2016-01-02 18:18:50 <nmeum> by the way: happy new year everyone
2016-01-02 18:18:56 <nmeum> is ncopa still on vacation?
2016-01-02 18:31:18 <tdtrask> ncopa's still not back
2016-01-03 01:44:44 <nmeum> should /usr/lib/*.so.[0-9] files always be splited into a -libs subpackage?
2016-01-03 01:49:08 <nmeum> and if I merge a patch and make slight modifications to it should I rebase my changes and the original patch into one commit or should I just ask the author of the patch to incorporate my changes?
2016-01-04 09:16:33 <leo-unglaub> ey :)
2016-01-04 09:16:40 <leo-unglaub> oh, there is an H missing
2016-01-04 09:16:43 <leo-unglaub> i meant Hey
2016-01-04 09:16:45 <leo-unglaub> :)
2016-01-04 09:27:04 <_mjones> Hello.
2016-01-04 09:46:21 <skarnet> Happy new year everyone \o/
2016-01-04 11:49:08 <ScrumpyJack> morning all. Happy Newness
2016-01-04 12:04:00 <shafire> happy new year
2016-01-04 12:04:14 <shafire> what will 2016 us bring?
2016-01-04 12:05:05 <ScrumpyJack> Alpine Happiness
2016-01-04 12:05:31 <skarnet> 2016, the year of alpeen
2016-01-04 12:06:28 <shafire> bring us*
2016-01-04 12:14:18 <barthalion> alpineess
2016-01-04 12:15:35 <skarnet> On the 1st day of Christmas, Santa gave to me: a brand new 3.3
2016-01-04 12:16:54 <shafire> yes, broken 3.3 :(
2016-01-04 12:18:08 <coredumb> Happy new year :)
2016-01-04 12:26:58 <clandmeter> 
ACTION just got his new seiko watch called Seiko Alpinist :)

2016-01-04 12:31:28 <fabled> clandmeter, got a link?
2016-01-04 12:31:28 <fabled> 
ACTION likes Seiko watches too.

2016-01-04 12:31:45 <fabled> i've been waiting for a nice looking Seiko Direct Drive watch to come out...
2016-01-04 12:32:04 <fabled> err... sorry. Spring Drive
2016-01-04 12:32:21 <clandmeter> yes its automatic
2016-01-04 12:32:50 <clandmeter> fabled: https://s-media-cache-ak0.pinimg.com/736x/de/41/f3/de41f30740c964024a2b18d1b77a2009.jpg
2016-01-04 12:32:51 <coredumb> what does "superseded" means in patchwork ?
2016-01-04 12:33:10 <fabled> coredumb, superseded = new patch was sent on the list that is better then the one you are looking
2016-01-04 12:33:55 <fabled> clandmeter, nice!
2016-01-04 12:33:58 <clandmeter> fabled: its japanese domestic, so its a bit hard to get.
2016-01-04 12:34:23 <clandmeter> http://www.amazon.com/Seiko-Mechaical-Alpinist-Automatic-Sarb017/dp/B000KG93BQ
2016-01-04 12:34:31 <clandmeter> but i dont think amazon will ship to eu.
2016-01-04 12:34:42 <coredumb> fabled: Ok great :)
2016-01-04 12:34:46 <coredumb> thx
2016-01-04 12:35:57 <coredumb> clandmeter: lately I was looking into maratac for a cheap mechanical watch
2016-01-04 12:36:52 <clandmeter> coredumb: did you checkout massdrop?
2016-01-04 12:37:11 <clandmeter> i got mine from massdrop.com
2016-01-04 12:38:03 <clandmeter> shipping was only 10 euro but Tax and import duty was 75 euro :|
2016-01-04 12:38:11 <coredumb> clandmeter: yep I'm a massdroper ^^
2016-01-04 12:38:22 <coredumb> clandmeter: no wonder about that :D
2016-01-04 12:38:43 <coredumb> friend of mine got two of the previous seiko drops
2016-01-04 12:39:04 <coredumb> I must say that they're quite OK for the price tag
2016-01-04 12:39:38 <clandmeter> fabled: massdrop currently has Seiko Cocktail Time SARB065, which is also very nice (except not alpinist :) )
2016-01-04 12:40:23 <coredumb> clandmeter: http://www.countycomm.com/sr3mid.html this one before I can stash enough for a stowa ^^
2016-01-04 12:41:02 <clandmeter> nice clean design
2016-01-04 12:41:45 <coredumb> http://www.stowa.de/cosmoshop/pix/a/z/fliegerunitas6498kleinesekunde/g/Flieger_Unitas_6498_kl_Orig.jpg < then this one
2016-01-04 12:41:57 <coredumb> but wife's gonna kill me then :D
2016-01-04 12:42:28 <barthalion> I need to buy something with day of the week
2016-01-04 12:42:50 <barthalion> calendar has ended in the one I got from grandfather
2016-01-04 12:42:50 <clandmeter> barthalion: visodate is nice
2016-01-04 12:43:16 <barthalion> fancy
2016-01-04 12:43:21 <barthalion> you got me now :D
2016-01-04 12:43:32 <clandmeter> and your wallet :p
2016-01-04 12:43:48 <barthalion> yeah
2016-01-04 12:44:22 <clandmeter> i think its around 500 in europe
2016-01-04 12:44:26 <coredumb> oh well tissot is still OK price :D
2016-01-04 12:44:28 <barthalion> "free shipping from 500zł" not so hard to achieve if something costs around 2000
2016-01-04 12:45:55 <coredumb> for 2k+ you have way more interesting brands than tissot imho
2016-01-04 12:46:37 <coredumb> I've always found Tissot lacking of "spirit" ... A bit too classical I guess
2016-01-04 12:47:18 <clandmeter> i think he talks about zlotties or however they are called.
2016-01-04 12:48:08 <coredumb> rofl
2016-01-04 12:48:51 <clandmeter> zlotys almost correct :D
2016-01-04 12:48:55 <coredumb> :D
2016-01-04 12:49:16 <barthalion> yeah, PLN
2016-01-04 12:49:23 <barthalion> zloties is good enough :P
2016-01-04 12:50:29 <coredumb> at least google found what it was about :P
2016-01-04 12:53:54 <skarnet> you guys should write for... (wait for it)
2016-01-04 12:53:54 <skarnet> ... distro-watch
2016-01-04 12:54:03 <skarnet> (ba-dum ksh!)
2016-01-04 13:00:16 <barthalion> :D
2016-01-04 13:27:07 <shafire> is it possible to get an option for the compression in the iso generation?
2016-01-04 13:38:00 <shafire> the modloop compression
2016-01-04 13:55:41 <fabled> shafire, http://git.alpinelinux.org/cgit/alpine-iso/tree/Makefile#n111
2016-01-04 13:59:04 <shafire> fabled: I see
2016-01-04 13:59:08 <shafire> I change always line 111
2016-01-04 13:59:10 <shafire> 	@$(MKSQUASHFS) $(MODLOOP_DIR)/lib $@ -comp xz
2016-01-04 14:03:08 <fabled> but yeah the image creation makefile/scripts could be improved. we had some ideas
2016-01-04 14:03:13 <fabled> no doer yet
2016-01-04 15:26:45 <ScrumpyJack> does anyone remember a problem with mysql on 2.6? I can't start it: Can't start server : Bind on unix socket: Permission denied
2016-01-04 15:30:53 <ScrumpyJack> permissions look fine. I'm in a LXC container running 2.6 :)
2016-01-04 15:37:05 <ScrumpyJack> found it. /var had mod 000
2016-01-04 15:49:14 <ScrumpyJack> clandmeter: could you bump lxd to 0.25? i think the pacakge has your name on it
2016-01-04 17:43:56 <laurentius77> Have anyone here Alpinelinux workin on RPi zero? Pls...
2016-01-04 19:57:05 <laurentius77> Please can you add Rpi zero support boot in next 3.3.1 Alpine image ?
2016-01-04 19:58:10 <_mjones> Yes, someone should put that in the bug tracker if it's not already there.
2016-01-04 20:28:57 <laurentius77> I added this request in bug track, as a feature. Thanks to developers for their great work in all this time
2016-01-04 20:32:51 <fabled> i'm not sure what the issue is, but 3.3.1 will have new firmware so chances are that it should work
2016-01-04 20:32:58 <fabled> they say the b+ overlay should work with zero
2016-01-04 20:33:14 <fabled> so i'm hopeful it's just matter of that
2016-01-04 20:33:24 <fabled> i hope we get 3.3.1 out this week
2016-01-04 20:33:30 <fabled> now that ncopa is back too
2016-01-04 21:02:04 <laurentius77> Thanks
2016-01-04 21:02:21 <laurentius77> welcome back to ncopa
2016-01-04 21:03:12 <laurentius77> Rpi zero is a great piece of hardware for embedded world
2016-01-04 21:05:35 <mitchty> seeing http://retrobsd.org/wiki/doku.php run on a pic32 impressed me a lot more if i'm honest :)
2016-01-04 21:39:05 <laurentius77> indeed...
2016-01-05 10:11:24 <ncopa> does anyone has idea what the problem was with apk cache? #4975
2016-01-05 10:11:35 <ncopa> what component is broken
2016-01-05 10:11:42 <ncopa> how do we fix it
2016-01-05 10:36:11 <ScrumpyJack> i saw this behaviour before on a usb stick with 3.2
2016-01-05 10:38:21 <ScrumpyJack> not very helpful, sorry
2016-01-05 11:17:27 <shafire> hi
2016-01-05 11:21:07 <shafire> ncopa: http://git.alpinelinux.org/cgit/mkinitfs/commit/initramfs-init.in?id=6aba975e73e7cea833640e8db180c0add01ecf33
2016-01-05 11:21:37 <shafire> maybe these changes are responsible for the behaviour of #4975
2016-01-05 15:15:06 <nmeum> fabled: is there any reason why you merged the libsodium upgrade (02a69d01fc9fe0becacd693cc4492d59a332b517) without merging the package rebuilds needed because the upgrade results in an soname bump?
2016-01-05 15:15:45 <nmeum> %1306, %1307, …
2016-01-05 15:24:49 <chris|> nmeum: imo the comment made for %1306 is accurate
2016-01-05 15:25:18 <nmeum> ah
2016-01-05 15:25:20 <nmeum> didn't see that
2016-01-05 16:47:10 <dowdle> Greetings from an Alpine Linux outsider.  I'm looking for an Alpine dev that goes by the IRC nick shafire.  shafire has been working on an Alpine-based LiveCD running the OpenVZ kernel and I've been doing some testing for him... but I don't know shafire's email address and I don't see shafire online today.  If someone could PM me the email address, I'd appreciate it.
2016-01-05 16:51:51 <ncopa> dowdle: he was here earlier today
2016-01-05 17:01:07 <dowdle> ncopa: I realize the reticense (sp?) to give out email addresses to strangers... and I'm here during work hours so I'll just keep an eye out for shafire.  Thanks.
2016-01-05 17:07:50 <skarnet> reluctance?
2016-01-05 17:14:48 <dowdle> skarnet: That is a synonym for the word... that appears to be spelled "reticence".  I had spell check turned on in my IRC program (Konversation) but no language selected.  Selecting a language makes it work now. :)
2016-01-05 17:16:11 <skarnet> ha! I thought réticence was only the French word. TIL it exists in English too. :)
2016-01-05 17:21:12 <ScrumpyJack> skarnet: I'm intrigued but the stuff dowdle was saying. What's this LiveCD you're working on?
2016-01-05 17:21:55 <skarnet> not me. shafire.
2016-01-05 17:23:11 <dowdle> ScrumpyJack: shafire is working on it.  Basically it is an extremely minimal LiveCD (about 100MB) that uses the stock OpenVZ kernel by default.  It isn't totally done yet, but it isn't that far.  You can find the last two builds here - http://dev.alpinelinux.org/~shafire/
2016-01-05 17:24:02 <skarnet> > extremely minimal
2016-01-05 17:24:06 <skarnet> > 100 MB
2016-01-05 17:24:10 <skarnet> does not compute
2016-01-05 17:24:24 <dowdle> There is a short quickstart text file with basic instructions.  The latest build doesn't have mkfs.ext4 so...
2016-01-05 17:24:33 <dowdle> skarnet: So 100MB is normal?  I'm used to media that is much bigger.
2016-01-05 17:24:42 <skarnet> (ok, ok, I get you have a graphical environment and all the shizznit. But still.)
2016-01-05 17:25:21 <dowdle> skarnet: I was completely unfamiliar with Alpine Linux before encountering shafire's OpenVZ-related project... but I guess since Alpine uses BusyBox... it is tiny to begin with, eh?
2016-01-05 17:25:36 <dowdle> skarnet: No, not graphical environment on that LiveCD.
2016-01-05 17:25:37 <ScrumpyJack> that's neat.
2016-01-05 17:25:55 <skarnet> I'd say 100MB is decent, and that you, as almost everyone else, are used to media that is horribly bloated.
2016-01-05 17:26:17 <skarnet> no graphical environment? then 100MB is too much.
2016-01-05 17:26:38 <skarnet> well ofc it depends on what there is on it.
2016-01-05 17:27:57 <skarnet> I remember tomsrtbt. That was a floppy disk. 1.9MB, kernel included. (with horrible compression tricks, but still.)
2016-01-05 17:28:26 <ScrumpyJack> yeah, and loaf
2016-01-05 17:28:33 <dowdle> skarnet: I make a Fedora LiveDVD and it is about 3.5GB. :P~~~
2016-01-05 17:29:07 <skarnet> dowdle: if you're trying to do anything small, stay the hell away from "mainstream" distributions.
2016-01-05 17:29:28 <skarnet> Fedora, Debian, Ubuntu - big names, and big (and bad) software, too.
2016-01-05 17:33:29 <ScrumpyJack>  [alpine] || die
2016-01-05 17:36:25 <_mjones> Alpine base install includes >100mb of kernel modules. More-specialised media can reduce that a lot.
2016-01-05 17:36:35 <_mjones> Alpine base install includes >100mb of kernel modules. More-specialised media can reduce that a lot.
2016-01-05 17:37:25 <dowdle> man not found.  Too small for me.
2016-01-05 17:37:45 <skarnet> Be the bigger man!
2016-01-05 17:39:10 <skarnet> _mjones: ugh, how big can the kernel get? Can you load ALL of these modules and use 100+Mb of kernel memory? >.>
2016-01-05 17:39:13 <skarnet> MB*
2016-01-05 17:40:04 <awilfox> I believe make allyesconfig can be about 700 MB or so on x86_64
2016-01-05 17:40:31 <awilfox> considerably less on arches that have less frilly stuff, think it's 300 MB or so on powerpc, but allyesconfig doesn't make sense even for a generic/distro kernel there
2016-01-05 17:40:44 <skarnet> 
ACTION converts to BSD

2016-01-05 17:41:04 <awilfox> mm, I managed to get NetBSD 6 to fit on a 1.2MB 5.25" floppy.
2016-01-05 17:41:07 <awilfox> kernel + small root
2016-01-05 17:41:27 <skarnet> nice.
2016-01-05 17:41:34 <awilfox> couldn't do that with linux, the smallest kernel I could make was 1.0 MB, and the minimalist root that just had musl + busybox was 300kb
2016-01-05 17:42:01 <skarnet> now you can boot your 386 with it, the only machine that still has a 5.25" floppy drive.
2016-01-05 17:42:06 <skarnet> :P
2016-01-05 17:42:27 <awilfox> http://blog.foxkit.us/2015/07/foxtoo-gentoo-musl-c-library-on-100mhz.html
2016-01-05 17:42:43 <awilfox> skarnet: my pentium 133MHZ desktop has a CD, DVD, 3.5" floppy, and 5.25" floppy
2016-01-05 17:42:48 <awilfox> skarnet: it does everything!
2016-01-05 17:43:14 <skarnet> heh. I had one until a few years ago, but the mobo died. ; ;
2016-01-05 17:44:36 <skarnet> I'm surprised you couldn't get busybox to less than 200k by cutting apps.
2016-01-05 17:45:11 <skarnet> At least, you could make one with just a shell and init=/bin/sh. It would boot.
2016-01-05 17:45:23 <skarnet> You wouldn't be able to do much with it, but... XD
2016-01-05 17:54:46 <ammunta> okay but at that poin you could also just not take input :&&
2016-01-05 18:06:45 <ScrumpyJack> maybe there is a demand for an alpine linux kernel with, like, 3 modules or something
2016-01-05 18:07:43 <ScrumpyJack> what do ting core do?
2016-01-05 18:08:22 <ScrumpyJack> oo oo oo, Alpine BSD
2016-01-05 18:47:15 <dowdle> Here's a demo video of shafire's livecd in action... from start to finish... about 11 minutes (no audio).  Amazingly, it's only a 1.7MB .webm file... so you have to be able to play vp9-based webm files... but contemporary Firefox and Google Chrome play it just fine - https://www.montanalinux.org/files/videos/alpine-based-openvz-livecd-demo.webm
2016-01-05 18:48:23 <dowdle> Obviously ffmpeg and webm/vp9 are amazing with compression of that genre of video.  > 480p at good quality... 11 minutes ~ 1.7MB.
2016-01-06 08:21:31 <clandmeter> morning climbers
2016-01-06 09:30:54 <ScrumpyJack> morning clandmeter
2016-01-06 09:31:32 <ScrumpyJack> hey if you have a moment, would you mind awfully bumping LXD package? (i don't mind doing it if you like)
2016-01-06 09:54:08 <clandmeter> ScrumpyJack: hi
2016-01-06 09:54:12 <clandmeter> let me take a look
2016-01-06 09:54:19 <clandmeter> iirc its go
2016-01-06 10:02:11 <clandmeter> ScrumpyJack: it has issues with gettext
2016-01-06 10:02:20 <clandmeter> ScrumpyJack: if you can get a patch, i can commit it.
2016-01-06 10:09:31 <clandmeter> ScrumpyJack: fix it. pusing now.
2016-01-06 10:13:23 <clandmeter> still not preferd way to add go apps to aports, if it ever would be moved to main/community it should have a tarball will all sources instead of using git to check them out.
2016-01-06 10:56:16 <clandmeter> hmm sprunge over quota again
2016-01-06 11:14:21 <shafire> I see at every login: "Kernel … on an x86_64" <- Is the an here correct?
2016-01-06 11:24:54 <ScrumpyJack> clandmeter: thanks dude
2016-01-06 11:25:43 <skarnet> shafire: it is phonetically correct, I think it's what counts
2016-01-06 12:18:44 <shafire> do you use "compile issues", "compiling issues" or "compilation issues"?
2016-01-06 12:59:45 <barthalion> all of them are clear, probably the last one is the best
2016-01-06 13:34:29 <laurentius77> Hello
2016-01-06 13:35:24 <laurentius77> I have an issue when trying to use ppp with a HSDPA modem
2016-01-06 13:35:43 <laurentius77> Jan  6 15:28:35 localhost daemon.err pppd[1952]: Couldn't open the /dev/ppp device: No such file or directory
2016-01-06 13:36:01 <laurentius77> Jan  6 15:28:35 localhost daemon.err pppd[1952]: You need to create the /dev/ppp device node by executing the following command as root: 	mknod /dev/ppp c 108 0
2016-01-06 13:36:36 <laurentius77> after issuing the recommended comand itțs working
2016-01-06 13:37:01 <laurentius77> it is working
2016-01-06 13:37:37 <laurentius77> tried modprobe ppp and says that ppp is not in modules.dep
2016-01-06 13:38:28 <laurentius77> Alpine version is 3.2.3, I tried on x86 and armhf, the same problem
2016-01-06 13:40:58 <laurentius77> Could be a mdev problem?
2016-01-06 13:43:42 <laurentius77> Or it is ppp problem? Should I write a bug ticket as a ppp issue?
2016-01-06 14:50:35 <shafire> so the mistake was the patch in mkinitfs?
2016-01-06 16:40:30 <BitL0G1c> the tinyssh / quicktun / fastd / dnscrypt patches can all be rejected
2016-01-06 16:59:40 <shafire> where can I find the repo of the website?
2016-01-06 16:59:57 <shafire> ah it's mksite
2016-01-06 17:28:23 <barthalion> shafire: some dowdle man was looking for you
2016-01-06 17:28:37 <shafire> barthalion: thanks, talked with him already
2016-01-06 17:28:56 <barthalion> k, np
2016-01-07 12:56:07 <ScrumpyJack> is alacerda still around?
2016-01-07 15:33:50 <joe42> Hello all. I installed PECL from apk and it complains that it can't find XML extension (which is also installed).  Any advice?
2016-01-07 16:40:19 <skarnet> grats on 3.3.1
2016-01-07 17:19:09 <nmeum> ncopa: since 3.3 has been released would you mind taking another look at my abuild patches from last year? Currently the man page hardlink rewrite still doesn't work under btrfs
2016-01-07 17:30:40 <nmeum> jomat: what is PECL?
2016-01-07 19:31:19 <jomat> nmeum: Why?
2016-01-07 19:31:27 <jomat> First hit says: "PHP Extension Community Library"
2016-01-07 19:31:39 <jomat> Oh... /me reads backlog :-)
2016-01-07 19:31:48 <nmeum> oh, sorry
2016-01-07 19:31:55 <nmeum> tabbing fail
2016-01-07 19:32:00 <jomat> Yeah:-)
2016-01-07 22:19:57 <jirutka> @ncopa: Hi, I’m happy to offer you another Alpine mirror – https://repository.fit.cvut.cz/mirrors/alpine/ – it’s located in Czech Republic, on fast academic network, uses trusted TLS certificate, synced every hour
2016-01-07 22:41:47 <ncopa> jirutka: very nice. I 'll add it tomorrow
2016-01-07 22:41:51 <ncopa> thanks!
