2015-05-01 04:38:49 woo, got gcc 5.1 working with all the patches 2015-05-01 05:13:13 Nemykal, cool 2015-05-01 05:13:35 we'll be happy to upgrade edge to gcc-5.1 soon after 3.2-stable branching 2015-05-01 05:17:59 fabled: nice. 2015-05-01 06:35:05 Nemykal, perhaps you could change the gcc changes/patch rebases to our mailing list? 2015-05-01 06:41:53 Nemykal: very nice! 2015-05-01 06:42:16 please wait til after v3.2 is release til apply those patches 2015-05-01 07:11:26 fabled: ncopa: thats the plan! I've been doing my own stuff, working on a distro based on arch and alpine... but I need to start upstreaming patches :D 2015-05-01 07:12:07 hoping to get more involved with actually using the bug tracker 2015-05-01 07:12:40 based on arch and alpine? 2015-05-01 07:12:50 what do you take from alpine and what to you take from arch? 2015-05-01 07:13:05 do you use glibc or musl? 2015-05-01 07:13:11 pacman or apk? 2015-05-01 07:13:49 ncopa: no weekend today? 2015-05-01 07:13:55 i have 2015-05-01 07:13:59 :) 2015-05-01 07:14:05 will leave soonish 2015-05-01 07:14:28 i know you concider us all pleasure not work ;) 2015-05-01 07:15:24 i have hard to separate those yes :) 2015-05-01 07:15:43 what is pleasure and what is work is difficult when you find pleasure doing your work :) 2015-05-01 07:16:17 true 2015-05-01 07:17:02 btw, when i move working things to main, will it still be included in 3.2? 2015-05-01 07:17:29 i guess so till its branched? 2015-05-01 07:18:24 yes 2015-05-01 07:18:44 if you want it available in v3.2, then it needs to be moved now 2015-05-01 07:19:15 when will you branch? 2015-05-01 07:19:33 the plan is: 1) make all packages build 2015-05-01 07:19:38 2) make rc1 2015-05-01 07:19:44 3) fix all bugs 2015-05-01 07:19:53 4) make rc2 2015-05-01 07:19:58 branch and tag v3.2 2015-05-01 07:20:01 where is the list? 2015-05-01 07:20:11 http://bld2.alpinelinux.org/buildlogs/build-3-2-x86_64.log 2015-05-01 07:20:17 i think there are some arm specific too 2015-05-01 07:20:32 nice. it's not missing many now 2015-05-01 07:20:36 yes, arm list is bigger. 2015-05-01 07:20:40 http://bld1.alpinelinux.org/buildlogs/build-3-2-x86.log 2015-05-01 07:20:42 i guess most of them are difficult ones 2015-05-01 07:20:44 i think for 3.1 we have it on continue-on-fail 2015-05-01 07:21:16 id like make it halt-on-fail for v3.2 2015-05-01 07:21:28 and those who does not build, we adjust arch=... 2015-05-01 07:21:31 pjproject is simple 2015-05-01 07:21:37 depend on the new speex 2015-05-01 07:21:42 speex was split recently 2015-05-01 07:21:47 speexdsp-dev 2015-05-01 07:21:59 i tried pjproject the other day 2015-05-01 07:22:05 there is a new version too 2015-05-01 07:22:10 and it broke on something else 2015-05-01 07:22:34 you want me to look at it? 2015-05-01 07:22:46 fabled: i dont concider your "simple" simple for me. 2015-05-01 07:22:49 would be nice 2015-05-01 07:23:12 ok 2015-05-01 07:23:14 test building now 2015-05-01 07:23:22 i'll upgrade while at it 2015-05-01 07:24:07 ncopa: was there a reason we didnt have GraphicsMagick? i remember we ones did. 2015-05-01 07:25:07 ncopa, pjsip 2.4 compiled on ttdev64 just fine 2015-05-01 07:25:16 any idea whay our issue was about? 2015-05-01 07:26:38 i'll just upgrade.. 2015-05-01 07:26:44 ok 2015-05-01 07:46:13 ncopa: Something like this. I figure I could make something nice for running in containers https://gist.github.com/Nemykal/ae533ed9d0b93aa69f17 2015-05-01 07:46:44 mostly it's just me making something i personally want to use 2015-05-01 07:48:23 Nemykal: nice! 2015-05-01 07:48:25 makes sense to me 2015-05-01 07:49:02 morning 2015-05-01 07:49:15 morning ScrumpyJack 2015-05-01 07:51:36 just running setup-alpine on edge 2015-05-01 07:52:26 what's the best way to only use a partition instead of the whole disk? 2015-05-01 07:52:26 setup-disk is probably broken 2015-05-01 07:52:48 depends on what you are doing 2015-05-01 07:52:54 yes, -L -uM is deprecated for sfdisk 2015-05-01 07:53:32 i've got two partitions, sda1 and sda2. I want to install alpine on sda1 :) 2015-05-01 07:53:51 during setup-alpine, selece "none" on disk 2015-05-01 07:54:02 after you mkfs.ext4 sda1 2015-05-01 07:54:09 mount sda1 /mnt 2015-05-01 07:54:19 and then you run: setup-disk /mnt 2015-05-01 07:54:39 ok cool. i'll try that 2015-05-01 07:55:31 there's been a lot of chat about setup-alpine and partitions, but i keep forgetting, and I have no browser at the moment :) 2015-05-01 07:55:53 i need to fix that 2015-05-01 07:56:01 we must fix it before v3.2 2015-05-01 07:56:13 oh btw 2015-05-01 07:56:20 the v3.2 repos can be used now 2015-05-01 07:56:36 clandmeter: re cgit checksum 2015-05-01 07:56:43 i think the tarball is generated on the fly 2015-05-01 07:57:01 so the tarball checksum will probably always change 2015-05-01 07:57:16 we should manually wget it to dev.a.o/archive/cgit/ 2015-05-01 07:57:40 i suppose we also could patch cgit to cache the generated tarballs 2015-05-01 07:59:26 ncopa: it was git's csum which didnt match 2015-05-01 07:59:35 oh, ok 2015-05-01 07:59:36 and thats just from their src 2015-05-01 07:59:40 but still strange 2015-05-01 07:59:43 yes 2015-05-01 07:59:47 yes, i saw the 3.2 repos appear on my mirror (which is why i got confused yesterday between 3.2 and edge) 2015-05-01 07:59:56 i searched for changes on that specific ver 2015-05-01 08:00:01 couldnt find anything 2015-05-01 08:00:13 ScrumpyJack: atm the v3.2 is a clone of edge repo 2015-05-01 08:00:27 ok i should probably go now 2015-05-01 08:00:31 see u monday 2015-05-01 08:03:10 have a good weekend 2015-05-01 08:13:43 hi 2015-05-01 08:14:29 hi 2015-05-01 08:54:00 can we move ifupdown to main? 2015-05-01 08:55:59 shafire: eveything confirmed working can be moved to main. 2015-05-01 08:56:19 fabled: does 3.2 builders take source from git.a.o? 2015-05-01 08:57:28 for me it is working 2015-05-01 09:13:55 shafire: ^ 2015-05-01 09:14:05 :-) 2015-05-01 11:20:08 new pkgs available, apr-1.5.2 2015-05-01 11:20:09 bcache-tools-v1.0.8 - https://github.com/g2p/bcache-tools.git 2015-05-01 11:31:29 fabled: still around? 2015-05-01 11:31:52 clandmeter, yes? 2015-05-01 11:33:20 can you fix building of eventlog? 2015-05-01 11:33:27 im not sure what its wrong 2015-05-01 11:33:33 maybe src is diff on 3.2? 2015-05-01 11:35:41 clandmeter, 3.2 builder follows git master 2015-05-01 11:35:44 until branching 2015-05-01 11:40:48 fabled: i mean distfiles location 2015-05-01 11:41:00 oh, i think it's separate 2015-05-01 11:41:10 sounds like upstream changed the tarball contents 2015-05-01 11:41:21 i updated src 2015-05-01 11:41:21 or it's cgit generated and has some gzip header different 2015-05-01 11:41:30 but i think it still has old src. 2015-05-01 11:41:34 but i cannto delete it 2015-05-01 11:41:39 cannot 2015-05-01 11:42:33 oh ok 2015-05-01 11:44:13 hmm 2015-05-01 11:44:28 it looks ok 2015-05-01 11:44:37 atleast i tried to delete it from git.a.o distfiles 2015-05-01 11:44:52 but that checksum is ok 2015-05-01 11:47:09 yes, checksum is ok 2015-05-01 11:47:11 i'll look at it 2015-05-01 11:53:04 and btrfs-progs-v4.0 released 2015-05-01 12:17:01 clandmeter, is it x86_64 only or also x86 ? 2015-05-01 12:19:47 it was bad download 2015-05-01 12:19:53 v3.2 seems to use different cache 2015-05-01 12:19:54 fabled: dont know. i only looked at 64 2015-05-01 12:20:02 yeah thats what i though 2015-05-01 12:20:16 but log says iet gets deleted, which i doubht 2015-05-01 12:20:19 ok 2015-05-01 12:20:24 on x86_64 only left is perl-mail-clamav 2015-05-01 12:20:28 yes, it does not delete it 2015-05-01 12:20:32 it's bug in abuild 2015-05-01 12:21:04 yes that should be the only one 2015-05-01 12:21:14 that build error didnt say me much 2015-05-01 12:21:43 looks weird 2015-05-01 12:22:07 its not used i think in aports 2015-05-01 12:22:53 i guess just move to unmaintained 2015-05-01 12:24:11 ncopa, 'abuild' does not delete properly the distfile in v3.2 when checksum failed 2015-05-01 12:24:24 and the download was partial on 3+ packages 2015-05-01 12:29:00 lame fails on x86, due to sse not enabled in toolchain, but the code trying to do sse stuff 2015-05-01 12:57:35 clandmeter, seems Inline::C is separate perl module now 2015-05-01 13:11:27 ok. x86_64 is now fully built it seem. 2015-05-02 07:24:20 ncopa, greetings, I need your help, let me know if you have a few minutes for me 2015-05-02 10:42:12 setup-alpine on edge seems to be broken, it breaks to start the hard disk partitioning, with an error of sfdisk 2015-05-02 10:42:38 any suggestions? 2015-05-02 10:42:40 it's already reported 2015-05-02 10:43:12 barthalion, any suggestions? 2015-05-02 10:43:40 no idea, I always install alpine from random livecd I've already downloaded 2015-05-02 10:45:31 barthalion, Since when is broken the setup-alpine? 2015-05-02 10:45:54 I would like to install edge 2015-05-02 10:46:31 oh, no idea 2015-05-02 10:46:44 I'd just use 3.1 iso and upgrade to edge afterwards 2015-05-02 10:46:44 oks 2015-05-02 10:47:22 ok, I'll try that way 2015-05-02 13:18:10 crow, I installed alpine 3.1 upgrade to 3.2 and then to edge and so did work blenc 2015-05-02 16:32:58 vitronic great :) 2015-05-03 06:46:26 Hi everyone. Is there a particular reason for not having /etc/nsswitch.conf in Alpine? 2015-05-03 06:49:02 I faced a wierd error in Java, when nsswitch.conf doesn't tell to use /etc/hosts before /etc/resolv.conf, Java fallbacks to /etc/resolv.conf and if `hostname` cannot be resolved Java fails with: 2015-05-03 06:49:02 java.net.UnknownHostException: `hostname`: unknown error 2015-05-03 06:50:37 Java got fixed once I've created /etc/nsswitch.conf and added the following line (copied from Ubuntu setup) there: 2015-05-03 06:50:37 hosts: files mdns4_minimal [NOTFOUND=return] dns mdns4 2015-05-03 08:33:54 Can anyone help me to solve a little problem 2015-05-03 08:34:05 with fwknop 2.6.6 2015-05-03 08:34:31 I upgraded from 2.6.5 and now it doesn't start as a daemon 2015-05-03 08:34:54 even the init script is in /etc/runlevels/default 2015-05-03 08:35:24 when I start it wit fwknopd command it starts 2015-05-03 08:36:19 after a warning which says /var/run/fwknop directory not exists and was succesfully created 2015-05-03 08:37:48 anyway in /var/log/messages there is no error in automatically starting the daemon 2015-05-03 08:38:09 but the daemon doesn't start 2015-05-03 08:44:14 ? 2015-05-03 08:55:34 frol: there is no nsswitch in AL because it doesn't use PAM. 2015-05-03 08:57:06 I found the reason...the init.d/fwknopd script starts with #!/sbin/openrc-run 2015-05-03 08:57:36 I think is supposed to start with #!/sbin/runscript 2015-05-03 08:59:50 can anyone please make the change in build scripts please...! 2015-05-03 08:59:53 giuliano77_ the edge is swithcing to openrc-run, stable is still old runscript 2015-05-03 09:00:01 so you are on edge? 2015-05-03 09:01:05 yes 2015-05-03 09:03:50 well: 57915 (13:55:20) (@ncopa) openrc-run is correct 2015-05-03 09:04:16 but the daemon doesn't start 2015-05-03 09:05:11 well thats other problem 2015-05-03 09:05:15 maybe it need some patches 2015-05-03 09:08:09 giuliano77_ with this commit its changed to openrc-run http://git.alpinelinux.org/cgit/aports/commit/?id=6bf81f56e8c0b362d9c5ed046d9a8cdb81d5d957 2015-05-03 09:08:15 I chenged the line and it is working with #!/sbin/runscript 2015-05-03 09:08:44 thank you 2015-05-03 09:08:51 I wil make the update 2015-05-03 09:12:32 what is the package name of that fwknop. 2015-05-03 09:12:46 fwknop-server 2015-05-03 09:12:58 is not in that commit 2015-05-03 09:13:32 i cannot find it in aports http://git.alpinelinux.org/cgit/amanison/aports/tree/ 2015-05-03 09:14:11 it has openrc-run in init.d script already 2015-05-03 09:14:26 but the script does nothing 2015-05-03 09:14:37 it doesn't start the daemon 2015-05-03 09:14:49 and it not trigger any error 2015-05-03 09:16:10 Actually I use main alpine distro and I have /alpine/edge/testing also in /atc/apk/repositories 2015-05-03 09:16:51 Should I change the distro to edge? 2015-05-03 09:20:53 i still can not see that package in this aports tree. where did you get it? 2015-05-03 09:21:41 duno what is your current stage, but i am not mixing repositories, on stable i use stalbe, and on edge i use edge 2015-05-03 09:22:32 you did repository pinning? http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management 2015-05-03 09:24:35 I didn't 2015-05-03 09:24:53 mixing repositories is not ok? 2015-05-03 09:25:31 I get it from /edge/testing 2015-05-03 09:25:46 is the only place where I found it 2015-05-03 09:26:22 I think I will stay to main and I will compile it myself 2015-05-03 09:26:51 and I will use sbin/runscript 2015-05-03 09:28:05 So mixing repositories is not a good practice...I think I should pin them and install just this package from @/edge/testing 2015-05-03 09:28:34 This is the only package which I need from edge repository 2015-05-03 09:34:43 giuliano77_ no mixing repositories is not a good practice. pinning is something support, but i dont use it either 2015-05-03 09:38:00 hi 2015-05-03 09:38:00 Solve it temporarly by pinning repository and changed the init script to runscript 2015-05-03 09:38:19 ifupdown gives me an error 2015-05-03 09:38:21 I will stay on main version 2015-05-03 09:38:29 # apk fix ifupdown (1/1) Reinstalling ifupdown (0.7.53.1-r0) Executing ifupdown-0.7.53.1-r0.post-upgrade busybox: 'busybox' is not an absolute path ERROR: ifupdown-0.7.53.1-r0.post-upgrade: script exited with error 1 Executing busybox-1.22.1-r15.trigger 1 errors; 145 MiB in 109 packages 2015-05-03 09:38:36 Can someone fix this please? 2015-05-03 09:38:37 http://git.alpinelinux.org/cgit/aports/tree/testing/ifupdown/ifupdown.post-upgrade?id=b32b52f051f062a695bbbcf58239eee6186abf1e 2015-05-03 09:38:39 and I will compile fwknop-server myself 2015-05-03 09:39:25 thank you for your advice crow 2015-05-03 09:39:46 I didn't know that mixxing repositories is not so OK 2015-05-03 09:40:09 I need that fwknop-server 2015-05-03 09:40:31 But I prefer to stay on main for now 2015-05-03 10:21:16 I'm not sure but I think bbsuid is broken in edge 2015-05-03 10:21:50 I can run su as root just fine, but as user I get "su: must be suid to work properly" 2015-05-03 10:22:49 same with other tools like passwd 2015-05-03 12:13:51 plueschopath: i'm on edge here and bbsuid has the correct rightsd 2015-05-03 14:35:15 we should write a big warning somewhere 2015-05-03 14:35:17 written in red 2015-05-03 14:35:53 "don't mix repositories, kthxbye" 2015-05-03 14:35:53 and blinking :v 2015-05-03 14:36:01 preferably in irc topic 2015-05-03 14:36:12 "ensure your repositories list is sane before you ask" 2015-05-03 16:32:09 hmm 2015-05-03 16:32:34 it's a new install ... 2015-05-03 16:33:13 the only repo in /etc/apk/repositories is: http://nl.alpinelinux.org/alpine/edge/main 2015-05-03 16:49:38 plueschopath: I wasn't complaining about you, sorry 2015-05-03 17:21:57 barthalion> "don't mix repositories, kthxbye" <- +1 :) 2015-05-03 17:22:33 barthalion: ah i see... I should just read the log next time XD 2015-05-03 20:09:51 barthalion: i think it would be better if ncopa had made it backwards compatible this openrc change. 2015-05-03 20:10:06 well, probably 2015-05-03 20:10:28 but it doesn't change the fact it's not the first time when someone has a problem due to mixed repositories 2015-05-03 20:10:49 sure, mixing means no support 2015-05-03 20:10:52 :) 2015-05-03 20:11:04 and not mixing eather :) 2015-05-04 06:00:49 clandmeter, fcolista: could someone add me some power in aport patchworks? 2015-05-04 06:33:50 main total built: 2009 2015-05-04 06:33:50 main total relevant aports: 2009 2015-05-04 06:33:58 looks like x86_64 builder is all done 2015-05-04 06:34:02 v3.2 2015-05-04 06:36:28 yes, i fixed it Fri 2015-05-04 06:37:01 x86 is almost fixed too 2015-05-04 06:37:08 seems farstream0.1 is left 2015-05-04 06:38:50 farstream looked like download fail 2015-05-04 06:38:55 retrying it 2015-05-04 06:48:42 done! 2015-05-04 06:48:44 wow 2015-05-04 06:48:45 cool 2015-05-04 06:49:05 x86 and x86_64 v3.2 builders are done! 2015-05-04 06:49:06 fabled: fixed 2015-05-04 06:50:40 clandmeter, how do change patch status? 2015-05-04 06:51:21 when you login it should be obvious 2015-05-04 06:51:35 you can use webif or pwclient 2015-05-04 06:51:52 ah 2015-05-04 06:51:54 now i see it 2015-05-04 06:51:58 needed relogin i suppose 2015-05-04 06:52:12 no i had to add you to project 2015-05-04 06:52:15 :) 2015-05-04 06:52:29 okay :) 2015-05-04 06:52:47 you can check pwclient here: http://patchwork.alpinelinux.org/project/aports 2015-05-04 06:52:53 or apk add pwclient 2015-05-04 06:53:22 don't forget to download http://patchwork.alpinelinux.org/project/aports/pwclientrc/ 2015-05-04 06:53:32 and edit it because it doesn 2015-05-04 06:53:34 argh 2015-05-04 06:53:39 doesn't set default project 2015-05-04 06:55:29 pwclient is pretty stupid 2015-05-04 06:55:52 but when you know how it works, it works just fine 2015-05-04 06:58:02 ncopa: is there a reason why zlib is in /lib? im seeting other disto's have it in /usr/lib 2015-05-04 07:01:31 probably some tool that are in /bin or /sbin needs it 2015-05-04 07:01:42 other distros requires that /usr is on same partition as / 2015-05-04 07:02:18 http://freedesktop.org/wiki/Software/systemd/separate-usr-is-broken/ 2015-05-04 07:03:02 "There is no way to reliably bring up a modern system with an empty /usr." 2015-05-04 07:04:31 this is probably bullshit 2015-05-04 07:04:51 its probably true for Fedora/redhat 2015-05-04 07:05:02 here too: http://www.freedesktop.org/wiki/Software/systemd/TheCaseForTheUsrMerge/ 2015-05-04 07:05:17 Myth #9: The /usr split is useful to have a minimal rescue system on the root file system, and the rest of the OS on /usr. 2015-05-04 07:05:17 Fact: On Fedora the root directory contains ~450MB already. This hasn't been minimal since a long time, and due to today's complex storage and networking technologies it's unrealistic to ever reduce this again. 2015-05-04 07:05:50 they are kinda arguing that it is not possible to make a minimal system with less that 450MB... 2015-05-04 07:12:43 :) 2015-05-04 07:14:13 ncopa: is nfs working in edge now? if not, do we have a related bug? 2015-05-04 07:14:37 i think it should work 2015-05-04 07:14:43 but i havent tested it really 2015-05-04 07:14:50 i suppose its time to do so 2015-05-04 07:15:04 it was working fine in stable, but when i tested it in edge it wasnt 2015-05-04 07:15:07 because i upgraded nfs 2015-05-04 07:15:25 had to remove multiple patches to let it pass 2015-05-04 07:16:05 i should realy upgrade my home 2.7 :) 2015-05-04 07:16:23 bad marketing 2015-05-04 07:16:58 could anyone help me with testing nfs in edge/v3.2? 2015-05-04 07:17:10 set up a vm with nfs server 2015-05-04 07:18:13 not sure i can. i ugess its still broken on vmware 2015-05-04 07:21:05 ok, I'll try do that later today then 2015-05-04 07:21:38 oh it works 2015-05-04 07:21:54 nice, we can close that bug for grsec on older vmware 2015-05-04 07:29:37 are there any 3.2 iso avaiable? last time like two weeks the "edge vanilla" would fail to ask for hostname and also could not create partition http://bugs.alpinelinux.org/issues/4072#change-11932 2015-05-04 07:31:29 i used workarround from clandmeter: ncopa just for the info: i used @stable repo pining but seemt setup-disk script adds and overwrites the installed sfdisk@stable: http://sprunge.us/ZETG 2015-05-04 07:32:06 crow: i will look at arm builders and start try fix those bugs 2015-05-04 07:32:17 i hope the setup-disk bug will be fixed this week 2015-05-04 07:51:32 why does nfs-utils pull in sqlite-libs... 2015-05-04 07:53:45 seems its broken 2015-05-04 07:58:20 bump, btrfs-progs-v4.0 and bcache-tools-v1.0.8 pls 2015-05-04 07:58:33 if possible 2015-05-04 08:27:33 ncopa: looks like rpc.statd has its background switch wrongly parsed 2015-05-04 08:27:48 it works vice-versa 2015-05-04 08:27:57 maybe they changed it recently? 2015-05-04 08:28:05 im using -h 2015-05-04 08:28:12 which should tell me the correct way i gyuess 2015-05-04 08:28:35 -F, --foreground Foreground (no-daemon mode) 2015-05-04 08:28:49 if i add it, then it goes the the background 2015-05-04 08:29:00 there is also https://github.com/facebook/watchman/ compared to http://patchwork.alpinelinux.org/patch/69/ 2015-05-04 08:49:35 ncopa, util-linux downgraded ? 2015-05-04 08:51:58 vkrishn: thats probably the easiest fix to setup-alpine :) 2015-05-04 08:53:59 or algibot having 'manic monday' 2015-05-04 08:56:39 and to add, apr-1.5.2 2015-05-04 08:57:35 nah, i think the fix is not that easy to implement in setup-disks. so for now its just easier to downgrade util-linux which holds cfdisk (or whatever its called). 2015-05-04 08:59:34 ok 2015-05-04 09:08:25 clandmeter: re statd: http://git.linux-nfs.org/?p=steved/nfs-utils.git;a=patch;h=17a3e5bffb7110d46de1bf42b64b90713ff5ea50 2015-05-04 09:08:37 they added fix upstream right after 1.3.2 release 2015-05-04 09:09:21 ah nice 2015-05-04 09:09:29 ncopa: we do not support v4? 2015-05-04 09:09:38 i now have to explicity specify v3 2015-05-04 09:09:51 clandmeter: i cannot see why we wouldnt support v4 2015-05-04 09:10:13 it says: mount.nfs: an incorrect mount option was specified 2015-05-04 09:11:43 seems like openbsd does not support v4 either 2015-05-04 09:11:52 http://openbsd-archive.7691.n7.nabble.com/nfsv4-td18690.html#a18712 2015-05-04 09:12:34 this stuff is weird 2015-05-04 09:12:48 if you dont specify version it wont mount 2015-05-04 09:13:09 but if you do, it doesnt matter which version you specify, it *will* mount 2015-05-04 09:13:19 but always as 3 2015-05-04 09:13:54 iirc, this behaviour has changed. i didnt need to specify version before. 2015-05-04 09:18:28 clandmeter, FYI: just rejected #73. md5sum was missing. 2015-05-04 09:18:41 patch n.73 2015-05-04 09:18:44 hi algitbot 2015-05-04 09:18:54 algitbot: i was not talking with you 2015-05-04 09:20:43 fcolista: we should setup a paypall account at address: donate_beer_when_patch_is_faulty@alpinelinux.org 2015-05-04 09:20:57 +1 2015-05-04 09:21:04 :) 2015-05-04 09:21:54 does patchwork send a mail to patch author about rejection? 2015-05-04 09:22:01 yes 2015-05-04 09:22:14 mkay 2015-05-04 09:23:10 sadly comments are not available yet explaining the reasons of rejection 2015-05-04 09:23:20 but it's a feature in their todo list 2015-05-04 09:25:16 you should send a mail to alpine-aports as well then 2015-05-04 09:25:57 yes, is suppose yes. 2015-05-04 09:27:57 same for Patch n.80 2015-05-04 09:28:17 feature request, algibot could listen to % 2015-05-04 09:28:28 well done vkrishn! 2015-05-04 09:28:43 %80 2015-05-04 09:28:52 algitbot, %80 2015-05-04 09:29:17 algitbot: wake up 2015-05-04 09:29:29 that was not magic wand request ! 2015-05-04 09:29:34 :D 2015-05-04 09:29:53 algitbot: stop sleeping! 2015-05-04 09:41:44 ncopa: nfsmount (idmapd) gives me: make sure DNOTIFY support is enabled ... 2015-05-04 09:44:13 hum 2015-05-04 09:44:23 didnt they add supoprt for inotify recently? 2015-05-04 09:44:40 maybe 2015-05-04 09:44:46 could be its just the init thats faulty 2015-05-04 10:36:51 there is rc releases , http://git.linux-nfs.org/?p=steved/nfs-utils.git;a=summary 2015-05-04 10:37:21 and found this, http://git.linux-nfs.org/?p=bfields/nfs4-acl-tools.git;a=summary 2015-05-04 11:10:05 still trying to get pNFS, http://git.linux-nfs.org/?p=bfields/linux.git;a=blob;f=Documentation/filesystems/nfs/pnfs-block-server.txt;h=2143673cf1544bfc18502b8e0e4ee469234e7aae;hb=c517d838eb7d07bbe9507871fab3931deccff539 2015-05-04 12:13:04 in order to debug a programm crash i compiled it with debug option, but now i should also set this "ulimit -c unlimited" but the user actually does not have shell. so how would i set this for an programm running as a deamon under user which does not have shell? 2015-05-04 12:16:14 clandmeter, can you please check why i'm not able to subscribe to alpine-aports ML ? 2015-05-04 12:16:39 no feedback received when sent email to alpine-aports+subscribe@lists.alpinelinux.org 2015-05-04 12:19:43 oh, i got a feedback now. 2015-05-04 12:23:58 crow: you can temporary change shell 2015-05-04 12:37:49 barthalion, clandmeter. I'm subscribed to alpine-aports, but i've no chance to reply to rejected patches without breaking the thread (if you have an hint on how to do that,i'm glad to hear). 2015-05-04 12:38:00 Would you mind to reply? 2015-05-04 12:38:24 Basically it's up to checksum that is removed. 2015-05-04 12:38:42 This is in the case of %73 2015-05-04 12:39:01 In case of %80 there's no updated checksum 2015-05-04 12:39:07 and patch does not apply, also 2015-05-04 12:39:31 %72 also has removed checksum 2015-05-04 12:39:36 and i'm going to reject it 2015-05-04 12:41:37 sending back a msg that checksum can be generated by `abuild fetch && abuild checksum` without doing full compile 2015-05-04 12:41:46 could help 2015-05-04 12:42:51 I don't think we should reject patches just like that 2015-05-04 12:43:08 I'd rather use "Changes requested" status 2015-05-04 12:43:19 and if there is no reply after week, then reject 2015-05-04 12:43:33 sounds reasonable. 2015-05-04 12:43:44 in my opinion it's our duty to merge the patch and fix things like checksums or merge conflict anyway 2015-05-04 12:43:54 these are not big things 2015-05-04 12:43:58 ok 2015-05-04 12:44:37 and people like to contribute to alpine because we're not bunch of smartasses that get angry if something is wrong with proposed patch 2015-05-04 12:45:40 yeah, you're right. That's not the spirit i would intended to promote for sure 2015-05-04 12:46:05 anyway, %72 has also the link broken 2015-05-04 12:46:13 fetch link i mean 2015-05-04 12:46:24 build/build-3-2-armhf 380/413 1639/1981 main/libcec 2.2.0-r0 2015-05-04 12:46:30 progress 2015-05-04 12:46:50 i suppose i should fix setup-disk 2015-05-04 12:46:54 then we can probably do rc1 2015-05-04 12:47:40 i can update the link and the checksum as well and apply patch. 2015-05-04 12:49:27 ncopa: didnt you fix it by dwongrading util-linux? 2015-05-04 12:50:13 oh 2015-05-04 12:50:19 i think that was unintentional 2015-05-04 12:50:36 the commit msg was a bit strange 2015-05-04 12:51:54 i think i did a typo 2015-05-04 12:51:58 24 instead of 26 2015-05-04 12:52:48 ok %72 is defenitly broken. No checksum, wrong link, and does not compile 'cause it has qmake now. APKBUILD needs other fixes. 2015-05-04 12:53:02 ah, manic monday! is over 2015-05-04 12:56:18 guys... there is one thing i'd like to say about those patches and patchwork 2015-05-04 12:56:27 big thank you! 2015-05-04 12:56:43 thank you for applying patches 2015-05-04 12:56:49 thank you for checking them 2015-05-04 12:57:01 than you for responding contributors 2015-05-04 12:57:05 thank* 2015-05-04 12:59:26 thank you! 2015-05-04 12:59:47 i still owe you 999 beers 2015-05-04 13:01:38 clandmeter, he prefers coffee 2015-05-04 13:01:52 maybe a nice can of coffee beans 2015-05-04 13:02:12 i do like beer ;) 2015-05-04 13:04:02 ncopa: it's not like we do even half of your work, so better thank yourself :p 2015-05-04 13:05:09 point is that i wouldnt be able to do it alone 2015-05-04 13:05:13 clandmeter and other way to do that, then changing shell? 2015-05-04 13:05:57 %72 rejected. 2015-05-04 13:15:47 fcolista: why did you reject it? 2015-05-04 13:16:16 which one ? 2015-05-04 13:16:42 72' 2015-05-04 13:16:54 <fcolista> ok %72 is defenitly broken. No checksum, wrong link, and does not compile 'cause it has qmake now. APKBUILD needs other fixes. 2015-05-04 13:17:24 but you didnt reply to ml? 2015-05-04 13:17:37 clandmeter, did you read the backlog? :) 2015-05-04 13:17:50 asked to you/barhalion to do it in my behalf 2015-05-04 13:18:09 'cause i've subscribed my account now to alpine-aports 2015-05-04 13:18:17 and i'm not able to reply without broken the thread 2015-05-04 13:18:32 i've asked for hint about it (if you know how to reply without branking the thread) 2015-05-04 13:18:32 i dont get it? 2015-05-04 13:18:37 could be :) 2015-05-04 13:18:53 am i missing something? 2015-05-04 13:19:21 I've asked to you or barthalion to reply to ML in my behalf. 2015-05-04 13:20:05 fcolista: i dont mind seding a mail on your behalf, i just dont get it why you cannot? 2015-05-04 13:20:43 dunno how to do it wihtout breaking the thread on mailing list, since i was not subscribed to the mailing list 2015-05-04 13:20:56 i can do it from patches from now on, since i got email. 2015-05-04 13:21:21 oh ok understand :) 2015-05-04 13:21:36 sorry for not being clear 2015-05-04 13:23:56 what the hell? 2015-05-04 13:25:01 ncopa, clandmeter ↑ remove the file from distfiles please 2015-05-04 13:25:50 owncloud-texteditor.tar.gz? 2015-05-04 13:26:21 clandmeter, thx for email reply to ml 2015-05-04 13:27:02 file deleted 2015-05-04 13:27:37 this starts to happen quite often now. 2015-05-04 13:27:46 how to restart the build process? 2015-05-04 13:28:16 ncopa: could we have a algitbot rebuild cmd again? 2015-05-04 13:28:22 or is it still here? 2015-05-04 13:29:13 ncopa, did we have kexec userland / kernel config yet? 2015-05-04 13:45:35 owncloud-texteditor.tar.gz: FAILED 2015-05-04 13:45:53 the content of that file changes all the time 2015-05-04 13:45:56 we cannot have it like that 2015-05-04 13:46:22 fabled: kexec? 2015-05-04 13:46:38 ncopa, reload new kernelimage without reboot 2015-05-04 13:46:51 https://www.kernel.org/pub/linux/utils/kernel/kexec/ 2015-05-04 13:46:52 i dont think we can do that with PAX 2015-05-04 13:47:02 hmm 2015-05-04 13:47:20 i havent checked but i would be surprised if that would work with pax 2015-05-04 13:47:54 i dont think we have the userspace tools either 2015-05-04 13:48:55 $pkgname-texteditor.tar.gz::https://github.com/owncloud/files_texteditor/archive/v$_texteditorver.tar.gz 2015-05-04 13:49:10 i think there should be a $pkgver in there too 2015-05-04 13:51:16 hmm.. apparently it depends on GRKERNSEC_KMEM 2015-05-04 13:52:43 yeah, 2015-05-04 13:52:43 config KEXEC 2015-05-04 13:52:43 bool "kexec system call" 2015-05-04 13:52:43 depends on !GRKERNSEC_KMEM 2015-05-04 13:53:06 do we want to support it on -vanilla and/or -rpi? 2015-05-04 13:53:24 um 2015-05-04 13:53:32 i think we want keep vanilla close to grsec config 2015-05-04 13:53:42 -rpi i dunno 2015-05-04 13:54:12 i suppose its an attack vector for rootkits 2015-05-04 13:54:39 hmm 2015-05-04 13:54:47 is it possible to switch_root instead of poweroff? 2015-05-04 14:06:02 fabled: i don tknow 2015-05-04 14:07:02 @ncopa you send your commit of change in the abuild to somewhere? 2015-05-04 14:08:08 ramonsoares: sorry? 2015-05-04 14:08:33 i dont follow 2015-05-04 14:09:30 commit about remove -doc and -dbg of -dev packages 2015-05-04 14:12:01 no 2015-05-04 14:12:16 i didnt send them anywhere and i dont think i committed them either 2015-05-04 14:12:18 not yet 2015-05-04 14:12:30 need to wait til after v3.2 is released 2015-05-04 14:12:31 ok 2015-05-04 14:12:57 but i think I'll apply that pretty soon after v3.2 is shipped 2015-05-04 14:15:44 Im creating a repository to use in docker containers and change some packages to make it small as possible 2015-05-04 14:19:43 i thought about -dev packages, and it should be relatively easy to make it optimal 2015-05-04 14:20:09 i think it should be enough to try resolve symlinks 2015-05-04 14:23:40 fabled, http://lkml.iu.edu/hypermail/linux/kernel/1504.2/03365.html 2015-05-04 14:45:00 will v3.2 updates have separate folder on repos ? 2015-05-04 14:45:14 ? 2015-05-04 14:45:31 now that we have kernel with unionfs 2015-05-04 14:45:40 we have separate folder for every stable release 2015-05-04 14:46:13 the idea is to preserve the version release in its entirety 2015-05-04 14:46:23 i.e v3.2.0 2015-05-04 14:46:56 and updates can happen in http://nl.alpinelinux.org/alpine/v3.2/main/updates/1 2015-05-04 14:47:14 or http://nl.alpinelinux.org/alpine/updates/v3.2/ 2015-05-04 14:47:56 the builders can be tricked to see as same folder by using unionfs 2015-05-04 14:50:23 vkrishn: you want that so you can roll back to earlier version if needed? 2015-05-04 14:51:04 that is one aspect, a versioned release also means it has to be there historically 2015-05-04 14:51:16 edge is rolling 2015-05-04 14:51:28 not "a versioned release" 2015-05-04 14:52:22 so I should be able to pkg v3.2.0 on a dvd its full release from the repo at any time 2015-05-04 14:52:34 or just release a dvd and archive it 2015-05-04 14:52:51 when there are updates 2015-05-04 14:53:30 ncopa: I just noticed haserl-0.9.35 is in testing (builds against Lua 5.3,5.2,5.1); but haserl-0.9.33 is in main 2015-05-04 14:54:05 the testing APKBUILD should be replace main, and testing can be deleted. 2015-05-04 14:54:48 http://lists.alpinelinux.org/alpine-aports/0242.html 2015-05-04 14:55:14 ok 2015-05-04 14:59:51 I mean tricked using "overlayfs" 2015-05-04 15:00:26 sounds hackish 2015-05-04 15:01:02 nangel: sorry, ididnt realize you sent a patch already 2015-05-04 15:01:44 there were 2 patches, well 3. My first one, (you applied), then your fix to my patch, then the final patch that I sent. 2015-05-04 15:01:47 it was confusing 2015-05-04 15:01:49 other method is using symlinks 2015-05-04 15:02:26 patchwork should make it easier to keep track of the patches 2015-05-04 15:03:35 And another tool for a PHB like me to get wrong. :) 2015-05-04 15:04:13 http://patchwork.alpinelinux.org/project/aports/list/ 2015-05-04 15:04:21 its mostly for us who applies the patches 2015-05-04 15:07:13 i suppose i need config my ~/.pwclientrc to point to patchwork.alpinelinux.org? 2015-05-04 15:07:15 cleaner method is, upload folder for builders is separate from nl.a.o, but this means replication 2015-05-04 15:15:14 since presently apk does not have auto logic to figure out update folders, one can add them to /etc/apk/repositories 2015-05-04 15:43:44 another extra step would be generating APKINDEX for updated pkgs only 2015-05-04 18:27:06 oscam just crashed, and i cant find dump. 2015-05-04 18:38:52 clandmeter, I sent some patches to aports ml, but i'd like to have this one applied sooner: http://sprunge.us/NZHG 2015-05-04 18:38:52 could you help me with this? (about the others that i sent to ml i can wait) 2015-05-04 18:39:26 mate-menu is breaking my Application menu 2015-05-04 18:49:52 so question is how to set ulimit to "ulimit -c unlimited" for an user which does not have login (is just for running oscam as daemon)? 2015-05-04 19:37:49 it seems i'll have to wait any way... 2015-05-04 19:37:50 :( 2015-05-04 22:56:24 alacerda: ^ 2015-05-05 05:38:16 ah!! why does http://thrysoee.dk/editline/ have such release num 2015-05-05 05:54:54 if this useful, http://www.tinycorelinux.net/6.x/x86_64/release/src/toolchain/bc-1.06.95-memory_leak-1.patch 2015-05-05 06:09:26 can it be applied if useful 2015-05-05 06:10:14 vkrishn: applied. thanks 2015-05-05 07:03:25 smplayer and smtube would be nice to see 2015-05-05 07:44:59 http://www.cdrkit.org/ not available 2015-05-05 07:45:32 I think debian maintains it 2015-05-05 08:43:28 seems active, http://cdrecord.org/ 2015-05-05 08:45:38 also pkged in http://sourceforge.net/projects/schilytools 2015-05-05 09:36:48 clandmeter, i've applied %88 but i had to do a minor change on the APKBUILD. 2015-05-05 09:37:01 It still looks as "New" in patchwork. 2015-05-05 09:37:15 this is what i've done: 2015-05-05 09:37:22 pwclient git-am 88 2015-05-05 09:37:29 git commit --amend -a 2015-05-05 09:37:34 make the change 2015-05-05 09:37:42 git pull --rebase && git push 2015-05-05 09:38:12 I think i can change the status to "Accepted" on webif. 2015-05-05 09:42:56 main/acct v6.6.2 available 2015-05-05 09:58:10 alacerda: do you think you could replace gtk+-dev with gtk+2.0-dev in your mate packages? 2015-05-05 09:58:16 i want remove gtk+ package 2015-05-05 11:10:05 ncopa, great... i'll do 2015-05-05 11:10:37 clandmeter, thanks for applying my patch ;) 2015-05-05 11:52:08 fcolista: do you have a hint how i config my pwclient to use http://patchwork.alpinelinux.org? 2015-05-05 11:54:25 http://patchwork.alpinelinux.org/project/aports/pwclientrc/ 2015-05-05 11:54:30 save it as ~/.pwclientrc 2015-05-05 11:54:43 and edit to use alpine-aports as default project 2015-05-05 11:54:58 that's it 2015-05-05 12:19:22 ssl.SSLError: [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed (_ssl.c:581) 2015-05-05 12:19:59 hm, I think I disabled https, I'm not sure 2015-05-05 12:24:51 does patchwork send a message to alpine-aports about patch being merged or rejected? 2015-05-05 12:34:35 dont think so but you can see the thread in patchwork: 2015-05-05 12:36:15 kinda sucks 2015-05-05 12:36:24 but I'll complain the other day 2015-05-05 12:42:12 ncopa: setup-keymap doesn't work for me in edge 2015-05-05 12:42:25 the uk keymaps give me a us keyboard layout 2015-05-05 12:43:44 ncopa, have you fixed the SSL issue? 2015-05-05 12:43:53 I had the same problem 2015-05-05 12:58:33 hi 2015-05-05 12:58:52 is it possible to cross-compile alpine? 2015-05-05 13:05:18 KBme, not really. only the core of packages needed to bootstrap self-build supports cross-compiling 2015-05-05 13:17:53 weird design decision. so you need to compile on a rpi for the rpi? 2015-05-05 13:25:00 KBme, we build arm images on wandboard 2015-05-05 13:25:21 KBme, cross-compiling is non-trivial; and especially since large percentage of upstream does not support it properly 2015-05-05 13:36:31 ok. 2015-05-05 13:36:54 yeah I know the difficulties of cross-compiling gnu 2015-05-05 13:40:50 I also know the pain of building on arm 2015-05-05 13:46:24 ScrumpyJack, ttf-inconsolata is in testing repo yet 2015-05-05 14:09:51 alacerda: yet? 2015-05-05 14:10:59 ncopa: can you move a few things from testing to main? twm works for me :( 2015-05-05 14:11:12 erm, i mean :) 2015-05-05 14:14:26 ScrumpyJack: do you want me to move twm from testing to main? 2015-05-05 14:15:07 hum 2015-05-05 14:15:32 this sfdisk thingy in setup-disk does not look too simple to resolve 2015-05-05 14:17:27 i just removed the -L and -uM i think 2015-05-05 14:18:00 yes please for twm and twm-doc 2015-05-05 14:21:06 ncopa: downgrading util-linux is a simple solution for setup-disks :) 2015-05-05 14:21:45 twm-doc? what am i talking about! 2015-05-05 14:21:59 ncopa: does or ca-cert support startssl certs? 2015-05-05 14:22:11 s/or/our 2015-05-05 14:24:43 ncopa: I've also been using sc snownews ldapvi rlog encfs sipcalc fvwm frotz and dillo successfully for a while now 2015-05-05 14:24:58 ncopa, you are the maintainer of libmatemixer ? :) 2015-05-05 14:26:32 alacerda: probably 2015-05-05 14:27:17 do you mind if I steal it from you? 2015-05-05 14:27:48 np 2015-05-05 14:27:51 ;) 2015-05-05 14:31:14 ScrumpyJack: sc says url is https://github.com/dkastner/sc but there are no release tags there 2015-05-05 14:31:36 source is then from a ibiblio mirror 2015-05-05 14:32:00 + a diff from launchpadlibrarian.net 2015-05-05 14:32:19 does not seem to be properly maintained by upstream 2015-05-05 14:33:02 https://github.com/dkastner/sc has only 1 commit 2015-05-05 14:33:17 saying "Initial commit - 7.13 source" 2015-05-05 14:33:24 but version number of package is 7.16 2015-05-05 14:33:42 its all very fishy to me 2015-05-05 14:34:02 the url is supposed to point to upstream maintainer 2015-05-05 14:34:08 where you can find security fixes 2015-05-05 14:34:14 and get in touch with upstream developer 2015-05-05 14:34:24 find where to send bugs 2015-05-05 16:27:52 ncopa: let me look into it. sc is very old 2015-05-05 16:59:05 ftp://metalab.unc.edu/pub/Linux/apps/financial/spreadsheet/ 2015-05-05 16:59:38 specifically: ftp://metalab.unc.edu/pub/Linux/apps/financial/spreadsheet/sc-7.16.tar.gz 2015-05-05 17:00:47 There's also: http://sourceforge.net/projects/scii/files/sc-7.16/ 2015-05-05 17:11:39 ncopa, please git pull http://git.alpinelinux.org/cgit/alacerda/aports/ 2015-05-05 18:34:01 hi 2015-05-05 18:34:03 There tutorial to start alpine arm on qemu? 2015-05-05 19:43:07 nangel: thanks for looking 2015-05-05 19:43:49 np. I was curious what sc was to start with... looks like visicalc! 2015-05-05 19:45:24 fwiw, I think the sf.net one is the one we would use. 2015-05-05 19:52:06 ok. 2015-05-06 07:43:31 i just got a quesstion on the license of the aports 2015-05-06 07:43:44 is MIT ok? 2015-05-06 07:44:01 I don't think there should be license at all, except for patches 2015-05-06 07:44:19 ncopa: whats the license for aports and the patches there? 2015-05-06 07:44:27 what license we have for patches? 2015-05-06 07:44:40 matching the project license? 2015-05-06 07:44:51 yeah 2015-05-06 07:45:07 packaging scripts don't look significant enough to license them to me 2015-05-06 07:45:24 it's more like WTFPL, except not written anywhere 2015-05-06 07:46:55 the APKGBUILDs are probably big enough to be covered by copyright, so in theory we should have a license for them 2015-05-06 07:47:55 MIT looks good to me then 2015-05-06 07:49:31 yeah 2015-05-06 08:02:52 ncopa: can you put alpine-2.20 into main? been using it daily 2015-05-06 08:03:31 yo dawg 2015-05-06 08:03:41 I heard you liek alpine so we put in alpine in your alpine 2015-05-06 08:03:49 lol 2015-05-06 08:46:51 barthalion: heh :) 2015-05-06 08:47:17 it used to be called pine 2015-05-06 10:52:43 ncopa: can you put ciwiki tsocks and darkstat into testing? patches sent to mailing list back in march :) 2015-05-06 11:03:14 http://sprunge.us/jhhe 2015-05-06 11:03:14 < I sent this to ml also 2015-05-06 11:03:27 s/ :) 2015-05-06 11:06:36 i think 3.2 is taking up most of the time 2015-05-06 11:41:11 ScrumpyJack: can you resend them to alpine-aports? 2015-05-06 11:46:09 clandmeter, ScrumpyJack: i think i can dig them up from my inbox 2015-05-06 11:47:23 >>> ciwiki: Fetching http://downloads.sourceforge.net/project/ciwiki/ciwiki-2.0.2.tar.gz 2015-05-06 11:47:23 % Total % Received % Xferd Average Speed Time Time Time Current 2015-05-06 11:47:23 Dload Upload Total Spent Left Speed 2015-05-06 11:47:23 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0curl: (22) The requested URL returned error: 404 Not Found 2015-05-06 11:47:30 thats probably why it was not applied 2015-05-06 11:49:38 let me fix that 2015-05-06 11:49:45 2.0.3 2015-05-06 11:49:46 and 2015-05-06 11:49:50 make install is broken 2015-05-06 11:50:14 kk 2015-05-06 11:50:27 we probaly also need set libexecdir 2015-05-06 11:51:16 do darkstat and tsocks build for you? 2015-05-06 11:53:10 ncopa: thanks for fixing ciwiki 2015-05-06 11:53:37 Latest News 2015-05-06 11:53:38 1.8 beta 5 is out! - 2002-10-23 21:02 2015-05-06 11:53:41 hehe 2015-05-06 11:53:49 having patchwork we only need to run ci against submitted patches and send results to ml :p 2015-05-06 11:54:29 those patches was before patchwork was set up 2015-05-06 11:55:30 ScrumpyJack: yes tsocks builds 2015-05-06 11:55:56 I know, I know, just a side note 2015-05-06 11:56:16 fortunately we won't miss any patch now 2015-05-06 11:56:24 yeah 2015-05-06 11:59:25 patchwork? 