2014-07-01 01:44:28 could nut be backported to 2.6? 2014-07-01 06:06:18 ncopa, for #3089 Mp5shooter asks to backport nut to 2.6. What's our policy for backport? Iirc is only for bugfixes but i'm not sure if is _only_ for that 2014-07-01 06:08:59 uh 2014-07-01 06:09:03 2.6 is bugfixes only 2014-07-01 06:09:27 or even more - security only 2014-07-01 06:09:28 http://wiki.alpinelinux.org/wiki/Alpine_Linux:Releases 2014-07-01 06:10:14 though, i think we have made _some_ exceptions 2014-07-01 06:14:55 we have made exceptions 2014-07-01 06:15:05 specially if its from brothers 2014-07-01 06:16:08 morning 2014-07-01 06:17:23 ...and we only backport things thats been tested 2014-07-01 06:18:50 any body know how to configure udhcpc? issue 3105 2014-07-01 06:18:56 #3105 2014-07-01 06:19:01 algitbot: hello 2014-07-01 06:19:07 algitbot: hi 2014-07-01 06:19:09 #3105 2014-07-01 06:19:16 http://bugs.alpinelinux.org/issues/3105 2014-07-01 06:20:09 new kernels are out again 2014-07-01 06:20:32 probably due to this lzo busyness 2014-07-01 06:20:36 business 2014-07-01 06:20:55 i wonder if we should do 3.0.2 soonish due to that 2014-07-01 06:21:08 ncopa, i'm waiting for #3090 feedback 2014-07-01 06:21:09 ncopa when in Bug #3105 maybe to check that what i wrote last night? 2014-07-01 06:21:34 i made a lua script for that 2014-07-01 06:21:57 but csv parsing with pattern matching with lua is quite uncomfortable for me 2014-07-01 06:22:18 especially if the csv ha double quotes with commas inside 2014-07-01 06:22:23 that should not be parsed 2014-07-01 06:22:53 with regexp is easier 2014-07-01 06:23:11 also the csv shipped contains a comma and a space at the end :-/ 2014-07-01 06:26:02 crow: looks weird 2014-07-01 06:26:14 do you have suggestion how i reproduce it? 2014-07-01 06:26:22 ncopa which ones from the above? 2014-07-01 06:26:29 lvm 2014-07-01 06:26:55 ncopa sure following this what i wrote http://wiki.alpinelinux.org/wiki/Installing_on_GPT_LVM 2014-07-01 06:27:23 this with edge repository is not needed anymore and just apk add -U gptfdisk 2014-07-01 06:27:35 ok, i kinda expected EFI for booting from GPT 2014-07-01 06:27:56 ncopa i did it inside vmware workstation, and this howto worked fine with alix 3d3 2014-07-01 06:28:03 k 2014-07-01 06:28:11 i should try it in qemu then 2014-07-01 06:28:14 ncopa syslinux works to if you have boot and bios partition 2014-07-01 06:28:19 ef02 2014-07-01 06:28:45 ncopa but actually you will get that msg before finishing, 2014-07-01 06:29:03 and if you complete install it will backtrace on boot 2014-07-01 06:29:07 (at least here) 2014-07-01 06:29:44 does syslinux require a bios partition? 2014-07-01 06:32:55 crow, that is interesting 2014-07-01 06:33:11 i didnt know it is possible to use gpt on bios mode disks :) 2014-07-01 06:40:23 ncopa for me it was also unknown, but i was in mude to try gtp and i dont have a efi system hier. but i found they thit that trick with ef02 partition 2014-07-01 06:50:16 crow: ok, i can reproduce it here 2014-07-01 06:59:03 crow, the mlock error is due to ulimit -l beeing to restrictive 2014-07-01 06:59:11 ulimit -l 1024 2014-07-01 06:59:17 and the error will disapear 2014-07-01 07:05:51 no, sorry, i was wrong its not 2014-07-01 07:08:44 crow: i think the lvm issue qualifies for a bug report 2014-07-01 07:30:41 ncopa :( 2014-07-01 07:30:56 should i fill a bug report or you did it already? 2014-07-01 07:31:09 crow: it would be nice if you could file it 2014-07-01 07:31:15 only the lvcreate part 2014-07-01 07:31:29 it appears to be reproduceable on any lvm 2014-07-01 07:31:40 i'm talking with musl devs 2014-07-01 07:34:53 ncopa i have it with 2.7.8 install but as that box is production i dont like to change/try it, but i could try in one VM 2014-07-01 07:35:34 it happens with alpine v2.7.8 too ? 2014-07-01 07:35:39 or only 3.0 2014-07-01 07:35:58 just lvcreate -n temp should trigger it 2014-07-01 07:36:13 so as long as you have a few MB free for test it should work 2014-07-01 07:36:23 it also happens on lvremove 2014-07-01 07:43:30 ncopa on 2.7.X there was no problem and i could easyly create lvm and also did wrote that howto after 2014-07-01 07:43:44 ncopa does it boot after on your side or on boot segfaults? 2014-07-01 07:43:52 ncopa bug reported: http://bugs.alpinelinux.org/issues/3107 2014-07-01 07:44:28 ncopa i time writing that howto (also stated in howto) i used alpine-2.7.2-x86.iso 2014-07-01 07:44:37 thanks 2014-07-01 07:44:44 i haven tried boot it yet i'm on it 2014-07-01 07:45:00 but i can reproduce the lvm issue 2014-07-01 07:45:18 it seems like ulimit -l is ignored 2014-07-01 07:45:26 so it smells kernel issue 2014-07-01 07:45:45 :( 2014-07-01 07:46:12 i cant try it on hardware as i cant create bootable usb 2014-07-01 07:47:18 (01:16:56) (crow) ncopa regarding editing syslinux.cfg seems empty i am not sure why.. http://pastebin.com/vnLKRtJM 2014-07-01 07:47:18 (01:17:41) (crow) i mounted iso to /mnt and did setup-bootable /mnt /dev/sdb2 and then mounted it manually to /media/sdb1 to edit syslinux.cfg 2014-07-01 07:48:14 ncopa i am not sure if reledet but on 3.0.1 boot first line is: mdev: bad line 73: 2 tokens found, 3 needed 2014-07-01 07:51:35 ok 2014-07-01 07:51:46 yes syslinux.cfg is empty 2014-07-01 07:51:48 thats ok 2014-07-01 07:51:54 actually, i think it should be removed 2014-07-01 07:52:01 we use /boot/extlinux.conf 2014-07-01 07:52:04 (for now) 2014-07-01 07:52:42 the mdev bad line error is fixed and will only happen when you boot the iso/usb 2014-07-01 07:52:52 so when you reboot from disk its gone 2014-07-01 07:53:03 (and the only effect it has is that /dev/snd/* has wrong premissions) 2014-07-01 07:53:21 my vm booted just fine 2014-07-01 07:53:35 i believe what happens with lvm is 2014-07-01 07:53:48 lvm tries to mlock all the memory it uses itself 2014-07-01 07:54:05 that is, it protects the lvm runtime to be swapped out 2014-07-01 07:54:37 so all the memory lvm uses, inclusive the dynamic linked objects (the .so libs) 2014-07-01 07:54:47 cannot get swapped out by memory manager 2014-07-01 07:54:51 that operation fails 2014-07-01 07:55:50 the only effect that will have is that if memory gets low, the kernel is allowed to swap out the memory used by lvm itself 2014-07-01 07:56:13 but there are no swap enabled at that time so that should not be a problem 2014-07-01 07:57:29 and i dont really undrestand why it should be a problem at all... might be they are afraid that the memory will never be allowed to swapped in again and it can never exit cleanly if you for example have memory leak or so 2014-07-01 08:02:18 crow: basically, it works for me, even if it give some error messages 2014-07-01 08:43:21 ncopa i can try it again, but i got backtrace as no root partition found or something 2014-07-01 08:49:05 ncopa i cant find the /boot/extlinux.conf inside this "bootalble" usb 2014-07-01 08:50:04 i have this usb (microusb to usb adapter is used) pluged on alix 3d3 with 2.7.7. so i can try to recreate bootable iso 2014-07-01 09:49:35 barthalion: i saw you backported zram-init to v2.7 2014-07-01 09:49:44 i dont think the kernel there works? 2014-07-01 09:53:45 ncopa yesterday i used setup-bootable to create an bootable usb, and there boot/syslinux/syslinux.conf was empty or with NULL inside, today i followed manual method to create it, and it seems fine. 2014-07-01 09:53:57 ncopa: yeah, I realized after the push 2014-07-01 09:54:02 i cant test if its boot, but i see contens of syslinux.conf 2014-07-01 09:54:12 barthalion: ok to remove it? 2014-07-01 09:54:16 yup 2014-07-01 15:03:12 ACTION needs to upgrade a box from 2.7 to 3.0 2014-07-01 15:03:37 anyone remember the doc for upgrading from uclibc to musl? 2014-07-01 15:04:53 its on a.lo 2014-07-01 15:04:56 a.o 2014-07-01 15:05:51 tdtrask: http://alpinelinux.org/edge-musl 2014-07-01 15:06:34 clandmeter: thank you 2014-07-01 15:06:50 for some reason, I couldn't find that by searching 2014-07-01 15:07:09 me nether 2014-07-01 15:07:17 i logged in an checked 2014-07-01 15:55:24 \o/ 2014-07-01 15:55:35 my brand new build server is working 2014-07-01 15:55:45 decided to abandon the old one :) 2014-07-01 15:59:16 are there docs about how to create a buildserver? 2014-07-01 16:00:02 not only build single packages ... build all packages and create buildlogs and so on 2014-07-01 16:35:51 pluesch_: I was just trying to build individual packages for testing before pushing 2014-07-01 16:36:08 but, there are some docs on the wiki 2014-07-01 16:37:00 http://wiki.alpinelinux.org/wiki/Creating_an_Alpine_package is for individual packages 2014-07-01 16:37:47 http://wiki.alpinelinux.org/wiki/Setting_up_a_compile_vserver descibes buildozer (from 1.5 years ago) 2014-07-01 16:47:03 tdtrask: ty 2014-07-01 16:47:08 second link looks interesting 2014-07-01 16:47:13 note that current infra uses lxc 2014-07-01 16:49:48 okay 2014-07-01 16:55:53 I don't think the current infra is documented anywhere 2014-07-01 16:58:16 jop... 2014-07-01 16:59:56 hmmm ... another question: how can I get a list of all languages that are available? the env variable I'm talking about is LANG= and it's defined in /etc/profile 2014-07-02 08:36:08 rnalrd, do we really want u-boot to be split for each subpkg? 2014-07-02 08:36:35 board types come and go 2014-07-02 08:36:38 well, package can be several MBs while boards are added 2014-07-02 08:36:42 especially when the SPL stuff changes 2014-07-02 08:36:50 so i thought of splitting it 2014-07-02 08:36:59 we don't want that installed 2014-07-02 08:37:03 it's like extlinux 2014-07-02 08:37:06 upgrade once 2014-07-02 08:37:19 and the boot loaders should really go to the install tarball 2014-07-02 08:37:47 though 2014-07-02 08:37:59 k, i just wondered how we wanted to distribute it 2014-07-02 08:38:07 me too 2014-07-02 08:38:15 there's no perfect solution 2014-07-02 08:38:20 imho, splitting it is a lot of work 2014-07-02 08:38:23 and gain is little 2014-07-02 08:45:07 rnalrd, but yes, we should figure out how to do arm release tarballs 2014-07-02 08:45:15 and do package split accordingly for the u-boot loaders 2014-07-02 08:45:21 yeah 2014-07-02 08:45:31 i think some distros just place raw u-boot binaries on the web 2014-07-02 08:45:44 and some package per architecture groups (armv7, armv8, etc) 2014-07-02 08:50:23 i think we should to release tarballs more or less as we do ISOs for x86(_64) 2014-07-02 08:52:03 well, since it contains only alpine-base, actually we don't need it to release it so often 2014-07-02 08:52:30 i'm just wondering how to install alpine/armhf on a sdcard 2014-07-02 08:52:57 should we have alpine-armhf.tar.gz that contains all u-boot variants + base; and setup-bootable can take "--board fooboard" 2014-07-02 08:53:07 or should setup-bootable only update .apks and kernel 2014-07-02 08:53:29 re setup-bootable 2014-07-02 08:53:33 and making the SDcard really bootable would be job for setup-arm-bootloader that gets the u-boot from internet/.apk/another .tar.gz ? 2014-07-02 08:53:54 i have been thinking of rewriting setup-bootable to be distro agnostic 2014-07-02 08:54:01 so you can download it on any distro 2014-07-02 08:54:18 and it will generate a bootable usb from an alpine.iso image 2014-07-02 08:54:44 secondly, i think we want some script that updates /boot incl modloop 2014-07-02 08:55:31 should we scrap .iso as medium for x86 / x86_64? or add .tar.gz as alternate download format? 2014-07-02 08:55:40 that script should be able to generate a working /boot for liveusbs 2014-07-02 08:56:00 and possibly also from kernel build tree 2014-07-02 08:56:14 nowadays nobody burns an iso anymore i guess 2014-07-02 08:56:14 but that script probably depends on apk 2014-07-02 08:56:35 the nice thing with .iso is that i can easily boot those from qemu 2014-07-02 08:56:40 right 2014-07-02 08:56:46 .iso is nice for virtualization 2014-07-02 08:56:49 and virtual mountable 2014-07-02 08:57:02 correct 2014-07-02 08:57:02 so iso + tar.gz for ARM 2014-07-02 08:57:11 i think iso does not make sense for arm 2014-07-02 08:57:15 nope 2014-07-02 08:57:29 .iso is only the archive format 2014-07-02 08:57:32 how about .img 2014-07-02 08:57:36 that are bootable 2014-07-02 08:57:42 raw bootable images 2014-07-02 08:57:42 osx uses .img 2014-07-02 08:58:02 we coudl maybe use ext4 or so for boot usb 2014-07-02 08:58:05 dd if=.img of= 2014-07-02 08:58:13 yes 2014-07-02 08:58:25 or using image writer for windows 2014-07-02 08:58:27 on ARM we still need to do special things for each board variant 2014-07-02 08:58:29 but you would have to make an image for every board 2014-07-02 08:58:49 do we want continue with FAT filesystem? 2014-07-02 08:59:01 i suppose we need FAT for uefi... 2014-07-02 08:59:01 ncopa, some ARM boards require FAT 2014-07-02 08:59:04 still it will be smaller than tipycal iso 2014-07-02 08:59:09 e.g. Rpi 2014-07-02 08:59:19 so might not be a problem after all 2014-07-02 08:59:23 the rom code boots from first FAT 2014-07-02 08:59:34 searching specificly named files 2014-07-02 08:59:36 the problem with fat is that you cannot extend it afterwards 2014-07-02 08:59:45 kernel can then mount secondary partition as root 2014-07-02 09:00:09 same for Bbone board 2014-07-02 09:00:25 u-boot partition is FAT 2014-07-02 09:00:29 FAT16 2014-07-02 09:00:50 so if we want dd if=alpine.img of=sdcard/usbstick we'd need either mutiple partitions with the boot as fat 2014-07-02 09:01:06 hm 2014-07-02 09:03:12 how about gpt? 2014-07-02 09:03:17 and uefi? 2014-07-02 09:03:23 lunch 2014-07-02 09:03:31 i think we want uefi boot too for v3.1 2014-07-02 11:04:45 ncopa, how are the security fixes handled? I mean: do you have a sort of checking of new cves and compare it with Alpine sw versions? Is automatic or manual? 2014-07-02 11:08:15 manual 2014-07-02 11:08:24 we have a contributor that monitors security lists 2014-07-02 11:08:45 then finds out which packages/git branches are affected 2014-07-02 11:08:48 8-| 2014-07-02 11:08:54 it's a huge work 2014-07-02 11:09:15 yup 2014-07-02 11:09:33 and fixing all the issues is fairly timeconsuming too 2014-07-02 11:09:40 for 4 branches... 2014-07-02 11:10:00 ok. I ask because i realized that i belong to alpine security, but i nevere did something for that and i would like to start.. 2014-07-02 11:10:14 :) 2014-07-02 11:11:24 while you are taking about setup-bootable , something is wrong there, as i cant do it correctly, and syslinux.conf is empty and can be used to boot, if i follow manuall way SD card is bootable 2014-07-02 11:12:48 and again i vote for iso hybrid which can be dd to SD/USB and boot from that device 2014-07-02 20:14:18 ncopa: xvidcore has new abi? 2014-07-02 20:18:51 I don't think so 2014-07-02 20:24:39 Error relocating /usr/lib/libavcodec.so.55: xvid_global: symbol not found 2014-07-02 20:27:38 barthalion: seems to be fixed now 2014-07-02 20:28:33 weird, it purged xvidcore after upgrade 2014-07-02 20:56:17 barthalion: its not abi thats the problem. they fixed linkage to the lib so the custom apkbuild linking breaks things. 2014-07-02 21:04:21 clandmeter: right, I've removed it from PKGBUILD as well 2014-07-03 07:08:54 hi. is there a way to "conflict" packages with each other? I want to build a package that should conflict with another package and the package manager should remove the conflicting package before installing the new one. I think "replaces" isn't what I'm looking for, because the "old package" will still be installed. 2014-07-03 07:10:58 this is another undocumented thing 2014-07-03 07:11:09 you can list conflicting packages in conflits="" 2014-07-03 07:12:45 thank you barthalion :) 2014-07-03 07:14:33 or depend="!foo" 2014-07-03 07:16:35 ncopa: the same effect? 2014-07-03 07:17:25 im not sure conflicts="" works.. 2014-07-03 07:17:45 ah, it should 2014-07-03 07:18:39 depends="!package" should have same effect 2014-07-03 07:18:50 okay. I like this :D 2014-07-03 07:18:51 good to know 2014-07-03 07:44:44 ncopa: what is the logic of mkinitfs.trigger 2014-07-03 07:45:08 mine just upgraded kernel but it changes my extlinux config 2014-07-03 07:45:17 it makes gsec default again 2014-07-03 08:33:15 maybe i forgot to push the fix 2014-07-03 08:33:41 clandmeter: which alpine release? edge, v3.0 or v2.7? 2014-07-03 08:36:38 ncopa: edge 2014-07-03 08:37:21 ncopa: http://sprunge.us/IeCJ 2014-07-03 08:37:38 yes 2014-07-03 08:37:43 someone reported it 2014-07-03 08:37:53 in fact multiple perons reported it and i got bitten too 2014-07-03 08:37:57 but i forgot to create a ticket 2014-07-03 08:38:01 so i forgot to fix it... 2014-07-03 08:38:05 :) 2014-07-03 08:38:09 im on the way 2014-07-03 08:43:29 ncopa: craeted another one for that vmware bug 2014-07-03 08:49:46 ncopa: where can i find the log from x86? 2014-07-03 08:50:38 htink i got it 2014-07-03 08:52:14 seems it doenst have recent logs.. 2014-07-03 08:53:17 ncopa: http://bld1.alpinelinux.org/buildlogs/build-edge-x86/main/zoneminder/ doesnt have logs after r9 2014-07-03 08:53:27 i guess this is the correct location? 2014-07-03 09:03:46 yes 2014-07-03 09:04:01 http://bld1.alpinelinux.org/buildlogs/build-edge-x86/main/zoneminder/zoneminder-1.27.0-r9.log 2014-07-03 09:04:12 zm_utils.cpp: In function 'void ssedetect()': 2014-07-03 09:04:12 zm_utils.cpp:208:3: error: PIC register clobbered by '%ebx' in 'asm' 2014-07-03 10:00:36 ncopa: yes but thats not the latest build 2014-07-03 10:00:39 its on r11 now 2014-07-03 13:35:54 barthalion: conflicts="package" doesn't work for me, but depends="!package" works :) 2014-07-03 13:36:07 tested on edge 2014-07-03 16:47:51 any thoughts on swapping udev with eudev so that we don't have to rely on a udev release from 2011? 2014-07-03 16:48:18 arent we on mdev? 2014-07-03 16:48:26 or do i mistake something? 