2016-01-07 22:42:02 <jirutka> you’re welcome!
2016-01-08 08:43:38 <shafire> hi
2016-01-08 08:45:20 <shafire> where can I find the rules which aport patches are acceptable?
2016-01-08 09:50:09 <ScrumpyJack> what do you mean by "which aport patches are acceptable?". How to make an acceptable patch?
2016-01-08 10:02:20 <shafire> yes
2016-01-08 10:59:42 <shafire> http://pastebin.com/56YwMgy5
2016-01-08 10:59:46 <shafire> what's wrong here?
2016-01-08 11:04:43 <skarnet> linux-2.6.32
2016-01-08 11:04:49 <skarnet> that's what's wrong :P
2016-01-08 11:05:24 <fcolista> :)
2016-01-08 11:07:17 <shafire> joke aside
2016-01-08 11:10:14 <skarnet> honestly I don't know why a header would be missing. Lack of asm/ link to the proper arch-specific asm directory maybe?
2016-01-08 11:10:58 <skarnet> (at the time of 2.6.32, those scripts were still a bit flakey)
2016-01-08 11:32:38 <shafire> I don't know
2016-01-08 11:32:53 <shafire> I have no clue
2016-01-08 12:38:31 <ncopa> nmeum: is %1358 ok?
2016-01-08 12:45:50 <skarnet> lgtm, but I'm not nmeum :)
2016-01-08 12:47:14 <ncopa> i ask because he is listed as maintainer
2016-01-08 12:47:33 <skarnet> ofc
2016-01-08 13:48:13 <shafire> skarnet: forgot the headers, but still the flags error is there http://pastebin.com/2Mem3d6z
2016-01-08 13:54:00 <skarnet> shafire: don't look at me for kernel debugging XD
2016-01-08 17:04:04 <mitchty> does build-base get added by default if makedepends is present in APKBUILD files
2016-01-08 17:04:21 <mitchty> trying to figure out how gcc is getting installed when I haven't requested it
2016-01-08 17:08:36 <mitchty> well that answers that, ok so how could one go about requesting that gcc/g++ aren't installed implicitly, i'm trying to build against clang but autoconf can detect that gcc is there and use it, rather not have to patch configure but hmm
2016-01-08 17:08:47 <nmeum> ncopa: Looks good on a first glance I will test it tomorrow
2016-01-08 17:21:49 <nmeum> mitchty: most build systems respect the CC environment variable just set it CC to clang
2016-01-08 17:22:21 <mitchty> nmeum oh i tried that already :)
2016-01-08 17:22:37 <mitchty> digging through m4 to find out what i need to patch away
2016-01-08 17:23:09 <mitchty> would be easier if it wasn't installed at all though
2016-01-08 17:23:25 <mitchty> or if there were a way to say don't install base-build
2016-01-08 19:48:28 <xen_roger> Hi guys, when I try to create a new iso-image on 3.3 I get "/bin/sh: isohybrid: not found" and I also could not found any pkg... Did I miss something ?
2016-01-08 19:50:04 <xen_roger> shame on me ... syslinux was missing.
2016-01-08 19:50:10 <xen_roger> ok, so solved xD
2016-01-08 20:36:35 <jirutka> @ncopa: Hi, have you added https://repository.fit.cvut.cz/mirrors/alpine/ to the mirrors list? I don’t see it in http://rsync.alpinelinux.org/alpine/MIRRORS.txt.
2016-01-08 20:47:46 <shafire> yes
2016-01-08 20:47:48 <shafire> he did
2016-01-08 20:48:02 <jirutka> where?
2016-01-08 20:48:31 <shafire> http://git.alpinelinux.org/cgit/aports/commit/?id=dca087c8bcbb5a286725b8c244cfd25c3ddc76e8
2016-01-08 20:49:25 <jirutka> Well, but why it’s not in http://rsync.alpinelinux.org/alpine/MIRRORS.txt (referenced in http://wiki.alpinelinux.org/w/index.php?title=Alpine_Linux:Mirrors)
2016-01-08 20:49:27 <jirutka> ?
2016-01-08 20:49:51 <shafire> I don't know for what the .txt is used
2016-01-08 20:51:02 <jirutka> I used it in new LXC template for Alpine Linux (https://github.com/lxc/lxc/pull/751), so if this txt file is deprecated, I should probably use another source for mirrors. :)
2016-01-08 20:54:06 <xen_roger> When I try to commit with lbu to an usb-stick created with dd ( hybrid-iso v3.3 ) I got an error that it cannot mount /dev/usbstick - is lbu working with the isohybrid stick ?
2016-01-08 20:54:27 <jirutka> Maybe I should fetch http://git.alpinelinux.org/cgit/aports/plain/main/alpine-mirrors/mirrors.yaml instead of the txt file and grep URLs from it.
2016-01-08 20:57:11 <shafire> or tell that here tomorrow again
2016-01-08 20:57:14 <shafire> they will add it then
2016-01-08 20:57:24 <jirutka> okay
2016-01-08 21:13:09 <skarnet> or not
2016-01-08 21:13:12 <skarnet> mt XD
2016-01-09 15:51:37 <jirutka> Does anyone know why build logs are not available for newer packages? It results to HTTP 404.
2016-01-09 15:52:01 <nmeum> clandmeter: would you mind if I move pulseaudio from testing/ to unmaintained/? The version is outdated, it has broken dependencies and it doesn't even build here anymore. Furthermore, it doesn't have a maintainer…
2016-01-09 15:52:08 <jirutka> correction: not newer, but some of
2016-01-09 15:52:16 <nmeum> jirutka:
2016-01-09 15:52:16 <nmeum> 04:24 -!- Frosh [uid4206@gateway/web/irccloud.com/x-duruqlamsgpsrapf] has quit [Ping timeout: 260 seconds]
2016-01-09 15:52:19 <nmeum> 04:26 -!- ovf [sid19068@gateway/web/irccloud.com/x-twficqbsimduhkrf] has quit [Ping timeout: 260 seconds]
2016-01-09 15:52:22 <nmeum> 04:26 -!- Frosh [uid4206@gateway/web/irccloud.com/x-jwlmmygfztcnjzkd] has joined #alpine-devel
2016-01-09 15:52:25 <nmeum> 04:30 -!- ovf [sid19068@gateway/web/irccloud.com/x-sdfsebqjuydxrged] has joined #alpine-devel
2016-01-09 15:52:28 <nmeum> 04:44 -!- blueness [~blueness@cpe-74-77-145-97.buffalo.res.rr.com] has quit [Ping timeout: 255 seconds]
2016-01-09 15:52:31 <nmeum> 05:05 -!- jirutka [~jirutka@95.85.215.48] has joined #alpine-devel
2016-01-09 15:52:34 <nmeum> 08:03 -!- blueness [~blueness@cpe-74-77-145-97.buffalo.res.rr.com] has joined #alpine-devel
2016-01-09 15:52:37 <nmeum> 09:00 -!- blueness [~blueness@cpe-74-77-145-97.buffalo.res.rr.com] has quit [Changing host]
2016-01-09 15:52:40 <nmeum> 09:00 -!- blueness [~blueness@gentoo/developer/blueness] has joined #alpine-devel
2016-01-09 15:52:43 <nmeum> 10:53 -!- StarWarsFan|afk is now known as StarWarsFan
2016-01-09 15:52:45 <nmeum> 11:13 -!- yth [~mark@unaffiliated/ytheleus] has joined #alpine-devel
2016-01-09 15:52:48 <nmeum> 11:18 -!- shafire [2573d886@gateway/web/freenode/ip.37.115.216.134] has quit [Ping timeout: 252 seconds]
2016-01-09 15:52:51 <nmeum> 11:21 -!- shafire [2573d886@gateway/web/freenode/ip.37.115.216.134] has joined #alpine-devel
2016-01-09 15:52:54 <nmeum> 11:28 -!- skarnet [~ska-skawa@ip-62.net-82-216-6.versailles2.rev.numericable.fr] has joined #alpine-devel
2016-01-09 15:52:57 <nmeum> 14:25 -!- bkleef [~bkleef@s5596adec.adsl.online.nl] has joined #alpine-devel
2016-01-09 15:52:57 <jirutka> what the hell is going on?
2016-01-09 15:53:00 <nmeum> oops
2016-01-09 15:53:03 <nmeum> sorry
2016-01-09 15:53:05 <nmeum> jirutka: http://dev.alpinelinux.org/buildlogs/
2016-01-09 15:53:07 <nmeum> copy & paste fail
2016-01-09 15:53:14 <nmeum> sorry about that
2016-01-09 15:53:57 <jirutka> how it should help me? I see some logs here, but it doesn’t answer my question
2016-01-09 15:54:36 <nmeum> you were looking for the buildlog, right?
2016-01-09 15:54:38 <jirutka> for example, https://pkgs.alpinelinux.org/package/main/x86_64/python → http://build.alpinelinux.org/buildlogs/build-edge-x86_64/main/python/python-2.7.11-r2.log 404
2016-01-09 15:55:32 <nmeum> buildlogs are not stored for ever
2016-01-09 15:55:45 <jirutka> hm, why not?
2016-01-09 15:56:32 <nmeum> no idea, ncopa might be able to answer that but he isn't here currently
2016-01-09 15:56:59 <jirutka> ok, I’ll ask him when he come online
2016-01-09 16:27:28 <nmeum> clandmeter: nevermind, I managed to fix the pluseaudio package
2016-01-09 20:24:51 <ammunta> hm feels like there's some value in at least keeping all the buildlogs for current versions of packages
2016-01-09 20:35:23 <nmeum> yeah
2016-01-09 22:58:42 <ballen> Taking a look at testing/linux-tools for cpupower. Cpupower is defined as a subpkg, but it should also have a cpupower-doc package for its man pages. How does one specify a $subpkg-doc sub-sub-package?
2016-01-09 23:38:04 <nmeum> ballen: currently the convention is that there should only be one -doc subpkg per package
2016-01-09 23:38:12 <ballen> gotcha
2016-01-09 23:38:37 <ballen> so in that case I should make a dedicated cpupower package
2016-01-09 23:38:54 <ballen> The replaces linux-tools-cpupower
2016-01-09 23:41:49 <nmeum> why?
2016-01-09 23:44:22 <ballen> because cpupower needs a -doc and -lang sub-package
2016-01-09 23:45:07 <ballen> and linux-tools is only creating a cpupower subpackage currently, so its redundant
2016-01-09 23:45:46 <nmeum> the manpages of the cpupower subpackage should be contained in the linux-tools -doc subpackage, but I just noticed that linux-tools doesn't have a -doc subpackage
2016-01-09 23:46:32 <ballen> is there a reason to have the meta linux-tools package at all? Instead of seperating out cpupower from any future linux tool like perf, etc
2016-01-09 23:46:43 <ballen> seems overly complicated
2016-01-09 23:46:50 <nmeum> looking at the package I am wondering as well
2016-01-09 23:47:01 <nmeum> it doesn't have a maintainer but clandmeter seems to have added it, ask him :)
2016-01-09 23:47:14 <ballen> indeed
2016-01-09 23:47:30 <ballen> I may just submit a patch for review
2016-01-09 23:47:41 <nmeum> that could work as well
2016-01-09 23:48:07 <nmeum> if you are interessted in maintaining it you might as well list yourself as the maintainer
2016-01-09 23:48:19 <ballen> hmm
2016-01-09 23:48:21 <ballen> perhaps
2016-01-09 23:48:32 <nmeum> but that's optional
2016-01-09 23:51:07 <ballen> is there any built-in way to reference the pkgver of the linux-grsec package in another packag's APKBUILD?
2016-01-09 23:51:44 <ballen> since cpupower is compiled from linux src it should be compiled against the same source tree
2016-01-09 23:52:03 <ballen> be sweet if it could follow along the linux-grsec package
2016-01-09 23:57:30 <nmeum> kind of: some packages simply source the linux-grsec APKBUILD zfs-grsec for instance http://git.alpinelinux.org/cgit/aports/tree/testing/zfs-grsec/APKBUILD#n17
2016-01-09 23:58:18 <nmeum> it's a rather dirty solution though (imho)
2016-01-10 00:00:03 <ballen> yea thats pretty ugly
2016-01-10 00:00:12 <ballen> functional though
2016-01-10 00:00:55 <ballen> appears zfs-grsec is mainly using it to ensure you're building against a kernel version you're expecting
2016-01-10 00:02:25 <nmeum> yeah
2016-01-10 00:02:42 <nmeum> (not sure if ncopa is happy if you use it for something else)
2016-01-10 00:21:22 <ballen> Does apk have an option I'm not finding to list the contents of a package?
2016-01-10 01:11:17 <ballen> nvm ... apk info -a
2016-01-10 02:18:16 <ballen> in default_dev() of abuild, why does it move the *.so lib links into the dev package? http://git.alpinelinux.org/cgit/abuild/tree/abuild.in#n1496
2016-01-10 02:28:17 <skarnet> because the .so link is only used at link time, so when you develop
2016-01-10 02:29:30 <skarnet> the real libraries (*.so.$major) are directly found by the dynamic linker, which has the soname
2016-01-10 02:29:43 <skarnet> so run-time does not need the .so link
2016-01-10 02:30:27 <ballen> ah cool thanks for the explaination
2016-01-10 21:26:41 <xen_roger> Hi guys, what is the reason for getting  "checking build system type... Invalid configuration `x86_64-alpine-linux-musl': machine `x86_64-alpine-linux' not recognized" ?
2016-01-10 21:35:59 <nmeum> do you get this error while using a GNU autoconf configure script?
2016-01-10 21:38:44 <xen_roger> adding update_config_sub fixed the problem
2016-01-10 21:38:52 <xen_roger> nmeum: ty anyway
2016-01-10 21:42:06 <nmeum> yeah I was about to suggest that
2016-01-11 02:58:44 <_mjones> There are a lot of packages in testing that probably need to be promoted so they're officially accessible by release versions, I think. Is there a criteria for promotion?
2016-01-11 02:59:17 <_mjones> 
ACTION is trying to dev on a stable box instead of running edge.

2016-01-11 07:52:02 <bpiotrowski> _mjones: it's usually because someone requested these and never visited us again to say if it works
2016-01-11 07:52:11 <bpiotrowski> so if someone works for you, feel free to move it
2016-01-11 07:53:45 <_mjones> bpiotrowski: excellent, thanks. I'm not confirming anything yet, but the package in question was sdl2. sdl-1.2.15 is from 2012 and presumably needs to be deprecated sooner than later.
2016-01-11 12:52:35 <fcolista> ncopa, re #5011, do you have any advice about how to test?
2016-01-11 12:53:20 <ncopa> dont know
2016-01-11 12:53:26 <fcolista> ok :)
2016-01-11 13:01:09 <littledude> hi
2016-01-11 13:01:21 <nmeum> ncopa: since 3.3 has been released would you mind taking another look at my abuild patches from last your especially http://lists.alpinelinux.org/alpine-devel/5085.html ?
2016-01-11 13:01:27 <littledude> I'm trying to package hping
2016-01-11 13:02:02 <littledude> but one of the source files has this: `#include <net/bpf.h>`
2016-01-11 13:02:29 <littledude> instead of `#include <pcap/bpf.h>`
2016-01-11 13:02:50 <littledude> what is the correct way to handle that in the APKBUILD?
2016-01-11 13:03:37 <littledude> on alpine (and linux in general I think?) this file is under /usr/include/pcap/bpf.h no /usr/include/net/bpf.h
2016-01-11 13:04:04 <ncopa> littledude: maybe you need linux-headers in makedepends?
2016-01-11 13:04:36 <ncopa> nmeum: i probably shoud, but my time is limited nowdays :-/
2016-01-11 13:05:23 <nmeum> alright, no problem
2016-01-11 13:05:39 <nmeum> don't you get paid to work on alpine anymore?
2016-01-11 13:05:48 <littledude> ncopa: thanks,but linux-headers installs /usr/include/linux/bpf.h unfortunately
2016-01-11 13:06:07 <ncopa> nmeum: i get paid to work with alpine
2016-01-11 13:06:23 <ncopa> for you who didnt know: I now work for docker :)
2016-01-11 13:07:25 <nmeum> oh
2016-01-11 13:07:32 <nmeum> nice
2016-01-11 13:07:41 <ncopa> it actually is
2016-01-11 13:11:06 <fcolista> littledude, patch the include file, then
2016-01-11 13:12:17 <ncopa> should probably also be reported upstream
2016-01-11 13:13:10 <littledude> yes there's actually a PR on the hping repo: https://github.com/antirez/hping/pull/13
2016-01-11 13:13:31 <littledude> but it's not maintained anymore so not sure this will get merged.
2016-01-11 13:16:38 <littledude> I'll as you suggest for now, fcolista. thanks for your help guys :)
2016-01-11 13:19:04 <nmeum> hm, do we really need two firefox packages in aports? One in testing/ (the most recent version) and one in community/ (the ESR version) I don't think that we have the man power to maintain two firefox packages
2016-01-11 13:19:59 <nmeum> the version in testing/ is constantly outdated
2016-01-11 13:49:11 <littledude> I'm getting a bunch of the following errors when compiling hping:
2016-01-11 13:49:14 <littledude> /usr/include/pcap/bpf.h:87:9: error: unknown type name 'u_int'
2016-01-11 13:50:11 <littledude> since this file is from libpcap-dev I'm not sure what to do here.
2016-01-11 13:51:13 <littledude> same with u_short, u_char, etc.
2016-01-11 13:54:58 <littledude> sorry seems like I should have googled this one before asking http://stackoverflow.com/questions/15393905/c-pcap-library-unknown-types-error
2016-01-11 14:12:52 <leo-unglaub> hey guys
2016-01-11 14:13:07 <jirutka> what’s up? :P
2016-01-11 15:54:36 <littledude> what should be the $pkgver of a package built out of the master branch of a git repo?
2016-01-11 17:24:18 <_mjones> Speaking of two packages, will openlldp be promoted out of testing and replace open-lldp?
2016-01-11 17:42:03 <barthalion> littledude: is there a tag in that repository?
2016-01-11 17:44:44 <littledude> barthalion: no: https://github.com/antirez/hping
2016-01-11 17:48:22 <barthalion> _mjones: sure, are you awake during CET working hours?
2016-01-11 17:48:25 <barthalion> if so, ping me
2016-01-11 17:48:35 <barthalion> if not, hope I don't forget
2016-01-11 17:48:56 <barthalion> littledude: then something like 0.r$(number of commits).$(hash) will do
2016-01-11 17:52:58 <littledude> oh so then I should checkout a specific commit instead of HEAD
2016-01-11 17:53:04 <littledude> thanks, I'm trying this.
2016-01-11 17:54:04 <barthalion> yeah, pulling HEAD is bad, but there is no easy way to handle that iirc
2016-01-11 17:54:15 <barthalion> some dev needs to put a snapshot on our servers
2016-01-11 17:54:50 <littledude> also would you (or someone else) review my APKBUILD before I submit it?
2016-01-11 17:55:16 <_mjones> barthalion: sometimes I'm around CET. tnx.
2016-01-11 17:55:41 <skarnet> What I would love for 3.4 is a way for apk to specify a git commit to package
2016-01-11 17:56:00 <skarnet> so we wouldn't have to wait for numbered releases
2016-01-11 17:58:31 <barthalion> you of course mean abuild
2016-01-11 17:58:43 <skarnet> yes, abuild, sry
2016-01-11 17:59:07 <barthalion> I think clandmeter implemented this ages ago
2016-01-11 17:59:14 <barthalion> we just don't use it for some reason
2016-01-11 17:59:26 <skarnet> last time I asked, I was told it wasn't possible
2016-01-11 17:59:32 <skarnet> that was a couple months ago
2016-01-11 17:59:38 <barthalion> hm
2016-01-11 18:01:01 <barthalion> gonna check it as well, tomorrow
2016-01-11 18:01:21 <skarnet> thanks
2016-01-11 18:04:42 <littledude> if someone wants to take a look, here it is: https://github.com/little-dude/alpine-hping
2016-01-11 18:20:29 <barthalion> littledude: no need to specify depends
2016-01-11 18:20:53 <barthalion> these are autodetected for dynamically linked binaries and libraries
2016-01-11 18:21:16 <littledude> ok
2016-01-11 18:21:18 <barthalion> the rest LGTM
2016-01-11 18:21:32 <littledude> I just noticed that `man hping` does not work though
2016-01-11 18:21:52 <littledude> alpinedev:/home/corentin/hping# man hping
2016-01-11 18:21:54 <littledude> /usr/share/man/mandoc.db: No such file or directory
2016-01-11 18:21:55 <littledude> man: No entry for hping in the manual.
2016-01-11 18:22:28 <littledude> the problem does not come from man, I just installed tree-doc and `man tree` works as expected.
2016-01-11 18:22:53 <littledude> although the APKBUILD does not seem to do anything specific: http://git.alpinelinux.org/cgit/aports/tree/main/tree/APKBUILD
2016-01-11 18:29:55 <barthalion> you probably need to trigger a mandb rebuild/update
2016-01-11 18:30:10 <barthalion> or maybe not
2016-01-11 18:31:27 <barthalion> I need to go, sorry
2016-01-11 18:31:42 <littledude> np, thanks a lot for your help
2016-01-11 18:31:49 <littledude> I'll try to figure it out
2016-01-11 18:41:46 <_mjones> Yeah I'm getting mandb errors on my old devbox and a new install too. Likely need to account for that in packaging/apk...
