2015-05-01 04:08:41 hello 2015-05-01 04:12:42 humm 2015-05-01 08:13:20 hi 2015-05-01 08:14:02 Can I force "apk cache clean" to remove unused packages? 2015-05-01 08:17:11 --purge worked :) 2015-05-01 09:32:51 Hello, I've been lurking in here for almost a month now and I'm thinking of using alpine as my main distro, however I've got a couple of questions 2015-05-01 09:33:46 hi 2015-05-01 09:34:04 I've looked in the wiki and can't seem to find anything related to key policy of /etc/apt/keys (if there is any) 2015-05-01 09:35:29 whose keys are in there by default? what's the signing policy of generated packages? 2015-05-01 09:42:13 i'm not great with fonts. last time i cared, it was the days of xfs 2015-05-01 09:43:10 how are they packaged in alpine linux? 2015-05-01 09:43:35 i see fontconfig 2015-05-01 09:44:19 and a search for font brings up all sorts. should I just add everything i find? is there a meta-package? 2015-05-01 09:52:23 ScrumpyJack: xfs was kinda easier hehe. normally you don't need all fonts, but it's not really that many. add them, and make sure you also have the ttf and scaling stuff 2015-05-01 09:52:57 in xfce (freebsd desktop here) i had to check some antialias option for things to play nice 2015-05-01 09:53:44 and the mkfontcache / fontconfig things you need to run via googling, after you add some fonts 2015-05-01 09:56:06 hi.. any hints how should I "properly" write contents of the alpine rpi .tar.gz package to an sd-card so that it's bootable on the R.Pi? 2015-05-01 09:58:58 or how to turn that .tar.gz into a single image file 2015-05-01 10:10:18 darkfader: thanks, i'll add fontconfig and cache 2015-05-01 10:10:40 what has replaced slim? 2015-05-01 10:11:08 tonic: you just need to untar the pi image to a properly setup sd card and boot 2015-05-01 10:12:24 any pointers how to "properly setup" it? I was just trying to go through how to build an image by instructions here: https://www.raspberrypi.org/forums/viewtopic.php?p=625202#p625202 2015-05-01 10:12:59 for the sd card, fdisk a partition as W95 FAT32 and toggle boot flag on 2015-05-01 10:13:10 however hitting a problem between losetup and mkfs.vfat, since the losetup doesn't make available /dev/loop0p1 and p2 but only loop0 2015-05-01 10:13:22 oh, so you can boot? you just want to add to the image you install? 2015-05-01 10:13:49 I have previously used just an image burned to an sdcard using win32diskimager 2015-05-01 10:14:11 so I do have a booting sd card for RPi, but it's a different distro 2015-05-01 10:15:30 so you haven't been able to boot your cd card right? 2015-05-01 10:15:32 ScrumpyJack, I don't have an image file yet to use with alpinelinux .. and would like to learn the steps what do I need to do to put a .tar.gz package of the alpine linux to an sd card, including the partition setup stuff 2015-05-01 10:15:47 ah ok 2015-05-01 10:16:10 try this: for the sd card, fdisk a partition as W95 FAT32 and toggle the boot flag on 2015-05-01 10:20:08 yes the card actually already has that (from the previous older image I had on it) 2015-05-01 10:22:11 untar the pri to the SD card should be enough now 2015-05-01 10:23:58 since I have two partitions on the card, W95 FAT32 one and a Linux one (shown by fdisk).. 2015-05-01 10:24:12 I have to mount those two separately 2015-05-01 10:25:52 i don't know if it will work with two partitions 2015-05-01 10:25:59 aren't some of the files in the .tar.gz supposed to go to the fat partition and rest of them to a root partition for the actual linux? 2015-05-01 10:26:19 nope. all on one partition 2015-05-01 10:26:33 everything on a single fat32 partition? 2015-05-01 10:26:37 yup 2015-05-01 10:27:35 oh. should I still keep the other one, and only have a small fat32 partition (100M or so) 2015-05-01 10:28:57 i think the rpi bootloader boots bootcode.bin or something. i don't know how it looks for it 2015-05-01 10:29:50 but if it's in partition 1, i can't see why you couldn't create more partitions 2015-05-01 10:36:44 heh.. this fat32 partition was too small. have to clear up the card and start again.. 2015-05-01 10:48:34 success, I got it written and it boots now. thanks. :) 2015-05-01 11:31:22 tonic, ? 2015-05-01 11:31:30 alpine default image is run from tmpfs 2015-05-01 11:31:46 so just extract rpi .tar.gz to SDcard with one FAT32 partition 2015-05-01 11:31:48 that's it 2015-05-01 11:36:28 fabled, yes, I got that far... and was just wondering about the tmpfs as well & reading some info from the website... 2015-05-01 11:36:52 I guess the "Upgrading Alpine Linux on other removable media (such as CF/USB)" applies to this case? 2015-05-01 11:37:12 mostly yes. 2015-05-01 11:38:30 I've used some typical desktop linux distros, but this is a bit more unfamiliar to me, so it takes a while to understand what's going on.. 2015-05-01 11:41:58 got the network running and installed some packages (e.g. ssh).. but now hitting some small issues like it won't actually let me login remotely as the time is totally wrong and only lets me to set a new password. :) 2015-05-01 11:43:11 I guess I should first worry about getting the system to retain the stuff I change 2015-05-01 12:05:27 tonic: you need to run lbu commit to save your changes to the sd card 2015-05-01 12:05:54 or they'll be lost when you reboot 2015-05-01 12:06:44 anypone know what csync-owncloud does? 2015-05-01 12:19:20 (i'm looking for a webdav client to sync from alpine linux to a webdav server) 2015-05-01 12:26:45 ScrumpyJack what kind of changes are covered by the lbu commit and where does it exactly then save them.. in some file in the fat partition? 2015-05-01 13:43:00 tonic, by default lbu saves /etc 2015-05-01 13:43:29 if you wanna save /home/tonic for example. I need to do: lbu inc /home 2015-05-01 13:43:36 and then lbu commit 2015-05-01 13:45:51 tonic, lbu commit takes lists from etc/apk/protected_paths.d and there's several different ways it looks for differences 2015-05-01 15:44:52 o hai 2015-05-01 15:45:14 Does anyone know why could cups-libs not install libcups.so? 2015-05-01 15:45:40 openfbtd: is it in origin? 2015-05-01 15:46:00 origin? 2015-05-01 15:46:15 yes, the pkg its originates from 2015-05-01 15:46:29 its probably split 2015-05-01 15:46:38 from cups or whatever the name is 2015-05-01 15:48:06 It's in no cups-related package 2015-05-01 15:49:36 http://pkgs.alpinelinux.org/package/main/x86/cups-libs 2015-05-01 15:49:42 check origin 2015-05-01 15:50:13 I'm downloading the source and checking if it's there 2015-05-01 15:51:19 acrroding to pkgs it does have /usr/lib/libcups.so.2 2015-05-01 15:51:43 Yes. But I've installed the package and it's completely empty 2015-05-01 15:51:50 and the symlink libcups.so is in -dev 2015-05-01 15:51:58 which repo? 2015-05-01 15:52:01 Just realised that 2015-05-01 15:52:20 3.1/main 2015-05-01 15:52:42 -Oh 2015-05-01 15:52:44 http://nl.alpinelinux.org/alpine/v3.1/main/x86_64/cups-libs-2.0.2-r0.apk 2015-05-01 15:52:50 apk upgrade and it fixed everything 2015-05-01 15:52:56 :) 2015-05-01 15:53:14 I forgot that the container I built this one from was old 2015-05-01 15:54:37 ACTION is setting up a minecraft server on alpine :3 2015-05-01 15:55:37 clandmeter: is there something the project can do about booting in bhyve? https://pastee.org/vcpwf 2015-05-01 15:56:41 tonic - yes 2015-05-01 15:57:08 was anyone working on elasticsearch ? 2015-05-01 16:01:11 clandmeter: that is trying to boot the iso 2015-05-02 02:52:56 hello 2015-05-02 02:53:12 well first results with nix package manager are poor 2015-05-02 02:53:20 unfortunately 2015-05-02 02:53:32 gonna keep experimenting with it 2015-05-02 02:53:50 but it make u notice how the 'linux' ecosystem is so inbred with GNU 2015-05-02 02:54:05 it creates these ridiculous dependencies 2015-05-02 02:54:15 and wastes space 2015-05-02 02:59:59 hmm 2015-05-02 05:42:04 Greetings, I have this problem with a package, http://pastebin.com/nUCDR13X, maybe someone can help me, I did not locate the fault 2015-05-02 08:13:57 vitronic you have there php-dev als depends_dev? i dont see such APKGBUILD in repository 2015-05-02 08:15:02 crow, http://git.alpinelinux.org/cgit/aports/tree/testing/php-blenc/APKBUILD 2015-05-02 08:16:40 crow, Debian runs smoothly, it seems to be a problem in the package alpine 2015-05-02 08:17:04 thank you for your time :-D 2015-05-02 08:18:01 I am poor with C so I have not been able to find the problem 2015-05-02 08:23:38 well i am also learning the packages system :) 2015-05-02 08:24:38 i saw the Devs added oft the "linux-headers" as makedepends=" 2015-05-02 08:24:41 can you try that to 2015-05-02 08:25:07 even the stuff are from zend.. 2015-05-02 08:25:40 perhaps this problem has happened before with some other program, I would be interested to solve 2015-05-02 08:27:01 here you can see requeres http://www.rpmfind.net//linux/RPM/remi/fedora/20/i386/php-pecl-blenc-1.1.4-0.1.b.fc20.remi.5.5.i686.html 2015-05-02 08:27:10 the truth I do not think it's a problem abuild, but still try 2015-05-02 08:27:40 well as i can see its your develop system you could try adding linux-headers and so on 2015-05-02 08:27:54 or wait for someone with more experience :) 2015-05-02 09:07:43 vitronic: could be it need to be rebuild against later php? 2015-05-02 09:08:39 I tried not to think about it :-( 2015-05-02 09:09:16 develop:/usr/src/blenc-1.1.4b# php -v 2015-05-02 09:09:16 PHP 5.6.5 (cli) (built: Jan 23 2015 14:21:11) 2015-05-02 09:09:41 I'm using the native version of alpine 2015-05-02 09:10:27 which version? stable or edge? 2015-05-02 09:10:54 stable 2015-05-02 09:11:13 develop:/usr/src/blenc-1.1.4b# cat /etc/alpine-release 2015-05-02 09:11:13 3.1.3 2015-05-02 09:11:40 but that pkg is only in edge? 2015-05-02 09:12:17 if but I have compiled in stable 2015-05-02 09:13:07 see also unstable package and has the same problem 2015-05-02 09:13:36 can you try using unstable? 2015-05-02 09:13:42 system is edge 2015-05-02 09:13:42 so I assume the problem is not my alpine 2015-05-02 09:14:30 I downloaded the package edge and you look ldd with the same results 2015-05-02 09:15:05 you are mixing stable and edge 2015-05-02 09:15:22 its possible some deps have upgraded in edge and is not in stable 2015-05-02 09:17:28 you said you compiled it yourself on 3.1 and it compiled fine? 2015-05-02 09:18:19 hmm 2015-05-02 09:18:50 looks like linking doesnt properly work 2015-05-02 09:22:15 vitronic you could try edge iso in one VM and check if that package works there. 2015-05-02 09:23:51 clandmeter, I compiling the package in stable 2015-05-02 09:24:10 I have not mixed nothing 2015-05-02 09:24:53 crow, I'll try 2015-05-02 09:25:41 crow, Where downloaded the iso unstable? 2015-05-02 09:26:07 http://nl.alpinelinux.org/alpine/edge/releases/x86_64/ 2015-05-02 09:26:31 i used the alpine-edge-150306-x86_64.iso 2015-05-02 09:26:48 i am not sure on which arch you are? x86_64 ? 2015-05-02 09:27:19 x86 for virtualbox 2015-05-02 09:27:38 ready 2015-05-02 09:27:39 alpine-edge-150306-x86.iso 2015-05-02 09:27:44 http://nl.alpinelinux.org/alpine/edge/releases/x86/ 2015-05-02 09:27:59 after install apk update && apk upgrade to be up2date 2015-05-02 09:28:46 Thanks, I'll try 2015-05-02 09:31:43 vitronic: the package is broken 2015-05-02 09:31:53 atleast thats what i think 2015-05-02 09:33:08 clandmeter, I agree, I'm trying to fix the problem 2015-05-02 09:34:16 I need blanc to run an application in alpine 2015-05-02 09:34:21 blenc* 2015-05-02 15:03:30 anyone has an idea why nginx reverse proxy with a lighthttpd backend would get slow after time? 2015-05-02 22:18:27 hello, is this the official alpine linux channel? 2015-05-02 22:18:44 <_ikke_> yumbox: it is 2015-05-02 22:19:04 can I mention a bug here? 2015-05-02 22:19:24 <_ikke_> You can, but adding it to bugs.alpinelinux.org is much more effective 2015-05-02 22:20:25 do I need to create an account for that? 2015-05-02 22:21:05 <_ikke_> It's supposed to support openid 2015-05-02 22:23:29 maybe it's already known bug, but if an IO error occurs during the testing of the mirrors at the installation, it still adds that mirror 2015-05-02 22:24:05 i.e. IO error occurs -> (faulty) mirror get's super low score -> (faulty) mirror get's added to repo config 2015-05-02 22:27:21 also, a timeout limit of something like 10 would be nice (one mirror took 24 sec) 2015-05-02 23:51:53 dalias: hey I've been trying to update aport unionfs-fuse to version 1.0 but it fails compilation because _IOC_SIZE() is not defined 2015-05-02 23:52:09 dalias: i found it defined in /usr/iclude/asm-generic/ioctl.h 2015-05-02 23:52:44 but that seemed like the incorrect thing to patch and add in to the unionfs.c file that needs the extra #include to have _IOC_SIZE defined 2015-05-02 23:53:59 hmm 2015-05-02 23:54:06 what are they using it for? 2015-05-02 23:54:54 lemme go check 2015-05-02 23:55:36 that line calls set_debug_path(debug_path, _IOC_SIZE(cmd)); 2015-05-02 23:55:52 which doesn't explain much but I'm going to try including it real quick and see if ti blows up or works :P 2015-05-02 23:56:00 thanks for quick response :D 2015-05-02 23:56:32 line 294 in src/unionfs.c 2015-05-02 23:56:44 my guess is that the use is rather nonsense :-p 2015-05-02 23:56:54 dalias: https://github.com/rpodgorny/unionfs-fuse/archive/v1.0.tar.gz 2015-05-02 23:57:05 probably :P 2015-05-02 23:57:06 is there a browsable online repo? 2015-05-02 23:57:24 yah 2015-05-02 23:57:46 dalias: github bro :P 2015-05-02 23:57:47 https://github.com/rpodgorny/unionfs-fuse/blob/master/src/unionfs.c 2015-05-03 00:03:31 dalias: that seems to make it compile well 2015-05-03 00:04:03 just including it, but I would like to investigate further to make sure it isn't just a shitty patch 2015-05-03 00:11:31 I'm trying to compile a program, but it uses ld -lc, which gives the error "cant find -lc" 2015-05-03 00:11:35 What's at fault here? 2015-05-03 00:17:52 the ld command is not the right way to link 2015-05-03 00:18:08 unless you're doing freestanding work like making a kernel or something 2015-05-03 00:18:28 the $CC command (e.g. gcc) is the right way to link with system libs for a hosted C program 2015-05-03 00:20:42 dalias: ah, I see. 2015-05-03 00:21:03 so how do I compile the program then? The program instructs me to compile with cmake 2015-05-03 00:21:58 systmkor, ok you found a major bug :) 2015-05-03 00:22:03 in their code 2015-05-03 00:22:23 set_debug_path takes a string and a string length as arguments 2015-05-03 00:22:43 some fool thought _IOC_SIZE(cmd) is the length of the string passed as an arg to the ioctl command 2015-05-03 00:23:05 this is akin to thinking char *s; sizeof(s) will tell you the length of a string :) 2015-05-03 00:23:44 in this case, cmd==UNIONFS_SET_DEBUG_FILE 2015-05-03 00:23:52 so _IOC_SIZE(cmd) is _IOC_SIZE(UNIONFS_SET_DEBUG_FILE) 2015-05-03 00:24:20 it's some number hard-coded into the ioctl command value for UNIONFS_SET_DEBUG_FILE 2015-05-03 00:25:33 hmm it's not that bad 2015-05-03 00:25:40 the ioctl actually does take a fixed-size buffer 2015-05-03 00:25:49 of size PATHLEN_MAX whatever that is 2015-05-03 00:26:00 so the code should just be using PATHLEN_MAX 2015-05-03 00:26:04 see https://github.com/rpodgorny/unionfs-fuse/blob/master/src/uioctl.h 2015-05-03 00:26:28 Apparently, I just needed to install alpine-sdk ? 2015-05-03 00:26:45 yumbox, i doubt that's the package you need 2015-05-03 00:27:02 well, I installed it and it didn't give that error anymore. 2015-05-03 00:27:05 probably need musl-dev (for libc dev files) and -dev packages for any other libs you use/need 2015-05-03 00:27:21 ah, well alpine-sdk probably has a dep on musl-dev and any other core lib dev packages 2015-05-03 00:29:51 yeah, alpine-sdk pulls in musl-dev and libc-dev 2015-05-03 00:32:23 dalias: okay cool, uhm give me a minute, i'll try to write a fix and send to them, unless since you really found that it's a bug you can 2015-05-03 00:32:29 yah it was confusing me when I started to look at it 2015-05-03 00:32:30 :P 2015-05-03 00:32:35 more closely 2015-05-03 00:35:33 yeah i don't think it's a practical bug 2015-05-03 00:35:45 but a major code smell 2015-05-03 00:35:52 i wonder why the size is even passed anyway 2015-05-03 00:36:08 perhaps there's a risk it's not null terminated 2015-05-03 00:37:54 well i'm confuddled because it checks size of cmd but cmd is in the switch case which is checked by UNIONFS_SET_DEBUG_FILE 2015-05-03 00:38:31 well i guess maybe I should do a light code audit before trying to get the version bump up on aports 2015-05-03 00:48:22 when I run autogen.sh do I need to add --prefix=/usr ? 2015-05-03 00:52:47 Apparently it installs to /usr/local by default. 2015-05-03 00:52:57 I guess that's fine too. 2015-05-03 00:55:03 dalias: well I submitted an issue, hopefully it gets fixed soon 2015-05-03 00:57:15 autogen.sh is not where you pass --prefix 2015-05-03 00:57:15 yumbox: if you check out the generated APKBUILD using newapkbuild using the option '-a' it will generate a decent reference point of what autoconf configurations you want 2015-05-03 00:57:19 configure is where you pass it 2015-05-03 00:57:40 for making packages, you need --prefix=/usr and maybe other options 2015-05-03 00:57:51 for installing outside the package management framework, the default is right 2015-05-03 01:01:05 APKBUILD? I'm not trying to make an apk package. I'm just trying to install some program. 2015-05-03 01:01:31 yumbox: oh okay, my bad 2015-05-03 01:01:57 systmkor: no, it's still helpful (for when I do want to make an apk package) 2015-05-03 01:10:03 CMake gives an error that REPLACE requires 4 arguments. 2015-05-03 01:10:35 is the CMake in alpine that different from the CMake in other distros? 2015-05-03 01:13:47 shouldn't be 2015-05-03 01:21:09 well, I did the exact same commands on Arch linux, and that worked fine. 2015-05-03 01:22:36 line 14 of this file is the offender here: https://github.com/Cloudef/wlc/blob/master/CMake/Findlibinput.cmake 2015-05-03 01:23:30 Is it because CMake is version 3.2.2 on Arch, but version 3.0.2 on Alpine? 2015-05-03 01:23:44 maybe 2015-05-03 01:24:06 How can I install a newer version of CMake? 2015-05-03 01:24:52 you can build it yourself from source 2015-05-03 01:25:08 is there a testing repo? 2015-05-03 01:26:13 yes but i dunno if cmake is in it 2015-05-03 01:26:54 hmm edge has 3.2.2 already 2015-05-03 01:27:03 what ver of alpine are you using? 2015-05-03 01:28:05 I don't know. 2015-05-03 01:28:30 3.1.3 I think? 2015-05-03 01:29:26 if you don't want to move to edge in general you could add a tagged repo for edge and then apk add cmake@edge 2015-05-03 01:29:50 is edge unstable? 2015-05-03 01:29:56 (in your experience) 2015-05-03 01:30:23 I'm using Alpine on the desktop, and if a program breaks, it's not that big of a deal for me. 2015-05-03 01:31:38 it's been ok for me 2015-05-03 01:32:07 How do I move to edge then? (in general) 2015-05-03 01:32:45 just edit /etc/apk/repositories and do apk update, apk upgrade ? 2015-05-03 01:35:46 oh, I will just follow this http://wiki.alpinelinux.org/wiki/Include:Upgrading_to_Edge 2015-05-03 01:42:49 okay, so I upgraded to Edge and got CMake version 3.2.2 2015-05-03 01:42:54 error is still the same :/ 2015-05-03 01:43:30 how is that even possible? 2015-05-03 01:48:46 no idea 2015-05-03 01:48:56 i hate cmake :-p 2015-05-03 02:45:28 yuuup 2015-05-03 02:45:57 dalias: what are you up to lately? 2015-05-03 07:07:44 systmkor, hi 2015-05-03 07:08:29 I also found a ref in comments for , _IOC_SIZE(cmd) in fuse-2.9.3 src pkg 2015-05-03 07:09:31 2. I would wait for next release for next release of unionfs as it is going to revert some binary name changes it did in v1.0 2015-05-03 07:10:12 https://github.com/rpodgorny/unionfs-fuse/commit/50ebad1b47c727e3fe3ddb16eda76455346140b4 2015-05-03 07:34:53 unless plan to use current master snapshot 2015-05-03 09:27:28 hi 2015-05-03 09:27:50 how can I switch from package@testing to package@edge or from package@edge to package? 2015-05-03 09:35:04 shafire http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management 2015-05-03 09:35:15 crow: where exactly? 2015-05-03 09:35:32 you could try repositories pinning 2015-05-03 09:35:50 I have it already 2015-05-03 09:35:53 how can I switch now? 2015-05-03 09:36:09 I have package app@testing installed that switched to edge 2015-05-03 09:40:32 shafire well then just: apk update && apk add packagename@testing 2015-05-03 09:41:08 I think the package is broken, see #alpine-devel 2015-05-03 09:41:16 thats what drives me crazy :D 2015-05-03 10:43:21 hey all, what would I need to do to get e.g. /dev/ttyACM0 owned by a different group when it is created? 2015-05-03 10:43:46 aha, mdev 2015-05-03 10:43:47 as you were :) 2015-05-03 11:56:34 so what's the fix for the icu-libs and libvpx conflicts on edge ? 2015-05-03 11:59:33 seems like apk del -f mplayer wasn't the solutiong 2015-05-03 12:00:19 http://pastebin.com/6B0y5ar9 ooops 2015-05-03 12:00:39 oops 2015-05-03 12:03:30 any idea ? 2015-05-03 12:05:00 Not me, I'm just trying to get Alpine to run under bhyve 2015-05-03 12:11:28 seems fixed :O 2015-05-03 17:56:19 Is there no way to install AlpineLinux without being able to write to a CD/DVD? I have a reader but not a burner. DD'ing the ISO to my flash drive doesn't work as it usually does and the only documentation on the site seems to suggest I HAVE to be able to burn the ISO to a cd/dvd.... 2015-05-03 18:08:50 <_ikke_> argv7: unetbootin can write it to a usb key 2015-05-03 18:09:49 <_ikke_> The iso is not hybrid like other ISOs so that you can both write them to optical media and usb disks 2015-05-03 18:10:05 <_ikke_> argv7: Note that, once you've created one usb key, you can dd that to an img and reuse it 2015-05-03 18:12:05 _ikke_: Thanks, I'll try unetbootin right now. 2015-05-03 21:08:32 fireglow: cool project, how close to working is it? :D 2015-05-03 21:10:04 ex-parrot: well, it's not something I have to do, rather somebody working on the project. Something panics the kernel. I'm not sure what, gentoo boots fine. 2015-05-03 21:10:17 fair enough 2015-05-03 21:10:29 I am really interested in bhyve, but there isn't much documentation of folks experiences online yet 2015-05-03 21:11:04 I spent some time over the weekend setting up alpine on a raspberry pi and I am so impressed! I am going to need to find more places to use alpine :D 2015-05-04 06:23:39 Hey, I'm trying to get nodejs running on alpine edge in x86_64, but it fails to run as follows: 2015-05-04 06:23:41 Error relocating /usr/lib/libstdc++.so.6: get_nprocs: symbol not found 2015-05-04 06:24:41 If I force libstdc++ *and* nodejs to be installed from latest-stable it works, but naturally that's not really what I'd like to be doing 2015-05-04 06:24:51 is there any quick workaround I can apply here? 2015-05-04 06:28:43 bawr: sounds like you are mixing edge repo with v3.1 2015-05-04 06:29:07 ncopa: no, nodejs fails to run when I *only* have edge 2015-05-04 06:29:20 hum 2015-05-04 06:29:26 ok, thats bug then 2015-05-04 06:29:43 i'll try reproduce here 2015-05-04 06:30:17 thanks 2015-05-04 06:30:39 (1/2) Installing libuv (1.4.2-r0) 2015-05-04 06:30:39 (2/2) Installing nodejs (0.12.2-r0) 2015-05-04 06:30:46 $ node 2015-05-04 06:30:46 > 2015-05-04 06:30:58 the hell. 2015-05-04 06:31:07 $ node --version 2015-05-04 06:31:07 v0.12.2 2015-05-04 06:31:17 bawr: try: apk upgrade -U -a 2015-05-04 06:31:22 or: apk fix 2015-05-04 06:35:12 ncopa: ah, -U -a fixed it, it was for a docker image actually, and before I was just doing "apk-install --update ca-certificates git go nodejs" 2015-05-04 06:35:32 chalk one up to me not reading the docs carefully enough ;) 2015-05-04 06:56:06 the v3.2 builders for x86 and x86_64 are done 2015-05-04 06:56:18 nice 2015-05-04 06:56:25 can i get a beer now? 2015-05-04 06:56:26 it should be possible to start use v3.2 repos 2015-05-04 06:56:35 clandmeter: you definitively deserve one 2015-05-04 06:56:43 but its a bit early for that :) 2015-05-04 06:56:52 i mean... a beer monday morning? :) 2015-05-04 06:56:56 with my back pain, it cant be too early ;-) 2015-05-04 06:57:29 back pain... then maybe whiskey would be better? 2015-05-04 06:58:17 ill have stomache pain instead 2015-05-04 07:01:19 ncopa: nice, so when do 3.2 repos get official? 2015-05-04 07:06:08 they are 2015-05-04 07:06:13 from now 2015-05-04 07:08:04 ncopa: funny thing is, I was browsing the dl server earlier today and saw 3.2 with no mention on the site, assumed those were unstable and got all confused since they weren't mentioned on the homepage at all 2015-05-04 07:09:20 currently v3.2 is a clone of edge 2015-05-04 07:10:13 but once v3.2 is released, edge will move forward, wile v3.2 will only get fixes 2015-05-04 08:13:12 ncopa: statd doesnt go to the background 2015-05-04 08:13:14 http://sprunge.us/WEGH 2015-05-04 08:15:17 "to a shutdown PID 1 process and finally a transition back to the initial ramdisk so that we can unmount the root file system even" 2015-05-04 08:15:21 interesting, http://lkml.iu.edu/hypermail/linux/kernel/1504.2/00241.html 2015-05-04 14:36:25 will v3.2 updates have separate folder on repos ? 2015-05-04 15:05:19 Has anybody had a look at the panic when booting the iso under bhyve? https://pastee.org/vcpwf 2015-05-04 15:13:02 fireglow: you should send this to freebsd-virt ml. AFAICT it seems like the hd emulation or virtio disk is not working properly 2015-05-04 15:13:12 royger: okay, thanks, I will 2015-05-04 15:13:28 I'm already talking to people on #bhyve, but it's rather low-traffic. 2015-05-04 15:13:34 I'll post to the ML later 2015-05-04 16:55:08 what is the root= setting for booting the iso? 2015-05-04 17:02:32 /dev/sro ? 2015-05-04 17:02:41 /dev/sr0 2015-05-04 17:14:32 fireglow: if you are using bhyve it's probably /dev/vd* because it will be attached using virtio 2015-05-04 19:36:14 meeh, I can't figure it out 2015-05-04 19:36:23 tried all of your suggestions 2015-05-04 19:48:22 virtio would be /dev/vda 2015-05-04 19:52:57 BitL0G1c: same issue. I've tried: /dev/cd0 /dev/cdrom /dev/vda /dev/vda0 /dev/vd0 /dev/sr0 2015-05-04 19:54:49 https://pastee.org/kebmg 2015-05-04 20:17:44 fireglow In order to boot operating systems other than FreeBSD, the sysutils/grub2-bhyve port must be first installed. http://www.freebsd.org/doc/en_US.ISO8859-1/books/handbook/virtualization-host-bhyve.html 2015-05-04 20:18:46 BitL0G1c: yeah it is, this is the panic I'm hitting: https://pastee.org/vcpwf 2015-05-04 20:20:01 according to some people in #bhyve, it's a matter of getting the root= right 2015-05-04 20:22:48 /dev/vda1 ? 2015-05-05 08:46:15 morning 2015-05-05 09:14:35 quiet here 2015-05-05 09:16:28 morn 2015-05-05 09:16:32 ay 2015-05-05 09:16:45 so tired i should head straight back home 2015-05-05 09:16:55 but alpine is not guilty ;) 2015-05-05 09:17:46 hey 2015-05-05 09:17:51 what's up guys 2015-05-05 09:18:10 what's alpine users' vt of choice? 2015-05-05 09:18:58 well i'm actually lfs user in limbo 2015-05-05 09:19:20 haha 2015-05-05 09:19:21 i took a detour and amde an lfs box and tried nix package manager on it 2015-05-05 09:19:42 after like 3 daz failing to install xfce i realized the nix project is futile... needs new leadership or sumt 2015-05-05 09:19:53 oh 2015-05-05 09:20:05 as for alpine, i never got that far... but the one running in vmware has lxde i think , so the default for that 2015-05-05 09:20:06 you just saved me some time i'd have put into playing with nixos 2015-05-05 09:20:20 and i reallly want to install lxqt on my next desktop 2015-05-05 09:20:25 oh dude 2015-05-05 09:20:36 got i lost so much time 2015-05-05 09:21:04 perfect criticism of Nix i just found like 30 mins ago 2015-05-05 09:21:13 after failing to build X after like 50 tries 2015-05-05 09:21:20 "The heart of the matter is that unix (ostensibly[0]) plays host to many different philosophies of how to go about programming. Languages differ in their approach to build systems, package management, syntax and foreign function interfaces. To change the way many software packages are built in a semi-automated, non-ast-walking fashion is insane. Not to discuss this in the design documents is insane. NixOS has 46 contributors and the 2015-05-05 09:21:20 documentation is sufficiently general as to give each contributor his own wildly different interpretation of the scope of the project. As far as I can tell, were one to extrapolate from the given information to a set of concrete requirements we see that NixOS plans to rewrite the build scripts for every version of every project on unix. Again, this is insane. The correct thing to do in this situation is to realize the utter impossibility of 2015-05-05 09:21:21 the task that has been set forth and re-evaluate one's approach." 2015-05-05 09:21:24 https://github.com/NixOS/nixpkgs/issues/4952 2015-05-05 09:21:52 --^ nix is a great idea, but has poor planning , leadership, execution, and usability 2015-05-05 09:22:02 thx :) 2015-05-05 09:22:03 and it's scope is not defined 2015-05-05 09:22:15 i'll talk apk any day 2015-05-05 09:22:20 yes! 2015-05-05 09:22:24 im serious 2015-05-05 09:22:38 ever since I've backed out of a xen host upgrade and it just managed that... 2015-05-05 09:22:40 and apk is decent.. not a breakthrough 2015-05-05 09:22:43 but it's respectable 2015-05-05 09:22:53 and after nix , apk seems like an oasis 2015-05-05 09:23:10 apk ~= apt-get in my mind anyway 2015-05-05 09:23:24 apk is more robust imo 2015-05-05 09:23:31 and faster 2015-05-05 09:23:35 well there's usability issuezz with it 2015-05-05 09:23:38 that's my poblem 2015-05-05 09:23:46 i agree it's built like a tank 2015-05-05 09:23:54 the code is incredibly impressive and organized 2015-05-05 09:24:00 i now cuz i wrote a patch to fix the usability haha 2015-05-05 09:24:01 rolling over our packages 2015-05-05 09:24:02 err 2015-05-05 09:24:03 :) 2015-05-05 09:24:34 but yeah i think apk is on a way better path than nix 2015-05-05 09:24:56 i need a decent vt 2015-05-05 09:24:59 also i told ncopa i would research nix to see if it was worthwhile for alpine... if anyone talks to him , say he should never let nix into alpine 2015-05-05 09:25:01 haha 2015-05-05 09:25:09 vt 2015-05-05 09:25:13 what do u mean 2015-05-05 09:25:20 like mate-terminal kinda thing 2015-05-05 09:25:22 terminal emulater 2015-05-05 09:25:31 what's the use case 2015-05-05 09:25:35 COM port 2015-05-05 09:26:02 terminal in X 2015-05-05 09:26:11 yeah k 2015-05-05 09:26:17 i just found one the other day that looked awesome 2015-05-05 09:26:19 lemme see ... 2015-05-05 09:26:22 mrxvt isn't doing it for me 2015-05-05 09:26:56 bitshark__: i did talk a bit with fabled about nix 2015-05-05 09:27:04 he was very sceptic 2015-05-05 09:27:04 here we go 2015-05-05 09:27:05 http://finalterm.org/ 2015-05-05 09:27:21 i havent tried it, but i found the homepage when was running into locale/utf bugs 2015-05-05 09:27:32 andi bookmarked it cause t looked pretty awesome 2015-05-05 09:27:40 what nix does is lets you install parallell versions of everthing - instead of fix the dependencies 2015-05-05 09:27:57 hey ncopa! 2015-05-05 09:27:58 so you end up with a separate copy of everything 2015-05-05 09:28:02 hi bitshark__ 2015-05-05 09:28:02 long time no see 2015-05-05 09:28:13 well the idea 'in general' is fine 2015-05-05 09:28:16 final term looks cool 2015-05-05 09:28:17 very general terms 2015-05-05 09:28:24 but in specifics it's a nightmare 2015-05-05 09:28:36 and in practice (what i did for a week or two) it's just insane 2015-05-05 09:28:44 heh 2015-05-05 09:28:45 ok 2015-05-05 09:28:45 it's completely unusable outside nixos 2015-05-05 09:28:50 good to know 2015-05-05 09:28:54 2nd time i see finalterm recommended 2015-05-05 09:29:01 seems it's on the way up 2015-05-05 09:29:08 so i think your best bet is improve apk 2015-05-05 09:29:23 and take the best ideas from other places and slowly incorporeate them 2015-05-05 09:29:29 yes, thats what think too 2015-05-05 09:29:42 ncopa, what good vt emulators are in alpine linux? 2015-05-05 09:29:46 like one idea i had (maybe this exists) is it'd be nice to be able to automaticaly populate apk from gentoo portage 2015-05-05 09:29:48 or another repo 2015-05-05 09:30:07 dunno if it's practical or not but i think it'd be cool 2015-05-05 09:30:07 ScrumpyJack: as in x11 terminal? i use xfce4-terminal 2015-05-05 09:30:15 I've finally made the just to alpine linux as my work desktop 2015-05-05 09:30:23 lets start with a defined package submission process, ideally automated :) 2015-05-05 09:30:42 well , now i'm once again torn between my lfs build and alpine 2015-05-05 09:30:47 i will probably install alpine tho 2015-05-05 09:30:59 cause i was counting on nix to not be a complete fail 2015-05-05 09:31:03 haha 2015-05-05 09:31:19 that's ok tho, apline is 100x more secure than my lfs build 2015-05-05 09:31:30 btw quick question.... does alpine use initramfs or no 2015-05-05 09:31:39 it uses initramfs 2015-05-05 09:31:45 k 2015-05-05 09:31:52 INITRD initramfs-grsec 2015-05-05 09:31:58 is there a reason there is no alpine 'live' desktop 2015-05-05 09:32:04 like philosophical reason 2015-05-05 09:32:12 keep-it-simple 2015-05-05 09:32:13 or just hadnt gotten to it 2015-05-05 09:32:17 okay 2015-05-05 09:32:34 the idea is you boot live cd/usb 2015-05-05 09:32:37 that's usualy a good policy but not always 2015-05-05 09:32:44 but i can see that 2015-05-05 09:33:00 it's very hard to probe for hardware from initrd and get it right 2015-05-05 09:33:06 from there it should be super easy to bring up a live system 2015-05-05 09:33:13 bitshark__: pulling a desktop together in alpine is a labour of love 2015-05-05 09:33:24 haha i know 2015-05-05 09:33:33 my first attempt at lxde failed 2015-05-05 09:33:50 i was like 'ill build an alpine lxde package today' 2015-05-05 09:33:52 no way 2015-05-05 09:33:56 i like that you have to pull your components together, so you get exactly what you want 2015-05-05 09:33:56 haha 2015-05-05 09:34:03 alpine is designed to make it easy to run 'live' systems 2015-05-05 09:34:25 i'd like a script so you can: setup-box lxde 2015-05-05 09:34:27 or similar 2015-05-05 09:34:29 OH ncopa i gotta tell u about this awesome innovation i found 2015-05-05 09:34:57 so im sure u know how many live distros use like aufs , overlayfs, or unionfs 2015-05-05 09:35:04 to make a coherent filesystem 2015-05-05 09:35:12 ncopa: setup-box would be awesome 2015-05-05 09:35:19 well i was playing around with those , and i started to look into cacheing 2015-05-05 09:35:25 ScrumpyJack: alacerda is working on it 2015-05-05 09:35:27 because the usb stick was slow for writes 2015-05-05 09:35:42 alacerda: you about? 2015-05-05 09:36:00 using a ramdisk with periodic flush()ing, i was able to get IO to the usb up by like 10x or more 2015-05-05 09:36:41 bitshark__: apine use tmpfs (diskless) and has a tool to save the config to a special apkovl.tar.gz 2015-05-05 09:36:55 well whati mean is i found a kernel module 2015-05-05 09:36:58 that is _really_ good at it 2015-05-05 09:37:10 hum 2015-05-05 09:37:12 i benchmarked it against bcache, flashcache, and dmcache 2015-05-05 09:37:14 and its faster 2015-05-05 09:37:39 at the expense of maybe clobbering the journal sometimes 2015-05-05 09:37:41 that's the tradeoff 2015-05-05 09:38:23 how is that different from running on ext4 or similar directly - and tweak the settings to make it flush very seldom? 2015-05-05 09:38:30 i guess your normal workload is so 'bursty', and linux page cache is so preoccupied that dedicated kernel module for usb IO makes it run like a ferrari 2015-05-05 09:38:43 i'm not sure 2015-05-05 09:38:58 interesting concept though 2015-05-05 09:39:02 i think because it's in kernel mode it doesn't have to have asmuch fh open 2015-05-05 09:39:09 so 'less overhead' 2015-05-05 09:39:30 what alpine does is run completely diskless 2015-05-05 09:39:31 i need to look at the code cuz i dont know if tis special 2015-05-05 09:39:40 yeah it's same idea 2015-05-05 09:39:45 and 'lbu commit' to save the state 2015-05-05 09:39:48 but it works on block device 2015-05-05 09:40:02 next reboot it installs system from scratch into tmpfs 2015-05-05 09:42:13 okay it's called "EPRD" for Eventually Persistant Ramdisk 2015-05-05 09:42:26 aand according to my tests with fio its faster than flashcache haha 2015-05-05 09:42:30 http://www.lessfs.com/wordpress/?p=699 2015-05-05 09:42:34 https://github.com/borland667/eprd 2015-05-05 09:43:16 u can use it on any block device too 2015-05-05 09:43:26 so its pretty flexible 2015-05-05 09:44:18 anyway how u been ncopa 2015-05-05 09:44:22 life okay? 2015-05-05 09:44:24 how do you list package dependancies? search -r gives me reverse 2015-05-05 09:44:39 (i should know this by now) 2015-05-05 09:44:58 bitshark_: im good thanks 2015-05-05 09:45:17 i'm desperately trying to get alpine v3.2 out the door 2015-05-05 09:45:50 i'm in the kill-the-engineers-and-just-ship-the-product-mode 2015-05-05 09:46:22 apk info -r 2015-05-05 09:46:33 apk info -R 2015-05-05 09:46:48 yes! 2015-05-05 09:46:55 ship first, ask questions later 2015-05-05 09:50:31 ncopa is it hard to translate packages from gentoo to alpine 2015-05-05 09:50:31 like from src 2015-05-05 09:50:47 or is it pretty straightforward and stable 2015-05-05 09:51:05 its not trivial to do it programatically 2015-05-05 09:52:11 k 2015-05-05 09:52:15 i meant any old way 2015-05-05 09:52:39 i'm deciding whether to install alpine or gentoo as base distro since by lfs/nix combination wont work 2015-05-05 09:53:03 i suppose i can always makea apk packages from src by hand 2015-05-05 09:53:11 if there's stuff not in the repos that i need 2015-05-05 09:53:46 no , so i meant by hand... i think src-> src translation like np complete haha 2015-05-05 09:53:50 we used to use gentoo to builds alpine 2015-05-05 09:54:03 alpine 1.8 and earlier was built with gentoo 2015-05-05 09:54:11 oh okay, rad 2015-05-05 09:55:03 but i realized that was more work than just build our own 2015-05-05 09:55:14 i see 2015-05-05 09:55:15 we also looked into use pacman and arch linux scripts 2015-05-05 09:55:16 interesting 2015-05-05 09:55:26 and ended up with something pretty similar 2015-05-05 09:55:58 i have a script that takes fedora spec file and generate an APKBUILD template 2015-05-05 09:58:19 oh okay so ncopa i ran into issues installing enlightenment from testing branch 2015-05-05 09:58:37 somehow apt put a bunch of identically named packages side byside 2015-05-05 09:58:47 shouldnt there be like a warning in that case 2015-05-05 09:58:49 heh 2015-05-05 09:59:05 cuz it took me awhile to realize i had clobbered the libs 2015-05-05 09:59:14 i think enlightenment is a bit outdated 2015-05-05 09:59:21 yeah it is 2015-05-05 09:59:28 i'm not using it after all 2015-05-05 09:59:50 i use xfce 2015-05-05 10:00:01 anyway why dont i just pull in pull request for the patches i made for apt 2015-05-05 10:00:08 and ifu likethem u can merge 2015-05-05 10:00:25 might be after v3.2 release 2015-05-05 10:00:26 if you dont i ccan keep them 2015-05-05 10:00:35 well yeah whenever 2015-05-05 10:00:58 i'll have a lunch now 2015-05-05 10:01:00 bbl 2015-05-05 10:01:00 i just wanted to share them to see if other ppl likethe ideazornot 2015-05-05 10:01:02 okay later 2015-05-05 10:16:56 ncopa: uk.bmap.gz seems not to work in edge. loadkmap < /etc/keymap/uk.bmap loads us keymap ("@~><\ are all in the wrong place) 2015-05-05 11:12:29 ScrumpyJack, about the setup-box: Yesterday i made some tests and I could install MATE desktop just doing: setup-box -d mate 2015-05-05 11:12:32 worked perfectly 2015-05-05 11:13:08 I am finishing some small things on it and will create its first aport version 2015-05-05 11:14:27 bitshark__, ^^ 2015-05-05 11:52:09 alacerda: awesome. can i take a look? is it written in a way that it's easy to add "desktop evironments" ? 2015-05-05 11:53:46 ScrumpyJack, sure... http://git.alpinelinux.org/cgit/alacerda/setup-box/ 2015-05-05 11:54:45 it is writing in a way that any one can add json files and have a new group 2015-05-05 11:55:27 i see that. 2015-05-05 11:56:26 i notice you pull all the video drivers down. no probing yet? 2015-05-05 12:00:51 not yet :( 2015-05-05 12:01:25 it is a TODO for a long time on setup-xorg-base 2015-05-05 12:01:40 indeed :) 2015-05-05 13:24:47 fabled: it's like an explosion of opennhrp activity! 2015-05-05 13:24:55 are there a lot of people using it? 2015-05-05 13:25:09 I suspect that there are probably relatively few users but some of them are probably really big users 2015-05-05 13:25:46 johnf1911, i'm not certain. i know a handful of big installs. 2015-05-05 13:25:54 but i don't know all the details 2015-05-05 13:26:07 there's also indirect users, as Vyatta uses opennhrp too 2015-05-05 13:26:27 the project I'm working on will be 836 devices 2015-05-05 13:26:31 (spokes) 2015-05-05 13:27:21 cool 2015-05-05 15:15:44 anyone recently treid to create an ubuntu container in alpine lxc? 2015-05-05 15:19:39 clandmeter - I can create one now to see if I have a problem - which ubuntu release ? 2015-05-05 15:25:13 ubuntu seems to fail http://hastebin.com/azebixopad.vhdl - I've run debian lxc containers in alpine without any problems 2015-05-05 15:33:47 clandmeter - 'mkdir /usr/share/keyrings && cd /usr/share/keyrings && wget /usr/share/keyrings 2015-05-05 15:34:15 extract the keys & lxc-create works for ubuntu 2015-05-05 15:35:06 http://hastebin.com/qoqexasele.vhdl 2015-05-05 15:35:35 *** wget http://archive.ubuntu.com/ubuntu/pool/main/u/ubuntu-keyring/ubuntu-keyring_2012.05.19.tar.gz 2015-05-05 15:38:26 BitL0G1c: yes i have that 2015-05-05 15:38:35 i was going to package those key's 2015-05-05 15:39:33 BitL0G1c: i was trying this: lxc-create -o lxc.log -t ubuntu -n ubuntu -- -r trusty -d 2015-05-05 15:39:42 but it fails very late 2015-05-05 15:40:44 yes http://hastebin.com/atotagunol.vbs 2015-05-05 15:41:05 <_ikke_> If you adjust a package with extra build flags and an extra dependency, is that considered an release update? 2015-05-05 15:41:05 yes 2015-05-05 15:42:25 BitL0G1c: probably soemthing in the template that goes worng 2015-05-05 15:43:10 yes - going to try building on a centos7 kvm host - see if it crashes there too 2015-05-05 15:43:51 dinner is ready 2015-05-05 15:44:00 ill be back shortly if the wife lets me :) 2015-05-05 15:44:44 :-D bon appetite 2015-05-05 16:13:58 BitL0G1c: which version of lxc are you using? 2015-05-05 16:14:21 1.0.7 2015-05-05 16:14:38 you are getting a diff error as me 2015-05-05 16:14:52 here is the problem http://hastebin.com/tahabequma.sm - CAP_SYS_ADMIN is required 2015-05-05 16:15:07 for an ubuntu template 2015-05-05 16:15:29 im not running grsec 2015-05-05 16:18:47 BitL0G1c: http://sprunge.us/aERZ 2015-05-05 16:19:38 I was using the 1.1.1-r0 templates - just upgraded to 1.1.2-r2 on edge 2015-05-05 16:23:14 looks like a similar error to what i was getting http://hastebin.com/atotagunol.vbs 2015-05-05 16:24:29 going to try with vanilla 2015-05-05 16:24:49 ive commented the trap in ubuntu template 2015-05-05 16:26:52 I found a bug in the lxc-download template & will be sending that upstream too (am still working on unprivileged lxc) - shadow-utils no longer segfaults 2015-05-05 16:27:58 which bug? 2015-05-05 16:28:24 the unprivileged subuid/gid will need to be a member of readproc for unprivileged containers to work 2015-05-05 16:28:49 if you have a small /tmp or it is mount noexec lxc-download fails 2015-05-05 16:29:10 going to propose a patch for it 2015-05-05 16:29:27 I have a feeling debootstrap has wrong return code 2015-05-05 16:30:25 it looks likes its finished, but still someting returns an error wihtout saying a thing 2015-05-05 16:33:32 on vanilla i get the same bus error 2015-05-05 16:34:52 going to try with an unprivileged image 2015-05-05 16:45:10 weird 2015-05-05 16:49:03 BitL0G1c: do you know where debootstrap leaves its packages? 2015-05-05 16:49:21 or does it delete them right after installing? 2015-05-05 16:50:08 /var/cache/lxc ? 2015-05-05 16:50:30 thats rootfs i think 2015-05-05 16:50:59 /var/cache/lxc/name for privileged containers & /var/cache/download/dist/release for unprivileged images 2015-05-05 16:52:57 it looks like debootstrap cleans up on failure - my /var/cache/lxc/precise was empty 2015-05-05 16:56:15 s/have backed up lxc-download :-( before upgrading 2015-05-05 17:21:57 <_ikke_> Is it normal that the pager in alpine cannot show ansi colors? (ie, -r / -R options seem to be missing). Notice it with git output 2015-05-05 17:25:33 _ikke_: I don't know if it is normal but I have the same effect. it gets rather annoying at times 2015-05-05 17:34:09 <_ikke_> right 2015-05-05 18:24:24 are you using busybox less? 2015-05-05 18:24:56 <_ikke_> ylVa333UKfLO: yess 2015-05-05 19:16:20 clandmeter - I got an ubuntu unprivileged container image running in Alpine 2015-05-05 19:17:13 just 'lxc-create -n xxxx -t download' & choose the ubuntu image you want 2015-05-05 20:03:53 <_ikke_> The wiki says patches should be send to alpine-aports, but I also see a lot of patches on the alpine-devel mailing list..? 2015-05-05 20:06:38 _ikke_: before we had the alpine-aports ml, everything went to alpine-devel 2015-05-05 20:06:53 so that's historical stuff. use alpine-aports 2015-05-05 20:10:16 <_ikke_> ok 2015-05-05 20:10:34 <_ikke_> I already send them to alpine-aports 2015-05-05 20:10:42 <_ikke_> Was just curious 2015-05-05 21:05:29 someone able to verify a bug in update-ca-certificates for me? i think the patch is super simple, just want to make sure someone else can replicate it 2015-05-05 21:08:25 try adding a custom CA certificate to /usr/local/share/ca-certificates/, and run update-ca-certificates to see if it gets added 2015-05-05 21:11:05 you can test with http://www.startssl.com/certs/ca.pem if you need a CA (just rename to something like /usr/local/share/ca-certificates/startssl-ca.crt) 2015-05-05 21:44:07 BitL0G1c: interesting 2015-05-05 21:44:10 i didnt know that option 2015-05-05 21:51:20 andyshinn: i can try that after work 2015-05-05 21:51:30 andyshinn: can you wait that long? 2015-05-05 21:51:40 clandmeter - yes - also as it downloads an image first it is a lot less hassle to debug 2015-05-05 21:52:40 eventually it will be nice to run unprivileged images as a user (not root) 2015-05-05 22:00:12 andyshinn - running update-ca-certificates (after downloading & renaming the PEM) gives "WARNING: ca-certificates.crt does not contain exactly one certificate or CRL: skipping" 2015-05-05 22:01:37 yea, that always happens, can you verify it created the symlink though? 2015-05-05 22:01:49 there should be a symlink to your cert in /etc/ssl/certs 2015-05-05 22:07:41 hold on 2015-05-05 22:09:39 no symlink 2015-05-05 22:30:54 on line 50 of update-ca-certificates it should be: for f in pairs(certlist) do, can you try changing it to: for key,fn in paris(certlist) do 2015-05-05 22:31:05 on line 50 of update-ca-certificates it should be: for f in pairs(certlist) do, can you try changing it to: for key,fn in paris(certlist) do 2015-05-05 22:32:17 can you change line 50 to be `for key,fn in pairs(certlist) do` and remove line 51 2015-05-05 22:39:54 ok hold on 2015-05-05 22:46:56 andyshinn - yes that seems to work http://hastebin.com/tugijiferu.coffee 2015-05-05 23:40:11 "for key,fn in paris(..)" such a lovely function, paris() :3 2015-05-06 00:18:03 hi there, I'm trying to build xcompmgr, based on archlinux PKGBUILD, but the build phase end with those errors: http://sprunge.us/YWHH do you know how to set XCOMPMGR_CFLAGS? 2015-05-06 00:23:43 fixed 2015-05-06 03:54:26 hey guys 2015-05-06 03:55:26 question regarding alpine... ne one around 2015-05-06 06:21:37 bitshark ask the question and maybe someoone will replay. 2015-05-06 06:22:04 +1 2015-05-06 06:31:25 I have many error like `fatal error: ft2build.h: No such file or directory` editing the the file and replacing the include directive to: 2015-05-06 06:31:27 #include 2015-05-06 06:31:34 fix the problem 2015-05-06 06:31:50 is there a more common way to do that? 2015-05-06 06:32:26 all libxft-dev is affected 2015-05-06 06:32:59 ACTION can send a patch for libxft-dev and freetype-dev if necesary 2015-05-06 06:57:23 Mo0O: i think you need add pkg-config freetype2 --cflags 2015-05-06 06:58:40 Mo0O: which file did you have to edit? 2015-05-06 07:00:04 hum 2015-05-06 07:01:02 does it help to edit xft.pc? 2015-05-06 07:01:03 ncopa: /usr/include/X11/Xft/Xft.h /usr/include/freetype2/ft2build.h /usr/include/freetype2/config/ftconfig.h /usr/include/freetype2/fttypes.h /usr/include/freetype2/ftsystem.h /usr/include/freetype2/ftimage.h and /usr/include/freetype2/fterrors.h 2015-05-06 07:02:01 those are the files I have edited 2015-05-06 07:02:03 the thing is that when useing any of those, CPPFLAGS (or CFLAGS) should have an -I/usr/include/freetype2 2015-05-06 07:02:30 CFLAGS="$CFLAGS $(pkg-config --cflags freetype2)" 2015-05-06 07:02:33 should fix it 2015-05-06 07:03:00 ok, I will try, thanks a lot 2015-05-06 07:03:04 i wonder if it helps to change "Requires: xproto" to "Requires: xproto, freetype2" in xft.pc 2015-05-06 07:03:30 hmm, not sure about that 2015-05-06 07:03:38 whenever pkg-config --cflags xft is used, then should also pkg-config --cflags freetype2 also be used 2015-05-06 07:04:02 because /usr/include/X11/Xft/Xft.h has "#include " 2015-05-06 07:04:34 no 2015-05-06 07:04:38 libxft is correct 2015-05-06 07:04:39 yes 2015-05-06 07:04:50 $ pkg-config --cflags xft 2015-05-06 07:04:50 -I/usr/include/freetype2 -I/usr/include/libpng16 2015-05-06 07:05:23 what package did you try build when you got the error? 2015-05-06 07:05:46 a custom version of dwm with xft patch 2015-05-06 07:06:06 no autoconf i suppose 2015-05-06 07:06:10 http://dwm.suckless.org/patches/xft this one 2015-05-06 07:06:44 I could patch config.mk to handle this change 2015-05-06 07:06:48 :) 2015-05-06 07:07:25 ACTION try 2015-05-06 07:07:45 XFT_CFLAGS := $(shell pkg-config --cflags xft) 2015-05-06 07:08:05 XFT_LIBS := $(shell pkg-config --libs xft) 2015-05-06 07:08:28 ACTION guess ncopa use golang at the moment ;) 2015-05-06 07:08:46 LIBS = LIBS = -L${X11LIB} -lX11 ${XINERAMALIBS} $(XFT_LIBS) 2015-05-06 07:09:18 INCS = -I${X11INC} $(XFT_CFLAGS) 2015-05-06 07:09:24 or similar 2015-05-06 07:09:29 no, i dont really use golang 2015-05-06 07:09:48 ok, just the `:=` syntax 2015-05-06 07:10:12 that is so you dont fork shell and pkg-config every time the varialbe is expanded 2015-05-06 07:10:28 the := makes it expand first time only 2015-05-06 07:10:34 nice 2015-05-06 07:11:40 is google broken or is it just me? 2015-05-06 07:12:09 ncopa: works here 2015-05-06 07:12:12 http://stackoverflow.com/questions/448910/makefile-variable-assignment 2015-05-06 07:12:14 google 2015-05-06 07:12:43 https://www.google.no/#q=gnu+make+variable+assignment 2015-05-06 07:12:46 great, thanks for the link :) 2015-05-06 07:13:01 when i click on second link (GNU make) it just hangs 2015-05-06 07:13:21 maybe internet is broken 2015-05-06 07:13:49 the second link works here too 2015-05-06 07:14:06 my internet tubes are probably broken 2015-05-06 07:17:21 you rox ncopa :) 2015-05-06 07:17:44 I've clean my libxft-dev install, and use your instructions 2015-05-06 07:17:50 it works like a charm 2015-05-06 07:18:10 LIBS = -L/usr/lib -lc -L${X11LIB} -lX11 ${XINERAMALIBS} -lXft $(XFT_LIBS) 2015-05-06 07:18:39 not the `-lXft`, to avoid another error I don't remember 2015-05-06 07:18:48 in case you want to use it 2015-05-06 07:31:28 yes you should replace -lXft with $(XFT_LIBS) 2015-05-06 08:26:46 ncopa: to install that patched version of nl.alpinelinux.org/alpine/edge/main/x86_64/claws-mail-plugins-rssyl-3.11.1-r2.apk 2015-05-06 08:26:51 do i have to run a whole edge system? 2015-05-06 08:26:57 or just claws-mail@edge 2015-05-06 08:27:20 sillysausage: mix edge with stable releases might not always work 2015-05-06 08:27:20 my repositories file is like so: 2015-05-06 08:27:23 /media/usb/apks 2015-05-06 08:27:24 sometimes it does 2015-05-06 08:27:25 http://nl.alpinelinux.org/alpine/v3.1/main 2015-05-06 08:27:27 @edge http://nl.alpinelinux.org/alpine/edge/main 2015-05-06 08:27:28 sometimes it doesnt 2015-05-06 08:27:29 @testing http://nl.alpinelinux.org/alpine/edge/testing 2015-05-06 08:27:31 ah 2015-05-06 08:27:39 however 2015-05-06 08:27:42 this is a bug 2015-05-06 08:27:48 yeah and it was a very minor patch 2015-05-06 08:27:48 and we fix bugs in stable trees 2015-05-06 08:27:53 so i pushed it to v3.1 2015-05-06 08:27:59 true 2015-05-06 08:28:08 oh so you did. 2015-05-06 08:28:10 <_ikke_> ncopa: what about rebuilds with additional flags? 