IRC Logs for #crux-devel Tuesday, 2012-01-03

*** sepen has joined #crux-devel01:29
*** prologic has quit IRC03:34
*** prologic has joined #crux-devel03:37
*** prologic is now known as Guest1479403:37
*** Guest14794 is now known as prologic04:35
*** prologic has joined #crux-devel04:35
jaegerteK_: amusingly after we both did all that UEFI stuff there's an article in Linux Journal this month about booting linux on EFI machines09:25
jaegerit's only part 1 of multiple but I still found it amusing09:25
jaegeryou used elilo, right?09:26
teK_i still do09:26
jaegerdoes it require the kernel to be in the EFI ESP or can it be loaded from anywhere?09:27
jaegerThe article kinda hinted it had to be in the ESP but didn't explicitly state that, sounded fishy to me09:27
jaegerI know the .efi executable needs to be there but doubt the kernel does09:28
teK_i put it on /boot09:29
teK_the kernel, too09:29
jaegerinteresting, so I'm wrong about the efi loader needing to be in the ESP09:29
jaegerunless you mounted your ESP at /boot? :)09:30
teK_wait ;)09:30
jaegerI used grub2 so just trying to compare infos09:30
sepenteK_: I was wondering about to move xfce/libmpd to opt, so you could remove contrib/libmpd09:31
teK_so it's what you call esp, I suppose09:31
teK_sepen: yeah, sure09:32
sepenteK_: are you agree?09:32
sepenok, thanks, and don't worry so I'll remove it from contrib too09:32
teK_just go ahead09:32
sepenok, also your port was a bit outdated09:32
jaegerteK_: ah, your ESP is mounted in /boot/EFI, I mounted mine at /mnt/EFI for no particular reason09:33
jaegerMy UEFI doesn't automatically pick up new .efi loaders, I had to temporarily replace the default one with grub, then when I was successfully booted via that, put the default one back and use efibootmgr to add the new one09:37
teK_there's a tool to do that09:38
teK_which, btw, is the reason why I did not write my stuff up09:39
teK_it's called efibootmgr09:39
teK_I packaged it, just like elilo, too09:39
jaegeryeah, I just mentioned it09:39
teK_I really was confused about that first place09:39
jaegerthe problem was that I had to do the install booted from BIOS, then figure out how to get booted via EFI to even use efibootmgr09:39
teK_all in all: it's fun09:40
jaegerIt was definitely an interesting experience09:40
teK_it sucked, lets be honest :p09:41
jaegerI will probably do it again from scratch sometime in the future to get some documentation down09:41
jaegeron a side note, gnome3 is complicated... so many little things that all have to be configured together09:44
jaegerother weird things... starting a gnome session fails the first time after a reboot, every time09:55
jaegersubsequence sessions work, though09:55
jaegersubsequent, event09:55
jaegerdamn it, I can't type today09:55
teK_strace! :p09:56
jaegerIt may come to that, heh, just trying to solve one thing at a time09:57
jaeger <-- it looks alright, at least, but needs a lot of integration work still09:57
jaegerI also need to do some testing to see if the PAM requirement can be removed09:58
jaegerI *think* the only thing that absolutely cannot work without PAM is gdm but I'm not 100% sure09:59
teK_I tried to build gnome keyring manager ONCE09:59
teK_after pulling in half of gnome.. NO09:59
jaegerIt's definitely a lot of junk, no doubt10:00
jaegerI still like it better than any other DE10:00
jaegerwell, I did at the time of 2.3x, I haven't made up my mind about 3.x yet10:01
jaegerEven if I end up not getting it working properly I learn a lot in the process about various things10:01
Romsterjaeger, what was it you had to strip out of CFLAGS etc to get gcc to compile correctly?10:01
jaegerPAM, policykit, gdk-pixbuf/gtk/pango helpers, all that jazz10:02
jaegerRomster: hrmm, you may have to refresh my memory, I don't think gcc had problems10:02
Romsteri remember you saying when you rebuild core for the iso gcc failed and you had to do something to get it to compile.10:03
Romsteri'll know soon enough recompiling gcc with libelf support.10:03
jaegerhrmm, wish I could remember10:03
jaegerWhat does libelf support add to gcc?"10:04
jaegerman, I have serious fatfinger today10:04
Romsterflags todo with the multilib 32bit part of gcc failing or something my memory is hazy on what exactly.10:04
RomsterLTO support.10:04
Romsterlink time optimizer10:05
