2014-06-01 21:26:47 hello 2014-06-01 22:02:28 hi all 2014-06-01 22:06:54 anyone here? 2014-06-02 01:45:16 hi all 2014-06-02 05:08:47 anyone here? 2014-06-02 05:17:49 DJDave, yes, many are in UTC+1/+2 so they'll be here in an hour or two 2014-06-02 06:23:16 hi 2014-06-02 06:23:39 Anaphaxeton: pong 2014-06-02 07:14:43 morning 2014-06-03 04:03:04 clear 2014-06-03 04:03:05 oops 2014-06-03 04:03:09 wrong window 2014-06-03 04:03:31 has anyone been using freeradius on an edge machine? 2014-06-03 04:03:48 for some reason it's saying the config is not OK even though the command that checks the config didn't return anything 2014-06-03 04:05:58 oh 2014-06-03 04:12:37 yeah never mind 2014-06-03 04:12:41 found it 2014-06-03 13:02:57 Looking for a little help, after install my startup gets stuck on "Loading hardware drivers" 2014-06-03 13:09:47 Ne0nd0g: sounds like kernel has problems with the hardware 2014-06-03 13:10:01 a driver is having problem 2014-06-03 13:10:10 right, wondering how I overcome it. When running just on live CD everything works fine 2014-06-03 13:10:16 hm 2014-06-03 13:10:22 thats a bit strange 2014-06-03 13:10:32 what livecd do you run? 2014-06-03 13:10:41 what version that is 2014-06-03 13:11:21 I was running the xen-2.7.8 x64 live cd. Right now I'm about to do a clean install with the standard x64 image 2014-06-03 13:12:18 normally it gets trickier in xen, not the other way run. heh. 2014-06-03 13:12:28 at boot prompt, you could try: noquiet versbose 2014-06-03 13:12:38 add that i mean 2014-06-03 13:12:45 so: grsec noquiet verbose 2014-06-03 13:12:51 and see if it gives any hints 2014-06-03 13:13:11 what kind of hardware is it? 2014-06-03 13:13:52 It's a Lenovo W520 2014-06-03 13:15:03 some of the drivers are broken i think 2014-06-03 13:15:26 for the Lenovo? 2014-06-03 13:15:38 yes, for your hardware 2014-06-03 13:15:48 it could be a problem in vanilla kernel 2014-06-03 13:15:54 or it could be triggered by grsecurity patch 2014-06-03 13:16:11 looks like it is stuck on USB hub 2014-06-03 13:16:20 might be the USB 3 ports 2014-06-03 13:16:47 i think we need a boot image with vanilla kernel 2014-06-03 13:17:29 usb 3-1: New USB device found, idVendor=8087, idProduct=0024 2014-06-03 13:17:47 will the "standard" image contain the vanilla kernel? 2014-06-03 13:17:51 no 2014-06-03 13:18:07 im thinking of an alpine-vanilla.iso 2014-06-03 13:18:26 which will be similar as mini, but with vanilla kernel 2014-06-03 13:19:24 Is alpine-vanilla.iso an image that exists or something that needs to be developed/created? 2014-06-03 13:19:34 something that needs to be created 2014-06-03 13:19:46 Is there no way to inject the drivers during the install? 2014-06-03 13:20:06 i dont remember really... 2014-06-03 13:20:13 i wonder if we have blacklist=.... 2014-06-03 13:20:16 What is that allows it to function when running from a liveCD that won't let it run after install? 2014-06-03 13:20:34 was it the standard livecd? 2014-06-03 13:20:43 or xen livecd 2014-06-03 13:20:48 Xen livecd 2014-06-03 13:21:02 then it runs in xen 2014-06-03 13:21:20 and you did the install from there too? 2014-06-03 13:21:25 yup 2014-06-03 13:21:33 hm 2014-06-03 13:21:49 so it should installed a xen hypervisor 2014-06-03 13:22:15 Yeah, my boot menu shows "Xen + Linux" 2014-06-03 13:22:18 and it shold be same version of the kernel as the livecd 2014-06-03 13:22:32 3.10.40 2014-06-03 13:29:25 is there a way to tell the system to boot without trying to load the drivers for USB? 2014-06-03 13:34:58 interestingly enough, I tried to run up the "standard" iso and it also freezes on "Loading Hardware". I put back in the Xen live CD and it boots up fine 2014-06-03 13:35:20 can I find out what the Xen liveCD is doing and have my installed version do the same? 2014-06-03 13:36:55 I looked through dmesg and it just keeps moving past the spot where it gets stuck on normal boot 2014-06-03 13:42:08 in the mean time i'll switch my install from sys to data I guess 2014-06-03 13:42:30 Ne0nd0g: one thing i wanted to note is that xen enables much less power management than normal kernel 2014-06-03 13:43:06 (unless you add cpuidle xen opt and the xen_acpi_processor module and some more things i don't know yet) 2014-06-03 13:43:20 so it might be getting stuck on a power management issue 2014-06-03 13:47:49 thanks for the info darkfaded 2014-06-03 13:48:28 so do you think it's getting stuck on power management when booting the installed "Xen + Linux" kernel? 2014-06-03 13:54:11 i also can't explain why it gets stuck with that one and not the livecd 2014-06-03 13:54:12 say 2014-06-03 13:54:17 how much time did you give it 2014-06-03 13:54:35 i know my systems sometimes take a minute to get over that point 2014-06-03 13:57:00 I went for a 20 minute walk and came back 2014-06-03 13:57:59 ok that counts 2014-06-03 13:58:55 is it possible to see what the boot statment is for the live cd and modify my installed version to match that? 2014-06-03 14:42:28 I added "nomodeset" to my boot options and everything loads up fine 2014-06-03 14:42:39 what does nomodeset do? 2014-06-03 14:50:28 Ne0nd0g: ask google? 2014-06-03 14:50:37 yeah, I got it 2014-06-03 14:56:50 we all have google 2014-06-03 15:02:20 i also had to add that... but i saw some screen flicker to hint me 2014-06-03 15:02:23 loooong ago 2014-06-03 15:02:34 it doesn't off the linux framebuffer smarties 2014-06-03 15:02:42 *turns off 2014-06-03 15:03:10 I'm having a battle with the onboard intel video chipset and the Nvidia Quadro chipset 2014-06-03 15:03:13 i have this: 2014-06-03 15:03:14 default_kernel_opts="quiet vga=normal nomodeset reboot=pci" 2014-06-03 15:03:33 how do I set that? 2014-06-03 15:03:34 reboot pci because otherwise it stucked at "requesting system reboot" 2014-06-03 15:04:24 yeah, mine get stuck at reboot too 2014-06-03 15:08:35 ok there a have it 2014-06-03 15:08:38 ;)) 2014-06-03 15:17:37 Ne0nd0g: sorry, didn't see your question. it goes to /etc/update-extlinux.cfg 2014-06-03 15:17:41 and then update-extlinux 2014-06-03 15:17:59 it should show up as you set it in /boot/extlinux.conf 2014-06-03 15:26:01 Thanks darkfader 2014-06-03 15:28:40 i need to get amd64-libs installed, does anyone know how to do that? It isn't in the repository 2014-06-03 15:29:09 what do those o? 2014-06-03 15:29:25 runtime libraries for amd64 architecture 2014-06-03 15:29:36 I think it is why i am having trouble installing my nvidia drivers 2014-06-03 15:30:33 ah, ok never tried gfx 2014-06-03 15:39:52 alpine is debain based? 2014-06-03 15:51:05 Ne0nd0g: check our about page 2014-06-03 15:54:16 I did, briefly skimmed over it and searched for debain 2014-06-04 08:38:37 Morning all! 2014-06-04 08:39:11 I realise this is a bit delayed, but I've been testing 3.0 rc3 and have no issues to report thus far 2014-06-04 08:39:30 ginjachris, thanks for report! we are planning to tag 3.0 release today 2014-06-04 08:39:43 we got the ipsec/gre/dmvpn stuff sorted out already too 2014-06-04 08:39:51 known issue is nfs apparently 2014-06-04 08:39:51 awesome! 2014-06-04 12:38:43 Hello again all, does anyone have any experience with unbound? I'm trying to figure out what "sendto failed: Resource temporarily unavailable" means in my logs. Online searches have only provided vague information 2014-06-04 12:39:14 Basically I'm trying to figure out if this is a problem my end, or something beyond my control. Any ideas will be appreciated 2014-06-04 12:39:23 and hello algitbot :) 2014-06-04 12:39:31 ginjachris, you have local firewall preventing outbound dns packets 2014-06-04 12:40:05 oh i should explain - no firewall.....and it's not a constant error. Just sometimes, Randomly it seems 2014-06-04 12:41:17 heavy load? perhaps nic outbound queue is full? 2014-06-04 12:41:59 oooh could be a nic problem i guess. Had only checked unbound queue, which has no jostles etc (i.e. no load issues) 2014-06-04 12:42:42 any idea on the best way to check interface next time it happens? 2014-06-04 12:42:58 (assuming i see it happen, which may not be possible) 2014-06-04 12:43:13 check ifconfig statistics on inc, and ethtool -S 2014-06-04 12:45:39 thanks fabled! I'll see if I can catch it next time, although it seems to happen in middle of night etc....so i initially thought it could be dos/flood attempt, but unbound stats suggest it isn't 2014-06-04 12:46:28 Hooray for v3! 2014-06-04 12:46:50 :-) 2014-06-04 12:48:10 I wish I could get a full time job working with Alpine. Most other stuff bores me to tears now and/or is lame 2014-06-04 12:49:02 if only it would be so simple :p 2014-06-04 12:49:51 yep. And simple is best! (My colleagues never believe me until their setup becomes so complicated they are unable to explain it or troubleshoot it lol) 2014-06-04 12:50:38 oh well, such is life I suppose 2014-06-04 14:06:04 \o/ 2014-06-04 14:06:06 woohooo 2014-06-04 14:07:35 ncopa: which kernel is shipped in 3.0? 2014-06-04 14:07:55 3.14.5-grsec 2014-06-04 14:10:21 (is there any chance that it will be upgraded to 3.15 later?) 2014-06-04 14:13:20 no, you would need a good reason 2014-06-04 14:13:32 and even if it, it would hit backports repository 2014-06-04 14:17:44 kocka: you can pin edge or newer release in /etc/apk/repositories for that 2014-06-04 14:24:22 just testing the 3.0 release, nothing negative to report so far :) 2014-06-04 15:36:26 Right I'm off, bye all! 2014-06-04 22:50:14 I can't find the answer on the website, so: does the "standard" live cd include X? 2014-06-05 00:08:59 Ok, having loopback mounted it, I see the X server is not present but the X libs are. 2014-06-05 08:20:40 Good morning everyone 2014-06-05 12:59:21 hi 2014-06-05 13:00:22 hi 2014-06-05 13:01:52 latest alpine 32bit virsh net-start default gives error "cannot check dnsmasq binary dnsmasq: No such file or directory" but dnsmasq is installed and /usr/sbin is in PATH, any solutions? 2014-06-05 13:01:59 poweroff 2014-06-05 13:02:31 and the guest powersoff? 2014-06-05 13:02:49 ncopa, no the network doesn#t start 2014-06-05 13:02:55 hm 2014-06-05 13:03:53 the dnsmasq path must be specified when bilding libvirt, howto rebuild the alpine pkg? 2014-06-05 13:04:00 iam new to alpine 2014-06-05 13:04:42 simplest is to: sudo apk add alpine-sdk && git clone git://git.alpinelinux.org/aports && cd aports/main/libvirt 2014-06-05 13:04:48 edit the APKBUILD 2014-06-05 13:04:51 and then: abuild -r 2014-06-05 13:05:04 i run libvirt here on my desktop 2014-06-05 13:05:07 its 64bit tough 2014-06-05 13:06:25 what happens if you run: dnsmasq --help 2014-06-05 13:06:50 do you get error message or help text 2014-06-05 13:08:19 how do we set dnsmasq path at buildtime? 2014-06-05 13:13:33 ncopa thans, will try that 2014-06-05 14:59:21 Hello all! I've been trying to test squidguard (not used it before) 2014-06-05 15:00:17 and I think it must do something like "mkdir -p /etc/squidGuard" at install....i think it needs to do "mkdir -p /etc/squidGuard/db" 2014-06-05 15:00:28 and the 'blockedsites' file needs to be in there 2014-06-05 15:01:27 that's the only way it works for me anyway 2014-06-05 15:01:48 otherwise the "squidGuard -C all" command never finishes 2014-06-05 15:03:16 ginjachris: please report it on our bugtracker 2014-06-05 15:03:40 it uses a db? 2014-06-05 15:03:51 databases normally goes into /var/lib/ 2014-06-05 15:04:48 i'll do some more testing and update bug tracker :) just wondering if anyone else had seen this. First time using squidguard, so prone to errors! 2014-06-05 15:09:53 looks like the db location is controlled in /etc/squidGuard/squidGuard.conf with the line "dbhome " which is default as "dbhome /etc/squidGuard/db" 2014-06-05 15:42:57 damnit I can't get the damn thing to block now...I think it's time to go home 2014-06-05 15:43:13 I'll rty to test more tomorrow and update the bug tracker 2014-06-05 18:34:39 hey, i wonder if anyone can reproduce this, but i cannot login as a regular user /o\ 2014-06-05 18:35:26 the thing seemed to be related to the fact, that both the hostname and the username are 1 char long. and passwd complains that the password is too similar to hostname/username. 2014-06-05 18:35:57 i even tried to set the password by copying the one from root to the user, but not even that worked. 2014-06-05 18:36:10 this is on edge 2014-06-05 18:39:28 and login works with other users 2014-06-05 18:52:35 for someone who uses 1-char names your irc nickname definitely looks too long 2014-06-05 18:52:59 by copying you mean to /etc/shadow? 2014-06-05 18:54:49 i mean copying the password around in shadow, yes. 2014-06-05 18:58:17 i don't have a live alpine system nearby, but your description smells of pam shenanigans 2014-06-05 18:58:33 in particularly something like pam_cracklib.so reject_username 2014-06-05 18:58:51 look around in /etc/pam.{conf,d/} 2014-06-05 18:59:00 there is no pam 2014-06-05 19:00:12 find /etc -name '*pam*' is equivalent to cat /dev/null 2014-06-05 19:08:18 sorry, too long since i used alpine it seems. then just read /sbin/login's source (busybox)? 2014-06-05 19:09:50 hmmm, i guess i'll do that! thanks for the pointer. 2014-06-05 19:17:10 busybox is very straightforward code. good chances too actually figure out what's going on. good luck doing that with gnu 2014-06-05 19:17:18 *to 2014-06-05 19:18:40 i guess pam is more related to sun than to the fsf 2014-06-05 19:28:48 well, the comparison was more to coreutils et al 2014-06-05 20:37:56 ACTION rawrs at TheWilfox http://i.imgur.com/jr5KQfj.gif 2014-06-05 22:44:42 bfv 2014-06-05 22:44:53 sorry, baby got the keyboard 2014-06-05 22:52:52 hello all! 2014-06-05 22:54:13 I have installed alpine linux in virtualbox with the xfce desktop environment. but i'm having troubles installing the guest additions. Anyone using alpine in vbox? 2014-06-05 23:25:22 does anyone know what package has ntlm_auth? 2014-06-06 06:23:06 morning 2014-06-06 06:23:13 vbox guest additions does not build :-/ 2014-06-06 06:23:17 rotten code 2014-06-06 06:23:47 like vmware code, i think the coders are organized in groups per arch 2014-06-06 06:23:59 or system 2014-06-06 06:24:04 so one group for windows 2014-06-06 06:24:14 other for (gnu) linux 2014-06-06 06:24:19 other for solaris, etc 2014-06-06 06:24:34 then the groups dont work together 2014-06-06 06:24:53 instead they do: #if __MYSYSTEM__ ... 2014-06-06 06:25:00 #else __OTHERSYSTEM__ 2014-06-06 06:25:03 ... 2014-06-06 06:25:15 #else if __YET_ANOTHER_SYSTEM___ 2014-06-06 06:25:18 ... 2014-06-06 06:25:20 #endif 2014-06-06 06:25:50 and none of the groups cares about any portability, as long as it builds/runs on their system 2014-06-06 06:26:35 it doesnt help that the __linux__ groups thinks that __linux__ == __GLIBC__ 2014-06-06 06:49:30 :( 2014-06-06 07:03:53 morning all! 2014-06-06 07:35:05 well, after messing around with Dansguardian and squidguard yeserday for ages (and a hell of alot of other troubleshooting), all due to my usual method of URL blocking as described on the wiki stopped working.....I finally got it working again. 2014-06-06 07:35:27 The cause? The one thing I thought hadn't changed, the url list, started with a "-" 2014-06-06 07:35:51 deleted that and BOOM! Squid loaded the configuration and it all works again! :D 2014-06-06 07:36:16 So now I personally have no need for squidguard/dansguardian :( 2014-06-06 07:37:22 that's just one of many exciting stories Uncle Ginja could tell 2014-06-06 07:47:11 Has anyone here installed mongodb through npm only for it to fail? 2014-06-06 07:54:09 I'm new to alpine and I consistently get compile errors for some reason. Is there some sort of setting or other that I need to disable? 2014-06-06 07:59:18 hmmm, I'm afraid I'm not going to be able to help with that. I'm more of a pre-packaged application user :( 2014-06-06 07:59:51 i have tried to build mongodb iirc 2014-06-06 08:00:01 it was not a pleasant experience... 2014-06-06 08:03:05 yeah, i tried to do that with nodejs and actually had to scrap my system just to fix it lol 2014-06-06 08:03:15 kept getting seg faults all day 2014-06-06 08:03:44 does dmesg say anything of interest? 2014-06-06 08:03:49 i think javascript uses a JIT 2014-06-06 08:03:59 I figured out what it is, though. Not sure why it's doing it but it seems npm cannot access the node dir 2014-06-06 08:04:00 JIT compiler 2014-06-06 08:04:16 apparently root doesn't even have access? 2014-06-06 08:04:34 gyp WARN EACCES user "root" does not have permission to access the dev dir 2014-06-06 08:06:50 even as root I dont have access. This is frikkin wonky 2014-06-06 08:06:59 ls 2014-06-06 08:07:01 oops 2014-06-06 08:18:19 I'm trying to get www.mean.io installed on my system and I keep getting that odd error. So far I haven't figured out exactly *what* it's doing but it doesn't seem to be causing me any issue anymore. 2014-06-06 10:04:49 ncopa: now, not being able to install mongodb is really a feature of alpine i can get behind! 2014-06-06 10:05:30 :D 2014-06-06 14:04:15 started to harmonize zfs with musl dunno how far i'll get. 2014-06-06 14:06:17 ovf i totally feel like mongo is the new mysql. however i also found out, that for non critcal bulk data processing it's quite useful. like you have only writes when you bulk load the data from another source, which means, acid and transaction and stuff is mostly useless in such cases 2014-06-06 14:08:52 like you download a huge and hierarchical dataset from some public source and want to do a few complex queryies on that 2014-06-06 14:24:45 hello 2014-06-06 14:46:41 algitbot really should answer to every hello 2014-06-06 15:02:54 hmm, ncopa working on the zfs/muse thing. and it seems the wrong sys/types.h is included, the system-wide one, and not the one supplied by libspl/include/sys/types.h 2014-06-06 15:03:08 s/muse/musl/ 2014-06-06 15:03:32 the __gnuc_asy to fix 2014-06-06 15:06:50 huh, bitgnomes ate some chars... /o\ 2014-06-06 15:07:02 __gnu_va_list was easy to fix 2014-06-06 15:15:38 hello, do you have a guide to report a bug? 2014-06-06 15:18:04 m0ikz, adding a bug is easy. Get an account here: http://bugs.alpinelinux.org/ 2014-06-06 15:20:15 once logged in, click on 'new issue' (http://bugs.alpinelinux.org/projects/alpine/issues/new) 2014-06-06 15:22:24 thanks ginjachris 2014-06-06 15:22:43 that's about it.....try to be as thorough and accurate as possible in bug/RFE description. 2014-06-06 15:22:54 no probs m0ikz 2014-06-06 15:23:17 I'll be leaving soon but if you get stuck I'm sure one of the other guys will be happy to help 2014-06-06 15:30:29 bye all, have a great weekend! 2014-06-06 17:12:06 what do you guys use for CA management? 2014-06-06 17:13:17 rm -rf :) 2014-06-06 18:57:26 ncopa: i managed to compile zfs-grsec 2014-06-06 18:57:34 with 3.14.5 2014-06-06 18:57:45 and zfs/spl from git. plus a ton of patches 2014-06-06 19:27:28 ncopa: http://pastebin.com/yyigWq5x my initial patch 2014-06-06 19:41:30 hi, does this distro support a Web browser? Not sure where the package info is kept. obv not in http://git.alpinelinux.org/ 2014-06-06 19:42:02 hendry: yes, we have firefox at least 2014-06-06 19:42:07 something smaller as well I guess 2014-06-06 19:42:31 packages can be found in aports tree or here: http://alpinelinux.org/packages 2014-06-06 19:45:33 be cool if you supported http://surf.suckless.org/ 2014-06-06 19:49:43 possible to do :) 2014-06-06 19:49:53 I can try tommorow, if I don't forget 2014-06-07 00:10:06 hi, all. (here new to alpinelinux) 2014-06-07 00:12:19 installing samba-dc, but samba-tool results in error:apline:~# samba-tool Traceback (most recent call last): File "/usr/bin/samba-tool", line 33, in from samba.netcmd.main import cmd_sambatool ImportError: No module named samba.netcmd.main 2014-06-07 00:13:04 I have tried version 2.7 and 3.0 x86_64, both same error. 2014-06-07 00:16:16 It seems is the python library found in samba source tree, where it is packaged in apk? 2014-06-07 00:22:57 ncopa: you might want to test the zfs-grsec with https://github.com/stef/aports/tree/master/testing/zfs-grsec 2014-06-07 08:12:58 hi. what package provides the chsh-command? 2014-06-07 08:13:37 i thought it was util-linux, but it isn't there 2014-06-07 08:18:23 https://github.com/karelzak/util-linux/blob/master/login-utils/chsh.c <- oO 2014-06-07 08:18:44 i guess it's compiled without chsh, if that is even possible 2014-06-07 08:24:16 --disable-login is in the APKBUILD 2014-06-07 13:05:42 pluesch_: could you please create a task on our bug tracker? 2014-06-07 13:05:55 I probably could re-enable it, but there may be something behind it 2014-06-07 13:06:08 hendry: I'll be pushing surf to testing in a few minutes 2014-06-07 13:06:18 hendry: let me know if it works so I could move it to main 2014-06-07 13:13:08 hendry: or I won't, our webkitgtk uses gtk3 2014-06-07 19:29:52 hey! no more vserver? 2014-06-07 20:01:06 barthalion: will do that, thanks :) 2014-06-07 20:12:13 Anaphaxeton: yep, replaced with lxc 2014-06-07 22:08:47 is it too late to get something backported to alpine 2.6? 2014-06-07 22:22:09 barthalion: how does webkitgtk using gtk3 affects things? 2014-06-08 09:30:04 hi, when i do startx with awesome/openbox i get http://ix.io/cRD and http://ix.io/cRE respectively 2014-06-08 12:28:17 goodmorning 2014-06-08 12:28:31 why has VServer been abandoned? 2014-06-08 13:17:36 anyone else has problems formatting btrfs partition? 2014-06-08 13:17:46 partitions* 2014-06-08 13:19:05 error checking /dev/sdaX mount status 2014-06-08 13:19:20 same for /dev/sda 2014-06-08 13:20:43 also mount expects -t to find an ext4 partition 2014-06-08 13:43:03 does it fail to read /proc/mount ? 2014-06-08 13:43:22 also i noticed there is no /dev/root 2014-06-08 13:44:44 ncopa ^ 2014-06-08 14:54:26 Anaphaxeton: we favour lxc instead of vserver 2014-06-08 14:59:17 ok 2014-06-08 14:59:29 is the kernel patched or anything else needed? 2014-06-08 15:00:22 btrfs , after an strace, seems to not be able to see only /dev but not tmpfs sysfs etc etc 2014-06-08 15:00:40 is it a grsecurity measure clandmeter ? 2014-06-08 15:15:27 Anaphaxeton: no its a bug afaik 2014-06-08 15:15:58 I think i faked something to format mine 2014-06-08 15:16:38 Anaphaxeton: lxc support is mainline 2014-06-08 15:17:03 because of the maintenance we try to use everything mainline. 2014-06-08 15:17:18 imho lxc is also nicer than vserver 2014-06-08 15:19:10 regarding btfr, i googled around and found somebody suggested to umount something, i cant really remember what i did. 2014-06-08 15:45:34 i think i am close 2014-06-08 15:45:48 only if that VM could start again :p 2014-06-08 15:46:01 someo msr error and a freeze 2014-06-08 15:47:21 yep, host problem brb 2014-06-08 15:50:13 why the hell, it is the same vm 2014-06-08 15:50:20 and the same iso 2014-06-08 16:09:09 vbox no worky anymore 2014-06-08 16:09:26 i give up for now 2014-06-08 16:36:56 does Adobe Flash work with alpine? 