2015-01-01 10:27:54 Happy New Year! :) 2015-01-01 15:24:48 thank you ncopa :) 2015-01-01 15:52:04 np 2015-01-02 01:17:03 in APKBUILD can I use "if [ `uname -m` = "x86_64" ]; " (with a $subpkg-lib32) - or should I make 2 x packages (arch=all) + 2nd as $pkg32 for arch=x86_64 ? 2015-01-02 06:50:35 there is a variable for that, check abuild.conf 2015-01-02 06:50:46 CHOST iirc 2015-01-02 11:18:30 ncopa, i'm giving a look to rt4 (Request Tracker) alpine package. Would be worthy to have a subpkg with lighttpd conf file and nginx conf file? If so, how can we call them? 2015-01-02 11:56:51 fcolista: how big are those? 2015-01-02 11:57:14 if they are big or pull in multiple dependencies, then yes, it might be worth create subpackage for them 2015-01-02 12:07:25 just a conf file for subpkg ncopa 2015-01-02 12:07:59 I'd put them in /usr/share/$pkgname/ 2015-01-02 12:08:04 or sth like that 2015-01-02 12:18:08 yeah, if the files are not very big, then there is not reason to subpkg them 2015-01-02 12:18:23 soudns good. 2015-01-02 12:18:33 you could also have an rt4-example-configs package where you collect them all 2015-01-02 12:19:02 rt4-4.2.9-r0 installed size: 2015-01-02 12:19:02 26054656 2015-01-02 12:19:18 i wouldnt split the package unless its 2MB or bigger 2015-01-02 12:28:01 ncopa: if i need qemu with sdl in the framebuffer (no x11) what would be a good way to do it? local packages, or qemu-sdl or qemu-sdl-fb subpackages? 2015-01-02 12:28:44 i'd say qemu-sdl-fb subpackages 2015-01-02 12:29:12 hmmm, someone made coreboot+debian(sid) to boot in 1.2 secs (with systemd) i guess i have to show those guys how to do this with runit or s6 on alpine... challenge accepted. 2015-01-02 12:29:49 ncopa: oki thx, i'll make qemu-sdl-fb packages then. 2015-01-02 12:30:01 also i will try to add libseccomp 2015-01-02 12:30:16 and will compile qemu with seccomp so we have the sandbox feature in alpine 2015-01-02 12:30:27 also maybe other packages should be checked for seccomp support. 2015-01-02 12:33:36 hmmm, ncopa you mean subpackage? as in built by the main/qemu/APKBUILD, so i have 2 builds in build()? or a seperate package like testing/qemu-sdl-fb? 2015-01-02 12:48:00 can you give an example apkbuild where you do two builds for similar reasons with different subpackages so i can learn/copy 2015-01-02 14:46:11 tvvdhSL5hEBn: i'd say a subpkg in qemu 2015-01-02 14:46:31 zabbix 2015-01-02 14:46:32 php 2015-01-02 14:46:35 are examples 2015-01-02 14:46:49 zabbix has multiple builds with different configs 2015-01-02 14:47:09 due to zabbix has db support as compiletime option instead of plugin 2015-01-02 14:47:29 i'd be interested in a qemu-gtk subpackage too 2015-01-02 15:12:37 thx! will try 2015-01-02 20:31:53 hmmm, i couldn't find a good description of how to do this compiling multiple versions. i started with sdl as this is a dependency to qemu, and it does link x11 by default, so like the qemu, i wanted to do a sdl-fb. but somehow the abuild process fails: /main/sdl/APKBUILD -> http://sprunge.us/gQSS 2015-01-02 20:33:17 this is the output of `abuild rootpkg`: http://sprunge.us/DTKh 2015-01-02 20:34:49 the dependencies of sdl-dev contain also sdl-fb for some reason, and sdl-fb has pkgconfig as a dep, but sdl itself doesn't. i don't think this is right. but i fail to see how to do this correctly. 2015-01-02 20:45:36 also how do i do a sdl-fb-dev pkg? if i say sdl-fb:dev that does not work as expected. 2015-01-03 07:17:43 ncopa: Evil_Bob new man w/ mdocml 1.13.2-r1 doesnt work when i do "man ls" 2015-01-03 07:18:01 i get the following error: "man: /usr/share/man/man1/ls.1: SYSERR: No such file or directory" 2015-01-03 07:18:27 if i symlink /usr/share/man/man1/ls.1p to /usr/share/man/man1/ls.1, then it works 2015-01-03 19:19:44 Hi all, I wonder if there is any way to generate the md5 and sha automatically on a APKBUILD 2015-01-03 19:20:44 I have a lot of patches in a package and is tedious to manually generate 2015-01-03 19:26:15 have you tried `abuild checksum` 2015-01-03 19:28:55 tvvdhSL5hEBn, thank you very much 2015-01-04 02:48:27 hi 2015-01-04 02:49:10 ncopa, http://sprunge.us/TQCh <-- package dietlibc 2015-01-04 02:50:36 ncopa, http://sprunge.us/hVbX <-- package runit (minor changes) to your consideration 2015-01-04 07:55:39 huh, dietlibc? 2015-01-04 07:55:42 we have musl already 2015-01-04 09:34:27 2deep4u 2015-01-04 10:29:26 are we switching from musl libc to dietlibc? 2015-01-04 10:30:37 plueschopath, I do not think so 2015-01-04 10:30:41 huh? no. we did the other way around. 2015-01-04 10:33:01 vitronic: why linking runit against dietlibc? 2015-01-04 10:34:08 tvvdhSL5hEBn, especially if used as replacement openrc 2015-01-04 10:34:48 to be No. 1 process, must be statically linked 2015-01-04 10:35:38 hmm, and musl cannot be statically linked? 2015-01-04 10:36:27 of course it can 2015-01-04 10:36:37 having dietlibc is pointless 2015-01-04 10:36:46 possibly, but the official developer suggests using dietlibc 2015-01-04 10:36:57 ah 2015-01-04 10:37:02 I just follow the suggestion 2015-01-04 10:37:31 we add dietlibc and we try to switch to another init system? 2015-01-04 10:37:32 and on the side note, it's not how we package standard libraries… 2015-01-04 10:37:32 does the developer suggest dietlibc vs musl or vs glibc? 2015-01-04 10:37:39 and dietlibc is pretty much dead 2015-01-04 10:37:45 so there is nothing to be discussing about 2015-01-04 10:38:14 anyway, my APKBUILD is just a suggestion, by no means is official 2015-01-04 10:38:36 I repeat, it is not official, only suggestion 2015-01-04 10:38:36 plueschopath: we don't add anything, we have musl, full stop 2015-01-04 10:38:47 can you give more context on "does the developer suggest dietlibc vs musl or vs glibc?" 2015-01-04 10:38:58 barthalion: I'm talking about the init system ... sorry 2015-01-04 10:39:11 tvvdhSL5hEBnv coreboot+debian(sid) to boot in 1.2 secs (with systemd) on an X60s ? 2015-01-04 10:39:27 runit was mentioned here before 2015-01-04 10:39:43 crow yeah i mentioned that a few days ago here 2015-01-04 10:40:14 crow in this context booting = until loginprompt 2015-01-04 10:40:20 is there anything wrong with openrc that we talk about changing init every month? 2015-01-04 10:40:31 barthalion: don't think so :D 2015-01-04 10:40:41 I'm happy with openrc :3 2015-01-04 10:40:49 my bet is that everyone is 2015-01-04 10:41:32 ACTION is a longtime runit user in the process of switching some boxes to s6 for experiments. 2015-01-04 10:42:01 (if anyone wants skalibs/execline/s6 apks ping me i have v2.0.0.0) 2015-01-04 10:42:05 I just runit use for qmail, but I decided to modify the package to make it more complete 2015-01-04 10:42:17 http://skarnet.org/software/s6/ this one? 2015-01-04 10:42:22 + 2015-01-04 10:42:40 and like this in testing, I seemed a good idea to share my APKBUILD 2015-01-04 10:43:09 tvvdhSL5hEBn: does it compile with musl? 2015-01-04 10:43:19 shortly publish my qmail package 2015-01-04 10:43:31 tvvdhSL5hEBn i have an x60s here prepared for something like that, but still no external flasher or clamp in case i broke bios chip 2015-01-04 10:43:34 vitronic: can you give some context on the devs recommendation on using dietlibc? was that in regards glibc or was he really also considering musl? 2015-01-04 10:43:51 tvvdhSL5hEBn, >http://smarden.org/runit/ 2015-01-04 10:43:59 tvvdhSL5hEBn, >http://smarden.org/runit/usedietlibc.html 2015-01-04 10:44:23 that is no recommendation, but a howto 2015-01-04 10:45:21 sorry, I can not understand, do not understand English well 2015-01-04 10:45:58 ie I do not understand English well 2015-01-04 10:46:18 ah skalibs ... I see 2015-01-04 10:46:21 vitronic: http://www.etalabs.net/compare_libcs.html you can easily see here that dietlibc is simply bad 2015-01-04 10:47:31 barthalion, understand, thanks for the clarification 2015-01-04 10:47:45 np 2015-01-04 10:48:40 crow: you can do coreboot also on other thinkpads, also even a buspirate is enough to flash the bios 2015-01-04 10:49:01 in any case, qmail not required to anything runit I is linked with dietlibc 2015-01-04 10:49:22 Parsing is a complex operation, and it is easy to make mistakes while designing and implementing a parser. :D 2015-01-04 10:49:39 wrong c&p 2015-01-04 10:49:43 This site is glibc-free, and uses no GNU software during its normal operation. :D 2015-01-04 10:51:20 that's the spirit :P 2015-01-04 10:53:48 hmm, i wonder how long it takes until someone comes along and posts glibc and systemd apks because some software recommends to link against those :) 2015-01-04 10:55:52 crow: this is what i was referring to: http://www.coreboot.org/pipermail/coreboot/2014-December/079053.html 2015-01-04 10:58:11 tvvdhSL5hEBn: I'm interested in the apks you've talked about. 2015-01-04 10:58:21 would like to experiment with s6 a bit 2015-01-04 10:59:00 crow: in case you want to solder a socket: http://www.coreboot.org/Soldering_a_socket_on_your_board 2015-01-04 11:03:22 plueschopath: http://sprunge.us/ieKQ http://sprunge.us/bbOJ http://sprunge.us/REZa 2015-01-04 11:03:29 apkbuild files. not packages 2015-01-04 11:03:38 i hope you can rebuild them yourself. 2015-01-04 11:03:46 src > bin 2015-01-04 11:04:48 thank you :) 2015-01-04 11:05:23 tvvdhSL5hEBn yes i am subscribed to that ML and saw that to. 2015-01-04 11:05:56 tvvdhSL5hEBn i know that buspirate is enough, but i dont have it. dont have any external programator, so in case of bad lack i habe soft bricked x60s 2015-01-04 11:07:01 meh. a buspirate is cheap and handy in many other situations. i would never leave home without it. especially the jtag support of it. <3 2015-01-04 11:07:11 how much is a buspirate nowadays? 30usd? 2015-01-04 11:07:40 softbricked, i'm sure it can be unbricked easily. ;) 2015-01-04 11:09:35 tvvdhSL5hEBn soldering an socket on laptop board is not that trivial, think easly would be some bus pirate and the POMONA SOIC Clip 2015-01-04 11:10:04 tvvdhSL5hEBn its actually not about money, but to leasy to order all parts :D 2015-01-04 11:12:11 I installed the last freeswitch apk and I encountered a problem. The freeswitch package look at /usr/sounds/ for sound files, but the sounds packages are installed in /usr/share/freeswitch/sounds. Please, can someone help me to solve this problem.... 2015-01-04 11:12:58 for a quick workaround, just make a symbolic link 2015-01-04 11:13:20 but please, report the issue on our bug tracker, so it will be fixed properly and the fix backported to the stable releases 2015-01-04 11:13:37 thank you 2015-01-04 11:13:47 I will do right that 2015-01-04 11:14:02 have a good day 2015-01-04 11:34:17 Full apkbuild to runit without dietlibc http://sprunge.us/GgTW me apologize for the shock, it was by no means my intention, I only intend to collaborate 2015-01-04 11:35:07 generated package is already fully usable 2015-01-04 11:35:12 didn't we have runit already without dietlibc? 2015-01-04 11:35:23 what's the difference to the original version? 2015-01-04 11:36:02 ah, it's a diff 2015-01-04 11:36:05 sorry for asking 2015-01-04 11:36:22 what are the changes, it's a bit big 2015-01-04 11:38:44 nice 2015-01-04 11:39:03 except for the whitespace and the s/"/'/ 2015-01-04 11:39:52 ncopa, http://sprunge.us/GgTW <-- finally, qmail to alpine , please notify if there is an error 2015-01-04 11:41:00 vitronic, does runsvdir.start work? dunno if alpine/openrc has start-stop-daemon 2015-01-04 11:42:32 what's the difference to the original version? <-- basically, now only is install and use , before not... 2015-01-04 11:46:40 vitronic: "nice. except for the whitespace and the s/"/'/" 2015-01-04 11:47:57 tvvdhSL5hEBn, you're right, I will include an init script in /etc/init.d/runit 2015-01-04 11:48:52 vitronic: "nice. except for the whitespace and the s/"/'/" <-- You can tell me the line number 2015-01-04 11:49:20 you changed whitespace in the APKBUILD and you changed a lot of " into ' 2015-01-04 11:49:24 see your diff 2015-01-04 11:52:02 tvvdhSL5hEBn> vitronic: "nice. except for the whitespace and the s/"/'/" <-- apology, still do not understand 2015-01-04 11:52:26 -pkgdesc="UNIX init scheme with service supervision" 2015-01-04 11:52:31 +pkgdesc='UNIX init scheme with service supervision' 2015-01-04 11:52:34 etc 2015-01-04 11:52:43 - echo "$CC $CFLAGS" > conf-cc 2015-01-04 11:52:45 - echo "$CC $LDFLAGS" > conf-ld 2015-01-04 11:52:47 + echo "$CC $CFLAGS" > conf-cc 2015-01-04 11:52:49 + echo "$CC $LDFLAGS" > conf-ld 2015-01-04 11:52:53 ^^^ whitespace 2015-01-04 11:55:56 tvvdhSL5hEBn, Now understand, I use single quotes when no variables declared, and double quotes where you declare variables, it is a good habit 2015-01-04 11:56:17 but I understand your reasoning 2015-01-04 11:59:02 why is that a good habbit? 2015-01-04 11:59:14 in terms of white space, I suppose, that are caused by IDE tab 2015-01-04 11:59:19 i use quotes for the purpose i need them, do i want variable expansion or not 2015-01-04 11:59:30 not the other way aroudn 2015-01-04 12:00:39 why is that a good habbit? <-- Saves CPU cycles by not having to check for variables in the chain 2015-01-04 12:01:18 hmm, makes sense i guess 2015-01-04 12:01:37 guess it's a habit acquired 2015-01-04 12:02:24 but also fits into the "premature optimization" category, i dunno, are there systems where "having to check for variables in the chain" is costly? 2015-01-04 12:02:40 execline makes more sense to me, avoiding forks. 2015-01-04 12:02:43 I am pleased to have clarified your question 2015-01-04 12:04:11 tvvdhSL5hEBn, that meaning has execline? 2015-01-04 12:06:02 I can imagine the context 2015-01-04 12:06:41 mea culpa, sorry :-D 2015-01-04 12:09:25 do you have measurements, how big the difference is between ' and " ? do you have these also across various shells ash/bash/zsh/etc? 2015-01-04 12:09:29 vitronic: ^^^^ 2015-01-04 12:11:35 tvvdhSL5hEBn, for me it is difficult to understand the question 2015-01-04 12:12:12 used the translator, but not helping me 2015-01-04 12:12:54 You may already spent my bedtime 2015-01-04 12:19:13 what is the time difference between using " and ' in various shells? 2015-01-04 12:19:17 vitronic: ^^^ 2015-01-04 12:19:34 did you measure this? do you have references who did measure this? 2015-01-04 12:20:32 I really do not know, only I used bash, ash and dash 2015-01-04 12:21:51 but I have no references, only theory 2015-01-04 12:22:14 I'm not compulsive 2015-01-04 12:24:22 so you optimize without knowing if it makes a difference? 2015-01-04 12:27:32 tvvdhSL5hEBn, If there are differences, but possibly in a small script there is none 2015-01-04 12:29:13 continue in #alpine-linux please 2015-01-04 12:31:45 i guess: case closed ;) 2015-01-04 12:33:40 tvvdhSL5hEBn, by the way, thanks for the suggestions on runit, thanks to that, I realized that this best script in /etc/init.d 2015-01-05 07:49:47 morning 2015-01-05 07:50:12 morning 2015-01-05 08:13:51 morning 2015-01-05 08:25:33 first day back at work :( 2015-01-05 08:25:46 I know that pain 2015-01-05 09:03:30 morning :) 2015-01-05 11:29:49 moin 2015-01-05 11:31:56 moin 2015-01-05 13:11:33 ncopa: remember me, wanted to do qemu-*-fb subpackages. but i got tangled up in the dependency rabbithole. i was looking at sdl, and saw it depends on X11 and co. so i thought, meh, i need to do an sdl-fb subpackage as well. which i kinda succeeded at, but i was unable to correctly do sdl-fb-dev subpackages, so i can use that to link qemu-fb. but i failed to find out how to do both sdl-dev and 2015-01-05 13:11:35 sdl-fb-dev as subpackages one linking x11 the other not. this is main/sdl/APKBUILD http://sprunge.us/ifXA - is it possible that this is not possible using subpackages? 2015-01-05 13:20:28 and this is the output of abuild -r : http://sprunge.us/UMQF 2015-01-05 13:21:05 sorry very busy around here cannot help you atm 2015-01-05 13:22:19 ncopa: no problem, can you have a look later this week? not urgent. 2015-01-05 13:22:21 i hope so 2015-01-05 13:23:05 is there anything we can help? 2015-01-05 16:53:40 i have a few packages that might be of interest, libseccomp, s6+dependencies, and percy++ with dependencies: http://www.ctrlc.hu/~stef/apkgs.txz 2015-01-05 16:56:05 for percy++ context see: http://cdn.media.ccc.de/congress/31C3/h264-hd/31c3-6140-en-de-DP5_PIR_for_Privacy-preserving_Presence_hd.mp4 2015-01-05 16:59:58 meh. actually percy++ does only build, but not package. i forgot to add that. sorry. but still enough for testing it out... 2015-01-06 11:07:37 http://hyperdex.org/ , possibly be used instead of mongodb(unmaintained) 2015-01-06 11:33:57 vkrishn, also redis 2015-01-06 11:35:31 wondering why there is no redis here, http://hyperdex.org/performance/ 2015-01-06 11:36:21 vkrishn: you know about postgres json support? since 9.4 no need for these immature tools 2015-01-06 11:36:31 ;) 2015-01-06 11:58:04 postgres ftw :D 2015-01-06 12:34:22 seems useful, http://www.ltrace.org/ 2015-01-06 13:41:56 http://git.alpinelinux.org/cgit/aports/tree/main/ltrace 2015-01-06 13:45:02 ah, thanks 2015-01-06 14:14:04 ncopa: did you saw my APKBUILD for mariadb? 2015-01-06 14:17:45 scadu: not yet 2015-01-06 14:18:40 scadu: where did you send it? 2015-01-06 14:19:06 ncopa: mail list. 2015-01-06 14:19:06 do we have support for newer usb spec, http://www.techspot.com/news/52321-usb-30-superspeed-update-to-eliminate-need-for-chargers.html 2015-01-06 14:19:14 http://www.usb.org/press/presskit/USBIF_MomentumPR_IDF2013_FINAL.pdf 2015-01-06 14:20:47 oh 2015-01-06 14:20:48 nice 2015-01-06 14:26:32 probably I could do better with subpackages. 2015-01-06 14:37:19 ncopa: let me know if you have any suggestions or sth. 2015-01-06 14:45:10 what I don't understand is the need for increased power delivery up to 100 watts 2015-01-06 14:50:13 and they don't mention if this would be childsafe/proof 2015-01-06 14:50:53 seems someone ported alpine to have llvm as native compiler: https://github.com/D-os/aports/tree/master/core 2015-01-06 14:50:58 was announced on musl devel list 2015-01-06 14:51:04 and they don't mention if this would be childsafe/proof 2015-01-06 14:51:30 http://www.openwall.com/lists/musl/2015/01/06/3 2015-01-06 15:02:17 that is very cool 2015-01-06 15:46:06 scadu: mariadb fails to build: 2015-01-06 15:46:09 CMake Error at storage/tokudb/ft-index/cmake_modules/TokuFeatureDetection.cmake:95 (message): 2015-01-06 15:46:09 Cannot find backtrace(), even with -lexecinfo. 