2015-05-06 08:28:23 _ikke_: which rebuilds? 2015-05-06 08:28:42 <_ikke_> ncopa: I sent in a patch yesterday for zabbix 2015-05-06 08:28:54 (5/9) Upgrading claws-mail-plugins-rssyl (3.11.0-r0 -> 3.11.1-r1) 2015-05-06 08:28:59 hmm so i assume that includes the patch 2015-05-06 08:29:09 it won't be r2 the pushed release would it 2015-05-06 08:29:17 becuase it was an upgrade from 3.11.0 to 3.11.1 2015-05-06 08:32:59 http://git.alpinelinux.org/cgit/aports/log/main/claws-mail?h=3.1-stable 2015-05-06 08:33:53 ah right 2015-05-06 08:34:41 right, looks like it is working kind of im guessing there's another date-style not catered for in that patch 2015-05-06 08:34:45 i might have to investigate 2015-05-06 08:37:52 seems that patch doesn't cater for dates like Wed, 06 May 2015 07:02:54 +0000 2015-05-06 08:46:57 sillysausage: it might be its a musl libc issue 2015-05-06 08:47:09 hmm actually that's a good point 2015-05-06 08:47:13 _ikke_: I'll have a look 2015-05-06 08:47:22 becuase i know with i3, the strftime package the %p and %P was different 2015-05-06 08:47:45 (the conf i use on arch didn't work on alpine for i3status) 2015-05-06 08:48:01 sillysausage: http://pubs.opengroup.org/onlinepubs/009695399/functions/strftime.html 2015-05-06 08:48:05 just looking through all my feeds and looking at the ones that didn't work 2015-05-06 08:48:14 to see if the options are different in the patch 2015-05-06 08:48:35 check strftime in claws source 2015-05-06 08:53:10 _ikke_: looks like your patches was applied 2015-05-06 08:53:14 but 2015-05-06 08:53:31 i wonder if we could use the configs from srouce package instead 2015-05-06 08:53:52 then we dont need update it in our aports tree every time its a new zabbix release 2015-05-06 09:01:23 . 2015-05-06 09:10:17 hmm 2015-05-06 09:10:23 well all the ones that are mucking up 2015-05-06 09:10:33 are either like: 2015-05-06 09:10:35 Mon, 26 Jan 2015 10:20:13 +0000 2015-05-06 09:10:37 or 2015-05-06 09:10:43 Wed, 06 May 2015 04:46:05 GMT 2015-05-06 09:10:45 which 2015-05-06 09:10:53 http://git.alpinelinux.org/cgit/aports/commit/main/claws-mail?h=3.1-stable&id=c662b8a7f2d165b10041dc58daf211acb5939f76 2015-05-06 09:10:56 looks to be this one: 2015-05-06 09:11:06 !strptime(date, "%a, %d %b %Y %H:%M:%S %Z", &t) && 2015-05-06 09:11:17 except %Z 2015-05-06 09:11:28 is%Z 2015-05-06 09:11:30 Replaced by the timezone name or abbreviation, or by no bytes if no timezone information exists. [ tm_isdst] 2015-05-06 09:14:08 hmm when i did man strftime 2015-05-06 09:14:10 there's another one 2015-05-06 09:14:13 %z The +hhmm or -hhmm numeric timezone (that is, the hour and minute offset from UTC). (SU) 2015-05-06 09:14:15 %Z The timezone name or abbreviation. 2015-05-06 09:18:00 okay i looked another one that works, and it has the date 2015-05-06 09:18:02 pubDate>Wed, 06 May 2015 05:38:26 GMT 2015-05-06 09:18:48 <_ikke_> ncopa: The only thing is the default locations should be adjusted 2015-05-06 09:19:10 <_ikke_> the default config places everything in /tmp/ 2015-05-06 09:19:16 ok 2015-05-06 09:19:22 how about we add a patch for it? 2015-05-06 09:19:33 <_ikke_> Sounds reasonably 2015-05-06 09:19:40 <_ikke_> reasonable 2015-05-06 09:19:52 then we also detect if they change it and patch no longer applies on upgrades 2015-05-06 09:20:52 <_ikke_> ncopa: did leonardo also apply the libxml2 fix, or only the config changes? 2015-05-06 09:21:34 http://git.alpinelinux.org/cgit/aports/log/main/zabbix 2015-05-06 09:22:25 <_ikke_> right, only the config updates 2015-05-06 09:30:05 anyone mounting iOS devices in Alpine Linux? 2015-05-06 09:35:35 Greetings Comrades! 2015-05-06 10:01:39 ncopa: https://dpaste.de/zGah/raw 2015-05-06 10:01:45 that's all you missed while you were gone 2015-05-06 10:01:50 isp issues 2015-05-06 10:02:03 http://pubs.opengroup.org/onlinepubs/009695399/functions/strftime.html 2015-05-06 10:02:08 is that the one alpinelinux uses? 2015-05-06 10:02:14 thanks 2015-05-06 10:02:39 basically any of the ones in here http://seclists.org/rss/ use the type: Fri, 01 May 2015 06:06:29 GMT 2015-05-06 10:03:10 https://www.archlinux.org/feeds/news/ uses: Mon, 26 Jan 2015 10:20:13 +0000 2015-05-06 10:03:45 g_warning("Invalid RFC822 date!\n"); 2015-05-06 10:03:46 neither of these seem to work, they show the Date: Thu, 1 Jan 1970 00:00:00 GMT 2015-05-06 10:03:50 do you get that warning? 2015-05-06 10:05:09 yes i do 2015-05-06 10:05:24 ** (claws-mail:2564): WARNING **: Invalid RFC822 date! 2015-05-06 10:05:38 i just refreshed archlinux and that's what i saw 2015-05-06 10:06:34 http://pubs.opengroup.org/onlinepubs/009695399/functions/strptime.html 2015-05-06 10:07:05 interestingly, there are no %z or %Z 2015-05-06 10:09:23 what's i nteresting is 2015-05-06 10:09:25 if i look at: 2015-05-06 10:09:36 http://feeds.theguardian.com/theguardian/world/rss 2015-05-06 10:09:38 it works 2015-05-06 10:09:40 i do see the same error 2015-05-06 10:09:47 *** (claws-mail:2564): WARNING **: Invalid RFC822 date! 2015-05-06 10:09:51 and the date is in the same format: 2015-05-06 10:10:09 Wed, 06 May 2015 10:02:56 GMT 2015-05-06 10:10:15 as the ones that were not working 2015-05-06 10:10:20 but it doesn't show up as 1970 2015-05-06 10:16:50 well when i do man strptime in alpinelinux 2015-05-06 10:16:52 i see 2015-05-06 10:16:54 %z An RFC-822/ISO 8601 standard timezone specification. 2015-05-06 10:16:56 %Z The timezone name. 2015-05-06 10:16:58 y 2015-05-06 10:17:18 that wouldn't be glibc's implementation would it 2015-05-06 10:18:06 http://git.alpinelinux.org/cgit/aports/commit/main/claws-mail?h=3.1-stable&id=c662b8a7f2d165b10041dc58daf211acb5939f76 2015-05-06 10:18:28 looking at that code i would say that it should only screw up when %z is used 2015-05-06 10:37:41 ncopa: the problem is that we don't have %z or %Z http://git.musl-libc.org/cgit/musl/tree/src/time/strptime.c 2015-05-06 10:43:18 i've raised it with the people in #claws 2015-05-06 10:43:40 basically that z and Z is used in glibc, llvm, newer versions of hp ux but not musl, and posix doesn't demand it 2015-05-06 10:44:09 seems to vary with hp-ux :) 2015-05-06 10:45:50 hopefully we can get another patch that works with musl :D 2015-05-06 10:57:37 sillysausage: do you think you could create an issue on bugs.alpinelinux.org? 2015-05-06 10:57:39 with all the details 2015-05-06 10:57:54 i can point musl devs to it 2015-05-06 10:58:00 im talking to them now 2015-05-06 10:58:07 basically they want a spec that is not inconsistent 2015-05-06 10:58:17 looking at how llvm vs glibc did it 2015-05-06 11:56:39 anyway ncopa i don't have the resources to fix this atm, (were you working on it?) 2015-05-06 11:57:12 if you're not i guess I'll have to put it on my to-do list when i get my thing fixed 2015-05-06 11:57:38 they said: get_rfc822_date() is just for generating string for current time ("now") 2015-05-06 11:57:43 so i'm not sure that is all that useful 2015-05-06 14:10:22 hello everyone 2015-05-06 14:10:41 hi 2015-05-06 14:11:31 may I ask for questions about Alpine? 2015-05-06 14:12:07 sure 2015-05-06 14:12:27 aslong its not about the email client, it should be fine. 2015-05-06 14:14:26 thanks, I couldn't figure out how to change keymap in syslinux menu? installed kbd-misc and added KDBMAP de.ktl to extlinux.conf, but didn't work! 2015-05-06 14:17:01 Guest96010: i didnt even know that you could change keymap in syslinux 2015-05-06 14:17:20 i suppose you need copy the de.ktl to some place syslinux finds it 2015-05-06 14:17:25 maybe in /boot/syslinux? 2015-05-06 14:18:10 there is no /boot/syslinux dir, and extlinux.conf is in /boot 2015-05-06 14:19:02 am guessing there is a way but its Alpine-specific.. 2015-05-06 14:32:04 Guest14930: kbdmap.c32 2015-05-06 14:32:17 you need kbdmap.c32 too 2015-05-06 14:32:22 some place it can find it 2015-05-06 14:32:49 http://www.syslinux.org/wiki/index.php/SYSLINUX#KBDMAP_keymap 2015-05-06 14:33:13 seems like its special keymap format 2015-05-06 14:36:13 thanks ncopa 2015-05-06 14:36:51 I will check it. 2015-05-06 14:39:08 alpine 3.2.o_rc1 is out 2015-05-06 14:39:11 please test it! 2015-05-06 14:42:04 am testing/tweaking 3.1.3 for production, sure why not :) 2015-05-06 14:59:31 what's the best way to create an APKBUILD that ships a custom user/group? 2015-05-06 15:05:56 have a look at udevil - it sets a plugdev group in the APKBUILD & creates the group in $pkg.pre-install 2015-05-06 15:06:34 pkggroups="plugdev" (in APKBUILD) 2015-05-06 15:07:10 #!/bin/sh 2015-05-06 15:07:11 addgroup -S plugdev 2>/dev/null 2015-05-06 15:07:11 exit 0 2015-05-06 15:08:40 The part I don't get is how do I chown things to that newly created user 2015-05-06 15:09:01 doing it in package() would add the UID to the tarball, so it might mismatch on the target system, right? 2015-05-06 15:09:18 you can set the chown in package() 2015-05-06 15:10:10 the PKGBUILD I'm deriving from uses pre-allocated UIDs (ugh...) 2015-05-06 15:11:00 also, how do I do arch-dependent entries in $source? 2015-05-06 15:11:20 makepkg has source_ for it 2015-05-06 15:11:43 see http://pastebin.com/bx61yXq4 2015-05-06 15:12:42 I think you can use $ARCH ? 2015-05-06 15:13:05 and then if $ARCH? won't that screw up checksumming? 2015-05-06 15:13:18 since it'll shadow out the arch that's not in use? 2015-05-06 15:14:46 hmm does tar actually store the ownership by name? I thought it just takes the UID/GID... 2015-05-06 15:15:44 if you set pkggroups="xxx" & chown in package() it will set the custom owners 2015-05-06 15:15:59 "When writing an archive, tar writes the user ID and user name separately. If it can't find a user name (because the user ID is not in `/etc/passwd'), then it does not write one. When restoring, it tries to look the name (if one was written) up in `/etc/passwd'. If it fails, then it uses the user ID stored in the archive instead." -- guess I was wrong :) 2015-05-06 15:16:43 I checked the udevil apk in xarchiver - it had the custom permissions set 2015-05-06 15:17:03 the arch is not stored in $ARCH 2015-05-06 15:18:51 does adduser create a corresponding group? 2015-05-06 15:19:01 BitL0G1c: what happend to those udevil patches? 2015-05-06 15:19:13 i cannot apply your update 2015-05-06 15:19:48 oh i think fcolista amended it 2015-05-06 15:20:22 i was missing the pkggroups=" & francesco fixed it for me before he applied it 2015-05-06 15:20:24 BitL0G1c: can you rebase your changes? 2015-05-06 15:20:56 the changes to the post-install files ? 2015-05-06 15:21:15 udevil is in aports now already 2015-05-06 15:21:33 yes that change 2015-05-06 15:21:53 networkmanager also has issues 2015-05-06 15:22:13 >>> ERROR: networkmanager: networkmanager.apk-post-msg is missing in checksums 2015-05-06 15:22:40 i can fix nm local 2015-05-06 15:22:49 ok - let me try again - I rebuilt all 3 packages before sending 2015-05-06 15:28:00 BitL0G1c: do you have a recent version of abuild? 2015-05-06 15:30:24 the latest i think 2015-05-06 15:30:49 i just git stashed & rebased ~3 2015-05-06 15:31:09 ok, cause pkgs from testing didnt have an updated init.d (openrc change) 2015-05-06 15:32:04 apk-tools 2.6.0_rc4, compiled for x86_64. 2015-05-06 15:32:56 I changed some packages to openrc-run - but this was causing issues for people running stable 2015-05-06 15:33:55 will change networkmanager 2015-05-06 15:57:36 ncopa I tried adding the kbdmap.c32 where applicable but not working yet.. I think the problem is in "keytab-lilo.pl" but my perl skills are mundane at best. thanks again. 2015-05-06 16:01:42 keytab-lilo is supposed to sort of conv de.map to de.ktl if you feed it us.map and de.map (found at /lib/kbd/keymaps/...)... but it returns keymap is empty (happend in other distros with old version of script) 2015-05-06 16:02:36 clandmeter - udevil post-install patch resent + nm initd fix 2015-05-06 16:03:30 udevil? what a cool name is that 2015-05-06 16:10:09 does this look good for pre-install? 2015-05-06 16:10:15 addgroup -S plex 2>/dev/null 2015-05-06 16:10:15 adduser -S -G plex -h /var/lib/plex -H -s /bin/false -D plex 2>/dev/null 2015-05-06 16:10:15 exit 0 2015-05-06 16:12:09 Also, anyone figured out how to use different sources for different archs? 2015-05-06 16:19:19 have a look at dnscrypt-proxy pre-install or bitcoin or namecoin - they all setup a system user 2015-05-06 16:20:03 it seems that abuild puts the arch in $CARCH 2015-05-06 16:21:46 you should prob include both arch downloads in $sources (as they will both be needed anyway) 2015-05-06 16:24:05 actually $CARCH would work 2015-05-06 16:34:27 This seems to work :D 2015-05-06 16:36:14 I need to put my apkbuilds up somewhere... 2015-05-06 16:56:56 why does http://wiki.alpinelinux.org/wiki/APKBUILD_Reference#install say that $install should be in $source and then abuild yells at me for doing so? 2015-05-06 17:03:07 heey yo 2015-05-06 17:03:12 anyone around 2015-05-06 17:04:07 hello 2015-05-06 17:05:20 what's up.. actually got some questions on apkbuilds 2015-05-06 17:05:45 first is how does apk detect what a package installs ... from what i'm looking at on some forms 2015-05-06 17:05:57 it's basically doing download / ./configure / make / make install 2015-05-06 17:06:22 on a tgz or git file ... so how is alpine tracking what make install is doing, where it is putting files 2015-05-06 17:07:36 or maybe its leaving the make install part out and putting in a prefix? 2015-05-06 17:07:45 to learn what files are produced? 2015-05-06 17:08:03 <_ikke_> bitshark: Yeah, everything is installed in the src dir in the package directory 2015-05-06 17:08:21 <_ikke_> I mean pkg dir 2015-05-06 17:09:08 okay so then what happens 2015-05-06 17:09:22 its gotta index that prior to unpacking it into the root fs 2015-05-06 17:09:34 even if its in an apk package... there's gotta be like a file list inthe package 2015-05-06 17:10:13 <_ikke_> you can verify that by unpacking the package 2015-05-06 17:10:21 and second, when i install a binary apk, it's gotta be tracking the files that get installed, either thru like a unionmount, or even just trusting the apk manifest is correct 2015-05-06 17:10:46 like so it knows to delete /var/lib/some/crazy/subdir 2015-05-06 17:10:54 if i uninstall it 2015-05-06 17:14:31 i suppose that's from the manifest oin the apk 2015-05-06 17:14:55 <_ikke_> There is no manifest 2015-05-06 17:15:05 <_ikke_> there is a .APKINFO but it just contains basic info 2015-05-06 17:15:12 <_ikke_> .PKGINFO* 2015-05-06 17:16:49 okay thank you ikke 2015-05-06 17:17:12 you know, this morning i was reviewing freebsd's and netbsd's package systems to get them running on linux 2015-05-06 17:17:15 <_ikke_> bitshark: There are some headers though that tar doesn't understand, but afaics, nothing regarding to the files 2015-05-06 17:18:12 and i realized two things: (1) netbsd's package mgmt sucks , and (2) freebsd pkg mgmt is pretty neat, and (3) alpine takes good ideas from freebsd and apt-get 2015-05-06 17:18:17 three things* 2015-05-06 17:19:44 anyway the problem that inspired me was this: If I want to install software that has no package in alpine or lfs or other microdistro (ie. a tar.gz file) ... if i do ./configure && make && make install ... how do I ensure that I know what gets installed and that I can remove it? 2015-05-06 17:19:49 <_ikke_> alpine is inspired by gentoo ports and archlinux pacman 2015-05-06 17:21:50 i came up with a few options... (1) build a package out of it... (2) run two diffs of the filesystem , (3) Remount / as RO, then run a union mount on top of it. Collect the files from the top layer to get a definiton of what's changed, then merge the union back down to / , or (4) use a tool like bup which is git based filesystem verisoning. you could use the deltas to define packages (and double as incremental backup) 2015-05-06 17:22:20 yeah, it's great. i think alpine will grow. 2015-05-06 17:23:09 i'm just wondering about a seperate problem... its along the lines of ... well if i install lxqt from source on alpine, how do i know what gets installed (so i can reverse the make install) WITHOUT having to build an apk package 2015-05-06 17:23:37 those four ways were the methods... i thinkk #3 and #4 are best from architecture standpoint 2015-05-06 17:23:57 bup is nice because you get snapshots independent of underlying filesystem 2015-05-06 17:24:31 OH that was the fifth way... take a snapshot at the FS layer if you are running XFS, ZFS, or other filesystem with cheap incremental snapshots 2015-05-06 17:25:00 so does everyone here run alpine on the desktop? 2015-05-06 17:25:44 <_ikke_> nope 2015-05-06 17:26:22 interesting.. me neither, not yet 2015-05-06 17:26:35 just picked up a new machine and i'm trying to pick my setup 2015-05-06 17:26:48 i'm definitely going to run f2fs on the ssd 2015-05-06 17:27:20 but i havent decided whether to use my own microdistro (lfs based or static ram based) .. or whether to use alpine 2015-05-06 17:27:52 my only hesitation with alpine is that i would prefer to use a completely staticly compiled linux, at least as a test 2015-05-06 17:28:20 static compiled distro would make dependencies go away haha 2015-05-06 17:28:52 so you'd just need a way of snapshoting and you're set 2015-05-06 17:29:07 what package contains rpm2cpio? 2015-05-06 17:29:48 i dunno i just use the cpio tool to make ramdisks 2015-05-06 17:30:25 not ramdisks, RHEL Package Manager format 2015-05-06 17:30:30 i think its just a gzipped archive that's like newc format 2015-05-06 17:30:45 it's some crazy magic as far as I'm concerned 2015-05-06 17:30:49 no that's what i mean.. like cant u just download cpio 2015-05-06 17:31:01 rpm is just a compressed package that's renamed 2015-05-06 17:31:16 I'm not so sure about that 2015-05-06 17:31:17 and then cpio is the same thing... its litereally just a different kind of tar.gz 2015-05-06 17:31:32 well it was when i ran RHEL5 10 years ago 2015-05-06 17:31:35 lol 2015-05-06 17:31:41 I know that DEB is an AR archive with a data.tar.xx inside, but no clue about RPM >.< 2015-05-06 17:31:58 oh you're right 2015-05-06 17:32:03 rpm is actually cpio 2015-05-06 17:32:13 same format as ramdisks are compressed with 2015-05-06 17:32:25 http://www.rpm.org/max-rpm/s1-rpm-file-format-rpm-file-format.html it seems to have a binary header >.< 2015-05-06 17:33:01 I'll just try to use the ubuntu package and go from there, DEBs are less painful to unpack 2015-05-06 17:37:14 okay so i think rpm2cpio is a tool 2015-05-06 17:37:21 provided by redhat 2015-05-06 17:37:27 yeah, but I didn't find it in aports 2015-05-06 17:37:56 its got alot of dependencies 2015-05-06 17:38:05 well if u want it i'll give u the src 2015-05-06 17:38:09 or a staticlinked version 2015-05-06 17:38:19 won't help 2015-05-06 17:38:19 if u are on x86 64 after kernel 3.16 2015-05-06 17:38:48 I'm writing a APKBUILD, and I'm not gonna ship stuff with it just because rpm 2015-05-06 17:41:06 heh 2015-05-06 17:41:15 i think u are reding too much into what im saying 2015-05-06 17:41:26 i have no idea what you're doing... i just know u wanted a copy of rpm2cpio 2015-05-06 17:41:40 i looked it up and found where the tool comes from ... that's all 2015-05-06 17:41:46 like a favor 2015-05-06 17:42:09 well, thanks, but it won't help much 2015-05-06 17:42:10 im not saying like distribute a random binary from redhat. i dont trust RH either 2015-05-06 17:42:42 That's not the problem, I need to work with what I have, not add a random binary cause I feel like it.. 2015-05-06 17:44:06 I wonder if this will run at all.. 2015-05-06 17:45:29 oh word! 2015-05-06 17:45:35 i think rpm2cpio is a shell script 2015-05-06 17:45:44 yup 2015-05-06 17:45:45 u want it? 2015-05-06 17:45:48 or do u have it 2015-05-06 17:46:05 its like 20 lines haha 2015-05-06 17:46:22 a bunch of sed and awk 2015-05-06 17:46:58 I'm questioning the viability of my project overall right now >.< 2015-05-06 17:49:17 yeah it totally workz 2015-05-06 17:50:43 I might have to include a copy of glibc >.< stupid binary distributions 2015-05-06 17:51:06 root@computer:/tmp/rpm-4.11.2/scripts# ./rpm2cpio.sh bash-4.3.33-3.fc23.x86_64.rpm > bash.cpio 2015-05-06 17:51:06 root@computer:/tmp/rpm-4.11.2/scripts# cpio -i -t < bash.cpio | head 2015-05-06 17:51:06 ./etc/skel/.bash_logout 2015-05-06 17:51:06 ./etc/skel/.bash_profile 2015-05-06 17:51:06 ./etc/skel/.bashrc 2015-05-06 17:51:07 ./usr/bin/alias 2015-05-06 17:51:11 ./usr/bin/bash 2015-05-06 17:51:29 nice^^ 2015-05-06 17:52:12 here i linked it for you 2015-05-06 17:52:13 http://pastebin.com/ykf8kcvW 2015-05-06 17:52:24 i pulled it out of the deb package 2015-05-06 17:52:40 just tested it... it only seems to like x86_64 rpms on my system tho 2015-05-06 17:52:44 but it workz 2015-05-06 17:53:24 the cpio format is literally just like an alternative to tar/gz , it's simple 2015-05-06 17:53:57 ik 2015-05-06 17:55:57 kk .. i just starting messing with cpio a few weekz ago when i was building an initramfs 2015-05-06 18:02:52 Hello! I am just checking out Alpine and was curious if apk has taken into account the TUF threat model https://github.com/theupdateframework/tuf/blob/develop/SECURITY.md? 2015-05-06 18:03:27 I didn't see anything in terms of security on the wiki page, just features/howto. 2015-05-06 18:04:05 Let me know if this would be better for #alpine-devel 2015-05-06 18:05:09 hey 2015-05-06 18:05:44 i think that's a great point about updates as an attack vector 2015-05-06 18:05:56 but all modern OS are vulnerable to that to varying degrees 2015-05-06 18:06:05 i think for a linux distro 2015-05-06 18:06:23 <_ikke_> What are you supposed to do against "Indefinite freeze attacks" 2015-05-06 18:06:53 the countermeasures would most likely involve authenticated encryption, using ssl with PFS, and most importantly.. perhaps actually using hte kernel functions and bios TCM 2015-05-06 18:07:22 <_ikke_> Well, all packages are signed, and they use hashes for file integrity 2015-05-06 18:07:45 yeah 2015-05-06 18:07:56 but that's got major flaws if not implemented correctly 2015-05-06 18:08:04 and 99% of the time its not 2015-05-06 18:08:16 ie. if i can mitm your server traffic to substitute my own pckages 2015-05-06 18:08:31 i can also mitm what is on your web page as 'authentic' sha sigs 2015-05-06 18:08:42 <_ikke_> they use rsa keys to sign packages, that's not trivial to mitm 2015-05-06 18:08:50 u have to essentially distribute the signing key to the clients 2015-05-06 18:09:00 well that's the question 2015-05-06 18:09:30 <_ikke_> But you have to use ssl indeed for the initial verification 2015-05-06 18:09:42 say i'm bob, non mallicious actor.. i have received alpine signing keyahead of time like when i downloaded the distro 2015-05-06 18:09:45 <_ikke_> alpinelinux does support https 2015-05-06 18:10:01 the alpine signing key is stored in the TCM in my BIOS via linux kernel 2015-05-06 18:10:11 so it cant be tampered with i get rooted 2015-05-06 18:10:50 now , IF i have that alpine public key prior to mallory initiating a Mitm attack on alpine's packages, i COULD validate the packages and see they were trojan 2015-05-06 18:10:57 but that's not from sha1 sums 2015-05-06 18:11:00 that's from digital sig 2015-05-06 18:11:04 _ikke_: For indefinite freeze attacks you should connect via TLS to multiple servers to collect package metadata. Return an error if they don't match. That way a single evil mirror couldn't execute this attack. 2015-05-06 18:11:05 <_ikke_> yes 2015-05-06 18:11:10 and its gotta be 'baked in' ya know? 2015-05-06 18:11:24 but if it's baked in is fine 2015-05-06 18:11:35 <_ikke_> bitshark: It is provided with the installation media 2015-05-06 18:11:36 and by 'baked in' i mean apk ALWAYS checks the sigs for you 2015-05-06 18:11:48 <_ikke_> if you can verify the integrity of the installation media, you are safe 2015-05-06 18:11:53 <_ikke_> bitshark: it does 2015-05-06 18:11:58 against a public signing key 2015-05-06 18:12:03 that's stored locally 2015-05-06 18:12:13 <_ikke_> rsa keys 2015-05-06 18:12:19 if that's the case you've prevented 90% + of common attacks 2015-05-06 18:12:31 yeah so that's a correct implementation , in the broad sense 2015-05-06 18:12:33 <_ikke_> well public keys, right 2015-05-06 18:12:43 what is an INCORRECT and common practice 2015-05-06 18:12:45 <_ikke_> openssl generated public / private keys 2015-05-06 18:12:53 is to put an SHA1 sig file next to a tgz 2015-05-06 18:12:59 that's completely pointless 2015-05-06 18:13:12 <_ikke_> right, everybody can provide the SHA1 sig 2015-05-06 18:13:22 its security theatre 2015-05-06 18:13:29 <_ikke_> you are dendend on ssl in that case 2015-05-06 18:13:33 <_ikke_> dependent* 2015-05-06 18:13:36 well like an attacker if he can fake that tgz he can change the sig 2015-05-06 18:14:29 <_ikke_> If you can verify the ssl certificate, you know the signature is not tampered with 2015-05-06 18:14:31 "Illusory security is terrible and is worse than no security" 2015-05-06 18:14:33 https://utcc.utoronto.ca/~cks/space/blog/tech/IllusorySecurityTerrible 2015-05-06 18:14:35 ---^ 2015-05-06 18:14:39 <_ikke_> yeah, I agree 2015-05-06 18:14:44 <_ikke_> false sense of security 2015-05-06 18:14:46 talks about how joyent did exactly what we are talking about 2015-05-06 18:14:50 on their home page 2015-05-06 18:14:53 hillarious 2015-05-06 18:15:01 Sorry to be a little ignorant on this: do you start a package upgrade by downloading a signed repo metadata file? 2015-05-06 18:15:07 they are like the 'docker'company 2015-05-06 18:15:09 dumbasses haha 2015-05-06 18:15:27 <_ikke_> fowlslegs: yes 2015-05-06 18:15:31 i dunno fowlslegsfrom what ikke says all packages are signed 2015-05-06 18:15:34 i think he's right 2015-05-06 18:15:50 the signing key is apparently part of the distro 2015-05-06 18:15:55 <_ikke_> if you don't have the correct public key, apk update (which downloads the meta data) will fail 2015-05-06 18:16:14 <_ikke_> try to download from http://alpine.ikke.info 2015-05-06 18:16:25 so that stops like all the common attacks 2015-05-06 18:16:33 it wont stop like a padding oracle attack 2015-05-06 18:16:40 <_ikke_> apk will refuse 2015-05-06 18:16:48 but it will stop 99% of ppl that would attack the package distribution 2015-05-06 18:17:10 its way more secure than like ubuntu which is like swiss cheeze imo 2015-05-06 18:17:18 padding oracle attack? 2015-05-06 18:17:24 yeah 2015-05-06 18:17:28 obscure math attack 2015-05-06 18:17:31 <_ikke_> bitshark: But don't you require for padding oracle to be able to alter the thing that gets downloaded? 2015-05-06 18:17:37 done by geniuses and the nsa 2015-05-06 18:17:41 no one else does those 2015-05-06 18:17:42 haha 2015-05-06 18:17:58 yeah well there's padding oracles that attack metadata 2015-05-06 18:18:11 <_ikke_> basically leaking data by altering the encrypted data in small ways 2015-05-06 18:18:29 but u need with a padding oracle.. . u need a screwed up server 2015-05-06 18:18:31 <_ikke_> the to be encrypted data 2015-05-06 18:18:33 in cbc mode usually 2015-05-06 18:18:46 that will tell u whether your guess of the message padding length is correct or not 2015-05-06 18:18:51 thru leaking bits over time 2015-05-06 18:19:04 eventually with enuf data (usually gb of data) u can recover the key 2015-05-06 18:19:29 <_ikke_> But that kind of attack is not really applicable in this case iirc 2015-05-06 18:19:42 but it's typically not seen in the wild outside of the blackhat genius crowd and the nation-states / defense complex 2015-05-06 18:19:48 <_ikke_> because an attacker cannot interact with the signing process 2015-05-06 18:19:49 u wont see some sccript kiddie doingit 2015-05-06 18:20:07 <_ikke_> bitshark: usually it will be provided through some prebuilt package 2015-05-06 18:20:17 ikke: i dont think its applicable here outside of the TLS implementation, no 2015-05-06 18:20:28 but TLS has already had like 3 or 4 padding oracle failz 2015-05-06 18:20:31 in openssl mainly 2015-05-06 18:20:43 most were ciphers inn cbc mode i think 2015-05-06 18:20:47 <_ikke_> right 2015-05-06 18:20:51 <_ikke_> poodle being the latest 2015-05-06 18:20:53 not like AES GCM 2015-05-06 18:20:58 or Salsa or Chacha 2015-05-06 18:21:05 those are immunze 2015-05-06 18:21:21 but like AES-CBC used to be vulneraable i think 2015-05-06 18:21:32 a couple of yearz ago 2015-05-06 18:21:53 its funny how all those bugs are in openssl 2015-05-06 18:21:55 Okay, but repodata is signed using DSA (I assume), so it's not vulnerable to this. 2015-05-06 18:22:00 Openssl isthe glibc of crypto haha 2015-05-06 18:22:15 uhm... so that's an excellent question 2015-05-06 18:22:23 that's actually not an easy answer 2015-05-06 18:22:33 it has to do with order of operations if i recall 2015-05-06 18:22:43 authenticate then encrypt or encrypt then authenticate 2015-05-06 18:22:47 one is vuln, the other is not 2015-05-06 18:23:07 BUT i dont think that applies here cuz we have a predistributed known signing key 2015-05-06 18:23:12 bitshark: Sorry, can you clarify who your last few posts were to? 2015-05-06 18:23:26 so even if TLS is cracked the endpoint can reject the downlod as malware 2015-05-06 18:23:44 uhm just rambling haha 2015-05-06 18:23:52 im talking about alpine abstractly 2015-05-06 18:23:57 why are packages not allowed to put things under /opt? D: 2015-05-06 18:24:01 cuz i dontknow the detailz 2015-05-06 18:24:17 ikkei think knows more on the specifics than i do for alpine linux 2015-05-06 18:24:29 im just speakn in broad terms about failures in crypto 2015-05-06 18:25:12 mostly historically they've been in TLS implementation 2015-05-06 18:26:11 bitshark: The problem with broken TLS though is that rollback/freeze/endless data/ slow retrieval/ malicious mirrors preventing upgrades all become possible w/ MitM. 2015-05-06 18:26:16 but in my opinion i think alpine is one of the most secure distros out there 2015-05-06 18:26:18 for linux anyway 2015-05-06 18:26:56 fowlslegs : sure of course that's true in theory , but the question is how practical is it 2015-05-06 18:27:15 how long can they mitm ... if they can do it for weeks undetected, yeah, we're screwed 2015-05-06 18:27:48 but most attackers can't sustain that ... it takes lots of resources, money/manpower 2015-05-06 18:28:23 exploits fetch like $80,000+ on the darknet so it's not a teenagers game anymore 2015-05-06 18:29:10 bitshark: All they need to do is wait until one huge remote attack vector comes, the MitM, then exploit that huge remote attack vector. 2015-05-06 18:29:20 but sure a nation state that can do knownplaintext/chosenplaintext attacks, indeterminate length mitm, .. yeah they will eventually own all the infrastructure if theyve got ppl and funding 2015-05-06 18:29:28 <_ikke_> bitshark: first authenticate, then decrypt 2015-05-06 18:29:37 ah thats right 2015-05-06 18:29:44 thanks ikke. 2015-05-06 18:29:47 <_ikke_> bitshark: so, encrypt, then calculate the MAC 2015-05-06 18:30:17 bitshark: They're generally not getting through with mathematical attacks. 2015-05-06 18:30:31 <_ikke_> bitshark: I don't know much more about alpine, I just know some basic things 2015-05-06 18:30:32 but anyway, if anything is backdoored right now... its the big binary distros 2015-05-06 18:30:36 ms win 2015-05-06 18:30:38 They being advanced attackers. Usually malware. 2015-05-06 18:30:46 cuz its so easy to hide malware in binariez 2015-05-06 18:31:05 yeah 2015-05-06 18:31:08 exactly 2015-05-06 18:31:12 On the topic, has Alpine considered working out a reproducible build spec? 2015-05-06 18:31:27 malware is the #1 vector... why break downthe vault door when they order a pizza 2015-05-06 18:31:31 aka malwarez! 2015-05-06 18:31:55 <_ikke_> fowlslegs: Aren't they dependendent on the actual build tools the projects use? 2015-05-06 18:32:01 And, also, is there a linux-libre kernel for Alpine? 2015-05-06 18:32:31 fowlslegs: the guy here ncopa is like one of the main dudez 2015-05-06 18:32:39 he's in europe but he could answer those questions 2015-05-06 18:32:47 i'm just a tangential party 2015-05-06 18:33:02 but i know that he has a build toolchain 2015-05-06 18:33:06 that was derived from gentoo 2015-05-06 18:33:23 bitshark: Thanks for the tip. 2015-05-06 18:33:33 <_ikke_> fowlslegs: I can't find any reference 2015-05-06 18:33:39 i highly doubt the alpine toolchain got malwared it was built from source ala linux from scratch or gentoo 2015-05-06 18:33:42 <_ikke_> fowlslegs: linux-grsec and linux-vanilla 2015-05-06 18:33:59 <_ikke_> bitshark: Unless the source is backdoored :P 2015-05-06 18:34:02 yeah 2015-05-06 18:34:06 that's always a risk 2015-05-06 18:34:27 <_ikke_> You usually call that a backdoor though 2015-05-06 18:34:30 Ah, systmkor, you're here. 2015-05-06 18:34:31 i just hope theres an autistic kid that reviews the main stuff 24/7 2015-05-06 18:34:34 haha 2015-05-06 18:34:49 <_ikke_> Yup, but some backdoors can be hidden in plain sight 2015-05-06 18:35:17 geez what was that one...there was a microsoft one that came out 2015-05-06 18:35:25 they had backdoored win7 or someting 2015-05-06 18:35:37 systmkor: I saw your post on the Qubes Google group the other week and made a note to contact you. 2015-05-06 18:35:39 but claimed it was just a mistake 2015-05-06 18:35:57 systmkor: Anyway, you're probably AFK. 2015-05-06 18:36:05 AH I REMEMBER 2015-05-06 18:36:17 okay so in windows theyd left a dll unstripped 2015-05-06 18:36:29 and it had a string called _NSAKEY 2015-05-06 18:36:31 hahahaha 2015-05-06 18:36:32 idiots 2015-05-06 18:36:50 http://cryptome.org/nsakey-ms-dc.htm 2015-05-06 18:37:45 _ikke_: What I was referring to is something like Debian's .buildinfo https://wiki.debian.org/ReproducibleBuilds 2015-05-06 18:38:10 <_ikke_> right 2015-05-06 18:38:37 _ikke_: Reproducible builds defend against (1) packagers whose keys are stolen and (2) packagers who are evil. 2015-05-06 18:39:07 <_ikke_> right 2015-05-06 18:39:09 do what do we derive a key from the toolchain ? 2015-05-06 18:39:11 <_ikke_> Makes sense 2015-05-06 18:39:18 <_ikke_> bitshark: what? 2015-05-06 18:39:30 im asking how reproducible builds work 2015-05-06 18:39:33 im guessing 2015-05-06 18:39:36 i have no idea how they work 2015-05-06 18:39:50 <_ikke_> preventing of variable data like dates 2015-05-06 18:39:53 i suspect maybe by including info on the parent cross compiler and the inputs of he source files 2015-05-06 18:39:58 but i dont know 2015-05-06 18:40:10 <_ikke_> bitshark: They do limit it to the same environment 2015-05-06 18:40:14 <_ikke_> (debian) 2015-05-06 18:40:22 ah k 2015-05-06 18:41:19 I personally can't give you a comprehensive answer and would say the Debian page I linked would. 2015-05-06 18:42:37 <_ikke_> alpine uses RSA keys 2015-05-06 18:42:37 Currently there are no reproducible distros, but Debian has 80% of their testing and unstable branch packages building reproducibly. 2015-05-06 18:42:41 <_ikke_> not DSA 2015-05-06 18:43:00 You come across some problems with the fact that certain packages use randomness in the compiling process. 2015-05-06 18:43:05 <_ikke_> fowlslegs: Do they also say how to setup the build environment? 2015-05-06 18:43:17 oh fowls 2015-05-06 18:43:18 <_ikke_> what version, etc? 2015-05-06 18:43:24 i looked into this via nix 2015-05-06 18:43:26 nix package manager 2015-05-06 18:43:32 its a neat idea but totally flawed 2015-05-06 18:43:50 i investigated it for alpine but we discarded it as flawed to the core 2015-05-06 18:44:07 but the idea has purely functional dependencies... so all builds are reproductible 100% 2015-05-06 18:44:11 in theory that is 2015-05-06 18:44:12 haha 2015-05-06 18:44:15 nixos.og 2015-05-06 18:44:21 channel is #nixoss 2015-05-06 18:44:22 _ikke_: Here's the page on their toolchain https://wiki.debian.org/ReproducibleBuilds/ExperimentalToolchain 2015-05-06 18:44:27 #nixos rather 2015-05-06 18:45:00 <_ikke_> fowlslegs: ok 2015-05-06 18:45:07 but all their toolchain has signatures for everything. . . and the sigs get passed forward like two or three compile generations 2015-05-06 18:45:21 its intersting idea but poorly managed project 2015-05-06 18:45:23 _ikke_: And here's how they (more or less) keep the build environment stable https://wiki.debian.org/ReproducibleBuilds/BuildinfoSpecification. 2015-05-06 18:45:28 got tons of scope creep 2015-05-06 18:45:43 thanks 2015-05-06 18:45:47 <_ikke_> fowlslegs: Not sure if it's the same thing, but apk has a .PKGINFO file inside each package 2015-05-06 18:46:15 its same idea... but image your pkginfo had sha1 sums of EVERYTHING that went into the build 2015-05-06 18:46:17 <_ikke_> fowlslegs: ok, it doesn't contain the version of each of the tools 2015-05-06 18:46:27 including your crosscompiler binary's sha1sum 2015-05-06 18:47:05 that's sort of the idea of 'reproducible builds' everything gets sha sumed so given inputs always generate the same exact binary output 2015-05-06 18:47:25 the problem is u cant represent it in strings or arrays 2015-05-06 18:47:52 u have to walk the AST of the binaries,, and the only language that's suitable for binary-binary translation is LISP 2015-05-06 18:48:06 so it because difficult to get perfect in practice 2015-05-06 18:48:07 https://wiki.debian.org/ReproducibleBuilds/BuildinfoSpecification 2015-05-06 18:48:09 but debian did a great job intheory 2015-05-06 18:48:12 ad practice 2015-05-06 18:48:20 even tho its not perfect it's still awesome 2015-05-06 18:48:41 Sorry, double-linked. 2015-05-06 18:48:53 i love debian ... i feel like debian is an ex girlfriend 2015-05-06 18:48:55 haha 2015-05-06 18:49:07 i dumped her because she got fat (aka systemd) 2015-05-06 18:49:23 bitshark: Yeah, especially would be nice if you could know when you were installing something that is not built reproducibly. 2015-05-06 18:49:28 heh systemd 2015-05-06 18:49:33 <_ikke_> I use arch, which has systemd for a while now 2015-05-06 18:49:49 I'm also on Arch^^ 2015-05-06 18:49:51 yeah i used ita bit too 2015-05-06 18:49:59 but i primarily hate the ppl at redhat 2015-05-06 18:50:09 and their coup on linux 2015-05-06 18:50:20 An awesome compromise for the fact that some packages may never be built reproducibly is to only distribute reproducible binaries, and then require your users to compile packages from multi-signature source. 2015-05-06 18:50:25 i'm all for modernizing the init system 2015-05-06 18:50:39 i just dont trust redhat and poettering whatever his name is to do it 2015-05-06 18:50:42 While I don't entirely disagree, don't you think you're over-generalizing just a bit? 2015-05-06 18:50:54 which 2015-05-06 18:51:03 to redhat? 2015-05-06 18:51:03 "the ppl at redhat" 2015-05-06 18:51:06 sure 2015-05-06 18:51:21 i mean 'the ppl at redhat' i speak of a very specific group 2015-05-06 18:51:34 involving poettering, the guy who wrote libc and co, etc 2015-05-06 18:51:47 the corporate leadership 2015-05-06 18:52:04 but as for like the employees and engineers... redhat had a fantastic innovation with lxc 2015-05-06 18:52:19 tbh, I'm no fan of systemd myself, but I still use it 2015-05-06 18:52:21 i use lxc instead of docker... best container technology ever 2015-05-06 18:52:27 <_ikke_> I tend to like 2015-05-06 18:52:28 lxc :D 2015-05-06 18:52:28 and that's redhat provied 2015-05-06 18:52:29 <_ikke_> it* 2015-05-06 18:52:50 depends on which lxc you're talking about 2015-05-06 18:53:07 i just mean like redhats main customers are the dod and military 2015-05-06 18:53:18 libvirt-lxc is indeed RH, but the "lxc" is actually Canonical 2015-05-06 18:53:20 and it kind of creeps into the leadership so they do sneaky crap 2015-05-06 18:53:36 even linus has chewed them out time and again 2015-05-06 18:53:53 for being overbearing 2015-05-06 18:54:10 oh i didnt realize lxc was canonical? 2015-05-06 18:54:19 i was 100% sure it started with two redhat engineers 2015-05-06 18:54:23 the userspace lxc tools are indeed canonical 2015-05-06 18:54:26 awesome brilliant guys 2015-05-06 18:54:28 oh right 2015-05-06 18:54:37 i mean the kernel code i think 2015-05-06 18:54:52 let me see 2015-05-06 18:54:54 well, the kernel code is the same for whatever container solution 2015-05-06 18:55:11 docker, lxc and libvirt-lxc use the same kernel interfaces 2015-05-06 18:55:27 I have to go unfortunately. Thanks for the conversation. 2015-05-06 18:56:20 There's literally one thing I dislike about the (canonical) lxc: It doesn't integrate with systemd-machined (which is kinda understandable..) 2015-05-06 18:56:50 bye fowlslegs^^ 2015-05-06 18:59:32 okay u were right , it was mostly userspace at ubuntu 2015-05-06 18:59:49 but alot of the mailing list traffic is between ubuntu and redhat engineers 2015-05-06 19:00:06 so its ubuntu and redhat were lxc and ubuntu engineers did the userspace part 2015-05-06 19:00:28 okay seeya 2015-05-06 19:00:28 that's it tho, there is no "the userspace part" 2015-05-06 19:00:36 there's multiple implementations 2015-05-06 19:00:39 well u need cgroups etc 2015-05-06 19:00:47 all that functionality took years to bake andperfect 2015-05-06 19:00:59 siince like 2.x that's been buildling no 2015-05-06 19:01:01 the ubuntu "lxc" is one of them, so is redhat's "libvirt-lxc" 2015-05-06 19:01:05 yeah 2015-05-06 19:01:21 im reading now 2015-05-06 19:01:27 https://www.redhat.com/archives/linux-audit/2014-May/msg00020.html 2015-05-06 19:01:37 i mean it seems like redhat did alot of the libvirt stuff 2015-05-06 19:01:48 I generally prefer the canonical version, cause libvirt is a pain to work with IMO 2015-05-06 19:02:02 i didnt realize there were two versions til just now 2015-05-06 19:02:04 haha 2015-05-06 19:02:13 yeah, libvirt is RH 2015-05-06 19:02:31 It's actually a nice interface to qemu when paired with virt-manager 2015-05-06 19:02:47 but working with libvirt without a gui is painful 2015-05-06 19:02:47 oh really? 2015-05-06 19:02:56 i just use qemu-system-x86_64 2015-05-06 19:03:06 They make you shuffle around XML left and right 2015-05-06 19:03:10 didnt even know there was abetter interface 2015-05-06 19:03:15 naw not really , not for basic stuff 2015-05-06 19:03:21 like testing your kernel 2015-05-06 19:03:30 bootloader, initrd, 2015-05-06 19:03:42 but yeah for large scale stuff it'd be a nightmare 2015-05-06 19:03:48 well, libvirt is more for setting up VMs, like VirtualBox, not testing kernels 2015-05-06 19:04:01 yeah i mean i set up vms with qemu 2015-05-06 19:04:04 for the latter, using qemu directly is probably the best solution 2015-05-06 19:04:04 but its simple to me 2015-05-06 19:04:20 i just do like dd nd make an 8gig file of zeros called like disk.raw 2015-05-06 19:04:29 mount it on loop0 with size limits 2015-05-06 19:04:37 then tell qemu that's an ide device 2015-05-06 19:04:44 well, "VMs" are of the complex variety in my book^^ 2015-05-06 19:04:46 oh do like mkfs /dev/loop0 2015-05-06 19:04:59 install the OS or whatever 2015-05-06 19:05:05 stuff like running a copy of windows with all the bells and whistles 2015-05-06 19:05:06 but i use virtualbox and vmware too 2015-05-06 19:05:17 wouldn't want to think up the qemu cmdline for that :P 2015-05-06 19:05:22 i prefer vmware for fullscale stuff 2015-05-06 19:05:50 I like using the tools Linux gives me, where viable :) 2015-05-06 19:05:52 since it has hotswappable usb, pci virtualization, the vmware tools 2015-05-06 19:06:02 hahah 2015-05-06 19:06:03 yeah 2015-05-06 19:06:31 well, have you ever tried qemu+kvm for the complex stuff? 2015-05-06 19:07:30 I think the GNOME guys even made a (which they tend to do these days...) so-simple-it-almost-hurts interface to qemu+kvm recently 2015-05-06 19:07:42 It's called Box or Boxed or something like that 2015-05-06 19:10:05 yes i use kvm 2015-05-06 19:10:06 i love i 2015-05-06 19:10:08 it 2015-05-06 19:10:18 iuse it mostly for sandboxes 2015-05-06 19:10:31 they are lower overhead than vmware if u dont need a gui and u ssh in 2015-05-06 19:10:43 like if there's no X11 2015-05-06 19:10:50 btw... different topic 2015-05-06 19:10:58 cheq this out... cool site i found 2015-05-06 19:11:06 http://sta.li/sandbox 2015-05-06 19:11:18 "all applications need dbus to be removed" 2015-05-06 19:11:21 i like the sound of that haha 2015-05-06 19:12:02 I only use kvm for Windows, I use containers for sandboxing linux stuff 2015-05-06 19:13:29 containers are no sandboxes 2015-05-06 19:14:19 containers are for incompetence, sandboxes for malice 2015-05-06 19:14:46 sure, but I don't need real "sandboxes" 2015-05-06 19:17:38 kvm is not a container brah 2015-05-06 19:17:51 Of course not 2015-05-06 19:17:55 not u 2015-05-06 19:18:10 ylVa333UKfLO was talking to me^^ 2015-05-06 19:18:13 oh 2015-05-06 19:18:15 :) 2015-05-06 19:18:16 nvm 2015-05-06 19:18:30 btw cheq this 2015-05-06 19:18:31 http://morpheus.2f30.org/ 2015-05-06 19:18:50 Is there a way to disable the library dependency auto-detect? 2015-05-06 19:18:57 man im gonna build statically linked linux 2015-05-06 19:19:09 what do u mean 2015-05-06 19:19:19 "library dependency auto-detect" 2015-05-06 19:19:36 what tool is that in and what does it do in your description 2015-05-06 19:19:46 abuild 2015-05-06 19:20:01 does an ldd on the binary, and sets those as deps 2015-05-06 19:20:47 I know 2015-05-06 19:20:51 I don't want it 2015-05-06 19:21:05 because I do magic with LD_LIBRARY_PATH 2015-05-06 19:21:15 (don't kill me for it D:) 2015-05-06 19:21:20 actually let me correct my statement: containers are for incompetence, sandboxes for incompetent malice 2015-05-06 19:22:17 ohh 2015-05-06 19:22:22 that sucks 2015-05-06 19:22:23 man 2015-05-06 19:22:41 i was so close to installing alpine linux on my new laptop thats coming tomorrow 2015-05-06 19:22:52 but now i really want to have a statically linked system 2015-05-06 19:22:58 100% static on musl 2015-05-06 19:23:13 with any dynamic stuff for compatibility goes in in a chroot or in an aufus layers 2015-05-06 19:23:25 but even X i want built statically 2015-05-06 19:23:52 maybe i'll try building my static linux then install apk on it 2015-05-06 19:23:55 I don't get why people are so fond of static builds... 2015-05-06 19:24:06 cause then i just need to pass in -static to ld 2015-05-06 19:24:13 well many reasons 2015-05-06 19:24:21 for me anyway 2015-05-06 19:24:34 (1) more secure, no trash hanging around your system 2015-05-06 19:24:45 (2) way faster, less bloat 2015-05-06 19:24:55 (3) most important: no dependency hell 2015-05-06 19:25:05 Hey guys quick question, i'm attemping to compile apt-cacher-ng but missing a lot of header files. (sys/socketvar.h for example). Is there a lib I should install first that provides these? 2015-05-06 19:25:27 can you do aslr with static linked binaries? 2015-05-06 19:25:31 ie. packages include what they need to run so it's not necessary to have nutty large scale dependency tracking 2015-05-06 19:25:50 all u have to track is what package puts what files where 2015-05-06 19:26:03 so package mgmt becomes either an aufs layer 2015-05-06 19:26:16 or it becomes version control of / of your system like a git repo 2015-05-06 19:26:25 with checkpoints at each package install 2015-05-06 19:26:32 but u dont use git u use bup 2015-05-06 19:26:35 heh 2015-05-06 19:26:45 those are my reasons... #3 is the biggest 2015-05-06 19:26:49 the other motivation is this: 2015-05-06 19:26:57 i ran an experiment on ubuntu 14 2015-05-06 19:27:02 let me ask again: can you do aslr with static linked binaries? 2015-05-06 19:27:06 Can anyone tell a noob what lib/package provides sys/socketvar.h? 2015-05-06 19:27:20 no 2015-05-06 19:27:35 are you answering me or Guest10351 ? 2015-05-06 19:27:37 ylV generally no ,but i thik a few smart ppl have figured out some trix 2015-05-06 19:27:46 there is some info on it onn that blog i linked 2015-05-06 19:27:47 u 2015-05-06 19:27:56 on the address space randomization of static linking 2015-05-06 19:28:03 u can randomize the stack and heap and thatsit 2015-05-06 19:28:37 some ppl have ideaz and claim there are tricks to it, 2015-05-06 19:28:41 there is a paper here: http://benpfaff.org/papers/asrandom.pdf 2015-05-06 19:28:53 "on the effectiveness of address space randomization" 2015-05-06 19:29:06 I'm sure you know that static linkage has drawbacks too^^ 2015-05-06 19:29:06 but im really sort of an idiot when it comes to that sort of thing 2015-05-06 19:29:19 i havent written exploits in 10 years 2015-05-06 19:29:26 well, it raises costs for the attacker. as a defense in depth mitigation it is something you don't want to go without, if done correctly 2015-05-06 19:29:38 well that's what i was getting at Oroch.. yeah it's a tradeoff in some respects 2015-05-06 19:29:45 security is clearly one 2015-05-06 19:30:04 but what i was saying is that on ubuntu the dynamic linked nmap binary is about 13M 2015-05-06 19:30:15 unstripped as default install from source 2015-05-06 19:31:09 Guest10351: use the package browser for that. 2015-05-06 19:31:33 seems like nothing provides that. 2015-05-06 19:31:47 the static linked unstripped nmap binary was 17m 2015-05-06 19:31:55 and the stripped static nmap binary was 6.8M 2015-05-06 19:32:01 smaller than the dynamic linked verison 2015-05-06 19:32:03 I don't see this endeavor going anywhere >.< 2015-05-06 19:32:40 huh 2015-05-06 19:32:58 u mean me or ylv or land 2015-05-06 19:33:01 cland* 2015-05-06 19:33:10 I mean myself 2015-05-06 19:33:40 @clandmeter I will definitely! Where can I find this tool? 2015-05-06 19:33:43 oh kk 2015-05-06 19:33:56 static linking advantages: http://wayback.archive.org/web/20090525150626/http://blog.garbe.us/2008/02/08/01_Static_linking/ 2015-05-06 19:33:59 I mean 2015-05-06 19:34:04 Guest10351: http://pkgs.alpinelinux.org/contents?filename=socketvar.h&path=&pkgname=&arch=x86 2015-05-06 19:34:14 clandmeter: 2015-05-06 19:34:37 but its probably glibc specific 2015-05-06 19:34:41 clandmeter: oh this is so awesome. where can I read more about this stuff so I understand alpine better? 2015-05-06 19:35:01 define "this stuff" 2015-05-06 19:35:33 guest your best bet is use the package browser if u are curiosu about missing stuff 2015-05-06 19:35:39 tools for compiling and developing on alpine 2015-05-06 19:35:50 its actually better than other distros imo... cuz u can serach missing files really quick 2015-05-06 19:36:00 well alpine is not like a desktop linux atm 2015-05-06 19:36:06 so u gotta get yr hands dirty 2015-05-06 19:36:07 So I'm missing socketvar.h but it doesn't appear to be in a package 2015-05-06 19:36:24 but the main thing to remember is this: u can put ubuntu or other linux binaries on alpine 2015-05-06 19:36:27 they wont run 2015-05-06 19:36:33 u need to either use the packages with apk 2015-05-06 19:36:42 or build from source using make / make install 2015-05-06 19:36:50 Right now I'm attempting to compile apt-cacher-ng and build a package 2015-05-06 19:36:51 thats cause alpine replaced bloated glibc 2015-05-06 19:36:54 oh kk 2015-05-06 19:36:57 u are fine then 2015-05-06 19:37:07 But I'm missing socketvar.h and can't find the library that provides it :( 2015-05-06 19:37:17 tthat's what cland linked you 2015-05-06 19:37:21 Guest10351: glibc provies it 2015-05-06 19:37:27 oh 2015-05-06 19:37:30 well wait 2015-05-06 19:37:31 So what would be the next step for someone like me? 2015-05-06 19:37:34 so musl doesnt? 2015-05-06 19:37:35 ohhh 2015-05-06 19:37:38 u need musl-dev 2015-05-06 19:37:41 thats what you need 2015-05-06 19:37:41 doesnt seem like it 2015-05-06 19:37:46 apk add musl-dev 2015-05-06 19:37:48 that's the one 2015-05-06 19:37:53 nah 2015-05-06 19:37:55 its not in there 2015-05-06 19:37:59 check pkgs.a.o 2015-05-06 19:38:10 oh u need the header 2015-05-06 19:38:11 already have that package. no dice 2015-05-06 19:38:20 i was gnna say musl-dev will have the object to link 2015-05-06 19:38:25 but the header hmm 2015-05-06 19:38:32 http://packages.ubuntu.com/search?searchon=contents&keywords=socketvar.h&mode=exactfilename&suite=trusty&arch=any 2015-05-06 19:39:12 check the file it includes 2015-05-06 19:39:20 could be its messed up ifdefs 2015-05-06 19:39:26 A lot of people recommending libc6-dev-i386 for ubuntu 2015-05-06 19:39:37 socketvar.h is not linux 2015-05-06 19:39:44 not posix i mean 2015-05-06 19:39:48 hm 2015-05-06 19:39:48 it's bsd derived 2015-05-06 19:40:03 looks like it comes from 4.3BSD 2015-05-06 19:40:12 I actually started downloading bsd headers such as this, but eventually compiling just complained, so I didn't think that was the best route 2015-05-06 19:40:24 and its on android too apparently? 