2015-05-06 11:59:42 http://patchwork.alpinelinux.org 2015-05-06 12:00:43 I also work on a patch to change status and delegation of patch basing on mailing list reply, but given my limited time, I have no idea when it's going to be ready 2015-05-06 12:01:33 tsocks fails on x86 :( 2015-05-06 12:02:20 tsocks.c:(.text+0x133): undefined reference to `__stack_chk_fail_local' 2015-05-06 12:02:23 weird... 2015-05-06 12:03:04 seems the patchwork dev isnt active lately. but there is activity from other sources. 2015-05-06 12:03:36 ncopa: remind me about --build=$CBUILD --host=$CHOST 2015-05-06 12:05:10 ccache gcc -fPIC -Os -fomit-frame-pointer -Wall -I. -nostdlib -shared -o libtsocks.so.1.8 tsocks.o common.o parser.o -ldl -lc 2015-05-06 12:13:10 when I add checking build system type... Invalid configuration `x86_64-alpine-linux-musl': machine `x86_64-alpine-linux' not recognized 2015-05-06 12:13:14 oops 2015-05-06 12:13:52 when I add $CBUILD and $CHOST to a new aport, it fails: checking build system type... Invalid configuration `x86_64-alpine-linux-musl': machine `x86_64-alpine-linux' not recognized 2015-05-06 12:15:47 i like the comments in the configure script i'm trrying to build: 2015-05-06 12:16:18 "# There might be people who depend on the old broken behavior: `$host' used to hold the argument of --host etc 2015-05-06 12:16:36 hmm 2015-05-06 13:25:54 ScrumpyJack, when we bump a package don't we need to reset pkgrel? 2015-05-06 13:27:11 no, no need 2015-05-06 13:27:19 oh, reset? 2015-05-06 13:28:00 sorry, i misread your msg 2015-05-06 13:32:45 done 2015-05-06 13:34:07 :) 2015-05-06 13:34:09 is there an X11 apps package with things like xclock, xkill, xcalc, xeyes, xedit etc? 2015-05-06 13:35:41 some X11 apps seem to have their own package. perhaps they all should for alpine linux 2015-05-06 13:53:30 the setup-disk issue should be fixed 2015-05-06 13:53:36 i think we can do rc1 then? 2015-05-06 13:56:30 ncopa: \o/ 2015-05-06 13:56:35 how did you fix it? 2015-05-06 13:58:07 you removed -L -uM :) 2015-05-06 13:59:52 and a little more 2015-05-06 14:00:03 yes, the awks :) 2015-05-06 14:00:10 -uM told sfdisk to use unit M 2015-05-06 14:00:28 i hada to add M suffix to all places there was size for sfdisk 2015-05-06 14:00:33 "give option to use lvm" very cool 2015-05-06 14:02:14 i think "Same as 'sys' but use logical volume manager for partitioning." can be confusing 2015-05-06 14:02:32 as a new user doesn't know what sys and data does 2015-05-06 14:02:51 and sys/data always seem to generate questions 2015-05-06 14:06:54 ok 2015-05-06 14:07:05 do you have suggestion how to explain it better 2015-05-06 14:09:27 This mode is a traditional disk install. The following logical volumes will be created on the disk: /boot, / (filesystem root) and swap. 2015-05-06 14:16:15 rnalrd: are you using patchwork already? 2015-05-06 14:16:21 nope 2015-05-06 14:16:36 could you register? 2015-05-06 14:16:38 how does it work? 2015-05-06 14:16:39 k 2015-05-06 14:16:41 so i can hook you up 2015-05-06 14:17:13 rnalrd: you can use both webif and pwclient 2015-05-06 14:17:32 registered 2015-05-06 14:17:42 pwclient brings some nice additions 2015-05-06 14:17:48 apk add pwclient 2015-05-06 14:17:49 ah, setup-disk it doesn't create volumes 2015-05-06 14:18:15 rnalrd: http://patchwork.alpinelinux.org/project/aports/ has a sampple pwclientrc 2015-05-06 14:19:15 ScrumpyJack: it will create lvm volumes if you run setup-disk -L 2015-05-06 14:19:31 or with alpien-conf-3.2, if you enter lvmsys or lvmdata 2015-05-06 14:19:34 or just 'lvm' 2015-05-06 14:20:11 i should finish reading before opening my bog mouth 2015-05-06 14:20:18 bog!!! 2015-05-06 14:20:24 big, naturally 2015-05-06 14:21:39 rnalrd: i think you should be able to change status now for patches 2015-05-06 14:22:03 I wonder why it didnt update the zabbix commit 2015-05-06 14:22:13 tnx 2015-05-06 14:23:45 rnalrd: did you amend that commit? 2015-05-06 14:23:51 yup 2015-05-06 14:24:06 needed pkgrel bump 2015-05-06 14:24:19 if you want it to automatic set status, then amend is not possible. 2015-05-06 14:24:26 i see 2015-05-06 14:24:33 it will change the checksum of the patch 2015-05-06 14:24:33 kinda sucks 2015-05-06 14:26:16 lets see if it manages to generate isos... 2015-05-06 14:27:23 rnalrd: if you can, please change the status in patchwork of the patches which need no follow-up anymore. 2015-05-06 14:27:39 sure 2015-05-06 14:28:49 done 2015-05-06 14:29:40 grazie 2015-05-06 14:29:55 prego :) 2015-05-06 14:34:22 whoops 2015-05-06 14:34:23 my bad 2015-05-06 14:34:27 thanks rnalrd 2015-05-06 14:34:45 whee! v3.2.0_rc1 is out 2015-05-06 14:35:48 does anyone acutally use the .xdelta files? 2015-05-06 14:35:49 \o/ 2015-05-06 14:35:50 \o/ 2015-05-06 14:36:22 oh well I use it, but I guess most don't :) 2015-05-06 14:38:05 ok 2015-05-06 14:38:36 it is currently less job to keep it :) 2015-05-06 14:38:47 its mostly for saving space on the mirrors 2015-05-06 14:47:23 barthalion: would be nice if pwclient would default to list -s new, and including the submitter. 2015-05-06 14:47:39 what are the new stuff that needs extra testing in v3.2? 2015-05-06 14:47:41 mariadb 2015-05-06 14:47:45 postfix 3.0 2015-05-06 14:47:48 what more? 2015-05-06 14:47:56 ruby/redmine? 2015-05-06 14:48:36 mariadb? not really 2015-05-06 14:48:43 we're still on 5.5, correct? 2015-05-06 14:48:47 yes 2015-05-06 14:49:08 so if apk does the replacement correctly, we're good 2015-05-06 14:49:23 i'd like smoeone to setup mysql with alpine v3.1 and test to do the actual migration 2015-05-06 14:49:36 I can do it once I'm back home 2015-05-06 14:50:14 anything else we'd like test? 2015-05-06 14:50:27 nfs 2015-05-06 14:50:30 ncopa: maybe you can mention alacerda work to bring mate to alpine? 2015-05-06 14:50:48 i think its new for v3.2 right? 2015-05-06 14:50:52 yes 2015-05-06 14:51:05 i'd like to have setup-box ready to 3.2 ... i am running 2015-05-06 14:51:08 nfs is working 2015-05-06 14:51:11 cool 2015-05-06 14:51:23 but i had to use optins to mount v3 2015-05-06 14:51:26 v4 i cannot get working 2015-05-06 14:51:32 i'm ok with that :) 2015-05-06 14:51:36 setup-alpine is out of scope for 3.2 2015-05-06 14:51:47 setup-box* 2015-05-06 14:51:57 probably yes :-/ 2015-05-06 14:52:10 setup-what? 2015-05-06 14:52:43 clandmeter: alacerda is working on a setup script to install desktop variants 2015-05-06 14:52:44 and similar 2015-05-06 14:52:50 ahw pity :( 2015-05-06 14:53:00 well, we want to replace everything that is setup-alpine related 2015-05-06 14:53:30 instead of writing $x variants of same script, you just write sequence of questions/commands in json and let the script do magic 2015-05-06 14:53:38 ncopa, barthalion does it means that setup-box will have to wait until 3.3 ? 2015-05-06 14:53:49 alacerda: not necessarily 2015-05-06 14:54:03 i might be able to look at it tomorrow or so 2015-05-06 14:54:06 ok 2015-05-06 14:54:07 but we will replace setup-xfce and similar for starters 2015-05-06 14:54:20 thats what setup-box currently do 2015-05-06 14:54:21 yes 2015-05-06 14:54:22 ok 2015-05-06 14:54:30 what more do we want people to test? 2015-05-06 14:54:41 things that got new major version 2015-05-06 14:54:49 and that might not have been tested much 2015-05-06 14:54:54 postfix is on my list 2015-05-06 14:56:15 btw, i think we may want remove some of that gnome-* stuff 2015-05-06 14:56:29 mate replaces it effectively 2015-05-06 14:57:25 ScrumpyJack: can you post your patches to alpine-aports instead of alpine-devel? 2015-05-06 15:07:10 ncopa: seems you didnt reference your fix for #4060 2015-05-06 15:55:11 ncopa, http://sprunge.us/PBEA 2015-05-06 15:55:11 please 2015-05-06 16:22:52 evening 2015-05-06 16:46:19 ncopa or clandmeter please: http://sprunge.us/HMRi 2015-05-06 16:46:39 haaaaaaaf, _o/ 2015-05-06 16:56:55 Have you seen this error message before? (trying to run 'mono' from /testing) Error relocating /usr/lib/libmonosgen-2.0.so.1: __flt_rounds: symbol not found 2015-05-06 16:59:50 I think it's fixed in musl already 2015-05-06 16:59:56 are you sure your system is up to date? 2015-05-06 17:03:04 No, I'm not... let me read about updating. 2015-05-06 17:03:25 apk upgrade -U 2015-05-06 17:05:00 Ok, testing. 2015-05-06 17:05:16 Now, there's a non-alpha released... How do I bump the package to 4.0.1? 2015-05-06 17:05:28 are you on edge or 3.1 btw? 2015-05-06 17:05:44 clone aports git repo 2015-05-06 17:06:07 I guess our wiki covers basic development 2015-05-06 17:06:09 I think I'm on 3.1. 2015-05-06 17:06:32 I'm on gliderlabs/alpine 2015-05-06 17:06:35 :latest 2015-05-06 17:06:51 Upgrading didn't help against the error. 2015-05-06 17:06:56 Same error 2015-05-06 17:07:11 well, one reason is that you're mixing stable release with edge now 2015-05-06 17:07:23 Ok 2015-05-06 17:07:44 It's just a PoC -- which should I choose and how do I choose it? 2015-05-06 17:07:54 folks from gliderlabs also provide edge image 2015-05-06 17:07:56 alpine:edge 2015-05-06 17:08:14 Let me try that. 2015-05-06 17:14:43 barthalion: that works, let me see how far I get =). What is the procedure for PRs to the arepos repo? 2015-05-06 17:16:05 I think we accepted few PR on github, but regular procedure is to send patches to alpine-aports mailing list 2015-05-06 17:18:40 wait... we can do PRs on the aports GitHub repo? 2015-05-06 17:19:48 was just futzing about with git format-patch and send-email... can someone confirm the patch i just emailed in is in plaintext? i'm not sure if my client lets me send plaintext :( 2015-05-06 17:21:37 these github PR were rather exceptions 2015-05-06 17:22:40 patch looks good here 2015-05-06 17:22:54 patchwork is also happy about it: http://patchwork.alpinelinux.org/patch/104/ 2015-05-06 17:23:10 (mindblown) the first i'm seeing patchwork... 2015-05-06 17:23:49 clandmeter and fabian_a set it up recently, because we started to overlook patches on mailing list 2015-05-06 17:24:03 this way we won't 2015-05-06 17:24:30 yea, one of the reasons i like GitHub PR workflow, definitely a good idea when it is email based 2015-05-06 17:25:00 i imagine it does basic linting and sanity check stuff, too? 2015-05-06 17:25:27 idiots like me who might send a rich text patch email :p 2015-05-06 17:25:28 oh, I don't think so 2015-05-06 17:25:50 I don't know what patchwork will do with wrongly formatted e-mail 2015-05-06 17:26:10 probably won't consider such message as a patch, who knows 2015-05-06 17:27:03 i wonder if Stuart manually formatted his subjects or if there is a way to get format-patch to do that 2015-05-06 17:27:44 I think it's git magic 2015-05-06 17:30:16 git send-email - --cover-letter --annotate does it, if we can trust google 2015-05-06 17:41:31 ACTION confirms 2015-05-06 17:42:36 but you wouldn't do send-email when you're tired :f 2015-05-06 20:22:01 greetings to all, alpine supports Raspberry Pi 2 Model B ? https://www.raspberrypi.org/products/raspberry-pi-2-model-b/ 2015-05-06 20:41:48 there is already model b of rpi2? 2015-05-06 20:41:52 jesus 2015-05-06 21:39:20 In an APKBUILD file, if it tells me "ERROR: fsharp: source 3.1.1.32.tar.gz needs to be renamed to avoid possible collisions" -- how do I go about renaming file? 2015-05-06 21:47:15 saveas- 2015-05-06 21:47:57 I might have a PR/patch for F# soon 2015-05-06 21:50:01 haaaaaaf: use :: instead 2015-05-06 21:50:21 at the end of the URI? 2015-05-06 21:50:31 filename::uri 2015-05-06 21:50:48 by default we use 2015-05-06 21:51:01 I see 2015-05-06 21:51:12 Ok, I will, thanks. 2015-05-06 21:51:21 $pkgname-$pkgver.tar.gz::http:// 2015-05-06 21:51:24 So I'm trying to figure this old-fashioned emailing out... 2015-05-06 21:51:41 I have a TLS-secured IMAP server with MD5-Challenge-Response authentication. 2015-05-06 21:51:42 submitting patches? 2015-05-06 21:51:54 Yes. 2015-05-06 21:51:58 No SMTP. 2015-05-06 21:52:50 not sure we have any doc on wiki about that. 2015-05-06 21:53:11 Can't I just give you this link?? https://patch-diff.githubusercontent.com/raw/alpinelinux/aports/pull/14.patch 2015-05-06 21:53:23 but i guess there should be enough info to be found online 2015-05-06 21:53:24 Just append .patch to a PR 2015-05-06 21:53:33 I know how to do PRs easily 2015-05-06 21:53:36 its not that we cannot handle them 2015-05-06 21:53:54 if more ppl do it, we get lost. 2015-05-06 21:54:29 Can I sign up to your mailing list without getting any e-mail? 2015-05-06 21:55:03 if its just one patch i can do it 2015-05-06 21:55:18 but we prefer mailing list 2015-05-06 21:59:41 There's F# too 2015-05-06 21:59:45 Two 2015-05-06 22:14:48 haaaaaaf: next time just send it to the ml with correct commit msg. please refer to aports git log. 2015-05-06 22:31:28 clandmeter: ok, I'll try. I can't use my mail server with git, I've discovered. 2015-05-06 22:46:23 & in your .bashrc or .profile or whatever you use: alias gitmail='git send-email --to alpine-aports@lists.alpinelinux.org HEAD^' 2015-05-06 22:46:57 then you just add your commit & "gitmail" from the terminal 2015-05-06 22:48:52 BitL0G1c: thank you 2015-05-06 22:50:55 np - an Alpine LXC container makes a very good dev environment too 2015-05-06 22:52:27 to test a desktop http://it-offshore.co.uk/linux/alpine-linux/30-alpine-linux-spice-kvm-desktop 2015-05-06 23:02:13 ok testing 2015-05-06 23:02:19 sent an e-mail 2015-05-06 23:05:37 if it was for "Add F#" it arrived 2015-05-06 23:06:03 ok great 2015-05-06 23:10:40 haaaaaaf: like i said before, please use a correct commit msg. testing/pkgname: new aport (or updated to) 2015-05-06 23:11:08 also, your package depends on mono, which is only build for x86_64 2015-05-06 23:11:36 so arch is transitive? 2015-05-06 23:11:55 set arch to x86_64 2015-05-06 23:12:08 mono doesnt build currently for x86 2015-05-06 23:12:29 atleast it didnt for alpha1 2015-05-06 23:12:41 I read in the docs that if a package doesn't have native components, it should be noarch 2015-05-06 23:13:01 correct 2015-05-06 23:13:21 So are you saying that if it depends on something that's not both 32 and 64 bit, it shouldn't? 2015-05-06 23:13:52 if you depends on a 64 package which doesnt exist, i guess you know that answer :) 2015-05-06 23:14:01 err is mean 32 2015-05-06 23:14:49 im also wondering why your email didnt end up in patchwork 2015-05-06 23:15:03 yeah, or one could assume it's the job of the package indexer to walk the dependencies and remove leafs that are orphans for a given architecture. 2015-05-06 23:15:24 because technically the package is noarch, but yeah, I understand what you're saying. 2015-05-06 23:15:45 I think I need to get some sleep, sorry. =) I'll be back tomorrow for more talk 2015-05-06 23:15:48 Cheers 2015-05-06 23:16:03 ok gnite. me too 2015-05-07 05:02:53 I've noticed that testing/py-numpy and testing/py-scipy are outdated (numpy 1.8.1 -> 1.9.1, scipy 0.14 -> 0.15). Can anyone bump those versions? 2015-05-07 05:05:58 *numpy 1.8.1 -> 1.9.2, scipy 0.14.0 -> 0.15.1 2015-05-07 05:21:00 There is also another question: why py-numpy depends on gfortran? I do expect dependency on libgfortran, but not gfortran itself 2015-05-07 05:22:26 (however, once I've done `ldd` on all .so files in py-numpy package, it showed no gfortran links) 2015-05-07 06:20:42 morning 2015-05-07 06:20:44 (14/64) Installing postgresql-bdr-libpq (9.4.1_p2-r3) 2015-05-07 06:20:44 ERROR: postgresql-bdr-libpq-9.4.1_p2-r3: trying to overwrite usr/lib/libpq.so.5 owned by libpq-9.4.1-r1. 2015-05-07 06:20:44 ERROR: postgresql-bdr-libpq-9.4.1_p2-r3: trying to overwrite usr/lib/libpq.so.5.7 owned by libpq-9.4.1-r1. 2015-05-07 06:20:48 we have problem with that 2015-05-07 06:21:12 the problem is that both provides a libpq with same soname 2015-05-07 06:21:30 any ideas how to solve that 2015-05-07 06:21:46 postgresql-bdr appears to be a patched version of postgresql 2015-05-07 07:02:28 ncopa: regarding openrc changes, can we have somekind of backwards compat mod? 2015-05-07 07:02:39 hum 2015-05-07 07:03:02 i think new openrc has support for old style runscript already 2015-05-07 07:03:07 gentoo is also having a symlink for bwsupport 2015-05-07 07:03:28 $ ls -la /sbin/runscript 2015-05-07 07:03:28 lrwxrwxrwx 1 root root 12 May 6 13:22 /sbin/runscript -> /sbin/openrc 2015-05-07 07:03:39 we have that too 2015-05-07 07:03:48 yes but not the other way around :) 2015-05-07 07:04:11 maybe i should say forward compat supprot 2015-05-07 07:04:43 i suppose my mistake was to migrate all runscript -> openrc-run 2015-05-07 07:05:12 and when doing so i should have added a versioned depend for openrc 2015-05-07 07:05:20 if you mix repo,s pkgs with openrc are broken 2015-05-07 07:05:47 what is new style? just shebang change or new syntax sugar? 2015-05-07 07:05:55 i should have added depend="openrc>0.13" 2015-05-07 07:06:01 then would the mixed repo worked 2015-05-07 07:06:11 dunno if i want rebuild all again :-/ 2015-05-07 07:06:18 just new shebang 2015-05-07 07:06:47 i suppose we could backport openrc-run to openrc-0.12 (for v3.1) 2015-05-07 07:07:04 otoh, in a couple of weeks v3.2 is out 2015-05-07 07:07:22 and edge with v3.1 will not be expected to work at that point 2015-05-07 07:07:29 but edge + v3.2 would 2015-05-07 07:07:37 yep 2015-05-07 07:07:53 well maybe next time we should be a bit more carefull :) 2015-05-07 07:08:12 i should probably have waited with rename the openrc-run til after v3.2 was released 2015-05-07 07:08:15 yes 2015-05-07 07:08:28 its easy to be afterwise 2015-05-07 07:08:30 pre-rc fun 2015-05-07 07:15:53 Hi guys! I've noticed that testing/py-numpy and testing/py-scipy are outdated (numpy 1.8.1 -> 1.9.2, scipy 0.14 -> 0.15.1). Can anyone bump those versions? 2015-05-07 07:16:53 I'm sorry, my IRC client has crashed 2015-05-07 07:17:57 ncopa: can you verify all certs are updated adn working? bugs/www/forum (although forum https revproxy crap is broken) 2015-05-07 07:19:45 https://bugs.alpinelinux.org works for me 2015-05-07 07:20:12 https://alpinelinux.org/ too 2015-05-07 07:20:56 wiki.a.o is broke 2015-05-07 07:22:02 https://forum seems to have correct cert at least but the css is bad 2015-05-07 07:22:06 i dont manager wiki 2015-05-07 07:22:10 -r 2015-05-07 07:22:22 atleast idont think i have access 2015-05-07 07:22:47 i can generate them though 2015-05-07 07:24:44 would be nice 2015-05-07 07:25:02 put them in alpine-ns2:/root/certs/ or something 2015-05-07 07:33:19 ncopa: done 2015-05-07 07:33:28 w.a.o works for me 2015-05-07 07:33:35 crt already combined for ngnix usage 2015-05-07 07:33:58 if you need a normal one get the ./other/wiki.crt 2015-05-07 07:34:15 thanks! 2015-05-07 10:45:35 ncopa, http://sprunge.us/PBEA http://sprunge.us/HMRi 2015-05-07 11:27:36 Executing shared-mime-info-1.4-r0.trigger 2015-05-07 11:27:54 if there are much disk io the shared-mime-info takes *ages* 2015-05-07 11:30:42 5mins 2015-05-07 11:35:16 ncopa, [setup-box]: I was thinking that intead of having a "gparted" desktop we should have an "openbox" desktop and an option to install gparted as an admin-tool 2015-05-07 11:35:39 what do you think? 2015-05-07 11:35:47 might work 2015-05-07 11:36:08 the idea with gparted desktop was to bring up a desktop as root with gparted as only application 2015-05-07 11:36:27 if openbox or other window manager is used is irrelevant 2015-05-07 11:37:08 but a general system rescue option might also work 2015-05-07 11:37:10 but if I do this way (the one I suggested) the user will need to execute: setup-box -d openbox -g gparted 2015-05-07 11:37:42 instead of setup-box -d gparted 2015-05-07 11:38:44 i suppose that woudl also make it possible to do setup-box -d xfce -g gparted 2015-05-07 11:38:50 yes 2015-05-07 11:39:59 what more 'boxes' do we have 2015-05-07 11:40:11 do we have non-desktop profiles too? 2015-05-07 11:40:14 (or... admin-tools (a meta group with a bunch of admin tools) vs admin-tools.gparted) 2015-05-07 11:40:20 Yes 2015-05-07 11:40:28 we have pentest for while 2015-05-07 11:40:28 i was thinking a setup-desktop tool 2015-05-07 11:40:49 setup-box -g pentest or... setup-box -s pentest.forensics 2015-05-07 11:41:02 is pentest with xorg? 2015-05-07 11:41:03 it wont bring any desktop 2015-05-07 11:41:06 ok 2015-05-07 11:41:06 no 2015-05-07 11:41:26 what does -g pentest do? 2015-05-07 11:41:42 does it do anything more than just install a group of packages? 2015-05-07 11:41:49 The way setup-box is designed it is possible, in the future, we do sth like: setup-box -g mail-server 2015-05-07 11:42:18 -g pentest just install a group of packages 2015-05-07 11:42:48 hum 2015-05-07 11:43:06 i am not overly excited about setup-box -d -g 2015-05-07 11:43:15 I want it to also to ask questions to users 2015-05-07 11:43:19 not only install packages 2015-05-07 11:43:22 yes 2015-05-07 11:43:30 desktop need set up a user account 2015-05-07 11:43:35 barthalion, what kind of questions? 2015-05-07 11:43:52 desired username, network configuration, and so on 2015-05-07 11:43:52 ncopa, great I can add this 2015-05-07 11:43:55 we should put algitbot in it 2015-05-07 11:44:03 :| 2015-05-07 11:44:04 everything that setup-* does, probably 2015-05-07 11:44:27 hm 2015-05-07 11:44:27 ncopa, this is when dialog takes place 2015-05-07 11:45:03 but for this first version I just wanted to make it work in a basic way 2015-05-07 11:45:08 i think we have 2 conflicting interests here 2015-05-07 11:45:25 one of the problem i am interested in solving is 2015-05-07 11:45:41 to make alpine like a live cd 2015-05-07 11:45:43 you boot it up 2015-05-07 11:45:57 and with very little extra effort you get a full live desktop 2015-05-07 11:46:22 (setup-box -d xfce isn't enough?) 2015-05-07 11:46:57 the target audience for this is people who deoes not know the difference of xfce and openbox 2015-05-07 11:47:13 hum... got it! 2015-05-07 11:47:20 by requireing -d xfce you force the user to 1) know that -d exists 2015-05-07 11:47:24 so you need read docs 2015-05-07 11:47:36 2) make a choice, xfce, openbox, mate, etc.. 2015-05-07 11:47:53 cant we add it to syslinux options? 2015-05-07 11:48:11 dunno 2015-05-07 11:48:18 http://hastebin.com/itewaqajiv.avrasm 2015-05-07 11:48:28 but i'd liek something that is 'bring me something graphical that looks nice' 2015-05-07 11:48:44 isn't it enough to tell the user to use -d option and pick up one desktop? 2015-05-07 11:48:59 that is one of the goals 2015-05-07 11:49:02 but we can have, for example, XFCE as default 2015-05-07 11:49:17 the second is giving more detailed choices to users 2015-05-07 11:49:31 but that requires more that the end user know what he wants 2015-05-07 11:50:28 why not just systlinux: start alpine linux with gui option. and it will postboot setup xfce and start it. 2015-05-07 11:50:44 and if user have opinion then is the chance bigger that he will not use the tool at all 2015-05-07 11:50:53 and just do: apk add 2015-05-07 11:51:17 ok... then we can just accept: setup-box -d and, if the user do not specify any environment, we install a default desktop? 2015-05-07 11:51:28 what could work is 2015-05-07 11:51:31 setup-box -d 2015-05-07 11:51:35 and a wrapper script 2015-05-07 11:51:38 setup-desktop 2015-05-07 11:51:52 with no options it would call setup-box -d xfce or similar 2015-05-07 11:52:07 i have some MIPS hardware coming 2015-05-07 11:52:11 i am going to port alpine to it 2015-05-07 11:52:20 i'd also be interested in a --quick,-q option 2015-05-07 11:52:28 bought 5 of those mips creator ci20 boards 2015-05-07 11:52:36 that sets up a user accound called 'liveuser' or similar 2015-05-07 11:52:40 kaniini: cool! 2015-05-07 11:52:41 yeah, I wanted a free sample to do the same 2015-05-07 11:52:54 but apparently they were uninterested 2015-05-07 11:53:34 how does MPIS compare to ARM? 2015-05-07 11:53:39 i have no experience with MIPS 2015-05-07 11:54:04 ncopa, i can change the setup-box to work with or without an arg for -d. 2015-05-07 11:54:12 doing this, is the bigger problem solved? 2015-05-07 11:54:45 I think that we cannot run away from -d, -g, -s and many other options 2015-05-07 11:55:27 i dont know 2015-05-07 11:55:40 personally, i have never really used 'mail server' profiles 2015-05-07 11:55:47 i think redhat used to have groups 2015-05-07 11:55:50 'mail server' 2015-05-07 11:55:53 'file server' 2015-05-07 11:55:56 web server 2015-05-07 11:55:57 etc 2015-05-07 11:56:04 and installed a bunch of apps for you 2015-05-07 11:56:12 i never really found that useful 2015-05-07 11:56:18 ubuntu has the same 2015-05-07 11:56:34 but thats probably mostly because i normally know exactly what tool i want 2015-05-07 11:56:47 'mail server' 2015-05-07 11:56:50 makes me wonder 2015-05-07 11:56:57 smtp only? 2015-05-07 11:56:59 imap? 2015-05-07 11:57:11 posfixadmin and a web server? 2015-05-07 11:57:14 apache? 2015-05-07 11:57:18 etc etc 2015-05-07 11:57:25 now, that might be me only 2015-05-07 11:57:28 i dont know about others 2015-05-07 11:57:37 might be others find such groups useful 2015-05-07 11:57:46 i dont that alpine has a lot of usercase for it. 2015-05-07 11:58:00 thats what i suspect too 2015-05-07 11:58:02 however 2015-05-07 11:58:13 setup-desktop is slightly different 2015-05-07 11:58:18 but for desktop, ive used it often 2015-05-07 11:58:33 because it involves installing a set of fonts, replace mdev with udev 2015-05-07 11:58:40 and some more 2015-05-07 11:58:58 set up a non-root user account 2015-05-07 11:59:01 etc 2015-05-07 11:59:06 yes, and maybe probing for hardware where possible 2015-05-07 11:59:35 i'd be interesting in a convenient way to do all those manual steps for me 2015-05-07 11:59:56 more than having a tool to suggest a group of tools for me for pentesting etc 2015-05-07 12:00:01 that said 2015-05-07 12:00:15 it might be useful to pull in a group of system rescue tools 2015-05-07 12:00:18 for rescue operations 2015-05-07 12:00:25 what I think is: if I have a mail-server group it is going to install what is choosen by "us" if the user has its own preferences he/she can install their server without the need of setup-box tool... 2015-05-07 12:00:30 another aplication: 2015-05-07 12:00:38 but what is a mail server? 2015-05-07 12:01:04 it can be a mail gateway 2015-05-07 12:01:14 for just filtering emails 2015-05-07 12:01:22 or it can be mailbox storage 2015-05-07 12:01:34 the same way: what is a desktop? :) 2015-05-07 12:01:46 true 2015-05-07 12:01:49 we have many choises... but we will chose one to be the default 2015-05-07 12:02:09 if the user wnats sth diferent.. he can go by himself 2015-05-07 12:02:32 as said, i am mostly interested in the tool to do the manual job of setting up user account and replace mdev with udev 2015-05-07 12:03:03 basically 2015-05-07 12:03:04 it all depends how much logic you build into setup-box. you can ask them what kind of mail server they want. 2015-05-07 12:03:19 the question I ask: what problem do we want to solve 2015-05-07 12:03:26 understood... I'll do the following then: focus on replace setup-* scripts and after that I can daydream about other features 2015-05-07 12:03:26 let's not make it a new ansible or puppet 2015-05-07 12:03:38 and focus on replacing current setup-* scripts 2015-05-07 12:03:45 barthalion, +1 2015-05-07 12:03:55 exactly, we need find the scope limit too 2015-05-07 12:03:58 so we can write just new json instead of new script 2015-05-07 12:03:59 where do we stop 2015-05-07 12:04:21 ok so we seem to want solve a bit different problems 2015-05-07 12:04:27 that are somewhat overlapping 2015-05-07 12:04:50 the person who creates/manages it, defines the limits. 2015-05-07 12:05:02 i am currently mostly interested in a setup-desktop that complements the current setup-alpine 2015-05-07 12:05:11 :) 2015-05-07 12:05:28 ncopa: will you maintain it also? or will all support come from alacerda? 2015-05-07 12:05:54 clandmeter: stop destroying a good flamewar with logic :) 2015-05-07 12:05:59 lol 2015-05-07 12:06:01 hahahaha 2015-05-07 12:06:14 what does algitbot has to say about it? 2015-05-07 12:06:20 lol 2015-05-07 12:06:34 clandmeter: i get your point 2015-05-07 12:06:41 ncopa: exactly my thoughts 2015-05-07 12:06:41 and i think you are right 2015-05-07 12:06:48 clandmeter: you ruin all the fun :p 2015-05-07 12:07:04 ok ill talk to algitbot again 2015-05-07 12:07:10 hahaha 2015-05-07 12:07:11 :) 2015-05-07 12:08:00 alacerda: i think clandmeter is right, since you do the actual work, your opinion is heavy 2015-05-07 12:08:08 clandmeter, thanks you rescued me from ncopa! :) 2015-05-07 12:08:14 do what you think is best 2015-05-07 12:08:33 my concern is that we trade "simple" shell scripts for "simple" json/yaml/whatever parsing 2015-05-07 12:08:44 i'll consider all that we have talked... 2015-05-07 12:08:45 i can probably live with setup-box -d xfce -g gparted 2015-05-07 12:08:45 we can't have both 2015-05-07 12:09:06 barthalion: i'd like rewrite the setup-* tools 2015-05-07 12:09:12 but not now 2015-05-07 12:09:34 ncopa, i'd like to replace setup-* tools with setup-box only 2015-05-07 12:09:48 dont know if it is possible 2015-05-07 12:09:49 thats not gonna happen before v3.2 2015-05-07 12:10:18 but i am willing to include a setup-desktop or a setup-box -d script 2015-05-07 12:10:29 which you'll have to run in addition to setup-alpine 2015-05-07 12:10:45 actually having separate scripts makes sense to me 2015-05-07 12:11:21 we have also kunkku's aconf which is supposed to replace acf 2015-05-07 12:11:28 but thats more designed for web interface 2015-05-07 12:11:31 how would setup-disks be implemented in yaml/json? 2015-05-07 12:12:34 i mean, is alacerda going to target that aswell? 2015-05-07 12:12:50 this is just my imagination at the moment 2015-05-07 12:13:13 I proposed replacing only setup-xfce like scripts for the beginning 2015-05-07 12:13:32 yeah 2015-05-07 12:13:42 I was thinking of just join the setup-disks into setup-box as a function. Basically everything that we can move into json, we'll do. What we cant we just create a function. But yes.. for while this is just imagination 2015-05-07 12:13:49 i dont think its worth aiming for replacing setup-* yet 2015-05-07 12:13:54 I don't know if you all agree with this 2015-05-07 12:13:57 if we tell alacerda to tackle all things at once, we won't achieve anything 2015-05-07 12:14:00 except tired alacerda 2015-05-07 12:14:17 yes, wait with thinking of replacing setup-* 2015-05-07 12:14:50 but focus on something that can replaces setup-xorg and setup-xfce and setup-desktop-gparted 2015-05-07 12:15:11 if alacerda want throw in 'pentest' there too, fine with me 2015-05-07 12:15:27 nice... we are on the same page now! :) 2015-05-07 12:15:29 i'm also ok to keep json and yaml out for now 2015-05-07 12:16:11 http://git.alpinelinux.org/cgit/alacerda/setup-box/tree/ 2015-05-07 12:16:21 json is already in use 2015-05-07 12:16:32 (or maybe i missunderstand what you said) 2015-05-07 12:16:38 btw, what with the box name? 2015-05-07 12:16:58 it makes me thing you want to setup some dropbox replacement 2015-05-07 12:17:05 2015-05-07 12:17:08 *think* 2015-05-07 12:17:26 2015-05-07 12:17:48 but i agree with clandmeter 2015-05-07 12:17:49 am i missing something? 2015-05-07 12:18:36 setup-box sounds like dropbox soundbox or 'box' https://play.google.com/store/apps/details?id=com.box.android 2015-05-07 12:18:58 which is why i like setup-desktop name better 2015-05-07 12:19:08 even if that is a non gui desktop for pen testing 2015-05-07 12:19:22 setup-alpine sounds even better 2015-05-07 12:20:07 but that will bring bigger issues then it will solve probably 2015-05-07 12:21:30 currently yes 2015-05-07 12:25:18 alacerda: you are parsing json with sh? 2015-05-07 12:25:40 it uses jq for parsing 2015-05-07 12:25:43 we're not insane yet 2015-05-07 12:26:21 :) 2015-05-07 12:31:58 wouldnt lua be a better match for such tool? 2015-05-07 12:32:02 ACTION hides 2015-05-07 12:36:00 the one who writes the code decides :p 2015-05-07 12:37:23 stateless: https://bugs.alpinelinux.org/issues/4075 2015-05-07 13:20:00 clandmeter, parsing with jq 2015-05-07 13:21:07 2015-05-07 13:21:08 :P 2015-05-07 13:21:23 hair* ¬¬ 2015-05-07 13:21:41 ncopa: are we having our 10th year anniversary this year? 2015-05-07 13:23:08 oh are we? 2015-05-07 13:23:34 yes 2015-05-07 13:23:44 2005-08-22 2015-05-07 13:23:46 wow 2015-05-07 13:24:50 we need BEER lots of it! 2015-05-07 13:26:24 yes! 2015-05-07 13:27:33 you make me feel old... 2015-05-07 13:52:49 ncopa, you ARE old :) 2015-05-07 13:53:45 :) 2015-05-07 14:02:50 ncopa: thanks applied your patch :) 2015-05-07 14:07:29 stateless: thanks! 2015-05-07 14:14:26 ncopa: I guess we can carry this patch until the 3.2 release and then bump up fortify to 0.5 2015-05-07 14:14:43 the reason I am saying this is because I also have a patch for 0.5 that may break things (trapping on out of bound reads for a few functions) 2015-05-07 14:23:31 ok 2015-05-07 14:23:33 sounds good 2015-05-07 14:41:11 do we want try make an automatic upgrade to mariadb? https://bugs.alpinelinux.org/issues/3843 2015-05-07 14:41:17 sounds risky to me 2015-05-07 14:43:08 if you do "apk upgrade -a" are mysql packages left? 2015-05-07 14:43:15 they shouldn't 2015-05-07 14:43:23 i think they are 2015-05-07 14:43:34 i dont know 2015-05-07 14:43:37 needs to be tested 2015-05-07 14:44:07 i suppose we could ship an empty mysl package that has mariadb as dependency. 2015-05-07 14:46:38 huh? main/linuc-grsec ? 2015-05-07 14:46:49 whoops... 2015-05-07 14:47:05 i think its the second typo in commit messages today 2015-05-07 14:47:18 my influence grows! 2015-05-07 14:47:28 :) 2015-05-07 14:53:00 my 2 cents: would there be value in adding "do you want to setup a desktop?" to setup-alpine and if yes call setup-desktop? 2015-05-07 14:54:19 thats one option 2015-05-07 14:54:29 can we implement that without adding a question? 2015-05-07 14:54:32 i mean 2015-05-07 14:54:44 "do you want setup a desktop?" yes 2015-05-07 14:54:51 "which desktop do you want?" ... 2015-05-07 14:54:57 that is 2 questions 2015-05-07 14:55:34 while "which desktop do you want setup? ('xfce', 'lxde', 'mate', or 'none') [none]: " 2015-05-07 14:55:49 does the same but with a single question 2015-05-07 14:56:12 =] 2015-05-07 14:56:14 i like that 2015-05-07 14:57:08 yes, the point was to re-use the existing wrapper, setup-alpine 2015-05-07 14:57:35 i read somewhere there is a interest in an alpine linux livecd 2015-05-07 14:57:44 yes 2015-05-07 14:57:47 what's the reasoning behind that? 2015-05-07 14:58:24 as a crowd pleaser? :) 2015-05-07 14:58:31 nice to have a boot cd/usb that give you network and a browser 2015-05-07 14:59:12 but i think its more a need for a convenient way to set up an alpine desktop 2015-05-07 14:59:24 that it runs 'live' is just what you get for free 2015-05-07 14:59:29 do you see it as boot into X out of the box? or boot, say, alpine-mini and run setup-something to get X 2015-05-07 14:59:46 latter 2015-05-07 14:59:50 kk 2015-05-07 14:59:52 setup-something to get X 2015-05-07 14:59:55 gotcha 2015-05-07 14:59:58 thats what i'd like 2015-05-07 15:00:03 we have setup-gparted-desktop already 2015-05-07 15:00:33 ok i gotta go 2015-05-07 15:00:36 2015-05-07 15:00:37 see u tm 2015-05-07 15:00:40 see u 2015-05-07 15:00:42 laters 2015-05-07 15:03:25 ACTION wonder what setup-xxx would do 2015-05-07 15:12:54 new releases, monit has v5.13, and networkmanage v1.0.2 2015-05-07 16:24:24 evening 2015-05-07 19:51:55 are they any ssh banner (asci) of alpinelinux ? :) 2015-05-07 20:04:39 official? I doubt it 2015-05-07 20:10:37 official yes, well also non official i dont find any 2015-05-07 21:40:31 crow: echo Alpine\ Linux | figlet 2015-05-07 21:46:31 uggedal, for i in "----------" "Alpine Linux 3.2.0" "----------"; do echo $i | figlet ; done 2015-05-07 21:46:32 :) 2015-05-07 21:49:24 i didn't know this figlet 2015-05-07 21:49:25 =] 2015-05-07 21:50:09 /usr/lib/ruby/gems/2.2.0/gems/therubyracer-0.12.2/lib/v8.rb:4:in `require': Error relocating /usr/lib/ruby/gems/2.2.0/extensions/x86_64-linux/2.2.0/therubyracer-0.12.2/v8/init.so: __vfprintf_chk: symbol - /usr/lib/ruby/gems/2.2.0/extensions/x86_64-linux/2.2.0/therubyracer-0.12.2/v8/init.so (LoadError) 2015-05-07 21:50:17 knows somebody a way to fix that? 2015-05-07 21:50:36 i'm not be able to install therubyracer gem that i need for a rails application :( 2015-05-08 00:09:36 anyone willing to test setup-box? I've just pushed 0.1.0_rc3 2015-05-08 00:11:18 setup-box -d mate -s desktop_extras.internet or setup-box -d xfce -s systools.disk or any other variation 2015-05-08 06:44:24 what are the blockers for v3.2 now? 2015-05-08 06:53:42 any possibility of updating monit v5.13, networkmanage v1.0.2, Mosquitto v1.4.2 2015-05-08 06:57:25 was there any discussion of not including both mariadb 10.0.x + 5.5.x ? 2015-05-08 06:57:45 that way it would give a base for migration 2015-05-08 07:11:58 uggedal thanks, i didnt know figlet to 2015-05-08 07:13:12 vkrishn: the discussion was about moving to mariadb 5.5 2015-05-08 07:13:26 because migration to 10 requires manual intervention 2015-05-08 07:13:42 and I'd personally wait for 10.1, as iirc it's 5.6 compatible 2015-05-08 07:14:08 we wiath with mariadb 10 2015-05-08 07:14:12 wait* 2015-05-08 07:15:27 *** The installed version of Flex, /usr/bin/flex, is too old to use with PostgreSQL. 2015-05-08 07:15:27 *** Flex version 2.5.31 or later is required, but this is flex 2.5.39. 2015-05-08 07:15:28 heh 2015-05-08 07:15:32 stupid warnings... 2015-05-08 07:17:12 lol 2015-05-08 07:18:22 ok 2015-05-08 07:21:38 MariaDB 10.1.4 now available 14 Apr 2015 2015-05-08 07:22:32 ok, its Beta release 2015-05-08 07:28:13 can someone on edge try this, https://mariadb.atlassian.net/browse/MDEV-7883 2015-05-08 07:34:46 and this, https://mariadb.atlassian.net/browse/MDEV-6892 2015-05-08 07:34:51 else upgrade to v5.5.43 2015-05-08 07:35:08 let me finish an upgrade for Arch first 2015-05-08 07:35:13 Alpine will follow 2015-05-08 07:35:19 ok, thanks 2015-05-08 07:58:32 I thought that buildbots go nuts when package is pulled back to testing from main 2015-05-08 07:58:37 is it no longer true? 2015-05-08 08:04:21 morning 2015-05-08 08:04:30 I get this error: 2015-05-08 08:04:33 ERROR: qt5-qttools-5.4.0-r0: trying to overwrite usr/bin/qdbus owned by qt-4.8.6-r2 2015-05-08 08:04:56 and 2015-05-08 08:04:59 ERROR: qt5-qdbusviewer-5.4.0-r0: trying to overwrite usr/bin/qdbusviewer owned by qt-x11-4.8.6-r2 2015-05-08 08:06:25 Arch resolves this renaming to 'qdbus{viewer}-qt5' 2015-05-08 08:18:45 AmatCoder: how does fedora solve it? 2015-05-08 08:19:11 ncopa: let me check... 2015-05-08 08:24:16 ncopa: fedora does same thing 2015-05-08 08:24:49 not sure, but vaguely recall on qt5 website that there are issues installing both qt4+qt5 2015-05-08 08:26:34 ncopa: really they install them on '/usr/lib{64}/qt/bin/qdbus{viewer}' and makes symlink to '/usr/bin/qdbus{viewer}-qt5' 2015-05-08 08:27:26 ok 2015-05-08 08:27:55 if both arch and fedora can start them with *-qt5, then we can do the same 2015-05-08 08:28:30 I think so 2015-05-08 08:33:15 that affects designer and liguist too? 2015-05-08 08:33:24 usr/bin/designer \ 2015-05-08 08:33:24 usr/bin/lconvert \ 2015-05-08 08:33:24 usr/bin/linguist \ 2015-05-08 08:33:24 usr/bin/lrelease \ 2015-05-08 08:33:24 usr/bin/lupdate \ 2015-05-08 08:33:25 usr/bin/pixeltool \ 2015-05-08 08:33:27 usr/bin/qcollectiongenerator \ 2015-05-08 08:33:31 usr/bin/qhelpconverter \ 2015-05-08 08:33:33 usr/bin/qhelpgenerator \ 2015-05-08 08:34:49 they rename them all 2015-05-08 08:35:06 so I believe so 2015-05-08 08:36:33 script build says: 2015-05-08 08:36:40 'for b in "${pkgdir}"/usr/lib/qt/bin/*; do' 2015-05-08 08:36:55 ' ln -s /usr/lib/qt/bin/$(basename $b) "${pkgdir}"/usr/bin/$(basename $b)-qt5' 2015-05-08 08:45:52 i've a beaglebone to play with, and i'm interested in booting it with alpine-uboot image available here: 2015-05-08 08:45:53 http://dev.alpinelinux.org/~tteras/ 2015-05-08 08:46:17 i'm gonna try this: http://dev.alpinelinux.org/~tteras/alpine-uboot-150424-armhf.tar.gz 2015-05-08 08:46:19 fcolista, please report back how it went 2015-05-08 08:46:30 yes, latest uboot-$date should be good to try 2015-05-08 08:46:49 fabled, i'd like to know how the installation dohuld be done 2015-05-08 08:46:50 http://archlinuxarm.org/platforms/armv7/ti/beaglebone-black 2015-05-08 08:46:57 i'm following that guide ^^^^^ 2015-05-08 08:46:58 there's README in side the .tar.gz 2015-05-08 08:47:02 oh ,ok 2015-05-08 08:47:04 perfect 2015-05-08 08:47:49 it's basically make SD a FAT 2015-05-08 08:47:51 extract .tar.gz 2015-05-08 08:47:56 and copy over the right bootloader files 2015-05-08 08:48:18 no big difference with normal usb installation 2015-05-08 09:30:22 bah, beaglebone won't boot. I've only one partition on SD (is this right? Arch wiki reports two partition, one for root and one for boot..anyway) which is bootable. FS is vfat. 2015-05-08 09:30:27 Disk is sde 2015-05-08 09:30:53 i've "cp am335x_boneblack/{MLO,u-boot.img} /media/sde1" 2015-05-08 09:31:02 no dd 2015-05-08 09:31:25 then, i've extracted the alpine-uboot-*.tar.gz into /media/sde1 2015-05-08 09:31:55 then, booted beagle, which results in black screen. 2015-05-08 09:32:23 (I booted from SD according with wiki, by pressing S2 button near the microsd slot) 2015-05-08 09:41:45 fabled. while you were off: 2015-05-08 09:41:46 bah, beaglebone won't boot. I've only one partition on SD (is this right? Arch wiki reports two partition, one for root and one for boot..anyway) which is bootable. FS is vfat. 2015-05-08 09:41:46 Disk is sde 2015-05-08 09:41:46 i've "cp am335x_boneblack/{MLO,u-boot.img} /media/sde1" 2015-05-08 09:41:46 no dd 2015-05-08 09:41:46 then, i've extracted the alpine-uboot-*.tar.gz into /media/sde1 2015-05-08 09:41:48 then, booted beagle, which results in black screen. 2015-05-08 09:41:50 (I booted from SD according with wiki, by pressing S2 button near the microsd slot) 2015-05-08 09:41:57 am i missing something? 2015-05-08 09:42:42 is the vfat partition marked 'bootable' in partition table? 2015-05-08 09:42:47 yes 2015-05-08 09:42:51 and they say it might need to be FAT16 2015-05-08 09:42:55 so there might be size limit 2015-05-08 09:43:09 gotta format in fat16 then. 2015-05-08 09:49:23 almost finished. Let you know how it goes. Formatted with F16, copied u-boot MLO and u-boot.img, then extracted alpine tar.gz 2015-05-08 09:52:09 doesn't boot. 3 leds on beagleboard of 4 are on 2015-05-08 09:52:23 D2,D3,D4,D5 2015-05-08 09:52:34 sorry, D5 is off 2015-05-08 09:58:07 fcolista, sorry. i have no idea how beagle works. but sounds like uboot issue. 