2015-07-01 00:25:21 crow: yeah but it totally misses the point. the really nasty parts of openssl are in libcrypto. 2015-07-01 06:00:14 kaniini well then they are doing work on wrong place :/ 2015-07-01 09:11:55 hey friends :) 2015-07-01 09:16:51 hi leo-unglaub 2015-07-01 09:17:24 did you manage to cleanup your room after that long discussion? :) 2015-07-01 09:17:54 clandmeter: no, i missed it ... and had to take my parents for there birthday dinner to a restaurant *g* 2015-07-01 09:19:37 lol 2015-07-01 09:20:11 its all kaniini faulth, lets blame it on him :) 2015-07-01 09:24:16 yeah, all his fault *g* 2015-07-01 09:24:25 hoe dare he stand up for his opinion ... 2015-07-01 09:24:30 :) 2015-07-01 09:53:12 are the buildlogs available somwhere? 2015-07-01 09:53:25 from the buildserver 2015-07-01 09:54:09 http://bld1.alpinelinux.org/buildlogs/ 2015-07-01 09:54:14 http://bld2.alpinelinux.org/buildlogs/ 2015-07-01 09:55:24 thanks 2015-07-01 09:58:10 np 2015-07-01 11:23:49 i'm wnondering how much bad is an APKBUILD containing this: 2015-07-01 11:23:50 pkgver=20150601 2015-07-01 11:23:50 _pkgver=$(echo $pkgver | sed -e 's/./-&/5' -e 's/./-&/8') 2015-07-01 11:24:31 would be possible to change 20150601 in 2015-06-01 atomically? 2015-07-01 11:25:00 some shell substring wizard can give an advice? 2015-07-01 11:29:00 umh 2015-07-01 11:29:04 export T=20150601 2015-07-01 11:29:06 echo ${T:0:4}-${T:4:2}-${T:6:2} 2015-07-01 11:29:11 this works 2015-07-01 11:31:37 _pkgver=2015-06-01 2015-07-01 11:31:54 pkgver=${_pkgver/-//} 2015-07-01 11:32:20 i tought about that 2015-07-01 11:32:29 but pkgver cannot be in this format 2015-07-01 11:32:32 abuild fails 2015-07-01 11:32:40 so you need to start from 20150601 2015-07-01 11:32:42 or whatever was the syntax for removing every occurence 2015-07-01 11:32:49 let's fix abuild instead? 2015-07-01 11:33:15 what's shorter? 2015-07-01 11:33:20 *faster 2015-07-01 11:33:29 yours way 2015-07-01 11:33:39 the second one 2015-07-01 11:33:51 yeah 2015-07-01 11:33:51 _pkgver= ${pkgver:0:4}-${pkgver:4:2}-${pkgver:6:2} 2015-07-01 11:33:58 but leave a comment what it does 2015-07-01 11:34:14 # convert YYYYMMDD to YYYY-MM-DD 2015-07-01 11:34:23 save someone a wtf 2015-07-01 11:34:29 yeah 2015-07-01 11:36:07 _pkgver= ${pkgver:0:4}-${pkgver:4:2}-${pkgver:6:2} 2015-07-01 11:36:15 does that work with busybox ash? 2015-07-01 11:36:19 yes 2015-07-01 11:36:45 oh, you mean the space 2015-07-01 11:36:52 no, it's a typo 2015-07-01 11:36:58 _pkgver=${pkgver:0:4}-${pkgver:4:2}-${pkgver:6:2} 2015-07-01 11:37:02 This works ^^^^ 2015-07-01 11:37:13 no space between = and $ 2015-07-01 11:37:17 you could also do: pkgver="YYYY.MM.DD"; _pkgver=${pkgver/.//-} 2015-07-01 11:37:47 right. This is even better. 2015-07-01 11:38:01 sorry 2015-07-01 11:38:06 pkgver="YYYY.MM.DD"; _pkgver=${pkgver//./-}; echo $_pkgver 2015-07-01 11:38:07 YYYY-MM-DD 2015-07-01 11:38:23 Just started from the assumption that pkgver should be the same of the package (same format, i mean) 2015-07-01 11:38:24 i had the slashed wrong in first example 2015-07-01 11:38:31 it's libre2 package. 2015-07-01 11:39:01 https://github.com/google/re2/releases 2015-07-01 11:39:14 oh 2015-07-01 11:39:17 yeah, i think pkgver=2015.07.01 looks nicer 2015-07-01 11:39:18 a new one :-/ 2015-07-01 11:39:56 royger: you have xen experience, do you know if it is common to start xen domains in gun screen? 2015-07-01 11:40:02 i see that gentoo does that 2015-07-01 11:40:08 gnu* screen 2015-07-01 11:40:12 they have support for it 2015-07-01 11:40:18 which alpine copied 2015-07-01 11:40:27 i even added support for tmux 2015-07-01 11:40:31 but i kind of regret that now 2015-07-01 11:54:57 ncopa: you shouldn't need to start them in screen or tmux, xl daemonizes itself 2015-07-01 11:55:19 does anything actually use libre2? 2015-07-01 11:56:00 chromium 2015-07-01 11:56:24 oh 2015-07-01 11:56:27 alright 2015-07-01 11:57:10 ncopa: and you can log all the console output to /var/log/xen/console/ or something similar, just needs to enable XENCONSOLED_TRACE=guest in xencommons init script 2015-07-01 12:08:21 ncopa: you sayed your firefox segfaults now on youtube 2015-07-01 12:08:25 is that still current? 2015-07-01 12:10:41 royger: thanks. i think i'll just remove gnu screen and tmux support. 2015-07-01 12:10:58 leo-unglaub: firefox youtbube works again 2015-07-01 12:11:08 with the recent gst-libav1 changes 2015-07-01 12:15:11 ah, you use gstreamer for irefox vieos 2015-07-01 12:40:22 does anybody know why im having getcontext issues with musl? 2015-07-01 12:42:01 hmm not implemented? http://wiki.musl-libc.org/wiki/Open_Issues#ucontext.h 2015-07-01 12:58:43 i thought we had a patch for it= 2015-07-01 12:58:45 ? 2015-07-01 12:59:05 nope. no one did it so far 2015-07-01 12:59:06 musl-dev provides ucontext.h 2015-07-01 12:59:08 ok 2015-07-01 12:59:19 yes, they added the header for some weird reason. but it's not implemented 2015-07-01 13:02:04 ncopa: we have one for uclibc 2015-07-01 13:02:44 yes, we had it patched in for musl 2015-07-01 13:02:46 err 2015-07-01 13:02:47 uclibc 2015-07-01 13:04:03 is there a macro to detect it? 2015-07-01 13:05:57 i guess just removing it will not be accepted from upstream. 2015-07-01 13:05:57 detect ucontext? not that i know 2015-07-01 13:08:12 if it aint posix anymore, i guess i can check for posix? 2015-07-01 13:09:17 i think it was removed in posix yes 2015-07-01 13:24:13 ncopa: what produces asus_atk0110.ko kernel module ? 2015-07-01 13:24:31 is there an gigabyte equivalent ? 2015-07-01 13:25:43 I am unable to get "Adapter: ACPI interface" via `sensors` on H81 MB 2015-07-01 13:26:44 i dont know? kernel srouces? 2015-07-01 13:28:20 yes, I guess 2015-07-01 13:28:29 modinfo asus_atk0110 2015-07-01 13:28:46 parm: new_if:Override detection heuristic and force the use of the new ATK0110 interface (bool) 2015-07-01 13:29:26 maybe something similar for Gigabyte MBs did not get to kernel source 2015-07-01 13:29:36 yet 2015-07-01 13:31:39 and I am loosing interest in esupport site of gigabyte, does not seem to work 2015-07-01 13:31:55 would try their forums 2015-07-01 13:36:54 :) 2015-07-01 13:38:09 vkris: what do you want to know from gigabyte? 2015-07-01 13:38:40 try sensors-detect && sensors 2015-07-01 13:39:04 is that a question? 2015-07-01 13:40:49 hmmm... I mean are y ou able to sensors-detect correctly ? 2015-07-01 13:40:59 vkris: i mean, what did you ask our (GIGABYTE) tech support? 2015-07-01 13:42:15 hmmm... ? 2015-07-01 13:42:39 ok would try the gigabyte forums ;) 2015-07-01 13:43:01 i dont think you get the point... 2015-07-01 13:45:06 is it possible that the drivers similar to asus_atk0110.ko hasmade into kernel source yet 2015-07-01 13:45:12 ? 2015-07-01 13:47:02 i dont think GIGABYE submits patches to kernel. atleast i didnt see any colleague ever do it. 2015-07-01 13:48:43 then how to get "ACPI interface" values, seems it is viewable via bios ? 2015-07-01 13:50:55 GIGABYTE does not support Linux, except for server business unit. 2015-07-01 13:51:05 hmmm 2015-07-01 15:12:36 have a nice weekend 2015-07-01 15:12:40 i'm back on tuesday 2015-07-01 15:14:23 you too 2015-07-01 15:51:04 kaniini: ping 2015-07-01 16:01:17 ACTION sits back and grabs some popcorn 2015-07-01 20:55:45 leo-unglaub: hi 2015-07-01 20:56:41 hey kaniini 2015-07-01 20:56:52 i was thinking about the mbedtls wrapper 2015-07-01 20:56:56 and what it should do 2015-07-01 20:57:20 and i would love your opinion on the following question 2015-07-01 20:57:50 if an application do to some reasons still wants to use insecure ciphres, ... should a wrapper allow that? 2015-07-01 20:58:06 in my opinion no, but that would kill the application in most cases 2015-07-01 20:58:31 i think it should for compatibility reasons, but perhaps it should write angry messages to stderr 2015-07-01 21:00:20 so you would allow SSL instead of TLS in a wrapper? 2015-07-01 21:00:49 it may be required for some services 2015-07-01 21:01:05 a wrapper is a wrapper, it is not an artistic statement. 2015-07-01 21:01:30 SSLv2/SSLv3 may have problems, but some legacy services do not speak TLS. 2015-07-01 21:01:43 we cannot control what those services do 2015-07-01 21:02:22 for example, my bank (which is a large bank in america) does not speak TLS on their online banking site 2015-07-01 21:02:30 should they fix it -- of course they should 2015-07-01 21:02:35 will they do so because i say so, probably not 2015-07-01 21:03:26 now, i think that insecure settings should be *explicit* 2015-07-01 21:03:51 i.e. the default should be at least TLS 1.2 with AES-128 in GCM mode 2015-07-01 21:04:44 in case of polarssl (i hate the name mbedtls) that would be a problem 2015-07-01 21:04:57 becasue he told me that SSL will be removed in the next version 2015-07-01 21:05:03 according to RFC 7568 2015-07-01 21:05:08 https://tools.ietf.org/html/rfc7568 2015-07-01 21:05:46 rfc7568 only deprecate SSLv3 2015-07-01 21:05:51 SSLv2 is more common 2015-07-01 21:06:35 ? 2015-07-01 21:06:53 do you know anyone using sslv2? 2015-07-01 21:07:36 well, TLS is backwards compatible with SSLv2, i.e. something that speaks TLS can speak to an SSLv2 with degraded security 2015-07-01 21:07:42 as far as i know even firefox and chrome kicked that one out 2015-07-01 21:07:44 SSLv3 is an incompatible update :p 2015-07-01 21:07:56 however. 2015-07-01 21:08:06 i think requiring TLS is probably ok. 2015-07-01 21:08:55 yeah, fully agreed 2015-07-01 21:09:11 i am currently hacking a little bit on such a wrapper 2015-07-01 21:10:06 hopefully under a license which allows it to be reused anywhere 2015-07-01 21:10:37 my default template contains the GPL3 2015-07-01 21:10:41 but i dont care about that 2015-07-01 21:10:57 since polarssl is moving to apache license, it should probably use apache license 2015-07-01 21:11:07 which is basically compatible with whatever 2015-07-01 21:11:13 for me a license is just a text file that i drop in to shup up lawyers 2015-07-01 21:11:45 sure, however, users building things with alpine have their own lawyers to shut up too :P 2015-07-01 21:11:54 *g* 2015-07-01 21:12:20 i think if we get this wrapper thing going finding a license.txt will be the smallest problem *g* 2015-07-01 21:12:42 well, it doesn't have to be a 100% compatible thing 2015-07-01 21:12:47 just enough to make porting feasible 2015-07-01 21:12:51 i could drop in the microsoft eula *g* would be the first time people would read that thing *g* 2015-07-01 21:13:21 i am focusing my wrapper completaly on the TLS part 2015-07-01 21:13:35 i leave out any key generation, .... thing 2015-07-01 21:15:08 that is probably a good starting point. 2015-07-01 21:15:34 i think a first proof of concept is getting curl running with minimal changes 2015-07-01 21:15:53 then work up slowly to stuff like openssh 2015-07-01 21:16:29 curl is a little bit of cheat, because it supports polarssl during compiletime *g* 2015-07-01 21:18:57 but it will not work without recompiling packages 2015-07-01 21:19:08 complete abi compat is impossible 2015-07-01 21:21:33 i did not say abi compat 2015-07-01 21:21:41 just curl-openssl building against the portability layer 2015-07-01 21:21:44 ;) 2015-07-01 21:21:53 i am just saying *g* 2015-07-01 21:22:40 i have to go and hit a burgerking, its nearly midnight 2015-07-01 21:22:42 see you laster 2015-07-01 21:22:44 later 2015-07-01 21:23:47 what is history limit from ash under alpinelinux? i hate it when i dont have "all what i typed" 2015-07-01 21:50:33 crow - my ~/.ash_history is 830 lines 2015-07-02 02:12:28 http://puu.sh/iKb78.txt 2015-07-02 02:12:32 anyone know what's going on here? o.O 2015-07-02 02:12:46 i changed 3.1 to 3.2 in /etc/apk/repositories and that happens when i try to upgrade 2015-07-02 06:23:55 BitL0G1c well here 871 but still i dont see all stuff, maybe because of tmux is used? 2015-07-02 06:24:06 and the command are not saved in history 2015-07-02 11:16:15 ncopa: ping 2015-07-02 11:17:16 i think there went something wrong with the latest pidgin update 2015-07-02 11:17:27 because now the jabber account type is missing 2015-07-02 11:33:05 clandmeter: why use cdn files on pkgs.alpinelinux.org, save bandwidth ? 2015-07-02 13:03:12 humm, is it a known issue that xf86-video-vesa crashes 2015-07-02 13:03:42 Assertion failed: key->initialized (../include/privates.h: dixGetPrivateAddr: 122) 2015-07-02 13:44:49 hi 2015-07-02 13:54:07 ACTION uses something other than virtualbox 2015-07-02 13:55:40 how do I create an iso from a specific version? 2015-07-02 13:56:09 Not possible, or? - without rebuilding all 2015-07-02 13:56:55 shafire: what is wrong with the iso's we have? 2015-07-02 13:57:15 I need a different modloop 2015-07-02 13:57:52 and you really need a specific version? 2015-07-02 13:59:27 Say, I want to use 3.2 and I am a week to late after release - then the kernel is already updated and I do not have the same release anymore 2015-07-02 14:05:53 i dont understand the issue 2015-07-02 14:10:11 shafire: alpine-iso will make an iso for you from the current used repo. edge or any stable repo (or custom repo) 2015-07-02 14:11:22 I know 2015-07-02 14:42:45 clandmeter: why is sandbox disabled for chromium ? 2015-07-02 15:18:41 kaniini: it doesnt work 2015-07-02 15:18:48 and im trying to find out why 2015-07-02 15:19:20 im a bit lost in that part 2015-07-02 15:19:35 maybe its musl related? 2015-07-02 15:31:01 clandmeter: what exactly happens 2015-07-02 15:31:38 it spits out some error msg. i didnt write it down (not on alpine atm) 2015-07-02 15:33:26 kaniini: i think it works on gentoo, so i guess its musl 2015-07-02 15:33:54 https://code.google.com/p/chromium/wiki/LinuxSandboxing 2015-07-02 15:34:21 i remember i got a msg about seccomp-bpf 2015-07-02 15:34:45 but disabling still prevents sandboxing 2015-07-02 15:37:12 so sandboxing is disabled, even though the chrome-browser complains that it's enabled? 2015-07-02 15:39:03 it complains it is disabled 2015-07-02 15:47:55 what provides xmllint? 2015-07-02 15:48:09 clandmeter: i must have misread it, sorry 2015-07-02 15:48:49 ScrumpyJack: http://pkgs.alpinelinux.org/contents?filename=xmllint&path=&pkgname=&arch=x86 2015-07-02 15:49:23 thanks 2015-07-02 15:54:09 bookmark it ;-) 2015-07-02 17:08:02 make succeeds under musl, but fails if i add --prefix=/usr to the configure script 2015-07-02 17:10:41 why would that happen? 2015-07-02 17:11:25 clandmeter, please: http://sprunge.us/MFBW 2015-07-02 17:12:12 clandmeter, forget it ^^ 2015-07-02 17:22:01 clandmeter, this one: http://sprunge.us/YLUT 2015-07-02 17:22:01 <- the previous one was with wrong version number :) 2015-07-02 17:29:02 ncopa, ref #4422 you are the mainteiner. Do you mind if I bump it? 2015-07-02 18:37:12 kernel module rtl8192eu seems missing in AL v3.1.2, added on bugs.a.o 2015-07-02 18:37:30 but not tried on edge yet 2015-07-03 16:34:44 any information related to Alpine vs ZFS ? 2015-07-03 16:38:46 alacerda: we used to have zfs in aports 2015-07-03 16:39:02 alacerda: I guess we still have 2015-07-03 16:39:23 http://pkgs.alpinelinux.org/: no item found... 2015-07-03 16:39:41 ahhw 2015-07-03 16:39:44 http://pkgs.alpinelinux.org/contents?filename=zfs&path=&pkgname=&arch=x86 2015-07-03 16:39:52 in testing 2015-07-03 16:44:51 thx barthalion 2015-07-03 17:23:10 alacerda: you're welcome 2015-07-03 17:37:23 alacerda: the biggest problem with zfs is that it requires rpc, which is in glibc, but not musl, and libtirpc is license incompatible with zfs. so legally it doesn't work 2015-07-03 18:30:05 i think the strongswan package is compiled against an old openssl... stopped working 2015-07-03 18:30:59 was it upgraded recently? 2015-07-03 18:32:14 2015-06-15 2015-07-03 18:32:26 vs 2015-06-11 2015-07-03 18:32:36 if you say it can help, I'll bump it 2015-07-03 18:34:30 not sure but when i told their channel about the problem i'm having they said it's probably compiled against old openssl 2015-07-03 18:35:02 (it stopped working after i did apk upgrade) 2015-07-03 18:37:26 pushed 2015-07-03 18:39:41 and built 2015-07-03 18:39:59 are we actually talking about edge or stable 2015-07-03 18:40:01 ? 2015-07-03 18:46:33 Mp5shooter: ↑ 2015-07-03 18:46:46 oh uhh 2015-07-03 18:47:08 v3.2 is the repo i use 2015-07-03 18:47:17 alright, let's check this one too 2015-07-03 18:51:13 Mp5shooter: pushed as well 2015-07-03 18:51:24 thank you 2015-07-03 18:51:35 Mp5shooter: should be there on nl.al.o already 2015-07-03 18:52:01 oh barthalion i forgot about this 2015-07-03 18:52:13 after i changed my repo to v3.2, this happens when i do apk upgrade.. http://puu.sh/iM7o8.txt 2015-07-03 18:52:33 do you know what's happening there? o.O 2015-07-03 18:52:55 uh, let me check that tomorrow 2015-07-03 18:53:07 oh ok 2015-07-03 18:53:49 I don't want to push anything more complicated than pkgrel bump after beer(s) 2015-07-03 18:53:54 haha 2015-07-03 18:56:19 although it looks like just cups-libs needs a rebuild 2015-07-03 18:57:13 hm, not really 2015-07-03 18:57:55 Mp5shooter: mind dropping me 'apk info' somewhere? I'll try to fix it tomorrow morning 2015-07-03 18:59:07 sure sec 2015-07-03 18:59:50 barthalion: http://puu.sh/iM7P6.txt 2015-07-03 19:00:29 thanks 2015-07-03 19:00:36 TY 2015-07-03 19:00:37 oops caps 2015-07-03 19:13:32 Mp5shooter: could you try now? 2015-07-03 19:13:40 sure 2015-07-03 19:14:00 looks like some leftover from previous release, but to be honest, I don't know apk-tools very well 2015-07-03 19:14:23 looks like it's updating now 2015-07-03 19:14:23 :D 2015-07-03 19:14:29 (y) 2015-07-03 19:14:36 thank you very much 2015-07-03 19:14:48 after this i shall try strongswan 2015-07-03 19:15:03 np 2015-07-03 19:15:35 we have an idiom in Poland, "a fool always has luck" :P 2015-07-03 19:16:44 haha :P 2015-07-03 19:26:16 it always impress me how apk-tools prevents all kinds of breakage 2015-07-03 19:26:22 haha yeah 2015-07-03 19:26:51 slowly doing that in Arch, but I'll probably never convince everyone to rely more on sodeps than just packages 2015-07-03 19:27:35 :( 2015-07-03 19:57:00 mmm 2015-07-03 19:57:00 http://puu.sh/iMbgR.txt 2015-07-03 19:57:06 o.O 2015-07-03 19:58:04 sorry, I know nothing about strongswan 2015-07-03 19:58:13 asking them about that now 2015-07-03 19:58:19 why did i upgrade haha 2015-07-03 19:58:25 ;) 2015-07-03 19:58:42 freeradius broke on me too lol 2015-07-03 20:00:22 looks like freeradius got split up into multiple packages 2015-07-03 20:03:28 seems so 2015-07-03 20:03:39 anything to fix on alpine side? 2015-07-03 21:14:09 not sure barthalion 2015-07-03 21:58:11 does strongswan have an install script which creates ipsec user and group? that was the problem barthalion 2015-07-03 21:58:42 the APKBUILD sets ipsec as user and group but it doesn't create them.. at least it didn't when i did the upgrade 2015-07-03 21:58:47 that's why it wasn't starting :P 2015-07-03 22:05:12 other than that strongswan is functional again :P 2015-07-03 22:49:50 Mp5shooter - this pre-install script should have run for strongswan http://hastebin.com/becuvujate.bash 2015-07-03 22:52:31 no ipsec user or group for me *shrug* 2015-07-03 22:56:08 does it trigger if it's an upgrade? 2015-07-03 23:06:43 no http://wiki.alpinelinux.org/wiki/APKBUILD_Reference#.24pkgname.pre-install 2015-07-03 23:08:38 just run the 2 commands manually 2015-07-03 23:09:09 addgroup -S ipsec 2015-07-03 23:09:22 adduser -S -H -h /var/empty -s /sbin/nologin -D -G ipsec ipsec 2>/dev/null 2015-07-03 23:41:35 that's what i ended up doing kk 2015-07-05 23:07:02 kaniini is this more what you were talking about regarding a ghc-bootstrap apkbuild? https://github.com/mitchty/alpine-linux-ghc-bootstrap/tree/apkbuild/ghc-bootstrap 2015-07-05 23:07:29 sorry for not a huge amount of updates but i can't spare a ton of time in the summer for things, have running and dancing to take up more of my time 2015-07-05 23:08:41 forewarning, this apk is huge 2015-07-05 23:08:42 195.7M ghc-bootstrap-7.10.1-r0.apk 2015-07-05 23:08:51 1.4g uncompressed 2015-07-05 23:10:17 so assuming that is what you meant by the ghc-bootstrap bit, i am assuming i can take that and use it to build ghc in its apkfile 2015-07-05 23:10:34 and have ghc replace ghc-bootstrap 2015-07-05 23:11:08 but does that mean I can makerequires ghc and ghc-bootstrap and that will work for both the first build and subsequent where I don't want the bootstrap apk? 2015-07-05 23:12:17 i want to keep the property that the ghc apk i build can build itself or at least make sure thats the case, otherwise its not much of a port 2015-07-05 23:23:48 but just to be sure i understood right, what i was going for was: use docker in an apkbuild file to build the ghc cross compiler to /usr 2015-07-05 23:24:22 take that apk and use it to build the normal ghc apk 2015-07-05 23:24:50 or is the idea we leave that as is for a while then release version 1 which drops the ghc-bootstrap dependency 2015-07-05 23:28:13 also might be a moot point as 7.10.2 will be out probably this week or next 2015-07-05 23:45:51 well have to go practice not stepping on toes for a bit, bbl 2015-07-06 11:01:39 hey guys! 2015-07-06 11:05:50 hi leo-unglaub 2015-07-06 11:32:29 fair point from distrowatch about lack of descriptions of isos 2015-07-06 11:33:34 hovering links to get a desc is not something I'd think about 2015-07-06 11:37:45 barthalion: yes, we should have a short desc about the types. 2015-07-06 11:39:37 i dont even know what the regular iso does. 2015-07-06 11:39:43 me neither 2015-07-06 11:39:46 I always use mini 2015-07-06 11:39:54 (maybe it should be used on front page instead) 2015-07-06 11:41:53 i wonder if we should explain we dont ship docs by default because we target more experienced users. 2015-07-06 11:46:36 well, it's pretty clear to me when livecd leaves you in siberian forest with one plastic bag 2015-07-06 11:46:50 that it's not friendly 2015-07-06 12:27:33 barthalion: now i understand your point, i didnt even see the title tags... 2015-07-06 13:17:59 clandmeter, nl.a.o under maintenance? 2015-07-06 13:20:45 ? 2015-07-06 13:21:39 fabled: what do you mean? 2015-07-06 13:22:02 hmm 2015-07-06 13:22:14 builders timed out on http get to it, but now it looks better again 2015-07-06 13:22:24 could've been just me too 2015-07-06 13:23:14 i saw lag in my irssi 2015-07-06 13:23:23 maybe somebody was hammering rsync:: 2015-07-06 13:54:09 can we have gzip as dependecies for lxc? because of this: http://sprunge.us/AjHH 2015-07-06 14:37:05 crow: please submit a bug report 2015-07-06 14:54:58 crow / clandmeter - I sent a patch adding gzip to lxc 2015-07-06 14:55:45 if resuming from hibernation will require modifications to the mkinitfs init script, should I submit a patch for that, or a patch to add hooks (and make everything slightly more complex)? 2015-07-06 14:56:03 also, pm-utils vs. suspend-utils 2015-07-06 14:56:11 (I only use the latter) 2015-07-06 14:57:45 ahills: probably best to submit a patch to our list. best way to discus it. 2015-07-06 14:58:37 ok 2015-07-06 14:58:52 I guess I've already done the work for one approach... 2015-07-06 15:43:42 BitL0G1c ok thanks which nr? 2015-07-06 15:44:39 crow - for lxc 1.1.2 2015-07-06 17:41:48 kaniini ping on the ghc-bootstrap stuff, i'm waiting on ghc to build under the ghc-bootstrap apk I generated, its gonna take a while the bootstrap compiler is mad slow but just making sure I'm going down the right path 2015-07-06 17:41:52 hi 2015-07-06 17:42:02 let me look 2015-07-06 17:42:53 mitchty: yes, i think so, however what all is in the tarball? 2015-07-06 17:43:07 mitchty: i don't think we want to distribute a full image of ubuntu for bootstrap-ghc :p 2015-07-06 17:43:17 kaniini just the cross compiler 2015-07-06 17:43:34 ok 2015-07-06 17:43:36 its unoptimized and static and stupid slow 2015-07-06 17:43:42 yeah that looks ok 2015-07-06 17:44:02 but it only needs to be around once per arch mostly 2015-07-06 17:44:35 what i didn't quite get is the makedepends parts if the idea was do a version zero with ghc-bootstrap as a prereq 2015-07-06 17:44:47 then update that to ghc after or just leave the bootstrap bits in 2015-07-06 17:45:00 mitchty: have it as a knob, see gcc APKBUILD for more info 2015-07-06 17:45:17 mitchty: that way we can just commit changes to the knob as needed 2015-07-06 17:45:31 mitchty: or manually upload the initial package 2015-07-06 17:45:36 kaniini cool will do, at work and this is going to take a while to compile so i'll look tonight 2015-07-06 17:46:06 well we could continue to test the bootstrap stuff 2015-07-06 17:46:37 the dockerfile in the base of that branch is kinda how I'm going to validate what I'm doing still works as far as cross compiling itself 2015-07-06 17:46:56 sure. 2015-07-06 17:47:10 do you want me to commit the ghc-bootstrap so that we can get it building? 2015-07-06 17:47:24 not until i validate this thing compiles itself 2015-07-06 17:47:33 i'm picky that way >.< 2015-07-06 17:48:02 as far as /dev/mqueue, we can get that into the kernel to make your life easier :p 2015-07-06 17:48:15 but yeah the bootstrap bit just requires that a docker daemon be up and running on 4243 2015-07-06 17:48:29 heh turns out i had a 3.1.4 image that didn't have the fix 2015-07-06 17:48:34 i switched to 3.2.0 and it worked 2015-07-06 17:48:43 after twiddling some sysctls 2015-07-06 17:49:14 rather the 3.1.4 image didn't have the config option set 2015-07-06 17:49:37 interesting. we don't have docker running on the builders. 2015-07-06 17:50:02 perhaps what we could do is just have the tarball as a source and just use package() to convert it into an apk? 2015-07-06 17:50:46 sure, thats basically all i'm doing in the apkbuild 2015-07-06 17:51:00 well, no, your APKBUILD invokes docker itself. 2015-07-06 17:51:22 i was thinking just a shell script so that i/you/whoever can run the build process and then package the result 2015-07-06 17:51:38 yes, thats mostly as i thought that is what you wanted the apkbuild to do 2015-07-06 17:52:04 it does work rather nice this way tbh, ouside of then requiring docker to be running on building system 2015-07-06 17:52:04 well, you basically got it 80% right, except i just want it to use a external tarball instead :p 2015-07-06 17:52:37 sure, but i don't want to make our infrastructure team angry :p 2015-07-06 17:52:49 ok well i'm going to update that tonight then after i validate this bulids itself 2015-07-06 17:52:59 it'll end up the same thing just a bit different 2015-07-06 17:53:17 yeah i think you get what i am saying 2015-07-06 17:53:37 just don't want to lose 2 hours of slow compile time :) 2015-07-06 17:53:50 this first build is... painful 2015-07-06 17:54:32 and my "compile" box is an older i3 with spinning rust 2015-07-06 17:54:44 oh. do you want access to something faster? 2015-07-06 17:55:09 i have an E5 sitting mostly idle 2015-07-06 17:55:19 dual E5, 128gb ram 2015-07-06 17:55:40 something like 16x ssd hardware raid 2015-07-06 17:55:50 err wow that would be a bit faster, but i probably won't be able to login until later tonight tbh 2015-07-06 17:55:57 oh, ok 2015-07-06 17:56:15 have to get back to the work grind 2015-07-06 17:56:42 the slow compile times have meant this takes a while to iterate through 2015-07-06 17:56:55 we'll run the final build on my box i guess 2015-07-06 17:56:56 which is the other reason for using docker, caching steps has been a godsend 2015-07-06 17:57:11 i'll build a vmware VM for it 2015-07-06 17:57:19 that feel when I built mariadb with dependencies on shitty mobile dual core with 2GB of RAM :3 2015-07-06 17:57:34 then there is dev.alpinelinux.org which i can upload the binaries to 2015-07-06 17:57:56 which is local to the actual builders :) 2015-07-06 17:58:10 ah cool 2015-07-06 17:58:24 yeah the bootstrap compiler builds in about an hour or so even on this i3 2015-07-06 17:58:30 its mostly i/o blocked 2015-07-06 17:58:40 yeah 2015-07-06 17:58:43 oh question on the make -j bit then 2015-07-06 17:58:45 the SSDs should help 2015-07-06 17:59:00 mitchty: vmware can only expose 8 VCPUs on the license i have 2015-07-06 17:59:25 if I fire off just a make it'll not use any parallelism but i'm not sure the apkbuild should specify a setting 2015-07-06 17:59:43 but without it the build times are nuts 2015-07-06 18:00:45 8 is fine, i've been running this all in vmware vm's with that amount anyway 2015-07-06 18:02:21 the builders override -j anyway 2015-07-06 18:02:31 see /etc/abuild.conf 2015-07-06 18:04:17 ah cool, ok i'll yank that out then 2015-07-06 20:23:10 BitL0G1c> crow / clandmeter - I sent a patch adding gzip to lxc <- i dont see the bug submited, do you habe bug # 2015-07-06 20:23:58 i didn't post a bug - just the patch 2015-07-06 20:25:27 ah ok 2015-07-06 20:26:17 what was the url for patchwork (to see submited and commited patches)? 