jaegeryeah, just googled it and read a bit10:06
Romstermuch like how kde could merge all the source files before compile and link. but this way avoids all the possible conflicts of that.10:09
jaegerhrmm, udev 176 looks like it could be annoying10:11
Romsteri think soon it requires the stupid /run directory.10:11
jaegerdevtmpfs and /run required10:11
jaegerI think we'll probably want to add a README to udev for the kernel config requirements10:12
Romsterdoes this devtmpfs really bring much improvements over the old way? kernel is able ot populate some nodes in devtmpfs?10:13
sepenis /run hardcoded?10:13
jaegerI don't know if it does but 176 will require it10:13
jaegerRomster: as a side note, I installed the binary VirtualBox from the website today and copied my VMs over from my Ubuntu partition, it's working well here with bridged networking without any difficulties10:15
Romsterdo you have to do anything special in crux to use that mode in vbox?10:16
Romsteri got static ip set to my host.10:17
Romstermodprobe e1000 off the crux iso10:17
Romstertried to bring up eth0 as a static ip failed to find eth010:18
jaegerYou definitely used to have to do stuff with tun/tap but it seems to be automatic now10:20
jaeger# CONFIG_TUN is not set10:20
Romsteri did read ages ago about tun/tap10:20
Romsterso how does it work in /etc/rc.d/net or should it just work transparently without any configuration in the guest?10:21
jaegertransparent in the guest10:21
jaegerif you set it to bridged networking then it acts just like it's on the same network as the host machine10:22
sepenfrom contrib/virtualbox -> usr/share/virtualbox/bin/VBoxTunctl10:22
Romsterhmm so it should of jsut wrked oh did try ping inNAT but that's VB NAT ot the same NAT i got on my gateway pc and that VB NAT needed DHCP configured. i'll try again with bridged, i did try that before with no luck.10:22
jaegerAll of my comments are related to the binary install from, for what that's worth, sepen10:23
Romsterthat's what i thought bridged meant.10:23
jaegerIf you set your guest to NAT then you do need to use DHCP, I believe10:23
sepenjaeger: ok, sorry for making noise10:23
jaegersepen: no worries, just didn't want to incite confusion :)10:23
Romsteri'm currently on the contrib-x86-64 virtualbox-bin10:23
jaegerWhat version of vbox is that?10:24
jaegerI guess I could check, heh10:24
jaeger4.1.2, ok10:24
Romstersepen, the contrib one failed for 64bit probably some simple path in the build()10:24
jaegerthat's fairly old, it's possible the networking has changed since then10:24
sependid you tried new 4.1.8?10:24
jaegerthough I would guess not10:24
sepenit failed to compile here (x86)10:24
Romsteroh i did notice that the virtualbox-bin has a newer version out too when i fired it up.10:24
Romsterthe 64bit one of course but i didn't think much of it not working with network.10:25
Romsterok gcc compiled fine jaeger so never mind about that brokenness it'snot broken at all.10:26
jaegerok, good... one less thing I have to dig up out of the recesses of my mind10:27
Romsteri jsut had a faint memory you had issues with that before.10:27
jaegerYou could be right, I just don't remember it10:27
Romsterbut that was when you had other flags set differently in pkgmk.conf10:27
jaegerLots of things changed over the course of figuring out the toolchain stuff10:27
Romsteryeah you've done a ton of work.10:28
Romsteri'm quite happy with the progress. but i'm still sorting out the remains of my hvl stuff.10:28
Romsterchanges i had.10:29
Romsterlike lto support being one.10:29
jaegerI'm quite happy with it as well, it's a fun project10:29
jaegeras well as relevant to my usage needs at work and at home, so :)10:29
Romsteri was playing red alert 2 yesterday in wine worked like a charm.10:30
jaegerI managed to get terraria working in wine yesterday :)10:31
jaegerNeed to bump wine today, .36 came out10:31
Romsteryeah i noticed i haven't bumped mine yet either.10:31
jaegerworking on work stuff at the moment, though10:31
jaegerI finished my vsphere 4.x to 5.x upgrades yesterday, now have to update virtual machine guest tools, etc.10:32
Romsterit's every friday morning around 3am they relase a new version every 2 weeks usually10:32
jaegerAh, good to know10:32
Romsterfor me it's around that time +11GMT10:32
Romstersometimes it's 3weeks or so depends on what's going on.10:32
