2014-03-03 09:05:16 clandmeter: would you be able to look if we could get this redmine feature: http://www.redmine.org/issues/5510 2014-03-03 09:25:57 tdtrask, could you have look again at form label changes 2014-03-03 09:26:04 page like, cgi-bin/acf/acf-util/password/edituser?userid=root&redir=/acf-util/password/status 2014-03-03 09:26:52 still have not labels 2014-03-03 09:34:58 ncopa: that bug is 4 years old 2014-03-03 09:35:10 bug/feature 2014-03-03 09:35:11 lol 2014-03-03 09:35:27 that was exatly my first reponse too! 2014-03-03 09:35:31 exactly* 2014-03-03 09:35:43 but it looks like they added it to roadmap 11 months ago 2014-03-03 09:38:37 sha512sum: can't open '0001-fibmap.f2fs-compile-fix-for-musl-libc.patch': No such file or directory 2014-03-03 09:38:37 0001-fibmap.f2fs-compile-fix-for-musl-libc.patch: FAILED 2014-03-03 09:38:50 ncopa, sounds like you want the 'write git hook to detect checksums' award ;) 2014-03-03 09:38:54 s/want/won/ 2014-03-03 09:38:58 AAAW!!!! 2014-03-03 09:39:07 \o/ 2014-03-03 09:41:31 lol 2014-03-03 09:43:03 fabian_a: ping 2014-03-03 09:43:46 fabian_a: http://bugs.alpinelinux.org/issues/2689 2014-03-03 09:44:00 seems like enabling xattrs in coreutils broke fakeroot 2014-03-03 09:44:12 do you mind if i disable xattrs for v2.7.5? 2014-03-03 09:51:56 ncopa: the latest release of coreutils is 8.22. perhaps an update fixes the issue with acl (but i doubt that because there are no details in changelog for 8.22), otherwise no objections to disable 2014-03-03 09:52:29 i dont think the issue is coreutils but i fakeroot 2014-03-03 09:52:53 we could try fix the support in edge, but for v2.7 i think we might want disable it 2014-03-03 09:53:02 disable acl in coreutils that is 2014-03-03 09:53:13 ncopa, i dug into it at great detail earlier. i remember that it's sort of gnulib issue 2014-03-03 09:53:32 in certain cases it's 'smart' and assumes the acl stuff does chmod/chown 2014-03-03 09:53:43 should be relatively easy to fix 2014-03-03 09:53:59 fabled: would you be able to have a look at it? 2014-03-03 09:54:22 i would prefer to fix it so we continue with acl support in v2.7 if possible 2014-03-03 09:54:47 if not easily fixable or if it is too timeconsuming, we can disable acl in v2.7 2014-03-03 09:55:02 i'd like it fixed soonish 2014-03-03 09:55:08 today or tomorrow 2014-03-03 09:55:16 so we can do release 2.7.5 this week 2014-03-03 09:56:15 was it just mode that's messed up, or also the owner/group ? 2014-03-03 09:57:15 i'm only aware of owner/group 2014-03-03 10:00:18 mmm. ok. i'll try to take a look at it. off for lunch now. 2014-03-03 10:01:51 i have backported aports-build (the mqtt based thingy) to v2.6 2014-03-03 10:02:04 which means that the zmq scripts are completely out of the picture now 2014-03-03 10:03:42 lets see iv v2.6 builders works... 2014-03-03 10:03:46 if* 2014-03-03 10:04:21 seems like it does :) 2014-03-03 10:09:34 files from v2.5.4-247-gf574def uploaded 2014-03-03 10:09:43 files from v2.4.11-195-g67acd4b uploaded 2014-03-03 10:09:46 files from v2.4.11-195-g67acd4b uploaded 2014-03-03 10:27:24 ncopa, mqtt has stopped msgs -- building main/ ? 2014-03-03 10:27:41 no? 2014-03-03 10:27:44 i think not 2014-03-03 10:27:53 ah 2014-03-03 10:28:00 building pkgname... 2014-03-03 10:28:00 hm 2014-03-03 10:28:14 maybe for stable branches 2014-03-03 10:28:18 i'll check that up 2014-03-03 10:28:19 it jumpped from pulling git -- building -- to rsync 2014-03-03 10:28:32 gobbled up "build main/...." 2014-03-03 10:28:45 yes it might be it does not show the stable branches 2014-03-03 10:28:55 i will check it up in a bit 2014-03-03 10:30:01 there is a new pkg testing/fossil for trying 2014-03-03 10:30:15 and grep for main 2014-03-03 10:30:17 :) 2014-03-03 10:30:40 whats that fossil version? 2014-03-03 10:31:27 2014-01-27 2014-03-03 10:31:31 Version 1.28 2014-03-03 10:31:48 http://www.fossil-scm.org/download/fossil-linux-x86-20140127173344.zip 2014-03-03 10:32:54 mqtt msg now ok 2014-03-03 10:33:01 i think its ok only for edge 2014-03-03 10:33:06 not for the stable builders 2014-03-03 10:35:21 files from v2.4.11-196-g6c587ad uploaded 2014-03-03 10:35:44 could try pushing newer man-pages to stable 2014-03-03 10:36:16 files from v2.4.11-196-g6c587ad uploaded 2014-03-03 10:37:02 files from v2.5.4-248-g32cb9c6 uploaded 2014-03-03 10:37:28 sh**, it seems that i was not persistent enough for my last pull request. it was possible to rebase my repo with some skips but can somebody please try to git pull git://git.alpinelinux.org/fab/aports 2014-03-03 10:38:09 some licenses specified and some pkgs updated 2014-03-03 10:39:12 and some new aports, too 2014-03-03 10:39:39 thanks 2014-03-03 10:51:19 how can i know what library i have to use if i know what is passed to the linker? 2014-03-03 10:51:23 Example: 2014-03-03 10:51:27 -lcrypt = openssl 2014-03-03 10:52:12 how can i know what library i need to use for -lcrypt if i don't know ? 2014-03-03 10:52:23 in general, -l will look for /usr/lib/lib.so 2014-03-03 10:52:24 another example: 2014-03-03 10:52:34 -lpthread = ? 2014-03-03 10:52:37 -ldl = ? 2014-03-03 10:52:43 -lm = ? 2014-03-03 10:53:06 so apk info --who-owns /usr/lib/libcrypt.so 2014-03-03 10:53:16 or in those cases it is in libc itself 2014-03-03 10:53:21 so /lib/libcrypt.so 2014-03-03 10:53:29 interesting 2014-03-03 10:53:49 libpthread, libdl libm and librypt are all libc 2014-03-03 10:53:56 libcrypto is openssl 2014-03-03 10:54:16 infact they are symlinks 2014-03-03 10:54:20 correct 2014-03-03 10:54:38 generally, then -l$libname = /lib/lib$libname 2014-03-03 10:54:44 yes 2014-03-03 10:54:47 godo to know 2014-03-03 10:54:49 *good 2014-03-03 10:54:58 thx ncopa 2014-03-03 10:54:59 or /usr/lib/lib$libname.so 2014-03-03 10:55:23 if gcc does not find the .so, it will look for .a 2014-03-03 10:55:28 then it will be statically linked 2014-03-03 10:55:58 which means that the contents of the needed functions in the .a file will be copied into the build binary 2014-03-03 10:56:06 yep 2014-03-03 10:56:43 you can see what libraries a compiled binary was linked to with readelf -d | grep NEEDED 2014-03-03 10:57:08 the NEEDED section tells what -l was used 2014-03-03 10:57:39 i was aware about the readeld 2014-03-03 10:57:41 *readelf 2014-03-03 10:57:50 but i need the contrary 2014-03-03 10:58:04 the /lib/lib$liname is ok 2014-03-03 10:58:10 thx a lot 2014-03-03 10:58:11 good 2014-03-03 10:58:15 yw 2014-03-03 10:58:50 hmpf, nss security fix needs newer nspr 2014-03-03 10:59:08 maintaining nss and nspr (or any mozilla blah) is painful 2014-03-03 11:14:25 fakeroot acl support is a joke, but should fix the issue 2014-03-03 11:14:53 basically it overrides few libacl symbols to return 'not supported'... that makes coreutils etc. to fallback to the regular syscalls 2014-03-03 11:15:02 so that should fix all acl related fakeroot issues 2014-03-03 11:59:06 fabled: thanks for fixing this 2014-03-03 12:31:15 fabled: that means acl is in the bootstrap chain for crosscompile. i suppose that does not matter 2014-03-03 12:31:25 ncopa, it does not matter 2014-03-03 12:31:46 thanks for fixing it 2014-03-03 12:31:58 when i was looking at it i tested attr-dev 2014-03-03 12:32:08 i was on wrong path :-/ 2014-03-03 12:39:27 wow. broadcom open source Video Core stuff 2014-03-03 12:56:39 stable builders still missing 'building/main....' msgs 2014-03-03 12:56:46 yes 2014-03-03 12:56:52 i havent tried to fix it yet 2014-03-03 12:57:09 i'm not sure i will bother fix it as it might require some abuild patching 2014-03-03 12:57:33 i will look into it later 2014-03-03 12:59:09 ok 2014-03-03 13:01:59 another option would be publish newer files rsync'd 2014-03-03 13:02:31 may use sfic 2014-03-03 13:03:05 ok here comes fabian_a's changes 2014-03-03 13:03:42 fabian_a: thanks! 2014-03-03 13:05:41 http://git.alpinelinux.org/cgit/aports/commit/?id=36b3b7df12a38987a54799e467ab1ed585dbbd15 2014-03-03 13:05:45 why the pkgrel bump? 2014-03-03 13:05:48 vkrishn: pages like /acf-util/password/edituser are unusual 2014-03-03 13:05:59 and who cares about quotes after empty variable 2014-03-03 13:06:06 basically, that should be a multi-select, but the haserl engine cannot handle them 2014-03-03 13:06:09 an empty variable is still empty 2014-03-03 13:06:17 so, I was forced to create a series of checkboxes 2014-03-03 13:06:22 barthalion: yeah... no need bump pkgrel 2014-03-03 13:06:33 algitbot: please... youre annoying... 2014-03-03 13:07:03 not sure how best to handle that, because there is one single label tag for the whole series of checkboxes 2014-03-03 13:07:44 ncopa: you are courageous...i thought that you want to release 2.7.5 soon ;-) 2014-03-03 13:07:56 i want... 2014-03-03 13:08:19 files from v2.5.4-250-g9876a50 uploaded 2014-03-03 13:09:57 tdtrask, then maybe hardcode in separate file and then fetch it 2014-03-03 13:10:19 ? 2014-03-03 13:13:03 fabian_a: if you're making stylistic changes don't bump pkgrel 2014-03-03 13:13:32 or multiselect multiple="multiple" 2014-03-03 13:14:13 vkrishn: the /acf-util/password/edituser page on my box has 182 errors and 178 warnings on w3 validator :( 2014-03-03 13:15:20 all of the checkboxes have the same id 2014-03-03 13:15:33 ACTION will look into it 2014-03-03 13:15:37 thanks 2014-03-03 13:16:54 ACTION really should look for a crawler / HTML validator to automatically check the entire ACF 2014-03-03 13:18:14 any suggestions? 2014-03-03 13:18:43 validator can be local installed 2014-03-03 13:19:27 cannot recall, there are some tools like htmltidy...etc that can check and report errors 2014-03-03 13:21:16 I thought of putting w3 validator in pkg request bugs.a.o 2014-03-03 13:25:08 ncopa: I guess you know that ruby 2.1.1 is out 2014-03-03 13:25:37 clandmeter: ^^^ 2014-03-03 13:25:49 I'm keeping away from it, not sure if someone is eager to get his hands dirty 2014-03-03 13:25:51 barthalion: ruby scares me... :) 2014-03-03 13:25:56 yeah, same here 2014-03-03 13:26:07 ACTION hides 2014-03-03 13:26:19 I thought so :D 2014-03-03 13:26:25 lol 2014-03-03 13:31:20 reading somewhere, ruby has chance of getting in LSB or weblang 2014-03-03 13:32:00 even git ML had some discussion of including it 2014-03-03 13:32:27 imho, the gems decs needs to do better job in release engineering 2014-03-03 13:32:52 just because felipec posted on git ML that ruby is zomg so great, that doesn't mean it will get included there… 2014-03-03 13:33:05 ;) 2014-03-03 13:37:43 files from v2.5.4-251-g925aedc uploaded 2014-03-03 13:40:00 files from v2.4.11-197-g934da98 uploaded 2014-03-03 13:40:00 files from v2.4.11-197-g934da98 uploaded 2014-03-03 14:10:35 files from v2.5.4-252-g99ae557 uploaded 2014-03-03 14:15:13 files from v2.4.11-198-g474e266 uploaded 2014-03-03 14:24:55 files from v2.4.11-199-g1143d93 uploaded 2014-03-03 14:25:11 files from v2.4.11-199-g1143d93 uploaded 2014-03-03 14:31:35 bah 2014-03-03 14:31:41 some uclibc specific issue 2014-03-03 14:38:38 found patch 2014-03-03 14:48:31 i think we want jump sideways to musl asap 2014-03-03 14:48:42 maintaining both uclibc and musl is more work 2014-03-03 15:18:43 vkrishn: check out the changes to acf-core and acf-lib to use HTML select with multiple flag rather than use series of checkboxes 2014-03-03 15:49:27 hum... is that readline upgrade API compat? 2014-03-03 15:52:25 seems not, sorry for not checking 2014-03-03 15:52:32 I'll push the rebuild in a minute 2014-03-03 15:54:50 ncopa: http://sprunge.us/cXXe does it look ok? 2014-03-03 15:55:47 whats that? 2014-03-03 15:56:08 rebuild list 2014-03-03 15:56:47 h 2014-03-03 15:56:48 hm 2014-03-03 15:57:00 i think readline is ABI compat 2014-03-03 15:57:11 oh 2014-03-03 15:57:14 maybe not 2014-03-03 15:58:00 hm 2014-03-03 15:58:03 wait a sec 2014-03-03 15:58:07 i think its ABI compat 2014-03-03 15:58:40 yup 2014-03-03 15:58:46 its other problem 2014-03-03 15:59:12 it could be that API changed, to confuse freeradius 2014-03-03 15:59:21 or its a depedency issue 2014-03-03 16:00:16 i dont know why it fails 2014-03-03 16:00:24 but i need to go now 2014-03-03 16:46:29 tdtrask, cgi-bin/acf/alpine-baselayout/cron/listjobs has 2 selects menu with same id 2014-03-03 16:47:08 thanks for select (multi) 2014-03-03 16:48:18 vkrishn: you're happy with the multi? 2014-03-03 16:48:58 if it works when form is posted then ok 2014-03-03 16:49:28 jqmobile has way of showing how many of them are selected 2014-03-03 16:49:30 yeah, the issue with haserl was very old 2014-03-03 16:49:34 apparently it works now 2014-03-03 16:50:10 I am currently trying to get the interface all working as intended by framework 2014-03-03 16:50:16 personally, I think multiple selects are difficult to read 2014-03-03 16:50:25 then sort out any backend issue or any 2014-03-03 16:50:37 ok, sounds good 2014-03-03 16:50:44 please keep informing me of issues you find 2014-03-03 16:50:47 its ok on jqmobile 2014-03-03 16:51:21 yeah, I figured it's a skin issue that could be fixed with CSS 2014-03-03 16:51:26 for the other skins 2014-03-03 16:51:30 I think I would upload the binary/sources for basic acf install, like earlier and let you see 2014-03-03 16:51:51 ok 2014-03-03 16:51:56 late today or tomorrow 2014-03-03 16:52:07 git patches would be nicer for merging 2014-03-03 16:52:47 having full pass the validator would be nice 2014-03-03 16:54:00 I think that should be priority for next AL release 2014-03-03 16:54:31 yes, I agree 2014-03-03 16:55:03 right now, going through HTML view files to make better use of htmlviewfunctions for consistency 2014-03-03 16:55:22 ACTION will try to get to crawling validator later today 2014-03-03 16:55:58 some distros like deb has pkgs for them already 2014-03-03 18:04:48 tdtrask, done, upload or prefer git patches ? 2014-03-03 18:49:32 tdtrask, src at http://download.tuxfamily.org/pmreader/alpine/src/acf/ 2014-03-03 18:49:46 also added aports for new pkgs 2014-03-03 18:50:44 I can send git patches on ML too, once its ok 2014-03-03 19:07:22 apk at http://download.tuxfamily.org/pmreader/alpine/edge/testing/x86/ 2014-03-03 19:12:23 and install script http://download.tuxfamily.org/pmreader/alpine/set-acf-alpine-v2.7.sh 2014-03-03 19:40:43 I'm disgusted, we have no ed in repositories! 2014-03-03 20:02:31 does anything in bison or flex link to readline? 2014-03-03 20:02:34 I don't think so 2014-03-03 22:20:20 vkrishn: we are no longer using DL/DT/DD in the latest ACF code in git 2014-03-03 22:20:41 so, we need a different way to replace the interfaces table 2014-03-03 22:21:10 also, I'm working on making the header level customizable by the skin 2014-03-03 22:21:21 instead of hardcoding to 1 or 3 2014-03-04 07:06:55 tdtrask, no issues if DL/DT/DD related code gets changed, I would update my local 2014-03-04 07:07:00 thanks for other changes 2014-03-04 07:10:05 also in page like 'alpine-baselayout/interfaces/read' I added
for subsection, have a look at it to see if its feasible for other pages 2014-03-04 07:11:19 this subsection helps in creating collapsible divs 2014-03-04 07:15:47 barthalion: i know whats wrong with readline 2014-03-04 07:15:59 its an underlinking issue 2014-03-04 07:16:47 https://dpaste.de/UTBP 2014-03-04 07:16:54 as you see, readline uses tgetnum 2014-03-04 07:17:15 $ nm -D /usr/lib/ 2014-03-04 07:17:15 libreadline.so | grep tgetnum 2014-03-04 07:17:15 U tgetnum 2014-03-04 07:17:31 but it is not linked with anything providing it 2014-03-04 07:18:02 $ readelf -d /usr/lib/libreadline.so | grep NEED 2014-03-04 07:18:02 0x0000000000000001 (NEEDED) Shared library: [libc.so.0.9.32] 2014-03-04 07:18:16 it might be intentional 2014-03-04 07:20:33 yup, as i suspected: https://bugzilla.redhat.com/show_bug.cgi?id=499837 2014-03-04 07:21:24 and yes, we force linked readline 6.2 it to ncurses 2014-03-04 07:21:30 i think thats what we need to do 2014-03-04 07:28:45 LXC 1.0 has been release 2014-03-04 07:35:15 and newer libvirt 2014-03-04 08:10:13 barthalion: I'm pretty sure readline is underlinked in arch linux too 2014-03-04 08:10:50 https://projects.archlinux.org/svntogit/packages.git/tree/trunk/PKGBUILD?h=packages/readline#n44 2014-03-04 08:11:02 as you see, they try to force SHLIB_LIBS=-lncurses 2014-03-04 08:11:18 but its not doing it 2014-03-04 08:11:35 check yourself with readelf -d /usr/lib/libreadline.so | grep NEEDED 2014-03-04 08:11:46 i bet ncurses is not there on arch linux 2014-03-04 08:13:04 i wonder if were should just replace gnu readline with something compat 2014-03-04 09:51:44 ncopa: it is underlinked here, I've just read backlog from #archlinux-tu 2014-03-04 09:52:45 ncopa: there is libedit in main already 2014-03-04 09:56:30 libedit is not API compat? 2014-03-04 09:56:50 i am pretty annoyed with what gnu readline does 2014-03-04 09:57:14 I don't know any other alternative 2014-03-04 09:57:20 i saw 3 2014-03-04 09:57:41 API compatible? 2014-03-04 09:57:48 libedit, linenoise, tecla 2014-03-04 09:57:52 i think they are not API compat 2014-03-04 09:58:09 might be tecla is 2014-03-04 09:58:20 http://stackoverflow.com/questions/2736248/simple-alternative-to-gnu-readline-library-not-gpl 2014-03-04 09:58:42 the intentional underlink in readline is really stupid 2014-03-04 09:58:54 it is so you can select either ncurses or termcap 2014-03-04 09:59:08 but it is only useful for binary only 2014-03-04 09:59:35 and readline is GPL so you are not really allowed to link binary-only with it 2014-03-04 11:13:17 fabled: did you already push a fix for that acl stuff from yesterday? 2014-03-04 11:13:27 clandmeter, yes, for edge only 2014-03-04 11:13:38 i get this on edge: 2014-03-04 11:13:38 dlsym(acl_get_fd): Symbol not found: acl_get_fd 2014-03-04 11:13:38 dlsym(acl_get_file): Symbol not found: acl_get_file 2014-03-04 11:14:05 thats musl 2014-03-04 11:14:30 its in the end of abuild, so i guess its fakeroot? 2014-03-04 11:19:02 clandmeter, yes, those are safe to ignore 2014-03-04 11:19:05 should fix it though 2014-03-04 11:20:03 its a little verbose :) 2014-03-04 11:20:53 i think it's musl issue 2014-03-04 11:21:05 or is it on uclibc too? 2014-03-04 11:22:17 yeah, it's musl only 2014-03-04 11:22:18 i'll nag about it on #musl 2014-03-04 12:20:18 files from v2.5.4-253-g22cc903 uploaded 2014-03-04 12:27:54 files from v2.4.11-200-g8928dd3 uploaded 2014-03-04 12:40:23 files from v2.5.4-254-g6eee885 uploaded 2014-03-04 12:40:34 files from v2.5.4-254-g6eee885 uploaded 2014-03-04 12:47:09 files from v2.4.11-201-g4fba928 uploaded 2014-03-04 12:47:15 files from v2.4.11-201-g4fba928 uploaded 2014-03-04 13:39:36 vkrishn: add "pageinfo.header_level=3" before the call to pageinfo.viewfunc in your template in order to use the dynamic header levels 2014-03-04 13:40:08 is it possible to add the