2016-01-11 18:43:37 <littledude> I solved the mandb error by installing mdocml-apropos
2016-01-11 18:44:26 <littledude> which does something like this: ` rm -f /usr/share/man/mandoc.db ;  /usr/sbin/makewhatis -a /usr/share/man`
2016-01-11 18:44:50 <littledude> but still, I get the `no entry for hping`
2016-01-11 18:45:26 <_mjones> I'm saying that I suspect the makewhatis needs to be triggered automatically now. By an apk trigger, I suppose.
2016-01-11 18:45:39 <_mjones> Didn't used to need it, but it seems like something changed.
2016-01-11 18:47:39 <littledude> http://lists.alpinelinux.org/alpine-aports/0802.html
2016-01-11 18:51:43 <_mjones> Oh, good catch littledude.
2016-01-11 19:01:07 <littledude>  _mjones: the man for hping was not found because of its name
2016-01-11 19:01:10 <littledude> mv /usr/share/man/man8/hping2.8.gz /usr/share/man/man8/hping.8.gz
2016-01-11 19:01:24 <littledude> that solved it.
2016-01-11 19:01:41 <littledude> not sure if it's a bug or if the name is really invalid
2016-01-11 19:04:05 <littledude> so hping.2.8.gz is valid but not hping2.8.gz
2016-01-11 19:10:35 <_mjones> That's unexpected. A manpage can't end with a digit?
2016-01-11 19:15:55 <littledude> on my archlinux it can. but archlinux uses `mandb`. I don't really how it works on alpine actually.
2016-01-11 19:16:22 <littledude> I don't really know* how it works
2016-01-11 19:19:11 <_mjones> That ideally should be logged as a bug so it doesn't get forgotten, IMO.
2016-01-12 01:14:46 <BitL0G1c> can imapsync be pushed please now all of it's dependencies are in the repos ?
2016-01-12 08:07:59 <nmeum> ncopa: by the way: even though you are very busy these days do you think that it would be possible to replace udev with something smaller in the 3.4 release? #4985
2016-01-12 08:19:09 <skarnet> nmeum: what's preventing Xorg from running with mdev ? I'll have that very problem to handle soonish
2016-01-12 08:39:01 <clandmeter> skarnet: i think there are some notes regarding mdev and xorg here https://github.com/slashbeast/mdev-like-a-boss
2016-01-12 08:40:20 <skarnet> clandmeter: thanks, bookmarked
2016-01-12 08:52:40 <ScrumpyJack> morning all
2016-01-12 08:58:23 <ScrumpyJack> anyone having trouble with sdcard devices since 3.3?
2016-01-12 09:00:09 <ScrumpyJack> kernel can't seem to see them any more (appropriate modules loaded, such as sdhci)
2016-01-12 09:03:23 <ScrumpyJack> dmesg shows the reader:
2016-01-12 09:03:26 <ScrumpyJack> scsi 1:0:0:2: Direct-Access     Generic- SD/MMC           1.00 PQ: 0 ANSI: 0 CCS
2016-01-12 09:24:14 <nmeum> skarnet: it's possible you probably have to compile some Xorg packages without udev support (xf86-input-evdev for instance) and you need to configure your input device manually
2016-01-12 09:24:32 <nmeum> however, it's not suitable for alpine because hotplugin and shit doesn't work
2016-01-12 11:17:11 <ncopa> nmeum: re %1283
2016-01-12 11:17:24 <ncopa> we dont want strong for x86?
2016-01-12 11:18:59 <ncopa> fabled: any opinion on enabling strong stack protector on kernel?
2016-01-12 11:19:14 <nmeum> ncopa: you can't somehow set in on x86
2016-01-12 11:19:21 <nmeum> ncopa: some grsecurity feature is preventing that
2016-01-12 11:19:29 <ncopa> ok
2016-01-12 11:19:38 <nmeum> menuconfig will tell you which one
2016-01-12 11:19:42 <nmeum> I don't remember anymore
2016-01-12 11:19:52 <ncopa> ok
2016-01-12 11:20:01 <ncopa> just checking so it was not an oversight
2016-01-12 11:21:53 <fabled> ncopa, no strong opinion from me
2016-01-12 11:22:15 <ncopa> pun intended? :)
2016-01-12 11:22:36 <fabled> yes ;)
2016-01-12 11:22:44 <ncopa> i'll enable it and test it in qemu
2016-01-12 11:22:49 <ncopa> and push it to edge
2016-01-12 11:22:55 <ncopa> i have a fix for overlayfs
2016-01-12 11:42:11 <skarnet> nmeum: can't input devices just use the hotplug mechanism (which will be handled by mdev) without depending on udev specifically?
2016-01-12 12:47:49 <ncopa> nmeum: (e)udev, xorg and hotplug is an issue
2016-01-12 12:47:54 <ncopa> not that easy to solve
2016-01-12 12:48:29 <ncopa> i was hoping for getting a proper udev alternative for alpine v3.4 but it does not look like I'll be able to :-/
2016-01-12 12:52:13 <skarnet> ncopa: I'll definitely have to deal with the subject in a couple months
2016-01-12 12:52:31 <skarnet> so I'll take every bit of information and attempt at alternatives as I can
2016-01-12 12:52:36 <skarnet> and I'll give back everything I can
2016-01-12 12:52:45 <ncopa> very nice
2016-01-12 12:52:58 <ncopa> the plan i had was to use nlplug as base
2016-01-12 12:53:18 <ncopa> which turned in to a special case tool for initramfs -> nlplug-findfs
2016-01-12 12:53:31 <skarnet> don't take it the wrong way, or personally, but I think nlplug needs to be rewritten from scratch :P
2016-01-12 12:54:04 <skarnet> did you mean to generalize nlplug(-findfs) as a hotplug manager?
2016-01-12 12:54:06 <ncopa> i was hoping for something based on netlink, similar to udev
2016-01-12 12:54:11 <ncopa> yes
2016-01-12 12:54:28 <skarnet> it almost exists
2016-01-12 12:54:41 <skarnet> it's the "mdev -i" command we talked about on the busybox ml last year
2016-01-12 12:56:02 <skarnet> it's easy to listen to the netlink (you and I have done it), it's almost easy (almost) to pair a netlink listener to a mdev.conf parser
2016-01-12 12:56:32 <skarnet> the big question is, what are the udev-specific interfaces Xorg uses that mdev isn't addressing
2016-01-12 12:57:37 <skarnet> patching mdev to accommodate Xorg's (and others') needs is doable
2016-01-12 12:58:14 <skarnet> but if mdev's model is insufficient for some reason, and something entirely different needs to be used
2016-01-12 12:58:25 <skarnet> then that's an order of magnitude harder to make happen
2016-01-12 13:04:39 <skarnet> the simplest would probably be to use libudev-compat from vdev
2016-01-12 13:27:01 <ScrumpyJack> Im getting "Clock skew detected with /etc/init.d/consolefont" while booting the RPi with a RTC clock installed
2016-01-12 13:27:21 <ScrumpyJack> I wonder why it does that
2016-01-12 13:31:28 <skarnet> there's no offline battery, or it's dead, or weak, or anything of the kind
2016-01-12 13:31:44 <skarnet> so the clock isn't accurate when you power on the box
2016-01-12 13:31:59 <ScrumpyJack> hmm
2016-01-12 13:32:02 <ScrumpyJack> new device
2016-01-12 13:32:20 <skarnet> powercycle it a few times
2016-01-12 13:32:33 <ScrumpyJack> i can see the device, get and set the device clock, and hwclock sets system time from it at boot
2016-01-12 13:32:44 <ScrumpyJack> ok, i'll try that
2016-01-12 13:32:53 <skarnet> check that /etc/init.d/consolefont doesn't have a time in the future
2016-01-12 13:33:04 <skarnet> that can happen if the clock was incorrect when it was last modified
2016-01-12 13:33:39 <ScrumpyJack> yes, it's the future (16 dec with no year)
2016-01-12 13:33:58 <skarnet> just touch it and you should be fine afterwards
2016-01-12 13:34:56 <ScrumpyJack> in fact evreything is either 1970 or december
2016-01-12 13:41:33 <ScrumpyJack> is everything in /etc/ added to apkovl by lbu?
2016-01-12 13:42:17 <ScrumpyJack> or do i need to at +/etc to protected_paths.d/lbu.list
2016-01-12 13:44:03 <ScrumpyJack> aha, not etc/init
2016-01-12 13:44:14 <ScrumpyJack> i remember that now
2016-01-12 13:46:04 <ScrumpyJack> how do i add something into etc/init.d into the apkovl with lbu?
2016-01-12 13:46:08 <fabled> symlinks in init.d go there
2016-01-12 13:46:19 <fabled> but if you add new files you need to lbu include them
2016-01-12 13:50:35 <ScrumpyJack> symlinks to/from where?
2016-01-12 13:50:59 <ScrumpyJack> stat thinks /etc/init.d/consolefont is a regular file
2016-01-12 13:51:31 <ScrumpyJack> that has a symlink (?)
2016-01-12 13:52:23 <ScrumpyJack> symlinks from /etc/runlevels?
2016-01-12 13:54:39 <fabled> ScrumpyJack, only the symlinks are stored; eg. if you have lxc.myinstance -> lxc
2016-01-12 13:55:09 <fabled> changes to init.d files is not preferred; if possible edit conf.d or ask the pkg init.d script to be updated
2016-01-12 13:55:18 <fabled> temporarily: you'd need to lbu include that file
2016-01-12 13:56:11 <ScrumpyJack> ok
2016-01-12 13:56:17 <ScrumpyJack> tryin that now
2016-01-12 13:57:57 <ScrumpyJack> +etc/init.d/consolefont in lbu.list made no diff
2016-01-12 14:04:22 <ScrumpyJack> (and etc/init.d/consolefont doesn't end up in the apkovl)
2016-01-12 14:04:45 <fabled> did you modify it? only changed files go there with +
2016-01-12 14:04:49 <fabled> if you want it included always
2016-01-12 14:05:14 <ScrumpyJack> i guess a touch wasn't enough then, i'll try again
2016-01-12 14:05:17 <fabled> use !etc/init.d/consolefont
2016-01-12 14:05:26 <fabled> yes, touch is not enough
2016-01-12 14:05:37 <fabled> the size or sha1 checksum needs to differ
2016-01-12 14:06:13 <ScrumpyJack> ah ok
2016-01-12 14:06:52 <ScrumpyJack> so ! is backup regardless?
2016-01-12 14:07:08 <fabled> ! is always store on overlay
2016-01-12 14:07:12 <fabled> it's mainly used for etc/apk
2016-01-12 14:07:21 <fabled> so that the repositories list, and keys are always on overlay
2016-01-12 14:07:27 <fabled> as they are needed to bootstrap the install
2016-01-12 14:07:50 <fabled> (when etc/apk/repositories.d/foo.list and the keys come from self built "policy" packages)
2016-01-12 14:09:59 <ScrumpyJack> clock skew still detected on /etc/init.d/consolefont, so i guess it's seeing the one in init?
2016-01-12 14:10:11 <ScrumpyJack> before the apkovl overlay?
2016-01-12 14:10:21 <fabled> clock skew can depend on the openrc cache file, and system time on bootup
2016-01-12 14:11:10 <ScrumpyJack> this is the RPi and RTC i meantioned earlier
2016-01-12 14:11:19 <fabled> maybe your system rtc time is in future, but later correct to real time with ntp after initial boot
2016-01-12 14:16:19 <ScrumpyJack> no ntp yet. openrc tells me that "Setting system clock using hardware clock"
2016-01-12 14:16:36 <ScrumpyJack> even though hwclock isn't added to any runlevel yet
2016-01-12 14:19:29 <ScrumpyJack> how do we avoid files in the future in the alpine-rpi.tgz?
2016-01-12 14:20:04 <ScrumpyJack> set the time on the build box for 1st JAN?
2016-01-12 14:22:36 <fabled> sounds like we need to fix ntp on the builders
2016-01-12 14:23:10 <fabled> ncopa, can you check arm 3.3-stable builders' time?
2016-01-12 14:24:57 <ncopa> build-3-3-armhf:~$ date
2016-01-12 14:24:57 <ncopa> Tue Jan 12 14:24:50 GMT 2016
2016-01-12 14:40:57 <ScrumpyJack> and files in the tar file have the date 2016-01-06
2016-01-12 14:49:12 <fabled> Jan 06 ?
2016-01-12 14:49:35 <fabled> i think it's Jan-6; not 1st of July
2016-01-12 14:49:42 <fabled> tar should say yyyy-mm-dd
2016-01-12 14:54:53 <ScrumpyJack> Jan 6, yes
2016-01-12 14:55:14 <ScrumpyJack> still can't find out why all these files at 1970 when i boot
2016-01-12 14:55:23 <ScrumpyJack> s/at/are
2016-01-12 14:56:02 <ScrumpyJack> they must be created before the time is set
2016-01-12 14:56:14 <fabled> sounds like that
2016-01-12 14:56:41 <ScrumpyJack> so i need to set the time earlier in the boot process
2016-01-12 14:56:42 <skarnet> where do those files come from? an archive or something?
2016-01-12 14:57:01 <ScrumpyJack> they are in ramfs
2016-01-12 14:57:11 <fabled> try "lbu include /lib/rc/cache/"
2016-01-12 14:57:14 <skarnet> bootstrapping time is har
2016-01-12 14:57:18 <fabled> that's the openrc cache
2016-01-12 14:57:21 <skarnet> hard*
2016-01-12 14:57:30 <fabled> and touch all files there too
2016-01-12 14:57:34 <fabled> to make sure it's current time
2016-01-12 14:57:49 <fabled> i do that on my rpi
2016-01-12 14:58:12 <fabled> mostly to speed up boot time some tens or hundreds of milliseconds (especially on rpi1)
2016-01-12 15:01:16 <ScrumpyJack> how is bootstrapping hardware RTC on the RPi? doesn't the kernel need to see the RTC first?
2016-01-12 15:02:18 <fabled> yes, dtoverlay is best way to do it
2016-01-12 15:02:44 <ScrumpyJack> that's what i use
2016-01-12 15:02:57 <fabled> is the module auto-loaded?
2016-01-12 15:04:03 <ScrumpyJack> i think so, with dtoverlay=i2c-rtc,pcf2127 in usercfg.txt
2016-01-12 15:06:11 <fabled> well then it's should be okish. maybe hwclock should just run earlier. or potentially the rtc modules should be made available in initramfs
2016-01-12 15:06:22 <fabled> though both are slightly problematic
2016-01-12 15:06:27 <ScrumpyJack> http://sprunge.us/TQHT
2016-01-12 15:06:41 <fabled> oh
2016-01-12 15:06:47 <fabled> you mean those directory timestamps?
2016-01-12 15:07:08 <fabled> yeah, apk creates them early during bootstrap
2016-01-12 15:07:15 <fabled> and for files the current time is used
2016-01-12 15:07:27 <fabled> err... for directories*
2016-01-12 15:07:34 <fabled> for files it's the timestamp on .apk
2016-01-12 15:07:53 <fabled> to fix that rtc would need to be set on boottime - or through someother means
2016-01-12 15:08:00 <ScrumpyJack> http://sprunge.us/ched
2016-01-12 15:08:12 <ScrumpyJack> all these are in the future :(
2016-01-12 15:09:49 <fabled> can you check with stat the timestamp's year?
2016-01-12 15:09:52 <fabled> maybe it's 2015?
2016-01-12 15:11:39 <fabled> or ls --full-time
2016-01-12 15:12:02 <fabled> i think ls uses the month if it's within 6 months from 'now'
2016-01-12 15:12:30 <fabled> to me those stamps look ok
2016-01-12 15:12:39 <fabled> if you get the clock skew error from openrc
2016-01-12 15:13:08 <fabled> it is likely because system time when running openrc is not yet updated, and it created cache file that has older timestamp than init.d files
2016-01-12 15:13:24 <fabled> lbu include of rc cache files should fix it
2016-01-12 15:18:41 <ScrumpyJack> ok, so i added hwclock to the sysinit runlevel and that was too early, but it's better at the boot runlevel
2016-01-12 15:19:20 <ScrumpyJack> http://sprunge.us/DSPI
2016-01-12 15:21:32 <ScrumpyJack> and these look better too http://sprunge.us/GOXG
2016-01-12 15:22:03 <fabled> no. i think your date is now wrong
2016-01-12 15:22:14 <fabled> i think hwclock does not work that early
2016-01-12 15:22:27 <ScrumpyJack> i still get clock skew detected with /etc/init.d consolefont though
2016-01-12 15:22:30 <fabled> the rtc module is loaded in 'hwdrivers' service
2016-01-12 15:22:38 <fabled> and apk runs in initramfs
2016-01-12 15:22:45 <fabled> which is three steps earlier
2016-01-12 15:23:52 <ScrumpyJack> i see, so the first of jan should have the 1970 date still in /etc ?
2016-01-12 15:25:32 <fabled> yes
2016-01-12 15:27:54 <ScrumpyJack> it's getting messy now, my hwclock has reset to 1970
2016-01-12 15:30:04 <ScrumpyJack> (that could have been me thought)
2016-01-12 15:32:50 <ScrumpyJack> i need the wiringpi libs. is it possible to build for the rpi in qemu-arm?
2016-01-12 15:37:52 <ScrumpyJack> fabled: I don't have lib/rc/cache/
2016-01-12 15:39:23 <ScrumpyJack> this is what a stat says
2016-01-12 15:39:25 <ScrumpyJack> http://sprunge.us/hEci
2016-01-12 15:45:38 <ScrumpyJack> heh, wiringpi is in edge/testing. neat!
2016-01-13 08:33:37 <blueness> this is perhaps the most important mgtow video i’ve every seen -> https://www.youtube.com/watch?v=4wpca1ZDIRQ
2016-01-13 08:35:03 <ScrumpyJack> morning all
2016-01-13 08:52:12 <coredumb> morning
2016-01-13 09:12:17 <blueness> oops wrong channel :)
2016-01-13 09:17:38 <fcolista> how do i know the package who owns blockdev ? looks that apk info -W blockdev does not work
2016-01-13 09:23:38 <clandmeter> fcolista: pkgs.a.o?
2016-01-13 09:24:34 <fcolista> clandmeter, yes, it works
2016-01-13 09:24:41 <fcolista> how about command line?
2016-01-13 09:26:52 <clandmeter> -W should work
2016-01-13 09:27:00 <clandmeter> if its installed by the package in the current system
2016-01-13 09:27:33 <fcolista> it is installed.
2016-01-13 09:27:34 <fcolista> 2ua4020qdl:~# apk info -W blockdev
2016-01-13 09:27:34 <fcolista> ERROR: blockdev: Could not find owner package
2016-01-13 09:27:34 <fcolista> 2ua4020qdl:~# apk info | grep util-linux
2016-01-13 09:27:34 <fcolista> util-linux
2016-01-13 09:27:34 <fcolista> util-linux-dev
2016-01-13 09:27:38 <clandmeter> apk only keeps a "contents" record of installed packages
2016-01-13 09:27:44 <fcolista> blockdev belongs to util-linux
2016-01-13 09:28:09 <fcolista> sounds like a bug
2016-01-13 09:29:01 <clandmeter>  apk info -W /sbin/blockdev
2016-01-13 09:29:01 <clandmeter> /sbin/blockdev is owned by util-linux-2.26.2-r0
2016-01-13 09:29:08 <clandmeter> works her
2016-01-13 09:29:09 <clandmeter> e
2016-01-13 09:29:19 <fcolista> try without the entire path
2016-01-13 09:30:02 <clandmeter> ?
2016-01-13 09:30:10 <fcolista> not apk info -W /sbin/blockdev
2016-01-13 09:30:16 <fcolista> but apk info -W blockdev
2016-01-13 09:30:28 <clandmeter> you cannot search like this. it will need full path.
2016-01-13 09:30:43 <fcolista> ah..
2016-01-13 09:30:48 <clandmeter> atleast not that i know.
2016-01-13 09:30:48 <fcolista> ok.
2016-01-13 09:31:01 <fcolista> sounds like a limitation, though
2016-01-13 09:31:09 <clandmeter> yes and im working on it.
2016-01-13 09:31:17 <fcolista> oh, ok
2016-01-13 10:22:08 <ScrumpyJack> my multicard read doesn't work automagically since kernel 4. I have to fiddle with the SCSI utils and /proc filesystem to get it to work
2016-01-13 10:23:54 <ScrumpyJack> i looks like CONFIG_SCSI_MULTI_LUN option got dropped from the 4 branch, which seemed enable probing of all the LUNs on a multicard reader.
2016-01-13 10:24:33 <ScrumpyJack> do we know what's replaced it, if anything?
2016-01-13 15:50:22 <faffolter> please git pull git://git.alpinelinux.org/user/fab/aports
2016-01-13 15:51:13 <faffolter> still the same stuff as last time ;-)
2016-01-14 07:59:49 <fcolista> clandmeter, has pkgs.alpinelinux.org API i can use for scripting?
2016-01-14 08:19:08 <clandmeter> fcolista: not yet, working on it.
2016-01-14 08:19:20 <fcolista> gr8 clandmeter
2016-01-14 08:19:21 <fcolista> thx
2016-01-14 08:21:28 <clandmeter> fcolista: when api is done, the idea is to create a lua script to query it.
