2015-12-01 08:32:26 morning. happy December! 2015-12-01 08:38:38 anyone tried to run docker inside an lxc container? 2015-12-01 08:56:50 I got error in pulling images 2015-12-01 08:56:58 ApplyLayer exit status 1 stdout: stderr: operation not permitted 2015-12-01 08:57:27 fcolista: is that even supported? I thought Docker liked to modify cgroups itself directly 2015-12-01 08:57:59 afaik, it is. 2015-12-01 08:58:04 with libcontainer 2015-12-01 08:58:27 well, then I don't know enough to provide any meaningful input 2015-12-01 08:59:31 i don't understand how this is supposed to work 2015-12-01 08:59:39 if you need a shared aufs dir with the host 2015-12-01 08:59:48 or you need to play with cgroups permissions 2015-12-01 09:00:21 you probably need to mount your own cgroup hierarchy inside the lxc. But then again, I don't know what I'm talking about 2015-12-01 09:15:20 skarnet, is definetly a cgroup issue fyi. 2015-12-01 09:28:42 yay for hunches 2015-12-01 09:29:22 that did the trick: 2015-12-01 09:29:29 lxc.cgroup.devices.allow = a 2015-12-01 09:29:36 # lxc.cgroup.devices.deny = a 2015-12-01 09:29:54 fcolista: that's pretty premissive :) 2015-12-01 09:29:59 permissive even 2015-12-01 09:30:08 ScrumpyJack, yes. It's a test env 2015-12-01 09:30:26 I need to figure the exact cgroup permissions 2015-12-01 09:30:38 The error i got was pretty vague 2015-12-01 09:30:49 so i had to understand first where the issue resides 2015-12-01 09:31:36 default lxc.cgroup config is "deny all" then "permit the strict necessary" 2015-12-01 09:31:47 that's the approach i like :) 2015-12-01 09:59:50 runtime: goroutine stack exceeds 1000000000-byte limit 2015-12-01 09:59:50 fatal error: stack overflow 2015-12-01 10:00:01 i'm hitting some limits 2015-12-01 10:07:10 fcolista: do you use the docker hub? 2015-12-01 10:07:25 yes 2015-12-01 10:07:41 the image is pulled from a docker hub, if this is what you mean 2015-12-01 10:07:42 does the alpine docker container work? 2015-12-01 10:07:47 ScrumpyJack, no 2015-12-01 10:07:55 oops 2015-12-01 10:08:17 what containers do you "pull"? 2015-12-01 10:09:05 i've tried different containers 2015-12-01 10:09:26 keep in mind i'm using docker inside an lxc container 2015-12-01 10:09:31 this is the issue i presume 2015-12-01 10:09:38 now i'm trying on a host 2015-12-01 10:09:52 alpine host? 2015-12-01 10:26:11 hey :) 2015-12-01 11:09:43 fcolista: Error pulling image (latest) from docker.io/library/ubuntu, ApplyLayer exit status 1 stdout: stderr: chmod /bin/mount: permission denied 2015-12-01 11:09:50 did you get this error at all? 2015-12-01 11:10:00 NO 2015-12-01 11:10:02 sorry 2015-12-01 11:10:03 No 2015-12-01 11:10:16 "fcolista> ApplyLayer exit status 1 stdout: stderr: operation not permitted" 2015-12-01 11:10:25 what images arey ou pulled successfully? 2015-12-01 11:10:36 s/arey/have 2015-12-01 11:10:41 no-one inside the container 2015-12-01 11:11:02 and with docker in a host? 2015-12-01 11:11:15 works. 2015-12-01 11:11:22 which containers? 2015-12-01 11:11:34 gondoi/unifi 2015-12-01 11:24:01 what version of Alpine? 2015-12-01 11:24:09 3.3rc1 2015-12-01 11:25:45 grsec: denied chmod +s of /var/lib/docker/overlay/... 2015-12-01 11:26:05 im on edge 2015-12-01 11:26:08 that's about the error that you got 2015-12-01 11:26:10 stdout: stderr: chmod /bin/mount: permission denied 2015-12-01 11:26:16 it's not the same i had 2015-12-01 11:26:22 stderr: operation not permitted" 2015-12-01 11:26:29 my issue is related to cgroup 2015-12-01 11:26:36 yes, grsec looks like it's bloking the chmod 2015-12-01 11:26:43 (for me) 2015-12-01 11:49:38 fcolista i'm new to docker. I have the alpine container. I'm a bit underwhelmed. 2015-12-01 11:50:19 i guess i'm used to lxc and getting an interactive shell "inside" the container 2015-12-01 11:51:14 how do i edit a config file in the docker container? 2015-12-01 11:51:24 docker run alpine vi /etc/apk/repositories ? 2015-12-01 12:19:12 oops, pacemaker package seems to have broken deps 2015-12-01 12:19:26 ERROR: unsatisfiable constraints:so:libgnutls.so.28 (missing):required by: pacemaker-1.1.12-r0[so:libgnutls.so.28] 2015-12-01 12:19:36 (edge) 2015-12-01 12:39:56 ScrumpyJack: no, if you use docker run vi, your change will not persist. You ideally want to change that at build time 2015-12-01 12:45:12 i see, so a docker container is pretty static? 2015-12-01 12:45:32 and it's a question of roll your own? 2015-12-01 13:07:11 the should be treated as static and I always roll my own 2015-12-01 13:39:22 would be nice to have docker images which works on alpine :-/ 2015-12-01 13:41:47 fcolista: what is broken? 2015-12-01 13:42:28 barthalion, just fighting with docker images for unifi controller that won't work. Mainly because of java. 2015-12-01 13:45:23 does not sound like a docker problem 2015-12-01 14:40:22 barthalion: what are your thoughts a docker abuild for arm? 2015-12-01 15:21:46 we don't have it? 2015-12-01 15:26:59 vkris: re curl and nghttp2 2015-12-01 15:27:16 i am sceptic, because curl is a part of bootstrap 2015-12-01 15:27:30 and nghttp2 depends on boost 2015-12-01 15:27:45 so with this change you will have to crosscompile boost to bootstrap things 2015-12-01 15:27:55 i am not too happy about that 2015-12-01 15:27:56 also 2015-12-01 15:28:04 nghttp2 is in testing 2015-12-01 15:28:25 so i cannot apply your patch as is 2015-12-01 15:28:43 and the nghttp2 package seem to ship some kind of http server 2015-12-01 15:28:47 but there are no init.d script 2015-12-01 15:28:54 so it feels "unfinished" 2015-12-01 15:29:05 I dont want ship unfinished stuff like that 2015-12-01 15:30:16 we can probably build nghttp2 as library only and disable boost etc 2015-12-01 15:30:23 the library is not linked to boost apparently 2015-12-01 15:30:33 however 2015-12-01 15:31:00 i will not spend time on doing that 2015-12-01 15:31:12 and i will not apply patches unless someone does things properly 2015-12-01 16:21:27 ncopa never forgives 2015-12-01 16:21:30 ACTION hides 2015-12-01 16:22:01 :) 2015-12-01 16:22:06 ok i gotta go 2015-12-01 16:22:07 see u 2015-12-01 17:04:46 barthalion: nah, it's not built for arm. the APKBUILD file is x86 x86_64 only 2015-12-01 17:05:19 I'm attempting to build it (docker) on a scaleway C1 at the moment (it's taking ages) 2015-12-01 17:16:27 barthalion: docker builds fine on Scaleway's armv7l C1. Would you be kind enough to update the APKBUILD file with arch="all" 2015-12-01 17:25:43 would be nice to have docker images which works on alpine :-/ 2015-12-01 17:25:50 you mean with Alpine as a host, not a guest? 2015-12-01 17:28:05 both :) 2015-12-01 17:28:31 it seems to work ok for me so far, but it's my first day with docker :) 2015-12-01 17:28:40 well Alpine guests have been a thing for a long time, and you probably already know :) 2015-12-01 17:29:32 and I only have alpine hosts, so that answers both questions :) 2015-12-01 17:29:53 now fcolista can sleep at night 2015-12-01 17:30:06 skarnet, i've alpine as host. Trying to make a docker image as guest. Looks that the issue is in the docker file 2015-12-01 17:30:51 https://github.com/just-containers/base-alpine 2015-12-01 17:30:53 the one i'm using are java-related 2015-12-01 17:31:00 *ones 2015-12-01 17:31:08 i need docker for unifi controller 2015-12-01 17:31:16 (uni ap pro) 2015-12-01 17:31:23 host has alpine 2015-12-01 17:31:29 with lxc container 2015-12-01 17:31:41 and in the lxc container i wanted to try docker 2015-12-01 17:32:10 but because of cgroup, first i wanted to start with easy thing, then going through the more complex 2015-12-01 17:32:14 so i moved docker on the host 2015-12-01 17:32:41 the images i'm using with java failed because of some limits this VPS has 2015-12-01 17:32:49 you don't need to run the docker daemon in a lxc container, it's pretty safe 2015-12-01 17:34:14 ncopa: ok, splitting the patch into more unit commit 2015-12-01 17:40:59 http://tpaste.us/36MB <- for enable unix-sockets 2015-12-01 17:42:19 since curl pkg is important (bootstrap), I would leave nghttp2 for now 2015-12-01 17:42:49 but got through its doc, may add the init-script and move to main/ 2015-12-01 17:43:00 go through* 2015-12-01 17:45:49 imo alpine works fine for docker on both sides 2015-12-01 17:49:05 sadly, we have some pretty bad advice in our own bugtracker 2015-12-01 18:01:04 I don't think you can actually run docker inside lxc 2015-12-01 18:04:30 I guess I need to bump pkgrel to make builders work 2015-12-01 18:23:25 ncopa, also while reading https://lwn.net/Articles/661424/, reminded me of 'nlplug-findfs and cryptsetup' issue 2015-12-01 18:23:28 not sure, but maybe like stall a process till some device shows up then bring back from 'wait' 2015-12-01 18:23:31 but this feature "simple wait queues" is for kernal v4.4+ 2015-12-01 20:32:00 nghttp2 has yet to implement Alt-Svc - doc says, 'Alt-Svc specification is not finalized yet' 2015-12-01 20:36:30 looks like "libnghttp2_asio is C++ library built on top of libnghttp2 and provides high level abstraction API to build HTTP/2 applications. It depends on the Boost::ASIO library..." 2015-12-01 20:36:34 libnghttp2_asio is not built by default. Use the --enable-asio-lib 2015-12-01 20:41:23 and I found a road-map point for AL-3.6 (after super stable AL v3.4) ;) 2015-12-01 20:41:23 encourage devs/upstreams for C11/C++11 compatibility/enhancements 2015-12-01 21:02:42 i booted (from hdd) with a usb optical reader attached with disk inserted and nlplug-findfs "struggled" a bit 2015-12-01 22:19:08 ncopa: I found out that I'm already using the musl repo on my voidlinux server. The virtual machine (still alpinelinux edge) is running stable since 2 days now, 2015-12-01 22:19:42 Therefore something must be different between qemu on alpinelinux and qemu on voidlinux (musl). 2015-12-02 07:02:16 morning 2015-12-02 07:02:23 pluesch: interesting! 2015-12-02 07:02:36 it might be the configuration then 2015-12-02 07:03:09 or seabios 2015-12-02 07:16:47 morning 2015-12-02 07:48:12 it could possibly also be the cflags? 2015-12-02 07:51:18 pluesch: can you compare seabios versions? 2015-12-02 07:51:27 i think we use bundled 2015-12-02 07:51:31 thta's the easiest one 2015-12-02 07:51:32 so it looks same 2015-12-02 07:51:47 i looked for differences 2015-12-02 07:52:01 they have libnfs and snappy support 2015-12-02 07:52:03 we dont 2015-12-02 07:52:05 then I would guess CFLAGS 2015-12-02 07:52:06 i will enable that 2015-12-02 07:52:08 yes 2015-12-02 07:52:11 thats the second thing 2015-12-02 07:52:27 those configurations code path aren't touched 2015-12-02 07:52:35 i am disabling our custom cflags 2015-12-02 07:52:43 and use whatever the qemu build scripts uses 2015-12-02 07:52:54 not much people use -Os for qemu I guess 2015-12-02 07:53:34 might be there are some required cflag that gets disabled by us 2015-12-02 07:53:45 so i will use upstream cflag 2015-12-02 07:55:00 ncopa: which cflags was it for alpine? 2015-12-02 07:55:20 export CFLAGS="-Os -fomit-frame-pointer" 2015-12-02 07:56:58 pluesch: do you have a bug report for it? 2015-12-02 07:57:00 on bugs.a.o 2015-12-02 07:57:27 i'd like to refer to the issue on bugs.a.o in commit message 2015-12-02 07:58:25 FYI, gentoo doesn't filter CFLAGS for qemu 2015-12-02 08:00:01 pluesch: Oh and I do hope your host is running 64-bit OS, right? 2015-12-02 08:01:47 xming: so gentoo also uses upsteam default cflag instead of user cflags? 2015-12-02 08:02:27 no filtering of cflags on gentoo means that ebuild doesn't restrict any user specified flags 2015-12-02 08:07:43 QEMU itself filters some cflags 2015-12-02 08:08:26 so if -O2 solves the issue, I think it should be upstreamed 2015-12-02 08:11:45 still very wierd, I've never seen qemu-kvm w/o printing any errors when something bad happens 2015-12-02 12:30:58 pluesch: i pushed a rebuild of qemu. which uses default/upstream cflags 2015-12-02 12:31:08 should be closer to what voidlinux does 2015-12-02 18:36:30 hugoleodev: the wiki should be a good place to start http://wiki.alpinelinux.org/wiki/Main_Page 2015-12-02 18:56:10 ncopa, xming_, sry. I'm busy around the day mostly with work... 2015-12-02 18:56:40 i'll try the new build once I find some time. thank you! :3 2015-12-02 21:50:33 hi, I'm doing my another round of what-I-would-like-to-have-upgraded-in-Alpine-if-I-will-finally-use-it-daily reports. I have a suggestion. wouldn't it be easier if Package upgrade issues were not be closed, but only marked as resolved, until another upgrade request comes, that would switch the status? 2015-12-02 21:50:52 ncopa: what do you think about it? 2015-12-02 22:13:38 zfsonlinux guys want to reconsider and apply my patchset from last year, i'm not sure i have time for this in the short term: https://github.com/zfsonlinux/zfs/pull/2604#issuecomment-161443661 2015-12-03 08:53:21 ping @ncopa 2015-12-03 08:53:28 skarnet: poing 2015-12-03 09:36:04 ncopa: is there anything in particular that needs testing in rc1 2015-12-03 09:52:44 przemoc: i like to close issues 2015-12-03 14:06:49 ncopa: booting alpine-3.3.0_rc1-x86.iso, I see "WARNING! modloop boot option is missing." 2015-12-03 14:07:01 should this be a warning? I thought this was the new normal 2015-12-03 14:08:30 its a bug 2015-12-03 14:08:34 i saw it 2015-12-03 14:08:36 and will fix 2015-12-03 14:08:42 and setup-alpine shows a list of 16 ramX options of 0.0GB for installing to disk 2015-12-03 14:08:58 (booting with VirtualBox) 2015-12-03 14:30:13 ncopa: can we also change the default repositories file to include a reference to the community repository? 2015-12-03 14:30:20 we can leave it commented out, if you want 2015-12-03 14:34:46 ah good catch 2015-12-03 14:34:50 yes we shoudl do that 2015-12-03 16:55:32 ncopa: ok, I'll be creating new upgrade issue every time (unless it hasn't been resolved yet, of course) 2015-12-04 08:40:51 ncopa, #4891 propose a php-redis package, but we already have that in testing. I'm replying that a package is already available. Is it ok to make it "Resolved" ? 2015-12-04 08:44:48 fcolista: yes. we need feedpack if its working to be able to move it to main 2015-12-04 08:45:19 Ok. I mark it solved, and add this to the reply. Thanks. 2015-12-04 08:56:46 good morning 2015-12-04 08:57:15 hi leo-unglaub 2015-12-04 08:58:51 ncopa: did you get the email with the mkdir boot error? 2015-12-04 09:02:41 i got it 2015-12-04 09:02:48 sorry for not responding 2015-12-04 09:03:02 is it tmpfs root? 2015-12-04 09:03:33 or is it root on disk (cryptsetup, or plain disk or mdam or lvm) 2015-12-04 09:03:48 i tried to reproduce but was not able to 2015-12-04 09:06:26 dont worry about not responding it. i was just curious if it went thru because my android mail app crashes all the time *g* 2015-12-04 09:06:40 basically its not a big deal becasue the system still boots normaly 2015-12-04 09:07:09 my root is not encrypted (sadly, because its not possible on alpine) but everything else is cryptsetup without luks 2015-12-04 09:08:23 but root is on disk 2015-12-04 09:08:26 not on tmpfs? 2015-12-04 09:08:47 root is on disk 2015-12-04 09:09:01 then you should probably disable the modloop "service" 2015-12-04 09:09:31 is it on by default? 2015-12-04 09:09:41 because it cannot remember turning it on by myself 2015-12-04 09:09:44 its only a hack for mounting loopback image with kernel modules, when you dont have the kernel installe don system 2015-12-04 09:09:48 eg tmpfs root 2015-12-04 09:10:18 it is on by default when booting on tmpfs root 2015-12-04 09:10:44 so its on when booting install media 2015-12-04 09:11:03 i see, i have removed it 2015-12-04 09:11:09 brb, reboot for testing 2015-12-04 09:14:16 re 2015-12-04 09:14:20 ncopa: it worked 2015-12-04 09:14:25 the error is gone, thanke 2015-12-04 09:15:51 is that gerlish? 2015-12-04 09:15:59 :) 2015-12-04 09:16:11 no, its misspellish 2015-12-04 09:17:05 :) 2015-12-04 09:17:38 yer wilkomish :) 2015-12-04 09:20:28 haha 2015-12-04 09:23:41 oh, very nice. out of the 4 openssl issues, libressl is only afffected twice 2015-12-04 09:23:50 http://marc.info/?l=openbsd-tech&m=144920913324251&w=2 2015-12-04 09:24:14 i am seriously thinking of switching to libressl for v3.4 2015-12-04 09:25:00 Yes please 2015-12-04 09:25:36 Actually, an entirely different implementation of SSL would be even better, but i guess too much software depends on the OpenSSL API :/ 2015-12-04 09:26:52 do you volunteer to fix all the apps using the openssl api? ;) 2015-12-04 09:27:07 we had this discussion with kaniini a long time ago 2015-12-04 09:27:09 I do, in the next 3 or 4 decades 2015-12-04 09:27:27 emulating the api with a complete new crypto backend is horrible 2015-12-04 09:27:50 then we schedule for replace openssl for the next 3 or 4 decades :) 2015-12-04 09:28:01 leo-unglaub: horrible because the existing API is horrible? 2015-12-04 09:28:14 or horrible because the compat layer is bloat? 2015-12-04 09:28:18 yeah, the existing api is horrible 2015-12-04 09:28:51 the reason most people screw up ssl/tls is, because the openssl api is sooo horrible 2015-12-04 09:29:05 hmmm 2015-12-04 09:29:06 thats why libressl has the old api and a new one they use in openbsd very successfully 2015-12-04 09:29:10 libtls 2015-12-04 09:29:34 then libressl sounds like the right thing to use 2015-12-04 09:29:39 http://www.openbsd.org/cgi-bin/man.cgi/OpenBSD-current/man3/tls_accept_fds.3?query=tls_init&sec=3 2015-12-04 09:30:02 skarnet: well, polarssl -> now mbedtls is still better 2015-12-04 09:30:09 but the api is complete different 2015-12-04 09:30:31 I don't doubt it is better - I meant, for programs using the OpenSSL API 2015-12-04 09:30:58 if libressl works for them, and authors can be convinced to switch to libtls in time 2015-12-04 09:31:11 skarnet: you dont have to switch 2015-12-04 09:31:18 because they also have the old api 2015-12-04 09:31:29 that's the point 2015-12-04 09:31:31 but internally in openbsd projects they use the new api 2015-12-04 09:32:06 but if the old api is causing security issues, it needs to be deprecated at *some* point 2015-12-04 09:32:18 so using an implementation providing both 2015-12-04 09:32:28 is a way to nudge people to switch apis slowly 2015-12-04 09:33:00 and much more efficient than saying "ok, now you must use the mbedtls API and you have 0 days to do it" :P 2015-12-04 09:36:28 ncopa: may i ask what changed your mind about maybe using libressl in 3.4? 2015-12-04 09:36:46 we have talked about libressl earlier 2015-12-04 09:37:04 there have been and still are some things that needs to be investigated 2015-12-04 09:37:07 yeah, and there you did not seam to be very in to it 2015-12-04 09:37:09 like hw crypto support 2015-12-04 09:37:21 i am not very against it either 2015-12-04 09:37:24 i am just sceptic 2015-12-04 09:37:47 whenever there is a new hype thing 2015-12-04 09:38:07 we cannot just jump on ever new bling bling 2015-12-04 09:38:14 so libressl is relatively new 2015-12-04 09:38:26 i ask myself, will libressl still exist next year? 2015-12-04 09:38:53 back then the answer was: i dont know, but probably yes 2015-12-04 09:39:04 now the answer is: most likely yes 2015-12-04 09:39:21 also, things like you show here 2015-12-04 09:39:24 is not first time 2015-12-04 09:39:38 there have been more CVEs that only affect openssl 2015-12-04 09:40:04 those things togehter makes me think that libressl is probably a good idea 2015-12-04 09:40:15 i also like what they have done: remove lots of code 2015-12-04 09:40:49 yeah, thats what i like the most about it. coding styles are relative ... but removing stuff you simple dont need is a good thing in my mind 2015-12-04 09:40:52 i like small things 2015-12-04 09:41:06 last time we talked about it i think the conclusion was "not now, we are too close to v3.3" 2015-12-04 09:41:21 exactly 2015-12-04 09:41:43 i am generally positive to libressl 2015-12-04 10:04:48 looking at #4914 : wireshark ships gui qt by default 2015-12-04 10:05:37 so i remove wireshark-qt in behalf of wireshark-gtk subpkg 2015-12-04 10:20:50 oh boy, this bug tracker is horrible 2015-12-04 10:20:56 how can i close a ticket? 2015-12-04 10:21:27 i would like to clone #4704 2015-12-04 10:21:47 ncopa, if for you it's ok: http://sprunge.us/CREb 2015-12-04 10:21:55 it's about #4914 2015-12-04 10:26:52 ncopa: did you change anything in gitolite? I can't push to alpine-iso 2015-12-04 10:27:09 fatal: remote error: access denied or repository not exported: /alpine-iso 2015-12-04 10:27:27 barthalion: what is your push url? 2015-12-04 10:27:37 fcolista: looks ok to me 2015-12-04 10:27:41 fcolista: push it 2015-12-04 10:27:46 oh, okay 2015-12-04 10:27:48 silly me 2015-12-04 10:27:54 ok. Thx ncopa 2015-12-04 10:28:05 thanks 2015-12-04 10:28:16 np :) 2015-12-04 10:41:17 ncopa: I noticed I am not getting any mail from the redmine. Is that to be expected? 2015-12-04 10:42:02 depends on your settings i suppose 2015-12-04 10:42:17 i can check the mail log 2015-12-04 10:53:21 gmail constantly categorizes alpine redmine mails as "Promotions" even though I've changed them many times already (it's supposed to be learning IIRC). guess will have to add a filter for it... 2015-12-04 10:54:35 BTW thanks for resolving some issues already, ncopa and fcolista 2015-12-04 10:57:38 przemoc: there is an easy solution to that, dont use gmail /nsamail ;) use your own mailserver with alpine on it *g* 2015-12-04 10:58:04 przemoc, np 2015-12-04 10:58:16 leo-unglaub, :) +1 2015-12-04 11:00:55 well, own mailserver is a hassle if you want it to be "respected" by other servers, otherwise your mails are easily treated as spam 2015-12-04 11:01:24 (haven't checked mailserver-related options available in alpine out-of-the-box yet, though) 2015-12-04 11:04:11 eheh, re sngrep (#4923). I've sent an email to the author telling him the availability of sngrep for alpine 2015-12-04 11:04:13 https://github.com/irontec/sngrep/wiki/Installing-Binaries#alpine-linux 2015-12-04 11:04:22 He already updated the wiki page. 2015-12-04 11:05:27 alpine already has some voip-related tools, so sngrep seemed very fitting 2015-12-04 11:08:32 yes 2015-12-04 11:08:42 we are still having issues with ncurses though 2015-12-04 11:08:49 sngrep segfaults 2015-12-04 11:09:12 [143467.396446] sngrep[18066]: segfault at 25255440 ip 00006edc444cb52e sp 000075c99d3d9c10 error 6 in libncurses.so.6.0[6edc444b6000+246000] 2015-12-04 11:09:12 [143467.396456] grsec: Segmentation fault occurred at 0000000025255440 in /usr/bin/sngrep[sngrep:18066] uid/euid:0/0 gid/egid:0/0, parent /bin/busybox[ash:17640] uid/euid:0/0 gid/egid:0/0 2015-12-04 11:09:12 [143467.396463] grsec: denied resource overstep by requesting 4096 for RLIMIT_CORE against limit 0 for /usr/bin/sngrep[sngrep:18066] uid/euid:0/0 gid/egid:0/0, parent /bin/busybox[ash:17640] uid/euid:0/0 gid/egid:0/0 2015-12-04 11:17:19 ncopa, #4917 it's a major release. We should wait for 3.3 completion, then push the new version, right? 2015-12-04 11:26:32 fcolista: that smells like the issue we are having with alsamixer 2015-12-04 11:26:52 i think ncopa was going to fix it, but probalby something got first. 2015-12-04 11:27:30 i remember it was some kind of abi issue with ncurses on musl. 2015-12-04 11:38:54 I'm planing on half-automating version tracking in near future, so I could report package upgrade issues early, not like now, i.e. "oh, it's already new alpine rc? I need to check it doesn't have too old stuff or lacks some too, so I'll finally smoothly switch to its upcoming version this time..." ;) 2015-12-04 11:44:01 (and when I'll finally do this friggin switch to alpine, then I'll be sending APKBUILD patches to ML myself, hopefully lessening maintenance burden a bit) 2015-12-04 11:45:34 workflow related question. in future would you prefer to have package upgrade issues in redmine filled even if I would be sending patches to ML? 2015-12-04 11:53:01 having old samba in 3.3 would be sad, but at least I still don't have any w10 in home. (I recommend Spybot Anti-Beacon, Destroy Windows Spying and GWX Control Panel if you deal with W7/8 and don't want to upgrade or accept some telemetry crap from redmond) 2015-12-04 12:04:44 przemoc, we have samba 4.2 for alpine 3.3 2015-12-04 12:05:18 oh, misread, sorry. I thought you were saying we had Samba 3.3 2015-12-04 12:12:58 I generally prefer to have as latest as possible versions (unless they're broken), because upstream always prefer dealing with bugs on recent version than the old ones. and newer versions are often (but not always, of course) simply better. 2015-12-04 12:51:50 fcolista: i tried samba 4.3 some time ago 2015-12-04 12:51:57 it was big job 2015-12-04 12:52:03 so i postponed it 2015-12-04 12:52:09 never managed to finish it 2015-12-04 12:52:13 ncopa, ok 2015-12-04 12:52:22 i've two patches that are working so far 2015-12-04 12:52:35 it's still building 2015-12-04 12:52:35 if we upgrade to samba 4.3 we will probably also need to rebuild alot 2015-12-04 12:52:40 yeah 2015-12-04 12:52:48 that's what i was thinking 2015-12-04 12:52:54 biggest problem is all the zillion libraries 2015-12-04 12:53:03 i think there are like 300 libraries or so 2015-12-04 12:53:25 problem is how to split the subpackages without causing circular deps etc 2015-12-04 12:53:35 i think i spent a day or two for samba 4.2 2015-12-04 12:54:26 i suggest we do samba 4.3 after v3.3 2015-12-04 12:56:07 przemoc: just submit patches 2015-12-04 12:57:00 300 libraries for samba? o.O 2015-12-04 12:57:03 what the 2015-12-04 13:10:34 maybe it was 400, i dont remember exactly 2015-12-04 13:13:17 ok, sounds like upgrading samba is a lot more hassle than it seemed. 2015-12-04 13:17:12 barthalion: so regarding my workflow-related question. if you'd prefer to avoid having issues in redmine (are you sure? ncopa said he likes to close issues :>) , then maybe alpine should switch redmine to something else if using it is a problem (but I cannot recommend any alternative, though, IMO all those trackers are too simple or overly complex) 2015-12-04 13:19:45 maybe we should do it the arch linux way add a button to pkgs.alpinelinux.org which allows users to flag packages as out of date and sends the maintainer an email? 2015-12-04 13:20:03 nmeum: we have that already 2015-12-04 13:20:15 really? 2015-12-04 13:20:34 i once in a while get an email that package is flagged out of date 2015-12-04 13:20:35 ncopa: we do? 2015-12-04 13:20:40 oh, I never got one 2015-12-04 13:21:16 so my carefully crafted package upgrade issues were total waste of time? :( 2015-12-04 13:21:23 clandmeter: didnt we have a 'flag outdate' button on pgs.a.o? 2015-12-04 13:21:26 no 2015-12-04 13:21:30 its not a waste 2015-12-04 13:21:38 we get notified about it 2015-12-04 13:21:52 and bugs.a.o makes us remember update before we ship 2015-12-04 13:22:04 email notification to me will risk drown in my inbox 2015-12-04 13:22:36 so your opinion is about different than barthalion? I mean re: 12:44 < przemoc> workflow related question. in future would you prefer to have package upgrade issues in redmine filled even if I would be sending patches to ML? 2015-12-04 13:23:00 s/about/a bit/ 2015-12-04 13:23:17 fabled: re #4913, it seems that "use modloop" does not work as expected 2015-12-04 13:23:53 submit patch is a good alternative. if test that it builds before you send it 2015-12-04 13:25:04 strace 4.10 has some arm related patch that needs to be rebased. it was non-trivial so i postponed it 2015-12-04 13:25:08 i want v3.3 out 2015-12-04 13:25:09 in fact 2015-12-04 13:25:18 monday i start new job 2015-12-04 13:25:35 so the fact that i was not able to get v3.3.0 out today is not good at all 2015-12-04 13:25:43 ok. what about patches sent too late in the cycle, do you queue them in mailbox or they would be forgotten without redmine issue? 2015-12-04 13:25:57 seems like the issues are coming in faster than we manage to resolve them 2015-12-04 13:26:10 patchwork.alpinelinux.org 2015-12-04 13:26:31 good to know 2015-12-04 13:27:00 at this stage i'd prefer only spend tme on looking at patches that fixes existing issues rater than create new ones 2015-12-04 13:27:06 ncopa: you found a new job? That's great \o/ 2015-12-04 13:27:08 "hte new versino X intrdoced foo" 2015-12-04 13:27:13 yup 2015-12-04 13:27:19 ncopa: is it alpine related? 2015-12-04 13:27:28 yup, i think so 2015-12-04 13:27:33 cool 2015-12-04 13:28:11 there was only rc1 so far, so it's rather early for final release. IIRC year ago there were a few RCs. 2015-12-04 13:28:23 yes, we are late 2015-12-04 13:28:25 very late 2015-12-04 13:29:31 if you would release rc0 earlier, then I would be possibly triggered earlier with checking alpine packages state. ;) 2015-12-04 13:29:32 yeah, 3.3 was actually supposed to be released in november, right? 2015-12-04 13:30:05 but I'll resolve it in future with my vracker 2015-12-04 13:34:04 yea. sorry for losing my job on an inconveninet time 2015-12-04 13:34:47 maybe new one will be much better at least? 2015-12-04 13:34:56 drbd has a patch that neesd to be rebased 2015-12-04 13:34:59 it will 2015-12-04 13:36:59 build system of drbd is different 2015-12-04 13:37:42 ======================================================================= 2015-12-04 13:37:42 With DRBD module version 8.4.5, we split out the management tools 2015-12-04 13:37:42 into their own repository at http://git.linbit.com/drbd-utils.git 2015-12-04 13:37:42 (tarball at http://oss.linbit.com/drbd) 2015-12-04 13:37:50 that kind of changes are too late for v3.3 2015-12-04 13:38:34 sure thing, postpone whatever needs fiddling. 2015-12-04 13:39:07 just a question, is it supported to have same package in main and testing (with different versions, of course)? 2015-12-04 13:41:18 or living on edge is the only option in alpine to get stuff that didn't get into release? 2015-12-04 13:43:04 ACTION is afk 2015-12-04 13:51:03 ncopa: yes, you can click the version number on pkgs, and flag it out of date. 2015-12-04 13:51:35 a list can be found here: https://pkgs.alpinelinux.org/flagged 2015-12-04 14:05:31 great 2015-12-04 14:05:50 clandmeter: why is nmap 6.47-r0 in that list we already upgraded it to 7.00 2015-12-04 14:06:15 and go 1.4.2 doesn't exist either anymore 2015-12-04 14:07:22 nmeum: it lists everything that ever got flagged. 2015-12-04 14:07:46 only regular listing gets updated on a pkg update. 2015-12-04 14:07:54 ah, wouldn't it make sense to remove packages from the database that have been upgraded? 2015-12-04 14:08:07 yes and no 2015-12-04 14:08:25 yes, in pkgs listing it gets reset. 2015-12-04 14:08:42 no, this way we can check which pkgs got flagged and for what reason. 2015-12-04 14:08:57 its just a reference. 2015-12-04 14:09:14 ok, makes sense 2015-12-04 14:09:16 i didnt add a link to the webif for others to actually find it. 2015-12-04 14:09:31 ncopa: mail is working now, thanks 2015-12-04 14:09:38 but you are right 2015-12-04 14:09:47 i think i should add a field that lists it got updated. 2015-12-04 14:10:10 i never really payed attention it it. 2015-12-04 14:10:17 accept now i find out there is a bug. 2015-12-04 14:16:31 clandmeter: how do I actually flag a package as out of date? 2015-12-04 14:16:45 the version number is a link 2015-12-04 14:16:59 ah 2015-12-04 14:17:03 green means ok, red means flagged. 2015-12-04 14:17:11 nice 2015-12-04 14:17:21 its not that obvious i know :) 2015-12-04 14:17:56 lol 2015-12-04 14:18:16 i just got a msg i flagged a package. 2015-12-04 14:18:27 i guess its the message from somebody else. 2015-12-04 14:18:32 lol? I just flagged one of my own packages :p 2015-12-04 14:18:36 why did you get the mail? 2015-12-04 14:19:16 no, i mean i got the html msg 2015-12-04 14:19:33 i pressed a link and got the msg i flagged something. 2015-12-04 14:19:37 bad coding... 2015-12-04 14:19:39 wow, that's interessting 2015-12-04 14:19:56 and i got the email aswell 2015-12-04 14:20:05 for android-tools? 2015-12-04 14:20:09 yes 2015-12-04 14:20:18 and its addressed to somebody else. :) 2015-12-04 14:20:18 that sounds…wrong 2015-12-04 14:20:23 Dear Sören Tempel 2015-12-04 14:20:25 it's addressed to me 2015-12-04 14:20:28 yeah that's me 2015-12-04 14:20:33 did you get it aswell? 2015-12-04 14:20:40 yep 2015-12-04 14:21:31 its pretty weird. 2015-12-04 14:22:11 ineed 2015-12-04 14:22:13 *indeed 2015-12-04 14:31:18 clandmeter, yeah, i got my email also 2015-12-04 14:31:22 but also bunch of others... 2015-12-04 14:31:43 fabled: but i guess all addressed to you? 2015-12-04 14:31:49 no 2015-12-04 14:31:54 <_mjones> BCC? 2015-12-04 14:31:59 yes, i got two as bcc 2015-12-04 14:32:25 i got emails about dpkg and android-tools 2015-12-04 14:32:41 they were address to ncopa, and nmeum; but i still got a blind copy 2015-12-04 14:32:56 :) 2015-12-04 14:33:02 i think i see the issue 2015-12-04 14:33:16 the SMTP RCPT TO is not reset? 2015-12-04 14:33:33 the object doesnt get reinitiated 2015-12-04 14:33:58 so the previous values are still set. 2015-12-04 14:34:20 you rely on lua garbage collection? 2015-12-04 14:34:24 <_mjones> interesting bug. Well, sorta. 2015-12-04 14:34:26 connection:close() 2015-12-04 14:34:53 it was my first lua scripting. 2015-12-04 14:34:58 which figures :) 2015-12-04 15:11:50 sorry ppl, i think i just fixed the issue about ppl get non related flagged emails. if it happens again please let me know. 2015-12-04 15:13:11 clandmeter, great. thanks. 2015-12-04 18:26:08 php v7.0 released !! 2015-12-05 12:08:51 hi friends 2015-12-05 12:20:49 hi yourself 2015-12-06 16:38:29 to mark issue as duplicate some special privilege is required in redmine? 2015-12-06 16:38:31 http://bugs.alpinelinux.org/issues/3950 2015-12-06 16:42:53 not sure about the permissions, but I marked it 2015-12-06 16:44:41 thanks. should the duplicated one (#3950) be closed in such case, so there is only one (#1068) related to this matter, or not? 2015-12-06 16:49:12 no idea, sorry 2015-12-06 17:42:35 please mark #4215 as duplicate of #3949 2015-12-06 17:50:11 are you sure you cant? you need to use the little add link on the right of the "related issues" listing 2015-12-06 17:55:29 I don't see any "related issue" item on the page at all. can you send some screenshot showing it? 2015-12-06 17:56:35 I'm blind or it's indeed a matter of some privileges in redmine (I'm Reporter only) 2015-12-06 17:57:47 chris|: ^ 2015-12-06 18:02:38 actually don't have any screenshot tool installed atm ^^ 2015-12-06 18:02:46 this should show it though: https://thumbtack-technology.github.io/redmine-issue-hot-buttons/screenshots/screenshot_time_tracker.png 2015-12-06 18:04:28 ok, I don't have it. related issue subsection is shown to me only when there are already some related issues. 