2015-05-08 09:58:17 otherwise you would get u-boot boot screen 2015-05-08 09:58:20 does serial work? 2015-05-08 09:58:49 i'm wondering if i need to dd the img or split the partition according with arch howto 2015-05-08 09:59:00 no serial available so far 2015-05-08 10:00:14 fcolista, arch splits the partition so that it can have ext4 rootfs 2015-05-08 10:00:22 only FAT primary is needed to boot 2015-05-08 10:00:27 you are free to try of course 2015-05-08 10:00:40 no, iìd preferr to keep it simple 2015-05-08 11:07:31 ncopa, I've tested both: http://sprunge.us/XVPY 2015-05-08 13:06:28 ncopa: https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/tree/Documentation/filesystems/overlayfs.txt?id=refs/tags/v4.0#n162 2015-05-08 13:06:38 i think that part is pretty interesting for us 2015-05-08 13:07:01 we can cut modloop into pieces 2015-05-08 13:09:53 keep in mind it's not in 3.18 2015-05-08 13:10:12 ncopa, 3.14.41-grsec is out 2015-05-08 13:10:33 so you need to stack these manually 2015-05-08 13:10:35 busy with other stuff today 2015-05-08 13:10:54 im out of office rest of today 2015-05-08 13:23:17 fcolista - hi - perl & perl-module-build both provide /usr/bin/config_data - is there another perl module that can be used in place of Module::Build for a new package Build.PL? 2015-05-08 13:33:40 or perhaps remove /usr/bin/config_data from module::build as it depends on perl ? I've compared the files in each & they are identical 2015-05-08 13:35:40 BitL0G1c: have you ever passed sound from an lxc container into alpine? 2015-05-08 13:40:37 not yet - I've been trying to get xpra to work - that would probably pass sound & video from a container 2015-05-08 13:41:39 I think my apkbuild is ok - the problem is with musl somewhere - perhaps i should send the patch ? 2015-05-08 13:42:32 it will be a nice solution - the applications run like a normal desktop application 2015-05-08 13:43:43 sure send it. 2015-05-08 13:43:54 ok will do 2015-05-08 13:44:31 i would really like to get chromium build on alpine 2015-05-08 13:44:37 but its a giant headache 2015-05-08 13:52:27 ncopa, http://sprunge.us/hceJ 2015-05-08 14:00:41 alacerda: Environments 2015-05-08 14:01:20 ? 2015-05-08 14:02:09 pkgdesc 2015-05-08 14:02:45 oops 2015-05-08 14:02:48 thanks 2015-05-08 14:03:38 alacerda: shouldnt it depends on jq? 2015-05-08 14:03:50 or however its called 2015-05-08 14:03:58 the script checks it ad install as a virtual package 2015-05-08 14:04:20 I think it is better than depends on it 2015-05-08 14:04:28 fixed: http://sprunge.us/iNAD 2015-05-08 14:04:56 clandmeter, thanks 2015-05-08 14:05:58 alacerda: you cannot use git.a.o as source 2015-05-08 14:06:08 no? 2015-05-08 14:06:20 :| 2015-05-08 14:06:22 the snapshots are not static like github does 2015-05-08 14:06:25 didn't know 2015-05-08 14:06:31 hum... 2015-05-08 14:06:38 setup-box-0.1.0.tar.bz2: FAILED 2015-05-08 14:07:43 fixing... 2015-05-08 14:09:41 clandmeter, shoudn't it appears on http://dev.alpinelinux.org/ ? 2015-05-08 14:10:11 no 2015-05-08 14:10:18 you should uplaod it manually 2015-05-08 14:10:22 :) 2015-05-08 14:10:28 but you dont have access 2015-05-08 14:10:31 only commiters have 2015-05-08 14:10:35 ahw! 2015-05-08 14:10:49 maybe you can mirror it on github? 2015-05-08 14:10:55 :( 2015-05-08 14:11:15 I think github caches snapshots 2015-05-08 14:11:26 so the checksums are alwas the same 2015-05-08 14:11:52 or have your own http/ftp source 2015-05-08 14:12:18 oh you have access to git.a.o right? 2015-05-08 14:12:25 else you cannot push your releases 2015-05-08 14:13:10 i have access to git.a.o 2015-05-08 14:13:26 ok, 2015-05-08 14:14:04 not sure about permissions, but try to create directory in /archive 2015-05-08 14:14:25 mkdir: can't create directory 'setup-box': Permission denied 2015-05-08 14:14:26 :) 2015-05-08 14:14:39 :) 2015-05-08 14:14:48 let me see 2015-05-08 14:15:53 i created the dir 2015-05-08 14:15:57 can you access it? 2015-05-08 14:16:11 yes 2015-05-08 14:16:57 i dont do permissions, so i just set group to you for that dir 2015-05-08 14:17:14 chmod g+w 2015-05-08 14:17:21 :_ 2015-05-08 14:17:22 :) 2015-05-08 14:17:37 try 2015-05-08 14:17:53 thanks 2015-05-08 14:17:57 working 2015-05-08 14:18:00 clandmeter - xpra patch sent 2015-05-08 14:18:17 i think you can even add a git hook, so it automatically creates a tarball 2015-05-08 14:18:26 BitL0G1c: nice ill check it 2015-05-08 14:18:37 BitL0G1c: it doenst build, or doesnt run? 2015-05-08 14:19:26 oh i see 2015-05-08 14:19:37 lazy loading... 2015-05-08 14:20:19 it builds 2015-05-08 14:20:54 clandmeter, http://sprunge.us/FPLi 2015-05-08 14:23:01 lol 2015-05-08 14:23:17 alacerda: did you delete the src and redownload it? 2015-05-08 14:23:51 oh i guess i didnt :) 2015-05-08 14:24:11 i fixed the source, then abuild checksum, then abuild -r (to test), then git add, then git --ammend 2015-05-08 14:24:23 the sprunge 2015-05-08 14:24:30 yes but it all against old src in your /var/cache 2015-05-08 14:24:31 then irc 2015-05-08 14:24:36 ahh 2015-05-08 14:24:37 :) 2015-05-08 14:25:03 rm /var/cache/distfiles/setup-box* 2015-05-08 14:27:57 shall i redo checksums? or will you give a new patch? 2015-05-08 14:28:18 ahhh this rm is what I need to do? 2015-05-08 14:28:27 hehe I thought it was you... sorry 2015-05-08 14:28:28 try it 2015-05-08 14:28:29 I'll do 2015-05-08 14:29:07 http://sprunge.us/XLZT 2015-05-08 14:29:46 \o/ 2015-05-08 14:30:02 thanks clandmeter ;) 2015-05-08 14:30:11 yw 2015-05-08 14:30:47 gonna test this afternoon 2015-05-08 14:31:12 but i still have a feeling jq should be in depends. 2015-05-08 14:33:05 but ill leave that up to ncopa 2015-05-08 14:46:49 hum... 2015-05-08 14:47:38 your point makes sense... 'cuz setup-box will install jq everytime it is executed and finally remove jq 2015-05-08 14:48:00 once in depends this won't happen 2015-05-08 14:49:46 there are probably senarios where its needed at run time 2015-05-08 15:17:18 jq is fairly small 2015-05-08 15:17:28 it should be in depends 2015-05-08 16:55:21 during the setup-box tests I foud that sed was crazy... I took the chance to add jq to depends: http://sprunge.us/aFSK 2015-05-08 16:55:24 clandmeter, ^^ 2015-05-09 00:28:30 Hello everyone, I just got my raspberry and wonder if anyone has any idea how to install alpine at 2015-05-09 07:41:30 please git pull git://git.alpinelinux.org/fab/aports 2015-05-09 07:42:15 well, still the same state as for the last requests to pull 2015-05-09 07:42:26 thanks 2015-05-09 08:09:55 fabian_a: can you rebase? 2015-05-09 08:11:20 there are some python packages in patchwork. could be they are also conflicting. 2015-05-09 08:11:36 ill wait with them and first do yours if i can merge cleanly 2015-05-09 09:07:03 any ideas why none of my xen dom0s reboot correctly? 2015-05-09 12:17:04 clever people, what musl support in gcc means to us? 2015-05-09 12:21:19 nothing yet, but it will clear up a few issues in the current gcc patches 2015-05-09 12:21:54 and hopefully no musl related patches will be needed to build gcc >=6.1 2015-05-09 12:23:40 once the patches are backported to earlier gcc versions it will fix a few corner cases (eg correct stdint types in the preprocessor that matters for gfortan, working libcilkrts, proper dynamic linker definition on mips and sh, etc) 2015-05-09 12:25:12 currently another blocking issue emerged though: libgcc_s.so is broken on x86, gcc 6.1 because the new multi versioning feature turned out to be broken and the fix involves symbol versioning hacks that musl dynamic loader does not like 2015-05-09 12:26:12 i'm not yet sure how that can be resolved.. 2015-05-09 12:26:46 ok, so no revolution, I thought so 2015-05-09 16:16:42 Hello BitL0G1c, now that I've had the time, I've tried to reproduce the nm problems I had in my old alpine installation (which I unfortunately no longer have). All I can say, is that it's working now. It most definitely was some problem in the interaction with udev; I think I may have messed something up with the dependencies in my first alpine install. Thanks for your help and time. 2015-05-09 16:19:28 np - many thx for testing 2015-05-09 19:20:00 barthalion: you should get some sleep 2015-05-09 19:20:08 sleep is for the weak 2015-05-09 19:20:13 what I need is pre-commit hook 2015-05-09 19:21:14 i hope you didnt use commas in pkgver :) 2015-05-09 19:21:18 ah, right, that was merged upstream 2015-05-09 19:22:47 but apparently not by Oracle 2015-05-09 19:25:51 clandmeter, scadu: there, with dots now 2015-05-09 19:25:55 :) 2015-05-09 19:26:11 my little hero. 2015-05-09 19:37:42 are we supposed to change init scripts to '#!/sbin/openrc-run' or not for the moment ? 2015-05-09 19:52:26 Idk, did ncopa do something about it? 2015-05-09 19:52:29 clandmeter? 2015-05-09 19:52:59 if he didn't, I guess so 2015-05-09 20:00:30 ok 2015-05-09 20:02:49 i recently updated the init script for supervisor to use /sbin/openrc-run and it was accepted, so i think the change is ok 2015-05-09 20:04:18 ;-) ok 2015-05-09 22:57:06 What is the software used for the mailing lists? 2015-05-10 03:35:11 ncopa, any feedback on the llvm/clang/compiler-rt patchset would be greatly appreciated - http://bugs.alpinelinux.org/issues/4068 2015-05-10 03:35:49 the patches are now against 3.6.1, due to be released shortly (freeze for testing was may 4th) 2015-05-10 03:36:29 i've also included PIE-by-default behavior when targeting alpine linux 2015-05-10 03:37:19 i'm not sure i want to bother with SSP-by-default, but could possibly be convinced to do so :p 2015-05-10 03:39:28 ncopa, i also need to know if i have the triples correct for what alpine supports. if i'm correct, that should be just... x86_64-alpine-linux-musl, i686-alpine-linux-musl, armv6-alpine-linux-muslabihf, and armv7-alpine-linux-muslabihf 2015-05-10 03:39:50 those are all i've added support for 2015-05-10 03:42:10 blueness, also ping in case you want to send that patchset over to gentoo's musl port. just two changes would be needed: 1) in one place, instead of checking for the "alpine" vendor string, just always return true (gentoo uses "pc" vendor string so it's a worthless check) and 2) the DistroCheck thing needs an entry for gentoo 2015-05-10 03:43:07 or for part 2 you could also just always use the checked behavior. whatever. i'm just a random person on IRC, not a cop. 2015-05-10 04:50:18 when is ncopa or other alpine devs usually awake? i think you guys are at least 6 time zones away from me 2015-05-10 07:19:22 ncopa, side-node - the patchset works with either autotools or cmake. i made sure the compiler-rt patch tweaked both. 2015-05-10 07:30:33 ACTION toots a horn 2015-05-10 07:30:36 ACTION sets off an alarm 2015-05-10 07:31:00 ACTION ...goes to bed 2015-05-10 09:02:50 can someone explain to me how the gcc bootstrap process works? when i abuild gcc, it compiles multiple times even when i don't set any environment variables 2015-05-10 15:51:12 is there any way to make abuild not strip debug info ? 2015-05-10 16:20:19 BitL0G1c: does it not copy debug info to /usr/lib/debug and creating *-dbg subpackages before stripping? 2015-05-10 16:20:56 BitL0G1c: could be the native build system strips before it's doing that though, like LDFLAGS=-s or straight calls to strip etc 2015-05-10 16:21:54 ok 2015-05-10 19:12:41 BitL0G1c i also did not find any global option, but for example oscam package have "sed -i '/STRIP/d' Makefile || return 1" , but i dont use that i use oscam nativ option: ./config.sh WITH_DEBUG 2015-05-10 19:18:06 BitL0G1c: you can create a -dbg subpackage 2015-05-10 19:18:52 barthalion, any time to review some patches to llvm/clang? ^_^ http://bugs.alpinelinux.org/issues/4068 2015-05-10 19:20:29 barthalion, am i correct in assuming that i only need SUID binaries to be PIE to pass the abuild sanity checks for them? and the like? 2015-05-10 19:20:53 or do they also need to be SSP? 2015-05-10 19:38:27 I leave things like that for ncopa and fabled 2015-05-10 19:38:57 I'm not the best guy around to handle such low-level stuff :) 2015-05-10 19:39:09 wait for tomorrow, they'll take a look at it 2015-05-10 19:41:59 fair enough 2015-05-10 19:49:41 I've built inotify-tools with a patch to add a --include & --includei regex (as well as the exclude regexes) - I forgot to increase the option array sizes 2015-05-11 06:04:13 libwebsockets - v1.4 2015-05-11 06:04:13 mosquitto - 1.4.2 - re-compiled 2015-05-11 06:04:13 task v2.4.4 2015-05-11 06:04:13 testing/taskd - v1.1.0 2015-05-11 06:04:29 recommended bug fixes, update pls, if possible 2015-05-11 07:35:57 ncopa: you said we dont micro optimize our build infra ;-) 2015-05-11 07:36:23 good morming btw 2015-05-11 07:59:33 morning 2015-05-11 12:47:50 ncopa: http://git.alpinelinux.org/cgit/aports/commit/?id=f7ac9edbb32386fdb47db354df45e47b89b0e999 2015-05-11 12:48:10 ncopa: ' Revert "main/qt5-qttools: add -qt5 suffix to exec files" ' 2015-05-11 12:48:19 ncopa: problems? 2015-05-11 12:50:34 yes 2015-05-11 12:51:07 there was some other qt5 package that broke 2015-05-11 12:51:25 was using an exe file 2015-05-11 12:52:26 ah, weird 2015-05-11 12:52:36 no 2015-05-11 12:52:43 i renamed all files 2015-05-11 12:52:49 i dont know if we need to do that 2015-05-11 12:56:05 seems like both fedora and arch linux installs the binaries in other dir than /usr/bin/ 2015-05-11 12:56:21 and symlink them with -qt5 suffix to /usr/bin/ 2015-05-11 12:56:51 yes 2015-05-11 13:06:23 we have plans for pcre2 for next AL release ? 2015-05-11 13:06:46 https://lists.exim.org/lurker/message/20150105.162835.0666407a.en.html , but was released in Jan 2015-05-11 13:07:50 "There is no C++ wrapper." 2015-05-11 13:08:02 does anything depend on it? 2015-05-11 13:08:44 unmaintained/mongodb 2015-05-11 13:08:54 then I don't think so 2015-05-11 13:08:59 :) 2015-05-11 13:09:14 also I'm sure mongodb uses some internal copy 2015-05-11 13:14:49 hmm, changes sounds like application using it may have to switch to pcre2 2015-05-11 13:15:17 rather than just pointing the builder to pcre2 2015-05-11 13:16:00 unless there is a pcre->pcre2 compat wrapper 2015-05-11 13:17:04 well, there is nothing use it at the moment 2015-05-11 16:09:51 http://patchwork.alpinelinux.org/patch/127/ - plans for java apps ? 2015-05-11 16:54:17 vkrishn - elasticsearch is needed to parse logs from nginx-naxsi - the pre-compiled ES packages have libraries built for libc - so I need to build ES 2015-05-11 16:54:35 *glibc 2015-05-11 16:55:01 and there are no specific plans for java apps, someone contributes, we commit 2015-05-11 16:55:10 for maven to work we need ca-certificates-java installed 2015-05-11 16:55:26 which s/be possible once maven is added 2015-05-11 16:57:49 for a couple pf perl modules I've found some conflicts with files which are already in main/perl - ok to move module files out of the way ? (file.module.name) - the files are identical 2015-05-11 16:58:10 or is there another way I s/be handling this conflict ? 2015-05-11 16:59:50 I remember Arch's perl maintainer handled that somehow 2015-05-11 17:00:02 I think he made the perl package provide these modules 2015-05-11 17:00:38 ok will check arch 2015-05-11 17:34:08 BitL0G1c, ok, elasticsearch is one of best, but we may also wanna stick to "Small. Simple. Secure" philosophy 2015-05-11 17:35:14 lol 2015-05-11 17:35:15 there are some search technologies brewing around http://reverbrain.com/ 2015-05-11 17:36:13 looking at aports from <=2.6 and today, you can call it puffy at best 2015-05-11 17:36:25 let people use proven tools 2015-05-11 17:36:41 sure, java is neither small, simple or secure 2015-05-11 17:36:52 yandex, facebook uses it 2015-05-11 17:37:13 worst part is who and how its gonna be maintained in future version 2015-05-11 17:38:10 even look at simple desktop like testing/enlightement, there is a new version that completely pkged differently 2015-05-11 17:38:41 that is the reason I even don't stress at adding destops 2015-05-11 17:39:14 best would be AL spinoff/blends completely done by individuals 2015-05-11 17:39:29 these blends make use of AL base 2015-05-11 17:41:32 that way one can install a different desktop by just adding repo alongwith AL's 2015-05-11 17:42:06 and best is , it will be updated and maintained to latest without giving stress to core AL team 2015-05-11 17:42:22 sure, first find people for that 2015-05-11 17:43:07 one thought is "AL hardware lease by user/time based" 2015-05-11 17:43:25 AL hardware is idle 12hrs min 2015-05-11 17:43:28 our workflow for non critical packages is pretty much streamlined, so I don't really see a problem here 2015-05-11 17:43:35 as long as users contribute 2015-05-11 17:43:49 contributing is ok, maintaining ! 2015-05-11 17:44:20 you can maintain package without push access 2015-05-11 17:45:15 with lease, one can get those aports reviewed and the user can trigger build 2015-05-11 17:46:01 maintain it... without stress on AL release cycle 2015-05-11 17:47:27 one can also start spinoffs like AL-go, AL-java 2015-05-11 17:47:32 with AL as base 2015-05-11 17:52:35 desktops are UI that has to look different with couple of releases, meaning some extra maintenance 2015-05-11 17:55:41 with current rising pkg nos + static nos Maintainers+ 3month rolling release cycle... ahem cough!! 2015-05-11 17:57:23 AL base = C, C++ , + absolute essentials (loosely, no hard rule) 2015-05-11 17:59:10 but AL[L] is welcome, I am now on my 6th Jessie source dvd download ;-))) 2015-05-11 18:07:54 suggestion , blends.a.o, bledit.a.o (cgit) 2015-05-11 18:07:56 gtg 2015-05-11 19:59:12 http://patchwork.alpinelinux.org/patch/127/ has already claimed Maintainer 2015-05-11 19:59:34 and infra seems streamline, just need to replicate 2015-05-11 19:59:46 for blends.a.o, blendit.a.o (cgit) 2015-05-11 20:00:29 idea.... ALDroid 2015-05-11 20:00:45 :) 2015-05-11 20:01:14 gtg 2015-05-11 20:47:01 sigh 2015-05-11 20:47:06 obviously it has 2015-05-11 20:47:19 you don't need to have push access to maintain a package 2015-05-11 20:47:33 you don't need another subdomain for that either 2015-05-12 01:32:03 hi 2015-05-12 01:32:29 is there any reason why I can see "kern.warn" from host inside lxc-container? 2015-05-12 01:33:41 kern.warn messages are presented in host ( /var/log/message) and also in containers log at the same position, log entry was generated by the host. 2015-05-12 07:26:28 ncopa, you there, main/php-phalcon has v2.0.1, but don't know why its taking insanely longer to build than present version 2015-05-12 07:29:38 http://sprunge.us/IBNK 2015-05-12 07:37:45 question for tmux users. how do you get around clients not redrawing properly when run in tmux? is it a char encoding issue? a termcap issue? 2015-05-12 07:38:22 my mail clients for exemple (pine/mutt) struggle 2015-05-12 07:39:17 never had an issue with irssi 2015-05-12 07:41:27 i've taken to using dtach with irssi for now until i figure out what's wrong 2015-05-12 07:43:53 character encoding is messed up. I'm not getting accents in french "Le président François Ho"e 2015-05-12 07:44:24 ncopa, I don't have edge to test, but any help would be nice 2015-05-12 07:45:31 ScrumpyJack: which client do you use? 2015-05-12 07:45:33 ScrumpyJack: what is TERM set to? 2015-05-12 07:45:51 export CHARSET=UTF-8 2015-05-12 07:45:51 export LANG=en.utf8 2015-05-12 07:45:53 might help 2015-05-12 07:46:12 i use putty, and i need to set it to utf8 exsplicitly. 2015-05-12 07:46:55 nepal earthquake still in foreplay now 2015-05-12 07:50:50 i'm in a remote alpine linux server. zsh. local client is xfce4-terminal. i access remote via ssh. lcoal and remote $TERM is screen-256color 2015-05-12 07:53:08 mutt redraws fine when not run inside tmux, but inside tmux it's a mess. mutt can't display accents inside or outside tmux 2015-05-12 08:00:06 setting CHARSET and LANG help 2015-05-12 08:04:09 actually it didn't :( 2015-05-12 08:08:37 ok, it helps if you actually export the $ :) 2015-05-12 08:22:05 does this tmux.conf file work for for anyone? https://github.com/yesmeck/tmuxrc 2015-05-12 08:22:40 it has neat arrows in the status bar that i can't seem to get 2015-05-12 09:07:56 ScrumpyJack: same as clandmeter i've no issue with weechat in tmux 2015-05-12 09:19:45 hi 2015-05-12 09:19:50 I made a new issue for openssh 2015-05-12 09:20:12 #4171 2015-05-12 09:20:22 whats the best way to find out, which keys should be generated? 2015-05-12 09:20:38 when all HostKey commented, generate all 2015-05-12 09:20:45 if only some, generate the uncommented HostKey? 2015-05-12 09:50:22 ScrumpyJack i tested it also now, and it seems not to work, but i dont have any X or something, just ssh and using "PuTTY Tray" https://puttytray.goeswhere.com/ . i have some bottom grey,black,blue arrows bars. 2015-05-12 10:07:34 <[xming]> Has anyone encounter this PHP Notice: Use of undefined constant GLOB_BRACE 2015-05-12 10:07:47 <[xming]> php.net says this might not work on some non GNU systems 2015-05-12 10:07:56 <[xming]> is this the case with musl? 2015-05-12 10:27:05 master does not have grsecurity-3.1-3.14.41-201505101121.patch in aports ! 2015-05-12 10:27:45 sounds terrible 2015-05-12 10:27:49 given how we use 3.18 2015-05-12 10:27:54 :p 2015-05-12 10:28:30 no just breaks normal way of including patch in aports 2015-05-12 10:28:44 :) 2015-05-12 10:29:05 why do you need it? 2015-05-12 10:29:10 we don't use 3.14 anymore 2015-05-12 10:29:46 oops, I meant grsecurity-3.1-3.18.11-201504190814-alpine.patch 2015-05-12 10:30:04 it's downloaded from our servers 2015-05-12 10:30:10 I still don't see any issue in that 2015-05-12 10:30:33 http://dev.alpinelinux.org/~tteras/grsec/grsecurity-3.1-3.18.13-201505021013-alpine.patch 2015-05-12 10:31:08 so what is this about? 2015-05-12 10:31:32 I thought all grsec patches go in aports 2015-05-12 10:31:45 there never was such rule 2015-05-12 10:32:41 awesome 2015-05-12 11:14:20 we add the to aports when patch is from grsecurity.net because they remove them once they are old 2015-05-12 11:14:34 so if you try build older kernel from aports 2015-05-12 11:14:37 then it would fail 2015-05-12 11:15:21 how are "we" on the 3.18 support not being a grsec stable ? 2015-05-12 11:17:49 there's already quite a few hicups on the stable one... I wouldn't dare running the test one in production 2015-05-12 11:18:29 https://github.com/alpinelinux/linux-stable-grsec 2015-05-12 11:18:43 oh you keep that one 2015-05-12 11:19:03 we maintain our own grsec patch for 3.18 2015-05-12 11:19:16 didn't read before commenting 2015-05-12 11:19:29 i dont think we wrote it any place 2015-05-12 11:19:32 you feel confident enough for the task ? 2015-05-12 11:19:49 well 2015-05-12 11:19:53 i meant "i read -stable- and thought you kept 3.14" 2015-05-12 11:19:55 :D 2015-05-12 11:20:07 i did 3.10 for a while 2015-05-12 11:20:13 its a bit risky 2015-05-12 11:20:15 yeah i know 2015-05-12 11:20:36 i said i will not be able to maintain 3.18 grsec alone 2015-05-12 11:20:43 i got some help 2015-05-12 11:20:54 and i remember seeing you struggling with upstream changes and not being able to add grsec additions 2015-05-12 11:20:55 fabled is pretty competent when it comes to kernel 2015-05-12 11:21:00 ok great 2015-05-12 11:21:34 we also monitor grsec changes now 2015-05-12 11:21:35 still 2015-05-12 11:21:36 cause there's still some painfull UAF on 3.14 that won't be solved anytime soon 2015-05-12 11:21:44 :( 2015-05-12 11:21:57 UAF? 2015-05-12 11:22:02 use after free 2015-05-12 11:22:19 gave me some nice kernel panics in production ^^ 2015-05-12 11:22:25 oh i though it was some university in alaska... http://www.uaf.edu/ 2015-05-12 11:22:28 :) 2015-05-12 11:22:29 ok 2015-05-12 11:22:29 lol 2015-05-12 11:22:36 hum 2015-05-12 11:22:36 ok 2015-05-12 11:22:44 i know nothing about alaska 2015-05-12 11:22:56 where those UAFs from grsec, pax or upstream? 2015-05-12 11:23:34 upstream detected by PAX 2015-05-12 11:23:37 https://forums.grsecurity.net/viewtopic.php?f=3&t=4150 2015-05-12 11:23:53 this is still pending and bites me in the a$$ every 3 days :( 2015-05-12 11:24:00 aw 2015-05-12 11:24:17 so you would have had it even with a vanilla kernel 2015-05-12 11:24:23 nope 2015-05-12 11:24:35 well you'd have it but you wouldn't see it 2015-05-12 11:24:37 :D 2015-05-12 11:24:42 quite the difference 2015-05-12 11:24:53 so attackers could use it unnoticed 2015-05-12 11:24:59 indeed 2015-05-12 11:25:12 or my service is down every 3 days 2015-05-12 11:25:14 you chose 2015-05-12 11:25:17 :D 2015-05-12 11:25:17 :) 2015-05-12 11:25:21 i know 2015-05-12 11:25:27 pick your poison 2015-05-12 11:25:34 run buggy software 2015-05-12 11:25:42 or dont run at all 2015-05-12 11:25:44 pick one 2015-05-12 11:25:55 well it's the kernel that's buggy in that case though 2015-05-12 11:26:06 but it's not like we don't know it ^^ 2015-05-12 11:26:23 i am pretty happy with pax though 2015-05-12 11:26:30 I am as well 2015-05-12 11:26:34 because those issues bubbles up to the surface 2015-05-12 11:26:39 and can get fixed 2015-05-12 11:26:43 otherwise they wouldnt be fixed 2015-05-12 11:26:54 but it sometimes get difficult to justify to the management 2015-05-12 11:26:54 but its a pain whilt it happens 2015-05-12 11:26:57 yes 2015-05-12 11:27:07 i understand that problem 2015-05-12 11:27:31 they already don't want me to switch on alpine :( 2015-05-12 11:27:34 those bastards 2015-05-12 11:27:37 :((( 2015-05-12 11:27:39 :) 2015-05-12 11:27:52 they probably dont deserve alpine :) 2015-05-12 11:28:29 yeah but i do 2015-05-12 11:28:45 i believe you are 2015-05-12 11:28:52 i believe you do 2015-05-12 11:28:57 maybe they don't deserve me either... but that's another issue :D 2015-05-12 11:30:18 that does not sound completely unlikely :) and yes. its other issue 2015-05-12 11:30:58 how do you reproduce that bug? 2015-05-12 11:31:01 would love to have buckloads of money to spend 2015-05-12 11:31:32 ncopa: i run consul agent on a grsec server, and wait for it to kill the consul user 2015-05-12 11:31:34 :D 2015-05-12 11:31:43 usually takes around 3 days to happen 2015-05-12 11:32:41 and as i use a pretty locked down grsec consul 2015-05-12 11:32:49 "Can you tell me what kind of file descriptors consul is putting into epoll? lsof output may be helpful here." 2015-05-12 11:32:53 I sometimes end up in panic 2015-05-12 11:33:07 ncopa: yeah has been answered on IRC actually 2015-05-12 11:33:09 ok 2015-05-12 11:33:11 good 2015-05-12 11:33:17 he's been working on it for a while 2015-05-12 11:33:25 sent a patch to lkml 2015-05-12 11:33:34 then this wasn't sufficient 2015-05-12 11:33:48 then haven't heard from him for a while 2015-05-12 11:34:49 he has a reproducer that can reproduce it quicker? 2015-05-12 11:35:11 "It withstood a 12 hour stress test with my local reproducer that otherwise triggered the bug within seconds." 2015-05-12 11:35:22 the first bug that he thought was the culprit 2015-05-12 11:35:39 http://git.mauras.ch/linux-grsecurity/commit/?h=linux-grsecurity-3.14.34-201502271838-coredumb-af_unix-no_dead_poll 2015-05-12 11:35:42 this fix 2015-05-12 11:36:03 yes i saw that 2015-05-12 11:36:12 so there might be more than one bug? 2015-05-12 11:36:18 yep definitely 2015-05-12 11:36:37 it was taking 1day to happen now it's randomly more than 3 2015-05-12 11:36:44 but still happend 2015-05-12 11:36:47 happens 2015-05-12 11:38:41 I'm testing the 3.14.41 right now 2015-05-12 11:42:22 even the stable patch changes too much :( 2015-05-12 12:05:48 ok found another one 2015-05-12 12:06:40 :(( 2015-05-12 14:13:56 ncopa: I will check ssh-keygen -A at the evening 2015-05-12 14:32:15 shafire: thanks 2015-05-12 14:32:39 i hope that ssh-keygen -A will check what you have enabled in your config 2015-05-12 15:06:18 what are the blocker for v3.2 now? 2015-05-12 15:06:36 i have openrc 'default' runlevel not pulling in services from 'boot' runlevel 2015-05-12 15:06:45 and qt5 conflicts with qt4 2015-05-12 15:06:48 anything else? 2015-05-12 16:01:36 ncopa, i'd like to upgrade musl but they still haven't tagged a new release :/ 2015-05-12 16:02:20 ncopa, if we get new release today/tomorrow, i'd hope to upgrade it 2015-05-12 16:02:34 even though it's slightly late. but it has bug fixes, i'd rather not have to cherry-pick 2015-05-12 16:14:55 will we get last/users and so one back soon? 2015-05-12 16:17:45 that's utmp stuff. probably not going to happen soon. unless some alternate mechanism is used to implement that. 2015-05-12 16:28:56 ok 2015-05-12 17:42:21 hi 2015-05-12 17:42:27 first CI20 arrived 2015-05-12 18:07:33 planning on porting Alpine to MIPS? 2015-05-12 18:33:20 hello everybody 2015-05-12 18:34:15 I have a little problem with some iptables ruleset 2015-05-12 18:35:22 I installed fwsnort which translates snort rules in iptables rules 2015-05-12 18:36:20 when I run fwsnort, the rules are translated succesfully 2015-05-12 18:38:41 but when I run command iptables-restore < /var/lib/fwsnort/fwsnort.save 2015-05-12 18:39:04 I get an error at last line COMMIT 2015-05-12 18:39:32 iptables-restore -t < /var/lib/fwsnort/fwsnort.save ? 2015-05-12 18:39:37 tried to restore fwsnort.save on another linux distro 2015-05-12 18:39:52 and is working 2015-05-12 18:40:04 I didn't try -t 2015-05-12 18:40:47 blind guess is you are missing a netfilter module 2015-05-12 18:41:28 But when I run fwsnort --ipt-sync the generated fwsnort.save is working 2015-05-12 18:41:56 with much less rules, obviously 2015-05-12 18:43:18 fwsnort recommendations don't include any -t option 2015-05-12 18:43:37 I will try it in one minute 2015-05-12 18:45:25 "-t" is "test these rules" 2015-05-12 18:47:15 crashedthe system, I can't issue any comand, even if I have the prompt 2015-05-12 18:47:50 mmm... I have /never/ seen that result 2015-05-12 18:48:13 can you try to replicate it please 2015-05-12 18:48:36 I'd need your ruleset. 2015-05-12 18:48:46 and which version of alpine? 2015-05-12 18:48:51 313 2015-05-12 18:48:57 3.1.3 2015-05-12 18:49:34 the ruleset is huge, I have no opened ports 2015-05-12 18:49:43 20000 lines 2015-05-12 18:49:51 mmm 2015-05-12 18:50:04 sounds like a place for "ipset" 2015-05-12 18:50:20 ? 2015-05-12 18:50:43 http://ipset.netfilter.org/ 2015-05-12 18:51:24 the problem is that rules generated by fwsnort are restored succesfully on a Puppy Linux distro 2015-05-12 18:52:05 puppy doesn't use a grsec/pax kernel 2015-05-12 18:52:19 yes, this is true 2015-05-12 18:52:32 dunno why it crashes your box tho. 2015-05-12 18:52:41 but with fewer rules is ok 2015-05-12 18:53:01 with 500 rules is working 2015-05-12 18:53:08 not with 20000 2015-05-12 18:53:41 and -t crashes the box. 2015-05-12 18:53:48 is there any way to have fwsnort rules as ruleset in iptables? 2015-05-12 18:54:08 yes 2015-05-12 18:54:18 I'd recommend filing a bug at bugs.alpinelinux.org 2015-05-12 18:54:37 I will...but for iptables or fwsnort? 2015-05-12 18:54:46 iptables 2015-05-12 18:54:52 I think is not fwsnort problem 2015-05-12 18:55:37 maybe not even iptables but grsec kernel 2015-05-12 18:56:01 i will try it on ARM which has not grsec kernel 2015-05-12 18:57:04 I will fill a bug...hope I will not do it in the wrong place 2015-05-12 19:00:55 the admins can move it if its in the wrong place. Just report it. :) 2015-05-12 19:01:19 thanks 2015-05-12 21:00:22 nangel: yes 2015-05-12 21:00:35 nangel: and running a couple of builders, but my order is only partially filled at the moment. 2015-05-12 21:03:39 cool. And I was about to recycle my fleet of Netgear WGT634U 2015-05-12 21:05:53 right now the setup is very rigged 2015-05-12 21:05:59 usb hubs galore 2015-05-12 21:06:01 bbl 2015-05-13 06:21:16 ---> Making bundle: dynbinary (in bundles/1.6.1/dynbinary) 2015-05-13 06:21:18 Created binary: /home/buildozer/aports/main/docker/src/docker-1.6.1/bundles/1.6.1/dynbinary/dockerinit-1.6.1 2015-05-13 06:21:20 # code.google.com/p/gosqlite/sqlite3 2015-05-13 06:21:22 /usr/lib/gcc/x86_64-alpine-linux-musl/4.9.2/../../../../x86_64-alpine-linux-musl/bin/ld: cannot find -lsqlite3 2015-05-13 06:21:24 collect2: error: ld returned 1 exit status 2015-05-13 06:21:26 >>> ERROR: docker: all failed 2015-05-13 06:21:28 >>> docker: Uninstalling dependencies... 2015-05-13 06:27:18 hi. Will consul be in 3.2? (i'm using it from @testing for about 2 weeks now) 2015-05-13 06:28:47 hum 2015-05-13 06:30:08 why does docker fail? 2015-05-13 06:30:13 sqlite-dev is in makedepends 2015-05-13 06:30:22 Jean-Scotch: I can look at consul soonish 2015-05-13 06:34:09 ok sqlite-dev is broke 2015-05-13 06:34:46 ncopa: I'm a very happy consul user on AL ;) and I'm writing a wiki page about it (currently in my talk section but will move it to main wiki when consul will move from edge/testing to edge/main) 2015-05-13 07:00:23 Jean-Scotch: i have a question about consul APKBUILD 2015-05-13 07:00:34 there are releases here: https://github.com/hashicorp/consul/releases 2015-05-13 07:00:39 but the APKBUILD uses a snapshot 2015-05-13 07:01:40 they also shipped a v0-5-0.json file: http://git.alpinelinux.org/cgit/aports/tree/testing/consul/APKBUILD#n18 2015-05-13 07:01:52 this one: http://git.alpinelinux.org/cgit/aports/tree/testing/consul/v0-5-0.json 2015-05-13 07:02:04 where does it come from? 2015-05-13 07:02:09 is it made manually? 2015-05-13 07:02:15 is it generated automatically? 2015-05-13 07:02:16 yes. it seems a bit complicated to build a go package. It was coredumb work. I'm trying to do the same for envconsul (without success yet) 2015-05-13 07:02:20 you sure it's part of tarball? 2015-05-13 07:02:21 are they any firewall on alpinelinux 3.1 vanilla ? 2015-05-13 07:02:25 we took it from git repository 2015-05-13 07:02:36 how do i generate a v0-5-1.json file? 2015-05-13 07:03:02 crow: you can apk add iptables or apk add awall 2015-05-13 07:03:08 oh 2015-05-13 07:03:14 ncopa: deps/ 2015-05-13 07:03:17 maybe you cannot use awall due to xtables-addons 2015-05-13 07:03:20 ncopa i dont need them just trying to debug issue i have. 2015-05-13 07:03:24 deps/v0-5-1.json 2015-05-13 07:03:42 no idea why someone added it to source array, it should be downloaded 2015-05-13 07:03:51 ok 2015-05-13 07:04:09 clandmeter knows more details, but we have own snapshot because go is retarded 2015-05-13 07:04:09 there are some sample files too there 2015-05-13 07:04:18 and fails to pull all deps with go get 2015-05-13 07:04:35 yep 2015-05-13 07:04:58 releases that can't be built without proper go deps file 2015-05-13 07:05:00 -_- 2015-05-13 07:05:09 it causes timeout here, so build is stuck until go get failed 2015-05-13 07:05:32 so we cannot use https://github.com/hashicorp/consul/archive/v0.5.1.tar.gz ? 2015-05-13 07:05:45 ncopa: nope 2015-05-13 07:05:46 well, clandmeter said it doesn't cause timeouts for him 2015-05-13 07:05:53 iirc 2015-05-13 07:05:53 barthalion: for me neither 2015-05-13 07:06:03 so feel free to ignore my gripe :P 2015-05-13 07:06:47 ncopa: actually yes it's possible but to compile it you must have the go deps files 2015-05-13 07:07:10 else go will fetch deps that are of a version too recent for the release 2015-05-13 07:07:19 it's nasty clearly -_- 2015-05-13 07:08:21 but the only way to compile it 2015-05-13 07:08:36 i think clandmeter regrouped everything in the snapshot 2015-05-13 07:08:42 deps + consul 2015-05-13 07:09:00 Rob Pike must hate packagers 2015-05-13 07:10:47 clearly 2015-05-13 07:11:25 I thought java and Ruby were the biggest packager haters but Go is actually a nice contender :D 2015-05-13 07:11:37 go wins easily with ruby 2015-05-13 07:12:26 being a Ruby hater myself i must be be biased :P 2015-05-13 07:12:44 ncopa: do you mind some late pushes to testing and then to main? what's the schedule for release? 2015-05-13 07:14:06 barthalion: i was hoping to get 3.2 out this week 2015-05-13 07:14:21 current blockers are qt5 conflict with qt4 2015-05-13 07:14:40 and openrc 'default' runlevel not pulling in services in 'boot' runlevel 2015-05-13 07:15:15 i dont mind pushing new to testing and then move to main as long as it does not have impact on current 'main' 2015-05-13 07:15:23 and does not have impact on toolchain 2015-05-13 07:15:44 so we dont need rebuidl alot of packages in main already 2015-05-13 07:15:46 yeah, it won't 2015-05-13 07:16:47 then i dont mind 2015-05-13 07:16:57 any volunteers to maintain consul? 2015-05-13 07:17:38 i think we need a comment in consul APKBUILD saying: # we cannot use because so we need create our own snapshot 2015-05-13 07:18:01 # to build new snapshot do: 2015-05-13 07:18:07 ncopa: well i created it i should maintain it 2015-05-13 07:18:49 coredumb: how do you generate snapshot for v0.5.1? 2015-05-13 07:19:22 dunno clandmeter did the snapshot part :D 2015-05-13 07:19:28 i'll check it and see 2015-05-13 07:19:48 must need the go deps json file for 0.5.1 2015-05-13 07:19:56 https://github.com/hashicorp/consul/blob/master/deps/v0-5-1.json 2015-05-13 07:20:38 yep 2015-05-13 07:21:06 clandmeter: did you need to manually edit the v0-5-0.json file? 2015-05-13 07:22:02 don't think he did 2015-05-13 07:22:35 then i wonder why he didnt just use the v0-5-0.json from the git clone? 2015-05-13 07:25:44 are some rights needed for oscam to "open" an port localy, netstat does now show me listenning port i specified in oscam.conf 2015-05-13 07:33:13 morning 2015-05-13 07:34:09 ncopa: the file is not in the release afaik 2015-05-13 07:34:21 its in master not in the tag 2015-05-13 07:36:18 $ curl -L --silent https://github.com/hashicorp/consul/archive/v 2015-05-13 07:36:18 0.5.1.tar.gz | tar -zt | grep deps 2015-05-13 07:36:18 consul-0.5.1/deps/ 2015-05-13 07:36:18 consul-0.5.1/deps/v0-3-0.json 2015-05-13 07:36:18 consul-0.5.1/deps/v0-3-1.json 2015-05-13 07:36:19 consul-0.5.1/deps/v0-4-0.json 2015-05-13 07:36:21 consul-0.5.1/deps/v0-4-1.json 2015-05-13 07:36:23 consul-0.5.1/deps/v0-5-0.json 2015-05-13 07:37:20 the release doesnt have a a git dir 2015-05-13 07:37:26 i think go neesd it somehow 2015-05-13 07:37:38 its a mess 2015-05-13 07:38:48 oh now i remember i think, i removed 2 repo's from the file. 2015-05-13 07:40:11 ncopa: https://github.com/hashicorp/consul/tree/0c7ca91c74587d0a378831f63e189ac6bf7bab3f/deps 2015-05-13 07:40:17 clandmeter: yeah release fail to compile if there's no .git dir 2015-05-13 07:40:29 they didnt include a proper deps file for 0.5.0 2015-05-13 07:40:30 that's why i initially added a patch to bypass that 2015-05-13 07:40:40 but clearly it's all a MESS 2015-05-13 07:41:07 go is complete mess, it even has undocumented switches. (or im blind and cannot find them) 2015-05-13 07:41:21 ok 2015-05-13 07:41:27 do we want maintain that? 2015-05-13 07:41:30 releasing a version that can't even compile because of a .git/ directory... come on 2015-05-13 07:41:54 ncopa: if you don't I will anyway on an alternate repo 2015-05-13 07:41:57 well go is completely git depended. 2015-05-13 07:41:59 so much needed ^^ 2015-05-13 07:42:13 so its better to generate a snapshot for each release 2015-05-13 07:42:21 i am ok to ship it in alpine main, if 2015-05-13 07:42:24 ncopa: well, coredumb wants to be maintainer, so why not? 2015-05-13 07:42:32 i agree with clandmeter 2015-05-13 07:42:34 if there is a comment in the APKBUILD how to generate the snapshot 2015-05-13 07:43:04 the build instructions also dont work 2015-05-13 07:43:17 so i had to do a regular go build, and modify some variables 2015-05-13 07:43:23 there is a snapshot() function, but it needs to explain why the v$pkgver.json file needs to be shipped separately 2015-05-13 07:43:45 i would prefer not ship all those .sample files too if possible 2015-05-13 07:43:45 it doesnt, if its available 2015-05-13 07:43:58 that would need to be cleaned up then 2015-05-13 07:44:29 we have to prepare for more of this crap 2015-05-13 07:44:49 if the file needs to be manually edited, then we need a comment how to do it 2015-05-13 07:44:57 I think the .sample files could go on the wiki instead of in the package... 2015-05-13 07:45:04 or even a sed line 2015-05-13 07:45:20 ncopa: it doesnt need to be edited, if they dont break remote git repo's 2015-05-13 07:45:40 there is no std rule to build it. 2015-05-13 07:45:41 oh my so .sample files made with love :( 2015-05-13 07:45:42 Jean-Scotch: its kinda nice to have minimal wokring config so we can smoke test it 2015-05-13 07:46:04 coredumb: i am ok to keep them if needed 2015-05-13 07:46:09 could i put them in /usr/share/consul ? 2015-05-13 07:46:26 i should have put them there first i guess 2015-05-13 07:46:27 i just want make sure that its not just a raw copy of some file that is shipped with the sources 2015-05-13 07:46:33 like they did with zabbix 2015-05-13 07:46:40 +1 2015-05-13 07:46:46 ncopa: no it's not 2015-05-13 07:46:53 it's my own config examples 2015-05-13 07:47:08 then i think we can keep them 2015-05-13 07:47:10 everytihg should come from src when possible, and patch it if needed. 2015-05-13 07:47:16 and those are good practice ones ;) 2015-05-13 07:47:17 yes 2015-05-13 07:47:33 but i totaly agree on putting them somewhere else 2015-05-13 07:47:33 sample configs are good 2015-05-13 07:47:36 abuild should yell if its broken :) 2015-05-13 07:47:42 /usr/share/consul should be better 2015-05-13 07:50:07 $ consul 2015-05-13 07:50:07 -ash: consul: Permission denied 2015-05-13 07:50:28 -rwxr-x--- 1 root consul 12191472 Apr 28 11:30 /usr/sbin/consul 2015-05-13 07:50:31 yes 2015-05-13 07:50:35 only root can read the file? 2015-05-13 07:50:36 much needed 2015-05-13 07:50:44 are there secrets in the consul binary? 2015-05-13 07:51:13 you don't want any user to be able to execute it 2015-05-13 07:51:28 ok 2015-05-13 07:51:41 cause executing it means you can run commands on the whole cluster 2015-05-13 07:51:51 whatta 2015-05-13 07:52:10 :| 2015-05-13 07:52:21 so limiting root and consul group - which includes consul user which is running the service - sounds like the only sane way 2015-05-13 07:52:30 now i got impressed 2015-05-13 07:52:41 is the entire cluster config compiled in to the binary? 2015-05-13 07:52:57 that's why there's .sample config preventing remote command execution 2015-05-13 07:53:17 that means that I can run commands on your cluster 2015-05-13 07:53:25 if you're joined to it yes 2015-05-13 07:53:47 how do i join to it? 2015-05-13 07:53:57 if other nodes are not configured to refuse remote commands 2015-05-13 07:54:05 config file + service running 2015-05-13 07:54:08 by entering the proper json data? 2015-05-13 07:54:10 ok 2015-05-13 07:54:28 then why dont we just change read permission on the config file? 2015-05-13 07:54:39 so it means that once service started any user of the system could, using the consul binary get/set infos on the cluster 2015-05-13 07:55:08 even if they dont have the read permission of the config? 2015-05-13 07:55:13 yes 2015-05-13 07:55:31 what if a local user as non-root downloads the consul binary from web? 2015-05-13 07:55:36 and installs it locally 2015-05-13 07:55:49 and runs his own built consul binary? 