2015-07-06 21:17:18 crow - http://patchwork.alpinelinux.org 2015-07-06 21:19:38 BitL0G1c thank you, do you maybe know if there is RSS for it? i dont see the link 2015-07-06 21:20:21 http://patchwork.alpinelinux.org/patch/241/ 2015-07-07 06:24:59 morning 2015-07-07 06:25:06 clandmeter: good job with chromium! 2015-07-07 06:43:06 hi ncopa 2015-07-07 06:43:13 enjoyed your long weekend? 2015-07-07 06:43:56 re %239 : can I commit? iirc we talked about it, and it's actually a good idea. 2015-07-07 06:53:22 fcolista: i dont think anyone would disagree with that patch. 2015-07-07 06:53:41 +1 2015-07-07 06:53:44 and its not a showstopper ;-) 2015-07-07 07:02:12 clandmeter: yes, i had a *great* weekend 2015-07-07 07:03:43 fcolista: i am ok with %239, just have someone english expert ACK that language is ok. 2015-07-07 07:04:08 ok. Just wondering where to commit :) 2015-07-07 07:04:47 morning. 2015-07-07 07:04:59 ncopa: would you mind rebuildinn encfs? 2015-07-07 07:05:36 and rlog :) 2015-07-07 07:05:43 they are broken in testing :( 2015-07-07 07:06:30 thats because we dont bump them when needed. so if it works, move it asap to main. 2015-07-07 07:07:24 fcolista: http://git.alpinelinux.org/cgit/alpine-baselayout/tree/motd 2015-07-07 07:07:35 ah, alpine-baselayout 2015-07-07 07:07:48 thx clandmeter 2015-07-07 07:08:26 fcolista: apk info -W would also work :) 2015-07-07 07:09:00 that's the package...i was not aware that was a distinct git repo 2015-07-07 07:09:37 ScrumpyJack: are you sure rebuilding encfs will fix it? 2015-07-07 07:09:59 ncopa: regarding chromium. im having 2 issues. 2015-07-07 07:10:06 bot rlog and encfs are rebuilt after gcc5 update 2015-07-07 07:10:15 i tried to add dbg pkg, but abuild fails to create it 2015-07-07 07:11:31 maybe i should update them locally 2015-07-07 07:12:33 ScrumpyJack: I suspect there is other problem 2015-07-07 07:12:38 what is the exact error msg? 2015-07-07 07:14:13 sprunge.com/PjbJ 2015-07-07 07:16:03 is an english-speaking guy wants to chekc this, i can commit: http://sprunge.us/LaGD 2015-07-07 07:16:34 i'm not able to type this morning 2015-07-07 07:19:30 looks ok to me 2015-07-07 07:22:01 doesnt sound completely correct imho, but im not native 2015-07-07 07:22:14 the system sounds a bit weird 2015-07-07 07:22:29 this system maybe 2015-07-07 07:27:10 ScrumpyJack: what does 'apk version rlog encfs' say? 2015-07-07 07:27:44 clandmeter: i also had the feeling that "You can setup the system with the command: setup-alpine" sounds a bit weird 2015-07-07 07:27:54 but i'm no english native 2015-07-07 07:28:38 To setup this system please run: setup-alpine 2015-07-07 07:29:19 Run 'setup-alpine' for install. 2015-07-07 07:29:21 i dunno 2015-07-07 07:29:53 encfs is r1 and rlog is r2 2015-07-07 07:30:33 hm 2015-07-07 07:30:39 its correct 2015-07-07 07:31:47 apk version boost 2015-07-07 07:32:46 i don't have it installed 2015-07-07 07:35:06 i find it weird 2015-07-07 07:35:07 I'm mixing 3.2 and edge here, perhaps i should bite the bullet and update to edge? 2015-07-07 07:35:14 oh 2015-07-07 07:35:21 its a mix of v3.2 and edge? 2015-07-07 07:35:22 yes 2015-07-07 07:35:25 then it is a poblem 2015-07-07 07:35:28 however 2015-07-07 07:35:36 it might be a better idea to backport it to v3.2 2015-07-07 07:35:53 ScrumpyJack: could you please set up a test box with encfs 2015-07-07 07:35:55 with edge 2015-07-07 07:36:00 verify it works as it should 2015-07-07 07:36:05 then we move it to main 2015-07-07 07:36:11 and finally we can backport it to v3.2 2015-07-07 07:36:33 v3.2 and edge does not play along too well 2015-07-07 07:36:38 yes, i already use it on edge, and have been using it successfully since i built the package :) 2015-07-07 07:36:40 due to gcc5 in edge 2015-07-07 07:38:34 ScrumpyJack: you should be able to build both on v3.2 to make it work 2015-07-07 07:38:39 there are several packages that I (and others) have been using successfully for a while now, and could be moved to main. (but to track those, there is no easy process yet) 2015-07-07 07:39:05 i don't have a 3.2 build box at the moment, only edge 2015-07-07 07:42:46 ok 2015-07-07 07:42:51 building one now 2015-07-07 07:43:18 libtool --finish "${pkgdir}"/usr/lib 2015-07-07 07:43:22 why is that needed? 2015-07-07 07:43:28 options="libtool" 2015-07-07 07:45:38 ok 2015-07-07 07:45:57 rlog now looks ok for moving to main 2015-07-07 07:48:08 encfs sounds like something you want for like dropbox 2015-07-07 07:49:47 does encfs need a init.d script? 2015-07-07 08:37:56 fabled: re #4429, we'll lose hw crypto for curl if we switch to mbedtls for curl? 2015-07-07 08:40:44 ncopa, well, at least it changes the hardware support level 2015-07-07 08:40:50 seems polarssl has some hw support 2015-07-07 08:40:53 e.g. it does padlock/aes 2015-07-07 08:40:59 but not padlock/sha1 2015-07-07 08:41:07 ok 2015-07-07 08:41:20 do we care? 2015-07-07 08:41:27 i think maybe not 2015-07-07 08:41:30 mmm 2015-07-07 08:41:35 for curl, not necessarily 2015-07-07 08:41:41 though 2015-07-07 08:41:50 it'll add a dependency 2015-07-07 08:41:53 yes 2015-07-07 08:41:55 we pull in openssl always now due to apk 2015-07-07 08:41:58 yes 2015-07-07 08:42:02 thats what i'm thinking too 2015-07-07 08:42:20 then we shoudl try switch to mbedtls for as much as possible 2015-07-07 08:42:28 to reduce number of deps 2015-07-07 08:42:41 yes. but i think it's not too widely supported -- unless it has openssl compat api 2015-07-07 08:42:49 i don thinkt it has 2015-07-07 08:43:01 i think the entire point with it is a better api :) 2015-07-07 08:44:37 :) 2015-07-07 08:44:54 yeah, it's api is different 2015-07-07 08:44:54 otoh, curl is not part of the base install 2015-07-07 08:46:15 what i mean, a compat api is kind of defeating the purpose 2015-07-07 08:46:38 ncopa: sorry about this, I've lost track. Are you putting rlog/encfs into main or am i supposed to be doing something? (sorry, i keep getting disturbed) 2015-07-07 08:46:52 ScrumpyJack: i can move them 2015-07-07 08:47:10 but i have a few questions about encfs 2015-07-07 08:47:38 sure 2015-07-07 08:47:42 why is options="libtool" needed? 2015-07-07 08:47:59 why is this needed: libtool --finish "${pkgdir}"/usr/lib 2015-07-07 08:48:09 arg, i can't remember. 2015-07-07 08:48:24 it seems that in the end it just purges the .la files anyways 2015-07-07 08:50:43 testing build without the libtool --finish 2015-07-07 08:55:41 that's fine, can you remove the line? 2015-07-07 09:16:55 there are no other package that uses libencfs 2015-07-07 09:17:08 there are no encfs-dev package 2015-07-07 09:17:20 so i dont think there are any point in splitting the encfs-libs 2015-07-07 09:24:10 ScrumpyJack: can you please file a ticket with : backport encfs to v3.2 or similar 2015-07-07 09:24:12 with target v3.2 2015-07-07 09:24:52 you can mention that edge introduces gcc5 so you can no longer mix v3.2 and edge 2015-07-07 09:24:56 so you want it backported 2015-07-07 09:25:44 its only for history so if i in future wonder: why did i backport this? who asked for it? then i git commit points to the issue and give me the needed answers 2015-07-07 09:27:02 hm, are we going to have backports repository for 3.2 like for 2.7? 2015-07-07 09:30:24 no 2015-07-07 09:30:27 i prefer not 2015-07-07 09:31:05 btw 2015-07-07 09:31:15 in the weekend i hacked a bit on nlplug 2015-07-07 09:31:30 a netlink listener for uevents 2015-07-07 09:32:00 it does not seem like busybox will ever get the read events from fd support in mdev 2015-07-07 09:32:21 so what i was thinking was 2015-07-07 09:32:35 have the nlplug filter the events 2015-07-07 09:32:50 i think the only events that makes sense to pass to mdev is those with DEVNAME set 2015-07-07 09:33:03 which will make them match a rule in mdev.conf 2015-07-07 09:33:23 then we handle modaliases directly in nlplug 2015-07-07 09:33:46 on my lappy there was like 200-300 events 2015-07-07 09:34:01 30-40 was modaliases 2015-07-07 09:34:59 i think we can simply fork/exec mdev for those events that creates /dev/* stuff 2015-07-07 09:35:06 to have mdev fix the permissions etc 2015-07-07 09:35:16 shoul speed up boot 2015-07-07 09:35:22 and reduce number of forks 2015-07-07 10:15:09 hey :) 2015-07-07 10:16:27 ncopa: i cannot reply in the bug tracker. so i do it here 2015-07-07 10:16:42 the openssl-dev package is only needed if you want to build in support for both 2015-07-07 10:16:48 if you only want one, you dont need it at all 2015-07-07 10:24:48 leo-unglaub: whats wrong with bug tracker? 2015-07-07 10:25:37 clandmeter: i have no reply button 2015-07-07 10:28:49 leo-unglaub: you should even be able to reply via email 2015-07-07 10:29:03 there should also be an update button 2015-07-07 10:29:08 https://i.imgur.com/r9uDb5W.png 2015-07-07 10:29:14 i dont see a reply button 2015-07-07 10:29:19 maybe i miss something 2015-07-07 10:29:55 ah, nevermind 2015-07-07 10:30:00 the german translation is bullshit 2015-07-07 10:30:28 bearbeiten == edit ... and that normaly means editing the own post and not creating a reply *g* 2015-07-07 10:31:53 hehe 2015-07-07 10:32:17 german is the issue ;-) 2015-07-07 10:32:29 tzzzzzz 2015-07-07 10:32:37 sure, blame german *g* 2015-07-07 10:32:46 greek does it to *g* 2015-07-07 10:33:36 im dutch, so i have to blame german. 2015-07-07 10:33:40 :p 2015-07-07 10:34:07 *g* 2015-07-07 10:41:48 kaniini: i am still guessing what the openssl bug is 2015-07-07 10:42:39 specially interresting to me is, that 0.9.8 and 1.0.0 are not affected. this means it has to be in the "new" code they added 2015-07-07 11:26:17 AL page on wikipedia updated: https://en.wikipedia.org/wiki/Alpine_Linux 2015-07-07 12:00:21 who is Łukasz Jendrysik in the channel here? 2015-07-07 12:08:54 scadu 2015-07-07 12:09:08 scadu: ping 2015-07-07 12:59:53 ncopa: i'll raise a ticket 2015-07-07 13:09:53 it's issue #4430, but it's under Alpine Package Keeper accidently, as i created with console elinks 2015-07-07 13:13:18 can anyone confirm for me that this bug: https://bugs.exim.org/show_bug.cgi?id=1638 2015-07-07 13:13:33 is fixed with this commit: http://vcs.pcre.org/pcre?view=revision&revision=1562 2015-07-07 13:14:02 there are nothing in the commit message that indicates that it fixes that bug 2015-07-07 13:14:19 there is nothing in the bugzilla that indicates exactly which commit fixes the issue 2015-07-07 13:14:48 changelog testcase looks similar to the one reported in the bugzilla http://vcs.pcre.org/pcre/code/trunk/ChangeLog?r1=1562&r2=1561&pathrev=1562 2015-07-07 13:14:54 and the dates match 2015-07-07 13:15:04 reported date, fixed date 2015-07-07 14:14:11 ncopa, what do you think of %242 ? 2015-07-07 14:14:16 is it possible ? 2015-07-07 14:19:56 sure 2015-07-07 14:20:05 we should probably cherry-pick i for v3.2 too 2015-07-07 14:29:36 alacerda: thanks! 2015-07-07 15:15:15 there is no tool similar to main/jq for yaml 2015-07-07 15:16:01 ncopa: any other issues with newsbeuter, kinda wanting to use it ! :) 2015-07-07 15:16:27 thinking of writing web-frontend for it 2015-07-07 15:25:59 already done some backend codes, need to move from liferea 2015-07-07 16:04:49 ncopa, ref 4422, fr-ca-june is left yet 2015-07-07 16:04:53 you are the maintainer 2015-07-07 16:28:34 vkrishn: yeah, I guess you need to use lua 2015-07-07 17:06:49 barthalion: sup? 2015-07-07 18:22:48 scadu: it's not me who pinged you, read more carefully 2015-07-07 18:25:48 he's offline and I was hoping that you know something ;f 2015-07-07 18:49:28 scadu: hey 2015-07-07 18:49:31 i pinged you 2015-07-07 18:50:16 and this is crazy 2015-07-07 18:50:38 (cmooon, I know you can finish this) 2015-07-07 18:51:04 I guess he's gone anyway, so maybe leave him a message on query 2015-07-07 18:51:22 barthalion: we have time *g* 2015-07-07 18:57:09 leo-unglaub: hi. 2015-07-07 18:57:25 hey scadu 2015-07-07 18:57:41 i have two comments to the mbedtls package that you maintain 2015-07-07 18:57:52 yes? 2015-07-07 18:59:23 you do some sed commands on the config file. this is not needed, because the archive already contains a script that does that for you in a simply way 2015-07-07 18:59:53 configScript="scripts/config.pl -f include/polarssl/config.h" 2015-07-07 19:00:06 $configScript unset/set 2015-07-07 19:00:27 it makes the script a little bit simpler and it handles new settings easiert 2015-07-07 19:00:33 the other and more important thing 2015-07-07 19:00:41 i would love it if you could set the following 2015-07-07 19:01:12 unset POLARSSL_DEBUG_C, unset POLARSSL_CIPHER_NULL_CIPHER, unset POLARSSL_ENABLE_WEAK_CIPHERSUITES, set POLARSSL_THREADING_C, set POLARSSL_THREADING_PTHREAD, set POLARSSL_REMOVE_ARC4_CIPHERSUITES 2015-07-07 19:01:53 well 2015-07-07 19:01:58 patches welcome I guess? 2015-07-07 19:02:16 I mean, I don't see why we wouldn't merge it even without scadu's approval 2015-07-07 19:02:36 barthalion: i always prefer talking to the maintainer first 2015-07-07 19:02:51 becasue i dont want to interrupt another persons work 2015-07-07 19:02:59 and to avoid bad blood i always ask first 2015-07-07 19:03:38 mbedtls doesn't have threading enabled by default? 2015-07-07 19:04:01 i think threading is enabled by scadu 2015-07-07 19:04:08 i was just pasting my entire config 2015-07-07 19:04:36 maybe I'll take a look tomorrow. and yes, patches are welcome. 2015-07-07 19:05:54 okay, then i put a patch in the bug tracker 2015-07-07 19:06:06 alpine-aports please 2015-07-07 19:06:22 so I (or someone else) could merge it tomorrow 2015-07-07 19:06:53 git email sending did not work for me at all 2015-07-07 19:07:04 thats why i uploaded other patches to the bug tracker 2015-07-07 19:07:54 I hope ncopa is tracking it 2015-07-07 19:07:58 or clandmeter 2015-07-07 19:08:16 ncopa: did in the past :) 2015-07-07 19:08:55 good 2015-07-07 19:10:49 but let me take another look at the git smtp tls code 2015-07-07 19:10:57 something there must be fucked up 2015-07-08 06:21:22 trying to build a package that requires termio.h, which seems to be in dev86, is that correct or should i find it somewhere else? 2015-07-08 06:55:23 how do I check what's missing for having *-dev package built ? 2015-07-08 07:14:03 ScrumpyJack: try to remove that include and see what you are missing. 2015-07-08 07:16:52 fcolista: ? 2015-07-08 08:00:52 fcolista: *-dev finds /usr/include /usr/lib/*.so which are symlinks 2015-07-08 08:01:02 only way to know exactly is to read abuild source 2015-07-08 08:01:05 default_dev 2015-07-08 08:21:03 umh...ok. I've compton package who does not create -dev package. I was wondering how to debug that. 2015-07-08 08:21:32 but probably it does not create .so or include files. 2015-07-08 08:24:04 Ok. I think i've figured. In Makefile, INCS = does not contains anything. 2015-07-08 08:24:07 OBJS = compton.o 2015-07-08 08:24:30 there are no .so nor include 2015-07-08 08:24:43 that's why -dev does not build. Am I right? 2015-07-08 08:27:52 fcolista: have you considered that there is simply no need for -dev package? 2015-07-08 08:28:19 barthalion, yes. I'd like to know how to know it in advance. 2015-07-08 08:28:32 I just build without subpackages first 2015-07-08 08:29:05 ..and then let abuild tell you which subpkg can be built? 2015-07-08 08:29:17 and then I let my tar list files inside 2015-07-08 08:29:32 I pretty much know what needs to be split, I don't trust abuild 2015-07-08 08:29:52 if there is /usr/include, I obviously need to add -dev 2015-07-08 08:32:19 thx for the suggestion barthalion 2015-07-08 11:29:25 seems there is problem with sensors-detect on 3.2.0 http://sprunge.us/FdKC hier i dont have -ash: cd: can't cd to /etc/modules-load.d/ 2015-07-08 11:30:24 crow: can you mkdir -p /etc/modules-load.d and see if it still errors? 2015-07-08 11:34:15 clandmeter it works after http://sprunge.us/IegG 2015-07-08 11:35:47 ok 2015-07-08 11:37:19 crow: whats inside the conf file? 2015-07-08 11:38:38 oh its in the sprunge 2015-07-08 11:38:39 sorry 2015-07-08 11:38:58 im not sure thats the correct location 2015-07-08 11:41:40 crow: it is the correct location 2015-07-08 11:44:31 we added support for that not too long ago 2015-07-08 11:45:02 ncopa: http://git.alpinelinux.org/cgit/aports/commit/?id=0b34e934c5ee509c197b159a51b14e9424f24470 2015-07-08 11:45:15 i guess openrc should create it? 2015-07-08 11:45:54 probably alpine-baselayout 2015-07-08 11:46:31 or the sensors-detect script could add it too 2015-07-08 11:47:31 well, having it at boot would make more sence, so ppl know we have support for it? 2015-07-08 11:48:03 in boot i mean in baselayout 2015-07-08 11:51:03 you are probably right 2015-07-08 11:52:15 pushed to edge 2015-07-08 12:11:49 clandmeter i belive so, there was before issue with loading modul and ncopa created that. sensors do output some value so its ok 2015-07-08 12:26:01 does stuffin.space work on chrome in AL? 2015-07-08 12:26:22 I can't get chromium to run at all 2015-07-08 12:26:27 (webgl - can't test right now, but just wondering) 2015-07-08 12:29:11 ahills: which error? 2015-07-08 12:40:43 I will have to look when I get home, sorry 2015-07-08 12:40:53 It gave about ten lines of errors 2015-07-08 12:50:34 ncopa: what is your firefox release plan? do you want to stay on esr or do you want a patch for 39.0? 2015-07-08 12:51:14 i want keep esr for main, but i dont mind doing 39.0 in testing/ 2015-07-08 12:51:40 okay, do you want a patch or do you want to bump the number yourself? 2015-07-08 12:52:01 i dont mind a patch ;) 2015-07-08 12:52:16 ncopa: okay, i create a ticket in the bug tracker and assign it to you 2015-07-08 12:52:43 one more thing, i have a package for the firefox-aurora build 2015-07-08 12:52:52 40.0b2 2015-07-08 12:53:00 would it make sence to put that in testing? 2015-07-08 12:53:11 so main is esr and testing is aurora? 2015-07-08 12:54:10 that sounds good 2015-07-08 12:54:35 great! 2015-07-08 12:55:32 firefox actually has another browser based on aurora, the developer edition 2015-07-08 12:55:33 https://www.mozilla.org/de/firefox/developer/ 2015-07-08 12:55:52 but i refuse to use that shit, because every application that messes up my custom theme can to fuck itself 2015-07-08 12:56:32 just disable it 2015-07-08 12:56:45 I don't really see any value in aurora in testing 2015-07-08 12:57:12 barthalion: if you disable it, you have a normal aurora 2015-07-08 12:57:14 :) 2015-07-08 12:57:46 ah 2015-07-08 12:57:57 I don't get the point of calling it "developer edition" then :P 2015-07-08 12:59:12 barthalion: marketing :) 2015-07-08 12:59:34 the download archive actually is still called *-aurora* 2015-07-08 12:59:34 *g* 2015-07-08 13:01:00 ncopa: before i have to go to a client, something is wrong with the hiawatha package. its build against an old version of mbedtls 2015-07-08 13:01:11 maybe the build server was not current? 2015-07-08 13:01:22 please could you bump that package to rebuild? 2015-07-08 13:02:46 mbedtls ships /usr/lib/libmbedtls.so.9 2015-07-08 13:02:52 but hiawatha requires /usr/lib/libmbedtls.so.8 2015-07-08 13:02:56 no idea why 2015-07-08 13:06:03 there was soname bump 2015-07-08 13:06:08 someone missed it 2015-07-08 13:06:34 leo-unglaub: edge? 2015-07-08 13:12:34 gosh, I not used to such colourful language in here :) 2015-07-08 13:15:21 hum? 2015-07-08 17:26:30 hi 2015-07-08 17:27:13 what does this mean? 2015-07-08 17:27:17 ERROR: unsatisfiable constraints: spl-grsec-3.18.17-r0: masked in: @testing satisfies: zfs-grsec-3.18.17-r0[spl-grsec] zfs-grsec-3.18.17-r0: masked in: @testing satisfies: world[zfs-grsec] 2015-07-08 17:27:22 want to build a zfs iso 2015-07-08 17:27:29 added zfs to modloop extra 2015-07-08 17:35:27 shafire: zfs is old and needs to be bumped 2015-07-08 17:35:31 ill check it later 2015-07-08 17:35:54 and you have testing pinned so you need to install it pinned 2015-07-08 17:36:13 MODLOOP_EXTRA = $(addsuffix -grsec, dahdi-linux xtables-addons zfs) 2015-07-08 17:36:16 MODLOOP_EXTRA = $(addsuffix -grsec, dahdi-linux xtables-addons zfs@testing) ? 2015-07-08 19:07:58 clandmeter: can you move the zfs-grsec-utils to zfs-utils? 2015-07-08 19:09:29 and update to 0.6.4.2 2015-07-08 19:11:00 clandmeter: can you split that in two packages? like here: http://pkgs.alpinelinux.org/packages?package=%25flashcache%25&repo=all&arch=x86 2015-07-09 06:53:56 hi 2015-07-09 07:07:56 shafire: i tried to update it yesterday but it fails to build 2015-07-09 07:08:05 could be im doing something wrong 2015-07-09 07:08:18 maybe as ncopa later when he fixed his build issues 2015-07-09 07:24:59 ok 2015-07-09 07:25:13 first watching Mr. Robot now :) 2015-07-09 07:25:16 do you know? 2015-07-09 07:28:03 ? 2015-07-09 07:32:09 http://www.imdb.com/title/tt4158110/ 2015-07-09 08:18:42 ah no i dont 2015-07-09 08:29:19 how to preserv the changed stuff in hier after reboot? /proc/sys/kernel/core_pattern and /proc/sys/kernel/core_uses_pid - on every reboot i need to reenter them 2015-07-09 08:34:14 put it in /etc/sysctl.conf or /etc/sysctl.d/my-tuning.conf 2015-07-09 08:53:24 ok ill try it 2015-07-09 10:28:39 hi 2015-07-09 10:28:53 clandmeter: there is a problem with ifupdown in .post-upgrade 2015-07-09 10:28:53 Thu:28 12:26:09 root:alpha~# apk fix (1/1) Reinstalling ifupdown (0.7.53.1-r0) Executing ifupdown-0.7.53.1-r0.post-upgrade busybox: 'busybox' is not an absolute path ERROR: ifupdown-0.7.53.1-r0.post-upgrade: script exited with error 1 Executing busybox-1.23.2-r0.trigger 1 errors; 141 MiB in 109 packages 2015-07-09 10:29:17 http://git.alpinelinux.org/cgit/aports/diff/main/ifupdown/ifupdown.post-upgrade?id=289d5519971f26fccff37293e0d1ff100f1df321 2015-07-09 10:31:12 mh 2015-07-09 10:31:18 or is it my mistake? 2015-07-09 10:31:19 https://bugs.alpinelinux.org/issues/4135 2015-07-09 10:31:25 why is that needed even? 2015-07-09 10:31:27 and I do not know how I solved it 2015-07-09 10:31:33 busybox trigger should call that automatic 2015-07-09 10:32:15 i think the post script should be just removed 2015-07-09 10:32:24 triggers="busybox.trigger=/bin:/usr/bin:/sbin:/usr/sbin:/lib/modules/*" 2015-07-09 10:32:47 and bb trigger will do "--install -s" when ever one of those paths change (except the modules one) 2015-07-09 10:32:47 .post-deinstall too? 2015-07-09 10:32:55 yes 2015-07-09 10:33:46 i think the scripts are old 2015-07-09 10:33:55 from time before triggers 2015-07-09 10:34:21 ah okay 2015-07-09 10:34:30 hey friends :) 2015-07-09 10:34:46 ncopa, boreutils, gawk, ifupdown, sed, wget and shadow all have same issue 2015-07-09 10:34:58 post-deinstall/post-upgrade calls busybox --install -s 2015-07-09 10:35:03 which is redundant with the bb trigger 2015-07-09 10:35:15 ncopa: sorry i did not send you the patch last night. but it was the first day after this fucking heat wave and i was so down 2015-07-09 10:36:29 fabled: should I open a ticket? 2015-07-09 10:36:44 shafire, that's a good idea yes 2015-07-09 11:04:22 hi leo-unglaub, no worries 2015-07-09 11:04:33 i have already forgotten what it was about :) 2015-07-09 11:20:55 ncopa: aurora 2015-07-09 11:21:12 oh that :) 2015-07-09 11:42:51 leo-unglaub: I think you haven't replied me about where you got that mbedtls error 2015-07-09 11:42:55 edge vs v3.2 2015-07-09 11:43:07 barthalion: ? 2015-07-09 11:43:52 15:02:37 mbedtls ships /usr/lib/libmbedtls.so.9 2015-07-09 11:43:54 15:02:43 but hiawatha requires /usr/lib/libmbedtls.so.8 2015-07-09 11:45:59 ah, that one 2015-07-09 11:46:23 i have not seen your reply/question there 2015-07-09 11:46:28 maybe i had a disconnect 2015-07-09 11:46:33 it was yesterday 2015-07-09 11:46:36 or two days ago 2015-07-09 11:47:32 i have nothing in my logs, could you please be so kind and paste it again? 2015-07-09 11:47:49 13:42:47 edge vs v3.2 2015-07-09 11:47:51 :P 2015-07-09 11:48:06 ah *g* i am on edge *g* 2015-07-09 11:48:16 alright 2015-07-09 11:48:37 lbu should have the possible to just revert one file 2015-07-09 11:48:48 possibility* 2015-07-09 11:53:55 leo-unglaub: should be ok now 2015-07-09 11:54:13 barthalion: awesome! thank you very much! 2015-07-09 11:54:52 not at all 2015-07-09 11:55:24 ncopa: ping 2015-07-09 11:55:46 and fabled probably too… 2015-07-09 11:56:06 ncopa, fabled: you had these fancy packages dependencies graph one day 2015-07-09 11:56:19 how can I generate it? 2015-07-09 11:56:31 with apk dot 2015-07-09 11:56:42 just a sec 2015-07-09 11:56:58 ah, it's built into apk 2015-07-09 11:57:42 barthalion, apk dot --errors | dot -Tpng -o errors.png 2015-07-09 11:58:46 thx 2015-07-09 12:06:51 fabled: does apk use wget still? 2015-07-09 12:07:00 or it only uses libfetch now? 2015-07-09 12:07:02 ncopa, no; libfetch onmly 2015-07-09 12:07:06 # apk uses wget so we must restore the busybox wget symlink early 2015-07-09 12:07:06 exec /bin/busybox --install -s 2015-07-09 12:07:22 so we can remove wget.post-dekinstall 2015-07-09 12:07:24 yes 2015-07-09 12:08:44 ... so I wonder how much did Microsoft gave OpenBSD foundation ... 2015-07-09 12:09:00 free money, nothing to worry about 2015-07-09 12:09:28 what is free money? 2015-07-09 12:09:40 the money from Microsoft for nothing 2015-07-09 12:10:53 perhaps they are hoping the money will help improve ssl 2015-07-09 12:14:38 ACTION could use some free money 2015-07-09 12:14:46 where's Microsoft's gift to Alpine LInux? 2015-07-09 12:26:39 ScrumpyJack: openssh 2015-07-09 13:22:35 is there any reason for abuild do not create pkgdir? http://pastebin.com/FQUXgZUz 2015-07-09 13:23:07 hi 2015-07-09 13:26:40 alacerda, public the apkbuild 2015-07-09 13:27:51 http://pastebin.com/dzVCuvDu 2015-07-09 13:28:24 alacerda, in APKBUILD try in function package(){ install -d $pkgdir/zabbix} 2015-07-09 13:29:17 vitronic, ¬¬" Worked! 2015-07-09 13:29:20 =] 2015-07-09 13:29:26 easy that way!!! 2015-07-09 13:30:27 alacerda, sometimes you have to manually create the directories 2015-07-09 13:30:40 vitronic, is there any explanation - so i can understand the why 2015-07-09 13:31:02 yes 2015-07-09 13:31:23 I try to explain 2015-07-09 13:31:26 hum... after this install do I need to "make install" again? 2015-07-09 13:32:53 ok. new openssl came, i'm working on it 2015-07-09 13:32:56 Sometimes the native font installer assumes that the directory already exists and does not bother to create it, so abuild fails when trying to move the directory, in these cases we must create the directories manually 2015-07-09 13:33:17 sadly, they say "it's security fix only" but it contains a bunch of other stuff; and the our parallel build fix patch no longer applies 2015-07-09 13:33:23 Sorry my bad English, but I use a translator :-( 2015-07-09 13:34:10 vitronic, i could grasp the idea 2015-07-09 13:34:11 ;) 2015-07-09 13:34:31 fabled: alpine is affected then? 2015-07-09 13:34:39 alacerda, great 2015-07-09 13:34:52 erw, https://www.openssl.org/news/secadv_20150709.txt 2015-07-09 13:35:18 fabled: yes, I've read it. I'm tracking vulnerable distributions. Not sure what Alpine carries 2015-07-09 13:35:21 fabled: FYI, I'm at Docker, we carry an image for alpine. I'm tracking the updates here: https://github.com/docker-library/official-images/issues/878 2015-07-09 13:35:30 we have 1.0.2c 2015-07-09 13:35:33 so yeah, update needed 2015-07-09 13:35:37 thanks 2015-07-09 13:35:45 it'll be out in a few minutes 2015-07-09 13:36:34 thanks. 2015-07-09 13:38:18 fabled: are you working on openssl issue? 2015-07-09 13:38:22 yes 2015-07-09 13:38:32 whatta 2015-07-09 13:38:36 let me know if you want me to cherry-pick anything to v3.2 2015-07-09 13:38:36 they also added a bunch of api calls 2015-07-09 13:38:43 ! 2015-07-09 13:42:35 fabled: i cerrypick that to v3.2? 2015-07-09 13:43:55 yeah, i think that it is good 2015-07-09 13:44:04 passed my mini-checks 2015-07-09 13:45:08 push to 3.1/3.0/2.7 too ? 2015-07-09 13:45:27 needs rebase 2015-07-09 13:45:31 hmmh 2015-07-09 13:45:33 i think i can handle that 2015-07-09 13:45:42 1 out of 4 hunks FAILED -- saving rejects to file Makefile.org.rej 2015-07-09 13:46:12 ah yes 2015-07-09 13:47:46 ncopa: you forgot to remove the post scripts in ifupdown (http://git.alpinelinux.org/cgit/aports/commit/?id=401bc15a8f7c492321211021cb9c0ebb50facdcd) 2015-07-09 13:48:09 i need to run in few minutes. thanks for handling the backports, ncopa 2015-07-09 13:48:17 thorduri: i hear you 2015-07-09 13:50:00 fabled: couldn't you just cherry-pick aae41f8c54257d9fa6904d3a9aa09c5db6cefd0d and 2aacec8f4a5ba1b365620a7b17fcce311ada93ad ? 