2014-06-08 16:37:52 or how do i install gstreamer? 2014-06-08 18:24:25 ok i switched from vbox to qemu 2014-06-08 18:24:40 strace show the same bizzare image 2014-06-08 18:24:54 cant stat stuff like tmpfs and sysfs 2014-06-08 19:17:10 clandmeter: it seems the issue is with loop devices 2014-06-08 19:22:01 after opening the /dev/loop it makes a series of assumptions that are not true 2014-06-08 19:23:25 like the existance of /sys/block/0/loop 2014-06-08 19:33:37 i need to get rid of this entry from /proc/self/mounts 2014-06-08 19:34:23 or present mkfs.btrfs with another file 2014-06-08 19:34:45 or use another distro to format, but that is not the good way 2014-06-08 19:49:50 from the btrfs side, what i see is a malformed line 2014-06-08 19:50:01 lets ask them 2014-06-08 20:01:13 hoi 2014-06-08 20:02:25 heyaho 2014-06-08 20:07:07 is there any paste utility? 2014-06-08 20:09:23 the btrfs people will appreciate an strace 2014-06-08 20:11:47 -o redirect the output to a file 2014-06-08 20:16:28 ok i transfered it via scp to the host 2014-06-09 07:35:40 my oh my 2014-06-09 07:35:59 ncopa i get the error with vanilla as well 2014-06-09 07:36:05 ACTION facepalms 2014-06-09 07:40:53 Anaphaxeton: sounds like bug in btrfs 2014-06-09 07:41:12 my favorite paste utility is sprunge 2014-06-09 07:41:14 apk add sprunge 2014-06-09 07:41:19 some-command | sprunge 2014-06-09 07:44:15 isn't the string paste anywhere in the description 2014-06-09 07:44:19 search didnt find it 2014-06-09 07:44:53 sprunge-0.2-r0 description: 2014-06-09 07:44:53 Helper script to paste things to http://sprunge.us 2014-06-09 07:45:28 hm 2014-06-09 07:45:32 interesting 2014-06-09 07:45:34 ncopa-desktop:~$ apk search -d 'paste' 2014-06-09 07:45:34 spice-gtk-tools-0.24-r0 2014-06-09 07:45:34 ipptool-1.7.3-r0 2014-06-09 07:45:34 st-0.4.1-r0 2014-06-09 07:45:34 spice-glib-0.24-r0 2014-06-09 07:45:42 does not show sprunge... 2014-06-09 07:47:17 fabled: any idea why apk search ' 2014-06-09 07:47:33 fabled: any idea why 'apk search --description paste' does not show sprunge? 2014-06-09 07:49:02 locale? 2014-06-09 07:49:11 shouldn matter 2014-06-09 07:49:41 k 2014-06-09 07:50:17 fails to parse the URL? 2014-06-09 07:50:51 -d should search the description 2014-06-09 07:53:36 k 2014-06-09 07:53:43 ACTION run out of ideas :p 2014-06-09 08:27:21 ncopa: i tried arch, which ships with btrfs-progs 3.14.1 and it worked 2014-06-09 08:27:41 now i can continue :D 2014-06-09 08:28:39 except if i cannot use the btrfs program 2014-06-09 09:37:12 Morning all! 2014-06-09 09:37:25 morning 2014-06-09 09:58:54 goodmorning 2014-06-09 10:01:53 Morning! 2014-06-09 10:03:20 moin 2014-06-09 11:09:20 no efi? 2014-06-09 11:11:12 you know, I've never had an EFI system to test on 2014-06-09 11:12:15 try vbox 2014-06-09 11:12:26 my desktop runs on efi 2014-06-09 11:12:51 it's easier than it seems 2014-06-09 11:13:23 does setup-alpine script work, or you need to make manual changes for EFI system? 2014-06-09 11:16:57 I'll try a quick test in virtual box 2014-06-09 11:19:57 wow that's very weird (efi) 2014-06-09 11:37:54 updated vbox 2014-06-09 11:39:56 arrgh, i still get dropped straight into a shell. No idea with EFI. I'll take a look some other time. Really need to continue revising for my exam tomorrow! 2014-06-09 11:40:26 setup-alpine does not handle efi 2014-06-09 11:41:32 oh I don't even get that far, just drops me straight to a shell without booting from cd-rom. It's ok, this is not pressing for me right now 2014-06-09 11:42:38 although I guess what would be really cool is setup-alpine script detecting EFI system and doing whatever it has to do differently from BIOS systems. I guess syslinux needs to be installed a little differently and it has to use GPT partitioning and so forth 2014-06-09 11:42:55 yup 2014-06-09 11:43:15 i dunno if you even can boot cdrom with EFI? 2014-06-09 11:43:21 i use efi on my laptop though 2014-06-09 11:43:37 using gummiboot 2014-06-09 11:44:09 there are a coupld of things i dont like with EFI 2014-06-09 11:44:17 i dunno. EFI is very different to me. Read about it, but this is the first time I used it (and in Vbox, which could i suppose introduce some more challenges) 2014-06-09 11:44:27 one thing is that it needs to be on a FAT partition 2014-06-09 11:44:27 yes 2014-06-09 11:44:43 it was new to me too when i got my laptop 2014-06-09 11:44:48 and i wanted do dualboot 2014-06-09 11:44:50 with windows 2014-06-09 11:45:05 took me a couple of weeks before i got it working 2014-06-09 11:45:33 syslinux cannot do chainloading of efi images 2014-06-09 11:45:34 noooo not FAT! 2014-06-09 11:45:43 the efi partition is FAT 2014-06-09 11:46:00 that almost seems a step backwards.... 2014-06-09 11:46:02 you could actually let kernel be your efi image 2014-06-09 11:46:24 then you dont need syslinux or any boot loader 2014-06-09 11:46:32 wow. 2014-06-09 11:47:33 i mount my EFI partition as /boot 2014-06-09 11:47:52 which means the kernel image and initramfs image is on fat 2014-06-09 11:48:24 I see....it makes sense 2014-06-09 11:48:29 just weird to me 2014-06-09 11:49:35 tbh i really like the alpine-setup script doing all the hard work for me (and raid etc if i have multiple drives). I came from Arch but get bored of having to do such a manual setup each time....hence why i like the setup script so much 2014-06-09 11:49:54 nice :) 2014-06-09 11:50:13 it certainly is! :D 2014-06-09 11:50:39 you know you can do: setup-disk -L /dev/sda /dev/sdb /dev/sdc 2014-06-09 11:50:43 and it will raid5 2014-06-09 11:50:48 and lvm on top of that ;) 2014-06-09 11:51:17 you can even do raid1 with a single disk: setup-disk -r 2014-06-09 11:51:21 Awesome! 2014-06-09 11:51:54 and if you run a diskless system already (using lbu for saving configs) 2014-06-09 11:52:08 and you later want install that system on disk, then you can simply do: setup-disk 2014-06-09 11:52:30 I have to say I haven't used LVM much either. I know more logical, for me personally it adds another layer of complexity. Complexity is my enemy lol! I guess I should start getting used to it though 2014-06-09 11:52:47 i use lvm on everything 2014-06-09 11:53:00 its an "insurance" thingy to me 2014-06-09 11:53:12 makes it easy to replace disks later 2014-06-09 11:53:45 plug in disk, add it to the volume group, tell lvm to move data to there (while still online) 2014-06-09 11:53:49 remove old disk 2014-06-09 11:53:53 put in new disk 2014-06-09 11:53:55 move data back 2014-06-09 11:54:23 and you have basically replaced the original disk with minimal downtime 2014-06-09 11:54:44 that's cool. I wrote the raid admin wiki page (http://wiki.alpinelinux.org/wiki/Raid_Administration) without lvm, I guess I should expand it to include lvm 2014-06-09 11:55:01 not enough hours in the day! 2014-06-09 11:55:27 full time job = not enough time for the important things in life (like Alpine) :) 2014-06-09 11:55:33 indeed :) 2014-06-09 11:56:11 i suppose we should try fix efi support for v3.1 2014-06-09 11:56:25 cool! :D 2014-06-09 12:31:47 greetings. I was here some months ago, mumbling about icedtea. I waited for the 3.0 release (musl, nice work), and just went on to test again, in VirtualBox. I have problems running icedtea-web with openjdk7; is there any information on this use? any experience with Mozilla openjdk usage? 2014-06-09 12:34:36 we have openjdk working, but icedtea-web i dont know 2014-06-09 12:35:37 by that, you mean tested? 2014-06-09 12:35:58 yes, i have not tested icedtea-web 2014-06-09 12:36:04 but i have tested openjdk7 2014-06-09 12:36:36 okay. well, then it's just the web scripts that may not be working. 2014-06-09 12:36:55 i'll test now 2014-06-09 12:37:24 do you have a good testsite with java applet? 2014-06-09 12:37:51 I searched up "icedtea web test" 2014-06-09 12:38:49 I installed icedtea-web-mozilla (difference from icedtea-web?), tried to run it in Firefox, but Firefox just stopped responding. I guess it timed out, because no Java applet was ever run, but I regained control of Firefox (XFCE also froze) after some time. 2014-06-09 12:39:51 hm 2014-06-09 12:39:54 same experience here 2014-06-09 12:40:15 icedtea-web-mozilla is the firefox plugin i think 2014-06-09 12:41:07 should be. it has more dependencies than -web, idk why 2014-06-09 12:41:30 afaik, it should just require an extra symlink to the mozilla plugin dir, forgot where that is 2014-06-09 12:41:58 but maybe mozilla requires some special interface 2014-06-09 12:43:40 heh 2014-06-09 12:43:43 it works actually 2014-06-09 12:43:47 http://www.oracle.com/technetwork/java/example1-142131.html 2014-06-09 12:44:06 i think the problem is a dependency only 2014-06-09 12:44:12 apk add icedtea-web 2014-06-09 12:44:16 and it started to work 2014-06-09 12:44:52 *testing* 2014-06-09 12:46:13 the fact that EFI reads FAT partitions is good. FAT is easy to implement in operating systems and that partition is meant to be kept ro so no journals, atomic filesystems and the such are needed 2014-06-09 12:46:56 but it seems that apart from the CD that doesn boot from EFI the kernel lacks EFI support as well 2014-06-09 12:46:59 Anaphaxeton: i miss support for symlinks and support for resizing it 2014-06-09 12:47:01 efivars is not there 2014-06-09 12:47:08 yes 2014-06-09 12:47:14 that was a major issue for me 2014-06-09 12:47:39 so what i did on my laptop was install gummiboot etc 2014-06-09 12:47:51 and then from windows i changed the boot manager 2014-06-09 12:47:55 ncopa: it is meant to be 512M so why resize? 2014-06-09 12:48:22 what if you want increase it in future? 2014-06-09 12:48:57 resize of fat was more a general feature missing 2014-06-09 12:49:07 symlinks is more relevant /boot though 2014-06-09 12:49:07 and put in it what? a blob linux distro? 2014-06-09 12:49:35 ncopa: installing icedtea-web did the trick. how can upstream be notified that icedtea-web{-mozilla,} dependency? 2014-06-09 12:49:48 Aspi: bugs.alpinelinux.org 2014-06-09 12:50:02 file a bug with v3.0 as target 2014-06-09 12:50:10 in this computer /boot is the esp 2014-06-09 12:50:49 when i update i remount it rw and afterwards back to ro 2014-06-09 12:51:33 maybe i should do that too 2014-06-09 12:51:35 and the ro thing is just a paranoia setting 2014-06-09 12:51:44 i have my /boot as rw fat 2014-06-09 12:51:53 sounds fine 2014-06-09 12:51:56 and it makes me feel uncomfortable 2014-06-09 12:52:11 mass paranoia then! 2014-06-09 12:52:18 i wonder if it would be an idea to implement pre/post upgrade hooks in apk-tools 2014-06-09 12:52:22 hehe do as i do then 2014-06-09 12:52:35 eg /etc/apk/pre-upgrade.d/ 2014-06-09 12:52:47 i dont know apk well yet :> 2014-06-09 12:52:54 and /etc/apk/post-uprade.d/ 2014-06-09 12:53:19 apk could run pre-upgrade.d/* before upgrading things 2014-06-09 12:53:28 and post-upgrade when done 2014-06-09 12:53:51 that way you could add a small script that mount -o remount,rw 2014-06-09 12:54:02 and in post-upgrade mount -o remount,ro 2014-06-09 12:57:55 makes sense 2014-06-09 12:58:10 it would be friendly to the various daemons too 2014-06-09 12:58:50 i dont like these cold turkey updates of archlinux 2014-06-09 12:58:59 ncopa: ah, wtf. I just realize who I mailed. I was lazy. sorry bout that xD 2014-06-09 13:14:43 anyone used to using icedtea-web (openjdk for web browsers)? 2014-06-09 13:18:12 I'm can make a particular site work using iceweasel and icedtea-web with openjdk7.55 on arch linux, while I can't make this site work with Firefox, icedtea-web{,-mozilla} at alpine. it runs, but has some problem and ends displaying an error itself. so, I'm guessing security settings et cetera 2014-06-09 13:20:53 i havent really used icedtea-web 2014-06-09 13:22:34 on anther note: I had no luck so far searching the docs for how lbu works. is it an overlay, or merely just replacing the changed files? 2014-06-09 13:22:52 merely replacing the changed files 2014-06-09 13:23:11 okay, nice. no aufs, so no greasy kernel :) 2014-06-09 13:23:17 creates a tarball of changed files in /etc 2014-06-09 13:29:39 okay, thanks for help. happy devving, very interesting project. 2014-06-09 13:33:56 hey, just saying that i packaged a few things, like mupdf-nox11, jfbview, i3wm, maybe others checkout github/stef/aports, also i have a patch to zfs-grsec that makes it work again. 2014-06-09 13:49:16 I have a query re: ipv6 configuration on busybox. It all works fine, but I couldn't get it to function with multiple static IPv6 addresses. Has anyone got such a setup working? (btw, won't have time to test this again right now, and i assume it's an upstream busybox problem) I seem to remember the IPv6 RFC stating an interface must be able to support multiple ipv6 addresses, although I don't recall if that explicitly refer 2014-06-09 13:49:54 It is possible without busybox (works in arch linux for example using netctl) 2014-06-09 13:54:29 WwQCn4HqahlD: nice! please join #alpine-devel 2014-06-09 14:26:28 byez, see you on #~-dev 2014-06-09 14:26:59 nice job 2014-06-09 14:34:50 ncopa: I might have found my issue. icedtea-web is working all right; the last problem is that it could not verify a signature 2014-06-09 14:35:22 ncopa: are you familiar with that? what/are signatures are bundled with alpine? 2014-06-09 14:35:55 icedtea at arch, installed ca-certificates-java (https://www.archlinux.org/packages/extra/any/ca-certificates-java/) 2014-06-09 14:36:03 depends on* 2014-06-09 14:36:07 aha 2014-06-09 14:36:59 we have what corresponds to https://www.archlinux.org/packages/core/any/ca-certificates/ 2014-06-09 14:37:15 i didnt know there is a spearate package for java 2014-06-09 14:37:25 I'll uninstall ca-*-java, try and see if that makes a difference 2014-06-09 14:37:52 i bet it does 2014-06-09 14:41:27 Aspi: if i dont reach fix it today, would you mind file a bug on bugs.alpinelinux.org? 2014-06-09 14:41:29 so i dont forget it 2014-06-09 14:41:34 thanks! 2014-06-09 14:41:50 o:-) I'll see what I can do 2014-06-09 14:42:07 it worked! now, it didn't work 2014-06-09 14:42:13 lol 2014-06-09 14:42:27 sounds like selfcontradict :) 2014-06-09 14:43:01 fantastic. prolly explains why very few of the lurking distros I've tried to get this working on, have been successful 2014-06-09 14:43:21 do you have an url? 2014-06-09 14:43:36 ah 2014-06-09 14:43:43 sorry, i misunderstood :) 2014-06-09 14:47:27 off I go then. cya round 2014-06-09 15:10:45 ncopa: java has (can have?) something called a keystore 2014-06-09 15:10:53 idk much about it 2014-06-09 15:11:13 the 'keytool' command in javahome/bin manages them 2014-06-09 15:21:33 right I'm off, bye all! 2014-06-09 16:33:32 goodevening :) 2014-06-09 16:40:00 ACTION , after a nap continues installing Alpine with a custom kernel! :p 2014-06-09 16:45:30 how come arent you using stack smashing? 2014-06-09 16:46:41 i thought it contributes to security. because of grecurity? 2014-06-10 03:25:37 for 3.0, do you need to change your repositories to edge or is v3.0 ok? 2014-06-10 06:52:32 morning 2014-06-10 06:53:06 jimmy99: yes, v3.0 is ok. but you need use apk.static upgrade --no-self-upgrade 2014-06-10 07:01:36 moin 2014-06-10 07:10:00 morning 2014-06-10 17:16:52 alert! 2014-06-10 17:17:01 ? 2014-06-10 17:17:13 ouxe!!! 2014-06-10 17:17:52 i just stumbled upon a compile error of linux-grsec 2014-06-10 17:18:38 in case you haven't noticed two grsec patches fail to apply 2014-06-10 17:19:26 or at least that is so with the grsecurity from git 2014-06-10 17:19:43 didnt think about trying the a release 2014-06-10 17:19:52 with* 2014-06-10 17:23:36 Anaphaxeton: witch patch? 2014-06-10 17:23:54 i mean which :D 2014-06-10 17:28:08 bpf_jit 2014-06-10 17:28:32 plus another that does not affect me at compile time 2014-06-10 17:31:42 after googling it seems it would be bypassed by disactivating some or all of netfilter 2014-06-10 17:32:07 but i will try the latest patch from grsecurity 2014-06-10 17:41:19 Hi. What am I supposed to install in order to able to compile a "int main(void) { return 1; }" C program under Alpine? I'm getting linker errors, like: /usr/lib/gcc/i486-alpine-linux-musl/4.8.2/../../../../i486-alpine-linux-musl/bin/ld: cannot find crti.o: No such file or directory 2014-06-10 17:47:05 have you tried the alpine-sdk meta package? 2014-06-10 17:47:24 No, I've never heard of it. 2014-06-10 17:48:16 Works now. Thank you! 2014-06-10 17:49:31 you're welcome 2014-06-10 17:50:05 Anaphaxeton: the kernel config in the aports git repo does not compile with the patches in that package? 2014-06-10 19:30:13 Anaphaxeton: on which architecture? 2014-06-10 20:45:47 Anybody seen crazy buffering and caching in 3.0? These shots are seconds apart http://i.imgur.com/jPeABJ8.png http://i.imgur.com/iAUXOzC.png 2014-06-10 20:50:54 n8@all 2014-06-10 22:48:08 kocka: x86 all 2014-06-10 22:48:29 i just patched 3.14.6 with the respective grsec 2014-06-10 22:48:33 they fixed it 2014-06-10 23:23:25 doh... gcc laments everywhere 2014-06-10 23:24:02 i wanted kexec so i removed the /dev/mem restriction of grsec. hope it wasnt a bad idea 2014-06-11 01:10:50 <^7heo> Moin people 2014-06-11 01:11:11 <^7heo> I wondered 2014-06-11 01:11:55 <^7heo> I'm interested in trying a musl-linux distro 2014-06-11 01:12:00 <^7heo> so I wanted to check Alpine 2014-06-11 01:12:32 <^7heo> however, is it possible to install Alpine from source, a-la-gentoo? 2014-06-11 01:25:29 from source i think not 2014-06-11 01:25:53 but regenerate it from source seems quit feasible 2014-06-11 01:26:13 like arch 2014-06-11 01:28:19 or maybe build an LFS following alpine recipes? 2014-06-11 01:28:32 i have partly made one 2014-06-11 01:30:05 <^7heo> okay 2014-06-11 01:30:37 <^7heo> thanks for the info :) 2014-06-11 01:30:40 <^7heo> I have to go 2014-06-11 01:30:41 <^7heo> laters :) 2014-06-11 01:33:22 bb 2014-06-11 03:44:33 is there a recommended way to setup on a system with no hard-wired network? iwconfig doesn't seem to be on the install cd 2014-06-11 03:54:20 hmm i see the apk's are available 2014-06-11 03:55:01 ncopa, is there any install guide for laptop installs with no wired network? 2014-06-11 04:48:34 zzz i'll be back tomorrow :) 2014-06-11 06:28:01 dalias: we dont have any install guides for wifi only laptops afaik. But i think something like: apk add wpa_supplicant && wpa_passphrase $SSID >> /etc/wpa_supplicant/wpa_supplicant.conf && /etc/init.d/wpa_supplicant start && setup-interfaces 2014-06-11 06:28:04 might work 2014-06-11 06:28:28 what i do in my laptop is using dhcpcd + wpa_supplicant 2014-06-11 08:27:54 Hello... Is there some sort of 'recovery' console in Alpine? I have a remote embedded machine, running of CF card and I have '/bin/bash' as shell, but it seems that 'bash' is not installed at bootup... 2014-06-11 08:38:55 danci1973: default is /bin/sh (busybox shell) 2014-06-11 08:42:17 clandmeter: I know. 2014-06-11 08:43:48 I can connect to the machine through IP based KVM, so I can try some boot parameters... 2014-06-11 08:46:46 where does it stop? 2014-06-11 08:47:04 you can get into initramfs with boot param 's' iirc 2014-06-11 08:47:39 boot params for increase verbosity: noquiet verbose 2014-06-11 08:53:11 It doesn't stop - I just can't log in after it finished booting. 2014-06-11 08:54:39 OK; got it into 'single' mode using 's' ... 2014-06-11 11:49:33 Weird - /media/sda1/apk-cache is missing ... 2014-06-11 14:17:21 ncopa, is it safe to just use apk add --force to install packages on the live system from the install cd? 2014-06-11 14:17:45 (to get wpa_supplicant, dhcp, etc. setup for running the install to hd) 2014-06-11 14:18:11 or should i just try to bypass anything that needs the network during install and wait til i have a real system on the hd? 2014-06-11 14:18:34 why you need --force? 2014-06-11 14:19:43 i get 2014-06-11 14:19:55 WARNING: Ignoring APKINDEX....tar.gz: No such file or directory 2014-06-11 14:20:28 ERROR: You tried to add a non-repository package to system, but it would be lost on next reboot. Enable package caching (apk cache --help) or use --force if you know what you are doing. 2014-06-11 14:20:39 maybe i should try apk cache --help :) 2014-06-11 14:20:47 ah, that message is stupid 2014-06-11 14:20:59 :) 2014-06-11 14:20:59 i think its a bug even 2014-06-11 14:21:08 is it a -rc iso? 2014-06-11 14:21:25 i think that should be fixed in the 3.0.0 iso 2014-06-11 14:21:28 alpine-3.0.0-x86_64.iso 2014-06-11 14:21:32 oh? 2014-06-11 14:21:33 hm 2014-06-11 14:21:42 i thought i fixed all those 2014-06-11 14:21:43 ok 2014-06-11 14:22:07 i think what you can do is before running setup-alpine, apk add wpa_supplicant 2014-06-11 14:22:11 i first ran setup-alpine, but it failed seting up the location for packages to download because there was no network device 2014-06-11 14:22:19 i think that broke things 2014-06-11 14:23:09 try get wpa_supplicant running from livecd before you start setup-alpine 2014-06-11 14:23:14 ok 2014-06-11 14:23:31 rc-update add wpa_supplicant (to make it autostart at boot) 2014-06-11 14:23:41 test with udhcpcd that you get ip addr 2014-06-11 14:23:44 btw wpa_supplicant does work, so no blocker there :) 2014-06-11 14:23:50 then kill udhcpcd 2014-06-11 14:23:54 and start setup-alpine 2014-06-11 14:24:15 it will infact clone the running system to the disk 2014-06-11 14:24:43 btw once i get a hd install, are there packages i should setup for better dynamic wifi config? or do i need to do that myself? :) 2014-06-11 14:24:59 this is my first time messing with wpa_supplicant directly 2014-06-11 14:25:27 but if i can automate it well, it looks A LOT better than working with all the low-quality existing tools like network-manager 2014-06-11 14:25:38 what i do, is mess with wpa_supplicant directly and let dhcpcd manage the link 2014-06-11 14:25:57 and i have a small dhcpcd-ui applet in xfce that shows the link status 2014-06-11 14:26:23 there is an ui for dhcpcd that should let us pick wifi net 2014-06-11 14:26:28 but i have never got that working 2014-06-11 14:26:52 just pick from existing configured ones, or is there an easy way to add/join new ones in the ui? 2014-06-11 14:27:10 so i simply just run wpa_passphrase and manage wpa_supplicant manually 2014-06-11 14:27:36 it can give you a list of existing ones and let you select 2014-06-11 14:27:41 but i dont know if that works 2014-06-11 14:27:47 i think it might 2014-06-11 14:27:54 presumably it's possible to install the usual desktop crapware, but of course i'd rather not... 2014-06-11 14:28:06 we have a pckage for networkmanager 2014-06-11 14:28:11 but i never used it actually 2014-06-11 14:28:14 ah 2014-06-11 14:28:35 the dhcpcd ui is fairly close to be usable i think 2014-06-11 14:28:36 might be fun to try and see how bloated/not-bloated it is with musl 2014-06-11 14:32:59 the biggest issue I have had with wpa_supplicant is when there are no encryption or wep 2014-06-11 14:33:13 otherwise it has been farily ok to manage manually 2014-06-11 14:34:20 oh? 2014-06-11 14:34:32 what issue? 