programmatically rather than statically? 2014-03-04 13:48:50 tdtrask, class="section4" can be class="??", haven't tried via js , what issues are that for such wrapper ? 2014-03-04 13:50:04 adding the div before the header is easy, adding the /div at the end means more changes in lots of files 2014-03-04 13:50:37 not a huge deal 2014-03-04 13:50:53 just more work for 30 ACFs to become compliant 2014-03-04 13:51:11 I can add them as I test each acf- 2014-03-04 13:51:46 ACTION would like to add them with equivalent of htmlviewfunctions.displayheader 2014-03-04 13:51:59 such as htmlviewfunctions.closeheader 2014-03-04 13:52:05 or some function like that 2014-03-04 13:52:22 maybe startsection / closesection 2014-03-04 13:52:41 sounds ok 2014-03-04 13:53:19 ACTION was looking briefly at data-native-menu and data-iconpos in all of the 'controller'.js files 2014-03-04 13:53:35 is there a way to make that generic so we don't need all those 'controller'.js files? 2014-03-04 13:53:49 that's the plan 2014-03-04 13:53:54 ok 2014-03-04 13:53:54 once all is done 2014-03-04 13:54:01 figure out a common ay 2014-03-04 13:54:03 way 2014-03-04 13:54:12 ACTION doesn't like the idea of changing the skin when new controllers are added 2014-03-04 13:54:17 :) 2014-03-04 13:54:26 I still don't know what's there to come in other modules 2014-03-04 13:54:41 understood 2014-03-04 13:55:16 ACTION is trying to make all modules consistent 2014-03-04 13:55:50 to make your job easier 2014-03-04 13:55:57 :) 2014-03-04 13:57:51 files from v2.5.4-255-gbb729dc uploaded 2014-03-04 13:58:05 files from v2.5.4-255-gbb729dc uploaded 2014-03-04 13:58:47 thanks for all the help vkrishn 2014-03-04 13:58:50 ACTION has to run 2014-03-04 13:58:55 also have a look at cgi-bin/acf/alpine-baselayout/interfaces/read, i.e how popup is called 2014-03-04 13:59:07 not sure why you used popup? 2014-03-04 13:59:22 would this be ok, interfaces-read-popups-html.lsp ? 2014-03-04 13:59:50 helps in not adding the code in original file and doing 2014-03-04 14:00:09 ACTION will look at it after my conference call 2014-03-04 14:00:14 ok 2014-03-04 14:09:26 files from v2.4.11-202-ge760d56 uploaded 2014-03-04 14:09:27 files from v2.4.11-202-ge760d56 uploaded 2014-03-04 15:25:42 barthalion: http://lists.gnu.org/archive/html/bug-readline/2014-03/msg00006.html 2014-03-05 07:58:17 files from v2.5.4-256-ga5c018c uploaded 2014-03-05 07:58:27 files from v2.5.4-256-ga5c018c uploaded 2014-03-05 08:36:51 files from v2.4.11-203-g18c70cf uploaded 2014-03-05 09:45:43 i think i definitively won the make git hook for pushing brokn checksum competition :-( 2014-03-05 10:38:00 files from v2.4.11-204-g1933d0f uploaded 2014-03-05 10:38:14 files from v2.4.11-204-g1933d0f uploaded 2014-03-05 10:52:17 files from v2.5.4-257-g98c0c11 uploaded 2014-03-05 10:59:25 files from v2.4.11-205-g4bac042 uploaded 2014-03-05 11:00:03 files from v2.4.11-205-g4bac042 uploaded 2014-03-05 11:01:03 files from v2.5.4-258-g10c1e0f uploaded 2014-03-05 11:01:37 files from v2.5.4-258-g10c1e0f uploaded 2014-03-05 11:01:46 files from v2.4.11-206-g3e6173a uploaded 2014-03-05 11:02:38 files from v2.4.11-206-g3e6173a uploaded 2014-03-05 13:58:32 fabled: can i trigger a specific package build for arm? 2014-03-05 13:58:46 clandmet1r, currently no 2014-03-05 13:58:52 we are working to make it follow git head 2014-03-05 13:58:55 ncopa, ^^^ 2014-03-05 13:59:06 clandmet1r, i can trigger it once if needed 2014-03-05 14:01:19 libtirpc has wrong sysconf dir which rpcbind dont like, but i can for now just move its config to /etc 2014-03-05 14:02:31 i can build that for you 2014-03-05 14:03:46 ok, thx 2014-03-05 14:06:04 clandmet1r, built 2014-03-05 16:32:22 fabled: still around? 2014-03-06 06:47:08 did anyone do wayland/weston for alpine? 2014-03-06 06:47:59 do we have anything capable of using it? 2014-03-06 06:48:44 probably not 2014-03-06 06:48:52 i'm looking into using Qt5 on raspberry pi 2014-03-06 06:48:55 I can take a look, but getting xwayland to work requires some patches for xorg-server 2014-03-06 06:48:57 two options 2014-03-06 06:49:23 using eglfs with broadcom platform plugin (i would like to do some patches to make that work nicely) 2014-03-06 06:49:25 or using wayland 2014-03-06 06:49:42 it seems qtwayland has relatively nice way to support different gl stacks 2014-03-06 06:50:13 then again... if using qtwayland's wayland compositor, it would mean that eglfs patching needs to be done anyway 2014-03-06 06:50:45 so maybe i should just fix qt5's platform plugin building to be more modular 2014-03-06 06:50:52 since that seems to be needed anyway 2014-03-06 07:12:49 ncopa: uh, I guess that your missing packages list doesn't include testing? 2014-03-06 07:14:09 correct 2014-03-06 07:21:19 how is that? it builds here 2014-03-06 07:21:32 hrm 2014-03-06 07:21:57 barthalion, it works on musl, but not uclibc 2014-03-06 07:22:02 you likely want 2014-03-06 07:22:17 !defined(__GLIBC__) || defined(__UCLIBC__) 2014-03-06 07:22:30 ah 2014-03-06 07:22:32 thank you 2014-03-06 07:22:49 defines on: 2014-03-06 07:22:50 glibc: __GLIBC__ 2014-03-06 07:22:54 uclibc: __GLIBC__ and __UCLIBC 2014-03-06 07:22:57 musl: (none) 2014-03-06 07:23:04 __UCLIBC__* 2014-03-06 07:29:35 let's see in uclibc chroot 2014-03-06 08:41:15 fabled: im now also getting this on uclibc http://sprunge.us/aXPM 2014-03-06 08:41:31 yes, it's safe to ignore 2014-03-06 08:41:34 need to still fix it 2014-03-06 08:41:37 it's annoying 2014-03-06 08:58:15 doh 2014-03-06 09:48:02 ncopa: I'll start working on eudev soon 2014-03-06 09:48:38 ncopa: and about moving away from mysql… we can switch to percona until mariadb 10 is released; users shouldn't notice anything more than new pkgname 2014-03-06 10:00:01 percona? 2014-03-06 10:02:55 http://www.percona.com/software/percona-server 5.5 branch 2014-03-06 11:40:15 /j #archlinux-security 2014-03-06 11:40:17 oops 2014-03-06 14:25:37 vkrishn: please see the latest changes to acf-core and acf-alpine-baselayout git repos, particularly to rc-status-html.lsp 2014-03-06 14:26:02 I'm not sure I like the code yet, but it is an attempt to replace the links with forms 2014-03-06 14:26:15 I figured it might look better in jquery mobile too 2014-03-06 14:26:23 please let me know what you think 2014-03-06 14:53:18 tdtrask, both links and button can be styled same in jqmobile 2014-03-06 14:54:34 but what would be interesting, if they are converted to select-menu , with text "action" 2014-03-06 14:54:59 less clutter 2014-03-06 14:56:02 possible to display it as icon only 2014-03-06 14:58:06 bug:

shows as 2014-03-06 14:58:54 OR another option if each is div wrapped, I can make it collapsible 2014-03-06 14:59:20 which are initially collapsed on page load 2014-03-06 15:07:39 gtg 2014-03-06 15:13:29 vkrishn: for the fix, you need the latest acf-core/lib/htmlviewfunctions.lua 2014-03-06 16:36:46 tdtrask, acf-core may require adjusting Makefile 'acf-util/skins-read-html.lsp' 2014-03-06 16:37:35 pageinfo.header_level=3 seems nice 2014-03-06 16:38:32 cgi-bin/acf/alpine-baselayout/rc/status - passed html5 validation, nice!! 2014-03-06 16:47:45 seeing "Application error occured" in cgi-bin/acf/alpine-baselayout/interfaces/read 2014-03-06 19:06:30 vkrishn: those two issues are now fixed in acf-core git repo 2014-03-07 11:00:30 files from v2.5.4-259-g13e92ef uploaded 2014-03-07 11:34:07 hello, everyone 2014-03-07 11:34:07 however, it seems that posix-spawn.c doesn't like uClibc spawn.h 2014-03-07 11:34:07 http://pastebin.com/zXgncFWq 2014-03-07 11:34:07 I am trying to install posix-spawn ruby gem in alpine 2014-03-07 11:42:51 sounds like it should be __restrict 2014-03-07 11:43:02 yeah, smeels uclibc bug 2014-03-07 11:45:18 :-D 2014-03-07 11:45:44 it seems that spawn.h included in ubuntu uses __restrict 2014-03-07 11:46:00 yes, both glibc and musl uses __restrict 2014-03-07 11:47:38 https://github.com/hwoarang/uClibc/blob/master-metag/include/spawn.h 2014-03-07 11:55:59 http://lists.uclibc.org/pipermail/uclibc/2012-April/046784.html - [PATCHv3] librt: Add posix_spawn support 2014-03-07 11:56:07 http://lists.uclibc.org/pipermail/uclibc-cvs/2012-March/030155.html - [git commit] spawn: use glibc style to keep differences down 2014-03-07 12:02:47 http://git.uclibc.org/uClibc/tree/include/spawn.h?id=20b69920b299585265eb100d0b67e1097ccb1092 2014-03-07 12:03:09 http://git.uclibc.org/uClibc/log/include/spawn.h 2014-03-07 12:05:07 could be that alpine does not include the patch that is already in uClibc? 2014-03-07 15:37:47 algitbot: build 2.7-stable 2014-03-07 16:06:18 http://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2014-0092 2014-03-07 16:49:14 alacerda: fixed in edge week or two ago, no idea about the rest 2014-03-07 16:55:04 alacerda: other releases are safe too 2014-03-07 16:59:13 barthalion: =] 2014-03-07 16:59:21 al being ahead 2014-03-07 17:00:33 well, not really 2014-03-07 17:00:55 the disclosure was known before all so popular websites started pretending to be smart about security 2014-03-07 17:06:14 I did a manjaro port on the chromebook, thinking of doing a alpine one too. I think the alpine one would be better for development processes because it would encourage me to get into packaging for alpine ;) 2014-03-07 17:07:35 if it's not ARM chromebook, I don't know why do you call it porting 2014-03-07 17:13:38 barthalion: maybe because I am not sure what to call it. 2014-03-07 17:14:41 barthalion: its not an ARM chromebook no, would it be easy to build an arm version of alpine? 2014-03-07 17:14:55 well, depends who you ask 2014-03-07 17:15:03 guess it wasn't too hard for fabled 2014-03-07 17:15:12 but generally speaking… it's not that simple 2014-03-07 17:15:32 and about getting into packaging, you don't need to run Alpine on your laptop 2014-03-07 17:15:38 just look for requests on our bug tracker 2014-03-07 17:18:05 barthalion: I know I do not 'need' to run Alpine. I 'want' to run alpine in crouton. I like how crouton works and think alpine will behave nicer than manjaro does because the debian systems use busybox and work fine. Systemd is troublesome. 2014-03-07 17:19:38 barthalion: crouton allows you to run a graphical chroot beside your chrome os install. It is a very neat concept. 2014-03-07 17:20:21 thus far it has allowed me to be able to run steam games on my chromebook :) 2014-03-07 17:21:02 I couldn't use ChromeOS as I couldn't use a Mac :) 2014-03-07 17:21:45 barthalion: I wanted a thinclient/dev box and it works absolutely perfect for that. 2014-03-07 18:51:15 blueness: if it's not a secret, did someone apply for embedded Gentoo w/musl already? 2014-03-07 22:10:39 algitbot: build edge 2014-03-07 22:11:25 algitbot: build edge-x86_64 2014-03-07 22:12:14 algitbot: build master 2014-03-07 22:18:24 names 2014-03-07 22:20:10 barthalion: you know how to kick those builders? 2014-03-07 22:20:47 larena_: do you know? 2014-03-07 22:21:45 algitbot: retry 2014-03-07 22:26:38 algitbot: build-edge-x86 2014-03-07 22:27:32 algitbot: build 2014-03-07 23:17:23 barthalion, nope, you can apply if you like 2014-03-08 04:13:17 main/owncloud-6.0.2-r0 was built, but builders did show - 'ERROR: owncloud: libc.so.6: path not found' 2014-03-08 04:13:51 I hope it is ok, cause it seems to have gone forward with "uploading packages" 2014-03-08 09:52:20 clandmeter: build edge 2014-03-08 09:53:09 i tried everything 2014-03-08 09:53:14 :) 2014-03-08 09:53:39 and the checksum fail was also weird 2014-03-08 12:58:19 blueness: my Gentoo skills are a little bit rusty 2014-03-09 14:36:58 barthalion, i'm not as worried about "gentoo skills" which i can lead you through as your coding skills 2014-03-09 14:37:07 if i have to actually code stuff for you then its a burdon 2014-03-09 14:37:29 but if i can say, you should look at catalyst, at this file, in this class or function, that's where XXX happens, fix it 2014-03-09 14:37:35 and you can fix it from there, then its help 2014-03-09 14:37:39 similarly with C 2014-03-09 14:37:41 or asm 2014-03-09 14:38:16 eg if you go to build some package in Gentoo, and it barfs on, say mkostmp() no symbol found, 2014-03-09 14:38:39 i can tell you what that function is (POSIX/non-POSIX GNU, etc) 2014-03-09 14:38:45 i can tell you were example code is 2014-03-09 14:38:59 if you can read that code and port to musl then that's good 2014-03-09 14:39:12 the lower level atomics might be more demanding 2014-03-09 14:39:32 but even if you know just x86 asm or x86_64 that helps 2014-03-09 14:40:03 barthalion, ^^^ up to you. i put it in the channel so others can read and think about it 2014-03-09 16:59:36 blueness: yeah, that was my another concert, I don't want to waste your time 2014-03-09 17:06:34 s/concert/concern/ 2014-03-10 02:03:34 vkrishn: several more updates to acf-core and acf-alpine-baselayout to make HTML views use htmlviewfunctions for consistency 2014-03-10 02:03:47 ACTION has not yet added the div around sections 2014-03-10 02:03:56 but, time to quit for tonight 2014-03-10 07:28:22 morning 2014-03-10 07:28:59 clandmeter: i think the algibot needs a little reconfig to work with mqtt 2014-03-10 07:29:23 i think its still trying to kick builders with zqm 2014-03-10 07:29:25 zmq 2014-03-10 09:18:42 ncopa: yeah i figured something was funky 2014-03-10 09:20:47 ncopa: any reason you didnt apply those openvswitch patches from ml? 2014-03-10 09:23:06 they didnt apply 2014-03-10 09:23:15 they need to be rebased 2014-03-10 09:23:22 and i simply didnt have time last week 2014-03-10 09:36:28 algitbot: build master 2014-03-10 09:38:03 algitbot: build master 2014-03-10 09:39:17 files from v2.4.11-207-g92f52c3 uploaded 2014-03-10 09:39:36 files from v2.5.4-260-ge8e76a7 uploaded 2014-03-10 09:39:45 algitbot: build 2.7-stable 2014-03-10 09:39:51 files from v2.5.4-260-ge8e76a7 uploaded 2014-03-10 09:40:19 clandmeter: will that do for now? 2014-03-10 09:41:21 i think so 2014-03-10 09:41:41 tbh, i think we should have a wiki page about it. 2014-03-10 09:44:08 btw, is there a reason not to use algitbot anymore for this? 2014-03-10 09:45:58 i need lua-mosquitto 2014-03-10 09:46:09 i think i posted a bug for lua5.2 support to that lib 2014-03-10 09:46:13 and it got stuck 2014-03-10 09:46:31 i could make a hack and call os.popen 2014-03-10 09:46:42 but popen from irc is risky business :) 2014-03-10 09:48:14 algitbot cannot exec mosquitto_pub? 2014-03-10 09:52:53 but popen from irc is risky business :) 2014-03-10 09:54:51 i could have a whitelist of supported branches (2.4-stable -> 2.7-stable + master) 2014-03-10 09:55:04 but probably better spend the time on fixing lua-mosquitto 2014-03-10 11:11:20 algitbot: build master 2014-03-10 11:11:49 algitbot: build 2.7-stable 2014-03-10 12:10:50 algitbot: build 2.7-stable 2014-03-10 12:11:08 algitbot: build master 2014-03-10 12:11:28 algitbot: build master 2014-03-10 12:14:45 algitbot: build master 2014-03-10 12:15:58 algitbot: build master 2014-03-10 12:17:28 Hello guys. Is there UPX for alpine linux, or any alternative ? 2014-03-10 12:21:37 zhivatma: dont think so, but should probably be simple to build upx 2014-03-10 12:22:22 I tried to build without success 2014-03-10 12:39:30 please git pull git://git.alpinelinux.org/fab/aports 2014-03-10 12:40:03 just a couple of updates and some new aports 2014-03-10 12:40:27 thanks 2014-03-10 13:14:30 Hi there, I sent my first patch on the mailing list earlier about ezstream package. But the attached file doesnt seem to show... What should I do? 2014-03-10 13:15:02 zhivatma: could you please file a package request on bugs.alpinelinux.org? 2014-03-10 13:16:03 ok doing it now. Thanks 2014-03-10 13:17:55 sinux: in general, I prefer that people use git send-email rather than attach them 2014-03-10 13:19:22 Sorry I tried but my email provider didnt want to let the email through from the alpine server. Should I try again using a different email address? 2014-03-10 13:19:46 thats good enough excuse :) 2014-03-10 13:20:13 i suppose you could also just sprunge it and paste the link here, when I'm here 2014-03-10 13:20:24 git format-patch --stdout -1 | sprunge 2014-03-10 13:20:43 the i can just 'curl $url | git am' 2014-03-10 13:23:11 i pushed it 2014-03-10 13:23:12 thanks 2014-03-10 13:23:22 lets see if it break anything :) 2014-03-10 13:23:28 thanks! 2014-03-10 13:23:46 :) it's my first package so it might ;) 2014-03-10 13:23:56 it looked okish 2014-03-10 13:25:06 let me know what I can do better for next time :) 2014-03-10 13:27:11 commit message 2014-03-10 13:27:51 sinux: look at your commit message compared to the others: http://git.alpinelinux.org/cgit/aports/log/ 2014-03-10 13:29:03 i also think we should set --build=$CBUILD \ 2014-03-10 13:29:03 --host=$CHOST \ 2014-03-10 13:29:19 and whitespace style 2014-03-10 13:29:36 other than that, pretty good to be first apkbuild 2014-03-10 13:29:41 :) 2014-03-10 13:30:32 thanks. ok I see so the commit message should be branch/package : message 2014-03-10 13:30:33 doesnt newapkbuild provide it? 2014-03-10 13:30:55 yes and yes 2014-03-10 13:31:13 as clandmeter points out, there is a script call newapkbuild 2014-03-10 13:31:23 that sets configure args 2014-03-10 13:31:39 but using newapkbuild is not a requirement 2014-03-10 13:32:08 ok I used it but just ran it "newapkbuild ezstream" 2014-03-10 13:33:10 then I removed what I thought wasnt necessary... I might have removed too much :) 2014-03-10 13:33:50 sinux: use -a Create autotools (use ./configure ...) 2014-03-10 13:34:19 or you can specify the url the tarbal which will fetch and check/add that automagically 2014-03-10 13:35:37 ok thanks. should I push a new patch with that? 2014-03-10 13:48:07 zhivatma: I'll take care of upx 2014-03-10 13:54:39 >>> ucl-dev*: Create ucl-dev-1.03-r0.apk 2014-03-10 13:54:39 awk: cmd. line:3: Division by zero 2014-03-10 13:54:42 huh 2014-03-10 13:58:35 hehe 2014-03-10 14:01:29 /usr/lib/gcc/i486-alpine-linux-musl/4.8.2/../../../../i486-alpine-linux-musl/bin/as: out of memory allocating 4072 bytes after a total of 0 bytes 2014-03-10 19:57:57 well, ok, screw upx for a while 2014-03-11 06:53:07 ncopa, i just pushed asterisk to edge and 2.7; i'm wondering what to do with 2.5 and 2.6 2014-03-11 06:53:15 as the security upgrade is only for latest release 2014-03-11 06:53:28 is the patch intrusive? 2014-03-11 06:53:30 should i try to find the patches; or just upgrade to latest 2014-03-11 06:53:41 i think i'd prefer backport the patch 2014-03-11 06:53:50 check if debian has backported patch 2014-03-11 06:54:19 i mean 2.7 is 11.8.x; 2.6 is 11.6.1 2014-03-11 06:54:24 11.x is already bug-fix only 2014-03-11 06:54:43 oh 2014-03-11 06:54:47 there's split out patches 2014-03-11 06:55:02 11.x.y 2014-03-11 06:55:07 x+1 sounds risky 2014-03-11 06:55:11 y+1 is ok 2014-03-11 06:55:22 in asterisk x+1 = bug fixes 2014-03-11 06:55:27 and y+1 = security fixes 2014-03-11 06:55:35 ok 2014-03-11 06:55:39 new features is major version bump 2014-03-11 06:55:48 but i found out the individual patches too 2014-03-11 06:55:50 then it should be ok to just upgrade 2014-03-11 06:56:00 if it does nto break anything 2014-03-11 06:56:48 v2.7 is in bugfix mode, while v2.4- 2.6 is in securityfix only 2014-03-11 06:57:12 that said if its simpler to just upgrade to a newer bugfix release 2014-03-11 06:57:16 then i am ok with that 2014-03-11 06:57:31 just try avoid surprises 2014-03-11 06:57:38 or changes that requires reconfiguration 2014-03-11 06:57:53 i think i'll just cherry-pick 2014-03-11 06:57:57 users should be able to apk upgrade withotu needing change anything 2014-03-11 06:57:59 and it should just work 2014-03-11 06:58:00 it'll encourage to upgrade to get the bug fix upgrade ;) 2014-03-11 06:58:06 yeah 2014-03-11 07:01:06 should those patches be URLs or files in git? 2014-03-11 07:01:20 asterisk ships links to patches in the security advisory 2014-03-11 07:02:54 mm 2014-03-11 07:22:48 i normally copy patches to aports git 2014-03-11 07:23:04 aaaand i just pushed it with the URLs 2014-03-11 07:23:06 :) 2014-03-11 07:23:23 np, as long the url dont break in 1 year 2014-03-11 07:23:51 i do it mostly because those patch urls broke or silently changed 2014-03-11 07:24:09 fabled: i wonder if we should implement error.h in musl 2014-03-11 07:24:20 its fairly trivally thing 2014-03-11 07:24:29 ncopa, yes, and i think rich was considering it to be added 2014-03-11 07:24:36 likely will not go in 1.0 2014-03-11 07:24:44 but could go in right after 1.0 2014-03-11 07:25:39 i think there are 10-15 packages atleast that uses it 2014-03-11 07:26:12 yes there is 2014-03-11 07:26:17 some have migrated away from using it 2014-03-11 07:26:18 some not 2014-03-11 07:26:23 it's still a GNU extension 2014-03-11 07:26:26 of course 2014-03-11 07:26:34 some also have configury magic to use wrappers if it's not there 2014-03-11 07:26:39 so it might effecting other packages too 2014-03-11 07:26:48 even if they build now, they'll build to be smaller next time 2014-03-11 07:27:09 i'm building moreutils 2014-03-11 07:27:25 #2358 2014-03-11 07:27:42 i wonder if i should just patch in an error.h in musl right away 2014-03-11 07:27:54 i'm ok to have it in 2014-03-11 07:27:56 ncopa: I pushed parallel to testing already 2014-03-11 07:28:01 oh 2014-03-11 07:28:03 apparently forgot to close the task 2014-03-11 07:28:10 or ref the issue 2014-03-11 07:28:17 I'm not very fond of moreutils 2014-03-11 07:28:30 ref #2358 in commit message 2014-03-11 07:28:51 i think in parallel the scary part is that it has 'gnu' in the name ;) 2014-03-11 07:29:13 Ah, I didn't know I need to use "ref" word. 2014-03-11 07:29:21 I just put issue number in commit msg. 2014-03-11 07:29:33 guess it didn't get parsed 2014-03-11 07:29:37 fabled: exactly :) 2014-03-11 07:29:42 anyway, it's 2402dd05b62054718038bba3cae9a408047dda1c 2014-03-11 07:29:52 barthalion: youre the man! 2014-03-11 07:29:54 thanks! 2014-03-11 07:30:39 np :) 2014-03-11 07:34:16 fabian_a: http://pkgs.fedoraproject.org/cgit/wavemon.git/tree/wavemon.spec 2014-03-11 07:34:19 download url is broke 2014-03-11 10:21:36 i think i will upgrade bld1.a.o and bld2.a.o 2014-03-11 10:26:52 fabled: possible apk issue on bld1 2014-03-11 10:27:03 bld1:~# apk version 2014-03-11 10:27:03 Installed: Available: 2014-03-11 10:27:03 curl-7.33.0-r0 < 7.33.0-r2 2014-03-11 10:27:17 bld1:~# apk upgrade -v 2014-03-11 10:27:17 OK: 124 packages, 359 dirs, 5904 files, 89 MiB 2014-03-11 10:27:23 (it does not get upgraded) 2014-03-11 10:27:38 bld1:~# apk add -u --simulate curl 2014-03-11 10:27:38 (1/3) Installing run-parts (4.4-r0) 2014-03-11 10:27:38 (2/3) Installing ca-certificates (20130906-r2) 2014-03-11 10:27:38 OK: 89 MiB in 124 packages 2014-03-11 10:27:38 (3/3) Upgrading curl (7.33.0-r0 -> 7.33.0-r2) 2014-03-11 10:27:51 why does not apk upgrade, upgrade it? 2014-03-11 10:56:29 weird 2014-03-11 10:56:37 some dependency blocks it? 2014-03-11 10:56:47 simulate should see it though 2014-03-11 10:57:01 maybe versioned world dependency? 2014-03-11 10:57:03 apk ugprade -a ? 2014-03-11 10:57:26 yeah 2014-03-11 10:57:39 use apk upgrade -a 2014-03-11 10:57:52 i think you have upgraded from v2.6 -> v2.7 2014-03-11 10:57:57 but not with upgrade -a 2014-03-11 10:58:03 so some packages are still from v2.6 2014-03-11 10:58:10 since they have identical version string 2014-03-11 11:00:29 though, it's still does not want to upgrade curl in upgrade -s -a 2014-03-11 11:00:32 interesting 2014-03-11 12:05:23 apk add -u -s curl seems to do it though 2014-03-11 12:06:06 ok, I'm gonna apk upgrade -U -a 2014-03-11 12:06:17 to replace the packages with same versions 2014-03-11 12:06:22 and see what happens after that 2014-03-11 12:07:06 same thing 2014-03-11 12:12:01 fabled: do you want to investigate it further? or I just apk add -u curl? 2014-03-11 12:12:09 it could be that it does not upgrade due to it pulls in new deps? 2014-03-11 12:12:17 ncopa, i want to investigate 2014-03-11 12:12:20 i'm on it 2014-03-11 12:12:25 (in few minutes) 2014-03-11 12:45:43 fabian_a: http://pkgs.fedoraproject.org/cgit/wavemon.git/tree/wavemon.spec 2014-03-11 12:45:43 download url is broke 2014-03-11 12:45:56 i pushed your changes too 2014-03-11 12:45:57 thanks! 2014-03-11 12:53:13 fabled: how is it goin? i'd like to reboot the bld1 today (and upgrade bld2 too) 2014-03-11 12:57:28 ncopa, sorry, i was interrupted with something. i'm on it again. 