2015-07-09 13:50:32 unless there is some issue with patching (vs upgrading the release) 2015-07-09 13:51:15 there's also some other commits that look like CVE worthy 2015-07-09 13:51:26 i think it was better to upgrade it 2015-07-09 13:51:36 the patch conflict was simple to resolve it 2015-07-09 13:51:42 fabled: interesting. If that were true though, they should have issued CVEs... 2015-07-09 13:52:32 like https://github.com/openssl/openssl/commit/4ce5ed5f3970a1973c4cba41901294632a3c617d 2015-07-09 13:55:08 erw, 4ce5ed sounds like remotely crash (DoS if OCSP gives malicious reply) 2015-07-09 13:55:24 oh 2015-07-09 13:55:26 fabled: they probably didn't CVE that because it's in the apps directory. It's not part of the library, but is part of the openssl client (or an example program) 2015-07-09 13:55:28 it's the apps 2015-07-09 13:55:39 yeah 2015-07-09 13:55:52 i wonder if https://github.com/openssl/openssl/commit/3c66a669dfc7b3792f7af0758ea26fe8502ce70c has security implications 2015-07-09 14:03:34 fabled: that does look bad for anyone using PSKs 2015-07-09 14:04:57 ok all stable branches has patched openssl 2015-07-09 14:05:52 nice 2015-07-09 14:05:53 thanks 2015-07-09 14:06:00 oh yes they were on 1.0.1 still 2015-07-09 14:08:23 fabled: so it was only the dev branch affected? 2015-07-09 14:12:00 also stables 2015-07-09 14:12:11 i think they were all affected 2015-07-09 14:12:20 yeah - all branches. we were up-to-date 2015-07-09 14:12:26 ok. need to run 2015-07-09 14:48:30 please, new aport: http://sprunge.us/AeHd 2015-07-09 14:55:00 w00t 2015-07-09 14:55:06 chromium works 2015-07-09 14:55:35 thanks to ncopa dalias and of course algitbot 2015-07-09 14:56:03 alacerda, we have zabbix already in edge? what's the difference? 2015-07-09 14:56:54 oops The diff is that I am making mistakes 2015-07-09 14:56:56 :s 2015-07-09 14:57:18 ttyl 2015-07-09 14:58:10 alacerda: lol 2015-07-09 14:58:17 kkk 2015-07-09 14:58:24 sorry for the noise! 2015-07-09 14:58:41 ACTION slaps alacerda with a i did-not-check-aports-stick 2015-07-09 14:59:52 alacerda: im glad i didnt make the same mistake with chromium.... or did it? 2015-07-09 14:59:55 :) 2015-07-09 14:59:58 ACTION is crying and saying: I did... I did! 2015-07-09 15:00:08 hehehhehe 2015-07-09 16:46:12 thanks for newsbeuter 2015-07-09 18:10:08 how 2015-07-09 18:10:09 hi 2015-07-09 18:10:42 abuild 2015-07-09 18:11:05 how to trigger a package build from build*? 2015-07-09 18:14:31 is builder broken? 2015-07-09 18:14:38 I do not see build messages from the last days 2015-07-09 18:15:42 build64-2-5: retry 2015-07-09 19:57:32 shafire: if you happen to watch the irc logs, build messages are now on #alpine-commits 2015-07-09 20:02:40 is the build service broken? 2015-07-09 20:06:33 shafire: if you happen to watch the irc logs, build messages are now on #alpine-commits 2015-07-09 20:06:45 ^^ is what I typed while you were gone :) 2015-07-09 20:20:41 ah, I see in logs 2015-07-09 20:20:42 thanks 2015-07-10 06:39:52 bah 2015-07-10 06:39:57 i think polkit is broken 2015-07-10 06:43:38 argh 2015-07-10 06:46:06 no, it works 2015-07-10 06:46:15 polkit just needed a restart 2015-07-10 06:46:39 dbus is not my favorite technology... 2015-07-10 15:31:23 hello, anyone here knows the equivalent operator [ =~ ] in ash 2015-07-10 16:28:39 none 2015-07-10 16:28:48 vitronic: what do you want to achieve? 2015-07-10 17:19:38 barthalion, comparing a regular expression 2015-07-10 17:20:27 bash is used =~ but this does not work in ash 2015-07-10 17:21:42 although I solved with [ echo $1 | grep -E '^[0-9]+$' ] 2015-07-10 17:23:14 so it appears that gcc isn't fully fpic with its object files 2015-07-10 17:23:18 https://gist.github.com/mitchty/a697dacadab59222634f 2015-07-10 20:22:06 vitronic: well, I know what =~ does, I was asking about exact case 2015-07-10 20:22:12 but good that you "fixed" it 2015-07-10 20:28:37 vitronic: drop those [ ] and add a -q to grep 2015-07-10 20:29:27 vitronic: your solution will fail when grep gives no output 2015-07-10 20:30:09 (could be solved with something hideous like: [ "$(echo $1 | grep ...)" ] 2015-07-10 20:30:50 uggedal, pitronic:~/pruebas# echo 123 | grep -E '^[0-9]+$' 2015-07-10 20:30:50 123 2015-07-10 20:30:50 pitronic:~/pruebas# echo $? 2015-07-10 20:30:50 0 2015-07-10 20:30:50 pitronic:~/pruebas# echo '23s' | grep -E '^[0-9]+$' 2015-07-10 20:30:50 pitronic:~/pruebas# echo $? 2015-07-10 20:30:52 1 2015-07-10 20:30:54 pitronic:~/pruebas# 2015-07-10 20:31:06 works 2015-07-10 20:32:45 yeah, but [ echo | grep ... ] does not make sence 2015-07-10 20:33:05 *sense 2015-07-10 20:33:56 use something like: if echo | grep ....; then ... fi 2015-07-10 20:35:56 my solution was this, correct me if that's wrong 2015-07-10 20:36:01 gre(){ 2015-07-10 20:36:01 echo $1 | grep -E '^[0-9]+$' 2015-07-10 20:36:01 return $? 2015-07-10 20:36:01 } 2015-07-10 20:36:31 if ! [[ $(gre $OCTET) ]]; then bla...bla 2015-07-10 20:36:58 vitronic: correct, but you can drop the return $? 2015-07-10 20:37:20 and [[ is not POSIX sh compatible, but a compile option for bash support in ash 2015-07-10 20:37:47 if ! [ "$(gre $OCTET) ]; then ... 2015-07-10 20:38:00 is "wrong" 2015-07-10 20:38:12 if ! gre $OCTET; then ... 2015-07-10 20:38:23 is better 2015-07-10 20:38:51 ok, grateful, I'll try :-D 2015-07-10 20:39:48 vitronic: like almost everyone new to shell, they think [ or [[ is somehow related to the if statement 2015-07-10 20:39:51 I had never worked with ash 2015-07-10 20:40:17 but [ is syntactic sugar for test(1) 2015-07-10 20:41:08 if you call programs/functions in if statements, drop the [ ] 2015-07-10 20:41:35 which returns non-zero on failure, that is 2015-07-10 20:48:33 uggedal, works, thanks for the tip 2015-07-10 21:20:22 is there sound support in alpine-rpi image ? 2015-07-10 21:22:25 how can I load sound module on Pi ? depmod is not working and insmod / modprobe will not load the needed module -.- 2015-07-10 21:23:10 ok got it ... prefix snd was missing ... my fail 2015-07-11 01:59:22 >>> WARNING: networkmanager-dev*: Could not find any provider for pc:dbus-glib-1 2015-07-11 01:59:22 >>> WARNING: networkmanager-dev*: dbus-glib-dev needs to be rebuilt 2015-07-11 03:12:07 I bumped dbus-glib & fixed $depends_dev warnings 2015-07-11 05:27:12 anyone know why the subshell of the Midnight Commander does not work in alpine? 2015-07-11 10:28:00 update main/libowfat pls to v0.30 2015-07-11 10:31:56 maybe also opentracker, it has moved to git, http://erdgeist.org/gitweb/opentracker/ 2015-07-11 10:37:58 there is also fork with releases https://github.com/flygoast/opentracker 2015-07-11 15:25:58 why is there no samba-tool with samba? o.O i have samba-common-tools installed too 2015-07-13 07:54:53 there is a newer acpica released 2015-07-13 09:57:29 ncopa: latest edge snapshot is from march? 2015-07-13 09:59:43 yes 2015-07-13 10:03:50 can you push a edge tag? 2015-07-13 10:34:42 alacerda_: around? 2015-07-13 10:58:25 clandmeter, yeap 2015-07-13 10:59:15 regarding setup-box 2015-07-13 10:59:22 2 comments 2015-07-13 10:59:34 would be nice to see which packages its trying to install 2015-07-13 10:59:46 would also be nice if i can later select which package i dont want to use. 2015-07-13 11:00:20 alacerda: they all belong to a virtual pacakge 2015-07-13 11:03:23 clandmeter, isn't it possible to apk del ? 2015-07-13 11:04:00 or are you suggesting to have a remove option on setup-box ? 2015-07-13 11:05:06 alacerda: i think its nice it sets up a base for me to work with 2015-07-13 11:05:18 but i should be able to remove some X drivers when i know they are not needed 2015-07-13 11:05:21 but currently im not? 2015-07-13 11:05:40 World updated, but the following packages are not removed due to: xf86-video-s3virge: .setup-box 2015-07-13 11:23:51 are there any official docker image? i would like to try let say irssi in docker with minimalistic docker image 2015-07-13 11:29:05 ncopa: have you seen this error when trying to start X? http://sprunge.us/SWHW 2015-07-13 11:31:59 crow: gliderlabs made one that is supposed to be "official" 2015-07-13 11:32:07 we don't support it per se, of course 2015-07-13 11:32:22 docker pull alpine:v3.2 2015-07-13 11:32:36 (or however they tag it) 2015-07-13 11:32:44 it's the best one around 2015-07-13 12:06:23 barthalion thank you, i am going to try it then 2015-07-13 12:06:50 https://github.com/gliderlabs/docker-alpine i suppose 2015-07-13 12:35:44 clandmeter: i get what your saying, but for video card drivers, you won't need to remove anything once we have probing sorted out. 2015-07-13 12:41:45 ScrumpyJack: what do you mean with probing? 2015-07-13 12:44:34 2015-07-13 12:46:36 clandmeter: video card probing to only install the relevant video drivers 2015-07-13 12:57:54 hi 2015-07-13 12:57:58 can please someone take a look at zfs? 2015-07-13 13:03:11 ScrumpyJack: you mention "once we have" who is we? and when will it happen? 2015-07-13 13:07:27 read "once Alpine Linux has it", which will happen when someone sits down to figure out how it is done. Is that what you meant? 2015-07-13 13:09:33 2015-07-13 13:16:35 ScrumpyJack: yes :) 2015-07-13 14:04:37 shafire: http://irclogger.com/.alpine-devel/2015-07-03#1435945035 2015-07-13 14:05:42 :-( 2015-07-13 14:06:08 thats the legal part 2015-07-13 14:06:59 it did (or still does) work 2015-07-13 14:18:58 so, zfs on linux is cddl 2015-07-13 14:19:12 and libtirpc is bsd 2015-07-13 14:21:41 yes im also checking 2015-07-13 14:21:52 seems libtirpc is bsd isntead of gpl2? 2015-07-13 14:22:11 so standard 3/4-clause bsd licenses aren't incompatible with cddl 2015-07-13 14:22:50 i have no idea aobut licenses, but freebsd is compatable, sounds very bsd like. 2015-07-13 14:23:08 hOGVgco2VEsY: ping 2015-07-13 14:23:24 whats about https://github.com/idunham/libdrpc ? 2015-07-13 14:23:56 libdrpc Port of RPC based on Android's libdrpc. Replaces parts of older glibc RPC functionality and/or libtirpc. 2015-07-13 14:53:09 clandmeter: pong 2015-07-13 14:53:37 oh i see 2015-07-13 14:53:40 let me check 2015-07-13 14:53:57 i think the license in apkbuild is wrong? 2015-07-13 15:13:31 clandmeter: i also hilite on stf btw, if you don't want to look up my current id here. 2015-07-13 15:21:13 :) 2015-07-13 15:21:23 i just take the first strange nick avaiable 2015-07-13 15:24:47 :) 2015-07-13 15:25:13 hOGVgco2VEsY: did you check the license? 2015-07-13 15:25:16 it seems i have been talking bullshit re libtrpc license. i just cloned the gitrepo and it seems 3 clause bsd 2015-07-13 15:25:45 i'm terribly sorry 2015-07-13 15:26:09 haha 2015-07-13 15:26:28 /o\ 2015-07-13 15:29:30 uh, here's the source of my false information: https://github.com/zfsonlinux/zfs/pull/2604#commitcomment-7429884 2015-07-13 15:29:53 yes i read it today 2015-07-13 15:30:04 WE license it as GPL 2015-07-13 15:30:25 hmmm the APK seems to be wrong then. 2015-07-13 15:30:35 http://pkgs.alpinelinux.org/packages?package=libtirpc&repo=all&arch=x86 2015-07-13 15:31:38 i wonder who did that 2015-07-13 15:31:41 probalby me 2015-07-13 15:31:42 :) 2015-07-13 15:32:11 http://git.alpinelinux.org/cgit/aports/commit/testing/libtirpc/APKBUILD?id=a50c70e1301adaa0886755bef95e290f4c64d531 2015-07-13 15:32:14 we can blame ncopa 2015-07-13 15:34:54 hOGVgco2VEsY: updated pkg to reflect correct license 2015-07-13 15:35:34 seems spl copies the modules to the wrong directory 2015-07-13 15:36:02 i just commented on the github PR again: https://github.com/zfsonlinux/zfs/pull/2604#issuecomment-120970539 2015-07-13 16:30:02 please, MATE upgrades: http://sprunge.us/DViU 2015-07-13 17:05:09 ncopa, clandmeter ^^ 2015-07-13 17:33:54 kaniini ok me again, for the ghc port I think i've got it all squared away, https://github.com/mitchty/alpine-linux-ghc-bootstrap/tree/apkbuild/ghc holds everything needed to bootstrap/build ghc apks, the caveat being the bootstrap requires docker, but the result is a ghc cross compiler targeting /usr 2015-07-13 17:34:14 then to build a ghc-bootstrap apk you do https://github.com/mitchty/alpine-linux-ghc-bootstrap/blob/apkbuild/Dockerfile#L39 2015-07-13 17:34:31 and to use that to build a ghc apk https://github.com/mitchty/alpine-linux-ghc-bootstrap/blob/apkbuild/Dockerfile#L47 2015-07-13 17:34:56 otherwise a normal abuild will work https://github.com/mitchty/alpine-linux-ghc-bootstrap/blob/apkbuild/Dockerfile#L67 2015-07-13 17:35:11 when you have the apk's built 2015-07-13 17:35:26 been using that dockerfile to validate the entire run/build 2015-07-13 17:36:03 so if that looks okish should be able to get that into the testing repo I think 2015-07-13 17:37:30 as for the bootstrap tar.xz file can use the one I generated or build another and add that into the apkbuild when its somewhere more "official", I just used an internal system at home 2015-07-14 07:51:59 fabled: should the kernel's include/generated/utsrelease.h match the $(uname -r)? 2015-07-14 07:52:17 the utsrelease.h has: 2015-07-14 07:52:18 #define UTS_RELEASE "3.18.18" 2015-07-14 07:52:25 i wonder if it should be: 2015-07-14 07:52:32 #define UTS_RELEASE "3.18.18-grsec-1" 2015-07-14 07:52:41 not sure 2015-07-14 07:53:31 itshould match makefile's KERNELRELEASE 2015-07-14 07:53:32 # Read KERNELRELEASE from include/config/kernel.release (if it exists) 2015-07-14 07:53:32 KERNELRELEASE = $(shell cat include/config/kernel.release 2> /dev/null) 2015-07-14 07:53:33 KERNELVERSION = $(VERSION)$(if $(PATCHLEVEL),.$(PATCHLEVEL)$(if $(SUBLEVEL),.$(S 2015-07-14 07:53:33 UBLEVEL)))$(EXTRAVERSION) 2015-07-14 07:54:03 so i think 3.18.18 is correct 2015-07-14 07:54:21 ok 2015-07-14 07:54:30 the is spl/zfs configure script broke 2015-07-14 07:54:34 then* 2015-07-14 07:54:46 they install the kernel modules in there 2015-07-14 07:54:52 but i'm not fully sure 2015-07-14 07:55:00 what it is normally? 2015-07-14 07:55:39 oh 2015-07-14 07:55:40 3.18.18-0-grsec 2015-07-14 07:55:46 in linux-grsec 2015-07-14 07:56:02 ttdev-edge-x86_64:~/aports/main/linux-grsec/src/build-grsec$ cat include/config/ 2015-07-14 07:56:02 kernel.release 2015-07-14 07:56:02 3.18.18-0-grsec 2015-07-14 07:58:23 yea, the question is if you can use utsrelease.h to figure out the dest dir for kernel modules 2015-07-14 07:58:39 like they do here: https://github.com/zfsonlinux/spl/blob/master/config/spl-build.m4#L133 2015-07-14 07:58:53 either they are doing it wrong, or our utsrelease.h is wrong 2015-07-14 07:59:04 and should include the -grsec-1 suffix 2015-07-14 08:01:21 i get the feeling that our utsrelease.h is wrong 2015-07-14 08:07:30 ok the utsrelease does not match KERNELRELEASE 2015-07-14 08:07:33 its a bug then 2015-07-14 12:24:04 ncopa, is this error known? 2015-07-14 12:24:06 apk add iftop 2015-07-14 12:24:06 (1/1) Installing libtasn1 (3.6-r0) 2015-07-14 12:24:06 0% [ ]wget: server returned error: HTTP/1.1 404 Not Found 2015-07-14 12:24:06 ERROR: libtasn1-3.6-r0: Bad address 2015-07-14 12:24:09 1 errors; 124 MiB in 96 packages 2015-07-14 12:24:22 cat /etc/alpine-release 2015-07-14 12:24:22 2.7.5 2015-07-14 12:24:35 apk update? 2015-07-14 12:24:54 :P ooops 2015-07-14 12:25:17 that solves.. 2015-07-14 12:28:41 ;) 2015-07-14 12:33:51 2.7!!! apk upgrade --everything ;-) 2015-07-14 13:02:48 at least before nov 1 2015-07-14 16:27:52 please, MATE upgrade: http://sprunge.us/CNaB 2015-07-14 22:38:14 good evening 2015-07-15 07:43:44 hum 2015-07-15 07:43:50 i'm looking at eudev 2015-07-15 07:43:57 we have a challenge 2015-07-15 07:44:08 eudev uses libkmod 2015-07-15 07:44:24 libkmod depends on other format of depmod output 2015-07-15 07:44:30 a binary ouput format 2015-07-15 07:44:43 the busybox depmod does not generate that 2015-07-15 07:45:03 which means that busybox depmod does not work with libkmod 2015-07-15 07:45:36 i wonder if we should just depend on kmod for kernels? 2015-07-15 07:54:30 sound much easier then hacking busybox 2015-07-15 07:56:15 well 2015-07-15 07:56:35 it would render the busybox depmod completely useless 2015-07-15 07:57:43 we need modules.dep.bin modules.alias.bin modules.symbols.bin modules.builtin.bin 2015-07-15 07:57:46 for libkmod to work 2015-07-15 07:57:54 also 2015-07-15 07:57:57 to be honest 2015-07-15 07:58:14 i am a bit scared that kmod will be sucked into systemd source tree 2015-07-15 07:58:35 what about toybox? 2015-07-15 07:59:04 we cannot replace busybox til toybox has equivalent functionality 2015-07-15 07:59:15 and i dont think we want depend on both busybox and toybox 2015-07-15 07:59:33 and i dont think tobybox depmod is better 2015-07-15 07:59:56 i think it is possible to disable libkmod from eudev 2015-07-15 08:00:09 but that means it would fork/exec for every modpropbe (i think) 2015-07-15 08:00:59 fabled did some hacking on busybox modprobe iirc 2015-07-15 08:01:40 yeh 2015-07-15 08:01:49 i actually put that on my todo 2015-07-15 08:01:53 to do the binary format 2015-07-15 08:01:55 never got to it 2015-07-15 08:02:00 it would speed up also modprobe 2015-07-15 08:02:12 the format is relatively simple 2015-07-15 08:02:20 cool 2015-07-15 08:02:25 alternatives we have: 2015-07-15 08:02:31 i remember looking at it ~1 month ago or so 2015-07-15 08:02:59 - fork exec modprobe (as is) - this is slow, but will work as is 2015-07-15 08:03:30 - make our own lib which is compat with libkmod 2015-07-15 08:03:39 - depend on kmod 2015-07-15 08:03:53 - fix busybox to support .bin format 2015-07-15 08:04:33 fixing busybox means that we need generate all 4: modules.dep.bin modules.alias.bin modules.symbols.bin modules.builtin.bin 2015-07-15 08:07:49 that's simple 2015-07-15 08:07:52 i think they have same format 2015-07-15 08:07:59 ok 2015-07-15 08:08:00 i could look at that 2015-07-15 08:08:10 probably a bit later this week or early next week 2015-07-15 08:08:21 would be great 2015-07-15 08:08:29 fixing busybox is the best i think 2015-07-15 08:08:37 agreed 2015-07-15 08:10:09 woudl also be awesome if modprobe could read aliases from stdin or other fd 2015-07-15 08:10:17 and timeout after a sec or 2 2015-07-15 08:10:30 oh? 2015-07-15 08:10:37 hmm 2015-07-15 08:10:56 maybe timeout is not needed 2015-07-15 08:11:38 that way my nlplug could use spawn "modprobe --stdin" and write each modalias 2015-07-15 08:11:38 you probably want to do timeout on the other end 2015-07-15 08:11:46 yes 2015-07-15 08:11:58 doing --stdin would probably be ok 2015-07-15 08:12:02 and not too complicated 2015-07-15 08:12:19 i looked a bit at it, i think modprobe needs some refactoring 2015-07-15 08:12:35 they kinda depend on a fixed arg array as it is now 2015-07-15 08:13:28 oh, ok 2015-07-15 08:13:46 if that is done, would the binary depmod files still be needed? 2015-07-15 08:14:13 for eudev 2015-07-15 08:14:18 ok 2015-07-15 08:14:20 but 2015-07-15 08:14:38 i would prefer make our own udev at some point 2015-07-15 08:14:45 i dont know if that is realistic though 2015-07-15 08:14:56 eudev can run with fork/exec modprobe too 2015-07-15 08:15:06 so it will work just fine, just be a bit slower 2015-07-15 08:15:29 same for my nlplug 2015-07-15 08:15:55 i might do nlplug with fork/exec both mdev and modprobe initially 2015-07-15 08:16:08 1) make it work, 2) make it fast 2015-07-15 08:16:42 actually 2015-07-15 08:16:59 i could probably patch eudev to pass modalis via pipe too 2015-07-15 08:18:49 i suppose binary fomrats would make bb modprobe faster too, so that would probably be a good start 2015-07-15 09:44:22 hum 2015-07-15 09:44:27 moving to eudev will not be easy 2015-07-15 10:24:54 did the profanity build? my screen died for some reason 2015-07-15 11:07:53 barthalion: yes, it built 2015-07-15 11:09:29 good 2015-07-15 11:09:31 thx 2015-07-15 11:21:55 what's the best way to handle optional dependencis in apkbuild? With post-message? Even if there's not yet officially README.alpine trigger? 2015-07-15 11:22:02 *dependencies 2015-07-15 11:28:17 doc it on wiki ;) 2015-07-15 11:29:25 ncopa, is README.alpine patch is not yet applied :) 2015-07-15 11:29:34 *the README.alpine 2015-07-15 11:29:39 dont remember 2015-07-15 11:29:45 no, it's %246 2015-07-15 11:30:13 i didnt like the "(revised)" in commit message 2015-07-15 11:30:18 makes extra work for me 2015-07-15 11:31:18 http://wiki.alpinelinux.org/wiki/Creating_patches 2015-07-15 11:31:50 i'll look at it later.. 2015-07-15 11:32:06 no hurry.. 2015-07-15 11:53:50 ok it seems like eudev actually works 2015-07-15 13:18:50 i think i'm gonna push eudev to testing 2015-07-15 13:19:03 it might break things for those who installs it 2015-07-15 13:27:59 I'm trying to build a new package, but abuild doesn't work... just says "all failed" 2015-07-15 13:28:10 is there a way to get more information about what its trying to do? 2015-07-15 13:40:26 Fooster, pastebin the output and APKBUIILD file 2015-07-15 13:40:46 I figured it out actually, it was that I wasn't building as a user (I was root) 2015-07-15 13:40:52 =] 2015-07-15 13:40:59 I have another issue, however. The source files are local (not downloaded) 2015-07-15 13:41:15 when it tries to build the files are not in /var/cache/disfiles 2015-07-15 13:41:25 I'm basically trying to make a custom package for rubyu 1.9.3 2015-07-15 13:41:39 how is your source="" in APKBUILD ? 2015-07-15 13:42:07 source="ruby-1.9.3-p448.tar.bz2 2015-07-15 13:42:07 $patches" 2015-07-15 13:42:30 my APKBUILD is pretty much the ruby/APKBUILD :) 2015-07-15 13:43:12 http://pastebin.com/ynACq8PJ 2015-07-15 13:43:58 And the failurehttp://pastebin.com/EwB6TJwv 2015-07-15 13:46:24 weird you abuild checksum successfully (it seems yes, but better ask than assume)? 2015-07-15 13:46:38 yes 2015-07-15 13:49:30 based on: aports/main/alpine-mirrors/APKBUILD 2015-07-15 13:49:40 yours is right 2015-07-15 20:04:58 bonsoir 2015-07-16 08:40:46 morning 2015-07-16 08:51:58 hi 2015-07-16 08:52:05 i am working on perl-5.22 now 2015-07-16 08:52:31 far out 2015-07-16 08:53:48 can i open a bug for edge? (edge isn't in the list of affected versions) - or is there a policy 2015-07-16 11:23:06 please, MATE upgrade: http://sprunge.us/DiFQ 2015-07-16 12:04:42 forget it ^^ 2015-07-16 12:05:02 going to send another one with other upgrades 2015-07-16 12:10:08 ncopa - want me to resubmit apk-post-messages / tinyssh / lshell without "(revised)" in the commit message ? 2015-07-16 12:11:48 BitL0G1c: http://wiki.alpinelinux.org/wiki/Creating_patches#Resend_an_updated_patch 2015-07-16 12:12:04 yes please 2015-07-16 12:12:26 otherwise it will get recorded in the git log 2015-07-16 12:12:31 which is not a big problem 2015-07-16 12:12:35 just looks ugly 2015-07-16 12:13:53 ok no problem - won't take long 2015-07-16 12:14:13 thanks 2015-07-16 12:14:19 sorry for being picky... 2015-07-16 12:14:23 but its my job ;) 2015-07-16 12:16:37 please, MATE upgrade: http://sprunge.us/iDbF 2015-07-16 12:17:53 ;-) np 2015-07-16 12:18:11 ops 2015-07-16 12:18:24 ncopa, i've applied the (revised) commit :-( 2015-07-16 12:18:28 sorry 2015-07-16 12:18:37 you already said that, but i forgot 2015-07-16 12:18:47 lol 2015-07-16 12:19:31 at least, i didn't applied lshell and apk-post-message 2015-07-16 12:19:39 so, spanking would be softer :) 2015-07-16 12:21:01 http://git.alpinelinux.org/cgit/aports/log/ 2015-07-16 12:21:37 what will people think 2 years in future, when they look at the log and see "(revised)"? 2015-07-16 12:21:58 it does not matter really 2015-07-16 12:22:04 i'm just picky 2015-07-16 12:22:37 anyone, that tested setup-box: I plan to move it into main/. Any objection? 2015-07-16 12:22:47 fcolista: if/when BitL0G1c sends the fixed patch, can you please take care of it? 2015-07-16 12:23:04 alacerda: i have not tested it yet :-( 2015-07-16 12:23:13 ncopa, sure. 2015-07-16 12:23:13 :( 2015-07-16 12:23:42 alacerda, is setup-box documented somewhere? 2015-07-16 12:24:07 it has a manpage - is it valid? 2015-07-16 12:24:30 good enough for me... 2015-07-16 12:24:50 soudns good. 2015-07-16 12:25:02 btw, i was hoping to push eudev soonish 2015-07-16 12:25:04 today or tomorrow 2015-07-16 12:25:12 i'm gonna push perl 5.22 first though 2015-07-16 12:26:35 fcolista - ok - will just send lshell & apk-post-messages again 2015-07-16 12:26:44 k BitL0G1c 2015-07-16 12:36:26 ncopa, please: http://sprunge.us/iDbF 2015-07-16 12:50:47 alacerda: thanks 2015-07-16 12:55:03 i'm pushing perl 5.22 now... 2015-07-16 12:55:21 Your branch is ahead of 'origin/master' by 95 commits. 2015-07-16 12:55:34 here we go... 2015-07-16 13:20:52 moving setup-box v1.0.1 to main: http://sprunge.us/IZVY 2015-07-16 13:21:09 ncopa, thank *You* 2015-07-16 14:06:09 i wonder if i should rebuild all makedepends=udev-dev packages in testing first 2015-07-16 14:06:27 so we can test that it does not break the world 2015-07-16 14:08:47 ncopa, +1 2015-07-16 14:17:24 hm 2015-07-16 14:17:45 $ grep udev-dev */APKBUILD | wc -l 2015-07-16 14:17:45 34 2015-07-16 14:17:49 atleast 34 packages 2015-07-16 14:18:00 its a big project 2015-07-16 14:18:24 it would be much simpler to just move eudev to main and rebuild the packages 2015-07-16 14:18:31 i run eudev on my lpatop already 2015-07-16 14:18:55 it works more or less once xorg and xf86-input is rebuilt 2015-07-16 14:19:21 pushing directly to main would save me for work 2015-07-16 14:21:18 so if it is working in your laptop it means that you've already tested it - same as testing/ 2015-07-16 14:21:19 :P 2015-07-16 14:21:39 exactly... :) 2015-07-16 14:21:56 i think it will be ok 2015-07-16 14:22:03 but it is a bit risky 2015-07-16 14:24:39 ncopa, if sth goes wrong in main/ I think you'll have more work than if you work with testing/ first. 2015-07-16 14:26:16 sorry, now I am the one being picky (read: craven) 2015-07-16 14:26:46 :) 2015-07-16 14:35:15 What do You think of it: somebody that knows the difference between the AL ISOs images avaliable could add a description to each one. Maybe in dowload page. http://www.alpinelinux.org/downloads/ 2015-07-16 14:36:20 this idea is based on the review here: http://distrowatch.com/weekly.php?issue=20150706#alpine 2015-07-16 14:44:01 i just wrote an email to hetzner online asking them to provide alpine linux in there iso collection for an easy install 2015-07-16 14:44:06 i let you know there response 2015-07-16 17:50:43 alacerda: a good idea. 2015-07-16 18:03:58 please: http://sprunge.us/YZZN 2015-07-16 18:09:28 kaniini, clandmeter, ncopa, or someone else could add this explanation to the download page. 2015-07-16 18:09:43 ACTION doesn't know the difference between them 2015-07-17 07:44:53 clandmeter: after i push eudev changes, chrome will need change udev-dev -> eudev-dev and rebuild 2015-07-17 07:58:20 eudev pushed... 2015-07-17 08:14:53 ok 2015-07-17 08:14:58 i think i can push it to main 2015-07-17 08:17:48 ncopa: moving to main will init a rebuild right? 2015-07-17 08:22:34 yes 2015-07-17 08:22:43 but you need change udev-dev -> eudev-dev too 2015-07-17 08:23:47 oh 2015-07-17 08:26:23 ncopa: no virtual pkg? 2015-07-17 08:26:36 or however thats called. 2015-07-17 08:26:41 i tried to avoid it 2015-07-17 08:26:44 for now 2015-07-17 08:26:48 i will probably add it later 2015-07-17 08:27:01 because i want packages to be fixed up 2015-07-17 08:27:33 so ihave to really cnange to uedev now i guess? 2015-07-17 08:27:43 yes 2015-07-17 08:28:01 apk upgrade will probably break 2015-07-17 08:28:30 maybe 2015-07-17 08:28:34 probably not 2015-07-17 08:30:06 ah you replaced all of them already in main 2015-07-17 08:31:51 and you kept the udev scriptname the same 2015-07-17 08:59:10 eudev pushed... 