2015-05-06 19:40:34 but i think on linux 2015-05-06 19:40:37 its in glibc 2015-05-06 19:40:45 and somehow made its way into glibc from bsd4.3 2015-05-06 19:40:55 so it'd be in the glibc headers 2015-05-06 19:41:08 musl probably didnt provide it because its not posix compliant 2015-05-06 19:41:18 or its buggy 2015-05-06 19:41:21 someting like that 2015-05-06 19:41:23 hm 2015-05-06 19:41:52 http://ftp.netbsd.org/pub/NetBSD/NetBSD-current/src/sys/sys/socketvar.h 2015-05-06 19:41:56 okay so what u can do i think 2015-05-06 19:42:07 is install glibc headers with socketvar and other crap 2015-05-06 19:42:25 and link against musl... thatmight work 2015-05-06 19:42:48 your more reliable option is to simply use a container like lxc or chroot and pass in glibc 2015-05-06 19:42:57 dont install glibc headers... 2015-05-06 19:43:02 and develop within there ... or in qemu, virtualbox, vmware 2015-05-06 19:43:05 yeah that's what i mean 2015-05-06 19:43:14 put them in a walled off part of the OS 2015-05-06 19:43:29 u can chroot and --bind all your root fs in read only 2015-05-06 19:43:32 How do I download the glibc headers? 2015-05-06 19:43:39 then install whatever u want without messing up the main OS 2015-05-06 19:43:48 honestly 2015-05-06 19:43:54 id go for an easier project to start 2015-05-06 19:43:56 Also, is this going to make it really complicated to build a pkg? 2015-05-06 19:44:01 like that's a tough project 2015-05-06 19:44:11 I feel like I can spend less time writing a cacher in go ;) 2015-05-06 19:44:18 porting incompatible glibc to musl is something that i'd consider hard / headache inducing 2015-05-06 19:44:29 and i'm pretty good devleoper even tho i cantz spell 2015-05-06 19:44:36 Well hopefully somebody stumbles onto this IRC history when attempting to compile apt-cacher-ng on alpine linux 2015-05-06 19:45:00 heheh 2015-05-06 19:45:01 I see squid is already built so I'll just use that 2015-05-06 19:45:06 well one thing u can do is this 2015-05-06 19:45:10 okay i got it 2015-05-06 19:45:28 (1) if u have another linux OS, thats glibc based, compile your apt-cacher-ng there 2015-05-06 19:45:39 just make sure that the kernel version is older than alpine 2015-05-06 19:45:48 okay when u compile pass in -static 2015-05-06 19:45:53 to LD 2015-05-06 19:46:20 so like ./configure LDFLAGS=-static 2015-05-06 19:46:20 interesting 2015-05-06 19:46:26 or make LDFLAGS=-static 2015-05-06 19:46:34 or even just gcc -c *.o and then 2015-05-06 19:46:46 gcc -static -o my_binary_elf obj1.o obj2.o etc 2015-05-06 19:46:50 -lc 2015-05-06 19:46:52 -lm 2015-05-06 19:46:54 whatever 2015-05-06 19:47:06 and u will get portable static linked binary that will run on any linux 2015-05-06 19:47:23 just make sure if u are x86_64 on alpine that u compile on an x86_64 distro 2015-05-06 19:47:38 same for 32 bit 2015-05-06 19:47:45 but besides that , your static binary will run on ANY linux with just the one file 2015-05-06 19:47:59 awesome I'll give that a shot 2015-05-06 19:48:01 u can check after compile to make sure that it's a clean static lib by doing 2015-05-06 19:48:06 ldd 2015-05-06 19:48:18 it should have nothin found as externals 2015-05-06 19:49:01 if u get to the ldd part and its clean, im 100% sure it will run on any linux with a kernel version later than the box u built it on 2015-05-06 19:49:11 recent kernel tho like 3.x series 2015-05-06 19:49:13 not 2.2 2015-05-06 19:49:16 haha 2015-05-06 19:49:41 you could try using linux-headers & #include instead 2015-05-06 19:50:31 actually is all you need 2015-05-06 19:51:00 socketvar.h is just #include 2015-05-06 19:51:42 there u go 2015-05-06 19:51:53 nice find! 2015-05-06 19:52:15 for things like this packages.debian.org helps a lot 2015-05-06 19:53:26 This little project of mine gets more ridiculous by the second ... 2015-05-06 19:54:06 <_ikke_> yak shaving? 2015-05-06 19:59:41 Alpine makes me realize how stupid binary-only blobs are >.< 2015-05-06 20:03:57 w=what are ya trying to do 2015-05-06 20:04:05 run PMS on alpine 2015-05-06 20:04:09 btw i think im gonna go for morpheus 2015-05-06 20:04:24 static linked minimal distro with aufs snapshots for pkg mgmt 2015-05-06 20:04:33 and it should run apk binaries too 2015-05-06 20:04:48 since it also uses mucl i thin 2015-05-06 20:04:53 PMS=?? 2015-05-06 20:05:01 Plex Media Server 2015-05-06 20:05:10 k neverheard of it 2015-05-06 20:05:17 but im not well versed in media/video 2015-05-06 20:05:29 it's binary only and obviously linked against glibc -.- 2015-05-06 20:05:32 what's the error its throwing 2015-05-06 20:05:42 oh man 2015-05-06 20:05:44 it complains about paths 2015-05-06 20:05:59 because it's not expecting to not be argv[0] 2015-05-06 20:07:10 well you've got a couple of options i think honestly... (a) run thru emulator either host or kernel (qemu), (b) run fully virtualized in vmware/virtualbox, or (c) use ldd to find the shared libraries it needs, pull them off another distro, and run it within a filesystem sandbox 2015-05-06 20:07:20 like a unionfs mount or whatever 2015-05-06 20:07:31 but since u dont have source, your options are limited 2015-05-06 20:07:32 yeah but libc is somewhat special 2015-05-06 20:07:49 the only other thing i can think is that there are programs to patch binaries linked against glibc 2015-05-06 20:07:49 because the path to ld-linux.so is hardcoded into executables 2015-05-06 20:07:58 right well that's what im saying 2015-05-06 20:08:01 theres a tool to change that 2015-05-06 20:08:05 right 2015-05-06 20:08:07 hang on let me find it 2015-05-06 20:08:12 that'd be my next stop 2015-05-06 20:08:26 or do u have it 2015-05-06 20:08:30 kk 2015-05-06 20:08:32 one sec 2015-05-06 20:08:46 alpine has http://kodi.tv for media 2015-05-06 20:08:54 https://github.com/NixOS/patchelf 2015-05-06 20:09:04 right now i'm trying to let the glibc ld-linux.so "interpret" the executable 2015-05-06 20:09:08 thats the one sane tingto come out of nixos 2015-05-06 20:09:09 haha 2015-05-06 20:09:11 sane thing* 2015-05-06 20:09:19 BitL0G1c: sure, but i'm talking about the Plex Media _Server_ 2015-05-06 20:09:19 the rest of it is bloatware 2015-05-06 20:09:35 but u can use that tool to patch the dynamic linker and replace the path 2015-05-06 20:09:58 this tool looks awesome, thanx :O 2015-05-06 20:10:00 i'd still put it in a sandbox of some sort if u could 2015-05-06 20:10:08 but it will probably work just ine without one 2015-05-06 20:10:26 if u get the exact binary versions of the libraries (and ALL their dependencies) 2015-05-06 20:10:30 its recursive 2015-05-06 20:10:51 that's why i'm so in favor of static linking. u dont have that problem. 2015-05-06 20:10:52 I like making software do stuff MY WAY, not letting it dictate the way I do stuff 2015-05-06 20:11:01 me too 2015-05-06 20:11:10 it already ships everything except libc 2015-05-06 20:11:17 hm? 2015-05-06 20:11:26 u mean alpine? 2015-05-06 20:11:29 no, pms 2015-05-06 20:11:35 oh k 2015-05-06 20:11:40 so its missing 1 library? 2015-05-06 20:11:48 that's totally doable 2015-05-06 20:12:05 it should be 2015-05-06 20:12:10 now remember u cant point it to musl 2015-05-06 20:12:14 wel most likely not 2015-05-06 20:12:20 yeah that doesn't work 2015-05-06 20:12:33 u gotta find the libc off like ubuntu's repo 2015-05-06 20:12:38 I already got the gnu stuff set up adn ready to go 2015-05-06 20:12:54 the other thing to remember for abi is u want to pick libc compiled against an OLDER kernel than alpine 2015-05-06 20:13:00 Ebisu:~/aports/custom/plex-media-server$ ls /glibc/lib 2015-05-06 20:13:00 ld-2.19.so libc-2.19.so libdl.so.2 libnss_compat-2.19.so libnss_hesiod.so.2 libpthread.so.0 libthread_db-1.0.so 2015-05-06 20:13:02 if u pick libc compiled against a newer kernel it will fail 2015-05-06 20:13:05 [snap] 2015-05-06 20:13:50 when i say libc i mean glibc that u are gonna splice into plex 2015-05-06 20:13:50 from ubuntu or whatever 2015-05-06 20:13:50 I'm running the Arch kernel, I'm not aware of any distro that ships a newer kernel than that by default :P 2015-05-06 20:14:26 yeah, I got glibc installed to /glibc/libs 2015-05-06 20:15:29 root@computer:/root/src# ls -al /lib/x86_64-linux-gnu/libc-2.19.so 2015-05-06 20:15:29 -rwxr-xr-x 1 root root 1840064 Feb 25 09:08 /lib/x86_64-linux-gnu/libc-2.19.so 2015-05-06 20:15:29 root@computer:/root/src# ldd /lib/x86_64-linux-gnu/libc-2.19.so 2015-05-06 20:15:29 /lib64/ld-linux-x86-64.so.2 (0x00007f1b297d2000) 2015-05-06 20:15:29 linux-vdso.so.1 => (0x00007fff6a7fa000) 2015-05-06 20:16:20 Ebisu:~/aports/custom/patchelf$ ls -l /glibc/lib/libc-2.19.so 2015-05-06 20:16:20 -rwxr-xr-x 1 root root 1729984 May 6 19:55 /glibc/lib/libc-2.19.so 2015-05-06 20:16:27 :P 2015-05-06 20:16:52 vdso i think might be optional 2015-05-06 20:16:55 not sure 2015-05-06 20:17:07 andldis the dynamic linker obvs ,,uve got that on alpine 2015-05-06 20:17:13 I just got the whole package off debian stable 2015-05-06 20:17:22 so theres no other dependencies for glibc 2015-05-06 20:17:53 yeah the only thing im not 100% sure on is vdso 2015-05-06 20:18:06 i know is virtual on my glibc based system but on musl i have no ide 2015-05-06 20:18:23 oh one last thing... remember to install musl-dev , gcc , and gcc dev stuff 2015-05-06 20:18:52 nothing compiles on alpine w/o musl-dev 2015-05-06 20:18:52 not proprly 2015-05-06 20:19:37 install alpine-sdk in an LXC container 2015-05-06 20:19:38 okay one last point.... i just realized this 2015-05-06 20:20:05 the dynamic linker normally has a config file on debian based systems in etc 2015-05-06 20:20:08 ld.so.conf or sumt 2015-05-06 20:20:22 that tells the linker where to look at runtime 2015-05-06 20:20:29 to splice stuff in 2015-05-06 20:20:56 ldconfig uses that config to rebuild the cache of what libs are where 2015-05-06 20:21:07 (FYI: i've been doing custom APKBUILDS for a couple months now ;) 2015-05-06 20:21:13 oh okay 2015-05-06 20:21:18 well i havent haha 2015-05-06 20:21:37 so most of this is trying to get my head around installing my static linux distro 2015-05-06 20:21:57 u clearly are firing on all cylinders im just thinking out loud 2015-05-06 20:22:05 greetings to all, alpine supports Raspberry Pi 2 Model B ? https://www.raspberrypi.org/products/raspberry-pi-2-model-b/ 2015-05-06 20:22:58 hey 2015-05-06 20:23:12 that's great.. i got a beaglebone black and i know its not easy 2015-05-06 20:23:15 since it's on ARM 2015-05-06 20:23:33 but that's cool 2015-05-06 20:24:43 man i think if my static linux goes well 2015-05-06 20:24:50 i'm going to make static alpine 2015-05-06 20:24:58 cuz that would be optimal 2015-05-06 20:25:09 static alpine with container virtualization for legacy stuff 2015-05-06 20:25:33 snapshots of whatever u want via block device, filesystem, or git deltacomp 2015-05-06 20:25:37 or augs 2015-05-06 20:25:41 that woulld be so cool 2015-05-06 20:31:39 do I pass --prefix=/usr or --prefix=$pkgdir/usr to ./configure? I forget >.< 2015-05-06 20:32:45 --prefix=/usr to configure 2015-05-06 20:33:05 $pkgdir/xxxx to make install 2015-05-06 20:33:25 in what way? 2015-05-06 20:33:36 like, how do I pass it to make install 2015-05-06 20:33:54 Never liked autotools much :/ 2015-05-06 20:36:43 for what 2015-05-06 20:36:57 i'd do --prefix=/usr or /usr/local 2015-05-06 20:37:22 where it goes exactly depends on the makefile 2015-05-06 20:37:37 what do in these cases is make an /opt diectory 2015-05-06 20:37:46 ./configure --prefix=/opt 2015-05-06 20:38:10 but u shold be fine with /usr... if u install to opt u gotta update your environment vars like path etc 2015-05-06 20:38:33 but it makes it easy to delete the code and reinstall it without hunting down the files 2015-05-06 20:38:45 that's what package managers are for 2015-05-06 20:38:51 (this was my whole idea behind the aufs package management) 2015-05-06 20:39:02 it would work for any build 2015-05-06 20:39:04 what's the best way to find all executables in a folder? 2015-05-06 20:39:21 find /path/to/folder -type f | filter 2015-05-06 20:41:28 I guess find /path -type f -perm /a+x? 2015-05-06 20:43:39 sorr 2015-05-06 20:43:54 i normally use a filter like sed or grep 2015-05-06 20:44:17 but heres a better way 2015-05-06 20:44:25 find /path/to/folder -type f -executable 2015-05-06 20:44:55 uhm... I dont think so? 2015-05-06 20:45:10 yeah, doesn't work 2015-05-06 20:45:29 well that was on the man page 2015-05-06 20:45:38 find unp/usr/lib/plexmediaserver/ -type f -perm /0111 2015-05-06 20:45:41 that works 2015-05-06 20:52:55 uuhm... it seems to work :D 2015-05-06 20:53:19 sure does :D 2015-05-06 20:55:17 sorry was afk 2015-05-06 20:55:18 itz find /chome/user/Downloads/pkgsrc -type f -perm /a=x 2015-05-06 20:55:18 thats the syntax to do it without pipes andfilters 2015-05-06 20:55:43 that finds any regular file recursively that matches user, group, or world as executable 2015-05-06 20:56:27 to do it for user execute only you'd do u=x instead 2015-05-06 20:56:38 the detailz are on the man page 2015-05-06 20:56:59 yeah okay u did the same thing in octal 2015-05-06 20:57:03 nice 2015-05-06 21:09:36 oh Oroch 2015-05-06 21:09:50 this might help u -- im reading about linux dynamic linking 2015-05-06 21:09:52 "The lovely land of modern Unix will allow you to have multiple versions of the same library installed at the same time. That's not the problem. The problem is that when you load the libraries all the symbols are resolved inside the same namespace. There can't be two versions in use of the same function with the same name, even if they are in different libraries (well, it won't work the way you want it to). You can't choose which one to see 2015-05-06 21:09:52 from different code. (see breadth first search description here for details)." 2015-05-06 21:09:59 https://blog.habets.se/2012/05/Shared-libraries-diamond-problem 2015-05-06 21:22:36 Alright now 2015-05-06 21:22:55 HOW DO I STOP abuild FROM ADDING ALL THOSE USELESS DEPENDENCIES D: 2015-05-06 21:23:07 (sorry for the caps >.<) 2015-05-06 21:31:02 haha 2015-05-06 21:31:09 by using bitshark static ram linux 2015-05-06 21:31:21 where all packages are static and self contained 2015-05-06 21:31:31 no srsly, I get 11 libs pulled in by installing a self-contained package 2015-05-06 21:31:35 that's ridiculous 2015-05-06 21:31:53 yeah 2015-05-06 21:31:54 well most linux's just hide it 2015-05-06 21:32:03 i tried using nix and nixos on a linux from scratch usb i made 2015-05-06 21:32:10 it pulled in 65000 files 2015-05-06 21:32:17 most were symlinks but still 2015-05-06 21:32:29 that was just to install xfce and X11 2015-05-06 21:33:00 so 11 is not bad... i think alpine is actually one of the best distros in terms of unneeded crap getting pulled in 2015-05-06 21:33:04 but its stil alot 2015-05-06 21:33:10 someone should fix the abuild/APKBUILD documentation >.< 2015-05-06 21:33:19 cuz if your binary even uses a single function from a 10mb shared library 2015-05-06 21:33:28 the whole 10mb library becomes a dependency 2015-05-06 21:33:34 even if u call 500 bytes of the thin 2015-05-06 21:33:49 yeah, agreed 2015-05-06 21:33:54 the point is, it DOESN'T USE the libraries at all, because it SHIPS CUSTOM VERSIONS and MODIFIES LD_LIBRARY_PATH 2015-05-06 21:33:59 i wanted to start getting into abuild / APKBUILDS 2015-05-06 21:34:09 but i got scared off cuz the docuentation is sparse 2015-05-06 21:34:11 of which abuild has no clue and pulls the system version in ON TOP OF IT 2015-05-06 21:34:21 you can say that again >.< 2015-05-06 21:34:24 what's there is well written, there's just not enohg of it 2015-05-06 21:34:35 like its like one page per topic 2015-05-06 21:34:36 for future reference: use the !tracedeps option 2015-05-06 21:34:42 ah nice 2015-05-06 21:35:20 This is more like it :D 2015-05-06 21:35:23 Shibuya:~# apk add plex-media-server@custom 2015-05-06 21:35:23 (1/2) Installing glibc-runtime@custom (2.19-r0) 2015-05-06 21:35:23 (2/2) Installing plex-media-server@custom (0.9.12.1.1079-r1) 2015-05-06 21:35:23 Executing plex-media-server-0.9.12.1.1079-r1.pre-install 2015-05-06 21:35:23 OK: 532 MiB in 177 packages 2015-05-06 21:35:31 oh wow 2015-05-06 21:35:33 that's impressive 2015-05-06 21:35:40 well done on the package 2015-05-06 21:36:01 still a bit messy but we'll see 2015-05-06 21:36:02 definitely merge that in if if u can 2015-05-06 21:36:05 that's cool 2015-05-06 21:36:11 i wish i learned howto do that haha 2015-05-06 21:36:17 sooner or later 2015-05-06 21:36:20 i wiill 2015-05-06 21:36:23 I don't think a patch that pulls in glibc would be accepted :P 2015-05-06 21:37:05 I need to put my custom APKBUILDs on github tho 2015-05-06 21:37:14 oh 2015-05-06 21:37:19 wel i think it would actually 2015-05-06 21:37:27 ud just hae to make it part of a containerization engine 2015-05-06 21:38:25 ie. if u pulled in glibc into like /emul 2015-05-06 21:38:37 where that's a dedicated directory to legacy emulation 2015-05-06 21:39:01 i have it in /glibc ^^ 2015-05-06 21:39:04 and maybe u do like mount / --bind -o ro /emul 2015-05-06 21:39:07 then chroot /emul 2015-05-06 21:39:14 and u have fs container more or less 2015-05-06 21:39:24 no chroots, that's what patchelf is for 2015-05-06 21:39:33 heh 2015-05-06 21:39:48 well the point is that u dont want to clutter up a nice clean system 2015-05-06 21:39:57 with the interlocking dependencies from somewhere else 2015-05-06 21:40:17 im sure your patch is fine but if everyone went down that road soon we'd have ubuntu 2015-05-06 21:40:18 find . -type f -perm /0111 -exec patchelf --set-interpreter /glibc/lib/ld-2.19.so \{\} \; 2015-05-06 21:40:19 :P 2015-05-06 21:40:42 plus the containerization stuff is such a buzz right now 2015-05-06 21:40:55 supporting it will help draw power users and developers capable of contributing 2015-05-06 21:41:01 I don't know what you're getting at but whatever 2015-05-06 21:41:16 As I said, I have everything glibc-related in /glibc/* 2015-05-06 21:41:33 i'm saying it's nice to (a) get your stuff merged and (b) have the OS u spend time on get better and have a good community 2015-05-06 21:41:38 that's what i was getting at 2015-05-06 21:41:57 not denying that 2015-05-06 21:42:14 if u just wanted to run video on alpine and that was it that's cool too 2015-05-06 21:42:14 and huzzah for getting it working 2015-05-06 21:42:27 thanks to nix's elfpatcher haha 2015-05-06 21:42:38 indeed 2015-05-06 21:42:58 That's actually a package that I might submit to the aports tree :) 2015-05-06 21:43:07 yeah man i say do it 2015-05-06 21:46:20 it seems to work :D 2015-05-06 21:46:27 plex-media-server [ started ] 2015-05-06 21:48:47 rad 2015-05-06 21:50:07 hm 2015-05-06 21:50:44 see the ideal use case for glibc on a clean secure distro is as a last resort 2015-05-06 21:50:54 preferably in some sort ofcontainer 2015-05-06 21:51:01 but ths is a good example of that use case 2015-05-06 21:51:05 binary only software 2015-05-06 21:51:24 is this a free thing i can download as well? 2015-05-06 21:51:44 i'd like to use it to test containerization of glibc in morpheus linux 2015-05-06 21:51:51 plex? yeah http://plex.tv 2015-05-06 21:52:03 (like alpine but even smmaller build, all stattic linked) 2015-05-06 21:52:35 just booted it in qemu and its really fast 2015-05-06 21:52:36 http://morpheus.2f30.org/ 2015-05-06 21:53:15 anyway i think this video stuff we discussed today would be a good test for namespace isolation for glibc 2015-05-06 21:53:22 "Separate /emul namespace for legacy or binary-blob software (skype, wine, robot-unicorn-attack)." 2015-05-06 21:53:42 thanks for the link, and glad u got it working! 2015-05-06 21:54:02 it's not working yet 2015-05-06 21:54:10 oh 2015-05-06 21:54:13 it's launching, but I get a HTTP 500 2015-05-06 21:54:15 well you'll get it 2015-05-06 21:54:39 it's probably missing python or wsgi or php 2015-05-06 21:54:45 some stupid thing like that 2015-05-06 21:55:09 attach a debugger to the process and see if it's segfaulting 2015-05-06 21:55:22 it's not 2015-05-06 21:55:48 other thing is to do strings to see if u can get hints on what it was written in 2015-05-06 21:55:52 and how it works 2015-05-06 21:56:13 and the last is netstat -a to see how its using tcp ports and unix pipes etc 2015-05-06 21:56:15 it's C++, cause it uses boost :P 2015-05-06 21:56:19 okay 2015-05-06 21:56:26 yeah so most likely actually 2015-05-06 21:56:29 hm 2015-05-06 21:56:35 i dont know what it is 2015-05-06 21:56:46 but is suspect if you compiled libboost against musl 2015-05-06 21:56:54 thats gonna be a problem 2015-05-06 21:57:00 musl doesnt work well at all with C++ 2015-05-06 21:57:06 so that's one idea 2015-05-06 21:57:14 it brings its own libboost 2015-05-06 21:57:17 k 2015-05-06 21:57:18 also, it's not crashing 2015-05-06 21:57:22 just not working 2015-05-06 21:57:27 another idea is to run wireshark on lo 2015-05-06 21:57:51 to see if you've got packets and if so wat they look ike 2015-05-06 21:57:53 wireshark or tcpdump dependin on if youve got X 2015-05-06 21:58:03 those are my 2c ideas heh 2015-05-06 22:08:27 May 06, 2015 22:04:06 [0x7f40a73ff700] ERROR - Exception handled: locale::facet::_S_create_c_locale name not valid 2015-05-06 22:08:30 wtf 2015-05-06 22:14:43 easy way to fix that is seen https://github.com/bitcoin/bitcoin/issues/4147 2015-05-06 22:15:55 export LANG=en.utf8 - will also fix it if you don't want to patch the c code 2015-05-06 22:16:42 I just put 'export LANG=en.utf8' in /etc/profile normally 2015-05-06 22:17:02 shipping it in the start_pms script should work 2015-05-06 22:18:02 it still happens tho >.< 2015-05-06 22:18:27 also, /etc/profile seems to have it by default 2015-05-06 22:22:32 is there a way to kill all processes a user is running using just busybox? 2015-05-06 22:23:43 still happens >.< 2015-05-06 22:25:09 setenv("LC_ALL", "C", 1); sets a default environment in c++ 2015-05-06 22:25:15 http://www.unix.com/aix/177466-how-find-out-kill-all-processes-user.html 2015-05-06 22:25:31 ohhh 2015-05-06 22:25:32 doesn't help, you don't get plex' source 2015-05-06 22:25:33 that's easy 2015-05-06 22:25:41 yeah the locale thing has screwed me before 2015-05-06 22:25:49 set LANG=POSIX 2015-05-06 22:25:52 or LANG=C 2015-05-06 22:25:58 (they are the same) 2015-05-06 22:26:07 WOO 2015-05-06 22:26:08 all posix operating systems have to support that 2015-05-06 22:26:16 IT F'ING WORKS 2015-05-06 22:26:21 even better is export LC_ALL=C 2015-05-06 22:26:26 that overrides global locale 2015-05-06 22:26:33 put it in .profile or .bashrc 2015-05-06 22:26:37 nice , hell yeah 2015-05-06 22:26:39 I'm running plex on alpine :D 2015-05-06 22:26:43 high five! 2015-05-06 22:26:58 ACTION gives bitshark a high five :) 2015-05-06 22:27:06 what did u do ... change the locale? 2015-05-06 22:27:13 im really interested now 2015-05-06 22:27:38 im thinking of porting apk to support a static distro with automatic sandboxing / containers 2015-05-06 22:27:46 of legacy crap 2015-05-06 22:28:33 so im curious as to the details so i can replicate this on morpheus linux , maybe with apk as pkg mgr 2015-05-06 22:45:42 awesome 2015-05-06 22:51:53 BitL0G1c: you were talking about make install and the destination path before, but I didn't quite get how I'm supposed to pass it to make install, care to explain it for me? 2015-05-06 22:53:13 make install DESTDIR="$pkgdir" || return 1 2015-05-06 22:53:25 aha that's exactly what I was looking for 2015-05-06 22:53:26 set that in package() normally 2015-05-06 22:57:00 grep -r "whatever" ~/aports - helps a lot when you're stuck 2015-05-06 22:57:14 probably true 2015-05-06 22:57:24 I whish I could just find some good docs tho 2015-05-06 23:05:55 All right, for anyone interested: https://github.com/Orochimarufan/custom-aports 2015-05-07 10:02:17 ncopa, is there any CPUs that rely on -march=i486 ? 2015-05-07 10:02:31 i was wondering should we bump that to i586, or i686 later after 3.2-stable branching 2015-05-07 10:04:25 i think there was some old alix boards 2015-05-07 10:04:55 would be nice to bump it yes 2015-05-07 10:05:17 alix says i586 2015-05-07 10:05:36 i think there was some old board that used cyrix cpu or simliar 2015-05-07 10:05:39 oh 2015-05-07 10:05:46 that used 486 instructions at 200Mhz 2015-05-07 10:06:39 i think i remember those. i never used them, but looked at the boxes one time 2015-05-07 10:07:29 soekris 2015-05-07 10:07:49 yes i have one 2015-05-07 10:07:56 close to the garbage can 2015-05-07 10:10:07 i wonder if ppl who use those boxes still upgrade to the latest release. 2015-05-07 10:13:17 yes 2015-05-07 10:13:26 good question 2015-05-07 10:24:50 i would be ok to default to i586 2015-05-07 10:25:03 i686 might be little bit too early still since ALIX is i586 2015-05-07 10:25:28 agreed, im still using my alix. 2015-05-07 10:44:29 i have also an Alix.3D3 here was some info about -march http://www.twam.info/hardware/chost-i586-vs-i486-on-amd-geode-lx 2015-05-07 11:04:28 ok, after v3.2 branch we change to i586 2015-05-07 12:50:52 anyone who's got one of those soekris boards 2015-05-07 12:50:55 net4501 etc 2015-05-07 12:51:01 has been running it for like, 10 years 2015-05-07 12:51:12 they've gotten their money's worth out of it. time they moved on. 2015-05-07 15:59:09 vk1xt i have apu4c :) 2015-05-07 16:20:50 good evening 2015-05-07 16:21:03 i am having weird issues while simply doing "reboots" 2015-05-07 16:21:16 i type reboot, it shuts down everything nice and tidy and then doesnt restart 2015-05-07 16:21:33 last thing is either deactivating swap or lvm (if lvm is enabled) 2015-05-07 16:22:02 the funny thing is, even after i see openrc shutting down all services, i still have a prompt 2015-05-07 21:02:13 ncopa: is anyone on the distro sec list already from alpine? 2015-05-07 21:15:07 hi guys 2015-05-07 21:15:34 I'm trying to build an alpine ISO but I'm having some trouble 2015-05-07 21:16:09 [snip] 2015-05-07 21:16:11 ERROR: Failed to set ownership on lib/modules/3.18.13-0-grsec/kernel/security/keys/encrypted-keys/encrypted-keys.ko.apk-new: Operation not permitted 2015-05-07 21:16:11 ERROR: Failed to set ownership on lib/modules/3.18.13-0-grsec/kernel/kernel/configs.ko.apk-new: Operation not permitted 2015-05-07 21:16:11 ERROR: Failed to set ownership on usr/share/kernel/grsec/kernel.release.apk-new: Operation not permitted 2015-05-07 21:16:11 13 errors; 216 MiB in 13 packages 2015-05-07 21:16:11 Makefile:82: recipe for target '/home/sasoiliev/alpinedev/alpine-iso/isotmp.alpine-mini/stamp.modloop.kernel.grsec' failed 2015-05-07 21:16:12 make: *** [/home/sasoiliev/alpinedev/alpine-iso/isotmp.alpine-mini/stamp.modloop.kernel.grsec] Error 13 2015-05-07 21:16:44 this is the output of `make PROFILE=alpine-mini iso` 2015-05-07 21:18:15 I've followed the Developer Documentation wiki pretty closely (Upgrading to Edge, Setting up the build environment on HDD and How to make a custom ISO) 2015-05-07 21:19:36 <_mjones> sasoiliev: huh. I've never tried to do that. 2015-05-07 21:20:08 <_mjones> sasoiliev: I'm guessing a problem with the buildroot/chroot or something like that. 2015-05-07 21:20:40 <_mjones> was that running as root or not? Can you try the other way? 2015-05-07 21:21:39 _mjones: I'm running this as a normal user - the docs explicitly said to create one and the make iso command wasn't sudo'ed 2015-05-07 21:21:51 but sure, I'll try with root/sudo 2015-05-07 21:24:53 I have an apk question as well - how can I list all files installed by a certain package (dpkg -L) and also how can I get the package a file belongs to (dpkg -S)? 2015-05-07 21:27:23 ok, on the make iso issue - running it with sudo got rid of the Operation not permitted errors, but now it says I don't have a private key (as it's in home dir and it's looking for it in /root) 2015-05-07 23:36:16 hey 2015-05-08 00:08:07 What is the process for getting updated versions of packages into apk? 2015-05-08 00:08:17 Specficially I'm looking at py-twisted 2015-05-08 00:11:37 ncopa, can I just replicate something similiar to, http://git.alpinelinux.org/cgit/aports/commit/?id=ed3fd090e194e33c856ace790a2e255fa0a2e3f2 ? 2015-05-08 00:16:36 ohai cnf 2015-05-08 00:16:51 hey 2015-05-08 00:16:58 i think u just issue a pull request 2015-05-08 00:17:06 there's docs on it on the web site somewhere 2015-05-08 00:17:10 i asked the same question last week 2015-05-08 00:31:14 ACTION creates a Dockerfile for alpine package development. 2015-05-08 00:49:46 Hmm, I have main and edge testing enabled, but I cannot install emacs 2015-05-08 00:51:04 https://gist.github.com/dustinlacewell/ce557289ca57ef182a1d 2015-05-08 00:51:10 What does this mean? 2015-05-08 00:52:49 bitshark_: any idea? 2015-05-08 01:00:30 I don't understand how to prevent packages from masking each other when using multiple repositories 2015-05-08 01:12:50 apk add emacs@testing 2015-05-08 01:14:35 BitL0G1c: but then its dependencies become masked 2015-05-08 01:14:42 until I peg eveything explicitly all the way down 2015-05-08 01:15:24 BitL0G1c: if I say emacs@testing then I have a bunch of curses libs 'masked in edge' 2015-05-08 01:16:08 you may need to upgrade to edge 2015-05-08 01:17:23 BitL0G1c: what do you mean 'upgrade' ? 2015-05-08 01:17:34 oh 2015-05-08 01:17:54 set your "main" to edge 2015-05-08 01:18:29 Yeah I think that works thanks 2015-05-08 01:18:36 http://wiki.alpinelinux.org/wiki/Edge 2015-05-08 01:19:29 <__number5__> so you can mix-match different version of alpine apk repo at the same time? that's nice 2015-05-08 01:19:57 yes you can pin repos too 2015-05-08 01:20:07 BitL0G1c: hmm now that I've set edge to be my main/untagged repo, it can't find alpine-sdk 2015-05-08 01:20:25 I tagged the stable repo as @stable and tried alpine-sdk@stable but that didn't work 2015-05-08 01:21:19 here are my edge settings: 2015-05-08 01:21:31 Okay I got it 2015-05-08 01:22:02 http://hastebin.com/izoqukatuk.avrasm 2015-05-08 01:22:13 if you run edge you don't pin 2015-05-08 01:23:00 you can also pin http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management#Repository_pinning 2015-05-08 01:23:45 that works for running things like LXC from edge on a stable system 2015-05-08 01:23:57 BitL0G1c: using docker at all? 2015-05-08 01:24:14 I tried it but i prefer LXC 2015-05-08 01:24:41 LXC with openvswitch is just simpler for exposing services 2015-05-08 01:25:02 I can SSH into all my containers 2015-05-08 01:25:04 BitL0G1c: for cross-host container networking? 2015-05-08 01:25:24 BitL0G1c: I just start a bash process in the namespace 2015-05-08 01:25:59 docker is ok if you don't need persistence 2015-05-08 01:26:16 volumes? 2015-05-08 01:26:24 <__number5__> BitL0G1c: do you have recommended lxc docs/tutorial? 2015-05-08 01:26:52 yes looked at data volumes but not tried them - LXC containers will use less space 2015-05-08 01:27:02 yes hold on 2015-05-08 01:27:14 BitL0G1c: why are lxc containers going to use less space? 2015-05-08 01:28:11 apparently docker doesn't fully free up space when a container is destroyed 2015-05-08 01:28:30 BitL0G1c: what? 2015-05-08 01:28:31 someone was testing using one as a build environment 2015-05-08 01:29:07 BitL0G1c: I built the DockerHub automated build service, and it doesn't have a problem cleaning up space. 2015-05-08 01:29:34 Some unionfs storage backends have had bugs, if that's what you mean. 2015-05-08 01:29:43 I don't have much experience with docker - for LXC I started with http://blog.scottlowe.org/2014/01/23/automatically-connecting-lxc-to-open-vswitch/ 2015-05-08 01:29:47 yes probably 2015-05-08 01:29:58 Docker containers are not inherently bigger than any other container format 2015-05-08 01:30:54 BitL0G1c: thanks for the link! 2015-05-08 01:31:14 np - it's a nice solution for exposing services 2015-05-08 01:31:28 simpler than messing with docker's nat 2015-05-08 01:31:41 BitL0G1c: yeah we're working on integrating openvswitch into docker first class 2015-05-08 01:32:10 abstracted as "networks" that you can manipulate like images, containers and volumes 2015-05-08 01:32:28 Lots of players in the industry have opinions on how we should do it, so its taking a while 2015-05-08 01:32:46 yes i can imagine 2015-05-08 01:33:19 will you also be able tun run docker images unprivileged ? 2015-05-08 01:33:45 you already can 2015-05-08 01:33:52 Oh 2015-05-08 01:33:54 nice 2015-05-08 01:34:09 What do you mean by unprivileged? 2015-05-08 01:34:21 Ithe docker process runs as a user not root 2015-05-08 01:34:41 Ah not yet. Something about user namespaces. 2015-05-08 01:35:20 http://www.wickedawesometech.us/2014/11/the-container-world-part-5-advanced.html - I've got that working with centos7 & confirmed the LXC container was running as a user 2015-05-08 01:35:22 <__number5__> BitL0G1c: thanks for the link 2015-05-08 01:36:54 np - I also proposed a patch for LXC to fix the download template https://github.com/lxc/lxc/pull/511 2015-05-08 01:37:29 running lxc-create -n xxxxxx -t download is probably a better way to install other distro containers 2015-05-08 02:57:41 hi there, do you know which lib got ? 2015-05-08 03:03:19 When building my changes to the twisted package I get the following: https://gist.github.com/dustinlacewell/2bb887a3b05aefa5cea8 2015-05-08 03:03:23 Not sure what it means. 2015-05-08 03:03:55 missing repository tags? 2015-05-08 03:09:31 ACTION scratches his head 2015-05-08 03:09:42 google hasn't a clue 2015-05-08 03:14:07 I think it has something to do with how I'm tagging my repos? 2015-05-08 03:14:14 as in alpine package repos? 2015-05-08 03:19:03 oh you wernt kidding when you said you don't tag if you're using edge 2015-05-08 03:29:21 <__number5__> ldlework: maybe you should try `pip install twisted` :P 2015-05-08 03:29:25 <__number5__> ACTION runs 2015-05-08 03:29:50 __number5__: it requires a bunch of stuff on the image to compile 2015-05-08 03:29:56 I want my base image to not have all that stuff 2015-05-08 03:30:00 which means I need a package 2015-05-08 03:32:01 <__number5__> ldlework: fully understand what you are doing, I'm just joking 2015-05-08 03:32:11 __number5__: hehe 2015-05-08 03:32:24 btw 2015-05-08 03:32:31 abuild -rF will essentially never work 2015-05-08 03:32:35 since -F isn't passed down 2015-05-08 03:51:16 hello 2015-05-08 03:53:22 hello 2015-05-08 03:56:46 howto build own vanilla kernel from source? are there any patches needed? 2015-05-08 03:57:33 when I tried it last time, make bzImage failed at the end with some strange errors 2015-05-08 04:26:50 Xe: when you build a package with abuild -r, where does it actually go? 2015-05-08 04:29:44 Also, now that I have my patch file where do I send it? 2015-05-08 04:40:40 okay apparently I sent the patch 2015-05-08 04:44:07 hmm but it doesn't show up on the mailing list 2015-05-08 04:59:42 How do I actually use the distfile that abuild creates? 2015-05-08 05:07:16 blehhhhhhhh 2015-05-08 05:10:43 someone save me 2015-05-08 05:34:28 ncopa: are you around? 2015-05-08 06:03:10 morning 2015-05-08 06:03:16 ldlework: i'm here 2015-05-08 06:10:11 <_ikke_> morning 2015-05-08 06:11:04 ncopa: hello :) 2015-05-08 06:11:32 ncopa: I tried to update the py-twisted package to 15.1.0 but have failed 2015-05-08 06:12:15 I updated the APKBUILD and ran 'abuild -r' and it appears to successfully compile twisted's python extension and so on, but I get no apk, just a tarfile in /var/cache/distfiles 2015-05-08 06:13:10 ldlework: so no error message? 2015-05-08 06:13:18 Nope 2015-05-08 06:13:27 ncopa: I have a dockerfile if you're privy to that 2015-05-08 06:13:27 <_ikke_> ldlework: the apk file isn't put in the current dir 2015-05-08 06:13:46 what is REPODEST set to in /etc/abuild.conf? 2015-05-08 06:14:08 oh 2015-05-08 06:14:27 well what do you know 2015-05-08 06:15:32 thanks. What is the command to install an apk file directly? 2015-05-08 06:15:39 <_ikke_> just apk add 2015-05-08 06:15:54 i normally add /home/ncopa/packages/main to my /etc/apk/repositories 2015-05-08 06:16:18 apk add will add it sticky, so apk upgrade will not upgrade it 2015-05-08 06:16:32 <_ikke_> makes sense 2015-05-08 06:16:41 ncopa: I'm just build docker images so that should be ok 2015-05-08 06:17:11 <_ikke_> I have a seperate docker container where I build my packages that I add as repo to other containers 2015-05-08 06:17:25 That's cool. 2015-05-08 06:17:34 <_ikke_> Well, one build container and one repo container sharing a volume 2015-05-08 06:17:49 that is cool indeed 2015-05-08 06:18:35 <_ikke_> Saves time in building the containers 2015-05-08 06:19:31 ncopa: you should have a patch from me to update twisted to 15.1.0 I hope 2015-05-08 06:19:34 _ikke_: docker pull -it -v /var/run/docker.sock:/var/run/docker.sock dlacewell/console 2015-05-08 06:19:52 ^ if you wanna try what I'm working on (a curses docker cli) 2015-05-08 06:20:22 brb 2015-05-08 06:20:41 <_ikke_> ldlework: cool 2015-05-08 06:20:51 <_ikke_> I'll try later 2015-05-08 06:21:06 well, alright 2015-05-08 07:48:23 abuild.conf says "# CHOST=powerpc-alpine-linux-musl " : powerpc, something we don't know about? :) 2015-05-08 08:11:55 ncopa: i'm wondering whether or not we can just fix the problems with thunderbird https://tails.boum.org/security/claws_mail_leaks_plaintext_to_imap/index.en.html 2015-05-08 08:11:58 that worries me 2015-05-08 08:12:19 we seem to have an icon theme faenza-icon-theme-thunderbird-1.3.1-r4 but not actually thunderbird/icedove 2015-05-08 08:12:37 or is it that thunderbird doesn't compile against musl? 2015-05-08 08:13:02 lol fuck google 2015-05-08 08:13:13 every time i search for musl it wants to correct it to muslim lol 2015-05-08 08:18:34 http://permalink.gmane.org/gmane.linux.distributions.alpine.devel/367 2015-05-08 08:18:38 looks like someone tried to do it 2015-05-08 09:08:35 hmm looks like newsbeauter doesn't work Error relocating /usr/bin/newsbeuter: __cxa_throw_bad_array_new_length: symbol not found 2015-05-08 12:00:46 how would i go about installing AL on a running linux distro? 2015-05-08 12:02:06 actually, this is for a rasperrypi so it might be really simple. not thought about it 2015-05-08 12:04:56 i have a rpi here in front on me with AL installed. I have a remote rpi where i want to install AL. 2015-05-08 12:05:49 i only did similar via a rescue system 2015-05-08 12:06:10 perhaps configure AL on local rpi with network settings of remote rpi. lbu commit, then somehow copy something from local rpi to remote rpi? 2015-05-08 12:06:53 yeah at times i just dd'ed over a live system, but really need to be angry and not caring to 2015-05-08 12:07:11 remote rpi is debian, and i just want to replace deb with AL 2015-05-08 12:07:48 you don't happen to have lvm on the sd card i guess 2015-05-08 12:07:49 darkfader: what can go wrong? :) 2015-05-08 12:08:08 local or remote sd? 2015-05-08 12:08:11 remote 2015-05-08 12:09:03 ah, i think i have a good / better idea now 2015-05-08 12:09:11 one that is deterministic 2015-05-08 12:09:15 no lvm 2015-05-08 12:09:25 make a ramboot initrd on local raspi 2015-05-08 12:09:38 one that you can place on remote raspi 2015-05-08 12:09:44 to boot it to ram 2015-05-08 12:09:57 then you can replace /oldroot easily 2015-05-08 12:11:00 i have some isos downloading atm, i'll go have lunch 2015-05-08 12:11:22 like, downloading via cfengine. i don't want to be around when it starts going all wrong 2015-05-08 12:15:10 and i want to do this from the local raspi :) 2015-05-08 12:15:44 so setup-networking on local with settings or remote, then lbu-commit 2015-05-08 12:16:17 then setup-networking on local again, and change back to local settings. 2015-05-08 12:16:31 what do i need to send to the remote raspi? 2015-05-08 12:16:39 I'm sure it's doable 2015-05-08 12:20:08 fabled: you about? 2015-05-08 12:25:06 maybe it's simpler than that. untar armhf-rpi into the remote SD card /dev/mmcblk0p1 (currently mounted on /boot) 2015-05-08 12:26:10 copy the pre-configured apkvol.tar.gz to /boot 2015-05-08 12:26:14 reboot :) 2015-05-08 12:27:24 ouch, on the remote rasbpi /dev/mmcblk0p1 56M 19M 38M 33% /boot 2015-05-08 12:27:29 that's too small 2015-05-08 12:28:52 http://sprunge.us/QSMQ 2015-05-08 12:29:19 that is the remote rasbpi :( 2015-05-08 13:14:58 is there an opennhrpctl command I can use to force opennhrp to re-register with a nhrp server 2015-05-08 13:15:51 fabled: 2015-05-08 13:17:27 purge 2015-05-08 13:18:19 ok, so I purge the neighbour address and it will reregister? 2015-05-08 13:20:21 ok, cool, I see how that works 2015-05-08 13:20:22 thanks 2015-05-08 13:20:55 on openwrt the opennhrp-script for peer-up seems to get called very aggressively 2015-05-08 13:21:03 even if the gre tunnel isn't present and configured 2015-05-08 13:21:21 how do you handle detection of an interface to decide if you want to try to register? 2015-05-08 13:26:59 that's probably not handled the best way. the new quagga/nhrp code will handle it all better 2015-05-08 13:27:27 i'm probably few weeks away of posting it. registration works, and traffic to hub already works. 2015-05-08 13:29:54 Is there a preference for indendation in aports APKBUILDs? tabs vs spaces? 2015-05-08 13:34:34 fabled: I'll take any pre-release you've got, I'm busy writing a script that watches the log 2015-05-08 13:34:37 so I see a 2015-05-08 13:34:45 CHILD_SA tun813{2} established with SPIs 2015-05-08 13:34:54 and I'm going to fire off a purge on that interface 2015-05-08 14:12:28 ldlework: you poked? ^^; 2015-05-08 14:42:30 ScrumpyJack: i installed alpinelinux on my rp2 2015-05-08 14:42:37 had to use the 3.2 version but it works nice 2015-05-08 14:43:59 Orochimarufan: spaces probably 2015-05-08 14:44:09 only noobs use tabs 2015-05-08 14:44:11 It actually seems to be tabs 2015-05-08 14:44:17 oh that's really bad then 2015-05-08 14:44:31 I'm just going to keep it consistent per file 2015-05-08 14:44:57 tabulators weren't designed for indenting, and they aren't consistently the same size across different editors 2015-05-08 14:45:09 correct 2015-05-08 15:24:40 Woo, look at this: http://www.phoronix.com/scan.php?page=news_item&px=Musl-Libc-GCC-Support :O 2015-05-08 15:30:08 actually i think this is a benefit not a drawback, that tabs are rendered according to the taste of the consumer, not the producer. while they don't make a difference for the machine. 2015-05-08 15:30:58 this why i can have all tabs shown as 2 spaces, while other more verbose purists can stick with more traditional values. 2015-05-08 15:31:13 In the end, it's a matter of taste 2015-05-08 15:31:16 not only 2015-05-08 15:31:22 also we have very different output devices 2015-05-08 15:31:30 rendering should depend on the renderer 2015-05-08 15:31:34 I hate it because it screws up alignment when consumer settings != creator settings 2015-05-08 15:32:10 why is that a problem? why do you care even? 2015-05-08 15:32:30 ooh, you use ws not for indentation but for aligning stuff? 2015-05-08 15:32:38 in that case indeed, spaces are the way to go 2015-05-08 15:32:43 but we talked indentation. 2015-05-08 15:32:53 because I don't like my code looking funny when viewed on GH 2015-05-08 15:33:25 also, I usually use spaces for everything. Except when contributing to projects that use tabs. In that case, I have no problem adapting 2015-05-08 15:41:04 cnf: just saying hi :) 2015-05-08 15:41:15 ohai1 2015-05-08 15:55:53 ylVa333UKfLO: yeah but then things don't like up nicely 2015-05-08 15:55:56 like function signatures 2015-05-08 15:56:00 i think that's the main reason 2015-05-08 15:56:58 what does musl offer ulibc doesn't? 2015-05-08 15:57:04 also, https://www.python.org/dev/peps/pep-0008/#tabs-or-spaces 'nuff said ;) 2015-05-08 15:57:49 http://www.etalabs.net/compare_libcs.html 2015-05-08 16:02:59 i think its mostly support 2015-05-08 16:04:22 from that chart, i'd say musl seems like it has the overall lead 2015-05-08 16:04:36 yeah 2015-05-08 16:05:24 i wonder if alpinelinux can boot on a mac 2015-05-08 16:05:43 if it's an intel mac, there shouldn't be any problems :O 2015-05-08 16:12:01 true it is. 2015-05-08 16:12:25 i've been thinking of going over to a model where i use alpine linux for everything with a debian LXC container 2015-05-08 16:12:32 or two, like qubes 2015-05-08 16:12:47 i like qubes, but i don't like how it comes with KDE 2015-05-08 16:12:53 i3wm ftw 2015-05-08 16:12:53 I'm running a Arch and Alpine hybrid system^^ 2015-05-08 16:13:04 yeah archlinux is what i use at the moment 2015-05-08 16:13:14 or maybe i could do it the other way round 2015-05-08 16:13:20 run archlinux, and use alpinelinux vms 2015-05-08 16:13:25 or containers 2015-05-08 16:13:41 i also think i prefer kvm to xen (qubes uses xen) 2015-05-08 16:13:54 I have Arch as a minimal host (cause bleeding edge kernel) and run a couple of alpine containers on it for system services, plus one big arch container for my desktop 2015-05-08 16:14:12 yeah Orochimarufan that's what i was thinking of doing 2015-05-08 16:14:17 i think xen due to qubes and joanna got more security love than kvm 2015-05-08 16:14:34 possibly 2015-05-08 16:15:11 what i really like about arch is lbu 2015-05-08 16:15:22 You need to be prepared for a couple of pits tho if you're gonna containerize your desktop^^ 2015-05-08 16:15:26 basically allows me to have a live distribution without having to build my own 2015-05-08 16:15:39 couple of pits? 2015-05-08 16:16:05 Mostly related to getting devices into the container 2015-05-08 16:16:10 oh yeah 2015-05-08 16:16:21 You probably want X to behave and USB drives to automount ;) 2015-05-08 16:16:29 yeah 2015-05-08 16:16:41 well automounting doesn't bother me too much tbh i never use usb drives 2015-05-08 16:16:49 It took me a reimplementation of udev to get there -.- 2015-05-08 16:16:56 oh :P 2015-05-08 16:16:58 Definitely not fun 2015-05-08 16:17:31 well at the moment i run archlinux on my workstation (no containers) 2015-05-08 16:17:37 and i've had jack shit of macosx 2015-05-08 16:17:45 apple hardware is nice, as in configuration 2015-05-08 16:17:57 https://github.com/Orochimarufan/cdev 2015-05-08 16:17:59 it's really difficult to find a fucking laptop with 16GB of ram that does not have optimus 2015-05-08 16:18:16 and a QHD screen 2015-05-08 16:18:26 (It has even less documentation than Alpine, sorry :P) 2015-05-08 16:19:08 i am hoping with the 6th generation intels people start making laptops with 16GB of ram 2015-05-08 16:19:22 they possibly will i read something about the possibility of DDR4 ram too 2015-05-08 16:19:29 there are already laptops with 16G :O 2015-05-08 16:19:40 but you really don't need that much 2015-05-08 16:19:48 they are rare with a QHD screen 2015-05-08 16:19:56 and i wasn't going to buy a dell precision thats like 5K 2015-05-08 16:20:02 heh 2015-05-08 16:20:12 try find a laptop basically like the macbook 2015-05-08 16:20:17 You can upgrade the RAM in most laptops 2015-05-08 16:20:30 sillysausage - get an AMD hpdv6z 2015-05-08 16:20:36 (Most meaning pretty much all, minus apple :P) 2015-05-08 16:20:36 do they address more than 8GB? in the lenovos T4XX or carbon X? 2015-05-08 16:20:47 i have 16gig of ram 2015-05-08 16:20:53 i was thinking of the dell XPS13 too 2015-05-08 16:21:36 [offtopic] BitL0G1c: high five for using AMD :D [/offtopic] 2015-05-08 16:21:41 heh 2015-05-08 16:21:52 ;-) 2015-05-08 16:22:40 (though I find it increasingly hard to justify AMD in a laptop... There's no Intel going into any of my desktops tho :) 2015-05-08 16:25:44 anyway i have one of the 1st gen macbook retinas 2015-05-08 16:25:58 so i think everything works nicely with that in linux 2015-05-08 16:26:20 If I'd get a MBP, I'd catch me one of the non-retina ones 2015-05-08 16:26:24 it's the one with the webcam that is still on the usb bus, not the pcie bus 2015-05-08 16:26:36 the retina ones are just too f'ing expensive 2015-05-08 16:27:09 oh i really like the retina screen 2015-05-08 16:27:23 Well, if you got the cash 2015-05-08 16:27:24 i spend most of my days in terminals so 2800x1600 is nice 2015-05-08 16:27:39 (Also, I don't like how everything is glued now) 2015-05-08 16:28:29 yeah i rather don't like that 2015-05-08 16:28:37 the carbon x1 looks nice 2015-05-08 16:28:56 I've been unscrewing and fixing my current laptop since f'ing 2004 now 2015-05-08 16:29:08 It still works (sort of hehe) 2015-05-08 16:29:27 but only has 2015-05-08 16:29:29 14" WQHD+ (2560 x 1440) Touch Screen (270 nit) 2015-05-08 16:29:39 I want a 15.6" 2015-05-08 16:29:47 touchscreen or no 2015-05-08 16:30:00 avoiding a touchscreen is basically impossible 2015-05-08 16:30:06 because of microsoft, not that i would use it 2015-05-08 16:30:11 most useless feature to have in a laptop lol 2015-05-08 16:30:21 who wants to dirty their screen 2015-05-08 16:30:52 well, it's nice now and again 2015-05-08 16:31:20 so the xps 13" has 2015-05-08 16:31:23 13.3-inch FHD (1920 x 1080) infinity display 2015-05-08 16:31:31 13.3-inch UltraSharp™ QHD+ (3200 x 1800) infinity touch display 2015-05-08 16:31:53 but i imagine that's overkill for a 13" 2015-05-08 16:32:03 very 2015-05-08 16:32:04 my hpdv6z-7200 boots to an alpine desktop - not figured out how to setup alsa yet 2015-05-08 16:32:41 I just dropped in pulse and every thing worked as expected 2015-05-08 16:32:47 (with all the issues too hehe) 2015-05-08 16:34:09 on debian pulse is a bit of a pain - am thinking of trying manjaro as it can use openrc (now that jessie is systemd only) 2015-05-08 16:34:29 you can use pulse on alpine 2015-05-08 16:34:51 be nice to get alsa working on it's own 2015-05-08 16:35:07 https://www.phoronix.com/scan.php?page=news_item&px=Linux-MacBook-GPU-Switching 2015-05-08 16:35:08 I never really got into ALSA 2015-05-08 16:35:10 i read that recently 2015-05-08 16:35:28 cause I use a couple of PA's more advanced features 2015-05-08 16:35:33 like network-transparency 2015-05-08 16:35:41 isn't manjaro that silly distro with people who get all butthurt on the forums 2015-05-08 16:35:57 i heard bad things about the dev and the ssl certificate some time ago too 2015-05-08 16:36:14 yeah the ssl cert was broken 2015-05-08 16:36:27 but other than that, it doesn't seem bad 2015-05-08 16:36:39 i have heard good things about antergos though 2015-05-08 16:37:09 i think this is it 2015-05-08 16:37:15 https://www.reddit.com/r/linux/comments/3241mj/if_you_want_some_good_entertainment_check_out_the/ 2015-05-08 16:37:16 made me lol 2015-05-08 16:37:28 You would expect the Manjaro users to be kind of upset and indeed they are - but not about this issue but about the people who criticize it. They seem to be completely paranoid and live in a world where all the other bad Linux users want to harm their distro out of jealousy. 2015-05-08 16:37:42 which init on antergos ? 2015-05-08 16:38:20 probably systemd 2015-05-08 16:38:23 as it's arch i think 2015-05-08 16:38:33 probably 2015-05-08 16:38:47 i don't use either i use vanilla arch 2015-05-08 16:38:55 same here 2015-05-08 16:38:57 mostly because i find installers don't install the way i want 2015-05-08 16:39:16 eg dmcrypt+luks with a gpt partition table and the legacy bios setup 2015-05-08 16:39:19 they all seem to do mbr 2015-05-08 16:40:07 i wrote a script to setup gpt / luks on alpine 2015-05-08 16:40:55 does the rp version of alpinelinux use grsec? 2015-05-08 16:41:04 rp? 2015-05-08 16:41:08 raspberry pi 2015-05-08 16:42:17 i didn't see any reference to it when i did uname -a 2015-05-08 16:45:02 anyway, brb dinner 2015-05-08 16:50:53 Why doesn't building an APKBUILD file pull in its dependencies? 2015-05-08 16:52:35 I have makedepends="python-dev" but it certainly doesn't pull python-dev before building the apk 2015-05-08 16:52:46 python-dev (missing): 2015-05-08 16:52:48 required by: world[python-dev] 2015-05-08 16:56:11 run 'apk update && apk upgrade' 2015-05-08 16:56:35 I had unsolvable dependencies when doing a patch earlier today 2015-05-08 16:57:14 musl32 [~/aports]$ apk search python|grep dev 2015-05-08 16:57:14 python-dev-2.7.9-r4 2015-05-08 16:57:14 python3-dev-3.4.3-r2 2015-05-08 16:57:19 I do.. 2015-05-08 16:57:44 python-dev is in my repos 2015-05-08 16:58:05 abuild -r, just doesn't seem to bother to get the packages in makedepends 2015-05-08 16:58:21 do abuild deps (iirc) 2015-05-08 16:58:27 perhaps you have it installed 2015-05-08 16:58:36 cat /etc/apk/w* 2015-05-08 16:59:53 if the host has a pkg installed - it will not get pulled into the fakeroot env by abuild 2015-05-08 17:00:52 https://gist.github.com/dustinlacewell/f0df4ccdf23e46ee422d 2015-05-08 17:02:57 no idea how to fix it 2015-05-08 17:03:06 me neither 2015-05-08 17:03:07 I'm having to manually install packages before running abuild 2015-05-08 17:03:26 you should probably see where those apk warnings come from 2015-05-08 17:04:13 ah 2015-05-08 17:04:30 that worked 2015-05-08 17:06:53 Nice, now I have pretty good alpine-dev dockerfile 2015-05-08 17:07:14 I just use a good old lxc container 2015-05-08 17:09:01 Ok 2015-05-08 17:13:54 When you do "apk update" where are the package lists stored? 