2016-01-14 08:22:20 <fcolista> clandmeter, that's exactly what i was thinking
2016-01-14 08:22:37 <fcolista> that would help _a lot_ on keep tracking of obsolete packages
2016-01-14 08:22:53 <ScrumpyJack> morning
2016-01-14 08:23:21 <fcolista> and the best would be having an upstream monitor...ncopa was working on it...fedora has is own that might be forked...
2016-01-14 08:23:43 <fcolista> but looks a big job engineer it
2016-01-14 08:23:50 <ncopa> morning
2016-01-14 08:24:01 <fcolista> hi ncopa!
2016-01-14 08:24:12 <ncopa> other option would be to include a function in APKBUILD, upstreamcheck or similar
2016-01-14 08:24:23 <ncopa> which would do the check
2016-01-14 08:24:31 <fcolista> yes
2016-01-14 08:24:36 <fcolista> that was an option
2016-01-14 08:25:03 <fcolista> and it would help a lot the same
2016-01-14 08:25:14 <clandmeter> what are "obsolete packages"?
2016-01-14 08:25:44 <fcolista> $oldpackages = $newpackageversion - 1
2016-01-14 08:26:10 <fcolista> monitor if a new release is available, in simple terms
2016-01-14 08:26:28 <clandmeter> you mean out of date packages?
2016-01-14 08:26:46 <fcolista> yes
2016-01-14 08:27:17 <clandmeter> i thought ncopa scanner was using archlinux's db
2016-01-14 08:27:30 <fcolista> yes
2016-01-14 08:27:41 <ncopa> it checks various sources
2016-01-14 08:28:04 <ncopa> if github.com is found in $source it will check git tags
2016-01-14 08:28:25 <ncopa> if gnome is found it will check ftp.gnome.org iirc
2016-01-14 08:28:44 <ncopa> otherwise it will fall back to checking what arch version is
2016-01-14 08:28:46 <ncopa> or similar
2016-01-14 08:30:11 <clandmeter> fcolista: you need an api for that to work? i think apk can already provide you with pkgname and pkgver?
2016-01-14 09:57:19 <rishoop1> is it easy to install the libressl package and fully replace openssl; also, does alpine plan to migrate officially from openssl into libressl (hope so) thx
2016-01-14 10:04:15 <rishoop1> must go - will look for replies in the chat logs thx
2016-01-14 10:04:22 <rishoop1> love alpine best distro
2016-01-14 19:39:24 <nmeum> rishoop1: nope, it's currently not possible it might change in the future though
2016-01-14 19:40:23 <nmeum> ncopa: Could you upgrade OpenSSH there has been a critical flaw in the ssh client code (see CVE-2016-0777 and CVE-0216-0778)
2016-01-14 19:40:54 <ammunta> snrk I was *just* looking at that and about to drop into to mention
2016-01-14 19:44:50 <skarnet> dropbear is still the default, right?
2016-01-14 20:32:22 <nmeum> skarnet: it's a client issue not a server issue
2016-01-14 20:32:31 <nmeum> dropbear only provides a server, doesn't it?
2016-01-14 20:32:37 <skarnet> it provides both
2016-01-14 20:32:50 <skarnet> (the name of the client is dbclient)
2016-01-14 20:34:54 <nmeum> oh, didn't know that
2016-01-14 20:48:05 <ncopa> nmeum: im on it
2016-01-14 20:48:10 <ncopa> thanks for the headsup
2016-01-14 20:50:58 <nmeum> no problem
2016-01-14 20:54:30 <ncopa> i can find patch for CVE-2016-0777 but nothing about CVE-2016-0778
2016-01-14 20:54:47 <ncopa> can anybody help find a patch for  CVE-2016-0778?
2016-01-14 20:57:24 <nmeum> http://ftp.openbsd.org/pub/OpenBSD/patches/5.8/common/010_ssh.patch.sig
2016-01-14 20:57:29 <nmeum> ^ should fix both CVE-2016-0777 and CVE-2016-0778
2016-01-14 20:57:51 <nmeum> it just disable the roaming feature by default
2016-01-14 20:57:55 <nmeum> *s
2016-01-14 21:19:02 <ncopa> yeah
2016-01-14 21:19:03 <ncopa> i figured
2016-01-14 21:19:09 <ncopa> thanks
2016-01-14 21:19:18 <ncopa> pushed to all maintained branches
2016-01-14 21:24:07 <mosez> awww... nodejs/npm seems to be broken
2016-01-14 21:24:16 <mosez> just starting a fresh system to verify...
2016-01-14 21:26:22 <mosez> my docker container for it fails to build for a few days
2016-01-14 21:27:59 <ncopa> i saw the issue some place
2016-01-14 21:34:12 <ncopa> mosez: same issue as this? https://github.com/gliderlabs/docker-alpine/issues/127
2016-01-14 21:35:12 <mosez> ncopa: yes... on edge
2016-01-14 21:35:40 <mosez> https://gist.github.com/tboerger/f1e38440411d9c7410f7
2016-01-14 21:37:59 <mosez> ncopa: so i'm forced to install the 3.2 package? :(
2016-01-14 21:38:36 <ncopa> i want fix it
2016-01-14 21:40:42 <mosez> ncopa: and i have no idea how to solve that
2016-01-14 21:40:56 <ncopa> neither have I
2016-01-14 21:40:57 <ncopa> yet...
2016-01-14 21:44:08 <mosez> to be able to install couchdb reproduceable i need to compile the unmaintained package and put it into a custom package index?
2016-01-14 21:59:00 <mosez> ncopa: will this be integrated into version 7 as well? https://bugs.alpinelinux.org/projects/alpine/repository/revisions/2445067072ee0b830308575d5d63ce0981a73de3
2016-01-14 22:02:56 <ncopa> no plans for that currently
2016-01-14 22:03:01 <ncopa> do you need it?
2016-01-14 22:03:29 <mosez> mkay... than i should try java8 for my minecraft/java containers... maybe i can get rid of my opensuse copy
2016-01-14 22:04:13 <mosez> currently i'm adding https://github.com/dockhippie/java/tree/master/rootfs/usr/lib/jvm/java-1.7-openjdk/jre/lib/security to my containers
2016-01-14 22:52:31 <ncopa> mosez: i know what the problem is with nodejs/openssl
2016-01-14 22:52:39 <ncopa> i dont know what the fix is though
2016-01-14 22:53:05 <ammunta> what's the problem
2016-01-14 22:53:15 <ncopa> i am emailing fabled and hopefully he will have a fix early tm morning
2016-01-14 22:53:43 <ncopa> problem is engine gets loaded twice
2016-01-14 22:53:47 <ncopa> crypto engine
2016-01-14 22:53:51 <ncopa> in openssl
2016-01-14 22:53:55 <ncopa> which is kinda of weird
2016-01-14 22:54:09 <ncopa> we do have a patch that autoloads padlock
2016-01-14 22:54:32 <ncopa> but it happens also when no padlock hw is available
2016-01-14 22:54:47 <ncopa> and removing the patch makes it work
2016-01-14 22:55:37 <ncopa> my guess is that fix is relatively simple
2016-01-14 22:56:09 <ncopa> probably we only need assign some var=NULL when padlock fails to load
2016-01-14 22:58:18 <ammunta> whoa what happened with giant load of patches to musl friday
2016-01-14 22:58:56 <ncopa> i need sleep now
2016-01-14 22:59:01 <ammunta> oh I see
2016-01-14 22:59:01 <ncopa> see u tm
2016-01-14 22:59:18 <ncopa> nodejs will be fixed early tm morning
2016-01-14 22:59:24 <ncopa> one way or the other
2016-01-15 11:36:06 <ncopa> we used to have aufs with our grsec kernel?
2016-01-15 11:56:16 <seamuz> Hi! Is there possibility to run alpine over PXE in diskless mode with mkinitfs-3.0.2 and nlplug-findfs utility without any medium for store apkovl's and modloop? I successfuly do it with mkinitfs-2.7.1 with little patch, apkovl was downloaded from my http server. But with 3.0.2 I am not sure how-to deal with it :-\
2016-01-15 12:23:17 <ncopa> seamuz: can i have a look ata your little patch?
2016-01-15 12:23:57 <ncopa> when i looked at the code i kinda thought that getting apkovl from http never had worked
2016-01-15 12:24:01 <ncopa> and assumed nobody used it
2016-01-15 12:24:46 <ncopa> seamuz: i believe it is fixable
2016-01-15 12:25:30 <ncopa> seamuz: did you use any specific boot option for it to work?
2016-01-15 12:27:00 <seamuz> ncopa: http://pastebin.com/HRgkaeUG
2016-01-15 12:27:37 <seamuz> I just do not declare "alpine_dev" boot opt
2016-01-15 12:27:44 <seamuz> and this work as sharm
2016-01-15 12:28:17 <seamuz> like a
2016-01-15 12:28:31 <ncopa> do you specify apkovl at boot option?
2016-01-15 12:28:35 <ncopa> and ip addr?
2016-01-15 12:28:42 <seamuz> yes
2016-01-15 12:28:48 <seamuz> apkovl=http://10.145.50.251/default.apkovl.tar.gz
2016-01-15 12:29:03 <seamuz> ip=dhcp
2016-01-15 12:29:14 <ncopa> and thats it?
2016-01-15 12:29:19 <seamuz> yes
2016-01-15 12:29:35 <seamuz> and alpine_repo= was specified
2016-01-15 12:29:48 <ncopa> to use repo from http
2016-01-15 12:29:56 <ncopa> i dont think it should be too hard to fix
2016-01-15 12:30:52 <seamuz> http://pastebin.com/2vz40GTJ
2016-01-15 12:31:19 <seamuz> but i do not know how
2016-01-15 12:32:10 <ncopa> nlplug-findfs will coldplug the hardward
2016-01-15 12:32:20 <ncopa> so it has to run
2016-01-15 12:32:42 <ncopa> it will time out on 250ms (or if you have usb storage attached it will be 1 or 2 sec)
2016-01-15 12:33:03 <ncopa> then it has to fetch the apkovl
2016-01-15 12:33:08 <ncopa> looks like that should work still
2016-01-15 12:33:49 <seamuz> I am miss any code in mkinitfs for http file retrieving
2016-01-15 12:34:13 <ncopa> its there
2016-01-15 12:34:19 <seamuz> nlplug also does find apkovl's on media
2016-01-15 12:34:28 <ncopa> ah
2016-01-15 12:34:31 <ncopa> it finds it on media
2016-01-15 12:34:48 <ncopa> when it should not
2016-01-15 12:34:51 <ncopa> that is a problem
2016-01-15 12:35:25 <ncopa> http://git.alpinelinux.org/cgit/mkinitfs/tree/initramfs-init.in#n423
2016-01-15 12:35:28 <ncopa> code is still there
2016-01-15 12:35:34 <ncopa> i tried to not break it
2016-01-15 12:35:39 <ncopa> but i didnt test it either
2016-01-15 12:36:45 <seamuz> hmm
2016-01-15 12:36:54 <ncopa> hum
2016-01-15 12:37:05 <ncopa> it should let KOPT_apkovl override
2016-01-15 12:37:20 <ncopa> and not use the found apkvol on any attached media
2016-01-15 12:37:59 <ncopa> does it get ip addr?
2016-01-15 12:38:03 <seamuz> can i just delete this line http://git.alpinelinux.org/cgit/mkinitfs/tree/initramfs-init.in#n400
2016-01-15 12:38:39 <ncopa> i think not. i think you need it to load the NIC driver
2016-01-15 12:39:02 <ncopa> ah
2016-01-15 12:39:18 <skarnet> ncopa: following on what you said yesterday or 2 days ago: is there a list somewhere of packages that don't build against libressl as opposed to openssl?
2016-01-15 12:39:31 <ncopa> it finds the boot repository?
2016-01-15 12:39:35 <ncopa> skarnet: not that i know
2016-01-15 12:39:46 <seamuz> nope
2016-01-15 12:40:06 <seamuz> script stopped by time out
2016-01-15 12:40:22 <ncopa> ok that is good
2016-01-15 12:40:31 <skarnet> ncopa: so the reason why there are no plans to switch to libressl at this time is just that nobody has looked into it so far ?
2016-01-15 12:40:44 <ncopa> skarnet: nobody looks at it i think
2016-01-15 12:41:02 <ncopa> skarnet: i would like to check hw crypto support in libressl
2016-01-15 12:41:09 <ncopa> eg if it supports via padlock
2016-01-15 12:41:26 <ncopa> not user if that is important anymore
2016-01-15 12:45:32 <seamuz> ncopa: there is no problem for me hold on 2.7.1, but I am migrated from 3.2 to 3.3 and boot proccess interrupt by dbus.post-install script with message "Failed to generate UUID: Could not open /dev/urandom...
2016-01-15 12:47:36 <seamuz> I tried patch init to create this block device, but with no luck. same message
2016-01-15 12:48:47 <seamuz> anyway.. Now I have to go, but I'll be back on Monday. Thank you
2016-01-15 12:53:09 <skarnet> ncopa: I'm now going to try and build some stuff (including git) using libressl instead of openssl, I'll report my findings
2016-01-15 12:53:56 <skarnet> (for armhf btw)
2016-01-15 12:54:26 <ncopa> perfect
2016-01-15 12:56:22 <fabled> libressl/openssl is going to be soon inevitable. there's new major openssl release in pipeline (pre2 released). and i think it requires full recompile
2016-01-15 12:57:41 <skarnet> I wish the Dutch government had funded libressl instead
2016-01-15 14:02:36 <ScrumpyJack> what modules do i need to load for the sound card on a rpi?
2016-01-15 14:09:34 <ScrumpyJack> snd_bcm2835
2016-01-15 15:14:39 <nmeum> ncopa: BTW: would you mind upgrading the firefox version in testing/?
2016-01-15 15:37:54 <ballen> Is testing/virtualbox-additions-grsec supposed to be the guest additions kernel modules, or the modules needed for for VirtualBox install itself on Alpine? Currently its the latter and equivalent to https://gitweb.gentoo.org/repo/gentoo.git/tree/app-emulation/virtualbox-modules/virtualbox-modules-5.0.12.ebuild.
2016-01-18 03:10:15 <mitchty> ok fired of my ghc patch to the alpine-devel list, if i screwed anything up let me know
2016-01-18 06:52:48 <seamuz> Boot proccess of Alpine Linux over PXE https://youtu.be/k3YG76cPfi8
2016-01-18 07:51:10 <seamuz> ncopa: hi
2016-01-18 07:51:29 <ncopa> seamuz: morning
2016-01-18 07:51:42 <seamuz> i successfuly boot over PXE
2016-01-18 07:51:50 <seamuz> with these pathes https://github.com/seamus-45/aports/commit/bfd51bcf07b630ca897038b676897f92edce323b
2016-01-18 07:51:50 <ncopa> nice!
2016-01-18 07:52:40 <seamuz> ncopa: please note about mac address issue
2016-01-18 07:53:31 <seamuz> $device variable was used out of scope
2016-01-18 07:57:23 <ScrumpyJack> morning folks
2016-01-18 07:59:50 <ncopa> seamuz: i suppose that was the main issue
2016-01-18 08:00:35 <seamuz> 0001 and 0004 patches fix main issues
2016-01-18 08:01:13 <seamuz> without 0001 boot proccess terminated and drop to busybox shell
2016-01-18 08:01:25 <ncopa> ah
2016-01-18 08:01:42 <ncopa> yes. makes sense
2016-01-18 08:02:54 <ncopa> seamuz: I suppose the modloop service does not work?
2016-01-18 08:03:12 <seamuz> I do not need it
2016-01-18 08:03:32 <ncopa> you have all the kernel modules in initramfs?
2016-01-18 08:03:37 <seamuz> yep
2016-01-18 08:04:14 <ncopa> it will fetch the apkovl too?
2016-01-18 08:04:19 <seamuz> yes
2016-01-18 08:04:45 <ncopa> then i think you could rc-update del modloop sysinit
2016-01-18 08:04:45 <seamuz> as you can see on video https://youtu.be/k3YG76cPfi8
2016-01-18 08:05:23 <seamuz> why? these services hardcoded in initramfs-init
2016-01-18 08:05:38 <ncopa> only if there is no apkovl
2016-01-18 08:05:47 <seamuz> ahh
2016-01-18 08:05:53 <ncopa> so only for first boot
2016-01-18 08:06:04 <ncopa> cool!
2016-01-18 08:06:15 <ncopa> pxe video is super cool :)
2016-01-18 08:08:12 <ncopa> re 0001 patch
2016-01-18 08:08:17 <ncopa> i think what we want to do is
2016-01-18 08:08:55 <ncopa> if boot option(s) for pxe is found, then we drop the -b $repofile and -a /tmp/apkvols options to nlplug-findfs
2016-01-18 08:09:07 <ncopa> and make it exit with success for timeout
2016-01-18 08:09:43 <seamuz> good
2016-01-18 08:12:47 <ScrumpyJack> seamuz: that's neat. can you explain what push_config does? saves to an apkovl on your PXE boot server?
2016-01-18 08:14:16 <seamuz> there is nfs share mounted: push_config='lbu commit && cp /tmp/*.apkovl.tar.gz /mnt/http/${MAC}.apkovl.tar.gz
2016-01-18 08:16:23 <ScrumpyJack> and how do you load /mnt/http/${MAC}.apkovl.tar.gz on the next boot?
2016-01-18 08:17:23 <seamuz> ScrumpyJack: i have "apkovl=http://${net0/next-server}/{MAC}.apkovl.tar.gz" in kernel cmdline
2016-01-18 08:18:14 <ncopa> alternatively we could have an -i <netiface> option
2016-01-18 08:18:26 <ncopa> which will coldplug until it finds the given network interface
2016-01-18 08:18:39 <ncopa> that way we save 250 ms at boot
2016-01-18 08:18:41 <ncopa> but hum
2016-01-18 08:18:49 <ncopa> you probably want coldplug it all
2016-01-18 08:18:51 <ncopa> for all devices
2016-01-18 08:18:56 <seamuz> ScrumpyJack: initramfs do the rest job
2016-01-18 08:19:04 <ncopa> since you have all modules in initramfs
2016-01-18 08:27:12 <ncopa> ok, ack for 0004 patch
2016-01-18 08:27:15 <ncopa> that will be applied
2016-01-18 09:16:59 <ncopa> seamuz: can you help me test a new mkinitfs patch?
2016-01-18 09:17:13 <ncopa> http://tpaste.us/Apew
2016-01-18 09:17:28 <seamuz> sure
2016-01-18 09:17:51 <ncopa> then in your apkovl, make sure that modloop is not starting
2016-01-18 09:18:06 <ncopa> eg rm /etc/runlevels/*/modloop or similar
2016-01-18 09:18:13 <ncopa> then i think it should be ok
2016-01-18 09:18:22 <seamuz> already done
2016-01-18 09:19:37 <ncopa> what the patch does: if alpine_repo boot opt is set, then don't consider it an error if no boot repository was found (eg prevent emergency shell to launch)
2016-01-18 09:19:42 <ncopa> we still catch other errors
2016-01-18 09:19:49 <ncopa> (if any)
2016-01-18 09:20:48 <ncopa> it might be that we should check for ip boot option in modloop service
2016-01-18 09:20:52 <ncopa> or similar
2016-01-18 09:21:08 <ncopa> so we dont bail out with error if no modloop was found
2016-01-18 09:41:03 <seamuz> ncopa: sorry for waiting. doing other stuff. but your patch is work very well!
2016-01-18 09:41:15 <seamuz> no problems detected :)
2016-01-18 09:50:01 <seamuz> say me when this patch be pushed to 3.3 branch
2016-01-18 10:02:05 <seamuz> when mkinitfs 3.0.3 released*
2016-01-18 10:19:58 <seamus-45> ncopa: there is no problem with modloop, we always can do "rc-update del modloop"
2016-01-18 10:19:59 <ncopa> seamus-45: thanks
2016-01-18 10:21:32 <ncopa> yes the idea is to include this for 3.3.2 release
2016-01-18 10:21:53 <ncopa> seamus-45: could you please file a bug on bugs.alpinelinux.org with target 3.3.2?
2016-01-18 10:21:56 <ncopa> just so i dont forget
2016-01-18 10:22:20 <seamus-45> ok
2016-01-18 12:11:34 <ScrumpyJack> what RPi devices does alpine-rpi support?
2016-01-18 12:12:27 <bpiotrowski> all of them
2016-01-18 12:12:54 <clandmeter> i thought we werent sure about the small one.
2016-01-18 12:12:58 <clandmeter> cant remember the name.
2016-01-18 12:13:39 <ScrumpyJack> for example, i see that temp fs/shm is  218M, which will be a squeeze in the RPi A+ (which has 256Mb shared with graphics)
2016-01-18 12:18:12 <clandmeter> isnt tmfs size determined by the available memory?
2016-01-18 12:34:57 <ScrumpyJack> and by what's installed?
2016-01-19 09:11:38 <ScrumpyJack> morning all
2016-01-19 09:31:48 <fcolista> lol re #5010
2016-01-19 12:08:48 <ScrumpyJack> what's blender? (can't look it up on terminal)
2016-01-19 12:10:00 <ScrumpyJack> ok, i got it
2016-01-20 00:28:43 <rishoop1> Dropping a URL for 'at' daemon - once said it was buildable on Alpine https://github.com/sillelien/http-workqueue/tree/master/at-3.1.16
2016-01-20 00:28:54 <rishoop1> Haven't used myself but seen a couple people ask Alpine about it - good tool btw - bye
2016-01-20 00:47:33 <skarnet> seriously, the at daemon ?