2014-03-11 12:59:29 ncopa: thanks 2014-03-11 13:09:38 ncopa: is it normal for an lxc container not to be able to see processor load in (h)top? 2014-03-11 13:27:19 ncopa, found the reason 2014-03-11 13:27:32 curl is installed, but not pulled in by anything in world currently 2014-03-11 13:27:39 so upgrade will not consider it 2014-03-11 13:27:47 but for some reason it did not remove it either 2014-03-11 13:27:50 it should have deleted it 2014-03-11 13:28:01 if i go and add curl to world with vi 2014-03-11 13:28:03 upgrade works 2014-03-11 13:43:24 ok 2014-03-11 13:43:26 makes sense 2014-03-11 13:43:48 a bit weird though 2014-03-11 13:43:59 and if we apk del curl? 2014-03-11 13:44:09 i just goes away I suppose 2014-03-11 13:44:33 # apk del --simulate curl 2014-03-11 13:44:33 OK: 202 MiB in 144 packages 2014-03-11 13:44:35 it does not 2014-03-11 13:45:32 ok, I'll apk add curl and delete it afterwards 2014-03-11 13:46:01 hm 2014-03-11 13:46:04 apk del curl does not work either 2014-03-11 13:46:12 strange 2014-03-11 13:46:56 seems liek apk add curl doesnot add it to world either 2014-03-11 13:48:44 weird 2014-03-11 13:48:53 seems like something is holding curl back 2014-03-11 14:28:22 same thing appears to happen with sqlite-libs on bld2 now too 2014-03-11 14:32:42 ok. that's a bug then 2014-03-11 14:32:50 it might miss to remove packages 2014-03-11 14:34:02 i'll test on bld1 2014-03-11 15:14:11 ncopa, i think i found the underlying bug 2014-03-11 15:14:21 install_if handling 2014-03-11 15:20:30 aha 2014-03-11 15:20:43 i suppose we should try add it to test suite? 2014-03-11 16:37:46 I wonder if someone uses 2.4 yet 2014-03-11 16:37:48 or 2.5 2014-03-11 16:38:37 files from v2.4.11-208-gd971b05 uploaded 2014-03-11 16:39:04 ncopa: seems that backporting fixes is way harder than bumping packages :) 2014-03-11 16:40:20 files from v2.5.4-262-g776792b uploaded 2014-03-11 16:40:28 ncopa, i think the above fix to install_if handling fixes the bld* issues; probably also the 'Huh? Error reporter did not find the broken constraints.' errors we saw earlier 2014-03-11 16:40:39 files from v2.5.4-262-g776792b uploaded 2014-03-11 16:40:44 i have one more commit coming up early tomorrow 2014-03-11 16:40:48 will tag new apk-tools after that 2014-03-12 12:51:43 new repository? 2014-03-12 12:51:53 yeah 2014-03-12 12:51:58 nice 2014-03-12 12:53:12 algitbot: build 2.7-stable 2014-03-12 12:54:22 algitbot: build 2.7-stable 2014-03-12 12:54:41 algitbot: build 2.7-stable 2014-03-13 07:14:08 files from v2.5.4-264-g0663abd uploaded 2014-03-13 07:14:09 files from v2.5.4-264-g0663abd uploaded 2014-03-13 07:14:10 files from v2.4.11-209-g67b9547 uploaded 2014-03-13 09:26:31 barthalion: thanks for backporting the sec fixes! 2014-03-13 09:26:41 again, youre the man! 2014-03-13 09:51:20 just to let you know: 2014-03-13 09:51:42 we will be at http://chemnitzer.linux-tage.de/2014/en/info/ as every year 2014-03-13 09:52:25 with http://www.eisfair.org/ and http://www.fli4l.de/ 2014-03-13 09:52:45 but this year will show eisfair-ng which is based on alpinelinux :-) 2014-03-13 09:53:24 see https://ssl.nettworks.org/wiki/display/eng/eisfair-ng+Wiki 2014-03-13 09:53:46 files from v2.5.4-266-g183b925 uploaded 2014-03-13 09:54:08 files from v2.5.4-266-g183b925 uploaded 2014-03-13 10:03:03 files from v2.4.11-211-g133948c uploaded 2014-03-13 10:57:41 files from v2.5.4-267-g553b4e0 uploaded 2014-03-13 10:59:39 files from v2.5.4-267-g553b4e0 uploaded 2014-03-13 12:09:12 i should migrate build-2-5 and build64-2-5 to new build servers 2014-03-13 12:09:22 and 2-4 builders too 2014-03-13 12:22:42 I moved build64-2-5 to bld2.a.o 2014-03-13 12:23:00 moving vserver to lxc is fairly straightforward 2014-03-13 12:23:38 i just create new guest, backup the inittab, rsync files (with --numeric-ids) 2014-03-13 12:23:47 configure /etc/network/interfaces 2014-03-13 12:23:51 and done! 2014-03-13 13:01:08 I wonder if I should set up gitolite for git.a.o 2014-03-13 13:01:19 it is a bit of a project... 2014-03-13 13:01:32 gitolite is neat, but do we really need it? 2014-03-13 13:01:40 sometimes devs need to put something in /archive 2014-03-13 13:01:46 oh, true 2014-03-13 13:02:27 i suppose we could start mess with ftp for that, but scp is handy 2014-03-13 13:03:00 ftp is mess and making sftp only account isn't much different from current setup 2014-03-13 13:03:10 yeah 2014-03-13 13:03:49 build-2-4 and build-2-5, x86 and x86_64 are migrated to bld1 and bld2 servers 2014-03-13 13:04:05 the only 2 left are git.a.o and distfiles.a.o 2014-03-13 13:04:14 then the old server can be shut down 2014-03-13 13:04:47 so i was thinking of maybe replace git.a.o with a x86_64 container while at it 2014-03-13 13:05:17 distfiles keeps all the source tarballs we used? 2014-03-13 13:05:26 thats the idea yes 2014-03-13 13:05:48 i think distfiles.a.o will need some work since it is now 2 different servers 2014-03-13 13:05:54 i am thinking of setting up nfs 2014-03-13 13:06:24 and let all share the distfiles dir 2014-03-13 13:06:40 then i need set up some cleanup script 2014-03-13 13:07:01 that looks over our current releases (stable branches) 2014-03-13 13:07:11 mark those as 'keep' 2014-03-13 13:07:27 and then delete everything older than 6 months or so 2014-03-13 13:07:42 makring current git master as 'keep' too 2014-03-13 13:12:27 ok 2014-03-13 13:12:34 one benefit with gitolite 2014-03-13 13:12:57 is that we could have lower bar on who gets a git repo 2014-03-13 13:13:34 also we could give some git repo without necessarily give archive access 2014-03-13 13:14:01 we could also give some ppl git master push access but not push access to stable branches 2014-03-13 13:14:25 i believe we will want gitolite in the long run 2014-03-13 13:14:28 but not today :) 2014-03-13 13:14:37 the last one sounds interesting 2014-03-13 13:14:54 yes 2014-03-13 13:15:05 we will have to think of what to do with archive though 2014-03-13 13:15:08 actually 2014-03-13 13:15:14 archive.a.o could be a separate server 2014-03-13 13:15:16 almost open edge development, let's say after 4-5 accepted contribution from IRC or mailing list we'd give access to edge 2014-03-13 13:15:26 with some guidelines how to format commit messages 2014-03-13 13:15:37 something like that yes 2014-03-13 13:15:41 and to actually test the builds before pushing… 2014-03-13 13:16:32 i am also thinking of doing what arch does with a core repo and a user built repo 2014-03-13 13:16:43 aug or waht they call it 2014-03-13 13:16:55 yaort 2014-03-13 13:17:10 community repo 2014-03-13 13:17:21 yeah, we have devs and TUs + AUR 2014-03-13 13:17:29 AUR 2014-03-13 13:17:35 where you have $x helpers for AUR to fetch all deps and build them 2014-03-13 13:17:50 right 2014-03-13 13:18:14 i think AUR is one of the things that made arch linux success 2014-03-13 13:18:18 we should review stuff more carefully, AUR is a mess nowadays 2014-03-13 13:18:26 oh ok :) 2014-03-13 13:18:31 lol 2014-03-13 13:19:43 the main advantage and disadvantage of AUR is that anyone can send a PKGBUILD here 2014-03-13 13:20:01 i just feel that maintaining our current 'main' is on the limit of what I/we can handle 2014-03-13 13:20:05 yeah 2014-03-13 13:20:07 i figured 2014-03-13 13:20:35 it doesnt really solve the real problem when i think of it 2014-03-13 13:20:55 the real problem is that there is not enough skilled ppl to deal with the packages 2014-03-13 13:21:15 many want contribute 2014-03-13 13:21:35 but many lacks the needed skills to be able to contribute with high quality stuff 2014-03-13 13:21:51 and shipping low quality will bite us later 2014-03-13 13:22:06 because it will create more maintenance work 2014-03-13 13:22:46 otoh, much of the maintenace work is not too difficult 2014-03-13 13:22:50 abump pkg-ver 2014-03-13 13:22:56 hm 2014-03-13 13:27:38 switching to musl will give you extra headaches 2014-03-13 13:27:43 ok we will have to think of it 2014-03-13 13:27:43 yes 2014-03-13 13:27:47 well 2014-03-13 13:27:49 thats stuff i cannot fix 2014-03-13 13:28:07 ones i need to fireup gdb, im lost 2014-03-13 13:28:33 yeah it gives us a little more work 2014-03-13 13:28:47 i think it will pay back in the long run 2014-03-13 13:28:50 it is a better libc 2014-03-13 13:29:03 for as long it will be maintainted. 2014-03-13 13:29:05 there are some bugs still 2014-03-13 13:29:19 musl is better maintained than uclibc 2014-03-13 13:29:43 one of the reasons we want switch 2014-03-13 13:29:54 sure, but isnt most of the work beeing done by dalias? 2014-03-13 13:30:25 nsz too 2014-03-13 13:30:44 they will drop a marketing bomb with 1.0 2014-03-13 13:30:57 and musl definitively have potensial 2014-03-13 13:31:20 i wouldnt be surprised if it will take over glibc in 5 years 2014-03-13 13:32:08 it probably won't, but i wouldnt be surprised if it did 2014-03-13 13:32:31 currently, supporting both uclibc and musl is a bit of a pain 2014-03-13 13:32:45 once we drop uclibc things will get a bit better i think 2014-03-13 13:33:09 also, once we get the current issues fixed in upstream, things gets better 2014-03-13 13:33:17 uclibc has been quite stable recently. atleast i didnt see many patches needed in aports. 2014-03-13 13:33:32 it has been stable because it has been stuck 2014-03-13 13:34:06 look at the amount of patches we need to get it where it is now 2014-03-13 13:34:15 from a developer pov, what are musl selling points? 2014-03-13 13:34:31 strictness 2014-03-13 13:34:34 clean code 2014-03-13 13:34:47 small/fast 2014-03-13 13:35:14 do many projects care about those 3 points? 2014-03-13 13:35:31 im not trying to be nagative, just curious 2014-03-13 13:35:36 some does 2014-03-13 13:35:48 the rest doesn't build 2014-03-13 13:35:49 ;D 2014-03-13 13:35:57 musl is gaining "market" from uclibc land i suppose 2014-03-13 13:35:58 oh 2014-03-13 13:36:00 one thing more 2014-03-13 13:36:06 the license 2014-03-13 13:36:14 mit/bsd like license 2014-03-13 13:36:23 that counts many places 2014-03-13 13:36:26 commercially 2014-03-13 13:37:28 clandmeter: the issue is that basically 2014-03-13 13:37:36 clandmeter: we have to maintain uclibc ourselves 2014-03-13 13:38:00 clandmeter: while that may make things slightly easier for you, everytime uclibc breaks, it's usually me, fabled and ncopa trying to put the world back together again 2014-03-13 13:38:47 with musl, we can say "hey dalias, there is a problem" and that reduces the amount of burden we have for maintenance significantly 2014-03-13 13:38:54 i think we should drop uclibc relatively soon... 2014-03-13 13:39:06 big question is how we do it 2014-03-13 13:39:18 i'd like that edge is musl 2014-03-13 13:39:29 well we need to find a way to safely transition 2014-03-13 13:39:40 yes 2014-03-13 13:39:44 right now if you go from edge to edge-musl, you trash your machine 2014-03-13 13:39:51 have you tried? 2014-03-13 13:39:55 yeah 2014-03-13 13:40:00 apk-tools breaks early in the process 2014-03-13 13:40:09 ah 2014-03-13 13:40:11 hm 2014-03-13 13:40:18 apk add apk-tools-static 2014-03-13 13:40:19 even if its static? 2014-03-13 13:40:30 2014-03-13 13:40:47 apk add -U -u apk-tools 2014-03-13 13:40:54 or apk fix apk-tools 2014-03-13 13:40:59 and then: apk upgrade -U -a 2014-03-13 13:41:03 something like that 2014-03-13 13:41:06 anyway 2014-03-13 13:41:13 http://mirrors.centarra.com/alpine/ 2014-03-13 13:41:16 ^^^ 10gbit 2014-03-13 13:41:19 also anycast 2014-03-13 13:41:27 cool! 2014-03-13 13:41:40 although there's only one node atm 2014-03-13 13:41:58 feel free to add it to alpine-mirrors package 2014-03-13 13:42:12 that's also running on an alpine system 2014-03-13 13:42:18 cool :) 2014-03-13 13:43:33 maybe we should cnam dl-6.alpinelinux.org to it? 2014-03-13 13:43:36 cname* 2014-03-13 13:44:06 remind me, we should probably start think of layering the mirrors 2014-03-13 13:44:09 tier1 mirrors 2014-03-13 13:44:13 and tier2 mirrors 2014-03-13 13:44:32 if you would like yeah 2014-03-13 13:44:37 where tier 1 mirrors gets synced with push notifications 2014-03-13 13:44:39 once we're done with this 2014-03-13 13:44:45 via mqtt 2014-03-13 13:44:47 there will be an anycast node 2014-03-13 13:45:03 in UK, three in USA, one in Australia 2014-03-13 13:45:50 nice 2014-03-13 13:45:57 re musl 2014-03-13 13:46:08 the current blocker is mkinitfs 2014-03-13 13:46:11 i need fix it 2014-03-13 13:46:26 to figure out the dynamic link deps 2014-03-13 13:46:40 so we dont need maintain the static list like we do now 2014-03-13 13:47:04 once that is in place were ready for migrating edge to musl 2014-03-13 13:47:29 i'll fix xen this weekend 2014-03-13 13:47:40 hopefully we can do so within a week or so 2014-03-13 13:52:45 i will shut down git.a.o for a few minutes. 15-mins or so 2014-03-13 14:00:06 and it appears that git.a.o is back 2014-03-13 14:00:25 better and faster than ever :) 2014-03-13 14:02:29 i will reboot bld1 2014-03-13 14:02:40 so all -x86 builders and git.a.o will be down 2014-03-13 14:05:19 ok rebooting now... 2014-03-13 14:18:03 kaniini: do you have experience with bridge over bond interfaces? 2014-03-13 14:18:28 lets say you have a virtual host with N nics 2014-03-13 14:18:33 you put them in a bond0 2014-03-13 14:18:40 then you bridge the network to the guests 2014-03-13 14:19:31 as i understand the only way to make that work is if you either use active-backup bonding mode or configure switch ports as a link aggregation (aka trunk) 2014-03-13 14:19:43 and/or with LACP 2014-03-13 14:20:16 but trunk config of switch does not really work when you connect nics to different switches for failoveer 2014-03-13 14:20:41 so now i use active-backup bonding mode 2014-03-13 14:21:31 is there some way to set it up so you get both performance gain of using both NICs for the guests *and* configure multiple switches for failover? 2014-03-13 14:22:05 the bridge code does not forward the PDUs 2014-03-13 14:22:21 we had to give up on using LACP with bridge, instead using a multipathing scheme 2014-03-13 14:41:26 ok, then its not just me :) 2014-03-13 14:41:47 i wonder if openvswitch can handle it properly 2014-03-13 14:42:03 if you configure the bond and bridge in ovs 2014-03-13 14:44:41 i will reboot bld2 now 2014-03-13 15:46:47 hm 2014-03-13 16:02:07 seems like git commits works again 2014-03-13 16:02:17 i'm going home 2014-03-14 06:08:10 morning 2014-03-14 07:07:05 morning 2014-03-14 10:41:03 algitbot: build master 2014-03-14 10:41:07 yeah, thanks 2014-03-14 10:45:01 1 means you are first in line? 2014-03-14 10:46:41 it means I'm the number one here, of course :p 2014-03-14 10:50:17 ncopa: we can use irc still or need to use msg.a.o? 2014-03-14 10:51:02 1 is the return code from lua-mosquitto's send 2014-03-14 10:51:12 algitbot sneds via msg.a.o now 2014-03-14 10:51:16 sends* 2014-03-14 10:51:28 the 1 means that messages was sent 2014-03-14 10:52:21 build-edge-x86 is missing in mosqtt 2014-03-14 10:52:39 https://paste.xinu.at/jR6/ 2014-03-14 10:54:29 seems like build-edge-x86 didnt start up at all after yesterdays reboot 2014-03-14 10:59:23 yeah 2014-03-14 10:59:26 networking was down 2014-03-14 10:59:47 there is something with networking not starting up proerply in lxc on reboots 2014-03-14 14:32:53 ncopa, noticed nodejs and libuv being moved to main 2014-03-14 14:33:25 interesting that I got build msg.a.o of nodejs building before libuv ? 2014-03-14 14:38:04 http://mpm-itk.sesse.net/ - has now patches for apache v2.4.7 2014-03-14 14:38:27 i pushed nodejs before libuv (my mistake) 2014-03-14 14:38:42 i realized it was bad, and pushed libuv afterwards 2014-03-14 14:39:26 but how come the build msg went ok 2014-03-14 14:40:19 though it seems ok for v2.7 builders 2014-03-14 14:42:27 can nginx spdy be somehow enabled and used on AL infra, maybe b.a.o ? 2014-03-14 14:42:46 isnt it? 2014-03-14 14:42:55 or wiki.a.o 2014-03-14 14:43:02 i mean in apkbuild 2014-03-14 14:43:41 oh... its enabled 2014-03-14 14:44:07 nope, I mean use it as installed 2014-03-14 14:44:24 newer browsers should be able to use it 2014-03-14 14:44:40 we use a reverse proxy 2014-03-14 14:44:44 which is based on ngnix 2014-03-14 14:44:49 thought so 2014-03-14 14:45:07 should it be enabled on the revproxy? 2014-03-14 14:46:01 have not tried it in proxy settings, or read about it 2014-03-14 14:46:38 would try to find more 2014-03-14 14:46:38 im not up2date on that proto 2014-03-14 14:47:07 we use lighty on the guests 2014-03-14 14:48:09 lighty has spdy implemented ? 2014-03-14 14:49:06 http://redmine.lighttpd.net/boards/3/topics/5201 2014-03-14 14:51:31 so need AL infra needs to wait 2014-03-14 14:56:09 http://www.mozilla.org/en-US/firefox/27.0.1/releasenotes/ - now has spdy 3.1 2014-03-14 14:56:53 vkrishn: is it only for ssl? 2014-03-14 15:01:33 From earlier readings I got confused at it states ssl as mandatory, now I am not sure from protocol point of view 2014-03-14 15:01:33 or implementation pov 2014-03-14 15:02:16 or does it always go over port 443? 2014-03-14 15:05:16 http://dev.chromium.org/spdy/spdy-protocol 2014-03-14 15:07:13 vkrishn: https://www.varnish-software.com/blog/why-i-dont-spdy 2014-03-14 15:07:56 I only like the multiplex part of the protocol 2014-03-14 15:10:53 or maybe use http://www.modpagespeed.com/ 2014-03-14 15:10:55 :) 2014-03-14 15:12:51 I tried mod_pagespeed on knoppix, at some point I disabled it 2014-03-14 15:14:37 or apache/lighttpd mod_concat 2014-03-14 15:32:51 SPDY only replaces the way the data is written to the network 2014-03-14 15:33:37 still don't know the ssl mandatory 2014-03-14 15:33:37 http://dev.chromium.org/spdy/spdy-protocol/spdy-protocol-draft3-1 2014-03-14 15:33:37 gtg, dinner ready 2014-03-14 15:38:38 i am powering off old build.alpinelinux.org 2014-03-14 15:38:49 which build alpine-1.x for us 2014-03-14 15:39:07 and alpine 2.0-2.6 2014-03-17 08:29:33 I've tested upgrading from 2.7 to edge musl 2014-03-17 08:29:43 mkinitfs is still not fixed 2014-03-17 08:29:52 ncopa: should I write a patch like for eglibc? 2014-03-17 08:29:59 is anyone running ntpd from busybox instead of openntpd? I was thinking of making an initscript for busybox-initscripts but since openntpd owns /etc/init.d/ntpd I can't see another solution than naming the init script /etc/init.d/busybox-ntpd (like gentoo does) 2014-03-17 08:31:56 barthalion: mkinitfs is on my list 2014-03-17 08:32:10 i should try fix mkinitfs today 2014-03-17 08:32:35 uggedal: very good question 2014-03-17 08:32:37 uggedal: it may be a good idea to create a split package 2014-03-17 08:32:45 uggedal: busybox-initscript-ntpd 2014-03-17 08:32:58 barthalion: that would work 2014-03-17 08:33:04 that is a possibility but feels like overkill 2014-03-17 08:33:22 I wouldn't say so 2014-03-17 08:33:35 user can still install busybox-initscripts and use any other ntp 2014-03-17 08:33:40 what about the other scripts? 2014-03-17 08:33:56 like watchdog, bb syslog, and all those 2014-03-17 08:34:09 does other scripts conflict with any package in repo? 2014-03-17 08:34:30 potensially yes 2014-03-17 08:34:36 but i dont know if they do reall 2014-03-17 08:34:48 I'm not that fond of running openntpd since the portable version has been unmaintained for ages. busybox's ntpd is based on openntpd but is maintained 2014-03-17 08:35:03 i suppose the ideal solution would be rename openntpd's init script to 'openntpd' 2014-03-17 08:35:08 yeah 2014-03-17 08:35:12 and call bb script ntpd 2014-03-17 08:35:42 that would break things for people currently running openntpd 2014-03-17 08:36:01 otoh, lots of breakage might be expected for the upcomong 3.0 release 2014-03-17 08:36:10 ncopa: about upgrading to musl… apk.static upgrade -U -a --allow-untrusted works fine 2014-03-17 08:36:23 great!!! 2014-03-17 08:36:26 there is one conflict 2014-03-17 08:36:30 uclibc-tools and musl-tools 2014-03-17 08:36:36 but it's easily fixed by apk fix afterwards 2014-03-17 08:36:50 i suppose we will have to do 'replaces' 2014-03-17 08:37:32 if we break a lot in 3.0, we really should make an announcement with info about possible incompatibilites 2014-03-17 08:37:35 so what do you guys think? we rename openntpd's script to 'openntpd' and create a bb ntpd script called 'ntpd'? 2014-03-17 08:37:39 yes 2014-03-17 08:37:47 we should make a list of incompatibilities 2014-03-17 08:37:58 yes, let's rename it your way 2014-03-17 08:38:10 i am not 100% comfortable iwth it either 2014-03-17 08:38:11 and while we're on breaking… let's switch to mariadb now :D 2014-03-17 08:38:46 and Qt 5.x ? :) 2014-03-17 08:39:06 reminds, i need to figure out how to do Qt for raspberry pi 2014-03-17 08:39:11 it needs separate build 2014-03-17 08:39:19 uggedal: what do you think about ntpd scripts name? 2014-03-17 08:39:28 just need to figure out how to minimize packages, and make sure the packages work nicely together 2014-03-17 08:39:48 ncopa: I think the best solution is renaming openntpd 2014-03-17 08:39:58 ok, good 2014-03-17 08:40:02 lets do that then 2014-03-17 08:40:12 i dont like breaking things for users 2014-03-17 08:40:26 since people probably are running syslog etc from bb I would guess bb ntpd would work better for most people 2014-03-17 08:40:27 but i think rename openntpd is what we really want 2014-03-17 08:40:34 I wonder if it'll break anything 2014-03-17 08:40:49 if busybox ntpd is a fork and provides same parameters… 2014-03-17 08:40:52 good question 2014-03-17 08:41:03 i just assumed it will break things :) 2014-03-17 08:41:20 yes, Murphy's law 2014-03-17 08:41:32 lets rename openntpd and make the best of it from there 2014-03-17 08:42:36 looks like bb ntpd has different flags and functionality 2014-03-17 08:42:49 -n vs -d for running in foreground 2014-03-17 08:43:07 anything that can go wrong, will go wrong 2014-03-17 08:43:28 bb has -q for quiting after clock is set while openntpd -s for hard-setting time at startup while still running afterwards 2014-03-17 08:44:53 and bb has no config file 2014-03-17 08:51:39 i wonder if we can do soemthing with pre/post upgrade scripts 2014-03-17 08:51:54 migrate the config from /etc/conf.d/ntpd to /etc/conf.d/openntpd 2014-03-17 08:52:04 and maybe echo some warning 2014-03-17 08:52:29 alpine-conf would need to be patched as well btw 2014-03-17 08:52:43 oh. true 2014-03-17 08:53:02 such magic is never a good solution… 2014-03-17 08:53:14 there is *always* some case you accidentally skip 2014-03-17 08:53:16 and then bam 2014-03-17 08:53:45 warning is fine, but if we write about it in announcement, it should be enough 2014-03-17 08:54:00 it sounds insane to upgrade to new release without looking at changelog first 2014-03-17 08:54:26 and people running edge should expect thhis kind of things 2014-03-17 08:54:31 exactly 2014-03-17 08:54:43 2.7 -> 3.0 2014-03-17 08:54:56 you *should* read release notes :) 2014-03-17 08:55:07 I don't see why someone would use edge without sitting here 2014-03-17 08:55:38 you never know 2014-03-17 09:14:44 ok 2014-03-17 09:14:48 i think pre-upgrade 2014-03-17 09:15:29 if /etc/conf.d/ntpd exists, then rename it to /etc/conf.d/openntpd 2014-03-17 09:16:13 in post-upgrade 2014-03-17 09:17:08 if /run/openrc/started/ntpd exists and is a symlink that points to /etc/init.d/ntpd, then rename it to openntpd 2014-03-17 09:17:23 that way the shutdown will be clean 2014-03-17 09:18:26 for i in /etc/runlevel/*/ntpd; if $i is symlink that points to /etc/init.d/ntpd, then rename 2014-03-17 09:18:44 i think that should handle all possible cases 2014-03-17 09:19:23 it should handle all relevant cases at least 2014-03-17 09:20:10 if you are switching runlevel and/or are starting/stopping services while you do the apk upgrade, then things might go unexpected 2014-03-17 09:26:10 should openntpd still use /etc/ntpd.conf and /usr/sbin/ntpd? 2014-03-17 09:35:21 uggedal: i think so yes 2014-03-17 09:38:46 ncopa: cool. I have patches locally for openntpd (minus the init/confd migration part), alpine-conf and busybox-initscripts 2014-03-17 09:39:03 very well 2014-03-17 09:40:05 i could try make the upgrade scripts 2014-03-17 09:40:12 wil you be able to test those? 