IRC Logs for #crux-devel Monday, 2009-07-20

*** pedja has quit IRC00:19
*** mike_k has joined #crux-devel01:06
*** sepen has joined #crux-devel01:20
*** tilman has quit IRC03:55
*** tilman has joined #crux-devel03:55
tilmanre. 64 bits stuff04:27
tilmanatm, the "pure 64" ports repos are an overlay for the plain i686 ports04:28
tilmanwould it make sense to have the multilib ports be an overlay for the pure64 overlay?04:28
nipuLnot a fan of the overlay idea04:37
nipuLall ports for a given arch should have an arch maintainer04:37
tilmanthat's what we thought initially04:38
tilman'pure' x86_64 is the same arch as multilib x86_64, no?04:39
tilmanso could we share those ports at least?04:39
tilmanwe'd have to take care that ports for foo and foo-compat32 don't diverge though04:39
nipuLyeah, the only ports that differ are in the toolchain04:40
nipuLplus all the compat32 stuff of course04:40
*** jue has joined #crux-devel04:53
aonmorning jue04:53
aonand yes, i agree that sbcl is best for stump04:54
tilmanhi jue04:56
jueaon: good to know for sure :)05:03
juebtw, I've marked 5 ports in nipuL's list with a *05:04
juemight take libexif myself because gphoto depends on it05:05
aonoh, and re: whatever you need to do with lib{attr,cap}, "just do it"05:07
jueaon: thx, but all the panic about udev was unnecesssary, so there's nothing left we have to change05:13
aonoh, ok05:14
juetilman: xz needs a huge, or better insane, amount of memory to decompress files05:31
juels -l metalog-1.tar.lzma | cut -f505:31
juexz -d metalog-1.tar.lzma05:32
juexz: metalog-1.tar.lzma: Memory usage limit reached05:32
juexz: Limit was 49 MiB, but 65 MiB would have been needed05:32
tilmanjue: yes. depends on how the file was compressed though05:33
juethat's on my router with 128M ram05:33
tilmanthe higher the compression ratio, the higher the dictionary size -> the higher the needed memory to decompress05:33
juewell, but 65M to decompress a 0.3M file :(05:34
juehmm, got the same message with coreutils (4MiB)05:38
tilmani thought you already tested extracting coreutils on your router?05:39
tilmansome weeks ago05:39
tilmanis this with unxz, or with pkgmk?05:39
jueunxz, resp. xz -d05:40
tilmannote that unxz might choose a different memory limit than libarchive05:40
tilmans/might/very likely will/05:41
juewill test it now05:41
tilmani think you can specify that limit on the command line somehow05:41
tilmanxzdec --memory=MAXNUMBYTES05:42
tilmanno idea whether unxz accepts that argument as well05:42
tilmanjue: by default, the memory limit is 40% of the available memory05:44
tilmanyou'll need 64 mb to uncompress the coreutils tarball05:44
tilman40% of 128 mb is less than 64 mb05:44
tilman-> fail05:44
jueyeah, 'xz -d --memory=65MiB coreutils-7.4.tar.xz' works05:52
juehmm, metalog-1.tar.lzma as well05:57
juelooks like a fixed lower limit05:57
sepenjue, which opt port needs libexif?05:58
sepenok, I'll mark hal-info on the list06:00
sepenoops, prt-get dependent libexif isn't working fine here06:02
juesepen: what's the problem?06:04
sepenprt-get dependent libexif should return at least libgphoto2, right?06:05
sepenI wrote a little script to obtain a list based on orphaned ports, so I'm using the prt-get dependent feature, this is my result
sepenthat's the script, and that's the orphaned list
juebug in your script?06:08
sepenoh sorry, remove '|'06:08
sepenI whould filter the output lines to be more tabbed, ;D06:09
sepenthe script ->
juetilman: works with new pkgutils :)06:40
juesorry for the noise06:40
juesepen: why hal-info?06:46
pitiIIohas anyone the orphaned list link to take a look?06:50 iirc06:50
pitiIIotilman: thank you06:51
*** mike_k has quit IRC07:11
sepenjue, hal-info is required by hal07:29
jueyeah, but hal is in the list as well07:32
jueso hal-info has to stay in opt only if hal remains in opt07:32
sepensame as dbus, and dbus-glib07:33
jueno, dbus-glib is a dependency of xulrunner07:33
sepenah ok I see the difference07:34
sepenjue, fixed it, thanks07:34
sepenjue, could you type $ prt-get dependent libexif ? I noted that it can't detect libgphoto2 if you have the prtdir to the opt collection in your prt-get.conf07:39
