IRC Logs for #crux Saturday, 2012-06-16

joe9libdrm version is 2.4.34 of tilman00:02
joe9in /usr/ports/opt00:02
joe9and xorg-xf86-video-nouveau of jue does not compile00:04
joe9that was a jue's port. let me try with alan's port.00:07
joe9compile error on that, too. missing file; nouveau_device.h00:09
joe9and I cannot find that header file anywhere else with prt-get fsearch.00:09
Romsterjoe9, it was told to you before libdrm 2.4.33 is the last version that works with nouveau00:36
*** vaddi has joined #crux01:39
cruxbot[opt.git/2.7]: gutenprint: update to 5.2.802:07
cruxbot[opt.git/2.7]: [notify] php: update to 5.3.1402:07
cruxbot[opt.git/2.7]: php-fcgi: update to 5.3.1402:07
cruxbot[opt.git/2.7]: php-fpm: update to 5.3.1402:07
cruxbot[opt.git/2.7]: php-mysql: update to 5.3.1402:07
cruxbot[opt.git/2.7]: mod_php: update to 5.3.1402:07
*** mike_k has joined #crux02:49
cruxbot[contrib.git/2.7]: php-bcmath: updated to 5.3.1403:38
cruxbot[contrib.git/2.7]: php-ftp: updated to 5.3.1403:39
cruxbot[contrib.git/2.7]: php-gd: updated to 5.3.1403:40
cruxbot[contrib.git/2.7]: php-gettext: updated to 5.3.1403:40
cruxbot[contrib.git/2.7]: php-mbstring: updated to 5.3.1403:41
cruxbot[contrib.git/2.7]: php-mcrypt: updated to 5.3.1403:42
cruxbot[contrib.git/2.7]: php-pdo-mysql: updated to 5.3.1403:42
cruxbot[contrib.git/2.7]: php-postgresql: updated to 5.3.1403:43
cruxbot[contrib.git/2.7]: php-sockets: updated to 5.3.1403:43
cruxbot[contrib.git/2.7]: php-xsl: updated to 5.3.1403:44
cruxbot[opt.git/2.7]: mod_perl: update to 2.0.703:51
*** c0x` has joined #crux03:55
*** c0x has quit IRC03:56
*** Rotwang has joined #crux04:01
*** c0x has joined #crux04:14
*** c0x` has quit IRC04:15
cruxbot[opt.git/2.7]: firefox: updated to 13.0.104:38
Romsteromg no 14.0 yet?05:01
nogagplz13.0.1 was to fix some spelling mistakes05:12
Romsterwould not surprise me05:21
*** c0x` has joined #crux05:44
*** c0x has quit IRC05:46
joe9Romster: thanks. i thought that was 2.4.3207:05
joe9let me try 2.4.3307:05
joe9romster, even 2.4.32 is missing the nouveau stuff.07:10
*** ente has joined #crux07:14
*** ente has joined #crux07:14
*** joe9 has quit IRC07:16
*** joe9 has joined #crux07:22
joe9romster, thanks. got my system back. I needed --enable-nouveau-experimental-api flag to configure.07:23
joe9and, as you mentioned it worked only with 2.4.3307:23
Romsteroh didn't think about that07:23
Romsterthe whole thing is a mess07:24
nogagplzgot rage ready to tape07:24
nogagplzgonna go shower07:24
jueI'm currently running a newer version of nouveau that works with drm >= 2.4.34, but no chance to get GLX (mesa) running with it07:28
juejoe9: sometimes a port update is more than just a version bump ->
jueno sure what I should do with the nouveau stuff, I'm leaning towards newer nouveau and to forget about OpenGL for now07:33
joe9jue: oh, ok. as long as it works, I could be your early adopter.07:34
joe9jue, please let me know if you want me to test something. would be glad to do so.07:34
Romsterno use to me until it supports 3d07:35
juegamer :)07:35
Romsteronly a little07:35
Romsterwant to more just bought star craft 2 and i haven't unwraped it yet07:36
*** pedja has quit IRC07:53
joe9romster, jue, need some expert advice, please? I have a system that keeps rebooting after starting up /sbin/init08:06
joe9I added setopt xtrace and setopt verbose to all /etc/rc.* scripts08:06
joe9but, that is not showing any additional details while it is booting.08:06
joe9it works fine with the initramfs and i manually startup /sbin/init and the same behaviour happens.08:07
joe9i figure it has to be something that /sbin/init does, I presume.08:07
joe9any thoughts, please?08:07
joe9rmull mentioned that there are init's that you can step through. but, i do not think crux's init allows that.08:08
