2014-07-01 12:02:15 nwmcsween: what feature of btrfs-progs is broken? and what release are you running? 2014-07-01 12:11:31 nwmcsween: hmm, I guess mkfs.btrfs is broken: 2014-07-01 12:11:38 stat("rootfs", 0x770615c4c228) = -1 ENOENT (No such file or directory) 2014-07-01 12:11:41 stat("proc", 0x770615c4c228) = -1 ENOENT (No such file or directory) 2014-07-01 12:11:44 stat("sysfs", 0x770615c4c228) = -1 ENOENT (No such file or directory) 2014-07-01 12:11:47 stat("mdev", 0x770615c4c228) = -1 ENOENT (No such file or directory) 2014-07-01 12:11:50 stat("devpts", 0x770615c4c228) = -1 ENOENT (No such file or directory) 2014-07-01 12:11:53 stat("shm", 0x770615c4c228) = -1 ENOENT (No such file or directory) 2014-07-01 12:22:22 yes mkfs.btrfs is borked 2014-07-01 12:22:34 had to do some trickery to get it to format 2014-07-01 12:35:54 can i mkfs.btrfs file.img to reproduce it? 2014-07-01 12:36:26 could anyone please file a bug on bugs.a.o with info how to reproduce and the details known so far 2014-07-01 12:36:40 just so we dont forget it 2014-07-01 12:38:05 https://www.google.com/about/appsecurity/patch-rewards/ 2014-07-01 12:43:45 hi to all, re: Bug #3105 2014-07-01 12:44:02 it happens here too 2014-07-01 12:44:36 cause is slow devices like r8169 (Realtek) 2014-07-01 12:45:10 I have a local patch for that: http://pastebin.com/raw.php?i=2UxiJLe4 2014-07-01 12:46:30 re: Bug #3105 , i have two network card, eth0 link down, and eth1 link up, after doing setup-alpine after network part,but before repository i get this msg: Error this script should be called from udhcpc 2014-07-01 12:47:07 AmatCoder actually it may be that what you did.. 2014-07-01 12:49:21 ncopa: dd if=/dev/zero of=/tmp/btrfs1.img bs=1G count=1; losetup /dev/loop1 /tmp/btrfs1.img; mkfs.btrfs /dev/loop1 2014-07-01 12:49:47 this is the offending syscall I think: open("/sys/block//1/loop/backing_file", O_RDONLY) = -1 ENOENT (No such file or directory) 2014-07-01 12:50:25 crow: yes, same here... I did: '/etc/init.d/networking restart' and 'setup-alpine' again 2014-07-01 12:50:31 I only have: /sys/block/loop1/loop/backing_file 2014-07-01 12:52:39 maybe we should patch busybox to have more retries by default 2014-07-01 12:52:41 AmatCoder thanks for hint 2014-07-01 12:53:02 i added a comment for #3105 2014-07-01 12:53:09 it looks like its configurable 2014-07-01 12:55:21 ncopa-desktop:~$ sudo mkfs.btrfs -f /dev/loop1 2014-07-01 12:55:21 WARNING! - Btrfs v3.14.2 IS EXPERIMENTAL 2014-07-01 12:55:21 WARNING! - see http://btrfs.wiki.kernel.org before using 2014-07-01 12:55:21 Performing full device TRIM (1.00GiB) ... 2014-07-01 12:55:21 Turning ON incompat feature 'extref': increased hardlink limit per file to 65536 2014-07-01 12:55:22 fs created label (null) on /dev/loop1 2014-07-01 12:55:24 nodesize 16384 leafsize 16384 sectorsize 4096 size 1.00GiB 2014-07-01 12:55:26 Btrfs v3.14.2 2014-07-01 12:57:39 i am not able to reproduce 2014-07-01 12:59:04 i was able to create btrfs on a temp lvm too 2014-07-01 13:00:02 ncopa: I will try 'udhcpc_opts' right now...I go to reboot 2014-07-01 13:00:54 I had issues with uclibc and btrfs, not sure about musl releases. 2014-07-01 13:07:43 seems like mkfs.btrfs work on alpine edge 2014-07-01 13:40:02 crow, ncopa: 'udhcpc_opts' fix the issue... 2014-07-01 13:40:39 'udhcpc_opts -t 5' is enough here 2014-07-01 13:41:25 with 'udhcpc_opts -t 3' (the default) fails 2014-07-01 13:42:23 very well 2014-07-01 13:46:53 AmatCoder thank you, how did you test it? did you make new iso or? 2014-07-01 13:48:22 no, I just edit '/etc/network/interfaces' and after 'lbu ci' 2014-07-01 13:48:41 i see :) 2014-07-01 13:53:00 AmatCoder: do you think you could add a comment on #3105 that 5 sec was enough for you? 2014-07-01 13:53:27 if 5 is a good value for others i'll patch busybox to use that as default 2014-07-01 13:53:49 ncopa: sure! 2014-07-01 14:18:14 mhm, looks like alsa-lib header (alsa/pcm_old.h) is referencing versioned symbols, even tough alsa-lib in alpine is compiled "--without-versioned". http://pastebin.com/uJuMxRGL 2014-07-01 14:19:04 some content of alsa/pcm_old.h: asm(".symver snd_pcm_hw_params_get_rate,snd_pcm_hw_params_get_rate@ALSA_0.9"); 2014-07-01 14:22:40 hm 2014-07-01 14:25:24 from very brief googling it is supposedly alsa-lib problem - old API is exposed, even when without-versioned doesn't provide it. fun stuff. 2014-07-01 14:50:52 fluidsynth is in testing, so maybe I'll use it instead of timidity 2014-07-01 15:01:57 clandmeter: how did you get btrfs-progs to format? 2014-07-01 15:02:28 i think i umounted something 2014-07-01 15:09:57 i dont remember what i search for on google, but something pointed me in that direction 2014-07-01 15:14:36 nwmcsween: i think if you umount proc, you should be able to fool btrfs 2014-07-01 15:15:17 nope 2014-07-01 15:16:04 thanks though 2014-07-01 15:16:33 umount it and create the mounts file in proc (something like that) 2014-07-01 15:17:21 the force option in btrfs should really do what it says though 2014-07-01 15:33:36 ah I got it working 2014-07-01 15:33:48 umount the /.modloop 2014-07-01 15:34:19 it's an issue with the current livecd 2014-07-01 15:40:50 hmm now mount won't mount it bleh 2014-07-01 15:45:15 also is it known that the kernel oops on the second boot under virtualbox? 2014-07-01 15:46:37 do_nmi... \n native_apic_mem_write * 4 2014-07-01 16:28:02 ncopa, for some reason i have a /gir-1.0 directory owned by udev package... 2014-07-01 16:28:05 this seems bogus 2014-07-01 17:37:17 Hi ! 2014-07-01 17:42:48 Is anybody here ? I want to install alpine on a computer, but I don't have any cdrom 2014-07-01 17:42:58 is it even possible ? 2014-07-01 17:46:30 yes 2014-07-01 17:47:01 use something like unetbootin to create a bootable usb drive from the iso 2014-07-01 17:48:40 Well, I don't have a "physical" USB drive in fact.. (I'm using drivedroid, an android app) 2014-07-01 17:49:59 I guess I'll have to seek an USB drive then 2014-07-01 17:51:21 that might work too 2014-07-01 17:53:38 I borrowed an USB drive anyway 2014-07-01 18:16:10 hmm, drivedroid looks interesting. it makes a virtual usb mass storage device and exports it via the phone's usb port? 2014-07-01 18:16:25 yep 2014-07-01 18:16:31 it works pretty nicely 2014-07-01 18:16:32 i didn't realize that was possible 2014-07-01 18:16:50 And provide an interface to download the common distros 2014-07-01 18:17:05 (as well as gparted, RIP, ...) 2014-07-01 18:17:09 i want an app that uses the same technique at the usb lebel.... 2014-07-01 18:17:39 ...but provides a virtual, read-only image of the photos dir that's generated on demand 2014-07-01 18:17:56 it can even emulate a CD drive 2014-07-01 18:18:04 (if your kernel uspports it) 2014-07-01 18:18:28 i.e. it only keeps the virtual FAT directory structure in memory, and when a file is accessed, pulls the data from the corresponding files on the android /sdcard fs 2014-07-01 18:18:49 mmmh 2014-07-01 18:18:50 so that you can read your data as usb mass storage (fast) rather than mtp (insanely slow and buggy) 2014-07-01 18:19:03 that'd be good 2014-07-01 18:19:45 technically, "when a file is accessed" means "when a sector corresponding to part of that file via the in-memory virtual FAT structure" 2014-07-01 18:20:27 being read-only, restricting the exported files to a user-chosen data range, etc. would also be a huge advantage 2014-07-01 18:20:40 e.g. you could safely plug your phone in to photo printing kiosks 2014-07-01 18:20:46 I'm a bit noobish with filesystems heh 2014-07-01 18:21:28 sorry, all that's rather OT 2014-07-01 18:21:37 nah, no problem ;) 2014-07-01 18:21:41 I started it all :p 2014-07-01 18:21:42 i was just surprised that what drivedroid does is possible -- i didn't realize it could be done 2014-07-01 18:21:54 and the fact that it's possible makes an app i seriously want also possible :) 2014-07-01 18:22:25 anyway, back to my installation 2014-07-01 18:22:32 it takes root though? 2014-07-01 18:22:40 yep 2014-07-01 18:22:43 yeah 2014-07-01 18:22:54 obviously something like that isn't going to be possible without root 2014-07-01 18:23:32 is unetbootin mandatory ? 2014-07-01 18:23:44 can't I just dd the iso right on the usb stick ? 2014-07-01 18:24:07 i think bootable iso images are specific to the media type 2014-07-01 18:24:11 I'm not sure 2014-07-01 18:24:20 in the past i was unable to get a cdrom iso to boot when burned to a dvd 2014-07-01 18:24:23 hybrid images are possible, but I don't know if the alpine one is hybrid 2014-07-01 18:24:32 erf 2014-07-01 18:24:35 i had to find an actual cd :( 2014-07-01 18:24:54 so i suspect you might have problems with usb sticks too 2014-07-01 18:25:01 I don't have one... I hope it will work with unetbootin then 2014-07-01 18:25:09 and my stick 2014-07-01 18:25:29 I've always used unetbootin from windows to get it done because I'm down to *one* linux box, a regular desktop, and one non-rooted android phone 2014-07-01 18:26:04 hehe 2014-07-01 18:27:07 less is more, sometimes 2014-07-01 18:27:15 but I still find room to do silly stuff with what I have 2014-07-01 18:28:53 btw, got a few questions about alpine 2014-07-01 18:29:11 I plan on using it as my personnal home server 2014-07-01 18:31:36 And I noticed that APKBUILDs looks really like Arch PKGBUILDs. But how can user share their ports ? 2014-07-01 18:33:48 i'm not sure that's so easy 2014-07-01 18:34:43 even if the package build format is similar (no idea if this is true), you're building for fairly different system environments 2014-07-01 18:35:22 how are they different ? 2014-07-01 18:35:36 the build method is the same isn't it ? 2014-07-01 18:35:40 most obvious difference is that alpine is using musl, arch is using glibc :) 2014-07-01 18:35:53 yeah sure 2014-07-01 18:36:18 so if there are packages that have problems building out of the box on musl, the arch package build will be difficult to mechanically transform to something that would work on alpien 2014-07-01 18:36:33 ncopa, do you have any figures on how many alpine packages (and what percentage) require some level of patching to build on musl? if it's low enough now this may not be a major concern :) 2014-07-01 18:36:47 Oh, yeah I understand that 2014-07-01 18:37:05 My question was probably not well formulated 2014-07-01 18:37:27 If I create an APKBUILD, how can I share it with the community ? 2014-07-01 18:58:33 z3bra: dd the iso is not supported (our iso is not hybrid) 2014-07-01 18:59:10 okay, thanks for the info :) 2014-07-01 18:59:21 z3bra: you can send apkbuild to our ml or pr on github 2014-07-01 18:59:43 creating an alpine usb is pretty easy 2014-07-01 18:59:51 just copy the contents of iso to usb 2014-07-01 18:59:57 and install syslinux bootloader 2014-07-01 19:00:26 Yeah I read that 2014-07-01 19:00:47 But I do not have syslinux installed on my current system 2014-07-01 19:00:58 so installing unetbootin was easier ;) 2014-07-01 19:01:11 whatever gets the job done :) 2014-07-01 19:01:15 exactly 2014-07-01 19:20:40 btw i noticed alpine has a libiconv package... is this used by anything? 2014-07-01 19:21:00 i would guess it's cruft from when uclibc was used (since uclibc lacked iconv) 2014-07-01 19:21:40 but if it's still needed, i'd like to know why so whatever deficiency it's making up for can be addressed in the upcoming improvements to musl iconv 2014-07-01 19:24:34 i think its leftover from ucibc 2014-07-01 19:26:10 dalias: http://alpinelinux.org/apk/main/x86_64/libiconv 2014-07-01 19:26:28 looks like only xen-libs 2014-07-01 23:00:29 hmm what i am doing wrong with lbu 2014-07-01 23:00:59 i specified the usb in LBU_BACKUPDIR in /etc/lbu/lbu.conf 2014-07-01 23:01:13 lbu ci does not write anything 2014-07-01 23:01:38 during setup-aline i did following: 2014-07-01 23:02:54 Avaiable disks are: 2014-07-01 23:02:56 sda (2.0 GB Multiple Card Reader ) 2014-07-01 23:02:58 Which disk(s) would you like to use?.... [none] 2014-07-01 23:03:00 Enter where to store configs (.....) [usb] 2014-07-01 23:03:02 Enter apk cache direcotry (.....) : none 2014-07-01 23:05:56 ok it does work /media/usb/ is /dev/usbdisk which is symlink to /dev/sda1 2014-07-02 09:26:26 hi 2014-07-02 09:31:40 Are the musl headers installed by default ? 2014-07-02 09:34:10 nevermind, found it alone 2014-07-02 10:16:33 i would like to do the 3.0 update today, which would be most "authoritative" howto for going from 2.7 to 3.0? 2014-07-02 10:17:05 oh, and might it be smart to test in a vm first? presumably, huh? 2014-07-02 10:23:35 darkfaded: http://alpinelinux.org/edge-musl 2014-07-02 10:24:07 ncopa: we should make that url more visable on www 2014-07-02 10:24:09 ty! 2014-07-02 10:24:21 actually *make* it visible 2014-07-02 10:24:36 it seems to be for edge users only/ 2014-07-02 10:24:38 ?? 2014-07-02 10:24:59 the process should be the same for 2.7 to 3.0 afaik 2014-07-02 10:25:27 most important part are the static tools 2014-07-02 10:25:49 you can always fix it with those installed. 2014-07-02 10:25:55 i wouldn't have thought of busybox-static :> 2014-07-02 10:26:10 so very helpful already 2014-07-02 10:26:43 if you can edit it, it would be good if you made clear it's generally the thing to use, not just for edge 2014-07-02 10:29:31 i recognize it's from the announcement mail or something back then :) 2014-07-02 11:07:42 clandmeter: yes you are right, we should write a "how to upgrade to v3.0" doc 2014-07-02 11:07:47 maybe on wiki? 2014-07-02 11:07:54 yes 2014-07-02 11:57:18 while using lbu, is there a way to do apk update && apk upgrade and write changes to disk? 2014-07-02 12:12:43 crow: apk update and upgrade always writes to memory on tmpfs install 2014-07-02 12:13:03 lbu_commit will write overlay changes to storage 2014-07-02 13:01:13 crow: when running from RAM, you can update packages with apk 2014-07-02 13:01:31 to use the updated packages on reboot, you must have package cache enabled 2014-07-02 13:01:38 which will write the packages to flash 2014-07-02 13:01:43 the exception is the kernel 2014-07-02 13:01:58 to update the kernel, you must update the boot media 2014-07-02 13:17:28 forgot in what package is mpstat 2014-07-02 13:17:33 fcolista: thanks for your quick and excellent work on Bug #3090 2014-07-02 13:18:07 clandmeter thanks 2014-07-02 13:18:15 np ginjachris 2014-07-02 13:18:18 tdtrask thanks to, i enabled cache like written here http://wiki.alpinelinux.org/wiki/Local_APK_cache 2014-07-02 13:18:45 i found this line quite weird in alpinelinux wiki: So you now have a run-from-RAM distro that can do a yum upgrade or apt-get dist-upgrade 2014-07-02 13:19:33 vmstat 2014-07-02 13:19:46 crow: agreed, that does look weird 2014-07-02 13:22:25 cant clean cache is now: 2014-07-02 13:22:26 apu4c:~# apk cache -v sync 2014-07-02 13:22:26 WARNING: Ignoring APKINDEX.9ac7f32e.tar.gz: No such file or directory 2014-07-02 13:24:30 indeed there is no such file or direcotry in /var/cache/apk but where is this info located 2014-07-02 13:26:11 crow: updated the wiki page 2014-07-02 13:42:48 tdtrask nice, you know what is this info about miss cache file 2014-07-02 14:04:42 crow: apk will usually check in /etc/apk/repositories......however, looks like you are on a diskless install maybe? Not done much with them but I have a vague memory of a similar message on a disk install 2014-07-02 14:04:53 I ignored it, it was benign and went away 2014-07-02 14:37:48 ginjachris yes its a diskless install, 2014-07-02 14:44:11 crow: whats in your repository file? 2014-07-02 15:00:12 cat /etc/apk/repositories 2014-07-02 15:00:12 /media/usb/apks 2014-07-02 15:00:12 http://nl.alpinelinux.org/alpine/v3.0/main 2014-07-02 15:00:34 and oher urls but they are commented 2014-07-02 15:05:22 /media/usb/apks is existing? 2014-07-02 15:53:43 clandmeter yes inside is an x86_64 directory 2014-07-02 16:48:28 booting 3.0.1, I get 'mdev: bad line 73: 2 tokens found, 3 needed'; both in Virtualbox and USB boot. unique? bad? 2014-07-02 17:03:32 rushed code 2014-07-02 17:25:07 is the default syslinux.cfg compressed? it's unreadable to me 2014-07-02 17:25:20 unable to boot. used setup-bootable 2014-07-02 17:28:25 this is the peculiarness: http://we.tl/sPy4xbKAVa 2014-07-02 17:40:10 so, I made my own customizing the default from the ISO, but still ... `setup-bootable` produced the ^ jabber, no good 2014-07-02 18:13:07 crow: I'm not sure about the cache sync WARNING 2014-07-02 18:13:22 where is your cache link pointing and what does it contain? 2014-07-02 19:06:29 Aspi: Perfectly ok syslinux.cfg here 2014-07-02 19:06:55 Though never used setup-bootable, I just munged with existing iso. 2014-07-02 19:07:18 meh, I should've just copied and then installed GRUB 2014-07-02 19:07:22 is syslinux faster? 2014-07-02 19:09:04 for the lazy ones (I know you're there), copy-paste'd at a bin: http://sebsauvage.net/paste/?6963d13808e47418#w9gxUWuTYh0HqeGg5tXyg42WYdhtV0rfDkOibmprcRU= 2014-07-02 19:09:15 it's the contents of the generated syslinux.cfg 2014-07-02 19:31:40 hi 2014-07-02 19:35:28 the bootable USB drive I just created with unetbooting don't boot up 2014-07-02 19:35:38 Did anybody encountered this issue recently ? 2014-07-02 19:36:01 where did it fail? 2014-07-02 19:36:14 I just did a UNetBooting install, with success 2014-07-02 19:36:19 mmmh 2014-07-02 19:36:22 from windows ? 2014-07-02 19:36:24 Windows host 2014-07-02 19:36:24 or linux ? 2014-07-02 19:36:28 okay 2014-07-02 19:36:35 I'm running azrchlinux atm 2014-07-02 19:36:58 the setup is really uncomplicated for alpine 2014-07-02 19:37:18 how does your / look atm? 2014-07-02 19:37:39 my / ? 2014-07-02 19:37:51 file system. what's gotten copied? 2014-07-02 19:38:02 oh, from the usb key 2014-07-02 19:38:05 my bad ^^ 2014-07-02 19:38:24 . .. .alpine-release apks boot ubnpathl.txt 2014-07-02 19:38:42 so, you should have everything you need 2014-07-02 19:38:42 2sec 2014-07-02 19:39:23 everything you need, yes. so prolly the bootloader is the problem, and it's prolly the thing most error-prone as well 2014-07-02 19:39:34 from an automatic program GUI shizz 2014-07-02 19:39:41 unetbootin^ 2014-07-02 19:39:43 I'll try dd'ing the syslinx mbr by hand 2014-07-02 19:40:15 why not. as you're an archer, I reckon installing syslinux by hand is no problem 2014-07-02 19:40:20 https://wiki.archlinux.org/index.php/Syslinux 2014-07-02 19:40:39 well, archlinux ncludes an install script :P 2014-07-02 19:40:59 But I know of to write a mbr ^^ 2014-07-02 19:41:01 naaaah, duuude! you're ruining the whole point of arch 8D 2014-07-02 19:41:06 haha 2014-07-02 19:41:08 also, wtf? 2014-07-02 19:41:24 I haven't. but I do use arch :P 2014-07-02 19:42:30 wish me luck 2014-07-02 19:42:43 meh :/ 2014-07-02 19:42:47 why not use syslinux tools? 2014-07-02 19:43:18 https://wiki.archlinux.org/index.php/Syslinux#Automatic_Install easy peasy 2014-07-02 19:44:11 but it will install syslinux on my hard drive.. 2014-07-02 19:44:53 anyway, does not boot :/ 2014-07-02 19:44:55 ah, I see. so you actually have to dd the mbr 2014-07-02 19:45:13 what /does/ happen with your USB when you try and boot it? 2014-07-02 19:45:39 nothing 2014-07-02 19:45:48 it just fall bakcs to my hard drive 2014-07-02 19:46:57 so you have no boot loader. 2014-07-02 19:47:25 my unetbootin installed syslinux 4.0* btw, while the embedded alpine setup installed 6.0* 2014-07-02 19:48:00 6.0 here too 2014-07-02 19:48:14 just to get the possibility out in the clear: if you can boot the ISO in some way, you can install from there. that's what I do 2014-07-02 19:48:27 I can't boot the iso :/ 2014-07-02 19:48:33 (I don't have any CD) 2014-07-02 19:48:38 do you have a windows? 2014-07-02 19:48:43 nope 2014-07-02 19:48:51 Please USB, you're our last hope.. 2014-07-02 19:48:52 I know it's terrible, but you could just use it to create the usb 2014-07-02 19:49:01 the doze 2014-07-02 19:49:07 yeah I know, but I don't :/ 2014-07-02 19:49:43 ok, you're up for an unorthodox way-more-difficult-than-necessary, but fun way? :P 2014-07-02 19:49:52 I always do when I get pissed 2014-07-02 19:49:59 let's go manual 2014-07-02 19:50:12 yeah, copying the iso, writing mbr by hand and such ? 2014-07-02 19:50:21 I'm doing it right now 2014-07-02 19:50:36 I just reformatted / partitionned my usb stick, just to be sure :P 2014-07-02 19:52:10 my supernatural universal alpine install tutorial guide: erase and reformat USB stick (fdisk) -> 'mkfs -F32 /dev/sd??' -> copy .iso contents to /dev/sd?? -> install ... the mother boot loader ... !!GRUB!! 2014-07-02 19:52:29 nononononononoooooo 2014-07-02 19:52:34 whynot? 2014-07-02 19:52:43 I don't know grub 2014-07-02 19:53:07 you /know/ syslinux? idk any of them, but I'm not afraid to use something new 2014-07-02 19:53:24 well, I used grub before 2014-07-02 19:53:38 I bricked a laptop because of it :/ 2014-07-02 19:53:47 and now, I'm scared ^^ 2014-07-02 19:53:53 in worst case, if you have two USB sticks, you can install GRUB on one of them, then use 'setup-bootable' alpine script to install to your other pen 2014-07-02 19:54:01 what, how did you do that? 2014-07-02 19:54:07 I.. 2014-07-02 19:54:09 I don't know 2014-07-02 19:54:27 plus, GRUB is now written from scratch, not the same code, if that's calming :9 2014-07-02 19:54:30 :) * 2014-07-02 19:54:59 and you can install grub on a stick without installing it on your system ? 2014-07-02 19:55:08 absolutely 2014-07-02 19:55:12 (I mean, with all the config generators it uses) 2014-07-02 19:55:32 idk how the heck syslinux is somehow unable, I can't really accept that either, *goes to read* 2014-07-02 19:55:44 oh and https://wiki.archlinux.org/index.php/Grub#Manually_creating_grub.cfg 2014-07-02 19:55:51 Aspi: I hate grub with burning sensation of thousand suns, but otherwise agree. 2014-07-02 19:55:57 oooooh fuck 2014-07-02 19:56:02 I know what's wrong ! 2014-07-02 19:56:11 le bootable flag 2014-07-02 19:56:14 sshame on me 2014-07-02 19:56:16 haah 2014-07-02 19:56:19 aaaaah 2014-07-02 19:56:22 so ashamed xD 2014-07-02 19:56:24 katlogic: what why? why can ppl hate it? 2014-07-02 19:56:46 Aspi: Also, bringing up archlinux in here is heresy. Archlinux is falled angel who strayed from the path of simplicity. 2014-07-02 19:56:55 It started with getting rid of simple /etc, ended up with systemd. 2014-07-02 19:57:02 GRUB is part of this heretic movement too. 2014-07-02 19:57:04 hahaha. the evil BIOS god got his daily doze of annoyal 2014-07-02 19:57:36 oh, is grub !simple? 2014-07-02 19:57:45 as said, I only use it, I don't know it 2014-07-02 19:57:49 Well, it is like gnome among bootloaders 2014-07-02 19:58:00 it is *easy* to use under good conditions 2014-07-02 19:58:00 now I've got the taste 2014-07-02 19:58:20 on my arch, I use awesome wm 2014-07-02 19:58:31 but now, have you tried to use gnome in alpine? 2014-07-02 19:58:36 nope? 2014-07-02 19:58:38 alpine is really bad conditions for gnome 2014-07-02 19:58:42 skip that ? 2014-07-02 19:58:54 I use xfce 2014-07-02 19:59:17 in alpine 2014-07-02 19:59:54 xfce almost as bad as gnome on alpine 2014-07-02 20:00:19 isn't alpine aimed at headless systems ? 2014-07-02 20:00:21 installed it, half of stuff not working, buttons refusing to work because no randr extension in xvfb etc etc 2014-07-02 20:00:30 icons missing 2014-07-02 20:01:02 so, that was a weird search: https://duckduckgo.com/?q=alpine+gnome 2014-07-02 20:01:03 again, its not really fault of xfce just that alpine is simply too spartan for something which spawns 20 helper processes and dbus daemons on desktop launch 2014-07-02 20:01:30 awesome works pretty ok, though not really what I wanted for terminal server 2014-07-02 20:01:43 talking about gnome our xfce? 2014-07-02 20:01:54 both were broken more or less same 2014-07-02 20:02:18 also both are about the same weight, xfce can be seen as gnome a bit stripped down 2014-07-02 20:02:21 ACTION crosses his fingers... 2014-07-02 20:02:29 damn it 2014-07-02 20:02:31 I'm very new to alpine atm; what would you recommend for a DE here? 2014-07-02 20:02:40 Aspi: well, only awesome works ok for me 2014-07-02 20:02:47 well, because its like syslinux/lilo 2014-07-02 20:02:52 it does not do much, thus not much to break 2014-07-02 20:02:59 IT'S ALIVE ! 2014-07-02 20:03:02 didn't know it was apk'd :) 2014-07-02 20:03:07 :) 2014-07-02 20:03:20 Aspi: get used to compiling stuff from source 2014-07-02 20:03:23 I forgot the boot flag, seriously... 2014-07-02 20:03:31 xvnc was tricky part getting running 2014-07-02 20:03:39 reinstalled the usb key like 10 times yesterday.. 2014-07-02 20:04:36 you kind of deserve it 8D 2014-07-02 20:04:43 yeah xD 2014-07-02 20:05:21 btw, is it hard to port applicatons to musl ? 2014-07-02 20:05:21 I can't remember to ever have had to set the boot flag myself. so either some tool has done it each time, or I've never encountered a machine which requires it 2014-07-02 20:05:39 I read something about it. it should be very easy 2014-07-02 20:06:20 I planning tu run things like a web server, quake3, and such... 2014-07-02 20:06:21 http://www.musl-libc.org/faq.html 2014-07-02 20:06:32 (I'm afraid about quake3 though...) 2014-07-02 20:09:35 katlogic: are you an experienced alpine user, or did you just not bother troubleshooting the DEs and went for something that worked? 2014-07-02 20:09:53 I find multiple xfce apks, maybe it'll all work if I use it correctly 2014-07-02 20:10:46 Aspi: They'll most likely work ok, I had rather unusual setup 2014-07-02 20:10:54 xvfb + xvnc 2014-07-02 20:11:03 and xrdp on top of that 2014-07-02 20:11:19 I'm using xfce4, works for the little I've tried it 2014-07-02 20:11:56 after startx, my xf86-video-intel is working; it is not before, so I suspect !dynamic module loading in alpine? 2014-07-02 20:11:58 I've yet to manage get xpra/xdummy working 2014-07-02 20:12:10 as those are supposedly better than chain of vnc hacks 2014-07-02 20:12:32 are you basically vnc-ing everything? even boot msgs and such 2014-07-02 20:12:38 nah, just x 2014-07-02 20:12:57 xvfb is some contradictionary name 2014-07-02 20:13:03 yeah :/ 2014-07-02 20:13:08 if it is X virtual frame buffer :P 2014-07-02 20:13:16 well, it is. 2014-07-02 20:13:26 though I might try run xen pv guest 2014-07-02 20:13:29 with vnc enabled 2014-07-02 20:13:37 vfb!=vt? 2014-07-02 20:13:47 not sure if xen's vnc is any good 2014-07-02 20:14:17 xen seems like a solid project, but that is maximum subjective 2014-07-02 20:14:29 oh boy, alpine+xen is pure love 2014-07-02 20:15:05 originally i got into alpine because it was first thing which managed to boot xen alpine ramdisk and load encrypted guests from local disk 2014-07-02 20:15:36 plausible deniability by local disks being just pure gibberish :) need pxe / usb to boot. this was impossible to accomplish with debian, archlinux ... 2014-07-02 20:15:52 and nearly native in alpine? O_o 2014-07-02 20:16:07 wtf is alpine really. I'm astonished, again and again 2014-07-02 20:16:12 how many devs are there? 2014-07-02 20:16:19 had to dig the internals enough to make that custom boot ramdisk 2014-07-02 20:16:31 was pleasantly surprised the guts are so small 2014-07-02 20:16:44 I have an impression this is a nearly single-guy hobby project by ncopa 2014-07-02 20:16:46 the setup- things are good read about how to go about things, though not particularly helpful for custom stuff 2014-07-02 20:17:05 which is probably good approach 2014-07-02 20:17:16 want fancy partition setup? read the script how things are done 2014-07-02 20:17:41 *taking a look* 2014-07-02 20:18:14 Well, there are hidden undocumented gems 2014-07-02 20:18:28 like want debootstrap? just `setup-disk /mnt/dir` 2014-07-02 20:19:47 do you know how I can load the intel video module at boot? 2014-07-02 20:22:23 don't blacklist it 2014-07-02 20:22:31 :) 2014-07-02 20:23:26 my intel driver is not loaded until I startx. after that, my terminal is using intel as well 2014-07-02 20:26:17 can vservers host any distro ? 2014-07-02 20:26:55 z3bra: what do you mean? 2014-07-02 20:27:15 I plan on running a quake3 server 2014-07-02 20:27:34 But i'm not sure it will build against musl 2014-07-02 20:27:39 z3bra: use lxc 2014-07-02 20:28:11 usually you need specially adjust the distro image in container 2014-07-02 20:28:14 I'm not used to this. what's the difference between lxc and vservers ? 2014-07-02 20:28:16 things like debian/centos work ok tho 2014-07-02 20:28:28 vserver is just too crappy :) 2014-07-02 20:28:34 ah ? 2014-07-02 20:28:35 otherwise the two are almost same thing 2014-07-02 20:28:49 also, lxc is native kernel feature, vserver is out-of-tree patch 2014-07-02 20:28:51 they're put in front on the wiki 2014-07-02 20:28:56 i more trust the first than the latter 2014-07-02 20:29:46 we do not support vserver anymore 2014-07-02 20:29:59 lxc is the way to go now 2014-07-02 20:30:01 okay 2014-07-02 20:30:11 I'll go with lxc then 2014-07-02 20:30:12 z3bra: The wiki usually lists whatever someone was doing, its not "official", in fact often outdated :( 2014-07-02 20:30:24 Ah... ok 2014-07-02 20:30:48 At least the installation method still works :P 2014-07-02 20:31:18 ACTION actually did full matryoshka doll with alpine 2014-07-02 20:31:27 xD 2014-07-02 20:31:28 xen->lxc->kvm 2014-07-02 20:31:47 I can't use any of them :P 2014-07-02 20:32:19 My needs are more basics anyway 2014-07-02 20:32:26 Back to my current problem: LVM 2014-07-02 20:32:53 I'm new to this too, so can I "wrap" an existing disk with LVM ? 2014-07-02 20:33:00 lvm is overkill unless you run dozens of guests and need to be serious about management of storage 2014-07-02 20:33:07 Or I'll need to wipe it in order to use LVM ? 2014-07-02 20:33:28 lvm is a bit like partitions inside a partition 2014-07-02 20:33:40 Yeah I get the concept 2014-07-02 20:33:54 so if you insist on using it, you first need to make real partition for it somehow 2014-07-02 20:34:04 usually resize the existing system partition 2014-07-02 20:34:26 or just backup what you have as .tar.gz and wipe disks and start afresh with lvm for everything including system 2014-07-02 20:34:31 So i need to use a virgin partition for it 2014-07-02 20:35:01 yeah, can't be backed by currently used/mounted partitons 2014-07-02 20:35:07 Let me explain my problem, that'll be easier 2014-07-02 20:35:17 well it can using losetup file but thats rather hackish :) 2014-07-02 20:35:59 My current home server runs Archlinux, and the /home is on a separate hard drive (with 3 partitions: /home, /backup, /var) 2014-07-02 20:36:38 I can't back up the /home, because it's >300Go, and I don't have a large enough hard drive 2014-07-02 20:36:46 sure just resize it then 2014-07-02 20:37:02 google for howto resize2fs and fdisk 2014-07-02 20:37:07 So I'd like to know if I can just "include" this partition in a new LVM setup, and resize it later then 2014-07-02 20:37:27 oooh 2014-07-02 20:37:30 thats even more tricky 2014-07-02 20:37:41 depends, what partition is in front of it? 2014-07-02 20:37:52 you need a bit of space in there unfortunately 2014-07-02 20:38:10 Okay, I'll just give up then... 2014-07-02 20:38:16 indeed 2014-07-02 20:38:24 just size it down to make space for new partition to use for lvm 2014-07-02 20:38:26 It contains all my valuable datas, and i don't want to compromise them 2014-07-02 20:39:11 i'll just keep this partition static then 2014-07-02 20:39:27 you can run lxc servers from there no prob 2014-07-02 20:39:34 it can work with directory 2014-07-02 20:40:47 http://sprunge.us/iLHR 2014-07-02 20:40:54 here's my CURRENT server map 2014-07-02 20:41:16 And I'll install alpine on another PC anyway, not on this sda drive 2014-07-02 20:41:50 well, it would be possible to lvmify sdc 2014-07-02 20:41:56 assuming you can afford to wipe /var 2014-07-02 20:42:03 I can 2014-07-02 20:42:36 1) backup partition table and write down at which sector sdc2 and sdc3 starts 2014-07-02 20:42:40 The only things that interest me here is /home 2014-07-02 20:43:01 2) wipe partition table clean, create single big partition 2014-07-02 20:43:01 (and /backup, if possible) 2014-07-02 20:43:08 3) create lvm on that partition 2014-07-02 20:43:28 now the trick is to create lv's on that lvm so that sector starts on original partitions 2014-07-02 20:43:40 smart 2014-07-02 20:43:43 this is rather nontrivial and you need to know lvm really well to not screw up 2014-07-02 20:43:54 I don't know LVM at all 2014-07-02 20:43:57 hehe 2014-07-02 20:44:05 Let's keep it static then ^^ 2014-07-02 20:44:20 Thanks for your help though 2014-07-02 20:44:36 ahh, installed awesome; I'm home :) 2014-07-02 20:45:12 has there ever been any consideration of linux-libre-grsec? 2014-07-02 20:45:39 yes, it does exist: https://parabolagnulinux.org/packages/libre/x86_64/linux-libre-grsec/ 2014-07-02 20:45:42 just patch it? 2014-07-02 20:45:45 or that 2014-07-02 20:45:46 :) 2014-07-02 20:46:06 can I just copy it over? I'm ashamed I've never LFS yet 2014-07-02 20:46:44 uh? 2014-07-02 20:46:56 well if its a package just install it :) 2014-07-02 20:47:41 ACTION never understood this obsession with being "pure-gnu" 2014-07-02 20:47:58 unless there is real threat from lawyers, it sounds awfully impractical 2014-07-02 20:48:08 especially whatever is covered by software patents is missing :/ 2014-07-02 20:50:26 the feeling of booting a system where you know "this system ... is completely clean of Uncle Sam ... he can try as much as he want to take this, but his own laws forbid him ... this system is infinitely hackable" 2014-07-02 20:50:42 Would it be easy to replace busybox with another set of coreutils ? 2014-07-02 20:50:48 ditto 2014-07-02 20:51:21 is busybox like awesome wm to gnome, or dropbear to openssh? 2014-07-02 20:51:32 ie. is it lacking, or just simple 2014-07-02 20:52:00 lacking some stuff compared to full blown coreutils 2014-07-02 20:53:50 Aspi: busybox is kinda so-so 2014-07-02 20:54:04 it works. that's it :) 2014-07-02 20:54:07 some of things in there range from ok to god-awful 2014-07-02 20:54:28 most of coreutils are apkd 2014-07-02 20:54:36 so when grep segfaults on you, just apk add grep 2014-07-02 20:54:48 are mdev and busybox two separate things here ? 2014-07-02 20:55:12 eg, apk remove busybox 2014-07-02 20:55:19 will it keep mdev ? 2014-07-02 20:55:46 ooh i'd not do that 2014-07-02 20:55:56 ah ? 2014-07-02 20:56:00 just keep it 2014-07-02 20:56:06 install whatever you miss on top of it 2014-07-02 20:56:24 why ? 2014-07-02 20:56:27 it will get rid of busybox version of that and you'll get the "better" version 2014-07-02 20:56:42 yeah, by overriding the symlink 2014-07-02 20:56:51 just apk install coreutils 2014-07-02 20:57:13 apk install grep, apk install psmisc etc etc 2014-07-02 20:57:22 its better to be selective, those are full featured but also somewhat bloated 2014-07-02 20:57:30 i'd say its more akin to openssh vs dropbear 2014-07-02 20:57:37 dropbear really sucks for everyday use, but is tiny 2014-07-02 20:58:08 in fact, I plan on using sbase, ubase and hbase form here http://git.2f30.org/ 2014-07-02 20:58:18 (creating my own package then) 2014-07-02 20:59:03 if anyone made base utils which are not as hostile as busybox ones 2014-07-02 20:59:11 yet it was tiny as busybox, that would be awesome 2014-07-02 20:59:54 http://alternativeto.net/software/toybox-linux-command-line-utilities-/ 2014-07-02 21:00:04 yeah i know the guy from #musl 2014-07-02 21:00:25 Oh, okay =) 2014-07-02 21:00:44 no apk either :/ 2014-07-02 21:01:10 apk REALLY looks like PKGBUILDs 2014-07-02 21:01:26 I personally packaged some stuff for archlinux and crux 2014-07-02 21:02:16 yeah, it is really sad there is no easy way to submit apks tho 2014-07-02 21:02:20 and after reading 2-3 APKBUILDs, It looks easy to create an apk for toybox 2014-07-02 21:02:23 ah, yeah :/ 2014-07-02 21:02:37 Maybe.. can user create their own online repo ? 2014-07-02 21:02:48 something like apt repositories and such ? 2014-07-02 21:02:50 i dont really want to muck with whole aports and git when i'm creating apk ad-hoc 2014-07-02 21:03:18 hmm, i'm not sure how the signing thing works 2014-07-02 21:03:59 gpg --sign ? 