:07:39 but wifi works :) and bluetooth ... and thats funny because i never knew that i had bluetooth on that laptop in the first place *g* 2016-04-12 08:24:08 leo-unglaub: what do you mean by 20% slower? 2016-04-12 08:28:26 scadu: the responsiveness of the system 2016-04-12 08:28:47 everyday tasks like: open file manager, move file x, install package x 2016-04-12 08:28:50 stuff like that 2016-04-12 08:40:39 firefox 45.0.2 is available :) 2016-04-12 09:19:23 clandmeter: ping 2016-04-12 09:20:55 pong 2016-04-12 09:21:18 clandmeter: you are propobly the only other hiawatha user here, right? 2016-04-12 09:21:28 no 2016-04-12 09:21:58 i never installed it. 2016-04-12 09:22:19 ah, sorry 2016-04-12 09:22:30 then i remembered it wrong 2016-04-12 09:22:51 not sure where you got that information from. 2016-04-12 09:22:59 i thought from you *g* 2016-04-12 09:23:06 hahaha 2016-04-12 09:23:07 i was using caddy 2016-04-12 09:23:10 or am 2016-04-12 09:26:39 ah, okay 2016-04-12 09:26:41 then nevermind :) 2016-04-12 09:27:00 i was just asking because i want to propose some changes to the config structure of hiawatha 2016-04-12 09:35:28 just updated and "ping: ping is not a valid applet" 2016-04-12 09:36:08 /bin/ping symlink target is owned by busybox-suid-1.24.2-r2 2016-04-12 09:36:40 ScrumpyJack: you are right, same here 2016-04-12 10:10:04 ncopa: what are your thoughts on a script that simply put wgets an entire mirror and checks all package signatures? 2016-04-12 10:10:14 usefull or simply bandwith burning? 2016-04-12 10:10:39 sounds like bw burning 2016-04-12 10:11:32 hmm, yeah ... i think it would be more usefull to validate the packages via cron job or in the rsync mirroring script directly if some mirror admin wants to 2016-04-12 10:13:40 leo-unglaub: what value does it add? 2016-04-12 10:14:55 ScrumpyJack: i was just talking to a friend of mine about the package difference that someone discovered last night 2016-04-12 10:14:56 I mean, I'm source rsyncing an alpinelinux.nl mirror. I trust that the files I'm rsyncing are valid. 2016-04-12 10:15:34 leo-unglaub: can you elaborate on what happened? 2016-04-12 10:16:08 ScrumpyJack: someone, i have forgetten who discovered that the same package from two different alpine mirrors was different 2016-04-12 10:16:16 and the signature was broken on one of the packages 2016-04-12 10:16:28 and that got me thinking about stuff like that 2016-04-12 10:16:53 "the same package from two different alpine mirrors was different" did you keep any evidence? 2016-04-12 10:16:56 dont get me wrong, there is no harm in a broken package, thats why there is a signature ... but as someone who deals in it security its still interresting to me 2016-04-12 10:17:26 it would be good to see what caused this 2016-04-12 10:17:45 ScrumpyJack: http://sprunge.us/IXdI 2016-04-12 10:17:48 that was the difference 2016-04-12 10:17:58 and yes, i am interrested in what caused it as well :) 2016-04-12 10:18:11 http://dl-5.alpinelinux.org/alpine/edge/main/x86_64/giblib-1.2.4-r8.apk 2016-04-12 10:18:11 http://mirror1.hs-esslingen.de/pub/Mirrors/alpine/edge/main/x86_64/giblib-1.2.4-r8.apk 2016-04-12 10:20:31 it looks like that package was updated, but the APKBUILD release variable was not bumped 2016-04-12 10:21:56 in this case, yes 2016-04-12 10:21:58 possible 2016-04-12 10:23:30 how does hs-esslingen.de pull from source? 2016-04-12 10:23:55 rsync shouldn't care about file names being the same 2016-04-12 10:24:25 my mirror has same version as dl-5, using rsync 2016-04-12 10:27:07 as an aside, giblib-1.2.4-r8.apk on hs-esslingen.de should not have failed a signature test. the sig in the apk tar file should work for the version of giblib-1.2.4-r8 on that host 2016-04-12 11:27:52 any can respond please 2016-04-12 11:27:54 Delivery to the following recipient has been delayed: alpine-devel+subscribe@lists.alpinelinux.org Message will be retried for 2 more day(s) Technical details of temporary failure: Google tried to deliver your message, but it was rejected by the server for the recipient domain lists.alpinelinux.org by mail.alpinelinux.org. [74.117.189.114]. The error that the other server returned was: 554 5.7.1 Service unavailable; 2016-04-12 11:28:38 Client host [209.85.161.195] blocked using bl.spamcop.net; Blocked - see http://www.spamcop.net/bl.shtml?209.85.161.195 2016-04-12 11:31:21 oneinsect_: your server is listed on multiple blacklsits 2016-04-12 11:31:21 https://mxtoolbox.com/SuperTool.aspx?action=blacklist%3a209.85.161.195&run=toolpage 2016-04-12 11:34:04 but i mailed from gmail 2016-04-12 11:40:57 well, use a good email provider then :) 2016-04-12 11:41:20 gmail, yahoo mail, hotmail ... all the same shit ... they only exist becaue they are huge ... 2016-04-12 11:41:50 hmmm 2016-04-12 11:43:52 on the other hand, it's a very bad idea to use centralized blacklists 2016-04-12 11:44:07 oneinsect_: you use alpine linux, its perfect for mail servers ;) 2016-04-12 11:44:11 so host your own ;) 2016-04-12 11:44:40 IP filtering has never been, is not, and will never be a reliable way to stop spam 2016-04-12 11:45:06 yeaa i am guessing the same 2016-04-12 11:45:06 host my own 2016-04-12 11:45:26 and it goes against net neutrality (I can't send e-mail from my home server, because my IP is dynamic so blacklisted by PBL) 2016-04-12 11:45:39 which gives ISPs way too much power 2016-04-12 11:47:14 oneinsect_: with opensmtpd its around 4 lines of config and then you have a very secure and robust mail server 2016-04-12 11:47:21 i highly recommend opensmtpd 2016-04-12 11:47:24 hmmm 2016-04-12 11:47:41 i have to get my own static ip 2016-04-12 11:48:13 yes, unless you have a public server somewhere where you can relay to 2016-04-12 11:49:50 i have a linode server 2016-04-12 11:49:56 probably i will run it there 2016-04-12 11:53:27 perfect :) 2016-04-12 11:54:06 my opensmtpd relay does around 20 000 mails in an hour and currently it needs around 120mb ram :) 2016-04-12 11:54:13 so hardware requirements are very low 2016-04-12 11:54:23 what about frontent 2016-04-12 11:54:25 honestly though, if a gmail server is listed on spamcop, it's bullshit 2016-04-12 11:55:42 it means that people who use gmail (i.e. a lot) can't send mail to servers who use spamcop lists? seriously? 2016-04-12 12:01:51 leo-unglaub: that giblib issue is interesting 2016-04-12 12:02:28 i am pretty sure that it was a leftover since uclibc days 2016-04-12 12:02:36 -depend = so:libc.so.0.9.32 2016-04-12 12:02:36 -depend = so:libm.so.0.9.32 2016-04-12 12:02:40 that is uclibc 2016-04-12 12:03:01 so something must have went wrong when we rsynced first musl build 2016-04-12 12:03:38 called it! automation snafu. :) 2016-04-12 12:05:43 ncopa: hmmm, that would mean there rsync script is horrible *g* 2016-04-12 12:12:41 or it could be that we were fixing the rsync script at the time 2016-04-12 12:12:51 or that something else went wrong 2016-04-12 12:12:55 i dunno 2016-04-12 13:18:13 leo-unglaub: re: mailservers. do you use anything between port 25 (or whatever) and your mailbox? 2016-04-12 13:18:31 (other than opensmtpd i mean) 2016-04-12 13:20:36 ScrumpyJack: yes, i use something similar to openbsd's spamd 2016-04-12 13:20:45 its self-written 2016-04-12 13:20:58 i have to ask my ex company if i can release it if you want 2016-04-12 13:21:30 http://man.openbsd.org/OpenBSD-current/man8/spamd.8 2016-04-12 13:21:57 i specially love there throtteling approach ... 2016-04-12 13:22:07 some connections randomly get throttled down 2016-04-12 13:22:17 spammers dont deal with that very well and the connection gets scrubbt 2016-04-12 13:22:27 but real mail servers dont have a problem with that 2016-04-12 13:22:40 its so simple, but amazing how much spam it keeps away 2016-04-12 13:23:31 some people also love https://rspamd.com/ 2016-04-12 13:23:34 i'm waiting for rspamd to hit community 2016-04-12 13:23:36 yes 2016-04-12 13:23:37 but for me thats way to much 2016-04-12 13:24:03 ScrumpyJack: it is alredy available 2016-04-12 13:24:16 https://pkgs.alpinelinux.org/package/testing/x86_64/rspamd 2016-04-12 13:24:27 yes, it's in edge 2016-04-12 13:24:34 I'm waiting for it go get to main 2016-04-12 13:24:54 ah, okay ... you are not on edge 2016-04-12 13:27:16 not on my mail server, no :) 2016-04-12 13:27:34 i'm using hermes, but it needs more work 2016-04-12 13:50:52 never heared of it ... but i googled it and they also do throtteling :) 2016-04-12 13:57:35 yes. and greylisting and spf tests. but the flow is all messed up imho. 2016-04-12 14:14:07 ncopa: you about? http://git.alpinelinux.org/cgit/aports/commit/?id=a3bc85c3cd190580d9a95526a4e89b6cdda0a9e0 removes ping and ping6 from bbsuid 2016-04-12 14:14:33 is that right? i get ping: ping is not a valid applet now 2016-04-12 14:15:31 it is correct 2016-04-12 14:15:56 i think you need to rm the ping symlink that is pointing to bbsuid 2016-04-12 14:16:01 and then run busybox --install 2016-04-12 14:17:40 then you need configure the ping_group_range 2016-04-12 14:17:51 in /proc/sys/net/ipv4/ping_group_range 2016-04-12 14:17:59 Socket(2) is restricted to the group range specified in 2016-04-12 14:18:00 "/proc/sys/net/ipv4/ping_group_range" (or the owner of CAP_NET_RAW). It 2016-04-12 14:18:00 is "1 0" by default, meaning that nobody except root may create ping 2016-04-12 14:18:00 sockets. Setting it to "100 100" would gain permissions to the single 2016-04-12 14:18:00 group, "0 65535" would enable it for the world. 2016-04-12 14:18:06 got it 2016-04-12 14:18:10 thanks 2016-04-12 14:19:57 that's neat by the way 2016-04-12 14:20:13 ncopa: does that mean that there's support for non-root raw sockets now? 2016-04-12 14:21:20 skarnet: correct 2016-04-12 14:21:32 no 2016-04-12 14:21:34 sorry 2016-04-12 14:21:54 it means it is support for icmp over datagram sockets 2016-04-12 14:22:06 so you dont need raw socket for ping 2016-04-12 14:22:12 yeah, so it's just for ping 2016-04-12 14:22:17 too bad 2016-04-12 14:22:47 ncopa: maybe this should be in a post-install script 2016-04-12 14:23:04 because it could cause a lot of confusion if people all of the sudden cannot ping anymore 2016-04-12 14:23:34 yes 2016-04-12 14:24:04 or post-upgrade 2016-04-12 14:24:58 yes, also possible 2016-04-12 14:30:12 leo-unglaub: what delay do you add when throttling? 2016-04-12 14:32:54 random, between 4 and 9 secounds 2016-04-12 14:33:57 that seems like a lot no? 2016-04-12 14:35:25 just looking at rspamd and it just seems huge :( 2016-04-12 14:37:27 35 configuration files, excluding the ones I have to create to override the default settings!!! 2016-04-12 14:38:19 leo-unglaub: pleaz can I haz your spamd 2016-04-12 14:56:55 ScrumpyJack: yeah, its fucking huge 2016-04-12 14:57:04 i am going to ask my old boss if its okay 2016-04-12 14:57:22 oh boy ... that c code must be 5 years old ... hehehe 2016-04-12 14:57:31 if ncopa would give me rust i could modernize it *g* 2016-04-12 14:57:32 hahahaha 2016-04-12 15:11:59 there's not much out there 2016-04-12 23:39:15 anyone have tried to install diskless mode into one partition of harddisk with GRUB as bootloader? 2016-04-12 23:51:07 I tried, but the /lib/modules seems missing. and one rc script running with usage failed. 2016-04-13 06:54:40 hey :) 2016-04-13 06:54:47 hey 2016-04-13 06:56:26 I have to provide IMAP to family. I sooo hate it. got dovecot working, but it writes so much crap to Maildir. Thank god i don't have to use it for me 2016-04-13 07:03:57 ScrumpyJack: What do you use for IMAP for you ? 2016-04-13 07:04:41 I don't. I use mutt locally on Maildir 2016-04-13 07:05:17 oh you make your smtp server deliver directly to Maildir 2016-04-13 07:06:09 no need for remote access then ? like mobile client and such ? 2016-04-13 07:11:39 what a day: customer wants to host a wordpress install and complains to me that the database host is localhost. localhost would be the default and therefore way to easy to guess for bots ... 2016-04-13 07:11:40 WTF 2016-04-13 07:11:55 :D 2016-04-13 07:12:13 i just replyed that he also could use 127.0.0.1 if thats lesser known in his opinion *g* 2016-04-13 07:12:25 there are some idiots out there ... un fucking beleavable 2016-04-13 07:12:26 hehehe 2016-04-13 07:13:30 I had a customer once complaining that my fils sharing platform was really slow 2016-04-13 07:13:40 file* 2016-04-13 07:13:48 tell them to use 2130706433 2016-04-13 07:13:55 uploading files was taking ages to her 2016-04-13 07:14:22 017700000001 also works 2016-04-13 07:14:26 indeed uploading 2GB with a GPRS connection is gonna take ages ... 2016-04-13 07:14:30 >_< 2016-04-13 07:16:48 coredumb: hahahahaha 2016-04-13 07:17:12 people have no idea how it works ... thats sooo sad because they really thing they know it 2016-04-13 07:17:24 but a good file sharing platform would make GPRS much faster 2016-04-13 07:18:00 i once had a customer who thought that all traffic would run thru the dns server ... litterally all her mails would traffel thru the dns server and because my dns servers are "slow" her company mail was slow 2016-04-13 07:18:01 *g* 2016-04-13 07:18:55 oh DNS it's always the DNS, blame DNS 2016-04-13 07:20:13 few years back we had a few consultants from Anderson, every time their stuffs didn't work, they would come to my office and blamed my DNS servers 2016-04-13 07:20:32 hahahahaha 2016-04-13 07:20:38 its always dns 2016-04-13 07:20:54 specially if they run windows servers with smb on it *g* 2016-04-13 07:40:01 correct, freaking WINS not my DNS, like I told you last time 2016-04-13 08:15:12 coredumb: I've never needed remote anything, but now everyone's concerned about gmail/hotmail mailboxes being cracked, so i've offered to host emails for family, meaning IMAP :( 2016-04-13 09:46:29 Hi! I've a question which would better fit in #alpine-linux, but nobody seems to know the answer over there. My question is: After installing alpine linux in sys mode on a Raspberry Pi 3, how can I use the new edge kernel? I could update all other packages from the version 3.3.0 branch to the edge branch without problems, but the new kernel in package "linux-rpi2" does not boot. I'm currently running 2016-04-13 09:46:35 edge software on top of a linux-rpi-4.1.20-r0 from version 3.3.0 2016-04-13 10:18:52 ScrumpyJack: gmail probably won't be cracked, its security is good. That's not the problem with gmail. 2016-04-13 10:19:00 The problem with gmail is data collection. 2016-04-13 10:19:27 But good luck using any service out there without getting your data collected anyway :/ 2016-04-13 10:30:25 hey :) 2016-04-13 10:52:30 ScrumpyJack: i did not know you are in #opensmtpd as well :) 2016-04-13 11:49:09 ncopa: is cdn mirror broken? 2016-04-13 11:51:02 not that i know? 2016-04-13 11:53:00 ncopa: ah i see, ngnix init scripts is a sep pkg 2016-04-13 12:11:53 is it possible to get a caddy-git or caddy-master package that gets build regulary from the bleeding edge? :) 2016-04-13 12:14:39 mosez: better to build it yourself if you need a git pkg. 2016-04-13 12:15:38 clandmeter: should just be interesting to get versions between releases from caddy. with caddy-git/caddy-master i just mean a more current version with latest features :) 2016-04-13 12:16:01 but it looks like i have to build it on my own to get my merged features from caddy running before the next release :) 2016-04-13 12:18:40 mosez: i know what you mean, but im not going to maintain it, and i dont think you will find (m)any devs that want to. 2016-04-13 12:19:35 if there are interesting new features mholt should release a new version. 2016-04-13 12:21:36 mosez: you can create your own Aports repository with quite little effort, see https://github.com/jirutka/aports#how-to-setup-your-own-repository 2016-04-13 12:21:37 mosez: then you can simple maintain more up-to-date packages built from the caddy master branch 2016-04-13 12:22:34 however, more systematic solution would be to convince the caddy maintainers to release their software more often… 2016-04-13 12:24:16 i am a caddy maintainer now :D 2016-04-13 12:24:29 but mholt wants to finish something different before the next release 2016-04-13 12:24:40 i think for now i will just build it on my own 2016-04-13 12:25:46 you are? for alpine? 2016-04-13 12:26:18 or you mean a caddy dev? 2016-04-13 12:26:34 no, for the project itself. i'm part of the maintainer/contributor team now ;) 2016-04-13 12:27:49 mosez: nice, then get my issue fixed! 2016-04-13 12:27:52 :) 2016-04-13 12:28:31 clandmeter: which one of the 56? :D 2016-04-13 12:29:24 XD 2016-04-13 12:29:34 can find only https://github.com/mholt/caddy/issues/649 which is closed :) 2016-04-13 12:30:01 Closing, due to inactivity. :D 2016-04-13 12:30:19 yes cause i think he didnt really get my point, and i was too busy fixing other things. 2016-04-13 12:30:48 probably due to my lack of clarity 2016-04-13 12:32:28 looks like i need to dig into the rewrite code :) 2016-04-13 12:32:58 i need a new caddy version to make my listing even more beautiful :P http://dl.webhippie.de/ 2016-04-13 12:34:46 and even there is still some space for improvement left as i want to inject folder specific files :) 2016-04-13 12:40:21 mosez: i need the revproxy code to support variable based redirects 2016-04-13 12:40:57 so our revproxy can redirect request to our redmine git repo to our cgit. 2016-04-13 13:50:06 i think my listing is even more beautiful: http://mirror.alpine.download.wayne-enterprises.company:8881/edge/main/x86_64/ 2016-04-13 14:10:02 leo-unglaub: no :D 2016-04-13 14:12:29 no 2016-04-13 14:12:46 however, I like the name wayne-enterprises :) 2016-04-13 14:15:42 idea: it would be useful if http://bugs.alpinelinux.org would automatically parse the package name from the bug title and assign it to the package maintainer 2016-04-13 14:25:45 can someone check the v3.3-stable x86 builder? 2016-04-13 14:25:50 clandmet1r, ^^ ? 2016-04-13 14:58:38 who is bruce carter? 2016-04-13 15:09:09 Hi, I'm trying to install alpine on harddisk partition as diskless mode boot by GRUB(not syslinux). Although it boot, but missing lib/modules, that let my network only display 'lo'. 2016-04-13 15:09:45 dlin: have you configured your network? 2016-04-13 15:12:26 fcolista: not me 2016-04-13 15:12:58 only limited ppl have access to builders. 2016-04-13 15:13:05 due to security 2016-04-13 15:14:39 it can not configure, because the lib/modules missing, no wlan0 & eth0 2016-04-13 15:16:14 This message is displayed during the boot process ( init many rc services) 2016-04-13 15:18:05 My /media/sda2 contains abox.apkovl.tar.gz.aes-256-cbc, apks/, boot/, cache/ , I guess I must missing something. I'm newbie in Alpine. 2016-04-13 15:18:56 dlin: what do you mean by installing to hd in diskless mode? 2016-04-13 15:19:26 I've successful install the Alpine in 'sys' mode. But, when I try to install as 'diskless' mode(in fact, I'm manually copy the contents from /media/cdrom into /media/sda2) 2016-04-13 15:19:50 modloop is what holds the modules 2016-04-13 15:20:16 init will mount it in didkless mode 2016-04-13 15:20:22 I want multiboot with sda1-ext4-ArchLinux, sda2-fat32-Alpine. 2016-04-13 15:21:19 i havent used grub in a long time 2016-04-13 15:22:09 make sure you provide the same options as we do in our syslinux.conf 2016-04-13 15:23:17 My grub.cfg works. That let me boot to a buggy diskless mode Alpine. Its setting is linux (xx)/boot/vmlinuz-grsec modlop=/boot/modloop-grsec modules=loop,squashfs,sd-mod-usb-storage quiet. I guess that is not the problem. 2016-04-13 15:23:50 Sorry, its modloop. 2016-04-13 15:24:53 I guess, maybe I missing some correct settings like /etc/apk/repository (the first line is /media/sda2/apks) 2016-04-13 15:25:43 my lbu.conf is 'LBU_MEDIA=sda2' 2016-04-13 15:26:16 dlin: can you start /etc/init.d/modloop ? 2016-04-13 15:28:07 It display the error message like "/lib/rc/sh/openrc-run.sh: eval: line 1: syntax error: unexpected "(" 2016-04-13 15:29:52 I know bash -x xxx can show script. But now, there is no bash. to debug the /etc/init.d/modloop script 2016-04-13 15:30:10 sh -x is the same 2016-04-13 15:30:35 but you can add --debug 2016-04-13 15:30:54 When I try to apk add bash, it shows many line of 'xxxx : temporary error (try again later) 2016-04-13 15:30:56 atleast that what i think it was 2016-04-13 15:31:14 why would oyu need bash? 2016-04-13 15:31:24 Or, is there correct process by manua 2016-04-13 15:31:56 my guess is 2016-04-13 15:32:01 I don't know ash well. How to debug the script, the first line of openrc-run.sh is correct, it is just #!/bin/sh 2016-04-13 15:32:19 your cmdline is different then normal and our init scripts dont like it. 2016-04-13 15:32:47 Or, is there a wiki to let me manual install into my sda2 with correct files and post-install processes? 2016-04-13 15:33:09 is there a reason you want to run it diskless? 2016-04-13 15:34:19 So, what's correct boot command line for diskless ? I want diskless is because my EEE PC's SSD seems often slow me down, I want all in RAM. 2016-04-13 15:35:34 dlin: check iso syslinux.conf, it should have it. 2016-04-13 15:36:38 dlin: we dont use grub here, so you are probably on your own. 2016-04-13 15:40:31 I double checked the syslinux & my gurb, both of them are same. 2016-04-13 15:41:24 so, if you use syslinux, what's the correct process to put into one harddisk partition? 2016-04-13 15:41:50 I just cp -a /mnt/cdrom/* into the /mnt/sda2 2016-04-13 19:27:11 I really love construbutors that don’t bother to look what code-style the project uses, ignores .editorconfig file and just enforce their code-style… 2016-04-13 20:03:40 oh lol, we really have .editorconfig 2016-04-13 20:04:13 tell it to all my APKBUILDs that use spaces for indentation 2016-04-13 20:05:12 :C 2016-04-13 20:05:18 jirutka redrum. 2016-04-13 20:47:41 jirutka1: Thanks for your time! I have learned a lot 2016-04-13 20:48:08 stwa: you’re welcome :) 2016-04-13 21:01:19 git.alpinelinux.org is down or repositories has been moved somewhere? 2016-04-14 07:05:04 morning. 2016-04-14 07:07:08 is there an issue with https://pkgs.alpinelinux.org/ ? or is it just me 2016-04-14 07:14:07 ScrumpyJack: works for me. 2016-04-14 07:14:30 but yesterday I was unable to connect with git.alpinelinux.org 2016-04-14 07:48:32 scadu: http://git.alpinelinux.org was down 2016-04-14 07:48:40 i have been cleaning up a bit 2016-04-14 07:48:50 apparently lighttpd had crashed 2016-04-14 07:48:53 i cound not figure out why 2016-04-14 07:49:17 disks was almost full so i have cleaned up things a bit 2016-04-14 08:38:35 hey :) 2016-04-14 08:47:58 hey 2016-04-14 08:49:42 ncopa: thanks for confirming. 2016-04-14 09:12:30 ScrumpyJack: pkgs always has issues. 2016-04-14 09:12:33 its by design. 2016-04-14 09:12:38 so you could call it a feature :p 2016-04-14 10:35:54 clandmeter: lol 2016-04-14 10:35:58 its not that bad ... 2016-04-14 10:36:33 just the gentoo, arch, debian, opensuse, redhat, voidlinux version is better 2016-04-14 10:36:47 see the positive .. at least its better than the mintlinux stuff *g* 2016-04-14 10:38:16 im open for suggestions 2016-04-14 10:38:28 leo-unglaub: and if you have them, now is the time :) 2016-04-14 10:39:23 clandmeter: just kidding :) i like the functionallity of it 2016-04-14 10:39:30 design is personal taste 2016-04-14 10:39:34 for me its to fancy 2016-04-14 10:39:42 but the features are great 2016-04-14 10:39:54 too fancy? 2016-04-14 10:40:03 default boostrap is too fancy? :) 2016-04-14 10:40:04 just that you choose % as a wildcard always throws me off *g* i am used to * :) 2016-04-14 10:40:12 i hate bootstrap *g* 2016-04-14 10:40:12 :) 2016-04-14 10:40:24 i am more of an oldschool plain text guy *g* 2016-04-14 10:40:26 wait for few days and see what happens :) 2016-04-14 10:40:44 I propose that clandmeter just snail mail us the apks every time we issue an apk upgrade :D 2016-04-14 10:40:57 http://www.openbsd.org/ <- best website design ever!! responsive, fast, low battery usage! 2016-04-14 10:42:57 ive rewrote most of pkgs now. 2016-04-14 10:43:07 so expect it to go online in few days 2016-04-14 10:43:13 nice :) 2016-04-14 10:43:15 leo-unglaub: there's a left-side menu and a picture 2016-04-14 10:43:18 but it will look the same 2016-04-14 10:43:19 that's bloated 2016-04-14 10:43:23 atleast almost :D 2016-04-14 10:43:45 scaleway doesn't accept paypal? 2016-04-14 10:43:47 skarnet: nah, one picture is fine as long as it is in a bad resolution 2016-04-14 10:43:57 and your beloved * will be there ;-) 2016-04-14 10:44:03 ... i am really a 4chan guy when it comes to websites *g* 2016-04-14 10:44:26 and its probably full of bugs, which are included for free of course 2016-04-14 10:44:36 leo-unglaub: skarnet.org is even more spartan :P 2016-04-14 10:44:54 yeah, i like the design! 2016-04-14 10:45:08 the design is called "I'm too lazy to learn CSS" 2016-04-14 10:45:47 isnt that what boostrap stands for? 2016-04-14 10:47:17 clandmeter: is the new version containing a OpenSearch description file ? 2016-04-14 10:48:10 :D 2016-04-14 10:48:30 hahaha 2016-04-14 10:48:59 2 options: A: i am good at guessing ... B: i hacked your laptop and read the changed myself and now i pretend to have a crystal ball *g* 2016-04-14 10:48:59 maybe a siteindex.xml or whatever its called what google wants. 2016-04-14 10:54:19 clandmeter: the sitemap.xml is not the opensearch description file 2016-04-14 10:54:33 that file tells the browser that you can use the browser search field to look thru the packages 2016-04-14 10:54:35 and search there 2016-04-14 10:55:28 can you file requests on the project tracker? 2016-04-14 10:55:36 then im sure i dont forget them. 2016-04-14 11:09:26 clandmeter: https://paste.debian.net/plainh/03616372 2016-04-14 11:09:38 put the first part in an xml file and store it on the server 2016-04-14 11:09:42 then add the link to the header 2016-04-14 11:10:05 and then the page will offer itself to add it as a browser search provider 2016-04-14 11:11:33 leo-unglaub: can you add it to https://github.com/clandmeter/aports-turbo/issues ? 2016-04-14 11:11:56 im trying to finish up some of the code now. 2016-04-14 11:12:11 is it possible to install a local apk? something like "apk add bitlbee-facebook-1.0.0-r0.apk" or do i need a APKINDEX.tar.gz? 2016-04-14 11:12:16 i hate github *g* 2016-04-14 11:12:19 but okay 2016-04-14 11:14:38 ok not... i get a UNTRUSTED signature error :) 2016-04-14 11:15:40 xsteadfastx: sure you cann install it without the apkindex 2016-04-14 11:15:59 but if the package is not an offician one its of course not signed by one of the keys 2016-04-14 11:16:12 if you trust the author, get his key and store it in your /etc 2016-04-14 11:16:39 i meant /etc/apk/keys 2016-04-14 11:16:57 or use apk add --allow-untrusted 2016-04-14 11:17:03 but i am not a big fan of that 2016-04-14 11:17:11 leo-unglaub: ah ok :) i will try that. its for a docker image i use and a new aport that is not accepted and merged yet 2016-04-14 11:17:18 but i want to use it already 2016-04-14 11:17:51 i see, it does not matte it its docker or not. the idea begind it is the same 2016-04-14 11:17:57 either use --allow-untrusted or get the key 2016-04-14 11:18:03 but i would recommend the key :) 2016-04-14 11:18:04 worked :) thank you 2016-04-14 11:18:10 you are welcome 2016-04-14 11:18:20 may i ask what package it is that is not in the official repos yet? 2016-04-14 11:18:27 maybe rust-lang??? ;) 2016-04-14 11:18:40 xsteadfastx: you can also create your own aports repository, incl. automatic building etc. see https://github.com/jirutka/aports#how-to-setup-your-own-repository 2016-04-14 11:19:06 leo-unglaub: http://patchwork.alpinelinux.org/patch/1851/ its a bitlbee facebook plugin. 2016-04-14 11:19:32 xsteadfastx: i know, i read it from the path above ... it was just an inside joke in case ncopa reads the backlog *g* 2016-04-14 11:19:43 jirutka: great. i will bookmark that :) 2016-04-14 11:20:07 hehe ok :) i hope it gets accepted and merged 2016-04-14 11:21:18 we should really remove depends_dev from the APKBUILD template, it apperantly confuses a lot people :/ 2016-04-14 11:23:58 and add builddir=${srcdir}/${pkgname}/${pkgver} to newapkbuild 2016-04-14 11:24:20 yeah 2016-04-14 11:24:20 builddir=${srcdir}/${pkgname}-${pkgver} even 2016-04-14 11:24:24 true... i dont get what depeneds_dev is for 2016-04-14 11:24:38 I’m gonna send patch 2016-04-14 11:24:48 this reminds me, @ncopa, are you here? 2016-04-14 11:24:54 im here 2016-04-14 11:24:58 have a chat to ncopa first 2016-04-14 11:25:05 arg, ninja'd 2016-04-14 11:26:11 ncopa: great, please open https://travis-ci.org/alpinelinux/aports/settings and enable “Build only if .travis.yml is present” … then it will not try to run build on v3.x branches that doesn’t contain .travis.yml 2016-04-14 11:26:41 jirutka: i already did 2016-04-14 11:26:58 ncopa: okay, great 2016-04-14 11:28:42 ncopa: I have a patch for abuild tool, but I’m not sure how patches that are not for aports should be send… there’s unfortunatelly no GitHub mirror and http://patchwork.alpinelinux.org seems to be just for aports; so I’ve attached it to the bug on http://bugs.alpinelinux.org/issues/5408 2016-04-14 11:33:28 ncopa: didnt you have some suggestions about pkgs.a.o? i remember something about flagged new version support. 2016-04-14 11:36:02 ncopa: and similar case, patch for the website (http://bugs.alpinelinux.org/issues/5386); who’s maintaining the website, i.e. who is the right person to discuss patches for website? 2016-04-14 11:44:10 jirutka: can you please test this patch on top of your abuild-sudo: http://tpaste.us/G0J6 2016-04-14 11:45:01 ncopa: sure, give me a minute 2016-04-14 11:45:07 we could make it simpler too 2016-04-14 11:45:24 name == NULL ? "(unknown)" : name 2016-04-14 11:46:05 in which situtation `getlogin()` may return null? 2016-04-14 11:46:12 on error 2016-04-14 11:46:16 for example 2016-04-14 11:46:25 if /etc/passwd is deleted 2016-04-14 11:46:28 i suppose 2016-04-14 11:46:32 huh, okay 2016-04-14 11:46:59 I’d personally prefer `name == NULL ? "(unknown)" : name`, it looks simpler and more straightforward 2016-04-14 11:47:04 yes 2016-04-14 11:47:59 http://tpaste.us/AaYo 2016-04-14 11:48:19 yeah 2016-04-14 11:48:22 lets do that instead 2016-04-14 11:49:13 i think i can test here 2016-04-14 11:49:23 i'll just create a dummy user 2016-04-14 11:49:26 su - dummy 2016-04-14 11:49:39 then remove it from /etc/passwd,shadow,groups 2016-04-14 11:49:45 and run abuild-apk 2016-04-14 11:51:56 ncdev-edge-x86_64:/home/dummy$ unset LOGNAME 2016-04-14 11:51:56 ncdev-edge-x86_64:/home/dummy$ abuild-apk 2016-04-14 11:51:56 abuild-apk: User (unknown) is not a member of group abuild 2016-04-14 11:53:19 pushed 2016-04-14 11:54:05 we havent setup dedicated mailing list and patchwork for abuild 2016-04-14 11:54:19 ncopa: great, thanks! you were faster than me :) 2016-04-14 11:54:26 patches can be sent to alpine-devel i think 2016-04-14 11:55:21 its me who maintains website code 2016-04-14 11:55:47 I’d be most happy for GitHub mirrors :) what do you think? 