2015-05-13 07:55:56 you should prevent your users to do that 2015-05-13 07:56:03 TPE 2015-05-13 07:56:14 sadly consul uses RPC 2015-05-13 07:56:36 so as long as it runs, any consul client connecting to the default RPC port would get access 2015-05-13 07:57:13 i got no other solutions :) 2015-05-13 07:57:32 is it just me who thinks that something is fundamentally wrong with this? 2015-05-13 07:57:43 no i do as well 2015-05-13 07:57:43 no 2015-05-13 07:57:51 but i control my environnement 2015-05-13 07:58:06 so binary rights is sufficient 2015-05-13 07:59:23 i just htink that disabling users from reading consul binary does not solve anything really 2015-05-13 07:59:41 another way of doing things is to use iptables owner module to prevent other users than consul one to access those ports 2015-05-13 07:59:42 as long as they can connect to default RPC port with their own client written in C 2015-05-13 08:01:08 i think it should been enough to just read protect the configs 2015-05-13 08:01:20 if thats not enough, something is fundamentally wrong 2015-05-13 08:01:33 <[xming]> morning 2015-05-13 08:02:33 morning 2015-05-13 08:03:46 <[xming]> is it possible to make /etc/freeswitch for freeswitch and freeswitch-sample-confg owned by freeswitch:freeswitch instead of root:root? 2015-05-13 08:04:07 [xming]: does freeswitch need to modify those at run time? 2015-05-13 08:05:14 <[xming]> yes and a lot of 3rd party GUI needs to modify them, so making php-fpm to run as freeswitch:freeswitch will make t hat possible 2015-05-13 08:05:42 <[xming]> I still dislike freeswitch and all those 3rd party GUI interms of security 2015-05-13 08:05:51 <[xming]> but currently it's the only way 2015-05-13 08:06:48 <[xming]> even /usrsharfreeswitch/sounds needs to be freeswitch:freeswitch but I don't like that 2015-05-13 08:07:14 <[xming]> so maybe freeswitch-sounds should be moved to /var/lib/freeswitch/sounds 2015-05-13 08:07:32 yes, sounds like that 2015-05-13 08:07:36 sounds like it needs to move to /var 2015-05-13 08:08:39 <[xming]> ncopa: can I make those changes then send a PR? 2015-05-13 08:08:46 also, re php needing write permission 2015-05-13 08:09:11 the files could have www-data group have write access 2015-05-13 08:09:25 <[xming]> I run php-fpm one pool with uid freeswitch 2015-05-13 08:09:33 ok 2015-05-13 08:09:57 <[xming]> I think that php-fpm + pools with diff uid is a better solution 2015-05-13 08:10:16 i think there was an issue filed for dir locations of freeswitch already 2015-05-13 08:10:32 https://bugs.alpinelinux.org/issues/4065 2015-05-13 08:11:37 i dont think freeswithc has support for the needed options at compiletime 2015-05-13 08:11:46 but i think they have the options at runtime 2015-05-13 08:11:59 so we might be able to edit the init.d accordingly 2015-05-13 08:12:05 <[xming]> I will check that 2015-05-13 08:12:31 <[xming]> FS like to put everything /PATH/freeswitch 2015-05-13 08:13:09 <[xming]> I used to put everything in /opt/freeswitch then symlink 2015-05-13 08:13:35 <[xming]> FS people are not sysadmin but telco admin, different kind of mindset :p 2015-05-13 08:14:36 yeah 2015-05-13 08:14:37 <[xming]> ncopa: I am currently installing fusionpbx, once t hat done if there are needs I will *try* to package it 2015-05-13 08:14:42 <[xming]> it's a hell 2015-05-13 08:15:07 <[xming]> ncopa: do you know anything about PHP's GLOB_BRACE not supported on musl? 2015-05-13 08:15:42 <[xming]> took me a week to figure out why fusionpbx doens't work and made a patch for it 2015-05-13 08:16:22 i didnt know that 2015-05-13 08:16:30 could you please file bug about it? 2015-05-13 08:16:59 <[xming]> it's know that some sunOS/solaris and uclibc do not have that 2015-05-13 08:17:11 sounds like a gnu extension 2015-05-13 08:17:33 <[xming]> according to php.net, it doesn't work on some non GNU system 2015-05-13 08:17:40 <[xming]> but apparently BSDs have that 2015-05-13 08:17:51 <[xming]> hench my honest question 2015-05-13 08:18:02 musl devs has normally good reasons to not add support for such things 2015-05-13 08:18:54 <[xming]> I understand, but it's weird, php could just use prce to achive t hat 2015-05-13 08:19:18 <[xming]> well php is weird :D 2015-05-13 08:21:21 yeah :) 2015-05-13 08:21:34 :) 2015-05-13 08:22:01 <[xming]> Phobia Home Page 2015-05-13 08:22:20 i think php disables sha512 and sha256 2015-05-13 08:22:25 on musl 2015-05-13 08:22:42 the configure check disables it if you have support for blowfish or similar 2015-05-13 08:22:44 its a bug 2015-05-13 08:22:52 php does stupid things 2015-05-13 08:23:23 <[xming]> we are lucky that non GNU, non x86 system are getting more popular 2015-05-13 08:23:46 <[xming]> otherwise it's damn difficult got get those things fixed upstream 2015-05-13 08:24:05 yes 2015-05-13 08:24:18 @ncopa | if thats not enough, something is fundamentally wrong < agree 2015-05-13 08:25:06 <[xming]> I still need to migrate all my home VMs to alpine, that project got side tracked by RL 2015-05-13 08:25:26 <[xming]> now I feel that I am doing everything from the beginning again 2015-05-13 08:26:49 <[xming]> lmao -> http://www.openwall.com/lists/musl/2013/09/15/3 2015-05-13 08:29:38 <[xming]> looks like that the lack of GLOB_BRACE is known http://git.alpinelinux.org/cgit/aports/plain/main/task/musl-fix.patch 2015-05-13 08:30:56 https://bug.tasktools.org/browse/TW-342 2015-05-13 08:31:01 apparently yes :) 2015-05-13 08:31:11 i wonder why we dont need patch for php 2015-05-13 08:31:18 they probably just silently ignore it 2015-05-13 08:31:48 <[xming]> probably configure check, and disabled that 2015-05-13 08:32:33 <[xming]> and the web GUI hides all error, even with E_ALL I couldn't see PHP spewing any eorr :( 2015-05-13 08:33:18 <[xming]> ACTION notes to himself, if some php apps won't work, check php weirdness on non GNU 2015-05-13 08:42:25 <[xming]> ncopa: what kind of h/w is running for the ARM compiles? 2015-05-13 08:44:07 [xming], wandboard quad 2015-05-13 08:46:45 <[xming]> oh nice board, 2GB RAM, I haven't much affordable SOCs with more than 1GB 2015-05-13 08:49:02 [xming], yeah, wandboard was the most feasible we started with a year ago or so. most ram, and quad core 2015-05-13 08:49:07 and ok price 2015-05-13 08:49:22 could use more ram; we need swap when compiling few of the big packages 2015-05-13 08:50:35 <[xming]> I can imaging that, saw some 4GB ARM going to be on sale 2015-05-13 08:50:46 <[xming]> I am one with at least 8GB :D 2015-05-13 10:01:39 ncopa: http://www.engadget.com/2015/05/12/firefox-38 i guess thats windows only? 2015-05-13 10:02:23 i dont know 2015-05-13 10:02:32 Currently, Adobe Primetime is only available in Microsoft Windows Vista and above when using 32-bit versions of Firefox. Mac OS X, Linux, Windows XP and 64-bit versions of Firefox are currently not supported. 2015-05-13 10:02:48 in the chrome case there was only 2 binary plugins needed 2015-05-13 10:02:57 that was closed source 2015-05-13 10:03:09 might be we can make it work with firefox 38 too 2015-05-13 10:03:17 if its the same binary plugins 2015-05-13 10:03:28 might be musl has good enough glibc compat for it too... 2015-05-13 10:03:38 which would be awesome 2015-05-13 10:37:14 \SB 2015-05-13 12:20:57 firefox 38 is an esr 2015-05-13 12:30:26 extra special release? 2015-05-13 12:34:58 clandmeter: yes. longtime support 2015-05-13 12:35:04 which means we can update main 2015-05-13 12:37:39 "Base for the next ESR release" 2015-05-13 12:37:42 im running from edge, so i guess no benefit here. 2015-05-13 12:38:23 but I am not sure, what they mean about in terms of LTS 2015-05-13 12:38:57 ncopa: question that rings my mind, how do you keep your time correct on your alpine desktop? mine is always wrong after sleep or reboot. 2015-05-13 12:45:34 https://www.mozilla.org/en-US/firefox/organizations/faq/ 2015-05-13 12:49:56 listener.c:(.text+0x3ee): undefined reference to `freehostent' 2015-05-13 12:50:17 any idea what provides it? 2015-05-13 12:57:21 "the organizations" page is all about ESR https://www.mozilla.org/en-US/firefox/organizations/ 2015-05-13 12:58:03 ok got it 2015-05-13 12:59:38 barthalion: that's an old deprecated interface, not sure if even glibc provides it anymore 2015-05-13 12:59:50 uh-uh 2015-05-13 13:00:09 I should buy a boat 2015-05-13 13:01:20 barthalion: it should normally be using get{addr,name}info() 2015-05-13 13:47:39 clandmeter: i use chrony 2015-05-13 13:47:51 but i dont know if if timesync works properly though 2015-05-13 13:47:58 i tried both 2015-05-13 13:48:13 chrony and ntpd 2015-05-13 13:49:13 ncopa: i guess your time is correct? 2015-05-13 13:49:22 sometimes :) 2015-05-13 13:49:29 ithink its set at boot 2015-05-13 13:49:56 i normally poweroff rather than suspend 2015-05-13 13:50:08 might also be that mi hw clock is more or less correct 2015-05-13 13:50:28 i guess its related to hwclock 2015-05-13 13:50:46 maybe because windows and linux set the clock different 2015-05-13 13:52:05 ncopa: have you added hwclock to your runlevel? 2015-05-13 13:52:27 i dont remember 2015-05-13 13:52:31 it might be the openrc thingy 2015-05-13 13:52:48 that openrc does not pull in 'boot' runlevel 2015-05-13 13:52:53 it used to do that automatically 2015-05-13 13:59:28 stateless: it's shitty ancient static library, so yeah, I can see why it's broken 2015-05-13 17:45:02 clandmeter you could set windows to utc to so there should then be no problem on booting windows and then into linux. 2015-05-14 09:16:22 wow 2015-05-14 09:16:29 openrc got support for s6 supervision 2015-05-14 09:17:25 cool 2015-05-14 09:17:32 we have it in testing 2015-05-14 09:17:40 nice 2015-05-14 09:31:08 mmmh virt-manager is broken on latest edge 2015-05-14 10:54:45 coredumb: i use virt-manager here? 2015-05-14 10:54:48 how is it broke? 2015-05-14 13:04:16 there is now an atom.xml on http://alpinelinux.org 2015-05-14 13:04:53 click on the news icon 2015-05-14 14:23:36 argh 2015-05-14 14:23:52 what is the current best practice for bootstrapping alpine? :) 2015-05-14 14:26:13 ah 2015-05-14 15:09:32 kaniini: we are in toolchain freeze. can you wait til after v3.2 is out? 2015-05-14 15:09:43 ncopa: oh. i thought 3.2 was already branched 2015-05-14 15:09:53 hum that change looks safe though 2015-05-14 15:09:56 no, its not branched 2015-05-14 15:10:54 mips.. 2015-05-14 15:10:57 awesome! 2015-05-14 15:11:02 have a nice evenin 2015-05-14 15:17:17 well this stuff is targeted for 3.3 anyway ;) 2015-05-14 15:25:45 i'll push the rest after 3.2 is branched 2015-05-14 15:36:05 rnalrd: you about? 2015-05-14 15:41:39 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62097 2015-05-14 15:41:41 yay 2015-05-14 15:41:43 :| 2015-05-14 16:00:16 also, 2015-05-14 16:00:18 /usr/mipsel-alpine-linux-musl/bin/ranlib: can't set BFD default target to `mipsel-alpine-linux-musl': Invalid bfd target 2015-05-14 16:00:20 :| 2015-05-14 16:59:55 ncopa: Error launching details: : unknown signal name: state-changed 2015-05-14 17:00:15 when double clicking any VM from any connection 2015-05-14 17:21:16 what this? /usr/lib/libX11.so.6: error adding symbols: DSO missing from command line (when building an aport) 2015-05-14 17:51:46 libx11-dev ? 2015-05-14 17:51:48 ScrumpyJack, ^^ 2015-05-14 17:52:33 missing flag to linker 2015-05-14 17:53:04 add -lX11 to C(XX)FLAGS 2015-05-14 17:53:16 or even better fix it in the project 2015-05-14 21:32:35 anyone tried KSM on alpine? 2015-05-14 23:33:34 barthalion: yes 2015-05-14 23:33:37 barthalion: it works fine 2015-05-15 06:02:04 cool 2015-05-15 07:00:47 barthalion: thanks, that helped 2015-05-15 07:01:23 no problem 2015-05-15 09:22:25 <[xming]> how do I list all the installed files by a package? 2015-05-15 09:23:01 apk info -L iirc 2015-05-15 09:23:53 <[xming]> yes that's it, ty 2015-05-15 09:24:44 <[xming]> it seems that I could never find the right siwthces/args I want with apk :( 2015-05-15 10:02:59 anyone know of "independant" clock system tray app and volume control system tray app, written in C 2015-05-15 10:12:33 ncopa thank you for the #4103 will test it after. 2015-05-15 10:23:13 can i see what packages are waiting for upgrade? if I do sudo apk update && sudo apk upgrade it will install them immediately 2015-05-15 10:30:19 --simulate 2015-05-15 10:31:15 -s 2015-05-15 10:31:25 (ninja'd by clandmeter) 2015-05-15 10:32:49 i've also got a question: new package has arrived on mirror, but apk add package says it's missing 2015-05-15 10:33:16 are you sure your mirror has it? 2015-05-15 10:33:25 yes, it's on the filesystem 2015-05-15 10:33:31 apk update shows v3.2.0_rc2-127-g9b68c74 2015-05-15 10:33:46 which repo do you use? 2015-05-15 10:33:56 here is the package http://liskamm.alpinelinux.uk/edge/testing/x86_64/owncloud-client-1.8.1-r0.apk 2015-05-15 10:34:21 my repo is http://liskamm.alpinelinux.uk/edge/main and http://liskamm.alpinelinux.uk/edge/testing 2015-05-15 10:34:57 perhaps if i pin the second repo, it might help? 2015-05-15 10:35:10 apk search -v package 2015-05-15 10:35:16 does it show the correct version? 2015-05-15 10:35:57 doesnt come up in a search 2015-05-15 10:36:03 you only need to pin testing if you have package name conflicts 2015-05-15 10:36:42 for instnace firefox is both in main and testing 2015-05-15 10:36:43 owncloud-client isn't in main so no confilt there 2015-05-15 10:37:07 if apk search doesnt list it, then its not in the apkindex 2015-05-15 10:37:51 i'll check 2015-05-15 10:38:21 owncloud-client-1.8.1-r0 2015-05-15 10:38:29 i have it here 2015-05-15 10:38:35 it's in the apkindex :( 2015-05-15 10:39:07 my apk repo file http://sprunge.us/NHVJ 2015-05-15 10:40:43 I can add other packages that are only in testing, like aha-doc 2015-05-15 10:42:11 i have two apkindex files locally. i presume one for each repo? 2015-05-15 10:42:21 can i delete them ? 2015-05-15 10:43:46 yup, that worked 2015-05-15 10:44:02 wierd 2015-05-15 10:46:37 if i delete the new apkindex files and replace with the old ones, i can't update them 2015-05-15 10:49:56 ok, i think i get it 2015-05-15 10:53:21 strace shows that apk does a newfstatat in the apkindex file (which i presume checks the time of the local apkindex) a 2015-05-15 10:53:52 does it compare it with the ctime of the apkindex on the mirror or something? 2015-05-15 10:54:05 the mirror is two hours ahead 2015-05-15 10:55:20 crow: apk upgrade -U -a -i 2015-05-15 10:55:31 for interactive 2015-05-15 10:59:07 i am very tempted to squeeze in xen 4.5 2015-05-15 10:59:37 yes please! 2015-05-15 11:00:19 apk uses libfetch and sends If-Modified-Since: 2015-05-15 11:00:45 and gets back: HTTP/1.1 304 Not Modified 2015-05-15 11:00:48 hmm 2015-05-15 11:01:01 interesting 2015-05-15 11:01:05 fabled: ^^^ 2015-05-15 11:01:09 which url? 2015-05-15 11:01:33 hmm 2015-05-15 11:01:42 the APKINDEX files for edge/main and edge/testing on liskamm 2015-05-15 11:01:50 sounds like mix up in local time / UTC 2015-05-15 11:02:05 apk update --force skips the If-Modified-Since detection 2015-05-15 11:02:30 cool 2015-05-15 11:02:32 thanks 2015-05-15 11:04:24 i think the setup-apkrepos does that 2015-05-15 11:04:45 gives error on liskamm mirror 2015-05-15 11:04:53 it downloads from all mirrors 2015-05-15 11:05:04 what sort of error? 2015-05-15 11:05:09 ERROR: ... 2015-05-15 11:05:15 thats the only it says 2015-05-15 11:05:22 i suppose that is the problem 2015-05-15 11:05:27 file is alreayd there 2015-05-15 11:05:43 which file? 2015-05-15 11:05:54 APKINDEX.tar.gz 2015-05-15 11:06:21 it should be there right? on the mirror? 2015-05-15 11:06:24 it does apk update ... || 2015-05-15 11:06:58 it *might* be that the local file, APKINDEX.tar.gz is there locally 2015-05-15 11:07:18 how do you mean? 2015-05-15 11:07:29 (it's debian) 2015-05-15 11:07:42 then apk does If-Modified-Since 2015-05-15 11:08:06 setup-apkrepos will try apk update on all mirrors on the list 2015-05-15 11:08:22 when trying liskamm mirror it give ERROR 2015-05-15 11:08:27 i never investigated why 2015-05-15 11:08:57 instead, i fixed apk and the script to ignore broken mirrors 2015-05-15 11:09:23 instead of 'oh this operation took only 0.00001 seconds! this is the fastest!' 2015-05-15 11:09:27 when the operation failed 2015-05-15 11:10:53 i'm not sure i fully follow why liskamm is broken, if indeed it is. 2015-05-15 11:13:58 thats kinda my point 2015-05-15 11:14:07 i am no longer sure if it is broken 2015-05-15 11:14:44 it sounds like the error comes from apk because the the liskamm mirror honor If-Modified-Since 2015-05-15 11:14:57 i thought it was broke 2015-05-15 11:15:16 but it sounds like it just says: HTTP/1.1 304 Not Modified 2015-05-15 11:15:53 btw, xen 4.5 builds 2015-05-15 11:17:55 alpxen:~# rc --sys 2015-05-15 11:17:55 XEN0 2015-05-15 11:17:55 \o/ 2015-05-15 11:18:15 and it boots in qemu 2015-05-15 11:18:29 i'll push it for v3.2 2015-05-15 11:18:53 i kinda like xen 2015-05-15 11:19:52 i'm very fond of it too. it's what brought me to alpine linux (when i was looking for a dom0) 2015-05-15 11:20:06 now alpine linux is even on my desktop 2015-05-15 11:20:13 cool :) 2015-05-15 11:20:25 you gotta be pretty crazy to run alpine on your desktop :) 2015-05-15 11:22:50 no no, it's brilliant! i use window managers, not a fan of the all-encompassing desktop-environment, so it works flawlessly, as i build the block myself, building packages as i go along. 2015-05-15 11:23:41 ScrumpyJack: "i use window managers, not a fan of the all-encompassing desktop-environment" -> you *are* crazy :) 2015-05-15 11:23:44 same here 2015-05-15 11:23:55 i'm crazy too and i love it :) 2015-05-15 11:25:48 hwere comes xen 4.5 2015-05-15 11:25:56 slim + twm + trayer ( + conky to make it pretty) 2015-05-15 11:26:07 xfce here 2015-05-15 11:26:32 ti have xfce4-terminal on your recommendatition. works well 2015-05-15 11:31:14 here is a scrot of alpine linux desktop http://st.ilet.to/scrot.png :) 2015-05-15 11:31:57 ScrumpyJack: why slim? is that a multi user system? 2015-05-15 11:32:14 nope, just me. office desktop 2015-05-15 11:32:31 so why a *dm then? 2015-05-15 11:32:45 *dm? 2015-05-15 11:32:55 the question is why not tty and startx :p 2015-05-15 11:32:56 i quit twm into slim 2015-05-15 11:33:04 [[ $(tty) = /dev/tty1 ]] && exec startx 2015-05-15 11:33:13 in my .zprofile ^^^ 2015-05-15 11:33:58 i use the dm like a lock screen with nice photo-themes :) 2015-05-15 11:34:45 aah, we differ here, i prefer `xlock -blank` 2015-05-15 11:35:26 aha! xlock isn't in the repos yet! you should share :) 2015-05-15 11:36:06 i like xtrlock 2015-05-15 11:36:27 and slim when i leave my desk :) 2015-05-15 11:38:28 we should get some neat screenshots of desktops together for alpinelinux.org 2015-05-15 11:39:56 https://i.imgur.com/uomkVIL.png this is how my alpine desktop looks 2015-05-15 11:40:06 although 404 error fits better 2015-05-15 11:42:42 lol. black png for me. if it's a joke, it's funny :) 2015-05-15 11:47:25 ah, everytime i share it turns out my kind of sharing is not the kind that is immediately welcome without wd40... :/ 2015-05-15 11:50:57 ugh 2015-05-15 11:51:00 now what... 2015-05-15 12:00:14 CMakeFiles/sems.dir/AmSipSubscription.o: In function `atomic_int64::get()': 2015-05-15 12:00:26 /home/ncopa/aports/testing/sipwise-sems/src/sems-mr3.8.1.1/core/./atomic_types.h:124: undefined reference to `__sync_bool_compare_and_swap_8' 2015-05-15 12:20:58 ncopa: iirc, i had the same issue with mono and had to raise arch. 2015-05-15 12:26:43 ScrumpyJack, did had a chance to test setup-box ? 2015-05-15 12:30:32 alacerda: i have a perfect usecase for setup-box now 2015-05-15 12:30:42 where can i get it? 2015-05-15 12:59:56 ncopa, it is in testing 2015-05-15 12:59:58 repo 2015-05-15 13:00:01 ncopa: apk add setup-box :) 2015-05-15 13:00:26 alacerda: i'll try it now too, i have a hour spare 2015-05-15 13:00:36 great! 2015-05-15 14:01:09 you ppl use for desktop the grsec or vanilla version of alpinelinux? 2015-05-15 14:03:31 i use grsec 2015-05-15 14:07:17 any reason to not do rc3 nw? 2015-05-15 14:07:18 now* 2015-05-15 15:06:18 what we have on AL for xen is unmodified guests only ? 2015-05-15 15:07:40 http://bhyvecon.org/bhyvecon2015-Roger.pdf nice to read from Roger 2015-05-15 15:36:53 vkrishn: Xen on AL can run any kind of guest (PV or HVM) 2015-05-15 15:37:21 vkrishn: thanks, glad you enjoyed the slides :) 2015-05-15 16:08:55 ncopa, what do you think about the ideia of an script execute "lbu commit" by itself ? 2015-05-15 16:09:12 after doing its job 2015-05-15 16:40:56 I have not tried installing but does "PV guests" require the kernel to be modified ? 2015-05-15 16:46:29 I mean vanilla linux (modified for xen as PV) ? 2015-05-15 16:48:25 vkrishn: not anymore (since 3.0 or even before), you just need to enable the options in the kernel, for exmaple the AL kernel is compiled with all that already, so it can run as PV or HVM guest 2015-05-15 16:49:58 ok :) 2015-05-15 16:51:02 ok found it kernelconfig.x86_64, :) looked in wrong place, kernelconfig.x86 2015-05-15 16:51:08 thanks 2015-05-15 16:53:24 royger: how does xen look nowadays? better than kvm? 2015-05-15 16:56:06 barthalion: depends on who you ask I guess. They are different beasts so it's hard to compare. It all biolds down to your workload I guess 2015-05-15 16:58:07 I don't need anything fancy, just a couple of general purpose virtual machines 2015-05-15 16:58:44 I'll probably use alpine for that, although smartos looks good (but no KSM :() 2015-05-15 17:06:45 royger, do you have a website where you store your slides (download format) ? also having hard to find the 4.5 manual 2015-05-15 17:07:59 though I am now downloading source now to check if it has the pdf 2015-05-15 17:09:18 vkrishn: you can find all the man pages and other docs online at http://xenbits.xen.org/ for each Xen release 2015-05-15 17:10:27 vkrishn: I've got some of them here https://people.freebsd.org/~royger/talks/ those are the more recent ones 2015-05-15 17:24:44 hi 2015-05-15 17:48:54 ncopa, https://github.com/jgm/cmark , also jgm has lua bindings for cmark 2015-05-15 18:04:06 do we use antyhing else on alpinelinux for aliases then the ~/.profile ? 2015-05-15 18:09:41 crow if you have some common settings, put them in /etc/profile 2015-05-15 18:12:08 vkrishn well this is just a single user system i guess on login it will load firstly /etc/profile then ~/.profile 2015-05-15 18:13:00 hmm.. does busybox->sh read ~/.profile / 2015-05-15 18:18:04 now I do `su - ` once I log in as root 2015-05-15 18:19:12 http://www.irclogger.com/.alpine-linux/2015-05-12#1431434368 2015-05-15 18:23:01 vkrishn i see so even a single user system if going to use root it may be better to use /etc/profile 2015-05-15 18:26:20 since `su - ` lets you re-read /etc/profile, adding a script there to read ~/. may be possible, but I have not tried 2015-05-15 18:27:34 but that is only for busybox->sh , if installing bash it probably get normal 2015-05-15 18:31:09 normal I mean as expected ;) 2015-05-15 18:34:46 dont need currtently anything then busybox ash 2015-05-15 18:34:53 thanks 2015-05-16 11:14:20 Hello everybody! 2015-05-16 11:16:34 I have a little question...is AlpineLinux ARM working on Raspberry Pi 2 (which is ARM v7 architecture)? I not...is there any plan to have a distro for RPi2? And are any plans to have GRSEC kernel on ARM platform? 2015-05-16 11:26:01 I think fabled mentioned something about rpi2 kernel with single kernel for all raspberries 2015-05-16 11:26:08 it might happen to be 3.2.0 2015-05-16 11:26:21 he also mentioned something about grsec, but I don't remember actually 2015-05-16 11:26:33 catch him on monday in CET 2015-05-16 19:13:18 hmm... where is rtorrent dissapear, http://pkgs.alpinelinux.org/packages?package=torrent&repo=all&arch=x86 2015-05-16 19:13:59 but is available for 'x86_64' 2015-05-16 19:17:49 probably need to do pkg count/compare between archs, though some we know are specific for particular arch 2015-05-17 00:09:23 ncopa, clandmeter: https://github.com/mvdan/pastecat something we could make use of 2015-05-17 00:09:35 except golang again :c) 2015-05-17 21:49:14 hmm, that is odd that rtorrent is missing in x86. 2015-05-17 22:46:24 e143087 2014-10-27 main/rtorrent: disable on x86. does not build [Natanael Copa] 2015-05-17 22:46:37 nobody noticed for 7 months 2015-05-18 06:20:38 telephony stack, seems active and nice, http://www.rowetel.com/blog/?page_id=452 2015-05-18 06:21:11 but no release yet, http://sourceforge.net/projects/freetel/?source=navbar 2015-05-18 06:46:01 fabled: so, i was trying to use your scripts to bring up mipsel port 2015-05-18 06:46:32 unfortunately it has a problem :/ 2015-05-18 06:46:34 http://turtle.dereferenced.org/~kaniini/toolchain-build.txt 2015-05-18 06:47:40 kaniini, it's been ~1 year since i used them to bootstrap arm 2015-05-18 06:48:07 ACTION nods 2015-05-18 06:48:11 but hum 2015-05-18 06:48:14 >>> ERROR: musl: up2date failed ? 2015-05-18 06:48:15 i was just wondering if you might have any idea about the error 2015-05-18 06:48:20 oh 2015-05-18 06:48:27 lemme see 2015-05-18 06:48:31 specifically, 2015-05-18 06:48:33 /usr/mipsel-alpine-linux-musl/bin/ranlib /home/kaniini/aports/main/gcc/pkg/gcc-pass2-mipsel-alpine-linux-musl/usr/lib/gcc/mipsel-alpine-linux-musl/4.9.2/libgcc.a 2015-05-18 06:48:35 /usr/mipsel-alpine-linux-musl/bin/ranlib: can't set BFD default target to `mipsel-alpine-linux-musl': Invalid bfd target 2015-05-18 06:48:53 seems gcc install fails 2015-05-18 06:49:04 yeah with that ranlib error 2015-05-18 06:49:15 but targmatch.h in binutils looks ok to me 2015-05-18 06:50:44 perhaps it's some configury thing not recognizing your host triplet 2015-05-18 06:51:20 mipsel-alpine-linux-musl should not be passed to ranlib 2015-05-18 06:51:33 it should've been mapped to some mips specific string by configure 2015-05-18 06:51:46 sounds like gcc configury needs patching to recognized mipsel-*-musl or similar 2015-05-18 06:52:00 did you apply additional mips+musl patches for gcc? 2015-05-18 06:52:05 no 2015-05-18 06:52:23 hmm 2015-05-18 06:52:32 i think musl support patch is there 2015-05-18 06:52:49 maybe it's incomplete 2015-05-18 06:53:17 or using mipsel-alpine-linux-musl needs to be added somewhere 2015-05-18 06:53:34 iiirc, gcc has tendency to use matches like mips*-*linux*gnu 2015-05-18 06:56:03 okay, i will look into it today 2015-05-18 06:56:06 thanks 2015-05-18 06:57:37 seems there is: 2015-05-18 06:57:43 mips*-*-linux* #Linux MIPS, either endian 2015-05-18 06:58:03 but also 2015-05-18 06:58:05 mips-*-elf* | mipsel-*-elf* | mipsr5900-*-elf* | mipsr5900el-*-elf* 2015-05-18 06:59:07 iirc, you also need to add --secure-plt or similar on mips for musl to work 2015-05-18 06:59:34 oh, no, that was for powerpc 2015-05-18 07:00:49 debian seems to use mipsel-linux-gnu 2015-05-18 07:02:20 most of that is in gcc/config.gcc and gcc/configure.ac 2015-05-18 09:34:56 setup-alpine, chosing openntpd doesn't seem to work on reboot 2015-05-18 09:36:12 rc-status show ntpd *and* openntpd in default runlevel 2015-05-18 09:42:40 rc-service ntpd seems to try to start ntpd binary from openntpd package with busybox ntpd options 2015-05-18 09:44:04 and chronyd starts too! 2015-05-18 09:44:36 should i expect openntpd, chronyd and busybox ntpd to all start? :) 2015-05-18 09:45:05 at once? :p 2015-05-18 09:46:04 ScrumpyJack chronyd starts hier fine on deftault runlevel 2015-05-18 09:46:05 yup 2015-05-18 09:47:38 and /usr/sbin/chronyd is owned by chrony-1.30-r0 2015-05-18 09:47:54 starts here fine too, i'm just pointing out that fresh install of AL edge, i chose openntpd for ntp and chrony + bb ntpd + openntpd are all set to start on default openrc runlevel! 2015-05-18 15:39:18 Hi guys! You would think I'm kidding as I have already asked for a number of compilers, but I want to ask you to package Julia (http://julialang.org/), Rust (http://rust-lang.org/) and Nim (http://nim-lang.org/ ; please package nimble (https://github.com/nim-lang/nimble) as well). 2015-05-18 15:45:20 frol__: the build files for apk are essentially shell scripts with a few required variables and functions. if you can make it compile from source against musl, you can create a package for it pretty easily. http://wiki.alpinelinux.org/wiki/Creating_an_Alpine_package 2015-05-18 15:46:16 just don't use any bashisms as it'll use the busybox ash as the bourne compatible shell 2015-05-18 15:47:03 bourne's slightly messy, but pretty simple to work with compared to some of the languages you just mentioned :p 2015-05-18 15:48:05 ttilley: Nim certanly compiles without any problems against musl. Here is a Dockerfile where it is compiled on image build: https://github.com/coopernurse/docker-nim/blob/master/Dockerfile 2015-05-18 15:49:02 Hi. I would like to prepare a package for a small tool I use already on some debian boxes. It's from sourceforge. It seems there is no tarball for source code but only a svn available. Previously it was not big deal as I did not build a packacge but just did the classic ./configure;make:make install invocation. What I do not know is how to define the source in APKBUILD for a svn checkout... any clue? 2015-05-18 15:49:46 i'm not an alpine dev. im just saying the build format is pretty simple, and it's likely to be accepted if you do the work upfront of creating the script. at least thrown into the testing repo. 2015-05-18 15:52:46 Jean-Scotch: it'd probably be easier to use git-svn to check out the project, then upload it to a git host that has source-checkout URLs in their public interface (using gitweb or the like, or hosting it on github which also has "check out this commit as a tarball" feature) 2015-05-18 15:52:56 ttilley: I'm somewhat familiar with the process of APK packaging, but when I do them myself it takes me a couple of days to make them properly... 2015-05-18 15:53:42 ACTION doesn't think the interface supports anything other than fetching a single file over http/https/ftp 2015-05-18 17:06:15 Jean-Scotch - just install subversion as a make depends & checkout the repo in prepare() - this worked for me with a git repo 2015-05-18 17:07:02 BitL0G1c: ok. was thinking that way. will do that 2015-05-18 17:20:22 it works. now, to fix install path and it will be good ^^ 2015-05-18 17:31:31 ttilley: it can fetch as many urls as you need 2015-05-18 17:32:04 fabled: i checked and mips*-*-linux* matches fine 2015-05-18 17:54:33 nginx-naxsi has missing SSL symbols http://hastebin.com/yuqagocoro.vbs - rebuild needed ? 2015-05-18 18:04:26 going to see if 1.90 works 2015-05-18 18:18:35 anyone has dualhead xorg config? 2015-05-18 19:21:20 ACTION hides 2015-05-18 19:30:23 actually this approach works better 2015-05-18 19:30:32 the other approach worked only for dlopen() files 2015-05-18 20:53:32 uggedal/alpine docker image needs to be rebuilt - has the same libssl.so symbol error 2015-05-18 21:14:03 use the 'alpine' image 2015-05-18 21:14:25 courtesy of gliderlabs 2015-05-18 21:15:21 trying that 2015-05-18 21:15:28 run is not mounted 2015-05-18 21:16:03 in LXC: run on /run type tmpfs (rw,nosuid,nodev,noexec,relatime,size=1024k,mode=755) 2015-05-18 21:16:16 andyshinn: ↑ 2015-05-18 21:17:11 trying to run sshd & nginx via supervisord fails - I think due to no /run 2015-05-18 21:17:12 or report it here https://github.com/gliderlabs/docker-alpine/issues 2015-05-18 21:17:22 right 2015-05-18 21:17:24 not our stuff, (un)fortunately 2015-05-18 21:18:23 What requires /run? 2015-05-18 21:18:58 my dockerbuild runs on debian - just not on alpine 2015-05-18 21:19:12 trying supervisord running sshd & nginx 2015-05-18 21:19:29 the daemons die every second 2015-05-18 21:20:10 on LXC /var/run is a link to /run which is tmpfs 2015-05-18 21:21:07 andyshinn: apk requires run 2015-05-18 21:21:58 yes 2015-05-18 21:22:34 this dockerbuild runs on debian jessie image 2015-05-18 21:23:07 do you get funny messages from openrc moaning about a chroot ? 2015-05-18 21:23:10 Can you give me a way to reproduce? I'm still not quite sure I understand.. 2015-05-18 21:23:34 http://hastebin.com/aqaxewuwaq.vbs 2015-05-18 21:23:43 let me get the dockerbuild 2015-05-18 21:24:25 Your running the Docker image under LXC? 2015-05-18 21:25:32 http://hastebin.com/otabuciqaq.vala 2015-05-18 21:25:39 no not in LXC 2015-05-18 21:25:54 in Debian jessie normal 2015-05-18 21:26:45 ok, so changing base to alpine breaks supervisord? 2015-05-18 21:27:11 i moved to my desk with a keyboard so i can test now :) 2015-05-18 21:27:24 (my primary computer is Android tablet) 2015-05-18 21:28:45 http://hastebin.com/avuruheqel.vbs supervisord.ini 2015-05-18 21:30:03 btw, sshd in container? 2015-05-18 21:30:18 use docker exec instead 2015-05-18 21:32:08 yea, i can give you a couple pointers 2015-05-18 21:32:15 but the root problem is missing nginx: [emerg] mkdir() "/tmp/nginx/client-body" failed (2: No such file or directory) 2015-05-18 21:32:37 i think this is a bug in the nginx package, i actually have to do this on all my alpine nginx images 2015-05-18 21:32:54 ah right 2015-05-18 21:33:04 but, i'd recommend just making your CMD nginx and getting rid of supervisord 2015-05-18 21:33:20 instead of ssh, use docker exec to do one off commands in the container as barthalion mentioned 2015-05-18 21:33:21 ok 2015-05-18 21:33:41 the only addition to nginx is you need "daemon off;" in the config, then it will run in foreground, and docker will be happy 2015-05-18 21:34:08 another tip, send access_log and error_log to /dev/stdout and /dev/stderr respectively, that way docker logging facility can catch nginx logging 2015-05-18 21:34:42 ok - very helpful :) 2015-05-18 21:34:48 then (shameless plug here) you can use gliderlabs/logspout to ship the docker logs somewhere else (like syslog) 2015-05-18 21:35:20 nice 2015-05-18 21:36:09 many thx for the pointers 2015-05-18 21:36:45 no problem! my first primary use case for alpine as a container was nginx, so i've definitely done it a bit 2015-05-18 21:36:59 a jessie based nginx is 249meg - on your image it's 53meg 2015-05-18 21:37:14 haha, get rid of supervisord and guess what? it will be 6mb :p 2015-05-18 21:37:29 going to use alpine image as a data only container too 2015-05-18 21:37:34 i've shipped entire static websites with it that were only 10mb 2015-05-18 21:37:58 and my deployment model becomes, literally, docker save | ssh "docker load" ;) 2015-05-18 21:39:03 for static websites that maybe sometimes need one or two extra packages, it can't be beat! unless you want to roll your own busybox image with buildroot - but my goodness... who wants to wait 45 minutes for buildroot to build?!?! 2015-05-18 21:39:20 totally 2015-05-18 21:39:32 does 'RUN rc-service xxxxx' in the Dockerbuild work ? 2015-05-18 21:39:43 nope, it won't 2015-05-18 21:39:53 ok 2015-05-18 21:40:01 typically, most init services do not work in Docker, because the init system is not pid 1 2015-05-18 21:40:25 your process is pid 1, so that shell command to run rc-service was pid1, then it exited, then the container exited... 2015-05-18 21:40:44 any way besides supervisord to run more than one service ? 2015-05-18 21:40:47 that is why you usually see supervisord or runit everywhere... 2015-05-18 21:41:21 yes, the one i typically deploy is actually a simple shell script, but one of our guys rewrote it into a static Go binary.. let me find it 2015-05-18 21:42:24 https://github.com/MattAitchison/codep this runs multiple processes, if one of the processes dies, it kills everything and exits 2015-05-18 21:42:51 you use it like: codep "nginx -c /etc/nginx/nginx/conf" "some_other_daemon" 2015-05-18 21:43:13 thx very much - will give it a go 2015-05-18 21:43:28 the simple shell example is https://gist.github.com/progrium/0ac0248f70e2adce964f 2015-05-18 21:43:50 http://hastebin.com/jupexasuha.parser3 - a script i use to start a data only container + a running container 2015-05-18 21:44:04 with optional fixed ip address 2015-05-18 21:44:25 but honestly, whatever you feel comfortable with is fine, just be aware of the implications of multiple processes in containers and what it can mean to have one process die and the other continue to run... sometime this means the entire container should exit, sometimes not 2015-05-18 21:45:04 wow interesting, i've never done static IPs for containers yet 2015-05-18 21:45:12 i've been using it to start nginx+ssh / 2nd build container 2015-05-18 21:45:27 the nginx stays up 2015-05-18 21:45:58 if you restart the script it will a 2nd container 'xxxxx2' 2015-05-18 21:46:06 then 'xxxxx3' etc... 2015-05-18 21:46:40 the nginx is for a local APT repo 2015-05-18 21:47:10 ah, neat 2015-05-18 21:50:07 handy for tidying up too - 'dki -r all' removes all images & containers 2015-05-18 21:58:19 BitL0G1c: barthalion: deleted my docker images, they were way out-of-date and the build script could not build edge 2015-05-18 21:58:43 ah ok 2015-05-19 07:26:32 fabled: i have a challenge in apk trigger script 2015-05-19 07:26:51 i want syslinux.trigger to not run update-extlinux in lxc, vserver etc 2015-05-19 07:26:57 so i tried to run openrc --sys 2015-05-19 07:27:00 to check the system 2015-05-19 07:27:02 however 2015-05-19 07:27:07 it fails to detect LXC 2015-05-19 07:27:21 and i'm pretty sure it will fail detect anything 2015-05-19 07:27:29 else if (file_regex("/proc/1/environ", "container=lxc")) 2015-05-19 07:27:29 return RC_SYS_LXC; 2015-05-19 07:27:39 the problem is that it runs in chroot? 2015-05-19 07:27:45 and /proc is not mounted? 2015-05-19 07:29:38 no, trigger is not in chroot; unless using apk --root /sysroot or similar 2015-05-19 07:30:52 Executing syslinux-6.03-r1.trigger 2015-05-19 07:30:52 + whoami 2015-05-19 07:30:52 root 2015-05-19 07:30:52 + openrc --sys 2015-05-19 07:30:52 + ls -l /proc/1 2015-05-19 07:30:52 ls: /proc/1: No such file or directory 2015-05-19 07:30:54 + openrc --sys 2015-05-19 07:30:56 + exec update-extlinux 2015-05-19 07:31:04 it cannot read /proc/1 2015-05-19 07:31:18 hmm 2015-05-19 07:31:28 maybe it does chroot then 2015-05-19 07:31:53 but'll just then chdir("/") and chroot(".") 2015-05-19 07:31:59 yes 2015-05-19 07:32:02 hum 2015-05-19 07:32:02 that can be enough to make it fail under grsec though 2015-05-19 07:32:13 probably yes 2015-05-19 07:32:42 yes. chroot is unconditional 2015-05-19 07:32:59 it'll basically chroot("/") 2015-05-19 07:33:07 yes 2015-05-19 07:33:14 i think thats ok 2015-05-19 07:33:15 but that might make grsec deny access to parts of /proc 2015-05-19 07:33:55 interesting 2015-05-19 07:33:56 yes 2015-05-19 07:33:59 it does 2015-05-19 07:34:07 ncdev-edge-x86_64:~# chroot / /bin/sh 2015-05-19 07:34:07 ncdev-edge-x86_64:/# ls /proc/1 2015-05-19 07:34:07 ls: /proc/1: No such file or directory 2015-05-19 07:34:37 http://sprunge.us/eKWP 2015-05-19 07:38:00 i suppose chroot("/") is a good thing 2015-05-19 07:38:20 as then it will behave the same as if apk --root ... 2015-05-19 07:39:50 you can set rc_sys=LXC hard in /etc/rc.conf 2015-05-19 07:42:06 i dont know if we want to depend on that rc_sys=LXC on all build boxes 2015-05-19 07:42:28 clandmeter: i think once InstallFlash is fixed with the suggestions i made in the thread (now that the necessary symlinks are installable via libc6-compat package) we can put it in main 2015-05-19 07:42:42 clandmeter: but we cannot legally ship flash in any archive, the license terms for flash forbid it 2015-05-19 07:43:04 kaniini: maybe we can have an apkbuild in non-free 2015-05-19 07:44:17 ncopa: well that is what InstallFlash does 2015-05-19 07:44:21 it fetches and installs flash 2015-05-19 07:44:31 its like b43-fwcutter or whatever 2015-05-19 07:44:52 but with apkbuild it can be properly managed by apk 2015-05-19 07:44:54 so i was thinking we just have an apk which runs it 2015-05-19 07:44:59 yup 2015-05-19 07:45:18 i mean 2015-05-19 07:45:18 basically an apk which contains installflash, and then post-install/deinstall scripts 2015-05-19 07:45:36 which invoke installflash and then the deinstall just rm's it 2015-05-19 07:45:46 no, i mean an apkbuild that installs it in "$pkgdir" 2015-05-19 07:45:56 so you can apk info --who-owns /path/to/flash.so 2015-05-19 07:46:01 and get the proper package name 2015-05-19 07:46:02 oh 2015-05-19 07:46:10 yes, we could do that. 2015-05-19 07:46:20 i think i do that with fwcutter too 2015-05-19 07:46:21 we just cant ship the binary itself ;) 2015-05-19 07:46:31 thats why non-free exists 2015-05-19 07:46:46 archlinux seems to ship it anyway 2015-05-19 07:46:54 but i mean 2015-05-19 07:46:58 that seems like a bad idea 2015-05-19 07:47:12 because technically that means you can have the archlinux repo taken down with a DMCA :| 2015-05-19 07:47:20 (or equivilant) 2015-05-19 07:47:35 heh 2015-05-19 07:47:37 except we have an agreement with Adobe iirc 2015-05-19 07:48:11 so it's not going to happen, sorry 2015-05-19 07:48:13 too bad you have agreement with Adobe. Otherwise I had a great idea how to kill competition ;) 2015-05-19 07:48:34 (thats what this DMCA is about, isnt, it? :) 2015-05-19 07:48:38 i dont think we have to worry about it 2015-05-19 07:48:41 all we have to say is 2015-05-19 07:48:45 "we don't use systemd" 2015-05-19 07:48:50 seems to be working quite well 2015-05-19 07:49:05 :) 2015-05-19 07:49:18 yeah, i was just kidding 2015-05-19 07:49:37 at this rate we will be the #1 linux distribution due to lack of systemd usage 2015-05-19 07:49:43 i prefer focus on make a good product over "how do we kill competition" 2015-05-19 07:50:26 it is unfortunate that Adobe makes deals with people to ship binary blobs 2015-05-19 07:51:00 it is unfortunate that Adobe ships binary blobs 2015-05-19 07:51:04 but what can you do... 2015-05-19 07:51:34 sure 2015-05-19 07:51:39 what is really unfortunate, now only Google can ship "modern" flash 2015-05-19 07:51:44 with Chrome, of course 2015-05-19 07:52:06 non-free could work as a solution 2015-05-19 07:52:24 just depends=libc6-compat and then fetch the plugin 2015-05-19 07:53:01 this flash business you are working on 2015-05-19 07:53:10 will that make netflix work with firefox 38? 2015-05-19 07:53:15 no 2015-05-19 07:53:19 that's silverlight :( 2015-05-19 07:53:24 no 2015-05-19 07:53:39 firefox 38 has netflix without silverlight 2015-05-19 07:53:47 with some adobe DRM thingy 2015-05-19 07:53:53 oh. that is only on Windows Vista or newer! 