2014-07-03 17:02:27 uggedal: I'd rather have eudev AND udev 2014-07-03 17:09:02 WwQCn4HqahlD: mdev is used for headless installs 2014-07-03 17:10:37 oooh, i see. i never done anything else. thanks for enlightening me 2014-07-03 17:12:12 barthalion: fine with me, but 28 apks depend on udev 2014-07-04 12:29:30 so, I have got the svg from tobias, converted the logo to a webfont and hacked up a new layout 2014-07-04 12:30:14 written a tiny script that generates .html from markdown sources with a yaml header 2014-07-04 12:30:30 and a makefile to generate the files that have been modified 2014-07-04 12:31:06 the general concept is found here: http://wwwtest.alpinelinux.org/ 2014-07-04 12:31:46 I havent figured out how to do the block elements with news, download and packages so they show up horizontal 2014-07-04 12:33:38 i have started to generate latest-releases.yaml too 2014-07-04 12:34:13 with the data needed for the download boxes to get generated 2014-07-04 12:54:29 would be nice to have some fluid/responsive design. 2014-07-04 12:54:54 i like it! 2014-07-04 12:55:51 yeah 2014-07-04 12:56:07 its 960px wide 2014-07-04 12:56:10 not sure i like that 2014-07-04 12:56:38 i was thinking of responsive design, but i doubt people will browse the site from cellphone 2014-07-04 12:57:21 you would be surprised how many people use smartphones and tablets to browse the internet 2014-07-04 12:58:14 if we have one site that would be used by mobile devices, it would be www. 2014-07-04 12:59:36 would be nice if we could get help from somebody with design experiance. 2014-07-04 12:59:56 uggedal: didnt you sign website? 2014-07-04 12:59:59 design 2014-07-04 13:31:40 like wwwtest.a.o :-) 2014-07-04 13:35:43 I have tried some of known responsive designs(css), http://insteps.net/pr/a/pmwiki/Cookbook/GridWorks 2014-07-04 13:35:44 any list of backend apps that would be used to generate the contents ? 2014-07-04 13:50:09 thanks for avfs update 2014-07-04 14:13:05 where are source files downloaded before bld[12] pkgs' it 2014-07-04 14:13:08 http://dev.alpinelinux.org/archive/ ? 2014-07-04 14:13:59 no 2014-07-04 14:14:04 distfiles.alpinelinux.org 2014-07-04 14:14:12 thanks 2014-07-04 14:14:21 but thats only bld2 2014-07-04 14:14:26 so its only fox x86_&4 2014-07-04 14:14:33 the idea is to share it with nfs... 2014-07-04 14:14:39 so both use same 2014-07-04 14:14:49 yes but sources would be same 2014-07-04 14:15:08 yup 2014-07-04 14:15:21 I just wanted to download some, that I having difficulty from orig src 2014-07-04 14:15:36 thats why distfiles.alpinelinux.org exists in first place 2014-07-04 14:16:58 I wonder I can mount it via sshfs, would it be safe ? 2014-07-04 14:17:21 it should be safe, bot not very efficient 2014-07-04 14:17:36 i think you can add a mirror in abuild.conf 2014-07-04 14:17:47 ok 2014-07-04 14:17:52 DISTFILES_MIRROR=http://distfiles.alpinelinux.org/distfiles 2014-07-04 14:17:57 will make it look there first 2014-07-04 14:19:24 does is support byte range, wget -c ? 2014-07-04 14:19:54 the webserver? 2014-07-04 14:20:08 ncopa: i thought it was small simple secure 2014-07-04 14:20:08 dont know, if not, I can enable it 2014-07-04 14:20:18 its lighttpd 2014-07-04 14:20:19 would be nice, thanks 2014-07-04 14:20:29 oh :) 2014-07-04 14:20:33 yeah 2014-07-04 14:20:38 :P 2014-07-04 14:20:49 thats why i ask for feedback ;) 2014-07-04 14:21:06 really, it felt wrong ... now i know why :) 2014-07-04 14:21:12 haha 2014-07-04 14:24:38 i think the 'about' page should have 3 sections, one for 'small', one for 'simple' and one for 'secure' 2014-07-04 14:27:15 its fixed now 2014-07-04 14:27:27 i should probably set up an autobuild hook 2014-07-04 14:29:05 on front page 2014-07-04 14:29:24 i think we want "Download latest" in the center or above 2014-07-04 14:29:43 maybe 2 blocks below with news and 'latest development' 2014-07-04 14:30:00 i also wonder if we want list git commits instead of a list with the last packages 2014-07-04 15:01:56 ACTION pushed acf-weblog to edge and backported to 3.0 2014-07-04 15:02:10 but, the push to edge failed because ncopa pushed before I did 2014-07-04 15:02:30 so, now the cherry-pick commit looks wrong :( 2014-07-04 15:02:42 ACTION should have moved slower, I suppose 2014-07-04 15:03:58 my v3.0 dev box is in a different container 2014-07-04 15:04:08 and i alwasy git fetch there 2014-07-04 15:04:11 and git log origin/master 2014-07-04 15:04:17 and use that when cherry-pick 2014-07-04 15:04:50 now its weekend 2014-07-04 16:40:22 yay weekend :D 2014-07-04 16:40:44 time to hack 2014-07-04 18:11:28 clandmeter: yeah, i made http://mediaqueri.es 2014-07-07 07:53:15 files from v2.5.4-316-gf6419b1 uploaded 2014-07-07 12:20:14 files from v2.5.4-317-g4e5e63d uploaded 2014-07-07 14:55:53 Hi 2014-07-07 18:14:10 hurray for alpine-xen! 2014-07-07 18:14:24 please git pull git://git.alpinelinux.org/fab/aports 2014-07-07 18:14:28 there are some updates 2014-07-07 18:15:18 thanks 2014-07-07 19:53:16 ACTION is looking for some news regarding musl and nfs 2014-07-07 19:53:27 but there seems to be no news :-( 2014-07-07 19:54:37 now three-quarter of my build-machines are not usable 2014-07-07 19:54:46 no edge-builds possible 2014-07-07 19:54:55 no 3.0-builds possible 2014-07-07 19:55:02 bad, very bad... :-( 2014-07-07 19:55:10 i'm getting stuck 2014-07-07 20:07:51 n8@all 2014-07-08 10:20:51 http://www.minnowboard.org/where-to-buy/ - :-) 2014-07-08 10:21:44 can I install AL on http://en.wikipedia.org/wiki/Intel_Galileo ? 2014-07-08 10:25:32 I think so 2014-07-08 10:30:07 x86 soc/sbc fills the gap, I hope it becomes as popular and easy diy like desktops 2014-07-08 12:26:20 ncopa, do you have some sort of trigger that when a new kernel is shipped it rebuilds all the $pkgname-grsec linked? 2014-07-08 12:27:29 no 2014-07-08 12:27:47 but i have a script that edites the APKBUILDs, builds and git add 2014-07-08 12:29:10 ok. I'm wondering how to deal with the usbip userspace, that is inside kernel tree, and should be shipped when a new kernel version is shipped. 2014-07-08 12:29:23 Dunno if the userspace is touched anyway 2014-07-08 12:29:33 i dont know if we want it build build with -grsec kernel 2014-07-08 12:30:13 not really. Is only that the userspace is inside the kernel tree, and i'd like to avoid to re-download the entire tree 2014-07-08 13:08:29 bah, i forgot to reset pkgrel when updating the kernel 2014-07-08 15:11:27 http://www.citadel.org - can this be packaged, the deps seems long 2014-07-08 15:30:39 is there a change log/new features page on http://alpinelinux.org? 2014-07-09 09:03:25 is SIGCLD on musl called SGCHLD ? 2014-07-09 09:06:26 http://infohost.nmt.edu/~eweiss/222_book/222_book/0201433079/ch10lev1sec7.html 2014-07-09 09:07:07 looks like its 2 different things 2014-07-09 09:13:13 yes 2014-07-09 09:29:05 umh 2014-07-09 09:29:31 include/linux/signal.h does not have SIGCLD 2014-07-09 09:29:45 and here: 2014-07-09 09:29:46 http://git.alpinelinux.org/cgit/aports/plain/main/openjdk7/icedtea-hotspot-musl.patch 2014-07-09 09:29:56 +#ifdef SIGCLD 2014-07-09 09:29:56 "CLD", SIGCLD, /* Same as SIGCHLD (System V). */ 2014-07-09 09:29:56 +#endif 2014-07-09 09:48:10 ncopa, can you take a look to this? http://sprunge.us/bgNG 2014-07-09 09:48:28 this is the patch http://sprunge.us/BBMU 2014-07-09 09:48:45 basically i reviewed what der-tiger did for usbip-utils 2014-07-09 09:48:54 i'm not sure about the versioning 2014-07-09 09:49:31 yeah, not sure about that either :) 2014-07-09 09:49:44 iirc they mention other version in configure.in 2014-07-09 09:49:49 2.0.0 2014-07-09 09:49:53 so --version will tell something else 2014-07-09 09:49:54 or something like that 2014-07-09 09:50:05 i dont know really 2014-07-09 09:50:19 Ok, i can use the configure.in (makes more sense) 2014-07-09 09:50:31 except that it makes it harder to do abump 2014-07-09 09:50:43 i think it makes more sense to use the version in configure.in 2014-07-09 09:50:53 +1 2014-07-09 09:51:01 what about downloading the kernel sources ? 2014-07-09 09:51:04 but what if they push minor commits/fixes 2014-07-09 09:51:15 without changing version number in configure.in? 2014-07-09 09:51:45 umh. 2014-07-09 09:51:59 Why they don't use a proper versioning? 2014-07-09 09:52:04 :-/ 2014-07-09 09:52:19 its conveninent to let others do the releasing for them 2014-07-09 09:52:38 they dont need do releases anymore 2014-07-09 09:52:58 re sources 2014-07-09 09:53:07 maybe do what we do in kernel builds 2014-07-09 09:53:26 pick the first major release and then add version patch 2014-07-09 09:54:01 i'd be ok to use kernel version number 2014-07-09 09:54:25 i saw that. What i need is only part of the kernel tree. But i'd like to avoid downloading the entire package. 2014-07-09 09:54:38 I've not find a proper way 2014-07-09 09:54:41 *found 2014-07-09 09:54:45 then we'd need to du surgery 2014-07-09 09:54:47 do* 2014-07-09 09:54:56 and ship our own tarball 2014-07-09 09:55:03 (eg do the release tarball for them) 2014-07-09 09:55:26 but this means that everyime we shoudl create a new tarball 2014-07-09 09:55:30 yes 2014-07-09 09:55:39 this means we need to do the releases for them 2014-07-09 09:55:43 and i dont think we want that 2014-07-09 09:55:50 i think i even complained about it upstream 2014-07-09 09:55:52 and we don't know in advance if a new kernel version ships a new usbip-utils modifies 2014-07-09 09:56:00 right 2014-07-09 09:56:06 i don't like that 2014-07-09 09:56:17 so w'ed need check git log 2014-07-09 09:56:21 and i dont like that either 2014-07-09 09:56:38 we could maybe even ship it as a subpackage to linux-vanilla 2014-07-09 09:56:38 a 3rd way could be use the kernel package already downloaded for linux-grsec 2014-07-09 09:56:47 yes 2014-07-09 09:56:51 or linux-vanilla 2014-07-09 09:56:55 right 2014-07-09 09:56:57 umh.. 2014-07-09 09:57:20 yes 2014-07-09 09:57:24 i like this 2014-07-09 09:57:26 usbip-utils users would get updates to the userland tool even if there are no changes there 2014-07-09 09:57:35 with every single kernel upgrade 2014-07-09 09:57:42 but thats a price we can pay i think 2014-07-09 09:57:47 for simpler maintenance 2014-07-09 09:57:56 a subpkg of linux-vanilla is ok for simpler maintenance. 2014-07-09 09:58:02 Or 2014-07-09 09:58:09 in source= 2014-07-09 09:58:22 i can put the linux-$kernerlversion 2014-07-09 09:58:31 but i've to set the path 2014-07-09 09:58:39 right path of the filesystem 2014-07-09 09:58:47 but is not an elegant solution 2014-07-09 09:58:52 other option is 2014-07-09 09:59:05 have a separate APKBUILD 2014-07-09 09:59:15 like the sprunge you showed me 2014-07-09 09:59:27 but only use linux-3.14.xz 2014-07-09 09:59:31 and cal the version 3.14 2014-07-09 09:59:40 and simply drop the zillion kernel bug fixes 2014-07-09 09:59:52 so usb-utils-3.14 2014-07-09 09:59:58 instead of 3.14.10 2014-07-09 10:00:00 3.14.11 2014-07-09 10:00:02 etc 2014-07-09 10:00:17 i think this is similar to what iproute2 does 2014-07-09 10:00:32 i think the might be bettter 2014-07-09 10:00:38 sounds good. This doesn't solve the problem of re-download the entire kernel package. 2014-07-09 10:00:45 But fix the versioning. 2014-07-09 10:01:11 i've the option of abuild git 2014-07-09 10:01:15 with snapshot() 2014-07-09 10:01:21 and creates the tarball 2014-07-09 10:01:25 upload it 2014-07-09 10:01:27 and use it 2014-07-09 10:01:30 actually 2014-07-09 10:01:38 it does solve re-download the kernel tarball 2014-07-09 10:01:53 How? 2014-07-09 10:02:20 source="http://ftp.kernel.org/pub/linux/kernel/v3.x/linux-$_kernver.tar.xz 2014-07-09 10:02:20 http://ftp.kernel.org/pub/linux/kernel/v3.x/patch-$pkgver.xz 2014-07-09 10:02:24 thats from kernel build 2014-07-09 10:02:33 not that we add a patch-$pkgver 2014-07-09 10:02:42 the $_kernver is: 2014-07-09 10:02:48 pkgver=3.14.11 2014-07-09 10:02:48 case $pkgver in 2014-07-09 10:02:48 *.*.*) _kernver=${pkgver%.*};; 2014-07-09 10:02:48 *.*) _kernver=${pkgver};; 2014-07-09 10:02:49 esac 2014-07-09 10:03:05 so _kernver=3.14 2014-07-09 10:03:32 which means we download linux-3.14.tar.xz + patch-3.14.y.xz 2014-07-09 10:03:44 and linux-3.14.tar.gz will be in download cache 2014-07-09 10:03:46 if i set the source is such way, do i avoid to redownload the package because it is already downloaded by linux-grsec? 2014-07-09 10:03:47 ok 2014-07-09 10:03:49 is in cahce 2014-07-09 10:03:51 *cache 2014-07-09 10:03:53 yes 2014-07-09 10:03:56 fine 2014-07-09 10:04:00 pkgname=usbip-utils 2014-07-09 10:04:00 pkgver=3.14.10 2014-07-09 10:04:04 use pkgver=3.14 2014-07-09 10:04:10 should solve things 2014-07-09 10:04:26 got it 2014-07-09 10:04:35 the new problem we get from that is 2014-07-09 10:04:45 if they in a 3.14 branch add fixes 2014-07-09 10:04:50 we will miss those 2014-07-09 10:05:19 i think the useland tools are not going to be modified very often 2014-07-09 10:05:23 *userland 2014-07-09 10:05:27 *nod* 2014-07-09 10:05:31 thats what i think too 2014-07-09 10:05:56 just if new modules implements cool feature.. 2014-07-09 10:06:10 but we are atalking about usb over ip... 2014-07-09 10:06:10 shouldnt implement new features 2014-07-09 10:06:48 here http://bugs.alpinelinux.org/issues/650 der-tiger did a sort of APKBUILD 2014-07-09 10:07:12 but is totally out-of-standard 2014-07-09 10:07:20 also it parses html :-0 2014-07-09 10:07:34 soudns like wrong thing to do 2014-07-09 10:07:40 definitely 2014-07-09 10:07:41 last commit to usb-utils 2014-07-09 10:07:45 v3.9-rc5-637-g70e90fb 2014-07-09 10:08:14 so usb-utils-3.9 and usb-utils-3.14 are exactly same thing... 2014-07-09 10:08:34 ok :) 2014-07-09 10:08:48 frustrating 2014-07-09 10:09:07 lazy devs who dont want do proper releasing 2014-07-09 10:09:11 github generation... 2014-07-09 10:09:37 eh eh 2014-07-10 08:39:07 i think i got my upstream version checker to find new versions from github 2014-07-10 08:39:12 github has an api 2014-07-10 08:39:36 but they only allow you to do 60 requests per hour unauthenticated 2014-07-10 08:39:49 so its simpler to parse the https... 2014-07-10 08:39:59 parse the html 2014-07-10 16:52:17 please git pull git://git.alpinelinux.org/fab/aports 2014-07-11 09:14:53 how do i change the keymap so i can swap caps in the console? the bmaps are bit too binary for my taste. 2014-07-11 09:36:43 WwQCn4HqahlD, setup-keymap? 2014-07-11 10:53:11 rnalrd: ok, but it offers only keyboard variants based on antiquated partitioning of the world, what i care about is the position of the caps lock key. sorry for being ignorant can you tell me how to affect that with setup-keymap? 2014-07-11 11:16:03 WwQCn4HqahlD, sorry i dunno 2014-07-11 11:53:50 :/ 2014-07-11 11:58:39 fabled, i got an email with report that 3.14 vanilla kernel has broken NIC driver for hyper-v: https://lists.debian.org/debian-kernel/2014/05/msg00263.html 2014-07-11 11:59:13 its fixed in upstream but does not appear to have made it to -stable 2014-07-11 11:59:23 i wonder if we should cherry-pick it ourselves 2014-07-11 11:59:49 hmm 2014-07-11 12:00:02 is it a confirmed fix? is there any side effects? 2014-07-11 12:00:02 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=af9893a3dc790ae0c4d3e68adde12bc3cb9c63fa 2014-07-11 12:00:30 its fixed in 3.15-rc-something 2014-07-11 12:00:40 so 3.15 kernel should have it 2014-07-11 12:00:58 i'm ok to cherry-pick too 2014-07-11 12:01:12 we could ask if it could go to -stable too 2014-07-11 12:01:23 i think it should 2014-07-11 12:01:53 apparently 3.14.y kernel is broke on yper-v 2014-07-11 12:02:43 ok, it does not apply cleanly 2014-07-11 12:03:35 then i'm probably leaning towards let upstream handle it 2014-07-11 12:21:04 hi 2014-07-11 12:22:47 ncopa: could you push flashcache to 3.0? facebook made a new release for kernel 3.14 2014-07-11 12:26:32 aha 2014-07-11 12:26:33 ok 2014-07-11 12:27:57 have there been any attempts to switch from gcc to clang? 2014-07-11 12:34:08 nothing serious 2014-07-11 12:34:12 bet we have talked about it 2014-07-11 12:34:52 rnalrd: im adding 3.10.48 as we speak... 2014-07-11 12:35:44 Linux localhost 3.10.48-0-grsec #1-Alpine SMP Fri Jul 11 11:47:40 UTC 2014 i686 Linux 2014-07-11 12:38:53 k ncopa 2014-07-11 12:39:27 im doing 3.14 kernel too for v3.0 2014-07-11 12:39:52 and add flashcache-grsec, so I can update to 3.0 soon :-) 2014-07-11 12:40:06 i'll look at it while at it yes 2014-07-11 14:24:43 ncopa: Can I use the current 3.0 iso image and flashcache-grsec? or must it be in the iso file too? 2014-07-11 14:27:34 depends on your install 2014-07-11 14:27:45 is you have a 'sys' disk install then it should work 2014-07-11 14:28:15 if you run from tmpfs you need create new /boot with kernel, initramfs and modloop 2014-07-11 14:28:36 (its on my list to write a tool for that) 2014-07-11 14:41:15 I run from tmpfs... So, still 2.7, but 2.7 has not the apk docker :( 2014-07-11 14:43:28 i am sceptic to move flashcache to main 2014-07-11 14:43:57 in the 3.14 kernel case, we got the fixed module now 2014-07-11 14:44:18 if we would depend on that, we would not have been able to ship v3.0 with 3.14 kernel 2014-07-11 14:44:30 (or we would need to fix it ourselves) 2014-07-11 14:45:04 missing 3.14 kernel would have been bad because it would have meant no zram and no LTS 2014-07-11 15:24:42 ncopa: mh... :( 2014-07-11 16:10:46 I get a mdev error: http://s7.directupload.net/images/140711/fggk8cuf.png 2014-07-11 16:10:52 starting 3.0.1 from usb 2014-07-11 16:44:44 for the record on another host having my prefered keybindings and being managed by kbd, i used busybox dumpkmap and used that on the other host. 2014-07-11 18:19:47 setup-bootable is broken for 3.0.0. I can not boot from usb 2014-07-11 18:19:58 Boot failure : No DEFAULT or UI configuration directive found! 