2016-04-14 11:55:50 and i think those patches can go to alpine-devel too, or as attachments on bugs.a.o 2016-04-14 11:56:02 dunno if its worth it 2016-04-14 11:56:27 what are downsides of it? 2016-04-14 11:56:58 more moving parts 2016-04-14 11:57:08 moving parts? 2016-04-14 11:57:21 i'd need configure up the mirror 2016-04-14 11:57:26 and the hook that uatopushes it 2016-04-14 11:57:31 autopushes* 2016-04-14 11:57:47 -> more complexity -> more things that can break 2016-04-14 11:57:59 + the time actually do it 2016-04-14 11:58:05 (not too much) 2016-04-14 11:58:38 the upside is that we'd have a backup 2016-04-14 11:58:49 well, I can do that… however, without access to alpine servers I’m limited just for pull strategy (i.e. periodically check for changes), but I suppose that it may be ok-ish for these not so frequented repositories 2016-04-14 12:01:08 jirutka: i added you to the alpinelinux proj on github 2016-04-14 12:01:36 ncopa: super, thanks! 2016-04-14 12:01:56 if you set up the initial repos and manually clone them i'll make them auto sync 2016-04-14 12:02:07 okay 2016-04-14 12:02:53 copy the key if possible: https://github.com/alpinelinux/aports/settings/keys 2016-04-14 12:03:25 I don’t have access to this page 2016-04-14 12:04:55 ncopa: I can’t create repos in AlpineLinux organization: You don’t appear to have permission to create repositories for this organization. Sorry about that. 2016-04-14 12:05:32 https://github.com/organizations/alpinelinux/settings/member_privileges → Allow members to create repositories for this organization 2016-04-14 12:06:16 added you to the owner role 2016-04-14 12:06:27 huh, that’s great responsibility! 2016-04-14 12:06:56 yes. dont mess it up :) 2016-04-14 12:07:33 okay, thanks for your trust :) 2016-04-14 12:09:10 also, dont push directly to aports on github 2016-04-14 12:09:24 but i assume you understood that :) 2016-04-14 12:09:56 yes, I understand that 2016-04-14 12:11:03 pushed your website patch: http://wwwtest.alpinelinux.org/ 2016-04-14 12:11:55 also to prodcution 2016-04-14 12:12:26 great, now it works over HTTPS without problem 2016-04-14 12:12:33 thanks! 2016-04-14 12:14:20 clandmeter: sorry, i missed your msg. 2016-04-14 12:14:31 yes, i have a request 2016-04-14 12:14:58 i'd like to get a list of packages that are flagged as out-date and not yet fixed 2016-04-14 12:15:08 already done 2016-04-14 12:21:46 jirutka: why you want our repos on github? 2016-04-14 12:21:54 asking for a friend... 2016-04-14 12:21:55 :) 2016-04-14 12:22:33 so I can easily send patches, via pull requests 2016-04-14 12:23:37 it makes them more accessible for contributors, easily to discover them etc. 2016-04-14 12:24:22 i'm ok with it 2016-04-14 12:24:49 i think we should be careful though, so we don't end up with a hard dependency on github 2016-04-14 12:24:57 pull requests are very good for code-review, you can see example here https://github.com/alpinelinux/aports/pull/53 2016-04-14 12:27:56 the upstream will be still on git.a.o and we’ve disabled issues on GitHub, so I think it should be okay 2016-04-14 12:33:26 ncopa: GitHub has one unpleasant limitations for deploy keys – they cannot be reused in multiple repositories; it seems that the only option how to use the same deploy key is to add it to user profile as a ordinary SSH key 2016-04-14 12:34:59 mailingslist are good for code-review as well besides you don't need a github account to send a patch to the mailling list 2016-04-14 12:35:27 yes 2016-04-14 12:35:28 you need to be a member on the mailing list 2016-04-14 12:35:44 nmeum: and beside you cannot attach comment to the specific line of the code and track changes… 2016-04-14 12:36:11 jirutka: you can comment on specific line of the code 2016-04-14 12:36:27 how? 2016-04-14 12:36:50 i mean comment on specific lines of the patch 2016-04-14 12:37:04 you quote the email and write your answer under the quote 2016-04-14 12:37:18 well, but there’s significant difference in clarity and comfort 2016-04-14 12:37:37 disadvantage is when you comment on a line, and somebody changes the commit. i beleive your comment is lost. 2016-04-14 12:37:57 no, i think comment remains 2016-04-14 12:38:07 it’s not, see these folded comments on https://github.com/alpinelinux/aports/pull/53 2016-04-14 12:38:07 are you sure? 2016-04-14 12:38:28 i think ive lost sdome commets on zfs repo 2016-04-14 12:38:30 jirutka: kernel devs has done code reviews on email long before github existed 2016-04-14 12:38:38 2016-04-14 12:38:48 lunch? 2016-04-14 12:38:52 you moved timezone? 2016-04-14 12:39:22 ncopa: I know, also people has travelled long before plains existed ;) 2016-04-14 12:40:25 jirutka: github is just a preference, i also like it, thats why we have a github org. 2016-04-14 12:41:12 jirutka: i was also looking in moving documentation to it, but i failed and never looked back due to time... 2016-04-14 12:41:21 github is convenient, but it should remain *convenient* and not *mandatory* 2016-04-14 12:41:34 of course, it wouldn’t be wise for a project to be hard-depend on GitHub 2016-04-14 12:42:22 or any commercial platform 2016-04-14 12:42:31 yeah 2016-04-14 13:26:20 I don't really think github is a good tool for code review 2016-04-14 13:26:45 it's good because it's popular, so it really keeps the contribution bar low 2016-04-14 13:26:58 vs sending patches via e-mail in correct format 2016-04-14 13:28:13 that’s not the point… sending pull requests that are not uter crap is also hard for many people :/ I see totally messed formatting very often 2016-04-14 13:28:35 sure, but we have the same on mailing list 2016-04-14 13:28:40 so yes, I know what's my point is, really 2016-04-14 13:28:53 the worst are people who uses IDE for everything, so they even don’t know what they’re committing… 2016-04-14 13:29:37 gerrit is terrific, but screws up any sane git workflow 2016-04-14 13:30:05 I haven’t tried that 2016-04-14 13:30:33 we’ve tried https://reviewable.io/ on one project, but I’m not fan of it 2016-04-14 13:30:44 well, I guess it's fine as long as you don't use branch view 2016-04-14 13:31:33 https://paste.xinu.at/7T1/ 2016-04-14 13:33:38 oh my… 2016-04-14 13:33:43 however, this screen looks like tig 2016-04-14 13:48:38 yeah, my point is how merge workflow looks 2016-04-14 13:48:45 I want to modify /etc/init.d/modloop to fix the problem of 'grub' support. I use 'lbu ci' to commit my modification, but it failed, is there any tip? 2016-04-14 13:48:54 like untar and tar? 2016-04-14 14:01:58 ncopa/fabled/anyone: so, are APKBUILDs supposed to behave correctly with cross-compilation in build(), or can I assume that build() is always native and we'll have a crossbuild() at some point ? 2016-04-14 14:10:34 dlin: i think apk considers /etc/init.d as 'program' data instead of config 2016-04-14 14:10:50 so it is intentionally excluded from apkovl 2016-04-14 14:10:53 ncopa: https://github.com/scaleway/image-alpine/issues/38#issuecomment-209961992 related to the scaleway issue where groupadd prompts for a password... it's fixed with passwd -d root, than i dont get any prompt -.- 2016-04-14 14:11:06 dlin: who does your diff look like? 2016-04-14 14:11:33 so shadow somehow requires a root password even when i'm already root if root have set a password 2016-04-14 14:11:54 ncopa/fabled/anyone: so, are APKBUILDs supposed to behave correctly with cross-compilation in build(), or can I assume that build() is always native and we'll have a crossbuild() at some point ? 2016-04-14 14:11:59 whoops 2016-04-14 14:12:05 mt, sorry 2016-04-14 14:13:30 (but you can still answer the question, even if I don't want to spam it! :D) 2016-04-14 14:15:01 mosez: just guessing, but could it be possible that there are 2 groupadd in different paths' and that PATH is set to pick the "wrong"? 2016-04-14 14:29:53 ncopa: nope 2016-04-14 14:50:42 https://github.com/scaleway/image-alpine/issues/38#issuecomment-209966053 so shadow@testing seems to have an issue if the root user got a password :( 2016-04-14 14:51:07 it's not possible to use configuration management tools on alpine if the root user got a password set :( 2016-04-14 14:54:30 i need a way to reproduce it to be able to fix it 2016-04-14 14:55:18 ah, i can just set a root password in a docker image to reproduce it? 2016-04-14 14:57:51 chris|: may I pull the s6-linux-utils maintainership from under your feet, or do you want to keep it? 2016-04-14 15:01:14 ncopa: exactly 2016-04-14 15:02:07 fwiw, i have an internal ticket for fixing useradd/groupadd http://lists.busybox.net/pipermail/busybox/2016-February/083907.html 2016-04-14 16:11:31 so for submitting patches to packages 2016-04-14 16:11:45 those get sent to http://lists.alpinelinux.org/alpine-aports/ 2016-04-14 16:11:47 ? 2016-04-14 16:32:00 mepholic: yes, but you can post them via github as well 2016-04-14 16:32:11 or just drop link to your fork/patches on IRC 2016-04-14 16:32:38 anything consumable by git works 2016-04-14 16:37:17 !@#$^% of *&%^#$%^ of @#$#$% of mail systems and spam databases 2016-04-14 16:37:51 mail.alpinelinux.org[74.117.189.114] said: 554 5.7.1 Service unavailable; 2016-04-14 16:37:51 Client host [82.216.111.38] blocked using dnsbl.sorbs.net; Currently Sending Spam 2016-04-14 16:38:01 this is my ISP's server. 2016-04-14 16:38:21 FUCK YEA my ISP's mail server is blocked because it's sending spam. 2016-04-14 16:38:51 and if I try to send to alpine-aports from my own server, it's blocked because it's a dynamic IP so it's listed in the PBL ! 2016-04-14 16:39:16 HOW THE HECK am I supposed to send mail ? 2016-04-14 16:40:21 Would someone kindly edit mail.alpinelinux.org's MTA config with a chainsaw ? 2016-04-14 16:48:04 skarnet: you should get yourself a proper mail server under your control :) 2016-04-14 16:48:28 This is just not a viable solution 2016-04-14 16:48:58 I do have a proper mail server under my control, it just so happens to be behind a dynamic IP 2016-04-14 16:49:35 and since mail.alpinelinux.org looks up the PBL, it will reject mails from it 2016-04-14 16:49:44 the other solution is to use my ISP's server 2016-04-14 16:50:09 and OH SURPRISE IT'S NOT WORKING EITHER BECAUSE STUPID CENTRALIZED SPAM DATABASE 2016-04-14 16:50:58 I *could* work around that by bunnyhopping via the skarnet.org server - with a little luck it won't be blacklisted - but I don't want to. 2016-04-14 16:51:05 ACTION puts on his ray-bans to counter the CAPS-LOCK glare 2016-04-14 16:51:07 How do other people do it? 2016-04-14 16:51:34 Does everyone have a fixed IP address? Is it a hard requirement to be able to send mail to alpine mailing-lists? Puh-lease. 2016-04-14 16:52:54 skarnet: are you doing this behind a DSL? 2016-04-14 16:53:17 i rent various hosts - all fixed IPv4 or IPv6 2016-04-14 16:54:12 I got fiber, but my ISP is stuck in 2002 and doesn't provide static IPs unless I fork $$$, which I won't because it sucks 2016-04-14 16:54:25 and changing ISPs is not an option, because they ALL suck 2016-04-14 16:55:05 dude, grab yourself a scaleway bare metal host, they are like 2.99 a month 2016-04-14 16:55:24 You don't get it. I do have a server already. THAT'S NOT THE POINT. 2016-04-14 16:55:38 (yes, I'm furious, in case you can't tell) 2016-04-14 16:56:08 the point is that neither my home server nor my ISP's server should be rejected. 2016-04-14 16:56:09 is this fibre to your home? 2016-04-14 16:56:15 yes. 2016-04-14 16:56:23 Both are perfectly legitimate servers. 2016-04-14 16:57:13 It is a MISTAKE for mail.alpinelinux.org to check the PBL or the DNSBL, and it's an even worse mistake to check both! 2016-04-14 16:57:24 sure they are. again, grab yourself a bare metal box and be done with it. you shouldn't be bothering with that stuff at home 2016-04-14 16:58:01 could you stop patronizing me when I'm raising a legitimate and SERIOUS problem ? 2016-04-14 16:58:20 sorry dude, i didn't mean to sound patronizing 2016-04-14 16:58:47 i'm just trying to change your mind 2016-04-14 16:58:49 then stop suggesting a "bare metal box". I do have a hosted server, with a fixed IP, which I could use as a relay. 2016-04-14 16:58:56 I DO NOT WANT TO. 2016-04-14 16:59:03 I should not have to do this. Nobody should. 2016-04-14 16:59:28 If a contributor wants to send patches to Alpine, the answer should not be "grab yourself a bare metal box with a fixed IP and configure a mail server". EVER. 2016-04-14 16:59:29 but your ISP won't give you a fixed IPv4 2016-04-14 16:59:37 so fucking what? 2016-04-14 16:59:45 net neutrality, do you speak it? 2016-04-14 17:00:02 fixed IP makes my life easier. no dynamic stuff 2016-04-14 17:00:35 Of course it makes life easier. That is still no excuse for enforcing it. 2016-04-14 17:01:15 I already raised the alarm about the PBL a few months ago. 2016-04-14 17:01:32 Nobody did anything, apparently. It was still OK since I could send mail via my ISP's server. 2016-04-14 17:01:38 I'm not enforcing it. you have a choice, I'm just trying to change your mind and go for least resistance 2016-04-14 17:01:43 But today this one is blocked too? this is ridiculous. 2016-04-14 17:02:20 No, I'm not going for least resistance. Alpine is supposed to do things right. Doing things right means not blocking my IP and my ISP's IP. 2016-04-14 17:02:59 Can't you see how braindead this is? My IP is blocked because it's dynamic and they say "use your ISP's server". OK, I'm using my ISP's server, and now IT IS BLOCKED TOO. 2016-04-14 17:03:53 The solution is easy. Edit the mail server's config and nuke that PBL and DNSBL shit - this is NOT how you stop spam. 2016-04-14 17:04:29 Telling contributors "Get your own mail server" IS NOT a solution. 2016-04-14 17:04:46 skarnet: you can bang your head against as many walls as you like. one thing you have going for you, is this is Alpine Linux and the crowd might do something about the MTA, but imagine if you were dealing with google, or MS. you'd go for the path of least resistance 2016-04-14 17:06:11 One thing I can tell: NEVER have Google or Microsoft told me to get my own mail server 2016-04-14 17:06:28 NEVER have I been in a situation where it was impossible to send mail to them 2016-04-14 17:07:10 and when they've done demented stuff, it was stuff I could entirely do without. 2016-04-14 17:07:11 dude, it's just something to consider. I'm not telling you to do anything! 2016-04-14 17:08:11 Good. Now what *I* would like you to do is to think about the absurdity of the situation and acknowledge I'm in the right. 2016-04-14 17:09:21 skarnet, you IP usually doesn’t appear on mail black lists for nothing… so how that happend? 2016-04-14 17:09:41 can't you guys ever stop blaming the victim? 2016-04-14 17:09:51 I’m not blaming you! 2016-04-14 17:10:16 I’m just asking how your IP address and IP address of your ISP appeard on the black list 2016-04-14 17:11:05 and of course, black list are stupid solution, but the problem is that there’s really no sane solution against mail spam; there are just bad and even worse solutions 2016-04-14 17:11:06 i get it dude, it sucks. we all have our own opinions on "spam control", much ink has been spilt over it. it's easier to get round other people's implementations rather than convince them you would do something differently and why. 2016-04-14 17:11:11 My IP is dynamic so it's on the PBL, a fascist list that says "if you're a dynamic IP, you can't send mail to us. If your ISP can't give you a static IP, too bad, you can't have your own mail server. Happy We Shit On Net Neutrality Day!" 2016-04-14 17:11:43 Ah, I understand now 2016-04-14 17:12:00 As for my ISP's mail server IP, how the fuck would I know how it ended listed by DNSBL? they suck, they relayed some spam at some point, someone reported them, it can go so fast 2016-04-14 17:12:22 And your ISP’s IP is on the black list because it’s operated by idiots that simply set open relay? 2016-04-14 17:12:41 s/the\ fuck/on\ earth/ 2016-04-14 17:13:13 I'm angry enough to swear, and you will not censor my swearing 2016-04-14 17:13:45 jirutka: I have no idea. No, it's not open relay, but some customer has probably sent spam, and bam, here we are 2016-04-14 17:13:51 well, could you propose some better solution to block spam than these black lists? 2016-04-14 17:13:57 Yes, I can. 2016-04-14 17:14:24 Forget blacklists, host a spamfilter. Such as rspamd or otherwise. 2016-04-14 17:14:56 Or just implement greylisting. It's cheap, easy and effective. 2016-04-14 17:15:04 Do both for bonus points! 2016-04-14 17:15:31 do you have some experience with rspamd? I’ve tried spamd (not sure if it has something in common) and not happy with it. 2016-04-14 17:15:32 But I will take craploads of spam over false positives *any day*. 2016-04-14 17:15:46 no, I don't. But I'm sure we can find someone who does. 2016-04-14 17:15:47 graylisting is the worst thing ever! 2016-04-14 17:16:01 nope. Can't be worse than centralized lists. 2016-04-14 17:16:08 yes it can 2016-04-14 17:16:51 imagine that you register on a site and waiting for fucking validation email so you can finally login… 2016-04-14 17:16:56 no. We have a database of legit senders. We can whitelist those. New ones get a confirmation request ONCE. 2016-04-14 17:17:15 jirutka: yes? I do that all the time. What's the big deal? 2016-04-14 17:17:54 skarnet I disagree: I am of the opinion that email filtering is a form of editting/censorship and should not be undertaken by the transport host. up to the end user to deal with it 2016-04-14 17:18:16 I'm kidding, but there is just no right opinion 2016-04-14 17:19:50 Maybe there's no right opinion, but blocking both my and my ISP's servers is *obviously* wrong 2016-04-14 17:20:26 and how are you going to get around that? 2016-04-14 17:21:10 By finding the admin of mail.alpinelinux.org and annoy the shit out of them until they remove that PBL check or at least the DNSBL check 2016-04-14 17:21:26 (let's imagine that you're not talking to the friendly bunch at alpine-devel who can do something about the MTA) 2016-04-14 17:21:54 ScrumpyJack: I'm not interested in your hypothetives. 2016-04-14 17:22:42 if it wasn't Alpine, I'd say "fuck this, they're just too moronic, there's no hope, I'll go do something else" 2016-04-14 17:22:50 with Alpine, I do still have hope. 2016-04-14 17:24:11 But if we can't fix this, I will have a very interesting story to tell to the Void Linux guys. :P 2016-04-14 17:24:13 Well, I'm sure you'll be able to convince someone. If it were me, I wouldn't do blacklist lookups. 2016-04-14 17:27:05 nevertheless, I'd still try to convince you to run a small mail server hosted somewhere for private stuff like this. I can't supply the same level of service at home than I can hosted somewhere 2016-04-14 17:29:13 ScrumpyJack: I could. And what are you going to tell to the next guy in my situation ? "sure you can contribute, it only costs 2.99€ a month and a shitload of time to configure a mail server, and then you're good to go!" ? 2016-04-14 17:49:26 who’s administrating Alpine’s mail server? 2016-04-14 17:49:58 barthalion: perfect 2016-04-14 17:57:15 jirutka: that can be nangel 2016-04-14 17:59:34 barhalion: Nathan Angelacos? 2016-04-14 18:00:23 yep 2016-04-14 18:01:12 hm, then skarnet is doomed… I’ve opened issue about invalid TLS certs few months ago and still without any response :/ 2016-04-14 18:01:16 I'm waiting for confirmation before sending him a mail with today's log archive :P 2016-04-14 18:02:24 write to him directly? 2016-04-14 18:02:34 I don't remember when was the last time I visited bugtracker 2016-04-14 18:02:51 bugtracker sends mails 2016-04-14 18:03:01 he’s assigned to these issues, so he should know about them 2016-04-14 18:03:25 if you say so… 2016-04-14 18:03:56 well, if he not disabled email notifications 2016-04-14 18:04:12 I’ll try to write him an email 2016-04-14 18:04:21 notifications are perceived in different way by regular people than human-made message 2016-04-14 18:05:10 even if both bring the same meaning 2016-04-14 18:06:25 I didn't know Alpine maintainers were regular people 2016-04-14 18:06:35 sad but true 2016-04-14 18:06:45 we even eat sometimes 2016-04-14 18:06:48 XD 2016-04-14 18:06:56 I'll believe it when I see it 2016-04-14 18:08:58 jirutka1: about these invalid certs… link plz 2016-04-14 18:13:06 barthalion: http://bugs.alpinelinux.org/issues/4982, http://bugs.alpinelinux.org/issues/4983, http://bugs.alpinelinux.org/issues/4981, 2016-04-14 18:21:57 welp, I 2016-04-14 18:22:09 welp, I'm afraid jirutka1 is right. 2016-04-14 18:22:28 about what? 2016-04-14 18:23:38 about certs. 2016-04-14 18:23:40 https://paste.xinu.at/lrB/ 2016-04-14 18:39:12 skarnet: may I ask you for a help? it’s C-related 2016-04-14 18:58:47 ACTION now accepting flames 2016-04-14 18:59:27 nangel: hi, could you please look at http://bugs.alpinelinux.org/issues/4982, http://bugs.alpinelinux.org/issues/4983, http://bugs.alpinelinux.org/issues/4981 ? 2016-04-14 18:59:50 hey guys 2016-04-14 18:59:52 https://github.com/alpinelinux/aports/pull/57 2016-04-14 19:00:13 I guess the PR name is awful and not standard 2016-04-14 19:00:17 sorry about that :| 2016-04-14 19:01:05 jirutka: yup. thanks. 2016-04-14 19:02:09 fixed 2016-04-14 19:04:24 jirutka: hey 2016-04-14 19:04:35 sorry, this is my first APKBUILD :P 2016-04-14 19:05:06 mepholic: hi 2016-04-14 19:05:32 no problem, it looks good except these two things I’ve commented 2016-04-14 19:05:37 should I kill this PR and remake it? 2016-04-14 19:05:59 no, just rebase and force push to the same branch, this’ll update the PR 2016-04-14 19:06:01 also, should the pkgrel only be bumped when it's being updated from a merged package? 2016-04-14 19:06:15 hm okay 2016-04-14 19:06:50 https://github.com/alpinelinux/aports/pull/57#discussion-diff-59773027 2016-04-14 19:06:50 about pkgrel, the base value is 0, so in the first commit it should be 0 and bumped in your second commit to 1 2016-04-14 19:07:11 okay, so it should get bumped per commit 2016-04-14 19:07:14 cool 2016-04-14 19:07:33 as far as that comment though 2016-04-14 19:08:02 it must be bumped with every modification of the package files, incl. APKBUILD, when pkgver is not bumped 2016-04-14 19:08:16 ah 2016-04-14 19:08:19 I get it 2016-04-14 19:08:26 :) 2016-04-14 19:09:10 one more thing 2016-04-14 19:09:12 20:05:58 no, just rebase and force push to the same branch, this’ll update the PR 2016-04-14 19:09:18 not very familiar with rebasing 2016-04-14 19:09:30 can you explain this process please 2016-04-14 19:10:04 I’m not sure how to do it when splitting changes into few commits; does anyone else now? 2016-04-14 19:11:15 nangel: hey dude. can you check postfix on mail.alpinelinux.org and disable PBL lookups for a bit? If nothing breaks (spam levels don't change), perhaps you could leave it. 2016-04-14 19:11:19 hm 2016-04-14 19:11:22 jirutka: shouldn' 2016-04-14 19:11:36 t I only need to rebase if HEAD on master has incremented? 2016-04-14 19:11:54 mepholic: he needs to edit his commits 2016-04-14 19:13:06 git rebase -i HEAD~2 # this will open an editor 2016-04-14 19:13:06 rewrite “pick” to “edit”, save & close the editor 2016-04-14 19:13:06 now your on the first commit, make changes, then git add . and git rebase --continue 2016-04-14 19:13:06 now your on the second commit, make changes, then git add . and git rebase --continue 2016-04-14 19:13:06 you’re done, now just git push --force 2016-04-14 19:13:59 warning: never ever do this when pushing to a public master, unless you know what you’re doing; for branches used only in PR it’s fine 2016-04-14 19:14:02 ah..... 2016-04-14 19:14:05 :D 2016-04-14 19:14:08 cool 2016-04-14 19:14:21 yeah 2016-04-14 19:14:23 obviously 2016-04-14 19:14:25 rofl 2016-04-14 19:14:27 ? 2016-04-14 19:14:33 I can see that ending terribly 2016-04-14 19:14:41 why? 2016-04-14 19:14:46 in a public master 2016-04-14 19:15:03 yeah, it’s rewriting the history 2016-04-14 19:15:10 it must be used with caution 2016-04-14 19:16:40 it’s acceptable when you’re sure that no one else pulled from that branch or when you warn others, but generally it should be avoided, it may cause a lot of trouble 2016-04-14 19:16:59 mepholic: what do you mean with “shouldn’t”? should he squash these commits into a single commit? 2016-04-14 19:23:17 mepholic: ignore the last message, I’ve forgotten who is who XD 2016-04-14 19:23:36 lololol 2016-04-14 19:23:37 ok 2016-04-14 19:23:40 It's rebased 2016-04-14 19:23:48 and corrected for the most part I think jirutka 2016-04-14 19:23:50 :) 2016-04-14 19:23:53 ty for the help! 2016-04-14 19:25:18 if you want me to squash em together 2016-04-14 19:25:19 let me know 2016-04-14 19:25:41 you’re welcome 2016-04-14 19:26:29 I don’t have rights for pushing into aports, we must ask ncopa if he wants it squashed 2016-04-14 19:26:44 the freeswitch perl thingy? 2016-04-14 19:26:59 yes please sqush them 2016-04-14 19:27:27 mepholic: then you can just reset pkgrev to zero 2016-04-14 19:27:39 okay 2016-04-14 19:27:42 rofl 2016-04-14 19:27:47 hold on 2016-04-14 19:28:01 I'm just gonna squash this fucker into one commit 2016-04-14 19:28:10 mepholic: yeah, sorry for that excercise with rebase :) 2016-04-14 19:28:13 also I'm adding the explicit checks back 2016-04-14 19:28:18 no, thank you actually 2016-04-14 19:28:22 taught me something new :) 2016-04-14 19:28:44 let me wait for travis lol 2016-04-14 19:28:50 need a smoke anyways 2016-04-14 19:29:00 ncopa: if I make few different changes in APKBUILD, should I always squash them into single commit, or it depends…? 2016-04-14 19:35:21 jirutka: ncopa 2016-04-14 19:35:25 squashed the commits 2016-04-14 19:35:31 and fixed the pkgrel 2016-04-14 19:35:43 I forgot to add the explicit " || return 1" back 2016-04-14 19:35:52 cause I do agree with ncopa that it con't hurt 2016-04-14 19:37:41 well, putting ":" on every second line doesn’t hurt either, but it also doesn’t make any sense… however, if ncopa is fine with it, then let it be :) 2016-04-14 19:39:00 i think it depends 2016-04-14 19:39:53 okay, so I can still apply a rule of thumb about single-purpose commits :) 2016-04-14 19:39:54 if the second is a fix of the first commit, then it makes sense to squshe them 2016-04-14 19:40:51 if there are a chance that things break, so that you potentially might want revert one change, but not the other, then it may makes sense to split them 2016-04-14 19:40:56 it was a fix? well, I’m probably too tired, I didn’t notice it 2016-04-14 19:41:24 I always prefer to squash in this case, but I overlooked that he just fixed mistake from the previous commit 2016-04-14 19:41:44 i think it was removal of unecessary dep that was introduced in his own commit 2016-04-14 19:41:51 i am tired too... 2016-04-14 19:42:14 re || return 1 on last line 2016-04-14 19:42:37 i have added that, incase someone wants extend the future later 2016-04-14 19:42:39 hello folks, i discovered that znc was built without iconv support, so it cannot recode irc messages on-the-fly :( it is a problem. 2016-04-14 19:42:44 20:41:43 <@ncopa> i think it was removal of unecessary dep that was introduced in his own commit 2016-04-14 19:42:47 yep 2016-04-14 19:43:03 i didn't realize that apk actually intelligently determined the runtime deps 2016-04-14 19:43:15 until after I committed it the first time 2016-04-14 19:43:17 but i dont have strong opinion either way re || return 1 or not on last line in func 2016-04-14 19:43:23 ok 2016-04-14 19:43:25 in that case 2016-04-14 19:43:32 I'll leave it as is 2016-04-14 19:43:53 but from now on, I think I'm just going to explicitly add || return 1 2016-04-14 19:44:00 even if it is the last line 2016-04-14 19:44:39 ideally we should run it as `set -e` so we could remove all || return 1 2016-04-14 19:45:17 ncopa: I’ve looked at this yesterday 2016-04-14 19:46:04 i dont want implement it til after v3.4 2016-04-14 19:46:58 ncopa: I think that it’d be best to polish some issues reported by shellcheck and then do few modifications required for set -e 2016-04-14 19:48:14 I didn’t read complete source yet, but from what I’ve read it seems that it will be quite easy to add set -e 2016-04-14 19:48:42 however, I’m afraid about compatibility with current abuilds… 2016-04-14 19:49:01 theoretically it shouldn’t break anything, but… 2016-04-14 19:50:01 ncopa: cool 2016-04-14 19:50:03 :) 2016-04-14 19:50:10 maybe it’d be better to make it optional, introduce some variable that will be set in newer abuilds, so we can be sure that it’ll not break anything 2016-04-14 19:50:22 ncopa: btw, I'm running this freeswitch package in production currently 2016-04-14 19:50:28 everything appears to work 2016-04-14 19:50:57 i'm slightly worried about the freetdm package, I'll test that a bit 2016-04-14 19:51:07 but it was definitely broken in the 1.6.6 package 2016-04-14 19:51:35 apparently, mjerris of the FS project fixed the build issue with 1.6.6 2016-04-14 19:53:05 ncopa: is abuild always run using ash? or it may be run with different shell? set -e behaves quite differently in every shell 2016-04-14 19:54:10 oh, you kno what 2016-04-14 19:54:35 it might have made sense to do the subpackage addition in separate commit 2016-04-14 19:54:38 but doesnt matter 2016-04-14 19:54:47 i'll take it as it is 2016-04-14 19:55:00 i can upgrade 1.6.7 in stable branch if needed 2016-04-14 19:55:10 that'd be cool :) 2016-04-14 19:55:13 except u 2016-04-14 19:55:15 h 2016-04-14 19:55:22 I haven't tested it with the er 2016-04-14 19:55:35 mepholic: if you think we should upgrade to 1.6.7 in 3.3-stable, then please file a bug on bugs.a.o with target 3.3.4 2016-04-14 19:55:39 so i dont forget 2016-04-14 19:55:44 okay 2016-04-14 19:55:49 otherwise i will forget 2016-04-14 19:55:57 ugh 2016-04-14 19:56:10 getting 502 bad gateway from thepkgs.alpinelinux.org 2016-04-14 19:56:30 okay, better now 2016-04-14 19:56:32 :P 2016-04-14 19:56:35 jirutka: i think we can assume ash 2016-04-14 19:56:42 ^ this happens to me quite often 2016-04-14 19:56:48 502 on pkgs.a.o 2016-04-14 19:56:57 many apkbuilds will break with dash 2016-04-14 19:57:00 bash might work 2016-04-14 19:57:03 so yeah ncopa 2016-04-14 19:57:10 the only thing I haven't really tested it with 2016-04-14 19:57:13 is this https://pkgs.alpinelinux.org/package/main/x86_64/acf-freeswitch 2016-04-14 19:57:41 i assume 1.6.6 -> 1.6.7 does not introduce any big revolutions 2016-04-14 19:58:05 okay 2016-04-14 19:58:22 i need to go 2016-04-14 19:58:26 many thank 2016-04-14 19:58:37 see ya! 2016-04-14 19:58:38 i'll apply it tm morning 2016-04-14 19:58:42 ty sir 2016-04-14 20:00:21 jirutka: sorry, I took a break to cool off. Sure, ask away :) 2016-04-14 20:01:08 nangel: what ScrumpyJack said. Could you please deactivate PBL lookups from the mail server? My very strong opinion, backed up by experience, is that PBL hurts more than it helps. 