2016-01-20 00:47:49 <skarnet> 1996 called, it wants its tools back
2016-01-20 07:24:27 <rishoop1> @skarnet, Yes the at daemon, as offered by many fine distros near you.  https://www.archlinux.org/packages/community/x86_64/at/ https://packages.gentoo.org/packages/sys-process/at https://packages.qa.debian.org/a/at.html http://packages.slackware.com/?r=slackware-current&p=at-3.1.12-i486-1.txz https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/7/html/System_Administrators_Guide/s1-autotasks-at-batch.html#sec
2016-01-20 07:24:45 <rishoop1> I saw two (2) people ask about at for Alpine. Package it already. Thx, bye.
2016-01-20 09:26:06 <fabled> clandmeter, any comments on http://patchwork.alpinelinux.org/patch/1387/ ?
2016-01-20 09:30:05 <clandmeter> fabled: go ahead and remove me.
2016-01-20 09:32:05 <clandmeter> would be nice to have a maintainer
2016-01-20 09:33:31 <fcolista> bah
2016-01-20 09:33:42 <fcolista> latex it's a nightmare
2016-01-20 10:40:58 <danilobuerger> hey guys, whats the difference between the community and testing package repos?
2016-01-20 11:00:13 <bpiotrowski> testing is only for edge users
2016-01-20 11:00:47 <bpiotrowski> and contains mostly untested packages
2016-01-20 11:01:05 <bpiotrowski> community is a repository for so-called trusted users
2016-01-20 11:01:24 <bpiotrowski> and besides that, it contains packages we don't want to maintain for 2 years in main
2016-01-20 11:01:30 <bpiotrowski> like golang projects
2016-01-20 11:02:51 <bpiotrowski> pretty much it
2016-01-20 11:03:41 <danilobuerger> ok, a while back someone mentioned i should consider moving the package i maintain from testing to community
2016-01-20 11:04:09 <danilobuerger> is there anything i should watch out for or just do a patch that moves it?
2016-01-20 11:05:12 <bpiotrowski> git mv is enough
2016-01-20 21:59:15 <dontbecold_> hey - apologies if this is the wrong place to ask, but i'm trying to build the ruby port for 2.3.0
2016-01-20 21:59:39 <dontbecold_> & it's complaining because <linux/random.h> is missing - is there a port that includes this?
2016-01-20 22:00:23 <skarnet> have you installed the linux-headers package?
2016-01-20 22:00:36 <dontbecold_> nope! sorry, bit new to ports in general
2016-01-20 22:02:27 <skarnet> np :) linux/*.h is a kernel header, same as foobar/*.h is generally a foobar header provided in the foobar-dev package :)
2016-01-20 22:03:13 <dontbecold_> amazing! installing the package worked
2016-01-20 22:03:23 <dontbecold_> thank you for your help :)
2016-01-20 22:04:26 <skarnet> gl!
2016-01-20 22:43:17 <dontbecold_> so, my package (ruby 2.3.0) builds and passes its tests successfully - what do I need to do to make an .apk file?
2016-01-20 22:43:46 <dontbecold_> abuild package just copies the package files to pkg/, google gives me a bunch of stuff about android
2016-01-21 08:30:37 <ScrumpyJack> morning
2016-01-22 03:49:08 <rishoop1> Re grsec 4 months now since news at http://forum.alpinelinux.org/forum/pax-grsecurity/grsec-patches-only-sponsors
2016-01-22 03:49:22 <rishoop1> Please update status someplace - Thx a lot - Bye
2016-01-22 08:49:44 <ScrumpyJack> morning climbers
2016-01-22 09:04:23 <ncopa> morning
2016-01-22 09:04:32 <ncopa> true climbers: http://www.dailymail.co.uk/news/article-1320728/Herd-mountain-goats-climb-160ft-near-vertical-Cingino-dam-Italian-Alps.html
2016-01-22 09:05:11 <wbx> fabled: hi
2016-01-22 09:05:56 <wbx> fabled: i know you are no longer interested in uClibc, but have you seen leonid email with a questin about a long outstandng bug you have fixed?
2016-01-22 09:10:38 <ncopa> wbx: url?
2016-01-22 09:11:39 <fabled> wbx, i probably saw it. i don't remember the details, we probably had it on alpine as patch since they did not apply it upstream for stream
2016-01-22 09:13:04 <ncopa> did we get gperftools run on alpine?
2016-01-22 09:14:37 <fabled> for some reason*
2016-01-22 09:18:53 <zenny1> morning
2016-01-22 09:20:34 <zenny1> Came across https://github.com/arachsys/containers which uses the user namespaces to isolate the container without requiring root privileges, completely in user space. Alpine's simplicity marries with this concept than docket, I guess. Have a look.
2016-01-22 09:21:55 <zenny1> ncopa: thanks for sharing the climber Ibex goats. I have seen in the high Himalayas similar goats thousands of meters high where no human can reach.
2016-01-22 09:21:57 <coredumb> user namespaces is hard blocked to non root users on Grsecurity because it'
2016-01-22 09:22:03 <coredumb> it's so unsecure
2016-01-22 09:22:05 <coredumb> *
2016-01-22 09:22:15 <wbx> http://lists.busybox.net/pipermail/uclibc/2013-June/047826.html
2016-01-22 09:22:37 <coredumb> Even Red Hat does block user namespace to non-root users by default
2016-01-22 09:23:15 <zenny1> coredumb, they do the same in debian/RH, but the way it has been deployed at https://github.com/arachsys/containers is interesting. Have a look.
2016-01-22 09:23:50 <coredumb> http://seclists.org/oss-sec/2015/q4/614
2016-01-22 09:24:22 <coredumb> zenny1: yeah except on RH you can revert by using a sysctl entry
2016-01-22 09:24:26 <coredumb> on grsec you can't
2016-01-22 09:24:45 <wbx> it might help to know which problem was fixed with the patch. leonid is doing some libdl cleanup/review.
2016-01-22 09:26:18 <coredumb> zenny1: I fail to understand how it's interesting when it relies on a subsystem that's been known to be highly unsecure and let loads of local root exploit over the years ;)
2016-01-22 09:28:55 <zenny1> coredumb: if you read the README, you will see how 'contain . /bin/bash' created an isolated zone without any access to the root fs.
2016-01-22 12:54:06 <coredumb> wget: can't execute 'ssl_helper': No such file or directory
2016-01-22 12:54:13 <coredumb> on a fresh 3.3.1 install
2016-01-22 12:54:19 <coredumb> is that a known issue ?
2016-01-22 12:55:08 <coredumb> oh ok wget doesn't have openssl in dep
2016-01-22 13:34:19 <ncopa> coredumb: i think ssl_helper is a helper program to give buybox wget https support
2016-01-22 13:34:28 <ncopa> without needing to link busybox with libssl
2016-01-22 13:34:44 <ncopa> i dont know if we provide an ssl_helper program
2016-01-22 13:34:55 <ncopa> in any case, you should be able to apk add wget
2016-01-22 13:35:00 <ncopa> to get gnu wget
2016-01-22 13:35:02 <ncopa> or use curl
2016-01-22 13:55:41 <coredumb> ncopa: actually I installed curl and it solved the issue
2016-01-22 13:56:46 <jirutka> ncopa: Talking about curl, I’ve used curl instead of wget in https://github.com/lxc/lxc/pull/751, but curl is not available by default on Alpine. Is it a problem?
2016-01-22 13:57:57 <ncopa> jirutka: i think not, because i suppose alpine itself does not need to fetch apk.
2016-01-22 13:58:44 <ncopa> jirutka: re bash vs sh, check checkbashisms
2016-01-22 13:58:55 <ncopa> which may help you convert it to posix shell
2016-01-22 13:59:33 <jirutka> well, that’s true, but I use it even to choose random mirror https://github.com/jirutka/lxc/blob/alpine-tmpl/templates/lxc-alpine.in#L126-L130 and resolve the latest relase branch https://github.com/jirutka/lxc/blob/alpine-tmpl/templates/lxc-alpine.in#L109-L115
2016-01-22 13:59:58 <jirutka> I will check it and replace bashisms, I just didn’t have time yet. :/
2016-01-22 14:00:28 <jirutka> I suppose that it’s okay if it will run with busybox sh (in default Alpine config)?
2016-01-22 14:01:02 <ncopa> if you use #!/bin/sh then it needs to work with dash i think
2016-01-22 14:01:12 <ncopa> as dash will be used on ubuntu/debian hosts
2016-01-22 14:02:20 <ncopa> hum
2016-01-22 14:02:37 <ncopa> curl may be a problem yes
2016-01-22 14:02:53 <jirutka> okay, I’ll try to replace it with wget
2016-01-22 14:03:01 <ncopa> busybox wget too
2016-01-22 14:03:12 <ncopa> so you get your arms tied behind your back :)
2016-01-22 14:03:27 <jirutka> Yeah, that the problem I have :(
2016-01-22 14:03:59 <jirutka> I’d like the script to be as clear, readable and robust as possible and that’s quite hard with such limited tools. But I’ll try it.
2016-01-22 14:05:39 <ncopa> looks like you cannot do the random_mirror_url that way with busybox wget
2016-01-22 14:05:47 <ncopa> i dont think it can give you the http headers
2016-01-22 14:06:45 <ncopa> it is a neat trick though
2016-01-22 14:12:42 <jirutka> Talking about mirrors, this list is not up-to-date; http://rsync.alpinelinux.org/alpine/MIRRORS.txt
2016-01-22 14:12:59 <jirutka> It’s mentioned on http://wiki.alpinelinux.org/wiki/Alpine_Linux:Mirrors
2016-01-22 14:15:56 <ncopa> better now?
2016-01-22 14:16:19 <jirutka> nope, repository.fit.cvut.cz is missing :P
2016-01-22 14:17:04 <jirutka> Also it would be very helpful to automatically create symlink for the latest apk-tools-static without the version number.
2016-01-22 14:17:43 <jirutka> It would allow us to get rid of this insanity https://github.com/jirutka/lxc/blob/alpine-tmpl/templates/lxc-alpine.in#L197-L201
2016-01-22 14:18:32 <ncopa> let me present some http get location insanity:
2016-01-22 14:18:36 <ncopa> echo -e "GET /cgi-bin/dl.cgi HTTP/1.0\r\nHost: wiki.alpinelinux.org\r\nConnection: close\r\n\r\n" | nc wiki.alpinelinux.org 80 | awk '$1 == "Location:" {print $2}'
2016-01-22 14:19:27 <ncopa> will spit out your random mirror from Location: http header using netcat only
2016-01-22 14:19:52 <jirutka> Looks cool, but what if you switch wiki to HTTPS only (that you surely should do ;)?
2016-01-22 14:20:30 <ncopa> i think it is available on https already
2016-01-22 14:20:42 <ncopa> but i dont think we will remove http support for a while
2016-01-22 14:21:01 <ncopa> in any case, if you need https, you'll need pull in some ssl library
2016-01-22 14:21:09 <jirutka> Nope, the cert is not valid, see http://bugs.alpinelinux.org/issues/4981
2016-01-22 14:21:09 <ncopa> busybox wget won't help you
2016-01-22 14:21:21 <ncopa> so you will need curl in any case there
2016-01-22 14:21:34 <jirutka> Ah, that’s right
2016-01-22 14:22:34 <jirutka> You use StartSSL CA, right?
2016-01-22 14:34:09 <ncopa> i think so
2016-01-22 14:34:29 <ncopa> we should probably better switch to that new https thingy which i forgot the name
2016-01-22 14:34:46 <jirutka> Let’s Encrypt?
2016-01-22 14:34:57 <ncopa> yes
2016-01-22 14:35:14 <ncopa> but last time i checked you need python on your production servers for it
2016-01-22 14:35:15 <jirutka> To be honest, I still use StartSSL, I’m not so excited from Let’s Encrypt.
2016-01-22 14:35:22 <ncopa> so i am waiting for c client
2016-01-22 14:35:27 <jirutka> There are already many alternative clients.
2016-01-22 14:35:39 <ncopa> probably
2016-01-22 14:35:57 <ncopa> i havent checked since it was in the news a few months ago
2016-01-22 14:36:08 <jirutka> This one is written in shell https://github.com/lukas2511/letsencrypt.sh
2016-01-22 14:36:14 <ncopa> cool :)
2016-01-22 14:36:44 <ncopa> bash
2016-01-22 14:36:46 <ncopa> non-cool :)
2016-01-22 14:36:48 <jirutka> It’s simpler than the official one that is huge monster trying to automagically handle all your configuration.
2016-01-22 14:37:17 <ncopa> i
2016-01-22 14:37:22 <jirutka> Then just install python and use https://github.com/diafygi/acme-tiny or https://github.com/felixfontein/acme-compact
2016-01-22 14:37:34 <ncopa> i'd rather use  lua over bash
2016-01-22 14:37:49 <ncopa> its less than half size
2016-01-22 14:37:53 <jirutka> I know.
2016-01-22 14:38:00 <jirutka> I don’t know about Lua client for ACME :(
2016-01-22 14:38:21 <ncopa> don't have time now though
2016-01-22 14:38:36 <jirutka> Then it’d be probably better to just stay with StartSSL.
2016-01-22 14:38:42 <jirutka> And just renew your certs.
2016-01-22 14:41:12 <ncopa> yes
2016-01-22 14:41:19 <ncopa> nangel should take care of that
2016-01-22 14:42:05 <ncopa> i cherry-picked the mirror list to 3.3-stable
2016-01-22 14:42:28 <ncopa> dont know how to deal with the mirrors properly
2016-01-22 14:42:49 <ncopa> i wish we had something that gave a close enough mirror on dns request
2016-01-22 14:43:14 <ncopa> so you could just use http://dl.alpinelinux.org everywhere
2016-01-22 14:43:33 <ncopa> and it would pick a good enough mirror
2016-01-22 14:43:47 <ncopa> as a sane default
2016-01-22 14:43:57 <ncopa> then we didnt need try pick one in any script
2016-01-22 14:45:14 <jirutka> If apk properly follows redirects, then it can be quite easily done, e.g. using Lua in nginx.
2016-01-22 14:46:08 <ncopa> it currently does not follow redirects?
2016-01-22 14:46:18 <jirutka> I dunno.
2016-01-22 14:46:27 <ncopa> it uses libfetch from bsd
2016-01-22 14:47:53 <jirutka> Sorry, my English kinda sucks, I’m not saying that it doesn’t handle it correctly now, only that if it handles it correctly, then it can be easily solved.
2016-01-22 14:48:20 <ncopa> where are you from?
2016-01-22 14:48:24 <jirutka> Czech Republic
2016-01-22 14:48:28 <ncopa> ok
2016-01-22 14:48:39 <ncopa> english is not my native lang either
2016-01-22 14:48:40 <ncopa> yes
2016-01-22 14:48:48 <ncopa> i think it handles redirects correctly
2016-01-22 14:49:01 <ncopa> the drawback is that it will do 2 http requests
2016-01-22 14:49:07 <ncopa> for each file
2016-01-22 14:49:51 <ncopa> i don't know how much impact that has
2016-01-22 14:50:07 <jirutka> Can I have a question about nss-pam-ldapd? I’m just installing Samba 4 on Alpine and I ran into problem with nss-pam-ldapd;
2016-01-22 14:50:07 <jirutka>  * Starting nslcd ...
2016-01-22 14:50:07 <jirutka> nslcd: Warning: NSS_LDAP module not loaded: Error loading shared library libnss_ldap.so.2: No such file or directory
2016-01-22 14:50:23 <ncopa> I think you actually can use http://wiki.alpinelinux.org/cgi-bin/dl.cgi/edge/main as repo
2016-01-22 14:51:03 <ncopa> i dont think musl supports nss
2016-01-22 14:51:58 <ncopa> pam should work
2016-01-22 14:52:30 <skarnet> indeed it doesn't, not as shared libraries loaded into the application anyway
2016-01-22 14:52:49 <ncopa> which is a good thing imho
2016-01-22 14:52:53 <jirutka> Hmm, then the package dependencies are incorrect; https://pkgs.alpinelinux.org/package/main/x86_64/nss-pam-ldapd
2016-01-22 14:53:25 <ncopa> i think they are correct
2016-01-22 14:53:34 <ncopa> just that we dont ship the nss part of it
2016-01-22 14:53:44 <ncopa> (what i gues at least)
2016-01-22 14:53:54 <ncopa> do you need the nss functionality?
2016-01-22 14:53:54 <jirutka> He? It’s named *nss*-pam-ldapd.
2016-01-22 14:54:11 <ncopa> yes, but i think it provides 2 different plugins
2016-01-22 14:54:25 <ncopa> one nss modules and one pam module
2016-01-22 14:54:26 <jirutka> Yeah, the only reason I need is to avoid explicitly creating entries in passwd for each Samba user.
2016-01-22 14:54:39 <ncopa> ok
2016-01-22 14:54:44 <ncopa> so you need the nss part
2016-01-22 14:55:19 <jirutka> Hmm, do you have any idea how to solve it differently?
2016-01-22 14:55:32 <ncopa> i think i have
2016-01-22 14:55:47 <ncopa> there is a protocol for nsdc(?)
2016-01-22 14:55:51 <ncopa> which is pretty simple
2016-01-22 14:56:02 <ncopa> that provides similar functionality
2016-01-22 14:56:07 <ncopa> i think musl supports it
2016-01-22 14:56:47 <ncopa> so basically, we only need a small prog that listens on a unix socket
2016-01-22 14:56:54 <ncopa> talking this protocol
2016-01-22 14:57:06 <ncopa> and returns passwd entries
2016-01-22 14:57:32 <ncopa> iow, there is a small service missing
2016-01-22 14:57:40 <ncopa> whcish should not be too hard to implement
2016-01-22 14:57:49 <jirutka> btw what about https://pkgs.alpinelinux.org/package/main/x86_64/samba-libnss-winbind ?
2016-01-22 14:58:02 <ncopa> i dont know really
2016-01-22 14:58:37 <ncopa> well i think it builds the plugin
2016-01-22 14:58:37 <ncopa> but i suspect its useless on a musl system
2016-01-22 14:58:48 <jirutka> But it has musl as a dependency, not any other libc lib.
2016-01-22 14:59:37 <ncopa> yes you can use any libc for *building* the plugin
2016-01-22 14:59:37 <ncopa> but you need another libc that actually loads it
2016-01-22 15:00:02 <ncopa> musl will happily build it, but it will not load it
2016-01-22 15:00:03 <jirutka> Ah, I understand. However, I think that it’d be very helpful to add this information to the description.
2016-01-22 15:00:39 <ncopa> i will not argue against that :)
2016-01-22 15:01:50 <ncopa> http://git.musl-libc.org/cgit/musl/tree/src/passwd/nscd_query.c
2016-01-22 15:01:56 <ncopa> "/var/run/nscd/socket"
2016-01-22 15:02:05 <jirutka> Does http://git.alpinelinux.org/ support something like pull requests?
2016-01-22 15:02:26 <ncopa> cgit does not support that
2016-01-22 15:02:27 <ncopa> but
2016-01-22 15:02:49 <ncopa> you can create pull request on http://github.com/alpinelinux/aports
2016-01-22 15:03:14 <ncopa> and we can git pull from github
2016-01-22 15:03:24 <jirutka> Ah, didn’t know about GH mirrors, great!
2016-01-22 15:03:59 <ncopa> for a single patch it is probably easier with git send-email
2016-01-22 15:04:26 <jirutka> Not for me, I use GitHub everyday, so I’m very comfortable with pull requests. :)
2016-01-22 15:04:31 <ncopa> if you get developers attention here on irc you can also do `git format-patch --stdout HEAD^ | tpaste`
2016-01-22 15:04:48 <ncopa> and I can just curl $url | git am
2016-01-22 15:05:03 <ncopa> works for trivial patches
2016-01-22 15:05:08 <jirutka> okay
2016-01-22 15:05:25 <ncopa> and only works if you have our attention here ;)
2016-01-22 15:05:37 <jirutka> I’m reading http://git.musl-libc.org/cgit/musl/tree/src/passwd/nscd_query.c, but I’m not sure what exactly it does
2016-01-22 15:05:55 <ncopa> i thin it is an nss alternative
2016-01-22 15:06:21 <ncopa> so that you can make the libc getpwent get the users from ldap
2016-01-22 15:06:33 <ncopa> i think thats the idea
2016-01-22 15:11:02 <jirutka> Ah, this commit message is useful:
2016-01-22 15:11:02 <jirutka> query a server over a Unix domain socket on /var/run/nscd/socket.
2016-01-22 15:11:02 <jirutka> when we fail to find the entry in the commonly accepted files,  we
2016-01-22 15:11:02 <jirutka> the protocol used here is compatible with glibc's nscd protocol on
2016-01-22 15:11:02 <jirutka> most systems (all that use 32-bit numbers for all the protocol fields,
2016-01-22 15:11:03 <jirutka> which appears to be everything but Alpha).
2016-01-22 15:16:59 <ncopa> right
2016-01-22 15:17:39 <ncopa> so we need a nsdc server talking the nsdc protocol, that uses libldap or samba to get the user account info
2016-01-22 15:20:11 <ncopa> hum
2016-01-22 15:20:23 <ncopa> i wonder if the nsdc thingy is actually used?