2014-07-11 18:23:46 same with 3.0.1 2014-07-11 18:23:50 what the hell :( 2014-07-11 18:24:54 I am trying to format with F16 instead of F32 2014-07-11 18:30:36 I tried to move from AL 2.4.6_x64 to 2.6.6_x86 just doing setup-bootable 2014-07-11 18:31:19 and now it is saying "/sbin/init not found in new root" 2014-07-11 18:32:47 any idea on how to fix it? 2014-07-11 18:44:58 seems that 3.0.x is broken when I want to run it from tmpfs 2014-07-11 19:10:50 Ok, I tried 2.7.0 and it works 2014-07-11 19:10:56 will now upgrade to 3.0, I will see... 2014-07-11 19:31:39 Okay, installing 2.7.0 on tmpfs and update to 3.0.0 worked 2014-07-11 19:31:43 now, I will try 3.0.1 2014-07-11 19:32:11 mkdosfs or syslinux from 3.0.0 is broken 2014-07-11 19:36:00 3.0.1 is broken 2014-07-11 21:09:05 shafire that mdev error is already fixed https://bugs.alpinelinux.org/issues/3097 2014-07-11 21:11:01 shafire yes seems setup-bootable is broken but if you wanna make an new iso bootable from SD or USB do it manually 2014-07-11 21:11:58 crow: okay, apk upgrade did not help 2014-07-11 21:12:10 So I will wait for 3.0.2 2014-07-11 21:12:47 http://s14.directupload.net/images/140711/vuahtfz9.png <-- what error is this? 2014-07-11 21:13:04 shafire http://pastebin.com/2NBnsdtH 2014-07-11 21:13:24 shafire you mean the cp one? 2014-07-11 21:13:38 yes 2014-07-11 21:13:53 duno i have had it to, was not problem on boot 2014-07-11 21:14:29 shafire in the above doku its to create new bootable usb from already running alpine (other linux should work to) 2014-07-11 21:14:36 and setup-bootable is broken in 3.0 ... http://s14.directupload.net/images/140711/m5w8vnsq.png 2014-07-11 21:14:44 shafire yes i wrote that to 2014-07-11 21:14:56 my workaround was to install 2.7.0 first and upgrade to 3.0 2014-07-11 21:15:06 crow: thanks 2014-07-11 21:15:42 shafire well 2.7 is uclibc based, 3.0 is musl so i prefered to do manualy way of creating bootable SD then upgrading 2014-07-11 21:16:20 crow: that's not a big problem :) 2014-07-11 21:16:56 because of uclibc -> musl change 2014-07-11 21:17:37 yeah, but running from tmpfs should not be a problem as long as you have not compiled something manually 2014-07-11 21:17:52 so upgrade is "safe" 2014-07-11 21:18:40 so i preffer clean iso :D 2014-07-11 21:18:47 its nice to have choice 2014-07-11 21:19:23 crow: your pasted iso is like "setup-bootable alpine-3.0.1-x86_64.iso /media/usb" 2014-07-11 21:19:26 it does the same 2014-07-11 21:20:03 shafire just this one works for me, and setup-bootable not 2014-07-11 21:20:11 why, what was your error? 2014-07-11 21:21:01 http://bugs.alpinelinux.org/issues/3137 2014-07-11 21:21:39 yeah, this is my bug too :D 2014-07-11 21:22:37 see :) 2014-07-11 21:23:09 but 2014-07-11 21:23:10 i needed to change syslinux.conf for sending console output to serial console (no vga on this device) 2014-07-11 21:23:18 the sysconfig.cfg was okay after install 2014-07-11 21:23:36 when I reboot, then it was unreadable 2014-07-11 21:24:29 i think it was here after install, because i need to change syslinux to be able to setup 2014-07-11 21:25:45 #3137 updated 2014-07-11 21:30:58 crow: are you using arm? 2014-07-11 21:32:57 no x86 (apu4c) 2014-07-11 21:34:26 i am not fan of arm 2014-07-12 16:48:41 hello! I'm an OpenRISC developer. If "bootstrapping" is defined as starting from 0 alpine packages, how hard would it be to bootstrap it to openrisc would you say? For debian it was quite a pain since because of the dependency cycles (gcc build-depends on latex which depends on everything..) 2014-07-12 16:50:21 surely less than debian 2014-07-12 16:50:27 and many other distributions around 2014-07-12 16:51:27 I guess what I'm wondering is how you handle those kind of dependencies and if you require alpine to build alpine. 2014-07-12 16:51:58 also, debian has wanna-build for automatic package building - I see you have algitbot which I assume is connected to some build system 2014-07-12 16:52:18 I guess you need to wait for fabled, he ported Alpine to ARM 2014-07-12 16:52:41 aha 2014-07-12 16:52:42 but we're avoiding dependency cycles when it's possible 2014-07-12 16:53:01 yes, algitbot is a message broker 2014-07-12 16:53:10 and various build bots are subscribers 2014-07-12 16:58:25 (well, skipping a lot of complexity behind the scenes made by ncopa) 2014-07-12 17:02:43 blueCmd_, i have relative good cross build scripts 2014-07-12 17:02:48 not perfect though 2014-07-12 17:03:07 if musl supports the architecture, the effort should be about 1-2 days 2014-07-12 17:03:22 basically adding the arch to apk-tools, and abuild 2014-07-12 17:03:26 and firing cross build 2014-07-12 17:03:39 and tuning necessary gcc flags 2014-07-12 17:03:44 dalias and stekern in #openrisc are working hard on morting musl, so that should be fine 2014-07-12 17:04:14 the (old) scripts i used are at http://dev.alpinelinux.org/~tteras/bootstrap/ 2014-07-12 17:04:29 our gcc isn't upstreamed yet, so to compile it you need special patches - is that a problem? for debian that is a bit of a pain 2014-07-12 17:04:49 how invasive they are? 2014-07-12 17:04:55 we have already relatively patched gcc 2014-07-12 17:05:00 as invasive as adding a new architecture is 2014-07-12 17:05:05 very isolated 2014-07-12 17:05:49 might need to consult ncopa about it 2014-07-12 17:05:59 silently I'm wishing you would say "yes, that's hard" so I would get a move-on to upstream it :P 2014-07-12 17:06:29 i was considering adding codesourcery arm patches 2014-07-12 17:06:39 but figured it's not worth tracking that set 2014-07-12 17:06:47 so i'm just waiting that stuff to get upstreamed 2014-07-12 17:06:59 so yes, having the openrisc stuff upstreamed would be simpler way to go 2014-07-12 17:08:07 brb 2014-07-12 17:35:03 libressl is finally out. 2014-07-12 17:51:06 hi fabled, libressl is out 2014-07-12 17:51:12 yes, i saw 2014-07-12 17:51:16 however... i think we will need to patch it a bit 2014-07-12 17:51:20 i wonder how usable it is. and if it's abi stable 2014-07-12 17:52:01 openssl people seem to have activated too 2014-07-12 17:52:07 so perhaps we wait for a while and see how it goes 2014-07-12 17:52:08 no, it's an abi bump for sure 2014-07-12 17:52:14 we would have to rebuild the distribution 2014-07-12 17:52:25 something we need to consider for alpine 3.1 2014-07-12 17:53:11 gcc-4.9 is out too; so if we bump it, it's worth to recompile everything 2014-07-12 17:53:59 though gcc-4.9 still had some bugs that we want fixed 2014-07-12 17:54:09 either wait 4.9.1 or cherry-pick some fixes 2014-07-12 18:00:41 i think we stay on gcc 4.8 for now 2014-07-12 18:00:49 would rather push to move to 100% clang 2014-07-12 18:01:08 good idea 2014-07-12 18:01:11 I love llvm 2014-07-12 18:01:28 yeah, clang is getting into usable state. 2014-07-12 18:01:36 and it has *way* better cross-compile support 2014-07-12 18:02:00 blueCmd_: is it possible to get openrisc hardware ? 2014-07-12 18:11:58 kaniini: it runs on FPGAs, so that's whatever FPGA board that you'd like 2014-07-12 18:12:53 is it fast on fpga? 2014-07-12 18:13:21 back in the old days, debian had an m68k/amiga port 2014-07-12 18:13:30 it was not fast enough to keep up with distribution changes 2014-07-12 18:13:45 this made release engineering of debian take weeks 2014-07-12 18:13:45 kaniini: #define fast :) 2014-07-12 18:13:57 blueCmd_: fast enough to keep up with the other builders 2014-07-12 18:14:13 currently we're not building on it, we're building in qemu-user chroots 2014-07-12 18:14:37 but with SMP coming along I think it would be fast enough 2014-07-12 18:26:35 fabled: yeah, let's wait for point release at least 2014-07-12 18:32:59 and for me its no problem with the clang switch, when you take ada out. I am using docker now :-) 2014-07-13 01:39:53 shafire: we will probably keep gcc around for specific packages requiring it like freebsd does 2014-07-13 13:49:02 I would like to request a package for Alpine Linux, is this the correct place? 2014-07-13 13:51:54 fnhail, we prefer that you'd file a bug in the issue tracker in bugs.alpinelinux.org 2014-07-13 13:52:11 All right, I will do that. 2014-07-13 13:52:14 Thank you. 2014-07-13 14:01:11 Done: http://bugs.alpinelinux.org/issues/3161 2014-07-13 15:09:57 Is Alpine Linux for ARM possible? 2014-07-13 15:12:22 maisal, we have arm build of edge 2014-07-13 15:12:30 bot not many boot images yet 2014-07-13 15:12:39 Nice 2014-07-13 15:13:03 latest for rasperry pi is at: http://dev.alpinelinux.org/~tteras/alpine-edge-140529-armhf.rpi.tar.gz 2014-07-13 15:13:08 What about packages? Even less than the few for x86 and x86_64, I guess? 2014-07-13 15:13:10 and i'm using also wandboard 2014-07-13 15:13:25 I was thinking about buying an ARM Chromebook 2014-07-13 15:13:29 its 100 or so less than x86 2014-07-13 15:13:39 Heh, not that much less. 2014-07-13 15:13:40 Nice 2014-07-13 15:13:51 so it's still work-in-progress 2014-07-13 15:13:58 but already usable on few chips 2014-07-13 15:14:08 bbl 2014-07-13 16:10:36 Looks like LibreSSL has already been released: http://ftp.openbsd.org/pub/OpenBSD/LibreSSL/ 2014-07-13 16:31:56 maisal, yes, we saw that. openssl folks are also stepping up activity. so we'll probably wait a bit and see what happens. 2014-07-13 16:32:09 Good news 2014-07-13 16:32:15 libressl is likely not abi compatible, so full rebuild would be needed if we switch 2014-07-13 16:32:26 but it's an option for alpine 3.1 2014-07-13 22:58:44 we would be making some changes to libressl too 2014-07-14 06:36:26 why switching for libressl (maybe in 3.1) is an option at all, same folks meintain it right, whats diff then portable version? 2014-07-14 07:00:01 crow, ? 2014-07-14 07:10:16 ups wrong window :) 2014-07-14 07:10:44 actually not, here was also talked about libressl 2014-07-14 07:11:41 yes, we talked it here too.. 2014-07-14 07:17:08 well my question is why an switch to libressl is needed? 2014-07-14 07:23:45 i dont know if it is *needed* 2014-07-14 07:23:52 but i think its worth considering 2014-07-14 07:24:22 and the reason we consider it is that openssl code quality has not been good enough 2014-07-14 07:27:52 is it not maintened from same team? 2014-07-14 07:29:55 nope 2014-07-14 07:30:04 http://www.tedunangst.com/flak/post/origins-of-libressl 2014-07-14 07:34:01 some of the things libressl have done: http://en.wikipedia.org/wiki/LibReSSL#Changes 2014-07-14 07:36:05 personally, i think we might want to switch if it's not too complicated 2014-07-14 07:36:15 even though openssl people have waken up and are fixing things 2014-07-14 07:36:29 the fact that in libressl they've deleted the 'bad' code sounds a whole lot better to me 2014-07-14 07:36:46 but as said, need to wait a bit, and the evaluate benefit/work 2014-07-14 07:45:48 yup, thats what i think too 2014-07-14 07:45:57 deleting lots of code sounds nice 2014-07-14 07:46:22 and we probably want wait a bit before we switch 2014-07-14 07:47:16 yes 2014-07-14 07:57:15 ok then my false, i was thinking its from same time 2014-07-14 07:58:25 what about boringssl http://it.slashdot.org/story/14/06/21/1126212/google-forks-openssl-announces-boringssl 2014-07-14 07:59:35 we saw that also 2014-07-14 07:59:41 it seems google just maintains a set of patches 2014-07-14 07:59:49 instead of trying to actually audit and clean up the code 2014-07-14 07:59:58 from that point libressl looks better candidate 2014-07-14 08:00:02 but time will show 2014-07-14 08:00:08 even openssl people have started to clean up things now 2014-07-14 08:00:25 they've been going through the rt backlog already to a decent number 2014-07-14 08:11:41 well libressl looks promising, only time will tell 2014-07-14 08:22:54 i'm looking at http://bugs.alpinelinux.org/issues/3158 2014-07-14 08:23:01 the rsync configure.ac test for ipv6 checks for __GLIBC__ 2014-07-14 08:23:02 sigh 2014-07-15 04:34:40 fabled: so i was thinking 2014-07-15 04:35:00 fabled: why don't we just write a stub library for libcrypto and libssl using libsodium 2014-07-15 04:35:17 fabled: then we could be done with openssl in it's entirety, crummy license, bad code and all 2014-07-15 05:08:00 looks interesting option 2014-07-15 05:08:08 kaniini, what libsodium adds to NaCl ? 2014-07-15 05:08:41 it just seems to have some portability gunk 2014-07-15 05:09:06 looking at libcrypto, there's not _many_ symbols that are public 2014-07-15 05:09:09 it seems doable 2014-07-15 05:09:22 as a first step we could replace libcrypto and then run openssl's tls code ontop of it 2014-07-15 05:10:07 after 5mins i kinda like NaCl... a lot. 2014-07-15 05:10:45 would be cool if we could do gnutls api also with it. we'd then have minimal footprint when merging code using different APIs 2014-07-15 05:10:46 i bet we could get some additional help from #musl etc. too 2014-07-15 05:10:50 so we wouldn't have to go it alone 2014-07-15 05:11:08 but seems NaCl is not as comprehensive 2014-07-15 05:11:14 not all those exotic ciphers 2014-07-15 05:11:20 yeah we might have to just write some of it ourselves 2014-07-15 05:11:22 not even edsa 2014-07-15 05:11:38 ecdsa* 2014-07-15 05:12:21 my keyboard's 'c' is starting to not function. i'm supposed to get my new laptop this week... 2014-07-15 05:12:58 is it an apple? 2014-07-15 05:13:06 on my previous equipment, my 8 key failed 2014-07-15 05:13:22 that was very annoying since 8 is * on a US keyboard 2014-07-15 05:14:57 I'm on Dell. 2014-07-15 05:15:23 the current one is _old_. i'm getting an XPS 13, Developer version (Ubuntu pre-installed) 2014-07-15 05:15:27 ah. 2014-07-15 11:38:51 mosquitto 1.3.2 released 2014-07-15 12:12:19 nice 2014-07-15 12:12:22 i'll bump that 2014-07-16 06:08:43 lwn.net: OpenSSL fork LibreSSL is declared “unsafe for Linux” 2014-07-16 06:09:05 http://arstechnica.com/security/2014/07/only-a-few-days-old-openssl-fork-libressl-is-declared-unsafe-for-linux/ 2014-07-16 06:09:46 at least it's getting looking-eyes and scrutiny 2014-07-16 06:09:47 :) 2014-07-16 06:11:34 i still like the fact that one of the goals are to reduce bloat 2014-07-16 06:16:00 asi understand, its a problem in arc4random implementation? 2014-07-16 06:17:25 also, it looks like they handle the __GLIBC__ case correct: https://github.com/libressl-portable/portable/commit/32d9eeeecf4e951e1566d5f4a42b36ea37b60f35 2014-07-16 06:18:21 i still think the apple approach is interesting 2014-07-16 06:18:29 core framework, stubs for the crappy libraries 2014-07-16 06:19:47 why are they using arc4random anyway 2014-07-16 06:19:55 they should just use fortuna on the portable version 2014-07-16 06:24:43 i think even if we adopt libressl, we should switch the PRNG to something better than arc4random. 2014-07-16 06:26:55 oh nevermind 2014-07-16 06:27:00 it's not actually arc4random 2014-07-16 06:27:02 weird. 2014-07-16 06:29:24 i like what I have seen from libressl so far 2014-07-16 06:33:19 to be fair, i think we were all expecting libressl 0day on the first couple of days of release 2014-07-16 06:34:10 you don't do massive refactoring of a codebase which is about as readable as hieroglyphics without there being some things 'lost in translation' 2014-07-16 06:34:25 that was discovered when people tried to refactor ircd 2.8 2014-07-16 06:35:24 yes 2014-07-16 06:36:33 https://github.com/libressl-portable/openbsd/commit/318450a310d065098b8dc6178302f18775878f1d 2014-07-16 06:36:46 at least this junk was removed 2014-07-16 06:50:25 fabled: we watch a little more. 2014-07-16 06:50:33 fabled: i will however, start preparing an APKBUILD. 2014-07-16 06:50:42 kaniini, yes, let's wait and see how it goes 2014-07-16 06:50:49 i suppose libressl is the easier way out 2014-07-16 06:51:09 though, having something decent with proper api glue, would be better 2014-07-16 06:54:47 what i think we do is 2014-07-16 06:54:50 libressl for 'now' 2014-07-16 06:55:04 common cryptographic framework 'later' 2014-07-16 06:55:07 well... 'now' being 'in 6 months or so' 2014-07-16 06:55:21 well right 2014-07-16 06:55:27 3.1 isnt coming out until november 2014-07-16 06:55:28 but yes, that sounds like a good plan forme too 2014-07-16 06:55:48 maybe the openbsd people will rewrite every line of code 2014-07-16 06:55:52 hmm.. 2014-07-16 06:55:54 and remove all those stupid openssl license things 2014-07-16 06:56:13 the openssl license is really a nightmare i'd like to do away with 2014-07-16 06:56:13 well - i think we evaluate 3-4 weeks before 3.1-branch preparations 2014-07-16 06:56:20 if libressl looks good/stable enough, we can go with it 2014-07-16 06:56:22 yeah, sounds good. 2014-07-16 06:56:42 interestingly 2014-07-16 06:56:54 i think if openbsd produced a libressl with the same license of their other products 2014-07-16 06:56:56 libressl did not get the security bug info from openssl for the earlier set of CVEs 2014-07-16 06:57:08 but at least google (boringssl) and libressl team are exchanging info 2014-07-16 06:58:00 we would not need to replace our core cryptographic framework 2014-07-16 06:58:47 but right now, as you are aware, a GPL application using openssl without an explicit exception 2014-07-16 06:58:54 is a bad situation 2014-07-16 07:12:35 i wouldnt mind if we wait til alpine 3.2 before we decide to switch to libressl/boringssl 2014-07-16 07:19:38 +1 2014-07-16 08:09:36 clandmeter: could you verify if that syslinux update fixes your #3138? 2014-07-16 08:09:39 its in edge 2014-07-16 18:50:49 What if I have two packages with the same name on /media/usb/cache/ ... how can i know which one AL is using? 2014-07-16 18:51:02 names like: 2014-07-16 18:51:03 bash-4.2.037-r0.d3ffc70b.apk 2014-07-16 18:51:17 bash-4.2.037-r0.ea4fc7e5.apk 2014-07-16 18:52:46 is there any way to find out which one is being used on startup? 