2015-01-06 15:46:45 I also see that jemalloc is used 2015-01-06 15:46:55 i wonder how musl libc's malloc compares 2015-01-06 15:47:08 might be the musl libc malloc is as fast 2015-01-06 15:58:42 ncopa: I tested on x86 and I didn't noticed problems during build 2015-01-06 16:02:43 except xtradb which is fixed already (see included patch) 2015-01-06 16:04:16 hum ok 2015-01-06 16:04:21 oh its late 2015-01-06 16:04:25 will continue tomorrow 2015-01-06 16:04:44 m'kay 2015-01-06 16:52:53 ncopa: I checked again and I can't reproduce this error 2015-01-07 07:03:04 http://wiki.alpinelinux.org/wiki/RPI_Video_Receiver 2015-01-07 07:03:11 i posted quick instructions 2015-01-07 07:03:15 should probably polish that a bit 2015-01-07 07:03:37 but it's nice, "pro looking" video decoder thing with Rpi and alpine 2015-01-07 07:43:06 hi@all 2015-01-07 07:43:40 did anyone else spot problems with differences between system time and used timestamps on log files? 2015-01-07 07:49:45 this seems to be a 3.x issue because on 2.x everything works fine 2015-01-07 07:50:35 it looks like this: 2015-01-07 07:50:38 root@builder06:~ date 2015-01-07 07:50:38 Wed Jan 7 08:50:13 CET 2015 2015-01-07 07:50:38 root@builder06:~ tail -n 1 /var/log/messages 2015-01-07 07:50:38 Jan 7 07:44:02 builder06 fcron[19334]: Job run-parts /etc/periodic/15min completed 2015-01-07 07:51:07 the date output is correct but on the log it seems to be UTC 2015-01-07 07:51:19 any ideas? 2015-01-07 07:59:19 hum 2015-01-07 07:59:42 maybe you have clock="UTC" in /etc/conf.d/hwclock 2015-01-07 08:02:21 i'll check.. 2015-01-07 08:03:03 exactly 2015-01-07 08:03:26 root@builder06:~ cat /etc/conf.d/hwclock 2015-01-07 08:03:26 ... 2015-01-07 08:03:26 clock="UTC" 2015-01-07 08:04:03 but this is the same on 2.7 and works as expected 2015-01-07 08:04:32 here's the output from a 2.7 box: 2015-01-07 08:04:33 root@builder07:~ date 2015-01-07 08:04:33 Wed Jan 7 09:03:42 CET 2015 2015-01-07 08:04:33 root@builder07:~ tail -n 1 /var/log/messages 2015-01-07 08:04:33 Jan 7 09:02:02 builder07 fcron[1516]: Job run-parts /etc/periodic/hourly completed 2015-01-07 08:05:59 ncopa, ping 2015-01-07 08:08:03 vitronic: pong 2015-01-07 08:08:12 StarWarsFan: i dont know really 2015-01-07 08:08:53 StarWarsFan: you could ask in #musl 2015-01-07 08:09:30 ncopa, I finished the qmail and vpopmail packages, I will show to see if you like 2015-01-07 08:10:27 I also made some modifications to runit package, such as the inclusion of a startup script 2015-01-07 08:10:44 regards, :-D 2015-01-07 08:13:44 ncopa: ok, thx 2015-01-07 10:02:34 fabled: neat 2015-01-07 10:11:24 ncopa: is it okay to take alpinelinux (I'm using about 30 packages and the apk-tools), modify it (I'm trying to make a very minimal version of it for kvm/xen virtual machines, with a custom kenrel, modified init, etc.) and publish the source on github under another name with credits to alpinelinux? 2015-01-07 10:11:29 fabled: can we put the fbsplash.ppm in the armhf.rpi.tar.gz? 2015-01-07 10:14:22 plueschopath: of course it is, it's open source after all 2015-01-07 10:20:11 plueschopath: yes. 2015-01-07 10:20:58 you are not required to credit alpine linux either 2015-01-07 10:21:03 but still nice if you do :) 2015-01-07 10:25:18 okay ty 2015-01-07 10:26:18 is it D-os? 2015-01-07 10:27:00 probably not 2015-01-07 10:27:16 in any case, i think its awesome :) 2015-01-07 10:29:38 D-os? 2015-01-07 10:31:45 i see 2015-01-07 10:31:52 no it's not D-os 2015-01-07 10:32:23 will post a link once I've made some more progress :D 2015-01-07 10:44:57 ncopa: the time-issue seems not to be new: http://www.openwall.com/lists/musl/2012/10/26/11 2015-01-07 10:46:23 i think they added /etc/localtime support since then 2015-01-07 10:46:49 is env TZ variable set? 2015-01-07 10:48:26 ah, this is interesting: http://www.openwall.com/lists/musl/2014/01/29/1 2015-01-07 10:49:36 no, TZ is not set 2015-01-07 10:51:32 i kinda agree with dalias there about using local time in syslog is questionable 2015-01-07 10:59:42 ScrumpyJack, no, fbsplash.ppm needs to be on bootmedia - it's looked early in the boot even before overlay is loaded 2015-01-07 11:09:03 fabled: i don't get it. 2015-01-07 11:09:13 ? 2015-01-07 11:10:32 fb.ppm is in / on the SD card no? 2015-01-07 11:10:45 yes 2015-01-07 11:11:11 and there is only one partition on the SD card 2015-01-07 11:11:15 yes 2015-01-07 11:11:46 so what can't the ppm go in the alpine rpi tar file that gets extracted to / on the SD card? 2015-01-07 11:12:04 maybe i'm missing something 2015-01-07 11:12:29 SD card is mounted in /media/mmcblk0p1 2015-01-07 11:12:49 and the splash screen is loaded in initramfs when we are still in bootstrap 2015-01-07 11:13:19 but the rpi-video-decoder.tar.gz 2015-01-07 11:13:24 is supposed to be extract to SD 2015-01-07 11:13:29 and it contains the .ppm 2015-01-07 11:13:33 so it ends up there 2015-01-07 11:13:56 rpi-video-decoder.tar.gz contains also the actual overlay (video-decoder.apkovl.tar.gz) which is not to be extracted 2015-01-07 11:14:02 not sure which tar.gz you mean 2015-01-07 11:14:35 alpine-rpi-3.1.1-armhf.rpi.tar.gz 2015-01-07 11:14:40 oh right 2015-01-07 11:14:44 we could maybe put it there too 2015-01-07 11:15:13 :) ok, sorry if i was vague 2015-01-07 11:15:29 there's only 3 different .tar.gz files involved ;) 2015-01-07 11:15:54 the only question is maybe if we want splash by default 2015-01-07 11:16:02 just having the .ppm file enable bootsplash 2015-01-07 11:16:29 i'm behind a proxy at work, so I don't think i can test you oxm build out 2015-01-07 11:16:45 unless it supports socks or works with tsocks 2015-01-07 11:18:35 i was thinking using local file also 2015-01-07 11:18:50 i think omxplayer honors http_proxy 2015-01-07 11:18:57 but not sure 2015-01-07 12:55:25 barthalion: iirc you mentioned something about issues in libffi 2015-01-07 12:55:32 do you remember what that was? 2015-01-07 12:56:20 not really 2015-01-07 12:56:48 enable support for PaX's EMUTRAMP (FS#42405) 2015-01-07 12:57:16 maybe that one: https://bugs.archlinux.org/task/42405 2015-01-07 12:57:19 ah, this one, not sure if I mentioned it here 2015-01-07 12:57:36 there was something you mentioned that made me hold the version bump 2015-01-07 12:57:45 https://projects.archlinux.org/svntogit/packages.git/commit/trunk?h=packages/libffi&id=d01f5c5c810cb69eedfbe3934f373d48fc9ad6c6 2015-01-07 12:57:50 maybe this? 2015-01-07 12:58:06 but it was merged in 3.2 2015-01-07 12:58:14 and 3.2.1 is out I guess 2015-01-07 12:58:18 yeah 2015-01-07 12:58:28 so I think there is no reason to hold it anymore 2015-01-07 12:59:02 ok 2015-01-07 12:59:08 i'll bump it and see what happens 2015-01-07 12:59:56 fabled: doesn't work for me. hangs on alpinelinux logo 2015-01-07 13:20:18 ScrumpyJack, it boots ~60 secs. and if fw blocks http, you don't see anything else 2015-01-07 13:20:24 no feedback on failure atm 2015-01-07 15:30:14 any thoughts on http://phabricator.org/ for redmine ? 2015-01-07 15:30:50 it has neat cli, http://phabricator.org/applications/arcanist 2015-01-07 16:42:38 it's great, but overcomplicated for us 2015-01-07 16:42:43 we don't need a platform for code review 2015-01-07 16:43:01 and we're fine hosting git repositories with the current approach 2015-01-07 16:43:31 we will probably move to gitolite if the team scaling becomes a problem, but there is no reason to think about it now 2015-01-08 07:13:53 morning 2015-01-08 07:14:00 yes, i have been thining of gitloite 2015-01-08 07:14:10 and i'd like to move to gitolite soonish 2015-01-08 08:26:18 this is kinda interesting: https://stribika.github.io/2015/01/04/secure-secure-shell.html 2015-01-08 08:26:38 the "System hardening" section more or less describes Alpine Linux :) 2015-01-08 08:52:06 morning 2015-01-08 08:52:50 moin 2015-01-08 09:03:38 morning 2015-01-08 09:03:54 ncopa: yeah, the article is very interesting :) 2015-01-08 09:06:57 I think we should harden sshd in alpine a bit more. ecdsa is still turned on. 2015-01-08 09:08:25 patch is welcome ;) 2015-01-08 09:59:45 morning 2015-01-08 10:01:20 ncopa: im having an issue with mpd. it segfaults on my alix x86 vanilla, but i cannot get a core dump. is there something im missing (i already set ulimit). 2015-01-08 10:04:50 clandmeter: do you have enough space for the dump? 2015-01-08 10:05:06 i think so 2015-01-08 10:05:24 shouldnt i get an error otherwise? 2015-01-08 10:06:35 i dont know... 2015-01-08 10:06:51 maybe you get a dump but in an unexpected dir? 2015-01-08 10:07:17 i remember i get a msg core dumped if it did. 2015-01-08 10:08:41 i did a find core to check, but couldnt find one. 2015-01-08 10:09:29 you can set the exact path in /proc/sys/kernel/core_pattern 2015-01-08 10:09:46 i think the default is ./core 2015-01-08 10:10:07 yeah thats what i thought 2015-01-08 10:10:20 maybe its because it changes user to mpd? 2015-01-08 10:10:30 could be 2015-01-08 10:10:44 you can set rc_ulimit="-c unlimited" in /etc/conf.d/mpd 2015-01-08 10:10:59 im running it from cmdline 2015-01-08 10:11:07 oh 2015-01-08 10:11:16 the user needs write permission in cwd 2015-01-08 10:11:32 check that too 2015-01-08 10:11:39 what alpine version is it? 2015-01-08 10:11:44 i can enable -dbg 2015-01-08 10:11:53 yeah, the user is the only thing i can think of currently. 2015-01-08 10:11:55 so we gan get a usefuc stackdump 2015-01-08 10:11:58 3.1.1 2015-01-08 10:12:00 ok 2015-01-08 10:12:04 I'll enable -dbg 2015-01-08 10:12:58 oh 2015-01-08 10:13:03 there is a .6 out 2015-01-08 10:13:07 0.19.6 2015-01-08 10:13:27 even 0.19.7 is out 2015-01-08 10:13:48 http://git.musicpd.org/cgit/master/mpd.git/plain/NEWS?h=v0.19.7 2015-01-08 10:13:54 does anything look related? 2015-01-08 10:14:15 what do we have as ffmpeg? 2015-01-08 10:14:22 cause the eror was on convert 2015-01-08 10:14:51 does not look like we have ffmpeg at all 2015-01-08 10:15:03 do you use nfs? 2015-01-08 10:15:19 I want to, but not yet configed 2015-01-08 10:15:23 ok 2015-01-08 10:15:41 do you nkow what file type it tried convert from? 2015-01-08 10:16:27 sounds like we need that -dbg 2015-01-08 10:18:55 its a web stream 2015-01-08 10:18:58 probably mp3 2015-01-08 10:19:23 ncopa, fcolista , can xapian be moved to main/ ? 2015-01-08 10:19:40 or still having issues ? 2015-01-08 10:23:27 so maybe bug is in libshout 2015-01-08 10:24:09 clandmeter: do you think you can add mpd@edge/main and mpd-dbg@edge/main for now? 2015-01-08 10:24:19 generate a dump 2015-01-08 10:24:50 send the dump to proper upstream - or lets have a look at it here 2015-01-08 10:25:00 you can post it on bugs.a.o too 2015-01-08 10:25:09 so we dont forget 2015-01-08 10:26:16 yes ill debug it. 2015-01-08 10:26:24 box is remote. 2015-01-08 10:37:35 bummer, not enough memory for gdb 2015-01-08 10:41:22 so it might be that you dont have enough memory for the dump 2015-01-08 10:44:11 i have the dump now 2015-01-08 10:44:37 dump issue because of user not allowed to dump 2015-01-08 10:44:44 i get this bt http://sprunge.us/BMab 2015-01-08 10:44:51 doesnt say much to me. 2015-01-08 10:51:05 ncopa: maybe this does say soemthing? http://sprunge.us/eeLW 2015-01-08 10:52:24 #0 0x210c253d in ?? () 2015-01-08 10:52:34 crash happens in unknown lib 2015-01-08 10:52:57 oh 2015-01-08 10:53:05 it could be stack size issue 2015-01-08 10:53:16 that is a common issue 2015-01-08 10:53:26 glibc has 8MB stack size by default 2015-01-08 10:53:34 musl libc has 80kb 2015-01-08 10:54:09 fabled: do you have any idea how we can detect stack overflows properly? 2015-01-08 10:54:21 ? 2015-01-08 10:54:30 mpd segfaults 2015-01-08 10:54:38 im running it fine on rpi 2015-01-08 10:54:43 i see "Thread..." 2015-01-08 10:54:54 how can i check if this is due to small stack? 2015-01-08 10:55:56 can you see what is the sigsegv subcode? 2015-01-08 10:56:27 how do you do that? 2015-01-08 10:56:40 its clandmeter who has the gdb and dump 2015-01-08 10:57:03 i'm not sure about if it's easily doable. sometimes the program prints it itself if it has sigsegv handler 2015-01-08 10:57:11 ok 2015-01-08 10:57:50 dmesg says decoder[1622]: segfault at 210c253d ip 210c253d sp b4f750f0 error 4 2015-01-08 10:58:28 can you run 'disassemble' on gdb 2015-01-08 10:58:45 sure 2015-01-08 10:59:00 and 'info registers' 2015-01-08 10:59:02 No function contains program counter for selected frame. 2015-01-08 11:00:05 http://sprunge.us/ieNM 2015-01-08 11:02:22 mmm... 'disassemble 0x210c2500,+0x80' 2015-01-08 11:03:53 http://sprunge.us/NMPJ 2015-01-08 11:05:24 huh 2015-01-08 11:05:39 mmm... 'disassemble 0x210c2528,+0x80' 2015-01-08 11:05:57 no need to paste if it's similar cannot access memory 2015-01-08 11:06:25 same 2015-01-08 11:06:27 ok 2015-01-08 11:06:30 sounds weird 2015-01-08 11:07:08 sounds like memory corruption, or just a programming bug 2015-01-08 12:14:49 clandmeter: can you reproduce it with 0.19.7 from edge? 2015-01-08 12:23:45 ncopa: yes 2015-01-08 12:34:51 clandmeter: can you please create an issue on bugs.a.o 2015-01-08 12:34:56 and post the backtrace there 2015-01-08 12:45:09 k 2015-01-08 12:52:13 ncopa: #3680 2015-01-08 17:09:32 oohmy. when was the last openssl update? https://www.openssl.org/news/secadv_20150108.txt 2015-01-08 17:09:46 ECDHE silently downgrades to ECDH [Client] (CVE-2014-3572) 2015-01-08 17:09:50 RSA silently downgrades to EXPORT_RSA [Client] (CVE-2015-0204) 2015-01-08 17:09:59 DH client certificates accepted without verification [Server] (CVE-2015-0205) 2015-01-08 17:10:06 Certificate fingerprints can be modified (CVE-2014-8275) 2015-01-08 17:10:10 Bignum squaring may produce incorrect results (CVE-2014-3570) 2015-01-08 17:10:22 OpenSSL 1.0.1 users should upgrade to 1.0.1k. 2015-01-08 18:13:04 here we go 2015-01-08 18:13:33 I can't test the build locally so brace yourselves 2015-01-08 18:22:33 sigh, need to go, I'll revert it for now and take care of it tommorow morning 2015-01-08 18:22:37 ncopa, fabled ↑ 2015-01-08 18:40:21 barthalion: did you suggest we do something back in December? 2015-01-08 20:39:08 ScrumpyJack: official docker + vagrant image, possibly with packer for bunch of other formats 2015-01-09 07:30:04 barthalion: ah yes, i remember 2015-01-09 08:41:24 ncopa: im still getting a symbol error on gst10 python bindings. ImportError: Error relocating /usr/lib/python2.7/site-packages/gst-0.10/gst/_gst.so: libxml_xmlDocPtrWrap: symbol not found 2015-01-09 12:39:43 ncopa: any news about mariadb? 2015-01-09 12:43:17 not had time yet sorry 2015-01-09 12:43:43 okay 2015-01-09 13:06:32 ncopa, i'm going to move xapian-* in main. vkrishn is asking for it since a while, and he tested it. 2015-01-09 13:07:27 ok 2015-01-09 13:07:29 sounds good 2015-01-09 13:08:11 ncopa, thanks for cherry-picking the openssl to -stables 2015-01-09 13:09:00 fabled: thanks for fixing it 2015-01-09 13:09:02 :) 2015-01-09 13:09:16 i also filed bugs for it on bugs.a.o 2015-01-09 13:09:50 the thing is that people thinks that good ssecurity = have lists and sec officers that keeps track and informs everyone about all security fixes 2015-01-09 13:10:35 people who wants a security oriented distro will go there and look for: 2015-01-09 13:10:43 1) do they have a big community? 2015-01-09 13:10:54 2) where are the security mailing list? 2015-01-09 13:11:04 3) how are scurity issues communicated? 2015-01-09 13:11:14 and they find nothing on alpine linux 2015-01-09 13:11:22 and think that alpine linux has bad security 2015-01-09 13:11:34 community/userbase size is bogus claim 2015-01-09 13:11:43 nod 2015-01-09 13:11:44 Debian is one of the biggest and had major openssl flaw due to local patch 2015-01-09 13:11:44 /nick alacerda 2015-01-09 13:12:05 much bogus indeed 2015-01-09 13:12:16 much perception 2015-01-09 13:12:37 they also think that "practical" security is all about finding and fixing security bugs 2015-01-09 13:13:01 and simply dont understand the prevent exploits approach 2015-01-09 13:13:26 https://github.com/nylira/prism-break/issues/1189 2015-01-09 13:14:05 Default installs of Debian stable for example can be considered quite secure. 2015-01-09 13:14:08 WAT 2015-01-09 13:14:32 depends on what you cmpare with 2015-01-09 13:14:40 i kinda regret that i even opened my mouth 2015-01-09 13:15:00 and i dont know if we want be on their list at all :) 2015-01-09 13:15:08 https://github.com/nylira/prism-break/issues/1189#issuecomment-68507022 2015-01-09 13:15:24 that is a good point though 2015-01-09 13:15:34 gave me a hint what we can do better 2015-01-09 13:15:41 and i think we should try do that better 2015-01-09 13:15:58 regardless if most of their claims are bogus or not 2015-01-09 13:16:22 so i wonder how do we improve? 2015-01-09 13:16:32 i think we could: 2015-01-09 13:16:51 1) try file a bug for every cve that affects us (maybe even if its already fixed) 2015-01-09 13:17:24 2) whenver bug is closed, we automatically send the info in the bug in email to a sec mailing list 2015-01-09 13:17:51 that would give us all the issues in a mailing list archive 2015-01-09 13:18:13 i think we already have an email address where youcan report sec issues 2015-01-09 13:18:18 but its not visible anywhere 2015-01-09 13:18:22 that we can fix also 2015-01-09 13:18:37 i'm open to other ideas 2015-01-09 13:18:51 but i really really really want keep admin work at minimum 2015-01-09 13:18:58 minimal extra manual work 2015-01-09 13:19:07 preferrible zero extra manual work 2015-01-09 13:20:04 funny to hear from gentoo dev about security, when they still don't sign portage tree 2015-01-09 13:20:13 or at least don't enable verification by default 2015-01-09 13:22:05 i really dont care what others do 2015-01-09 13:22:23 i have enought with caring about alpine :) 2015-01-09 13:24:15 when talking of improving alpine 2015-01-09 13:24:17 https://github.com/nylira/prism-break/issues/1189#issuecomment-69180242 2015-01-09 13:24:26 we could evaluate some of the points he lists 2015-01-09 13:24:49 sure, some 2015-01-09 13:24:53 but in overall, it's a FUD 2015-01-09 13:25:01 *nod* 2015-01-09 13:25:06 and many bogus claims 2015-01-09 13:25:18 and they totally miss the big imporant bits 2015-01-09 13:25:24 like what PaX do 2015-01-09 13:25:33 there are maybe 3 points that make any sense 2015-01-09 13:25:46 the rest is more like "we do it so we must be right" 2015-01-09 13:25:50 i actually believe that the freeradius bug mentioned there does not even segfault on glibc 2015-01-09 13:26:07 which means glibc just runs the vulnerable code just happy 2015-01-09 13:26:26 with some nasty memory corrution going on unnoticed 2015-01-09 13:26:58 end users will notice the difference: on ubuntu runs. on alpine it crashes. 