Romsterrun level 1 for single user root only08:10
Romsterbut i forget how to even switch run levels now -_-08:10
Romsterinit 1 or something.08:10
joe9romster, is it "init 1"08:10
joe9shouldn't "setopt xtrace" work in the rc.* files?08:12
joe9i have both "setopt xtrace" and "setopt verbose" in all the rc.* files.08:12
joe9but, I do not see any messages to the console while init is starting up.08:12
joe9or doing its' work.08:12
joe9i can see that the rc.* scripts use #!/bin/bash08:13
joe9is it some version of minimal bash that does not support these options or something like that?08:13
*** deus_ex has joined #crux08:13
Romsterhaven't done that and i haven't had random reboots on boot up08:20
joe9i figured out that init is running rc.shutdown on startup08:20
joe9is there something that would cause that.08:20
Romsteris it at switch_root or pivot_root that hapens?08:20
joe9i see that /etc/inittab has 2 as the default08:20
Romsterso is roto mounted before switch_root fine?08:21
joe9yes, it mounts fine. I can change the files there and all that.08:21
Romster/sbin/init is the stock crux one?08:21
joe9yes, i think so.08:21
joe9hope so, at the very least.08:21
Romsterin the root files system you switch to08:21
joe9interestingly the /sbin/init is 0 bytes file.08:22
joe9oh, sorry08:22
joe9it is 3122408:22
Romsterboot off the crux cd mount real root, mount /dev /sys /proc chroot to real root see if that works ok.08:22
Romsteryeah same size as the crux one is.08:23
joe9ok, thanks.08:26
*** c0x` has quit IRC08:57
frinnstRomster: wrt firefox: relax. they are already on beta 7 of v.14 :)09:04
Romsteryeah i saw a beta of 14 floating about.09:05
*** jdolan has quit IRC09:07
*** c0x has joined #crux09:09
cruxbot[contrib.git/2.7]: glib-networking: fix dependency09:18
joe9is there a static fsck.ext4? without all the library dependencies.09:57
frinnstcrux doesnt supply one by default09:59
frinnstcant mount / ?10:00
joe9i cannot see how to build it from e2fsprogs either.10:00
joe9wondering if there is a way to do so.10:00
frinnstzero experience with initramfs here10:01
joe9i just want to build a static version of fsck.ext410:02
joe9no library dependencies.10:02
joe9with mdadm,  make CXFLAGS="$CFLAGS" all mdadm.static10:02
joe9builds the mdadm.static10:02
joe9curious if there was some such way to do it for fsck too..10:03
joe9 found it.10:06
Romsterjoe9, add -static to LFDLAGS mightbe enough to build static binaries see how i did nano in romter repo.10:08
joe9romster, thanks.10:08
joe9romster, thanks. that is very helpful.10:09
joe9if I package it, would you be interested in taking this (e2fsprogs w/static) port into your repo?10:09
Romsteri guess so but why haven't you started your own ports collection? or no where to host it?10:10
joe9i have no host. but, I have a github collection of ports
joe9but, important things, I pass onto someone more reliable like you or rmull.10:12
Romsterwell i got gitolite here setup i can add a joe9.git repo for you if you like.10:13
Romsternot very fast but it's reliable.10:13
Romsternogagplz, has his ports on it too. i'll have faster speeds later on though.10:13
joe9maybe, some time in the future. i need to get some thing working now.10:13
Romsterports don't take much time to transfer either.10:13
joe9ok, sounds like a very good idea.10:14
Romsterwell i can do it right now.10:14
joe9ok, then lets do it.10:14
Romsteryou already have a key here10:14
joe9what do you need.10:14
joe9yes, I do with your opt and contrib collections.10:14
nogagplzhe needs you to sign a waiver10:14
Romsteryour already using git locally?10:16
Romsterif you are you can do this "git remote add fire git@fire:joe9.git"10:17
Romstergit push fire10:17
