IRC Logs for #crux-devel Wednesday, 2013-02-27

Romsterheadsup frinnst i found a small issue in nss01:00
Romsteryou may want to double check if you add that, that it works on your system too. i have CC and CXX set on mine.01:07
*** pitillo_ has joined #crux-devel01:24
*** pitillo_ has quit IRC01:37
frinnsthm, rsyncing opt broke?
frinnst3.0 is gone from git-to-rsync-working-copy/opt01:46
frinnstthanks Romster01:56
frinnstgood reason to fix opt also :)01:57
Romsteroh that happens when someone pushes at the same time as a ports -u01:57
Romsteror is this something else?01:57
Romsteryou did test my patch on your system i hope <<01:58
Romsterbecasue i didn't test in my chroot just on my system.01:58
Romstertrying to build my firefox-32 again for runescape the thing is a pain to get to compile also this jpeg is a pain too...01:59
Romsteryeah that error looks like what i described.02:01
pitillowin 202:15
frinnstyes if you rsync just as somebody pushes, yes02:22
Romsteryeah i reported that ages ago too02:22
frinnstbut it just dissapeared02:22
frinnstand didnt return02:22
Romsterdid jaeger break that?02:22
frinnstdunno what happened02:23
Romsterdamn firefox i never had this much trouble to get a 32bit version02:23
Romster/usr/ports/work/firefox-32/src/mozilla-release/js/src/jsarray.cpp:3741:1: fatal error: opening dependency file .deps/jsarray.o.pp: No such file or directory02:24
Romsteri'm using -j102:24
frinnstmkdir .deps02:24
frinnst    #mozilla sucks: BZ #82438102:25
frinnst    mkdir $SRC/mozilla-release/js/src/.deps02:25
Romsteri sware it's got a line somewhere to asplode on multilib 32bit mode.02:27
Romster17.01 is compiling so far ok02:27
Romstertried 19 and 18 both died.02:27
*** sepen has joined #crux-devel03:17
sepenjaeger: around here?03:17
teK_Romster: rrdtool still uses perl 5.12 ins its footprint03:39
Romstergrep -rl site_perl/5.12 /usr/ports/contrib/*/.footprint03:45
Romsternot only that one03:45
teK_spamassassin, too! shame on me!03:47
Romsterafter i eat i'll get onto that03:47
teK_still wrestling with the aftermath of my 2.8->3.0 upgrade03:48
teK_libtool: link: cannot find the library `/usr/lib/' or unhandled argument `/usr/lib/'03:48
Romsteri got around that by building everying in a chroot then pkgadded those03:48
teK_and I can't seem to find the package beingbroken and the libstub-breakage is still tehre, too03:49
Romstergrep though the *.la files03:49
teK_I think I had a scrit for that03:49
Romsteri didn't do this this time but last time i changed the version in the la files and rebuilt stuff03:50
Romstersed :)03:50
teK_I think it was stupid libwcnk03:51
teK_now exo builds so I can use xfce-Terminal again.. yay03:53
Romsterit's a pain we have both la and pc files03:53
teK_dunno but this upgrade went anything but smooth :)03:54
Romsteri was already multulib on 2.803:54
teK_btw.. .pc is for pkgconfig, right?03:54
Romsteryeha pkg-config03:55
teK_hi jue04:04
teK_anybody mind (re)building xeces-c for fpmismatch?04:05
jueteK_: I have s script like pkginfo -o, but it accepts several file arguments and you can do something like the following to find all ports you have to rebuild:04:11
juefileowner $(grep -l libpng15 /usr/lib/*.la) | sort | uniq04:11
jueteK_: if you need it I can paste it04:11
juewell, libpng14 in your case ;)04:12
teK_I found a snippet from jaeger in the logs that did just that, jue. thank you!04:21
teK_ah same here, frinnst04:21
teK_I'll file a bug against sepen  :P04:21
jueteK_: np, just in case it's here ->
teK_uh, someone who can script awk04:24
jueteK_: any other problems with the upgrade than the library mess04:25
teK_not as far as I can tell04:25
teK_i.e. only libstub and png1.4 messup04:26
jueok, good to hear04:26
teK_and quite some footprint mismatches but I'm slowly checking for those now04:26
teK_we already spoke about this.. I get a fp mismatch everytime I build glibc due to my restrictive umask04:27
Romsteri really need to learn awk more it's a neat program very powerful but also very cryptic to me.04:27
teK_at least I think so.. or was this about the permission problems with /var/lib/pkg/db..?04:28