2015-01-09 13:27:07 and ofcourse draw the wrong conclusions 2015-01-09 13:27:35 thanks to the crash early we found a bug 2015-01-09 13:27:38 a real bug 2015-01-09 13:27:45 that likely affects everyone 2015-01-09 13:27:50 we got it fixed 2015-01-09 13:28:01 the big-user-base-distros didnt even notice it 2015-01-09 13:29:06 and thanks to upstream does not care (or think its embarrassing) they probably will continu run their vulnerable code 2015-01-10 16:22:38 afternoon 2015-01-10 16:23:25 lxc-create with a choice of backing store fails when creating the filesystem 2015-01-10 16:25:30 strace shows dbev.c looking for mkfs, which doesn't come with e2fsprogs 2015-01-10 16:26:20 mke2fs comes with that 2015-01-10 16:28:23 ls -s /sbin/mke2fs /bin/mkfs fixed that. perhaps the symlink should be in the e2fsprogs? 2015-01-10 18:30:51 but then it wouldn't work for non-ext{2,3,4} filesystems? 2015-01-11 09:35:28 dne: mkfs.vfat and mkfs.ext3 etc are hardlinks to mke2fs 2015-01-11 12:18:33 ScrumpyJack: mkfs is part of util-linux, maybe the lxc package should depend on that? 2015-01-11 16:59:40 dne: it's possible that someone would want to run lxc containers without needing to create a new filesystem, so perhaps not add the dependancy 2015-01-12 00:34:08 scadu: do you know sth. about mariadb and alpine? 2015-01-12 07:01:55 well, we want to switch 2015-01-12 07:02:11 since 2.6 or 2.7, just nobody cared enough to make it/there are important tasks 2015-01-12 07:02:55 jomat: ↑ 2015-01-12 07:47:58 morning 2015-01-12 07:49:45 barthalion: took a look at docker last night and it seems that ncopa has put a docker image in their repository alpinelinux/base 2015-01-12 07:49:48 https://registry.hub.docker.com/u/alpinelinux/base/ 2015-01-12 07:54:16 I don't think it's ncopa 2015-01-12 07:54:32 ncopa: ↑ is it you? 2015-01-12 07:54:45 jomat: why do you ask? if ncopa have time, we'll start a migration, I hope. 2015-01-12 07:55:54 ScrumpyJack: I hardly think that ncopa would 1) add a repo on bitbucket 2) not use "ncopa" username for that 2015-01-12 07:56:40 and additionally make an image with bash, cmon, smells fishy 2015-01-12 07:56:45 I just assumed that it was his as the dockerfile has his name in it 2015-01-12 07:57:21 indeed 2015-01-12 07:57:42 still weird, and I still don't like the way this image is created 2015-01-12 07:58:00 we can do better than extracting premade rootfs 2015-01-12 08:00:49 fair point 2015-01-12 08:03:21 morning 2015-01-12 08:03:24 what have i done? 2015-01-12 08:04:18 i got brand new cat 6 cobber here over the weekend 2015-01-12 08:04:31 my irc feels much more snappy now! 2015-01-12 08:04:47 and it makes me happy that the new cables are blue 2015-01-12 08:05:00 instead of the previous old, boring grey 2015-01-12 08:05:34 i cannot see the cables of course, but it makes me feel good knowing the cables are blue :) 2015-01-12 08:06:16 so whats this ncopa bitbucket business about? 2015-01-12 08:07:27 all our cat6 is beige 8-( 2015-01-12 08:07:41 that still better than grey 2015-01-12 08:11:02 ncopa: well, just asking if it's you 2015-01-12 08:12:27 if its me what? 2015-01-12 08:12:48 https://registry.hub.docker.com/u/alpinelinux/base/ 2015-01-12 08:13:12 no, thats not me 2015-01-12 08:13:19 thought so 2015-01-12 08:13:28 we should request removal of it somehow 2015-01-12 08:13:33 ok? 2015-01-12 08:13:40 same goes for bitbucket repo where it's hosted 2015-01-12 08:13:55 or renamed? 2015-01-12 08:13:58 yeah 2015-01-12 08:14:32 there are official ubuntu and debian repos 2015-01-12 08:14:37 https://registry.hub.docker.com/search?q=library 2015-01-12 08:16:03 https://registry.hub.docker.com/search?q=alpine&searchfield= 2015-01-12 08:17:15 https://bitbucket.org/alpinelinux/base/commits/edcf4bb328beb6d83dd380bbfffec3a95027c62a/raw/ 2015-01-12 08:17:18 hum 2015-01-12 08:17:19 ok 2015-01-12 08:17:23 i see the problem 2015-01-12 08:17:25 i didnt do that commit 2015-01-12 08:19:48 does anyone know who did it? 2015-01-12 08:19:56 no 2015-01-12 08:20:04 it's not uggedal, he hosts his image under his name 2015-01-12 08:20:08 yes 2015-01-12 08:20:18 and he uses his own name in his git commits 2015-01-12 08:22:59 seems you joined #bitbucket already, I'll leave it for you 2015-01-12 08:23:05 I'll handle docker hub 2015-01-12 08:23:09 thanks 2015-01-12 08:24:20 i can send signed emails if you need confirmation that the bitbucket repo is untrusted 2015-01-12 08:25:56 I guess sending it to info@docker.com will do 2015-01-12 08:26:04 about that image I mean 2015-01-12 08:27:30 someone on #docker suggested support@docker.com 2015-01-12 08:27:43 it might have more sense actually 2015-01-12 08:39:53 ncopa, you moved to Amsterdam, https://bitbucket.org/alpinelinux ;) 2015-01-12 08:40:54 barthalion, update calcurse pls, 2 new versions 2015-01-12 08:40:59 vkrishn: thats not me 2015-01-12 08:41:08 I know ;) 2015-01-12 08:41:50 as long they are not signing/committing with your email, it should be ok 2015-01-12 08:42:47 and they are using old AL logo 2015-01-12 08:44:36 they are using our name, it's enough to get rid of it 2015-01-12 08:46:52 they are committing in my name 2015-01-12 08:47:20 author and committer is my name and email 2015-01-12 08:47:22 i didnt do it 2015-01-12 08:47:41 thats enough to make you suspicious of their intentions 2015-01-12 08:49:14 I think bitbucket.org has nice management working and should be helpful 2015-01-12 08:55:26 ncopa, if you need a email writeup to be send, I can draft one 2015-01-12 09:11:42 Do you know what's the way to compile a package with a different version of gcc? E.g. OpenVAS requires gcc 4.3. What kind of problmes could arise (if gcc support this king of legacy compilation)? 2015-01-12 09:12:10 vkrishn: could you please write a email draft for docker? 2015-01-12 09:12:16 i sent to bitbucket 2015-01-12 09:13:50 ncopa, receive attachment 2015-01-12 09:14:28 its for bitbucket, but replace it with docker.com 2015-01-12 09:23:43 fcolista: I'm sure you can build it with modern gcc 2015-01-12 09:23:58 fcolista: we have it in Arch, and ship only the newest toolchain 2015-01-12 09:24:26 arh does not use musl :) 2015-01-12 09:24:28 *Arch 2015-01-12 09:25:36 fcolista: i think he means that Arch has openvas build with latest gcc 2015-01-12 09:26:07 we should try make it build with latest gcc too, even it upstream only support gcc-4.3 2015-01-12 09:26:18 yes, that's what I mean 2015-01-12 09:26:27 ok, sorry for misunderstanding barthalion 2015-01-12 09:26:41 old gcc won't solve code portability issues 2015-01-12 09:26:57 any volunteers for analyzing the suspicious alpinelinux docker image? 2015-01-12 09:27:16 it is potentially evil 2015-01-12 09:29:13 it doesn't look malicious at the first glance 2015-01-12 09:29:39 it doesnt 2015-01-12 09:29:52 it could be a misunderstanding 2015-01-12 09:30:24 but it could also be a carefully planned malicious thing 2015-01-12 09:44:43 ok i have sent email to support@bitbucket.com and security@docker.com 2015-01-12 09:44:44 thanks! 2015-01-12 09:46:46 https://bitbucket.org/alpinelinux/base/src 2015-01-12 09:46:56 there is a root.tar.xz image checked in with binaries 2015-01-12 09:47:27 i'd like to verify those 2015-01-12 09:50:40 haven't seen anything out of place, but didn't check files integrity against the repo 2015-01-12 09:51:06 check if keys in there are same as official keys 2015-01-12 09:51:17 i'm looking at the image too now 2015-01-12 09:51:49 i will run apk audit after verifying that keys are not modified 2015-01-12 09:53:06 upgraded calcurse: https://github.com/scadu/aports/commit/e4c4cbc85d4e0f79cf87c768e0bee56cfb8b4f7f 2015-01-12 09:54:12 scadu: set me as a contributor and leave maintainer as empty, please 2015-01-12 10:01:23 barthalion: done 2015-01-12 10:02:17 and squash your changes… 2015-01-12 10:02:27 it will make things easier and git history cleaner 2015-01-12 10:04:05 wow, that docker thing is really bad 2015-01-12 10:05:10 why impersonate alpine/ncopa if you don't have nefarious intentions 2015-01-12 10:05:59 uggedal: exactly 2015-01-12 10:06:06 fortunally 2015-01-12 10:06:16 we sign all binary packages 2015-01-12 10:06:50 whats makes it more higly suspicious is that the bitbucket source repo has a tarball with binaries 2015-01-12 10:07:03 we are analyzing those now 2015-01-12 10:07:12 well, my docker image also used a checked in tarball 2015-01-12 10:07:59 it makes it easier to pull of an attack like this 2015-01-12 10:08:18 ship a tarball that looks legitimate, modify one of the binaries 2015-01-12 10:09:02 i have requested both the docker image and the bitbucket repository to be removed 2015-01-12 10:09:24 so if anyone wants analyze the binaries, you should download it now 2015-01-12 10:09:37 what also is a bit bad is that there are 400+ that has downloaded it 2015-01-12 10:11:59 barthalion: sorry for that. fixed already ;) 2015-01-12 10:16:15 scadu: thanks 2015-01-12 10:25:54 ok, the only difference i found in the docker image and trusted binaries are in libcrypto 2015-01-12 10:26:07 that is likely due to the recent openssl security fix 2015-01-12 10:32:11 I think as long as commits are --reset hard and re-recomitted with their name/ids and no-use of log (or get permission) or any keywords that may confuse user of its origin, then should be ok 2015-01-12 10:32:23 logo* 2015-01-12 10:38:00 hmm... does --reset hard leave dangling objs ? 2015-01-12 11:11:45 :wq 2015-01-12 11:11:49 oops... 2015-01-12 11:24:12 gotta go 2015-01-12 14:02:05 stable kodi 14.0 has been released which makes xbmc unmaintained. the question is somebody use it on alpine. if yes, I'll prepare an APKBUILD. 2015-01-12 14:09:10 scadu: it would be cool to have kodi - i'd like to think i could make a pvr one day :) 2015-01-12 14:09:56 tvh works nicely on alpine 2015-01-12 14:10:44 scadu: makes sence to move do it. 2015-01-12 14:11:01 just update download url and pkgname, it will build 2015-01-12 14:11:10 drop xbmc completely 2015-01-12 14:11:26 barthalion: by moving to unmaintained? 2015-01-12 14:12:23 by removing it 2015-01-12 14:12:34 what's the point of keeping it? kodi should provide/replace xbmc 2015-01-12 14:12:43 agree 2015-01-12 14:12:47 just rename it 2015-01-12 14:14:21 probably add replaces=xbmc 2015-01-12 14:15:22 yeah, right. 2015-01-12 14:16:30 it doesn't look good either 2015-01-12 14:16:37 does kodi provide the same files as xbmc? 2015-01-12 14:16:43 given the rename, I doubt it 2015-01-12 14:17:12 so unless $replaces behavior changed, it gives no profits 2015-01-12 14:19:34 does $replaces implies that users will get seamless xbmc → kodi transition? 2015-01-12 14:20:04 and we can be sure it won't be so straightforward, unless we provide compatibility symlinks for binaries and other directories 2015-01-12 14:20:52 right, makes no sence. i guess they name changed everything. 2015-01-12 14:23:17 so let's just remove xbmc, package kodi, and provide the info in release notes for 3.2 2015-01-12 14:23:27 +1 2015-01-12 14:24:21 seems legit. 2015-01-12 14:27:48 by the way, we could be writing release notes on some etherpad 2015-01-12 14:27:51 or anything similiar 2015-01-12 14:28:05 so we don't forget anything on the freeze 2015-01-12 14:28:18 would be nice yes 2015-01-12 14:28:50 we do that in git releasenotes already? or we use wiki? 2015-01-12 14:29:12 git releasenotes? 2015-01-12 14:29:45 i mean, we could strat write the release notes already now 2015-01-12 14:29:52 we jsut dont publish it on front page yet 2015-01-12 14:30:06 right 2015-01-12 14:31:13 i suppose wiki.alpinelinux.org is the simplest atm 2015-01-12 14:39:24 https://github.com/scadu/aports/commits/master I've upgraded few packages. 2015-01-12 14:45:33 scadu: merged, thanks 2015-01-12 14:46:06 barthalion: you're welcome 2015-01-12 15:10:20 ncopa: is it possible to see mariadb in 3.2? 2015-01-12 15:11:56 scadu: i hope so yes 2015-01-12 15:12:11 that and python3 2015-01-12 15:12:13 would be nice 2015-01-12 15:13:12 do you know if mariadb-10.x are compatible with mysql-5.5? 2015-01-12 15:13:47 no 2015-01-12 15:13:50 well, ABI is kept 2015-01-12 15:13:51 ncopa: nope. that's why I wrote APKBUILD for 5.5. 2015-01-12 15:14:00 but manual intervention is required 2015-01-12 15:14:09 but mariadb-10 is compatible with mariadb-5.5? 2015-01-12 15:14:17 libraries are 2015-01-12 15:14:23 but not the server itself 2015-01-12 15:14:27 ok 2015-01-12 15:14:37 you can't use mariadb 10.0 with 5.5 databases 2015-01-12 15:14:53 so you need dump/restore 2015-01-12 15:15:11 or restart mysqld and run mysql_upgrade 2015-01-12 15:15:12 we need have that in releasenotes too then 2015-01-12 15:15:24 this is the route we went in Arch 2015-01-12 15:15:46 but basically yes, the best way is to dump&restore 2015-01-12 15:15:58 ncopa: so we'll migrate to 10.x, not 5.x? 2015-01-12 15:16:22 let's switch to 5.5 for the beginning 2015-01-12 15:16:44 yeah, that'd wiser solution. 2015-01-12 15:21:37 we want to go to 10.x eventually 2015-01-12 15:21:47 but i dont know which is the best way to get there 2015-01-12 15:21:55 mariadb-5.5 sounds good to me 2015-01-12 15:22:07 me neither, thus I suggested 5.5 to scadu 2015-01-12 15:22:57 anyways, I'm going home, see you tommorow 2015-01-12 15:23:14 bb 2015-01-12 15:26:23 ncopa: my APKBUILD still fails to build for you? 2015-01-12 15:27:51 i didn't try again, been busy with other stuff lately 2015-01-12 15:34:11 moreover, we have that mess with docker images. sorry for being intrusive. 2015-01-12 15:34:25 np 2015-01-12 15:49:19 i'll try look at mariadb tm 2015-01-12 15:49:26 what do we do with mysl pkg? 2015-01-12 15:49:29 i just remove it? 2015-01-12 15:50:26 if mariadb-5.5 libmysql is ABI compatible, then we shouldnt need rebuild everything linked with libmysqlclient 2015-01-12 15:51:43 there's no sense to keep mysql. 2015-01-12 15:53:44 thats what i think too 2015-01-12 18:08:26 http://www.lua.org/versions.html#5.3 2015-01-12 19:13:09 alacerda: didn't realize the official release of 5.3 was today ^^^ 2015-01-12 19:13:19 but, I knew it was coming 2015-01-12 19:13:28 part of the reason for trying to drop Lua 5.1 2015-01-12 19:13:39 new table. functions 2015-01-12 19:13:40 ;) 2015-01-12 19:18:12 not nearly as many incompatibilities as when 5.2 was introduced 2015-01-12 19:18:29 maybe I should push for 5.3 to be included in AL 3.2 2015-01-12 19:55:36 ncopa: there seems to be huge logging problems with musl 2015-01-12 19:56:01 especially with postfix 2015-01-12 19:56:28 so we'd like to ask for a new 2.7 version! 2015-01-12 19:56:38 Postfix 2.11.3 2015-01-12 19:56:50 and by the way, ClamAV could be updated too 2015-01-12 19:57:02 to 0.98.5 2015-01-13 06:54:50 morning 2015-01-13 06:55:01 StarWarsFan: huge loggin problems? 2015-01-13 06:59:33 StarWarsFan: could you please create a ticket on bugs.a.o with the details 2015-01-13 06:59:47 i would like to fix the logging problems in v3.1 if possible 2015-01-13 07:00:14 hi 2015-01-13 07:00:24 great! 2015-01-13 07:00:29 ok, will create a ticket 2015-01-13 07:00:44 v2.7 has postfix-2.10 2015-01-13 07:00:57 i would prefer not to upgrade to postfix-2.11 in a stable branch 2015-01-13 07:02:23 i could upgrade postfix to 2.10.5 though 2015-01-13 07:03:02 it would be nice with a ticket though, so I in the commit message can document why and when the upgrade was requested 2015-01-13 07:03:13 i c 2015-01-13 07:03:14 and by who 2015-01-13 07:06:16 v3.0, v2.7 and v2.6 are in "security fixes only" maintenance mode 2015-01-13 07:06:30 but i can do bugfix updates on request too 2015-01-13 07:06:35 right, but 3.1 is not usable in production 2015-01-13 07:06:45 would be nice to fix it :) 2015-01-13 07:06:52 because of these log problems on musl 2015-01-13 07:07:21 the UTC timestamps in logs? 2015-01-13 07:07:38 no, that's another thing 2015-01-13 07:07:42 ok? 2015-01-13 07:07:46 in short: uclibc uses buffer size of 1024 for the log entries 2015-01-13 07:07:52 musl only 256 2015-01-13 07:08:09 my collegue has a patch for that 2015-01-13 07:08:13 ok? i didnt know 2015-01-13 07:08:18 is it reported to musl devs? 2015-01-13 07:08:33 so especially postfix writes long log messages with the important data at the end of each line 2015-01-13 07:08:40 and these content is lost on musl 2015-01-13 07:08:40 understand 2015-01-13 07:08:44 i didnt know that was a problem 2015-01-13 07:09:12 it would be very very nice if you could post the details in a ticket on bugs 2015-01-13 07:09:20 i'm on it 2015-01-13 07:09:20 then i can have the musl devs to look at that 2015-01-13 07:09:34 we have a very good relation with musl devs 2015-01-13 07:09:40 have to translate to english before ;) 2015-01-13 07:09:47 a cool! good to know 2015-01-13 07:10:15 I am sure they will be interested in a fix 2015-01-13 07:51:27 ncopa can we have an update iso from edge ? http://nl.alpinelinux.org/alpine/edge/releases/x86_64/ 2015-01-13 08:04:42 crow, i want a fix for busybox modprobe first 2015-01-13 08:21:06 ncopa i see ok :) 2015-01-13 08:22:34 i have a workaround already that we use but i want a fix from upstream 2015-01-13 08:42:48 ok there is a parallel build issue in kamailio 2015-01-13 08:55:54 ncopa: 2015-01-13 08:55:56 http://bugs.alpinelinux.org/issues/3692 2015-01-13 08:55:58 http://bugs.alpinelinux.org/issues/3693 2015-01-13 08:55:59 http://bugs.alpinelinux.org/issues/3694 2015-01-13 08:56:05 [x] done :-) 2015-01-13 08:56:22 StarWarsFan: thanks! 