joe9Romster: yes, I use git locally.10:17
Romsterbut before you push i'll set up a hook to make the REPO file10:18
joe9ouch, the push started10:19
joe9 romster10:19
*** ente has quit IRC10:20
joe9romster, I do "git push fire master" everytime I push a patch to github. Is there anything more I need to do?10:20
*** ente has joined #crux10:20
*** ente has joined #crux10:20
joe9and, does this make the ports available in crux ports browser.10:20
teK_ente ente ente ente10:21
joe9Romster: ^^10:21
*** ente has quit IRC10:21
Romsterok should be right now.only need to specify master once then after that just git push fire10:22
Romsterif oyu've pushed already only the next push will show the REPO file.10:22
Romsterand you get a listing here generated when you git push fire10:23
joe9i do not see anything there. does it take some time?10:24
joe9"Everything up-to-date" -- i get this with push10:24
joe9i will probably need to email someone to get the list of these ports in the crux ports browser, correct?10:25
*** ente has joined #crux10:25
*** ente has joined #crux10:25
Romsterno you pushed before i finsihed setting up the post-recive hook script. so the next commit and pugh to fire you do will generate it.10:26
joe9romster, ok, cool. thanks. will push the new e2fsprogs in a little while.10:28
joe9romster, how can I get this listed in the crux ports browser?10:28
Romsteri'll fire a email off and sepen will add it and then it'll show up at next cron job up to 24 hours depending on time of run.10:29
Romsterfor /etc/ports/10:30
joe9Romster: cool, thanks a lot.10:30
Romsteremail sent.10:33
Romsterremember you can still push to github if you like and you'll definitely want to push to fire for portsdb10:33
Romsterand it's all automated so little effort on my part.10:33
enteteK_: indeed10:38
joe9Romster: thanks a lot. I will push to both, initially. and drop the github at some point10:39
teK_how'sit going?10:40
entebeen setting up an ircd10:40
Romsterunrealircd is my personal fave.10:41
enteand rebooting my server, because of that sysret bug10:41
entewe run inspircd10:41
entepersonally, I like irc2, because it's oldschool10:42
enteyou might have noticed that I like deprecated things from the 70s10:42
Romsterwell you might be interested in romster/bopm for your ircd.10:43
entehow's your bopm different from others?10:44
entemy networks are both pretty small10:44
entealso, do you run your own network, or are you part of a bigger one?10:44
joe9romster, quick question, have you tried gentoo?10:45
Romsterit's not it's just packaged for crux.10:45
Romsterjoe9, nope10:45
Romsterbut i do troll there ebuilds.10:46
joe9i look at gentoo/exherbo lovingly everytime I cannot find a port in crux.10:46
enteyou could use pkgsrc for things that aren't packaged, if you're absolutely not willing to do that10:48
ente(package them)10:48
joe9ente: oh, really. will check out how it works.10:51
Romsteronce you do a few hundred it's not that different to do another except a few have some way different build methods.10:57
nogagplzwhat trolling Romster11:00
teK_ente: fine. I used the quakenet-stack back in 200311:01
Romsterebuilds when i can't find how something builds or for patches11:01
nogagplzdo you mean trawling11:02
Romsteri guess11:06
joe9romster, need your quick opinion. I get this error when trying to build a static e2fsck11:10
joe9wondering if you have come across such error before.11:11
joe9google'ing on it is not coming up with something pertinent.11:11
enteI kind of like how they did it in pkgsrc11:14
entesince most packages use autoconf anyway11:14
Romsterthat means that glibc hasn't got static *.a files or something else hasn't11:15
entebut crux ports are fine too. the duplication is a disadvantage one gets for not adding another abstraction layer11:15