2015-12-06 18:08:36 thanks, chris| 2015-12-06 18:09:27 sure 2015-12-06 18:10:33 you might try pinging ncopa tomorrow, I think its silly you can close issues mit not tag duplicates. Probably only a configuration issue 2015-12-06 18:13:45 ok. just going through some issues in Alpine's redmine to clean it a bit. 2015-12-06 18:43:16 could you move issue #4605 from Alpine Documentation to Alpine Linux 2015-12-06 18:56:02 sorry, I don't have any permissions on the documentation project either 2015-12-06 20:41:42 ncopa, chris|: we need to talk after 3.3 leaves the nest 2015-12-06 20:42:32 about what? :) 2015-12-06 20:53:17 kidneys, of course 2015-12-06 20:53:45 does Valery use IRC? 2015-12-06 20:56:40 no idea, sorry 2015-12-06 21:02:47 I perfer mine with onions, btw. 2015-12-07 08:37:54 For a package that I would like on alpine linux, but I do not own the project, would it be better to file an issue or submit a package to the alpine-devel mailing list? 2015-12-07 08:38:05 Here it is libunwind 2015-12-07 08:40:58 if you have ready APKBUILD, sending a patch is enough 2015-12-07 08:41:21 if you don't, and you do not want to write it on your own, I guess you should file an issue 2015-12-07 08:42:41 I do have a ready APKBUILD, I guess I'll send the patch then 2015-12-07 08:45:02 sorry, not devel, aports, but yeah, same idea 2015-12-07 08:48:53 morning 2015-12-07 08:49:31 bpye: what's package? 2015-12-07 08:49:48 libunwind? 2015-12-07 08:49:54 libunwind? It's a dependency for .NET core which is ultimately what I'm working on 2015-12-07 08:49:57 Yeah 2015-12-07 08:50:42 does it build for you? 2015-12-07 08:51:12 Once I disable the tests, yes 2015-12-07 08:51:46 The only build failure is in a test, the release versions are already part configured which makes disabling a single test non trivial though so I disable all tests. They don't run when building in any case.. 2015-12-07 08:53:34 ScrumpyJack: I'm curious, sounds like you've tried building it yourself, what was your failure? 2015-12-07 08:55:41 nope, never tried building that 2015-12-07 09:00:41 Ah okay 2015-12-07 09:03:26 clandmeter: you about? 2015-12-07 09:03:36 yes 2015-12-07 09:04:34 you're down as the contributor for kodi :) 2015-12-07 09:04:50 just wondering why is has openjdk as a deps 2015-12-07 09:05:26 can you remember? 2015-12-07 09:05:30 sure 2015-12-07 09:05:45 its needed to gererate something at build time. 2015-12-07 09:06:01 im not sure what exactly 2015-12-07 09:06:50 hmm. I'm trying to build it on armhf without a jdk,so we'll soon see :) 2015-12-07 09:07:02 http://forum.kodi.tv/showthread.php?tid=140526 2015-12-07 09:17:34 there are some posts from ncopa on that thread. boostrapping on another machine would work if we could boostrap and x86_64 and move that to arm. If java is only used for binding stuff before build in order to make docs, perhaps we can build kodi on arm without the docs? 2015-12-07 09:17:55 boostrapping? what's that? 2015-12-07 09:18:21 s/boost/boots of course 2015-12-07 09:19:39 (or we build openjdk for arm? is there an issue there as well?) 2015-12-07 09:34:27 <_mjones> Are you asking what is bootstrapping? 2015-12-07 09:44:31 _mjones: nope. just trying to build kodi on arm. it requires java during bootstrap, which we don't have on arm. 2015-12-07 09:45:39 so either we boostrap kodi elseware and move it to arm to build, or remove whatever requires java (the docs i think) or build java for arm :) 2015-12-07 09:45:51 or any other ideas 2015-12-07 09:46:51 <_mjones> oh, boostrapping. I see what you did there. I misread that four times. 2015-12-07 09:47:16 <_mjones> openjdk is unfortunately a pain. I only tried to upgrade the version once, and I never finished. 2015-12-07 09:50:08 How can I get the contents of an .apk without installing it? 2015-12-07 09:50:21 pkgs.alpinelinux.org 2015-12-07 09:50:27 untar 2015-12-07 09:50:48 Ah it's a tar not zip, my guess was wrong 2015-12-07 09:51:12 zip, oh my :p 2015-12-07 09:53:01 Thanks though, just wanted to check that the contents of the package I submitted was correct 2015-12-07 09:53:17 ScrumpyJack: Gentoo can build kodi w/o java 2015-12-07 09:53:32 ooo, i'll take a look 2015-12-07 09:53:36 it downloads pregenerated stuff 2015-12-07 09:53:38 !java? ( http://mirrors.kodi.tv/releases/source/${MY_P}-generated-addons.tar.xz )" 2015-12-07 09:54:23 yes, that's one of the options - boostrapping elsewhere 2015-12-07 09:54:29 but how/where? 2015-12-07 09:54:44 ScrumpyJack: I have no clue if it actually works 2015-12-07 09:54:54 ScrumpyJack: it seems that kodi is providing that 2015-12-07 09:55:05 and it doesn't build on arm in gentoo 2015-12-07 09:55:23 (or they haven't tried) 2015-12-07 09:55:34 https://packages.gentoo.org/packages/media-tv/kodi 2015-12-07 09:57:57 http://forum.kodi.tv/showthread.php?tid=238645 2015-12-07 09:58:14 built with OpenGL which is wrong, but it builds :p 2015-12-07 09:58:50 but he probably had java 2015-12-07 09:59:32 from that thread: "cross compile. Every other method is unsupported here" 2015-12-07 10:01:10 unsupported doens't mean it won't work, I know you can accomplish this :D 2015-12-07 10:02:44 needing java for non java package is pretty insane 2015-12-07 10:05:58 ScrumpyJack: ncopa&kodi http://forum.kodi.tv/showthread.php?tid=140526 2015-12-07 10:06:34 I think you can do w/o java with the pregenerated stuffs 2015-12-07 10:16:22 what is this pregenerate stuff? and where to find it? 2015-12-07 10:17:33 ScrumpyJack: it doesnt build with openjdk7? 2015-12-07 10:19:36 seems vanished from kodi's mirrors, but void has it 2015-12-07 10:19:36 http://repo2.voidlinux.eu/distfiles/ 2015-12-07 10:19:40 it's pregenerated bindings 2015-12-07 10:22:14 _mjones: here the 2015-12-07 10:22:22 _mjones: here the WIP page http://wiki.alpinelinux.org/wiki/Alpine_on_ARM 2015-12-07 10:23:25 <_mjones> xming_: yes I saw. I decided to start with Debian to make sure I had the qemu part working properly. But I can't actually get Debian to work. Sigh. 2015-12-07 10:32:57 _mjones: I had debian initramfs booting and working 2015-12-07 10:35:36 _mjones: the alpine's arm kernel has issues not booting 2015-12-07 10:35:40 with qemu 2015-12-07 10:37:26 <_mjones> xming_: I have what seem to be simple parameters in qemu, but I'm getting absolutely nothing after starting qemu. I used debootstrap to build the sysimage but I'm not seeing anything obviously wrong with it. 2015-12-07 10:42:38 hi there! 2015-12-07 10:43:26 hi guys 2015-12-07 10:43:46 i need a path to download alpinelinux logo with transparent background 2015-12-07 10:43:47 _mjones: which kernel? 2015-12-07 10:45:48 hopefully my ttl cable will arrive today 2015-12-07 11:04:46 fcolista: http://git.alpinelinux.org/cgit/alpine-mksite/plain/_static/alpine-logo.png 2015-12-07 11:05:00 thx nmeum 2015-12-07 11:12:20 clandmeter: I was looking at the wrong repo. it seems we hve sdk7 in edge :) 2015-12-07 11:12:57 I'll give that a shot once I get distcc working on a bunch of scaleway hosts 2015-12-07 11:36:40 reporter in redmine cannot mark issue as duplicate - I noticed it yesterday. please fix it. (not sure whether ncopa is the only one with powers to do so or not) 2015-12-07 12:52:33 fcolista: alos here: http://alpinelinux.org/alpine-logo.png 2015-12-07 13:22:39 ncopa, i've created an ansi and ascii alpine linux logo and added to linux_logo. 2015-12-07 13:22:48 Any feedback are welcome 2015-12-07 13:23:15 where can i find it? 2015-12-07 13:23:25 testing/linux_logo 2015-12-07 13:23:48 what's the best way to find out which packages need to be rebuild after an soname bump? 2015-12-07 13:25:43 apk search -r --quiet --origin so:libsomething.so.0 2015-12-07 13:25:48 thanks 2015-12-07 13:25:53 where libsomething.so.0 is the old so name 2015-12-07 13:26:08 right 2015-12-07 13:29:20 ncopa: isnt that what checkapk is for? 2015-12-07 13:29:56 +1 2015-12-07 13:30:36 checkapk tells you if/which library has ABI change 2015-12-07 13:30:46 but don't tell you what packages needs rebuild 2015-12-07 13:30:53 which is what nmeum is looking for 2015-12-07 13:31:04 so, it's a mix of both 2015-12-07 13:31:23 checkapk + apk search -r --quiet --origin so:libsomething.so.0 2015-12-07 13:31:36 ah you are right, i didnt read the question correctly. 2015-12-07 15:14:26 anyone tried distcc ? 2015-12-07 15:15:13 clandmeter: why did you set arch to empty string in the clisp APKBUILD? 2015-12-07 15:17:10 nmeum: i have no idea 2015-12-07 15:17:17 bad commit msg.. 2015-12-07 15:21:17 nmeum: could be it builds locally, but it fails on builder (due to grsec/pax) 2015-12-07 15:22:00 maybe you should move it to unmaintained/ then? 2015-12-07 15:26:21 nmeum: your wish is my cmd (to some extent..) 2015-12-07 15:26:52 anyone had success in using lxc-attach ? 2015-12-07 15:26:53 grsec: use of CAP_SYS_ADMIN in chroot denied for /usr/bin/lxc-attach[lxc-attach:9547] 2015-12-07 15:27:05 i've this error when try to spawn a shell from the host 2015-12-07 15:27:15 try setting the kernel.grsecurity.chroot_caps to 0 2015-12-07 15:27:20 *sysctl 2015-12-07 15:27:33 what's the backfire of it? 2015-12-07 15:28:11 btw: with that command, lxc-attach works. 2015-12-07 15:29:00 chroot_caps probably disables capget(2) and capset(2) in chroots 2015-12-07 15:41:21 I always use lxc-console 2015-12-07 15:43:14 or chroot /path/to/container /bin/bash (for a debian container to set a root passwd) 2015-12-07 15:44:30 i want to check the ip address of the lxc container 2015-12-07 15:44:35 which is set via dhcp 2015-12-07 15:48:34 fcolista: there are python tools for that, which we have disabled. i suppose we should implement those in lua or something 2015-12-07 15:49:02 ncopa, are those disabled in lxc package? 2015-12-07 15:49:50 yes 2015-12-07 15:50:07 i dont want pull in python dependency 2015-12-07 15:50:31 if those tools are enabled i think lxc-start and/or other similar are replaced with a python implementation 2015-12-07 15:51:14 you're wise 2015-12-07 15:57:40 price is some missing feauters like this, which can be handy 2015-12-07 15:57:56 i was actually thinking of implementing an lxc-stat tool, similar to vserver-stat 2015-12-07 15:58:11 but have not had the time :-/ 2015-12-07 16:10:58 does anybody have any why alsamixer is segfaulting? 2015-12-07 16:11:12 since this is a pretty common programm we probably want to fix that before 3.3 is released 2015-12-07 16:15:04 oh, indeed 2015-12-07 16:15:11 i have no idea why it happens 2015-12-07 16:15:26 and yes, we probably want fix it before v3.3 2015-12-07 16:25:55 > #0 0x0000707f0769f52e in waddch_literal (win=win@entry=0x2d76ddf81f0, ch=2097696) at ../../ncurses/base/lib_addch.c:385 2015-12-07 16:25:58 > 385../../ncurses/base/lib_addch.c: No such file or directory. 2015-12-07 16:26:00 interessting… 2015-12-07 16:26:20 so it's probably an ncurses problem and not an alsamixer problem it seems 2015-12-07 19:34:35 ncopa: libreoffice doesn't seem to finish building on armhf but there is no buildlog in http://dev.alpinelinux.org/buildlogs/build-edge-armhf/main/libreoffice/ any ideas what's going wrong? 2015-12-07 21:06:50 good evening gentlemen 2015-12-08 05:24:34 what size are the repos roughly ? (to run a mirror) 2015-12-08 07:23:41 nmeum: it needs a alot of memory and arm is probably swapping 2015-12-08 07:23:59 it will probably need another day 2015-12-08 07:44:23 ncopa. re: alpine security issues. 2015-12-08 07:44:45 for applying patches in other alpine's version 2015-12-08 07:45:02 is enough change the branch and commit there? 2015-12-08 07:45:17 Or i need to have a buildenv for each alpine version? 2015-12-08 07:45:46 I thing changing the branch would work. But i'm not sure about the build.. 2015-12-08 07:46:22 you need a buildenv for every branch 2015-12-08 07:47:04 OK 2015-12-08 07:47:06 Ok 2015-12-08 07:47:07 you could in theory bind mount your home dir or aports dir in vaorious different build env 2015-12-08 07:47:21 then you could git checkout 2015-12-08 07:47:29 i've two lxc containers atm 2015-12-08 07:47:33 both for edge 2015-12-08 07:47:37 one x64 and one x86 2015-12-08 07:48:09 so i need to create other buildenv. 2015-12-08 07:48:15 i have x64 for edge, 3.2, 3.1 and 3.0 2015-12-08 07:48:31 and x86 for v2.7 2015-12-08 07:48:46 iirc there's a script done by kaniini that creates easily a buildenv 2015-12-08 07:48:47 i alos have a x86 edge env 2015-12-08 07:48:56 good 2015-12-08 07:49:16 my main OS in the WS is alpine edge 2015-12-08 07:49:20 i was thinking of build env in docker 2015-12-08 07:49:30 that's the same i was thinking as well :) 2015-12-08 07:49:46 would make it super simple to set up a compile env for 2015-12-08 07:49:54 exactly. 2015-12-08 07:50:02 In this case, docker would work _a lot_ 2015-12-08 08:41:51 good morning 2015-12-08 08:47:13 morning 2015-12-08 08:47:56 has anyone of you managed to get access to the files on an android phone while beeing on alpine? 2015-12-08 08:48:04 is there a special package to use? 2015-12-08 08:56:59 morning 2015-12-08 09:02:13 hey ScrumpyJack 2015-12-08 09:04:22 leo-unglaub, i think you need MTP support 2015-12-08 09:07:48 hmm, thats gvfs stuff, right? 2015-12-08 09:08:01 hmm, damn ... i wanted to get rid of that layer of bloat .... 2015-12-08 09:08:03 but okay ... 2015-12-08 09:16:48 reboot 2015-12-08 09:48:38 leo-unglaub: does adb need mtp? or does adb pull not satisfy? 2015-12-08 09:49:23 clandmeter: i have no idea 2015-12-08 09:49:44 i just need to push a file on my phone to reflash it because the current cyogenmod updater is broken :( 2015-12-08 09:50:20 im not sure on linux, but on windows adb would be enough. 2015-12-08 09:51:44 leo-unglaub: you cannot just download the latest cyanogenmod release and flash it from recovery? 2015-12-08 09:51:56 i mean download it from your phone directly 2015-12-08 09:52:06 you dont need mtp but adb doesn't have right access rights to the usb device by default, there are some udev rules or you run adb as root 2015-12-08 09:53:48 leo-unglaub: i had the same issue with my N5, the updates wants to upgrade to Cyanogenmod v13 which is Android 6.0, and it needs the 6.0 bootloader and radio to able to flash it. 2015-12-08 09:54:35 clandmeter: dont upgrade to android 6 2015-12-08 09:54:49 everything is broken, i could not do calls anymore, the gapps crashed every 10 secounds, ... 2015-12-08 09:54:59 :) 2015-12-08 09:55:11 i turned back to vanilla 5.1.1 2015-12-08 09:55:14 and i lost all contacts and calendar entries 2015-12-08 09:55:25 6.0 vanilla is also horrible. 2015-12-08 09:55:38 dialer has issues 2015-12-08 09:55:45 yeah, google really did a horrible quality control this time 2015-12-08 09:55:48 and my battery doesnt last a day. 2015-12-08 09:56:06 oh and 6.0 has a bluetooth bug aswell 2015-12-08 09:56:20 making my carkit go crazy 2015-12-08 09:56:42 ncopa, what's the best way to share a dir among containers? 2015-12-08 09:56:54 as much as i hate apple, there iphone works better than android ... 2015-12-08 09:57:10 fcolista: i mount them from the config 2015-12-08 09:57:41 what about the permissions? 2015-12-08 09:57:54 uid/gid might be different 2015-12-08 09:57:56 volume i think 2015-12-08 09:58:25 you might need copy the user id 2015-12-08 09:59:15 did nfs work on guests as client? 2015-12-08 10:00:33 leo-unglaub: i dont hate apple, but i must agree, ios seems much more stable. 2015-12-08 10:31:35 uh? 2015-12-08 10:31:36 ERROR: http://nl.alpinelinux.org/alpine/v2.7/main: UNTRUSTED signature 2015-12-08 10:32:26 I'm still on android 4.4.4 on my Z1C. constantly ignoring available update to 5, which is UX downgrade IMHO from what I've seen in others phones/tablets. the only complain I have is inability to delete stuff on uSD card, but it's not big deal. 2015-12-08 11:22:15 hmmm, getting my build environment to run on my laptop is harder than i thought ... those lxc containers are strange 2015-12-08 11:22:33 is it normal that only root can use lxc-console to connect to a container? 2015-12-08 11:26:23 ncopa: is there any reason for lxc-templates to depend on tar? 2015-12-08 11:26:31 busybox already ships tar 2015-12-08 11:26:54 because now tar has overriden the busybox tar 2015-12-08 11:35:12 leo-unglaub - it is needed for the download template (unprivileged containers) 2015-12-08 11:35:40 BitL0G1c: but why not use the busybox tar? 2015-12-08 11:35:47 its basically the same, just a lot smaller 2015-12-08 11:35:55 see the commit message 2015-12-08 11:36:01 it is missing an option 2015-12-08 11:37:33 doesn''t have 'tar --anchored' http://git.alpinelinux.org/cgit/aports/commit/?id=9e6bffefe49c4aa06be92dcd70ca0a20ee8f8c93 2015-12-08 11:43:49 hmm, just for the tar --anchored option? 2015-12-08 11:43:58 thats sume stupid gnu invention ... 2015-12-08 11:44:04 hmmm 2015-12-08 12:03:10 leo-unglaub: see git log 2015-12-08 12:03:57 http://git.alpinelinux.org/cgit/aports/log/main/lxc 2015-12-08 12:04:11 http://git.alpinelinux.org/cgit/aports/commit/main/lxc?id=9e6bffefe49c4aa06be92dcd70ca0a20ee8f8c93 2015-12-08 12:05:17 thanks 2015-12-08 15:00:18 yuhuuuuu, my lxc container finaly has networking :) 2015-12-08 15:44:57 Hello 2015-12-08 15:47:00 I need some guiding help in compiling ipop-project https://github.com/ipop-project/Documentation/wiki/Building-the-code-for-Linux for Alpinelinux 2015-12-08 15:50:53 I hope one day to have a working APKBUILD. Please advice me if it willl be posible to compile it or this package cant be compiled for Alpinelinux 2015-12-08 15:51:16 thanks 2015-12-08 16:56:36 fakeroot question. what is the effective id of a process running with fakeroot? 2015-12-08 16:57:17 ? as in what it sees itself as having? or what it actually runs with 2015-12-08 16:57:46 root righT? 2015-12-08 16:58:54 it's messing up distcc when using distcc to abuild alpine packages 2015-12-08 16:58:58 0, yeah 2015-12-08 16:59:40 distcc tries to create a bunch of files (lock files, temp files) in /root 2015-12-08 17:02:16 fun 2015-12-08 17:12:01 could someone please apply this version bumb? http://dpaste.com/1PDS1G7.txt 2015-12-08 17:12:05 thanks! 2015-12-08 17:15:12 i think i might have the wrong end of the stick https://github.com/distcc/distcc/issues/166 2015-12-08 17:15:36 Arch and gentoo seem to use distcc fine in fakeroot 2015-12-08 20:31:17 ScrumpyJack: IIRC there are some DISTCC_DIRC_* which you can set 2015-12-09 02:42:28 How much work is it typically to patch glibc code for musl? 2015-12-09 02:43:44 <_mjones> 'Depends'. 2015-12-09 02:43:55 OK average package? 2015-12-09 02:43:55 <_mjones> Am doing a bit right now. 2015-12-09 02:44:19 <_mjones> Big complicated C++ beast with many dependencies is the worst case I've had (not completed successfully by me). 2015-12-09 02:44:52 <_mjones> Smallish C that already works on a BSD as well as Linux, probably easy to trivially easy. 2015-12-09 02:45:30 <_mjones> musl is strict with header includes and posix requirements for headers, and it's not always obvious which unless you have some experience or insight. 2015-12-09 02:45:41 Am impressed that Alpine builds LibreOffice on musl 2015-12-09 02:45:58 <_mjones> I recall there are a few patches. 2015-12-09 02:46:15 <_mjones> It helps to look at the patches for existing patches. I don't always remember to do that when I'm stuck. 2015-12-09 02:46:34 <_mjones> openjdk was a pain last I tried. Let me look at where the package is at now. 2015-12-09 02:47:18 <_mjones> Bumped a bit since I tried but not on openjdk8 yet unfortunately. :-/ 2015-12-09 02:48:11 <_mjones> #musl is typically rather active and helpful for porting issues (which are usually involving musl). 2015-12-09 02:48:40 Well I am considering Alpine but need RetroShare a big C++ app needs Qt and SSL galore 2015-12-09 02:49:27 <_mjones> openssl or libressl or gnutls should be no problem at all. 2015-12-09 02:49:42 <_mjones> qt and c++, well. 2015-12-09 02:49:42 And I assume Alpine already has Qt someplace built 2015-12-09 02:49:46 yah 2015-12-09 02:49:59 <_mjones> Let me look. I currently don't run any headed machines. (I should fix that.) 2015-12-09 02:50:39 <_mjones> Yes, plenty of qt. 2015-12-09 02:51:39 <_mjones> Package 'qt' is qt4; the qt5 packages are broken out with the base package being qt5-qtbase 2015-12-09 02:52:28 <_mjones> There are 34 qt5 packages and 75 packages for qt in all (including consuming apps). 2015-12-09 02:53:49 Any ETA on release candidate 2 ? 2015-12-09 02:54:33 <_mjones> Not that I know of, but I expect soonish. Haven't done RCs in the past afaik. 2015-12-09 02:54:57 <_mjones> Many of the main developers are on in the morning and during the day western european time. 2015-12-09 02:55:30 Well ncopa does so many packages he prolly needs a release mgr 2015-12-09 02:55:40 Also Alpine could use buildbot.net 2015-12-09 02:55:47 Automation = efficiency 2015-12-09 03:01:09 Oven just chimed, will go eat, but leave a Q here 2015-12-09 03:01:24 Does grsec kernel (which I LOVE) cause any code porting issues? 2015-12-09 03:05:31 <_mjones> Porting issues no. But some packages need pax markers applied to disable certain features, after the build. 2015-12-09 03:06:12 <_mjones> For instance nodejs needs to be able to execute 'data' because it's an interpreter. This is common to any of the interpreted languages with VMs. 2015-12-09 03:06:28 <_mjones> I don't know how the automation works; been meaning to ask about that. 2015-12-09 03:31:38 http://vlcore.vectorlinux.com/buildbot/one_line_per_build 2015-12-09 03:55:49 Hay every one I was checking the Internet and came across this http://opencatalog.darpa.mil/MEMEX.html My ? is how can I install in in kali ? Is it many packages or is there a place just to install just one big package. 2015-12-09 03:57:03 <_mjones> This isn't Kali. 2015-12-09 03:57:14 <_mjones> There is a lot of different software listed there. 2015-12-09 03:58:23 <_mjones> The normal procedure is to install any dependencies, download the code, make any changes required for musl. Then optionally submit an APKBUILD for the package. 2015-12-09 06:47:46 kids these days 2015-12-09 06:47:54 03:55:41 Also Alpine could use buildbot.net 2015-12-09 06:48:07 he should propose jenkins 2015-12-09 06:48:31 screw our automation that works well, java everywhere 2015-12-09 06:58:21 <_mjones> ha 2015-12-09 08:04:43 _mjones: thanks for the pointer, but setting DISTCC_DIR and I still see distcc[26464] (dcc_mkdir) ERROR: mkdir '/root/.distcc' failed: Permission denied 2015-12-09 08:07:03 DISTCC_DIR isn't passed to the fakeroot, so maybe I need to set it inside the fakeroot :) 2015-12-09 08:08:25 are vars in /etc/abuild.conf used inside the fakeroot? 2015-12-09 09:04:01 how can i install an lxc container with a specific alpine version? 2015-12-09 09:06:25 -- -R 2015-12-09 09:06:27 iirc 2015-12-09 09:18:08 :q 2015-12-09 09:19:15 fcolista: lxc-create -t alpine -n thename -- -R 3.0 2015-12-09 09:19:31 ncopa, thx 2015-12-09 09:20:17 oh 2015-12-09 09:20:22 lxc-alpine has this flag 2015-12-09 09:20:26 stupid me :/ 2015-12-09 09:28:59 ncopa, how can i install all the packages of alpine? apk add $(apk search) doesn't work 'cause of version, and apk search does not support a flag where the version is not showed 2015-12-09 09:29:30 i can hack with sed|awk, but it's not precise because of packages name with dash 2015-12-09 09:32:20 <_mjones> Anybody used CoreOS's rkt (Rocket) containers yet? 2015-12-09 09:32:57 <_mjones> I got the host version hosted on Alpine but the docs are sufficiently poor that I don't know which subsets Alpine would want to ship. 2015-12-09 09:33:37 <_mjones> Certain parts are heavily tied to systemd, but not all of it. 2015-12-09 09:36:54 fcolista: apk search -q 2015-12-09 09:37:04 btw, it will most probably fail 2015-12-09 09:38:18 clandmeter, yes 2015-12-09 09:38:34 fcolista: why do you want to install them all? 2015-12-09 09:38:47 for cvechecker 2015-12-09 09:39:05 it checks CVE entries againsts packages installed 2015-12-09 09:39:12 i've an lxc container where i'm testing it 2015-12-09 09:39:42 you created the cve checker? 2015-12-09 09:39:58 i've done the package, yes 2015-12-09 09:40:03 and maintain it 2015-12-09 09:42:04 fcolista: its https://github.com/sjvermeu/cvechecker? 2015-12-09 09:42:13 yes 2015-12-09 09:42:21 http://wiki.alpinelinux.org/wiki/Cvechecker 2015-12-09 09:43:21 maybe have a chroot to install check and uninstall a package? 2015-12-09 09:44:02 i guess if you install them all apk will complain about overwriting files. 2015-12-09 09:44:30 yes 2015-12-09 09:44:38 that's what i was wondering 2015-12-09 09:44:59 a script that install package, run cvechecker against it, and uninstall it 2015-12-09 09:45:07 but is not very performing 2015-12-09 09:52:39 :) 2015-12-09 10:41:21 i think it would be useful for CVE security fix 2015-12-09 11:04:41 ncopa: so how is your new job? 2015-12-09 11:05:52 przemoc: so far its great! :) 2015-12-09 11:05:54 thanks 2015-12-09 11:06:52 good to hear that 2015-12-09 11:07:27 ncopa: what do you do now? 2015-12-09 11:07:48 you only have to make alpine in its very centre of operation, so you'll have more time to make it even better 2015-12-09 11:08:03 ;) 2015-12-09 11:08:12 if it would be that simple :f 2015-12-09 11:16:36 off-topic question: is anyone aware of some free CIs (extra points for being free for private repos too) that 1) does not require some configuration file in the repo and 2) have decent network connectivity? I've quickly tested drone.io yesterday, but its net is rather slow (or at least downloading 11MB of kernel headers took about several seconds IIRC) 2015-12-09 11:23:16 3) unlimited build time (or absurdly high build time) would be great too 2015-12-09 11:24:24 (drone's free tier allows 15 minutes only) 2015-12-09 13:04:19 hey 2015-12-09 14:18:18 scadu, przemoc: yes, my new work will be much related alpine 2015-12-09 14:18:37 oh, nice. 2015-12-09 14:19:15 ncopa: any details? if you want to tell ofc. 2015-12-09 14:33:04 ncopa: +1, do tell us about your new job 2015-12-09 14:39:10 fcolista: i would a) create a repo mirror specifically for this purpose 2015-12-09 14:40:54 b) a shell script that gunzips and untars each package into a chroot and cve checks the files 2015-12-09 14:41:07 c) cleans up and moves onto the next package 2015-12-09 14:42:09 why a mirror specific for this purpose? 2015-12-09 14:42:42 so that the mirror is not used for other purposes, like for apk add 2015-12-09 14:44:59 perhaps I'd even explore things avfs, which can untar etc transparently, so all you'd have to do is run check cve against an avfs mounted repo 2015-12-09 14:46:03 "explore things *like* avfs", even 2015-12-09 14:47:52 ScrumpyJack, sounds a nice idea actually. 2015-12-09 14:48:50 :) 2015-12-09 14:49:16 what's the avfs support for alpine? 2015-12-09 14:49:39 it's there 2015-12-09 14:49:43 why use a script to untar? 2015-12-09 14:49:51 whats wrong with apk? 2015-12-09 14:50:08 clandmeter: nothing :) 2015-12-09 14:50:21 i think this is what he was wondering 2015-12-09 14:50:30 gunzip+untar = apk 2015-12-09 14:50:40 i forgot apk can chdir of course 2015-12-09 14:50:44 use the power of apk 2015-12-09 14:51:22 you can use apk inside or outside the chroot 2015-12-09 14:51:40 i'd explore avfs first 2015-12-09 14:53:23 mountavfs edge/main cvedir 2015-12-09 14:53:29 cvecheck cvedir 2015-12-09 14:53:34 or something like that 2015-12-09 14:53:43 avfs doc is lame 2015-12-09 14:54:15 what happens when two apk's provide the same file? 2015-12-09 14:54:46 i think cve check should be done per-category 2015-12-09 14:55:18 avfs on alpine does not have init script 2015-12-09 14:55:47 i think avfs adds a path, so you have cvedir/apk1/usr/ and cvedir/apk2/usr ... 2015-12-09 15:00:29 fcolista: what happens if you just run mountavfs ? 2015-12-09 15:02:16 2ua4020qdl:~$ mountavfs 2015-12-09 15:02:16 Mounting AVFS on /home/fcolista/.avfs... 2015-12-09 15:03:32 ok, so your whole / will be remounted in ~/.avfs and you should be able to cd into tars, or gzips etc 2015-12-09 15:05:14 doesnt cvescanner need to the files in its original location? (i have no idea how it works) 2015-12-09 15:05:23 -to 2015-12-09 15:07:58 fcolista: it looks like you'll have to tell avfs how to handle .apk files, but once that's done it should be a doddle 2015-12-09 15:09:22 ScrumpyJack, is not clear to me how avfs works 2015-12-09 15:09:49 it mounts / into ~fcolista/.avfs 2015-12-09 15:10:11 cd if you cd in .avfs , you'll be in / 2015-12-09 15:11:17 and then? 2015-12-09 15:11:30 then cd into a gz file for example 2015-12-09 15:11:39 ah ok 2015-12-09 15:11:39 or a tar file 2015-12-09 15:11:42 it's like a dir 2015-12-09 15:11:43 k 2015-12-09 15:11:47 apk is not handled 2015-12-09 15:12:24 you may need to add a # at the end on the file.gz 2015-12-09 15:12:43 like this cd archive.tar# 2015-12-09 15:16:48 good doc here -> http://delogics.blogspot.nl/2013/01/thiss-virtual-fs-to-browse-through.html 2015-12-09 15:18:50 cd alpine-baselayout-2.3.1-r5.apk#uz#utar 2015-12-09 15:18:54 that does the trick 2015-12-09 15:19:35 awesome. there must be a way to configure it so avfs can handle .apk on it's own 2015-12-09 15:20:51 ok i need to download all apk in main 2015-12-09 15:21:42 that's where the mirror comes in :) 2015-12-09 15:21:59 yes, i got it :) 2015-12-09 15:22:22 yes, sorry, i'm stating the obvious :| 2015-12-09 15:22:29 now i'm doing something very preistoric: apk fetch $(apk search -q) 2015-12-09 15:22:44 i'm gonna download all the apks 2015-12-09 15:23:13 just few, for testing purpose 2015-12-09 15:23:42 /usr/bin/rsync -pruav --delete --timeout=600 --delay-updates --delete-after rsync://rsync.alpinelinux.org/alpine/ ~fcolista/cverepo/ 2015-12-09 15:24:06 then make some coffee :) 2015-12-09 15:24:14 rsync://mirror.nl.leaseweb.net/alpine 2015-12-09 15:24:18 use that one pls 2015-12-09 15:24:30 +1 2015-12-09 15:25:23 clandmeter: sure. no problem, i'll change my sync. Out of interest, any reason? 2015-12-09 15:25:40 i think bw usage 2015-12-09 15:25:48 init sync 2015-12-09 15:25:57 better to use leaseweb 2015-12-09 15:26:36 they have a 4GE connection 2015-12-09 15:26:40 compared to 100mbit here 2015-12-09 15:27:24 I have asked them to sync more often. 2015-12-09 15:27:39 no problem, done. I think i used rsync.alpinelinux.org as I was going to test push updates with ncopa but we never got round to it 2015-12-09 15:28:22 rsync.a.o is 5 meters away from me. 2015-12-09 15:28:25 :) 2015-12-09 15:28:47 *leave that power button alone* ;) 2015-12-09 15:29:31 we might change rsync to leaseweb 2015-12-09 15:29:51 or any other server that sync regular. 2015-12-09 15:30:17 currently everytning is pulled directly from rsync/nl.a.o 2015-12-09 15:30:42 most ppl also use nl.a.o as default apk mirror. 2015-12-09 15:35:57 fcolista: /usr/lib/avfs/extfs/README if you are feeling brave :) 2015-12-09 15:36:58 it looks like the handlers are in /usr/lib/avfs/extfs 2015-12-09 15:37:37 ok here comes 3.3.0_rc2 2015-12-09 15:38:01 \o/ 2015-12-09 15:38:06 ScrumpyJack: i'll tell about new job later 2015-12-09 16:28:32 fcolista: will you let me know if avfs works for cvechecking? 2015-12-09 17:13:12 ncopa: you might want to adjust the topic 2015-12-09 17:41:53 ncopa: can you trigger a rebuild of openexr? 2015-12-09 17:42:04 (in edge) 2015-12-09 18:22:26 ScrumpyJack: I am currently working on an update to 2.2.0 2015-12-09 18:22:29 wait a few minutes 2015-12-09 18:38:31 apk-tools really needs better error handling, I just tells me that an error occured but it doesn't tell me what the error was... 2015-12-09 18:39:53 <_mjones> During what operation? 2015-12-09 18:39:59 apk add 2015-12-09 18:40:16 > 1 errors; 2806 MiB in 556 packages 2015-12-09 18:40:24 <_mjones> I'm making a mental note of that. 2015-12-09 18:40:36 <_mjones> Maybe it shows up with -v -debug? 2015-12-09 18:41:36 apk has a --debug flag? 2015-12-09 18:42:03 --verbose doesn't tell me what the error is 2015-12-09 18:57:44 ScrumpyJack: openexr was just upgraded and should have been rebuild 2015-12-09 20:40:21 nmeum: IexMathFpu.cpp:276:28: fatal error: asm/sigcontext.h: No such file or directory 2015-12-09 20:40:24 on x86 2015-12-09 20:40:52 huh, interessting 2015-12-09 20:41:01 I will look into this 2015-12-09 20:43:29 depending on linux-headers should fix it 2015-12-09 20:47:04 http://dev.alpinelinux.org/buildlogs/build-edge-x86/testing/ilmbase/ilmbase-2.2.0-r1.log 2015-12-09 20:47:28 fixed it, sorry for noticing that too late 2015-12-09 20:56:35 that's okay, just pointing out it failed 2015-12-10 07:51:52 morning alpine guys 2015-12-10 07:52:09 morning. 2015-12-10 07:54:03 hum 2015-12-10 07:54:09 my eth0 and eth1 swapped 2015-12-10 08:14:44 morning 2015-12-10 08:14:53 ncopa: have you tried net.ifnames=0? 2015-12-10 08:16:14 no? 2015-12-10 08:16:39 i assume it is eudev that swapped the ifnames 2015-12-10 08:17:06 it usually is unless you rearranged h/w 2015-12-10 08:17:56 my quad ethenet (sun hme) always gives me troubles, those NIC doesn't even have fixed MAC under Linux 2015-12-10 08:18:49 okay finally got my ttl cable and got my ARM board to boot with mainline Linux, but only well enough with 4.4-rc 2015-12-10 08:24:52 that's all w/o initramfs 2015-12-10 08:25:12 I am missing a lot, does Alpine do fsck in the initramfs? 2015-12-10 08:31:24 no 2015-12-10 08:31:37 openrc does fsck 2015-12-10 08:31:46 rc-update 2015-12-10 08:31:52 you should have fsck there some place 2015-12-10 08:32:52 morning 2015-12-10 08:34:51 ncopa: thanks, found it 2015-12-10 08:34:58 hey ScrumpyJack morning 2015-12-10 08:35:03 I've done with svechecker 2015-12-10 08:35:05 *cvechecker 2015-12-10 08:35:23 xming_: initramfs should mount / as read-only so it can be fscked 2015-12-10 08:35:31 I'm doing few cosmetics 2015-12-10 08:35:51 I pass append ro 2015-12-10 08:36:03 gotta make the script little bit more robust when a cd $DIR#uz#utar doesn't work 2015-12-10 08:44:18 fcolista: can i see the script? 2015-12-10 08:45:56 fcolista: scratch that. there is nothing worse than someone asking to see your stuff and make suggestions. 2015-12-10 08:46:08 no 2015-12-10 08:46:09 http://sprunge.us/jFCi 2015-12-10 08:46:11 that's the script 2015-12-10 08:46:22 I'm not going to look :) 2015-12-10 08:46:25 it's fine, i'm open to suggestion 2015-12-10 08:46:56 there's one part commented out, because i'm running this script inside a container 2015-12-10 08:47:11 and the dir where the apks resides is in a mounted dir in the host 2015-12-10 08:47:19 and in the host avfs is running 2015-12-10 08:47:27 ok 2015-12-10 08:47:37 also rsync is commented since this is a test 2015-12-10 08:47:53 but i think i'll do it in lua 2015-12-10 08:48:44 so this is the thing that fails sometimes? cd $package\#uz\#utar 2015-12-10 08:49:05 if the package is broken 2015-12-10 08:49:21 it stucks on trying to cd 2015-12-10 08:49:52 how does find $package\#uz\#utar behave? 2015-12-10 08:50:14 what do you mean? 2015-12-10 08:50:26 skip the cd 2015-12-10 08:50:35 ah 2015-12-10 08:50:40 right. 