2015-07-17 09:03:05 hum 2015-07-17 09:04:22 things didnt go so well in qemu... 2015-07-17 09:04:29 lets se how things are on my lappy 2015-07-17 09:06:09 :-( 2015-07-17 09:06:48 xorg's input modules does not work 2015-07-17 09:10:36 ok 2015-07-17 09:10:53 seems like the udevadm trigger does not force the input devs to be created 2015-07-17 09:11:07 guys be careful when reboot now 2015-07-17 09:14:50 barthalion: i pushed eudev 2015-07-17 09:14:58 having issues with modules not beeing loaded at boot 2015-07-17 09:15:06 i'm investigating 2015-07-17 09:16:27 cool 2015-07-17 09:16:39 not cool it's broken, but cool in general 2015-07-17 09:16:47 :) 2015-07-17 09:16:48 yeah 2015-07-17 09:18:42 oh 2015-07-17 09:18:46 i think we have a problem 2015-07-17 09:18:50 /o\ 2015-07-17 09:21:49 ncopa: no lunch time for you? 2015-07-17 09:26:09 sigh, something is completely screwed on Arch or on Debian 2015-07-17 09:26:25 not sure yet if it's actually distro problem or mosh sucks 2015-07-17 09:27:05 or maybe everything is broken and we all should have become medicine doctors 2015-07-17 09:32:09 and now I can't even trigger this 2015-07-17 09:32:10 argh 2015-07-17 09:32:35 I have noticed it always break when I await a message from someone or pushed something to aports 2015-07-17 09:35:38 fabled: it seems that will will need bb depmod fixes sooner than later :-/ 2015-07-17 09:36:02 apparently eudev without libkmod will simply not load modaliases at all 2015-07-17 09:36:35 i wrongly assumed it would fall back to fork/exec modprobe 2015-07-17 09:46:12 i see 2015-07-17 09:46:37 i have few things still in my queue. i can probably look at it later today / by monday 2015-07-17 09:47:59 we could also make it depend on kmod for now 2015-07-17 09:48:10 kmod's depmod 2015-07-17 09:48:22 then there is no hurry 2015-07-17 09:49:03 and meanwhile, i will force loading of "hwdrivers" 2015-07-17 09:49:12 that way we dont totally break xorg for now 2015-07-17 10:48:26 i'm trying to play with alpine-iso, I have a package list, and run Make. 2015-07-17 10:49:27 please: http://sprunge.us/YZZN 2015-07-17 10:50:01 iso build fine, but my package file turns out to be a list of packages to put in /media/cdrom/apks/ ... 2015-07-17 10:50:51 id like the packages to be installed, and in a apkvol so that they load at boot. 2015-07-17 10:50:59 any advice? 2015-07-17 10:51:53 i think you can generate a ovl and specify to include it. check makefile. 2015-07-17 10:54:06 ok 2015-07-17 10:55:29 so i have to generate the ovl manually seperatly? 2015-07-17 10:56:20 i think so 2015-07-17 10:56:29 i never tried it, but ive seen it mentioned before. 2015-07-17 10:57:15 what's the best way to generate an ovl? 2015-07-17 10:58:58 boot iso make changes and lbu_commit 2015-07-17 10:59:26 or analyze how its build and generate one by hand 2015-07-17 11:00:32 I was afraid you might say that :) 2015-07-17 11:05:43 an apkfile is baiscally only a tarball of modified files in /etc 2015-07-17 11:05:51 you can: mkdir /etc 2015-07-17 11:05:58 you can: mkdir tmp/etc 2015-07-17 11:06:03 edit files in there 2015-07-17 11:06:20 then cd tmp/ && tar -zcf host.apkovl.tar.gz 2015-07-17 11:06:23 then cd tmp/ && tar -zcf host.apkovl.tar.gz etc/ 2015-07-17 11:16:35 doh 2015-07-17 11:16:50 i though xorg was broke in qemu 2015-07-17 11:16:55 it turns out that it is just slow 2015-07-17 11:17:11 it waits for a few minutes and then suddenly it works 2015-07-17 11:17:21 if a patch does not apply, I should set it as "rejected" or "Not Applicable" ? 2015-07-17 11:17:22 i saw it on my eee pc too 2015-07-17 11:17:35 i use "Not Applicable" 2015-07-17 11:17:38 k 2015-07-17 11:17:51 ok nice, mate actually works in qemu 2015-07-17 11:30:08 new aport: http://sprunge.us/DZQP 2015-07-17 11:31:30 fabled: if there are 2 different packages that has a trigger for a given dir, will both triggers execute? 2015-07-17 11:31:44 ncopa, yes 2015-07-17 11:31:46 ncopa: so how do i get packages to actually load when i boot an iso? 2015-07-17 11:31:53 do i put them in world? 2015-07-17 11:32:02 s/put/list 2015-07-17 11:32:08 ScrumpyJack: add them to /etc/apk/world 2015-07-17 11:33:38 so, put them in world and make a apkovl which will catch that. cool 2015-07-17 11:33:45 ncopa, are you testing mate desktop? 2015-07-17 11:33:52 alacerda: yes 2015-07-17 11:33:57 :) 2015-07-17 11:33:59 i tested your setup-box script too 2015-07-17 11:34:02 ncopa: what did you use? 2015-07-17 11:34:10 xfce 2015-07-17 11:34:24 no, i mean what did you use to install. setup-box, cool 2015-07-17 11:34:51 ScrumpyJack, I've moved setup-box to main yesterday 2015-07-17 11:35:04 apk add setup-box (v1.0.1 in in edge) 2015-07-17 11:35:40 cool. most of the subpackages are in main too, so perhaps we could look at that subpackage patch 2015-07-17 11:35:55 only rawtherapee in not in main 2015-07-17 11:36:00 oh 2015-07-17 11:36:02 that too... 2015-07-17 11:36:10 there was a dep too 2015-07-17 11:36:20 ScrumpyJack, you mean that one ove moving some packages to main ? 2015-07-17 11:36:31 there was some dep that looked wrong 2015-07-17 11:36:35 ncopa: was going to have a look 2015-07-17 11:36:38 in rawtherapee 2015-07-17 11:36:42 in rawtherapee? 2015-07-17 11:36:43 ok 2015-07-17 11:37:05 libjpeg 2015-07-17 11:37:20 i dont know wtat that does there 2015-07-17 11:37:41 my guess it should be open-jpeg-turbo-dev instead 2015-07-17 11:37:42 in makedeps, hmm 2015-07-17 11:37:49 yes, you are right 2015-07-17 11:37:53 sorry 2015-07-17 11:38:03 i'll change it now and test 2015-07-17 11:38:09 we also need a maintainer for it 2015-07-17 11:38:43 libiptcdata also needs a maintainer 2015-07-17 11:39:06 ok. i leave maintainer empty on all my aports until I update them, in case someone else does it first :) 2015-07-17 11:51:02 ncopa, patchwork job is done so far 2015-07-17 11:51:23 fcolista: thanks! 2015-07-17 11:51:50 np. I've replied in ML for what's not working 2015-07-17 11:52:00 if you have any advice, are very welcome 2015-07-17 11:54:24 alacerda: good job witih mate! 2015-07-17 11:54:59 alacerda: also big thanks for high quality patches 2015-07-17 11:55:06 they always look good 2015-07-17 11:55:09 they always apply 2015-07-17 11:55:45 alacerda: i have given you git push access to aports 2015-07-17 11:56:59 thanks ncopa. You and all that use to help with my questions. 2015-07-17 11:57:22 clandmeter: re splitting the aports/main 2015-07-17 11:57:25 and thanks for the credit too 2015-07-17 11:57:47 i'd like to have 'core' and or long-term-support dir 2015-07-17 11:57:52 vs community dir or similar 2015-07-17 11:58:07 i dont know if we want those in different repos, but we probably do 2015-07-17 11:58:19 somethings like 'core' and 'extras' ala arch 2015-07-17 11:58:29 maybe submodule? 2015-07-17 11:58:44 like categories? 2015-07-17 11:59:04 what i want is first class and second class packages 2015-07-17 11:59:15 and that we only provide long term support for the firstclass 2015-07-17 11:59:39 so we get rid of testing? 2015-07-17 11:59:40 so we can add packages to a second class repo without worrying too much about maintenance 2015-07-17 11:59:45 could be 2015-07-17 11:59:57 testing might be good though 2015-07-17 12:00:00 for testing 2015-07-17 12:00:10 we might need a place where we can push things early 2015-07-17 12:01:58 or we could have a community/ repo 2015-07-17 12:02:13 but how to manage it? 2015-07-17 12:02:20 the problem is how should we actually split packages 2015-07-17 12:02:38 and how to name the dirs.. 2015-07-17 12:02:40 what decides about "coreness" of package 2015-07-17 12:02:47 i have some ideas 2015-07-17 12:02:50 it's not so clear even in Arch with [core] 2015-07-17 12:03:13 but yes it is very good question 2015-07-17 12:03:18 what i am thinking is 2015-07-17 12:03:31 the packages we can and want provide support for 2015-07-17 12:03:34 for 2 years 2015-07-17 12:04:06 basically, it the maintainer can provide support for it, including sec fixes 2015-07-17 12:04:09 for 2 years 2015-07-17 12:04:19 then its in core/main/whatever we call it 2015-07-17 12:04:32 if its not, the its in "extras" or what we end up calling it 2015-07-17 12:04:45 "main" is good and should stay 2015-07-17 12:04:52 I'm not sure about extra(s) 2015-07-17 12:04:59 the name you mean? 2015-07-17 12:05:01 it still looks like it's fully supported by us 2015-07-17 12:05:03 i like 'main' too 2015-07-17 12:05:04 yeah 2015-07-17 12:05:18 so community seems to be better choice 2015-07-17 12:05:25 we could call it 'community' 2015-07-17 12:05:28 or similar 2015-07-17 12:05:57 I have also another concern – build errors in "community" can't block "main" builds 2015-07-17 12:06:11 that is also a dilemma 2015-07-17 12:06:24 in fact, nothing in main can depend on community 2015-07-17 12:06:55 and another question is whether community should have its testing counterpart 2015-07-17 12:07:06 yes 2015-07-17 12:07:18 I think testing should be renamed to "staging" to be honest 2015-07-17 12:07:19 or if the 'testing' repo could be used for both main and community 2015-07-17 12:07:19 omg sounds complicated :) 2015-07-17 12:07:29 some packages sit there for many releases 2015-07-17 12:07:31 clandmeter: yes that too 2015-07-17 12:07:46 how do we solve this all without making it too complicated 2015-07-17 12:07:47 :) 2015-07-17 12:08:14 ok 2015-07-17 12:08:35 we could do: new pkg -> testing -> community -> main 2015-07-17 12:08:47 and maybe even rename testing to staging 2015-07-17 12:08:52 i kinda like that 2015-07-17 12:09:11 :) 2015-07-17 12:09:11 I like this last option 2015-07-17 12:09:13 +1 2015-07-17 12:09:36 if we resolve the issue with blocking builds, we can make push access less restricted to community 2015-07-17 12:10:03 you mean if someone push something to community repo 2015-07-17 12:10:10 and chokes builder 2015-07-17 12:10:15 it should still build and upload 'main' 2015-07-17 12:10:20 exactly 2015-07-17 12:10:28 that should be easy to solve 2015-07-17 12:10:30 trivial 2015-07-17 12:10:41 well, then I guess we have a resolution 2015-07-17 12:10:56 i would like to give it a weeks thought 2015-07-17 12:11:01 sure 2015-07-17 12:11:23 submit a concept on the ml 2015-07-17 12:11:29 i will 2015-07-17 12:11:43 I won't reply until monday though, just saying 2015-07-17 12:11:47 how do we handle permissions on community? 2015-07-17 12:12:14 that also need thought 2015-07-17 12:12:22 we have another ml where ppl can submit requests? 2015-07-17 12:12:25 if someone doesn't seem crazy and contributed a couple of times in general or one specific matter (erlang, or whatever), he's in 2015-07-17 12:12:39 we could either give push access to the packages he maintains 2015-07-17 12:12:41 assuming he does that regularly 2015-07-17 12:12:54 but, as ncopa says, we can just let people push changes to particular files 2015-07-17 12:13:00 or we could give push access to the entire 'community' repo 2015-07-17 12:13:10 gitolite allows that, but that makes thing even more complicated 2015-07-17 12:13:38 http://gitolite.com/gitolite/vref.html#NAME 2015-07-17 12:13:40 would be nice to avoid maintain the maintainer list in 2 places, both in APKBUILD and in gitolite config 2015-07-17 12:15:20 can we group packages by maintainer? 2015-07-17 12:15:27 it would be nice if we could have only 2 groups, maindevs and communitydevs 2015-07-17 12:15:47 and communitydevs gets git push access to community repo 2015-07-17 12:15:57 and maindevs gets access to both 2015-07-17 12:17:05 it woudl be nice if we can avoid being too granular on the permissions 2015-07-17 12:17:29 i mean, i prefer giving a little more permissions than strictly needed 2015-07-17 12:17:45 yeah, I guess we can just give access to entire community 2015-07-17 12:17:46 so if a maintainer is away for some reason, someone else can step in and push 2015-07-17 12:18:18 one thing more, while at it 2015-07-17 12:18:31 i'd like to move the alpine-iso scripts into the aports repo 2015-07-17 12:18:38 but thats a different topic 2015-07-17 12:18:51 maybe into a dir called 'release-scripts' 2015-07-17 12:19:20 that way we can tag/branch the package lists for the releases etc 2015-07-17 12:22:12 can't we just make a tag in aports when we do a release? 2015-07-17 12:22:55 we can 2015-07-17 12:23:08 but the scripts building iso is in different repo 2015-07-17 12:23:22 so if i change package list, i need git pull manually first 2015-07-17 12:23:55 well, i suppose that means 'no, we can't just make a tag in aports when we do a release' 2015-07-17 12:23:58 :) 2015-07-17 12:24:08 i have good news 2015-07-17 12:24:12 eudev works 2015-07-17 12:32:21 \o/ 2015-07-17 13:33:43 ncopa, how did eudev affected setup-box -d mate when you tested it? 2015-07-17 13:34:16 you had to manually apk add eudev ? 2015-07-17 13:34:18 what i tested was run setup-box to install mate desktop in qemu 2015-07-17 13:34:30 then i tested just apk upgrade 2015-07-17 13:34:47 i made a qemu snapshot first 2015-07-17 13:34:52 then i tested upgrade 2015-07-17 13:34:57 fixed the issues 2015-07-17 13:35:02 reverted to saved snapshot 2015-07-17 13:35:09 and repeated til it all was fixed 2015-07-17 13:35:25 so i have not tested setup-box with eudev yet 2015-07-17 13:48:36 i added a setup-udev script that will remove mdev and add needed services for eudev 2015-07-17 13:52:32 alacerda: i'm looking at setup-box now 2015-07-17 13:53:12 i dont think you should add udev service to any runlevel except sysinit 2015-07-17 13:53:35 and why is dbus needed in sysinit? 2015-07-17 13:55:31 oops, in fact it is needed in default 2015-07-17 13:56:33 thats what i epxect 2015-07-17 14:57:20 2015-07-17 15:04:42 alacerda: btw, i added a setup-udev script to eudev package 2015-07-17 15:05:12 it will rc-update the needed udev serivces to sysinit and will remove mdev and hwdrivers 2015-07-17 15:05:51 ok im going home 2015-07-17 15:06:01 im pretty happy with eudev 2015-07-17 15:06:14 it lets us move forward 2015-07-17 15:06:18 have a nice weekend 2015-07-17 16:11:59 2015-07-17 18:25:36 pnutzh4x0r, do you use abcde? 2015-07-17 18:30:09 oops, wrong channel 2015-07-17 18:30:11 alacerda: yes 2015-07-17 18:30:24 is it stable? 2015-07-17 18:30:56 i've only done basic ripping w/ it (ie. default options w/ mp3) 2015-07-17 18:31:04 it mostly works, depends on the disk 2015-07-17 18:31:14 if it is scratched, it can stall for a while 2015-07-17 18:31:22 otherwise, it's pretty straightforward 2015-07-17 18:43:32 pnutzh4x0r, is it a bug? 2015-07-17 18:50:07 alacerda: no, that's just the way it is w/ scratched disks 2015-07-17 18:52:11 I am trying to find a reason of why it is still in testing repo 2015-07-17 18:56:32 i never asked to move it 2015-07-17 18:56:40 i keep forgetting to ask to move my packages 2015-07-17 19:00:33 ahw :) 2015-07-17 20:42:12 clandmeter: hi, do I need spl also for zfs? 2015-07-17 20:42:30 I saw it in #4434 2015-07-18 09:15:42 ncopa: i just upgraded my destkop on edge. seems broken. 2015-07-18 09:20:23 clandmeter: saw the mail about eudev? 2015-07-18 09:20:37 nope 2015-07-18 09:21:23 eeks 2015-07-18 09:21:54 ill check it out later 2015-07-18 09:22:20 i think we will run into more issues 2015-07-18 09:22:32 kmod is not compatible with bb modprobe 2015-07-18 09:22:56 for instance alsa uses -l switch 2015-07-18 09:23:36 probably some of the init scripts have been patches for bb modprobe 2015-07-18 14:21:42 hi 2015-07-18 14:21:53 how do I pack something from sourceforge? because of the download url 2015-07-18 14:58:56 grep sourceforge ~/aports/*/*/APKBUILD 2015-07-18 15:00:22 something like http://downloads.sourceforge.net/$pkgname/$pkgname-$pkgver.tar.gz 2015-07-18 15:11:44 ok thanks 2015-07-18 16:42:48 hm... redmine thinks -blah- should be crossed out, clicking "pre" (preformatted text) inserts some [[ ]] around -blah- which results in blah ... how am I supposed to write -blah-? 2015-07-18 16:43:48 ah.. @foo@ or even 
foo
... 2015-07-18 16:44:28 clicking the left button would work too... [[]] actually was the one for hyperlinks %) 2015-07-18 17:03:25 ncopa, any comment about %271 ? I never seen "go get github" before in the APKBUILD..i'd like your comment to know if i can apply that patch. 2015-07-19 09:09:56 hey guys, for me the latest xorg-update broke my laptop 2015-07-19 09:10:08 either xfce or i3wm are broken 2015-07-19 09:10:15 the system freezes up as soon as i startx 2015-07-19 09:10:42 i am working on getting a stacktrace right now 2015-07-19 09:47:23 hmmm 2015-07-19 09:47:31 even eth0 is gone after that update 2015-07-19 09:47:38 what the hell changed? 2015-07-19 09:49:47 i get houndrets of evbug messages 2015-07-19 10:04:39 leo-unglaub: http://lists.alpinelinux.org/alpine-devel/4931.html 2015-07-19 10:07:56 AmatCoder: shit, i did not read the ml during my vacation and just did an update 2015-07-19 10:09:08 :/ 2015-07-19 10:09:25 networking and x are broken 2015-07-19 17:06:16 basically guys, don't use edge if you don't need to… 2015-07-19 17:06:46 except containers, I guess 2015-07-19 19:56:15 quick question, under which conditions may packages become promoted from testing to main (assuming that packages may not be included in a stable release from testing)? 2015-07-19 20:01:06 chris|: user has to report that the package works 2015-07-19 20:01:10 pretty much it 2015-07-19 20:01:23 most of packages rot there because we never hear back from package requester 2015-07-19 20:03:59 sounds easy enough 2015-07-19 20:04:40 ist there any designated channel for this or is alpine-devel list sufficient? 2015-07-19 20:05:06 you can even ask for it here 2015-07-19 20:05:40 usually during working hours in CET, but as you can see, I sometimes stay late 2015-07-19 20:06:35 we dislike bureaucracy ;) 2015-07-19 20:06:48 in that case I would like to request that skalibs, execline and s6 are moved over please :) 2015-07-19 20:15:02 sure thing 2015-07-19 20:15:23 cool, thank you :) 2015-07-19 20:16:17 np, done 2015-07-19 20:17:06 one more question: I am trying to figure out how to best port my postgres base to alpine but since alpine does not ship the old stable version of postgres, there is a broken upgrade path as pg_upgrade requires the binaries of bis the old and the new version. Any advice? 2015-07-19 20:30:21 probably nobody thought of that 2015-07-19 20:30:29 chris|: is arch solution satisfactionary? 2015-07-19 20:30:41 did I just invented new word? 2015-07-19 20:30:53 no, it's so old that nobody uses it 2015-07-19 20:30:58 enjoy my Engrish 2015-07-19 20:31:06 yes and maybe :) 2015-07-19 20:31:21 chris|: basically Arch ships a package called postgresql-old-upgrade 2015-07-19 20:31:33 I am using arch right now, worked fine so far 2015-07-19 20:31:55 that is one release behind postgresql and ships tools needed for upgrade in /opt/ 2015-07-19 20:32:13 ok, I'll push it tomorrow, if there is nothing like that in aports 2015-07-19 20:33:07 awesome, I'll give it a try as soon as I can 2015-07-20 06:33:36 morning 2015-07-20 06:33:51 good morning! 2015-07-20 06:33:53 for upgrade postgrsql i'd recommend doing dump/restore of the database 2015-07-20 06:34:02 that is the only safe way to do upgrade 2015-07-20 06:34:17 leo-unglaub: sorry for breaking your x 2015-07-20 06:34:29 can you check if you have /dev/input/* devices? 2015-07-20 06:34:48 i also noticed the evbug messages 2015-07-20 06:35:06 which indicates that /etc/modprobe.d/blacklist.conf is no longer respaected 2015-07-20 06:35:10 respected* 2015-07-20 06:35:19 i need to fix those things today 2015-07-20 06:35:32 ncopa: not your fault. it was my fault of running the upgrade without reading the ml :) 2015-07-20 06:36:35 i *try* avoid this kind of breakages 2015-07-20 06:36:48 and i thought it would just work in most cases 2015-07-20 06:36:52 apparently i was wrong 2015-07-20 06:36:58 i thought so too 2015-07-20 06:36:59 :) 2015-07-20 06:37:03 good morning 2015-07-20 06:37:06 morning 2015-07-20 06:37:10 lets fix this eudev mess now 2015-07-20 06:37:21 hey, at least this time we didn't lose 70% of packages in repositories 2015-07-20 06:37:26 lol 2015-07-20 06:37:56 that woud probably been easier to recover from 2015-07-20 06:38:00 ok 2015-07-20 06:38:15 so the plan is to fix busybox depmod 2015-07-20 06:38:19 fabled will look at that 2015-07-20 06:38:38 ncopa: do i need to cleanup some udev rules after the upgradE? 2015-07-20 06:38:41 well, he already started look at it and told me he has a plan 2015-07-20 06:38:44 or why did my nic get renamed? 2015-07-20 06:38:48 clandmeter: i dont know 2015-07-20 06:39:04 ncopa: run away to Mexico? 2015-07-20 06:39:10 i got this in private too: "after updating my alpine linux edge, wlan0 became wlp18s0b1" 2015-07-20 06:39:31 yeah, my interfaces are renamed to 2015-07-20 06:39:34 i dont think that happened to my laptop 2015-07-20 06:39:36 that's easy to fix 2015-07-20 06:39:41 there is a flag that you can set to avoid that 2015-07-20 06:39:43 assuming they pulled it from systemd 2015-07-20 06:39:51 yes 2015-07-20 06:39:57 leo-unglaub: you can also symlink some location to /dev/null 2015-07-20 06:40:01 easier to provide in package 2015-07-20 06:40:12 i think this is what systemd devs has decided to name the interfaces 2015-07-20 06:40:38 apparently they have a much better naming scheme 2015-07-20 06:40:54 this is the price for pulling in systemd bits... 2015-07-20 06:41:19 well, because it makes sense on desktops, and probably in some complicated networking setups on servers 2015-07-20 06:41:19 systemd ... i hate that word 2015-07-20 06:41:21 but anyway 2015-07-20 06:41:23 ln -s /dev/null /etc/udev/rules.d/80-net-setup-link.rules 2015-07-20 06:41:47 anyone affected, try this and reboot 2015-07-20 06:41:52 i said the s* word, sorry :) 2015-07-20 06:42:03 thanks barthalion 2015-07-20 06:42:50 ncopa: np, you should put this into eudev package for backwards compat 2015-07-20 06:45:42 i dont see any 80-net-setup-link.rules in eudev package? 2015-07-20 06:46:23 ncopa: yes, i have /dev/input devices 2015-07-20 06:46:28 I don't know what eudev does 2015-07-20 06:46:38 event0 -> event 14 2015-07-20 06:46:47 what systemd-udev will do after detecting that this path points to /dev/null, is reverting to old interface naming scheme 2015-07-20 06:46:47 and mice, mouse0, mouse1 2015-07-20 06:47:19 leo-unglaub: ok. good. when i discovered that coldplugging didnt work those were missing on my system, causing Xorg to "hang". once that was fixed, things worked 2015-07-20 06:48:59 we also need figure out why blacklist.conf is not respected 2015-07-20 06:50:27 hmmm, any idea why i worked for you and not for me? 2015-07-20 06:50:37 no 2015-07-20 06:50:49 one possibility might be that you ahve soemthing blacklisted 2015-07-20 06:51:06 but i doubt that is the problem here 2015-07-20 06:52:51 ok 2015-07-20 06:52:58 re net interfaces name 2015-07-20 06:53:20 leo-unglaub: tried that symlink? 2015-07-20 06:53:39 ln -s /dev/null /etc/udev/rules.d/80-net-name-slot.rules 2015-07-20 06:53:44 is the correct symlink 2015-07-20 06:54:10 so they pulled code from systemd-udev and forgot to update it afterwards 2015-07-20 06:54:38 or just kept it as it was for backwards compat on gentoo 2015-07-20 06:55:18 i think udev will let /etc/udev/rules.d/ override /lib/udev/rules.d/ 2015-07-20 06:55:29 so symlinking it to /dev/null makes udev ignore it 2015-07-20 06:55:52 i wonder if we should put that file in a subpackage for backwards compat 2015-07-20 06:56:13 i mean, if you want new net names, then you do: apk add eudev-netnames or similar 2015-07-20 06:57:14 i personally dont see any benefit in the new names 2015-07-20 06:57:22 its just different without any reason 2015-07-20 06:57:49 ncopa: in Arch we kept old interfaces for existing installs and enabled new naming for new ones 2015-07-20 06:58:09 i bet they have a good reason for it 2015-07-20 06:58:11 ncopa: and I honestly think it's way to go, although I don't know how it fits networking configuration in alpine 2015-07-20 06:59:01 iirc the problem is that net interfaces got names byt the order kernel found them 2015-07-20 06:59:27 so what nic is eth0, eth1 et2 etc completely depends on which order kernel detect them 2015-07-20 06:59:49 so if nic driver changes or something else changes, they got renamed 2015-07-20 06:59:55 we had that issues on alpine too 2015-07-20 07:00:22 the way we solved (or worked around it) was that we sort the modaliases 2015-07-20 07:00:34 that way the kernel modules always got loaded in same order 2015-07-20 07:00:43 and thus, eth0 always ended up as eth0 2015-07-20 07:01:34 i suspect different naming scheme of NIC is a way to solve this problem 2015-07-20 07:01:51 and it is likely a better way to solve it too 2015-07-20 07:03:24 how is this new name constructed? 2015-07-20 07:03:35 i dont know yet :) 2015-07-20 07:03:50 wl p18 s0 b1 2015-07-20 07:04:03 pci port 2015-07-20 07:04:17 probably something like that 2015-07-20 07:04:51 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames/ 2015-07-20 07:04:58 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames/ 2015-07-20 07:05:00 yeah :) 2015-07-20 07:05:46 hey dont steal my links 2015-07-20 07:05:56 :) 2015-07-20 07:06:02 copying is not stealing! 2015-07-20 07:06:04 :) 2015-07-20 07:06:26 http://cgit.freedesktop.org/systemd/systemd/tree/src/udev/udev-builtin-net_id.c#n20 2015-07-20 07:06:38 hehe i was waiting you copied that ;) 2015-07-20 07:07:23 so it actually makes sence 2015-07-20 07:07:35 it normally does 2015-07-20 07:07:47 they have this kind of policy in systemd? 2015-07-20 07:07:59 yes 2015-07-20 07:08:00 sorry, couldnt resist... 2015-07-20 07:08:20 ... i am leaving this channel until the s word discussion is over *g* 2015-07-20 07:09:12 i would have liked that kernel handled it 2015-07-20 07:09:33 i dont think we have much s lovers here, if we do i apologize. 2015-07-20 07:10:15 problem is 2015-07-20 07:10:30 we don't have any good alternative to eudev 2015-07-20 07:10:45 so we kinda have to deal with it, one way or the other 2015-07-20 07:10:49 that is sad but true 2015-07-20 07:11:03 and to be honest 2015-07-20 07:11:10 i do have a bad taste in my mouth 2015-07-20 07:19:09 ok 2015-07-20 07:19:20 re network interfaces names 2015-07-20 07:19:33 we will have to make it work similar with or without eudev 2015-07-20 07:20:23 we will need a way to make sure that interface name does not chan (eth0 -> eth1 and the other way) 2015-07-20 07:33:30 clandmeter, you around? 2015-07-20 07:36:38 a UUID for network interfaces would be the right way to go 2015-07-20 07:37:51 but that is kernel side 2015-07-20 07:37:56 they would have to do that 2015-07-20 07:41:19 I'm not sure about %271 patch: the usage of git 2015-07-20 07:41:49 i thinks is fragile 2015-07-20 07:42:18 if github repo changes for some reasons, build breaks. 2015-07-20 07:47:44 it uses tarball based on a tag 2015-07-20 07:47:52 and we have this "idiom" already in aports 2015-07-20 07:48:27 "go get github.com/robfig/glock" this is more wrong, actually 2015-07-20 07:50:00 this is what i'm referring to 2015-07-20 07:51:27 well, you said about git, not go 2015-07-20 07:52:15 i dont know what the best way is the handle go. everytime i bump into it, i try to solve it, but in the end its a mess. 2015-07-20 07:53:07 i have no idea how to deal with go either :) 2015-07-20 07:53:32 leo-unglaub: i agree that it would be nice if kernel solved it for us 2015-07-20 07:53:38 maybe based on MAC addr 2015-07-20 07:53:44 maybe we should just say NO to GO and let the user figure it out. 2015-07-20 07:53:52 they is what they want anyway 2015-07-20 07:54:00 s/they/this 2015-07-20 07:54:00 go kinda provide their own package manager yes 2015-07-20 07:54:14 ncopa: based on the mac is not a good idea 2015-07-20 07:54:22 leo-unglaub: ok? 2015-07-20 07:54:25 because every time you spoof your mac you have a different device 2015-07-20 07:54:29 true 2015-07-20 07:54:38 so how to generate the UUID? 2015-07-20 07:54:46 by random? 2015-07-20 07:54:51 the same way hdd vendors do it 2015-07-20 07:55:05 nic serial number 2015-07-20 07:55:07 cant you generate it always from the *real* mac? 2015-07-20 07:55:09 would be my guess 2015-07-20 07:55:28 isnt the serial number and Mac address same thing mroe or less? 2015-07-20 07:55:37 i am not sure about that 2015-07-20 07:56:31 added a section to w.a.o/Patch Workflow 2015-07-20 07:56:32 waaaa, debgging x sucks ... debugging x on a laptop with a 800x600 resolution sucks even more .... 2015-07-20 07:56:46 otoh, it is kinda nice to know that if you have a single ethernet NIC, the name will always be eth0 2015-07-20 07:57:05 bsd has named the interface based on what driver it uses 2015-07-20 07:57:15 makes it somewhat consistent i suppose 2015-07-20 07:57:18 fcolista: the idea is, the binary should always be the same for each version. GO could suddenly have a dep upgraded, and the result will be different. 2015-07-20 07:57:44 fcolista: so generate a tarball seems most logic to me. 2015-07-20 07:57:45 this mess up the version 2015-07-20 07:57:53 +1 for tarball 2015-07-20 07:58:08 so we can reply to use abuild git snapshot feature 2015-07-20 07:58:34 and maybe its me, but when i read GO docs, im always missing switches which are used in other dists. 2015-07-20 07:58:52 ACTION when reads GO docs vomits 2015-07-20 07:59:03 +1 2015-07-20 07:59:36 btw clandmeter 2015-07-20 07:59:48 patchwork is not able to catch patches sent as attachment 2015-07-20 07:59:52 but only inline 2015-07-20 08:00:01 yes 2015-07-20 08:00:04 so I added this note to w.a.o/patch workflow 2015-07-20 08:00:08 by design 2015-07-20 08:00:39 this is kinda bad, since we don't have a complete patch workflow management 2015-07-20 08:00:46 ive seen ncopa ask them as attachements when the inline ones dont work for some reason 2015-07-20 08:00:48 but..i can live with that 2015-07-20 08:01:56 we should have them inline, and try to prevent as attachment. 