2015-01-13 08:56:27 i will try fix asap 2015-01-13 08:56:36 already contacted #musl 2015-01-13 09:31:10 ncopa, would be better if http://www.pell.portland.or.us/~orc/Code/discount/ this was seperate pkg and mksite called the bin/markdown ? 2015-01-13 09:35:27 we embed a yaml header 2015-01-13 09:35:33 how does it handle that? 2015-01-13 09:36:16 I compiled and it has mkd2html, markdown bins 2015-01-13 09:36:58 I also emailed http://asbradbury.org author to update, but no reply yet 2015-01-13 09:37:47 lua-discount-1.2.10.1 has bug, that seems fixed in http://www.pell.portland.or.us/~orc/Code/discount/ 2015-01-13 10:20:45 vkrishn: there was some trademark issues with markdown 2015-01-13 10:20:58 i think thy wanted createa common standard 2015-01-13 10:21:08 commonmark 2015-01-13 10:21:19 http://commonmark.org/ 2015-01-13 10:21:25 can we switch to commonmark ? 2015-01-13 10:21:30 i think we can 2015-01-13 10:21:38 nice ! 2015-01-13 10:21:47 i think thats what e'd want 2015-01-13 10:22:39 and try stick to text/markups that would parse without issues 2015-01-13 10:22:57 till proper commonmark tools are available 2015-01-13 10:24:04 https://github.com/jgm/CommonMark 2015-01-13 10:24:56 nice :-) 2015-01-13 10:25:45 would need to carefully check "Differences from original Markdown" section 2015-01-13 10:28:13 im reading 2015-01-13 10:28:27 to me the differences makes sense so far 2015-01-13 10:31:09 is there any discussion on markdown's trademark issues ? 2015-01-13 10:31:45 i saw some article about it 2015-01-13 10:31:54 they wanted to standardize markdown 2015-01-13 10:31:56 I too recall vaguely 2015-01-13 10:32:12 have a spec that descibed how implementations should deal with cornercases 2015-01-13 10:32:20 i think they even asked the original author 2015-01-13 10:32:29 got no response 2015-01-13 10:32:31 they moved ahead 2015-01-13 10:32:44 and suddenly toe original author woke up and was angry 2015-01-13 10:32:52 so they just renamed it to commonmark 2015-01-13 10:34:17 I think that would go for any 'named' markup if you try add something new and remove something to/from it 2015-01-13 10:34:38 similar sparks would come 2015-01-13 10:34:53 yeah 2015-01-13 10:35:58 if CommonMark distorts the way we have come to like markdown and its variants, then it needs be jointly evaluated 2015-01-13 10:36:36 other wise it just a strong assertion trying to fadeOff something old 2015-01-13 10:40:39 http://www.irclogger.com/.alpine-linux/2014-12-11#1418312378 2015-01-13 10:41:33 yes 2015-01-13 10:41:40 i tend to agree with that 2015-01-13 10:41:58 but commonmark is not really intended for complex technical docs 2015-01-13 10:42:11 http://spec.commonmark.org/dingus.html 2015-01-13 10:42:37 i think the reference implementation already handles the inline html better than current lua-discount for our alpinelinux.org site 2015-01-13 10:45:22 we still don't have nice and lengthy quickstart/tech docs/pdf for Al, but more popular it get, need would be more 2015-01-13 10:45:43 i'd say we use asciidoc for that 2015-01-13 10:46:05 ok 2015-01-13 11:19:17 ncopa: how are you getting on with tmux xen console integration. what that just for you? :) 2015-01-13 11:59:11 download link on http://wiki.alpinelinux.org/wiki/Create_Alpine_Linux_PV_DomU is broken 2015-01-13 12:12:41 ScrumpyJack: tmux xen console should work with v3.1 2015-01-13 12:13:17 set XENDOMAINS_CONSOLE=tmux in /etc/conf.d/xendomains 2015-01-13 12:19:37 ScrumpyJack: the download link should be fixed after next sync (in 40mins) 2015-01-13 12:45:22 ScrumpyJack: download link should be ok now 2015-01-13 12:48:04 thanks 2015-01-13 12:48:09 and thanks :) 2015-01-13 12:50:18 ncopa: the mirror i've been running is looking good. can we add it to the list? shall i submit a patch to alpine-mirrors or do you want to add it? 2015-01-13 12:51:06 ScrumpyJack: which mirror is it? 2015-01-13 12:51:14 the url? 2015-01-13 12:51:17 yes 2015-01-13 12:51:25 liskamm.alpinemirror.uk 2015-01-13 12:51:34 name, location, urls 2015-01-13 12:51:57 liskamm.alpinemirror.uk does not open for me 2015-01-13 12:52:05 my typo 2015-01-13 12:52:10 liskamm.alpinelinux.uk 2015-01-13 12:54:06 name? localtion? bandwidth? 2015-01-13 12:54:40 name = liskamm, location = france, bandwidth = 100, url = http://liskamm.alpinemirror.uk 2015-01-13 12:54:52 duh! 2015-01-13 12:55:09 url = http://liskamm.alpinelinux.uk 2015-01-13 12:55:16 :) 2015-01-13 12:55:33 maybe i'll register alpinelinux.fr for that mirror 2015-01-13 12:56:12 i suppose I could point dl-6.alpinelinux.org to it too 2015-01-13 12:56:19 i'm looking for a uk host but they are all quite expensive 2015-01-13 12:56:52 this one is in northern france, kimsufi :) 2015-01-13 12:59:46 like this? http://sprunge.us/UVFY 2015-01-13 13:03:19 yes. do you think other methods may be useful? like rsync and https? 2015-01-13 13:04:41 (i would love to see mqtt in the design - but i guess that's not a priority at the moment) 2015-01-13 13:08:10 i have an mqtt triggered mirror 2015-01-13 13:08:14 for testing 2015-01-13 13:08:22 oooh! 2015-01-13 13:08:23 the current problem is that i want flock -w 2015-01-13 13:08:26 can i try? 2015-01-13 13:08:31 sure 2015-01-13 13:11:33 http://wiki.alpinelinux.org/wiki/User:Ncopa/Alpine_Mirror_with_MQTT 2015-01-13 13:11:39 apk add mqtt-exec 2015-01-13 13:12:03 ln -s mqtt-exec /etc/init.d/mqtt-exec.alpine-mirror 2015-01-13 13:12:37 there is a problem though 2015-01-13 13:12:44 the flock might hang forever 2015-01-13 13:13:16 so i'd like to have a flock -w $timeout ... 2015-01-13 13:22:07 i which conditions would the flock hang? 2015-01-13 13:23:55 if the lockfile already exists? 2015-01-13 13:37:30 ncopa: I was looking at posixtz, and seems that the post-receive hook is broken on git.alpinelinux.org 2015-01-13 13:37:49 also, do I have your permission to push fixes to posixtz.lua and bump release? 2015-01-13 13:38:13 (post-receive hook seems to be ownership issue) 2015-01-14 07:19:03 morning 2015-01-14 07:21:17 tdtrask: please feel free to take over posixtz if you still want use it 2015-01-14 07:21:25 i have no need for it with musl libc 2015-01-14 07:21:56 i changed the group permissions on posixtz.git so you should be able to push to posixtz.git 2015-01-14 07:22:17 tdtrask: re git hook, what permission problem do you have? 2015-01-14 07:24:22 StarWarsFan: re musl bugs you filed. is it enough to fix it in v3.1? or do you want me to fix it in v3.0 too? 2015-01-14 08:42:16 morning@all 2015-01-14 08:42:43 ncopa: 3.1 would be fine, no need for 3.0 2015-01-14 08:56:06 perfect, thanks 2015-01-14 10:32:36 cool Llua 5.3 released 2015-01-14 10:32:39 Lua* 2015-01-14 14:20:38 ncopa: I believe the permissions issue prevented the git hook from creating the archive tar.bz 2015-01-14 14:21:30 ncopa: what were you using posixtz for? Why did that change with musl? 2015-01-14 14:21:58 I ask, because maybe I can drop it and do my work another way 2015-01-14 14:26:13 it was used to generate the /etc/TZ for uclibc 2015-01-14 14:26:51 during setup-alpine it would use the tzdata to convert a posix tz string 2015-01-14 14:27:54 with musl we just copy the tzdata file iirc 2015-01-15 08:57:55 just wanted to say, the installation method in alpine is the best i've ever used 2015-01-15 08:58:34 much better than regular ncurses/graphical installation interface things 2015-01-15 08:59:05 Tsutsukakushi: thanks :) 2015-01-15 08:59:13 i am missing a way to move backwards 2015-01-15 08:59:21 change things afterwards 2015-01-15 08:59:32 any ideas how that could be implemented? 2015-01-15 08:59:33 an option for using linux-libre would be cool 2015-01-15 08:59:45 C-c 2015-01-15 08:59:50 and start it again 2015-01-15 08:59:56 that's what i did 2015-01-15 09:00:10 it was the most logical installation script i've seen 2015-01-15 09:00:24 inspired from openbsd installer 2015-01-15 09:00:38 i have tried hard to reduce number of questions to a minimum 2015-01-15 09:01:21 how about more advanced installer script that asks some extra questions? 2015-01-15 09:01:52 it has support for some advanced features 2015-01-15 09:01:58 like lvm or raid support 2015-01-15 09:02:09 you can run setup-disk afterwards 2015-01-15 09:05:15 supporting grsec+linux-libre would require you roll your own kernels tho 2015-01-15 09:11:41 ncopa: you should email the busybox guys to add alpine to http://www.busybox.net/products.html 2015-01-15 09:15:44 indeed :) 2015-01-15 10:08:31 ncopa: good morning. could you check that my rlog patch sent to alpine-aports works? i've changed my mail client's behaviour 2015-01-15 10:12:34 ScrumpyJack: rlog? 2015-01-15 10:13:05 yes please 2015-01-15 10:13:20 foudn it 2015-01-15 10:13:49 Applying: testing/rlog: new aport 2015-01-15 10:13:50 /home/ncopa/aports/.git/rebase-apply/patch:17: trailing whitespace. 2015-01-15 10:13:50 pkgdesc="A flexible message logging facility for C++ programs and 2015-01-15 10:13:50 fatal: corrupt patch at line 18 2015-01-15 10:13:50 Patch failed at 0001 testing/rlog: new aport 2015-01-15 10:14:00 your mailer wraps the lines 2015-01-15 10:14:09 i thought i'd fixed that 2015-01-15 10:14:16 thanks 2015-01-15 10:17:13 https://www.kernel.org/doc/Documentation/email-clients.txt 2015-01-15 10:17:24 Alpine (TUI) 2015-01-15 10:17:24 Config options: 2015-01-15 10:17:24 In the "Sending Preferences" section: 2015-01-15 10:17:24 - "Do Not Send Flowed Text" must be enabled 2015-01-15 10:17:24 - "Strip Whitespace Before Sending" must be disabled 2015-01-15 10:17:24 When composing the message, the cursor should be placed where the patch 2015-01-15 10:17:26 should appear, and then pressing CTRL-R let you specify the patch file 2015-01-15 10:17:28 to insert into the message. 2015-01-15 10:19:57 lol 2015-01-15 10:20:02 Lotus Notes (GUI) 2015-01-15 10:20:03 Run away from it. 2015-01-15 10:21:20 ScrumpyJack: i manually removed a couple of newlines and it finally applied 2015-01-15 10:23:09 ncopa: thank you. i'm trying to fix the mailer with the maintainer 2015-01-15 10:23:44 it looks like it wrapped line 18 2015-01-15 10:23:45 +pkgdesc="A flexible message logging facility for C++ programs and 2015-01-15 10:23:46 libraries." 2015-01-15 10:23:48 :( 2015-01-15 10:24:48 yup 2015-01-15 10:24:55 i manually removed that 2015-01-15 10:25:01 and in checksums too 2015-01-15 10:25:05 ok 2015-01-15 10:25:38 I'll sprunge future patches until I can get mailer working 2015-01-15 10:26:11 ScrumpyJack: there is a seciton here: https://www.kernel.org/doc/Documentation/email-clients.txt 2015-01-15 10:26:20 that tells how to configure Alpine 2015-01-15 10:26:40 - "Do Not Send Flowed Text" must be enabled 2015-01-15 10:26:40 - "Strip Whitespace Before Sending" must be disabled 2015-01-15 10:26:49 i'm pretty sure that will solve it for you 2015-01-15 10:34:07 scadu: interestingly, mariadb does build on x86 but it fails on x86_64 2015-01-15 10:36:25 jt should work with default settings. 2015-01-15 10:36:27 However, it's a good idea to set the "send_charset" to: 2015-01-15 10:36:29 set send_charset="us-ascii:utf-8" 2015-01-15 10:36:37 oh 2015-01-15 10:36:40 sorry 2015-01-15 10:36:45 accidental paste 2015-01-15 10:43:32 scadu: i know why it fails on x86_64 2015-01-15 10:43:37 TokuFT is supported on 64-bit Centos, should work on other 64-bit linux 2015-01-15 10:43:37 distributions, and may work on OSX 10.8 and FreeBSD. TokuFT is not 2015-01-15 10:43:37 supported on 32-bit systems. 2015-01-15 10:43:56 so on 32bit tokuft is disabled 2015-01-15 10:53:27 hrm, how do you execute a command on new mqtt msg? 2015-01-15 10:54:49 barthalion: mqtt-exec 2015-01-15 10:55:18 https://github.com/ncopa/mqtt-exec 2015-01-15 10:55:33 ncopa: i see you're busy, so just a short question: should we schedule some time to care about the qemu x11less and gtk subpackages? 2015-01-15 10:56:01 tvvdhSL5hEBn: maybe in an hour or so? 2015-01-15 10:56:09 i have lunch in 5mins 2015-01-15 10:56:15 i got mariadb building 2015-01-15 10:56:42 i am slightly interested in testing if jemalloc actually gives any benefit 2015-01-15 11:01:13 huh, in two hours would be better, hour lunchtime is not in sync 2015-01-15 11:04:02 s/hour /our / 2015-01-15 11:06:23 ncopa: testing/encfs: new aport http://sprunge.us/KjJD 2015-01-15 11:06:44 this needs rlog to build, which you just put in testing :) 2015-01-15 11:29:49 ncopa: thx 2015-01-15 11:52:31 barthalion: mind if i submit a new monkey (webserver) abuild? updated version, with --musl-mode, added ssl support 2015-01-15 11:52:48 barthalion: you're down as the maintainer 2015-01-15 11:59:18 sure, feel free to get rid of my name too 2015-01-15 12:00:24 how is monkey doing these days? 2015-01-15 12:03:46 quite good 2015-01-15 12:06:07 ScrumpyJack: we cannot really use github.com/.../master.tar.gz 2015-01-15 12:06:20 next upstream commit and the checksumbreaks for next person 2015-01-15 12:12:40 ncopa: where did i use master? 2015-01-15 12:13:09 source="..../master.tar.gz" 2015-01-15 12:13:43 ahm encfs 2015-01-15 12:13:51 ok, i'll fix that 2015-01-15 12:13:51 you are releasing current git master as 1.8 2015-01-15 12:14:07 so package encfs-1.8 will be a random git snapshot 2015-01-15 12:14:47 i can use archive/release 2015-01-15 12:15:06 http://sprunge.us/AGSi 2015-01-15 12:15:43 i wonder how htat rlog is used 2015-01-15 12:16:03 so:librlog.so.5 2015-01-15 12:16:04 ok 2015-01-15 12:16:39 there should have been a rlog-dev subpkg 2015-01-15 12:16:43 i'm fixing that 2015-01-15 12:19:10 is fuse really needed in the depends? 2015-01-15 12:19:19 dude, you should send my patches back if they are missing things. how will i learn? :) 2015-01-15 12:19:35 or is the so:libfuse.so.2 enough? 2015-01-15 12:19:40 ok 2015-01-15 12:19:47 i didnt pay attention enough on rlog 2015-01-15 12:21:00 fuse provides fusermount with is needed to umount your encfs 2015-01-15 12:21:13 ok 2015-01-15 12:21:18 thats correct then 2015-01-15 12:22:07 this should be ok: http://sprunge.us/hYGA 2015-01-15 12:22:24 note that i set pkgver=1.8_rc1 2015-01-15 12:22:37 there are no 1.8 release tag in github 2015-01-15 12:22:43 but there is a 1.8-rc1 2015-01-15 12:22:58 nice trick +_ver=${pkgver/_/-} 2015-01-15 12:23:04 apk tools does not understand the version string 1.8-rc1 2015-01-15 12:23:10 but it understand 1.8_rc1 2015-01-15 12:23:11 i noticed 2015-01-15 12:23:31 due to we used gentoo once upon a time (alpine 1.7 something) 2015-01-15 12:24:11 second, it looks like the homepage is moving to https://vgough.github.io/encfs/ 2015-01-15 12:24:44 oh 2015-01-15 12:24:47 one thing more 2015-01-15 12:24:57 looks like the main applications are gplv3 2015-01-15 12:25:03 and libs lgpl 2015-01-15 12:25:18 so license varible could be improved 2015-01-15 12:26:15 then libintl and gettext are only build time deps and those are pulled in by gettext-dev 2015-01-15 12:26:24 so only gettext-dev is needed 2015-01-15 12:26:31 might be its not needed at all 2015-01-15 12:26:44 looks like it is only used for autoconf part 2015-01-15 12:27:19 i added the --build=$CBUILD --host=$CHOST for consistancy 2015-01-15 12:27:24 we do that most other places 2015-01-15 12:27:35 and might be usefule if we want support crosscompile in future 2015-01-15 12:28:46 ncopa, the last patch that I sent has a typo: s/mksir/mkdir/ if you prefer: http://sprunge.us/eGiX 2015-01-15 12:29:05 ScrumpyJack: could you curl http://sprunge.us/hYGA | git apply && git add APKBUILD && git commit --amend 2015-01-15 12:29:15 and then resend that patch? 2015-01-15 12:29:27 ScrumpyJack: or do you want me to just apply it? 2015-01-15 12:29:43 i'll curl it 2015-01-15 12:31:31 patch didn't apply. can you apply it and i'll git pull it back here 2015-01-15 12:31:47 what's --build=$CBUILD --host=$CHOST for ? 2015-01-15 12:32:01 i added the --build=$CBUILD --host=$CHOST for consistancy 2015-01-15 12:32:01 we do that most other places 2015-01-15 12:32:01 and might be usefule if we want support crosscompile in future 2015-01-15 12:32:07 we do that on most other apkbuilds 2015-01-15 12:32:17 its mainly to be able to do crosscompiling 2015-01-15 12:32:31 which we dont support currently 2015-01-15 12:32:55 ie, it is useful if we ever would try build arm on x86_64 build machines 2015-01-15 12:33:17 alacerda: i just apply that? 2015-01-15 12:33:58 + mkdir -p "${pkgdir}"/usr/var/run 2015-01-15 12:34:01 that looks wrong 2015-01-15 12:34:18 >>> ERROR: xrdp*: Found /usr/var, localstatedir is most likely wrong 2015-01-15 12:34:18 >>> ERROR: xrdp*: prepare_subpackages failed 2015-01-15 12:34:18 >>> ERROR: xrdp: all failed 2015-01-15 12:35:41 alacerda: http://sprunge.us/RQUF 2015-01-15 12:35:47 i'll try to remember all those things 2015-01-15 12:35:50 i havent checked if the other paths are ok 2015-01-15 12:36:14 --build=$CBUILD --host=$CHOST looks like its in about 25% of packages though :( 2015-01-15 12:36:25 not all are autoconf 2015-01-15 12:37:14 alacerda: http://www.pathname.com/fhs/pub/fhs-2.3.html#THEUSRHIERARCHY 2015-01-15 12:37:15 ncopa, ok.. my abuild -r worked due to my typo... 2015-01-15 12:37:20 (what's --build=$CBUILD --host=$CHOST for ?) ah, that's what i was looking for: it's autoconf specific. I did not know that 2015-01-15 12:37:24 basically /usr is readonly 2015-01-15 12:38:14 ncopa, yeah... i'll need to fix the xrdp code as well then. 'cau their smart-code searchs for .pid files in /usr/var/run 2015-01-15 12:38:26 or the program doesn't start 2015-01-15 12:38:38 you should not need change that 2015-01-15 12:38:47 the --localstatedir=/var should be enough 2015-01-15 12:39:02 let me see ./configure --help 2015-01-15 12:39:07 a secnd... 