2014-07-02 21:04:06 or is it too simple ? :p 2014-07-02 21:04:35 nah how whole key management is handled 2014-07-02 21:04:41 obviously i'd have to install keys somewhere 2014-07-02 21:04:57 and there is no gpg by default so it does verify it internally somehow 2014-07-02 21:05:51 Well, idk then 2014-07-02 21:06:02 maybe some proprietary signing algorithm :P 2014-07-02 21:06:17 built right in the apk binary * 2014-07-02 21:06:21 hehe 2014-07-02 21:07:35 yeah it is right in the binary, though pretty standard otherwise 2014-07-02 21:07:50 not in the binary apk, i mean the package binary 2014-07-02 21:08:21 /etc/apk/keys 2014-07-02 21:08:39 apparently just add there 2014-07-02 21:08:57 easy =) 2014-07-02 21:09:09 and it will check wether a pakcage is trusted ? 2014-07-02 21:09:16 yup 2014-07-02 21:10:17 what are the partitions created by setup-alpine ? 2014-07-02 21:10:45 the usual dark triad, /, /boot and swap, setup-alpine is rather basic 2014-07-02 21:10:58 okay 2014-07-02 21:11:10 vim `which setup-disk` 2014-07-02 21:11:14 cause I'll have to reuse my old /var an /home 2014-07-02 21:11:24 (from my external drive) 2014-07-02 21:11:28 for that do what clandmeter said 2014-07-02 21:11:32 use setup-disk for custom installations 2014-07-02 21:12:16 I see 2014-07-02 21:12:51 it does the whole disc formatting, but if you setup-disk /dir it will just put stuff into that dir 2014-07-02 21:13:00 yeah 2014-07-02 21:13:10 but the partition will have to be mounted then 2014-07-02 21:13:25 So I'll have to put the server down.. 2014-07-02 21:13:34 regarding busybox, i dont think you can remove it even if you install coreutils (or you need to hack deps) 2014-07-02 21:13:45 deps ? 2014-07-02 21:13:46 indeed, busybox seemed rather essential 2014-07-02 21:13:57 some tools require busybox ? 2014-07-02 21:14:07 also things would catch fire immediately the second you'd mkinitfs 2014-07-02 21:14:14 many things in alpine depend on busybox 2014-07-02 21:14:15 as it expects you to have busybox around 2014-07-02 21:14:31 it also provides our shell 2014-07-02 21:14:33 So it depends on busybox, not on the tools it provide 2014-07-02 21:14:36 got it :) 2014-07-02 21:14:48 the shell ? 2014-07-02 21:14:48 nah, i'm sure it would be easy to fix hard deps if there are any 2014-07-02 21:15:00 its just that busybox provides *lot* of stuff 2014-07-02 21:15:05 you'd have to replace it all 2014-07-02 21:15:26 anyway, that's for a server, and alpine is small enough for keeping busybox installed :P 2014-07-02 21:15:28 no ash? just use bash etc :) but there are dozens of things like that 2014-07-02 21:15:31 abuild also adds it when any script is found 2014-07-02 21:15:34 Was just curious 2014-07-02 21:15:35 things which are not coreutils but busybox provide em 2014-07-02 21:16:11 but with out busybox trigger, you can just install and replace busybox features (symlinks) 2014-07-02 21:16:12 z3bra: yeah, one would get rid of busybox out of sheer hate :) 2014-07-02 21:16:26 or by replacing it with something similiar, let's see how far the toybox thing will fare 2014-07-02 21:17:08 ^^ 2014-07-02 21:17:40 http://landley.net/toybox/status.html 2014-07-02 21:17:49 it seems it might *almost* work 2014-07-02 21:17:55 eh, I hope I'll be able to use all the software I used before 2014-07-02 21:18:36 bind 9, thttpd, umurmur, cgit, ... 2014-07-02 21:18:46 bind 9? eeew :) 2014-07-02 21:18:56 anyone of you run encrypted? I'll do later on 2014-07-02 21:18:57 run tinydns like true alpine hipster you are! 2014-07-02 21:19:00 katlogic, i don't know anything else 2014-07-02 21:19:09 katlogic, I'll be glad too :P 2014-07-02 21:19:29 I installed bind9 for training purpose only 2014-07-02 21:19:32 z3bra: just kidding, well those alternatives are rather newbie hostile :) 2014-07-02 21:19:43 It's the only soft I'll be glad to throw out 2014-07-02 21:19:49 but if all you need is jut caching dns, dnscache/dnsmasq works just well if not better 2014-07-02 21:20:03 if you have complex zone files to host, use bind and dont bother with tinydns 2014-07-02 21:20:10 nah I have my own domain . machines 2014-07-02 21:20:15 it's for my local network 2014-07-02 21:20:28 the zone is not really complex tough 2014-07-02 21:20:39 Just staic ip -> hostnames 2014-07-02 21:20:48 Aspi: i see some ppl using encrypted over lvm. they had some issues migrating to 3.0 with initramfs issues (probably solved now) 2014-07-02 21:21:28 Oooooh nice !!! 2014-07-02 21:21:36 umurmur and cgit are already packaged :D 2014-07-02 21:22:00 git.alpinelinux.org duh ;-) 2014-07-02 21:22:01 Aspi: yeah, encrypted llvm LV work out of box 2014-07-02 21:22:17 Aspi: encrypted llvm PV is more secure (as in obfuscated), but you'll have to muck with mkinitfs 2014-07-02 21:22:26 s/llvm/lvm/g damnit 2014-07-02 21:22:40 :) 2014-07-02 21:22:52 Aspi for me setup-bootable does produce also junks so i did it manually using these steps http://wiki.alpinelinux.org/wiki/Create_a_Bootable_USB#Manual 2014-07-02 21:23:28 (yeah, I peeked there too. and looked at the default config) 2014-07-02 21:24:28 katlogic: PV is abbreviation of? 2014-07-02 21:24:50 ah, Physical Volume. how is it different, what is the point? 2014-07-02 21:25:05 I'd like an encrypted fs, not necessarily the partition table 2014-07-02 21:25:06 PV is the "top level", the physical partition lvm 2014-07-02 21:25:20 yeah, in that case encrypted LV is ok 2014-07-02 21:25:41 aaalso, I installed awesome, and indeed it is; my fonts are all awkward, any idea on how to fix? 2014-07-02 21:26:45 encrypted PV basically entails `cryptsetup tcryptOpen --tcrypt-hidden /dev/sda` 2014-07-02 21:27:00 ie super obfuscated, but nobody can tell what the disk really is at is made of pure gibberish 2014-07-02 21:27:41 okay, so font problem is fixed through 'apk add ttf-dejavu' 2014-07-02 21:27:41 nice 2014-07-02 21:27:42 (also makes booting the fun part, usually pxe & usb is used) 2014-07-02 21:28:02 truecrypt is potentially compromised, no? 2014-07-02 21:28:11 this is just truecrypt emulation 2014-07-02 21:28:16 not truecrypt itself 2014-07-02 21:28:22 Aspi i also needed to edit syslinux.conf regarding serial console, and only way was the manuall way. seesm setup-bootalbe borked 2014-07-02 21:28:24 so truecrypt is a good idea still? 2014-07-02 21:28:29 I mean, I have a bad feeling about it now 2014-07-02 21:28:31 you could use openPlain too, but that does not support key management 2014-07-02 21:28:51 Aspi maybe to submit a bug report 2014-07-02 21:28:53 Aspi: on linux there is 0 reason to use actual truecrypt, as cryptsetup implements it fully. 2014-07-02 21:28:57 I'll do dm-crypt if I crypt 2014-07-02 21:29:11 Aspi: the truecrypt format is useful for advanced things 2014-07-02 21:29:40 I go simple 8D 2014-07-02 21:29:48 you heretic 2014-07-02 21:29:54 and another huge bonus: those disks are readable in windows too :) 2014-07-02 21:30:26 crow: do you have an account? 2014-07-02 21:30:27 ie boot with my trusty windows xp + truecrypt usb to play battlefield :> 2014-07-02 21:30:32 ohmy, I must create one one day 2014-07-02 21:30:43 hahaha :P 2014-07-02 21:30:58 windows /on/ your usb? 2014-07-02 21:31:13 yeah 2014-07-02 21:31:31 the disks are complete gibberish, nothing can boot from it 2014-07-02 21:31:34 always need usb to boot 2014-07-02 21:32:26 crow: do you know boot loaders well? is it possible that the reason I don't get no video drivers at boot is because it's not in 'initrd modules='? 2014-07-02 21:33:04 Aspi: or just /etc/modules 2014-07-02 21:33:28 modules= is to get module in early, as in make initrd do stuff 2014-07-02 21:33:45 crypt needs there btw then? 2014-07-02 21:35:01 ah, so I need i915 modules 2014-07-02 21:35:04 module* 2014-07-02 21:36:00 Aspi: for cryptsetup you might need to mkinitfs anyway 2014-07-02 21:36:21 i'm not sure theres even wiki for alpine FDE 2014-07-02 21:37:01 also careful with that stuff, unless your cpu does aes-ni it tends to be slow :( 2014-07-02 21:39:19 are you aware of any package that would handle upnp on alpine ? 2014-07-02 21:39:46 client or server? 2014-07-02 21:40:35 upnpc or clever shell script doing a bit of netcat abuse :> 2014-07-02 21:40:42 server 2014-07-02 21:40:57 miniupnpd 2014-07-02 21:41:05 you'll have to compile from souce i'm afraid 2014-07-02 21:41:10 that's for sharing medias with the multimedia box provided by my ISP 2014-07-02 21:41:19 i don't mind compiling 2014-07-02 21:41:28 as long as it compiles agains musl :> 2014-07-02 21:41:44 oh, dlna :/ 2014-07-02 21:41:48 yeah 2014-07-02 21:42:03 ie you want to be dlna server 2014-07-02 21:42:13 servio is best, but is rather huge 2014-07-02 21:42:15 java :> 2014-07-02 21:42:35 i'd rather forget sharing 2014-07-02 21:42:36 :) 2014-07-02 21:42:48 just samba cifs like rest of us normal people? :) 2014-07-02 21:43:18 I hope the box handle this 2014-07-02 21:43:23 any respectable device can play from that, dlna is just pure evil 2014-07-02 21:43:40 unfortunately sometimes necessary for tvs which can decode only two codecs 2014-07-02 21:43:44 at least, it works oob 2014-07-02 21:44:56 if tv is crap, get raspi/mk802 + xbmc 2014-07-02 21:45:13 xbmc is way cooler anyway 2014-07-02 21:47:18 rpi+xbmc: I always dreamt of having a full-blown HTPC, though always feared the time it would take. I had >3 rpis when I discovered RaspBMC. it was not a very hard decision to sacrifice one for the media center :) 2014-07-02 21:47:36 nah tv has nothing to do with it 2014-07-02 21:47:56 I have an independent media server 2014-07-02 21:48:18 i have huge steaming box showed in closet 2014-07-02 21:48:32 as 20TB of family photo albums make a lot of noise 2014-07-02 21:48:40 and raspi as frontend to that 2014-07-02 21:49:02 that's what I want in fact 2014-07-02 21:49:18 my server (alpine) will be your streaming box 2014-07-02 21:49:28 and the media center from my ISP will be your raspi 2014-07-02 21:49:33 i run alpine and debian in lxc there 2014-07-02 21:49:37 (a Freebox V6) 2014-07-02 21:49:49 i had to keep the debian thing for servio 2014-07-02 21:49:57 as my neighbor has only crap tv (we share lan) 2014-07-02 21:50:31 hehe 2014-07-02 21:50:58 Aspi yes i have an account, 2014-07-02 21:51:11 Aspi i dont have VGA on device where i run alpinelinux sorry 2014-07-02 21:51:13 anyway, I'll need to find a way to stream media using upnp 2014-07-02 21:52:06 ok so how do I write to my usb O:) 2014-07-02 21:52:35 dd 2014-07-02 21:52:38 write what ? 2014-07-02 21:52:40 does alpine run in such a way that I can safely pull my USB after booting, or does it require it still? 2014-07-02 21:52:50 run-from-ram 2014-07-02 21:52:53 yes 2014-07-02 21:53:02 it means the OS is loaded into ram 2014-07-02 21:53:10 wait, I'll take katlogic's advice: I'll look at lbu script 2014-07-02 21:53:17 But the USB is still needed for persistent changes I guess 2014-07-02 21:53:41 I'd like to write to the USB. alpine writes to USB sometimes, such as when writing to apk cache 2014-07-02 21:54:00 then you need to keep the usb plugged in 2014-07-02 21:54:12 we are geniuses, aren't we? 2014-07-02 21:54:13 (it's the "data" install iirc) 2014-07-02 21:54:38 you see, I use RAM-install; still, I can write to the cache. I'd like to do that :) 2014-07-02 21:56:00 eh, i don't know this enough to help you I'm afraid 2014-07-02 21:56:03 Aspi if you will need to use some extra packages, you will need to enble cache for apks (i found this today) and thanks to tdtrask and clandmeter i did it, so packages are loaded after reboot 2014-07-02 21:56:16 Aspi kernel is a bit tricky 2014-07-02 21:58:39 I did it :) katlogic seems to work, because all I needed was to read /sbin/lbu: I needed to remount /media/usb as rw (`mount -o remount,rw /media/usb`), do my stuff, and -o remount,ro again. easy peasy 2014-07-02 21:59:02 Aspi yes as written in wiki :) 2014-07-02 21:59:35 http://wiki.alpinelinux.org/wiki/Local_APK_cache 2014-07-02 21:59:36 -.- I've barely glanced at the wiki. sinning at first day 2014-07-02 21:59:47 congratulation :) 2014-07-02 22:00:00 yes, I have the apk cache up and running. setup-alpine calls setup-apk-cache 2014-07-02 22:00:23 Aspi yea i did not enabled it during setup-alpine 2014-07-02 22:00:28 ah 2014-07-02 22:01:03 well, now I have a very nice desktop up and running. I only miss a couple of apks before use, and I'd love for example Iceweasel 2014-07-02 22:01:50 Good night everybody ! 2014-07-02 22:02:47 good night :D 2014-07-02 22:05:43 for installing alsa, do I install alsaconf or alsa-utils? 2014-07-02 22:22:29 ACTION is calling it a night as well 2014-07-02 22:22:40 cya 2014-07-03 03:51:03 ncopa, you around? 2014-07-03 06:48:22 morning 2014-07-03 06:48:27 dalias: i'm here 2014-07-03 06:49:18 hey 2014-07-03 06:49:30 have you seen gcc 4.9.0's -fstack-protector-strong ? 2014-07-03 06:49:55 once 4.9.x is no longer broken, i think it would make sense to switch alpine from using -fstack-protector-all to -fstack-protector-strong 2014-07-03 06:50:15 i havent seen it 2014-07-03 06:50:15 as an example of why, see the hideous code gcc generates with -fstack-protector-all for: 2014-07-03 06:50:27 double f(double x, double y) { return x/y; } 2014-07-03 06:50:48 (this function has utterly no use for stack protector because it has no use for a stack) 2014-07-03 06:50:57 he 2014-07-03 06:51:16 so what is the diff with -strong? 2014-07-03 06:51:29 they skip stackprotector when its useless? 2014-07-03 06:53:10 basically 2014-07-03 06:53:27 -fstack-protector has the wrong condition for using protection: "buffer 8 bytes or larger" 2014-07-03 06:53:44 -fstack-protecot-all also has a wrong condition: "use it all the time, even when it's provably useless" 2014-07-03 06:54:31 -fstack-protector-strong has almost the right condition: "use it when the address of any automatic var is taken" 2014-07-03 06:54:47 however now that i think about it, that's still not right 2014-07-03 06:55:13 because the stack could be clobbered by writing through a pointer passed by the caller 2014-07-03 06:55:57 the right condition is really: "any function that writes via a pointer, or calls another function that takes a pointer argument" 2014-07-03 06:56:34 it's pretty amazing how gcc keeps getting this wrong... 2014-07-03 06:57:01 lol 2014-07-03 06:57:19 it sounds like they are using a try-and-fail approach 2014-07-03 06:57:47 and for each fail add another -fstack-protector-newtry 2014-07-03 06:57:55 yes..... 2014-07-03 06:58:25 hopefully we'll eventually get -fstack-protector-done-right :) 2014-07-03 06:58:47 i don't see why they keep adding new ones, such that software has to be aware of whatever the latest broken trial is... 2014-07-03 06:58:57 rather than just fixing -fstack-protector to do the right thing 2014-07-03 06:59:17 and adding -fstack-protector=obsolete1 etc. with the old behavior if you really want it 2014-07-03 06:59:27 probably becase fixing it might break things that depend on the broken behaviour... 2014-07-03 06:59:28 yeah 2014-07-03 06:59:43 this looks like a candidate for ewontfix :) 2014-07-03 06:59:52 sounds like that yes :) 2014-07-03 07:00:50 i suppose -fstack-protector-strong is not supported in gcc-4.8? 2014-07-03 07:00:55 aright 2014-07-03 07:00:57 right 2014-07-03 07:01:03 it was added in 4.9 2014-07-03 07:01:11 and actually i wouldn't switch from -all to -strong 2014-07-03 07:01:20 if you want maximum hardening, it's a regression 2014-07-03 07:01:25 e.g. consider 2014-07-03 07:01:47 yes, we saw it 2014-07-03 07:02:01 i remember talking with ncopa about it 2-3 months ago 2014-07-03 07:02:05 void foo() { char buf[10]; bar(buf); } void bar(char *p) { gets(p); } 2014-07-03 07:02:21 and i think we were undecided, but leaning towards -strong instead of -all 2014-07-03 07:02:29 bar happily returns to the attacker-provided return address 2014-07-03 07:03:15 oops, gets won't do it 2014-07-03 07:03:23 because it only overflows up 2014-07-03 07:03:34 but anything that overflows down would 2014-07-03 07:03:43 maybe the gcc ideas is that overflow-down is rare 2014-07-03 07:03:44 sounds like to me that the approach is "right thing is hard, so let's do the approximation that i can do easier. i'll fix it later if it becomes an issue." 2014-07-03 07:03:56 imo the right thing is actually easiest 2014-07-03 07:04:02 in long run yes 2014-07-03 07:04:07 "function uses pointers" is basically the right condition 2014-07-03 07:04:42 for any function that does not use pointers, ssp is provably useless 2014-07-03 07:05:16 and for any function that performs writes via pointers, proving it's safe against stack smashing is very hard 2014-07-03 07:05:56 that sounds like a candidate for ewontfix 2014-07-03 07:06:05 which hopefully gcc dvs will read 2014-07-03 07:06:08 devs* 2014-07-03 07:10:25 so i dont forget: http://bugs.alpinelinux.org/issues/3136 2014-07-03 07:11:05 ha 2014-07-03 07:11:10 "Have you considered a “valued argument”, i.e. -fstack-protector=foo (where foo in {regular, strong, all})? Seems more extensible." 2014-07-03 07:14:33 fyi 4.9 is not presently usable 2014-07-03 07:14:39 it has bugs that affect compiling musl 2014-07-03 07:20:14 ok 2014-07-03 07:20:30 we hold it back then, unless we have patches to fix it 2014-07-03 07:29:03 ncopa, https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61144 2014-07-03 07:30:13 ncopa it seems something is wrong with setup-bootable 2014-07-03 07:30:20 if you check log from last nichtg 2014-07-03 07:30:42 Aspi have had same problem, should i submit a bug report 2014-07-03 07:34:28 crow: yes please, file a bug report, i will likely not have time to look at it today :-/ 2014-07-03 07:35:56 ncopa no problem 2014-07-03 07:36:06 manualy way described in wiki works fine 2014-07-03 07:45:19 ncopa https://bugs.alpinelinux.org/issues/3137 2014-07-03 08:00:02 hi ! 2014-07-03 08:25:58 Does alpine have a "default" group for all users 2014-07-03 08:38:26 z3bra: not that i know 2014-07-03 08:39:11 okay 2014-07-03 09:51:19 Morning all! 2014-07-03 09:52:24 Is it me or is Bug #3135 at best, confusing as hell? And probably in fact nonsensical? Has anyone seen a setup like this in reality? I certainly haven't 2014-07-03 09:53:27 ginjachris: yeah, didnt even know that works 2014-07-03 09:53:49 but apparently that is what some isps does... 2014-07-03 09:55:58 It's news to me. I've asked around and none of my colleagues have seen that. Is the ISP just breaking the rulebook for the hell of it? 2014-07-03 09:56:23 I don't understand how you can reach a default gateway on another network....to do so you need to go via a local hop 2014-07-03 09:57:09 I'm wondering if the guy simply needs to configure a seond interface? 2014-07-03 09:58:04 anyway I'll refrain from adding a comment on bug tracker....it may come across as rude! 2014-07-03 09:59:09 i suppose he could add pre-up ... hacks to get it working 2014-07-03 10:01:29 I'll leave that at your discretion. 2014-07-03 10:01:36 Seems very strange to me though! 2014-07-03 11:04:16 i think 'onlink 2014-07-03 11:04:47 err... in "ip route" you can add 'onlink' attribute to force assuming the IP is onlink even if there's no matching subnet route 2014-07-03 11:04:53 but that's really a hack 2014-07-03 11:05:00 it's broken network design 2014-07-03 11:06:07 is that a point to point route? 2014-07-03 11:06:18 i have bad memories of those 2014-07-03 11:06:47 onlink pretend that the nexthop is directly attached to this link, even 2014-07-03 11:06:47 if it does not match any interface prefix. 2014-07-03 11:10:29 that all sounds dubious to me. It may work, but it's not what I'd call best practice :S 2014-07-03 11:11:05 +1 2014-07-03 11:14:10 correct 2014-07-03 11:14:24 it probably will work. but is essentially broken network design. 2014-07-03 11:29:19 I have a pretty fresh system. I have installed xorg'n'stuff, and xf86-video-intel. before startx, I only have a low-res terminal, but after running startx and exiting the wm, I have a full-res terminal. any ideas on what modules I'm missing? 2014-07-03 11:30:14 I lsmod'ed, and found that i915 was not loaded before startx; though, `modprobe i915` freezes the whole display (while system must be working, as ACPI shutdown works. 2014-07-03 11:30:37 maybe fbcon? 2014-07-03 11:30:43 modprobe fbcon 2014-07-03 11:31:40 *trying* 2014-07-03 11:32:35 thank you sir :) fbcon with i915 it is 2014-07-03 11:32:46 you can add those to /etc/modules 2014-07-03 11:32:50 yep 2014-07-03 11:33:09 order is important? 2014-07-03 11:33:16 i think you can even load them early by adding 'kms' to features in /etc/mkinitfs/mkinitfs.conf 2014-07-03 11:33:27 i dont think order matters, but i dont know 2014-07-03 11:36:14 I see cramfs and squashfs is used from place to place, which one is in use now? 2014-07-03 11:44:53 squashfs 2014-07-03 11:45:11 long time since we dropped cramfs 2014-07-03 11:45:14 is it loaded? 2014-07-03 11:45:19 might i kept it for compat 2014-07-03 11:45:50 it is in default mkinitfs.conf 2014-07-03 11:45:58 and no squashfs 2014-07-03 11:46:12 any wiki for mkinitfs? 2014-07-03 11:56:46 I can't find one. 2014-07-03 11:57:18 but you can regenerate the initfs just by doing "mkinitfs" 2014-07-03 11:58:52 run "mkinitfs -L" for a list of available features 2014-07-03 11:59:51 these can be added to /etc/mkinitfs/mkinitfs.conf 2014-07-03 12:01:37 *testing* 2014-07-03 12:01:55 "Booting kernel failed: Invalid argument" 2014-07-03 12:02:09 what did you do? 2014-07-03 12:02:11 should the default mkinitfs.conf work, or does it want some tampering? 2014-07-03 12:02:56 I have a RAM install booted from USB, then I remounted /media/usb as rw, and ran `mkinitfs -o /media/usb/boot/grsec`. I have backups 2014-07-03 12:03:36 ah, I'm running a disk install 2014-07-03 12:03:47 default .conf is fine 2014-07-03 12:04:26 added in 'bootchart', did 'mkinitfs', then rebooted. All fine 2014-07-03 12:04:55 I'll have to pass you over to those more experienced than I! 2014-07-03 12:05:01 what is notable, is that the generated /boot/grsec is 10MB in size, while default is 2MB; grsec.gz is however identical to the default one (sha1-tested) 2014-07-03 12:05:51 hmmm...with -o option do you need to add the .gz extension? 2014-07-03 12:06:02 ah, maybe 2014-07-03 12:06:39 also, when you run mkinitfs, do you get lots of "No such file or directory. Skipped"? 2014-07-03 12:06:46 I daren't use a different outfile. I'll likely break something knowing my luck 2014-07-03 12:07:31 cp /boot/grsec{,.bak} ? 2014-07-03 12:08:08 ok, the mentioned identical grsec.gz was because I used cp instead of mv -.- 2014-07-03 12:09:14 do you get the "directory skipped" msgs? I suspect they're the problem 2014-07-03 12:10:38 /usr/bin/{ac,lastcomm} /usr/sbin/{dump-utmp,dump,acct,accton,sa} and /lib/modules/3.14.8-1-grsec/kernel/drivers/block/cpqarray* are the ones missing, so skipped 2014-07-03 12:20:50 another thing is that I need both grsec and grsec.gz; mkinitfs only creates one. does it create both at your's? 2014-07-03 12:38:39 mkinitfs and kernel upgrade is not really supported on live-install/run-from-ram/tmpfs-install 2014-07-03 12:39:27 its on my todo list to write a script that can apk fetch new kernel and generate new initramfs image and new modloop 2014-07-03 12:40:11 could I generate it on a sys and transfer it? 2014-07-03 12:40:33 yes 2014-07-03 12:40:36 well 2014-07-03 12:40:43 i think you need to do it manually 2014-07-03 12:40:56 ? transfer? 2014-07-03 12:41:08 generate the modloop 2014-07-03 12:41:35 how do you generate the initrds for the ISO? 2014-07-03 12:41:36 i am normally to lazy for that so i simply use the alpine-iso makefile to generate it 2014-07-03 12:41:43 ah, gotcha 2014-07-03 12:41:55 for livecd/liveusb you need 2014-07-03 12:41:58 1) kernel 2014-07-03 12:42:01 so I could customize the ISO, then use that 2014-07-03 12:42:01 2) initramfs 2014-07-03 12:42:04 3) modloop 2014-07-03 12:42:07 we cheat 2014-07-03 12:42:31 instead of installing all kernel driver in tmpfs, we mount /lib/modules on squashfs at boot 2014-07-03 12:42:40 and use that at bootup 2014-07-03 12:42:50 you can /etc/init.d/modloop stop after it booted up 2014-07-03 12:43:00 and then you can remove the boot usb/cdrom if you want 2014-07-03 12:43:37 you'll not be able to load drivers for hotplugged devices (ethernet over USB dongles etc) though 2014-07-03 12:43:59 because of no kms? 2014-07-03 12:44:09 no because the modloop squashfs image 2014-07-03 12:44:31 this is how we generate the .iso: http://git.alpinelinux.org/cgit/alpine-iso/tree/Makefile 2014-07-03 12:44:51 modloop rules: http://git.alpinelinux.org/cgit/alpine-iso/tree/Makefile#n94 2014-07-03 12:45:56 I'm not experienced in this field. for now, I'll /etc/modules my modules needs 2014-07-03 12:46:08 your approach, does it yield faster booting, is that the reason? 2014-07-03 12:46:08 yeah 2014-07-03 12:46:28 no, its for getting the higres earlier 2014-07-03 12:46:35 to mount the mods at a loop dev 2014-07-03 12:46:44 so i'd say use /etc/modules 2014-07-03 12:46:46 instead of installing 2014-07-03 12:46:51 yeah 2014-07-03 12:46:52 yeah, I know I know :) 2014-07-03 12:46:58 ah, okay 2014-07-03 12:47:31 I must LFS some time. this is still one of the dark fields for me, how linux kernel modules work 2014-07-03 12:48:42 I like the idea of automatic module need detection. I have the impression that that is kms. ohwell, nothing to bother with today 2014-07-03 12:49:12 did you read the posts about corrupt syslinux.cfgs from setup-bootable yesterday? 2014-07-03 12:51:39 i saw the posts yes 2014-07-03 13:26:51 found this the other day, a bit old but very useful: http://rute.2038bug.com/index.html.gz 2014-07-03 13:30:56 (in case you're wondering it's "LINUX: Rute User's Tutorial and Exposition", not some dodgy link) 2014-07-03 15:12:27 bye all, have a great evening! 2014-07-03 15:35:43 in what package is fsck.vfat 2014-07-03 15:47:12 is it possible to auto load a module when bringing up an interface? 2014-07-03 15:49:42 not sure if busybox supports pre-up/post-down like debian but you could try it 2014-07-03 15:50:03 pre-up /sbin/modprobe $MODULE in /etc/network/interfaces 2014-07-03 15:51:06 do you think it's possible with udev to load module based on usb bus address 2014-07-03 15:51:33 z3bra: I've thinking of packaging sbase/ubase myself. one problem though: since the utils are super minimal I guess you won't be able to build many of our APKBUILDS with it 2014-07-03 16:03:49 uggedal, see... 2014-07-03 16:04:03 At least, it would not hurt to install them 2014-07-03 16:10:07 with udev is it possible to , I have two identical usb wifi devices and I want to load each one with a different module they sit on different usb addresses different usb ports 2014-07-04 00:04:12 are there expected errors that the installer normally produces and can be ignored? I get "cp: can't stat '/var/run/*': no such file or directory" and "/lib/rc/sh/runscript.sh: line 215: can't create /sys/fs/cgroup/openrc/syslog/tasks: nonexistent directory" 2014-07-04 06:04:43 uniquenick well if unsure check bugs.alpinelinux.org and if there is no report about that, then you can submit a bug 2014-07-04 06:08:42 uniquenick i think there is also that first line on boot about 'mdev: bad line 73: 2 tokens found, 3 needed' which should be fixed https://bugs.alpinelinux.org/issues/3097 2014-07-04 07:11:49 Good Morning everyone! 2014-07-04 08:46:57 morning 2014-07-04 08:47:02 moin 2014-07-04 12:14:04 http://www.arm.com/products/tools/development-boards/versatile-express/juno-arm-development-platform.php this our new arm builder? 2014-07-04 12:28:19 those are paid in blood? 2014-07-04 21:01:43 hi ! 2014-07-04 21:07:14 I plan on installing alpine on my home server this week end, but I'd like to reuse my /var and /home 2014-07-04 21:07:26 will the setup-disk keep them safe ? 2014-07-04 21:08:14 I'm pretty sure it will keep my /home intact, but what about the /var ? 2014-07-04 21:08:38 will the install overwrite my directories ? 2014-07-04 21:32:01 z3bra: Yes, if you setup-disk without partitioning (you have to do that manually beforehand) 2014-07-04 21:34:04 I plan on running setup-disk on its own yes 2014-07-04 21:34:20 katlogic: with setup-disk can i set up disk encryption before the installation? 2014-07-04 21:34:32 (setup-disk /mnt, with my /var correctly mounted under /mnt) 2014-07-04 21:35:01 But I don't know what setup-disk will do 2014-07-04 21:35:24 maybe it just rm -rf /mnt/* or something before installing 2014-07-04 21:35:37 I could not find the content of the script online 2014-07-04 21:36:14 kocka: again, setup encrypted partition first, then setup-disk /mnt 2014-07-04 21:36:31 it works exactly like debootstrap if you're familiar with debian 2014-07-04 21:36:42 I'm not personnally :/ 2014-07-04 21:37:38 yeah, setup-disk is rather spartan, still not sure if thats for the worst or better 2014-07-04 21:38:39 if it were babysitting and provided fancy stuff like resizing ntfs partitions, manage custom partition layout and cryptsetup and whatever else ubuntu does, inexperienced people would shoot themselves in foot with that anyway 2014-07-04 21:38:53 yeah totally ^^ 2014-07-04 21:39:06 now it does either "delete fucking everything" or "i'll put stuff where you tell me to, but prepare it manually beforehand" 2014-07-04 21:40:02 that being said, fancy menus in debian installer are time savers :) 2014-07-04 21:40:04 But as my /var partition holds many valuable data (git repos, DNS zone files and quake3 server configs) I'd rather be sure that it will not just erase them 2014-07-04 21:40:30 well, i'd not disk-setup /var 2014-07-04 21:40:41 cant you just use use clean partition for that? 2014-07-04 21:40:59 if not, move everything you have there in new top level directory and it should be ok but meh 2014-07-04 21:41:21 It would be a time saver to just use my current /var as-is 2014-07-04 21:41:40 the problem would be potential directory collsions 2014-07-04 21:41:53 if you have something like, say, /var/usr now 2014-07-04 21:41:59 it would obviously clash 2014-07-04 21:42:12 what if I remove everything EXCEPT "special" directories ? 2014-07-04 21:42:14 so make it like /var/old_root/usr (ditto for all old top dirs there) 2014-07-04 21:42:33 My valuable dires are: /var/named, /var/git, /var/games 2014-07-04 21:42:48 blah 2014-07-04 21:43:00 just do: cd /var; mkdir old_var; mv * old_var 2014-07-04 21:43:08 and then setup-disk /var 2014-07-04 21:43:27 you can reorg stuff back from old_var after alpine is installed .. 2014-07-04 21:43:47 That was my question in fact, will setup-disk wipe the partition ? or just copy things to it ? 2014-07-04 21:43:59 just copy if you give it directory like that 2014-07-04 21:44:08 okay thanks 2014-07-04 21:44:22 another question though 2014-07-04 21:44:30 you said "setup-disk /var" 2014-07-04 21:44:43 but you probably meant "setup-disk /mnt" ? 2014-07-04 21:44:46 don't you 2014-07-04 21:44:48 yeah 2014-07-04 21:44:58 (with my "old" /var mounted to /mnt/var 2014-07-04 21:45:08 Okay, just to be sure =) 2014-07-04 21:45:31 but well, in that case setup-disk /mnt/var ... 2014-07-04 21:45:45 ah ? 2014-07-04 21:45:50 blah 2014-07-04 21:45:59 what mount command you'll use? 2014-07-04 21:46:09 mount /dev/sdc3 /mnt/var ? 2014-07-04 21:46:15 mount /dev/sda3 / 2014-07-04 21:46:19 huh? 2014-07-04 21:46:29 mount /dev/sda3 /mnt 2014-07-04 21:46:30 sorry 2014-07-04 21:46:35 oh 2014-07-04 21:46:49 so var is not partition, just directory in old root fs? 2014-07-04 21:46:49 mount /dev/sdb1 /mnt/var 2014-07-04 21:46:54 mount /dev/sdb2 /mnt/home 2014-07-04 21:46:58 ah i see 2014-07-04 21:47:01 yeah 2014-07-04 21:47:06 in that case dont do that 2014-07-04 21:47:08 but it's a partition on its own 2014-07-04 21:47:17 (on an external drive) 2014-07-04 21:47:19 just mount /dev/sdb1 /mnt 2014-07-04 21:47:39 those other dirs are super confusing and not really relevant 2014-07-04 21:47:57 as you are installing alpine *only* on /var right? 2014-07-04 21:48:01 no 2014-07-04 21:48:10 That will be a full install 2014-07-04 21:48:22 why not use sda3 then? 2014-07-04 21:48:27 its the old root anyway 2014-07-04 21:48:46 I plan on installing alpine on /dev/sda (sda1 -> /boot, sda2 -> swap, sda3 -> /) 2014-07-04 21:48:55 i see 2014-07-04 21:48:58 in that case 2014-07-04 21:49:02 forget about /var :) 2014-07-04 21:49:12 and use an external disk for /var (/dev/sdb1) and /home (/dev/sdb2) 2014-07-04 21:49:15 alpine wont touchit if you wont mount it 2014-07-04 21:49:31 just sda3, move dirs in old_root there, setup-disk, voial 2014-07-04 21:49:45 /boot is not really necessary either, it can use root partition for that 2014-07-04 21:50:08 yeah but I'd like to keep it separate 2014-07-04 21:50:18 as a "good practice" 2014-07-04 21:50:23 well, you have to install bootloader somehow later anyway 2014-07-04 21:50:32 what setup-disk /mnt does 2014-07-04 21:50:35 won't alpine install it by itself ? 2014-07-04 21:50:36 is create /mnt/bin etc etc 2014-07-04 21:50:39 *nothing else* 2014-07-04 21:50:59 yeah but doesn't it create directories in /mnt/var ? 2014-07-04 21:51:07 it does, so? 2014-07-04 21:51:11 that partition is not mounted 2014-07-04 21:51:24 you cant mix and match directories from old system like that 2014-07-04 21:51:31 just keep the old partitions tossed away 2014-07-04 21:51:35 deal with it after install 2014-07-04 21:51:50 the moment you overwrite those with installer you overthinked it :) 2014-07-04 21:52:22 mmh 2014-07-04 21:52:36 so I can reuse my old /var under alpine without problem ? 2014-07-04 21:53:05 after install, /var will be just directory on sda3 2014-07-04 21:53:22 naturally you can then put your old var in fstab, copy alpine structure there etc 2014-07-04 21:53:28 but do that *after* you booted alpine 2014-07-04 21:53:38 not before, it will screw things up royally :) 2014-07-04 21:53:49 okay that's the answer I wanted :p 2014-07-04 21:54:20 mmmh... I could also copy everything I want to keep on my old /home, and put it back in /var later... 2014-07-04 21:54:49 depends where you want to put your alpine / really 2014-07-04 21:55:15 so that I could mount /dev/sdb1 on install, wipe it and have setup-disk create a new /var on it 2014-07-04 21:55:17 rule of thumb: never ever setup-disk on more than one mount 2014-07-04 21:55:38 more than one mount ? 2014-07-04 21:55:41 how so ? 2014-07-04 21:55:44 or well, you can do that, but you have to make sure to fix fstab right away etc 2014-07-04 21:55:51 it wont boot 2014-07-04 21:55:51 yeah I know 2014-07-04 21:55:56 because it wont see the dir 2014-07-04 21:56:00 ah oki then 2014-07-04 21:56:10 well, I'll just edit /mnt/etc/fstab before reboot 2014-07-04 21:56:12 z3bra: it might also barf on the directory already existing 2014-07-04 21:57:00 ah.. 2014-07-04 21:57:54 anyhow, check install_mounted_root in setup-disk 2014-07-04 21:58:08 the problem is, it assumes mounted structure by dumb partitioner 2014-07-04 21:58:09 I tried to find setup-disk source, but couldn't 2014-07-04 21:58:14 so if there is something complex mounted beneath 2014-07-04 21:58:19 it might get too smart 2014-07-04 21:58:28 z3bra: just alpine git :) 2014-07-04 21:58:40 http://sprunge.us/hVbI 2014-07-04 21:58:45 will this work, for example ? 2014-07-04 21:59:06 (assuming /dev/sdb1 was just formatted and is empty) 2014-07-04 22:00:01 http://git.alpinelinux.org/cgit/alpine-conf/tree/setup-alpine.in 2014-07-04 22:00:02 there 2014-07-04 22:00:07 err 2014-07-04 22:00:10 setup-disk.in 2014-07-04 22:00:12 you get the idea :) 2014-07-04 22:00:19 ah, thanks 2014-07-04 22:00:54 that paste sounds awfuly dangerous 2014-07-04 22:01:15 why ? 2014-07-04 22:01:16 but its just opinion, perhaps you know perfectly well what you're doing, idont really know what you want to keep where 2014-07-04 22:01:52 well, nothing in that case. I would backup anything to another drive 2014-07-04 22:01:56 and put it back later 2014-07-04 22:02:58 z3bra: anyhow, /boot will be probably ok 2014-07-04 22:03:04 assuming you wipe it before anyway 2014-07-04 22:03:10 yeah 2014-07-04 22:03:18 and then just extlinux -i /mnt/boot 2014-07-04 22:03:23 no setup-alpine :) 2014-07-04 22:03:28 oh, okay 2014-07-04 22:03:31 (syslinux.cfg has to be there too) 2014-07-04 22:03:47 so I'll need to install syslinux by hand ? 