2015-07-20 08:02:11 some mailers mess things up 2015-07-20 08:02:12 like gmail 2015-07-20 08:02:15 ok. So write it on a doc would be the 1st step :) 2015-07-20 08:02:19 maybe find out the cause of the failures 2015-07-20 08:02:25 gmail finds out that it should wrap lines 2015-07-20 08:02:31 and put that on wiki 2015-07-20 08:02:52 what I have done when they send patch as attachment is 2015-07-20 08:03:06 i have to revert the eudev changes on my laptop 2015-07-20 08:03:24 i need that laptop to work tomorrow because i am out in the field with the red cross and i need it to be working 2015-07-20 08:03:31 after that i have time to debug it 2015-07-20 08:03:32 people should simply use git send-email 2015-07-20 08:03:36 opend it in claws mail, "view source", ctrl-a for mark all, ctrl-c for copy 2015-07-20 08:03:53 then in terminal i do: git am 2015-07-20 08:03:55 +1 with git send-mail 2015-07-20 08:04:03 and paste it there, and ctrl-d 2015-07-20 08:04:14 ncopa: we keep those special cases just for you ;-) 2015-07-20 08:04:24 :) 2015-07-20 08:04:30 i think git send-mail via gmail also breaks things 2015-07-20 08:04:46 simple solution: dont use gmail! 2015-07-20 08:04:51 yeah.. 2015-07-20 08:04:52 i just used gmail 2015-07-20 08:04:59 try out if it broke something :p 2015-07-20 08:05:07 patchwork 278 2015-07-20 08:05:19 shafire, last patch you sent (rng-tools) returns whitespace error 2015-07-20 08:05:23 but applied, though 2015-07-20 08:05:25 leo-unglaub: anything i can do to help? 2015-07-20 08:05:27 *applies 2015-07-20 08:05:30 so its gmail 2015-07-20 08:05:31 i want fix eudev 2015-07-20 08:05:52 seems so 2015-07-20 08:05:53 whitespace error doesnt need to be gmail issue 2015-07-20 08:06:07 i have tested eudev in qemu and on my laptop 2015-07-20 08:06:10 both places works 2015-07-20 08:06:40 only time things broken have been when something has hold back the old udev while libs uses new libudev 2015-07-20 08:06:45 umh 2015-07-20 08:06:52 rng-tools has textrels on x86 2015-07-20 08:06:56 uff 2015-07-20 08:07:31 http://pastebin.com/4ijy2V8e what should I do with this question? 2015-07-20 08:07:33 found a patch 2015-07-20 08:07:41 ncopa: hmmm, i am not sure what the problem is 2015-07-20 08:07:46 that makes it hard to debug 2015-07-20 08:11:20 i have evdev, evdev-libs, evdeb-libgudev, udev, udev-init-scripts isntalled 2015-07-20 08:11:24 so no old libs 2015-07-20 08:12:19 leo-unglaub: how are you going to revert? 2015-07-20 08:12:45 most of the packages are now build againts eudev 2015-07-20 08:13:47 clandmeter: reinstall with the latest stable release 2015-07-20 08:13:53 only way is probably stable rel 2015-07-20 08:14:01 or a lvm snapshot :) 2015-07-20 08:17:22 leo-unglaub: what happens when it freezes? 2015-07-20 08:17:38 can you ctrl-alt-f1 to get a terminal? 2015-07-20 08:17:51 no, i have to hard reset 2015-07-20 08:20:28 https://paste.debian.net/plain/284057 2015-07-20 08:32:13 this log is not helpful 2015-07-20 08:32:21 because it does not explain why there is an io error 2015-07-20 08:34:51 i think i had the same issue recently with vmware and xorg. 2015-07-20 08:36:21 leo-unglaub: what chipset? 2015-07-20 08:36:35 its an intel 2015-07-20 08:36:52 hum 2015-07-20 08:37:03 i have intel on my laptop 2015-07-20 08:37:34 hmmmmm 2015-07-20 08:37:43 it worked perfectly before the update 2015-07-20 08:37:47 so something must have changed 2015-07-20 08:38:41 intel is the only chipset i never had issues with (untill this weekend) 2015-07-20 08:39:20 leo-unglaub: are you already on stable? or still trying to fix edge? 2015-07-20 08:39:35 i am still on edge 2015-07-20 08:39:52 i plan to reinstall tonight 2015-07-20 09:01:21 fcolista: I think -dev can be removed from knot 2015-07-20 09:01:57 why? 2015-07-20 09:48:35 because there is nothing to put in -dev, or? 2015-07-20 09:49:17 rebooting my main desktop with eudev 2015-07-20 09:49:28 hopefully im back before lunch... 2015-07-20 09:55:33 i am hoping ncopa comes back *g* 2015-07-20 09:55:49 but that reboot already takes a loooong time *g* 2015-07-20 10:00:44 ha, he is back :) 2015-07-20 10:01:01 wb ncopa 2015-07-20 10:01:10 yeah 2015-07-20 10:01:19 not without a bit drama 2015-07-20 10:01:38 xorg didnt want start 2015-07-20 10:02:00 problem was that i915 kernel module was loaded without modeset=1 2015-07-20 10:02:11 basically /etc/modprobe.d/ is not consulted 2015-07-20 10:02:35 which also makes pcspkr and evbug load at boot, even if its blacklisted 2015-07-20 10:02:46 i need figureout why that happens 2015-07-20 10:02:55 this sounds like something that could cause my problem as well 2015-07-20 10:03:02 then after that, mouse did not work 2015-07-20 10:03:15 the bluetooth keyboard did work 2015-07-20 10:03:45 after unplugging and plugging my mouse, it started to work again 2015-07-20 10:03:55 so something is probably broke during coldplug 2015-07-20 10:04:21 but now i have lunch 2015-07-20 10:04:37 will try find out why /etc/modprobe.d is not loaded after lunch 2015-07-20 10:04:56 sure, no hurry. have a good lunch 2015-07-20 10:06:04 i think i found it... 2015-07-20 10:09:07 yeah 2015-07-20 10:09:12 sysconfdir is wrong 2015-07-20 10:09:14 in kmod 2015-07-20 10:09:46 pushed fix 2015-07-20 10:09:53 see you after lunch 2015-07-20 10:13:26 fyi: next pkgconf version will be pkgconf-1 instead of pkgconf 0.x, and the fake --version output is probably going away :) 2015-07-20 10:14:29 i am trying the fix as soon as it is replicated to the mirror servers 2015-07-20 10:14:32 :) 2015-07-20 10:20:19 use nl.alpinelinux.org 2015-07-20 10:20:24 it's the master mirror 2015-07-20 10:27:32 i normally avoid using the master mirror to keep the load low 2015-07-20 10:31:35 yeah, ncopa it works again with that fix! 2015-07-20 11:07:28 nice! 2015-07-20 11:31:24 someone running armhf on rpi from sd card in memory? 2015-07-20 11:39:25 shafire, yes 2015-07-20 11:39:43 i am. several. :) 2015-07-20 11:40:13 how did you do that? 2015-07-20 11:49:39 using the image online 2015-07-20 11:49:45 ;) 2015-07-20 11:49:50 what you mean exactly? 2015-07-20 11:50:07 basically this config: http://wiki.alpinelinux.org/wiki/RPI_Video_Receiver 2015-07-20 11:51:32 hmmmm 2015-07-20 11:51:48 the "problem" with the new interface names is now, that on servers you have the old ones 2015-07-20 11:51:55 but on desktops you have the new ones 2015-07-20 11:52:30 leo-unglaub: i sent email about that. i don't know what we want do there 2015-07-20 11:52:36 do we want keep old names everywhere? 2015-07-20 11:53:01 do we want deskop's iface names different from server setups (eg depending on if you run udev or not) 2015-07-20 11:53:18 or do we want try adapt to new names everywhere 2015-07-20 11:53:32 we will need some way to make the iface name persistant over reboots in any case 2015-07-20 11:53:45 i would have preferred that they solved it in kernel instead of in userspace 2015-07-20 12:00:07 fabled: ah okay 2015-07-20 12:00:10 thanks 2015-07-20 12:15:47 hmmm, i am not sure what the best solution is 2015-07-20 12:16:35 fcolista: im going to nitpick a bit 2015-07-20 12:16:52 fcolista: http://git.alpinelinux.org/cgit/aports/commit/?id=22306dd8 2015-07-20 12:17:01 ACTION grabs popcorn 2015-07-20 12:18:23 that commit msg, its a bit vague about which issue is actually fixed. 2015-07-20 12:20:05 clandmeter, yeah..just had in mind my reply in http://lists.alpinelinux.org/alpine-aports/0713.html 2015-07-20 12:20:12 barthalion: salty ot sweet? 2015-07-20 12:20:19 or.. 2015-07-20 12:20:30 but i should have been more wordly 2015-07-20 12:20:36 and accurate in msg 2015-07-20 12:20:38 +1 2015-07-20 12:22:49 sweet popcorn? eew 2015-07-20 12:44:07 could the following packages be moved from testing to main: http://sprunge.us/NITF 2015-07-20 12:52:40 ncopa, i'm wondering to move aircrack-ng from testing to main. 2015-07-20 12:55:06 fcolista: i have no objection if the APKBUILD looks ok and if its tested to work 2015-07-20 12:55:29 and that it has all needed .initd scripts etc if any 2015-07-20 12:56:39 gotta test all the packages involved, before pushing to main, then. 2015-07-20 12:58:16 yes 2015-07-20 12:58:42 check also that there is a maintainer 2015-07-20 12:58:59 if i am the maintainer, it might be that i move it to community later 2015-07-20 12:59:29 ncopa: did you fix eudev? 2015-07-20 13:00:45 clandmeter: i think so 2015-07-20 13:00:59 actually, i should reboot my desktop to verify 2015-07-20 13:01:08 rebooting 2015-07-20 13:01:09 inc kmod changes? 2015-07-20 13:01:12 compated to bb 2015-07-20 13:01:15 yes 2015-07-20 13:01:16 grr 2015-07-20 13:01:17 ncopa: i have an idea and are interrested in your feedback 2015-07-20 13:01:17 compared 2015-07-20 13:01:48 leo-unglaub: brb, i'll just reboot 2015-07-20 13:01:48 what do you think about an apk flad that always automatically installs the -doc package as well if available? 2015-07-20 13:02:02 that would be very helpful for development computers 2015-07-20 13:06:54 leo-unglaub: i have been thinking about that too 2015-07-20 13:07:15 soemthing like: apk add docs 2015-07-20 13:07:30 and it pulls in -doc if available 2015-07-20 13:07:35 via install_if 2015-07-20 13:08:13 it would meant that every -doc package would get an install_if= condition 2015-07-20 13:08:18 i dont know if we want that 2015-07-20 13:08:23 ncopa: dont get me wrong, i LOVE that docs are in a seperate package! it keeps my servers, routers, ... clean 2015-07-20 13:08:43 butr on my laptop i actually love al docs because sometimes i forgett about them and then i miss them when beeing on the train 2015-07-20 13:09:53 fabled: what do you think about to every -doc package add: install_if="docs $pkgname=$pkgver-r$pkgrel" 2015-07-20 13:09:55 $ apk search --exact '*-doc' | wc -l 2015-07-20 13:09:55 1829 2015-07-20 13:10:13 would have inpact on apk index size 2015-07-20 13:10:24 and index processing speed 2015-07-20 13:10:47 i dont know if it is significant impact though 2015-07-20 13:12:24 hmm, i dont think thats a problem 2015-07-20 13:12:36 because if you install dev packages you have even more dependencies 2015-07-20 13:14:33 or what about the other way around? 2015-07-20 13:14:35 or echo 1 > /etc/apk/doc 2015-07-20 13:15:03 apk add curl does and internal check if curl-doc is available and then isntall it directly 2015-07-20 13:15:12 that way you dont have to modify the doc package 2015-07-20 13:15:25 not sure i like that 2015-07-20 13:15:40 it means special handling for -doc 2015-07-20 13:16:00 apk would need treat some packages special depending on the name 2015-07-20 13:16:06 no, why? 2015-07-20 13:16:19 apk just looks if there is a doc package and if there is, isntall it as well 2015-07-20 13:16:42 what if you dont want doc packages? 2015-07-20 13:16:50 like for tmpfs installs 2015-07-20 13:17:00 ncopa: you only do that if if you enable it via config flag 2015-07-20 13:17:09 by default no doc packages are isntalled 2015-07-20 13:17:17 only if /etc/apk/docs is 1 2015-07-20 13:17:20 apk add doc sounds fancy, but i would prefer to have a config option. 2015-07-20 13:17:43 apk add docs has the probem that it always has to check all packages even it i just insall one new one 2015-07-20 13:17:43 what about -lang packages? 2015-07-20 13:18:05 apk does not work like that 2015-07-20 13:18:16 it check all deps even if you only add one package 2015-07-20 13:18:19 i would use /etc/apl/{lang,docs,dev} 2015-07-20 13:18:26 apk add docs looks cool 2015-07-20 13:18:28 or have /etc/apk/extra and list which should be auto pulled in 2015-07-20 13:18:52 but I come from the land of pacman, so disregard my big eyes 2015-07-20 13:19:13 apk would need calculate the dep graph anyway 2015-07-20 13:19:19 so 2015-07-20 13:19:23 if you apk add curl 2015-07-20 13:19:32 without setting /etc/apk/docs 2015-07-20 13:19:41 then later you change /etc/apk/docs 2015-07-20 13:19:57 should it pull in curl-doc at that point or not? 2015-07-20 13:20:01 hmm, you are right, if it changes you have to check everything again 2015-07-20 13:20:42 apk has a feature for what you are asking 2015-07-20 13:20:45 and its install_if 2015-07-20 13:21:02 but it needs to be used with care 2015-07-20 13:21:22 for example we use it for kernel modules 2015-07-20 13:21:26 so if you apk add linux-grsec 2015-07-20 13:21:38 and then apk add dahdi-linux 2015-07-20 13:21:49 it will automatically pull in dahdi-linux-grsec 2015-07-20 13:21:54 the module for that kernel flavor 2015-07-20 13:22:16 we do that for lua modules too 2015-07-20 13:22:26 apk add lua-filesystem lua5.3 2015-07-20 13:22:38 will pull in lua5.3-filesystem 2015-07-20 13:22:56 if you later add lua5.2 2015-07-20 13:23:06 it will also pull in lua5.2-filesystem 2015-07-20 13:23:34 if you apk del lua5.3 it will also purge lua5.3-filesystem 2015-07-20 13:23:52 we could do that with apk add docs 2015-07-20 13:24:16 if you did that, it could pull in -doc packages too 2015-07-20 13:25:34 sounds nice 2015-07-20 13:26:19 it is a relatively simple change in abuild 2015-07-20 13:26:31 and rebuild of all packages 2015-07-20 13:26:53 it'd be available in the v3.3 rebuild 2015-07-20 13:27:12 however, i dont know if we want that for absoultely all packages 2015-07-20 13:27:18 how long does then build servers need for a complete rebuild? 2015-07-20 13:27:30 due to the increased apkindex 2015-07-20 13:27:44 for edge i think we'd do it as the packages got updated 2015-07-20 13:28:06 full rebuild for x86_64/x86 are probably 2-3 days or so 2015-07-20 13:28:29 depends on how many packages are broke 2015-07-20 13:28:46 i see 2015-07-20 13:29:05 i agree its a solution, and maybe the best we have. but i do think a real config option would be more clear from a user point of view. 2015-07-20 13:29:13 can you think of packages that should not have that install_if thing? 2015-07-20 13:34:07 packages that does not have -doc at all? 2015-07-20 13:34:36 if a package has no -doc at all, why would that make the index bigger? 2015-07-20 13:34:46 it would not get installed, because it does not exist 2015-07-20 13:34:50 it wouldnt 2015-07-20 13:34:50 or do i get that wrong? 2015-07-20 13:34:59 its just that we have so main -doc packages 2015-07-20 13:35:03 so many* 2015-07-20 13:35:27 yes, but it would only install the -doc packages for packages i have installed 2015-07-20 13:35:41 i dont need the libreoffice doc package if i just have wget and curl 2015-07-20 13:36:11 apk reads the entire index 2015-07-20 13:36:52 in any case, fabled's opinion here matters 2015-07-20 13:37:20 i have no strong opinion 2015-07-20 13:37:30 might be good to have it 2015-07-20 13:37:58 i'm worried about that many many added install_if 2015-07-20 13:38:10 will have impact on dep graph calculation time 2015-07-20 13:38:21 and size increase of apkindex 2015-07-20 13:38:29 i dont know if that matters 2015-07-20 13:38:41 i dont know if the impact is significant i mean 2015-07-20 13:38:58 hard to say. it'll be more data, so it'll increase both. but i don't think it's significant increase. 2015-07-20 13:39:20 it wil not make things smaller and faster, thats for sure 2015-07-20 13:39:55 i think it could be a handy feature 2015-07-20 13:40:18 if a user does not install docs, none of the install_if checks would be executed, right? 2015-07-20 13:40:35 or would they be executed even if the docs package is not installed 2015-07-20 13:41:25 not sure 2015-07-20 13:41:36 the apkindex size wuld increase at least 2015-07-20 13:41:41 but probably not significant 2015-07-20 13:41:43 and achtyally 2015-07-20 13:41:45 actually 2015-07-20 13:41:59 if we split out a "community" repo 2015-07-20 13:42:26 then you'd get smaller index if you only used main 2015-07-20 13:43:09 so for the typical small installs where you dont have community, it wouldnt matter 2015-07-20 13:44:26 then the next question is 2015-07-20 13:44:41 do we do it as config option or as metapackage? 2015-07-20 13:44:54 eg /etc/apk/docs or apk add docs? 2015-07-20 13:45:15 i think I'd prefer: apk add docs 2015-07-20 13:45:44 that is your desicion :) 2015-07-20 13:46:14 you propobly know better whats best for the project 2015-07-20 13:47:11 i have no strong opinion ;) 2015-07-20 13:48:45 ok good :) 2015-07-20 13:48:47 i have now installed every -doc package i could find for my desktop installation 2015-07-20 13:48:55 and there is no speed difference 2015-07-20 13:49:16 only difference is that /lib/apk/db/installed is grown to 4 mb 2015-07-20 13:49:27 from? 2015-07-20 13:49:34 from 2,1 2015-07-20 13:49:46 okm but thats file lists 2015-07-20 13:49:50 that make that diff 2015-07-20 13:50:07 i suppose the extra install_if is insignificant 2015-07-20 13:50:28 on a normal desktop cpu you dont see a difference 2015-07-20 13:50:37 but on a raspery pi or something like that 2015-07-20 13:50:41 you meight see a difference 2015-07-20 13:50:48 but i dont have one to test it 2015-07-20 13:51:23 on x86/x86_64 it doesnt matter 2015-07-20 13:51:34 it might make difference on armhf 2015-07-20 13:51:42 but i doubt that too 2015-07-20 13:51:52 i dont know, sadly i cannot test that here 2015-07-20 13:51:52 dont forget my alix 2015-07-20 13:52:04 :) 2015-07-20 13:52:16 my precious 2015-07-20 13:52:20 :) 2015-07-20 13:53:15 hum 2015-07-20 13:53:22 i have a pending change 2015-07-20 13:53:27 -static subpackages 2015-07-20 14:14:57 leo-unglaub: i pushed testing/docs 2015-07-20 14:15:41 and patched abuild 2015-07-20 14:15:44 tis houdl work now 2015-07-20 14:16:35 ncopa: holy shit ... you are fast 2015-07-20 14:16:47 sometimes... 2015-07-20 14:17:01 i had tought of this for a while actually 2015-07-20 14:17:20 damn ... and i thougt i am the smart mastermind behind that idea *g* 2015-07-20 14:17:21 and patch for abuild was a one-liner 2015-07-20 14:17:29 :) 2015-07-20 14:17:35 i installed docs 2015-07-20 14:17:42 now i wait for new rebuild packages :) 2015-07-20 14:17:47 yes :) 2015-07-20 14:17:52 :) 2015-07-20 14:17:57 algitbot: rebuild world 2015-07-20 14:19:03 haha 2015-07-20 14:19:29 i wonder how long it takes for spammers to find this backdoor 2015-07-20 14:19:48 probably forever :) 2015-07-20 14:20:04 i'm pretty sure we'll notice when they do 2015-07-20 14:20:15 :) 2015-07-20 14:20:24 and we will think its you 2015-07-20 14:20:28 and its on my list to fix it 2015-07-20 14:24:57 leo-unglaub: anything small you use that i can rebuild as a test? 2015-07-20 14:26:48 curl 2015-07-20 14:26:56 or hiawatha 2015-07-20 14:27:16 or xfdesktop 2015-07-20 14:27:26 it actually needs a version bumb to 4.12.3 2015-07-20 14:27:38 yup 2015-07-20 14:29:22 and its pushed 2015-07-20 14:36:20 i got lost there. what's with the docs? a flag to auto install docs? 2015-07-20 14:36:32 ScrumpyJack: apk add docs 2015-07-20 14:36:53 will install all -doc packages 2015-07-20 14:37:03 ... at least the rebuild packages 2015-07-20 14:37:05 (1/4) Upgrading btrfs-progs (4.0.1-r0 -> 4.1.2-r0) 2015-07-20 14:37:05 ah ok. 2015-07-20 14:37:06 (2/4) Installing btrfs-progs-doc (4.1.2-r0) 2015-07-20 14:37:06 (3/4) Upgrading xfdesktop (4.12.2-r0 -> 4.12.3-r0) 2015-07-20 14:37:06 (4/4) Installing xfdesktop-doc (4.12.3-r0) 2015-07-20 14:37:09 awesome! 2015-07-20 14:37:54 I moved 'docs' to main 2015-07-20 14:38:00 ncopa: docs should depend on "man" 2015-07-20 14:38:13 hum 2015-07-20 14:38:13 ues 2015-07-20 14:38:40 as long as we can still pick and choose which docs to install ... :) 2015-07-20 14:38:59 ScrumpyJack: nothing changed if you dont install the "docs" package 2015-07-20 14:39:01 ScrumpyJack: no, it pulls in all -docs 2015-07-20 14:39:06 you still get the clean version 2015-07-20 14:39:09 yes 2015-07-20 14:39:23 only if you install the docs package you get all doc packages for all installed packages 2015-07-20 14:40:40 what about the man pages that aren't -docs ? 2015-07-20 14:40:56 like man-pages 2015-07-20 14:41:05 not pulled in 2015-07-20 14:41:16 you'd need apk add man-pages for those 2015-07-20 14:43:10 is it patern matching? 2015-07-20 14:43:23 no 2015-07-20 14:43:26 change in abuild 2015-07-20 14:43:36 that generates the -doc subpackages 2015-07-20 14:43:47 i just added install_if 2015-07-20 14:44:23 ncopa: could you move the following packages from testing to main http://sprunge.us/NITF 2015-07-20 14:45:09 would it catch things like py-paramiko-docs ? 2015-07-20 14:45:33 ah i get it 2015-07-20 14:45:54 pnutzh4x0r: abcde has no maintainer 2015-07-20 14:46:07 same with acpi 2015-07-20 14:47:23 pnutzh4x0r: figlet moved 2015-07-20 14:51:37 ncopa: i can take any that have no maintainer 2015-07-20 14:52:09 affects vairous of the py- too 2015-07-20 14:52:56 pnutzh4x0r: can you please create patch? you can git format-patch --stdout origin/master | sprunge 2015-07-20 14:53:10 also 2015-07-20 14:53:24 why is pyflakes not named py-flakes? 2015-07-20 14:53:32 just for the maintainership? or for the move to? 2015-07-20 14:53:39 pnutzh4x0r: both 2015-07-20 14:53:56 ok, i ll do that in a bit... i ahve a meeting soon 2015-07-20 14:53:57 also verify that nothing moved to main depends on anything in testing 2015-07-20 14:54:06 send me the patch to morrow 2015-07-20 14:54:12 no problem 2015-07-20 16:32:27 good evening :) 2015-07-20 16:32:35 ncopa: this docs package is awesome :) 2015-07-20 16:32:37 i love it 2015-07-20 17:33:45 leo-unglaub - apk-post-messages will cat /usr/share/doc/$pkg/README.alpine if it exists (these should contain helpful post install info) 2015-07-20 17:34:19 BitL0G1c: good to know 2015-07-20 17:34:26 but why are you writing it to me? 2015-07-20 17:36:04 just saw your comment about the other doc pkg 2015-07-20 17:36:34 BitL0G1c: yes, but that doc package has nothing todo with the post messages 2015-07-20 17:36:45 thats why you confused me 2015-07-20 17:36:46 :) 2015-07-20 17:37:54 apk-post-messages is a trigger for pkg-doc 2015-07-20 17:39:17 yeah, but i still dont see the connection to the docs package 2015-07-20 17:39:42 the docs package is a new package from today that if in stalled pulls down all doc packages for all installed packages 2015-07-20 17:41:14 it uses install_if statements do to a change in abuild 2015-07-20 17:41:23 ncopa did that today after i asked him for it 2015-07-20 22:32:40 if u dont run edge, u dont have to deal w/ the udev to eudev switch 2015-07-21 04:47:01 ncopa: here are the patches for moving the packages from testing to main http://sprunge.us/eSSf 2015-07-21 04:47:35 looks like some of them are out-of-date... i will send in update patches once they are moved to main 2015-07-21 04:47:58 anyway, im going to bed... let me know if anything needs fixing 2015-07-21 04:55:51 oops, i missed a package 2015-07-21 04:55:56 here is the updated patch: py-rsa 3.1.4 3.1.4 False 2015-07-21 04:56:08 http://sprunge.us/aKHg 2015-07-21 05:45:21 regarding the package split, i propose splitting main into 3 sections 2015-07-21 05:46:13 core - packages which are used only for building and booting the OS. build-base, *vanilla* kernel, openrc, syslinux, busybox, apk-tools 2015-07-21 05:47:06 main - standard features of alpine, which have a support commitment for the entire lifecycle of a distribution. mainly mature stuff, alpine-desktop packages, firefox esr, etc. 2015-07-21 05:48:04 community - everything else with the guarantee that it will receive support for at least the current period, no long-term support, security updates are best effort after the next point release, lower QA requirements 2015-07-21 05:48:29 also low barrier to contribution 2015-07-21 05:48:46 then staging (or perhaps, 'experimental') for experimental packages 2015-07-21 05:48:59 then orphaned/graveyard/unmaintained/whatever for dead packages 2015-07-21 05:49:23 splitting core from main is mainly to make it simpler to bring alpine up on new boards/archs 2015-07-21 05:49:45 that's why grsec doesn't live in core, but in main, because grsec can get in the way of porting sometimes :) 2015-07-21 05:57:06 pretty much what we proposed, except core 2015-07-21 05:58:27 sounds good, mimics BSD ports, not sure about "main" name though 2015-07-21 05:59:03 you're native so you know better, but "main" and "core" sound pretty similar to me 2015-07-21 06:13:57 good morning! 2015-07-21 06:18:50 morning 2015-07-21 06:18:54 hi 2015-07-21 06:19:02 do we have a metagroup of web servers? 2015-07-21 06:19:04 like http? 2015-07-21 06:19:06 or web? 2015-07-21 06:19:11 *for 2015-07-21 06:19:16 metagroup? 2015-07-21 06:20:27 umh 2015-07-21 06:20:38 a group who has inside users of webservers 2015-07-21 06:20:56 maybe metagroup is not the correct sentence :) 2015-07-21 06:21:09 i think we have www-data 2015-07-21 06:21:25 www-data works for nginx, lighttpd etc? 2015-07-21 06:21:26 ok 2015-07-21 06:21:32 going to check 2015-07-21 06:21:36 addgroup lighttpd www-data 2>/dev/null 2015-07-21 06:21:41 barthalion: well, core being a group of packages to bootstrap the distribution :) 2015-07-21 06:21:56 it could be called boot or bootstrap or whatever 2015-07-21 06:22:27 the idea being, if you are porting to a new arch/board (say openrisc, for example) 2015-07-21 06:22:28 fcolista: grep www-data */*.pre-install 2015-07-21 06:22:34 seems like apache2 is missing 2015-07-21 06:22:35 ncopa, yes, got it 2015-07-21 06:22:43 then you can build the stuff in core/bootstrap/whatever 2015-07-21 06:22:46 apache2 uses www-data by default iitc 2015-07-21 06:22:48 *iirc 2015-07-21 06:22:52 ACTION checks 2015-07-21 06:22:55 and from there you have a system image you can load onto real hardware 2015-07-21 06:23:12 and use that to build the rest of the distribution 2015-07-21 06:23:25 kaniini: yeah, I'm just nitpicking 2015-07-21 06:23:32 apache2 does not use www-data 2015-07-21 06:23:45 fcolista: thats a bug, should be fixed 2015-07-21 06:23:52 going to do it 2015-07-21 06:23:59 all web server user accounts should be in www-data group 2015-07-21 06:24:04 apache2.pre-install 2015-07-21 06:24:08 addgroup apache www-data 2>/dev/null 2015-07-21 06:24:34 kaniini: i understand what you mean 2015-07-21 06:24:40 with a 'core' repo 2015-07-21 06:24:44 to simplify bootstrap 2015-07-21 06:25:02 it could have like abuild and all its dependencies 2015-07-21 06:25:16 the bare minimum to build a selfhosting system 2015-07-21 06:25:30 right, basically core would be what you get in alpine-mini today plus build-base, (perhaps we could call it alpine-core, even) 2015-07-21 06:25:52 also, can we have a virtual for all -dev packages too? 2015-07-21 06:25:54 :) 2015-07-21 06:26:24 do we need that? 2015-07-21 06:26:26 hm 2015-07-21 06:26:32 it might be nice for the enduser 2015-07-21 06:26:39 ncopa: i have created two patches for you and created tickets with them 2015-07-21 06:26:39 who wants gcc etc to "just work" 2015-07-21 06:26:44 i hope the patches are correct 2015-07-21 06:27:05 kaniini: re 'core' repo 2015-07-21 06:27:06 yea, virtual dev makes sense 2015-07-21 06:27:20 to prevent debian-like rage 2015-07-21 06:27:46 it would also mean that all end users would need to have both 'core' and 'main' in their repository list 2015-07-21 06:27:55 and many will probably also use 'community' 2015-07-21 06:28:08 ncopa: core and main would be the default 2015-07-21 06:28:19 community could be enabled out of the box for edge 2015-07-21 06:28:19 ncopa: i don't think that is a huge problem 2015-07-21 06:28:26 ncopa: i suggest, perhaps, 'core' and 'base' 2015-07-21 06:28:51 ncopa: http://sprunge.us/FbHE 2015-07-21 06:29:11 maybe i can remove whitespaces 2015-07-21 06:29:29 fcolista: i think the script should also create the 'apache' user 2015-07-21 06:29:40 in case it does not exist already 2015-07-21 06:29:58 ok 2015-07-21 06:31:59 i would like to wait a bit with the devs meta package for -dev 2015-07-21 06:32:03 sure 2015-07-21 06:32:08 see how -doc plays out 2015-07-21 06:32:17 -doc works awesome :) 2015-07-21 06:32:27 atm yes 2015-07-21 06:32:43 i want see how things goes when we have all packages using it 2015-07-21 06:32:46 and when you upgrade 2015-07-21 06:32:47 etc 2015-07-21 06:32:51 sure :) 2015-07-21 06:32:58 the dependency graph becomes a lot moe complicated 2015-07-21 06:33:05 i just bumped some about 100 packages locally 2015-07-21 06:33:10 and it worked very well last night 2015-07-21 06:33:15 good 2015-07-21 06:33:17 reminds me 2015-07-21 06:33:35 mdocml needs a feature 2015-07-21 06:33:48 currently we recreate the sqlite db from trigger 2015-07-21 06:34:03 because there are no functionality to purge entries 2015-07-21 06:34:32 in the apropos db 2015-07-21 06:34:43 ncopa: http://sprunge.us/WKUg 2015-07-21 06:35:04 fcolista: looks good 2015-07-21 06:35:05 thanks 2015-07-21 06:35:32 we should probably do the same with mini_httpd 2015-07-21 06:35:58 I'm going to check 2015-07-21 06:45:49 ncopa: could you please merge #4455, it is kind of urgend because there is a bugfix in there that is causing a problem 2015-07-21 06:47:35 leo-unglaub: please send patches to aports ml, its easier for us to apply them. 2015-07-21 06:47:51 clandmeter: i would, but i still have the smtp tls issue 2015-07-21 06:48:05 sprunge is also ok for me 2015-07-21 06:48:19 git format-patch --stdout origin/master | sprunge 2015-07-21 06:48:22 works too 2015-07-21 06:48:38 we can try to ofload ncopa as much as possible :) 2015-07-21 06:48:43 +f 2015-07-21 06:49:01 i really have to dig into that tls issue 2015-07-21 06:49:17 because only with that git send thing it does not work, every other client works 2015-07-21 08:01:00 morning 2015-07-21 08:01:34 ncopa: could you apply http://sprunge.us/XaNH and put rawtherapee into main, unless you have other recommendations of course 2015-07-21 08:05:05 ScrumpyJack: i think i did that yesterday 2015-07-21 08:05:07 http://git.alpinelinux.org/cgit/aports/commit/?id=f67ab95801a57702024c80c8658a961ef1584733 2015-07-21 09:12:20 ncopa: oh cool! thanks 2015-07-21 10:32:13 when i try to create an ISO with a custom apkovl file, i get 2015-07-21 10:32:22 /path/to/apkovl.tar.gz: Scheme missing 2015-07-21 10:33:00 anything more specific? 