2015-12-10 08:50:45 It might work 2015-12-10 08:50:53 so i can remove the check 2015-12-10 08:50:55 find $APKPATH/$package\#uz\#utar -type f -perm -o+x 2015-12-10 08:51:08 good suggestion ScrumpyJack 2015-12-10 08:52:40 let's move this to /query, we're polluting here 2015-12-10 09:17:15 is this something what could be used from alpinelinux http://www.phoronix.com/scan.php?page=news_item&px=BUS1-In-Development 2015-12-10 09:31:32 "made by systemd developers" 2015-12-10 09:31:35 stopped reading there 2015-12-10 09:34:45 lol 2015-12-10 09:35:24 thats kinda interesting though 2015-12-10 09:35:54 kdbus work was initiated due to systemd (I assume) 2015-12-10 09:36:18 not really 2015-12-10 09:36:43 it was initiated because lots of people use D-Bus and found it was slow 2015-12-10 09:36:44 i think dbus people wanted it themselves 2015-12-10 09:36:54 (including the authors themselves, yes) 2015-12-10 09:36:56 now, systemd developers feels its obsolete even before it is included in mainline 2015-12-10 09:37:04 ERROR: unsatisfiable constraints: 2015-12-10 09:37:04 openjdk7-jre-base-7.91.2.6.3-r0: 2015-12-10 09:37:04 conflicts: openjdk7-jre-base-7.91.2.6.3-r0[so:openjdk7:libjvm.so=0] 2015-12-10 09:37:04 openjdk7-jre-base-7.91.2.6.3-r0[so:openjdk7:libjvm.so=0] 2015-12-10 09:37:09 and they thought "it's slow? then we need to put it in the kernel!" 2015-12-10 09:37:23 not even thinking that it may be slow because they suck at coding 2015-12-10 09:37:40 this bus1 thing is just them throwing a tantrum 2015-12-10 09:37:42 systemd needed dbus to work 2015-12-10 09:37:52 "we didn't get our toy included in the kernel? then we're gonna make a new one" 2015-12-10 09:38:04 but since systemd is pid 1, dbus was not running early enough 2015-12-10 09:38:12 that too 2015-12-10 09:38:23 they don't know how to bootstrap either 2015-12-10 09:38:28 "Kay Sievers" 2015-12-10 09:38:34 the primadonna 2015-12-10 09:38:37 :) 2015-12-10 09:39:09 "Compared to KDBUS, BUS1 doesn't seem to be at least heavily based on D-Bus." 2015-12-10 09:39:10 Red Hat should send those guys back to engineering school instead of paying them a salaty 2015-12-10 09:39:16 ncopa, any idea why openjdk7-jre-base provides might get same soname twice? 2015-12-10 09:39:16 and moronix should stop talking about them 2015-12-10 09:39:19 as should we 2015-12-10 09:39:19 sounds like a step in right riection 2015-12-10 09:39:39 salary, not salaty (but I like salaty) 2015-12-10 09:52:41 I don't understand why Greg is push kdbus 2015-12-10 09:53:36 Greg KH is lost to the cause 2015-12-10 09:55:33 IMHO he shouldn't have messed in the troubling water of kdbus 2015-12-10 09:56:29 ncopa exactly seems it doesn't s looks that heavily based on d-bus, but i was just posting some news i read 2015-12-10 09:57:41 my guess is that if this turns out to have a future, it will be eaten up by systemd 2015-12-10 09:57:52 and will not be possible to build outsude systemd 2015-12-10 09:58:04 like they did with Kay's other project, udev 2015-12-10 10:03:20 there's *nothing* in the linked github, just some stubs and the start of a rbtree implementation 2015-12-10 10:03:31 talk about vaporware 2015-12-10 10:03:54 it will be a long time before we have one more shit bus to worry about. 2015-12-10 10:29:30 what's the best way to get a list of the $packagename without -lang and -doc subpackage? 2015-12-10 10:29:34 and -dev 2015-12-10 10:29:55 (besides regexp along wiht ls of course) 2015-12-10 10:36:43 apk search | sed "s/\-lang//; s/\-doc//; s/\-dev//" 2015-12-10 10:45:48 or apk search -q ... | grep -E -v '(-lang|-doc|-dev)$' 2015-12-10 10:48:53 ncopa, it's still a beta: http://sprunge.us/PNED 2015-12-10 10:49:16 cool 2015-12-10 12:35:48 hey :( 2015-12-10 12:41:35 hi 2015-12-10 12:41:43 why the sad face? 2015-12-10 12:43:59 it worked: it got your attention 2015-12-10 12:44:22 hehe, the sad face is because rust still does not compile on alpine ;) hahaha 2015-12-10 12:47:03 i pushed openjdk7 fix 2015-12-10 12:47:16 arm builder will be busy for a couple of days now 2015-12-10 12:54:11 yay for arm apks 2015-12-10 12:59:37 it fails on x86_64 2015-12-10 12:59:42 i think its parallel build problem 2015-12-10 12:59:48 :-( 2015-12-10 13:13:22 to what adress should i send apk patches? 2015-12-10 13:14:46 alpine-aports@lists.alpinelinux.org 2015-12-10 13:18:15 apk? 2015-12-10 13:18:25 i think alpine-devel is better for that 2015-12-10 13:18:35 where do you want apk patches fabled? 2015-12-10 13:18:55 alpine-devel so far is probably best 2015-12-10 13:19:06 if we start to get lot of them, it might be good to setup patchworks for it 2015-12-10 13:19:24 complicated subjects might be good in issue tracker 2015-12-10 13:19:33 but no strict rules there 2015-12-10 13:23:53 i ask because i am subscribed to all alpine lists but never get any mails 2015-12-10 13:24:03 i think my activation status meight be pending 2015-12-10 13:24:20 and since i lost my job today i have a lot of time for packaging and creating patches .. 2015-12-10 13:24:37 and skarnet, thats why i had the sad face 2015-12-10 13:36:18 ncopa started a trend 2015-12-10 13:36:30 i hope you get a new one as fast as him though. 2015-12-10 13:37:16 i hope you get it faster 2015-12-10 13:38:34 i hope so to .. i was never without a job ... feels strange to be now complete useless 2015-12-10 13:48:48 fabled: by the way: is there anyway to get more error information from apk-tools? For instance when a conflict was encountered during apk upgrade you will get an error notice everytime you invoke apk add but apk add doesn't tell you what the error actually is 2015-12-10 13:49:16 > 1 errors; 2806 MiB in 556 packages 2015-12-10 13:53:14 nmeum, mmm. at least "apk fix --simulate" will tell you what it'd try to do it 2015-12-10 13:54:01 but yeah, those package content related errors are not reported nicely 2015-12-10 13:54:06 yeah 2015-12-10 14:58:22 leo-unglaub: unless money's really tight for you, I don't wish you to find a new job very quickly. I wish you to take a step back and a deep breath, a small vacation, and to enjoy your free time and realize your intrinsic worth isn't tied to any job. 2015-12-10 14:59:20 Then, when your self-confidence is mended and you feel like kicking ass again, I wish you to look for and find a new job. You'll be in a great position to start it. 2015-12-10 14:59:41 skarnet: * leo-unglaub has quit (Ping timeout: 272 seconds) 2015-12-10 14:59:42 ... and I'm blabbering to the wind. 2015-12-10 14:59:48 yeah, just realized. 2015-12-10 14:59:48 :) 2015-12-10 15:00:22 Someone copypaste that when he comes back! I hate to see my lyrical efforts go to waste. :P 2015-12-10 15:00:32 :D 2015-12-10 15:02:37 as you probably have noticed, getting lyrical is a way to keep my sanity when I'm dealing with crap such as gcc builds :) 2015-12-10 15:02:53 indeed :) 2015-12-10 15:06:32 leo-unglaub: wb 2015-12-10 15:06:39 you missed the best part. 2015-12-10 15:07:31 Now who is the *real* leo-unglaub? 2015-12-10 15:07:35 Please stand up. 2015-12-10 15:07:46 lol 2015-12-10 15:08:42 ... they were both fake 2015-12-10 15:14:36 re 2015-12-10 15:14:43 wb 2015-12-10 15:20:53 wb :) 2015-12-10 15:20:59 leo-unglaub: did you get my private paste? 2015-12-10 15:21:20 no, i did not 2015-12-10 15:26:51 you got it now 2015-12-10 15:26:56 skarnet got lyrical :) 2015-12-10 15:29:07 yeah, very smart words 2015-12-10 15:29:13 thanks skarnet 2015-12-10 15:29:57 ncopa: could you please apply this: http://dpaste.com/1PDS1G7.txt 2015-12-10 15:29:59 thanks :) 2015-12-10 15:30:05 it builds fine here 2015-12-10 15:30:25 this does not mean that hiawatha is still broken, but its broken on a newer version now *g* 2015-12-10 15:30:33 the patch to fix it in mbedtls is comming later 2015-12-10 15:33:04 leo-unglaub: pushed. thanks 2015-12-10 15:34:02 thanks. now comes the funny part :) watching #alpine-commits for th build server to finish 2015-12-10 15:34:04 hehe 2015-12-10 15:37:41 new kernel is done, brb reboot 2015-12-10 16:29:29 Hello alpine ..! what is this new buz-word .. "recursive static route" ?? how we do it in alpine linux ?? 2015-12-10 17:47:16 any ideas why "rcs" will build on void musl http://paste.pound-python.org/show/y5MWJYtvXEVVnzxq66v4/ but not on Alpine http://hastebin.com/etunafowir.vbs 2015-12-10 17:55:45 BitL0G1c: void has a musl version?? 2015-12-10 18:11:43 yes 2015-12-10 20:26:35 ncopa: did you share whom are you working for/what are you doing now? I didn't pay much attention to backlog in this and previous week 2015-12-11 07:34:06 Hello 2015-12-11 07:34:22 If I want to update a package that got into unmaintained 2015-12-11 07:34:37 should I provide a patch putting it to community ? 2015-12-11 07:36:30 I guess it doesn't make much sense to add it to testing 2015-12-11 07:37:08 coredumb: if it builds then feel free to move it back to testing 2015-12-11 07:38:16 ncopa: ok then 2015-12-11 07:38:32 well I wouldn't push it if it wouldn't build :D 2015-12-11 07:38:58 we normally move things to unmaintained if things does not build and we dont want spend time on it 2015-12-11 07:39:07 oh ok 2015-12-11 07:39:37 the version is pretty old and there's much to change to update the package 2015-12-11 07:39:46 cfengine that is 2015-12-11 07:40:06 oh 2015-12-11 07:40:10 you are working on cfengine 2015-12-11 07:40:19 that was my plan to update to 3.7 2015-12-11 07:40:36 http://tpaste.us/AvNB 2015-12-11 07:40:40 I see that at least a part of the patch has been pushed mainstream 2015-12-11 07:40:46 i started on it some time ago 2015-12-11 07:40:49 and reported upstream 2015-12-11 07:40:59 please fell free to test build it 2015-12-11 07:41:08 and if it builds as is then i'll move it ot testing for you 2015-12-11 07:41:08 oh brilliant 2015-12-11 07:41:23 i think it needs init.d scripts 2015-12-11 07:41:33 yep 2015-12-11 07:41:39 would be nice to include masterfiles 2015-12-11 07:41:54 and tokyocabinet is not recommended anymore 2015-12-11 07:41:58 ok 2015-12-11 07:42:05 lmdb is prefered 2015-12-11 07:42:16 the provided apkbuild is a good starting point anyway 2015-12-11 07:42:28 let me start from yours and send you the modifications 2015-12-11 07:42:34 perfect 2015-12-11 07:42:34 thanks 2015-12-11 07:43:24 barthalion: no i didnt share yet 2015-12-11 07:44:46 oh we don't have a lmdb package ... 2015-12-11 07:45:50 ok gonna add the package then :) 2015-12-11 07:46:51 its from samba iirc 2015-12-11 07:47:17 you can probably use the tdb APKBUILD as base 2015-12-11 07:47:31 yep 2015-12-11 07:48:28 apparently lmdb is from openldap 2015-12-11 07:49:05 there was some huge leak issues with tokyo cabinet in the past 2015-12-11 07:49:18 that's why cfengine moved to lmdb 2015-12-11 08:09:01 good morning climbers, and Happy Friday! 2015-12-11 09:01:37 what's with pulseaudio? 2015-12-11 09:03:33 morning 2015-12-11 09:13:23 can it be that a lot of patches here have not been merged? how to you keep track of what got merged and what did not? 2015-12-11 09:13:29 http://lists.alpinelinux.org/alpine-aports/ 2015-12-11 09:14:01 leo-unglaub: http://patchwork.alpinelinux.org/project/aports/list/ 2015-12-11 09:14:16 ah, now thats cool 2015-12-11 09:14:23 thanks clandmeter 2015-12-11 09:14:30 yw 2015-12-11 10:14:44 ncopa: could you take a look at %1120 and %1119 ? 2015-12-11 10:14:59 *%1200 2015-12-11 10:15:55 andyshinn: re su and securetty, if /etc/securetty is missing, all ttys are accepted 2015-12-11 10:16:18 i wonder if we should provide a default securetty, either with busybox or with bbsuid 2015-12-11 10:16:53 nmeum: %1120 seems to be applied 2015-12-11 10:18:10 nmeum: re %1199 dont we need a root-hints in default config? 2015-12-11 10:18:25 maybe we can install it as a heredoc? 2015-12-11 10:18:48 eg cat >$pkgdir/etc/unbound/configuration < server: 2015-12-11 10:18:51 .... 2015-12-11 10:18:53 EOF 2015-12-11 10:19:21 as microoptimisation. fewer files in git -> faster 2015-12-11 10:23:47 I think that's why the centos repos are split by packages 2015-12-11 10:24:17 looks crazy at first but then when you have a lot of files + lot of commits it starts to make sense 2015-12-11 10:46:24 ncopa: sent new lmdb and updated cfengine patches 2015-12-11 10:46:42 let me know if I messed up something 2015-12-11 10:50:58 dont worry. if you messed up, I'll let you know :) 2015-12-11 10:51:56 :D 2015-12-11 10:52:35 now that makes me wonder what the Alpine population is using as configuration management tool ? 2015-12-11 11:11:24 any ideas why "rcs" will build on void musl http://paste.pound-python.org/show/y5MWJYtvXEVVnzxq66v4/ but not on Alpine http://hastebin.com/etunafowir.vbs ?? 2015-12-11 11:16:20 ncopa: /etc/unbound/root.hints seems to be the defautl value for root-hints, but you are right it would be a good idea to set it explicitly 2015-12-11 11:16:32 will update it later 2015-12-11 11:39:04 hmm, groupadd is not playing ball for me 2015-12-11 11:39:05 kitten:~# apk info -L --who-own /usr/bin/groupadd 2015-12-11 11:39:05 ERROR: /usr/bin/groupadd: Could not find owner package 2015-12-11 11:39:52 sorry, ignore the who-owns, it belongs to shadow 2015-12-11 11:41:03 anyway, why does shadow groupadd prompt for Password: 2015-12-11 11:43:56 it's called addgroup 2015-12-11 11:45:24 that's the busybox modules. virtualbox needs groupadd, which is in the shadow package 2015-12-11 11:58:02 ah, alright 2015-12-11 12:31:03 bah 2015-12-11 12:31:07 inkscape crashes 2015-12-11 12:35:11 coredumb: lmdb: installs man pages in /usr/man 2015-12-11 12:35:16 should be /usr/share/man 2015-12-11 12:35:21 ooops 2015-12-11 12:35:27 see I messed up something :P 2015-12-11 12:35:34 + we should have $pkgname-doc subpackage 2015-12-11 12:35:45 i am a bit worried about this too: 2015-12-11 12:35:49 pkg/lmdb/usr/lib/liblmdb.so 2015-12-11 12:35:56 the library is not verioned 2015-12-11 12:36:03 what happens if ABI breaks? 2015-12-11 12:36:10 that is an upsteram problem though 2015-12-11 12:36:29 yeah ... 2015-12-11 12:36:33 we could also consider adding a lmdb-tools subpackage 2015-12-11 12:36:39 with the usr/bin 2015-12-11 12:36:47 for the bins only 2015-12-11 12:36:49 why not 2015-12-11 12:37:00 other than that, good 2015-12-11 12:37:02 we could also version the lib ourselves ? 2015-12-11 12:37:11 we could but thats risky too 2015-12-11 12:37:17 i think i'll push as is for now 2015-12-11 12:37:26 and you can provide incremental fix 2015-12-11 12:37:29 ok I'll send updates then 2015-12-11 12:37:45 thanks 2015-12-11 12:39:29 I had to battle a bit with the cfengine-masterfiles subpackage :P 2015-12-11 12:42:28 ncopa: could you please merge my patch from last night, i would need it in edge on a server :) 2015-12-11 12:42:31 thanks sooo much! 2015-12-11 12:44:32 http://patchwork.alpinelinux.org/patch/1203/ 2015-12-11 12:54:52 hum 2015-12-11 12:54:55 patch is corrupt 2015-12-11 13:04:43 ncopa: we agree then that lmdb package would only contain the library right ? 2015-12-11 13:04:47 what does anyone make of this: 2015-12-11 13:04:53 /opt/VirtualBox/VirtualBox: setuid, setgid ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked 2015-12-11 13:04:56 coredumb: yes 2015-12-11 13:04:59 ok 2015-12-11 13:05:00 ldd: /opt/VirtualBox/VirtualBox: Not a valid dynamic program 2015-12-11 13:05:35 ScrumpyJack: readelf -d /opt/VirtualBox/VirtualBox 2015-12-11 13:05:43 its a precoplied binary 2015-12-11 13:07:07 http://sprunge.us/JfNN 2015-12-11 13:07:22 it's dyn linked 2015-12-11 13:20:26 ncopa: any reason for the -doc subpkg to correctly get man pages and compress them, but not remove them from pkgname ? 2015-12-11 13:21:34 ok my bad 2015-12-11 13:21:37 works as expected 2015-12-11 13:29:14 ScrumpyJack: [libc.so.6] 2015-12-11 13:29:22 gnu libc 2015-12-11 13:30:37 ncopa: sent update for lmdb 2015-12-11 13:40:11 yeah. I made a fool of myself in #vbox I thought virtualbox was closed source 2015-12-11 14:07:53 :) 2015-12-11 14:51:37 should we ship /etc/securetty with busybox-suid or with busybox itself? 2015-12-11 14:51:53 or have a separate package for it to avoid conflict with pam 2015-12-11 14:53:14 I guess a separate package would be easier ? 2015-12-11 14:57:49 I think bb-suid would be incorrect or does bb login not read it? 2015-12-11 15:02:56 we used to have a bbsuid helper package 2015-12-11 15:03:38 i think it got renamed to busybox-suid 2015-12-11 15:03:56 busybox-suid-1.24.1-r5 contains: 2015-12-11 15:03:56 bin/bbsuid 2015-12-11 15:04:05 bbuisd used to be shipped with busybox itself 2015-12-11 15:04:11 it is su and login that reads it 2015-12-11 15:04:15 oh 2015-12-11 15:04:22 coreutils login? 2015-12-11 15:04:33 what pkg provides gnu login? 2015-12-11 15:05:12 shadow 2015-12-11 15:05:38 and shadow does not seem to install any /etc/securetty 2015-12-11 15:06:21 i was thinking shipping it with busybox-suid because it is not useful without suid 2015-12-11 15:06:24 but i think thats wrong 2015-12-11 15:06:33 login can use it, without suid root 2015-12-11 15:06:44 so maybe better ship it with busybox itself? 2015-12-11 15:07:21 seems like no pakcage ships any securetty 2015-12-11 15:07:24 https://pkgs.alpinelinux.org/contents?filename=securetty&path=&pkgname=&repo=all&arch=x86_64 2015-12-11 15:38:28 ncopa: imo busybox would be the right place, I don't see any sense in creating a new configuration package just yet 2015-12-11 16:21:04 what is the purpose of the -suid package? 2015-12-11 16:21:13 it is anyway required by alpine-base 2015-12-11 16:28:03 to be able to install busybox without having to have and suid binary on your system 2015-12-12 18:03:46 good evening friends :) 2015-12-12 18:33:45 good evening 2015-12-12 19:34:35 Good evening. 2015-12-12 19:34:56 Is there anyone here that can answer some of my curiousities regarding the workings of the Live CD? 2015-12-12 19:42:35 I wish there were an Alpine Linux IRC channel that anyone actually used. 2015-12-12 20:15:04 mswan: this channel is used a lot 2015-12-12 20:15:12 just ask your question 2015-12-12 20:15:28 and if someone reads the backlog and knows the answer he will answer 2015-12-14 08:34:59 morning 2015-12-14 08:48:47 <_mjones> Evening. 2015-12-14 09:12:55 hey 2015-12-14 09:43:56 morning 2015-12-14 09:59:47 _mjones: what TZ are you in? 2015-12-14 11:54:39 someone had success with adobe flash x64 with alpine? 2015-12-14 11:55:04 Looks that firefox was not able to read the plugin 2015-12-14 11:55:24 ldd libflashplayer.so showed that he was looking for ld-linux-x86-64.so.2 2015-12-14 11:55:29 which does not exists 2015-12-14 11:55:37 so i : 2015-12-14 11:55:38 ln -s /lib/ld-musl-x86_64.so.1 /lib/ld-linux-x86-64.so.2 2015-12-14 11:55:51 firefox now is happy 2015-12-14 11:55:56 and flash works 2015-12-14 11:55:58 did you try the compat apk? 2015-12-14 11:56:04 nope 2015-12-14 11:56:15 libc-compat or something like that. 2015-12-14 11:56:38 there's a libc-compat flash player package? 2015-12-14 11:56:47 no 2015-12-14 11:56:48 I just took the archlinux ones 2015-12-14 11:56:58 the link from PKGBUILD 2015-12-14 11:57:03 there is a musl subpkg that provides it 2015-12-14 11:57:46 ah 2015-12-14 11:57:47 libc6-compat 2015-12-14 11:58:10 with this package there's no need to make the symlink by hand then 2015-12-14 11:58:22 http://git.alpinelinux.org/cgit/aports/commit/main/musl?id=0ee848857083de56a36285494071e75a3d557139 2015-12-14 11:59:03 I totally miss that 2015-12-14 11:59:03 it was added when were trying to get flash working. 2015-12-14 11:59:18 I never used flash :) 2015-12-14 11:59:28 ok 2015-12-14 11:59:31 good to know 2015-12-14 11:59:33 thx clandmeter 2015-12-14 12:01:59 what's the way to make flash work in chromium ? 2015-12-14 12:03:21 https://blog.chromium.org/2014/11/the-final-countdown-for-npapi.html 2015-12-14 12:20:06 fcolista: Yeah I meant on alpine :D 2015-12-14 12:20:44 right now I haven't find a working way 2015-12-14 12:21:56 afaik chrome embeds a flash plugin but so far it's not running on chromium 2015-12-14 12:22:06 alpine's chromium that is 2015-12-14 12:23:39 coredumb: for the rest chrome is working ok? 2015-12-14 12:23:44 i havent tried it anymore. 2015-12-14 12:24:18 clandmeter: it's supposed to, I couldn't make chrome work on alpine 2015-12-14 12:24:51 coredumb: not at all or? 2015-12-14 12:25:11 but I checked the way debian uses the flash plugin - pepper something - on chromium 2015-12-14 12:25:18 clandmeter: not at all 2015-12-14 12:25:34 there's missing symbols in loads of gnome deps 2015-12-14 12:25:39 or the like 2015-12-14 12:25:43 ? 2015-12-14 12:25:46 for chrome? 2015-12-14 12:25:49 yeah 2015-12-14 12:25:55 the RPM version of it 2015-12-14 12:26:04 huh? 2015-12-14 12:26:09 i mean chrome on alpine 2015-12-14 12:26:19 without flash 2015-12-14 12:26:19 yeah same 2015-12-14 12:26:25 you mean chromium ? 2015-12-14 12:26:34 oh yes sorry 2015-12-14 12:26:36 chromium 2015-12-14 12:26:38 yeah 2015-12-14 12:26:48 ofc chrome doesnt work :) 2015-12-14 12:27:06 tried chromium with the pepper thing extracted from official chrome 2015-12-14 12:27:14 but chromium wouldn't see anything :( 2015-12-14 12:27:23 ok 2015-12-14 12:27:27 not sure if it's alpine's compilation flags or something 2015-12-14 12:27:40 haven't dug deeper 2015-12-14 12:27:44 but for the rest its still working ok? 2015-12-14 12:27:48 sure 2015-12-14 12:28:00 i had some issues when using font hinting 2015-12-14 12:28:06 but i havent used chromium for months 2015-12-14 12:28:14 i think i should also upgrade it. 2015-12-14 12:28:32 but atleast somebody is using it... 2015-12-14 12:28:54 is flash still used that much? 2015-12-14 12:29:18 on windows it just works. 2015-12-14 12:34:48 I think flash got name changed 2015-12-14 12:35:57 "Animate CC" 2015-12-14 12:52:15 do we have somewhere the perl-cryptx (Freenode SASL auth - http://freenode.net/sasl/sasl-irssi.shtml ) 2015-12-14 12:52:36 i searched in pkgs.a.o but i am not able to find it 2015-12-14 12:54:08 good thing i use alpine linux :) https://twitter.com/RichFelker/status/675360594346819584 2015-12-14 13:11:58 >_< 2015-12-14 16:27:40 ncopa - I also built perl-par-packer with git-email removed from my build environment to be sure the APKBUILD was not missing any depends 2015-12-15 08:17:30 morning folks 2015-12-15 08:36:16 morning climber 2015-12-15 08:37:06 morning alpiners 2015-12-15 08:39:36 fcolista: how is cvechecker coming on? 2015-12-15 08:39:53 ScrumpyJack, good 2015-12-15 08:40:06 but is not very performing 2015-12-15 08:40:12 oh dear 2015-12-15 08:40:43 so you have a local mirror of the repos now? 2015-12-15 08:40:49 yes 2015-12-15 08:41:08 and you're running a find in aufs? 2015-12-15 08:41:10 to check one version tree it takes ~7 hrs 2015-12-15 08:41:24 avfs 2015-12-15 08:41:25 yes 2015-12-15 08:41:41 how big is a version tree? 2015-12-15 08:42:04 ~4800 packages 2015-12-15 08:42:35 v3.3 is 19G 2015-12-15 08:43:05 have you got a better understanding of the output? 2015-12-15 08:43:27 what do you mean? 2015-12-15 08:44:05 well, we were scratching our heads with the cvechecker 2015-12-15 09:09:04 fcolista: i guess you only need to scan updated/new packages? 2015-12-15 11:01:37 Is a 3.3.0RC2 blurb forthcoming on the website or mailing lists 2015-12-15 11:07:33 Been looking for it but only saw it in the chatroom topic bar, downloading now, thank you ncopa 2015-12-15 11:07:57 Alpine is very chatcentric put more stuff on the website to let people know it's very ALIVE 2015-12-15 11:08:02 Over and out 2015-12-15 11:09:08 PS GnuPG sign the RCs as well as main stable releases 2015-12-15 11:34:34 http://imapsync.lamiral.info/ - is this open source %1228 ? 2015-12-15 11:34:34 also could not find if this is official site for https://github.com/imapsync 2015-12-15 11:36:57 alternate, http://isync.sourceforge.net/ 2015-12-15 12:51:59 are there any more issues with setup-alpine? 2015-12-15 12:52:14 i have fixed setup-disk to not show ramX 2015-12-15 12:52:29 and i have fixed the detection of fastest mirror 2015-12-15 12:52:33 it should be much faster now 2015-12-15 12:54:35 anything else that needs fixing? 2015-12-15 12:56:43 kaniini: what was the reason that we dont upgrade libbsd? #4613 2015-12-15 12:57:10 oh, alpine mail avalanche coming into my mail ;) 2015-12-15 12:57:48 oh, rc2 is out, nice 2015-12-15 12:58:04 yeah, i should have announced those on web page 2015-12-15 12:58:08 and on ml 2015-12-15 12:58:25 i think rc3 may be out today 2015-12-15 13:02:42 not too fast? people won't manage to report problems from rc2. 2015-12-15 13:08:04 assuming I would install rc2 (in vm for now), what's the update path toward further rc releases and final one? is it described somewhere, maybe wiki? 2015-12-15 13:12:28 http://nl.alpinelinux.org/alpine/v3.3/ has empty armhf, x86, x86_64 directories. they shouldn't be here right? 2015-12-15 13:13:52 so whether I'm on rc or not it doesn't matter, because I'll simply have path to v3.3 repo in config 2015-12-15 13:18:52 correct 2015-12-15 13:19:04 i have fixed the setup scripts 2015-12-15 13:19:26 oh, i have one more issue to fix 2015-12-15 13:26:26 ncopa: I need some abuild advice. What licence shall I put in an APKBUILD file for this http://tpaste.us/Avoo 2015-12-15 13:26:30 for now i just put other 2015-12-15 13:28:16 please google it 2015-12-15 13:28:31 ok 2015-12-15 13:29:49 seems a bit similar to MIT, but reworded and shorter 2015-12-15 13:30:02 google points me towards MIT, but it's not word for word. 2015-12-15 13:31:04 i don't know much about MIT licence. there seems to be a general idea, worded as you wish. would that be correct? 2015-12-15 13:38:13 license in APKBUILD takes only specific values or can have anything? 2015-12-15 13:38:29 I would call it MIT-like ;) 2015-12-15 13:39:51 otherwise Other sohuld be fine 2015-12-15 13:39:54 there is no check on license 2015-12-15 13:41:18 przemoc, you can also set it to CUSTOM and then copy the license file in package() 2015-12-15 13:41:23 btw I have sometimes wondered what the license the aports repo itself carries 2015-12-15 13:41:50 if its up to me then it'd be MIT 2015-12-15 13:41:52 we discsussed that recently 2015-12-15 13:41:58 I think it is not mentioned anywhere 2015-12-15 13:42:17 i think we mention it somwhere, just dont remember where. 2015-12-15 13:42:32 BTW packages in Alpine Linux distinguish between MIT (aka Expat) and MIT/X (aka X11) licenses? the latter has non-endorsement clause. 2015-12-15 13:42:51 wouldn't it be good to use GPL3 to encourage corporate use? :) 2015-12-15 13:43:17 Personally I prefer the one without non-endorsement clause, as it's limit freedom of the user. 2015-12-15 13:44:37 s/it's limit/the clause limits/ 2015-12-15 13:44:54 My view about GPL changed in last 10 years or so. GPL is in fact restricting freedom and who is going to abuse it will abuse it anyway. Who has sharing mindset will share changes/improvements even if she or he is not forced to do so by the license. MIT is much simpler and also eases reusing your code in your workplace. 2015-12-15 13:44:59 no strong opinion there 2015-12-15 13:45:19 perhaps we should put LICENSE file to aports 2015-12-15 13:45:59 for some time I started calling files LICENSE.NAME, e.g. LICENSE.MIT, so people can now at first glance what it is (and verify later just in case) 2015-12-15 13:46:09 s/now/know/ sorry 2015-12-15 13:47:03 MIT and BSD can be confusing, atleast thats my experiance. 2015-12-15 13:47:49 Licences can be confusing indeed, except perhaps the WTFPL 2015-12-15 13:49:15 surely some lawyer would find loopholes also in WTFPL :) 2015-12-15 13:49:25 :) 2015-12-15 13:50:52 if a license is not clear at first glance, there is seriously something wrong with it. (or maybe i watches too many episodes of suits) 2015-12-15 13:52:37 some more complex licenses have actually useful clauses w.r.t. contributions 2015-12-15 13:52:52 e.g. ASL2.0 5§ 2015-12-15 13:53:31 means you don't need to manage contributor license agreements 2015-12-15 13:53:54 perhaps we should write an Alpine Licence 2015-12-15 13:54:03 please no 2015-12-15 13:54:25 like the wtfpl, but with the cursing removed :) 2015-12-15 13:55:54 echo "Public Domain" > LICENSE 2015-12-15 13:57:00 or CC0, which is a bit longer 2015-12-15 13:59:10 yes that's even better due to explicit liability and IPR disclaimer 2015-12-15 14:22:21 +1 for LICENSE file in aports, as we also have official aports here https://github.com/alpinelinux/aports 2015-12-15 14:24:59 I strongly suggest to name it LICENSE.MIT 2015-12-15 14:33:06 what's the difference between 3.3.0_rc2 and edge-3.3.0_rc2 ISOs? 2015-12-15 14:33:52 http://git.alpinelinux.org/cgit/aports/log/?ofs=50 2015-12-15 14:34:21 rc1 is here: http://git.alpinelinux.org/cgit/aports/log/?ofs=300 2015-12-15 14:34:33 and how often edge ISO is built? 2015-12-15 14:34:41 daily (nightly)? 2015-12-15 14:34:51 whenever i tag 2015-12-15 14:35:21 i figured that ther is no point in doing nightly as normally the stuff included in iso has not changed 2015-12-15 14:37:14 ncurses problem has been fixed already? 2015-12-15 14:38:30 there were some segfaults IIRC in e.g. sngrep (pkg that fcolista added upon my request in redmine), but reportedly it was not sngrep only thing 2015-12-15 14:39:20 ok? url? 2015-12-15 14:39:26 there was an issue with alsamixer too 2015-12-15 14:39:57 don't have any url, heard about issues on the channel only 2015-12-15 14:40:17 difficult to fix then 2015-12-15 14:41:20 fcolista: you remember the problem maybe? 2015-12-15 14:44:01 afaik, somekind of abi issue. 2015-12-15 14:44:42 przemoc, it's a ncurses issue 2015-12-15 14:46:05 ncopa: you're a bit active, so let me ask/request one redmine thing. recently I wanted to (but couldn't) mark some issues as duplicated. supposedly reporter role lacks some permissions. could you fix it? 2015-12-15 14:49:55 i changed permissions for role "reporter" 2015-12-15 14:50:05 to have the perms to modify issue relations 2015-12-15 14:54:10 thanks 2015-12-15 14:54:41 hopefully there won't be any abusers 2015-12-15 15:42:51 what are nice switches to use when creating a diff patch for packages? 2015-12-15 15:53:41 as in git? 2015-12-15 15:57:36 ScrumpyJack maybe near the -u option 2015-12-15 16:01:27 I'm using -buB, but perhaps there is better. I keep catching things that haven't changed 2015-12-15 16:07:13 grr, chunks failing 2015-12-15 16:11:27 (this is busybox diff) 2015-12-15 20:00:59 so I'm continuing my lone journey of recategorizing package requests as package requests, providing current version information when possible 2015-12-15 20:26:30 ncopa: http://tpaste.us/2Kxo could you have a look at this if its ok ? 2015-12-15 21:13:57 fcolista: http://bugs.alpinelinux.org/issues/4443#note-2 2015-12-15 21:14:35 sometimes I'm speechless at people's creativity... 2015-12-15 21:48:24 damn cgit, it lacks file search 2015-12-15 22:17:40 fcolista: thanks for xml-starlet and html-xml-utils 2015-12-16 08:13:22 good morning 2015-12-16 08:19:05 morning climbers 2015-12-16 08:19:23 looking forward to the holidays? 2015-12-16 08:21:19 good morning ScrumpyJack 2015-12-16 08:24:27 yes and no, i need to watch my weight... 2015-12-16 08:30:46 a package i'm building requires qt5serialport and we see to have qtserialport-0_git20130311-r0 2015-12-16 08:30:53 that's qt4 only right? 2015-12-16 10:56:15 that's interesting. 2015-12-16 10:56:23 http://nl.alpinelinux.org/alpine/edge/testing/x86_64/ contains py-characteristics 2015-12-16 10:56:27 sorry 2015-12-16 10:56:33 py-characteristic 2015-12-16 10:56:43 in my buildenv i've that very repo 2015-12-16 10:56:56 but apk search doesn't find py-characteristic 2015-12-16 10:57:02 i did apk update 2015-12-16 10:57:06 and repos are correct 2015-12-16 10:57:16 x86_64-edge:~# cat /etc/apk/repositories 2015-12-16 10:57:16 #/media/usb/apks 2015-12-16 10:57:16 http://nl.alpinelinux.org/alpine/edge/community 2015-12-16 10:57:16 #http://dl-3.alpinelinux.org/alpine/v3.2/main 2015-12-16 10:57:16 http://nl.alpinelinux.org/alpine/edge/main 2015-12-16 10:57:17 http://nl.alpinelinux.org/alpine/edge/testing 2015-12-16 10:57:25 apk update 2015-12-16 10:57:38 x86_64-edge:~# apk search py-char 2015-12-16 10:57:38 py-chardet-2.3.0-r0 2015-12-16 11:03:25 perhaps try apk update --force 2015-12-16 11:04:40 nah 2015-12-16 11:04:50 i've deleted also the APKINDEX in /var/cache/apk 2015-12-16 11:07:24 i don't see py-characteristic on my mirror 2015-12-16 11:09:16 ok, it's from today 2015-12-16 11:10:59 hmm, rsync to leaseweb is slow today 2015-12-16 11:27:35 mirror rsynced, and i can see py-characteristic-14.3.0-r0 2015-12-16 11:44:25 build server had issues 2015-12-16 11:44:27 i broke gcc 2015-12-16 11:44:44 should be ok now 2015-12-16 12:16:00 bah...looks that build.a.o -> nl.a.o takes some time 2015-12-16 12:16:02 packages built i mean 2015-12-16 12:16:39 that's why apk search charact didn't return py-characteristic 2015-12-16 13:59:21 so, how can i get qt serialport for qt5? 2015-12-16 14:25:11 we figured out the (stupid) mistake why our arm stuff did not work on qemu 2015-12-16 14:25:17 or probably anything else than wandboard 2015-12-16 14:25:29 i forgot the early serial console stuff on 2015-12-16 14:25:42 will push soon new kernels 2015-12-16 14:25:56 qemu with vexpress profile worked for me on that image 2015-12-17 07:50:59 edge/armhf builder is off line for a while 2015-12-17 07:51:05 i'm migrating the disk to SSD 2015-12-17 07:57:35 cool 2015-12-17 07:58:14 awesome! 2015-12-17 07:58:21 so 2015-12-17 07:58:29 3.3.0 rc3 is out 2015-12-17 07:58:35 i taged it yesterday 2015-12-17 07:58:38 tagged* 2015-12-17 07:58:56 if there are things that you think *must* be fixed before 3.3.0, now is the time to let me know 2015-12-17 08:12:38 ncopa was the ncurses issue fixed? 2015-12-17 08:12:46 no 2015-12-17 08:12:57 neither was inkscape 2015-12-17 08:13:18 ncopa: if I make a bugfix release for s6 this morning and a new APKBUILD, is it possible to integrate the new version? (It won't break anything.) 2015-12-17 08:13:22 we want it fixed before 3.3 i suppose 2015-12-17 08:14:13 skarnet: yes 2015-12-17 08:14:24 fcolista: yes we want fix ncurses too 2015-12-17 08:14:27 thanks. on it. 2015-12-17 08:14:32 k 2015-12-17 08:20:01 Policy question. There's no /libexec in FHS. Programs use /usr/libexec; when they need to be on the rootfs, they sometimes use /lib/$foobar (which is ugly). Should I move unexported programs to /lib/s6, or is it ok to keep /libexec (as it is now) ? 2015-12-17 08:20:49 i prefer /lib/$foobar 2015-12-17 08:20:54 ok. 2015-12-17 08:22:06 we have /lib/{apk,mdev,rc,udev,device-mapper} 2015-12-17 08:22:28 hm looks like we even have a /lib/systemd. I wonder what that is for :) 2015-12-17 08:22:50 from avahi apparently 2015-12-17 08:22:51 morning 2015-12-17 08:23:01 ncopa, which programs segfault because of ncurses, beside alsamized? 