2016-01-22 15:21:37 <ncopa> yes, getgrouplist
2016-01-22 15:22:46 <ncopa> getpwnam too
2016-01-22 15:22:47 <ncopa> ok
2016-01-22 15:22:50 <ncopa> should work
2016-01-22 15:23:29 <jirutka> btw they have “Non-glibc-based nscd-protocol backend for LDAP (and perhaps NIS)” in the roadmap http://wiki.musl-libc.org/wiki/Roadmap
2016-01-22 15:23:47 <ncopa> yes
2016-01-22 15:24:06 <ncopa> i dont think it is hard to implement
2016-01-22 15:24:34 <jirutka> I just wonder if I can implement it in Lua, Python, or… anything else than C/C++ :)
2016-01-22 15:26:01 <jirutka> I know many langs, but all of them are high-level langs, no one for system programming.
2016-01-22 15:26:44 <ncopa> lua is ok
2016-01-22 15:26:49 <ncopa> but unix socket in lua...
2016-01-22 15:26:59 <ncopa> i think it is easier to do in C
2016-01-22 15:27:19 <ncopa> i shoudl get back to debug python libffi issue here
2016-01-22 15:27:46 <ncopa> oh lua-socket supports unix socket iirc
2016-01-22 15:27:55 <ncopa> i think it should be doable
2016-01-22 15:28:17 <jirutka> I can’t find any documentation for the nscd protocol, just C sources.
2016-01-22 15:28:25 <ncopa> :)
2016-01-22 15:28:27 <ncopa> same :)
2016-01-22 15:29:12 <skarnet> yes.
2016-01-22 15:29:31 <skarnet> I have looked for a while, and Denys' unscd source is probably the most readable "documentation"
2016-01-22 15:29:34 <ScrumpyJack> i don't think nscd is "protocol" related. it's just a [in]convenience :)
2016-01-22 15:30:45 <ScrumpyJack> "protocol" related, in so far as "this is the way you should write a caching daemon"
2016-01-22 15:32:09 <jirutka> You mean this https://github.com/keymon/unscd/blob/master/nscd-0.47.c ?
2016-01-22 15:32:29 <ScrumpyJack> or do you mean "how to add a cache to nscd"?
2016-01-22 15:32:36 <skarnet> the protocol is just the way the client and server communicate over the unix socket.
2016-01-22 15:32:42 <jirutka> I know
2016-01-22 15:32:50 <skarnet> You need to understand it to write a daemon. :P
2016-01-22 15:34:15 <skarnet> jirutka: apparently that's an old version. I was thinking of https://busybox.net/~vda/unscd/nscd-0.52.c
2016-01-22 15:34:25 <jirutka> Okay, we should probably talk about format of the messages, not protocol.
2016-01-22 15:35:41 <skarnet> that's exactly what "protocol" means!
2016-01-22 15:36:44 <jirutka> Hmm, that’s right… I’m confused now, so what did you mean with your previous comments?
2016-01-22 15:38:26 <skarnet> the latest unscd source I know is 0.52, which I linked. It's probably the best thing to reverse-engineer to understand the details of the nscd protocol.
2016-01-22 15:38:26 <skarnet> That's all I meant.
2016-01-22 15:38:40 <jirutka> ah, okay
2016-01-23 12:12:45 <Bim60> What is roadmap on 3.3.2 release
2016-01-23 12:28:11 <ncopa> http://bugs.alpinelinux.org/projects/alpine/issues?fixed_version_id=104&set_filter=1&status_id=o
2016-01-23 13:34:53 <skazz> Hello, I noticed a problem running grsec on hyper-v (not sure if it affects other visualization too), but vanilla works fine (3.3.1), I would be nice if this was mentioned somewhere for others trying to get it working in a VM.
2016-01-24 11:13:00 <faffolter> ncopa: can you please pull git://git.alpinelinux.org/user/fab/aports
2016-01-24 11:14:39 <faffolter> those updates are now sitting there for quite awhile
2016-01-24 11:14:55 <faffolter> thanks
2016-01-24 23:40:33 <AlecTaylor> hi
2016-01-24 23:40:51 <_mjones> Hello.
2016-01-24 23:41:09 <AlecTaylor> Just made some minor changes (version bump) to an APK. I have a patch file ready. How do I join the mailing-list to post it?
2016-01-24 23:41:54 <_mjones> So, I'm not certain if the team is favouring mailing-list or github pull-requests at the moment.
2016-01-24 23:42:22 <_mjones> maillist info here: https://alpinelinux.org/community/
2016-01-24 23:42:38 <AlecTaylor> Oh, there's a github as well? - I cloned git://git.alpinelinux.org/aports
2016-01-24 23:53:41 <AlecTaylor> Emailed patch through 6 minutes ago.
2016-01-25 00:15:38 <_mjones> Good for you. Better off than me, who has whole piles of such changes I haven't submitted because $reasons.
2016-01-25 00:33:28 <AlecTaylor> xD
2016-01-25 00:33:29 <AlecTaylor> Well it's such
2016-01-25 02:47:33 <mitchty> ok question on getting alpine linux up on an unsupported arm soc
2016-01-25 02:48:27 <mitchty> there are debian and arch ports to this board so no worries there, but alpine linux is missing the dtb files for this currently and it uses its own fork of uboot
2016-01-25 02:49:51 <mitchty> any suggestions/pointers on where I should start?
2016-01-25 02:50:47 <mitchty> current plan was to copy the dtb's from debian into 3.3.1 and build the uboot files myself and try those out, but if anyone has suggestions i'm all ears
2016-01-25 02:53:07 <mitchty> but lets say that works, what has to happen to get this into the generic arm tar.gz?
2016-01-25 03:14:44 <_mjones> one sec mitchty for me to look that over.
2016-01-25 03:20:19 <mitchty> so what i'm trying to get running is https://www.solid-run.com/freescale-imx6-family/hummingboard/
2016-01-25 03:20:21 <mitchty> those guys
2016-01-25 03:20:31 <mitchty> they show up as hummingboard2 on the dtbs
2016-01-25 03:21:01 <_mjones> ok, so essentially I was trying to figure out the same thing.
2016-01-25 03:21:18 <_mjones> I was working with qemu before hardware, planning on a odroid board.
2016-01-25 03:21:48 <_mjones> But I think you can use the same uboot from debian?
2016-01-25 03:21:55 <_mjones> Let me think about that for a minute.
2016-01-25 03:22:37 <_mjones> Yes, the dtb is definitely the same and can copy over.
2016-01-25 03:23:29 <mitchty> so that was basically my plan for the moment to get it running
2016-01-25 03:23:49 <mitchty> but would like to get it building and in the normal arm distribution if possible
2016-01-25 03:24:24 <_mjones> Hopefully the dtb is redistributable, so I don't see a problem there.
2016-01-25 03:24:57 <_mjones> To tell you the truth I'm not sure what the strategy is going to be for supporting additional armhf machines. I'm still figuring it out myself.
2016-01-25 03:52:50 <mitchty> well its got 4g ram that is pretty much my main reason for buying it
2016-01-25 03:53:11 <_mjones> Very nice. That's a very open soc.
2016-01-25 03:53:30 <mitchty> i'm also going to make it a gps time source cause why not
2016-01-25 03:53:32 <_mjones> I'm waiting for some more armv8 boards. Even so the best I've seen so far is 3G.
2016-01-25 03:53:50 <mitchty> yeah this isn't super fast cpu wise but memory was my main concern
2016-01-25 03:54:07 <mitchty> the 2g boards hit linker oom killer issues
2016-01-25 03:54:18 <_mjones> With ghc? ;-)
2016-01-25 03:54:36 <mitchty> heh, well actually lto and llvm
2016-01-25 03:54:46 <_mjones> I have a 128mb armel box that, before I bricked it,  was working the toolchain surprisingly well.
2016-01-25 03:54:47 <mitchty> but i can get 2.5g process size in ghc easy
2016-01-25 03:55:03 <_mjones> lto I can dig.
2016-01-25 03:55:12 <mitchty> there are a few quadratic things in the compiler
2016-01-25 03:55:20 <mitchty> if you hit them 8g can happen fast
2016-01-25 03:55:39 <_mjones> The 900mb install footprint is fun too.
2016-01-25 03:55:40 <mitchty> generally bugs though
2016-01-25 03:55:45 <_mjones> Thank everybody for that one.
2016-01-25 03:55:50 <mitchty> half that is profile archive libraries
2016-01-25 03:56:18 <mitchty> i could compile it with default and threaded support only and that would halve
2016-01-25 03:56:36 <mitchty> wouldn't be as useful though
2016-01-25 03:56:48 <mitchty> i like to be able to profile code
2016-01-25 03:58:02 <mitchty> but anyway time to cross my fingers
2016-01-25 03:59:07 <_mjones> Do you have the board yet or just ordering?
2016-01-25 03:59:13 <mitchty> nope got it
2016-01-25 03:59:39 <mitchty> came from israel
2016-01-25 04:00:09 <mitchty> hold on let me ssh into their mini linux ramdisk thing
2016-01-25 04:05:15 <_mjones> If you're just interested in getting it up and running, and not so much in making clean installers, you could replace the stock dist with Alpine easily enough, I imagine.
2016-01-25 04:05:37 <_mjones> I guess mkinitfs will just work.
2016-01-25 04:10:12 <mitchty> finicky sd card
2016-01-25 04:10:13 <mitchty> https://gist.githubusercontent.com/mitchty/bb2fe51afd44c3f36ebc/raw/f60f02abdc60f443e573bb41a3dc526546a63077/gistfile1.txt
2016-01-25 04:11:16 <mitchty> i wasn't sure if it would only show 3g ram due to some weird 2/1 split
2016-01-25 04:20:23 <mitchty> heh, well this might be a bit more involved, need to figure out their boot specifics
2016-01-25 04:20:40 <mitchty> it kinda worked
2016-01-25 04:20:51 <mitchty> just didn't find the kernel
2016-01-25 04:21:16 <_mjones> the odroids have a reputation for being finicky about sdcards, but the truth of the matter is that sdcards vary dramatically in performance, and it can make a big difference.
2016-01-25 04:21:50 <_mjones> Ah, for a moment I thought there was a new algitbot xkcd.
2016-01-25 04:29:37 <mitchty> yeah, i'm just going to nfs mount things and compile off that
2016-01-25 04:29:46 <mitchty> it sounds crazy but its faster
2016-01-25 04:30:55 <_mjones> It's pretty typical with pro boards actually.
2016-01-25 04:31:20 <_mjones> easier to bring up ethernet and nfs than hw, block layer and filesystem. ;-)
2016-01-25 12:49:39 <clandmeter> pcket lss.. lots of it
2016-01-25 12:53:08 <clandmeter> nl.a.o bugs and the likes will have issues.
2016-01-25 19:29:39 <_mjones> Is there a reason we don't have /proc/config.gz enabled in the kernel build? Super-handy piece of self-documentation. And yes, I know about /boot/config-grsec etc.
2016-01-25 19:44:03 <skazz> It's there on the pi, is that not valid?
2016-01-25 20:50:11 <barthalion> _mjones: you need a module for that…
2016-01-25 20:50:42 <barthalion> but I don't remember the name
2016-01-25 20:51:25 <_mjones> Yes, it's kernel compile-time option. I haven't put in a bug, but I'd like to have it on by default.
2016-01-25 20:51:57 <_mjones> Debian doesn't, but they're recalcitrant about it for no reason. There's no downside of which I'm aware, but if there is I'd like to know it.
2016-01-25 20:56:05 <barthalion> it's kernel compile-time option that we have enabled as a module
2016-01-25 20:56:19 <skarnet> _mjones: I guess the reason is "it's only useful to people who will know how to 'modprobe configs' if they need it"
2016-01-25 20:56:57 <skarnet> the module is "configs"
2016-01-25 20:57:12 <barthalion> skarnet: domo arigato
2016-01-25 20:57:22 <barthalion> just googled it, you were faster though
2016-01-25 20:58:31 <skarnet> barthalion: grep config /lib/modules/*/modules.dep :)
2016-01-25 20:59:35 <_mjones> skarnet, barthalion: oh, I missed what you were saying when you said module. I assumed it wasn't present as a .ko and didn't look.
2016-01-25 21:02:26 <_mjones> danke gents.
2016-01-25 21:02:57 <_mjones> Although I still scratch my head about why not compile it in. Must be like 1k of code or something.
2016-01-25 21:03:37 <skarnet> that's 1k of RAM that can never be swapped.
2016-01-25 21:03:58 <skarnet> why compile it in if most people don't need it and the people who need it know how to modprobe ? :P
2016-01-25 21:07:31 <_mjones> But you can say that about everything. Except maybe filesystems.
2016-01-25 21:07:38 <_mjones> It's features all the way down. ;-)
2016-01-25 21:09:39 <skarnet> well yes... a lsmod on my Alpine actually shows a lot of modules :P
2016-01-25 21:11:14 <skarnet> (there should be a "turtle" module that depends on itself. Really.)
2016-01-26 10:13:37 <ScrumpyJack> morning all
2016-01-26 11:06:39 <skarnet> ncopa (or anyone who knows): what are the minimal requirements for running abuild on a from-scratch system? i.e. what does abuild depend on ?
2016-01-26 11:07:24 <skarnet> (context: I'm building a machine from scratch but would like to use apk as a package manager, so I just need to build from hand until I can make abuild work; I need to know where that point is.)
2016-01-26 11:07:40 <ncopa> hard dependencies is busybox iirc
2016-01-26 11:07:59 <ncopa> you will probably also need build-deps for building anything
2016-01-26 11:08:08 <ncopa> for gcc, make libc-dev etc
2016-01-26 11:08:22 <skarnet> build-deps?
2016-01-26 11:08:30 <clandmeter> build-base?
2016-01-26 11:08:47 <skarnet> busybox is a given, build environment (libc et al.) is a given
2016-01-26 11:08:49 <ncopa> build-base
2016-01-26 11:09:03 <ncopa> for https you will need curl
2016-01-26 11:09:09 <ncopa> or gnu wget
2016-01-26 11:09:23 <skarnet> I have no Alpine packages atm, just the stuff I've built from scratch myself. I can't install "build-base", I need to know what components are in it.
2016-01-26 11:09:23 <ncopa> i think you also need gnu tar and openssl
2016-01-26 11:09:53 <skarnet> I have busybox, gnu tar, openssl (from libressl), curl and gnu wget
2016-01-26 11:10:05 <ncopa> then i think you sould be ok
2016-01-26 11:10:07 <skarnet> and a native musl toolchain
2016-01-26 11:10:26 <ncopa> i did try to keep number of deps low
2016-01-26 11:10:41 <ncopa> fakeroot
2016-01-26 11:10:46 <ncopa> you need fakeroot
2016-01-26 11:10:59 <skarnet> ugh
2016-01-26 11:11:05 <ncopa> it is so permissions in tarball is root:root even if you dont run as root
2016-01-26 11:11:19 <skarnet> I hate that stuff
2016-01-26 11:11:28 <ncopa> im not too fond of it either
2016-01-26 11:11:48 <skarnet> gnu tar allows that
2016-01-26 11:12:06 <ncopa> in theory, what we want is: create tarball, set this and this owner/permission in tarball for this and that file
2016-01-26 11:12:08 <ncopa> hm
2016-01-26 11:12:10 <skarnet> tar --owner=0 --group=0 --numeric-owner
2016-01-26 11:12:31 <ncopa> can you set owner on given files?
2016-01-26 11:12:42 <skarnet> that will set every file in the tarball as root:root
2016-01-26 11:13:01 <ncopa> sometimes we need some files be owned by something else
2016-01-26 11:13:01 <skarnet> good question
2016-01-26 11:13:04 <ncopa> also
2016-01-26 11:13:27 <skarnet> but if fakeroot doesn't need special permissions to run, then there's no reason why tar can't do it
2016-01-26 11:13:28 <ncopa> tar format has both numeric id + user/group name
2016-01-26 11:13:57 <ncopa> correct, there is nothing that prevents us to fix tar, or write our own tar
2016-01-26 11:14:25 <ncopa> i think we already have partial own tar
2016-01-26 11:14:29 <skarnet> let me see if tar can't already do it
2016-01-26 11:14:31 <ncopa> for embedding checksums and signatures
2016-01-26 11:14:52 <ncopa> we probably want to provide some permission map file or so
2016-01-26 11:15:02 <skarnet> what
2016-01-26 11:15:10 <skarnet> you patched tar?
2016-01-26 11:15:12 <ncopa> so you can have a list of files
2016-01-26 11:15:29 <ncopa> no
2016-01-26 11:15:31 <ncopa> well yes
2016-01-26 11:15:35 <ncopa> but thats other thing
2016-01-26 11:15:46 <skarnet> anyway this is to bootstrap abuild, so I have to use vanilla packages up to there
2016-01-26 11:15:51 <ncopa> http://wiki.alpinelinux.org/wiki/Include:Abuild-tar
2016-01-26 11:16:11 <ncopa> you need build the tools in abuild repository
2016-01-26 11:16:47 <ncopa> well, some of them at least
2016-01-26 11:17:21 <ncopa> http://git.alpinelinux.org/cgit/abuild/tree/abuild-tar.c
2016-01-26 11:17:36 <skarnet> I can build anything C
2016-01-26 11:18:06 <skarnet> sure
2016-01-26 11:18:20 <skarnet> but I want to avoid playing with LD_PRELOAD
2016-01-26 11:18:30 <skarnet> so, work w/o fakeroot
2016-01-26 11:18:57 <skarnet> because I need my stuff to work with static builds
2016-01-26 11:19:05 <skarnet> and I can't guarantee I'll have a libc.so
2016-01-26 11:19:06 <ncopa> curently i think we have fakeroot hardcoded in abuild
2016-01-26 11:19:16 <ncopa> might be it will work if you run it as root
2016-01-26 11:19:36 <ncopa> i also have a stupid abuild-fetch.c helper
2016-01-26 11:19:47 <ncopa> to create lockfiles on nfs
2016-01-26 11:19:51 <ncopa> http://git.alpinelinux.org/cgit/abuild/tree/abuild-fetch.c
2016-01-26 11:20:03 <ncopa> flock and nfs is very nasty on linux
2016-01-26 11:20:34 <skarnet> it's 2016, why do you think I care about nfs
2016-01-26 11:21:32 <skarnet> but yeah, you have a rootpkg() function
2016-01-26 11:22:01 <skarnet> I need to understand when this is needed...
2016-01-26 11:28:09 <skarnet> respect for you increasing tenfold when I realize the amount of shell scripting you had to do
2016-01-26 11:30:29 <skarnet> so, the command executed under fakeroot is abuild itself, which is a shell script
2016-01-26 11:30:39 <skarnet> and the shell is bb
2016-01-26 11:31:04 <skarnet> so that means I need to dynamically link my bb, that's out of the question
2016-01-26 11:31:42 <skarnet> I guess I could have a specific dynamic shell somewhere and use it as an interpreter for only abuild
2016-01-26 11:31:59 <skarnet> but no, that wouldn't work since spawned commands would be static
2016-01-26 11:36:27 <skarnet> eh, I'll build as root for now, and figure out the hard stuff later on
2016-01-26 11:39:50 <skarnet> hm, you also have a dependency to pkg-config in the Makefile, for the openssl paths
2016-01-26 11:40:19 <skarnet> if you had a ?= instead of a = (just as for prefix and other variables) then it would be easily overridable
2016-01-26 12:20:13 <skarnet> is there a way to install Alpine on a rpi in non-diskless mode ?
2016-01-26 12:53:46 <ncopa> i suppose apk add --root /mounted/mmc --initdb --arch alpine-base linux-rpi
2016-01-26 12:53:52 <ncopa> or similar
2016-01-26 12:54:06 <ncopa> --arch armhf
2016-01-26 12:54:08 <ncopa> btw
2016-01-26 12:54:25 <ncopa> did you see that there are some unofficial armel5 port?
2016-01-26 12:54:42 <ScrumpyJack> ncopa: how is the new job going?
2016-01-26 12:54:59 <ncopa> ScrumpyJack: good thanks
2016-01-26 12:58:35 <ScrumpyJack> can you tell us more about it? :)
2016-01-26 12:59:33 <skarnet> ncopa: what do you mean? my colleague followed the instructions at http://wiki.alpinelinux.org/wiki/Raspberry_Pi , at what point should he do something different to install the system on the mmc ?
2016-01-26 13:01:13 <ncopa> oh , i didnt know about that wiki page
2016-01-26 13:12:03 <clandmeter> ncopa: i dont think you can install on rpi like that?
2016-01-26 13:12:10 <clandmeter> boot needs to be on fat32
2016-01-26 13:14:52 <ncopa> ok :)
2016-01-26 13:15:54 <clandmeter> or maybe uboot images work differently?
2016-01-26 13:23:43 <skarnet> clandmeter: do you have a quick recipe to turn a diskless rpi installation into a sd card installation?
2016-01-26 13:44:52 <clandmeter> skarnet: i never tried that, maybe fabled knows.
2016-01-26 13:45:44 <skarnet> thanks. I'll wait for fabled.
2016-01-26 16:17:40 <laurentius77> Good evening
2016-01-26 16:18:56 <laurentius77> Ițm trying to compile ipop-tincan for Alpine Linux using this tutorial which is a litlle outdated https://github.com/ipop-project/documentation/wiki/Building-the-code-for-Linux
2016-01-26 16:20:07 <laurentius77> Please is anyone who can help me in this somehow...even some ideeas are welcomed
2016-01-26 16:21:26 <laurentius77> it seems that gclient config --name=trunk http://webrtc.googlecode.com/svn/branches/3.52 is no longer an option, I get an error on that step
2016-01-26 16:22:43 <laurentius77> It seems that must be replaced by fetch chromium
2016-01-26 16:23:13 <laurentius77> or fetch webrtc
2016-01-26 16:24:17 <laurentius77> but the fetching proccess is finished with an error
2016-01-26 16:24:43 <laurentius77> please give me some help
2016-01-26 21:46:34 <PHLAK> I submitted a bug and a patch for the OpenVPN package a week ago and haven't seen any responses to either. Can someone take a look and tell me if I need to do anything else or what my next steps should be?