2014-06-11 14:34:38 most of the networks i use are open -- basically all except my home network 2014-06-11 14:35:37 basically, when i have been in brazil and there have been an open wifi, i have simply not known the config options for wpa_supplicant 2014-06-11 14:36:01 i think it should be possible 2014-06-11 14:36:23 and i think i might have got it working, i just dont remember how i did and i never have the documentation handy when i need it 2014-06-11 14:36:27 :) 2014-06-11 14:36:40 ... and it happens seldom enough for me to forget... :) 2014-06-11 14:37:28 i could never understand why wpa_supplicant requires a config file at all 2014-06-11 14:37:49 why not just do everything through wpa_cli (or a saner equivalent thereof) 2014-06-11 14:38:15 well, its practially a db of known networks 2014-06-11 14:38:30 wpa_cli is hideous too 2014-06-11 14:38:37 i actually think you can use wpa_cli and config wpa_supplicant to be able to overwrite the config file 2014-06-11 14:38:45 i have no idea why it has its own prompt 2014-06-11 14:38:56 rather than just invoking each wpa_cli command from the shell 2014-06-11 14:39:10 because it waits for events 2014-06-11 14:39:17 and displays those 2014-06-11 14:39:29 yes, it prints those over top of the line you're trying to type.... 2014-06-11 14:39:35 anyone know what a section mismatch is? 2014-06-11 14:39:36 due to not using the terminal cleanly 2014-06-11 14:39:55 this kernel has quite a lot 2014-06-11 14:40:02 Anaphaxeton: in what context? 2014-06-11 14:40:10 kernel compilation 2014-06-11 14:40:31 http://stackoverflow.com/questions/8563978/what-is-kernel-section-mismatch 2014-06-11 14:40:36 ty 2014-06-11 14:41:16 dalias: i think wifi is an area where we could do better in alpine linux 2014-06-11 14:41:25 i simply dont know how we want it to work yet 2014-06-11 14:41:59 ncopa: archlinux's netctl (nee netcfg) sort of works for me 2014-06-11 14:42:15 basically it's a wpa_supplicant.conf management script 2014-06-11 14:42:44 #1008 2014-06-11 14:42:51 issue #1008 2014-06-11 14:42:52 ncopa, *nod* 2014-06-11 14:42:57 http://bugs.alpinelinux.org/issues/1008 2014-06-11 14:42:59 this isn't a complaint about alpine 2014-06-11 14:43:17 actually sane interactive wifi config is one of my big "new platform" open problems 2014-06-11 14:43:31 imo so far only android does it remotely sanely 2014-06-11 14:43:48 and even then you need a paid third-party app to deal with networks that require you to login 2014-06-11 14:44:02 :-/ 2014-06-11 14:44:20 i don't think android does anything special? 2014-06-11 14:44:25 i am absolutely interested in ideas 2014-06-11 14:44:38 otoh the developer of that app is great and responds immediately when you find a network the scripts don't work with 2014-06-11 14:44:50 if we had funding for the new platform project i'd hire him ;-) 2014-06-11 14:44:53 it has a list of known networks it's trying to connect to all the time, and then lists others so you can add them to the known list 2014-06-11 14:45:21 ovf, what it does special is just that it works :) 2014-06-11 14:45:31 lol 2014-06-11 14:46:10 what system has failed you? 2014-06-11 14:46:18 every time i've used network-manager it either fails to realize i moved away from the access point and won't re-search for an ap until i cycle the wifi off/on, or (opposite end of the spectrum) it constantly drops and reconnects even when i'm at 90% signal strength 2014-06-11 14:46:42 funny, i never used network-manager 2014-06-11 14:46:45 what distros? 2014-06-11 14:46:49 ncopa, debian :) 2014-06-11 14:47:13 i think networkmanager is just a frontend to wpa_supplicant 2014-06-11 14:47:26 no, it's much uglier 2014-06-11 14:47:33 if it were that, it'd be great 2014-06-11 14:47:34 from what i heard it's a rogue AI 2014-06-11 14:47:44 ugh.. 2014-06-11 14:47:45 ok 2014-06-11 14:48:14 really you kinda want something like nm tho, where wpa_supplicant isn't the "master" 2014-06-11 14:48:20 because nm supports multiple types of connections 2014-06-11 14:48:43 i am getting thousands of them 2014-06-11 14:48:47 wifi, wired, dialup, vpn, bluetooth, usb tethering to a phone, ... 2014-06-11 14:49:08 i don't think you really want to support connection switching at distro level 2014-06-11 14:49:10 aha you want systemds networkd :-P 2014-06-11 14:49:27 if they're all under the control of different tools/subsystems, it's a pain to switch from one to another 2014-06-11 14:50:07 ideally if they're all managed by one _good_ manager (NOT anything related to systemd :) then you could have features like auto-joining vpn on insecure networks, etc. 2014-06-11 14:50:38 i don't think that's the way to go 2014-06-11 14:50:52 might have been if you were designing a system for handheld devices 2014-06-11 14:50:57 ovf, i don't think a general-purpose distro (especially one made to be suitable for server use) should impose a particular connection switching system 2014-06-11 14:51:07 indeed 2014-06-11 14:51:21 but such a system should exist as an option independent of distro 2014-06-11 14:51:31 you want it for any device that's not stuck-in-place 2014-06-11 14:51:53 that means not just phones and tablets, but laptops/netbooks, cars, robots, ... :) 2014-06-11 14:52:49 ncopa: i took a look at wpa_supplicant some time ago. findings: 1) it was surprisingly configurable. 2) it probably has all the intelligence you want -- so you most likely need just some ui glue 2014-06-11 14:53:04 ovf: thats my impression too 2014-06-11 14:53:23 yeah 2014-06-11 14:53:27 wpa_supplicant looks impressive 2014-06-11 14:53:33 this is my first time playing with it directly 2014-06-11 14:53:37 dhcpcd fills in the missign bits with link detection, dhcp configuration and bridge to ui 2014-06-11 14:53:41 in the past i've just been lazy and let nm do all the work 2014-06-11 14:55:13 ncopa: i think the major problem here is psychological: you have to give in to wpa_supplicant and agree to store your configuration with it 2014-06-11 14:55:59 because if you begin storing your config elsewhere and generating wpa_supplicant.conf you're heading down the uncanny valley 2014-06-11 14:56:39 i think there is a socket connection so you can tell wpa_supplicant to store its config itself 2014-06-11 14:56:49 that was unclear 2014-06-11 14:57:05 there is a config option that allows wpa_supplicate overwrite the config file 2014-06-11 14:57:25 then you can connect via client (socket? dbus?) 2014-06-11 14:57:27 reconfig 2014-06-11 14:57:36 and make wpa_supplicant save the config 2014-06-11 14:57:40 yes, though i never tested this 2014-06-11 14:58:02 so if you write an UI for it, you could use that 2014-06-11 14:58:03 so indeed, i think the solution to #1008 should indeed look like a wpa_supplicant client 2014-06-11 14:58:29 i think dhcpcd-ui uses that 2014-06-11 14:58:50 i found dhcpcd-ui halfdone and halfworking but it might be fixable 2014-06-11 14:59:02 i kinda like dhcpcd 2014-06-11 14:59:32 i can plug in my usb ethernet adapter and it just works 2014-06-11 14:59:55 ncopa, i think probably you want to end up having a persistent wpa_supplicant that handles the wifi interface 2014-06-11 15:00:24 and i have a nice status icon on my desktop that tells me if i have link/address or not 2014-06-11 15:00:30 yes, doesn't seem to be too complicated: http://roy.marples.name/projects/dhcpcd-dbus/artifact/d64ac452e44eb788 2014-06-11 15:00:33 and a separate daemon that sees the interface going up/down, disables it when it's not needed, etc. 2014-06-11 15:00:40 dalias: might be 2014-06-11 15:01:06 so wpa_supplicant wouldn't be in control of the whole switchable network link system, but in control of its device 2014-06-11 15:01:30 as opposed to network-manager, which (as far as i can tell) invokes a new wpa_supplicant each time it connects to a network 2014-06-11 15:01:42 urgh.. 2014-06-11 15:01:50 ok i gotta go 2014-06-11 15:01:52 :) 2014-06-11 15:01:59 thanks for you input 2014-06-11 15:01:59 sorry i made you throw up ;-) 2014-06-11 15:02:45 see u 2014-06-11 15:05:46 ncopa: if i don't forget i might revisit this in near term from a less-talk more-code angle as i'm interested in this, so if you come up with something please keep me up to date. 2014-06-11 15:08:47 <^7heo> so, alpine linux is using syslinux? 2014-06-11 15:40:53 ^7heo: yes 2014-06-11 15:46:14 <^7heo> okay :) 2014-06-11 15:52:31 ncopa: i told gcc to uninline inline functions and the warnings went away. i was compiling 3.14.6 with latest grsec testing 2014-06-11 15:52:44 with my host's toolchain btw 2014-06-11 15:52:52 gcc 4.9.0 2014-06-11 15:53:09 the kernel will be loaded by qemu 2014-06-11 15:54:05 4.9.0 has several code generation bugs -- i'd be wary of compiling a kernel with it 2014-06-11 16:01:44 really? but it is not a production vm 2014-06-11 16:09:56 and then i wonder what toolchain does archlinux uses for its kernel 2014-06-11 16:10:15 the one it ships or something else.. 2014-06-11 16:10:48 intel's! 2014-06-11 16:10:58 that used to be able to compile it 2014-06-11 18:04:47 bleh 2014-06-11 18:04:51 during install i get 2014-06-11 18:05:13 /dev/sda1 alignment is offset by 3584 bytes. This may result in very poor performance. 2014-06-11 18:05:20 wtf is the fdisk utility doing wrong? :( 2014-06-11 18:33:28 <^7heo> how do you create your partition table? 2014-06-11 18:34:41 setup-alpine created it 2014-06-11 18:35:08 apparently busybox fdisk just sucks and doesn't know how to align partitions right 2014-06-11 18:53:26 <^7heo> dalias: if you're having an UEFI, maybe the busybox fdisk is a bit old 2014-06-11 18:54:01 is it uefi related? 2014-06-11 18:54:06 <^7heo> 'cause 3584+512=4096 2014-06-11 18:54:10 <^7heo> 4096 = 4k 2014-06-11 18:54:19 <^7heo> and the new AF disks boundaries have to be placed at 4k 2014-06-11 18:54:23 <^7heo> not at 512 bit 2014-06-11 18:54:29 <^7heo> byte 2014-06-11 18:54:34 <^7heo> my bad 2014-06-11 18:54:58 <^7heo> moreover, one sector in AF format is 4k, not 512 byte 2014-06-11 18:55:38 <^7heo> therefore, if you're having partitions at 512 when they are supposed to be at 4k, or the other way around, you might have a problem related to AF/UEFI. 2014-06-11 19:19:39 ^7heo, any idea how to fix it? 2014-06-11 19:19:54 <^7heo> I don't have enough info on your problem. 2014-06-11 19:20:08 what info could i provide that would help? 2014-06-11 19:20:56 <^7heo> first, what BIOS/UEFI you have 2014-06-11 19:21:01 <^7heo> then the fdisk output. 2014-06-11 19:21:15 <^7heo> but I'm watching footage from the E3 at the same time 2014-06-11 19:21:20 <^7heo> so it could take time. 2014-06-11 19:21:33 if it's just a performance issue, it might not matter; this is just /boot which doesn't really have to perform well 2014-06-11 19:22:45 <^7heo> http://www.ibm.com/developerworks/linux/library/l-4kb-sector-disks/ 2014-06-11 19:30:54 ok based on my reading of this it probably doesn't matter 2014-06-11 19:31:16 sda2 (swap) and sda3 (root fs) are aligned 2014-06-11 19:32:48 thanks for the link 2014-06-11 19:32:54 i think i'll try and see if it boots... 2014-06-11 19:34:03 ok it boots :) 2014-06-11 23:30:30 hmm alpine's grsec kernel cripples /proc/$pid/maps, etc.... is there a way to change this setting without recompiling the kernel? 2014-06-11 23:30:39 it makes it impossible to evaluate memory usage 2014-06-12 00:37:44 anyone on musl edge using openldap? 2014-06-12 00:37:50 it segfaults for me 2014-06-12 00:43:56 no, but have you tried gdb or strace to see what it's doing when it crashes? 2014-06-12 00:45:36 hmm 2014-06-12 00:47:37 --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0x117c8} --- 2014-06-12 00:47:37 +++ killed by SIGSEGV +++ 2014-06-12 00:47:37 Segmentation fault 2014-06-12 00:47:38 ACTION shrugs 2014-06-12 00:49:28 does that mean anything to you dalias 2014-06-12 00:50:35 hm 2014-06-12 00:51:15 it's an invalid pointer dereference with a low but nonzero address 2014-06-12 00:51:41 it could be memory corruption, or accessing a moderate-size (~70k) offset from a null pointer 2014-06-12 00:51:50 gdb backtrace might show more interesting 2014-06-12 00:54:06 Reading symbols from /usr/sbin/slapd...(no debugging symbols found)...done. 2014-06-12 01:04:48 hmm dhcpcd-ui seems to work 2014-06-12 07:11:27 dalias: I am not aware of a way to disable /proc/$pid/maps 2014-06-12 07:26:47 dalias, would be nice if gdb backtrace worked now... :) 2014-06-12 07:36:54 Hello 2014-06-12 07:37:05 fabled: +1 2014-06-12 07:37:11 morning 2014-06-12 07:42:20 when running diskless from iso, there is a gotcha when adding an internal hardisk for apk cache and lbu commits ? on reboot the iso dosen't know about /media/sda1 and since the ISO is read only all settings aren't loaded on reboot 2014-06-12 07:43:22 unless there is a cmdline parameter I can pass that some script takes care of reading the drive and extracting the files 2014-06-12 07:44:48 Mp5shooter: could you please file a bug for it? 2014-06-12 07:53:03 alpine_dev= '/dev/storagename':fsystem ? 2014-06-12 07:57:07 Good morning... 2014-06-12 07:58:53 On a embeded Alpine system running of a CF card, I have a local package repository /media/sda1/apks/x86 containing basic packages. Then there's a 'apk-cache' directory, where packages installed from a remote repo are saved... 2014-06-12 07:59:23 Why is it, that some packages appear in the 'apk-cache' even though the same package (including version) is available in the local repo? 2014-06-12 07:59:40 I also expected that 'apk cache clean' would get rid of those... 2014-06-12 08:28:19 danci1973, do you have /media/sda1/apks in /etc/apk/repositories? 2014-06-12 08:28:27 Yes. 2014-06-12 08:29:51 And it works too - I can install packages from it and these don't get copied to 'apk-cache'. 2014-06-12 08:29:55 if for some reason was not there before, i don't think "apk cache sync" would get rid of those in cache that exist in local repo 2014-06-12 08:30:23 rnalrd: It seems that way, yes... 2014-06-12 08:30:43 i think apk never remove packages from cache 2014-06-12 08:31:00 rnalrd: So what does 'apk cache clean' do? :) 2014-06-12 08:31:42 accord to help output should do exactly that :) 2014-06-12 08:33:14 It does remove 'old versions' of a certain package... For example - there were 'bash-4.2.010-r0' and 'bash-4.2.037-r0' in my 'apk-cache' - cache clean removed the older one... 2014-06-12 08:33:31 So it does something good. :) 2014-06-12 09:19:40 goodmorning 2014-06-12 09:51:15 morning 2014-06-12 10:02:12 how does alpine-iso know where to look for hostname.tar.gz (made with lbu) on boot 2014-06-12 10:03:34 it looks for *.apkovl.tar.gz 2014-06-12 10:04:01 and if it finds multiple it gets problems (i dont remember exactly what it does then...) 2014-06-12 10:04:21 zxd: http://git.alpinelinux.org/cgit/mkinitfs/tree/initramfs-init.in#n50 2014-06-12 10:04:39 echo "ERROR: More than one apkovl file was found on $(basename $mnt). None will be read." >&2 2014-06-12 10:04:43 thanks clandmeter :) 2014-06-12 10:05:13 should you be off for lunch? :) 2014-06-12 10:11:52 ncopa: how does it know on which disk to look for the file ? 2014-06-12 10:14:14 it seems hardcoded to for i in usb floppy cdrom; do ... ovl=$(find_ovl /media/$i) 2014-06-12 10:23:14 I really like alpine-linux lbu feature thing 2014-06-12 10:23:48 and running the rest from ram 2014-06-12 10:25:26 zxd: you can specficy ovl device by boot param 2014-06-12 10:28:38 oh 2014-06-12 10:30:26 right ovl_dev 2014-06-12 10:45:54 zxd: it looks for first found usb dev and whatever it booted from iirc 2014-06-12 10:46:02 i'd like refaktor that 2014-06-12 10:46:22 i think it will only find usb sticks with vfat currently 2014-06-12 11:28:57 sure ncopa 2014-06-12 11:34:44 yes ovl_dev works beautiful 2014-06-12 11:34:50 clandmeter: thanks 2014-06-12 11:35:18 ovl_dev=sda1:ext4 2014-06-12 11:38:39 cool :) 2014-06-12 12:35:37 Mp5shooter: thanks. i cannot reproduce it 2014-06-12 13:27:26 o 2014-06-12 13:46:47 ncopa, here's the option that breaks /proc/$pid/maps: GRKERNSEC_PROC_MEMMAP 2014-06-12 13:46:55 see http://en.wikibooks.org/wiki/Grsecurity/Appendix/Grsecurity_and_PaX_Configuration_Options 2014-06-12 13:47:32 the claims about it being essential in order for ASLR not to be useless are nonsense. the files it disables are visible only by the user who owns the process 2014-06-12 13:47:57 but there doesn't seem to be any way to fix it at runtime :/ 2014-06-12 13:48:18 yes, i think i bumped into this when running some memory usage script some time ago 2014-06-12 13:48:40 if you'd like i can do a review of the grsec options in use for (a) ones that are likely to break things, and (b) ones where the grsec description of their option is misleading or outright lies about its usefulness/necessity :) 2014-06-12 13:49:18 or i can ask spender if he can make it runtime configurable 2014-06-12 13:49:50 *nod* 2014-06-12 13:49:59 that would work too 2014-06-12 13:50:15 you need it to analyze memory usage, right? 2014-06-12 13:50:19 yes 2014-06-12 13:50:23 you could also apk add linux-vanilla 2014-06-12 13:50:27 also /proc/$pid/smaps doesn't even exist 2014-06-12 13:50:42 ah that might be a better solution 2014-06-12 13:51:04 these days it seems like grsec has maybe <10 features that aren't in mainline kernel but actually useful for security 2014-06-12 13:53:49 i am mosly interested in the PaX features 2014-06-12 13:55:13 i suppose the GRKERNSEC_PROC_MEMMAP is so you dont get a random address that you can easily read from /proc 2014-06-12 13:55:51 i suppose the idea with a random address is so that you should not be able to know where the address is at runtime 2014-06-12 13:56:17 so if you can read it from /proc you know where the stuff is... 2014-06-12 13:56:48 only protects against local exploits 2014-06-12 13:58:02 there are also problems with JIT compilers and grsecurity 2014-06-12 13:58:15 for those we disable the pax protections :-/ 2014-06-12 13:58:33 no, it doesn't protect against local exploits 2014-06-12 13:58:51 how so? 2014-06-12 13:59:00 if you're reading /proc/.../maps locally, the process you're attacking has the same perms you do 2014-06-12 13:59:04 so there's no point 2014-06-12 13:59:15 it mildly protects against remote 2014-06-12 13:59:27 if you could somehow trick a webserver to serve its own /proc/self/maps... 2014-06-12 13:59:33 you could use that to bypass aslr 2014-06-12 13:59:43 i think that's what they mean 2014-06-12 14:00:09 anyway the attack vector is getting a privileged process to reveal its own maps to you 2014-06-12 14:00:15 not manually reading maps with your own permissions 2014-06-12 14:02:08 hm 2014-06-12 14:02:10 might be 2014-06-12 14:06:13 should I ask spender about disable MEMMAP runtime? 2014-06-12 14:06:52 anyway i think it could break lots of things.. glibc's pthread_getstack for instances parses /proc/self/maps to find the main thread's stack. musl doesn't use it for anything, but apps might, and maps (or preferably smaps) are the only ways to measure memory usage on linux 2014-06-12 14:07:11 yeah, runtime-disable should be fine 2014-06-12 14:07:18 btw what did you do about the arm kuser thing? 2014-06-12 14:07:29 i'm looking to solve that problem more robustly on musl's side now 2014-06-12 14:07:36 arm kuser? 2014-06-12 14:07:48 its fabled that has been working on arm mostly 2014-06-12 14:07:55 ok i'll ask him 2014-06-12 14:08:00 haven't seen him in a few days tho 2014-06-12 14:08:09 i think he have been busy last days 2014-06-12 14:08:59 fabled: ^^^ 2014-06-12 14:19:22 dalias: did you get wpa_supplicant working? how did yo solve it? 2014-06-12 14:20:18 yeah, i forget which stage of the problem i was at when i last talked to you 2014-06-12 14:20:48 for some reason, even once wpa_supplicant was configured right, it would see the network but refuse to connect until i issued "iwlist scan" from another terminal.... 2014-06-12 14:20:48 hmm dhcpcd-ui seems to work 2014-06-12 14:21:02 but after that i got things setup 2014-06-12 14:21:11 and now i switched to dhcpcp-ui, yes 2014-06-12 14:21:20 sounds like driver or firmware issue 2014-06-12 14:21:21 and it's working for my wpa-psk ap at home 2014-06-12 14:21:26 nice 2014-06-12 14:21:35 haven't tried anywhere else 2014-06-12 14:21:50 i don't really have enough installed/configured yet for it to be useful as a main laptop yet 2014-06-12 14:22:00 but what i do have is working great 2014-06-12 14:22:07 good 2014-06-12 14:22:10 firefox plays html5 youtube at 20% cpu 2014-06-12 14:22:23 vs 130% cpu for chrome+flash on my old (but comparable) laptop 2014-06-12 14:22:41 flash... ugh 2014-06-12 14:23:06 i think i have been able to play most videos on youtube 2014-06-12 14:23:16 once in a while i bump into something that does not play 2014-06-12 14:23:26 then its normally possible to play it in vlc 2014-06-12 14:23:43 oh btw, vlc works much better in musl than in uclibc 2014-06-12 14:23:49 it used to hang with uclibc 2014-06-12 14:23:53 some threading issue 2014-06-12 14:24:00 ..done wrong in uclibc 2014-06-12 14:24:50 clandmeter: i would not be surprised if freeswitch runs better with musl than with uclibc due to better pthread implementation 2014-06-12 14:25:15 well, i just bit a bug i think 2014-06-12 14:25:36 yeah, depending on which implementation you're using, uclibc's pthread ranges from bad to unusable... 2014-06-12 14:26:02 the webif loads very slow and will spit an error in fs.log 2014-06-12 14:26:33 2014-06-12 14:08:21.718259 [ERR] switch_xml.c:1715 Error[[error near line 8]: unexpected closing tag