2015-12-17 08:23:08 *alsamixer 2015-12-17 08:23:20 i only know about alsamixer 2015-12-17 08:23:31 umh 2015-12-17 08:23:33 inkscape also segfaults at startup 2015-12-17 08:24:21 alsamixer works for me now 2015-12-17 08:24:27 it does? 2015-12-17 08:24:33 not for me 2015-12-17 08:24:38 Segmentation fault (core dumped) 2015-12-17 08:24:38 ncopa-desktop:~$ 2015-12-17 08:24:59 i'm on edge x86_64 2015-12-17 08:26:07 http://st.ilet.to/alsa.png 2015-12-17 08:26:19 Im on edge 64 too 2015-12-17 08:27:37 ScrumpyJack, are you on vanilla? 2015-12-17 08:27:42 grsec 2015-12-17 08:27:54 this is interesting 2015-12-17 08:28:00 4.1.13-0-grsec 2015-12-17 08:28:16 what do you have in $LANG 2015-12-17 08:28:18 echo $LANG 2015-12-17 08:28:24 and echo $LC_ALL 2015-12-17 08:28:41 ACTION has: alpinetest:~# echo $LANG 2015-12-17 08:28:41 en.utf8 2015-12-17 08:28:41 alpinetest:~# echo $LC_ALL 2015-12-17 08:28:43 http://tpaste.us/Aj1v 2015-12-17 08:29:02 did you build it yourself? 2015-12-17 08:29:02 LANG=en_GB.utf8 2015-12-17 08:29:06 nope 2015-12-17 08:29:21 do you use edge or v3.3 repo? 2015-12-17 08:29:26 apk policy alsamixer 2015-12-17 08:29:34 edge 2015-12-17 08:29:43 i think its alsa-utils 2015-12-17 08:30:02 ncopa-desktop:~$ apk policy alsa-utils 2015-12-17 08:30:02 alsa-utils policy: 2015-12-17 08:30:02 1.1.0-r1: 2015-12-17 08:30:02 lib/apk/db/installed 2015-12-17 08:30:02 http://rsync.alpinelinux.org/alpine/edge/main 2015-12-17 08:30:06 /usr/bin/alsamixer is owned by alsa-utils-1.1.0-r1 2015-12-17 08:30:14 yes 2015-12-17 08:30:56 ScrumpyJack: what is $TERM 2015-12-17 08:32:17 i usually set it to screen-256color 2015-12-17 08:32:38 or linux or vt100 depending on the app's support 2015-12-17 08:33:36 ok, that didnt make any diff here 2015-12-17 08:34:14 soo happy it works again, i use it all the time 2015-12-17 08:38:22 oh no! i restarted X (for other reasons) and now alsamixer seg faults! 2015-12-17 08:44:31 ok, i can get it to work, but this is very wierd 2015-12-17 08:44:38 it is weird indeed 2015-12-17 08:45:56 if i open a terminal window in x (mate-terminal) with the default size (80x24) and run alsamixer, it seg faults 2015-12-17 08:47:39 if I "maximise" the terminal window with my window mananger and run alsamixer, it works 2015-12-17 08:48:43 segfaults always here 2015-12-17 08:50:22 if I maximise a terminal x client window, run alsamixer, then slowly make the window smaller, there is a point where is seg faults 2015-12-17 08:50:45 let me trace that 2015-12-17 08:56:42 eh, I botched my patch separation again. Mind if I send a unique patch for 4 apkbuilds ? >.> 2015-12-17 08:58:04 skarnet: ok 2015-12-17 08:58:20 you can git format-patch --stdout -1 | tpaste 2015-12-17 08:58:26 and i'll push it 2015-12-17 08:59:48 sent 2015-12-17 08:59:52 to alpine-aports 2015-12-17 09:00:06 brb 2015-12-17 09:03:34 right. I can start alsamixer in a maximised terminal window with screen-256color but not linux or vt100 2015-12-17 09:06:02 aso, to reproduce this, can anyone open a mate-terminal, export TERM=screen-256color (you might need the term definitions) maximise the X client window and run alsamixer? 2015-12-17 09:09:08 Segmentation fault (core dumped) 2015-12-17 09:09:08 ncopa-desktop:~$ 2015-12-17 09:09:15 xfce4-terminal 2015-12-17 09:12:57 yes, same for me on xfce4-terminal, but it's ok as full screen, mate-terminal and screen-256color 2015-12-17 09:47:33 ncopa: could you take a look at the package upgrades I submitted? 2015-12-17 09:52:05 yeah, i should 2015-12-17 09:53:31 yeah 2015-12-17 09:53:35 we want those 2015-12-17 09:55:01 ah, you already upgraded perl and wget 2015-12-17 10:00:04 sorry for that :) 2015-12-17 10:00:08 thanks! 2015-12-17 10:00:11 pushed the patches 2015-12-17 10:00:17 re mirrors 2015-12-17 10:00:27 we need some automatic check that can verify that they work 2015-12-17 10:00:37 and we need some routine to remove those who does not work 2015-12-17 10:00:39 arm edge still syncing disks to ssd. should be back online in few hours time. 2015-12-17 10:00:56 i have build-3-3-armhf on other box :) 2015-12-17 10:01:06 currently i am more interested in 3.3 than edge ;) 2015-12-17 10:01:17 :) 2015-12-17 10:01:22 very good 2015-12-17 10:09:18 ncopa: no need to be sorry, packages upgrades are trivial most of the time 2015-12-17 10:09:40 yeah 2015-12-17 10:10:10 can apk fetch from https? 2015-12-17 10:56:47 you should be able to, yes 2015-12-17 11:00:25 ok. arm edge is back up. 2015-12-17 11:40:20 fabled: can you please give me the link to your arm bootstrap scripts repo again? 2015-12-17 11:40:36 skarnet, http://dev.alpinelinux.org/~tteras/bootstrap/ 2015-12-17 11:40:44 but it's incomplete. i have work-in-progress to fix that 2015-12-17 11:40:59 thanks. When you have a complete version please ping me :) 2015-12-17 11:42:28 (I managed to get a native armhf build environment under qemu, but it was some work, and it's still very manual. I'm currently industrializing it.) 2015-12-17 11:47:08 hey 2015-12-17 11:48:31 ncopa: I'm slowly continuing my cleanings of package request/update related issues in redmine, like recategorizing, providing latest version information or closing old ones that received package but never got feedback, etc. keeping tickets for years only leads to bigger clutter. I hope you're fine with it. 2015-12-17 11:50:04 I try to reassing package request that already got into alpine after 3.2 as 3.3 too. 2015-12-17 11:50:35 s/request/&s/ 2015-12-17 11:50:49 fcolista: any idea to tackle #4443 issue? having non-rebuildable APKBUILD seems quite bad. 2015-12-17 11:53:09 alpine doesn't have source packages. it's double-edged sword in such cases. 2015-12-17 11:55:09 firefox 43 is avaliable 2015-12-17 12:12:03 przemoc: super! 2015-12-17 12:12:13 many thanks for helping us clean up the bugs.a.o 2015-12-17 12:14:14 oh so he just make new release with same version number? 2015-12-17 12:14:19 thats i pretty bad 2015-12-17 12:14:21 exactly 2015-12-17 12:14:25 I'm speechless 2015-12-17 12:14:40 if he dont want maintain his stuff, why should we do it? 2015-12-17 12:15:03 maintain, reproduceable builds, security fixes 2015-12-17 12:15:06 stable abi 2015-12-17 12:15:08 you name it 2015-12-17 12:15:15 "i have a bug" 2015-12-17 12:15:21 "ok what version is it?" 2015-12-17 12:15:23 "1.4.2" 2015-12-17 12:16:18 file a bug on it 2015-12-17 12:16:26 well, I haven't filled issue in his repo, but if he's doing this kind of mess for a longer time already, I'm afraid it would be hard to make him change such behavior. 2015-12-17 12:16:51 even if he dont change it, it servers as documentation 2015-12-17 12:16:57 for future generations 2015-12-17 12:17:15 we can point future users to the issue and explain why things are like they are 2015-12-17 12:18:14 it might also be that he simple dont know how to do proper release engineering :) 2015-12-17 12:20:15 przemoc: are there any specific issues you think we should prioritize for v3.3.0 release? 2015-12-17 12:20:48 otherwise, could you set target to 3.4.0? 2015-12-17 12:20:56 on stuff you think could wait 2015-12-17 12:21:03 and if unsure, ask 2015-12-17 12:24:43 ncopa, i know what's wrong with ncurses 2015-12-17 12:24:46 but not yet sure how to fix it 2015-12-17 12:24:56 causing alsamixer crash etc. 2015-12-17 12:24:59 it's pretty bad 2015-12-17 12:25:05 ncurses is bad? 2015-12-17 12:25:08 or alsamixer 2015-12-17 12:25:11 or both 2015-12-17 12:25:11 ncurses 2015-12-17 12:25:17 im not surprised 2015-12-17 12:25:19 and it requires building all apps 2015-12-17 12:25:23 what happens 2015-12-17 12:25:23 aw 2015-12-17 12:25:36 is that alsamixer is built against the regular version (non-wide) 2015-12-17 12:25:40 but the header go into wide mode 2015-12-17 12:25:51 and the inline macro writes/reads from invalid memory 2015-12-17 12:26:06 they changed wide-mode detection 2015-12-17 12:26:22 should file bug upstream then 2015-12-17 12:26:59 or is it the way we package things that is broke 2015-12-17 12:27:03 basically, NCURSES_WIDECHAR is what is used to define the mode for the libraries 2015-12-17 12:27:11 it's a #define 2015-12-17 12:27:13 previously it was 2015-12-17 12:27:18 if not defined -> go regular 2015-12-17 12:27:20 if defined -> go wide 2015-12-17 12:27:28 przemoc, back now 2015-12-17 12:27:30 now they changed that NCURSES_WIDECHAR is always defined 2015-12-17 12:27:34 0 = regular, 1 = wide 2015-12-17 12:27:39 and if at #include time it's not defined 2015-12-17 12:27:54 the default is based on defined(_XOPEN_SOURCE_EXTENDED) || (defined(_XOPEN_SOURCE) && (_XOPEN_SOURCE - 0 >= 500)) 2015-12-17 12:28:10 i think we/musl sets _XOPEN_SOURCE practically always 2015-12-17 12:28:13 iow, they gratuitously broke compatibility 2015-12-17 12:28:16 yes 2015-12-17 12:28:29 we should spank them 2015-12-17 12:28:44 yes 2015-12-17 12:28:51 but that's not a bug stricto sensu 2015-12-17 12:28:56 well 2015-12-17 12:28:58 actually 2015-12-17 12:29:05 i think it was: 2015-12-17 12:29:06 -#if (defined(_XOPEN_SOURCE_EXTENDED) || defined(_XPG5)) && defined(WACS_ULCORNER) 2015-12-17 12:29:14 no 2015-12-17 12:29:20 wacs_ulcorner me harder 2015-12-17 12:30:26 that was different 2015-12-17 12:30:28 it used to be: 2015-12-17 12:30:30 #undef NCURSES_WIDECHAR 2015-12-17 12:30:30 #if defined(_XOPEN_SOURCE_EXTENDED) || defined(_XPG5) 2015-12-17 12:30:30 #define NCURSES_WIDECHAR 2015-12-17 12:30:30 #endif 2015-12-17 12:30:43 so yeah, you guys can complain upstream but I think fabled is right and the solution is to rebuild 2015-12-17 12:30:58 i think the change was good 2015-12-17 12:31:09 they just forgot to -DNCURSES_WIDECHAR=[01] in the pkgconfig file 2015-12-17 12:31:16 the thing is 2015-12-17 12:31:21 we did rebuild 2015-12-17 12:31:24 everything 2015-12-17 12:31:30 ncopa, the problem is, the headers get it wrong 2015-12-17 12:31:34 yeah. And any change (even a good one) that breaks compatibility needs to come with flashing red neon signs 2015-12-17 12:31:49 the rebuild we made, produced binaries that corrupt stuff 2015-12-17 12:31:56 because ncurses messed up upstream 2015-12-17 12:32:03 so we shoudl complain 2015-12-17 12:32:06 yes 2015-12-17 12:32:17 the application should now always explicitly -DNCURSES_WIDECHAR=0|1 2015-12-17 12:32:26 based on what lib it wants to link to 2015-12-17 12:32:34 so the .pc file should really provide that #define 2015-12-17 12:33:15 the new headers do: 2015-12-17 12:33:16 #ifndef NCURSES_WIDECHAR 2015-12-17 12:33:16 #if defined(_XOPEN_SOURCE_EXTENDED) || (defined(_XOPEN_SOURCE) && (_XOPEN_SOURCE - 0 >= 500)) 2015-12-17 12:33:16 #define NCURSES_WIDECHAR 1 2015-12-17 12:33:16 #else 2015-12-17 12:33:20 #define NCURSES_WIDECHAR 0 2015-12-17 12:33:22 #endif 2015-12-17 12:33:24 #endif /* NCURSES_WIDECHAR */ 2015-12-17 12:33:29 I suppose it would be faster to modify the ncurses APKBUILD and get the correct .pc manually for 3.3 2015-12-17 12:33:29 ncopa: I'll try to list some stuff today. generally everything that has package upgrade issues and doesn't require any bigger changes (beside removing patches that were already incorporated upstream) would be really nice being upgraded before releasing 3.3. 2015-12-17 12:33:32 so if pkg-config defines NCURSES_WIDECHAR it goes ok 2015-12-17 12:33:39 faster than wait for the upstream change 2015-12-17 12:33:45 the failure is that they broke the default 2015-12-17 12:33:57 defined(_XPG5) was changed to (defined(_XOPEN_SOURCE) && (_XOPEN_SOURCE - 0 >= 500)) 2015-12-17 12:33:58 ncopa: #4925 for starters 2015-12-17 12:34:11 but it does not affect gnu libc i suppose 2015-12-17 12:34:19 or uproar would arise 2015-12-17 12:34:42 i think this ncurses issue is 3.3.0 blocker 2015-12-17 12:34:50 yes 2015-12-17 12:34:51 that, or nobody using glibc has switched to the latest ncurses 2015-12-17 12:35:01 fedora switched 2015-12-17 12:35:04 arch linux switched 2015-12-17 12:35:14 i'll ask void linux 2015-12-17 12:35:39 ncopa, arch packages widec mode only 2015-12-17 12:35:43 i think most do that 2015-12-17 12:35:50 widec is needed in utf8 anyway 2015-12-17 12:35:52 debian is on 6 in stretch too (I'm on jessie/stretch/sid, though) 2015-12-17 12:35:58 we should probably also do that 2015-12-17 12:36:07 package only widec mode, and drop the regular one 2015-12-17 12:36:16 my point is, if you want 3.3.0 out before Xmas, better patch the ncurses apkbuild yourselves so it correctly advertises for pkg-config than wait for ncurses upstream to fix it 2015-12-17 12:36:27 yes 2015-12-17 12:36:49 ok 2015-12-17 12:36:51 that's what we'll do. just trying to figure out how to patch it 2015-12-17 12:36:53 so we drop regular mode 2015-12-17 12:36:57 and go for widec only? 2015-12-17 12:37:16 i think so 2015-12-17 12:37:16 i am tempted to do so 2015-12-17 12:37:18 what are the common usages of regular mode ncurses? 2015-12-17 12:37:29 regular works only with 8-bit locales 2015-12-17 12:37:33 is tehre some way to detect the regula mode apps? 2015-12-17 12:37:34 but musl is utf8 default 2015-12-17 12:37:45 ncopa, it's source compatible 2015-12-17 12:37:59 we can compile the app against either 2015-12-17 12:38:00 is anything in alpine NOT using utf8? 2015-12-17 12:38:01 and it just works 2015-12-17 12:38:11 seems ncurses 2015-12-17 12:38:19 so yeah; even from correctness point 2015-12-17 12:38:23 yes, i just wonder how to generate the list of packages needs to be rebuilt 2015-12-17 12:38:24 dropping regular is the right thing to do 2015-12-17 12:38:24 seems like 2015-12-17 12:38:35 pkgdesc="Ncurses wide character libraries" 2015-12-17 12:38:36 depends="$pkgname-terminfo-base" 2015-12-17 12:38:36 mkdir -p "$subpkgdir"/usr/lib 2015-12-17 12:38:36 mv "$pkgdir"/usr/lib/lib*w.so.* "$subpkgdir"/usr/lib/ 2015-12-17 12:39:10 so we apk search -r so:lib*.so which are not lib*w.so.* 2015-12-17 12:39:16 yes 2015-12-17 12:39:20 might be there are not so many pkgs 2015-12-17 12:39:45 arch does https://projects.archlinux.org/svntogit/packages.git/tree/trunk/PKGBUILD?h=packages/ncurses 2015-12-17 12:39:46 apk doesn't have apt-get rdepends equivalent? 2015-12-17 12:40:10 s/get/cache/ 2015-12-17 12:40:14 apk search -r 2015-12-17 12:40:54 i think that's why the regular mode broke 2015-12-17 12:40:58 no one is really using it anymore 2015-12-17 12:41:06 everyone uses only the wide mode 2015-12-17 12:41:12 since any system is utf8 2015-12-17 12:41:16 sane* 2015-12-17 12:41:48 goody, lets fix it 2015-12-17 12:42:08 przemoc: strace needs rebase the arm patches 2015-12-17 12:42:33 fabled: many thanks for investigating that. i think i can work on ncurses 2015-12-17 12:42:49 thanks 2015-12-17 12:43:08 fabled: if you have time, and want, could you have a look at strace 4.10 2015-12-17 12:43:12 i dont think its difficult 2015-12-17 12:43:15 i fix kdelibs for arm first 2015-12-17 12:43:19 the arm patches needs to be rebase 2015-12-17 12:43:25 kdelibs can be moved to unmaintained 2015-12-17 12:43:31 not worth it 2015-12-17 12:43:35 its way outdated 2015-12-17 12:43:45 and updating it is a bigger project 2015-12-17 12:44:01 ERROR: unsatisfiable constraints: 2015-12-17 12:44:01 udev-dev (missing): 2015-12-17 12:44:01 required by: world[udev-dev] 2015-12-17 12:44:03 ? 2015-12-17 12:44:11 should be eudev-dev 2015-12-17 12:44:15 but i tried that too 2015-12-17 12:44:17 does not build 2015-12-17 12:44:25 arm or anything? 2015-12-17 12:44:28 anything 2015-12-17 12:44:44 ah, then it should probably go to unmaintained 2015-12-17 12:45:09 nmeum: do you think you could help us with that? 2015-12-17 12:45:21 ok. so strace then 2015-12-17 12:45:29 nmeum: move kdelibs and anything depedning on kdelibs to unmaintained? 2015-12-17 12:45:52 coffee first 2015-12-17 12:47:39 ACTION reads the backlog 2015-12-17 12:48:03 here are the ncurses package list: http://tpaste.us/37oL 2015-12-17 12:48:24 aw... llvm 2015-12-17 12:48:47 samba 2015-12-17 12:48:54 you missed some packages from testing, like sngrep 2015-12-17 12:48:56 make those builders work 2015-12-17 12:49:00 arm builder will need a couple of days 2015-12-17 12:49:19 go go greenhouse effect 2015-12-17 12:49:27 :D 2015-12-17 12:50:21 did apk search -r gave incomplete result? 2015-12-17 12:50:45 gah! asking just in case, any qemu experts in here? 2015-12-17 12:51:24 ACTION is pulling his hair out trying to get a NIC recognized. Not much hair left. 2015-12-17 12:52:39 ncopa: I can move kdbelibs and everything depending on it to unmaintained in the evening. I am currently busy at my uni (sorry) 2015-12-17 12:52:53 ok. np 2015-12-17 12:52:55 and thanks 2015-12-17 12:53:26 unless someone else with git push access beats you to it ;) 2015-12-17 12:55:01 re ncurses 2015-12-17 12:55:16 other option is to rebuild it with patched pkg-config 2015-12-17 12:55:25 and then we rebuild as things are found broken 2015-12-17 12:55:27 or similar 2015-12-17 12:55:43 til now we only have alsamixer afaik 2015-12-17 12:56:06 sngrep was also misbehaving according to fcolista 2015-12-17 12:56:19 przemoc, right. 2015-12-17 12:56:28 That was the package that i was trying to remember 2015-12-17 12:56:30 thx :) 2015-12-17 12:57:14 I already mentioned it 7 minutes ago, though ;) 2015-12-17 12:57:25 regarding incomplete ncurses apps list 2015-12-17 12:58:35 i was @lunch :) 2015-12-17 12:58:57 np 2015-12-17 13:00:10 przemoc: re: #4443, current git log --decorate=full still shows v1.4.2 with previous commit(046075924941) dated Jul 10 19:57:05 2015 2015-12-17 13:00:45 but I found something weird in NEWS fils seems changed 2015-12-17 13:00:51 vkris, przemoc, what is ULib at the moment used for? 2015-12-17 13:00:58 hum 2015-12-17 13:01:09 ncurses might have been a packaging mistake 2015-12-17 13:01:16 why devs are so lazy in versioning?!?! 2015-12-17 13:01:22 C++ application development framework 2015-12-17 13:01:24 github generation... 2015-12-17 13:01:57 so maybe github web interface was lieing to me 2015-12-17 13:01:59 fcolista: I have yet to do some app on it, but by docs seems nicce 2015-12-17 13:02:21 I have to admit that I'm a lot lazier with numbered versions since I've adopted git. Telling people "use the latest git" is *very* comfortable for a dev. 2015-12-17 13:03:28 uff 2015-12-17 13:03:43 przemoc: thanks for taking notice, I would try get in touch with the developer and github people 2015-12-17 13:03:52 might be a github issue 2015-12-17 13:04:05 or git bug 2015-12-17 13:04:58 ok. I propose to use git snapshot feature until this dev doesn't start to do versioning correctly. 2015-12-17 13:05:23 oh. It's already shipped v1.4.2 2015-12-17 13:05:23 but it does seems dev is not concerned about making best of git 2015-12-17 13:05:40 I think many a dev would welcome a git snapshot feature with open arms 2015-12-17 13:05:44 vkris, we can tell him...as ncopa did several times.. 2015-12-17 13:06:01 vkris: if you can investigate it further and correct my misaccusations (if they are mis), please do. I prefer to go through bugs.a.o, then file issues on github. 2015-12-17 13:06:07 skarnet: the problem is that developers dont want maintain their stuff 2015-12-17 13:06:26 git snapshot move the maintenance to packager 2015-12-17 13:06:29 "you want to package my stuff ? use that git commit" as opposed to "meh, that means I have to tag a numbered release" 2015-12-17 13:06:54 i dont want maintain a random git snapshot for 2 years 2015-12-17 13:07:06 what if next week the developer does a refactor 2015-12-17 13:07:07 but does git commit a sequential version number 2015-12-17 13:07:09 this is true whether or not the feature exists 2015-12-17 13:07:21 and then security fix is needed 2015-12-17 13:07:23 pkging is versioning in sequence 2015-12-17 13:07:28 and developers who do want to maintain would love a snapshot feature 2015-12-17 13:07:53 what good if git makes it so simple 2015-12-17 13:08:05 i mean, making a release is not harder thatn git tag, done 2015-12-17 13:08:09 well the problem will occur whether the version you use is a tagged release or a snapshot 2015-12-17 13:08:11 they're sometimes temporarily unable to maintain/release for various strange reasons. like my releasing first (practically ever) metastore release after becoming official maintainer being delayed continuously... 2015-12-17 13:08:30 tagging release makes you think a bit on how you plan to maintain it 2015-12-17 13:08:40 does it ? o.O 2015-12-17 13:08:45 ACTION whistles 2015-12-17 13:08:56 well, it encourage it 2015-12-17 13:09:05 semantic versioning 2015-12-17 13:09:20 x.y.z, new feature, y++ 2015-12-17 13:09:29 bugfix only, z++ 2015-12-17 13:09:31 Oh, definitely 2015-12-17 13:09:38 major rewrite, x++ 2015-12-17 13:09:54 unless ofc, you are a ruby gem dev... 2015-12-17 13:10:11 I use l.x.y.z with major rewrite=l++ and x++=API change 2015-12-17 13:10:38 vkris: in https://github.com/stefanocasazza/ULib/releases/tag/v1.4.2 I see: released this 7 days ago and commit is: f4cc822, which is: committed 7 days ago 2015-12-17 13:10:42 because sometimes you need to do some minor fixes that still break compatibility 2015-12-17 13:10:56 in any case, the underlying problem is unwillingness to maintain the code 2015-12-17 13:11:02 vkris: maybe you haven't updated tags in your local clone? git fetch --all -p 2015-12-17 13:11:04 regardless if yo do git snapshots or releases 2015-12-17 13:11:11 exactly 2015-12-17 13:11:12 przemoc: try git clone then git log --decorate=full 2015-12-17 13:11:47 and git snapshots are probably more recent than the last numbered release 2015-12-17 13:12:01 especially if the dev whips out a release every 5 years 2015-12-17 13:18:03 vkris: $ git log --all --decorate --oneline | head -1 2015-12-17 13:18:04 f4cc822 (HEAD, tag: v1.4.2, origin/master, origin/HEAD, master) some fix 2015-12-17 13:19:05 vkris: $ git show HEAD | grep Date 2015-12-17 13:19:05 Date: Thu Dec 10 18:10:25 2015 +0100 2015-12-17 13:19:58 so github web interface is (thankfully) not lying 2015-12-17 13:20:25 (which would be more than terrible, if it was the case) 2015-12-17 13:22:19 let me recheck, I did git pull onto old clone (at, 0460759249417) 2015-12-17 13:22:56 he may be rewriting history too, I wouldn't be surprised 2015-12-17 13:23:19 once again, do: git fetch --all -p 2015-12-17 13:24:18 or maybe -t was needed too? I think --all was enough, but maybe I'm misremembering things 2015-12-17 13:27:15 anyway, I did a clean clone and the tag is pointing to recent commmit, which couldn't be the same in the past, yet tag name in APKBUILD is the same, so his releasing approach is definitely broken 2015-12-17 13:29:17 vkris: please report the problem upstream, if you're interested in this package 2015-12-17 13:30:36 oh, I miswrote earlier. "I prefer to go through bugs.a.o than file issues on github." (it was then) 2015-12-17 13:33:38 retrying - git fetch --all -p on fresh old copy 2015-12-17 13:33:48 why projects still use sourceforge after all the crap they did this year is beyond me... 2015-12-17 13:33:54 :) 2015-12-17 13:34:28 fresh old? sounds like oxymoron 2015-12-17 13:34:50 lol 2015-12-17 13:35:11 I mean copy from by archive 2015-12-17 13:35:15 from my 2015-12-17 13:35:51 it is on a dvd, I ejected it now getting back again 2015-12-17 13:36:10 :) 2015-12-17 13:36:12 I bet you haven't upgraded it locally for a long time, because you thought it's still 1.4.2 on github, so there is no need to. :) 2015-12-17 13:36:54 which is latest version? 1.4.2? good, same as i run. no need to update... 2015-12-17 13:37:06 i am still running the fresh old 1.4.2 version 2015-12-17 13:37:09 :) 2015-12-17 13:37:41 dvd dt is 8th Aug 2015 2015-12-17 13:38:26 to be honest I don't remember seeing ever such a releasing curiosity. drop tags at all if you want to retag same tag to latest commit... 2015-12-17 13:39:32 nope, same issue, this was done using git version 1.7.10.4 2015-12-17 13:41:59 ok would try later with newer git and ask on git channel 2015-12-17 13:42:20 vkris: what are you fighting with, because I don't understand? 2015-12-17 13:44:58 as I wrote, maybe -t is required to get updated tags during fetch. just don't remember whether --all is enough. 2015-12-17 13:45:18 algitbot: this is getting tiresome, you dumb bot 2015-12-17 13:46:57 algitbot: please stop annoy przemoc 2015-12-17 13:47:10 \o/ 2015-12-17 13:47:43 but I hope he won't stop linking bugs... 2015-12-17 13:48:22 i cannot find any packages in testing or community that links to libncurses.so.6 2015-12-17 13:48:35 or any of the other non-widec libs 2015-12-17 13:49:06 for i in $(apk info -L ncurses-libs | grep 6$ | sed 's,.*/,so:, 2015-12-17 13:49:06 '); do sudo apk search -U --repositories-file /dev/null -X http://rsync.alpinelinux.org/alpine/edge/community -r --exact --origin -q $i ; done | sort -u 2015-12-17 13:49:37 sngrep is in testing 2015-12-17 13:50:29 huh 2015-12-17 13:50:30 ok 2015-12-17 13:51:14 that's why I asked earlier whether your searching is faulty 2015-12-17 13:51:24 apparently it is 2015-12-17 13:51:30 hmm... could be a bug in git pull/fetch or I need to upgrade git and read docs 2015-12-17 13:52:53 vkris: please state your problem with git, what's wrong? 2015-12-17 13:54:27 simpler would be to send you archive copy of ULib and let you do git pull/fetch 2015-12-17 13:56:49 if it's _really_ simpler then I guess I have no choice. upload it somewhere and I'll download it in my debian vm and will try fetch from it. 2015-12-17 13:58:46 ok 2015-12-17 13:58:53 but knowing what's your problem is would be helpful anyway 2015-12-17 13:59:23 s/is // 2015-12-17 14:13:17 przemoc: sent links 2015-12-17 14:20:18 i'm gonna git push now 2015-12-17 14:20:25 partial rebuilt 2015-12-17 14:20:29 and push the rest later 2015-12-17 14:20:40 so rebuild gets started 2015-12-17 14:21:13 hm 2015-12-17 14:21:19 maybe wait for samba first 2015-12-17 14:22:29 ncopa: v3.3 would have gcc v5.3.0 ? 2015-12-17 14:22:36 yes 2015-12-17 14:22:40 we pushed gcc-5.3 2015-12-17 14:22:56 its a bugfixes-only release 2015-12-17 14:23:05 ok 2015-12-17 14:23:46 but all pkgs were build using 5.2.0 2015-12-17 14:26:11 przemoc: found it ! yes, there seems a difference in how `git fetch --all -p -t` is done in git version 2.6.3 and 1.7.10.4 2015-12-17 14:26:38 v2.6.3 behaves as expected 2015-12-17 14:27:27 vkris: sorry, didn't have time to check your files earlier, but it's good that you solved the problem already 2015-12-17 14:27:55 thanks 2015-12-17 14:28:02 git has many dormant bugs 2015-12-17 14:28:38 some examples are in recent post from github: http://githubengineering.com/move-fast/ 2015-12-17 14:36:51 ncurses update pushed 2015-12-17 14:46:20 ncopa: should I move packages depending on kdelibs to unmaintained/ in one commit or should I do one commit for each package? 2015-12-17 14:46:28 can anyone test whether xmlsec works? #4554 2015-12-17 14:46:41 nmeum: doesnt matter really 2015-12-17 14:47:20 just make sure that anything that is left in testing does not depend on anything moved to unmaintained 2015-12-17 14:47:59 yeah 2015-12-17 14:48:24 move 15 packages so far, I believe that's it 2015-12-17 14:48:27 *moved 2015-12-17 14:48:32 sounds good 2015-12-17 14:49:27 ncopa: pushed 2015-12-17 14:50:34 thanks! 2015-12-17 14:51:41 no problem 2015-12-17 14:54:35 nmeum: you are climbing: http://git.alpinelinux.org/cgit/aports/stats/?period=w&ofs=10 2015-12-17 14:54:57 you passed by fcolista 2015-12-17 14:55:00 \o/ 2015-12-17 14:55:29 :) good job nmeum 2015-12-17 14:55:49 yes, good job indeed 2015-12-17 14:55:51 those statistics don't mean much to me but it's nice to know that I managed to fix some thinks this week :) 2015-12-17 14:55:58 *things 2015-12-17 15:03:36 in 2012, i did 70% of the commits 2015-12-17 15:03:44 in 2015, i did less than 50% 2015-12-17 15:04:09 in 2015 there were 200% more commits? 2015-12-17 15:04:36 :) 2015-12-17 15:04:49 2012 was 5233 commits 2015-12-17 15:04:57 2015 was 7129 2015-12-17 15:05:17 only 40% more 2015-12-17 15:08:40 przemoc: are they any tickets i should fix for v3.3.0? 2015-12-17 15:08:46 except strace 2015-12-17 15:09:10 oh, i have an mkinitfs/awk issue not reported... 2015-12-17 15:16:08 ncopa: 2015-12-17 15:16:12 #3613 2015-12-17 15:16:20 #3614 2015-12-17 15:16:29 #4613 2015-12-17 15:17:38 #4964 2015-12-17 15:22:11 there was some issue with libbsd 2015-12-17 15:22:21 i think libbsd is very gnu centric 2015-12-17 15:22:35 i dont remember what it was, but there was som reason to not upgrade it 2015-12-17 15:25:19 are ok with pasting above 3 lines to redmine as comment? 2015-12-17 15:25:27 s/are/& you/ 2015-12-17 15:25:29 ncopa: ^ 2015-12-17 15:26:15 I'll set it to 100% and there will be new issue targetting 3.4, ok? 2015-12-17 15:28:19 ok 2015-12-17 15:30:13 i set target to 3.4.0 2015-12-17 15:30:37 no, there will be new issue for 3.4 2015-12-17 15:30:46 let's have trace of upgrade in 3.3 2015-12-17 15:30:58 ACTION has to vanish for a few minutes 2015-12-17 15:31:13 kaniini: do you remember why we dont want update libbsd? 2015-12-17 15:32:08 ncopa: already updated upgrade request for libbsd, will create new one later 2015-12-17 15:32:14 ACTION will really vanish now 2015-12-17 15:32:33 thanks! 2015-12-17 15:38:09 it doesnt build 2015-12-17 15:38:11 dont wanna deal with it 2015-12-17 15:38:13 ;p 2015-12-17 15:38:21 oh ok :) 2015-12-17 15:39:03 it built here 2015-12-17 15:39:10 probably recent musl fixes 2015-12-17 15:42:25 ncopa: but there is 0.8.1 2015-12-17 15:42:33 as I already wrote in the comment earlier ;) 2015-12-17 15:42:49 http://bugs.alpinelinux.org/issues/4613#note-4 2015-12-17 15:43:15 bah. 2015-12-17 15:43:30 and thanks for upgrading mentioned stuff 2015-12-17 15:43:37 sounds like i need vacation 2015-12-17 15:43:51 thanks for help follow things up 2015-12-17 15:51:30 I have so many issue tabs open that I'm no longer sure if we're clear enough ;) 2015-12-17 15:51:43 there are still some issues I have to update 2015-12-17 15:54:34 anyone looked at #4492? it may reappear in 3.3 -> edge scenario too 2015-12-17 16:00:07 I suggest changing this pesky ULib's APKBUILD to use currect https://github.com/stefanocasazza/ULib/archive/f4cc822edd7797a0dd4b47379f555e05b9ab8990.tar.gz for now as v1.4.2 disguise, so it's clear which source is used 2015-12-17 16:00:23 #4443 2015-12-17 16:00:31 s/currect/current/ 2015-12-17 16:11:18 fcolista: you there? 2015-12-17 16:11:36 i do 2015-12-17 16:12:31 would you be that kind and upgrade nanomsg to 0.8-beta? #3632 2015-12-17 16:13:34 umh 2015-12-17 16:13:40 i'm not able to download it 2015-12-17 16:13:52 gotcha 2015-12-17 16:14:04 https://github.com/nanomsg/nanomsg/archive/0.8-beta.tar.gz 2015-12-17 16:14:47 building 2015-12-17 16:16:37 great, thanks! 2015-12-17 16:16:37 done 2015-12-17 16:16:50 v4l-utils continue to fails 2015-12-17 16:17:07 some musl-related stuff? 2015-12-17 16:17:17 umh 2015-12-17 16:17:29 i remember i've been trying few weeks ago 2015-12-17 16:17:45 but i was not able to build it correctly 2015-12-17 16:19:28 ah ok 2015-12-17 16:19:36 'cause of ncurses, it will took a while 2015-12-17 16:19:37 ok 2015-12-17 16:25:41 fcolista: I don't want to burden you too much with this v4l-utils. 1.6.x is simply quite old, so it would be really nice having more recent branch in new alpine release. if you think it's unfixable in reasonable time, then go ahead and revert the commit bumping the version up. 2015-12-17 16:26:37 przemoc, i think first we need to wait for ncurses rebuild 2015-12-17 16:28:04 ok. in such case, will you be here for a bit more today anyway? may I bother you with some other minor upgrades like nanomsg before? 2015-12-17 16:29:23 go ahead 2015-12-17 16:30:31 are build bot logs accessible anywhere for public? 2015-12-17 16:30:51 yes...via mosquitto iirc 2015-12-17 16:31:05 one sec 2015-12-17 16:38:16 umh 2015-12-17 16:38:28 i'm not able to find the note about mosquitto buildlog 2015-12-17 16:38:38 perhaps clandmeter knows 2015-12-17 16:38:41 or ncopa :) 2015-12-17 16:49:55 I installed mosquitto on my debian vm at least 2015-12-17 16:49:58 never used it, though 2015-12-17 16:52:28 clandmeter: I guess you don't want to upgrade chromium any more before releasing 3.3? http://bugs.alpinelinux.org/issues/285#note-18 2015-12-17 16:52:40 oh shit 2015-12-17 16:53:05 >>> chromium*: Create chromium-47.0.2526.106-r0.apk 2015-12-17 16:54:19 oh, indeed, there was another one 2 days ago 2015-12-17 16:54:23 hatseflats! 2015-12-17 16:55:10 i forgot i started a build yesterday. 2015-12-17 16:55:23 did it finish? ;) 2015-12-17 16:55:31 yep, just pushed it. 2015-12-17 16:55:50 ok, I'll update the issue then 2015-12-17 16:55:58 ok thx! 2015-12-17 16:56:23 (did you test that it works? :>) 2015-12-17 16:56:43 no, i have no X here... 2015-12-17 16:57:13 I sent another minor apkbuild patch to alpine-aports (I had a tag issue with s6-2.2.4.1, had to make another release to fix it. Nothing serious.) 2015-12-17 16:57:18 i always wanted to setup a lxc container with xorg 2015-12-17 16:57:24 if it can make it into 3.3, good. If it can't, no big deal. 2015-12-17 16:58:00 przemoc: are you on alpine(desktop)? 2015-12-17 16:58:17 clandmeter: I will do that at some point early next year. If you get to it, I'll be interested in your experience. 2015-12-17 16:59:32 no, I even don't have alpine yet. :) but I'll change it hopefully really soon. it will be server-oriented installation, though, so rather without X. 2015-12-17 17:00:36 i think a few here have. 2015-12-17 17:01:29 I have an older notebook here which i tried to get X working without any luck. 2015-12-17 17:02:26 is X stuff in alpine generally second-class citizen, does barely anyone run X on alpine? 2015-12-17 17:02:47 ScrumpyJack has X for sure, I've seen his screenshot 2015-12-17 17:02:51 ncopa and fabled do i believe 2015-12-17 17:03:08 probably more who dont talk about it 2015-12-17 17:04:23 did ncopa or fabled tested earlier chrome builds? 2015-12-17 17:05:03 chrome is quite massive beast, so it's easy to have something broken in it 2015-12-17 17:05:16 chromium yes 2015-12-17 17:05:26 yup, sorry, chromium I meant 2015-12-17 17:05:40 i talked to somebody about it few days ago 2015-12-17 17:05:43 he says it works 2015-12-17 17:05:50 except flash didnt 2015-12-17 17:06:13 flash is a pain to make work anywhere 2015-12-17 17:06:28 it works on alpine and firefox 2015-12-17 17:06:48 if you install libc6-compat 2015-12-17 17:06:48 hm 2015-12-17 17:07:04 how well it works i dunno. 2015-12-17 17:07:20 so setting done tot 100%, thanks 2015-12-17 17:07:38 s/tot/to/ 2015-12-17 17:07:40 so, glibc necessary to make flash work? 2015-12-17 17:07:59 not really, we can fool it :) 2015-12-17 17:08:13 but yes it links to glibc 2015-12-17 17:08:37 i have no need for it, so i didnt really test it. 2015-12-17 17:09:10 we can fool it? as in, link it against musl while pretending it's glibc? 2015-12-17 17:09:13 my desktop at home has alpine wiht X, but im never behind it anymore now my wife is made it her office... 