2016-04-14 20:01:46 hey skarnet: welcome back 2016-04-14 20:01:53 ish 2016-04-14 20:01:58 skarnet: I have to go now, the building I work in is closing at 22.00 and doorman is quite angry when I’m here late 2016-04-14 20:02:08 skarnet: I’ll write later 2016-04-14 20:02:16 np 2016-04-14 20:02:55 (praise that doorman who encourages a healthier work-life balance!) 2016-04-14 20:03:32 skarnet: I have disabled zen.spamhaus.org RBL for you. 2016-04-14 20:03:50 I'll put it back if we start getting spam flooded though. 2016-04-14 20:04:38 nangel: what other anti-spam stuff is setup on postfix m.a.org?? 2016-04-14 20:04:53 nangel: thanks. Is it possible to be specific in your config? i.e. only disable PBL and not the other blacklists? PBL is the most offensive. 2016-04-14 20:05:43 reject_non_fqdn_recipient, 2016-04-14 20:05:43 reject_unknown_recipient_domain, 2016-04-14 20:05:43 permit_mynetworks, 2016-04-14 20:05:43 permit_sasl_authenticated, 2016-04-14 20:05:43 reject_unauth_destination, 2016-04-14 20:05:43 check_sender_access hash:/etc/postfix/sender_checks, 2016-04-14 20:05:45 DNSBL is also bad - my ISP is on it, gmail (!) is on it - but it's less of a political issue 2016-04-14 20:05:45 reject_rbl_client bl.spamcop.net, 2016-04-14 20:05:47 reject_rbl_client dul.dnsbl.sorbs.net, 2016-04-14 20:05:49 reject_rbl_client dnsbl.sorbs.net, 2016-04-14 20:05:51 check_policy_service inet:127.0.0.1:5525 2016-04-14 20:05:53 the paste of doom 2016-04-14 20:05:53 # Clamsmtp 2016-04-14 20:05:55 content_filter = scan:[127.0.0.1]:10025 2016-04-14 20:06:00 rip #alpine-devel 2016-04-14 20:06:10 :) 2016-04-14 20:06:21 shorter than I expected :( 2016-04-14 20:06:31 :/ 2016-04-14 20:06:33 hasn't been groomed recently 2016-04-14 20:07:54 and it's already public: http://wiki.alpinelinux.org/wiki/ISP_Mail_Server_3.x_HowTo 2016-04-14 20:09:21 ok, let me experiment... 2016-04-14 20:09:50 skarnet: yes, and yahoo gets on it quite frequently as well 2016-04-14 20:10:26 yahoo terrible 2016-04-14 20:10:31 +100 2016-04-14 20:10:37 I ate "is" 2016-04-14 20:10:39 wasn't tasty 2016-04-14 20:10:56 no - you just talk like a yahoo user. :) 2016-04-14 20:11:01 heh 2016-04-14 20:11:06 it recently blocked one of Arch's IPs 2016-04-14 20:11:11 would you consider adding spf + greylist + nolisting + ipfilter block taiwan+north corea and see what happens? (monitor RBL hits and if they slow to a halt, remove the lookups) 2016-04-14 20:11:11 for no good reason 2016-04-14 20:11:22 e-mail to postmaster@ are apparently redirected to /dev/null 2016-04-14 20:11:39 postmaster should work; if not I'll fix that 2016-04-14 20:11:50 I mean yahoo, not us 2016-04-14 20:11:57 we're cool I guess 2016-04-14 20:12:00 oh. yeah 2016-04-14 20:13:18 ScrumpyJack - I can look into it. I've got a project to move the mail server; but getting TLS certificates needs to happen first, and that's already 3 months old. :( 2016-04-14 20:13:21 apparently, dul.dnsbl.sorbs.net is the one that implements the PBL 2016-04-14 20:13:47 nangel: do you want help with that? 2016-04-14 20:13:51 could you please deactivate that one? I can live with the other ones 2016-04-14 20:14:46 skarnet - done. You have dial-up? I have a dynamic IP and am not blocked. 2016-04-14 20:15:12 I have fiber. With a dynamic IP. The whole range of IPs that my ISP provides is in the PBL. 2016-04-14 20:15:23 testing. 2016-04-14 20:15:25 sigh... 2016-04-14 20:15:46 my ISP blocks port 25 2016-04-14 20:15:46 I added zen back in; so if you are still blocked I'll need to remove it again 2016-04-14 20:16:05 one way to avoid getting blacklisted... 2016-04-14 20:16:20 yup, still blocked 2016-04-14 20:16:23 by zen 2016-04-14 20:16:24 any bl are terrorists - they want money for listing-out innocent mail-servers 2016-04-14 20:16:45 kunkku - I never even try port 25 anymore. port 587 or bust. 2016-04-14 20:17:22 Time to destroy the whole mail infrastructure and rebuild it from scratch -_- 2016-04-14 20:17:49 irc & email is for really old people like us 2016-04-14 20:17:56 skarnet updated 2016-04-14 20:18:01 thanks, trying again... 2016-04-14 20:18:10 kunkku: speak for yourself :) 2016-04-14 20:18:28 ACTION should probably just whitelist skarnet, and keep the RBL's 2016-04-14 20:18:37 then if we get spam, we'll know who it came from. :D 2016-04-14 20:18:37 the young use "smartphone apps" for communication 2016-04-14 20:18:56 I'm in early twenties, man 2016-04-14 20:18:59 don't bury me yet 2016-04-14 20:19:02 nope, still blocked, this time by dnsbl.sorbs.net. Why so much redundancy? 2016-04-14 20:19:31 Those 3 domains obviously access the same lists... why are they different >.> 2016-04-14 20:19:39 also, no, the point is not just to whitelist me 2016-04-14 20:19:50 the point is to make an infrastructure that will work for any Alpine contributor 2016-04-14 20:20:30 I understand that 2016-04-14 20:20:53 but there is "the correct solution" and the "help a guy out today" 2016-04-14 20:21:07 skarnet: you don't need to use "Alpine contributor" as leverage :) you have the attention now :)) 2016-04-14 20:21:31 I'll whitelist you today and work on the infra later, ok? 2016-04-14 20:21:50 ScrumpyJack: I don't care about the attention, I care about fixing stuff 2016-04-14 20:22:28 nangel: sorry, no, not ok. I know all to well what "I'll just do that workaround now and work on the real fix later" means, and you know it too. 2016-04-14 20:22:58 I'd rather keep my case unfixed and keep nagging you every week. 2016-04-14 20:23:03 fair enough. 2016-04-14 20:23:52 you got to hand it to skarnet for persistance 2016-04-14 20:26:57 nangel: so, can we work on a plan to fix this? Not necessarily tonight, since you're now aware that it's a real issue impacting real people. 2016-04-14 20:28:12 I've seen dynamic ranges blacklisted quite thoroughly. I'm rather surprised it worked for you so far. 2016-04-14 20:29:21 It never worked for me, but I was basely using a quick fix, i.e. using my ISP's server. 2016-04-14 20:29:32 If you can't get sensible PTR for your mailserver, you're going to have a bad time. 2016-04-14 20:29:48 Until today when I was reminded that the quick fix wasn't reliable. 2016-04-14 20:31:22 CcxCZ: Oh, because PTR checks are a thing too? how many buggy techniques can people pile on top of one another just to make things more annoying? 2016-04-14 20:33:15 (even if I could get a sensible PTR, that wouldn't guarantee shit. Case in point: my ISP's server, gmail, Yahoo.) 2016-04-14 20:33:20 There's SPF & DKIM, iirc google now refuses mail if you don't have records for those. 2016-04-14 20:33:56 email is *PAIN* 2016-04-14 20:34:20 CcxCZ: I don't think so. spf is advisory only, not blocking if non existant 2016-04-14 20:34:30 spf is bad anyway. 2016-04-14 20:34:46 dkim a bit less so, but is heavier. 2016-04-14 20:34:58 ScrumpyJack: that's what the standard says, not what google does 2016-04-14 20:35:07 (as usual) 2016-04-14 20:35:45 skarnet, actually i think spf is not that bad. it enables *me* to say who can send mail for *me* 2016-04-14 20:36:03 I find SPF rather okay myself 2016-04-14 20:36:07 e-mail is horrible, because it's quick fix on top of quick fix all the way down, and at the bottom there's just an old misdesigned protocol that's the cause of all the problems. 2016-04-14 20:36:33 http://david.woodhou.se/why-not-spf.html 2016-04-14 20:36:49 skarnet: I'm open to suggestions 2016-04-14 20:37:34 nangel: I suggest using a spam filter and doing away with the blacklists entirely. 2016-04-14 20:37:49 that certainly requires more processing power. 2016-04-14 20:38:04 I'm ready to contribute financially to that, if necessary. 2016-04-14 20:40:24 that article is BS, that's why there's distinct envelope and header senders 2016-04-14 20:41:34 I want to ask you what part of the article, and discuss it, but I've argued enough for one day - can we do that tomorrow? 2016-04-14 20:42:39 Sure. And the problem it solves (more or less) is backscatter. 2016-04-14 20:51:14 CcxCZ: does toybox build on musl? 2016-04-14 20:51:47 it would be good to have a choice 2016-04-14 20:51:54 ScrumpyJack: that's the main platform actually 2016-04-14 20:52:38 Or more precisely, the author maintains aboriginal which is musl+toybox+toolchain 2016-04-14 20:53:37 email is not horrible when it was over uucp, it's horrible over IP :p 2016-04-14 20:54:47 (It's supposed to be the most minimal self-hosting distribution) 2016-04-14 20:56:14 I'll read about it. Baking tomorrow's bread at the mo 2016-04-14 21:02:29 CcxCZ: facy having a bash as packaging toybox for AL? :P 2016-04-14 21:02:44 s/facy/fancy 2016-04-14 21:05:09 uhh, while I wouldn't mind doing that I kinda lack any semblance of free time for that kind of stuff in the near future 2016-04-15 08:05:36 ncopa: i created the munin pull request on github. can you help me? you wrote you can not verify the checksums, but even if i download the files by hand and compute the checksum with sha512sum it gives me correct results. how is this possible? the commit builds just fine 'on my machine' 2016-04-15 08:06:18 ncopa: abuild checksum and abuild verify also works 2016-04-15 08:25:34 ncopa: https://bugs.alpinelinux.org/issues/5419 2016-04-15 08:25:42 think I did that right 2016-04-15 09:41:08 Anyone here know why I am getting lxc-start: cgfs.c: cgroup_rmdir: 207 Resource busy - cgroup_rmdir: failed to delete /sys/fs/cgroup/... I see references to AppArmour so I assume its something to do with it being an unprivaliged container or grsec? 2016-04-15 09:41:24 unprivileged* 2016-04-15 10:09:29 youw blowed it up! 2016-04-15 10:09:34 also apparmor? 2016-04-15 10:09:46 does alpine use apparmor as well? 2016-04-15 10:10:04 also is the grsec RBAC enabled? 2016-04-15 10:15:38 Sorry I used the wrong channel, its not devel related 2016-04-15 14:04:36 leo-unglaub: around? 2016-04-15 14:13:57 clandmeter: yes sir! 2016-04-15 14:14:07 check pkgs 2016-04-15 14:14:25 and dont tell me you dislike the interface, i know that already :p 2016-04-15 14:15:01 No item found... 2016-04-15 14:15:12 ah, maybe because i was looking for rust ;) 2016-04-15 14:15:27 hehe 2016-04-15 14:16:05 looks good, responds very fast! 2016-04-15 14:16:49 cached from nginx 2016-04-15 14:17:09 yeah, please can you add "server_tokens off;" to the nginx config 2016-04-15 14:17:32 turbo -> nginx(cache) -> caddy 2016-04-15 14:18:37 maybe there should be an arch preselected 2016-04-15 14:18:49 otherwize there are always 3 times the same items visible 2016-04-15 14:19:57 yes, im thinking of it. 2016-04-15 14:20:06 not sure what is the clearest solution. 2016-04-15 14:21:01 i think use the most popular arch as default ... and removing the filter is pretty easy thanks to the x next to the selection 2016-04-15 14:21:22 so i think its kind of a noob filter. if people dont know how to remove the filter, then they are banned form irc as well *g* 2016-04-15 14:21:53 and the * is back, just for you. 2016-04-15 14:39:59 leo-unglaub: what’s the situation around rust? you’ve written rust package and waiting to be merged? 2016-04-15 14:45:01 what happend to pkgs.alpinelinux.org? "%" is not a wildcard anymore? 2016-04-15 15:00:28 Hi. I asked a few days ago how to get the linux-rpi2 kernel booting on an Rasberry Pi 3. It turned out that the .dtb files (device tree blobs) were not automatically updated 2016-04-15 15:01:35 These .dtb files are located at "/usr/linux-$pkgver-$pkgrel-$flaver/" (linux-4.4.5-0-rpi2 in my case), but they should be stored on the boot partition of the Pis sdcard 2016-04-15 15:03:40 As also the linux kernel must be stored on the boot partition and is installed at "/boot/", where I have mounted my boot partition on my sys install, I would suggest to install the dtb files to "/boot/" instead of "/usr/lib/...". This was kernel update with apk will be fully automatically and painless, as long as the boot partition is mounted under /boot, which I believe to be the default anyway 2016-04-15 15:04:07 *This way 2016-04-15 15:36:06 Hi, i just send some patches using git-email 2016-04-15 15:36:37 however I see on http://patchwork.alpinelinux.org/project/aports/list/ it shows the names of every contributor, but for me it shows my email address for some reason 2016-04-15 15:41:15 onodera_: did you configure your name in git? 2016-04-15 15:41:45 $ git config --global user.name "John Doe" 2016-04-15 15:41:45 yeah I did 2016-04-15 15:41:45 $ git config --global user.email johndoe@example.com 2016-04-15 15:43:04 yup, I just double checked my gitconfig and it's in there 2016-04-15 15:45:49 weird 2016-04-15 15:46:29 brb gotta restart x 2016-04-15 16:00:57 https://aur.archlinux.org/cgit/aur.git/tree/?h=firefox-esr-privacy 2016-04-15 16:01:26 ^ there are some really interesting privacy related patches in this repo that could (and should imo) be added to the alpine firefox ports :) 2016-04-15 17:03:58 http://unlicense.org/ <- what should I put in license for packages using this license/? 2016-04-15 23:02:26 I've finally installed AL, but haven't set it up fully yet. I'll be writing down some of my AL experiences on wiki: http://wiki.alpinelinux.org/wiki/User:Przemoc/Notes 2016-04-16 10:49:28 I'm trying to use a diskless alpine on 1GB RAM PC. I installed the 3.3.3 firefox, but it is very strange, the address bar can not display any word, it seems displayed as '1 pt' size font. 2016-04-16 10:49:54 Sorry, I mean alpine 3.3.3 main repository. 2016-04-16 10:57:15 I found the 1GB RAM only left 500MB for / diskspace. It is soon full. The apk command become very strange. I don't know how to proper reduce the ram disk usage 2016-04-16 10:59:04 I type 'apk add midori', it installed but failed by disk full. Since that, 'apk del dillo' will trying to do the 'apk add midori' again. 2016-04-16 12:11:39 that problem sounds like apk doesn't properly rollback failed operations 2016-04-16 12:12:16 so it has to complete them before moving on to other things 2016-04-16 12:13:01 (have I ever mentioned that Unix primitives make it hard to implement filesystem transactions? :/) 2016-04-16 12:30:03 skarnet: please, go ahead, I don't mind at all. maintainership is not something I care deeply about 2016-04-16 12:33:02 chris|: ok. Anyway the changes are in my local repo, and I can't send them until the mailserver config issue gets fixed, so they're patiently waiting. 2016-04-16 12:33:55 skarnet: you can send them via GitHub as a pull request ;) :P 2016-04-16 12:35:26 jirutka: I'd use my own static-IP mail server before I'd stoop that low! ;) 2016-04-16 12:35:39 XD 2016-04-16 12:36:22 skarnet: or you can just print them and send using local postal service :P 2016-04-16 12:36:48 https://www.ietf.org/rfc/rfc1149.txt 2016-04-16 12:37:03 sorry, can't help you there, I may only commit to community and testing 2016-04-16 12:37:44 skarnet: I love IETF april jokes, my favourite is Hyper Text Coffee Pot Control Protocol :) 2016-04-16 14:01:00 2016-04-16 14:01:02 please git pull git://git.alpinelinux.org/user/fab/aports 2016-04-16 14:01:12 just a couple of updates 2016-04-16 14:01:14 thansk 2016-04-16 15:50:12 Any one know how do list package installed by manual? (excluding the dependency, like Arch's pacman -Qe) 2016-04-16 15:50:44 cat /etc/apk/world ? 2016-04-16 15:54:28 Thanks, that's what I want. 2016-04-16 15:54:54 np 2016-04-16 15:55:12 I use 'apk info openbox', but it displayed two 'openbox', is this a bug? or that's caused by my /etc/apk/repository? 2016-04-16 15:56:10 this probably means that you have two repositories that provides openbox, probably different versions 2016-04-16 15:57:04 My /etc/apk/repositories contains "/media/sda2/apks", "http://.../alpine/v3.3/main", "http://.../alpine/v3.3/community" 2016-04-16 15:57:37 what’s the output of apk info openbox? 2016-04-16 15:58:02 But, both of them are openbox-3.6.1-r0 but with different installed size, one is 1622016, the other is 1617920. 2016-04-16 15:58:24 My /etc/apk/repositories also contains pin repository of testing & edge. 2016-04-16 15:59:17 the /media/sda2/apks is your own local repo? 2016-04-16 16:00:03 I use diskless install, that's /media/sda2 is one partition. I boot alpine from there. 2016-04-16 16:02:16 the /media/sda2/apks is copied from v3.3.3 alpine ISO file. 2016-04-16 16:06:23 Or, should I remove the /media/sda2/apks ? I've made a symoblic link /etc/apk/cache to /media/sda2/cache. 2016-04-16 16:28:53 Is there a command to query openbox-3.6.1-*.apk 's detail info? 2016-04-16 16:29:23 I want to find out which two apk files cause the 'apk info openbox' display two packages. 2016-04-16 18:21:30 hey :) 2016-04-16 18:22:41 ho 2016-04-17 02:50:34 clandmeter: what do you think about packaging https://github.com/Eloston/ungoogled-chromium as alternative (more privacy-oriented) variant of Chromium in Alpine? 2016-04-17 05:27:32 I like the idea 2016-04-17 07:49:36 already behind upstream releases 2016-04-17 07:49:53 can't wait til it gets abandonded 2016-04-17 07:50:36 if not using *optional* Google integration in some software is too much for someone, I doubt we can help them in any way 2016-04-17 07:53:47 so, up to you… but we shouldn't put it further than in community 2016-04-17 08:07:22 barthalion: the problem is that this "integration" isn't optional as you think 2016-04-17 08:08:03 or, as they want you to think about it 2016-04-17 08:08:54 sure, more 3rd party patches to everyone 2016-04-17 08:09:14 of course 2016-04-17 08:09:41 but unfortunately those patches are mostly removing 3rd party stuff, instead of adding something new 2016-04-17 08:09:49 it's not you who's going to keep it building, so no burden for you 2016-04-17 08:10:31 as usually, testing is the way to go 2016-04-17 08:11:17 I don't have any problems with trying to maintain that, as I did with some other projects, but I wanted to ask actual chromium maintainer about opinion before actual preparation 2016-04-17 08:14:33 also keep in mind this repo also comes with building recipe for debian, so there's not too much work to do except checking those patches against our musl patchset, and we haven't to reinvent a wheel 2016-04-17 08:14:57 ncopa: do you remember who has access to gitolite conf? just you and me? 2016-04-17 08:58:01 hey :) 2016-04-17 09:49:05 sup 2016-04-17 12:14:40 clandmeter: I've got "[web.lua] Error in RequestHandler, thread: 0x41d04fd8 is dead." when I tried to flag dejagnu package as old. (http://pastebin.com/AY6Rqu3w) 2016-04-17 12:37:47 ncopa, do you know how java is bootstrapped in alpine ? i'm trying since 3 days to get gnu classpath compiled with different ecj and gcc versions to use with jamvm, but to no avail 2016-04-17 13:05:21 why 'apk info -w /lib/firmware/iwlwifi-100-5.ucode' could not find owner package? 2016-04-17 13:12:57 because -w is not a short version of --who-owns 2016-04-17 13:13:20 -w is short for --webpage 2016-04-17 13:16:03 if you're lazy, then you can also go the long and slow way with: apk info -vL '*' | grep iwlwifi-100-5.ucode 2016-04-17 13:16:19 dlin: ^ 2016-04-17 13:25:10 Sorry, I make typo '-W', I mean I want to check the owner, but those file can not found the owner 2016-04-17 13:26:17 I've just do apk info -vL '*' | grep ucode and got none. 2016-04-17 13:32:01 dlin: does this file exist on the filesystem? 2016-04-17 13:32:30 apk checks local database 2016-04-17 13:36:01 Yes, it is in my disk. You can check any file in your /lib/firmware directory 2016-04-17 13:40:29 realpath /lib/firmware/iwlwifi-100-5.ucode -> /.modloop/modules/firmware/iwlwifi-100-5.ucode 2016-04-17 14:19:26 who here uses xfce as there desktop environment? 2016-04-17 16:27:54 greetings earthlings 2016-04-17 21:21:45 does awall has any support for custom rules? is there any recommended way of defining them? 2016-04-18 00:39:01 ncopa: when trying to build using the alpine-iso tools http://imgur.com/a/mj7S1 2016-04-18 00:42:10 it appears .default_boot_services no longer has any effect? 2016-04-18 00:42:55 I'm afraid ncopa will got about hundred notifications when he come back here 2016-04-18 01:38:39 I'm working on deadbeef 0.7.0 -> 0.7.1 update. Additionally I added some missing deps to make it able to play WAV, AAC and streams via curl. But while deadbeef itself is very modular, I wonder it's worth to split it into many other subpackages with separate plugins 2016-04-18 01:39:03 like deadbeef-aac, deadbeef-flac, deadbeef-curl, deadbeef-dumb 2016-04-18 01:41:37 skrzyp: if you're able to do it, its ideal, if those can be stripped away it keeps things lighter 2016-04-18 01:45:52 Adran: can I use standard shell functions in APKBUILDs? I think it'll be better to do something like "for x in curl aac wavpack flac mp3 ; do $x() { mv $pkgdir/usr/lib/deadbeef/$x.so $subpkgdir/usr/lib/$x.so } ; done" for 2016-04-18 01:47:09 ~/ % apk info deadbeef -L | grep -i 'lib/.*\.so' | wc -l 2016-04-18 01:47:11 40 2016-04-18 01:47:18 except that snippet won't work, yes 2016-04-18 01:47:32 that's not an actual code 2016-04-18 01:47:33 we usually have some helper functions in case of many split packages 2016-04-18 01:48:05 nice, wiki doesn't say anything about them 2016-04-18 01:48:33 documentation lacks some stuff, unfortunately 2016-04-18 01:49:14 but as long as it's valid POSIX-compliant shell script, we're cool 2016-04-18 01:49:34 any example of apkbuild with those helper functions? 2016-04-18 01:54:06 look at claws-mail 2016-04-18 01:56:49 thanks 2016-04-18 02:01:45 Abuild will track dependencies of main package AFTER splitting it into subpackages, right? 2016-04-18 02:04:34 yes, for each package 2016-04-18 02:05:05 if it doesn't detect something, it's not linked dynamically 2016-04-18 02:06:15 exactly 2016-04-18 02:07:05 I know about this dependence tracking schema a bit, mainly from Void, but they have other approach on subpkgs 2016-04-18 08:15:42 hey :) 2016-04-18 08:36:36 hey :) 2016-04-18 08:38:33 morning 2016-04-18 08:43:09 ScrumpyJack: have you seen the avr related patches? 2016-04-18 08:57:13 stwa: yes, that's great dude. 2016-04-18 08:57:21 thank you very much 2016-04-18 10:22:32 I got segfault calling rc-status today. :) I had `find stuff -type f -exec md5sum {} + >MD5SUMS` running (around 2M files) and calling rc-status during that led to grsec: denied resource overstep by requesting 4096 for RLIMIT_CORE against limit 0 for /sbin/openrc[rc-status:5420] 2016-04-18 10:25:28 ncopa, which tood we have for ruby package? 2016-04-18 10:25:31 *tool 2016-04-18 10:25:43 something like apkbuild-cpan? 2016-04-18 10:31:54 fcolista: we dont have 2016-04-18 10:32:14 atleaast not that i know of. 2016-04-18 10:32:43 i remember a stuff made by kunkku about ruby 2016-04-18 10:32:55 but i don't remember what actually was used for 2016-04-18 10:33:03 probably upgrade the packages we already have 2016-04-18 10:33:08 and not for building new ones 2016-04-18 10:36:47 unreliable segfault reproduction: cat $(seq 1 1041) 2>/dev/null & rc-status # maybe it's not common thing to have such long cmdline, but I think rc-status shouldn't die because of that 2016-04-18 12:23:29 clandmeter: ping 2016-04-18 12:23:57 dong 2016-04-18 12:24:48 clandmeter: do you know why on pkgs the only select shortened is the branch one? 2016-04-18 12:24:54 all other ones are visible 100% 2016-04-18 12:24:58 but that one is not 2016-04-18 12:25:20 http://i.imgur.com/6aDpJJ7.png 2016-04-18 12:25:24 yes because the items in the select's are not wide enough to match 2016-04-18 12:25:40 and there is no min-with 2016-04-18 12:25:45 width 2016-04-18 12:26:31 ah okay 2016-04-18 12:26:36 not that important, just noticed it :)) 2016-04-18 12:26:37 before i had latest-stable as v3.3 2016-04-18 12:26:44 then the issue is gone 2016-04-18 12:26:58 another thing i noticed is that on the "flagged" list there is not package name tooltip describing the package 2016-04-18 12:27:13 right its a bug 2016-04-18 12:27:23 kind of 2016-04-18 12:27:27 as its not a package name 2016-04-18 12:27:34 its origin name 2016-04-18 12:27:46 but still it has a description. 2016-04-18 12:28:11 Im also trying to fix the message 2016-04-18 12:28:31 as hyperlinks are outside the div 2016-04-18 12:28:35 need to break them. 2016-04-18 13:50:02 leo-unglaub: where did you get the opensearch template from? 2016-04-18 13:51:51 clandmeter: the top of my head 2016-04-18 13:52:05 i looked at the s pec 2016-04-18 13:52:12 thats why there is propobly a typo in there :) 2016-04-18 13:52:22 i just typed it down in the github editor 2016-04-18 13:59:25 leo-unglaub: looks like the param and method have changed in recent spec. 2016-04-18 14:01:05 ah there are extensions... 2016-04-18 14:08:23 clandmeter: possible ... i am off that web development shit for years ... 2016-04-18 14:08:37 ACTION is still ashamed of having done web development for such a long time 2016-04-18 14:08:45 not as bad as apps, but still shit *g* 2016-04-18 14:53:45 ncopa, are you here? 2016-04-18 14:54:42 hi 2016-04-18 14:54:46 hi 2016-04-18 14:55:00 could you please look at https://github.com/alpinelinux/aports/pull/55 and https://github.com/alpinelinux/aports/pull/56 ? 2016-04-18 14:55:18 patches: https://github.com/alpinelinux/aports/pull/56, https://patch-diff.githubusercontent.com/raw/alpinelinux/aports/pull/56.patch 2016-04-18 15:18:34 ncopa: you have a chance to look at that strangeness with the alpine-iso? 2016-04-18 15:19:00 i'm stumped 2016-04-18 15:30:58 clandmeter: perfect 2016-04-18 15:30:59 works! 2016-04-18 15:38:46 any one know how to login console with default user without password? 2016-04-18 15:39:15 dlin: type the username and enter 2016-04-18 15:40:45 Ha, I just want skip the password input, and directly login into X windows. 2016-04-18 15:41:26 In my Arch Linux, it is through systemd settings then use a .bashrc to automatic enter X window 2016-04-18 15:42:01 you can use the .xinitrc 2016-04-18 15:42:53 https://wiki.archlinux.org/index.php/Automatic_login_to_virtual_console 2016-04-18 15:43:42 .xinitrc is later called. How to skip and jump into the first console login. 2016-04-18 15:45:48 seems find some tip here https://wiki.manjaro.org/index.php?title=Autostart_X_at_Login_-_OpenRC_%26_SystemD 2016-04-18 15:45:49 dlin: You want to change /etc/inittab ... like: tty2::respawn:/bin/su - -- USERNAME -l -c '/usr/bin/xinit /dev/null 2>&1' 38400 tty2 2016-04-18 15:46:11 Ya, thanks skazz. 2016-04-18 15:47:36 ncopa: thanks for returning from holidays :P 2016-04-18 15:50:39 looks like you pulled in all updates 2016-04-18 16:05:22 any tutorial wiki about ACF? I skim http://wiki.alpinelinux.org/wiki/Category:ACF, but don't know what's proper. 2016-04-18 16:08:46 In Arch Linux, I often use 'yaourt -C' to compare config file and manual merge them. What's the proper way in Alpine? 2016-04-18 16:09:25 update-conf maybe? 2016-04-18 16:32:16 Getting a odd issue with LXC, I think its something to do with grsec... I can execute "lxc-start -n NAME -F -- /bin/echo foo" the first time it works, second time it does not, and then it does again 2016-04-18 16:39:40 Are cgroup instances of a container suppose to still be there after the container has stopped? 2016-04-18 16:39:53 I have 8 folders for the same container in /sys/fs/cgroup/cpu/lxc 2016-04-18 16:46:14 ncopa, an OT question: what's the best approach of having a docker image interactive? e.g. a script that run inside a container that takes variable from stdin in the host 2016-04-18 16:47:18 docker run -it ... 2016-04-18 16:47:55 uh 2016-04-18 16:47:56 right 2016-04-18 16:47:58 stupid me 2016-04-18 17:09:06 ncopa: you around? 2016-04-19 00:39:32 ncopa: looks like Docker is broken out of the box, or I did something wrong 2016-04-19 00:39:47 skrzyp: did you disable the grsec stuff? 2016-04-19 00:39:57 Adran: I'm on vanilla 2016-04-19 00:40:12 running docker service fails and drops this line in docker.log: 2016-04-19 00:40:14 "level=fatal msg="Failed to connect to containerd. Please make sure containerd is installed in your PATH or you have specificed the correct address. Got error: exec: \"docker-containerd\": executable file not found in $PATH" " 2016-04-19 00:45:51 Adran: any suggestions? 2016-04-19 00:47:27 i can debug a bit later 2016-04-19 00:47:51 ok. 2016-04-19 02:44:07 What if I want to automatically install -doc and -dev subpkgs on current system? 2016-04-19 05:47:59 morning 2016-04-19 05:48:11 fabled: you familiar with alpine-iso/ 2016-04-19 05:48:20 yes 2016-04-19 05:48:32 fabled: been getting this strange issue http://imgur.com/a/mj7S1 2016-04-19 05:49:54 that looks more like you are trying to install kernel package on tmpfs install 2016-04-19 05:50:19 use update-kernel to update kernel on boot media of tmpfs installs 2016-04-19 05:50:35 fabled: sure, but thats with this configuration of alpine-iso 2016-04-19 05:52:25 fabled: https://gist.github.com/Ttech/6c574cea4182df8903f629da438870d1 2016-04-19 05:52:26 normally our builders are full disk install. so that might be a bug with the script. 2016-04-19 05:52:46 those errors occur onl when a ovl is specified 2016-04-19 05:52:50 ovl 2016-04-19 05:53:03 anything in dmesg? 2016-04-19 05:53:05 and when starting setup-alpine 2016-04-19 06:01:26 fabled: also keep getting the error unsatisfiable constraints 2016-04-19 06:02:03 those can be due to packing errors, and not recompiling all packaging when some base library gets abi change 2016-04-19 06:02:36 i fully rm the tmp directories 2016-04-19 06:03:03 and then rebuild 2016-04-19 06:03:18 i just don't get why there are those read only errors 2016-04-19 06:37:41 Adran, skrzyp: package is broken, nothing to do with grsec 2016-04-19 06:38:36 barthalion: is there a patch or such? 2016-04-19 06:38:45 ncopa said he's working on it 2016-04-19 06:38:53 skrzyp: yes docker is currently broken 2016-04-19 06:38:56 and yes i am working on it 2016-04-19 06:38:59 :) 2016-04-19 06:39:11 docker now needs 4 bins 2016-04-19 06:39:22 runc and containerd 2016-04-19 06:39:31 and containerd-shim 2016-04-19 06:39:39 they are in different git repos/tarballs 2016-04-19 06:39:42 runc is part of docker package now ? 2016-04-19 06:39:53 yes and no 2016-04-19 06:40:08 its not part of the docker git repe 2016-04-19 06:40:10 repo 2016-04-19 06:40:17 its a separate "package" 2016-04-19 06:40:24 but official docker bundles it 2016-04-19 06:40:49 interesting 2016-04-19 06:41:07 https://github.com/docker/docker/blob/master/Dockerfile#L250 2016-04-19 06:41:41 in theory i could make them separate packages and make docker depend on them 2016-04-19 06:41:56 but i think they (want) depend on specific version/git commit 2016-04-19 06:42:15 so they bundle it and ship it as docker-runc 2016-04-19 06:42:15 yeah then it's better to put them in the same package 2016-04-19 06:42:23 i think so too 2016-04-19 06:42:26 however 2016-04-19 06:42:31 i am struggeling with go 2016-04-19 06:42:31 ncopa: is docker in 3.3 affected by this? 2016-04-19 06:42:36 Adran: no 2016-04-19 06:43:11 here is my work in progress: http://tpaste.us/3YeP 2016-04-19 06:43:40 but it complains on my gopath 2016-04-19 06:44:02 fabled: confirmed, ovl / lbu file is what breaks live iso. :/ 2016-04-19 06:44:11 i have never really understood how to re-package go packages 2016-04-19 06:44:16 which does pose a slight issue :| 2016-04-19 06:44:32 ncopa: they're kinda static binaries right? 