2016-01-26 21:46:40 <PHLAK> Patch: http://lists.alpinelinux.org/alpine-aports/2195.html
2016-01-26 21:47:56 <PHLAK> I believe you are the maintainer @ncopa
2016-01-26 21:48:02 <PHLAK> ncopa: ^
2016-01-26 21:48:25 <ncopa> PHLAK: hi
2016-01-26 21:48:26 <ncopa> sorry
2016-01-26 21:48:32 <ncopa> i have hard time to keep up
2016-01-26 21:48:50 <PHLAK> no problem, I've never contributed before so I might not have done things properly
2016-01-26 21:49:16 <ncopa> its %1430 i suppos
2016-01-26 21:50:27 <ncopa> ok
2016-01-26 21:50:35 <ncopa> i know why i didnt apply it :)
2016-01-26 21:50:55 <ncopa> you also need to do:
2016-01-26 21:50:59 <ncopa> abuild checksum
2016-01-26 21:51:05 <ncopa> and increase the pkgrel
2016-01-26 21:51:16 <PHLAK> how do I do that?
2016-01-26 21:51:25 <ncopa> edit the APKBUILD
2016-01-26 21:51:36 <ncopa> and change the value of pkgrel to +1
2016-01-26 21:51:57 <ncopa> otherwise the package would be built with exact same version number
2016-01-26 21:52:14 <ncopa> and in fact, due to that the builder will not even try rebuild it
2016-01-26 21:52:24 <ncopa> i can do it for you this time
2016-01-26 21:53:18 <PHLAK> thanks
2016-01-26 21:53:27 <ncopa> like this: http://tpaste.us/36nr
2016-01-26 21:53:41 <ncopa> i will include that in your commit
2016-01-26 21:53:48 <ncopa> so you will stay as author
2016-01-26 21:54:16 <PHLAK> great
2016-01-26 21:54:20 <PHLAK> anything else I need to do?
2016-01-26 21:54:32 <ncopa> i think not
2016-01-26 21:54:38 <ncopa> just to explain the workflow a bit
2016-01-26 21:54:46 <ncopa> git master is the development branch
2016-01-26 21:55:06 <ncopa> we normally don't touch the stable branches (eg for v3.3 repository) unless explicitly asked to
2016-01-26 21:55:20 <ncopa> in this case we probably want fix it for v3.3 too
2016-01-26 21:55:25 <ncopa> and not only 'edge'
2016-01-26 21:55:38 <PHLAK> right
2016-01-26 21:56:10 <ncopa> the Fixes #5027 in commit message will mark the ticket as resolved in bugtracker
2016-01-26 21:56:13 <ncopa> as it is now
2016-01-26 21:56:20 <ncopa> hum
2016-01-26 21:56:21 <ncopa> no
2016-01-26 21:56:44 <ncopa> probably because it needs to be 'fixes' with lowercase and not 'Fixes'
2016-01-26 21:56:57 <ncopa> ah, no
2016-01-26 21:57:16 <ncopa> the syntax for autocolose it is: 'fixes #5027'
2016-01-26 21:57:22 <ncopa> ok
2016-01-26 21:57:26 <ncopa> np problem then
2016-01-26 21:57:28 <PHLAK> ah okay
2016-01-26 21:58:05 <ncopa> there is a risk doing so, and that is that it will not show up on the list of bugs needs to be looked at before next 3.3.x release
2016-01-26 21:58:22 <ncopa> so whn i commit in git master i normally do 'ref #5027'
2016-01-26 21:58:38 <ncopa> and when I cherry-pick it to stable brnach i change it to 'fixes ...'
2016-01-26 21:58:49 <ncopa> i am not 100% happy with how redmine handles that
2016-01-26 21:58:54 <ncopa> but thats another story :)
2016-01-26 21:58:59 <PHLAK> okay,
2016-01-26 21:59:00 <ncopa> thank you for your patch!
2016-01-26 21:59:04 <PHLAK> I'll keep that in mind for the future
2016-01-26 21:59:08 <PHLAK> you're welcome
2016-01-26 21:59:59 <ncopa> nice commit message
2016-01-26 22:00:33 <ncopa> i love it when commit message it bigger than the change :)
2016-01-26 22:01:56 <kunkku> is there any policy on how pkgrel should be handled when fixing a problem in stable branch only?
2016-01-26 22:02:43 <kunkku> it can happen that multiple branches have the same pkgver
2016-01-26 22:02:58 <kunkku> but different pkgrel
2016-01-26 22:03:14 <kunkku> even such that an older branch has greater pkgrel
2016-01-26 22:03:21 <ncopa> the different branches are to be thought as separate entities
2016-01-26 22:03:33 <ncopa> thats how i have viewed it so far
2016-01-26 22:04:02 <ncopa> so yes, the pkgrel may differ for same fix in different branches
2016-01-26 22:04:17 <kunkku> so you are always supposed to use the -a flag when upgrading between branches?
2016-01-26 22:04:33 <ncopa> PHLAK: do we want upgrade openvpn version too for v3.3?
2016-01-26 22:04:42 <ncopa> there is 2.3.10 out
2016-01-26 22:05:06 <ncopa> kunkku: yes, if you upgrade from one branch to another you need to use -a
2016-01-26 22:05:22 <kunkku> some distros use multi-level pkgrel
2016-01-26 22:05:23 <ncopa> which will result in full reinstall
2016-01-26 22:05:27 <PHLAK> ncopa: I wont worry about that, I'm fine with it as-is
2016-01-26 22:05:37 <ncopa> kunkku: yes, we dont atm
2016-01-26 22:05:38 <PHLAK> just needed the down script to work properly
2016-01-26 22:05:56 <ncopa> kunkku: but we may want to change that
2016-01-26 22:06:15 <kunkku> for any other reasons than ^^ ?
2016-01-26 22:06:30 <ncopa> may help with pinning
2016-01-26 22:06:37 <ncopa> eg pick one package from one repo
2016-01-26 22:06:39 <ncopa> and mix them
2016-01-26 22:06:47 <ncopa> i am not too happy with mixing repos
2016-01-26 22:06:53 <ncopa> as it creates some nasty problems
2016-01-26 22:06:59 <kunkku> indeed
2016-01-26 22:07:11 <ncopa> like with libstdc++ abi breakage
2016-01-26 22:08:19 <ncopa> PHLAK: do you mind if i do upgrade openvpn while at it?
2016-01-26 22:09:37 <kunkku> btw my log file permission hardening project got stuck to an apk-related problem
2016-01-26 22:10:13 <ncopa> actually, we can not upgrade openvpn in v3.3 branch
2016-01-26 22:10:19 <ncopa> they removed --enable-password-save
2016-01-26 22:11:17 <kunkku> I wanted to set mode 2755 on /var/log
2016-01-26 22:11:27 <PHLAK> ncopa: I don't mind
2016-01-26 22:11:38 <PHLAK> it should be upgraded at some point anyway
2016-01-26 22:11:59 <PHLAK> ahh, okay yeah don't upgrade then
2016-01-26 22:12:07 <ncopa> https://community.openvpn.net/openvpn/wiki/ChangesInOpenvpn23#OpenVPN2.3.9
2016-01-26 22:12:11 <ncopa> yeah
2016-01-26 22:12:17 <kunkku> this is fine until you install a package that creates subdirectory in /var/log
2016-01-26 22:12:39 <ncopa> kunkku: it will reset the permissions on /var/log?
2016-01-26 22:12:45 <kunkku> then /var/log mode reverts to 0755, root:root
2016-01-26 22:12:56 <ncopa> because that is what it is in the .apk
2016-01-26 22:13:01 <kunkku> yes
2016-01-26 22:13:08 <ncopa> i dunno how to solve that
2016-01-26 22:13:16 <ncopa> should mention that for fabled
2016-01-26 22:13:33 <ncopa> maybe he can come up with some brilliant solution :)
2016-01-26 22:13:43 <kunkku> he suggested adding directory permission priority to .apk
2016-01-26 22:13:56 <ncopa> sounds like adding complexity to me
2016-01-26 22:13:57 <ncopa> :)
2016-01-26 22:14:15 <kunkku> to me also, but I don't know any better solution
2016-01-26 22:14:51 <ncopa> reminds me of one time i complained to him that apk is not 'simple'
2016-01-26 22:15:01 <kunkku> use 'tar d' to take /var/log out?
2016-01-26 22:15:12 <ncopa> he responded: "the problem it solves is complex"
2016-01-26 22:15:45 <ncopa> i think the "correct" thing to do is set permission in APKBUILD on /var/log
2016-01-26 22:15:47 <ncopa> but hum
2016-01-26 22:15:58 <ncopa> thats not practical
2016-01-26 22:16:03 <kunkku> then that has to be done everywhere
2016-01-26 22:16:10 <ncopa> not even possible actually
2016-01-26 22:16:21 <ncopa> because what happens if you miss it one single place?
2016-01-26 22:16:25 <ncopa> then things goes boom
2016-01-26 22:16:34 <kunkku> right...
2016-01-26 22:16:38 <ncopa> so i think we actually need some kind of priority
2016-01-26 22:17:04 <ncopa> what do do if different packages say that permissions should be different
2016-01-26 22:17:07 <ncopa> who wins?
2016-01-26 22:17:43 <kunkku> for files, there is such a priority setting already
2016-01-26 22:18:21 <ncopa> makes sense to do it for dirs too then
2016-01-26 22:18:22 <ncopa> ok
2016-01-26 22:18:25 <ncopa> i better go to sleep
2016-01-26 22:21:13 <ncopa> good night
2016-01-27 03:18:21 <mitchty> quick question, if i built 4.4 without grsecurity will that still be ok to run the userland?
2016-01-27 04:35:41 <_mjones> yes.
2016-01-27 04:35:57 <_mjones> There's also the -vanilla kernel in repos, that doesn't have pax/grsec.
2016-01-27 04:36:08 <_mjones> You don't have to build it if you don't want.
2016-01-27 08:29:21 <ScrumpyJack> morning
2016-01-27 08:52:02 <ScrumpyJack> speaking of grsec, what's the news on that?
2016-01-27 08:52:17 <ScrumpyJack> will it be in Alpine's future?
2016-01-27 09:00:57 <bpiotrowski> nothing new so far
2016-01-27 09:26:45 <ScrumpyJack> there was some noise a while ago about trying to get some elf binary signing support on Alpine. Can anyone remember the method?
2016-01-27 10:29:34 <skarnet> ncopa: I had to do this for abuild to build nicely and configurably. might want to review and apply it. http://pastebin.com/u25v3AED
2016-01-27 10:30:18 <skarnet> in particular: CFLAGS-$@ is used, and it resolves to CFLAGS-abuild-tar.o instead of CFLAGS-abuild-tar :)
2016-01-27 10:30:32 <ncopa> lgtm
2016-01-27 10:30:33 <skarnet> hence the variable name change.
2016-01-27 10:30:54 <ncopa> git commit && git format-patch -1 --stdout | tpaste
2016-01-27 10:31:09 <ncopa> so you get logged as author
2016-01-27 10:31:25 <skarnet> sucks that pkg-config doesn't have a separate --ldflags, but meh.
2016-01-27 10:32:48 <skarnet> meh, it's not on my regular Alpine machine. idc about being logged as author. If it's important to you (for git blame :)) then give me a minute...
2016-01-27 10:33:11 <ScrumpyJack> grr. what's the nicest way to build from this repo? https://git.drogon.net/?p=wiringPi
2016-01-27 10:33:51 <ScrumpyJack> no tarball release
2016-01-27 10:34:30 <ncopa> ScrumpyJack: i'd create a tarball form a git commit
2016-01-27 10:34:35 <ncopa> and upload it some place
2016-01-27 10:35:31 <ScrumpyJack> do we have an official place for that?
2016-01-27 10:35:56 <ScrumpyJack> (trust and all that)
2016-01-27 10:37:49 <skarnet> ... why won't patch apply a patch obtained by git diff? o.O
2016-01-27 10:41:11 <skarnet> anyway, here you are: http://tpaste.us/ARpY
2016-01-27 10:41:11 <ScrumpyJack> you mean if you don't have git installed?
2016-01-27 10:42:20 <skarnet> ScrumpyJack: no, I copied the patch to my Alpine dev machine, git pulled, then tried to "patch -p1 < patchfile", and it got rejected.
2016-01-27 10:42:39 <skarnet> probably wrong options. I copied the entire file anyway, that's not a problem.
2016-01-27 11:04:37 <ScrumpyJack> :)
2016-01-27 11:25:19 <skarnet> So, I successfully built abuild from scratch. And now, same questions with apk-tools!
2016-01-27 11:25:42 <skarnet> What are the dependencies? It looks like pkg-config is a hard dependency, but I would like to work around it if possible.
2016-01-27 11:26:38 <skarnet> openssl and zlib are dependencies - that's ok. But can I do without lua ? and what make variables should I set to avoid pkg-config ?
2016-01-27 11:27:42 <skarnet> @ncopa @fabled
2016-01-27 11:28:07 <awilfox> skarnet: https://code.foxkit.us/adelie/packages/blob/master/sys-apps/apk-tools/apk-tools-2.6.4.ebuild
2016-01-27 11:30:12 <skarnet> awilfox: thanks, but that only tells me about the explicit dependencies. ebuild could (and very likely does) hide a truckload of hidden dependencies that even the ebuild authors have forgotten about.
2016-01-27 11:30:29 <awilfox> skarnet: well I am the ebuild author, and this is a sys-* pacakge so it can't.
2016-01-27 11:30:35 <awilfox> skarnet: those are the only deps it has
2016-01-27 11:30:41 <awilfox> skarnet: and also, it tells you how to disable Lua
2016-01-27 11:30:47 <awilfox> lines 38-39
2016-01-27 11:31:29 <skarnet> ok, then how do you get around pkg-config ?
2016-01-27 11:31:42 <skarnet> and what does emake do around make ?
2016-01-27 11:32:04 <awilfox> 'emake' just ensures that MAKEFLAGS are honoured and adds -jX if available
2016-01-27 11:33:39 <awilfox> god
2016-01-27 11:33:49 <awilfox> I just noticed that apk-tools makefile hardcodes /usr/lib/libfetch.a
2016-01-27 11:34:05 <awilfox> I can't WAIT to try and get that to work in a cross-compilation environment
2016-01-27 11:34:44 <skarnet> ok, now you understand what I'm talking about
2016-01-27 11:35:06 <skarnet> I'm working on the bare Makefiles, not on wrapper tools, because wrapper tools hide that stuff from me
2016-01-27 11:35:08 <awilfox> I think ebuild may have a thin script for when pkg-config is not available
2016-01-27 11:35:16 <awilfox> not sure
2016-01-27 11:35:29 <awilfox> I certainly don't have pkg-config (or pkgconf, which is much better) in this chroot and apk-tools built
2016-01-27 11:35:38 <fabled> skarnet, you can build without lua; but libfetch is needed currently.
2016-01-27 11:35:49 <fabled> we had talk of embedded libfetch to apk
2016-01-27 11:36:00 <skarnet> then ebuild is definitely doing something
2016-01-27 11:36:20 <awilfox> embedded libs are bad :(
2016-01-27 11:37:17 <skarnet> fabled: I see that libfetch has an Alpine package. Which... has a bootstrap problem.
2016-01-27 11:37:50 <fabled> yeah, generally we don't want to embed things
2016-01-27 11:37:58 <skarnet> oh, I can't wait to try compiling NetBSD stuff on Linux...
2016-01-27 11:38:01 <awilfox> skarnet: well, not sure if this helps https://code.foxkit.us/adelie/packages/blob/master/dev-libs/libfetch/libfetch-2.33.ebuild
2016-01-27 11:38:08 <fabled> but libfetch is not widely used, and there's various different versions of it
2016-01-27 11:38:44 <awilfox> the patch just makes it so that it can be cross-compiled https://code.foxkit.us/adelie/packages/blob/master/dev-libs/libfetch/files/libfetch-2.33-remove-hardcoded-compiler.patch
2016-01-27 11:39:05 <awilfox> actually since I made myself my own upstream of it, I could just roll 2.34 with that patch applied, I guess.
2016-01-27 11:39:29 <awilfox> it hasn't seen any activity in NetBSD in many moons
2016-01-27 11:43:06 <skarnet> guyz
2016-01-27 11:43:19 <skarnet> in the APKBUILD for libfetch
2016-01-27 11:43:29 <skarnet> source="Makefile
2016-01-27 11:43:29 <skarnet>         https://sources.archlinux.org/other/libfetch/libfetch-$pkgver.tar.gz
2016-01-27 11:43:35 <awilfox> yeah
2016-01-27 11:43:42 <awilfox> archlinux.org hasn't used it since 2011 and dropped the mirror
2016-01-27 11:43:45 <awilfox> I had to FIND the upstream
2016-01-27 11:43:49 <awilfox> which is an obscure NetBSD mirror
2016-01-27 11:43:54 <awilfox> hence I put it on distfiles.foxkit.us
2016-01-27 11:44:10 <awilfox> as shown in the libfetch ebuild I linked.
2016-01-27 11:44:36 <skarnet> so if I want to build apk, I need to depend either on an obscure NetBSD mirror, or on arch linux, or on Adélie ?
2016-01-27 11:44:42 <awilfox> yes.
2016-01-27 11:45:19 <skarnet> Well, as loathe as I am to say it, that means that Alpine *has* to maintain its own version of it.
2016-01-27 11:45:39 <skarnet> If a distro isn't self-sufficient, it *will* break at some point
2016-01-27 11:46:37 <skarnet> doesn't mean it has to be difficult, if awilfox wants to take over maintainership
2016-01-27 11:46:43 <skarnet> (since you're doing it for Adélie anyway)
2016-01-27 11:48:32 <skarnet> awilfox: actually, your patch isn't needed if you provide CC, LD and stuff as make variables instead of environment variables
2016-01-27 11:48:33 <awilfox> actually the overall goal is to port apk to something with a more active upstream than libfetch
2016-01-27 11:49:51 <skarnet> what is libfetch anyway? a HTTP/FTP client?
2016-01-27 11:49:56 <skarnet> may as well fork a busybox wget
2016-01-27 11:50:17 <awilfox> skarnet: it's a library.  it'd be better to use libcurl or such.
2016-01-27 11:50:56 <skarnet> yeah, except that bb is easy to build whereas libcurl depends on the kitchen sink
2016-01-27 11:51:13 <awilfox> uh?  libcurl only requires openssl and zlib
2016-01-27 11:51:19 <awilfox> which apk already requires anyway
2016-01-27 11:52:21 <skarnet> libfetch doesn't, right?
2016-01-27 11:52:24 <skarnet> that means apk handles the SSL layer itself
2016-01-27 11:52:29 <awilfox> libfetch does require openssl
2016-01-27 11:52:34 <skarnet> oh
2016-01-27 11:56:45 <skarnet> awilfox: distfiles.foxkit.us: network unreachable
2016-01-27 11:57:06 <awilfox> loads fine here; you need IPv6
2016-01-27 11:57:29 <awilfox> I don't have enough money to buy more IPv4 addresses
2016-01-27 11:57:40 <awilfox> and my provider is upset enough at the fact I already fully used my /29
2016-01-27 12:00:04 <ncopa> i agree that bundling libraries is not a good thing, generally
2016-01-27 12:00:57 <ncopa> in libfetch case we can either fork+bundle it or fork it as separate project
2016-01-27 12:01:15 <ncopa> i dont think its worth a separate project
2016-01-27 12:01:30 <ncopa> we have to fork it either way
2016-01-27 12:01:57 <ncopa> so bundle it makes sense imo
2016-01-27 12:02:29 <ncopa> skarnet: you can make LUAAPK=no iirc
2016-01-27 12:03:13 <skarnet> awilfox: ah, makes sense, my VM doesn't have IPv6. I'll use a manual proxy.
2016-01-27 12:03:16 <ncopa> if you want avoid pkg-config due to freedesktop's implementation requries bootstrapping with glib, then we have a different implementation: pkgconf
2016-01-27 12:03:26 <ncopa> which is more sane
2016-01-27 12:04:13 <skarnet> ncopa: I don't doubt it is (it's not exactly hard to be saner than anything coming from fdo) but what bothers me is that it's ONE MORE dependency
2016-01-27 12:04:36 <ncopa> understand
2016-01-27 12:05:14 <skarnet> the bootstrap chain is long enough as it is
2016-01-27 12:06:33 <skarnet> and pkgconf is an apk package, and the source is: http://rabbit.dereferenced.org/~nenolod/distfiles/pkgconf-$pkgver.tar.bz2
2016-01-27 12:06:51 <skarnet> allow me to be suspicious of the perenniality of that URL
2016-01-27 12:07:36 <awilfox> skarnet: you might want to use https://github.com/pkgconf/pkgconf/releases
2016-01-27 12:07:56 <ncopa> author is in this channel, kaniini
2016-01-27 12:08:24 <ncopa> skarnet: apk-tools makefile looks simple enough to patch
2016-01-27 12:08:30 <ncopa> i think we want disable lua by default
2016-01-27 12:08:35 <skarnet> nenolod is kaniini? nice :)
2016-01-27 12:09:07 <skarnet> ncopa: the apk-tools makefile is EVIL
2016-01-27 12:09:12 <awilfox> ^
2016-01-27 12:09:20 <awilfox> as a sort-of 'downstream', yes it is
2016-01-27 12:09:24 <skarnet> it looks so nice and clean and simple
2016-01-27 12:09:25 <ncopa> :)
2016-01-27 12:09:36 <skarnet> but all the nasty stuff is hidden in included makefiles
2016-01-27 12:09:44 <skarnet> and the nasty stuff IS nasty
2016-01-27 12:09:56 <ncopa> like what?