2015-12-17 17:09:15 fcolista: please fix ULib's APKBUILD by using https://github.com/stefanocasazza/ULib/archive/f4cc822edd7797a0dd4b47379f555e05b9ab8990.tar.gz 2015-12-17 17:09:44 skarnet: look at libc6-compat (musl subpkg) 2015-12-17 17:10:44 https://pkgs.alpinelinux.org/contents?pkgname=libc6-compat&arch=x86_64&repo=main 2015-12-17 17:11:12 I only have one installation and I'd like to keep it clean of hacks AND of everything even remotely looking like glibc, so I'll stand by my guess. :) 2015-12-17 17:11:35 we'll think how to deal with ULib more properly after 3.3 release 2015-12-17 17:11:59 but let's have truly known source which we base the build on 2015-12-17 17:12:01 clandmeter: I guess all those files are links to libc.so ? 2015-12-17 17:12:25 i think so 2015-12-17 17:12:57 clandmeter: maybe you know about mosquitto buildlog? 2015-12-17 17:13:17 what do you want to know? 2015-12-17 17:13:28 how to look at build logs 2015-12-17 17:13:35 (if they're public) 2015-12-17 17:13:44 pkgs.a.o has them 2015-12-17 17:13:59 only the ones already finished 2015-12-17 17:14:10 or just go to http://build.alpinelinux.org/buildlogs/ 2015-12-17 17:14:43 http://build.alpinelinux.org/buildlogs/build-edge-x86_64/ and enter the pkgdir 2015-12-17 17:15:06 previously there was a main log file 2015-12-17 17:15:09 im not sure there still is one. 2015-12-17 17:16:27 so, for instance, how can I check whether ncurses and its rdeps building already finished? 2015-12-17 17:16:49 is there any easy way beside going into each package directory for recent build? 2015-12-17 17:18:03 http://build.alpinelinux.org/buildlogs/build-edge-x86_64.v3.3.0_rc3-105-g8e2fd38.log failing xen sounds bad 2015-12-17 17:21:03 http://build.alpinelinux.org/ <- really nice overview 2015-12-17 17:21:35 i dont think there is an easy way. except you can check pkgs.a.o. it will list build pkgs. 2015-12-17 17:21:43 I don't see anything on http://build.alpinelinux.org/ 2015-12-17 17:21:58 just : 2015-12-17 17:21:59 # host activity aport built aport total 2015-12-17 17:22:03 nothing else 2015-12-17 17:22:05 wait a bit 2015-12-17 17:22:27 it should fill with 16 hosts and their activity 2015-12-17 17:22:45 nope 2015-12-17 17:23:46 maybe try in chrome/chromium 2015-12-17 17:23:55 i've been trying chrome and firefox 2015-12-17 17:25:04 might be related to the fact that i'm behind a proxy? 2015-12-17 17:25:12 definitely 2015-12-17 17:25:42 there is some javascript mosquitto magic 2015-12-17 17:25:43 fcolista: yes proxy makes it not work 2015-12-17 17:25:50 how that table is filled? 2015-12-17 17:25:54 cause msoquito will try to connect to another port 2015-12-17 17:25:58 ah ok 2015-12-17 17:26:03 it's mosquitto 2015-12-17 17:26:08 just check the src 2015-12-17 17:26:08 this explains. 2015-12-17 17:26:22 should it update real-time? 2015-12-17 17:26:28 yes 2015-12-17 17:26:33 it *should* 2015-12-17 17:26:47 gotta try proxychains 2015-12-17 17:27:00 openvpn :) 2015-12-17 17:27:08 or ssh with socks 2015-12-17 17:27:49 yeah...ssh tunnel it's better. 2015-12-17 17:28:06 if v4l-utils failed, why it's still shown there? 2015-12-17 17:28:25 is it stuck? 2015-12-17 17:30:52 are 3.3 builds triggered automatically from edge before release or there is some additional gatekeeping? 2015-12-17 17:31:08 yes 2015-12-17 17:31:16 it follows master 2015-12-17 17:31:19 untill we branch 2015-12-17 17:32:01 its same as edge, except rebuild from scratch. 2015-12-17 17:32:19 and patched up to fix build errors 2015-12-17 17:33:57 so if -x86/-x86_64 does nothing, why it did not build updated nanomsg yet? 2015-12-17 17:35:49 fcolista: thanks for ULib fix 2015-12-17 17:36:37 pkgrel did not need to be increased? 2015-12-17 17:37:42 pkgver remains 1.4.2, so shouldn't pkgrel be bumped to trigger the build? 2015-12-17 17:38:24 or _gitsnapshot overrides pkgver? 2015-12-17 17:42:53 Does the binary change? 2015-12-17 17:44:02 yes, the commit I gave you is the one currently pointed by the tag. I don't know (and possibly no one knows) what commit was used for tagging it back then. 2015-12-17 17:44:40 that's why I wanted to fix this mess 2015-12-17 17:45:04 yeah ok. I though they were the same. 2015-12-17 17:45:09 Bumped pkgrel. 2015-12-17 17:46:02 I explained it in the issue. http://bugs.alpinelinux.org/issues/4443#note-2 2015-12-17 17:46:06 thanks 2015-12-17 17:58:10 clandmeter: maybe in fact only one host is doing all the builds and that's why something is blocked on http://build.alpinelinux.org/? 2015-12-17 18:03:42 fcolista: did you test that this recent ULib builds locally before pushing? 2015-12-17 18:36:29 przemoc: mosquitto_sub -h msg.alpinelinux.org -t 'build/#' -v 2015-12-17 18:36:51 I don't think they are active 2015-12-17 18:52:36 vkris: thanks! 2015-12-17 18:53:37 but why xen and v4l-utils are blocking edge-x86*? shouldn't they build other stuff in queue after failing? 2015-12-17 19:53:36 BTW: alsamixer still segfaults here 2015-12-17 20:46:26 outch 2015-12-17 20:46:33 its a big mess 2015-12-17 20:48:22 now it's readline? 2015-12-17 20:49:03 its all of it 2015-12-17 20:49:18 anybody has a arch linux box? 2015-12-17 20:49:21 what is bash linked to? 2015-12-17 20:49:34 libncurses.so.6? 2015-12-17 20:49:44 or libncursesw.so.6 2015-12-17 20:50:15 I have some vm, not updated recently, though. 2015-12-17 20:50:18 give me a minute 2015-12-17 20:50:59 libncursesw.so.5 2015-12-17 20:51:04 damn, I need to upgrade 2015-12-17 20:51:52 713 MB of downloads... 2015-12-17 20:53:27 https://projects.archlinux.org/svntogit/packages.git/tree/trunk/PKGBUILD?h=packages/bash 2015-12-17 20:53:58 w in explicit depends 2015-12-17 20:56:57 how they acomplish that? 2015-12-17 20:57:05 gcc -L./builtins -L./lib/readline -L./lib/readline -L./lib/glob -L./lib/tilde -L./lib/sh -Wl,--as-needed -rdynamic -Os -fomit-frame-pointer -o bash shell.o eval.o y.tab.o general.o make_cmd.o print_cmd.o dispose_cmd.o execute_cmd.o variables.o copy_cmd.o error.o expr.o flags.o jobs.o subst.o hashcmd.o hashlib.o mailcheck.o trap.o input.o unwind_prot.o pathexp.o sig.o test.o version.o alias.o array.o arrayfunc.o assoc.o braces.o bracecomp.o 2015-12-17 20:57:05 bashhist.o bashline.o list.o stringlib.o locale.o findcmd.o redir.o pcomplete.o pcomplib.o syntax.o xmalloc.o -lbuiltins -lglob -lsh -lreadline -lhistory -lcurses -ltilde -ldl 2015-12-17 20:57:19 /bin/ld: cannot find -lcurses 2015-12-17 20:57:56 ah 2015-12-17 20:58:05 https://projects.archlinux.org/svntogit/packages.git/tree/trunk/PKGBUILD?h=packages/ncurses#n35 2015-12-17 20:58:05 isn't it readline dependency? 2015-12-17 20:59:38 doesn't look nice, but if it's shortest path to get things finally build... 2015-12-17 20:59:54 shortest path is rebuild it all again 2015-12-17 21:05:54 oh well... 2015-12-17 21:07:40 ncopa: BTW were stuck packages in build.a.o, like failing xen, related to curses (nomen omen :>) 2015-12-17 21:07:52 yes 2015-12-17 21:07:58 they all broke because of this 2015-12-17 21:11:02 non-wide ncurses should be really ditched, I guess. (not sure whether there are apps meant/working only for non-wide version, though.) 2015-12-17 21:12:25 thats what i did 2015-12-17 21:12:31 i ditched non-wide 2015-12-17 21:12:41 but the non-wide variant got stuck 2015-12-17 21:13:19 and made the wide-variant-only non-installable 2015-12-17 21:13:24 it picked the old 2015-12-17 21:13:27 and rebuilt against the old 2015-12-17 21:13:30 and uploaded 2015-12-17 21:13:36 and purged the old 2015-12-17 21:14:05 so rebuilt apps were built against libncurses.so.6 instead of libncursesw.so.6 2015-12-17 21:17:24 so build w with nonw name and link w to it (and curses), just like arch did, to avoid fixing all APKBUILDs to link with proper libncurses 2015-12-17 21:17:37 thast what im doing 2015-12-17 21:18:02 er 2015-12-17 21:18:03 no 2015-12-17 21:18:06 that will not work 2015-12-17 21:18:35 y? 2015-12-17 21:18:36 i am making -lncurses link to libncursesw 2015-12-17 21:18:41 so i need to rebuild 2015-12-17 21:18:52 it must rebuild or apk will not pick the correct runtime dep 2015-12-17 21:23:02 upgraded my arch 2015-12-17 21:23:09 I misread what arch did 2015-12-17 21:23:40 libncurses.so has simply INPUT(-lncursesw) 2015-12-17 21:24:21 I didn't know there was such notation for textual "redefinition" 2015-12-17 21:25:30 and of couse similarly for libncurses++ 2015-12-17 21:27:21 is there any win of doing it that way over normal symlink? 2015-12-17 21:28:01 readelf -d 2015-12-17 21:28:17 will show NEEDS libncursesw.so.6 2015-12-17 21:28:29 instead of libncurses.so.6 2015-12-17 21:28:41 which is what we need 2015-12-17 21:29:34 right 2015-12-17 21:36:08 newt and nethack has keyword 'lncurses' 2015-12-17 21:37:04 when you are it, bump newsbeuter+stfl pls 2015-12-17 21:37:44 what are the new versions? 2015-12-17 21:38:59 ncopa: is it possible to make reporter being able to set status to resolved, rejected or feedback? when issues are new, I can only change them to closed 2015-12-17 21:39:32 I mean rebuild for widechar 2015-12-17 21:39:54 just guessing it might 2015-12-17 21:40:01 vkris: are they linked against ncurses.so.6? 2015-12-17 21:40:04 i think they are not 2015-12-17 21:41:35 przemoc: i changed your role to developer 2015-12-17 21:42:41 libncursesw.so.6 2015-12-17 21:42:49 so no need to rebuild 2015-12-17 21:43:43 ok 2015-12-17 21:44:09 thanks! in fact I will be one, I promise to start sending patches in year 2016 instead of playing with issues only. 2015-12-17 21:47:04 how much space alpine mirror takes nowadays? 2015-12-17 21:47:10 i think we have more people sending patches than messing with issues 2015-12-17 21:47:55 # du -sh . 2015-12-17 21:47:55 173.7G . 2015-12-17 21:51:05 what's recommended way of creating local Alpine build environment? 2015-12-17 21:51:25 http://wiki.alpinelinux.org/wiki/Installing_Alpine_Linux_in_a_chroot - this one? 2015-12-17 21:57:31 vkris: does Wt works for you in alpine? 2015-12-17 21:58:17 you requested that package 5 months ago #4444 2015-12-17 21:58:43 przemoc: I would check Wt and ULib after v3.3 release 2015-12-17 21:59:29 planning some websocket related applications 2015-12-17 22:00:36 ULib has been updated by fcolista upon my request (in APKBUILD, not built yet), so we have crystal clear source (by commit sha1 instead of "living" tag) 2015-12-17 22:00:47 if you're not willing to test it, then I'll mark it as resolved 2015-12-17 22:02:28 resolved would not put that in main/, it needs "is working" said by some user 2015-12-17 22:03:02 so in case of bugs in future (if there will be any) in ulib or wt, please open new issue and not reopen this one 2015-12-17 22:03:14 we have to start managing issues somehow 2015-12-17 22:03:20 right now there are too many of them 2015-12-17 22:03:21 ok 2015-12-17 22:03:32 and waiting indefinitely for feedback will not help at all 2015-12-17 22:04:09 so issue was package request, package was created and user does not complain soon, then it's assumed to be working 2015-12-17 22:07:17 hmm... pkg-request means adding to testing/ , no workability assumed, unless maintainer is also testing 2015-12-17 22:07:36 testing is staging area after all. if there is clear interest in using it by users and it's working in a stable manner it can be promoted, to main or community. 2015-12-17 22:07:50 +1 2015-12-17 22:07:59 user requests package 2015-12-17 22:08:05 there is no point in linking pkg request to obtainig main/community level 2015-12-17 22:08:05 we ship it to testing 2015-12-17 22:08:16 and we are done 2015-12-17 22:08:28 users file new issue or ping us once its tested 2015-12-17 22:08:51 that way we can clean up the staging area once in a while 2015-12-17 22:09:00 and remove stuff nobody intend to use 2015-12-17 22:10:07 testing the built package is the least we can expect from requester 2015-12-17 22:16:45 maybe we should create another category for testing->main/community promotion in the future? Package promotion. with at least 2 other users stating in such issue that package works, it would be green for community at least, for instance. 2015-12-17 22:26:54 possibly 2015-12-17 22:27:08 but i dunno if i want require users to file a bug for it 2015-12-17 22:27:36 I don't think that we should do so 2015-12-17 22:27:39 but new category is a good idea 2015-12-17 22:27:48 or more like 2015-12-17 22:27:56 rethink categories is a good idea 2015-12-17 22:28:20 definitely. Package upgrade for instance would be nice. 2015-12-17 22:28:40 i'm not sure i want package upgrades as issue 2015-12-17 22:28:49 they are relatively expensive 2015-12-17 22:28:54 i mean 2015-12-17 22:28:58 filling issue is "expensive" 2015-12-17 22:29:01 takes time 2015-12-17 22:29:05 won't deny it 2015-12-17 22:29:17 and resolving it may also 2015-12-17 22:29:28 while the fix itself is very simple 2015-12-17 22:29:38 and the request too 2015-12-17 22:29:55 where was the flagging in pkgs.a.o shown? 2015-12-17 22:30:06 sent as email to maintainer 2015-12-17 22:30:27 pkgs.a.o already has that feature, maybe expand on that 2015-12-17 22:30:28 i dont know if it is possible to get a list of them 2015-12-17 22:30:33 there should be a page showing all outdated packages 2015-12-17 22:30:36 yes 2015-12-17 22:30:40 there probably is too 2015-12-17 22:30:49 clandmeter would know 2015-12-17 22:31:01 let me grep logs 2015-12-17 22:32:12 https://pkgs.alpinelinux.org/flagged 2015-12-17 22:32:42 ah nice 2015-12-17 22:32:44 but how to clear them? 2015-12-17 22:33:25 what would be nice was if user had a input of version string 2015-12-17 22:33:35 which is validated as valid version 2015-12-17 22:33:52 then we could version compare 2015-12-17 22:33:56 and autoclear 2015-12-17 22:34:22 and the upgradeing could be half automatic 2015-12-17 22:34:36 clandmeter wrote before: 2015-12-17 22:34:38 15:07 < nmeum> ah, wouldn't it make sense to remove packages from the database that have been upgraded? 2015-12-17 22:34:42 15:07 <@clandmeter> yes and no 2015-12-17 22:34:44 15:07 <@clandmeter> yes, in pkgs listing it gets reset. 2015-12-17 22:34:47 15:07 <@clandmeter> no, this way we can check which pkgs got flagged and for what reason. 2015-12-17 22:34:50 15:08 <@clandmeter> its just a reference. 2015-12-17 22:35:06 still not sure how to clear it right now ;) 2015-12-17 22:35:12 possibly only clandmeter can do it 2015-12-17 22:35:40 i need to go 2015-12-17 22:35:44 I tsill believe that it would be better to reset the flag as soon as the package gets upgraded 2015-12-17 22:35:56 I think the same 2015-12-17 22:36:07 the list gets bigger and bigger if you don't do so 2015-12-17 22:36:24 i think most of the ncurses stuff will be rebuilt and ok 2015-12-17 22:36:32 i'lld do the rest tm morning 2015-12-17 22:36:38 good night 2015-12-17 22:36:44 good night, ncopa 2015-12-17 22:37:01 good night 2015-12-17 22:37:38 still, the issues already in redmine, has to be tackled ;) lex retro non agit 2015-12-17 22:39:09 quick version upgrade is part of rolling release (fast edgy distro), and should not be considered as a bug 2015-12-17 22:39:45 exactly 2015-12-17 22:39:45 alpine is not rolling release, even though edge may seem so 2015-12-17 22:40:05 edge is rolling 2015-12-17 22:40:12 edge is, but yeah alpine is not a rolling release distro 2015-12-17 22:40:18 v3.3 would be stable 2015-12-17 22:44:31 http://lm-sensors.org/ strange, was it always that empty? 2015-12-17 22:45:32 site been down for awhile 2015-12-17 22:47:15 so where you can download their sources now? 2015-12-17 22:50:31 any new release after v3.4.0 ? 2015-12-17 22:51:26 no, just wanted to check latest i2c-tools version 2015-12-17 22:51:42 https://web.archive.org/web/20150803224145/http://www.lm-sensors.org/wiki/I2CTools 2015-12-17 22:52:20 edge is more rolling breakage 2015-12-17 22:52:59 that's the definition of living on the edge 2015-12-17 22:53:01 ;) 2015-12-17 22:53:21 that's why called edge... anything higher or tipper ;) 2015-12-17 22:54:11 maybe it should be renamed cut-throat-razor to make it more clear 2015-12-17 22:54:34 or maybe 'hoverboard' 2015-12-17 23:10:10 przemoc: http://distfiles.alpinelinux.org/distfiles/i2c-tools-3.1.2.tar.bz2 2015-12-17 23:11:50 didn't know there was distfiles.a.o 2015-12-17 23:11:54 thanks 2015-12-18 00:42:58 can anyone check whether binary from xmlsec package at least runs? #4554 2015-12-18 06:55:31 morning 2015-12-18 06:55:34 przemoc, you around? 2015-12-18 07:01:47 ncopa, apk search -r libv4l2.so returns: 2015-12-18 07:01:47 motion-3.2.12-r2 2015-12-18 07:01:48 sane-1.0.24-r1 2015-12-18 07:01:48 guvcview-2.0.1-r1 2015-12-18 07:01:48 ffmpeg-libs-2.8.3-r0 2015-12-18 07:01:48 ptlib-2.10.10-r2 2015-12-18 07:01:52 gst-plugins-good1-1.6.1-r0 2015-12-18 07:01:54 qv4l2-1.8.1-r0 2015-12-18 07:01:57 we upgraded v4l-utils to 1.8.1 2015-12-18 07:02:04 from 1.6 2015-12-18 07:02:33 do we need to rebuild those packages as well? Dunno if 1.8.x has ABI change 2015-12-18 07:13:49 i couldnt se any ABI change when i tested 2015-12-18 07:14:17 i think its ok 2015-12-18 07:17:28 k 2015-12-18 07:17:30 thx 2015-12-18 07:17:44 i'm working on https://pkgs.alpinelinux.org/flagged 2015-12-18 07:21:26 we should clean up that list 2015-12-18 07:24:04 atop new version has makefile broken 2015-12-18 07:24:10 needs patch 2015-12-18 07:24:17 if [ ! -d /var/log/atop ]; \ 2015-12-18 07:24:17 then mkdir -p /var/log/atop; fi 2015-12-18 07:24:17 mkdir: can't create directory '/var/log/atop': Permission denied 2015-12-18 07:24:28 post-pone it 2015-12-18 07:24:45 or patch it: $(DESTDIR)/var/log/atop 2015-12-18 07:24:53 yeah 2015-12-18 07:34:41 bah 2015-12-18 07:34:48 we've in main packages withou maintainer 2015-12-18 07:34:52 *without 2015-12-18 07:35:19 oh dear 2015-12-18 07:35:20 which 2015-12-18 07:35:26 mtr 2015-12-18 07:35:32 atop 2015-12-18 07:37:25 atop has ncopa listed? 2015-12-18 07:38:13 ammunta, yes, i was wrong. I'm working on: https://pkgs.alpinelinux.org/flagged 2015-12-18 07:38:21 it was letsencrypt without maintainer 2015-12-18 07:38:22 mtr looks like it was submitted with maintainer blank and never got one added when it moved to main 2015-12-18 07:38:26 and mtr 2015-12-18 07:38:34 atop has ncopa. 2015-12-18 07:38:48 letsencrypt is in main alreadY? 2015-12-18 07:38:54 nah, community 2015-12-18 07:39:02 atm only mtr 2015-12-18 07:39:19 I'm taking maintainership 2015-12-18 07:39:31 what's flagged, no maintainer? 2015-12-18 07:39:37 or 2015-12-18 07:39:45 version 2015-12-18 07:40:08 ah 2015-12-18 07:40:54 i could take mtr 2015-12-18 07:40:59 mtr is a nice tool 2015-12-18 07:41:03 ok ncopa 2015-12-18 07:41:09 but the dns implementation is bad 2015-12-18 07:41:28 yes. 0.86 won't compile because of the same issue you had with https://github.com/traviscross/mtr/issues/59 2015-12-18 07:41:42 patch needs to be refactored with the new version 2015-12-18 07:41:43 they solved that in a feature branch i think 2015-12-18 07:42:05 newdns branch 2015-12-18 07:42:34 k i'll use that branch 2015-12-18 07:42:41 version is the same 2015-12-18 07:42:46 huh. mtr seems like it has an awful build 2015-12-18 07:42:54 fcolista: ? 2015-12-18 07:42:56 I was not expecting that 2015-12-18 07:43:14 i think we want upstream merge the newdns branch and make new release 2015-12-18 07:43:17 ncopa, forget it 2015-12-18 07:43:18 yes 2015-12-18 07:43:42 so atm we can: 2015-12-18 07:43:45 1. wait 2015-12-18 07:44:00 2. fix the configure 2015-12-18 07:44:10 check the changelog, anything major in the version? 2015-12-18 07:44:20 .2 is if we want to upgrade to 0.86 2015-12-18 07:45:18 i'd file another issue 2015-12-18 07:45:19 to mtr 2015-12-18 07:45:28 sayin 0.86 does not build with musl lib 2015-12-18 07:45:38 and ask if they can merge newdns 2015-12-18 07:46:35 is there any reason newdns has been sitting there for a year 2015-12-18 07:48:32 https://github.com/openwrt/packages/pull/1425 looks like openwrt decided half a year ago that not worth waiting on 2015-12-18 07:49:45 eheh 2015-12-18 07:51:26 i'm tempted to do a nasty thing. 2015-12-18 07:52:03 diff the two branches 2015-12-18 07:52:07 produce a patch 2015-12-18 07:52:17 apply the patch 2015-12-18 07:52:26 at that point can't we just build from newdns 2015-12-18 07:52:55 newdns is hosted on github, and the zipfile has checksum issues because of timestamp 2015-12-18 07:53:08 we should download it and put on dev.a.o/mtr 2015-12-18 07:53:19 but this is not in-line with the actual package. 2015-12-18 07:53:42 which is master branch, rather newdns branch. This should be differenciated somewho imho 2015-12-18 07:53:44 *somehow 2015-12-18 07:55:29 wait, github generated tarballs change timestamp? 2015-12-18 07:55:35 yes. 2015-12-18 07:55:41 It's generated at runtime. 2015-12-18 07:56:08 And it adds timestamp, which makes the abuild fail each time it calculates the checksum. 2015-12-18 07:56:25 looks to me like the timestamps are all for the last modified time of the files 2015-12-18 07:56:39 or the revision anyway 2015-12-18 07:56:54 or is it something else 2015-12-18 07:57:11 whatever is what github adds, it makes checksum fail 2015-12-18 07:57:38 checksum seems to be fine downloading the same file three times but that might also just be caching 2015-12-18 07:58:25 yes, might be. We already faced that issue before. 2015-12-18 07:58:52 damn 2015-12-18 09:00:21 przemoc, nmeum, ncopa: the flagged url should not be reset. its not made as an part of the interface, its just to list what has ever been flagged. there needs to be a new function in listing to sort on flagged packages (or only list them). If thats added then we could probably use the /flagged uri for it. 2015-12-18 09:01:51 ACTION is around 2015-12-18 09:01:59 fcolista: ^ 2015-12-18 09:02:06 hi 2015-12-18 09:02:13 yo 2015-12-18 09:02:16 forgot why i was looking for you 2015-12-18 09:02:17 :) 2015-12-18 09:02:40 clandmeter: what we want is a list of packages that we need to update 2015-12-18 09:02:50 things that are already done is not interesting anymore 2015-12-18 09:02:55 ncopa: check your mailbox. 2015-12-18 09:02:59 and should be filtered out 2015-12-18 09:03:00 that what the functions is for 2015-12-18 09:03:09 btw: https://pkgs.alpinelinux.org/flaggedthat is not updated automatically 2015-12-18 09:03:19 i have 8964 things in my inbox 2015-12-18 09:03:21 correct and it shouldnt 2015-12-18 09:03:28 I'll ask again: can someone test whether xmlsec at least runs? ;) 2015-12-18 09:03:47 it IS updated in the regular pkg list 2015-12-18 09:03:56 the version becomes red if flagged 2015-12-18 09:03:58 green when ok 2015-12-18 09:04:07 ok 2015-12-18 09:04:19 can i get a regula pkg list with only red flags? 2015-12-18 09:04:22 there is just no way to list *only* them 2015-12-18 09:04:27 clandmeter, some of that packages has been updated but they are still "red" 2015-12-18 09:04:33 :) 2015-12-18 09:04:45 i think we finally have defined our feature request :) 2015-12-18 09:04:45 fcolista: show me 2015-12-18 09:05:02 letsencrypt show version 0.1.0 2015-12-18 09:05:06 ncopa: its a known feature :) 2015-12-18 09:05:10 while i've updated it to 0.1.1 2015-12-18 09:05:13 show me the uri please 2015-12-18 09:05:16 same for atop 2015-12-18 09:05:17 k 2015-12-18 09:05:29 oh ok 2015-12-18 09:05:34 ;-) 2015-12-18 09:05:34 i got what you meant 2015-12-18 09:05:35 https://pkgs.alpinelinux.org/packages?name=letsencrypt&arch=x86_64 2015-12-18 09:05:37 here is red 2015-12-18 09:05:40 sorry 2015-12-18 09:05:41 green 2015-12-18 09:05:44 while the list is red 2015-12-18 09:05:51 https://pkgs.alpinelinux.org/flagged 2015-12-18 09:05:57 i know you didnt get it. thats why i asked for the uri. 2015-12-18 09:05:59 its confusing 2015-12-18 09:06:03 and its my faulth 2015-12-18 09:06:05 this one imho should be updated. 2015-12-18 09:06:17 but i still want to get paid! 2015-12-18 09:06:24 ahaha 2015-12-18 09:06:28 ok 2015-12-18 09:06:29 :) 2015-12-18 09:06:40 i'm ok with the red...i love that color ;-) 2015-12-18 09:06:48 :) 2015-12-18 09:06:58 i thought italians like blue better? 2015-12-18 09:07:04 lol 2015-12-18 09:07:07 :) 2015-12-18 09:07:07 fcolista: enough chit-chat. v4l-utils doesn't look good 2015-12-18 09:07:29 przemoc, what is wrong with that? 2015-12-18 09:07:58 clandmeter, well, there are no shared preferences about color :) 2015-12-18 09:08:35 clandmeter: i have a feature request for flagging. add a entry field for version string, which will be validated, and possibly an url field for url to new source package 2015-12-18 09:09:33 if we have a valid version field, then we could automate things, eg, generate an: 'abump $pkg-$ver' line 2015-12-18 09:09:40 which could be copy/paste 2015-12-18 09:10:01 https://pkgs.alpinelinux.org/contents?pkgname=v4l-utils&arch=x86_64&repo=main - this should be some subpackage. debian names it ir-keytable, for instance. but where are command-line binaries? v4l2-ctl and so on? 2015-12-18 09:10:28 I haven't checked how old v4l-utils looked previously, though 2015-12-18 09:11:11 ncopa: please add a feature resuest on the project page. 2015-12-18 09:11:26 ok 2015-12-18 09:11:26 and anybody else that have suggestions. 2015-12-18 09:11:33 url is in footer. 2015-12-18 09:12:45 and arch has one big pile of junk with everything in v4l-utils pkg, so it's bad reference ;) 2015-12-18 09:13:29 anyone here has experiance with mongodb query syntax? 2015-12-18 09:14:30 please dont fight to answer, one by one please. 2015-12-18 09:14:53 no, but I liked ncopa comment re mongodb > worst bloat I've seen in loooong time.... 2015-12-18 09:14:56 ;) 2015-12-18 09:15:08 :) 2015-12-18 09:15:11 2 years old, though 2015-12-18 09:15:25 maybe it's got better? 2015-12-18 09:15:45 i didnt say i was working with mongodb ;-) 2015-12-18 09:16:39 im currently using sqlite 2015-12-18 09:16:52 fcolista: any idea where command-line binaries get lost in v4l-utils? 2015-12-18 09:17:13 przemoc, not really. 2015-12-18 09:17:15 Let me check. 2015-12-18 09:17:18 i was thinking of switching to ejdb 2015-12-18 09:17:31 morning 2015-12-18 09:17:32 subpackages="$pkgname-dev $pkgname-doc qv4l2 $pkgname-dvbv5" 2015-12-18 09:18:21 no, they're not there 2015-12-18 09:18:27 hm 2015-12-18 09:18:44 (I'm not sure they were built, though) 2015-12-18 09:19:01 looks like they are in v4l-utils? 2015-12-18 09:19:18 https://pkgs.alpinelinux.org/contents?pkgname=v4l-utils&arch=x86_64&repo=main ? 2015-12-18 09:19:41 apk fetch --stdout v4l-utils | tar -zt 2015-12-18 09:20:03 $ apk fetch --stdout v4l-utils | tar -zt | grep usr/bin | tpaste 2015-12-18 09:20:03 http://tpaste.us/AW5M 2015-12-18 09:20:10 ok, so web ui is unreliable? 2015-12-18 09:20:34 it probably has a max-entries 2015-12-18 09:20:40 for pagination 2015-12-18 09:20:53 send feature request ;) 2015-12-18 09:20:56 and pagination is where? :) 2015-12-18 09:21:00 i am thinking of adding an api to pkgs.a.o, so we can have apk-tools lua scripts to query contents of packages. 2015-12-18 09:21:18 that would be awesome 2015-12-18 09:21:29 yes i think so too 2015-12-18 09:21:30 :) 2015-12-18 09:21:35 so you can search which package has file foo 2015-12-18 09:21:43 yep 2015-12-18 09:21:52 and others could build on it 2015-12-18 09:22:00 I'll use it from vracker (which is not there yet, but stay tuned) 2015-12-18 09:22:43 I'm going to build super simple version tracker using free CI like travis to track version number and store it in separate repository 2015-12-18 09:23:00 upstream and distro-specific, like alpine 2015-12-18 09:23:28 the thing is, these collections are large, so i need an better way to store them then sqlite. 2015-12-18 09:23:32 all in bash, jq, xmlstarlet, html-xml-utils, etc. 2015-12-18 09:25:07 being able to query versions from pkgs.a.o and get some JSON response would be nice 2015-12-18 09:25:21 that part is pretty simple 2015-12-18 09:25:36 actually already had a simple implemetnation 2015-12-18 09:26:51 acpica makefile is also broken 2015-12-18 09:26:53 /o\ 2015-12-18 09:26:55 uff 2015-12-18 09:28:24 but simple is probably not good in an api. so i prefer to follow something like http://jsonapi.org 2015-12-18 09:29:35 fcolista: could you add subpackage in v4l-utils, so all these rc_keymaps are in it? ir-keytable will be a good name. there should be /etc/rc_maps.cfg, /lib/udev/rc_keymaps/*, /lib/udev/rules.d/*ir-keytable.rules, /usr/bin/ir-keytable, and ir-keytable man page 2015-12-18 09:30:54 so v4l-utils would have only basic command-line tools (without mentioned ir-keytable) and their man pages. this will be much cleaner 2015-12-18 09:31:54 or is anyone against such improvement and have really good reason for that? :) 2015-12-18 09:33:46 sorry, in alpine you always put manpages in -doc? if so, there should be ir-keytable-doc too? 2015-12-18 09:34:05 ir-keytable doesn't work as function name. 2015-12-18 09:34:22 it should be irkeytable 2015-12-18 09:34:38 or ir_keytable 2015-12-18 09:36:09 what's the current practice of naming packages that have dash in other distros? if you replace them with underscore, then sure, let's go with ir_keytable. (ir-keytable would be the best, though) 2015-12-18 09:36:47 yes, it would be the best. 2015-12-18 09:37:09 other distros call it as ir-keytable actually 2015-12-18 09:37:26 yes, I already wrote that 2015-12-18 09:37:42 i know, it's just for confirming :) 2015-12-18 09:38:09 you have stuff in -doc that belong to subpackages, like dvb*, so maybe let's keep all man pages in v4l-utils-doc, without additional doc packages 2015-12-18 09:38:28 not sure what are best alpine practices regarding that 2015-12-18 09:38:51 there is info on our wiki for that. 2015-12-18 09:39:31 you mean http://wiki.alpinelinux.org/wiki/APKBUILD_Reference? 2015-12-18 09:40:04 there should be more 2015-12-18 09:41:18 it says nothing about subpackages having their own binaries. I mean common use case. you install ir-keytable. want it's man page. install ir-keytable-doc. it's not there? oh, I have to install v4l-utils-doc for man page ir-keytable... 2015-12-18 09:41:28 s/it's/its/ 2015-12-18 09:41:51 s/it's not there?/it doesn't exist?!/ 2015-12-18 09:42:00 i was refering the the function names 2015-12-18 09:42:06 like we do with subpkgs 2015-12-18 09:42:49 I'm simply not sure that one doc package is always the best solution 2015-12-18 09:43:21 but I guess it's the only alpine-way of doing it nowadays 2015-12-18 09:46:13 przemoc, theoretically that's the best thing to do, as you said. But atm i'm not aware of subpck documentation so far. 2015-12-18 09:46:25 i might be wrong, though 2015-12-18 09:46:44 ok, let's leave -doc improvements for future 2015-12-18 09:47:09 i an create a subpackage called ir_keytable-doc and just move the man page there. 2015-12-18 09:47:23 but it's not worthy at the moment. 2015-12-18 09:47:40 It's better that this way is engineered in a uniform way for other use-cases imho. 2015-12-18 09:47:47 I agree 2015-12-18 09:48:21 having separate ir_keytable subpkg is worthy at the moment, though :) 2015-12-18 09:49:03 I hate bloated one main packages 2015-12-18 09:49:23 and will postpone pagination feature, because binaries will be visible on first page of v4l-utils :) 2015-12-18 09:50:17 przemoc, aports:master |Francesco Colista| main/v4l-utils: added ir_keytable subpackage | h 2015-12-18 09:50:35 thank you! 2015-12-18 09:51:57 fcolista: you have to bump pkgrel, because otherwise we still have bloated v4l-utils main pkg 2015-12-18 09:52:07 i've bumped pkgrel 2015-12-18 09:52:27 0 -> 1 2015-12-18 09:52:31 https://pkgs.alpinelinux.org/contents?pkgname=ir_keytable&arch=x86_64&repo=main 2015-12-18 09:52:46 http://git.alpinelinux.org/cgit/aports/commit/?id=cf2a4fe1 2015-12-18 09:53:02 przemoc, ^^^ 2015-12-18 09:53:06 that's the commit. 2015-12-18 09:53:31 git is more reliable ;-) 2015-12-18 09:55:51 fcolista: are you saying my app is not reliable? ;-) 2015-12-18 09:55:54 I just wanted to see the content of the _built_ package, not the intention. am I that weird with such attitude? ;) 2015-12-18 09:56:22 clandmeter, i would never say that! 2015-12-18 09:56:36 przemoc, not at all :) 2015-12-18 09:57:13 but if you want to check changes, git is the best place...isn't it ;) 2015-12-18 09:57:17 ? 2015-12-18 09:57:21 pkgs updates every /etc/periodic/15min 2015-12-18 09:57:46 oh, it isn't triggered by builds? 2015-12-18 09:57:50 that I didn't know 2015-12-18 09:57:54 nope 2015-12-18 09:59:12 its possible to do it differently, but that would need more logic. 2015-12-18 10:00:06 fcolista: yes, aports cgit page is what I have opened at least in a few tabs non-stop for a few days ;) 2015-12-18 10:00:26 maybe I should pin it 2015-12-18 10:00:37 never used that feature in chrome 2015-12-18 10:01:46 clandmeter: but package list is updated ~real-time 2015-12-18 10:03:58 fcolista: I may possibly bother you today a few times more. hope you're becoming fine in being my benign right hand. 2015-12-18 10:04:21 ;) 2015-12-18 10:06:29 fcolista: have you maybe verified that sngrep is no longer segfaulting? just to be sure it was only ncurses thing. 2015-12-18 10:08:26 ir_keytable has content now, but v4l-utils still have old list file 2015-12-18 10:08:46 no, it's almost ok 2015-12-18 10:09:21 przemoc, yes. It works with ncurses fix, as well as alsamixer. 2015-12-18 10:10:39 fcolista: there is no /etc in ir_keyremote 2015-12-18 10:10:46 yet it's in v4l-utils 2015-12-18 10:12:02 przemoc, ? 2015-12-18 10:12:07 there is. 2015-12-18 10:12:47 /etc/rc_maps.cfg should be in ir_keytable 2015-12-18 10:13:10 you make mistake in commit 2015-12-18 10:13:14 s/make/made/ 2015-12-18 10:13:15 ah 2015-12-18 10:13:15 yes 2015-12-18 10:13:51 it's often easier to spot mistake on output than in input (intentions...) 2015-12-18 10:14:08 there are some libs in v4l-utils pkg 2015-12-18 10:14:25 maybe it should be moved to other subpkg too 2015-12-18 10:16:58 or maybe let's leave it for now, as there would be more than one pkg. 2015-12-18 10:17:05 os simply go with one -libs 2015-12-18 10:17:20 and main pkg have to depend on it 2015-12-18 10:18:18 and other binary subpkgs would have to depend on it too 2015-12-18 10:19:08 are there any pkgs in alpine depending on v4l-utils (beside its subpks)? 2015-12-18 10:20:18 motion-3.2.12-r2 2015-12-18 10:20:19 sane-1.0.24-r1 2015-12-18 10:20:19 guvcview-2.0.1-r1 2015-12-18 10:20:19 ptlib-2.10.10-r2 2015-12-18 10:20:19 ffmpeg-libs-2.8.3-r0 2015-12-18 10:20:19 gst-plugins-good1-1.6.1-r0 2015-12-18 10:20:21 qv4l2-1.8.1-r1 2015-12-18 10:20:25 those ^^^^ 2015-12-18 10:20:27 oh my... 2015-12-18 10:21:26 going with -libs subpkg is the right thing to do, as these above should not depend on command-line utilities... 2015-12-18 10:21:36 but is it acceptable to rebuild them after such change now? 2015-12-18 10:21:58 (with fixed then deps on v4l-utils-libs, of course) 2015-12-18 10:22:46 i think ncopa should say that, since we're close to 3.3 release, maybe he wants to postpone that for 3.4 2015-12-18 10:22:55 but it's only an opiniong 2015-12-18 10:22:58 *opinion 2015-12-18 10:23:07 it would not take too much time doing that. 2015-12-18 10:23:41 its 680k 2015-12-18 10:23:55 does the libs need the udev rules and configs? 