2016-04-19 06:44:32 seems like you are not supposed to install go stuff with distro packages 2016-04-19 06:44:36 yes 2016-04-19 06:45:00 i think you cal build and link against dynamic C libs 2016-04-19 06:45:35 ncopa: clearly Go is not here to help distro packagers ... 2016-04-19 06:45:41 http://tpaste.us/G55N 2016-04-19 06:45:43 ncopa: https://groups.google.com/forum/#!topic/golang-nuts/-UG-lF_Ey-o/discussion 2016-04-19 06:46:10 i think i saw that long time ago 2016-04-19 06:47:10 "go tool has no concept of DESTDIR other then GOPATH. so you need to do 2016-04-19 06:47:10 some real trickery with GOPATH to emulate DESTDIR." 2016-04-19 06:47:16 i think that is what i am struggeling with 2016-04-19 06:48:25 sounds like chroot time 2016-04-19 06:48:50 the problem with chroot is that it require root permission 2016-04-19 06:49:04 i dont want require that all packages are built as root 2016-04-19 06:49:25 i don't want run scripts (configure/make) downloaded from internet as root 2016-04-19 06:55:46 looks like most go projects has their homebuilt script 2016-04-19 06:56:23 syncthing does: go run build.go ... 2016-04-19 06:56:43 hub does: ./script/build 2016-04-19 06:57:00 runc itself ships a makefile (that does not work?) 2016-04-19 06:57:02 yep PITA 2016-04-19 06:57:53 godep does not ship anything so you have to set up the fake dirs for GOPATH manually and do: go build 2016-04-19 07:19:49 huh, so confirmed, alpine-iso doesn't like it very much if you make a lbu ovl that contains .default_boot_services 2016-04-19 10:04:56 is the problem "go decided that things would be done the One True Way, and then nobody agreed" 2016-04-19 10:05:56 \o/ 2016-04-19 10:06:05 i finally got docker running properly 2016-04-19 10:06:25 is that "it worked" or "yay everything's on fire" 2016-04-19 10:07:20 its "it finally works and i am so happy that i am jumping up and down" 2016-04-19 10:09:16 morning 2016-04-19 11:57:14 ncopa: well you working there now ... we did expect it to work :P 2016-04-19 12:40:35 ncopa, are you here? 2016-04-19 12:45:21 Hello, I'm trying to get a touchscreen driver working on the rpi2 with alpine 3.3.3, I managed to compile the module but the binaries seem to have missing dependencies... I used ldd and it indicated ld-linux-armhf.so.3 was missing, so I linked that to /lib/ld-musl-armhf.so.1 and it seemd to work... But the next thing I am getting is symbol not found for: __fdelt_chk, __strcpy_chk, __strcat_chk, etc. 2016-04-19 12:46:31 Is there a compatability package for those functions? 2016-04-19 12:47:37 I see there was a patch for musl "implement glibc *_chk intefcaces for ABI compat..", was this implemented? 2016-04-19 12:48:05 jirutka: hi, im here 2016-04-19 12:57:13 ncopa: could you please look at https://github.com/alpinelinux/aports/pull/56 ? it’s needed for https://github.com/alpinelinux/aports/pull/58 2016-04-19 12:58:46 drats. i didnt think you would notice i was ignoring those... :) 2016-04-19 12:59:18 drats? 2016-04-19 12:59:28 ncopa: drats? 2016-04-19 13:00:00 who’s algibot? 2016-04-19 13:00:09 a friend of mine 2016-04-19 13:00:23 human or machine? :P 2016-04-19 13:00:27 he can be a bit annoying at times 2016-04-19 13:01:34 algitbot is never annoying \o/ 2016-04-19 13:01:50 lol :) 2016-04-19 13:02:03 yeas, ’cause everytime when someone wrote “drats”, you’ve responded with “sorry about that”… and also you have “bot” in your name :P 2016-04-19 13:02:07 XD 2016-04-19 13:02:35 algitbot: please behave! 2016-04-19 13:03:08 XD 2016-04-19 13:03:31 its a stupid bot \o/ 2016-04-19 13:03:37 skazz: the rpi kernel should contain the drivers for touchscreen, is that not the case? 2016-04-19 13:04:01 ScrumpyJack: This is a acustic wave touch, does not work out the box, I do have another capacitive touch screen and it does work 2016-04-19 13:04:10 algibot: are you annying? ;) 2016-04-19 13:05:04 algibot: how can be machines tired? 2016-04-19 13:05:51 algitbot: what do you think about that^^^ 2016-04-19 13:07:12 i think its broke 2016-04-19 13:08:02 ACTION pokes algitbot with a stick 2016-04-19 13:09:07 algitbot: are you jelly? 2016-04-19 13:09:18 so its not 'are you' =o 2016-04-19 13:09:39 ncopa, how difficult is having #5011 done? 2016-04-19 13:09:52 #1337 2016-04-19 13:10:29 Sorry... Anyway, anyone know aboit 2016-04-19 13:10:46 CVE-2014-6271 2016-04-19 13:10:46 about these __fdelt_chk, __strcpy_chk, __strcat_chk functions? 2016-04-19 13:11:20 sounds like glibc specific issue 2016-04-19 13:11:52 There was a patch for musl to implement them, but I'm not sure if it took (from 2015) 2016-04-19 13:11:53 it's ASLR 2016-04-19 13:11:54 skazz: do you have the source? 2016-04-19 13:12:20 it should be rebuilt against musl 2016-04-19 13:12:21 No, the touchscreen people only provide the source for the kernel module (which I had to add an include for) 2016-04-19 13:12:39 file a bug to them 2016-04-19 13:12:50 say that you you use musl libc 2016-04-19 13:12:57 I'll see if I can get them to compile it, I was just wondering if there was some sort of compatability library which I could add, they are only minor functions at a glance (6 lines each) 2016-04-19 13:13:24 skazz: musl provides partial glibc abi compat 2016-04-19 13:13:31 but its not complete 2016-04-19 13:18:53 algibot: what is the current state of Rust in Alpine Linux? :P /cc leo-unglaub 2016-04-19 13:20:49 btw #5386 is already fixed, so it should be closed 2016-04-19 13:23:45 #5390 is also fixed 2016-04-19 13:26:56 ncopa: is there a reason newapkbuild doesnt correctly set builddir anymore? 2016-04-19 13:27:44 clandmeter: what do you mean with “correctly set”? what it do now? 2016-04-19 13:28:02 builddir="$srcdir"/ 2016-04-19 13:28:37 hm, it should be `builddir="$srcdir/$pkgname-$pkgver"` 2016-04-19 13:30:15 iirc previously it would use the root dir in the archive 2016-04-19 13:34:54 its probably due to the change from _builddir to builddir 2016-04-19 13:34:58 havent really checcked yet. 2016-04-19 13:35:09 thought that maybe it changed for some reason. 2016-04-19 13:36:06 btw any reason why the directory isn’t changed automatically to $builddir prior to invoking user-defined package/build/etc functions? 2016-04-19 13:37:03 jirutka, $builddir was previously $_builddir and not used in the abuild library functions. this was recently changed for edge when default prepare function was added 2016-04-19 13:37:23 I know 2016-04-19 13:37:32 so it's historical reason 2016-04-19 13:38:14 so isn’t there any reason why _not_ to implement this feature, since we have $builddir? 2016-04-19 13:38:43 i think no 2016-04-19 13:38:46 not all set $builddir yet. but other than that not really. 2016-04-19 13:38:57 i probably didnt pay attention to newapkbuild while at it 2016-04-19 13:42:22 in which case you dont use builddir? 2016-04-19 13:42:38 and are those cases more then 5% of the time? 2016-04-19 13:44:08 i remember for python modules (or some other module) i didnt use it, but if newapkbuild detects autocrap, why not set it correctly? 2016-04-19 16:21:38 ncopa: I've checked, it seems mkinitfs isn't caring about .default_boot_services 2016-04-19 16:29:14 putting the lbu ovl file seems to be making the file system read only 2016-04-19 16:54:16 pretty sure it has to do with initramfs-init, ncopa, where could I debug this further? 2016-04-19 17:55:23 fails at the apk add stage of setup-alpine, trying to travce where init-mkinitfs is having issues 2016-04-19 18:25:57 hi fabled 2016-04-19 18:26:09 can you tell me what's needed to bootstrap java ? 2016-04-19 18:26:41 sh4rm4^bnc, gcc-java -> openjdk7 -> openjdk8 2016-04-19 18:27:09 i tried to use gjc (from 4.6-4.8), with all available ecj*.tar but could never get classpath compiled 2016-04-19 18:27:57 we have java-gcj-compat package to ship the precompiled java 2016-04-19 18:30:52 how can i fetch that ? 2016-04-19 18:31:01 it doesnt have an URL in it, fabled 2016-04-19 18:31:25 it uses gcc-java shipped stuff only 2016-04-19 18:31:52 is that gcj provided by the vanilla gcc ? 2016-04-19 18:32:10 yes 2016-04-19 18:32:20 which ecj.jar do you use ? 2016-04-19 18:33:00 the precompiled java classpath of gcj lives in libgjc.tar 2016-04-19 18:33:11 but it's not compatible with jamvm 2016-04-19 18:33:43 (lacking stuff (value array) in java.lang.String) 2016-04-19 18:34:17 huh. sounds like some configure error. 2016-04-19 18:35:07 openjdk7 we build with icedtea patches 2016-04-19 18:35:32 there are 2 String.java in gcc tree, the classpath one which would be good, and another gcc one which is used for the classpath 2016-04-19 18:35:47 java-gcj-compat does: 2016-04-19 18:35:48 gcj -Wl,-Bsymbolic -findirect-dispatch -o ecj \ 2016-04-19 18:35:48 --main=org.eclipse.jdt.internal.compiler.batch.Main \ 2016-04-19 18:35:48 /usr/share/java/ecj.jar -lgcj \ 2016-04-19 18:35:48 || return 1 2016-04-19 18:35:55 and ships that 2016-04-19 18:36:30 and which ecj.jar is that ? 2016-04-19 18:37:20 the one from gcc-java shipped as part of gcc 2016-04-19 18:37:33 which is ftp://sourceware.org/pub/java/ecj-latest.jar 2016-04-19 18:37:38 aha 2016-04-19 18:37:52 which is a symlink to ecj-4.5.jar 2016-04-19 18:38:27 do you drop ecj.jar in the gcc build tree before the build, or do you use it later ? 2016-04-19 18:39:23 so once you have this gcj compiled ecj binary, you set JAVAC to ecj and use that to compile icedtea ? 2016-04-19 18:41:21 and which gcc version do you use ? i found that gcc 530 with java enabled breaks with -j > 1 2016-04-19 18:41:34 i.e. parallel makefile broken 2016-04-19 18:44:08 (i looked at those packages in alpine but none of them have a download URL) 2016-04-19 18:45:17 fabled, ^ 2016-04-19 19:03:35 ok my first question is answered by by line 241 of the gcc APKBUILD mv "$srcdir"/ecj-latest.jar ecj.jar 2016-04-19 19:46:14 OK, so once i got java-gcj-common reproduced, how do i build openjdk,icedtea or whatever ? 2016-04-19 19:46:26 i cannot find the alpine recipe that does it 2016-04-19 19:48:45 community/openjdk7 uses the icedtea package to build it 2016-04-19 19:49:24 you said something about a precompiled package ? (i assume its java -> class -> jar) 2016-04-19 21:33:56 sorry, for some inexplicable reason, I sent aports patch to alpine-devel@ instead of alpine-aports@... 2016-04-19 21:34:32 but I'll avoid resending unless told otherwise 2016-04-19 23:51:10 wow, nice, someone fixed docker 2016-04-19 23:51:28 who was that generous? 2016-04-19 23:52:03 skrzyp: huh, really? so it’s not a security disaster with absurdly high complexity anymore? :P 2016-04-19 23:52:59 jirutka: but I only use it to play touhou on 32-bit wine on void (while alpine has only wine64) so it's not connected to the network 2016-04-19 23:53:17 skrzyp: that’s okay :) 2016-04-19 23:53:40 skrzyp: sadly many people use this sh*t for running production applications 2016-04-19 23:54:00 it was mean to be developer only tool 2016-04-19 23:54:09 exactly 2016-04-19 23:54:24 but looks like someone sees a huge business in selling this 2016-04-19 23:54:34 yeah, a lot of $$$ 2016-04-19 23:54:39 https://www.youtube.com/watch?v=PivpCKEiQOQ 2016-04-19 23:55:00 I’ll look at it later, I’m now on mobile connection :/ 2016-04-19 23:55:05 okay 2016-04-19 23:55:23 but it's a "must watch" for every Docker user :D 2016-04-20 07:49:51 ncopa: so, you're planning for another release, and I *still* can't send patches to alpine-aports. Nice. 2016-04-20 07:52:44 skarnet: complain to nangel? 2016-04-20 07:52:52 or fix your email server 2016-04-20 07:52:55 ACTION hides 2016-04-20 07:53:21 skarnet: lets fix this now 2016-04-20 07:53:23 barthalion: did you miss the tantrum I threw a few days ago, and the ensuing discussion with nangel? 2016-04-20 07:53:34 yes, and I thought he fixed it then 2016-04-20 07:53:39 what is your exact error message? 2016-04-20 07:54:35 ncopa: my IP is in the PBL, because it's dynamic. spamhaus (or sorbs) apparently doesn't allow you to check against one blacklist and not against another. 2016-04-20 07:54:45 So the only real solution is to disable blacklisting *entirely*. 2016-04-20 08:04:18 I would say "I'll raise awareness on the alpine-devel mailing-list and hope to fan a discussion there" but that would only work if I could, uh, send mail to that mailing-list. :P 2016-04-20 08:05:05 do we have default prepare() now? 2016-04-20 08:05:27 ncopa, clandmeter and whoever packaged something recently ↑ 2016-04-20 08:05:32 nmeum too 2016-04-20 08:05:41 pinging random people always works 2016-04-20 08:07:01 barthalion: default prepare() is there :) 2016-04-20 08:07:17 with default patchlevel set to 1? 2016-04-20 08:07:20 barthalion: yes 2016-04-20 08:07:25 good 2016-04-20 08:07:27 thx 2016-04-20 08:07:36 you can override it by some (undocumented) var 2016-04-20 08:07:42 patch_flags or similar 2016-04-20 08:07:46 i dont remember 2016-04-20 08:07:47 Going on an errand. Please make sure I don't have to turn all big, green and angry again when I'm back. 2016-04-20 08:07:58 :) 2016-04-20 08:09:57 it's patch_args i think 2016-04-20 08:21:07 re skarnets dnsbl problem 2016-04-20 08:21:32 i think the technical correct solution is to disable dns blacklist 2016-04-20 08:21:44 possible do greylist only 2016-04-20 08:22:09 however, i do have a serious spam problem 2016-04-20 08:22:29 i spend more time on deleting spam than answering emails 2016-04-20 08:23:08 i think a proper content based spam filter is needed 2016-04-20 08:23:17 but, dnsbl is very tempting 2016-04-20 08:23:21 because its cheap 2016-04-20 08:23:34 ncopa: is this for lists.a.o only? 2016-04-20 08:23:48 for *.alpinelinux.org 2016-04-20 08:23:50 greylisting + spamassassin should be sufficient no ? 2016-04-20 08:23:58 whitelisting works well for mailing lists 2016-04-20 08:24:36 spamassassin uses cpu 2016-04-20 08:24:39 and bw 2016-04-20 08:25:03 i am interested in other spam filters, maybe dspam or similar 2016-04-20 08:25:21 but nothing is as cpu/bw effective as dnsbl 2016-04-20 08:25:26 skarnet | So the only real solution is to disable blacklisting *entirely* < or not use the list that contain PBL ... SBL should be sufficient 2016-04-20 08:25:33 same here. I don't use bayers antispam, too heavy 2016-04-20 08:26:02 PBL/SBL? 2016-04-20 08:26:30 ncopa: I had spamass + dspam in the past, it's quite hungry as well ... 2016-04-20 08:26:57 ncopa: https://www.spamhaus.org/ 2016-04-20 08:27:05 what's the mailing list mananager? 2016-04-20 08:27:18 do you by any chance use zen.spamhaus.org as dnsbl ? 2016-04-20 08:27:22 we use zen spamhaus 2016-04-20 08:27:26 yes 2016-04-20 08:27:29 yeah you shouldn't 2016-04-20 08:27:40 as it includes dynamic IPs as well 2016-04-20 08:27:57 so what should i use? 2016-04-20 08:28:05 switch to sbl.spamhaus.org and it should be OK 2016-04-20 08:28:47 I think they also have a sbl-xbl.spamhaus.org alias if you want to check XBL as well 2016-04-20 08:29:56 zen.spamhaus.org replaces sbl-xbl.spamhaus.org in most configurations. If you are currently using sbl-xbl.spamhaus.org you should replace sbl-xbl.spamhaus.org with zen.spamhaus.org. 2016-04-20 08:30:23 but sbl-xbl shouldn't include PBL 2016-04-20 08:30:44 cause there's some people like skarnet that want to send emails from home :P 2016-04-20 08:31:27 i think sending emails from home is a reasonable request 2016-04-20 08:31:35 i think its reasonable to expect that to work 2016-04-20 08:31:54 yep 2016-04-20 08:32:05 so sbl only should be better 2016-04-20 08:32:51 "The Spamhaus PBL is a DNSBL database of end-user IP address ranges which should not be delivering unauthenticated SMTP email to any Internet mail server except those provided for specifically by an ISP for that customer's use." 2016-04-20 08:33:27 so its a filter of ip ranges that "should not send email" 2016-04-20 08:34:02 i think removing PBL is a good thing 2016-04-20 08:34:44 I've had some customers being big asia cable providers having their mail servers in the range of their end-user IP 2016-04-20 08:34:50 gave good results :P 2016-04-20 08:36:46 is it possible to use sbl+xbl+dbl, or does that not make sense 2016-04-20 08:42:02 coredumb: where can i read about the sbl-xbl? 2016-04-20 08:42:49 back 2016-04-20 08:43:16 i also think we should whitelist mailing list subscribers 2016-04-20 08:43:26 https://pkgs.alpinelinux.org/package/edge/testing/x86_64/rspamd any good? 2016-04-20 08:43:42 I use greylisting and have amazing results - I still get spam, but most of it is correctly handled by my client's filter. 2016-04-20 08:43:58 we use gross for that 2016-04-20 08:44:35 also, yes, the only way to get consistently correct result is to use a spam filter. Yes, it's CPU-heavy. No, I don't care. CPU exists for a reason: to do the right thing. 2016-04-20 08:44:51 as I already said, I'm ready to contribute financially if that's a problem. 2016-04-20 08:45:33 but in the short term, not using the PBL would already be a *huge* improvement, so thank you for considering it. 2016-04-20 08:45:55 i agree with you that PBL is wrong 2016-04-20 08:46:00 ncopa: have you considered using an external provider for incoming mail? 2016-04-20 08:46:23 i mean who are they to decide who is supposed to send emails 2016-04-20 08:46:27 yth: external providers are using blacklists more and more 2016-04-20 08:46:28 yth: not really 2016-04-20 08:46:52 that would only put the problem outside of Alpine's control 2016-04-20 08:47:16 skarnet: could you please try send again? 2016-04-20 08:47:42 still nope 2016-04-20 08:47:58 Client host [82.216.x.y] blocked using safe.dnsbl.sorbs.net; 2016-04-20 08:48:00 safe.dnsbl.sorbs.net disagrees 2016-04-20 08:48:06 sigh 2016-04-20 08:48:15 nangel already tried things 2016-04-20 08:48:18 it didn't work 2016-04-20 08:48:27 skarnet: that's true, but the better mail providers also dedicate a lot more time and resource to getting it right (in terms of false positives, tuning greylisting etc) than us - cost/benefit... 2016-04-20 08:48:42 as long as one of those !@#$%^&* servers is listed, they use all their blacklists, you can't select them 2016-04-20 08:49:12 yth: that's true, and it's a possible solution *if* they're committed to never using blacklists 2016-04-20 08:49:56 i'm removing sorbs 2016-04-20 08:50:53 hm 2016-04-20 08:51:30 hm? 2016-04-20 08:51:38 it sorbs has actually blocked 14 mails last 24 hours 2016-04-20 08:51:53 that was not blocked by spamhaus 2016-04-20 08:52:02 including the one I just sent 2016-04-20 08:52:09 yes 2016-04-20 08:52:22 i bet you are going to receive a lot more spam in favor of letting users sending mail from home. 2016-04-20 08:52:34 clandmeter: right, there's no quick fix 2016-04-20 08:52:58 whiltelist subscribers? 2016-04-20 08:53:12 that works 2016-04-20 08:53:15 thats probably better 2016-04-20 08:53:21 i think its possible 2016-04-20 08:53:31 and probably not so difficult to implement 2016-04-20 08:53:40 greylisting dynamic IPs? 2016-04-20 08:53:43 or let them use dyndns 2016-04-20 08:54:29 i think we can simply whitelist subscribers 2016-04-20 08:55:01 and if we get spam on the list, we can remove them from the ml 2016-04-20 08:55:12 its not as safe as dyndns 2016-04-20 08:55:20 uh 2016-04-20 08:55:35 something with "dns" in the name can't be safe 2016-04-20 08:55:59 safer doesnt say its safe :p 2016-04-20 08:56:10 depending on the mlm, it can be tricky. you'll need to let *anyone* subscribe, so the mlm needs to pass the whitelist to the mta 2016-04-20 08:57:24 or maybe use a moderated subscription MTA (different "host" than the actual mailing list) 2016-04-20 09:02:34 anyone can subscribe 2016-04-20 09:02:43 but you need to confirm the subscription 2016-04-20 09:03:05 which means that the spammer needs to use an email he has control over 2016-04-20 09:03:48 so i think that should work 2016-04-20 09:03:55 I'm using qconfirm for my mailing-lists, I can confirm it works very well to keep the spam away (but that precise implementation is qmail-only) 2016-04-20 09:05:53 ncopa: ...or spoof the from address of an existing subscriber, which isn't uncommon 2016-04-20 09:06:48 if a subscriber replies "yes" to a confirmation email they don't know the origin of, it's on them 2016-04-20 09:07:09 yes it does not solve the spoof problem 2016-04-20 09:07:22 (still, that's probably a small part of the total picture) 2016-04-20 09:07:49 i think whitelist subscribers is a step forward 2016-04-20 09:08:10 what's the mlm on lists.a.o? 2016-04-20 09:08:24 mlmmj 2016-04-20 09:08:35 (nothing in the headers) 2016-04-20 09:13:56 spamhaus SBL + greylisting should be amply sufficient 2016-04-20 09:14:41 coredumb: it's apparently not possible to only use the SBL and not the PBL. That's the main issue here. 2016-04-20 09:14:58 well it should 2016-04-20 09:15:23 issue here is athat we also use sorbs in addition to spamhaus 2016-04-20 09:15:48 spamhaus alone uses the PBL as well 2016-04-20 09:15:51 and apparently sorbs has blocked 13 spams last 24 hours 2016-04-20 09:16:13 ncopa: I'm not sure if you consider this negligible or good 2016-04-20 09:17:13 well, i prefer 100 spams in my inbox over 113 per day 2016-04-20 09:17:31 if we can work around theproblem by whitelisting subscribers, then i prefer try that first 2016-04-20 09:17:52 befor accepting 13 more spams / day 2016-04-20 09:17:52 I agree 2016-04-20 09:18:46 when it comes to spamhaus, it appears that the vast majority of the blocked emails there came from SBL 2016-04-20 09:19:12 actually whitelisting only subscribers and discarding the rest would solve 99.9% of the issues 2016-04-20 09:19:35 thats what i'm hoping for 2016-04-20 09:21:52 skarnet: Whenever possible, we encourage applications to query zen.spamhaus.org and then parse the return code(s) to determine whether to block an IP. This prevents unnecessary queries and speeds processing on your application. If your application cannot parse return codes, you can query sbl.spamhaus.org to determine whether an IP address is on the SBL, and xbl.spamhaus.org to determine whether an IP 2016-04-20 09:21:55 address is on the XBL. Either of these zones returns 127.0.0.2 if the IP address is on that blocklist. 2016-04-20 09:22:14 so yes either sbl, xbl, sbl-xbl, or zen to match all the lists at once 2016-04-20 09:24:01 and now external providers like office 365 emails are blacklisting by default IP they've never seen... 2016-04-20 09:24:31 no matter if you're not blacklisted anywhere if they haven't seen your IP the don't want to talk to you 2016-04-20 09:24:33 >_< 2016-04-20 09:25:20 case in point: external providers generally can't be trusted to do the right thing 2016-04-20 09:25:30 ah clearly 2016-04-20 09:25:53 I'm still considered as a spammer by Gmail ... 2016-04-20 09:25:55 :D 2016-04-20 09:26:07 get a dkim key :P 2016-04-20 09:27:47 yeah yeah ... this and that... 2016-04-20 09:30:01 skarnet: can you please retry send? 2016-04-20 09:30:12 wait 2016-04-20 09:30:57 did it work? 2016-04-20 09:31:05 I didn't receive a bounce this time 2016-04-20 09:33:33 yep directly in /dev/null now problem solved :P 2016-04-20 09:34:39 %1927 2016-04-20 09:34:48 yay 2016-04-20 09:34:49 thanks! 2016-04-20 09:35:00 thanks for you patience skarnet 2016-04-20 09:35:07 can I send the rest now, or do you need more time to finish configuring stuff? 2016-04-20 09:35:13 no, thank *you* for fixing this. 2016-04-20 09:35:27 i think you can send the rest 2016-04-20 09:36:16 ^^ 2016-04-20 09:37:06 all sent! 2016-04-20 09:37:50 all received 2016-04-20 09:40:54 let this be a lesson: when you want something done (and can't do it yourself), being a constant pain in the ass to the people who can get it done *works*. :D 2016-04-20 09:57:26 ^^ 2016-04-20 10:17:11 skarnet: ay! 2016-04-20 11:00:22 uff 2016-04-20 11:00:26 gsad segfaults 2016-04-20 11:00:27 [ 1967.420502] grsec: From 10.44.64.12: denied resource overstep by requesting 4096 for RLIMIT_CORE against limit 0 for /usr/bin/gpg2[gpg:3061] uid/euid:0/0 gid/egid:0/0, parent /bin/busybox[ash:2007] uid/euid:0/0 gid/egid:0/0 2016-04-20 11:00:27 [ 2952.107413] gsad[3454]: segfault at 77c54bab5fb8 ip 000077c54a5c9b6f sp 000077c54bab5f90 error 6 in libopenvas_omp.so.8.0.7[77c54a5c7000+207000] 2016-04-20 11:00:27 [ 2952.107425] grsec: From 10.44.64.12: Segmentation fault occurred at 000077c54bab5fb8 in /usr/bin/gsad[gsad:3454] uid/euid:0/0 gid/egid:0/0, parent /bin/busybox[init:1] uid/euid:0/0 gid/egid:0/0 2016-04-20 11:00:29 [ 2952.107447] grsec: From 10.44.64.12: denied resource overstep by requesting 4096 for RLIMIT_CORE against limit 0 for /usr/bin/gsad[gsad:3454] uid/euid:0/0 gid/egid:0/0, parent /bin/busybox[init:1] uid/euid:0/0 gid/egid:0/0 2016-04-20 11:01:07 it's on x86_64 2016-04-20 11:01:55 gsad it's greenbone-security-assistant 2016-04-20 11:02:05 this happens when you try to login in the webif 2016-04-20 11:09:47 in libopenvas_omp.so.8.0.7 2016-04-20 11:10:20 does it use threading? 2016-04-20 11:10:39 a common issue is that they dont allocate stack space 2016-04-20 11:10:53 musl have 80k stack space for threads by default 2016-04-20 11:10:58 glibc has 8MB irc 2016-04-20 11:12:11 huh, it looks like its gpg 2016-04-20 11:37:00 gsad: GNU sad 2016-04-20 11:37:24 that's sadness with more autoconf 2016-04-20 11:38:08 ncopa, would an strace help? 2016-04-20 11:39:38 dunno 2016-04-20 12:12:19 i've reported the segfault to the upstream 2016-04-20 12:58:21 ncopa: anything we can do to speed up the dovecot package with pigeonhole plugin integration in stable ? 2016-04-20 14:22:41 what's the best way to add/remove a user post/pre installation in a package? 2016-04-20 14:54:56 ncopa: you around? 2016-04-20 14:56:37 are you ok with this: 2016-04-20 14:56:38 https://dpaste.de/OhcC 2016-04-20 14:57:04 adding gnutls-dev libmicrohttpd will support https 2016-04-20 15:08:08 hi, I'd like to submit a patch for this bug: https://bugs.alpinelinux.org/issues/4855 2016-04-20 15:08:33 should I use the aport ML or the alpine-devel mailing list for this? 2016-04-20 15:09:48 also, should I contact the maintainer of the package first, or is it ok to directly submit a patch? 2016-04-20 15:18:01 littledude: there’s also the third option, you can send pull request on GitHub https://github.com/alpinelinux/aports 2016-04-20 15:19:54 oh that's just perfect :) thanks jirutka1 2016-04-20 15:20:04 ACTION grumbles 2016-04-20 15:22:55 skarnet: why? it's much easier than sending emails via git 2016-04-20 15:24:01 I don't like it when people advise using github (or any other commercial platform). I don't like it when people rely on it more. 2016-04-20 15:24:06 I actually have some APKBUILD that I never sent because it was tedious. Now that I can make PRs I'll submit them. 2016-04-20 15:24:12 But you can safely ignore me, I'm a grumpy dinosaur. 2016-04-20 15:24:18 :) 2016-04-20 15:24:26 I also have multiple APKBUILDS because effort to send 2016-04-20 15:24:29 Pull Reuests ++ 2016-04-20 15:24:32 skarnet makes me laugh 2016-04-20 15:24:54 that's at least one, thanks fcolista :'-) 2016-04-20 15:25:06 :-DD 2016-04-20 15:25:21 skarnet, love your webpage 2016-04-20 15:26:04 which one, the /index.html? there's not much to love there (and not much to hate either XD) 2016-04-20 15:26:20 where is it? 2016-04-20 15:26:26 yes, the homepage..but basically there's no much more :D 2016-04-20 15:27:01 littledude, skarnet.org 2016-04-20 15:27:15 there's a lot of stuff under software/s6/ and such :P 2016-04-20 15:27:33 thanks 2016-04-20 15:27:41 yeah..the intersting stuff is there 2016-04-20 15:45:37 looks like busybox change from 3.3 to edge causes /etc/init.d/cron to become /etc/init.d/crond 2016-04-20 15:45:50 this breaks ACF, and probably causes upgrade problems 2016-04-20 15:49:07 ncopa ^^ 2016-04-20 15:51:15 #5462 2016-04-20 15:51:27 assigned to ncopa 2016-04-20 16:02:08 skarnet, Adran, littledude: I started contributing to Gentoo after they had finally migrated to git and started accepting pull requests. Sending patches via email is like using Usenet for me. GitHub is really great platform and makes contributing so much easier and fun. That said, I fully understand skarnet concerns about relaying on a commercial platform. I think that the current state, when we use GitHub just as a mirror, is good compromis 2016-04-20 16:02:34 I like the Github pull capabilties. 2016-04-20 16:02:37 no need to convince me. :) 2016-04-20 16:02:53 It would be functionally the same if gitlab was used 2016-04-20 16:03:03 But the ease is what is great 2016-04-20 16:03:15 I’m not convincing, just wanted to say it :) 2016-04-20 16:05:22 I know GitLab very well, I’m administrating GitLab for ~2k users on our faculty and three other small instances. It’s very good for internal projects, but I always prefer GitHub for OSS projects, because it’s much more accessible for others (all open-source in one place). 2016-04-20 16:07:01 is gitlab free for self hosting? 2016-04-20 16:07:11 yes, it’s open-source project 2016-04-20 16:07:31 they also provide enterprise edition that is paid and a hosted platform 2016-04-20 16:07:38 the enterprise edition has some additional features 2016-04-20 16:07:43 nice. I tried gogs recently. Super easy to set up, but no PR 2016-04-20 16:08:20 I'll try gitlab then 2016-04-20 16:08:22 Gogs – *Go* Git service – I’ve stopped here. 2016-04-20 16:09:15 I’m preparing Alpine bundle for GitLab, it’ll be hopefully ready in few days. 2016-04-20 16:18:37 coredumb: i dont know how to speed up things :-/ 2016-04-20 16:18:56 fcolista: can you add explicit --enable-gnutls or similar configu option? 2016-04-20 16:19:13 nmeum: could you please have a look at #5462 2016-04-20 16:21:21 ncopa, yes 2016-04-20 16:21:44 --enable-https 2016-04-20 16:21:56 by default is "auto" 2016-04-20 16:22:00 add that too 2016-04-20 16:22:04 ok 2016-04-20 16:22:09 better be explicit 2016-04-20 16:22:29 ok 2016-04-20 16:22:30 donw 2016-04-20 16:22:32 done 2016-04-20 16:22:32 so that if gnutls is broken or something, we will get error instead or build without https 2016-04-20 16:22:39 ah 2016-04-20 16:22:43 that's why 2016-04-20 16:22:48 yes 2016-04-20 16:23:23 and if we would want it without, then we should explicitly have added --disable-gnutls 2016-04-20 16:23:30 so that it is not pulled in by mistake 2016-04-20 16:23:49 https://dpaste.de/GYqU 2016-04-20 16:23:51 here we go 2016-04-20 16:24:01 pushit 2016-04-20 16:24:21 done 2016-04-20 16:24:29 ncopa, a question re github aports 2016-04-20 16:24:40 what should be the workflow? 