2014-03-17 09:41:49 untested openntpd.pre-upgrade: http://sprunge.us/VSHF 2014-03-17 09:42:30 woohoo, hplip has been built 2014-03-17 09:43:06 arg!!!!!!!!!!! 2014-03-17 09:43:21 i think i am the new checksumerror champing 2014-03-17 09:43:26 champion 2014-03-17 09:43:50 ncopa: yes, I can test them. at work atm so could take some time 2014-03-17 09:44:34 ncopa: should the default for alpine-conf still be chrony? 2014-03-17 09:44:53 i think so yes 2014-03-17 09:45:01 ncopa: and should we call the busybox ntpd 'ntpd' or 'busybox' in alpine-conf? 2014-03-17 09:45:17 or 'busybox-ntpd' 2014-03-17 09:45:35 maybe busybox? 2014-03-17 09:45:45 yep, that was what I went with 2014-03-17 09:45:52 good :) 2014-03-17 09:46:42 failed to build v2.5.4-268-g8fe86ba. http://distfiles.alpinelinux.org/distfiles/build-2-5-x86_64.txt 2014-03-17 09:46:54 oops 2014-03-17 09:47:06 2.5? 2014-03-17 09:47:13 yep 2014-03-17 09:47:35 ok 2014-03-17 09:48:03 still getting this error 2014-03-17 09:48:04 remote: /usr/local/bin/algitbot.sh: line 67: can't create /tmp/algitbit.out: Permission denied 2014-03-17 09:48:29 thats just debug thingy 2014-03-17 09:48:32 i'll remove it 2014-03-17 09:48:41 in a sec 2014-03-17 09:48:46 k, thanks 2014-03-17 09:48:47 nothing to worry about 2014-03-17 09:50:02 404? 2014-03-17 09:51:02 try bld2 2014-03-17 09:51:11 i need fix config there too 2014-03-17 09:51:26 try: http://bld2.alpinelinux.org/buildlogs/build-2-7-x86_64.v2.7.5-25-g72cac4b6.log 2014-03-17 09:51:54 sigh, will fix it in a moment 2014-03-17 09:51:59 /lib/libpci.a(init.o): relocation R_X86_64_PC32 against symbol `pci_malloc' can not be used when making a shared object; recompile with -fPIC 2014-03-17 09:52:17 ncdev-edge-musl-x86:~/aports/main/openntpd$ sprunge < openntpd.pre-upgrade 2014-03-17 09:52:17 http://sprunge.us/XfPT 2014-03-17 09:52:17 ncdev-edge-musl-x86:~/aports/main/openntpd$ sprunge < openntpd.post-upgrade 2014-03-17 09:52:17 http://sprunge.us/QFNL 2014-03-17 09:52:23 uggedal: ^^^ 2014-03-17 09:52:34 i think they shoudl work 2014-03-17 09:52:55 algitbot: build 2.6-stable 2014-03-17 09:53:00 guess it will fail too 2014-03-17 09:53:01 curl http://sprunge.us/XfPT > openntpd.pre-upgrade 2014-03-17 09:53:19 curl http://sprunge.us/QFNL > openntpd.post-upgrade 2014-03-17 09:56:33 build-2-7-x86_64 config shoudl be fixed now 2014-03-17 09:56:40 for error log url 2014-03-17 09:59:39 ncopa: is there a way to force openntpd to be upgraded before busybox-initscripts? 2014-03-17 10:00:30 if not people would have to: apk upgrade && apk fix busybox-initscipts if busybox-initscript was upgraded before openntpd 2014-03-17 10:01:26 fabled: ^^^ 2014-03-17 10:01:29 i think not 2014-03-17 10:01:45 but i think we can do: replaces='openntpd' 2014-03-17 10:01:52 in busybox-initscripts 2014-03-17 10:02:09 i *think* ti should not overwrite /etc/conf.d/ntpd if it is modified 2014-03-17 10:24:34 seems like net-snmpd was difficult 2014-03-17 10:25:58 sigh, I need to setup amd64 builder… 2014-03-17 10:26:23 I'll fix it in a couple of hours, I'm a bit short of time now 2014-03-17 10:26:44 ok, np 2014-03-17 10:28:53 it could be a parallel build issue 2014-03-17 10:29:09 hm 2014-03-17 10:29:09 no 2014-03-17 10:29:14 we already do make -j1 2014-03-17 10:33:16 ncopa: replaces results in /etc/conf.d/ntpd.apk-new but no /etc/conf.d/ntpd 2014-03-17 10:33:47 but the migration parts of openntpd worked 2014-03-17 10:34:06 even with changes to /etc/conf.d/ntpd 2014-03-17 10:34:32 ok 2014-03-17 10:34:37 good 2014-03-17 10:34:42 i think thats all acceptable 2014-03-17 10:34:55 i don think we bothere try fixup the ntpd.apk-new 2014-03-17 10:35:26 because if you have it, you likely are already running openntpd 2014-03-17 10:35:31 and will continue to do so 2014-03-17 10:37:14 so i think it is acceptable 2014-03-17 10:37:45 i suppose only problem is if you have openntpd installed, but not yet configured, do apk upgrade 2014-03-17 10:37:52 and intend to continue with bb ntpd 2014-03-17 10:38:13 in which case you'll have to rename ntpd.apk-new 2014-03-17 10:38:43 woudl it be possible to have sensible defaults in init.d so that it will run without an existing conf.d/ntpd? 2014-03-17 10:40:52 ncopa: good idea. something like: command_args="${NTPD_OPTS:--N -p pool.ntp.org}" ? 2014-03-17 10:41:08 yes 2014-03-17 10:41:17 looks good 2014-03-17 10:43:14 lunch break now. I'll look over the patches and test them again before sending them 2014-03-17 10:43:33 great! bon apetit :) 2014-03-17 10:49:41 I don't understand what's wrong 2014-03-17 10:49:52 I've the same build environment here 2014-03-17 10:49:59 the only difference is it's slower 2014-03-17 10:50:04 but anyway it builds locally 2014-03-17 10:50:05 :/ 2014-03-17 10:50:29 it is indeed weird 2014-03-17 10:50:36 it builds here too, in musl libc 2014-03-17 10:50:54 I can have a look at it after lunch 2014-03-17 10:51:27 it apparently build on 2.4, 2.5 and 2.6 too 2014-03-17 10:52:07 sounds like a problem with toolchain or similar 2014-03-17 11:15:12 ncopa: if /etc/conf.d/ntpd (old openntpd) has no changes before upgrade /etc/conf.d/openntpd will be the one from busybox-initscript and /etc/conf.d/openntpd.apk-new will contain the packages openntpd version... 2014-03-17 11:15:31 s/packages/packaged 2014-03-17 11:17:33 so 'replaces' will break openntpd 2014-03-17 11:34:59 ncopa: this openntpd.pre-upgrade handles both cases: http://sprunge.us/PLXV 2014-03-17 11:35:38 but it's kinda hacky 2014-03-17 11:36:35 and I think it will remove a user configured config if --clean-protected is used 2014-03-17 11:44:27 another option would be to use BUSYBOX_NTPD_OPTS="" in busybox-initscripts and `grep ^NTPD_OPTS= /etc/conf.d/ntpd` to determine whether to move the file 2014-03-17 12:07:04 hm 2014-03-17 12:10:47 i think it should be something like: if /etc/conf.d/ntpd and it is the openntpd's file, then move it 2014-03-17 12:11:59 yeah i think grepping it is an ok solution 2014-03-17 12:12:21 i dont know if we can apk info --who-owns /etc/conf.d/ntpd 2014-03-17 12:12:47 ncopa: will test 2014-03-17 12:12:52 we probably cannot do that from pre-upgrade 2014-03-17 12:13:18 i think since the apk db is not yet flushed to disk, it will say that it belongs to openntpd 2014-03-17 12:14:35 we could grep for NTPD_HOME 2014-03-17 12:15:03 yes 2014-03-17 12:15:07 that sounds good 2014-03-17 12:16:16 i think the service will not run unless NTPD_HOME is set 2014-03-17 12:42:27 ncopa: patches for openntpd, busybox-initscripts and alpine-conf sent to alpine-devel@a.o 2014-03-17 12:46:04 looks good 2014-03-17 12:46:18 i think what i'll do, apply openntpd now 2014-03-17 12:46:22 the rest tomorrow 2014-03-17 12:46:55 actually, shouldnt really matter 2014-03-17 12:47:00 it should just work... 2014-03-17 12:50:25 in theory... 2014-03-17 13:10:11 hm, clamav is still outdated 2014-03-17 13:10:23 would be nice if the new version is available asap 2014-03-17 13:10:50 oops, I forget 2014-03-17 13:10:52 sorry 2014-03-17 13:10:59 StarWarsFan: edge + 2.7? 2014-03-17 13:11:04 ack 2014-03-17 13:11:23 2.7 is blocked for a while… 2014-03-17 13:11:47 StarWarsFan: what's the latest release? 2014-03-17 13:12:56 Version : 0.98-r1 Build date : 2013-10-30 09:30:26 2014-03-17 13:13:04 ah, only edge is up to date 2014-03-17 13:13:09 now I know everything 2014-03-17 13:17:09 ok 2014-03-17 13:20:55 StarWarsFan: ↑ 2014-03-17 13:21:13 :-) 2014-03-17 13:25:32 ncopa: I've put something in 2.7's backports, but apparently it hasn't been build… 2014-03-17 13:55:57 seems like x86 builder fell off 2014-03-17 13:56:51 yup 2014-03-17 13:59:03 i think i should try get that web interfaces of buildserver status up 2014-03-17 14:23:29 that openvswitch install script 2014-03-17 14:23:45 i wonder if it would be better to try update db from the init.d script 2014-03-17 16:21:52 vkrishn: please see latest acf-core and acf-alpine-baselayout git for changes to add div around each section 2014-03-17 16:22:01 let me know what you think 2014-03-17 17:10:30 tdtrask, thanks, just check looks nice, solves most collapsible issues 2014-03-17 17:10:52 also the previous skins seems now better/fixed 2014-03-17 17:15:13 I think most of jqmobile related issues are done, 2014-03-17 17:15:36 Just tried DNS module, I guess it will take a while to update other modules too 2014-03-17 17:16:39 meanwhile I would try to refine using acf-core and acf-alpine-baselayout 2014-03-17 17:19:42 yes, I only updated those two packages 2014-03-17 17:19:56 was waiting for feedback before tackling the others 2014-03-17 17:20:23 didn't change skins recently 2014-03-17 17:24:51 is there any pagination available ? 2014-03-17 17:25:07 thinking for /acf/apk-tools/apk/available 2014-03-17 17:25:42 tablesorter does have pagination functionality 2014-03-17 17:25:47 but, I haven't implemented it yet 2014-03-17 17:26:14 I mean paginated data from server end 2014-03-17 17:26:33 yes, I wanted to look into using ajax calls from tablesorter 2014-03-17 17:26:40 ok 2014-03-17 17:27:16 there is some rough pagination code for some pages, but I would like to standardize it using tablesorter 2014-03-17 17:27:24 instead of the custom code I had written 2014-03-17 17:28:05 (that's why I was concerned with making tablesorter work with jquery mobile) 2014-03-17 17:28:19 (since I'm relying more heavily on tablesorter now) 2014-03-17 17:29:35 I am also tried as standalone javve/list.js, tristen/tablesort on github 2014-03-17 17:31:51 currently using https://github.com/Mottie/tablesorter/archive/v2.10.8.tar.gz 2014-03-17 17:58:00 I would prefer css over inline styling eg. tables in - acf/alpine-baselayout/interfaces/read 2014-03-17 18:01:30 what lower versions of browsers are targeted ? 2014-03-17 18:03:07 for non jqmobile skins 2014-03-17 18:11:22 vkrishn: since we're moving to HTML5, haven't determined which browsers to support 2014-03-17 18:11:29 used to shoot for IE6 :/ 2014-03-17 18:12:01 XP is dead, let's kill IE6 support too 2014-03-17 18:12:12 yup, done 2014-03-17 18:12:48 not worried so much about older browsers anymore 2014-03-17 18:12:59 and yes, I know about removing inline styling 2014-03-17 18:13:06 haven't gotten around to it yet 2014-03-17 18:13:18 please feel free to help with that :) 2014-03-17 18:14:50 inline styling also used for alpine-baselayout/health/* 2014-03-17 18:15:22 tdtrask, I can write css for removed inline styles, in some places a class to relevant table/td may be required 2014-03-17 18:15:44 if needed redo all the styles... not an issue 2014-03-17 18:16:38 if someone wants to create a new skin (jpeg) with sliced icon/background imgs I can css/style them too 2014-03-17 18:18:05 I think at this point we may wanna see a merge point for interface with up comming acf2 2014-03-17 18:19:11 it is possible to write templates look and feel similar 2014-03-17 18:35:19 vkrishn: thanks, I would appreciate the help with the inline styles 2014-03-17 18:35:43 would like something generic that can be used on multiple pages/skins 2014-03-17 18:36:36 ACTION is not familiar with acf2 progress 2014-03-17 18:37:10 acf is not being replaced anytime soon 2014-03-17 18:37:14 will leave acf2 aside for the moment 2014-03-17 18:37:47 ACTION is guessing they will continue parallel for a while 2014-03-17 18:38:54 I am guessing for tables its only the preset columns width at some pages that would vary 2014-03-17 18:39:28 for that reasons I allowed in template a page wise css to be inserted 2014-03-17 18:40:04 but that would mean modules writing/including that css file 2014-03-17 18:40:28 and add it in skins dir at install 2014-03-17 18:45:45 or maybe a common www folder for theme agnostic styles (structural only) 2014-03-17 18:46:31 eg is static folder 2014-03-17 19:52:42 sorry vkrishn, I didn't see your last comments before you left 2014-03-17 19:53:38 for the preset column widths, we can just set a global setting in .css and not make it page specific 2014-03-17 19:54:14 ACTION is more concerned about how to display the charts in health 2014-03-17 19:54:35 might have to replace table with inline css with another thing entirely 2014-03-18 06:20:49 ncopa: pushed f2fs support to mkinitfs 2014-03-18 06:22:02 ncopa: I'm not going to do this, but just in case, to make a release I need to do git tag -a v$VERSION -m "==== release $VERSION ===="? 2014-03-18 07:26:18 mornin 2014-03-18 07:26:33 barthalion: iirc, you change version in Makefile firs 2014-03-18 07:26:48 and that that commit with the above, yes 2014-03-18 09:25:44 hm, how can I automatically restore apkovl? 2014-03-18 09:31:23 ah, found it 2014-03-18 11:28:08 failed to build v2.4.11-212-g86d6c29. http://distfiles.alpinelinux.org/distfiles/build64-2-4.txt 2014-03-18 11:32:33 files from v2.4.11-213-g5540f95 uploaded 2014-03-18 11:32:38 files from v2.4.11-213-g5540f95 uploaded 2014-03-18 12:05:04 I created group "Alpine Linux" on Linkedin.: http://lnkd.in/bFuMCtD 2014-03-18 12:50:05 ok i ported lddtree to bb shell 2014-03-18 12:50:25 that should help us with mkinitfs fix 2014-03-18 13:41:45 hm 2014-03-18 13:41:56 i wonder if i should just rewrite the mkinitramfs from scratch 2014-03-18 13:42:05 including the initramfs's init 2014-03-18 13:42:19 I was thinking about using/forking mkinitcpio 2014-03-18 13:44:16 i requires bash? 2014-03-18 13:45:00 yeah 2014-03-18 13:45:20 i'd prefer not depend on bash 2014-03-18 13:45:32 what other options are there? 2014-03-18 13:45:36 that's why I used "forking" 2014-03-18 13:45:38 genkernel? 2014-03-18 13:45:48 dracut? 2014-03-18 13:45:59 I guess this is all 2014-03-18 13:46:17 i think debian/ubuntu has something too 2014-03-18 13:46:34 there are some old mkinitrd for older redhats too 2014-03-18 13:46:39 initramfs-tools 2014-03-18 13:53:39 initramfs-tools looks fairly nice 2014-03-18 14:06:48 indeed 2014-03-18 14:07:37 what we want in any case is some framework 2014-03-18 14:07:44 something modular 2014-03-18 14:07:55 our current mkinitfs is kinda static 2014-03-18 14:18:32 ok 2014-03-18 14:18:38 i think for now we continue with mkinitfs 2014-03-18 14:18:42 but i think I'll change it a bit 2014-03-18 14:19:16 instead of /etc/mkinitfs/modules.d/ and /etc/mkinitfs/files.d/ 2014-03-18 14:19:53 i think we should do something like: /etc/mkinitfs/features.d/.modules 2014-03-18 14:20:21 and /etc/mkinitfs/features.d/.files 2014-03-18 14:20:37 I wouldn't complain if we just use wildcard in Makefile to get everything from features.d/ 2014-03-18 14:21:37 we coudl also in future add features.d/.hook something 2014-03-18 14:22:04 and maybe .depends 2014-03-18 14:22:12 so you could run the hooks in dep order 2014-03-18 14:22:24 i dont know if we need premount postmount etc 2014-03-18 14:24:05 but are you ok with replacing {modules.d,file.d}/$feature with features.d/$feature.{modules,files} ? 2014-03-18 14:24:15 yeah, sounds ok 2014-03-18 14:24:22 i think it'll make the mkinitfs script simpler 2014-03-18 14:24:36 the easier to add something new, the better 2014-03-18 15:00:28 ncopa: build-edge-x86_64 not happy at the moment? 2014-03-18 15:02:25 seems it has not been happy since last reboot 2014-03-18 15:02:31 i think i have fixed it 2014-03-18 15:02:41 build/build-edge-x86_64 building main/python-2.7.6-r2 2014-03-18 15:03:05 i should also try get the build server status page up and running... 2014-03-18 15:03:37 ACTION can test out x86 for now 2014-03-18 15:15:02 Can ZRAM be added to the kernel as a module ? - it works very well on machines with lower RAM - my script for it was added to Antix last summer so it's had a bit of testing https://wiki.gentoo.org/wiki/Zram 2014-03-18 15:17:11 BitL0G1c: I'll take a look 2014-03-18 15:18:36 many thx 2014-03-18 15:19:37 if it's added I'll submit a patch with an init & conf.d script 2014-03-18 15:20:16 I'll just take what Gentoo has 2014-03-18 15:20:27 I'll probably backport it to 2.7 too 2014-03-18 15:20:53 ok 2014-03-18 15:31:05 i wonder if the move of /etc/mkinitfs/files.d to /etc/mkinitfs/features.d/ is acutally worth it 2014-03-18 15:44:03 ncopa: btw, we should enable tmpfiles service by default and stop using mkdir in daemons 2014-03-18 15:45:03 tmpfiles service? 2014-03-18 15:45:30 openrc implemented tmpfiles.d from systemd 2014-03-18 15:45:41 you drop file in /usr/lib/tmpfiles.d/*.conf 2014-03-18 15:45:56 let's say you want to create /var/cache/apk on boot 2014-03-18 15:46:11 d /var/cache/apk chmod user group 2014-03-18 15:46:49 got you 2014-03-18 15:47:26 additionally /etc/tmpfiles.d/ takes precedence if user want to customize it 2014-03-18 15:48:02 there is more options, but I'm mostly complaining about bunch of mkdir I've seen in our init.d scripts 2014-03-18 15:50:04 is it possible to check which packages provide a daemon? 2014-03-18 16:26:09 yay 2014-03-18 16:49:31 BitL0G1c: done; I'll push ninit script later 2014-03-18 16:53:10 anyone use the own cloud package? 2014-03-18 16:56:38 Mp5shooter: _0/ I do... some months ago i installed it on a client 2014-03-18 16:56:46 oh cool 2014-03-18 16:57:00 working smoothly ; ) 2014-03-18 16:57:05 hehe 2014-03-18 16:57:07 just installed it 2014-03-18 17:15:22 BitL0G1c: and everything done 2014-03-18 17:17:18 ;-) gracias 2014-03-18 17:25:54 hey alacerda, when you installed owncloud did it say anything about not having a locale installed? 2014-03-18 17:27:22 Mp5shooter: not that i remember! 2014-03-18 17:34:59 hm 2014-03-18 17:37:48 kernel = 3.10.32-1 & modules install to /lib/modules/3.10.33-1-grsec 2014-03-18 17:39:58 hm? 2014-03-18 17:41:53 & kernel module seems to be missing from modules.dep 2014-03-18 17:45:12 kernel is 3.10.33, see pkgver 2014-03-18 17:45:19 I'll take a look at modules.dep later 2014-03-18 17:46:27 root@web [/lib/modules]# uname -r 2014-03-18 17:46:27 3.10.32-1-grsec 2014-03-18 17:47:10 how do you set locales? 2014-03-18 17:47:54 mm 2014-03-18 17:47:57 someone else had this problem 2014-03-18 17:47:58 http://alpinelinux.org/forum/installation/owncloud-locale-error 2014-03-18 17:50:43 BitL0G1c: vb-alpine27:~# uname -r 2014-03-18 17:50:46 3.10.33-1-grsec 2014-03-18 17:54:59 module didn't get installed, not sure why 2014-03-18 17:55:49 ncopa: can you take a look? ↑ 2014-03-18 18:19:39 not sure why I show 3.10.32 then - but it has happened before - I think when I was having problems with the initramfs not rebuilding for LUKS (which is fixed now btw - many thx) 2014-03-18 18:19:58 a symlink fixes it 2014-03-18 19:33:38 what do we do if something checks if a locale is installed? 2014-03-18 19:35:09 we patch the software to avoid the check 2014-03-18 19:36:40 :s 2014-03-18 19:36:48 own cloud won't let you proceed with setup 2014-03-18 19:36:57 owncloud* 2014-03-18 19:37:29 Setting locale to en_US.UTF-8/fr_FR.UTF-8/es_ES.UTF-8/de_DE.UTF-8/ru_RU.UTF-8/pt_BR.UTF-8/it_IT.UTF-8/ja_JP.UTF-8/zh_CN.UTF-8 failed 2014-03-18 19:37:29 Please install one of theses locales on your system and restart your webserver. 2014-03-18 19:37:49 I read that owncloud 5 just warned you about it but owncloud 6 doesn't let you continue 2014-03-18 19:43:00 Mp5shooter: tried to comment 481st line of lib/base.php and 28th of settings/admin.php? 2014-03-18 19:43:18 hmmm 2014-03-18 19:43:56 I'll give you a patch in a moment 2014-03-18 19:45:22 Mp5shooter: https://paste.xinu.at/IPqa/ 2014-03-18 19:46:00 I have no idea if it works 2014-03-18 19:46:35 if not, you may want to restore '\Patchwork\Utf8\Bootup::initLocale();' above 'return true;' 2014-03-18 19:50:32 elo :) 2014-03-18 19:50:45 looks like it worked barthalion :P 2014-03-18 19:51:12 good 2014-03-18 19:51:18 thank you 2014-03-18 19:52:04 I was going nuts 2014-03-18 19:52:10 trying to find a way to get it to let me through 2014-03-18 19:52:11 haha 2014-03-18 19:52:24 the patch worked alone or you needed to restore the line I mentioned? 2014-03-18 19:53:22 it worked alone 2014-03-18 19:54:05 nice :) 2014-03-18 19:54:59 yeah 2014-03-18 19:55:01 thanks again :P 2014-03-18 19:55:20 you're welcome 2014-03-18 19:59:15 Mp5shooter: ping rnalrd about including it in the package 2014-03-18 19:59:22 ok 2014-03-18 20:46:03 hmm just build a example iso from http://bit.ly/1ePzCO5, and when booting the final iso getting "/sbin/init not found in new root" issue, any ideas ? 