2015-01-15 12:41:12 ncopa, hum... yes! Let me fix the APKBUILD 2015-01-15 12:42:48 ncopa, http://sprunge.us/LJLI 2015-01-15 12:44:28 find /var/run -name xrdp-sesman.pid -exec rm {} \; 2015-01-15 12:44:39 sorry but i find that kinda funny :) 2015-01-15 12:46:39 iirc the pidfile was invented to make it easier to have multiple instances of same application running on same host 2015-01-15 12:47:01 Well... i don't know why, but hey do this in their code. I just simplified their code. If they have a reason to do so(?!) 2015-01-15 12:47:25 ncopa: can you look at this before i git add it? http://sprunge.us/JRLM 2015-01-15 12:47:31 sure 2015-01-15 12:47:36 ncopa: yeah, I just saw your message on mailing list. I'm not able to test x86_64 builds 2015-01-15 12:48:04 purging all pidfiles instead of a specific kinda makes it impossible to run multiple instances on same box 2015-01-15 12:48:30 which kinda make the pidfile useless 2015-01-15 12:48:43 you could just killall xrdp 2015-01-15 12:49:58 alacerda: one more question. is xrdp installed in /usr/sbin or /sbin ? 2015-01-15 12:54:32 ncopa, I placed it in /sbin and worked. But reading the pathname.com site that you showed me I found: 2015-01-15 12:54:34 "/sbin contains binaries essential for booting, restoring, recovering, and/or repairing the system in addition to the binaries in /bin. [18] Programs executed after /usr is known to be mounted (when there are no problems) are generally placed into /usr/sbin" 2015-01-15 12:54:43 exactly 2015-01-15 12:54:50 so it should go to /usr/sbin/ 2015-01-15 12:54:55 Sou I should change it to /usr/sbin isnt it? 2015-01-15 12:54:57 great 2015-01-15 12:55:00 yup 2015-01-15 12:55:02 also 2015-01-15 12:55:04 while at it 2015-01-15 12:55:24 - if test -e /var/run/xrdp-sesman.pid 2015-01-15 12:55:24 - then 2015-01-15 12:55:24 - rm /var/run/xrdp-sesman.pid 2015-01-15 12:55:24 - fi 2015-01-15 12:55:24 + find /var/run -name xrdp-sesman.pid -exec rm {} \; 2015-01-15 12:55:33 remove those hunks 2015-01-15 12:56:05 do you think I should replace with a "killall" or just drop it off? 2015-01-15 12:56:30 i was thinking just drop 2015-01-15 12:56:33 but hang on a sec 2015-01-15 12:56:39 is this an init.d script? 2015-01-15 12:57:03 yup 2015-01-15 12:57:08 its a sysv init script 2015-01-15 12:57:24 we should replace it with an openrc script 2015-01-15 12:57:55 well... it is placed in /etc/xrdp/xrdp.sh (ugly) 2015-01-15 12:58:49 it looks very much like an init.d script 2015-01-15 12:59:04 and behaves like an initrc script 2015-01-15 12:59:09 #!/bin/sh 2015-01-15 12:59:10 # xrdp control script 2015-01-15 12:59:15 could we replace it with an openrc script? 2015-01-15 12:59:22 or maybe 2 2015-01-15 12:59:32 xrdp-sesman and xrdp 2015-01-15 12:59:50 ;) ok... 2015-01-15 13:01:04 xrdp is widely used but is very ugly =/ 2015-01-15 13:01:17 yeah 2015-01-15 13:03:47 this might be enough as xrdp.initd http://sprunge.us/YeUZ 2015-01-15 13:07:26 cool... testing 2015-01-15 13:41:35 ncopa, for xrdp-sesman.initd I used the "same" file... 2015-01-15 13:41:38 http://sprunge.us/XaQV 2015-01-15 13:41:48 much better than the first one =] 2015-01-15 13:42:23 tell me if there is anything else that should be improved 2015-01-15 13:43:20 the 2 init.d files are so similar 2015-01-15 13:43:30 so i think they could even be the same 2015-01-15 13:43:38 unless one needs to depend on the other 2015-01-15 13:44:00 no, they do not depend on each other.. 2015-01-15 13:44:04 ok 2015-01-15 13:44:22 then you could just let one be a symlink of the other 2015-01-15 13:44:37 command=/usr/sbin/$SVCNAME 2015-01-15 13:44:43 do you think it could be useful keep them splited in case the user need to change sesman config, for eg, he can restart only the sesman 2015-01-15 13:44:59 pidfile=/var/run/$SVCNAME.pid 2015-01-15 13:45:26 ln -s xrdp $pkgdir/etc/init.d/xrpd-sesman 2015-01-15 13:45:56 then you can still start and stop them individually 2015-01-15 13:46:03 nice 2015-01-15 13:59:38 ncopa, http://sprunge.us/gVHg 2015-01-15 14:00:20 2015-01-15 14:08:17 2015-01-15 14:09:55 ncopa, i'm a bit late. wanna reschedule? should i summarize? 2015-01-15 14:11:35 ah, i see in the other chan you're busy 2015-01-15 14:11:40 and interesting topic :0 2015-01-15 14:15:12 i'll go look for coffe 2015-01-15 14:15:34 then I'll flush the current patch queue from sprunge 2015-01-15 14:15:38 and then i'll look at qemu 2015-01-15 14:15:40 ok? 2015-01-15 14:19:37 hmmm, i might have to switch locations, can be back in about 2h 2015-01-15 14:38:39 ok, i'm off back in 90min or so. o/ 2015-01-15 14:48:32 2015-01-15 14:49:12 alacerda: last patch look good 2015-01-15 14:49:23 cool 2015-01-15 14:50:35 ncopa, great... now i gonna focus on vino port then 2015-01-15 14:52:56 alacerda: i have used remmina for rdp 2015-01-15 14:53:08 not sure it does what you need 2015-01-15 16:04:54 ncopa: i'm back 2015-01-15 16:06:29 tvvdhSL5hEBn: and i'm going home... :) 2015-01-15 16:18:04 hehe 2015-01-15 16:38:19 tomorrow we'll be more in sync i hope. i don't have torelocate 2015-01-15 17:19:20 the fi keymap seems to be broken 2015-01-15 17:32:03 'us' keymap is only way to go. 2015-01-15 18:11:46 ncopa: ping 2015-01-15 18:38:40 ncopa, reminna acts as a rdp client i need an rdp server 2015-01-16 08:33:16 looks like there is an issue with alpine lxc-containers 2015-01-16 08:33:24 details here http://sprunge.us/cPRO 2015-01-16 08:33:44 I booted it flawlessly like yesterday 2015-01-16 08:33:58 not on grsec though 2015-01-16 08:52:59 what busybox version do you have? 2015-01-16 08:53:09 i found yet another bug in busybox modprobe 2015-01-16 08:53:29 o its alpine 3.1 2015-01-16 08:54:01 hum 2015-01-16 08:54:10 it mdev mounts that fails 2015-01-16 08:54:24 you shouldnt really run mdev mount in containers 2015-01-16 08:54:40 probably not acpid either 2015-01-16 09:51:25 ncopa: that is a fresh lxc 2015-01-16 09:57:02 also, you can't set the hostname 2015-01-16 10:01:07 ncopa: any chance to look at http://bugs.alpinelinux.org/issues/3680 ? 2015-01-16 10:01:47 clandmeter: I have issues with busybox modprobe atm 2015-01-16 10:02:15 http://lists.busybox.net/pipermail/busybox/2015-January/082186.html 2015-01-16 10:02:30 my laptop only gave a black screen 2015-01-16 10:02:57 and i had to troubleshoot blind with only HDD led as indicator that something happened 2015-01-16 10:03:24 i knew it booted and that i had a login because when typing 'poweroff' and it did poweroff :) 2015-01-16 10:03:37 :) 2015-01-16 10:04:55 so i had to blindly write a text to morse translator in shell so i could pipe all commands to produce morse blinking on the HDD led 2015-01-16 10:05:00 :) 2015-01-16 10:05:21 j/k 2015-01-16 10:05:33 what i did was dmesg > tmpfile 2015-01-16 10:05:46 and booted another usb and analyzed that 2015-01-16 10:06:14 i noticed evbug was running - which only could mean that blacklisting was broke 2015-01-16 10:06:40 apk add kmod should fix it :) 2015-01-16 10:07:44 i actually did that 2015-01-16 10:08:00 but it didnt fix it 2015-01-16 10:08:15 because i915 kernel module is loaded with same broken busybox from initramfs 2015-01-16 10:08:24 ah ok 2015-01-16 10:13:29 now, modprobe code makes my head hurt... 2015-01-16 11:05:44 ncopa: why did you ruin the story 2015-01-16 11:05:52 ncopa: it was really cool and interesting 2015-01-16 12:36:13 ncopa: the gpg key link on the downloads page is broken 2015-01-16 12:39:14 ah 2015-01-16 12:39:20 thanks. will fix 2015-01-16 12:41:07 fixed 2015-01-16 12:41:08 thanks! 2015-01-16 12:48:43 #3699 2015-01-16 12:48:54 for the LXC mess 2015-01-16 12:49:18 apologies for the crappy bug report 2015-01-16 12:53:25 :) 2015-01-16 12:53:33 same problem a while ago 3057 2015-01-16 12:53:45 #3057 2015-01-16 13:04:24 man should depend on man-pages 2015-01-16 13:05:04 or perhaps *-doc should depend on man-pages 2015-01-16 13:05:09 apk add man 2015-01-16 13:05:15 apk add openrc-doc 2015-01-16 13:05:36 apk info -L openrc-doc -> nothing 2015-01-16 13:05:43 apk del openrc-doc 2015-01-16 13:05:48 apk add man-pages 2015-01-16 13:05:51 apk add openrc-doc 2015-01-16 13:05:57 apk info -L openrc-doc -> all man pages 2015-01-16 13:06:39 perhaps man-pages sets up /usr/share/man/man* that needs to be there for other man packages 2015-01-16 13:17:11 clandmeter: re mpd 2015-01-16 13:17:17 i have issues setting up the db 2015-01-16 13:17:25 for mpd 2015-01-16 13:18:53 i have started mpd with default config 2015-01-16 13:19:18 i copy saewill.mp3 to /var/lib/mpd/music 2015-01-16 13:19:27 ncopa-desktop:/var/lib/mpd/music$ mpc update 2015-01-16 13:19:27 error: No database 2015-01-16 13:28:20 ncopa: i think i did touch database 2015-01-16 13:29:16 i edited /etc/mpd.conf 2015-01-16 13:29:18 now it works 2015-01-16 13:29:34 music_directory "/var/lib/mpd/music" 2015-01-16 13:29:34 db_file "/var/lib/mpd/mpd.db" 2015-01-16 13:29:51 i then do mpc update 2015-01-16 13:29:57 i add the file 2015-01-16 13:30:01 mpc update 2015-01-16 13:30:14 mpc add saewill.mp3 2015-01-16 13:30:20 and then mpc play 2015-01-16 13:30:24 its playing here... 2015-01-16 13:30:45 x86? 2015-01-16 13:30:50 x86_64 2015-01-16 13:31:23 kinda sucks that you have to manually edit /etc/mpd.conf to make things work 2015-01-16 13:31:27 it works on arm 2015-01-16 13:31:45 ok 2015-01-16 13:31:57 sounds like x86 specific issue then 2015-01-16 13:43:49 ncopa: i got to go now. happy weekend! 2015-01-16 13:46:30 ok have a nice weekend you too 2015-01-16 16:14:13 i fixed an old eee pc laptops cpu fan 2015-01-16 16:14:20 it had old alpine edge with uclibc 2015-01-16 16:14:24 so i did an upgrade 2015-01-16 16:14:32 which basically means full reinstall 2015-01-16 16:14:47 731 packages installed 2015-01-16 16:15:36 its running the triggers now 2015-01-16 16:16:02 12.5 mins 2015-01-16 16:16:22 2.5GB in 915 packages 2015-01-16 16:16:36 try that with ubuntu ;) 2015-01-16 16:16:39 have a nice weekend 2015-01-16 17:12:07 if i need to execute a command just before starting a service... May I add the command in start() function in the init script? 2015-01-16 17:12:38 it is a "gsettings set" command 2015-01-16 17:26:08 ScrumpyJack: you set LXC container hostnames in /var/lib/lxc/xxxx/config - not inside the container in /etc/hostname 2015-01-16 17:31:53 your other problems are probably due to trying to mount things from inside the container & NOT from /var/lib/lxc/xxxx/config 2015-01-16 17:34:59 LXC is fine for me in 3.1.1 running a container with unbound / dnscrypt & with the host having openvswitch bridging wifi to the guest with the help of NAT 2015-01-16 19:58:07 morning 2015-01-16 19:58:25 what's the current way to build isos? alpine-iso on 3.1.1 doesn't seem to work very well :s 2015-01-16 20:01:13 never mind, i had the wrong repo ;) 2015-01-16 21:14:05 BitL0G1c: im running stock apline LXC. hostname is called by klogd and syslog and barfs. there should be nothing wrong with modifying hostname inside the container. 2015-01-16 21:46:42 ScrumpyJack - hostname is configured via 'lxc.utsname = ' in the container config - to change it from within the container you would need to give the container capability to do it 2015-01-16 21:59:14 lxc.utsname is indicative. you should be able to change it in the container 2015-01-16 21:59:30 just looking 2015-01-16 22:01:18 The utsname section defines the hostname to be set for the container. That means the container can set its own hostname without changing the one from the system. That makes the hostname private for the container. 2015-01-16 22:02:05 hmm, that's a bit ambiguous. 2015-01-16 22:14:04 anyway, you can change hostname in a container. hostname in guest is trivial, ie not something the host cares about 2015-01-16 22:46:27 ScrumpyJack - see these notes for LXC http://comments.gmane.org/gmane.linux.kernel.containers.lxc.general/5338 - certain configurations s/be done outside the container. The whole point of running a container is to drop privileges 2015-01-16 22:52:56 BitL0G1c: you've lost me there. changing the hostname inside a container doesn't/shouldn't be a problem. 2015-01-16 22:55:53 BitL0G1c: there should be a way of changing a hostname in an LXC container without it affecting the host. don't you think? 2015-01-16 23:05:28 BitL0G1c: anyway, i dont know enough about cgoups and priv. sep. in general to be able to fix it 2015-01-17 03:14:53 net-snmp-tools should be another dependency for cacti package 2015-01-17 03:14:59 since snmpwalk is in there and cacti wants it 2015-01-17 03:15:34 and a bunch of other things cacti wants :p 2015-01-17 21:59:17 Hello all, this is my first foray into Alpine and I'm trying to compile some drivers for a USB->Gigabit adapter. 2015-01-17 21:59:52 However, I am getting errors when trying to make and I believe it is due to the kernel headers being a different version than my installed version 2015-01-17 22:00:58 I found the correct version on the package browser website, but can't seem to get them to show up on my box. I'm assuming I'm missing something in my repository file, but can't find any info. 2015-01-17 22:03:37 Av8or1ab: it's quite late, so forgive me if I stop to reply abruptly 2015-01-17 22:04:04 IIRC, the package search on our website is tracking our development release, edge 2015-01-17 22:06:41 ok, I enabled the edge/main repository, but not the edge/testing. However, after I update and search for headers it only infds 3.12.6 2015-01-17 22:07:30 maybe your mirror is out of date? 2015-01-17 22:07:45 nl.alpinelinux.org should be ok 2015-01-17 22:08:04 ok, I ma using the repos.lax-noc.com, I'll make the switch now 2015-01-17 22:08:39 ah, ok 2015-01-17 22:08:41 I get it now 2015-01-17 22:08:52 no, it's not the mirror 2015-01-17 22:09:01 you're right about linux-headers being out of date 2015-01-17 22:10:32 I think you want linux-grsec-dev instead 2015-01-17 22:11:55 Av8or1ab: could you try it and see whether the driver compiles now? 2015-01-17 22:12:40 ok, trying that now. Rebooting quickly to match kernel versions 2015-01-17 22:14:54 great, that seems to work. Now just some other errors that I have to try and handle. Thanks for the quick response 2015-01-17 22:17:58 you're welcome :) 2015-01-17 22:18:03 and good night 2015-01-17 22:18:24 You as well from the West Coast of the US 2015-01-18 00:20:33 https://www.ctrlc.hu/~stef/aports.txz <- s6+deps (skarnet.org/software), and percy++ (inkl deps), +bonus libseccomp - aports/testing/*/APKBUILD files and requisites 2015-01-18 00:22:07 also i have switched over to s6 for all serivces (disabled all in openrc), and call s6 with /etc/local.d/s6.start. my thinkpad now reaches login prompt in 3secs. 2015-01-18 00:22:18 (i don't use x11 though) 2015-01-18 00:23:18 altough i guess if you have to cover all the corners a distro has to, then it would be slower i guess. 2015-01-18 00:23:36 (oh, i have coreboot, i guess that counts as well into the bootspeed) 2015-01-18 00:24:12 regarding percy++ see this: http://media.ccc.de/browse/congress/2014/31c3_-_6140_-_en_-_saal_1_-_201412291715_-_dp5_pir_for_privacy-preserving_presence_-_ian_goldberg_-_george_danezis_-_nikita_borisov.html 2015-01-18 00:24:28 and libseccomp would be nice to be added to qemu as a dependency 2015-01-18 00:24:45 latter enables a sandboxing mode, which is quite nice. 2015-01-18 03:35:33 hi, i am trying to use ocfs2, but ran into the issue where it isn't available. I came across this issue https://bugs.alpinelinux.org/issues/2982 - is this due to musl, and will this be able to build at some point? I have tried to build it from aports, but didn't manage to get past ./confgure due to lots of dependencies such as PyGTX etc. 2015-01-18 03:36:36 if it won't be able to build in any case due to musl I will have to park the alpine project for now as ocfs2 is a hard dependency for my project, I am on a tight deadline so limited time to learn a new distro and sort our complex build issues. 2015-01-18 03:36:59 appreciate any comments and feedback! 2015-01-18 20:36:52 /join #musl 2015-01-18 20:36:57 meh 2015-01-18 20:40:45 hey I can't create account on wiki 2015-01-18 20:51:35 Is there any deb package for Alpine 2.20 ? There's non in Ubuntus repo other then the previous version 2.10. Besides 2.10 is buggy when it comes to tls/ssl since it sends everyhing in plain. 2015-01-18 21:21:54 K350: it's distribution channel, not mail client :p 2015-01-18 21:21:58 Roosevelt: ask clandmeter 2015-01-18 21:30:57 barthalion: Oups..I go tconfused for a min there :-) 2015-01-19 10:08:50 what the best way to get round abuild not liking my $pkgver and giving me " ... is not a valid version" 2015-01-19 10:09:21 hard to say without saying the value of $pkgver 2015-01-19 10:09:35 pkgver=1.8build5 2015-01-19 10:10:32 morning 2015-01-19 10:10:52 ncopa: have you found any issue/solution for mpd? 2015-01-19 10:13:09 ScrumpyJack: maybe you can do pkgver=1.8_p5 2015-01-19 10:13:34 _ver=${pkgver/_p/build} 2015-01-19 10:14:20 clandmeter: no, i havent but i managed to fix my old eee pc so i should have a box to test it 2015-01-19 10:14:37 nice 2015-01-19 10:14:40 no hurry 2015-01-19 10:14:54 i did pkgver=1.8 and _ver=${pkgver}build5 and put $_ver in URL. Is the ok? 2015-01-19 10:15:11 how would you do build6 when that comes? 2015-01-19 10:15:39 1.8_build5 should be ok 2015-01-19 10:16:05 im not sure apk understands _build 2015-01-19 10:16:07 when build6 comes out, update _ver ? 2015-01-19 10:16:27 but package is: foo-1.8-r0.apk 2015-01-19 10:16:42 maybe _b then? 2015-01-19 10:17:03 barthalion: i dont think it undertands that either 2015-01-19 10:17:09 sounds silly 2015-01-19 10:17:09 ncopa: i don't follow 2015-01-19 10:17:27 string is string 2015-01-19 10:17:30 it undertands _alpha, _beta _pre, _git 2015-01-19 10:17:50 _alpha2 < _beta1 2015-01-19 10:18:26 what is newest version 1.0_build2 1.0_string3 ? 2015-01-19 10:18:52 apk also understands _p as patch level 2015-01-19 10:19:06 which appears to correpond with "build" in this case 2015-01-19 10:19:10 ScrumpyJack: do you package something binary? 