2015-05-19 07:53:56 yeah 2015-05-19 07:54:11 :| 2015-05-19 07:54:37 Adobe Content Decryption Module (CDM) 2015-05-19 07:54:57 i'm sure it is just a matter of time before someone reverse engineers it and makes a replacement 2015-05-19 07:55:21 chrome has 2 .so files too 2015-05-19 07:55:25 where is DVD_Jon when you need him 2015-05-19 07:55:27 ;) 2015-05-19 07:55:29 it's going to be ported to other OSes as well 2015-05-19 07:55:29 :) 2015-05-19 07:55:37 adobe CDM? 2015-05-19 07:55:41 they just started from Windows, you can guess why 2015-05-19 07:56:26 i know you can run google chromimum (the opensource variant) if you just copy 2 .so files from the closed source package 2015-05-19 07:56:40 that's true as well 2015-05-19 07:56:53 not that I care, netflix is unavailable here anyway 2015-05-19 07:57:24 so if we had chromium for alpine we could maybe use those + libc6-compat 2015-05-19 07:57:28 oh. 2015-05-19 07:57:30 well. 2015-05-19 07:57:38 i doubt our libc6-compat is good enough to run chromium :P 2015-05-19 07:57:53 it only need to be good enough for those 2 .so files 2015-05-19 07:58:05 or? 2015-05-19 07:58:09 hmm, true 2015-05-19 07:58:17 i doubt the CDM uses that advanced of functionality 2015-05-19 07:58:20 and someone has to be good enough to fix shit in chromium code 2015-05-19 07:58:20 anyway. 2015-05-19 07:58:33 barthalion: been there, tried that 2015-05-19 07:58:39 barthalion: yeah, thats probably the bigger problem :) 2015-05-19 07:58:41 wound up screaming obscenities for hours 2015-05-19 07:59:07 probably easier to reverse engineer the adobe CDM :) 2015-05-19 07:59:20 unfortunately adobe has lawyers that scare me 2015-05-19 07:59:48 where is DVD_jon? 2015-05-19 07:59:52 :) 2015-05-19 07:59:54 however, i might do it anyway. after all the DMCA allows reverse engineering for the purpose of "interoperability" :) 2015-05-19 08:00:07 i dunno, i figured he got sued into oblivion over DeCSS 2015-05-19 08:03:53 oh wow, not sued, but criminal charges 2015-05-19 08:03:55 i forgot about that 2015-05-19 08:17:47 kaniini: im all ok with any solution, free or non-free. except nobody has non-free added to their repo file. 2015-05-19 08:20:30 I don't think we even build packages from there 2015-05-19 08:21:12 is there a safe way to have the .so on dev.a.o and post-install it? 2015-05-19 08:21:45 and just add a function to apkbuild that uploads a new version when bumped 2015-05-19 08:26:01 clandmeter, the non-free is only to make it easy for people to build their own repo 2015-05-19 08:26:20 build their own repo with packages that we are not allowed to ship 2015-05-19 08:27:32 ncopa: ok i understand. so then its better just to do it post-install? 2015-05-19 08:27:59 or like InstallFlash does 2015-05-19 08:28:04 no, the idea is that we build a proper apk of it in non-free 2015-05-19 08:28:13 be we cannot build that package for end users 2015-05-19 08:28:28 so end users can git clone aports 2015-05-19 08:28:31 cd non-free 2015-05-19 08:28:33 abuild -r 2015-05-19 08:28:39 and they have their own build .apk 2015-05-19 08:28:44 is that user friendly? 2015-05-19 08:28:46 with the non-free stuff 2015-05-19 08:28:54 probably not 2015-05-19 08:29:04 but that is more technically correct 2015-05-19 08:29:13 im fine with having it. but it doesnt solve the problem. 2015-05-19 08:29:47 i suppose it is possible to ship a install-flash.apk with main repo 2015-05-19 08:29:55 which pulls it from post-install 2015-05-19 08:30:25 or even make a deal with adobe 2015-05-19 08:31:11 we make deals with the darkside? 2015-05-19 08:31:32 but adding a install-flash package to non-free does not solve it either, and the way it does not solve it is also techically incorrect :) 2015-05-19 08:32:18 depends on what you mean with "it aka the problem" 2015-05-19 08:33:00 what do you mean with "the problem"? 2015-05-19 08:33:10 what proble are you trying to solve 2015-05-19 08:33:19 user wants to use flash 2015-05-19 08:33:27 he wants to install it 2015-05-19 08:33:36 and he has no knowledge how to build an apk 2015-05-19 08:34:09 then we have 2 options 2015-05-19 08:34:12 he has 2 solutions, 1. go to arch 2. learn aports 2015-05-19 08:34:23 :) 2015-05-19 08:34:49 or we provide him option 3 2015-05-19 08:34:55 we can help him use an InstallFlash program (shell script or go-lang program) 2015-05-19 08:35:12 or we can ship a package for him that does it with post-install 2015-05-19 08:35:22 in testing or main repo 2015-05-19 08:35:52 but it makes no sense to install it from post-install in a package that is in non-free 2015-05-19 08:36:08 ofcourse 2015-05-19 08:36:14 that was not what i was saying 2015-05-19 08:36:19 ok 2015-05-19 08:36:22 it should go in main 2015-05-19 08:36:25 then we are on same page :) 2015-05-19 08:36:46 i was suggesting a solution for people who do know how to build an .apk 2015-05-19 08:36:56 or more 2015-05-19 08:37:20 i was suggesting a more technical correct solution, but it requires that end user builds his own package 2015-05-19 08:37:47 download src from apkbuild, and upload the so to dev.a.o, and create a post-install to download it with apk. 2015-05-19 08:38:41 my only concern is pulling .so from a "non verified" source 2015-05-19 08:38:52 cannot the post-install just download from adobe directly? 2015-05-19 08:39:04 it could 2015-05-19 08:39:20 but thats even more untrusted for me. 2015-05-19 08:39:22 we are not allowed to redistribute from dev.a.o 2015-05-19 08:39:43 yup 2015-05-19 08:39:48 untrusted 2015-05-19 08:39:56 you have to make choices in life :) 2015-05-19 08:39:57 but that was what the end-user wanted? 2015-05-19 08:40:02 yes :) 2015-05-19 08:40:45 my reasoning is 2015-05-19 08:40:56 well we should then add a msg when install the pacakge and add a script for a next step 2015-05-19 08:41:02 closed source stuff like flash is bad 2015-05-19 08:41:15 we should not encourage it 2015-05-19 08:41:42 but i am ok if users want use it anyway 2015-05-19 08:41:47 i might even do it myself 2015-05-19 08:42:13 so i dont object to a non-free package en user can build himself 2015-05-19 08:43:19 maybe we should just have a pacakge in main that makes the host compatible to install the so 2015-05-19 08:43:31 and msg an url to our wiki how its done. 2015-05-19 08:43:33 but shipping a package with main, that is super easy to install but difficult to trust... 2015-05-19 08:44:02 i have my doubts about making it too easy to install untrusted stuff 2015-05-19 08:44:11 agree 2015-05-19 08:44:26 post-install sounds scary 2015-05-19 08:44:38 we are not a scary dist 2015-05-19 08:44:39 and is technically ugly 2015-05-19 08:44:49 aw i forgot my coffe 2015-05-19 08:44:51 brb 2015-05-19 08:44:56 go to arch if you wnat something scary 2015-05-19 08:45:00 ;-) 2015-05-19 08:46:43 yeah 2015-05-19 08:47:09 i dont mind that the OMG! ubuntu crowd goes somewhere else :) 2015-05-19 08:47:31 what about just a flash-compat package 2015-05-19 08:47:41 and a wiki entry 2015-05-19 08:47:50 isnt libc6-compat enough? 2015-05-19 08:48:05 i dont think so 2015-05-19 08:48:10 we needto symlink it 2015-05-19 08:48:14 i think 2015-05-19 08:48:31 isnt that what kaniini did withe the libc6-compat? 2015-05-19 08:48:51 let me check, i htink he still mentioned it needed a sym 2015-05-19 08:49:38 $ ls -l /lib/ld-linux-x86-64.so.2 2015-05-19 08:49:38 lrwxrwxrwx 1 root root 26 May 19 10:48 /lib/ld-linux-x86-64.so.2 -> /lib/libc.musl-x86_64.so.1 2015-05-19 08:50:00 oh ok 2015-05-19 08:50:02 then fine 2015-05-19 08:50:26 lets just add a wiki page how to make flash working. 2015-05-19 08:50:35 anybody can download and extract a file 2015-05-19 08:50:44 if thats not the case, go to windows 2015-05-19 08:52:01 :) 2015-05-19 09:20:10 our wiki is driving me nuts 2015-05-19 09:20:27 its keeps resetting to previous changes when i edited a page 2015-05-19 09:20:38 anyways http://wiki.alpinelinux.org/wiki/Installing_Adobe_flash_player_for_Firefox 2015-05-19 09:22:44 mkdir -p /usr/lib/mozilla/plugins && curl $url | sudo tar -zx -C /usr/lib/mozilla/plugins libflashplayer.so 2015-05-19 09:23:05 :) 2015-05-19 09:23:09 i was thinking the same thing 2015-05-19 09:23:19 but then i though, lets not make it more complicated :) 2015-05-19 09:23:22 yeah 2015-05-19 09:25:48 ncopa: could it be the caching reverse proxy of wiki that caches the wiki form and provides the previous edit. its kind of frustrating to keep loosing last changes. (or maybe its just me....) 2015-05-19 09:30:11 nangel_: ^^^ 2015-05-19 09:31:33 clandmeter: it's not just you... As an avid wiki user, I now have the habit to always refresh before starting to type anything... 2015-05-19 09:32:48 nangel_: maybe add an exception when =edit is part of the url? 2015-05-19 09:42:20 ncopa: you're a heavey tmux user right? tmux isn't diplaying my characters properly. i can see them fine in normal shell. have you come across this? 2015-05-19 09:44:48 ScrumpyJack: i dont have that problem 2015-05-19 09:44:57 what is your LANG set too? 2015-05-19 09:45:06 $ echo $LANG 2015-05-19 09:45:06 en.utf8 2015-05-19 09:45:16 $ echo $TERM 2015-05-19 09:45:16 screen 2015-05-19 09:45:38 also, what shell is it 2015-05-19 09:45:47 busybox ash or bash/zsh? 2015-05-19 09:46:15 zsh, LANG=en.utf8, CHARSET=UTF-8 2015-05-19 09:46:20 what's yours? 2015-05-19 09:46:33 what's your TERM? 2015-05-19 09:46:34 term is screen-256color 2015-05-19 09:46:37 $ echo $CHARSET 2015-05-19 09:46:37 UTF-8 2015-05-19 09:46:48 my shell is busybox ash 2015-05-19 09:46:49 ScrumpyJack: I have the same (with mksh), works fine here on -edge 2015-05-19 09:47:07 maybe its zsh that is lacking utf8? 2015-05-19 09:47:08 what do you mean iti s not displaying the chars properly? 2015-05-19 09:47:15 does it happen with plain ascii? 2015-05-19 09:47:40 "plain" ascii is fine until I get to characters with accents 2015-05-19 09:47:44 right 2015-05-19 09:47:51 check your zsh manual 2015-05-19 09:47:57 ok 2015-05-19 09:47:58 I would be surprised if it did not support utf8 2015-05-19 09:48:40 i can see them fine when not using tmux 2015-05-19 09:49:56 there is a utf8 command for tmux 2015-05-19 09:50:02 I have never needed to use it though 2015-05-19 09:50:38 setw -g utf8 on or so 2015-05-19 09:51:12 or set -g utf8 on 2015-05-19 09:51:15 something like that 2015-05-19 09:52:46 i have this in my putty LANG=en_CA.utf8 exec tmux att -d 2015-05-19 09:53:57 some screenshots showing the problem 2015-05-19 09:54:01 http://st.ilet.to/inside_tmux.png 2015-05-19 09:54:08 http://st.ilet.to/outside_tmux.png 2015-05-19 09:54:34 can you force utf8 for that window and see if it works? 2015-05-19 09:54:48 ok, i'll try that 2015-05-19 09:55:16 \o/ 2015-05-19 09:55:18 http://www.adobe.com/no/software/flash/about/ 2015-05-19 09:55:26 tells me: Adobe flash player sucessfully installed 2015-05-19 09:56:07 now i need websites that requires flash 2015-05-19 09:58:45 ncopa: firefox? 2015-05-19 09:58:56 yes 2015-05-19 10:00:41 ScrumpyJack: try with tmux -u 2015-05-19 10:01:14 ScrumpyJack: you will have to start with a clean server though 2015-05-19 10:02:39 kk 2015-05-19 10:04:53 -u worked! i see it guesses. I'll need new tmux server :( 2015-05-19 10:04:55 bye! 2015-05-19 10:05:13 /quit 2015-05-19 10:06:30 that worked! thanks 2015-05-19 10:08:13 ncopa: is it so hard to find a flash website these days? :) 2015-05-19 10:08:20 ScrumpyJack: you can tweak your environment so that these checks matches: https://github.com/ThomasAdam/tmux/blob/833fe5bdee449b04afe8844774be66374e1baf9c/tmux.c#L282 2015-05-19 10:08:32 so that you don't have to force utf8 with -u 2015-05-19 10:09:00 i'll definitly look at that, thanks 2015-05-19 10:09:36 (now if i could only get urlview to build woth musl :) 2015-05-19 10:09:53 i think i also have tmux started with -u, just forgot about it. maybe put it on the wiki. 2015-05-19 10:10:07 if either LC_ALL, LC_TYPE or LANG contains UTF-8/UTF8 ignoring case you should not need -u 2015-05-19 10:10:43 i start it about ones a year, so i dont remember :) 2015-05-19 10:11:09 well, urlview builds, but just seg faults when it tries to call elinks 2015-05-19 10:11:11 http://sprunge.us/fUSO 2015-05-19 10:12:35 i have it mapped to ctrl-a u in tmux, and urlview pops up an list the URLs in the tmux window (irc in this case). when i select one, seg fault :( 2015-05-19 10:15:01 ScrumpyJack: what about the one from github? 2015-05-19 10:16:19 checking 2015-05-19 10:16:42 which one? :) https://github.com/search?utf8=%E2%9C%93&q=urlview 2015-05-19 10:17:25 https://github.com/sigpipe/urlview/archive/08767aa863cd27d1755ba0aff65b8cc1a0c1446a.zip 2015-05-19 10:17:46 no releases :( 2015-05-19 10:20:40 ScrumpyJack: if it works i can add it. 2015-05-19 10:20:53 not sure its based on the same source though 2015-05-19 10:20:56 "add it" to what? 2015-05-19 10:21:19 add it to testing 2015-05-19 10:21:28 its based on 0.9 2015-05-19 10:21:55 ah cool. if it worked, i would have sent a patch :) 2015-05-19 10:22:08 i got the taiwan url from gentoo 2015-05-19 10:22:13 https://packages.gentoo.org/package/net-misc/urlview 2015-05-19 10:22:57 s/taiwan/austria 2015-05-19 10:39:22 arch linux get it from debian https://aur.archlinux.org/packages/ur/urlview/PKGBUILD 2015-05-19 10:44:58 ScrumpyJack: it works for me, but it adds %0A to the end of the uri 2015-05-19 10:54:19 clandmeter: can you sprunge me your APKBUILD file? 2015-05-19 10:57:43 ScrumpyJack: try it 2015-05-19 10:57:59 not sure i added bash though 2015-05-19 10:58:03 the script wants it. 2015-05-19 11:01:44 i'll try it now 2015-05-19 11:13:39 clandmeter: \o/ 2015-05-19 11:13:42 works for me 2015-05-19 11:23:55 ScrumpyJack: nice, if you want to you can take over maintenance and move it to main. 2015-05-19 11:24:41 if the .sh file is really usefull, bash should be added. 2015-05-19 12:02:01 clandmeter: we'll leave it in your name, it's your work :) 2015-05-19 12:02:18 ScrumpyJack: i dont use it :) 2015-05-19 12:10:38 ncopa: installflash depends on xz 2015-05-19 12:15:00 open /usr/lib/mozilla/plugins/libflashplayer.so: no such file or directory 2015-05-19 12:17:50 ScrumpyJack: installflash is broken 2015-05-19 12:18:28 :) oops 2015-05-19 12:19:11 ScrumpyJack: http://wiki.alpinelinux.org/wiki/Installing_Adobe_flash_player_for_Firefox 2015-05-19 12:19:25 or build it yourself from non-free 2015-05-19 12:21:06 what was ncopa excited about earlier? 2015-05-19 12:21:46 he got flashplugin working in ff i guess 2015-05-19 12:21:54 he didnt confirm it worked though 2015-05-19 12:22:10 ah, i thought i was a packaged being pushed shortly after that 2015-05-19 12:22:41 but i have no logs (i disconnected for tmux -u and http://dev.alpinelinux.org/irclogs/ not yet updated) 2015-05-19 12:23:41 creating the package was not such an achivement, having musl abi compatible enough that it works is :) 2015-05-19 12:24:47 ScrumpyJack: i did create a package in non-free 2015-05-19 12:24:58 so you'd have to build it locally 2015-05-19 12:25:04 and it does work 2015-05-19 12:25:11 ok cool, awesome 2015-05-19 12:25:22 i'll build 2015-05-19 12:27:29 no libc6-compat deps? 2015-05-19 12:29:20 libc6-compat has been added to edge today 2015-05-19 12:33:47 yay! "You have version 11,2,202,460 installed" 2015-05-19 12:34:47 how does chrome get an updated flash player? does google maintain it? 2015-05-19 12:35:20 something like that 2015-05-19 12:35:32 Adobe delegated Flash development for Linux to Google 2015-05-19 12:35:46 Google told everyone to fuck off and you need Chrome for it 2015-05-19 12:35:56 or Chromium + extracted flash from chrome tarball 2015-05-19 12:36:51 although there is freshplayerplugin, so it's possible to use it in firefox 2015-05-19 12:36:57 (and it works quite well) 2015-05-19 12:38:58 i see. thanks for the info. 2015-05-19 12:40:26 (because plot is thickened and Flash used in Chrome doesn't use NPAPI) 2015-05-19 12:42:56 got it 2015-05-19 12:43:10 and found info here http://tinyurl.com/olxewkv 2015-05-19 13:00:24 hello 2015-05-19 13:01:07 i try this channel because there is nobody on the channel apline-linux 2015-05-19 13:02:04 i have to ask something about repositories can i ask here ? 2015-05-19 13:02:15 don't ask to ask 2015-05-19 13:02:29 if there is someone who knows the answer, he surely will reply 2015-05-19 13:02:40 ok thanks 2015-05-19 13:04:08 i use for test Alpine Linux from yesterday for a private project on raspeberry pi and today i have a pb with repositories. I try to install packages i got for answer IO ERROR 2015-05-19 13:04:22 could u give me a way to resolve this 2015-05-19 13:05:02 I have seen this problem two times: first was when the drive was in RO mode and the other when I had no free space on my device 2015-05-19 13:05:40 i think it is a pb with space on m vm i m going to inspect this in first 2015-05-19 13:16:46 you need to remount shm with more space 2015-05-19 13:18:44 mount -o remount /dev/shm size=512m or something 2015-05-19 13:19:32 on rpi1? 2015-05-19 13:20:31 yeah 2015-05-19 13:22:56 interesting 2015-05-19 13:22:57 mount -o remount,size=512m /dev/shm 2015-05-19 13:44:11 not on rpi1 but rpi2 for my needs i think rpi1 is not powerful enough 2015-05-19 13:52:40 jck_ - run 'setup-apk-repos' & try a different repo - also first run 'setup-interfaces' 2015-05-19 13:55:19 you probably have not started networking - it is disabled for new installations by default 2015-05-19 23:28:10 so, i think regarding EME in firefox, the solution is to implement the EME algorithms and then make it so the user provides a key (presumably that he would extract from the proprietary Adobe plugin) 2015-05-20 10:26:57 i need to build libiptcdata, do i call it iptcdata-lib or libiptcdata for alpine linux's repo? 2015-05-20 10:27:29 libiptcdata 2015-05-20 10:47:24 seems nice, http://www.anytun.org/ , we have anything similar ? 2015-05-20 11:18:29 rnalrd: can you add neon-dev to the makedepends of owncloud? 2015-05-20 13:11:29 hum 2015-05-20 13:11:43 why is glm failing on build servers? 2015-05-20 13:11:47 i cannot reproduce it here 2015-05-20 13:13:09 ncopa, fwiw here glm build correectly. 2015-05-20 13:13:13 but i got a warning: 2015-05-20 13:13:13 >>> WARNING: glm*: No arch specific binaries found so arch should probably be set to "noarch" 2015-05-20 13:14:14 glm only provides headers apparently 2015-05-20 13:14:48 http://sprunge.us/cahP 2015-05-20 13:15:00 Can i push? 2015-05-20 13:15:05 i dont know if that is correct 2015-05-20 13:15:18 but i suppose it does not hurt 2015-05-20 13:15:27 the -dev package might not be noarch 2015-05-20 13:15:43 "GLM is a header only library, there is nothing to build, just include it." 2015-05-20 13:15:53 this is what readme.txt state 2015-05-20 13:15:56 *states. 2015-05-20 13:16:08 there are tests 2015-05-20 13:16:17 [ 16%] /home/buildozer/aports/main/glm/src/glm/glm/detail/dummy.cpp:1:0: sorry, unimplemented: 64-bit mode not compiled in 2015-05-20 13:16:17 /////////////////////////////////////////////////////////////////////////////////// 2015-05-20 13:16:17 ^ 2015-05-20 13:16:31 aaaaah 2015-05-20 13:20:53 ok 2015-05-20 13:20:57 now it fails here too 2015-05-20 13:35:41 have you figured how to reproduce? 2015-05-20 13:39:19 updating 3.1 to edge i get this: http://sprunge.us/UKRb 2015-05-20 13:39:30 can i safely ignore and reboot? 2015-05-20 13:40:39 looks fishy 2015-05-20 13:40:44 is it pre-3.0 install? 2015-05-20 13:40:51 yes 2015-05-20 13:41:01 its definitively pre-3.0 install 2015-05-20 13:41:23 ScrumpyJack: look for *.apk-new in /etc/mkinitfs 2015-05-20 13:42:03 i just upgrade from v3.1.4-19-g00ade24 2015-05-20 13:42:11 but it may have been older 2015-05-20 13:42:23 the files containing *uClibc* should be edited 2015-05-20 13:42:25 uptime up 127 days 2015-05-20 13:42:40 so it's probably several updates without a reboot :) 2015-05-20 13:43:00 normally it should work remove the /lib/lib* files 2015-05-20 13:43:11 its leftovers from v2.x 2015-05-20 13:44:23 nothing containing uClibc in /etc/mkinitfs/mkinitfs.conf.apk-new 2015-05-20 13:44:33 as expected 2015-05-20 13:45:09 this is output of ls /lib/lib* http://sprunge.us/iYOG 2015-05-20 13:45:16 so i should delete all that? 2015-05-20 13:45:29 no 2015-05-20 13:45:31 no no no 2015-05-20 13:45:39 :) 2015-05-20 13:45:59 find /etc/mkinitfs -type f | xargs grep uClibc 2015-05-20 13:46:00 mv /etc/mkinitfs/mkinitfs.conf.apk-new /etc/mkinitfs/mkinitfs.conf 2015-05-20 13:46:17 be careful with that barthalion 2015-05-20 13:46:28 well, it's 3.1 already 2015-05-20 13:46:34 so somehow it managed to boot so far 2015-05-20 13:46:34 ncopa: http://sprunge.us/iSEE 2015-05-20 13:46:34 need compare that the needed 'features' are in there 2015-05-20 13:46:42 in case he uses lvm root or similar 2015-05-20 13:46:59 and I believe ScrumpyJack knows that he needs to "move" his customizations 2015-05-20 13:47:16 mv /etc/mkinitfs/features.d/base.files.apk-new /etc/mkinitfs/features.d/base.files 2015-05-20 13:47:23 no lvm in this alpine, it's a hvm guest (alpine-xen host \o/ ) 2015-05-20 13:47:24 should work 2015-05-20 13:47:42 you need fix your /etc/mkinitfs/features.d/base.files 2015-05-20 13:47:55 i believe using the .apk-new variant of it should just work 2015-05-20 13:48:05 unless you do funky stuff in initramfs 2015-05-20 13:48:34 alpine:~# cat /etc/mkinitfs/mkinitfs.conf.apk-new | sprunge 2015-05-20 13:48:34 http://sprunge.us/MOiM 2015-05-20 13:48:51 is that one ok to use? 2015-05-20 13:49:24 it depends on your modifications to /etc/mkinitfs/mkinitfs.conf 2015-05-20 13:49:46 features="ata base ide scsi usb virtio ext4" 2015-05-20 13:50:35 ide is missing from apk-new 2015-05-20 13:51:27 then just remove .apk-new 2015-05-20 13:51:39 you can go over all the .apk-new files and diff with current 2015-05-20 13:51:42 with update-conf 2015-05-20 13:52:16 you will want to 'zap' new in most cases i suppose 2015-05-20 13:53:21 May 29 2014 mkinitfs.conf 2015-05-20 13:53:27 Mar 10 09:25 mkinitfs.conf.apk-new 2015-05-20 13:53:45 i only have one new 2015-05-20 13:54:49 ah you mean .apk-new modules? 2015-05-20 13:58:03 what if i just rename all the apk-new ? 2015-05-20 13:58:17 /etc/mkinitfs/features.d/base.files.apk-new to /etc/mkinitfs/features.d/base.files etc 2015-05-20 13:58:46 then you'll lose your /etc/passwd and user accounts etc 2015-05-20 13:58:50 so you dont want that :) 2015-05-20 13:59:30 but the base.files.apk-new can most likely just be renamed to base.files 2015-05-20 13:59:33 ah right, i thought it was just in /etc/mkinitfs/ 2015-05-20 14:00:01 so how did all the apk-new files come about? 2015-05-20 14:00:04 no. update-conf will search all /etc/ 2015-05-20 14:00:11 when you upgrade 2015-05-20 14:00:16 and you have a configfile 2015-05-20 14:00:19 which you have modified 2015-05-20 14:00:28 and there comes a new version of package 2015-05-20 14:00:38 with new default config, it will create the .apk-new 2015-05-20 14:01:05 so you have a chance to compare your own config with what is the new default 2015-05-20 14:02:36 so which package "backed-up" this lot? http://sprunge.us/igZh 2015-05-20 14:02:39 :) 2015-05-20 14:03:02 i didnt change anything in /etc/mkinitfs/ 2015-05-20 14:33:30 reboots fine now. by the way is was a pv guest with pv-grub-x86_64 - very cool 2015-05-20 16:04:54 ncopa: llvm 3.6.1 was tagged in svn last night. just waiting on tarballs and release announcement to submit an updated package. then i'd need some code review if possible. 2015-05-20 16:29:21 ScrumpyJack: please send your patches to alpine-aports next time 2015-05-20 16:29:51 ttilley: I'll go through after madness at my company ends 2015-05-20 16:30:02 through patches* 2015-05-20 16:30:12 I hope this weekend 2015-05-20 17:51:05 barthalion: oops, yes i forgot there were seperate mailing lists 2015-05-20 21:16:44 found list of patches, http://www.linuxfromscratch.org/patches/lfs/7.7/ some pkgs version match with edge pksgs 2015-05-21 06:26:48 ttilley: i think i can just bump the version? 2015-05-21 06:26:56 for llvm 2015-05-21 06:27:05 whenever the 3.6.1 tarballs are available that is 2015-05-21 06:31:40 speaking of clang 2015-05-21 06:31:55 the clang 3.6.0 is able to build hello world 2015-05-21 06:32:04 so i think i just move it to main? 2015-05-21 07:20:09 ncopa: yeah go for it 2015-05-21 07:21:13 ncopa: i was thinking we should put some "hints" in our livecd 2015-05-21 07:21:28 ncopa: a common thing i notice is people landing at a login prompt and not knowing what to do next 2015-05-21 07:28:37 kaniini: we could add a "run setup-alpine" to /etc/motd 2015-05-21 07:29:00 ncopa: a good starting point but we might want to also put a notice in /etc/issue too 2015-05-21 07:29:09 and as last step of setup-alpine we could update motd 2015-05-21 07:32:16 i know lets make Alpine GNU/HURD 2015-05-21 07:32:17 ACTION ducks 2015-05-21 07:33:00 ACTION throws a hurd at kaniini and beats him with a gnu 2015-05-21 07:33:44 actually hurd's translators are cool 2015-05-21 07:37:31 if we could get that going on alpine that would be cool 2015-05-21 07:37:44 ;) 2015-05-21 07:48:54 gnu/hurd 2015-05-21 07:48:57 is painfully slow 2015-05-21 08:02:24 kaniini: have you had a chance to try it? how? 2015-05-21 08:02:51 debian gnu/hurd? 2015-05-21 08:03:20 yes 2015-05-21 08:03:27 this is just awful 2015-05-21 08:03:41 oh dear :( 2015-05-21 08:03:45 like 2015-05-21 08:03:47 there is iowait 2015-05-21 08:03:52 reading from the virtual cdrom 2015-05-21 08:06:08 are you using the qemu image? 2015-05-21 08:06:32 no 2015-05-21 08:06:34 did an install 2015-05-21 08:06:44 that's brave! :) 2015-05-21 08:13:05 how do you do make V=1 with a cmake project? 2015-05-21 08:13:22 i want see the exact cmdline 2015-05-21 08:17:51 experimental support for SATA devices was added in May 2013. On the other side, support for character devices (like sound cards) and other hardware (USB, multicore) is mostly missing. 2015-05-21 08:17:53 wow 2015-05-21 08:17:59 SATA but no SMP 2015-05-21 08:18:02 hilarious 2015-05-21 08:24:30 su apk something doesn't pick up the proxy profile :( 2015-05-21 08:35:19 kaniini: bare metal debian gnu/hurd install? 2015-05-21 08:37:26 nah vmware 2015-05-21 09:18:46 ncopa, i've some doubt if apply some patches. Ex: 125,126,127,129 2015-05-21 09:19:04 125 has an example.sh script in APKBUILD 2015-05-21 09:22:31 126 the funcion build() does not return with 0 2015-05-21 09:23:03 127 adds a .sh script in profile.d 2015-05-21 09:24:31 129, well, it's in main...then i'd like your feedback before apply 2015-05-21 09:24:43 even if looks fine to me 2015-05-21 09:34:56 fcolista: reply on mailing list and set patch status to changes requested 2015-05-21 09:40:02 to which packages? 2015-05-21 09:40:17 I mean: are my doubt correct or not? 2015-05-21 09:43:16 125: I'd get rid of example script 2015-05-21 09:43:24 if upstream doesn't provide it, dump it 2015-05-21 09:44:04 build() in 127 is wrong 2015-05-21 09:44:19 if it's needed at all, it should only have "true" inside 2015-05-21 09:44:46 same about 126 2015-05-21 09:45:17 I don't see a problem with 126 adding a file to profile.d, java is insane 2015-05-21 09:45:43 129 is ok, except commit message 2015-05-21 09:46:06 yes, noticed that. 2015-05-21 09:46:09 Ok. 2015-05-21 09:47:19 bah. amend with a typo :-/ 2015-05-21 09:47:44 ACTION shrugs 2015-05-21 12:13:02 fabled: is there a minimum firmware version number required on the rasbpi to boot alpine linux? 2015-05-21 12:13:37 ScrumpyJack, we ship the compatible firmware 2015-05-21 12:13:40 in the image 2015-05-21 12:15:43 so i need to update the firmware on the pi with what's shipped? 2015-05-21 12:17:04 ScrumpyJack, rpi loads firmware on boot 2015-05-21 12:17:25 so yes, you should use the ones from the image 2015-05-21 12:17:42 but no, there's no separate "i need to flash firmware with special procedure" step 2015-05-21 12:19:46 fabled: thanks. according to my rpi host, SD card i sent then with alpine linux boots on one of their rasbpis but not the one they host for me 2015-05-21 12:20:13 maybe he didn't insert it right :) 2015-05-21 12:20:17 ScrumpyJack, is it official image? or the one in ~tteras ? 2015-05-21 12:20:23 3.1-release or edge-snapshot? 2015-05-21 12:20:44 3.1-release image will not run Rpi version 2 boards 2015-05-21 12:53:19 fabled: it's an official image. the host said it boots if he removes the usb stick :) 2015-05-21 12:56:42 barthalion, clandmeter, ncopa: what do you think about 106? The raspicom module. 2015-05-21 12:57:58 fcolista: you can assign it to fabled, he has commented on previous patch. 2015-05-21 12:58:08 sounds good. 2015-05-21 13:06:46 umh. There's a new feature of apktool related to post-msg? 2015-05-21 13:08:15 not yet iirc 2015-05-21 13:08:39 there are some patches that rely on that afaiu 2015-05-21 13:08:49 like 120 2015-05-21 13:08:55 118 2015-05-21 13:08:57 because we decided to add it and think about such feature later 2015-05-21 13:09:09 ah, ok.. 2015-05-21 13:09:20 i remeber ncopa commented it few days ago 2015-05-21 13:09:29 but never seen commit about that feature. 2015-05-21 13:09:54 its a proposed feature for apk 2015-05-21 13:10:00 should be somwheer on the ml 2015-05-21 13:10:04 somewhere 2015-05-21 13:10:29 ok. So applying that patches is ok 2015-05-21 13:10:49 i think ncopa had some concerns 2015-05-21 13:11:19 my concerns are about using printf and colors.. 2015-05-21 13:11:20 but i dont want to speak for him, better to check ml or ask him. 2015-05-21 13:11:26 +1 2015-05-21 13:11:36 hi 2015-05-21 13:11:46 i dont want spend time no new apk-tools features now 2015-05-21 13:11:49 hi ncopa 2015-05-21 13:11:54 on* 2015-05-21 13:12:02 i want get 3.2 out 2015-05-21 13:12:27 anything that slows down 3.2 release will not make me happy ;) 2015-05-21 13:13:04 :) Do you think is safe to apply patches that relies on that feature that might be added we don't know when in the future? 2015-05-21 13:13:05 anything we can help to speed it up? 2015-05-21 13:13:28 check open issues: http://bugs.alpinelinux.org/versions/91 2015-05-21 13:13:51 there are some package upgrades 2015-05-21 13:13:58 i think some of them can be closed 2015-05-21 13:14:13 the package requests can be moved to 3.3.0 target 2015-05-21 13:14:22 fcolista: we already merged some of them 2015-05-21 13:14:29 fcolista: there is nothing unsafe about it 2015-05-21 13:14:35 just a file that's not used anyway 2015-05-21 13:14:51 yeah, /me would mean safe for clean :) 2015-05-21 13:14:53 ok then 2015-05-21 13:17:37 I'm going to test apache mod-perl 2015-05-21 13:17:43 #2244 2015-05-21 13:19:55 fcolista: didnt you recently use #2592? 2015-05-21 13:20:19 yup 2015-05-21 13:20:24 for pwclient 2015-05-21 13:20:50 it's in main now 2015-05-21 13:20:55 proxychains-ng 2015-05-21 13:21:41 i guess you can close it 2015-05-21 13:21:45 if you verified it works 2015-05-21 13:21:56 i'm usgin it 2015-05-21 13:21:58 *using it 2015-05-21 13:26:08 ncopa: can resolved bugs be closed? 2015-05-21 13:26:18 i think so 2015-05-21 13:26:24 ok 2015-05-21 13:26:28 but you should check them first 2015-05-21 13:26:30 i mean 2015-05-21 13:26:39 ofc 2015-05-21 13:26:43 there have been some tickets that are marked as resolved 2015-05-21 13:26:51 but have comment that say that its not 2015-05-21 13:26:54 http://bugs.alpinelinux.org/issues/3921 2015-05-21 13:27:11 thats a clear case for close 2015-05-21 13:27:15 :) 2015-05-21 13:27:40 specially if it was some days it was marked resolved 2015-05-21 13:30:42 i am working on lm_sensors 2015-05-21 13:33:17 whats wrong with it? 2015-05-21 13:33:25 i think i have it running local 2015-05-21 13:33:36 sensors-detect 2015-05-21 13:33:50 i know i had to change confd 2015-05-21 13:33:54 for module loading 2015-05-21 13:34:19 http://bugs.alpinelinux.org/issues/4186 2015-05-21 13:36:40 i also changed confd conf 2015-05-21 13:41:10 anyone know of a good calendaring program to use? something that docks in a systray and shows the time etc. preferably with caldev support 2015-05-21 13:41:51 ScrumpyJack: i would be interested in that too 2015-05-21 13:42:08 i use orage 2015-05-21 13:42:14 but it has no caldav afaik 2015-05-21 13:42:42 and i only use it to display which weekday a date has 2015-05-21 13:43:15 would be awesome to have a solution for syncing with my samsung table s-calendar 2015-05-21 13:43:27 with owncloud or similar 2015-05-21 13:43:50 i'm trying to sync with owncloud on desktop too 2015-05-21 13:43:58 preferable not google 2015-05-21 13:44:09 there was other cal server 2015-05-21 13:44:19 radicale or something 2015-05-21 13:44:20 ncopa: have you see Mate in alpine linux? it looks pretty good 2015-05-21 13:44:28 (from setup-box) 2015-05-21 13:44:35 http://radicale.org/ 2015-05-21 13:44:37 i havent! 2015-05-21 13:44:56 post screenshot! 2015-05-21 13:46:56 hang on, building some more apps to use with Mate so that it has default app groups with stuff actually in the groups :) 2015-05-21 13:47:00 13th May was released mod_perl 2.0.9-rc1. It should fix #2244 2015-05-21 13:47:20 it's building now.. 2015-05-21 13:47:31 we should add a bug for algitbot 2015-05-21 13:48:06 (the calendar app is for twm - Mate already has one) 2015-05-21 13:48:33 remove its personal preference to display bugs rather then always display them. 2015-05-21 13:49:19 :) 2015-05-21 13:49:42 ok, lm_sensors shoudl be done 2015-05-21 13:49:54 then i think i only have freeswitch paths 2015-05-21 13:49:56 as blocker 2015-05-21 13:50:02 i'd like to test xen iso too 2015-05-21 13:50:09 i have a xen host 2015-05-21 13:50:28 so unless someone else give me feedback on xen i'd have to run a test myself 2015-05-21 13:50:32 didnt freeswitch get fixed today 2015-05-21 13:50:41 other issue 2015-05-21 13:50:45 oh 2015-05-21 13:51:07 i'm thinking of #4065 2015-05-21 13:51:36 oh paths, thought you made a typo paths/patchs :) 2015-05-21 13:53:30 nice, they have fixed configure options for default paths 2015-05-21 13:53:33 upstream that is 2015-05-21 13:56:04 that was about time 2015-05-21 13:56:16 that issue is from previous centrury 2015-05-21 13:56:33 radicale doesn't do webdav :( 2015-05-21 13:58:50 yeah 2015-05-21 13:58:58 mod-perl doesn't segfault :) 2015-05-21 14:02:08 \o/ 2015-05-21 14:02:36 mod_perl works 2015-05-21 14:15:06 this error ring any bells during a cmake? error: expected ')' before '(' token 2015-05-21 14:28:39 I'm getting these errors when building darktable http://sprunge.us/FMAY 2015-05-21 14:29:09 any tips before i throw my ignorance at it? 2015-05-21 14:29:21 nice sprunge 2015-05-21 14:29:35 hey, git guru: after git-am, i went to abuild -r but the patch author forgot to update the sha sum. What shoud i do? git stash, modify and git stash apply? 2015-05-21 14:30:30 i would just abuild checkum and git add/commit it. 2015-05-21 14:30:32 clandmeter: oops 2015-05-21 14:30:54 clandmeter, does not appear in such way that 'm the commiter? 2015-05-21 14:30:58 *I'm 2015-05-21 14:31:06 you are 2015-05-21 14:31:17 ..actually.. 2015-05-21 14:31:21 :) 2015-05-21 14:31:34 :) 2015-05-21 14:31:35 and its also clear you had to correct it 2015-05-21 14:31:41 ^echo^cat http://sprunge.us/LOEZ 2015-05-21 14:31:51 in the hope the submitter wont do it again. 2015-05-21 14:32:35 ncopa for freeswitch upstream seems to be just one commit, at least listend in your bug report. Bug #4065 2015-05-21 14:32:51 that's the issue i had doubt 2015-05-21 14:33:01 ERROR: nginx-naxsi: Failed to build 2015-05-21 14:33:02 crow: i just learned that there is an --enable-fhs 2015-05-21 14:33:29 'cause the checksum was wrong 2015-05-21 14:33:47 fcolista: your checksum still wrong 2015-05-21 14:33:50 ncopa but you wrote this to: PS. The --enable-layout=FHS is broken in several ways. the configure script passes over --enable-layout=FHS to apr and apr-util which does not have the FHS layout (which is not FHS compliant anyways). 2015-05-21 14:34:05 could also be the source is not static 2015-05-21 14:34:34 abuidl checksum and git diff does not show a difference 2015-05-21 14:34:52 crow: they did --enable-fhs which set their paths. it is different from the autotools' --enable-layout=FHS 2015-05-21 14:35:06 ah now i see : 2015-05-21 14:35:17 fcolista: the source on our builders is probably different 2015-05-21 14:35:37 crow: i think we want use that. mod dir moved though 2015-05-21 14:35:40 i'm testing it now 2015-05-21 14:35:44 hum 2015-05-21 14:35:58 it also no longer generate the timezones.conf.xml for some reason 2015-05-21 14:36:01 clandmeter, wait. 2015-05-21 14:36:02 upstream-fair.zip::https://github.com/gnosek/nginx-upstream-fair/archive/master.zip 2015-05-21 14:36:10 sysguard.zip::https://github.com/alibaba/nginx-http-sysguard/archive/master.zip 2015-05-21 14:36:13 fcolista: thats broken 2015-05-21 14:36:17 yeah 2015-05-21 14:36:18 you canno use master.zip 2015-05-21 14:36:22 that's github 2015-05-21 14:36:32 i had to reject that patch 2015-05-21 14:36:34 :-/ 2015-05-21 14:37:07 you can change the source to a _sha 2015-05-21 14:37:28 uh? 2015-05-21 14:37:42 i use a script to fetch the latest sha and get the date to set it as pkgver 2015-05-21 14:38:11 care to share? 2015-05-21 14:38:47 ACTION hates devs who doesn't do versioning 2015-05-21 14:39:38 http://sprunge.us/bQYI 2015-05-21 14:40:14 source would be https://github.com/$_project/archive/$_sha.tar.gz 2015-05-21 14:43:10 fcolista: i used it here http://git.alpinelinux.org/cgit/aports/tree/testing/urlview/APKBUILD 2015-05-21 14:43:45 clandmeter, got it, but in this case we have 2 master.zip 2015-05-21 14:46:27 then reject it and let submitter figure it out. 2015-05-21 14:46:36 I'm wondering how this worked before. 2015-05-21 14:46:57 it works local, cause the master.zip doesnt get downloaded anymore 2015-05-21 14:47:17 local = alpine build server 2015-05-21 14:47:44 abuild first checkes if the source is already on disk 2015-05-21 14:47:59 it might be that rolling back to the previous version, this had error the same. 2015-05-21 14:48:13 so, i think we need to disable this package. 2015-05-21 14:48:57 yes, the submitter should take care of proper source. 2015-05-21 14:49:29 ok. I'm going to set arch = "", push it, then write to ml 2015-05-21 14:49:35 https://github.com/alibaba/nginx-http-sysguard/issues/5 2015-05-21 14:49:40 saw it 2015-05-21 14:50:24 actually it can also be done differently 2015-05-21 14:51:06 https://aur.archlinux.org/packages/ng/nginx-custom/PKGBUILD 2015-05-21 14:51:14 this is how arch does 2015-05-21 14:51:35 there's no sysguard but there's upstream-fair 2015-05-21 14:51:49 i might use _sha only for sysguard 2015-05-21 15:02:59 sir, this is not what Arch does 2015-05-21 15:03:05 this is what some random Arch user uploaded 2015-05-21 15:03:43 yes, you're right barthalion 2015-05-21 15:05:05 remember me: how to disable compile for a package? 2015-05-21 15:05:57 arch="" 2015-05-21 15:52:07 i think someone need to delete upstream-fair.zip and sysguard.zip from build server 2015-05-21 15:56:42 you should rename it 2015-05-21 16:05:13 (y) 2015-05-21 16:05:41 it's ugly having this modules with the shasum as name. 2015-05-21 16:06:20 but is in testing... 2015-05-21 16:15:44 barthalion, 113 was not applied because the pkgver was wrong. mantainer resend the patch and this was applied. 2015-05-21 16:15:56 113 should be set as superseded or rejected ? 2015-05-21 16:16:27 did the author send a new patch or you modified it? 2015-05-21 16:16:33 new patch 2015-05-21 16:16:49 i didn't touched the patch. Just informed the author via ml 2015-05-21 16:16:57 and he sent a new patch. 2015-05-21 16:17:10 so, the old one is in the tree as "new". 2015-05-21 16:18:31 superseded then 2015-05-21 16:18:41 doesn't matter to be honest 2015-05-21 16:19:07 yes, doesn't change anything, but is just to be consistent.. 2015-05-21 16:24:04 ok, only 5 patches in patchwork 2015-05-21 16:24:08 2 of them i don't like 2015-05-21 16:24:18 3 needs more insight 2015-05-21 16:24:22 ttl 2015-05-21 16:24:24 bye 2015-05-21 18:01:01 ACTION tested #4065 and looks much better. Thanks ncopa / tteras. 2015-05-21 22:24:00 ncopa: ping. two things. 1) the llvm build you committed was my first submit and not my more refined third. the first is slightly buggy and can't compile itself. should i bother submitting the fixed build considering i'll have a 3.6.1 build to you within a day or two? (major distributions are already uploading 3.6.1 binaries for the project to include with their src release announcement, they just haven't hit the public server yet) 2015-05-21 22:25:12 ncopa: and 2) is PIE-by-default sufficient to pass abuild's security checks for SUID binaries? my updated patchset adds PIE-by-default functionality to clang when compiling for a triple with the vendor set to "alpine". if SSP is also required, i can add that before submitting the 3.6.1 build. 2015-05-21 22:27:38 ACTION would like to compile a variant of alpine that doesn't include gcc, essentially for nothing more than giggles and boredom... too bad libc++ is absolutely bugshit with musl, so gcc's libstdc++ is still necessary 2015-05-22 02:28:24 ttilley: we're trying to avoid SUID binaries when possible 2015-05-22 06:42:48 ttilley: i dont mind if you rebase your patches against current aports git master 2015-05-22 06:43:28 you can wait for 3.6.1 if you want, or you can submit your fixes before that if you think that the patches will still work against 3.6.1 2015-05-22 06:43:47 PIE-by-default and SUID check are different things 2015-05-22 06:43:55 and not related at all 2015-05-22 06:45:12 nd sorry for using wrong patches to begin with 2015-05-22 06:45:42 it was hard for me to see the difference 2015-05-22 06:45:54 i prefer if you submit patches against the aports git 2015-05-22 07:00:24 <[xming]> ncopa: for which arch is the x86 compiled for? 586? 2015-05-22 07:00:34 [xming] i486 2015-05-22 07:01:15 <[xming]> ncopa: thanks, I am having a hard time to get alpine to boot in a js emu 2015-05-22 07:08:17 <[xming]> hmm hz=300 2015-05-22 07:08:54 <[xming]> then fan is spinning like crazy 2015-05-22 07:09:28 js emu? 2015-05-22 07:09:31 sounds fun :) 2015-05-22 07:10:06 <[xming]> http://copy.sh/v86/ 2015-05-22 07:10:26 <[xming]> tiny core boots fine 2015-05-22 07:10:32 <[xming]> and relatively fast 2015-05-22 07:13:05 have you tried alpine-vanilla? 2015-05-22 07:14:51 <[xming]> yes normal, mini and vanilla and diff versions 2015-05-22 07:16:21 <[xming]> the goal is to try to make a decent image and see if I have make interactive CLI tutorials 2015-05-22 07:18:00 [ 0.739999] LAPIC pending interrupts after 512 EOI 2015-05-22 07:18:05 then it just hangs there 2015-05-22 07:19:49 maybe we should have a linux-tinygrsec kernel? 2015-05-22 07:19:49 <[xming]> it's just slow as dog 2015-05-22 07:19:52 for x86 2015-05-22 07:19:59 <[xming]> not hanging 2015-05-22 07:20:14 any idea why it is that slow? 2015-05-22 07:20:26 we have too much into kernel? 2015-05-22 07:20:33 lxc etc... 2015-05-22 07:20:34 <[xming]> the speed on a modern CPU is like a 2~3Mhz Pentium 2015-05-22 07:20:50 <[xming]> so it shouldn't be *that* slow 2015-05-22 07:21:16 <[xming]> lxc uses namespace, does that slows things down? 2015-05-22 07:21:35 we have memory controller 2015-05-22 07:21:42 which is said to slow things down 2015-05-22 07:21:49 i dont know if the lack of PAE makes things slow too 2015-05-22 07:22:12 cool 2015-05-22 07:22:16 it is actually booting 2015-05-22 07:23:04 <[xming]> I never got it to boot :( my CPU is too slow :D 2015-05-22 07:23:17 <[xming]> waited like 15min and gave up 2015-05-22 07:23:24 Loading hardware drivers ... 