Romsteri'm in #winehackers -_-10:32
jaegerI just glance at the page now and then, don't really follow it with any tools or RSS readers10:32
jaegerat some point I will need to set up ck4up again or similar, haven't gotten to it yet10:33
Romstersame jsut check it every Saturday morning should be good to go.10:33
Romsteri've still got to resume work on my version sort, it's been postboned recently due to changing to multilib. and the fact i know what i need to do to get around some issues that needs some major work.10:34
Romsterck4up is nice though.10:35
Romsterjust got my own idea to try out.10:35
Romsteri'm nearly thinking of a Makefile way to write out 32 and 64bit Pkgfiles for the -32 and normal stuff, for my own personal use. than to try and track changes. but not sure if i want to do that yet.10:43
sepenwell, if helps this is my automated-ck4up's stuff
sepenits useful to have a mail every time a new update is available10:44
Romsterwas thinking a step ahead, isn't thee some generic ports language and modules to write out Pkgfiles ebuilds etc already in existence? i haven't spotted any.10:44
Romsteroh mail i didn't know it could do that.10:44
sepenRomster: cron does it10:45
Romsterah you made it do it.10:45
Romsteri'm thinking too complex again i know.10:47
sepenyep and also I've a great script to find new -unmaintained- ports, called 'repock4up'10:47
Romsteri'm liking your pkg-not10:47
Romsterthere is probably 1 or 2 in opt i can pick up but they arn't broken currently so i haven't touched them yet.10:48
sepenRomster: sepen/crux-repo-utils but until portdb shows my new domain name (I moved to
sepenalso this tarball contains 'repoverify', the tool I used to generate repo-logs (
*** sepen has quit IRC10:53
Romsteroh i totally forgot about that10:55
Romsterwonder how many of mine have warnings.10:55
jaegerlooks like a few of mine show up in the list for redundant deps, heh10:56
Romsterneeds to be regenerated hasn't got me in opt.10:57
Romsterfew junk files and a libexec path10:57
Romsterminor issues i'll fix them later on.10:57
Romsterquite suprising considering the number of ports i'm maintaining.10:58
Romsterthe only big issue with x86-64 and multilib is you can't cd into a directory and pkgmk -r -u and expect it to work after.11:05
jaegerwhy not?11:06
jaegeroh, are you thinking of something like cd into /usr/ports/core and then doing that?11:07
jaegeryeah, the overlays would cause issues there11:07
Romsterbecause some in core like glibc in core-multilib is not the same as in core11:07
Romsterthinking of a tool to merge the REPOs for opt core contrib into there own single REPO and to remove any in a lower REPO that exists in the higher REPO line.11:08
Romsterthat way the prefer higher option is also safe to use in prt-get.conf11:08
Romsternot sure if it's a good idea doing that though or warrants it.11:08
jaegerNot worth putting a lot of effort into that until official 64-bit support is decided11:09
jaegerI had an interesting idea on that, though11:10
Romsteractually order can stay as is in prt-get.conf just need a tool to zap (remove) ports in a lower repo that exist in the one above it. after each ports -u.11:10
jaegerto build it into ports or ports drivers11:10
jaegermaybe a directive in the ports file that says overlay=core or something11:10
jaegerJust idle speculation, I've done nothing to make it happen11:11
Romsterso if core-multilib/glib exists it's removed out of core-x86-64 and core11:11
jaegermakes sense, though I was thinking of taking it even further11:11
Romsterbuilding it into ports would be a nice idea actually.11:11
jaegercore-multilib wouldn't exist in /usr/ports11:11
jaegerit would overwrite core/* with core-multilib/* where required11:11
Romsterdn't forget we have 3 layers there is x86_64 in the middle of this.11:12
Romsterbut that idea of merging into one repo on the host with the aid of ports command is a interesting idea.11:13
jaegertrue, though that wouldn't be the case if we smoothed out all the 64-bit stuff as jue was talking about doing, perhaps11:13
Romsteri know it's not official but it would even help us out.11:13
Romsterwhere i can i'm fixing up ports in contrib and opt to be transparent between 32 and 64bit.11:14
Romsterlike libfame i recently fixed.11:14