2015-01-19 10:19:22 "build" is kinda bad name on a source tarball :) 2015-01-19 10:19:26 yeah 2015-01-19 10:20:40 did i say buid? sorry, it's beta 2015-01-19 10:21:58 1.8build5 to be precise 2015-01-19 10:22:07 you said build again 2015-01-19 10:22:15 duh, i did it again 2015-01-19 10:22:15 now you said build again 2015-01-19 10:22:18 :D 2015-01-19 10:22:18 :D 2015-01-19 10:22:19 lol 2015-01-19 10:22:19 1.8beta5 !!! 2015-01-19 10:22:26 that makes it easier 2015-01-19 10:22:30 1.8_beta5 2015-01-19 10:22:33 yes 2015-01-19 10:22:35 ok 2015-01-19 10:22:36 thanks 2015-01-19 10:22:44 ${pkgver/_} if you want to get rid of _ somewhere 2015-01-19 10:22:56 _ver=${pkgver/_/} 2015-01-19 10:23:12 you dont need the second / there? 2015-01-19 10:23:18 yeah, I think so 2015-01-19 10:23:47 you need it if you want to replace the match with something 2015-01-19 10:23:54 but by default it just removes 2015-01-19 10:24:12 cool 2015-01-19 10:24:17 i didnt know 2015-01-19 10:24:23 so i learned something new today :) 2015-01-19 10:29:37 and of course, it unpacks to pkg-1.8 so what's the prefered _builddir change? pkgname-1.8 or create another _cddir=${pkgver/_beta5} and _builddir=$pkgname=$_cddir 2015-01-19 10:30:19 with ${pkgver%_*} you can get the part before _ 2015-01-19 10:30:23 _builddir=${pkgname}-${_cddir} even - butterfingers 2015-01-19 10:30:31 so 1.8_build5 becomes 1.8 2015-01-19 10:30:45 barthalion: clever 2015-01-19 10:31:22 bread and butter of ours 2015-01-19 10:36:28 you mean you bothered to learn your shell ;) 2015-01-19 10:40:30 I would be insane if I were using bash or POSIX compliant sh, I use zsh :p 2015-01-19 10:41:46 ncopa: some packages for testing: https://www.ctrlc.hu/~stef/aports.txz <- s6+deps (skarnet.org/software), and percy++ (inkl deps), +bonus libseccomp - aports/testing/*/APKBUILD files and requisites 2015-01-19 10:42:27 more bg info about this in scrollback buffer at 01:19:56 on Sun, 18 Jan 2015 2015-01-19 10:43:27 also i have a bugfix for musl/getspnam_r https://www.ctrlc.hu/~stef/spnam.patch - reported and also submitted upstream, but i guess dalias is still evaluating the patch. 2015-01-19 10:45:41 so, now configure is looking for sys/poll.h - if i sed sys/ out of configure, that's fine. is there a better way? 2015-01-19 11:15:16 ScrumpyJack: hacking configure is rarely a solution 2015-01-19 11:15:22 you should check config.log why it failed 2015-01-19 11:41:23 ncopa, you can move xrdp to main when you want... I've tested it here in lab and installed in a production box. Since friday it is working perfectly 2015-01-19 11:42:07 and regarding the vino patch let me know if it is fancy or need any ajustment. (I am already using it in production here also) 2015-01-19 11:52:43 it fails looking for poll.h Normally, i'd see a redirecting of sys/poll.h to poll.h and everyone is happy, but for some reason configure failed nevertheless 2015-01-19 11:52:46 http://sprunge.us/XahD 2015-01-19 12:01:58 change it in configure.ac and run autoreconf in prepare() 2015-01-19 13:12:32 barthalion: I'll give that a go 2015-01-19 13:36:11 ncopa would you put my apkbuilds into testing? s6+deps could even go into main, as i use them. i'm thinking of adding s6 initscripts to the packages i already support in my setup. or should i do extra unbound-s6-init.apks? 2015-01-19 13:37:03 using s6 and some extensive kernel boot params (about 600chars in total) i managed to boot to a prompt within 2s. 2015-01-19 13:37:30 (also i use coreboot, which also counts for speedups) 2015-01-19 13:42:27 also if you want we can chat about qemu, i'll be around all day 2015-01-19 14:59:26 tvvdhSL5hEBn: do you think you could send your apkbuilds as patches? 2015-01-19 15:02:32 sure they are all new 2015-01-19 15:02:39 a sec 2015-01-19 15:10:02 ncopa: all of them together: https://www.ctrlc.hu/~stef/aports.patch, and seperately: https://www.ctrlc.hu/~stef/libseccomp.patch https://www.ctrlc.hu/~stef/percy++.patch https://www.ctrlc.hu/~stef/s6.patch 2015-01-19 15:10:56 tvvdhSL5hEBn: i was thinking somethign that i can git am 2015-01-19 15:11:29 ah. ok. takes a bit of time, i need to fetch a clean aports of yours then. 2015-01-19 15:12:39 is it ok as one big patch, or you want 3 seperate commits? 2015-01-19 15:12:50 or single commits/pkg? 2015-01-19 15:13:44 one patch with 3 commits or patch per pkg 2015-01-19 15:14:16 they are the same actually, both are simple to git am 2015-01-19 15:14:40 yeah. i was mixing commits with patches. my bad 2015-01-19 15:19:08 good day to all 2015-01-19 15:19:13 o/ 2015-01-19 15:33:02 ncopa: like this? https://www.ctrlc.hu/~stef/s6percyseccomp.patch 2015-01-19 15:46:49 ncopa: I pushed new tag for posixtz, but no tarball created 2015-01-19 15:47:08 the post-receive hook still wasn't updated with chown 2015-01-19 15:48:05 ncopa: nevermind, it was created, just no log in this channel like I expected 2015-01-19 15:50:59 tvvdhSL5hEBn: the configure script is stupid 2015-01-19 15:51:04 Checking whether system has /dev/random... 2015-01-19 15:51:10 i think its waiting for entropy 2015-01-19 15:52:20 hmm, yes i noticed that myself. 2015-01-19 15:52:33 but by the time i figured it out, it went ahead 2015-01-19 15:52:48 how do we handle that with our build servers? 2015-01-19 15:53:15 can you pls help me remember which pkg was that again? 2015-01-19 15:53:15 also, nitpick, we normally do ... || return 1 instead of ... || exit 1 2015-01-19 15:53:21 skalibs 2015-01-19 15:53:24 oh, sorry! 2015-01-19 15:53:27 ok it moved on finally 2015-01-19 15:54:04 patching it out from the configure script is one option. i guess in parallel i'll contact the dev of s6 about this? 2015-01-19 15:54:25 tvvdhSL5hEBn: do you think you can clean up the the || exit 1? 2015-01-19 15:54:36 yes, i think we can simply patch out the test 2015-01-19 15:54:50 sure 2015-01-19 15:54:53 might be we can set some magiv env var too 2015-01-19 15:55:17 --enable-force-devr assume /dev/random exists and is valid [autodetection takes time] 2015-01-19 15:55:25 ! 2015-01-19 15:55:30 add that 2015-01-19 15:55:34 i'll add that 2015-01-19 15:56:53 i uploaded a new patch to the earlier url. with the s/exit/return/ and the -enable-force-devr 2015-01-19 16:02:02 libsecomp fails 2015-01-19 16:02:12 libseccomp.pc (/home/ncopa/aports/testing/libseccomp/pkg/libseccomp//usr/lib/pkgconfig) 2015-01-19 16:02:12 install: can't stat '/bin/bash 2015-01-19 16:02:13 -c 2015-01-19 16:02:13 libseccomp.pc': No such file or directory 2015-01-19 16:02:13 Makefile:95: recipe for target 'install' failed 2015-01-19 16:02:13 make: *** [install] Error 1 2015-01-19 16:03:09 hmm. checking/fixing 2015-01-19 16:04:59 i added bash as a makedependency 2015-01-19 16:05:05 updated the url with the patch 2015-01-20 01:10:01 hello 2015-01-20 01:11:51 Is it possible to create "File System Navigator (fsn; pronounced "fusion") is an experimental application to view a file system in 3D, made by SGI for IRIX systems." 2015-01-20 01:12:58 The one on Jurassic Park back in the day. I have being working on such an idea. 2015-01-20 10:43:09 Tried to compile openjdk. The bug described in https://bugs.gentoo.org/show_bug.cgi?id=534118 also applies to alpine linux. 2015-01-20 10:43:46 Error: time is more than 10 years from present: 1104530400000 2015-01-20 10:44:12 funny error. Triggered since 01-01-2015 2015-01-20 11:04:25 heh... https://bugs.gentoo.org/show_bug.cgi?id=534118#c7 2015-01-20 13:01:07 ncopa: you forgot about jemalloc 2015-01-20 13:02:25 oh 2015-01-20 13:02:56 but hum 2015-01-20 13:03:22 scadu: there is one more thig i would like to test if you have time to help me 2015-01-20 13:03:44 i would like to run some mariadb benchmark 2015-01-20 13:03:51 with jemalloc and without jemalloc 2015-01-20 13:04:35 i suspect that jemalloc does not improve perfomance significantly 2015-01-20 13:04:39 ncopa: mariadb didn't depend on jemalloc in past 2015-01-20 13:04:53 ncopa: okay, I'll check it. 2015-01-20 13:05:00 i think jemalloc makes sense on glibc 2015-01-20 13:05:12 should do significant better performance 2015-01-20 13:05:16 it depends on number of cores 2015-01-20 13:05:34 jemalloc is awesome on >4 threaded CPUs IIRC 2015-01-20 13:05:47 there are benchmarks somewhere 2015-01-20 13:05:59 I don't think we care much about it 2015-01-20 13:06:11 it's trivial to make mariadb use any malloc library without rebuilding 2015-01-20 13:06:51 http://www.percona.com/blog/2013/03/08/mysql-performance-impact-of-memory-allocators-part-2/ 2015-01-20 13:07:46 (and part 1 too, obviously) 2015-01-20 13:10:03 sadbarthalion.jpg 2015-01-20 13:10:29 well, I have only two threads and small amount of RAM. my benchmarks will be useless. 2015-01-20 13:11:19 scadu: paste ncopa what I said when he's back, I'm going back to work 2015-01-20 13:19:32 http://locklessinc.com/benchmarks_allocator.shtml 2015-01-20 13:19:36 there is a graph at the end 2015-01-20 13:20:43 ok they use sysbench 2015-01-20 13:21:20 14:05:14 it depends on number of cores 2015-01-20 13:21:23 14:05:32 jemalloc is awesome on >4 threaded CPUs IIRC 2015-01-20 13:21:25 14:05:46 there are benchmarks somewhere 2015-01-20 13:21:27 14:05:57 I don't think we care much about it 2015-01-20 13:21:29 14:06:09 it's trivial to make mariadb use any malloc library without rebuilding 2015-01-20 13:21:31 14:06:49 http://www.percona.com/blog/2013/03/08/mysql-performance-impact-of-memory-allocators-part-2/ 2015-01-20 13:21:33 14:07:12 build-edge-x86_64: files from v3.1.0-466-gec681e0 uploaded 2015-01-20 13:21:35 14:07:31 build-edge-x86: files from v3.1.0-466-gec681e0 uploaded 2015-01-20 13:21:37 14:07:44 (and part 1 too, obviously) 2015-01-20 13:21:44 thanks. isp issues 2015-01-20 13:26:48 scadu: you can add the line -DWITH_JEMALLOC=no 2015-01-20 13:26:56 i wonder if we just disable jemalloc for now 2015-01-20 13:30:21 ncopa: I just wanted to commit changes. 2015-01-20 13:30:27 barthalion: sorry, I was afk 2015-01-20 13:30:56 I expell you from our wizard school 2015-01-20 13:31:02 BURN 2015-01-20 13:31:09 :( 2015-01-20 13:31:25 scadu: ok lets drop mysql performance tests for now 2015-01-20 13:31:34 scadu: https://drawception.com/pub/panels/2012/4-4/KfncHgTjQ8-8.png 2015-01-20 13:32:12 barthalion: you don't even know how hard I laugh now 2015-01-20 13:32:15 ncopa: noted. 2015-01-20 13:34:02 cool! https://github.com/ryonkn/multithread-malloc 2015-01-20 13:34:09 barthalion: but still I have this: https://i.imgur.com/03lx8Rq.png 2015-01-20 13:34:25 LOL 2015-01-20 13:35:57 http://www.quickmeme.com/img/f6/f6c61e04c0b3478f72b4606587c0d16324c3bdafbcd532cf2b4dcaa67ac415cb.jpg 2015-01-20 13:37:31 apt so fast, wow 2015-01-20 13:42:14 apt is the right tool if you want to die waiting until it'll finish upgrade 2015-01-20 13:42:45 uhm, I screwed up this sentence. 2015-01-20 14:17:13 seems like jemalloc is ~ double speed compared to musl malloc when threads > 8 and above 2015-01-20 14:17:27 single thread is musl libc slightly faster 2015-01-20 14:22:20 rnalrd: you should probably move libmemcached to main 2015-01-20 14:22:30 y 2015-01-20 14:22:34 fossil (20150119112900) , monit (5.11) update pls 2015-01-20 14:22:43 im running malloc benchmarks as we speak 2015-01-20 14:23:01 so any builds will disrupt the stats 2015-01-20 14:23:17 so if you please could hold it 5mins 2015-01-20 14:23:30 Threads: 18 2015-01-20 14:23:30 malloc avg: 23.4929 2015-01-20 14:23:30 jemalloc avg: 9.24303 2015-01-20 14:23:40 nice 2015-01-20 14:24:06 that's huge difference 2015-01-20 14:24:25 probably higher memory usage as well 2015-01-20 14:24:45 but I doubt it matters when you have 18 threads and want your db to be fast 2015-01-20 14:25:59 I was rethinking if I could persuade some ram/flashdisk manufacturer for usb based ramdisk ! 2015-01-20 14:26:20 would be nice to run alpine using it 2015-01-20 14:27:30 would try on gskill forum, any suggestions ? 2015-01-20 14:28:14 heh rnalrd could hold it more than 3 mins... :) 2015-01-20 14:28:51 ? 2015-01-20 14:29:00 im running malloc benchmarks as we speak 2015-01-20 14:29:00 so any builds will disrupt the stats 2015-01-20 14:29:00 so if you please could hold it 5mins 2015-01-20 14:29:13 oh sry 2015-01-20 14:29:13 it doesnt matter 2015-01-20 14:29:26 the number will just be less accurate 2015-01-20 15:30:08 re the jemalloc test 2015-01-20 15:30:22 it allocates big block 2015-01-20 15:30:56 size[i] = ( rand() % (mallocSize * 10) + 1 ) * 100 * 1000; 2015-01-20 15:31:30 so it callcluates random(1-10) * 100kb 2015-01-20 15:31:43 if i do small allocs 2015-01-20 15:31:48 *100 only 2015-01-20 15:32:27 100 -> 1000 bytes instead 2015-01-20 15:32:36 then musl malloc wins *big* time 2015-01-20 15:32:41 ~30x speed 2015-01-20 15:33:00 Threads: 24 2015-01-20 15:33:00 malloc avg: 0.0194295 2015-01-20 15:33:00 jemalloc avg: 3.0301 2015-01-20 15:37:42 my bad 2015-01-20 15:37:47 bad test 2015-01-20 15:49:53 jemalloc is still ~ double as fast 2015-01-20 16:04:41 alacerda: i dont know if vino should have a init.d script 2015-01-20 16:04:56 maybe it can be autostarted on login? 2015-01-20 16:05:45 ncopa, if we do not have one... hum, i need to have it started so xrdp-sesman can talk to it through TCP 5900 2015-01-20 16:07:07 ok, i dont know 2015-01-20 16:33:04 ncopa: so jemalloc comes back to apkbuild or not? 2015-01-20 16:39:00 not my call, but it should not 2015-01-21 12:41:51 ncopa, it seems that we are have some troubles with this http://lists.alpinelinux.org/alpine-devel/4581.html ... err, if the package builds successfuly, how can I know that it depends on bsd-compat-headers (id abuild didn't complain about it) ? 2015-01-21 12:48:26 you can build it with edge and make sure that bsd-compat-headers is not installed during build time 2015-01-21 12:48:42 or you can grep the sources 2015-01-21 12:48:59 grep '' ... 2015-01-21 12:49:45 i try to keep my dev box as clean as possible 2015-01-21 13:17:50 ncopa, fossil (20150119112900) , monit (5.11) 2015-01-22 10:34:42 ncopa, mount -o loop also would not work from within lxc ? 2015-01-22 10:38:00 depends if you give the necessary permissions in lxc config 2015-01-22 10:38:09 ok 2015-01-22 10:38:11 ncopa, will there be an interest in adding SGD(super grub disk) to AL isos ? 2015-01-22 10:38:12 i dunno if cap_admin is needed or something 2015-01-22 10:38:31 what is super grub disks? 2015-01-22 10:38:40 http://www.supergrubdisk.org/index.php?pid=5 2015-01-22 10:38:43 we have tried to stay away from grub 2015-01-22 10:38:57 the booting will be syslinux 2015-01-22 10:39:13 but can load the grubdisk(single file) 2015-01-22 10:39:26 ~ 367 MB Size 2015-01-22 10:39:36 it would double the size 2015-01-22 10:39:49 only 4mb 2015-01-22 10:39:52 gzipped 2015-01-22 10:39:53 kernel memdisk 2015-01-22 10:39:54 append initrd=/bootdisk/grubdisk.img floppy raw 2015-01-22 10:40:21 oh that other thing 2015-01-22 10:40:22 ok 2015-01-22 10:40:30 :) 2015-01-22 10:40:46 Helps you to fix your GRUB or GRUB2 2015-01-22 10:40:52 it offers all grub features 2015-01-22 10:41:03 grub2 features 2015-01-22 10:41:08 i dont know why we should ship that with alpine iso if we don't even support grup in first place 2015-01-22 10:41:11 2.0.2 2015-01-22 10:41:57 why would we be interested in adding that to AL iso? 2015-01-22 10:42:22 what value would it add for alpine linux? 2015-01-22 10:42:41 coz it adds easy way to detect other partitions/disk and boot from it 2015-01-22 10:43:17 but not much , just a little 2015-01-22 10:43:46 just a thought, would be nice if you tried it sometime when free 2015-01-22 10:44:12 also nice if someone wants to remaster the cd 2015-01-22 10:44:56 grubdisk also has nice direct iso file detect/boot feature 2015-01-22 10:46:58 I also notices /syslinux folder in 2.7 iso cds ? 2015-01-22 10:47:05 noticed* 2015-01-22 10:47:25 should it be isolinux? or its not a generall convention ? 2015-01-22 10:50:17 isolinux will look in both isolinux and syslinux subdirs 2015-01-22 10:50:32 the idea with renaming it to syslinux is that the usb can reuse it 2015-01-22 10:50:53 i'd like a single syslinux.cfg for both isolinux (cd boot) and usb boot 2015-01-22 10:52:56 yes, I thought that, easy on install script 2015-01-22 11:45:31 doing some stuff on on armhf builder 2015-01-22 11:45:34 taking it off for a while 2015-01-22 12:44:22 grand idea, AL super pendrive (usb ramdisk + pendrive), in one device, any thoughts ? ;-)) 2015-01-22 12:44:41 we can label and sell :)) 2015-01-22 12:51:09 ncopa, what is alpine-3.1.0-x86_64.iso.sig file ? 2015-01-22 12:51:47 and why there is no alpine-3.1.1-x86_64.iso.sig ? 2015-01-22 12:54:09 the .sig was renamed to .asc 2015-01-22 12:55:52 ok, thanks 2015-01-22 13:17:08 ACTION pokes ncopa for some mpd love 2015-01-22 13:17:45 thanks :) 2015-01-22 13:18:00 :) 2015-01-22 13:18:09 the x86 lappy is right here now 2015-01-22 13:18:16 so i have no excuse anymore 2015-01-22 13:18:21 ah i have one excuse 2015-01-22 13:18:26 my bt keyb is broken 2015-01-22 13:18:34 and i hate to work on this dell keyb 2015-01-22 13:18:46 i was hoping to fix that first... 2015-01-22 13:47:58 and I just ordered rapoo X1800, wireless kb 2015-01-22 13:48:10 I mean yesterday ;) 2015-01-22 14:05:00 vkrishn: bluetooth? 2015-01-22 14:07:47 wireless, but not bt 2015-01-22 14:09:21 http://www.rapoo.com/ProductShow.aspx?PType=8fQpYH%2b%2b7i8%3d&PID=eFGeQHLufoU%3d 2015-01-22 14:10:20 I was planning a compact size for using as config(carry anywhere admin kb) 2015-01-22 14:10:30 8000 Wireless Mouse & Keyboard Combo, seems ok 2015-01-22 14:11:36 but don't know if I could do alt+insert 2015-01-22 14:21:20 ;), and I also have a grand-new-kb design, but don't know where to put 2015-01-22 22:02:24 http://www.flockport.com/new-micro-containers-based-on-alpine-linux/ 2015-01-23 07:01:20 ok 2015-01-23 07:01:27 i think i now have usable wandboard image 2015-01-23 07:01:29 http://dev.alpinelinux.org/~tteras/alpine-uboot-3.1.1-test-armhf.rpi.tar.gz 2015-01-23 07:01:33 in case someone wants to test 2015-01-23 07:02:02 there's also some beaglebone support, but it's probably not working yet (u-boot script is not fixed) 2015-01-23 07:41:29 i will hopefully get my sd cards for booting my wandboards today 2015-01-23 08:14:21 fabled: commit 881d55a53a111ea06854f4596266c3b044f5eb53 2015-01-23 08:14:21 Author: Timo Teräs 2015-01-23 08:14:21 Date: Wed Jun 25 14:05:30 2014 +0300 2015-01-23 08:14:21 main/php: fix php_crypt implementation 2015-01-23 08:14:30 what was that about? 