juegot libgphoto2, core/opt/xorg in prt-get.conf07:42
sepenhmm, 2.5?07:43
nipuLprt-get dependent --all libexif07:44
nipuLwithout the all it only checks installed ports07:44
sepennipuL, thanks I was wrong07:44
sepennipuL, yeah like pkginfo -o for only installed ones07:45
nipuLgah, util-linux-ng is a pain to build a compat32 for07:46
nipuLspecify --libdir=/usr/lib32 and it still puts stuff in /usr/lib07:46
nipuLand there's not --syslibdir :\07:46
nipuLanyhoo, i better start training my self for earlier nights (and as a result, getting up earlier), as i'm back at uni next week07:48
nipuLi'm going to go head butt a pillow until i pass out07:48
sepena list of maintainer candidates for nipuL's orphaned ports
sepen$ grep -v Lucas, ;D07:54
juenice, thanks sepen07:58
jaegerI don't think I've mentioned it but I'm back to where I have some more time to put into crux stuff if I'm needed to do a bit more08:04
juejaeger: glad to hear that!08:09
jueone thing that came instantly to my mind is the handbook for 2.608:11
jueesp. the chapter about available kernel drivers08:12
jaegerI'll take a look08:13
tilmani'll fix the handbook to mention ext4 is available, too08:13
jaegerjue: section 3.1?08:13
jaegerAh, I was looking at Handbook2-608:15
jueI've converted the handbook to 2.6, done, hopefully, all s/2.5/2.6/, but nothing more08:16
jaegerIn the handbook it says 160MB memory is needed to install, how did we test that for 2.5?08:17
juejaeger: btw, thanks for test1, I guess you've read about the bug I've found?08:17
jaegerif you mean xz missing from the injects, yes08:18
jueyes, that was it08:18
juejaeger: IIRC mike_k did some tests and added 64k or so to be sure08:20
jaegerI just wondered how the tests were done... in a VM with successively lower limits?08:20
jueat least it works for me with the i586 with 128M08:20
juesorry, don't remember08:20
jaegerok, no worries08:20
sepenjaeger, I think it was discussed here before 2.5 was released08:21
sepenmaybe in the ircbacklog08:21
jaegerseems like it was, yeah08:21
tilmanjue: do you think cpnt will mind if i patch opt/slim to build with g++ 4.4?08:25
jueI think you can go for it08:30
sepenjaeger, I think I found the conversation (or at least a part of it)
jaegerHrm... I wonder how much AoE support would add to a boot CD08:36
tilmanata over ethernet?08:36
jaegersepen: ok08:38
jaegertilman: yeah.. I was thinking again about network installs. the netinst iso hasn't been updated any time recently but AoE could be nice for it08:38
jaegerfor example to mount a block device with packages on it from another host without nfs, etc.08:40
tilmansounds useful :]08:42
jaegerI'd like to get the netinst ISO back in shape, I'll look into that soon as well08:43
tilmanno idea how widespread aoe is, but i guess you do08:43
jaegerI'm not sure it's widespread at all but I've used it myself a few times08:43
tilmanthat's good enough a reason to offer it :) (well, depends on how much work it is to implement08:46
jaegerIf I remember correctly it's 2 ports and a kernel tweak08:49
jaegerit wasn't too complicated08:49
*** Rotwang has joined #crux-devel09:12
*** pedja has joined #crux-devel10:06
*** sepen has quit IRC11:01
tilmanjue: i think you said pkgmk should recognize *.tar.lzma, too?12:27
tilmanjust noticed that libpng tarballs are available as tar.lzma as well12:27
juetilman: yeah, would be nice, imagemagick is using .lzma too13:01
juebtw, installing xz _before_ the update seems the best solution to the test1/xz issue13:02
tilmanjue: i wonder whether we should add *.tar to that list13:20
tilmanwell,maybe not13:20
*** jue has quit IRC14:37
jaegerdang it, there you go with opt updates after a bootstrap :)15:31
*** Rotwang has quit IRC17:00
nipuLfscking autostools21:07
nipuLfinally got my new core sorted, now to make some public repos, this ports tree will have a working ports port.... yo dawg21:34
nipuLwow, opt is going to take even longer :\22:16
nipuLfortunately, the merged repos are compatible with my existing ones22:17
nipuLgit ftw22:17

Generated by 2.11.0 by Marius Gedminas - find it at!