2014-03-18 21:04:49 own cloud is too much of a pain in the butt to set up 2014-03-18 21:16:06 n8@all 2014-03-19 07:00:55 Mp5shooter, i know, owncloud-5@edge is brokem 2014-03-19 07:01:03 broken* 2014-03-19 07:01:25 owncloud6 i mean 2014-03-19 07:01:46 i'm figuring out if I should patch it or not 2014-03-19 07:08:17 le^zeek: "/sbin/init not found in new root" means that it failed to unpack the packages into tmpfs for some reason 2014-03-19 07:08:47 you can try add noquiet verbose to boot cmdline 2014-03-19 07:09:03 eg at boot prompt type: grsec noquiet verbose 2014-03-19 07:09:15 should display errors 2014-03-19 07:32:39 will check, fell asleep last night :) 2014-03-19 09:18:54 hmpf i think we'll have to fork lddtree.sh 2014-03-19 09:19:10 https://bugs.gentoo.org/show_bug.cgi?id=504954 2014-03-19 09:21:10 can you take a look at linux-grsec? I enabled CONFIG_ZRAM, but somehow it's missing from the package 2014-03-19 09:21:53 I'm thinking it may be grsec fault 2014-03-19 09:23:46 do you have any 3rd party kernel modules? 2014-03-19 09:23:57 if you have it might not have been upgraded 2014-03-19 09:24:23 no, it's our main/linux-grsec 2014-03-19 09:25:38 depends on BLOCK && SYSFS && ZSMALLOC 2014-03-19 09:26:03 build would fail if deps were not met 2014-03-19 09:26:26 are you sure about that? 2014-03-19 09:27:01 this is what it used to do back when I was compiling my kernels 2014-03-19 09:27:49 and it requires disabled ZSMALLOC, not enabled 2014-03-19 09:27:59 $ grep ZSMALLOC .config 2014-03-19 09:28:00 # CONFIG_ZSMALLOC is not set 2014-03-19 09:28:10 Depends on: BLK_DEV [=y] && BLOCK [=y] && SYSFS [=y] && ZSMALLOC [=n] 2014-03-19 09:28:49 so it depends on ZSMALLOC which is disabled 2014-03-19 09:28:59 ah, so I misread it 2014-03-19 09:29:00 i think make silentoldconfig will resolv the deps 2014-03-19 09:29:18 and simply remove zram 2014-03-19 09:29:55 http://bugs.alpinelinux.org/issues/2076 2014-03-19 09:30:11 specifically: http://bugs.alpinelinux.org/issues/2076#note-1 2014-03-19 09:30:56 it's in staging since ages and perfectly usable 2014-03-19 09:31:10 why is it still in staging? 2014-03-19 09:31:17 why not moved out from theree? 2014-03-19 09:31:23 no idea, it was going to be out in 3.10 2014-03-19 09:32:11 as i explained, it will result in a tainted kernel 2014-03-19 09:32:30 and basically lose any posibility to ask for support upstream 2014-03-19 09:32:39 we cannot report bugs upstream after that 2014-03-19 09:33:03 Linux devs doesn't support kernels with staging enabled? 2014-03-19 09:33:24 i think they dont support kernels wich are 'tainted' 2014-03-19 09:33:34 so why do we even enable CONFIG_STAGING? 2014-03-19 09:33:54 i dont think the kernel ends up as 'tainted' if its only modules 2014-03-19 09:34:24 so if zsmalloc was a module, then we'd already have it 2014-03-19 09:34:33 isn't kernel "tainted" only if code is non-GPL? 2014-03-19 09:35:07 hm, no 2014-03-19 09:35:10 11: 'C' if a staging driver has been loaded. 2014-03-19 09:35:25 I guess modules counts too 2014-03-19 09:35:39 right 2014-03-19 09:35:54 so if you modprobe a staging module 2014-03-19 09:36:02 it'll get tanited 2014-03-19 09:36:10 if you dont modprobe it its not tainted 2014-03-19 09:36:35 but if its compiled into kernel, there is no way to run it not tainted 2014-03-19 09:36:58 have we ever reported an issue on lkml? 2014-03-19 09:37:04 yes 2014-03-19 09:37:09 various 2014-03-19 09:37:29 even if they would bring up that we have tainted kernel, it's matter of unsetting one thing in config to prove that it doesn't matter 2014-03-19 09:37:38 well 2014-03-19 09:37:42 yes 2014-03-19 09:37:56 but it also means that we need to recompile kernel for that 2014-03-19 09:38:43 i suppose we often do that anyways, since while debugging kernel we should use vanilla 2014-03-19 09:38:48 well, ok, I'll investigate why it wasn't moved out of staging yet 2014-03-19 09:39:03 thats the "correct" way forward yes 2014-03-19 09:39:26 my guess is that the reason is that 2014-03-19 09:39:37 something is still not resolved in there 2014-03-19 09:39:48 or maintainer is not wanting to maintain it anymore 2014-03-19 09:39:50 https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git/commit/?id=ab41c488c652be9ed08e5b1660727147df27e8cd 2014-03-19 09:39:50 or similar 2014-03-19 09:40:04 there is no reason actually 2014-03-19 09:40:11 it's moved in mainline 2014-03-19 09:40:21 ha! 2014-03-19 09:40:25 great! 2014-03-19 09:40:37 that's why I thought it's done already… 2014-03-19 09:40:41 will it be included in 3.14? 2014-03-19 09:41:08 http://www.phoronix.com/scan.php?page=news_item&px=MTU0NzA 2014-03-19 09:41:10 seems so 2014-03-19 09:41:22 i suggest we try get 3.14 kernel for alpine 3.0 then 2014-03-19 09:42:05 yes, it was merged in Jan 30 by Torvalds 2014-03-19 09:45:46 that brings us back to what kernel to use for v3.0 2014-03-19 09:46:09 we generally want do long-term kernels as much as possible 2014-03-19 09:46:41 so since 3.12 was announced longterm, there might be an idea to try stick to that 2014-03-19 09:46:51 yep 2014-03-19 09:46:56 grsecurity has already dropped support for it though 2014-03-19 09:47:04 and i dont know what ubuntu lts will use 2014-03-19 09:47:11 I guess that grsec will stick to what Ubuntu LTS will use 2014-03-19 09:47:17 and it will be 3.13 or 3.14 2014-03-19 09:47:21 thats what i think too 2014-03-19 09:47:44 and it might be an idea to aim for that 2014-03-19 09:47:49 but hum 2014-03-19 09:48:04 ẅasnt it an ubuntu dev that was supposed to maintain 3.12? 2014-03-19 09:48:09 https://lists.ubuntu.com/archives/ubuntu-devel/2014-March/038129.html I guess they hope for 3.14 2014-03-19 09:48:10 so 2014-03-19 09:48:30 since 3.12 was longerm, the chance for 3.14 become longterm is smaller 2014-03-19 09:48:37 now 2014-03-19 09:48:45 Jiri Slaby maintains 3.12 2014-03-19 09:48:47 since we are trying to get musl out 2014-03-19 09:48:53 there were 2 optoins 2014-03-19 09:48:54 he's employed by SUSE 2014-03-19 09:48:58 do 2.8 2014-03-19 09:49:12 if we do alpine v2.8, then we definitively want longterm kernel 2014-03-19 09:49:36 but me and fabled talked about since musl will be a bit buggy 2014-03-19 09:49:48 we could consider 3.0 as a 'beta' 2014-03-19 09:49:55 and maybe extend v2.7 support a bit 2014-03-19 09:50:14 that would also meant that we probably will reduce the support time for 3.0 2014-03-19 09:50:28 ...which leads to... 2014-03-19 09:50:42 in that case it doesnt really matter if kernel is not longterm 2014-03-19 09:50:59 and then we will likely want an as new kernel as possible 2014-03-19 09:51:00 2.8 + 3.0 sounds fine 2014-03-19 09:51:13 i dont think I'll have capacity to do both 2.8 and 3.0 2014-03-19 09:51:56 but i think we want do 3.0 2014-03-19 09:51:58 what help is required? backporting bug and security fixes or you mean "release engineering" 2014-03-19 09:52:17 bug and security fixes 2014-03-19 09:52:19 also 2014-03-19 09:52:44 it means that all packages will have to build/work on both musl and uclibc 2014-03-19 09:52:58 this is what we do with edge now 2014-03-19 09:53:00 if we simply drop uclibc and jump, it doesnt matter if it dont build on uclibc after that 2014-03-19 09:53:05 yes 2014-03-19 09:53:18 thatas what we do on edge, and thats what i want stop doing :) 2014-03-19 09:53:30 i think what i really want 2014-03-19 09:53:30 I'm more concerned with 2.8 + 3.0 + edge than just 2.8 and 3.0 2014-03-19 09:53:41 1) we jump to musl asap 2014-03-19 09:53:47 (and drop uclibc) 2014-03-19 09:54:01 2) we do only alpine 3.0, with as new kernel as possible (3.14) 2014-03-19 09:54:42 we annouce 3.0 as a first musl release, can be considered beta, or whatever 2014-03-19 09:54:50 and also say that it will have shorter support time 2014-03-19 09:54:56 maybe only to 3.1 is out 2014-03-19 09:55:23 i don think we will do both v2.8 *and* v3.0 2014-03-19 09:55:42 we can extend 2.7 support and do 3.0 short-time release 2014-03-19 09:55:53 thats what i'd prefer 2014-03-19 09:55:57 something like 6 or 12 months 2014-03-19 09:56:01 exactly 2014-03-19 09:56:43 during that time we can polish upgrading from 2.7 and possible musl problems 2014-03-19 09:56:54 yes 2014-03-19 09:57:18 so right, extedned 2.7 and short-time 3.0 sounds the best so far 2014-03-19 10:04:17 2.7 is also a good candidate for extended because it already has a good longterm kernel 2014-03-19 10:20:50 re musl, the current blocker is mkinitfs' files.d/* 2014-03-19 10:21:05 the dependencies are static 2014-03-19 10:21:17 so there is no way to make a list that works on both uclibc and musl 2014-03-19 10:21:25 so the fix is to resolv the dependencies 2014-03-19 10:21:39 so you dont need to list the .so files in there 2014-03-19 10:21:49 and i am almost there... 2014-03-19 10:22:15 but it seems like i have to fork pax-utils' lddtree.sh 2014-03-19 10:22:50 we can just remove it from pax-utils and ship as additional package 2014-03-19 10:22:55 what about that? 2014-03-19 10:22:58 yeah 2014-03-19 10:23:03 thats what we need to do 2014-03-19 10:23:19 vaiper@gentoo didnt want support hte shell version 2014-03-19 10:23:28 and they provide a better python version 2014-03-19 10:23:33 but i dont want depend on python 2014-03-19 10:23:55 which would mean that python would be a dep for the kernel 2014-03-19 10:24:13 we cant have that... 2014-03-19 10:24:50 ugh 2014-03-19 10:25:04 they should drop lddtree.sh, if they don't want to support it 2014-03-19 10:26:36 thats what i thought too 2014-03-19 10:26:57 but otoh, it was nice that they didnt yet do it - or i wouldnt have found it 2014-03-19 10:31:11 reminds me, i didnt push uggedals patches... 2014-03-19 10:31:14 should do that now 2014-03-19 10:34:25 uggedal: there was checksum failure on budybox-initscripts, could you please doublecheck that you actually sent the thing you intended to? 2014-03-19 10:34:43 if it was I'll just git commit --amend the checksum 2014-03-19 10:35:00 ntpd.initd: FAILED 2014-03-19 10:35:00 sha512sum: WARNING: 1 of 1 computed checksums did NOT match 2014-03-19 10:35:48 uggedal: btw, still using Arch? :) 2014-03-19 11:34:51 barthalion: only on my Macbook Pro 2014-03-19 11:35:33 my distro timeline has been RedHat, Gentoo, Debian, Arch, Alpine, Gentoo, Alpine 2014-03-19 11:36:01 ncopa: will check the checksum failure 2014-03-19 11:37:33 ncopa: seems i added the provide ntp-client 2014-03-19 11:37:44 line before committing and forgot to abuild checksum 2014-03-19 11:38:10 ncopa: should i send a new patch with fixed checksum or will you fix it? 2014-03-19 11:39:29 ncopa: nevermind, skipped your --amend comment 2014-03-19 11:41:02 reviewing a 65000 line patch at work so my head is not functioning properly 2014-03-19 11:48:05 barthalion: oh, and the server running http://mediaqueri.es is still Arch... 2014-03-19 12:06:40 65000 line patch! yeah... drives you nut... 2014-03-19 12:07:07 ok so the initscript is good but checksum is bad and not the other way around 2014-03-19 12:11:16 ncopa: thanks 2014-03-19 12:12:38 ncopa: in java. building the whole project takes 45min. so quite painful 2014-03-19 12:13:54 i feel with you... 2014-03-19 12:44:46 I don't think that it replaces pax-utils 2014-03-19 12:44:53 it depends on scanelf from it 2014-03-19 12:45:10 it provides only lddtree, not symtree or dumpelf 2014-03-19 13:04:08 'replaces=foo' means basically, 2014-03-19 13:04:35 allow overwriting files that is owned by 'foo' 2014-03-19 13:04:47 otherwise apk will barf, file X is owned by 'foo' 2014-03-19 13:04:55 and thats it bascially 2014-03-19 13:10:02 what!? 2014-03-19 13:10:10 First, rewinding head to replay your work on top of it... 2014-03-19 13:10:10 fatal: cannot create directory at 'main/lddtree': Permission denied 2014-03-19 13:10:10 could not detach HEAD 2014-03-19 13:10:21 build-edge-musl-x86_64:~/aports$ ls -la 2014-03-19 13:10:21 total 132 2014-03-19 13:10:21 drwxr-xr-x 7 root root 4096 Jan 27 08:56 . 2014-03-19 13:10:34 how did that happen? 2014-03-19 13:12:37 ah, I thought it works like in Arch 2014-03-19 13:12:44 here replaces mean literally replace 2014-03-19 13:13:09 jbilyk__, jbilyk__: there is somethning wrong with wiki? 2014-03-19 13:14:45 i wonder what is changing permission to aports to root:root ? 2014-03-19 13:17:56 i think its goffice 2014-03-19 13:26:01 ncopa: have you tried lxc on nfs? 2014-03-19 13:26:15 nope 2014-03-19 13:27:11 seems it cannot lock over nfs and fails. 2014-03-19 13:27:24 on the topic of lxc, are you guys using veth with a bridge, nat with iptables or just sharing network with host? 2014-03-19 13:27:34 doesnt nfs has some lockd something? 2014-03-19 13:27:48 i use veth with bridge 2014-03-19 13:27:58 and some places macvlan 2014-03-19 13:28:18 beware that bridge over bonding might give unexpected issues 2014-03-19 13:28:37 i'd like to investigate openvswitch, but has not had time to do so 2014-03-19 13:28:40 i read flock doesnt work over nfs. there is a patch to use fcntl which should be supported on nfs v3 2014-03-19 13:29:24 i'm trying to figure out why building goffice caused various dirs to be owned by root... 2014-03-19 13:30:22 ncopa: do you use static ips or run dnsmasq or some other dhcpd on the host? 2014-03-19 13:31:26 tryting to get lxc running on rpi :) 2014-03-19 13:31:30 trying 2014-03-19 13:32:39 uggedal: i use static some places, but since its a bridge, i use the same dhcp server as the host uses 2014-03-19 13:32:46 eg, not running dhcp server on host 2014-03-19 14:11:12 ncopa: http://git.alpinelinux.org/cgit/aports/commit/main/nfs-utils/0003-replace-__attribute_malloc__-with-the-more-portable-.patch?id=35038f4a2af51efed7265a064fd4a43ffd8a4b95 2014-03-19 14:11:35 could that be the cause of: Mar 19 14:05:09 rpi daemon.err rpc.statd[9837]: my_svc_run() - select: Invalid argument 2014-03-19 14:12:41 i doubt 2014-03-19 14:13:09 __attribute_malloc__ and __attribute__((__malloc__)) are supposed to be synonyms 2014-03-19 14:13:20 unless arm should not use it for some reason 2014-03-19 14:13:46 it spams my syslog and makes nfs not usable 2014-03-19 14:14:56 did it work before that? 2014-03-19 14:15:04 bfore that patch? 2014-03-19 14:15:07 i dont know 2014-03-19 14:15:13 i never tried nfs on rpi before 2014-03-19 14:15:24 barthalion: why did you add gettext-dev to tevent? 2014-03-19 14:15:46 ./configure complained without it 2014-03-19 14:16:47 i could try to build it, but i need some kind of storage on the pi to build anything :) 2014-03-19 14:39:30 /home/ncopa/aports/main/tevent/src/tevent-0.9.21/lib/replace/wscript:403: error: library gettext not found, try specifying the path to it with --with-gettext= or --without-gettext to build without 2014-03-19 14:39:46 barthalion: do you mind if we try --without-gettext? 2014-03-19 14:41:36 ncopa: go ahead, I've just picked the easiest way 2014-03-19 14:42:21 good. thanks! 2014-03-19 21:14:45 ncopa: do you mind if I add overlayfs patch to linux-grsec? 2014-03-20 07:47:16 barthalion: i am sceptic to patching kernel more than necessary 2014-03-20 07:48:00 i'm mostly scared of future maintenance of it 2014-03-20 07:48:21 so if possible, i'd prefer wait til its in mainline kernel 2014-03-20 07:48:43 aye; I doubt it will ever get merged 2014-03-20 07:49:06 and haven't tested with grsec, overlaying two filesystems can be too big abuse for that 2014-03-20 07:49:11 I'll just compile it locally 2014-03-20 08:17:11 ACTION is also interested in overlayfs 2014-03-20 09:22:59 alpine linux will soon be on front page of slashdot.org 2014-03-20 09:23:10 mentioned in musl 1.0 release 2014-03-20 09:24:17 hopefully :) 2014-03-20 10:58:05 ncopa: I'm wondering if it'd be possible to implement all non-vanilla kernels as linux-vanilla subpackages 2014-03-20 10:58:28 everything is possible 2014-03-20 10:58:48 but keep in mind we want simplify maintenance if possible 2014-03-20 10:59:13 wouldn't it simplify after all? 2014-03-20 10:59:27 it would simplify some things 2014-03-20 10:59:30 complicate others 2014-03-20 10:59:57 so it would give possibly low profit 2014-03-20 11:00:07 correct 2014-03-20 11:00:40 one of the problem is that you cannot really have different maintainers for the different kernel flavors 2014-03-20 13:13:54 clandmeter: just a heads up... 2014-03-20 13:14:06 http://slashdot.org/submission/3422139/musl-libc-hits-10-milestone 2014-03-20 13:14:30 they mention alpinelinux.org in that slashdot submission 2014-03-20 13:14:40 just in case we get slashdotted... 2014-03-20 13:14:52 :) 2014-03-20 13:16:08 i suppose we could vote for that story... 2014-03-20 13:17:24 done 2014-03-20 14:04:28 clandmeter: ok to upgrade to ruby 2.1.1? 2014-03-20 14:05:34 http://www.redmine.org/issues/16194 2014-03-20 14:10:41 ncopa, so cherry-pick that patch for 2.1.1 ? 2014-03-20 14:11:02 clandmeter, the reason i'm interested is that libreadline upgrade broke ruby build due to https://bugs.ruby-lang.org/issues/9578 2014-03-20 14:11:19 oh but 2.1.1 is also affected 2014-03-20 14:11:26 ncopa, so just simple ugprade will not fix it 2014-03-20 14:12:41 k 2014-03-20 14:12:54 bah 2014-03-20 14:12:55 ok 2014-03-20 14:18:58 actually new readline broke a lot, I think I didn't patched it yet in Alpine 2014-03-20 14:19:28 fabled: https://projects.archlinux.org/svntogit/packages.git/tree/trunk/0001-Fix-undeclared-identifier-error-by-using-the-actual-.patch?h=packages/ruby this is want you want 2014-03-20 14:19:49 barthalion, yes, looks like it 2014-03-20 14:19:55 ncopa, ^^^ you cherry pick that or me? 2014-03-20 14:24:38 clandmeter, musl (also Alpine Linux link) is on slashdot frontpage now 2014-03-20 14:26:17 barthalion: yeah i found the oneliner 2014-03-20 14:26:18 it won't fix build failures, but at least REPL apps will work 2014-03-20 14:26:22 fabled: cpu can still handle it :) 2014-03-20 14:27:03 would be interesting to know if that increases our web traffic a lot or not 2014-03-20 14:29:30 ill monitor our analicts 2014-03-20 14:30:23 we should move to piwik :p 2014-03-20 14:30:56 from reddit comments 2014-03-20 14:30:57 "Assuming this is a media player (and given no information to indicate the contrary), how does this compare to VLC?" 2014-03-20 14:31:01 lol 2014-03-20 14:31:10 lol 2014-03-20 14:31:59 that's actually great question, but it's beyond my brain to write a clever answer to it 2014-03-20 14:39:45 i guess he needed a screenshot to find out its not a media player. 2014-03-20 19:30:32 we are developping with python3 on Alpine-2.8 and we have small but annoying problems with time method in python, something does not woriking well in Alpine (maybe the uclibc), here is a simple python test to illustrate the issue : 2014-03-20 19:30:39 import time 2014-03-20 19:30:46 t = (2009, 2, 17, 17, 3, 38, 1, 48, 0) 2014-03-20 19:30:51 t = time.mktime(t) 2014-03-20 19:30:57 print("time : ", time.strftime("%b %d %Y %H:%M:%S", time.gmtime(t))) 2014-03-20 19:31:06 which work well on any distro/python version and give nothing on Alpine 2014-03-21 07:26:10 barthalion: please be careful with kernels in stable branches 2014-03-21 07:26:17 http://bugs.alpinelinux.org/issues/2776 2014-03-21 07:27:24 basically, if you rebuild kernel (with pkgrel bump) you need rebuild all the 3rdparty modules - always 2014-03-21 12:42:13 ncopa, http://sprunge.us/YaGP <- mosquitto upgrade to 1.3.0 2014-03-21 12:49:54 It would be nice to have pastebin.a.o with 3months+ expiry 2014-03-21 12:58:37 cool 2014-03-21 12:58:42 i'll upgrade mosquitto 2014-03-21 13:02:06 vkrishn: the patch is wrong, we shouldnt add c-ares to depends 2014-03-21 13:02:13 only to makedepends 2014-03-21 13:02:25 resend ? 2014-03-21 13:02:32 i think i cna just bump it 2014-03-21 13:02:47 unless you want your name on the commit 2014-03-21 13:02:55 nope, ok 2014-03-21 13:03:48 adding only to makedepends will install c-ares when `apk add mosquitto` is done ? 2014-03-21 13:03:49 thanks! 2014-03-21 13:03:56 yes 2014-03-21 13:03:59 ok 2014-03-21 13:04:11 abuild will detect the elf dependencies with scanelf 2014-03-21 13:04:28 mosquitto-1.3-r0 depends on: 2014-03-21 13:04:28 busybox 2014-03-21 13:04:28 so:libc.musl-x86.so.1 2014-03-21 13:04:28 so:libcrypto.so.1.0.0 2014-03-21 13:04:28 so:libgcc_s.so.1 2014-03-21 13:04:29 so:libssl.so.1.0.0 2014-03-21 13:04:31 hm 2014-03-21 13:04:35 it didnt 2014-03-21 13:04:38 I had this little confusion for a while 2014-03-21 13:04:49 thanks 2014-03-21 13:04:59 it didnt pull in -cares 2014-03-21 13:05:13 -lcares 2014-03-21 13:05:37 ifeq ($(WITH_SRV),yes) 2014-03-21 13:05:37 LIB_CFLAGS:=$(LIB_CFLAGS) -DWITH_SRV 2014-03-21 13:05:37 LIB_LIBS:=$(LIB_LIBS) -lcares 2014-03-21 13:05:37 endif 2014-03-21 13:06:14 its optional somehow 2014-03-21 13:06:25 i think it needs be built with WITH_SRV=yes 2014-03-21 13:06:48 I tried setting WITH_SRV:=yes in config.mk to "no", did not work 2014-03-21 13:07:39 # Build with SRV lookup support. 2014-03-21 13:07:39 WITH_SRV:=yes 2014-03-21 13:08:11 but on knoppix with no changes I installed libc-ares-dev and it compilled 2014-03-21 13:08:27 looks like we should use cmake 2014-03-21 13:08:54 yes, thats what I did, mkdir build, cd mkdir, cmake .. , make 2014-03-21 13:12:43 so either comment out -lcares or use cmake 2014-03-21 13:13:14 i think we want srv support? 2014-03-21 13:14:23 that is for fast DNS resolve ? 2014-03-21 13:14:40 async 2014-03-21 13:18:31 hm 2014-03-21 13:18:43 i wonder how to disable memory traicking with cmake 2014-03-21 13:24:08 so 2014-03-21 13:24:11 i asked the devs 2014-03-21 13:24:16 cmake is for win only 2014-03-21 13:24:21 and i am stupid :) 2014-03-21 13:24:30 cares *is* linked in 2014-03-21 13:24:41 but not against the server 2014-03-21 13:24:47 its linked to the libmosquitto 2014-03-21 13:25:14 mosquitto-libs-1.3-r0 depends on: 2014-03-21 13:25:14 so:libc.musl-x86.so.1 2014-03-21 13:25:14 so:libcares.so.2 2014-03-21 13:25:14 so:libcrypto.so.1.0.0 2014-03-21 13:25:14 so:libgcc_s.so.1 2014-03-21 13:25:15 so:libssl.so.1.0.0 2014-03-21 13:25:26 it has bugs though :| 2014-03-21 13:25:40 that's fine then, as longas you can fix the mosquitto_pub keepalive=1 problem, 2014-03-21 13:35:53 I could try converting APKBUILD to use cmake, if its better 2014-03-21 14:00:06 no 2014-03-21 14:00:13 the current is correct 2014-03-21 14:00:43 i was looking wrong place 2014-03-21 14:00:47 its all fine 2014-03-21 14:29:57 pskfile is used by bin mosquitto (not by utils) 2014-03-21 14:30:13 but bin mosquitto_passwd is packed in utils 2014-03-21 14:31:10 sub and pub users may not use mosquitto_passwd at all 2014-03-21 14:33:53 ? 