2015-01-23 08:14:46 php-fix-crypt.patch 2015-01-23 08:15:01 it does not apply with php-5.6.5 2015-01-23 08:15:15 might be fixed upstream, there was a bug about it 2015-01-23 08:15:25 php by default ended up using the non-thread safe crypt() 2015-01-23 08:15:33 lol 2015-01-23 08:15:38 https://bugs.php.net/bug.php?id=45430 2015-01-23 08:15:57 i also filed something i think 2015-01-23 08:16:00 just a sec 2015-01-23 08:16:23 https://bugs.php.net/bug.php?id=67512 2015-01-23 08:17:29 they have really messed up their crypt stuff 2015-01-23 08:17:32 i had a similar bug 2015-01-23 08:17:35 https://bugs.php.net/bug.php?id=67827 2015-01-23 08:17:41 which is just *stupid* 2015-01-23 08:18:46 yes, and the code they had is broken horribly 2015-01-23 08:18:55 the *default* is *empty* function 2015-01-23 08:19:13 meaning you end up with random crashes and horribly broken setup 2015-01-23 08:19:19 only if you get magic defines from system, it works 2015-01-23 08:19:57 you want me to rebase the patch? 2015-01-23 08:20:20 seems like the fixed it upstream 2015-01-23 08:20:32 no theyhavent 2015-01-23 08:20:42 if _REENTRANT is not set 2015-01-23 08:20:57 it does not do anything 2015-01-23 08:22:32 what they changed is: 2015-01-23 08:22:36 crypt_res = crypt_r(password, salt, &buffer); 2015-01-23 08:22:39 if (!crypt_res) { 2015-01-23 08:22:40 return FAILURE; 2015-01-23 08:22:40 } else { 2015-01-23 08:22:48 the if condition changed to: 2015-01-23 08:23:10 crypt_res = crypt_r(password, salt, &buffer); 2015-01-23 08:23:10 if (!crypt_res || (salt[0] == '*' && salt[1] == '0')) { 2015-01-23 08:23:10 return FAILURE; 2015-01-23 08:23:10 } else { 2015-01-23 08:23:10 *result = estrdup(crypt_res); 2015-01-23 08:23:10 return SUCCESS; 2015-01-23 08:23:13 } 2015-01-23 08:23:38 right 2015-01-23 08:23:44 patch applies with fuzz 3 2015-01-23 08:23:48 -F3 2015-01-23 08:24:10 that'd be enough to refresh a new patch 2015-01-23 08:24:14 i suppose i can do patch surgery 2015-01-23 08:24:17 or just hand-edit that one changed line 2015-01-23 08:24:34 add reference to https://bugs.php.net/bug.php?id=67512 while at it 2015-01-23 08:24:40 so we don't forget why the patch is there 2015-01-23 08:25:43 brb 2015-01-23 11:44:03 ncopa, I have a bootable usbdisk(vfat), syslinux, If I manually install AL on it ? 2015-01-23 11:44:40 meaning can I ? 2015-01-23 11:45:57 it seems to be loading vmlinux,initramfs but the stalls after few msgs 2015-01-23 11:48:36 http://i.imgur.com/WoHzNZL.png?1 2015-01-23 13:44:37 can I do a usb install and boot similar to cdrom, i.e "modloop-grsec" mount method ? 2015-01-26 12:26:08 uggedal: do you think you could upgrade docker? 2015-01-26 12:29:54 we're still at go 1.3? 2015-01-26 12:31:23 I posted last week about a problem in openjdk. Did someone reproduce this problem? 2015-01-26 12:31:38 if we are, I can take care of it, but there is a regression with newest docker + go 1.4 2015-01-26 12:53:24 there are some CVEs 2015-01-26 12:53:26 with docker 2015-01-26 12:53:37 i thikn we still have 1.2 2015-01-26 12:53:43 i'm ok with upgrade to 1.3.x 2015-01-26 13:20:53 let me spin up container to test the build 2015-01-26 13:26:01 barthalion: i want resolve this: http://bugs.alpinelinux.org/issues/3749 2015-01-26 13:26:14 and thanks a bunch for looking at it 2015-01-26 13:29:23 if I only know my password to see it :c) 2015-01-26 13:29:27 some CVE? 2015-01-26 13:30:58 yup 2015-01-26 13:31:24 docker: multiple issues (CVE-2014-6407, CVE-2014-6408) 2015-01-26 13:31:44 CVE-2014-6407: 2015-01-26 13:31:44 Docker before 1.3.2 allows remote attackers to write to arbitrary files and execute arbitrary code via a (1) symlink or (2) hard link attack in an image archive in a (a) pull or (b) load operation. 2015-01-26 13:31:44 CVE-2014-6408: 2015-01-26 13:31:44 Docker 1.3.0 through 1.3.1 allows remote attackers to modify the default run profile of image containers and possibly bypass the container by applying unspecified security options to an image. 2015-01-26 13:31:57 CONFIRM: https://docs.docker.com/v1.3/release-notes/ 2015-01-26 13:31:57 https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2014-6407 2015-01-26 13:31:57 https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2014-6408 2015-01-26 13:33:41 barthalion: do you want me to reset your password on bugs.a.o? 2015-01-26 13:41:27 no, I don't have it remembered on my work laptop 2015-01-26 13:41:34 I saved it on personal one 2015-01-26 13:41:44 and I don't use fx sync 2015-01-26 13:51:23 how is that fancy package with linux headers called? 2015-01-26 13:56:38 linux-headers 2015-01-26 13:56:55 the other is bsd-compat-headers 2015-01-26 13:57:01 for sys/queue.h 2015-01-26 13:57:09 ah, ok 2015-01-26 13:57:17 yeah, figured out it would be linux-headers 2015-01-26 13:57:23 is it out of date for purpose? 2015-01-26 13:58:11 no 2015-01-26 13:58:18 well, we have 3.14 kernel still 2015-01-26 13:58:28 yeah, but it's 3.12 atm 2015-01-26 13:58:33 oh 2015-01-26 13:58:42 ok, we hsould update that then 2015-01-26 13:58:54 http://sprunge.us/gDIG 2015-01-26 13:59:18 I took the liberty to ditch docs subpkg 2015-01-26 13:59:21 do we want push 1.4 to v3.1-stable ? 2015-01-26 13:59:26 probably 2015-01-26 13:59:31 ok for me with no docs 2015-01-26 13:59:49 it builds and starts, need to check some container yet 2015-01-26 14:00:57 mhm, ok, I'm having issues because I run docker inside lxc 2015-01-26 14:01:17 :) 2015-01-26 14:03:57 can you check it on bare metal/vm? apk add docker, rc-service docker start, docker pull busybox; docker run --rm busybox echo foobar 2015-01-26 14:05:15 I'm quite sure it works anyway, but I have too many vms booted to test it 2015-01-26 14:05:28 im on it 2015-01-26 14:06:08 it should download busybox image, run echo in container and remove it afterwards 2015-01-26 14:06:12 ---> Making bundle: dynbinary (in bundles/1.4.1/dynbinary) 2015-01-26 14:06:12 go build runtime/cgo: no buildable Go source files in /usr/lib/go/src/pkg/runtime/cgo 2015-01-26 14:06:12 >>> ERROR: docker: all failed 2015-01-26 14:06:30 >>> docker: docker.init-ulimit-fix.patch 2015-01-26 14:06:31 patching file contrib/init/openrc/docker.initd 2015-01-26 14:06:31 # WARNING! I don't seem to be running in the Docker container. 2015-01-26 14:06:31 # The result of this command might be an incorrect build, and will not be 2015-01-26 14:06:31 # officially supported. 2015-01-26 14:06:31 # 2015-01-26 14:06:33 # Try this instead: make all 2015-01-26 14:06:35 # 2015-01-26 14:06:37 ---> Making bundle: dynbinary (in bundles/1.4.1/dynbinary) 2015-01-26 14:06:39 go build runtime/cgo: no buildable Go source files in /usr/lib/go/src/pkg/runtime/cgo 2015-01-26 14:06:41 >>> ERROR: docker: all failed 2015-01-26 14:06:45 hum? 2015-01-26 14:06:49 it builds here 2015-01-26 14:07:27 I mean, I installed the resulting package like 10 minutes ago 2015-01-26 14:07:42 maybe it builtds because it is in a lxc container 2015-01-26 14:07:52 and it maybe thinks it runs in a docker container due to that 2015-01-26 14:08:06 it doesn't 2015-01-26 14:08:12 I've got the same warnings 2015-01-26 14:08:30 every distro gets it, I remember it from Arch and Fedora 2015-01-26 14:08:58 http://sprunge.us/QaZE 2015-01-26 14:09:59 ccache 2015-01-26 14:10:06 unset CC makes it build 2015-01-26 14:12:58 ncopa-desktop:~/aports/main/docker$ sudo docker run --rm busybox echo foobar 2015-01-26 14:12:58 foobar 2015-01-26 14:13:00 ok it runs 2015-01-26 14:13:23 barthalion: feel free to push it as is 2015-01-26 14:13:49 kk 2015-01-26 14:13:53 actually 2015-01-26 14:13:56 add unset CC 2015-01-26 14:20:19 huh, in the mean time i found out that probably the libseccomp sandboxes will not quite work with libseccomp. :/ as musl uses different syscalls than glibc. so the qemu sandbox might not work either. but i wasnt able to test this yet, as i had not enough diskspace on my build vm 2015-01-26 14:21:43 ok, here comes mariadb 2015-01-26 14:21:51 big thanks to scadu 2015-01-26 19:02:11 ncopa: you're welcome. 2015-01-27 08:41:48 morning 2015-01-27 15:02:26 clandmeter: git log says you added mailx from ftp://ftp.archlinux.org/other/mailx/ but i cannot find any arch linux build scripts 2015-01-27 15:02:44 the url points to the posix spec http://pubs.opengroup.org/onlinepubs/007908799/xcu/mailx.html 2015-01-27 15:02:56 which says nothing about where this mailx implementation comes from 2015-01-27 15:03:12 i have a feeling it is some variant of bsd mailx 2015-01-27 15:03:32 the arch linux tarball says -fixed.tar.gz 2015-01-27 15:03:42 but says nothing about how or what it was fixed with 2015-01-27 15:04:07 the issue at hand is that there are some CVE that affects mailx 2015-01-27 15:04:14 http://seclists.org/oss-sec/2014/q4/1066 2015-01-27 15:04:18 and the patches does not apply 2015-01-27 15:04:27 so i was thinking of doing update 2015-01-27 15:04:34 but i have not idea how to do that 2015-01-27 15:05:14 we could look at using bsd-mailx from debian or ubuntu 2015-01-27 15:05:40 but those appear to re-introduce the liblockfile dependency (which i assume the arch linux variant got rid of) 2015-01-27 15:05:49 any ideas how to deal with this? 2015-01-27 15:06:04 i cannot remember anymore 2015-01-27 15:06:12 must be a long time ago 2015-01-27 15:06:18 2011 something :) 2015-01-27 15:06:41 wasnt there talk about it here some time ago? 2015-01-27 15:06:51 oh that might be 2015-01-27 15:07:10 http://lists.alpinelinux.org/alpine-devel/4104.html 2015-01-27 15:12:08 couldnt find anything usefull in backlog 2015-01-27 15:14:38 ncopa: s-nail seems activly maintained 2015-01-27 15:29:18 yes 2015-01-27 15:29:24 that looks like the best option 2015-01-27 17:17:13 Is this expected to be finished by 3.2? 2015-01-27 17:17:13 https://bugs.alpinelinux.org/issues/1008 2015-01-27 17:17:32 It got some delays in the past years, but I'm pretty excited about it 2015-01-27 17:17:37 Hope it makes it on 3.2 2015-01-27 21:12:52 is there any way to use nfnl_osf on alpine-linux? 2015-01-27 21:33:34 xen_roger1: nfnl_osf didn't exits in any package half a year ago... i bet that didn't change until now 2015-01-27 21:35:22 damn ... can't get it ... any idea how to build ... :( 2015-01-27 21:35:50 jomat: could you help? 2015-01-27 21:42:56 xen_roger1: Im too busy these months, sorry :-( 2015-01-27 21:43:30 jomat: ty anyway ... hope to get it working 2015-01-28 06:15:25 ncopa: you forgot to rebuild qt5-qtbase against mariadb :> 2015-01-28 06:54:18 the fact that it pulls mysql is stupid 2015-01-28 07:10:16 barthalion: ncopa said that he'll rebuild qt5-qtbase. 2015-01-28 07:11:10 what's the difference? it'll pull mariadb instead 2015-01-28 07:11:14 still stupid 2015-01-28 07:13:30 I realized that few seconds before your answer. hm, Arch has only libmariadbclient in makedepends. inb4 you don't say 2015-01-28 07:18:14 i will fix it 2015-01-28 07:18:33 and upgrade qt in the same time 2015-01-28 07:19:22 fabled: net-v2-gre-fix-the-inner-mac-header-in-nbma-tunnel-xmit-path.patch is included in 3.14.30? 2015-01-28 07:19:27 yes 2015-01-28 07:19:30 thanks 2015-01-28 08:14:23 scadu: i am slightly afraid of pushing qt5 5.4 right now 2015-01-28 08:14:35 i woudl like to push v3.1.2 release 2015-01-28 08:15:16 and qt upgrade would keep the arm builder busy for long time 2015-01-28 08:15:28 it is already busy with webkit since yesterday 2015-01-28 09:07:04 ncopa: what's wrong with 5.4? 2015-01-28 09:08:12 scadu: nothing 2015-01-28 09:08:27 the problem is that i am trying to get v3.1.2 out and arm builder are busy building 2015-01-28 09:08:37 arm is slow 2015-01-28 09:08:52 i push fixes and new kernel releases 2015-01-28 09:09:06 and before i reach to tag new release, new issues and new kernel releases comes in 2015-01-28 09:09:37 if i push qt 5.4 it will likely keep the arm builder busy for another day 2015-01-28 09:10:13 actually, the real problem is that both edge and v3.1 builders are using same wandboard hardware 2015-01-28 09:10:29 i have new hardware here but not had time to set it up 2015-01-28 09:10:40 I thought there's something besides. at least you know how I feel building someting bigger ;f 2015-01-28 09:10:52 yeah... 2015-01-28 09:10:55 something* 2015-01-28 09:11:15 arm v3.1 builder was offline for a while due to reboot 2015-01-28 09:11:35 so it didnt get the webkit update 2015-01-28 09:11:47 it has been building webkit since yesterday 2015-01-28 09:13:12 oh 2015-01-28 09:13:24 arm edge is likely building openjdk too... 2015-01-28 09:13:47 lol 2015-01-28 09:15:10 no it isnt 2015-01-28 09:16:15 load average: 3.98, 3.94, 3.95 2015-01-28 09:16:25 yeah, java is not enabled on arm currently. it builds, but produces TEXTRELs 2015-01-28 09:16:30 ironic 2015-01-28 09:16:50 they added a new JVM mode 2015-01-28 09:16:58 called ZeroVM 2015-01-28 09:17:06 because it was supposed to have zero assembly, and be portable 2015-01-28 09:17:12 but TEXTRELs come from arm assembly 2015-01-28 09:17:12 i need coffe 2015-01-28 09:17:18 lol 2015-01-28 09:32:51 clandmeter, http://bugs.alpinelinux.org/issues/3680 is a gcc compiler bug 2015-01-28 09:34:09 wow 2015-01-28 10:31:18 fabled: ah, bug still present? 2015-01-28 10:49:17 yes, i got a small test case now 2015-01-28 10:49:24 i'm filing gcc bug soon 2015-01-28 10:51:29 virt-grsec flavour on its way 2015-01-28 10:52:00 rnalrd: note that i replace the copy of the patch with a symlink 2015-01-28 10:52:25 yeah, noticed 2015-01-28 10:54:37 ah 2015-01-28 10:54:41 found matching bug 2015-01-28 10:54:43 it's caused by https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64037 2015-01-28 10:55:00 triggered by -Os 2015-01-28 10:55:54 fixed for 4.9.3 2015-01-28 10:56:07 which is not released yet 2015-01-28 10:59:19 updated ticket 2015-01-28 11:02:21 ncopa, or you think we should cherry-pick the compiler fix ? 2015-01-28 11:02:37 could fix the ada description someone mentioned on alpine-devel list also 2015-01-28 11:03:46 um 2015-01-28 11:03:51 might be worth fixing it 2015-01-28 11:03:58 we compile all with -Os 2015-01-28 11:04:18 Unless some blocker bug is found, GCC 4.9.3 should be released 2015-01-28 11:04:19 in March or April. 2015-01-28 11:04:31 yeah i think we should cherry-pick it 2015-01-28 11:04:31 ~2-3 months 2015-01-28 11:04:34 ok 2015-01-28 11:04:36 i'll do that 2015-01-28 11:04:50 but would probably be nice to do that after v3.1.2 2015-01-28 11:04:57 to not put more preasure on arm builder 2015-01-28 11:05:12 hum 2015-01-28 11:05:18 maybe doesnt matter 2015-01-28 11:05:42 arm has load-based make limits 2015-01-28 11:05:44 it should be ok 2015-01-28 11:05:47 ok 2015-01-28 11:06:07 lunch here 2015-01-28 12:16:51 hey guys, I have recently started working on a standalone fortify source implementation (to be used with musl primarily but it works on other systems as well) - it is very bare bones at the moment so I am looking for some feedback: http://git.2f30.org/libfort/tree/ 2015-01-28 12:17:03 might be useful for alpine linux as well :) 2015-01-28 12:18:58 stateless, yes, we are interested on it 2015-01-28 12:20:34 though, i would hope it gets built into musl.so, or potentially to do it as .a 2015-01-28 12:20:43 depending on how many functions need to be overrided 2015-01-28 12:20:53 making it a separate thin .so sounds redundant 2015-01-28 12:20:59 or a littlebit bloat 2015-01-28 12:27:19 fabled: dalias suggested putting everything in the .h 2015-01-28 12:27:27 I have not investigated that yet 2015-01-28 12:27:35 stateless, i was about to suggest the same 2015-01-28 12:27:55 I will have a look at that 2015-01-28 12:29:51 i wonder if dprintf()+__builtin_trap could be exported from musl as some sort of "__abort_with_error" 2015-01-28 12:30:13 perhaps, I am not happy with dprintf() either 2015-01-28 12:30:30 if inlined fully in header, that'd made it a little bit size efficient 2015-01-28 12:31:01 I also want to find a way to generating compile time warnings for strlcpy/strlcat, I guess there is a gcc attribute that can be used for that purpose 2015-01-28 12:31:59 yes 2015-01-28 12:32:15 there's builtins to check if an expression is build time constant 2015-01-28 12:32:23 yeah I am using those at the moment 2015-01-28 12:32:41 ah, yes 2015-01-28 12:32:46 for strlc* 2015-01-28 12:32:48 yeah 2015-01-28 12:33:23 I also want to generate a warning when n > bos though (at compile time) - at the moment for strl* it will only be caught at runtime 2015-01-28 12:34:04 I am also not sure if naming the original strlcpy, as strlcpy_orig() is acceptable 2015-01-28 12:34:10 from an app point of view 2015-01-28 13:21:22 fabled: thx, will try when i get home! 2015-01-28 13:23:53 ncopa: haven't we talked about making linux-virt-grsec a subpackage of linux-grsec once? 2015-01-28 13:27:10 that would be nice yes 2015-01-28 13:28:35 hups 2015-01-28 13:29:08 that was wrong version :-( 2015-01-28 13:57:22 ncopa: would be nice if the kernels all got upgraded in onego 2015-01-28 14:10:12 fabled: after a quick investigation, I found out that it is possible to put it in the .h but you can no longer use __builtin___foo_chk as those assume the symbol __foo_chk has external visibility (as opposed to using static inline __foo_chk in the header) 2015-01-28 14:10:49 so this works for strl* fine but not for memcpy let's say (it is possible to implement the memcpy checks the same way it is done for strl*) 2015-01-28 14:11:12 should I go down that route? 2015-01-28 14:13:27 stateless, can compiler generate references to __foo_chk if FORTIFY is set by itself? e.g. from the builtin memcpy() ? 