] 2014-06-12 14:27:35 could be snprintf format issue or similar 2014-06-12 15:45:03 is it possible to save a wildcard for lbu and have lbu commit ls the wildcard with every commit? 2014-06-12 15:45:24 ah sec 2014-06-12 15:46:04 will lbu include '/root/*' work? 2014-06-12 15:47:02 looks like it's good 2014-06-12 19:10:29 congrats for the mesa packaging 2014-06-12 19:10:34 really 2014-06-12 20:29:18 if i can pay star citizen on alpine some day i'll buy a round for the whole channel (based on current size hehe) 2014-06-12 21:34:34 ? 2014-06-12 22:10:33 darkfader: what stops you? 2014-06-12 22:11:05 Anaphaxeton: they didn't do the linux port yet 2014-06-12 22:11:16 but it's already planned 2014-06-12 22:11:19 so maybe nothing 2014-06-12 22:11:25 we'll see in a year i guess 2014-06-12 22:11:38 mesa support will raise the chances a lot 2014-06-12 22:15:20 oh the problem is the proprietary drivers 2014-06-12 22:15:48 i am lucky to have been avoiding them for years hehe 2014-06-12 22:20:19 i think i have the oss drivers on my ubuntu box 2014-06-12 22:20:29 because, moving windows and stuff is shitty like in win95 :) 2014-06-13 05:32:41 darkfader, if moving windows is slow, the problem is that you have NO DRIVER, not an open source driver 2014-06-13 06:40:36 morning 2014-06-13 06:42:28 cgit does not handle clone-url 2014-06-13 06:42:30 i dunno why 2014-06-13 06:46:00 goodmorning 2014-06-13 07:16:46 dang 2014-06-13 07:17:09 directfb has difficulties with musl it seems 2014-06-13 07:18:06 how well does musl work with glibc binary compatability as shipped by you guys? 2014-06-13 07:33:04 i dont think its compatible things like with nivdia/ati blobs 2014-06-13 07:33:38 with things like... 2014-06-13 07:45:11 we dont really provide any glibc ABI compat 2014-06-13 07:45:19 heh 2014-06-13 07:45:32 from reading musl's docs itself seemed like a very incomplete implementation 2014-06-13 07:45:46 actually, i think musl tries to be paritally ABI compat 2014-06-13 07:45:47 i know i asked this before and sorry for asking again but does anyone still have that eglibc based iso floating around? 2014-06-13 07:46:10 i think their ultimate goal is to have ABI compat but dear god that's a chore and a half 2014-06-13 07:46:22 i have been thinking of providin a glibc-compat package 2014-06-13 07:46:24 that is 2014-06-13 07:46:33 how would you go about that? 2014-06-13 07:46:33 provide glibc binary only 2014-06-13 07:46:36 arch based chroot? 2014-06-13 07:46:36 no headers 2014-06-13 07:46:39 no 2014-06-13 07:46:43 we simply build glibc 2014-06-13 07:46:50 and ship libc.so.6 2014-06-13 07:46:56 but not the headers 2014-06-13 07:47:23 should be enough to get atleast some glibc compat 2014-06-13 07:47:30 yeah 2014-06-13 07:47:39 anything that tries for libc.so direct could just use an LD_PRELOAD no? 2014-06-13 07:48:02 nothing should try libc.so direct 2014-06-13 07:48:59 0x0000000000000001 (NEEDED) Shared library: [libc.so.6] 2014-06-13 07:49:05 should look like that 2014-06-13 07:49:50 0x0000000000000001 (NEEDED) Shared library: [ld-linux-x86-64.so.2] 2014-06-13 07:49:50 0x000000000000000e (SONAME) Library soname: [libc.so.6] 2014-06-13 07:50:14 so if we provide ld-linux-x86-64.so.2 and libc.so.6 2014-06-13 07:50:17 we should be fine 2014-06-13 07:50:22 for most things 2014-06-13 07:50:44 what should it be calling for eglibc then? 2014-06-13 07:50:54 sorry? 2014-06-13 07:51:10 AH NVM 2014-06-13 07:51:15 ah nvm* 2014-06-13 07:51:19 looks like libc.so.0.9.32 2014-06-13 07:51:28 ACTION was looking in wrong paths 2014-06-13 07:51:33 yes 2014-06-13 07:51:44 i have an eglibc iso but its old. 2014-06-13 07:51:57 and musl has: /lib/libc.musl-x86_64.so.1 2014-06-13 07:52:16 i know people are trying to run some binary only things 2014-06-13 07:52:30 i have even myself needed one once for some printer/cups driver 2014-06-13 07:52:35 clandmeter: can you please link me? 2014-06-13 07:52:40 i dont remember what it was 2014-06-13 07:52:55 old is not an issue I just want something eglibc+apk based to use as a starting point for a project of mine 2014-06-13 07:52:55 but i remember simply copied libc.so.6 from debian and got it working 2014-06-13 07:54:11 stitch: http://alpine.nethq.org/releases 2014-06-13 07:54:19 <3 2014-06-13 07:54:21 ty 2014-06-13 07:54:31 http://alpine.nethq.org/aports/eglibc/edge/main/ 2014-06-13 07:54:43 dont ask for support :) 2014-06-13 07:55:02 OMG CAN YOU PLEASE HALP ME GET THIS INSTALLED I BE CLUELESS WITH LINUX WHERE DO I PUT IN THE ACTIVATION KEY?!?!? 2014-06-13 07:55:22 lol 2014-06-13 07:58:00 you are expecting it actually works :p 2014-06-13 07:59:17 clandmeter: more or less i'm trying to roll my own distro 2014-06-13 07:59:23 as long as apk, abuild,g cc and perl work 2014-06-13 07:59:25 i'll be happy 2014-06-13 07:59:56 i figure that should at least be enough to avoid the mess of creating cross-compiliers on a debian system or what not 2014-06-13 08:28:56 yeah signal.h seems wierd 2014-06-13 08:29:34 i will try sdl 2014-06-13 08:59:52 devs, now that wayland 1.5 is out, please consider mesa-dev in makedepends and mesa-egl in depends, pleeeease 2014-06-13 09:13:11 Anaphaxeton: i tried to build wayland with mesa-dev installed 2014-06-13 09:13:16 but it did not pick it up? 2014-06-13 09:13:28 does not look like there are any mesa related configure opts either? 2014-06-13 09:14:16 ah 2014-06-13 09:14:20 its the other way around 2014-06-13 09:17:35 well you have egl.pc 2014-06-13 09:17:42 it should have picked it 2014-06-13 09:18:00 no, i think you should build mesa with wayland support 2014-06-13 09:18:02 and indeed it does not appear in configure 2014-06-13 09:18:03 not the other way around 2014-06-13 09:18:11 hmm 2014-06-13 09:18:13 indeed 2014-06-13 09:18:23 http://wayland.freedesktop.org/building.html 2014-06-13 09:18:26 you build wayland 2014-06-13 09:18:29 there are the targets: drm,egl,wayland-egl iirc 2014-06-13 09:18:42 then you build mesa with --with-egl-platforms=x11,wayland,drm 2014-06-13 09:19:09 and wayland-egl appears? 2014-06-13 09:19:11 wow 2014-06-13 09:25:17 i remember building directfb with egl drm. at least that one went forward 2014-06-13 09:26:12 but then it makes sense since directfb contains a drawing api that would like to know what it is drawing with 2014-06-13 09:26:48 too bad they are so introverse 2014-06-13 09:34:50 i'll build mesa with libwayland-egl in a minute 2014-06-13 09:35:17 ok 2014-06-13 09:35:24 it seems necessary 2014-06-13 09:35:50 if we are going to take wayland seriously 2014-06-13 09:36:02 my host is fine with xorg :p 2014-06-13 09:36:14 building wayland was like 10 sec 2014-06-13 09:36:21 compared to minutes of xorg 2014-06-13 09:36:27 but the guest (alpine) feels more sexy 2014-06-13 09:36:40 seems like wayland is much less bloated 2014-06-13 09:36:45 wayland isnt much if i understand correctly 2014-06-13 09:36:59 try weston 2014-06-13 09:37:45 it is some sort of a helper lib and i dont know if it can do what xorg (or directfb) does 2014-06-13 09:41:50 hm 2014-06-13 09:41:51 right 2014-06-13 09:41:58 its weston that is the server 2014-06-13 09:42:23 for example what i want to do now is build sdl2 with wayland support and draw with that 2014-06-13 09:43:16 hm i dont know if i can get freetype this way... 2014-06-13 09:47:31 it seems i cannot get text at all this way! 2014-06-13 10:07:28 Hello all 2014-06-13 11:03:14 l 2014-06-13 12:47:05 hello again all 2014-06-13 12:51:10 I think I've located a bug with 3.0.0, and setup-disk script when you feed it multiple disks (doesn't happen with single disk) 2014-06-13 12:51:31 install goes through to initramfs, then xxxx blocks 2014-06-13 12:52:28 then I get "/lib/modules/3.14.5-0-grsec/kernel/drivers/block/cpqarray*: No such file or directory. Skipped" 2014-06-13 12:53:04 then "xxx blocks" "/boot is device md0" (i.e. all normal except the no such file or directory line) 2014-06-13 12:53:23 happens whether or not I include a repo 2014-06-13 12:53:51 in other words also happens with 3.14.6-0-grsec kernel 2014-06-13 12:56:07 I don't suppose anyone can replicate? 2014-06-13 12:58:38 does it boot up after reboot? 2014-06-13 12:58:48 oh, also happens if I do setup-disk -L /dev/sda /dev/sdb for example 2014-06-13 12:58:54 nope 2014-06-13 12:58:59 hm 2014-06-13 12:59:00 kernel panic :S 2014-06-13 12:59:17 well, drops to a prompt first 2014-06-13 12:59:30 'exit to continue booting; leads to kernel panic 2014-06-13 13:01:48 I should apologize tbh, I should have checked this ages ago! 2014-06-13 13:08:44 would you like me to raise a bug? 2014-06-13 13:13:19 yes maybe 2014-06-13 13:14:18 I'll give you some more time, just let me know if you want me to raise a bug :) 2014-06-13 15:27:20 bye all, have a great weekend! 2014-06-13 18:21:42 how can i search for installed packages if i installed in hdd? 2014-06-13 21:28:02 I'm working on making some Alpine tarballs for openvz systems 2014-06-13 21:28:36 the only problem I'm running into so far is that I need to have some commands run on boot 2014-06-13 21:32:30 n8@all 2014-06-13 21:55:05 how can i search for installed packages if i installed in hdd? 2014-06-14 11:45:36 how can i search for installed packages if i installed in hdd? 2014-06-14 15:45:16 how can i search for installed packages if i installed in hdd? 2014-06-14 15:45:18 do 2014-06-14 15:45:21 doh 2014-06-14 15:45:24 stupid weekends 2014-06-14 16:12:31 firefox crashes pretty often.. it'd be nice to figure out what's causing this 2014-06-14 19:43:27 hmm, it seems the cause of firefox crashing is SIGKILL from PAX for attempting to execute code from an anonymous mapping (probably js jit stuff?) 2014-06-14 19:50:00 ncopa, is installing linux-vanilla supposed to update extlinux.conf, or do i need to do that manually? 2014-06-14 20:22:02 *sigh* linux-vanilla has badly broken config 2014-06-14 20:22:31 for instance CONFIG_MOUSE_PS2_ELANTECH is disabled (makes laptop touchpads crippled) 2014-06-14 20:22:40 and CONFIG_SECCOMP is inexplicably off 2014-06-14 20:23:02 (looking at diff -u config config-grsec) 2014-06-15 23:42:24 When I'm trying hostname -F /etc/hostname I get operation not permitted. What could cause that? 2014-06-15 23:43:22 du 2014-06-15 23:53:34 are you root? 2014-06-16 06:15:09 monring 2014-06-16 06:15:14 morning 2014-06-16 06:15:25 dalias: i think SECCOMP beeing disabled in vanilla is a bug 2014-06-16 06:15:29 i'll fix that asap 2014-06-16 06:16:01 same with CONFIG_MOUSE_PS2_ELANTECH 2014-06-16 06:16:22 diff of the config's should be empty except for GRSEC stuff, no? 2014-06-16 06:16:28 yes 2014-06-16 06:16:32 i think there are a lot of other differences... 2014-06-16 06:16:34 well 2014-06-16 06:16:47 there are a few things that are not available when gresc is enabled 2014-06-16 06:18:23 well aside from them 2014-06-16 06:19:13 the diff file is 842 lines here :) 2014-06-16 06:20:39 and 113 after all the comments/blank lines/GRKERNSEC/PAX lines are removed 2014-06-16 06:22:06 can you 'apk add sprunge' and then ... | sprunge and paste the url here? 2014-06-16 06:22:20 i'm fixing seccomp and elantech 2014-06-16 06:22:33 looks like elantech was fixed in gresc but not fixed in vanilla 2014-06-16 06:23:18 we should probably build both kernel flavors from same APKBUILD and have its base in same config 2014-06-16 06:24:42 http://sprunge.us/WFEb 2014-06-16 06:25:22 hm 2014-06-16 06:25:33 maybe sort before diff? 2014-06-16 06:25:42 ok 2014-06-16 06:26:09 hm 2014-06-16 06:26:31 http://sprunge.us/TZcV 2014-06-16 06:27:34 thanks. I'll look over it 2014-06-16 06:27:51 i thought zram was enabled already 2014-06-16 06:27:54 thanks 2014-06-16 06:28:00 but it is not even shown 2014-06-16 06:28:09 so it must be some dependency thats disabled 2014-06-16 06:28:18 btw i think the firefox crashes are an incompatibility with some grsec features 2014-06-16 06:28:21 not musl-related 2014-06-16 06:28:24 yes 2014-06-16 06:28:25 at least the ones i've seen so far 2014-06-16 06:28:44 i noticed it too since ff-30 2014-06-16 06:28:55 probably something in their JIT 2014-06-16 06:28:56 i think you can workaround it with paxctl -p 2014-06-16 06:28:59 i think so too 2014-06-16 06:29:06 i am pusing a fix for that 2014-06-16 06:29:23 i just pushed it 2014-06-16 06:29:33 together with elantech and seccomp fix 2014-06-16 06:29:47 i'll look over the diff now 2014-06-16 06:29:50 still need to get around to setitng more stuff up, but so far alpine is pretty impressive 2014-06-16 06:30:56 ...thanks to musl ;) 2014-06-16 06:34:49 dalias: this is x86_64 right? 2014-06-16 06:38:43 ff starts almost instantly 2014-06-16 06:38:43 which is odd considering that musl doesn't have all of the over-engineered dynamic linker stuff glibc added specifically to make firefox load fast.... 2014-06-16 06:40:36 :) 2014-06-16 06:40:45 that is actually interesting 2014-06-16 06:41:38 is that with grsec kernel or vanilla? 2014-06-16 06:41:56 i wonder how much that is due to musl linker 2014-06-16 06:44:02 yes 2014-06-16 06:44:20 x86_64 2014-06-16 06:44:26 grsec kernel 2014-06-16 06:44:28 i wodner also if RELRO can have something to do with it 2014-06-16 06:44:42 i'm not sure because musl dynamic linker is not at all optimized 2014-06-16 06:44:54 relro shouldn't affect performance 2014-06-16 06:44:56 it could be more efficient shared memory too 2014-06-16 06:45:07 lack of lazy binding could affect performance tho 2014-06-16 06:45:30 conventional wisdom says lacking lazy binding would hurt performance 2014-06-16 06:45:34 but i suspect it helps 2014-06-16 06:45:48 since you do all the symbol resolution while it's fresh in the cache 2014-06-16 06:46:04 rather than repeatedly jumping back and forth between symbol resolution and program logic 2014-06-16 06:46:38 anyway tho musl's dynamic linker is insanely un-optimized 2014-06-16 06:46:53 well, it apparently do something right :) 2014-06-16 06:46:56 like we should have a check whether reloc # N+1 refers to the same symbol name that reloc N referred to 2014-06-16 06:47:03 and skip looking it up again 2014-06-16 06:47:22 ff-30 is slightly broke though 2014-06-16 06:47:25 i wonder why 2014-06-16 06:47:31 (afaik the linker does some sort of sorting to put multiple relocs against the same symbol adjacent to one another) 2014-06-16 06:48:17 session restore in firefox broke in 30.0 2014-06-16 06:48:41 yes, lots of other stuff is broken too 2014-06-16 06:48:48 none of the things on the start page work 2014-06-16 06:48:55 correct 2014-06-16 06:49:00 in 29.0 it worked 2014-06-16 06:49:17 actually 2014-06-16 06:49:27 i filed a bug about it some time ago: https://bugzilla.mozilla.org/show_bug.cgi?id=949375 2014-06-16 06:49:49 but when i installed firefox-jspdf it started to work again 2014-06-16 06:49:51 in 29.0 2014-06-16 06:50:21 with 30.0 the symptoms came back 2014-06-16 06:50:28 firefox-pfdjs 2014-06-16 06:50:42 i splitted out the internal pdf viewer into a separate package 2014-06-16 06:50:53 it worked just fin earlier 2014-06-16 06:51:18 but with 29.0 it caused session restore and about:home stuff break 2014-06-16 06:51:32 installing pdfjs appeared to solve things 2014-06-16 06:51:35 til 30.0.... 2014-06-16 06:52:35 we earlier had some other issues due to js doing various glibc assumptions 2014-06-16 06:53:39 one of the things was that they did dlopen('libc.so.6') for their FFI... 2014-06-16 06:54:13 which also broke things for freebsd's libc.so.7.... 2014-06-16 06:55:20 haha 2014-06-16 06:55:31 that actually should work for musl 2014-06-16 06:55:52 dlopen("libc.*") returns a handle to the internal libc 2014-06-16 06:55:55 only if we have a symlink named libc.so.6 i suppose 2014-06-16 06:55:59 ah 2014-06-16 06:56:02 really? 2014-06-16 06:56:03 cool 2014-06-16 06:56:21 they have soemthing similar in firefox 2014-06-16 06:56:33 but for 'a.out' 2014-06-16 06:56:36 yes. it's sorta a workaround for the fact that the kernel provides no information on the identity of the PT_INTERP it mapped to the process 2014-06-16 06:56:58 so it can't know if a given file in the filesystem is the same as the interpreter that's already loaded 2014-06-16 06:57:17 (it could by parsing /proc/self/maps, but of course that's a bad idea, see grsec :) 2014-06-16 06:57:24 hehe 2014-06-16 06:57:25 yes 2014-06-16 06:57:54 anyway since -lc is reserved by posix, it seems reasonable to intercept "libc.*" and treat it special 2014-06-16 06:58:06 a libc.so with a pathname won't be treated special tho 2014-06-16 06:58:13 only plain libc.* 2014-06-16 06:58:16 ok 2014-06-16 06:59:04 i find it interesting that removing pdfjs earlier gave same symptoms that we have now 2014-06-16 06:59:48 yes... 2014-06-16 07:01:12 re kernel config 2014-06-16 07:01:24 i must admit i find it challenging to do kernel config right... 2014-06-16 07:01:37 it has become such big/complicated task 2014-06-16 07:02:27 morning all! 2014-06-16 07:02:51 morning 2014-06-16 07:13:30 @ncopa: I'm just doing some more testing re: setup-disk script problem for mulitple disks I mentioned on friday. 2014-06-16 07:16:58 hmm, same thing under vbox as vmware server, so I don't think it is a freak host based occurance 2014-06-16 07:19:32 damn, same error under 3.0 rc1 too. "/lib/modules/3.14.4-0-grsec/kernel/drivers/block/cpqarray*: No such file or directory. Skipped" 2014-06-16 07:19:54 and I can't spell my name correctly either...not a good start to the day! 2014-06-16 07:20:32 cpqarray error is not really a problem 2014-06-16 07:20:49 it just does not find some drivers it thought would be there 2014-06-16 07:21:27 ok let me run it again and boot into the installed version....maybe I won't get kernel panic in RC1? 2014-06-16 07:28:31 yeah same issue with RC1.....boot into the newly installed system, after syslinux countdown I get "Mounting root failed. initramfs emergency recovery shell launched. Type 'exit' to continue boot sh:can't access tty; job control turned off" 2014-06-16 07:28:51 and doing 'exit' gives me a kernel reboot 2014-06-16 07:28:59 *panic 2014-06-16 07:39:55 so I'm lost really 2014-06-16 07:43:31 definitely don't get the problem with single disk sys install. 2014-06-16 07:44:29 I guess, to narrow it down (and state the obvious) could be a musl issue, or could be something in the kernel version (although 3.14-6 had the same issue)? 2014-06-16 07:46:24 skal jeg heve en bug på bug tracker? (shall i raise a bug on bug tracker?) 2014-06-16 08:02:20 moin 2014-06-16 08:02:47 which package is missing if openvpn gives the following: 2014-06-16 08:02:48 ERROR: Cannot open TUN/TAP dev /dev/net/tun: No such file or directory (errno=2) 2014-06-16 08:02:52 ? 2014-06-16 08:02:56 Good Morning StarWarsFan 2014-06-16 08:03:25 hi 2014-06-16 08:03:41 ah, i should read http://wiki.alpinelinux.org/wiki/Setting_up_a_OpenVPN_server ;-) 2014-06-16 08:03:42 have you configured a tun/tap device? 2014-06-16 08:04:06 sounds like a missing modprobe tun or similar 2014-06-16 08:04:38 right, was missing that part ;-) 2014-06-16 08:27:54 ok, created Bug #3040 for setup-disk issue 2014-06-16 08:28:39 thanks algitbot. So very helpful :) 2014-06-16 08:28:56 Can algitbot make tea too? 2014-06-16 08:49:31 algitbot: sudo make tea 2014-06-16 08:49:46 algitbot: sudo make sandwich 2014-06-16 08:50:29 hm 2014-06-16 08:50:35 radio adapter support in kernel 2014-06-16 08:50:36 is off 2014-06-16 08:50:41 do we want enable it? 2014-06-16 08:50:50 i suppose we just let it off til someone asks for it 2014-06-16 08:54:30 lol, I so want one now! 2014-06-16 11:47:19 I have a quick question re: raising bugs/feature requests. There is obviously the option to assign the ticket to someone...is that for the end user to decide, or presumably it's better to leave blank and let the dev's decide? 2014-06-16 11:51:46 I usually leave it blank anyway; I don't think I have the authority to assign tickets out! 2014-06-16 11:52:46 if its a bug for a certain package, you can assign it to the maintainer (if y ou are allowed to do that) 2014-06-16 11:52:50 not sure y ou can by default. 2014-06-16 11:53:46 ok thanks. If I am in any doubt I will continue to leave it blank 2014-06-16 12:37:57 hoi tg 2014-06-16 15:27:18 bye everyone, have a good evening! 2014-06-16 16:18:50 hello 2014-06-16 18:49:11 hello, anybody using Xen server on alpine here? 2014-06-16 18:50:17 I've actually got some VM's running, but I'm having network troubles, I see it is best to have each VM with a seperate IP, but I don't quite have that luxury 2014-06-16 18:50:48 so I was leaning towards using some sort of NAT ing. there is where the experiment ended... 2014-06-16 18:52:58 so if anyone has had experience, or a link to a good wiki - thaat would be greatly appreciated. 2014-06-16 20:40:45 hi, i just installed alpine 3.0 (first time) and I cant work out what packages I need to compile code, installed gcc but I get cannot find Scrt1.o 2014-06-16 20:42:09 try the alpine-sdk metapackage 2014-06-16 20:42:33 ah thanks. also just found musl-dev which helped... 2014-06-16 21:39:57 ACTION rawrs at awilfox|phone 2014-06-17 06:06:46 justincormack: apk add build-base 2014-06-17 07:08:56 Hello all 2014-06-17 15:25:00 in what package is ping ? 2014-06-17 15:25:14 not busybox 2014-06-17 15:27:53 probably iputils at a guess 2014-06-17 17:38:36 ping belongs to bb 2014-06-17 17:38:42 but its probably a symlink 2014-06-17 17:38:48 which is not in the apk db 2014-06-17 17:39:07 it even says busybox when you run it without args. 2014-06-17 17:54:06 oh btw 2014-06-17 17:54:21 alpine seems to have ping as a symlink to bbsuid 2014-06-17 17:54:33 it would be nice to instead apply the patch for make ping work without root on modern kernels 2014-06-17 17:54:58 using SOCK_DGRAM IPPROTO_ICMP sockets 2014-06-17 17:56:14 busybox's suid framework is clearly insecure -- for example ping does all the program logic without dropping root rather than opening the raw socket as the first line of the applet's main then dropping root 2014-06-17 17:59:36 hi dalias 2014-06-17 17:59:48 how is alpine? 2014-06-17 18:00:14 overall i like it so far 2014-06-17 18:00:23 need to spend more time setting stuff up still 2014-06-17 18:00:31 i've been busy working on musl tho 2014-06-17 18:00:44 ok nice 2014-06-17 18:01:01 i'm really curious about how/why firefox is so fast starting up 2014-06-17 18:01:29 when you have some free time, take a look at nfs. Seems ncopa is unable to fix it. 2014-06-17 18:01:43 since it's supposedly notorious for having huge relocation overhead at startup, and musl's dynamic linker isn't optimized at all for this situation 2014-06-17 18:01:58 yet the totally unoptimized code seems to be doing better than glibc's 2014-06-17 18:02:03 ok 2014-06-17 18:02:14 ncopa, can you fill me in on the current status of nfs work? 2014-06-17 18:03:04 i guess he is afk, ill msg him tomorrow morning. 2014-06-18 06:21:06 morning 2014-06-18 06:21:22 dalias: basically nfs compiles (try apk add nfs-utils) 2014-06-18 06:21:52 i runs (try config /etc/exports and /etc/init.d/nfs start) 2014-06-18 06:22:07 but when you from a client try mount it, it just hangs 2014-06-18 06:22:35 i have tried start rpcbind and the rpc.mountd daemons in foreground with debugging on 2014-06-18 06:22:44 i can see that there is rpc activity 2014-06-18 06:23:26 i can see that nfsd detects the client (and caches it?) 2014-06-18 06:23:33 but then it just hangs 2014-06-18 06:23:40 i don't know where... 2014-06-18 06:24:27 have you tried strace or gdb? 2014-06-18 06:40:31 i tried strace but it didnt really tell me anything 2014-06-18 06:40:34 gdb i havent tried 2014-06-18 06:45:01 i did try with nfsvers=2 to avoid the idmap and gss stuff 2014-06-18 06:55:03 dalias: btw, bbsuid is our own binary so we dont need busybox be suid root 2014-06-18 06:55:12 moin 2014-06-18 06:55:37 re ping as non-root, i didnt think the ping as non-root patch made it to mainline kernel? 2014-06-18 06:56:43 moin kocka 2014-06-18 06:56:51 sup there? 