2016-04-20 16:24:42 I mean 2016-04-20 16:24:51 someone does a pull request to github 2016-04-20 16:25:02 then, we pull it in our local repo 2016-04-20 16:25:29 copy the patch in our local master repo 2016-04-20 16:25:32 push it? 2016-04-20 16:26:20 to me github is just another way to get a patch 2016-04-20 16:26:38 you could git format-patch --stdout -1 | tpaste 2016-04-20 16:26:53 and git am the paste 2016-04-20 16:26:59 or get it from email 2016-04-20 16:27:13 or use github.com/..../*.patch 2016-04-20 16:27:21 or git pull from pull url 2016-04-20 16:27:41 the benefit with git pull is that you can grab a set of patches 2016-04-20 16:27:47 so its convenient if there are more than one 2016-04-20 16:28:30 and then all pull requests are automatically built on CI 2016-04-20 16:28:36 *then → that 2016-04-20 16:28:40 what i do is: git pull $url && git pull --rebase 2016-04-20 16:28:43 so i rebase the patches 2016-04-20 16:29:13 yes, the CI will tell you if the patch builds 2016-04-20 16:29:35 which is nice 2016-04-20 16:29:36 for example https://travis-ci.org/alpinelinux/aports/builds/124509627, you can click on “Building package main/dhcp” and see full log 2016-04-20 16:33:31 ok, sorry but i'm not that clever. ncopa, you can get the patch with git format-patch from where? I mean: you clone the github repo locally, then git format-patch --stdout etc.. 2016-04-20 16:33:31 ? 2016-04-20 16:34:32 fcolista: for example, https://patch-diff.githubusercontent.com/raw/alpinelinux/aports/pull/59 … just append “.patch” to the end of the url, https://patch-diff.githubusercontent.com/raw/alpinelinux/aports/pull/59.patch 2016-04-20 16:34:32 If i'm $user and i want to submit a patch, i do a pull request to github aports. 2016-04-20 16:34:48 pardon, this https://github.com/alpinelinux/aports/pull/59 2016-04-20 16:34:51 and https://github.com/alpinelinux/aports/pull/59.patch 2016-04-20 16:35:08 ok 2016-04-20 16:35:16 you wget it? 2016-04-20 16:35:30 in local repo? 2016-04-20 16:35:37 then git am and git push ? 2016-04-20 16:41:31 curl https://github.com/alpinelinux/aports/pull/59.patch | git am 2016-04-20 16:41:39 or wget -O - https://github.com/alpinelinux/aports/pull/59.patch | git am 2016-04-20 16:41:50 yes ok.. 2016-04-20 16:41:55 makes sense 2016-04-20 16:57:35 ncopa: I will take a look at #5642 tonight. I am rather busy these days sorry about that 2016-04-20 17:06:36 I accidentally sent aports patch (main/openrc modules.initd fix) to alpine-devel instead of -ports yesterday. should I resend it to -ports? not sure how picky AL-devs are about workflow (but I read from IRC backlog that aports are also on GH, so I guess the answer is: not much). 2016-04-20 17:07:23 s/\/a&/ of course 2016-04-20 17:11:13 fcolista: regarding GH PR you may also read: https://help.github.com/articles/checking-out-pull-requests-locally/ - it's convenient and useful that PRs outlive forks they're originating from. 2016-04-20 17:11:52 thx przemoc 2016-04-20 17:13:09 that way you can fork, do changes, send PR and remove your fork if it's no longer needed 2016-04-20 17:20:03 ncopa / nmeum: if the change to crond was on purpose, I can always update ACF 2016-04-20 17:20:11 just let me know 2016-04-20 17:27:50 from the commit message alone it can be concluded the change was intentional http://git.alpinelinux.org/cgit/aports/commit/?id=64c6f1e9d0c4 2016-04-20 17:28:31 true, but I want ncopa's signoff, if possible 2016-04-20 17:28:35 sure thing 2016-04-20 18:14:50 tdtrask: yes it was on purpose. The problem is that cron is also the name of a virtual OpenRC service thus the cron service was renamed to crond in order to prevent this conflict 2016-04-20 18:15:08 so if you update ACF accordingly that would be nice 2016-04-20 18:15:55 sorry for not checking ACF when I made that change 2016-04-20 18:17:46 no problem 2016-04-20 18:17:57 how will upgrades from 3.3 work? 2016-04-20 18:18:19 ACTION tried upgrading and didn't see crond in rc-status output 2016-04-20 18:19:12 you upgraded from 3.3 to edge? 2016-04-20 18:22:24 either way you need to manually replace the cron service with crond when upgrading the busybox package which should include that in the release notes for 3.4 2016-04-20 18:22:40 s/which/we/ 2016-04-20 18:43:55 Hi could not figure out docker-compose in testing,edge and community repos. Is there any? 2016-04-20 18:44:02 I pinned all repos, fyi 2016-04-20 18:47:33 sorry nmeum, I stepped away 2016-04-20 18:47:53 yes, upgrade from 3.3 to edge did not automatically account for cron -> crond change 2016-04-20 18:48:12 ACTION things that would be important 2016-04-20 18:48:16 thinks 2016-04-20 18:50:01 we could add a post-upgrade script to the busybox-initscripts package which does that 2016-04-20 18:50:37 ACTION agrees 2016-04-20 19:19:39 ncopa: can I do something ? 2016-04-20 19:20:02 that's quite a broad question. :) 2016-04-20 19:27:41 you think so ? 2016-04-20 19:34:50 coredumb: are you French perhaps? 2016-04-20 19:35:19 damn :( 2016-04-20 19:35:30 trying to not show it 2016-04-20 19:35:32 :D 2016-04-20 19:37:04 this very annoying French-specific space (incorrect in English and most other languages) before question mark is my only hunch right now, though. 2016-04-20 19:38:56 oh really? :D 2016-04-20 19:39:07 cause I'm not really consistent on this one 2016-04-20 19:39:18 sometimes a space, sometimes not ... 2016-04-20 19:39:34 You don't need to be French to put spaces before question marks. You're not *supposed* to do it in English, but some people prefer it. 2016-04-20 19:40:11 you say that because you're french as well :D 2016-04-20 19:40:20 if you want to be typographically pedantic, the rules for French are a *semi* space before question/exclamation marks. 2016-04-20 19:40:24 so... 2016-04-20 19:40:35 it's just as valid to put it and not to put it. :P 2016-04-20 19:40:42 lol 2016-04-20 19:41:42 I see people putting space before question marks in every language, sadly, but that doesn't change the fact it's incorrect in most languages other than French 2016-04-20 19:45:07 I really don't have a strong attachment to correct typography. I tend to see it as a purely aesthetical thing and consider its rules mostly religious. 2016-04-20 19:45:24 przemoc: tu es un raciste ! 2016-04-20 19:45:34 And if we're talking religion, mine disagrees with yours. 2016-04-20 19:45:46 no just kidding.. I'm not even French, but just been studying it since I was very young 2016-04-20 19:45:59 you're just a showoff :D 2016-04-20 19:46:10 what just happened here. 2016-04-20 19:46:26 I find national typography as kind of heritage that should be preserved and cultivated. 2016-04-20 19:47:26 I have that kind of attachment to spelling, but not to typography. 2016-04-20 19:47:46 orthography also shouldn't be neglected 2016-04-20 19:48:52 I really don't care much :P 2016-04-20 19:49:11 s/much/at all/ 2016-04-20 19:53:02 Trying to build and compile from https://github.com/hyperhq/hyperd.git, but I am getting musl-related issues (http://hastebin.com/xubewiromi.vbs). Any input appreciated! 2016-04-20 19:54:35 ACTION has 0 go experience 2016-04-20 19:57:52 The error with a copy of Makefile is posted in http://hastebin.com/evikapojav.mel for reference. 2016-04-20 20:44:35 how quickly http://lists.alpinelinux.org/alpine-devel/ is updated? I sent some message a few minutes ago, but I don't see it there. yesterday it was visible within same timeframe. 2016-04-20 20:45:13 przemoc I see your message. 2016-04-20 20:45:18 fine then 2016-04-20 20:45:33 ;) 2016-04-20 20:47:30 sample of volume 1 is not good in any sense, but still promising (if the answer is positive) 2016-04-20 20:50:01 (just being cautious as apparently some people, skarnet mostly form what I've seen, have problems with mailing AL MLs) 2016-04-20 20:50:52 email issues should be fixed now. I made a point of getting it fixed for everyone, not just myself. 2016-04-20 20:51:03 <3 2016-04-20 20:53:37 well, I don't send mails using home mail server, as I don't have one (and don't plan on having one, too much hassle to be "respected" by gmail & co, especially when you're on dyn IP), but you never know when things may become broken in the mail provider you use (I'm using zoho for mailing list stuff) 2016-04-20 21:04:04 https://github.com/golang/go/issues/15394 2016-04-20 21:13:35 kunkku: does awall has any support for custom rules (e.g. --ttl-set)? is there any recommended way of defining them? or I should simply remember to add lacking lines after `awall activate` (possibly wrapping it into my own script)? 2016-04-20 21:20:12 there is no easy way at the moment 2016-04-20 21:20:31 is that the only thing you would need? 2016-04-20 21:23:54 I have been considering such an extension mechanism where you could define custom chains and jump there 2016-04-20 21:24:18 being able to provide custom rules where variables could be used would be good enough I guess 2016-04-20 21:27:09 but usually I have ended up implementing native support where there are new needs :) 2016-04-20 21:28:17 re custom rules: I would allow defining chain each one is added too. it would be kind of power user feature anyway, so there would be no point in enforcing them in some special chains only. 2016-04-20 21:29:41 so you say you want details to implement exact use-case? :) 2016-04-20 21:29:57 well that's one option... 2016-04-20 21:30:24 though an extension mechanism would be of more use in future 2016-04-20 21:31:21 that goes without saying and it wouldn't block you from implementing other concrete use cases in fully awall way 2016-04-20 21:31:39 in future 2016-04-20 21:34:08 would you like to describe in more detail how you think the extension thing should look like? 2016-04-20 21:47:12 I would simply add top-level attribute "custom" with list of rules that should be added. Each custom rule would have two attributes: "chain" and "rule". chain is self-explanatory, and rule would be like what you get from one iptables-save entry, e.g. "-A OUTPUT -j ACCEPT" (bad but short example), with additional power of being able to use variables there. 2016-04-20 21:47:56 damn, not chain but table 2016-04-20 21:50:52 if you'd be more verbose, you could add "chain" attribute too, but being able to define whether it's insertion or append would be still useful, that's why I suggest this minimalist approach which easily maps to iptables-save form 2016-04-20 21:53:47 kunkku: so how does it sound for you? 2016-04-20 21:56:00 "custom": [ { "table": "filter", "rule": "-A OUTPUT -j ACCEPT" } ] // more complete example 2016-04-20 21:57:37 or it could be per table rules, maybe it would be even more convenient 2016-04-20 21:58:15 "custom": [ { "table": "filter", "rules": [ "-A OUTPUT -j ACCEPT", ... ] } ] 2016-04-20 22:00:42 my motivation for first way is that you may want to add rules for different tables, yet have them grouped together, because they consist of one logic block. 2016-04-20 22:01:48 then you would also need a mechanism to specify their location w.r.t. other rules 2016-04-20 22:02:27 I think that putting custom rules after normal awall rules would be expected 2016-04-20 22:02:49 though you could use the "after" and "before" directives on policy level 2016-04-20 22:05:55 I was thinking something like 2016-04-20 22:05:58 rules within list would be appended to rules.save in order they appear in it (like in rest of awall), so yes, grouping wouldn't be always possible, but it would be more flexible than { "table":..., "rules": [...] } way 2016-04-20 22:06:51 there is also a third way: having /etc/awall/custom.rules in iptables-save format that are appended to what awall outputs, but with logic to deal with tables properly (I'm not sure whether iptables-save format accepts repeating tables) 2016-04-20 22:06:55 you have normal rules which contain "action": "custom", "target": "xxx" 2016-04-20 22:08:41 then something like you propose but without table+chain 2016-04-20 22:09:04 that would kind of avoid the ordering problem 2016-04-20 22:09:31 but would produce extra chains 2016-04-20 22:10:19 that at least used to come with performance penalty (which is why awall in general avoids creating new chains), but I'm not sure if it matter so much nowadays 2016-04-20 22:14:08 aren't you internally queueing output for rules-save per-table anyway? I somehow assumed you did, that's why "table"+"rule" attributes seemed like a (moderately) low hanging fruit. 2016-04-20 22:15:15 yes, but what if somebody wants to have custom stuff in between some normal rules? 2016-04-20 22:17:00 specifying "target": "xxx" in a normal rule and then "custom": { "xxx": [ "-j XXX --yyy zzz" , ... ] } would nail it down 2016-04-20 22:18:14 hm 2016-04-20 22:18:28 instead of target-level "custom" there could be "custom.pre" and "custom.post", put accordingly before and after normal awall rules. it would deal with 80% of cases, I guess. I'm looking at it also from point of converting existing iptables rules to such awall-custom rules. 2016-04-20 22:20:22 one thing to remember is that awall's top-level dicts do not map 1-to-1 to tables 2016-04-20 22:20:41 that's why I'm using "table" 2016-04-20 22:20:57 the thing is to be able to write any possible rule within iptables 2016-04-20 22:21:01 so iptables' filter get input from multiple awall "tables" 2016-04-20 22:22:32 one option would be to expose some internal awall data structures in the JSON level 2016-04-20 22:22:51 exposing internal stuff always leads to bad things 2016-04-20 22:23:30 but in such a way that you could override certain parts in what awall generates 2016-04-20 22:24:45 { "out": ..., "service": ..., "ipt-action": "--set-ttl 7" } 2016-04-20 22:25:30 service is already a bother for quick rules, to be honest 2016-04-20 22:25:31 { "out": ..., "service": ..., "action": { "target": "TTL", "params": "--set-ttl 7" } } 2016-04-20 22:27:04 { "out": ..., "ipt-match": "-p tcp --dport xxx", "action": { "target": "TTL", "params": "--set-ttl 7" } } 2016-04-20 22:30:07 well, I need to think the altenatives 2016-04-20 22:30:11 "custom.pre", "custom.post" (both with "table" and "rule" attrs), as explained before, and "custom.usernamed" (with rule only as string) creating chain named "usernamed", that one can jump from awall rules, should be able to describe everything power user may need, without requiring tedious json-ification. 2016-04-20 22:34:10 it's "I want to play fair with awall, but I have these rules I'm not able to express there, how I can easily achieve that" kind of thing that would be nice in awall. maybe simple non-json text-files /etc/awall/rules.{pre,post} {pre,ap}pended per-table to rules-save generated by awall would be the best after all for starters 2016-04-20 22:41:50 going back for a minute into your suggestions, ipt-match and params sound's inconsistent. if you'd allow providing custom stuff anywhere, then name such attribute the same everywhere. service entry is not mandatory after all, so it's not about checking whether there is service or ipt-match. 2016-04-20 22:43:21 e.g. { "out": ..., "params": "-p tcp --dport xxx", "action": { "target": "TTL", "params": "--set-ttl 7" } } 2016-04-20 22:44:34 BTW is there now perhaps any way to provide '-p ... --dport ...' instead of service? git grep 'ipt-match' didn't reveal anything. 2016-04-20 22:48:56 s/sound's/sounds/ sorry for this terrible mistake. my only explanation is that it's getting late in PL, so I have to go sleep. 2016-04-20 22:49:02 gn 2016-04-20 22:49:37 and it actually should be sound without s, doh 2016-04-20 23:15:22 przemoc: ahhhh almost worst than the space before ? 2016-04-20 23:15:25 :D 2016-04-20 23:15:53 worse* 2016-04-20 23:15:55 ^^ 2016-04-21 07:19:43 barthalion: i am interested on your comment on this: http://lists.alpinelinux.org/alpine-devel/5254.html 2016-04-21 07:20:03 the suggestion is to rename python -> python2 and python3 -> python 2016-04-21 07:20:12 i think you originally did that in arch linux 2016-04-21 07:20:16 but later renamed it again? 2016-04-21 07:26:17 ncopa, i got notified about py-lxml upgrade from 3.4.4 to 3.5.0 for branch 3.3. 2016-04-21 07:26:22 This is a major release 2016-04-21 07:26:34 so i don't think we do this backport, right? 2016-04-21 07:27:00 we normally dont do that, yes. 2016-04-21 07:27:07 ok 2016-04-21 07:27:08 unless there are very good reasons to do so 2016-04-21 07:27:17 got think about how to reply... 2016-04-21 07:27:33 https://pkgs.alpinelinux.org/flagged# 2016-04-21 07:27:34 but you have to check the py-lxml 2016-04-21 07:27:57 what the update includse 2016-04-21 07:28:05 might be it is 100% backwards compat 2016-04-21 07:28:10 then it would be okish 2016-04-21 07:28:27 specially if there are good reasons 2016-04-21 07:28:36 it might be that the 3.4.4 is completely useless 2016-04-21 07:28:41 then i'm ok with upgrade 2016-04-21 07:28:57 ACTION is reading https://pkgs.alpinelinux.org/flagged# 2016-04-21 07:29:04 http://lxml.de/3.5/changes-3.5.0.html 2016-04-21 07:29:07 wrong paste :) 2016-04-21 07:29:56 might be py-lxml-3.4 only works with python-3.4 and not 3.5 2016-04-21 07:30:03 then we need upgrade it 2016-04-21 07:30:09 sounds correct to upgrade it 2016-04-21 07:30:28 ok 2016-04-21 07:30:44 btw...has been released 3.6.0 for edge 2016-04-21 07:33:34 ncopa, how do we got notifies about CVE re : $package ? 2016-04-21 07:33:39 *notify 2016-04-21 07:34:11 emaiml from bugtracker? 2016-04-21 07:35:31 there's someone that do this check? 2016-04-21 07:35:56 reporting issues? 2016-04-21 07:35:58 yes 2016-04-21 07:36:10 yes, reporting issues. Ok. 2016-04-21 07:39:38 i wonder if we can remove ruby-* packages 2016-04-21 07:39:42 and only geep ruby 2016-04-21 07:39:44 keep* 2016-04-21 07:39:49 and let people use gem install 2016-04-21 07:39:59 i think they are mostly broken anyway 2016-04-21 07:43:52 i've been working on metasploit on alpine with docker, ruby packages are a nightmare. 2016-04-21 07:45:39 i think we really need to fix the python > python3 thingy 2016-04-21 07:46:01 scadu volunteered to do the job 2016-04-21 07:46:09 we just need to figure out how we want it 2016-04-21 07:47:04 i vote for py2 and py3, basically just to avoid to rewrite the APKBUILD logic when python4 will be available 2016-04-21 07:47:10 my current thinking is that the best would be to have python2 + python3 and py2-* + py3-* 2016-04-21 07:47:25 +1 2016-04-21 07:47:27 ideally the py2- and py3- would be built from same apkbuild 2016-04-21 07:47:32 similar to lua-* 2016-04-21 07:47:32 yes, like lua 2016-04-21 07:47:46 i've the same thought 2016-04-21 07:47:51 however, i dont know how easy that is 2016-04-21 07:47:55 i mean 2016-04-21 07:47:59 its pretty big job 2016-04-21 07:48:09 arch does 2016-04-21 07:48:21 and currently, i think the job is so big that it will prevent it to happen at all 2016-04-21 07:48:40 so i wonder if we should simply add py3-* apkbuilds 2016-04-21 07:49:07 that would increase a lot the number of packages 2016-04-21 07:49:13 yes 2016-04-21 07:49:14 hm 2016-04-21 07:49:22 especially the ones working with both python version 2016-04-21 07:49:28 not only py-* 2016-04-21 07:49:37 but other packages relying on py-* 2016-04-21 07:49:50 oh true 2016-04-21 07:49:53 those are the problem 2016-04-21 07:49:57 right 2016-04-21 07:49:59 the dependencies 2016-04-21 07:50:30 so you would have $package with python2 and python3. How would you name $package ? 2016-04-21 07:50:43 No, it can't work 2016-04-21 07:50:49 $package? 2016-04-21 07:50:51 like what? 2016-04-21 07:51:27 like 2016-04-21 07:51:29 avahi-ui 2016-04-21 07:51:33 apache2-mod-wsgi 2016-04-21 07:51:36 ansible 2016-04-21 07:51:38 etc 2016-04-21 07:51:45 will ansible work with python3? 2016-04-21 07:52:00 this specific package dunno, but some of them are. 2016-04-21 07:52:12 they can work with both py versions 2016-04-21 07:52:25 so how would you name that? 2016-04-21 07:52:46 avahi-ui for example 2016-04-21 07:52:53 is an application 2016-04-21 07:52:57 right? 2016-04-21 07:53:08 yes. 2016-04-21 07:53:25 it does not really matter which python version it uses, as long as the app works 2016-04-21 07:53:39 in that case it does not make sense to have a python2 + python3 variant 2016-04-21 07:53:49 we just pick one python version and stick to it 2016-04-21 07:54:13 ok. Right. 2016-04-21 07:54:25 We need to rewrite APKBUILD with the python version it works 2016-04-21 07:54:33 yes 2016-04-21 07:54:38 and fix the deps that way too 2016-04-21 07:54:43 speaking of avahi 2016-04-21 07:54:45 it needs an update 2016-04-21 07:54:48 yes. 2016-04-21 07:54:50 but new version does not build 2016-04-21 07:55:39 seems like ansible does not support python3 2016-04-21 07:55:58 i'm wondering how many things will break when we will rename the py packages 2016-04-21 07:56:09 from py-* to py2-* and/or py3-* 2016-04-21 07:56:18 i think alot will break 2016-04-21 07:56:45 which is why i am starting to think that we could append the py3-* apkbuilds 2016-04-21 07:56:54 makes sense having a transition with py-* metapackage ? 2016-04-21 07:57:40 i dont want py-*/APKBUILD + py2-*/APKBUILD + py3-*/APKBUILD 2016-04-21 07:57:45 if thats what you mean 2016-04-21 07:57:48 but what we could do 2016-04-21 07:57:56 right. But py-*/APKBUILD will be removed 2016-04-21 07:58:01 rename py-* to py2-* and add "provides=py-*" 2016-04-21 07:58:07 ah 2016-04-21 07:58:13 good point. 2016-04-21 07:58:14 Yes. 2016-04-21 08:01:03 181 APKBUILDs has py-[a-z] match 2016-04-21 08:01:38 of those 131 is pkgname=py-* 2016-04-21 08:02:06 that is in main only 2016-04-21 08:02:21 i think we cannot break compat 2016-04-21 08:02:56 right 2016-04-21 08:03:53 ok 2016-04-21 08:03:58 it's not that complicate then 2016-04-21 08:05:18 a simple test: http://tpaste.us/2KnY 2016-04-21 08:05:21 did not work at all 2016-04-21 08:06:02 does py-imaging work with python3 ? 2016-04-21 08:31:41 apparently not 2016-04-21 08:48:54 gradm -E fails to start on edge 2016-04-21 08:49:14 here's a log: http://sprunge.us/UgRD 2016-04-21 08:49:20 (config wasn't edited, of course) 2016-04-21 10:00:27 skrzyp: would be nice if you could help us with a suggestion for a fix 2016-04-21 10:53:39 ncopa: but I'm not an grsec expert ;__; 2016-04-21 12:38:38 google is clearly not using alpine or busybox ;) 2016-04-21 12:38:39 tar xzvfC /tmp/AndroidStudio1FirstRun/android-sdk_r22.6.2-linux.tgz /tmp/AndroidStudio1FirstRun/android-sdk_r22.6.2-linux.tgz-unpacked 2016-04-21 12:38:40 tar: can't change directory to 'f': No such file or directory 2016-04-21 12:38:40 Unable to unpack file android-sdk_r22.6.2-linux.tgz: Unable to unpack archive file. 2016-04-22 09:56:34 hey :) 2016-04-22 10:42:38 hi 2016-04-22 10:42:45 hi 2016-04-22 11:12:45 a very interresting read: https://tonyarcieri.com/would-rust-have-prevented-heartbleed-another-look 2016-04-22 11:14:16 :) 2016-04-22 11:14:54 btw how it looks with Rust in Alpine? I’m just learning Rust and writing some program that I’d like to use on Alpine… 2016-04-22 11:15:15 # compile on Alpine 2016-04-22 11:15:28 hahahaha, ncopa i swear i did not put jirutka up to it !!!! 2016-04-22 11:16:04 that's a lie, a dirty lie 2016-04-22 11:16:13 I know there's collusion between you two! 2016-04-22 11:16:43 That’s right, I’m interested in Rust for a long time. 2016-04-22 11:18:28 So where’s the problem with Rust on Alpine? 2016-04-22 11:21:27 leo-unglaub: Do you know about https://github.com/uutils/coreutils? It looks very promising :) 2016-04-22 11:22:12 jirutka: thats an inside joke ;) 2016-04-22 11:22:18 i want rust on alpine very bad 2016-04-22 11:22:27 the problem is, that it is not buoldable at the moment 2016-04-22 11:22:37 buoldable -> buildable 2016-04-22 11:22:43 I like buoldable 2016-04-22 11:22:47 why not? 2016-04-22 11:23:05 something with llvm and musl, I'd guess 2016-04-22 11:23:43 hm, Rust should be supported on musl, just not guaranted 2016-04-22 11:24:29 https://github.com/rust-lang/rust/issues/28667 2016-04-22 11:25:13 yeah, that's it 2016-04-22 11:25:27 gotta love projects that *embed* code instead of *depending on* it 2016-04-22 11:25:47 they embed llvm?! why the hell? 2016-04-22 11:26:02 jirutka: because they protect against trusted trust attacks 2016-04-22 11:26:09 a little bit overkill in my mind, but okay .. 2016-04-22 11:26:44 between trusting the system's llvm and duplicating maintenance, they chose to duplicate maintenance, which makes trust more than twice harder 2016-04-22 11:26:47 those people are SMART 2016-04-22 11:27:17 calm down, people behind Rust are smart 2016-04-22 11:27:46 that said, I agree that this is quite bad solution for the trust problem 2016-04-22 11:27:54 they are very smart ... in theory its a good idea 2016-04-22 11:27:55 lots of software people are smart - doesn't mean they're smart for all aspects of software 2016-04-22 11:28:11 true 2016-04-22 11:29:00 a surprisingly high amount of otherwise smart people just don't realize the cost of duplicating maintenance 2016-04-22 11:29:02 I have to go for lanch… see ya 2016-04-22 11:31:55 leo-unglaub: man, you're a cutie (saw your github profile picture) 2016-04-22 11:32:25 skarnet: well, my only good picture ;) 8 years old *g* 2016-04-22 11:32:54 don't worry about it, my only good picture is 10 years old. And black and white. XD 2016-04-22 11:33:31 yeah :) 2016-04-22 12:29:54 skarnet: sounds like they are smart but unexperienced 2016-04-22 12:30:10 yeah 2016-04-22 12:30:40 when they are old enough to realize it, new smart people show up 2016-04-22 12:30:45 and take over the show 2016-04-22 12:31:20 skarnet: do you remember busybox init reap children problem? 2016-04-22 12:31:32 which gave a delay of 1 sec 2016-04-22 12:31:42 the old guard won't let young whippersnappers take over the show! where's my cane? 2016-04-22 12:31:49 yes, I remember it 2016-04-22 12:31:52 what about it? 2016-04-22 12:32:03 do you have any idea how to fix buysbox init? 2016-04-22 12:32:26 apart from turning it into a real state machine, no 2016-04-22 12:32:46 ok. thanks 2016-04-22 12:32:53 honestly it would need much tearing apart 2016-04-22 12:33:20 can you hold out with bb init for 10 more months 2016-04-22 12:33:22 ? 2016-04-22 12:33:32 i suppose so 2016-04-22 12:34:57 when you have a state machine but want to avoid writing it as a real state machine and take shortcuts, problems *always* arise :/ 2016-04-22 12:35:32 it's the case with bb init, it's the case with all existing implementations of printf, it's even the case with djb's supervise 2016-04-22 14:56:48 isaac dunham impress me 2016-04-22 14:56:54 is he here? 2016-04-22 14:58:49 I don't think so. But yes, he's a competent guy :) 2016-04-22 15:08:19 ncopa: adduser/addgroup improvements got delayed, because 1) I couldn't automate the changes as much as I wanted initially, so the stats in my mail were a bit off, 2) couldn't work on it in one go, 3) started making all of them more consistent (w/o changing options, though). in the end there will be more than one commit, because it was the only sensible way to make changes reviewable and I would lik 2016-04-22 15:08:25 e them to be reviewed and not blindly applied, because in during later commits I was finding incompleteness in previous ones and had to amend them... I haven't mailed them yet, as I'm doing review during `git add -p` for the last (7th) commit now. 2016-04-22 15:11:04 we can take it in chunks 2016-04-22 15:11:09 10 at the time if you want 2016-04-22 15:11:24 or similar 2016-04-22 15:11:39 all of these commits work on whole tree 2016-04-22 15:13:23 and I don't mess with pkgrel in any of these 2016-04-22 15:41:01 i think you need bump pkgrel 2016-04-22 15:41:07 or package will not be rebuilt 2016-04-22 15:44:59 ok its weekend 2016-04-22 15:45:03 have a nice weekend everyone 2016-04-22 15:45:09 not all of my changes give behavioral changes 2016-04-22 15:46:31 question is whether such packages should also be rebuilt just because of cleaned scripts? 2016-04-22 15:47:24 or whether they can wait till software upgrade? 2016-04-22 15:48:04 the restuling .apk will be different 2016-04-22 15:48:13 that is enough for bumping pkgrel 2016-04-22 15:50:47 ok, so one bumping commit for each pkg is ok? 2016-04-22 15:51:00 one commit per one pkg I meant 2016-04-22 15:51:32 I'll leave bumping after the whole tree changes 2016-04-22 15:51:45 ACTION is vanishing for 1,5 hour 2016-04-22 15:53:01 one commit per package is cool. it would be much easier to include over9000 changes in one commit, but it is a mess. 2016-04-22 17:35:30 mass changes were done in 6 commits, it's only pkgrel bumping that will happen per package. 2016-04-22 17:41:16 and because it will be a lot of them, I think I'll send only those 7 (6 + 1 fix) patches to the -aports with cover letter pointing to place where pkgrel bumping patches can be obtained. there will be 188 of them. 2016-04-22 17:42:13 4 would be in unmaintained part, so they can be dropped I guess 2016-04-22 17:45:29 scadu: I'm always against dumping stuff in one commit and prefer one commit per logical change if possible. that's why I didn't put all changes in 1 commit, but 6 of them. reviewing is also easier that way. git --color-words is very helpful in checking that kind of changes. 2016-04-22 17:53:05 przemoc: I see. 2016-04-22 17:53:31 przemoc: I thought there is much more stuff in single commit ;) 2016-04-22 18:09:55 well, they touch lot of files, so in a sense they're cannot be called small ;) 2016-04-22 18:43:55 I truly noticed it a bit late, but there are 2 packages coexisting in main and testing: zabbix (3.0.2, 3.0.0_rc2) and gdnsd (1.11.5, 2.2.2). in case of gdnsd there is a point in having newer version in testing, but what's the point of old rc version of zabbix in testing? 2016-04-22 19:01:31 oh no, some movements in aports... 2016-04-22 19:01:37 another rebase needed 2016-04-23 07:05:33 Good morning. 2016-04-23 07:08:26 I am trying to compile hyperd (from github) and reportedly a problem with musl was encountered. Posted an issue to https://github.com/golang/go/issues/15394 which pointed to https://github.com/golang/go/issues/13678. The latter categorically states that it was because of the bug in libc. Since I am trying to build in alpine which uses musl, how can this be addressed? Any inputs appreciated! 2016-04-23 07:10:10 Go-lang people categorically say it is not a go-lang issue, but due to "an assembler that generates a R_X86_64_REX_GOTPCRELX relocation but your linker doesn't understand it., fyi. 2016-04-23 11:27:01 o/ 2016-04-23 11:30:30 clandmeter: what's the state of threading messages on http://lists.alpinelinux.org/? somehow I thought it was supported, at least some Re: are indented with '-->', but my mails sent in reply to cover letter (using its Message-Id in In-Reply-To, References) are not. 2016-04-23 13:14:32 please don't apply the xpra patch - I have a bunch of other fixes to add to it 2016-04-23 16:34:37 Repost: I am trying to compile hyperd (from https://github.com/hyperhq/hyper) and reportedly a problem with musl was encountered. Posted an issue to https://github.com/golang/go/issues/15394 which pointed to https://github.com/golang/go/issues/13678. The latter categorically states that it was because of the bug in libc. Since I am trying to build in alpine which uses musl, how can this be addressed?Go-lang people categorically 2016-04-23 17:19:16 hey :) 2016-04-23 17:19:22 how is everyone? 