teK_Romster: if you're doing sed.. | awk | ... you're almost always doing it wrongly :>04:28
Romsterbut pkgmk has umask 022 in the pkgmk file iirc teK_04:28
jueRomster: yeah, awk is very nice and useful04:28
Romsteri do sed awk alot...04:29
teK_or even better use shell builtins over external binaries :)04:29
Romsterprefer perl or python i do too much bash04:29
Romsterawk is one of them cut sed grep all in ones that i've never learned more than a few simple tasks04:30
juethe basic awk usage is very simple: awk '/pattern/ {action}'04:30
jueif you omit one of them it's using the default04:30
Romsteri came unstuck when i was doing arrays and string changing04:30
teK_the most advanced awk-use I ever did was this:04:31
teK_wget -qO - | sort -n -r | awk -F: '/stable/ { gsub(" ","",$2); print $2; exit }'04:31
Romsterawk can sort too iirc04:31
teK_which is .. grep sed and tail/head in one :)04:31
teK_damn it (wrt sorting) ! :P04:32
Romstercurl -s |04:32
teK_let's not argue over 3 characters :P04:32
Romstererr hehe04:32
juebtw, prtverify is mostly a awk script ;)04:34
teK_I was suprised to learn that ldd is a shell script04:35
Romsteryeah i had looked though that04:36
jaegersepen: I was asleep then, what's up?04:51
sepenwell, I pushed to 3.0 and 2.8 almost simultaneously and that causes git hooks to fail due to concurrence on the same dir (opt/, next push in opt.git solve the problem04:55
jaegerIf you'll recall I didn't alter the hook that updates the rsync repos04:57
jaegerMaybe we need a more robust one but that one hasn't changed04:57
jaegerAnyway, getting ready to go to work, I can take a look at it later04:58
Romsteryeah but wondered if you did anything or not04:58
Romsterit's just what sepen said concurrency04:59
teK_updating glibc during a revdep is a bad idea :P05:02
teK_/usr/bin/revdep: line 27: /usr/bin/grep: No such file or directory05:02
teK_/usr/bin/revdep: line 27: /usr/bin/file: No such file or directory05:02
teK_/usr/bin/revdep: line 21: /usr/bin/awk: No such file or directory05:02
teK_/usr/bin/revdep: line 21: /usr/bin/awk: No such file or directory05:02
teK_/usr/bin/revdep: line 21: /usr/bin/awk: No such file or directory05:02
teK_etc :D05:02
teK_btw jue was what I meant05:06
teK_umask is 07705:06
sepensorry was on phone05:09
sepenjaeger: could be the case that two committers make a push at the same time but in different branches of the same repository, and since our post-update can't know the target branch then it runs the script for all branches (2.8, 3.0) and hence the problem of concurrency with temporary dirs, which should be resolved by the hook in my opinion05:10
sepenthat could be fixed easily with something like s|.new|.new.$$| so the pid will be used as part of the temporary dir to avoid concurrence between different processes05:10
sepenanother way could be to move the export_repo stuff to post-receive hook which takes the same information as the pre-receive hook does on its standard input05:11
sepenI wrote something based on post-update contents, but will require to translate it to perl if we wan't to integrate the export_repo stuff in post-receive hook
Romsteroh that could be the solution05:17
sepenor a solution ;)05:18
teK_once again I hit the red borders problem in chromium .. *sigh*05:22
sepenmust be solved around a year ago?05:28
sepenwe don't have cairo with qt4 support, so what happened?05:30
teK_I really don't know yet05:31
teK_building pixbuf in a chroot first05:32
teK_ok my chroot is b0rked. rebuild chroot FIRST05:33
sepenooops, thought jaeger was here but he's gone to his office :P05:34
teK_anyone mind rebuilding squid for build errors?05:39
jaegersepen: If you have a fix you don't need my permission :) I didn't even write that hook05:55
sepennah, the fact is if could be better to move the export_repo stuff to post-receive05:56
sepenI think we should only run export_repo for the target branch05:56
sepenpost-update hooks are executed from all branches05:57
sepenpost-receive is only executed from the target branch where the committer makes the push05:57