2014-06-18 07:02:35 ncopa, ping as nonroot is in mainline 2014-06-18 07:02:58 and i know the main busybox binary isn't suid, but bbsuid is still dangerous imo 2014-06-18 07:03:17 yes, im not happy with it either 2014-06-18 07:03:28 we should fix ping then 2014-06-18 07:03:54 you need to set a file in /proc/sys to control which gids can use SOCK_DGRAM ping tho 2014-06-18 07:04:08 default is noone, not even root :) 2014-06-18 07:04:13 heh 2014-06-18 07:04:22 does it need special kernel config option? 2014-06-18 07:04:28 it might 2014-06-18 07:04:59 we have an issue in our bugtracker for fping too 2014-06-18 07:05:00 i don't see a CONFIG item for it 2014-06-18 07:06:58 the one i looked at last time was this: http://lwn.net/Articles/420800/ 2014-06-18 07:09:11 ncopa, commit c319b4d76b9e583a5d88d6bf190e079c4e43213d 2014-06-18 07:09:17 on 2011 it was applied upstream 2014-06-18 07:09:21 net: ipv4: add IPPROTO_ICMP socket kind 2014-06-18 07:09:58 nice 2014-06-18 07:38:48 ha 2014-06-18 07:39:26 it looks like fping actually have support for the SOCK_DGRAM ping 2014-06-18 07:39:36 but guess what happens when you try use it? 2014-06-18 07:40:24 [418536.560360] BUG: unable to handle kernel NULL pointer dereference at (nil) 2014-06-18 07:43:15 oh? 2014-06-18 07:43:27 it's working for me on 3.15 2014-06-18 07:43:53 https://gist.github.com/ncopa/79318d70414d8f6d23b0 2014-06-18 07:44:06 it could be a regression or it could be something gresc does 2014-06-18 07:44:18 dalias: did you try fping? 2014-06-18 07:44:23 on 3.15 kernel 2014-06-18 07:45:32 busybox ping with the patches 2014-06-18 07:45:35 i don't have fping here 2014-06-18 07:45:47 do you have patch for bb ping for it? 2014-06-18 07:46:18 yes 2014-06-18 07:46:52 http://sprunge.us/eLDM 2014-06-18 07:47:33 thanks! we want this for alpine 2014-06-18 07:47:54 it should match the two patches sent to the bb list (one applied on the other) 2014-06-18 07:48:28 sadly it's in limbo upstream because it "adds too much bloat" :( 2014-06-18 07:48:37 i hope someone can debloat it a bit so we can get it upstream 2014-06-18 07:50:25 anyway i'm going to sleep now 2014-06-18 07:53:41 good night 2014-06-18 11:30:19 it does look interesting: http://0pointer.de/blog/projects/stateless.html 2014-06-18 11:30:33 similar concept as our tmpfs or 'livecd' mode 2014-06-18 11:41:21 ncopa: sounds interesting, yes. i don't trust them anyway 2014-06-18 11:41:43 "look we wrote a bunch of critical features for embedded use in our basement" 2014-06-18 11:42:23 it could be really nice. but personally i'd avoid testing it 2014-06-18 11:45:21 what i find interesting is that they add features that is similar what we have had all the time 2014-06-18 11:45:52 and they have a solution for a problme we have too 2014-06-18 11:45:56 the sysusers thingy 2014-06-18 11:59:26 but apparently, its sounds like its easy to mess up your system with systemd, and if you do, its non-trivial to fix it 2014-06-18 11:59:37 so they need a simple way to do 'factory reset' 2014-06-18 12:16:29 was there any discussion of moving /[s]bin to /usr/[s]bin ? 2014-06-18 12:16:38 http://www.freedesktop.org/wiki/Software/systemd/TheCaseForTheUsrMerge/ 2014-06-18 12:16:46 or still early, as only few distro have implemented it 2014-06-18 12:30:15 the stateless or statechange can also done by supplying the path to overlay disk/partition/file 2014-06-18 12:30:32 at boot time/param 2014-06-18 12:31:27 I have been doing this on knoppix for maybe more than 4yrs now 2014-06-18 12:31:29 i absolutely don't understand why if you want to merge */bin you'd choose /usr/bin as canonical location and not /bin 2014-06-18 12:31:59 idea to get a vendor specific distro in one folder 2014-06-18 12:32:24 while making a symlink of /bin -> /usr/bin 2014-06-18 12:33:04 so what's wrong with doing it the other way around? 2014-06-18 12:33:17 /usr is a kludge 2014-06-18 12:33:26 the main benefit I see is packages from different distros that add same file to either /bin or /usr/bin 2014-06-18 12:33:43 would be unified 2014-06-18 12:34:17 it kinda make sense to move read-only stuff to /usr 2014-06-18 12:34:26 coz what is a USER bin cannot be called system bin but visa versa is ok 2014-06-18 12:34:42 but varius things they mention on http://www.freedesktop.org/wiki/Software/systemd/TheCaseForTheUsrMerge/ does not apply to us 2014-06-18 12:36:17 sorry, to me it just looks like the sort of faecal matter all systemd people appear to be thinking 2014-06-18 12:37:23 i understand why you'd want to have one /bin, but why not get rid of /usr while you're at it? 2014-06-18 12:37:53 Myth #11: Instead of merging / into /usr it would make a lot more sense to merge /usr into /. 2014-06-18 12:37:53 Fact: This would make the separation between vendor-supplied OS resources and machine-specific even worse, thus making OS snapshots and network/container sharing of it much harder and non-atomic, and clutter the root file system with a multitude of new directories. 2014-06-18 12:38:09 i kinda like the idea of having all vendor specifics in /usr 2014-06-18 12:38:29 but myth 9 and 10 are bogus 2014-06-18 12:38:49 Fact: On Fedora the root directory contains ~450MB already. 2014-06-18 12:39:20 what is alpine's vendor? 2014-06-18 12:39:44 well on alpine you can have a fully usable system in less than 10MB as root 2014-06-18 12:39:50 alpine is the 'vendor' 2014-06-18 12:40:30 I think this ~450MB is the reason that has triggered all this 2014-06-18 12:40:33 so if alpine, as vendor, is in /usr, who's got /? 2014-06-18 12:41:03 that was my general feeling too 2014-06-18 12:41:18 "we cannot make it smaller than 450MB so why bother..." 2014-06-18 12:42:10 i havent bothered move things to /usr yet 2014-06-18 12:42:23 i dont have any specific plans for doing it either 2014-06-18 13:03:13 aufs also supports delete(whiteouts), a feature I think is not yet fully exploited 2014-06-18 13:03:43 used ;) 2014-06-18 13:03:52 "snapshots of them become atomic" - feature something that is inherent in Btrfs 2014-06-18 13:04:19 low backup overhead 2014-06-18 13:05:15 but I am skeptic about /lib → /usr/lib and /lib64 → /usr/lib64 2014-06-18 13:05:33 i'm sceptic too 2014-06-19 08:06:36 morning all! 2014-06-19 08:30:42 morning 2014-06-19 08:30:47 moin 2014-06-19 10:55:50 Any reason why we have libstdc++, but not gcc-c++ in Alpine ? 2014-06-19 10:59:24 lsp:~$ apk search g++ 2014-06-19 10:59:24 g++-4.8.2-r1 2014-06-19 10:59:27 it's g++ 2014-06-19 10:59:35 hmm 2014-06-19 10:59:44 Then I'm an idiot :) 2014-06-19 11:00:27 I searched for gcc, since must distro's prefix the c++ with the compiler version 2014-06-19 11:00:29 Ah well.. 2014-06-19 11:04:08 igmar: i don think that qualifies for "idiot" :) 2014-06-19 11:14:24 I should have thought of that :) 2014-06-19 11:14:55 its always easy to be "after-smart" :) 2014-06-19 11:15:05 Also true 2014-06-19 11:15:23 captain obvious to the rescue :) 2014-06-19 11:15:38 Time to dig trough the bugs database 2014-06-19 11:26:37 set auto_away off 2014-06-19 11:26:52 sorry, this BitchX is kinda virgin 2014-06-19 11:28:49 that's a bad thing? ;) 2014-06-19 11:35:35 In this case : Yes :-) 2014-06-19 11:38:44 ncopa: any ideas how can i tinker with wpa_supplicant from a virtual machine? looks like none of the usual suspects implement any 802.11 hardware 2014-06-19 11:43:55 ovf: you will have to passthrough an actual wifi interface I think, not sure if Qemu is able to emulate on of those, neither how would it work 2014-06-19 11:44:52 unless there is some virtio-something magic around it i don't think you can pass through an operating system object to a virtual machine 2014-06-19 11:45:29 i'd prefer an emulated 802.11 chip though 2014-06-19 11:45:38 you need pci passthought I think 2014-06-19 11:47:55 you might get away with a wireless USB device 2014-06-19 11:58:52 pci passthrough is probably the most viable option 2014-06-19 12:00:29 ovf: by passthrough I mean PCI passthrough of a real device 2014-06-19 12:03:37 yep, the host is a laptop, so should work 2014-06-19 12:10:26 would haev been nice with a virtual wifi hw 2014-06-19 12:13:45 This bamboos script makes me cry 2014-06-19 12:24:44 i think implementing something like orinico wavelan is feasible, but then you'll also want to test wpa... 2014-06-19 12:24:55 orinoco 2014-06-19 13:20:39 something in the combination of host archlinux, virtualbox, iwlwifi and guest alpinelinux is uncooperative in doing pci passthrough. go figure 2014-06-19 13:29:15 hai 2014-06-19 13:33:27 ncopa: looks like from alpine's chroot apk can't fetch anything so I have to use apk static from outside chroot to add for example python 2014-06-19 13:33:56 sabre:~# apk add perl 2014-06-19 13:33:56 ERROR: unsatisfiable constraints: 2014-06-19 13:33:56 perl (missing): 2014-06-19 13:33:56 required by: world[perl] 2014-06-19 13:35:26 slashbeast: did you do apk update? 2014-06-19 13:35:43 do you have anything in /etc/apk/repositories? 2014-06-19 13:39:33 no repositories file at all. 2014-06-19 13:39:45 ah, I skipped the part with echo. 2014-06-19 13:39:49 totally my fault 2014-06-19 13:44:23 i'd kinda recommend using lxc instead of pure chroot 2014-06-19 13:45:42 I do use proot 2014-06-19 13:45:58 lxc is too insecure to run as root there. 2014-06-19 13:46:06 proot does the trick with close to minimal perofrmance lost 2014-06-19 13:46:23 proot -0 -R rootfs/ busybox sh -l 2014-06-19 13:47:15 ncopa: the target for alpine is just a routers and deployments like that or you do presume that some may found its use as general purpose server system? 2014-06-19 13:49:12 i know people uses it for general purpose server systems too 2014-06-19 13:49:29 http://alpinelinux.org (drupal) runs alpine 2014-06-19 13:49:39 bugs.alpinelinux.org redmine runs alpine 2014-06-19 13:49:48 Executing busybox-1.22.1-r5.trigger 2014-06-19 13:49:48 ERROR: busybox-1.22.1-r5.trigger: script exited with error 1 2014-06-19 13:49:51 how can I debug it btw? 2014-06-19 13:50:02 mailing list, wiki all runs alpine 2014-06-19 13:50:04 strace? 2014-06-19 13:50:12 i use it for mail+dns+web(mail) 2014-06-19 13:50:13 debuging a script with strace? 2014-06-19 13:50:13 wow 2014-06-19 13:50:25 i run alpine on my work desktop 2014-06-19 13:50:39 and my private laptop 2014-06-19 13:51:23 looking at the trigger script 2014-06-19 13:51:40 strace can tell you if the script is actually execute or not 2014-06-19 13:51:49 surprisly I can't find ith with simple find / -name busybox-1.22.1-r5.trigger 2014-06-19 13:52:27 its executed from a temp dir 2014-06-19 13:52:37 so apr-util-dev installed postgresql and openldap 2014-06-19 13:53:01 anyway, i was asking because I found it surprisly suitable for small virtual machines and such. 2014-06-19 13:53:01 apk fetch busybox --stdout | tar -zt 2014-06-19 13:53:44 apk fetch busybox --stdout | tar -zx -O .trigger 2014-06-19 13:54:18 http://git.alpinelinux.org/cgit/aports/tree/main/busybox/busybox.trigger 2014-06-19 13:54:55 do you have any roadmap for alpine, like where it is going? 2014-06-19 13:55:03 and do you offer any commercial support to this distro? 2014-06-19 13:55:11 if the script executes at all, then the only thing that can cause it is if /bin/busybox --install -s 2014-06-19 13:55:37 i havent written down any general direction roadmap 2014-06-19 13:55:56 not mre than this: http://bugs.alpinelinux.org/projects/alpine/roadmap 2014-06-19 13:56:03 comercial support start at 1 million USD. 2014-06-19 13:56:03 which is just tickets :) 2014-06-19 13:56:14 no commercial support afaik 2014-06-19 13:56:28 :) 2014-06-19 13:57:00 but I wouldnt mind if anyone started sell commercial support 2014-06-19 13:57:29 would be awesome if someone could help making alpine better and gets paid to do it :) 2014-06-19 13:57:47 thats the questions I will be asked if I try to force alpine in my company 2014-06-19 13:57:51 so thats why I asked, about the support 2014-06-19 13:57:54 but i believe all us core developers at this time are unable to do it at this time 2014-06-19 13:58:06 yes 2014-06-19 13:59:51 is the backtrace issue with gdb / issue with valgrind described somewhere ? 2014-06-19 14:01:51 do you guys strip libtool archives from packages? 2014-06-19 14:02:01 I have hard times linking subversion with sqlite on alpine, staticly 2014-06-19 14:05:27 yes we strip .la files 2014-06-19 14:07:03 http://git.alpinelinux.org/cgit/aports/tree/main/subversion/APKBUILD#n65 2014-06-19 14:08:43 :< 2014-06-19 14:09:00 I need to build gazillion of tools, staticlly, and musl or uclibc is the only way 2014-06-19 14:09:07 but that's not big issue anyway. 2014-06-19 14:09:44 I can treat alpine like aboriginal linux. 2014-06-19 14:16:56 i suppose you could git fork our aports tree and remove the rm *.la lines 2014-06-19 14:17:17 you could probably also add the LDFLAGS you want in /etc/abuild.conf 2014-06-19 14:17:24 for -static 2014-06-19 14:17:29 but i havent tried that really 2014-06-19 14:19:33 I do build subversion on a side, but it try to link with db and sqlite3 and if there's no libtool archives linker can't proceed with -static 2014-06-19 14:19:41 but I can handle it whatsoever. 2014-06-19 15:19:08 what use are three different digest? 2014-06-19 15:19:12 digests 2014-06-19 15:21:34 wow apk update is fast when you're used to "apt-get update" ... 2014-06-19 15:22:08 one thing i've been pressing debian for for years now, which might be an idea for alpine if the package index grows tho... 2014-06-19 15:22:28 updating it via rsync rather than full download or diffs 2014-06-19 15:23:08 ncopa, i've got a couple more questions if you're around :) 2014-06-19 15:23:43 when i installed, i just manually started wpa_supplicant dhcpcd, etc. rather than setting them up thru the installer 2014-06-19 15:23:53 and now they don't auto-start on startup 2014-06-19 15:24:06 what's the right way to add things to startup? 2014-06-19 15:24:06 dalias: I use Alpine on my docker images 2014-06-19 15:42:35 another question: how do i (without editing scripts that should be package-provided) get busybox ntpd not to block-then-fail when the network is not yet up? 2014-06-19 15:42:49 what i want is for it to go in the background and start working as soon as there is a network link 2014-06-19 16:03:39 ncopa, i found out why wpa_supplicant wasn't doing anything until i manually run "iwlist scan" 2014-06-19 16:04:28 the default wpa_supplicant.conf contains freq_list=5180 as an "example config" that's not commented out (i.e. only use channel 36) 2014-06-19 17:32:36 I'm having a bit of trouble with the alpine initramfs. It's not automatically loading the ext4 module before trying to mount the root filesystem, so it fails. 2014-06-19 17:33:23 I'm skimming through the init shell script in the initramfs, I came across "myopts" which seems to have a "modules" argument... is that the best way to go about making sure ext4 is loaded? 2014-06-19 17:35:04 Wait, it reads /etc/modules 2014-06-19 17:35:46 I mean, of course it does, god damn it. 2014-06-19 18:20:36 k, filed the bug report. 2014-06-20 06:38:57 good morning 2014-06-20 06:39:22 moin 2014-06-20 09:17:18 what's the right way to add things to startup? 2014-06-20 09:17:23 rc-udpate add 2014-06-20 09:17:28 rc-update* 2014-06-20 09:27:44 rc-udpate: no such command 2014-06-20 09:28:02 rc-update 2014-06-20 09:28:04 rc-update probably 2014-06-20 09:28:04 ncopa: y u lie to me! 2014-06-20 09:28:08 i did a typo 2014-06-20 09:28:09 sorry 2014-06-20 09:28:13 you lied! 2014-06-20 09:28:24 rc-update* 2014-06-20 09:28:34 i tried fix it... sorry 2014-06-20 09:28:55 ACTION will settle for a boop 2014-06-20 09:43:49 right, setup-xorg-base xf86-video-fbdev worked, startx after it threw up an xsession for me, a restart has broke that, with xorg failing on failing to open /dev/fb0 2014-06-20 09:44:09 thus no screens are found 2014-06-20 09:46:46 modprobe fbdev? 2014-06-20 09:47:05 maybe: echo 'fbdev' >> /etc/modules 2014-06-20 09:49:57 nope 2014-06-20 09:50:36 is there a way to reset the xorg base so that setup-xorg-base works again, right now rc-service is just failing to work 2014-06-20 09:52:13 Do you have a /dev/fb0 ? 2014-06-20 09:55:11 "Failed to load /usr/lib/xorg/modules/drivers/fbdev_drv.so: Error relocating /usr/lib/xorg/modules/drivers/fbdev_drv.so: fbdevHWProbe: symbol not found" 2014-06-20 09:55:33 getting this after a second reboot 2014-06-20 09:56:38 huh 2014-06-20 09:57:30 xorg configure is just totally failing 2014-06-20 09:57:34 vesa fails too 2014-06-20 09:57:46 same issue except this time it's "VBEExtendedInit" that it cannot relocate 2014-06-20 09:58:22 i just pushed new kernel, did you apk upgrade? 2014-06-20 09:58:37 could it be that your installed kernel is different than running kernel? 2014-06-20 09:58:59 http://sprunge.us/ZaSY 2014-06-20 09:59:24 this is after a reboot 2014-06-20 09:59:43 is there something I must do to make it boot into new kernels 2014-06-20 09:59:52 should not 2014-06-20 10:00:18 apk version linux-grsec 2014-06-20 10:00:32 3.14.6-r0 2014-06-20 10:01:02 Looks like xorg is miscompiled 2014-06-20 10:01:28 some of the drivers depends on lazy loading which is not supported in musl 2014-06-20 10:01:32 vesa, fbdev, modesetting all fail to resolve symbols for some reason 2014-06-20 10:01:51 those are the only sensible three drivers one would use with xorg 2014-06-20 10:01:53 Ah, that makes sense them 2014-06-20 10:02:31 Wonder why the dlopen() flags that provide that are in place, and why it doesn't error out 2014-06-20 10:02:34 I like how it sometimes works 2014-06-20 10:02:43 just rebooted again and x starts up 2014-06-20 10:02:48 kill it, try and start it again and it fails 2014-06-20 10:02:51 !!? 2014-06-20 10:03:14 Probably depends on the order the libs are loaded 2014-06-20 10:04:33 yes 2014-06-20 10:04:55 we do a hack in /etc/X11/xorg.conf.d 2014-06-20 10:05:00 http://sprunge.us/XBhO 2014-06-20 10:05:31 [ 17.722] (EE) Failed to load /usr/lib/xorg/modules/input/evdev_drv.so: Error loading shared library /usr/lib/xorg/modules/input/evdev_drv.so: Exec format error 2014-06-20 10:05:51 not very stable :P 2014-06-20 10:06:06 Exec format error 2014-06-20 10:06:25 apk audit --system 2014-06-20 10:06:32 That means the kernel doesn't recognise the file format. Arch differences maybe 2014-06-20 10:06:47 or disk corruption or something 2014-06-20 10:06:50 what does the audit do exactly 2014-06-20 10:06:55 I see a bunch of U and Xs 2014-06-20 10:07:01 verifies the checksum 2014-06-20 10:07:05 of index 2014-06-20 10:07:32 2014-06-20 10:08:35 ld says that evdev_drv.so is "file not recognized" 2014-06-20 10:08:40 "File truncated" 2014-06-20 10:09:07 oh wow, it's 0 bytes in size 2014-06-20 10:09:09 wtf 2014-06-20 11:03:29 apk fix xf86-input-evdev 2014-06-20 12:05:42 there should be an apk fix install 2014-06-20 12:17:14 fix it all? 2014-06-20 12:17:42 apk info | xargs apk fix 2014-06-20 13:36:02 why apk instead of rpm / deb / whatever format ? 2014-06-20 13:42:36 igmar: originally we had tar.gz, no dependency handling 2014-06-20 13:42:55 i thought we could do better and added dependency handling into the shell script 2014-06-20 13:43:02 fabled though we can do better 2014-06-20 13:43:06 we looked a pacman 2014-06-20 13:43:08 deb 2014-06-20 13:43:10 rpm 2014-06-20 13:43:12 and others 2014-06-20 13:43:21 but since we were running from tmpfs 2014-06-20 13:43:33 and basically installing the system from scratch on every reboot 2014-06-20 13:43:45 we wanted read from disc once, write once 2014-06-20 13:44:03 we wanted it even for networking stuff 2014-06-20 13:44:27 and we wanted caclulating checksums on the fly, while waiting for io 2014-06-20 13:45:07 so we ended up with fabled writing apk-tools 2014-06-20 13:45:21 Makes sense indeed. 2014-06-20 13:45:55 oh, we also wanted the db format be possible to inspect (and even edit!) with a text editor 2014-06-20 13:46:05 and we wanted be able to extract files with tar 2014-06-20 13:46:20 so you can do tar -zxf package-1.0.apk 2014-06-20 13:56:53 Hi to all, I am reading in mailing list that icewm package has been submited... 2014-06-20 13:57:12 ... but I can not find it in /testing 2014-06-20 13:59:12 me neither 2014-06-20 13:59:13 i'm quite happy to have a package manager that runs instantly rather than thrashing the disk for 10 seconds (dpkg/apt) 2014-06-20 13:59:55 :) 2014-06-20 13:59:58 of course i have no idea how apk scales. maybe it would be just as bad with the number of packages debian has 2014-06-20 14:00:24 Amatcoder: not sure what happend 2014-06-20 14:00:33 im building it local to see what happens 2014-06-20 14:00:42 fails to build 2014-06-20 14:01:22 clandmeter: thanks for looking 2014-06-20 14:01:54 dalias, it scales more or less O(n). if it ever becomes an issue, i can take a look at it. 2014-06-20 14:02:20 my 5yr old laptop has 3GB in 1223 installed packages, and it's still lightning fast. 2014-06-20 14:02:31 Amatcoder: did you submit it? 2014-06-20 14:02:38 bah 2014-06-20 14:02:40 Nope 2014-06-20 14:02:47 http://bld2.alpinelinux.org/buildlogs/build-edge-x86_64/testing/icewm/icewm-1.3.8-r1.log 2014-06-20 14:02:50 i know whats wrong 2014-06-20 14:03:00 they use #!/bin/bash in some build script 2014-06-20 14:03:02 bash is mising 2014-06-20 14:03:11 and i happened to have bahs installed when i tested it 2014-06-20 14:05:12 fabled, O(n) in installed packages or total packages? 2014-06-20 14:05:21 depends on the action 2014-06-20 14:05:21 i think debian's problem is thast it's O(n) in total packages 2014-06-20 14:05:28 for all operations :) 2014-06-20 14:06:32 i think also debian does: fetch to disk ( read from net, write data to disk), verify (read from disk), unpack (read once again from disk) 2014-06-20 14:06:44 it loads the installed and available databases in memory, so it's O(n) to load it. solving dependencies, installation etc are all one time hash look ups and fast. 2014-06-20 14:07:09 so yes, there's some affect with number of total packages, but it affects mostly RAM usage; not overall performance 2014-06-20 14:08:09 what makes it additional fast is that the index (available / installed) is single flat file 2014-06-20 14:08:23 not a directory tree with gazillion files 2014-06-20 14:08:33 like pacman 2014-06-20 14:08:52 that was the deal breaker for pacman, gazillion files 2014-06-20 14:08:58 *nod* 2014-06-20 14:08:59 it's slow to open/read/close myriad files, and it also wastes lot of memory on tmpfs 2014-06-20 14:09:19 i think debian also has way too much data in their flat file 2014-06-20 14:09:25 huge details on every package 2014-06-20 14:09:34 rather than just the name/version/deps (minimal stuff) 2014-06-20 14:09:40 right. yes, that we tried to keep minimal too. 2014-06-20 14:09:41 we intentionally tried to keep the data stored minimal 2014-06-20 14:09:48 and the field names are single letters, not multiple words 2014-06-20 14:10:14 and the db is relatively easy to parse with awk ;) 2014-06-20 14:15:15 ginjachris: I'm trying to reproduce your raid issue 2014-06-20 14:16:01 indeed... 