Romsternot sure exactly what but to link staticly everything needs the .a files to link into the final static binary11:16
entegetting into pkgsrc packaging is much harder than crux porting11:16
enteI expect it to get easier after you've done a few 100, though ;)11:16
enteon crux, you don't need to know more than bash and maybe a how building software works11:16
enteon pkgsrc you need to know the entire pkgsrc framework, which is a load of makefiles including more makefiles11:17
entethe advantage of pkgsrc is, though, that it's portable11:17
entea subset even works on windows (interix, but cygwin or mingw are wanted)11:17
entebut I think I'm repeating myself (and also the pkgsrc guide)11:18
enteI just obviously like it :)11:23
Romsterjoe9, you should be able to get everything working with dietlibc i think glibc is hard to link fully staticly11:23
entejoe9: that's a glibc thing11:23
Romsterand you rarely want the bloat of glibc for initramfs11:23
enteglibc doesn't like static linking because (official reason) /etc/nsswitch.conf support11:23
joe9Romster: ente never tried dietlibc, will just copy over all the ldd libs.11:23
ente(inofficial reason) the glibc maintainers don't like static linking11:23
Romsterit's really stupid that it can't11:24
entedieltmusl should work too11:24
Romsteror uclibc11:24
joe9is it easy to get dietlibc or uclibc installed and get going on crux?11:24
entedietlibc has few features, but it should be enough to bring the system up11:24
joe9or, is it a pita..11:24
entewouldn't recommend it as everyday libc though11:24
entedietlibc and musl are actually pretty easy11:24
joe9no, just for an initrd fsck.11:24
enteyou compile them, and they provide a xxx-gcc wrapper around gcc, which then links against the correct libc11:25
entexxx as in {musl,dietlibc}11:25
entebut personally, I don't see the point why a libc shouldn't support static linking just because nsswitch11:26
entethere are better solutions for that11:26
tilmanente: now that drepper is gone, maybe things will improve in that area as well =)11:27
enteyeah, maybe :)11:27
enteI heard he's working for goldman-sachs these days11:27
teK_according to linkedin, yes11:28
teK_he's listed on their website, too11:28
ente I blame drepper :P11:28
entetilman: at least, there will be less WONTFIX bugs11:29
entefor "unsupported" "carp architectures" like arm11:29
joe9 code in /etc/rc was causing the reboot.11:30
jueente: if you follow the mailing list a bit, you will see that the glibc development is a complete different story now11:30
joe9i do not understand why sulogin -p would not give a shell and automatically drop to the reboot.11:31
joe9any thoughts ,please?11:31
entejue: good!11:31
joe9this line: /sbin/sulogin -p11:31
jueyeah, indeed :)11:31
joe9is not giving me a maintenance shell.11:31
Romsterconfigure glibc with --enable-static-nss.11:34
joe9Romster: that's it? that is a simple solution.11:35
entewell, it does work...11:37
entemusl does some dynamic linking optimisations too11:38
entelike putting the dynamic loader in the libc, which saves startup time for things which are only linked against libc/libm (which is also in libc)11:38
entei.e. /lib/ is a symlink to /lib/libcmuslwhateveritscalled.so11:39
Romsterit's stupid how glibc deals with NSS11:39
enteyeah, intedeed11:39
teK_glibc worries about bloat. sounds non-credible11:39
entebecause you wouldn't have to sacrifice static linking for it11:40
Romsternot like you link everything static just a few things.11:40
enteto reduce bload, you could e.g. speak sasl to something else iff the configuration is not in local files11:41
juethe idea is to have a plugin system, which allows ldap an similar crap to add libs to nss11:41