2015-07-21 10:39:05 ok, perhaps is was not gzipped, just tarred 2015-07-21 10:40:51 to start again, i rm -rf isotmp 2015-07-21 10:41:33 this deletes all the apk that where downloaded. is there a better way, so that i don't have to download them again? 2015-07-21 10:51:54 ScrumpyJack: try hostname.apkovl.tar.gz? 2015-07-21 10:55:26 clandmeter: was about to try that, and got errors about modules not found locally. probably due to an apk upgrade. 2015-07-21 10:55:49 I rebooted, but eudev has renamed my network interfaces 2015-07-21 10:57:39 ping barthalion 2015-07-21 10:57:56 were you talking sth about mate-control-center? 2015-07-21 10:59:16 http://sprunge.us/GSCR 2015-07-21 11:00:02 what's the point of all these conflicts? it shouldn't install them in the first place 2015-07-21 11:00:40 you should pass --disable-update-mimedb to configure 2015-07-21 11:01:19 while bluefish should have '--disable-update-databases --disable-xml-catalog-update' there 2015-07-21 11:01:38 ScrumpyJack, alacerda: ↑ 2015-07-21 11:01:48 gotchya 2015-07-21 11:02:07 barthalion, nice ;) 2015-07-21 11:02:10 gonna fix 2015-07-21 11:02:14 thx 2015-07-21 11:10:12 depmod: WARNING: could not open /part/to/isotmp/tmp.initfs.grsec/lib/modules/3.18.18-2-grsec/modules.order: No such file or directory 2015-07-21 11:10:56 I get this error during initramfs creating the grsec 2015-07-21 11:11:04 can i ignore it? 2015-07-21 11:37:59 is there any policy regarding lua libraries and lua versions? minimal 5.1 or both 5.1 and 5.2? 2015-07-21 11:38:21 i even see some have 5.[123] 2015-07-21 11:41:47 fabled: is it possible to let apk search show repository? 2015-07-21 11:42:05 anything is possible ;) 2015-07-21 11:42:11 "apk policy" shows it at the moment 2015-07-21 11:43:28 wouldnt it make sence to prepend it, atleast if its pinned? 2015-07-21 11:46:06 what you suggest? can you give a mockup print what you'd like it to look like? 2015-07-21 11:48:07 actually append it like pkgname@repo (without version) 2015-07-21 11:48:42 hmm 2015-07-21 11:49:59 ncopa, do we have any scripts using 'apk search' output ? 2015-07-21 11:50:37 i think alpine-iso uses it 2015-07-21 11:51:40 i guess it uses -q 2015-07-21 11:52:24 maybe regular search could do pkgname@repo-version. 2015-07-21 11:54:29 vkris, it's you related to #4444 ? 2015-07-21 11:54:53 ncopa, in a iso.conf.mk, how does the file referred by APKOVL get pulled in? 2015-07-21 11:57:31 if i use an absolute path, or file://path i get Unsupported scheme 2015-07-21 11:58:05 does it only pull with http or ftp 2015-07-21 12:01:49 looks like it currently only pull via wget yes 2015-07-21 12:02:01 $(APKOVL_STAMP): 2015-07-21 12:02:01 @if [ "x$(APKOVL)" != "x" ]; then \ 2015-07-21 12:02:01 (cd $(ISO_DIR); wget $(APKOVL)); \ 2015-07-21 12:02:01 fi 2015-07-21 12:02:01 @touch $@ 2015-07-21 12:02:23 shoudl be trivial to fix it 2015-07-21 12:04:57 ScrumpyJack: something like: http://sprunge.us/jKXR 2015-07-21 12:52:25 sigh, anyone running dovecot with single passwd-file for userdb and passdb? 2015-07-21 12:53:07 looks like dovecot completely ignores userdb's default_fields 2015-07-21 12:53:27 so login works if I have all fields specified, but doesn't if I leave just username and password 2015-07-21 13:03:36 ncopa: or perhaps http://sprunge.us/cGFc 2015-07-21 13:14:33 somethign like that 2015-07-21 13:14:37 or maybe even 2015-07-21 13:15:00 if [ -e "$(APKOVL)" ]; then cp ....; else wget ... ; fi 2015-07-21 13:18:01 barthalion, mate-control-center fixed 2015-07-21 13:46:49 booting a custom iso, i get st.ilet.to/boot.png 2015-07-21 13:48:26 the only error i got when building was http://sprunge.us/YFFc 2015-07-21 14:57:02 i've packaged wt 2015-07-21 14:57:07 which fixes #4444 2015-07-21 14:57:09 it's a monster 2015-07-21 14:57:58 wt or c++? 2015-07-21 14:58:01 :) 2015-07-21 14:58:03 wt 2015-07-21 14:58:05 :) 2015-07-21 17:43:29 fcolista: thanks for wt 2015-07-21 17:55:48 if it becomes maintenance issue would request to push it to unmaintained 2015-07-21 21:23:20 solus uses yaml for their package format: https://wiki.solus-project.com/Package.yml 2015-07-21 21:23:47 could be interesting to look at if alpine does change the APKBUILD format 2015-07-22 06:21:28 pnutzh4x0r, there's a reason why pyflakes is called like that rather than py-flakes ? 2015-07-22 06:23:08 looks that rename pyflakes to py-flakes means change the dependency on main/py-flakes8 and testing/py-setuptools-flakes 2015-07-22 06:23:23 If you agree i'm going to rename it 2015-07-22 06:26:20 ncopa, do you agree with that? ^^^^^ 2015-07-22 06:27:12 pyflakes is not a library 2015-07-22 06:27:18 it's a tool 2015-07-22 06:27:35 we don't add py- prefixes for executable scripts 2015-07-22 06:30:25 barthalion, is this a standard? 2015-07-22 06:30:37 morning 2015-07-22 06:30:41 fcolista: yes 2015-07-22 06:31:01 looking at it, there is a program called "pyflakes" 2015-07-22 06:31:13 yes 2015-07-22 06:31:15 so naming the package pyflakes is correct 2015-07-22 06:31:20 ok, good to know 2015-07-22 06:31:53 same with any other python app 2015-07-22 06:32:01 for example, mercurial is a python program 2015-07-22 06:32:08 we dont call it py-mercurial 2015-07-22 06:32:27 however 2015-07-22 06:32:34 what about django? 2015-07-22 06:32:45 Or you mean the executable name ? 2015-07-22 06:32:58 django is a library… 2015-07-22 06:33:10 if "flakes" would have been a library, and pyflakes would have been a python binding to the "flakes" library, then we'd call it py-flakes 2015-07-22 06:33:17 django is a framework 2015-07-22 06:33:47 hum 2015-07-22 06:34:01 there is a usr/bin/django-admin 2015-07-22 06:34:09 yes 2015-07-22 06:34:18 django is a fw 2015-07-22 06:34:22 but contains executables 2015-07-22 06:34:35 i'd say py-django is a greyzone 2015-07-22 06:34:40 it's not a greyzone 2015-07-22 06:34:43 it's a library 2015-07-22 06:34:56 just because it provides convenience tools, it doesn't make it a standalone tool 2015-07-22 06:35:04 right 2015-07-22 06:35:33 it is a set of librarys 2015-07-22 06:35:40 makes more sense calling it py-django 2015-07-22 06:35:58 also because its heavily python oriented 2015-07-22 06:36:02 i eamn 2015-07-22 06:36:04 i mean 2015-07-22 06:36:13 py-flakes8 then should be pyflakes8 2015-07-22 06:36:19 *pyflake8 2015-07-22 06:36:30 good morning 2015-07-22 06:37:01 fcolista: might make sense yes 2015-07-22 07:29:14 ncopa: http://sprunge.us/cGFc was just for an error message to help the user 2015-07-22 07:31:17 ScrumpyJack: it also looks like it requires absolute file path 2015-07-22 07:43:31 yes 2015-07-22 07:46:33 there is a new binutils release 2015-07-22 08:38:54 ncopa: can you rebuild cadaver against the new openssl? currently broken. I tested a local rebuild and it works 2015-07-22 08:53:51 openssl? 2015-07-22 08:54:00 openssl should be abi compat 2015-07-22 08:56:52 what is the exact error message you get? 2015-07-22 08:57:47 i pushed a rebuild 2015-07-22 09:09:18 The error was "SSL handshake failed: SSL disabled due to library version mismatch" 2015-07-22 09:09:28 It worked after a rebuild 2015-07-22 09:22:00 ncopa: some apps have internal checks 2015-07-22 09:22:12 and openssl sometimes break compat without bumping soname… 2015-07-22 09:22:20 no idea if fabled is fixing it 2015-07-22 09:22:45 ah 2015-07-22 09:22:53 stupid 2015-07-22 09:23:04 i remeber freeradius had that issue 2015-07-22 09:23:12 and iirc it was fixed in freeradius 2015-07-22 09:58:16 just poking around at the isos. what's the motivation for putting freeswitch dahdi-linux kamailio espeak quagga and the like in the base iso? 2015-07-22 10:00:36 ScrumpyJack, I guess they are very needed packages for building a sip router or L3 router 2015-07-22 10:01:03 dadhi is needed for accurate timing 2015-07-22 10:01:15 with freeswitch/asterisk 2015-07-22 10:02:25 also t1/e1 cards 2015-07-22 10:02:46 but doesn't that express a bias towards a particular Alpine Linux build? What the motivation for that bias? 2015-07-22 10:03:46 iso contains all packages for http://alpinelinux.org/about/ 2015-07-22 10:03:58 ScrumpyJack: i dont really understand how, the alpine non-mini is supposed to be a basic server distribution 2015-07-22 10:06:50 I don't think there should be an alpine non-mini 2015-07-22 10:07:04 and the alpine-mini should become the alpine 2015-07-22 10:08:21 feel free to send your rationale to alpine-devel :) 2015-07-22 10:08:58 for the record i agree with you, i think even people using it with run-from-ram would be better off setting up an cache repo 2015-07-22 10:09:12 so they can be sure to get the security updates 2015-07-22 10:09:36 and probably the lack of properly setting up a cache repo is harming the ability to get/keep security updates 2015-07-22 10:09:51 including the packages on the disc makes it tempting to skip setting up a repo 2015-07-22 10:10:05 choices were made in the packages for alpine-standard, which shows a bias. for me, Alpine is all about small base on which I can apk add what i want 2015-07-22 10:10:49 yes, but alpine isn't just you, and the overall strategy of the product has changed over time 2015-07-22 10:10:54 ;) 2015-07-22 10:11:14 i mean i'm not saying you're wrong, and i agree that from the direction we're approaching things now, it is probably a dinosaur that should go away 2015-07-22 10:11:39 soon we will have a alpine-systemd.iso ;-) 2015-07-22 10:11:49 over my commit access 2015-07-22 10:12:28 s/for me/IMHO/ 2015-07-22 10:12:48 clandmeter: lol 2015-07-22 10:13:23 systemd would have to be a lot different than it is now 2015-07-22 10:13:28 for that to ever happen 2015-07-22 10:13:35 kaniini: how is mips going? 2015-07-22 10:13:46 clandmeter: well maybe the new binutils will fix the problem 2015-07-22 10:13:59 does wayland have systemd deps? 2015-07-22 10:15:39 doesn't seem to 2015-07-22 10:15:54 (sorry, just rambling) 2015-07-22 11:03:53 ScrumpyJack, are u facing problems with eudev in your AL desktop? 2015-07-22 11:04:46 kaniini: how is the raid tool project going? 2015-07-22 11:05:30 ScrumpyJack: as kaniini said, strategy has changed a bit. alpine iso is old 2015-07-22 11:06:11 kaniini, how's your raid tool going? 2015-07-22 11:09:58 oh, looks that i made the same question of ncopa :) 2015-07-22 11:39:43 ncopa: i'm excited about that 2015-07-22 11:40:13 alacerda: yes, eudev renamed my NIC card. not a big deal, just messy 2015-07-22 11:40:47 i dont know how to deal with NIC renames properly 2015-07-22 11:40:59 i can see no perfect solution 2015-07-22 11:41:01 rules.d i think 2015-07-22 11:42:17 ScrumpyJack, maybe i am doing sth wrong then... now setup-box doesnt work anymore here on my lab 2015-07-22 11:42:28 even if I apk add eudev && apk del udev 2015-07-22 11:43:38 i haven't tried it since eudev 2015-07-22 11:44:54 ncopa: cant you copy the logic from eudev to mdev? or doesnt mdev decide devnames? 2015-07-22 11:46:10 mdev does not touch interface names 2015-07-22 11:46:17 it only manages /dev/* 2015-07-22 11:46:28 and optionally $MODALIAS 2015-07-22 11:46:54 so its the kernel that handes it 2015-07-22 11:46:59 handles* 2015-07-22 11:48:01 eudev just overides it 2015-07-22 11:49:48 I also think mini iso should be "default" one 2015-07-22 11:50:59 clandmeter: yes, eudev will rename the interfaces when they are {hot,cold}plugged 2015-07-22 12:19:56 over on #gentoo-udev: "eudev is basically udev", which means that it should honor udev rules 2015-07-22 12:20:52 so adding a rules to the eudev package that persistantly preserves the NIC dev name should work 2015-07-22 12:40:10 ScrumpyJack, when you have the chance to test the "after-eudev" setup-box, please, let me know the result 2015-07-22 12:42:46 ScrumpyJack: the problem is the udev rules that preserves the NIC name 2015-07-22 12:42:59 it means you would have to create a state 2015-07-22 12:43:10 a state file or state database 2015-07-22 12:43:27 with the current hwaddr -> NIC mapping 2015-07-22 12:43:40 make sure that this database is preserved for reboot 2015-07-22 12:43:48 that is what udev did earlier too 2015-07-22 12:44:20 but i dont think that is idea either 2015-07-22 12:45:29 how the nics are ordered first time is also a bit unpredictable 2015-07-22 12:46:20 a preserver nic name udev rule means: restore how NICs was named last time 2015-07-22 12:48:22 i don't think you need to bind the current hwaddr to NIC in a rule. I think you can just tell eudev to use eth[0-9] but i could be wrong 2015-07-22 12:48:49 just checking http://reactivated.net/writing_udev_rules.html 2015-07-22 12:52:24 i would have preferred that kernel did it all 2015-07-22 12:53:52 i am working on build logs 2015-07-22 12:54:06 do we need historical build logs? 2015-07-22 12:54:27 the idea is that after a package is built, it should copy the build log to a logfile server 2015-07-22 12:54:51 so error messages could include a working url to the build log 2015-07-22 12:55:01 do we want keep older buildlogs? 2015-07-22 12:57:13 the way I inted to do this is via plugins.d/*.lua 2015-07-22 12:57:35 each plugin can have a prebuild() and postbuild() function 2015-07-22 12:58:04 from there we can make it post build status messages over mqtt 2015-07-22 12:58:15 and copy/clean build logs 2015-07-22 12:59:18 the question is, do we want $repo/$pkgname/$pkgname-$pkgver-r$pkgrel.log 2015-07-22 12:59:29 or is $repo/$pkgname.log enough? 2015-07-22 12:59:47 i think the latter is enough, specially if we copy it to other location once its built 2015-07-22 13:02:53 on the topic of eudev, to stop eudev renaming net devices, it should be enough to create an empty /etc/udev/rules.d/80-net-setup-link.rules file 2015-07-22 13:03:15 ScrumpyJack: thats what barthalion said too 2015-07-22 13:03:19 and it works indeed 2015-07-22 13:03:29 i suggested to ship that file in separate package 2015-07-22 13:03:46 for anyone who wants to have his pastings hosted on alpine infra, we now have tpaste.us 2015-07-22 13:03:53 so you could do: apk add eudev-weird-interface-names 2015-07-22 13:04:03 to get the weird names 2015-07-22 13:04:17 clandmeter: nice! 2015-07-22 13:05:47 ncopa: if you ship that file (an empty /etc/udev/rules.d/80-net-setup-link.rules file) in a package, it will disable eudev renaming, not enable it 2015-07-22 13:06:55 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames/ 2015-07-22 13:07:15 clandmeter: is tpaste.us yours? 2015-07-22 13:07:47 http://github.com/clandmeter/turbo-paste cool! 2015-07-22 13:08:13 :) 2015-07-22 13:08:30 turbo-paste is awesome :) 2015-07-22 13:09:13 i want one! :) 2015-07-22 13:09:24 ScrumpyJack: I meant to ship a package with the /lib/udev/rules.d/80-net-setup-link.rules 2015-07-22 13:09:42 so you dont have the renaming rules there in first place 2015-07-22 13:09:54 ncopa: yay! 2015-07-22 13:09:56 unless you explicitly add them 2015-07-22 13:10:25 the thinking is: why do you need *add* a file to *remove* a feature 2015-07-22 13:10:37 when you can just not add the feature in the first place :) 2015-07-22 13:11:33 barthalion: we have some sort of alpine paste now, so i hope you can sleep again. 2015-07-22 13:11:49 mini_httpd run with "nobody" user. ncopa, do you want me to add "nobody" to www-data as well? 2015-07-22 13:12:19 hu? 2015-07-22 13:12:36 that sounds like a bad idea. 2015-07-22 13:12:53 for me too. It's better to create minihttpd user 2015-07-22 13:12:57 and add it to www-data 2015-07-22 13:13:07 dunno "nobody" from who is used 2015-07-22 13:14:52 fcolista: do not add "nobody" to any group 2015-07-22 13:15:04 create mini_httpd or minihttpd user 2015-07-22 13:16:02 http://sprunge.us/FRSO 2015-07-22 13:19:17 ACTION slaps fcolista with tpaste.us 2015-07-22 13:20:26 clandmeter, sounds better? http://tpaste.us/AQr3 2015-07-22 13:20:34 :D 2015-07-22 13:20:38 \o/ 2015-07-22 13:20:38 :D 2015-07-22 13:21:00 $ curl --silent http://sprunge.us/FRSO | 2015-07-22 13:21:01 tpaste 2015-07-22 13:21:01 http://tpaste.us/AogA 2015-07-22 13:21:54 :) 2015-07-22 13:22:04 lloks good to me fcolista 2015-07-22 13:22:08 thanks for cleaning it up 2015-07-22 13:22:11 np 2015-07-22 13:22:21 just wondering if i should chown the log dir 2015-07-22 13:22:55 dunno if its needed 2015-07-22 13:23:04 is the current logdir owned by "nobody"? 2015-07-22 13:23:21 I'm firing up edge vm for chcking this 2015-07-22 13:23:28 it could be mini_httpd open logfile before dropping privs 2015-07-22 14:43:34 how to deal with bit suid set? 2015-07-22 14:43:35 >>> ERROR: thttpd*: Found executable files with SUID bit set: 2015-07-22 14:43:35 /home/fcolista/aports/testing/thttpd/pkg/thttpd/usr/sbin/makeweb 2015-07-22 14:43:56 iirc there were a thread on ML about this 2015-07-22 16:09:19 fcolista: you could remove it and if needed set it after pkg install via post command/script ? 2015-07-22 16:09:28 just an idea 2015-07-22 19:07:21 clandmeter: tpaste.us? 2015-07-22 19:07:35 clandmeter: it improves my life a lot, thanks man 2015-07-22 19:08:37 ncopa: I'll be resting in Spain for some time, I'll be back in two weeks 2015-07-22 19:11:37 clandmeter: yay lua 2015-07-22 19:12:06 clandmeter: redis is kinda worring but I guess we won't lose anything important 2015-07-22 19:12:40 :D 2015-07-22 19:12:48 also be sure to use mongodb 2015-07-22 19:14:16 yeah 2015-07-22 19:14:19 WEB SCALE 2015-07-22 22:44:51 why is /bin/nologin need, wouldn't it be enough to set the shell of a user you shouldn't be able to login to /bin/false? 2015-07-22 22:48:28 i think /sbin/nologin does some extra logging that /bin/false doesn't 2015-07-22 23:01:06 I am not a C expert, but looking at the code http://git.alpinelinux.org/cgit/aports/tree/main/busybox/nologin.c it doesn't seem to do any extra logging 2015-07-22 23:04:23 They only 'advantage' I see is that it allows you to configure a custom message shown to the users who tried to login 2015-07-22 23:06:00 It was added in 1a7dacd2b946f0fce327179a2eec84c4aaeb60d3 by ncopa maybe he could explain that later? 2015-07-22 23:26:08 fcolista: not to bikeshed, but pyflake8 should probably be flake8 as the command is flake8 and the upstream project is flake8 2015-07-22 23:27:06 archlinux also has the package named flake8 2015-07-22 23:32:45 nmeum: it is basically for /etc/nologin.txt only 2015-07-22 23:33:05 nmeum: if you don't want to use it, feel free to use /bin/false 2015-07-22 23:34:08 I didn't want to criticise that I was just interested in the purpose of that program 2015-07-23 06:03:28 morning 2015-07-23 06:03:37 barthalion: enjoy vacation! 2015-07-23 06:06:10 nmeum: iirc /sbin/nologin vs /bin/false is an error message 2015-07-23 06:06:48 if something tries do do something with user account that has /bin/false, then it would just fail 2015-07-23 06:07:12 with nologin it would give a message: "this account have been disabled" 2015-07-23 06:07:18 i think that was the idea 2015-07-23 07:09:30 good morning 2015-07-23 08:09:06 ncopa: any reason lua5.3-dev doesnt pull lua5.3? 2015-07-23 08:15:42 lua5.3-dev-5.3.0-r3 depends on: 2015-07-23 08:15:42 lua5.3-libs=5.3.0-r3 2015-07-23 08:15:42 pkgconfig 2015-07-23 08:15:57 it strictly does not need lua5.3 2015-07-23 08:16:37 so this is a change in recent abuild? 2015-07-23 08:17:11 i mean, before i always trust all subpkgs are installed. 2015-07-23 08:17:14 yes 2015-07-23 08:17:30 what is the advantage? 2015-07-23 08:17:47 #3642 2015-07-23 08:17:56 and the related #4109 2015-07-23 08:18:50 the -dev package only pulls it what it actually needs 2015-07-23 08:19:13 some feedback from my usage of the "docs" package 2015-07-23 08:19:28 i have it on my home computer and my office computer and my laptop and it works fine on all of them 2015-07-23 08:19:33 no limitation of speed 2015-07-23 08:19:48 on my servers running alpine i dont have it installed and also there no difference 2015-07-23 08:19:59 good 2015-07-23 08:20:00 thanks 2015-07-23 08:20:16 the mdocml trigger is slow 2015-07-23 08:20:24 but a note from my female coworker 2015-07-23 08:20:38 she loves the change to, but she would prefer to have a woman instead of a man command 2015-07-23 08:20:45 lol 2015-07-23 08:21:13 but i think a simple bash alias will do *g* 2015-07-23 08:21:24 i think that is a good idea actually 2015-07-23 08:21:31 leo-unglaub: http://linux.die.net/man/1/ln 2015-07-23 08:21:38 a 'man' alternative 2015-07-23 08:22:09 a better, smarter manual reader: woam 2015-07-23 08:22:12 woman* 2015-07-23 08:23:29 i love the idea :) 2015-07-23 08:24:02 alpine could ship man as a plain text man viewer and woman, a colorful, more pretty manual viewer 2015-07-23 08:29:48 i think it would be for upstream mdocml http://mdocml.bsd.lv/ 2015-07-23 08:32:57 Any comments on the patches supplied here http://bugs.alpinelinux.org/issues/4457 ? 2015-07-23 08:35:53 BerndLa: thanks for working on it 2015-07-23 08:37:01 are busybox setfont/loadfont compatible with the GNU tools? 2015-07-23 08:37:20 so things does nto break if you apk add 2015-07-23 08:38:57 will it continue work if you do: apk add kbd? 2015-07-23 08:38:58 There is no gnu setfont but yeah if you install kbd now you have a problem because kbd also installs a setfont utility? How should I resolve this? 2015-07-23 08:39:15 are they compatible? 2015-07-23 08:39:26 define 'compatible' 2015-07-23 08:40:12 you can also load psf2 and raw fonts with kbds setfont but kbds setfont has more functionality 2015-07-23 08:40:56 the idea is that if you use the consolefont init.d script, and apk add kbd 2015-07-23 08:41:00 will it continue work 2015-07-23 08:41:25 if they are compatible enough for making the init.d script work, then all is good 2015-07-23 08:43:12 i suppose its just to test :) 2015-07-23 08:43:24 I didn't that test but the init.d script should work with kbd too. The problem is that you don't need to specify the .psf.gz extension in conf.d when using kbd because kbd appends that to the font name automatically 2015-07-23 08:43:53 so if you adjust the conf.d file when using kbd than it should work 2015-07-23 08:44:12 but the problem I see is that kbd installs a binary (setfont) into $PATH that is already installed by busybox, isn't that a problem? 2015-07-23 08:44:23 no 2015-07-23 08:44:41 that is a working feature 2015-07-23 08:44:54 busybox creates symlinks of non existing binaries 2015-07-23 08:45:08 so if you are not happy with the busybox tool, you can always apk add 2015-07-23 08:45:13 ah, cool 2015-07-23 08:45:18 and it will simply disable the busybox link 2015-07-23 08:45:29 if you apk del , a trigger will restore the link 2015-07-23 08:45:33 that just works 2015-07-23 08:46:17 what happens if you specify the full path with kbd setfont? including the .psf.gz extention 2015-07-23 08:46:32 will kbd try load .psf.gz.psf.gz? 2015-07-23 08:47:04 we could try set the path in init.d script 2015-07-23 08:47:14 in any case, enabling setfont in busybox is first step 2015-07-23 08:47:39 Author: Anonymous 2015-07-23 08:48:02 I suppose i can add myself as author 2015-07-23 08:48:11 yeah, do that 2015-07-23 08:48:28 if you specify the full path it should definitly work, but again: it didn't test it with kbd yet 2015-07-23 08:48:47 the extension is only added if you specify a relative path (I guess) 2015-07-23 08:49:26 I could test it if with kbd if you want me to, but then you shouldn't merge this yet ;) 2015-07-23 08:49:45 i can add the busybox patch 2015-07-23 08:50:04 and add the other patch later? 2015-07-23 08:50:46 yeah after tested 2015-07-23 08:51:19 alright 2015-07-23 08:51:40 the idea is to fix init.d script to work with both busybox and kbd setfont 2015-07-23 08:52:22 yeah I got that 2015-07-23 08:54:24 I need to go to work now. I will test it with kbd and modify the init.d script as need when I come back home 2015-07-23 08:54:34 thanks 2015-07-23 12:29:05 its just amazing how slow ubuntu is (lubuntu) 2015-07-23 12:29:21 i thought a virtual lubuntu would be a quick and easy way to get a glibc system... 2015-07-23 12:36:00 probably better to install an arch linux system 2015-07-23 12:41:33 antix is fast (& debian) 2015-07-23 12:41:53 & no systemd 2015-07-23 12:46:19 i think problem is that apt-get is slow 2015-07-23 12:46:25 its just painful 2015-07-23 12:50:16 ftp.nl.debian.org is fast 2015-07-23 12:50:50 i'm sure it is. too bad apt-get itself is slow :) 2015-07-23 12:51:28 :) 2015-07-23 13:22:26 ncopa: oh yeah, apt-get can be soooo unbeleavable slow. longest time for a dist-upgrade of a desktop systen on an SSD is 45 minutes ... 2015-07-23 13:22:30 on an ssd ... 2015-07-23 13:23:04 its simply useless... 2015-07-23 13:23:23 ncopa: why do you think i am here now after 8 years of debian? 2015-07-23 13:23:30 :) 2015-07-23 13:45:19 alone apt-get update sometimes takes aeons… 2015-07-23 13:49:21 hum 2015-07-23 13:49:46 after my eudev migration my /dev/bus/usb/*/* devices has 660 permissions instead of 664 2015-07-23 13:49:59 so my virt-manager usb redirection fails 2015-07-23 13:50:06 due to no read permission 2015-07-23 14:10:11 clandmeter: nice for tpaste.us 2015-07-23 14:10:52 does it support syntax highlighting and any limit on expiry ? 2015-07-23 14:11:20 there is main/highlight 2015-07-23 14:12:47 ah add ?hl=true, nice 2015-07-23 14:16:17 tpaste.us is snappy 2015-07-23 16:25:19 ncopa: regarding the consolefont service and kbd: kbd install setfont to /bin/setfont while busybox installs it to /usr/sbin/setfont thus busybox setfont implementation isn't replaced when installing kbd 2015-07-23 16:39:03 ncopa: Why did you decide to move kbd binaries to /bin in the first place? I don't believe that they are need before /usr is mounted 2015-07-23 18:56:00 ncopa: What's your opinion on moving the kbd files as implemented here http://bugs.alpinelinux.org/issues/4457#note-4 ? 2015-07-23 18:58:19 I am off now, please comment in the issue if you have any objections 2015-07-23 20:33:09 hello, I have just done a fresh install of alpine 3.2.2 2015-07-23 20:33:30 I installed xorg, xinit, xf86-input-*, etc. and dwm as my window manager 2015-07-23 20:33:46 but I have no input inside X11, no synaptics, no mouse, no keyboard 2015-07-23 20:34:01 could someone help me debug this? 2015-07-23 20:37:27 eudev? 2015-07-23 20:38:28 no, I was having problems with edge and eudev 2015-07-23 20:38:38 that`s why I did a clean install of v3.2.2 2015-07-23 20:38:47 everything is default 2015-07-23 20:43:15 are you sure you have udev running? 2015-07-23 20:43:22 almost sure 2015-07-23 20:43:31 how can I double-check? 2015-07-23 20:43:38 rc-status -a and ps 2015-07-23 20:44:18 oh, looks like udev isn`t there 2015-07-23 20:44:20 just mdev 2015-07-23 20:44:23 so... 2015-07-23 20:44:27 rc-update add udev ? 2015-07-23 20:44:32 sysinit 2015-07-23 20:44:42 rc-update add udev sysinit 2015-07-23 20:44:45 rc-update add udev sysinit 2015-07-23 20:44:52 got it! let me try 2015-07-23 20:45:13 ive seen it fail to start before. not sure you have hitting that issue. 