2015-05-08 17:15:10 no clue.. 2015-05-08 17:16:12 I have some in /var/cache/apk (aka /etc/apk/cache) but I doubt those are the ones you want 2015-05-08 17:16:59 abuild really needs a command to invoke apk update 2015-05-08 17:17:01 or flag 2015-05-08 17:17:08 why 2015-05-08 17:17:21 Because convenience? 2015-05-08 17:17:29 So I can start up a container, and just build it with a single command 2015-05-08 17:17:40 Knowing that I wont have any problems with deps 2015-05-08 17:18:03 It's the docket hype again :O 2015-05-08 17:18:09 I must be getting old.. 2015-05-08 17:18:35 Old people can like standard interfaces too 2015-05-08 17:19:00 (I'm not really old) 2015-05-08 17:19:22 I don't know.. I don't see myself using docker 2015-05-08 17:19:27 use LXC for the build environment 2015-05-08 17:19:32 Well that doesn't mean you have to dig on everyone who does 2015-05-08 17:19:55 BitL0G1c: No, thanks. 2015-05-08 17:19:57 there is no advantage to using docker to build in as abuild cleans up after itself 2015-05-08 17:20:13 BitL0G1c: That's not really the point 2015-05-08 17:20:19 I'm actually quite young and I still feel like the "good old days" were better than what we have now -.- 2015-05-08 17:20:45 BitL0G1c: does it also clean up dependencies? 2015-05-08 17:20:46 Orochimarufan: oh yeah, I loved deploying a variety of different arbitrary software and writing scripts for each one rather than having a single deployable asset type. 2015-05-08 17:20:56 yes it does 2015-05-08 17:21:02 hmm 2015-05-08 17:21:22 ldlework: well, I never had to deploy stuff 2015-05-08 17:21:36 Orochimarufan: maybe that's why you feel unserved by a tool you don't need 2015-05-08 17:21:46 I was mostly referring to the desktop situation with that particulary remark 2015-05-08 17:23:15 There are going to be more and more docker users in here as alpine becomes more popular as a container foundation 2015-05-08 17:24:10 I hope they're welcomed, because alpine is pretty awesome 2015-05-08 17:24:42 FYI: I'm in no way representing anyone here :P I'm just a random guy on the internet^^ 2015-05-08 17:25:08 You're 1/7th of my interaction with the alpine community so far! :D 2015-05-08 17:26:49 > "good old days" 2015-05-08 17:26:54 i get like that when i think about social media 2015-05-08 17:26:57 if you are going to use docker as a build environment you will need http://container42.com/2014/11/18/data-only-container-madness/ 2015-05-08 17:27:09 and how the internet is used by morons these days 2015-05-08 17:27:27 which is part of the reason i refuse to use any such 'sites' 2015-05-08 17:27:28 BitL0G1c: why? 2015-05-08 17:27:39 i'm also young. 2015-05-08 17:28:04 BitL0G1c: I have a host checkout of aports, I volume mount that in, I have a small sh script that lets me name a package, when I start the container it builds the package and spits out an apk 2015-05-08 17:28:11 It works, exactly as intended. 2015-05-08 17:28:56 ok 2015-05-08 17:30:32 maybe keep a copy of /var/cache/apk too to fix the other problems 2015-05-08 17:31:15 the other problem was simply not running upgrade (was only running update) 2015-05-08 17:31:23 it works fine now 2015-05-08 17:37:39 Argh I can't find that awesome caption about memories.. 2015-05-08 17:53:35 https://drive.google.com/file/d/0B5I8UGgSqvvucWFPT3ZuVlUwcHc/view?usp=sharing that's what I was looking for :) 2015-05-08 18:06:32 Should I just be able to do "apk add some.apk" ? 2015-05-08 18:08:41 yes 2015-05-08 18:08:59 I just set up a custom local repo^^ 2015-05-08 18:09:16 add your public key to /etc/apk/keys 2015-05-08 18:13:32 BitL0G1c: aw so I can't just give people an apk and have them install it? 2015-05-08 18:13:48 just give them the key too 2015-05-08 18:13:52 :( 2015-05-08 18:14:06 hmm 2015-05-08 18:14:25 i think there is a switch to allow an unsafe install 2015-05-08 18:14:43 Oh I didn't see one 2015-05-08 18:14:44 --allow-untrusted 2015-05-08 18:14:49 beautiful 2015-05-08 18:14:50 apk add --allow-untrusted 2015-05-08 18:15:47 thank you 2015-05-08 18:16:06 np 2015-05-08 18:16:19 if you're gonna ship, I'd still ship the key alongside, personally 2015-05-08 18:16:37 I only need to do this until my patch to update the official repo lands to update Twisted to 15.1.0 2015-05-08 18:16:43 Otherwise I wouldn't do this at all 2015-05-08 18:16:49 The tool is only being used interally right now 2015-05-08 18:16:56 nice 2015-05-08 18:19:55 I wonder if alpine is small enough that you could make an inforgraphic explaining what a base install consists of 2015-05-08 18:20:06 probably 2015-05-08 18:20:18 there's pretty much only busybox, musl and linux 2015-05-08 18:20:30 and apk of course 2015-05-08 18:20:49 Its the kind of distro that is clean enough that it kind of makes me want to sit down and learn the linux kernel 2015-05-08 19:18:45 i don't even think the base alpine package depends on Linux... another reason it makes a great base for a container image :) 2015-05-08 19:20:57 no base should depend on linux, otherwise its broken in my book 2015-05-08 19:21:05 containers don't need a kernel 2015-05-08 19:23:31 an lxc container is 5 meg 2015-05-08 19:24:07 correct 2015-05-08 19:24:45 (not to mention that it's very lightweight to run too, thanks to openrc) 2015-05-08 19:42:42 BitL0G1c you have some real example of docker on alpinelinux and data only container? 2015-05-08 19:45:41 crow: there's plenty of alpine based docker images 2015-05-08 19:45:53 and data-containers are entirely unrelated to alpine 2015-05-08 19:54:05 crow - I've not really used docker much - I was going to use it as a build env for debian - i was going to use these instructions http://container42.com/2014/11/18/data-only-container-madness/ for a data container 2015-05-08 20:10:24 BitL0G1c i would like to use it also for build env, currently have vm 2015-05-08 20:10:57 crow: I just built a docker-based build env 2015-05-08 20:11:30 crow: but I haven't packaged it up and pushed it to DockerHub yet 2015-05-08 20:11:37 if you can give me a few hours I can make it available to you 2015-05-08 20:11:49 crow: by build env, you mean "for alpine package development" right? 2015-05-08 20:12:20 ldlework yea for test mystuff and such. 2015-05-08 20:12:46 ldlework if you pust it to dockerhub i could try it this weekend 2015-05-08 20:18:59 crow: I'll ping you if you're around 2015-05-08 20:26:10 ldlework ok great thank you 2015-05-08 22:40:25 hey anyone around 2015-05-09 00:28:33 Hello everyone, I just got my raspberry and wonder if anyone has any idea how to install alpine at 2015-05-09 01:46:34 *sigh* i lost track of the person who was working on chromium package for alpine 2015-05-09 01:57:07 hmm found the latest draft here: http://sprunge.us/KaNA 2015-05-09 14:09:29 so, did i understand last days' conversations correctly that i could actually reinstall my raspi1/2 with alpine now and *use* them the way i already do? 2015-05-09 14:25:57 dalias: funny, i was looking at the same thing yesterday. 2015-05-09 14:31:26 dalias: i think this is the corresponding patch http://sprunge.us/cGcL 2015-05-09 16:57:12 clandmeter, anything doing #ifdef _GNU_SOURCE is a bug 2015-05-09 16:57:35 _GNU_SOURCE is defined by the application to request stuff from the implementation; testing #ifdef _GNU_SOURCE in application code is nonsense 2015-05-09 16:59:31 thanks for digging it up tho 2015-05-09 17:45:22 dalias: i guess he just wanted it to work locally and didnt think it through 2015-05-09 17:45:52 dalias: im adding some deps to testing which seems to be needed now for chromium 2015-05-09 17:46:54 I wonder if chromium team are eager to accept patches for musl 2015-05-09 18:35:42 clandmeter, i think they would be ok with it 2015-05-09 21:20:12 clandmeter: did you remove my small refactor to the APKBUILD on py-twisted because that file is automatically generated, etc? 2015-05-09 22:36:20 glurg, no py-cffi 2015-05-10 16:29:17 how to find the $OSTYPE in sh? 2015-05-10 16:34:23 uname -o 2015-05-10 20:08:50 hi, I get an error ": temporary error (try again later) when I run apk update 2015-05-10 20:09:03 tried different mirrors, same result 2015-05-10 20:09:13 anyone knows whats up? 2015-05-10 20:09:48 no dns? 2015-05-10 20:10:22 how do I check that? 2015-05-10 20:13:28 can you ping www.cnn.com 2015-05-10 20:13:47 nope, returns bad address 2015-05-10 20:13:54 see 2015-05-10 20:13:59 no dns 2015-05-10 20:14:32 but ping 8.8.8.8 also returns an error: sendto: network unreachable 2015-05-10 20:14:49 that might be an explanation why you have no dns :P 2015-05-10 20:15:00 haha, yeah 2015-05-10 20:15:08 I'm using QEMU. 2015-05-10 20:15:28 qemu fux that up 2015-05-10 20:15:30 yes 2015-05-10 20:15:34 set /etc/resolv.conf 2015-05-10 20:15:38 it worked before I rebooted :( 2015-05-10 20:16:56 you need to override the nameserver in the dhcp response from qemu 2015-05-10 20:17:06 to what do I need to set it? 2015-05-10 20:17:07 if you're using the usermode networking stack 2015-05-10 20:17:37 whatever your nameserver is. however, stuff only bound to on your host 127.0.0.1 will obviously be confusing 2015-05-10 20:17:37 networking worked fine before I rebooted 2015-05-10 20:18:07 hmm. 2015-05-10 20:18:14 i'm talking rubish 2015-05-10 20:18:25 but I can't ping 8.8.8.8 either 2015-05-10 20:18:27 you need to fix the net, not the resolv.conf 2015-05-10 20:18:35 yeah, thats what I think too 2015-05-10 20:26:47 what should I check if networking doesnt work? 2015-05-10 20:38:00 yumbox - for a new install 'setup-interfaces && rc-service networking start' 2015-05-10 20:38:46 (& rc-update add networking) 2015-05-10 20:43:54 BitL0G1c: thanks 2015-05-10 20:44:03 I think I didnt have the service running 2015-05-10 20:44:15 rc-service networking start fixes dns 2015-05-10 20:44:23 but I still can't ping 2015-05-10 20:44:54 however, I can run apk update and apk upgrade fine. hmm weird. 2015-05-10 20:46:14 BitL0G1c: thanks! 2015-05-10 20:49:57 ;-) np 2015-05-10 20:50:45 for the ping check your firewall - you may have set a default deny on iptables 2015-05-10 20:50:48 a different question: do the services start concurently by default? 2015-05-10 20:51:15 i think openrc tries to set off things in parallel 2015-05-10 20:51:21 where possible 2015-05-10 20:51:58 https://wiki.gentoo.org/wiki/OpenRC 2015-05-10 20:52:38 ah, okay. 2015-05-10 20:52:47 well, it boots very fast, so its not really an issue anyway 2015-05-10 20:53:11 https://wiki.gentoo.org/wiki/Comparison_of_init_systems 2015-05-10 20:57:13 BitL0G1c: cool, thanks. 2015-05-10 21:41:48 When I enable PARALLEL in /etc/rc.conf, hwclock waits on modules, while modules waits on hwclock 2015-05-10 21:42:00 this increases boot time to 60 seconds 2015-05-10 21:42:08 (because of timeout) 2015-05-10 21:44:51 perhaps make one depend on the other so one loads first 2015-05-10 22:00:35 I added rc_need="!hwclock" to /etc/conf.d/modules 2015-05-10 22:00:40 but it didnt work 2015-05-10 22:01:48 I will ask in #linux 2015-05-10 22:49:39 yumbox - see an example depends http://hastebin.com/utoxugetoc.vhdl 2015-05-10 22:50:26 you need to set the depends in initd not confd 2015-05-10 22:50:36 I tried both. 2015-05-10 22:51:04 ok 2015-05-10 22:51:37 i did it wrong probably 2015-05-10 22:52:54 hwclock has: before * 2015-05-10 22:54:03 and modules has: before hwclock 2015-05-10 22:54:55 I tried editing hwclock to: 2015-05-10 22:54:57 before * 2015-05-10 22:54:59 after modules 2015-05-10 22:55:03 but that didnt work 2015-05-10 22:55:31 even non parallel it is fast 2015-05-10 22:56:12 it is. I wanna make it concurent though :P 2015-05-10 22:56:21 :) 2015-05-10 22:56:25 (parallel isn't the right word tbh) 2015-05-10 22:56:48 im surprised openRC uses that word in the program 2015-05-10 22:57:35 but it doesnt really matter, i guess 2015-05-11 01:40:49 good morning 2015-05-11 07:29:16 hello 2015-05-11 07:31:55 i've just install alpine on raspberry pi (model B) but i must use lbu commit if i make any change. is there another solution to apply automaticaly all change on sd card ? 2015-05-11 07:33:09 Chipster1, we consider that more feasible in most cases; as then SD card is read-only mounted, and there's no danger of FS corruption 2015-05-11 07:33:28 alternative is to partition SD, and install rootfs to the second partition 2015-05-11 07:33:30 ok 2015-05-11 07:33:33 like most other distributions do 2015-05-11 07:33:41 ok, i understand 2015-05-11 07:34:10 thx for your help fabled 2015-05-11 07:45:40 fabled, how to do because if I reboot, I lose all the data in RAM. 2015-05-11 07:45:49 Before each reboot I have to make a lbu commit or it is dangerous to proceed like this? 2015-05-11 07:50:40 in principle you should commmit after each batch of changes. you could however add it to cron or run it before a reboot/poweroff 2015-05-11 08:37:12 fabled: morning. you about? i have a breif raspi related question 2015-05-11 08:38:04 i'm off for lunch soon 2015-05-11 08:38:08 but go ahead 2015-05-11 08:38:15 thx clandmeter 2015-05-11 08:46:52 Hi. I'm investigating alpine as a foundation for my docker containers. I need a python 3 build environment set up, but it seems pip is only available for python 2. How do I fix that (or where do I look for further information) 2015-05-11 09:51:44 fabled: i'm trying to replace debian on a remote raspi with alpine linux. all i have is ssh access 2015-05-11 09:56:14 i'm wandering what your thoughts on that are 2015-05-11 10:07:39 i have a local raspi with alpine linux installed and configured as i want the remote to look, and all the data stored in a apkvol 2015-05-11 10:08:54 my main problem is that my remote SD card is partitioned like this 2015-05-11 10:09:53 http://sprunge.us/KSFO 2015-05-11 10:10:56 ScrumpyJack: what's the issue? 2015-05-11 10:11:30 the partition table 2015-05-11 10:11:44 why? 2015-05-11 10:12:29 not enough space on the first partition to put alpine linux 2015-05-11 10:13:26 the first partition is 56M 2015-05-11 10:13:38 well, you should be able to fit a kernel and initrd in there 2015-05-11 10:13:58 then add the ovl_dev=mmcblk0p2 to cmdline 2015-05-11 10:14:22 what about modloop? 2015-05-11 10:14:51 i suppose you need that, too 2015-05-11 10:15:35 modloop + vmlinuz + initramfs > 56M :( 2015-05-11 10:15:40 :/ 2015-05-11 10:52:05 sdlarsen - pip is included with python 3.4+ 2015-05-11 11:06:35 BitL0G1c: Thanks. Now onto 'Broken toolchain: cannot link a simple C program'... 2015-05-11 11:08:04 I've just sent in a patch to update pip to 6.1.1 2015-05-11 11:10:49 It seems I'm missing basic C libraries - libc libm etc. So i need gcc + what package to get that? 2015-05-11 11:11:07 install alpine-sdk 2015-05-11 11:11:47 see http://wiki.alpinelinux.org/wiki/Developer_Documentation 2015-05-11 11:11:57 BitL0G1c: Thanks, will do 2015-05-11 11:13:07 an Alpine LXC container makes a good build environment - if using docker remember you need to setup data containers if you need persistence 2015-05-11 11:13:58 BitL0G1c: It's the environment I'm after. I'm already familiar with docker and using it in production. But thanks anyway. I'm new to alpine though ;) 2015-05-11 11:15:35 ;-) ok 2015-05-11 11:16:20 if you can get docker guests running unprivileged - please let me know 2015-05-11 11:23:03 BitL0G1c: I'll let you know. Hmm. Seem like I'm missing psycopg2 and the package database link on the wiki is pointing to something that is down atm. 2015-05-11 11:24:50 Fortunately, pkgs.alpinelinux.org/packages works :) 2015-05-11 11:33:59 'apk search copg' 2015-05-11 11:34:46 in alpine python packges are prefixed with py- & for perl prefixed with perl- 2015-05-11 11:35:12 'apk add py-psycopg2' 2015-05-11 11:35:59 http://wiki.alpinelinux.org/wiki/Package_policies 2015-05-11 12:17:50 the af_packet module wants to load from /etc/modules but doesn't exists in modloop on my raspi 2015-05-11 12:18:00 what added the ap_packet module in wonder? 2015-05-11 16:20:42 hey folks, I am having an rather odd issue. I had this issue few years when I installed alpine first time but seems to be present. I am seeing a performance drop in alpine linux as for other distros. For example, php: phpbench in nearly all distros scorses a 2.5 seconds, alpine linux scores like 3.6 seconds. Same 5.6 version and this is just an example I have performed many tests and I see a performance drop 2015-05-11 16:20:54 am I missing something here? 2015-05-11 16:21:46 misconfiguring something? perhaps blaming the kernel ? 2015-05-11 16:26:00 all other distros use glibc, alpine uses musl, phpbench probably uses lot of regex (musl implementation is slow but bounded complexity, glibc implementation is fast but worst-case exponential), or string functions (memcpy is highly optimized in glibc), or some other bits you have to perf test yourself 2015-05-11 16:26:34 I see 2015-05-11 16:26:43 to be honest I would like to build a simple hypervisor 2015-05-11 16:26:46 i can also think about malloc being slower on musl, but really you have to measure yourself 2015-05-11 16:26:52 so I need to check qemu kvm performance 2015-05-11 16:27:17 not much I can do about it can I ? 2015-05-11 16:27:49 dont use php? 2015-05-11 16:28:01 i thought there were faster implementations than the official one.. 2015-05-11 16:28:18 I will siplu use php in kvm guest 2015-05-11 16:28:39 but if the performance is slowed by qemu as well and same rule applies than ... 2015-05-11 16:29:00 if you can identify the bottleneck then it may be possible to fix 2015-05-11 16:29:25 (phpbench is most likely not a usual workload your site would generate) 2015-05-11 16:29:49 I know but's it provides a very simple benchmark 2015-05-11 16:29:59 cpu mostly 2015-05-11 16:30:55 so on the exact same server , alpine is ~30% slower. empty server, clean install. 2015-05-11 16:31:40 I don't have access right now to the machine but will do later, perhaps is just php acting up. Altough it's php on other distros. kvm might work fine 2015-05-11 16:31:40 most likely a difference in musl vs glibc.. might be easy to fix if you report the bottleneck.. 2015-05-11 16:31:57 I will post a report 2015-05-11 16:33:00 could qemu have the same issue, technically? 2015-05-11 16:34:32 qemu running on alpine will use musl too, but qemu does not use libc much: if your workload is cpu bound and you use kvm then you should not see slow-down 2015-05-11 16:36:22 that would be great for me as I am planning to build a very low weight hypervisor, kvm + libvirt and that is all 2015-05-11 16:40:22 although the whole distro is based on musl, is glib at least an option ? 2015-05-11 16:41:26 you mean glibc? 2015-05-11 16:41:33 (glib is something else) 2015-05-11 16:42:33 the libc is linked into every binary with c code: you would need 2 sets of packages for everything, so distros that do binary packages only support one libc in practice 2015-05-11 16:43:49 I see 2015-05-11 16:44:02 the php issue is worth reporting anyway 2015-05-11 16:44:13 as said, I have seen it a while ago 2015-05-11 17:32:40 nice, at $9, I want half-dozen of them, https://www.kickstarter.com/projects/1598272670/chip-the-worlds-first-9-computer 2015-05-11 19:32:58 Samba AD seems to be completely broken on Alpine 2015-05-11 19:38:42 hello, how come when i do apk update/upgrade I don't get the latest owncloud version? 2015-05-11 19:38:54 according to this we are on 8 already but I'm still on 7 2015-05-11 19:38:54 http://pkgs.alpinelinux.org/package/main/x86/owncloud 2015-05-11 19:44:47 hi robin666: owncloud is 7 in the 3.1 repositories; 8 in the 3.2 repositories 2015-05-11 19:45:06 but 3.2 is still in "release candidate" mode; lots of stuff still being fixed 2015-05-11 19:45:21 does that mean I will need to reinstall owncloud, or can I just upcrade when 3.2 is finished 2015-05-11 19:46:15 I don't know. I don't use owncloud 2015-05-11 19:50:13 robin666 - check the wiki for repo pinning & 'apk add owncloud@edge' 2015-05-11 19:50:35 maybe run that with --simulate to see what other packages it pulls in 2015-05-11 19:51:15 if it doesn't have too many dependencies running owncloud from edge on a stable system may be ok 2015-05-11 19:52:43 http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management#Repository_pinning 2015-05-11 19:54:33 it should be ok - it only really has php as a depends http://hastebin.com/puxicefoso.coffee 2015-05-11 19:54:42 hi there, what is the vanilla edition ? 2015-05-11 19:55:12 there is a linux-vanilla kernel 2015-05-11 19:55:26 & a vanilla iso 2015-05-11 19:55:44 aaa so same as mini but without grsec 2015-05-11 19:57:02 same as main but with linux-vanilla i think 2015-05-11 19:57:22 it has 80 megs so don't think it's main 2015-05-11 19:57:31 same as mini 2015-05-11 19:58:15 exactly 2015-05-11 20:01:26 hi how i can change the keyboard ... i ´ve run the alpinelinux setup and it doesnt change ... thanks in advance 2015-05-11 20:12:48 question for custom partition is there a special procedure? 2015-05-11 20:12:59 currently only asked me for sys or data and then partitioned itself 2015-05-11 20:16:11 which is not even close to what I want 2015-05-11 20:42:52 Roland-: there's some option you can give to let it do manually i think; there's nothing where you can enter and define a server-like layout 2015-05-11 20:43:20 but the idea is either to add a script for your needs, or idk 2015-05-11 20:43:32 i rebuilt it to be the way i liked it 2015-05-11 20:45:23 will try 2015-05-11 20:46:18 rc-update enable lvm boot (or lvm2? don't know offhand and make sure the right modules make it into the in initrd 2015-05-11 20:48:25 hello I have spoken to somebody earlier about the php slow thing? 2015-05-11 20:51:07 for a custom partition scheme see http://wiki.alpinelinux.org/wiki/LVM_on_LUKS 2015-05-11 20:51:51 I've also made some scripts to make it a lot easier linked at the top of the page 2015-05-11 20:52:37 will work for lvm / custom partitions with or without encryption 2015-05-11 20:53:05 setup-disk -m sys /mnt basically will copy alpine there right? 2015-05-11 20:53:22 yes 2015-05-11 20:53:42 still many things to setup by hand 2015-05-11 20:53:52 considering there is no ssh daemon in arch install 2015-05-11 20:53:54 use the scripts 2015-05-11 20:53:56 so I can ssh later 2015-05-11 20:54:10 i can setup a system in 5 minutes 2015-05-11 20:54:24 yes, because you made it. 2015-05-11 20:54:26 https://asciinema.org/a/18354 2015-05-11 20:54:35 no - because it's simple 2015-05-11 20:54:52 ACTION watches 2015-05-11 20:54:58 https://asciinema.org/a/18130 2015-05-11 20:55:27 I do not have lvm and do not plan to use lvm 2015-05-11 20:55:37 don't need to 2015-05-11 20:56:03 setup-lvm is not what name says? 2015-05-11 20:56:06 partition how you want lvm - no lvm - encryption or not - whatever 2015-05-11 20:57:50 yes but it's not just for lvm 2015-05-11 20:58:00 ok will test now 2015-05-11 20:58:29 see this one too https://asciinema.org/a/18354 2015-05-11 21:06:11 ok there is some issue here 2015-05-11 21:06:25 I just created a partition with cfdisk but not listed in /dev 2015-05-11 21:06:27 as sda1 2015-05-11 21:07:31 perhaps you are using virtio & it's /dev/vda1 2015-05-11 21:07:48 no no it's a raid controller 2015-05-11 21:07:54 I can see sda, I just partitioned it 2015-05-11 21:08:29 i've not tested on raid - so yes prob an issue 2015-05-11 21:08:40 hardware raid, sda 2015-05-11 21:09:16 fdisk sees it, working fine, also created one partition, worked fine, deleted partition, created another one, not showing up 2015-05-11 21:10:22 ok restarted mdev 2015-05-11 21:11:46 and now it's up 2015-05-11 21:13:22 maybe install udev - what you install in the iso session gets installed into the new install 2015-05-11 21:13:57 i normally install ssh 2015-05-11 21:14:14 before starting the installation 2015-05-11 21:16:07 cannot login as root via ssh 2015-05-11 21:16:28 although changed the root password 2015-05-11 21:16:41 says password incorrect, edited sshd config to allow root, same 2015-05-11 21:20:35 maybe the password is blank to start 2015-05-11 21:23:34 blank pass via ssh uhm 2015-05-11 21:23:34 looking at a script i made for ssh - i chroot into the new install to add a ssh user before rebooting 2015-05-11 21:23:35 ok 2015-05-11 21:24:29 http://hastebin.com/wucikaxivi.bash 2015-05-11 21:25:59 it also changes the root password 2015-05-11 21:32:30 ACTION on it 2015-05-11 21:36:38 the libraries for this small system have more than 250 megs 2015-05-11 21:36:54 statically ? 2015-05-11 21:40:43 most are shared - there is a static busybox & apk-tools for major system upgrades (such as uclibc to musl) 2015-05-11 21:52:02 ok finished installing 2015-05-11 21:54:22 success! 2015-05-11 22:02:25 ah ipmitool is missing completely 2015-05-11 22:14:22 I hope I can build a complex network, bridge on top of bond 2015-05-11 22:15:00 documentation is very ambiguous 2015-05-11 22:15:07 but looks like debian 2015-05-11 22:15:56 look at openvswitch & libteam 2015-05-11 22:20:53 yes well not working 2015-05-11 22:21:01 at least bridge over bond 2015-05-11 22:27:01 ah working now 2015-05-11 22:31:27 question: who keep creating my shm partition ? 2015-05-12 04:58:38 hey guys is it possible to enable btrfs (so I can use the mkfs.btrfs) during install? 2015-05-12 07:19:50 DeeEff well you can prepare disk manually and just then do setup-disk -m /mnt 2015-05-12 07:20:38 DeeEff check this wiki http://wiki.alpinelinux.org/wiki/Installing_on_GPT_LVM from there you can use part you need yourself 2015-05-12 11:52:26 hello 2015-05-12 11:54:12 i search which file was modified with setup-keymap. Have someone an idea ? 2015-05-12 11:57:46 s/Have/Has/ 2015-05-12 12:01:13 ChipsterOne: i think it modifies /etc/conf.d/keymaps 2015-05-12 12:01:32 and /etc/keymap/* 2015-05-12 12:05:55 in /etc/keymap there is only a .gz 2015-05-12 12:06:14 but i go to see conf.d/keymaps 2015-05-12 12:06:22 thx for your help 2015-05-12 12:07:19 by default we use busybox format of the keymaps 2015-05-12 12:07:36 so we dont need keep all keymaps in memory all times 2015-05-12 12:07:43 eg the entire keymap database 2015-05-12 12:07:45 i don't know busybox 2015-05-12 12:07:50 ok 2015-05-12 12:07:57 i understand 2015-05-12 12:08:07 i think you can install kbd and use the "normal" keymaps 2015-05-12 12:08:10 if you need 2015-05-12 12:08:37 i search keymap since i try to find an issue 2015-05-12 12:09:32 i make alias in /etc/profile since i must do "lbu ci" before reboot or halt 2015-05-12 12:09:52 otherwise i lost all changes 2015-05-12 12:10:45 yup 2015-05-12 12:10:55 unles you install it on disk 2015-05-12 12:11:02 i can't 2015-05-12 12:11:14 it's a raspberry pi 2015-05-12 12:12:07 alias I did not work or are not read 2015-05-12 12:16:16 ncopa, have you an idea where i can put my alias ? 2015-05-12 12:16:31 i'm not sure if bb ash supports alias 2015-05-12 12:16:33 i can test 2015-05-12 12:16:46 alias cmd return void 2015-05-12 12:17:10 alias should work 2015-05-12 12:17:33 it does not help add you alias into /etc/profile? 2015-05-12 12:17:43 i'd expect that to work 2015-05-12 12:17:45 if i define alias on shell, it works but if i put into /etc/profile, no work 2015-05-12 12:18:23 you need make sure its re-read 2015-05-12 12:18:28 eg logouto and login again 2015-05-12 12:18:41 i already try 2015-05-12 12:18:46 but not working 2015-05-12 12:19:40 and i try to find an issue :p 2015-05-12 12:38:43 ncopa i find my error 2015-05-12 12:39:33 i use "su" and alias are dropped. If i use "su -" alias work 2015-05-12 12:41:09 su - will do 'login shell' i think 2015-05-12 12:41:17 which reads /etc/profile 2015-05-12 12:42:05 yes 2015-05-12 12:42:41 thx for your help and sorry for my poor english 2015-05-12 13:25:08 It seems that samba (or possibly krb5) doesn't like how musl doesn't do relative dns queries (domain, search in resolv.conf) 2015-05-12 13:53:28 <_ikke_> ncopa: just out of curiosity, when does a package migrate from testing to main? 2015-05-12 13:56:12 _ikke_: once someone report that they work 2015-05-12 13:56:23 it has traditionally been like: 2015-05-12 13:56:33 user requests a package 2015-05-12 13:56:43 developer create a package and pushes it to testing 2015-05-12 13:57:02 developer does not test if it works 2015-05-12 13:57:10 user who requested the package test it 2015-05-12 13:57:13 and report back 2015-05-12 13:57:18 and developer move it to main 2015-05-12 13:57:39 <_ikke_> right 2015-05-12 13:57:41 unfortunally, way to often the user requested it does not report back 2015-05-12 13:57:49 so packages gets stuck in testing 2015-05-12 13:58:33 that means if user==dev you can skip the testing step to avoid the being stuck in testing part? 2015-05-12 13:58:33 btw ncopa, anything you'd be interested in having upstream: https://github.com/Orochimarufan/custom-aports ? I might go ahead and clean it up and submit a patch 2015-05-12 13:59:09 ylVa333UKfLO: normally i push to testing first anyway, to verify that the package build by build server works 2015-05-12 13:59:24 It should be noted that I've yet to get samba4 to behave correctly on alpine :( 2015-05-12 13:59:59 Orochimarufan: what is mpd difference? 2015-05-12 14:00:08 built with ffmpeg 2015-05-12 14:00:14 and PA 2015-05-12 14:00:17 see readme 2015-05-12 14:01:24 I think the most relevant ones would probably be pulseaudio and ntp 2015-05-12 14:05:08 backup='etc/ntp.conf' 2015-05-12 14:05:11 we dont have that 2015-05-12 14:05:14 I know 2015-05-12 14:05:32 I was probably too lazy to remove it 2015-05-12 14:05:38 :) 2015-05-12 14:05:42 :) 2015-05-12 14:06:02 _pkgver=4.2.8p2 2015-05-12 14:06:02 pkgver="`echo $_pkgver | sed s/p/.p/`" 2015-05-12 14:06:13 i'd like that to be: 2015-05-12 14:06:27 pkgver=4.2.8_p2 2015-05-12 14:06:39 _pkgver=${pkgver/_p/p} 2015-05-12 14:06:50 that's the least of problems ;) 2015-05-12 14:11:46 Anything else? 2015-05-12 14:19:46 I should probably split the ntp package.. 2015-05-12 14:27:43 speaking of ffmpeg, it has silly dependancies at the momemt :( 2015-05-12 14:28:46 wayland and a loads of mesa stuff 2015-05-12 14:29:31 and libx11 2015-05-12 14:33:51 what's the policy on -doc packages? always? only when the docs are large? 2015-05-12 14:41:37 Orochimarufan: always :) 2015-05-12 14:42:07 that seems kinda overkill 2015-05-12 14:43:55 & any post-install messages to /usr/share/docs/$pkg/README.alpine 2015-05-12 14:47:26 not overkill, consistant :) 2015-05-12 14:48:54 I've noticed on perl modules no warning is given is files that should be in $pkg-dev exist & no $pkgname-dev subpkg is specified 2015-05-12 14:50:57 also, any clues on getting krb5 to behave withough domain and search? 2015-05-12 14:53:15 Samba 4 is literally the only thing I've not gotten to work on Alpine yet 2015-05-12 15:35:52 After upgrading 3.1 to 3.2 I can't use postfix anymore: https://0.jmt.gr/?983d12237d82ee38#3Y+Oa1+OZd+jUZj2vwLXJEXR1NCEnFVTkwceMh0mUpM= 2015-05-12 15:37:38 My libpq also seems to be the right version: https://0.jmt.gr/?5774cce71fc96119#dpbT7moiQD3J29gKfB9tOTaq+yzmDvhwoFh+m2KlBF4= 2015-05-12 15:48:29 Oh and yes, postfix can start, but logs this: 2015-05-12 15:48:30 May 12 15:47:45 mx0 postfix/proxymap[20962]: fatal: load_library_symbols: dlopen failure loading /usr/lib/postfix/postfix-pgsql.so: Error relocating /usr/lib/postfix/postfix-pgsql.so: PQresultErrorMessage: symbol not found 2015-05-12 15:49:29 strings /usr/lib/libpq.so.5|grep PQresultErrorMessage <- has two results 2015-05-12 16:00:49 same happens on my notebook with alpine 3.2 2015-05-12 17:07:07 modprobe kvm and qemu -enable-kvm gives me: 2015-05-12 17:07:13 Could not access KVM kernel module: No such file or directory 2015-05-12 17:07:18 failed to initialize KVM: No such file or directory 2015-05-12 17:07:39 has someone seen this? 2015-05-12 17:10:15 ScrumpyJack: does lsmod|grep kvm show up something? 2015-05-12 17:10:26 ScrumpyJack: can you modprobe kvm? 2015-05-12 17:10:40 ScrumpyJack: Did you update your kernel but didn't reboot? 2015-05-12 17:11:27 ScrumpyJack: Is there a directory /lib/modules/`uname -r` 2015-05-12 19:22:21 can one of you help me through switcing to openvswitch? 2015-05-12 19:22:32 never tried and would like to have not too many failures on the route 2015-05-12 22:20:04 hi folks 2015-05-12 22:22:21 question: what is the status of zfs on alpine? 2015-05-12 22:22:52 <_ikke_> sasoiliev: afaik, still experimental 2015-05-12 22:23:09 I see that there is a zfs-grsec aport in testing, so I guess it's not tested much 2015-05-12 22:41:27 hey 2015-05-12 22:41:34 i need some help wtih getting alpine running on a dell inspiron laptop 2015-05-12 22:42:05 i cant seem to get (a) X to work in Alpine (tried all sorts of combinations, from vesa to native driver etc) 2015-05-12 22:42:50 and (b) when i load the i915 (intel video kernel module) in Alpine, it sort of freezes up and screen goes black... but the same kernel module in Ubuntu or Sabotage linux , no problem and no hangs 2015-05-12 22:43:22 anyone got any tips or tricks on getting video working on Alpine , ie a laptop intel video card? 2015-05-12 22:53:10 look at the xorg logs 2015-05-12 22:53:59 then if it looks fine use a failsafe .xinitrc (eg statx, or whatever terminal you have installed) 2015-05-12 22:54:13 i checked xorg logs 2015-05-12 22:54:17 s/statx/xterm/ 2015-05-12 22:54:31 it's wierd 2015-05-12 22:55:20 so there's no framebuffer device in alpine... that basically leaves the option of probably vesa and/or the intel driver 2015-05-12 22:56:07 with vesa, the xorg.0.log looks okay, it learns the correct resolution i think, startx (with exec startxfce4) causes the display to jump and i can see X start up for a second 2015-05-12 22:56:10 but then the system hangs 2015-05-12 22:56:48 i915 works here 2015-05-12 22:57:51 hm my xorg logs are full of dbus spam 2015-05-12 22:58:02 ... so thats #1... then there's #2... i noticed on ubuntu , X is working with the i915 kernel module and the xf86-intel driver... so i set that up on alpine (unblacklisted i915) .... and no matter how i load it, the i915 kernel module immediately hangs the system in alpine 2015-05-12 22:58:15 but (same laptop) in two other distros i915.ko works fine and is stable 2015-05-12 22:58:41 im starting to wonder whether its like a grsec thing ... or maybe ubuntu patches the i915 kernel module without telling anyone 2015-05-12 22:58:42 i dunno 2015-05-12 22:58:54 try a non grsec kernel? 2015-05-12 22:59:28 i know ubuntu works with that i915.ko kernel module, and a custom 'from scratch' linux i compiled works fine with xf86 and the generic framebuffer device... but i cant get alpine to work with anything 2015-05-12 23:00:28 yeah i might toss that grsec kernel ... i actually wanted to drop the kernel in from the working distro (the framebuffer one) but i need to make a custom initramfs so im thinking i'll just compile the kernel w/o grsec and with the modules i need built in 2015-05-12 23:00:40 I'd try a non-grsec kernel on alpine 2015-05-12 23:00:52 okay 2015-05-12 23:00:57 i think that's a good idea 2015-05-12 23:01:10 I seriously doubt selinux is the cause but might be a worthehile as well 2015-05-12 23:01:26 that's the next step ... vanilla kernel with the framebuffer, and if that works, maybe ill try vanilla kernel with i915.ko 2015-05-12 23:02:16 i really dont know... i gotta wonder tho if it's something specific to alpine tho 2015-05-12 23:02:24 does anyone here run alpine on a laptop? 2015-05-12 23:03:24 and also does anyone in here know what in the hell kernel mode set is and how that relates to X 2015-05-12 23:03:50 yes 2015-05-12 23:04:10 hmmm actually i wonder if alpine has the *wrong* kernel framebuffer driver loaded , and maybe that's cocking everything up like i915 or vesa 2015-05-12 23:04:36 (i'm running alpine on a laptop.. i didnt have to do anything magic just install the drivers with apk) 2015-05-12 23:04:41 yeah 2015-05-12 23:04:43 okay 2015-05-12 23:05:01 so u did like apk add alpine-desktop xfce4 or something 2015-05-12 23:05:08 then startx and it worked more or less 2015-05-12 23:05:16 without any crazy configs 2015-05-12 23:05:18 i dont you any fancy desktop 2015-05-12 23:05:27 i did setup-xorg i think 2015-05-12 23:05:29 yeah 2015-05-12 23:05:37 k i did that too 2015-05-12 23:05:47 then i searched for xorg drivers 2015-05-12 23:06:14 i'm not sure how to list which one is installed 2015-05-12 23:06:33 yeah 2015-05-12 23:06:39 are you on the laptop now? 2015-05-12 23:07:23 yes 2015-05-12 23:07:46 i have xf86-video-intel-2.99.917-r1 2015-05-12 23:09:05 i had to install another package for touchpad 2015-05-12 23:09:10 and that was it 2015-05-12 23:12:50 okay 2015-05-12 23:12:51 thx 2015-05-13 01:47:12 hey all, I'm trying to get wifi to work, and can't seem to get wpa_supplicant to behave. any advice? 2015-05-13 03:14:58 hey 2015-05-13 03:15:19 anyone around? 2015-05-13 03:16:04 ncopa?? 2015-05-13 03:16:30 i noticed a pretty major bug in alpine that's also replicated across to redhat as well 2015-05-13 03:16:57 i see there's a post from March 17 on the issue... https://www.mail-archive.com/busybox@busybox.net/msg21396.html 2015-05-13 03:17:43 also here: http://lists.busybox.net/pipermail/busybox/2015-March/082667.html 2015-05-13 03:18:38 basicly it seems like a kernel segmentation fault that is close to a kernel panic 2015-05-13 03:18:43 involving the hotplugging of devices 2015-05-13 03:21:12 it seemed to start after i replaced mdev with udev so i could have X autoconfigure... but now there are regular kernel warnings resulting in dump_stack() that has something to do with the linux drm.ko video driver and/or events and/or interrupts in general 2015-05-13 03:21:15 drivers/gpu/drm/drm_irq.c:1130 2015-05-13 03:21:46 i see identical issue to the redhat issue open here... also identical to the recent bugs open by ncopa 2015-05-13 05:56:45 morning 2015-05-13 05:57:00 jomat: do you have a minimalistic config to reproduce the postfix/pgsql issue? 2015-05-13 06:06:50 jomat: i think i know the problem 2015-05-13 06:06:57 TYPE NEEDED FILE 2015-05-13 06:06:57 ET_DYN libc.musl-x86_64.so.1 pkg/postfix-pgsql/usr/lib/postfix/postfix-pgsql.so 2015-05-13 06:07:07 it is not linked with libpq 2015-05-13 06:11:49 I just installed Alpine and I'm getting IO ERROR when trying to do apk update. It's just regular v3.1 2015-05-13 06:12:21 Actually I got that while installing too but was able to install anyways, still cant update after booting into the install though 2015-05-13 06:12:42 what do you have in /etc/apk/repositories? 2015-05-13 06:13:29 just http://nl.alpinelinux.org/alpine/v3.1/main but I also tried dl-2 with same result 2015-05-13 06:13:48 Wait nevermind 2015-05-13 06:14:34 I think it's network problems 2015-05-13 06:19:16 Yeah... that was it 2015-05-13 06:21:05 Was just missing resolv.conf 2015-05-13 06:28:44 how to restart this "udhcpc -R -n -p /var/run/udhcpc.eth0.pid -i eth0 -x hostname srv01" after i changed hostname inside /etc/network/interfaces ? 2015-05-13 06:32:05 ok found it /etc/init.d/networking is this :) 2015-05-13 06:32:14 ifdown eth0; ifup eth0 2015-05-13 06:38:15 jomat: i think postfix-3.0.1-r2 shoudl have the fix for your postfix issue 2015-05-13 06:48:21 what for is /etc/TZ used? i have had date/time issue, which i fixed with setup-timezone (i think i select UTC durring install) , here some log http://sprunge.us/NTXa 2015-05-13 06:50:45 is this something we should think over ? https://lists.ubuntu.com/archives/ubuntu-devel/2015-May/038761.html 2015-05-13 06:51:01 "stateless persistant network interface names" 2015-05-13 07:28:02 crow: i think /etc/TZ is leftovers since uclibc 2015-05-13 07:28:13 i dont think it is needed anymore for musl libc 2015-05-13 07:30:37 <_ikke_> vkrishn: Idea is nice, but those names are anoying :-( 2015-05-13 07:31:08 <_ikke_> enp5s0 2015-05-13 07:31:42 vkrishn: we kinda solved that by sorting the modailases 2015-05-13 07:32:04 at least when using mdev 2015-05-13 07:32:33 is this a problem for us? 2015-05-13 07:32:59 ncopa so it could be ignored and removed 2015-05-13 07:33:13 crow: i think so yes 2015-05-13 07:34:11 jomat: yes to all those questions :) 2015-05-13 07:40:21 _ikke_, may wait to see how other distros do, coz one issue it that causes admins to generate configs it may be a migration issue 2015-05-13 07:41:00 issue is, if it * 2015-05-13 07:42:29 iirc, even openbsd has different names 2015-05-13 07:42:46 hard to remember 2015-05-13 07:43:44 lsmod show kvm module loaded but qemu -enable-kvm says 2015-05-13 07:43:49 Could not access KVM kernel module: No such file or directory 2015-05-13 07:43:54 failed to initialize KVM: No such file or directory 2015-05-13 07:45:41 an strace show that qemu is trying to open /dev/kvm 2015-05-13 07:48:36 i have to load kvm_intel too 2015-05-13 07:48:41 and it works :) 2015-05-13 07:49:04 <_ikke_> strace ftw 2015-05-13 07:51:32 qemu loaded, next problem is i have no focus in spicec :( 2015-05-13 08:03:27 ScrumpyJack: you need kvm-{intel,amd} as you said and you might need to add your user to 'kvm' group 2015-05-13 08:05:00 crw-rw-rw- 1 root kvm 10, 232 May 13 07:53 /dev/kvm 2015-05-13 08:05:03 looks fine 2015-05-13 08:05:54 the mouse focus in spice seems to be for a specific iso, works with others 2015-05-13 08:06:20 although i once i have mouse focus in the spice client, i can't get it back lol 2015-05-13 08:12:47 was using the wrong vga option. now i can get mouse focus in the guest that wasn't working 2015-05-13 08:13:42 can't unfocus the mouse though. it stays stuck in the guest. anyone remember de default keystrokes to release the mouse? :) 2015-05-13 08:20:45 ctrl+alt+G 2015-05-13 08:22:17 nope :( 2015-05-13 08:24:53 docs say Shift + F12 2015-05-13 08:26:57 ncopa: i can confirm musl does not use /etc/TZ 2015-05-13 08:27:01 feel free to remove it 2015-05-13 08:27:03 ;p 2015-05-13 08:29:05 spice is flaky with mouse focus. sometimes it will grab the mouse and focus, other times it won't. 2015-05-13 08:31:43 ScrumpyJack: i added qemu-gtk too 2015-05-13 09:02:03 ncopa: thanks. that works wonders. shame spice doesn't work well 2015-05-13 09:02:40 not tried vnc yet. do we have a good vnc client? 2015-05-13 09:02:54 ScrumpyJack: spice works for me in virt-manager with win7 guest 2015-05-13 09:03:26 i've not used virt-manager? 2015-05-13 09:03:37 with a ? on the ned 2015-05-13 09:03:39 end even 2015-05-13 09:07:09 i just use spicec to connect 2015-05-13 09:14:55 if I toggle full screen on then off for the spice client, I cat get mouse focus in the guest 2015-05-13 09:39:09 05:56:55│@ncopa> jomat: do you have a minimalistic config to reproduce the postfix/pgsql issue? <-- I reproduced it in a test-container, just drop me a ssh key 2015-05-13 09:40:57 jomat: i think i found the issue 2015-05-13 09:41:07 problem was that plugin was not linked with -lpq 2015-05-13 09:41:22 i think it is fixed now 2015-05-13 09:41:26 could you please test? 2015-05-13 09:42:00 I'll try 2015-05-13 09:46:47 ah ! and I thought it was ruby-lang(scripting), https://hacks.mozilla.org/2015/03/ruby-support-in-firefox-developer-edition-38/ 2015-05-13 09:47:20 ncopa: I stumbled over another problem right now: #!/sbin/openrc-run 2015-05-13 09:47:39 the old init script used #!/sbin/runscript 2015-05-13 09:47:50 /sbin/openrc-run doesn't exist here 2015-05-13 09:48:06 hum 2015-05-13 09:48:09 due to mixed repos 2015-05-13 09:48:41 do you dare upgrade to a pure v3.2? 2015-05-13 09:49:12 I did :-) 2015-05-13 09:49:33 you should have gotten a newer version of openrc with it 2015-05-13 09:49:41 looks better now :-) 2015-05-13 10:12:01 why can't perl release the scaRRy version 6, say ooops something went wrong, we are correcting it and releasing version 7 ! 2015-05-13 10:14:28 or best version 6 was our internal release that had lots of scarry stuffs 2015-05-13 11:00:34 hmm, i get http://sprunge.us/HVZB when launching gvncviewer 2015-05-13 11:02:18 i get at-spi-bus-launcher problems too 2015-05-13 11:02:23 i havent digged up why it happens 2015-05-13 11:02:32 i htink its gtk accessibility something 2015-05-13 11:14:06 do we have 3.2 iso already ? 2015-05-13 11:20:07 crow: did you check our repo's? 2015-05-13 11:20:26 last time i did was not there 2015-05-13 11:20:30 1-2 weeks before 2015-05-13 11:20:50 http://nl.alpinelinux.org/alpine/v3.2/releases/x86_64/ 2015-05-13 11:20:56 you can click the link 2015-05-13 11:22:49 its still rc (forgot to write stable) 2015-05-13 11:23:44 stable is not out yet 2015-05-13 12:49:10 thats what i am asking for :) 2015-05-13 14:05:08 any suggestion for a nice free email hosted in EU ? 2015-05-13 14:12:25 <_ikke_> In a PKGBUILD, is it possible to specify a name for a repository downloaded through the git protocol? saveas-git://.../name tries to download it through curl 2015-05-13 14:25:55 no its not possible 2015-05-13 14:26:22 you have to write your own function to checkout git 2015-05-13 14:26:56 _ikke_: is it from github? 2015-05-13 14:37:20 <_ikke_> clandmeter: It was from sourceforge, so I used the normal sourceforge url 2015-05-13 14:37:30 <_ikke_> git.code.sf.net 2015-05-13 14:49:25 _ikke_: i dont know about their snapshots, but the ones from github are static. 2015-05-13 14:58:54 <_ikke_> clandmeter: I don't mind 2015-05-13 15:09:24 _ikke_: https://wiki.archlinux.org/index.php/VCS_PKGBUILD_Guidelines 2015-05-13 15:09:27 it is supported 2015-05-13 15:09:55 <_ikke_> Is it possible to build packages without abuild expecting to be in the aports repository? It now currently complains about not being in one 2015-05-13 15:10:50 I don't know, so I always just clone aports and do my work in there 2015-05-13 15:11:00 I usually intend to eventually submit my packages back :P 2015-05-13 15:11:40 <_ikke_> hehe 2015-05-13 15:58:37 _ikke_ ~/packages acts as a local repo while you are building something new with a few new depends 2015-05-13 16:48:49 hi there 2015-05-13 16:49:53 when I try to install some go tools using `go get` I have the following error: /usr/lib/go/pkg/tool/linux_amd64/6l: corrupt input 2015-05-13 16:50:07 eg. go get github.com/mailgun/godebug 2015-05-13 16:50:19 do you know how to fix that? 2015-05-13 17:16:11 fixed :) 2015-05-13 17:16:29 was more go related 2015-05-13 17:17:54 by the way I have another problem which is alpine related, when I try to install github.com/derekparker/delve/cmd/dlv using `go get` too, I have the following error: fatal error: sys/debugreg.h: No such file or directory 2015-05-13 17:18:08 do you know how can I install sys/debugreg.h ? 2015-05-13 17:41:51 <_ikke_> BitL0G1c: I don't have ~/packages, I do have ~/aports 2015-05-13 17:43:04 _ikke_ once something compiled in ~/aports will be packaged and copied to ~/packages (i think) 2015-05-13 17:44:33 yes 2015-05-13 18:02:53 <_ikke_> A, I have a custom dir for that 2015-05-13 18:03:33 <_ikke_> Just find it anoying it's complaining about wanting a git repo all the time 2015-05-13 18:35:06 <_ikke_> http://wiki.alpinelinux.org/wiki/Writing_Init_Scripts says I should add #!/sbin/runscript to initd files, but in aports, it's /sbin/openrc-run, has it changed? 2015-05-13 18:42:35 _ikke_, that was a recent change. this week or the previous one 2015-05-13 18:42:38 wiki should be updated 2015-05-13 18:43:42 <_ikke_> right 2015-05-13 18:45:51 <_ikke_> If a project offers things like vim plugins, should those be added to a seperate sub package? 2015-05-13 18:52:33 that would probably be preferable 2015-05-13 18:53:34 <_ikke_> right 2015-05-13 18:53:50 <_ikke_> Is the convention like $pkgname-vim ? 2015-05-13 18:55:47 i'm not sure if there is convention 2015-05-13 18:55:58 that would probably do 2015-05-13 18:56:07 <_ikke_> right 2015-05-13 21:53:27 Mo0O: /usr/include files are always in *-dev packages, not sure where that one is but I always start with libc-dev 2015-05-13 22:06:14 ahills: hmmm, it looks like it's not in libc-dev, do you know how can I search for a lib in packages? 2015-05-13 23:37:14 _ikke_: openrc-run is preferred, it was changed from runscript because some packages also ship a binary called runscript 2015-05-14 05:08:12 hmm, the alpine lxc template won't download a static apk 2015-05-14 05:08:22 or it will, but the signature fails, impossible to tell exactly what's happening 2015-05-14 07:05:20 ahills: which host distro do you use? 2015-05-14 07:22:46 <_ikke_> I've downgraded a custom built package, but the packages repo still contains the newer version. Is there a way to remove the newer version? 2015-05-14 08:09:27 _ikke_: do you mean your local repo? 2015-05-14 08:09:31 your cache? 2015-05-14 08:11:06 apk cache clean ? 2015-05-14 08:26:10 <_ikke_> ScrumpyJack: The one built by abuild 2015-05-14 08:59:34 <_ikke_> abuild automatically adds and signs the APKINDEX, but it doesn't seem trivial to remove an older package without rebuilding the repo completely 2015-05-14 08:59:41 <_ikke_> older or newer 2015-05-14 13:40:53 ncopa: I'm trying to do it from Arch Linux in this case 2015-05-14 13:41:05 ahills: ok 2015-05-14 13:41:22 my dedicated server host will not provide alpine to run as my host OS 2015-05-14 13:41:37 so I've got an arch+openrc setup 2015-05-14 13:41:49 i dont know why it fails 2015-05-14 13:42:02 Sorry I just woke up, I'm hacking the /usr/share/lxc/templates/lxc-alpine script now 2015-05-14 13:42:07 ok 2015-05-14 13:42:11 going to add some more detailed error messages 2015-05-14 13:48:27 <_ikke_> ncopa: How should I handle downgrading packages in a repository? The newer version would be still there and alpine then prefers the newer version over the older version 2015-05-14 13:48:30 ncopa: are you the author of this script? 2015-05-14 13:48:43 I think I saw you submitting a patch at least on an lxc mailer 2015-05-14 13:53:02 apk upgrade -U -a 2015-05-14 13:53:41 if you want keep a lower version of a package, lets say package foo should not be 1.0 or newer then: 2015-05-14 13:53:50 apk add 'foo<1.0' 2015-05-14 13:54:05 maybe apk add -u 'foo<1.0' 2015-05-14 13:54:19 i think i wrote majority of that script yes 2015-05-14 13:55:44 it has the same sort of clear and concise style as the alpine setup scripts, so I had my suspicions that at least someone on this project was involved 2015-05-14 13:56:17 i think kunkku also did parts of it 2015-05-14 13:57:27 the failing command is the << openssl dgst -verify $rootfs/etc/apk/keys/$keyname -signature "$APK.SIGN.RSA.$keyname" "$APK" >> 2015-05-14 13:57:42 lxc keeps deleting the rootfs when it fails lol 2015-05-14 13:58:13 can you try run the command manually? 2015-05-14 13:58:18 do you get an error message? 2015-05-14 13:59:42 oh, i get your problem about purging the rootfs... 2015-05-14 13:59:43 heh 2015-05-14 14:00:05 fixed that by replacing "return 1" with "read" :P 2015-05-14 14:00:14 openssl is, of course, completely unhelpful 2015-05-14 14:00:21 Verification Failure is the only message 2015-05-14 14:00:44 this is signing v3.1's apk.static 2015-05-14 14:01:43 er, not signing, of course... checking the signature... 2015-05-14 14:04:07 for one foolish moment I was wondering why I couldn't produce the same signature on my own 2015-05-14 14:04:37 any particular reason you use openssl instead of to sign? 2015-05-14 14:04:44 ubiquity I'm guessing 2015-05-14 14:05:57 in any case, /sbin/apk.static.SIGN.RSA.alpine-deve 2015-05-14 14:05:57 l@lists.alpinelinux.org-4a6a0840.rsa.pub is an invalid signature for /sbin/apk.static as downloaded from the v3.1 repository, I just tested it on another machine to be certain 2015-05-14 14:13:21 <_ikke_> ncopa: thanks 2015-05-14 14:13:39 <_ikke_> ncopa: So there is no easy way to remove a package from a repository (APKINDEX)? 2015-05-14 14:14:10 i do: abuild cleanoldpkg 2015-05-14 14:14:10 _ikke_: "rm" 2015-05-14 14:14:43 <_ikke_> ahills: Just removing the package is enough? 2015-05-14 14:14:51 you need update the index too 2015-05-14 14:15:04 so i do that from aports tree 2015-05-14 14:15:13 there is a command to update the index 2015-05-14 14:15:29 abuild update_aportsrepo or similar 2015-05-14 14:15:34 maybe its abuild index 2015-05-14 14:16:09 <_ikke_> I think so 2015-05-14 14:16:10 <_ikke_> ok 2015-05-14 14:18:03 apk also has a command to build the index 2015-05-14 14:18:25 it'll just build it from the tree of packages you give 2015-05-14 14:24:39 I personally use a custom lxc template with support for shared package caches and external /var mountpoints (don't want those on ssd) 2015-05-14 14:27:28 Orochimarufan: how did you arrange the shared package cache? symlink out of rootfs or some other linking scheme? 