2015-12-18 10:24:05 what does fedora? 2015-12-18 10:24:18 udev rules are already moved out 2015-12-18 10:24:21 to ir_keyremote 2015-12-18 10:24:37 http://pkgs.fedoraproject.org/cgit/v4l-utils.git/plain/v4l-utils.spec 2015-12-18 10:24:38 and key_maps and all these stuff 2015-12-18 10:24:53 yeah 2015-12-18 10:24:56 -libs should be ok 2015-12-18 10:25:41 its a simple change 2015-12-18 10:25:55 ok, glad you're fine with that 2015-12-18 10:26:18 I don't like postponing such cleanings 2015-12-18 10:26:27 include usr/lib/libv4l dir too 2015-12-18 10:27:40 /usr/lib/libv4l/ov511-decomp doesn't seem to be lib (from the name, don't have the file here), but you want it in -libs? 2015-12-18 10:28:08 it should be libexec possibly 2015-12-18 10:28:26 ship it with -libs 2015-12-18 10:28:29 as is 2015-12-18 10:28:35 kay, you heard, fcolista 2015-12-18 10:28:37 fedora does something similar 2015-12-18 10:29:30 I don't think that fedora always does good things, but v4l-utils will be already in much saner state for 3.3 than they were before, so I'm fine with that 2015-12-18 10:29:49 agree 2015-12-18 10:31:30 przemoc, ncopa, basically you want in -libs all the content of /usr/lib ? 2015-12-18 10:31:52 yes 2015-12-18 10:31:59 k 2015-12-18 10:32:17 also v4l-utils-dev has few .so that needs to be moved imho 2015-12-18 10:32:25 ./v4l-utils-dev/usr/lib/v4l2convert.so 2015-12-18 10:32:25 ./v4l-utils-dev/usr/lib/libv4l2rds.so 2015-12-18 10:32:25 ./v4l-utils-dev/usr/lib/v4l1compat.so 2015-12-18 10:32:25 ./v4l-utils-dev/usr/lib/libv4l2.so 2015-12-18 10:32:25 ./v4l-utils-dev/usr/lib/libv4lconvert.so 2015-12-18 10:32:30 those should stay there 2015-12-18 10:32:39 these are for building only 2015-12-18 10:32:41 ok. 2015-12-18 10:32:55 built apps don't use them but so.* 2015-12-18 10:33:12 correct 2015-12-18 10:34:55 can someone check if xmlsec at least runs? PLEASE :) 2015-12-18 10:35:21 it displays help text :) 2015-12-18 10:36:01 let's say it's good enough 2015-12-18 10:36:14 ncopa: can you check if chromium runs at least? i cant verify it after the upgrade. 2015-12-18 10:36:49 we atleast ship something that kind of works. 2015-12-18 10:36:59 Chromium 47.0.2526.106 (Developer Build) (64-bit) 2015-12-18 10:37:09 clandmeter: is that correct version? 2015-12-18 10:37:18 I already marked it as done and resolved, please create new bug issue if not :) 2015-12-18 10:37:33 Version 47.0.2526.106 m 2015-12-18 10:37:38 that what i have in windows 2015-12-18 10:38:02 not sure why it states developer build 2015-12-18 10:38:32 ncopa, remember me. If i want that a package (like v4l-utils-libs) is picked as dependency from my buildenv packages, and it is not pushed on main tree, i need to add my packages repo to /etc/apk/repositories right? 2015-12-18 10:40:41 yes, if you want install it locally 2015-12-18 10:40:46 umh 2015-12-18 10:40:53 or if you want use it from testing or community builds 2015-12-18 10:40:54 WARNING: Ignoring /home/fcolista/packages/main/x86_64/APKINDEX.tar.gz: UNTRUSTED signature 2015-12-18 10:40:54 WARNING: Ignoring /home/fcolista/packages/testing/x86_64/APKINDEX.tar.gz: UNTRUSTED signature 2015-12-18 10:40:54 WARNING: Ignoring /home/fcolista/packages/community/x86_64/APKINDEX.tar.gz: UNTRUSTED signature 2015-12-18 10:41:14 do you have your abuild key installed on system? 2015-12-18 10:41:39 sudo cp ~/.abuild/*.pub /etc/apk/keys 2015-12-18 10:41:49 ah. 2015-12-18 10:41:51 right. 2015-12-18 10:42:23 clandmeter: chromium works 2015-12-18 10:42:30 but hum 2015-12-18 10:42:45 chromium will probably delay arm builder 2015-12-18 10:43:06 ok basically i've done with v4l-utils-libs. Now i'm going to add this to packages who depends from v4l-utils. 2015-12-18 10:43:13 no 2015-12-18 10:43:21 fcolista: you dont need too 2015-12-18 10:43:47 why not? 'cause is just -dev ? 2015-12-18 10:43:54 those packages should have a depend=so:libv4l.so.0 2015-12-18 10:44:11 the v4l-util-libs should get a: provides=so:libv4l.so.0 2015-12-18 10:44:23 and apk should pick the v4l-util-libs automatically 2015-12-18 10:44:37 ok, so we don't need explicit -libs dependance? 2015-12-18 10:44:41 this "provides=so:libv4l.so.0" it's done automatically? 2015-12-18 10:44:42 we dont need set depend=v4l-util-libs 2015-12-18 10:44:45 yes 2015-12-18 10:44:46 nice 2015-12-18 10:44:47 Or i should specify it to APKBUILD? 2015-12-18 10:44:48 abuild does it 2015-12-18 10:44:49 no 2015-12-18 10:44:53 oh 2015-12-18 10:44:57 sounds new to me 2015-12-18 10:45:04 no, its relatively old feature :) 2015-12-18 10:45:09 cool :) 2015-12-18 10:45:10 1-2 years :) 2015-12-18 10:45:28 so, we need to bump pkgrel only in case of abi change. 2015-12-18 10:45:37 yes 2015-12-18 10:45:39 packages needs to be rebuilt. 2015-12-18 10:45:40 Ok 2015-12-18 10:45:41 gotcha. 2015-12-18 10:45:47 abi-split, I would say ;) 2015-12-18 10:46:33 ok 2015-12-18 10:46:44 i dont think arm builder will be ready today :-( 2015-12-18 10:47:22 it's a good thing after all, one can say, more things can be fixed in the meantime :> 2015-12-18 10:47:38 ncopa: chromium will delay arm? 2015-12-18 10:47:46 isnt it already build since yesterday? 2015-12-18 10:48:26 if you make sure you always fix big things, you can postpone the release forever- and turn alpine into a rolling releases only :) 2015-12-18 10:48:37 clandmeter: it has built since yesterday 2015-12-18 10:48:48 but it is not ready with that yet 2015-12-18 10:48:57 ncopa, i think that sooner or later we should move to a rolling-release.. 2015-12-18 10:48:58 I don't intend to destroy your distro, rest assured 2015-12-18 10:49:16 fcolista: did you know we have a abuild libs subpkg helper? 2015-12-18 10:49:31 fcolista: we already have a rolling release, its called 'edge' 2015-12-18 10:49:32 clandmeter, no 2015-12-18 10:49:49 ncopa...yes...from this point of view, yes :) 2015-12-18 10:49:53 try to remove the function and see what happends ;-) 2015-12-18 10:50:03 having stable milestones is a good thing. it's reassuring to have where to go back if something doesn't work 2015-12-18 10:50:38 clandmeter: it will not copy the dir so i think we need libs function anyway 2015-12-18 10:51:43 copy which dir? 2015-12-18 10:51:56 v4l subdir in /usr/lib 2015-12-18 10:52:03 I guess 2015-12-18 10:52:16 i cannot see that from the apkbuild. 2015-12-18 10:52:27 it's mv. 2015-12-18 10:52:41 it includes the subdir 2015-12-18 10:52:43 but normally using $pkgname-libs should be enough in my experiance. 2015-12-18 10:52:48 correct 2015-12-18 10:52:51 normally its enough 2015-12-18 10:53:05 well, atleast now he knows we have it :) 2015-12-18 10:53:11 yeah :) 2015-12-18 10:53:19 in this case we also want the usr/lib/libv4l 2015-12-18 10:53:26 so yes, you both are right :) 2015-12-18 10:54:10 maybe we should bump the depending pkgs just in case, to be sure that nothing blows? 2015-12-18 10:54:21 i am about to 'kill the engineers and just ship the product' 2015-12-18 10:54:25 you guys better hide... 2015-12-18 10:54:35 no, we dont bump pkgs 2015-12-18 10:54:39 not needed 2015-12-18 10:54:48 and if things blows up, we know who to blame :) 2015-12-18 10:54:50 they were built againts 1.6.3, not 1.8.1 2015-12-18 10:55:19 the soname didnt change 2015-12-18 10:55:33 but it would be nice to test some of the apps using it though 2015-12-18 10:55:41 it should work 2015-12-18 10:55:47 but, we cannot polish forever 2015-12-18 10:55:50 we need ship 2015-12-18 10:56:05 ship without warrantry 2015-12-18 10:56:18 we can reduce the price... 2015-12-18 10:56:19 :) 2015-12-18 10:56:22 :) 2015-12-18 10:56:25 well 2015-12-18 10:56:30 what's blocking for this ship? 2015-12-18 10:56:33 thats what the release candidates are for: testing 2015-12-18 10:56:38 the arm builder 2015-12-18 10:56:42 I know, but it's in main, and rebuilding this 5 or 6 pkgs is not big deal, yet will make us sure it's at least building :) 2015-12-18 10:57:02 przemoc: feel free to tes build them locally 2015-12-18 10:57:07 test* 2015-12-18 10:57:09 hehe 2015-12-18 10:57:14 and report if something breaks 2015-12-18 10:57:35 he will first need to install it somewher 2015-12-18 10:57:53 i am serious. if we dont ship today, we will probably not ship til next year 2015-12-18 10:58:02 next blocker is release notes 2015-12-18 10:58:09 ship it 2015-12-18 10:58:10 draft it 2015-12-18 10:58:16 yes 2015-12-18 10:58:48 ncopa: there are some issues I would like you to look at whether you're going to do anything with them 2015-12-18 10:58:54 so what else changed except initramfs? 2015-12-18 10:59:00 what's your view about externally build modules? 2015-12-18 10:59:09 clandmeter: gcc 5 libstdc++ ABI breakage 2015-12-18 10:59:10 #4427 2015-12-18 10:59:37 przemoc: push to 3.4.0 2015-12-18 10:59:40 or possible to 3.3.1 2015-12-18 10:59:44 if you think its important 2015-12-18 10:59:51 that kind of things we need patches for... 2015-12-18 11:00:25 i dont have time to deal with it 2015-12-18 11:00:37 and ideally, the dirver should go to mainline linux 2015-12-18 11:00:49 normally when mainline dont include a driver or patch 2015-12-18 11:00:51 I'm asking whether it's possible at all to include such external kernel module in future in alpine or not 2015-12-18 11:00:53 they have good reasons to 2015-12-18 11:01:00 it is 2015-12-18 11:01:00 yes, users should buy hardware that is support, not let us fix it for them. 2015-12-18 11:01:15 so maybe close it as WONT-FIX? ;) 2015-12-18 11:01:18 supported.. 2015-12-18 11:01:26 przemoc: we have some supported 3rd party kernel modules 2015-12-18 11:01:33 not much IIRC 2015-12-18 11:01:34 but we try keep those down 2015-12-18 11:01:39 close it with FIXITYOURSELF :) 2015-12-18 11:01:40 yes, I understand 2015-12-18 11:02:05 one person cannot fix everything. 2015-12-18 11:02:13 we need more ppl to do such kind of things. 2015-12-18 11:02:17 i dont know if we keep it or close it 2015-12-18 11:02:30 but i agree, we need start close things... 2015-12-18 11:02:32 ok, postponed for now and removed target 2015-12-18 11:02:35 ok 2015-12-18 11:02:44 thanks 2015-12-18 11:03:08 and will remove targets from other 3.3 requests stuff that is not included in alpine already 2015-12-18 11:03:31 there is no point in postponing something for n straight releases... 2015-12-18 11:03:45 sounds good 2015-12-18 11:03:48 many thanks! 2015-12-18 11:03:55 realtek wifi chipsets are *supposed* to be supported 2015-12-18 11:03:55 im super happy for this help 2015-12-18 11:04:02 it's just that the mainline driver doesn't work 2015-12-18 11:04:08 heh 2015-12-18 11:04:17 so you can't blame that one on users :) 2015-12-18 11:05:10 i go for lunch and if arm is not ready then, well.... you better hide... :) 2015-12-18 11:05:35 he'll blame me. > that's why we haven't fixed it, przemoc removed the target, I didn't know it's important 2015-12-18 11:05:57 ncopa: you'll botnet our machines to distcc qemu arm build ? 2015-12-18 11:06:38 has anyone used texlive in alpine? 2015-12-18 11:06:59 przemoc, i've started to package it 2015-12-18 11:07:02 #3920 2015-12-18 11:07:02 it's a nightmare 2015-12-18 11:10:25 who is Isaac, is he hanging out here? 2015-12-18 11:11:32 fcolista: it doesn't look like we can fix it for 3.3, right? 2015-12-18 11:11:43 przemoc, i gave up 2015-12-18 11:11:55 too much time to spend on it 2015-12-18 11:11:56 i tried to package texlive, but you really need a good understanding of *tex to package it properly, and i'm just not a *tex user 2015-12-18 11:12:04 exactly. 2015-12-18 11:12:11 I'm not a tex user too 2015-12-18 11:12:14 I was TeX user 2015-12-18 11:12:51 ok, we'll try to fight with it for 3.4 2015-12-18 11:13:25 can we assign #3920 to przemoc :) 2015-12-18 11:13:31 not yet 2015-12-18 11:13:34 :) 2015-12-18 11:13:36 I don't have alpine yet 2015-12-18 11:13:39 after releasing 3.3 we'll create bug for that 2015-12-18 11:18:56 not having alpine is a bug indeed 2015-12-18 11:20:12 I'm just very smart. fixing alpine before starting using it. :> 99.999% people go the wrong way. 2015-12-18 11:20:28 lol 2015-12-18 11:21:14 (does alpine has >= 100K users?) 2015-12-18 11:21:24 that's a clever perspective, I like it 2015-12-18 11:21:49 a bit tiresome, but everybody wins 2015-12-18 11:23:05 with a pain in heart, I closed texlive package request 2015-12-18 11:23:38 why? 2015-12-18 11:25:04 because package request per se has been fulfilled in the past. it's just somewhat broken, and as I stated before, we'll work on it after 3.3 release, there will be bug issue and we'll try to solve it then 2015-12-18 11:28:01 ncopa: have uefi things been improved for 3.3? #4424 2015-12-18 11:28:18 ok, but package was built wrongly. not even close to being a texlive package. it might as well have been empty and should probably be remove from the repos (it doesn't meet any standard). I think it would have been better to change the target for that package request 2015-12-18 11:28:54 no offence to the original packager of course 2015-12-18 11:29:43 I'm asking, because sometimes things get fixed, but have no trace in redmine 2015-12-18 11:32:19 ScrumpyJack: actually you convinced me, but I cannot unclose it ;) 2015-12-18 11:32:29 :) 2015-12-18 11:32:32 just having the package doesn't justify closing the request… it has to be working 2015-12-18 11:32:42 right. 2015-12-18 11:33:27 ok, in that case there was feedback, so sorry, but in most cases there is no feedback and we cannot keep package request issues because of that forever. 2015-12-18 11:33:32 ncopa agreed with that 2015-12-18 11:33:58 we have to bring down number of issues 2015-12-18 11:34:23 przemoc: there is no commercial pressure here to do that :) 2015-12-18 11:34:44 there is pressure to make in manageable 2015-12-18 11:34:49 it's NOT as it is now 2015-12-18 11:35:19 most alpine devs don't even look at issues 2015-12-18 11:35:25 that's how it looks from the outside 2015-12-18 11:35:43 fair point 2015-12-18 11:35:45 well, sure 2015-12-18 11:35:47 and why? because there are too many of them 2015-12-18 11:36:05 and you're lazy too, but that part I understand actually ;) 2015-12-18 11:36:07 mostly because we don't have anyone to hunt packagers to actually handle these 2015-12-18 11:36:31 and ncopa doesn't count in that, because he's already holding at least 60% of distro on his back 2015-12-18 11:36:57 if I were constantly bothered by bug wrangler to fix my stuff, I would 2015-12-18 11:38:19 so you'll possibly be after 3.3 release 2015-12-18 11:38:21 and others too 2015-12-18 11:38:55 RE: texlive we should move that package to unmaintained it's not even close to a decent texlive package and nobody is working on it 2015-12-18 11:39:21 I agree if it's really that bad 2015-12-18 11:39:38 can we just axe it 2015-12-18 11:40:12 because it may need to be done from scratch anyway? 2015-12-18 11:40:15 that's an option as well but moving it to unmaintained will hopefully stop people from filling issues against it 2015-12-18 11:40:22 yeah 2015-12-18 11:42:40 so who'll do the honors? 2015-12-18 11:44:05 I just moved it 2015-12-18 11:44:25 thanks 2015-12-18 11:44:47 FTFY s/from/to/ :) 2015-12-18 11:44:56 yeah, I fucked that up 2015-12-18 11:45:14 actually the commit message was supposed to be unmaintained/texlive: move from testing 2015-12-18 11:45:18 (sorry) 2015-12-18 11:47:21 ah, well, the long message explains correctly 2015-12-18 11:47:46 (and surely {testing => unmaintained}/texlive/APKBUILD also) 2015-12-18 11:50:07 yeah. Hopefully people will stop sending me emails about this package now… 2015-12-18 11:53:59 I suspect at least *one* of those is unlikely to stop but 2015-12-18 11:55:10 do you mean James? ;) 2015-12-18 12:00:44 barthalion: is upgrading mongodb to 3.2 a big deal? are there any rdeps? 2015-12-18 12:01:09 przemoc: uefi situation has improved yes 2015-12-18 12:01:31 sould be fully possible now 2015-12-18 12:01:37 but i havent tested it 2015-12-18 12:01:51 well, i boot alpine with uefi on my laptop... 2015-12-18 12:02:21 also isohybrid works now 2015-12-18 12:02:27 ISO work with uefi too, right? will you deal with the mentioned issue? 2015-12-18 12:02:41 push it to 3.4.0 2015-12-18 12:02:50 i can comment 2015-12-18 12:02:50 k 2015-12-18 12:02:54 hang on 2015-12-18 12:02:56 i'll do it 2015-12-18 12:03:55 done 2015-12-18 12:04:30 thx 2015-12-18 12:05:02 anyone know why partclone was moved to unmaintained? was it in such a bad shape as texlive? 2015-12-18 12:12:14 barthalion: another thing. http://bugs.alpinelinux.org/issues/2448#note-6 can you ship it quickly or no? 2015-12-18 12:13:46 fabled: did you look into strace 4.10 maybe? 2015-12-18 12:15:01 przemoc, not yet 2015-12-18 12:15:25 is there any chance you look into it today? 2015-12-18 12:15:37 realistically 2015-12-18 12:15:55 i can take a quick look. if it's simple then it's doable. 2015-12-18 12:16:55 ok, if you fail, then I'll set it to target as 3.3.1, if ncopa won't be against 2015-12-18 12:19:31 there's a bug about it? 2015-12-18 12:20:31 no 2015-12-18 12:20:36 its a simple upgrade to 4.10 2015-12-18 12:20:56 actually there is 2015-12-18 12:21:04 #4925 2015-12-18 12:22:24 btw 2015-12-18 12:22:28 did anyone test mariadb? 2015-12-18 12:22:34 there are many bug fixes, that's why having it at least in 3.3.1 would be nice 2015-12-18 12:22:40 i think init.d script refactor broke things 2015-12-18 12:23:58 is the setup-alpine script ok now? 2015-12-18 12:24:00 and setup-disk? 2015-12-18 12:24:02 and setup-apkrepos? 2015-12-18 12:24:04 ncopa: if fabled won't manage to make strace before 3.3, do you accept targetting it for 3.3.1? 2015-12-18 12:24:11 probably 2015-12-18 12:26:08 this strace patch works for me: http://skarnet.org/tmp/strace-4.10-musl.patch.txt 2015-12-18 12:26:29 I can cross-compile strace for ARM too. 2015-12-18 12:26:47 fabled: ^^^ 2015-12-18 12:27:33 ok. i'm on it 2015-12-18 12:28:05 barthalion: if mongodb 3.2 is too much to handle, then please at least try last 3.0.x, i.e. 3.0.8, should be simple pkgver change I hope 2015-12-18 12:28:38 and please say whether you're alive or I'm writing to /dev/null ;) 2015-12-18 12:29:24 skarnet, thanks 2015-12-18 12:29:38 przemoc, i just complete the arm test build, and push strace update 2015-12-18 12:29:52 great 2015-12-18 12:29:55 can anyone test that apk can actually use the australia mirror? %1256 2015-12-18 12:30:03 np, credit to someone on #musl (I don't remember whom exactly) who sent me that patch 2015-12-18 12:30:24 probably void linux folks 2015-12-18 12:30:36 skarnet, the patch contains credits 2015-12-18 12:30:49 :) 2015-12-18 12:30:53 possibly, I remember suggesting looking into void patches earlier too 2015-12-18 12:31:00 seems it's upstream commit hcerry-picks 2015-12-18 12:31:08 credits for the authors of the patch, yes, but not for the full chain that brought it to you :) 2015-12-18 12:31:47 fabled: you're still pushing it, right? as I don't see it in aports 2015-12-18 12:32:37 just pushed 2015-12-18 12:32:48 k 2015-12-18 12:35:14 does not look like apk works with https repos 2015-12-18 12:35:39 fcolista: barthalion seems inactive, despite complaining earlier about closing issues :), so could you check that mongodb builds fine with pkgver bump to 3.0.8 and pushed it if so? and if barthalion shows up he may try to bring it to 3.2 if he manage to do it quickly 2015-12-18 12:35:47 hey guys :) 2015-12-18 12:36:10 hi leo-unglaub 2015-12-18 12:36:13 s/pushed/push/ 2015-12-18 12:40:26 hm 2015-12-18 12:40:29 we need update ruby 2015-12-18 12:40:31 nightmare 2015-12-18 12:40:34 https://www.ruby-lang.org/en/news/2015/12/16/unsafe-tainted-string-usage-in-fiddle-and-dl-cve-2015-7551/ 2015-12-18 12:41:09 sorry guys, I have to vanish for an hour or two, and later again 2015-12-18 12:41:20 there are still issues to tackle with, unfortunately 2015-12-18 12:41:21 thanks!!! 2015-12-18 12:43:09 ttp://bugs.alpinelinux.org/projects/alpine/issues?c%5B%5D=project&c%5B%5D=tracker&c%5B%5D=status&c%5B%5D=fixed_version&c%5B%5D=done_ratio&c%5B%5D=priority&c%5B%5D=subject&c%5B%5D=assigned_to&c%5B%5D=updated_on&f%5B%5D=status_id&f%5B%5D=fixed_version_id&f%5B%5D=done_ratio&f%5B%5D=&group_by=&op%5Bdone_ratio%5D=%3C%3D&op%5Bfixed_version_id%5D=%3D&op%5Bstatus_id%5D=o&per_page=100&set_filter=1&utf8=%E2%9C 2015-12-18 12:43:14 %93&v%5Bdone_ratio%5D%5B%5D=99&v%5Bfixed_version_id%5D%5B%5D=96 2015-12-18 12:45:06 ncopa: maybe you could postpone 3.3 release for monday at least? 2015-12-18 12:45:16 doing it today seems rushed 2015-12-18 12:45:40 im unavailable all next week 2015-12-18 12:45:55 and the week after that i will shut down my office here and physcally move 2015-12-18 12:46:06 so then it'll be 2016... 2015-12-18 12:46:22 ok, that settles it 2015-12-18 12:46:25 we are already 1.5 month late with the release 2015-12-18 12:46:54 we'll do 3.4 much better 2015-12-18 12:47:04 I promise, because I'll be involved earlier ;) 2015-12-18 12:47:10 so will I 2015-12-18 12:47:13 if you stay til v3.4 then yes... 2015-12-18 12:47:31 but then again, I'm sure that "we'll do 3.3 much better" is what everyone said when releasing 3.2 2015-12-18 12:47:36 lol 2015-12-18 12:47:40 exactly :) 2015-12-18 12:47:43 well 2015-12-18 12:47:52 i have a "valid" excuse 2015-12-18 12:48:00 i lost sponsor/job 2015-12-18 12:48:05 have gotten a new job 2015-12-18 12:48:13 I don't think anybody's blaming you for anything, ncopa 2015-12-18 12:48:14 no one blames you, ncopa 2015-12-18 12:48:20 i do :) 2015-12-18 12:54:51 my right hand fcolista is inactive too, maybe I'll adopt some left hand. anyone willing to check if bumping mongodb from 3.0.4 to 3.0.8 builds fine and push it? 2015-12-18 12:55:18 przemoc, working on it 2015-12-18 12:55:23 oh, great 2015-12-18 12:55:42 if anyone is starting working on something mentioned by other one, please say so 2015-12-18 12:57:05 we will introduce such regime before 3.4, that it will be best release ever ;) 2015-12-18 13:02:14 ncopa: #3937 you didn't finish it in the end, right? 2015-12-18 13:02:21 przemoc: I'm at work, nobody wanted me to pay for packaging unfortunately 2015-12-18 13:02:32 minor releases of mongo should be fine 2015-12-18 13:02:45 przemoc: correct 2015-12-18 13:03:22 we need to mention the community repo in release notes 2015-12-18 13:03:23 oh 2015-12-18 13:03:39 setup-apkrepos should add a commented community url 2015-12-18 13:04:23 ok, really vanishing now. sorry, didn't manage to retarget/comment/etc. all issues 2015-12-18 13:04:30 ncopa: just trying latest rc, but gettting bootloop after installing. 2015-12-18 13:05:26 very early in kernel loading 2015-12-18 13:05:34 weird it didnt happen in boot media 2015-12-18 13:05:47 what hardware? 2015-12-18 13:05:52 which rc? 2015-12-18 13:06:46 RC3 2015-12-18 13:07:00 pentium notebook, pretty old 2015-12-18 13:07:10 with nvidia gfx 2015-12-18 13:07:25 let me try vanilla 2015-12-18 13:07:53 x86? 2015-12-18 13:07:56 hybrid iso is working nicely. greatwork 2015-12-18 13:08:02 x86_64 2015-12-18 13:11:19 dmesg http://tpaste.us/3gbv 2015-12-18 13:11:56 [ 89.726235] usb-storage 2-2:1.0: USB Mass Storage device detected 2015-12-18 13:11:56 [ 89.726339] scsi host8: usb-storage 2-2:1.0 2015-12-18 13:11:56 [ 89.730682] blk_update_request: I/O error, dev loop0, sector 90552 2015-12-18 13:11:56 [ 89.730699] blk_update_request: I/O error, dev loop0, sector 90554 2015-12-18 13:12:08 sounds like bad disk? 2015-12-18 13:12:27 it boots fine from sdcard via usb 2015-12-18 13:12:32 i can install it correctly 2015-12-18 13:12:41 when i reboot it fails to boot from hd 2015-12-18 13:13:24 how do we explain about the alpine iso image renames? 2015-12-18 13:13:28 for release notes 2015-12-18 13:15:47 ncopa: is there a way i can debug such early boot failure? 2015-12-18 13:16:02 the paste shows squashfs 2015-12-18 13:16:06 did you install it to disk? 2015-12-18 13:16:07 and why does it work via usb, but not via sata.. 2015-12-18 13:16:08 s 2015-12-18 13:16:12 sys? 2015-12-18 13:16:15 yes 2015-12-18 13:16:24 why does it try open modloop then? 2015-12-18 13:16:50 it should not try squashfs 2015-12-18 13:17:21 because its a dmesg of usb boot? 2015-12-18 13:17:32 :) 2015-12-18 13:17:32 i cannot boot from hd, so there is no way to get a dmesg 2015-12-18 13:17:37 ok 2015-12-18 13:17:51 without silent, i only see 3 lines 2015-12-18 13:17:52 can you boot with 'single' as boot option? 2015-12-18 13:18:02 just single? 2015-12-18 13:18:07 grsec single 2015-12-18 13:18:12 or '1' 2015-12-18 13:18:20 should give you initramfs 2015-12-18 13:18:22 single cpu? 2015-12-18 13:18:23 do you use lvm? 2015-12-18 13:18:28 no 2015-12-18 13:18:28 mdadm? 2015-12-18 13:18:30 regular install 2015-12-18 13:18:39 just sys sda 2015-12-18 13:18:45 hm 2015-12-18 13:18:55 i tested it in qemu i think 2015-12-18 13:19:01 i dont know what goes wrong 2015-12-18 13:19:52 should there be a *dot* infront of linux in boot line? 2015-12-18 13:22:21 ok as release notes? http://tpaste.us/Ae65 2015-12-18 13:22:30 anything else tht should be mentioned? 2015-12-18 13:24:00 "Happy Holidays" 2015-12-18 13:24:54 (i.e. mention that release is a Christmas present, while keeping it agnostic) 2015-12-18 13:26:44 i think is those release notes are fine 2015-12-18 13:26:49 *that 2015-12-18 13:27:13 i thikn i forgot something 2015-12-18 13:27:17 but i dont know what :) 2015-12-18 13:27:22 eheh 2015-12-18 13:27:28 i have the feeling of forgotten something 2015-12-18 13:27:39 i've the same feeling...was thinking something related to arm 2015-12-18 13:27:48 but can't figure out :) 2015-12-18 13:31:14 were there any other major upgrades that might require attention while upgrading? 2015-12-18 13:32:33 ncopa: i just installed to a usb drive, it has the same issues. 2015-12-18 13:32:34 perl 5.22 2015-12-18 13:32:42 hum 2015-12-18 13:32:48 after installing, reboot bootloop 2015-12-18 13:32:57 sounds bad 2015-12-18 13:34:35 i just verified, it works on another system. 2015-12-18 13:34:42 so its system specific 2015-12-18 13:34:50 do you think you can track it down? 2015-12-18 13:35:01 other linux and windows versions work just fine on this laptop 2015-12-18 13:35:06 does it happen after initramfs? 2015-12-18 13:35:21 what is weird is that you can boot from usb 2015-12-18 13:35:27 with silent disabled i only see 3 lines. 2015-12-18 13:35:38 it is 2015-12-18 13:35:50 it alomst feels like i try to boot wrong arch 2015-12-18 13:36:32 3 lines of kernel i mean 2015-12-18 13:36:34 not init. 2015-12-18 13:36:47 so it fails loading the kernel or something. 2015-12-18 13:37:04 is syslinux installed correctly? 2015-12-18 13:37:24 it boots ok on another system 2015-12-18 13:37:24 it should be. hm 2015-12-18 13:37:39 i dont know what could be wrong 2015-12-18 13:37:45 do you think you can track it down? 2015-12-18 13:38:16 if you add 'sinle' to boot option you should get emergency shell from initramfs 2015-12-18 13:38:22 before disk is mounted 2015-12-18 13:39:24 the kernel doesnt load 2015-12-18 13:39:33 so i dont get to initramfs 2015-12-18 13:39:43 ok 2015-12-18 13:40:00 so it has to be syslinux/extlinux 2015-12-18 13:42:12 it loads only the first part 2015-12-18 13:42:17 so i dont think its the bootloader 2015-12-18 13:42:23 i would say its the kernel 2015-12-18 13:42:36 but i guess it installs the same kernel as it boots? 2015-12-18 13:42:54 i give up. its not that important. just contiue with your release. 2015-12-18 13:43:05 ill install ubuntu 2015-12-18 13:43:08 ACTION hides 2015-12-18 13:44:00 ACTION sends a pie to clandmeter's head 2015-12-18 13:46:50 ACTION throws a two-headed chicken at clandmeter 2015-12-18 13:47:12 pie and chicken hmm 2015-12-18 13:47:18 http://wwwtest.alpinelinux.org/posts/Alpine-3.3.0-released.html 2015-12-18 13:47:53 "the ISO images has been renamed" 2015-12-18 13:47:56 -> have 2015-12-18 13:48:38 New repository named, "community" -> remove the comma 2015-12-18 13:48:51 or use a colon and no quotes 2015-12-18 13:49:26 The full lists of changes -> The full list of changes 2015-12-18 13:50:39 Also, will the "commit statistics" part be filled in? 2015-12-18 13:50:47 yes 2015-12-18 13:51:14 i'm afraid to add the commit stats now 2015-12-18 13:51:21 as i may forget update it later 2015-12-18 13:51:41 is that builder still working? 2015-12-18 13:51:49 add them as the last operation before hitting the big release button :) 2015-12-18 13:52:42 refresh 2015-12-18 13:52:48 http://wwwtest.alpinelinux.org/posts/Alpine-3.3.0-released.html 2015-12-18 13:53:02 really weird, installing latest stable does work on this notebook. 2015-12-18 13:53:16 it could be syslinux or something 2015-12-18 13:53:19 typos not fixed in the refresh 2015-12-18 13:53:31 or new sfdisk... 2015-12-18 13:54:25 i get types fixed, with ctrl-r in firefox 2015-12-18 13:54:47 maybe i would make new image names 2015-12-18 13:55:44 now they're fixed for me, but it took time 2015-12-18 13:55:59 maybe my ISP is ninja proxying me -_- 2015-12-18 13:56:12 blame nsa... 2015-12-18 13:56:25 https wwwtest please 2015-12-18 13:56:45 not today 2015-12-18 13:56:59 ;) 2015-12-18 13:57:27 alpine-rpi is not iso, it's tar.gz 2015-12-18 13:57:33 and we have alpine-uboot for arm too 2015-12-18 13:57:36 so include both or neither 2015-12-18 13:57:40 LGTM presentation-wise 2015-12-18 13:58:05 and rpi2? 2015-12-18 13:58:18 fabled: how do you even use a tar.gz to install 2015-12-18 13:58:23 ncopa: i just upgraded the booted stable to 3.3 and boom bootloop. 2015-12-18 13:58:39 its kernel thing then 2015-12-18 13:58:41 skarnet, on arm it's just extracted to vfat (or ext4 on uboot). bootloader installation is board specific. 2015-12-18 13:58:43 so i think its something with the kernel 2015-12-18 13:58:53 clandmeter: and linux-vanilla? 2015-12-18 13:58:56 hm 2015-12-18 13:58:58 using it 2015-12-18 13:59:14 I'd love a way to boot Alpine on a qemu-system-arm 2015-12-18 13:59:17 report it upstream :) 2015-12-18 13:59:24 skarnet, it works now 2015-12-18 13:59:34 i have qemu-system-arm line that will work with latest images 2015-12-18 13:59:48 i'm on mobile. fcolista, how is mongodb progressing? 2015-12-18 13:59:52 we had kernel config issue that broke things. but it got fixed or _rc3 iirc 2015-12-18 14:00:08 though, the initramrd might have still issue in it. the next tag is expected to fix all issues. 2015-12-18 14:00:10 przemoc, no. 2015-12-18 14:00:14 fabled: great, please mention the line somewhere in the docs. Would be a nice item to have in the release notes, I can't be the only one interested in this 2015-12-18 14:00:30 got an error : 2015-12-18 14:00:30 : In member function 'void mongo::WiredTigerSessionCache::waitUntilDurable(mongo::WiredTigerSession*)': 2015-12-18 14:00:38 which block the build 2015-12-18 14:00:42 3.0.8 does not build? 2015-12-18 14:01:16 skarnet, something like: qemu-system-arm -kernel boot/vmlinuz-grsec -initrd boot/initramfs-grsec -dtb boot/dtbs/vexpress-v2p-ca9.dtb -M vexpress-a9 -cpu cortex-a9 -serial stdio -m 1G -append "console=ttyAMA0" 2015-12-18 14:01:26 no przemoc does not build. 2015-12-18 14:01:44 try 3.0.6 then, if you can 2015-12-18 14:01:51 http://wwwtest.alpinelinux.org/ 2015-12-18 14:01:59 i mean 2015-12-18 14:02:01 http://wwwtest.alpinelinux.org/posts/Alpine-3.3.0-released.html 2015-12-18 14:02:09 though, that will give you emergency shell since boot media is not exported as sd/mmc or anything else 2015-12-18 14:02:12 fabled: so the new vmlinuz and initramfs work with vexpress-a9? that's cool 2015-12-18 14:02:24 skarnet, or a15 2015-12-18 14:02:40 (I'm working on a9 atm) 2015-12-18 14:02:41 skarnet, the issue we had was that i accidentally had left wandboard earlydebug on 2015-12-18 14:02:56 build-3-3-armhf is still building linux-rpi 2015-12-18 14:02:59 so the uboot image was wandboard specific even though it was generic build 2015-12-18 14:03:08 in order to properly mount the boot media I should edit the initramfs? 2015-12-18 14:03:11 ncopa, is it on rpi or rp2 kernel build now? 2015-12-18 14:03:18 why do ppl like to have snow on their website i dont know. 2015-12-18 14:03:22 make -C /home/buildozer/aports/main/linux-rpi/src/build-rpi 2015-12-18 14:03:38 skarnet, no, it's just preparing proper disk image with the .tar.gz contents, and adding that as mmc disk to the image 2015-12-18 14:03:45 clandmeter: agree. i prefer RAIN! 2015-12-18 14:04:22 come to NL we have enough of that . 2015-12-18 14:04:36 fabled: so why do you get an emergency shell? what's going wrong? 2015-12-18 14:04:38 skarnet, so prepare disk image, and add "-sd mmc.img" or similar 2015-12-18 14:04:46 ah 2015-12-18 14:04:51 you need an image 2015-12-18 14:04:54 yes 2015-12-18 14:05:15 qemu loads kernel+initram+dtb directly from host FS to target's memory 2015-12-18 14:05:32 to make initramfs load the system you need to have bootmedia attached to target 2015-12-18 14:05:43 yup, I know that, but what I'm looking for is a way to have a base Alpine image as /dev/mmcblk0p1 :P 2015-12-18 14:06:02 make disk image, partition it 2015-12-18 14:06:07 and "-sd foo.img" 2015-12-18 14:06:09 to qemu 2015-12-18 14:06:21 yes, but how do I get the Alpine baselayout on the image? 2015-12-18 14:06:47 fdisk; mkfs.vfat; mount -o loop; tar xzvf alpine-image; umount; qemu 2015-12-18 14:06:53 looks that 3.0.6 mongodb with patch from 3.0.8 is building.. 2015-12-18 14:07:50 oh, mount -o loop might need magic to calculate the partitions offset 2015-12-18 14:07:51 fabled: so that alpine-image.tar.gz contains all the layout I need to run apk and stuff? 2015-12-18 14:07:55 yes 2015-12-18 14:08:10 you just extract the uboot image and it's good 2015-12-18 14:08:35 cool. This needs to be automated for 3.4 though 2015-12-18 14:08:37 the only special step is for real hardware where the uboot bootloader needs to be installed with specific file name, or specific offset on the disk 2015-12-18 14:09:08 fcolista: lgtm then, thx 2015-12-18 14:09:53 we'll nag barthalion to work on it after 3.3 rel ;) 2015-12-18 14:10:43 ncopa, anything else you need before release? 2015-12-18 14:10:53 i dont think so 2015-12-18 14:10:57 hum 2015-12-18 14:11:06 possibly check why inkscape segfault 2015-12-18 14:11:06 anyway, mongodb is at 3.2.0 2015-12-18 14:11:17 not 3.0.8 2015-12-18 14:11:31 well... i have ~30mins and then i'm off to see Luke using the source. 2015-12-18 14:11:48 ok 2015-12-18 14:11:52 dang, I go on Monday 2015-12-18 14:12:06 someone already spolilered on alpine-linux channel 2015-12-18 14:12:11 also on #musl 2015-12-18 14:12:20 some idiot must've been doing that on all channels 2015-12-18 14:12:26 the Internet is a wild place 2015-12-18 14:12:39 I should see it ASAP to avoid spoilers 2015-12-18 14:12:54 (I averted my eyes just in time on #musl) 2015-12-18 14:13:07 i think i'm only one in the channel who never watched star wars saga. 2015-12-18 14:13:19 someone wants chicken and pie 2015-12-18 14:13:50 ncopa, how to reproduce inkscape? 