2014-06-20 14:17:06 seems like problem is that mdadm utility is not inclued in initramfs 2014-06-20 14:21:58 @ncopa: thanks, I've been scratching my head over that one, I feel like it's something I should have been able to find the fix for......but alas I did not succeed 2014-06-20 14:22:18 it is definitively a bug 2014-06-20 14:22:47 i have found atleast one thing that is definitively broken 2014-06-20 14:23:01 I can replicate at will on VM so if you need me to do anything (like collect x, y, z outputs) just let me know 2014-06-20 14:23:09 i can replicate too 2014-06-20 14:23:16 so no problem 2014-06-20 14:23:22 ok cool :) 2014-06-20 14:24:35 also, i think the way we currently do the raid is stupid 2014-06-20 14:25:00 we should raid the sda sdb and create partitions on md0 2014-06-20 14:25:09 atleast for raid1 2014-06-20 14:25:40 but that wont work for raid5... 2014-06-20 14:26:38 I see....currently the raid is built from sda1, sdb1 -->md0 sda2, sdb2--->md1 etc. I have no problem with it that way, especially if it adds complications for Raid 5 2014-06-20 14:26:47 i always think simple is best :) 2014-06-20 14:27:03 its nice to have it similar raid5 yes 2014-06-20 14:28:40 ok i think i found it 2014-06-20 14:29:32 and i think i have done a mistake kindof 2014-06-20 14:30:05 mkinitfs has features (addons) 2014-06-20 14:30:17 so you can enable/disable things you use in mkinitfs 2014-06-20 14:30:39 there is a 'feature' called 'raid' 2014-06-20 14:31:07 the problem is that it covers both hardware raid (eg cciss kernel modules) 2014-06-20 14:31:16 and soft raid, eg mdadm 2014-06-20 14:31:42 that means that you dont get the mdadm utility just because you enable the 'raid' feature 2014-06-20 14:32:05 so you manually have to add /sbin/mdadm to your config 2014-06-20 14:32:10 setup-disk does this for you though 2014-06-20 14:32:26 however, we moved the location of the config files recently 2014-06-20 14:32:38 and didnt update the setup-disk script 2014-06-20 14:33:12 i think the real solution is to separate the mdadm and hwraid feature 2014-06-20 14:33:47 Amatcoder: should be build now. 2014-06-20 14:35:18 clandmeter: Yes, I'm testing right now, thanks 2014-06-20 14:36:05 hm... difficult to separate those out without breaking compat with 'raid' feature 2014-06-20 14:36:20 i suppose we could add 2 new 'features', hwraid and mdadm 2014-06-20 14:36:45 alternatively, simply add /sbin/mdadm to 'raid' 2014-06-20 14:40:41 Cool, icewm is working fine: http://s27.postimg.org/n3fzj09pv/screenshot_icewm.png 2014-06-20 14:40:52 nice! 2014-06-20 14:41:22 it looks pretty nice too 2014-06-20 14:41:46 Thanks to clandmeter for fixing and Isaac Dunham for aport ! 2014-06-20 14:42:41 are those alps on the background? ;-) 2014-06-20 14:43:45 @ncopa: You Sir, are a genius! 2014-06-20 14:44:12 ginjachris: i'm the idiot who caused the bug in the first place.... :-D 2014-06-20 14:45:00 a genious is the one who fixes nfs on x86_64 :) 2014-06-20 14:45:05 clandmeter: Not sure if Alps or Himalaya ;-) 2014-06-20 14:45:10 indeed :) 2014-06-20 14:45:25 1. create buggy software 2014-06-20 14:45:32 2. make people trip on bugs 2014-06-20 14:45:37 3. fix the bug 2014-06-20 14:45:41 no no, far from an idiot! Just temporarily overcome by the complexity of some changes and the move to musl 2014-06-20 14:45:43 ... 2014-06-20 14:45:48 $$$ profit!!! 2014-06-20 14:46:19 hahaha, you sound like a true security vendor! 2014-06-20 14:46:26 well when i look at the code now, years later, i think "what idiot wrote this?" :) 2014-06-20 14:46:34 "...oh it was me..." 2014-06-20 14:46:41 git log :p 2014-06-20 14:46:46 git blame 2014-06-20 14:46:58 git punch 2014-06-20 14:50:51 ncopa: "alternatively, simply add /sbin/mdadm to 'raid'" - this does seem simple, I'm all for simple. If this is easy and doesn't cause any problems elsewhere, go for it! 2014-06-20 14:53:48 looking at things.. 2014-06-20 14:54:02 hm 2014-06-20 14:54:46 the only problem i can see with it is that hwraid setups will get it too 2014-06-20 14:54:58 and if it is found at boot, it will be executed 2014-06-20 14:55:04 so it will give error on thos boxes 2014-06-20 14:55:10 we dont want that i think 2014-06-20 14:55:40 at least not in the default config 2014-06-20 14:55:48 but i think for setup-disk we can just do it 2014-06-20 14:56:18 i think i do that for now 2014-06-20 14:58:28 yeah that makes sense, hwraid shouldn't need to have multiple disks passed to setup-disk. When I've used it, hwraid just makes everything appear as sda 2014-06-20 14:58:52 and that had 4 500GB disks 2014-06-20 14:59:01 so I think that's a good solution 2014-06-20 14:59:18 raid10 would be nicer :) 2014-06-20 15:00:13 ok time is up 2014-06-20 15:00:18 i'll fix it on monday 2014-06-20 15:00:25 and hopefully we have new release tuesday 2014-06-20 15:49:21 thanks @ncopa. I've got to go, have a good weekend all 2014-06-20 21:35:57 the .iso is not a hybrid type? i.e. it can't be imaged to a USB stick? Is that right? 2014-06-20 21:43:20 I'm not sure what exact type it is, but it's trivial to get it over to an usb and boot it that way 2014-06-20 21:43:29 unetbootin did the trick just fine for me 2014-06-20 21:45:19 Hi! Is the syntax of the /etc/network/interfaces file the same as other distros like debian? 2014-06-20 21:46:14 The specific case is that I want to create a bonded nic on active-passive mode... 2014-06-20 21:46:26 I just want to make sure that it works... 2014-06-20 22:04:53 Yeah! it does. 2014-06-20 22:05:01 Sweet! 2014-06-20 23:54:05 ncopa, so the issue seems to be disk corruption 2014-06-20 23:54:15 it just keeps fucking up the disk on reboots 2014-06-21 03:21:46 hmm 2014-06-21 03:21:51 pergatorio:~# apk add iwlwifi-5000-ucode 2014-06-21 03:21:51 (1/1) Installing iwlwifi-5000-ucode (8.24.2.12-r1) 2014-06-21 03:21:51 ERROR: Failed to extract lib/firmware/iwlwifi-5000-2.ucode: Read-only file system 2014-06-21 03:21:51 wget: short write 2014-06-21 03:21:51 ERROR: iwlwifi-5000-ucode-8.24.2.12-r1: Read-only file system 2014-06-21 03:21:51 1 errors; 30 MiB in 73 packages 2014-06-21 03:21:54 any ideas? 2014-06-21 03:23:01 check your disk for errors 2014-06-21 03:23:15 shouldn't go read only without reason 2014-06-21 03:23:28 failing any real issues/errors mount -o remount,rw / 2014-06-21 03:24:28 i had that a few times if the barrier test failed 2014-06-21 03:25:11 I think because it's the /lib directory 2014-06-21 07:53:42 hmm, my mouse is inverted 2014-06-21 08:25:52 any suggestions how to benchmark my alpine linux machine? Just curious to compare to Archlinux. Would musl bring performance improvements? 2014-06-21 15:23:06 hendry: what kind of workload? 2014-06-21 18:37:23 hi 2014-06-21 18:37:51 is alpine ready to boot from usb formated with ntfs ? 2014-06-21 18:38:26 nevermind I see in the wiki 2014-06-21 18:39:37 brb 2014-06-21 18:43:33 it dosen't work it keeps rebooting, ntfs module is in grsec modloop but not in initrd file - grsec.gz does it need to be there aswell? 2014-06-21 18:51:58 brb 2014-06-21 21:30:44 hey guys 2014-06-21 21:30:57 does the mini disk allow sys-type install? 2014-06-21 21:31:22 I tried and it's telling me I'm missing some packages (sfdisk, e2fsprogs, lvm2, syslinux) 2014-06-22 18:27:30 ncopa: virtio now booting fine 2014-06-22 18:27:34 with latest patch 2014-06-22 23:10:30 ncopa, CONFIG_PROC_PAGE_MONITOR should be enabled in the kernel 2014-06-22 23:10:41 it's key to measuring memory usage 2014-06-22 23:10:59 disabling that is what killed /proc/$pid/smaps 2014-06-22 23:11:40 btw busybox top aborts when you press 's' to switch to memory-usage view if this option is disabled 2014-06-23 06:15:41 morning 2014-06-23 06:15:43 knitting_g 2014-06-23 06:16:06 knitting_legend: yes alpine-mini supports 'sys' disk install, but you need network 2014-06-23 06:17:11 kocka: seems like they went for 3.15 kernel now :-/ 2014-06-23 06:17:40 mornin 2014-06-23 06:45:53 ncopa: i think they'll keep 3.14 as stable 2 :P 2014-06-23 06:46:02 moin 2014-06-23 06:47:55 it woudl be nice 2014-06-23 06:53:46 at least i think spender meant that on twitter 2014-06-23 06:59:51 https://twitter.com/grsecurity/status/480750611336814592 2014-06-23 07:00:01 https://twitter.com/grsecurity/status/480884290654597120 2014-06-23 07:10:54 nice 2014-06-23 07:51:29 dalias: i suppose CONFIG_PROC_PAGE_MONITOR is for vanilla kernel 2014-06-23 14:40:46 ncopa, it seems to be off for both, but grsec might force it off anyway 2014-06-23 14:41:45 yes, not possible to enable it on grsec unless i disable the grsec/pax protections 2014-06-23 14:42:13 i suppose i should ask the grsec ppl what the proper(tm) way is to calculate the memory usage 2014-06-23 14:42:30 dalias: i pushed the fix to edge/linux-vanilla 2014-06-23 14:42:57 it would be nice if blocking these things was a runtime option on grsec 2014-06-23 14:43:05 even if only root could see them it would be an improvement 2014-06-23 14:43:32 making alpine linux get out of the way for dalias' work on musl is high prio ;) 2014-06-23 14:43:43 :) 2014-06-23 14:46:02 it's not really "in the way" because i do most of my development on my headless server not laptops 2014-06-23 14:46:35 but having alpine is an opportunity to do some development tasks that were difficult before 2014-06-23 14:47:16 because it's got a moderately large base of tool and library dependencies i otherwise wouldn't bother to build, but that are needed for testing software and trying to reproduce issues ppl report 2014-06-23 14:53:33 i was hoping you would eat your own dogfood on your desktop too at some point ;) 2014-06-23 14:55:20 i'd just been waiting for a distro that wouldn't make me spend all my time waiting for things to compile :) 2014-06-23 14:55:56 my development system has been dogfood since before the initial release 2014-06-23 14:56:33 heh... i used gentoo earlier so i know what waiting for compile means... 2014-06-23 14:56:59 you used it on your desktop too? 2014-06-23 14:57:10 i need to reboot 2014-06-23 14:57:12 brb 2014-06-23 15:04:29 gotta go now, have a nice evening 2014-06-23 15:04:41 hopefullly 3.0.1 comes out tomorrow 2014-06-23 15:05:05 o/ 2014-06-24 02:04:18 ncopa, one more issue to report: CONFIG_FEATURE_SEAMLESS_XZ is off in busybox 2014-06-24 02:04:21 this makes tar a pain to use 2014-06-24 02:04:54 and i don't see an option for explicit unxz when running tar 2014-06-24 02:05:52 one other issue 2014-06-24 02:06:08 dropbear and openssh-client conflict (both provide scp) 2014-06-24 02:06:21 but i want dropbear for server and openssh for client (since dbclient sucks :-p) 2014-06-24 03:12:18 *sigh* no luck compiling emacs on alpine yet 2014-06-24 06:17:10 morning 2014-06-24 06:17:19 i wonder how to solve dropbear scp 2014-06-24 06:17:25 its a symlink only 2014-06-24 06:17:40 i dont think its worth a subpackage for it 2014-06-24 06:17:47 hm 2014-06-24 06:17:50 maybe it is 2014-06-24 06:28:46 moin 2014-06-24 06:30:32 ncopa: http://git.gnupg.org/cgi-bin/gitweb.cgi?p=gnupg.git;a=commitdiff;h=014b2103fcb1 2014-06-24 06:31:54 1999 :) 2014-06-24 06:32:14 dalias: i pushed updated busybox and dropbear 2014-06-24 08:18:37 ncopa: http://www.openwall.com/lists/oss-security/2014/06/24/8 2014-06-24 08:18:45 2.0.24 arrives later today 2014-06-24 08:19:04 (gnupg) 2014-06-24 08:22:43 we dont do gpupg2 for musl 2014-06-24 08:22:55 we want gnupg 2.1 i think 2014-06-24 08:23:12 oh 2014-06-24 08:23:15 due to libpth crazyness 2014-06-24 08:23:28 we use gnupg1 for musl 2014-06-24 08:24:02 libpth is really ... hmm ... dunno how to say it diplomatically correct 2014-06-24 08:24:06 "interesting" 2014-06-24 08:24:38 no. it's not "interesting". it more like... "utter stupidity" 2014-06-24 08:25:09 it solved a valid problem by creating a bigger problem. 2014-06-24 08:25:38 i suppose its not posssible to express libpth diplomatically correct :) 2014-06-24 08:25:59 "amazingly stupid" was on my mind 2014-06-24 08:26:18 as i understand, the only thing it is supposed to provide is portability 2014-06-24 08:27:03 but we cannot use it on musl due to it not portable... 2014-06-24 08:27:44 :/ 2014-06-24 08:36:51 gnpug2 has been dissapointing due to pth 2014-06-24 11:35:44 Hi Guys, is it foreseeable when the Alpine 3.0 Xen Version will be released? 2014-06-24 11:36:20 i was hoping today 2014-06-24 11:36:28 there is a known issue with hvmloader though 2014-06-24 11:36:35 which we havent yet fixed 2014-06-24 11:36:46 but it should work with pvs 2014-06-24 11:37:12 Is the Issue documented somewhere so i can see if i would be affected? 2014-06-24 11:37:56 kaniini: do we have the musl tolchain+hvmloader documented some place? maybe file a bug with the details? 2014-06-24 11:38:20 tze: if you use hvmloader you are affected (eg windows guests) 2014-06-24 11:38:56 ncopa: That's a shame, i am currently using a FreeBSD HVM as a Storage Domain ... 2014-06-24 11:39:04 i was thinking to ship a broken 3.0.1 release anyways, since its easier to test it that way 2014-06-24 11:39:15 hm 2014-06-24 11:39:22 might be i could test it anyways actually 2014-06-24 11:39:40 we have some ideas what the problem might be 2014-06-24 11:39:49 but I have not had the time to follow it up really 2014-06-24 11:39:50 sorry 2014-06-24 11:41:33 ncopa: No Problem really, i have time, am just eager to test Alpine as Dom0 2014-06-24 11:43:01 ncopa: I'll be happy to test a RC as soon as the Issue is found 2014-06-24 11:45:30 we more or less think know what the issue is 2014-06-24 11:45:49 when building hvmloader it pulls in the system libc headers for some reason 2014-06-24 11:45:53 which is 64bit 2014-06-24 11:46:10 so it mixes in 64bit headers in the 32 bit build 2014-06-24 11:46:21 that is what we think happens 2014-06-24 12:00:56 Thanks for the Info! 2014-06-24 15:11:41 hi 2014-06-24 15:11:49 ncopa, i've debugged the emacs issues 2014-06-24 15:11:58 unfortunately i don't see any easy fix 2014-06-24 15:12:12 anyway I can see why on boot it gets stuck at loading drivers 2014-06-24 15:12:19 the basic issue is how emacs unexec dumper works 2014-06-24 15:12:57 because loading all the lisp at startup is insanely slow, they load it once then dump a new ELF file with the loaded state 2014-06-24 15:14:17 however their code to generate the new ELF file sucks and can't see memory that was allocated via mmap rather than brk 2014-06-24 15:14:50 they have a hack for glibc (and their own gnu malloc) to disable use of mmap for allocating lisp objects, but obviously this doesn't work with musl 2014-06-24 15:39:09 it gets stuck at boot on "rc_add syslog boot" 2014-06-24 15:39:15 booting from usb thumb drive 2014-06-24 15:40:35 save 2014-06-24 15:40:48 sorry about that 2014-06-24 16:24:50 are packages for x64 2.7 alpine compatible with 3.0 x64 ver ? 2014-06-24 16:25:06 one uses ulibc other musl correct? 2014-06-24 16:26:58 i don't think they'd be compatible directly, but it might be possible to install the uclibc libs in an alternate location such that the binaries work 2014-06-24 16:33:17 zxd: No 2014-06-24 17:21:45 Is there any plan to integrate openzfs into Alpine? 2014-06-24 17:40:38 ncopa, i have a fix for emacs but it's hideous (it makes the binary 400 megs :) 2014-06-24 17:48:39 What's the apk equivalent of rpm -ql ? 2014-06-24 17:51:05 -L 2014-06-24 17:51:08 never mind 2014-06-24 17:52:40 That doesn't work 2014-06-24 17:55:35 anybody had their alpine 3.0.0 reboot randomly? 2014-06-24 18:36:32 Can I send github pull requests ? 2014-06-24 18:37:02 forget that one 2014-06-24 18:40:39 blakes, i haven't heard of that sort of problem in the channel, but stick around and see if someone else knows 2014-06-24 19:25:41 dalias: I will try to stick around, but random unknown reboots are kind of thwarting that :-P (alpine is my router) 2014-06-24 19:25:49 :( 2014-06-24 21:15:46 ncopa, i've gotten interest from the emacs side in a permanent solution to the dumper problem :) 2014-06-24 21:16:14 see http://lists.gnu.org/archive/html/emacs-devel/2014-06/msg00452.html 2014-06-24 21:17:21 here's a thread on the same topic from earlier this year, but with no proposed solution except making emacs statup 150 times slower: 2014-06-24 21:17:24 http://lists.gnu.org/archive/html/emacs-devel/2014-01/msg01015.html 2014-06-24 21:30:09 i suspect it'll take a while to make progress on this 2014-06-24 21:30:39 but in the mean time i have a hackish patch to make emacs work on musl 2014-06-24 22:18:05 how to make wlan interface names permanent across reboots 2014-06-24 22:31:03 zxd: what's changing? 2014-06-24 22:31:36 you can get any file to persist with lbu add and lbu commit 2014-06-25 01:15:28 does anyone know what the nut package is called? 2014-06-25 01:15:30 (it's not nut) 2014-06-25 06:08:38 blakes: what arch and what kernel version? 2014-06-25 06:08:50 uname -r 2014-06-25 06:09:45 dalias: wow 2014-06-25 06:12:57 they even have special glibc hacks to make that work in emacs... 2014-06-25 06:13:16 It still requires substantial ugly hacks, e.g. there is code in glibc to 2014-06-25 06:13:16 serialize and deserialize malloc state whose sole purpose is to support 2014-06-25 06:13:16 Emacs dumping, and which cannot be changed since that would force Emacs 2014-06-25 06:13:16 to be redumped when glibc was upgraded. It seems plausible that this 2014-06-25 06:13:16 might eventually retard glibc allocator development :( 2014-06-25 06:13:30 thats just.. wow... 2014-06-25 06:13:50 :) 2014-06-25 06:14:01 i've never seen that code and i wonder if it really exists... 2014-06-25 06:14:09 but yes it's rather insane 2014-06-25 06:14:33 just get over emacs! 2014-06-25 06:17:36 :) 2014-06-25 06:17:45 and use...? 2014-06-25 06:17:50 vim? 2014-06-25 06:17:56 :-D 2014-06-25 06:17:58 >_< 2014-06-25 06:18:02 i use vim... 2014-06-25 06:18:13 jmacs is at least usable 2014-06-25 06:18:21 vim is nice 2014-06-25 06:18:26 the funny thing is 2014-06-25 06:18:31 i used to hate vim 2014-06-25 06:18:38 and i used emacs for a while 2014-06-25 06:18:45 (many years ago) 2014-06-25 06:18:54 then you really started to love vim? 2014-06-25 06:19:05 but the default gnu coding style made me leave emacs 2014-06-25 06:19:19 by default the .c files use the gnu coding style 2014-06-25 06:19:23 i can't stand modality and the illogical way vim handles the cursor position (and how it's inconsistent between insert mode and outside insert mode) 2014-06-25 06:19:42 dalias: it's confusing until you get used to it 2014-06-25 06:19:58 a way to think about it is like this 2014-06-25 06:20:04 in the same way as numbering arrays from 1 would be: i.e. nobody sane should do it :) 2014-06-25 06:20:12 you only need to be in insert/append mode when you are editing text 2014-06-25 06:20:16 short bursts 2014-06-25 06:20:18 and see lua 2014-06-25 06:20:20 :P 2014-06-25 06:20:21 its still kinda confusing... :) 2014-06-25 06:20:33 and yes Lua uses array from 1 2014-06-25 06:20:36 given 2014-06-25 06:20:37 (and i hate that too...) 2014-06-25 06:21:01 ncopa: I've recently learned to appreciate busybox vi 2014-06-25 06:21:11 it's a very decent implementation 2014-06-25 06:21:14 lua is for people who think there are 3 hours between 2 oclock and 4 oclock :) 2014-06-25 06:21:28 dalias: there can be with time zones! 2014-06-25 06:21:30 yes bb vi is okish 2014-06-25 06:21:32 2,3,4 2014-06-25 06:21:59 and you might have daylight saving changes around that time too... 2014-06-25 06:23:11 $ apk search --exact -v mg 2014-06-25 06:23:11 mg-20140414-r0 - mg is Micro GNU/emacs, this is a portable version of the mg maintained by the OpenBSD team. 2014-06-25 06:23:17 ncopa: you got a chance to screw with docker yet? 2014-06-25 06:23:18 i have never used it 2014-06-25 06:23:25 Xe: nope 2014-06-25 06:23:30 i think uggedal has 2014-06-25 06:23:34 ncopa: it makes lxc easy 2014-06-25 06:23:37 ah 2014-06-25 06:23:43 ok? 2014-06-25 06:23:49 I took his alpine images and consolidated them into one repo 2014-06-25 06:24:10 making lxc easy might be worth a look 2014-06-25 06:24:23 https://registry.hub.docker.com/u/xena/alpine/ 2014-06-25 06:24:35 it's a golang program though 2014-06-25 06:24:53 so unless you have golang in aports it's going to be interesting to port over 2014-06-25 06:25:08 from what I have heard, Go is doing many things "right" 2014-06-25 06:25:21 i think uggedal ported and maintains Go 2014-06-25 06:25:22 It's interesting 2014-06-25 06:25:45 It's pretty much the plan 9 people remaking C to be cloud scale 2014-06-25 06:26:12 as such there are some plan9-isms in there 2014-06-25 06:26:24 such as dial being the name of the socket connection function 2014-06-25 06:27:10 ncopa, mg is cool in principle but last i checked it has ZILCH for unicode support 2014-06-25 06:27:27 i looked at fixing it but it seemed hard 2014-06-25 06:27:30 dalias: doesn't alpine proper have sketchy unicode support? 2014-06-25 06:27:38 xe, no 2014-06-25 06:27:41 or am I thinking 2.x? 2014-06-25 06:27:46 2.x most likely 2014-06-25 06:27:49 ah 2014-06-25 06:27:51 nm then 2014-06-25 06:27:57 under musl everything is always utf-8 2014-06-25 06:28:05 *shrug* i never tested mg. i just saw that someone sent a patch for a mini emacs... 2014-06-25 06:28:16 i'd love mg if it worked 2014-06-25 06:28:26 but i use emace e.g. as my editor for mutt 2014-06-25 06:28:27 ncopa: what sold me for docker was the ability to create an anonymous ubuntu vm to play with in a single command 2014-06-25 06:28:35 and lack of unicode for email is a show-stopper 2014-06-25 06:28:51 function anonubuntu { 2014-06-25 06:28:52 true 2014-06-25 06:28:54 docker run --rm -it ubuntu:trusty /bin/bash 2014-06-25 06:29:48 ncopa: another interesting thing about go is the fact that all compiled golang programs are static binaries 2014-06-25 06:30:30 dalias: seems like mg-20140414 still not support utf8 2014-06-25 06:31:36 ncopa, another busybox bug report :) 2014-06-25 06:31:57 CONFIG_LAST_SUPPORTED_WCHAR should be INT_MAX 2014-06-25 06:32:04 or rather 0x10ffff 2014-06-25 06:32:07 rather than 767 2014-06-25 06:32:18 my ash on alpine is showing everything as ?'s 2014-06-25 06:32:23 because of that 2014-06-25 06:32:27 hm i thought i fixed that? 2014-06-25 06:32:34 oh maybe i just need to update :) 2014-06-25 06:32:42 no, long time ago 2014-06-25 06:33:45 seems broken still 2014-06-25 06:33:49 yes 2014-06-25 06:33:54 i must be dreaming then 2014-06-25 06:34:04 checking git log... 2014-06-25 06:34:55 tze: there is a patch for zfs 2014-06-25 06:35:05 ok 2014-06-25 06:35:14 i set it to 0x10ffff then? 2014-06-25 06:36:48 yeah 2014-06-25 06:37:12 also.. every time i upgrade kernels, /boot/extlinux.conf gets overwritten with a version that only has grsec in it 2014-06-25 06:37:27 this happens when i install/upgrade vanilla 2014-06-25 06:37:29 any idea why? 2014-06-25 06:37:36 edit default in /etc/updated-extlinux 2014-06-25 06:37:39 edit default in /etc/updated-extlinux.conf 2014-06-25 06:37:56 # default 2014-06-25 06:37:56 # default kernel to boot 2014-06-25 06:37:57 default=grsec 2014-06-25 06:38:45 what should it be set to for vanilla? 