2014-07-04 22:03:48 basically a lot of instructions in http://wiki.alpinelinux.org/wiki/Install_to_disk apply 2014-07-04 22:03:54 yeah, mbr and syslinux 2014-07-04 22:04:18 again, you must know how is this done, setup-disk is *just* like cp -va /zygote_system /mnt 2014-07-04 22:04:19 from chroot then I guess 2014-07-04 22:04:21 it does nothing else for you 2014-07-04 22:04:55 well, it seems to generate fstab and such too 2014-07-04 22:05:05 yeah it will do that 2014-07-04 22:05:13 for / guaranteed, for the rest not sure ymmv 2014-07-04 22:05:22 do tell if correct entries show up there 2014-07-04 22:05:42 anyway if i wanted to preserve /var for sure i'd not dare to mount it until everything is over :) 2014-07-04 22:05:55 /var and /home probably 2014-07-04 22:06:42 okay 2014-07-04 22:06:54 thnkq for your help 2014-07-04 22:06:57 :) 2014-07-05 06:23:39 have not checked v3.x iso, does syslinux work from booted iso ? 2014-07-05 06:30:15 if yes, then one can also prep a usb with qemu 2014-07-05 06:30:19 qemu-system-x86_64 -cdrom alpine-2.7.0-x86.iso -hdb /dev/sdb -boot d 2014-07-05 10:16:44 ncopa: what are you using as a mail client on your alpine desktop? 2014-07-05 21:54:25 errm, all my hvm domUs get stuck at "bios" after the 3.x update. any hints? 2014-07-05 21:57:34 due to the pygrub bug in 2.7 i had switched all to hvm. so, i'm doomed twice... 2014-07-06 00:13:21 different way of asking it: does anyone have a working hvm domU on 3.0? 2014-07-06 02:38:40 Yay got it on the third try 2014-07-06 14:37:32 anyone got an idea what's causing this: 2014-07-06 14:37:33 http://hastebin.com/ivemanovac.avrasm 2014-07-06 15:34:57 hi 2014-07-06 15:44:52 hi 2014-07-06 15:47:09 how long will be 3.0 beta? 2014-07-06 15:47:14 until 3.1? 2014-07-06 16:03:30 shafire: hopefully until hvm xen works ;) 2014-07-06 16:32:19 please, how can I download a package and build it locally ? 2014-07-06 16:32:39 (and also, create my own package) 2014-07-06 16:42:56 hi 2014-07-06 16:51:31 z3bra: http://wiki.alpinelinux.org/wiki/Aports_tree plus http://wiki.alpinelinux.org/wiki/Development_using_git should have most info 2014-07-06 16:51:52 and command reference: http://wiki.alpinelinux.org/wiki/Abuild 2014-07-06 16:52:11 thanks 2014-07-06 16:55:15 z3bra: are you using alpine 2.x or 3.x? 2014-07-06 16:55:44 3.0.1 2014-07-06 19:13:55 I have question regarding package creation 2014-07-06 19:14:08 Let's say my package provide a man page 2014-07-06 19:14:54 should I had "rm ..../$pkgname.1.gz" at the bottom, and create a package that will ONLY provide that manpage ? 2014-07-06 19:21:16 nevermind, I found by myself ^^ 2014-07-06 19:38:42 I have no success with the setup-bootable -U switch for creating UUIDs. how should an UUID entry look? 2014-07-06 19:40:14 I have success with 'append initrd=/boot/grsec.gz slpine_dev=UUID=*the uuid*:vfat ...', but that results in a nasty /media/UUID=*the uuid* device. is it nevertheless the right way? 2014-07-06 19:46:00 I like nasty. I'll keep the /media/UUID=$UUID :) 2014-07-07 06:43:20 morning 2014-07-07 06:43:28 kocka: I use claws-mail 2014-07-07 06:48:11 distrowatch is down.. whoops... 2014-07-07 06:48:20 "This website is currently unavailable due to account issues." 2014-07-07 06:49:32 yeah i'm surpised they managed to get by on some form of shared hosting 2014-07-07 06:52:04 Hi ! 2014-07-07 07:20:20 ncopa: ah thanks 2014-07-07 07:20:54 it can handle pgp keys and messages? 2014-07-07 07:21:01 yes 2014-07-07 07:21:50 cool 2014-07-07 07:22:17 i heard about this app earlier but then i forgot about it :/ 2014-07-07 07:22:43 on the weekend i setted up a mail vm on my desktop machine :) 2014-07-07 07:24:40 but i think i'll re-create it on a full disk encrypted alpine :) 2014-07-07 13:20:19 Hi ! 2014-07-07 13:20:33 which program would your recommend to fetch mails ? 2014-07-07 13:20:50 I used fdm before, which is great, but it does not compile against musl 2014-07-07 13:31:50 trying alpine i get: "who are you: unable to ... username" 2014-07-07 13:34:16 root 2014-07-07 13:34:25 LoCuteAs: ^ 2014-07-07 13:35:36 huh? 2014-07-07 13:36:12 have you installed alpine? 2014-07-07 13:36:22 or booting the ISO? 2014-07-07 13:37:45 i have a bshellz.net account. allpine installed. ssh'd in with my username. running alpine gives the above error. 2014-07-07 13:40:04 ah, then I'm less sure I can help 2014-07-07 13:40:53 you know anyting about install linux on a mac without a cd? net install? tftp? 2014-07-07 13:41:20 not on macs, sorry 2014-07-07 13:43:34 already got a pc linux cd. 2014-07-07 14:17:49 i think ti should be possible to install it with usb on MAC, but will likely require manual hackery 2014-07-07 14:19:21 getting the right modules might be tricky? 2014-07-07 14:19:45 anyways, he was running a iBook, PowerPC processor, said in private chate 2014-07-07 14:19:47 chat* 2014-07-07 14:21:43 Why can't I install umurmur through apk ? 2014-07-07 14:21:51 (it's in main/) 2014-07-07 14:24:23 did you do 'apk update' first? 2014-07-07 14:24:51 nope 2014-07-07 14:25:02 but it waas in main/ before I installed my system 2014-07-07 14:25:27 anyway, it does not work 2014-07-07 14:27:31 what do you get if you do "apk -U search umurmur"? 2014-07-07 14:28:48 segfault 2014-07-07 14:29:46 delete the umurmur.apk in /apks or /cache? 2014-07-07 14:29:51 redownload 2014-07-07 14:29:55 if everything else works 2014-07-07 14:30:59 there is something I don't get yet 2014-07-07 14:31:34 when doing "apk search", it will search in a remote repo 2014-07-07 14:31:55 and those repo are SUPPOSED to package everything in main/, right ? 2014-07-07 14:32:19 it will look in the locations specified in /etc/apk/repositories 2014-07-07 14:32:44 okay 2014-07-07 14:32:44 so if you have main repo defined in there, then yes. 2014-07-07 14:33:01 Yeah I have one 2014-07-07 14:33:22 But is it possible that my repo do not have a package from main/ ? 2014-07-07 14:33:28 ok. Are you using version 3.x? (musl in other words?) 2014-07-07 14:33:35 yeah musl 2014-07-07 14:33:46 i guess the repo might not have sync'd from the others yet 2014-07-07 14:33:54 okay 2014-07-07 14:34:02 but that seems unlikely tbh 2014-07-07 14:34:18 looks like the package was built: Monday, June 30, 2014 - 16:19 2014-07-07 14:34:19 How can I rerun the "which repo is the fastest" script ? 2014-07-07 14:34:32 setup-repo 2014-07-07 14:34:34 i think 2014-07-07 14:34:35 lemme check 2014-07-07 14:34:47 nope 2014-07-07 14:35:03 setup-apkrepos, maybe ? 2014-07-07 14:35:04 silly me 2014-07-07 14:35:07 exactly 2014-07-07 14:36:02 manaully check /etc/apk/repositories after you do. From memory it will append the new servers 2014-07-07 14:36:27 yep 2014-07-07 14:36:28 thanks 2014-07-07 14:36:32 I hate vi so always use nano for manual changes to files 2014-07-07 14:36:54 apk add nano && nano /etc/apk/repositories 2014-07-07 14:37:15 I just install vim ;) 2014-07-07 14:37:48 If I understand corectly, I can add my $REPODEST to this file ? 2014-07-07 14:38:03 so that every package I create will be available to "apk add" 2014-07-07 14:38:19 (Cause I'm tired of providing full pathes..) 2014-07-07 14:38:31 ah, I've never done that I'm afraid :( only ever done long pathname 2014-07-07 14:38:33 sorry 2014-07-07 14:39:03 ah, it needs an APKINDEX.tar.gz 2014-07-07 14:39:11 no idea what this file is :/ 2014-07-07 14:40:25 easiest way is to take a look, e.g. nl.alpinelinux.org/alpine/v3.0/main/x86_64/APKINDEX.tar.gz 2014-07-07 14:41:24 this might help: http://wiki.alpinelinux.org/wiki/Apkindex_format 2014-07-07 14:41:34 ah, thanks 2014-07-07 14:41:58 I guess this too: http://wiki.alpinelinux.org/wiki/Abuild 2014-07-07 14:42:13 and maybe eveything in here lol http://wiki.alpinelinux.org/wiki/Developer_Documentation 2014-07-07 14:42:21 I'm really no Dev :S 2014-07-07 14:43:13 thanks anyway ;) 2014-07-07 14:43:17 I'll sort things out 2014-07-07 14:43:27 good luck! 2014-07-07 14:46:29 I have sooo many things to setup right now ^^ 2014-07-07 14:46:38 But I'm already liking alpine 2014-07-07 14:46:57 Alpine is awesome! :D 2014-07-07 14:47:23 it's a good project 2014-07-07 14:48:04 It's the first distro I'm aware of that use busybox and musl together within an easy to use distro 2014-07-07 14:48:15 (and full-featured) 2014-07-07 17:43:50 hi. anyone using quagga/zebra on alpinelinux? I can't configure ipv6 with it and it throws only one error in my log (netlink-cmd error: Invalid argument, type=RTM_NEWROUTE(24), seq=7, pid=0). 2014-07-07 19:43:08 pluesch_: I saw an YouTube video on it, seen that? I'm not using it, or even ipv6 consciously yet 2014-07-07 20:15:48 Is that because of a misconiguration, or is tinydns REALLY slow to answer queries ? 2014-07-07 20:24:55 Aspi: i saw that video. but ipv6 static address configuration isn't demostrated in it. 2014-07-07 20:25:06 need to find a way to debug this... -.- 2014-07-07 20:26:11 idk how you do that, sorry. tcpdump and wireshark are net log tools, I'm proficient in none of them 2014-07-07 21:56:09 I have a problem with ash 2014-07-07 21:56:19 /bin/sh: /usr/bin/markdown: not found 2014-07-07 21:56:21 ─ g ── which markdown 2014-07-07 21:56:23 /usr/bin/markdown 2014-07-07 22:27:18 head -1 /usr/bin/markdown 2014-07-07 22:27:50 perl shebang 2014-07-07 22:28:23 Oooh... 2014-07-07 22:28:30 perl is not installed oO 2014-07-07 22:28:54 thanks xD 2014-07-07 22:42:29 :) 2014-07-07 22:42:46 ENOENT for exec operations is confusing 2014-07-07 22:42:56 lots of times it pertains to the interpreter 2014-07-07 22:43:02 #! style or PT_INTERP 2014-07-07 22:43:26 e.g. the same can happen if you try to run a glibc or uclibc binary on alpine :) 2014-07-08 06:21:03 Hi 2014-07-08 06:21:22 Why would a system mount itself as read-only ? 2014-07-08 06:23:24 disklees install (tmpfs) ? 2014-07-08 06:23:43 nope, it's a sys install 2014-07-08 06:24:00 And it worked great until yesterday 2014-07-08 06:24:37 and mount -o remount,rw / does not work 2014-07-08 06:26:30 z3bra, some filesystems remount automatically 'ro' if there's bad damange on disk or corruption on filesystem 2014-07-08 06:26:43 check 'dmesg' 2014-07-08 06:26:49 and logs 2014-07-08 06:28:10 http://sprunge.us/gQhW 2014-07-08 06:28:14 you were right 2014-07-08 06:32:03 It might be related to the latest kernel update 2014-07-08 06:34:07 I'm gonna have hard times... 2014-07-08 06:34:24 I'll need to downgrade the kernel, on a read-only FS :/ 2014-07-08 06:45:37 Morning everyone 2014-07-08 06:46:39 morning 2014-07-08 06:46:54 Good morning... 2014-07-08 06:47:40 well... rebooting the serv fixed it 2014-07-08 07:04:59 Is the perl package working for any of you ? 2014-07-08 07:05:13 z3bra: which fs are you using? 2014-07-08 07:05:19 ext4 2014-07-08 07:05:41 i think thats pretty stable, never had fs issues before on alpine. 2014-07-08 07:05:51 im using perl 2014-07-08 07:06:07 yeah that's weird... rebooting the server ent fine though 2014-07-08 07:06:08 anyways. 2014-07-08 07:06:47 when I start `perl`, it just prints out ~20 blank lines and return "127" 2014-07-08 07:07:13 what repo? 2014-07-08 07:07:40 http://mirror.bpiotrowski.pl/alpine/v3.0/main 2014-07-08 07:09:18 http://sprunge.us/WiiU 2014-07-08 07:09:24 but this is on edge. 2014-07-08 07:09:39 are you sure your hd isnt damaged? 2014-07-08 07:09:46 yeah 2014-07-08 07:10:08 better to run fsck and apk fix 2014-07-08 07:10:15 need fsck? "touch /forcefsck && reboot" 2014-07-08 07:10:17 I can't atm 2014-07-08 07:10:34 (connected through ssh) 2014-07-08 07:12:22 is it on a vm? 2014-07-08 07:12:44 nope 2014-07-08 07:12:54 try apk fix perl 2014-07-08 07:12:55 nevermind, reinstalling it fixed it 2014-07-08 07:13:19 then i wouldnt trust that isntall anymore 2014-07-08 07:13:30 and probably also your disks 2014-07-08 07:13:58 weird... that's a computer someone gave me. It worked great before :/ 2014-07-08 07:14:20 yeah thats why he gave it to you 2014-07-08 07:14:29 ;-) 2014-07-08 07:14:40 that's my grand mother, she don't need a computer anymore :P 2014-07-08 07:16:18 anyway, I'll fsck the partitions when I have physical access 2014-07-08 07:16:40 touch /forcefsck then ? 2014-07-08 07:16:58 that just creates a file 2014-07-08 07:17:06 the fsck won't happen until the next reboot 2014-07-08 07:17:16 yeah I get it 2014-07-08 07:17:20 cool :) 2014-07-08 07:17:33 probs worth doing anyway, just to be on the safe side 2014-07-08 07:17:33 So it will run fsck on what ? 2014-07-08 07:17:42 everything in fstab ? 2014-07-08 07:17:47 all partitions which are to be mounted at boot 2014-07-08 07:18:02 based on fstab then 2014-07-08 07:18:08 aye 2014-07-08 07:18:35 I hope my install's not fucked up :/ 2014-07-08 07:19:27 yeah, never a good thing! 2014-07-08 07:19:54 I was proud to bring my server back up in two days... 2014-07-08 07:19:55 meh 2014-07-08 07:20:28 ginjachris: are you sure thats supported on alpine? 2014-07-08 07:21:42 ok i see its added in /etc/init.d/fsck 2014-07-08 07:26:59 ncopa: does busybox fsck actually support ext4? 2014-07-08 07:28:05 no 2014-07-08 07:28:10 well 2014-07-08 07:28:23 i think fsck is just a frontend to fsck.* 2014-07-08 07:28:34 ok i found a ref 2014-07-08 07:28:36 so it will exec fsck. 2014-07-08 07:28:41 you need e2fsprogs installed 2014-07-08 07:28:45 correct 2014-07-08 07:29:12 z3bra: make sure you have that installed before you reboot. 2014-07-08 07:29:58 ncopa: i asked, because it looks like it does have ext[23] support (if its enabled) 2014-07-08 07:30:22 I *think* e2fsprogs is installed by default 2014-07-08 07:30:31 yep 2014-07-08 07:30:40 at least, I never explicitly installed it and my install has it 2014-07-08 07:30:51 but yeah, defo a good idea to check! 2014-07-08 08:01:21 Erf, alpine enforce a notation in packages version ? 2014-07-08 08:01:29 eg: I can't use 0.99.beta19 2014-07-08 08:05:49 here's a quick one: /etc/fstab on new install to disk, using default parameters. boot mountpoint is "//boot" not "/boot". Still seems to work though...... 2014-07-08 08:08:33 and of course still works if I change it to "/boot" 2014-07-08 08:16:59 ginjachris_: i think e2fsprogs is a leftover from setup-alpine (setup-disk to be precise) 2014-07-08 08:18:13 just received my apu's :) http://tinypic.com/r/2qup2tz/8 2014-07-08 08:24:39 ah yes, I (almost) always use setup-disk/setup-alpine 2014-07-08 08:24:44 love the scripts 2014-07-08 08:25:12 I can't get to tinypic to take a look at the apu's :( 2014-07-08 08:25:47 nice clandmeter 2014-07-08 08:40:07 clandmeter welcome to club :D 2014-07-08 08:50:04 now i need to score some sd cards :) 2014-07-08 08:50:18 didnt expect them to arrive so soon 2014-07-08 08:51:29 can you take the sd card out of your mobile phone? ;) 2014-07-08 08:54:33 try it with nexus 5 ;-) 2014-07-08 08:54:43 wow sd cards are cheaper than I remember, unless you start looking at 128GB http://www.ebuyer.com/store/Storage/cat/Memory-Cards/subcat/SD-----SDHC-Cards/?limit=30&page=1 2014-07-08 08:55:23 i wanted to get 2 from amazon, but now ill just get 2 cheap ones from local shop 2014-07-08 08:56:03 nexus 5! I'm still with the Samsung Galaxy S3 :S 2014-07-08 08:56:18 "old faithful" :) 2014-07-08 09:07:08 :) i have my "old" s3 in the closet. samsung software makes me cry 2014-07-08 09:08:36 awwww :( 2014-07-08 09:08:53 I always like Samsung stuff. Good build quality 2014-07-08 09:09:04 my wife used my "old" N4 after using the S3, she doesnt want to go back to the samsung *lagship* 2014-07-08 09:09:11 lol 2014-07-08 09:09:41 I've never had a problem with it 2014-07-08 09:09:59 I was very disappointed my gf got an Apple phone (shudder) 2014-07-08 09:10:18 you got an US s3? 2014-07-08 09:10:24 but, each to their own I guess 2014-07-08 09:10:36 oh I'm in UK, so I guess not 2014-07-08 09:10:48 soc is different 2014-07-08 09:11:47 EU has samsung exynos 2014-07-08 09:12:25 that sounds like some kind of super strength glue! 2014-07-08 09:12:49 it is, it glues the bits to the soc and wont let loose :D 2014-07-08 09:13:33 lol 2014-07-08 09:13:42 but i guess if you never used a nexus, you wont know the speed difference. (im spoiled) 2014-07-08 09:14:36 so long as my phone can make calls and send/rec texts I'm usually happy. Web browsing and gaming is a bonus really.....I have a pc for that sort of thing! :D 2014-07-08 09:14:43 (I know, I'm weird) 2014-07-08 09:15:15 you actually use it what its made for :) 2014-07-08 09:15:24 hahaha yeah 2014-07-08 09:31:16 hmm 2014-07-08 09:31:19 https://sandstorm.io/ 2014-07-08 09:31:22 looks nice 2014-07-08 10:53:35 clandmeter: have you bought them directly from pcengines or via a reseller? 2014-07-08 10:54:22 and btw: congrats :3 the apu's are very nice. 2014-07-08 11:03:35 proud owner of an apu4c to :) 2014-07-08 11:15:43 pluesch_: directly 2014-07-08 11:57:45 Can someone help me get my head around tinydns / dnscache ? 2014-07-08 11:57:50 I'm a bit confused.. 2014-07-08 11:59:09 I used to run a BIND server on my home server. It was authoritative for my local network (domain.lan), and recursive for everything else (forwarding to root servers) 2014-07-08 11:59:40 I'm actually trying to reproduce the same behavior using tinydns and dnscache. How Am I supposed to do this ? 2014-07-08 12:00:53 Actually, tinyDNS is listening on "0.0.0.0", and dnscache listens on 127.0.0.1 and answer requests on 0.0.0.0 2014-07-08 12:01:36 I then use the local ip of this box in the resolv.conf of all machines under this network 2014-07-08 12:01:46 and it's slow as hell... 2014-07-08 12:33:13 Hey, guys... I'm trying to compile som stuff (libreswan) and it seems /usr/include/glob.h is missing some stuff (GLOB_BRACE, GLOB_NOMAGIC) - should these be somewhere else? 2014-07-08 12:33:52 i think those are gnu extensions not implemented in musl 2014-07-08 12:34:06 you could see what we do with strongswant alternatively 2014-07-08 12:34:12 *strongswan 2014-07-08 12:36:47 danci1973, I had the same problem while compiling fdm 2014-07-08 12:36:58 it's not implemented in musl 2014-07-08 12:37:03 as said ncopa 2014-07-08 12:37:28 ncopa: Will have a look, thanks. 2014-07-08 12:37:46 damn, I'd never though that boot fast could be a problem... 2014-07-08 12:38:10 My /home partition is located on an external drive 2014-07-08 12:38:17 But alpine boots too fast 2014-07-08 12:38:27 It does not have the time to "boot" 2014-07-08 12:38:51 so when alpine mount partitions, my /home can't be mounted 2014-07-08 12:39:09 sounds like a bug 2014-07-08 12:39:21 is it on the usb bus? 2014-07-08 12:39:49 idk 2014-07-08 12:39:52 ok 2014-07-08 12:39:59 i need to reboot 2014-07-08 12:40:02 new kernel 2014-07-08 12:40:03 brb 2014-07-08 12:40:06 I'd have to check better next time 2014-07-08 12:40:16 but mount -a mounts /home without much problem 2014-07-08 12:51:24 ncopa: What do you mean by ' what we do with strongswant alternatively' ? Do you mean I should check how strongswan is compiled / packaged? 2014-07-08 12:56:56 ncopa: are you settling for 3.14 kernel? as it's LTSed 2014-07-08 12:57:52 danci1973: maybe he means strongswan is already packaged for alpine, you could try and see if you can achieve your goals using that alternative 2014-07-08 13:00:32 Aspi: He probably did mean that - but I'd like to use KLIPS with it's 'ispec' interface model, which is provided either by Openswan or Libreswan ... So far I could get neither of those to compile on Alpine 3.0 . 2014-07-08 13:04:38 then you need to replace the dependency on glibc with some musl compatible feature. patch it. http://www.musl-libc.org/faq.html some docs 2014-07-08 13:04:38 danci1973: i meant you can look how we built strongswan 2014-07-08 13:05:04 might be we have fixed the issue there already 2014-07-08 13:05:30 and you could ofcourse consider using strongswan 2014-07-08 13:07:20 ncopa: The only 'special' thing in strongswan's APKBUILD file is a '-D_GNU_SOURCE' flag and a bunch of 'configure' options ... 2014-07-08 13:07:28 ok 2014-07-08 13:07:40 i think they might have fixed the musl related things upstream then 2014-07-08 13:07:42 I considered strongswan many times, but I keep coming back to KLIPS and it's convenient 'ipsecX' interface. :D 2014-07-08 13:07:51 ok good 2014-07-08 13:08:58 z3bra: browser crashed so only saw part of conversation; I've not used dnscache/tinydns but unbound is good (and fast) as a cache/recursive lookup 2014-07-08 13:09:10 and maybe try NSD as authoritive server? 2014-07-08 13:10:08 have nsd listen on loopback address, unbound forward (normally as a stub zone) authorive (local.com or whatever) to nsd, everything else to public dns 2014-07-08 13:13:08 It can be done with tinydns + dnscache if I understand things correctly 2014-07-08 13:13:19 I'll have to try it later 2014-07-08 13:13:23 yes certainly, I just have no experience of using them 2014-07-08 13:13:36 having tinydns listen on loopback, and dnscache handling forwarding 2014-07-08 13:13:39 I can definitely vouch for unbound being great 2014-07-08 13:13:52 But I'll keep nsd / unbound in mind, just in case ;) 2014-07-08 13:13:53 thanks 2014-07-08 13:14:06 cool :) 2014-07-08 13:14:35 anyway, I'm struggling with local repositories now :P 2014-07-08 13:14:41 I doubt it will affect you (for a home network) but from memory, tinydns has a built in 200 query per second limit? 2014-07-08 13:14:47 (so many things to do upon new install ^^) 2014-07-08 13:15:25 ginjachris, yeah I know, but that's not a problem at all on myhome network 2014-07-08 13:15:35 (< 10 machines :P) 2014-07-08 13:16:08 yeah, if you hit 200 qps I'd look for something dodgy on the network! :D 2014-07-08 13:16:23 exactly ^^ 2014-07-08 13:17:05 Speaking about package building, what is "$repo" ? 2014-07-08 13:17:24 I found this 2014-07-08 13:17:28 where $repo is the name of the parent directory of $startdir. 2014-07-08 13:17:37 but I have no clue what startdir is... :/ 2014-07-08 13:18:04 is it the top dir for packages buildings ? 2014-07-08 13:21:04 I would assume so...but one of the Devs can hopefully confirm 2014-07-08 13:55:59 this is going to be interesting 2014-07-08 13:56:01 http://www.phoronix.com/scan.php?page=news_item&px=MTczNjI 2014-07-08 13:56:14 basically, it will not be possible to run udev without systemd 2014-07-08 13:56:25 (almost is that way already) 2014-07-08 13:57:09 I thought about alpine when I read that article 2014-07-08 13:57:32 no udev means no hotplug of keyboards/mouse in xorg 2014-07-08 13:57:38 libudev is used 2014-07-08 13:57:57 we already use an overdue outdated udev version 2014-07-08 13:58:34 surely, the gentoo user group is too large to ignore this; so accepting systemd or forking/maintaining patches for udev are the ways to survive 2014-07-08 13:58:37 i thought we used mdev by default? 2014-07-08 13:59:46 we use mdev by default 2014-07-08 13:59:53 but i used udev on desktop systems 2014-07-08 13:59:57 mdev is non-optimal 2014-07-08 14:00:01 i have no problem with systemd, but openRC is way cleaner 2014-07-08 14:00:18 ah, so it's more about speed? 2014-07-08 14:00:35 no 2014-07-08 14:00:47 its about hotplug keyboards mouses in xorg 2014-07-08 14:00:55 you need libudev for that 2014-07-08 14:03:22 so, I've never attended systemd flaming. I must say, even though the whole project is scary and non-unix, it has played well where I've used it, and it's both manageable and fast for me as a user (dunno how it is to work with as a pkg manager) 2014-07-08 14:04:48 I kind of agree with them, that as long as the system is booted, it does not matter how it's done. it's not done in the traditional script way, but that yields only advantages except very clear readability. it's perfectly usable if you get into the system though 2014-07-08 14:06:43 this reminds me of the story of compilers: people rejected the first programming languages and compilers like the plague in the beginning. who is so mad that they'll let some /system/, with human errors and autoshizz, translate in-best-case perfect code to our precious 1s and 0s? no, let's write them all out like we've always done, that way we can maintain 100% control 2014-07-08 14:06:47 How can I force reinstallation of a package ? 2014-07-08 14:06:57 z3bra: apk fix 2014-07-08 14:07:26 does not work 2014-07-08 14:07:34 I'll detail a bit 2014-07-08 14:07:34 whats the error message? 2014-07-08 14:07:45 I'm installing umurmur 2014-07-08 14:07:51 try "apk del --purge somepackage" 2014-07-08 14:07:59 but the initd file is broken 2014-07-08 14:08:13 Aspi: what scares me most is the narsissistic management style 2014-07-08 14:08:19 So I updated the APKBUILD and rebuilt the package 2014-07-08 14:08:22 theyre trying to take away the options 2014-07-08 14:08:36 now I'm trying to replace the current package with the new one 2014-07-08 14:08:47 apk upgrade -U -a 2014-07-08 14:08:54 but that will reinstall all rebuilds 2014-07-08 14:09:14 erf 2014-07-08 14:09:21 I can't reinstall a single package ? 2014-07-08 14:09:30 apk fix is supposed to do that 2014-07-08 14:09:41 I'm all for OpenRC if it's working (parallellization), the fact that it's multi-platform is a requierement in my ideology world, but I'm not as scared of systemd 2014-07-08 14:09:58 but i think it will not consider a recompile as 'reinstall' 2014-07-08 14:10:01 its an upgrade 2014-07-08 14:10:08 management style is a good point, ncpåa 2014-07-08 14:10:13 ncopa* 2014-07-08 14:10:16 damn what is up with my connection today 2014-07-08 14:10:19 ncopa, yeah sure 2014-07-08 14:10:31 but I'm trying to fix the packaged 2014-07-08 14:10:34 package 2014-07-08 14:10:41 I already changed pkgrel 2014-07-08 14:11:02 But I forgot to change something in the initd 2014-07-08 14:11:21 Am I supposed to increment pkgrel again ? 2014-07-08 14:15:28 it's all much fun poking around with the dhcp client, time sync service, disk mounter, et cetera, but to be completely honest, if taking a step back and watching my computer as a system -- you know -- I don't /really/ care about them other than my system running well, and if I'm in a perfectionist mood, that my system is running the best possible system software 2014-07-08 14:16:08 then I can focus on /using/ the system yet again 2014-07-08 14:16:49 nuff' idea ranting for today :) 2014-07-08 14:29:50 ncopa: http://www.reddit.com/r/linux/comments/2a2tz5/when_kdbus_will_become_merged_in_kernel_using/cir13uk ? 2014-07-08 14:30:10 no, wtf, sorry 2014-07-08 14:49:44 How can I deal with a file shipped as .tgz, but is in fact a .tar ? 2014-07-08 14:50:20 tar xvf whatever.tgz? 2014-07-08 14:50:32 in an APKBUILD* 2014-07-08 14:50:40 you cant 2014-07-08 14:50:58 its not supposed to fix stupidity 2014-07-08 14:51:07 sure 2014-07-08 14:51:21 but maybe I could override some untar() function, or whatever 2014-07-08 14:51:34 why not repack the archive and upload it soemwhere? 2014-07-08 14:51:58 Yeah that's an other solution 2014-07-08 14:52:10 I was just wondering if there was a way to deal with it 2014-07-08 14:52:37 it makes no sence why it should 2014-07-08 14:53:03 clandmeter: hey that apu is a nice piece of serial port-having hardware! 2014-07-08 14:53:20 shame there's no intel i[357] version 2014-07-08 14:54:36 ovf: itx with serial is no option? 2014-07-08 14:55:51 clandmeter, the function "unpack()" can be overridden :) 2014-07-08 14:56:34 clandmeter: even that is hard to come by 2014-07-08 14:56:46 ovf: i think we have some 2014-07-08 14:57:02 z3bra: its a shell script. you can hack it whatsoever 2014-07-08 14:57:09 you can even print a xmas tree if you like 2014-07-08 14:57:13 yeah sure 2014-07-08 14:57:26 Just noticed the wiki states it "officially" 2014-07-08 14:58:11 you should fix the problem, not find a way around it. 2014-07-08 14:59:15 ovf: http://www.gigabyte.eu/products/product-page.aspx?pid=4918#ov 2014-07-08 14:59:18 2x serial 2014-07-08 14:59:20 cpu on board 2014-07-08 14:59:27 yeah, but first I'll see if I can build the package, and such 2014-07-08 14:59:35 then, I'll contact them to report the error 2014-07-08 15:01:46 have there already been self-modifying APKBUILDs? 2014-07-08 15:02:36 say you could redefine package() in prepare() or something 2014-07-08 15:03:41 clandmeter: thanks, doesn't look bad at all. 2014-07-08 15:03:43 i think you can even use whateverfil.tar.gz::http://site.com/whatever.zip 2014-07-08 15:04:40 ovf: gigabyte mostly targets industrial with their on board cpu boards 2014-07-08 15:12:47 z3bra, try file utils for file type guessing 2014-07-08 15:14:12 I don't need it to de dynamicaly checked 2014-07-08 15:14:49 now that I know the archive is a .tar 2014-07-08 15:14:51 :) 2014-07-08 15:15:58 :) 2014-07-08 15:25:24 clandmeter: do you guys offer colocation? :-) you know how nokia had all their symbian devices on the premises and you could connect to them and run your code over the web. 2014-07-08 15:30:48 cgroups will soon be pretty difficult without systemd too 2014-07-08 15:31:40 I don't really mind systemd, but it's still kind of annoying that linux is so tightly tied to gcc and soon systemd 2014-07-08 15:32:04 if only Minix were progressing more rapidly... 2014-07-08 15:32:35 it's freaking annoying linux is tied to the cpu arch -.- 2014-07-08 15:32:53 speaking of systemd, i have a somewhat funnily named unit file that when i do systemctl disable doesn't print anything. likely i'm escaping the funny chars wrong. any way to see what's happening? 2014-07-08 15:33:01 tanenbaum all the way 2014-07-08 15:33:25 Aspi: a machine-independent operating system wouldn't make much sense 2014-07-08 15:33:40 ^? 2014-07-08 15:34:17 ovf: ls /etc/systemd/system/multi-user.target.wants 2014-07-08 15:34:24 how do you see an operating system not "tied to the cpu arch"? 2014-07-08 15:34:26 if it's not there, you already have it disabled :) 2014-07-08 15:35:00 I see a kernel abstracting the hardware to the operating system 2014-07-08 15:35:41 Aspi: perhaps you'd prefer NetBSD? It's fairly well abstracted from the hw architecture internally 2014-07-08 15:35:56 thanks, looks like indeed i have it disabled 2014-07-08 15:36:07 i'd like some more informative output in such cases though 2014-07-08 15:36:17 so that when arm128 appears, it's only a matter of developing a compatible lowest level kernel 2014-07-08 15:36:33 linux is literally the lowest level kernel 2014-07-08 15:36:52 as it runs on its own, save for some runtime hooks 2014-07-08 15:37:12 it's huge, though. 2014-07-08 15:37:38 and not particularly well abstracted internally, presumably for performance but I'm guessing also out of a certain level of laziness. 2014-07-08 15:38:03 Aspi: seriously, though, you should check out NetBSD. It is extremely easy to port to new architectures. 2014-07-08 15:38:12 most of the bloat are device drivers, and they're necessarily tied to the hardware they service 2014-07-08 15:38:19 how does NetBSD relate to OpenBSD and FreeBSD really, never touched them 2014-07-08 15:38:24 It is also the only other kernel to support Xen's dom0 2014-07-08 15:38:50 Aspi: it's another descendant of BSD, probably the most closely descended of any of them 2014-07-08 15:38:54 oh, and a compiled application should never be tied to an architecture 2014-07-08 15:39:11 It's the third most widely used BSD if you don't count OS X (and you shouldn't) 2014-07-08 15:39:15 after FreeBSD and OpenBSD 2014-07-08 15:39:22 Aspi: this one's easy 2014-07-08 15:39:27 we have computers that can take the little overhead of translating on the fly. JavaScript performance anyone? and how many times has that been translated? 2014-07-08 15:39:33 compile with cat, run with make && ./a.out 2014-07-08 15:40:11 waitwut, compile with cat? :P 2014-07-08 15:40:13 Aspi: that's doable using something like PNaCl, which uses portable LLVM bitcode 2014-07-08 15:40:24 Psalt? :P 2014-07-08 15:40:35 Aspi: it's pretty hard to make a JIT work in such a system, though 2014-07-08 15:40:50 it's fine if you don't care about the performance of dynamic languages 2014-07-08 15:41:28 ohwell, I do. just 90% performance is fine with me, if it's all portable 2014-07-08 15:41:47 Aspi: what's wrong with that? fits your bill 2014-07-08 15:41:49 Aspi: you can use qemu to dynamically translate any architecture to any other architecture, you know 2014-07-08 15:41:54 Aspi: even transparently to you 2014-07-08 15:42:12 it will even translate the syscall arguments 2014-07-08 15:42:28 hehe :) still, you know what I mean. something has to run qemu 2014-07-08 15:42:44 give me qemuOS! 2014-07-08 15:43:09 I still don't understand your cat-compiler 2014-07-08 15:43:16 ovf: ^ 2014-07-08 15:43:32 Given how hard the CPU manufacturers are competing on performance, I don't think you're going to see OSes switching to CPU-independent binaries anytime soon 2014-07-08 15:43:52 The closest thing is the Java universe. And I don't think you want to live in that universe. 2014-07-08 15:44:08 What happens when you come up with a portable format is people start trying to make CPUs that directly support that format 2014-07-08 15:44:09 I - do - not! 2014-07-08 15:44:15 Aspi: basically just carry your program in source form (and assume it's compiled), and compile directly prior to execution. how's that for not tied to an architecture? 2014-07-08 15:44:16 java is ugly 2014-07-08 15:44:20 And then everything goes to hell. 2014-07-08 15:44:24 This is the whole point of a compiler 2014-07-08 15:44:42 Anyway, soon ARM will die and all that will be left is x86-64. 2014-07-08 15:44:49 arm64? 2014-07-08 15:44:56 And then you'll have your portable binary format. 2014-07-08 15:45:00 http://www.phoronix.com/scan.php?page=news_item&px=MTcyNjU 2014-07-08 15:45:17 Hah. Russia. 2014-07-08 15:45:24 gotta love em 2014-07-08 15:45:25 They're not even part of the G8 anymore. 2014-07-08 15:45:25 arm64 is a slower x86_64 2014-07-08 15:45:28 so hipsters 2014-07-08 15:45:30 So who gives a shit about what they do? 2014-07-08 15:45:49 It's just a dig at the US. 2014-07-08 15:46:14 I see GNU entering the stage with a new kernel some time 2014-07-08 15:46:18 Don't get me wrong; I like the other architectures, Intel is just too awesome for anyone to compete with. 2014-07-08 15:46:26 ditto 2014-07-08 15:46:28 Aspi: err what? They already have one: Hurd. 2014-07-08 15:46:45 absolutely. it's not functional though 2014-07-08 15:46:46 There were even attempts to port it to a modern microkernel. 2014-07-08 15:47:07 And maybe now that seL4 is getting open sourced that work will get some new life. 2014-07-08 15:47:10 I'm in part referring to the Hurd. but it must be worked out to enter the stage. 2014-07-08 15:47:24 I doubt it. It seems more likely to me that everything will just switch to Linux. 2014-07-08 15:47:37 And anyway, GNU has kind of made itself irrelevant by being so hostile to commerce. 2014-07-08 15:48:00 There's a reason MacOS is not Linux. 2014-07-08 15:48:13 And why Apple dumped GCC for LLVM 2014-07-08 15:48:18 everything switching to Linux is like everything switching to systemd, for you init folks 2014-07-08 15:48:36 And there will be similar kicking and screaming and then everyone will do it anyway. 2014-07-08 15:48:44 :) 2014-07-08 15:48:49 Actually, I just convinced myself tha teverything WON'T switch to Linux. there's still the matter of Linux's being non-free. 2014-07-08 15:49:00 (which I say only to troll GPL fanatics) 2014-07-08 15:49:11 The GPL is a major problem. 2014-07-08 15:49:29 but I don't like how Linux is gluing the holes instead of doing it right: doing drivers in kernel space instead of creating proper comm and letting them be independent 2014-07-08 15:49:39 well, sometimes worse is better 2014-07-08 15:49:41 Usually, in fact. 2014-07-08 15:49:47 pragmatic 2014-07-08 15:49:57 do the simplest thing that could possibly work, and then stop when it works well enough 2014-07-08 15:50:01 and go on to the next thing 2014-07-08 15:50:12 you know, I see a time where energy is free, so all labour is done by machines, so everybody worships GPL :_ 2014-07-08 15:50:45 until then, Linux is a step in the right direction 2014-07-08 15:50:56 well for some value of free 2014-07-08 15:51:03 I don't see where worshipping GPL follows from that though 2014-07-08 15:51:10 I wonder how many times Linus have regreted GPLing Linux instead of BSDing it 2014-07-08 15:51:27 One would think he would have mentioned something by now if he regretted it 2014-07-08 15:51:50 there are those who claim that the BSD license is the reason for the fragmentation of the BSD space 2014-07-08 15:51:52 he'd kill the project 2014-07-08 15:52:08 i don't see GNU introducing a new kernel. they already tried once and they're still trying 30 years later. 