2014-03-21 14:34:13 so pskfile should be packed in what package? 2014-03-21 14:34:32 sorry , pwdfile 2014-03-21 14:35:56 /bin/mosquitto_passwd is admin tool or can be used by util users also ? 2014-03-21 14:36:11 admin = broker 2014-03-21 14:40:19 I suppose admin user could apk add mosquitto-utils if needed 2014-03-21 14:40:51 ok 2014-03-21 14:40:58 but um 2014-03-21 14:41:05 probably good point though 2014-03-21 14:41:19 might be an idea to ship it with 'mosquitto' package, with the server tool 2014-03-21 14:41:29 bcoz, util users need bother about broker files 2014-03-21 14:41:50 the -utils package is just /usr/bin/* 2014-03-21 14:43:19 need not* 2014-03-21 14:43:29 i think youre right 2014-03-21 14:43:55 we could probably also rename the mosquitto-utils to mosquitto-clients 2014-03-21 14:44:26 ok sounds better 2014-03-24 06:55:59 well 2014-03-24 06:56:02 that is interesting 2014-03-24 06:56:07 my musl patch broke uclibc 2014-03-24 06:56:09 lol 2014-03-24 07:13:09 morning 2014-03-24 07:13:16 kaniini: so did the qemu patch 2014-03-24 07:13:48 i wonder if we should just jump over to musl so we dont need support both uclibc and musl in parallel 2014-03-24 07:13:53 and do that soonish 2014-03-24 07:18:39 ncopa: honestly 2014-03-24 07:18:49 ncopa: i'm all for just going all in on musl 2014-03-24 07:19:06 ncopa: i upgraded my uclibc system to musl without destroying it 2014-03-24 07:19:43 ok. good 2014-03-24 07:20:01 http://kaniini.dereferenced.org/2014/03/23/upgrading-to-alpine-musl.html 2014-03-24 07:20:05 i even wrote about how i did it 2014-03-24 07:20:29 i mean good to know 1) I am not the only one who thinks just jumping over is a good idea 2) someone has actually tested it 2014-03-24 07:20:58 i worked on the mkinitfs thingy last week 2014-03-24 07:21:03 i have all the bits there 2014-03-24 07:21:09 just integration work needs to finish 2014-03-24 07:22:19 kaniini: while i have you here, i'd like your opinion of an mkinitfs change i did. I renamed /etc/mkinitfs/files.d/* -> /etc/mkinitfs/features.d/*.files 2014-03-24 07:22:38 and /etc/mkinitfs/modules.d/* -> /etc/mkinitfs/features.d/*.modules 2014-03-24 07:22:47 i think it's a good idea 2014-03-24 07:22:55 good thanks! 2014-03-24 07:23:09 it's a lot more straightforward 2014-03-24 07:23:16 that was the only think i was a bit in doubt 2014-03-24 07:23:29 but then we are atleast 3 who is ok with it :) 2014-03-24 07:23:43 (yes your opinion counts...) 2014-03-24 07:26:43 fabled: btw, i save musl devs sent some patches to linux kernel for fixing up the header mess 2014-03-24 07:27:06 i think we could apply those and that should make ebtables and other packages just work 2014-03-24 07:27:54 so basically, i think i just need fix mkinitfs, and we are ready to switch 2014-03-24 07:27:57 to musl 2014-03-24 07:28:45 kaniini: btw, i dont think we will make any v2.8 2014-03-24 07:28:58 i agree 2014-03-24 07:29:04 lets just go straight into 3.0 2014-03-24 07:29:16 you don't have to sell me on it, i've experienced UTF-8 on IRC 2014-03-24 07:29:19 in Alpine 2014-03-24 07:29:26 :) 2014-03-24 07:29:26 i'm sold based on that alone 2014-03-24 07:29:29 lol 2014-03-24 07:29:43 instead of seeing 2014-03-24 07:29:46 ????????? 2014-03-24 07:29:50 i actually see glyphs 2014-03-24 07:30:01 there was someone who reported a python3 bug here on irc last week 2014-03-24 07:30:09 i was able to reproduce 2014-03-24 07:30:14 but on musl: no problem 2014-03-24 07:30:38 ok. i better try hurry up 2014-03-24 07:30:44 yeah, lets just do 3.0 2014-03-24 07:30:54 i'll test xen and kvm here locally 2014-03-24 07:31:30 xen does not build on uclibc currently? 2014-03-24 07:31:45 so edge builder choked? 2014-03-24 07:31:58 if so, maybe we could just disable xen on edge for now 2014-03-24 07:34:01 ncopa, interesting on the kernel header mess... 2014-03-24 07:34:09 ncopa, but yes, let's email re release plan on alpine-devel 2014-03-24 07:34:17 i'm all for doing 3.0 too 2014-03-24 07:34:33 though we should declare that 'beta' and make 2.7 LTS 2014-03-24 08:06:45 yes 2014-03-24 08:06:46 i agree 2014-03-24 08:07:01 2.7 will be supported until 3.2 or 3.3 i guess 2014-03-24 08:47:29 ncopa: ffmpeg 2.2 is out; I'll take care of it 2014-03-24 08:58:13 barthalion: super. thanks! 2014-03-24 09:23:54 fabled, barthalion, kaniini, clandmeter, rnalrd: does this look ok? https://dpaste.de/LEGY 2014-03-24 09:24:07 i plan to send that to alpine-devel list 2014-03-24 09:25:15 ncopa, looks ok to me. perhaps add that we'll do armhf too, and ask for feedback on which boards to support? 2014-03-24 09:25:27 yep, looks fine 2014-03-24 09:26:31 ncopa, k to me 2014-03-24 09:28:05 fabled: somethign like: "We are also working on armhf support for this release and would like 2014-03-24 09:28:05 some feedback on what boards people are interested in." 2014-03-24 09:28:11 is that enough? 2014-03-24 09:28:18 yes 2014-03-24 09:29:35 ncopa: 4ecf0f9e ugh 2014-03-24 09:29:57 ncopa: enabling non-free always means that the build cannot be distributed in any way 2014-03-24 09:30:12 and is not GPL anymore 2014-03-24 09:30:34 oh 2014-03-24 09:30:37 I'll remove it now 2014-03-24 09:30:42 yeah 2014-03-24 09:30:44 and do so in previous releases yet today 2014-03-24 09:30:56 ok 2014-03-24 09:31:26 what are included in the non-free? 2014-03-24 09:32:29 https://git.videolan.org/?p=ffmpeg.git;a=blob;f=LICENSE;h=1f757aa782ca1c9e25ebcb5e40a808744067a42e;hb=HEAD 2014-03-24 09:32:33 the last section 2014-03-24 09:33:15 I think it's faac in our package 2014-03-24 09:33:38 the other story is… 2014-03-24 09:33:42 /usr/lib/gcc/i486-alpine-linux-musl/4.8.2/../../../../i486-alpine-linux-musl/bin/as: out of memory allocating 4072 bytes after a total of 0 bytes 2014-03-24 09:34:04 any suggestions how to fix it? 2014-03-24 09:34:48 I've had similiar issue with upx, but I have no idea how to handle it 2014-03-24 09:35:04 barthalion, which kernel you are running? 2014-03-24 09:35:33 fabled: it's a chroot on my Arch box, 3.13.6 2014-03-24 09:35:41 barthalion, that's a problem 2014-03-24 09:35:58 barthalion, you need -grsec kernel, or alpine patched almost vanilla kernel :) 2014-03-24 09:36:18 its pie mem patch? 2014-03-24 09:36:18 it's long story 2014-03-24 09:36:20 yes 2014-03-24 09:36:25 :( 2014-03-24 09:36:28 alpine builds all PIE enabled 2014-03-24 09:36:40 and vanilla kernel does not handle them completely right; and musl does not like that 2014-03-24 09:36:49 well it's debatable if kernel is doing the right thing or not 2014-03-24 09:36:50 do we have some remote box where I could build before push? 2014-03-24 09:36:53 but musl does not like it 2014-03-24 09:37:17 http://git.alpinelinux.org/cgit/aports/tree/main/linux-grsec/fix-memory-map-for-PIE-applications.patch 2014-03-24 09:37:39 there is no way I can change my OS now, I have no disk space for that 2014-03-24 09:39:48 this silence sounds like no :D 2014-03-24 10:07:27 ncopa: is there a list of packages that fail to build with musl with the errors produced? 2014-03-24 10:09:43 royger: not really 2014-03-24 10:09:52 i have a list of pacakges that does not build 2014-03-24 10:10:03 i can fish out error messages 2014-03-24 10:10:11 but nothing automatically generated i think 2014-03-24 10:10:17 or maybe we have... 2014-03-24 10:10:53 actually 2014-03-24 10:11:01 i have per-package buildlogs 2014-03-24 10:14:00 ncopa: as you can guess I'm mainly interested in the Qemu/Xen failures :) 2014-03-24 10:15:03 :) 2014-03-24 10:15:20 i got qemu built, but it broke uclibc build 2014-03-24 10:15:36 so the qemu patches should be fixed to be included upstream 2014-03-24 10:15:57 i think kaniini worked on xen with musl this weekend 2014-03-24 10:16:32 http://git.alpinelinux.org/cgit/aports/log/main/qemu 2014-03-24 10:16:44 i reverted the musl build fix 2014-03-24 10:16:53 but the patches are there, in the git history 2014-03-24 10:17:47 http://git.alpinelinux.org/cgit/aports/commit/main/qemu?id=e68aa4b6118b001c4752c54e65dbe9829b2e362c 2014-03-24 10:17:48 hm 2014-03-24 10:17:55 patches are not that big 2014-03-24 10:20:17 ncopa: https://paste.xinu.at/j6zq/ 2014-03-24 10:21:28 a tarball with patches? 2014-03-24 11:40:45 yep, I can't build ffmpeg, thus cannot test it 2014-03-24 12:08:39 got you :-/ 2014-03-24 12:12:50 there should be no build problems if ffmpeg builds 2014-03-24 12:12:59 there is soname bump in only one library 2014-03-24 12:17:08 is there any paper explaining why musl in replacement of uClibc? 2014-03-24 12:17:52 Why the choice 2014-03-24 12:18:54 alacerda: i dont think we have anything officially written 2014-03-24 12:20:04 i can write something to ml 2014-03-24 12:21:07 i'd like to send a twitter msg about this changing. Maybe sth like: "Alpine Linux 2.7 will be the latest branch based on uClibc." 2014-03-24 12:21:25 ncopa: it would be great if you do so 2014-03-24 12:22:32 ACTION brb 2014-03-24 12:23:12 fabled: do you have some things in uclibc that we wanted to fix but figured its not worth it? 2014-03-24 12:23:30 ncopa, there's was atleast one static linking bug 2014-03-24 12:23:42 mmaped syscalls or similar? 2014-03-24 12:24:00 locales? 2014-03-24 12:24:29 ncopa, http://bugs.alpinelinux.org/issues/1579 2014-03-24 12:37:15 back 2014-03-24 12:37:22 What do you think this msg ??/ 2014-03-24 18:20:19 it's because on uclibc, -lutil has openpty 2014-03-25 05:18:51 wtffffff 2014-03-25 05:19:08 ffffffffffffffff. 2014-03-25 05:19:29 Y U DO THIS?????? 2014-03-25 05:20:38 buffer 35 2014-03-25 05:20:40 oops 2014-03-25 05:46:06 algitbot: build master 2014-03-25 05:46:24 algitbot: rebuild master 2014-03-25 05:46:25 algitbot: build master 2014-03-25 05:46:31 algitbot: build edge 2014-03-25 05:46:37 algitbot: build build-edge-x86_64 2014-03-25 08:56:12 i think i finally got mkinitfs working 2014-03-25 08:56:25 was much more tricky than originally thought 2014-03-25 09:04:50 fabled: i think we should probably rebuild our edge-musl again from scratch 2014-03-25 09:04:53 with the new gcc 2014-03-25 09:04:58 yes 2014-03-25 09:05:04 needed for relro to take affect 2014-03-25 09:05:08 maybe wait til afternoon though 2014-03-25 09:05:12 though musl does not support relro yet 2014-03-25 09:05:21 i was planning to code that feature - it's relatively simple to add 2014-03-25 09:05:22 oh 2014-03-25 09:05:23 btw, what's your opinion on using gcc snapshots? 2014-03-25 09:05:36 well 2014-03-25 09:05:36 yeah, add it to musl first 2014-03-25 09:05:49 barthalion: why you want gcc snapshots? 2014-03-25 09:05:51 my view is this 2014-03-25 09:05:58 i'd rather make clang the system compiler 2014-03-25 09:06:04 and ditch gcc altogether 2014-03-25 09:06:13 ncopa: I do not, just asking; they provide 4.x snapshots with some fixes backported 2014-03-25 09:06:42 barthalion, you mean 4.8.x -branch snapshots, or trunk snapshots ? 2014-03-25 09:06:54 4.8.x 2014-03-25 09:07:16 i dont have any strong opinions 2014-03-25 09:07:43 i think that might be ok 2014-03-25 09:07:57 and yes, would be nice to use clang as system compiler if possible, but i dont think we're there yet 2014-03-25 09:08:10 i was even considering adding linaro patchset, but decided against that after all 2014-03-25 09:08:42 but yes, clang sounds like easier to live with 2014-03-25 09:09:01 not sure how much there's difference on code generation 2014-03-25 09:09:02 then again 2014-03-25 09:09:12 grsec requires gcc since it has custom gcc plugins 2014-03-25 09:09:54 i think grsec devs are interested in clang too 2014-03-25 09:10:22 i think pipacs did some work with getting kernel compile with clang some time ago 2014-03-25 09:10:27 i dont know the status there 2014-03-25 09:37:42 I've been using clang with FreeBSD for some time now, and I have to say it works quite well 2014-03-25 14:26:05 ha! 2014-03-25 14:26:16 i booted my first musl iso image 2014-03-25 14:26:29 nice 2014-03-25 14:27:35 sweet 2014-03-25 15:12:49 ooo, will i be able to update from like 2.3 to 3 musl? 2014-03-25 15:13:59 in theory, any "update" to musl will technially be a reinstall 2014-03-25 15:14:00 probably not directly 2014-03-25 15:14:24 fabled: did you see the ODROID-U3 CE? 2014-03-25 15:18:37 o cool, in theory do I even need to update? everything works perfectly 2014-03-25 15:22:31 2.4 has only two months of support left 2014-03-25 15:22:54 and 2.5 – 8, but it's up to you 2014-03-25 15:23:09 2.7 will get longer support than initially planned 2014-03-25 15:23:11 ncopa: what about tmpfs installs? probably setup-bootable upgrade of media + reboot would do the trick? 2014-03-25 15:23:36 Frosh: so there is no hurry 2014-03-25 15:30:17 jbilyk: yes, thats what i expect, but since the current /etc/apk/repositories points to old location it might require additional steps 2014-03-25 15:30:24 right 2014-03-25 15:30:42 might need edit repositroies and apk upgrade -U -a first 2014-03-25 15:30:55 and then lbu ci and reboot 2014-03-25 15:39:07 ncopa: what about arm? 2014-03-25 15:41:55 clandmeter: what about it? 2014-03-25 15:42:02 we dont have offical arm builder yet 2014-03-25 15:42:21 and i even think that we havent decided what boards to support yet 2014-03-25 15:42:27 so no official 3 release for arm? 2014-03-25 15:42:35 dont know 2014-03-25 15:42:42 i hope we can make offical arm 2014-03-25 15:42:56 i wonder if i should buy that odroid 2014-03-25 15:42:59 its cheap and fast 2014-03-25 15:43:11 dunno 2014-03-25 15:43:31 i think fabled wanted feedback on what boards we should support 2014-03-25 15:52:43 that mkinitfs change is slightly scary 2014-03-25 15:53:08 but thats was the blocker for generating initramfs images properly with musl 2014-03-26 08:17:01 ncopa, thanks for fixing mkinitfs. what are we missing from musl build? 2014-03-26 08:17:15 i think we just make musl builders now follow git? 2014-03-26 08:17:30 and try to fix remaining packages / tag the ones that do not build 2014-03-26 08:17:45 and prepare to swap edge to musl sometime early next month? 2014-03-26 08:18:46 jfsutils is missing IIRC 2014-03-26 08:27:08 i was hoping to do the edge -> musl this week 2014-03-26 08:27:33 i think whats left is builder script 2014-03-26 08:46:51 ubuntu 14.04 will use 3.13 kernel 2014-03-26 08:47:01 i suppose we should do the same for alpine 3.0 2014-03-26 08:47:54 main total built: 1621 2014-03-26 08:47:54 main total relevant aports: 1738 2014-03-26 08:48:07 testing total built: 861 2014-03-26 08:48:07 testing total relevant aports: 1116 2014-03-26 08:48:23 3.13 sounds great, maybe I'll be able to install Alpine 3.0 on my laptop 2014-03-26 08:48:24 ncopa, which stats those are? musl-x86_64-edge? 2014-03-26 08:48:40 fabled: yes, musl x86_64 2014-03-26 08:48:51 we have new failed packages then 2014-03-26 08:48:57 i thought we had passed over 1650 before.. 2014-03-26 08:48:57 it used to be more 2014-03-26 08:49:00 yes 2014-03-26 08:49:06 x86 builds 1706 2014-03-26 08:49:08 i have fixed a few already 2014-03-26 08:49:10 parted 2014-03-26 08:49:14 sipp 2014-03-26 08:49:16 musl-x86_64 was at 1678 earlier 2014-03-26 08:49:20 ok 2014-03-26 08:49:30 sounds some dependency to many packages 2014-03-26 08:49:31 i think some are readline 2014-03-26 08:49:33 parted was readline 2014-03-26 08:49:43 ok, yes. sounds readline upgrade broke many 2014-03-26 08:50:05 I can take care of it if you give me a list 2014-03-26 08:50:17 I broke, I fix :) 2014-03-26 08:50:35 barthalion, i fixed quagga earlier already. and ruby was fixed too. 2014-03-26 08:50:39 i think most are done already 2014-03-26 08:50:42 not sure which ones are broke still 2014-03-26 08:50:44 i fixed parted today 2014-03-26 08:51:01 i have a list of broken but dont knwo if its due to readline 2014-03-26 08:51:14 readline breakage aint that bad, easy to fix 2014-03-26 08:51:20 worth it 2014-03-26 08:54:01 but we have various new packages too 2014-03-26 08:54:10 qemu, libvirt 2014-03-26 08:54:14 sipp 2014-03-26 08:54:42 ah libvirt might be x86 only, unless xen builds 2014-03-26 09:10:48 118/120 1682/1738 acf-postfix 2014-03-26 09:11:15 we are breaking that previous 1678 number of packages 2014-03-26 09:15:37 main total built: 1685 2014-03-26 09:15:37 main total relevant aports: 1738 2014-03-26 09:15:48 testing total built: 868 2014-03-26 09:15:48 testing total relevant aports: 1116 2014-03-26 09:17:01 i'd like to drop readline in exchange for bsd libedit 2014-03-26 09:17:07 same 2014-03-26 09:17:14 since some things cannot use readline 2014-03-26 09:17:23 and there's no point in having two things which have the same exact api 2014-03-26 09:17:35 is libedit exact same api? 2014-03-26 09:17:39 as far as i know 2014-03-26 09:17:43 is it api compat? 2014-03-26 09:17:45 debian made the same change already 2014-03-26 09:17:51 then we should do it 2014-03-26 09:18:02 i mean, i am sure there is some package we will have to patch 2014-03-26 09:18:13 but it is probably doing something it shouldn't, like mucking with readline internal data 2014-03-26 09:18:18 if it did, what is readline6 package doing? 2014-03-26 09:18:41 maybe there's some packages not yet changed 2014-03-26 09:18:56 i haven't followed it closely, i know that some people were working on it 2014-03-26 09:19:39 http://sprunge.us/gMYC 2014-03-26 09:19:47 packages that has readline-dev 2014-03-26 09:20:07 is libedit uptodate? 2014-03-26 09:20:26 lasts time i checked it was outdated 2014-03-26 09:20:48 depends what you mean 2014-03-26 09:21:01 newest upstream release is 20140213-3.1 2014-03-26 09:21:11 we have 20130712-3.1 2014-03-26 09:22:29 anyway 2014-03-26 09:22:42 I'd really like to see the discussion in Debian first, I can't google it 2014-03-26 09:23:33 well 2014-03-26 09:23:37 i could be hallucinating this 2014-03-26 09:23:43 but i recall irc discussion about it 2014-03-26 09:23:56 last time i looked at readline code my impression was "i want get rid of this" 2014-03-26 09:24:12 either way, we'll look at it for 3.1 2014-03-26 09:24:23 i don't want to deal with that sort of a transition right now 2014-03-26 09:24:25 well, the same applies to bash probably 2014-03-26 09:24:42 well, my opinion there is probably controversial 2014-03-26 09:24:50 i think we should make fish the default user shell 2014-03-26 09:24:52 ;p 2014-03-26 09:25:18 i'm ok to leave readline as package, but make as much as possible use libedit instead 2014-03-26 09:25:41 i'm willing to patch things to eradicate readline 2014-03-26 09:26:10 sqlite looks trivial to patch: http://trac.macports.org/ticket/31346 2014-03-26 09:26:33 of course, i'd also like toybox to be ... you're kidding 2014-03-26 09:26:38 readline is GPLv3???? 2014-03-26 09:26:57 yeah, that's just more reason to nuke that package from alpine 2014-03-26 09:27:08 the code is awful, the license is awful 2014-03-26 09:27:53 the number of packages is not that many: http://sprunge.us/gMYC 2014-03-26 09:28:09 bash i dont care if we leave to use readline 2014-03-26 09:28:56 but the rest should be doable to replace i think 2014-03-26 09:29:02 most of them 2014-03-26 09:29:33 well, i do not think it is a good idea to make packages GPLv3 as a side-effect of using readline 2014-03-26 09:30:02 that seems like a can of worms we want to get away from 2014-03-26 09:30:49 files from v2.4.11-214-ge8f74d4 uploaded 2014-03-26 09:32:30 I'm actually surprised that fish is smaller than bash 2014-03-26 09:32:42 at least on glibc system 2014-03-26 09:32:54 i havent used fish 2014-03-26 09:32:57 3.8MB vs 6.1MB 2014-03-26 09:33:25 I did, it's cool for beginners 2014-03-26 09:36:14 it's cool for non-beginners too 2014-03-26 09:36:19 it's basically like 2014-03-26 09:36:22 powershell vs. cmd.exe 2014-03-26 09:36:39 one is pleasant to use, the other is cmd.exe 2014-03-26 09:37:05 the most clever thing I've seen in fish is parsing man pages to generate tab completion 2014-03-26 09:37:23 is it snappy? 2014-03-26 09:37:32 it is pretty responsive 2014-03-26 09:37:40 i find bash completion slow me down on ubuntu 2014-03-26 09:37:54 as fast as zsh here 2014-03-26 09:38:06 anyway 2014-03-26 09:38:15 and i hate that it does not autocomplete thinkgs mlike mount -o loop image.iso 2014-03-26 09:39:12 http://i.imgur.com/pRSiO0v.png 2014-03-26 09:39:20 it autocompletes everything *but* -o loop :( 2014-03-26 09:40:05 will it aoutcomplete tar -ztf apkfile-1.0.apk ? 2014-03-26 09:40:30 bash completion things that .apks should not be completed when using tar 2014-03-26 09:40:46 yeah, it will 2014-03-26 09:40:52 ok good 2014-03-26 09:41:05 the mount -o loop 2014-03-26 09:41:22 with bash the problem is that i type first chars in file name and 2014-03-26 09:41:30 http://i.imgur.com/Ngb2KHD.png 2014-03-26 09:41:33 and it just dont want help me with it 2014-03-26 09:41:50 nice! 2014-03-26 09:43:01 anyway, i'm all for getting rid of readline (at least against non-GPL packages) 2014-03-26 09:43:15 great! 2014-03-26 09:44:18 i would argue that libsqlite pulling in libreadline is a Bad Thing, even 2014-03-26 09:44:44 fix it or simply remove readline support 2014-03-26 09:44:53 removing realine is ok with me 2014-03-26 09:45:03 replacing it with libedit is even better 2014-03-26 09:45:08 because anything libreadline touches becomes GPLv3 2014-03-26 09:45:14 bad.. 2014-03-26 09:45:25 and if it links against OpenSSL... 