IRC Logs for #crux-devel Tuesday, 2010-05-04

*** nthwyatt has quit IRC00:32
*** nthwyatt has joined #crux-devel00:37
*** nthwyatt has quit IRC00:47
*** nthwyatt has joined #crux-devel00:49
*** jhw has joined #crux-devel01:08
*** acrux_ has joined #crux-devel02:47
*** jhw has quit IRC07:32
*** jhw has joined #crux-devel11:30
*** jhw has quit IRC12:01
*** jhw has joined #crux-devel12:03
*** pitillo_ has joined #crux-devel12:07
tilmanfrinnst: running xorg server 1.8.1-rc2 (902)12:30
*** acrux_ has quit IRC12:30
tilmanfrinnst: it's using /usr/share/X11/xorg.conf.d -- did we want that?12:30
tilmanmmh, i guess it's okay12:31
teK_(omg)12:31
tilmaninput hotplug works without gudev/udev extras!12:32
tilman\o/12:32
tilmanfont path still messed up though12:32
teK_what do you mean by hotpluggin12:33
teK_i.e. which device?12:33
tilman1.) unplugged mouse12:33
tilman2.) startx12:33
tilman3.) plugged in mouse12:33
tilman-> mouse worked12:33
tilman(no mouse mentioned in xorg.conf)12:33
teK_is this supposed to work with our current udev setup?12:33
tilmandunno. but it does :P12:33
teK_or does this only work by mentioning the mouse for 1.7.x setups?12:34
tilman[  1671.607] (II) config/udev: Adding input device Logitech USB-PS/2 Optical Mouse (/dev/input/mouse0)12:34
teK_212:34
tilmanteK_: with 1.7.x, you need to have hal for this afaik12:34
teK_ah, k12:34
teK_I heard HAL was dead. This *plugging is one complicated beast :)12:35
tilmanffs12:35
tilmanx linked to gcrypt. probably for the sha1 stuff12:35
tilmanteK_: do me a favor? does 1.7 link to gcrypt or ssl?12:35
tilmanresp libcrypto12:35
teK_% ldd =Xorg | grep cryp libcrypto.so.0.9.8 => /usr/lib/libcrypto.so.0.9.8 (0x00007f1f5e118000)12:36
teK_?12:36
tilmanyes, thanks12:36
teK_oh12:36
teK_% ldd =Xorg | grep ssl  libssl.so.0.9.8 => /usr/lib/libssl.so.0.9.8 (0x00007f06bb3a6000)12:36
tilman  --with-sha1=libc|libmd|libgcrypt|libcrypto|libsha1|CommonCrypto12:36
teK_brb12:36
tilmanlibgcrypt was moved up a spot i guess12:36
*** jue has joined #crux-devel13:31
*** sepen has joined #crux-devel13:32
sepenhi13:32
tilmanevening13:33
sepenuea13:33
teK_sepen: wuddup?13:33
sepenteK_, its about postfix13:33
sepenpre-install script requires an update13:34
teK_oh13:34
sepenhttp://crux.nu/gitweb/?p=ports/core.git;a=commit;h=cdfe88c5eabab2b565534e397a4cdb1fcbf27ba913:34
sepenwe switched to exim by default13:34
teK_>_>13:34
sepenso sendmail should be replaced imho13:35
teK_uhm13:35
sepenalso the daemon rc script doesn't work for 'status' argument13:35
teK_package guidelines state that no packages should be rm'd at all13:35
teK_iirc.13:35
sepenyou could use directly postfix status13:35
sepenteK_, well the fact is that now it has13:35
sepenteK_, maybe you could move pre-install to a section in a readme file13:37
teK_this is the preferred solution, I thnik13:37
tilmanyeah13:37
teK_REQUIREMENTS13:38
teK_1. Remove Sendmail.  Postfix and Sendmail can NOT coexist.13:38
teK_well ;)13:38
teK_s/sendmail/exim/13:38
tilmanoui13:38
tilmanfrinnst: i fixed the font issue by using --with-fontrootdir=/usr/lib/X11/fonts; seems okay now13:38
sepenopenbsd and others have a little script like our 'gl-select' to switch between both13:39
juetilman: objection to gcc 4.4.4 update?13:39
tilmanno13:39
tilmandid anyone else test gcc 4.4.4 and find problems?13:40
frinnstyes and no14:02
frinnst:)14:02
j^2:)14:05
j^2hey all14:06
sepencould someone confirm these tickets? #472 #57814:06
sepenhi!14:06
tilmansepen: are you saying #472 can be closed?14:07
sepenyep14:07
j^2still on for the "meeting" in 3ish hours?14:07
tilman3 hours?14:07
tilmanomg14:07
sepen?14:07
juej^2: no14:08
tilmani'll be asleep by then :)14:08
j^2k, i'll send out the note14:08
tilmanmmh14:09
tilmansepen: err.14:09
tilmansepen: i closed the bug. but i just realized that's not what you wanted me to do, right?14:09
tilman:o14:09
sepenehe np14:09
juesepen: 4.6.1 works for me without having clucene installed14:09
sepenyeah I'll close 578 too14:10
sepenbut danny seems opossited (again)14:10
juewell, I think we should start to look at some points in TODO2714:12
jueIMO the most important ones are gcc opt. and acl/attr support14:13
tilmanit's okay to add the graphite stuff14:13
tilmanimo14:13
jueand the LTO stuff?14:15
jaegerno preference on those here14:16
tilmanoops, i didn't check out LTO yet14:16
tilmanwhoa.14:17
juetilman: is it all worth the effort? IIRC the author of the tests are not very convinced14:17
tilman(LTO) "To use this feature, you must be using gold as the linker [...]"14:17
tilmanjue: it's not worth it _yet_14:18
jueGraphite gives small improvements ...it seems not mature yet14:18
tilmanwe could postpone the graphite stuff by a year14:18
tilmanLTO is probably worth checking out though14:19
tilmanjue: i'll merge those pkgutils patches ASAP (this weekend, i guess)14:21