sepenwell I understand that from manuals05:57
jaegerI know all this, did that research when reworking the other 2 hooks :)05:57
jaegerI've got no objections to testing it in the post-receive hook05:58
sepenyeah, you know that, just want to be sure I understand right the concept05:58
sepenok, I'll try to create a patch for the post-receive05:58
jaegermoving it to post-receive and using .new.$$ are probably both good05:58
sepenfor now +105:58
sepenpost-receive has more contents05:59
sepenpatch post-update with the $$ thing would be enough for now05:59
teK_just like the last time.. pixbuf from a chroot build fixes chromium05:59
sepenteK_: that means we should find a solution for pixbuf?05:59
sepento force not detection of qt4 maybe?06:00
teK_I'm still not sure as the bug report says it's cairos qt4-support (or did you mean that?)06:00
sepen"The problem is with gtk not recognizing .png as a png image because x11-libs/gdk-pixbuf segmentation faults because x11-libs/cairo has a bug when compiled with qt4 support."06:01
sepenthats from gentoo guys06:01
jaegereven if the hook is run for each branch using the pid to make the temporary more unique is a good idea, in my opinion06:01
teK_so what happens if cairo has no qt4 support, will this break things?06:01
teK_ puzzles me06:02
sepenI'm not sure how this affects to us06:02
teK_but maybe it's just the same bug caused by [another] seg fault06:02
sepenthats from 21dec2011, should not be solved?06:03
teK_solved but pixbuf segfaults again because of another cairo-qt4 issue06:03
sepenpfff my ubuntustudio still has the glibc bug06:06
sepen*** glibc detected *** grep: malloc(): memory corruption: 0x08609e80 ***06:06
sepenI received too many updates via apt but I'm still waiting this one :(06:07
sepenohhh, my 2.8 box too06:08
jaegerIs there already another flash player plugin md5sum mismatch?06:08
sepenI've installed glibc-2.16.0-3 and I got the glibc bug too06:09
sepenjaeger: thats my fault06:09
jaegerMISSING   a95a8d4049bde90e5dc8e2e394275c42  install_flash_player_11_linux.x86_64.tar.gz06:09
sepenjaeger: I missed to delete the old source file06:09
jaegerNEW       b4640c55eb2e09cfd8a4cd46ebd81fdd  install_flash_player_11_linux.x86_64.tar.gz06:09
jaegerAh, ok06:09
jaegerno worries, just checking06:09
sepenyeah, my fault06:09
sepenjaeger: any idea about the glibc bug?06:09
jaegerSomeone linked a patch for it recently, right? Maybe frinnst? I don't remember for sure06:10
jaegerAre you wanting to patch your ubuntustudio glibc?06:10
sepenI was doing a fake test ;D06:10
teK_btw.. this flash ports source file issue is an architectual problem on our side, right?06:10
sepenthats the other thing, I'm waiting for ubuntu patch06:10
sepenteK_: yeah, due to PKGMK_SOURCE_DIR06:11
jaegerteK_: I would say no, personally, because they never change the version in the tarball name06:11
teK_well jaeger ;)06:11
teK_that's one thing we can't solve06:11
sepenboth of you have reason06:11
jaegerWe could probably add some logic that would account for that but I don't like working around upstream idiocy06:11
teK_including the tar.gz in our ports would be the only possible solution I think. Which stinks :\06:11
jaegerSometimes it's required but I don't have to like it :D06:12
sepennah, maintainer fault ;D06:12
jaegerAnother option would be mirroring the source tarball on crux.nu06:12
jaegerIf we can even do that06:12
sepen*err, fault x 2 (both branches)06:12
teK_you're right jaeger, that'd be the preferablesolution, imho06:13
teK_hosting the tarball on crux.nu06:13
teK_IFF that's legal06:13
jaegerAt least that way we have a bit more control. It's another step for the maintainer, though06:14
sepenso I should repackage the tarball with something like $name-$version.tar.xz?06:14
jaegerno need to repackage it06:14
sepenI could do that, and after a considerable amount of time I can remove old files06:15
jaegerMy point was that if we host a copy of it we know when the md5sum changes06:15
jaegerI don't know if it's legal, though06:15