2014-07-17 08:31:30 hi@all 2014-07-17 08:31:46 question: why is quassel not available for 2.7? 2014-07-17 08:34:43 it was still in testing back then? 2014-07-17 08:34:52 and it was segfaulting iirc? 2014-07-17 08:35:26 ah it needs pax marking 2014-07-17 08:35:34 do you want me to backport quassel for v2.7? 2014-07-17 08:43:08 StarWarsFan: we develop things in edge and fixed things gets only with next release 2014-07-17 08:43:13 we can do backports on request 2014-07-17 08:43:31 and nobody requested a quassel backport ;) 2014-07-17 08:43:33 thats why 2014-07-17 08:43:42 i can push it to v2.7 if you want 2014-07-17 08:49:02 that would be great 2014-07-17 08:49:38 if i remember right, it should be released on 2.7 2014-07-17 08:49:57 but might be forgotten... 2014-07-17 08:49:59 never mind 2014-07-17 08:50:21 commit 0ec92a577cc67f7d51c9616bcc26e9e18bb8b41f 2014-07-17 08:50:21 Author: Natanael Copa 2014-07-17 08:50:21 Date: Wed Feb 5 12:06:53 2014 +0000 2014-07-17 08:50:21 main/quassel: moved from testing 2014-07-17 08:50:35 cool! thx a lot 2014-07-17 08:50:37 git log says it went to main in Feb 2014-07-17 08:50:45 so it was for v3.0 2014-07-17 08:50:49 but its backported now 2014-07-17 08:50:59 nice, thx a lot 2014-07-17 08:51:20 we're using it on 2.7 as 3.x is not an option at the moment 2014-07-17 08:52:04 i kinda feel i owed you that since i have not been able to fix nfs :-/ 2014-07-17 08:53:24 nfs is indeed one of our major issues 2014-07-17 08:54:26 and as long as this is not fixed i'm unable to build any other packages for 3.x 2014-07-17 08:54:47 because nfs is fundamental on our build nodes 2014-07-18 10:36:53 morning 2014-07-18 14:52:19 sry forgot new deps 2014-07-18 14:57:04 bad idea backporting 5 mins before leaving :-/ 2014-07-18 14:58:11 ncopa, any way to look at build log? 2014-07-18 14:58:16 URL 404 2014-07-18 14:59:38 http://bld2.alpinelinux.org/buildlogs/build-3-0-x86_64.v3.0.1-75-g958c25f.log 2014-07-18 14:59:49 http://bld2.alpinelinux.org/buildlogs/build-3-0-x86_64/ 2014-07-21 07:01:35 question: 2014-07-21 07:01:50 why is clamav not always build with the latest version from upstream? 2014-07-21 07:04:44 this is a package which should always be backported 2014-07-21 07:05:55 on 2.7-repo there is 0.98.1-r2 2014-07-21 07:06:09 but on 3.0-repo there is 0.98.4-r1 2014-07-21 07:06:27 and an up2date anti virus package is a must have... ;-) 2014-07-21 07:07:11 oh, i forgot something important: good morning@all :-) 2014-07-21 07:08:41 morning 2014-07-21 07:09:40 normally, for security updates, there is a CVE related 2014-07-21 07:10:07 when there is a CVE, we normally update 2014-07-21 07:10:18 but yeah, i guess you have a point 2014-07-21 07:23:02 ah i c 2014-07-21 07:23:33 but as i said, in my opinion, such a tool should be always up2date 2014-07-21 07:23:46 or with other words, 2014-07-21 07:24:01 every update of anti virus software is a security update ;-) 2014-07-21 07:32:09 clandmeter, r u around? What's the issue with xbmc on alpine 3? 2014-07-21 07:36:55 StarWarsFan: feel free to file an issue if you feel that some package should be updated 2014-07-21 07:37:12 ok, thx 2014-07-21 07:37:21 chances are big i forget if you only mention it here in irc 2014-07-21 07:37:36 thanks! 2014-07-21 07:37:41 yeah, have this problem too ;-) 2014-07-21 07:44:13 Morning 2014-07-21 07:45:11 morning 2014-07-21 07:57:48 ncopa: http://bugs.alpinelinux.org/issues/3197 2014-07-21 08:04:20 thanks! 2014-07-21 09:23:20 asus zenfone - has intel atom 2014-07-21 11:28:43 i think i will rename gstreamer to gstreamer0.10 2014-07-21 11:28:58 and gst-plugins-* to gst-plugins-*0.10 2014-07-21 11:29:36 i dont know if i bother rename gsteamer1 to gstreamer though 2014-07-21 11:30:14 i'll remove gnome deps too while at it 2014-07-21 11:30:22 gconf and gnome-vfs 2014-07-21 13:45:43 ncopa: I'm having a bit of trouble building a package because the source, a zip file, doesn't have a direct link 2014-07-21 13:45:55 have to browse to http://software.cisco.com/download/release.html?mdfid=283998771&flowid=&softwareid=282463187&release=1.3.3&relind=null&rellifecycle=null&reltype=null and think click on buton 2014-07-21 13:46:00 suggestions? 2014-07-21 13:46:08 s/buton/button/ 2014-07-21 13:46:42 I could manually place the file on dev.a.o/archives/ 2014-07-21 13:46:50 thats an option 2014-07-21 13:46:56 does the license allow that? 2014-07-21 13:46:57 but, can't automate with wget 2014-07-21 13:47:22 https://secure-global.esd.cisco.com/files/swc/sec/4_SDSP_758178_1387273043374/1/SPA112-SPA122_1.3.3_015_FW.zip?ip=84.202.203.3&dtrTag=4_SDSP_758178_1387273043374_e063938187dfd12cb57a8e6c36b5939c&userid=null&tenant-id=sdsp&__gda__=1406036781_15a8c694507e8c68980e772c4b5025f78558516c19823e56f22cc1e4d0b03fdb&__gdb__=exp=1405950444~hmac=03b1d21882ed9366eb76af662ea4a131dd6bda99c5fa0e91a9b9937dcf63fb88&xac=4-GQ1l4gcy0Ty73vewWs1Ff6WLoOD4Vhp0DKPJs% 2014-07-21 13:47:22 2FVwU0xiWtkX63cDfup%2F9nOX2lFf 2014-07-21 13:47:35 hm 2014-07-21 13:47:49 seems like they let the dl link expire 2014-07-21 13:47:55 ACTION doesn't know about the license 2014-07-21 13:48:47 check if we are allowed to repack it and re-distribute it 2014-07-21 13:49:03 seems they use lots of GNU licensed software in it 2014-07-21 13:49:05 http://www.cisco.com/c/dam/en/us/td/docs/voice_ip_comm/csbpvga/spa100-200/release/SPA100_OSD_78-20420-02.pdf 2014-07-21 13:49:16 GPL that is 2014-07-21 13:49:39 then you can request the soruce code from them 2014-07-21 13:50:05 yup 2014-07-21 13:50:11 and: "We protect your rights with two steps: (1) copyright the software, and 2014-07-21 13:50:11 (2) offer you this license which gives you legal permission to copy, 2014-07-21 13:50:11 distribute and/or modify the software." 2014-07-21 13:50:24 i suppose best alternative is to host it on dev.a.o 2014-07-21 13:50:40 but check if the actually license allows us to do that 2014-07-21 13:50:56 might be only parts of it is gpl 2014-07-21 13:52:56 ACTION is not a lawyer :/ 2014-07-21 13:53:17 the doc above mentions the licenses of various software used in the product 2014-07-21 13:53:26 don't see anything about the entire package 2014-07-21 13:54:22 i suppose all licenses mentioned needs to allow us to redistribute 2014-07-21 14:05:29 seems like they do 2014-07-21 14:05:32 ok, thanks ncopa 2014-07-21 16:01:07 afternoon 2014-07-21 16:01:51 i have alpine 3.0.0 and my repo is configured for edge. i did apk upgrade --update-cache --available but /etc/alpine-release still says 3.0.0 2014-07-21 16:01:58 should i expect 3.0.1? 2014-07-21 16:15:16 no. i think edge has 3.0.0 in it 2014-07-21 23:05:31 02:27:02 StarWarsFan | every update of anti virus software is a security update ;-) 2014-07-21 23:05:40 well our current policy does not really agree with that 2014-07-21 23:05:58 we could have some optional backports repository or perhaps something like the newer debian volatile repository 2014-07-22 07:15:31 fcolista: regarding xbmc, i dont know. i didnt try it. fabled looked at it some time ago and patched it. 2014-07-22 07:52:13 kaniini: thx for the info 2014-07-22 07:52:33 i think in case of clamav it might be a little bit special 2014-07-22 07:53:01 because there are quite often updates necessary if they change their internals 2014-07-22 07:53:35 so if the internal anti virus engine gets updated, 2014-07-22 07:53:44 its an update of the entire tool 2014-07-22 08:13:38 StarWarsFan: I updated it in v2.7 branch 2014-07-22 08:13:59 i can do so in future too on request, but i will not monitor clamav specifically 2014-07-22 08:14:27 ncopa: yeah, i've seen it. thx a lot! 2014-07-22 08:14:41 thanks for your feedback :) 2014-07-22 08:15:13 no problem, if this way is ok for you, we can do it like that 2014-07-22 09:18:42 firefox-31.0 is released and is an ESR (lontime) release 2014-07-22 09:18:57 i think we should stick to it til v3.1 atleast 2014-07-22 12:57:19 I wanted to use Chromium on Alpine Linux, so after poking around, I found it in the "unmaintained" repository 2014-07-22 12:57:35 Does that mean we won't get updates? 2014-07-22 12:58:10 it meanst that it currently does not build 2014-07-22 12:58:46 But is somebody working on it? 2014-07-22 12:59:00 not that i knkow 2014-07-22 12:59:05 What a pity 2014-07-22 12:59:15 At least we have firefox 2014-07-22 12:59:18 if its unmaintained it basically means nobody works on it 2014-07-22 12:59:32 I hope someone does, one day 2014-07-22 12:59:33 there used to be some serious issues with chromium 2014-07-22 12:59:43 Really? Like what? 2014-07-22 13:01:14 http://ostatic.com/blog/making-projects-easier-to-package-why-chromium-isnt-in-fedora 2014-07-22 13:02:04 https://code.google.com/p/chromium/issues/detail?id=28287 2014-07-22 13:02:06 That's from 2009, has nothing changed? 2014-07-22 13:02:11 it has 2014-07-22 13:02:19 they are slowly working on it apparently 2014-07-22 13:02:24 Nice 2014-07-22 13:02:34 but thats the reason we gave it up 2014-07-22 13:03:49 the https://code.google.com/p/chromium/issues/detail?id=28287 has the exact status and details 2014-07-22 13:04:13 I will keep an eye on that, then 2014-07-22 13:04:22 Hope it's solved soon. Thanks 2014-07-22 13:07:12 Another question, as Alpine isn't a rolling release, should I always update when there's a new version? Or is just updating once every 6 months or once a year fine? 2014-07-22 13:07:36 we provide a rolling branch, 'edge' 2014-07-22 13:07:52 each 6 months we cut a release from edge 2014-07-22 13:08:07 which we maintain for 2 years (thats the goal atleast) 2014-07-22 13:08:47 So, if I install a stable release, I should be good to go for 2 years? 2014-07-22 13:08:51 we normally only do fixes for the latest stable 2014-07-22 13:08:57 and sec fixes only after that 2014-07-22 13:08:59 yes 2014-07-22 13:09:02 Nice. 2014-07-22 13:09:08 you should be good for 2 years 2014-07-22 13:09:24 What about 'edge'? Is it recommended to use it or are there too many bugs? 2014-07-22 13:09:41 edge is good forever, if I get it right? 2014-07-22 13:09:45 i use it on my desktop 2014-07-22 13:09:54 and on my laptop 2014-07-22 13:10:08 once in a while things break, but hopefully not too often 2014-07-22 13:10:13 we try avoid breakages 2014-07-22 13:10:20 and fix them normally quick 2014-07-22 13:10:30 for desktop usage i'd say edge is ok 2014-07-22 13:10:53 Thank you, nice to hear that 2014-07-22 13:11:02 specially if you are interested in contribute back 2014-07-22 13:11:30 I wish I was capable of contributing 2014-07-22 13:19:33 what's your advice for monitoring updates to source code for the packages you are maintaining? 2014-07-22 13:20:29 suscribe to some rss feed or announce mailing list 2014-07-22 14:09:50 yeah, i was just wondering if there was a tool to centralise all the different methods that i haven't come across 2014-07-22 14:16:41 i have a project on that 2014-07-22 14:17:10 http://git.alpinelinux.org/cgit/ncopa/upstream-monitor/ 2014-07-22 14:17:27 but it currently only checks gnome, github and archlinux 2014-07-22 14:17:59 would be nice if we could add a hint in APKBUILD how to look for updated version 2014-07-22 14:24:47 What's Alpine's command shell? 2014-07-22 14:24:57 busybox 2014-07-22 14:25:06 ash 2014-07-22 14:25:24 Does ash support auto-completion? 2014-07-22 14:25:32 no 2014-07-22 14:25:45 sorry akoma 2014-07-22 14:25:46 What a pity 2014-07-22 14:25:49 wrong window 2014-07-22 14:25:58 yes, ash supports autocompletion 2014-07-22 14:26:04 Oh, nice! 2014-07-22 14:26:11 You got me worried for a while 2014-07-22 14:26:24 i was writing on another window :) 2014-07-22 14:26:25 sorry 2014-07-22 14:26:31 it does not do the smart bash style autocompletion 2014-07-22 14:26:42 but it autocomplete files 2014-07-22 14:26:52 and paths 2014-07-22 14:27:08 which imho is better than the full 'smart' auto complete 2014-07-22 14:27:19 What about packages? On arch, I can do something like: 'pacman -S fire' and it goes like 'pacman -S firefox' 2014-07-22 14:29:05 Well, I don't think that's too important, it's not like I'm going to install too many packages 2014-07-22 14:31:00 no, it cannot 2014-07-22 14:31:44 ncopa, i'm playing with a "mypackages.lua" script, so i can monitor what packages i have...this is the script: http://pastie.org/9412296 so far, i don't know how manage the "nil" values. 2014-07-22 14:32:32 i mean: on multiple string.match value 2014-07-22 14:32:45 how can i print the values that are not nil ? 2014-07-22 14:33:09 should i do "if variable then print (variable) end" for every variable? 2014-07-22 14:34:01 print(foo, (variable or "n/a")) 2014-07-22 14:34:12 if variable is nil it will print "/na" 2014-07-22 14:34:16 "n/a" 2014-07-22 14:34:36 oh 2014-07-22 14:34:42 fine 2014-07-22 14:35:16 but pkgver might be $_something 2014-07-22 14:35:29 so you actually need to execute the APKBUILD in a shell 2014-07-22 14:35:37 i have a lua-aports module 2014-07-22 14:35:50 which will load the entire aports db 2014-07-22 14:35:57 and give you intrators 2014-07-22 14:36:01 iterators 2014-07-22 14:36:39 http://git.alpinelinux.org/cgit/lua-aports/ 2014-07-22 14:38:45 ok, installed 2014-07-22 21:14:13 ncopa, I see you are the maintainer of scrot 2014-07-22 21:14:19 Could I make a suggestion? 2014-07-22 21:15:35 A friend of mine rewrote scrot, since the original version was convoluted as hell and, according to him, "Gnome level shit" 2014-07-22 21:15:51 His version compiles even with -Wall -pedantic -Werror 2014-07-22 21:16:05 Check out for yourself and tell me what you think: https://github.com/mahkoh/scrot 2014-07-22 21:16:31 I think it'd be greatly beneficial to have that scrot on Alpine instead of the current one 2014-07-22 21:41:40 Hey again. Lost connection 2014-07-23 01:32:40 ncopa, ping 2014-07-23 15:17:54 hi 2014-07-23 15:18:06 will be setup-bootable (with syslinux) be okay in 3.0.2? 2014-07-23 15:29:01 -be 2014-07-23 16:08:44 I am having an issue with Alpine's installation and posted about it at #alpine-linux, but nobody answered. Can I post it here too? I'm stuck. 2014-07-23 16:10:08 do you use setup-bootable? 2014-07-23 16:10:49 I am trying to install Alpine Linux through a CD, but it says "Mounting boot media failed""and throws me at an emergency shel 2014-07-23 16:10:57 I searched a little and, apparently, I have to edit syslinux.cfg, is that so? 2014-07-23 16:11:03 Where can I find that file? I am currently at the emergency shell 2014-07-23 16:19:38 Another question is the edge branch the only one using musl or is 3.0.0 as well? 2014-07-23 16:20:08 snce 3 2014-07-23 16:20:12 since 3 2014-07-23 16:20:36 Right, so I will just stick to 3.0.0 since it's stable 2014-07-23 16:20:43 If I manage to install it, that is... 2014-07-23 16:20:54 try to use a usb stick instead? 2014-07-23 16:21:06 I don't have one right now 2014-07-23 16:21:45 And isn't CD installation almost always preferred? I don't know why it's different with Alpine 2014-07-23 23:13:40 StarWarsFan: I am getting the following during boot: 2014-07-23 23:13:44 wlan0 . . . 2014-07-23 23:13:54 Error; this script should be called from udhcpc [ !! ] 2014-07-23 23:14:02 Did you find a proper way to fix that? 2014-07-24 07:37:47 we've several package in main without maintainer 2014-07-24 07:37:56 *packages 2014-07-24 08:21:38 sounds bad 2014-07-24 08:57:24 yes 2014-07-24 09:00:57 how do i list the files provided by an installed package? 2014-07-24 09:02:03 apk info -L $package 2014-07-24 09:02:53 thanks 2014-07-24 09:02:57 yw 2014-07-24 09:03:09 i had just found info -a 2014-07-24 09:03:18 apk info --help 2014-07-24 09:03:23 has a lot of useful options 2014-07-24 09:03:56 ah i see! 2014-07-24 09:04:12 apk COMMAND [-h|--help] 2014-07-24 09:04:24 that's not obvious at first 2014-07-24 09:04:45 apk -h doesnt show -L or -a for info 2014-07-24 09:05:22 apk help shows that several commands are available 2014-07-24 09:05:29 (other than options) 2014-07-24 09:05:49 info is among the commands 2014-07-24 09:05:59 and apk command -h isn't intuative in a linux shell 2014-07-24 09:06:19 and i haven't seen a man page 2014-07-24 09:08:00 yes, there are no manpages atm for apk 2014-07-24 09:08:56 ScrumpyJack: "apk command -h isn't intuative in a linux shell" i disagree so many commands have the same semantics, that it is very obvious. 2014-07-24 09:09:28 right 2014-07-24 09:09:34 git uses the same semantic 2014-07-24 09:09:40 openssl as well 2014-07-24 09:10:12 ScrumpyJack, is only matter of reading carefully the output of --help ;-) 2014-07-24 09:13:51 can i have more exemples of commands? that's two ;) 2014-07-24 09:14:46 actually its three (you ignore apk) there's more, i think these are more widely known. 2014-07-24 09:16:42 iproute2, subversion 2014-07-24 09:17:05 those are the ones coming in my mind.. 2014-07-24 09:17:28 brctl 2014-07-24 09:19:19 to me, it's intuative on a swtich or router, but on the linux cammand line, my reflexes are moo --help and man moo :) 2014-07-24 09:20:15 adapt. 2014-07-24 09:29:28 is there a man page for apk? 2014-07-24 09:29:56 apk --help is for me quite enough 2014-07-24 09:33:04 or rather apk command --help ;) 2014-07-24 09:33:29 can awk returns two variables that i can reuse? Something like that: 2014-07-24 09:33:32 cat test.list | grep -A1 Download | awk -F":" '{down=$1; up=$2}' 2014-07-24 09:33:54 and i have $down and $up that i can use 2014-07-24 09:36:36 how about down=`cat test.list | grep -A1 Download | awk -F":" '{print $1}` 2014-07-24 09:36:55 no 2014-07-24 09:37:01 i don't want to repeat the command 2014-07-24 09:37:12 ok, found the way 2014-07-24 09:37:35 cat test.list | grep -A1 Download | awk -F":" '{print $1 $2}' | while read down up do echo $down $up done 2014-07-24 12:33:47 StarWarsFan: ping 2014-07-24 12:36:51 I am getting the same error you did some months ago: 2014-07-24 12:36:57 wlan0 . . . 