jaegereven if the 3 layers didn't go away we could name /etc/ports/*.rsync with numbers like rc scripts or something11:14
Romsternow works on 64 and 32bit11:14
jaeger01_core.rsync 02_core-x86_64.rsync 03_core-x86_64-multilib.rsync11:14
jaegersomething like that11:14
Romstereh then ports -u name wouldn't work.11:15
jaegerports would automatically give higher numbers higher priority11:15
Romsterhmm or that.11:15
jaegerso then running 'ports -u core' would update all 3 of those11:15
jaegerbut make it transparent(ish) to the user11:15
jaegerand all would end up in /usr/ports/core11:15
Romsteractually that wouldn't be a bad way to get the order right.11:15
Romsterso oyu were thinking about something like that before, i even mentioned it?11:16
jaegeryes, though I've not tried to implement any of it11:17
jaegerwas waiting to see what happens on the "official 64-bit CRUX support" front11:17
Romsterit's all thoughts currently, though i somehow have a feeling that it's not simple enough for CRUX to be accepted.11:17
jaegerperhaps not11:17
Romsteri've started testing and getting more 64bit support going. but i don't think 64bit is ready fro prime time yet unless more start to use 64bit that maintain ports n the official/semi-official repos.11:19
Romsterit may come down to it'll be only us on multilib that's not official and needs some ports patch11:20
jaegerI think it's a bit of chicken and egg, partially11:20
Romsterdepending on the reactions11:20
jaegerI think more people would use 64-bit if more people were already making 64-bit ports, etc.11:20
Romsterprobably right there.11:21
Romsteri'll keep weeding out 64biut issues as i go though them.11:21
frinnstunfortunately i use quite few ports, so its difficult to test and maintain lots o' random ports11:22
Romsterthere shoulnd't be that many though11:22
Romsterhence why i've got access in x86-64 opt/contrib but i haven't committed anything there as of yet, i know i got netpbm to sort out11:23
Romsteri still wonder why we don't have more applying for contrib11:24
Romstermaintainers have shrunk the past couple of years11:24
Romsterbeen wondering where would crux be if i wasn't helping out.11:25
jaegerI wonder if our lack of official 64-bit support has caused the user base to dwindle11:25
Romsterwhos mostly been working on 64bit frinnst ?11:25
frinnstwell, if i were shopping around for a new distro today and i came across a i686 one, i doubt i'd try it out :)11:26
Romsterso the census is to get 64bit mainstream official but what happens to i68611:27
jaegerAs far as I know only frinnst11:27
jaegerthere may be others running it but I don't think others are working on it, though I could be wrong11:27
Romsterwell we can add to 64bit11:27
jaegeror I could go look at the git activity, I guess11:27
Romsterhmm i should make gitstatus of them git trees too.11:28
frinnstpedja, tek and rotwang helps with a few ports too11:28
jaegerI'll admit I never asked for x86_64 access because I wanted multilib and therefore didn't run pure64 anywhere except one machine11:28
jaegerAh, good to know11:28
Romsteri liked 64bit but needed to be able to run some 32buit and when jaeger made the multilib i quickly jumped on board.11:29
Romsterand nogagplz as well so 3 confirmed on multilib11:30
jaegerI believe that officially supporting 64-bit would make CRUX more popular, regardless of whether or not it's multilib, though I could be wrong11:30
Romster still not all done.11:30
frinnstbut maintaining x86_64 isnt that much work11:30
frinnstthough it feels like a waiting-game at the moment11:31
Romsterkeeping i686 offical along with x86_64 will need someone that's still running i686 to properly test.11:31
frinnsti wonder how many still runs crux on non-64bit-capable hardware11:32
Romsterit's nearly at the point where like jue stopped building the i586 iso than the i686 will be not worth it in a few more years time.11:32
Romsteri still have a few 32bit cpu's here but i've retired them all for linux use.11:32
jaegerI think it's worth keeping 32-bit around, I'd just rather see 64-bit be the "default"11:33
Romsterwho actually make up crux thesedays tek jue tilman?11:33
frinnstyeah and jaeger.. but hes "retired" :D11:34
jaegerI offered to step back in, though, now that I have some more time11:34
jaegerIf they want me to help more they'll ask :)11:34