*** horrorSt1uck has joined #crux-devel06:16
sepenI did not get the point, didn't notice that md5sum changed?06:16
teK_I do that formplayer btw06:17
teK_because their releasemodel is a) screwed and b) to have no real model06:17
teK_hmm I could release mplayer.. it's been some time :>06:17
sepenok I'll wait to you to see an example, for now I'll fix flashplayer06:18
*** horrorStruck has quit IRC06:18
jueteK_: squid builds just fine for me06:25
jaegerunrelated to crux stuff I switched up my workstation setup at work yesterday, so much happier with it now06:26
sepenfor flash we could use the way: flash-plugin-$version-release.x86_64.rpm06:27
jaegerwent from glossy LCDs to matte and from that 2x Xeon 2.26GHz with 16 threads and 12GB RAM to a 1x Xeon 3.6GHz with 8 threads and 32GB RAM06:27
jueteK_: we had the same error already IIRC, wasn't it a problem of the -O3?06:34
teK_you are right :|06:35
jaegerit builds fine for me as well06:35
teK_so many ports and so many borkage :p sysup is down to 5 ports yay..06:35
sepenjue: would you agree if I cherry-pick php and php-sqlite for 2.8?06:49
juewrt the filesystem port: I'd suggest to use 'install -m 0644' instead of plain 'cp' for the /etc files to avoid problems with strange umasks like 077?06:50
juesepen: not if you do it for php-{fcgi,fpm,mysql} and mod_php as well ;)06:51
teK_would be better. I do cp and mv myself too :\06:51
jueeven though only teK_ has problems like this :)06:51
teK_my system is more secure  than yours!!106:52
sepenthat would be nice as a note in package guidelines06:52
teK_I even hand-patched the netlink vuln in 3.8 :P06:53
sepeninstall -m / -o / -g is prefered instead of cp06:53
sepento aboid teK's umasks06:53
sepenjue: so I should merge all php* stuff? ;D06:54
sepenthat would be easy but in this case I'll test after build them06:55
juesepen: I'd say so, IMO it makes no sense if some of the php ports are on a older version06:56
sepenwell the fact is that you last commits to php makes sqlite3 classes usable06:56
juehmm, should work with the old version as well, it's using a bit older sqlite version, the one that comes with php, but that's the only difference06:58
juewasn't the problem you've reported recently not a issue with the deprecated DB class?07:00
sepenyes it was, but I tried with SQlite3 class too07:01
sepennow I'm not sure, I should downgrade php to try it again07:01
sependoes not matter, the fact is that I like how you left the issue07:02
juewell, it's more consistent to use the system sqlite as the one bundled with php, that was the reason for me to change the ports07:04
*** joe9 has quit IRC07:09
juesepen: thanks07:20
sepennah, was your work thanks to you too07:22
jaegerI wonder how hard it would be to get crux onto a zfs mirror root07:50
jaegerI will try it. :)07:50
*** pitillo has quit IRC08:05
*** pitillo has joined #crux-devel08:07
sepenI forked webtools.git for testing sqlite3 changes;a=summary08:16
sepenteK_: ^^08:16
sepennever played before with git remotes/upstream, its funny ;D08:17
*** sepen has quit IRC08:36
*** sepen has joined #crux-devel08:37
*** joe9 has joined #crux-devel08:54
*** acrux has quit IRC09:58
teK_but not from our ISO, right? ;)11:02
jaegerno :P12:13
jaegerI did it from an already-working crux install on another drive12:14
jaegerrpool/PORTS  compressratio         5.20x                  -13:01
jaegerthat's pretty decent13:01
jaegerdaedalus/root /usr/ports # du -sh13:24
jaeger9.6M    .13:24
*** mike_k has quit IRC13:42
teK_nice also14:16
teK_so this is 100% in kernel of fuse-based zfs?14:16
jaegerIt's the zfsonlinux project... so native in-kernel14:17
jaegerwell, modules :)14:17
teK_I remember, it was split in 3-4 separate packages14:18
jaegerIt's 2 now but I haven't used it long14:18
*** sepen has quit IRC17:13
*** sepen has joined #crux-devel17:13
*** sepen has quit IRC18:46
*** mavrick61 has quit IRC19:25
*** mavrick61 has joined #crux-devel19:26
*** mavrick61 has quit IRC22:55
*** mavrick61 has joined #crux-devel22:55
*** mike_k has joined #crux-devel23:06

Generated by 2.11.0 by Marius Gedminas - find it at!