2016-04-23 17:46:25 ncopa: I'll reply to this on mailing list 2016-04-23 17:46:45 but no, I didn't handle python migration in Arch, I think I did lua 2016-04-23 18:22:27 my linux syscall knowledge is a bit rusty ... whats currently the best way to get a random value from the kernel? 2016-04-23 18:22:47 is it still getrandom() 2016-04-23 18:23:04 or did they finally bring in the better OpenBSD arc4random one? 2016-04-23 18:23:10 leo-unglaub: do you mean rusty, or Rusty? :P 2016-04-23 18:23:42 hehe, in this case i really mean rusty ;) 2016-04-23 18:24:05 i only did c on OpenBSD for the last 4 years ... 2016-04-23 18:26:46 getrandom() was introduced only 1,5y ago in linux 2016-04-23 18:27:17 s/,/./ 2016-04-23 18:30:15 so its still the way to go? 2016-04-23 18:32:40 I'm not aware of any other way (beside manually opening and reading /dev/[u]random, obviously) 2016-04-23 18:34:53 nah, opening /dev/[u]random is highly dangerous ... 2016-04-23 18:35:08 so i am going with the getrandom() syscall 2016-04-23 18:36:19 dangerous is only fd exhaustion, but it is real, true, yet until 3.17 it was the only option 2016-04-23 18:37:57 fd exhausting is by far not the only danger to it 2016-04-23 18:38:15 chroot environments that just get a fake /dev/random linked in, ... 2016-04-23 18:38:20 and a lot of other shit ... 2016-04-23 18:40:35 ok, fair enough, I wasn't talking about fucked up setups. when there is getrandom() there is no longer any reason to go with dev-based way 2016-04-23 18:44:04 there is libbsd, which provides arc4random API (along with getentropy() based getrandom(), of course) 2016-04-23 18:46:52 przemoc: yeah, i thought about it but it just adds bloat ... and i am just rewriting a small rust programm in c so i can use it on alpine 2016-04-23 18:47:36 libbsd is there on alpine 2016-04-23 18:47:56 but it reminds me now :) back in school when we played a prank on our teacher and gave him a fake chroot for his presentation and linked in /dev/zero as /dev/random and then he had weak keys :) 2016-04-23 18:50:44 and while I understand your no-dep sentiment (I'm of similar mind), it's sometimes detrimental, as you may want to reimplement or embed everything yourself. then you can remain vulnerable to later discovered bugs that are fixed in libs, but not in your no-dep program. 2016-04-23 18:51:37 libraries are there to be used. if not, the go with stali instead of alpine :) 2016-04-23 18:53:02 s/the/&n/ 2016-04-23 18:59:18 re libbsd. IIRC they're still using those ugly glibc macros __BEGIN_DECLS in headers. not sure whether anyone reported it to them. I remember I was going to do it once, but had other stuff piled up and never did so... 2016-04-23 19:03:24 przemoc: in general i am 100% with you 2016-04-23 19:03:31 but my whole programm is 600 lines 2016-04-23 19:03:36 so its really fucking small 2016-04-23 19:03:43 its just a letsencrypt client 2016-04-23 19:05:03 so remember only to provide statically build binaries using musl for x86, x86_64 and arm in the end :) 2016-04-23 19:10:18 (and I'm not bashing stali, it seems interesting distro on its own, but haven't used it, though) 2016-04-23 19:12:01 so openbsd is your OS of choice, leo-unglaub? so why are you using AL too? 2016-04-23 19:12:06 has anyone succeeded already at running alpine on arm64 allwinner socs? 2016-04-23 19:12:26 przemoc: i use both 2016-04-23 19:12:30 depending on the usecase 2016-04-23 19:12:59 alpine is fast, clean and very small 2016-04-23 19:13:34 always use the right tool for the job ;) 2016-04-23 19:13:59 and openbsd is just clean ;) 2016-04-23 19:14:47 nooooo 2016-04-23 19:14:54 joking 2016-04-23 19:14:55 somehow I never really got into BSDs, even though I planned to 2016-04-23 19:15:08 openbsd is messy and bloaded ... but its damn secure and perfectly mixed together 2016-04-23 19:15:24 clean from security standpoint 2016-04-23 19:15:33 clear from a "thought about it" standpoint 2016-04-23 19:15:38 let me give you an example 2016-04-23 19:16:39 on linux you have ssh running and someone is bruce forcing on it. so ssh writes to /var/log about it. then you have to use something like file2ban, parse the logs and then execute iptables commands ... that almost always fails becasue log parsing is ugly 2016-04-23 19:16:45 the same case on openbsd looks like this 2016-04-23 19:16:58 the firewall has multiple source tables that it reads and watches for changes 2016-04-23 19:17:13 ssh writes and attack to that table and its automatically in the firewall 2016-04-23 19:17:28 thats what i call clean design of multiple toold working perfectly together 2016-04-23 19:18:18 sounds cool 2016-04-23 19:18:51 yeah, it is. the same with smtpd, ftpd, httpd, its very clean 2016-04-23 19:19:09 but if you get around to system tools like the package manager, the sys upgrades, ... the general speed 2016-04-23 19:19:12 fuck, thats horrible 2016-04-23 19:19:25 alpines package manager is 1000 times faster than the openbsd one 2016-04-23 19:19:58 well, but apk is 1000 times faster than _any_ package manager… 2016-04-23 19:20:09 yes, thats alpine awesomeness! 2016-04-23 19:20:31 i think i sayed that one in the past before. OpenBSD core with Alpine usertools ... 2016-04-23 19:20:38 and like infinitely faster than Portage :( 2016-04-23 19:20:38 that would be the optimal 2016-04-23 19:20:53 leo-unglaub: can you give links to docs or some tutorials on those BSD cool features, like the one mentioned. the kind of "best to sell it" 2016-04-23 19:21:06 s/\./?/ 2016-04-23 19:21:13 przemoc: thats not existing ... 2016-04-23 19:21:17 thats a huge problem 2016-04-23 19:21:24 you have to find it out yourself ... 2016-04-23 19:21:30 but all the man pages are awesome! 2016-04-23 19:21:36 :) 2016-04-23 19:21:38 they have huge pride in there man pages 2016-04-23 19:21:45 they are a lot more complete than the linux ones 2016-04-23 19:21:53 but the getting started it horrible 2016-04-23 19:22:02 man pages are great when you know what to look for 2016-04-23 19:22:49 if you don't know, then you may try some man -K with hope of finding anything relevant by keywords... 2016-04-23 19:22:53 I think that OpenBSD guyes simply don’t want new users… most of OpenBSD projects have sites like this http://www.openntpd.org/ … it’s like traveling in time back to '90 2016-04-23 19:23:28 przemoc: no, thats wrong 2016-04-23 19:23:42 they use man pages for everything, even for gereral purposes 2016-04-23 19:23:50 like man afterboot 2016-04-23 19:23:53 man afterinstall 2016-04-23 19:23:55 ... 2016-04-23 19:24:11 about the website design, i love it ... i hate all that new fancy shit 2016-04-23 19:24:16 so i love there website :) 2016-04-23 19:24:23 uh… okay… 2016-04-23 19:24:23 man beforeinstall? ;) 2016-04-23 19:25:19 thats a redirect to the mailinglist ... where you get called a noob and then you try it yourself anyway ;) 2016-04-23 19:26:10 I’ve already said that… they simple don’t want new users ;) 2016-04-23 19:26:32 no, they want new users ... some of them are just idiots .. 2016-04-23 19:26:40 they really work hard on getting rid of them! 2016-04-23 19:27:06 back in uni one of my groupmates was openbsd user. he was a sec freak too. his notebook needed CD with his keys and long password to boot. 2016-04-23 19:30:33 he also wrote his own OS with own GUI back in high school. he was working for Palm last time I heard of him, IIRC, RE their apps for l10n and stuff like that. he was a smart guy. 2016-04-23 19:30:43 he apparently passed OpenBSD test :> 2016-04-23 19:32:41 :) 2016-04-23 19:33:27 re apk, I also like it. apt-get is still acceptable. but yum, yum is not acceptable... 2016-04-23 19:43:12 kunkku: any chance of getting some awall improvements from you next week regarding custom rules? just asking. 2016-04-23 19:44:17 I'm talking about next week, because, as I guess, it's theoretically the last call before 3.4, which hopefully will be delayed as always :> [which will make more fixes possible] 2016-04-23 19:48:19 I 2016-04-23 19:48:31 'm afraid they won't make it to 3.4 2016-04-23 19:49:48 I'm too busy for the next 2 months 2016-04-23 19:49:58 np 2016-04-23 21:01:47 argh 3.4 next week ? 2016-04-23 21:02:12 ncopa: can we have the pigeonhole plugin integrated in dovecot for 3.4 ? 2016-04-23 21:02:15 :) 2016-04-23 21:02:23 please please please 2016-04-23 21:02:36 where should I send this 6 pack ? 2016-04-23 21:03:01 ncopa: what are the rules for promoting abuilds from testing to community or main? 2016-04-23 21:22:11 jirutka1: 1) it works 2016-04-23 21:22:27 it goes to community if we can't guarantee support for 2 years 2016-04-23 21:22:58 because upstream doesn't and/or we're not interested in doing it on our own 2016-04-23 21:23:09 ad 1) it must work even when it’s accepted to testing, shouldn’t it? 2016-04-23 21:23:34 for people without push access, guess so 2016-04-23 21:24:01 we have long history of pushing buildable packages that are broken… 2016-04-23 21:24:17 that’s really bad 2016-04-23 21:24:24 I'm joking 2016-04-23 21:24:46 but if someone requests a package, then he should be the one who verifies if it works 2016-04-23 21:26:05 I thought that it’s obvious to verify package even before sending patch to testing… 2016-04-23 21:26:39 well, we have people here that don't need to send patches 2016-04-23 21:26:42 I mean, I always verify that it actually works, at least breifly 2016-04-23 21:27:10 what do you mean with “don’t need to send patches”? 2016-04-23 21:27:15 and it's still requester's duty to test a package he wanted 2016-04-23 21:27:20 I mean push access 2016-04-23 21:27:55 jirutka1: requesters aren't always packagers and packager doesn't always test what (s)he adds thoroughly 2016-04-23 21:28:13 it's requester's duty 2016-04-23 21:28:50 genau 2016-04-23 21:30:52 sometimes requester vanishes (or doesn't even have AL, like me for a long time), but sometimes some other people come and say that pkg in testing works fine and: "please move it!" 2016-04-23 21:31:17 and then they're ignored and pkg lives in testing for eternity! 2016-04-23 21:31:19 HA! 2016-04-23 21:33:22 lol 2016-04-23 21:34:32 aha, I think that we have a little misunderstanding here… I don’t have push access (only to GitHub repo, that is just a mirror), so I’m in the role of the requester (if I understand your terminology correctly) 2016-04-23 21:36:34 and I’m conscientious, so it’s matter of course for me to verify packages I’m pushing to the repository :) that’s why I found it strange when you wrote that the package must work to be promoted to community or main 2016-04-23 21:37:06 yeah, I'm spoiled 2016-04-23 21:41:58 it never hurts to spell out what seems obvious 2016-04-23 21:42:31 you probably know as well as I do how often what seems obvious to us isn't to many people 2016-04-23 21:43:07 skarnet: well, that sadly true 2016-04-23 21:51:00 skarnet: so how is progressing the idea (and its implementation) of being able to easily use s6 alone in alpine? when we last chatted I suggested some discussion on -devel ML, but it never happened, did it? you're targetting 3.5? 2016-04-23 21:51:45 I'm targetting "some point in the future when I'm able to set aside a whole lot of fulltime to work on it and do it right" 2016-04-23 21:52:13 which basically means "when I can afford to stop working for a year or so" 2016-04-23 21:52:23 which should happen in 2017 2016-04-23 21:52:48 so 2016 is/will be that good for you? :) 2016-04-23 21:53:08 my original plan was to start in November-ish 2016 2016-04-23 21:54:18 but since I currently have a nice customer with a project that should help me have a clearer view of what's involved in doing this for Alpine 2016-04-23 21:55:14 and that once I leave his project to work on Alpine there's little chance he'll take me back 2016-04-23 21:55:29 I see 2016-04-23 21:55:39 so it will be around 3.6, good to know 2016-04-23 21:55:46 I'm inclined to binge on money as long as I can, and start working on the transition early 2017 2016-04-23 21:55:51 say Feb-Mar 2017 2016-04-23 21:56:22 the hard rule of free software funding :/ 2016-04-23 21:58:01 well, maybe not 3.6 but 4.0, it would be one of highlights ;) 2016-04-23 21:58:21 now, this is the *starting time* of the project 2016-04-23 21:58:40 yeah, yeah, I get it 2016-04-23 21:59:09 I don't know how much time it will take, but I'm guessing it could take up to a year, given how deep the openrc entanglement goes 2016-04-23 21:59:48 and since I want to give the user a choice of service managers rather than arbitrarily replace one with another 2016-04-24 08:57:43 morning 2016-04-24 12:01:10 question out of curiosity, barthalion, do you actually use default pl variant of pl keymap? I'm wondering about this default, as I always preferred left Alt to be usable for stuff other than inputting polish letters, because otherwise it becomes clunky (you know, reaching Esc for all shortcuts). 2016-04-24 12:15:52 przemoc: you mean xorg? yes, I do 2016-04-24 12:16:10 what shortcuts are available via Esc? 2016-04-24 12:16:13 no, I mean console 2016-04-24 12:16:20 I have pl2 for that 2016-04-24 12:16:34 but I usually forgot to change keymap in console anyway 2016-04-24 12:16:47 I don't spend much time there that requires me to use polish accents 2016-04-24 12:17:13 actually I don't remember last time when I had to enter such in tty 2016-04-24 12:34:21 I have physical machine that won't have X (at least not in base install, maybe in some lxc later), so I switched to pl1, for better Alt, but you may be right, that I used to use pl2 in the past on other distros. I thought pl1 was good enough from quick test. there is a comment in pl2 "changed by Adam Tla/lka to obtain working left Alt normally", but I'm not sure yet what works abnormally in pl1 2016-04-24 12:36:24 I checked that in the old Slack I have pl and pl2 only 2016-04-24 15:59:18 no idea about the difference, sorry 2016-04-24 15:59:39 there was some long time ago, but I forgot since then 2016-04-24 17:39:27 hey :) 2016-04-24 17:42:37 hey, how is electrolysis progressing, finished? 2016-04-24 17:45:11 electrolysis is used for rust removal, if you didn't catch that 2016-04-24 17:49:20 haha 2016-04-24 17:49:30 thanks for the hint, because i really did not get it .. 2016-04-24 17:53:49 s/ is / can be / (as it has other uses too) 2016-04-24 17:54:33 przemoc: i worked all night at the red cross ... so i did not sleep ... maybe with some sleep i would have gotten it :) 2016-04-24 17:57:34 oh, altruist, how noble (no sarcasm) 2016-04-24 18:06:11 nah, just a simple paramedic ;) 2016-04-24 18:14:57 still noble :) 2016-04-24 22:31:36 does anyone tested irssi's perl support in alpine? /LOAD returns no perl. perhaps some perl-* pkgs are needed beside perl? 2016-04-24 22:48:10 s/does anyone tested/did anyone test/ too many mistakes, too late, time to sleep 2016-04-24 23:16:02 przemoc: i have it working, also have some perl-* pkgs beside perl, but no clue if any of those are related to irssi 2016-04-24 23:25:22 please ignore the xpra patches sent to alpine-devel in error - have resent the mails to aports 2016-04-25 07:25:38 przemoc: thank you for the great pre-install cleanup patches 2016-04-25 07:25:42 good job! 2016-04-25 08:31:59 hey :) 2016-04-25 08:43:06 hi :) 2016-04-25 09:17:34 hi 2016-04-25 09:25:27 morning :) 2016-04-25 09:35:08 a bit OT question. I need a pdf with a drawing where the mount holes (under) for 2.5 disk i located 2016-04-25 09:35:36 the idea is that i will make 4 holes in a wandboard case 2016-04-25 09:35:53 and attach the 2.5 disk directly to the case 2016-04-25 09:39:18 morning 2016-04-25 09:47:04 don't mix it up - don't make the holes into the disk 2016-04-25 09:47:12 :) 2016-04-25 09:54:39 tinc needs an update, personally I would prefer to 1.1-pre12 2016-04-25 09:54:59 if there are objections then I will bump it to 1.0.28 2016-04-25 10:03:41 do you know what the ETA for 1.1 is? 2016-04-25 10:03:59 if its expected in a week or two then we can go for 1.1.-pre12 2016-04-25 10:04:34 if its 2+ months then maybe do 1.0.28 (which i just pushed) 2016-04-25 10:08:42 no ETA, it's been pre for years 2016-04-25 10:09:31 1.1-pre has all the new features, and it's 1.0 backwards compatible (except for bugs) 2016-04-25 10:11:31 I don't mind staying on 1.0.x, I will just to roll my own 2016-04-25 10:12:00 this weekend i have created new aports for crda/wireless-regdb to get hostapd working on 5ghz. can i submit them or are there any concerns? i am asking because there is #3910 which has never been merged… 2016-04-25 10:41:59 i think this is a must have package for alpine linux :) way more important than rust at the moment :) http://goodies.xfce.org/projects/panel-plugins/xfce4-eyes-plugin 2016-04-25 10:42:20 lol 2016-04-25 10:45:39 hi 2016-04-25 10:45:42 you're welcome, ncopa. I agree that adding users/groups (and their home dirs, if needed) should become part of APKBUILD, and pre-install/upgrade scripts fallback should be used only for some strange/bizzare cases. 2016-04-25 10:57:12 przemoc: the problem is that the current support in APKBUILD doesn’t allow to specify home dir, shell, … any parametrs for new user/group 2016-04-25 11:08:06 I know that, I wrote "should become" meaning that stuff allowing that should be added in future to apk tooling. 2016-04-25 11:44:39 I would also suggest introducing some kind of defuidgid mapping (in file?), so we could maintain clean uid/gid for all known installable services for new installations, to make AL even more deterministic (i.e. making order of package installation irrelevant) 2016-04-25 11:46:05 why uid/gid matters? 2016-04-25 11:49:43 it doesn't matter per se, i.e. everythings works whatever uid/gid you have behind name as long as you use the name, but it's nice to have same ids for same services on different hosts if possible. obviously older hosts upgrading to AL having such feature wouldn't benefit from such default mapping, but new installation would. 2016-04-25 11:51:23 As perfectionist I agree that it’s a nice to have same UID/GID across hosts, but to be honest… does it really matter? 2016-04-25 11:51:57 I mean, is it really worth it? 2016-04-25 11:54:35 AL strives for minimalism and cleanliness. I haven't seen that kind of global mapping (that would be used during adding user/group) on other distros, but I think it's at least worth to be considered. such thing would be possible (well, sensible is better word) only after introducing user/group-stuff into APK. we could have one file package that would provide such default "suggested" mapping and ap 2016-04-25 11:54:41 k would use that. 2016-04-25 11:55:57 Gentoo uses predefined gid/uid for _some_ packages. 2016-04-25 11:56:22 However I don’t know how it handles possible conflicts 2016-04-25 12:42:02 arm builder doesn't look in a good shape.. 2016-04-25 13:23:37 is it possible to boot AL iso without quiet provided to kernel or with noquiet provided to it? I mean w/o creating custom ISO. 2016-04-25 13:25:15 inital v3.4 packages: http://fr.alpinelinux.org/alpine/v3.4/main/armhf/ 2016-04-25 13:25:38 przemoc: it should be possible to type "noquiet" at boot prompt 2016-04-25 13:25:56 but i think isolinux has a bug or something 2016-04-25 13:27:06 bah 2016-04-25 13:27:09 definitely I couldn't get noquiet working from isolinux in 3.3.3 (not sure about older releases) 2016-04-25 13:27:16 i can't understand what's happening with build server 2016-04-25 13:29:03 and I haven't gone through creating custom ISOs yet (as my migration from old slack is still not complete) 2016-04-25 13:29:13 noquiet could be a bit helpful in case of #5479 2016-04-25 13:29:47 fcolista: which of them 2016-04-25 13:30:09 i set up build-3-4-armhf before the weekend 2016-04-25 13:30:18 i think some of the deps are failing 2016-04-25 13:30:22 now prosody and lizardfs 2016-04-25 13:30:32 i can give you a list 2016-04-25 13:30:41 http://tpaste.us/GxB7 2016-04-25 13:30:55 but what's happening is that while i fix them, other packages pops up unable to build 2016-04-25 13:31:06 yes 2016-04-25 13:31:20 e.g. prosody: 2016-04-25 13:31:29 now a patch that was there now no longer applies 2016-04-25 13:31:36 huh? 2016-04-25 13:31:41 yes 2016-04-25 13:32:01 prosody.cfg.lua.patch 2016-04-25 13:32:36 testing/[various]: bump pkgrel for pre-install fixes 2016-04-25 13:32:41 uff 2016-04-25 13:33:18 how it can conflict even? 2016-04-25 13:34:16 patching file /home/ncopa/aports/testing/prosody/pkg/prosody/etc/prosody/prosody.cfg.lua 2016-04-25 13:34:16 Reversed (or previously applied) patch detected! Assume -R? [n] 2016-04-25 13:34:37 i know what happens 2016-04-25 13:34:47 its picked up by the new default_prepare 2016-04-25 13:36:43 fcolista: if you want help clean up: http://tpaste.us/GxB7 2016-04-25 13:36:57 oi 2016-04-25 13:36:58 ok 2016-04-25 13:37:17 build logs are in http://build.alpinelinux.org/buildlogs/build-3-4-armhf/main/ 2016-04-25 13:37:34 i think a bunch of the build errors comes from missing deps 2016-04-25 13:37:55 so prioritize libs 2016-04-25 13:38:10 automagic stuff always strikes back! ;) 2016-04-25 13:38:19 yeah 2016-04-25 13:38:26 it was kind of expected 2016-04-25 13:38:53 but i think it was a bad thing to apply patch in package() anyways 2016-04-25 13:39:02 inconsistent 2016-04-25 13:39:30 prepackage() would sound better 2016-04-25 13:43:25 ncopa: is anyone beside you familiar with setting up builders similarly to what is done on build.a.o, if one would like to set up local builders for himself? 2016-04-25 13:44:35 short version: apk add aports-build 2016-04-25 13:45:44 almost sounds too good to be fully true 2016-04-25 13:45:47 edit /etc/abuild.conf, create a build user and make sure it has abuild keys (created by abuild-keygen) 2016-04-25 13:46:12 and you may need to edit /etc/conf.d/mqtt-exec.aports-build 2016-04-25 13:46:35 there you set which git branch you want monitor etc 2016-04-25 13:46:40 https://github.com/andyshinn/docker-alpine-abuild looks cool 2016-04-25 13:46:55 i have been looking for something similar 2016-04-25 13:47:25 -e RSA_PRIVATE_KEY="$(cat ~/.abuild/mykey.rsa)" \ 2016-04-25 13:47:34 will leak the signing key to local users 2016-04-25 13:47:53 it will be visible in 'ps xa' list 2016-04-25 13:48:07 andyshinn: ^ 2016-04-25 13:48:23 didn't use it anyway :f 2016-04-25 13:48:42 ncopa, how do you want to cleanup awstat? 2016-04-25 13:48:54 *awstats ? 2016-04-25 13:49:03 removing prepare() ? 2016-04-25 13:49:19 create patch in behalf of sed ? 2016-04-25 13:49:33 what is the error? 2016-04-25 13:49:36 or moving in /usr/share/webapps ? 2016-04-25 13:49:37 http://build.alpinelinux.org/buildlogs/build-3-4-armhf/main/awstats/awstats-7.5-r0.log 2016-04-25 13:49:46 looks like checksum error 2016-04-25 13:49:51 oh, ok 2016-04-25 13:49:58 but 2016-04-25 13:50:00 i wonder why 2016-04-25 13:50:09 dont just update the checksum 2016-04-25 13:50:13 i was thinking you wanted to refactor the APKBUILD 2016-04-25 13:50:29 it had error: http://build.alpinelinux.org/buildlogs/build-3-4-armhf/main/awstats/awstats-7.5-r0.log 2016-04-25 13:50:40 yes, ok 2016-04-25 13:50:42 that list is packages that was not built for some reason 2016-04-25 13:50:49 ok gotcha 2016-04-25 13:50:56 this is an interesting case though 2016-04-25 13:51:02 the tarball has changed 2016-04-25 13:51:37 "last minute" fixes, perhaps? 2016-04-25 13:51:42 7.5 is beta 2016-04-25 13:51:46 7.4 is stable 2016-04-25 13:51:57 i wonder if you could grab the current tarball from http://distfiles.alpinelinux.org/distfiles 2016-04-25 13:52:00 so might be that is actually a fixes 2016-04-25 13:52:06 and compare 2016-04-25 13:52:11 might have fixes yes 2016-04-25 13:52:13 hum 2016-04-25 13:52:35 why they dont call the tarball -beta1.tar.gz -beta2.tar.gz then? 2016-04-25 13:52:38 oh well 2016-04-25 13:53:20 not everyone pays attention to such "minor" details. it's probably perpetual beta, until it becomes final release. ;) 2016-04-25 13:53:38 Beta version is 7.5 - 2016-04-11 14:55 2016-04-25 13:53:38 This may be a not stable version. I put here for download for test purpose. 2016-04-25 13:53:46 yup 2016-04-25 13:53:47 ok 2016-04-25 13:53:50 we cannot use that 2016-04-25 13:53:54 we should go back to 7.4 2016-04-25 13:54:40 +1 ncopa 2016-04-25 13:57:13 sh: ./APKBUILD: line 16: netsurf.sh: not found 2016-04-25 14:23:07 https://dpaste.de/ObLm#L8,11 2016-04-25 14:23:15 ncopa, gexiv2 fails with fortify 2016-04-25 14:28:10 i have not read the entire backlog 2016-04-25 14:28:17 but what is going on with the arm builder? 2016-04-25 14:31:11 fcolista: i dont think its fortify 2016-04-25 14:31:25 seems like libtool error: /usr/bin/libtool: line 2460: printf: write error: Bad file descriptor 2016-04-25 14:31:41 or maybe /dev/null is wrong 2016-04-25 14:32:31 bah 2016-04-25 14:32:35 leo-unglaub: i set up build-3-4-armhf 2016-04-25 14:32:40 last week 2016-04-25 14:32:48 its build errors for 3.4 2016-04-25 14:32:49 it fails also in figure out the right python version :/ 2016-04-25 14:33:11 ah, i see! 2016-04-25 14:33:12 thanks! 2016-04-25 15:09:49 ncopa, interestin: 2016-04-25 15:09:50 /usr/bin/libtool: line 2460: printf: write error: Bad file descriptor 2016-04-25 15:10:04 libtool on row 2460: 2016-04-25 15:10:05 printf '%s\n' "$progname $scriptversion" 2016-04-25 15:10:31 maybe is redirected to a wrong FD ? 2016-04-25 15:10:45 ls -l /dev/null 2016-04-25 15:10:45 crw-rw-rw- 1 root root 1, 3 Apr 15 06:56 /dev/null 2016-04-25 15:10:50 /dev/null exists 2016-04-25 15:11:09 echo test > /dev/null 2016-04-25 15:11:09 x86_64-edge:~/aports/main/gexiv2$ 2016-04-25 15:21:35 i dont know why 2016-04-25 22:07:24 hey friends :) 2016-04-25 22:07:27 how are you? 2016-04-25 22:57:15 hi :) 2016-04-25 22:58:59 is there any reason why the alpine syslog-ng pkg splits the config and generates the real config on startup? 2016-04-25 23:08:45 it doesn't make sense for me, e.g. if i want to add remote logging like "source net { udp(); };" i can't add it to the source config file because it is wrapped by "source s_all { ... };". therefore i have to put it into destination, filter or log config file which makes the naming schemes useless. 2016-04-25 23:13:47 i would prefer using @include and give full control over the structure of the config files to the admin 2016-04-25 23:43:32 stwa: i am not sure about your problem since i dont use remote logging 2016-04-25 23:43:49 but in general config is trimmed out of packages to keep them small 2016-04-26 07:11:19 ncopa, iv'de done with nelinux.org/buildlogs/build-3-4-armhf/community/libvirt-glib/libvirt-glib-0.2.3-r0.log 2016-04-26 07:11:30 sorry: http://tpaste.us/GxB7 2016-04-26 07:43:52 bah 2016-04-26 07:43:57 mlocate is still broken 2016-04-26 07:44:01 and ruby stuff as well 2016-04-26 07:48:46 leo-unglaub: but there are 9 config files inside the package 2016-04-26 07:54:15 fcolista: i'm thinking of removing ruby-everything 2016-04-26 07:54:21 ruby-* 2016-04-26 07:54:27 not ruby itself 2016-04-26 07:54:37 that would not be a bad idea. Let ruby install dependencies 2016-04-26 07:55:24 but i think that stuff will break. For each ruby package we need to ship also alpine-sdk to allow a successfull build 2016-04-26 08:39:32 fcolista: wouldn't build-base be enough for that? 2016-04-26 08:41:03 I doubt ruby mess needs abuild, cdrkit or mkinitfs for instance, but you never know. ;) 2016-04-26 08:41:58 przemoc, yes might be enough 2016-04-26 10:08:15 ncopa: i dont know about you, but as someone who is deeply interrested in security i get really horney when i read that: https://marc.info/?l=openbsd-tech&m=146159002802803&w=2 2016-04-26 10:09:30 greetings 2016-04-26 10:09:36 fellows 2016-04-26 10:09:48 when is our next alpine release being planned 2016-04-26 10:19:24 oneinsect_, this w/e ncopa has configured builder for 3.4. So, very soon 2016-04-26 10:19:52 excellent fcolista: 2016-04-26 10:19:57 Dunno which blockers we have so far 2016-04-26 10:20:06 none seems to be interested in releasing alpine for other arm distros 2016-04-26 10:20:07 i think python2/3 2016-04-26 10:21:08 hmmm 2016-04-26 10:34:15 we still have not determined details in python2/3 case. 2016-04-26 10:34:53 not yet, right 2016-04-26 13:09:49 hi 2016-04-26 13:09:59 do we have i2c tools 2016-04-26 13:10:01 available? 2016-04-26 13:10:04 package* 2016-04-26 13:13:13 yes https://pkgs.alpinelinux.org/package/edge/testing/x86_64/i2c-tools 2016-04-26 13:28:36 okie great there is an arm package 2016-04-26 13:28:37 lovely 2016-04-26 16:02:49 friends 2016-04-26 16:02:58 bootargs earlyprintk /uimage modules=loop,squashfs,sd-mod,usb-storage modloop=/modloop-sunxi-img console=${console} 2016-04-26 16:03:05 is there a way i can add a rootwait 2016-04-26 16:03:09 somewhere in the above 2016-04-26 16:03:15 in the bootargs??? 2016-04-27 07:21:40 morning 2016-04-27 09:28:51 rnalrd, http://sprunge.us/RPGF 2016-04-27 09:29:11 virt-manager needs virtlogd to work 2016-04-27 09:29:30 this is the same approach gentoo uses 2016-04-27 09:29:38 it's fine with you if i push this change? 2016-04-27 09:30:05 fcolista, I'm listed as maint, but I'm not really maitaining it 2016-04-27 09:30:10 ask @ncopa 2016-04-27 09:31:58 rnalrd just get the credit :) 2016-04-27 09:32:04 :D 2016-04-27 09:32:56 fcolista: i'm ok with it. just remember to push the files too (they are not included in the diff you posted) 2016-04-27 09:33:05 yes 2016-04-27 09:35:11 lol 2016-04-27 09:35:12 http://git.alpinelinux.org/cgit/aports/stats/ 2016-04-27 09:35:25 for the first time in the history, someone else has committed more than ncopa 2016-04-27 09:35:32 wow! 2016-04-27 09:35:42 congrats! 