2014-07-24 12:36:58 Error; this script should be called from udhcpc [ !! ] 2014-07-24 12:37:16 Did you find a proper fix for it? 2014-07-24 13:54:11 ehuwa: http://bugs.alpinelinux.org/issues/3105 2014-07-24 13:54:51 ncopa: Someone pointed me to that link yesterday, but I couldn't fix it. 2014-07-24 13:55:09 basically, it means you didnt get an ip addr 2014-07-24 13:55:27 I added the udhcpc_opts -t 20 line to /etc/network/interfaces under wlan0, but it only takes longer to give me the error 2014-07-24 13:55:40 -t 4 -T 6 -b didn't work either 2014-07-24 13:56:35 are you sure you have link? 2014-07-24 13:56:45 is your dhcp server working? 2014-07-24 13:56:56 it could be something with wpa_supplicant 2014-07-24 13:57:04 does your nic need firmware? 2014-07-24 13:57:07 what does dmesg say? 2014-07-24 13:58:14 I get network access through ethernet, but haven't tried a wireless connection yet, since it gives me this error on wlan0 2014-07-24 13:58:56 Maybe it;s a missing firmware that gives me this error? 2014-07-24 14:01:47 lspci returns only some lines of the form "00:00.0 Class 0600: 8086:0044" 2014-07-24 14:02:16 apk add pciutils 2014-07-24 14:02:26 wlan0 is wireless 2014-07-24 14:02:44 did you configure ssid and password etc? 2014-07-24 14:02:52 you need set up wpa_supplicant 2014-07-24 14:03:29 If I do that, will the boot error stop? 2014-07-24 14:03:35 If I set up a wireless network? 2014-07-24 14:03:44 likely yes 2014-07-24 14:03:57 you could also comment out the auto wlan0 in /etc/network/interfaces 2014-07-24 14:04:42 Well, I will need to use wireless eventually 2014-07-24 14:04:47 I will try to get it working 2014-07-24 14:04:52 Looking into nmcli-dmenu right now 2014-07-24 14:07:21 There's no network-manager package? 2014-07-24 14:07:58 hi 2014-07-24 14:09:41 I have to go, be back soon 2014-07-24 14:10:38 for which date is the 3.0.2 release scheduled? 2014-07-24 14:11:35 i was hoping to get it out today 2014-07-24 14:11:39 or this week 2014-07-24 14:11:50 will likely not be today 2014-07-24 14:12:00 but hopefully tomorrow or eraly next week 2014-07-24 14:12:35 the biggest problem is that i have 30C in my office now so things are working slow... 2014-07-24 14:12:53 understandable 2014-07-24 14:13:21 did anyone try: http://bugs.alpinelinux.org/issues/3137? 2014-07-24 14:13:35 ncopa if i remmber correctly you could reproduce it? http://bugs.alpinelinux.org/issues/3107 2014-07-24 14:13:46 i did some tests but it would be nice if someone could verify that setup-bootable works from edge 2014-07-24 14:14:04 oh 2014-07-24 14:14:07 thats true 2014-07-24 14:14:09 this one is [Alpine Linux - Bug #3107] (New) lvm creation mlock error 2014-07-24 14:14:22 that does actually not make things break 2014-07-24 14:14:28 oh ok i could post just bug nummber 2014-07-24 14:14:56 here on my test system was not bootable 2014-07-24 14:15:05 i could test it next week 2014-07-24 14:15:11 again* 2014-07-24 14:15:17 i believe that was due to something else 2014-07-24 14:15:40 i did fix a bunch of things though 2014-07-24 14:15:44 in setup-disk iirc 2014-07-24 14:16:03 ncopa: was #3137 a bug since 2.x? 2014-07-24 14:16:21 yes 2014-07-24 14:16:40 seems like it has not really manifested itself until v3.0 2014-07-24 14:16:42 what changed that we could "see" that bug? 2014-07-24 14:16:46 i dont know 2014-07-24 14:16:59 its either something musl libc related or its kernel 2014-07-24 14:17:25 i belive the flushing of the filesystem happens later now 2014-07-24 14:17:53 what have you changed to fix the bug? I only see a patch deleted: http://git.alpinelinux.org/cgit/aports/commit/?id=83897f40b280e7949b2c3b95512903d61b114edb 2014-07-24 14:18:16 yes, the fixes are in alpine-conf 2014-07-24 14:18:27 http://git.alpinelinux.org/cgit/alpine-conf/log/ 2014-07-24 14:18:37 http://git.alpinelinux.org/cgit/alpine-conf/commit/?id=5856a288a732b2a5de650983854d684e6dd81c55 2014-07-24 14:18:41 thats the real fix 2014-07-24 14:19:26 basically, we need unmount the filesystem before we can run syslinux 2014-07-24 14:20:12 I see :-) 2014-07-24 14:20:41 so it is just me that have been stupid... :-/ 2014-07-24 14:57:33 ncopa thanks :) 2014-07-24 16:00:28 Does anyone know how to get a driver for the wireless card BCM4313 on Alpine? 2014-07-24 16:34:12 Power and out, I am back. Did anyone answer about BCM4313? 2014-07-24 16:35:16 ehuwa: irclogger.com :-) 2014-07-24 16:35:49 Nice. Thanks 2014-07-24 16:36:45 Seems I'm out of luck so far. 2014-07-24 16:37:08 Where are you from? 2014-07-24 16:37:26 Why do you ask? 2014-07-24 16:37:41 Which timezone do you have? 2014-07-24 16:38:11 It's 13:38 here 2014-07-24 16:38:34 most people talk here between 9-17 UTC 2014-07-24 16:38:43 Oh, I see 2014-07-24 16:38:52 Will keep that in mind, thank you 2014-07-24 16:39:45 so, you are from brazil? 2014-07-24 16:39:57 Yes, you /whois'd me? 2014-07-24 16:40:08 Your timezone is unique for Brazil :p 2014-07-24 16:40:20 hehe 2014-07-24 16:41:00 your whois says the same :p 2014-07-24 16:41:36 I will fix that soon, just poking around Alpine for now 2014-07-24 16:41:56 Do you like it? 2014-07-24 16:42:33 I quite love the concept of simplicity and straight-forwardness 2014-07-24 16:42:48 I like it that it runs fully in ram from my usb :-) 2014-07-24 16:42:52 I am using alpine since 2.6 and now in 3.x with docker 2014-07-24 16:43:05 dropped kvm for docker 2014-07-24 16:43:10 I waited for the musl migration to try it out 2014-07-24 16:43:20 Installed yesterday and am playing around 2014-07-24 16:43:37 Only things to be fixed for me are wireless and font rendering 2014-07-24 16:44:20 docker seems nice, I didn't know it 2014-07-24 16:46:17 I'm not too knowledgeable on all this stuff, but I'm using it mostly due to the concept and ideas Alpine employ 2014-07-24 16:46:43 I used Arch before, but they've gone astray of their original premises since quite some time 2014-07-24 16:48:02 Hopefully I can get everything working and just lay back for 1 year or so, until I install a new version 2014-07-24 16:48:55 I gotta go to the movies, see you later 2014-07-24 16:49:08 have fun 2014-07-24 16:49:10 Thanks 2014-07-24 16:49:19 And just reposting, in case anyone comes: 2014-07-24 16:49:20 Does anyone know how to get a driver for the wireless card BCM4313 on Alpine? 2014-07-24 21:38:58 I am trying to install the b43 driver 2014-07-24 21:39:01 I did "First, apk add alpine-sdk, then git clone aports from git.alpinelinux.org, switch to the aports/non-free/b43-firmware folder, then run abuild -r. Install the generated apk file. Run fwcutter, and you should be good to go." 2014-07-24 21:39:08 Worked as intended, it seems 2014-07-24 21:39:38 But lspci says "Kernel driver in use: bcma-pci-bridge" 2014-07-24 21:40:24 I tried adding "blacklist bcma" to /etc/modprobe.d/blacklist.conf but even then bcma-pci-bridge keeps being used instead of b43 2014-07-24 21:40:27 Any ideas? 2014-07-24 21:43:03 Firefox crashed, sorry. I'm back 2014-07-25 07:43:41 files from v2.5.4-329-g690b96b uploaded 2014-07-25 08:05:27 morning 2014-07-25 12:45:47 what do folks use to record cds in alpine? i can't find any tools such as mkisofs, wodim or cdrecord etc 2014-07-25 12:46:31 ok, found cdrkit 2014-07-25 12:47:43 it has'em all! yay! i'm spoilt for choice 2014-07-25 13:48:37 i have used xfburn... 2014-07-25 13:59:38 ncopa: what's your timezone? 2014-07-25 14:00:19 ScrumpyJack: Europe/Oslo 2014-07-25 14:17:46 yesterday i tried to setup pcsc-lite, and i packaged ccid for it. but i have some weird problem. i get: ERror relocating /usr/lib/libfl.so.2: yylex: symbol not found 2014-07-25 14:18:16 and indeed if i nm -D the two ccid libraries none of them has much exported symbols 2014-07-25 14:21:00 (i also submitted two pcsc-lite related bugs) 2014-07-25 15:22:43 compiled tcsh, but when i run it, it core dumps. anyone familiar with this? 2014-07-25 15:22:46 http://sprunge.us/YSeF 2014-07-25 15:23:22 strace: http://sprunge.us/LjgX 2014-07-25 16:20:04 apparently libfl.so.2 is a dependency of libccid and libfl expects to have a yylex function available 2014-07-25 17:58:56 what are the desktop environments currently in aports? i see gnome and xfce for now. any others? 2014-07-25 18:22:43 huh, a firefox fork that removes the guifuckery and most importantly the DRM: http://www.palemoon.org/info.shtml 2014-07-25 19:24:14 hi@all 2014-07-25 19:24:32 is there any chance to bring openjdk7 to 2.7? 2014-07-25 20:43:44 hm, compiling openjdk7 on 2.7 failed :-( 2014-07-25 20:51:56 so i opened https://bugs.alpinelinux.org/issues/3242 2014-07-25 20:57:00 just took a look on the roadmap on https://bugs.alpinelinux.org/projects/alpine/roadmap 2014-07-25 20:57:36 how about 2.6.7 and 2.5.5? 2014-07-25 20:58:09 ok, need some sleep now... 2014-07-25 20:58:11 c ya 2014-07-26 07:01:52 kaniini: straight to main? 2014-07-26 07:02:13 sure 2014-07-26 07:02:23 it meets all of the requirements: 2014-07-26 07:02:24 1. it works 2014-07-26 07:02:28 2. it's got a maintainer 2014-07-27 07:42:39 hi@all 2014-07-27 07:42:59 some weeks ago i run into a strange problem with my alpine lxc containers 2014-07-27 07:43:40 (if one is interested in the background, see https://issues.jenkins-ci.org/browse/JENKINS-22651) 2014-07-27 07:43:47 but the current question is: 2014-07-27 07:44:09 why is /dev/null on the lxc containers set to root:root 644? 2014-07-27 07:44:26 this results in the fact, that scp with non-root did not work 2014-07-27 07:44:40 jenkins@flieger:~$ scp -i .ssh/builder07 tmp/foobar.txt alpine_2_7_x86:~/ 2014-07-27 07:44:40 Couldn't open /dev/null: Permission denied 2014-07-27 07:44:40 lost connection 2014-07-27 07:44:40 jenkins@flieger:~$ 2014-07-27 07:45:00 after changing /dev/null to 666 everything works as expected 2014-07-27 07:45:42 any ideas? 2014-07-27 21:35:00 n8 @all 2014-07-28 07:42:09 anyone else getting xorg freeze with grsec resource overstep of RLIMIT_NICE xinit on edge? 2014-07-28 07:42:49 requesting 21 for RLIMIT_NICE against limit 0 2014-07-28 08:06:22 uggedal: i heard someone report that Xorg freezes 2014-07-28 08:06:39 it sounds like xorg-1.16 introduced something 2014-07-28 08:07:11 the only issue I have atm with xorg is my intel chipset driver that wants to access /sys 2014-07-28 08:07:40 http://forums.grsecurity.net/viewtopic.php?f=3&t=4012 2014-07-28 08:08:15 so every reboot I have to: find /sys -type d -exec chmod o+rx {} \; 2014-07-28 08:09:44 uggedal: could you please file a bug on bugs.a.o with the details? 2014-07-28 08:10:19 ncopa: yes, debugging it a bit first 2014-07-28 08:11:18 I can't find any setpriority() or nice() usage in xorg-server, maybe it's in a driver 2014-07-28 08:11:28 probably yes 2014-07-28 08:11:48 my guess is your xf86-video-* driver 2014-07-28 08:11:53 and i bet its not intel 2014-07-28 08:12:05 maybe nvidia? 2014-07-28 08:16:04 vesa 2014-07-28 08:16:09 vbox guest 2014-07-28 08:17:01 and I could not find any setpriority() usage in any of my installed xf86-* packages 2014-07-28 08:18:19 StarWarsFan|afk: try this http://sprunge.us/hIcZ (untested) 2014-07-28 08:23:08 ScrumpyJack: thx 2014-07-28 08:23:19 morning@all 2014-07-28 08:24:04 ScrumpyJack: ncopa: what do you think, should i file an issue on b.a.o for http://sprunge.us/hIcZ ? 2014-07-28 08:24:23 i think this should be fixed on the lxc setup 2014-07-28 08:24:24 did it fix /dev/null? :) 2014-07-28 08:24:41 !? we dont create /dev/null? 2014-07-28 08:24:53 as i wrote above, i did this by hand and it works 2014-07-28 08:25:09 ncopa: right, but it should be created 2014-07-28 08:25:31 maybe apk creates it 2014-07-28 08:26:06 on a bare metal machine everything is ok 2014-07-28 08:26:27 but on an lxc container /dev/null has wrong permission 2014-07-28 08:26:53 lxc-create -t alpine -n a1 2014-07-28 08:26:59 there must be a change sometime in the past arround the date i've created the jenkins issue mentioned above 2014-07-28 08:27:17 ~ $ ls -l /var/lib/lxc/a1/rootfs/dev/null 2014-07-28 08:27:17 crw-rw-rw- 1 root root 1, 3 Jul 28 10:25 /var/lib/lxc/a1/rootfs/dev/null 2014-07-28 08:27:23 because it was working since i've setup the evironment until then 2014-07-28 08:27:46 hm, interesting 2014-07-28 08:28:50 ok, if it is working with a fresh created container, then there must have changed the permission in the past 2014-07-28 08:28:54 but then it's ok 2014-07-28 08:28:57 :-) 2014-07-28 08:29:20 i will verify this too 2014-07-28 08:29:33 yes, please do that 2014-07-28 08:29:40 its trivial to test 2014-07-28 08:29:59 /dev/null has the right default perms for me too 2014-07-28 08:30:36 i think we have had issues with /dev/null in the past 2014-07-28 08:31:03 and then my problem was a result of them 2014-07-28 08:31:07 so ti might be a regression that get triggered by something 2014-07-28 08:31:10 need some time to figure it out... 2014-07-28 08:31:18 ncopa: can you remind me how to create a nice patch with git for a new package? (i'll tatoo it to my forehead this time) 2014-07-28 08:32:19 git add ../testing/$d 2014-07-28 08:32:19 git commit -m "testing/$d: new aport 2014-07-28 08:32:19 $pkgdesc 2014-07-28 08:32:19 $url 2014-07-28 08:32:19 " 2014-07-28 08:32:36 should be a blank line before $pkgdesc 2014-07-28 08:33:29 i suppose we should have an example of a good commit message in wiki 2014-07-28 08:33:55 hmm, the setpriority is done by xinit itself and is not the reason for the crash 2014-07-28 08:33:57 i've committed, but that didn't create a patch 2014-07-28 08:34:21 (that i can see) :) 2014-07-28 08:34:44 ScrumpyJack: then git format-patch -1 2014-07-28 08:34:56 or git format-patch HEAD^ 2014-07-28 08:35:17 or for sprunge it: git format-patch -1 --stdout | sprunge 2014-07-28 08:36:19 http://sprunge.us/gdFc 2014-07-28 08:36:22 like so? 2014-07-28 08:38:11 remove leading whitespace in commie message 2014-07-28 08:38:20 git commit --amend 2014-07-28 08:38:52 remove the whitespace in front of every line 2014-07-28 08:38:59 http://sprunge.us/NBON 2014-07-28 08:39:46 oops 2014-07-28 08:39:48 hm, there is still whiespace... 2014-07-28 08:39:50 also 2014-07-28 08:39:57 there is trailing whitespace in APKBUILD 2014-07-28 08:40:08 under 'done' in prepare() {} 2014-07-28 08:40:35 row number 30 2014-07-28 08:41:01 it is also a good thing to wrap long lines 2014-07-28 08:41:33 http://sprunge.us/gNCS 2014-07-28 08:42:07 ok 2014-07-28 08:44:04 like this? http://sprunge.us/CQLU 2014-07-28 08:44:34 ooh, that's all over the place 2014-07-28 08:45:45 how do i get rid of the first message in the first commit? 2014-07-28 08:47:31 ? 2014-07-28 08:47:46 if you want fix commit message you do: git commit --amend 2014-07-28 08:48:42 http://sprunge.us/dWNC 2014-07-28 08:49:06 ah yes, i didn't see the message at the top, i just added a message at the bottom 2014-07-28 08:58:30 commit message is ok now 2014-07-28 08:58:56 could you also please fix the trailing whitespace in APKBUILD 2014-07-28 08:59:04 and wrap the long lines in package()? 2014-07-28 08:59:22 this is the fix: http://sprunge.us/gNCS 2014-07-28 08:59:51 (so you can cheat with: curl http://sprunge.us/gNCS | git apply && git add APKBUILD && git commit --amend 2014-07-28 09:02:27 i thought i did that 2014-07-28 09:03:32 which whitespace? 2014-07-28 09:05:00 got it 2014-07-28 09:05:14 i forgot git add 2014-07-28 09:06:19 http://sprunge.us/aQgd better? 2014-07-28 09:11:45 can you check this? http://wiki.alpinelinux.org/wiki/Creating_an_Alpine_package#Commit_your_work 2014-07-28 09:19:53 ncopa: another aport http://sprunge.us/QDZK 2014-07-28 09:23:43 ScrumpyJack: yes that looks nice 2014-07-28 09:24:19 I'm flattered! 2014-07-28 09:24:24 ;) 2014-07-28 09:24:31 thanks :) 2014-07-28 09:25:03 /home/ncopa/aports/.git/rebase-apply/patch:52: trailing whitespace. 2014-07-28 09:25:03 "$pkgdir/usr/share/man/man1" "$pkgdir/usr/share/locale" 2014-07-28 09:25:14 yup 2014-07-28 09:25:35 there is a single whitespace there 2014-07-28 09:25:40 in package() finction 2014-07-28 09:25:55 i thinkt --prefix=/ looks wrong... 2014-07-28 09:26:05 is taht correct 2014-07-28 09:26:30 in fvwm, is it correct that --prefix=/ ? 2014-07-28 09:26:34 shouldnt it be /usr ? 2014-07-28 09:27:43 i've removed the trailing whitespace 2014-07-28 09:28:14 prefix is good, because make DESTDIR="$pkgdir"/usr 2014-07-28 09:30:46 why not do: --prefix=/usr and then later make DESTDIR="$pkgdir"? 2014-07-28 09:30:58 like we do in every other apkbuild 2014-07-28 09:31:27 yes, it makes sense, i'll do that 2014-07-28 09:33:37 done http://sprunge.us/QFQh 2014-07-28 09:33:41 might be you dont need set the MANDIR 2014-07-28 09:35:03 hm 2014-07-28 09:35:19 i dont thin ktehre should be \ when inside "" 2014-07-28 09:35:38 so i dont think it should be used in depends or depends_Dev 2014-07-28 09:35:46 i'd also like to shorten the pkgdesc so it fits on one line 2014-07-28 09:35:52 ok 2014-07-28 09:36:27 why don't i need MANDIR? i want abuild to create a package for the man pages 2014-07-28 09:36:37 fedora uses: "Highly configurable multiple virtual desktop window manager" 2014-07-28 09:36:41 as pkgdesc 2014-07-28 09:36:54 no, i mean that make install should do the right thing when --prefix=/usr 2014-07-28 09:37:37 pkgdesc="FVWM is a powerful desktop window manager for the X Window system" 2014-07-28 09:38:18 good enough 2014-07-28 09:38:21 oh 2014-07-28 09:38:27 i think url should include http:// 2014-07-28 09:38:39 hm 2014-07-28 09:38:47 i didint check that in prev apkbuild 2014-07-28 09:38:48 also 2014-07-28 09:38:49 depends 2014-07-28 09:38:55 does not need anything i think 2014-07-28 09:40:04 i alos tent to like to have each configure option on its own line 2014-07-28 09:40:17 makes it easier to temporary enable/disable things 2014-07-28 09:42:10 i added the depends as they would be annoying to figure out if you wanted to run fvwm locally 2014-07-28 09:42:30 ? 