2014-07-08 15:52:11 You think that many people on the project are that fanatical about the GPL? 2014-07-08 15:52:19 they really don't have the technical expertise to do so 2014-07-08 15:52:21 They'd just make a GPL fork if they cared that much 2014-07-08 15:52:34 on Linux project? 2014-07-08 15:52:44 no, hurd :) 2014-07-08 15:53:01 last i checked it boots and runs ok with 1990s-era hardware... 2014-07-08 15:53:02 any open source project needs to have a need driving it 2014-07-08 15:53:06 and the need just isn't there with Hurd 2014-07-08 15:53:19 there is the philosophical. like OpenRC 2014-07-08 15:53:47 and what would the Hurd philosophy be? 2014-07-08 15:53:54 oh, and there can appear folks revolutionizing the field. from GNU. take Guix: https://www.gnu.org/software/guix/ 2014-07-08 15:54:02 the major reasons for having a microkernel are reliability and security through isolation 2014-07-08 15:54:20 freakazoid: that is the problem! thank you. there is no philosophy. only that it must be perfect 2014-07-08 15:54:40 guix is just a clone of NixOS 2014-07-08 15:54:48 it has no point in existing. 2014-07-08 15:54:54 it's GNU :) 2014-07-08 15:55:03 no, but it makes it more standard 2014-07-08 15:55:05 right, and that isn't good enough. 2014-07-08 15:55:14 nobody gives a shit about guile 2014-07-08 15:55:29 is that true? I've never used it. 2014-07-08 15:55:35 in fact there are people who simply won't use scheme because it has too little syntax for human consumption 2014-07-08 15:55:43 just the GNU trying to make it omniscient? 2014-07-08 15:56:10 GNU has NIH syndrome big time 2014-07-08 15:56:19 yes. very 2014-07-08 15:56:19 and it's part of the reason they've ceased to be relevant 2014-07-08 15:56:39 plus a lot of people simply won't use their stuff because of its restrictive licenses 2014-07-08 15:56:57 even people who don't care about selling their code 2014-07-08 15:57:12 just because they don't want their license dictated to them 2014-07-08 15:57:37 oh. ohwell, we choose ourselves 2014-07-08 15:57:52 when I see GPL, I see green light 2014-07-08 15:58:04 nih makes sense only when (1) there's an actual problem with the existing solution that was n.i.h. and (2) your replacement actually fixes that problem. 2014-07-08 15:58:27 dalias: :) 2014-07-08 15:58:58 init is a good example, linux/rhel sysv init was much worse than the ones on most unixes. they make a shitty clone. they make another shitty clone... and i think now there's 15 or so 2014-07-08 15:59:08 in the case of hurd, their perveived problem with the existing solution was simply the lack of gnu branding, and they failed to fix any of the actual problems with linux 2014-07-08 15:59:17 it makes sense when the existing community is not interested in your use case 2014-07-08 15:59:30 perceived* 2014-07-08 15:59:47 hurd is technically inferrior to linux in almost every way 2014-07-08 16:00:03 especially their poor use of microkernel design makes it less reliable, not more 2014-07-08 16:00:12 the end result has been that now there are multiple non-GNU userspaces to choose from that are just as good if not better 2014-07-08 16:00:27 because you have more pieces that can fail non-atomically with respect to each other, and no way to handle the failure 2014-07-08 16:00:28 FreeBSD doesn't even need GCC anymore 2014-07-08 16:01:25 should never need it. like systemd should not need linux. 2014-07-08 16:01:50 you know, maths is at the bottom of all this. 2014-07-08 16:01:52 since they don't have PATH_MAX, i suspect you can load the fs down to being unusable just by constantly accessing billions-deep directory trees via absolute pathnames... 2014-07-08 16:01:55 microkernel servers need to be written by people who understand the Erlang philosophy 2014-07-08 16:02:29 where anything can crash at any time and come back into a known state 2014-07-08 16:03:02 or just write everything except the core in a language that lacks crashes :) 2014-07-08 16:03:24 of course there are no such languages, so you'd first have to invent one 2014-07-08 16:06:23 sometimes crashing is the right thing to do, becuase you will always have bugs 2014-07-08 16:06:32 and that can cause you to get into states that you shouldn't try to recover from 2014-07-08 16:07:21 the main reason some programs try to recover is that they're monolithic beasts where the only type of crash available is the entire program and you lose a bunch of state in the process that's still known good 2014-07-08 16:08:37 you even want a filesystem that doesn't require a single server - it should be able to have a server per open file 2014-07-08 16:09:02 of course, the normal means of doing isolation in hardware isn't particularly compatible with that many processes 2014-07-08 16:09:47 freakazoid, not really 2014-07-08 16:09:49 the other option is to put all the handling of filesystem operations into a library and have the kernel control which disk blocks you can access only 2014-07-08 16:10:01 which is the exokernel approach 2014-07-08 16:10:04 you have to think about interprocess use of files and atomicity requirements 2014-07-08 16:10:14 doing that with multiple fs servers is difficult 2014-07-08 16:10:36 you do, but serializing everything is the brute force approach to that problem 2014-07-08 16:10:55 there are well-known ways to handle it in a distributed fashion 2014-07-08 16:11:09 are you sure? 2014-07-08 16:11:13 mmap greatly complicates it 2014-07-08 16:11:35 well it's possible you need to throw POSIX out the window 2014-07-08 16:11:41 but NFS already does that :) 2014-07-08 16:11:48 that's why nfs is unusable crap 2014-07-08 16:11:51 and 9p is the right solution 2014-07-08 16:12:01 eh, only if your code relies on POSIXisms 2014-07-08 16:12:11 9p is also distributed 2014-07-08 16:12:21 but 9p lacks the problems (aka caching) of nfs 2014-07-08 16:12:25 sure 2014-07-08 16:12:45 anyway this kind of bad design is also why we have so many utter shit fuse filesystem drivers 2014-07-08 16:12:46 I would probably use 9p for this hypothetical multiprocess approach :) 2014-07-08 16:13:05 anyway, it was just an example 2014-07-08 16:13:09 fuse encourages bad implementation 2014-07-08 16:13:15 to avoid having the entire filesystem server crash 2014-07-08 16:13:29 basically a good microkernel-like fs layer would have two classes/tiers of fs driver 2014-07-08 16:14:12 the better one would not implement read/write at all, purely mmap and file metadata operations 2014-07-08 16:14:33 hmmm 2014-07-08 16:14:34 and the real kernel would do all the read/write logic via the mmaps 2014-07-08 16:14:55 it seems like hiding that much from the app is frequently not a good idea 2014-07-08 16:15:18 but then again it's not like the app knows the layout of blocks on disk anyway 2014-07-08 16:15:29 "the app"? 2014-07-08 16:15:43 this is about the fs driver and the kernel, not apps 2014-07-08 16:15:43 the end user of the API 2014-07-08 16:15:55 oh 2014-07-08 16:15:59 the end user app uses the normal posix io operations (either read/write or mmap) 2014-07-08 16:16:15 but if it uses read/write, it's not calling into the fs driver (except when the kernel needs to access new pages of the file it didn't already) 2014-07-08 16:16:23 it's just asking the kernel to memcpy to/from those pages 2014-07-08 16:16:29 but mmap is not a lower level operation than read/write 2014-07-08 16:16:37 something still needs to actually write to the disk 2014-07-08 16:16:37 logically, it is 2014-07-08 16:17:09 on a sane modern implementation that (1) has mmap, and (2) does fs caching, mmap is the logically lower-level operation 2014-07-08 16:17:15 ok, I understand what you're saying 2014-07-08 16:17:24 read and write are not disk operations 2014-07-08 16:17:29 they're fs cache operations 2014-07-08 16:17:34 *nod* 2014-07-08 16:17:36 so they're not particularly useful 2014-07-08 16:18:21 btw this is why it's stupid that glibc's stdio adapts its buffer size and alignment to the fs/disk blocksize of the underlying device.... :) 2014-07-08 16:18:32 read/write are operations on the kernel's fs cache, not on the disk 2014-07-08 16:18:35 you still want barriers of some kind with mmap, no? 2014-07-08 16:19:06 and those barriers either need to interact with battery-backed cache or the disk itself, to ensure data integrity in the event of a power failure or system crash 2014-07-08 16:19:26 yes, that kind of operation is needed if the fs implements fsync/O_SYNC/etc. 2014-07-08 16:21:30 I'm hugely excited about the musl switch, btw 2014-07-08 16:21:58 I'm guessing it means upgrading my firewall is going to be non-trivial, though 2014-07-08 16:22:14 ask him: http://templeos.org/ 2014-07-08 16:22:52 he knows it all. he's the best programmer on the planet. God said so. 2014-07-08 16:22:57 I'd prefer to avoid that guy at all costs. 2014-07-08 16:23:28 I'm listening to his music. https://youtu.be/1okW1RTPZ7Q 2014-07-08 16:26:29 ohmy, it boots in VirtualBox. he's got mouse and kbd support. 2014-07-08 16:27:04 Replace all instances of "God" with "me" or "I" 2014-07-08 16:27:29 or "he" or "she" if referring to something someone else said 2014-07-08 16:27:34 and suddenly it all makes sense 2014-07-08 16:27:44 or "some random guy in the past" 2014-07-08 16:27:47 I thought so too 2014-07-08 16:27:53 if you're referring to scripture 2014-07-08 16:28:01 since God is all in people's heads 2014-07-08 16:28:13 http://www.gamefaqs.com/boards/916373-pc/67825081?page=2 2014-07-08 16:28:29 I was referring to the God at his page 2014-07-08 16:28:43 I'm startled at the guy. fascinated 2014-07-08 16:29:25 Insanity can be fascinating. 2014-07-08 16:29:48 I have a friend I've known since he was 20 who started developing schizophrenia around 30 2014-07-08 16:30:02 his stories are fascinating 2014-07-08 16:30:30 you can sort of map how his brain works by listening to him and noting the places he suddenly starts using very imprecise language 2014-07-08 16:30:44 like "the government" or "law enforcement" 2014-07-08 16:30:51 and he can't name a name or describe a person 2014-07-08 16:31:17 schizophrenia seems like an extreme form of biases we all have 2014-07-08 16:31:23 like, he can't remember, or he can't tell 2014-07-08 16:31:27 like the delusion that we're more important than we actually are 2014-07-08 16:31:43 well they're narratives that explain things about the world 2014-07-08 16:31:58 many of them, in fact, are around why he can't hold down a job 2014-07-08 16:32:12 so it's very meta because literally he's explaining his way around his own schizophrenia 2014-07-08 16:32:39 he got a disability letter from a doctor, but he reads the word "disabled" in it as a transitive verb 2014-07-08 16:32:49 so he considers it "proof" that "people" are "disabling" him. 2014-07-08 16:33:00 but he remembers the text of the letter exactly 2014-07-08 16:33:07 he just interprets it differently than a sane person would 2014-07-08 16:33:41 it seems like schizophrenia is just an overactive self-importance nerve 2014-07-08 16:33:47 very fascinating 2014-07-08 16:34:02 or gland, or whatever you want to call the system that deludes into thinking we matter. 2014-07-08 16:34:09 deludes us* 2014-07-08 16:34:21 (maybe all of us at IRC has a bit of shizo) 2014-07-08 16:34:33 it seems like everyone is, and a little bit of asperger's 2014-07-08 16:34:40 which in a way is a little bit of autism 2014-07-08 16:34:54 I've certainly been diagnosed with ADHD, but it's not a binary condition 2014-07-08 16:34:56 so it's a positive thing? :) 2014-07-08 16:35:21 heh, ohwell 2014-07-08 16:35:34 well it ultimately depends on whether it interferes with your ability to be happy and productive 2014-07-08 16:35:54 it's not the conditions/properties that matters, it's your own ability to live your life 2014-07-08 16:36:26 ok, shower time 2014-07-08 16:36:33 certainly didn't expect this in #alpinelinux today 2014-07-08 16:36:43 thanks though 2014-07-08 16:37:13 the problem in my friend's case is that he blames all failures on external parties and takes no responsibility for his own success 2014-07-08 16:37:18 he's mostly just given up on life 2014-07-08 16:37:27 won't seek treatment 2014-07-08 16:37:41 talking about leaving the country to escape these people 2014-07-08 16:38:07 reminds me so much of Beatiful Mind 2014-07-08 16:38:11 and when it doesn't help and he's even farther away from people who care about him, he'll believe they followed him 2014-07-08 16:38:19 I have little experience with schizophrenia 2014-07-08 16:38:40 things are going to get pretty interesting when we really understand the brain and can start manipulating it precisely 2014-07-08 16:38:57 I wonder if the human ego can survive what we're going to discover about how simple we really are 2014-07-08 16:39:16 of course, most people will just ignore it 2014-07-08 16:39:22 the ego is good at protecting itself 2014-07-08 16:39:26 ok, really going now 2014-07-08 16:39:28 catch ya later 2014-07-08 16:39:41 cya :D 2014-07-08 17:19:02 hooray 2014-07-08 17:19:14 my local DNS works ! \o/ 2014-07-08 17:20:10 On question though... 2014-07-08 17:20:23 My /etc/resolv.conf contains the following lines 2014-07-08 17:20:32 search domain.lan 2014-07-08 17:20:39 domain domain.lan 2014-07-08 17:20:44 nameserver 127.0.0.1 2014-07-08 17:21:23 when I do "ping random_machine.domain.lan" it works great. But it does not when I do "ping random_machine" 2014-07-08 17:22:01 (While it works from another linux box, using the same resolv.conf, except from the nameserver) 2014-07-08 17:26:17 Ah, looks like it's musl fault 2014-07-08 17:26:27 http://www.marshut.com/ivmnkq/gethostbyname-and-resolv-conf.html 2014-07-08 17:27:22 tldr: "search" is not implemented by musl yet because they're discussing the way it should be handled (parallel/serie search, etC...) 2014-07-08 17:41:08 hey 2014-07-08 17:41:20 z3bra: same problem here 2014-07-08 17:41:47 z3bra: thank you very much for the link and info 2014-07-08 17:41:49 :) 2014-07-08 17:42:00 actually the discussion was pretty much finished, but the conclusion was that search is mostly harmful and not a priority 2014-07-08 17:42:06 no prob :) 2014-07-08 17:42:10 Ah, okay 2014-07-08 17:42:16 so no plan to add it then ? 2014-07-08 17:42:16 for lan uses like this, just putting appropriate lines in your /etc/hosts file achieves the same thing 2014-07-08 17:42:21 well, maybe 2014-07-08 17:42:37 if there's demand, it could be added. but it won't be optimized with parallel query and stuff. 2014-07-08 17:43:15 I see :) 2014-07-08 17:43:18 thansk for clarifications 2014-07-08 17:43:52 until now i hadn't seen anyone asking for it 2014-07-08 17:44:05 and when i raised the issue of whether we want it, the response was all negative :-p 2014-07-08 17:44:14 haha, I see ^^ 2014-07-08 17:44:33 it's just make things easier to ping local machines for me 2014-07-08 17:44:48 ping gavroche, instead of ping gavroche.z3bra.lan 2014-07-08 17:45:00 yeah 2014-07-08 17:45:01 btw. will alpine switch to systemd or stay with openrc? 2014-07-08 17:45:15 i was saying you could just put those in /etc/hosts tho 2014-07-08 17:45:22 unless they're dynamic-ip 2014-07-08 17:45:25 Oh, yeah forgot about that 2014-07-08 17:45:36 na those are statics 2014-07-08 17:45:49 also hosts can do link-local addresses with ipv6 2014-07-08 17:45:52 whereas dns can't 2014-07-08 17:46:10 I don't do ipv6 yet ;) 2014-07-08 17:50:31 pluesch_, if alpine wanted systemd it would need to either maintain a lot of patches or convince lennart to support musl 2014-07-08 17:54:57 i hope alpine stays with openrc ... 2014-07-08 17:55:30 I'm not fond of openRC myself 2014-07-08 17:55:37 (i'm discovering it with alpine) 2014-07-08 17:55:46 But I find it better than systemd though 2014-07-08 18:05:51 being one of the few non-systemd distros will be an asset 5 years from now when all the big names are trying to figure out how to extricate themselves from the systemd mess 2014-07-08 18:14:45 >>> ERROR: detox: Missing license in APKBULID 2014-07-08 18:14:51 There is a typo =) 2014-07-08 18:14:55 APKBULID 2014-07-08 18:18:36 :) 2014-07-09 06:35:19 My guess is that when the big names are trying to extricate themselves from the systemd mess, a lot of people will be thinking about how to extricate themselves from the Linux mess. 2014-07-09 06:38:38 morning 2014-07-09 06:46:30 morning 2014-07-09 06:46:41 pluesch_, if alpine wanted systemd it would need to either maintain a lot of patches or convince lennart to support musl 2014-07-09 06:46:54 that is one big issue 2014-07-09 06:47:40 when presenting a portability problem for systemd devs their response have been: we dont care. your libc should mimic glibc. EOD 2014-07-09 06:47:55 that obviously won't work for us 2014-07-09 06:48:53 since the entire point of using musl libc is get rid of the gnu libc broken behaviors and get rid of the bloat 2014-07-09 06:49:36 so even if we got musl libc compatible enough to run systemd, do we want systemd devs dictate how musl libc should behave? 2014-07-09 06:49:41 no. definitively not 2014-07-09 06:49:49 second issues is the dbus thingy 2014-07-09 06:50:27 we will not consider systemd with the current dbus dependency 2014-07-09 06:50:52 i am open to atleast consider it again once kdbus is in place, but not a second before that 2014-07-09 06:51:51 i really really really don't want end up with systemd 2014-07-09 06:52:05 openrc is not great, but it does the job for now 2014-07-09 06:52:40 systemd solves some of the problems with openrc 2014-07-09 06:52:52 but it also creates a set of new problems 2014-07-09 06:53:33 i was kind hoping something else would show up 2014-07-09 06:54:10 did you consider runit ? 2014-07-09 06:54:12 we might consider something combined with runit or s6-init 2014-07-09 06:54:35 runit is probably the strongest candidate atm 2014-07-09 06:55:27 What's the problem with openrc btw ? 2014-07-09 06:57:00 you cannot tell when a service is ready to use 2014-07-09 06:57:11 so if you start service foo 2014-07-09 06:57:20 and service bar depends on foo 2014-07-09 06:57:28 foo will fork and bootup continue 2014-07-09 06:57:39 so service dependency 2014-07-09 06:57:45 bar will start up but might fail due to foo is not ready 2014-07-09 06:57:46 yes 2014-07-09 06:57:53 okay 2014-07-09 06:57:56 systemd solves that 2014-07-09 06:57:59 but in wrong way 2014-07-09 06:58:03 http://ewontfix.com/15/ 2014-07-09 06:58:07 in in weird way though, but yeah 2014-07-09 06:58:54 what i think systemd does good (well, i'm on thin ice since i havent really used it) 2014-07-09 06:59:10 is declarative service descriptions 2014-07-09 06:59:14 i kinda like that 2014-07-09 06:59:30 you mean, the unit files ? 2014-07-09 06:59:38 http://en.wikipedia.org/wiki/Declarative_programming 2014-07-09 06:59:39 yup 2014-07-09 07:00:23 well, it's a bit weird to use in fact 2014-07-09 07:00:28 not sure i like .ini files but i kinda like the declarative idea 2014-07-09 07:00:42 the "Everything goes in a variable" approach kinda shocked me 2014-07-09 07:01:04 i havent actually tried it but i like the idea 2014-07-09 07:01:17 ExecStart="ip link set wlan0 up" 2014-07-09 07:01:19 what i dont like is that pid is huge 2014-07-09 07:01:27 pid 1 2014-07-09 07:01:28 ExecStart"wpa_supplicant ..." 2014-07-09 07:01:41 ExecStart="dhcpcd wlan0" 2014-07-09 07:01:53 openrc has something simliar 2014-07-09 07:01:56 command="foo" 2014-07-09 07:02:03 command_args="--blah" 2014-07-09 07:02:21 start_stop_daemon_args="--pidfile=/var/piD" 2014-07-09 07:02:36 you can run your openrc script the that 2014-07-09 07:02:38 well, yeah, but if your service can't fit in a variable, you can override the start() function at least 2014-07-09 07:02:45 correct 2014-07-09 07:03:12 also, the systemd pid 1 is not *that* huge 2014-07-09 07:03:26 lol 2014-07-09 07:03:28 It does its job as pid 1 2014-07-09 07:03:33 when i think of it 2014-07-09 07:03:40 but will spawn thousand of subprocesses 2014-07-09 07:03:48 alpine linux's pid 1 is likely bigger :) 2014-07-09 07:04:16 yeah, haha 2014-07-09 07:04:28 >800K ;) 2014-07-09 07:05:53 Oh, btw, I have a problem with my alpine install 2014-07-09 07:06:03 It boots too fast x) 2014-07-09 07:06:34 my /home partition is on an external hard drive, and it slower to boot than my computer 2014-07-09 07:06:43 if booting to fast is a problem, then thats a good indication that something is done wrong 2014-07-09 07:06:58 so mdev won't find it upon mounting everything from /etc/fstab 2014-07-09 07:07:44 (if I power on my drive BEFORE the server, everything goes fine) 2014-07-09 07:07:58 thats a problem i would like to fix 2014-07-09 07:08:07 but i dont think it is fixable with openrc 2014-07-09 07:08:20 well, i suppose its possible to hack around it 2014-07-09 07:08:23 well, I read something about hutplug 2014-07-09 07:08:26 *hotplug 2014-07-09 07:08:33 but i dont think its properly fixable with openrc 2014-07-09 07:08:34 yes 2014-07-09 07:08:35 But I'm not familiar with it at all 2014-07-09 07:08:49 i actually ahve some ideas there 2014-07-09 07:09:04 well, the "best" one would be to slow down extlinux 2014-07-09 07:09:08 (atm) 2014-07-09 07:09:25 because something like 3-4 more seconds would be enough 2014-07-09 07:28:15 you can do that 2014-07-09 07:29:02 you can edit timeout option in /etc/update-extlinux.conf 2014-07-09 07:29:18 yep 2014-07-09 07:29:37 so what i am thinking 2014-07-09 07:29:59 systemd wants to start things on hotplug event 2014-07-09 07:30:03 i like that 2014-07-09 07:30:34 what if your disk doesn't show up ? 2014-07-09 07:30:41 in my case, it's /home 2014-07-09 07:30:59 i think you shoudl be able to declare mount point dependency 2014-07-09 07:31:00 but one could use /var, or /usr 2014-07-09 07:31:12 so lets say, a service could depend on /home to be mounted 2014-07-09 07:31:25 you put your /home in fstab 2014-07-09 07:31:44 you have a boot time process that scans /etc/fstab 2014-07-09 07:31:54 then it knows what to expect to be there at boot time 2014-07-09 07:32:02 Ah, I see 2014-07-09 07:32:10 services depending on mount points 2014-07-09 07:32:24 yes 2014-07-09 07:32:32 and when you start that service, it would simply lock 2014-07-09 07:32:51 (or it would not start at all til the mount point is available) 2014-07-09 07:33:38 once all expected mounts are found, it can simply exit and leave a 'jobe dome sucessfully' mark after itself 2014-07-09 07:33:52 yep 2014-07-09 07:34:10 systemd probably handle that 2014-07-09 07:34:17 so if you later start services it will know that the needed mounts are there 2014-07-09 07:34:21 yes i think it does 2014-07-09 07:34:24 however 2014-07-09 07:34:46 it also handles the issue when you hot plug a disk afterwards 2014-07-09 07:34:54 and i think it will use libmount for that too 2014-07-09 07:35:03 which means that libmount needs to be in memory all the time 2014-07-09 07:35:10 erh 2014-07-09 07:35:11 that part i dont like 2014-07-09 07:35:12 yeah 2014-07-09 07:35:30 (virtual memory should make that swap out etc so its not big deal) 2014-07-09 07:35:39 and its not much memory 2014-07-09 07:35:42 but still 2014-07-09 07:35:46 i hate bloat :) 2014-07-09 07:36:02 so what i'd like is 2014-07-09 07:36:08 yeah ^^ 2014-07-09 07:36:17 that's why there is alpine anyway :P 2014-07-09 07:36:23 exactly 2014-07-09 07:36:57 so i'd like to have the exepected boot mounts done with libmount, libblkid etc 2014-07-09 07:37:14 but once everything to be expected (eg /etc/fstab) is done 2014-07-09 07:37:16 it exists 2014-07-09 07:37:28 if you need handling of later-on plugged disks 2014-07-09 07:37:31 basically, a "mount" service 2014-07-09 07:37:44 you'll have to fork/exec something 2014-07-09 07:37:50 yes 2014-07-09 07:38:14 i like that sysmted tries to avoid fork/exec at boot time 2014-07-09 07:38:53 now, in your /home case 2014-07-09 07:38:58 where you have external disk 2014-07-09 07:39:19 if you would start up the machine wihtout the disk plugged 2014-07-09 07:39:25 it would start up normally 2014-07-09 07:39:32 but services depending on /home would not start 2014-07-09 07:39:37 yep 2014-07-09 07:39:52 since its an 'expected' mount 2014-07-09 07:39:59 the service would not exit 2014-07-09 07:40:13 and libmount would stay in memory 2014-07-09 07:40:25 forever or til you plug the disk 2014-07-09 07:40:36 yeah 2014-07-09 07:40:44 now.. what hit me is 2014-07-09 07:40:51 what happens if disk is there at boot 2014-07-09 07:40:54 it gets mounted 2014-07-09 07:41:07 and everything runs fine 2014-07-09 07:41:13 but you unplug the disk? 2014-07-09 07:41:22 i suppose thats food for thought :) 2014-07-09 07:41:56 Mmmh.. 2014-07-09 07:42:07 I guess you don't want this to occur 2014-07-09 07:42:11 ^^ 2014-07-09 07:42:38 do we want solve that? 2014-07-09 07:42:43 i think sysmted tries to do that 2014-07-09 07:43:03 you could have an exec-on-demand service 2014-07-09 07:43:37 well, I think that if one of your server's disk get unplugged, even systemd can't help you 2014-07-09 07:43:49 your services relying on it will crash 2014-07-09 07:44:01 that fork/execs on first request, stays in memory for a while, and exits on no activity on timeout 2014-07-09 07:44:01 And you'll have to replug the disk, and restart the service 2014-07-09 07:44:12 i suppose so 2014-07-09 07:44:21 and i thinkt that might be acceptable 2014-07-09 07:44:22 i mean 2014-07-09 07:44:31 if you unplug a system disk... 2014-07-09 07:44:35 what can you expect 2014-07-09 07:44:43 or if a system disk fails 2014-07-09 07:44:56 because that is what your disk is now, a system disk 2014-07-09 07:45:04 yeah 2014-07-09 07:45:08 you're screwed 2014-07-09 07:45:14 and systemd can't solve that 2014-07-09 07:45:22 no matter how hard it tries 2014-07-09 07:45:44 now, i the fork/exec-on-demand feature is a kinda nice thing 2014-07-09 07:45:55 At least, it will stop your services instead of watching them crash (if it manages to stop them in time) 2014-07-09 07:45:58 and i think this is exactly what dbus provides 2014-07-09 07:46:24 well, it *could* hande it if you cleanly umount it 2014-07-09 07:46:31 but uh, you cannot unmount it if its in use... 2014-07-09 07:46:37 yep 2014-07-09 07:47:17 so you would have to tell the (a) service, shut done this mount and all the services depending on it 2014-07-09 07:48:37 now 2014-07-09 07:48:41 this is mounts 2014-07-09 07:48:49 what if we do similar with network? 2014-07-09 07:49:00 network mounts, you mean ? 2014-07-09 07:49:02 no 2014-07-09 07:49:06 netwrok serivces 2014-07-09 07:49:09 oh 2014-07-09 07:49:09 like sshd, httpd etc 2014-07-09 07:49:28 eh, you'd not want that, I guess 2014-07-09 07:49:41 well 2014-07-09 07:49:44 having your services stopped at each deconnection... 2014-07-09 07:49:58 you could get a similar problem as your slow disk 2014-07-09 07:50:05 yeah 2014-07-09 07:50:11 some network interfaces could be really slow before they show up 2014-07-09 07:50:14 like ppp0 2014-07-09 07:50:43 so we could have a similar approach 2014-07-09 07:50:52 openrc handle this case 2014-07-09 07:51:06 you can put a dependency on service net.ppp0 2014-07-09 07:51:11 right 2014-07-09 07:51:29 b-b-b-but 2014-07-09 07:51:36 can't we do the same for mounts ?! 2014-07-09 07:51:38 i think we kinda support it too in alpine 2014-07-09 07:51:41 mount.home 2014-07-09 07:52:13 the problme is that the disk device is not there when the service wants to start 2014-07-09 07:52:18 that will not return untill "mount | grep -q '/home'" is 0 2014-07-09 07:52:40 so it should be triggered with a uevent uevent 2014-07-09 07:52:57 instead of polling with a loop, is /home there? nope, ok wait and try again... 2014-07-09 07:53:16 netlink uevent* 2014-07-09 07:53:17 well, implementation is another question anyway 2014-07-09 07:53:26 but this might work 2014-07-09 07:53:57 i think its hard to get it right with openrc 2014-07-09 07:54:05 yeah sure 2014-07-09 07:54:18 but as i said, i think you can hack it :) 2014-07-09 07:54:26 I'm convinced that linux need a new init system 2014-07-09 07:54:32 same 2014-07-09 07:54:33 but it's definitely not systemd 2014-07-09 07:54:40 thats what i want too 2014-07-09 07:54:51 i want a new init system, but not systemd 2014-07-09 07:55:05 something simple 2014-07-09 07:55:09 yup 2014-07-09 07:55:10 REALLY simple 2014-07-09 07:55:22 well, it cannot be too simple 2014-07-09 07:55:47 because the problem it needs to solve might not be that simple 2014-07-09 07:55:55 init -> int main (void) { execv("/sbin/rc", "init", NULL); return 0; } 2014-07-09 07:56:01 ah 2014-07-09 07:56:04 yes 2014-07-09 07:56:07 the pid1 2014-07-09 07:56:16 that i completely agree with 2014-07-09 07:56:44 the mount problem and services starting/monitoring and dependency handling must be out of pid 1 2014-07-09 07:56:55 and then, /sbin/rc would consider mounts, net interface and services as files 2014-07-09 07:57:25 device is mounted ? -> create /var/svc/device.mount 2014-07-09 07:57:53 eth0 is us ? -> create /var/svc/eth0.iface 2014-07-09 07:58:05 mpd is started ? -> /var/run/mpd/pid 2014-07-09 07:58:35 is address 1.2.3.4 assigned any interface? -> /var/svc/net.ipv4.addr.1.2.3.4 2014-07-09 07:58:41 then, to handle dependency -> test -f /var/svc/$file.$type 2014-07-09 07:59:09 maybe use inotify to trigger services to start? 2014-07-09 07:59:22 ncopa, or maybe : grep '1.2.3.4' /var/svc/*.iface 2014-07-09 07:59:22 lets say you want bind httpd to addr 1.2.3.4 2014-07-09 07:59:34 ooooh yeah ! 2014-07-09 07:59:35 inotify ! 2014-07-09 07:59:42 but the problem is 2014-07-09 07:59:57 you might need depend on more than one prerequesite 2014-07-09 08:00:03 example httpd 2014-07-09 08:00:09 you want bind to 1.2.3.4 2014-07-09 08:00:16 but you als depend on /var to be mounted 2014-07-09 08:00:28 well, in your httpd script, just add more tests 2014-07-09 08:01:01 no you dont want httpd to even try start before all preconditions are there 2014-07-09 08:01:19 test -f /var/svc/http/mount && grep -q 1.2.3.4 /var/svc/*.iface 2014-07-09 08:01:27 once all prereqs are there, then you fire off httpd 2014-07-09 08:01:33 yeah sure 2014-07-09 08:02:09 if your services are shell script, you could think of a "depends()" function 2014-07-09 08:02:33 right 2014-07-09 08:02:54 execpt i'd like to avoid describing the service depends in shell 2014-07-09 08:03:04 i like declarative remember ;) 2014-07-09 08:03:16 depends() { 2014-07-09 08:03:16 while ! test -f /var/svc/http.mount; do sleep 1; done 2014-07-09 08:03:16 while ! grep -q '1.2.3.4' /var/svc/*.iface; do sleep 1; done 2014-07-09 08:03:16 } 2014-07-09 08:03:27 Could be made declarative too 2014-07-09 08:03:34 the above wont work 2014-07-09 08:03:41 thats no how openrc works 2014-07-09 08:03:57 it will in fact run all the serivces' depend function 2014-07-09 08:04:08 long time before it tries to start anything 2014-07-09 08:04:15 and it wil actualy cache the result 2014-07-09 08:04:27 gen-depends.sh or so 2014-07-09 08:04:39 well, a better initrc would run it right before starting the script :) 2014-07-09 08:05:14 Anyway, I'm sure this could be a nice idea 2014-07-09 08:05:15 and i think i'd like to avoid the .pid files 2014-07-09 08:05:18 yes 2014-07-09 08:05:25 why ? 2014-07-09 08:06:09 i think the entire daemonize idea comes from trying to work around the drawbacks of sysv 2014-07-09 08:06:27 work around what sysv init could not solve 2014-07-09 08:06:37 i think we want proper process monitoring 2014-07-09 08:06:37 Oh, so how would you handle it ? 2014-07-09 08:06:39 threads ? 2014-07-09 08:06:55 no, likely simliar to what runit does 2014-07-09 08:06:58 (and systemd) 2014-07-09 08:07:13 how do they work in depth ? 2014-07-09 08:07:36 fork wrappers to them ? 2014-07-09 08:08:04 something like "systemd-run /usr/bin/mpd $MPD_OPT" 2014-07-09 08:12:26 service mpd start 2014-07-09 08:12:28 or similar 2014-07-09 08:13:08 systemctl start mpd 2014-07-09 08:13:14 yeah 2014-07-09 08:13:17 http://smarden.org/runit/runsvdir.8.html 2014-07-09 08:13:19 but I mean, how does systemd handle this internally ? 2014-07-09 08:13:47 http://smarden.org/runit/runsv.8.html 2014-07-09 08:13:57 supervising 2014-07-09 08:14:04 was the word i was looking for 2014-07-09 08:14:22 I see 2014-07-09 08:14:36 thats better than .pid files imho 2014-07-09 08:14:42 yeah sure 2014-07-09 08:19:15 Morning all! 2014-07-09 08:19:36 morning 2014-07-09 08:19:45 ncopa, I'm writing all of this down 2014-07-09 08:19:50 just in case ;) 2014-07-09 08:20:24 there was some discussion yesterday about udev and systemd etc...i probs missed half of it at least due to connection issues. Anyway there is hope, Gentoo have eudev project: http://www.gentoo.org/proj/en/eudev/ 2014-07-09 08:21:03 yeah, and it works fine 2014-07-09 08:21:13 crux use it already 2014-07-09 08:21:22 cool 2014-07-09 08:24:01 ncopa, I think that shell script could still be nice to use for services 2014-07-09 08:24:18 But make them declarative first 2014-07-09 08:24:28 As ar openrc scripts 2014-07-09 08:24:30 *are 2014-07-09 08:25:07 you can create a script only by fullfilling variables (command, command_opt, ...) or by overriding functions) 2014-07-09 08:25:09 i dont hinkg we use eudev yet 2014-07-09 08:25:18 mdev is nice 2014-07-09 08:25:34 it makes alpine extremly light and fast 2014-07-09 08:25:48 but mdev does not provide libudev which is needed for hotlpugging keyboards and mouse in xorg without reconfiguration 2014-07-09 08:26:07 eudev is an option 2014-07-09 08:26:31 but i dont like the idea of a try-to-catch-up clone 2014-07-09 08:26:52 id like a new libkueventd 2014-07-09 08:27:04 with a libudev-compat.h or similar 2014-07-09 08:27:30 similar API so you can port from libudev 2014-07-09 08:28:50 You're a bit of a revolutionnary guy, aren't you? ^^ 2014-07-09 08:31:40 not really 2014-07-09 08:31:46 i just things to be done properly :) 2014-07-09 08:32:02 well, building a distro without gnu libc might be considered revolutionary... 2014-07-09 08:33:06 haha no 2014-07-09 08:33:12 that's common sense 2014-07-09 08:35:02 Hi looking for package name py-m2crypto 2014-07-09 08:35:03 http://wiki.alpinelinux.org/wiki/User:Fab/Salt 2014-07-09 08:35:08 where is it :) ? 2014-07-09 08:35:32 in edge/testing 2014-07-09 08:35:43 someone needs confirm that it works so it can be moved to main 2014-07-09 08:36:05 will it compile on main ? 2014-07-09 08:36:10 3.0.1 2014-07-09 08:36:27 I will test it now, I need it dependency for building crda 2014-07-09 08:36:30 which is also missing 2014-07-09 08:40:15 is there a wiki explaining how to quickly add testing repository and have it not conflict with main 2014-07-09 08:40:34 or maybe allow only specific packages and it dependices to be updated 2014-07-09 09:05:29 good morning 2014-07-09 09:06:35 dbparse module for python anyone where to get it? 2014-07-09 09:07:59 never mind 2014-07-09 09:19:39 in what package is install command? 2014-07-09 09:20:44 Good morning... I'm trying to build a kernel module package for libreswan - with help of #swan I got it to compile, now I'm stuck at 'abuild rootpkg' which gives a couple of 'permission denied' errors... 2014-07-09 09:20:57 zxd: I think it's in 'coreutils'. 2014-07-09 09:22:21 install -m 644 -t //usr/lib/crda/pubkeys/ root.key.pub.pem; \ 2014-07-09 09:22:21 fi 2014-07-09 09:22:21 invalid option -- t 2014-07-09 09:22:25 k 2014-07-09 09:22:50 install is probably from busybox 2014-07-09 09:22:53 looks good 2014-07-09 09:23:05 yes busybox install missing -t 2014-07-09 09:23:14 install coreutils 2014-07-09 09:23:43 danci1973> Can you put a pastebin somewhere ? 2014-07-09 09:24:06 zxd, or correct the install line 2014-07-09 09:24:45 igmar: http://paste.ubuntu.com/7769466/ 2014-07-09 09:24:51 install -m 644 root.key.pub.pem /usr/lib/crda/pubkeys/root.key.pub.pem 2014-07-09 09:26:00 igmar: I can't remember if there's anything more I'd need to setup for 'fakeroot'... 2014-07-09 09:26:32 igmar: It's working on my old 2.3.6 alpine-devel box... 2014-07-09 09:27:37 Looks like fakeroot isn't working at all 2014-07-09 09:32:30 does Makefile:126: recipe for target 'install-libreg' failed 2014-07-09 09:32:30 make: *** [install-libreg] Error 1 2014-07-09 09:32:37 does anyon know how to get more debug info from make 2014-07-09 09:32:47 -d I guess 2014-07-09 09:34:50 make 4 should be a lot more verbose if asked 2014-07-09 09:35:17 make 4 ? 