2016-01-27 12:11:12 <skarnet> like the fact that Make.rules is unreadable?
2016-01-27 12:11:51 <ncopa> i think its copied from linux kernel
2016-01-27 12:12:07 <skarnet> yeah, that's not exactly a reference for readability ;)
2016-01-27 12:16:42 <ncopa> this may work: http://tpaste.us/A97l
2016-01-27 12:16:58 <ncopa> i suspect that fabled wanted avoid forking pk-config for each file it compiles
2016-01-27 12:20:48 <skarnet> make still forks a cargo freighter of pkg-config after applying the patch
2016-01-27 12:20:59 <skarnet> nvm, I'll build pkgconf...
2016-01-27 12:21:58 <skarnet> kaninii's github picture is incredibly cute, makes me want to pet him.
2016-01-27 12:22:08 <skarnet> kaniini, even.
2016-01-27 12:25:13 <awilfox> j'ai une peu diff, mais il ne fonctionne pas tout de même
2016-01-27 12:25:17 <awilfox> err
2016-01-27 12:25:20 <awilfox> wrong chan sorry
2016-01-27 12:26:56 <skarnet> awilfox: I didn't know you spoke French :)
2016-01-27 12:29:23 <awilfox> skarnet: indeed!  not an expert, but I can function with it
2016-01-27 12:51:04 <skarnet> are you fking kidding me
2016-01-27 12:51:28 <skarnet> the pkgconf-0.9.12 tarball doesn't come with a configure script
2016-01-27 12:51:38 <skarnet> if has a configure.ac and an autogen.sh
2016-01-27 12:51:54 <skarnet> but there's NO WAY I'm installing autotools to get this done
2016-01-27 12:52:12 <skarnet> kaniini: could I please have a configure script in the numbered releases ?
2016-01-27 12:53:32 <ncopa> isnt that what http://rabbit.dereferenced.org/~nenolod/distfiles/ tarballs are for?
2016-01-27 12:54:42 <skarnet> sigh
2016-01-27 12:54:50 <skarnet> ok then, I'll use those
2016-01-27 12:55:18 <skarnet> but the tarballs on github with the *same names* don't have the same contents
2016-01-27 12:55:21 <skarnet> you might want to change that...
2016-01-27 13:46:17 <skarnet> success: apk compiled!
2016-01-27 13:46:43 <skarnet> I had to patch src/Makefile though: /usr/lib/libfetch.a is hardcoded (as awilfox said)
2016-01-27 13:46:43 <skarnet> changing it to -lfetch works
2016-01-27 13:59:02 <bpiotrowski> ncopa: what about libcurl?
2016-01-27 14:09:43 <ncopa> libcurl has a bunch of deps
2016-01-27 15:00:00 <faffolter> please pull git://git.alpinelinux.org/user/fab/aports
2016-01-27 15:00:36 <faffolter> ncopa: was the workflow change for doing packaging stuff?
2016-01-27 15:03:24 <faffolter> s/change/changed
2016-01-27 15:15:46 <kaniini> skarnet: github sux
2016-01-27 15:16:17 <kaniini> skarnet: there is no way to opt out of their system
2016-01-27 15:31:19 <skarnet> kaniini: yeah, github suxx. But I forgive them, because they turn you into a cute rabbit.
2016-01-27 15:35:05 <kaniini> this problem wouldnt exist if you could turn off the 'releases' feature
2016-01-27 15:38:48 <skarnet> you can't ? I have nothing similar in my github projects. It is because I'm only using github as a mirror?
2016-01-27 15:39:40 <kaniini> you can't
2016-01-27 15:43:38 <elfranne> any dev present ?
2016-01-27 15:54:00 <clandmeter> kaniini: cant you upload your releases to github? that would make things clearer.
2016-01-27 15:54:20 <kaniini> i do
2016-01-27 15:54:27 <kaniini> then it does this.
2016-01-27 15:55:32 <clandmeter> hmm ok
2016-01-27 15:57:18 <clandmeter> so what actually happens here? you upload your release and github replaces it with a git tag archive?
2016-01-27 15:58:09 <kaniini> https://www.dropbox.com/s/41f2h3k8gjgyg6s/Screenshot%202016-01-27%2010.58.07.png?dl=0
2016-01-27 15:58:14 <kaniini> which is the source code?
2016-01-27 15:59:08 <kaniini> people go and click the "Source code (tar.gz)" and get fucked over
2016-01-27 16:02:05 <clandmeter> ah ok
2016-01-27 16:04:02 <kaniini> skarnet: as for the perenniality of that URL, it has existed since 2009
2016-01-27 16:32:53 <skarnet> kaniini: ok, ok. I didn't know pkgconf was yours. :P
2016-01-27 16:52:30 <kaniini> pkgconf 0.9.12 may be insufficient for your needs.  https://github.com/pkgconf/pkgconf/commit/7e6fa325eb668c3462981a16fb4c36270832e00f
2016-01-27 16:52:54 <kaniini> we found out about some undocumented pkg-config features.
2016-01-27 16:53:04 <kaniini> that apparently are used by some build systems.
2016-01-27 16:53:18 <kaniini> it's the gift that keeps on giving.
2016-01-27 16:54:43 <skarnet> kaniini: thanks, but I only intended to use it to build apk. When abuild+apk is bootstrapped, I'm washing my hands. :)
2016-01-27 16:54:53 <kaniini> hah
2016-01-27 16:55:07 <kaniini> probably a good thing
2016-01-27 16:55:45 <kaniini> pkgconf was made out of necessity (terrible maintainance strategy of pkg-config), not because i really love pkg-config
2016-01-27 16:55:46 <kaniini> ;p
2016-01-27 16:57:10 <skarnet> it's a testament to your sanity
2016-01-27 16:58:12 <kaniini> (well i do find pkg-config to be an upgrade over the libfoo-config scripts that everyone wishes to ship.  but in reality, the toolchain should just be aware of .pc files and use them internally, like with OS X's -framework flag, which accomplishes the same thing basically)
2016-01-27 17:00:02 <skarnet> that, or people could accept that file placement is a policy question and every user, including distros, will *know* where the software is, so pkg-config is ultimately useless.
2016-01-27 17:01:17 <kaniini> sure, but specifying "-framework Gtk+3.0" is much cleaner than specifying a wide series of CFLAGS/LDFLAGS/LIBS etc.
2016-01-27 18:15:16 <mitchty> also objc #import vs #include ... #ifndef stuff guards
2016-01-27 18:16:32 <kaniini> that one is solved with #pragma once
2016-01-27 18:19:50 <ScrumpyJack> anyone tried alpine-rpi on a RPi A+?
2016-01-27 18:23:46 <mitchty> kaniini: only didn't work with sun compilers https://en.wikipedia.org/wiki/Pragma_once
2016-01-27 18:24:00 <mitchty> though i don't miss solaris that much, except dtrace
2016-01-28 01:23:29 <skazz> Can someone point me in the right direction for compiling a Kernel Module for Alpine Raspberry Pi 2?
2016-01-28 01:23:53 <jirutka> 👍
2016-01-28 01:25:35 <skazz> That is a thumb not a point =D
2016-01-28 01:25:48 <skazz> Also searching by directory does not seem to work : https://pkgs.alpinelinux.org/contents?filename=%25&path=%2Flib%2F%25&pkgname=%25&repo=all&arch=all
2016-01-28 01:32:33 <skazz> Guess I need to clone https://github.com/raspberrypi/linux but github is down, agh, of all the times lol
2016-01-28 07:02:09 <ncopa> morning
2016-01-28 07:02:19 <ncopa> firefox-44.0 segfaults often
2016-01-28 07:15:17 <ncopa> i think we need to try upstream the firefox patches we have
2016-01-28 08:50:49 <ScrumpyJack> morning
2016-01-28 08:55:57 <ncopa> morning
2016-01-28 08:56:08 <ncopa> ok i have a specific webpage that make firefox crash
2016-01-28 08:56:12 <ncopa> http://streambuilder.pro/docs/how-tos/check-iqsv/
2016-01-28 09:07:09 <ScrumpyJack> that page works for me on 42.0 so i'll wait before updating :)
2016-01-28 09:16:45 <ncopa> what happened with samba package? http://rsync.alpinelinux.org/alpine/latest-stable/main/x86/
2016-01-28 09:16:48 <ncopa> its gone
2016-01-28 09:23:57 <ncopa> ok firefox sefault still happens even with the bundled cairo
2016-01-28 09:31:43 <ScrumpyJack> I'm building at (atd) for #5030
2016-01-28 09:31:57 <ScrumpyJack> where shall we put the spool?
2016-01-28 09:32:23 <ScrumpyJack> some folk put it in /var/spool/cron/ and some in /var/spool/at
2016-01-28 09:32:57 <ncopa> i started on it soem time ago: http://tpaste.us/GZJl
2016-01-28 09:33:33 <ncopa> $ tpaste < parallel-make.patch
2016-01-28 09:33:33 <ncopa> http://tpaste.us/3lO6
2016-01-28 09:33:42 <ScrumpyJack> lol, i didnt' check aports, at the issue was unassigned :)
2016-01-28 09:33:47 <ncopa> $ tpaste < is_leap_year.patch
2016-01-28 09:33:47 <ncopa> http://tpaste.us/GdZL
2016-01-28 09:33:53 <ncopa> i havent checked it in anywhere
2016-01-28 09:34:07 <ncopa> $ tpaste < configure-pam.patch
2016-01-28 09:34:07 <ncopa> http://tpaste.us/24OX
2016-01-28 09:34:12 <ncopa> feel free to continue from there
2016-01-28 09:35:04 <ScrumpyJack> ah i see now
2016-01-28 09:37:47 <ScrumpyJack> I'll try, but don't hold your breath :)
2016-01-28 11:14:42 <clandmeter> ncopa: have we ever concidered changing our apk compression alg?
2016-01-28 11:16:09 <ncopa> questino have come up once in a while
2016-01-28 11:16:36 <ncopa> i think gz algo is a pretty good middle ground
2016-01-28 11:16:51 <clandmeter> LZ4 HC seems nice
2016-01-28 11:16:52 <ncopa> speed vs compression/decompression ratio
2016-01-28 11:17:26 <clandmeter> about 7x faster at decompression
2016-01-28 11:17:40 <ncopa> compression time is not totally irrelevant
2016-01-28 11:17:50 <ncopa> i suppose compression time is bigger
2016-01-28 11:18:07 <clandmeter> https://github.com/Cyan4973/lz4
2016-01-28 11:19:18 <skazz> Maybe I should be asking in here, compling the rpi2 kernel headers for a module, getting errors, can someone help me please?
2016-01-28 13:32:38 <ncopa> stateless: your fortify implementation caught a firefox 44 issue
2016-01-28 13:32:40 <ncopa> good job!
2016-01-28 13:32:45 <ncopa> https://bugzilla.mozilla.org/show_bug.cgi?id=1243724
2016-01-28 13:51:58 <ScrumpyJack> could someone remind me where I can find the build logs for arm packages?
2016-01-28 13:57:12 <ncopa> http://build.alpinelinux.org/buildlogs/build-edge-armhf/
2016-01-28 14:01:59 <ScrumpyJack> thanks. neat bit of bug crushing there btw
2016-01-28 14:03:17 <ScrumpyJack> does no log means success?
2016-01-28 14:05:20 <ScrumpyJack> slock builds on x86_64 but not on arm, due to a checksum error http://build.alpinelinux.org/buildlogs/build-edge-armhf/testing/slock/slock-1.2-r1.log
2016-01-28 14:07:30 <ScrumpyJack> would triggering a rebuild slock help?
2016-01-28 14:33:34 <mosez_> plans somebody to add a gecode package to alpine? 8)
2016-01-28 14:33:40 <mosez_> s/gecode/gecode3/
2016-01-28 14:50:52 <stateless> ncopa, very nice!
2016-01-28 15:40:09 <ScrumpyJack> oops, just pulled firefox 444 r1
2016-01-28 15:41:36 <ScrumpyJack> i mean 44.0 r0
2016-01-28 16:04:40 <ncopa> r1 has the fix
2016-01-29 08:55:40 <ScrumpyJack> morning
2016-01-29 08:55:52 <ScrumpyJack> Firefox 44.0 crashes here -> http://www.raspberrypi-spy.co.uk/2014/07/raspberry-pi-model-b-3-5mm-audiovideo-jack/
2016-01-29 08:56:01 <ScrumpyJack> is it just me?
2016-01-29 08:58:36 <ScrumpyJack> ignore me. I thought I had r1 of firefox or something
2016-01-29 09:07:54 <ncopa> firefox-44.0-r0 crashes yes
2016-01-29 09:08:01 <ncopa> -r1 should not crash
2016-01-29 09:13:58 <ScrumpyJack> ooo, r1 is on it's awy to me now :)
2016-01-30 01:21:45 <mitchty> curious if there is anything I can do for the ghc patch I submitted for testing a couple weeks ago, haven't heard anything back so wondering if I didn't follow the guidelines right or what might be amiss
2016-01-30 01:23:53 <mitchty> patch i submitted was http://sprunge.us/CPAa
2016-01-30 01:25:24 <mitchty> I know its a doozy but it probably won't get any simpler sadly
2016-01-30 01:30:15 <mitchty> as a note, the llvm patches are the same as in main/llvm mostly, but need to build the right llvm more for prep for 8.0.1 and arm support
2016-01-30 01:33:22 <mitchty> just curious as i'm starting to work on 8.0.1 and adding building from git this weekend
2016-01-30 01:45:27 <_mjones> mitchty: mailing list or github pr? I can't do anything, I just want to see if it's being tracked.
2016-01-30 01:45:52 <mitchty> _mjones ml post was this http://lists.alpinelinux.org/alpine-devel/5135.html
2016-01-30 01:46:04 <_mjones> 
ACTION just found another part of boost that needs to be fixed. C++, sigh.

2016-01-30 01:46:41 <_mjones> mitchty: holding off on armhf for now I assume?
2016-01-30 01:46:50 <_mjones> I know it presented bootstrapping problems.
2016-01-30 01:47:29 <mitchty> yeah
2016-01-30 01:47:38 <mitchty> i managed to get a mostly working binary
2016-01-30 01:47:41 <mitchty> ish
2016-01-30 01:47:58 <mitchty> but there were some messed up symbols in a shared library so it just segfaulted
2016-01-30 01:48:17 <mitchty> i kind a threw up my hands then and worked on llvm stuff
2016-01-30 01:50:29 <_mjones> mitchty: I'm going to make a guess that because it was a big port of a boostrapping language that no-one looked at it right away and it has fallen through the cracks.
2016-01-30 01:50:46 <_mjones> ncopa in particular has been very busy recently I think.
2016-01-30 01:50:56 <mitchty> _mjones: yeah no worries just thought i'd ping to see whats up
2016-01-30 01:51:13 <mitchty> its a huge patch and an odd beast in general
2016-01-30 01:52:25 <mitchty> but once its bootstrapped we can just bootstrap off alpine linux instead of an ubuntu bootstrap or whatever
2016-01-30 01:52:36 <mitchty> building natively is a lot less finicky
2016-01-30 01:53:04 <_mjones> mitchty: you want to ping in central european time if you can.
2016-01-30 01:53:19 <_mjones> mitchty: I don'
2016-01-30 01:53:31 <_mjones> t know if there's policy on boostrapping from bins either.
2016-01-30 01:54:02 <_mjones> What little I've worked with languages was nodejs, iojs and jdk where there's no boostrapping issue.
2016-01-30 01:55:03 <mitchty> would've been nice if there wasn't a bootstrapping issue, but have to live with what is there
2016-01-30 01:55:28 <mitchty> what was in the wiki was way off in what needed to happen and was more confusing
2016-01-30 01:56:01 <mitchty> porting ghc made me appreciate how easy c is to cross compile
2016-01-30 01:56:31 <mitchty> also found ellcc that way which is my new favorite cross compiler
2016-01-30 01:57:39 <mitchty> also need to port upx http://upx.sourceforge.net
2016-01-30 02:00:21 <_mjones> Most for PE binaries, isn't it? Not sure of the use case on posix.
2016-01-30 02:01:15 <_mjones> Oh, I see it will compile shell scripts. That's somewhat interesting.
2016-01-30 02:01:39 <_mjones> I'm sure it just wraps each line in system(), but still.
2016-01-30 02:02:43 <mitchty> just basically lz compresses the binary then at runtime reexecs that after it decompresses
2016-01-30 02:03:25 <_mjones> I know.
2016-01-30 02:05:18 <_mjones> In practice I've only ever seen it used for binary obfuscation, and primarily with PE binaries on win32, although ISTR some packed malware-related binaries for Linux some time back.
2016-01-30 02:05:41 <mitchty> i use it to shrink giant binaries
2016-01-30 02:05:59 <mitchty> some elf binaries are full of 0's all over
2016-01-30 02:06:36 <mitchty> i just use it to compress static binaries for work, makes it easier to rsync
2016-01-30 02:08:45 <mitchty> can get tmux to about 250k in size, and when sysadmins won't install anything for whatever reason means your home directory isn't huge
2016-01-30 02:12:07 <mitchty> i'll go pink on the ml tomorrow, or probably sunday i suppose
2016-01-30 02:12:12 <mitchty> ping
2016-01-30 02:23:26 <_mjones> The biggest binaries I see are ++ and unstripped.
2016-01-30 02:23:54 <_mjones> But if you get tmux to half size...
2016-01-30 02:24:58 <mitchty> heh, you need to see a fully static pandoc
2016-01-30 02:25:03 <mitchty> its 50 megs unstripped
2016-01-30 02:26:27 <mitchty> err stripped
2016-01-30 02:26:58 <mitchty> https://gist.githubusercontent.com/mitchty/c8a4c9200775f97830c2/raw/565b137f37c76b668254d9e084a09d985685031b/gistfile1.txt
2016-01-30 02:27:15 <mitchty> to be honest a fully static pandoc is about 99% of why i ported ghc
2016-01-30 02:28:43 <_mjones> I know mitchty.
2016-01-30 02:29:08 <_mjones> I almost had an aneurysm when I installed pandoc and it took 1536mb, of which 900mb was ghc.
2016-01-30 02:29:45 <_mjones> I've had to suspend recommending RST for the time being while I ponder that.
2016-01-30 02:30:41 <_mjones> Damnit now you're making me want to try upx. But I don't know if I want to mess with porting this old UCL library.
2016-01-30 02:30:53 <_mjones> s/try/try porting/
2016-01-30 02:32:46 <mitchty> if it helps here is my current docker file for building things https://github.com/mitchty/alpine-static-tmux
2016-01-30 02:33:01 <mitchty> but yeah ucl is some odd beast
2016-01-30 02:36:43 <_mjones> I was building some new compression things earlier though.
2016-01-30 03:51:03 <mitchty> what kind of compression things
2016-01-30 04:08:48 <_mjones> Goggle's zopfli and brotli.
2016-01-30 04:25:39 <mitchty> ah
2016-01-30 04:27:38 <_mjones> zopfli gets a few percent tighter compression on gzip at the cost of much longer compression time. But I'm thinking this would be a very worthwhile tradeoff for .apk packages on media and in mirrors.
2016-01-30 04:31:29 <mitchty> how does it compare to using xz
2016-01-30 04:52:36 <_mjones> Not as good as. But using .xz for packages would presumably require a new package format. I'm interested in the response to that idea too.
2016-01-30 21:14:04 <jarth> hi
2016-01-30 21:16:01 <jarth> i'm having trouble compiling packages on alpine, formost i wanted to get the linux graphics driver gui to compile
2016-01-30 21:16:24 <jarth> Such as here https://01.org/linuxgraphics/downloads/intel-graphics-installer-linux-1.2.1
2016-01-30 21:19:01 <Roland-> oh hi
2016-01-30 21:41:38 <jarth> Hi
2016-01-30 21:41:42 <jarth> well, it got fixed in the meantime
2016-01-30 21:41:56 <jarth> /usr/local/sbin is now populated with brctl
2016-01-31 00:21:07 <jarth> almost there but struggling to have libiconv compiled
2016-01-31 00:21:44 <jarth> make
2016-01-31 00:21:44 <jarth> builddir="`pwd`"; cd libcharset && make all && make install-lib libdir="$builddir/lib" includedir="$builddir/lib"
2016-01-31 00:21:44 <jarth> make[1]: Leaving directory '/opt/libs/libiconv-1.11/libcharset'
2016-01-31 00:21:44 <jarth> make[1]: *** No rule to make target 'all'.  Stop.
2016-01-31 00:21:44 <jarth> make[1]: Entering directory '/opt/libs/libiconv-1.11/libcharset'
2016-01-31 00:21:44 <jarth> Makefile:41: recipe for target 'lib/localcharset.h' failed
2016-01-31 00:21:44 <jarth> make: *** [lib/localcharset.h] Error 2
2016-01-31 10:21:26 <ammunta> I feel like the compiling gag gets old after a bit tbh
2016-01-31 10:42:34 <skarnet> stop criticizing and share the joy \o/