2015-05-22 07:23:35 Running: 7m 18s 2015-05-22 07:23:45 <[xming]> ncopa: try this and you see the difference http://distro.ibiblio.org/tinycorelinux/6.x/x86/release/Core-current.iso 2015-05-22 07:24:08 <[xming]> 1 min and I get shell 2015-05-22 07:24:54 ncopa: have i read this right? do we get #3504 in 3.2 ? 2015-05-22 07:25:26 ScrumpyJack: yes 2015-05-22 07:25:36 ah 2015-05-22 07:25:38 only for x86_64 2015-05-22 07:25:43 ACTION hugs ncopa  2015-05-22 07:25:53 x86_64 will do nicely :) 2015-05-22 07:27:24 <[xming]> ncopa: can I install alpine from a chroot? (ala Gentoo) 2015-05-22 07:27:32 <[xming]> if so then I can use a simple kernel 2015-05-22 07:28:03 <[xming]> otherwise I will just fire up qemu, install alpine swap the kernel and be with it 2015-05-22 07:28:16 <[xming]> but installing in a browser is fun :p 2015-05-22 07:28:18 [xming]: like this? http://wiki.alpinelinux.org/wiki/Replacing_non-Alpine_Linux_with_Alpine_remotely 2015-05-22 07:31:07 <[xming]> ScrumpyJack: that might work __Install Alpine cd-rom image on hard disk__ 2015-05-22 07:32:41 i'm confused. what's your goal? 2015-05-22 07:35:37 <[xming]> ScrumpyJack: trying to have a working alpine in a browser 2015-05-22 07:36:27 [xming]: it does work 2015-05-22 07:36:33 i have it booted here 2015-05-22 07:36:42 alpine-vanilla 2015-05-22 07:36:45 with 128 MB ram 2015-05-22 07:37:01 re: %121 i don't like very much print with colors in post-install 2015-05-22 07:37:51 <[xming]> ncopa: how long to boot? 2015-05-22 07:38:01 7min maybe? 2015-05-22 07:38:04 i didnt pay attention 2015-05-22 07:39:34 <[xming]> thne ineed more patince 2015-05-22 07:40:13 where can i find the tinycore kernel config? 2015-05-22 07:41:06 tinycore image booted in 2 mins 2015-05-22 07:41:23 virtio 2015-05-22 07:41:24 ha 2015-05-22 07:41:42 <[xming]> ncopa: http://distro.ibiblio.org/tinycorelinux/6.x/x86/release/src/kernel/config-3.16.6-tinycore 2015-05-22 07:42:39 <[xming]> 1~2min is perfectly acceptable 2015-05-22 07:42:50 <[xming]> 7~15 min is aweful :( 2015-05-22 07:43:00 fcolista: re %121 he is removing the colored output 2015-05-22 07:43:19 what he does is good 2015-05-22 07:43:52 <[xming]> especially the fist 5 min it seems that it hangs 2015-05-22 07:45:29 <[xming]> ncopa: what do you think about an interactive tutorial to install apline? :D 2015-05-22 07:45:40 sounds cool :) 2015-05-22 07:45:49 just that 2min boot time is a bit much 2015-05-22 07:46:57 <[xming]> tinycore booted in 1 min here 2015-05-22 07:47:15 <[xming]> why does it takes longer on your machine and alpine takess less? 2015-05-22 07:47:28 <[xming]> Running: 8m 07s 2015-05-22 07:47:28 dunno 2015-05-22 07:47:35 <[xming]> [ 0.000000] tsc: Fast TSC calibration failed 2015-05-22 07:47:45 <[xming]> still there after 8 min 2015-05-22 07:47:53 <[xming]> ncopa: FF or chromium? 2015-05-22 07:47:59 FF 2015-05-22 07:48:01 under alpine 2015-05-22 07:49:05 <[xming]> ncopa: I saw some nifty sites pre-booting the js VM, show some intro and when the vm is ready they unhide the element 2015-05-22 07:49:32 <[xming]> that feels like instantaneously 2015-05-22 07:50:29 <[xming]> I think chrom[e|ium] is faster, as v86 doens't use asm.js 2015-05-22 07:52:33 <[xming]> and my CPU isn't the fastest (amd 6600K) 2015-05-22 08:01:20 ncopa, yes, the colored output is what i dislike. I'll tell him on ml 2015-05-22 08:02:07 fcolista: the patch removes the colored output 2015-05-22 08:02:34 oh, yes. You are right. 2015-05-22 08:02:46 ACTION needs a pair of glasses 2015-05-22 08:05:49 <[xming]> Running: 26m 21s still no progress 2015-05-22 08:12:37 [xming]: re setting up a chroot 2015-05-22 08:12:41 it should be very simple 2015-05-22 08:13:31 apk add -U --root $rootfs --initdb --keys-dir /etc/apk/keys --repositories-file /etc/apk/repositories alpine-base 2015-05-22 08:18:41 [xming] cant you somehow check "serial" output while installin in browser, for that there is need to append few stuff to kernel boot line, but i am not sure how to check serial in that .js 2015-05-22 08:22:29 <[xming]> ncopa: ty 2015-05-22 08:22:54 <[xming]> crow: possibly, but I haven't figured out how serial works (or not) 2015-05-22 08:23:51 http://dev.alpinelinux.org/~ncopa/alpine/alpine-virtgrsec-150522-x86.iso 2015-05-22 08:24:04 tiny image that boots the virtgrsec kernel 2015-05-22 08:24:10 http://dev.alpinelinux.org/~ncopa/alpine/alpine-virtgrsec-150522-x86.iso.sha512 2015-05-22 08:24:41 it is also slow to boot 2015-05-22 08:27:27 Alpine Linux Nano! 2015-05-22 08:29:00 ncopa can we have with 3.2 serial output in iso already? 2015-05-22 08:30:59 do you need to boot a real cdrom? 2015-05-22 08:31:14 otherwise its just to add a few lines in syslinux.cfg 2015-05-22 08:31:38 ncopa http://sprunge.us/ZLKX /boot/extlinux.conf 2015-05-22 08:31:59 yes 2015-05-22 08:32:00 ncopa well exactly, i always forget to add these feq lines to extlinux.conf 2015-05-22 08:32:09 yes before was syslinux.cfg 2015-05-22 08:32:26 i think you can add them to your /etc/update-extlinux.conf 2015-05-22 08:32:37 i do create a bootable usb stick, and install from it 2015-05-22 08:32:56 i dont have here that file 2015-05-22 08:34:12 but nice to know that it could be added there. https://github.com/Barthalion/aports/blob/master/main/syslinux/update-extlinux#L8 2015-05-22 08:34:49 ncopa there should be just the diff or whole extlinux.conf 2015-05-22 08:34:54 this repository is old as hell 2015-05-22 08:35:13 I haven't touched it since I've push access 2015-05-22 08:36:48 barthalion well it was about update-extlinux file so i guess its still valid 2015-05-22 08:42:19 ncopa: is that modloop the same as the ones used in other images? 2015-05-22 08:42:42 (the one in alpine-virtgrsec.iso) 2015-05-22 08:45:16 no 2015-05-22 08:45:24 the modloop is with virtgrsec kernel 2015-05-22 09:07:04 it's big :) 2015-05-22 11:26:14 ncopa, re %125, having an example.sh directly in the APKBUILD is not very clean. What's the alternative? barthalion said tht since is not included in the upstrem, we should avoid. I don't have strong opinion on that. 2015-05-22 11:26:48 put it in $source and then install from package() ? 2015-05-22 11:26:54 i prefer having an example.sh directly in the APKBUILD rather than having an additional example.sh in $source 2015-05-22 11:27:14 can i ask the reason? 2015-05-22 11:27:37 From tech point of view i'm interested 2015-05-22 11:27:37 the file is very small 2015-05-22 11:28:21 many few tiny files is slower than fewer slightly bigger files 2015-05-22 11:28:25 you can't set permissions in this way, so should be done manually later 2015-05-22 11:28:43 you can set permissions with umask 2015-05-22 11:28:49 ncopa: I think we shouldn't have example.sh at all 2015-05-22 11:28:53 it's not part of upstream tarball 2015-05-22 11:28:58 user can look it up in docs 2015-05-22 11:29:05 yeah 2015-05-22 11:29:15 that would probably be better 2015-05-22 11:29:35 but in general, if you have a 5 line file you want ship 2015-05-22 11:29:41 i prefer inline it 2015-05-22 11:29:52 rather than create another file 2015-05-22 11:30:07 the aports tree is already slow due to many files 2015-05-22 11:30:41 so to answer fcolista's question: to keep number of files down 2015-05-22 11:46:43 Thx ncopa for the explanation. I'm going to apply the patch as-is, then. 2015-05-22 11:52:17 ok only 3 patches in patchwork. Those should be reviewed from someone else (like fabled and ncopa) :) 2015-05-22 11:56:24 ncopa, http://bugs.alpinelinux.org/issues/4235 2015-05-22 11:59:46 where was that page that showed all the alpine packages and their reverse dependencies? 2015-05-22 12:01:54 maybe this? http://pkgs.alpinelinux.org/packages 2015-05-22 12:02:32 -- The CXX compiler identification is unknown 2015-05-22 12:02:32 CMake Error at CMakeLists.txt:25 (project): 2015-05-22 12:02:32 The CMAKE_C_COMPILER: 2015-05-22 12:02:32 ccache gcc 2015-05-22 12:02:32 is not a full path and was not found in the PATH. 2015-05-22 12:03:07 ncopa, weird 2015-05-22 12:03:17 i think the cmake script is a bit stupid 2015-05-22 12:03:36 it probably check if "$CXX" exists 2015-05-22 12:03:53 i just used that build to build 3.6.0, and then (after tweaking prepare() to use svn checkout) used the same build to compile 3.6.1 2015-05-22 12:04:28 ncopa, and yeah, that was the page i was looking for. thanks :) 2015-05-22 12:05:29 ncopa, the APKBUILD file sets CC and CXX if they're not -already- set to something... defaulting to gcc since that seems sane. then it builds the bootstrap compiler and _resets_ those variables to point to the bootstrap compiler 2015-05-22 12:05:48 ncopa, it probably needs a full path for CC/CXX 2015-05-22 12:07:19 i have my CXX="ccache g++" 2015-05-22 12:07:24 which is not a full path 2015-05-22 12:08:08 yeah. i can only assume that what it wants is an absolute path 2015-05-22 12:09:00 why does it not just use $CXX? 2015-05-22 12:09:00 it might also consider it as a singular argument, so it cant find a literal "ccache g++" as a singular file 2015-05-22 12:09:09 yes 2015-05-22 12:09:19 for some reason it tries to be smart 2015-05-22 12:09:37 ncopa, it does use $CXX or you wouldn't be seeing that error message :p 2015-05-22 12:10:05 no, it does not use it 2015-05-22 12:10:14 it tests if it exists 2015-05-22 12:11:36 ok 2015-05-22 12:11:42 simply removing that logic makes it work 2015-05-22 12:12:04 huh 2015-05-22 12:12:13 http://sprunge.us/BPKI 2015-05-22 12:12:24 ah. my quoting around might have made it interpret that as a single argument 2015-05-22 12:12:24 cmake finds CXX itself 2015-05-22 12:13:02 fair enough 2015-05-22 12:13:04 i think it takes it as a single argument when you use -DCMAKE_CXX_COMPILER="${CXX}" 2015-05-22 12:13:20 the rest should be ok 2015-05-22 12:13:34 ncopa: i made sure the entire build file was well commented 2015-05-22 12:13:53 ncopa: please let me know what you think and any code review is greatly appreciated 2015-05-22 12:14:28 that build also works as-is for 3.6.1. 2015-05-22 12:14:34 very nice 2015-05-22 12:29:23 Expected Passes : 11816 2015-05-22 12:29:23 Expected Failures : 75 2015-05-22 12:29:23 Unsupported Tests : 181 2015-05-22 12:29:23 Unexpected Failures: 140 2015-05-22 12:30:07 some of the unit tests fails 2015-05-22 12:30:48 [ FAILED ] AllocationTests/MappedMemoryTest.ZeroSizeNear/5, where GetParam() = 2015-05-22 12:30:48 117440512 2015-05-22 12:31:11 0.o 2015-05-22 12:31:11 which test suite was that? 2015-05-22 12:31:20 various 2015-05-22 12:31:34 [37566411.442695] PAX: From 79.160.13.133: execution attempt in: ing>, 7c59e7868000-7c59e786a000 7c59e7868000 2015-05-22 12:31:34 [37566411.442700] PAX: terminating task: /home/ncopa/aports/main/llvm/src/build/bin/lli(lli):4851, uid/euid: 1000/1000, PC: 00007c59e7869010, SP: 00007ef519af0158 2015-05-22 12:31:49 ncopa, note that i'm using a vanilla kernel 2015-05-22 12:31:52 seems like its some jit or something 2015-05-22 12:32:01 pax is not happy with the testes 2015-05-22 12:32:13 blargh 2015-05-22 12:32:37 i am ok to skip the unit tests 2015-05-22 12:32:59 ncopa, i haven't run the test suites on armhf 2015-05-22 12:33:18 so someone should probably run those, under a vanilla kernel, on arm 2015-05-22 12:33:20 i think the llvm does soem JIT testing 2015-05-22 12:33:23 which fails 2015-05-22 12:33:30 yes, MCJIT 2015-05-22 12:33:44 JIT and PaX does not work well together 2015-05-22 12:33:58 it's a core component of LLVM and clang can take advantage of 2015-05-22 12:34:07 optionally 2015-05-22 12:34:24 pax makes sure you cannot execute from memory location you can write to 2015-05-22 12:34:44 and it makes sure you cannot write to memory locations you can execute from 2015-05-22 12:35:07 the concept of JIT defeats that 2015-05-22 12:35:26 must make modern javascript engines... fun 2015-05-22 12:35:57 we disable pax for those 2015-05-22 12:36:18 <[xming]> hehe this feels like 20 years ago 2015-05-22 12:36:38 <[xming]> got a self compiled kernel without any modules 2015-05-22 12:36:51 <[xming]> booting alpine in less than 2 min 2015-05-22 12:37:00 cool :) 2015-05-22 12:37:16 <[xming]> I cna't mount by UUID w/o intramfs 2015-05-22 12:37:30 <[xming]> UUID isn't written to partition table during setup? 2015-05-22 12:38:19 you need mdev do its thing to get /dev/by-uuid or something 2015-05-22 12:38:21 <[xming]> I was afaid that the js emu won't write to the disk image, but when it's local it does write, which is good news :D 2015-05-22 12:38:45 <[xming]> I meant the root=PARTUUID=xxxxxx 2015-05-22 12:38:50 <[xming]> no mdev yet 2015-05-22 12:39:09 yes 2015-05-22 12:39:50 <[xming]> Oh only EFI partitions 2015-05-22 12:39:53 i think mount utility uses blkid for mount UUID=.... 2015-05-22 12:40:33 <[xming]> an MSDOS * partition using the format SSSSSSSS-PP, where SSSSSSSS is a zero- * filled hex representation of the 32-bit "NT disk signature", and PP 2015-05-22 12:41:15 <[xming]> it makes sense, I never knew that MSDOS partition table had space for that 2015-05-22 12:41:45 <[xming]> but he, I've never mount by UUID in my life except on those auto install boxes 2015-05-22 12:42:18 ncopa, i have a friend with a raspberry pi i could probably borrow for armhf testing, but otherwise i have no other way to run the llvm/clang tests on arm/armhf. (unless there's a sane emulator?) 2015-05-22 12:42:40 i have a couple of wandboards 2015-05-22 12:42:40 if you think that qemu is sane… 2015-05-22 12:43:12 the 3.2 builder is currently busy with unittesting libreoffice... 2015-05-22 12:44:13 <[xming]> ttilley2: there is an ARM emu in js :p 2015-05-22 12:45:22 gem5 can emulate arm too 2015-05-22 12:45:23 <[xming]> I hope someone will make a 4GB/8GB RAM SoC with the new mediatek deca-cores 2015-05-22 12:45:28 ncopa, the triple for your arm builds, what is it? i had assumed armv6-alpine-linux-musleabihf and the same for armv7 2015-05-22 12:45:31 but qemu is probably better 2015-05-22 12:45:37 ncopa, that's all i support in the patchset 2015-05-22 12:45:43 ncopa, is that correct? 2015-05-22 12:47:18 (clang almost all behavior off of the triple passed in, and triple set as the default. the rest is based on the detected distribution of the host for guessing what features binutils supports) 2015-05-22 12:49:39 make check-clang passed though 2015-05-22 12:49:50 i am slightly scared to push those tests 2015-05-22 12:49:58 it will probably keep the arm builder busy for days... 2015-05-22 12:50:03 ncopa, lol 2015-05-22 12:50:28 ncopa, i'll see if i can borrow a raspberry pi and let -that- sit in a corner for days instead :p 2015-05-22 12:50:31 i would have prefered that the arm builder could catch up before we do the 3.2 release 2015-05-22 12:53:32 ncopa: is there an ETA on 3.2 btw? 2015-05-22 12:54:17 ncopa: ok, i'll have a raspberry pi within a few hours. i'll set that bitch up and start running the llvm/clang/compiler-rt test suite locally 2015-05-22 12:55:05 ncopa: just want your confirmation that the above triples are what you use (armv7-alpine-linux-musleabihf) 2015-05-22 12:55:33 <[xming]> oh my, even network works, there is a default websocket gateway 2015-05-22 12:57:21 erethon: i was hoping to do 3.2 today 2015-05-22 12:57:31 but arm has not catched up 2015-05-22 12:57:39 so maybe monday 2015-05-22 12:58:12 fabled: is armv7-alpine-linux-musleabihf correct? 2015-05-22 12:58:22 ? 2015-05-22 12:58:36 no we use armv6 currently 2015-05-22 12:58:49 i think i had armv7 preliminarily added to gcc apkbuild 2015-05-22 12:58:52 both are supported 2015-05-22 12:58:53 but builders use armv6 2015-05-22 12:59:27 i think the question was which triplet we use for armhf 2015-05-22 12:59:32 http://bugs.alpinelinux.org/issues/4235 2015-05-22 12:59:43 its for clang/llvm 2015-05-22 12:59:54 ncopa: thnx, guess I'll be installing alpine on my work pc on monday then :) 2015-05-22 12:59:58 fabled: i just wanted to make sure i got the right triple for alpine. i added support for armv6-alpine-linux-musleabihf and the same for armv7. 2015-05-22 13:00:11 but it has to be exactly that triple to trigger everything 2015-05-22 13:00:32 ttilley, armv6-alpine-linux-musleabihf is the one currently in use on arm builds 2015-05-22 13:00:50 i think my plan was to later add armv7-alpine-linux-musleabihf, if we start doing thumb2 2015-05-22 13:00:57 awesome, then everything is good to go and i just need to run the test suite 2015-05-22 13:01:17 arg, successfully built darktable, but it seg faults when importing images :'( 2015-05-22 13:01:42 ncopa: fabled: i'll have a raspberry pi in a few hours and run the test suites for llvm/clang/compiler-rt so it doesn't clog the buildbot for 3.2. i just want to make sure the test suites are run 2015-05-22 13:04:38 <[xming]> aw disk img isn't persistent, it was cached somehow :( 2015-05-22 13:05:17 <[xming]> only way is use plan9 fs which stores changes in the browser file storage 2015-05-22 13:05:34 <[xming]> this is too bad 2015-05-22 13:06:41 ncopa, i also want to see if i can bump isl to 0.14 and cloog to a version that works with 0.14 without breaking gcc's graphite support... then i can include polly with clang which does pretty much the same thing. but THAT is a task for another day 2015-05-22 13:08:23 ttilley: i dont want bump isl and cloog til after v3.2 is released 2015-05-22 13:08:42 should gdb depend on python? 2015-05-22 13:08:43 fair enough 2015-05-22 13:09:56 argh 2015-05-22 13:11:54 <[xming]> lol rage quit? :D 2015-05-22 13:11:54 now what... 2015-05-22 13:12:14 :-O 2015-05-22 13:13:04 <[xming]> or maybe kicked? :p 2015-05-22 13:14:44 why would he rage quit? 2015-05-22 13:14:55 he just left 2015-05-22 13:17:20 i wonder what happened 2015-05-22 13:17:28 i tried it here 5mins ago 2015-05-22 13:17:30 and it worked 2015-05-22 13:17:35 now it does not 2015-05-22 13:19:23 missing deps? :) 2015-05-22 13:20:04 -- Performing Test HAVE_U_INT64_T - Success 2015-05-22 13:20:04 CMake Error at cmake/config-ix.cmake:280 (message): 2015-05-22 13:20:05 libffi includes are not found. 2015-05-22 13:20:10 $ pkg-config --cflags libffi 2015-05-22 13:20:10 -I/usr/lib/libffi-3.2.1/include 2015-05-22 13:20:43 ncopa: can you chuck my aports into testing? :) 2015-05-22 13:21:01 $ abuild -r 2015-05-22 13:21:02 Package libffi was not found in the pkg-config search path. 2015-05-22 13:21:02 Perhaps you should add the directory containing `libffi.pc' 2015-05-22 13:21:24 i know whats wrong 2015-05-22 13:21:39 _cmake_common_flags="\ 2015-05-22 13:21:40 ${_cmake_flags} \ 2015-05-22 13:21:40 -DCOMPILER_RT_TEST_TARGET_TRIPLE=$CBUILD \ 2015-05-22 13:21:40 -DLLVM_DEFAULT_TARGET_TRIPLE=$CBUILD \ 2015-05-22 13:21:40 -DLLVM_HOST_TRIPLE=$CHOST \ 2015-05-22 13:21:40 -DFFI_INCLUDE_DIR=$(pkg-config --cflags-only-I libffi | sed 's|^-I||g') \ 2015-05-22 13:21:42 -DLIBXML2_INCLUDE_DIR=$(pkg-config --cflags-only-I libxml-2.0 | sed 's|^-I||g') \ 2015-05-22 13:21:46 is in global scope 2015-05-22 13:21:56 can I disable grsec at boot with a flag? 2015-05-22 13:22:22 so it tries to set set the variable before the deps are installed 2015-05-22 13:22:35 ScrumpyJack: it is a kernel build 2015-05-22 13:22:47 so you'd have to run another kernel 2015-05-22 13:22:54 eg. linux-vanilla 2015-05-22 13:23:17 i think there might be certain grsec features you can disable at boot time 2015-05-22 13:26:28 thought so. what's the easiest way to get linux-vanilla along-side linux-grsec? 2015-05-22 13:30:43 ttilley: why do you need to set -DFFI_INCLUDE_DIR 2015-05-22 13:32:14 ncopa: habit from the autoconf build. i think cmake has logic to check a set of specific possible locations, but doesn't actually check pkg-config first. i could be wrong. 2015-05-22 13:32:27 lloks like you are right 2015-05-22 13:32:32 same for libxml2 2015-05-22 13:32:44 sigh 2015-05-22 13:33:01 so i check pkg-config for it 2015-05-22 13:33:02 heh 2015-05-22 13:33:13 -DFFI_INCLUDE_DIR=$(pkg-config --cflags-only-I libffi | sed 's|^-I||g') \ 2015-05-22 13:33:19 i find that very ugly 2015-05-22 13:33:46 isnt there a -DFFI_CPPFLAGS 2015-05-22 13:33:53 or -DFFI_CFLAGS 2015-05-22 13:34:10 nope 2015-05-22 13:34:36 just FFI_INCLUDE_DIR and FFI_INCLUDE_PATH (for listing individual files i think) 2015-05-22 13:35:45 For a Release build, the default flags will be "-O3 -NDEBUG". 2015-05-22 13:35:57 ncopa: i take it you haven't even looked at the commented out hack to get the header directories for libsubc++... lol 2015-05-22 13:36:09 its commented out 2015-05-22 13:36:11 sigh 2015-05-22 13:36:14 i regrep pushing it 2015-05-22 13:36:23 regret* 2015-05-22 13:36:27 ncopa: if you want -Os then you can use MinSizeRel 2015-05-22 13:36:34 now i have to clean it up 2015-05-22 13:36:44 i want whatever user wants 2015-05-22 13:36:49 eg. set in /etc/abuild.conf 2015-05-22 13:37:18 if someone wants to build world with -O3 then he shoudl be able to set it one place 2015-05-22 13:37:23 and not sed every APKBUILD 2015-05-22 13:37:43 nor should he need rebuild a custom toolchain 2015-05-22 13:38:03 ncopa: i asked for code review... 2015-05-22 13:38:07 ^^; 2015-05-22 13:38:14 :) 2015-05-22 13:38:18 yeah 2015-05-22 13:38:30 ok 2015-05-22 13:38:55 the _variable=$(pkg-config ...) in global scope will not work 2015-05-22 13:39:18 because the build time dep is not installed at the time that is parsed 2015-05-22 13:39:30 aubild will parse the APKBUILD 2015-05-22 13:39:33 find makedepends 2015-05-22 13:39:39 install those 2015-05-22 13:39:42 and the build the stuff 2015-05-22 13:40:04 ah. so i need to set the cmake flags from within the build function itself 2015-05-22 13:40:08 yes 2015-05-22 13:40:16 i am cleang that up 2015-05-22 13:40:35 also, i find the pkg-config | sed very ugly 2015-05-22 13:40:43 i'd prefered patch the cmakelists.txt 2015-05-22 13:41:17 or just set CFLAGS="$CFLAGS $(pkg-config --cflags libffi) ..." 2015-05-22 13:42:03 ncopa: in an hour or so i 'm heading to a friend's place to molest their raspberry pi to make sure the tests all pass on armv6-alpine-musleabihf. i can do the cleanup myself if you leave me feedback on the bug so i'm not wasting your time 2015-05-22 13:42:25 ok 2015-05-22 13:43:31 actually, thats him now. gotta head out. be back in a few :) 2015-05-22 13:50:01 uh 2015-05-22 13:50:35 umh..here build. 2015-05-22 13:58:32 I'm wondering how can i replicate this error here 2015-05-22 14:13:26 ncopa, what's change between the 3.2 and edge builder? 2015-05-22 14:14:02 fcolista: only the building container 2015-05-22 14:14:04 oh 2015-05-22 14:14:09 3.2 does not build testing at all 2015-05-22 14:14:10 on edge trayer doesn't build (it returns error on systray.o). 2015-05-22 14:14:14 ah 2015-05-22 14:14:22 3.2 only main builds 2015-05-22 14:14:41 umh..in my edge build env trayer builds 2015-05-22 14:21:42 http://bld2.alpinelinux.org/buildlogs/build-edge-x86_64/testing/trayer/trayer-1.0-r1.log 2015-05-22 14:22:00 that looks like a parallel build issue 2015-05-22 14:22:14 how do you figure? 2015-05-22 14:22:19 i can set make -j1 then 2015-05-22 14:22:27 make: *** No rule to make target 'systray/systray.o', needed by 'trayer'. Stop. 2015-05-22 14:22:57 looks like the build script expect that the file is just there 2015-05-22 14:23:00 ... at that point... 2015-05-22 14:23:11 a correct make file would add that as a dep 2015-05-22 14:23:20 oh 2015-05-22 14:23:23 it is a dep 2015-05-22 14:23:29 it just does not know how to build it 2015-05-22 14:23:33 this is a bug since 2005 2015-05-22 14:24:03 http://www.archivum.info/bug-wranglers@lists.gentoo.org/2005-03/00323/%28bug-wranglers%29-%28Bug-84675%29-New-trayer-doesn%27t-merge-because-of-an-error-in-the-makefile-make---No-rule-to-make-target-%60s.html 2015-05-22 14:24:11 well, at least is the same error. 2015-05-22 14:24:36 But makefile has target for systray/systray.o 2015-05-22 14:25:46 ncopa, looks you are right 2015-05-22 14:25:48 "Nicolas, try lowering your MAKEOPTS to -j2." 2015-05-22 14:26:06 "It seems that there is a bug in Makefile causing race condition in parallel make (that's why compilation works only sometimes). Temporarily I have added -j1 to emake but I will report it to the upstream." 2015-05-22 14:26:13 nice that this is 10 years old 2015-05-22 14:27:06 ACTION is still wondering how ncopa figured that this was a parallel build issue 2015-05-22 14:28:01 it needed systray/systray.o 2015-05-22 14:28:06 but didnt know how to buidl it 2015-05-22 14:28:14 so it is probably build in another subdir or so 2015-05-22 14:28:44 ACTION realize how much needs to learn 2015-05-22 14:29:10 -j1 is lame :) 2015-05-22 14:29:20 :) at least -j2... 2015-05-22 14:29:28 -j2 is unsafe 2015-05-22 14:29:31 it's just to be sure 2015-05-22 14:29:49 -j3 fails 2015-05-22 14:30:01 which option do we have? :) 2015-05-22 14:30:24 https://gist.github.com/ncopa/89b6a92893e544ee302f 2015-05-22 14:30:29 look at the makefile 2015-05-22 14:31:12 all: $(TARGET) 2015-05-22 14:31:26 SUBDIRS = systray 2015-05-22 14:31:39 subdirs is: cd systray && make 2015-05-22 14:31:42 more or less 2015-05-22 14:32:10 i think you could have done: make subdirs && make 2015-05-22 14:32:44 interesting 2015-05-22 14:33:04 weird that it does not work 2015-05-22 14:33:32 subdirs: $(SUBDIRS) 2015-05-22 14:33:36 that is the problem 2015-05-22 14:33:53 if 'systray' (directory) exists 2015-05-22 14:33:56 so this subdirs does not do "cd systray && make( 2015-05-22 14:33:59 the dependency is satisfied 2015-05-22 14:36:37 bah 2015-05-22 14:36:42 the makefile is broken on many levels 2015-05-22 14:36:52 he probably had no clue what he was doing 2015-05-22 14:37:15 :) 2015-05-22 14:37:34 whats that sed about? 2015-05-22 14:37:38 sed -i 's/$(LIBS) $(OBJ) $(SYSTRAYOBJ)/$(OBJ) $(SYSTRAYOBJ) $(LIBS) -lX11/' Makefile 2015-05-22 14:42:08 ncopa, it's re linking x11 libs, i tought it was a bug in Makefile 2015-05-22 14:42:31 i saw this on aur package 2015-05-22 14:42:44 so it should be patched in that case 2015-05-22 14:43:00 alternatively: make LIBS+='-lX11' or similar 2015-05-22 14:43:16 yeah, it's the same 2015-05-22 14:43:24 maybe is cleaner 2015-05-22 14:47:47 <[xming]> barthalion: that was just a joke on the timing of things 2015-05-22 15:44:25 have a good weekend campers. it's a three day weekend here in the UK! \o/ 2015-05-22 16:40:52 ncopa: uhhh.... so... this raspberry pi has 512m of memory, 300m of which is already in use by tmpfs filesystems. how am i supposed to actually compile anything? (excuse my noobness here, this is the first time i've even touched a pi, and i'm just doing it to test the llvm build) 2015-05-22 16:41:16 i have a 30G+ card in there, nothing should be using tmpfs 2015-05-22 17:01:36 how do you guys typically build the armv6 packages? this raspberry pi just doesn't have enough memory 2015-05-22 17:01:43 cross-compile? 2015-05-22 17:17:25 no, we have more powerful hardware 2015-05-22 17:17:29 with 4GB ram iirc 2015-05-22 17:22:57 barthalion: i'm about to install raspian on the thing and install alpine into a chroot just so it's not trying to install everything to a tmpfs filesystem ^^; 2015-05-22 17:23:17 you can install alpine as normal OS as well, I guess 2015-05-22 17:23:28 (no idea to be honest, but why wouldn't it be possible) 2015-05-22 17:23:37 I haven't bought my rpi2 yet 2015-05-22 17:24:26 barthalion: doesn't seem to work, as that depends on syslinux and syslinux won't boot on pi 2015-05-22 17:24:33 they have their own custom 3 stage bootloader 2015-05-22 17:24:44 fabled might know more about it 2015-05-22 17:25:19 but then again, even with 512m of ram, i'm not sure llvm can compile 2015-05-22 17:25:38 wonder if i can set up a qemu style armv6 emulation environment that isn't asstacular 2015-05-22 17:32:50 yeah, fuck it. emulation it is 2015-05-22 18:32:33 barthalion: is there a sane way to check if you're using a PaX kernel in a script? i have no idea, i've just stuck to vanilla 2015-05-22 18:32:57 i want to run the tests if you're not using a PaX kernel, but skip them if you are 2015-05-22 18:39:27 uname output, maybe? 2015-05-22 20:38:22 ncopa: barthalion: what's the cutoff for 3.2? cuz setting up a reasonable armhf testing setup is more painful than i expected ;) 2015-05-22 20:38:52 and i still want to actually run the full test suite on armhf, which might take a day itself >_> 2015-05-22 20:43:06 probably monday 2015-05-22 20:43:26 I'm not the guy who cuts releases here, but ncopa wanted to do it today 2015-05-22 20:43:55 ok, i'll be good by monday for sure 2015-05-22 21:58:36 maybe push it before monday, http://www.postgresql.org/about/news/1587/ if possible 2015-05-22 23:49:02 https://weakdh.org/sysadmin.html#openssh 2015-05-22 23:49:17 we should probably generate new DH groups at keygen time 2015-05-23 23:36:12 does testing/jreen require testing/qca, does not show in deps, http://pkgs.alpinelinux.org/package/testing/x86_64/jreen ? 2015-05-24 01:03:10 so I just discovered alpine linux… really nice job, guys 2015-05-24 01:11:19 so is doing something like ‘for pkg in $(apk info); do cd “path/to/$pkg” && abuild -R; done’ if I want to rebuild all installed packages with a different, say, optimization level, an okay approach? 2015-05-24 09:21:22 probably not 2015-05-24 09:21:46 different "pkgbases" can provide different packages 2015-05-24 19:37:37 Hey there. 2015-05-24 19:37:46 Where can I download the setup-bootable script? 2015-05-24 20:00:43 valera_rozuvan - http://hastebin.com/adivowuqew.parser3 2015-05-24 20:02:34 actually where it lives? 2015-05-24 20:02:40 aports? 2015-05-24 20:03:05 BitL0G1c: Thanks. 2015-05-24 20:03:13 install an LXC container http://wiki.alpinelinux.org/wiki/LXC#Alpine_Template 2015-05-24 20:03:22 you find them all under /sbin 2015-05-24 20:03:23 barthalion: I couldn't find it in the .iso file. 2015-05-24 20:03:37 BitL0G1c: it doesn't really answer my question 2015-05-24 20:04:10 /main/apk-config i think 2015-05-24 20:04:34 we have such package? :o 2015-05-24 20:04:55 /main/apk-tools 2015-05-24 20:05:11 BitL0G1c: In the wiki it says http://wiki.alpinelinux.org/wiki/Alpine_setup_scripts#setup-bootable : This page summarizes the low-level behavior of the /sbin/setup-* scripts on the Alpine ISO (and in a normal Alpine install). 2015-05-24 20:05:31 I don't think it's in apk-tools 2015-05-24 20:05:49 /sbin/setup-* scripts - where is this /sbin folder in the .iso files that are found on https://www.alpinelinux.org/downloads/ ? 2015-05-24 20:06:10 they are under /sbin on the iso 2015-05-24 20:06:20 Nope. Just downloaded - not there. 2015-05-24 20:06:24 & under /sbin in alpine lxc containers 2015-05-24 20:06:27 it's not in the aports 2015-05-24 20:06:50 For example - http://wiki.alpinelinux.org/cgi-bin/dl.cgi/v3.1/releases/x86/alpine-3.1.4-x86.iso . 2015-05-24 20:07:07 valera_rozuvan: if they are not on the iso, something went really, really wrong 2015-05-24 20:07:18 Extracted the archive. There are only /boot and /apk folders. 2015-05-24 20:07:30 OK. Than this is a bug. 2015-05-24 20:07:40 ah, no, it's fine 2015-05-24 20:07:58 it's in one of packages under /apk 2015-05-24 20:08:21 those are regular tar archives, so you can list files with tar tf 2015-05-24 20:08:24 Then should the wiki page be updated? 2015-05-24 20:08:30 OK - thx. 2015-05-24 20:08:59 (so it has to be in aports actually…) 2015-05-24 20:09:10 What is "aports"? 2015-05-24 20:09:39 git repository containing build scripts for all our packages 2015-05-24 20:09:58 Is this git repository accessible publicly? 2015-05-24 20:09:58 the setup scripts live @ http://dev.alpinelinux.org/archive/alpine-conf/ 2015-05-24 20:10:31 pkg is alpine-conf not apk-tools 2015-05-24 20:10:38 valera_rozuvan: sure thing, http://git.alpinelinux.org/cgit/aports/ 2015-05-24 20:10:41 BitL0G1c: ah 2015-05-24 20:10:45 BitL0G1c: thabks 2015-05-24 20:10:47 thanks even 2015-05-24 20:10:55 ;-) 2015-05-24 20:11:34 I could use some sleep, apparently 2015-05-24 20:55:50 <[xming]> is there any other bootloader for x86 besides extlinux/syslinux? 2015-05-24 20:55:58 probably not 2015-05-24 20:56:02 <[xming]> syslinux is pulling 5MB dependencies 2015-05-24 20:56:25 <[xming]> that's almost 50% of my image :( 2015-05-24 20:57:02 <[xming]> well pygrub for xen doesn't count :D 2015-05-24 20:57:59 <[xming]> I've got alpine to boot in less than 1min in js emu 2015-05-24 20:58:12 <[xming]> and instant on (restore) too :D 2015-05-24 20:58:21 what's there except syslinux and grub? 2015-05-24 20:58:46 <[xming]> grub2, lilo 2015-05-24 20:59:00 <[xming]> and rdev of course, but technically it's not a boot loader 2015-05-24 20:59:14 <[xming]> I have used rdev in 2 decades 2015-05-24 20:59:23 <[xming]> haven't* 2015-05-24 20:59:52 <[xming]> lilo is so outdated, grub is legacy and grub2 is just to complex IMHO 2015-05-24 20:59:59 <[xming]> syslinux is the best choice 2015-05-24 21:00:19 <[xming]> just taking too much space in this particular case 2015-05-24 21:01:04 <[xming]> maybe syslinux can be split in different packages, whne one want extlinux, one doesn't need mtools 2015-05-24 21:04:27 <[xming]> and mkinitfs could be optional 2015-05-25 06:46:56 hum 2015-05-25 06:47:00 mesa is broke 2015-05-25 06:47:10 http://bld1.alpinelinux.org/buildlogs/build-edge-x86/main/xorg-server/xorg-server-1.17.1-r2.log 2015-05-25 06:48:46 weird 2015-05-25 06:48:49 that does not make sense 2015-05-25 06:49:58 hum, mesa -r2 should have pulled in llvm-3.6 http://bld1.alpinelinux.org/buildlogs/build-edge-x86/main/mesa/mesa-10.5.5-r2.log 2015-05-25 06:50:03 but it didnt, it used llvm-3.5 2015-05-25 11:46:26 those fails on armhf: http://sprunge.us/EKHO 2015-05-25 11:47:12 do we want/need qemu on arm? 2015-05-25 11:52:05 alacerda: seems like mate 1.10 is out 2015-05-25 11:52:14 i suppose we want that instead of 1.9.90 2015-05-25 11:54:06 ugh 2015-05-25 11:54:26 source="..../1.9/..." 2015-05-25 11:54:45 1.9 is hardcoded there so 'abump mate-common-1.10.0' does not work 2015-05-25 11:54:53 alacerda: i'll leave it to you clean up 2015-05-25 11:56:04 i'd like to: /^source=/s/1\.9/${pkgver%.*}/ 2015-05-25 12:04:10 actually, i can sed mate-* packages 2015-05-25 12:08:53 <[xming]> ncopa: http://xming.github.io/alpine-v86/demo2/instant.html 2015-05-25 12:09:17 yay 2015-05-25 12:09:44 we should have something like that on our website 2015-05-25 12:09:54 cool! 2015-05-25 12:10:12 <[xming]> once image and state are d/l it's instant 2015-05-25 12:10:30 <[xming]> only the first time without cache it takes like 30min 2015-05-25 12:10:41 it was much faster here 2015-05-25 12:10:43 <[xming]> about 60MB to d/l 2015-05-25 12:10:48 less than minute 2015-05-25 12:10:50 ncopa, good =] thanks for the tip 2015-05-25 12:10:52 <[xming]> barthalion: okay I have a low line :D 2015-05-25 12:10:53 2015-05-25 12:11:00 [xming]: i'm on my work pc (and network) and it took ~5 seconds 2015-05-25 12:11:11 <[xming]> barthalion: that's the goal to have a interactive tutorial 2015-05-25 12:11:46 <[xming]> ncopa: I've got it to cool boot for less than 1 min 2015-05-25 12:12:04 thats awesome 2015-05-25 12:12:14 kernel changes only? 2015-05-25 12:12:16 <[xming]> but a restore is even faster 2015-05-25 12:12:25 <[xming]> very minimal kernel 2015-05-25 12:12:53 you know what 2015-05-25 12:12:53 <[xming]> ncopa: I would also like a boot loader which takes less disk space 2015-05-25 12:12:58 i'm impressed :) 2015-05-25 12:13:02 <[xming]> syslinux is taking 5MB with its dependencies 2015-05-25 12:13:17 <[xming]> also disabled hardware detection 2015-05-25 12:13:24 <[xming]> won about 5 sec 2015-05-25 12:13:51 <[xming]> maybe split syslinux into ext4linux and syslinux? 2015-05-25 12:14:12 isnt it same binary? 2015-05-25 12:14:23 this is crazy 2015-05-25 12:14:36 you can actually run alpine in the browser 2015-05-25 12:15:15 <[xming]> syslinux is pulling mtools and mkinitfs 2015-05-25 12:15:34 <[xming]> mtools is certainly not needed for extlinux 2015-05-25 12:15:45 <[xming]> mkinitfs is debateble 2015-05-25 12:15:49 ncopa: to be fair, you can run anything on the brower 2015-05-25 12:16:20 people are emulating win 3.1 in the browser to play games :P 2015-05-25 12:16:25 i suppose you can 2015-05-25 12:18:29 <[xming]> here are all the files needed in case someone wants them 2015-05-25 12:18:32 <[xming]> https://github.com/xming/alpine-v86/tree/gh-pages/demo2 2015-05-25 12:20:21 this is worth a tweet :) 2015-05-25 12:21:23 <[xming]> ncopa: not the same bin 2015-05-25 12:21:27 <[xming]> alpine-v86:~# ls -l /sbin/extlinux -rwxr-xr-x 1 root root 215712 Nov 12 2014 /sbin/extlinux alpine-v86:~# ls -l /usr/bin/syslinux -rwxr-xr-x 1 root root 207560 Nov 12 2014 /usr/bin/syslinux alpine-v86:~# 2015-05-25 12:21:34 ok 2015-05-25 12:22:25 <[xming]> it's kernel w/o modules nor initramfs 2015-05-25 12:23:16 <[xming]> disk image is 12MB without any compression, if syslinux can reoganized, I can cut down an other 4.5MB 2015-05-25 12:24:33 wow 2015-05-25 12:24:35 ok 2015-05-25 12:24:39 we will do that after v3.2 release 2015-05-25 12:24:42 this week 2015-05-25 12:25:10 <[xming]> should I put a link somewhere on the wiki? 2015-05-25 12:26:24 sure 2015-05-25 12:26:59 <[xming]> but where? Do you want it on the Mai_Page? 2015-05-25 12:27:08 <[xming]> Main_Page* 2015-05-25 12:27:12 i have no idea :) 2015-05-25 12:27:38 I would add a link on www.alpinelinux.org called "demo" 2015-05-25 12:27:45 yeah 2015-05-25 12:27:52 next to community maybe 2015-05-25 12:28:04 i don't think can be related to wiki somehow 2015-05-25 12:28:20 <[xming]> great I don't have to do anyting then :p 2015-05-25 12:28:55 here comes mate 1.10 2015-05-25 12:29:10 \o/ 2015-05-25 12:29:53 alacerda: could you please verify that I didnt forgot any mate package? 2015-05-25 12:33:30 ncopa, where's the alpinelinux.org git repo ? 2015-05-25 12:33:49 ah 2015-05-25 12:33:51 found 2015-05-25 12:33:53 http://git.alpinelinux.org/cgit/alpine-mksite 2015-05-25 12:33:56 mksite-alpine 2015-05-25 12:34:06 http://git.alpinelinux.org/cgit/alpine-mksite/tree/README 2015-05-25 12:34:37 cool thx 2015-05-25 12:35:50 2015-05-25 12:36:06 ncopa, checking... 2015-05-25 12:38:35 [xming], would it be possible to write a script that takes the current modloop and generate a small-compact iso ? 2015-05-25 12:38:43 for js-emu 2015-05-25 12:39:26 <[xming]> should be possible 2015-05-25 12:40:09 <[xming]> but with disk image and when network is implemented, one can do apk update && apk upgrade 2015-05-25 12:40:26 wow, network is possible in current demo ? 2015-05-25 12:40:46 ah, ok 2015-05-25 12:40:48 <[xming]> no, because I don't have suitable ws:// server 2015-05-25 12:41:14 <[xming]> but with a ws:// server then network works, the ws server of v86's author is down ATM 2015-05-25 12:41:39 <[xming]> in any case I don't think overloading his server with our users is nice 2015-05-25 12:41:59 can we have a similar ws server on AL infra ? 2015-05-25 12:42:17 <[xming]> if someone wants to setup that up then yes 2015-05-25 12:42:39 we have nginx with ws support 2015-05-25 12:44:03 ncopa, http://sprunge.us/GcaA 2015-05-25 12:44:08 <[xming]> vkrishn: a moment I will dig up the ws implementation, it's doing ethernet frames inside ws 2015-05-25 12:45:05 <[xming]> vkrishn: https://github.com/benjamincburns/websockproxy 2015-05-25 12:45:15 <[xming]> that's the one which needs to be installed 2015-05-25 12:45:20 hum 2015-05-25 12:45:22 i have more... 2015-05-25 12:45:53 ok 2015-05-25 12:45:57 <[xming]> and then one needs dns/dhcp and NAT/PAT for the internet 2015-05-25 12:46:03 alacerda: those got not applied for some reason: http://sprunge.us/haJJ 2015-05-25 12:46:34 ncopa, gonna check 2015-05-25 12:46:35 i'll fix those 2015-05-25 12:47:56 <[xming]> vkrishn: or great a massive LAN for alpine users do we can all test firewall, metasploit, snort, ... on each other :p\ 2015-05-25 12:48:35 :) 2015-05-25 12:49:03 <[xming]> that will be fun :D 2015-05-25 12:49:54 <[xming]> if you don't want to give internet access, at least allow access to *.alpinelinux.org 2015-05-25 12:50:00 does participant need to sign a disclaimer 2015-05-25 12:51:16 <[xming]> like? "I promise that I waon't hack you?" 2015-05-25 12:51:23 :)) 2015-05-25 12:51:41 <[xming]> or I am 1337, I will pwn u 2015-05-25 12:51:51 or "On event someone or something ... etc.. " 2015-05-25 12:52:16 <[xming]> as long as you don't allow internet access, it's all private 2015-05-25 12:52:51 <[xming]> it's like a VPN but unencrypted :D 2015-05-25 12:53:01 then "I promise not to mention it to anyone or anywhere and forgo any hackback" ;)) 2015-05-25 12:54:20 <[xming]> vkrishn: if you guys want to host everything, here are the files https://github.com/xming/alpine-v86/tree/gh-pages/demo2 2015-05-25 13:18:58 ncopa, weird... 2015-05-25 13:19:04 http://sprunge.us/LPHO 2015-05-25 13:19:04 worked fine here 2015-05-25 13:19:18 oops 2015-05-25 13:19:25 i've already solved 2015-05-25 13:19:43 You've* 2015-05-25 13:20:12 yes i pushed it 2015-05-25 13:20:16 sorry 2015-05-25 13:20:17 <[xming]> hmm maybe I should put distcc on the image, then we'll have a compile farm ^_^ 2015-05-25 13:20:59 alacerda: i'm not sure what to do with mate-themes though 2015-05-25 13:21:35 what about mate-terminal? 2015-05-25 13:23:09 i had it locally here 2015-05-25 13:23:10 ncopa, mate-themes is not 1.10.0 yet, it remains 1.9. 2015-05-25 13:23:13 but i'll use your commit 2015-05-25 13:23:38 there are som gtk3 themes: http://pub.mate-desktop.org/releases/1.10/ 2015-05-25 13:23:54 i don tknow what the difference is 2015-05-25 13:24:29 thanks 2015-05-25 13:24:38 last time i talked to mate-developers they told me not to trust on gtk3 yet 2015-05-25 13:24:51 alacerda: ok so we just wait with mate-themes? 2015-05-25 13:25:18 ncopa, I would say yes 2015-05-25 13:25:30 ok 2015-05-25 13:25:53 lets do so then 2015-05-25 13:31:11 [xming], how do you generate the .img file? 2015-05-25 13:33:02 oh, the img is the hdd 2015-05-25 13:33:11 <[xming]> with qemu 2015-05-25 13:33:11 while the bin is the bootloader 2015-05-25 13:33:22 <[xming]> bin is te state file 2015-05-25 13:33:50 <[xming]> bootloader is on the mbr of .img 2015-05-25 13:34:11 <[xming]> really just like regular qemu 2015-05-25 13:34:18 ok got it 2015-05-25 13:34:32 i tought it was something done ad-hoc 2015-05-25 13:35:35 <[xming]> it's a full PC emulator, quite impressive 2015-05-25 13:35:47 indeed. 2015-05-25 13:36:10 ncopa: i have an llvm 3.6.1 build in the ticket, but it seems we've collided mid-air. why try to patch the sanitizers btw? we don't even compile those. 