2014-06-25 06:39:06 and is there a way to make it put both in the boot menu still? 2014-06-25 06:39:20 or if i want that should i just turn off overwriting and edit it myself? 2014-06-25 06:39:29 it should put both there 2014-06-25 06:39:35 and it shoudl give you a menu at boot 2014-06-25 06:39:35 it doesn't 2014-06-25 06:39:43 it only puts grsec 2014-06-25 06:39:54 if i edit /boot/extlinux.conf i get both 2014-06-25 06:40:20 the /boot/extlinux.conf is generated 2014-06-25 06:40:31 from a trigger in syslinux package 2014-06-25 06:40:49 so whenever any package puts anything in /boot/ the trigger is executed 2014-06-25 06:41:01 but hum 2014-06-25 06:41:02 indeed 2014-06-25 06:41:12 script doesnt find vanilla 2014-06-25 06:41:16 i saw that the other day 2014-06-25 06:41:17 ok 2014-06-25 06:41:18 its a bug 2014-06-25 06:41:24 i'll fix it 2014-06-25 06:41:28 i think it's that the kernel name is just vmlinuz 2014-06-25 06:41:31 yes 2014-06-25 06:41:32 rather than vmlinuz-something 2014-06-25 06:41:44 i actually fixed ith the other day, just didnt push it 2014-06-25 06:41:48 ah 2014-06-25 06:42:12 and i think i fixed it in a vm, which i likely has overwritten by now... 2014-06-25 06:42:14 tell me when it's pushed and i'll upgrade 2014-06-25 06:42:19 sure 2014-06-25 06:42:22 :/ 2014-06-25 06:42:40 oh, this is the reason why i prefer that users use bugs.a.o to file bugs 2014-06-25 06:42:55 i tend to forget things when only reported in irc... 2014-06-25 06:43:02 unless i get to fix it immediatly 2014-06-25 06:43:05 *nod* 2014-06-25 06:43:24 re bb utf 2014-06-25 06:43:36 why the magic number 0x10ffff? 2014-06-25 06:43:41 that's the end of unicode/ucs 2014-06-25 06:43:48 a larger number would be ok too 2014-06-25 06:43:52 ok 2014-06-25 06:44:02 but in case it's helpful for busybox to know that nothing larger will appear, that's safe 2014-06-25 06:44:12 no, its ok, i just want it in the commit message 2014-06-25 06:44:18 that should be 1114111 decinal 2014-06-25 06:44:31 hah nice 2014-06-25 06:44:47 i'd never realized that 2014-06-25 06:45:09 neither did i, but apparently bb config only supports decimal... 2014-06-25 06:45:53 i hope it does not alloc an array with chars[CONFIG_LAST_SUPPORTED_WCHAR] or something.... 2014-06-25 06:46:00 ah i think it accepts 0 as no limit 2014-06-25 06:46:11 afaik it doesn't do anything like that 2014-06-25 06:46:16 good 2014-06-25 06:46:24 you prefer 0 for no limit? 2014-06-25 06:46:29 i think 0x10ffff is ok 2014-06-25 06:46:34 i dunno, maybe read the config help? 2014-06-25 06:46:56 btw, afaik this "feature" is just to prevent attempting to display CJK on primitive terminals and messing up the cursor position... 2014-06-25 06:47:40 seems so yes 2014-06-25 06:47:54 Any character with Unicode value bigger than this is assumed │ 2014-06-25 06:47:54 │ to be non-printable on output device. Many applets replace │ 2014-06-25 06:47:54 │ such chars with substitution character. 2014-06-25 06:48:13 │ The idea is that many valid printable Unicode chars are │ 2014-06-25 06:48:13 │ nevertheless are not displayed correctly. Think about │ 2014-06-25 06:48:14 │ combining charachers, double-wide hieroglyphs, obscure │ 2014-06-25 06:48:14 │ characters in dozens of ancient scripts... │ 2014-06-25 06:48:14 │ Many terminals, terminal emulators, xterms etc will fail │ 2014-06-25 06:48:14 │ to handle them correctly. Choose the smallest value 2014-06-25 06:48:42 4351 (0x10ff) - there are no double-wide chars in [0..4351] range, │ 2014-06-25 06:48:42 │ code is ~300 bytes smaller for this case. 2014-06-25 06:49:12 maybe we should use 0x10ff instead of 0x10ffff? :) 2014-06-25 06:49:23 │ 12799 (0x31ff) - nearly all non-ideographic characters are │ 2014-06-25 06:49:23 │ available in [0..12799] range, including │ 2014-06-25 06:49:23 │ East Asian scripts like katakana, hiragana, hangul, │ 2014-06-25 06:49:23 │ bopomofo... │ 2014-06-25 06:49:23 │ 0 - off, any valid printable Unicode character will be printed. │ 2014-06-25 06:49:54 i'd say we use 0x31ff, 12799 2014-06-25 06:50:59 why? 2014-06-25 06:51:18 because thats 'nearly all' 2014-06-25 06:51:19 their info is bogus 2014-06-25 06:51:23 but second thought 2014-06-25 06:51:30 plenty of languages need characters outside that range 2014-06-25 06:51:39 i think we do 0x10ff and increase on request 2014-06-25 06:51:47 i've already requested 2014-06-25 06:51:52 yes 2014-06-25 06:51:53 full range 2014-06-25 06:52:26 so you actually meant 0x10ffff and not 0x10ff 2014-06-25 06:52:26 ok 2014-06-25 06:52:54 lets do that then 2014-06-25 06:52:58 yes i meant 0x10ffff (all of unicode 2014-06-25 06:53:06 np 2014-06-25 06:53:15 0x10ff excludes basically everything but european langs and indian langs 2014-06-25 06:53:50 and it's even missing the extended latin and greek blocks 2014-06-25 06:53:59 ~1e00/1f00 2014-06-25 06:54:13 and all unicode punctuation (~2000) 2014-06-25 06:58:40 dalias: fixed bb pushed 2014-06-25 06:58:42 thanks! 2014-06-25 06:58:57 and so it was update-extlinux 2014-06-25 06:59:04 ? 2014-06-25 06:59:36 vanilla kernel in extlinux.conf 2014-06-25 07:01:24 which package needs to be updated for it to work tho? 2014-06-25 07:01:32 kernel or something else? 2014-06-25 07:02:08 syslinux, which ships the update-extlinux script 2014-06-25 07:02:12 i found the fixed script 2014-06-25 07:02:15 ah ok 2014-06-25 07:03:11 doesn't seem to be pushed..? 2014-06-25 07:03:19 i'm working on it 2014-06-25 07:03:22 ah ok 2014-06-25 07:03:30 np 2014-06-25 07:03:38 thanks for your patience ;) 2014-06-25 07:03:43 :) 2014-06-25 07:05:06 its always scary to mess with the boot. if i mess up i'll have 1000 angry users.. 2014-06-25 07:05:34 yes.. 2014-06-25 07:06:12 I learned the hard way to /never/ fuck with the bootloader without 3 good reasons and recovery commands written down 2014-06-25 07:07:27 reason 1) dalias want vanilla kernel. reason 2) i want be able to boot vanilla kernel reason 3) hyper-v users want be able to boot vanilla kernel 2014-06-25 07:07:31 :) 2014-06-25 07:07:51 touche 2014-06-25 07:07:59 dalias: syslinux should be fixed now 2014-06-25 07:08:03 thanks 2014-06-25 07:08:16 thanks for the reminder, sorry for forgetting it... 2014-06-25 07:08:43 ncopa: btw, I never was able to track down the problem with that system that had the fucked apk 2014-06-25 07:08:50 hmm apk add --upgrade syslinux isn't finding anything new 2014-06-25 07:09:05 dalias: if you can confirm that it works (apk upgrade -U && cat /boot/extlinux.conf) 2014-06-25 07:09:19 what repo do you have in /etc/apk/repositories ? 2014-06-25 07:09:23 I ended up having to reinstall that server and I just slapped Ubuntu server 14.04 on it 2014-06-25 07:09:27 v3.0 and edge 2014-06-25 07:09:31 nl.alpinelinux.org ? 2014-06-25 07:09:37 oh no, a mirror 2014-06-25 07:09:41 i suppose i should change that 2014-06-25 07:09:42 use nl.a.o 2014-06-25 07:10:07 the others gets synced approx every hour or so 2014-06-25 07:10:41 (i have idea how to make them sync instantly...) 2014-06-25 07:10:55 what's the full url i should have? 2014-06-25 07:11:11 http://nl.alpinelinux.org/alpine/edge/main 2014-06-25 07:11:35 ah got it now 2014-06-25 07:12:34 vanilla shows up as "label 0" 2014-06-25 07:12:45 how do i make it default? default=0 ? 2014-06-25 07:13:10 yeah that seems to work 2014-06-25 07:13:13 i guess so 2014-06-25 07:13:38 but vanilla kernel showed up, right? 2014-06-25 07:13:42 yes 2014-06-25 07:14:10 hmm one other issue 2014-06-25 07:14:16 "networking" service fails to start 2014-06-25 07:14:22 this has been going on a while 2014-06-25 07:14:36 ip: RTNETLINK answers: File exists 2014-06-25 07:15:04 do you run dhcpcd service in a 'catch all' mode? 2014-06-25 07:15:16 it could be the 'lo' interface is configured double 2014-06-25 07:15:20 in whatever the default mode is 2014-06-25 07:15:36 i thought lo was autoconfigured anyway by the kernel nowdays tho 2014-06-25 07:15:58 that could be the case yes 2014-06-25 07:16:12 that kernel autconfigures it and then 'networking' does it for you too 2014-06-25 07:16:29 hm 2014-06-25 07:16:31 i have 2014-06-25 07:16:36 auto lo 2014-06-25 07:16:36 iface lo inet loopback 2014-06-25 07:17:19 dalias: do you run the dhcpcd service? 2014-06-25 07:17:39 it could be that dhcpcd and udhcpc is fighting about the interface? 2014-06-25 07:17:51 yes 2014-06-25 07:18:20 dhcpcd wants take control of it all 2014-06-25 07:18:53 anyway this had dropbear server broken until recently 2014-06-25 07:19:08 i just removed dropbear's dependency on networking from the service file.. 2014-06-25 07:19:21 actually the dependency doesn't make sense to me 2014-06-25 07:19:31 there's no reason it can't run perfectly fine without networking up 2014-06-25 07:19:45 (just binding to :: and/or 0.0.0.0) 2014-06-25 07:23:33 do you have ipv6? 2014-06-25 07:23:53 not especially configured, just link-local addresses 2014-06-25 07:24:02 i dont think you need start 'networking' serivice at all if you use 'dhcpcd' service 2014-06-25 07:24:08 which work amazingly well with musl :) 2014-06-25 07:24:26 both 'networking' and 'dhcpcd' should provide 'net' 2014-06-25 07:24:36 well dropbear depends on networking i think 2014-06-25 07:24:47 ok, that might be a bug then 2014-06-25 07:24:50 and so does busybox ntpd 2014-06-25 07:24:52 i think it should depend on 'net' 2014-06-25 07:25:19 it could also be that i added 'networking' because of 'lo' interface 2014-06-25 07:25:22 i dont remember 2014-06-25 07:25:52 re dropbear, i got a report someone that had ipv6 only interface 2014-06-25 07:25:58 and services could not bind 2014-06-25 07:26:04 due to ipv6 dad 2014-06-25 07:26:33 similar to this: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=705996 2014-06-25 07:27:09 i dunno if its related though 2014-06-25 07:27:51 that should only matter for programs which bind to a specific ip 2014-06-25 07:28:16 binding to :: or 0.0.0.0 should always work 2014-06-25 07:29:20 btw it looks to me like DAD should just be turned off -- according to that bug report it adds 2 second latency before the interface becomes usable, and it's unlikely to ever have any effect 2014-06-25 07:29:49 oh, true 2014-06-25 07:49:20 btw glad to find out about accept_dad 2014-06-25 07:49:37 that's an option i need to fix on any systems that'll be using ipv6 2014-06-25 07:56:30 i think the wrokaround was to disable it 2014-06-25 07:57:01 i think i need get ipv6.. 2014-06-25 07:57:17 too many people asks about it and i have basically zero experience with it 2014-06-25 08:35:51 kocka: Where can i find the zfs Patches? 2014-06-25 08:36:51 https://github.com/zfsonlinux/zfs/pull/2380 2014-06-25 08:51:56 kocka: nice! 2014-06-25 08:52:17 hey is there a ready script to turn off all HDs on run from ram setup 2014-06-25 08:52:56 I am using hdparm commands manually 2014-06-25 08:53:14 what do you have on your hds? 2014-06-25 08:53:16 the system? 2014-06-25 08:53:30 do you have anything of importance under /var (except logs) 2014-06-25 08:54:31 it depends completely on the apps you run 2014-06-25 08:55:18 running from usb thumb drive 2014-06-25 08:55:26 using laptop as router 2014-06-25 08:55:35 cache is on usb 2014-06-25 08:55:41 and lbu commits on usb 2014-06-25 08:55:44 ah so you dont actually use the disks 2014-06-25 08:55:48 no 2014-06-25 08:55:52 you just want to power them off 2014-06-25 08:55:57 yes 2014-06-25 08:56:11 but if there is a smart script that tests to see if the devicename is a real hardisk would be nice 2014-06-25 08:56:12 i suppose google hdparm 2014-06-25 08:56:30 before trying to shut it down with hdparm -Y 2014-06-25 08:57:52 i think we ship a hdparm init.d script so you can se hdparms at boot 2014-06-25 08:58:12 if you dont mount any disk, then it should be no problem powering off the disk with hdparm 2014-06-25 08:58:39 i supose you could also physically remove the disk 2014-06-25 09:02:14 kocka: Thanks a lot! I will give it a try in the next couple of days! 2014-06-25 09:06:17 thanks to stef for the patch 2014-06-25 09:17:38 so weird for some reason iptables -P INPUT DROP freezes connections to dnsmasq, I did set ACCEPT rules to allow all traffic from appropiate addresses I think 2014-06-25 09:54:13 it isn't possible to completly turn off machine and have it wake up at specific time correct? 2014-06-25 09:54:22 I remember there used to be an alarm clock in the bios in the old days 2014-06-25 09:54:45 depends on your bios yes 2014-06-25 09:54:58 or you can do wake-on-lan 2014-06-25 09:55:03 should still be there 2014-06-25 09:55:12 on most bios vresion 2014-06-25 09:55:15 versions 2014-06-25 09:59:04 atleast it does on our motherboards, cant vouch for other brands. 2014-06-25 09:59:54 yea, they don't have it on this laptop, I thought it was permently removed by some newer tech I missed 2014-06-25 10:59:54 say, I installed a bunch of packages I don't want to commit with lbu commit, but there are some files I want to save, how can I do that? 2014-06-25 11:00:09 if I do lbu include files I need to lbu commit that will also save the packages to world ? 2014-06-25 11:06:16 if you do lbu commit is will store /etc/apk/world yes 2014-06-25 11:07:24 best apk very world 2014-06-25 11:07:30 only for you 2014-06-25 16:04:03 ah, nothing like random alpine reboots :-P 2014-06-25 16:04:19 going to have to setup serial output to get to the bottom of it, since it's diskless and doesn't keep logs 2014-06-26 07:19:00 pk add bash 2014-06-26 07:19:00 ERROR: unsatisfiable constraints: 2014-06-26 07:19:00 .setup-apkrepos (missing): 2014-06-26 07:19:00 required by: world[.setup-apkrepos] 2014-06-26 07:19:21 what is this .dot package .setup-apkrepos /etc/apk/world 2014-06-26 07:42:04 ncopa: is this patch fixed the problem? 2014-06-26 07:42:26 its a temp 'virtual' or empty package 2014-06-26 07:42:32 which only is there to pull in its deps 2014-06-26 07:42:36 http://forums.grsecurity.net/viewtopic.php?t=3982#p14104 2014-06-26 07:42:55 zxd: i suppose you can apk del .setup-apkrepos 2014-06-26 07:43:50 baiscally the idea is to pull in temp repos, lets say you need a package while doining something from script 2014-06-26 07:44:02 you want install it before you start and delete it when ready 2014-06-26 07:44:16 apk add mypkg; ... do something; ... apk del mypkg 2014-06-26 07:44:34 if mypkg was installed before you stared you get problem - it will be deleted at end 2014-06-26 07:44:54 if someone else do apk del mypkg while you 'do something' you get problem too 2014-06-26 07:45:15 so what you do is: apk add --virtual .temp mypkg 2014-06-26 07:45:29 the .temp gets installed as an empty package, that depends on mypkg 2014-06-26 07:45:52 is someone apk del mypkg, apk will complain: mypkg is needed by .temp 2014-06-26 07:46:28 if mypkg was installed from before, apk will say that deps are ok and it will not install it 2014-06-26 07:46:36 when youre done you do: apk del .temp 2014-06-26 07:46:49 so its a way to 'lock' packages temproary 2014-06-26 07:46:59 we do that when installing deps from abuild 2014-06-26 07:47:41 kocka: http://grsecurity.net/changelog-stable2.txt 2014-06-26 07:47:51 its in grsecurity-3.0-3.14.8-201406222110.patch 2014-06-26 07:48:42 commit acccccdea02e1bdde09f762fc4bf934682c936fd 2014-06-26 07:48:42 Author: Brad Spengler 2014-06-26 07:48:42 Date: Thu Jun 19 13:12:03 2014 -0400 2014-06-26 07:48:42 Fix bad git merge caused by recent upstream backport, reported by ncopa: 2014-06-26 07:48:42 http://forums.grsecurity.net/viewtopic.php?t=3982&p=14104#p14104 2014-06-26 07:48:44 net/ipv4/ping.c | 2 +- 2014-06-26 07:48:46 1 files changed, 1 insertions(+), 1 deletions(-) 2014-06-26 07:49:37 ok, looking for 3.2.60 2014-06-26 07:49:58 oh cool 2014-06-26 08:02:06 hmm 2014-06-26 08:02:15 maybe it's not fixed in 3.2.60? 2014-06-26 08:02:24 can't see any changes 2014-06-26 08:02:34 in net/ipv4/ping.c 2014-06-26 08:03:05 could you give me the name of the previous 3.4.18 patch? 2014-06-26 08:07:28 3.14.8? 2014-06-26 08:07:52 grsecurity-3.0-3.14.8-201406191347.patch 2014-06-26 08:08:14 thx 2014-06-26 08:18:57 hm 2014-06-26 08:18:59 nevermind 2014-06-26 08:19:16 i cant see any changes but somehow its been fixed 2014-06-26 08:25:56 Bonjour all! 2014-06-26 08:27:17 bom dia, ginjachris! 2014-06-26 08:55:16 fala portugues? 2014-06-26 08:59:44 as vezes 2014-06-26 10:11:34 good morning 2014-06-26 10:28:14 Good morning Anaphaxeton 2014-06-26 11:30:10 what's with "mdev: bad line 73: 2 tokens found, 3 needed" message at boot? (Not that I've noticed it causing any problems) 2014-06-26 11:37:24 ginjachris: latest release? 2014-06-26 11:38:18 yes indeedy! 3.0.1, 64 bit 2014-06-26 11:38:42 Noticed it too with 3.0.0 after upgrade.....I guess a change in busybox somewhere? 2014-06-26 11:40:33 let me upgrade myself and see if i have the same issue 2014-06-26 11:42:50 got the same issue 2014-06-26 11:42:55 looks like SUBSYSTEM=sound;.*root:audio 0660 2014-06-26 11:45:01 hm, I always use alpine as a serious server platform, so I have never used sound with it! 2014-06-26 11:45:35 not sure what the right syntax is 2014-06-26 11:46:21 does this help: http://git.busybox.net/busybox/tree/docs/mdev.txt 2014-06-26 11:46:40 the mdev.conf bit? 2014-06-26 11:47:39 is the semi colon needed? 2014-06-26 11:48:31 i think there needs to be a whitespace before root 2014-06-26 11:49:28 yep 2014-06-26 11:49:29 ah i see.....yes quite possibly! This is mostly beyond me 2014-06-26 11:49:33 looks like it works 2014-06-26 11:49:37 let me commit that 2014-06-26 11:49:38 sweet :) 2014-06-26 11:49:40 thanks! 2014-06-26 12:02:45 ginjachris: its commited to edge, waiting for ncopa to add it to stable (i dont have 3.0 repo by hand) 2014-06-26 12:03:18 awesome, thanks very much, that was super quick! 2014-06-26 14:39:51 Does anyone have any recommendations for books/online resources re: learning C coding? I have an idea! 2014-06-26 14:41:13 or, is C++/C# recommended over C? 2014-06-26 14:50:10 i dont think anyone here prefers C++ 2014-06-26 14:54:41 are you only considering languages named with letter "C" and some punctuational suffix? 2014-06-26 14:55:19 oh i suppose I would consider other languages 2014-06-26 14:55:31 C seems to be common 2014-06-26 14:56:07 I would need something that can take highly accurate time measurements (I think C function gettimeofday does that?) 2014-06-26 14:56:39 My only knowledge so far is using something based on TCL 2014-06-26 14:56:53 and some very basic bash scripting 2014-06-26 14:59:15 if you share your idea, maybe somebody can provide you a proper advise. 2014-06-26 14:59:40 good point.....it may be a very dumb idea though! 2014-06-26 14:59:47 so, entropy 2014-06-26 14:59:59 basically a poor man's entropy daemon 2014-06-26 15:01:03 using variance in ntpd updates, and possibly sleeps (like http://www.vanheusden.com/te/), to create entropy and add it to the usual pool 2014-06-26 15:01:19 in fact it would probably be on demand rather than a daemon 2014-06-26 15:01:51 I'm sure a bunch of other smaller things could be added too 2014-06-26 15:04:37 even if the entropy generated is poor, generating a few streams and XOR'ing together (whether that's before adding to the entropy pool or otherwise - I think the linux entropy pool might do that anyway, but i need to learn some coding in order to read the code i guess) can only result in better entropy 2014-06-26 15:04:52 i'm no expert on the matter, but the linked article doesn't seem to make much sense under preemptive multitasking environment like linux. 2014-06-26 15:05:47 I can't find the source anymore, but there was some clever maths done in the 80's that apparently proved getting 2 poor entropy sources and XORing together can only result in better entropy 2014-06-26 15:06:40 ovf: basically you'd run it to generate entropy on a system with low entropy, like before generating keys for example (maybe during new install for instance) 2014-06-26 15:07:40 ginjachris: if you want a language that is really helpful with parallelism try Ada 2014-06-26 15:07:49 it's fast and safe 2014-06-26 15:08:05 or stick to C! 2014-06-26 15:08:19 i get that, but if the page is referring to sleep as in usleep(2), the actual sleep time you get is most dependent on the system scheduler, not on any timer-specific properties 2014-06-26 15:10:08 I couldn't tell you if it refers to usleep(2) or not I'm afraid 2014-06-26 15:12:23 I always thought (maybe incorrectly) it did 'sleep 1' or whatever, then calculated how long the sleep had taken in microseconds (or some suitably accurate time measurement) 2014-06-26 15:12:40 Thanks Anaphaxeton! 2014-06-26 15:18:37 right I need to go, it's been a long day....my idea may seem less attractive after some sleep 2014-06-26 15:18:39 bye all 2014-06-26 15:21:28 did he do a pretty decent pun by accident? 2014-06-26 16:58:15 wget download in alpine acting as router is sgnificantly faster vs download from box routed via alpine 1.8MB/s vs 500KB/s~ any ideas why this could be? 2014-06-26 17:07:27 100mbit between box and alpine I did a test it's 10MB/s 2014-06-26 17:23:12 could hardware, tso/gso issue, or bad cable/switch. 2014-06-26 19:18:17 "oh boy, /sbin/init not found in new root. Launching emergency recovery shell" 2014-06-26 19:18:23 Type exit to continue boot 2014-06-26 19:19:33 are you booting off the wrong partition? 2014-06-26 19:30:35 Loading user settings from /media/sda1/localhost.apkovl.tar.gz 2014-06-26 19:30:35 Installing packages to root filesystem [ 17...] EXT4-fs (sda1): re-mounted. Opts: data=ordered 2014-06-26 19:30:35 OK: 0Mib in 0 packages 2014-06-26 19:30:35 18.5] EXT4-fs (sda1) re-mounted.Opts: data=ordered ok 2014-06-26 19:30:53 sounds like something got screwed in media/sda1/localhost.apkovl.tar.gz ? 2014-06-27 05:59:09 I came in here a while back and was having issues compiling NodeJS 2014-06-27 05:59:40 I ended up installing alpine vanilla and it solved all my segmentation fault issues so I can only assume it was a kernel issue 2014-06-27 05:59:49 **kernel security issue 2014-06-27 06:54:09 BrockLee: dmesg will probably tell you if grsec jumped in. 2014-06-27 06:56:20 oh what 2014-06-27 06:56:23 I had no idea 2014-06-27 06:56:34 Well, time to go check 2014-06-27 08:15:06 where are virtual packages located? I think I removed .setup-alpine virtual package fromw orld with apk del and it broke my ovl ? foo.apkovl.tar.gz 2014-06-27 08:39:06 Morning all! 2014-06-27 09:12:36 what does --overlay-from-stdin in apk mean Read list of overlay 2014-06-27 09:12:46 what are overlay files 2014-06-27 09:16:49 overlays are for non-disk installs 2014-06-27 09:16:56 run from ram via USB for instance 2014-06-27 09:17:33 one of many things on my list to experiment with. I've been purely disk installs thus far 2014-06-27 09:19:45 what are overlays files 2014-06-27 09:21:48 read this: http://wiki.alpinelinux.org/wiki/Alpine_local_backup 2014-06-27 09:22:15 saves changes of a run from ram system 2014-06-27 09:22:24 so changes are not lost between reboots for instance 2014-06-27 09:25:58 so what does it mean when apk read overlay from stdin ? it's just a tarball with /etc no ? apk expects .apk files ? 