IRC Logs for #crux-devel Tuesday, 2008-11-18

*** sepen has joined #crux-devel01:39
sepennipuL, ping01:56
*** cptn has joined #crux-devel02:09
sepennice to see you again cptn02:20
cptnthanks :-)02:27
*** mike_k has joined #crux-devel02:48
*** rehabdoll has quit IRC03:03
*** rehabdoll has joined #crux-devel03:03
nipuLsepen: pong03:16
*** fetid has joined #crux-devel03:20
sepennipuL, just I'd like to maintain gl-select03:22
sepenI'm the co-author of gl-select, so sometimes I need to modify it, and I don't want to bother you every time to change it03:24
nipuLoh, go for it03:32
sepenok I'll do for the next update, thanks03:33
sepennipuL, are you busy now?03:34
sepenor better, I'll write a mail to you03:35
nipuLnot particularly03:36
nipuLas long as i'm not required to think, i'm on painkillers for a chronic headache03:37
sepenportdb is generating rsync files now?03:38
nipuLit always did03:38
sepenah yes03:38
sepenany plain for edit your own repo properties?03:39
nipuLi'm planning a next gen portdb03:39
sepennipuL, I decided to use python at office instead of php03:40
nipuLpython is my weapon of choice03:40
sepennow I'm playing with .psp files, just rocks03:40
nipuLmostly with django at the moment03:41
sepenI'm also using pygresql, do you know a persistent wrapper for it?03:41
sepennipuL, I read about django but we decided to learn also about 'pylons'03:42
nipuLyeah, i've played with a heap f them03:42
nipuLcherrypy, pylons, web.py03:42
sepenwe are developing network appliances, ATM a traffic shaper box, so which one could be the best?03:43
nipuLit all depends03:43
sepenfor what I know django seems more webcontents oriented03:43
nipuLdjango is very complete03:43
sepenhmm but maybe bigger?03:44
nipuLyeah, cherrypy and are more lightweight also more "pythonic"03:44
sepennipuL, and a thing I can't found, python + server_session like phpsession03:46
sepenis that available on python, or is not need it03:46
nipuL"sessions are bad"03:46
sepenhtaccess instead¿03:46
nipuLdjango has server side session support built in though03:47
sepenhmmm well I need to investigate a lot, many thanks for the info03:47
nipuLif your building network apps rather than web apps, i'd recommend twisted03:48
sepenhmmm interesting03:48
nipuLbut that's also quite heavy03:49
sepenTwisted_Sister ---> heavy metal band03:49
nipuLI WANNA ROCK03:49
sepenyes its too heavy03:49
nipuLhmm, might go play som FoF now03:50
nipuL(frets on fire)03:50
sepenI've on my repo03:50
sepenbut my repo is ATM unavailable, just I'm writing a mail to you with the new .rsync file, etc.03:51
sepennipuL, FoF is also written in python ;-)03:51
*** fetid has quit IRC03:52
*** DarkNekros has joined #crux-devel03:53
sepennipuL, are you maitaining also this page?
nipuLi've asked to be an alias fo contrib-admin04:05
nipuLbut for now i'm nore intersted in trying to fix my toolchain04:08
nipuLsomething broken when i tried to upgrade to glibc 2.8-104:09
nipuLluckily i had a backup of glibc, but noy glibc-compat3204:09
nipuLand unfortunately i thin this is the only crux box with multilib in existence04:10
sepenour new appliance at office (the shaper) works on centos 5.2 x6404:10
sepencentos has multilib04:10
sepennipuL, I could help you with the portdb if you want04:14
sepenalso I think it makes more sense 'portdb-admin' instead of 'contrib-admin'04:15
sepennipuL, did you have an automated tool for building your toolchain?04:16
sepenwould be interesting to share a toolchain script for ppc, x64, arm, etc04:24
sepeniirc acrux worked on a ppc iso, pitillo and me and others from crux-es are working on the arm one too04:25
* mike_k wonders if wii hardware is fun enough to install crux on04:32
nipuLcool, 64bit flash plugin has been released. about f'ing time04:45
teKpre-release quality04:50
*** DarkNekros_ has joined #crux-devel05:13
*** DarkNekros has quit IRC05:15
sepennipuL, did you know it?