2014-03-26 09:46:15 now, we could argue that openssl is core system software (because apk-tools uses it) 2014-03-26 09:46:27 but, that's still awful either way 2014-03-26 09:47:02 well, we can use polarssl… 2014-03-26 09:47:32 well, i'd rather get rid of openssl too, for other reasons 2014-03-26 09:47:33 like 2014-03-26 09:47:43 a lot of the code in openssl is about the same quality as readline 2014-03-26 09:48:11 fabled: just commented openssl code recently 2014-03-26 09:48:18 i wasnt aware that it was that bad 2014-03-26 09:48:24 should we favor gnutls? 2014-03-26 09:48:38 or something else? 2014-03-26 09:48:42 http://arstechnica.com/security/2014/03/critical-crypto-bug-leaves-linux-hundreds-of-apps-open-to-eavesdropping/ 2014-03-26 09:48:51 ncopa: well, it uses uninitialized variables as 'entropy' 2014-03-26 09:49:00 :-O 2014-03-26 09:49:06 yeah i saw that gnutls bug... 2014-03-26 09:49:09 ncopa: which is just a nono 2014-03-26 09:49:41 and, i suspect with our hardening patches 2014-03-26 09:49:47 that 'entropy' is predictable 2014-03-26 09:49:54 i.e. 0 2014-03-26 09:50:39 but 2014-03-26 09:50:45 polarssl is GPLv2 2014-03-26 09:50:49 not LGPLv2 2014-03-26 09:50:56 so, i don't think it's a good replacement 2014-03-26 09:51:05 http://nacl.cr.yp.to/features.html 2014-03-26 09:51:14 simply because things shouldn't become GPL just by linking to a library 2014-03-26 09:53:50 libsodium 2014-03-26 09:54:52 kaniini, tropicssl was suggested on #musl 2014-03-26 09:54:57 https://github.com/axic/tropicssl 2014-03-26 09:55:17 well 2014-03-26 09:55:20 it's basically similar to polarssl but is BSD licensed before polarssl forked 2014-03-26 09:55:26 it's 3 years out of date 2014-03-26 09:55:40 ok 2014-03-26 09:55:48 that's true 2014-03-26 09:55:51 i mean, it seems kind of unmaintained :( 2014-03-26 09:56:06 i suppose there's no good bsd/mit/similar licensed ssl lib 2014-03-26 09:56:10 https://gitorious.org/tropicssl 2014-03-26 09:56:13 i think NSS is the better way out 2014-03-26 09:56:28 or it moved to gitorious 2014-03-26 09:56:29 it's GPL/LGPL/MPL tri-license 2014-03-26 09:56:51 last change may 2011 2014-03-26 09:57:15 ncopa, it's still last real change 1+ years ago 2014-03-26 09:57:21 ok 2014-03-26 09:57:37 anyways, replacing openssl is bigger thing 2014-03-26 09:57:42 and wont happen for 3.0 2014-03-26 09:57:45 my concern with tropicssl would be that it'd be our next uclibc 2014-03-26 09:57:54 we dont want that, yes 2014-03-26 09:58:05 libedit is possible for 3.0 2014-03-26 09:58:50 yeah - openssl is workable with, just not nice always 2014-03-26 10:49:40 list of packages that does not build with musl: http://sprunge.us/FgGF 2014-03-26 10:49:49 what should I prioritize? 2014-03-26 10:49:57 let's drop gnome 2014-03-26 10:50:04 thats what i think too 2014-03-26 10:50:05 we should package MATE for 3.1 2014-03-26 12:25:18 any plans to go with 3.13 or 3.14 on edge? 2014-03-26 12:27:09 or possibly apply the following patch to 3.12.x? https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=1f7f4dde5c945f41a7abc2285be43d918029ecc5 2014-03-26 12:29:31 uggedal: the current plan is 3.14 2014-03-26 12:30:44 ncopa: cool, I'll just keep a localy patched 3.12 for the time being 2014-03-26 12:31:16 i suppose i should upgrade the kernel soonish 2014-03-26 12:31:19 to 3.13 2014-03-26 12:31:38 or alternativly just patch with that for now 2014-03-26 12:36:06 seems it's queued to -stable 2014-03-26 14:55:09 fabled: i'm upgrading kernel and have updated x86 and x86_64 config 2014-03-26 14:55:27 oh, we probably dont use -grsec kernel for arm anyways 2014-03-26 14:55:31 nope 2014-03-26 14:55:38 musl + grsec do not yet work together 2014-03-26 14:55:38 was thinking what to do with arm, but then i dont need to worry 2014-03-26 15:04:12 ncopa, pushing 3.13? 2014-03-26 15:05:39 i will soon 2014-03-26 15:05:47 k 2014-03-26 15:05:50 will just try the 3rdparty kernel modules first 2014-03-26 15:05:55 you want a pre-patch? 2014-03-26 15:06:36 The request to API call datastore_v3.Put() was too large. 2014-03-26 15:06:43 cant prunge it 2014-03-26 15:06:48 sprunge* 2014-03-26 21:04:26 hi 2014-03-26 21:04:36 adamsutton: lo 2014-03-26 21:04:44 ACTION ist still struggling with the wlan-problem after an update of an edge box 2014-03-26 21:04:46 lol 2014-03-26 21:04:55 StarWarsFan: lo 2014-03-26 21:05:03 -t 2014-03-26 21:05:23 wlan was working very nice 2014-03-26 21:05:32 the box is running on edge for some months 2014-03-26 21:05:46 then some days ago i've updated the machine 2014-03-26 21:05:53 (113 apk's) 2014-03-26 21:06:04 and now during boot, an ip is associated 2014-03-26 21:06:11 but afterwards, the interface has no IP 2014-03-26 21:06:14 any ideas? 2014-03-26 21:06:20 here's the boot: 2014-03-26 21:06:29 http://goo.gl/2A6pFJ 2014-03-26 21:08:04 looks like it got an ip 2014-03-26 21:09:08 right 2014-03-26 21:09:12 can you sprunge ifconfig? 2014-03-26 21:09:19 but if boot is finished, it looks like this: 2014-03-26 21:09:20 http://goo.gl/PP1xnL 2014-03-26 21:09:57 nothing? 2014-03-26 21:09:59 sprunge might be a little bit difficult without net... 2014-03-26 21:09:59 i see nothing 2014-03-26 21:10:04 exactly 2014-03-26 21:10:09 StarWarsFan: you got eth0 :) 2014-03-26 21:10:17 yeah, just a second 2014-03-26 21:10:25 have to search a wire... ;-) 2014-03-26 21:10:42 the second pic doesnt work 2014-03-26 21:10:49 right 2014-03-26 21:10:53 that's the problem 2014-03-26 21:11:03 ah, mom... 2014-03-26 21:15:23 ok, got a wire now... 2014-03-26 21:17:11 here we go 2014-03-26 21:17:12 http://sprunge.us/cUII 2014-03-26 21:17:54 during boot wlan0 got 192.168.42.63 2014-03-26 21:17:59 bot now there's no ip 2014-03-26 21:19:52 are you using udev? 2014-03-26 21:20:17 hm, dmesg has a lot of these entries: 2014-03-26 21:20:18 http://sprunge.us/WGaL 2014-03-26 21:20:49 udev? if this is default, then yes. never changed something explicitly 2014-03-26 21:21:06 apk info |grep udev 2014-03-26 21:22:59 smallone:~# apk info | grep udev 2014-03-26 21:22:59 libudev 2014-03-26 21:22:59 libgudev 2014-03-26 21:22:59 udev 2014-03-26 21:23:46 so i think, yes 2014-03-26 21:26:47 ah, found this on the web: https://bbs.archlinux.org/viewtopic.php?id=164279 2014-03-26 21:30:27 is it possible that there's a mismatch between some network managers? 2014-03-26 21:38:27 the log of my accesspoint show the corresponding part: 2014-03-26 21:38:28 Mar 26 22:35:50 hostapd: wifi0vap0: STA 00:24:2b:d9:94:e3 IEEE 802.11: associated 2014-03-26 21:38:28 Mar 26 22:35:50 hostapd: wifi0vap0: STA 00:24:2b:d9:94:e3 IEEE 802.11: disassociated 2014-03-26 21:47:42 ACTION needs some sleep 2014-03-26 21:47:55 n8@all 2014-03-27 06:05:02 Morning! 2014-03-27 07:50:21 kaniini: i disabled ipfw-grsec for now, since it does not build with 3.13.7 2014-03-27 07:50:54 ncopa: i will look at it 2014-03-27 07:50:59 thanks! 2014-03-27 07:51:15 ncopa: obviously i would like to keep ipfw/netmap around 2014-03-27 07:51:16 ;) 2014-03-27 07:51:27 clandmeter: i did the same with spl/zfs 2014-03-27 07:52:07 kaniini: yeah, i figured. 2014-03-27 08:38:00 ncopa: you've read the backlog from yesterday evening? 2014-03-27 08:38:33 starting 22:04 2014-03-27 08:54:23 hi StarWarsFan 2014-03-27 08:54:24 hm 2014-03-27 08:54:33 what wlan chipset is it? 2014-03-27 08:55:39 "I'm beginning to think this is a driver or firmware issue. Not only are there many posts with similar issues with this card (intel 6235) but--" 2014-03-27 08:57:14 StarWarsFan: http://sprunge.us/WGaL 2014-03-27 08:57:26 looks like a box I had here too 2014-03-27 08:57:45 my box is an "acer aspire one" 2014-03-27 08:57:57 i think i had it with eee pc 2014-03-27 08:58:07 atheros wlan chipset 2014-03-27 08:58:09 grrr, sprunge is blocked here :-/ 2014-03-27 08:58:12 athk5 or similar 2014-03-27 08:58:40 what i find strange in your case is that it worked 113 days ago 2014-03-27 08:58:45 and now it doesnt 2014-03-27 08:59:19 what kernel driver is it for wlan0? 2014-03-27 08:59:32 not 113 days! 2014-03-27 08:59:40 113 apk's where updated 2014-03-27 08:59:48 oh :D 2014-03-27 09:00:10 do you remember what kernel you had before the update? 2014-03-27 09:00:11 i'm using it last time around two months ago 2014-03-27 09:00:31 was it 3.10 kernel before? 2014-03-27 09:00:40 it could be 3.12 broke something 2014-03-27 09:00:55 btw, my eee pc is broke 2014-03-27 09:00:58 cpu fan dont start 2014-03-27 09:01:11 so after 5mins it gets overheat and auto shut down 2014-03-27 09:02:05 I'll make a new snapshot of kernel firmware... 2014-03-27 09:02:05 i can't say which kernel... :-/ 2014-03-27 09:02:49 i just upgraded kernel to 3.13 2014-03-27 09:03:20 other possibitily could be that the wlan_supplicant timeout values are not good enough 2014-03-27 09:03:42 i saw fedora has some patch that increase timout before it diconnects or similar 2014-03-27 09:04:08 do you think you could try 3.13 kernel? 2014-03-27 09:12:18 sure, but not before the evening 2014-03-27 09:12:22 box is @home 2014-03-27 11:09:42 ncopa, can nodejs be used same as main/js ? 2014-03-27 11:10:11 coz, if js has issues building on musl, can be moved to testing or unmaintained 2014-03-27 11:10:21 and nodejs used instead 2014-03-27 11:17:44 mosquitto 3.1.1 available 2014-03-27 11:18:08 1.3.1 2014-03-27 11:34:38 what builds package g++ ? 2014-03-27 12:16:29 vkrishn: gcc builds g++ 2014-03-27 12:24:42 hmm, 3.13 or the mkinitfs changes broke my box: http://sprunge.us/jdJc 2014-03-27 12:27:01 ncopa: ↑ 2014-03-27 12:31:37 aw 2014-03-27 12:31:42 i suspect mkinitfs 2014-03-27 12:31:47 it broke my laptop too 2014-03-27 12:32:24 [25999010.742580] Failed to execute /init (error -2) 2014-03-27 12:32:24 [25999010.742688] Kernel panic - not syncing: No working init found. Try passing init= option to kernel. See Linux Documentation/init.txt for guidance. 2014-03-27 12:32:59 uggedal: can you please add the following boot options: noquiet verbose 2014-03-27 12:33:23 in my case it broke before 3.13 kernel 2014-03-27 12:33:45 and i think it was due to kernel was not included in mkinitfs 2014-03-27 12:34:21 and i though it was before the e1f4fcf83fb502adee53388e7722721f2c11d2e3 2014-03-27 12:36:10 you could possibly also add debug_init as boot option 2014-03-27 12:36:45 my laptop wasnt easy to debug since it is EFI so no console output in initfs 2014-03-27 12:37:38 i can see if i can reproduce in vm 2014-03-27 12:37:50 this is under xen on linode.com 2014-03-27 12:47:53 ncopa: http://sprunge.us/fLeW 2014-03-27 12:48:23 was unable to get more than the 100 last lines 2014-03-27 12:49:55 ncopa: this one should have all lines of kernel messages: http://sprunge.us/fGYY 2014-03-27 12:52:37 and I'm not using musl btw 2014-03-27 12:55:03 hm 2014-03-27 12:55:16 seems like it failed to find any init in initramfs 2014-03-27 12:55:50 somethign must have went wrong during initramfs creation 2014-03-27 12:57:50 i can try to extract the initramfs 2014-03-27 13:00:33 think /bin/sh might be missing and or some deps 2014-03-27 13:00:43 you could try boot with init=/bin/sh too 2014-03-27 13:01:16 ok 2014-03-27 13:01:24 i think i maybe found the problem 2014-03-27 13:02:10 ==> initramfs: creating /boot/initramfs-grsec 2014-03-27 13:02:10 cpio: ld-uClibc.so.0.9.32: No such file or directory 2014-03-27 13:02:10 18405 blocks 2014-03-27 13:02:10 OK: 333 MiB in 78 packages 2014-03-27 13:02:37 so, what I have done in mkinitfs is to change the way things are copied 2014-03-27 13:02:50 in stead of for i in $every_file 2014-03-27 13:02:59 i use cpio to copy the dir tree 2014-03-27 13:03:17 but, if cpio fails to find a file, things will boom 2014-03-27 13:03:23 i think thats what happened 2014-03-27 13:03:31 cpio: ld-uClibc.so.0.9.32: No such file or directory 2014-03-27 13:03:50 i got that error 2014-03-27 13:05:17 i think i need echo loud warnings if cpio fails 2014-03-27 13:05:25 WARNING!!!! 2014-03-27 13:05:41 WARNING!!!! your box will not boot!!!!! 2014-03-27 13:08:29 + lddtree -R / -l --no-auto-root /bin/busybox /bin/sh /lib/libcrypt-0.9.33.2.so /lib/libcrypt.so.0.9.32 /lib/libcrypto.so.1.0.0 /lib/libdl-0.9.33.2.so /lib/libdl.so.0.9.32 /lib/libm-0.9.33.2.so /lib/libm.so.0.9.32 /lib/libuClibc-0.9.33.2.so /lib/libc.so.0.9.32 /lib/libz.so.1 /lib/libz.so.1.2.8 /lib/ld-uClibc-0.9.33.2.so /lib/ld-uClibc.so.0.9.32 /lib/mdev/usbdev /lib/mdev/usbdisk_link /lib/mdev/xvd_links /lib/mdev/ide_links /lib/mdev/dvbdev 2014-03-27 13:08:29 /sbin/apk /etc/modprobe.d/aliases.conf /etc/modprobe.d/blacklist.conf /etc/modprobe.d/i386.conf /etc/modprobe.d/kms.conf /etc/mdev.conf 2014-03-27 13:08:29 cpio: ld-uClibc.so.0.9.32: No such file or directory 2014-03-27 13:11:22 ok 2014-03-27 13:11:27 i can reproduce 2014-03-27 13:11:31 i'm getting closer 2014-03-27 13:11:40 a-edge:~# lddtree -R / -l --no-auto-root /lib/libcrypt-0.9.33.2.so /lib/libcrypt 2014-03-27 13:11:40 .so.0.9.32 2014-03-27 13:11:53 a-edge:~# 2014-03-27 13:12:12 a-edge:~# lddtree -R / -l --no-auto-root /lib/libcrypt-0.9.33.2.so /lib/libcrypt 2014-03-27 13:12:12 .so.0.9.32 | sprunge 2014-03-27 13:12:12 http://sprunge.us/hHaI 2014-03-27 13:12:36 as you see in that sprunge link, there is an: /ld-uClibc.so.0.9.32 2014-03-27 13:12:48 that is what chokes cpio 2014-03-27 13:13:31 its a bug in lddtree in other words 2014-03-27 13:13:33 hm 2014-03-27 13:17:01 it's giving wrong file name 2014-03-27 13:17:02 ? 2014-03-27 13:17:31 libdl.so.0.9.32 => /lib/libdl.so.0.9.32 (interpreter => /ld-uClibc.so.0.9.32) 2014-03-27 13:17:51 it probably gets interpreter wrong 2014-03-27 13:18:01 i think its the script 2014-03-27 13:18:08 that prefixes with $ROOT 2014-03-27 13:18:31 well 2014-03-27 13:19:23 # scanelf -qF '#F%i' /lib/libcrypt-0.9.33.2.so 2014-03-27 13:19:23 ld-uClibc.so.0.9.32 2014-03-27 13:19:32 there is no / prefixed 2014-03-27 13:20:33 hi been in RHEL land what alpine linux 2014-03-27 13:21:29 getting closer: http://sprunge.us/fdfL 2014-03-27 13:21:39 hi madlat, welcome 2014-03-27 13:22:31 thanks 2014-03-27 13:22:34 elf_specs=$(elf_specs "${resolved}") 2014-03-27 13:22:34 interp=$(scanelf -qF '#F%i' "${resolved}") 2014-03-27 13:22:34 [ -n "${interp}" ] && interp="${ROOT}${interp#/}" 2014-03-27 13:22:34 2014-03-27 13:23:19 seems like interp gets wrong when its not full path 2014-03-27 13:24:00 it unconditionally prefixes with $ROOT 2014-03-27 13:27:46 fabled: do you know how that interp thing is uspposed to work? 2014-03-27 13:27:55 ncopa, what do you mean? 2014-03-27 13:28:03 if the interp is not an absolute path 2014-03-27 13:28:12 no. it should be absolute path always. 2014-03-27 13:28:22 its not in uclibc 2014-03-27 13:28:31 ? 2014-03-27 13:28:34 thats why it breaks 2014-03-27 13:28:49 scanelf -q -i /lib/* 2014-03-27 13:28:59 sounds like bug in uclibc build then 2014-03-27 13:29:04 iterp is: ld-uClibc.so.0.9.32 2014-03-27 13:29:12 then again, it does not matter since interp is not used for .so files 2014-03-27 13:29:30 basically 2014-03-27 13:29:35 what it does is 2014-03-27 13:29:42 on executing file 2014-03-27 13:29:58 kernel will check if the ELF has interp set, if yes, it will load the 'interp' file and use that to execute the executable 2014-03-27 13:30:32 and for normal ELFs it points to dynamic linker? 2014-03-27 13:30:40 ld.so 2014-03-27 13:30:52 dynamic linked apps that is 2014-03-27 13:31:19 yes 2014-03-27 13:31:28 on .so files you don't need it 2014-03-27 13:31:33 since they are not intended to be run directly 2014-03-27 13:31:51 but uclibc set it to something bogus 2014-03-27 13:32:01 so uclibc set it to something bogus 2014-03-27 13:32:10 but noone cares since its not used anyways 2014-03-27 13:32:24 it's also possible to use "libc.so /path/to/executable" 2014-03-27 13:32:27 except lddtree who actually use it to scan for elf deps 2014-03-27 13:32:28 to test out new dynamic loader 2014-03-27 13:32:33 that's why interp is set in libc 2014-03-27 13:33:03 it's basically feature for testing new library without installing it as system default 2014-03-27 13:33:08 new c-library* 2014-03-27 13:33:14 interesting 2014-03-27 13:33:25 i wonder if that wors at all in uclibc then 2014-03-27 13:33:29 probably not 2014-03-27 13:33:42 since interp is not absolute path 2014-03-27 13:33:42 $ /lib/ld-uClibc.so.0.9.32 /bin/busybox 2014-03-27 13:33:42 Killed 2014-03-27 13:34:01 and its kernel that does does that, right? 2014-03-27 13:34:06 probably 2014-03-27 13:34:07 yes 2014-03-27 13:34:15 you just told me it is... 2014-03-27 13:34:22 kernel will check if the ELF has interp set, if yes, it will load the 'interp' file and use that to execute the executable 2014-03-27 13:34:25 yes 2014-03-27 13:34:33 i meant 'probably killed by kernel due interp missing' 2014-03-27 13:34:37 and kernel will not look in any $PATH 2014-03-27 13:35:00 yes 2014-03-27 13:35:13 and it'd be security issue, etc. 2014-03-27 13:35:20 interp needs to be trusted 2014-03-27 13:35:23 makes sense 2014-03-27 13:35:24 sigh 2014-03-27 13:35:39 i'd just ignore interp for anything in /lib 2014-03-27 13:35:45 or ignore if it does not exist 2014-03-27 13:35:54 i'd love to switch to a more correct libc... 2014-03-27 13:35:56 :) 2014-03-27 13:36:11 yeah. 2014-03-27 13:36:26 will hopefully save us for stupidity like this 2014-03-27 13:36:29 yes 2014-03-27 13:36:44 i think that what i will do is 2014-03-27 13:36:48 did it make cpio stop? 2014-03-27 13:36:54 yes 2014-03-27 13:36:57 lol 2014-03-27 13:37:09 because i use cpio to copy dir struct 2014-03-27 13:37:12 yes 2014-03-27 13:37:13 performance reasons 2014-03-27 13:37:38 i used to loop the dir struct and mkdir ${file%/*}; cp $file... 2014-03-27 13:37:46 copy with cpio is faster 2014-03-27 13:37:52 well 2014-03-27 13:37:54 quick fix 2014-03-27 13:38:05 lddtree, in LIST mode should list $interp only if it exists 2014-03-27 13:38:06 filter non-existing files 2014-03-27 13:38:34 list $interp only if it exists and has leading / 2014-03-27 13:38:41 right 2014-03-27 13:38:50 slightly hackish, but works 2014-03-27 13:38:57 well 2014-03-27 13:39:17 if it does not start with / then will not kernel accept it anyways, right? 2014-03-27 13:39:24 i think so 2014-03-27 13:39:36 but it started with / 2014-03-27 13:39:43 nope it didnt 2014-03-27 13:40:04 oh 2014-03-27 13:40:04 the lddtree script prefixed it with $ROOT which happened to be / 2014-03-27 13:40:22 not sure what kernel does then 2014-03-27 13:40:30 interp=$(scanelf -qF '#F%i' "${resolved}") 2014-03-27 13:40:30 [ -n "${interp}" ] && interp="${ROOT}${interp#/}" 2014-03-27 13:40:37 yes 2014-03-27 13:40:38 $ scanelf -qF '#F%i' /lib/libc.so.0.9.32 2014-03-27 13:40:38 ld-uClibc.so.0.9.32 2014-03-27 13:40:49 i suppose thats was my real question 2014-03-27 13:41:02 what does kernel when interp is not absolute path 2014-03-27 13:41:23 i think in any case, non-absolute path can be considered 'invalid' 2014-03-27 13:44:34 seems kernel considers it relative to current working directory 2014-03-27 13:44:49 so non-absolute interp, is absolutely bogus 2014-03-27 13:45:07 $ /lib/libc.so.0.9.32 /bin/busybox 2014-03-27 13:45:08 ash: /lib/libc.so.0.9.32: not found 2014-03-27 13:45:25 /lib$ /lib/libc.so.0.9.32 /bin/busybox 2014-03-27 13:45:25 Killed 2014-03-27 13:46:15 [ 4082.925615] PAX: From 10.65.65.1: execution attempt in: , 00000000-000da000 00000000 2014-03-27 13:46:15 [ 4082.925619] PAX: terminating task: /lib/ld-uClibc-0.9.33.2.so(ld-uClibc.so.0.):4715, uid/euid: 0/0, PC: 00006fa0, SP: 5b667484 2014-03-27 13:46:36 could be pax that forbids that 2014-03-27 13:46:49 and it maybe works with vanilla kernel 2014-03-27 13:47:22 but i think we can safely filter out interps with non-absolute path 2014-03-27 13:47:53 my guess that it does work with vanilla kernel 2014-03-27 13:48:02 and that for uclibc devs its a feature 2014-03-27 13:48:31 because then they can run things directly from their working tree (given the .so ends up in there) 2014-03-27 13:48:44 so 2014-03-27 13:48:47 fixes: 2014-03-27 13:49:05 1) filter non-absolute interps 2014-03-27 13:49:26 2) verify that file exists and if not, echo to stderr instead of stdout 2014-03-27 13:49:55 3) check exist status of cpio copy and exit with big noise if it fails 2014-03-27 13:50:08 right 2014-03-27 13:50:24 any of those would have save uggedal from a non-bootable box 2014-03-27 13:51:53 ncopa: there was another person in #alpine-linux with similar issue. 2014-03-27 14:32:19 ok, that lddtree update should fix it 2014-03-27 14:32:42 apk upgrade -U and apk fix mkinitfs 2014-03-27 14:38:07 ugh 2014-03-27 14:38:17 ncopa, mkinitfs got updated, so no separate 'fix' needed 2014-03-27 14:38:17 problem is bigger than that 2014-03-27 14:38:29 yeah 2014-03-27 14:38:32 thats why i did it 2014-03-27 14:38:34 right 2014-03-27 14:38:39 but its still broken 2014-03-27 14:38:39 so something is still broke? 2014-03-27 14:38:42 yes 2014-03-27 14:38:47 dunno what 2014-03-27 14:38:57 but i think its the thing that broke my musl laptop 2014-03-27 14:39:20 but since i have it in qemu without EFI i can now troubleshoot 2014-03-27 14:40:33 ncopa, does mkinitfs on uclibc now work? as in, is the further issue musl specific or not? 2014-03-27 14:41:31 i think its broken everywhere 2014-03-27 14:41:43 it think its kernel module deps that broke 2014-03-27 14:42:06 ext4 module is included 2014-03-27 14:42:10 but cannot be modprobe 2014-03-27 14:42:33 so it must be the awk script in mkinitfs that autodetects kernel module deps thats broken 2014-03-27 14:42:53 why would it suddenly break 2014-03-27 14:43:16 because i messed with the code 2014-03-27 14:43:36 and did not test well enough it was actually working 2014-03-27 14:43:41 oops 2014-03-27 14:46:16 http://git.alpinelinux.org/cgit/mkinitfs/commit/mkinitfs.in?id=b5a2220b10ae071cb518ef1301a04710efb3341a 2014-03-27 14:46:22 i guess its there 2014-03-27 14:46:25 it was broken already 2014-03-27 14:46:36 but there are many tricky cornercases 2014-03-27 15:33:19 ok 2014-03-27 15:33:23 that should fix 2014-03-27 15:33:33 anyone has an edge box not yet rebooted? 