tilmanie your selectable-compression patch and the --ignore-new one14:21
juefine, please use the patch I've linked in TODO27, cause it splits ignore-new and error-codes14:23
tilmanyeah14:23
jueother opinions to the gcc stuff?14:24
tilmanfwiw, debian doesn't seem to have graphite support either: http://packages.debian.org/squeeze/gcc14:25
jueit's optional in gentoo14:29
tilman(arch has it, but meh.)14:29
sepenor just wait to 2.8 when be mature14:31
juewell, if we are uncertain I'd suggest to go the simple way and postpone graphite and lto14:32
tilmansepen: yes, we can "re-evaluate" (sounds a bit too sophisticated :D) the graphite situation then..14:32
frinnsthas anybody here run gcc with graphite?14:32
tilmanme14:33
frinnstissues/benefits?14:33
tilmanfrinnst: it "only" optimizes nested loops. stuff.14:33
juefrinnst: ... mostly degradation on x86_64 ...14:33
tilmanfrankly i don't see it as being ziomg-awesome for c code ;)14:33
tilmanin all of their examples, they used fortran. coincidence?14:34
tilmanright14:34
juefrinnst: you've read the gcc mail?14:34
tilmanhttp://gcc.gnu.org/ml/gcc/2010-04/msg00948.html says something about graphite gaining him 2% on x8614:34
tilmanand -10% on x86_64 or so14:34
frinnstnew and nothing exciting then.. so waiting for a future release sounds sane14:34
frinnstjue, no14:34
tilmanfrinnst: the link above is probably the one jue meant, too14:35
jueyeah14:35
sepenbbl I'm preparing some food14:36
jueso our decicion is to postpone for now?14:41
tilmanyes.14:41
jueok14:41
tilmanbut: i'd like to have a a closer at LTO14:41
tilmani think libelf isn't that bad a dependency14:41
tilmanat least it seems easier to handle than the graphite stuff =)14:42
juelibelf is simple, I've a port already14:43
tilmanditto :D14:43
tilmanoh no, i confused it with elfutils14:44
tilmanah, elfutils includes libelf14:44
tilman:)14:44
juehmm, seems that it would be helpful to create our 2.7 branches soon, at least for core and maybe opt14:49
* frinnst just created a core-x86_64 2.7 branch14:49
juefwiw, the ports I've changed so far are here -> http://jue.ods.org/repos/2.7-wip/14:51
frinnstgentoos bugtracker tracks openssl 1.0 issues here: http://bugs.gentoo.org/show_bug.cgi?id=30427914:53
juetilman: what's the plan with LTO?14:54
tilmangive me another week to have a look14:54
tilmandoes that sound okay?14:54
juesure14:54
juewe are not in hurry :)14:55
tilmanfrinnst: that looks slightly frightening14:55
tilmanbut postponing the openssl update, too would be a bad idea i think14:55
frinnsti think most issues are more or less ironed out14:56
tilmanmmh14:57
tilmanthat bug has been open since february 10th14:57
tilman~10 bugs in 3 months isn't so bad14:57
tilmani wonder how many ports we'll have to fix due to the libpng and libjpeg API breaks though14:58
frinnsti think i patched one or two ports for one or the other when i bumped here14:59
jueok, next is the acl/capability stuff15:00
jueit's more intrusive cause many ports will link against that15:01
*** jhw has quit IRC15:01
tilmani'm not familiar yet with the posix file capabilities thing :O15:02
*** jhw has joined #crux-devel15:03
tilmanjue: having coreutils link to those three libs doesn't seem that worrying though. those aren't very new packages/not experimental anymore i think15:04
jueno15:05
juebut we have to look which other ports link against it if acl etc. is installed15:06
jueby chance I saw that vim is one15:07
tilmanmh, right15:08
juebut at all I'd say its a good thing that some people will like and it doesn't hurt15:10
tilmanshould't hurt at least =)15:11
tilmani agree that we  should give it a try15:11
jue:)15:11
jueok15:11
jue.oO  man-pages?15:13
tilmanthe statistics you gathered make me think we shouldn't move them around yet15:13
frinnst/usr/man ftw!15:14
frinnst:)15:14
tilman"You're call going from 20 hundred hours to 21 hundred hours pushing /back/" wtf?15:14
jueit's ok for me as well because it's much less work ;)15:15
tilmanerm15:21
tilmanwhat's the shortest way to say "x o'clock local time in central europe respecting DST"? :p15:22
tilmann/m15:27
tilmangoing to bed now. let me know what you think about talking about 2.7 things weekly15:28
tilman:}15:28
juegood idea15:29
jueone final note:15:29
jueI've added our decicions to TODO2715:30
jueI'll recompile core with acl/cap the next days on my test-drive15:31
jueif someone needs packages, to have an easier start with the new stuff, let mne know15:32
*** jhw has quit IRC15:53
frinnstzzz16:02
*** jue has quit IRC17:09
*** sepen has quit IRC18:21
*** nthwyatt has quit IRC21:15
*** nthwyatt has joined #crux-devel21:20
*** _mavrick61 has quit IRC21:34
*** _mavrick61 has joined #crux-devel21:35

Generated by irclog2html.py 2.11.0 by Marius Gedminas - find it at mg.pov.lt!