*** jtnl has joined #crux-devel05:24
nipuLyeah, i'm used it for a job a few weeks ago05:24
sepenbut seems not uptodate05:25
sepenIm looking for a template engine, or similar05:25
sepenalso I'm evaluating to use 'genshi'05:25
*** jtnl has quit IRC05:41
sepenand this other
nipuLthere's ots of pythong template engines05:51
nipuLpylons uses mako05:51
*** DarkNekros_ has quit IRC05:57
*** jtnl has joined #crux-devel06:50
*** DarkNekros has joined #crux-devel07:13
*** mike_k has quit IRC08:03
sepenjaeger, around?08:31
sepenwell, gnome and xfce are also using gstreamer and gst-plugins-base from contrib08:39
sepenany idea? I think I could maintain both ports in opt if Romster is agree with that08:40
sepenI need gst-plugins-base for the xfce4-mixer port, just a nice to have port on a desktop08:40
sepens/are also using/are using08:41
jaegerI have no objection, I usually only put overlapping things in if they need different library support or something08:44
sepenyep, same for xfce08:46
*** mike_k has joined #crux-devel09:03
*** jtnl has quit IRC09:31
*** jtnl_ has joined #crux-devel09:31
*** Rotwang has joined #crux-devel09:41
*** namenlos has joined #crux-devel10:36
*** pedja has joined #crux-devel10:46
*** sepen has quit IRC10:57
*** jtnl_ has quit IRC11:04
*** jtnl has joined #crux-devel11:05
*** jue has joined #crux-devel11:20
tilmanhi jue11:20
*** jtnl has quit IRC11:26
juetilman: what do you think, the xorg-server bug is quite uggly, shouldn't we provide a patched version on our iso11:28
tilmanjue: the one i emailed the crux ml about?11:28
tilmancan't we just advise people to use the evdev driver?11:29
jueiirc, if you use no config X use the kbd driver and than you have a running X without keyboard ...11:30
cptnanyone around for an IRC meeting tonight? :-)11:30
tilmannamenlos wanted to know whether the git ports driver will be in 2.511:31
jueyes, but with limited time only11:31
tilmanjue: mmmmmmh11:31
namenlosi got to go at 190011:32
namenlosi have training... sry.11:32
jueis the git-driver in any way related to a new CRUX release?11:35
tilmanno, we can easily make a new core/ports release any time11:35
juecptn: btw, welcome back :-)11:36
tilmani don't remember whether we thought the git driver was cool or not11:36
cptnjue: thanks :-)11:36
cptnI'll be afk for a while, but I'll be back around eight11:36
namenlostilman: afaik we agreed to include it (which doesn't neccessarily mean, that crux core/opt/... have to use it)11:37
juetilman: dont remember either, might be a nice extension, but I think rsync will be the default anyway11:37
namenlosit would only help lazy people (like me) to let people download the ports directly from
juenamenlos: I don't get it why it is important to have it on the cd?11:38
namenlosjue: it doesn't have to be on the cd11:39
namenlosafaik the drivers are distributed with the ports they use.11:39
namenlosso imho it should be in the git port11:39
tilmanoh, indeed11:40
tilmannamenlos: send me a patch for opt/git and i'll try to test and then include it11:41
namenlosafaik cptn did some additions to my initial implementation. so he should have ne newest version.11:41
cptnI think we actually agreed to create a merged version11:42
cptnto check whether it is efficient both during checkout and disk usage (not the whole history)11:42
namenlosif he hasn't i will have to look at my pc at home in tyrol, but i only get there at friday 000011:42
cptnI've not been able to verify that11:42
tilmanokay, just send me the script when you're done :D11:43
cptnI think if those two things aren't given, we shouldn't encourage replacing rsync by git11:43
namenlosin general, or for core/opt...11:44
cptnin general11:44
cptnwell, it really is the same11:44
cptnif we use it in core/opt, users will adapt11:45
cptn^ this is what I have in my ~/src11:46
namenlosjust an argument: we dont use httpup in core/opt either ;)11:47
cptnwe do11:47
tilmancptn: s/prestine/pristine/ i think11:47
cptnwe offer most repos as httpup too11:48
cptnnamenlos: besides, that's not the point I was trying to make11:48
cptnnamenlos: it's just that we shouldn't encourage a bad solution11:48
cptnjust to make it easier for lazy maintainers11:48
cptnhttpup may be slow, but it doesn't checkout any history11:49
cptnnor does it store it11:49
jueand it can be used everywhere11:49
cptnand as far as I remember, "git-gc --quiet" doesn't remove history11:49
cptnit just optimized it11:49
*** namenlos has quit IRC11:49
juewhich was the main reason for providing core/opt with httpup as well11:49
*** namenlos has joined #crux-devel11:50
cptntilman: fixed :-)11:50
namenloswhat does the {+,-}set do?11:51