teK_the oracle binary is 100mb of size :>11:41
entethat would reduce bloat from libc and use code that's alreadyu present11:41
entegrr, this latency11:42
enteI get a 5s delyay11:42
enteyay... exactly11:42
RomsterNote: your uClibc must have been built with UCLIBC_HAS_FTW=y (and probably UCLIBC_HAS_NFTW=y too), otherwise e2fsprogs will complain on "ftw.h".11:46
Romsternot tested it but it should work better that glibc for static linking.11:46
Romsterelse just copy what ldd says it's using to the initramfs.11:47
joe9 Romster: yes, that is what I am doing.11:47
joe9just copying the libs to the initramfs11:47
Romsterreally really stupid static means static not link everything static but still requiring shared glibc libs11:48
jueRomster: that's not possible because of the reason I gave above11:50
Romsteryeah i know.11:50
juebut if you don't need any extension it just works, but at link time you will get the warning11:52
juesee the build of pkgadd for example11:52
jue.oO at least with our glibc and the revert patch ;)11:55
Romsterdon't know how the modules work or interact.11:55
juehehe, me neither11:56
*** ente has quit IRC12:21
teK_I only had to recompile python to  have non-standard unicode=uc4 instead of  uc2 to make this crapload of vbox-python bindings work. Curious what's broken now..12:22
Romsterdon't ask it's best not to.12:25
*** ente has joined #crux12:50
*** ente has joined #crux12:50
*** Rotwang has quit IRC12:50
joe9this piece of code in /etc/rc is being problematic:
joe9causing my system to reboot all the time automatically.13:29
joe9it does not give the console as "sulogin -p" is supposed to.13:29
joe9Romster:  would you have any thoughts on what could be going wrong.13:29
joe9I checked the filesystems and everything is fine.13:30
joe9if I comment the code in the if block, everything works fine.13:30
joe9i checked the error code of fsck:
joe9and it is 0, as it should be.13:30
joe9any other suggestions on what I could be missing, please?13:30
joe9jue, any thoughts, please?13:32
Romstersorry no idea, you better commit and push something so sepen can see everything is there when he adds it to portsdb, it's 4:34am i'm heading to bed13:35
joe9ok, thanks.13:39
joe9Romster: done.13:42
joe9Romster: looks like I need to change the structure of my ports directory13:43
*** joacim has quit IRC13:44
*** mike_k has quit IRC14:35
*** clb has joined #crux15:13
*** hawksbill has quit IRC15:38
*** hawksbill has joined #crux15:43
*** jdolan has joined #crux15:59
*** ChanServ sets mode: +o jdolan15:59
frinnsthow can a mail containing the words "viagra" anc "cialis" get by my spamfilter?!16:38
joe9i am going through my /etc/rc script and am trying to figure out who mounts tmpfs?16:47
joe9my login shell is zsh, and in the rc script, the "sulogin -p" when started16:47
joe9needs to create a file in /tmp/ -- which has not been mounted with tmpfs16:48
joe9and as the root is ro fs, gives an error.16:48
*** joacim has joined #crux16:58
*** vaddi has quit IRC17:49
*** jdolan has quit IRC18:15
*** jdolan has joined #crux18:15
*** ChanServ sets mode: +o jdolan18:15
*** joacim has quit IRC18:34
*** joacim has joined #crux18:35
*** tilman has quit IRC18:36
*** tilman has joined #crux18:37
*** ChanServ sets mode: +o tilman18:37
teK_joe9: mount -o remount,rw /18:39
*** Rotwang has joined #crux18:41
*** Rotwang has quit IRC19:36
joe9teK_: thanks, that was it. but, i just mounted tmpfs in initramfs script19:59
*** mavrick61 has quit IRC21:42
*** mavrick61 has joined #crux21:43
Romsterjoe9, fixed now on next push22:34
Romsternot sure why but the postrevice file i cop[ied had ports in a ports/ subdirectory now it'll work with your layout.22:35

Generated by 2.11.0 by Marius Gedminas - find it at!