2015-12-18 14:13:50 dualhead chicken and quadcore pie 2015-12-18 14:14:02 ah. just run it? 2015-12-18 14:14:05 fabled: apk add inkscape && inkscape 2015-12-18 14:14:10 ncopa, how to reproduce inkscape? 2015-12-18 14:14:10 dualhead chicken and quadcore pie 2015-12-18 14:14:11 fcolista: https://www.youtube.com/watch?v=oBQVvEMc-VQ there 2015-12-18 14:14:18 i go tomorrow on sw 2015-12-18 14:14:57 fcolista: .0.6 is still better than .0.4 2015-12-18 14:15:15 przemoc, why not 3.2 ? 2015-12-18 14:15:40 because you cannot build it? 2015-12-18 14:16:05 ncopa: re building on arm, I tried grouping together some scaleway C1s to use distcc but it doesn't work very well 2015-12-18 14:16:06 i've tried 3.0.8 2015-12-18 14:16:09 not 3.2 2015-12-18 14:16:15 you asked 3.0.8 2015-12-18 14:16:25 i thought it was the last version. 2015-12-18 14:16:47 ScrumpyJack: i have 2 wandboards but only one runs atm 2015-12-18 14:16:47 because it seemed easier for obv reason 2015-12-18 14:16:59 distcc tend to complicate things 2015-12-18 14:18:38 and it failed, so 3.2 will likely fail too, but if you can test and want to check, then please do 2015-12-18 14:19:34 umh. 2015-12-18 14:19:50 3.0 -> 3.2 might include patches that we need to apply now 2015-12-18 14:19:57 anyway 3.0.6 is still building 2015-12-18 14:26:46 the link to the git logs is broken in that announcment 2015-12-18 14:27:23 s/the first release in v3.3 stable series/the first stable release in v3.3 series 2015-12-18 14:27:51 ncopa, not sure if i get inkscape rebuilt with debug info in time 2015-12-18 14:27:56 how long it compiles normally? 2015-12-18 14:27:58 s/Current images are now/The|Our current images are now 2015-12-18 14:28:55 s/slightly more packages available in the repository/slightly more packages available in the download|image 2015-12-18 14:29:36 s/same as before/No change/g 2015-12-18 14:30:48 mongodb bison has been build. Pushed. 2015-12-18 14:30:53 *built. 2015-12-18 14:31:00 s/Some of the desktop applications that got upgraded and are available for v3.3:/Some of the upgraded desktop applications are: 2015-12-18 14:31:08 Now i'm gonna try 3.2 2015-12-18 14:31:28 anything I can do? I have a few hours ahead of me 2015-12-18 14:40:00 fcolista: .0.8 has been released later than .2.0, but maybe it wiil build fine. thx for trying. 2015-12-18 14:52:32 fabled: http://tpaste.us/2knz 2015-12-18 14:54:43 up 2015-12-18 14:54:58 (gdb) up 2015-12-18 14:54:58 #1 0x000070fccedecea8 in __strcoll_l (loc=, 2015-12-18 14:54:58 r=0x740b11636c0 "sans-serif", l=0x0) at src/locale/strcoll.c:8 2015-12-18 14:54:58 8 src/locale/strcoll.c: No such file or directory. 2015-12-18 14:55:34 tries to strcmp null 2015-12-18 15:00:10 yes, looks like it's some setlocale() / glib / musl thing 2015-12-18 15:00:18 or just unconvertable text 2015-12-18 15:00:21 i need to run 2015-12-18 15:00:27 ok 2015-12-18 15:00:29 thanks 2015-12-18 15:02:53 hm 2015-12-18 15:02:59 its building rpi2 kernel now 2015-12-18 15:36:32 ACTION is back 2015-12-18 15:37:14 I'll try to go through some remaining issues till I leave again for good 2015-12-18 15:37:31 and I'm not going tomorrow on star wars, but today :P 2015-12-18 15:43:36 fcolista: building mongodb 3.2 failed? 2015-12-18 15:43:43 yes 2015-12-18 15:43:57 i don't have time now to check 2015-12-18 15:44:03 np 2015-12-18 15:57:39 ncopa: #4965 I believe it should be fixed in final release, as all ncurses problems were solved in the end, right? should I mark it as resolved (and then closed) or feedback only? 2015-12-18 15:59:57 i think it is fixed by now 2015-12-18 16:01:55 there is one more ncurses fix actually 2015-12-18 16:02:04 i just found another issue 2015-12-18 16:02:06 trivial 2015-12-18 16:02:21 what is it this time? 2015-12-18 16:02:32 symlink in wrong package 2015-12-18 16:02:43 libcurses.so is in -libs package instead of -dev 2015-12-18 16:03:15 no real harm in it either 2015-12-18 16:03:37 it just causes ncurses-dev to be pulled in when not needed 2015-12-18 16:05:05 so I'll mark it as resolved, in worst case it will be created again 2015-12-18 16:05:59 should I resolved and close later some stuff (there are some issues that are resolved, but has not been marked like it, I found one so far), or you'll be closing yourself 2015-12-18 16:06:08 ? 2015-12-18 16:06:37 oh, I received texlive mail ;) 2015-12-18 16:15:57 from MarK? 2015-12-18 16:16:08 (with a lower case K) :) 2015-12-18 16:16:23 Marc Collin 2015-12-18 16:16:37 it seems Issac has a APKBUILD file 2015-12-18 16:16:54 I'll ask him to send it and i'll try it out in edge 2015-12-18 16:18:04 unless you want to deal with it przemoc 2015-12-18 16:18:12 from the image of mail (sic! :>) it looked like it's not necessary fully working thing, but rather mostly working 2015-12-18 16:18:23 no, I don't have alpine yet, so it's hard to do any testing 2015-12-18 16:18:37 ah yes, I forgot 2015-12-18 16:18:39 :) 2015-12-18 16:18:45 remind me why? 2015-12-18 16:21:46 because I didn't have time to switch to it yet. I wanted to alpinize last year, bought hardware for new home server, but there are always other things with higher priority. server is assembled but without disks yet. ;) I hope to install 3.3 soon finally. 2015-12-18 16:23:27 not even Alpine in some kind of VM? 2015-12-18 16:23:34 not yet 2015-12-18 16:23:36 sorry 2015-12-18 16:23:44 *blink* 2015-12-18 16:23:53 but I'll fix my behavior soonish 2015-12-18 16:24:23 do you want to borrow a hosted Alpine Linux install? 2015-12-18 16:24:49 what do you mean by borrowing? 2015-12-18 16:27:19 you can have a root shell on a hosted Alpine Linux VM if it helps. Something like an LXC container with ssh forwarding from the internet and NAT to the internet if you like? It costs me nothing 2015-12-18 16:31:39 sounds nice. what's the bandwidth? 2015-12-18 16:33:09 python3 is working in alpine? 2015-12-18 16:34:37 what's the bandwidth? depends where I put you 2015-12-18 16:34:47 should work fine 2015-12-18 16:34:49 not sure what are current release notes draft, but python3 is one of bigger things too (if it works :>) 2015-12-18 16:35:38 py3 packages are a different story though 2015-12-18 16:36:51 we should tackle that for 3.4 2015-12-18 16:37:27 so we cannot treat python 3 package request as solved? 2015-12-18 16:38:16 main python stuff is working, right? so I would call it resolved after all. 2015-12-18 16:40:14 python3 is working and some externals like boost are built with py3 support, but general python packages are py2 only atm i think 2015-12-18 16:41:29 ncopa: what you think re #3524? should it be 100% resolved, as we have python3 already, or not resolved and target 3.4, because not all python packages support py3? 2015-12-18 16:43:05 przemoc: /q me if you want some Alpine Linux cpu cycles until you get yours sorted out 2015-12-18 16:43:12 have a nice weekend all 2015-12-18 16:43:43 ok, ScrumpyJack. will consider your generous offer. 2015-12-18 16:47:00 ACTION is leaving to watch star wars in 1,5h 2015-12-18 19:41:36 can someone explain noarch vs all in APKBUID? 2015-12-18 19:49:20 ScrumpyJack: sure: http://git.alpinelinux.org/cgit/abuild/tree/abuild.in#n1571 2015-12-18 19:51:18 apparently arch=all means there are arch-specific binaries that can be built on all relevant archs 2015-12-18 19:51:35 or rather, you set all if your apk includes binaries but builds on all archs, and noarch if your apk includes only things that are not architecture sprecific, like a shell script 2015-12-18 19:52:03 this leaves a weird gap for things that include architecture-specific things which are not binaries 2015-12-18 19:52:41 e.g. the check for # we dont want -dev package go to noarch 2015-12-18 19:53:19 I ran into this earlier with ljsyscall 2015-12-18 19:58:12 yeah, the check is not quite right for that 2015-12-18 20:02:54 not sure whether I should just set it =all next time and ignore the warning 2015-12-18 20:04:32 or set options=!archcheck ? 2015-12-18 20:05:42 I agree, noarch seems wrong for it so all + !archcheck is probably the way 2015-12-18 20:49:47 ncopa: re: release note, maybe mention addition of new mirrors and 'leaseweb' (superfast) ! 2015-12-18 21:02:17 and what does superfast mean in numbers? :) 2015-12-18 22:29:12 I didn't even know about archcheck 2015-12-18 23:59:50 is the behavior of the restart policy from docker-compose used differently in Rancher? 2015-12-19 00:04:09 oops, sorry wrong channel! 2015-12-19 00:42:03 urf... I am back 2015-12-19 00:47:51 #3621 please someone test that after adding docbook2x to makedepends in main/lxc/APKBUILD, lxc-doc finally has man pages and if so, then push it to aports. 2015-12-19 00:53:50 clandmeter: I guess you haven't touched subject of #4452? postponing it to 3.4.0, but please provide comment there if some of it is already in 3.3. 2015-12-19 01:16:29 #2038 if someone would be willing to create libmongo-client pkg and extend rsyslog pkg, test it and verify it, then please push it to aports if it works 2015-12-19 16:59:20 im doing v3.3.0 now 2015-12-19 17:34:51 ncopa: great. sadly I haven't managed to tackle with all issues yet. tried hard to squeeze time to work more on them today... 2015-12-19 17:35:16 without much results, though 2015-12-19 18:16:05 extra folders in nl.a.o/alpine/v3.3/ 2015-12-19 18:43:47 vkris: what extra folders? 2015-12-19 19:52:03 seems removed now, but still on mirrors, eg http://mirror.yandex.ru/mirrors/alpine/v3.3/ 2015-12-19 19:56:10 http://imgur.com/O3IYJub 2015-12-20 10:16:16 @ammunta thanks for packaging ljsyscall for alpine BTW 2015-12-20 12:18:58 congrats on the 3.3 release :) 2015-12-20 12:45:22 ncopa: please create 3.3.1 target when you'll be able to. at least some (very) low hanging fruits/fixes that didn't manage to come into 3.3, and were postponed to 3.4 before you released, should went into 3.3.1. like #3621 2015-12-20 12:47:54 not having qemu pushed by anyone, even though it was on ML, is also a bit sad. 2015-12-20 13:15:37 oh, downloads page still have 3.2.3 ISO/tar.gz. is it somehow automated to mirror syncing? i.e. they're not synced yet, so there is no 3.3? 2015-12-20 13:16:18 s/automated/related/ 2015-12-20 13:26:59 release date in http://alpinelinux.org/posts/Alpine-3.3.0-released.html is wrong, yesterday was Dec 19 ;) 2015-12-20 13:31:22 in future I would suggest keeping non-final ISO/tar.gz in subdirectory, maybe rc? they unnecessarily clutter releases/ARCH/ directories 2015-12-20 14:28:35 are new iso already build? download page does not show them http://alpinelinux.org/downloads/ 2015-12-20 15:33:01 crow: isos are already available http://rsync.alpinelinux.org/alpine/v3.3/releases/x86_64/ 2015-12-20 15:40:32 please git pull git://git.alpinelinux.org/user/fab/aports 2015-12-20 15:40:42 only updates ;-) 2015-12-20 17:44:06 chris|: yes but no on official page :) 2015-12-21 02:39:33 currently I can't install ncurses-dev on edge, can anybody reproduce it? http://sprunge.us/LRVB 2015-12-21 02:44:50 nmeum: http://sprunge.us/KFHJ no problem here 2015-12-21 02:46:02 is that problem on update? 2015-12-21 07:52:06 morning 2015-12-21 07:54:37 same problem here 2015-12-21 09:51:31 hi 2015-12-21 11:23:04 anyone on edge? 2015-12-21 11:23:55 there seems to be an issue with ncurses-widec-libs-6.0-r2 conflicting 2015-12-21 11:24:02 a climber that is on the edge? 2015-12-21 11:24:08 heh 2015-12-21 11:25:17 apk just purged 258 packages :( 2015-12-21 11:25:30 I did apk upgrade --force :) 2015-12-21 11:25:46 ive had that before 2015-12-21 11:26:01 are they gone from world? 2015-12-21 11:26:20 force is evil 2015-12-21 11:26:29 no they are will there 2015-12-21 11:26:54 yes, I used force knowing it would probably break stuff, so i blame myself :) 2015-12-21 11:27:09 I used the force? go me! 2015-12-21 11:27:36 im guesing you are having apk issues like mentioned today? 2015-12-21 11:27:47 go to the cinemas :D star wars 2015-12-21 11:28:03 but star wars is booked out for the next two weeks :( 2015-12-21 11:28:37 libncursesw.so.6 issues? yes 2015-12-21 11:29:32 fabled: any idea with apk ncurses issues? 2015-12-21 11:29:54 i guess ncopa is out today? 2015-12-21 11:30:08 i think he said he was out for the next 2 weeks 2015-12-21 11:30:23 i'm on edge, so perhaps I should move to 3.3 instead 2015-12-21 11:30:50 i have no idea. i saw fabled around, he may know whats going on. 2015-12-21 11:31:01 possible the rebuild of packages did not use fixed ncurses due to edge builder having tmux or similar installed 2015-12-21 11:31:16 i had to manually rebuild some packages on arm/edge 2015-12-21 11:31:41 would be good to know what packages need rebuild 2015-12-21 11:31:46 for my edge / x86_64 seems ok 2015-12-21 11:31:56 but i might not have all packages installed you have 2015-12-21 11:32:05 "apk upgrade --latest" error output would be helpful 2015-12-21 11:33:15 http://sprunge.us/QZci 2015-12-21 11:35:21 try: apk del ncurses-widec-libs 2015-12-21 11:35:25 i think ncurses-widec-libs got removed 2015-12-21 11:35:35 we should probably add the ncurses-libs to provide that name 2015-12-21 11:36:49 ok cool. moc stopped apk from deleting ncurses-widec-libs i think 2015-12-21 11:37:06 moc what pinned from older repo 2015-12-21 11:56:37 fabled: thanks for your help 2015-12-21 12:37:02 I don't think a lots of rebuild and direct final release without RC release was recommended 2015-12-21 12:37:28 but v3.3 was already late and would have got delayed more 2weeks... 2015-12-21 13:06:53 vkris, agreed. and i think ncopa is off this week. so we would have been late 3-4weeks without release last week 2015-12-21 13:13:25 ScrumpyJack: I've created another issue for TeX Live as it will be new attempt. please let old one be closed. 2015-12-21 13:43:36 fabled: i think we need to update www for the iso uri's and also update the latest-stable symlink on the mirrors. 2015-12-21 13:45:15 clandmeter, oh, if you know how, please go ahead. i'm not sure. 2015-12-21 13:45:30 +1 clandmeter 2015-12-21 14:14:58 indeed, http://alpinelinux.org/downloads/ all point to 3.2.3 :( 2015-12-21 14:16:00 i remember we have a checklist to follow before pushing out a new alpine version 2015-12-21 14:20:34 I believe ncopa was releasing 3.3 in a hurry, so cutting corners was inevitable 2015-12-21 14:21:06 as long as there are others with privileges to do necessary changes on servers, there is no harm 2015-12-21 14:22:00 corners? They were cross country shortcuts ;) 2015-12-21 14:26:03 he manages that in git... 2015-12-21 14:26:07 bah. Forgot to mention $repo/$package in commit message. Stupid me. :-/ 2015-12-21 14:27:08 clandmeter: managing in git doesn't imply the workflow, deployment procedures, whether there are hooks and what not. 2015-12-21 14:28:35 is it described on wiki or is it insiders knowledge? 2015-12-21 14:29:00 you have a chance of 10% it is described on wiki 2015-12-21 14:31:00 I've found only http://wiki.alpinelinux.org/wiki/Alpine_Release_Testing_Checklist 2015-12-21 14:36:29 its reading some yaml file 2015-12-21 14:36:36 probably provided by some pkg 2015-12-21 14:40:31 i have no idea how he manages it 2015-12-21 14:40:39 better to send him an email 2015-12-21 14:41:31 my guess is, you check out the scritps, modify it, push the changes and it will post upload it. 2015-12-21 14:42:58 ah i see now 2015-12-21 14:44:11 i updated the latest-stable uri 2015-12-21 14:45:46 ok 2015-12-21 14:45:52 seems to work 2015-12-21 14:46:02 only need to push some commit 2015-12-21 14:51:57 seem i cannot commit 2015-12-21 14:52:23 is it some public repo? 2015-12-21 14:54:24 hmm 2015-12-21 14:54:39 i updated the readme, but it didnt post push the site 2015-12-21 14:56:00 you posted to master, and it's possibly working from production 2015-12-21 14:58:19 yeah, do that on production branch 2015-12-21 14:58:24 nope 2015-12-21 14:58:26 doesnt matter 2015-12-21 14:58:30 meh 2015-12-21 14:58:40 we should make ncopa document it all 2015-12-21 14:58:44 maybe he only tracks md files 2015-12-21 14:58:59 strangely last ncopa commit in production is not in master, possibly another rushed change ;) 2015-12-21 14:59:04 oh yeah, that could be it 2015-12-21 15:02:31 where are the git repos moved ot? 2015-12-21 15:02:32 to 2015-12-21 15:02:34 on disk that is 2015-12-21 15:03:35 we moved to gitolite 2015-12-21 15:03:48 found it 2015-12-21 15:03:49 /var/lib/git, git* maybe 2015-12-21 15:03:51 checking post scripts 2015-12-21 15:04:06 I don't have access there anyway, except gitolite-admin repo 2015-12-21 15:04:15 oh, I have it 2015-12-21 15:07:18 damn, somekind of mqtt msg 2015-12-21 15:07:26 i give up before i destroy anything 2015-12-21 15:09:48 if ncopa ever leaves this building, an earthquake will happen. 2015-12-21 15:20:35 optimist 2015-12-21 15:21:12 i prefer the word realist 2015-12-21 15:22:15 well, realist would rather say: earth will blow up 2015-12-21 15:22:53 soon, when we hit 21 degrees this month 2015-12-21 15:24:14 it's ~7C in Warsaw, Poland right now 2015-12-21 15:24:20 still more than expected, though 2015-12-21 17:05:24 przemoc: dude, I appreciate that you are trying help with the issue management on bugs.a.o, but please ask around before closing an issue that is being talked about/people are working on. #3920 was fine, shouldn't have been closed. I asked to re-open it, and you zapped it and opened a new one :( 2015-12-21 17:07:51 please don't take this the wrong way, again, your help is appreciated, but people are tracking these issues (via scripts or atom/rss feeds) and now they have to change their scripts/feeds/whateverr with a new issue number for the same package. And any other implication that closing/opening issue numbers may have on the community 2015-12-21 17:13:57 Am i being unreasonable here? 2015-12-21 18:12:10 you are not… package request is not closed till the package actually works 2015-12-21 19:26:32 ScrumpyJack: it was not talked for about 8 straight months. old attempt was deemed a failure. closing with 0% reflected in the end texlive being moved to unmaintained with 3.3. originally it was meant to be closed as package was made, and it only needed some fixing, so it should be done in new bug issue and go into 3.3.1, that's how I've seen it, but then I heard on this channel that texlive is to 2015-12-21 19:26:38 o broken to even keep in testing and nmeum was tired of some mails regarding it, so it was moved to unmaintained. was that really so harmful? so far there was practically no issue management, just read some issues, like you assigning barthalion for something and issue staying untouched for many months. or people asking for package and never reporting back whether it's broken - if someone is intere 2015-12-21 19:26:45 sted in package, he should be relatively prompt about its quality or at least try to behave like interested. anyone wanting to tackle with all these will be blamed for lot of things. so far I can keep with such complaints, but if I'm really more harmful than helpful, then maybe I shouldn't strain myself. be also reasonable about these supposed heavy trackers... changing issue id in some script is 2015-12-21 19:26:51 no big deal, is it? 2015-12-21 19:38:54 None of that matters. The only thing that really matters to me is that you ask around before doing stuff that involves other people. i don't think I have ever assigned an issue# to someone, barthalion included, without asking them first etc. People were involved with #3920, dispite the relative quiet on b.a.o. Just ask around, it's called a community for a reason [tm]. 2015-12-21 19:39:10 and this is not personal, I think you're cool 2015-12-21 20:02:03 ok, upgrading to 3.3 2015-12-21 21:07:01 why I am noticing a 30%+ increase in all pkgs size from previous versions, irrespective of being upgraded pkg or not !? 2015-12-21 21:07:30 I hope this is ok ? 2015-12-21 21:07:58 some or even 50%+ 2015-12-21 21:10:24 check the binaries have correctly been stripped 2015-12-21 21:10:31 (just a wild guess) 2015-12-21 21:13:07 comparing between 3.2.x and current v3.3 on nl.a.o 2015-12-21 22:08:25 vkris: do you have results? 2015-12-21 22:21:40 ScrumpyJack: ok. I'm sorry for prematurely closing the issue, it wasn't meant to harm anyone and I didn't want to sound harsh for you. if you think it's necessary, then I can repeat my apology in bugs.a.o too. once again, closing came mostly from original plan of leaving it at 90% and making it fully working in a bug-fix release. back then I didn't know it was in such a bad shape. but when movin 2015-12-21 22:21:46 g to unmaintained came out, I thought that there is no reason to back out as 1) closing it marks some milestones (even if bad one) and 2) opening new one will bring some fresh attention to it, because it was really dormant, as I wrote before (if there was some progress along the way, there is no excuse to report about it in the issue). actually it was meant to involve people, because dealing with is 2015-12-21 22:21:52 sues ultimately means involving them (devs included too), especially if we want to assure perfectly working stuff. 2015-12-21 22:43:35 ScrumpyJack: the texlive aport was seriously unfinished and nobody wanted to finish it. That's the reason it was moved to unmaintained incase people start working on it again we can move it back, it's not an issue 2015-12-21 23:02:27 and I can understand why nobody wanted to finish this tex is serious pita to package 2015-12-21 23:04:55 yeah, absolutly 2015-12-21 23:25:38 gosh right now I'm hatting elasticsearch sooooooo much >_< 2015-12-22 09:11:05 ok, let's move on from the texlive issue issue :) 2015-12-22 09:23:50 any objections to pulling back openjdk6 into the repos for those who need legacy java support? 2015-12-22 09:24:20 i guess we can stuff it in community 2015-12-22 09:24:53 though it's probably full of known security issues... so dunno if that needs some tagging as 'use at your own risk' 2015-12-22 09:27:37 perhaps it could go in unmaintained? (which seems wrong, as unmaintained is a graveyard) 2015-12-22 09:28:33 unmaintained is not compiled 2015-12-22 09:28:50 but you can of course compile it yourself 2015-12-22 09:33:43 unmaintained isn't compiled, I hadn't thought of that 2015-12-22 09:34:06 so community then? 2015-12-22 10:09:47 hello, I have somes questions about error during software install, is it the right chan for this ? 2015-12-22 10:53:57 Bishop: yeah 2015-12-22 10:55:10 I want to install powerdns recursor on alpine 2015-12-22 10:55:21 .configure is ok 2015-12-22 10:56:02 but during make i got error like undefined reference to `swapcontext' 2015-12-22 10:56:08 undefined reference to `makecontext' 2015-12-22 10:56:42 I understand that's libc things and muslc don't use it 2015-12-22 10:57:00 musl does not support those 2015-12-22 10:57:06 is there a solution ? 2015-12-22 10:57:25 if powerdns requires them, then not really. except to write the code to musl 2015-12-22 10:58:25 bad news 2015-12-22 10:58:41 iirc, posix deprecated those functions; so it has not been a priority on adding them 2015-12-22 10:58:52 I read that :) 2015-12-22 10:59:24 we have few other packages that would benefit of them 2015-12-22 11:00:14 I guess, but if it's deprecated, dev shouldn't use it 2015-12-22 11:01:34 ok then, thanks for your time =) 2015-12-22 11:04:30 Bishop, but *context stuff is wanted for gcc-go and few others too 2015-12-22 11:06:03 ? 2015-12-22 11:08:18 i don't get it 2015-12-22 11:09:42 gcc's libgo uses makecontext/setcontext also 2015-12-22 11:09:53 so gcc-go is disabled currently 2015-12-22 11:10:03 there's few other apps also using it 2015-12-22 11:10:04 the thing is 2015-12-22 11:10:34 the reason why *context was obsoleted is stupid; and there's no real replacement 2015-12-22 11:10:42 ok 2015-12-22 11:10:57 then muslc is the problem ? 2015-12-22 11:11:03 they tell to use threads, but e.g. for go where there can be 100k+ go routines, it's not viable to use threads 2015-12-22 11:11:03 in this case ? 2015-12-22 11:11:09 well 2015-12-22 11:11:23 posix is the problem - they deprecated it, but provided no real alternate 2015-12-22 11:11:41 musl generally wants to do posix stuff only. i suppose *context might get in if it was done properly 2015-12-22 11:11:56 but since it's obsoleted, it's not a priority for the musl author 2015-12-22 11:12:20 i get it 2015-12-22 11:13:25 and that's not bad for alpine software pool ? 2015-12-22 11:14:23 if it's really important for thread, does this mean a lot a software won't compile on alpine because of this ? 2015-12-22 11:15:08 fabled: what was the reason for deprecating *context in POSIX? simply telling that you can redo any *context stuff with threads? 2015-12-22 11:18:35 dalias would remember better. i think it was just some "formalistic" reason. 2015-12-22 11:18:58 i'm trying to find the rationale 2015-12-22 11:19:36 ah 2015-12-22 11:19:36 void makecontext(ucontext_t *ucp, void (*func)(), int argc, ...); 2015-12-22 11:19:36 is making use of an obsolescent feature of the ISO C standard. Therefore, a strictly conforming POSIX application cannot use this form. Therefore, use of getcontext(), makecontext(), and swapcontext() is marked obsolescent. 2015-12-22 11:19:45 http://pubs.opengroup.org/onlinepubs/009695399/functions/makecontext.html 2015-12-22 11:20:08 use of function declarators with empty parentheses is an obsolescent feature 2015-12-22 11:20:16 so it's the "void (*func)()" that is obsolete 2015-12-22 11:20:28 instead of adding '(void)' they decided to remove it 2015-12-22 11:25:55 hey :) 2015-12-22 11:28:19 well, i suppose the entry function requires some arguments 2015-12-22 11:29:00 i suppose they would've needed to specify new makecontext2 or similar that takes callback with (int argc, ...) or similar 2015-12-22 12:10:04 well, having well-specified arguments for function taken as pointer is definitely a good thing, so makecontext2 would be the only way to go, unless they would be willing to break compatibility with redefining existing makecontext... 2015-12-22 13:39:11 someone here was looking for powerdns? 2015-12-22 13:39:16 Or it was alpine-linux? 2015-12-22 13:40:19 no Bishop was trying to package powerdns I guess 2015-12-22 13:40:31 > 11:55 < Bishop> I want to install powerdns recursor on alpine 2015-12-22 13:40:32 ah ok 2015-12-22 13:40:38 Bishop, still around? 2015-12-22 13:40:43 powerdns is available in testing. 2015-12-22 13:40:43 yep 2015-12-22 13:40:46 hi 2015-12-22 13:40:46 Please test it. 2015-12-22 13:40:49 hi. 2015-12-22 13:40:49 :) 2015-12-22 13:40:58 O.O 2015-12-22 13:41:06 sure i'll do that 2015-12-22 13:41:10 thanks 2015-12-22 13:41:18 np 2015-12-22 13:41:40 fcolista: how did you fix the swapcontext issue that Bishop ran into? 2015-12-22 13:42:10 i didn't had that issue, probably he was trying to package a different version. 2015-12-22 13:42:24 wich version it is ? 2015-12-22 13:42:31 3.4.7 2015-12-22 13:43:10 the last one. 2015-12-22 13:43:43 that's the autorithative server 2015-12-22 13:44:00 did you try the recursor only ? 2015-12-22 13:44:08 Bishop, no 2015-12-22 13:44:10 only auth 2015-12-22 13:44:53 so downloads on www is ok now? 2015-12-22 13:45:14 did ncopa fix it or did it do that automagically? 2015-12-22 13:45:21 anyway it's good news that's will be my next software to compile 2015-12-22 13:50:21 can't try just right now 2015-12-22 13:50:29 i hate the mozilla build scripts: https://paste.debian.net/plain/354405 2015-12-22 13:50:30 i will soon 2015-12-22 13:50:41 does anyone see an reason for the error? 2015-12-22 13:57:28 Bishop, swapcontect issue is in powerdns-recursor 2015-12-22 13:58:52 but in the auth, the recursor part doesn't give the error ? 2015-12-22 13:59:11 in the auth there's not recursor part. 2015-12-22 14:00:35 i think i read this on the powerdns site 2015-12-22 14:11:06 fcolista: you are right, i miss read the doc 2015-12-22 14:11:51 looks that so far we can have only authoritative powerdns. 2015-12-22 14:12:03 yep 2015-12-22 14:12:03 recursor is not possible for the moment. 2015-12-22 14:12:22 i know 2015-12-22 14:14:55 why would a DNS recursor (whatever that means) even use swapcontext 2015-12-22 14:18:10 skarnet, i think fabled spoke about that before 2015-12-22 14:18:25 but i'm not a C-coder, so his works flying up on my head. 2015-12-22 14:19:06 he said that threads were not always practical from an efficiency point 2015-12-22 14:19:25 which may be true for goroutines, but for a fucking DNS cache this makes no sense at all 2015-12-22 14:20:16 (I mean, what fabled said makes sense, but using swapcontext, or even threads, for a DNS cache, is idiotic) 2015-12-22 14:20:48 sounds like they used it as cheap co-routine approach 2015-12-22 14:20:50 yes, i got your point. 2015-12-22 14:21:42 ncopa said that dnscache (from djbdns) wasn't used in Alpine because it wasn't efficient enough 2015-12-22 14:21:57 I'd be interested in comparative benchmarks and profiling 2015-12-22 14:21:58 we have unbound in repos 2015-12-22 14:22:15 yeah 2015-12-22 14:22:18 but it has issues with scalability as well. 2015-12-22 14:22:50 I'm really wondering where dnscache loses efficiency, and I'm interested in learning more 2015-12-22 14:23:24 (I still have plans to add a cache to s6-dns at some point, just because djbdns is unmaintained) 2015-12-22 14:23:53 (so if dnscache lacks scalability, I want to learn from it and avoid the same mistake) 2015-12-22 14:25:27 fcolista: I thought unbound was used for big sites where scalability was the primary factor? 2015-12-22 14:29:32 skarnet, yes, i had issues with unbound when n. of queries becomes high, it needed to be restarted. It was a thread or fd issue, but an upgrade fixed the issue. It's not a design problem. 2015-12-22 14:31:18 I see, thanks. 2015-12-22 14:31:31 dunno about maradns, we have it in repos too 2015-12-22 14:31:42 I only just now realized, when you mentioned s6, that you're skarnet of s6 fame. I've never got into trying s6, though. I like runit, but I dislike djb's awful compile scripts heritage used in it. 2015-12-22 14:32:50 przemoc: since January 2015, I've switched to a configure/make/make install model, so you're welcome to try s6 :) 2015-12-22 14:33:49 oh, what a wise decision. well-appreciated change. 2015-12-22 14:34:37 fcolista: I'd love a serious benchmarking initiative of existing DNS resolver software (from BIND and dnscache to maradns and unbound via powerdns "recursor"). I just can't do it myself, I lack the knowledge and infrastructure to do that. 2015-12-22 14:37:13 przemoc: I liked the djb approach because the only configure scripts I knew were autoconf-generated and there was no way I was going to use that. But people were giving me a lot of shit for it (and rightfully so, it appears); and then I met a nice, simple, well-designed hand-written configure script that I could use as a model, and so I made the change. It was hard work, but I'm happy to have done it. 2015-12-22 14:38:37 yeah, I'm also agaist autohell. I usually go for Makefile and add simple hand-made configure script only if necessary 2015-12-22 14:41:14 I depend on GNU make in practice, though, as it have some useful enhancements. it's not a problem in practice, as it's available for most platforms. 2015-12-22 14:41:52 yes. I actually have a lot of praise for GNU make. 2015-12-22 14:47:02 I have to say that idea of runit-based OS like void linux is somewhat appealing for me. has anyone tried that, how well it works in practice? maybe alpine could become s6-based. ;) 2015-12-22 14:48:25 przemoc: shhh, don't reveal my master plan 2015-12-22 14:49:10 (open secret master plan, I guess :>) 2015-12-22 14:49:55 fcolista: if you have some free quarter today, maybe you'd be willing to take on http://bugs.alpinelinux.org/issues/3621#note-9 2015-12-22 14:50:01 Changing inits is easy enough, I've done it plenty of times, it's a detail. Changing the *init scripts* is a whole other enchilada, and requires a lot of work 2015-12-22 14:50:52 and no upstream project provides runit/s6-oriented init scripts 2015-12-22 14:51:06 My objective was to provide s6-rc as an alternative service manager (and maybe even the default) for Alpine 3.4 2015-12-22 14:51:16 but I had to take a job, and so did ncopa 2015-12-22 14:51:24 so it's probably going to delay things 2015-12-22 14:52:12 it's not on upstream projects to provide init scripts, be it s6-rc, sysvrc, openrc or systemd-oriented ones 2015-12-22 14:52:21 it's purely a policy decision and belongs to the distro 2015-12-22 14:52:23 such stuff delay things significantly, usually, unless you're able to do your open source staff in work 2015-12-22 14:53:29 it's true to some extent, but in practice big projects provide their init scripts templates, supporting sometimes some additional actions beside start/stop/reload 2015-12-22 14:53:47 I know, and they're wrong. 2015-12-22 14:54:00 They should provide the *mechanism* for the actions, but not the scripts themselves 2015-12-22 14:54:38 because using their scripts more or less enforces the model they've chosen 2015-12-22 14:54:46 you can treat their init scripts as example how to work with their mechanisms, as it's not always that obvious 2015-12-22 14:54:48 and it's not on upstream to choose an init script model 2015-12-22 14:55:02 so I say they can be helpful 2015-12-22 14:55:03 oh yes, examples are totally fine 2015-12-22 14:55:07 that they are 2015-12-22 14:55:12 but they must not be the official API 2015-12-22 14:55:21 agreed 2015-12-22 14:55:54 usually there are some *ctl or SIGUSR1 etc. stuff, so it's manageable in most cases 2015-12-22 14:56:44 it's always manageable. I just get tired of it feeling hackish and unsupported. I want alternate daemon management models to become first-class citizens. 2015-12-22 14:58:08 distros simply install init scripts along with pkgs, as there are no pkg-init-runit, pkg-init-s6, etc. subpackages 2015-12-22 14:58:31 I had a discussion about that on the devuan ML 2015-12-22 14:59:50 and my conclusion is that it would actually be a good idea to have foobar-systemd, foobar-s6, foobar-openrc packages along with foobar and foobar-dev 2015-12-22 15:00:19 when you install foobar, the package manager tests what your current init system is, and installs the proper dependency 2015-12-22 15:00:43 so you have the script that matches your chosen service manager 2015-12-22 15:01:13 and that's independent from the mechanism itself, contained in the foobar package 2015-12-22 15:01:17 and I also think it would be nice that way. but I would call it with init- in between, to make it explicit it's one of init subpkgs and usually you don't want other one if you have already one 2015-12-22 15:02:18 sure, naming can reflect that 2015-12-22 15:03:03 and ideally it wouldn't hurt to have foobar-init-s6 if you're using openrc, or the other way around - it would just be unnecessary. 