cptnah, that's debugging code only11:52
Rotwangnamenlos: -e means will fall on errors11:53
Rotwang-x echoes whats being executed11:53
Rotwanglike in pkgmk build()11:54
namenlosRotwang: thx for the explanation11:57
namenlosok, i got to go.11:57
*** namenlos has quit IRC11:57
*** pitillo has joined #crux-devel12:08
juetilman: 'X -configure' on my laptop ->
tilmanjue: did you already find a patch for that problem?12:14
juesorry no, hadn't looked at it12:14
*** jue has quit IRC12:27
*** jue_ has joined #crux-devel12:27
*** jue_ is now known as jue12:29
*** sepen has joined #crux-devel12:39
cptnso... :-)12:59
tilmanjue: i hate to patch x13:00
tilmanmaybe we should just stop trying to give users the best possible crux experience :>13:00
cptnlet's release 2.5-test1 then ;-)13:02
tilmandoes anyone else have an opinion on this particular issue?13:02
tilmanpatch xorg-server: yes/no?13:02
cptnwas the patch added upstream?13:03
tilmanno idea whether anyone has a patch13:03
juetilman: it was just an idea, I have no problem if we leave it as it is13:03
tilmani could try just reverting the commit that introduced this behaviour (linked above)13:03
cptnwhat's the effect?13:05
cptncan I read it up in today's log?13:06
tilmanxorg-server 1.5.3 input driver problems13:06
tilman^^^ subject of a mail i sent to crux@13:06
cptnah, that was on the ml13:06
tilmanjue: maybe we can stop shipping xf86-input-{mouse,keyboard} :)13:06
cptnI'd say we can live with this13:06
jueif you forget to set 'AllowEmptyInput' to false and used the kbd driver your keyboard is dead13:06
tilmani guess xorg -configure would pick up evdev then13:07
tilmani'll try that on the laptop13:07
jueI'm using kbd, because not all keys are working with evilwm13:07
tilmanwith evdev?13:07
jueCrtl-Alt-<cursor keys>13:08
tilmanmmh, i think i've seen someone talk about this on irc13:08
tilman#xorg-devel or so13:08
tilmanmaybe this is fixed by building a hotplug-capable xorg-server13:12
tilmanwhich i wanted to talk about anyway13:12
tilmanwe *could* build xorg-server with dbus and hal support13:12
tilmanwhich would also fix some xfce problems iirc13:13
jueother benefits13:14
sepentilman, but not sure, the new xfce 4.6 release don't requires dbus support for X13:14
tilmanyou can hotplug input devices ;)13:14
sepenbut imho its a nice to have feature13:15
juehmm, we have dbus anyway because of firefox so the step to hal isn't that big13:17
jueI hate it to have more and more stuff like that, though13:18
tilmani'd rather have a 200 mb iso, too :D13:19
sepenso remove xorg and opt ;-)13:20
tilmanthen it's 60 mb or so ;)13:20
jueno but firefox13:20
tilmanjue: +113:20
tilmanthere's no _real_ alternative though :(13:20
sepensurely people want to build their own ffox port13:20
juetilman: yes, that's true13:21
juecptn: opinion?13:24
cptngood question13:24
cptnnever lacked hotplugging13:24
cptnbut tilman's should make the call13:24
tilmanthe big distros enable it these days13:24
cptnokay, let's go ahead13:25
tilmanso wed be less likely to hit bugs that they didn't see13:25
cptnyeah, makes sense13:25
tilmanhello, grammar13:25
cptnso, looking at the TODO25 page, I don't think there's a lot keeping us from releasing13:27
jueok, so I'll add a user/group for haldaemon to filesystem and add hal/hal-info to packages.opt13:27
tilmanwould be nice if we could fix the udev bugs13:28
tilmanbut i'm having a real hard time motivating myself to look at that sh*t13:28
juecptn: IMO yes, if that xorg-change isn't that too problematic13:28
cptntilman: 361 should be easy13:29
cptnfor 270, it's easily done, however I'm not sure if we want it13:30
cptnI'd say if the other distros do, we should too13:30
cptnfor once! :-)(13:30
cptnnot as a general rule13:30
tilmanthrice also mentioned LFS released a new rules tarball13:30
cptnmaybe we could also try to integrate the rules from udev upstream13:31
cptnhowever I'm not sure if we should delay 2.5 for that13:32
cptnlooking at the distro specific rules in udev 133, it looks like they don't need a lot of not standard rules13:34
jueto be sure, the hotplug-capable xorg-server was decided, right?13:44
tilmani'll experiment with that13:45
tilmanand if there's no huge problems -> let's ship it :D13:45
jueok, so I'll wait with the other changes until you gives me a go13:46
cptnI just built udev 133 without any specific rules13:59
cptnand it booted fine at least13:59
cptnsound seems to be broken though :-)13:59
cptnno /dev/snd14:00
cptndevice links are directly in /dev14:00
mike_ktilman: btw, udev 133 also fixes the issues I had (with stupid deprecated options in kernel config)14:08
tilmancool :)14:08
cptntilman: I tried the LFS rules, udev-config-20081015.tar.bz214:11