2016-04-27 09:36:03 ACTION blushes 2016-04-27 09:36:41 :q 2016-04-27 09:42:45 re mlocate 2016-04-27 09:43:04 something is messing up the fortify headers 2016-04-27 09:43:09 yeah 2016-04-27 09:43:10 yes 2016-04-27 09:43:12 some of the checks are invalid i think 2016-04-27 09:43:23 it fails to find strnlen for example 2016-04-27 09:43:28 i've been trying yesterday to fix it but it's quite a nightmare 2016-04-27 09:46:08 seems like it is gnulib that fails 2016-04-27 09:52:03 hey :) 2016-04-27 15:22:01 i think i finally nailed the samba subpackage splitting 2016-04-27 15:22:10 spent 2 days on it 2016-04-27 15:22:53 there are like 241 libraries 2016-04-27 15:22:56 shared libs 2016-04-27 15:22:58 sounds like serious PITA 2016-04-27 15:23:08 guess if it is! 2016-04-27 15:23:21 those 241 libs has inter deps 2016-04-27 15:23:30 so they depen on each other 2016-04-27 15:23:47 then i want group clients and servers 2016-04-27 15:23:52 so if you apk add samba-lient 2016-04-27 15:24:06 you will not pull in stuff needed by server 2016-04-27 15:24:25 and if you only run server, you wont pull in libs only needed by clients 2016-04-27 15:24:30 but it does not stop there 2016-04-27 15:24:53 extreme modular programming demands extreme modular distribution :) 2016-04-27 15:25:03 then we have a set of tools that might be needed by both server and client 2016-04-27 15:25:10 so i put those in samba-common-tools 2016-04-27 15:25:15 (so does defora) 2016-04-27 15:25:20 fedora* 2016-04-27 15:25:52 defora sounds nice, but has a note of depreciation 2016-04-27 15:26:09 now we have a set of common-libs that is needed by common-tools and clients 2016-04-27 15:26:22 and another set that is needed by common-tools and servers 2016-04-27 15:26:35 oh, and we have 2 different servers, samba-server and samba-winbind 2016-04-27 15:26:54 so you can install only winbind if you dont need the file/print sharing 2016-04-27 15:27:18 what winbind alone is useful for? 2016-04-27 15:28:08 provides a service for ntlm auth 2016-04-27 15:28:42 i think you can for example, run squid instance 2016-04-27 15:28:47 with auth 2016-04-27 15:28:53 I see 2016-04-27 15:28:58 and have the auth backend from windows servers 2016-04-27 15:29:07 you need to "join" the windows domain 2016-04-27 15:29:15 i htink winbind can do this for you 2016-04-27 15:30:07 as a consequence i have: samba-common-server-libs and samba-common-client-libs 2016-04-27 15:30:47 samba-common-client-libs is the libs needed by both common-tools and the clients 2016-04-27 15:30:52 why not samba-server-common-libs and ...-client-...? 2016-04-27 15:31:23 ok, to have common samba-common 2016-04-27 15:31:29 yes 2016-04-27 15:31:37 same with client 2016-04-27 15:31:50 that way, when installing client+ common tools 2016-04-27 15:31:59 you dont get the libs needed by server 2016-04-27 15:32:08 saves you 4Mb or so 2016-04-27 15:32:13 sounds nice 2016-04-27 15:32:20 hopefully upgrade will be pain free 2016-04-27 15:32:29 i havent bothered do opposite 2016-04-27 15:32:48 samba-server will pull in common-client-libs in addition 2016-04-27 15:33:22 oh and i also have a separate libwbclient and libsmbclient and a samba-libs for the core libraries 2016-04-27 15:33:23 so 2016-04-27 15:33:36 guess how fun it was to untwist the deps there 2016-04-27 15:33:57 and there are apparently a handful libs that is not used by anything :) 2016-04-27 15:33:57 I can imagine 2016-04-27 15:34:02 :D 2016-04-27 15:34:03 oh and testing tools 2016-04-27 15:34:17 some libs are only used by testing tools 2016-04-27 15:34:57 i remved the samba dc support 2016-04-27 15:35:01 i dont think it works anyway 2016-04-27 15:35:16 "some 3rd party may still use these unused libs, let's don't stop shipping them, better be on the safe side..." 2016-04-27 15:35:31 right 2016-04-27 15:35:51 the leftover libs are stored in the package "samba" 2016-04-27 15:36:08 you should store them in samba-leftover 2016-04-27 15:36:12 and depend on them 2016-04-27 15:36:28 i could have them in samba-libs (core libs) 2016-04-27 15:36:38 or 2016-04-27 15:36:39 actually 2016-04-27 15:36:46 i dont want to touch it anymore 2016-04-27 15:36:53 i just want to ship it 2016-04-27 15:37:21 samba-leftover would be nice tongue in cheek 2016-04-27 15:37:35 hm 2016-04-27 15:37:45 they seem to only be 212kb 2016-04-27 15:37:49 that aint bad 2016-04-27 15:38:02 that's how all bloat starts 2016-04-27 15:38:13 "it's only N KB!" 2016-04-27 15:38:18 yup 2016-04-27 15:38:36 the 'samba' package is the leftover package 2016-04-27 15:38:53 i'm ok with that 2016-04-27 15:39:00 but also convenient one with deps to -server and -client, right? 2016-04-27 15:39:14 ? 2016-04-27 15:39:37 samba is w/ deps? 2016-04-27 15:39:42 w/o 2016-04-27 15:39:58 i think samba will pull in server 2016-04-27 15:40:06 yup 2016-04-27 15:40:11 it pulls in server only 2016-04-27 15:40:40 we should probably pull in -server, -client and -common-tools 2016-04-27 15:40:48 so if you apk add samba 2016-04-27 15:40:51 you get it all 2016-04-27 15:42:06 that's why I'm saying that leftover (w/o rdeps) stuff should have separate package, so if samba pulls client and server you can do: apk add samba; apk del samba-leftover :) 2016-04-27 15:42:33 samba-compat 2016-04-27 15:42:55 it's nice having documented what libs are unused by other samba stuff 2016-04-27 15:43:17 i supspect some of the libs are only used by DC or something else that we dont support 2016-04-27 15:43:20 or build 2016-04-27 15:43:33 oh, that may be the case too, true 2016-04-27 15:43:43 leftovers: http://tpaste.us/2bD1 2016-04-27 15:44:19 i think i just leave them there 2016-04-27 15:44:26 we have more than enough subpackages already 2016-04-27 15:44:51 i am considering adding another subpackage though 2016-04-27 15:44:58 samba-common-server-client-libs 2016-04-27 15:45:43 which is needed by all except samba-winbind-clients 2016-04-27 15:45:58 (or they could go to samba-libs) 2016-04-27 15:46:31 that would save ~3-4MB for server install i think 2016-04-27 15:46:40 i'm not sure its worth it 2016-04-27 15:48:32 nah 2016-04-27 15:48:43 i think i just rename the common-client-libs to common-libs 2016-04-27 16:07:27 it is worth it IMO, as great customisability and minimalism is what define alpine linux 2016-04-27 16:07:48 s/define/&s/ 2016-04-27 16:17:48 oh, mails from bugs.a.o changed look a bit 2016-04-27 16:18:03 was redmine upgraded recently? 2016-04-27 16:21:06 yes i think so 2016-04-27 16:21:19 ncopa: now that there is a new release on the horizon, any chance the alpine-iso stuff could be looked at? 2016-04-27 16:21:58 it still doesn't seem to allow images with the ovl to work as installers 2016-04-27 16:24:55 ncopa: #5479 - can I change target version to 3.4 and comment that it should be backported later to 3.3.4? not sure whether anyone is willing to dive into that, and I haven't played with alpine-iso at all yet. 2016-04-27 16:28:15 przemoc: sounds like a good idea 2016-04-27 16:28:41 Adran: did you file a bug with the details? i dont remember the exact issue 2016-04-27 16:28:55 ncopa: no, I can go file a bug 2016-04-27 16:29:04 thanks 2016-04-27 16:31:03 ncopa: https://bugs.alpinelinux.org/issues/5503 2016-04-27 16:33:49 Adran: can you be more specific? 2016-04-27 16:34:05 is missing sfdisk the only error? 2016-04-27 16:34:18 Let me get you some screenshots 2016-04-27 16:34:18 what happens if you add the sfdisk package to your iso? 2016-04-27 16:34:24 already is added 2016-04-27 16:34:59 should failing iso boot be categorized as "Boot sequence"? 2016-04-27 16:35:16 actually, we should make new, better categories 2016-04-27 16:35:16 it doesn't fail to boot, technically it fails at alpine setup scripts 2016-04-27 16:35:34 (I was talking about other issue than yours, Adran) 2016-04-27 16:35:38 oh sorry 2016-04-27 16:35:49 przemoc: didn't see the conversation switch, my bad. :D 2016-04-27 16:35:54 no problem 2016-04-27 16:36:07 IRC is double edged sword :> 2016-04-27 16:36:12 :) 2016-04-27 16:37:19 ncopa: attached screenshots, those are of the errors before the sfdisk error 2016-04-27 16:38:12 ok 2016-04-27 16:38:24 those errors has actually nothing to do with the installer 2016-04-27 16:38:52 i think you have linux-grsec in your /etc/apk/world in your ovl 2016-04-27 16:39:02 remove that and those errors should go away 2016-04-27 16:40:04 we were talking about better categories for some time, IIRC, but I'm not sure if there was presented any new coherent and complete list of such categories 2016-04-27 16:41:20 probably better discuss that on alpine-devel 2016-04-27 16:41:52 i think it would make sense to have one issue tracker for every git repo 2016-04-27 16:41:58 i think we currently dont 2016-04-27 16:43:39 gcc-6.1.0 is out 2016-04-27 16:43:52 i don't think we even try get that in for v3.4 2016-04-27 16:44:02 or we will get delayed with a couple of weeks 2016-04-27 16:45:20 yeah, changing compiler shortly before release is not a good idea 2016-04-27 16:46:16 but in edge it could be done shortly after release, so there would be lot of time to fix stuff and make switching to higher 6.x later easier 2016-04-27 16:46:33 yes 2016-04-27 16:46:34 (just for the record, Paul Roland in the end workarounded problem of beforementioned issue by installing 3.2 and upgrading it to 3.3, so he wasn't muchy pushy about it, but it's nice to install straight from latest iso.) 2016-04-27 16:47:00 what was that issue about? 2016-04-27 16:47:27 Mounting boot media failed. 2016-04-27 16:47:39 #5479? 2016-04-27 16:47:56 I initially thought it's i8042, but I was mistaken, as this issue has been solved long time ago 2016-04-27 16:48:23 yup, that one 2016-04-27 16:51:41 i'll continue tm. have a nice evening 2016-04-27 16:52:21 thanks and you too, ncopa 2016-04-27 17:41:39 ncopa: nice, interesting fix 2016-04-27 18:32:35 yay, a new firefox version in apk :) 2016-04-27 18:32:42 i am always happy to install it :)( 2016-04-27 18:34:31 it took me a second to properly parse :)(, but then it should be rather -)( (it it's a whole stickman) 2016-04-27 18:35:44 oh shit ... firefox looks different ... did they switch to gtk3 in this release?? 2016-04-27 18:35:47 it looks broken now 2016-04-27 18:35:51 damn you gtk3 ... 2016-04-27 18:38:01 maybe it was esr before and it's not now? 2016-04-27 18:48:28 i renamed the firefox in stable to firefox-esr 2016-04-27 18:50:08 ncopa: its in the changelog 2016-04-27 18:50:19 GTK3 integration (GNU/Linux only) 2016-04-27 18:50:29 those idiots really switched to gtk3 ... 2016-04-27 18:50:39 no wonder it eats 100mb mor ram 2016-04-27 18:50:39 I have no idea why 2016-04-27 18:52:49 firefox is awfull these days 2016-04-27 18:53:02 works far better on windows than Linux 2016-04-27 18:53:05 such a shame 2016-04-27 18:53:06 browsers are awful these days 2016-04-27 18:53:15 quite true 2016-04-27 18:53:56 When just a single browser is larger than my entire OS install and takes 30x the amount of memory on a server running a single service you know something is wrong 2016-04-27 18:55:09 websites are also roughly 100x-1000x more bloated than they were 15 years ago 2016-04-27 18:56:19 So much over engineering 2016-04-27 18:59:30 and to be honest, some web apps can be really complex, but they're build on layers^Wpiles of indirection, so costs have to be spend somewhere... 2016-04-27 19:03:56 s/complex/pointlessly complex/ ftfy 2016-04-27 19:04:39 and WebAssembly won't improve anything, it will only lead making life harder for js-copy-pasters and co. no, wait, it will possibly even worsen things, because people will be pushing more and more junk into web. 2016-04-27 19:05:34 I'm just glad the js hipsters stay to the web. Easier to avoid them 2016-04-27 19:06:20 no they don't they get everywhere 2016-04-27 19:06:42 those hipsters were also ruby hipsters and recently go hipsters 2016-04-27 19:07:05 Hey now watch yourself with Go 2016-04-27 19:07:08 lol 2016-04-27 19:07:29 static binaries are very anti-hipster 2016-04-27 19:09:42 poptart: I used static binaries before they were cool 2016-04-27 19:09:49 oh wait... they still aren't cool 2016-04-27 19:10:08 hipster doesn't care about static or dynamic, it's all about trends, and programming in go is fashionable nowadays 2016-04-27 19:10:39 I actually like Go. I haven't liked any other popular language recently 2016-04-27 19:11:01 same here :) 2016-04-27 19:11:23 I think that only Rust brings useful stuff to the table, but didn't have to time to dig into it properly 2016-04-27 19:11:38 you would be wrong lol. Rust is nasty 2016-04-27 19:12:29 Good concept with an extremely "mozilla" execution 2016-04-27 19:12:38 does go actually produce static binaries, or just the go-side dependencies 2016-04-27 19:12:46 Actually static 2016-04-27 19:12:54 Unless you use cgo 2016-04-27 19:23:01 my point is that go is quite C-ish to the bone, so why bother with it and don't go with C directly? if you like channels for instance, then there is libmill for you (haven't used it, though, so maybe it's too incomplete now for serious use). what are those other great benefits of using go? serious question. 2016-04-27 19:23:25 memory safety 2016-04-27 19:24:13 It prevents idiots from shooting themselves in the face 2016-04-27 19:24:23 but it is also super easy to write 2016-04-27 19:24:54 i.e. Google Java 2007 2016-04-27 19:25:07 Except those aren't static binaries 2016-04-27 19:25:36 static binaries ~ write once, run anywhere 2016-04-27 19:26:05 But it does mean I can compile it on linux and run it on same architecture linux 2016-04-27 19:26:16 I never said it did either 2016-04-27 19:27:48 I mean, neither did java really 2016-04-27 19:28:00 "write once, debug everywhere" 2016-04-27 20:05:07 go static binaries aren't all that interesting, easy to do in c too tbh 2016-04-27 20:05:29 i like static binaries too but they come with their own warts 2016-04-27 20:05:52 note: i hate shared libraries with a passion, but they do have uses 2016-04-27 20:06:06 mostly rootkits ;) 2016-04-27 20:06:51 ooh that too 2016-04-27 20:07:13 i was thinking more you can LD_PRELOAD things in between and change behavior on things you can't control with libraries easier 2016-04-27 20:07:19 but thats pretty much the same thing 2016-04-27 20:07:45 I know exactly what you meant, I've only seen LD_PRELOAD be abused in real life though 2016-04-27 20:08:33 i used it to swizzle around some hokey old tcl/tk thing to fix it working on some old rhel 4/5 box 2016-04-27 20:08:50 I'm so sorry lol 2016-04-27 20:09:14 my hatred of shared libraries has reasons >.< 2016-04-27 20:10:00 My real problem is I don't trust my own C skills lol 2016-04-27 20:10:43 i work in kernel c code for a living, I don't trust mine either 2016-04-27 20:11:20 but i'm a huge fan of http://www.idris-lang.org 2016-04-27 20:11:48 i'm pretty sure the future of computers is being less throw spaghetti at walls and more mathematical, but i could be wrong 2016-04-27 20:12:28 latest release of idris makes it easy to call from c, sooo, i'm tempted to try using it to help test some of the fs code I work on 2016-04-27 20:13:32 I don't want to rain on your parade, but the future of computers has been less throw spaghetti at walls and more mathematical for a good 20 years now 2016-04-27 20:13:56 I wouldn't want to write a webapp in C honestly with my skillfu 2016-04-27 20:14:02 and pasta vendors are still richer than academics 2016-04-27 20:14:09 ^^^^^ 2016-04-27 20:14:14 can do that correctly in 10minutes in Go 2016-04-27 20:14:35 was fun writing a state machine that describes simple bind/listen/accept and makes it impossible to compile things that do things wrong 2016-04-27 20:15:16 from my pov, rust doesn't do all that much interesting compared to say ocaml/ml/idris/haskell et al 2016-04-27 20:19:16 Anyway if I wanted to make an aport for the tinc 1.1 branch is there any formal way that you guys are handling packages with the same name? I can't seem to find any good examples 2016-04-27 20:36:27 leo-unglaub: are you here? 2016-04-27 20:39:19 yes 2016-04-27 20:39:56 whats up? 2016-04-27 20:40:48 leo-unglaub: Today I’ve speak with one friend of mine who works at RedHat on the Fedora project and the talk comes to Rust. He told me that they don’t have package for Rust, because of issue with bundled LLVM. It’s against their rules, the same is for Gentoo and probably other distros. 2016-04-27 20:41:36 yes, i know 2016-04-27 20:41:39 it is stupid 2016-04-27 20:41:45 So it seems that Rust will be eventually forced to provide some sensible way how to use system-provided LLVM instead of they stupidly bundled… 2016-04-27 20:41:49 the bundled llvm sucks hard!! 2016-04-27 20:41:49 they’re 2016-04-27 20:42:16 the currrent problem i see in rust is that they try to please to many people at once 2016-04-27 20:42:23 the language is getting way to complex at the moment 2016-04-27 20:42:28 because they put in everything 2016-04-27 20:42:40 the nice thing about c is that you can learn it in a day or two 2016-04-27 20:42:47 there is a couple of keywords and thats it 2016-04-27 20:42:47 LOLO 2016-04-27 20:42:58 Maybe syntax, but that’s just very little part… 2016-04-27 20:43:01 :O 2016-04-27 20:43:01 but with rust ... they have bloaded it up latley and i dont know why 2016-04-27 20:43:47 Well, it’s quite complex, but I think that it’s manily due to their inovative memory-management. 2016-04-27 20:44:05 no, has nothing todo with memory management ... 2016-04-27 20:44:15 the language itself is getting to complex 2016-04-27 20:44:22 ^^ 2016-04-27 20:44:25 Could you provide some examples of unnecessary complexity? 2016-04-27 20:45:09 stuff like the new try 2016-04-27 20:45:18 what’s wrong with it? 2016-04-27 20:45:32 it sucks ;( 2016-04-27 20:45:57 because thats not a propper way to deal with errors because it just inlines a lot of shit 2016-04-27 20:46:05 and then it becomes dublicate 2016-04-27 20:46:17 because you have the same try in multiple places and people just inline it 2016-04-27 20:47:08 the current state of error handling is not ideal; it’s great that they uses strong concepts from FP, but from my very little experience, it needs to be improved 2016-04-27 20:47:30 it needs to be less strict in my mind 2016-04-27 20:47:57 As I see it now, the main problem is that the error types are not… how to say it… composable? 2016-04-27 20:49:00 When I need to return various error types (like io::Error, Utf8Error, …), then it becomes really pain in the ass. 2016-04-27 20:49:25 And the solutions I found are more workarounds then proper solutions IMO. 2016-04-27 20:50:19 Anyway, the new try is not accepted yet, is it? 2016-04-27 20:50:28 *isn’t it 2016-04-27 20:50:46 i am not sure because it dont really use it 2016-04-27 20:50:58 i just have two places where i really check for those kind of errors 2016-04-27 20:51:06 when working with IO and while opening sockets 2016-04-27 20:51:12 everything else does not really need it 2016-04-27 20:51:26 really? 2016-04-27 20:52:06 Do you have some code online so I can see it? I’d like to learn how to deal with it better, I have just few lines of code now and quite a lot error handling. 2016-04-27 20:53:15 i am oldschool, i am not one of those "test driven development" kind of programmers 2016-04-27 20:53:48 what does it mean? that you write spaghetty code and tests are too fancy for you…? :P 2016-04-27 20:53:49 i test things that are dangerous because other people can influence it but i dont test a method that does a modulo on an integer ;) 2016-04-27 20:54:13 sure, but we’re talking about error handling, not tests now… 2016-04-27 20:54:39 yeah, but its the sme principle. i dont really handle an error from a method that is calculating x -y 2016-04-27 20:54:54 i just handle errors from a method that does something that really can go wrong 2016-04-27 20:55:19 so how do you deal with Result and Option? 2016-04-27 20:55:31 there are almost everywhere in the Rust API, you can’t just ignore them 2016-04-27 20:55:50 sure you can ;) 2016-04-27 20:55:59 and please don’t tell me that you write unwrap() everywhere, this is definitely wrong 2016-04-27 20:56:27 ACTION is not saying anything anymore *g* 2016-04-27 20:56:28 hahaha 2016-04-27 20:56:57 sorry, I didn’t want to be mean 2016-04-27 20:57:25 but I can’t imagine how else can you “ignore” error handling 2016-04-27 20:58:50 i can show you if you want 2016-04-27 20:58:55 yes please :) 2016-04-27 20:58:58 but not tonight, i am very busy at the moment 2016-04-27 20:59:19 okay, no problem :) 2016-04-27 20:59:58 i have a big exam on saturday and i still did not start learing 2016-04-27 21:00:07 so i am reading the scriptum right now *g* 2016-04-27 21:00:13 saturday? 2016-04-27 21:00:38 you have distance (part-time) study? 2016-04-27 21:01:05 nah, a new extension for my paramedic 2016-04-27 21:01:26 aha 2016-04-28 01:53:58 https://twitter.com/marcan42/status/724745886794833920 hey anyone in the mood to attempt panicking a kernel 2016-04-28 01:54:04 (navigating internet ego contests optional) 2016-04-28 08:40:11 ammunta: I'm not sure i understand twitter. what's going on? 2016-04-28 08:44:43 grsec updated their overflow plugin, found a lot of new territory to fix, also in the tty code, but the latest patch did not fix everything, and now it crashes. 2016-04-28 08:44:56 not grsecs fault really 2016-04-28 09:03:54 i've got an rpi with a disk install. i just need to get modloop mounted. does init mount modloop? 2016-04-28 09:04:23 i guess it has to get mounted pretty damn early 2016-04-28 09:22:35 it's mounted by openrc 2016-04-28 09:26:05 hey :) 2016-04-28 09:37:53 i just unsquashed modloop to finish the disk install 2016-04-28 09:37:57 leo-unglaub: hey 2016-04-28 09:45:11 ScrumpyJack: all the modules are in linux-rpi (or -rpi2) anywya 2016-04-28 10:00:43 yth: the kernel executable? 2016-04-28 10:08:22 buildlogs for 3.4 armfh show a lot of build failures. what gives? 2016-04-28 10:09:07 work needs to be done to fix 3.4 arm build 2016-04-28 10:09:23 i'm setting up v3.4 builders for x86_64 and x86 too now 2016-04-28 10:13:29 ping --help 2016-04-28 10:13:38 ah, damn ... wrong window 2016-04-28 10:41:10 ncopa, gcc 6.1 is out with upstream support for musl and --enable-default-pie 2016-04-28 10:41:24 need to look at it once we get stable branched 2016-04-28 10:41:49 oh, that sounds like really nice features 2016-04-28 10:42:09 that's actually much ado about nothing 2016-04-28 10:42:52 you'll just need less hacks to make ***-alpine-linux-musl work 2016-04-28 10:43:11 and that's about the extent of the "support" 2016-04-28 11:13:12 still less hacks. 2016-04-28 11:20:01 oh, sure, it's always nice. But it's nothing to get overexcited about. 2016-04-28 11:45:04 I think nobody's overexcited ;) 2016-04-28 12:17:27 it looks like armhf 3.4 is being built first. what's the thinking behind that? 2016-04-28 12:18:32 it's the slowest one, so better start early, right? 2016-04-28 12:24:07 ncopa, what can I do for movine some my packages from testing to main or community before releasing v3.4? 2016-04-28 12:24:14 s/movine/moving/ 2016-04-28 12:25:22 i am very exited about gcc 6.1 2016-04-28 12:25:40 what’s great about it? 2016-04-28 12:25:48 leo-unglaub: well, you're excited about Rust, so it doesn't mean much 2016-04-28 12:25:54 ACTION ducks 2016-04-28 12:25:58 hahaha 2016-04-28 12:26:14 well, you are right, i am also exited about s6 ... so it really doesnt mean anything *g* 2016-04-28 12:26:22 ACTION ducks 2016-04-28 12:26:24 definitely meaningless 2016-04-28 12:26:34 leo-unglaub: +1 XD 2016-04-28 13:02:25 ScrumpyJack: i'm looking at gexiv2 2016-04-28 13:02:30 i have no idea why it fails 2016-04-28 13:02:38 could be new gobject-introspection 2016-04-28 13:03:24 http://tpaste.us/Apox 2016-04-28 13:08:19 py-twitter needs py-oauthlib, but one of the py-oauthlib files has got the wrong perms for py-twitter 2016-04-28 13:08:32 oauthlib-0.6.3-py2.7.egg-info/PKG-INFO' 2016-04-28 13:08:53 i'm wondering if it's a fakeroot issue 2016-04-28 13:09:05 http://build.alpinelinux.org/buildlogs/build-3-4-armhf/main/py-twitter/py-twitter-2.2-r0.log 2016-04-28 13:10:01 if i abuild deps py-taothlib, chmod oauthlib-0.6.3-py2.7.egg-info/PKG-INFO and abuild -r py-twitter, it works 2016-04-28 13:10:20 s/py-taothlib/py-oauthlib 2016-04-28 13:10:36 check what fedora and others do 2016-04-28 13:12:17 ncopa, what can I do for moving some my packages from testing to main or community before releasing v3.4? 2016-04-28 13:12:54 make a pr, with all the packages you need 2016-04-28 13:13:29 jirutka: are they all tested? :) 2016-04-28 13:13:35 or git format-patch --stdout HEAD~$N | tpaste 2016-04-28 13:13:44 where $N is the number of patches you want move 2016-04-28 13:33:45 it doesn't seem to be a problem for gentoo or arch 2016-04-28 13:40:33 ScrumpyJack: seems that oauthlib 1.1 is out 2016-04-28 13:40:55 maybe we can upgrade to 1.1.0 2016-04-28 13:41:57 here is the error 2016-04-28 13:41:59 http://sprunge.us/SWcc 2016-04-28 13:42:24 the files in oauthlib-0.6.3-py2.7.egg-info have the wrong perms for fakeroot 2016-04-28 13:43:16 i upgrade it to 1.1.0 2016-04-28 13:43:20 seems to fix it 2016-04-28 13:43:26 lol - nice 2016-04-28 13:46:29 can you rebuild py-requests-oauthlib and py-flask-oauthlib just in case? 2016-04-28 14:03:18 updated py-django-reg. 2016-04-28 14:03:24 ncopa: what's next? 2016-04-28 14:04:37 where/how did you fix py-django-reg? 2016-04-28 14:05:38 patchwork 2016-04-28 14:05:53 i mean it's in the patchwork queue 2016-04-28 14:07:17 ah there it is 2016-04-28 14:10:44 lots of ruby build failures in armhf/main 2016-04-28 14:14:07 i think we prefer the source="$pkgname-$pkgver.tar.gz::https://..." over "save-as-https" 2016-04-28 14:15:58 ok cool, I'll change it 2016-04-28 14:16:02 no 2016-04-28 14:16:04 i'm on it 2016-04-28 14:16:35 hm 2016-04-28 14:16:39 checksum will differ 2016-04-28 14:18:24 there are a few more with saveas, not many mind you http://sprunge.us/OFTE 2016-04-28 14:18:32 I'll change the wiki 2016-04-28 14:18:41 thanks!!! 2016-04-28 14:22:25 i should write to alpine-devel 2016-04-28 14:22:31 re ruby-* 2016-04-28 14:23:11 why is so much stuff failing in armhf/community? 2016-04-28 14:30:01 i thought it was due to missing deps in main 2016-04-28 14:30:32 speaking of comunity 2016-04-28 14:30:49 i think there is a new mate version out 2016-04-28 14:31:21 yup 2016-04-28 14:31:23 http://pub.mate-desktop.org/releases/1.14/ 2016-04-28 14:34:20 the build order: http://tpaste.us/AWZ0 2016-04-28 14:40:13 I also have to finally go through testing pkgs that were added upon my request and test them to see if they deserve move toward community or main. I'm scarce on time lately, though. have to do taxes today and also move all the stuff from notebook, so I could carry it in to some warranty service (as it expires in days). 2016-04-28 14:41:16 ncopa, what is your rough deadline for aports patches? 2016-04-28 14:43:01 i'd say ~May 1 2016-04-28 14:43:07 depending on what the patch does 2016-04-28 14:43:20 that sounds really bad 2016-04-28 14:43:27 i will do my best to get the patches in 2016-04-28 14:43:36 but i will prioritize stuff that fixes things 2016-04-28 14:43:43 rather than introduces new stuff 2016-04-28 14:44:03 that’s good, so I have all weekend 2016-04-28 14:44:40 things that are not in main after the weekend will likely not make it 2016-04-28 14:44:53 i want fix the dovecot stuff 2016-04-28 14:46:12 I'm out of town for most of Saturday, but I think I'll manage to touch some AL stuff on Sunday at least. 2016-04-28 14:48:58 generally speaking, i want fix the things we have now, instead of adding more stuff 2016-04-28 14:49:26 i will try get the x86_64 and x86 builders up before weekend 2016-04-28 14:49:41 tinc has a 1.1 branch that has been in "beta" for like 4 years if I wanted to submit a package to testing for that version is there any standard for naming? 2016-04-28 14:55:32 poptart: tinc-1.1_betaN 2016-04-28 15:00:34 ncopa: what's left to fix in main? (excluding ruby*) 2016-04-28 15:01:14 ncopa: please keep few “core” packages like ruby-bundler 2016-04-28 15:01:50 maybe this is the only one that should be really keeped, but I’ll check it later 2016-04-28 15:02:51 are there any ruby-* packages that contains some fixes for native extensions to compile on musl? if so, then I think that these are candidates to be keeped too 2016-04-28 15:04:01 um 2016-04-28 15:04:25 yes I expect that there might be needed a few ruby-* packages 2016-04-28 15:04:46 ncopa: could you create 3.5.0 version in bugs.a.o before weekend, so we could move or create stuff that won't make into 3.4? 2016-04-28 15:04:58 przemoc: good idea 2016-04-28 15:05:05 i'll do so now 2016-04-28 15:05:51 done 2016-04-28 15:05:59 ok, thx 2016-04-28 15:10:11 hum 2016-04-28 15:10:18 i would have liked to have mate-1.14 2016-04-28 15:10:31 seems like we should build it with gtk3 2016-04-28 15:11:43 from talks on #alpine-linux I already see that adding resolvconf would be very good thing, but adding pkg alone is half of the problem as there are scripts that must be made compatible with it. I doubt I manage to go through this thoroughl this weekend (or someone else would volunteer) as it's not my top priority and it's rather too late for such changes, but I'll add proper ticket for it for 3.5 la 2016-04-28 15:11:50 ter this week. 2016-04-28 15:12:06 i thought the whole point of mate was to stick to gnome2 and gtk2, no? 2016-04-28 15:12:19 gnome2 feel 2016-04-28 15:12:21 not libs 2016-04-28 15:13:18 apparently mate-system-monitor no longer builds with gtk2 2016-04-28 15:13:36 I stand corrected "Improved gtk3 support" 2016-04-28 15:14:59 i wonder if its usable with gtk3 support yet 2016-04-28 15:17:26 where's alacerda when we need him :) 2016-04-28 15:17:33 i think he left us 2016-04-28 15:17:46 :( 2016-04-28 15:17:59 ncopa: yes purge ruby-* I vote for 2016-04-28 15:26:50 thanks ncopa! 2016-04-28 15:27:10 does gem work for everyone? 2016-04-28 15:27:58 dunno 2016-04-28 15:28:14 the nice thing with it is that we can simply push issues to upstream 2016-04-28 15:28:20 gem? not our business 2016-04-28 15:34:21 ruby is such a mess anyway :D 2016-04-28 15:35:05 exactly 2016-04-28 15:35:07 :) 2016-04-28 15:45:07 I’m trying to make package ruby21 for packages that depends on old ruby version. The only problem I have is -dev subpackage, how to modify ruby-21.pc (http://haste.fit.cvut.cz/avufahu) to move just libruby-static.a (LIBRUBY_A) and libruby.so.2.1.9 (LIBRUBY_SO) do directory /usr/lib/ruby21 (defualt is /usr/lib)? 2016-04-28 15:46:34 I can’t just change --libdir, because it’s a base directory also for ruby stuff like gems, vendor_ruby etc. which already contain a version number (ABI version). 2016-04-28 15:47:04 jirutka: that is an option yes 2016-04-28 15:47:09 we do something simliar with lua 2016-04-28 15:47:18 i think we als call it ruby2.1 2016-04-28 15:47:35 So I want to keep /usr/lib/ruby/2.1.0, /usr/lib/ruby/gems/2.1.0, /usr/lib/ruby/vendor_ruby/2.1.0, but I need to move libruby-static.a and libruby.so to avoid conflicts 2016-04-28 15:47:44 yeah, sorry, ruby2.1 2016-04-28 15:48:14 I’m looking to lua5.1 for inspiration, but I don’t understand package config and other C stuff 2016-04-28 15:50:26 If I don’t miss something, then it should be ok-ish just to patch generated ruby-2.1.pc and move .a and .so files, so I can avoid messing with configure and autohells… 2016-04-28 15:50:47 but I don’t understand .pc, so I don’t know what I can change without breaking it 2016-04-28 15:51:38 you can find the ruby-2.1.pc file on http://haste.fit.cvut.cz/avufahu 2016-04-28 16:08:30 please, could someone help me with this? just push me in the right direction 2016-04-28 16:56:41 ncopa, I’ve just got email: “just reaching out to ask if there's any plan to have `apache-ant` in main or community at Alpine? We really appreciate your work there and I'd like to get rid of building Ant in our image. Can we support you somehow in this matter?” from Nils Magnus at IO.GmbH :) 2016-04-28 17:52:40 I’ve eventually figured out that ruby2.1 thing… Gentoo is always good inspiration. :) 2016-04-28 17:53:35 I wonder, should I send ruby2.1 to the testing branch, or keep it just in my “overlay”? 