2015-07-23 20:45:34 setup-xorg-base or what's it called is doing all this for you 2015-07-23 20:45:35 rc-update told me udev was already added to sysinit 2015-07-23 20:45:52 hOGVgco2VEsY: I ran setup-xorg-base 2015-07-23 20:46:05 but input isn`t working inside xorg 2015-07-23 20:46:26 usually it's the udev thing that interferes 2015-07-23 20:46:34 clandmeter: rc-status -a doesn`t show udev but rc-update add udev sysinit says udev is added to sysinit 2015-07-23 20:46:36 weird 2015-07-23 20:49:44 sudo ps | grep udev shows udev 2015-07-23 20:49:57 any idea? 2015-07-23 20:54:31 i dont know, ive had that before 2015-07-23 20:54:45 something with openrc i guess 2015-07-23 20:57:39 damn, it`s a clean install 2015-07-23 20:57:43 I don`t know what to do 2015-07-23 20:58:28 do I have to be on any groups to use keyboard, mouse, synaptics, etc? 2015-07-23 20:58:32 I`m only on wheel 2015-07-23 20:58:56 you can try setup-box 2015-07-23 20:59:09 maybe you missed something 2015-07-23 20:59:43 I don`t remembering ever running setup-box 2015-07-23 21:00:24 setup-box: command not found 2015-07-23 21:00:41 you have to install setup-box first i think 2015-07-23 21:00:41 apk add setup-box 2015-07-23 21:00:47 it's like tasksel in debian 2015-07-23 21:01:52 oh, what does that command do? 2015-07-23 21:02:04 last time I installed alpine I didn`t need it 2015-07-23 21:02:09 back in 3.1 I tink 2015-07-23 21:02:45 it just makes installing groups of packages easier 2015-07-23 21:03:32 apk didn`t find setup-box 2015-07-23 21:03:44 is it on testing or something? 2015-07-23 21:04:31 http://pkgs.alpinelinux.org/package/main/x86_64/setup-box 2015-07-23 21:04:41 im not sure its in stable yet 2015-07-23 21:04:44 probably not... 2015-07-23 21:05:18 "A tool to setup a variety of Alpine Linux Environments" 2015-07-23 21:05:23 probalby doenst work on stable due to eudev 2015-07-23 21:05:33 should this really help me? I just want dwm and it`s already installed 2015-07-23 21:06:48 did you add xf86-input-evdev? 2015-07-23 21:08:13 yes 2015-07-23 21:09:23 rc-status -a doesn`t show udev and evdev 2015-07-23 21:09:40 can you start udev? 2015-07-23 21:09:44 then startx 2015-07-23 21:09:47 but rc-update says udev is added to sysinit already when I try to add it 2015-07-23 21:10:04 start udev? like /etc/init.d/udev restart ? 2015-07-23 21:10:41 yes or just start if its not started yet 2015-07-23 21:11:08 it worked! 2015-07-23 21:11:20 /etc/init.d/udev start 2015-07-23 21:11:25 udev wasn`t started 2015-07-23 21:11:51 now just to figure why this wasn`t happening by default on boot, since it was a clean install 2015-07-23 21:11:54 and how to fix it 2015-07-23 21:12:15 rc-update doesn`t let me add udev to sysinit because it tells me it`s already there 2015-07-23 21:12:41 it starts mdev in favour of udev 2015-07-23 21:12:51 maybe try remove mdev 2015-07-23 21:12:56 but i dont think it will matter 2015-07-23 21:13:20 something starting mdev as some dep 2015-07-23 21:14:10 is system dbus running 2015-07-23 21:17:11 kaniini: rc-status -a doesn`t list dbus 2015-07-23 21:17:30 it lists mdev and mdev-mount on dynamic syslevel "needed" 2015-07-23 21:20:39 and to think I did a new install because of all the issues I was having due to the udev->eudev transaction... oh, the irony! 2015-07-23 21:20:52 but looks like we are close to solving this. any ideas? 2015-07-23 21:21:39 I could do "sudo /etc/init.d/udev start" every time I boot, but that`s pretty hackish and a tad inconcenient 2015-07-23 21:22:19 rc-update add udev sysinit 2015-07-23 21:22:23 rc-update del mdev sysinit 2015-07-23 21:22:29 rc-update del mdev-mount sysinit 2015-07-23 21:22:32 reboot 2015-07-23 21:22:34 :) 2015-07-23 21:23:34 rc-update add udev sysinit tells me it`s already added to it 2015-07-23 21:23:41 just ignore it? 2015-07-23 21:24:01 yep 2015-07-23 21:24:55 rc-update del mdev sysinit tells me mdev isn`t in runlevel sysinit 2015-07-23 21:25:02 same for rc-update del mdev-mount sysinit 2015-07-23 21:25:17 rc-status -a tells me it`s in "dynamic runlevel needed" 2015-07-23 21:25:39 rc-update del mdev needed tells me needed isn`t a valid runlevel 2015-07-23 21:25:49 so... no command worked :( 2015-07-23 21:27:31 weird 2015-07-23 21:28:34 it`s a clean install. downloaded the 3.2.2 standard iso from the website and ran setup-alpine 2015-07-23 21:28:36 that was it 2015-07-23 21:28:59 sudo apk del mdev? 2015-07-23 21:29:58 if mdev is preventing udev from loading, what if I uninstall mdev? would that have any bad consequences 2015-07-23 21:38:53 mdev is part of busybox 2015-07-23 21:39:04 oh 2015-07-23 21:39:06 any other idea? 2015-07-23 21:39:12 no 2015-07-23 21:39:17 damn :( 2015-07-23 21:39:19 create an issue 2015-07-23 21:39:24 explain what you did 2015-07-23 21:39:30 assign it to ncopa 2015-07-23 21:39:40 okay, I`ll try doing that 2015-07-23 21:39:46 but if anyone has any idea, please tell me 2015-07-23 21:39:50 i talked to him about this issue 2015-07-23 21:40:02 we need a way to reproduce it 2015-07-23 21:40:10 those issues use to take a while to be solved 2015-07-23 21:40:22 when asked on irc people usually have ideas more quickly 2015-07-23 21:40:51 it doesnt seem you have the right idea now 2015-07-23 21:41:05 so do the next best thing 2015-07-23 21:41:09 :) 2015-07-23 21:42:24 this issue is a bit weird, as it happens time to time 2015-07-23 21:43:22 what do you think is the culprit? openrc? mdev? udev? 2015-07-23 21:43:35 openrc is the problem most probably 2015-07-23 21:43:58 but it doesnt matter what i think. I cannot give you a solution. 2015-07-23 21:44:23 can we get a new init? :) suckless has a good one, based on dalias work http://git.suckless.org/sinit/ 2015-07-23 21:44:36 it`s stable, 1.0 was released 2015-07-23 21:44:43 sure, if you rewrite all of our init scripts :) 2015-07-23 21:45:07 if I could I would\ 2015-07-23 21:45:08 damn 2015-07-23 21:46:07 openrc isnt pefect, but it aint bad. 2015-07-23 21:46:53 it`s because I had another issue too and I think it might have been openrc fault`s 2015-07-23 21:46:57 try for yourself 2015-07-23 21:47:07 1) install standard 3.2.2 alpine iso 2015-07-23 21:47:20 2) setup-alpine 2015-07-23 21:47:38 3) edit /etc/apk/repositories to edge instead of v3.2 2015-07-23 21:47:52 just that breaks the boot process 2015-07-23 21:48:08 lots of errors about mdev, mdev-mount, devfs, modloop, and other stuff 2015-07-23 21:48:24 apk add eudev && apk add udev doesn`t help 2015-07-23 21:48:38 meant 2015-07-23 21:48:41 apk add eudev && apk deludev 2015-07-23 21:58:35 clandmeter, kaniin: I found something that can help 2015-07-23 21:58:38 https://bugs.alpinelinux.org/projects/alpine/repository/revisions/d70b3fb8f136dfd320f6ec4cbc129990f3a26932 2015-07-23 21:58:55 "On LXC, disable acpid, klogd, mdev-mount, mdev and rdate." 2015-07-23 21:59:11 "disable [...] mdev-mount, mdev" 2015-07-23 21:59:23 that`s what we are trying to do, right? 2015-07-23 22:00:57 should I edit /etc/init.d/mdev and /etc/init.d/mdev-mount in some way? Maybe delete them? 2015-07-23 22:04:13 another hint 2015-07-23 22:04:14 https://bugs.alpinelinux.org/projects/alpine/repository/revisions/d70b3fb8f136dfd320f6ec4cbc129990f3a26932/entry/main/busybox-initscripts/mdev.initd 2015-07-23 22:04:22 check line 13 2015-07-23 22:04:29 "ewarn "Skipping mdev as udev requested in kernel cmdline" 14" 2015-07-23 22:04:45 so mdev should really be disabled if udev is to be used, right? 2015-07-23 22:10:11 I renamed mdev to mdev2 and mdev-mount to mdev-mount2 and rebooted. This cause the EXACT SAME errors I had when I upgraded to edge. 2015-07-23 22:10:21 this may give some more hints on that error 2015-07-23 22:10:31 but doesn`t help solve my current errors... 2015-07-23 22:15:40 ksaoa: sinit does not provide actual service lifecycle management 2015-07-23 22:15:48 ksaoa: openrc would still be needed 2015-07-23 22:16:35 too bad :( I should have imagined, sinit was so few lines of code, no way it would do everything needed 2015-07-23 22:29:03 okay, another issue with openrc... 2015-07-23 22:29:12 I added wpa_supplicant to runlevel boot 2015-07-23 22:29:24 but it doesn`t load (and rc-status also doesn`t show it) 2015-07-23 22:29:41 but when I try rc-update add wpa_supplicant boot again, it says it`s already added 2015-07-23 22:29:53 it`s the same thing that happens with udev 2015-07-23 22:31:00 so now I have to manually run 2 commands after login (wpa_supplicant -D wext -i wlan0 and /etc/init.d/udev start) 2015-07-23 22:31:30 can I at least make this commands run automatically? It`s a hackish fix, but at this point it`s better than nothing 2015-07-23 22:32:35 I found s6 on the aports http://git.alpinelinux.org/cgit/aports/tree/main/s6/APKBUILD maybe later on I will try to use it 2015-07-23 22:34:35 so, is there a way to make "/etc/init.d/udev start" and "/etc/init.d/wpa_supplicant start" run automatically on boot WITHOUT using rc-update as it doesn`t work for me? 2015-07-23 22:39:50 wait, I got it! 2015-07-23 22:40:08 rc-update -u 2015-07-23 22:40:09 -u, --update Force an update of the dependency tree 2015-07-23 22:40:30 I did some commands including that one above and I think it was what fixed it 2015-07-23 22:41:15 maybe that plus DISABLING mdev and mdev-mount (I found a way!) 2015-07-23 22:41:27 what I did was to first add mdev and mdev-mount to default 2015-07-23 22:41:31 then I could del them 2015-07-23 22:41:49 I couldn`t del before because they were on a "dynamic runlevel" whatever that is 2015-07-23 22:42:00 but after adding them to default, I could disable them 2015-07-23 22:42:11 openrc is too complex to me :( 2015-07-23 22:42:21 but it`s nice to see I managed to find a solution 2015-07-23 22:42:31 thanks for clandmeter and kaniini that helped me 2015-07-23 22:42:53 when I have some time I will try to use s6 init instead 2015-07-23 22:43:01 hopefully it`s more simple 2015-07-23 22:52:35 have a good night, everyone! 2015-07-23 23:03:34 when ksaoa comes back http://it-offshore.co.uk/linux/alpine-linux/46-alpine-linux-automatic-wireless-configuration & I'm having a lot of success with runit to supervise services (just managed to get dnscrypt / unbound & tinyssh all running) 2015-07-24 06:03:18 morning 2015-07-24 06:03:47 did ksaoa use disk install or tmpfs install? maybe he didnt lbu commit before reboot? 2015-07-24 06:18:30 i wrote a tiny setup-udev script that is shipped with eudev 2015-07-24 06:19:06 it will remove mdev from sysinit runlevel 2015-07-24 06:19:15 and add udev services 2015-07-24 06:19:22 including udev-trigger 2015-07-24 06:42:43 ncopa: it sounded like the exact same issue i have had multiple times. 2015-07-24 09:51:13 hey guys :) 2015-07-24 09:51:28 just wanted to say a quick hello before the weekend :) 2015-07-24 09:54:31 good moring leo-unglaub 2015-07-24 09:54:36 morning even 2015-07-24 09:57:49 is someone workig on kernel 4.1.3? 2015-07-24 09:58:58 its lts? 2015-07-24 09:59:19 not, sure 2015-07-24 09:59:22 https://www.kernel.org/category/releases.html 2015-07-24 09:59:27 3.18 2015-07-24 09:59:49 so you only build lts? 2015-07-24 10:01:11 i htink so 2015-07-24 10:01:36 ah, i thougt only stable if grsecurity is stable as well 2015-07-24 10:01:36 and we need grsec patches of course 2015-07-24 10:01:56 clandmeter: yes, of course, but they are available, thats why i asked 2015-07-24 10:04:50 anyway, i have to go 2015-07-24 10:04:53 see you later! 2015-07-24 10:12:45 ncopa: Did you see http://bugs.alpinelinux.org/issues/4457#note-4 ? 2015-07-24 10:54:41 no, i didnt 2015-07-24 10:56:12 so 2015-07-24 10:56:37 either we require that /usr is on same partition as / 2015-07-24 10:57:08 or you will have to wait with getting the console font til after /usr is mounted 2015-07-24 10:57:14 are we ok with that? 2015-07-24 10:58:01 i suppose eudev requires that /usr is available from start anyway 2015-07-24 10:58:39 i think the latter makes more sense, i would feel uncomfortable if the availability of fonts restricts my freedom in partitioning my system. that is something i'd expect from macosx 2015-07-24 11:03:35 i suppsoe this also breaks setup-keymap script 2015-07-24 11:03:58 hum 2015-07-24 11:04:00 no it does not 2015-07-24 11:05:27 looks like openrc's consolefont are started after localmount anyway 2015-07-24 11:05:52 so you will not have your consolefont loaded til after fsck is done (if its triggered) 2015-07-24 11:05:58 i suppose that is ok 2015-07-24 11:06:53 I am not sure about the keymaps though 2015-07-24 11:07:03 if you actually depend on keymaps from kbd 2015-07-24 11:07:11 and something goes wrong early boot 2015-07-24 11:07:30 then you will not have your keymap loaded 2015-07-24 11:07:47 i think that is why keymaps was installed in /lib/kbd earlier 2015-07-24 11:08:42 BerndLa: are we ok with that users will have to use the default keyboard (english keyboard) to do early boot troubleshooing in case there are problems with /usr partition? 2015-07-24 11:10:39 ncopa: alpine doesn't rely on kbd for keymaps the initramfs and the keymap service use busyboxs loadkmap 2015-07-24 11:11:12 hum true 2015-07-24 11:11:12 thus it doesn't break the setup-keymap script and consolefont is started after localmount (as you said) 2015-07-24 11:11:41 you can add busybox keymap to initramfs if you want 2015-07-24 11:12:00 but this will be different config that kbd config 2015-07-24 11:12:01 hum 2015-07-24 11:12:42 ncopa: hm? The initramfs already uses busybox keymap? http://git.alpinelinux.org/cgit/mkinitfs/tree/initramfs-init.in#n423 2015-07-24 11:12:51 i just wonder why we should move /lib/kbd to /usr/share/ instead of fixinf busybox link 2015-07-24 11:13:34 yes we support busybox keymaps in initramfs 2015-07-24 11:14:39 I don' see why the busybox link needs to be fixed…setfont isn't needed before /usr is mounted. Thus it makes more sense to store setfont in /usr/sbin 2015-07-24 11:15:20 regarding the move from /lib/kbd to /usr/share just revert that if you don't like it but I think it makes more sense to store that data in /usr/share 2015-07-24 11:16:40 the reasoning is 2015-07-24 11:16:47 kbd has been there for years 2015-07-24 11:17:03 i have no idea what users use or does 2015-07-24 11:17:43 i mean, i have no idea if moving /lib/kbd will break things fomr someone running a whacko setup 2015-07-24 11:18:18 the busybox setfont utility on the other hand, is a new feature 2015-07-24 11:18:40 so im pretty sure nobody uses that 2015-07-24 11:18:49 I don't see how this change could brake something: The worst case is that you don't have the correct keymap set which can be easily fixed and that's only the case if you modified the initscripts manually 2015-07-24 11:18:55 *break 2015-07-24 11:19:10 i dont see how it would break things either 2015-07-24 11:19:15 its just a small risk 2015-07-24 11:20:08 it also makes it impossible to load kbd's keymaps before /usr is mounted. currently it is possible 2015-07-24 11:20:16 even if we dont do it in practice 2015-07-24 11:21:04 as I said: If you disagree with that change just revert it. All I am saying is that it does make more sense to store it in /usr/share 2015-07-24 11:21:19 well if you want a kbd keymap to be used before /usr is mounted simply copy it to /etc/keymaps 2015-07-24 11:21:21 i am thinking cost/benefit 2015-07-24 11:21:33 setup-keymap currently does that as well with binarykeymaps 2015-07-24 11:23:31 Currently you would be able to access all kbd keymaps before /usr is mounted, right? But actually you just need one so manually copying the one you need to /etc/keymaps as suggest by BerndLa does make more sense to me 2015-07-24 11:24:02 thats what we do with bb keymaps too 2015-07-24 11:24:42 and if you do have /usr on separate partition, then you probably want your / (root) small 2015-07-24 11:24:50 excactly 2015-07-24 11:25:14 *exactly 2015-07-24 11:27:48 -# Contributor: Orion 2015-07-24 11:27:48 +# Contri: Orion 2015-07-24 11:27:56 that was probably not intentional 2015-07-24 11:28:36 oops, nope 2015-07-24 11:28:37 vi fuckup 2015-07-24 11:36:20 æi wonder if we should implement the updated consolefont openrc script as a patch instead 2015-07-24 11:36:32 i suppose it does not matter 2015-07-24 11:40:40 BerndLa: do you mind if i add "Patch from Bernd Lauert" to commit message? 2015-07-24 11:42:43 no, I don't mind 2015-07-24 11:43:15 pushed thanks 2015-07-24 11:44:07 no problem 2015-07-24 12:11:03 would it be an idea to always gzip our build logs? 2015-07-24 13:30:38 ncopa, http://dl.2f30.org/releases/fortify-headers-0.7.tar.gz 2015-07-24 13:31:31 pushed 2015-07-24 13:31:32 thanks 2015-07-24 13:32:40 cool :) 2015-07-24 13:32:56 ncopa, any more issues with fortify? I have not followed this chan for a while 2015-07-24 13:33:10 um 2015-07-24 13:33:15 nothing specific 2015-07-24 13:33:22 cool so everything mostly builds now I suppose 2015-07-24 13:33:37 the gobjct-introspection have been complaining 2015-07-24 13:34:00 there was some package that had issues, but i think it was gcc5 issues 2015-07-24 13:34:05 dont remember which package it was 2015-07-24 13:34:23 ah right ok 2015-07-24 13:36:43 openwrt also uses fortify now so it will get some more testing 2015-07-24 13:37:00 nice! 2015-07-24 13:37:19 there was an issue 2015-07-24 13:37:31 but its ABI issue 2015-07-24 13:37:57 the google crome DRM extesion needs some *_chk symbols 2015-07-24 13:38:09 right 2015-07-24 13:38:14 which musl does not (yet) provide 2015-07-24 13:38:16 but that's to be provided by musl I think (not yet merged?) 2015-07-24 13:38:24 yes 2015-07-24 13:38:30 but it is fortify related 2015-07-24 13:38:32 I really like that our fortify does not impose an ABI :) 2015-07-24 13:39:06 as everything is inlined 2015-07-24 13:39:37 there is some more work to do on fortify to ensure that crashes happen only when an overflow would occur (not possible with all interfaces but some still need fixing) 2015-07-24 13:39:53 in some cases valid input would trigger a crash 2015-07-24 13:39:56 http://tpaste.us/GPe2 2015-07-24 13:40:08 those are typical fortify functions 2015-07-24 13:40:17 ah right 2015-07-24 13:40:45 fortify-headers is not the right place for these 2015-07-24 13:40:50 it should work well when musl imports them 2015-07-24 13:40:53 so i thought it might be an idea to add fortify suport directly in musl 2015-07-24 13:41:12 my understanding was that that _chk syms will be added to musl for abi compat only 2015-07-24 13:41:18 ok 2015-07-24 13:41:18 fortify-headers will continue to exist as an overlay 2015-07-24 13:41:29 makes sense 2015-07-24 13:41:33 main reason people don't want full blown fortify support in musl is compiler dependence 2015-07-24 13:41:46 it uses many gcc builtins/extensions 2015-07-24 13:43:58 the snprintf_chk() implementation in your diff is not ideal I think 2015-07-24 13:44:09 the idea is to only crash the process if an overflow would occur 2015-07-24 13:44:15 probably 2015-07-24 13:44:18 sure if you pass n > b that's probably undefined behaviour 2015-07-24 13:44:19 it was just a hack 2015-07-24 13:44:28 I am in the process of fixing fortify-headers for these kind of issues 2015-07-24 13:44:33 to make netflix work in chromium 2015-07-24 13:44:38 yeah 2015-07-24 13:44:40 it didnt work anyway 2015-07-24 13:44:54 hm.. how so? 2015-07-24 13:45:00 did it crash? 2015-07-24 13:46:14 yes 2015-07-24 13:46:24 inside the checker? 2015-07-24 13:46:31 i dont know 2015-07-24 13:46:33 ah ok 2015-07-24 13:46:49 was difficult to get a strace 2015-07-24 13:46:58 if it crashed with illegal instruction it is probably inside the _chk() funcs 2015-07-24 13:47:09 otherwise something totally random 2015-07-24 13:47:09 :P 2015-07-24 13:50:15 I will let you know how the fortify changes go, it will make things slightly less restrictive (without affecting the provided security) so it shouldn't affect alpine 2015-07-24 13:50:48 and potentially avoiding to crash programs when input is actually legal (not easy or possible to do for every interface for various reasons) 2015-07-24 13:51:33 I have some ideas on how to tackle the latter but they are not nice :( 2015-07-24 14:25:30 Is there any plan to implement a "kickstart" like way of installation ? 2015-07-24 14:35:15 coredumb: what's that supposed to mean? there is setup-alpine… 2015-07-24 14:47:56 nmeum: how do you call it from boot command line ? 2015-07-24 14:48:29 how do you call it from a system where you can't boot alpine iso? 2015-07-24 14:53:06 nmeum: with kickstart I can reinstall and configure an OS from any other distrib by just dropping a kernel and an initrd and specifying a kickstart file in the boot option 2015-07-24 14:53:12 that _very_ convenient 2015-07-24 16:21:27 hi 2015-07-26 01:43:48 right so i updated my ghc port on alpine linux to the (not quite exactly/technically) released 7.10.2, it bootstraps fine and includes a few fixes for things I hit here https://github.com/mitchty/alpine-linux-ghc-bootstrap 2015-07-26 01:44:33 so testing it out i've not hit any issues with it like i had with 7.10.1, and its pretty much self contained on bootstrapping 2015-07-26 01:44:51 pretty much git clone that with docker installed, type make, wait 2015-07-26 01:45:49 the apkbuild would just need the generated ghc-bootstrap xz tar file stored somewhere, for now i just pass it in as an env variable 2015-07-26 01:47:53 the ghc/cabal-install directories there have what i'd like to submit up for the testing repo I'd presume 2015-07-26 09:25:42 mitchty: if you want we can host that file on dev.a.o 2015-07-26 16:38:35 clandmeter: sure, i just threw it up on s3 because it was easiest, but regardless if this setup is kosher it would be nice to get it into alpine linux proper 2015-07-26 19:32:17 any recommended build environment to build packages for alpine ? 2015-07-26 19:37:18 rdenis - LXC is sufficient & works very well 2015-07-27 06:48:21 i've been working on thttpd package in order to make it www-data compliant. I've upgraded it, and did some polish of patches already included in the new version. 2015-07-27 06:48:24 http://tpaste.us/24wG 2015-07-27 06:48:43 looks that makeweb is installed with suid 2015-07-27 06:49:00 which makes abuild fails 2015-07-27 06:49:19 so i had to add the README.alpine section in the APKBUILD 2015-07-27 06:50:15 i'm going to push it 2015-07-27 06:50:23 if nobody has something in contrary :) 2015-07-27 06:50:43 especially the makeweb suid...dunno if it can be managed differently. 2015-07-27 06:58:05 fcolista: what about nginx? 2015-07-27 06:58:12 did you change anything? 2015-07-27 06:58:18 it's already done 2015-07-27 06:58:32 what was the default user before? 2015-07-27 06:58:49 clandmeter, i didn't touched it 2015-07-27 06:58:56 I found out it runs as nobody on our rev proxy 2015-07-27 06:59:02 'cause i saw that post-install had already www-data 2015-07-27 06:59:29 it's edge ? 2015-07-27 06:59:39 i dont think so 2015-07-27 06:59:46 we mostly use stable for our infra 2015-07-27 07:00:04 maybe the config was old, i dont know. 2015-07-27 07:00:29 i had to change it to nginx user and chown /usr/lib/nginx 2015-07-27 07:00:37 init and apkbuild shows that the user shoudl be nginx 2015-07-27 07:01:40 umh 2015-07-27 07:01:49 maybe is only matter of default nginx.conf 2015-07-27 07:02:03 it has #user nobody; 2015-07-27 07:02:16 it's commented out 2015-07-27 07:02:27 but is suppose is the default 2015-07-27 07:02:54 we might patch nginx.conf to have nginx user by default. 2015-07-27 07:03:12 let's see first of all if nginx on edge starts with default settings 2015-07-27 07:03:19 i think it should be default nginx 2015-07-27 07:03:49 --user=nginx 2015-07-27 07:06:52 clandmeter, what do you think about http://tpaste.us/24wG ? 2015-07-27 07:07:01 (please note the usage of tpaste ;-) 2015-07-27 07:10:37 fcolista: ;-) 2015-07-27 07:11:07 that syntaxhighlighter is a pos 2015-07-27 07:11:14 it detects perl... 2015-07-27 07:12:02 isnt www-data part of baseinstall? 2015-07-27 07:12:54 where is syntaxhighlighter? 2015-07-27 07:13:57 ?hl=true 2015-07-27 07:14:02 add that to the paste 2015-07-27 07:15:23 it's a question? I mean: i should add ?h1=true when i cat $file | tpaste ? 2015-07-27 07:15:47 no 2015-07-27 07:15:56 http://tpaste.us/24wG?hl=true 2015-07-27 07:16:02 rtfm ;-) 2015-07-27 07:16:45 fcolista: did you check http://pkgs.fedoraproject.org/cgit/thttpd.git/tree/ 2015-07-27 07:17:18 clandmeter, no. What should i check ? 2015-07-27 07:18:13 it has additional patches 2015-07-27 07:18:37 it's for a previous version 2015-07-27 07:18:43 2.25b 2015-07-27 07:18:56 i'm gonna push the 2.26 2015-07-27 07:18:58 oh ok 2015-07-27 07:19:02 sorry 2015-07-27 07:19:05 np 2015-07-27 07:19:19 also, a couple of patches does not longer apply 2015-07-27 07:19:33 'cause is already implemented in upstream 2015-07-27 07:20:08 do you think is fine the message re: makeweb ? 2015-07-27 07:20:31 my both alpine webservers have nginx running as nginx from the default alpine config 2015-07-27 07:20:41 Makefile install it as suid, and abuild fails since it found a suid binary when packages. 2015-07-27 07:20:55 jomat, thanks for the feedback. 2015-07-27 07:22:26 fcolista: i dont see that patch. 2015-07-27 07:24:16 which one? 2015-07-27 07:24:26 suid one 2015-07-27 07:24:45 i guess this is git diff 2015-07-27 07:24:46 suid patch is not pushed yet. 2015-07-27 07:24:55 so what are you asking? 2015-07-27 07:24:57 yes it's git diff 2015-07-27 07:24:59 :) 2015-07-27 07:25:23 ok i don't explain correctly myself. Reformulate. 2015-07-27 07:25:37 Makefile installs makeweb binary as suid 2015-07-27 07:25:47 this cause abuild fails. 2015-07-27 07:26:03 s 2015-07-27 07:26:05 so 2015-07-27 07:26:07 i did: 2015-07-27 07:26:08 http://tpaste.us/3M8A 2015-07-27 07:26:37 http://tpaste.us/3M8Ahl=true 2015-07-27 07:26:45 http://tpaste.us/3M8A?hl=true 2015-07-27 07:27:05 and added to APKBUILD: 2015-07-27 07:27:05 cat > $pkgdir/usr/share/doc/$pkgname/README.alpine < To use makeweb, please set suid to makeweb binary with: 2015-07-27 07:27:06 ------------------------------------------------------- 2015-07-27 07:27:06 chmod 2755 /usr/sbin/makeweb 2015-07-27 07:27:06 EOF 2015-07-27 07:27:12 that section. 2015-07-27 07:27:29 what i'm asking if: 2015-07-27 07:27:44 there's a better way to handle the suid binary ? 2015-07-27 07:28:09 there's a better way to handle suid binary when they comes from upstream in such way? 2015-07-27 07:33:13 what does it actually do? 2015-07-27 07:33:42 if it needs it, maybe better to subpkg it. 2015-07-27 07:34:37 you will need to add suid option to apkbuild 2015-07-27 07:36:12 aaah 2015-07-27 07:36:15 what's the option? 2015-07-27 07:36:49 options="suid" i think 2015-07-27 07:44:45 interesting 2015-07-27 07:46:13 options="suid" did the trick 2015-07-27 07:50:23 we try to prevent suid where possible 2015-07-27 08:02:08 there is an interessting discussion on alpine-devel ml about udev/mdev and alternatives 2015-07-27 08:02:08 so, what's your suggestion? Simply add the options "suid", or making a subpackage with makeweb ? 2015-07-27 08:02:22 With suid options? 2015-07-27 08:16:13 i wonder if it would be an idea to use both toybox and busybox 2015-07-27 08:16:35 that is, we enable those tools that toybox deso good in toybox, and use busybox for the rest 2015-07-27 08:24:25 ncopa: nah, I don't think that's a good idea. I assume toybox and busybox don't mix well 2015-07-27 08:25:01 they just implement some standard tools 2015-07-27 08:25:20 to be more specific 2015-07-27 08:25:36 there are a few things i am not too happy with in busybox 2015-07-27 08:25:55 like mdev code 2015-07-27 08:26:07 mdev in it self is mostly good 2015-07-27 08:26:19 but i think lots of the things in busybox mdev could be simplified 2015-07-27 08:26:33 and i really really want reading events from netlink 2015-07-27 08:26:41 :) 2015-07-27 08:27:27 i'd like mdev and modprobe work better together 2015-07-27 08:28:17 i think atleast depmod is mostly fableds code 2015-07-27 08:28:24 probably modprobe too 2015-07-27 08:28:44 hm.... 2015-07-27 08:29:07 other thing I have been thinking of is containers 2015-07-27 08:29:14 they often dont need some of the kernel tools 2015-07-27 08:29:17 like modprobe 2015-07-27 08:29:28 you dont load kernels from containers 2015-07-27 08:29:37 you probably dont need hotplugger either 2015-07-27 08:30:02 so if we moved those bits out of busybox, then we could make the container base even smaller 2015-07-27 08:30:16 which would make the docker ppl happier :) 2015-07-27 08:31:57 ncopa: there is currently a thread on the devel mailinglist regarding mdev, did you see it? There are a few good alternatives like smdev for instance 2015-07-27 08:32:07 yes 2015-07-27 08:32:15 that is what i have been reading this morning 2015-07-27 08:33:38 my gmail is terrrible slow 2015-07-27 08:41:35 ncopa: you mention you need udev for xorg only for hotplugging? 