2014-07-09 09:35:42 make version >= 4.00 2014-07-09 09:36:44 GNU Make 4.0 2014-07-09 09:38:19 use --trace in that case 2014-07-09 09:44:30 Makefile:126: target 'install-libreg' does not exist 2014-07-09 09:44:30 echo ' INSTALL libreg' 2014-07-09 09:44:30 INSTALL libreg 2014-07-09 09:44:30 mkdir -p /usr//lib 2014-07-09 09:44:30 cp libreg.so /usr//lib/ 2014-07-09 09:44:30 ldconfig 2014-07-09 09:44:31 Makefile:126: recipe for target 'install-libreg' failed 2014-07-09 09:44:31 make: *** [install-libreg] Error 1 2014-07-09 09:44:58 why does it say it dosen't exist the target 2014-07-09 09:45:20 Does the target really exist ? 2014-07-09 09:45:34 grep 'install-libreg' Makefile 2014-07-09 09:45:48 install-libreg-headers: 2014-07-09 09:45:48 install-libreg: 2014-07-09 09:45:48 install: install-libreg install-libreg-headers crda crda.8.gz regdbdump.8.gz 2014-07-09 09:46:36 that's target no ? install-libreg: , if it didn't exist how would it have known to print the lines underneath the target 2014-07-09 09:46:40 that supposdly is missing 2014-07-09 09:48:19 what doesn make install-libreg say ? 2014-07-09 09:49:06 same 2014-07-09 09:49:06 Makefile:126: target 'install-libreg' does not exist 2014-07-09 09:51:00 http://pastebin.com/SM74Xq0X < Makefile 2014-07-09 09:53:52 weird 2014-07-09 09:59:55 Looks like something in that target is failing 2014-07-09 10:46:03 ncopa: i totally agree with you about systemd :) thanks for clarification. 2014-07-09 10:49:40 what about systemd ? 2014-07-09 10:57:55 it's a nice try 2014-07-09 10:58:14 but a failed attempt at a new init system 2014-07-09 10:59:57 true, it still can't make coffee 2014-07-09 11:00:27 I boot my system like once every month 2014-07-09 11:01:17 it will 2014-07-09 11:01:24 soon. 2014-07-09 11:21:22 Hope that also includes cappucino and black coffee 2014-07-09 11:43:35 wow, init scrollback overload 2014-07-09 11:44:29 i had a brain/idea dump sorry :) 2014-07-09 11:48:10 daemontools, runit, etc are pretty simple: don't double fork to reparent the child process but in stead fork and exec and supervise the child by lisyening to SIGCHLD 2014-07-09 11:49:03 something like that is what i was thinking of instead of .pid files 2014-07-09 11:49:09 yup 2014-07-09 11:49:17 thats the sane way to do it 2014-07-09 11:49:24 and where openrc does it wrong 2014-07-09 11:50:59 i've been meaning to run alpine on runit, but haven't had the time to test it yet 2014-07-09 11:51:15 or s6 2014-07-09 11:51:48 like sabotage or ignite for archlinux 2014-07-09 11:52:05 i find s6 a bit too "weird" but i like the ideas (mostly) 2014-07-09 11:52:08 even voidlinux are migrating to runit 2014-07-09 11:52:45 runit is the strongest candidate for openrc replacement atm 2014-07-09 11:52:59 yeah, runit from bb will likely be a good 0lace to start 2014-07-09 11:53:11 *nod* 2014-07-09 11:53:34 It's better to pick systemd 2014-07-09 11:53:47 too bad they dont support musl libc 2014-07-09 11:53:48 damn, typing on a phome keyboard is hard 2014-07-09 11:53:51 ACTION awaits the rant 2014-07-09 11:54:07 doesn't systemd ? 2014-07-09 11:54:13 igmar: nope 2014-07-09 11:54:25 and if you politely try ask 2014-07-09 11:54:34 guess what the devs say? 2014-07-09 11:54:40 I absolutely fucking hate ubuntu, and I hate ubuntu's cloud-init even more 2014-07-09 11:54:41 I've used systemd since arch swithched som years ago and I'm not impressed 2014-07-09 11:55:13 i like many of the technical ideas behind systemd 2014-07-09 11:55:43 I dont like the black box approach 2014-07-09 11:56:02 super hard to understand what's going on and debug 2014-07-09 11:56:04 gah stop mentioning systemd 2014-07-09 11:56:33 but supervision and socket activation and declerative conf is neat 2014-07-09 11:56:37 yeah, maybe we should stop use the s-word 2014-07-09 11:56:39 it's like solaris' smf dumped into a pool of shit 2014-07-09 11:56:59 smf is xml... 2014-07-09 11:57:00 i'm ok we talk about constructive ideas 2014-07-09 11:57:04 let me put it like this: if we need(ed) anything new - it should be GOOD 2014-07-09 11:57:08 not "good enough" 2014-07-09 11:57:17 i don't have any time for good enough. 2014-07-09 11:57:17 darkfader: exactly 2014-07-09 11:57:17 and sml does not run ax pid 1 2014-07-09 11:57:41 openrc is good enough.. for now 2014-07-09 11:57:51 but i think we can do better 2014-07-09 11:58:05 uggedal: but it doesn't have unhandled races that won't let you login to fix a broken fstab entry 2014-07-09 11:58:28 even if systemd ran as uid "GOD on mars" it would still be very very badly designed 2014-07-09 11:58:31 uggedal: i agree with you, supervision, socket activation and declarative conf is nice 2014-07-09 12:00:05 systemd isn't badly designed. I just want to pick what I want to use, and what not. 2014-07-09 12:00:14 ncopa: if you come along and say you wanna have an init in erlang that was specced out before coding i'd even shell out a few hundred $$$ for it 2014-07-09 12:00:29 erlang ? Please not 2014-07-09 12:00:33 lol 2014-07-09 12:00:35 but the next thing that fails at reliably booting... don't need. 2014-07-09 12:01:18 igmar: i was just grabbing the craziest example. no worry :) 2014-07-09 12:02:43 a super simple systemd unit fike to runit run scripts translator in abuild would be a good start for getting easy support for daemons under runit 2014-07-09 12:02:46 also, how many network config frameworks can you use for systemd on arch right now? 3? 2014-07-09 12:02:49 I hardly boot machines, so I don't care about the init system 2014-07-09 12:03:20 I just hate Ubuntu 2014-07-09 12:03:24 :)) 2014-07-09 12:03:51 rm /etc/profile.d/* helps a bit ;)) 2014-07-09 12:04:13 or was it /etc/update.motd.d/* ? 2014-07-09 12:04:22 uggedal: yes, i have been thinking that we want something that helps us reuse the unit files .ini 2014-07-09 12:04:34 in an ideal world supdrvision would not be needed, but linux oom and stupid userland devs means things wkll crash 2014-07-09 12:04:56 either by using them as is, or with only minor modification, or by make script to convert them 2014-07-09 12:05:35 and i have been thinking of maybe use either s6-init or runit as init (pid 1) and then bold on other supervision daemon 2014-07-09 12:05:37 uggedal: it's also a core feature of init since ever, so it should be there (:respawn) 2014-07-09 12:05:45 doesn't need to be default though 2014-07-09 12:05:54 bolt* 2014-07-09 12:05:56 people might run real clusters 2014-07-09 12:06:02 darkfader: inittab is a pita to use though 2014-07-09 12:06:16 kaniini talked about writing a supervisiond 2014-07-09 12:06:36 i'd vote for runit, in case 2014-07-09 12:06:40 i've already written one 2014-07-09 12:07:09 but like runit as pid 1 better 2014-07-09 12:07:25 The only thing that fixes Ubuntu is rm -rf / 2014-07-09 12:07:28 what i'd like to hear is that you spend a few hours on design before writing anything (not "you" but anyone tackling it) 2014-07-09 12:08:34 for embedded / industrial / ha setups it's a pretty basic task to look at distros inits, get gray hair and then rip it all apart because it's so "i'll write us an init"-ish 2014-07-09 12:11:06 darkfader: thats what i want too 2014-07-09 12:11:30 discussion on how to design things 2014-07-09 12:11:34 ideas on the table 2014-07-09 12:11:42 write a spec what ideas to keep 2014-07-09 12:11:48 and what to throw out 2014-07-09 12:12:04 Isn't it better to describe objectives ? 2014-07-09 12:12:17 nad then pick the best-fit 2014-07-09 12:12:51 i suppose you are right 2014-07-09 12:12:57 but there are no best-fit atm 2014-07-09 12:13:15 The always is. The fix might be small however 2014-07-09 12:16:17 objective: small simple secure init system 2014-07-09 12:16:52 that reliably can bring up the system without introducing bloat and unnecessary complexity 2014-07-09 12:17:18 (but keep in mind that the problem we want to solve might not be simple so we will need some complexity) 2014-07-09 12:17:38 What's wrong with sysvinit then ? :) 2014-07-09 12:17:49 it includes init (pid 1) 2014-07-09 12:18:02 it includes hotplug manager (ala udev) 2014-07-09 12:18:04 igmar: /var/lock/subsys for one 2014-07-09 12:18:22 ... which likely means we need some sort of libudev compat 2014-07-09 12:18:32 it includes a supervision daemon 2014-07-09 12:18:51 it will likely need some communication system between the different parts 2014-07-09 12:19:38 and it should try to be "tecnically correct" 2014-07-09 12:19:45 which is where sysvinit fails 2014-07-09 12:20:27 if you need 'sleep ...' then you do something wrong 2014-07-09 12:21:44 whats wrong with sysvinit is that you need things like: while ! service_is_up_and_running; do sleep 1; done 2014-07-09 12:21:54 same issue with openrc 2014-07-09 12:23:04 here is the proof that we currently do things wrong: http://sprunge.us/dMSZ 2014-07-09 12:23:29 i'm putting it in an etherpad, removing the timestamps from irc takes a moment still 2014-07-09 12:23:35 The socket based launch from systemd isn't a bad idea 2014-07-09 12:23:47 agree 2014-07-09 12:23:50 that is a nice thing 2014-07-09 12:24:12 Same with the dependencies if you ask me 2014-07-09 12:24:24 that might be true too 2014-07-09 12:24:29 i would expect it to be true 2014-07-09 12:25:25 as previously said, my major issues with systemd is: no musl libc support, dbus dependency (will be fixed with kdbus) 2014-07-09 12:25:40 and the narcissistic management style 2014-07-09 12:30:29 https://etherpad.mozilla.org/pH2iKzR5ln 2014-07-09 12:31:01 i'll need to go buy something that's why i dumped my notes in it. but feel free to just delete them if it is too negative 2014-07-09 12:33:41 dependencies are nice yes. i liked the "subsystem" thinking from smf there, in systemd i just didn't find a pattern really 2014-07-09 12:33:54 --all | grep seems to be the pattern ;) 2014-07-09 13:00:03 re 2014-07-09 14:25:40 there is a bug in ldconfig script that makes it not exit with 0 2014-07-09 14:25:49 . /sbin/ldconfig 2014-07-09 14:41:37 what happens if i lbu exclude /etc/apk/world 2014-07-09 14:47:10 it will not store the world in config 2014-07-09 14:47:27 and you will basically dont have any packages installed at boot up 2014-07-09 14:51:36 will it still load what I had in previous world file in /etc/apk/world stored in apkvol.tar.gz 2014-07-09 14:51:50 I mean will it load the world file that I already have in apkvol.tar.gz 2014-07-09 14:52:12 what I basically want to do is to commit some changes but not the ones to world 2014-07-09 14:53:16 zxd_: no, it will create a new apkovl and rename the old one (if backups are enabled) 2014-07-09 14:53:44 zxd_: since apkovl files are tar.gz files, you can manually edit the apkovl to removed packages from world 2014-07-09 14:54:01 pk 2014-07-09 14:54:07 okie 2014-07-09 14:54:10 brb 2014-07-09 15:05:13 should I disable mdev with rc-service if I enable udev udev-postmount 2014-07-09 15:06:46 it dosen't seem like udev is working 2014-07-09 15:07:09 don't see any events with udevadm monitor --environment kernel 2014-07-09 15:14:19 I read a thread about how /usr must be mounted before init 2014-07-09 15:15:22 like crda /sbin/crda 2014-07-09 15:22:24 brb 2014-07-09 15:31:22 well it seems that udev and modules parameters are read after apkvol.tar.gz is unpacked, so I guess also is cfg80211 module that calls CRDA 2014-07-09 15:32:03 why isn't udevadm monitor --environment kernel capturing any events when I do iw reg set 2014-07-09 15:32:09 why isn't udevadm monitor --environment kernel capturing any events when I do iw reg set IL 2014-07-09 15:35:25 open("/sys/class/ieee80211/reg/index", O_RDONLY|O_LARGEFILE) = -1 ENOENT (No such file or directory) 2014-07-09 16:57:28 what happens if I remove mdev from sysinit 2014-07-09 17:01:54 nothing 2014-07-09 17:02:10 it will not populate your /dev, and thus, will not boot 2014-07-09 17:02:47 cool 2014-07-09 17:02:58 brb 2014-07-09 18:20:35 sounds like sh MAKEDEV all is on its way 2014-07-09 23:31:16 when trying to build any abuild, i get the following error:$ abuild -r 2014-07-09 23:31:34 usr/bin/abuild: .: line 1969: : not found 2014-07-09 23:32:23 building the package without abuild works, and i have this problem with all abuilds 2014-07-10 08:36:05 Hello all 2014-07-10 08:36:12 hi 2014-07-10 09:41:14 wiki.alo is down? 2014-07-10 09:50:21 http://wiki.alpinelinux.org? I can reach it, seems operational 2014-07-10 10:26:57 'morning ... Did 'abuild rootpkg' change in recent versions? It seems like old version (from Alpine 2.3.6) ran 'build' and then 'package', while the new version (Alpine 3.0.1) only does 'package' - is this right? 2014-07-10 12:12:47 danci1973: it depends a bit on the APKBUILD 2014-07-10 12:13:04 if there was no package() function it would run build() instead 2014-07-10 12:13:30 you can do: abuild deps clean unpack prepare build rootpkg 2014-07-10 12:14:29 ncopa: I had the same APKBUILD on both Alpine versions - the old one worked as it first ran build() and then package(), the new one didn't work, cause it ran just package() ... 2014-07-10 12:14:51 yes, i think that could be the case 2014-07-10 12:15:12 ncopa: Turns out export INSTALL_MOD_PATH="$pkgdir" was defined in build() but not in package(); 2014-07-10 12:15:24 ok 2014-07-10 12:16:46 How would I best setup my build environment so I could build 'kernel module' packages for an USB based Alpine install? 2014-07-10 12:17:58 Is it possible to build these packages with different kernel versions, so people could install them without upgrading the kernel? 2014-07-10 18:34:10 hi 2014-07-10 18:34:25 where can I see status about a specific package 2014-07-10 18:34:42 feature request 2014-07-10 18:34:50 like compile with option CONFIG_ACS=y in hostapd 2014-07-10 18:37:11 hi 2014-07-10 18:37:32 fill an issue on bugs.alpinelinux.org 2014-07-10 19:10:36 hi was looking for way to force device name with udev, and found most distros have this file /lib/udev/write_net_rules that generates the udev rules for persistent interface name, it seems this is missing in alpine-linux 2014-07-10 19:10:42 is it maybe in some different package? 2014-07-10 21:11:25 zxd_ i have udev rules inside: /etc/udev/rules.d/52_reader.rules 2014-07-10 22:56:28 Anybody tried compiling or installing golang on alpine? 2014-07-11 09:30:17 Good Morning everyone 2014-07-11 14:54:32 it would be cool to get phoronix test suit running on alpine: https://github.com/phoronix-test-suite/phoronix-test-suite/ 2014-07-11 14:55:19 make make some stupid simple setup script that even phoronix editor could use 2014-07-11 14:55:48 (they could not handle arch linux, but needed gui installer from manjaro) 2014-07-11 14:57:13 ouch 2014-07-11 14:57:36 i always forget how to set the disk it should run the disk tests agains 2014-07-11 14:57:37 t 2014-07-11 14:59:23 hehe: 2014-07-11 14:59:23 error: Your local changes to the following files would be overwritten by merge: main/perl-io-tty/APKBUILD 2014-07-11 14:59:26 ACTION sitting there 2014-07-11 14:59:31 wtf did i do... 2014-07-11 14:59:54 ah, changed the version 2014-07-11 15:00:05 git diff 2014-07-11 15:00:09 i need a clone with a fresh brain 2014-07-11 15:00:28 its 28.2 celcius here 2014-07-11 15:00:37 and its friday 2014-07-11 15:00:39 let me ask nagios lol 2014-07-11 15:00:46 and its time to go... 2014-07-11 15:00:47 ... 2014-07-11 15:00:49 HOME! 2014-07-11 15:00:51 :) 2014-07-11 15:01:02 you're a smart man, sir 2014-07-11 15:01:15 enjoy the weekend 2014-07-11 15:01:17 mv $self $HOME 2014-07-11 15:01:19 you too 2014-07-11 15:01:29 have a nice weekend 2014-07-11 16:30:48 o/ 2014-07-11 23:01:50 hi 2014-07-11 23:02:03 eeer stupid question: i never used alpine linux different than do a full system install on the harddrive 2014-07-11 23:02:18 is there a way to boot a memory image of the distro that runs in memory and just stores the stuff on the harddrive? 2014-07-11 23:02:30 i've seen it during setups that i can do a config only setup, but how do i boot it? 2014-07-11 23:06:19 hm, i guess i'm looking for a mixture of diskless and data mode 2014-07-11 23:12:17 grimeton: you want to run from memory? 2014-07-11 23:12:42 it's more or less that, let me explain 2014-07-11 23:12:49 i've got a small atom box that came with an 8gb ssd 2014-07-11 23:13:00 if you ask me this ssd is more or less an msata usb thumbdrive 2014-07-11 23:13:18 so what i'd like todo is to boot alpine from it, run it in memoy and store my settings on an r/w partition on the ssd 2014-07-11 23:13:27 what i'm actually looking into is todo the following: 2014-07-11 23:13:31 have one big partition that is r/w 2014-07-11 23:13:35 put the iso of alpine on it 2014-07-11 23:13:50 let grub boot the iso into memory and run alpine from the iso in memory and then let alpine store the data on the r/w partition 2014-07-11 23:14:02 which is what i'd call diskless-data mode 2014-07-11 23:14:03 what most do is 2014-07-11 23:14:36 so i don't need to use an extra drive or usb thumbdrive or something to boot alpine 2014-07-11 23:14:44 and the ssd isn't messed with too much 2014-07-11 23:15:02 i would format it fat32 2014-07-11 23:15:08 copy cd to it 2014-07-11 23:15:12 install syslinux 2014-07-11 23:15:13 boot 2014-07-11 23:15:18 and then tell alpine to store the settings in the fat32 partition? 2014-07-11 23:15:27 with the unix sec add on? 2014-07-11 23:15:42 alpine will detect the disk and mount it in /media 2014-07-11 23:15:47 unix sec add on -> that add on that allows to store file permissions and stuff on a fat filesystem 2014-07-11 23:15:51 you set that media in lbu.conf 2014-07-11 23:16:05 so it will store overlay on the disk 2014-07-11 23:16:24 disk will mount rw to write it and afterwards mount it ro again 2014-07-11 23:16:44 yeah but that requires me to run the lbu stuff before i reboot or after i changed something - right? 2014-07-11 23:16:58 yes 2014-07-11 23:17:40 my setup tends to have the data mode version (http://wiki.alpinelinux.org/wiki/Installation ) with the iso inside the data partition 2014-07-11 23:18:13 because i know myself 2014-07-11 23:18:20 and trust me, i'll forget to run lbu by default 2014-07-11 23:18:59 thats the cost of running from ram. 2014-07-11 23:19:27 you can also setup cron to do it hourly or so 2014-07-11 23:19:38 or daily 2014-07-11 23:19:47 yeah but the data mode should do that for me and let me run from ram 2014-07-11 23:19:50 or did i get something wrong? 2014-07-11 23:20:35 you can mount var on disk on a ram based ssystem 2014-07-11 23:20:45 but then etc is still not stored 2014-07-11 23:21:15 if you want to run from ram and not use the disk that much, then lbu is your only friend 2014-07-11 23:21:33 hm i see 2014-07-11 23:21:45 lbu it is then 2014-07-11 23:21:46 thanks 2014-07-11 23:22:08 you could also add an init script to lbu_commit on reboot/poweroff 2014-07-11 23:23:18 hmm 2014-07-11 23:30:24 sometimes I create 2 partitions, one which holds kernel and pkgs and config. and other will mount as /home 2014-07-11 23:30:53 by default lbu only backup etc 2014-07-11 23:34:08 time for bed. gnite. 2014-07-11 23:34:12 nite 2014-07-11 23:34:13 and thanks 2014-07-12 02:34:29 my apk fu just fails me 2014-07-12 02:34:52 how do i find out which package contains a specific file? 2014-07-12 03:59:59 is there a reason why rsync is built without ipv6 support? 2014-07-12 04:10:37 probably just a mistake, easily overlooked since most uses of rsync are with ssh providing the transport 2014-07-12 04:11:00 i'm running an rsync server on dl-5.alpinelinux.org and just wondered why it isn't reachable via v6 2014-07-12 04:11:10 and figured that one out =) 2014-07-12 04:11:28 perhaps file a bug report against the package 2014-07-12 04:13:10 yeah guess that's best 2014-07-12 11:36:44 Testsuite summary for libressl 2.0.0 / TOTAL: 41 / PASS: 41 2014-07-12 11:38:04 on Alpine Edge, with two minor patches (sys/sysctl.h and gnu/libc-version.h headers) and CFLAGS="-Wno-error=attributes" 2014-07-12 14:48:53 did anyone try to build the ceph-common pieces on alpine? 2014-07-12 14:49:26 i'm just interested in having rbd working but don't think i can make it work 2014-07-12 16:20:44 hello, i find this distro intretsing but find the documenation lacking (at least on bridges), 2014-07-12 16:23:20 or to say it another way, for the novice the page of bridges is a little confusing.... it says you dont assign ip addresses manually to the ports. in my example i have a modem on eth0, and a switch attached to eth1, eth0 has static ip so i have to configure it and i dont have any idea where, since i need to add a dhcp server to eth1 and possibly give it a ip, but again the documenbtaion justs says you assign a ip address to the bridge at br0, wtf? 2014-07-12 16:24:24 btw , the switch attached to eth1 is dumb, it has no web interface, it merely is the most basic of switches, i'm not even sure it has a microprocessor! 2014-07-12 16:25:10 Bridging is then configured in /etc/network/interfaces with the bridge-ports keyword. Note that you normally don't assign ip addresses to the bridged interfaces (eth0 and eth1 in our example) but to the bridge itself (br0). <- the actual quote from the web page 2014-07-12 16:25:52 so if eth0 has to be configured as static ip and eth1 needs a gateway address, where on earth does the static ip info go\ 2014-07-12 16:28:10 at this point i am very tempted to look to see how my dd-wrt box handles it 2014-07-12 16:36:50 ACTION dumps coca-coca all over everything getting the pc out of closet self because documenation  2014-07-12 16:37:08 is not sufficent and he is clumsly 2014-07-12 16:37:41 auto br0 2014-07-12 16:37:42 iface br0 inet static 2014-07-12 16:37:42 bridge-ports eth0 eth1 2014-07-12 16:37:42 bridge-stp 0 2014-07-12 16:37:42 address 192.168.0.1 2014-07-12 16:37:42 netmask 255.255.255.0 2014-07-12 16:37:57 is the example on the wiki..... 2014-07-12 16:38:40 i would love to know just where this address 192.,168.0.1 is, it is on both ports , how does one confiture eth0 if he cant do it directly \ 2014-07-12 16:43:30 great i erased the hd dd-wert was on 2014-07-12 16:44:36 you know, i been using computers since the sinclair zx81 and the documenation these days makes so little sense sometimes that one should be told to go screw oneself and not bother instead 2014-07-12 16:45:40 i learned two programming languages, used various 8 - bit machines, had a couple of amigas, hated windows, got into linux and struggled for years and the problem is the same every time. the documentaion is very unclear with phrashes like the above and it sucks \ 2014-07-12 16:47:22 Bridging is then configured in /etc/network/interfaces with the bridge-ports keyword. Note that you normally don't assign ip addresses to the bridged interfaces (eth0 and eth1 in our example) but to the bridge itself (br0). <----- yeah, great ,, 2014-07-12 17:16:03 ACTION screams in agony  2014-07-12 17:16:53 seems the debian bridge software is not the same as alpiine\ 2014-07-12 17:19:28 arrgh420: i can give you a working config from one system, the problem is i have used it for another one and there it didn't work so far. not sure if it'd help you or make it even more complicated right now 2014-07-12 17:20:32 i've done it a few times, sometimes on lacp and with vlans, and it seems to be horrible on any distro (at least to me) 2014-07-12 17:21:44 i had it working on dd-wrt ;( 2014-07-12 17:29:03 if it helps: it'll work in the end 2014-07-13 09:13:42 Have anyone succeded to run last version of Asterisk (12.4.0) on Alpine 3.0.1? 2014-07-13 09:16:16 Tried on multiple machines and it crashed at the beginning with the error "unable to get real address of SSL_library_init: Symbol not found: SSL_librari_init" 2014-07-13 09:17:01 Tried running from ram or sys install.. the result is the same 2014-07-13 09:17:48 It used to work without problems on alpine 2.7 version 2014-07-13 09:18:24 please...I need little help with this 2014-07-13 13:44:55 laurentius1977, seems i can reproduce it. 2014-07-13 13:45:08 i tested asterisk on musl on earlier version; seems this a new issue 2014-07-13 13:47:49 sounds like underlinking issue 2014-07-13 14:01:21 i have a hunch on the issue. i'll take a look at it a bit later. 2014-07-13 14:28:23 thank you very much 2014-07-13 14:39:26 laurentius1977, i pushed a fix to edge 2014-07-13 14:39:56 you are great 2014-07-13 14:40:12 many thanks again 2014-07-13 14:40:28 can you test it, if it works, i'll backport it to 3.0-stable too 2014-07-13 14:40:55 I will right now 2014-07-13 15:03:42 Thank you 2014-07-13 15:03:50 works now 2014-07-13 15:04:25 You were very fast 2014-07-13 15:06:56 I have a question...if somebody can help me with an answer 2014-07-13 15:07:27 can googleearth aplication be ported to alpine linux desktop? 2014-07-13 15:07:53 i found no source files for this proprietary app 2014-07-13 15:08:27 can it be ported from debian version somehow? 2014-07-13 15:47:36 laurentius1977: no, without source it can't be ported 2014-07-13 15:48:01 you can probably run it in a chroot or lxc though 2014-07-13 15:48:35 thank you 2014-07-13 15:49:05 I will try to run it in chroot or lxc 2014-07-14 16:30:07 Is there any reason why alsa-lib isn't configured and compiled with --with-cards=aloop by default? I ask, since I would welcome snd-aloop on Alpine (soundcard without hardware capture mixing). 2014-07-14 16:42:04 Wrong package, kernel modules are build altogether, so "# CONFIG_SND_ALOOP is not set" -> "CONFIG_SND_ALOOP=m" 2014-07-14 19:00:08 look like urxvt is broken after recent perl update on Edge, some gdb info: http://pastebin.com/1GqGR2uZ 2014-07-14 20:59:36 melville maybe better to report it on bugs... url 2014-07-15 09:55:11 melville: i'm working on it 2014-07-15 09:55:19 on the perl-5.20 updates 2014-07-15 11:04:11 <^7heo> Moin @ 2014-07-15 11:05:27 <^7heo> I've been told by a friend of mine that the about page of the website (http://alpinelinux.org/about) was featuring a lie :D 2014-07-15 11:05:36 <^7heo> "You can probably get away with just using the tools that are there, although crudely. When combined, tools like sh, awk, sed, and grep can do everything Perl can — seriously." 2014-07-15 11:06:04 <^7heo> This is true in the context of shell scripting and general operating system scripting. 2014-07-15 11:06:25 <^7heo> However, perl can do MUCH more than that, if the context is "writing a complete application". 2014-07-15 11:06:47 either is turing complete, so i fail to see how perl does much more... 2014-07-15 11:06:56 <^7heo> openGL. 2014-07-15 11:07:00 <^7heo> that's just one example. 2014-07-15 11:07:25 <^7heo> So, I think that for the greater good, some clarification would be best: this single sentence is a troll nest in itself. 2014-07-15 11:07:35 <^7heo> just saying. 2014-07-15 11:08:02 you can do opengl from shell script :) 2014-07-15 11:08:11 you can even implement opengl in shell script 2014-07-15 11:08:17 not that it would be a good idea... 2014-07-15 11:08:59 <^7heo> you can actually do opengl with punched cards 2014-07-15 11:09:15 <^7heo> but not in a reasonable timespan (not in one lifespan) 2014-07-15 11:09:18 anyway i think the desired "fix" would be something like 2014-07-15 11:10:39 replacing "everything Perl can" (which while true, is not helpful or practical) with "all the tasks that Perl is typically used for" 2014-07-15 11:11:14 or something similar 2014-07-15 11:11:16 <^7heo> That might do it, but imho, it's still a bit too general. 2014-07-15 11:11:36 <^7heo> the point of this sentence is to state the reason for not using perl in the distro scripts, right? 2014-07-15 11:12:08 let me see... 2014-07-15 11:12:53 the text "although crudely" is already there 2014-07-15 11:13:06 <^7heo> so, whereas perl is, or not, equivalent (in terms of possibilities, not in terms of syntax) to shell/grep/sed/awk, for something other than the distro scripts, is out of scope... 2014-07-15 11:13:18 that makes it clear that using these tools may not produce results as clean/elegant as using a single powerful script lang 2014-07-15 11:13:36 <^7heo> that yes, but it's not the pont. 2014-07-15 11:13:38 <^7heo> point* 2014-07-15 11:13:49 i don't think that's the point of this text 2014-07-15 11:14:17 <^7heo> the point is that perl can do much more, right out of the box, than ash/*sh combined to awk grep, and maybe sed, 2014-07-15 11:14:38 it's just telling you that the base system is very small, but that you can write scripts that run on the base system without needing extra packages like perl installed, if you want 2014-07-15 11:14:49 <^7heo> whatever the point of this text is, it taunts perl users, and therefore, is really a bad idea. 2014-07-15 11:14:57 <^7heo> I agree that perl sucks personally. 2014-07-15 11:15:01 <^7heo> it's a write-only language. 2014-07-15 11:15:17 http://www.etalabs.net/erik-perl.txt 2014-07-15 11:15:17 <^7heo> and more than that, it's useless in most applications. 2014-07-15 11:15:20 :-) 2014-07-15 11:15:52 <^7heo> however, I also think that one should let other people discover by themselves why their favorite language suck 2014-07-15 11:16:21 <^7heo> or they will just defend themselves, and consider themselves right, and will be less likely to evolve. 2014-07-15 11:16:51 <^7heo> it's the whole point of force/taunt people never works; but showing them why, while backed up by facts, works. 2014-07-15 11:17:15 honestly i don't see this as "taunting" perl users. if a perl fan is upset by the (true) statement that you can do anything perl does with lesser tools, even when it's mentioned that such usage may be "crude", i think it's the perl user who's being unreasonable 2014-07-15 11:18:38 <^7heo> it's taunting perl users because it's a wrong fact, which is right within the implied scope; but without that scope explicited. 2014-07-15 11:18:41 <^7heo> Nothing more. 2014-07-15 11:18:55 as far as i can tell, all this text is doing is telling you not to expect the most featureful experience on a minimal install, but that this minimal install can still be used (albeit crudely) to do the same things you would do with more powerful tools 2014-07-15 11:19:17 i'm missing what you claim is factually wrong. 2014-07-15 11:19:17 <^7heo> I just report that here because I have a good friend who loves perl, and told me "that is false, and I feel urged to troll by that sentence." 2014-07-15 11:19:41 it sounds to me like your friend is a troll/perl-fanatic :) 2014-07-15 11:20:03 <^7heo> that whatever general shell (commonly used as a shell in your /etc/passwd file) has the same capabilities without years of work than perl. 2014-07-15 11:20:17 <^7heo> This is true in the context of OS utilies, but this is wrong in the general context. 2014-07-15 11:20:43 <^7heo> anyone knowing perl enough can back that up with facts. 2014-07-15 11:20:55 <^7heo> unfortunately, not liking perl that much, I can't do so. 2014-07-15 11:21:00 "without years of work" did not appear in the statement 2014-07-15 11:21:49 <^7heo> it didn't, because you did say (with lots of bad faith imho) that they are equivalent no matter what. 2014-07-15 11:22:11 fyi it's not me saying it 2014-07-15 11:22:12 <^7heo> so I wanted to add infomation in order to avoid more people trying to make up edge cases 2014-07-15 11:22:26 and i don't represent alpinelinux 2014-07-15 11:22:30 <^7heo> I know 2014-07-15 11:22:34 <^7heo> but still 2014-07-15 11:22:42 <^7heo> I have NO interest in defending perl 2014-07-15 11:22:58 <^7heo> I love alpine, I code in C/shell/python (less for the latter) 2014-07-15 11:23:06 <^7heo> and I really dislike perl. 2014-07-15 11:23:24 obviously some things are easier in one language than another, and some languags have larger existing library corpuses than others 2014-07-15 11:23:33 <^7heo> but from what I know of perl, it is really different from any bash scripting language, both in terms of syntax, features, and everything. 2014-07-15 11:23:45 <^7heo> would it be only for the existing collection of librairies. 2014-07-15 11:24:24 btw the biggest hidden gem in that statement is awk 2014-07-15 11:24:27 <^7heo> and I really am a huge fan of "letting people believe in their own religions" aka. "Please avoid starting religion wars just by taunting other people." 2014-07-15 11:24:32 these days almost nobody even knows awk exists 2014-07-15 11:24:37 ^7heo: I intend to rephrase that 2014-07-15 11:24:42 <^7heo> ncopa: okay :) 2014-07-15 11:24:49 <^7heo> ncopa: can I suggest a rephrasing? 2014-07-15 11:24:59 <^7heo> so you can directly replace it if it suits you... 2014-07-15 11:25:07 but it's comparably expressive to perl for most of the things perl is used for 2014-07-15 11:25:12 sure 2014-07-15 11:25:15 and tiny and universally available on all posix systems 2014-07-15 11:25:23 i actually inted to rewrite the entire page 2014-07-15 11:25:28 simplify it alot 2014-07-15 11:25:54 <^7heo> okay :) 2014-07-15 11:26:08 <^7heo> ncopa: anything I can do to help? 2014-07-15 11:26:40 ncopa, it would probably make a lot of sense to move much of the content to a "history" page or similar 2014-07-15 11:27:04 yeah 2014-07-15 11:27:42 ^7heo: the idea is to make the site content tracked in git and statically generated 2014-07-15 11:28:01 it's interesting but outside the scope of what you want to put in front of someone to convince them to try alpine (or let them quickly determine that it's not for them) 2014-07-15 11:28:09 when that is in place it should be easy to send patches for content 2014-07-15 11:28:15 dalias: exactly 2014-07-15 11:28:27 what i am thinking to put on 'about' page is 2014-07-15 11:28:47 short overview one paragraph 2014-07-15 11:28:51 + 3 sections 2014-07-15 11:29:19 one section for small, one for simple and one for secure 2014-07-15 11:29:29 ## small 2014-07-15 11:29:37 alpine linux is small becaseu blalbla 2014-07-15 11:29:41 ## simple 2014-07-15 11:29:49 alpine linux tries ge out of the way blabla 2014-07-15 11:29:53 ## secure 2014-07-15 11:30:01 grsec+pax \o/ 2014-07-15 11:30:02 alpine linux is secure becase blabla 2014-07-15 11:30:08 and thats it 2014-07-15 11:30:19 pax >_< 2014-07-15 11:30:25 :) 2014-07-15 11:30:35 source of endless breakage :-p 2014-07-15 11:30:50 :P 2014-07-15 11:30:52 as long as exploits breaks... 2014-07-15 11:31:15 i like how top crashes when you switch to the memory view 2014-07-15 11:31:17 and dont forget to mention musl 2014-07-15 11:31:20 no systemd 2014-07-15 11:31:30 top is bb bug 2014-07-15 11:31:39 well the crashing is a bug 2014-07-15 11:31:43 <^7heo> ncopa: the content of the site can be generated with a hook from markdown 2014-07-15 11:31:45 but it can't display any info with pax 2014-07-15 11:31:51 <^7heo> ncopa: and tracked through git 2014-07-15 11:31:56 <^7heo> I can implement/set that for you. 2014-07-15 11:31:58 <^7heo> want it? 2014-07-15 11:32:03 because pax removed all the files from /proc 2014-07-15 11:32:34 ^7heo: that was exactly what i had in mind: work in progress: http://git.alpinelinux.org/cgit/ncopa/mksite-alpine/ 2014-07-15 11:32:39 for "no systemd" to be mentioned there, i think there would need to be a long-term commitment to no-systemd :) 2014-07-15 11:33:08 ncopa, i want to do something similar for the new musl site on libc.org 2014-07-15 11:33:21 <^7heo> ncopa: do youkknow that: http://git.suckless.org/swerc? 2014-07-15 11:33:27 <^7heo> s/youkknow/you know/ 2014-07-15 11:33:42 (all content markdown, makefile to generate static html) 2014-07-15 11:34:06 ^7heo: i was inspired by suckless site indeed 2014-07-15 11:34:16 i saw some other fork of it 2014-07-15 11:34:33 and uggedals page, uggedal.com 2014-07-15 11:34:36 similar concept 2014-07-15 11:34:41 <^7heo> yeah 2014-07-15 11:34:46 <^7heo> in any way 2014-07-15 11:34:49 <^7heo> I have some time right now 2014-07-15 11:34:53 <^7heo> I am willing to help 2014-07-15 11:34:59 <^7heo> so I can do it, if you want. 2014-07-15 11:35:16 i mixed in a lua script for markdown + lustache as template engine 2014-07-15 11:35:18 <^7heo> but if I do it right as it is now, it will be with a github repo, if I code anything. 