2015-05-14 14:27:39 mount 2015-05-14 14:28:02 nice 2015-05-14 14:28:03 I make heavy use of btrfs subvols 2015-05-14 14:28:07 ah 2015-05-14 14:28:13 I'm not ready to experiment with btrfs yet 2015-05-14 14:28:35 is your mounting any more complex than binding the same location to a bunch of rootfs's? 2015-05-14 14:29:06 I have Package/ subvols so I can mount /dev/sdd2 to the package cache location with -o subvol=Packages/ 2015-05-14 14:29:38 that's fancy 2015-05-14 14:29:56 it's also pretty easy for you to version distros that way 2015-05-14 14:30:13 I'm looking forward to proper deduplication in btrfs :) 2015-05-14 14:30:43 btrees are indeed a powerful category of data structure 2015-05-14 14:30:55 And since both the container / and /var are btrfs subvols, I can also easily snapshot them, both at once or separately 2015-05-14 14:35:55 I snapshot with dd 2015-05-14 14:57:29 https://github.com/Orochimarufan/lxc-tools/blob/master/share/templates/lxc-alpine2 that's what I use 2015-05-14 14:57:49 the template itself isn't all that different 2015-05-14 15:00:19 https://github.com/Orochimarufan/lxc-tools/blob/master/share/template.functions that's where the real magic happens 2015-05-14 15:05:26 thanks 2015-05-14 15:05:26 the default alpine template doesn't set a root password, does it? 2015-05-14 15:05:35 nope 2015-05-14 15:05:54 It doesn't really make sense to 2015-05-14 15:16:54 it does if you're going to run sshd in the container 2015-05-14 15:17:53 yeah but what should it set it to 2015-05-14 15:18:12 oh I thought you meant it doesn't make sense to set a root password in a container 2015-05-14 15:18:14 just run lxc-attach -n -- passwd once after creating it 2015-05-14 15:18:38 lxc-attach: attach.c: lxc_attach: 632 failed to get the init pid 2015-05-14 15:18:48 using a default password is in no way more secure than setting none 2015-05-14 15:18:53 I agree 2015-05-14 15:18:54 you need to lxc-start it first 2015-05-14 15:19:01 it is already started 2015-05-14 15:19:13 then something is wrong with your lxc :O 2015-05-14 15:19:21 you can try lxc-console 2015-05-14 15:19:28 what I'm trying to do here is create a VM without the overhead of qemu, etc, since I don't need the added security and such 2015-05-14 15:19:51 huh, that one worked, but attach doesn't... 2015-05-14 15:20:02 yeah attach is quite magical 2015-05-14 15:20:10 something's up with my setup, I think 2015-05-14 15:20:16 but console has its own issues 2015-05-14 15:20:28 I think I might pass over lxc for this purpose and just eat the extra resources with a kvm guest 2015-05-14 15:20:34 generally, attach does the job better IF it works at all 2015-05-14 15:20:53 I'd stick to lxc for linux guests 2015-05-14 15:21:03 But that's just my personal stance 2015-05-14 15:21:31 well, if LXC works, then I will... 2015-05-14 15:21:49 but I only have until the end of the day until my mail is permanently bounced 2015-05-14 15:22:08 if you're going to use ssh anyway, lxc-attach doesn't really matter, does it? 2015-05-14 15:22:17 bounced? 2015-05-14 15:22:38 also: 2015-05-14 15:22:45 ACTION hints at #lxcontainers 2015-05-14 15:23:01 Though the activity there is generally quite low :( 2015-05-14 15:24:55 I'll need to attach to install sshd :P 2015-05-14 15:25:21 true enough 2015-05-14 15:25:29 you get about three days of your mail server being down before messages stop getting resent 2015-05-14 15:25:49 I'm moving services away from one host with xen to a new host, and I wanted to try LXC 2015-05-14 15:25:51 fun fact: usually you can stop the container and do some fiddling using chroot 2015-05-14 15:26:05 yeah, apk doesn't work as expected in the chroot though 2015-05-14 15:26:13 unless your setup is too crazy (like mine :P) 2015-05-14 15:26:18 can't find and/or write to the package database 2015-05-14 15:26:20 that's a problem then 2015-05-14 15:26:44 lxc-console is working fine 2015-05-14 15:26:51 that's good 2015-05-14 15:26:53 I'm not sure what's missing that apk needs... 2015-05-14 15:27:00 what's it saying? 2015-05-14 15:30:40 # apk update 2015-05-14 15:30:40 ERROR: Unable to lock database: No such file or directory 2015-05-14 15:30:40 ERROR: Failed to open apk database: No such file or directory 2015-05-14 15:30:59 hmm 2015-05-14 15:31:03 in the console it works fine, although I must have broken something with ip forwarding 2015-05-14 15:31:23 well I meant in the console 2015-05-14 15:31:34 ip forwarding shouldn't be too hard to set up 2015-05-14 15:31:39 what did you do? 2015-05-14 15:32:10 sysctl -w net.ipv4.ip_forwarding=1 2015-05-14 15:32:30 oh, in the console apk works fine, I must just have misconfigured the bridge or something 2015-05-14 15:33:22 well yeah but what did you do with the bridge 2015-05-14 15:33:49 and did you set up the veth endpoint in the container (does it have an ip assigned etc) 2015-05-14 15:34:21 Not having the veth device up in the container is a pit that I've fallen into a few times 2015-05-14 15:34:58 yeah, I can ping between the guest and host 2015-05-14 15:35:02 it's broken routing 2015-05-14 15:35:31 sorry--the bridge is not attached to the host's actual ethernet controller, only a dummy 2015-05-14 15:35:35 do you have your external network device added to the bridge or are you trying to do subnet routing? 2015-05-14 15:35:45 I'm doing the routing on my host 2015-05-14 15:35:53 So NAT? 2015-05-14 15:36:22 In that case, did you set up the iptables rules? 2015-05-14 15:36:50 yep 2015-05-14 15:37:06 hmm, I've run out of ideas then 2015-05-14 15:37:09 I have a few bridges created, I think I typo'd in one of the few places 2015-05-14 15:37:31 late last night, after a few beers, I was too ambitious and wanted to migrate all my services at once 2015-05-14 15:37:45 shit happens^^ 2015-05-14 15:38:16 I just threw out my whole system a month ago and spent almost a week setting everything back up 2015-05-14 15:38:33 that's what I'm doing, except not entirely voluntarily... 2015-05-14 15:38:57 I wasn't happy with my old system at the time 2015-05-14 15:39:11 Unfortunately, I'm not exactly happy with the current iteration either >.< 2015-05-14 15:39:59 Haven't gotten Samba4 to work properly yet and I don't have the slightest idea why it's broken 2015-05-14 15:40:51 I'm guessing it has to do with how musl handles DNS but I have no proof to back that claim 2015-05-14 15:41:21 I try to avoid mixing DNS and essential services 2015-05-14 15:41:32 I use rsync+/etc/hosts for important stuff 2015-05-14 15:41:51 ahh, much better now that I have the correct outgoing interface specified 2015-05-14 15:42:02 silly me thinking it would be named eth0 2015-05-14 15:42:04 Well, DNS is an intricate part of Active Directory, there's no way around it 2015-05-14 15:43:22 And musl has some differences in how it handles name resolution 2015-05-14 15:43:27 oof, you're stuck in an actual AD network? 2015-05-14 15:43:52 No, I'm trying to set one up so I can keep the windows machines in check 2015-05-14 15:44:26 My goal is to centralize as much as I can 2015-05-14 15:44:36 are you running physical machines or KVM? 2015-05-14 15:44:45 the clients? 2015-05-14 15:44:45 or other virt 2015-05-14 15:44:51 any of the windows machines 2015-05-14 15:44:54 both 2015-05-14 15:45:11 are the physical ones workstations or something, or could they be virtualized? 2015-05-14 15:45:37 nah, there's no way they'll be virtualized 2015-05-14 15:46:14 It's hard enough trying to convince people that centralizing is a good idea 2015-05-14 15:47:15 Seeing as I'm the only linux guy anyway, trying to force everyone to switch to virtualizing windows doesn't sound like it'll help the case at all 2015-05-14 15:47:34 I thought commenting out the entire /etc/network/interfaces file would let me start networking to satisfy daemons without modifying the existing setup, but the networking script doesn't like that... 2015-05-14 15:47:40 not to mention that it'll defeat the purpose of some of the windows machines 2015-05-14 15:47:52 just use a custom script 2015-05-14 15:48:09 I don't want updates to override it 2015-05-14 15:48:24 name it differently 2015-05-14 15:48:28 https://github.com/Orochimarufan/lxc-tools/blob/master/misc/openrc_network.in 2015-05-14 15:48:32 that's what I use 2015-05-14 15:48:44 note the depend() part 2015-05-14 15:49:35 how does openrc choose which provider? 2015-05-14 15:49:41 it doesnt 2015-05-14 15:50:08 it will need to, since networking will also provide it 2015-05-14 15:50:24 you just tell it to start your script (rc-update add ) 2015-05-14 15:50:38 hmm, ok 2015-05-14 15:50:42 that way, the dependency is already satisfied and it doesn't try to start networking 2015-05-14 15:51:16 in your script, you're actually configuring your interface 2015-05-14 15:51:47 yeah, I don't use the lxc setup options 2015-05-14 15:51:58 why not? 2015-05-14 15:52:28 dunno, I've just started doing it this way when I started using lxc 2015-05-14 15:52:33 ok 2015-05-14 15:52:42 I don't see any reason to choose one over the other 2015-05-14 15:53:31 generally, the guest distro expects to need to set up the network, so I'd need to deal with that in any case 2015-05-14 15:53:49 As you can tell from your current endeavor ;) 2015-05-14 15:55:42 haha yeah 2015-05-14 15:56:42 are there musl update backports for 3.1.4? i'm only seeing upgrade for alpine-base, alpine-conf, and apk-tools packages 2015-05-14 15:57:30 backports for what was just released? 2015-05-14 15:58:01 He means if any musl updates were backported from 3.2 and included in the 3.1.4 release 2015-05-14 15:58:31 yea, in the news article on site i see two commits for main/musl, but the package doesn't seem updated? 2015-05-14 15:58:46 I don't know, since I don't think I have a 3.1 container anymore 2015-05-14 16:00:14 3.2 isn't actually out yet, is it? 2015-05-14 16:00:21 nope, it's in rc 2015-05-14 16:00:38 3.2.0_rc2-126-gdf42cb0 2015-05-14 16:02:25 oh, i see now a ===== release 3.1.4 ===== marker in the changelog... maybe everything under that is 3.2+? 2015-05-14 16:02:58 I think there were some backports, but not a musl upgrade 2015-05-14 16:03:13 http://git.alpinelinux.org/cgit/aports/log/main/musl?h=v3.1.4&id=4f9aab14c673986a852e0bb58faa3ef2f8c0244e&showmsg=1 2015-05-14 16:04:39 Orochimarufan: but those are already all in 3.1.3 branch as well, correct? 2015-05-14 16:05:00 seems like the 3.1.4 changelog includes some stuff that was in 3.1.3, that is why i am a little confused 2015-05-14 16:05:10 pretty sure 3.1.3 was tagged before the top 2 commits 2015-05-14 16:05:54 http://git.alpinelinux.org/cgit/aports/log/main/musl?h=v3.1.3 2015-05-14 16:07:39 expect to see more than one musl commit with "cherry-pick upstream" 2015-05-14 16:07:48 As you can see, there definitely were patches to musl, just no new upstream import 2015-05-14 16:09:38 ok, i have 1.1.5-r4 on my 3.1.3 image... i wonder how i managed that... 2015-05-14 16:10:04 which mirror is closest to zürich? 2015-05-14 16:11:06 I don't know what the repo tracks, but it's probably the 3.1 branch, not the tags 2015-05-14 16:11:15 So you get rolling updates 2015-05-14 16:11:40 I'd imagine that the actual releases just determine what goes onto an install iso or something like that 2015-05-14 16:11:44 rolling updates within major.minor 2015-05-14 16:12:25 yea yea ok, so i just happened to build this 3.1.3 image after those packages were pushed, and then 3.1.4 is just more of a rollup of what changed since 3.1.3 started 2015-05-14 16:12:42 yeah 2015-05-14 16:12:56 ok i feel less crazy now, thanks! :) 2015-05-14 16:14:47 hmm, maybe it is better to track the latest commit on 3.1-stable at build time rather than release version, then there is a more clear point in time of what the image has 2015-05-14 16:15:36 you mean the other way around? 2015-05-14 16:16:14 oh I see what you mean 2015-05-14 16:16:28 you want to version your images 2015-05-14 16:16:44 yeah, the 3.1-stable HEAD is probably your best bet 2015-05-14 16:17:22 it might still be slightly off since the build might not have finished and pushed to the repos after the change was pushed 2015-05-14 16:19:02 I wonder if there's a way to get the APKINDEX hash 2015-05-14 16:20:53 curl .../APKINDEX.tar.gz | md5sum 2015-05-14 16:21:43 well, it already has some kind of hash, I wonder what that is: 2015-05-14 16:22:02 -/var/cache/apk/APKINDEX.1ebd6aac.tar.gz /var/cache/apk/APKINDEX.66222e8e.tar.gz /var/cache/apk/APKINDEX.cbed5262.tar.gz /var/cache/apk/APKINDEX.f47f7f90.tar.gz /var/cache/apk/APKINDEX.fbe88a9c.tar.gz 2015-05-14 16:26:00 oh hmm 2015-05-14 16:26:41 it might be the same as git's, which is the last n digits of the sha1 2015-05-14 16:26:51 (in this case 8) 2015-05-14 16:27:15 that's what I'm guessing 2015-05-14 16:27:21 but where does apk get it from 2015-05-14 16:27:59 apk's caching renames the files it downloads 2015-05-14 16:28:09 all of the apks (if you enable the cache) have a hash appended too 2015-05-14 16:29:08 true 2015-05-14 16:29:41 the question is, what does that hash correspond to? the file hash? the commit on the repo it was built from? 2015-05-14 16:30:09 well, it's not the file hash, I just tried all the *sum commands 2015-05-14 16:32:13 it's not the commit hash either 2015-05-14 16:32:19 er, commit ID 2015-05-14 16:32:50 the commit id does seem to be embedded in the APKINDEX tho 2015-05-14 16:33:03 so you could try to extract it from there 2015-05-14 16:35:25 I don't think it is... 2015-05-14 16:35:37 it's somewhere in there 2015-05-14 16:35:48 v3.2.0_rc2-126-gdf42cb0 2015-05-14 16:36:24 not in 3.1.4 2015-05-14 16:36:34 I'm looking at the raw tar in vim tho (cause the prepended signature breaks tar -xf) 2015-05-14 16:36:44 tar -xzf 2015-05-14 16:36:54 busybox tar doesn't autodetect 2015-05-14 16:37:03 ugh.. 2015-05-14 16:37:36 tar Ozf APKINDEX.448df425.tar.gz DESCRIPTION 2015-05-14 16:37:50 yeah the DESCRIPTION one has it 2015-05-14 16:38:13 it probably uses git describe, so it won't have the hash if it matches the tag exactly 2015-05-14 16:38:20 in 3.1.4 it's jus "v3.1.4" 2015-05-14 16:38:23 it's the name of the release 2015-05-14 16:38:50 but the commit tagged v3.1.4 is not 448df425 2015-05-14 16:39:18 well yeah I don't think the filenames have the commit 2015-05-14 16:39:29 but the DESCRIPTION just might 2015-05-14 16:39:50 what? 2015-05-14 16:40:13 what is in the description file? 2015-05-14 16:40:24 just v3.1.4 2015-05-14 16:40:30 ah 2015-05-14 16:40:33 for me, it's v3.2.0_rc2-126-gdf42cb0 2015-05-14 16:40:40 right, because that's the name of the release 2015-05-14 16:41:10 I still think that's because 3.1-stable == v3.1.4 right now 2015-05-14 16:41:14 this mystery could probably be solved by ten minutes looking at apk source code but I am going to return to building my system 2015-05-14 16:46:31 anyone know why when i start virt-manager i see this https://dpaste.de/cHQ6/raw 2015-05-14 16:46:36 i'm in the kvm and libvirt group 2015-05-14 16:47:02 i instaled: 2015-05-14 16:47:04 virt-manager 2015-05-14 16:47:07 virt-install 2015-05-14 16:47:09 virt-viewer@testing 2015-05-14 16:47:11 libvirt 2015-05-14 16:47:13 libvirt-qemu 2015-05-14 16:47:15 libvirt-client 2015-05-14 16:47:17 bridge-utils 2015-05-14 16:47:19 qemu 2015-05-14 16:47:21 and i've put libvirtd in the default runlevel 2015-05-14 16:47:23 i've also added the kvm module to /etc/modules and have rebooted 2015-05-14 16:47:38 (new to KVM) so i'm having to adapt the installation from other distributions 2015-05-14 16:47:50 --description "$repo $(cd $startdir && git describe)" 2015-05-14 16:47:57 it does git describe after all 2015-05-14 16:48:08 http://git.alpinelinux.org/cgit/abuild/tree/abuild.in#n1236 2015-05-14 16:48:11 anyway, dinner 2015-05-14 16:50:54 i must be missing something 2015-05-14 16:51:02 because sudo virsh -c qemu:///system list 2015-05-14 16:51:06 only works if i use sudo 2015-05-14 16:51:22 error: internal error: Unable to get DBus system bus connection: Failed to connect to socket /var/run/dbus/system_bus_socket: No such file or director 2015-05-14 16:51:34 maybe i need to be in the dbus group 2015-05-14 16:52:14 http://bugs.alpinelinux.org/issues/2686 2015-05-14 16:52:25 $ id 2015-05-14 16:52:27 uid=1000(user) gid=1000(user) groups=18(audio),34(kvm),102(libvirt),1000(user) 2015-05-14 16:52:32 so am i missing the group? 2015-05-14 16:52:59 is it safe to add myself to messagebus or will that have unintended consequences 2015-05-14 16:53:56 suppose ill try that 2015-05-14 16:57:45 sounds a bit like this https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=765045 2015-05-14 16:59:10 oh do i need polkit agent? 2015-05-14 17:01:43 yeah maybe i will install lxpolkit and see if that works 2015-05-14 17:01:50 i think you need an agent for users for start VMs 2015-05-14 17:10:44 ncopa: do you use lxpolkit? 2015-05-14 17:11:32 if anyone has any experience with this please speak up 2015-05-14 17:12:13 I'm using it over a ssh to root, so I can't comment 2015-05-14 17:12:45 ah 2015-05-14 17:13:15 $ virsh -c qemu:///system list 2015-05-14 17:13:16 error: failed to connect to the hypervisor 2015-05-14 17:13:18 error: authentication failed: no agent is available to authenticat 2015-05-14 17:13:26 i was under the impression i was seeing that because of no polkit agent is started 2015-05-14 17:14:09 i installed lxpolkit because i'm using lxdm 2015-05-14 17:15:40 what im suprised is there's no policykit daemon 2015-05-14 17:15:42 in /etc/init.d 2015-05-14 17:23:47 sillysausage: polkit is a mechanism to allow users to invoke superuser functionality in programs, but I don't know how libvirt uses it, sorry 2015-05-14 17:23:58 I manage my VMs as root 2015-05-14 17:24:05 okay i suppose that's okay 2015-05-14 17:24:15 its not unsafe? 2015-05-14 17:24:20 the system only has one VM anyway 2015-05-14 17:24:23 I also haven't used KVM yet, only Xen 2015-05-14 17:24:44 It's unsafe if you don't know what you're doing ;) 2015-05-14 17:25:19 mm 2015-05-14 17:25:22 But the end effect of any method is still superuser activity 2015-05-14 17:25:28 yeah 2015-05-14 17:25:35 Polkit is just an interface for managing it 2015-05-14 17:25:49 as long as i don't need superuser to use something like virt-client 2015-05-14 17:26:06 A bit more fine-grained than sudo 2015-05-14 17:26:39 For what you're doing it sounds like it doesn't matter, just use sudo 2015-05-14 17:26:50 hmm okay 2015-05-14 17:27:31 libvirt 2015-05-14 17:27:33 libvirt-qemu 2015-05-14 17:27:35 libvirt-client 2015-05-14 17:27:37 bridge-utils 2015-05-14 17:27:39 qemu 2015-05-14 17:27:41 virt-manager 2015-05-14 17:27:43 virt-install 2015-05-14 17:27:45 that's the packages i installed 2015-05-14 17:27:53 basically i'm trying to just use kvm to run a single VM, i used to use virtualbox but i heard kvm had better performance 2015-05-14 17:31:08 <_ikke_> What OS family would Alpine fall under? 2015-05-14 17:31:20 what do you mean 2015-05-14 17:31:50 it's linux, own distribution, but does use openrc scripts so those come from gentoo mostly 2015-05-14 17:31:55 <_ikke_> right 2015-05-14 17:32:09 <_ikke_> Many tools report a OS family 2015-05-14 17:32:16 <_ikke_> debian for ubuntu / debian, etc 2015-05-14 17:32:26 yeah it doesn't come from a family really 2015-05-14 17:32:46 <_ikke_> Yeah, I gathered 2015-05-14 17:33:26 <_ikke_> Archlinux has osfamily Archlinux, so I guess it would be Alpinelinux then 2015-05-14 17:33:33 yeah 2015-05-14 17:34:39 i found a log file with people who had the same issue :P http://dev.alpinelinux.org/irclogs/%23alpine-linux-2015-03.log 2015-05-14 17:36:53 Mo0O: did you get virt-manager to work? 2015-05-14 17:41:52 brb 2015-05-14 17:46:59 did anyone say anything? 2015-05-14 17:50:16 https://help.ubuntu.com/lts/serverguide/libvirt.html#libvirt-installation 2015-05-14 17:50:21 i more a less tried to follow that 2015-05-14 17:52:36 <_ikke_> sillysausage: No message in between 2015-05-14 17:54:38 i guess if i can't use virt-manager 2015-05-14 17:54:49 i could use virsh with root 2015-05-14 17:54:52 ie via sudo 2015-05-14 17:55:13 and i could still use virt-viewer (ie remote-viewer) to locally run the VM 2015-05-14 17:56:08 https://wiki.archlinux.org/index.php/Libvirt#Set_up_authentication i think that's probably why it wasn't working, not sure if Mo0O got it work though 2015-05-14 17:56:17 virt-manager wouldn't work as root via sudo either 2015-05-14 17:56:25 where as virsh did 2015-05-14 17:56:44 sillysausage: yes it works here 2015-05-14 17:56:50 how did you do it! 2015-05-14 17:56:55 what's your problem ? 2015-05-14 17:57:01 im getting that same error you got 2015-05-14 17:57:09 Error starting Virtual Machine Manager: 'module' object has no attribute 'VIR_DOMAIN_BLOCKED' 2015-05-14 17:57:28 ah, you need to update py-libvirt 2015-05-14 17:57:43 ACTION check his version 2015-05-14 17:58:05 $ apk search py-libvirt 2015-05-14 17:58:08 py-libvirt-glib-0.2.0-r0 2015-05-14 17:58:10 py-libvirt-1.2.15-r0 2015-05-14 17:58:18 py-libvirt-1.2.14-r0 2015-05-14 17:58:28 so you have an older version :P 2015-05-14 17:58:39 looks like ^^ 2015-05-14 17:58:47 what is weird is when i installed py-libvirt 2015-05-14 17:58:50 i saw this 2015-05-14 17:58:54 $ sudo apk add py-libvirt 2015-05-14 17:58:56 (1/2) Installing libaio (0.3.110-r0) 2015-05-14 17:58:58 (2/2) Installing qemu (2.1.2-r3) 2015-05-14 17:59:00 ERROR: qemu-2.1.2-r3: No error information 2015-05-14 17:59:02 OK: 784 MiB in 405 packages 2015-05-14 17:59:25 weird error 2015-05-14 18:00:04 did you install polkit and set that up? 2015-05-14 18:00:07 qemu-2.3.0-r0 here 2015-05-14 18:00:11 i think you need that for virt-manager 2015-05-14 18:00:15 ah 2015-05-14 18:00:27 I've on edge 2015-05-14 18:00:29 so your qemu is newer im just using what was in stable 2015-05-14 18:00:31 ah im not 2015-05-14 18:00:36 I'm* 2015-05-14 18:00:37 so maybe its broken in stable 2015-05-14 18:00:44 maybe 2015-05-14 18:00:54 I've made a patch for py-libvirt 2015-05-14 18:01:08 to update it to a version without this error 2015-05-14 18:01:16 https://help.ubuntu.com/community/KVM/Installation 2015-05-14 18:01:20 that was the other guide i used 2015-05-14 18:01:31 Mo0O: oh that's good could you submit a bug for that 2015-05-14 18:01:43 http://lists.alpinelinux.org/alpine-devel/4769.html 2015-05-14 18:01:47 this one 2015-05-14 18:02:16 it's already merged 2015-05-14 18:02:18 yeah seems to be the thing 2015-05-14 18:02:26 yeah but not in stable that's probably why im seeing it 2015-05-14 18:03:00 how do you connect to your VMs? 2015-05-14 18:03:04 do you use remote-viewer? 2015-05-14 18:03:08 localy 2015-05-14 18:03:16 that's the binary 2015-05-14 18:03:26 but the actual package is http://pkgs.alpinelinux.org/package/testing/x86/virt-viewer 2015-05-14 18:03:39 took me a while i was looking for a virt-viewer package 2015-05-14 18:06:03 oh looks like virt-manager can connect to graphical console 2015-05-14 18:06:36 guess i'll use that then 2015-05-14 18:07:01 yes there's one 2015-05-14 18:07:34 virt-manager seems very much like virtualbox that i'm used to 2015-05-14 18:08:43 yes, it's realy easy to use :) 2015-05-14 18:09:04 but i guess i will need that package either backported 2015-05-14 18:09:08 or ill try using the one from edge 2015-05-14 18:09:13 see if it works on my stable system 2015-05-14 18:10:16 Mo0O: so what are the base packages 2015-05-14 18:10:24 libvirt virt-manager libvirt-common-drivers 2015-05-14 18:10:47 and maybe bridge-utils 2015-05-14 18:10:55 that right Mo0O? 2015-05-14 18:11:21 not a lot of specific documentation for alpinelinux 2015-05-14 18:14:13 mmm yeah mixing edge with stable aint going work :) 2015-05-14 18:17:13 yes mixing edge with stable is realy dificult to maintain 2015-05-14 18:17:30 maybe i should move to edge i wonder how likely it is to break on me 2015-05-14 18:17:49 or how long before the edge becomes the stable 2015-05-14 18:18:43 it more package related, depending on the package's feedback 2015-05-14 18:19:12 yeah 2015-05-14 18:19:25 well this machine only is going to be running something pretty cut down 2015-05-14 18:19:33 i3, and a debian VM 2015-05-14 18:19:44 more for testing software on different distributions 2015-05-14 18:19:46 dwm here :) 2015-05-14 18:19:50 or dwm :P 2015-05-14 18:19:58 though ive been using my i3 config 2015-05-14 18:20:04 and 1MB vs 3 doesn 2015-05-14 18:20:10 doesn't make much difference :) 2015-05-14 18:20:20 yes ;) 2015-05-14 18:20:32 I was using i3 before 2015-05-14 18:20:49 my understanding was dwm required you to compile patches in for it to be actually useful 2015-05-14 18:20:51 but my fingers and my brain are more confortable with dwm 2015-05-14 18:20:54 and i'd rather not maintain that :P 2015-05-14 18:21:12 yes, I've made a lot of patch for my dwm 2015-05-14 18:21:46 I also use st, which works almost the same way 2015-05-14 18:21:56 st? 2015-05-14 18:21:57 suckless powa ^^ 2015-05-14 18:22:06 suckless terminal 2015-05-14 18:22:11 st 2015-05-14 18:22:14 oh i use urxvt-unicode 2015-05-14 18:22:28 i do use dmenu i like that. 2015-05-14 18:22:31 I was using it for a long time before too 2015-05-14 18:22:40 i used to use KDE some time ago 2015-05-14 18:22:44 I'm using dmenu too 2015-05-14 18:22:48 but im actually finding a window manager less of a hassle 2015-05-14 18:22:52 and easier to use 2015-05-14 18:23:04 i3's documentation made it really easy to learn the shortcuts 2015-05-14 18:23:05 ACTION remeber he needs to patch it to support xft fonts 2015-05-14 18:23:18 Mo0O: yeah that's why i didn't use dwm :P 2015-05-14 18:23:19 i did try it 2015-05-14 18:23:36 but i wouldn't want to use something like awesome that's bloated at 9MB of memory :D 2015-05-14 18:23:49 dwm support xft fonts, I was talking about dmenu ;) 2015-05-14 18:23:58 oh 2015-05-14 18:24:03 +1 for awesome 2015-05-14 18:24:13 dwm is realy lighter 2015-05-14 18:24:15 i tried awesome but didn't like it 2015-05-14 18:24:31 I was using for 2 years 2015-05-14 18:24:39 the multi monitor stuff was a big plus in i3 2015-05-14 18:24:49 didn't like how awesome handled that 2015-05-14 18:25:33 yeah, this is the only wm which manage it that way 2015-05-14 18:25:33 the main thing i like about i3 over awesome was being able to create windows in the way i wanted them 2015-05-14 18:25:38 i didn't like the presets thing 2015-05-14 18:26:19 I alway use the fullscreen layout :D 2015-05-14 18:26:22 so.. 2015-05-14 18:26:23 and once i got used to meta+h and meta+v 2015-05-14 18:26:32 yeah well i have a 30" or so screen 2015-05-14 18:26:44 so i have browser and term on workspace 1 usually 2015-05-14 18:26:44 huge 2015-05-14 18:26:49 mail and IM on workspace 2 2015-05-14 18:26:57 and something else on workspace 3 2015-05-14 18:27:07 usually more terminals 2015-05-14 18:27:15 sometimes i have terminals and a movie on a workspace 2015-05-14 18:27:56 the nifty thing i found about awesome is you could drag windows around 2015-05-14 18:28:01 gnu screen, ffx, others..., virt-manager at the end 2015-05-14 18:28:08 instead of using the keyboard, but i don't tend to move the windows once i've created them 2015-05-14 18:28:21 ewww screen 2015-05-14 18:28:26 i threw that out ages ago for tmux 2015-05-14 18:28:45 tmux on server, screen localy 2015-05-14 18:28:50 servers* 2015-05-14 18:29:04 why that way? 2015-05-14 18:29:42 no conflict short cut and I like the way screen manage splits 2015-05-14 18:29:53 hmm i wonder if it's fixed in 3.1.4 2015-05-14 18:30:04 probably not 2015-05-14 18:30:08 what? 2015-05-14 18:30:16 the virt-manager problems 2015-05-14 18:31:00 i wonder how long until alpine 3.2 is released 2015-05-14 18:31:11 that has support for rp2 2015-05-14 18:31:12 iduno 2015-05-14 18:32:21 well i might try edge 2015-05-14 18:33:09 oh looks like edge is in RC2 2015-05-14 18:33:19 yes, it is 2015-05-14 18:33:32 do RCs indicate something is almost ready for release? 2015-05-14 18:34:10 yeah rc is "Release Candidate" 2015-05-14 18:34:20 looks like ill try that out then :D 2015-05-14 18:34:27 basically it's "ready to release unless something breaks" 2015-05-14 18:36:29 i actually am really liking alpinelinux 2015-05-14 18:36:34 there's really nothing quite like it 2015-05-14 18:36:37 except for tiny core 2015-05-14 18:37:05 what i did notice is the RP2 builds aren't in there http://nl.alpinelinux.org/alpine/v3.2/releases/x86_64/ 2015-05-14 18:39:42 sillysausage: I now remember that you need ebtables and dnsmasq to fix the `virNetworkCreate libvirtError: out of memory` ;) 2015-05-14 18:39:46 just in case 2015-05-14 18:39:57 oh 2015-05-14 18:40:08 do i just need them or do i have to configure them in some way? 2015-05-14 18:40:27 no configuration required, just install them :) 2015-05-14 18:40:32 kiss way 2015-05-14 18:40:54 ebtables and dnsmasq for the default NAT/DHCP networking. 2015-05-14 18:40:59 ah yes i saw that in the arch wiki 2015-05-14 18:44:50 sillysausage: so, you're on edge now? 2015-05-14 18:45:05 not yet downloading it still 2015-05-14 18:45:25 ok, let me know if you have some issues 2015-05-14 18:45:29 ACTION afk 2015-05-14 18:58:46 i'm having some horrible error on upgrade vom 3.0 to 3.1: 2015-05-14 18:58:51 Why is mysql a dummy package? What am I supposed to use if I need mysql? I still see many wiki articles referencing mysql 2015-05-14 18:58:52 extlinux: no previous syslinux boot sector found 2015-05-14 18:59:59 my mysql server has those installed: 2015-05-14 19:00:11 http://hastebin.com/elerifayay.hs 2015-05-14 19:02:23 Okay thanks, I think that was it. Some of the wiki articles might need to be updated, I'll look at them later 2015-05-14 19:10:33 Half the time I think I just get confused because I get such different results on the package search page as compared to apk search. Really need to stick to apk search 2015-05-14 19:11:02 Like, I didn't see mariadb before I came here, somehow. I would have just used that to start with.. 2015-05-14 19:13:14 :/ i had been asking for the package search a long time, didn't even yet have time to try it out 2015-05-14 19:15:54 I don't know if it's just me but it seems to often say give less or no packages, when there are actually packages. 2015-05-14 19:16:04 What's that vom that you are having problems with? 2015-05-14 19:17:01 ah i found it already, that was during update - my /boot wasn't mounted and the device name in /etc/fstab was wrong 2015-05-14 19:17:32 now i'm looking stupidly since xen says it's version 4.3.3 but the update said i was going to find 4.4.x installed 2015-05-14 19:17:45 i'm not feeling smart today, at all 2015-05-14 19:19:47 I know what you mean, I was definately there last night. I just set the keyboard down then came back today and the immediately found the solution to the problem I had been stuck on (just some stupid file permissions thing) 2015-05-14 19:20:25 with that non-mounted /boot i got the xen4 in /boot (on root) and 4.3 in /boot (partition) 2015-05-14 19:20:37 i need to have a talk with my monitoring 2015-05-14 19:21:03 but, first time in months i do some update at home, no suprise there is a little fallout 2015-05-14 19:24:47 iwearsable: on the good side, my alpine db server is running so absolutely smooth, i can recommend that setup a lot 2015-05-14 19:25:02 just having recent versions of software is worth a lot 2015-05-14 19:27:30 xen_version : 4.4.2 2015-05-14 19:27:32 yes! 2015-05-14 19:27:47 I'm going to try Mariadb, only because that is what I was using last (before moving to alpine). It's just a LEMP server for Wordpress. I'm going to try using edge for this server (it's not really essential) 2015-05-14 19:32:02 bcache is weird 2015-05-14 19:32:31 it gave a invalid argument when i re-registered all disks and then it works 2015-05-14 19:33:02 gonna buy the cachecade upgrade for my controller as soon as i find a cheap one 2015-05-14 19:33:07 sick of software ssd caching 2015-05-14 19:42:51 [ 303.410039] EXT4-fs (xvda3): error count: 16 2015-05-14 19:42:53 fuck 2015-05-14 19:42:56 this 2015-05-14 19:42:58 bcache 2015-05-14 19:42:59 crap 2015-05-14 20:24:50 hmm i'm on edge now, seems that my paackages aren't all installing 2015-05-14 20:24:52 if i go apk update 2015-05-14 20:24:57 and apk upgrade it pulls them in 2015-05-14 20:24:58 from cache 2015-05-14 20:25:05 i forgot why that occurs 2015-05-14 20:25:29 i tried doing lbu commit -d afterwards, but next time i reboot it didn't install the packages 2015-05-14 20:26:34 i also noticed when i start lxdm 2015-05-14 20:26:40 i lose the mouse/keyboard 2015-05-14 20:32:14 oh wait wasn't it that USB dev 2015-05-14 20:32:16 thing 2015-05-14 20:32:47 y eah alpine_dev=UUID=XXXXXX-XXXXXX 2015-05-14 20:32:52 need to change that to usbdev don't i 2015-05-14 20:33:15 no idea :/ 2015-05-14 20:33:25 oh its when you use setup-bootable 2015-05-14 20:35:43 okay, so http://wiki.alpinelinux.org/wiki/Edge explains how to upgrade to edge - so, when done the system would be running edge/main right? 2015-05-14 20:36:15 i just booted the newest image in a vm and used setup bootable on it 2015-05-14 20:36:18 then copied my apkprov 2015-05-14 20:37:01 and its working now :) 2015-05-14 20:37:11 can I have an edge/main system with few selected packages from edge/testing? 2015-05-14 20:37:17 also Mo0O virt-manager works yay! 2015-05-14 20:37:25 sasoiliev: yes you can do pinning 2015-05-14 20:37:27 but 2015-05-14 20:37:29 it often does not work it depends 2015-05-14 20:37:34 sometimes you have to go all edge 2015-05-14 20:37:46 i'd say edge is probably close enough RC2 it's going to go stable soon 2015-05-14 20:38:08 you can often use packages from @testing 2015-05-14 20:38:21 like for example i3 and i3status i found worked nicely 2015-05-14 20:38:39 but using claws from @edge or virt-manager wasn't possible as it had too many dependencies 2015-05-14 20:40:11 right, thanks, sillysausage 2015-05-14 20:41:04 okay, another one - how does a package get promoted from @testing to @edge? 2015-05-14 20:41:27 I mean - for packages that are only in @testing 2015-05-14 20:41:32 other way around i think 2015-05-14 20:41:36 edge to testing to stable 2015-05-14 20:42:40 hm, looking at http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management#Repository_pinning it looks to me like @edge (edge/main) is more stable than @testing (edge/testing) 2015-05-14 20:42:57 edge is just the next release 2015-05-14 20:43:08 testing is where packages go when they are being tested 2015-05-14 20:43:24 in this case edge is very stable because its just about to be released i think RC2 2015-05-14 21:28:46 critical bug in MariaDB 5.5.43 & 10.0.18 2015-05-14 21:29:43 <_ikke_> are new versions released? 2015-05-14 21:56:00 _ikke_ there is a 10.0.19 so yes 2015-05-14 21:56:47 http://hastebin.com/icogewosob.vhdl for details 2015-05-14 21:59:21 <_ikke_> Ok, so happens either when upgrading, or sswitching products 2015-05-14 23:39:13 Hey I was wondering if someone could help me? just installed alpine on a pc 2015-05-14 23:41:31 never mind i figured it out :P 2015-05-14 23:51:27 hello there, I have the lsi megaraid manager software, try to stat it says no such file but it's there, strings say it has glibc 2015-05-14 23:51:31 no way to run it? 2015-05-14 23:52:02 not yet. working to reverse engineer these cards and release an open tool soon :) 2015-05-14 23:52:38 well I don't play much to it but it has some cool alerting features 2015-05-14 23:53:13 it's the only thing I am missing in my small box 2015-05-14 23:55:06 you could probably run it from a chroot of debian 2015-05-14 23:55:22 hmz, actually 2015-05-14 23:55:33 chroot of an bootstrap should do it 2015-05-14 23:57:37 but that will beat the mini distro purpose 2015-05-14 23:58:19 unfortunately that's all that i can say right now 2015-05-14 23:58:37 i do not yet have a spare 9260 series card 2015-05-15 00:04:58 alright thank you 2015-05-15 01:38:13 hey hey 2015-05-15 01:51:06 hi 2015-05-15 02:04:15 what u guys up to 2015-05-15 02:04:24 i just set up my new laptop 2015-05-15 02:04:33 i'm going back and forth between sabotage and alpine heh 2015-05-15 02:04:49 they are on two 128gig partitions 2015-05-15 02:05:30 (both are musl-based distros that are fiercly independent haha) 2015-05-15 02:10:12 <__number5__> sabotage? 2015-05-15 02:11:07 <__number5__> nvm, found it 2015-05-15 02:16:30 yeah 2015-05-15 02:16:38 it's got alot in common with alpine 2015-05-15 02:16:46 so much so i think the binaries might be cross-compatible 2015-05-15 02:16:53 (gotta try that heh) 2015-05-15 02:17:30 well anyway i was like 95% of the way into alpine as my laptop distro, cuz building from source on sabotage was getting boring, and i couldnt get X working 2015-05-15 02:17:37 i spent like 3 days on it in alpine 2015-05-15 02:18:24 anyway i switch back to sabotage since X was working in that... 2015-05-15 02:18:51 i've been waiting like an hour for QT5 to compile haha 2015-05-15 02:18:59 no wonder ncopa has a build server 2015-05-15 02:19:40 btw 2015-05-15 02:19:51 i'm running kernel 4.1 (latest) 2015-05-15 02:20:05 i switched out of desperation to get my i915 driver stable 2015-05-15 02:20:15 kernel 4.1 is elite 2015-05-15 02:20:32 ive got it running with f2fs on a 512gb ssd 2015-05-15 02:20:48 f2fs is a solid state hard drive filesystem... it's so awesome 2015-05-15 02:21:12 besides being fast as hell... .f2fs is smart enuf to like manage the wear on your ssd 2015-05-15 02:21:28 <__number5__> hmm, I have f2fs on my nexus 7... 2015-05-15 02:21:28 it deletes stuff in the erase block size only (usually about 0.5MB to 16MB) 2015-05-15 02:21:49 and it writes stuff in the segment size (usually 4kb up to 32kb) 2015-05-15 02:22:12 anyway it prevents solid-state hd fragmentation 2015-05-15 02:22:51 i was skeptical but it's really really stable so far.... and it's at least 20% to 50% faster than ext4 and xfs dependening on the benchmark 2015-05-15 02:27:06 bitshark, what video hardware? 2015-05-15 02:27:10 intel? 2015-05-15 02:27:18 it's been broken for a long time :( 2015-05-15 02:27:28 i have it pinned to the version from 3.0 which works 2015-05-15 02:28:03 ncopa, everyone seems to be experiencing broken intel video drivers :( 2015-05-15 02:28:31 i don't think i have the time to spend tracking it down, but i know the regression took place soon after 3.0 2015-05-15 02:28:52 the version from 3.0 worked, and edge broke somewhere before 3.1 2015-05-15 02:31:32 i really doubt that sabotage is abi compatible with alpine 2015-05-15 02:31:37 we have some modifications 2015-05-15 02:31:43 oh sorry 2015-05-15 02:31:45 afk dalias 2015-05-15 02:31:57 intel i915 driver based hardware... i think pci was like 00:02 2015-05-15 02:32:18 specifically one of their integrated video driver series... this might have been intel video hd4000 series or sumt 2015-05-15 02:32:36 i sort of found the issue (somewhat)... there's more than one issue 2015-05-15 02:33:25 (1) i915 has been buggy for years, on and off depending on the kernel release (some are fine, some are broken) ... and it also depends on the hardware combination 2015-05-15 02:34:14 (2) i915 bugs can include failz involving interrupts and hotplugging so broad scope 2015-05-15 02:35:02 (3) linux changed structure of framebuffer driver around since about kernel 2.6 thru kernel 3.19 ... the details im not sure... but u need a /dev/fb0 even for the intel-specific X driver 2015-05-15 02:35:57 so the first two i resolved by upgrading to non-grsec kernel 4.1 and recompiling it with my own optionz (unfortunately not on alpine, on sabotage... but i bet the fixes would work if i boot into alpine) 2015-05-15 02:37:03 and for the framebuffer thing for my specific video combination ... well i seemed to have fixed that by enabling all the intel framebuffer options as Y , and enabling all other intel video stuff as M 2015-05-15 02:52:40 do you know what's the prupose of the pacman package? I know that it's the archlinux package manager, but why it's disponible on alpine? 2015-05-15 02:59:33 <__number5__> maybe it's pacman the game? 2015-05-15 02:59:38 <__number5__> #justkidding 2015-05-15 03:01:10 __number5__: looks not http://pkgs.alpinelinux.org/package/main/x86_64/pacman 2015-05-15 03:01:12 ;) 2015-05-15 03:12:10 kaniini, ? 2015-05-15 03:12:19 what would be abi incompatible vs sabotage? 2015-05-15 03:16:58 pacman is abi incompatible 2015-05-15 03:17:00 for sure 2015-05-15 03:17:10 the binaries would be linked against glibc , right? 2015-05-15 03:17:26 but from what i hear , arch has great build-from-source type instructions 2015-05-15 03:17:34 so u could use pacman to expand the package universe 2015-05-15 03:17:35 Mo0O: so you can install pacman if you want! 2015-05-15 03:17:44 well the ones from arch binary repo would be 2015-05-15 03:17:46 IF u dont use arch binary repos 2015-05-15 03:17:52 that will prob fail 2015-05-15 03:17:56 but presumably you could use pacman without that, i guess 2015-05-15 03:17:59 yeah 2015-05-15 03:18:01 you obviously wouldn't use it with the arch binary repo... 2015-05-15 03:18:06 that's what i was gonna do acutally 2015-05-15 03:18:09 here on sabotage linux 2015-05-15 03:18:10 it's really handy to have pacman around if you want to make an arch linux container 2015-05-15 03:18:18 install pacman and use it for source 2015-05-15 03:18:38 b/c the sabotage package universe is only like 679 source packages 2015-05-15 03:18:45 but pacman is way bigger 2015-05-15 03:18:56 sounds like funn heh 2015-05-15 03:24:10 ahills: yes, but for doing what? :D 2015-05-15 03:24:42 ah ok 2015-05-15 03:28:47 dalias: hmm, it appears it should work 2015-05-15 03:28:57 we just add additional symbols 2015-05-15 03:29:02 __stack_chk_fail_local 2015-05-15 03:35:06 __stack_chk_fail_local is hidden, not part of dynamic ABI, and it's from gcc anyway 2015-05-15 03:35:33 at present my understanding is that alpine's musl libc.so doesn't have any symbols not in upstream 2015-05-15 03:35:40 oh, that's good 2015-05-15 03:35:52 i know in the 1.0 era we had our own getopt_long() and some other things 2015-05-15 03:39:35 :) 2015-05-15 03:39:50 well getopt_long wasn't a new symbol just a replacement impl 2015-05-15 03:39:55 right 2015-05-15 03:39:56 but yes there were new symbols 2015-05-15 03:40:09 they were all eventually upstreamed in one form or another 2015-05-15 03:40:36 that's good, because fabled and i were stuck with a giant mess with uclibc ;) 2015-05-15 03:52:00 i can imagine :) 2015-05-15 06:05:46 hey 2015-05-15 06:05:56 how do i download an alpine package without apk 2015-05-15 06:06:03 like here http://pkgs.alpinelinux.org/package/main/x86_64/coreutils 2015-05-15 06:33:07 bitshark, wget from mirror 2015-05-15 07:12:46 so Mo0O now that virt-manager is working 2015-05-15 07:12:53 did you use policykit at all? 2015-05-15 07:12:59 sudo virsh -c qemu:///system list 2015-05-15 07:13:04 for example only works when i use sudo 2015-05-15 07:13:15 otherwise i get an error like: error: failed to connect to the hypervisor 2015-05-15 07:13:17 error: authentication failed: no agent is available to authenticate 2015-05-15 07:13:35 now i'm using i3wm with lxdm, so i was thinking lxpolkit? 2015-05-15 07:14:36 which is why virt-manager gives me something like: 2015-05-15 07:14:38 Unable to connect to libvirt. 2015-05-15 07:14:40 authentication failed: no agent is available to authenticate 2015-05-15 07:14:42 Libvirt URI is: qemu:///system 2015-05-15 07:14:52 it is running, and i didn't want to run virt-manager as root 2015-05-15 07:15:05 uid=1000(user) gid=1000(user) groups=18(audio),34(kvm),101(messagebus),102(libvirt),1000(user) 2015-05-15 07:15:10 my user is actually a part of all these groups 2015-05-15 07:17:24 i suspect the problem is lxdm isn't creating a lxpolkit agent 2015-05-15 07:17:28 im wondering how i make it do that 2015-05-15 07:18:30 https://bugs.mageia.org/show_bug.cgi?id=11956 2015-05-15 07:20:48 user ~ $ pkexec virt-manager 2015-05-15 07:20:50 ==== AUTHENTICATING FOR org.freedesktop.policykit.exec === 2015-05-15 07:20:52 Authentication is needed to run `/usr/bin/virt-manager' as the super user 2015-05-15 07:20:54 Authenticating as: root 2015-05-15 07:20:56 password: 2015-05-15 07:20:58 ==== AUTHENTICATION COMPLETE === 2015-05-15 07:21:00 user ~ $ 2015-05-15 07:21:02 (virt-manager:2382): Gtk-CRITICAL **: gtk_settings_get_for_screen: assertion 'GDK_IS_SCREEN (screen)' failed 2015-05-15 07:21:04 so the problem might be there that we don't have a policy kit file for it 2015-05-15 07:21:06 maybe we can steal one from another distribution 2015-05-15 07:25:51 dalias: I use intel video driver on my desktop and laptop 2015-05-15 07:26:54 and i have never had any issues 2015-05-15 07:27:06 i think it might be problem with some intel chipsets 2015-05-15 07:27:20 since i dont have those, i cannot really track it down either :-/ 2015-05-15 07:32:28 <_ikke_> Trying to port puppet to alpine, but running in some issue due to useradd not being available / compattible. Any ideas if this can be solved? 2015-05-15 07:37:35 ln -s adduser useradd :) 2015-05-15 07:38:26 rnalrd: hi 2015-05-15 07:38:43 hi 2015-05-15 07:39:00 rnalrd: you're maintaining owncloud right? :) 2015-05-15 07:39:08 looks like so :) 2015-05-15 07:39:41 cool. have you had any luck building the owncloud sync client ? 2015-05-15 07:40:05 never attempted 2015-05-15 07:41:07 oops did i miss anything 2015-05-15 07:41:21 it would be cool to be able to sync the webdav in owncloud with alpine linux. if you have a moment, could you give it a go? 2015-05-15 07:41:30 k 2015-05-15 07:41:44 could you please open a package request on bug tracker? 2015-05-15 07:41:55 so that i don't forget :) 2015-05-15 07:42:15 certainly. it's called owncloudcmd 2015-05-15 07:42:51 also if you have work-in-progress APKBUILD you post that as well 2015-05-15 07:42:53 thanks! 2015-05-15 07:43:04 you may post* 2015-05-15 08:12:22 rnalrd: done, and assigned to you : 2015-05-15 08:12:26 http://bugs.alpinelinux.org/issues/4189 2015-05-15 08:12:31 tnx 2015-05-15 08:40:52 hey guys 2015-05-15 08:40:54 ne1 around 2015-05-15 08:41:18 i'm doing a similar task... tryin to get vmware working on musl/busybox distro 2015-05-15 08:41:35 might have to ditch it for kvm cuz vmware wont install 2015-05-15 08:44:26 btw why use virt-manager over sumt else? 2015-05-15 08:44:29 <_ikke_> Interesting problem, need to make a VMDK image bootable from usb. I've already managed to convert it to a raw img file, but now running into problems getting the USB media bootable. Is it even possible to do it this way? 2015-05-15 08:44:32 virt-manager == redhat 2015-05-15 08:44:36 yuck 2015-05-15 08:44:55 im thinking like this for a kemu gui 2015-05-15 08:44:56 http://sourceforge.net/projects/aqemu/ 2015-05-15 08:45:27 is virt-manager just so awesome ppl look past the redhat part 2015-05-15 08:45:47 ikke: YES! 2015-05-15 08:45:56 ikke: I did that exact task over past 2 days 2015-05-15 08:46:04 <_ikke_> bitshark: Oh, cool! 2015-05-15 08:46:09 to make customized linux frmo scratch on USB key 2015-05-15 08:46:21 different varients of lfs 2015-05-15 08:46:31 <_ikke_> It is an alpine system 2015-05-15 08:46:34 i built it in a vm, then moved it to usb and made it bootable 2015-05-15 08:46:38 <_ikke_> ok 2015-05-15 08:46:41 on bare metal 2015-05-15 08:46:45 <_ikke_> yeah, that's the goal 2015-05-15 08:46:48 okay so what vm u using 2015-05-15 08:46:57 virtualbox? 2015-05-15 08:47:01 <_ikke_> It came from vmware 2015-05-15 08:47:05 <_ikke_> I have virtualbox 2015-05-15 08:47:07 is i like virt-manaer 2015-05-15 08:47:09 yeah okay 2015-05-15 08:47:10 *virt-manager 2015-05-15 08:47:18 i would like to know though how to not run it as root 2015-05-15 08:47:18 do u want to customize it? 2015-05-15 08:47:23 i think it needs a polkit agent 2015-05-15 08:47:31 silly: i never used virt-manager i heard good things 2015-05-15 08:47:34 running, and it seems lxdm isn't starting lxpolkit 2015-05-15 08:47:41 silly: i'll give it a try 2015-05-15 08:47:51 oh i have lxde 2015-05-15 08:47:56 and xfce tho 2015-05-15 08:48:00 yeah virt-manager can do qemu, kvm, bhyve, kvm, lxc, xen 2015-05-15 08:48:01 but i prefer moving to qt5 2015-05-15 08:48:06 i like the idea of having the one interface 2015-05-15 08:48:09 oh lxC? 2015-05-15 08:48:11 that's elite 2015-05-15 08:48:13 didnt know that 2015-05-15 08:48:15 <_ikke_> bitshark: What I've done now is to create 2 partitions on the usb drive, and copy the boot and root filesystem 2015-05-15 08:48:18 now i got to install it! 2015-05-15 08:48:22 <_ikke_> the boot partition is made bootable 2015-05-15 08:48:22 thx sillysausage haha 2015-05-15 08:48:26 okay ikke 2015-05-15 08:48:38 the only problem i am having with it is connecting to a local host 2015-05-15 08:48:40 whoa okay confused ikke 2015-05-15 08:48:42 wants to run as root 2015-05-15 08:48:51 even though im in the libvirtd group 2015-05-15 08:48:54 and kvm group 2015-05-15 08:49:03 pretty sure its because i don't have a polkit agent 2015-05-15 08:49:25 bitshark: is your kvm host remote? 2015-05-15 08:59:55 hmm so i manually started polkit ie 2015-05-15 09:00:02 : /usr/libexec/lxpolkit & 2015-05-15 09:00:11 and now it asks for the root password 2015-05-15 09:01:01 ScrumpyJack, i have owncloud-client built locally against qt5 2015-05-15 09:01:14 is that okay for you or need qt4? 2015-05-15 09:02:16 Unable to connect to libvirt. 2015-05-15 09:02:18 internal error: Cannot find suitable emulator for x86_64 2015-05-15 09:02:20 i wonder why that is 2015-05-15 09:03:12 i made sure qemu-x86_64 was installed 2015-05-15 09:10:16 I use qemu-system-x86_64 to launch KVM on AL... 2015-05-15 09:11:27 yeah i was just trying to figure out why virt-manager was complaining 2015-05-15 09:48:17 rnalrd: qt5 is fine 2015-05-15 09:48:49 are the deps huge? 2015-05-15 09:49:15 (i am qt free at the moment) 2015-05-15 09:49:43 dunno 2015-05-15 09:49:46 i haven't tested it 2015-05-15 09:49:54 but it's on mirror already 2015-05-15 09:50:03 what is the best way to start lxpolkit on boot? ubuntu had a rc service iirc 2015-05-15 09:50:06 pls feel free to report anything broken 2015-05-15 09:51:07 as in: /usr/libexec/lxpolkit & 2015-05-15 09:51:13 there must be a correct way of doing it 2015-05-15 09:51:27 i thought it was lxdm that could do it 2015-05-15 09:51:30 it hasn't hit my mirror yet 2015-05-15 09:51:42 sorry i meant primary mirror 2015-05-15 09:51:46 nl.a.o 2015-05-15 09:52:14 i mirror that too (liskamm.alpinelinux.uk) 2015-05-15 09:52:18 within ~1h should hit all other mirrors 2015-05-15 09:52:41 Set lxpolkit as default policy agent 2015-05-15 09:52:43 In the menu, go to Preferences > Default applications for LXSession (or run lxsession-default-apps in a terminal). In the first section "Running applications" (Update: Default apps manager 14.10 is different. Use the Core applications tab.), go to the option for Polkit agent and make sure lxpolkit is selected. 2015-05-15 09:52:48 i wonder if that applies to AL 2015-05-15 10:02:04 okay back 2015-05-15 10:02:30 man qt is elite 2015-05-15 10:02:35 way better than gtk 2015-05-15 10:02:41 qt5 especially 2015-05-15 10:16:38 anyone know of "independant" clock system tray app and volume control system tray app, written in C 2015-05-15 11:05:22 hmm maybe ill stop using lxdm and try lightdm 2015-05-15 11:06:23 as i can't seem to get lxdm to run lxpolkit 2015-05-15 11:08:55 in any case Mo0O that seems minor to resolving this error i was having earlier: 2015-05-15 11:08:58 internal error: Cannot find suitable emulator for x86_64 2015-05-15 11:09:14 i have qemu-x86_64-2.3.0-r0 2015-05-15 11:14:35 rnalrd: owncloud seems to work. the system tray icon doesn't seen to support transparancy though 2015-05-15 11:15:29 and the monochrome icons are blank 2015-05-15 11:17:16 ScrumpyJack, i'm pretty clueless about graphics bugs 2015-05-15 11:18:35 are you using xcfe? 2015-05-15 11:20:30 no 2015-05-15 11:20:43 twm :) 2015-05-15 11:24:58 would be nice to test it with other desktop env 2015-05-15 11:25:43 you know what is missing 2015-05-15 11:25:44 compton 2015-05-15 11:26:33 i should package that some time 2015-05-15 11:29:52 i see i missed discussion on virt-manager 2015-05-15 11:29:56 i use virtmanager here 2015-05-15 11:29:59 works for me 2015-05-15 11:30:07 its a bit memory consuming though 2015-05-15 11:30:31 eats almost a gig of memory, before you start any guest 2015-05-15 11:30:53 i pushed qem-gtk recently 2015-05-15 11:31:07 which is qemu with minimalistic gtk interface 2015-05-15 11:32:24 ah 2015-05-15 11:32:34 ncopa: what DE do you use? 2015-05-15 11:32:41 basically ive got lxdm running i3wm 2015-05-15 11:32:53 but the problem i was having was getting lxpolkit to run 2015-05-15 11:33:04 without having to manually type "lxpolkit &" 2015-05-15 11:33:06 every time 2015-05-15 11:33:08 i boot 2015-05-15 11:33:38 i use lxdm + xfce 2015-05-15 11:34:32 there is an /etc/xdg/autostart/lxpolkit.desktop 2015-05-15 11:34:49 i think xfce4 autostarts that for me 2015-05-15 11:35:55 this might be the case 2015-05-15 11:36:09 i do have that btw 2015-05-15 11:36:51 actually 2015-05-15 11:36:58 i should probably have polkit in my i3 config 2015-05-15 11:37:01 eg 2015-05-15 11:37:05 #exec --no-startup-id /usr/lib/fingerprint-gui/fingerprint-polkit-agent -d 2015-05-15 11:39:09 https://archeando.wordpress.com/2014/05/27/tiling-window-manager-i3/ 2015-05-15 11:43:18 i'll try that 2015-05-15 12:04:11 i wonder if it's more appropriate that it be in .xinitrc 2015-05-15 12:04:16 https://github.com/mschristiansen/dotfiles/blob/master/.xinitrc 2015-05-15 12:16:45 <_ikke_> Aparently I'm dealing with a really old alpine installation (comparing packages makes me suspect it's v1.9). 