2014-07-28 09:42:31 are they all added by setup-xorg? 2014-07-28 09:42:55 no 2014-07-28 09:43:15 but abuild should figure out the shared-object dependencies 2014-07-28 09:43:32 >>> fvwm*: Tracing dependencies... 2014-07-28 09:43:32 so:libICE.so.6 2014-07-28 09:43:32 so:libSM.so.6 2014-07-28 09:43:32 so:libX11.so.6 2014-07-28 09:43:32 so:libXcursor.so.1 2014-07-28 09:43:33 so:libXext.so.6 2014-07-28 09:43:35 it would? 2014-07-28 09:43:35 so:libXft.so.2 2014-07-28 09:43:37 so:libXpm.so.4 2014-07-28 09:43:39 so:libXrender.so.1 2014-07-28 09:43:43 so:libc.musl-x86_64.so.1 2014-07-28 09:43:45 so:libfontconfig.so.1 2014-07-28 09:43:47 so:libintl.so.8 2014-07-28 09:43:49 so:libpng16.so.16 2014-07-28 09:43:51 yup 2014-07-28 09:43:53 it will scanelf 2014-07-28 09:43:55 and find the 'NEEDED' secion 2014-07-28 09:44:15 that is pretty cool! 2014-07-28 09:44:30 and then add depends="so:libpng16.so.16" 2014-07-28 09:44:48 the libpng package will have a provides="so:libpng16.so.16" 2014-07-28 09:45:02 that way things will work even if we rename the libpng package 2014-07-28 09:45:11 neat. dependancies removed 2014-07-28 09:45:22 or if the libpng16.so.16 moves to another package 2014-07-28 09:45:41 http://sprunge.us/XJSd 2014-07-28 09:46:05 MANDIR gets correct once you have --prefix=/usr 2014-07-28 09:47:04 i'm looking at you ldapvi pacakge again 2014-07-28 09:47:17 :) 2014-07-28 09:47:19 i'd like to rename the gnumakefile.diff file to .patch 2014-07-28 09:47:27 ok 2014-07-28 09:47:27 and make it work with -p1 2014-07-28 09:47:58 simply adding ./ to --- and +++ lines should work 2014-07-28 09:48:19 or: --- a/GNUmakefile.in 2014-07-28 09:48:28 +++ b/GNUmakefile.in 2014-07-28 09:48:38 can we finish fvwm first? :) 2014-07-28 09:48:42 ok 2014-07-28 09:48:48 lets do fvwm first 2014-07-28 09:49:47 http://sprunge.us/WaYW 2014-07-28 09:52:36 should work 2014-07-28 09:52:45 just split the long depends_dev line 2014-07-28 09:52:56 you dont need any \ for that 2014-07-28 09:53:01 as it is within "" 2014-07-28 09:54:28 http://sprunge.us/JaGO 2014-07-28 09:55:25 without the \ 2014-07-28 09:55:33 um 2014-07-28 09:55:34 also 2014-07-28 09:55:42 duh! 2014-07-28 09:55:45 is libxslt needed build time? 2014-07-28 09:55:57 what happens if you remove libxslt 2014-07-28 09:56:19 it doesn't compile. let me check 2014-07-28 09:56:20 normally only -dev package are needed 2014-07-28 09:56:32 yes, are both in? 2014-07-28 09:56:51 no only libxslt are there 2014-07-28 09:56:57 libxslt-dev would make sense 2014-07-28 09:57:28 rebuilding with -dev ... 2014-07-28 09:57:35 you can also remove the mkdir line(s) in package 2014-07-28 09:57:39 its not needed 2014-07-28 09:59:13 how does abuild know which dirs to make in $pkgdir? 2014-07-28 10:02:59 make install does the mkdir -p 2014-07-28 10:03:27 it know what dirs it needs due to ./configure 2014-07-28 10:03:32 http://sprunge.us/hMQC 2014-07-28 10:03:52 try with --stdout ;) 2014-07-28 10:04:11 lol, sorry 2014-07-28 10:04:20 http://sprunge.us/gEOF 2014-07-28 10:06:10 that shoudl be good enough 2014-07-28 10:06:55 you could remove one leading tab 2014-07-28 10:07:04 but i feel i am very picky now :) 2014-07-28 10:08:02 nitpick: http://sprunge.us/iQjd 2014-07-28 10:08:08 one thing more 2014-07-28 10:08:17 i am not convinced libxslt is needed 2014-07-28 10:08:25 >>> fvwm*: Tracing dependencies... 2014-07-28 10:08:25 so:libICE.so.6 2014-07-28 10:08:25 so:libSM.so.6 2014-07-28 10:08:25 so:libX11.so.6 2014-07-28 10:08:25 so:libXcursor.so.1 2014-07-28 10:08:27 so:libXext.so.6 2014-07-28 10:08:29 so:libXft.so.2 2014-07-28 10:08:31 so:libXpm.so.4 2014-07-28 10:08:33 so:libXrender.so.1 2014-07-28 10:08:35 so:libc.musl-x86_64.so.1 2014-07-28 10:08:37 so:libfontconfig.so.1 2014-07-28 10:08:39 so:libintl.so.8 2014-07-28 10:08:41 so:libpng16.so.16 2014-07-28 10:08:45 >>> fvwm*: Package size: 5.9 MB 2014-07-28 10:08:47 i would have expected to see libxslt there 2014-07-28 10:09:24 could be it is needed for html documentation 2014-07-28 10:10:40 yup 2014-07-28 10:10:44 libxslt-dev can be removed 2014-07-28 10:11:01 ah yes, i remember now. libxslt and perl were needed to create the man pages, so libxslt still needed 2014-07-28 10:15:14 ah 2014-07-28 10:15:14 yes 2014-07-28 10:15:18 that actually makes sense 2014-07-28 10:15:26 Build man pages? no: No xsltproc found in PATH 2014-07-28 10:16:06 yup :) i did this on friday and have forgotten it all over the weekend :) 2014-07-28 10:16:20 then libxslt makes sense and not libxslt-dev 2014-07-28 10:16:27 yup 2014-07-28 10:17:02 http://sprunge.us/KAjL 2014-07-28 10:20:28 whhops i forgot my lunch... 2014-07-28 10:33:08 ncopa: yum yum in my tum :p 2014-07-28 11:01:08 hmm, what provides #include 2014-07-28 11:04:18 http://man7.org/linux/man-pages/man3/mcheck.3.html 2014-07-28 11:04:23 CONFORMING TO top 2014-07-28 11:04:23 These functions are GNU extensions. 2014-07-28 11:27:37 ncopa: yes i realise that, i meant what provided this in the musl c libraries (if at all)? 2014-07-28 11:33:35 mcheck.h seems tp provide the functions mtrace and muntrace 2014-07-28 11:41:11 . 2014-07-28 11:42:12 on another note, i get a lot of undefined reference to `libintl_gettext' when compiling stuff. does this ring a bell? 2014-07-28 11:43:51 is the solution to add -liconv -lintl ? 2014-07-28 11:46:38 yes 2014-07-28 11:46:59 the correct question is, does musl c provide mcheck.h at all 2014-07-28 11:47:14 i assume it does not, since you ask (and since it is a gnu extension) 2014-07-28 12:01:42 ncopa: can I remove spawn-fcgi.init/confd from the lighttpd package and push a separate fcgi-spawn package? 2014-07-28 12:01:54 spawn-fcgi is no longer part of lighttpd it seems 2014-07-28 12:02:17 so, the lighttpd package no longer providers /usr/bin/spawn-fcgi 2014-07-28 12:16:41 uggedal: oh, then yes, it should be removed 2014-07-28 12:17:07 what provides spawn-fcgi nowdays? 2014-07-28 12:18:51 http://redmine.lighttpd.net/projects/spawn-fcgi/wiki/WikiStart 2014-07-28 12:20:18 since it was broken I used a newer init/confd from gentoo with multi instance support 2014-07-28 12:24:21 ncopa: here is another one http://sprunge.us/ZEYN 2014-07-28 12:25:31 normally we patch configure.ac and re-run autoconf/automake 2014-07-28 12:25:41 which makes the patch possible to send upstream 2014-07-28 12:25:51 he doesn't use autotools 2014-07-28 12:26:05 configure is a short perl script 2014-07-28 12:26:08 oh 2014-07-28 12:26:10 indeed 2014-07-28 12:27:25 applied. thanks! 2014-07-28 12:27:47 how are we doing with fvwm? 2014-07-28 12:28:30 i think i pushed it? 2014-07-28 12:28:39 aports:master | ScrumpyJack | testing/fvwm: new aport | http://tinyurl.com/l3x88r7 2014-07-28 12:28:39 aports:master | Natanael Copa | testing/fvwm: move perl and libxslt to makedepends | http://tinyurl.com/n46q447 2014-07-28 12:30:35 so you put non-dev packages on makedeps? 2014-07-28 12:31:07 depends_dev is a hack so we can add dpeends on a -dev subpackage 2014-07-28 12:31:33 but fvwm does not have any fvwm-dev so it does not matter 2014-07-28 12:31:55 you could in theory removed the depends_dev line 2014-07-28 12:31:59 and only use makedepends 2014-07-28 12:32:44 that's getting complicated ;) 2014-07-28 12:32:51 but lets say you have a library, libfoo, that in foo.h uses some defines from libbar 2014-07-28 12:32:59 so libfoo depends on libbar 2014-07-28 12:33:10 the foo.h has a line: #include 2014-07-28 12:33:33 this means that you cannot build anything with libfoo, unless bar.h is installed 2014-07-28 12:33:55 thus, your libfoo-dev package will need to have a depends="libbar-dev" 2014-07-28 12:34:01 so bar.h gets pulled in 2014-07-28 12:34:09 sometimes that is needed, sometiems not 2014-07-28 12:34:42 so libfoo will need: depends_dev="libbar-dev" 2014-07-28 12:35:04 that will add the depends=libbar-dev to libfoo-dev 2014-07-28 12:35:26 that is why the depends_dev variable exist in first place 2014-07-28 12:35:33 ncopa: can i maintain the taskwarrior package you created? I'm making a taskd package and maintaining both makes sense 2014-07-28 12:35:42 sure 2014-07-28 12:36:13 (oh and i was listening to that) 2014-07-28 14:42:49 ncopa: i have a depend questiion for a package *sigh*. I'll get there eventually 2014-07-28 14:44:27 cdw is a frontend to wodim cdrecord mkisofs etc, but it is useless on it's own 2014-07-28 14:45:15 my depends for it are cdrkit ncurses libcdio libburn 2014-07-28 14:45:20 does that make sense? 2014-07-28 14:46:58 ScrumpyJack: i guess it links to ncurses libcdio and liburn 2014-07-28 14:47:12 those are and it uses cdrkit 2014-07-28 14:47:21 -those are 2014-07-28 14:49:09 sounds like ncurses libcdrio and libburn should not be in depeds yes 2014-07-28 14:49:13 maybe cdrkit? 2014-07-28 14:49:41 ncurses libcdio libburn are run dependancies 2014-07-28 14:50:00 they must be installed for cdw to run 2014-07-28 14:51:34 i'm just wordering if cdrkit should be there. it's not needed to run cdw, but cdw is useless without it. 2014-07-28 14:51:54 it should be the only depends 2014-07-28 14:52:08 the rest is auto detected 2014-07-28 14:52:17 if you add the headers of course and the pkg find them. 2014-07-28 14:52:49 ncurses should be autodetected yes 2014-07-28 14:53:04 i'd expect that to happen with libcdio and libburn too 2014-07-28 14:53:07 so i should check Tracing dependencies ... 2014-07-28 14:53:15 ok cool 2014-07-28 14:53:20 just adding ncurses-dev, libcdio-dev and libbur-dev to makedepends should be enough 2014-07-28 14:53:48 i also need md5sum to burn cd/dvd but 2014-07-28 14:53:49 alpine-dev-edge:~/cdw$ apk info --who-owns /usr/bin/md5sum 2014-07-28 14:53:49 ERROR: /usr/bin/md5sum: Could not find owner package 2014-07-28 14:54:01 they are probably hard deps, it wont build if you dont install those headers. 2014-07-28 14:54:19 its busybox 2014-07-28 14:54:32 ls -al /usr/bin/md5sum 2014-07-28 14:56:33 cool 2014-07-28 14:56:54 the symlinks are not in apk db 2014-07-28 14:57:21 they are created afterwards so other pkgs can overwrite them. 2014-07-28 14:59:09 and on every apk disk operation, apk will regenerate them where possible (when you remove a pkg which replaced one of the busybox applets). 2014-07-28 15:09:11 ncopa: ok here is cdw http://sprunge.us/gJdJ 2014-07-28 15:14:19 ScrumpyJack: i think bison should not be depends_dev 2014-07-28 15:15:21 if you need it at runtime, then its depends, if you need it for building its makedepends 2014-07-28 15:54:14 ok, so i put everything in makedepends, unless i want to build a pkgname-dev ? 2014-07-28 17:08:25 ScrumpyJack: no 2014-07-28 17:09:09 bison doesnt get linked to your pkg, so its either depends or makedeps 2014-07-28 17:10:31 ScrumpyJack: our wiki has some information regarding this. 2014-07-28 17:13:19 http://wiki.alpinelinux.org/wiki/APKBUILD_Reference#depends_dev 2014-07-28 17:30:35 clandmeter: i don't get it. are you sure you read me right? 2014-07-28 17:31:41 when i create a package, i put stuff that is needed at build time in makedeps - unless i am building a -dev package 2014-07-28 17:33:13 and if i need something for a package at run time, i put it in depends (but only if abuild hasn't figured out what i need with scanelf), right? 2014-07-28 18:57:32 ScrumpyJack: use depends for run time deps, makedeps for tools that are needed to build the package and headers/shared libs needed for linking (in *-dev packages) 2014-07-28 18:58:46 use depends_dev if there is a -dev subpackage that needs another shared lib 2014-07-28 19:11:45 lovely, thanks 2014-07-28 19:51:34 kaniini, around ? 2014-07-28 19:52:25 hi 2014-07-28 19:52:46 hi 2014-07-28 19:53:04 hi 2014-07-28 19:53:04 I was thinking of build compact mini-pc using i3-4130T or i3-3220T 2014-07-28 19:53:10 hi barthalion 2014-07-28 19:53:22 can advice of 'T' models 2014-07-28 19:53:42 its low TDP model, rare and hard to find in market 2014-07-28 19:54:06 wanting to know if its worth the effort ? 2014-07-28 19:55:21 probably not 2014-07-28 19:55:34 planning to use mini-ITX board 2014-07-28 19:55:53 barthalion, any explanation ? 2014-07-28 19:56:23 energy isn't usually that expensive to waste time and money for rare model 2014-07-28 19:56:35 if it's really rare, it probably costs more as well 2014-07-28 19:57:01 I think I can get at same range price as normal model 2014-07-28 19:57:08 but just speculating 2014-07-28 19:57:14 if so, go for it 2014-07-28 19:58:39 hmmm 2014-07-28 20:16:24 performancecopilot.org - could be usefull 2014-07-29 07:13:56 moinmoin 2014-07-29 07:47:16 ncopa: i've verified the /dev/null-issue 2014-07-29 07:47:23 hi, yes? 2014-07-29 07:47:28 by creating a fresh lxc container 2014-07-29 07:47:36 and the permissions on /dev/null are correct 2014-07-29 07:47:48 so everything is fine 2014-07-29 07:47:49 do you have any idea what caused the bad permissions? 2014-07-29 07:47:57 no 2014-07-29 07:48:09 ok, lets keep our eyse open 2014-07-29 07:48:17 unfortunately not 2014-07-29 07:48:22 sure 2014-07-29 07:48:43 i think it must be one of the updates around the date i've created the jenkins-issue 2014-07-29 07:49:06 got the problem on both edge- and release-nodes 2014-07-29 07:50:18 but all in all i've learned something about the internals :-) 2014-07-29 08:18:36 morning 2014-07-29 08:48:27 fabled: the "main/quagga: fix msghdr initializers" fixes #3244? 2014-07-29 08:48:44 ncopa, no, it does not fix it yet 2014-07-29 08:48:51 ok 2014-07-29 08:49:00 it's need, but not full fix. there's a configure.ac thing i need to fix too. i think i know what is the final problem 2014-07-29 08:51:14 hi 2014-07-29 08:52:41 hi 2014-07-29 09:07:37 clandmeter: we need update those: http://weblog.rubyonrails.org/2014/7/2/Rails_3_2_19_4_0_7_and_4_1_3_have_been_released/ 2014-07-29 09:08:23 since we have both ruby-rails* and ruby-redmine-rails* we will need to fix 2 different rails 2014-07-29 09:08:33 and we should do that without breaking anything... 2014-07-29 09:08:37 any idea how we do that? 2014-07-29 09:12:26 i think we need to update redmine and related gems to match them. the non redmine gems can be upgraded without redmine (afaik) 2014-07-29 09:13:24 do you know if there are any way we can check that we have dont introduce any breakage? 2014-07-29 09:13:52 like: SOMEPATH=$pkgdir" ruby -r actionpack '' 2014-07-29 09:19:40 rails is a nightmare to package 2014-07-29 09:47:21 i need to get some free time go check it 2014-07-29 09:49:00 it has always been trial and error to get it running :( 2014-07-29 09:57:07 ok 2014-07-29 09:57:28 clandmeter: i tested ruby-redmine-* in v3.0 2014-07-29 09:57:33 and i think it works 2014-07-29 09:57:58 and i think i finally have an idea how to split up main repo 2014-07-29 09:58:17 or introduce an 'extras' repo or similar 2014-07-29 09:58:43 the extras repo would only get security fixes for latest stable 2014-07-29 09:58:54 while main would get for 2 years 2014-07-29 09:59:19 there are some package that we cannot really give support for 2 years 2014-07-29 09:59:29 redmine + rails is one 2014-07-29 09:59:33 actually all ruby* 2014-07-29 09:59:38 qemu is another 2014-07-29 10:00:06 while things like samba, postgresql, xen are not really any problem provide longer support for 2014-07-29 10:00:45 flashcache? :-) 2014-07-29 10:34:59 fabled: any change you could look into the issue we have with nfs on musl? 2014-07-29 10:36:35 chance even :) 2014-07-29 10:40:24 i kinda swamped with things. and i don't even have a test box for that. 2014-07-29 10:58:48 hm, what a pity 2014-07-29 12:08:41 was there some reason we have not upgraded to gcc-4.8.3 already? 2014-07-29 12:10:52 it was probably released to close v3.0 release 2014-07-29 12:12:57 busybox top works in edge now? #3243 2014-07-29 12:30:13 ncopa, works for me in edge 2014-07-29 12:30:34 ok good 2014-07-29 12:34:25 lbu diff fix does not work as expected 2014-07-29 12:34:44 symlinks to files (eg rc-update ...) does not show anything at all 2014-07-29 12:37:32 was lbu also ment for created user? i did have installed oscam package, user was created with the install, did lbu ci, but on reboot package and the user was not there 2014-07-29 12:37:51 fabled: do you think this will work for dahdi? http://git.alpinelinux.org/cgit/aports/commit/?id=2bc0e0bf93ebc8486cda686b659f71a8bae87faa 2014-07-29 12:38:33 crow: the user should be there (in /etc/passwd) but the home dir is not taken care of by lbu unless explicitly told so 2014-07-29 12:38:40 ncopa, yes, i think so 2014-07-29 12:38:48 ok, i just cerry pick it 2014-07-29 12:41:37 ncopa this user seems does not have /home . this is from /etc/passwd oscam:x:1000:1000:Linux User,,,:/home/oscam:/sbin/nologin 2014-07-29 12:42:21 ls -ltr /home gives total 0 2014-07-29 12:42:27 after reboot, the oscam user is expected to be in /etc/passwd 2014-07-29 12:42:43 but lbu will not manage anything under /home by default 2014-07-29 12:42:46 will check it on my dev machine again 2014-07-29 12:42:56 so the homedir is not expected to be there after reboot 2014-07-29 12:43:23 if you want /home/oscam to be there after reboot you will have to: lbu include /home/oscam 2014-07-29 12:43:35 well actually as this user is nologin, there is no need for lbu a home dir 2014-07-29 12:43:44 ok 2014-07-29 12:43:53 so problem is not that /home is empty then? 2014-07-29 12:44:21 fabled: trying to build openjdk for 2.7 now as you suggested on #3242 2014-07-29 12:44:43 StarWarsFan, good luck with that one...;) 2014-07-29 12:44:52 but as the initial problem is solved now (issue with /dev/null) this is not that much important anymore... 