2014-07-15 11:35:31 <^7heo> hmm 2014-07-15 11:35:33 let me see if i pushed the latest 2014-07-15 11:36:32 <^7heo> what I had in mind is a C markdown parser, piping data to a C webpage generator (which supports templates) 2014-07-15 11:36:54 lua discont is a lua module in C 2014-07-15 11:36:59 <^7heo> but the parts that annoyed me was the piping stuff, as one cannot really get simpler than markdown anyway 2014-07-15 11:37:15 i also want a header with metadata 2014-07-15 11:37:17 <^7heo> so you end up parsing markdown, and piping an equivalent of it... 2014-07-15 11:37:28 <^7heo> what do you call 'header with metadata' ? 2014-07-15 11:37:30 <^7heo> in HTML? 2014-07-15 11:37:30 and i want it generated with gnu make 2014-07-15 11:37:32 no 2014-07-15 11:37:42 <^7heo> bsdmake would be preferable, but it's the same idea. 2014-07-15 11:37:43 <^7heo> :P 2014-07-15 11:37:44 yaml header above the markdown 2014-07-15 11:38:31 <^7heo> ah 2014-07-15 11:38:38 i think make, lua, discount, lustache and yaml is a nice combo 2014-07-15 11:38:39 <^7heo> yeah, that would make sense. 2014-07-15 11:38:45 <^7heo> why not json, btw? 2014-07-15 11:39:05 because it will be written by a human 2014-07-15 11:39:12 <^7heo> json can :P 2014-07-15 11:39:14 <^7heo> unlike xml 2014-07-15 11:39:20 isn't yaml just a gratuitously-incompatible variant of json? 2014-07-15 11:39:29 yaml is nicer for humasn than json 2014-07-15 11:39:29 <^7heo> but okay, yaml is probably more elegant, granted. 2014-07-15 11:39:38 <^7heo> until you forget to indent... 2014-07-15 11:39:52 json is a subset of latest yaml i think 2014-07-15 11:40:07 <^7heo> yeah 2014-07-15 11:40:13 <^7heo> my main problem with yaml is indentation. 2014-07-15 11:40:15 i could have used json, yaml is just slightly nicer 2014-07-15 11:40:24 <^7heo> 'cause people don't necessarily pay attention to that. 2014-07-15 11:40:44 <^7heo> or worse, they don't always see tab/spaces differences. 2014-07-15 11:41:19 my vim config should mark whitepace errors with red 2014-07-15 11:41:34 <^7heo> yeah 2014-07-15 11:41:38 <^7heo> that's a good practice. 2014-07-15 11:41:39 i'll upload my latest wip 2014-07-15 11:41:45 <^7heo> but don't rely on good practices... 2014-07-15 11:41:51 <^7heo> or you will consider C++ to be a sane language... 2014-07-15 11:41:53 i just pushed the git 2014-07-15 11:41:55 <^7heo> ACTION hides 2014-07-15 11:42:09 <^7heo> okay, lemme check it 2014-07-15 11:43:35 http://wwwtest.alpinelinux.org/ 2014-07-15 11:43:41 the generated page 2014-07-15 11:43:56 <^7heo> VERY beautiful. 2014-07-15 11:44:01 <^7heo> I like it. 2014-07-15 11:44:08 the really cool thing about it is that the logo is implemented as a webfont 2014-07-15 11:44:15 <^7heo> works well. 2014-07-15 11:44:20 <^7heo> niice :) 2014-07-15 11:44:20 so it scales nicely on your highres screen 2014-07-15 11:44:27 <^7heo> yeah. 2014-07-15 11:44:31 <^7heo> that's awesome mate. 2014-07-15 11:44:49 <^7heo> I wondered also, maybe it's written in the about page and I missed it but... 2014-07-15 11:44:59 <^7heo> is the "alpine" linux name chosen because of the alps? 2014-07-15 11:45:11 <^7heo> because it originated from a place in/near the alps? 2014-07-15 11:45:12 <^7heo> or? 2014-07-15 11:45:37 Why the Name Alpine? 2014-07-15 11:45:37 "Alpine" originally stood for A Linux Powered Integrated Network Engine. The idea was that the distribution would be focused on networking, and be a tiny "engine" or framework, which larger systems could be built upon. 2014-07-15 11:45:37 Today, "Alpine" is nothing more than a name, despite continuing to live up to its original name. 2014-07-15 11:45:52 at the bottom of about page 2014-07-15 11:46:01 <^7heo> ah 2014-07-15 11:46:01 its just a name basically 2014-07-15 11:46:03 <^7heo> okay 2014-07-15 11:46:08 <^7heo> I was born in the alps 2014-07-15 11:46:12 cool :) 2014-07-15 11:46:16 ncopa, logo implemented as webfont? i don't see that in wwwtest 2014-07-15 11:46:17 <^7heo> so I wanted to know if I had to be proud of something :P 2014-07-15 11:46:20 <^7heo> or not yet :D 2014-07-15 11:46:25 dalias: it is 2014-07-15 11:46:34 you dont see the logo? 2014-07-15 11:47:03 i see the logo 2014-07-15 11:47:05 http://wwwtest.alpinelinux.org/font.css 2014-07-15 11:47:15 hmm is the content in the :before ? 2014-07-15 11:47:19 yup 2014-07-15 11:47:20 i can't highlight it as text 2014-07-15 11:47:28 that's why i assumed it was non-font 2014-07-15 11:47:30 i think its because its inserted with css 2014-07-15 11:47:33 right 2014-07-15 11:48:36 btw is there a reason to do a webfont rather than .svg image? 2014-07-15 11:48:45 IE i suppose 2014-07-15 11:48:47 is it more compatible? 2014-07-15 11:48:51 should be yes 2014-07-15 11:48:53 ah 2014-07-15 11:48:56 nice to know 2014-07-15 11:49:04 i figured it would be less compatible 2014-07-15 11:49:23 <^7heo> using fonts for logos is a very widely used practice. 2014-07-15 11:49:33 <^7heo> and corporate products use it a lot. 2014-07-15 11:49:33 i suppose its also lighter 2014-07-15 11:49:37 <^7heo> yes 2014-07-15 11:49:45 <^7heo> in both terms of bandwith and implementation. 2014-07-15 11:50:17 <^7heo> because the fonts are already implemented for the normal functionality 2014-07-15 11:50:29 <^7heo> so nothing more is necessary (in contrast to implementing SVG rendering for instance) 2014-07-15 11:50:45 you could have the worst of both worlds: svg-font webfonts :-p 2014-07-15 11:50:59 i think that was an option yes 2014-07-15 11:51:08 (most idiotic and useless feature ever :-p) 2014-07-15 11:51:12 :) 2014-07-15 11:51:31 i'd like the 'latest development' be a feed of last git commits i think 2014-07-15 11:51:34 (svg fonts are huge and don't support any of the features necessary for proper typography) 2014-07-15 11:51:44 they dont? 2014-07-15 11:51:59 no, they're basically just an svg image per character 2014-07-15 11:52:07 heh, ok 2014-07-15 11:52:14 so they can't do arabic or indic scripts at all 2014-07-15 11:52:22 sorry for interruption. speaking of perl, seems that urxvt on edge x86_64 is broken after perl update ;) 2014-07-15 11:52:25 nor decomposed korean (or really decomposed anything) 2014-07-15 11:52:33 melville: i pushed fix 2014-07-15 11:52:40 try apk upgrade 2014-07-15 11:52:49 melville: sorry about that... 2014-07-15 11:53:10 great, thanks ncopa and no problem - this is to be expected from edge i guess 2014-07-15 11:53:21 it should not happen too often though 2014-07-15 11:53:35 but we have reserved us the right to risk breakages in edge yes 2014-07-15 11:53:36 btw. is there any reason why snd-aloop isn't build by default as module? 2014-07-15 11:53:48 whats snd-aloop? 2014-07-15 11:54:05 alsa loopback driver 2014-07-15 11:54:26 kernelconfig.x86_64:# CONFIG_SND_ALOOP is not set 2014-07-15 11:54:37 probably nobody ever asked for it 2014-07-15 11:54:46 i can enable it if you want/need it 2014-07-15 11:54:47 yeah, I have seen it and wonder why it isn't "CONFIG_SND_ALOOP=m" 2014-07-15 11:54:54 ncopa, I was thinking if on wwwtest.a.o "download latest" could be moved to top right area of content ? 2014-07-15 11:55:20 yeah, without hardware capture mix support for soundcard it is hard to record audio from desktop without alsa loopback 2014-07-15 11:55:37 melville: I'll enable it right away for you 2014-07-15 11:55:50 what it does it redirects audio through loopback and then to soundcard, so loopback can be used as capture device 2014-07-15 11:55:56 in a box , while the 'top left' has more space for texts 2014-07-15 11:56:18 2/3 - 1/3 ratio 2014-07-15 11:56:24 vkrishn: good point, i'll think of it 2014-07-15 11:56:36 and let the text float around it 2014-07-15 11:56:47 i originally wanted it centered 2014-07-15 11:56:55 thanks again 2014-07-15 11:57:27 <^7heo> ncopa: I shall look in your work. Would I have any contribution to make, should I send git patches? 2014-07-15 11:57:35 yes please 2014-07-15 11:57:37 <^7heo> okay 2014-07-15 11:57:49 the current 'puch line' can l be centered on top left 2014-07-15 11:57:55 punch* 2014-07-15 11:58:05 i could put it on github if it makes things easier 2014-07-15 11:58:23 i havent put it on github because i was not too happy with the mksite-alpine project name 2014-07-15 11:58:28 but havent found any better either 2014-07-15 11:59:46 I have the svg original for the logo, the icons are from font awesome and i use icomoon web service to generate the webfont 2014-07-15 11:59:55 no "contact link" ? or same as 'community' ? 2014-07-15 12:00:10 the idea was to put it under community yes 2014-07-15 12:00:18 ok 2014-07-15 12:00:36 <^7heo> ncopa: I am thinking of a name 2014-07-15 12:00:42 <^7heo> lemme think a bit more about it :P 2014-07-15 12:00:47 <^7heo> and then I shall tell you :P 2014-07-15 12:01:02 got inspiration from here: https://github.com/clehner/mksite 2014-07-15 12:01:13 he was also inspired from suckless 2014-07-15 12:01:29 how about a powered by alpinelinux image at bottom ? 2014-07-15 12:01:39 since the site would run on AL 2014-07-15 12:02:02 heh, we could probably use busybox httpd to serve it :) 2014-07-15 12:02:11 wwwtest runs darkhttpd 2014-07-15 12:03:00 ^7heo: i want minimal with javascript, but I am ok with some 2014-07-15 12:03:19 been thinking of using discuss for enable comments (like nikola does) 2014-07-15 12:03:20 +1 for darkhttpd :) 2014-07-15 12:04:59 <^7heo> ncopa: I come from the suckless comunity, btw. 2014-07-15 12:05:09 ^7heo: oh! nice! welcome! :) 2014-07-15 12:05:20 <^7heo> thanks :) 2014-07-15 12:05:26 <^7heo> I didn't really contribute to anything there. 2014-07-15 12:05:35 <^7heo> but I respect their work/philosophy very much 2014-07-15 12:05:38 the suckless wiki did impress me 2014-07-15 12:05:48 yes, they do lots of good stuff 2014-07-15 12:05:53 <^7heo> and I'd be okay to contribute to something here (it's less... intimidating, to me) 2014-07-15 12:06:02 :) 2014-07-15 12:07:07 i can setup hooks so script can auto-regenerate on git push 2014-07-15 12:07:23 both for website content (eg news...) 2014-07-15 12:07:26 and for git commits 2014-07-15 12:08:04 twitter, etc links in - community ? 2014-07-15 12:09:30 yes, thats needed 2014-07-15 12:09:44 i havent done anything serious with the content yet 2014-07-15 12:10:02 work has mostly been on the scripts, makefile, and stylesheet 2014-07-15 12:10:22 i'd like to have the package browser there too 2014-07-15 12:10:29 generated from makefile 2014-07-15 12:17:38 and the current build with blinky images ;) 2014-07-15 12:18:10 and scrolling current pkg being build 2014-07-15 12:18:48 worldmap with blinking dot as who pushed from where ;))) 2014-07-15 12:19:39 some generated build status/report/stats 2014-07-15 12:23:15 build server status i want other location 2014-07-15 12:24:07 dev.a.o / build.a.o ? 2014-07-15 12:24:56 build[n].a.o being particular server 2014-07-15 12:25:05 while build.a.o common portal 2014-07-15 12:42:35 yeah 2014-07-15 12:42:38 i like that 2014-07-15 12:43:09 do we want the development log on front page be implemented as javascript? 2014-07-15 12:43:16 eg like this: https://github.com/uggedal/uggedal.com/blob/master/static/feed.js 2014-07-15 12:43:30 or do we want regenerate the index.html page for every git commit 2014-07-15 12:44:19 <^7heo> re-generating an html page isn't a big problem, is it? 2014-07-15 12:44:25 no 2014-07-15 12:44:33 thats probably simpler too 2014-07-15 12:44:38 easier to implement 2014-07-15 12:44:39 hm 2014-07-15 12:44:56 <^7heo> and using js for ANYTHING but cosmetic effects is a bad thing imho. 2014-07-15 12:45:07 development log - only in form of who pushed what pkg 2014-07-15 12:45:17 <^7heo> having js off shouldn't change anything about the availability of the data. 2014-07-15 12:45:20 may linked to cgit commit 2014-07-15 12:45:45 i was thinking only publish commit message 2014-07-15 12:45:47 the subject 2014-07-15 12:45:51 and possibly the date 2014-07-15 12:46:28 build logs shows a server image with blinking status with commit status/message running in a box below 2014-07-15 12:46:28 currently http://alpinelinux.org/ shows the last 4 packages in x86 2014-07-15 12:46:40 and may be linked to elaborate resource 2014-07-15 12:48:13 last 4pkg is on home page is different thing 2014-07-15 12:48:37 yes build server status is other thing 2014-07-15 12:48:49 i'm talking about 4 pkg on home page 2014-07-15 12:48:59 when one re-visits(refresh/clean cookie) build.a.o the history is not seen 2014-07-15 12:49:00 i'd like replace it with 4 or 5 last commits 2014-07-15 12:49:23 but scrolls from current state 2014-07-15 12:49:26 build.a.o can wait for now 2014-07-15 12:49:29 ok 2014-07-15 12:52:17 the current 4 pkg does not show status for other arch 2014-07-15 12:52:39 maybe just pkg name and when one moves mouse over it gets more info 2014-07-15 12:52:45 <^7heo> ncopa: what about using RSS/atom for "streams" of information? 2014-07-15 12:52:59 that way more pkg names can fit in less space 2014-07-15 12:55:02 I think both last 4/5 commits and last 4/5 pkg in compact box can be done, that way one knows how far is last pkg from current commit 2014-07-15 12:56:01 <^7heo> ncopa: how about "miocene" for your alpine-site generation software? 2014-07-15 12:56:01 means date stamp needs to be there 2014-07-15 12:57:05 miocene hmm 2014-07-15 12:57:15 <^7heo> it's when the alps did form. 2014-07-15 12:57:34 <^7heo> I wanted to come up with the name of the alps-formation event... 2014-07-15 12:57:40 <^7heo> but it appaears that there isn't any. 2014-07-15 12:57:42 cool 2014-07-15 12:57:48 :) 2014-07-15 12:57:56 <^7heo> so miocene is the closer I can get to "alps formation" 2014-07-15 12:58:02 <^7heo> and if you see the alps as a site... 2014-07-15 12:58:15 <^7heo> then, "the alpine site did form during miocene" is correct :P 2014-07-15 12:58:17 box maybe titled 'latest build news' 2014-07-15 12:58:31 basically, it will be yet another general site generator then 2014-07-15 12:58:46 <^7heo> meh 2014-07-15 12:59:04 <^7heo> why not do a standard markdown-to-XML generator then? 2014-07-15 12:59:22 vkrishn: i think instead of latest build we show latest git activity 2014-07-15 12:59:25 <^7heo> 'cause then it's not "another general site generator", it's git hook scripts that use a markdown to XML engine 2014-07-15 12:59:45 we need some special feature i think 2014-07-15 12:59:51 for news index box 2014-07-15 12:59:58 and git log box 2014-07-15 13:00:04 and package browser 2014-07-15 13:00:26 i wonder if its worth it 2014-07-15 13:00:39 so "latest development news" 2014-07-15 13:00:39 <^7heo> ncopa: or "techtonics" also 2014-07-15 13:00:45 to separate out the site generator scripts from the data, the actual site 2014-07-15 13:00:58 <^7heo> (techtonics have been the name of the process of the alps creation :)) 2014-07-15 13:01:02 vkrishn: exactly 2014-07-15 13:01:10 <^7heo> (but it's not really directly related) 2014-07-15 13:01:26 <^7heo> or "plate techtonics" :P 2014-07-15 13:01:46 ^7heo: those names sounds like good candidates for abuild/aports 2014-07-15 13:01:57 the scripts that builds the distro, not the site 2014-07-15 13:02:22 <^7heo> I understood :P 2014-07-15 13:02:30 if we want keep the scripts bundled with the actual site, then i'd say alpine-site is good enough :) 2014-07-15 13:02:42 <^7heo> yeah 2014-07-15 13:02:53 if we want do it as a yet another static site generator tool 2014-07-15 13:03:01 then those names are to be considered 2014-07-15 13:03:04 good names 2014-07-15 13:03:13 <^7heo> :) 2014-07-15 13:03:44 maybe we just bundle the scripts with the content for now 2014-07-15 13:03:47 as a start 2014-07-15 13:03:53 makes it easier to get started 2014-07-15 13:04:04 or maybe name it "alfrost" 2014-07-15 13:04:11 frostal 2014-07-15 13:04:52 miocene - sounds medicine 2014-07-15 13:05:12 <^7heo> it's not a medical term, it's a geological one. 2014-07-15 13:05:12 <^7heo> :) 2014-07-15 13:05:17 :) 2014-07-15 13:05:42 <^7heo> and it's more or less "obvious" to people who did study at least a bit of geology. 2014-07-15 13:06:30 or maybe 'bigfoot' 2014-07-15 13:06:37 <^7heo> http://dictionary.reference.com/browse/-cene 2014-07-15 13:07:58 who perhaps lives in hidden caves in Alpine 2014-07-15 13:08:07 alsite 2014-07-15 13:08:15 or alpine-site 2014-07-15 13:08:18 works for now 2014-07-15 13:08:37 <^7heo> right 2014-07-15 13:08:49 <^7heo> it's more important to have working stuff than names. 2014-07-15 13:08:53 yeah 2014-07-15 13:09:27 :) 2014-07-15 13:10:04 freezals 2014-07-15 13:10:28 i just realised alpine linux isn't named after where its developers live 2014-07-15 13:10:49 <^7heo> well 2014-07-15 13:11:08 <^7heo> I was wondering for some time. 2014-07-15 13:11:13 <^7heo> but I can say that I'm not surprised. 2014-07-15 13:11:24 devs lives different places 2014-07-15 13:11:35 <^7heo> People in the alps can get very good, technically, but 99.9999% of them are dumbasses. 2014-07-15 13:11:49 <^7heo> so that doesn't leave much people for making a linux distro :D 2014-07-15 13:11:55 since your name sounded italian to me, i just assumed we're talking italian alps and never doubted it 2014-07-15 13:12:20 we have some italian devs 2014-07-15 13:12:28 some in finland 2014-07-15 13:12:32 some in norway 2014-07-15 13:12:38 netherlands 2014-07-15 13:12:42 usa 2014-07-15 13:13:39 :) 2014-07-15 13:14:14 did you get a chance to look into wifi? i just recently had the first major netctl disaster in my life, so i'm much closer to do doing something about it now 2014-07-15 13:15:48 ovf: i havent looked at wifi recently 2014-07-15 13:15:57 but i'm open to ideas 2014-07-15 13:17:42 setup-wifi script would be amazing (but I bet it would be a nightmare to get working with so many different wifi adaptors about) 2014-07-15 13:18:07 btw ncopa dhcpcd-gtk does not seem to be working for me... attempting to connect to an ap with it does nothing, but it does show the status of connection to existing known aps 2014-07-15 13:18:08 we coudl do wpa_cli something 2014-07-15 13:18:16 any idea where i can get some support for it? 2014-07-15 13:18:44 dalias: not really, you might need enable write to config file in wpa_supplicant.conf 2014-07-15 13:18:53 but i havent really been able to get it working either 2014-07-15 13:19:01 i manage the config manually 2014-07-15 13:19:16 which is a pain if you have many different wifi nets... 2014-07-15 13:19:32 <^7heo> ovf: my name isn't italian 2014-07-15 13:19:36 ncopa, write is enabled 2014-07-15 13:19:38 <^7heo> but I know a bit of italian. 2014-07-15 13:20:10 last i checked the dhcpcd-ui site didn't even have contact info for developer, much less a list, but i could probably find it in the source... 2014-07-15 13:20:30 i was just wondering if you know any community using it that might be helpful for support 2014-07-15 13:20:43 its this guy: http://roy.marples.name/blog/ 2014-07-15 13:20:51 i dont know any community though 2014-07-15 13:20:59 i dont know if anyone except us use it either :) 2014-07-15 13:21:03 ^7heo: i meant ncopa's 2014-07-15 13:21:15 <^7heo> ah okay :) 2014-07-15 13:21:28 http://roy.marples.name/projects/dhcpcd-ui/index 2014-07-15 13:22:22 dalias: seems like it is discussed on dhcpcd mailing list 2014-07-15 13:33:04 btw, dhcpcd/dhcpcd-ui guy is the same guy who gave us openrc 2014-07-15 13:33:46 oh? kinda surprising that openrc guy is using dbus, no? :-p 2014-07-15 13:34:48 yeah 2014-07-15 13:34:49 well 2014-07-15 13:34:51 <^7heo> KILL WITH FIRE! 2014-07-15 13:34:58 <^7heo> sorry, I just read "dbus:\ 2014-07-15 13:35:01 <^7heo> damn 2014-07-15 13:35:04 <^7heo> "dbus"* 2014-07-15 13:35:06 Hello everyone 2014-07-15 13:35:12 <^7heo> moin dvmacias 2014-07-15 13:35:17 im kinda ok with dbus for ui stuff... 2014-07-15 13:36:33 <^7heo> I don't see the point of it. 2014-07-15 13:38:42 Was wondering how to use rcp in alpine. (Alpine Linux2.6) 2014-07-15 13:38:50 ran apk add heimdal-rcp 2014-07-15 13:39:15 and heimdal-libs 2014-07-15 13:39:20 rcp? isnt scp the way to go? 2014-07-15 13:39:26 i have never used rcp 2014-07-15 13:39:30 yeah i know but for a test 2014-07-15 13:39:57 heimdal rcp is rcp with kerberos support 2014-07-15 13:40:19 i dont know if we even provide normal rcp 2014-07-15 13:40:30 ^7heo: don't go that route, you'll start contemplating the point of living not before long... 2014-07-15 13:40:42 hehe 2014-07-15 13:40:45 oh ok 2014-07-15 13:40:54 <^7heo> ovf: I have, already... 2014-07-15 13:41:13 <^7heo> remember, I come from suckless. 2014-07-15 13:41:22 <^7heo> and I have a dog, so I'm still alive. 2014-07-15 13:41:23 basically i wanted to see scp progress output 2014-07-15 13:41:35 was trying cv but not in alpine 2014-07-15 13:42:00 scping and image to a switch 2014-07-15 13:42:08 an image* 2014-07-15 13:46:23 well thanks guys for curteous response. take care 2014-07-15 13:51:07 hmm it seems dhcpcd-ui "works" in that it adds the network i select and sets the key for it 2014-07-15 13:51:27 however it doesn't tell wpa_supplicant to switch to that network 2014-07-15 13:56:10 sounds like a bug to me 2014-07-15 14:05:14 i get the feeling wpa_supplicant doesn't like to be told what network to use, tho. 2014-07-15 14:05:41 it just wants to automatically hop on the most-preferred one or whatever (which, short of manual configuration, is just whichever one you added first...) 2014-07-15 14:23:44 how does networkmanager and others do it? 2014-07-15 14:26:36 nm does not communicate with an existing wpa_supplicant daemon 2014-07-15 14:26:59 it does everything itself and invokes wpa_supplicant with a per-ap config once it's decided on a network to connect to 2014-07-15 14:27:08 hm, ok 2014-07-15 14:27:26 that might be the reason why they do it that way 2014-07-15 14:28:12 also with dhcpcd-ui, when i make changes via the gui, wpa_supplicant begins a new scan and doesn't do anything until the scan finishes (5+ seconds later) 2014-07-15 14:28:26 rather than taking some immediate action 2014-07-15 14:28:37 however, i'm 99% sure wpa_supplicant can be used correctly 2014-07-15 14:28:42 because android seems to do so 2014-07-15 14:28:48 aha 2014-07-15 14:29:26 i suppose sending an email to dhcpcd-discuss might be the way to go 2014-07-15 14:40:44 btw this topic is very interesting to me not just for making my alpine laptop more practical, but also for the "new platform" plan 2014-07-15 14:41:08 where sane auto and manual network configuration is one of the big wanted items 2014-07-15 14:41:37 like for new init system 2014-07-15 14:41:44 and/or network config 2014-07-15 14:43:47 i have been thinking of triggering service start on ip addr found, any/inet4/inet6 2014-07-15 14:44:29 for the boot system/supervision suite 2014-07-15 14:44:53 ? 2014-07-15 14:45:06 like, a service you wnat running 2014-07-15 14:45:08 like httpd 2014-07-15 14:45:16 ah 2014-07-15 14:45:29 i'm thinking of different problems mainly from a client perspective 2014-07-15 14:45:33 you could say, start this as soon there shows up a valid any addr 2014-07-15 14:45:37 ok 2014-07-15 14:45:56 i've been thinking of init system recently :) 2014-07-15 14:46:11 s6 has some really interesting stuff 2014-07-15 14:46:30 like what happens when your browser happens to restart at the moment you're getting connected to some shitty access point with a login screen and all your tabs get replaced with login screens 2014-07-15 14:46:55 lol 2014-07-15 14:47:45 how does andoid handle that? 2014-07-15 14:47:49 they don't 2014-07-15 14:47:53 it happens on android :( 2014-07-15 14:48:16 sounds like a difficult problem to solve 2014-07-15 14:48:16 the most graceful solution would be virtualizing the user-visible network interface/addresses and only connecting them up to the physical interface once it's safely logged in and ready 2014-07-15 14:48:41 actually, sounds like it is a problem that shouldnt be there in first place 2014-07-15 14:48:53 proper thing to do would be fix acess point 2014-07-15 14:48:55 but i think you could also solve it with firewall rules that drop all packets except ones from a special config-process uid or pid until the network is up 2014-07-15 14:50:05 yes but good luck with that 2014-07-15 14:50:42 every municipality/coffee shop/mall/retail store/university/etc. wants to redirect you to their login crap, even if access is free and open 2014-07-15 14:51:07 so they can make you click through some meaningless agreement and/or view some ads 2014-07-15 14:51:45 imo an ideal system would not only get you past these logins without them being able to mess up your browser state... 2014-07-15 14:52:13 ...it would also automatically connect your vpn (if you have one configured and request this feature) when it detects a crap network 2014-07-15 14:52:45 what happens when you refresh the page after logged in? 2014-07-15 14:52:46 so that you'd be safe from ongoing content manipulation/ad injection/etc. 2014-07-15 14:53:30 normally the pre-login mode on the access point sends you an http three-oh-whatever to redirect you to the login page in such a manner that your original location is lost 2014-07-15 14:53:42 'good' ones redirect you back after login 2014-07-15 14:54:02 but most redirect you to some useless site for the provider of the access point 2014-07-15 14:54:08 (with more ads :) 2014-07-15 14:54:44 for some uses you just don't connect to random low-quality access points 2014-07-15 14:55:03 but i'm thinking of things like phones where you might want to not have a voice plan at all, and use voip over whatever access points are available 2014-07-15 14:55:36 if we're lucky, this kind of crap _might_ end somewhere in the next 3-5 years 2014-07-15 14:56:07 with most major sites shifting to forcing https, it's no longer possible to do these content manipulations 2014-07-15 14:56:31 and the user experience is severely hindered by users never seeing the login page (if none of the sites they access are plain http) 2014-07-15 14:56:38 and thus not being able to login 2014-07-15 14:56:58 my guess is that they won't abandon this login nonsense, but instead it might shift to a standard protocol 2014-07-15 14:57:22 where instead of redirecting all http, the browser detects a non-working connection and tries a special standardized http url for access-point login 2014-07-15 14:58:50 that sounds like the proper solution yes 2014-07-15 14:59:43 a standardized login url that the browser tries if connection fails, beofer giving final error message 2014-07-15 14:59:54 the more likely solution: phones come preinstalled with a rogue CA for access points to hijack your https... :( 2014-07-15 15:00:31 ok, i gotta go 2014-07-15 15:00:34 ok 2014-07-15 15:00:37 later 2014-07-15 15:00:38 i subscribed to dhcpcd-discuss ml 2014-07-15 15:00:53 i'm interested if you find a solution for wpa_supplicant 2014-07-15 15:01:00 *nod* 2014-07-15 15:01:08 i'll probably get involed discussing it there later 2014-07-15 15:01:08 c u 2014-07-15 15:01:13 going to work on some other things first tho 2014-07-16 10:40:10 looks like upcoming version of mdocml will provide full-blown package for manpages: apropos, man, makewhatis with dependency on sqlite3 2014-07-16 10:41:15 and also web interface written in C, how it looks now: http://www.openbsd.org/cgi-bin/man.cgi 2014-07-16 11:14:56 nice 2014-07-16 11:16:46 maybe we should have an man.alpinelinux.org site? 2014-07-16 11:17:29 we could apk add *-doc or similar 2014-07-16 11:17:45 I was thinking about it - even started writing mail ;) 2014-07-16 11:18:17 i suggest we wait til next mdocml is out 2014-07-16 11:18:57 mdocml should stabilize soon I think, since they are getting ready for OpenBSD 5.6 release. 2014-07-16 11:19:18 i dont want start any new projects til we have closed some already running 2014-07-16 11:19:29 like new website, build server monitor site 2014-07-16 11:21:50 There is no hurry. Just since Apline Linux doesn't provide man pages by default (makes sense, since it is targeting small devices like routers on top of other things) having acces to them via web interface would be nice. 2014-07-16 11:22:28 indeed 2014-07-16 11:46:52 hum syslinux appears to be broken 2014-07-16 11:47:28 hm, no 2014-07-16 11:47:52 but bad things happens if you syslinux /dev/sda1 while still mounted apparently 2014-07-16 11:48:53 hi 2014-07-16 11:56:48 hi 2014-07-16 12:10:15 my alpine install is fucked up :( 2014-07-16 12:11:34 my sda remounts itself in RO all the time, because of I/O errors :/ 2014-07-16 12:15:04 bad disk? 2014-07-16 12:15:18 bad everything 2014-07-16 12:15:25 ── pgrep vim 2014-07-16 12:15:25 Bus error 2014-07-16 12:15:38 /bin/sh: /usr/bin/markdown: I/O error 2014-07-16 12:15:58 sounds like your disk is dying 2014-07-16 12:16:15 yeah 2014-07-16 12:16:18 :/ 2014-07-16 12:16:18 does dmesg show errors 2014-07-16 12:16:23 yep 2014-07-16 12:16:26 I/O errors 2014-07-16 12:16:28 everywhere 2014-07-16 12:16:40 it just vomits I/O errs 2014-07-16 12:17:26 I'll just leave it alone until I'm bakc home 2014-07-16 12:17:45 And I'll fsck it (twice) 2014-07-16 12:17:46 smartmontools? to check health of hdd and run S.M.A.R.T. self-tests if possible? 2014-07-16 12:17:57 And start reinstalling 2014-07-16 12:18:02 melville, what's that ? 2014-07-16 12:19:01 http://en.wikipedia.org/wiki/S.M.A.R.T. 2014-07-16 12:21:25 there is also graphical inteface for smartmontools called GSmartControl, but it isn't packaged for Alpine. 2014-07-16 12:30:05 does it work with mounted disks ? 2014-07-16 12:32:47 yes. S.M.A.R.T. is provided by most disks right now I think and it is done at lower level. good way to diagnose if you have problems with hardware. 2014-07-16 12:37:29 Oh, yeah I see 2014-07-16 12:37:50 my drive is like, 7 yo 2014-07-16 12:38:04 So I think it's just dying 2014-07-16 12:38:58 if you run short self check for smart and it ends with errors, where is high chance of that. 2014-07-16 12:39:55 takes about 2 minutes. extended check can take up to few hours. 2014-07-16 12:41:51 will the data still be available during the check ? 2014-07-16 12:42:07 if so, I'll just run it now, and check the logs tonight 2014-07-16 12:43:22 ah, it's a daemon and all :/ 2014-07-16 12:45:54 I don't think you need smartd - it is intented for continous monitoring when needed 2014-07-16 12:46:20 https://wiki.archlinux.org/index.php/S.M.A.R.T. 2014-07-16 12:46:48 so guidance 2014-07-16 12:46:49 404 2014-07-16 12:47:04 ah, missed last dot 2014-07-16 12:52:10 I just tried "sudo smartctl -t short /dev/sda", waited a minute and then "sudo smartctl -l selftest /dev/sda", result: 2014-07-16 12:52:30 # 1 Short offline Completed without error 2014-07-16 13:15:55 I hope 'ill be able to complete the installation 2014-07-16 13:42:04 installation? you can "setup-interfaces" from alpine cd and then "apk add " afaik 2014-07-17 06:40:18 hi 2014-07-17 06:40:45 i'm having this weird problem trying to boot the xen dom0 image from usb 2014-07-17 06:41:47 so far i've tried several unetbootin versions (supposedly different syslinux releases) 2014-07-17 06:42:35 and whenever i try to boot it says "mboot.c32: not a COM32R image" 2014-07-17 07:02:06 bvcx: hi 2014-07-17 07:03:31 if i want to boot from usb, I just format the usb as fat32 and then copy the release to it. then you need to install the syslinux bootloader manually and maybe clear the mbr. 2014-07-17 07:27:43 mboot.c32 error message is likely due to syslinux version mismach 2014-07-17 07:27:58 copy the .c32 files from same syslinux version you are using 2014-07-17 07:28:23 please note that hvmloader is still broken with v3.0 2014-07-17 08:12:20 ncopa: 2014-07-17 08:12:22 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=3cf521f7dc87c031617fd47e4b7aa2593c2f3daf 2014-07-17 08:12:37 CVE-2014-4943: Linux privilege escalation in ppp over l2tp sockets 2014-07-17 08:15:52 kocka: what 3.10/3.14 kernel versions includes the fix? 2014-07-17 08:17:43 i dont think its already released 2014-07-17 08:55:48 clandmeter: righto i installed the standard image without unetbootin and built other stuff on that 2014-07-17 08:57:52 ncopa: thanks. seemed like it 2014-07-17 09:00:22 ncopa: why is /boot/syslinux.cfg all f'ed up in the xen iso? theres a few short lines of nonsense followed by a long line, no sensible text at all 2014-07-17 09:01:36 ncopa: but the file is fine in the standard iso 2014-07-17 09:03:19 its a bug 2014-07-17 09:03:54 http://bugs.alpinelinux.org/issues/3137 2014-07-17 09:04:16 basically, it me who have done something stupid in the setup-bootable script 2014-07-17 09:04:29 and have been "lucky" that its gone undiscovered for years... 2014-07-17 09:05:14 i wonder if it is the new kernel or musl that makes the bug bubble up to the surface 2014-07-17 09:11:54 right, right. that happened after i did setup-bootable from loop 2014-07-17 09:22:32 i am working on a fix as we speak 2014-07-17 09:23:59 ncopa thanks :) 2014-07-17 09:25:12 thanks! 2014-07-17 10:10:21 ncopa "to clarify, this is a bug in setup-bootable. It should run run syslinux on a mounted device, which it currently does. It is pure 'luck' that this has previously worked." <- it should NO run syslinux on a mounted device? 2014-07-17 10:10:25 erm NOT 2014-07-17 10:43:05 ncopa: https://twitter.com/grsecurity/status/489717185992855552 2014-07-17 10:51:38 spender really sucks at this, sperging about silent fixes 1day after its cved 2014-07-17 11:11:44 his communication skills suck badly, but i think it's good to slap the kernel devs when their practices turn towards sh--sh--sh 2014-07-17 11:14:06 darkfader: Nah, this is outright queen drama. The CVE has been issued almost at the same time the fix landed, can't really say how is this bad. 2014-07-17 11:14:33 katlogic: because they're trying to keep it out of the commit log for ego reasons 2014-07-17 11:14:55 i've worked with a few devs and it's a pattern really 2014-07-17 11:15:14 even if there's a cve they'll still try to not let it show in the log 2014-07-17 11:15:21 Doubtful. Whatever gets in the commit stays in the commit, there are not "they". Just the author. 2014-07-17 11:16:03 and you don't see the difference if the commit msg says "fix for cve xxxx" or doesn't? 2014-07-17 11:16:15 like, if you look at 2000 patches for, i.e. making release notes 2014-07-17 11:16:31 sometimes they put the cve in the stable branches 2014-07-17 11:16:31 darkfader: You already know the answer. The CVE is result of many eyes scouring every commit for bugs. 2014-07-17 11:16:43 When it is inverse, the CVE is usually mentioned in the commit. 2014-07-17 11:16:52 yes 2014-07-17 11:17:17 True silent fix is, when someone points out exploitable condition, it is silently fixed and no reported cve for months. 2014-07-17 11:17:33 That usually signs negligence, its not that often (last time in 2012 i think?) 2014-07-17 11:17:57 i think upstream dont want spend time in investigating "is this bug worth a CVE or not" and let others file CVEs 2014-07-17 11:18:02 i dont blame them for that 2014-07-17 11:18:40 I think spender is just mad because this bug is infoleak as well, thus readily exploitable on grsec :> 2014-07-17 11:19:16 i still havent seen any CVE for bugs introduced by pipacs/spender either... 2014-07-17 11:19:27 i believe there have been atleast one 2014-07-17 11:19:41 that potensially could be exploitable 2014-07-17 11:20:00 anyways 2014-07-17 11:20:09 Nah, spender have almost zero knack for finding bugs, he's however very good at writing exploits (and plugging the usual routes). 2014-07-17 11:20:25 its good when the CVE gets mentioned in git log 2014-07-17 12:05:35 <^7heo> ncopa: about that markdown parser stuff 2014-07-17 12:05:54 hi 2014-07-17 12:06:13 <^7heo> ncopa: do you want to continue with your small lua script which is reusing 3 other components, or do you think that a single markdown-to-HTML transformer would be preferrable? 2014-07-17 12:06:17 <^7heo> hi btw :D 2014-07-17 12:06:36 <^7heo> 'cause I've been wanting to implement such a thing for a long time 2014-07-17 12:07:07 <^7heo> with flags to add meta and other tags, and support for templates. 2014-07-17 12:07:30 hm 2014-07-17 12:07:36 how do you intend implement it? 2014-07-17 12:07:50 with shell/awk? Lua? Python? 2014-07-17 12:08:12 no external deps? like i do, with lua-discount, lustache and lua-yaml 2014-07-17 12:08:34 i think those 3rdparty deps are small enough 2014-07-17 12:08:41 they are not like 40MB python... 