2015-05-15 12:17:10 <_ikke_> /etc/alpine-release is a broken symlink 2015-05-15 12:17:40 <_ikke_> I need to be able to add vlan + bonding, but cannot find those packages for that version 2015-05-15 12:22:53 <_ikke_> aparently it's done through modprobe 2015-05-15 12:26:17 _ikke_: whats via modprobe? 2015-05-15 12:26:29 i know the bonding mode has to be done via modprobe to be certain 2015-05-15 12:26:41 especially if you have multiple bonds running different modes 2015-05-15 12:26:54 <_ikke_> No, just one bond 2015-05-15 12:27:13 <_ikke_> We usually do it just via /etc/network/interfaces 2015-05-15 12:28:32 i only started in 2.x, but modprobe is pretty safe anyway 2015-05-15 12:30:46 <_ikke_> ok 2015-05-15 12:30:54 <_ikke_> do you also know if vlan tagging can be done somehow? 2015-05-15 12:39:26 _ikke_: i dont remember if we had vconfig back then 2015-05-15 12:40:11 i dont think busybox had it atleast 2015-05-15 12:41:47 _ikke_: do you have vconfig there? 2015-05-15 12:42:51 _ikke_: yes 2015-05-15 12:43:08 i only do it manually though lol 2015-05-15 12:43:12 have a script 2015-05-15 12:43:21 The Unix Way 2015-05-15 12:44:20 <_ikke_> ncopa: Note sure 2015-05-15 12:44:49 <_ikke_> ncopa: I don't see it in the repository 2015-05-15 12:44:50 i dont remember which busybox version it was added 2015-05-15 12:44:56 as a busybox applet 2015-05-15 12:44:59 <_ikke_> ok 2015-05-15 12:45:05 <_ikke_> I have the filesystem here 2015-05-15 12:45:05 $ vconfig 2015-05-15 12:45:05 BusyBox v1.23.2 (2015-03-23 14:45:58 GMT) multi-call binary. 2015-05-15 12:45:05 Usage: vconfig COMMAND [OPTIONS] 2015-05-15 12:45:07 ... 2015-05-15 12:45:19 <_ikke_> It's at least not present in /usr/bin 2015-05-15 12:45:35 in /sbin/vconfig? 2015-05-15 12:45:44 $ which vconfig 2015-05-15 12:45:45 <_ikke_> yes, it's there 2015-05-15 12:45:47 kaniini: re: uclibc, was the giant mess maintaining a compatibility layer because upstream wouldn't accept patches, or what? 2015-05-15 12:46:34 ahills: uclibc was a giant mess 2015-05-15 12:46:51 regardless what you tried to do with the code... 2015-05-15 12:46:53 Ah, bummer 2015-05-15 12:47:16 A little too ambitious for its authors, perhaps? 2015-05-15 12:47:16 they almost never did releases 2015-05-15 12:47:31 people did big jobs 2015-05-15 12:47:41 <_ikke_> ncopa: How would I persist the config if I'd use vconfig? 2015-05-15 12:47:48 which to years to get merged into mainline uclibc tree 2015-05-15 12:48:07 pre-up in /etc/network/interfaces 2015-05-15 12:48:11 <_ikke_> ok 2015-05-15 12:48:21 ncopa: is that why AL went with musl? 2015-05-15 12:48:23 pre-up vconfig .... 2015-05-15 12:48:24 Haha, I love the modern fly-by-wire no-release git-based communities we have today, like in the webdev world 2015-05-15 12:49:09 ahills: i wouldnt say uclibc was 'modern' 2015-05-15 12:49:43 i think the maintainer was busy, and didnt really want do maintenance of uclibc 2015-05-15 12:49:54 and the uclibc code is horribly messy 2015-05-15 12:49:55 It's just a pattern I see everywhere and it makes me chuckle, as a systems programmer 2015-05-15 12:50:16 sillysausage: yes, musl code was *way* cleaner 2015-05-15 12:50:17 Messier than glibc? 2015-05-15 12:50:44 much of uclibc is copied directly from glibc 2015-05-15 12:51:02 oh so might as well use glibc then :P 2015-05-15 12:51:16 so yeah, the result was a bigger mess than glibc 2015-05-15 12:51:28 uclibc did strip out lots of things 2015-05-15 12:51:42 so it was same mess as glibc, just smaller 2015-05-15 12:52:12 i was scared for lacking of gnu extensions when migrating to musl 2015-05-15 12:52:26 but the move went much smoother than feared 2015-05-15 12:52:49 and many bugs that was "unfixable" in uclibc, just went away 2015-05-15 12:53:44 switching to musl was was of the better things we ever did :) 2015-05-15 12:58:05 ahills: not the way i like it; basically /etc/network/interfaces only does the phys. intefaces and bond and the host's IP 2015-05-15 12:58:36 I don't want to have 100 of vlan configs in the same process, all likely to break main IP config 2015-05-15 12:59:02 one of two reasons that I'm eyeing openvswitch 2015-05-15 13:00:21 lol 2015-05-15 13:00:54 my vlan configs are the result of extensive whiteboarding, so I don't view each one as likely to break the main config :P 2015-05-15 13:01:01 hehe 2015-05-15 13:01:06 I wonder how long until my arrogance catches up to me 2015-05-15 13:01:15 nah that's sensible somewhat 2015-05-15 13:01:20 you should do what I accidentally did to solve my container problem 2015-05-15 13:01:30 and use python's shlex library to write a mini-language 2015-05-15 13:02:53 i just have one very beefy box at home to use and test and play and have labs on 2015-05-15 13:03:02 so vlans come and go and get forgotten all the time 2015-05-15 13:11:57 nice 2015-05-15 13:55:34 hmm this polkit thing is so annoying 2015-05-15 13:56:04 exec --no-startup-id i3-msg '/usr/libexec/lxpolkit' 2015-05-15 13:56:08 well that doesn't work at all 2015-05-15 14:01:15 why do you need polkit? 2015-05-15 14:01:31 i needed it so that i could use virt-manager not as root 2015-05-15 14:01:46 <_ikke_> I'm trying to setup bonding on alpine 1.9 by using /sys/class/net/bond0/bonding/.. 2015-05-15 14:01:47 and it does seem to fix it when lxpolkit is running, but i've had some difficulty running it automatically 2015-05-15 14:02:06 <_ikke_> But after I set it up, I don't get any response from the gateway 2015-05-15 14:02:26 <_ikke_> If I look at dmesg, I see a lot of martian warning from a different submnet 2015-05-15 14:02:28 <_ikke_> subnet 2015-05-15 14:02:37 <_ikke_> (I'm using bonding and vlan tagging) 2015-05-15 14:03:57 i might also try qemu-gtk 2015-05-15 14:04:02 seeing as ncopa recommends it 2015-05-15 14:04:42 _ikke_: do you have multiple virtual interfaces behind (eg lxc or virtual) 2015-05-15 14:04:54 you might want try set bonding mode to failover mode 2015-05-15 14:04:55 <_ikke_> ncopa: they're blades 2015-05-15 14:05:03 <_ikke_> using active-backup bonding mode 2015-05-15 14:05:09 that should work 2015-05-15 14:06:02 <_ikke_> So I modprobe bonding and 8012q 2015-05-15 14:06:17 <_ikke_> I add eth0 and eth1 as slaves to bond0 2015-05-15 14:06:51 <_ikke_> I set bond mode to active backup, and miimon to 100 2015-05-15 14:07:24 <_ikke_> then ip link add link bond0 name bond0.123 type vlan id 123 2015-05-15 14:07:38 <_ikke_> I bring up bond0 + bond0.123, assign an ip address 2015-05-15 14:07:47 <_ikke_> Then when I try to ping, I don't get a response 2015-05-15 14:10:38 okay that's weird 2015-05-15 14:10:46 May 15 14:08:51 palanthas kern.info kernel: [ 42.024805] lxpolkit[1923]: segfault at 8 ip 00000c812d4f22c5 sp 00007a4690ab56c0 error 4 in lxpolkit[c812d4f0000+4000] 2015-05-15 14:10:47 May 15 14:08:51 palanthas kern.alert kernel: [ 42.024829] grsec: Segmentation fault occurred at 0000000000000008 in /usr/libexec/lxpolkit[lxpolkit:1923] uid/euid:1000/1000 gid/egid:1000/1000, parent /bin/busybox[sh:1920] uid/euid:1000/1000 gid/egid:1000/1000 2015-05-15 14:10:49 May 15 14:08:51 palanthas kern.alert kernel: [ 42.024859] grsec: denied resource overstep by requesting 4096 for RLIMIT_CORE against limit 0 for /usr/libexec/lxpolkit[lxpolkit:1923] uid/euid:1000/1000 gid/egid:1000/1000, parent /bin/busybox[sh:1920] uid/euid:1000/1000 gi 2015-05-15 14:10:53 i think that's why polkit isn't working, it crashes 2015-05-15 14:11:04 but it works if i call it from urxvt once i3 has started 2015-05-15 14:12:05 maybe i should try mate's polkit 2015-05-15 14:12:06 instead 2015-05-15 14:13:22 sillysausage: it might just be that polkit requires you to disable that grsec feature 2015-05-15 14:13:37 yeah but why does it work from inside i3wm 2015-05-15 14:13:38 _ikke_: ip link det dev up? 2015-05-15 14:13:43 ie when i just do it in urxvt 2015-05-15 14:13:44 check link status 2015-05-15 14:13:54 if it was grsec i would have thought it wouldn't have worked anywhere 2015-05-15 14:14:02 <_ikke_> ncopa: The bond links are up 2015-05-15 14:14:50 I don't know what RLIMIT_CORE is, so if you looked that up, it would probably shed some light 2015-05-15 14:15:25 Question: I just tried booting up a machine on 3.2_rc2, but I get this error (and can't boot): 2015-05-15 14:15:25 "Loading /boot/grsec.gz failed: No such file or directory" 2015-05-15 14:15:30 ncopa: ^ 2015-05-15 14:15:49 <_ikke_> ncopa: the slave interfaces are also up 2015-05-15 14:16:15 _ikke_: and you confirmed that the switch vlan config is ok? 2015-05-15 14:16:28 _ikke_: you could try unplug one of the cables too 2015-05-15 14:16:42 <_ikke_> ncopa: well, if I just add vlan tagging on one interface, it works 2015-05-15 14:16:47 mhavela: vm? 2015-05-15 14:16:59 ahills: Physical hardware 2015-05-15 14:17:06 mhavela: you need update your syslinux.cfg 2015-05-15 14:17:10 <_ikke_> I can't verify the switch setup myself 2015-05-15 14:17:25 did you upgrade or just plug in a new drive? 2015-05-15 14:17:39 re update syslinux.cfg - Strange. I thought the 'setup-bootable' would solve that for me 2015-05-15 14:17:48 Thanks for feedbac ncopa/ahillis 2015-05-15 14:17:55 I try fixing syslinux.cfg 2015-05-15 14:18:06 mhavela: i think i changed the setup-bootable to not touch syslinux.cfg unless you give it a flag for it 2015-05-15 14:18:14 Ah.. 2015-05-15 14:18:27 that is preferrable behavior 2015-05-15 14:18:33 ok I could try that flag. I am ok with having syslinux.cfg updated 2015-05-15 14:18:34 <_ikke_> I do keep getting these martion packages from a different subnet 2015-05-15 14:19:14 mhavela: i'm not sure it is able to fix the grsec.gz -> initramfs-grsec change 2015-05-15 14:19:20 you might need to do that manually 2015-05-15 14:19:42 wasn't that a 3.1 change? 2015-05-15 14:19:49 or copy the syslinux.cfg from the iso before doing: setup-bootable -u -U 2015-05-15 14:19:59 ahills: it was 2015-05-15 14:20:13 so mhavela is probably upgrading from older than 3.1 2015-05-15 14:20:30 would be nice if that just worked... 2015-05-15 14:20:31 ACTION is currently running: setup-bootable -vfu /home/mhavela/Download/alpine-3.2.0_rc2-x86_64.iso /dev/sdb1 2015-05-15 14:20:50 I was upgrading from 2.7 2015-05-15 14:20:57 i dont think it will fix that 2015-05-15 14:21:08 is the preferred upgrade method running a new iso or changing the repo path? 2015-05-15 14:21:14 but the -f said it will replace syslinux.cfg 2015-05-15 14:21:28 ah 2015-05-15 14:21:30 that might work then 2015-05-15 14:21:43 ACTION gives it a new try - brb 2015-05-15 14:21:49 ahills: depends on how you run it 2015-05-15 14:22:18 is changing the repo path, apk update, apk upgrade not a complete one? 2015-05-15 14:22:19 if you have diskless then you need run setup-bootable to replace boot repo and kernel 2015-05-15 14:22:25 ah right 2015-05-15 14:22:25 not if you run diskless 2015-05-15 14:22:42 otherwise, repo path apk upgrade should be enough 2015-05-15 14:22:50 No good :-/ 2015-05-15 14:23:02 mhavela: edit the syslinux.cfg manually 2015-05-15 14:23:07 or copy it folrm the iso 2015-05-15 14:23:08 Ok. I try that 2015-05-15 14:23:24 so syslinux.cfg is broken in 3.2_rc2 iso? 2015-05-15 14:23:31 change boot/grsec too boot/vmlinuz-grsec 2015-05-15 14:23:42 Ok. Thanks - I try that 2015-05-15 14:23:51 and grsec.gz to initramfs-grsec 2015-05-15 14:23:56 that should be enough 2015-05-15 14:24:04 no, i dont think syslinux.cfg is broken in rc2 2015-05-15 14:24:16 i think the setup-bootable in v2.7 does not handle it properly 2015-05-15 14:24:28 Aaah.. Yeah. That might be it 2015-05-15 14:24:40 I am using 2.7 when running setup-bootable 2015-05-15 14:24:42 mhavela: FYI, rc3 is just out 2015-05-15 14:24:44 ACTION edits manually 2015-05-15 14:24:46 Oh... 2015-05-15 14:24:54 ACTION downloads rc3 2015-05-15 14:25:12 i think it woudl be nice if upgrading from v2.7 worked 2015-05-15 14:25:24 i shoudl probably try look at it 2015-05-15 14:25:33 i have no idea how to do that though 2015-05-15 14:25:43 you would always need your upgrade path to include new upgrade procedures from online 2015-05-15 14:26:02 i think i fixed the v2.7 setup-bootable to handle the upgrade to v3.0 2015-05-15 14:26:14 ahills: yes 2015-05-15 14:26:17 that is a good point 2015-05-15 14:26:27 so tehcnically 2015-05-15 14:26:31 update repos 2015-05-15 14:26:32 apk upgrade 2015-05-15 14:26:33 the best you can do is make 3.2 and future upgrade seamlessly 2015-05-15 14:26:44 setup-bootable 2015-05-15 14:26:52 but with diskless 2015-05-15 14:27:00 it should be possible create new boot usb 2015-05-15 14:27:06 and simply copy over the apkovl 2015-05-15 14:27:28 and after next reboot, edit repos to v3.2 and apk upgrade -U -a 2015-05-15 14:27:34 then you should be ok 2015-05-15 14:27:51 i am open for ideas how to do upgrades smooth 2015-05-15 14:27:54 ncopa: Should I change 'kernel' and 'initrd' values to /boot/vmlinuz-grsec? 2015-05-15 14:28:03 yes 2015-05-15 14:28:15 I will need to actually try a serious diskless install first before I can contribute meaningfully 2015-05-15 14:28:26 kernel frmo 'grsec' to vmlinuz-grsec 2015-05-15 14:28:30 shouldn't initrd be initramfs- 2015-05-15 14:28:44 and initrd to initramfs-grsec 2015-05-15 14:28:46 yes 2015-05-15 14:28:51 ahills is right 2015-05-15 14:28:59 initrd is initramfs-grsec 2015-05-15 14:29:13 my memory of my upgrade surprise is still with me 2015-05-15 14:29:22 <_ikke_> ncopa: Ok, even with just one interface it's not working, so it's not necessarily a bonding issue 2015-05-15 14:34:59 hmm 2015-05-15 14:35:07 the mate polkit seems to start 2015-05-15 14:35:29 daemon.notice dbus[1726]: [system] Activating service name='org.freedesktop.PolicyKit1' (using servicehelper) 2015-05-15 14:35:31 daemon.info polkitd[1934]: started daemon version 0.105 using authority implementation `local' version `0.105' 2015-05-15 14:35:33 daemon.notice dbus[1726]: [system] Successfully activated service 'org.freedesktop.PolicyKit1' 2015-05-15 14:35:35 daemon.notice dbus[1726]: [system] Activating service name='org.freedesktop.ConsoleKit' (using servicehelper) 2015-05-15 14:35:37 daemon.notice dbus[1726]: [system] Successfully activated service 'org.freedesktop.ConsoleKi't 2015-05-15 14:35:39 well i see that, but there seems to be no process 2015-05-15 14:35:43 and virt-manager doesn't seem to be able to connect to it 2015-05-15 14:35:56 ie i did: 2015-05-15 14:35:58 exec --no-startup-id /usr/libexec/polkit-mate-authentication-agent-1 2015-05-15 14:36:44 (polkit-mate-authentication-agent-1:2154): polkit-mate-1-WARNING **: Unable to determine the session we are in: GDBus.Error:org.freedesktop.ConsoleKit.Manager.GeneralError: Unable to lookup session information for process '2154' 2015-05-15 14:37:49 maybe ill try polkit-qt 2015-05-15 14:38:58 actually looks like that needs KDE 2015-05-15 14:51:03 <_ikke_> ncopa: they gave me the wrong subnet, it's working now :P 2015-05-15 14:51:30 ncopa/ahills: Thanks for your feedback on the syslinux.cfg modifications - after fixing a typo I made in the file, it started working. 2015-05-15 14:51:48 happy linuxing! 2015-05-15 14:52:04 Now I bumped into another problem (it couldn't modprobe some modules) - but that might be resolved in the rc3 (that I still havent tested) 2015-05-15 14:52:32 So I upgrade the USB stick to 3.2_rc3 and hopefully it's able to modprobe 'bonding' etc. 2015-05-15 14:58:51 have a nice weekend! 2015-05-15 15:00:28 rc3 works better than rc2 in regards to modules 2015-05-15 15:01:00 During boot (rc3) I get a error that it tries mounting /dev/sd?1 on /media/usb but it isn't able because the resource is busy 2015-05-15 15:01:25 But other than that I haven't seen anything that fails (not so far at least) 2015-05-15 15:01:30 Have a nice weekend all! 2015-05-15 15:26:43 well looks like it actually works 2015-05-15 15:26:49 i might write a kvm article for AL 2015-05-15 15:27:10 https://ask.fedoraproject.org/en/question/45805/how-to-use-virt-manager-as-a-non-root-user/ 2015-05-15 15:27:13 i found that worked nicely 2015-05-15 16:06:58 something i noticed about alpinelinux is as soon as i startx 2015-05-15 16:07:03 all my other ttys go black 2015-05-15 16:07:07 even if i quit x 2015-05-15 16:07:27 graphics is intel 2015-05-15 16:11:14 http://bugs.alpinelinux.org/issues/3917 2015-05-15 16:11:17 looks like that bug 2015-05-15 16:11:58 Got it to work after "echo fbcon >> /etc/modules". 2015-05-15 16:12:01 ah maybe i missed that 2015-05-15 16:21:33 so i've figured out why polkit mate or lxpolkit won't start when executed in .xinitrc 2015-05-15 16:21:39 before i3 is executed 2015-05-15 16:21:55 (polkit-mate-authentication-agent-1:2004): polkit-mate-1-WARNING **: Unable to determine the session we are in: GDBus.Error:org.freedesktop.DBus.Error.ServiceUnknown: The name org.freedesktop.ConsoleKit was not provided by any .service files 2015-05-15 16:22:06 seems to be when you're running it not in i3, it doesn't know what session to run it in 2015-05-15 16:22:34 ie: 2015-05-15 16:22:36 sh ~/.local/bin/fehbg & 2015-05-15 16:22:38 /usr/libexec/polkit-mate-authentication-agent-1 & 2015-05-15 16:22:40 exec dbus-launch i3 2015-05-15 16:30:18 xfce gives the same errors 2015-05-15 16:30:34 right. 2015-05-15 16:30:44 but does it actually work once xfce is loaded? 2015-05-15 16:30:55 i might use polkit-gnome, less dependencies and smaller 2015-05-15 16:30:57 but i can run xfce through spice & have copy & paste functionality with no problems 2015-05-15 16:31:08 ah 2015-05-15 16:31:47 http://it-offshore.co.uk/linux/alpine-linux/30-alpine-linux-spice-kvm-desktop 2015-05-15 16:32:05 should work for xen too 2015-05-15 16:32:13 yeah im running kvm 2015-05-15 16:33:06 i use libvirt / virsh (not virt-manager) on kvm 2015-05-15 16:33:16 yeah i might just use virsh 2015-05-15 16:33:18 instead 2015-05-15 16:33:22 & virt-viewer 2015-05-15 16:34:03 on debian you need to build virt-viewer 2.0 for spice functionality (jessie has 1.0) 2015-05-15 16:37:29 brb 2015-05-15 17:34:50 isn't consolekit the polkit for sane people, or am I confused? 2015-05-15 17:57:59 iirc consolekit is also just another deprecated poettering effect 2015-05-15 18:07:21 bonsoir 2015-05-15 18:32:00 ahills: no, we had no real opinion on ABI compatibility with uclibc itself 2015-05-15 18:32:30 ahills: let me put it this way, uclibc did not have a functioning x86_64 port 2015-05-15 18:32:41 ahills: until i came along and fixed it 2015-05-15 18:37:54 :) 2015-05-15 18:50:44 kaniini: I'm glad you've chosen to dedicate your skills to a worthier project 2015-05-15 18:51:19 the last time I was as happy to discover a linux distribution as I was for alpine was arch, back when I was living in a debian/fedora/gentoo hell 2015-05-15 18:51:26 and to be fair, I liked gentoo, I just ran out of free time :P 2015-05-15 18:51:34 i do not think arch belongs on server 2015-05-15 18:51:39 I agree 2015-05-15 18:51:49 alpine has replaced mostly debian boxes for me 2015-05-15 18:52:13 actually I turned eight debian VMs into twenty alpine VMs, and it's still using less memory ;) 2015-05-15 18:52:14 alpine has strong debian roots anyway 2015-05-15 18:52:23 most of the people who work on alpine used to work on or admin debian machines 2015-05-15 18:52:26 ;) 2015-05-15 18:52:28 yeah, it has the same attitude towards correctness 2015-05-15 18:52:38 I also notice some command similarities with apk 2015-05-15 18:53:22 i use/work on both alpine and freebsd 2015-05-15 18:53:26 but oh my god, after dealing with apt-get, and finding apk, it's like being in heaven 2015-05-15 18:53:58 apk is basically 2015-05-15 18:54:03 well 2015-05-15 18:54:07 there's 2 different apk's 2015-05-15 18:54:18 old apk is basically some tools ncopa made on his own 2015-05-15 18:54:28 that more modeled the old bsd package system 2015-05-15 18:54:42 the new apk-tools 2.x/3.x 2015-05-15 18:54:56 basically we looked at apt and really it's kind of a 2015-05-15 18:54:58 I replaced debian with arch on my laptop because I didn't want to go through the work of learning alpine as a desktop OS, but now that I have a mission to try the diskless setup, that will probably change 2015-05-15 18:55:04 "what if we wrote apt-get today?" exercise 2015-05-15 18:55:14 I didn't find alpine until v3.0.1, so apk was already very mature 2015-05-15 18:55:42 I have a special kind of love for developers who focus on making elegant command line interfaces 2015-05-15 18:55:58 just out of curiosity, why did you choose to integrate git so heavily? 2015-05-15 18:56:08 dunno, that wasn't my decision :) 2015-05-15 18:56:19 s/you/y'all/ 2015-05-15 18:56:40 i think dpkg source packages are terrible though 2015-05-15 18:56:48 yes... they are... 2015-05-15 18:56:54 going the PKGBUILD route was very savvy 2015-05-15 18:57:12 yeah, making things easier for development 2015-05-15 18:57:36 I wanted a stable version of arch so badly for that reason 2015-05-15 18:57:46 no more installs to /usr/local/bin for every bit of software 2015-05-15 18:58:06 (obviously I'm not a BSD user at heart) 2015-05-15 19:04:48 alpine has spawned a lot of big things in FOSS too 2015-05-15 19:05:03 OpenNHRP, pkgconf, etc 2015-05-15 19:07:29 the way abuild cleans up after itself is very nice - on debian i use a docker build environment for the same effect 2015-05-15 19:08:21 many people I know haven't heard of it, so I'm always surprised when I find references to it sprinkled around the internet 2015-05-15 19:09:21 that is changing slowly 2015-05-15 19:09:29 i was here when #alpine-linux had 9 users 2015-05-15 19:09:31 now it has 123 2015-05-15 19:09:33 ;) 2015-05-15 19:11:08 alpine hipster 2015-05-15 19:11:22 when it gets to 250 users it's mainstream 2015-05-15 19:11:26 and you'll have to switch to sabotage 2015-05-15 19:11:39 heh 2015-05-15 19:13:10 :) 2015-05-15 19:18:20 ahills: i do not think sh4rm4 and i agree on basically any aspect of distribution design 2015-05-15 19:18:50 ;) 2015-05-15 19:19:43 where's bitshark to tell us about how he's going to mix aptitude, pacman, and apk on sabotage 2015-05-15 19:22:28 but its funny that debian is now working on stuff we've had going for years 2015-05-15 19:22:32 like reproducible builds 2015-05-15 19:24:11 The main grief I have with the debian build system is its complexity 2015-05-15 19:24:29 APK/PKGBUILDs are a children playground in comparison 2015-05-15 19:26:01 i need to add an additional power circuit for my closet 2015-05-15 19:37:01 Is that related to the Debian build system? 2015-05-15 19:40:11 no 2015-05-15 23:18:04 for those who are interrested to install flashplugin on alpine, I've made this: https://github.com/WnP/InstallFlash 2015-05-15 23:21:18 feedbacks are welcome 2015-05-15 23:49:37 i'm interested. 2015-05-15 23:50:05 i don't get what the lib from glibc is for tho 2015-05-15 23:50:11 that sounds unlikely to be safe/reliable 2015-05-15 23:55:16 iduno too, but it's mandatory for flashplugin 2015-05-15 23:57:36 gcc -fPIC -shared -nostartfiles -O3 -x c /dev/null -o ld-linux-x86-64.so.2 2015-05-15 23:57:46 (make an empty .so by that name) 2015-05-15 23:58:00 that should work instead 2015-05-15 23:58:46 ACTION try dalias 2015-05-16 00:00:00 i'll try 2015-05-16 00:01:04 yeah, it works :) 2015-05-16 00:01:11 ACTION update the code 2015-05-16 00:06:10 dalias: could I mention you as contributor? 2015-05-16 00:06:20 if you like 2015-05-16 00:06:47 nice, do you have a github page or anything else to link? 2015-05-16 00:13:10 musl-libc.org ;-) 2015-05-16 00:15:19 and yes, it seems to work here 2015-05-16 00:16:09 dalias: code updated 2015-05-16 00:30:33 dalias: flashplugin will never, ever be safe/reliable, jail it as much as possible 2015-05-16 00:30:56 I would consider apparmor at least to be a non-negotiable prerequisite to running that code on my machine 2015-05-16 00:31:02 :) 2015-05-16 00:31:20 click-to-play is a pretty good approach 2015-05-16 01:27:11 sup 2015-05-16 15:12:45 i was thinking of buying this: http://www.element14.com/community/docs/DOC-68907/l/shim-rtc-realtime-clock-accessory-board-for-raspberry-pi 2015-05-16 15:13:01 i wonder if the AL RP2 kernel supports that 2015-05-16 15:13:11 just ordered a rpi to put alpine on 2015-05-16 15:13:22 should be here next week 2015-05-16 15:19:22 RP2? 2015-05-16 15:20:08 yep 2015-05-16 15:20:16 also a new xenon flash bulb so I can reboot it 2015-05-16 15:28:34 xenon flash bulb? 2015-05-16 15:28:41 i use mine as a home router 2015-05-16 15:30:36 haha 2015-05-16 15:30:43 but it's kind of irritating because i don't have an RTC clock 2015-05-16 15:30:43 I can't imagine using a pi for that 2015-05-16 15:30:54 heh. well my internet is pretty shitty 2015-05-16 15:30:54 I have a proper router running openbsd 2015-05-16 15:31:00 7mbit 2015-05-16 15:31:03 ah, bummer 2015-05-16 15:31:07 and a RP can route up to 30mbit 2015-05-16 15:31:12 I finally got over 110Mbit here 2015-05-16 15:31:16 lulz 2015-05-16 15:31:22 you obviously live in a nice country 2015-05-16 15:31:41 probably romania 2015-05-16 15:31:45 i hear they have good internet 2015-05-16 15:33:45 ACTION cheers, "USA! USA! USA!" 2015-05-16 15:34:18 I actually dropped the cost of my internet by $170/mo and got 10Mbps extra out of it 2015-05-16 15:34:52 oh ahills 2015-05-16 15:34:57 i live in ausfailure 2015-05-16 15:35:13 ausfailurestralia or ausfailuretria? 2015-05-16 15:35:31 the one with the corrupt government that is trying to crawl up the ass of mining companies 2015-05-16 15:35:45 well, I know australia is corrupt from top to bottom 2015-05-16 15:35:46 you know because everyone wants coal, even when they don't 2015-05-16 15:36:14 and because the average australian is retarded 2015-05-16 15:36:19 don't really know anything about austria except that it looks beautiful from the ICE trains and arnold comes from there 2015-05-16 15:36:21 and doesn't know what is good for them voted in this government 2015-05-16 15:36:31 it's not like they had a choice 2015-05-16 15:36:42 yeah but it's changing rapidly 2015-05-16 15:36:47 adopting all the bad things from the US 2015-05-16 15:36:57 as opposed to staying with it's european roots 2015-05-16 15:37:13 eg, war on public healthcare, war on minimum wage 2015-05-16 15:37:19 war on clean energy 2015-05-16 15:37:25 war on drugs 2015-05-16 15:37:30 yeah that too and terrorism 2015-05-16 15:37:35 also I hear video games are censored there 2015-05-16 15:37:38 jack thompson should move 2015-05-16 15:37:42 yeah he has 2015-05-16 15:37:48 it was a retarded attorney general we had 2015-05-16 15:38:02 http://en.wikipedia.org/wiki/Michael_Atkinson 2015-05-16 15:38:15 wouldn't let them implement an R18 rating 2015-05-16 15:38:24 so anything that would be R18 elsewhere in the world 2015-05-16 15:38:30 got banned here because no rating existed for it 2015-05-16 15:38:37 hahahahaha 2015-05-16 15:38:56 https://en.wikipedia.org/wiki/Michael_Atkinson#Media_classification_and_censorship 2015-05-16 15:39:00 basically he's a fuckwit 2015-05-16 15:39:29 Unanimity from Atkinson and his fellow state and federal Attorneys-General is required for the introduction of the rating (or a change to that requirement).[10 2015-05-16 15:39:32 the others supported one 2015-05-16 15:39:36 but he kept blocking it 2015-05-16 15:40:04 because he is an old man who only plays wi tennis and thinks everyone else should too 2015-05-16 15:41:00 actually the setup we have here is almost identical to the UK with david cameron 2015-05-16 15:41:14 im waiting for the day when they declare all encryption that they cannot break is bad 2015-05-16 15:41:19 don't worry 2015-05-16 15:41:23 old people eventually dio 2015-05-16 15:41:25 die 2015-05-16 15:41:32 usually we let our counterparts US/UK do all the legwork before we implement laws here 2015-05-16 15:41:55 it's like the once good FVEY countries have been infested with conservative fuckwits 2015-05-16 15:42:06 who are just greedy and want to mess everything up 2015-05-16 15:42:39 re-igniting the crypto wars is a pet hate of mine 2015-05-16 15:42:46 because it is fucking stupid 2015-05-16 15:43:11 the only way to stay in office is to give your population something to be afraid of, right? 2015-05-16 15:43:19 yeah 2015-05-16 15:43:24 for us its men in turbans 2015-05-16 15:43:29 and halal meat 2015-05-16 15:43:41 isn't halal meat most meat that isn't pork 2015-05-16 15:43:43 there are politicians here that won't shut up about how it funds terrorism 2015-05-16 15:44:05 nah it's a certification thing like the jewish cosher 2015-05-16 15:44:25 the federal police investigated it and found that there's no links to terrorism 2015-05-16 15:44:34 but it doesn't stop politicians from still saying otherwise 2015-05-16 15:44:49 i've worked in a meat place 2015-05-16 15:44:54 basically all it is is a certification 2015-05-16 15:44:59 the workers tend to go on break 2015-05-16 15:45:17 your name has new meaning to me now 2015-05-16 15:45:28 and a grubby old man comes into the abattoir and says some muslim bullshit from his book 2015-05-16 15:45:30 then he goes home 2015-05-16 15:45:36 and the meat can be marketed as "halal" 2015-05-16 15:45:40 lol 2015-05-16 15:45:50 is it an extra dollar a pound or something? 2015-05-16 15:45:53 they often don't really care how the meat is killed 2015-05-16 15:45:59 well the licensing is 2015-05-16 15:46:04 but we can then export to indonesia 2015-05-16 15:46:09 who now hate us 2015-05-16 15:46:18 because you know we spied on their president and got caught doing it 2015-05-16 15:46:28 I'm glad to hear your corner of the world is just as fucked up as mine 2015-05-16 15:46:35 let's all give up and move to europe 2015-05-16 15:46:47 yeah if david cameron doesn't ruin that 2015-05-16 15:46:54 thankfully the EU struck down mandatory data retention 2015-05-16 15:46:59 so some countries are safe to VPN to 2015-05-16 15:47:11 I don't really consider the UK part of europe, it's very different culturally 2015-05-16 15:47:17 correct 2015-05-16 15:47:20 UK and former colonies are much closer 2015-05-16 15:47:24 but it doesn't stop them pushing that bullshit 2015-05-16 15:47:31 of course, of course 2015-05-16 15:47:36 it was the UK that was trying to get the EU to do data retention in the first place 2015-05-16 15:47:51 because they are a part of FVEY and tasked with being the US's special outpost 2015-05-16 15:48:01 like australia is supposed to spy on the asians 2015-05-16 15:48:24 australia is in the asian region yet our culture is nothing like theirs 2015-05-16 15:49:41 surprise! 2015-05-16 15:53:19 gnublin-develop-kernel.googlecode.com/git-history/c6fe84097da8ecd91b0fd4e48556cf56d1e7e17f/linux-2.6.33-lpc313x/drivers/rtc/rtc-mcp7940.c 2015-05-16 15:53:24 i wonder if that ever got merged in the kerne 2015-05-16 15:53:26 l 2015-05-16 15:53:35 2.6.33... 2015-05-16 15:54:19 yeah but that's the gnublin kernel 2015-05-16 15:55:44 Shouldn't be too hard to patch it in 2015-05-16 15:55:51 I really doubt the RTC driver framework has changed at all 2015-05-16 15:55:58 http://tom.meinlschmidt.org/2014/02/19/mcp7940-rtc-with-pi/ 2015-05-16 15:56:07 nah it hasn't but i didn't want to recompile the kernel :P 2015-05-16 15:56:26 but i guess it could be merged into the AL one 2015-05-16 15:56:35 as it is a standard module used across many RPs 2015-05-16 15:56:45 haha what a silly use of python 2015-05-16 15:56:59 http://www.element14.com/community/docs/DOC-68907/l/shim-rtc-realtime-clock-accessory-board-for-raspberry-pi?ICID=rpiaccsy-access-products#blogposts 2015-05-16 15:57:03 well documented there 2015-05-16 15:57:07 don't be lazy 2015-05-16 15:57:20 cross-compiling an arm kernel will take you like half an hour to set up and one minute to build 2015-05-16 15:57:26 unless you're on a pentium 4 2015-05-16 15:59:27 http://www.piface.org.uk/assets/piface_clock/PiFaceClockguide.pdf i saw a link to 2015-05-16 15:59:29 https://raw.githubusercontent.com/piface/PiFace-Real-Time-Clock/master/install-piface-real-time-clock.sh 2015-05-16 15:59:44 i wonder if it will work on AL 2015-05-16 16:00:08 it doesn't talk about obtaining the patch 2015-05-16 16:00:13 modprobe i2c:mcp7941x 2015-05-16 16:00:13 so im thinking it is merged into the kernel 2015-05-16 16:00:21 yeah 2015-05-16 16:00:25 it should have a config option 2015-05-16 16:00:44 grep your config for 'mcp(.+)?7941' 2015-05-16 16:00:53 grep -i obviously 2015-05-16 16:01:06 might already be there 2015-05-16 16:01:21 I guess you can just type modprobe since you have a pi, eh? 2015-05-16 16:01:28 yeah 2015-05-16 16:01:35 and it's got alpine linux on it 2015-05-16 16:01:53 do note that for the RP2 you need the edge version 2015-05-16 16:02:01 I'm getting a pi because it has an analog audio out jack, and the beaglebone black doesn't 2015-05-16 16:04:25 http://dev.alpinelinux.org/~tteras/alpine-rpi-150424-armhf.rpi.tar.gz 2015-05-16 16:04:31 so yeah that's the link to the version you need for the RP2 2015-05-16 16:04:49 i remember trying the one from the download page and it didn't work 2015-05-16 16:05:03 and fabled i think mentioned you need the newer one with the newer kernel for the RP2s 2015-05-16 16:05:16 so just keep that in mind when you get your pi :) 2015-05-16 16:05:23 I don't need a realtime clock 2015-05-16 16:05:28 I'm just going to use NTP 2015-05-16 16:05:29 no but to boot 2015-05-16 16:05:34 oh yeah 2015-05-16 16:05:34 you wont be able to boot it 2015-05-16 16:05:46 without using that version ^ 2015-05-16 16:05:56 I want to hack up my alpine ISO already anyways 2015-05-16 16:06:07 ah 2015-05-16 16:06:18 way i looked at it 2015-05-16 16:06:26 RP was the best inexpensive option for a router 2015-05-16 16:06:34 because anything else would have been more than $100 2015-05-16 16:06:42 I have a buffalo I've had for years now 2015-05-16 16:06:46 most of our country is on ADSL, on really bad phone lines 2015-05-16 16:06:47 I think I got it in 2010 2015-05-16 16:06:55 yeah, I actually got it when I was on DSL 2015-05-16 16:06:58 not even ADSL 2015-05-16 16:06:59 the medium ADSL speeds australians get is 6.5/7Mbit 2015-05-16 16:07:02 1Mbit upload 2015-05-16 16:07:05 when I was literally two hours' drive from where I am now 2015-05-16 16:07:09 1Mbit download 2015-05-16 16:07:12 700Kbit upload 2015-05-16 16:07:13 so even if i move its a good chance i wont need anything more powerful 2015-05-16 16:07:16 on a good day 2015-05-16 16:07:41 and thanks to the current government they canned the FTTP project that was in the works 2015-05-16 16:07:44 I used it as a reverse wifi hotspot so I could get on my neighbors' cable connection at night 2015-05-16 16:07:52 so if you're in one of the lucky areas that got it when the previous government was in then yay 2015-05-16 16:07:55 otherwise you get nothing 2015-05-16 16:08:00 or maybe some shitty fibre to the node 2015-05-16 16:08:04 but nobody has gotten that yet 2015-05-16 16:08:29 i think it's due to the fact that Telstra who is one of the only cable providers 2015-05-16 16:08:38 sells HFC in monopoly in a lot of areas at 100mbit 2015-05-16 16:08:53 and our prime minister is rupert murdoch's private butt buddy 2015-05-16 16:09:05 good ol' murdoch 2015-05-16 16:09:21 so providing 100Mbit FTTP to compete with that yeah, you can see how a 50/50% company like FoxTel would not like that 2015-05-16 16:09:38 I need to do non-computer things, be back later for some alpine 2015-05-16 16:09:39 so in return for sabotaging it he got positive press in rupert's papers 2015-05-16 16:10:24 meanwhile we're now paying telstra more money than the FTTP network would have cost to begin with to maintain the copper for us to continue using crappy adsl on 2015-05-16 16:10:38 (over the next 30 years) 2015-05-16 16:17:27 and yes those modules are there i was able to probe them both 2015-05-16 16:20:21 i2c_bcm2708 5364 0 2015-05-16 16:20:23 rtc_ds1307 9378 0 2015-05-16 16:20:25 i2c_dev 5894 0 2015-05-16 16:20:27 apparently was all you needed 2015-05-16 17:16:24 Anyone else can't seem to download the gpg signature files for the latest version? 2015-05-16 17:16:36 Is this a temporary thing? 2015-05-17 09:01:17 hey 2015-05-17 16:15:38 wlan0 at boot says "udhcpc failed to get a DHCP lease [!!]" 2015-05-17 16:15:39 then I manually do "sudo udhcpc -i wlan0" after boot and get internet, but I wanted it automatically as it was before 2015-05-17 16:15:52 I'm on alpine linux edge, can someone help me please? thanks 2015-05-17 16:25:46 anyone know? 2015-05-17 17:54:37 teigo - I use these instructions http://linuxcommando.blogspot.nl/2013/10/how-to-connect-to-wpawpa2-wifi-network.html + the comment near the bottom from "Emerson Prado" 2015-05-17 17:57:04 I set my wireless that way many weeks ago, BitL0G1c 2015-05-17 17:57:19 it was working but suddenly stopped working 2015-05-17 17:57:59 now, all of a sudden, I get "udhcpc failed to get a DHCP lease" on boot 2015-05-17 17:58:50 after boot if I do "udhcpc -i wlan0" it works, but I wanted it to be automatic on boot like it was before :( 2015-05-17 18:02:10 check dmesg & the logs 2015-05-17 18:12:59 Hey, the alpine live cd hangs at boot; I'm told this is frequently a GPU issue and can be avoided by passong "nomodeset"... but how do I boot with options in isolinux? naively emtering "nomodeset noquiet" as the boot string makes it say "Loading nomodeset failed: No such file or directory" 2015-05-17 18:15:53 nvm, got it. tab completion 2015-05-17 18:24:42 BitL0G1c: dmesg tells me wlan0 associated (but I don't know if the message is from during boot or after) 2015-05-17 18:25:08 I also get some weird messages after that, but I don't know if they are relevant 2015-05-17 18:25:10 brcmsmac bcma0:1: wl0: brcms_c_d11hdrs_mac80211: \xffffffd07^A3^A\xffffff88\xffffffff\xffffffff txop exceeded phylen 159/256 dur 1778/1504 2015-05-17 18:33:09 brcmsmac is the broadcom wireless driver http://www.linuxwireless.org/en/users/Drivers/brcm80211/ 2015-05-17 18:35:38 this is a similar error http://ubuntuforums.org/showthread.php?t=2226075 - "exceeded phylen" 2015-05-17 18:40:33 if it comes up when you run it manually perhaps try putting a script in /etc/local.d/ 2015-05-17 18:50:37 I go into single-user mode and liyerally set my root password to blank 2015-05-17 18:50:48 and I still get "login imcorrect" 2015-05-17 18:51:44 minus the typos; I'm on my phone 2015-05-17 18:51:58 are you logging in via ssh? 2015-05-17 18:52:12 if so, ssh does not allow blank root password 2015-05-17 18:53:03 no, local box 2015-05-17 18:53:58 besides, who actually allows keyboard-interactive ssh auth pfft 2015-05-17 18:59:50 is there a good way to keep alpine up-to-date on a long-running machine that is seldom logged into? 2015-05-17 19:00:19 something like yum-cron or cron-apt -- just want to keep security patches up-to-date, but nobody really directly interacts with this system 2015-05-17 19:03:30 BitL0G1c: putting a script in /etc/local.d/ sounds like a hackish solution, am I out of options besides that? 2015-05-17 19:03:43 it used to work just fine before 2015-05-17 19:04:13 kaniini: 2015-05-17 19:04:15 any ideas? 2015-05-17 19:04:35 not sure :( 2015-05-17 19:06:52 khm: you can at least wrap around apk upgrade --simulate to make it a little safer 2015-05-17 19:10:22 khm - use 'apk-cron' 2015-05-17 19:11:54 teigo - try the linux-vanilla kernel - to at least rule out grsecurity causing this on a new kernel 2015-05-17 19:19:01 BitL0G1c: what if I downgrade the kernel? would that work? how do I do it? 2015-05-17 19:23:18 if you are on edge you probably need to download the kernel from main & install it manually 2015-05-17 19:27:49 or try starting wireless from /etc/local.d 2015-05-17 21:06:19 some use docker here? trying to make it up and running but always have this error: http://sprunge.us/Cjge 2015-05-17 21:07:06 fresh install from edge, linux-vanilla kernel 2015-05-17 21:08:48 by the way in this example it create a b2cf4f549bc96a86ed19731766c57ba2845b0178b5c17acf4aefe8ebb8f5b96e folder in /var/lib/docker/containers, each time I run this command it create a new folder with a new hash, but the error still present 2015-05-17 21:08:53 any idea? 2015-05-17 21:12:50 ncopa: hello - i got 404 for alpine-3.1.4-x86.iso.asc - was there a published 3.1.4 sig for official iso? 2015-05-17 21:13:12 right now i'm looking at http://dl-3.alpinelinux.org/alpine/v3.1/releases/x86/ and i don't see it there listed 2015-05-17 21:13:48 it also does not work from https://alpinelinux.org/downloads/ 2015-05-17 21:14:16 i see it's a fresh release (few days old) so when signature will be published? 2015-05-17 22:24:41 Mo0O - check you have /dev/mqueue http://bugs.alpinelinux.org/issues/4010 2015-05-18 04:32:53 do you know how to fix this error when starting named deamon? -> named: -u with Linux threads not supported: no capabilities support or capabilities disabled at build time 2015-05-18 04:34:19 is that bind? 2015-05-18 04:35:10 yes dalias 2015-05-18 04:35:57 sounds like it was built wrong or something 2015-05-18 04:36:14 looks like 2015-05-18 04:48:23 I guess the problem is here: http://git.alpinelinux.org/cgit/aports/tree/main/bind/APKBUILD?id=737279b1#n58 2015-05-18 04:49:38 maybe 2015-05-18 04:49:43 wonder why it fails without caps tho 2015-05-18 04:52:21 I've found only this related topic https://forums.gentoo.org/viewtopic-t-891650-view-previous.html?sid=04de280cb68ba985aa0f72b0edca7ea8 2015-05-18 06:19:17 for SSD users 2015-05-18 06:19:18 http://www.jedec.org/sites/default/files/Alvin_Cox%20%5BCompatibility%20Mode%5D_0.pdf 2015-05-18 06:19:18 http://www.dailytech.com/Seagate+Senior+Researcher+Heat+Can+Kill+Data+on+Stored+SSDs+/article37346.htm 2015-05-18 08:32:40 morning 2015-05-18 08:50:34 Hi guys, is there a way to upgrade the kernel? I'd like to try overlayfs in kernel 3.18 2015-05-18 08:51:48 Good morning 2015-05-18 08:52:09 ncopa: Re 3.2_rc3: Running 'last' will give me this error: 2015-05-18 08:52:31 last: can't open '/dev/null/wtmp': Not a directory 2015-05-18 08:56:31 HowardMei: we have 3.18 in edge 2015-05-18 08:57:17 Thanks! It's good to know :) 2015-05-18 09:06:20 mhavela: thats a known missing feature in musl 2015-05-18 09:06:27 Ah. Ok. 2015-05-18 09:08:04 Seems 'who' is also broken :-/ 2015-05-18 09:08:06 utmpx is missing 2015-05-18 09:08:07 yes 2015-05-18 09:08:26 ncopa: i am worried about liskamm mirror issue when using setup-apkrepos 2015-05-18 09:08:56 ScrumpyJack: i think latest alpine-conf should ignore 'broken' mirrors 2015-05-18 09:09:09 but I might need add --force 2015-05-18 09:14:59 kk 2015-05-18 09:53:09 uh, i have a conflict between nettle and gnutls on one of my hosts, looks like a packaging/repo thingy anyone an idea how to resolve this?http://sprunge.us/bcUa 2015-05-18 09:53:19 repos are edge only 2015-05-18 14:31:50 gs63vW9tNVr9: try apk upgrade -U -a 2015-05-18 14:33:49 \o/ thx ncopa! 2015-05-18 15:32:25 hello 2015-05-18 19:34:15 do you know where are declared default start() stop() and status() function for openrc initscripts? 2015-05-18 20:08:07 evening 2015-05-18 20:08:22 hi 2015-05-18 20:34:08 I noticed something while configuring mariadb that I don't understand, maybe somebody here knows 2015-05-18 20:37:14 If I set innodb_buffer_pool_size=1G I cannot start mariadb. I get InnoDB: mmap(1089208320 bytes) failed; errno 12 2015-05-18 20:37:53 I have to go all of the way down to innodb_buffer_pool_size=360M to be able to start mariadb 2015-05-18 20:38:31 This is on a 32 bit system with 2GB of ram, only about 100M in use (confirmed with free, top, htop) before attempting to start mariadb 2015-05-18 20:39:43 I don't see anything in ulimit that would seem to be a problems (ulimit -v is unlimited). Also have 4GB of swap enabled (none in use) 2015-05-18 20:41:16 I googled around and I'm not seeing anybody else having a problem that meets this description. I guess it's not a huge problem because I'm happy with the performance of my site, but I'm still a bit curious why I can't increase the buffer pool size to say a gigabyte, because it seems I have plenty enough free ram 2015-05-18 20:41:43 Is it some PaX related thing maybe? 2015-05-18 20:45:06 After I start mariadb with 360m buffer_pool_size it has a VSZ of 1018m and 50%VSZ so is there a limit that a single process cannot take more than 50%VSZ? 2015-05-18 20:45:58 have you run dmesg? 2015-05-18 20:48:16 with grsec aslr it is possible that the address space got fragmented 2015-05-18 20:48:33 and the largest continous address range is 360M 2015-05-18 20:48:58 simply disable aslr and see if that makes it work 2015-05-18 20:50:04 printf 0 >/proc/sys/kernel/randomize_va_space 2015-05-18 20:50:20 I don't see anything relevant in dmesg but let me try the disable aslr thing that sounds like it might work 2015-05-18 20:53:01 Same error 2015-05-18 20:53:44 can you run it in gdb? 2015-05-18 20:54:02 or just stop the process in some way after the mmap fails.. 2015-05-18 20:54:30 then cat /proc//maps 2015-05-18 20:54:57 or in gdb: i proc maps 2015-05-18 20:55:26 I'm not familiar with gdb but give me a few minutes to read up 2015-05-18 21:03:18 Lol, now I can't even start it with 360M size. Time to reboot 2015-05-18 21:04:40 I guess it's not important immediately so I'll just spend some time reading up about gdb and /proc//maps 2015-05-18 21:05:23 It was just something I've never encountered before. Also, I'm good again at 360M after the reboot 2015-05-18 21:05:50 is it a continously running process? 2015-05-18 21:06:29 then paste /proc/pid/maps somewhere 2015-05-18 21:12:47 Yes it's continuously running http://sprunge.us/PiMa 2015-05-18 21:12:50 i'm guessing it's fragmented address space 2015-05-18 21:13:05 *sigh* 2015-05-18 21:13:14 grsec dummied out all the data in maps 2015-05-18 21:13:27 so there's no way to tell what's going on 2015-05-18 21:16:12 It doesn't *really* matter as my application runs fine, I was just kind of baffled. Figured maybe there was something to learn 2015-05-18 21:17:02 Do you think the issue would be gone with the vanilla kernel? 2015-05-18 21:17:26 if it's aslr then it's not fine: it can randomly fail 2015-05-18 21:20:17 this thing is running >10 threads 2015-05-18 21:20:54 if they get their stack mapped evenly in the address space that means <400M continous area 2015-05-18 21:22:39 dalias: but grsec is perfect in every way 2015-05-18 21:22:57 at least if you don't say that spender will come beat you like a wife or something 2015-05-18 21:23:17 iwearsable, grsec increases randomization 2015-05-18 21:23:26 and the main executable is probably PIE, right? 2015-05-18 21:23:35 everything in alpine is PIE 2015-05-18 21:23:41 so you're likely to have stuff spread out all over VM 2015-05-18 21:23:53 in which case you'd be lucky to get more than a few hundred megs contiguous 2015-05-18 21:24:33 grsec is a thing that really needs to be forked by someone more responsible 2015-05-18 21:24:43 mmm... PIE 2015-05-18 21:25:34 PaX has nice features that we then go around gimping with paxmark because things we want to support are incompatible with PaX 2015-05-18 21:29:11 i think for some use-cases grsec is probably good, but i think for the average user it may be too aggressive (like with this contiguous space thing) 2015-05-18 21:29:14 Well, it sounds like I could use vanilla kernel if it came down to it. Although that kind of defeats the point of using Alpine 2015-05-18 21:29:30 how does it defeat the point? 2015-05-18 21:30:09 Well, isn't the grsec/pax a main distinguishing feature? I know it's not the only one 2015-05-18 21:30:43 Musl being the other one. Anyways I didn't mean any offense 2015-05-18 21:31:01 i wouldn't say grsec is the point 2015-05-18 21:31:23 iwearsable: i think we provide adequate ASLR on vanilla kernel 2015-05-18 21:31:39 the single biggest contribution to security for one-task servers is the lack of sprawling dependencies that result in having hundreds of unneeded packages installed (and thus a huge attack surface) 2015-05-18 21:32:25 yep 2015-05-18 21:32:34 alpine also gives you PIE, relro, fortify (only in latest versions), etc. hardening 2015-05-18 21:32:39 That's true, alpine is very small and doesn't end up with a bunch of junk you dont need 2015-05-18 21:33:05 there should be an easy way to go into "debug mode" on the grsec kernel 2015-05-18 21:33:06 most of grsecurity's perceived security advantages really depend on PIE/relro/etc. 2015-05-18 21:33:16 a philosophy of avoiding/eliminating suid would also help. so far alpine is only so-so in that area 2015-05-18 21:33:24 i'd like to see alpine adopt more ideas from Owl for that 2015-05-18 21:33:28 dalias: me too 2015-05-18 21:33:33 dalias: i want to work on that in 3.3 2015-05-18 21:33:50 i also want to see us perhaps start deprecating grsecurity 2015-05-18 21:34:09 re: deprecating grsec... 2015-05-18 21:34:12 i'm not convinced grsec provides meaningful protection (other than perhaps syscall exploit countermeasures, but i mean that's so rare) 2015-05-18 21:34:26 i think it would make sense to do some analysis of what features grsec is adding that modern vanilla kernel doesn't have a way to do 2015-05-18 21:34:28 (also their countermeasures are dubious at best) 2015-05-18 21:34:29 yes 2015-05-18 21:34:31 precisely 2015-05-18 21:34:52 and see if there are better tools (LSMs or whatever) that could meet the gap 2015-05-18 21:34:59 chromium os has already done the work for us 2015-05-18 21:35:02 :) 2015-05-18 21:35:03 they have this Yama thing 2015-05-18 21:35:14 which brings over the missing grsec features that really matter 2015-05-18 21:35:20 i'm not convinced PaX really matters anymore 2015-05-18 21:35:20 i'm a bit skeptical of theirs too tho :) 2015-05-18 21:35:21 afaik, grsec has not been approved into the so called mainline kernel 2015-05-18 21:35:38 (yama) 2015-05-18 21:35:40 there might be a reason for it... 2015-05-18 21:36:04 i think alpine's main security advantage has nothing to do with grsec 2015-05-18 21:36:22 and everything to do with the fact that everything we do is based on simpler, more robust components 2015-05-18 21:36:25 too many of the kernel hardening efforts have approached it with an attitude that you can just remove or turn off interfaces and it's automatically a security enhancement 2015-05-18 21:36:45 but if an app misbehaves because the interface is changed or disabled, or if it uses a less-secure or less-tested fallback because of it.... 2015-05-18 21:36:51 that could actually _reduce_ security 2015-05-18 21:37:27 I gotta go move some furniture but I'll catch up on the conversation later. I think I understand a few things a little bit more, thank you all 2015-05-18 21:37:34 :) 2015-05-18 21:37:43 kaniini: alpine uses musl, not glib, it's a big change in and on itself already 2015-05-18 21:37:49 my first encounter with yama was similar to grsec: "where did my core dump go?" 2015-05-18 21:37:51 yes, like i said 2015-05-18 21:37:58 nsz, :) 2015-05-18 21:38:03 "simple, more robust components" 2015-05-18 21:38:40 actually we should add security+/cve rate to libc comparison... 