2014-06-27 09:27:50 where exactly are you seeing this message? What are you trying to achieve 2014-06-27 09:29:12 apk --help , --overlay-from-stdin Read list of overlay files from stdin 2014-06-27 09:29:12 , i was debugging boot process with debug_init saw a line with apk and this 2014-06-27 09:29:16 was wondering what it was doing 2014-06-27 09:35:15 Well if you have a disk install, it shouldn't be relevant. 2014-06-27 09:35:55 I am loading from ram 2014-06-27 09:36:02 hmm from usb 2014-06-27 09:36:06 no disk 2014-06-27 09:39:31 I've no experience with run from ram, but I'm guessing it just gives you the option to provide an overlay file location from stdin, rather than searching for apkovl on the configured location (usually the USB) 2014-06-27 09:41:46 ah I see 2014-06-27 10:34:15 zxd: http://git.alpinelinux.org/cgit/apk-tools/commit/?id=d0909569c639a033a6f1c47a95745886ebe3e093 2014-06-27 11:41:56 anyone here using vlc? 2014-06-27 11:46:08 I would, if I built an Alpine desktop machine 2014-06-27 11:46:15 but I'm not at the moment 2014-06-27 11:46:37 I prefer vlc to alot of other audio/video players 2014-06-27 11:46:44 im wondering how well it will work without any plugins installed. 2014-06-27 11:47:52 web plugins? Or literally no plugins 2014-06-27 11:48:27 no plugins at all 2014-06-27 11:51:01 I couldn't say for sure, but probably wouldn't be a good experience 2014-06-27 11:52:39 I think it would break alot of stuff.....no mp3 decoding for instance 2014-06-27 11:52:54 (I'm just looking at the plugin list on a Windoze machine) 2014-06-27 11:54:39 yeah i think no plugins would remove alot of vlc's functionality 2014-06-27 11:56:40 http://sprunge.us/MKWA 2014-06-27 11:56:45 those are the plugins im talking about 2014-06-27 11:58:56 I guess some of those might be removable, but the ones like video and audio output I expect would be essential 2014-06-27 12:00:49 sadly i am no expert in the subject though! 2014-06-27 15:18:11 bye all, have a great weekend! 2014-06-27 17:06:26 <^7heo> guys 2014-06-27 17:06:32 <^7heo> does anyone here use alpine on a server? 2014-06-27 17:06:43 <^7heo> not a home server 2014-06-27 17:06:50 <^7heo> a company-grade server. 2014-06-27 17:06:55 <^7heo> running web services and all 2014-06-27 18:02:39 ^7heo: yes 2014-06-27 18:03:08 <^7heo> kocka: I do too, now. 2014-06-27 18:03:14 <^7heo> =] 2014-06-27 18:03:19 <^7heo> I love alpine. 2014-06-27 18:03:25 <^7heo> it just works, and that rocks :D 2014-06-27 18:03:31 <^7heo> and it uses musl. 2014-06-27 18:03:33 <^7heo> \o/ 2014-06-27 18:03:40 <^7heo> yeah \o/ bot! 2014-06-27 18:03:55 <^7heo> it's easy to cheer up the bot here :) 2014-06-27 18:05:00 :D 2014-06-27 18:05:10 i haven't upgraded yet to 3.0 2014-06-27 18:05:31 that machine is still 2.7 2014-06-27 18:06:22 <^7heo> well, musl is worth it imho. 2014-06-27 18:19:32 is there a step by step howto how to upgrade from 2.7.9 to 3.0.1? 2014-06-27 18:26:41 <^7heo> anyone kn ows if sshd can use PAM on alpine? 2014-06-27 18:26:49 <^7heo> 'cause the configuration option isn't known from the daemon 2014-06-27 18:38:20 i don't think it's built to, but that would probably be a reasonable feature request 2014-06-27 21:18:30 ^7heo: for starters, busybox is not built with pam support for login(1) in alpine 2014-06-27 21:52:02 I'm building diskless XEN server, I'm wondering how to go about modloop. Perhaps make the rc script copy modules from initrd if it fails to find squashfs image, or would there be better way? The idea is to not depend on NFS as the dom0 will passthru the network interface to guests. 2014-06-27 21:52:48 (the initrd is custom anyway, so necessary modules would be added at mkinitfs time) 2014-06-27 23:37:36 <^7heo> thanks to alpine linux, I pay $2.5 a month for a VPS that is incredibly fast. 2014-06-27 23:37:50 <^7heo> (not ONLY thanks to alpine linux but partly thanks to it) 2014-06-27 23:40:23 hey guys, how screwed I am if I want to switch from the 2.6 to the 3.0 repo? 2014-06-27 23:40:54 wouldn't the logical approach be to go 2.7, then 3.0? 2014-06-27 23:41:14 ah, that's what I wanted to know 2014-06-27 23:41:50 uclibc to musl makes it extra important to be careful when upgrading to 3.0, as far as I know 2014-06-27 23:42:06 ACTION has been too lazy to get it done 2014-06-27 23:42:25 uhh, if I'm not planning on rebooting in the next couple of months I guess that'd be a partykiller here 2014-06-27 23:42:48 <^7heo> personally, I re-install. 2014-06-27 23:43:02 I... can't 2014-06-27 23:45:39 thanks for the info guys 2014-06-28 08:30:04 <^7heo> nc is very useful :) 2014-06-28 08:31:11 <^7heo> (I say that here because the wget of alpine did fail me more than once) 2014-06-28 19:01:04 is there really no xen support in 3.0 yet? 2014-06-28 19:01:31 i've broken so many things today i could even update 2014-06-28 22:08:22 so uhm, I guess there's no package for dsniff? 2014-06-28 22:08:44 are there any other things that do arp spoofing that are packaged for alpine? 2014-06-28 22:09:30 ACTION is trying to think creatively to find ways around a stupid networking issue 2014-06-29 16:47:27 It is said that "Alpine Linux is a community-developed operating system designed for x86 Routers, Firewalls, VPNs, VoIP and servers." 2014-06-29 16:47:58 But the project interested me and I wanted to try it out as my x64 Desktop and Laptop. 2014-06-29 16:48:04 Is that a bad idea? 2014-06-29 16:55:06 i find it quite nice for laptop use 2014-06-29 16:57:19 My main concern is lack of packages. I couldn't even find XMonad. 2014-06-29 16:57:37 Compiling from source and manually maintaining packages is really a hassle 2014-06-29 17:00:58 the set of X packages is fairly minimal 2014-06-29 17:02:14 Looks like so. Is there any will to expand alpine to desktop/laptop use? If so, I might check it back in the future when there are more packages, such as XMonad, spacefm, chromium, etc. 2014-06-29 17:16:07 that's a good question 2014-06-29 17:16:17 are there plans to expand the set of gui packages? 2014-06-29 17:22:52 wb 2014-06-29 17:23:20 i'm using xfce and firefox and they work quite well. fastest-starting firefox i've ever seen. 2014-06-29 17:24:20 I think firefox is a good chromium alternative, I can deal with that. 2014-06-29 17:24:37 But the lack of Xmonad and SpaceFM is too much for me to take. 2014-06-29 17:24:52 I guess I will just compile those myself, if no packages for them arrive any time soon. 2014-06-29 17:28:47 supposedly chromium is reasonably feasible to compile from source 2014-06-29 17:29:26 i have unconfirmed reports of (non-alpine) musl-libc users successfully building chromium 2014-06-29 17:30:15 Neat 2014-06-29 21:23:34 Hi there. I'm bit buffled how to set 'ulimit -l' (RLIMIT_MEMLOCK) during boot time for specific user or system wide - desktop system. Would appreciate any hints. 2014-06-30 05:56:01 I think something is broken with lbu status -v, it lbu status shows a small list of changed files 2014-06-30 05:56:20 lbu status -v shows a big list even for files not in my exclude/include list 2014-06-30 06:01:36 btw, does anyone know a way to turn off the screen it never goes to powersaving 2014-06-30 06:22:49 zxd: in xorg? or terminal? 2014-06-30 06:29:24 melville: echo 'rc_ulimit=-l ...' >> /etc/rc.conf 2014-06-30 06:29:30 will set it systemwide 2014-06-30 06:43:40 ncopa in terminal 2014-06-30 06:51:31 zxd: setterm? 2014-06-30 06:52:19 setterm -powersave off 2014-06-30 06:52:20 maybe? 2014-06-30 06:52:25 thats what google indicates.. 2014-06-30 06:52:32 google setterm 2014-06-30 06:56:56 ncopa: in what package is setterm? 2014-06-30 06:57:43 hm 2014-06-30 06:57:49 i thought busybox provided it 2014-06-30 06:57:56 looks like it does not. util-linux 2014-06-30 06:58:12 set 2014-06-30 07:04:20 ncopa, did you see niao's questions? 2014-06-30 07:04:41 i'm interested too -- are there plans to expand the set of available desktop-app-type packages? 2014-06-30 07:05:10 ncopa, btw re: setterm, i'm pretty sure it just echos linux-console-specific escapes 2014-06-30 07:05:14 see man console_codes 2014-06-30 07:05:46 you should be able to generate the same escapes with the 'printf' command or 'echo -ne' (latter is non-portable tho) 2014-06-30 07:06:01 setterm: cannot (un)set powersave mode: Not a tty 2014-06-30 07:06:23 ncopa, one other thing: did you ever figure out the weird behaviors in firefox? (e.g. home screen not working, not restoring tabs) 2014-06-30 07:06:50 zxd, setterm has to be run on a virtual-terminal tty 2014-06-30 07:07:15 dalias: is there a way to specify tty1 on the command line, or with the printf escape sequence like you said 2014-06-30 07:07:29 maybe >/dev/tty1 2014-06-30 07:09:28 tried from virtual terminal tty1 it runs prints no error but dosen't turn off display 2014-06-30 07:09:50 I read somewhere that APM features are disabled from alpine linux kernel, don't know if it's related 2014-06-30 07:10:48 dalias: i saw the question re xorg apps. I dont have any specific roadmap on X apps we want add 2014-06-30 07:11:13 but we can add on request for specific apps 2014-06-30 07:11:42 most likely we will need someone to maintain them 2014-06-30 07:12:48 zxd: i could probably enable apm if you are sure that will solve your issue 2014-06-30 07:13:07 i'm trying to disable things and enable on request only 2014-06-30 07:13:36 dalias: re firefox, i havent looked at it yet 2014-06-30 07:14:47 i asked in -devel, instead here, 2014-06-30 07:15:15 is there a way to tell 3.0.1 iso to output to serial console? 2014-06-30 07:15:27 its a bit hard to install on headless server (apu4c) 2014-06-30 07:25:26 hmm anyone know how to make suspend work in alpine? 2014-06-30 07:25:34 i installed pm-utils and running pm-suspend manually works 2014-06-30 07:25:46 but neither the Zz button nor the lid makes the laptop suspend 2014-06-30 07:25:47 dalias: i have actually never tried it, alpine boots fast enough for me :) 2014-06-30 07:26:10 suspend isn't about boot time but about not rebooting at all 2014-06-30 07:26:18 (so you have your working state undisturbed) 2014-06-30 07:26:46 it lose the network connetions? 2014-06-30 07:27:10 no, i just mean anything you're in the middle of 2014-06-30 07:27:19 editor sessions, shells, browser state, ... 2014-06-30 07:27:54 yes, i understand that 2014-06-30 07:28:00 i think there is a tool for xfce 2014-06-30 07:28:10 power manager 2014-06-30 07:28:17 the xfce items are all grayed out :/ 2014-06-30 07:28:29 aha 2014-06-30 07:28:34 grsec kernel? 2014-06-30 07:28:35 but on debian the Zz button works even on virtual consoles, i think 2014-06-30 07:28:38 non-grsec 2014-06-30 07:28:58 ok, then its likely some missing package 2014-06-30 07:28:58 running the pm-suspend command manually works fine tho 2014-06-30 07:29:14 it's just lacking the automatic handling of Zz button and lid 2014-06-30 07:29:28 i think xfce needs system dbus and upower installed 2014-06-30 07:29:42 i installed upower thinking it might help 2014-06-30 07:29:48 do i have to add a service to run it? 2014-06-30 07:29:55 i think system dbus might need be running 2014-06-30 07:29:56 rc-service -l didn't show anything that looked interesting 2014-06-30 07:30:05 rc-update add dbus 2014-06-30 07:30:15 and /etc/init.d/dbus start 2014-06-30 07:30:24 lemme check on my lappy 2014-06-30 07:31:25 i suppose you could hack something with a keypress event listener daemon and execute a given command 2014-06-30 07:31:37 but it would be nice if it "just worked" 2014-06-30 07:34:19 dalias: I'm migrating my laptop from arch to alpine and I've packaged most X apps I use 2014-06-30 07:35:11 its a pretty minimal setup though, but now alpine support everything I use except spotify 2014-06-30 07:36:50 Xmonad would be hard to ort th 2014-06-30 07:37:08 err, hard to port due to Haskell 2014-06-30 07:37:30 and SpaceFM is abandoned upstream 2014-06-30 07:40:07 is there any sane way to bootstrap haskell? 2014-06-30 07:40:32 iirc it requires its own binaryware to build itself 2014-06-30 07:42:01 as far as I understand you have to cross compile a ghc binary from another system 2014-06-30 07:42:37 i think someone looked at it half year or a year ago 2014-06-30 07:42:48 its nontrivial iirc 2014-06-30 07:44:17 :/ 2014-06-30 07:45:05 http://wiki.alpinelinux.org/wiki/Porting_GHC_to_Alpine 2014-06-30 07:45:27 more than a year ago 2014-06-30 07:46:11 we did something similar for ada not too long ago 2014-06-30 07:46:11 go is porting its compiler from c to go and will likely loose its bootstrap capability in the future as well 2014-06-30 07:46:25 urgh... 2014-06-30 07:47:07 but, in theory you could start with the last non-go compiler version and build the next compiler and so on 2014-06-30 07:47:40 i think one of the nicest bootstraps out there is ocaml's 2014-06-30 07:48:31 it begins by compiling a bytecode interpreter written in c, then it runs the packaged bytecode compiled compiler to build everything else, including a native compiler 2014-06-30 07:52:13 gcc has the same issue 2014-06-30 07:52:29 they rewrote it in C++ 2014-06-30 07:52:40 so now you can't bootstrap directly 2014-06-30 07:52:55 you need to first build an older gcc with a c++ compiler, than build current gcc 2014-06-30 07:53:09 and yes 2014-06-30 07:53:18 ocaml's bootstrap is really nice 2014-06-30 07:53:38 you can even use the bytecode compiler, and make 'binaries' with it 2014-06-30 07:53:59 if the 'real' compiler doesn't work on your system 2014-06-30 07:54:10 i had ocaml bytecode working way back around musl 0.5.0 2014-06-30 07:54:41 and actually way back in 2005 on the previous-gen work that later became musl 2014-06-30 08:24:38 dalias: suspend is grayed out here too.. hmm.. i dont know what is needed so make suspend work properly 2014-06-30 08:30:15 bash 2014-06-30 08:30:17 bah 2014-06-30 08:30:35 upower has deprecated .Suspend 2014-06-30 08:30:55 they say: use org.freedesktop.login1.Manager.Suspend 2014-06-30 08:31:10 iirc the login manager is systemd 2014-06-30 08:31:43 ... 2014-06-30 08:32:10 is there any fix/workaround? 2014-06-30 08:32:48 i suppose its possible to hack something to execute pm-utils commands 2014-06-30 08:33:29 but the proper way would be implement the dbus interface for logind i suppose (which also would allow us to run gnome iirc) 2014-06-30 08:35:24 maybe lightdm implements it? 2014-06-30 08:38:42 Sorry for interruption, I'll repeat yesterdays question if you don't mind. I'm bit buffled how to set 'ulimit -l' (RLIMIT_MEMLOCK) during boot time for specific user or system wide - desktop system. Would appreciate any hints. 2014-06-30 08:41:02 melville: you can set rc_ulimit='-l unlimited' in /etc/rc.conf 2014-06-30 08:41:06 for system wide 2014-06-30 08:42:27 Tried it already with rc_ulimit='-l 256', still ulimit -a shows it is 64. 2014-06-30 08:43:38 hm.. maybe it only affects the services? 2014-06-30 08:43:54 I think so, at least comments say so. 2014-06-30 08:44:25 it is under "# SERVICE CONFIGURATION VARIABLES" 2014-06-30 08:45:08 I tried also script in local.d, but that was long shot and it seems just wrong. 2014-06-30 08:45:47 hm 2014-06-30 08:45:59 limits.conf seems to be a pam thingy 2014-06-30 08:46:37 Yeah, and pam isn't used by default - been that route too /etc/security/limits.conf. 2014-06-30 08:47:28 Will try rc_ulimit with unlimited in a minute and give it reboot. I'm not in hurry. 2014-06-30 08:47:48 its the first time someone asks for it for a user... 2014-06-30 08:47:59 normally it has been question of ulimits for services 2014-06-30 08:48:17 i suppose it depends on functionality in busybox then 2014-06-30 08:49:30 Grsec is spamming my dmesg with denial of resource oversteps - firefox and such. 2014-06-30 08:49:47 hm 2014-06-30 08:49:51 firefox works for me 2014-06-30 08:50:21 I use it in safe mode, since otherwise it crashes a lot for me. 2014-06-30 08:50:43 it looks like setting ulimits for users is not implemented :-( 2014-06-30 08:50:48 i cannot find any way to do it 2014-06-30 08:50:55 melville: is it x86 or x86_64? 2014-06-30 08:51:13 x86_64 2014-06-30 08:51:23 how much memory does your box have? 2014-06-30 08:52:39 1G - I had plans to upgrade, but so far it is enough for me. 2014-06-30 08:52:58 it should be enough... 2014-06-30 08:54:21 I use "firefox -safe-mode" on daily basis and even with gst plugins for h.264 it stable as it gets. 2014-06-30 08:54:44 Don't think I had single crash in safe mode. 2014-06-30 08:55:26 hey, I used this guide to setup sensors http://wiki.alpinelinux.org/wiki/Setting_up_lm_sensors , but I get 2014-06-30 08:55:27 localhost:~# sensors-detect 2014-06-30 08:55:27 No i2c device files found. 2014-06-30 08:55:40 anyone played with sensors in alpine? for configuring fan speeds and such? 2014-06-30 08:56:18 maybe the driver is missing 2014-06-30 08:56:29 dalias: i think i found a workaround for lid -> suspend 2014-06-30 08:56:46 mkdir -p /etc/acpid/LID 2014-06-30 08:57:43 echo -e "#!/bin/sh\nexec pm-suspend" > /etc/acpid/LID/00000080 2014-06-30 08:57:57 sensors works for me, but I don't have needs beside procfs: http://pastebin.com/8xVkD2jP 2014-06-30 08:57:57 chmod +x /etc/acpid/LID/00000080 2014-06-30 08:58:52 no, its /etc/acpi/LID/00000080 2014-06-30 09:01:32 Then again, it is Thinkpad, so most things works just out of box - I didn't have to modprobe anything here for sensors. 2014-06-30 09:02:03 it could be that the needed i2d driver is not compiled, but i dont know really 2014-06-30 09:02:53 I have i2c-dev loaded like it says 2014-06-30 09:03:01 hmm 2014-06-30 09:04:24 I would check dmesg, if anything is detected. 2014-06-30 09:05:58 or lsmod for something like "hwmon 3804 1 thinkpad_acpi" 2014-06-30 09:08:13 no there is nothing 2014-06-30 09:11:36 sudo modprobe i2c-dev? or sensors-detect is doing that already? 2014-06-30 09:13:28 I did modprobe i2c-dev and sensors-detect I get nothing 2014-06-30 09:13:40 No i2c device files found. 2014-06-30 09:15:19 dalias: http://wiki.alpinelinux.org/wiki/Suspend_on_LID_close 2014-06-30 09:17:59 ncopa, any idea for the keyboard button? 2014-06-30 09:18:02 it's not as important 2014-06-30 09:23:38 dalias: can you run 'acpid -f -d' and see if it generates any acpid events? 2014-06-30 09:24:24 ncopa: that's cool is there way to just turn off the screen aswell ? setterm -powersave powerdown dosen't do it 2014-06-30 09:24:40 not to put the whole machine to sleep just turn off display 2014-06-30 09:25:10 zxd: i dont know. on my laptop it does turn off the screen when i close the lid 2014-06-30 09:25:35 if you find a command to turn off the screen, then you could use that in your /etc/acpi/LID/00000080 instead 2014-06-30 09:27:55 ncopa: it turns off the lid but does the laptop go to sleep? 2014-06-30 09:28:00 I don't want it to goto sleep 2014-06-30 09:28:11 ncopa: maybe it's doing a full suspend 2014-06-30 09:32:59 I noticed there is no crda package for wireless, other than limitng the channels and transmit power there isn't a real use for it correct? 2014-06-30 09:33:23 they say without crda it runs with a default restrictive settings 2014-06-30 12:42:38 hi to all, re: firefox 2014-06-30 12:43:02 I was able to install firefox from nix packager manager 2014-06-30 12:43:43 and it has same symptoms than firefox from alpine 2014-06-30 12:44:11 so musl is not guilty :) 2014-06-30 12:45:09 Has somebody tried with vanilla kernel? 2014-06-30 12:48:11 Forget it, grsec kernel is not either... 2014-06-30 12:48:47 they has same issue on NixOS: https://github.com/NixOS/nixpkgs/issues/2950 2014-06-30 12:49:36 they downgraded to 29.0.1 2014-06-30 12:50:11 interesting 2014-06-30 12:50:47 what libc is nixOS using? 2014-06-30 12:51:05 glibc 2.19 2014-06-30 12:51:22 the its an upstream bug 2014-06-30 12:54:53 then* 2014-06-30 12:56:28 hmm, but on Arch it's working... 2014-06-30 12:57:36 is there some way to look at the config options for nixos? 2014-06-30 12:57:44 the config options for firefox on nixos 2014-06-30 12:58:06 https://github.com/NixOS/nixpkgs/blob/e66d3953ae79c478b5d16bc5cccd5baaeb7a039b/pkgs/applications/networking/browsers/firefox/default.nix 2014-06-30 13:00:22 Arch has this: https://projects.archlinux.org/svntogit/packages.git/tree/trunk/mozconfig?h=packages/firefox 2014-06-30 13:01:37 and this patch: https://projects.archlinux.org/svntogit/packages.git/tree/trunk/firefox-install-dir.patch?h=packages/firefox 2014-06-30 13:03:26 i think what triggers it is when libxulrunner is built separately 2014-06-30 13:05:10 are there other distros that build xulrunner separately? 2014-06-30 13:05:59 debian as far as I know 2014-06-30 13:06:47 but they has firefox 29 2014-06-30 13:07:42 *iceweasel 2014-06-30 13:10:35 no, sorry, debian sid has version 30 2014-06-30 13:13:01 doesn't seem to depend on xulrunner: https://packages.debian.org/jessie/iceweasel 2014-06-30 13:13:42 fedora switched to standalone firefox on somewhere around 26 2014-06-30 13:15:35 hmm, ovf is right 2014-06-30 16:23:06 anyone also see an instapanic with flashcache in writeback mode? 2014-06-30 17:56:06 hi, domain and search in /etc/resolv.conf seems to have no effect in alpine linux? I've tried to add search mydomain.tld to /etc/resolv.conf but ping reports 'bad address' instead of a lookup and ping 2014-06-30 18:47:38 someone with ipv6 on alpine linux here? 2014-06-30 18:48:58 it works very well for me, but I'm confused about this route entry: "unreachable default dev lo metric -1 error -101" ... what is it? why is it there? it's actually two times in the output of ip -6 route 2014-06-30 19:16:53 btrfs-progs is broken on alpine? 2014-06-30 19:19:13 strace shows all stats returning ENOENT 2014-06-30 19:19:18 thus broken 2014-06-30 20:43:19 hi I noticed with command free that there is no cached column, I assume because alpine loads all programs to ram, how does it do it exactly? 2014-06-30 20:43:27 is it how the application was compiled? 2014-06-30 21:00:03 in what package is vmstat 2014-06-30 21:04:08 zxd: sysstat normally 2014-06-30 21:05:27 nope 2014-06-30 21:06:14 oh ... procps 2014-06-30 21:06:59 zxd: it's procps ... sorry 2014-06-30 21:14:32 server irc.mozilla.org 2014-06-30 21:14:35 ... 2014-06-30 21:20:30 thanks ! 2014-06-30 21:21:01 btw wanted to update closing my laptop lid does put screen to sleep goes blank 2014-06-30 21:21:08 so it's good 2014-06-30 21:22:13 brb 2014-06-30 21:22:19 Hmm, I'm wondering why procps `top` reports 0% utilization. 2014-06-30 21:22:31 htop and busybox seem to work just fine