2015-12-22 15:03:56 exactly, as long as each init solution goes with its own directory structure 2015-12-22 15:04:51 that's easily enforceable at the distro level :) 2015-12-22 15:06:06 having init subpkgs would be really nice (and even alpinish, I would say). AL already have man pages separated in -doc subpks, so it would be obvious next step. 2015-12-22 15:07:07 conceptually that's not hard, but as with everything that requires manually scanning the whole package db, that's *a lot* of work 2015-12-22 15:07:41 it could be automated to some extent, I presume 2015-12-22 15:08:29 anyway, before any switch, these -init-openrc subpkgs would need to appear first in all pkgs with daemons 2015-12-22 15:08:48 not only with daemons. With anything that requires an init script. 2015-12-22 15:09:35 examples of init scripts not dealing with daemons? 2015-12-22 15:09:46 but yes, making those subpackages first would make it a lot easier to integrate alternate service managers. 2015-12-22 15:10:18 anything that is a oneshot. Seeding randomness, setting the hostname, setting up the network, etc. 2015-12-22 15:10:25 ok, fair enough 2015-12-22 15:12:35 when ncopa will be around, then maybe some mail to alpine-devel should be posted about such init subpks concept to discuss these matters. 2015-12-22 15:13:47 I think he may be against it as it would make non-standard upgrade path for 3.4. 2015-12-22 15:14:15 I'm not sure about the preferred procedure for Alpine maintainers. Should the discussion happen on the ML? or should it happen here with a ML post to sum up the conclusions? 2015-12-22 15:14:19 We'll see. 2015-12-22 15:14:55 mails have advantage of being better worded and more well-thought than ad hoc discussions 2015-12-22 15:15:31 being able to install pkg without any of its init scripts may also seem super useful in some cases 2015-12-22 15:15:38 what do you mean, "non-standard upgrade path" ? 2015-12-22 15:16:15 again, 3.4 may be a little to close a goal - but if it's just about splitting policy from mechanism, it may be doable 2015-12-22 15:16:20 too close* 2015-12-22 15:17:55 people would need to install all these -init subpkgs for their daemons and what not. to made it automagical, apk should gain some new coffee powers, and it should went into last 3.3.x stable release, to make future transition as smooth as possible 2015-12-22 15:18:56 main pkg should not depend on -init subpkg, that's why otherwise it would not install without some additional magic 2015-12-22 15:19:49 I'm not sure about the extent of apk's coffee powers, but I'm sure it's manageable 2015-12-22 15:21:53 for instance, if renaming is allowed with apk, you could rename foobar to foobar-openrc, and make foobar-openrc a virtual package containing foobar and foobar-init-openrc 2015-12-22 15:23:38 that approach would multiply pkg number by ~3 or more (if other init variants would show up). it seems a bit too bloated at first sight. 2015-12-22 15:25:03 um, no, additional packages are only created for stuff that involve init scripts. 2015-12-22 15:25:13 sure thing 2015-12-22 15:25:25 and the virtual packages can disappear after the release. 2015-12-22 15:26:00 I don't know about the apkindex structure, but I don't think virtual packages add much bloat. 2015-12-22 15:26:23 removing them later sounds better, but there is edge too, so it's not simple on/off things I guess 2015-12-22 15:26:37 anyway, I suggest discussing whole init subpkg thing on ML in new year, as not everyone is on IRC at the same time 2015-12-22 15:27:34 of course. And not right away - let ashes fall down first, maintainers will have plenty of work with 3.3.0 feedback 2015-12-22 15:29:00 many of them have said "we're interested in changing inits, but it's not a priority" so let's not start a construction site before the 3.3 one isn't packed up yet XD 2015-12-22 15:29:18 gah, I can't English today 2015-12-22 15:29:28 (but you get my meaning) 2015-12-22 15:44:18 bye 2015-12-22 19:08:17 hi 2015-12-22 19:08:25 can please someone confirm this bug: 2015-12-22 19:09:38 1.) boot alpine 3.3 from usb 2) setup alpine with lbu 3) install zsh or any package, which needs to use the apkcache 4) lbu commit the changes 5) reboot 6) any packages in apkcache will not be installed after reboot, even it's in etc/apk/world mentioned and available in the cache directory 2015-12-22 19:14:59 i think this is a great idea. http://www.openbsd.org/cgi-bin/man.cgi/OpenBSD-current/man2/pledge.2?query=pledge 2015-12-22 19:14:59 any thoughts from you? 2015-12-22 19:45:28 #4975 :-( 2015-12-22 20:06:00 what's the difference between testing/libxss and main/libxscrnsaver? The use the same source... 2015-12-22 20:06:29 open a ticket, one of them will be then removed 2015-12-22 20:06:46 I can remove them myself the question is which one of them should be removed 2015-12-22 20:07:22 there is actually no packages which depends on libxss so I guess removing libxss would be the better option 2015-12-22 20:07:26 and libxscrnsaver is already in main/ 2015-12-22 20:13:44 fabled: Is there any way to tell apk that it should install a package from a specific repository? For instance firefox is available in community/ and in testing/, is it possible to explicitly install the version from community/ (firefox ESR)? 2015-12-22 21:16:18 #4975 is annoying 2015-12-23 00:44:09 hmmm.. https://pkgs.alpinelinux.org/contents?pkgname=adwaita-icon-theme-lang&arch=x86_64&repo=main 2015-12-23 00:44:42 No item found... -- but 74.37 kB 2015-12-23 00:45:47 did something go weird with gzip..ing ? 2015-12-23 00:50:21 ok seems it only has empty folders 2015-12-23 07:34:26 how long is ncopa in holidays? 2015-12-23 08:09:47 morning climbers 2015-12-23 08:16:36 !last 2015-12-23 08:17:12 ScrumpyJack: morning 2015-12-23 08:19:15 shafire: i cannot reproduce your issue 2015-12-23 08:19:56 clandmeter: I tried many times, I will try it now again 2015-12-23 08:20:03 with cdrom, it's working, just usb not 2015-12-23 08:20:04 wait 2015-12-23 08:20:08 i know what is diff 2015-12-23 08:20:18 took the wrong iso from my dl folder :| 2015-12-23 08:23:00 shafire, i think ncopa is away at least all of this week. possible it was also next week. 2015-12-23 08:25:03 fabled: i can reproduce, apk does not install cache after a reboot on tmpfs 2015-12-23 08:26:41 fabled: i guess this is done in initramfs? 2015-12-23 08:27:34 nmeum, http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management#Repository_pinning 2015-12-23 08:27:58 clandmeter, sounds like the symlink is pointing to wrong place 2015-12-23 08:28:12 /media/sda1 during runtime, /media/something else during boot time 2015-12-23 08:31:07 fabled: i guess i cannot set anything at boot? 2015-12-23 08:31:37 on 3.2 it was derived from alpine_dev etc. 2015-12-23 08:31:48 on 3.3 i think it should honor fstab from apkovl 2015-12-23 08:31:59 I saw that v3.3 does not use alpine_dev anymore 2015-12-23 08:32:17 this should someone mention in the upgrade process from v3.2 to v3.3 2015-12-23 08:32:20 yeah, alpine_dev needed to go 2015-12-23 08:32:23 yes 2015-12-23 08:32:38 iirc. if you have the proper mount line in fstab, it should be honored 2015-12-23 08:32:41 so that syslinux.cfg should be upgraded too (without -u?) 2015-12-23 08:33:04 do we have a boot opt to get in emergency shell? 2015-12-23 08:33:17 single 2015-12-23 08:36:08 thx 2015-12-23 08:39:43 fabled is right 2015-12-23 08:39:54 the symlinks is pointing in wrong direction 2015-12-23 08:40:15 shafire: you can fix it by not using usbdisk device 2015-12-23 08:40:33 change it in fstab to /dev/sd** 2015-12-23 08:40:45 and update the symlink in /etc/apk 2015-12-23 08:42:43 problem is that it changes between two everytime 2015-12-23 08:43:02 sdd1 and sde1 2015-12-23 08:43:17 on each boot its diff? 2015-12-23 08:43:28 you can use uuid i guess? 2015-12-23 08:43:57 I think, uuid should work, can I also just update fstab? 2015-12-23 08:44:20 will the behavior in future versions corrected like it was in 3.2? 2015-12-23 08:44:40 im not sure what the correct way of handling is. 2015-12-23 08:44:44 ncopa should look into it. 2015-12-23 08:45:05 but i think you should be able to fix it. 2015-12-23 08:45:45 aslong cache dir points to "a" correct location. 2015-12-23 08:49:12 hmm that wouldnt matter 2015-12-23 08:49:24 it only matters where initramfs mounts it 2015-12-23 08:49:40 shafire: does initramfs keep changing it? 2015-12-23 08:52:40 I think not 2015-12-23 08:53:22 then just change symlink to the mount point from initramfs 2015-12-23 08:53:26 it's just the cdrom drive from KVM AMT, if there is something mounted it's sde, else sdd 2015-12-23 08:54:15 i guess thats why before it used /dev/usbdisk so you dont have that issue. 2015-12-23 08:55:16 fabled: are rpi images broken now? 2015-12-23 09:02:48 hi :) 2015-12-23 09:02:54 any clue http://pastie.org/10648697 ? 2015-12-23 09:05:26 no network? 2015-12-23 09:08:34 hetii : port 80 firewalled? dns issue? 2015-12-23 09:10:34 hmm its odd cause alpine is downloaded 2015-12-23 09:10:39 also I can open http://dl-4.alpinelinux.org/alpine/v3.2/main 2015-12-23 09:10:55 try change repo 2015-12-23 09:11:14 I change to 3.1 with the same result 2015-12-23 09:11:32 and as far as I remmeber one or two hour before all works fine 2015-12-23 09:11:36 http://nl.alpinelinux.org/alpine/v3.2/main 2015-12-23 09:11:40 try this 2015-12-23 09:11:53 in browser I can open it 2015-12-23 09:12:02 are you using proxy? 2015-12-23 09:12:13 no 2015-12-23 09:12:42 here is my docker file http://pastie.org/10648705 2015-12-23 09:13:08 this error is _inside_ the docker container? 2015-12-23 09:13:29 its happen when I build that docker on my host machine 2015-12-23 09:14:23 how I can set in dockerfile to use this second link that you put ? 2015-12-23 09:15:43 ok will reboot my PC and see if that change something... 2015-12-23 09:17:34 its time for alpine-docker :) 2015-12-23 09:17:43 #alpine-docker... 2015-12-23 09:31:01 clandmeter, sort of yes, you need to remove the console=ttyAMA0 from cmdline*.txt to boot the 3.3.0 image 2015-12-23 09:31:11 it's fixed in git, but not in release tarballs yet 2015-12-23 09:31:37 seems it was introduced in a late change in initramfs related to securetty handling 2015-12-23 09:32:35 maybe we sould send a mail to the list with these 2 issues. 2015-12-23 09:33:00 perhaps 2015-12-23 09:33:01 if its still going to be a atleast a week before ncopa is back. 2015-12-23 10:56:37 can someone reject #4971 on b.a.o.? 2015-12-23 10:56:43 dpkg package is already present in main repository. 2015-12-23 10:56:57 And is in the wrong project: "Alpine Documentation" 2015-12-23 10:57:15 i cannot change status of that request in document project 2015-12-23 11:11:44 fcolista, i changed it to alpine project 2015-12-23 12:03:18 fcolista: re #4971 I cannot mark it as rejected. I can only change state to: assigned, resolved, feedback or close. 2015-12-23 12:55:20 przemoc, no problem. I've marked it as resolved. 2015-12-23 12:56:20 I could mark it as resolved too, but you wanted rejected so I avoided changing status. 2015-12-23 12:56:58 I think I've seen having Rejected in list with some bugs, but maybe I'm mistaken. 2015-12-23 12:57:52 i think rejected is a sub-type of resolved state 2015-12-23 12:57:55 or maybe not 2015-12-23 13:00:26 i can't set rejected too. Would it be matter of permissions maybe? 2015-12-23 13:03:12 I think, only ncopa can do that 2015-12-23 13:24:35 I believe that matters not concerning any particular upcoming release (or concerning past releases, but which were solved in already released stuff) should be closed. 2015-12-23 13:24:51 fcolista: do you mind if I close this issue now? 2015-12-23 13:25:11 przemoc, no 2015-12-23 13:25:15 go ahead 2015-12-23 13:25:47 s/matters/resolved &/ 2015-12-23 13:25:47 ok 2015-12-23 15:38:30 I used UUID instead of /dev/usbdisk 2015-12-23 15:38:33 still not working 2015-12-23 15:49:16 clandmeter: http://bugs.alpinelinux.org/attachments/download/317/2_using-uuid-in-fstab.png 2015-12-23 15:49:25 fabled: ^ 2015-12-23 16:59:55 shafire, yeah, UUID is probably not working. i wonder why it did not work with the 'usbdisk' thing 2015-12-23 17:12:42 I am using UUIDs in my /etc/fstab and it works perfectly 2015-12-23 18:05:54 nmeum: v3.3 and lbu/apkcache? 2015-12-23 18:06:37 edge and sys install 2015-12-23 18:06:57 but it shouldn't matter much in this case 2015-12-23 18:07:27 yeah, it's about #4975 2015-12-23 18:40:06 so I will use 3.2 until 3.3 is repaired 2015-12-23 18:40:12 3.2 uses its own edge or not? 2015-12-23 19:06:47 there is only the one edge 2015-12-23 19:07:21 because ncopa told that 3.3 is compiled with gcc 5 and 3.2 with gcc 4 and there will be problems :( 2015-12-23 19:08:48 all true 2015-12-23 19:10:30 can someone backport libvirt-daemon to 3.2? 2015-12-23 19:10:53 1.2.15 has some bugs which are already fixed in newer versions 2015-12-23 19:33:24 shafire: same problem here 2015-12-23 19:33:49 alpine does not load packages from /etc/apk/world at boot 2015-12-23 19:34:15 diskless system, of course 2015-12-23 19:34:58 I found no solution 2015-12-23 19:35:19 I blame this: http://git.alpinelinux.org/cgit/mkinitfs/commit/initramfs-init.in?id=6aba975e73e7cea833640e8db180c0add01ecf33 2015-12-23 19:36:15 so, all the rc's had the same mistake? 2015-12-23 19:36:24 yes 2015-12-23 19:36:34 we need better testing :S 2015-12-23 19:37:06 yes :( 2015-12-23 19:55:28 I could not make UUIDs work on fat 2015-12-23 19:55:38 but on ext3 works 2015-12-23 21:36:37 I won't be online tomorrow, so merry christmas folks 2015-12-23 21:42:04 merry christmas 2015-12-23 21:48:39 merry christmas to you too 2015-12-23 21:51:07 merry christmas to all of you 2015-12-24 05:13:24 hello friends 2015-12-24 05:13:51 are there any instructions anywhere that can help me compile alpine-rpi from source to release? 2015-12-24 08:20:22 shafire: uuid will not work, you need to use the device name which is used in initramfs. you can boot using grsec single to get in initramfs shell. 2015-12-24 08:21:00 shafire: if your device name keeps changing also in initramfs, then you need to wait for a fix. 2015-12-24 09:08:49 clandmeter: I will wait for a fix, I hope, it will be behavior like in 3.2 again 2015-12-24 09:10:20 only in Honolulu, it's still not marry christmas 2015-12-24 09:20:58 someone tried to build openvz on alpine? 2015-12-24 09:23:47 next time, please no big releases before holidays:) 2015-12-24 09:41:00 hey :) 2015-12-24 09:46:17 hi 2015-12-24 10:37:02 hi 2015-12-24 10:52:48 ho 2015-12-24 10:58:28 ho 2015-12-24 11:08:56 ho ho ho 2015-12-24 11:09:11 and Merry Christmas, if you are that way inclined 2015-12-24 12:44:41 3.3 breaks btrfs installs: "btrfs: found compressed data, cannot continue! invalid or corrupt kernel image." 2015-12-24 12:50:26 have you tried rootflags=compress=? 2015-12-24 12:51:32 clandmeter, claws-mail-3.13.1 package while was downloading got interrupted. Therefore the checksum fails. 2015-12-24 12:51:37 http://build.alpinelinux.org/buildlogs/build-edge-armhf/community/claws-mail/claws-mail-3.13.1-r0.log 2015-12-24 12:51:55 i don't have root access to git.a.o for deleting it 2015-12-24 13:12:35 chris| - I've not compressed anything - perhaps it is the compression used in the 4.1 kernel ? 2015-12-24 13:13:11 fortunately I took a snapshot I can roll back to 2015-12-24 13:22:44 Takes 5 minutes to roll back a broken system 2015-12-24 16:48:42 is it just me or is / not properly remounted if it has been mounted by the initramfs already? 2015-12-24 16:49:14 at least mountoptions for / specfied in my /etc/fstab seem to be ignored 2015-12-25 08:06:12 hi 2015-12-25 13:40:49 currently there is no way to install multiple versions of ruby/nodejs/python/go, right? 2015-12-25 13:51:12 maybe i should create separate package trees for that :D 2015-12-25 21:29:46 mosez: I mean I'm p sure py2 and py3 are still separate but that's probably not actually your concern 2015-12-25 22:40:41 mosez: no, alternatives system are pretty much limited to Debian and Gentoo 2015-12-25 22:40:54 you would need to go arch way, creating package like 'ruby2.1' for example 2015-12-26 11:43:18 i've an issue with trh_hdr 2015-12-26 11:43:27 it is declared in if_tr.h 2015-12-26 11:43:39 but this include does not appear in kernel 4.1.13 2015-12-26 11:46:12 i mean this file does not appear in the includes 2015-12-26 11:50:53 ah ok 2015-12-26 11:51:00 token ring is dead. 2015-12-27 12:20:11 hi 2015-12-28 09:40:46 hey :) 2015-12-28 09:40:55 merry x-mas and so on :) 2015-12-28 09:52:11 hi 2015-12-28 09:52:17 all still in holidays, or? 2015-12-28 09:52:57 yes, and you should relax and enjoy the holidays too! :P 2015-12-28 09:53:30 I am already relaxed :) 2015-12-28 09:53:42 all kids are gone, finally :p 2015-12-28 14:10:24 both ncopa and fabled are not around, so I created #4979 without talking to them 2015-12-28 14:10:35 not a critical issue 2015-12-28 15:18:12 I don't think it'll get fixed for 2.x though ;) 2015-12-28 17:01:25 what do I do when my package which I am building needs a package which is in edge and testing? 2015-12-28 17:01:31 I get "ERROR: unsatisfiable constraints:" :( 2015-12-28 17:17:06 well, there is no way to get it out of edge before the next release 2015-12-28 17:59:06 can pkgver also have text? 2015-12-28 18:26:44 I could suggest http://pandas.pydata.org/ instead of sleuthkit 2015-12-28 18:27:02 from AL usage/purpose point of view 2015-12-28 20:23:41 or better the full stack, http://www.scipy.org/stackspec.html 2015-12-28 23:05:09 shafire: afaik no 2015-12-28 23:21:02 I need it :S 2015-12-29 07:52:55 hi guys _o/ 2015-12-29 07:57:24 hi 2015-12-29 07:57:33 whats the easiest way to create a patch? 2015-12-29 07:58:13 abuild unpack, change, and get patch diff? 2015-12-29 08:01:44 shafire, i use this way. 2015-12-29 08:01:58 abuild fetch && abuild unpack 2015-12-29 08:02:05 then go into the source dir 2015-12-29 08:02:10 git init . 2015-12-29 08:02:14 git add * 2015-12-29 08:02:20 git commit -m "." 2015-12-29 08:02:36 now, you have a git repository inside src 2015-12-29 08:02:42 sorry 2015-12-29 08:02:47 src/$pkgname-$pkgver 2015-12-29 08:02:52 but you get the point. 2015-12-29 08:03:06 Then, modify the file(s) you need 2015-12-29 08:03:33 and then git diff > ~/path/to/aports/repo/package/patchfile.patch 2015-12-29 08:03:43 good, thanks 2015-12-29 08:04:49 morning 2015-12-29 08:05:08 i'll be much away this week 2015-12-29 08:05:14 but will pop in once in a while 2015-12-29 08:05:23 how are everyone doing? 2015-12-29 08:05:28 how are things going? 2015-12-29 08:05:58 ncopa: I had an issue with the new release 2015-12-29 08:06:13 ok? 2015-12-29 08:06:32 #4975 2015-12-29 08:08:27 shafire: did you change your /etc/apk/repositories? 2015-12-29 08:08:41 no 2015-12-29 08:08:52 so it probably still has v3.2 2015-12-29 08:09:01 change that to v3.3 2015-12-29 08:09:09 and apk upgrade -U -a 2015-12-29 08:09:13 mh, it's not that 2015-12-29 08:09:17 I also used a fresh iso 2015-12-29 08:09:22 without upgrade 2015-12-29 08:09:31 and no apk cache? 2015-12-29 08:09:34 clandmeter and two others could reproduce 2015-12-29 08:09:37 ok 2015-12-29 08:09:44 then should i be able to do so too 2015-12-29 08:10:14 I opened the issue after clandmeter could reproduce 2015-12-29 08:10:21 wanted to be sure it's not me 2015-12-29 08:10:37 so i suppose, simply installing alpine and setting up apkcache is broke 2015-12-29 08:11:34 19:35 I blame this: http://git.alpinelinux.org/cgit/mkinitfs/commit/initramfs-init.in?id=6aba975e73e7cea833640e8db180c0add01ecf33 2015-12-29 08:15:34 ncopa: yes its broken 2015-12-29 08:15:53 seems it mounts it differently in initramfs 2015-12-29 08:15:59 so it cannot find cache dir 2015-12-29 08:17:18 sounds plausible 2015-12-29 08:17:39 i also see rpi images are broken, i think fabled fixed the issue, not sure the images are updated already. 2015-12-29 08:17:52 clandmeter do you think you (or someone else ) could give the info in the ticket? 2015-12-29 08:18:08 eg, how does apk expect it to be mounted 2015-12-29 08:18:16 and how is it actually mounted in initramfs 2015-12-29 08:18:44 yes fabled emailed me about the broken rpi stuff 2015-12-29 08:18:45 and the fix 2015-12-29 08:18:48 need make new image 2015-12-29 08:18:50 hum 2015-12-29 08:18:51 by device name in initramfs 2015-12-29 08:19:05 so, like /media/sda1 2015-12-29 08:19:08 i think the real problem is if oyu plugin some device and device names reorder 2015-12-29 08:19:41 and also using usbdisk 2015-12-29 08:20:08 maybe it would be better to mount it with uuid? 2015-12-29 08:20:56 if it would mount it both in initramfs and after with uuid, the mount point would always be the same. (if thats possible) 2015-12-29 08:25:11 ncopa: I also had to fix the latest-stable symlink on mirror, and update the www for new releases. 2015-12-29 08:25:28 i think some notes about how to maintain www would be usefull. 2015-12-29 08:26:06 i guess they are updated periodically? 2015-12-29 08:42:22 ok 2015-12-29 08:42:30 yes, i forgot about latest-stable 2015-12-29 08:42:34 it needs to be done manually 2015-12-29 08:42:52 "update the www for new releases"? 2015-12-29 08:43:01 how did you do that? 2015-12-29 08:43:43 it finds the info in some yaml file 2015-12-29 08:43:44 hum 2015-12-29 08:44:08 we should have some more text about the different variants on alpinelinux.org/downloads 2015-12-29 08:44:27 people does not discover the hover text 2015-12-29 08:44:53 i also saw this: 2015-12-29 08:44:56 https://news.ycombinator.com/item?id=10782897 2015-12-29 08:45:09 which is a bit interesting 2015-12-29 08:45:25 people dont know or dont see alot of the work that is done 2015-12-29 08:45:32 we need make some of the work more visible 2015-12-29 08:48:31 raising awareness is important 2015-12-29 08:48:48 yes 2015-12-29 08:48:50 but it takes time away from coding and doing tech stuff 2015-12-29 08:48:57 so we need more people to do that :P 2015-12-29 08:49:01 yes :) 2015-12-29 08:49:08 but it is kind of nice too 2015-12-29 08:49:20 becuase it means you dont need to be a coder to contribute 2015-12-29 08:49:25 yes 2015-12-29 08:49:38 that's why we need non-coders to do it for coders :P 2015-12-29 08:49:43 and it means that we *need* non-coders 2015-12-29 08:49:48 yes 2015-12-29 08:50:11 i would also like to improve the documentation 2015-12-29 08:50:19 i don tknow how 2015-12-29 08:50:51 i think that might be somehing for the people in between 2015-12-29 08:51:00 who might have some coding skills 2015-12-29 08:51:05 or at least readonly skills 2015-12-29 08:51:18 translate it into useful documentation for non-coders 2015-12-29 08:51:49 https://developer.atlassian.com/blog/2015/08/minimal-java-docker-containers/ 2015-12-29 08:51:51 :) 2015-12-29 08:52:52 they use oracle jdk8 on alpine? 2015-12-29 08:52:56 I thought that's not possible 2015-12-29 08:54:57 i think they do 2015-12-29 08:55:05 i think they also copy glibc inthere 2015-12-29 08:55:24 i mean how does the www get updated? 2015-12-29 08:55:28 i know its based on git 2015-12-29 08:55:39 i thought i had to push to update it 2015-12-29 08:55:51 some post push script 2015-12-29 08:56:01 but that wouldnt make sence to update the news. 2015-12-29 08:56:58 push to this git repo: http://git.alpinelinux.org/cgit/alpine-mksite/ 2015-12-29 08:57:04 i did 2015-12-29 08:57:05 there are 2 branches 2015-12-29 08:57:09 but it didnt update 2015-12-29 08:57:17 'master' makes it visible on wwwtest.alpinelinux.org 2015-12-29 08:57:29 yes i thought so 2015-12-29 08:57:34 when push to the branch 'production' 2015-12-29 08:57:43 it goes to alpinelinux.org 2015-12-29 08:57:51 hover text needs to be changed on the download website 2015-12-29 08:57:56 and the extended title is missing 2015-12-29 08:58:00 yup 2015-12-29 08:58:04 we need fix that 2015-12-29 08:58:09 but it doesnt update on a push 2015-12-29 08:58:16 so there must be something else 2015-12-29 08:58:20 then something is broke 2015-12-29 08:58:28 its a mqtt message that deals with it 2015-12-29 08:58:38 it did in the end 2015-12-29 08:58:46 i just dont know what made it change 2015-12-29 10:14:17 Am I eligibled to get write access to the community repo? 2015-12-29 11:15:04 is there a reason why /usr/local/lib is in the /etc/ld-musl-${ARCH}.path by default? 2015-12-29 12:46:10 what to do with all the compiler warnings when compiling? 2015-12-29 12:46:37 mostly warning: called from here [-Winline] 2015-12-29 12:51:43 I ignore them and move on with my life 2015-12-29 13:01:39 good choice 2015-12-29 13:12:24 same as barthalion here 2015-12-29 13:12:26 :) 2015-12-29 13:20:59 do we move all files from lib64 to lib? 2015-12-29 13:31:19 no idea, catch ncopa or fabled for that 2015-12-29 13:31:50 shafire: yes, we normally do 2015-12-29 13:42:27 barthalion: btw. you wanted to talk about something after 3.3 was out. that still relevant? 2015-12-29 13:53:44 does abuild delete something? 2015-12-29 13:53:59 I am moving *.la files to the -dev package 2015-12-29 13:54:02 mv -v shows me the moving 2015-12-29 13:54:16 but I cannot find them in -dev anymore 2015-12-29 13:55:17 also it deletes symlinks 2015-12-29 13:55:26 from lib 2015-12-29 13:57:38 and what is the option in options="!strip" doing? 2015-12-29 13:57:50 there is no doc in http://wiki.alpinelinux.org/wiki/Creating_an_Alpine_package 2015-12-29 14:02:01 !strip will prevent abuild from stripping symbols from compiled binaries 2015-12-29 14:34:42 I think "after 3.3 is out" means "after 3.3 is out, we've put the fires out, fixed the obvious bugs and taken a few days of vacation" 2015-12-29 14:37:12 well, turns out it was a short discussion 2015-12-29 15:09:23 Tracing dependencies... >>> ERROR: libncursesw.so.5: path not found 2015-12-29 15:09:35 There is only a package which provides libncursesw.so 2015-12-29 15:10:14 I tried to create a symlink at the package function in APKBUILD 2015-12-29 15:13:40 libncursesw.so.5 is provided by ncurses5-widec-libs 2015-12-29 15:14:26 how did you find it? I used the package search 2015-12-29 15:14:46 https://pkgs.alpinelinux.org/contents?filename=%25libncursesw.so.5%25&path=&pkgname=&repo=all&arch=x86_64 2015-12-29 15:14:48 now it works 2015-12-29 15:15:05 before there was no result :( 2015-12-29 15:15:13 thanks 2015-12-29 15:18:36 sure 2015-12-29 16:06:51 I have an update for nginx-naxsi to send 2015-12-29 16:12:03 it enables https://www.nginx.com/blog/thread-pools-boost-performance-9x/ 2015-12-29 18:47:32 what's the new way to create the isos? 2015-12-29 19:07:43 I may have just submitted the only use of !archcheck in the repos 2015-12-29 19:09:36 link 2015-12-29 19:10:27 %1321 ? 2015-12-29 19:11:00 :D 2015-12-29 19:11:02 what is it doing? 2015-12-29 19:11:06 where do you have this command from? 2015-12-29 19:13:52 it's something that came up a while ago about packages that have architecture deps but no binaries I think? 2015-12-29 19:17:02 I just never got around to it 2015-12-29 19:17:27 oh hey when'd apk get --no-cache 2015-12-29 19:17:43 why is !archcheck and arch=all better than arch=noarch 2015-12-29 19:18:26 #4950 2015-12-29 19:18:29 #4905 2015-12-29 19:18:36 it's really not 2015-12-29 19:20:11 but it's legitimately architecture-dependent 2015-12-29 19:20:34 I don't think anything gets /done/ about that but like, you can't just build it anywhere and expect it to work everywhere 2015-12-29 19:21:02 although that's mostly because I went and got rid of the files for other archs to save space 2015-12-29 19:21:09 but that's a goal, so 2015-12-29 20:16:29 I built an openvz iso image 2015-12-29 20:16:55 how is the cgroup filesystem mounted? 2015-12-29 20:23:00 shafire: it's mounted by the mount_cgroups() function of the sysfs service 2015-12-29 20:23:09 take a look at /etc/init.d/sysfs for details 2015-12-29 20:26:53 ls /sys/fs/cgroup is empty that's why it's not mounted 2015-12-29 21:03:22 nmeum: the initial "cgroup_root 10.0M 0 10.0M 0% /sys/fs/cgroup" is created by this ? 2015-12-29 21:05:06 https://gitweb.gentoo.org/proj/openrc.git/tree/init.d/sysfs.in#n113 2015-12-29 21:07:36 I mean when AL boots, its has cgroup_root mounted, but I could not find what calls init.d/sysfs 2015-12-29 21:13:09 nm, I was thinking something else 2015-12-29 21:16:03 /etc/runlevels/sysinit/mdev: need sysfs dev-mount 2015-12-30 06:59:38 good morning 2015-12-30 07:10:00 where can I find the possible values for options in APKBUILD? 2015-12-30 07:15:59 probably only in abuild itself 2015-12-30 07:16:38 shafire: 2015-12-30 07:16:39 http://wiki.alpinelinux.org/wiki/APKBUILD_Reference 2015-12-30 07:16:43 this would not suffice? 2015-12-30 07:17:19 options Build-time options for the package. Can be: !strip - to avoid stripping symbols from binaries. 2015-12-30 07:17:34 but there is not only !strip or? 2015-12-30 07:17:44 Yesterday someone told there is also !archcheck 2015-12-30 07:17:56 ah ok 2015-12-30 07:17:57 sorry 2015-12-30 07:18:11 misunderstood. 2015-12-30 07:19:11 I have the feeling that abuild does some magic (at least thats my feeling) 2015-12-30 07:19:30 symlinks are deleted, files I moved are being deleted? 2015-12-30 07:20:45 https://plain.greata.uk/options.txt these are the ones actually being used 2015-12-30 07:22:24 thanks 2015-12-30 07:25:52 okay so !configsub isn't actually useful anymore 2015-12-30 07:26:27 and !libc_uclibc deprecated? 2015-12-30 07:45:36 I mean, I'm p sure it's not useful in the one package still marked that but I didn't touch it 2015-12-30 09:17:36 do all .so files in usr/lib go to the -dev package? 2015-12-30 09:21:04 so don't go anywhere 2015-12-30 09:21:16 -dev keeps headers and maybe static libraries, if enabled 2015-12-30 09:25:54 mh 2015-12-30 09:26:07 because .so stayed in the packaged, but the symlinks moved to -dev 2015-12-30 09:26:11 -d 2015-12-30 09:29:20 can abuild show me, because of which files he traced a -dev dependency? 2015-12-30 11:56:54 error: #warning usage of non-standard #include is deprecated 2015-12-30 11:57:13 how to correct this? 2015-12-30 12:40:29 don't #include ? :P 2015-12-30 12:41:57 just remove? 2015-12-30 13:34:08 shafire: its a warning, just leave it. 2015-12-30 14:54:42 any reason to not have enabled GPG support in sylpheed ? 2015-12-30 14:55:00 so that I know if it's worth sending a patch or not :) 2015-12-30 14:57:05 check git log for that pkg, that should give you some info. 2015-12-30 14:57:37 if there is a bug regarding it, it should have a ref. 2015-12-30 14:59:29 clandmeter: seems not apparently 2015-12-30 14:59:45 then go ahead 2015-12-30 15:00:54 testing binary right now see if it works as expected 2015-12-30 15:05:09 alpine becomes popular or not? 2015-12-30 15:05:13 a lot of blog posts lately 2015-12-30 15:06:01 link? 2015-12-30 15:07:00 the link from ncopa yesterday 2015-12-30 15:07:01 https://news.ycombinator.com/item?id=10782897 2015-12-30 15:11:25 thx for the link 2015-12-30 15:11:56 shafire: noticed it recently 2015-12-30 15:12:12 clandmeter: ok sent patch 2015-12-30 16:01:30 I will bring out a new flavor soon 2015-12-30 17:31:15 Hi! I have a bunch of aports patches to get the OCaml OPAM package manager working reliably; https://github.com/alpinelinux/aports/compare/master...avsm:ocaml-ports.diff 2015-12-30 17:31:33 Not sure what the current place to send the patch is, as there havent been many aports patches sent to alpine-devel in the past few months... 2015-12-30 17:31:40 Or are github PRs accepted? 2015-12-30 17:32:07 (and how does a package get promoted from testing? The OCaml port works pretty well for me) 2015-12-30 17:33:09 avsm: aport patches are sent to alpine-aports 2015-12-30 17:33:42 aha! thanks chris| — i didnt find that in http://wiki.alpinelinux.org/wiki/Alpine_Linux:Mailing_lists 2015-12-30 17:34:20 yeah, it's on http://wiki.alpinelinux.org/wiki/Creating_patches 2015-12-30 17:36:03 got it, thanks! 2015-12-30 17:36:19 sure 2015-12-30 19:40:30 avsm: oh, you got around to packaging the stuff opam wants 2015-12-30 19:40:45 for what it's worth 2015-12-30 19:41:08 I deliberately avoided putting a bunch of things in opam's depends= 2015-12-30 19:41:37 ammunta: if ocaml is in the depends, then camlp4 should be as well 2015-12-30 19:41:46 but both can be skipped and opam can be initialised without a system compiler 2015-12-30 19:42:04 generally we've found that making the opam package depend on a system compiler is a reasonable default though 2015-12-30 19:42:30 but it could make more sense for alpine's minimal ethos to skip the dependency entirely. 2015-12-30 19:43:00 I skipped out on the depends but kept it as a buld dependency 2015-12-30 19:43:17 to avoid unnecessary bootstrapping 2015-12-30 19:44:08 is the ocaml apk fully statically linked to musl? 2015-12-30 19:44:36 if it is, then having no runtime depends does make more sense 2015-12-30 19:44:54 (also, do you use opam on alpine btw? I'm creating this port for a bulk build infrastructure for opam) 2015-12-30 19:45:04 dynamically linked, I think 2015-12-30 19:45:21 haven't used it for a while 2015-12-30 19:45:46 I just packaged it a while ago 2015-12-30 19:47:30 ammunta: cool, will look into statically linking it. Would be really handy to be able to build standalone binaries for containers using this version of OPAM 2015-12-30 19:48:01 although I suspect I should at least put a depends= line in for wget 2015-12-30 19:48:25 i can do that…just sending the full PR now 2015-12-30 19:49:04 since it flat-out doesn't function without it, and with busybox wget it makes assumptions that aren't true 2015-12-30 19:49:25 ah, i should fix that in opam itself then 2015-12-30 19:49:49 i have a patch for a builtin cohttp/ocaml fetcher somewhere that removes the dependency 2015-12-30 19:49:59 ah, that would also be good 2015-12-30 19:51:44 hm I also didn't check whether the package fetching opam does before building is done safely 2015-12-30 19:52:47 ah, it's md5summed 2015-12-30 19:53:02 that's ... probably not good enough tbh 2015-12-30 19:56:45 there's package signing coming up soon; http://opam.ocaml.org/blog/Signing-the-opam-repository/ 2015-12-30 19:58:53 has ocaml good concurrency support, avsm? 2015-12-30 19:58:53 with a clean aports checkout with the new packages, is there an abuild that will recursively build packages from source that are in aports but not yet binary packages? 2015-12-30 19:59:01 to conveniently build all the new dependencies for aspcud 2015-12-30 20:13:50 oh right I think the main reason I didn't get around to writing up the other rec ports is because we didn't have a convention for ocaml packages (because we didn't have any) and I didn't want to make one up 2015-12-31 01:36:21 <_mjones> avsm: yes, one of the -k or -K will recursively build aports that are not binaries.