2015-05-25 13:37:05 <[xming]> fcolista: state file is for resume, otherwise a cool boot takes +/- 1min, now it's matter of seconds 2015-05-25 13:37:22 http://bugs.alpinelinux.org/issues/4235 2015-05-25 13:39:45 ttilley: build failed on x86 2015-05-25 13:39:54 eep 2015-05-25 13:39:55 i had to get the build server up and run 2015-05-25 13:40:01 actually 2015-05-25 13:40:05 i messed up 2015-05-25 13:40:20 llvm was disabeld for x86 2015-05-25 13:40:26 but we need it for mesa 2015-05-25 13:40:39 so i had to make it build at least 2015-05-25 13:41:39 ...but the musl-no-sanitizers patch specifically disables building the sanitizer code, even sanitizer_common... unless that header is being included elsewhere :/ 2015-05-25 13:41:47 <[xming]> fcolista: see how long this takes -> http://xming.github.io/alpine-v86/demo2/coldboot.html 2015-05-25 13:46:03 not so much, actually 2015-05-25 13:46:27 I used to run alpine on a way lower hardware than that :p 2015-05-25 13:46:27 1min 30 sec 2015-05-25 13:46:31 slower* 2015-05-25 13:46:42 <[xming]> less than 1min here on chromium 2015-05-25 13:46:43 ttilley: interception 2015-05-25 13:46:46 <[xming]> FF is slower 2015-05-25 13:47:10 what about working network? is it possible? 2015-05-25 13:47:18 <[xming]> hm, ws server is up, I wonder why I don't get network 2015-05-25 13:47:28 <[xming]> ACTION digs 2015-05-25 13:47:55 ttilley: the http://bugs.alpinelinux.org/attachments/download/248/compiler-rt-0002-musl-no-sanitizers.patch does not comment out "interception" which failed on sizeof(::OFF_T) != sizeof(off_t) 2015-05-25 13:48:05 on glibc and android off_t is 32 bit 2015-05-25 13:48:09 on musl its 64 bit 2015-05-25 13:49:30 ncopa: fair enough 2015-05-25 13:49:57 ncopa: though the updated patchset disables the sanitizers via cmake instead of just commenting out directories in a makefile 2015-05-25 13:50:18 ok 2015-05-25 13:50:41 the whats in current git does not yet use cmake 2015-05-25 13:51:51 ncopa: sorry for the delay in responding to code review btw. had a kid's birthday to go to and i swear to god 3yos run on plutonium or something. they don't get tired. EVER. ._. 2015-05-25 13:54:02 ttilley: np 2015-05-25 13:54:07 kids first 2015-05-25 13:54:47 i mean... i dont pay you :) 2015-05-25 13:55:24 heh 2015-05-25 14:08:34 np = no pay ;) 2015-05-25 14:08:47 <[xming]> barthalion: I guess the author doesn't want his ws server being abused so he hides some little details 2015-05-25 14:09:21 <[xming]> barthalion: it's working now, try /etc/init.d/networking restart 2015-05-25 14:09:44 <[xming]> you hsould get an IP, proper default gw, dns, etc 2015-05-25 14:09:51 <[xming]> ping google.com works from here 2015-05-25 14:11:44 <[xming]> energizer batteries 2015-05-25 14:13:35 <[xming]> so the server is actually wss and not ws, but it's ws:// everywhere 2015-05-25 14:14:13 <[xming]> only if we had our own ws server 2015-05-25 14:14:21 <[xming]> ACTION pokes vkrishn  2015-05-25 14:15:57 :) 2015-05-25 14:16:14 could request ncopa or clandmeter 2015-05-25 14:16:54 I plan to get new harwares but kinda sad story here for couple of years 2015-05-25 14:17:23 <[xming]> oh I thought you co-managed AL infra 2015-05-25 14:17:33 ncopa has old AL harware lying around iirc 2015-05-25 14:17:53 <[xming]> a lxc will do 2015-05-25 14:19:31 <[xming]> people say that v86 run rather fast on an iPad Air2 2015-05-25 14:21:12 http://wiki.alpinelinux.org/wiki/Alpine_Linux:IRC_Availability see infra admins 2015-05-25 14:22:08 <[xming]> I have never seen teras 2015-05-25 14:22:32 fabled = Teräs 2015-05-25 14:22:59 <[xming]> ah 2015-05-25 14:24:19 ttilley: do you think we should hold alpine v3.2 release for llvm 3.6 fixes? 2015-05-25 14:24:38 ttilley: i am about to shoot the engineers and just ship the product 2015-05-25 14:25:24 ncopa: dont ask me man, you're the guy in charge. :p 2015-05-25 14:26:04 ncopa: my updated llvm 3.6.1 build works beautifully on x86-64, might need your additional patch for x86, and i have no sane way of testing on armhf 2015-05-25 14:26:18 ok. then i think i'll just shoot myself and ship v3.2... 2015-05-25 14:26:23 or it will never come out 2015-05-25 14:26:47 ncopa: can almost ship the fix as an update 2015-05-25 14:26:56 s/almost/always 2015-05-25 14:27:05 yeh 2015-05-25 14:27:09 yeah 2015-05-25 14:27:58 its a 6month release cycle , right ? 2015-05-25 14:27:58 ncopa: part of why i do it in a two-stage build now... to catch subtle screwups where clang passes its tests yet somehow is unable to compile itself (which is the state of that first patchset i sent you) 2015-05-25 14:28:29 ...which i guess is what's currently in the tree actually 2015-05-25 14:28:43 vkrishn: yes 2015-05-25 14:29:11 release 3.1.0 was dec 10 2015-05-25 14:29:31 it was 1 month delayed :-( 2015-05-25 14:29:38 we aim for May/Nov 2015-05-25 14:29:44 ok 2015-05-25 14:30:26 hmm... overlaping rolling release (ORR) ;) 2015-05-25 14:38:59 ncopa: what about python3-related stuff? I mean we'll move to p3 excepting packages that we can't port to p3 or sth. 2015-05-25 14:39:46 scadu: python3 didnt make it for v3.2 2015-05-25 14:39:51 but that is a prio for v3.3 2015-05-25 14:40:23 i havent thought much how to solve it 2015-05-25 14:41:25 oh I see. 2015-05-25 14:51:32 people in the real world still use python 2.7 anyways... at least if you're doing web development with django/flask/whatever 2015-05-25 14:53:41 also in the HPC world... where you are stuck w/ RHEL 2015-05-25 14:53:45 i still stick to py2 2015-05-25 14:54:35 ncopa: when you're done with your release dance, i'd still like code review on that ticket. ;) 2015-05-25 14:54:45 ncopa: ...whenever that ends up being 2015-05-25 14:55:07 ACTION hears the fabled "release dance" tends to leave one exhausted 2015-05-25 14:57:21 ok 2015-05-25 14:57:25 lets ship v3.2 tomorrow 2015-05-25 14:57:37 woohoo 2015-05-25 14:58:02 <[xming]> I bet ncopa has the whole release ritual with dance, food, drinks and herbs 2015-05-25 15:43:56 [xming]: ncopa: don't forget the shaking of dead chickens 2015-05-25 15:44:51 lol 2015-05-25 15:49:01 what just happened here. 2015-05-26 06:25:42 fabled, ncopa 2015-05-26 06:26:20 i'd like to look at what grsec is actually netting us for 3.3 and see if we could perhaps just merge the interesting features (if any) into vanilla (and i'm not convinced PaX is an interesting feature anymore because we have the other mitigations) 2015-05-26 06:26:46 not being able to upgrade the kernel with kexec is kind of sad 2015-05-26 06:27:10 as well as possibly this new ksplice-like thing 2015-05-26 06:29:26 beeing able to modify a running kernel sounds like rootkit invitation to me 2015-05-26 06:29:36 kpatch is useless at the moment 2015-05-26 06:30:10 well, kexec is still nice either way when you have minutes of downtime between reboots 2015-05-26 06:30:12 ttilley: web developers don't really care about py2 vs py3 2015-05-26 06:30:29 ACTION is working on hardware raid management 2015-05-26 06:30:52 we could maybe enable kexec in the vanilla kernel? 2015-05-26 06:31:31 well, what i mean is. i think at this point, the vanilla kernel is probably secure enough that it could be the default. 2015-05-26 06:31:44 and then if you want the extra features of grsecurity you enable them by switching to that kernel 2015-05-26 06:35:44 will have to think of that after v3.2 release 2015-05-26 06:36:00 that is what i mean, for 3.3 :) 2015-05-26 06:36:03 but in general, i like the PaX features 2015-05-26 06:37:06 there have been recently instances where the PaX features have been hurting users 2015-05-26 06:37:10 by overfragmenting the heap 2015-05-26 06:37:38 for example, a user with 4GB ram on his machine could not use any memory space larger than 300MB for mysql cache 2015-05-26 06:37:44 because of lack of contiguous pages 2015-05-26 06:38:00 this is why we have bug trackers? 2015-05-26 06:38:17 the bug can't be fixed without turning off the PaX protections 2015-05-26 06:38:38 see #alpine-linux from few days ago (maybe a week, i'll fetch a date in a moment) 2015-05-26 06:39:25 and of course, debugging the issue is made difficult because grsecurity (not PaX) obfuscates /proc information 2015-05-26 06:40:32 i know dalias also feels that the grsecurity modifications have some questionable things 2015-05-26 06:43:05 i think i still prefer tight/secure by default, and lose up when needed 2015-05-26 06:43:16 rather than lose by default and tighten up when wanted 2015-05-26 06:43:41 but lets talk about that later 2015-05-26 06:43:44 my question is, "is it really tight" 2015-05-26 06:44:05 that is of course a good question 2015-05-26 06:44:26 a valid question 2015-05-26 06:44:32 i think real-world security issues are better mitigated by doing things like removing SUID binaries 2015-05-26 06:44:48 because no doubt, grsecurity comes with a 'cost' 2015-05-26 06:44:50 more work 2015-05-26 06:44:58 various things does not work out of the box 2015-05-26 06:45:00 no JIT 2015-05-26 06:45:09 and to get JIT you have to throw out grsecurity anyway 2015-05-26 06:45:11 and no kexec... 2015-05-26 06:45:13 yes 2015-05-26 06:45:44 so the question is valid: what benefit do we get for this cost 2015-05-26 06:46:18 my point basically is, we have hardening in many more areas which have basically removed a lot of the benefit of PaX 2015-05-26 06:46:26 fortify removes buffer-overflows 2015-05-26 06:46:34 everything is PIE 2015-05-26 06:46:43 musl itself doesn't allow unsafe relocation types 2015-05-26 06:47:35 maybe in kernelspace we are still in theory getting something, but a vmsplice type thing is rare *and* we can likely mitigate that by slipping in some of grsecurity's special GCC plugin stuff into the main tree 2015-05-26 06:48:21 i think we disable kernexec, due to xen 2015-05-26 06:49:16 of course spender will probably say that grsecurity still protects against theoretical attacks like ASLR bruteforce but i've never seen any actual realworld ASLR bruteforce 2015-05-26 06:49:33 pax's pagexec probably do a big difference on x86 2015-05-26 06:49:49 but x86 is not as relevant as it once was 2015-05-26 06:50:12 yes, hince my view that perhaps it's good to keep around as an option, but the mainline stuff we have is probably better than alpine 2.x + grsec anyway in terms of realworld impact 2015-05-26 06:50:28 protects against things that is not yet realworld problems... 2015-05-26 06:50:42 that is actually one of the things i like with grsecurity 2015-05-26 06:50:51 the problem with "not yet realworld problems" is that they may never become realworld problems 2015-05-26 06:51:25 but when they do, its kinda nice to have it covered 2015-05-26 06:51:34 like insurance calculations 2015-05-26 06:51:41 sure, hince keeping it around for people who want it 2015-05-26 06:52:14 lets take that discussion later 2015-05-26 06:52:27 "If you want extra security at the possible cost of losing compatibility or having difficult to debug issues, you can install the linux-image-grsec package." 2015-05-26 06:52:49 but really what i want more than anything is a single APKBUILD that handles all of the kernel variants 2015-05-26 06:52:53 or something along those lines 2015-05-26 06:53:02 so we're shipping the same linux base kernel no matter what variant 2015-05-26 06:53:03 in theory, if you should have maximum benefit of grsecurity, you should buidl your kernel yourself 2015-05-26 06:53:59 that said, i still dont like the idea about shipping vanilla kernel by default 2015-05-26 06:54:20 well perhaps vanilla install media 2015-05-26 06:54:28 so you can choose to have either grsec or vanilla install 2015-05-26 06:54:30 but we should consider if it still is worth it 2015-05-26 06:54:36 you have alpine-vanilla iso image 2015-05-26 06:54:50 http://alpinelinux.org/downloads/ 2015-05-26 06:54:52 pick vanilla 2015-05-26 06:55:31 hmm, true 2015-05-26 06:56:03 i've also noticed as i said previously a lot of people are confused when they are dropped into a server-oriented livecd 2015-05-26 06:56:26 we might want to do what freebsd does and ask if you want LiveCD or Install mode 2015-05-26 06:58:44 like this: https://www.freebsd.org/doc/en/books/handbook/bsdinstall-start.html#bsdinstall-choose-mode 2015-05-26 06:58:56 yeah, but like at boot time so it can default to livemode 2015-05-26 06:59:08 (or perhaps we can have it not prompt if there's an apkovl?) 2015-05-26 07:02:02 yes, the question shoudl not come up if there is an apkovl 2015-05-26 07:04:08 about the installer: https://news.ycombinator.com/item?id=8355125 2015-05-26 07:04:18 >Alpine Linux 2015-05-26 07:04:19 How? The installer doesn't even work. 2015-05-26 07:04:29 it may be time to perhaps look at building a more robust install framework too :P 2015-05-26 07:04:41 yes 2015-05-26 07:04:42 i know 2015-05-26 07:05:09 244 days ago 2015-05-26 07:05:12 i've been kind of playing around with building a networkd type thing 2015-05-26 07:05:31 to replace busybox ifupdown which is fragile 2015-05-26 07:05:33 :/ 2015-05-26 07:07:21 in a way, i kind of understand why systemd guys have replaced a lot of things. the predecessors in some cases (like ifupdown) are quite fragile. 2015-05-26 07:07:49 yes 2015-05-26 07:07:55 the concept of pid files too 2015-05-26 07:08:31 but right now RAID is really the big thing i want to get going in 3.3 2015-05-26 07:08:40 some open management tools for RAIDs 2015-05-26 07:09:28 right now the 3.2 readme is what i want get going :) 2015-05-26 07:09:38 3.2 release notes 2015-05-26 07:28:34 for the raid tools, i'm starting with the LSI SAS1078/2108 chipset because it's the most common one in the world 2015-05-26 07:33:47 Controller: Smart Array P410i 2015-05-26 07:34:17 that one's not SAS1078/2108 2015-05-26 07:34:31 thats whats on the build servers 2015-05-26 07:34:33 SAS1078 is DELL PERC5 2015-05-26 07:34:38 Lenovo whatever 2015-05-26 07:34:41 Fujitsu whatever 2015-05-26 07:34:52 IBM ServeRAID (on pSeries) 2015-05-26 07:34:52 build servers are HP 2015-05-26 07:35:15 i think i have a box with dell perc5 2015-05-26 07:35:24 i have 5 PERC5i cards 2015-05-26 07:35:29 sitting in my cabinet 2015-05-26 07:35:44 i just ordered 2 more off ebay 2015-05-26 07:36:02 # cat /sys/block/sdb/device/model 2015-05-26 07:36:02 PERC 6/i 2015-05-26 07:36:14 the perc6 is the SAS2108 2015-05-26 07:36:29 it's the same commands for management 2015-05-26 07:36:30 i think there are some closed source tool for it 2015-05-26 07:36:33 yup 2015-05-26 07:36:43 i'm going to trace the ioctls to figure out what the closed source tool does 2015-05-26 07:36:45 ;) 2015-05-26 07:36:48 :) 2015-05-26 07:37:13 sounds fun :) 2015-05-26 07:41:38 then i will get an adaptec and an hp card and repeat the process 2015-05-26 07:41:57 that should provide sufficient base that other people will send patches 2015-05-26 07:47:24 yeah 2015-05-26 08:00:45 even better: netbsd has already reverse engineered the HP Smart Array driver 2015-05-26 08:00:56 or rather tool 2015-05-26 08:01:33 The CISS interface is defined in the document entitled CISS Command 2015-05-26 08:01:34 Interface for SCSI-3 Support Open Specification, Version 1.04, Valence 2015-05-26 08:01:36 Number 1, Compaq Computer Corporation, 2000/11/27. 2015-05-26 08:01:45 wow, it's even openly documented 2015-05-26 08:01:48 how great that card will be 2015-05-26 08:16:49 there is also cciss_vol_status 2015-05-26 08:19:03 yep. my general idea is to build something like bioctl on openbsd 2015-05-26 08:19:13 one tool to provide it for all cards 2015-05-26 08:21:09 aha 2015-05-26 08:21:14 sounds creat to me 2015-05-26 08:21:34 i think that can be big 2015-05-26 08:21:47 i mean, that is a real missing piece 2015-05-26 08:24:14 http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/dev/ic/mfireg.h?rev=1.8&content-type=text/x-cvsweb-markup&only_with_tag=MAIN 2015-05-26 08:24:22 this seems interesting, it's for megaraid cards 2015-05-26 08:25:00 all the drivers basically have an ioctl that is "pass-through to firmware" 2015-05-26 08:49:11 ok 2015-05-26 08:49:20 what do we write as news for v3.2? 2015-05-26 08:49:26 Some of the new features are: 2015-05-26 08:49:26 - mysql is replaced with mariadb 2015-05-26 08:49:26 - Lua 5.3 support 2015-05-26 08:49:26 - Ruby 2.2 2015-05-26 08:49:26 - Xen 4.5 2015-05-26 08:49:53 freeswitch 3 2015-05-26 08:50:05 freeradius i mean 2015-05-26 08:53:01 new apk? :) 2015-05-26 08:54:26 http://sprunge.us/YITY 2015-05-26 08:55:10 were there other major version numbers? 2015-05-26 08:55:27 libreoffice? 2015-05-26 08:55:37 yes we ship libreoffice now 2015-05-26 08:55:42 we didnt for v3.1 2015-05-26 08:55:56 Version: 4.4.3.2 2015-05-26 08:56:45 then kernel 3.18.14, postfix 3.0.1 ruby 2.2.0 xen 4.5.0 too? 2015-05-26 08:57:17 there is a lack of photo apps in AL so I packaged rawTherapee (in testing but works for me). you could add that if you like? 2015-05-26 08:57:59 also packaged darktable, but that sega faults :( 2015-05-26 08:58:24 sega faults? woah. 2015-05-26 08:58:33 :) 2015-05-26 08:59:15 samba 4.2 2015-05-26 08:59:36 also added geekie as an image viewer/manager, but it's not a very active project 2015-05-26 08:59:49 need to find another one 2015-05-26 09:00:00 alpine has generated lots of cool things 2015-05-26 09:00:15 opennhrp, pkgconf, now raidctl or whatever i am calling it, acf, etc. 2015-05-26 09:00:16 :p 2015-05-26 09:00:37 the plan is to put some nice default apps into setup-box (for Mate at the moment) 2015-05-26 09:00:47 kaniini: indeed 2015-05-26 09:00:55 pkgconf has been awesome btw 2015-05-26 09:01:20 just works 2015-05-26 09:01:24 never gets in the way 2015-05-26 09:01:48 and if the unlikely event of problem, it has gotten fixed within 24 hours 2015-05-26 09:01:50 :) 2015-05-26 09:02:34 ah 2015-05-26 09:02:54 we need mention gcc version and fortify 2015-05-26 09:03:35 kodi (former xbmc) was moved in main in 3.1 ? Or 3.2 ? 2015-05-26 09:03:44 I think if it is 3.2 is worthy mention it 2015-05-26 09:04:04 oh we have kodi 2015-05-26 09:04:05 yes 2015-05-26 09:05:18 postgres with bdr ? 2015-05-26 09:14:53 ncopa: what would your opinion be on patching busybox and perhaps PAM to support openwall-style tcb shadow so we can reduce SUID permissions 2015-05-26 09:18:54 first impression: sounds like a good idea to me 2015-05-26 09:27:31 there's a lot of APKBUILDs which request suid exemption 2015-05-26 09:27:59 ncopa, what's the easier way to get the latest x86 alpine stable iso? e.g. http://dl-3.alpinelinux.org/alpine/.latest.x86.txt 2015-05-26 09:28:10 but what happens if dl-3 is down? 2015-05-26 09:28:17 we do have dl.cgi 2015-05-26 09:28:32 which redirects automatically to one of the MIRROR afaiu 2015-05-26 09:29:12 would it be possible to have something like dl.cgi/stable/.latest.x86.txt ? 2015-05-26 09:29:46 so you have redundancy in case one of the mirror is down 2015-05-26 09:38:12 fcolista: we dont have that atm 2015-05-26 09:38:16 but its a good idea 2015-05-26 09:38:21 we need something like that 2015-05-26 09:38:51 just for avoiding to parse the yaml file and build the logic if one of the mirror is down 2015-05-26 09:41:27 what happens if the machine hosting dl.cgi goes down? 2015-05-26 09:48:00 i think we notice it since is the link putted on the webpage 2015-05-26 09:50:16 generally, imho is better to have a link to dl.cgi rather than having hardcoded in a script one of the mirrors. They also might change in the future. 2015-05-26 09:50:35 it's consistent. 2015-05-26 10:06:49 ncopa: i just ordered an HP Smart Array P410i controller as well 2015-05-26 10:07:09 it's coming from hong kong though so it won't get here for a while 2015-05-26 11:13:36 2015/05/26 11:13:14 unsupported GOARCH armv6 2015-05-26 11:13:53 thanks go-lang for that... 2015-05-26 11:15:39 ncopa, Go really supports only x86_64 2015-05-26 11:16:02 there's been issues on 32-bit architectures, not sure if they got them fixed 2015-05-26 11:16:15 thats what i thought too 2015-05-26 11:16:15 but 2015-05-26 11:16:25 >>> syncthing*: Tracing dependencies... 2015-05-26 11:16:25 busybox 2015-05-26 11:16:25 so:libc.musl-armhf.so.1 2015-05-26 11:16:25 >>> syncthing*: Package size: 20.9 MB 2015-05-26 11:16:25 >>> syncthing*: Compressing data... 2015-05-26 11:16:34 it builds... 2015-05-26 11:16:43 case "$CTARGET" in 2015-05-26 11:16:44 arm*) _arch="-goarch armv6";; 2015-05-26 11:16:44 esac 2015-05-26 11:16:47 i just removed that 2015-05-26 11:18:30 ncopa, it needed that earlier. sounds like they fixed go in that regard 2015-05-26 11:18:39 yup 2015-05-26 11:18:42 seems so 2015-05-26 11:18:59 they did close that 'GC broke on 32-bit architectures' bug too 2015-05-26 11:19:06 maybe it works better nowadays 2015-05-26 11:19:12 at least it compiled 2015-05-26 11:19:34 i really hate that go produce such horribly bloated binaries 2015-05-26 11:19:49 its even worse than autoconf... 2015-05-26 11:25:22 hum, raspberrypi fails to build 2015-05-26 11:25:36 ncopa, ? 2015-05-26 11:25:59 problem 1: linux-headers 2015-05-26 11:26:00 you have new alpine-iso in use? 2015-05-26 11:26:06 the package 2015-05-26 11:26:09 oh 2015-05-26 11:26:11 main/raspberrybi 2015-05-26 11:26:31 should maybe update the snapshot too 2015-05-26 11:26:43 problem 2: 2015-05-26 11:26:45 [ 46%] Building C object host_applications/linux/libs/debug_sym/CMakeFiles/debug 2015-05-26 11:26:45 _sym.dir/debug_sym.c.o 2015-05-26 11:26:45 23f9b6005/host_applications/linux/libs/debug_sym/debug_sym.c: In function 'Acces 2015-05-26 11:26:45 sVideoCoreMemory': 2015-05-26 11:26:48 23f9b6005/host_applications/linux/libs/debug_sym/debug_sym.c:606:34: error: 'PAG 2015-05-26 11:26:50 E_MASK' undeclared (first use in this function) 2015-05-26 11:26:52 memOffset = vcMemAddr & ~PAGE_MASK; 2015-05-26 11:26:54 ^ 2015-05-26 11:27:50 i thought i had patch for that 2015-05-26 11:29:25 sounds like linux-headers changed or something 2015-05-26 11:29:37 no PAGE_MASK comes from libc headers iirc 2015-05-26 11:32:29 grep cannot find it 2015-05-26 11:32:41 it's in bits/user.h 2015-05-26 11:32:48 nor with bsd-compat-headers 2015-05-26 11:33:15 $ apk info -W /usr/include/bits/user.h 2015-05-26 11:33:15 /usr/include/bits/user.h is owned by musl-dev-1.1.9-r2 2015-05-26 11:34:25 http://git.musl-libc.org/cgit/musl/tree/arch/arm/bits/user.h 2015-05-26 11:34:26 ? 2015-05-26 11:34:58 hmmm... i checked x86_64 2015-05-26 11:35:19 on arm its not there 2015-05-26 11:37:13 ncopa, i'll take new snapshot 2015-05-26 11:37:24 i think it found it earlier due to linux-headers 2015-05-26 11:37:32 but it probably got removed from 3.18 or someething 2015-05-26 11:39:30 'make all-release' should work on arm, right? 2015-05-26 11:41:51 release-notes: http://sprunge.us/VAMd 2015-05-26 11:41:59 are there other packages that got renamed? 2015-05-26 11:43:42 linux-virtgrsec 2015-05-26 11:44:39 gmp? 2015-05-26 11:44:46 libeventlog 2015-05-26 11:44:49 gst* -> gst*0.10 2015-05-26 11:45:02 but gst not worth mention i think 2015-05-26 11:45:09 nor gmp 2015-05-26 11:47:39 re: raspberrypi 2015-05-26 11:47:51 i think it's because PAGE_SIZE was not maybe defined before 2015-05-26 11:48:01 the file defines PAGE_MASK locally only if PAGE_SIZE is not defined 2015-05-26 11:48:39 sounds wrong 2015-05-26 11:48:51 #if ! defined(PAGE_MASK) ... 2015-05-26 11:56:38 fabled: are you fixing raspberrypi? 2015-05-26 11:56:51 i think it will build release once that is done 2015-05-26 11:57:10 we should probably do rc5 for that :-( 2015-05-26 12:00:47 hmm 2015-05-26 12:00:49 ok 2015-05-26 12:00:56 yes, i was working on it, but got interrupted 2015-05-26 12:00:58 continuing now 2015-05-26 12:01:55 and it needs linux-headers ? 2015-05-26 12:04:40 yes 2015-05-26 12:04:59 ok, i'm pushing update in a second 2015-05-26 12:10:22 wasn't kaniini working on glusterfs for v2.7 ? 2015-05-26 12:13:59 either of one "glusterfs, ceph, lustre" + mariadb-10x + python 3x would make AL v3.2 superbase for all migrations and new things 2015-05-26 12:17:21 rtorrent now ok for x86 / 2015-05-26 12:18:23 ? 2015-05-26 12:19:12 vkrishn: i think rtorrent at least builds 2015-05-26 12:19:14 for x86 2015-05-26 12:19:30 ok, thanks 2015-05-26 12:19:44 AL v3.2 overall looks cool 2015-05-26 12:22:33 those does not yet build on arm: http://sprunge.us/GNEP 2015-05-26 12:22:59 libreoffice actually builds, but hangs on unittest 2015-05-26 12:23:07 which is slightly worrying 2015-05-26 12:23:25 i suppose video-vmware makes no sense on arm? 2015-05-26 12:23:47 you could probably use qmeu with -vga vmware 2015-05-26 12:26:07 lets see if it is able to upload the release... 2015-05-26 12:26:10 arm that is 2015-05-26 12:26:26 http://nl.alpinelinux.org/alpine/v3.2/releases/armhf/ 2015-05-26 12:30:27 <[xming]> There might be some other things using video-vmware, IIRC qemu/kvm could use that 2015-05-26 12:30:45 yeah 2015-05-26 12:30:55 thats what i'm trying to say 2015-05-26 12:31:41 <[xming]> so it's not totally non sense on non-x86(_64) 2015-05-26 12:31:55 <[xming]> man too much negatives 2015-05-26 12:32:14 <[xming]> but my sentence is correct :p 2015-05-26 12:32:56 i understand what you say :) 2015-05-26 12:36:26 http://nl.alpinelinux.org/alpine/v3.2/releases/armhf/ 2015-05-26 12:36:28 \o/ 2015-05-26 12:37:02 \o/ 2015-05-26 12:37:30 [xming], care to re-paste the alpine-v86 source link ? 2015-05-26 12:38:34 <[xming]> fcolista: https://github.com/xming/alpine-v86/tree/gh-pages/demo2 2015-05-26 12:38:45 thx [xming] 2015-05-26 12:38:54 <[xming]> will update afeter 3.2 2015-05-26 12:39:02 <[xming]> and lot's of other things too 2015-05-26 12:39:35 xf86-video-vmware fails on arm du to missing outl() 2015-05-26 12:39:41 <[xming]> since it's all binary, I will probably just delete that branch and re-create it 2015-05-26 12:39:46 sounds very x86'ish 2015-05-26 12:40:46 ncopa, how would be possible with alpine to have an unattended installation? 2015-05-26 12:41:14 a kickstart file 2015-05-26 12:41:53 well, the equivalent. Umh...it might be an apkovl. 2015-05-26 12:42:04 ACTION is thinking louder. 2015-05-26 12:42:47 no, apkovl not. 2015-05-26 12:42:48 fcolista: we have some issue/ti ket on that 2015-05-26 12:42:50 yes 2015-05-26 12:42:55 we want something nice for that 2015-05-26 12:43:06 dunno how to implement it properly 2015-05-26 12:43:48 just thinking about a qemu img that update himself when a new stable iso get released 2015-05-26 12:44:14 or you boot directly the iso (easier) 2015-05-26 12:44:40 or you create a custom img, with packages pre-configured. 2015-05-26 12:44:41 nah, I like rtorrent, I can active throttle speed 2015-05-26 12:45:14 fabled: xf86-video-qxl build error on arm: 2015-05-26 12:45:16 qxl.h:635:5: error: implicit declaration of function 'outb' [-Werror=implicit-function-declaration] 2015-05-26 12:45:16 outb(qxl->io_base + port, val); 2015-05-26 12:45:16 ^ 2015-05-26 12:45:22 <[xming]> fcolista: stock kernel is stlow as hell on the js emu 2015-05-26 12:45:22 outb() and outl() 2015-05-26 12:45:30 vkrishn, do you still need anytun? 2015-05-26 12:45:41 fabled: those are x86/x86_64 specifics? 2015-05-26 12:46:03 we could probably make a correspoding func for arm, but will it work on that platform? 2015-05-26 12:46:04 <[xming]> outb() outl() sound really x86 to me 2015-05-26 12:46:07 do we have anything similar ? otherwise it seems nice 2015-05-26 12:46:16 vkrishn: iirc you asked about it few days ago 2015-05-26 12:47:29 <[xming]> hmm ARM has that? http://marc.info/?l=git-commits-head&m=129651481103139 2015-05-26 12:47:59 I may not have immidiate use, or till I understand it fully, i got interested coz I did not see a tool like it before 2015-05-26 12:48:52 Its probably a recent RFC implementation 2015-05-26 12:49:14 even JSON is making lots of RFC these days 2015-05-26 12:49:33 i've an anytun package on testing, not yet pushed. 2015-05-26 12:49:36 gotta test it 2015-05-26 12:49:47 outb() and outl() are traditionally x86 specific 2015-05-26 12:50:02 some platform provides wrappers for them 2015-05-26 12:50:03 should be ok to push to testing 2015-05-26 12:50:26 it'd probably be easier to just not build it on arm 2015-05-26 12:51:31 can we tag AL v3.2 as HTTP/2 ready ? 2015-05-26 12:57:59 nice, thanks 2015-05-26 13:00:20 ../deps/v8/src/base/atomicops_internals_arm_gcc.h:258:4: error: #error "Your CPU's ARM architecture is not supported yet" 2015-05-26 13:00:20 # error "Your CPU's ARM architecture is not supported yet" 2015-05-26 13:00:20 ^ 2015-05-26 13:00:24 thats nodejs 2015-05-26 13:00:41 ok 2015-05-26 13:00:46 i'll go look for coffe 2015-05-26 13:00:54 and then I'll spend rest of day with git branch 2015-05-26 13:01:08 please dont push anything til v3.2.0 is out 2015-05-26 13:04:58 http://wwwtest.alpinelinux.org/posts/Alpine-3.2.0-released.html 2015-05-26 13:05:40 no mention of nfs ? 2015-05-26 13:06:50 what should we say about nfs? 2015-05-26 13:06:54 "it works"? 2015-05-26 13:13:05 please dont push anything yet... 2015-05-26 13:16:50 \o/ 2015-05-26 13:19:27 maybe a group "Fixed since previous major release" 2015-05-26 13:19:52 vkrishn: thats not a bad idea 2015-05-26 13:19:57 ok 2015-05-26 13:20:07 kinda negative 2015-05-26 13:22:45 kodi - (previously known as xbmc) 2015-05-26 13:23:11 mariadb replaces mysql 2015-05-26 13:24:11 the point with this list 2015-05-26 13:24:12 is 2015-05-26 13:24:18 if you run xbmc already 2015-05-26 13:24:22 and want to upgrade 2015-05-26 13:24:33 then you need apk del xbmc and apk add kodi 2015-05-26 13:24:59 kodi 14.2 (previously known as xbmc) 2015-05-26 13:25:25 does "kodi 14.2 (previously known as xbmc)" make it clear that you need to do things if you currently run xbmc? 2015-05-26 13:26:40 better? http://sprunge.us/MPfX 2015-05-26 13:27:22 kodi (previously known as xbmc) 14.2 ;) 2015-05-26 13:29:14 http://sprunge.us/DYUT 2015-05-26 13:29:18 or a wiki page describing migration from xbmc to kodi 2015-05-26 13:29:27 we should have that too 2015-05-26 13:29:43 kodi 14.2 (previously known as xbmc), see 2015-05-26 13:30:40 i cannot post a wiki url that does not exisst 2015-05-26 13:30:49 i need publish this page in 15 mins or so... 2015-05-26 13:31:06 so unless someone write that article *now* i cannot do that 2015-05-26 13:33:13 15mins, how many cups of coffee you had ;) 2015-05-26 13:33:32 the release is uploaded 2015-05-26 13:33:36 the branching is done 2015-05-26 13:33:42 i'm cleaning up bugs.a.o 2015-05-26 13:33:52 :) 2015-05-26 13:34:05 so unless i get delayed, that is 15 mins... 2015-05-26 13:37:28 does this still exsits http://wiki.alpinelinux.org/wiki/Apk.lua ? 2015-05-26 13:37:31 I can't find in acf-core/lib 2015-05-26 13:40:21 hum... ok. git log answer: "Simplified apk.lua by moving functions to acf-apk-tools version 0.3.0 2015-05-26 13:40:21 " 2015-05-26 13:50:12 that module has a conflict with apk.so from lua-apk 2015-05-26 13:50:19 from apk-tools 2015-05-26 14:08:25 ewe! has algibot stop loging alpine-linux irc ? 2015-05-26 14:09:05 algitbot, wake up _o/ 2015-05-26 14:10:06 sleeping since "2015-05-18 21:58:38" 2015-05-26 14:43:04 kunkku: something broke: http://bld2.alpinelinux.org/buildlogs/build-edge-x86_64/main/apache2/apache2-2.4.12-r2.log 2015-05-26 14:44:13 ok? http://wwwtest.alpinelinux.org/posts/Alpine-3.2.0-released.html 2015-05-26 14:44:40 any volunteer to fix up http://wiki.alpinelinux.org/wiki/Alpine_Linux:Releases ? 2015-05-26 14:45:00 v3.1.3, 3.1.4 and 3.2.0 is missing 2015-05-26 14:52:13 "checking whether to enable mod_xml2enc... no (disabled)" 2015-05-26 14:52:30 "checking whether to enable mod_proxy_html... no (disabled)" 2015-05-26 14:59:37 ncopa: updated wiki, u should check the end of support stuff though 2015-05-26 14:59:54 pnutzh4x0r: thanks! 2015-05-26 15:01:09 hmm 2015-05-26 15:01:12 my changes disappeared 2015-05-26 15:01:15 lol 2015-05-26 15:01:18 i ll change them again 2015-05-26 15:04:03 need to refresh the edit page in subsequent edits, make sure you save the edits text locally 2015-05-26 15:04:31 generally I do F5 2015-05-26 15:04:35 yeah 2015-05-26 15:04:37 that's what happened 2015-05-26 15:04:42 it's fixed now 2015-05-26 15:06:05 i have heard about that issue 2015-05-26 15:06:12 it shoudl be reported on bugs.a.o 2015-05-26 15:06:15 ok i gotta go 2015-05-26 15:06:22 will check in later to sign the release 2015-05-26 15:06:33 thanks all for the big job on 3.2 release 2015-05-26 15:06:47 i couldn't done it without you 2015-05-26 15:07:38 I'm wondergin for the next release to test aircrack-ng and move it from testing to main 2015-05-26 15:07:55 setup-box has already pentest profile 2015-05-26 15:08:16 Also porting openvas would be nice, but it's a huge task 2015-05-26 15:08:21 gotta go 2015-05-26 15:08:22 c u 2015-05-26 15:09:04 yay 3.2 2015-05-26 15:37:31 <[xming]> *3* *dot* *2* 2015-05-26 15:37:58 <[xming]> horray \o/ 2015-05-26 15:38:04 WEE 2015-05-26 15:43:28 upgrading alpine-xen to 3.1 to 3.2 broke xen for me :( 2015-05-26 15:44:02 ScrumpyJack: did you reboot? it worked for me 2015-05-26 15:44:22 yes, rebooted. i get permission problems 2015-05-26 15:44:28 libxl: error: libxl.c:5044:libxl_get_physinfo: getting physinfo: Permission denied 2015-05-26 15:44:38 libxl_physinfo failed. 2015-05-26 15:44:44 libxl: error: libxl.c:5534:libxl_get_scheduler: getting domain info list: Permission denied 2015-05-26 15:46:31 uptime 14 min 2015-05-26 15:47:03 when did that errmsg show up? 2015-05-26 15:47:12 i see it's happend before 2015-05-26 15:47:23 runnig any xl command 2015-05-26 15:47:35 is all the needed xen services running 2015-05-26 15:47:43 openrc --sys 2015-05-26 15:47:50 does it say DOM0? 2015-05-26 15:48:11 :( no DOM0 running 2015-05-26 15:48:14 hmm 2015-05-26 15:49:06 altough its in process table 2015-05-26 15:49:10 usr/lib/xen/bin/qemu-system-i386 -xen-domid 0 -xen-attach 2015-05-26 15:49:45 ScrumpyJack: missmatch between xen kernel version and tools? 2015-05-26 15:50:04 that's what google says, but i can't see that here 2015-05-26 15:50:19 ScrumpyJack: xl info works? 2015-05-26 15:51:27 it fails between machine and xen_major 2015-05-26 15:52:15 it's still giving me 4.4.1 :( 2015-05-26 15:52:24 3.14.26-0-grsec 2015-05-26 15:53:59 by the way why is my DOM0 running under qemu-system-i386 and not x86_64 ? 2015-05-26 15:54:00 ScrumpyJack: I bet you either need to update the xen kernel or the xen tools. Do you have /proc/xen populated? 2015-05-26 15:54:27 yes 2015-05-26 15:55:10 ScrumpyJack: Dom0 is not running under Qemu, this is just there to perform local disk attach. Arch doesn't matter because we don't use the cpu emulation from qemu 2015-05-26 15:55:47 xen just needs the device emulation which is exactly the same 2015-05-26 15:55:52 nice to know, thanks for the explaination 2015-05-26 15:56:38 xen and xen libs come up as 4.5 2015-05-26 15:58:03 ScrumpyJack: need to update syslinux/pxelinux... whatever to pick the right xen kernel? 2015-05-26 15:59:37 syslinux is the default 2015-05-26 15:59:48 have names changed? append /boot/xen.gz 2015-05-26 16:00:32 hmm, no xen-4.5 2015-05-26 16:02:31 at the end of the upgrade, i got mount: mounting /dev/usbdisk on /media/usb failed: Resource busy 2015-05-26 16:02:55 so i guess the xen kernel wasn't writen to the usb before I rebooted? 2015-05-26 16:04:38 mmm maybe? no idea really, also I'm not sure of the layout in alpine, but Xen usually places some links so xen-4.5.gz -> xen.org (or the other way around) you should check those are correct regarding the config you have in syslinux 2015-05-26 16:05:44 it's xen.gz 2015-05-26 16:10:49 yik yik yik, python has slipped in again as a dependancy :( 2015-05-26 16:12:10 lbu vommit doesn't work ;) 2015-05-26 16:13:42 ok, i'll try tomorrow. thanks for the help so far 2015-05-26 16:26:41 congratulation on the 3.2.0. will there be vanilla iso from final version to? i dont see it on dl page http://dl-4.alpinelinux.org/alpine/v3.2/releases/x86_64/ or on dl-1 2015-05-26 16:39:45 who is the one that add Alpine Linux releases to distrowatch ? 2015-05-26 16:52:53 2015-05-26 06:54:50 http://alpinelinux.org/downloads/ 2015-05-26 16:52:53 2015-05-26 06:54:52 pick vanilla 2015-05-26 16:53:10 seems it was not build and thus missing on download page 2015-05-27 06:03:51 morning 2015-05-27 06:04:12 crow: yes, the release building broke 2015-05-27 06:04:21 i fixed the iso last night 2015-05-27 06:04:29 the scripts are not very robust 2015-05-27 06:21:52 ncopa, could you change "Raspberry Pi" heading to "ARM" 2015-05-27 06:22:11 since the uboot tarball is not for rpi, but for wand+others 2015-05-27 06:25:49 ok 2015-05-27 06:25:59 i though i didt but apparently i didnt push it 2015-05-27 06:26:37 fabled: something like this? http://sprunge.us/cBbL 2015-05-27 06:26:55 maybe i should mention wandboard there 2015-05-27 06:27:17 dunno about 'universal' at this point 2015-05-27 06:27:33 thats what uboot page said... 2015-05-27 06:27:45 "General ARM"? 2015-05-27 06:28:03 or just ARM? 2015-05-27 06:28:32 mmm.. "Generic ARM" or "ARM" 2015-05-27 06:29:29 oh 2015-05-27 06:29:42 and you have typo 2015-05-27 06:29:53 it's "Raspberry". you are missing the 'p' in it 2015-05-27 06:30:45 "Generic ARM" maybe 2015-05-27 06:30:58 http://sprunge.us/jeDG 2015-05-27 06:31:16 sure 2015-05-27 06:31:36 mm 2015-05-27 06:31:41 maybe in rpi build say 2015-05-27 06:31:53 (all models including v2) 2015-05-27 06:31:57 or similar 2015-05-27 06:35:37 or just (version 1 and 2) 2015-05-27 06:49:11 http://sprunge.us/KLhh 2015-05-27 06:51:34 good enough for now. thanks! 2015-05-27 07:36:15 hum 2015-05-27 07:36:30 seem like recent mesa broke my eee pc's xorg 2015-05-27 07:41:32 morning 2015-05-27 07:42:04 i'm going to reinstall alpine-xen instead of upgrading (tests failed last night for me) 2015-05-27 07:42:33 all my alpine=xen hypervisors are on USB sticks, so it will be easy enough to do 2015-05-27 07:45:24 ncopa thanks for fixing builders, i see now iso, but the size of grsec and vanilla is almost 210MB, is that realy so much? 2015-05-27 07:45:36 note i did not tested isto, just saw your msg 2015-05-27 07:46:33 grsec and vanilla? 2015-05-27 07:46:38 http://alpinelinux.org/downloads/ 2015-05-27 07:46:48 normal iso is 298MB 2015-05-27 07:47:03 vanilla iso is 86MB 2015-05-27 07:48:06 so it is ok that diff :) never look at that trought 2015-05-27 08:12:48 love setup-bootable :) 2015-05-27 08:16:23 panic when booting alpine-xen usb using qemu :( 2015-05-27 08:17:15 should i even be able to boot alpine-xen with qemu? 2015-05-27 08:18:36 ScrumpyJack to update usb stick system? yes its nice tool 2015-05-27 08:19:50 can you just copy your apkvol file back to the usb stick? 2015-05-27 08:21:14 to new created usb stick? yes that should work 2015-05-27 08:21:39 well, im upgrading 3.1 to 3.2 where apk upgrade failed 2015-05-27 08:24:15 was it 'sys' install or tmpfs? 2015-05-27 08:45:11 tempfs 2015-05-27 08:45:23 onto usb sticks 2015-05-27 08:46:25 anyone successfully booted alpine-xen using qemu? 2015-05-27 08:47:43 yay! i have xen-4.5 on /boot now 2015-05-27 08:57:26 ScrumpyJack then just copy apkvol back, you may need to install some apps trough, check what is inside that apkvol 2015-05-27 09:35:53 crow: yeah, i needed to reinstall some apps 2015-05-27 09:36:11 evrey boot complains that /media/usb is busy thought :( 2015-05-27 09:42:47 http://sprunge.us/ZDYP 2015-05-27 09:43:02 lbu commit doesn't work either :( 2015-05-27 09:43:12 probably or the same reason 2015-05-27 09:50:29 and xen has a new dependancy in python, which won't install at boot 2015-05-27 09:51:36 new dep? 2015-05-27 09:51:39 ugh 2015-05-27 09:51:42 for what? 2015-05-27 09:57:41 ncopa: xen depends on python again i beleive 2015-05-27 09:58:06 ScrumpyJack: which part of xen 2015-05-27 09:58:43 i don't know yet. when i boot i get errors relating to xen 2015-05-27 09:59:07 i'm not in front of the host at the momoent 2015-05-27 09:59:52 xen01:~# apk info xen -R 2015-05-27 09:59:52 xen-4.5.0-r0 depends on: 2015-05-27 09:59:53 ... 2015-05-27 09:59:58 so:libpython2.7.so.1.0 2015-05-27 10:00:06 yeah something pulls in python 2015-05-27 10:16:23 ncopa: boo :( 2015-05-27 10:17:06 so, in a usb data install, why does apk complain that /media/usb is busy 2015-05-27 10:17:20 does it try to unmount it if it's mounted? 2015-05-27 10:17:42 or does it try to mount it and it's already mounted? 2015-05-27 10:18:31 i might just erase the usb stick and to setup-bootable on a black stick. my previous setup-bootable was on top of a 3.1 install 2015-05-27 10:23:59 hmm, something wrong with the usb stick 2015-05-27 11:43:09 setup-bootable on clean usb stick, then copy over apkvol.tar.gz 2015-05-27 11:44:07 anything else i should do to get this to work? apk complains that /media/usb is busy 2015-05-27 12:31:04 ScrumpyJack you did setup cache? 2015-05-27 12:44:50 [xming], you around? 2015-05-27 12:47:44 crow: yes, but only after doing an apk update and apk upgrade so that apk would get what was in apk/world 2015-05-27 12:47:58 i'll try again but doing setup-cache first 2015-05-27 12:51:47 try that first. i think apk update would not be able to save anything 2015-05-27 12:51:53 but i am not expert in this :) 2015-05-27 13:21:28 ncopa, did you see the mail on musl; i thought we had gnu hashing on in gcc, but apparently not 2015-05-27 13:21:46 seems gcc default is sysv-only 2015-05-27 14:15:14 fabled: i didnt see it 2015-05-27 14:19:42 <[xming]> fcolista: what's up? 2015-05-27 14:20:00 did you try alpine 3.2 with v86 ? 2015-05-27 14:20:42 <[xming]> not yet 2015-05-27 14:21:04 ok 2015-05-27 14:21:10 i'm trying but it stucks 2015-05-27 14:21:18 <[xming]> being doing other stuff, like serial console, 9p fs etc 2015-05-27 14:21:53 it stucks with "MPTABLE: no processors registered" 2015-05-27 14:22:02 "BIOS bug, MP table errors detected" 2015-05-27 14:22:09 <[xming]> alpine default kernel doesn't boot (on my PC) or boot very slow (15 min on ncopa's) 2015-05-27 14:22:24 gotta try with vanilla then 2015-05-27 14:22:26 <[xming]> oh that's new :D 2015-05-27 14:23:10 <[xming]> that's one of the reasons of the custom kernel 2015-05-28 07:01:01 ncopa, remove lazy from LDFLAGS for mesa (and all others too) 2015-05-28 07:01:34 musl uses 'now' anyway; and using lazy just prevents full relro 2015-05-28 07:01:38 that another issue 2015-05-28 07:01:50 ok 2015-05-28 07:01:51 yes, just noticed it from the commit 2015-05-28 07:02:27 seems we have it in mesa, and bunch of xf86-* 2015-05-28 07:02:51 $ grep lazy */*/APKBUILD|wc 2015-05-28 07:02:51 36 144 2515 2015-05-28 07:04:21 we should work for gcc-5.1 too 2015-05-28 07:07:53 fabled: to enable hash-style=gnu by default 2015-05-28 07:08:03 do we need patch/rebuild gcc or bintutils 2015-05-28 07:08:04 for both 2015-05-28 07:08:10 rebuild gcc with new configure option 2015-05-28 07:08:50 hmm... i wonder if that could be done in binutils 2015-05-28 07:09:01 --with-linker-hash-style=both ? 2015-05-28 07:09:51 i wondered the same, if its possible to do it with binutils 2015-05-28 07:11:35 it should be 2015-05-28 07:11:41 it's ld option which is from binutils 2015-05-28 07:12:18 gcc calls ld and maybe add those options? 