2015-01-28 14:20:46 fabled: I think so, I mean you call __builtin___foo_chk(), it will eventually call the runtime version as well __foo_chk() if it can determine the size of dst/src/size 2015-01-28 14:21:16 then header-only implementation seems impossible 2015-01-28 14:21:29 __builtin___foo_chk() will also spits out compile time info 2015-01-28 14:21:36 compile time warnings I mean 2015-01-28 14:22:12 but you can reimplement your own __builtin___foo_chk() using __builtin_object_size() and __builtin_constant_p() 2015-01-28 14:23:26 when I do that, the only issue is: warning: ‘__memcpy_chk’ is static but used in inline function ‘memcpy’ which is not static 2015-01-28 14:23:44 memcpy() cannot be static as that will conflict with #include_next which is declared as non-static 2015-01-28 14:26:27 stateless, using #define would work. but not sure if that is advisable 2015-01-28 14:36:51 fabled: http://sprunge.us/KXQa 2015-01-28 14:36:54 that works :) 2015-01-28 14:37:49 will remove the _orig stuff, not needed anymore 2015-01-28 14:37:53 you might want to check with dalias too if he has more comments on it 2015-01-28 14:38:06 yeah just talking to him 2015-01-28 14:38:21 the _orig stuff is not needed anymore because the macro is undefined at that point :) 2015-01-28 14:58:00 fabled: i agree re: #3280 - but i don't understand the last part re: apk-tools should be fixed to support capabilities. 2015-01-28 14:58:27 tvvdhSL5hEBn, capabilities are xattrs 2015-01-28 14:58:35 meta-data for inodes 2015-01-28 14:58:38 yes 2015-01-28 14:58:45 but what about apk-tools? 2015-01-28 14:58:47 apk-tools does not currently preserve it 2015-01-28 14:58:51 aah 2015-01-28 14:58:58 that's something i want for alpine 3.2 2015-01-28 14:59:08 gotit 2015-01-28 14:59:10 to get PaX flags as xattrs, and capabilities xattr preserved 2015-01-28 14:59:15 very cool! 2015-01-28 14:59:22 i wrote the hard bits already 2015-01-28 14:59:37 support for "pax tar" (not to be confused with PaX) header 2015-01-28 15:00:41 *reading pax(P)* 2015-01-28 15:01:01 i just need to write the code to write that data to FS (simple), and have apk audit check it also (requires xattrs or hash of xattrs to be stored in apk's file db) 2015-01-28 15:01:12 so it's kinda half-way implemented 2015-01-28 15:05:52 i can confirm the proposed fix does work since back then. i can access my smartcards easily 2015-01-28 15:07:39 tvvdhSL5hEBn, nice 2015-01-28 15:07:51 i actually had the same issue and ended up running pcsc as root for the timebeing 2015-01-28 15:11:37 i don't like processes running as root :) 2015-01-28 15:18:44 fabled: http://git.2f30.org/fortify/tree/include/string.h :) 2015-01-28 15:19:05 not tested thoroughly yet but the overall structure should be ok now :) 2015-01-28 15:19:13 interesting 2015-01-28 15:19:24 (some checks may be enhanced I suspect) 2015-01-28 15:19:31 i'll look at it more tomorrow. i'm off for today. 2015-01-28 15:19:37 ok :) thanks for your input fabled 2015-01-28 17:14:08 ncopa: any progress with fake docker image? 2015-01-29 06:49:24 hmmh 2015-01-29 06:49:34 Assembler messages: 2015-01-29 06:49:34 Fatal error: can't create output/pjnath-i486-alpine-linux-uclibc/errno.o: No such file or directory 2015-01-29 06:49:34 /home/buildozer/aports/main/asterisk/src/asterisk-11.15.1/res/pjproject/build/rules.mak:135: recipe for target 'output/pjnath-i486-alpine-linux-uclibc/errno.o' failed 2015-01-29 06:49:39 is that like... disk full? 2015-01-29 07:59:43 fabled: where was that? 2015-01-29 08:00:04 ncopa, build-2-7-x86: failed to build v2.7.9-170-g277af68: http://bld1.alpinelinux.org/buildlogs/build-2-7-x86.v2.7.9-170-g277af68.log 2015-01-29 08:36:55 there is a bug in alpine-iso's Makefile 2015-01-29 08:37:26 fabled: mpd working nicely now. Thx! 2015-01-29 08:37:36 APK_OPTS := should also have "--arch $(ALPINE_ARCH)", otherwise i can't make images for armhf on x86_64 :s 2015-01-29 08:38:01 cuz it will always use the default, not the one specified in $(PROFILE).conf.mk 2015-01-29 08:38:21 i'll open pull req for it if it's ok 2015-01-29 08:39:16 ah it's not on shithub, also good :P 2015-01-29 08:39:22 could anyone fix it? 2015-01-29 08:43:31 fbettag, can you send git commit via email or pastebin? 2015-01-29 08:50:27 fabled: i think so, i will first try to get my img to finish 2015-01-29 08:50:56 currently initramfs breaks somehow, but i don't want to send a patch for something that might be my mistake ;) 2015-01-29 08:50:58 ok. we native build the images, but if cross-building the images can bedone simply, we'll be happy to add the patch. 2015-01-29 08:51:03 so i'm gonna follow through and let you know 2015-01-29 08:51:44 fabled: well for building x86 packages from x86_64 should work fine i think 2015-01-29 08:51:50 s/packages/imgs/ 2015-01-29 08:52:18 rpi i'm about to find out, but that answers my question as to wether you guys build the imgs on the rpis or do crossbuilds, so thanks =) 2015-01-29 08:52:48 nice 2015-01-29 08:53:03 fbettag, all arm building is done on the wandboard builder 2015-01-29 08:53:09 so it's arm building it, but not a rpi 2015-01-29 08:53:32 ah okay, i would've come to the chicken/egg question otherwise ;) 2015-01-29 08:54:21 the initial build was cross-compiled from x86_64 to bootstrap 2015-01-29 08:54:27 and the boot image was hand built 2015-01-29 08:55:02 kk 2015-01-29 08:58:05 fabled: http://el8.nl/~fbettag/alpine_rpi_img.png this is where i'm at 2015-01-29 08:58:22 gonna check some more later why it's trying to use x86_64 stuff there for initramfs 2015-01-29 09:02:02 u see two problems in that 2015-01-29 09:02:22 1) initramfs building seems to use lddtree which does not work cross-platform 2015-01-29 09:02:41 ncopa, would know the details. i think we would need to use readelf or scanelf instead 2015-01-29 09:03:10 2) it seems to also take host's initramfs file list 2015-01-29 09:03:25 i also wonder why it's taking linux-grsec and not the linux-rpi kernel 2015-01-29 09:07:25 that's my config probably 2015-01-29 09:07:41 but why is there a grsec kernel in the armhf repo then? :> 2015-01-29 09:09:36 fbettag, it's for armv7 boxes 2015-01-29 09:09:50 wandboard, cubieboard, etc. 2015-01-29 09:09:50 damn 2015-01-29 09:10:00 i was hoping to be flaunting about my homeautomating using grsec 2015-01-29 09:10:01 it will not work on rpi currently 2015-01-29 09:10:11 :P 2015-01-29 09:10:21 i think there was some rpi support added to mainline earlier 2015-01-29 09:10:25 any cheap armv7 boxes you can recommend for home automation? 2015-01-29 09:10:40 depends what it needs to do 2015-01-29 09:10:47 there's plenty of alternatives 2015-01-29 09:10:48 a few usb ports and gpio 2015-01-29 09:11:13 the only arm devices i really know are rpis (cuz they're easy to get here) 2015-01-29 09:11:32 we are running alpine on rpi and wandboard only currently 2015-01-29 09:11:39 supporting others should not be too complicated 2015-01-29 09:11:45 but would probably require some work 2015-01-29 09:12:10 you can check http://en.wikipedia.org/wiki/Comparison_of_single-board_computers for a list 2015-01-29 09:16:18 fbettag, rpi has the slowest arm core, the gpu is quite fast though 2015-01-29 09:16:40 if you want general purpose thing, it's probably good idea to get armv7 device 2015-01-29 09:17:04 then again, if it's just very dumb gpio unit, then it doesn't make a big difference 2015-01-29 09:19:20 fabian_a: hi. i've been looking your alpine-rescue custom iso. do you still maintain it? 2015-01-29 09:52:12 fabled: thanks! 2015-01-29 09:55:40 fbettag: did you check the new odroid C1? 2015-01-29 09:56:11 http://www.pollin.de/shop/dt/MTI2OTgxOTk-/Bausaetze_Module/Entwicklerboards/ODROID_C1_Einplatinen_Computer_1_5_GHz_QuadCore_4x_USB_1_GB.html 2015-01-29 09:57:07 clandmeter: uuuuuh very nice 2015-01-29 09:57:09 thanks! 2015-01-29 10:47:59 fabled: that asterisk issue looks like parallel build issue 2015-01-29 11:17:08 clandmeter, ODROID-C1 looks nice 2015-01-29 11:17:17 and cheap too :) 2015-01-29 11:26:51 fabled: yeah, already ordered one 2015-01-29 11:39:40 i'm gonna try running alpine on an omap ti. i have a kernel ready for exactly this hw. what obstacles might i run into, and do you have any good resources to prepare for what i want to do? 2015-01-29 11:40:19 fabled: could your recent gcc commit breaking building kernel modules? 2015-01-29 11:45:20 clandmeter, should not. it should affect only g++ program using some C++11 features 2015-01-29 11:45:49 i cannot build spl-grsec in testing 2015-01-29 11:47:08 can you paste the error? 2015-01-29 11:47:21 could be some change in recent kernel upgrade too 2015-01-29 11:47:38 http://sprunge.us/HHWE 2015-01-29 11:47:42 config.lo 2015-01-29 11:47:43 g 2015-01-29 11:48:01 its probably me who is doing soething wrong 2015-01-29 11:48:10 i can only see: code model kernel does not support PIC mode 2015-01-29 11:49:59 sounds like broken configure script 2015-01-29 11:50:12 it tries to use CFLAGS with the kernel compiler command 2015-01-29 11:50:17 could be also APKBUILD issue 2015-01-29 11:50:30 clandmeter, try unsetting CFLAGS 2015-01-29 11:51:01 i had also a weird error yesterday, had to unset ldflags to get it build. 2015-01-29 11:52:12 fabled: weird thing is, the latest build for spl is yesterday 2015-01-29 11:52:20 hmmh 2015-01-29 11:55:17 it fails in configure? 2015-01-29 11:55:43 it builds here 2015-01-29 11:55:52 on x86_64 2015-01-29 11:56:22 yes 2015-01-29 11:56:55 freshly updated edge x86_64, and spl-grsec builds ok 2015-01-29 11:57:13 http://sprunge.us/NIFc 2015-01-29 12:52:50 clandmeter, dependencies are installed? 2015-01-29 12:53:08 clandmeter: apk upgrade? 2015-01-29 12:53:22 yes did all of that. 2015-01-29 12:53:39 im thinking of setting up a new container. 2015-01-29 12:54:10 well, a new lxc hosts with new containers. 2015-01-29 12:57:13 seems btrfs wasnt a good choice as lxc storage. 2015-01-29 12:59:57 lvm works well for me as a backend 2015-01-29 13:01:03 lvm is nice, but i dont like resizing. 2015-01-29 13:01:13 zfs is nicer 2015-01-29 13:01:15 (part from those error messages when lxc-create does lvcreate) 2015-01-29 13:01:41 i think we already had this conversation :) 2015-01-29 13:01:56 :) 2015-01-29 13:02:04 not sure zfs will performe that well though 2015-01-29 13:02:17 can you shrink a zpool now? 2015-01-29 13:02:51 i dont know 2015-01-29 13:42:33 why do i need to be root to run make PROFILE=alpine-rescue iso? 2015-01-29 13:48:07 (1/12) Installing musl (1.1.5-r2) 2015-01-29 13:48:07 ERROR: Failed to set ownership on lib/libc.musl-x86_64.so.1.apk-new: Operation not permitted 2015-01-29 13:48:17 this is when running make iso 2015-01-29 14:00:01 hehe 2015-01-29 14:00:05 ncopa: ^ 2015-01-29 14:00:22 ScrumpyJack: i just had the same issue 2015-01-29 14:00:35 you can use fakeroot 2015-01-29 14:09:26 clandmeter: are you running make in a git clone of alpine-iso? 2015-01-29 14:11:34 ScrumpyJack: yes 2015-01-29 14:13:13 ncopa said it was always needing (fake)root, but i couldnt rememeber i ever needed/used it. 2015-01-29 14:13:16 would it be preferable that the Makefile used a fakeroot? 2015-01-29 14:14:21 or aleast warn about it 2015-01-29 14:16:48 what's your temporary solution? 2015-01-29 14:27:37 doesnt fakeroot make PROFILE=alpine-rescue iso work? 2015-01-29 15:13:50 clandmeter: fakeroot worked 2015-01-29 15:23:13 Is there something like https://github.com/bruenig/rankmirrors for Alpine Linux? 2015-01-29 15:34:23 rade: setup-apkrepos does something similar: http://git.alpinelinux.org/cgit/alpine-conf/tree/setup-apkrepos.in#n46 2015-01-29 15:56:38 ncopa, neat 2015-01-29 16:00:28 ncopa, any chance of us getting https://bugs.alpinelinux.org/issues/1008 soon? 2015-01-29 16:04:14 hum 2015-01-29 16:04:25 yeah it would be nice to fix that 2015-01-29 16:04:31 but not today... 2015-01-29 16:04:35 see you tm 2015-01-29 16:09:24 all right, no hurry 2015-01-29 16:38:50 clandmeter, you around ? 2015-01-29 18:00:07 fabian_a: ping 2015-01-29 21:02:29 now that mysql is replaced by mariadb, I'm not sure how properly to handle acf-mysql 2015-01-29 21:02:50 it's strange because the mysql package name and init.d script have changed to mariadb 2015-01-29 21:03:34 but the binaries, libs, and config are still 'mysql' 2015-01-29 21:04:53 considering changing acf-mysql package name to acf-mariadb, but not sure if I should change /mysql/mysql/* URL 2015-01-29 21:05:17 probably should leave that the same for backward compatibility 2015-01-29 21:28:20 hm 2015-01-29 21:28:32 error trying to install lua-sql-mysql on edige 2015-01-29 21:32:30 ncopa: http://sprunge.us/Qfbf 2015-01-29 21:33:30 not sure why lua-sql-mysql is trying to use mysql packages 2015-01-29 21:33:35 instead of mariadb 2015-01-29 21:35:02 hm, seems it was using the package from the boot media rather than from the edge repo 2015-01-29 21:35:06 sorry for false alarm 2015-01-29 21:43:38 still, there are some problems with mariadb 2015-01-29 21:44:56 the file /etc/init.d/mariadb contains a reference to (the non-existant) /etc/init.d/mysql file 2015-01-29 21:46:19 running '/etc/init.d/mariadb setup' fails: http://sprunge.us/YXGT 2015-01-29 21:59:26 not sure if this is a problem with the mariadb build that it can't find /usr/bin/my_print_defaults or that the /etc/init.d/mariadb init script needs to pass basedir 2015-01-29 23:13:25 ncopa: cloning alpine-iso, created something.package and something.conf.mk but make PROFILE=something iso doesn't contain my custom list of packages. 2015-01-29 23:13:58 also I can't boot a non-grsec iso build (boot: grsec not found) 2015-01-30 10:33:06 ncopa: is the pre-c version of apk available somewhere? 2015-01-30 10:33:43 uggedal: you mean the shell version? 2015-01-30 10:34:27 http://sourceforge.net/projects/apk-tools/?source=directory 2015-01-30 10:35:08 cool, the old website is still there: http://apk-tools.sourceforge.net/ 2015-01-30 10:42:50 in default_unpack() in abuild.in there is a switch statement that basically unpacks based on the type of archive, would it be possible to use gzip/bzip2/xz instead and pipe over to tar? I can cook up a patch later 2015-01-30 10:43:21 the reason I am asking is because I am using sbase to build stuff on Alpine linux (as a means to test sbase) and even though I did add support to unpack via tar using the z|j flags I'd like to discourage that behaviour 2015-01-30 10:43:28 (it uses external programs to handle that case) 2015-01-30 10:56:31 ncopa: thanks :) 2015-01-30 10:58:04 stateless: i'm ok with pipe 2015-01-30 11:34:55 ncopa: cool :) 2015-01-30 12:47:05 fabled: while you're at apk-tools 2015-01-30 12:47:20 could you please look into fixing apk audit --root ... 2015-01-30 12:47:28 ncopa, is there a ticket? 2015-01-30 12:47:34 no.. :) 2015-01-30 12:47:48 ok i'll file a ticket 2015-01-30 12:47:50 care to file one? :) 2015-01-30 12:48:42 #3840 2015-01-30 12:48:49 algitbot: hi 2015-01-30 12:48:52 #3840 2015-01-30 12:49:05 algitbot: please wake up 2015-01-30 12:49:24 i think i should punish algitbot 2015-01-30 12:49:33 spank him 2015-01-30 13:12:04 ncopa: have you seen '/etc/init.d/mariadb setup' work on edge? 2015-01-30 13:12:20 or am I the only one who has a problem with it? 2015-01-30 13:13:13 i havent teted i think i only tested use pre-existing mysql db 2015-01-30 13:13:38 so its probably broke for everyone 2015-01-30 13:14:14 http://sprunge.us/YXGT 2015-01-30 13:14:27 i need complete the v3.1.2 release first 2015-01-30 13:14:55 my_print_defaults is now installed by mariadb instead of mysql, but it's in the same location 2015-01-30 13:15:06 not sure why the setup script doesn't find it 2015-01-30 13:15:33 should I file a bug to remind you? 2015-01-30 13:16:08 tdtrask: yes please 2015-01-30 13:20:06 ok? http://wwwtest.alpinelinux.org/posts/Alpine-3.1.2-releases.html 2015-01-30 13:22:36 ncopa: #3842 2015-01-30 13:23:44 +1 for release notes 2015-01-30 13:25:15 ncopa: releasenotes look ok. 2015-01-30 13:25:42 thanks 2015-01-30 13:25:43 pushed 2015-01-30 13:28:23 ncopa: when you get a chance, can you please move acf-mariadb repo into cgit and set up the post-receive hook? 2015-01-30 13:39:24 should be done now 2015-01-30 13:43:10 ERROR: qt5-qtquick1-5.4.0-r0: trying to overwrite usr/bin/qmlviewer owned by qt-x11-4.8.6-r1. 2015-01-30 13:43:10 ERROR: qt5-qttools-5.4.0-r0: trying to overwrite usr/bin/qdbus owned by qt-4.8.6-r1. 2015-01-30 13:43:22 ncopa, ^^^ -> when trying to install both qt4 and qt5 2015-01-30 13:45:41 ncopa: grazie 2015-01-30 14:54:48 i want to compile alpine for armv7l, has anyone done similar efforts? or should i start with the armv6 port? 2015-01-30 14:55:20 maybe not all alpine, but a minimal system. 2015-01-30 15:03:47 tvvdhSL5hEBn, we ship arm build 2015-01-30 15:04:11 well 2015-01-30 15:04:26 you mean you want a full armv7+ optimized build? 2015-01-30 15:05:00 then yes, start with armv6 2015-01-30 15:05:25 armv7 can run armv6 just fine 2015-01-30 15:06:17 i actually was considering to build 'armv7' with the proper flags; perhaps in thumb2 mode 2015-01-30 15:06:23 but have not gotten there yet 2015-01-30 15:06:39 maybe you can reply to the mailing list mail thread regarding ARM build? 2015-01-30 15:06:45 i'm off for the weekend 2015-01-30 15:07:01 hmmm. i'm not (maybe yet) on the mailinglist 2015-01-30 15:07:15 will try the arm6 build first. thanks! have a nice weekend! 2015-01-30 15:07:36 http://lists.alpinelinux.org/alpine-devel/4674.html 2015-01-30 15:07:42 for the mail thread start 2015-01-30 15:08:09 thanks a lot! 2015-01-30 15:08:37 i run the arm6 build on my wandboard (armv7) 2015-01-30 15:08:45 the linux-grsec kernel is built for armv7+ 2015-01-30 15:09:03 and the userland is armv6+ to support rpi (it has it's own linux-rpi kernel build) 2015-01-30 15:58:59 ACTION is trying to see how upgrading from 3.1 to edge will cause mysql to be replaced by mariadb 2015-01-30 15:59:02 doesn't seem to work 2015-01-30 15:59:24 apk upgrade -U doesn't do anything :( 2015-01-30 16:32:47 ACTION guesses that there will have to be a manual step to uninstall mysql and install mariadb when upgrading to Alpine Linux 3.2 2015-01-30 17:23:14 ACTION created #3843 for upgrading mysql to mariadb 2015-01-31 21:30:27 maybe you've noticed already ? 'ERROR: linux-grsec-3.14.30-r0: lib/modules/3.14.30-0-grsec/kernel/drivers/net/ethernethp/hp100e.ko: no dirent in archive' http://hastebin.com/isagukuseb.avrasm 2015-01-31 21:30:46 from /sbin/setup-disk