2014-07-29 12:45:03 but the oscam would not start, without not so good error message, so i did apk add osam and startet custom /etc/init.d/oscam.server services (as file are in /etc/conf.d/oscam.server and in /etc/oscam.server/ ) 2014-07-29 12:45:06 i tried it once and failed. though i think i did all the fixes in openjdk later on that made it work with musl 2014-07-29 12:45:15 so if you could take a look at nfs, that would be awesome! ;-) 2014-07-29 12:45:53 someone *still* uses NFS? hey, this is 2014! ;) 2014-07-29 12:46:15 ACTION uses nfs 2014-07-29 12:46:22 any suggestions for a replacement? 2014-07-29 12:46:56 i do also, 2014-07-29 12:47:12 i wonder the same, i want share /var/cache/distfiles for bld1/bld2.a.o 2014-07-29 12:47:15 same quesstion like StarWarsFan , any better suggestions? 2014-07-29 12:47:40 i suppose people in 2014 use google drive and such... :) 2014-07-29 12:47:48 dropbox 2014-07-29 12:48:09 well for my NAS i still do nfs, 2014-07-29 12:59:32 that's not an option... 2014-07-29 13:00:02 depends on network bandwidth... well, i've noticed the ";)" on fabled's answer ;-) 2014-07-29 13:00:26 so don't disturb him, maybe he's already digging into it... 2014-07-29 13:00:37 i still don't have a test box 2014-07-29 13:02:11 local i could use samba but that should even be worse then nfs, so is there realy an alternativ? 2014-07-29 13:02:56 anyone tried Code? 2014-07-29 13:02:58 Coda* 2014-07-29 13:05:28 Andrew filesystem looks interesting too 2014-07-29 13:08:34 fabled: im sure somebody is able to provide you a test box :) 2014-07-29 13:08:52 that includes me 2014-07-29 13:28:50 fabled: what's it for? i have something i could put a alpine lxc container on if that helps 2014-07-29 14:52:21 woah 2014-07-29 14:53:00 is http://nl.alpinelinux.org/alpine/v3.0/main/x86_64/ broken atm? 2014-07-29 14:54:08 http://www.downforeveryoneorjustme.com/http://nl.alpinelinux.org/alpine/v3.0/main/x86_64/ 2014-07-29 14:54:53 ncopa: there are packages missing 2014-07-29 14:55:24 oh 2014-07-29 14:55:26 indeed 2014-07-29 14:55:31 ^^ 2014-07-29 14:56:40 edge seems to be fine 2014-07-29 14:59:56 x86 too apparently 2014-07-29 15:00:52 argh 2014-07-29 15:00:52 ok 2014-07-29 15:01:23 problem found? :) 2014-07-29 15:01:52 yes 2014-07-29 15:02:14 and i in last minute probably managed to make it stop sync to dl-3 2014-07-29 15:02:20 so i have a copy of it 2014-07-29 15:02:24 hopefully 2014-07-29 15:02:34 but it should spread to all other mirrors as we speak :-( 2014-07-29 15:02:41 :| 2014-07-29 15:04:47 yes x86 is also affected 2014-07-29 15:04:55 this is bad 2014-07-29 15:08:01 i don't know if this helps, but here is a mirror that seems to be fine... http://mirror1.hs-esslingen.de/pub/Mirrors/alpine/v3.0/main/x86_64/ 2014-07-29 15:14:51 plueschopath: thanks a bunch 2014-07-29 15:15:02 you saved a major disaster 2014-07-29 15:15:08 now its only a minor :-/ 2014-07-29 15:16:39 you're welcome. 2014-07-29 15:19:02 i managed to stop dl-3 sync with 1-2 mins margin :) 2014-07-29 15:19:14 thanks to that i can just copy data back to nl.a.o 2014-07-29 15:19:21 instead of total rebuild 2014-07-29 15:19:47 \o/ 2014-07-29 15:19:52 :3 2014-07-29 15:20:08 this was triggered by a bug in the newer build scripts 2014-07-29 15:20:21 + me beeing stressed while pushing kernel update earlier today 2014-07-29 15:20:47 (i did verify that kernel boots though) 2014-07-29 15:20:57 (both in vm and on physical) 2014-07-29 15:21:19 :) 2014-07-29 15:21:38 and i know where to look to make sure it never happens again 2014-07-29 15:21:43 total rebuild would be complicated i guess... because of all the dependencies 2014-07-29 15:21:51 thats good :) 2014-07-29 15:21:53 not really complicated 2014-07-29 15:22:07 but it would take a day or 2 2014-07-29 15:22:26 and packages would be unavailable meanwhile 2014-07-29 15:23:19 okay 2014-07-29 15:28:02 should be back on nl.a.o now 2014-07-29 15:30:50 http://nl.alpinelinux.org/alpine/v3.0/main/x86_64/APKINDEX.tar.gz isn't the right one i think 2014-07-29 15:31:45 that is probably true yes 2014-07-29 15:31:55 i need copy the packages to the builders too 2014-07-29 15:32:00 and trigger a rebuild 2014-07-29 15:32:08 and then we should be all ok 2014-07-29 15:32:24 okay :) 2014-07-29 15:41:21 main deleted: 3885 2014-07-29 15:41:28 that was what happened :-( 2014-07-29 15:42:17 oh no... not again.. 2014-07-29 15:46:48 ok 2014-07-29 15:46:53 its finally building properly 2014-07-29 15:47:04 i will have to go for a while now 2014-07-29 15:47:16 but it should be ok on nl.a.o in 20-30 mins? 2014-07-29 15:47:43 if it manages to build kernel in 12 mins it will reach next sync period 2014-07-29 15:47:56 otherwise the other mirrors will have to wait another hour 2014-07-29 15:48:17 i'll come by later (in 3-4 hours) to check the progress 2014-07-29 15:48:18 see you 2014-07-29 16:09:40 thank you ncopa :) 2014-07-29 18:01:38 https://paste.xinu.at/O2eFx/ for some reason it fails on make install 2014-07-29 18:01:53 because $pkgdir/blablah/usr/bin exists, and apparently is not a directory 2014-07-29 18:02:18 are my packaging skills so rusty or configure.ac is screwed? 2014-07-29 18:02:25 clandmeter, uggedal ↑ 2014-07-29 18:06:18 Is Alpine simple, minimalist and sane because it was originally designed for x86 Routers, Firewalls, VPNs, VoIP and servers? 2014-07-29 18:06:49 Or is it a philosophy it follows? And ideology? 2014-07-29 18:07:28 both I guess 2014-07-29 18:07:36 we just prefer sanity 2014-07-29 18:07:52 and keeping things minimalistic comes from or want of simplicity 2014-07-29 18:07:59 there are things obviously we can't help 2014-07-29 18:08:05 My concern is if Alpine will continue following this flow of thought, praising minimalism and sanity, or if it might want to try to expand usage to desktops and in the process become bloated 2014-07-29 18:08:35 you can start using alpine on desktop anytime 2014-07-29 18:08:51 I know, I do on mine 2014-07-29 18:09:36 we aren't going to change the philosophy, even if we expand usage in more or less new areas 2014-07-29 18:10:12 Great to hear that! 2014-07-29 18:10:40 I think a page explaining that on the wiki or website would be great 2014-07-29 18:11:10 barthalion: works for me 2014-07-29 18:11:19 "Alpine Philosophy" or "Alpine Ideology" or "Trivial Solutions" or whatever 2014-07-29 18:11:32 clandmeter: weird, my chroot must be broken then 2014-07-29 18:11:41 check dmesg? 2014-07-29 18:11:52 or is it non grsec 2014-07-29 18:12:01 non grsec 2014-07-29 18:12:35 I'll push it to repos and let the builders deal with it 2014-07-29 18:12:43 it's testing anyway 2014-07-29 18:13:29 "The trivial solution in math is the simple, obvious answer. The fable that NASA spent millions on a space pen while Russia used a pencil is an example." I think Alpine could have a page explaining how it follows this idea. 2014-07-29 18:14:19 Like the (RIP) Arch Way page, from when Arch actually followed that 2014-07-29 18:15:49 well, we as in Arch still follow it… 2014-07-29 18:17:27 oh well 2014-07-29 18:55:28 barthalion: imho arch is a bit bloated 2014-07-29 18:56:08 sure, the core is decently sized, but most packages enable every option under the sun to cater to its wide audience 2014-07-29 18:58:02 and systemd is a pita if you actually care what your system is doing 2014-07-29 18:59:26 :) 2014-07-29 18:59:36 uggedal: you are the maintainer of docker, right? Do you use it successfully? 2014-07-29 19:00:31 shafire: no, I don't like its one app per container philosophy... 2014-07-29 19:01:02 simple things become compilicated if you split it up into dozen docker containers in my experience 2014-07-29 19:01:20 do you use lxc instead? 2014-07-29 19:01:40 yeah, in some cases 2014-07-29 19:02:39 are the templates okay in alpine? last time I checked, all were broken except alpine 2014-07-29 19:03:16 my servers are linode xen instances anyway, so I can compartmentalize them well without the need for lxc 2014-07-29 19:03:44 I haven't tried running non-alpine distros under alpine lxc 2014-07-29 19:04:46 my guess is that some of the might need coreutils and bash 2014-07-29 19:04:54 s/the/them/ 2014-07-29 19:08:34 ubuntu was working last i tried 2014-07-29 19:09:05 so i guess that should work for most of the debian family 2014-07-29 19:10:53 Ok, thanks 2014-07-29 19:13:21 uggedal: just like every binary distro 2014-07-29 19:13:44 uggedal: if you need to slimmed down packages use gentoo, using binary distribution doesn't make much sense if you care about such things 2014-07-29 19:14:15 and about systemd… haters gonna hate, I'm not here to convince anyone to it or to openrc in Arch 2014-07-29 19:24:35 barthalion: nah, I don't like openrc either 2014-07-29 19:26:11 I actually like most of the init parts of systemd. socket activation, supervision, declarative format 2014-07-29 19:50:36 hi 2014-07-29 19:50:39 just checking in 2014-07-29 19:50:47 seems like packages are back on v3.0 2014-07-29 19:52:28 uggedal: i agree with you that systemd has nice ideas 2014-07-29 19:52:51 and that openrc, while is 'ok' and gets the job done, is not really great 2014-07-29 19:53:04 i think openwrt has some nice things going on 2014-07-29 19:53:14 ubus, ubox etc 2014-07-29 20:02:52 ok i better go again 2014-07-29 20:02:54 see u tm 2014-07-29 20:04:35 good night 2014-07-30 05:28:42 good morning 2014-07-30 06:42:57 morning 2014-07-30 07:42:11 is there an easy way to run the offical jvm from oracle on alpine? 2014-07-30 07:42:21 I know only the lxc/docker way 2014-07-30 07:45:01 i guess those are the only ones or real virtualization like kvm and xen 2014-07-30 08:06:49 i had a quick look at nfs with ncopa just now 2014-07-30 08:06:54 it looks non-trivial to fix 2014-07-30 08:07:09 need to email linux-nfs / musl and ask what to do 2014-07-30 08:07:29 fundamentally it's bad mix of nfs-utils, musl and kernel assumptions 2014-07-30 08:07:50 nfs-utils talk to kernel via virtual devices in /proc/net/rpc 2014-07-30 08:08:05 those devices support only read(2) and write(2); not e.g. writev(2) 2014-07-30 08:08:21 nfs-utils uses the c-library stdio (FILE) to access them 2014-07-30 08:08:31 and musl implements it in optimized way using writev(2) 2014-07-30 08:08:38 so those three won't work together 2014-07-30 08:11:30 ncopa: any chance to get flashcache-grsec build for current kernel 3.14.8-1-grsec? 2014-07-30 08:12:20 I tried to use it from edge (3.14.14-0) but without success 2014-07-30 08:23:54 sigh, ok 2014-07-30 08:24:10 i can cherry-pick it to 3.0-stable 2014-07-30 08:24:23 you will still need to build it yoursef from testing/ 2014-07-30 08:25:10 I just unpack the apk and insmod :) 2014-07-30 08:25:23 its in testing 2014-07-30 08:25:30 we dont build testing for stable branches 2014-07-30 08:25:46 i'd have to move it to main 2014-07-30 08:26:09 and if i move it to main i am afraid to get future maintenance pain 2014-07-30 08:26:47 was it so bad in 2.x? 2014-07-30 08:26:58 what was bad was 2014-07-30 08:27:05 i wanted make v3.0.0 2014-07-30 08:27:13 i was late in making v3.0.0 2014-07-30 08:27:21 we were running 3.14 kernel 2014-07-30 08:27:33 and there was no support for 3.14 kernel whatsoever at that point 2014-07-30 08:27:52 i think i even spent some time on it (causeing v3.0 get even more delayed) 2014-07-30 08:28:00 and finally gave up and moved it to "unmaintained" 2014-07-30 08:28:18 after v3.0 was released they made a new upstream release 2014-07-30 08:28:46 i can move it to main/ in v3.0 2014-07-30 08:29:03 but i cannot guarantee that flashcache will be included in v3.1 2014-07-30 08:30:08 Ok 2014-07-30 08:33:28 is 3.14.14 the current kernel? 2014-07-30 08:35:01 mh, looks like that 2014-07-30 08:44:12 test 2014-07-30 08:49:16 I still get the mdev bad line error on startup 2014-07-30 08:49:38 only from initramfs right? 2014-07-30 08:49:52 yes 2014-07-30 08:50:05 with 3.0.1 you got it twice, first from initramfs, and second from openrc 2014-07-30 08:50:23 if you apk ugprade, you'll get the openrc fixed 2014-07-30 08:50:34 apk upgrade does not automatically upgrade initramfs 2014-07-30 08:50:43 3.0.2 will include the fix 2014-07-30 08:50:50 fabled: thanks for looking into it. 2014-07-30 08:51:21 Still waiting for 3.0.2 because of the syslinux error 2014-07-30 08:51:42 fabled: that pretty much explains why it does not work and what to do to fix it 2014-07-30 08:51:58 i think it does not make sense to use buffered IO for talking with kernel 2014-07-30 08:52:10 so i think nfs-utils needs some refactoring... 2014-07-30 08:52:20 yes. they are jumping through hoops to make it non-buffered using setvbuf() etc. 2014-07-30 08:52:30 i think they use FILE* stuff only to be able to fprintf to it or similar 2014-07-30 08:53:00 morning 2014-07-30 08:53:09 they could in theory do snprintf + write 2014-07-30 08:53:19 for which date have you 3.0.2 planned? 2014-07-30 08:53:30 shafire: yesterday :) 2014-07-30 08:53:40 yes, if the write() atomicity matters, they should handle the write buffer themselves 2014-07-30 08:54:09 ncopa: :-) 2014-07-30 08:55:12 Someone needs to update http://nl.alpinelinux.org/alpine/MIRRORS.txt, it is outdated 2014-07-30 08:55:28 the only thing left to actually tag 3.0.2 is linux-vanilla update and linuv-virt-grsec 2014-07-30 08:56:39 rnalrd: could you please cherry-pick linux-virt-... 2014-07-30 08:56:43 thanks :D 2014-07-30 08:57:05 :) 2014-07-30 09:00:49 fabled: the fix has to come from the nfs-utils guys, or do you think you can fix it? 2014-07-30 09:01:11 clandmeter, i'd at least like to discuss it with all involved parties on how it should be fixed 2014-07-30 09:01:42 it can be fixed in nfs-utils (by not using stdio), kernel (supporting writev in nfs stuff) or musl (be rewriting stdio, not likely to happen) 2014-07-30 09:01:45 lunch 2014-07-30 09:01:47 if i understand correctly, this issue should happen on all archs? 2014-07-30 09:02:56 possibly 2014-07-30 09:35:11 now, just linux-vanilla left? 2014-07-30 10:06:58 ok 3.0.2 is baking in the oven... 2014-07-30 10:10:46 :-) 2014-07-30 10:17:05 ncopa: will be flashcache in initramfs in v3? 2014-07-30 10:36:45 who build the isos? 2014-07-30 10:37:32 built* 2014-07-30 10:38:21 shafire, the build system autobuilds them on tagging 2014-07-30 10:39:09 cool 2014-07-30 10:45:05 40 minutes to download 3.0.2 :( 2014-07-30 10:46:07 i'd complain, if i wasn't at the office :) 2014-07-30 10:46:20 OK 2014-07-30 10:46:38 upgraded to 3.0.2 with -u -s, rebooted, everything OK :-) 2014-07-30 10:46:46 no mdev error more 2014-07-30 11:23:44 ncopa, read on aufs ml about some patch for aufs-musl, any interest in getting it to AL ? 2014-07-30 11:26:05 huh 2014-07-30 11:27:20 how to delete a raid1 device? (https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/5/html/Deployment_Guide/s2-raid-manage-removing.html) 2014-07-30 11:27:47 If I do --stop, I can not --remove it anymore, because /dev/md0 disappeared 2014-07-30 11:28:04 fabled thank you for exmplaining what everything is involved in to nfs problem 2014-07-30 11:30:30 fabled: i think it was me who messed up the cups sec fix commit 2014-07-30 11:57:19 mh, someone using btrfs in v3? 2014-07-30 11:57:31 I can not use mkfs.btrfs, always getting "Error: error checking /dev/sda mount status" 2014-07-30 11:58:13 i heard someone having issues with it yes 2014-07-30 11:58:19 dunno who or why 2014-07-30 12:00:14 shit :( 2014-07-30 12:00:49 correct 2014-07-30 12:02:21 you could boot a live cd, format and then use alpine. not the most elegant solution though. 2014-07-30 12:08:13 running alpine-setup (3.0.2), selected dhcp for eth0, dhcp for wlan, i get "Error: this script should be called from udhcpc" 2014-07-30 12:12:15 ScrumpyJack, it happens when udhcpc timeouts 2014-07-30 12:12:42 pls see http://bugs.alpinelinux.org/issues/3105 for options 2014-07-30 12:16:48 cool, it's bound to timeout anyway trying dhcp on wlan. there aren't any tools to configure a wifi connection 2014-07-30 12:17:45 (which is fine) 2014-07-30 12:26:20 after installing sys, i get 2014-07-30 12:26:38 ERROR: unsatisfiable contraints: 2014-07-30 12:26:55 .setup-apkrepos (missing): 2014-07-30 12:27:04 required by world etc 2014-07-30 12:28:35 (this is after i cancelled setup-alpine when my mistyped my proxy settings, re-ran setup-alpine and entered the correct proxy settings) 2014-07-30 12:29:33 could it be left over from the first attempt to update package index? i'll reboot and try again 2014-07-30 12:31:37 Mh, just want to know, do I need a partition on my usb stick to use setup-bootable? 2014-07-30 12:31:47 or can I use directly /dev/sd* without 1/2/3... 2014-07-30 12:33:20 shafire: i think you need partition it yes 2014-07-30 12:38:17 worked fine when i got my proxy settings correct. looks like there must have been some stray wgets 2014-07-30 12:39:32 busybox ntpd starts way faster than chrony! \o/ 2014-07-30 12:55:49 running setup-xorg-base installs xinit (xorg-server depends on) 2014-07-30 13:03:43 ignore that 2014-07-30 13:05:22 would a xorg-base meta package make sense instead of setup-xorg-base? 2014-07-30 13:08:55 ScrumpyJack: just use ntpdate with cron 2014-07-30 13:09:37 cannot ntpdate make time jump backwards? 2014-07-30 13:11:01 i'm using busybox ntpd 2014-07-30 13:13:51 ncopa: potentially 2014-07-30 13:15:52 i remember writing a tiny sntp client that makes sure time does not jump backwards 2014-07-30 13:16:14 for running linux in old hyper-v 2014-07-30 13:21:46 http://pastebin.com/17SGrwpx <- the btrfs error logfile 2014-07-30 13:28:28 ncopa: it's because of modloop 2014-07-30 13:28:36 1) rc-service modloop stop, 2) btrfs.mkfs, 3) rc-service modloop start 2014-07-30 13:28:40 this worked 2014-07-30 13:30:45 thats what i read from the strace too 2014-07-30 13:30:47 weird... 