2014-07-17 12:08:52 <^7heo> in C 2014-07-17 12:09:01 <^7heo> fully in C. 2014-07-17 12:09:07 <^7heo> but brb, gotta go for a few minutes 2014-07-17 12:09:10 depend on discount? yaml? 2014-07-17 12:09:21 i doubt its worth the time to write it in C... 2014-07-17 12:09:26 <^7heo> no depends, but will maybe reuse code. 2014-07-17 12:09:28 <^7heo> brb 2014-07-17 12:11:46 oh there is such tool already 2014-07-17 12:11:50 makepage 2014-07-17 12:11:53 from discount 2014-07-17 12:11:59 https://github.com/Orc/discount/blob/master/makepage.1 2014-07-17 12:12:57 the thing is, i think i want split up things like front page 2014-07-17 12:13:10 and let parts of it be generated by something else 2014-07-17 12:13:14 like the git log 2014-07-17 12:13:21 and download release 2014-07-17 12:13:55 <^7heo> back 2014-07-17 12:15:18 and theme 2014-07-17 12:15:30 this seems to do what you want: https://github.com/Orc/discount 2014-07-17 12:15:43 i'm ok with using the lua modules 2014-07-17 12:15:54 <^7heo> I know discount, but... I wanted to re-implement mine 2014-07-17 12:16:00 <^7heo> in a different way (technically) 2014-07-17 12:16:06 <^7heo> but following the same idea. 2014-07-17 12:16:22 ok... 2014-07-17 12:16:26 as lib too? 2014-07-17 12:16:31 <^7heo> no 2014-07-17 12:16:35 <^7heo> just as a binary 2014-07-17 12:16:46 <^7heo> I wanted to make it very small 2014-07-17 12:16:52 <^7heo> and not use any XML library. 2014-07-17 12:17:19 <^7heo> I actually wanted to allow people to define their own output format. 2014-07-17 12:18:15 <^7heo> markdown is a very simple format, but the problem is supporting html in it... 2014-07-17 12:18:25 <^7heo> (for stuff such as tables, etc.) 2014-07-17 12:18:40 hm 2014-07-17 12:18:41 yes 2014-07-17 12:18:55 <^7heo> 'cause discounts supports that afaik. 2014-07-17 12:19:10 <^7heo> but I don't really think that supporting XML-based extensions is a good idea at all. 2014-07-17 12:19:13 i am thinking to maybe implement parts of generated page in not markdwn 2014-07-17 12:19:15 html only 2014-07-17 12:19:23 <^7heo> html is XML-based... 2014-07-17 12:19:35 the front page does not really make sense to implement in markdown 2014-07-17 12:19:43 <^7heo> (even if it's derived from it more or less far, considering the HTML norm you consider) 2014-07-17 12:19:53 <^7heo> (depending on the HTML norm you consider, I meant) 2014-07-17 12:20:10 <^7heo> hmm 2014-07-17 12:20:36 <^7heo> yeah right 2014-07-17 12:21:23 <^7heo> and it would be a chore to describe that in a simple way 2014-07-17 12:21:47 <^7heo> I mean, it's basically 3 "frames" 2014-07-17 12:22:25 <^7heo> and I like it MUCH more than the "page based" counterpart... 2014-07-17 12:22:37 <^7heo> however... 2014-07-17 12:22:41 <^7heo> css can make this better 2014-07-17 12:22:56 <^7heo> It's already what you're using actually, I didn't realize. 2014-07-17 12:23:12 <^7heo> so it's possible to define the front page in markdown, and it's even nicer, imho. 2014-07-17 12:24:42 <^7heo> but, yeah, it can also be interesting to consider having a first page which is dynamically generated... 2014-07-17 12:25:45 <^7heo> (which can be done with your markdown -> html lua script) 2014-07-17 12:25:52 im not overly happy with not beeing able to mix in html 2014-07-17 12:26:07 <^7heo> yeah I can understand that. 2014-07-17 12:26:14 i will need import current release notes 2014-07-17 12:26:16 from drupal 2014-07-17 12:26:19 <^7heo> but markdown should be enough for 99% of the uses. 2014-07-17 12:26:24 they are all written in html 2014-07-17 12:26:28 <^7heo> yeah I know 2014-07-17 12:26:45 <^7heo> markdown is a very nice idea; but it's incomplete. 2014-07-17 12:26:50 so it would be nice to just import them as news-date.md 2014-07-17 12:26:53 yes, i figured 2014-07-17 12:27:00 asciidoc is more complete 2014-07-17 12:27:08 can generate tables and all 2014-07-17 12:27:10 <^7heo> yeah but less attracting imho 2014-07-17 12:27:33 asciidoc is nice for manpages and documentation 2014-07-17 12:27:39 <^7heo> yeah 2014-07-17 12:27:47 <^7heo> but I don't find it as readable as markdown 2014-07-17 12:27:49 but for our news posts markdown is just find 2014-07-17 12:27:50 yes 2014-07-17 12:27:51 agree 2014-07-17 12:28:00 <^7heo> especially for the headers. 2014-07-17 12:28:27 why i want markdown is that i hope that i can just copy paste the written markdown article directly to the announce email 2014-07-17 12:28:36 <^7heo> haha okay :) 2014-07-17 12:28:37 wihtout needing massage the text anything 2014-07-17 12:28:42 i'm lazy... :) 2014-07-17 12:28:44 <^7heo> yeah 2014-07-17 12:29:26 the html in here: http://git.alpinelinux.org/cgit/ncopa/mksite-alpine/tree/index.md 2014-07-17 12:29:30 is just for the testing 2014-07-17 12:29:41 the 3 blocks will be generated 2014-07-17 12:30:14 <^7heo> well 2014-07-17 12:30:15 so they will likely be replaced with {{newsblock}} {{downloadblock}} {{devactivityblock}} 2014-07-17 12:30:21 <^7heo> by changing the css just a little 2014-07-17 12:30:32 <^7heo> (i.e. specifying a different css for different pages) 2014-07-17 12:30:37 i gotta go bbl 2014-07-17 12:30:48 <^7heo> you can actually manage to do that much simpler 2014-07-18 09:09:19 dalias: i am looking at connman: https://wiki.archlinux.org/index.php/Connman 2014-07-18 09:09:34 https://01.org/connman 2014-07-18 09:09:51 but it appears to depend on res_ninit/res_nclose 2014-07-18 10:46:21 mhm, I remember reading about disabling grsecurity, but can't find relevant command. any help (figuring out why specific application doesn't work properly - if it is grsec related)? 2014-07-18 10:49:30 melville: that normally shows up in dmesg 2014-07-18 10:49:43 with some "denied privilege" error or so 2014-07-18 10:50:29 i think almost all can be configured via sysctl, otherwise i don't know if there is a boot time switch 2014-07-18 10:50:38 alternatively one could use the vanilla kernel 2014-07-18 10:54:59 funny that ncopa created apkbuild for dosbox yesterday - I have been playing with it for some time. this is where issue is. 2014-07-18 10:56:03 dynamic recompilation in dosbox doesn't work on x86_64 - same applies to SVN version of dosbox. 2014-07-18 10:56:32 Setting excute permission on the code cache has failed \ Exit to error: DRC64:Unhandled memory reference 2014-07-18 10:57:38 when running with dynamic core. not sure if this could be grsec or pax (pax being more likely candidate I think). 2014-07-18 11:05:46 if its grsecurity it will likely leave traces in dmesg 2014-07-18 11:05:58 you can paxmark it too with paxctl 2014-07-18 11:06:38 ah 2014-07-18 11:06:50 seems like: paxctl -c -m is needed 2014-07-18 11:07:11 https://aur.archlinux.org/packages/linux-pax-flags/?comments=all 2014-07-18 11:07:15 dosbox is mentioned 2014-07-18 11:07:56 with "-m" I get only "Exit to error: DRC64:Unhandled memory reference" 2014-07-18 11:08:21 will try "paxctl -c -m" on SVN build 2014-07-18 11:08:33 and dmesg? 2014-07-18 11:08:41 i havent tried it 2014-07-18 11:09:41 [11808.170049] grsec: denied resource overstep by requesting 135168 for RLIMIT_MEMLOCK against limit 65536 for /usr/bin/dosbox[dosbox:1488] uid/euid:1000/1000 gid/egid:1000/1000, parent /bin/busybox[ash:14484] uid/euid:1000/1000 gid/egid:1000/1000 2014-07-18 11:12:09 I don't think this is issue, since core=normal generates exact same warning, but it works 2014-07-18 11:14:07 I get DRC64 on SVN build of dosbox with paxctl -c -m. 2014-07-18 11:21:15 fabled, dalias, connman does not build due to missing res_ninit and friends 2014-07-18 11:21:39 when asked about using c-ares the response was musl should implement libresolv 2014-07-18 11:22:04 when asking about gio's gresolver (https://developer.gnome.org/gio/2.40/GResolver.html) theresponse was: no 2014-07-18 11:23:08 seems we also patch asterisk for res_ninit 2014-07-18 11:23:59 though, res_n* seems BIND interface that glibc has later implemented 2014-07-18 11:29:16 fabled: if you want look at it, i have some patches: http://sprunge.us/eaIS 2014-07-18 11:29:40 and http://sprunge.us/EPVC 2014-07-18 12:24:07 ncopa, dynamic core works in dosbox when compiled with "CXX=clang++ CXXFLAGS=-mcmodel=medium ./configure" and "paxctl -c -m" 2014-07-18 12:24:47 with clang! 2014-07-18 12:24:51 he 2014-07-18 12:25:26 what does -mcmodel=medium do? 2014-07-18 12:25:39 i also wonder if gentoo has patches 2014-07-18 12:26:17 without it, linker complains about some symbols "relocation truncated to fit" 2014-07-18 12:27:14 could you please file a bug on bugs.a.o with the details? 2014-07-18 12:28:06 http://pastebin.com/sJmuYYBE 2014-07-18 12:28:56 report than dosbox works just fine with clang and not with gcc? 2014-07-18 12:29:06 yes 2014-07-18 12:29:12 report that the package shipped does not work 2014-07-18 12:29:15 and needs paxctl 2014-07-18 12:29:23 and that even with paxctl it fails 2014-07-18 12:29:29 and that it does work with clang++ 2014-07-18 12:29:42 which is kind of interesting 2014-07-18 12:36:42 ncopa: setattr instead of paxctl? 2014-07-18 12:37:41 we will 2014-07-18 12:37:48 once apk tools has support for xattrs 2014-07-18 12:59:44 ncopa: https://bugs.alpinelinux.org/issues/3189 2014-07-18 13:00:32 thanks 2014-07-18 13:01:15 do you have a dos .exe file i can test with? 2014-07-18 13:03:32 any dos game should do, since by default dosbox is switching to "core=dynamic", when "core=auto" (default config) is used 2014-07-18 13:05:14 i'll give something in a minute 2014-07-18 13:14:45 ncopa: demo version of "Blood & Magic": https://archive.org/details/BloodMagic_1020 2014-07-18 13:18:39 just unzip it and run "BAM.EXE" is dosbox. 2014-07-18 13:18:46 *in 2014-07-18 13:22:52 lack of similar reports (or complains from that matter) from gentoo/arch or other distributions that use recent GCC is kinda weird. 2014-07-18 13:50:03 ncopa, musl's resolver is stateless so res_ninit makes no sense 2014-07-18 13:50:17 ncopa, what is he trying to use it for? 2014-07-18 13:51:01 dalias: connman 2014-07-18 13:51:30 which looks like best alternative for wifi management 2014-07-18 13:51:57 http://git.kernel.org/cgit/network/connman/connman.git/tree/gweb/gresolv.c 2014-07-18 13:53:28 this is the error i get: https://gist.github.com/ncopa/e7cb811c500b3b2e5af0 2014-07-18 13:58:47 yes i've heard of connman 2014-07-18 13:58:50 it looked promising 2014-07-18 13:59:25 i got a bit dissapointed of the initial response from the devs 2014-07-18 13:59:37 tobur> would be better to patch musl to support resolv api 2014-07-18 14:00:38 but hum 2014-07-18 14:00:38 i'm confused why they're using res_ninit and a local resolver state since there don't seem to be any other calls that rely on it... 2014-07-18 14:00:57 it looks like they only use the struct 2014-07-18 14:01:10 they do the send here: http://git.kernel.org/cgit/network/connman/connman.git/tree/gweb/gresolv.c#n580 2014-07-18 14:02:07 we support the traditional part of the resolver api, just not some relatively recent things that seem to be glibc extensions 2014-07-18 14:02:19 and which are not useful/don't make sense for musl's stateless resolver 2014-07-18 14:02:20 it looks like they only use the struct 2014-07-18 14:02:45 @ line 1025 they're using it to inspect the list of nameservers being used 2014-07-18 14:02:48 no idea why they care 2014-07-18 14:02:59 i think they have their own resolver 2014-07-18 14:03:41 and use the __res_state struct for it 2014-07-18 14:03:41 oh, are they trying to probe what nameservers the system one uses and duplicate that? 2014-07-18 14:03:52 not sure 2014-07-18 14:04:27 yes that seems to be what they're doing 2014-07-18 14:05:04 btw this is all hideous code copied out of glib.... 2014-07-18 14:05:20 i think 2014-07-18 14:05:25 or maybe it's just using glib, i dunno 2014-07-18 14:05:39 yeah it looks not quite as bad as actual glib source... 2014-07-18 14:05:46 :) 2014-07-18 14:05:52 nice dose of UB... 2014-07-18 14:05:53 struct _GResolv { 2014-07-18 14:06:14 i think it is a resolver for glib implementation 2014-07-18 14:06:24 using the async features from glib 2014-07-18 14:06:38 so it fits into the glib mainloop 2014-07-18 14:06:56 it would be so much easier and more reliable to just make a thread that calls getaddrinfo... 2014-07-18 14:07:40 maybe they want support libc's with broken thread implementation? (eg uclibc) 2014-07-18 14:07:55 i see what they're _trying_ to do is duplicate what the system will do, but with their own resolver 2014-07-18 14:08:07 using the res state to load the nameserver list 2014-07-18 14:08:16 but obviously there's lots of stuff this can't duplicate 2014-07-18 14:08:17 dalias, the problem is that glibc does multi-megabyte stack allocation for new threads 2014-07-18 14:08:31 so you glibc+embedded = no threads 2014-07-18 14:08:31 like supporting hosts file, search suffix, ... 2014-07-18 14:08:36 fabled, :) 2014-07-18 14:08:42 pthread_attr_setstacksize 2014-07-18 14:08:46 i know 2014-07-18 14:08:55 programmers in general do not 2014-07-18 14:09:22 or somehow they think it's not portable 2014-07-18 14:09:41 what they currently do is not portable 2014-07-18 14:09:41 or maybe glibc just crashes with small stacks... ;) 2014-07-18 14:09:48 :) 2014-07-18 14:10:16 maybe they just generally think threads are bad 2014-07-18 14:10:52 parsing resolv.conf themselves would be arguably less bad than what they're doing now 2014-07-18 14:11:32 doesnt glibc has some nss.conf something? 2014-07-18 14:12:04 yes 2014-07-18 14:12:09 but thats maybe only for passwd etc 2014-07-18 14:12:21 if the underlying system is using something other than plain dns for hostname resolving, via nss 2014-07-18 14:12:28 this connman hack won't work at all 2014-07-18 14:12:39 that's part of why i say it's broken 2014-07-18 14:13:11 yeah, looks like nss.conf does not specify resolv.conf 2014-07-18 14:13:23 hosts, and passwd can be configured 2014-07-18 14:14:21 and they use getaddrinfo too: http://git.kernel.org/cgit/network/connman/connman.git/tree/gweb/gresolv.c#n785 2014-07-18 14:15:06 hmm their code is also broken in that it does not suppot ipv6 2014-07-18 14:15:09 they think it does, but it doesn't 2014-07-18 14:15:15 :) 2014-07-18 14:15:21 nsaddr_list[] has type struct sockaddr_in[] 2014-07-18 14:15:27 so it cannot store ipv6 addresses 2014-07-18 14:15:32 i thought ipv6 was the reason they dont use gio's Gresolver 2014-07-18 14:16:23 there's space at the end of the __res_state struct with room for ipv6 nameserver addresses 2014-07-18 14:16:38 and it looks like it's made so you could setup access to it via macros 2014-07-18 14:17:07 oh they do use it... 2014-07-18 14:17:30 resolv->res._u._ext.nsaddrs[i] 2014-07-18 14:17:36 how hideous 2014-07-18 14:17:44 i really doubt _u._ext was meant to be used 2014-07-18 14:19:58 i suspect their code may also fail to find new nameservers when resolv.conf changes 2014-07-18 14:20:46 those are the only calls made to their resolveR: 2014-07-18 14:20:55 gweb.c: g_resolv_cancel_lookup(web->resolv, session->resolv_action); 2014-07-18 14:20:55 gweb.c: web->resolv = g_resolv_new(index); 2014-07-18 14:20:55 gweb.c: g_resolv_unref(web->resolv); 2014-07-18 14:20:55 gweb.c: g_resolv_set_debug(web->resolv, func, user_data); 2014-07-18 14:20:55 gweb.c: g_resolv_set_address_family(web->resolv, family); 2014-07-18 14:20:58 gweb.c: g_resolv_add_nameserver(web->resolv, address, 53, 0); 2014-07-18 14:21:00 gweb.c: session->resolv_action = g_resolv_lookup_hostname(web->resolv, 2014-07-18 14:22:17 i wonder if that g_resolv_set_address_family is the reason why they dont use gio's gresolver 2014-07-18 14:22:26 which can lookup hostnames async 2014-07-18 14:30:45 in theory it would probably be possible to factor things such that res_ninit could be provided 2014-07-18 14:31:04 do we want that in musl? 2014-07-18 14:31:53 i wonder why they cannot simply use https://developer.gnome.org/gio/2.39/GResolver.html#g-resolver-lookup-by-name-async 2014-07-18 14:32:39 i'd basically follow the normal inclusion/exclusion principles 2014-07-18 14:33:04 there's no strong reason for exclusion. the api is ugly but not fundamentally unusable 2014-07-18 14:33:42 so it mostly goes by the inclusion criteria: is there major demand for the interface that makes including it easier/better than patching programs not to use it? 2014-07-18 14:34:01 i have one more app that i believe wants it 2014-07-18 14:34:03 libspf2 2014-07-18 14:34:12 i might remember wrong 2014-07-18 14:34:13 oh 2014-07-18 14:34:18 zabbix might want it too 2014-07-18 14:35:23 zabbix issue with uclibc: https://support.zabbix.com/browse/ZBX-5197 2014-07-18 14:35:23 and asterisk prefers it, we do some patching now to it wrt. res_ninit 2014-07-18 14:42:40 yup libspf2 wants it 2014-07-18 14:42:42 libtool: link: ccache gcc -Os -fomit-frame-pointer -DNETDB_SUCCESS=0 -Wall -Wl,--as-needed -o .libs/spfquery spfquery.o ../../src/libspf2/.libs/libspf2.so -lpthread -lintl -lresolv 2014-07-18 14:42:42 ../../src/libspf2/.libs/libspf2.so: undefined reference to `res_close' 2014-07-18 14:42:42 ../../src/libspf2/.libs/libspf2.so: undefined reference to `ns_msg_getflag' 2014-07-18 14:42:42 collect2: error: ld returned 1 exit status 2014-07-18 14:42:42 Makefile:427: recipe for target 'spfquery' failed 2014-07-18 14:43:43 i think i also carved out res_* from either nfs-utils or one of its deps 2014-07-18 19:17:29 hahaha, how ontopic: 2014-07-18 19:17:29 19:15 < keen> heh 2014-07-18 19:17:29 19:16 < keen> i mean, really, a tool designed to start and keep a process up that a) often fails to do that, and b) can't figure out how to shut one down - and even worse, when it fails to shut it down, it just gives up and reports it down. 2014-07-18 19:17:33 19:17 < keen> so if you try to start it later, guess what? conflict!19:15 < keen> heh 2014-07-18 19:17:36 19:16 < keen> i mean, really, a tool designed to start and keep a process up that a) often fails to do that, and b) can't figure out how to shut one down - and even worse, when it fails to shut it down, it just gives up and reports it down. 2014-07-18 19:17:45 sorry, i mispasted the second time 2014-07-19 08:44:05 hey 2014-07-19 09:16:31 can i ask for help? 2014-07-19 09:19:43 you can ask always, but don't expect instant answer or answers for every question ;) 2014-07-19 09:23:09 hahah i know. anyway, i'm installing alpine on my laptop, how can i use a separated /boot (on /dev/sda3) and a separated / (/dev/sda5) /home (/dev/sda6) and a swap on /dev/sda1 ? 2014-07-19 09:24:19 http://wiki.alpinelinux.org/wiki/Setting_up_disks_manually i found this earlier but i still don't know how to set a separated /boot 2014-07-19 09:26:54 "If you want a different layout you can manually create the parititions, filesystems and mount them up on /mnt (or any other mount point) and then run: setup-disk /mnt" 2014-07-19 09:28:17 so you mount /dev/sda5 as /mnt, then mkdir -p /mnt/boot /mnt/home, mount /dev/sda3 as /mnt/boot, /dev/sda6 as /mnt/home 2014-07-19 09:29:40 for swap you use swapon /dev/sda1. setup-disk should pick up fstab entries by itself after that. 2014-07-19 09:29:40 and run setup-disk -m sys /mnt ? 2014-07-19 09:29:53 ah ok, thanks 2014-07-19 09:32:57 and when i run setup-disk /mnt, there will be changes on /home? 2014-07-19 09:33:29 i mean, the partition will be formatted or not? 2014-07-19 09:35:15 my answer is no, but I'm not 100% sure. if you are hesitant you can just mount / and /boot, you can always add /etc/fstab entry for /home later on 2014-07-19 09:37:11 okay, thanks for the help! 2014-07-19 09:37:19 ncopa sorry for bothering (I could just send mail or file report), but it seems that firefox is not working after xorg drivers update: 2014-07-19 09:37:30 [ 502.060603] firefox[1715]: segfault at 8 ip 000062345d03196d sp 0000708edcc282c0 error 4 in i915_dri.so[62345cde4000+65f000] 2014-07-19 09:38:06 as normal user, "sudo firefox" works. permission issues with xf86-video-intel after update? 2014-07-19 09:40:57 scraps from id: "groups=18(audio),27(video)," 2014-07-19 09:46:12 u7knv9h_, let me know how it goes. problem is that installation of alpine takes like 5 minutes - hard to remember that :] 2014-07-19 09:50:30 oh, i think i need to fix the mbr, when i try to reboot instead of alpine there is grub rescue mode, even though the bootloader of alpine is extlinux 2014-07-19 09:50:46 but i don't know how, i'm searching in the wiki 2014-07-19 09:52:07 i figured how to do, now i try 2014-07-19 09:58:19 to copy /usr/share/syslinux/mbr.bin into the mbr is right to do " dd if="/usr/share/syslinux/mbr.bin" of="/dev/sda" " ? 2014-07-19 10:01:35 /dev/sda 2014-07-19 10:02:07 it won't delete the whole hard disk, right? 2014-07-19 10:02:50 it shouldn't, here http://wiki.alpinelinux.org/wiki/Setting_up_LVM_on_GPT-labeled_disks at bottom you have similar instructions, just for gpt 2014-07-19 10:08:15 oh. thanks! everything works fine now 2014-07-19 10:09:00 just, how i add /home in /etc/fstab? the other partitions have the uuid, how can i get it? 2014-07-19 10:11:39 "sudo blkid" 2014-07-19 10:12:22 or as root 2014-07-19 10:26:11 thanks for all the help, but after the UUID, the mountpoint and the filesystem, what should i put? 2014-07-19 10:26:43 rw,relatime,datme,data=ordered 0 0 ? 2014-07-19 10:26:53 as / and /boot ? 2014-07-19 10:29:33 for /home parition? just copy "/" entry and adjust UUID and change / to /home (that is if you use same filesystem) 2014-07-19 10:31:37 thanks for the help, and sorry for bothering you! 2014-07-19 10:31:45 after introducing new entry you can check if it is working without rebooting by using "mount /home" 2014-07-19 10:32:03 np 2014-07-19 10:33:57 just a question, what you intend to use alpine for? 2014-07-19 10:36:06 daily use, as desktop 2014-07-19 17:38:10 is setup-alpine broken in the 3.0.1-x86_64 iso on the site? 2014-07-19 17:38:30 i get http://i.imgur.com/HyzTODe.png when installing in an ESXi VM 2014-07-19 17:38:50 and upon startup 2014-07-19 17:39:17 as far as i can tell it installs just fine 2014-07-19 17:39:36 i also get http://i.imgur.com/yBjXyz8.png 2014-07-19 17:45:05 Rylee first stiff is already fixed 2014-07-19 17:45:35 is that fix reflected in the ISO available on the site? 2014-07-19 17:45:51 in https://bugs.alpinelinux.org/issues/3097 2014-07-19 17:46:13 okay 2014-07-19 17:46:14 regardind udhcpcd 2014-07-19 17:46:25 its started to fast for some network 2014-07-19 17:46:37 interesting 2014-07-19 17:46:43 well it is a virtual network 2014-07-19 17:47:13 I'm considering Alpine as a lightweight firewall and router in front of every other internal VM 2014-07-19 17:47:15 run setup-alpine once more and i think it should be ok 2014-07-19 17:48:18 okay 2014-07-19 17:48:20 thanks crow 2014-07-19 17:54:54 Rylee maybe this can help you http://bugs.alpinelinux.org/issues/3105 2014-07-19 17:55:37 actually configuring my dhcpd would help too >_> 2014-07-19 17:55:40 thanks for the help crow 2014-07-19 17:58:09 well if you use auto then timeout could fix your problem. i have had such case to with apu4c 2014-07-19 19:12:13 http://en.wikipedia.org/wiki/UClibc still refers to Alpine, and musl does not 2014-07-19 20:19:10 Rylee 2.7.* is still uClibc, 3.0x is musl 2014-07-19 20:29:07 what do I need to install to get truncate? 2014-07-19 20:29:45 (or any other tool that'll let me expand a sparse file?) 2014-07-19 20:32:33 found it in coreutils :) 2014-07-20 22:16:16 <^7heo> ncopa: I was wondering 2014-07-20 22:16:28 <^7heo> wouldn't it be great to have a img that you can dd on a volume? 2014-07-20 22:16:38 <^7heo> (as in usb drive) 2014-07-20 22:16:49 <^7heo> isn't it even preferable to an iso these days? 2014-07-20 22:17:04 <^7heo> (I know I would prefer) 2014-07-20 22:17:53 <^7heo> (since you can install a netbsd, for example, and just dd your image on a usb drive, without any third party tool needed) 2014-07-20 23:36:47 ncopa, you around? 2014-07-21 00:41:46 <^7heo> seems not 2014-07-21 07:03:18 morning 2014-07-21 07:05:13 ^7heo: youre not the first one to ask for that 2014-07-21 07:05:26 need to think how to do that properly 2014-07-21 07:05:59 the reason we dont use images for dd already is that we currently depend on FAT 2014-07-21 07:06:13 and for tmpfs installs the config is stored there 2014-07-21 07:06:48 which means that we'd either have a fixed size image, regardless how big your usbstick is 2014-07-21 07:06:58 which should old the apkovl and the apk cache 2014-07-21 07:07:21 ...or we'd need to have a way to add a second partition for apkovl and apk cache 2014-07-21 07:07:40 the problem is FAT, and that you cannot resize it 2014-07-21 07:08:10 FAT is kinda convenient for windows users that want backup/restore apkovl 2014-07-21 10:28:38 <^7heo> ncopa: is the dependency on FAT such a big problem? 2014-07-21 17:36:54 i need to figure out a way to put Alpine on a VPS somewhere 2014-07-21 17:36:57 any good VPSes which can work with Alpine? 2014-07-21 17:47:45 <^7heo> vultr. 2014-07-21 17:47:53 <^7heo> one word, vultr. 2014-07-21 17:47:59 <^7heo> that's what I use right now. 2014-07-21 17:48:03 <^7heo> It's running alpine. 2014-07-21 17:48:28 <^7heo> and you can get freaking cheap prices if their sale is still on. 2014-07-21 17:48:39 <^7heo> and if you get one, would you care to use my referential? 2014-07-21 17:48:45 <^7heo> (so I get free credit out of it :)) 2014-07-21 17:49:36 <^7heo> asie: ^ 2014-07-21 17:51:49 what sale? 2014-07-21 17:52:01 <^7heo> you can get $1 free for $1 of credit. 2014-07-21 17:52:08 <^7heo> so I got the cheapest VPS they sell 2014-07-21 17:52:11 <^7heo> I paid 75E 2014-07-21 17:52:15 <^7heo> and I have it for more than 2 years. 2014-07-21 17:52:18 <^7heo> pre-paid. 2014-07-21 17:52:21 link? 2014-07-21 17:52:22 oh 2014-07-21 17:52:23 <^7heo> mom 2014-07-21 17:52:45 <^7heo> https://www.vultr.com/news/1/Vultr-Promotion!-Matching-up-to-$100-of-initial-account-funding! 2014-07-21 17:52:52 <^7heo> but please use my referential then :) 2014-07-21 17:53:51 <^7heo> http://www.vultr.com/?ref=6806103 2014-07-21 17:54:14 <^7heo> if you use that link, I will get money if you get a VPS ;) 2014-07-21 17:54:18 <^7heo> so please, do so. :D 2014-07-21 17:54:42 vultr looks pretty sweet 2014-07-21 17:54:49 <^7heo> (and the best thing is, with vultr, you can actually upload (or download from another location) an iso, and use it when you install your VPS) 2014-07-21 17:54:52 <^7heo> well, I can tell you 2014-07-21 17:55:03 <^7heo> I did my homework before getting that one. 2014-07-21 17:55:11 i think i will go there 2014-07-21 17:55:12 <^7heo> KVM based (not that cheap openVZ shit) 2014-07-21 17:55:16 i know i know 2014-07-21 17:55:21 <^7heo> 12 locations, including japan and US 2014-07-21 17:55:22 i did have quite a few VPSes before this 2014-07-21 17:55:29 <^7heo> and AMSTERDAM for godsake 2014-07-21 17:55:30 stop before you sound like you're a salesman 2014-07-21 17:55:33 <^7heo> :P 2014-07-21 17:55:34 <^7heo> sorry 2014-07-21 17:55:34 amsterdan is common 2014-07-21 17:55:39 tokyo, japan is the surprising one from my expeirence 2014-07-21 17:55:42 <^7heo> not in the cheap VPS 2014-07-21 17:55:48 yes in the cheap VPS 2014-07-21 17:55:50 digitalocean 2014-07-21 17:55:54 <^7heo> yeah okay. 2014-07-21 17:55:56 <^7heo> this one okay. 2014-07-21 17:56:01 <^7heo> but I looked into other ones 2014-07-21 17:56:16 i also have the infinite bandwidth advantage over at DigitalOcean 2014-07-21 17:56:19 as I have a grandfathered acc 2014-07-21 17:56:19 <^7heo> they were from obscure places, and not from places where you have huge IXes 2014-07-21 17:56:25 yeah 2014-07-21 17:56:26 that is useful 2014-07-21 17:56:55 <^7heo> but the feature that killed me was this "upload or download is from another location and boot it in 1 minute" 2014-07-21 17:57:05 <^7heo> I didn't expect that. 2014-07-21 17:59:46 TinyCoreLinux! www.tinycorelinux.com 2014-07-21 18:01:45 <^7heo> room098: is it using musl? 2014-07-21 18:24:50 <^7heo> meh 2014-07-21 18:25:04 <^7heo> I should ignore tiny as an expression 2014-07-21 18:36:29 ^7heo: https://vpsboard.com/topic/4372-vultr-smtp-restriction-requirements-are-unreasonable/ 2014-07-21 18:36:35 a friend linked this to me 2014-07-21 18:37:05 <^7heo> ah 2014-07-21 18:37:24 <^7heo> I wasn't aware of that, sorry. 2014-07-21 18:37:31 and i do use SMTP for my personal email 2014-07-21 18:38:04 <^7heo> I don't like their requirements... 2014-07-21 18:38:07 <^7heo> I really don't. 2014-07-21 18:38:23 <^7heo> please disregard my recommendation. 2014-07-21 18:38:36 <^7heo> too bad. 2014-07-21 18:46:39 <^7heo> sending email is "free", up to 200 emails a day, according to http://sendgrid.com/ 2014-07-21 18:46:43 <^7heo> thats bullshit. 2014-07-21 18:49:08 <^7heo> asie: from that thread you linked: "Using a verified PayPal account will be enough in most cases and won't require any scanned documents." 2014-07-21 18:49:15 <^7heo> and then "I've added a few funds to my account and staff liftet the block just a few minutes later." 2014-07-21 18:56:09 mhm, acme + edbrowse with some scripting is not bad: http://i.imgur.com/3PCWKxw.png 2014-07-21 18:57:18 i managed to compile plan9port against musl, but there are some issues (occasional crashes) 2014-07-21 19:38:30 hi 2014-07-21 19:39:27 I'm trying to use debootstrap in a chroot but I keep getting permission denied 2014-07-21 19:39:30 in dmesg I get 2014-07-21 19:40:16 grsec: use of CAP_SYS_ADMIN in chroot denied for /home/aphrael/bin/mount[mount:27429] uid/euid:0/0 gid/egid:0/0, parent /usr/bin/debootstrap[debootstrap:24177] uid/euid:0/0 gid/egid:0/0 2014-07-21 19:42:18 sysctl kernel.grsecurity.chroot_caps? 2014-07-21 19:43:11 that's set to 1 2014-07-21 19:43:51 should I set it to something else melville 2014-07-21 19:43:59 try setting it to zero to turn off grsec for this: http://en.wikibooks.org/wiki/Grsecurity/Appendix/Grsecurity_and_PaX_Configuration_Options#Capability_restrictions 2014-07-21 19:44:05 ah. 2014-07-21 19:44:07 excellent 2014-07-21 21:27:51 sysctl helped Elizacat? would be nice to know in future 2014-07-21 22:03:10 my nfs is sliiiightly broken, any ideas? http://hastebin.com/ravokuluyi.sm 2014-07-21 22:04:35 oh my, it seems there's a bug for it 2014-07-21 23:09:48 melville, yes 2014-07-21 23:09:50 melville, :) 2014-07-21 23:23:04 Good evening. After downloading both the Mini and Standard editions of Alpine Linux and installing it onto Virtual Machines I find that they are identical. I must be wrong. What is different? Is the standard edition detecting that I have an active internet connection and then proceeds to work as a mini edition, thus eliminating the need for local packages? 2014-07-21 23:35:01 Hey all. Was wondering if anyone else received "error checking /dev/vdXY mount status" when trying to create a btrfs file system on 3.0.1 x86_64? 2014-07-21 23:39:09 busk more packages in standard edition, so it can be better if you don't do network installation. it can be also more helpful as rescue cd - so you don't need network to add packages. 2014-07-21 23:39:33 But assuming I have network access. There is no practical differnce? 2014-07-21 23:41:09 if you have network you can boot cd and 'setup-network' and 'apk add' (you can do that since it is stored into ram). 2014-07-21 23:41:37 Alright, thanks 2014-07-23 11:30:48 Hello everyone, I've got a simple question about how to setup ZNC on a box running Alpine Linux. How can I create the config files for running znc globaly. (su znc doesn't work) Thank you in advance. 2014-07-23 11:37:13 apk info -L znc | head -n3 2014-07-23 11:37:15 etc/init.d/znc 2014-07-23 11:37:22 etc/conf.d/znc 2014-07-23 11:37:37 or you mean something different by running it globally? 2014-07-23 11:39:25 I want to run it using service znc start (/etc/init.d/znc start) but it crashes because of a missing config file in /var/lib/znc/configs/znc.conf which I normaly would create with znc --makeconf 2014-07-23 11:39:35 but I can' t access the user znc 2014-07-23 11:39:46 without changing the user 2014-07-23 11:41:16 run znc --makeconf as normal user and cp ~/.znc/configs/znc.conf /var/lib/znc/configs/znc.conf ? 2014-07-23 11:41:32 is this the recommended way of doing it? 2014-07-23 11:43:49 znc is probably a no-login user, so you can't log to it 2014-07-23 11:44:48 right 2014-07-23 11:45:26 i'm still unsure if /var/lib is proper directory for config though - doesn't it use global config in /etc like the rest of the world? 2014-07-23 11:46:29 mhm 2014-07-23 11:47:35 /var/lib is the path set in /etc/conf.d/znc by default 2014-07-23 11:48:16 actually it seems that you can run "znc --makeconf --datadir /var/lib/znc" as su I guess 2014-07-23 11:48:47 okay... I did it now by running it as local user and copying files and setting owner 2014-07-23 11:49:06 I didn't do it with --datadir because of the permissions of these files 2014-07-23 11:49:51 Thank you after all for your help 2014-07-23 11:51:06 i'm not znc user, just it is common to copy configs from /usr/share/examples to /etc. not sure if this is recommended way for znc. 2014-07-23 11:55:27 I don't think so 2014-07-23 15:27:07 I am trying to install Alpine Linux through a CD, but it says "Mounting boot media failed""and throws me at an emergency shell 2014-07-23 15:27:33 I searched a little and, apparently, I have to edit syslinux.cfg, is that so? 2014-07-23 15:27:49 Where can I find that file? I am currently at the emergency shell 2014-07-23 16:23:34 dhua: hit tab at the boot menu 2014-07-23 16:23:47 boot time edit for the boot command line 2014-07-23 16:24:01 it should just work though. 2014-07-23 16:24:42 How do I get to the boot menu? 2014-07-23 16:24:49 I am on the emergency shell right now 2014-07-23 16:25:01 it is the first thing loaded at boot 2014-07-23 16:25:07 if that doesn't load, there is something wrong 2014-07-23 16:25:23 So, I reboot? 2014-07-23 16:25:28 The I get boot: 2014-07-23 16:25:43 but after a while I am throw to the emergency shell 2014-07-23 16:26:20 and that's all? 2014-07-23 16:26:25 i had that once 2 weeks back 2014-07-23 16:26:32 it was on virtualbox on ubuntu 2014-07-23 16:26:49 i just switched to my laptop and there the same iso worked without issue 2014-07-23 16:26:59 it sounds pretty similar :/ 2014-07-23 16:27:09 Here's how it goes: 2014-07-23 16:27:11 ISOLINUX blah blah blah 2014-07-23 16:27:15 boot: 2014-07-23 16:27:26 Then, after a few seconds: 2014-07-23 16:27:36 Mounting boot media failed 2014-07-23 16:27:38 yep, same issue. idk what caused it 2014-07-23 16:27:54 are you in a vm or is this on a real system? 2014-07-23 16:27:56 initramfs emergency recovery shell launched blah blah 2014-07-23 16:28:08 A real system 2014-07-23 16:28:12 damn :/ 2014-07-23 16:28:29 i'd recommend you do the same thing but use a 2.7 iso 2014-07-23 16:28:44 I wanted to stay up to date and use musl 2014-07-23 16:28:50 upgrading is a bit tricky from 2.7 to 3.0 but no big deal 2014-07-23 16:29:02 take like 2 minutes 2014-07-23 16:29:08 Meh, I will do that if I can;t come across this issue 2014-07-23 16:29:25 I need to edit syslinux.cfg apparently 2014-07-23 16:29:37 but you're in a ramdisk, i don't see how that can work 2014-07-23 16:29:48 maybe it can 2014-07-23 16:30:28 Here's my error: http://wiki.alpinelinux.org/wiki/Create_a_Bootable_USB#Wrong_Device_Name 2014-07-23 16:30:34 But that concerns USB 2014-07-23 16:30:42 I'm using a CD 2014-07-23 16:50:26 I guess I will just buy a USB stick, be back in 20 minutes. 2014-07-23 16:50:34 I hope this CD error gets fixed soon... 2014-07-23 17:44:41 Great, USB installation worked 2014-07-23 17:44:55 But I'm getting the following error on boot: 2014-07-23 17:45:13 Starting networking ... 2014-07-23 17:45:20 lo ... [ok] 2014-07-23 17:45:30 eth0 ... [ok] 2014-07-23 17:45:41 wlan0 ... 2014-07-23 17:45:49 Error: this script should be called from udhcpc [ !! ] 2014-07-23 17:46:01 Any ideas? 