2015-07-27 08:41:51 yes 2015-07-27 08:41:58 i think you can use xorg without udev 2015-07-27 08:42:11 but then you need to manage all your input devices manually in xorg.conf 2015-07-27 08:47:22 https://xkcd.com/963/ 2015-07-27 08:57:28 ncopa, if I push atomically a package with pkgrel=0 who does not build, and the same package with pkgrel=1 which builds, what happens? It fails and then builds, or it builds directly? 2015-07-27 09:03:10 ? 2015-07-27 09:03:44 nmeum: lol 2015-07-27 09:03:58 nmeum: yes. i want solve that 2015-07-27 09:04:21 i am prepared to hack xorg code to get rid of libudev too 2015-07-27 09:04:27 but we need an alternative then 2015-07-27 09:16:29 ncopa: openblas does not build (patch %304) because of missing linux-headers. So i added linux-headers in makedepends and bumped pkgrel. If i push the package, does it build? Or it fails because of missing linux-headers, and then build because i've added the missing dependencies? 2015-07-27 09:18:35 you can infact add linux-headers to makedepends without bump pkgrel 2015-07-27 09:18:59 ok. 2015-07-27 09:19:00 if something does not build, you dont need bump pkgrel to push the fix 2015-07-27 09:19:10 however, if something already is built 2015-07-27 09:19:19 and you push soemthing 2015-07-27 09:19:35 then it will not try rebuild unless you bump pkgrel 2015-07-27 09:20:08 ok 2015-07-27 09:20:28 so in this case, the git-am update the git tree with the missing dependency. 2015-07-27 09:20:39 I add the dependency and commit. 2015-07-27 09:20:52 When i push, it builds. 2015-07-27 09:21:10 that's it. 2015-07-27 09:24:00 ncopa: I wouldn't mind editing xorg.conf every now and then 2015-07-27 09:27:29 i kinda do mind. when xscreensaver has locked my screen and wireless keyboard runs out of battery 2015-07-27 09:27:48 the is it nice to be able to plug in an usb keyboard... 2015-07-27 09:29:17 yeah, hotplugging of input devices is kind of useful 2015-07-27 09:35:03 +1 2015-07-27 11:19:56 hey there, i just wanted to ask if it would be possible to add the realip, dav and dav_ext modules to the nginx package 2015-07-27 11:23:36 sigh 2015-07-27 11:23:43 problem with nginx is that it is not modular 2015-07-27 11:23:57 yea, i noticed that 2015-07-27 11:24:38 we have 3 different nginx packages already 2015-07-27 11:24:47 do we have "conflicts" and "provides" in APKBUILD? 2015-07-27 11:25:05 we have provides 2015-07-27 11:25:08 but does it make that much of a difference to fill it with modules? footprint for most of the modules should be quite small, and most of them should be configurable in nginx config 2015-07-27 11:25:24 conflicts are done as: depends=!conflictingpackage 2015-07-27 11:25:55 ok. Gotta change %312 2015-07-27 11:26:08 lukas2511: you would end up with a bloated nginx binary 2015-07-27 11:26:18 where bloat is code that you dont need 2015-07-27 11:26:40 i think the proper solution is to make nginx be able to load plugins 2015-07-27 11:27:04 dynamic plugins 2015-07-27 11:27:11 which are dlopen'ed 2015-07-27 11:27:18 there is a nginx fork that does it 2015-07-27 11:27:23 tengine iirc 2015-07-27 11:28:02 i don't know... 2015-07-27 11:28:58 as long as it's not an official feature this doesn't really feel like the right thing to do 2015-07-27 11:29:19 i saw it is listed in the nginx bug tracker 2015-07-27 11:29:21 or roadmap 2015-07-27 11:29:55 i dont know what is the proper way to handle it meanwhile 2015-07-27 11:30:05 yea it's on 1.9 roadmap 2015-07-27 11:30:26 but 2.0 would be the next stable version if i remember nginx version numbering correctly 2015-07-27 11:30:36 (or 1.10) 2015-07-27 11:31:08 we probably need do something before that 2015-07-27 11:32:06 what is wrong with tengine? 2015-07-27 11:32:27 its not named 'nginx' :) 2015-07-27 11:32:27 nothing wrong with tengine, but it should be handled as tengine, and not as nginx 2015-07-27 11:33:12 i mean, i'd probably use it if there would be a package in the repo 2015-07-27 11:33:19 but i wouldn't call it "nginx" 2015-07-27 11:33:37 who cares about a name 2015-07-27 11:33:44 we have it in testing 2015-07-27 11:34:18 the patches we have for nginx work for tengine? 2015-07-27 11:34:28 i think so 2015-07-27 11:34:52 atleast the ipv6.patch apperas to be there 2015-07-27 11:34:59 its not the first time we do things different :) 2015-07-27 11:36:05 if it actually works, we should move it to main. 2015-07-27 11:36:16 i pushed 2.1.0 release to testing 2015-07-27 11:36:23 i have not tested if it works or not 2015-07-27 11:36:26 ncopa: there was also an issue with our revproxy 2015-07-27 11:36:34 ok? 2015-07-27 11:36:53 ngnix couldnt write to usr/lib/nginx 2015-07-27 11:37:02 seems it was running as nobody... 2015-07-27 11:37:08 could be it was my bad though. 2015-07-27 11:37:28 why would it want write to /usr/lib 2015-07-27 11:37:32 oh and btw. not sure if this was fixed in edge but i had a problem with abuild, it seems to query apk for dependencies, but if apk fails it doesn't print the error message and just fails without any further notice why it did so 2015-07-27 11:37:34 bigger post sizes would use tmp 2015-07-27 11:37:57 would be nice to actually see the error message there 2015-07-27 11:38:07 the perl error you got? 2015-07-27 11:38:10 yea 2015-07-27 11:38:14 i dont know why it failed 2015-07-27 11:38:37 abuild normally tells why it fails 2015-07-27 11:39:08 ncopa: there were also errors coming from redmine 2015-07-27 11:39:13 i guess also post size related. 2015-07-27 11:39:22 probably attaching log files 2015-07-27 11:39:24 or the likes 2015-07-27 11:40:03 I'm noticing that several contributors are sending patches of new aport without "maintaniner" set. 2015-07-27 11:40:09 anyways, its fixed and tpaste works with bigger pastes 2015-07-27 11:40:30 fcolista: you can put your name in it ;-) 2015-07-27 11:40:44 I can put your name also :) 2015-07-27 11:40:54 :p 2015-07-27 11:40:57 ehe 2015-07-27 11:41:04 fcolista: only main really needs it 2015-07-27 11:41:07 i suppose you can push it to testing without maintainer 2015-07-27 11:41:21 but dont move it to main unless someone wants to maintain it 2015-07-27 11:41:30 yes, that's what i'm doing. 2015-07-27 11:41:38 ok 2015-07-27 11:41:45 I'll continue in this way 2015-07-27 11:41:54 (and put clandmenter sa maintainer :) 2015-07-27 11:42:22 im planning to add a "Out of Date" feature at pkgs 2015-07-27 11:42:38 It's kinda useful on AUR 2015-07-27 11:42:50 I think we can implement it as well 2015-07-27 11:44:32 anyone anything against pkgs to send mail notifications when somebody tags "Out of date"? 2015-07-27 11:45:15 will you set a flag in the db? 2015-07-27 11:45:23 yes 2015-07-27 11:45:26 so that once its sent, we wont get it again? 2015-07-27 11:45:36 :) of course 2015-07-27 11:45:50 and once we push new version of it, the flag is automatically reset... 2015-07-27 11:46:04 mh, so weird, i can't even reproduce the problem i had 2015-07-27 11:46:48 lukas2511: could be that lib was a leftover when you canceled abuild 2015-07-27 11:47:05 yea, but that doesn't explain why i can't reproduce it 2015-07-27 11:47:10 i created that file manually 2015-07-27 11:47:21 and tried building perl, and it worked without problems 2015-07-27 11:47:29 not sure what is going on 2015-07-27 11:47:38 if i encounter it again i'll do a bit more testing 2015-07-27 11:47:58 i was just happy that i found that one line in the 564 MB of strace log so quickly :D 2015-07-27 11:48:21 it was looking for /usr/lib/libdb.so something? 2015-07-27 11:49:07 yea it was looking for /usr/lib/libdb-5.3.so or something like that, but didn't find the package owning it, and that resulted in apk failing, which resulted in abuild failing, i guess (don't have the log anymore) 2015-07-27 11:49:27 i have it 2015-07-27 11:49:55 mh? 2015-07-27 11:50:01 http://tpaste.us/GEV2 2015-07-27 11:50:17 i had that build log in an open window 2015-07-27 11:50:21 yea 2015-07-27 11:50:27 but not the 564MB strace log 2015-07-27 11:50:33 which contained the actual error message 2015-07-27 11:50:55 that is the console output (stdout and stderr) of the package building 2015-07-27 11:51:00 which didn't contain the error message 2015-07-27 11:51:47 i hope the next time i encounter this problem it's a package with a lot faster packaging step 2015-07-27 11:51:53 ah 2015-07-27 11:52:06 i thikn i know what the problem is 2015-07-27 11:52:36 $ apk info --quiet --who-owns /usr/lib/libdb-nonexisting.so 2015-07-27 11:52:36 ERROR: /usr/lib/libdb-nonexistsing.so: Could not find owner package 2015-07-27 11:52:50 yes, that was the error message i found in strace 2015-07-27 11:53:04 $ apk info --quiet --who-owns /usr/lib/libdb-nonexisting.so >/dev/null 2015-07-27 11:53:12 does not print anything 2015-07-27 11:53:22 oh, it got printed to stdout? well 2015-07-27 11:53:22 it sends error message to stdout and not to stderr 2015-07-27 11:53:28 that should be fixed 2015-07-27 11:54:33 fabled_: is it intentional that 'apk info --quiet --who-owns /nonexisting' sends error to stdout instead stderr? 2015-07-27 11:57:48 looks like it just uses the same log function but adds a "ERROR" in front of the message 2015-07-27 11:58:00 was introduced here: http://git.alpinelinux.org/cgit/apk-tools/commit/src/print.c?id=517378721855280d2e23d25d7529e6b9cbae9136 2015-07-27 11:58:54 looks like everything was logged to stderr before, and now everything is logged to stdout 2015-07-27 11:58:57 this seems wrong 2015-07-27 11:59:06 yup 2015-07-27 11:59:14 warn and err should go to stderr 2015-07-27 12:02:22 http://tpaste.us/Aal2 2015-07-27 16:39:20 ncopa - qt5-qtbase is built with -reduce-relocations but at the moment this means executables need to be built with -fPIC https://codereview.qt-project.org/#/c/111787/ - which means we lose the address layout randomization from PaX. This was an issue in Bitcoin / Arch https://github.com/bitcoin/bitcoin/pull/6248#issuecomment-110491401 but Arch have since changed to -no-reduce-relocations in qt5-base https://projects.archlinux.org/svntogit/packages.gi 2015-07-27 16:40:23 I can get bitcoin to build with -fPIC but think it is better to have the features from PaX 2015-07-28 07:32:46 in order to revert specific commit, it's enough git revert $COMMIT_ID right ? 2015-07-28 07:32:54 the git pull --rebase and git push 2015-07-28 07:32:58 *then 2015-07-28 07:33:40 git revert is an incremantal revert 2015-07-28 07:33:49 git revert $hash 2015-07-28 07:34:27 will remve the lines $hash adds, and add the lines $hash removes 2015-07-28 07:34:42 to it will show up as a new commit 2015-07-28 07:35:02 if you want revert seomthine that not yet is in the public repository (ie only in your local clone) 2015-07-28 07:35:30 then you can git reset HEAD^ to "undo" the last sommit 2015-07-28 07:35:32 commit* 2015-07-28 07:35:48 or you can use 'git rebase -i' and remove commits you want remove 2015-07-28 07:36:16 this will change the commit history so you cannot do that with things that is already pushed to the public repository 2015-07-28 07:36:39 wow 2015-07-28 07:36:49 git rebase -i is actually what i was looking for 2015-07-28 07:37:06 since i've applied ucspi-xx patches 2015-07-28 07:37:22 but the website is not available from where download the source 2015-07-28 07:37:32 in the meanwhile i've added other patches 2015-07-28 07:37:41 so ucspi-xx was "in the middle" 2015-07-28 07:37:46 git rebase -i did the trick 2015-07-28 07:37:56 just removed what i needed to remove 2015-07-28 07:38:01 and git log is clean 2015-07-28 07:43:00 would be correct to set the patch "deferred" if source file is not available 'cause the website is down? 2015-07-28 07:43:12 or is better "Awaiting Upstream"? 2015-07-28 07:43:46 i think you can just leave it as "new" 2015-07-28 07:43:58 or "assigned" 2015-07-28 07:44:22 Ok. I'm replying to the ML that we wait for the source website to be available again, or (maybe) look for another place where to download the source. 2015-07-28 07:49:50 clandmeter, ncopa: there's no "assigned" status in patchwork. Care if i add it? 2015-07-28 07:55:05 nah, if its not there, then dont add it 2015-07-28 07:55:23 ther is an "under review" that is similar 2015-07-28 07:56:12 k 2015-07-28 08:11:18 anything in contrary in applying 321 and 322? 2015-07-28 08:24:02 321 looks ok 2015-07-28 08:24:43 322 is good too 2015-07-28 08:24:47 thanks 2015-07-28 08:25:41 ok. What is in main and in critical packages is better that you crosscheck 2015-07-28 08:26:00 like 323,324. 2015-07-28 08:26:45 323 looks fine, 324 dunno. 2015-07-28 08:27:11 i have my doubt re 323 2015-07-28 08:27:37 we create the /tmp explicitly 2015-07-28 08:28:12 there must be a reason why we create /tmp in the first place 2015-07-28 08:33:19 i'm sure there's a reason... 2015-07-28 08:33:46 something in contrary in moving erlang on main? 2015-07-28 08:35:44 looking at history 2015-07-28 08:36:10 the maintainer marlus sariva seems to have taken good care of it 2015-07-28 08:36:20 i'd say we can move it to main 2015-07-28 08:36:21 yes, i agree 2015-07-28 08:36:38 just make sure that there are no depends in testing 2015-07-28 08:37:00 i think all deps are in main already 2015-07-28 08:37:06 ACK for moving it 2015-07-28 08:37:15 ok thx ncopa 2015-07-28 08:37:22 I'm building erlang now 2015-07-28 09:17:19 i'll mark 323 as rejected. 2015-07-28 13:44:32 ncopa: what's your opinion on removing the log rotation feature from busybox syslog and instead advise users to install logrotate? 2015-07-28 13:47:26 personally i think bb syslog is pretty nice 2015-07-28 13:49:16 nmeum: it is relatively simple to reconfigure it to use logrotate in stead 2015-07-28 13:50:00 you have to recompile it do that, right? At least that's what I am currently doing 2015-07-28 13:50:09 echo 'SYSLOGD_OPTS="-s 0"' > /etc/conf.d/syslog 2015-07-28 13:50:15 no, you dont need recompile 2015-07-28 13:50:18 oh 2015-07-28 13:50:26 ok forget about that suggestion then 2015-07-28 13:50:49 -s SIZE Max size (KB) before rotation (default:200KB, 0=off) 2015-07-28 13:51:04 yeah, I obviously missed that option 2015-07-28 13:51:34 there was a patch about disable klogd in syslog or similar 2015-07-28 13:52:22 "klogd should not be used together with syslog to kernel printk buffer." 2015-07-28 13:52:26 why is that? 2015-07-28 13:53:06 there are two ways to log kernel printk message, either syslog with printk buffer support or klogd at the moment alpine linux busybox uses both even though that's discouraged 2015-07-28 13:53:22 so it logs double 2015-07-28 13:53:29 ? 2015-07-28 13:54:09 no, somehow it doesn't but it's still discouraged to enable both options 2015-07-28 13:54:21 in the .config file it explicitly says 'klogd should not be used together with syslog to kernel printk buffer' 2015-07-28 13:55:07 yes. i just wonder why it is discrouaged 2015-07-28 13:55:16 how bad is it to have them both enabled 2015-07-28 13:55:26 is it so bad that we should apply the fix to v3.2 too? 2015-07-28 13:55:40 I don't know 2015-07-28 13:55:44 probably not 2015-07-28 13:55:58 *looks at source code* 2015-07-28 13:56:49 we don't enable the -K syslogd flag by default, right? 2015-07-28 13:57:06 it only logs double if you run klogd and syslogd with -K 2015-07-28 13:57:52 ah 2015-07-28 13:58:33 gotta go, see you later 2015-07-28 13:58:36 thats probably why 2015-07-28 13:59:07 looks that the new amule version (patch %306) won't build after crypto++ update. I'm going to defer the patch, i'll take a look later. 2015-07-28 14:03:05 so, klogd grabs kernel messages and sends them to syslog 2015-07-28 14:03:15 syslogd -K does the opposite 2015-07-28 14:03:36 it takes syslog messages and send them as kernel messages 2015-07-28 14:04:05 so you can get syslog via dmesg instead of a logfile 2015-07-28 14:04:14 if you enable both you create a loop :) 2015-07-28 15:27:47 fcolista: It looks like the source tarballs for ucspi-{ssl,tcp6} are available again. 2015-07-28 18:45:14 quick question, are there any objections and/or guidelines regarding packaging of software without a release version? 2015-07-28 18:46:47 I don't think that the aport repo contains software without a release version so far 2015-07-28 18:50:44 chris|: it's obviously preferred to use a tagged release, but it's not always possible 2015-07-28 18:50:58 sometimes we're forced to use a git snapshot or similar 2015-07-28 18:52:59 tdtrask: yes, that's exactly my problem. how do you handle the pkgver in those cases? 2015-07-28 19:48:38 chris|: there are several APKBUILD files with "snapshot" functions that will grab a git snapshot and upload to dev.alpinelinux.org/archive 2015-07-28 19:48:50 you must be a developer to save the snapshot 2015-07-28 19:51:10 so, we save on our infrastructure the snapshot that we use to build the package 2015-07-28 19:51:30 it's not ideal, but it works 2015-07-28 19:57:18 tdtrask: I'll look into that, thanks 2015-07-28 20:03:06 np 2015-07-29 06:45:33 re 332 and 333, Isaac wanted to delete package in unmaintanied. Anything in contrary? 2015-07-29 06:45:44 It's ok for me 2015-07-29 07:17:58 fcolista: why would you want to do that? I would keep them in unmaintained so if somebody needs any of the unmaintained package he can take over maintenance for them 2015-07-29 07:18:30 nmeum, we already have chromium 2015-07-29 07:18:44 ah 2015-07-29 07:19:02 and lua-sqlite3 alternative called lua-sql-sqlite3 2015-07-29 07:19:35 but actually, lua-sqlite3 could be kept for the reasoning you wrote 2015-07-29 07:19:55 so make sense only remove chromium and keep lua-sqlite3 2015-07-29 07:20:22 we do have 111 packages in unmaintained 2015-07-29 08:14:00 fcolista: i guess you read the ml? 2015-07-29 08:14:35 apropos of...? 2015-07-29 08:15:41 I do...are you referring to something specific, clandmeter ? 2015-07-29 08:16:37 yes, regarding the repo discussion 2015-07-29 08:18:50 Yes, i saw the discussion, but not read a definitive solution 2015-07-29 08:19:33 correct, me neither 2015-07-29 08:19:51 maybe we should wait cleaning up stuff before we are sure how we move forward 2015-07-29 08:20:10 I've just cleaned chromium 'cause we already have 2015-07-29 08:20:33 so does not make any sense to have it on unmaintained also 2015-07-29 08:20:45 but sure, i'm not going to clean other things 2015-07-29 08:21:49 yes thats fine 2015-07-29 08:21:57 im just suggesting somethhing 2015-07-29 08:22:08 i agree with you 2015-07-29 08:22:13 i think ncopa should come with a defenite conclusion 2015-07-29 08:22:27 and we should write the ideas/rules on wiki 2015-07-29 08:22:37 summarize our workflow 2015-07-29 08:22:49 +1 2015-07-29 08:54:01 i will try finish up the 3.18.19 kernel and then look at the ml re repos 2015-07-29 08:56:14 i've applied the other patch which removes lua-sqlite3, since like chromium we already have this package in testing 2015-07-29 09:44:40 good, thanks 2015-07-29 11:14:35 has 'xorg-fiasco' been fixed on edge? 2015-07-29 11:15:04 I sent two patches and they has been rejected 2015-07-29 11:15:19 I think that it is because I am on main instead of edge when abuild them 2015-07-29 11:15:43 I would like to come back to edge but I am afraid 2015-07-29 11:23:25 which patches are you refering to? 2015-07-29 11:24:02 have a link on patchwork? 2015-07-29 11:24:05 http://patchwork.alpinelinux.org/patch/306/ 2015-07-29 11:24:10 http://patchwork.alpinelinux.org/patch/307/ 2015-07-29 11:25:14 AmatCoder, how those patches are related to xorg-fiasco? 2015-07-29 11:25:54 well, here they build fine but I am on main now since xorg-fiasco 2015-07-29 11:26:20 dunno if this is the case of mednafen 2015-07-29 11:26:23 but not for amule 2015-07-29 11:26:55 also, since the new packages are in edge, in order to be sure they build you should use an edge build environment 2015-07-29 11:27:20 the amule issue i think is related to crypto++ update 2015-07-29 11:28:25 hmm I will try again 2015-07-29 11:28:58 maybe I don't get correct crypto++ 2015-07-29 11:29:22 ok, but please try in an edge environment 2015-07-29 11:31:22 that is why I am asking about xorg-fiasco 2015-07-29 11:32:27 How do I test them without xorg? 2015-07-29 11:33:16 ah now i get it 2015-07-29 11:33:23 i thought xorg-fiasco is a pkg 2015-07-29 11:33:28 haha 2015-07-29 11:33:38 ACTION too 2015-07-29 11:33:52 you can try edge 2015-07-29 11:33:54 it should be working 2015-07-29 11:34:01 mine was broken 2015-07-29 11:34:07 but seems ncopa fixed it 2015-07-29 11:34:49 ok, I will try 2015-07-29 11:34:51 ive been trying to get a headless xorg installation 2015-07-29 11:35:01 on lxc 2015-07-29 11:35:13 something with xrdp. 2015-07-29 11:35:39 so i could test x packages when im at work 2015-07-29 11:39:22 ok, I am going to reboot...croosing fingers ;) 2015-07-29 11:47:01 all ok, I am on edge now :) 2015-07-29 11:47:13 wb 2015-07-29 11:47:18 \o/ 2015-07-29 11:48:06 thanks to both 2015-07-29 11:48:39 I am going to look into those patches 2015-07-29 11:48:48 ok 2015-07-29 11:49:29 seems that only patch %324 is pending 2015-07-29 11:49:41 ncopa replied via email, though 2015-07-29 15:55:23 clandmeter: I am thinking of adding ?format= to tpaste 2015-07-29 15:55:27 any thoughts 2015-07-29 16:02:01 AL 3.2.2, is this a bug?: http://pastebin.com/EH0vZeeT 2015-07-29 16:27:32 alacerda, i think 'who' is based on utmp which is not supported in musl 2015-07-29 16:32:45 that is why on AL 2.4.11 it is working :) 2015-07-29 17:16:43 clandmeter: done, sending patch or maybe github pr, not forked yet 2015-07-29 17:30:59 clandmeter: patch for turbo-paste, http://tpaste.us/3gmx 2015-07-29 17:35:23 I think highlight2.tpl can be used for hl=? too 2015-07-30 06:05:16 good morning! 2015-07-30 07:11:33 it seems that i'm no longer able to push to git.alpinelinux.org. isn't the url git@git.alpinelinux.org:user/USER/REPO? 2015-07-30 07:12:27 hmmm, or my key is gone 2015-07-30 07:19:32 does anybody use multiple ssh keys specified in a ~/.ssh/config file with git? 2015-07-30 07:39:42 vkris: create a pr and we discuss on github. 2015-07-30 07:53:40 morning 2015-07-30 07:54:19 faffolter: we moved to gitolite not too long ago 2015-07-30 07:54:45 i might have missed your key while doing it 2015-07-30 07:56:07 faffolter: i do use multiple keys 2015-07-30 07:56:23 ncopa: it's also possible that some is wrong with my configuration 2015-07-30 07:56:34 i think the key is just missing 2015-07-30 07:56:41 you probably used password login earlier 2015-07-30 07:57:39 could you please scp your public key(s) to your git.alpinelinux.org:.ssh/authorized_keys ? 2015-07-30 07:57:58 sure 2015-07-30 08:08:03 while specifying a key i get 'Too many authentication failures' from git.alpinelinux.org 2015-07-30 08:08:43 would be a good time to check my .ssh/config 2015-07-30 08:33:26 3.18.19 kernel is building 2015-07-30 08:58:45 ncopa: yey :) 2015-07-30 08:58:49 i love new kernels :) 2015-07-30 09:00:42 ncopa: at the red cross i had to work last week with chrome books. horrible things, but two things are impressive. bootup time and the battery. the battery is nothing that alpine can change, but the bootup time is something where we could tweak some stuff 2015-07-30 09:01:03 one idea would be to not have the network manager wait for connections that are not available at the moment 2015-07-30 09:01:51 specially for laptops where the ethernet is not alway awailable it would be a huge speed boost 2015-07-30 09:05:42 using s6 i'm able to boot within 2 secs 2015-07-30 09:06:02 see my example https://github.com/stef/s6-services 2015-07-30 09:06:26 here's some readme for this https://www.ctrlc.hu/~stef/s6.txt 2015-07-30 09:28:59 bY4JHxo6GVhj: how would you compare it against runit ? 2015-07-30 09:37:58 leo-unglaub: i use dhcpcd 2015-07-30 09:38:15 which will delay networking til link is detected 2015-07-30 09:38:26 makes my lappy boot fast(er) 2015-07-30 09:39:01 s6 reimplements runit/daemontools but with some improvements, like their own non-forking script lang execline, some service is up messaging 2015-07-30 09:39:22 it's like runit-experimental 2015-07-30 09:39:24 :) 2015-07-30 09:39:29 i find execline a bit "weird" 2015-07-30 09:39:36 totally weird. 2015-07-30 09:40:13 but less weird than systemd init rules ;) 2015-07-30 09:41:05 # uname -a 2015-07-30 09:41:05 Linux localhost 3.18.19-0-grsec #1-Alpine SMP Thu Jul 30 08:25:58 GMT 2015 x86_64 Linux 2015-07-30 09:41:12 it boots in qemu atleast... 2015-07-30 09:41:27 sorry for the lowest baseline comparison possible 2015-07-30 09:45:44 :) 2015-07-30 09:52:55 is it considered stable enough to not kill your dog when booting ? 2015-07-30 09:53:25 or is it worth the hassle of rewriting all services myself ? 2015-07-30 09:53:37 would like to try it but I'm lazy 2015-07-30 09:54:04 i use it daily for now more than 6 months on my laptop 2015-07-30 09:54:20 and you can check my github for the implemented services running a minimal alpine system. 2015-07-30 09:57:28 yep saw it 2015-07-30 22:00:05 hmmm, where did git://git.alpinelinux.org/bbsuid dissapear ? 2015-07-30 23:08:44 Any HHVM package? 2015-07-30 23:43:03 Any HHVM package? 2015-07-31 06:26:27 good morning 2015-07-31 06:35:54 morning 2015-07-31 06:49:02 hi 2015-07-31 07:18:03 hello 2015-07-31 07:21:44 早安 2015-07-31 07:22:31 algitbot: haha got you! 2015-07-31 07:22:50 ncopa: how would one debug a lua module? 2015-07-31 07:25:30 :) 2015-07-31 07:25:35 google lua debugging 2015-07-31 07:25:51 hum 2015-07-31 07:26:07 does not really tell how to do the debugging 2015-07-31 07:26:21 its a c module 2015-07-31 07:26:35 lua-sqlite 2015-07-31 07:26:45 does it segfault? 2015-07-31 07:26:52 no just hang 2015-07-31 07:27:04 one some queries 2015-07-31 07:27:19 its not that important, was just wondering 2015-07-31 07:27:21 can you do the same query from cmdline? 2015-07-31 07:27:41 it works on pkgs now 2015-07-31 07:27:47 with dbi module 2015-07-31 07:28:00 doing the same with the other sql module will fail 2015-07-31 07:28:22 weird thing is, if i list * then it will not display some pkgs 2015-07-31 07:28:41 if i add those missing pkgs to the query, it hangs. 2015-07-31 07:29:30 makes me thing it has something to do with musl 2015-07-31 07:29:34 think... 2015-07-31 08:17:28 ncopa: hiawatha got a new release: 9.13 -> 9.14 2015-07-31 08:17:48 i cannot send you a patch right now, because i am on my laptop and i have no dev tools here 2015-07-31 08:27:39 ok 2015-07-31 08:27:41 hum 2015-07-31 08:27:42 does not build 2015-07-31 08:27:48 also mbedtls-2.0 came out 2015-07-31 08:30:39 ok 2015-07-31 08:30:51 hiawatha does not build with mbedtls-1.3 2015-07-31 08:31:20 need to update mbedtls first 2015-07-31 08:31:26 but it is not API compatible 2015-07-31 08:31:29 so all apps needs porting 2015-07-31 08:35:51 i thught alpine already ships mbedtls-2.0 2015-07-31 08:36:06 my bad, must have confused it during lesting with my local mbedtls version+ 2015-07-31 08:36:09 sorry about that 2015-07-31 09:25:16 leo-unglaub: is this the official upstream? https://github.com/hsleisink/hiawatha 2015-07-31 09:25:30 yeah 2015-07-31 09:25:31 it is 2015-07-31 09:25:39 it has underlinking problem 2015-07-31 09:25:53 ncopa: yeah, its the official upstream 2015-07-31 09:25:59 why is it underlinking? 2015-07-31 09:27:01 hiawatha and wigwam binaries uses symbols in libmbedx509 and libmbedcrypto 2015-07-31 09:27:20 but it does not add -lmbedx509 -lmbedcrypto 2015-07-31 09:27:39 it probably works on his system due to -lmbed indirectly pulls those in 2015-07-31 09:28:20 eg hiawatha links to libmbed which is linked to libmbedx509 and libmbedcrypto 2015-07-31 09:29:03 ncopa: dont bother wasting your time with it 2015-07-31 09:29:13 i do it tonight and send you a patch as soon as i am at home 2015-07-31 09:29:21 you have more important stuf todo :) 2015-07-31 09:29:56 too late... 2015-07-31 09:30:48 i pushed it 2015-07-31 09:31:14 but you could send the patch i made upstream 2015-07-31 09:31:35 http://tpaste.us/3JME 2015-07-31 09:32:01 will do, i send him an email 2015-07-31 09:32:02 problem explain: http://wiki.rosalab.com/ru/index.php/Underlinking 2015-07-31 09:32:05 thanks 2015-07-31 09:35:03 ncopa: patch+explonation send to the developer 2015-07-31 09:35:24 thanks 2015-07-31 09:38:26 thank you 2015-07-31 09:40:45 ncopa: got an mail back, he will fix it in the next release 2015-07-31 09:41:07 that was quick :) 2015-07-31 09:42:21 yeah :) we know each other so he knows to read my mails first *g* 2015-07-31 09:44:06 :) 2015-07-31 10:09:49 i have to go to the red cross. see you guys later!