2016-04-28 21:09:12 jirutka send it to testing 2016-04-29 00:08:33 https://grsecurity.net/~spender/grsecurity-3.1-4.5.2-201604281949.patch contains the riproprap from paxteam mitigation against ROP vectors 2016-04-29 00:25:58 Here is the announcement too: https://grsecurity.net/rap_announce.php 2016-04-29 00:26:14 Have you ever been so angry at a shitty twitter conversation that you make a kernel patch to prevent ROP exploits? 2016-04-29 00:26:19 > just spender things 2016-04-29 00:35:15 rap is from paxteam and has been announced quite some time ago, this is the release 2016-04-29 00:49:35 I realize these, it just happened to coincide with the twitter bullshit. It was a joke 2016-04-29 07:19:39 morning 2016-04-29 07:21:17 morning 2016-04-29 07:30:43 can you show us a list of broken packages this morning? 2016-04-29 07:31:23 ruby-* ? 2016-04-29 07:31:24 :) 2016-04-29 07:31:27 lol 2016-04-29 07:31:49 py-fuse/src 2016-04-29 07:32:11 acf-provisioning-snom/src py-cryptography/src 2016-04-29 07:32:13 the buildozer logs are still showing lots of failures in community. 2016-04-29 07:32:26 ok 2016-04-29 07:32:31 let me run a manual build 2016-04-29 07:35:31 hmm, wierd. take evince. it just failed. how is gtk+3.0-dev missing? 2016-04-29 07:35:48 I see it for armhf 2016-04-29 07:36:05 I don't use 9p-fs, does it now work ok with current kernel ? 2016-04-29 07:36:42 read about having issue with 4.2, https://github.com/vincentbernat/eudyptula-boot/blob/master/eudyptula-boot 2016-04-29 07:37:33 but saw some fixes in kernel, b5891cfab08fe31 2016-04-29 07:38:14 ncopa: there's something wrong. 2016-04-29 07:39:37 just checking randomly, xpdf, youtube-dl, lensfun, have missing build deps, but i see the deps in arm repo 2016-04-29 07:39:52 ScrumpyJack: you are right 2016-04-29 07:42:32 i figured out whats wrong witih the ruby stuff 2016-04-29 07:42:40 >>> ruby-globalid: Analyzing dependencies... 2016-04-29 07:42:40 ERROR: unsatisfiable constraints: 2016-04-29 07:42:40 ruby-activesupport4.2 (missing): 2016-04-29 07:42:40 required by: world[ruby-activesupport4.2] 2016-04-29 07:43:22 there is not ruby-activesupport4.2 2016-04-29 07:43:54 its in community 2016-04-29 07:44:06 ah 2016-04-29 07:45:39 why can't buildozer see it? don't tell me it doesnt have community in /etc/apk/repositories 2016-04-29 07:46:41 btw, unmaintained/mongrel2 is active again 2016-04-29 07:51:04 anyone using community chromium ? 2016-04-29 07:51:08 ncopa: does 32bit AL needs fixing for y2038 issue ? 2016-04-29 07:51:10 https://github.com/schwern/y2038/wiki 2016-04-29 07:51:10 doesn't work here ... 2016-04-29 07:51:42 or just assume everbody would move to 64bit before 2038 ;) 2016-04-29 07:54:49 everybody would move to bsd ;) 2016-04-29 07:54:59 :O 2016-04-29 07:56:18 vkris: i dont know 2016-04-29 07:57:50 thankfully AL doesn't support releases for over 22 years yet, so it doesn't require any hotfix before 3.4. 2016-04-29 07:59:10 my current problem is to get 3.4 release out, which hopefully will happen long before year 2038... 2016-04-29 07:59:42 optimist 2016-04-29 08:01:14 coredumb: chromium works for me on x86_64 2016-04-29 08:02:33 ncopa: you stopped dozer, found something? 2016-04-29 08:02:44 i stopped dozer 2016-04-29 08:03:17 i think i will start it as soon as i have removed the ruby stuff 2016-04-29 08:03:23 and i think i will make it halt on first error 2016-04-29 08:03:36 ok 2016-04-29 08:06:38 bump main/gzip to 1.8 pls 2016-04-29 08:08:10 clandmeter: are you ok that we move redmine to unmaintained? 2016-04-29 08:08:25 and install/maintain it with bundler/gem? 2016-04-29 08:08:43 clandmeter: or do you want keep the redmine.apk? 2016-04-29 08:10:52 vkris: can it wait? gnu.org is down for me 2016-04-29 08:11:09 ok 2016-04-29 08:13:26 testing/opennebula goes out too 2016-04-29 08:13:32 oh 2016-04-29 08:13:42 one thing i'd like help with 2016-04-29 08:14:10 we should purge all unmaintained/* that is older that v3.3.0 2016-04-29 08:17:39 heh - emacs is in unmaintained. 2016-04-29 08:17:48 it does not build 2016-04-29 08:18:04 and noone wants to fix it :) 2016-04-29 08:18:18 i think dalias looked at it 2016-04-29 08:18:52 but it does nasty stuff 2016-04-29 08:19:05 gnu-isms 2016-04-29 08:19:30 emacs is a nice OS. it only needs a decent editor 2016-04-29 08:19:44 it would be a nice OS on a LISP machine 2016-04-29 08:20:08 but on post-1990 computers? it's obsolete. :P 2016-04-29 08:20:16 lol 2016-04-29 08:20:25 right, so, unmaintained ... what do you want to happen? 2016-04-29 08:20:58 git rm -r unmaintained/ 2016-04-29 08:21:09 ok 2016-04-29 08:21:39 my current plan: get build-3-4-x86/x86_64 started and start bootstrapping the toolchain 2016-04-29 08:22:22 i am waiting for an ACK from clandmeter that we can remove redmine/ruby-* 2016-04-29 08:22:41 i don't want purge it without his ACK, since he has set up and maintains bugs.a.o 2016-04-29 08:24:50 ACTION hides 2016-04-29 08:25:39 well, that was at least "no! lets keep redmine!" 2016-04-29 08:30:20 I would say that redmine per se is ok-ish, doesn't have this "heavy" feel that comes with many other issue trackers 2016-04-29 08:31:14 we will continue use redmine, but not install it with .apk 2016-04-29 08:31:21 and i agree, redmine is ok-ish 2016-04-29 08:31:52 have not found better alternatives yet 2016-04-29 08:32:51 like the most of the best open source stuff, it's just not written yet ;) 2016-04-29 08:33:08 ncopa: delete everything in unmaintained *before* 3.3-stable? 2016-04-29 08:33:31 i.e. moved there before tagging 3.3 2016-04-29 08:33:37 do you mean *in* 3.3-stable? or in 3.2-stable (before 3.3) 2016-04-29 08:34:29 do the git rm on git master 2016-04-29 08:34:37 only stuff keep stuff added to master recently? 2016-04-29 08:34:44 find everething that existed in unmaintained before 3.3-stable was branched 2016-04-29 08:34:47 that's a lot of stuff to delete 2016-04-29 08:34:47 and purge it 2016-04-29 08:35:00 why delete things from unmaintained? 2016-04-29 08:35:13 to clean up 2016-04-29 08:35:21 its unmaintained 2016-04-29 08:35:23 the idea was that we keep things in unmaintained in 6 months 2016-04-29 08:35:27 then we purge it 2016-04-29 08:35:42 i dont want keep things there forever 2016-04-29 08:35:57 it is not deleted technically 2016-04-29 08:35:59 i disagree 2016-04-29 08:36:04 its just hidden 2016-04-29 08:36:07 ACTION hold fire 2016-04-29 08:36:09 thats the whole point 2016-04-29 08:36:21 you can get it back from git history 2016-04-29 08:36:21 well you guys just need to decide what exact semantics "unmaintained" has 2016-04-29 08:36:27 ok 2016-04-29 08:36:30 hide things that used to be interesting will hide it forever. 2016-04-29 08:36:35 nobody will ever check gitlog 2016-04-29 08:37:02 skarnet: right. 2016-04-29 08:37:07 hm 2016-04-29 08:37:14 ok 2016-04-29 08:37:28 suggestion: do what ncopa suggests, but have a web page where you list all packages that you delete 2016-04-29 08:37:37 i think i have pulled things from unmaintained before. 2016-04-29 08:37:39 so if someone wants it, they know where to look 2016-04-29 08:37:58 i think i also have ressurected from unmaintained 2016-04-29 08:38:06 ok, int that the same as having unmaintained? 2016-04-29 08:38:14 its just not a website... 2016-04-29 08:38:39 the thing is that keeping packages prepared according to older AL standards can be somewhat detrimental. it's better to not have old stuff overly exposed. if someone is willing to bring something to live, she/he can always look in aports's history, but should always package according to current standards 2016-04-29 08:39:00 how about zap from pre 3.0-stable 2016-04-29 08:39:00 I mean something like: foobar: deleted from unmaintained/foobar on 2016-13-47 on commit 1234deadbeef 2016-04-29 08:39:15 (reason: unmaintained for 6 months) 2016-04-29 08:39:41 ok i rest my case, im outnumbered! 2016-04-29 08:39:42 :) 2016-04-29 08:39:44 easy to grep, explains why it has been deleted, tells people what to do if they want to resurrect it 2016-04-29 08:39:49 i mean clandmeter, you have cinnamon in there. will you ever resurrect that? :) 2016-04-29 08:40:04 ScrumpyJack: why not? 2016-04-29 08:40:34 if somebody is really interested in it and wants to maintain it. 2016-04-29 08:40:43 whats the drawback of keeing unmaintained/* 2016-04-29 08:40:54 someonee would probably start from scratch no? 2016-04-29 08:41:11 ncopa: there's a difference between "unmaintained" and "unmaintained for more than 6 months/1 year/2 years" 2016-04-29 08:41:27 the former is waiting for a maintainer, the latter is abandoned and nobody cares 2016-04-29 08:41:30 the standard APKBUILD file has evolved a bit 2016-04-29 08:41:48 ScrumpyJack: i dont mean move it back and be happy 2016-04-29 08:41:58 some apkbuilds have more then just configure --whatever 2016-04-29 08:42:12 it can be used as a basis for someone that wants to pick it up 2016-04-29 08:42:32 that was the intention 2016-04-29 08:42:41 but i also think that we should not keep it forever there 2016-04-29 08:42:46 i just wonder what the cost is like ncopa said. 2016-04-29 08:42:49 if we keep it. 2016-04-29 08:42:52 or rename it "landfill" 2016-04-29 08:43:00 "graveyard" 2016-04-29 08:43:22 burning pyre! 2016-04-29 08:44:46 ncopa: its a non issue for me. do what you think is right. 2016-04-29 08:45:05 clandmeter: i think the motivation for deleting it would be to: 1) hide stuff that does not work or is outdated 2016-04-29 08:45:13 2) hide stuff that nobody cares about 2016-04-29 08:45:43 ok, so who would look in unmaintained? the original contributor/maintainer - to pick something up again; and someone looking to see if the package they want to build has been built before and is now unmaintained. 2016-04-29 08:45:44 3) motivate people to take action if the care about it 2016-04-29 08:46:09 so can we provide that some other way? 2016-04-29 08:46:47 try and auto-build stuff in unmaintained once in a while, and if it's been broken for N months, delete it? 2016-04-29 08:46:59 add a hook to newapkbuild that will check gitlog for */$pkgname/APKBUILD and sound an alarm ;-) 2016-04-29 08:47:18 skarnet: we normally move stuff to unmaintained because it does not build 2016-04-29 08:47:21 if you care about an absent package, unamaintained makes no difference. you'd build one anyway 2016-04-29 08:47:55 i think clandmeters point is to avoid double work 2016-04-29 08:47:56 ncopa: so stuff that still builds by chance can remain in testing even if it hasn't been looked at for ages? o.O 2016-04-29 08:48:10 someone wants a package, and we used to have it 2016-04-29 08:48:20 skarnet: i want fix that too 2016-04-29 08:48:43 currnetly we dont have any routine for cleaning up testing repo either 2016-04-29 08:48:51 but i figured that we start with unmaintained 2016-04-29 08:48:58 and then we do testing 2016-04-29 08:49:04 suggestion: separate "broken" from "graveyard" 2016-04-29 08:49:17 i'd like to do something similar with testing 2016-04-29 08:49:40 regularily remove thigns from testing 2016-04-29 08:49:40 shouldn't unmaintained be "it does not build and noone wants to fix it"? 2016-04-29 08:49:56 if something does not build, move it to "broken"; if it stays broken for N months, move it to "graveyard" 2016-04-29 08:49:57 ScrumpyJack: that is what it currently is in practice 2016-04-29 08:50:15 the thing is 2016-04-29 08:50:27 if its broken, then we should not keep it in git at all 2016-04-29 08:50:42 (in theory) 2016-04-29 08:51:04 testing -> obscure -> unmaintained -> broken -> graveyard 2016-04-29 08:51:08 ;) 2016-04-29 08:51:39 but if you only want to keep stuff that's not broken, it means you need to fix it, i.e. find a maintainer, which is precisely the problem 2016-04-29 08:52:16 which is why it is named "unmaintained" 2016-04-29 08:52:30 so you shouldn't expect "unmaintained" to not be broken 2016-04-29 08:52:50 its not are requirement 2016-04-29 08:52:55 also, a package that's been in unmaintained for ages could confuse someone. there are meta packages in there like gnome-base. there's mysql, which might be there for religious reasons. 2016-04-29 08:53:00 its just how it has worked in practice 2016-04-29 08:53:32 we build main/community/testing, somethign breaks, we want ship, nobody has time to fix the broken pkg and we move it to unmaintained 2016-04-29 08:54:49 i think i'd liek something like 2016-04-29 08:55:06 AIUI "unmaintained" has 3 different semantics: A: the package happens to work but has no maintainer and needs one asap. B: the package doesn't build or doesn't work, but people might still care. C: the package is apparently abandoned, has been broken for a long time. 2016-04-29 08:55:10 maybe cherry pick what stays in unmaintaned? 2016-04-29 08:55:17 ideally you'd want to separate A, B and C 2016-04-29 08:55:29 przemoc's suggestion isn't half bad :D 2016-04-29 08:56:09 separating A,B from C is why i want purge stuff in unmaintained 2016-04-29 08:56:30 which means C is only found in the git log 2016-04-29 08:57:26 category C is dead packages 2016-04-29 08:57:30 really dead packages 2016-04-29 08:57:37 they are so dead that they start to stink 2016-04-29 08:58:07 so i want bury those 2016-04-29 08:58:08 so, either bury them ("graveyard") or cremate them (git rm) 2016-04-29 08:59:00 bury = git rm 2016-04-29 08:59:10 calibre has been in unmaintained since 2.6 2016-04-29 08:59:16 if you want it back you can still dig it up from git history 2016-04-29 09:00:16 (sorry, 2.7) 2016-04-29 09:00:49 so, what i'd like is 2016-04-29 09:00:56 ncopa: I'm not opposed to that, but I think an obituary would be nice 2016-04-29 09:01:28 packages should not stay in testing 2016-04-29 09:01:39 either they move to main or community 2016-04-29 09:01:49 that is if someone wants to maintain them 2016-04-29 09:01:51 if not 2016-04-29 09:01:56 it goes to unmaintained 2016-04-29 09:02:08 after N months, if nobody wants to pick it up 2016-04-29 09:02:11 we git rm it 2016-04-29 09:02:35 testing could be renamed 'staging' too 2016-04-29 09:03:18 s/after N months/after N months of not building or not working/ and I'm with you 2016-04-29 09:03:51 no need to remove stuff that just works and is simply awaiting a maintainer 2016-04-29 09:04:13 renaming testing will cause apk breakages for those on edge and who don't follow what's going on 2016-04-29 09:04:37 ScrumpyJack: yes, thats why i havent renamed it. the cost for doing so is high 2016-04-29 09:04:53 skarnet: i wonder if we can use "unmaintained" for that 2016-04-29 09:05:13 its not removed, its just moved out of the way for the builders 2016-04-29 09:05:17 that's exactly what unmaintained means 2016-04-29 09:06:11 but again, "unmaintained but works" and "broken" are different things, even if it's easy for a package in the former to fall into the latter 2016-04-29 09:06:57 yes and that is also why i want move things out of from testing 2016-04-29 09:07:00 i mean 2016-04-29 09:07:14 sometimes i upgrade things that require rebuild of things 2016-04-29 09:07:19 anyway, those are my outsider inputs, feel free to ignore them, you're the boss :D gotta afk for a while. 2016-04-29 09:07:23 it slwos down the upgrade 2016-04-29 09:07:50 at this point i often try to build it in testing 2016-04-29 09:08:03 and if it fails, then its a convenient time to move it to unmaintained 2016-04-29 09:12:48 other issue: ERROR: virt-manager-1.3.2-r1: trying to overwrite usr/share/glib-2.0/schemas/gschemas.compiled owned by shotwell-0.23.0-r0. 2016-04-29 09:13:10 ACTION leaves unmaintained alone 2016-04-29 09:13:38 ScrumpyJack: do you have stats? 2016-04-29 09:13:53 how many packages are there that are older than 3.0? 2016-04-29 09:18:01 lets remove packages in unmaintained thats older than 3.0-stable branch 2016-04-29 09:18:06 i think thats a good compromise 2016-04-29 09:18:14 if package has been there 2 years 2016-04-29 09:18:19 then its clearly dead 2016-04-29 09:18:24 and i think we can remove it 2016-04-29 09:19:38 but i wonder if we should git rm one by one, so they get listed in git log 2016-04-29 09:21:38 something like: "unmaintained/foobar: deleted (unmaintained for more than 24 months)" 2016-04-29 09:21:49 or similar 2016-04-29 09:22:44 then you can get an overview on what has been there 2016-04-29 09:24:50 how about a history-<20160429>.readme file that contains deleted files list with checkpoint, this gives a starting point for someone searching of purged pkgs ? 2016-04-29 09:25:09 i think its unecessary double work 2016-04-29 09:25:13 the info is in git log 2016-04-29 09:26:01 I mean for some easy grepping 2016-04-29 09:26:21 git grep? 2016-04-29 09:26:24 back. stuff in since 3.0 http://sprunge.us/CUIF 2016-04-29 09:26:26 ok 2016-04-29 09:27:49 ScrumpyJack: i say purge it 2016-04-29 09:27:56 purge it 2016-04-29 09:27:56 perhaps post to alpine-devel with the list first? 2016-04-29 09:28:07 ok 2016-04-29 09:28:08 coffee 2016-04-29 09:28:09 yes 2016-04-29 09:28:10 do that 2016-04-29 09:28:31 good idea 2016-04-29 09:28:39 the coffee? thanks :) 2016-04-29 09:28:47 :) 2016-04-29 09:28:49 that too 2016-04-29 09:28:52 post to alpine-devel 2016-04-29 10:01:15 py-django-registration failed checksum. it works here. 2016-04-29 10:07:20 have you pulled to dozer since last py-django-registration commit? 2016-04-29 10:48:36 hey friends :) 2016-04-29 10:49:00 how is everyone? 2016-04-29 10:49:09 have you played around already with the new gcc? 2016-04-29 11:04:18 ncopa: https://mta.openssl.org/pipermail/openssl-announce/2016-April/000069.html 2016-04-29 11:07:06 the real news would be if there were NO new security holes in OpenSSL 2016-04-29 11:09:16 i am more interrested in seeing how much of those problems are currently in libressl 2016-04-29 11:09:17 :) 2016-04-29 11:09:49 my intuition says "none" 2016-04-29 11:12:17 maybe one 2016-04-29 11:12:20 but not more 2016-04-29 11:24:54 skarnet: you have some knowledge about pid1 and so on 2016-04-29 11:25:08 how is your knowledge around bootloaders, ... 2016-04-29 11:25:09 ? 2016-04-29 11:26:12 hi 2016-04-29 11:26:30 any ideas on how to include a thirdparty .ko file to load up during boot 2016-04-29 11:26:39 i tried putting the link to /etc/module 2016-04-29 11:26:44 but doesnt pick it up 2016-04-29 11:28:08 leo-unglaub: if it's not user-space, I know next to nothing about it :/ 2016-04-29 11:29:41 hmm, same problem here 2016-04-29 11:30:38 also, to answer your tweet: I can't make heads or tails of the OpenBSD repository, and I've already decided I'm not going to invest time to fix things that obviously fall under someone else's responsibility 2016-04-29 11:30:51 so I reported the bug, and I expect *them* to fix it 2016-04-29 11:31:20 skarnet: my question had nothing todo with the tweet. i also wanted ted to fix it and post a patch 2016-04-29 11:31:39 my question had a different background 2016-04-29 11:31:40 they can do it in 5 minutes, whereas taking the time to learn things, install OpenBSD and craft a patch... would take me a day or so 2016-04-29 11:33:12 leo-unglaub: I think you should clarify you're asking *Ted* for a patch, because your tweet could be interpreted as you asking *me*, and I don't want Ted to wait for a patch that will never come unless he writes it :P 2016-04-29 11:33:47 skarnet: i know him, he propobly knows that he has to do it himself *g* 2016-04-29 11:33:55 I see :D 2016-04-29 11:34:14 also, "probably" 2016-04-29 11:34:41 he is the first in the @ order ... so he should know that he has to do it *g* 2016-04-29 11:34:49 if he does not, i am going to ping him again 2016-04-29 11:34:50 *g* 2016-04-29 11:35:09 yeah, theoretically that's how it works, but bad faith is so easy to come by when you don't want to do something ;) 2016-04-29 11:37:41 hehe 2016-04-29 11:37:49 we will see if he does something about it 2016-04-29 11:38:22 I'm counting on you to nag him until he does! 2016-04-29 11:38:51 becuase i tend to agree with you, that the current behavour of ppol is wrong 2016-04-29 11:38:54 (yes, the bug breaks s6 on OpenBSD) 2016-04-29 11:39:15 hehe, as much as i love OpenBSD ... as much i hate it *g* 2016-04-29 11:39:54 my biggest problem currenty, the old and outdated compiler in base ... 2016-04-29 11:40:14 you can basically dont use any "new" c features because the compiler is so out of date 2016-04-29 11:40:30 none of the gcc5 features is available 2016-04-29 11:40:56 so, no C11 2016-04-29 11:41:01 sadly no :( 2016-04-29 11:41:14 and i normally code c11 ... so thats a huge problem 2016-04-29 11:41:18 meh, as long as it does C99 I'm ok with it :P 2016-04-29 11:42:20 skarnet: i joked to ncopa once, that he should to an AlpineBSD :) use all the good stuff from OpenBSD (kernel, security, utilities, bootloader) and then put on top the awesomeness and cleanness from alpine linux like apk, the file structure, .... 2016-04-29 11:42:56 that's not even too far-fetched 2016-04-29 11:43:40 the problem would be low-level utilities such as routing stuff etc. 2016-04-29 11:45:11 i am not seeing a big problem there 2016-04-29 11:45:27 routing, device nodes, ... works all very well on openbsd 2016-04-29 11:45:38 the kernel APIs are different 2016-04-29 11:45:44 the only problem that i have is the static /dev/ folder 2016-04-29 11:45:48 so the userland utilities have to be different, too. 2016-04-29 11:46:00 it pisses me off every time i have to generate a new device manually 2016-04-29 11:46:43 i would use a lot of userland utilities from them already, because most of them are very great integrated and have full pledge support 2016-04-29 11:47:07 but you would want alpines package manager, the alpine cleanness and some of the alpine confug structure 2016-04-29 11:47:23 and you want s6 instead of the current openbsd one ;) 2016-04-29 11:47:37 that's obviously the first thing to change :P 2016-04-29 11:47:41 *g* 2016-04-29 11:55:49 i really need help ... taking about alpine mixed with openbsd makes me horney *g* hahahahaha 2016-04-29 11:55:55 thats clearly not normal 2016-04-29 11:55:56 heheh 2016-04-29 12:23:50 >i really need help ... taking about alpine mixed with openbsd makes me horney *g* hahahahaha 2016-04-29 12:23:52 wut 2016-04-29 12:28:43 nanopi m1 quad-core arm + 148f:7601 Ralink Technology, Corp. MT7601U in AP mode + Alpine = most advanced, stable, crash proof, wi-fi router, firewall 2016-04-29 12:28:50 plus the cheapest 2016-04-29 12:28:55 under $10 2016-04-29 12:29:03 woaaa who could have guessed the power of Alpine 2016-04-29 12:32:40 oneinsect_: $10? 2016-04-29 12:32:55 almost ...yes 2016-04-29 12:33:00 1.5$ for wifi 2016-04-29 12:33:07 around 10$ for board 2016-04-29 12:33:21 and $1 for sdcard 2016-04-29 12:33:35 should be around $10 with orange pi 2016-04-29 12:33:38 in volume 2016-04-29 12:41:56 build-3-4-x86 is up 2016-04-29 12:43:46 is the APKBUILD file on dozer release 0 or 1? 2016-04-29 12:44:01 ? 2016-04-29 12:46:20 sorry. ignore me, it's r1 main/py-django-registration 1.0-r1 2016-04-29 12:48:16 grep you grep sha512sum /home/buildozer/aports/main/py-django-registration/APKBUILD if it makes sense? 2016-04-29 12:49:34 on build-edge-armhf? 2016-04-29 12:49:46 i cannot easily log in to that atm 2016-04-29 12:49:52 i think it has the old tarball in the cache 2016-04-29 12:50:01 so simply removing it from the cache should fix it 2016-04-29 12:51:22 doesn't a sum fail delete the file? 2016-04-29 12:52:39 apparently not 2016-04-29 12:54:21 bah 2016-04-29 12:54:26 circular dep 2016-04-29 12:54:32 eudev and util-linux 2016-04-29 12:55:26 argh 2016-04-29 12:58:25 why does util-linux depend on eudev? o.O 2016-04-29 12:58:35 and the other way around 2016-04-29 12:58:47 apparently it is not a real need 2016-04-29 12:59:50 checking for BLKID... no 2016-04-29 12:59:58 I use mdev everywhere, eudev is really not needed for containers and servers. 2016-04-29 13:00:21 i think eudev needs blkid for libgudev 2016-04-29 13:09:47 have a good weekend all o/ 2016-04-29 13:15:43 ScrumpyJack: you too! 2016-04-29 13:42:02 oh no /o\ 2016-04-29 13:42:26 apk add samba-dev && ldd /usr/lib/libsmbclient.so 2016-04-29 13:42:33 libsmbclient is broke 2016-04-29 13:43:24 hm 2016-04-29 13:43:30 might need to set rpath 2016-04-29 13:43:55 no, rpath is set 2016-04-29 13:59:19 ncopa, what does it mean? "stderr: chmod /usr/bin/abuild-sudo: permission denied" 2016-04-29 13:59:22 i'm root 2016-04-29 14:00:06 when does it happen? 2016-04-29 14:00:16 it is not allowed to chmod abuild-sudo 2016-04-29 14:00:30 is it in a chroot with grsec kernel? 2016-04-29 14:00:31 whys so, since i'm root? 2016-04-29 14:00:34 Ah 2016-04-29 14:00:39 yeah 2016-04-29 14:00:47 the disable_grsec in docker conf.d 2016-04-29 14:00:52 always forget :-/ 2016-04-29 14:01:04 thx ncopa, sorry 2016-04-29 14:01:05 Remember: no Docker, no problems ;) 2016-04-29 14:01:31 i have a problem with samba 2016-04-29 14:01:40 libsmbclient is broken 2016-04-29 14:04:35 and i think i have fixed it 2016-04-29 14:04:37 phew 2016-04-29 14:32:35 3.4 builders are up 2016-04-29 14:32:46 and will build over the weekend 2016-04-29 14:32:55 next we we clean up what failed 2016-04-29 14:33:09 nice 2016-04-29 14:36:05 what does “Redundant /usr/lib in rpath found” actually mean? 2016-04-29 14:36:50 it means that something ws linked with rpath set to /usr/lib/ 2016-04-29 14:36:58 its not needed because thats the default 2016-04-29 14:37:23 rpath? 2016-04-29 14:37:55 you can in ELF set a search path for the libs 2016-04-29 14:38:05 ah, this path 2016-04-29 14:38:08 readelf -d $path/lib.so 2016-04-29 14:38:34 builder status: http://build.alpinelinux.org/ 2016-04-29 14:38:47 nice :) 2016-04-29 14:39:35 btw do you have enought resources for build servers etc? I have some servers on faculty which I might use for Alpine 2016-04-29 14:48:16 we have 2 big servers 2016-04-29 14:48:33 each with 48GB ram and 12 cores 2016-04-29 14:48:43 its enough so far 2016-04-29 14:48:48 okay :) 2016-04-29 14:49:05 how do you build ARM packages? 2016-04-29 14:49:10 wandboard 2016-04-29 14:49:23 we have 2 wandboard quads 2016-04-29 14:49:30 and a third is here on my desk 2016-04-29 14:49:41 nice :) 2016-04-29 14:49:43 i'm mounting a ssd disk via sata 2016-04-29 14:49:59 the current arm builder uses usb disk 2016-04-29 14:50:03 which is slow 2016-04-29 14:50:44 i only need to make 4 holes in the case where i will attach the disk 2016-04-29 14:50:52 one colleague of mine is currently building some demo computation cluster for Spark on Raspberry Pi 2016-04-29 14:50:59 cool :) 2016-04-29 14:51:24 I’ve told them that Raspberry is not the best option… of course he has problems with RAM, Spark is very hungry 2016-04-29 14:52:15 and I have to go to train, see ya 2016-04-29 14:57:27 have a nice weekend everyone! 2016-04-29 15:33:31 nooooo 2016-04-29 15:44:52 skarnet: he alredy fixed it ;) 2016-04-29 16:53:10 is there any maxscale package available? 2016-04-29 16:53:19 or does nobody of you cluster mariadb? :) 2016-04-29 17:11:33 Hello guys, I am actually new to this community and ready to contribute, but I still lack some communication skills 2016-04-29 17:13:31 I basically have two questions: a. I posted a patch to the devel ml to add a package to the testing tree, how should I procede to keep contributing? b. I am fixing a post-install script on alpine:edge, where should I send my patches? 2016-04-29 17:20:23 kaiyou: everything related to package patches should be sent to aports mailing list 2016-04-29 17:22:27 okay, subsribing now and I will lurk through the archives before I send anything 2016-04-29 17:22:28 thanks 2016-04-29 17:24:57 there's a page on wiki on how to submit patches 2016-04-29 17:29:32 kaiyou: you probably don't need to subscribe to alpine-aports; doing so will spam you with all the incoming patches. 2016-04-29 17:29:42 Subscribing to alpine-devel should be enough. 2016-04-29 17:30:10 is it OK to send patches to -devel ? 2016-04-29 17:30:28 I know it works but I always thought that it was misplaced... 2016-04-29 17:30:44 afaik no, but you can send to -aports without being subscribed 2016-04-29 17:31:09 oh yeah until the whitelist mode really happens :D 2016-04-29 17:31:51 unless ncopa made a special exception for me, I think it has happened 2016-04-29 17:34:59 then you can't send anymore without subscribing right ? 2016-04-29 17:37:36 I'm not sure. You probably can, but you're subject to the spam blacklists, I guess. 2016-04-29 17:40:25 I did not subscribe anyway and just sent these patches, thank you all 2016-04-29 19:22:25 anyone knows if new stable nginx 1.10 ahs been picked up for 3.4 ? 2016-04-29 19:23:10 it includes new dynamic modules that could result in sub packages and less deps for the main one 2016-04-29 19:23:58 also things like naxsi or modsecurity could potentially - haven't given much try - become subpackages 2016-04-29 19:31:55 clandmeter: https://pkgs.alpinelinux.org/packages?name=nginx&branch=&repo=&arch=&maintainer= - look at only arm being green ;) that's why I suggested some basic tests, because clearing with each new build is too general 2016-04-29 19:32:55 this means that the package is outdated, doesn’t it? 2016-04-29 19:33:08 red means it has been flagged 2016-04-29 19:33:16 flagges as outdated 2016-04-29 19:33:24 s/ges/ged/ 2016-04-29 19:33:38 Yeah, so what tests do you mean? 2016-04-29 19:33:55 that version actually changed 2016-04-29 19:34:10 ? 2016-04-29 19:34:54 wait a moment, shouldn’t this flag be for whole package, not just specific platform? 2016-04-29 19:35:04 s/platform/architecture/ 2016-04-29 19:37:23 new build is what clears the flag. it makes sense, because you cannot clear flag for all archs if they aren't ready yet. the problem is it clears unconditionally 2016-04-29 19:43:17 clandmeter: but the flag date is apparently past build date, so I'm not sure why it became green. was it related to some ncopa's fiddling with builders? but if so, the build date should be more recent too, right? 2016-04-29 19:47:45 I wonder why your mail, ScrumpyJack, isn't shown in http://lists.alpinelinux.org/alpine-devel/. it is in my mailbox, though. 2016-04-30 13:59:46 hey :) 2016-04-30 13:59:54 has someone of you packaged Tixati? 2016-04-30 14:00:08 https://www.tixati.com/ 2016-04-30 14:48:30 kmod-usb-audio