2015-05-28 07:13:25 gentoo patches ld 2015-05-28 07:15:07 i cannot find patch? https://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/sys-devel/binutils/ 2015-05-28 07:16:45 it comes from the toolchain eclass 2015-05-28 07:16:46 http://dev.gentoo.org/~vapier/dist/binutils-2.25-patches-1.2.tar.xz 2015-05-28 07:19:12 debian patches too 2015-05-28 07:22:30 debian patch is http://sprunge.us/ZLJP 2015-05-28 07:23:07 do we need rebuild gcc with 'both'? 2015-05-28 07:24:02 seems gentoo changed defeault to "gnu" 2015-05-28 07:24:10 debian has "both" for some builds, and "gnu" for some 2015-05-28 07:26:08 i think gcc rebuild is not needed 2015-05-28 07:26:47 i can test 2015-05-28 07:27:25 should we do 'gnu' or 'both' ? 2015-05-28 07:27:46 i think dalias said if we dont include sysv its not an elf 2015-05-28 07:29:11 In the sense of "known issues" I would say no, but I would still 2015-05-28 07:29:11 strongly discourage this. A valid dynamic ELF file requries the sysv 2015-05-28 07:29:11 (standard ELF) hash table; the original reason musl added support for 2015-05-28 07:29:11 GNU hash was for ABI-compat to be able to run binaries from distros 2015-05-28 07:29:11 that were producing "invalid" ELF files. 2015-05-28 07:29:56 right 2015-05-28 07:30:03 yes, some newer distros use 'gnu' 2015-05-28 07:30:11 gentoo and recent debian/ubuntu to new a few 2015-05-28 07:30:16 apparently it saves few kBs 2015-05-28 07:30:18 per binary 2015-05-28 07:30:27 hum 2015-05-28 07:30:30 thats tempting 2015-05-28 07:31:14 http://www.openwall.com/lists/musl/2015/05/27/18 2015-05-28 07:31:25 yes, i got the mail 2015-05-28 07:31:56 any tips on getting supend to work on a laptop with AL? 2015-05-28 07:33:04 ScrumpyJack: apk add pm-suspend? 2015-05-28 07:33:21 i'll give that a shot 2015-05-28 07:33:53 http://wiki.alpinelinux.org/wiki/Suspend_on_LID_close 2015-05-28 07:34:31 pm-utils 2015-05-28 07:35:10 ncopa, just tested. patching binutils is enough. seems gcc will use the default if the --with-default-hashstyle is not set 2015-05-28 07:36:16 ok good 2015-05-28 07:37:09 now to just decide if we want "both" or "gnu" 2015-05-28 07:37:17 debian seems to have clean patch for both 2015-05-28 07:37:34 that is it has two separate patches, one for "both" and one for "gnu" 2015-05-28 07:38:56 it suspends :) but doesn't wake :( I'll try again later 2015-05-28 07:44:06 <[xming]> gcc-5.1 and enable lto builds? :D 2015-05-28 07:46:21 <[xming]> grrr, now I remember why I hated SGML 2015-05-28 07:47:52 <[xming]> I can never find the right dtd, xslt, blah blah 2015-05-28 07:54:49 yes, gcc-5.1 would be also good candidate for edge now 2015-05-28 07:54:59 someone said having patches for that 2015-05-28 07:57:04 i wonder if i should recreate edge from v3.2 2015-05-28 07:57:11 to have new fresh packages for all 2015-05-28 07:57:30 too late now, they've already diverged 2015-05-28 07:57:36 can do that for v3.3 2015-05-28 07:57:36 do edge rebuild rather if needed 2015-05-28 07:57:55 i'm considering hash 'gnu' for edge, and potentiall hash 'both' for 3.2-stable 2015-05-28 07:58:42 it seems the sysv hash is bloated (and slow) partly due to it containing lot of symbols in the hash that are not needed there 2015-05-28 07:58:52 gnu hash other improvements too, though 2015-05-28 07:59:12 (such as better hash function, and bloom filter) 2015-05-28 08:00:31 i think I'll change the x86 builer to i586 2015-05-28 08:00:47 change CHOST=i586... 2015-05-28 08:00:53 that'd be valid reason for rebuild 2015-05-28 08:00:57 along with hash-style 2015-05-28 08:01:11 and then rebuild gcc 2015-05-28 08:01:13 http://sprunge.us/FZQi 2015-05-28 08:01:18 with the above 2015-05-28 08:01:33 then we fix binutils to do hash-style=gnu 2015-05-28 08:01:48 i have a few things more i'd like to do before full rebuild 2015-05-28 08:02:02 i586 or pentium-mmx ? 2015-05-28 08:02:10 i think i586 2015-05-28 08:02:11 <[xming]> for 32. or edge? 2015-05-28 08:02:18 for edge, x86 2015-05-28 08:02:23 i suppose i586 2015-05-28 08:02:34 i think i586 was the conclusion from ml 2015-05-28 08:02:49 [xming], yes, x86 only. x86_64 is newer than any of those ancient things... 2015-05-28 08:02:51 yes 2015-05-28 08:02:55 i586 would probably be it 2015-05-28 08:03:08 abuild's way of creating deps for -dev 2015-05-28 08:03:18 i'd like to fix so it only pulls in whats needed 2015-05-28 08:03:22 ok 2015-05-28 08:03:25 <[xming]> ACTION votes for i586 2015-05-28 08:03:33 which means we would need rebuild all -dev packages 2015-05-28 08:03:41 which means practically a world rebuild 2015-05-28 08:04:11 but that does not stop us from doing i586 now 2015-05-28 08:04:24 or hash change 2015-05-28 08:04:29 yes 2015-05-28 08:04:33 ok 2015-05-28 08:04:52 ncopa, you remember who worked on gcc 5.1? 2015-05-28 08:05:02 no, i dont, sorry 2015-05-28 08:06:01 <[xming]> switching to gcc-5.1 is already a rebuild of all c++ pkgs 2015-05-28 09:04:14 laptop suspends, but i still get a suspend failed messages 2015-05-28 09:08:29 apk upgrade keeps giving me: 2015-05-28 09:08:31 mount: mounting /dev/sdg1 on /media/usb failed: Resource busy 2015-05-28 09:13:45 ScrumpyJack: and you did setup-cache before? 2015-05-28 09:20:31 i think it tries to remount rw before writing to apk cache, and then again remount to read-only 2015-05-28 09:21:46 i'm gonna push i586 gcc now 2015-05-28 09:22:20 i also changed the /etc/abuild.conf to use i586-alpine-linux-musl instead of i486-... 2015-05-28 09:36:11 ScrumpyJack: do you maybe have that usb as apk repository ? 2015-05-28 09:55:23 crow_: yes i did 2015-05-28 09:55:51 nope, the only repo is liskamm 2015-05-28 09:58:35 it only seems to do that once when freshly booted, then never again 2015-05-28 09:58:49 this is new to me: Continuing the upgrade transaction with new apk-tools: 2015-05-28 13:06:48 i have a techincal question for how to seralize multiple runs of rsync for alpine mirrors 2015-05-28 13:07:06 currently mirrors normally run from cron, every hour or so 2015-05-28 13:07:18 i want have some mirrors to respond on MQTT 2015-05-28 13:07:25 problem is 2015-05-28 13:07:41 how do we deal with builer x86_64 is done first, uplaod 2015-05-28 13:07:47 and trigger mqtt 2015-05-28 13:07:50 rsync starts 2015-05-28 13:08:06 second after the x86 builder is done 2015-05-28 13:08:24 and trigger new sync, before first run is complete 2015-05-28 13:08:35 we dont want 2 parallel rsync jobs 2015-05-28 13:08:43 so they need to be queued some how 2015-05-28 13:08:57 how to do that? 2015-05-28 13:09:39 the mqtt trigger will simply fork/exec the rsyncscript and let kernel auto-reap the child 2015-05-28 13:09:59 so it will spawn the second run while first is active 2015-05-28 13:10:22 we currently have a flock thingy to make the script exit if lock is there already 2015-05-28 13:10:43 however, we want run the rsync again, directly after the first run is done 2015-05-28 13:12:54 i can do flock without -n so it waits 2015-05-28 13:13:07 however, what happens if the lockfile gets stale? 2015-05-28 13:13:12 then will it never recover 2015-05-28 13:13:37 so we need flock with a wait timer, which is not implemented in busybox flock 2015-05-28 13:13:53 so i am thinking, maybe there are better ways? 2015-05-28 13:14:13 i could for example make mqtt-exec serialize the runs 2015-05-28 13:14:33 that is, make it wait til child is done before forking the next child 2015-05-28 13:14:56 and queue the incoming new mqtt events 2015-05-28 13:16:35 other option is make a script with lua or so with librsync, and have that queue the events 2015-05-28 13:16:58 i also consider pass as subdir with the event, so the rsync run does not need sync it all 2015-05-28 13:17:08 only the relevant arch/subdir 2015-05-28 13:17:22 in that case, it does not matter if x86 and x86_64 run in parallel 2015-05-28 13:17:52 but it would be kinda nice to not run them in parallel anyway 2015-05-28 13:18:57 any thoughts of pkging, transmission-web 2015-05-28 13:19:36 this is the mirror script: https://dpaste.de/EWfS 2015-05-28 13:20:14 why not parallel rsync jobs ? 2015-05-28 13:20:51 they could be running for x86 or/and x86_64 2015-05-28 13:21:06 yes that would work 2015-05-28 13:21:13 but i'd prefer not 2015-05-28 13:21:28 to be nice with the link 2015-05-28 13:22:07 if the rsync server has a bwlimit, then we want respect that and dont run 2 parallel runs 2015-05-28 13:22:20 also, we cannot run 2 parallel rsync of same dirs 2015-05-28 13:22:24 then run rsync with bwlimit 2015-05-28 13:22:46 ncopa: different approach: send a notification to mqtt when all builders are done 2015-05-28 13:22:56 yes that is why the mqtt msg will say which to trigger 2015-05-28 13:23:18 barthalion: i dont like that 2015-05-28 13:23:31 then would you wait for arm a week... 2015-05-28 13:23:36 opr a day or two 2015-05-28 13:23:40 or* 2015-05-28 13:23:49 we probably want the x86_64 before that 2015-05-28 13:23:57 or ignore ARM 2015-05-28 13:24:20 it can be synced when there is an opportunity 2015-05-28 13:24:30 i think i have a better solution 2015-05-28 13:24:34 so it would be always slightly behind x86 and x86_64 2015-05-28 13:24:39 so it sucks 2015-05-28 13:24:41 (y) 2015-05-28 13:24:57 https://dpaste.de/EWfS#L8,25 2015-05-28 13:25:13 this script currently waits for flock forever 2015-05-28 13:25:24 so if might hang 2015-05-28 13:25:36 and if the lockfile get stuck it will not recover 2015-05-28 13:25:43 so we want flock -n i think 2015-05-28 13:25:46 just exit 2015-05-28 13:26:33 then we have another tmp file: /tmp/sync-again 2015-05-28 13:26:45 before doing the flock 2015-05-28 13:26:57 do: touch /tmp/sync-again 2015-05-28 13:27:00 I wonder if we should bother with this at all 2015-05-28 13:27:28 let mirrors use cron if they want, rest can subscribe to "sup new packages are ready" mqtt 2015-05-28 13:27:49 then while [ -e /tmp/sync-again ]; do rm /tmp/sync-again; rsync .... ; done 2015-05-28 13:27:51 and sync immediately or after more than $x packages were uploaded 2015-05-28 13:28:30 i am thinking how to do the sync immediately thingy properly 2015-05-28 13:28:32 this is only for nl.a.o or all mirrors ? 2015-05-28 13:28:49 what i want is tier1 mirrors 2015-05-28 13:28:59 who sync asap from master rsync 2015-05-28 13:29:09 ok tier1 group of mirrors 2015-05-28 13:29:18 then we could do dns round robin on those 2015-05-28 13:29:39 in arch we have one or two tier0 mirrors 2015-05-28 13:29:56 tier1 mirrors are allowed to sync from those, and they are also supposed to let tier2 sync from them 2015-05-28 13:30:24 but it doesn' 2015-05-28 13:30:26 argh 2015-05-28 13:30:29 i consider tier0 the machine you upload to 2015-05-28 13:30:33 yes 2015-05-28 13:30:39 but it doesn't solve mqtt problem 2015-05-28 13:30:43 so you upload to 2 different hosts? 2015-05-28 13:30:55 hm, I think there is just one tier0 2015-05-28 13:31:00 right 2015-05-28 13:31:20 I don't know many details about Arch infra, so I can be wrong 2015-05-28 13:31:28 but that makes sense 2015-05-28 13:31:37 thats what i want too 2015-05-28 13:31:50 the issue with this is that mirrors are getting out of sync 2015-05-28 13:32:04 let's say that tier1 syncs 15 minutes after tier0 got a package 2015-05-28 13:32:21 then tier2 syncs 15 minutes after 15 minutes 2015-05-28 13:32:31 so it's 30 minutes late 2015-05-28 13:32:39 you cannot prevent the go out of sync 2015-05-28 13:32:47 but we want that time minimal 2015-05-28 13:33:05 30 mins is unacceptable 2015-05-28 13:33:07 yeah, mqtt can solve it 2015-05-28 13:33:24 mqtt is just the mecahnism to tell: new packages. please sync now 2015-05-28 13:33:30 I know, there are a lot of tier2 mirrors like that in Arch :( 2015-05-28 13:33:47 I constantly change mirrorlist on my friends' machines due to that 2015-05-28 13:34:05 because they use outdated ones, and I remember 2-3 that are fine 2015-05-28 13:34:06 problem is, how to handle a sync request burst 2015-05-28 13:35:00 i think the while [ -e /tmp/sync-again ]; do rm /tmp/sync-again; rsync ...; done solves it 2015-05-28 13:35:24 and create that /tmp/sync-again before flock 2015-05-28 13:44:41 this should work: https://dpaste.de/hbY7 2015-05-28 13:52:11 nice. looks like it works 2015-05-28 13:52:39 its still slightly non-optimal because it will scan the entire tree 2015-05-28 13:52:45 instead of only relevant arch 2015-05-28 13:52:53 but thats ok i suppose 2015-05-28 13:54:44 there is also https://github.com/fagga/transmission-remote-cli 2015-05-28 13:56:42 vkrishn: transmission-daemon seems to ship the web interface too 2015-05-28 13:59:12 fabled: nice! 2015-05-28 14:01:42 ah, nice thanks 2015-05-28 14:33:02 I'm going to try alpine 3.2 with ebaglebone. 2015-05-28 14:33:06 *beaglebone. 2015-05-28 14:33:45 Is this iso ok? http://nl.alpinelinux.org/alpine/v3.2/releases/armhf/alpine-uboot-3.2.0-armhf.tar.gz 2015-05-28 14:34:51 http://hastebin.com/hobewaselo.lua 2015-05-28 14:34:57 any idea? ^^ 2015-05-28 14:36:07 alacerda, it's due to FORTIFY 2015-05-28 14:36:32 something like that might work. 2015-05-28 14:36:33 .http://git.alpinelinux.org/cgit/aports/plain/main/fortify-headers/0001-fix-realpath-when-stdlib.h-is-included-before-limits.patch 2015-05-28 14:37:11 ncopa already fixed that issue with minidlna 2015-05-28 14:38:44 what is the simplest way to create (or request) new packages? looking for ruby-mechanize 2015-05-28 14:39:10 shreeve: file a bug on bugs.alpinelinux.org and mark it as request 2015-05-28 14:39:14 is there a docker container or virtualbox image with a dev environment already set up? 2015-05-28 14:39:30 we don't provide docker images 2015-05-28 14:39:40 ncopa: hello! sorry I just emailed you last night... i'll do as you suggest 2015-05-28 14:39:47 you can use gliderlab's image and set up dev environment following wiki 2015-05-28 14:39:52 shreeve: i just read the email 2015-05-28 14:40:10 we have some tool for ruby gems packaging iirc 2015-05-28 14:40:24 ncopa: how hard is that request for ruby-mechanize? I'm happy to help in any way, just don't have a dev env setup yet 2015-05-28 14:40:45 barthalion: do you have a link or info to that? 2015-05-28 14:41:08 my brain is a pinata, but I can tell that kunkku wrote it 2015-05-28 14:41:57 i'll searfch 2015-05-28 14:43:34 barthalion: this page has a reference to that (possibly)... http://dev.alpinelinux.org/irclogs/%23alpine-devel-2015-03.log 2015-05-28 14:44:22 barthalion: this type of thing? -> https://github.com/anatol/gem2arch 2015-05-28 14:56:46 ncopa: another nice/similar effort -> https://github.com/fedora-ruby/gem2rpm 2015-05-28 14:58:18 ncopa: I see a reference to this also => gem2rpm -> rpm2aport (ncopa's hidden script) 2015-05-28 14:58:28 and this => git pull git://git.alpinelinux.org/fab/aports 2015-05-28 15:04:03 shreeve, check this ^^^^^ 2015-05-28 15:08:09 algitbot: holy hotness! i'll check it out now 2015-05-28 15:08:33 algitbot: how did you do that? 2015-05-28 15:09:12 uh... am i talking to a bot? 2015-05-28 15:09:16 hehe 2015-05-28 15:09:21 you are... 2015-05-28 15:09:22 :) 2015-05-28 15:09:23 fcolista: how did *you* do that? 2015-05-28 15:09:42 fcolista: are you sure it actually works? 2015-05-28 15:09:50 I doubt it 2015-05-28 15:09:55 barthalion, no, that's why i asked to test 2015-05-28 15:09:57 fcolista: not yet... ;-) what do I need to type to add the edge repo??? 2015-05-28 15:10:10 I don't need to even install it 2015-05-28 15:10:11 https://rubygems.org/gems/mechanize/versions/2.7.3 2015-05-28 15:10:14 unmet dependencies 2015-05-28 15:10:54 shreeve: I believe it was something like gem2arch, but I wasn't really around back then, so I don't remember 2015-05-28 15:11:53 barthalion: ok... i see gem2arch, gem2rpm... 2015-05-28 15:13:06 how do i add that @testing repo? 2015-05-28 15:20:24 barthalion: fcolista: ncopa: this seems super close -> https://github.com/anatol/gem2arch/blob/master/gem2arch.rb 2015-05-28 15:21:26 hmm... go 1.5 can have deps 2015-05-28 16:47:09 alpine-xen problem: i can see the vmx flag in cpuinfo in alpine 3.1.4 but not in alpine-xen 3.2 2015-05-28 16:57:44 does xen hide that in some way? 2015-05-28 17:04:22 Xen sees VMX http://sprunge.us/ANKC 2015-05-29 06:51:12 ncopa, https://wiki.postgresql.org/wiki/May_2015_Fsync_Permissions_Bug 2015-05-29 06:52:15 section, "Should I not apply the updates" 2015-05-29 06:56:31 vkrishn: did it bite you? 2015-05-29 06:56:51 does it affect us? 2015-05-29 06:57:08 do we have root owned files in $PGDATA? 2015-05-29 06:57:16 the page suggests setting file permissions 2015-05-29 06:57:25 as a temp workaround yes 2015-05-29 06:57:35 Will you be fixing it soon? 2015-05-29 06:57:35 The PostgreSQL project expects to release another update very soon which addresses the file permissions issue. The tentative release date for this update is June 4. 2015-05-29 06:57:47 so a proper fix will come soon 2015-05-29 06:58:19 is it ssl related ? 2015-05-29 06:58:32 "FATAL: could not open file "/etc/ssl/certs/ssl-cert-snakeoil.pem": Permission denied" 2015-05-29 06:58:57 so basically, they recommend that you change file permissions and then do apk upgrade to get the fixed postgres 2015-05-29 06:59:49 isnt that due to symlink? 2015-05-29 06:59:51 have one or more files or directories, or symlinks to one or more files or directories, not owned or writeable by the "postgres" user (or other installation owner) under the postgres data directory (PGDATA). 2015-05-29 07:00:26 sounds to me that they have a symlink in PGDATA to /etc/ssl/certs 2015-05-29 07:00:40 and that /etc/ssl/certs are not writable by postgres user 2015-05-29 07:01:01 but vkrishn, is this a problem for us? 2015-05-29 07:01:03 I don't use postgres much 2015-05-29 07:01:03 does it affect us? 2015-05-29 07:03:00 would try to install, and check 2015-05-29 07:04:13 seems like we dont have anything not writeable by postgresql in $PGDATA in default setup 2015-05-29 07:04:55 i think we can just wait til jun 4 2015-05-29 07:05:01 no stress 2015-05-29 07:05:16 ok 2015-05-29 08:06:26 any change of seeing my aports pushed to testing? 2015-05-29 08:06:52 i added the prepare to moc. thanks for the tip ncopa :) 2015-05-29 08:08:11 I'm playing around gem2arch. 2015-05-29 08:08:31 I've a basic working gem2apk which generates APKBUILD 2015-05-29 08:09:07 what's not working is the automatic resolution of dependencies 2015-05-29 09:03:42 uh? 2015-05-29 09:03:43 make: i486-alpine-linux-musl-g++: Command not found 2015-05-29 09:04:44 thats probably due to our recent i586 change 2015-05-29 09:04:53 yes 2015-05-29 09:05:08 umh 2015-05-29 09:06:03 i will need to fix that :-/ 2015-05-29 09:06:28 also 2015-05-29 09:06:31 that package or this behaviour? 2015-05-29 09:06:41 all testing/ruby-* are broke 2015-05-29 09:06:48 ..nice.. 2015-05-29 09:06:51 or most of them 2015-05-29 09:07:23 i think all testing/ruby-* packages that has not been rebuild since ruby 2.2 update needs a rebuild 2015-05-29 09:07:47 abuild bump of all of them? 2015-05-29 09:07:54 yes 2015-05-29 09:08:11 i think you can do git log main/ruby 2015-05-29 09:08:19 find the commit of the update 2015-05-29 09:08:57 then git log .. -- testing/ruby-* 2015-05-29 09:09:21 those are the testing/ruby-* packages that does not need to rebuild 2015-05-29 09:09:33 the packages not listed needs bump pkgrel 2015-05-29 09:09:36 and testbuild 2015-05-29 09:10:50 hum 2015-05-29 09:11:09 i think we woudl have gotten away if we we would have used --build=$CBUILD \ 2015-05-29 09:11:09 --host=$CHOST \ 2015-05-29 09:11:38 does ruby gem install supports that? 2015-05-29 09:11:52 support what? 2015-05-29 09:12:05 those flags: --build && --host 2015-05-29 09:12:32 i think gem use what the built ruby says 2015-05-29 09:12:45 ruby says whatever was used at build time 2015-05-29 09:12:46 oh you mean for ruby package 2015-05-29 09:12:48 yes 2015-05-29 09:12:51 ok 2015-05-29 09:17:35 usr/lib/ccache/bin/i486-alpine-linux-musl-c++ 2015-05-29 09:17:35 ===== ccache-3.2.2-r0.apk 2015-05-29 09:18:07 i'm searching for similar breakages 2015-05-29 09:18:32 usr/lib/kodi/system/libsse4-i486-linux.so 2015-05-29 09:18:32 ===== kodi-14.2-r3.apk 2015-05-29 09:19:47 usr/lib/ruby/2.2.0/i486-linux-musl/psych.so 2015-05-29 09:19:48 ===== ruby-libs-2.2.2-r0.apk 2015-05-29 09:19:52 this one we already knew 2015-05-29 09:36:14 I'm curious if ruby-mechanize works now 2015-05-29 10:20:05 <[xming]> o/~ Ruby Ruby Ruubiii o/~ 2015-05-29 10:21:03 I'm having a problem with libtool while packaing 2015-05-29 10:21:08 packaging even 2015-05-29 10:21:53 <[xming]> fcolista: if you want to script auto-update the v86 image, you don't even need qemu, just mount with kpartx (or loop with offset) then chroot and apk update :D 2015-05-29 10:23:03 libtool: link: gcc -o dictzip dictzip.o data.o etc etc but gcc cant see any of the object files 2015-05-29 10:23:32 they are definitly in .libs 2015-05-29 10:23:46 how do I point gcc to them? 2015-05-29 10:24:11 <[xming]> I don't know how libtool works 2015-05-29 10:24:42 <[xming]> for plain gcc you just need the right paths 2015-05-29 10:24:55 <[xming]> I suppose libtool should do that 2015-05-29 10:25:16 sure, just wondered where to specify the paths 2015-05-29 10:34:18 <[xming]> probably in the auto* or configure.ac, which package is it? 2015-05-29 11:04:42 libtool uses .la wrappers that should point to .libs 2015-05-29 11:37:26 ncopa: should I tag fortify 0.5? 2015-05-29 11:37:32 now that 3.2.0 has been released 2015-05-29 11:39:41 http://dl.2f30.org/releases/fortify-headers-0.5.tar.gz 2015-05-29 11:51:04 stateless: sure 2015-05-29 12:22:19 ncopa, did you update edge x86 builder to new toolchain already? 2015-05-29 12:22:31 to i586 yes 2015-05-29 12:22:41 you rebuilding everything? 2015-05-29 12:22:44 no 2015-05-29 12:22:46 ah, ok 2015-05-29 12:23:11 might be an idea to do it at somepoint now that we have hash=gnu default, and i586 2015-05-29 12:23:16 but you had some additional changes still? 2015-05-29 12:23:17 yes 2015-05-29 12:23:25 i was thinking til after gcc-5.1 2015-05-29 12:23:33 and i have some abuild changes i'd like 2015-05-29 12:23:42 i'd like -dev subpackage be more optimized 2015-05-29 12:23:47 oh, update x86 kernel config too 2015-05-29 12:23:51 right 2015-05-29 12:23:57 that we also need to to 2015-05-29 12:24:14 i have a llvm patch pending too 2015-05-29 12:24:16 CONFIG_X86_MINIMUM_CPU_FAMILY=4 --> 5 2015-05-29 12:24:23 which makes it bootstrap itself 2015-05-29 12:24:37 i was thinking of do llvm today 2015-05-29 12:24:50 the firefox x86 issue is a bit critical too 2015-05-29 12:24:53 but nontrivial 2015-05-29 12:24:58 maybe i should just enable -dbg 2015-05-29 12:25:19 new musl is due out soon(ish) 2015-05-29 12:25:34 there is new fortify-headers too 2015-05-29 12:35:39 ahills: nice to have you here :) 2015-05-29 12:36:17 ahills: i did a minor clean up for xbindkeys: http://dup.pw/844d4c8cd9b9 2015-05-29 12:36:52 note that you dont need libc-dev in makedepends 2015-05-29 12:36:52 Ah yes, I keep accidentally setting et in vim 2015-05-29 12:37:01 and depends=libx11 2015-05-29 12:37:22 abuild will use scanelf to detect the .so dependencies 2015-05-29 12:37:28 I'm going to get build container up before more aports 2015-05-29 12:38:00 Yeah I saw, I just thought it was a huge for the packager 2015-05-29 12:38:04 Guide 2015-05-29 12:38:16 Sorry I'm on my phone as I test eudev 2015-05-29 12:38:25 :D 2015-05-29 12:38:51 i know exactly what you mean: talk in phone, hack code + type irc.... 2015-05-29 12:38:53 I'm unemployed until the end of June so I should have plenty of time... 2015-05-29 12:38:53 :) 2015-05-29 12:38:57 cool! 2015-05-29 12:39:16 not cool to be unemployed 2015-05-29 12:39:23 but cool to have time :) 2015-05-29 12:39:25 I should say funemployed 2015-05-29 12:39:39 No I quit my job at Cisco, it sucked 2015-05-29 12:39:58 did you code there? 2015-05-29 12:40:20 I did a lot of things... 2015-05-29 12:40:57 familiar with dmvpn/flexvpn etc? 2015-05-29 12:41:20 ahills: i feel your pain 2015-05-29 12:41:26 with job that sucks 2015-05-29 12:41:37 did you know that fabled wrote opennhrp, the dmvpn for linux 2015-05-29 12:41:40 we've all had those :) 2015-05-29 12:43:30 Not familiar no 2015-05-29 12:44:18 I was tools engineer for group of 400ish, but it turned into an operations job, and I have no patience for 3am calls from India 2015-05-29 12:48:49 My group was responsible for AnyConnect VPN client 2015-05-29 12:49:01 Reading dmvpn page on alpine wiki 2015-05-29 12:54:48 Wow, are you using this setup currently? 2015-05-29 12:55:09 I need to buy some more network hardware... 2015-05-29 12:58:16 ahills, i'm aware of at least 4 large install 2015-05-29 12:58:27 me being involved running one ofthem 2015-05-29 12:59:46 I'm always jealous of people who get to play with this kind of equipment 2015-05-29 13:18:43 OK, still not able to get kms at boot 2015-05-29 13:19:09 mkinitfs's kms.modules file lists some directories, does it just pull in everything under? 2015-05-29 13:26:15 Got it, needed to add i915 to /etc/modules 2015-05-29 13:28:30 Creating a wiki account: 2015-05-29 13:28:36 Alpine Linux could not send your confirmation mail. Please check your email address for invalid characters.Mailer returned: Unknown error in PHP's mail() function. 2015-05-29 13:29:09 ugh 2015-05-29 13:30:54 managing a PHP application is one of the reasons I quit my job ;) 2015-05-29 13:31:07 :) 2015-05-29 13:31:16 the account created fine though 2015-05-29 13:33:38 <[xming]> all cook people here 2015-05-29 13:33:47 <[xming]> cool* :D 2015-05-29 13:43:19 ncopa: are you the mediawiki master? is aports a specially linked thing or whatever? 2015-05-29 13:43:26 "interwiki" 2015-05-29 13:45:21 i'm not the mediawiki master 2015-05-29 13:45:37 the mediawiki master is on vacation i think 2015-05-29 13:45:48 i think i can fix the mailer problem though 2015-05-29 13:46:48 I don't care about the mailer, I was just wondering if I should refer to aports on a wiki page in any special way 2015-05-29 13:47:27 dunno to be honest :) 2015-05-29 13:52:51 ahills: copy paste your edit to safe local doc, the wiki has a tendence to lose your changes 2015-05-29 13:53:56 heh, thanks for the warning 2015-05-29 13:54:21 enterprise web applications have already made that a habit of mine 2015-05-29 13:55:12 the mailer should work now 2015-05-29 13:58:57 it is haproxy in front 2015-05-29 13:59:18 i wonder if the timeouot values or something makes it lose the edits 2015-05-29 14:01:27 sigh 2015-05-29 14:01:33 llvm broke again in x86 2015-05-29 14:04:47 /usr/bin/ld: cannot find crtbeginS.o: No such file or directory 2015-05-29 14:07:05 is it bootstrapped llvm or built with gcc? 2015-05-29 14:07:41 bootstrapped 2015-05-29 14:07:46 ithink i know why 2015-05-29 14:10:36 god I hate the mediawiki format 2015-05-29 14:17:20 seems nice for job/process queue, http://kr.github.io/beanstalkd/ 2015-05-29 14:34:34 ok, how can I have multiple lines in a {{Cmd|}} thing 2015-05-29 14:34:46 sometimes it works, but sometimes the text just shows up as {{{1}}} 2015-05-29 14:34:51 I hate PHP and everything around it 2015-05-29 14:40:11 {{cat|/etc/mkinitfs/mkinitfs.conf|features="some stuff"}} always yields {{{2}}} 2015-05-29 14:40:14 whyyyyy 2015-05-29 15:12:29 <[xming]> ahills: I think it's the pipe 2015-05-29 15:30:39 I figured it out 2015-05-29 15:30:46 the = needs {{=}} 2015-05-29 15:30:51 otherwise it's a keyword argument 2015-05-29 15:31:13 of course it took me an hour to figure out that it's a template and not an extension/plugin 2015-05-29 15:31:14 silly me 2015-05-29 15:34:18 ok, added a few wiki pages; kms is a stub but I have an amd card around here somewhere that I should be able to use to test that 2015-05-29 16:33:45 if as a long-term project I start compiling manuals for the various busybox commands, is there any interest in adding a busybox-doc package? 2015-05-29 16:34:51 mostly for the ones that don't have a -h option, or that aren't exactly a command 2015-05-29 16:34:56 (e.g. man 5 interfaces) 2015-05-29 16:36:29 +1 2015-05-29 16:38:05 +1, but send them to upstream if applicable 2015-05-29 16:38:25 I will need to do some research beforehand to see if upstream wants them 2015-05-29 16:38:31 otherwise I will just start a project and become mr. upstream 2015-05-29 16:46:54 ahills: it would be a better use of your time to probably contribute to toybox instead of busybox 2015-05-29 16:51:37 Alpine will be switching? 2015-05-29 16:51:44 at some point in the future yes 2015-05-29 16:52:11 How distant might that point be? 3.3? 2015-05-29 16:52:26 probably in 3.5-3.6 time period 2015-05-29 16:53:20 busybox has gotten really bloated and has various design flaws (like our needing to ship bbsuid(1) for example to work around them), which could be corrected in toybox instead 2015-05-29 16:53:23 ;) 2015-05-29 17:02:33 Is toybox mature enough yet? 2015-05-29 17:02:48 no 2015-05-29 17:02:54 that's why we're still on busybox 2015-05-29 17:07:58 ahills: it's getting there, but not yet. hince "probably in 3.5-3.6 time period" 2015-05-29 17:10:05 I guess my concise question would be, is it just a matter of waiting for it to mature, and Alpine will switch as soon as possible? 2015-05-29 17:11:18 it's a complicated situation. some things we want need to be replaced outside toybox (such as ifupdown, which should really be a daemon so that state is properly managed. going to probably work on this after the raid stuff.) 2015-05-29 17:11:52 busybox is a strange situation because it's the entire kitchen sink, usually implemented oddly. 2015-05-29 17:12:02 hince why it's common for alpine users to install coreutils. 2015-05-29 17:12:08 and the other packages 2015-05-29 17:13:33 if you want a hand with the ifupdown stuff, color me interested 2015-05-29 17:13:53 well what i am thinking is a lightweight daemon which handles all networking functions 2015-05-29 17:13:55 I've been pretty displeased with connman, networkmanager, etc 2015-05-29 17:14:01 so we can justify having it in base 2015-05-29 17:14:13 (since it will replace the need for bridge-tools et al.) 2015-05-29 17:14:33 and it should handle launching wpa_supplicant over wireless too, or no? 2015-05-29 17:14:39 s/over/for/ 2015-05-29 17:14:52 yes ideally. or just have the functionality built in. 2015-05-29 17:14:53 c.f. openbsd's /etc/hostname 2015-05-29 17:15:10 i don't really care either way -- so whichever makes more sense 2015-05-29 17:15:19 wpa_supplicant is a mess of bad code but it has a lot of features... 2015-05-29 17:15:37 yes, such an integration would involve studying which parts are relevant of course. 2015-05-29 17:16:07 the systemd guys aren't wrong in that a lot of this stuff should be daemons and have clean well-defined interfaces 2015-05-29 17:16:19 it's just everything else about systemd that seems, "hmm" to me 2015-05-29 17:17:19 never going back after my journald fiasco 2015-05-29 17:18:15 +1 2015-05-29 17:18:15 oh well, alpine is proof that you don't have to really go along with that crowd. 2015-05-29 17:18:32 there are many ideas behind systemd that i like 2015-05-29 17:18:39 the technical ideas 2015-05-29 17:18:55 but as said, evertying around it makes me go "hmm" :) 2015-05-29 17:19:51 anyway, raid is more interesting at the moment, because if we have hw raid support out of the box, then that's really interesting vs. everyone else 2015-05-29 17:20:39 I am not much interested in hw raid because I don't have any (anymore) 2015-05-29 17:20:47 reminds me 2015-05-29 17:21:28 one of the server hw managers asked me to add zabbix monitoring for the raid on our build servers 2015-05-29 17:21:39 this is the config: UserParameter=raid.status,cciss_vol_status /dev/sda |grep OK |wc -l 2015-05-29 17:22:13 alpine 3.3 shipping with *free* raid management tools will be something that will be a really huge thing and we'll probably get a lot of attention for it 2015-05-29 17:22:32 when looking at it, there was no clean way to make zabbix tell which of the disks was failed 2015-05-29 17:22:37 (i mean as in 'freedom' there) 2015-05-29 17:22:58 i looked at how to do it in my dell server here 2015-05-29 17:23:05 there was no clean way to do it 2015-05-29 17:23:25 does zabbix support plugins? 2015-05-29 17:23:31 i think so 2015-05-29 17:23:37 i havent worked with it enough 2015-05-29 17:23:41 we could probably provide a zabbix plugin then for it 2015-05-29 17:23:46 once it's done of course. 2015-05-29 17:24:05 the various raid cards i ordered, start arriving next week 2015-05-29 17:24:07 haha 2015-05-29 17:26:22 https://www.zabbix.com/documentation/1.8/manual/tutorials/extending_agent 2015-05-29 17:26:25 ugly 2015-05-29 17:26:50 that is just a white screen here 2015-05-29 17:27:13 oh dear 2015-05-29 17:27:19 that is awful 2015-05-29 17:27:36 they fork/exec | grep | sed etc and call that a plugin :) 2015-05-30 02:16:08 So, the mumble package (and its murmur subpackage) are broken 2015-05-30 02:16:19 the abuild succeeds because of a missing || return 1 2015-05-30 02:16:55 Is there a reason not to just set -e when evaluating the APKBUILD functions? 2015-05-30 02:22:58 ahills - if you need a vnc viewer - try virt-viewer - it has spice & vnc support 2015-05-30 02:23:23 with spice you also get copy & paste support http://it-offshore.co.uk/linux/alpine-linux/30-alpine-linux-spice-kvm-desktop 2015-05-30 02:25:57 oh nice 2015-05-30 02:26:11 yeah I think I'm abandoning tigervnc, even the version I've built has some strange behavior 2015-05-30 02:27:38 alpine's spice also has support for LZ4 compression 2015-05-30 02:28:26 sweet 2015-05-30 02:28:33 I feel like this has improved the usefulness of my computer 2015-05-30 02:28:35 thanks 2015-05-30 02:30:39 I didn't test vnc in virt-viewer but it should work - was no reason to use it with copy & paste in spice 2015-05-30 02:31:12 I tested it 2015-05-30 02:31:26 well, I didn't test copy/paste 2015-05-30 02:31:34 can you use spice over network? 2015-05-30 02:31:44 yes 2015-05-30 02:31:52 the vnc protocol is not good 2015-05-30 02:32:16 you'd need to setup an ssl certificate on the remote machine 2015-05-30 02:33:05 i think you can use ssh with spice too 2015-05-30 02:35:09 yeah I already tunnel my VNCs, I'd probably even have it listen on the same ports... 2015-05-30 02:37:31 ugh installing probably an ancient version of spice-vdagent on debian... bleh 2015-05-30 02:37:43 I guess VNC's one advantage is that it hasn't changed since the '90s 2015-05-30 02:38:47 i managed to compile the latest spice everything on debian 2015-05-30 02:39:09 yeah but I don't care about these machines enough :P 2015-05-30 02:39:21 :-D 2015-05-30 02:40:16 the virt-viewer in debian wheezy did not have spice support 2015-05-30 02:41:08 fortunately, only my vncservers are on debian, and my laptop is running beautiful alpine 2015-05-30 08:10:22 ahills: posix shells don't have 'set -e' 2015-05-30 08:10:28 it's a bashism 2015-05-30 09:30:52 you sure? dash and ash have it 2015-05-30 09:31:18 now when you said so, no 2015-05-30 09:32:24 pretty sure it's posix, but I'm digging up the standard 2015-05-30 09:32:30 oh well, I was wrong my entire life 8) 2015-05-30 09:32:34 haha 2015-05-30 09:32:52 ncopa: as ahills says, any reason why we don't use set -e? 2015-05-30 10:54:56 hmm, I see awall is GPL2, but I see no license for apk-tools, etc 2015-05-30 10:55:11 is GPL2 preferred license for alpine tools? 2015-05-30 11:06:18 apk-tools is GPL. 2015-05-30 11:06:42 GPL is "ok", BSD-like is better 2015-05-30 11:06:49 GPL3 is no-go 2015-05-30 11:17:14 why do you think so? 2015-05-30 11:22:19 it's not a matter of thinking; that's the policy for alpine tools 2015-05-30 11:23:06 too many consumers are concerned with using GPL3 components in their projects due to the various political features 2015-05-30 11:23:25 we do not want that, we want anyone to use alpine for any project they want 2015-05-30 11:24:15 so licenses which are permissive (or at least, well-understood) are what is required 2015-05-30 11:24:38 ok 2015-05-30 11:24:50 apk-tools doesn't have a LICENSE in the tree on git.alpinelinux.org 2015-05-30 11:25:31 the comments are GPLv2 2015-05-30 11:25:37 in the individual files 2015-05-30 11:25:40 ahh ok 2015-05-30 11:25:41 it should probably be corrected. 2015-05-30 11:25:56 hopefully the whole license isn't in each file ;) 2015-05-30 11:59:49 kaniini: thx for the answer 2015-05-30 13:00:26 should .a files go in a lib package or a lib-dev package? 2015-05-30 13:00:51 I would use -dev, but what is currently the system? 2015-05-30 13:15:34 ahills - in the APKBUILD put: subpackages="$pkgname-dev" - & abuild creates the dev pkg automatically 2015-05-30 13:16:20 yes, but I have a more complex setup, as I'm building eudev, and it has a libudev and libgudev subpackage, both of which have their separate headers and such 2015-05-30 13:17:44 but: .a goes in -dev? 2015-05-30 13:18:22 i just normally set the $pkg-dev & don't worry about the contents 2015-05-30 13:20:27 sometimes *.la files need to be deleted in package() 2015-05-30 13:20:55 yeah, already covered 2015-05-30 13:21:24 as soon as I test this, I'll submit it, and you can critique my methods 2015-05-30 13:21:49 maybe it's as simple as letting dev be automatic and having the other -dev subpackages copy from $pkgdir-dev 2015-05-30 13:21:56 but I don't want to depend on the order unless it's guaranteed 2015-05-30 13:54:10 has anyone got a good setup for cross-compiling to x86/armhf via abuild? 2015-05-30 13:54:24 it's not by any chance as simple as setting CARCH, is it? 2015-05-30 13:54:48 and/or CHOST 2015-05-30 15:26:57 ncopa: any luck with clang on x86? the build works for me, but my abuild config is pretty conservative and i'm not using ccache or anything funky. 2015-05-30 17:04:44 Hi 2015-05-30 17:05:11 mesa-dri-nouveau seems broken... 2015-05-30 17:05:21 ERROR: unsatisfiable constraints: 2015-05-30 17:05:26 so:libLLVM-3.6.so (missing): 2015-05-30 17:05:33 required by: mesa-dri-nouveau-10.5.6-r2[so:libLLVM-3.6.so] 2015-05-30 17:06:51 'llvm-libs-3.6.1-r1.apk' has not a 'libLLVM-3.6.so' file 2015-05-30 17:11:34 AmatCoder: I don't know if it's related, but llvm build was broken for a bit 2015-05-30 17:11:43 not sure if it's fixed 2015-05-30 17:13:26 hmm it seems not... 2015-05-30 17:14:53 3.5.1 has a 'libLLVM-3.5.so' as I remember 2015-05-30 17:15:25 you may want to pin your nouveau package in the meantime 2015-05-30 17:16:08 late... I clean the cache :( 2015-05-30 17:17:08 not too late, just grab the aports repo, revert to previous revision of APKBUILD and build it 2015-05-30 17:17:24 could take a while if you cleaned llvm too 2015-05-30 17:19:29 yes I would need build llvm and mesa...at least 2015-05-30 17:20:13 oh, and llvm builds clang also... 2015-05-30 17:20:37 actually you just need to build llvm, unless others are broken 2015-05-30 17:20:43 not fun but at least it's automated 2015-05-30 17:22:10 you mean to edit APKBUILD to avoid building clang? 2015-05-31 03:16:35 AmatCoder: you could do that... I would probably just wait 2015-05-31 03:20:58 mesa probably just needs a rebuild. 2015-05-31 06:04:04 ncopa: any luck with clang on x86? the build works for me, but my abuild config is pretty conservative and i'm not using ccache or anything funky. 2015-05-31 11:14:10 http://turtle.dereferenced.org/~kaniini/interfaces.conf 2015-05-31 11:14:12 opinions ? 2015-05-31 11:14:22 oops. 2015-05-31 11:14:24 http://turtle.dereferenced.org/~kaniini/interfaces.conf.txt 2015-05-31 11:15:31 zomg 2015-05-31 11:15:42 that's everything that systemd-networkd should be but it is not 2015-05-31 11:16:01 +1 2015-05-31 11:18:35 well, the idea beyond that is to allow additional functionality to be provided via plugins 2015-05-31 11:18:45 for example starting openvpn or whatever 2015-05-31 11:19:46 then you use ifup/ifdown/ifstat as you would now. (well ifstat is new) 2015-05-31 12:27:08 barthalion: i think networkd-1 will be available as say, a preview, in 3.3, and then we will cut over in 3.4 2015-05-31 13:54:53 kaniini: is that an existing format or did you just craft that example? 2015-05-31 13:55:39 ahills: it is somewhat based on /etc/network/interfaces of today 2015-05-31 13:55:46 ahills: just converted into a proper tree syntax 2015-05-31 13:55:59 ahills: the format itself is used by BIND for example 2015-05-31 13:58:58 Yes and nginx 2015-05-31 13:59:23 Is your networkd written in C? 2015-05-31 13:59:46 that is the plan 2015-05-31 13:59:51 i havent written anythign yet 2015-05-31 13:59:58 That's what I meant 2015-05-31 14:00:01 i am just kind of sketching out how the ux is going to be 2015-05-31 14:00:25 I was just recoiling in horror at the idea of a tree parser in sh 2015-05-31 14:01:04 ahills: networkd is planned to be a daemon which manages the network state 2015-05-31 14:01:14 you can't do daemons in sh ;) 2015-05-31 14:01:16 This format looks easy to generate, which I think is important to not lose moving away from /e/n/i 2015-05-31 14:01:16 well 2015-05-31 14:01:19 i mean, you can 2015-05-31 14:01:35 ahills: it is easy to generate and more easy to parse correctly 2015-05-31 14:01:56 Yes you can ;) https://github.com/ahills/beard 2015-05-31 14:02:35 Will you define a formal grammar for one of those parser tools like yacc? 2015-05-31 14:02:56 ahills: no, i have a parser for that format already 2015-05-31 14:03:09 ahills: i will literally just copy it into place 2015-05-31 14:03:25 You wrote, or grabbed out of other software? 2015-05-31 14:04:19 https://github.com/atheme/libmowgli-2/blob/master/src/libmowgli/ext/confparse.c 2015-05-31 14:04:21 I want a parser for this format for some other projects... Have you published/licensed it? 2015-05-31 14:04:35 yes, above 2015-05-31 14:05:28 OK enough of this phone, where is a real computer... 2015-05-31 14:05:45 if you want an already standalone version 2015-05-31 14:06:17 grab one from https://github.com/atheme/atheme-6.0/blob/master/src/confparse.c 2015-05-31 14:06:30 and, https://github.com/atheme/atheme-6.0/blob/master/include/confparse.h 2015-05-31 14:09:01 there is also https://github.com/vstakhov/libucl and libconfig 2015-05-31 14:10:28 libucl is really overengineered 2015-05-31 14:11:42 i should make a really standalone version of that parser that you can just git submodule into place 2015-05-31 14:12:55 the code is simple and lightweight vs those libs 2015-05-31 14:13:01 and has been used in a lot of places 2015-05-31 14:24:13 mine also has the advantage of having compact boolean values just by specifying the name of the parameter 2015-05-31 14:35:00 kaniini, I want to see inside, it would be possible ?, some git repository? 2015-05-31 14:35:39 kaniini has been linking git repositories 2015-05-31 14:41:55 kaniini: libmowgli looks pretty cool, why not just pull that in as a dependency? 2015-05-31 14:42:08 seems like you guys have already put a lot of effort into it 2015-05-31 15:00:11 also, for networkd, do you have a daemon framework you like, or were you just going to create that on the fly