2014-07-23 17:50:38 dhua: http://bugs.alpinelinux.org/issues/3105 2014-07-23 17:51:28 Thank you! 2014-07-23 19:22:08 I see Alpine has no C compiler by default 2014-07-23 19:22:30 What do you guys suggest? gcc, clang, tcc or something else? 2014-07-23 19:44:54 I tried installing gcc but when I try compiling a file get " not found" 2014-07-23 19:49:21 didn't 'apk add gcc', pull in musl-dev also? 2014-07-23 19:51:06 I am on Alpine 3.0.0 and did 'apk add gcc' but cannot compile anything, since stdlib.h is not found 2014-07-23 19:52:04 libc-dev 2014-07-23 19:52:35 i'm curious myself how does one apt-file search in alpine 2014-07-23 19:53:13 What is libc-dev and how does it relate to musl? 2014-07-23 19:54:15 katlogic: http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management#Search_for_Packages 2014-07-23 19:54:58 yeah, no file search 2014-07-23 19:55:04 sucks but understandable, the db would be huge 2014-07-23 19:56:17 One thing I liked on arch was to type "pacman -S firef" and it autocompleted to "pacman -S firefox" 2014-07-23 19:56:50 And "pacman -S fi" and it listed all packages starting with "fi" 2014-07-23 19:57:22 But that's just a gimmick, no real need 2014-07-23 19:57:36 Even though it's useful 2014-07-23 20:16:19 katlogic: apk search -x 'fi*' is actually more useful, since it will go for exact search and not substring, listing only packages that start with fi and not with fi in the middle 2014-07-23 20:17:31 as for db, netbsd has apropos implementation that does full text search on man pages using sqlite - i would believe that is bigger db than file lists for packages 2014-07-23 20:17:43 nah, i just want files 2014-07-23 20:17:46 apt-file search 2014-07-23 20:51:19 Can I edit my /etc/apk/repositories to edge,install a package that's only available there and go back to 3.0? 2014-07-23 20:58:51 don't think so. you could build that package using apkbuild by yourself on 3.0 i guess. 2014-07-23 21:02:00 How do I install a package that I have the apkbuild for? 2014-07-23 21:07:25 makepkg? 2014-07-23 21:08:11 Nevermind,it's abuild 2014-07-23 21:11:57 dhua: http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management#Repository_pinning 2014-07-23 21:12:23 Thank you, tdtrask 2014-07-23 21:12:41 I will try that if the abuild method fails 2014-07-23 21:26:17 All right, I succesfully installed using abuild 2014-07-23 21:26:29 But I can't run the program for some reason 2014-07-23 21:26:37 It isn't found 2014-07-23 21:48:30 apk -L name-of-package to check what it contains? 2014-07-23 21:49:29 which name-of-binary to check if it is in PATH? 2014-07-23 22:22:33 I will try that soon, just screwed my syslinux config and booted up a live usb to try to recover it 2014-07-23 22:46:25 melville: apk tells me -L is an invalid option 2014-07-23 22:46:50 which name-of-binary returns nothing 2014-07-23 22:47:15 when I try abuild again, it says package is up to date 2014-07-23 22:47:22 But I can't run it 2014-07-23 22:47:30 It isn't found 2014-07-23 22:49:15 apk info -L 2014-07-23 22:49:22 sorry about that 2014-07-23 22:50:16 returns nothing 2014-07-23 22:52:49 Any idea? 2014-07-23 22:53:32 did you apk add package? 2014-07-23 22:54:02 No, I just abuild it 2014-07-23 22:54:39 so it isn't installed - abuild builds in chroot? (didn't use it in quite some time) 2014-07-23 22:55:00 I thought abuild would install it? 2014-07-23 22:55:08 Just like make install or something 2014-07-23 22:56:27 oh, it created an apk 2014-07-23 22:56:43 an .apk 2014-07-23 22:56:49 I think I got it 2014-07-23 22:58:36 Got it. abuild created a .apk and I had to apk add --allow-untrusted that .apk file 2014-07-23 22:58:40 Thanks for the help 2014-07-23 23:05:09 There's no infinality for Alpine? 2014-07-23 23:05:29 How do you guys keep your font rendering gorgeous? 2014-07-23 23:48:22 hi all I'm having trouble getting asterisk to do anything on alpine 3.0.1. Did apk add asterisk and asterisk-sample-config, /etc/init.d./asterisk start and can actually connect to the command interface with asterisk -r. But 'sip show peers' isn't recognized, and 'module show' tells me 0 modules are loaded. =? 2014-07-24 06:03:35 fabled: have you got asterisk sip running with musl libc? ^^^ 2014-07-24 06:09:12 i don't remember if i tested that well or not 2014-07-24 06:09:27 there was ssl issue i fixed earlier 2014-07-24 06:10:09 so the version .iso might be broke 2014-07-24 09:44:51 whee 2014-07-24 09:44:55 open-vm-tools finally built 2014-07-24 09:45:03 there are just one zillion patches... 2014-07-24 09:50:04 cool 2014-07-24 09:50:29 the nox11 version i assume? 2014-07-24 09:50:36 (never needed the other one) 2014-07-24 10:01:48 fighting the kernel modules now 2014-07-24 10:01:52 after lunch... 2014-07-24 10:35:40 cool! 2014-07-24 16:21:55 Hi! What is the latest no musl stable release? 2014-07-24 16:23:38 2.7.9 2014-07-24 16:23:53 Thanks! 2014-07-24 16:24:22 Any particular reason why you are moving to musl? what is the main benefit? 2014-07-24 16:26:12 Got it... I am reading the list 2014-07-24 19:21:50 Hi 2014-07-24 19:23:35 does anyone know what this means in gdb 2014-07-24 19:23:36 warning: Cannot call inferior functions, Linux kernel PaX protection forbids return to non-executable pages! 2014-07-24 19:30:43 pax is overstepping, what application do you use? 2014-07-24 19:31:07 paxctl -c some_flags your_binary should help 2014-07-24 19:31:38 I am learning C, and they just showed a quick intro to gdb, it's some custom program I wrote it does it with every program 2014-07-24 19:35:14 is this only warning or it prevents you from using gdb? 2014-07-24 19:38:55 http://pastebin.com/N8yA6RWR 2014-07-24 19:39:34 http://pastebin.com/74cNNaUB 2014-07-24 19:42:24 for breakpoints you need to turn off pax mprotect if i remember 2014-07-24 19:42:59 apk add paxctl, then paxctl -c -m your-binary, check then gdb 2014-07-24 19:45:03 ok 2014-07-24 19:45:30 localhost:~# paxctl -c -m decrypt 2014-07-24 19:45:30 file decrypt had a PT_GNU_STACK program header, converted 2014-07-24 19:46:44 it works now 2014-07-24 19:46:54 I still get this warning : warning: Cannot call inferior functions, Linux kernel PaX protection forbids return to non-executable pages! 2014-07-24 19:46:59 but it works the breakpoints 2014-07-24 19:47:54 may be relevant: http://wiki.gentoo.org/wiki/Hardened/Debugging 2014-07-24 19:51:43 thanks 2014-07-24 20:00:10 I am trying to use this APKBUILD: http://pastebin.com/raw.php?i=CMkPySKF 2014-07-24 20:00:29 It returns: >>> ERROR: fontconfig-infinality-ultimate-git: fontconfig-ultimate.install: unknown install script suffix 2014-07-24 20:00:40 Any ideas on how to fix this? 2014-07-24 20:25:21 ehuwa http://wiki.alpinelinux.org/wiki/APKBUILD_Reference#install 2014-07-24 20:26:51 So, I just change install=fontconfig-ultimate.install to install=fontconfig-ultimate.pre-install? 2014-07-24 20:26:57 Let me try that 2014-07-24 20:27:51 converting aur pkgbuild? 2014-07-24 20:28:29 >>> ERROR: fontconfig-infinality-ultimate-git: fontconfig-ultimate.pre-install: install script does not match pkgname or any subpackage 2014-07-24 20:28:43 melville: trying to convert, hehe 2014-07-24 20:29:22 It's for the infinality patches of freetype2, that makes font rendering gorgeous 2014-07-25 11:46:24 had somebody success with unpriviliged lxc containers? 2014-07-25 11:55:51 Hello There 2014-07-25 11:56:48 I just install Alpine Linux 3.0.1 on x86_64; as per the Wiki al lwent fine, at staartxfce4, I get a frozen Mouse Desktop with no Keyboard and Mouse. 2014-07-25 11:57:28 GRUB boot kernel commandline has these modules=sd-mod,usb-storage,etx3, as it boots iptables, etc dont appear in mesages 2014-07-25 11:58:21 Anybody there? 2014-07-25 12:11:51 Hello There 2014-07-25 12:12:21 I just installed Alpine Linux on HD, did all as per Wiki, I get XFCE4 desktop, but it is frozen 2014-07-25 12:19:05 bluesky: what hardware is it? 2014-07-25 12:19:10 intel graphics? 2014-07-25 12:21:41 ncopa; It is AMD64 with NVIDIA FX310 or so Graphics Card. Xorg -configure, cannot find xf86-nouveau, modesettings,vesa,fbdev, error so no Xorg.conf, but first time with startxfce4 I got the desktop setup backgroud, panel, Now it just freezes, no LXDDM etc 2014-07-25 12:22:25 can you log in via ssh or do ctrl-alt-f1 to switch to terminal? 2014-07-25 12:23:12 no ctrl-alt-f1:f12 2014-07-25 12:23:20 Just reset 2014-07-25 12:24:44 I installed iptables-firewall config, it does not show up during boot, rc-service LXDM start, rc-update add LXDM ; LXDM does not appear, 2014-07-25 12:31:09 ncopa, Are you still there? 2014-07-25 12:37:15 i'm here 2014-07-25 12:37:39 i believe you are having issues with either kernel or xorg driver 2014-07-25 12:37:48 it coudl be grsec kernel is causing it 2014-07-25 12:37:57 but it could also be something else 2014-07-25 12:38:26 you could try install the vanilla kernel and see if you have problem there too 2014-07-25 12:39:13 I belive mprotect of grsec has issues with X; I have had that problem with ArchLinux, no ALSASOOUND on Gentoo with Grsec 2014-07-25 12:39:33 after a reboot, does /var/log/Xorg.0.log say anything? 2014-07-25 12:40:11 with latest xorg and intel driver we got issues with grsecurity kernel due to driver trying to access /sys 2014-07-25 12:40:45 http://forums.grsecurity.net/viewtopic.php?f=3&t=4012 2014-07-25 12:40:48 but that is other issue 2014-07-25 12:41:03 i dont know what is causing your problem 2014-07-25 12:42:02 with reboot, I get that cannot load xf86-nouveau.so, xf86-modesetting.so, xf86-vesa.so, xf86-fbdev.so unloading nouveau, modesetting, vesa, fbdev, so it is grsec and kernel 2014-07-25 12:45:29 ncopa, Thanks will purse this some other time Thanks 2014-07-26 21:47:44 hey guys is there a problem with the download link atm? Cause it does not work for me. 2014-07-26 21:48:22 trying to download alpine-3.0.1-x86.iso 2014-07-26 21:50:36 works for me 2014-07-26 21:50:50 anyway try a different mirror 2014-07-26 21:53:47 thx, I used one of the links at the bottom and it seems to work 2014-07-26 22:50:41 bye 2014-07-26 23:29:12 can someone tell me if alpine-3.0.1-x86 should work with virtual box 4? 2014-07-26 23:29:39 cause I get "Booting kernel failed: Invalid argument" 2014-07-27 00:05:40 bye 2014-07-27 05:11:25 hi 2014-07-27 05:12:21 does anybody know, if quagga 0.99.23 will be official included in alpine 2.7.9? 2014-07-27 05:13:38 there is a bug in 0.99.22-4 which forces me to upgrade to alpine 3 on all machines 2014-07-27 05:17:50 https://bugzilla.quagga.net/show_bug.cgi?id=799 2014-07-27 06:27:08 Hello 2014-07-27 06:28:02 I am trying to compile suricata on Alpine Linux, I get typedef error cpu_t_set cpuset_t at util-affinity.h 2014-07-27 06:28:21 Are there apkbuids for suricata and xombrero 2014-07-27 09:14:04 hiy guys, I'm trying to start 3.0.1 in Virtual Box but I get "boot: Booting kernel failed: Invalid argument" when booting. Any hints? 2014-07-27 15:16:05 Hi 2014-07-27 15:16:28 I have a process stuck /bin/busybox mount /dev/sdc1 /mnt/ , because of a usb storage error : usb-storage: Error in queuecommand_lck: us->srb = c1c67400 2014-07-27 15:16:32 how do I force kill this process 2014-07-27 15:16:34 kill -9 won't do 2014-07-27 15:19:09 zxd: you can't kill a process stuck in kernel io 2014-07-27 15:19:15 you can try umount -l /mnt 2014-07-27 15:19:24 but i think it will not work 2014-07-27 15:19:47 it's STAT is D 2014-07-27 15:19:48 in ps 2014-07-27 15:20:06 crap 2014-07-27 17:14:23 hi 2014-07-27 17:14:52 is here anybody there who is familar with alpine quagga package? 2014-07-27 17:15:10 is there a chance to get 0.99.23 running on alpine 2.7.9 ? 2014-07-29 09:56:22 any suggestions for upgrading from 2.7 to 3.0.1 ? 2014-07-29 09:59:21 wouldn't do it if you need nfs ;-) 2014-07-29 10:30:15 I don't use NFS 2014-07-29 10:38:20 "apk add busybox-static" in case something goes wrong. 2014-07-29 10:40:58 goes wrong = you end up with musl libc and busybox linked against uclibc, rendering it useless 2014-07-29 10:41:59 yea somehow I managed to hose apk 2014-07-29 10:42:17 and of course I did this without testing on my most complex box.. woo for two partitions 2014-07-29 10:43:59 apk-tools-static - forgot about those 2014-07-29 10:45:47 i didn't do upgrade from 2.7 to 3.0 myself. i switched from slackware to alpine edge, after 3.0 release 2014-07-29 11:37:58 NameBrand: apk add busybox-static apk-tools-static 2014-07-29 11:38:12 the busybox.static will help you if things goes bananas 2014-07-29 11:38:21 argh... what's up with my network today!? 2014-07-29 11:38:29 then edit /etc/apk/repositories replacee v2.7 with v3.0 2014-07-29 11:38:47 then apk.static update 2014-07-29 11:39:06 apk.static upgrade --no-selfupgrade --simulate 2014-07-29 11:39:18 and finally: apk.static upgrade --no-selfupgrade 2014-07-29 12:46:57 seems I have to launch sshd manually after that 2014-07-29 20:44:01 hey 2014-07-29 20:44:23 can someone help me with fancontrol? 2014-07-29 20:45:16 i got lm_sensors and sensord started as a service 2014-07-29 20:45:33 now i´m not sure if i need also fancontrol 2014-07-29 20:46:04 i cannot start it without /etc/fancontrol 2014-07-29 20:46:21 is there a simple was to get a defalt /etc/fancontrol ? 2014-07-29 20:50:18 pwmconfig? for creating /etc/fancontrol? 2014-07-29 20:52:25 melville: right... just came over it 2014-07-29 20:52:55 melville: but "there are no pwm-capable sensors moduels installed" ... 2014-07-29 20:58:41 is "$ sensors" showing anything (elevated priviliges are need for it)? 2014-07-29 20:59:24 melville: yes it does 2014-07-29 21:01:28 it is showing temp or also fans? 2014-07-29 21:01:57 melville: only temp 2014-07-29 21:04:23 i'm not sure if i can help you beside that - maybe you need to load additional kernel modules or pass some options to kernel module. depends on hardware. google? 2014-07-29 21:04:52 did you use sensors-dectect? it is in separate package. 2014-07-29 21:05:13 melville: thx 2014-07-29 21:05:20 melville: yes i used it 2014-07-29 21:05:42 melville: it just found one module to load 2014-07-29 21:05:47 melville: "coretemp" 2014-07-29 21:06:58 just out of curiosity, is it laptop? 2014-07-29 21:07:34 melville: no, its a HP proliant microserver 2014-07-29 21:10:31 melville: thx for the hints. i think i will life with the noise 2014-07-30 06:36:34 Hi 2014-07-30 11:56:46 yesterday, my whole USB drive got corrupted 2014-07-30 11:57:18 aw :-( 2014-07-30 11:57:21 had been rebooting prolly 50 times, no problem. but I powered down, returned some hours later, and would not recognize anything. fsck doesn't work 2014-07-30 11:57:27 haven't tried testdisk yet 2014-07-30 11:57:34 I was running frugal with savefile 2014-07-30 11:57:48 known phenomena? 2014-07-30 11:58:28 not really 2014-07-30 11:58:39 also, I have a little bug for you. in /etc/init.d/slim, it spells "needs" instead of "need" 2014-07-30 11:58:54 might be why it didn't pull dbus, because slim would not start, and it was lacking dbus 2014-07-30 11:59:35 wouldn't 'require' be more appropriate? 2014-07-30 11:59:45 as a wording 2014-07-30 11:59:52 this is OpenRC, their wording 2014-07-30 12:00:08 and firefox' been very crashy. I was running edge with FF 31.0 2014-07-30 12:00:25 I know I should file bugs, but I usually don't before I'm and running with a distro 2014-07-30 12:04:09 i use ff 31.0 2014-07-30 12:04:26 the only issue i have is with xorg and intel dri driver 2014-07-30 12:04:46 you use alpine as your everyday distro? 2014-07-30 12:04:55 i have to: find /sys -type d -exec chmod o+rx {} \; 2014-07-30 12:04:56 yes 2014-07-30 12:05:11 it was crashing on me every ~3min of muse 2014-07-30 12:05:13 both on my work desktop and on my private laptop 2014-07-30 12:05:17 use* not amuse :) 2014-07-30 12:05:22 :) 2014-07-30 12:05:32 do you have anything in dmesg? 2014-07-30 12:05:38 in Xorg.0.log? 2014-07-30 12:05:44 any indication what goes wrong? 2014-07-30 12:05:52 didn't look. I should've. it's corrupted now 2014-07-30 12:06:13 I remember starting FF from terminal, and it was spitting out long error msgs from time to time 2014-07-30 12:06:28 I'll try testdisk ... 2014-07-30 12:06:49 no, fk my work, I'll reinstall :) 2014-07-30 12:07:25 I also &'ed chrony, whatta ya think of that? it was very annoying not having desktop starting before time was updatd 2014-07-30 12:07:28 updated* 2014-07-30 12:10:21 and for dhcp, do you have any was to quicken things up? udhcpc is darn slow, dhcpcd with noarp (https://wiki.archlinux.org/index.php/Dhcpcd#Speed_up_DHCP_by_disabling_ARP_probing) is faster, but compared to (the dreadful around here) systemd-networkd (<1ms), well ... 2014-07-30 12:10:25 way 2014-07-30 12:10:27 * 2014-07-30 12:10:43 https://plus.google.com/+TomGundersen/posts/eztZWbwmxM8 2014-07-30 12:10:58 it's not reliable atm, but darn that's fast 2014-07-30 12:12:44 Aspi: "firefox -safe-mode" solved the crashes for me 2014-07-30 12:14:26 :) some addons (Self Destructing Cookies) did thwarten things up, yes. so maybe disabling them would help around, didn't think of that 2014-07-30 12:15:17 I got some strage issues with a ADSL-modem 2014-07-30 12:16:01 unfortunately i cant do something right no because its a production system so i will do something later 2014-07-30 12:16:21 i got a d-likg 321B ADSL-Modem 2014-07-30 12:16:40 this works great with a router or my macbook attached 2014-07-30 12:17:00 but when i want to attach the alpine-linux-server 2014-07-30 12:17:17 when does alpine pivot_root? after the [#####] is done? 2014-07-30 12:17:49 and i want to connect via dhcp to the modem i got "RTNETLINK answers: Network is unreachable" 2014-07-30 12:18:49 anybody a hint what i have to look for? google doesn´t help me a lot right now 2014-07-30 12:18:52 ncopa, just out of curiosity what "$ groups" you have for user. i've some suspicions that crashes could be related to insufficient privileges (not talking about grsec issues on edge). 2014-07-30 12:19:27 (firefox never ran as a user here. I had to run it as root. yuck!) 2014-07-30 12:20:32 ncopa adm wheel audio cdrom video netdev readproc kvm vde2-net abuild 2014-07-30 12:20:40 readproc might make difference 2014-07-30 12:20:44 that is gsec problem in edge Aspi - newer xorg drivers are touching sysfs, where grsec is setting permissions 2014-07-30 12:20:49 it allows me to read /proc 2014-07-30 12:20:58 mirac: "Network is unreachable" is some ICMP return message. it's some router telling you it can't reach the network. CORRECT ME WHERE I'M WRONG ;) 2014-07-30 12:21:31 mirac problem sounds like dhcp client not waiting long enough? 2014-07-30 12:22:04 melville: but i get an ip from my ISP 2014-07-30 12:22:16 mirac: https://en.wikipedia.org/wiki/ICMP_Destination_Unreachable#Destination_unreachable 2014-07-30 12:22:19 prolly 2014-07-30 12:22:45 idk, I very might be wrong 2014-07-30 12:23:20 melville: ah, ok. remember reading something about X11 becoming nonroot, yes. come forth, Wayland, come 2014-07-30 12:30:49 mirac: here's something as well http://www.wildpackets.com/resources/compendium/tcp_ip/unreachable 2014-07-30 12:31:14 I got that msg when the address was blocked, I think 2014-07-30 12:31:33 no, idk, don't remember 2014-07-30 12:40:51 anyone use any automounting solution? 2014-07-30 12:40:59 well, other than thuar 2014-07-30 12:41:04 n 2014-07-30 12:55:41 it's very strange that the whole USB disk just corrupted. at that poweroff, I didn't even save anything, just discarded it all. 2014-07-30 12:56:22 testdisk found a partition (type Linux), but was not able to recover anything from the FS 2014-07-30 12:56:23 fs 2014-07-30 13:31:32 hey... just got disconnected 2014-07-30 13:31:46 one simple question. 2014-07-30 13:32:12 ist there a way to bring up/down only one interface form /etc/network/interface ? 2014-07-30 13:33:06 i got to bridges in there and when i restart init.d/interfaces they don´t come up well 2014-07-30 13:33:10 ip i set dev down 2014-07-30 13:33:25 ifup eth0 2014-07-30 13:33:28 ifdown eth0 2014-07-30 13:33:29 :) 2014-07-30 13:33:49 Aspi1: thx. does it also dhcp when this is in the interfaces? 2014-07-30 13:33:55 s/ip i/ip l/ O:) 2014-07-30 13:33:59 ifup/ifdown requires that interface is configured in /etc/network/interfaces 2014-07-30 13:34:06 no. ncopa's way is better 2014-07-30 13:34:14 ncopa: thx 2014-07-30 13:34:20 i will try later... 2014-07-30 13:35:21 also, it should be `ip link set dev up/down`. whatever, look around 2014-07-30 13:35:28 folks, how compatible is musl with glibc? 2014-07-30 13:35:48 http://www.musl-libc.org/faq.html 2014-07-30 13:35:58 is it correct that a lot of stuff doesn't compile/is buggy with musl? 2014-07-30 13:36:06 Aspi1: I've read that:) 2014-07-30 13:36:20 there are some stuff that does not compile with musl 2014-07-30 13:36:46 most builds out of the box 2014-07-30 13:36:54 others builds with minor patching 2014-07-30 13:37:13 others again needs major refactoring (like nfs-utils) 2014-07-30 13:37:30 mostly it fails during buildtime 2014-07-30 13:37:56 but it happens it fails during runtime too 2014-07-30 13:38:20 ouch 2014-07-30 13:38:30 how common is this in practice? 2014-07-30 13:38:33 not too often 2014-07-30 13:38:51 sysklogd 2014-07-30 13:39:01 there was one i saw today 2014-07-30 13:39:05 ulogd something 2014-07-30 13:39:37 quagga i think 2014-07-30 13:39:43 hm 2014-07-30 13:39:52 tbh it sounds kinda iffy:) 2014-07-30 13:40:01 glibc sucks big time though.. 2014-07-30 13:40:07 nfdump 2014-07-30 13:40:13 those are the packages we have fixed 2014-07-30 13:40:30 and often those issues could have showed up on glibc too 2014-07-30 13:40:42 often it is real bugs that glibc covers 2014-07-30 13:41:34 i'd say we have found less than 10 of 1500 packages or so til now 2014-07-30 13:41:44 nice! 2014-07-30 13:42:03 there are likely some more 2014-07-30 13:42:14 obviously:) 2014-07-30 13:42:20 overall, the general impression is 2014-07-30 13:42:28 musl does not like ugly code 2014-07-30 13:42:44 so its mostly ugly code that breaks 2014-07-30 13:42:53 high quality code tends ro run 2014-07-30 13:42:59 to* 2014-07-30 13:43:13 problem is, there's quite some ugly code around:) 2014-07-30 13:43:18 yeah :) 2014-07-30 13:43:41 and it's getting worse;[ 2014-07-30 13:43:44 i think musl does good though 2014-07-30 13:43:57 neat 2014-07-30 13:44:11 was far less than originally expected 2014-07-30 13:44:11 ACTION is going to have a look at the pkg creation process 2014-07-30 13:44:38 DEFAULT_CIPHER doesn't work in lbu. ENCRYPTION does 2014-07-30 13:44:54 i think there are some things that might break with glibc that will work with musl 2014-07-30 13:45:12 probably 2014-07-30 13:45:13 and i know for sure that there are some things that are broken in uclibc that works with musl 2014-07-30 13:45:21 vlc for example 2014-07-30 13:45:30 threading cornercases 2014-07-30 13:45:41 http://wiki.alpinelinux.org/wiki/APKBUILD_examples:Simple nice! 2014-07-30 13:46:22 yeah 2014-07-30 13:46:33 tried to keep it visible what is going on 2014-07-30 13:46:44 the package splitting is not that obvois though 2014-07-30 13:47:03 I'm used to crux's Pkgfiles 2014-07-30 13:47:08 ah 2014-07-30 13:47:12 i did look at those 2014-07-30 13:47:16 re: musl differences - http://wiki.musl-libc.org/wiki/Functional_differences_from_glibc 2014-07-30 13:49:04 ncopa: do you use adblock? it was about the only addon I had 2014-07-30 13:49:23 generally build errors are due to application is missing #include's and it works on glibc due to other system headers "leaking" the definitions used 2014-07-30 13:49:25 Aspi1: yes, Adpblock Plus 2014-07-30 13:49:59 mkay. I'm setting up a new env. will report 2014-07-30 15:14:20 Afternoon all 2014-07-30 15:16:07 I don't have much time :( but I was wondering if anyone had used cjdns? Any thoughts on it? 2014-07-30 15:17:18 is alpine-desktop doing any power saving? 2014-07-30 15:19:50 corrupt again. seriously, wtf 2014-07-30 15:19:57 might be the usb disk 2014-07-30 15:37:45 anything in dmesg before corruption did happen? i don't use any automounters and i'm probably bit overprotective by using sync(1), when copying to external device and then doing manual unmount. 2014-07-30 15:41:10 the reason I suspect the disk, is that this time I was just moving apkvols around, actually in Windows for a moment. it worked, I wrote the data, I safe ejected and the pulled the drive. it wouldn't boot, and when I put in Windows machine again, it wouldn't recognize it either 2014-07-30 15:59:00 ncopa: firefox crashed. PAX and grsec. "denied resource overstep by requesting 4096 for RLIMIT_CORE against limit 0 for /usr/lib/firefox-29.0.1" and more 2014-07-30 15:59:42 only have firefox installed, and could be noted I'm testing fonts, ttf-ubuntu-font-family at the crash 2014-07-30 16:01:53 Aspi unimportant warning, since it only means that firefox couldn't dump a core (ulimit -c). 2014-07-30 16:02:31 I'm not running this in xfce, but shouldn't make any difference? 2014-07-30 16:02:58 don't think so 2014-07-30 16:04:15 you could set ulimit -c unlimited to get core dump, not sure how useful would that be 2014-07-30 16:30:32 p/buffer 22 2014-07-30 16:30:34 oops 2014-07-30 16:30:52 Aspi: no powersaving... 2014-07-30 16:33:48 I was toying around with cpufreq 2014-07-30 16:34:39 is it possible to do "if speedstep is available, load one module. else, load another"? 2014-07-30 16:35:23 like "try and load intel_pstate. if failed, load p4-clockmod" 2014-07-30 17:13:17 Hey. when I install an acf package by command line the web portal doesn't add the package to view until I logoff and logon in the portal. Anyone knows a way of refreshing the view by command line so that the packages appears in the portal without having to logoff and logon? 2014-07-30 17:40:46 ncopa: I think ff might started crashing after installing HTTPS Everywhere 2014-07-30 17:41:12 also, running it from cmdline yields "Failed to find nss3 in installed directory, checking system paths". okay? 2014-07-30 17:41:32 is that nothing to worry about* 2014-07-30 18:05:11 Hi everyone 2014-07-30 18:05:36 I'm trying to boot X86 alpine off of a CD 2014-07-30 18:05:47 However, it just hangs on ISOLINUX 2014-07-30 18:06:13 I've burned a couple CDs and DVDs and its still happening 2014-07-30 18:06:20 Any idea as to why 2014-07-30 18:12:20 ? 2014-07-30 18:22:52 urras, I remember someone reporting similar issue here few days ago - can't check, since i erased logs. 2014-07-30 18:23:27 i think he used usb in the end 2014-07-30 18:24:20 what version do you use and which mirror? 2014-07-30 18:24:58 alpine-3.0.1-x86 2014-07-30 18:42:17 melville: Its still doing it on 3.0.2 2014-07-30 18:45:39 I'm going to try to USB boot it 2014-07-30 18:59:22 urras, you should report it, especially since it doesn't seems unique http://bugs.alpinelinux.org/projects/alpine/ 2014-07-30 19:06:37 Alright, its at http://bugs.alpinelinux.org/issues/3273 2014-07-30 19:10:46 Updated by Jeff Bilyk 3 minutes ago: Is this on hardware setup or a VM? 2014-07-30 20:05:13 Alright, I'll see if unetbootin to USB works 2014-07-30 20:07:18 Huh 2014-07-30 20:07:46 Now it just says "selected boot device not available 2014-07-30 20:17:22 urras if you have an older alpine installation you could do setap-bootable 2014-07-30 20:18:03 http://wiki.alpinelinux.org/wiki/Create_a_Bootable_USB#Manual 2014-07-30 20:18:12 manual way worked for me last time, 2014-07-30 20:18:23 setup-bootable should be fixed with 3.0.2 2014-07-30 20:24:25 So if I install 3.0.2 to USB with unetbootin it should work? 2014-07-30 20:33:14 dont know i dont use unetbootin 2014-07-30 20:36:14 i used unebootin for edge cd like a month ago - worked 2014-07-30 21:00:49 melville: Cant seem to find an edge iso 2014-07-30 21:00:56 Perhaps you know where they are? 2014-07-30 21:08:50 http://nl.alpinelinux.org/alpine/edge/releases/x86/ 2014-07-30 21:10:49 which seems to be a bit older than 3.0.0 2014-07-30 21:54:00 melville: Still no luck 2014-07-31 01:26:24 Finally got it to install 2014-07-31 01:26:51 I had to get an extra laptop hard drive, swap it out, and install Alpine on it 2014-07-31 01:27:15 Then I had to hotswap it into my desktop, where I made a clonezilla image of it and installed it on my Desktop HDD 2014-07-31 01:28:16 The Alpine way℗ ©™ 2014-07-31 05:50:27 how do you get a runlevel you created yourself to show up in rc-status? 2014-07-31 05:50:38 -a ? 2014-07-31 05:51:02 oh haha 2014-07-31 08:46:49 ncopa: I suspect my ff crashing problems was due to using awesome. it uses xcb. 2014-07-31 08:48:01 hm ok, i would not expect that trigger crashes in ff 2014-07-31 08:48:16 would be nice if you could figure out where it craches 2014-07-31 08:48:28 in what lib 2014-07-31 08:49:23 after the crash, console output was: https://rof.li/p/1436 2014-07-31 08:50:25 make that https://rof.li/p/1437 2014-07-31 09:04:30 that could be awesome or xcb killing firefox yes 2014-07-31 09:05:25 fyi, I tried with alpine-desktop, and didn't die for 15min use 2014-07-31 09:05:53 I'll see, maybe I need to go with xfce instead 2014-07-31 09:06:05 sorry about that 2014-07-31 09:06:13 but it does smell like a threading issue with awesome 2014-07-31 09:07:24 you think xcb is involved? 2014-07-31 09:07:51 looks like that 2014-07-31 09:07:57 [xcb] Most likely this is a multi-threaded client and XInitThreads has not been called 2014-07-31 09:09:08 I'm not much of a fanboy, screw awesome, I just love using a lightweight tiling wm 2014-07-31 09:09:23 maybe I'll look into i3 or dwm 2014-07-31 09:09:36 i think uggedal pushed a tiling wm recently 2014-07-31 09:09:51 dont remember the name of it though 2014-07-31 09:10:52 I realize it's turbulent times in X world. look forward to some some years ahead 2014-07-31 09:11:35 herbstluftwm 2014-07-31 09:11:44 just found it as well :) 2014-07-31 09:16:45 nice name 2014-07-31 09:17:04 https://bbs.archlinux.org/viewtopic.php?pid=1005392#p1005392 :_ 2014-07-31 09:19:57 most icons are missing in xfce 2014-07-31 09:20:16 even if you select the icon theme? 2014-07-31 09:20:24 i have seen that issue 2014-07-31 09:20:30 pretty sure 2014-07-31 09:20:31 icons does not show by default 2014-07-31 09:20:33 i dont know 2014-07-31 09:20:39 i dont know why it happens 2014-07-31 09:21:10 kay. strange 2014-07-31 09:21:34 it is normally fixed if you open settings -> appearance -> icons and select icon theme 2014-07-31 09:26:42 ff never crashes when idle, just when interacting. seems to crash most at new page or heavy page. 2014-07-31 09:27:08 just crashed in xfce. didn't start from terminal -.- 2014-07-31 09:27:50 ah, I think grsec killed it. the same 4096 RLIMIT_CORE I mentioned yesterday 2014-07-31 09:51:48 ncopa: icon bug happens because /etc/xdg/xfce4/xfconf/xfce-perchannel-xml/xsettings.xml has no IconThemeName set. idk if that should be set manually, or a program should set it, but setting type="string" value="Tango" yields likeable results 2014-07-31 09:56:07 ha 2014-07-31 09:56:10 i think you are right 2014-07-31 09:56:47 yesh, I could be of help B) 2014-07-31 09:56:57 wonderful 2014-07-31 09:57:07 could you please create a ticket for it? 2014-07-31 09:57:10 so i dont forget.. 2014-07-31 09:57:25 hahaha 2014-07-31 09:57:27 this is what fedora does: http://pkgs.fedoraproject.org/cgit/xfce4-settings.git/tree/xfce4-settings-4.9.4-fedora.patch 2014-07-31 09:57:34 I'll write it down 2014-07-31 09:57:39 thanks 2014-07-31 09:57:52 i tend to forget things that is only mentioned in irc 2014-07-31 09:58:11 unless i drop what i currently have in my hands and just fix it immediatly 2014-07-31 09:58:52 ha! 2014-07-31 09:58:53 that's called singletasking. prolly a factor of alpine being functional? :) 2014-07-31 09:59:01 https://projects.archlinux.org/svntogit/packages.git/tree/trunk/xfce4-settings-xml-4.10.0.patch?h=packages/xfce4-settings 2014-07-31 09:59:05 archlinux also patch it 2014-07-31 09:59:13 thats why it works out of the box for those 2014-07-31 09:59:16 and not for us 2014-07-31 09:59:38 i have been wanting fix that for long time but didnt know where to look 2014-07-31 09:59:55 idk if xfce should choose the first available icon theme, or let the user choose. it does autoselect for normal theme, no? 2014-07-31 10:00:30 i think i patch it to pick a default theme 2014-07-31 10:00:44 the retarded bug is really that I couldn't choose Tango if no other options was available -.- so I had to add faenza first 2014-07-31 10:00:59 but apparently it does not pick icontheme by default 2014-07-31 10:01:09 thats why other distros are patchin it 2014-07-31 10:01:32 then I shamelessly used the Settings Browser app, but realized it must've saved it in ~/.config. and so, I `find / -name xsettings.xml` and found it 2014-07-31 10:02:09 i dont know if we want use Tango icon theme as default 2014-07-31 10:02:11 it does not pick anything by default. hmm. I tend to lean more and more towards dynamic solutions 2014-07-31 10:02:19 Tango lacks some icons, like Run 2014-07-31 10:02:25 ok 2014-07-31 10:02:33 i'm open for other suggestions too 2014-07-31 10:02:48 maybe someone has an opinion. I haven't 2014-07-31 10:02:55 arch uses 'gnome' 2014-07-31 10:03:02 fedora has a "Fedora" icon theme 2014-07-31 10:03:12 ok its lunch 2014-07-31 10:03:15 thanks! 2014-07-31 10:03:22 np :) 2014-07-31 10:03:35 (little less annoyed now? haha) 2014-07-31 10:05:37 my only say is that having a small icon theme apk would be nice, I imagine they can grow pretty huge (https://www.archlinux.org/packages/community/any/faenza-icon-theme/) 2014-07-31 10:11:47 Aspi: I've packaged herbstluftwm for alpine (it's a manual tiler though) 2014-07-31 10:12:43 nevermind, replied before reading the whole scrollback 2014-07-31 10:13:14 it seems nice, but alpine will not be my primary distro, not sure I want to spend the work 2014-07-31 10:13:58 much work required to setup? 2014-07-31 10:13:58 I really like it (used dwm, awesome, i3, ratpoison, bspwm before) 2014-07-31 10:14:41 Aspi: depends on what you wan't. the default config is ok. mine is here: http://git.uggedal.com/dotfiles/tree/.config/herbstluftwm 2014-07-31 10:15:14 I'll try it again 2014-07-31 10:15:46 but if you don't want to configure things i3 is probably the easiest 2014-07-31 10:16:18 I figured i3 uses XCB as well, so no-go for now 2014-07-31 10:16:36 the multithreading issue, is it a quick or extensive fix? 2014-07-31 11:11:42 Aspi: dont know since I dont know what the exact problem is 2014-07-31 12:06:23 for icon fix in desktop I could suggest meta pkg 2014-07-31 12:06:58 eg. alpine-desktop-xfce-faeza or alpine-desktop-xfce-tango 2014-07-31 12:07:31 these pkgs would fix or take care of any missing entry in config files 2014-07-31 12:08:15 i think we should have a sane default 2014-07-31 12:09:28 default means when desktop is installed the default theme/icon-set would install too 2014-07-31 12:10:48 as long as un-installing themes is easy 2014-07-31 12:16:07 ok, the meta pkgs may become complicated, i.e preserving old settting eg. if one installs faeza then does some custom settings and installs tango 2014-07-31 12:16:32 yep, it's just setting a default instead of no default set at all. one can choose from the xfce menu no different from before 2014-07-31 12:18:56 and i think i want drop a meta package 2014-07-31 12:19:06 and use s setup script instead for desktop 2014-07-31 12:19:18 becase with meta package you cannot uninstall if you want replace component 2014-07-31 12:21:09 so setup-desktop-xfce script that takes command line options ? 2014-07-31 12:21:24 like theme=tango 2014-07-31 12:22:01 yeah 2014-07-31 12:22:57 it likely be setup-desktop -e xfce 2014-07-31 12:23:21 i havent really planned for setting up a default systemwide icontheme 2014-07-31 12:23:56 this would create .xfce in /etc/world ? 2014-07-31 12:24:23 i think i'll still have the xfce4 meta package 2014-07-31 12:24:29 just not the alpine-desktop meta package 2014-07-31 12:24:43 coz un-installing the desktop and adding anther should be easy 2014-07-31 12:25:09 I was looking around for icon themes. Tango seems like a reasonable choice. not too large (<2MB) and liked by some Tango standard 2014-07-31 12:25:14 and looks okay imo 2014-07-31 12:25:38 what's hicolor-icon-theme? 2014-07-31 12:26:31 these two also seems okay, but I must say I prefer tango: 2014-07-31 12:26:31 https://www.archlinux.org/packages/community/any/tangerine-icon-theme/ 2014-07-31 12:26:31 https://www.archlinux.org/packages/community/any/human-icon-theme/ 2014-07-31 16:52:48 what about moka-icon-theme xfce theme? 2014-07-31 16:53:24 https://github.com/moka-project/moka-icon-theme 2014-07-31 19:01:05 I'm looking at Linux equivalents of SmartOS and Alpine looks ideal. How secure is the LXC stuff though? I've heard conflicting reports but with GRSEC as standard Alpine seems to take security seriously. 2014-07-31 19:03:32 I was also wondering if anyone was using ZFSonLinux for the permanent storage in a data mode install? Can this work?