cptnseems fine14:11
tilmani'll try them too, and will push if it worksforme, too14:12
cptnalsa, they use a cdrom group in there14:12
cptnof course, their cdrom.rules doesn't use the cdsymlinks script14:13
tilmanmeaning? ;)14:14
cptnI assume you wouldn't get /dev/dvd-1.0 -> /dev/dvd symlinks14:15
cptnlet me try that14:15
tilmandvd-0.0 -> hda, dvd-0.1 -> hdb here14:15
tilmanwith plain 13314:16
tilmanoh well14:16
cptnyeah, same with 131 + new LFS rules14:18
mike_kI guess, utf-8 integration will not be addressed in 2.5?14:27
tilmanyou can generate an utf8 locale and use that14:27
tilmanreally integrating utf8 into the system is complicated/annoying/etc14:27
mike_kjust generate and pray that your favourite software will work with it =)14:28
tilmanit will14:28
mike_kso, with removing prebuilt locales that ticket can be marked as 'wont fix'?14:29
tilmanmmh, probably14:29
mike_kotherwise flyspray report and TODO25 looks a little confusing together14:31
tilmani closed the bug, thanks14:32
mike_kthat was fast14:32
mike_kI guess, there is no time to try integrating dmcrypt (fs#341), which teK seems to implement...14:34
tilmani'm not even sure whether tek was finished with it14:35
mike_kor he'd let you know if it was that ready14:35
tilmani think he didn't get much feedback on it14:35
mike_ka patch or something like that... and some of us could test it.14:36
mike_kso, can you mark them for 2.6, maybe?14:36
tilmanteK: ping14:36
mike_k[teK] is away (aasdasd)14:37
mike_koh, one more thing (asking before creating a ticket): why not to add a "/usr/sbin/runjobs /etc/cron/hourly" to core/dcron?14:43
mike_kit can be empty for now14:43
juefor what do you need that?14:46
mike_kpoint: some (all?) major distributions ship logrotate. and  the have the hourly entry in crontab or crontab.d, which allows logrotate to be executed. with contrib/logrotate, I also have to ship README for every user to modify his crontab.14:46
cptngotta go14:49
cptngood night folks :-)14:49
juenight cptn14:49
mike_kon a busy server weekly or even daily rotation might be not enough. and sysklogd/rotatelog looks not very safe.14:49
mike_kcptn: bye14:49
juemike_k: have you seen that latest dcron has support for a cron.d directory14:51
mike_kyeah, just've noticed that...14:55
mike_kok, seems like a good place to place an entry if we have no 'hourly' directory.14:56
juedon't get me wrong, I have nothing against adding an hourly dir, but IMO most users would benefit more from a switch to fcron15:00
juenight tilman15:01
mike_kjue: no problem15:02
juebut well, it's not a big issue, I'll add the dir and the crontab entry15:05
jueleaving now, night all15:05
mike_kjue, thanks for that and bye15:06
*** jue has quit IRC15:17
*** pitillo has quit IRC15:44
*** pitillo has joined #crux-devel15:49
*** DarkNekros has quit IRC15:52
*** Rotwang has quit IRC16:32
*** Rotwang has joined #crux-devel16:34
*** Rotwang has quit IRC16:37
*** Rotwang has joined #crux-devel16:39
*** Rotwang has quit IRC16:42
*** Rotwang has joined #crux-devel16:42
nipuLsepen: re portdb maintainance, pdbcacher.php is +w for the opt group16:56
nipuLdoes your fof port build from source or download the binary version?17:08
*** Rotwang has quit IRC17:09
sepenbinary, I tried to compiled it but it had an error in their code17:09
sepenalso required more stuff to be ported too17:09
nipuLi looked at it too, horrible build script17:09
nipuLbut it wont run on 64bit due to 32bit python modules17:10
nipuLmaking a python-compat32 just seems silly17:10
nipuLso i'm forced to play it under wine17:11
sepenfof is one of the ports I have in my todo ;-)17:11
sepenmy box runs slow with the binary and still having graphic acceleration17:12
nipuLhmm, that might explain why it runs slow on my media center17:13
nipuLnvidia fx520017:13
nipuLshould be fast enough to play the game, but doesn't17:14
nipuLmight see how it plays on wine when i get home17:14
sepenshit! I should update my port too17:14
*** mike_k has quit IRC17:20
sepennipuL, the cedega port has a missing .md5 file17:20
nipuLhmm i should drop that one17:26
nipuLi wish people would stop calling me first when their internet stops working17:33
nipuLwe charge to fix it, their ISP will do it for free if possible17:34
sepennipuL, fof related
nipuLwhy am i looking at this?17:45
sepenthat their stuff right?17:46
sepenit has changed a lot from when I ported the first time17:46
*** sepen has quit IRC20:47
*** fetid has joined #crux-devel22:12
*** fetid has quit IRC23:38
*** Rotwang has joined #crux-devel23:39
*** fetid has joined #crux-devel23:39

Generated by 2.11.0 by Marius Gedminas - find it at!