Romsterso we could go the other way around and put most ports in the official 64bit repos and then i686 can overlay the top of that with the 32bit ports.11:34
Romsterfunny how jaeger is retired but your still pretty active11:35
frinnsthehe yeah.. always building the iso :)11:35
teK_Romster: you forgot yourself, sepen, ...11:36
jaegerI do build the ISOs mostly but I don't do much port work or toolchain work11:36
teK_ente and all I forgot :>11:36
teK_it would be awesome to have some jenkins or so for various automated builds11:36
teK_once my root server gets upgrade to that multicore monster, I will provide a VM for crux (backup + this stuff)11:37
frinnstoh why did i upgrade my eeepc to wheezy? gnome3 D:11:37
jaegerfrinnst: try the fallback session, maybe?11:37
jaegerit'll look more like gnome211:37
Romsteroh right am i even a dev now lol11:40
jaegerteK_: my buildbot setup is easy to break (bad source URL, etc.) but it's a good start for automated ISO creation11:40
Romstersepen yes11:40
enteteK_: hm?11:41
Romster4;42am i should probably nap before the sun comes up.11:42
Romsternapping bbl11:50
frinnstjaeger: why are you bothering with gnome3? :D11:53
frinnstgod its aweful11:53
teK_ente: what's unclear?11:55
jaegerfrinnst: haven't used it enough yet to decide if it's awful :)11:55
teK_hi ju12:13
enteteK_: you hilighted me :P12:14
* teK_ throws some context at ente by point to the backlog12:14
entemh :/12:15
enteI don't have a CRUX machine currently12:15
entealso not enough time to update things12:15
enteplus, I came to the conclusion that duplicating the packaging efforts for all distributions doesn't really help12:15
enteif I were on crux, I'd probably install all of core and use pkgsrc for the rest12:16
juejaeger: thanks for your offer, but what do you mean with "step back in"?12:25
juejaeger: another question, who is "they"?12:26
juefrinnst: are you using glibc 2.15?12:28
jaegerjue: I meant that I could devote some more time to helping with crux stuff you guys needed it13:13
jaegerjue: "they" is whoever's in charge these days :) you, tilman, tek? whoever else13:13
jaegers/stuff /stuff if /13:24
teK_what's your username?13:30
tilmanteK_: pedja probably13:30
teK_whois? ;)13:30
tilmanteK_: what? deusex is pedja13:31
teK_deus_ex: please retry13:35
*** deus_ex is now known as pedja13:35
teK_tilman: I alluded to the fact that I could have found out via /whois who deusx is ;)13:35
pedjateK_: still the same13:36
teK_ok. one moment please :-)13:36
pedjaI am not going anywhere :)13:37
teK_try again13:37
pedjateK_: works now, thanks.13:38
teK_sudo chmod -R 777 /home fixes almost every permission problem13:38
pedjatilman: i have a git patch, if you want it :)13:41
rmullteK_: Ouch >.<13:42
tilmanpedja: email please13:42
pedjatilman: sent.13:51
pedjaeither git format-patch does not honor -N switch (no numbered), or I am not using it properly...13:55
pedjaoh, well.13:55
frinnstjue: no.. still waiting for a tarball14:49
frinnstjust glanced the libc-alpha ml.. still lots of problems15:01
juejaeger: thanks, good to hear :)15:20
juefrinnst: read that too, looks like glibc isn't usable without a bunch of patches nowadays15:21
juemaybe we should try the fedora branch? Looks like Andreas Jaeger is doing a good job improving Drepper's marvelous work15:30
teK_should he not be ruining some finance corp? :>15:31
jueTBH, I have no idea what other strange fedora ideas he is adding as well15:31
teK_Lehman that was.15:32
jueoff to sleep, night all15:39
Romsteris this my fault or something wrong with permissions on files on the server?
Romsterdeus_ex> I still can't push to contrib64 git.
Romsterjsut noticed deus_ex has the same issue.19:21
Romsteronly for contrib-x86_64 though19:22
RomsterteK_, ^19:22
Romsteris the umask set wrong on some script?19:32
jaegersomeone set it to the "NOTyoumask"19:33
Romsterthat's pretty obvious hehe19:34
Romsterobviously doesn't want me to commit :P19:40
Romsterwell i can commit but not push until that's fixed, now that i know that's on the remote end i can safely commit changes locally.19:41
RomsterLTO is coming along nicely rebuildng everything fixing a few ports that break.21:02
*** prologic has quit IRC21:26
*** prologic has joined #crux-devel21:29
*** prologic is now known as Guest1418621:29

Generated by 2.11.0 by Marius Gedminas - find it at!