2015-05-18 21:38:43 inb4 alpine switches to unix kernel 2015-05-18 21:39:09 i'm not happy with the few CVEs we've had to issue for musl 2015-05-18 21:39:21 still better than glibc :p 2015-05-18 21:39:30 but if you look at the big picture over time i think it's a lot fewer per-unit-time, and total-over-lifetime, than glibc 2015-05-18 21:39:31 or postfix 2015-05-18 21:39:50 or uclibc: 0 CVEs but thats because they do not even really maintain it 2015-05-18 21:39:54 kaniini, :) 2015-05-18 21:40:03 yeah for uclibc it would have to just be a red "no data' 2015-05-18 21:40:14 same for dietlibc too i think :) 2015-05-18 21:40:34 if we can add bionic to the comparison there might be interesting data for it 2015-05-18 21:41:05 I'd love to see alpine on MIPS and other more obscure stuff to replace OpenWRT :3 2015-05-18 21:41:10 Diftraku: working on it 2015-05-18 21:41:17 <3 2015-05-18 21:43:21 :) 2015-05-18 21:43:33 i wouldn't necessarily see it as a replacement for openwrt 2015-05-18 21:43:57 the interesting part of openwrt, from my perspective at least, is the router config/admin panel/etc. stuff 2015-05-18 21:44:12 which could presumably be reused on an alpine base 2015-05-18 21:44:23 well there is ACF which has the same features 2015-05-18 21:45:08 what exactly is ACF? 2015-05-18 21:45:14 MIPS is more interesting for "micro-server" type deployments tho 2015-05-18 21:45:27 it's a configuration framework, part of which provides a web ui 2015-05-18 21:46:00 is it comparable to end-user router config panel tho? 2015-05-18 21:46:10 there are modules which provide that 2015-05-18 21:46:24 btw i also suspect openwrt is more practical for really-tiny-storage/mem router hardware 2015-05-18 21:46:33 yeah 2015-05-18 21:46:36 since it's stripped down specifically for that 2015-05-18 21:46:50 they even use fully-shared libgcc and softfloat 2015-05-18 21:48:30 bbiab 2015-05-18 21:49:08 they increased the min ram requirement to 16M not too long ago 2015-05-18 21:49:44 BitL0G1c: i just realized that conversation was probably better had here instead of devel chan D: 2015-05-18 21:49:57 openwrt minimum is 4mb of flash 2015-05-18 21:50:22 yes 2015-05-18 21:56:51 creating dockers with optional static ip's https://gist.github.com/itoffshore/9e30f6c857ab51f05d9b 2015-05-18 21:57:56 needs pipework & a bridge http://it-offshore.co.uk/linux/debian/36-giving-docker-containers-static-ip-addresses 2015-05-18 21:58:38 asenna gentoo 2015-05-26 14:53:02 I'm running Alpine 2.7 and I have/use alpine-desktop 2015-05-26 14:53:15 Now I'm about to upgrade to 3.2 (my HDD installed laptop) 2015-05-26 14:53:44 Does 3.2 have a functional desktop envonment? 2015-05-26 14:53:53 environment 2015-05-26 14:58:00 yes 2015-05-26 15:01:14 \o/ 2015-05-26 15:02:01 hum 2015-05-26 15:02:19 why dont they show up on http://alpinelinux.org/downloads/ 2015-05-26 15:03:28 These are bootable iso, you can then install whichever DE you want (and is available) 2015-05-26 15:03:56 i mean, why did not 3.2. iso show up on downlaods 2015-05-26 15:05:55 oh :) 2015-05-26 15:06:15 Surely because the webpage hasn't been updated yet, wait 2015-05-26 15:29:15 ok, here goes test upgrade to alpine-xen 3.2! 2015-05-26 15:29:45 Upgrading xen (4.4.2-r1 -> 4.5.0-r0) 2015-05-26 15:29:49 \o/ 2015-05-26 15:30:31 hmm mount: mounting /dev/usbdisk on /media/usb failed: Resource busy 2015-05-26 15:33:20 libxl: error: libxl.c:5044:libxl_get_physinfo: getting physinfo: Permission denied 2015-05-26 15:33:23 failed to free memory for the domain 2015-05-26 15:33:26 ops 2015-05-26 15:34:10 xl list 2015-05-26 15:34:11 libxl: error: libxl.c:669:libxl_list_domain: getting domain info list: Permission denied 2015-05-26 15:34:13 libxl_list_domain failed. 2015-05-26 15:35:43 i guess upgrading from 4.4.2 to 4.5.0 isn't as smooth as I had hoped 2015-05-26 15:39:45 woooo 3.2! 2015-05-26 15:39:55 celebrating my first day of voluntary funemployment with a new alpine release 2015-05-26 15:48:13 http://alpinelinux.org/downloads/ 2015-05-26 15:48:16 updated 2015-05-26 15:48:24 even the uboot showed up :) 2015-05-26 15:48:41 hum 2015-05-26 15:49:04 where is the xen image? 2015-05-26 16:06:39 \o/ 2015-05-26 16:11:32 \o/ 2015-05-26 17:02:50 Hmm, webkitgtk doesn't have a package with debug symbols? 2015-05-26 17:03:16 This package takes years to build... 2015-05-26 17:40:07 ahills, do you really intend to debug webkitgtk' 2015-05-26 17:40:10 s/'/? 2015-05-26 17:41:20 (but if you do, pleuse keep the package at hand :>) 2015-05-26 17:44:53 I really, really want to find out why I can't run surf 2015-05-26 17:44:56 It's the browser I need 2015-05-26 17:45:09 I might just be setting up a chroot for it, but that would be unfortunate 2015-05-26 17:49:07 Humm, it runs fine here, what's the error? 2015-05-26 17:50:14 surf? 2015-05-26 17:50:17 It just segfaults 2015-05-26 17:50:24 This is on x86_64, haven't tried other arch 2015-05-26 17:55:21 quinq: Are you building surf with any special flags? 2015-05-26 17:58:58 Sorry, I was heating food 2015-05-26 18:00:15 x86_32 here, no additional flag 2015-05-26 18:00:39 It segfaults as soon as you launch it? 2015-05-26 18:01:07 Where does the segfault happen, in surf? 2015-05-26 18:03:13 (using git version btw) 2015-05-26 18:05:21 Yeah, I'm using git 2015-05-26 18:05:35 I can't really tell; I think it's in webkitgtk, but the backtrace is all ?? 2015-05-26 18:06:12 Well, you can build sur with debugging symbols and at least see where it segfaults from there 2015-05-26 18:06:33 s/sur/&f 2015-05-26 18:13:38 I did that, hold on one sec 2015-05-26 18:14:21 http://sprunge.us/RRTJ 2015-05-26 18:14:32 that's the output of bt 2015-05-26 18:14:36 in gdb ./surf 2015-05-26 18:15:06 Oh yes right, sorry 2015-05-26 18:15:14 That's a grsec issue/feature 2015-05-26 18:15:26 Oh, forgot about that 2015-05-26 18:15:35 I can turn that off via PaX though, right? 2015-05-26 18:15:38 You have to authorize somewhat the binary to be scanned by gda 2015-05-26 18:15:39 gdb 2015-05-26 18:15:40 yeah 2015-05-26 18:17:09 Wait, when I build it by hand it segfaults :> 2015-05-26 18:18:04 How were you running it before? Is there a package/ 2015-05-26 18:18:22 Well, I made one 2015-05-26 18:21:33 ok sorry, nevermind what I said, I was running the vanilla kernel on which I'm testing things… 2015-05-26 18:22:02 It segfaults on grsec indeed 2015-05-26 18:30:26 It doesn't segfaults when disabling MPROTECT 2015-05-26 18:30:43 But it doesn't quit as it should, the process hangs 2015-05-26 18:32:41 kill -KILL 1666, that's an evil process 2015-05-26 18:34:46 lol 2015-05-26 18:35:00 so webkit is doing something weird then... 2015-05-26 18:36:19 It happens in runscript() 2015-05-26 18:36:52 oh wow, MPROTECT actually makes it work perfectly 2015-05-26 18:37:02 disabling mprotect that is 2015-05-26 18:37:17 yeah? 2015-05-26 18:37:24 Yeah... you're a genius 2015-05-26 18:37:41 Here, the process is still running, though the window has been destroyed 2015-05-26 18:37:49 Not for me 2015-05-26 18:37:57 Let me try in a more advanced setup, in tabbed and so 2015-05-26 18:38:24 Do you know if other webkitgtk-based browsers works? 2015-05-26 18:38:30 (I'm new here) 2015-05-26 18:38:47 No, I only use surf 2015-05-26 18:38:59 I have a wrapper that works nicely for me 2015-05-26 18:39:11 https://github.com/ahills/browse 2015-05-26 18:39:14 (shameless plug, etc) 2015-05-26 18:41:20 I keep all in ~/.surf ^^ 2015-05-26 18:42:06 I prefer to keep my web applications separate 2015-05-26 18:42:32 Still working on a way to block those really slow ad scripts, probably should just do /etc/hosts adblocking 2015-05-26 18:43:25 midori works without segfaults 2015-05-26 18:43:55 Personally, I only block ads on website I visit often with css styles 2015-05-26 18:44:03 Well, I hide them, don't really block them 2015-05-26 18:44:43 hehe, MPROTECT is disabled on midori 2015-05-26 18:48:20 in the midori APKBUILD: # PAX does not like the JIT, paxmark -mp "$pkgdir"/usr/bin/midori 2015-05-26 18:50:12 Yeah, probably anything with JIT will need PaX disabled... 2015-05-26 19:03:59 Is my apk syntax wrong? I want to install specific versions of redis, like 3.0 or 2.9, but apk complains that versions other than 2.8.17-r0 "breaks: world". 2015-05-26 19:06:26 And what's your apk syntax? 2015-05-26 19:19:23 apk add --update redis=3.0.1-r0 2015-05-26 19:19:48 results in: https://www.irccloud.com/pastebin/xVkmNck0 2015-05-26 19:28:20 I get a similar "breaks: world" error when I try to `apk add --update ruby=2.0.0_p481` 2015-05-26 19:30:19 preyalone: if you've already installed it and/or other installed packages have an explicit version dependency, you may need to resolve that first 2015-05-26 19:32:22 i have not... i'm doing this in a docker container FROM alpine (nothing but OS) 2015-05-26 19:33:29 check the /etc/apk/world file for redis/ruby? 2015-05-26 19:39:42 cat /etc/apk/world => alpine-base 2015-05-26 19:41:11 In alpine v3.1, alpine-base doesn't include redis or ruby by default, right? 2015-05-26 19:43:32 strange, if i apk add --update =, most of the versions listed on aports are defaulted to something else. like, redis 3.0 is somehow interpreted as redis 2.8.17-r0, followed by "breaks: world"... 2015-05-26 19:48:21 preyalone - you may need to pin the edge / testing repos & apk add pkg@edge 2015-05-26 19:49:31 http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management#Repository_pinning 2015-05-26 19:52:30 redis-3.0.1-r0 is in edge 2015-05-26 19:59:12 redis-2.8.17-r0 in alpine 3.1 & 3.0.1-r0 in Alpine 3.2 (& in edge) 2015-05-26 20:21:26 Oh in that case it's likely preyalone only has the edge repo 2015-05-26 20:49:33 ahills: Thanks, I'll try that! 2015-05-26 20:51:18 so of all the redis versions, only 2.8 ever made it past edge into the main repo? 2015-05-26 20:52:07 No, 3.0.1 is now in main 3.2 repo 2015-05-26 20:52:39 lol, i'll ping the docker hub maintainers to update their alpines to 3.2 2015-05-27 01:56:53 whaletechno: I need to know the origin of your name 2015-05-27 02:58:32 ahills: I like techno 2015-05-27 11:54:49 git add -p gives an error: git: 'add--interactive' is not a git command. See 'git --help'. 2015-05-27 11:55:50 (obviously also git add -i) 2015-05-27 11:55:59 <_ikke_> Seems like a broken installation 2015-05-27 11:56:05 apk add git ;) 2015-05-27 11:56:26 <_ikke_> what does ls $(git --exec-path) return 2015-05-27 11:57:09 <_ikke_> ie, does it return git-add--interactive? 2015-05-27 11:58:10 No, is that there in a normal installation? 2015-05-27 11:58:54 Maybe this one is included in the git-perl subpackage, hold on 2015-05-27 11:59:37 Weird, my git package is broken in a different way... it won't upgrade to 2.4.2 from 2.4.1 2015-05-27 12:01:57 apk add -u git yields "OK: 1315 MiB in 536 packages" 2015-05-27 12:02:16 the installed version is definitely 2.4.1, and the latest available is definitely 2.4.2 2015-05-27 12:02:36 <_ikke_> the exec-path should contain all subcommands 2015-05-27 12:02:56 <_ikke_> But it can indeed be part of the perl packge 2015-05-27 12:03:19 I'm surprised that git-add--interactive is a separate subcommand 2015-05-27 12:03:28 But, I am no expert on git internals 2015-05-27 12:04:26 I can't install git-perl because apk is confused about versions... 2015-05-27 12:05:40 <_ikke_> ahills: Historically, some commands were implemented in perl or bash 2015-05-27 12:05:53 <_ikke_> For rapid implementation 2015-05-27 12:06:18 <_ikke_> ahills: Slowly, these scripts are being implemente in native c 2015-05-27 12:07:03 Right, I just don't know the specifics of how they break down commands 2015-05-27 12:07:11 e.g. double hyphen in a command seems strange 2015-05-27 12:07:32 I have definitely looked at the Perl source of git commands to figure out what's going on 2015-05-27 12:07:49 git-perl does indeed have the command 2015-05-27 12:07:53 <_ikke_> ahills: Some commands are split in several sub sub commands 2015-05-27 12:08:00 <_ikke_> they are not supposed to be called by users 2015-05-27 12:08:21 <_ikke_> ie, you don't do git add--interactive 2015-05-27 12:09:08 <_ikke_> git rebase is another 2015-05-27 12:09:25 <_ikke_> rebase--am rebase--interactive and rebase--merge 2015-05-27 12:10:05 I suppose the naming scheme differentiates a sub-sub-command from a sub-command with an extra hyphen 2015-05-27 12:10:12 e.g. vs. git-add-interactive 2015-05-27 12:10:52 <_ikke_> yes 2015-05-27 12:12:48 now... which package is gitk in? 2015-05-27 12:16:31 <_ikke_> Can't seem to find it 2015-05-27 12:17:24 Oh man, git has officially moved to github! 2015-05-27 12:17:35 kernel.org is losing the popularity battle! 2015-05-27 12:23:41 Ah, it's in the makeopts 2015-05-27 12:23:48 NO_TCLTK=YesPlease 2015-05-27 12:24:10 <_ikke_> ah, i see 2015-05-27 12:24:33 <_ikke_> But i guess it brings extra depedencies 2015-05-27 12:24:38 <_ikke_> tk 2015-05-27 12:30:26 what are the lowest lowend hardware (cpu) people are using alpine for? 2015-05-27 12:30:41 we need change -march=i486 to at least -march=i586 2015-05-27 12:31:35 the lowest of mine is amd64... 2015-05-27 12:32:08 which package depends on i586 specifics? 2015-05-27 12:32:25 mesa 2015-05-27 12:32:28 mono 2015-05-27 12:32:35 sems something iirc 2015-05-27 12:40:28 unrelated, would you accept (in the distant future) a git aport with git-gui and gitk split into a subpackage? 2015-05-27 12:45:32 or separate subpackages, either is easy 2015-05-27 13:13:30 Well ncopa, please don't go above i586, i486 builds let me install alpine on my old Geode‐based hardware 2015-05-27 13:13:51 (IIRC it doesn't event support the full i586 set) 2015-05-27 13:15:37 quinq, ALIX is Geode 2015-05-27 13:15:47 it should be ok with i586... 2015-05-27 13:17:26 It's not an ALIX 2015-05-27 13:17:35 It's an IBM ThinClient 2015-05-27 13:17:49 Not AMD Geode, Cyrix Geode 2015-05-27 13:19:05 oh 2015-05-27 13:19:14 yes, cyrix had those funky things 2015-05-27 13:19:22 probably the same as in soekris 2015-05-27 13:23:48 Well, that's the processor before Net4801 2015-05-27 13:24:07 Which uses NSSC1100 2015-05-27 13:24:22 But should be very similar, yes 2015-05-27 13:29:51 On the other hand, that's not really a wide usage, I'd understand if it'd be left behind in favor of 5/686 2015-05-27 14:13:29 quinq: so you do use that geode based i486 2015-05-27 14:14:15 quinq: can you use -march i586? 2015-05-27 14:41:57 IIRC it's i586 less some instructions 2015-05-27 14:42:05 Or that was 686 less some instructions… 2015-05-27 14:42:22 na 2015-05-27 14:45:57 I'll have to check, I just brung it back to life after a few years 2015-05-27 14:46:12 maybe we should let it rest in peace :) 2015-05-27 14:46:40 Yeah, nevermind me and this thing :) 2015-05-27 14:46:51 If I need, I'll build its packages 2015-05-27 14:47:51 ok 2015-05-27 14:48:16 well, if there are many of those 2015-05-27 14:48:25 then we might want reconsider keep i486 2015-05-27 14:54:30 cflags-$(CONFIG_MGEODEGX1) += -march=pentium-mmx 2015-05-27 14:54:47 That's how Linux builds it 2015-05-27 14:55:08 So it should supports full i586 2015-05-27 14:56:30 yup 2015-05-27 14:56:34 good 2015-05-27 14:56:43 we should enable i586 asap 2015-05-27 15:02:24 ncopa: will it come into effect next revision of each package or will you be rebuilding *? 2015-05-27 15:03:08 next revision of each package 2015-05-27 15:03:16 we might want force rebuild some package 2015-05-27 15:03:31 we will rebuild * for v3.3 2015-05-27 15:03:47 how long does that take? 2015-05-27 15:04:00 for arm a week maybe 2015-05-27 15:04:05 x86_64 a few days 2015-05-27 15:04:09 oh, that's not bad 2015-05-27 15:04:15 not that bad 2015-05-27 15:04:27 so we can trigger a rebuild of * if we want 2015-05-27 15:04:31 ok gotta go 2015-05-27 15:04:35 I have some spare machines here I could set up if you need extras 2015-05-27 15:23:07 I know most of us despise pulseaudio. But is there an alternative for network sound support ? I want my laptop to be able to play sound from and to forward mic capture to remote machine on my LAN... 2015-05-27 15:25:38 In fact from serveral machines on the LAN. 2015-05-27 15:27:39 jackd? 2015-05-27 15:28:15 there is an APKBUILD for pa in ~/aports/unmaintained - I've build it successfully before & so did someone else 2015-05-27 15:29:43 thanks. jack is in repo. and will have a look at this pa APKBUILD. 2015-05-27 15:30:22 http://jackaudio.org/faq/netjack.html 2015-05-27 15:30:29 bbut first I need to have some sound output on my laptop. still nothing till now. And there is nothing about sound in the wiki. 2015-05-27 15:33:05 hey, anyone tried using alpine as an openvz ct0? 2015-05-27 20:01:36 alpinelinux: 300+ followers on twitter 2015-05-27 20:38:19 hi there, do you know how to enable sound for kvm-quemu-libvirt hosts? 2015-05-27 20:39:16 I've set `vnc_allow_host_audio = 1` in /etc/libvirt/qemu.conf but it doesn't fix the issue 2015-05-27 21:04:10 alacerda: I hope twitter isn't a primary means of distributing news 2015-05-27 21:49:11 is there a way to make /dev/shm read/write from a chroot env? 2015-05-27 22:36:28 What's the best way to execute a command at startup? 2015-05-27 22:37:26 Is there an "rc.local" equivalent that's already in init.d? 2015-05-27 22:40:21 /etc/conf.d/local 2015-05-27 22:43:23 thanks 2015-05-27 22:47:44 Isn't there /etc/local.d/my_command.start ? 2015-05-27 22:48:10 Yeah, I found /etc/local.d by accident, and the README is clear 2015-05-28 03:40:29 Hi all, I can't install python-bcrypt through pip on Alpine. I'm getting a compliation error "unknown type name 'u_int8_t'" 2015-05-28 03:41:23 trevorj: most things require libc-dev to build 2015-05-28 03:41:33 ahills: Is that no part of build-base? 2015-05-28 03:41:34 trevorj: build-base is a good metapackage (if I didn't get the name wrong) 2015-05-28 03:41:38 not* 2015-05-28 03:41:40 Oh, I don't know 2015-05-28 03:41:43 I assume it is 2015-05-28 03:42:01 I'd assume so as well. u_iny8_t is usually om sys/types iirc 2015-05-28 03:42:04 You probably also need bcrypt-dev, if bcrypt is a package 2015-05-28 03:42:14 Well, bcrypt is not a package 2015-05-28 03:42:15 u_int8_t is a convoluted name; the C standard has uint8_t 2015-05-28 03:42:50 I see, so u_int8_t is a glibcism perhaps? 2015-05-28 03:43:20 maybe? I've never seen that before except in shitty custom type definitions at work 2015-05-28 03:43:25 by people who hate standards 2015-05-28 03:43:39 Lol. That sounds horrible. 2015-05-28 03:43:44 literally #typedef u_int8_t uint8_t... I hate programmers 2015-05-28 03:43:52 Ugh 2015-05-28 03:43:53 er 2015-05-28 03:43:55 ignore the # 2015-05-28 03:44:00 Yeah ;) 2015-05-28 03:44:36 has pip been made not bad yet? 2015-05-28 03:44:51 or is it still impossible to fetch and build sources without installing? 2015-05-28 03:46:19 Why would you want to fetch and build without installing? 2015-05-28 03:46:26 to inspect the build 2015-05-28 03:46:55 Ah, I just do that in an ephemeral virtualenv 2015-05-28 03:46:56 pip has given me nothing but trouble, just like every other language's package manager 2015-05-28 03:47:10 yeah, but having to set up a venv every time I want to do anything with a python package is such a pain in the ass 2015-05-28 03:47:14 pip is horrible, just like every other language's package manager 2015-05-28 03:47:37 someday one of them will understand that languages shouldn't have them 2015-05-28 03:47:51 but as long as people are using crippled operating systems, I guess we're stuck... 2015-05-28 03:48:02 The issue is time to delivery I believe 2015-05-28 03:48:28 and the solution is to litter my /usr/local with shit 2015-05-28 03:48:44 Oh, that I disagree with fully 2015-05-28 03:48:45 I guess I don't mind golang's solution 2015-05-28 03:49:11 pip supports user installs as well, plus every language has virtualenv-esque support these days 2015-05-28 03:49:21 I like the way node handles it to be honest 2015-05-28 03:49:23 heh, rbenv... 2015-05-28 03:49:26 oh god 2015-05-28 03:49:33 yeah I love how npm installs each dependency separately 2015-05-28 03:49:36 even if it's the same package 2015-05-28 03:49:52 I've had one package with many dependencies end up downloading the exact same code and building it over a hundred times 2015-05-28 03:50:00 then I never used node again and I've been so, so happy with computers 2015-05-28 03:50:24 Lol, at least npm has signature checking nowadays doesn't it? 2015-05-28 03:50:40 Only took years of everyone screaming about it 2015-05-28 03:50:42 lol 2015-05-28 03:50:58 nope, pip still isn't leaving the failed build anywhere on my disk... 2015-05-28 03:51:49 There's an option for that 2015-05-28 03:51:49 --no-clean 2015-05-28 03:51:55 Yeah 2015-05-28 03:52:11 god knows why the default behavior for failure isn't to leave it there... 2015-05-28 03:54:22 probably some older version of Python.h included sys/types.h 2015-05-28 03:54:28 and this depends on it because it's poorly written 2015-05-28 03:54:39 is there a system you're able to build this on? 2015-05-28 03:55:08 Every system that's not Alpine ;) 2015-05-28 03:55:42 really? 2015-05-28 03:55:55 can you be sliiiiightly more specific? 2015-05-28 03:56:06 Ubuntu 14.04.1, CentOS 6 2015-05-28 03:56:08 I'm not able to build this on my arch box either 2015-05-28 03:56:18 Let me verify before speaking about others 2015-05-28 03:57:11 Yeah, just tested on Ubu 14.04.1 2015-05-28 04:03:09 in pyport.h, line 347, I see this comment: 2015-05-28 04:03:21 /* NB caller must include */ 2015-05-28 04:03:51 so, I claim incorrect behavior for python-bcrypt, and silly behavior for other systems 2015-05-28 04:04:26 That's there on my ubu system as well 2015-05-28 04:04:43 I'm not seeing anything special in the APKBUILD for python either 2015-05-28 04:05:57 ok, so the build command is this: 2015-05-28 04:06:08 gcc -fno-strict-aliasing -Os -fomit-frame-pointer -DNDEBUG -Os -fomit-frame-pointer -fPIC -I/usr/include/python2.7 -c bcrypt/bcrypt_python.c -o build/temp.linux-x86_64-2.7/bcrypt/bcrypt_python.o 2015-05-28 04:06:29 can you drop into the build directory on ubuntu and add -E to that, so we can see where it's getting defined? 2015-05-28 04:09:40 sys/types.h is included ifdef HAVE__GETPTY in pyport.hgcc -fno-strict-aliasing -Os -fomit-frame-pointer -DNDEBUG -Os -fomit-frame-poin 2015-05-28 04:09:55 Sorry, pasted wrong 2015-05-28 04:10:06 ah, hmm 2015-05-28 04:10:08 /usr/include/x86_64-linux-gnu/sys/types.h 2015-05-28 04:10:29 so HAVE__GETPTY is defined... where? 2015-05-28 04:10:36 Oh, it's not 2015-05-28 04:10:41 That was something I typed that I figured out was wrong 2015-05-28 04:10:43 is _getpty a glibc function? 2015-05-28 04:10:53 oh 2015-05-28 04:10:57 i don't think so 2015-05-28 04:11:00 not sure tho 2015-05-28 04:11:02 Then I pasted it in by accident, lol 2015-05-28 04:11:59 I don't really know what -E does, but I'm assuming it shows a tree of defs? 2015-05-28 04:12:16 it does preprocessing 2015-05-28 04:12:24 so all #defines, #includes, etc 2015-05-28 04:12:31 but then doesn't compile, just spits out the text 2015-05-28 04:12:45 Ah, I've always wondered how to find where a define comes from 2015-05-28 04:13:07 # 1 "/usr/lib/gcc/x86_64-linux-gnu/4.8/include/stddef.h" 1 3 4 2015-05-28 04:13:11 # 147 "/usr/include/x86_64-linux-gnu/sys/types.h" 2 3 4 2015-05-28 04:13:23 That's what's above the def for u_int8_t 2015-05-28 04:13:44 it must be included via Python.h 2015-05-28 04:13:52 is it possible to trace it back further than that? 2015-05-28 04:14:02 actually, could you sprunge/pastebin/whatever the whole output? 2015-05-28 04:14:08 Sure 2015-05-28 04:14:17 I don't really understand the details of the output format enough to provide "general guidance" of any sort 2015-05-28 04:14:24 Hehe, understand 2015-05-28 04:16:19 http://paste.ubuntu.com/11404479/ 2015-05-28 04:29:43 this is the first time I've wished I had an ubuntu machine around 2015-05-28 04:29:51 I used to have many, but I quit my job 2015-05-28 04:30:23 don't feel like going through the installation process... 2015-05-28 04:30:36 docker ;) 2015-05-28 04:33:39 maybe when I have more energy... it's getting late... 2015-05-28 04:33:54 anyways, this file makes it clear that stdlib.h is pulling in sys/types.h on ubuntu 2015-05-28 04:34:11 could you take a peek in there? 2015-05-28 04:34:26 this is probably a glibc thing... 2015-05-28 04:34:56 stdlib.h pulling in sys/types.h is a "known glibc feature" 2015-05-28 04:35:05 and sys/types.h in turn pulls in sys/sysmacros.h 2015-05-28 04:35:12 which defines macros named major() and minor() 2015-05-28 04:35:17 your best bet may be to create a quick apkbuild that patches in a quick #include into bcrypt/bcrypt_python.c 2015-05-28 04:35:26 unless you need it in a venv 2015-05-28 04:36:09 is musl aware of this difference, and have they commented on their stance? 2015-05-28 04:37:12 ? 2015-05-28 04:37:36 you mean not pulling in sys/types.h 2015-05-28 04:37:38 ? 2015-05-28 04:37:45 yeah 2015-05-28 04:37:49 Ah, thanks guys 2015-05-28 04:38:02 since it's behavior different from glibc, even though glibc is (as usual) incorrect 2015-05-28 04:38:07 That's the issue, is that I'm pulling this as a python dependency for a custom python package 2015-05-28 04:38:38 Into a virtualenv as a part of my deployment process 2015-05-28 04:38:44 musl aims to reproduce actual interfaces/apis in _BSD_SOURCE(default)/_GNU_SOURCE profiles, but not to reproduce bug-compatibility & namespace-pollution-cruft 2015-05-28 04:38:47 :) 2015-05-28 04:39:06 so, is the recommended solution here a patch? 2015-05-28 04:39:07 it could be added to differences from glibc on the wiki if it's not already there 2015-05-28 04:39:19 or can pulling that nonsense in be accomplished via a simple -D 2015-05-28 04:39:20 the recommended solution is to include sys/types.h if it's needed 2015-05-28 04:39:38 right, i.e., compliant C code 2015-05-28 04:39:42 but even better would be using the standard C types from stdint.h (uintN_t) rather than the legacy nonstandard ones (u_intN_t) 2015-05-28 04:41:14 I might actually have access to one of the maintainers of python-bcrypt 2015-05-28 04:41:56 btw it should just expose libc crypt_r's bcrypt if available, no? 2015-05-28 04:42:35 probably, but with python developers, who knows... 2015-05-28 04:43:05 trevorj: any particular reason you're using python-bcrypt instead of bcrypt? 2015-05-28 04:43:26 the latter seems to be more actively maintained 2015-05-28 04:44:02 also it builds for me :) 2015-05-28 04:44:11 it's ridiculous that musl is the only linux libc with bcrypt... 2015-05-28 04:44:27 should add that to the libc-comparison 2015-05-28 04:44:33 but it gives us hope for the future that at least someone cares 2015-05-28 04:44:47 how many core developers on the musl project? 2015-05-28 04:46:24 trevorj: the github page for the module you're using literally describes the project as "Deprecated and out of date" 2015-05-28 04:46:29 https://github.com/DisruptiveLabs/python-bcrypt 2015-05-28 04:46:46 trevorj: strongly recommend switching libraries 2015-05-28 04:47:35 ahills, according to openhub's stats, i'd say just 1-2 2015-05-28 04:48:04 but there are a number of additional contributors who aren't actively writing code but involved in discussion of development 2015-05-28 04:48:25 or who have written code but only contribute to isolated parts of musl 2015-05-28 04:49:47 maybe about 10 2015-05-28 04:51:11 dalias: how many people with push access, I guess is the question 2015-05-28 09:10:07 hiu 2015-05-28 09:10:08 hi 2015-05-28 09:10:18 I get this error when installing shadow on edge 2015-05-28 09:10:20 ERROR: shadow-4.2.1-r0: trying to overwrite sbin/nologin owned by util-linux-2.26.2-r0. 2015-05-28 09:11:20 Adran, found you 2015-05-28 11:18:10 Bureaucat: could you please file a bug for that? 2015-05-28 11:18:15 so i dont forget it 2015-05-28 11:18:39 shadwo is not in 3.2? 2015-05-28 11:18:41 only edge? 2015-05-28 11:19:10 dont bother file bug then 2015-05-28 11:19:12 i'll just fix it 2015-05-28 11:24:38 Bureaucat: shoudl be fixed now 2015-05-28 11:27:52 darkfader, kexec is working for vanilla, you were interested iirc 2015-05-28 12:38:09 ncopa: did $CBUILD/$CHOST change in v3.2? 2015-05-28 12:38:35 ahills: no 2015-05-28 12:39:12 hmm, some configures are complaining about x86_64-alpine-linux-musl having an extra - 2015-05-28 12:48:20 example? 2015-05-28 12:49:14 http://wiki.osdev.org/Target_Triplet 2015-05-28 12:51:35 Figured it out; it's just an ancient config.sub 2015-05-28 12:51:48 that doesn't recognize linux-musl as a kernel-os part 2015-05-28 12:52:48 god I hate autotools 2015-05-28 13:00:15 indeed :) 2015-05-28 13:41:47 thanks for merging imapfilter 2015-05-28 13:42:03 I have a patch for the flushing problem, will submit in a sec 2015-05-28 13:42:15 I forgot to supersede my old one 2015-05-28 13:42:49 my aports tree got a little messy when I accidentally committed a bunch of stuff to master instead of another branch :/ 2015-05-28 13:58:27 autotools can make our 24 core buildserver as slow as my eeepc 2015-05-28 14:01:43 haha 2015-05-28 14:01:54 running thousands of instances of m4 2015-05-28 14:03:10 ... sequentially 2015-05-28 14:03:41 if only replacing configure in existing packages were as "easy" as replacing glibc with musl 2015-05-28 14:04:22 it is not 2015-05-28 14:04:26 and yeah 2015-05-28 14:04:54 it would be *awesome* to be able to take a configure.ac and convert it to something that works fast 2015-05-28 14:09:31 ahills: >>> xbindkeys: Fetching http://www.nongnu.org/xbindkeys/xbindkeys-1.8.6.tar.gz 2015-05-28 14:09:31 % Total % Received % Xferd Average Speed Time Time Time Current 2015-05-28 14:09:31 Dload Upload Total Spent Left Speed 2015-05-28 14:09:31 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0curl: (22) The requested URL returned error: 500 Internal Server Error 2015-05-28 14:09:59 :/ sorry, nothing I can do about that... 2015-05-28 14:11:01 can you just leave it and hope savannah starts working again? I'll message the owners 2015-05-28 14:11:28 ok 2015-05-28 14:13:59 heh, had to register on savannah to get his email... 2015-05-28 14:21:41 ncopa: it could use http://git.savannah.gnu.org/cgit/xbindkeys.git/snapshot/xbindkeys-1.8.6.tar.gz instead, but I don't know whether savannah cgit is more or less reliable than the other part of the site... 2015-05-28 14:24:39 we can wait a bit 2015-05-28 14:25:29 the author's email is @free.fr, so hopefully he is awake and checking mail 2015-05-28 14:45:23 ahills: Yeah, it's a dependency of flask-bcrypt 2015-05-28 14:45:42 ahills: I personally use bcrypt 2015-05-28 14:46:35 tell them to update! 2015-05-28 14:46:45 FOSS ftw, send more emails 2015-05-28 14:47:15 I'll just make a pull request 2015-05-28 14:47:53 I was just trying to avoid kludging something over a minor dependency of a dependency 2015-05-28 18:30:17 I find that amazing to find bashisms in linux kernel build scripts 2015-05-28 18:33:27 ... 2015-05-28 18:33:34 that's not the worst 2015-05-28 18:33:50 they periodically add perl to the build scripts... 2015-05-28 18:33:57 then someone complains and it gets removed again 2015-05-28 18:33:58 repeat 2015-05-28 18:33:59 :( 2015-05-28 18:34:13 haha 2015-05-28 18:39:53 I'd rather have Perl than bash 2015-05-28 18:41:20 Bash only has half-assed attempts at implementing basic language features and it's already too bloated to use as a shell, so you may as well use a full language 2015-05-28 18:42:17 Fallacy of the false dilemma, etc 2015-05-28 18:44:31 Is busybox ash actually dash? 2015-05-28 18:44:52 no 2015-05-28 18:45:00 they both have common history (ash) 2015-05-28 18:45:04 dash is "debian ash" 2015-05-28 18:45:49 ah, thanks 2015-05-28 18:45:58 I wish ash had a manual page... 2015-05-28 18:46:50 http://pubs.opengroup.org/onlinepubs/9699919799/utilities/sh.html and http://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html ;-) 2015-05-28 18:47:00 Maybe http://pubs.opengroup.org/onlinepubs/9699919799/xrat/V4_xcu_chap02.html 2015-05-28 18:47:07 oups, too slow ^^ 2015-05-28 18:47:21 :) 2015-05-28 18:49:18 does debian ash deviate far from POSIX or could its man page be a good reference? 2015-05-28 18:50:27 the wikipedia page is unsurprisingly unhelpful 2015-05-28 18:51:25 http://gondor.apana.org.au/~herbert/dash/ “DASH is a POSIX-compliant implementation of /bin/sh” 2015-05-28 18:51:59 Right, but does POSIX compliance mean no additional features? 2015-05-28 18:52:22 No, it means that it supports all POSIX features and behaves as specified 2015-05-28 18:52:52 OK, so what I'm looking for is a list of features that exceed the POSIX specification, if any... 2015-05-28 18:52:55 So if you write POSIX shell scripts, it's supposed to run fully on dash 2015-05-28 18:53:00 Right 2015-05-28 18:53:02 Why would you want that? 2015-05-28 18:53:45 I want to know how the dash shell is different from the standard, including features it includes as extras 2015-05-28 18:53:54 I've been surprised by features of shells in the past... 2015-05-28 18:54:04 Yes, but why do you want to know that? 2015-05-28 18:54:14 Because surprises are always bad 2015-05-28 18:54:24 I'd rather have this document for ash, but busybox is very light on documentation 2015-05-28 18:54:35 How could you be surprised if you write POSIX shell grammar? 2015-05-28 18:54:51 I really don't understand what you mean 2015-05-28 18:55:23 dash is (at least it says it is) POSIX compliant, how could you encounter “surprises”? 2015-05-28 18:55:50 If it misinterpret your scripts, then it's a bug 2015-05-28 18:56:44 The short version is that, if it's exactly POSIX-compliant, no more, no less, then apk add dash-doc is sufficient for documenting busybox ash, which is supposedly also POSIX-compliant 2015-05-28 18:57:04 busybox ash has one conformance bug if you turn on bash compatibility 2015-05-28 18:57:34 compliant is compliant, there's no “more or less”, or it's a joke (or bug) 2015-05-28 18:57:35 foo&>bar gets parsed as the tokens 'foo', '&>', 'bar' where the middle is some compound-redirection hack 2015-05-28 18:58:09 So, if something adds a feature, say an extra command, is it out of compliance? 2015-05-28 18:58:15 it should be parsed as 'foo', '&', '>', 'bar' (run foo in the background with no redirection, and truncate bar in for foreground) 2015-05-28 18:58:26 no it isn't 2015-05-28 18:58:39 compliance means at least support what's required 2015-05-28 18:58:47 ahills, no, extra commands are ok as long as the standard commands work and you can invoke programs via full pathnames or such 2015-05-28 18:59:50 OK, so let's say dash adds another command, and I have a mistake in a script where I am naming, e.g., a function after that command, but I leave out the function--if my script does not fail because that function is missing, because it is instead executing the shell's command, that is a "surprise feature" I would like to avoid, just by knowing ahead of time 2015-05-28 18:59:50 (via a complete manual) 2015-05-28 19:00:22 It's not a surprise, it's your fault for writing non‐POSIX compliant script 2015-05-28 19:00:44 It's a surprise when the name is of an existing command, instead of failing because of the missing symbol 2015-05-28 19:01:25 Obviously there is a bug in the script--the reason I want to know the extent of the differences between the POSIX specification and the shell's complete feature set is to anticipate "silent failures" of this sort 2015-05-28 19:01:40 Something that doesn't fail when it ought to is a very nasty surprise 2015-05-28 19:02:50 Understanding the complete behavior of a program is why I prefer reading documentation over stabbing in the dark (and is one of the reasons I'll probably always be using OpenBSD on at least one machine) 2015-05-28 19:03:18 Well, if you really want to be sure, then read the full sources of dash 2015-05-28 19:03:33 You can't be really sure else (and even if you do) 2015-05-28 19:04:17 Undocumented features are a bug in the documentation 2015-05-28 19:04:51 Yep, that's why I'm suggesting to you to read the code 2015-05-28 19:05:01 s/to/ 2015-05-28 19:05:47 Sure, and let's just get rid of manuals altogether while we're at it 2015-05-28 19:05:55 If it can't be perfect, let's just not do it at all 2015-05-28 19:06:47 A shell is not a script testing tool 2015-05-28 19:07:14 No, but it is a program that is complex enough that documentation is incredibly useful 2015-05-28 19:07:15 If you want to be sure your script if fully POSIX compliant, check it 2015-05-28 19:07:39 POSIX compliant and bug-free are not equivalent 2015-05-28 19:09:05 Well, the bug would be in your script semantics, a shell won't solve that 2015-05-28 19:09:30 Right, but a fully documented shell would make it easy to understand ahead of time the sorts of things I might run into 2015-05-28 19:09:56 This is why, to answer your original question, I'd want documentation of deviations from POSIX that include features added 2015-05-28 19:10:38 Yes, I understood that you want to use dash as a script testing tool :) 2015-05-28 19:10:47 POSIX is not some pleasant dogma that enlightens me; it is just a standard protocol for command execution 2015-05-28 19:10:58 No, I want to use the dash manual as an ash manual... 2015-05-28 19:11:56 A script that calls a non-existent function is fully POSIX-compliant, but a POSIX-compliant shell may not indicate a failure if, e.g., it adds a command by that same name 2015-05-28 19:12:23 That behavior is more or less unavoidable, but it's ideal to have a reference for it 2015-05-28 19:13:03 That behavior is fully unavoidable if you read posix documentation 2015-05-28 19:13:22 How would reading the POSIX specification prepare me for a non-POSIX feature? 2015-05-28 19:14:02 Because you'll know then that this non-POSIX feature you use is not POSIX 2015-05-28 19:14:48 Right, but how will I know I'm using the feature if it's not documented? 2015-05-28 19:14:50 It's trial by fire 2015-05-28 19:15:03 Shells are exactly the place on my system where concise and complete documentation is most wanted 2015-05-28 19:16:13 If it's not documented in POSIX, then you know that your script isn't compliant 2015-05-28 19:16:40 How would this scenario be the fault of the script? A script that calls a non-existent function is POSIX-compliant, and it should fail 2015-05-28 19:17:01 no 2015-05-28 19:18:06 http://pubs.opengroup.org/onlinepubs/9699919799/utilities/sh.html << looking at this specification 2015-05-28 19:21:50 dalias, I agree that busybox shamely lacks documentation, but I think your approch is wrong 2015-05-28 19:22:28 Write your scripts as they should be, check them, then again. 2015-05-28 19:22:35 I think you mean ahills 2015-05-28 19:22:46 Yes I did :) 2015-05-28 19:23:11 My approach is to understand my tool when I use it 2015-05-28 19:23:26 My preferred method of gaining this understanding is to read its reference 2015-05-28 19:23:26 If they misbehave somewhere (they shouldn't, so it's a bug on your script or the shell), then check again, and patch either 2015-05-28 19:23:53 After some years of writing software, strict compliance is not all that I require of my tools 2015-05-28 19:23:58 Oh, so you intend to write busybox ash scripts, not posix scripts ? 2015-05-28 19:24:09 I intend to write both, now and in the future 2015-05-28 19:24:18 Ok, I didn't understand that 2015-05-28 19:24:21 I also intend to write compliant Perl, Python, Ruby, ... 2015-05-28 19:24:50 In bash more than any other tool, I've encountered unexpected behavior not explained by the documentation, with the additional disadvantage that even in compliance mode it does not comply to the standard 2015-05-28 19:25:10 and I rely on my shell(s) to keep my systems together 2015-05-28 19:25:21 Yeah, don't use bash 2015-05-28 19:25:42 had I known about dash/ash fifteen years ago, I may have been using them instead 2015-05-28 19:25:50 c'est la vie 2015-05-28 19:26:53 the reason I still prefer Lisp to Ruby is that Lisp has proper debugging tools built in 2015-05-28 19:27:05 but the debugger is not really part of the language 2015-05-28 19:28:06 http://www.in-ulm.de/~mascheck/various/ash/#busybox 2015-05-28 19:29:55 http://git.busybox.net/busybox/tree/shell/README 2015-05-28 19:30:25 suddenly sonera decided to cut me completely 2015-05-28 19:30:50 also, as an aside: got rpi running alpine 2015-05-28 19:31:26 me too! 2015-05-28 19:31:38 quinq: thanks, these are useful 2015-05-28 19:31:55 I hope, that README is five years old ^^ 2015-05-28 19:32:22 so, my worst case scenario is now five years of git commit logs, instead of thirty 2015-05-28 19:32:26 Let's hope that nothing has been added on removed on user side 2015-05-28 19:32:41 haha yes 2015-05-28 19:35:59 now if only I could get ash to get ^W to stop on the / character, I would need nothing more from this shell... 2015-05-28 19:44:01 btw, how shitty is it to crosscompile glibc code into musl for alpine? 2015-05-28 19:46:18 ? 2015-05-28 19:46:27 crosscompile glibc code into musl? 2015-05-28 19:46:35 more or less 2015-05-28 19:46:41 i don't get what that means 2015-05-28 19:47:05 derp, I mean C that has been written with glibc in mind into muslc 2015-05-28 19:48:39 ok, so not "glibc code" but "code written for use on glibc" 2015-05-28 19:48:48 by crosscompile do you actually mean crosscompile? 2015-05-28 19:48:54 or just native compile on alpine? 2015-05-28 19:49:03 ahills, seems like you'll have to patch libbb/linedit.c 2015-05-28 19:49:07 while (cursor > 0 && BB_isspace(command_ps[cursor-1])) 2015-05-28 19:49:18 It's hardcodded to stop on space 2015-05-28 19:52:40 Yeah, I looked into it a while ago, but then just installed mirksh instead 2015-05-28 19:52:54 haha 2015-05-28 19:52:59 er, mksh 2015-05-28 19:53:15 a big overhead for one feature, but... I have a fast computer 2015-05-28 20:01:43 ahills, http://ix.io/iOf should do the trick 2015-05-28 20:02:04 But there's more work to be done for other vi-like keybindings 2015-05-28 20:03:47 Thanks, I was actually instead going to just test command_ps[cursor-1] != '/', and just not publish the patch 2015-05-28 20:06:05 Ok, but that would have broke word separation 2015-05-28 20:06:17 anyway 2015-05-28 20:08:35 I meant adding that instead of ispunct 2015-05-28 20:09:32 oh right, sorry 2015-05-28 20:38:00 ncopa: xbindkeys site seems to be up again 2015-05-28 21:47:29 Is there a way to instruct apk to always install a -doc package, if available? 2015-05-28 22:17:56 quinq: thx! 2015-05-29 01:37:36 oh man, I can't believe no one's ported sl yet 2015-05-29 01:37:37 proud to be the first 2015-05-29 01:39:46 lol 2015-05-29 01:53:02 i'm trying to use iojs in alpine, but /usr/bin/iojs complains "Error relocating /usr/bin/iojs: _ZSt24__throw_out_of_range_fmtPKcz: symbol not found" 2015-05-29 01:55:59 i'm using the semi-official docker library "alpine" base OS 2015-05-29 01:58:44 it's only the iojs --version flag that triggers this; i can actually use the iojs repl just fine 2015-05-29 02:07:58 preyalone: that looks like C++ munging, good luck debugging... 2015-05-29 02:08:33 unlikely that it's glibc but you never know... do you have libstdc++ installed and all that? 2015-05-29 02:08:38 by the way, could pkgs.alpinelinux.org please make it easier to list *all* versions of a package, not just the latest in main? 2015-05-29 02:09:11 ahills: libstdc++ is a dependency of iojs: http://pkgs.alpinelinux.org/package/testing/x86_64/iojs 2015-05-29 02:09:32 I'm stabbing in the dark here; C++ dependencies can be a nightmare 2015-05-29 02:09:48 might want to build it yourself in case there's a build version problem 2015-05-29 02:09:49 could the iojs require a certain version of alpine? i'm using 3.1 2015-05-29 02:10:02 some C++ software needs to be rebuilt any time a single dependency is built 2015-05-29 02:10:41 gross 2015-05-29 02:11:49 preyalone: if u installed iojs from testing, then u will need 3.2 2015-05-29 02:12:05 try upgrading if that is possible 2015-05-29 02:30:19 pnutzh4x0r: BOOM! that fixed it thanks 2015-05-29 02:32:40 np 2015-05-29 03:01:37 how do i prevent services from automatically starting? e.g., i want to use postgres in an alpine 3.2 docker container 2015-05-29 03:02:40 i dont think anything is automatically started unless u enable them 2015-05-29 03:02:44 alpine uses openrc 2015-05-29 03:03:03 so u can do rc-update/rc-service/etc 2015-05-29 03:06:17 preyalone: rc-update alone will list every automatically enabled service and which runlevel it's set to start in 2015-05-29 03:18:51 my mistake, my postgres configuration was just wrong :P 2015-05-29 03:56:51 which vnc client do you guys use? 2015-05-29 03:56:57 porting tigervnc is a bitch 2015-05-29 07:08:41 ahills: I use remmina 2015-05-29 07:08:53 Is it possible to have a custom apkovl automatically detected by a cdrom launch in KVM? maybe with an usb stick along? 2015-05-29 07:13:09 I want to launch 4 KVM from the same ISO. But I would like to be able to pass them a specific apkovl each... 2015-05-29 07:14:35 And I need to add nomodeset in the boot params... 2015-05-29 07:15:42 Jean-Scotch: I'm not sure I follow 2015-05-29 07:16:08 Jean-Scotch: yes 2015-05-29 07:16:24 kaniini's y'a man :) 2015-05-29 07:17:03 Jean-Scotch: you can actually specify a network URI to download an apkovl from, even. 2015-05-29 07:17:13 Jean-Scotch: which could check what to send based on IP or whatever. 2015-05-29 07:17:55 kaniini: is there some wiki page about thet? 2015-05-29 07:18:12 just for my benefit, how are you launching 4 KVM from the same ISO? 2015-05-29 07:18:34 Jean-Scotch: this will require remastering the CD, see http://wiki.alpinelinux.org/wiki/How_to_make_a_custom_ISO_image 2015-05-29 07:19:04 ScrumpyJack: http://wiki.alpinelinux.org/wiki/User_talk:Jch#My_laptop_setup ;) 2015-05-29 07:19:28 there is also, http://wiki.alpinelinux.org/wiki/PXE_boot 2015-05-29 07:20:00 I was thinking about PXE as I did contribute to that page ;) 2015-05-29 07:20:31 but I wanted to avoid the need of a dhcp server in my laptop ;) 2015-05-29 07:20:49 i dunno. "cloudy" type things really work best with PXE in alpine 2015-05-29 07:21:16 I suppose I will have to follow that path... 2015-05-29 07:21:38 That's what I do with my servers anyway 2015-05-29 07:21:58 and I have bout 80 of them now ;) 2015-05-29 07:28:36 Jean-Scotch: what do you want to use consul for? 2015-05-29 07:30:16 I use mostly consul for naming on the lan. naming services makes life easier ^^ 2015-05-29 07:30:45 I suppose I don't need it on my laptop internal lan with only 4 machines ;) 2015-05-29 07:30:56 ahills: xbindkeys pushed. thanks 2015-05-29 07:31:16 seems like control-w in bash deletes to space too and not to / 2015-05-29 07:31:17 but I will also have a bunch of LXC 2015-05-29 07:32:21 I'm building my laptop a bit like qubeOS with a lot of things contained 2015-05-29 07:33:26 nothing runs on bare metal but openvswitch and qemu 2015-05-29 08:09:22 hey folks 2015-05-29 08:09:49 I was wondering what are the busybox equivalents of "w" and "last" commands ? 2015-05-29 08:28:10 Jean-Scotch: consul RPC protection is coming soon :P 2015-05-29 08:29:01 codedumb: great news ^^ 2015-05-29 08:29:36 I hope for envconsul someday ;) 2015-05-29 08:34:40 Jean-Scotch: ok i get it now, you are booting 4 qemus on a laptop :) 2015-05-29 08:35:33 ScrumpyJack: that's it. but I also manage my servers that way ;) but with PXE in that case. 2015-05-29 08:36:16 pxe in qemu? 2015-05-29 08:36:31 to pxe boot alpine linux in qemu? 2015-05-29 08:37:46 and your desktop is one of your qemu? *blink* 2015-05-29 08:38:34 yes ;) 2015-05-29 08:38:45 how's sound working? 2015-05-29 08:39:17 sound is not woking yet. but I have to path to solution ;) 2015-05-29 08:39:23 I will document it 2015-05-29 08:39:49 libvirt alsa? 2015-05-29 08:39:56 that all sounds a bit crazy :) 2015-05-29 08:40:13 he is crazy indeed 2015-05-29 08:40:18 no libvirt, too high-level for me ;) 2015-05-29 08:41:15 Jean-Scotch: you've got to tell me, why are you doing all this on a laptop? :) 2015-05-29 08:41:38 To have a very very paranoid setup ;) 2015-05-29 08:41:47 so that his wife can't use it 2015-05-29 08:42:19 I also have poweroff on LID cloe, poweroff of insertion/extraction of any peripheral 2015-05-29 08:42:45 so you've got X11 running from a qemu which i presume you'll access remotely? what window manager or desktop environment will you be using? 2015-05-29 08:42:57 only run-from-ram and LUKS volume whend HDD is really needed 2015-05-29 08:43:16 For my self I use awesome 2015-05-29 08:43:26 but I also hace a LXDE desktop 2015-05-29 08:43:41 ok, i was hoping you could try setup-box and mate ) 2015-05-29 08:43:46 I use x2go to access that desktop 2015-05-29 08:43:59 I can try any DE in fact ;à 2015-05-29 08:44:10 One at a time in different KVM ;) 2015-05-29 08:44:18 same $HOME 2015-05-29 08:44:44 Just I cannot have 3D acceleration 2015-05-29 08:44:58 but I don't use it anyway 2015-05-29 08:46:18 foe now, eveything is still running from a usb stick. but when all will be ready I will do a setup-bootable to a small partition on the HDD. 2015-05-29 08:48:23 how is flash videos going ? 2015-05-29 08:48:38 HD videos over x2go ? 2015-05-29 08:55:41 I dunno, I don't use it for multimedia stuff... but I will look soon. But for regular web surfing, even with some light anims, it works well 2015-05-29 09:17:56 /buffer fanf 2015-05-29 09:17:59 oop 2015-05-29 09:36:52 btw upping my question 2015-05-29 09:37:01 I was wondering what are the busybox equivalents of "w" and "last" commands ? 2015-05-29 09:39:01 last works but is defaulting to /dev/null/wtmp 2015-05-29 09:39:13 and /var/log/wtmp is not populated anyway... 2015-05-29 09:39:18 thats a know issue with musl 2015-05-29 09:39:28 or feature if you want 2015-05-29 09:40:16 ah 2015-05-29 09:40:37 http://www.openwall.com/lists/musl/2012/03/04/6 2015-05-29 09:40:45 we should probably start using that libutmpx 2015-05-29 09:42:23 this thread is relatively new: http://www.openwall.com/lists/musl/2015/03/29/1 2015-05-29 09:42:26 wouldn't changing musl default utmp path sufficient ? 2015-05-29 09:43:38 oh ok 2015-05-29 09:44:39 i believe it is intentinally not implemented due to it is impossible to implement it properly or something 2015-05-29 09:45:37 ok 2015-05-29 09:46:48 it's kind of really really really useful to be able to have last working 2015-05-29 09:47:03 instead of having to grep trough logs :P 2015-05-29 09:49:46 yes 2015-05-29 09:49:50 i miss that feature too 2015-05-29 09:56:11 Jean-Scotch: what's your sound workaround ? 2015-05-29 09:56:35 I'm quite interested in this to get sound from VMs on the host 2015-05-29 09:59:08 one "easy" solution would be to use pulseaudio to redirect audio from guests to pulseaudio server on host but ... well pulseaudio 2015-05-29 10:03:23 ncopa: is there any limitation having pulseaudio working on alpine ? or is it just a matter of choice ? 2015-05-29 10:03:46 iirc pulseadui are doing some glibc specific hacks 2015-05-29 10:04:06 so it i think it does not build out of the box 2015-05-29 10:04:28 ok wouldn't surprise me 2015-05-29 10:05:46 https://bugs.freedesktop.org/show_bug.cgi?id=85319 2015-05-29 10:05:49 nice 2015-05-29 10:06:34 padsp does the interception by redefining the open() function. 2015-05-29 10:06:55 thats the problem 2015-05-29 10:07:23 i suppose we could just skip padsp 2015-05-29 10:08:54 http://lists.openembedded.org/pipermail/openembedded-core/2015-April/103615.html 2015-05-29 10:09:03 seems someone already worked on the subject 2015-05-29 10:11:44 i see that 3.2.0 of alpine-linux is out, when about will we see the RP2 version? 2015-05-29 10:11:59 would i be right to assume alpine-rpi-3.2.0-armhf.rpi.tar.gz does not work on RP2? 2015-05-29 10:12:21 i thought fabled said that it was going to happen with the 3.2.X release but i might have been wrong 2015-05-29 10:12:35 sillysausage, it should work 2015-05-29 10:12:42 oh 2015-05-29 10:13:37 we even updated the tooltip ;) 2015-05-29 10:13:52 oh so you did :D 2015-05-29 10:13:59 btw, does that include grsec? 2015-05-29 10:14:03 no 2015-05-29 10:14:10 the u-boot arm image is grsec 2015-05-29 10:14:16 but does not really support rpi 2015-05-29 10:14:16 right 2015-05-29 10:14:30 have not had time to make rpi and grsec patches work together 2015-05-29 10:14:50 fair enough 2015-05-29 10:15:44 ah apparently jack is able to do network audio as well 2015-05-29 10:15:53 should be simple as wiping the sdcard (which has alpine-rpi-150424-armhf.rpi.tar.gz) on it and extracting alpine-rpi-3.2.0-armhf.rpi.tar.gz with the apkprov present. 2015-05-29 10:17:26 yes 2015-05-29 11:26:11 ncopa / coredumb: suckless has got a utmp implementation 2015-05-29 11:26:51 ahills: i saw a minimalistic implementaion in the mailing list up there too 2015-05-29 11:26:58 ncopa: yeah most shells ^W to space, but OpenBSD's ksh doesn't, and that's preferable to me 2015-05-29 11:27:18 Heh that mailing list... 2015-05-29 11:27:30 i think ash mimic bash 2015-05-29 11:27:37 Yeah 2015-05-29 11:27:49 ^W deletes to / sounds much more sane to me :) 2015-05-29 11:28:01 Bash is technically configurable but... Good luck 2015-05-29 11:28:08 heh 2015-05-29 11:28:20 i looked at ash this morning 2015-05-29 11:28:41 and found a way to save 35 bytes :) 2015-05-29 11:28:42 Yeah it helps a lot when you're using tab completion to explore, etc 2015-05-29 11:28:50 Where? 