2014-03-27 15:33:39 and are willing to test it for me? 2014-03-27 15:33:46 I've, on virtualbox 2014-03-27 15:34:02 my laptop is unbootable 2014-03-27 15:34:10 my virtual edge is unbootable 2014-03-27 15:34:17 have not had time to rescue those yet 2014-03-27 15:34:37 doing upgrade right now 2014-03-27 15:34:37 i could test on my work desktop but if it breaks I'll not reach to fix it today... 2014-03-27 15:34:41 before you reboot 2014-03-27 15:34:44 please do this: 2014-03-27 15:34:56 gunzip -c /boot/initramfs-grsec | cpio -t | grep -E 'fs/(ext|mb)' 2014-03-27 15:35:12 k 2014-03-27 15:35:12 if you use ext? that is 2014-03-27 15:35:27 it should list ext4 and 2014-03-27 15:35:39 mbcache.ko 2014-03-27 15:35:59 problem was that the dependency mbcache.ko was not included in the initramfs 2014-03-27 15:37:10 both are here 2014-03-27 15:37:32 (although it's linux-virt-grsec if it make a difference) 2014-03-27 15:37:39 no, should be the same 2014-03-27 15:37:47 but your root is on ext4 right? 2014-03-27 15:37:50 yes 2014-03-27 15:37:52 ok 2014-03-27 15:37:55 and rebooted succesfully 2014-03-27 15:37:55 try reboot 2014-03-27 15:37:59 phew! 2014-03-27 15:38:01 thanks! 2014-03-27 15:38:10 np :) 2014-03-27 15:38:11 i'll send an email about it 2014-03-27 15:39:29 ncopa, good work. 2014-03-27 15:42:15 not really... 2014-03-27 15:42:24 i should have tested more before pushing things to edge... 2014-03-27 18:00:05 ncopa, cool. edge-musl needs updating too, though. 2014-03-27 18:02:04 I think 2014-03-27 18:02:23 oh there we go 2014-03-27 18:03:04 was using a stale mirror :) 2014-03-27 18:04:10 cpio: --no-auto-root Do not automatically prefix input ELFs with ROOT: No such file or directory 2014-03-27 18:04:11 ncopa, ^ wtf 2014-03-27 18:04:46 ncopa, then it says "1 blocks" 2014-03-27 18:04:50 ncopa, this is on edge-musl 2014-03-27 18:06:10 Elizafox, which arch? 2014-03-27 18:06:15 x86_64 2014-03-27 18:06:22 note lddtree just updated 2014-03-27 18:06:33 right. it should be updated on x86_64 2014-03-27 18:06:41 lddtree-1.25-r0 2014-03-27 18:06:49 along with mkinitfs 2014-03-27 18:10:38 ncopa, using the magic of set -x, it's trying to run lddtree -R / --no-auto-root 2014-03-27 18:10:51 with no other ELF files 2014-03-27 18:10:51 wtf 2014-03-27 20:18:37 ACTION prods ncopa  2014-03-27 20:19:09 ACTION hangs a piece of cheese in front of ncopa :3 2014-03-27 20:19:38 lol 2014-03-27 20:20:35 ACTION thinks ncopa is just sleeping :P 2014-03-27 20:45:16 jbilyk: or he's not 2014-03-27 20:58:44 heh 2014-03-27 21:05:47 re 2014-03-27 21:06:02 ncopa: i've updated the box with the wlan-problem 2014-03-27 21:06:11 but no change, same problem :-( 2014-03-27 21:06:39 before the update: 3.12.8-1-grsec #2-Alpine SMP Tue Mar 18 16:15:58 UTC 2014 i686 Linux 2014-03-27 21:06:49 now: 3.13.7-0-grsec #1-Alpine SMP Thu Mar 27 07:48:51 UTC 2014 i686 Linux 2014-03-27 21:07:12 but dmesg is still flooded with these messages: 2014-03-27 21:07:56 http://sprunge.us/LGCX 2014-03-27 22:24:50 n8@all 2014-03-28 06:37:15 moinmoin 2014-03-28 06:45:18 that's a wiki! 2014-03-28 06:56:11 :) 2014-03-28 06:59:55 StarWarsFan: do you know what wlan driver it is? 2014-03-28 07:04:13 StarWarsFan: do you use WEP or WPA? Do you use ipv6? 2014-03-28 07:05:22 "So in my case, the error=3 error has occured when I’ve specified one type of encryption, and wpa_supplicant didn’t like it. Possible because my wireless router doesn’t support it? Not sure why. Obscure." 2014-03-28 07:09:57 StarWarsFan: https://bugs.gentoo.org/show_bug.cgi?id=501828 2014-03-28 07:10:07 do you think you could try get a log from wpa_supplicant? 2014-03-28 08:39:24 re 2014-03-28 08:41:53 ncopa: will try to get a log 2014-03-28 08:42:10 unfortunately i forget to check for the chipset yesterday evening :-/ 2014-03-28 08:42:30 im using WPA but not ipv6 2014-03-28 08:42:37 i think its something that got triggered by wpa_supplicant 2.1 2014-03-28 08:42:45 gentoo has a bug there 2014-03-28 08:43:01 i also see that fedora has some patches for wpa_supplicant, but they are still on 2.0 2014-03-28 08:43:36 so i think what would be nice is to try get som wpa_supplicant logging 2014-03-28 08:43:42 verbose logging of some sort 2014-03-28 08:43:52 and the google error message from there 2014-03-28 08:44:00 then we'll likely find the correct patch 2014-03-28 08:44:38 other option is that we just try random patches, but i'd prefer have a more specific plan 2014-03-28 08:47:22 StarWarsFan: do you use WPA Enteprise (EAP?) 2014-03-28 08:47:31 https://bugs.archlinux.org/task/38996 2014-03-28 08:47:37 thats the only patch arch linux has applied 2014-03-28 08:48:36 have to check in the evening. long(er) time ago i've setup the box... 2014-03-28 08:51:24 but all in all i used the guide from http://wiki.alpinelinux.org/wiki/Connecting_to_a_wireless_accesspoint to setup the box 2014-03-28 08:53:01 yeah, but i think there are regressions in the wpa_supplicant 2014-03-28 08:53:10 depending on the specific crypto you are using 2014-03-28 08:53:16 i dont think you use EAP 2014-03-28 08:53:27 it seems to be certificate check 2014-03-28 08:53:49 (eg you install certificates on the clients instead of a password) 2014-03-28 09:00:10 main total built: 1692 2014-03-28 09:00:10 main total relevant aports: 1737 2014-03-28 09:00:18 thats musl x86_64 2014-03-28 12:49:07 finally got go and docker running on alpine 2014-03-28 12:49:54 thankfully musl made some patches unneccesary 2014-03-28 12:51:44 uggedal: sorry about making your box unbootable 2014-03-28 12:52:02 i wonder if i should write a short incident report on alpine-devel mailing list 2014-03-28 12:52:13 how it happened 2014-03-28 12:54:14 i'm working on the git hook that generates those tarballs 2014-03-28 12:56:22 gonna make another 2014-03-28 13:22:37 i have a new buildrepo script 2014-03-28 13:22:43 ithink i should push it now 2014-03-28 13:26:53 uggedal: care to send your patches to alpine-devel? 2014-03-28 13:27:02 ncopa: np, the fix was easy after you found the bugs 2014-03-28 13:27:09 as we'll switch edge to musl soon, uclibc shouldn't be an issue 2014-03-28 13:27:47 uggedal: i'm glad to here you could fix it 2014-03-28 13:27:58 it shouldnt have happened, but it did 2014-03-28 13:28:03 there were 2 bugs 2014-03-28 13:28:16 i tested mkinitfs and booting new iso in musl 2014-03-28 13:28:24 and i didnt test on uclibc before pushing 2014-03-28 13:28:27 barthalion: they need some polish, but I'll send them soon 2014-03-28 13:28:37 i assumed if it works on musl it'l work on uclibc 2014-03-28 13:28:46 and one of the bugs only affected uclibc 2014-03-28 13:29:04 ncopa: I'm on musl now ;) 2014-03-28 13:29:13 second issue was kernel module dependency 2014-03-28 13:29:33 but all kernel modules needed for booting iso are listed 2014-03-28 13:30:08 so it booted my iso 2014-03-28 13:30:31 but ext4 module deps was not pulled in 2014-03-28 13:30:35 so boot from disk broke 2014-03-28 13:30:54 uggedal: how did the migration to musl go? 2014-03-28 13:31:28 i think i should migrate my main work desktop too 2014-03-28 13:32:40 ok i have another scary git push coming up 2014-03-28 13:32:45 new buildrepo in lua 2014-03-28 13:32:58 it should be faster than current 2014-03-28 13:33:04 ncopa: your desktop is xfce? 2014-03-28 13:33:15 it shoudl be able to show progress status 2014-03-28 13:33:22 Frosh: yes 2014-03-28 13:34:24 ncopa: without any issues. had to apk fix libc and some 2 other packages 2014-03-28 13:34:24 ACTION just set a network status display at $$ with alpine XFCE desktop - works well enough for that on current 2.7 x86_64 2014-03-28 13:35:04 i use xfce alpine musl on my priv laptop 2014-03-28 13:35:07 booting EFI 2014-03-28 13:35:12 barthalion: the unfinished abuilds for go and docker are here: https://github.com/uggedal/aports/tree/master/extra 2014-03-28 13:35:50 does go build without ucontext? 2014-03-28 13:36:12 oh we have ucontext in uclibc i think 2014-03-28 13:36:42 ncopa: no idea 2014-03-28 13:37:28 that is, it builds fine on musl but some tests fail 2014-03-28 13:38:03 and I had to force -fno-PIC when go build links with ld 2014-03-28 13:38:29 ok 2014-03-28 13:38:34 but it built with musl 2014-03-28 13:38:35 nice 2014-03-28 13:39:47 ncopa: I actually got it building with uclibc last year, but had to patch more stuff 2014-03-28 13:40:13 ncopa: and never got linking with cgo working until I found out about -fno-PIC 2014-03-28 13:41:35 nice 2014-03-28 13:41:47 but i assueme that go does not produce PIEs 2014-03-28 13:41:55 position independent executables 2014-03-28 13:42:16 we got gcc-go built earlier 2014-03-28 13:42:35 but i think we dropped it for now due to missing ucontext in musl 2014-03-28 13:44:00 I don't really understand the whole situation but I think go produces ELFs which are neither PIE nor PIC 2014-03-28 13:46:01 http://code.google.com/p/go/issues/detail?id=6940 2014-03-28 13:47:05 under openbsd go links with -nopie (as recommended on the Hardened Gentoo wiki), but I could not get it to work 2014-03-28 13:48:50 "The Go linker does not currently support building objects that may be linked into a PIE." 2014-03-28 13:50:41 so yes, we need -no-pie and -fno-PIC for now 2014-03-28 13:50:53 would be nice to add support for it at some point 2014-03-28 13:50:57 but not today :) 2014-03-28 13:59:50 ok, i'm gonna replace the buildrepo script now 2014-03-28 13:59:56 i hope it does not break the builders... 2014-03-28 14:00:06 scare since i replace it while its building... 2014-03-28 14:00:09 scary... 2014-03-28 14:01:05 so far so good 2014-03-28 14:03:00 bah 2014-03-28 14:07:11 i think it works now 2014-03-28 14:07:34 single package builds should be noticeable faster 2014-03-28 14:11:09 nice 2014-03-28 14:13:59 ha! 2014-03-28 14:14:09 deadly fast 2014-03-28 14:14:25 i think ti biggest bottleneck is running the configure script 2014-03-28 14:14:46 im gonna look for coffe 2014-03-28 14:49:59 ncopa, would it be better to have nodejs with --shared-v8 ? 2014-03-28 14:50:05 no 2014-03-28 14:50:28 i dunno 2014-03-28 14:50:29 having anything "shared" is stupid, maybe openssl is an exception 2014-03-28 14:50:34 and compile v8 with shell + libv8.so 2014-03-28 14:50:46 nodejs is good at keeping their static deps up to date 2014-03-28 14:50:47 so shell give functionality like js 2014-03-28 14:51:02 build v8 from nodejs dir 2014-03-28 14:51:05 actually as much as possible shared is generally a good idea 2014-03-28 14:51:11 not in nodejs case 2014-03-28 14:51:12 but in v8 case i'm in doubt 2014-03-28 14:51:31 because if your v8 is not the very same version that nodejs use, it'll be terribly broken 2014-03-28 14:51:32 build v8 from nodejs dir, tired and works 2014-03-28 14:51:46 barthalion, build v8 from nodejs dir 2014-03-28 14:51:53 as long nodejs is the only that depends on v8 then it does matter 2014-03-28 14:52:00 vkrishn: building v8 as shared from nodejs sources is pointless 2014-03-28 14:52:07 since I noticed main/js has no deps at present in mains 2014-03-28 14:52:09 anyway, been there, done that 2014-03-28 14:52:11 so should be ok 2014-03-28 14:52:20 don't use shared v8 please 2014-03-28 14:52:57 not unless we ship v8 and v8-dev as subpackages for nodejs 2014-03-28 14:53:01 point is to able to use v8 shell 2014-03-28 14:53:07 but sounds ugly in any case 2014-03-28 14:53:33 vkrishn: there is no point of packaging v8 from nodejs sources, when newer release is availabl 2014-03-28 14:53:36 and move main/js to testing or unmaintained 2014-03-28 14:53:36 available 2014-03-28 14:54:19 barthalion, if we see newness of v8 as more important then yes 2014-03-28 14:54:37 but it would like have not depends than nodejs 2014-03-28 14:54:45 likely have no* 2014-03-28 14:55:24 note that main/js does not build for musl presently 2014-03-28 14:55:39 so let's drop it, if you say that nothing depends on it 2014-03-28 14:56:58 extra is shell cmd 2014-03-28 14:57:51 ncopa: is it possible to override the pkgver of subpackage? 2014-03-28 15:01:11 barthalion: nope 2014-03-28 15:01:13 just for info, newer v8 and icu4c as depend 2014-03-28 15:01:18 well, ok, let's leave it 2014-03-28 15:01:23 we can package v8 separately 2014-03-28 15:01:23 has* 2014-03-28 15:01:44 ok, its already in testing 2014-03-28 15:01:58 vkrishn: yes but it does not build either i think 2014-03-28 15:02:05 i looked at it a couple of weeks ago 2014-03-28 15:02:11 recent v8 is *horrible* to build 2014-03-28 15:02:15 gyp blah 2014-03-28 15:02:33 ncopa, it depends on its own flavor of icu4c 2014-03-28 15:02:44 so, i actually tried to build a shared v8 2014-03-28 15:02:46 hosted at google code 2014-03-28 15:02:47 but gave up... 2014-03-28 15:02:51 urgh 2014-03-28 15:02:53 ok 2014-03-28 15:03:11 google is on the fork-everything-and-your-mother path 2014-03-28 15:05:32 newer monit v5.8 available update pls 2014-03-28 15:07:07 need to have gyp installed in v8/build dir 2014-03-28 15:08:25 see dependencies in Makefile 2014-03-28 15:10:27 hum 2014-03-28 15:11:35 moar spam 2014-03-28 15:11:47 sorry bout that 2014-03-28 15:11:56 np 2014-03-28 15:12:01 we are bringing musl builders online to follow git master 2014-03-28 15:12:06 some clever irssi plugin takes care of it 2014-03-28 15:12:08 needs little finetuning still 2014-03-28 15:12:51 vkrishn: I'm doing v8 test build locally now 2014-03-28 15:12:57 gee, looks fancy on clog`ed mosquitto 2014-03-28 15:13:05 barthalion, ok 2014-03-28 15:13:12 I should just write PKGBUILD→APKBUILD converter one day 2014-03-28 15:13:30 or at least something more automated than I do now when I'm lazy 2014-03-28 15:16:46 musl x86 and x86_64 builders should be online now 2014-03-28 15:16:51 and i think they work too 2014-03-28 15:18:47 how do I paste image , any pastebin ? 2014-03-28 15:19:36 https://imgur.com/ 2014-03-28 15:20:45 http://imgur.com/LawUaOt 2014-03-28 15:23:09 algitbot: build master 2014-03-28 15:45:29 algitbot: build master 2014-03-28 15:46:19 algitbot: build master 2014-03-28 15:51:52 codegen.cc:(.text._ZN2v88internal13CodeGenerator16MakeCodeEpilogueEPNS0_14MacroAssemblerEjPNS0_15CompilationInfoE+0x5f): undefined reference to `v8::internal::Assembler::GetCode(v8::internal::CodeDesc*)' 2014-03-28 15:52:02 so this is all I guess, at least about newest release 2014-03-28 16:07:18 barthalion, v8-3.24 ? 2014-03-28 16:18:10 whee 2014-03-28 16:18:13 arm builder online too 2014-03-28 16:18:16 it's slow 2014-03-28 16:18:18 but works 2014-03-28 16:18:22 and follows git masters 2014-03-28 16:18:33 the build logs are not available though 2014-03-28 16:18:37 it's natted box 2014-03-28 16:23:06 the arm box has also NIC issue so network connectivity is slow for downloading. need to patch kernel. 2014-03-28 16:23:59 relocating 2014-03-28 16:36:24 huh 2014-03-28 16:39:13 i killed it 2014-03-28 16:39:18 something is not right 2014-03-28 16:40:04 need to look at it later 2014-03-28 20:29:11 ncopa: ok, trying to determine the wlan chipset now... 2014-03-28 20:32:39 http://en.wikipedia.org/wiki/Acer_Aspire_One 2014-03-28 20:32:50 Atheros 802.11b/g WLAN 2014-03-28 20:33:55 lspci -k tell's me: 2014-03-28 20:34:06 02:00.0 Class 0200: 168c:001c ath5k 2014-03-28 20:35:16 seems to be "Atheros AR5007EG" 2014-03-28 20:52:11 ok, here's the debug output: 2014-03-28 20:52:12 http://sprunge.us/FEBN 2014-03-28 20:52:56 used ssid should be "swf" 2014-03-28 22:09:52 hello guys :) 2014-03-28 22:29:36 was wondering if anyone tried adopting zfs on linux packages on alpine ? 2014-03-28 22:39:43 le^zeek: we used to have zfs modules, they're disabled temporarily 2014-03-28 22:39:59 any special reason ? 2014-03-28 22:40:18 I mean on edge 2014-03-28 22:40:21 if there's some show stopper and noone looking at it, i could probably try to help 2014-03-28 22:40:38 yeah, build failures, but everything should be fine on stable releases 2014-03-28 22:41:05 doing some weird setup for a internal project, and found alpine really awesome for it 2014-03-28 22:41:15 playing around with it for a good part of last week :) 2014-03-28 22:41:31 :) 2014-03-30 22:54:04 ncopa, do you guys have a alpine-musl image yet? 2014-03-31 01:03:58 blueness: http://dev.alpinelinux.org/~tteras/alpine-140103-x86_64.iso 2014-03-31 01:04:16 thats the only musl iso i know of 2014-03-31 01:04:31 and x86 http://dev.alpinelinux.org/~tteras/alpine-140103-x86.iso 2014-03-31 03:22:54 has anyone encrypted their root partition during the installation 2014-03-31 06:24:50 blueness: we don't have any official iso yet 2014-03-31 06:29:23 we should probably be able to tag a development snapshot iso today though 2014-03-31 07:19:51 thanks arch 2014-03-31 07:19:56 did you know that 2014-03-31 07:20:03 arch's systemd 2014-03-31 07:20:07 depends on packages in extra 2014-03-31 07:20:11 top quality 2014-03-31 07:21:23 kaniini: [core] is completely self-contained, you probably mean optdepends or build deps 2014-03-31 07:24:50 barthalion: false. 2014-03-31 07:24:55 barthalion: systemd depends on libseccomp 2014-03-31 07:25:00 barthalion: libseccomp is in extra 2014-03-31 08:04:48 files from v2.4.11-215-gef9cae4 uploaded 2014-03-31 08:06:54 doh 2014-03-31 08:07:07 clamav initd is broken from edge 2014-03-31 08:07:14 eheh 2014-03-31 08:08:11 it seems none needs an antivirus 2014-03-31 08:11:52 rnalrd: http://is.gd/diE9pY 2014-03-31 08:12:13 is it main.cvd or main.cld? 2014-03-31 08:12:20 cld 2014-03-31 08:12:26 ok 2014-03-31 08:12:38 i think is used to be cvd, but got renamed 2014-03-31 08:13:11 ok 2014-03-31 08:18:19 cool 2014-03-31 08:18:21 ^^^ 2014-03-31 08:18:24 \0/ 2014-03-31 08:20:38 mm 2014-03-31 08:21:19 clamav web sites reports main.cvd, but on my boxes there's no trace of it 2014-03-31 08:24:00 k 2014-03-31 08:24:16 main.cvd is downloaded from web site 2014-03-31 08:25:00 please bear with bogus musl-arm messages. we are looking what's wrong. 2014-03-31 08:25:25 when it gets updated it becomes main.cld 2014-03-31 08:34:00 ... ok we know what is broke in arm. 2014-03-31 08:34:10 hope we get the thing online today 2014-03-31 08:50:11 kaniini: I'll move it for you 2014-03-31 09:14:24 barthalion: i already worked aorund it 2014-03-31 12:31:15 cool 2014-03-31 12:32:17 i'd like an mqtt plugin for conky 2014-03-31 13:41:01 ncopa, I am planning to test xapian and sphinxsearch, is it ok if I request for both to be moved to "mains" once test goes ok ? 2014-03-31 13:41:42 next major version of sphinxsearch has some nice features planned 2014-03-31 13:43:39 fcolista, can sphinxsearch be updated to 2.1.7 pls 2014-03-31 13:44:23 sure vkrishn 2014-03-31 13:44:29 thanks 2014-03-31 13:48:52 and xapian too - v1.2.17 2014-03-31 13:49:11 yes, it's building 2014-03-31 13:49:27 vkrishn: use the shell i gave you and start hacking ;-) 2014-03-31 13:50:14 clandmeter, yes, that was one of main reasons for requesting the server 2014-03-31 13:50:44 whats holding you back? 2014-03-31 13:50:51 fcolista, can nodejs binding for xapian be added ? 2014-03-31 13:51:11 ok, i'll take a look 2014-03-31 13:51:19 clandmeter, xapian/sphinx test planned today 2014-03-31 13:55:17 vkrishn: you could add aports and alpine-sdk and build it yourself. then if it works you could send them to github for us to fetch. 2014-03-31 13:56:31 xapian - nodejs binding ? 2014-03-31 13:56:33 ok 2014-03-31 13:57:13 vkrishn: i mean in general. 2014-03-31 13:57:21 yes, I did same for mosquitto new version, used the server 2014-03-31 13:57:31 then send patch 2014-03-31 13:57:33 :) 2014-03-31 13:57:40 :) 2014-03-31 14:05:22 for git, I think I would use repo.or.cz, I would do the initial hard work of cloning aport 2014-03-31 14:05:29 it would* 2014-03-31 14:06:29 present clone is over 120mb 2014-03-31 14:25:56 vkrishn: if you use the shell, the clone of aports takes around 20 seconds 2014-03-31 14:26:44 atleast thats what i remember. 2014-03-31 14:29:56 and pushing the cloned it to github ? 2014-03-31 14:30:30 depens a bit on gh, but usually as fast. 2014-03-31 14:31:21 or I can fork github.com/ncopa/aports 2014-03-31 14:31:29 you dont have to worry about the bw, i think the box can manage. 2014-03-31 14:32:19 ok 2014-03-31 14:33:18 uhmm, there is no github.com/ncopa/aports 2014-03-31 14:50:18 here we go vkrishn 2014-03-31 14:50:48 thanks 2014-03-31 14:51:04 i can find nodejs binding in xapian-bindings 2014-03-31 14:51:10 *can't 2014-03-31 14:52:50 github.com/mtibeica/node-xapian 2014-03-31 14:53:00 ah 2014-03-31 14:53:00 ok 2014-03-31 15:04:48 fcolista, looks like there is some patch code in node-xapian, if not ok the binding can be postponed 2014-03-31 15:05:30 i'm gonna look at that l8r if is not urgent for you 2014-03-31 15:06:37 just that if it can be done easily it would get tested today 2014-03-31 15:06:43 ok 2014-03-31 15:09:35 i'd like to create a subpkg for node-xapian 2014-03-31 15:10:09 i mean a subpkg for xapian-bindings 2014-03-31 15:10:23 that includes nodejs 2014-03-31 15:12:11 umh 2014-03-31 15:12:23 the naming is not consistent with the other bindings 2014-03-31 15:16:11 eg. xapian-bindings-lua would install lua ? 2014-03-31 15:17:00 if yes, then a meta pkg xapian-bindings-node would be better 2014-03-31 15:17:22 yes 2014-03-31 15:17:25 it install lua 2014-03-31 15:17:30 that installs node-xapian + nodejs 2014-03-31 15:18:06 lua is the only subpkg that install the interpreter 2014-03-31 15:18:22 and maybe this could be an error 2014-03-31 15:18:36 since in depends_dev there is already lua-dev 2014-03-31 15:19:36 bah 2014-03-31 15:19:45 no versioning for node-xapian 2014-03-31 15:19:50 umpf 2014-03-31 15:19:59 maybe postpone for now 2014-03-31 15:21:58 also the code last update was 10month 2014-03-31 15:22:37 yup 2014-03-31 15:24:14 there's no makefile 2014-03-31 15:25:22 gyp 2014-03-31 15:28:02 we don't have gyp pkg in aports 2014-03-31 15:28:15 right 2014-03-31 15:28:30 i'm looking for an alternative of node-xapian 2014-03-31 15:29:57 I guess it would be better, I would also give a search 2014-03-31 15:31:08 there's a long 3d here: http://lists.xapian.org/pipermail/xapian-discuss/2011-October/008459.html