2014-07-30 13:31:18 [16:30] k, we had another alpine user with the same issue 2014-07-30 13:31:25 [16:30] so we have to be more clever parsing the loop device name than just chopping the basename 2014-07-30 13:32:13 http://paste.debian.net/104076/ 2014-07-30 13:32:16 thats the patch 2014-07-30 13:35:07 shafire: its against master? 2014-07-30 13:38:36 clandmeter: [16:37] should apply to master and anything else 2014-07-30 13:43:50 can't get X to work on a Dell 620 Integrated graphics (Intel Media Accelerator 950) 2014-07-30 13:44:30 shafire: please try this and let me know if it works 2014-07-30 13:44:30 i've installed xf86-video-intel 2014-07-30 13:45:09 if it works we can push it to stable 2014-07-30 13:46:01 I still have another problem with btrfs 2014-07-30 13:46:14 failed to open /dev/btrfs-control skipping device registration: No such device fs created label disk on /dev/sda 2014-07-30 13:46:27 But: # mknod /dev/btrfs-control c 10 234 mknod: /dev/btrfs-control: File exists 2014-07-30 13:48:16 shafire: please ask in #btfrs 2014-07-30 13:48:30 or whatever channel you are in 2014-07-30 13:49:30 i dont have a filesystem to play with. im adding patches to it blindly which i dont prefer but in this case can accept (it isnt working anyway). 2014-07-30 14:29:40 shafire: btrfs works for me ... but I've only tested it in a xen vm 2014-07-30 14:29:47 /dev/xvd* and so on 2014-07-30 14:30:01 to be more specific: linode vm 2014-07-30 14:30:27 as expected :) 2014-07-30 14:31:11 oO 2014-07-30 14:31:50 the bug only occurs when you have mounted the alpine iso 2014-07-30 14:32:18 or running alpine from tmpfs 2014-07-30 14:33:13 or have any loopback device mounted 2014-07-30 14:58:53 whee! i think i have cleaned up my inbox 2014-07-30 14:59:07 please ping me if i have forgotten any patch sent to mailing list 2014-07-30 15:10:48 ncopa: ping 2014-07-30 15:14:03 http://sprunge.us/gJdJ 2014-07-30 15:14:08 http://sprunge.us/UiVO 2014-07-30 15:14:13 http://sprunge.us/IRaB 2014-07-30 16:46:41 Can I call a dev by uuid? 2014-07-30 16:46:57 something like cat /dev/get-by-uuid/% 2014-07-30 17:45:08 in which file can I put a command, which should be called on startup? 2014-07-30 19:00:55 shafire: http://wiki.gentoo.org/wiki/Local.d 2014-07-30 19:15:20 do we have a hdd performance testing tool? 2014-07-30 19:20:23 uggedal: thanks 2014-07-30 19:29:39 shafire: http://alpinelinux.org/apk/main/x86/hdparm 2014-07-30 19:30:13 http://www.cyberciti.biz/tips/how-fast-is-linux-sata-hard-disk.html 2014-07-30 19:33:49 looks good 2014-07-30 19:34:00 /dev/sda => 100mb/s 2014-07-30 19:34:12 /dev/sdc => 280mb/s (ssd) 2014-07-30 19:34:22 /dev/bcache0 => 200mb/s 2014-07-30 19:35:30 just reading, can not find writespeed 2014-07-30 20:20:24 shafire: use dd 2014-07-30 21:42:14 quick question. the efforts to create an ARM flavor of alpine were abandoned? 2014-07-30 21:48:12 no 2014-07-31 02:18:50 If I make an APKBUILD that pulls directly from Git, should I name the package "packagename" or "packagename-git"? 2014-07-31 02:33:29 Also, I'm getting this: 2014-07-31 02:33:32 abuild-apk: Not a member of group abuild 2014-07-31 02:33:40 But I just checked, and I am a member of abuild 2014-07-31 04:12:09 urras: have you logged out and logged in? 2014-07-31 04:14:36 Wow, that worked 2014-07-31 04:14:38 thanks 2014-07-31 04:41:32 urras: it should be called packagename 2014-07-31 04:42:13 urras: also, if possible, stable release versions is preferable if available from upstream 2014-07-31 06:10:20 ScrumpyJack: http://sprunge.us/iYWJ 2014-07-31 06:11:06 i did abuild checksum and git commit --amend and am pushing it now 2014-07-31 06:11:06 thanks 2014-07-31 06:29:01 urras: I guess you saw already that I uploaded tarballs with given hashes to dev.al.org, I could do the same with newer ones if you tell me commits 2014-07-31 06:35:55 do we have the package iozone? 2014-07-31 06:37:51 nope 2014-07-31 06:53:11 can I use udev rules from tmpfs? 2014-07-31 07:20:49 clandmeter: looks like it works (btrfs) 2014-07-31 08:11:36 ncopa: what about stripping connman a bit (I've been meaning to package it myself)? 2014-07-31 08:11:59 with --disable-wispr you don't need the gnutls dep 2014-07-31 08:13:07 what is this wispr? 2014-07-31 08:14:09 some kind of wifi hotspot 2014-07-31 08:14:32 ok 2014-07-31 08:14:35 lets disable it 2014-07-31 08:15:05 and only consider enable if someone asks 2014-07-31 08:15:26 yeah, some kind of standard for roaming between different different hotspots with captive web login portals 2014-07-31 08:16:55 woudl be nice with some kind of gtk ui for it 2014-07-31 08:17:57 ncopa: https://github.com/tbursztyka/connman-ui 2014-07-31 08:18:10 not a full gui, but a tray icon thingy 2014-07-31 08:18:40 This is an alpha stage software. Though it has proven to be stable, the look and 2014-07-31 08:18:40 feel require still lot of work, thus contributions are welcome! 2014-07-31 08:18:42 yeah 2014-07-31 08:18:46 looks promising 2014-07-31 08:19:15 no release 2014-07-31 08:29:26 package enhanceio-grsec/enhanceio-tools don't work, project looks dead too 2014-07-31 08:30:20 kaniini: enhanceio is yours ^^^ 2014-07-31 08:30:51 creating a cache device with enhanceio prints "signed integer is greater than maximum Cache creation failed (dmesg can provide you more info)". kernel logs shows nothing. 2014-07-31 08:30:59 shafire: lask commit is May 19 2014-07-31 08:31:26 strace? 2014-07-31 08:31:44 file a bug for it and have kaniini to fix it ;) 2014-07-31 08:33:43 Is strace something new? 2014-07-31 08:33:59 Never heard of it before yesterday 2014-07-31 08:34:21 strace is old 2014-07-31 08:34:30 traces what syscalls an app does 2014-07-31 08:34:33 i have used it for years 2014-07-31 08:35:19 shafire i use udev rules from tmpfs, need to restart udev and replug devices but rule worked 2014-07-31 08:36:12 How did you do it? I can not get it. Do you have a small howto/tutorial? 2014-07-31 08:39:13 just installed udev, created rule, restarted udev und pluged device 2014-07-31 08:42:15 shafire: so the patch made it work? 2014-07-31 08:42:41 shafire you can check device and rules with udevadm info -a -p $(udevadm info -q path -n /dev/ttyUSB0) 2014-07-31 08:44:41 ncopa, enlightenment has built-in support for connman. i should try again E at somepoint. 2014-07-31 08:45:08 yeah 2014-07-31 08:45:15 i also read about deepin desktop 2014-07-31 08:45:48 looks nice but bloaty http://distrowatch.com/weekly.php?issue=20140728#feature 2014-07-31 09:19:41 ncopa: if that patch fixes mkfs-btrfs, maybe its a good idea to apply it to 3x 2014-07-31 09:23:23 yeah 2014-07-31 09:24:11 btw, its not a musl-compat issue really 2014-07-31 09:24:19 i think its caused by our mdev config 2014-07-31 09:25:54 so i didnt work on uclibc either? 2014-07-31 09:26:10 i think we got reports after the musl switch 2014-07-31 09:26:53 or just coincidence 2014-07-31 09:51:37 sry, I did a mistake. it won't work :-( 2014-07-31 10:05:13 argh 2014-07-31 10:05:14 it's okay 2014-07-31 10:07:42 clandmeter: I did not see, you added the patch already and I added it manually too and broke all 2014-07-31 10:08:21 I built your aport from git and btrfs-progs-3.14.2-r1.apk worked for me 2014-07-31 10:15:10 shafire: why did you build it manually? 2014-07-31 10:15:51 I build it with abuild 2014-07-31 10:17:09 the one in repo isnt good enough? :) 2014-07-31 10:19:02 is it the same? 2014-07-31 10:20:50 oh 2014-07-31 10:20:56 it's in edge, you did not say that 2014-07-31 10:21:02 so I did not check :( 2014-07-31 10:21:59 you can see the commits here in this channel :) 2014-07-31 10:23:46 [16:44] aports:master | Carlo Landmeter | main/btrfs-progs: musl compat | http://tinyurl.com/ktsvjpz 2014-07-31 10:23:48 oh no :D 2014-07-31 10:24:02 can you backport it to 3.0? 2014-07-31 10:24:24 i would, if you can verify the pkg works 2014-07-31 10:26:58 I do it again 2014-07-31 10:43:03 clandmeter: positive 2014-07-31 10:43:18 nice 2014-07-31 10:43:30 sorry about the miscommunication 2014-07-31 10:43:38 should have told you i pushed it. 2014-07-31 10:44:02 ncopa: can you push it to 3.x? 2014-07-31 10:45:15 clandmeter: now I know how abuild works :p 2014-07-31 10:46:01 \o/ 2014-07-31 10:46:58 ncopa: i am just going to delete enhanceio support, it is broken 2014-07-31 10:49:13 kaniini: you tested it too? 2014-07-31 10:50:46 yeah. it does not work anymore, and we don't even use flash caching at new $dayjob anyway 2014-07-31 10:53:44 OK 2014-07-31 10:54:45 enhanceio was bought by HGST 2014-07-31 11:06:15 ok 2014-07-31 11:06:24 lets remove it then 2014-07-31 11:21:17 will you backport btrfs-progs to 3.x? 2014-07-31 11:56:23 yeah i should 2014-07-31 11:56:26 it works right? 2014-07-31 11:56:52 the only thing i dont like with the patch is the name of it :) 2014-07-31 11:59:23 ncopa: you didnt answer my question :) 2014-07-31 11:59:50 else i would have renamed it already. 2014-07-31 12:01:26 sorry... 2014-07-31 12:01:34 yeah i can push it to 3.c :) 2014-07-31 12:01:37 3.x 2014-07-31 12:02:09 thanks for taking care of it 2014-07-31 12:08:04 I was refering to another qeustion 2014-07-31 12:08:11 ScrumpyJack, thanks for taskd 2014-07-31 12:08:30 ncopa: 11:25 <@clandmeter> so i didnt work on uclibc either? 2014-07-31 12:08:46 oh.. 2014-07-31 12:08:57 i think it was a cooincident 2014-07-31 12:09:08 i dont think it would have worked with uclibc either 2014-07-31 12:09:23 so i'd say it is a regression in newver version of btrfs-tools 2014-07-31 12:09:57 thats my conclusion based on the strace i saw 2014-07-31 12:10:18 what i did was trying to resolve symlink in /dev and map it to /sys 2014-07-31 12:10:26 it* 2014-07-31 12:11:24 my guess is that with recent udev there are /dev/loop0 + a symlink /dev/loop/0 pointing to /dev/loop0 2014-07-31 12:11:31 i think in mdev we do the opposite 2014-07-31 12:11:54 so btrfs-tools does readlink and gets the "real" device /dev/loop/0 2014-07-31 12:12:05 and try map /sys/..../loop/0 2014-07-31 12:12:13 which does not exist 2014-07-31 12:12:26 i suppose we could have fixed the default mdev.conf too 2014-07-31 12:13:03 (i am guessing/assuming things - i havent checked it that corresponds with reality) 2014-07-31 12:22:28 well if mdev is configurable, then it sounds like btrfs shouldnt rely on it? 2014-07-31 12:33:27 I have seen on some routers physical switch saying WISP, hope its related to wispr ? 2014-07-31 12:33:57 'wireless Internet service provider' 2014-07-31 13:00:20 btrfs should not assume things like it does anyways so i think the patch is good 2014-07-31 13:31:31 rnalrd: do you mind if I rebuild umurmur against polarssl and add some key/cert autogeneration to initd? 2014-07-31 13:31:51 that would be very cool 2014-07-31 13:32:01 ACTION wasn't aware we had polarssl 2014-07-31 13:32:09 we don't, we will have :) 2014-07-31 13:32:16 kk :) 2014-07-31 13:52:55 uggedal: great job maintaining docker images, thanks 2014-07-31 13:59:25 ncopa: no decision was taken about tmpfiles I guess? 2014-07-31 13:59:33 barthalion: no problem. i don't use docker myself at the moment, so they can be a bit outdated 2014-07-31 14:07:29 barthalion: tmpfiles? 2014-07-31 14:08:26 ncopa: systemd thing to create directories in /run for services, also implemented in openrc 2014-07-31 14:08:35 ah 2014-07-31 14:08:43 mainly to avoid playing with mkdir in daemons 2014-07-31 14:09:05 what is the openrc equivalent? 2014-07-31 14:09:24 to avoid checkpath from init.d scripts? 2014-07-31 14:09:49 do you have link to systemd documentation? 2014-07-31 14:09:51 as, so there is checkpath for that? 2014-07-31 14:09:54 it does the same thing? 2014-07-31 14:10:20 https://github.com/OpenRC/openrc/blob/master/sh/tmpfiles.sh.in http://0pointer.de/public/systemd-man/tmpfiles.d.html 2014-07-31 14:10:25 check path verifies if the tmp dir exists and have proper permissions 2014-07-31 14:10:37 and if not, then it fails? 2014-07-31 14:10:46 or creates it? 2014-07-31 14:10:50 creates 2014-07-31 14:11:05 let's stick to the checkpath then, I didn't know about it 2014-07-31 14:11:39 I'm sitting too often in systemd playground… 2014-07-31 14:11:59 i have not current plans implementing tmpfiles.d 2014-07-31 14:13:51 sure 2014-07-31 14:14:08 protobuf-c appears missing from edge x86, any idea why? 2014-07-31 14:14:25 at least on nl.al.org 2014-07-31 14:15:58 https://gist.github.com/ncopa/9ea730dd842040b61765 2014-07-31 14:16:36 hm 2014-07-31 14:16:39 fum 2014-07-31 14:16:41 fun even 2014-07-31 14:16:44 same happens on x86_64 now 2014-07-31 14:17:00 something with grsec? it built fine in my vanilla chroot 2014-07-31 14:17:18 no, i rather think its some recent musl libc change 2014-07-31 14:23:08 no, it smells more like a parallel build issue 2014-07-31 14:26:01 barthalion, hiawatha could possibly also be build with polarssl ? 2014-07-31 14:27:52 I think so 2014-07-31 14:28:12 ncopa: probably, as I said, on my laptop it's been built correctly 2014-07-31 14:32:46 just kidding, but I am scared when the site says "Our doors are always open" 2014-07-31 14:32:56 https://polarssl.org/source-code 2014-07-31 14:33:33 specially when a secure software 2014-07-31 14:34:51 :D 2014-07-31 14:40:51 and at the bottom: "Let's be friends!" 2014-07-31 14:52:11 barthalion: https://github.com/protobuf-c/protobuf-c/issues/156 2014-07-31 14:52:42 ncopa: thanks 2014-07-31 15:00:20 ugh 2014-07-31 15:00:32 there is a protobuf-c-1.0.0 2014-07-31 15:00:45 but umurmur does not appear to build with it 2014-07-31 15:00:45 oh 2014-07-31 15:00:56 well, that's something I can take care of 2014-07-31 15:01:02 and my workaround does not build with 0.15.. 2014-07-31 15:01:31 http://sprunge.us/Reaj 2014-07-31 15:01:41 thats the fix for protobuf-c 2014-07-31 15:01:52 so: curl http://sprunge.us/Reaj | git am 2014-07-31 15:01:58 and then fix umurmur 2014-07-31 15:02:01 and then git push 2014-07-31 15:02:17 aye, thank you again 2014-07-31 15:02:20 thanks! 2014-07-31 15:02:34 will do tommorow, need to move my server first 2014-07-31 15:02:50 ok 2014-07-31 15:03:08 i havent saved the http://sprunge.us/Reaj patch though... 2014-07-31 15:21:16 http://linuxgizmos.com/atom-based-industrial-pc-has-five-gigabit-ports/ 2014-07-31 15:22:01 added to my wishlist 2014-07-31 15:32:28 vkwhat are you doing with it? 2014-07-31 15:37:35 http://thread.gmane.org/gmane.comp.file-systems.btrfs/37349 2014-07-31 15:37:37 juhu 2014-07-31 18:59:33 When building a package, is it okay to set pkgdir to /usr? 2014-07-31 19:02:00 it's not okay to set $pkgdir to anything 2014-07-31 19:02:06 it's not variable you can fiddle with 2014-07-31 19:02:12 and if you do, you're doing something wrong 2014-07-31 19:02:42 I'm probably doing something terribly wrong then 2014-07-31 19:03:20 http://lpaste.net/108511 2014-07-31 19:03:48 you know that there is testing/tox already? 2014-07-31 19:04:06 s/sudo// 2014-07-31 19:04:28 $_builddir stores path to directory with your source code, $pkgdir to package directory 2014-07-31 19:04:40 so you don't do -C $_builddir, but -C $pkgdir 2014-07-31 19:05:08 still, it has been written already 2014-07-31 19:05:45 ncopa: seems that protobuf-c-rpc has been moved to separated project 2014-07-31 19:06:04 http://lpaste.net/108513 Output of apk add tox 2014-07-31 19:07:25 barthalion: I have already installed libsodium 2014-07-31 19:07:27 blame ncopa, he didn't bump pkgrel after updating libsodium to 0.5.0 2014-07-31 19:07:33 Oh 2014-07-31 19:07:42 let me finish something and I'll trigger the rebuild 2014-07-31 19:07:51 Ok, thank you 2014-07-31 19:08:32 clandmeter: spacefm failed on build servers too 2014-07-31 19:08:38 Also, Toxic no longer uses autofools as a build system 2014-07-31 19:09:21 urras: I know, but I disgress 2014-07-31 19:09:34 feel free to send the patches 2014-07-31 19:09:49 I'm actually almost done with it 2014-07-31 19:12:40 barthalion: Is that pkgver correct? 2014-07-31 19:12:51 what's wrong with it? 2014-07-31 19:13:00 0.2052.d0e3712 2014-07-31 19:13:10 it's better than date 2014-07-31 19:13:36 I don't disagree 2014-07-31 19:14:03 it could be better of course, but I was too stupid back then 2014-07-31 19:15:05 let's hope that the first stable release will be greater than 0 2014-07-31 19:15:55 barthalion: The source is using a really old core 2014-07-31 19:16:10 Would it be better to pull directly from git? 2014-07-31 19:16:55 when in Rome, do as the Romans do 2014-07-31 19:17:11 we don't want to have git on build servers 2014-07-31 19:17:26 (it's a thing I don't understand, so don't ask me) 2014-07-31 19:18:24 What if we do it like toxics apkbuild? 2014-07-31 19:18:45 toxic uses tag, it's different story 2014-07-31 19:18:45 Where it pulls the latest in a zip or a tarball 2014-07-31 19:18:51 Oh yeah, 2014-07-31 19:19:05 I misunderstood it for a second 2014-07-31 19:19:26 does github allow to pull zip/tarball with given hash? 2014-07-31 19:19:43 I think it did, but such tarball has different md5sum every time 2014-07-31 19:19:48 (or maybe it was cgit) 2014-07-31 19:20:14 What if we use the Jenkins builds? 2014-07-31 19:20:35 they are static 2014-07-31 19:20:39 iirc 2014-07-31 19:20:52 Thats right... 2014-07-31 19:21:27 going back to work, I need to finish something before midnight 2014-07-31 19:21:42 Ok, thanks for your help 2014-07-31 19:24:26 what probably would be a good solution for vcs packages is reimplementing support for them like it has been done in makepkg 2014-07-31 19:24:44 i.e. git://repo.tld/repo.git#commit=qwerty 2014-07-31 19:24:56 and requiring maintainers to put their vcs tool in makedeps 2014-07-31 19:25:06 I guess it doesn't work like that now 2014-07-31 19:32:27 Does 3.0.2 satisfy the pre-requisites listed here for unprivileged containers? https://www.stgraber.org/2014/01/17/lxc-1-0-unprivileged-containers/