2015-05-29 11:28:56 And... Why? 2015-05-29 11:29:07 http://lists.busybox.net/pipermail/busybox/2015-May/082999.html 2015-05-29 11:29:22 i looked at send a patch to bb ml with the / fix 2015-05-29 11:29:53 I doubt many will want the / fix 2015-05-29 11:30:00 thats what i figured 2015-05-29 11:30:06 so i gave that up 2015-05-29 11:30:08 but 2015-05-29 11:30:23 I'll do it and save your reputation 2015-05-29 11:30:28 if send patch to busybbox increases size, it will meet resistance 2015-05-29 11:30:39 if you add new feature and together with it you reduce size 2015-05-29 11:30:47 then you have bigger chance to get it accepted :) 2015-05-29 11:31:19 after finding the size reduction i realized that the / fix wouldnt make it anyway 2015-05-29 11:31:32 I'll say, hey look guys, I found something to fill those 35 bytes ncopa saved 2015-05-29 11:31:48 i can send a patch 2015-05-29 11:31:52 [RFC] ... 2015-05-29 11:32:26 RFC: lineedit: use let ^W delete to / 2015-05-29 11:32:44 the patch is trivial, but i dont think they will like it 2015-05-29 11:32:54 if bash did it, then it would be other thing 2015-05-29 11:34:14 ksh does it! 2015-05-29 11:34:44 how does it deal with : ; . etc? 2015-05-29 11:34:44 I bet you at least a dozen openbsd users on the list have been secretly hoping for it 2015-05-29 11:35:12 I forget which characters, I would make a BB_isword or so 2015-05-29 11:35:47 !BB_isalnum ? 2015-05-29 11:35:59 the vi mode does some funky stuff 2015-05-29 11:36:54 !isalnum would also delete spaces, so not sufficient for ^W 2015-05-29 11:37:26 i'd like read the ksh source 2015-05-29 11:37:37 yeah I'm digging it up now 2015-05-29 11:37:43 I think mksh is the best one right now... 2015-05-29 11:37:46 http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/bin/ksh/ 2015-05-29 11:38:04 ugh cvs 2015-05-29 11:38:32 ksh has vi mode too 2015-05-29 11:38:37 http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/bin/ksh/edit.c?rev=1.40&content-type=text/x-cvsweb-markup 2015-05-29 11:39:40 #define IS_WORDC(c) !( ctype(c, C_LEX1) || (c) == '\'' || (c) == '"' || \ 2015-05-29 11:39:40 (c) == '`' || (c) == '=' || (c) == ':' ) 2015-05-29 11:40:17 now where is C_LEX1 2015-05-29 11:41:25 sh.h 312: #define C_LEX1 BIT(2) /* \0 \t\n|&;<>() */ 2015-05-29 11:42:16 ascii hackers 2015-05-29 11:42:25 indeed :) 2015-05-29 11:43:41 i wonder if this is smaller: strchr("\0 \t\n|&;<>()\\\"`=:", c) == NULL 2015-05-29 11:44:09 probably, call to strchr is unlikely to be even a longjmp 2015-05-29 11:44:55 is there a tool to compare the new size? 2015-05-29 11:45:04 I'm going to test that patch right now 2015-05-29 11:45:21 hmm, actually, let me make a 3.2 usb disk just in case... 2015-05-29 11:45:28 still haven't tested eudev 2015-05-29 11:49:24 ahhhh it seems that the arm image is missing ppp-pppoe from the image 2015-05-29 11:49:50 thats a bug 2015-05-29 11:49:51 the rpi one? 2015-05-29 11:49:52 if I download the package from the mirror can I sideload it into cache somehow 2015-05-29 11:49:56 yes 2015-05-29 11:50:16 I'm stuck on my phone unable to dial my PPP connection lol 2015-05-29 11:50:16 you can just shove it in /apks/armhf/ 2015-05-29 11:50:18 i think its tricky to download anythign if you dont have network in the first place 2015-05-29 11:50:24 not really 2015-05-29 11:50:32 we sign the index 2015-05-29 11:50:38 --allow-untrusted ;) 2015-05-29 11:50:40 I don't think it has any dependencies 2015-05-29 11:50:44 yeah 2015-05-29 11:51:07 let the NSA in ;) 2015-05-29 11:51:23 you're already in fiveeyes 2015-05-29 11:51:33 that's true 2015-05-29 11:51:50 but they want me to fight terrorists 2015-05-29 11:51:58 they need my router for that 2015-05-29 11:54:41 md5sum ppp-pppoe-2.4.7-r3.apk 2015-05-29 11:54:43 a7527becd264077e591a2d929b09ab40 ppp-pppoe-2.4.7-r3.apk 2015-05-29 11:55:04 http://nl.alpinelinux.org/alpine/v3.2/main/armhf/ppp-pppoe-2.4.7-r3.apk 2015-05-29 11:55:47 so what's the command to install that package untrusted? 2015-05-29 11:56:26 apk add --allow-untrusted ppp-ppp.apk? 2015-05-29 11:56:53 in the current dir as the apk 2015-05-29 12:03:10 I saw this example apk add cherokee --update-cache --repository http://dl-3.alpinelinux.org/alpine/edge/testing/ --allow-untrusted 2015-05-29 12:03:35 how do I not specify a repository 2015-05-29 12:04:05 try just a path to the file 2015-05-29 12:04:08 ./ 2015-05-29 12:04:12 not for repository 2015-05-29 12:04:21 oh ok 2015-05-29 12:04:37 as in ./something.apk 2015-05-29 12:04:41 I'm guessing here, I haven't had to do this 2015-05-29 12:05:26 I think I need to use force 2015-05-29 12:05:35 I tried 2015-05-29 12:06:50 ncopa is there a way to avoid the APK index problems? 2015-05-29 12:07:05 in an emergency, you can "tar xzf something.apk" in / 2015-05-29 12:07:07 and fix it later 2015-05-29 12:07:21 yes 2015-05-29 12:07:29 might be an idea get network up first 2015-05-29 12:07:29 yeah that might be the only way 2015-05-29 12:07:42 I wondered if that would work 2015-05-29 12:07:44 that's what I would do, then you can update the repos, force overwrite 2015-05-29 12:07:46 then do apk add from network + cache 2015-05-29 12:07:59 then remove /.WHATEVER, I think there are two files 2015-05-29 12:08:27 .PKGINFO and .SIGN.RSA..rsa.pub 2015-05-29 12:08:30 ahills: seems like mksh's ^W does something similar to isalpha_or_underscore 2015-05-29 12:08:42 ncopa: oh, is it not the same as OpenBSD's? 2015-05-29 12:09:40 i mean ^W eats up to space, /,&%¤#"!`+?:. 2015-05-29 12:09:53 but it does eat up _ 2015-05-29 12:09:56 mm I think ppp-pppoe installs something else 2015-05-29 12:10:12 is so I get /use/bin/pon 2015-05-29 12:10:27 alnum includes á, ü, etc? 2015-05-29 12:10:38 dunno 2015-05-29 12:10:54 I would just be worried about not being explicit, but I suppose I haven't had a problem 2015-05-29 12:11:01 ppp-daemon it is 2015-05-29 12:11:21 silly_sausage1: we need add the ppp-* stuff to the image 2015-05-29 12:11:31 silly_sausage1: there are like eight subpackages 2015-05-29 12:11:32 could you please open a bug for it and set target to 3.2.1 2015-05-29 12:12:24 I think I need to make a minimal build machine for alpine, to prevent missing so many dependencies 2015-05-29 12:12:30 yes 2015-05-29 12:12:38 lxc 2015-05-29 12:12:41 or I should stop making packages after midnight 2015-05-29 12:12:43 is what i use 2015-05-29 12:12:45 :) 2015-05-29 12:12:54 damn this insomnia 2015-05-29 12:13:01 yeah I've been meaning to set up a container 2015-05-29 12:13:13 but then I wanted to fix my xorg/intel problems first 2015-05-29 12:13:15 and got distracted 2015-05-29 12:13:19 in any case, i think it would be nice to fix busybox ash's ^W 2015-05-29 12:13:36 would it be worth keeping a patch, or only if it's upstream? 2015-05-29 12:13:39 ahills: what xorg/intel problems do you have? 2015-05-29 12:13:43 busybox already has several, this would add to that load 2015-05-29 12:13:55 with udev, quitting xorg makes my screen black, can't get it back 2015-05-29 12:13:56 i would prefer to get it into upstream 2015-05-29 12:13:59 no errors in Xorg.0.log 2015-05-29 12:14:16 ahills: could you please file a bug with that? 2015-05-29 12:14:18 I need hotplugging which is why I was going to just try eudev 2015-05-29 12:14:23 and with as much info as possible 2015-05-29 12:14:30 udev, what arch 2015-05-29 12:14:34 x86_64 2015-05-29 12:14:40 specifically I have the 1st gen x1 carbon from lenovo 2015-05-29 12:14:42 lspci | grep VGA output 2015-05-29 12:14:50 lsmod 2015-05-29 12:14:52 etc etc 2015-05-29 12:14:55 yep 2015-05-29 12:15:04 i hear others have issues too 2015-05-29 12:15:20 mdev it worked almost fine 2015-05-29 12:15:26 I also want this stupid intel driver to do kms 2015-05-29 12:15:28 and i might have attention from someone who knows the code, but he needs be able to reproduce 2015-05-29 12:15:38 but even though it's set in modprobe.conf... 2015-05-29 12:15:42 ah ok 2015-05-29 12:15:59 i have an issue with an intel 32 bit too 2015-05-29 12:16:12 the lxdm hangs for a while 2015-05-29 12:16:13 2 mins 2015-05-29 12:16:18 then suddenly it shows up 2015-05-29 12:16:37 but that does not happen on my more modern 64bit boxes 2015-05-29 12:16:41 with mdev it would work except I couldn't get keyboard/mouse via hotplugging, and kms would happen after xorg was launched, so when I quit it would look how I wanted 2015-05-29 12:16:55 oh, one thing 2015-05-29 12:17:00 do you load fbcon? 2015-05-29 12:17:07 you need it for console 2015-05-29 12:17:10 in xorg or in modprobe? 2015-05-29 12:17:17 modprobe 2015-05-29 12:17:25 echo fbcon >> /etc/modules 2015-05-29 12:17:27 looks like I don't, thanks 2015-05-29 12:17:38 that will make screen black after exit xorg 2015-05-29 12:17:47 well, it's already black after exit... 2015-05-29 12:17:58 fortunately I can press power button to have openrc shutdown cleanly 2015-05-29 12:18:03 and unmount my cryptroot 2015-05-29 12:18:34 i mean, if you dont have fbcon module loaded, screen will just be black after quitting xorg 2015-05-29 12:18:38 oh 2015-05-29 12:18:52 well maybe I don't have any problems... maybe instead of a bug I'll be adding to the wiki 2015-05-29 12:19:01 test that 2015-05-29 12:19:03 brb, testing... 2015-05-29 12:19:46 Oh my god, I could kiss you, how was my solution so simple 2015-05-29 12:20:40 we had bug about that earlier 2015-05-29 12:20:42 fixed :) 2015-05-29 12:20:46 It's still not kms on boot 2015-05-29 12:20:48 thanks. 2015-05-29 12:20:51 Only after xorg 2015-05-29 12:21:25 i *think* that if you add kms as a feature in /etc/mkinitfs/mkinitfs.conf and regenerate initramfs 2015-05-29 12:21:38 might be a good idea to include all the PPP packages with the armhf image it won't make it much bigger 2015-05-29 12:21:49 it will start your kms and fbcon early 2015-05-29 12:22:17 silly_sausage1: yes. ppp-*, except ppp-doc and ppp-dev maybe 2015-05-29 12:22:45 yeah 2015-05-29 12:23:08 ok, I need to put this on the wiki somewhere 2015-05-29 12:24:15 # apk upgrade 2015-05-29 12:24:17 1 errors; 1507 MiB in 479 packages 2015-05-29 12:24:29 while it sill upgraded everything 2015-05-29 12:24:52 might some packaging error (same file in two packages; etc.) 2015-05-29 12:24:57 coredumb: apk fix? 2015-05-29 12:25:01 "apk fix" will try to reinstall the broken packgr 2015-05-29 12:25:18 ok got it 2015-05-29 12:25:36 thx :) 2015-05-29 12:26:00 abuild was failing for no reason because of that ^^ 2015-05-29 12:26:06 yeah 2015-05-29 12:26:08 well it's a good reason 2015-05-29 12:26:23 you don't want to be installing dependencies when there might be a serious problem with apk database 2015-05-29 12:26:58 ahills: "for no reason" when it actually didn't show any error and was installing deps correctly i meant 2015-05-29 12:27:00 :) 2015-05-29 12:27:48 Right but it can only know the return code 2015-05-29 12:29:14 ncopa: I did add kms feature but kms only happens after xorg still 2015-05-29 12:29:46 Do you know the Intel driver's module name? 2015-05-29 12:33:01 915 maybe? 2015-05-29 12:34:13 ahills: feel free to join #alpine-devel 2015-05-29 12:34:53 Didn't realize it was a channel 2015-05-29 15:56:55 is there a way to use /etc/networking/interfaces to configure a bridge via brctl? 2015-05-29 16:04:51 qemu-system-x86[12505]: segfault at 7ffa0a10fff8 ip 00000000f5757fb0 sp 00007ffa0a10bf30 error 6 in qemu-system-x86_64[f565d000+403000] 2015-05-29 16:04:53 mmmh 2015-05-29 16:30:55 ahills: yes 2015-05-29 16:32:49 ACTION does a documentation dance for the busybox gods 2015-05-29 16:35:12 mmmh anyone else having qemu segfaulting on edge ? 2015-05-29 16:54:07 hey, this is a silly question... but where can i find the build script for a particular apk package? 2015-05-29 16:58:10 lol. nm. found it 2015-05-29 16:59:30 *sigh* 2015-05-29 16:59:32 # CONFIG_ALX is not set 2015-05-29 17:00:03 why does alpine's kernel have an important wired ethernet driver disabled? :( 2015-05-29 17:04:45 ncopa, can we get CONFIG_ALX added to the kernel? 2015-05-29 17:05:35 hi 2015-05-29 17:05:40 which of the kernels? 2015-05-29 17:05:48 config-grsec.armhf:# CONFIG_ALX is not set 2015-05-29 17:05:48 config-grsec.x86:CONFIG_ALX=m 2015-05-29 17:05:48 config-grsec.x86_64:CONFIG_ALX=m 2015-05-29 17:06:21 oh vanilla kernel has not 2015-05-29 17:06:23 I'll fix 2015-05-29 17:06:37 ah 2015-05-29 17:07:13 what is ALX? 2015-05-29 17:07:19 wifi chipset? 2015-05-29 17:07:26 03:00.0 Ethernet controller: Qualcomm Atheros AR8162 Fast Ethernet (rev 10) 2015-05-29 17:07:30 wired 2015-05-29 17:07:34 or maybe both 2015-05-29 17:08:00 do you want it for v3.2 too? 2015-05-29 17:08:04 how long will it take to get an updated kernel i can apk update to? trying to work on some stuff i need it for, but if there'll be some latency i can just go plug stuff directly into router 2015-05-29 17:08:09 i use edge 2015-05-29 17:08:56 working on this sh2 dev board and it would be a lot more convenient to have the board plugged directly to my laptop rather than having to go back and forth to the router every time i want to change something 2015-05-29 17:09:04 I normally testbuild here first and test boot in qemu 2015-05-29 17:09:14 but this is so trivial so i think I just push 2015-05-29 17:09:41 yeah if the driver is enabled and works on grsec i don't think there's any serious risk of breakage adding it to vanilla 2015-05-29 17:09:44 btw 2015-05-29 17:09:59 wouldn't it be a good idea to have some automated check of vanilla config against grsec config? 2015-05-29 17:10:07 they should only differ in grsec/pax options, not drivers 2015-05-29 17:10:15 yes 2015-05-29 17:10:22 it is a good idea :) 2015-05-29 17:10:30 problem is time to implement it 2015-05-29 17:10:37 :) 2015-05-29 17:11:07 diff -u | grep ^[+-] | grep -Ev 'PAX|GRSEC' 2015-05-29 17:11:10 or something like that ;-) 2015-05-29 17:11:16 yes 2015-05-29 17:11:32 problem is that some options are disabled when grsec is enabled 2015-05-29 17:12:00 probably just need a whitelist of those to add to the grep -v 2015-05-29 17:12:07 yeah 2015-05-29 17:12:11 someething like that 2015-05-29 17:13:09 your kernel is in the oven 2015-05-29 17:13:56 you were lucky i was here at this time 2015-05-29 17:14:33 :) 2015-05-29 17:14:59 i hadnt even noticed it was missing because i don't normally used wired eth 2015-05-29 17:15:25 actually i don't even know if it will help -- i don't have a crossover cable 2015-05-29 17:15:37 but i think modern ethernet devices auto-crossover 2015-05-29 17:16:11 last time i used a crossover cable was probably in 98 or 99 :-p 2015-05-29 17:16:37 same 2015-05-29 17:16:40 all my devices auto crossover 2015-05-29 17:20:25 so hopefully this will just work 2015-05-29 17:21:21 i've got an fpga board setup as a starting point for bringing musl to nommu targets :) 2015-05-29 17:22:11 dalias: kernel done 2015-05-29 17:28:06 :) 2015-05-29 17:28:27 ncopa, will updating linux-vanilla get the new one now? 2015-05-29 17:28:39 it should yes 2015-05-29 17:28:45 i use nl.alpinelinux.org 2015-05-29 17:28:48 yes 2015-05-29 17:28:55 you should get the new kernel 2015-05-29 17:28:59 3.18.14-r1 2015-05-29 17:29:06 btw it would also be nice to add /proc/config.gz support to the kernel 2015-05-29 17:29:13 i think we have it 2015-05-29 17:29:22 nope :-p 2015-05-29 17:29:22 modprobe config or similar 2015-05-29 17:29:25 at least not in vanilla 2015-05-29 17:29:42 modprobe configs 2015-05-29 17:30:05 how can i check deps of a .apk ? 2015-05-29 17:30:17 is it in .PKGINFO? 2015-05-29 17:30:21 apk info -R 2015-05-29 17:30:37 ah it's a module, ok 2015-05-29 17:30:51 tar -O - -zxf file.apk .PKGINFO | grep depend 2015-05-29 17:31:09 :) 2015-05-29 17:31:21 I love simple standards 2015-05-29 17:31:22 i love how apk's are just tgz 2015-05-29 17:31:28 cool thx 2015-05-29 17:31:42 apk once broke itself and it was so easy to just untar the base packages to fix it 2015-05-29 17:32:03 tar -O -zxf file.apk .PKGINFO | grep ^depend 2015-05-29 17:32:06 is the correct 2015-05-29 17:32:13 yes, that was by design :) 2015-05-29 17:32:17 (2/2) Upgrading linux-vanilla (3.18.11-r1 -> 3.18.14-r1) 2015-05-29 17:32:20 is that the right ver? 2015-05-29 17:32:24 yes 2015-05-29 17:32:28 cool 2015-05-29 17:32:34 also, the db format is plaintext 2015-05-29 17:33:00 plaintext is fine as long as you're not parsing and loading the whole thing into memory with perl ;-) 2015-05-29 17:33:00 you can manually edit the installed db worstbase 2015-05-29 17:33:03 ACTION looks at debian 2015-05-29 17:33:03 worstcase 2015-05-29 17:33:05 ok pulseaudio pkg almost ready for testing 2015-05-29 17:33:17 or you can read with with awk 2015-05-29 17:33:36 coredumb: i think we have pa package in unmaintained 2015-05-29 17:33:39 when will it be systemd-sound ;P 2015-05-29 17:33:39 unixists 2015-05-29 17:33:42 soundd* 2015-05-29 17:33:54 ncopa: rofl 2015-05-29 17:34:32 dalias: .apk was also designed so that you can read from stream (eg network) and write directly to target 2015-05-29 17:34:37 1 read. 1 write 2015-05-29 17:34:46 nice 2015-05-29 17:34:55 so no need to save/buffer 2015-05-29 17:34:58 and do the checksum calculations while wiating for io 2015-05-29 17:35:10 of course for that to be safe you need a reliable stream 2015-05-29 17:35:23 drat no packages over udp then 2015-05-29 17:35:28 :) 2015-05-29 17:35:28 makes most sense in boot process i guess 2015-05-29 17:35:33 with no local storage 2015-05-29 17:35:33 yes 2015-05-29 17:35:35 but 2015-05-29 17:35:42 ncopa: I'll merge the APKBUILDs 2015-05-29 17:35:45 it will write to .apk-new files 2015-05-29 17:36:04 once checksum/sig confirmed, rename all .apk-new 2015-05-29 17:36:10 ah cool 2015-05-29 17:36:38 I actually for feature less build 2015-05-29 17:37:17 would mean a lot more subpackages to support all that from unmaintained 2015-05-29 17:37:17 apk is probably the fastest package manager there is 2015-05-29 17:37:38 that has support for dependencies and crypto signatures 2015-05-29 17:37:50 <_ikke_> apk is amazingly fast 2015-05-29 17:38:24 deb/rpm/whatever: 1 download, 2 check, 3 unpack 2015-05-29 17:38:35 apk does it in one 2015-05-29 17:38:49 <_ikke_> Really love that 2015-05-29 17:38:51 so when deb/rpm/whatever is done with step 1, apk is done with all 2015-05-29 17:39:06 built-in rsa signing is also a very strong feature 2015-05-29 17:39:18 <_ikke_> most have that, don't they? 2015-05-29 17:39:26 most use gnupg i think 2015-05-29 17:39:34 <_ikke_> ah yeah, ok 2015-05-29 17:39:56 gnupg is more flexible though 2015-05-29 17:40:01 <3 apk 2015-05-29 17:40:35 have a look at /lib/db/apk/installed 2015-05-29 17:40:47 oh the checksums 2015-05-29 17:40:49 for all files 2015-05-29 17:40:57 its ok we are going to rewrite it in python and use rpm 2015-05-29 17:41:01 rebooting :) 2015-05-29 17:41:08 brb 2015-05-29 17:41:10 you can actually veryfy that your systems 2015-05-29 17:41:14 apk audit --system 2015-05-29 17:41:51 ncopa: is that normal that deps for subpackages aren't calculated automatically by abuild ? 2015-05-29 17:42:00 no 2015-05-29 17:42:08 ok i gotta go 2015-05-29 17:42:09 it scans subpkg too 2015-05-29 17:42:14 for sonames 2015-05-29 17:42:30 have a nice weekend everyone 2015-05-29 17:42:37 <_ikke_> ncopa: you too 2015-05-29 17:42:49 got eth0 :) 2015-05-29 17:43:05 ncopa: you too 2015-05-29 17:43:51 kaniini: yep my bad 2015-05-29 17:45:43 bleh no ping from the board :/ 2015-05-29 17:46:31 do you have link 2015-05-29 17:46:35 might need crossover 2015-05-29 17:46:48 the really cheap ethernet silicon doesn't do mdi-x 2015-05-29 17:46:49 hmm tcpdump showed a dhcp packet 2015-05-29 17:46:59 so i don't think it's honoring the static ip that was configured 2015-05-29 17:47:15 i'll add a udhcpd here 2015-05-29 17:53:17 hmm udhcpd isn't getting the request, but tcpdump shows it 2015-05-29 17:57:06 uhg no alpine is running its own dhcp client on the eth :/ 2015-05-29 17:57:16 why would it be doing that? eth0 isn't in /etc/network/interfaces 2015-05-29 17:59:51 dalias: ps aux | grep udhcpc 2015-05-29 17:59:57 maybe from a prior configuration? 2015-05-29 18:05:08 no 2015-05-29 18:07:32 i don't understand why dhcpcd is probing eth0 tho 2015-05-29 18:07:39 i only have one dhcpcd process 2015-05-29 18:07:43 and wifi should be dhcp 2015-05-29 18:13:39 if you don't pass the interface it tries to do every interface iirc 2015-05-29 18:13:46 I haven't used dhcpcd in a long time though 2015-05-29 18:14:07 uhg, i see 2015-05-29 18:14:13 fixed that with denyinterfaces 2015-05-29 18:15:06 now nothing on the ethernet 2015-05-29 18:15:14 so it might be that auto-crossover doesn't work 2015-05-29 18:15:21 or the board is not booting :( 2015-05-29 18:15:44 i could dig out an old wired switch and plug that in between 2015-05-29 18:19:01 auto-crossover only needs to work on one device 2015-05-29 18:19:12 or at least only on one device when one of them was my t60 2015-05-29 18:19:53 *nod* 2015-05-29 18:22:04 which fpga are you using? 2015-05-29 18:28:56 weird, util-linux's column isn't actually formatting my text into columns... 2015-05-29 19:06:52 it's a xilinx based board ("lx9 microboard") but all the fpga stuff was pre-setup for me 2015-05-29 19:06:56 and it's supposed to boot from the provided micro sd card 2015-05-29 19:13:45 do you have any interface to it except ethernet? 2015-05-29 20:51:57 dalias: I'm looking to get alpine running on a i586 (Pentium 1 like CPU) 2015-05-29 20:52:09 just for curiosity ;3 2015-05-29 21:18:47 should work fine 2015-05-29 21:19:03 hope it's not an x1000 tho 2015-05-29 21:37:34 it has really bad bugs that are nearly impossible to workaround 2015-05-29 21:38:04 i've been advising someone from intel on the glibc dev list about ideas for a kernel workaround 2015-05-29 21:38:12 but it's just an atrociously bad chip :( 2015-05-29 21:48:05 *sigh* my eth0 works fine for crossover with another laptop 2015-05-29 21:48:14 so i think the test board just isn't booting 2015-05-29 22:13:03 Diftraku: i had a hang on boot on some old i586 laptop, didn't yet retry; i felt it was so slow that the root mount expired, but not really sure 2015-05-29 22:16:27 dalias: elimination of unnecessary (and justification of all) SUID binaries is a release goal for 3.3 2015-05-29 23:05:02 hi, i have a little problem installing alpine, running setup-alpine after choosing the disk and the option sys to install the system, when the script is setuping the partitions it stops saying that "The file does not exist and no size was specified" 2015-05-29 23:05:28 is there a way to find out whats the issue? i can see that the partitions where created 2015-05-29 23:06:39 i0m installing from a usb stick with the 3.2.0 vanilla image 2015-05-29 23:25:40 kaniini, :) :) :) 2015-05-29 23:26:15 btw one of those is polkit 2015-05-29 23:27:00 what would you think of a project to, one by one with initially compatible outward apis, replace all of the fdo components like that? :) 2015-05-29 23:31:44 ok, i've mountted the partition manually and used setup-disk mounoint and is working but there is something wrong with the setup script. i'll try to find out what it is. 2015-05-29 23:33:12 m0ikz, sorry, i'm not ignoring you, i'm just not the best qualified to give installation help 2015-05-29 23:33:35 no problem :) 2015-05-29 23:38:21 dalias: do they need replacements? 2015-05-29 23:41:53 I suppose some of fdo's software is useful 2015-05-29 23:53:53 the problem i was having is that the disk had a gpt partition table :) 2015-05-29 23:54:32 dalias: well 2015-05-29 23:54:35 dalias: considering i made pkgconf 2015-05-29 23:54:46 dalias: which is a replacement for an fdo component 2015-05-29 23:54:52 dalias: i would be generally supportive 2015-05-29 23:55:21 luzie: the problem is, a lot of the old fd.o components, are being replaced by systemd-food 2015-05-29 23:55:50 polkit, for example is likely being replaced by systemd-logind at some point in the future 2015-05-29 23:57:08 kaniini, :) 2015-05-29 23:57:35 but presumably it will still provide the same apis that crap is depending on 2015-05-29 23:57:55 replacing those (dbus-based) apis with something that lets you remove polkit/systemd-logind is the important thing 2015-05-30 00:46:30 Does anyone have any idea why the murmur package does not contain the murmur binary? http://pkgs.alpinelinux.org/contents?pkgname=murmur&arch=x86 2015-05-30 01:29:20 bontibon: that is indeed strange, have you looked at the APKBUILD to see what's up? 2015-05-30 02:12:09 actually, the mumble package is also broken, it seems... 2015-05-30 02:48:42 *sigh* this board isn't booting :/ 2015-05-30 02:48:58 hooked up serial and got errors at the bootloader stage 2015-05-30 08:21:29 hi, i have a problem with cups and ghostscript 9.16... 2015-05-30 08:21:52 missing devices 2015-05-30 08:22:11 Available devices: 2015-05-30 08:22:11 bbox bit bitcmyk bitrgb bitrgbtags bmp16 bmp16m bmp256 bmp32b bmpgray 2015-05-30 08:22:11 bmpmono bmpsep1 bmpsep8 ccr cfax cif cups cups devicen dfaxhigh dfaxlow 2015-05-30 08:22:11 eps2write faxg3 faxg32d faxg4 fpng inferno ink_cov inkcov jpeg jpegcmyk 2015-05-30 08:22:11 jpeggray mag16 mag256 mgr4 mgr8 mgrgray2 mgrgray4 mgrgray8 mgrmono miff24 2015-05-30 08:22:11 nullpage pam pamcmyk32 pamcmyk4 pbm pbmraw pcx16 pcx24b pcx256 pcx256 2015-05-30 08:22:13 pcx2up pcxcmyk pcxgray pcxmono pdfwrite pdfwrite pdfwrite pgm pgmraw pgnm 2015-05-30 08:22:15 pgnmraw pkm pkmraw pksm pksmraw plan plan9bm planc plang plank planm plib 2015-05-30 08:22:17 plibc plibg plibk plibm png16 png16m png256 png48 pngalpha pnggray 2015-05-30 08:22:19 pngmono pnm pnmraw ppm ppmraw ps2write psdcmyk psdcmykog psdrgb pwgraster 2015-05-30 08:22:21 sgirgb spotcmyk sunhmono tiff12nc tiff24nc tiff32nc tiff48nc tiff64nc 2015-05-30 08:22:25 tiffcrle tiffg3 tiffg32d tiffg4 tiffgray tifflzw tiffpack tiffscaled 2015-05-30 08:22:27 tiffsep tiffsep1 txtwrite xcf xpswrite 2015-05-30 08:22:37 but on my other system i have: Available devices: [24/1841] 2015-05-30 08:22:37 alc1900 alc2000 alc4000 alc4100 alc8500 alc8600 alc9100 ap3250 appledmp 2015-05-30 08:22:38 atx23 atx24 atx38 bbox bit bitcmyk bitrgb bitrgbtags bj10e bj10v bj10vh 2015-05-30 08:22:38 bj200 bjc600 bjc800 bjc880j bjccmyk bjccolor bjcgray bjcmono bmp16 bmp16m 2015-05-30 08:22:40 bmp256 bmp32b bmpgray bmpmono bmpsep1 bmpsep8 ccr cdeskjet cdj1600 cdj500 2015-05-30 08:22:42 cdj550 cdj670 cdj850 cdj880 cdj890 cdj970 cdjcolor cdjmono cdnj500 cfax 2015-05-30 08:22:44 chp2200 cif cljet5 cljet5c cljet5pr coslw2p coslwxl cp50 declj250 deskjet 2015-05-30 08:22:46 devicen dfaxhigh dfaxlow dj505j djet500 djet500c dl2100 dnj650c epl2050 2015-05-30 08:22:48 epl2050p epl2120 epl2500 epl2750 epl5800 epl5900 epl6100 epl6200 eplcolor 2015-05-30 08:22:50 eplmono eps9high eps9mid epson epsonc epswrite escp escpage faxg3 faxg32d 2015-05-30 08:22:54 faxg4 fmlbp fmpr fs600 gdi hl1240 hl1250 hl7x0 hpdj1120c hpdj310 hpdj320 2015-05-30 08:22:56 hpdj340 hpdj400 hpdj500 hpdj500c hpdj510 hpdj520 hpdj540 hpdj550c 2015-05-30 08:22:58 hpdj560c hpdj600 hpdj660c hpdj670c hpdj680c hpdj690c hpdj850c hpdj855c 2015-05-30 08:23:00 hpdj870c hpdj890c hpdjplus hpdjportable ibmpro ijs imagen inferno inkcov 2015-05-30 08:23:02 iwhi iwlo iwlq jetp3852 jj100 jpeg jpegcmyk jpeggray la50 la70 la75 2015-05-30 08:23:04 la75plus laserjet lbp310 lbp320 lbp8 lex2050 lex3200 lex5700 lex7000 2015-05-30 08:23:06 lips2p lips3 lips4 lips4v lj250 lj3100sw lj4dith lj4dithp lj5gray lj5mono 2015-05-30 08:23:08 ljet2p ljet3 ljet3d ljet4 ljet4d ljet4pjl ljetplus ln03 lp1800 lp1900 2015-05-30 08:23:10 lp2000 lp2200 lp2400 lp2500 lp2563 lp3000c lp7500 lp7700 lp7900 lp8000 2015-05-30 08:23:12 lp8000c lp8100 lp8200c lp8300c lp8300f lp8400f lp8500c lp8600 lp8600f 2015-05-30 08:23:14 lp8700 lp8800c lp8900 lp9000b lp9000c lp9100 lp9200b lp9200c lp9300 2015-05-30 08:23:16 lp9400 lp9500c lp9600 lp9600s lp9800c lps4500 lps6500 lq850 lxm3200 2015-05-30 08:23:18 lxm5700m m8510 mag16 mag256 md1xMono md2k md50Eco md50Mono md5k mgr4 mgr8 2015-05-30 08:23:20 mgrgray2 mgrgray4 mgrgray8 mgrmono miff24 mj500c mj6000c mj700v2c mj8000c 2015-05-30 08:23:24 ml600 necp6 npdl nullpage oce9050 oki182 oki4w okiibm oprp opvp paintjet 2015-05-30 08:23:26 pam pamcmyk32 pamcmyk4 pbm pbmraw pcl3 pcx16 pcx24b pcx256 pcx256 pcx2up 2015-05-30 08:23:30 and some more... but i need for example ljet4 2015-05-30 08:24:01 and i can not find any solution to make it available... 2015-05-30 08:24:10 and i can not find any solution how to make it available... 2015-05-30 08:24:52 anyone know where may be a problem? 2015-05-30 08:35:06 those output was from: gs -h 2015-05-30 09:22:41 please use a pastebin for pasting multiline stuff, sprunge.us seems to be favoured here 2015-05-30 09:23:12 ok, sorry 2015-05-30 09:31:54 bolo_: ljet4 is in that list... 2015-05-30 09:32:04 I can see it on my screen right now 2015-05-30 09:34:55 http://sprunge.us/ZQRc ;] this is the full output from the alpine system 2015-05-30 09:36:10 you probably see ljet4 in the output from my second system (gentoo) 2015-05-30 09:36:29 this is why you shouldn't paste to channel ;) 2015-05-30 09:36:59 right;] 2015-05-30 09:37:42 ok, well, if you look at the APKBUILD it's built with configure --enable-drivers=FILES,cups 2015-05-30 09:37:46 how is it built on gentoo? 2015-05-30 09:40:47 --with-drivers=ALL :( 2015-05-30 09:41:10 ok, well, the easiest thing to do is clone aports, cd aports/main/ghostscript, edit APKBUILD, abuild -r 2015-05-30 09:41:38 I'm trying it now, to see if there's a reason it won't build with all 2015-05-30 09:42:00 thanks 2015-05-30 09:43:00 I guess it's not as simple as I said if you haven't run abuild yet 2015-05-30 09:43:09 I think it generates a key for you the first time... probably there is a guide on the wiki 2015-05-30 09:44:08 i am new in alpine, i do not know how to use abuild .. yet;] 2015-05-30 09:45:41 it's pretty simple 2015-05-30 09:49:01 still building :) this laptop is not very speedy 2015-05-30 09:49:19 if it works, I will submit a patch, but I don't know how long it will take to make it upstream 2015-05-30 09:49:35 are you on x86_64? 2015-05-30 09:51:59 no.. x86 2015-05-30 09:57:15 ok, I won't be able to build a package for you, but I can sprunge the modified APKBUILD, and you can follow http://wiki.alpinelinux.org/wiki/Abuild_and_Helpers 2015-05-30 09:58:10 unless your machine is much faster than mine, though, it will take a long time... hopefully you weren't planning on using gs with ljet4 in the next ten minutes 2015-05-30 09:58:36 ok, thanks a lot, i probably can do that myself;] 2015-05-30 09:59:09 http://sprunge.us/dWFB 2015-05-30 09:59:59 after you use that APKBUILD, and build the package, use "apk add ghostscript=9.16-r1" 2015-05-30 10:00:23 I think even if it gets that version in remote repos, it won't affet you 2015-05-30 10:00:53 http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management#Holding_a_specific_package_back 2015-05-30 10:00:56 i should change that: --with-drivers=ALL right? 2015-05-30 10:01:05 oh, did I not send the right one? 2015-05-30 10:01:08 haha oops 2015-05-30 10:01:11 yes, just do that 2015-05-30 10:01:18 that and the pkgrel are the only things I changed 2015-05-30 10:01:38 ok, my build succeeded 2015-05-30 10:01:38 mhm 2015-05-30 10:01:46 great;] 2015-05-30 10:02:02 without errors? 2015-05-30 10:02:15 what is the output from gs -h? 2015-05-30 10:03:21 without errors 2015-05-30 10:03:44 ... is it any chance that this version appears in the official repo?;] 2015-05-30 10:03:54 I did a manual configure/make, I'm using the updated apkbuild now and will look at the actual package 2015-05-30 10:04:08 there is always a chance, I will submit the patch and then maybe it'll make it into 3.2.1 or something 2015-05-30 10:04:19 if it gets accepted it will be in edge almost right away 2015-05-30 10:04:51 thanks for reminding me that I need to get my other build machine set up so I can use my laptop while I build :P 2015-05-30 10:08:00 BitL0G1c: which program do you use for spice server? or do you just use it with kvm? 2015-05-30 10:14:04 spice server? 2015-05-30 10:14:30 afk.. 1h 2015-05-30 10:17:01 bolo_: http://sprunge.us/DSHI 2015-05-30 10:17:10 good luck with the build 2015-05-30 10:18:53 bolo_: please report whether you have success on x86 2015-05-30 12:02:07 ahills - on alpine it is spice-server & on debian libspice-server1 - I've only tried it with kvm but it should work with xen thru libvirt too 2015-05-30 12:02:40 what if it's just a bare host? 2015-05-30 12:02:54 I'll do it via xen if I have to but I'd rather not 2015-05-30 12:03:19 i dont think libvirt is a requirement - it just makes the configuration more automatic 2015-05-30 12:03:46 I mean, is it possible to use it as a direct replacement for vncserver 2015-05-30 12:03:58 where I just run a server daemon that listens, which gives me a persistent desktop 2015-05-30 12:04:09 yes 2015-05-30 12:04:20 no xen, no kvm, no qemu 2015-05-30 12:04:23 with or without libvirt - all libvirt does is expose a port 2015-05-30 12:04:49 ok 2015-05-30 12:05:24 https://packages.debian.org/jessie/libspice-server1 - no libvirt depends 2015-05-30 12:06:26 ok, as expected, this does not give me an executable... only libspice-server.so.1 2015-05-30 12:07:10 I should quit being so lazy and just build spice-server 2015-05-30 12:26:04 has anyone got hibernation working with pm-utils? 2015-05-30 12:26:21 I get "pm-utils does not know how to hibernate on this system." 2015-05-30 12:34:44 with or without grsec? 2015-05-30 12:36:32 do you have /sys/power/disk 2015-05-30 12:50:42 stf: with; no I don't 2015-05-30 12:50:49 is this as "simple" as a kernel issue? 2015-05-30 12:52:55 gs63vW9tNVr9: do you highlight on "stf"? 2015-05-30 12:55:47 no i don't i should, good point! 2015-05-30 12:56:35 ok, then it's grsec disabling hibernation because of security concerns, because people seem to hibernate on non-fde partitions.... 2015-05-30 12:57:34 HIBERNATION is depending on !GRKERNSEC_KMEM and !PAX_MEMORY_SANITIZE drop those, and you'll have a /sys/power/disk and hibernation magic 2015-05-30 12:57:34 well, I'm fde... 2015-05-30 12:58:13 can't I just paxctl /boot/vmlinuz-grsec :P 2015-05-30 12:58:23 nope :/ 2015-05-30 12:58:27 just kidding 2015-05-30 12:58:42 :) 2015-05-30 12:58:50 I guess it's time to maintain the linux-grsec-hibernation aport 2015-05-30 13:03:54 ahills: would you try again hightlighting me with stf? 2015-05-30 13:04:18 stf: PAX_MEMORY_SANITIZE is already unset; looking for docs on GRKERNSEC_KMEM to see what it actually is before disabling... 2015-05-30 13:05:09 \o/ 2015-05-30 13:06:22 huh, http://en.wikibooks.org/wiki/Grsecurity/Appendix/Grsecurity_and_PaX_Configuration_Options doesn't explicitly mention hibernation... are /dev/kmem and/or /dev/mem required for it? 2015-05-30 13:06:30 stf: stf: stf: stf: stf: stf: stf: 2015-05-30 13:07:05 hmm, does it also highlight on stfu? ;) 2015-05-30 13:07:27 try again with stfu 2015-05-30 13:07:35 stfu 2015-05-30 13:07:43 ACTION has been banned 2015-05-30 13:07:53 no does not highlight on stfu 2015-05-30 13:09:21 all the documentation I've found thus far on the hibernation feature just assumes /sys/power/disk exists 2015-05-30 13:17:56 ahills: you need to revert this patch from the grsec-patch: https://www.ctrlc.hu/~stef/hibernate.patch 2015-05-30 13:18:23 oh, is it just a user-safety check? 2015-05-30 13:18:42 like a seatbelt light 2015-05-30 13:18:55 I can still drive the kernel car without it? 2015-05-30 13:19:20 yes, afaik this has been introduced to protect people from hibernating to non-fde partitions 2015-05-30 13:20:10 that is very annoying... 2015-05-30 13:20:24 if I've got fde, it should be easier than rebuilding my kernel to fix that... 2015-05-30 13:20:26 i think this has been introduced a year ago 2015-05-30 13:20:37 oh well, one small patch is no big deal, let's try it 2015-05-30 13:20:46 what's a good suffix for linux-grsec? 2015-05-30 13:21:02 be aware the patch i posted is the one you need to revert, not apply. 2015-05-30 13:22:14 haha yes, that much is obvious both from your message and from the content of the patch 2015-05-30 13:25:10 ok, just trying to avoid you having to restart your experiment due to ambiguous wording of mine 2015-05-30 13:26:07 abuild would catch it if it wouldn't apply cleanly 2015-05-30 13:27:15 sure, but it would still be wasted time 2015-05-30 13:27:45 ah, the good old days of hand-editing diffs 2015-05-30 13:27:59 memories of my first job out of college are coming back to me.. 2015-05-30 13:28:07 :) 2015-05-30 13:29:18 if that's a criteria, then i'm still in college 2015-05-30 13:29:32 meh i mean 1st job after college 2015-05-30 13:29:36 haha 2015-05-30 13:29:46 no, it's just something I happened to do there 2015-05-30 13:30:02 i still do it when i patch things 2015-05-30 13:30:06 it was a defense contractor, and most of the guys there were too old to use even old unix tools 2015-05-30 13:31:13 it still failed, because copying and pasting from a website gives me spaces instead of tabs 2015-05-30 13:34:59 oh. that's my copypasting. not yours 2015-05-30 13:35:10 you can take this section out of the grsec patch 2015-05-30 13:35:15 with tabs 2015-05-30 13:35:22 sorry, for still wasting your time 2015-05-30 13:36:10 if you curl it you still get the spaces, so i'm to blame 2015-05-30 13:39:20 lol 2015-05-30 13:39:43 it is a beautiful saturday, I'm on my balcony with my laptop 2015-05-30 13:39:51 I could not care less that I had to take twenty seconds to :s in my editor 2015-05-30 13:40:29 can't sit on the balcony now, the sun is out there trying to kill me. 2015-05-30 13:40:50 I'm in the shade until about 3pm 2015-05-30 13:42:08 the sun tricked me in to this appartment that is strategically pointed south, so she can attack me all day... 2015-05-30 13:44:36 hahaha 2015-05-30 13:44:43 so devious 2015-05-30 13:44:58 does anyone have an x86 machine that wants to try out grsec + hibernation? 2015-05-30 13:47:20 sorry can't help you with that. last time i updated my alpine on x86 i forgott to disengage the cpu fan, and right in the middle of the kernel update the cpu overheated, and i still didn't bother to recover the system. 2015-05-30 13:47:33 oh no 2015-05-30 13:47:53 disengage? or (undis)engage? 2015-05-30 13:48:06 make it spin uncontrolledly fast 2015-05-30 13:48:28 is this a pentium 4 or something? 2015-05-30 13:48:31 echo disengaged >/proc/acpi/ibm/fan 2015-05-30 13:48:49 I won't be doing that, but thanks for the tip 2015-05-30 13:48:51 no a laptop 2015-05-30 13:48:57 core2 2015-05-30 13:49:04 oh, I guess I do have one of those 2015-05-30 13:49:11 I could put alpine on it, it's just sitting on a shelf 2015-05-30 13:49:27 good old t60 works just as well as it did in 2006... 2015-05-30 13:49:46 you can try it out, and then revert back using `echo auto >/proc/acpi/ibm/fan` 2015-05-30 13:50:25 "invalid argument" on my newer thinkpad 2015-05-30 13:50:28 this laptop has a legendary bad thermal design 2015-05-30 13:50:57 even back when it was "modern" there was special dockingstations with fans just to make it bearable 2015-05-30 13:51:18 hm? 2015-05-30 13:51:25 try cat /proc/acpi/ibm/fan 2015-05-30 13:51:30 should give you all the options 2015-05-30 13:51:45 no, only the current status, speed, and level 2015-05-30 13:51:48 level is "auto" 2015-05-30 13:52:00 status is fortunately "enabled" 2015-05-30 13:52:18 disengaged is speed, not stauts 2015-05-30 13:52:26 here at least 2015-05-30 13:53:01 I think I'll leave this one alone... if my t60 dies, it will be sad, but if this one dies, it's an expensive replacement 2015-05-30 18:36:58 ahills, i want to inform you, that ghostscript works fine now;] thank you for your help;] 2015-05-30 18:37:13 and that abuild is a nice tool;] 2015-05-30 21:53:27 If anyone wanted to get Alpine running on Digital Ocean, I just wrote up a guide on how to do so: https://github.com/bontibon/digitalocean-debian-to-alpine 2015-05-30 22:35:58 \o/ 2015-05-30 23:39:38 Has anybody used Alpine on their laptop? I was considering installing it on my macbook, I'm running Gentoo right now. 2015-05-30 23:46:14 phoeniciavore: me, i love it 2015-05-30 23:46:50 ahills: how have you gotten around the whole issue of certain packages not building under musl? 2015-05-30 23:47:05 I think the idea behind musl is great, but I'm a little scared to try it 2015-05-30 23:49:24 Initially I tried building a Gentoo installation using muslibc but I had to give up. It was too finnicky 2015-05-30 23:50:52 Haven't run into it phoeniciavore, except for guile, which I do without 2015-05-30 23:52:02 Looks like I ought to give it a shot then :) 2015-05-30 23:52:12 what WM are you using? 2015-05-31 03:09:15 i am getting this error: gyp WARN EACCES user "undefined" does not have permission to access the dev dir "/root/.node-gyp/2.1.0" 2015-05-31 03:09:51 on 'docker-compose build'. this is the line in my Dockerfile: RUN npm install anvil-connect -g 2015-05-31 03:10:54 it is something to do with installing an NPM package called bcrypt 2015-05-31 03:11:04 phoeniciavore: dwm 2015-05-31 03:11:18 ../node_modules/nan/nan.h:401:81: note: in definition of macro 'NanReturnValue' 2015-05-31 03:12:43 owen1: if you're having trouble building, identify the missing headers and install corresponding -dev packages 2015-05-31 03:13:33 ahills: thanks. despite of those warning the docker build was complete. i'll see if i have any issues. 2015-05-31 03:14:10 In general with npm, good luck... 2015-05-31 03:22:13 ahills: (: bcrypt was not installed. how do i find the missing headers? 2015-05-31 03:23:21 apk search bcrypt and see if there's a bcrypt-dev or libbcrypt-dev or such 2015-05-31 03:24:17 ok. btw, here is some of the errors - http://pastebin.com/gk6beg8d 2015-05-31 03:26:08 cryptsetup-libs-1.6.6-r3, libcrypto1.0-1.0.1m-r1, freerdp-libs-1.1.0_beta2013071101-r0, libcrypto1.0-1.0.1m-r1 2015-05-31 03:26:10 no dev 2015-05-31 03:27:22 I have no idea what that output means 2015-05-31 03:27:59 npm is shitty, it makes finding errors extremely difficult 2015-05-31 03:28:09 Also fixing them... 2015-05-31 03:28:14 (: 2015-05-31 03:28:23 sorry if i confused you 2015-05-31 03:29:07 should i ask in #node ? 2015-05-31 03:31:15 Probably, find out why npm doesn't report any errors... 2015-05-31 03:31:52 hahah. ok 2015-05-31 03:32:23 You sure it's not installed? 2015-05-31 03:33:03 ahills: actually i see the folder bcrypt so maybe it's ok 2015-05-31 03:34:07 let me se 2015-05-31 03:34:09 see 2015-05-31 04:05:16 https://www.flockport.com/a-quick-introduction-to-alpine-linux/ 2015-05-31 05:23:50 pnutzh4x0r: great into! thanks 2015-05-31 06:09:44 pnutzh4x0r: http://i.imgur.com/LXM1hEa.png why does it not manage its tools using apk? 2015-05-31 10:51:33 hi 2015-05-31 10:53:22 hi 2015-05-31 11:54:28 http://turtle.dereferenced.org/~kaniini/interfaces.conf.txt 2015-05-31 11:54:30 opinions ? 2015-05-31 12:10:22 kaniini, It seems Vyatta , am I right? 2015-05-31 12:11:10 vitronic: kind of, but no. it's meant to replace our current /etc/network/interfaces with something that has an actually stable grammar 2015-05-31 12:11:54 vitronic: i did choose to go with a tree-like structure for the configuration elements in order to be familiar to system administrators 2015-05-31 12:12:36 like json 2015-05-31 12:12:55 i use alpine on a raspberry pi. I try to upgrade on latest alpine (3.1.x -> 3.2) but all system is on ram so upgrade failed since note enough memory (314Mo). I try to create swap.img on sdcard but i can't too. Have someone an idea ? 2015-05-31 12:13:38 Chipster1: i believe you are supposed to flash to the new image 2015-05-31 12:13:49 however, i do not have any rpi devices 2015-05-31 12:13:56 so i could be wrong. 2015-05-31 12:13:59 kaniini, It looks good, although I have no problems with the current /etc/network/interfaces 2015-05-31 12:14:09 kaniini it's the ultimate solution :D 2015-05-31 12:14:44 I would rather do otherwise 2015-05-31 12:14:52 vitronic: there are technical problems with it which are hidden by the fact that the current ifupdown infrastructure is basically a giant hack 2015-05-31 12:16:35 kaniini, So like I said it looks pretty good, like json 2015-05-31 12:24:54 Chipster1, I have a similar problem in my p2, detects only 700Mb of RAM, I configured the gpu_mem = 32 in config.txt in the / boot unsuccessfully with raspbian no problem 2015-05-31 12:28:11 has anyone considered looking at http://bugs.alpinelinux.org/issues/1831 2015-05-31 12:28:27 it would help stimulate user interaction i think 2015-05-31 12:28:58 probably a good idea 2015-05-31 12:29:11 oh wait my bad 2015-05-31 12:29:13 http://alpinelinux.org/atom.xml 2015-05-31 12:29:17 how did i not see that before 2015-05-31 12:29:27 well, that is just releases on the main-site. 2015-05-31 12:29:39 i think #1831 is more about creating an aggregator of blogs that discuss alpine 2015-05-31 12:29:47 like planet.debian.org 2015-05-31 12:30:13 ah true maybe 2015-05-31 13:02:40 Hello, ijs it possible to use proprietary nvidia drivers on alpine? 2015-05-31 13:03:30 And also does alpine follows a rolling release model? 2015-05-31 13:04:01 it does not 2015-05-31 13:08:10 kaniini: owen1 i didnt write that article, i just found it 2015-05-31 13:12:34 ok thanks 2015-05-31 13:18:56 ok vitronic 2015-05-31 13:19:16 i have the same problem with arch and alpine :/ 2015-05-31 13:57:25 Chipster1: with rpi you just extract .tar.gz over the old one, your apkovl will me untouched 2015-05-31 14:29:52 ahills ok, i try this 2015-05-31 14:29:55 thx 2015-05-31 14:31:37 ahill and if i want to create a swap.img with 2Go size, how i must proceed since i have only 314Mo ? 2015-05-31 14:31:52 Chipster1: make it directly on disk 2015-05-31 14:32:03 Chipster1: you can also make it a sparse file, like this 2015-05-31 14:32:50 ok 2015-05-31 14:33:04 Chipster1: dd if=/dev/zero of=/media/mmc0blk0p1/swap.img bs=1M count=0 seek=1024 (for 1024MB) 2015-05-31 14:33:18 you mean to use swap.img as a swapfile? 2015-05-31 14:33:27 yes 2015-05-31 14:33:43 ok... so you understand why creating a swapfile in memory would be silly, right? 2015-05-31 14:33:52 i use this but i didn't use /media/mm… 2015-05-31 14:34:03 yeah sure, put it on any storage you have 2015-05-31 14:34:16 e.g. I have a big fat 2TB usb disk attached at /dev/sda 2015-05-31 14:34:24 ;) 2015-05-31 14:35:17 thx for you help ahills 2015-05-31 14:35:53 no problem 2015-05-31 14:36:03 oh, don't forget to update /etc/fstab and lbu commit too 2015-05-31 14:36:12 when I forget to lbu commit I get pretty mad at myself 2015-05-31 14:37:48 ahills i use a cron to sync with lbu 2015-05-31 14:38:18 i had some surprise :D 2015-05-31 14:39:04 I thought about making an /etc/local.d script to run on shutdown 2015-05-31 14:39:25 but I also like being able to reset my mistakes by rebooting 2015-05-31 14:39:35 I guess I could make backups and all that nonsense 2015-05-31 14:39:46 phew, sounds like work though, putting two lines in a script instead of one ;) 2015-05-31 15:28:23 ahills /media/mm… is mounted ro 2015-05-31 15:28:25 :/ 2015-05-31 15:29:37 Chipster1: mount -o remount,rw /media/whatever 2015-05-31 15:29:44 then when you're done, mount -o remount,ro 2015-05-31 15:29:54 this is what lbu does, and a few other scripts possibly 2015-05-31 15:29:56 already done ;) 2015-05-31 15:29:59 apk too 2015-05-31 15:30:10 ok 2015-05-31 15:30:21 if you have the cache enabled, that is 2015-05-31 15:30:39 every time I think of how difficult my task is going to be, it turns out an alpine tool has already taken care of it 2015-05-31 15:30:50 it's as if the developers have been reading my mind for the last decade 2015-05-31 15:32:10 ok 2015-05-31 15:32:20 Chipster1: sometimes when I'm working actively on the configuration I just leave it mounted rw 2015-05-31 15:32:29 ok 2015-05-31 16:25:19 anyone used zfs on alpinelinux? 2015-05-31 16:26:45 i was thinking of doing something like 2015-05-31 16:26:48 http://linux.arantius.com/installing-gentoo-into-a-luks-encrypted-zfs-root 2015-05-31 16:29:42 i did, but running lvm for luks under zfs kinda defeats some of the benefits of zfs, considering also the workarounds to hack code engineered into solaris to make it stick to the evolved codebase of linux kernels, i stoped my experiments 2015-05-31 16:30:07 also there's licensing issues 2015-05-31 16:30:38 well the only reason i'd be doing it would be to run my vms and use a nice zfs raidz2 pool 2015-05-31 16:30:47 with my vms underneath 2015-05-31 16:30:57 i don't really like md raid devices 2015-05-31 16:31:05 it felt unnatural for me the zfs + linux thingy 2015-05-31 16:31:25 and ZoL is pretty stable now 2015-05-31 16:31:25 if you want luks, you need lvm anyway, no? 2015-05-31 16:31:31 you can compile it into the kernel 2015-05-31 16:31:34 no you don't 2015-05-31 16:31:41 no? 2015-05-31 16:31:46 i've seen that way of doing it ^ which does not use LVM 2015-05-31 16:31:55 let me read that 2015-05-31 16:32:28 which bootloader support luks/zfs? 2015-05-31 16:32:37 you do however need a small disk or something to put your UEFI partition (vfat 2MB) and /boot on 2015-05-31 16:32:43 grub does 2015-05-31 16:32:50 I don't use lvm at all, but I use luks everywhere 2015-05-31 16:32:58 https://thestaticvoid.com/post/2013/10/26/how-i-do-encrypted-mirrored-zfs-root-on-linux/ 2015-05-31 16:33:05 i also saw that but that method does use lvm 2015-05-31 16:33:32 wait no it doesn't. 2015-05-31 16:33:36 but it uses a md device 2015-05-31 16:33:39 aha 2015-05-31 16:33:51 only because hes' done what he's done with the keyfile 2015-05-31 16:34:01 and doesn't have a separate disk for his /boot and UEFI partition 2015-05-31 16:34:40 the first guide the guy just bought a CF card and put it into an IDE converter 2015-05-31 16:34:54 its not like you use that volume after your system is booted and the kernel is loaded into memory 2015-05-31 16:35:21 what was interesting about the second guide is the use of dracut and puppet 2015-05-31 16:35:54 but i'll probably go with the first method as i don't want to use a keyfile 2015-05-31 16:36:16 i think gentoo is probably the best supported distribution for zfs on root 2015-05-31 16:36:23 (besides FreeBSD) 2015-05-31 16:37:14 but i tend to prefer linux to bsd 2015-05-31 16:37:26 bhyve is still very new 2015-05-31 16:39:38 isn't zfs fuse only ? 2015-05-31 16:40:15 no 2015-05-31 16:40:19 not ZoL 2015-05-31 16:41:05 https://wiki.gentoo.org/wiki/ZFS 2015-05-31 16:45:39 Coolgeek here ? :D 2015-05-31 16:46:08 and unlike btrfs raid 6 i probably won't lose all my data ;) 2015-05-31 16:46:58 hehe Chipster1 :) 2015-05-31 16:47:54 Coolgeek do you abandon arch ? 2015-05-31 16:48:02 "cool geek" and "c hipster" 2015-05-31 16:48:10 xD 2015-05-31 16:48:40 nope, I just lurk here, maybe for a side project 2015-05-31 16:50:39 ok 2015-05-31 17:35:49 which usb ethernet devices have good linux compatibility? 2015-05-31 17:41:33 atheros will probably be well supported 2015-05-31 17:47:10 ahills - see http://www.reddit.com/r/linux/comments/1eihho/quick_question_usb_ethernet_adapter_with_linux/ 2015-05-31 17:48:29 thanks 2015-05-31 20:06:40 I came across this today: http://troglobit.com/finit.html it seems to fit the paradigm of alpine pretty well (it's small and does what it claims) anyone else played around with it before? 2015-05-31 20:09:57 no, I have been using daemontools to manage my processes that require supervision 2015-05-31 20:16:24 I'm going to try it with debian right now (having systemd in a container makes systemd take up more resources than the actual apps >.<) 2015-05-31 20:17:16 debian base install adds lots of overhead other than systemd 2015-05-31 20:17:40 true 2015-05-31 20:17:58 if it's good, you should make an aport for it! 2015-05-31 20:18:05 Possibly 2015-05-31 20:18:58 It's kinda funny, I have my debian base install slimmed down as much as possible 2015-05-31 20:21:07 To a point where I have it use busybox for most things (unfortunately, due to some issues with bootstrapping, coreutils is kinda required...) 2015-05-31 20:42:50 Why wouldn't you just use a different distribution? 2015-05-31 20:52:16 if you are ok with 32 bit antix 15 is debian without systemd http://antix.mepis.org/index.php?title=Main_Page 2015-05-31 20:53:02 ahills: because I know my way around debian 2015-05-31 20:53:32 As to why I'm not using Alpine, Samba seems to have issues with musl libc when running in ADDC mode 2015-05-31 20:53:51 Or maybe I'm too stupid to set it up correctly 2015-05-31 20:53:56 what's ADDC? 2015-05-31 20:54:06 Active Directory Domain Controller 2015-05-31 20:54:30 oh haha 2015-05-31 20:54:42 I just know that I had it working on debian before and i couldn't get it to work on alpine 2015-05-31 20:54:45 I only use samba for unix-to-unix cifs with posix extensions 2015-05-31 20:58:14 hi 2015-05-31 20:58:27 nameservers from al are broken 2015-05-31 20:58:29 http://www.digwebinterface.com/?hostnames=alpinelinux.org&type=&compare=on&useresolver=8.8.4.4&ns=auth&nameservers= 2015-05-31 20:59:27 good night 2015-05-31 21:40:45 Just the apt index is almost 100MB D: