IRC Logs for #crux-devel Thursday, 2015-08-20

*** Workster has quit IRC01:29
*** qismrrgyenjgcjnw has joined #crux-devel01:29
*** Workster has joined #crux-devel01:29
nomiusHey guys02:14
nomiusAny link for the 3.2 test iso?02:14
*** ________mavric has quit IRC02:45
*** ________mavric has joined #crux-devel02:46
frinnstnomius: dont use it07:53
teK__let him? :)07:54
teK__the file name indicates not to use it. :P07:54
frinnstI dont remember the url anyways07:58
teK__I grepped it just two days ago. nomius: interested?07:59
nomiusyeah. want to give it a try :-)08:02
teK__21:12 < jaeger> http://jaeger.morpheus.net/linux/crux/files/drop/crux-3.2-notready.iso if you're feeling adventurous.08:03
nomiusthanks!08:03
nomiusqq08:07
nomiusI didnt see the code of SSD, but does it send sigkill after a couple of seconds if sigterm didn't kill the process?08:08
teK__welcome08:08
teK__I have no idea08:08
nomiusok, i'll look in the source after. thanks!08:09
teK__sure08:10
nomiusMay I ask why using dash as /bin/sh instead of bash?08:12
frinnstlot faster08:16
frinnstand /bin/sh gets called loads of times when building packages etc08:16
frinnstwe still rely on /bin/bash for scripts and pkgutils08:17
nomiusThat's interesting. Thanks!08:17
nomiusAre you guys thinking into moving fully to dash in the future?08:18
teK__some day I will convince^Wprovide code replacing bashisms in pkgutils ;)08:32
frinnstyeah its a very low prio08:53
Romsteryou won't gain much09:50
Romstersome makefiles call /bin/sh like 20,000 or so times09:51
Romsterwhere rc scripts only a few09:51
Romsterusing finit or runit would speed up running services in parallel09:52
RomsterSSD takes care of killing and pid files09:53
teK__I manually add '&' to scripts that can be started in background09:56
Romsterah09:57
Romsteri should do that09:57
prologicI actually really like finit10:08
prologicI'm using it val vallinux10:09
prologicit's quite fast and easy to use and configure10:09
prologiccould really fit CRUX nicely too IHMO10:09
teK__frinnst: did you google about enable-zsh-mem?11:34
teK__s/google/$SEARCHENGINE/11:34
*** Workster has quit IRC11:42
newbie__teK__ meant to say: frinnst: did you $SEARCHENGINE about enable-zsh-mem?11:43
teK__thank you11:44
*** Workster has joined #crux-devel12:00
frinnstnot really, no12:49
frinnsti built it locally first with just ./configure to find any debug stuff12:50
frinnstand then it just worked12:50
teK__alright, It slipped my eye ;)12:55
nomiusSo dash is actually the fastest POSIX shell right now or the fastest one with the most compatibility with bash?13:09
teK__ fuck comaptibility with bash. If the script says /bin/sh, it needs to only use features of sh.13:11
nomiusThen dash is the fastest POSIX shell right now. Cool :-)13:27
jaegerfrinnst: the git version of xorg-xf86-video-intel did work fine on the machine that was giving me problems with the released driver, no need to use modesetting13:53
teK__jaeger: did you do git log | grep -i '(mode|setting)'?14:22
jaegerno14:23
jaegerI didn't particularly care about the modesetting driver, just intel14:23
jaegerdid we ever sort out the libncurses.so.5 issue?15:03
jaegerI'm seeing the .so.5 file on my 3.2 iso machine and I don't recall what the end result of that discussion was15:03
teK__I planned to, frinnst had no time and I did not investigate alone; I don't think it is resolved15:06
jaegerok15:06
jaegerodd that libncurses.so.5 has no owner but still exists15:07
teK__that was caused by the sources installing it outside of $PKG15:07
teK__.. did you (re)build it at all?15:08
jaegerin 3.1 or 3.2?15:08
teK__if not, this will be interesting15:08
teK__depends.. is this an upgraded or freshly installed instance?15:08
jaegerI don't remember, to be honest :)15:08
jaegerIt's a VM that I could replace if needed, so not a big deal15:08
teK__heh15:08
teK__is there a quick way to find out.. timestamps in /etc?15:09
jaegerncurses has been rebuilt on it, though15:09
teK__or even easier: ls /boot15:09
teK__if you rebuilt it, chances are, that: < teK__> that was caused by the sources installing it outside of $PKG15:09
teK__s/chances/it is very likely, that/15:09
jaegerso I could likely just remove it if nothing links against it15:10
teK__as long as you did not (re)build something that depends on it afterwards: yes15:11
jaegerat present it appears vim is the only thing linking against it15:11
jaegerso I'm betting this was an upgraded system15:11
jaegerI've rebuilt a lot of stuff on it (and do so repeatedly) because it's an ISO builder15:11
teK__it will be best to wait for frinnst to report about the latest state of affairs or look it up from the logs15:12
teK__proposal: remove the .so.5 file and rebuild ncurses once more15:12
teK__(let's be crazy today)15:12
jaegerI've got to check other linkage first, I was just looking in /bin at first15:13
teK__that does not matter too much, imho15:13
teK__which essential tool is going to break without ncurses? ;)15:13
jaegerit will if something (like bash) is linked against it15:14
teK__there is cat and echo, for christ's sake!15:14
jaegerbash isn't but that's the idea15:14
teK__and sed :D15:14
jaegerdialog, htop, mutt, nano15:14
jaegerprobably safe but I'll rebuilt those too15:14
jaegers/rebuilt/rebuild/ god damn it15:15
jaegerI can't type today15:15
teK__I need to redistribute my private keys more planfully.. I dont have access to my 3.2 host atm. It's killing me to find out if the current package spills the .so.5 file to /usr/lib :x15:16
jaegeroops, heh15:16
jaegerhere's a really odd thing - bash is linked against libncurses.so.6 but if I remove libncurses.so.5 bash won't work15:33
jaegermaybe that's bash linking against libreadline, though15:35
jaegerI'll keep poking at it15:35
teK__nice one15:35
jaegerah, that was it. now bash works without .so.515:35
teK__but either way, we have to fix the no-owner problem, too15:35
jaegeryes, we should. At the moment I just want clean linking :D15:36
teK__and: see if it was caused by the -5 or -6 version of the sources (I guess -5 from CRUX 3.1)15:36
jaegerpresumably the 5 version since it's the .so.515:36
teK__very likely15:36
teK__if this is the case, we ought to fix the 3.1 package and clean/rename the file in setup on our ISO15:37
teK__as an upgrade step15:37
jaegeragreed15:37
teK__maybe move and give notice in the handbook15:38
frinnstJue responded to my ticket regarding ncurses15:44
frinnstkinda waiting for him to show up and magically solve everything as usual :D15:45
frinnstI might be a bit inactive in the following weeks - my bathroom will be replaced so i'll probably move into my dads guest house15:46
frinnstand horrible internetz there15:46
teK__hahaha15:46
teK__ok, have fun15:46
teK__I'll be mostly afk starting Aug 27th through Sep 7th due to wedding, vacation and a new job :=15:47
frinnstI'll probably go crazy and move back home. I'll just have to shit in a bucket :)15:48
frinnstbut what does one not do for 100/100mbit?15:49
jaegerhe's probably correct, ldconfig will create symlinks if it thinks it needs to15:49
teK__srsly?15:50
teK__frinnst: internet > ?15:50
teK__But I am not in a position to judge, because zoidberg will be hit up with 100MBit, too, right? :>15:50
frinnstI already have 100mbit15:53
frinnstwhat I will be moving to is 2/512k15:53
frinnstand really shitty wifi15:54
teK__oh the other way round.15:55
frinnsthttps://www.google.se/maps/@57.7006401,16.726172,3a,37.5y,209.69h,82.47t/data=!3m6!1e1!3m4!1sWHhsPYiyUntmSJqWpXj2nw!2e0!7i13312!8i665615:55
frinnstyou can just spot the guesthouse on the right behind some brush15:56
teK__I think Google stopped doing streetview in Germany because of too many complaints15:57
frinnstyeah it is pretty creepy15:58
teK__even DHL or UPS sometimes are unable to find my house; so I'll be safe16:02
teK__and they have GPS and shit :D16:02
jaegerlooks like someone already fixed ncurses in 3.116:12
jaegerlrwxrwxrwx      root/root       lib/libncurses.so.5 -> libncurses.so.5.916:12
jaeger-rwxr-xr-x      root/root       lib/libncurses.so.5.916:12
jaegerlrwxrwxrwx      root/root       lib/libncursesw.so.5 -> libncursesw.so.5.916:12
jaeger-rwxr-xr-x      root/root       lib/libncursesw.so.5.916:12
jaegerthat's in the .footprint16:12
teK__https://crux.nu/gitweb/?p=ports/core.git;a=history;f=ncurses;hb=HEAD suggests not16:14
jaegerhttps://crux.nu/gitweb/?p=ports/core.git;a=blob;f=ncurses/.footprint;h=f438deecde1bb653f17b3eda57ec1b017f6062cd;hb=refs/heads/3.1 shows the symlinks in the .footprint there, though16:15
teK__so we will have to fix the ldconfig issue only16:16
jaegerI'm not sure there is an issue16:16
jaeger$ pkginfo -o /lib/libncurses.so.516:17
jaegerPackage  File16:17
jaegerncurses  lib/libncurses.so.516:17
jaegerncurses  lib/libncurses.so.5.916:17
jaegerthat's from a 3.1 install, it looks right16:17
teK__ok but if ldconfig creates .so.5 on 3.2 systems?16:17
jaegernope, heh16:18
jaegerAt least it doesn't for me16:18
jaegerit only would if libncurses.so.5.x existed16:18
jaegerso I think this problem would only have happened if a 3.1 system is upgraded via ports rather than ISO16:18
jaegernot exactly sure, though, since the ncurses pkgadd upgrade should have removed the .so.516:20
jaegerwill have to test a 3.1 to 3.2 upgrade from the ISO but I don't think it'll happen there16:22
teK__18:20 < jaeger> not exactly sure, though, since the ncurses pkgadd upgrade should have removed the .so.5 << this16:30
jaegerI'll test an ISO upgrade and a ports-based upgrade when I get time and see if I can't narrow it down16:37
jaegerI've removed the .so.5 files from my 3.2 ISO build VM now and sorted linking, at least16:38
teK__but ldconfig will not restore the link, if not  libncurses.so.5.x existed correct?16:39
jaegerassuming it is working properly, yes16:39
teK__ok16:41
jaegereven back in 2008 ncurses had these symlinks. at this point I have no idea what caused the orphaned symlinks but I'll try to figure it out when I do upgrade tests16:42
jaegerhttps://crux.nu/gitweb/?p=ports/core.git;a=commitdiff;h=214a87e3ef1c6b3fcb8ce3f84b6cc7291e86ab8f <- even back to this one16:42
teK__you might set a fnotify watch :P16:47
teK__*set up16:47
jaegerworth a try16:47
*** heroux has quit IRC19:44
*** heroux has joined #crux-devel19:45
jaegerugh. I'm going to have to run a custom openssh port here at work because I have networking hardware and other crap with old ssh setups20:00
frinnstyeah my android rsync+ssh client doesnt work either :(20:07
frinnstbtw, anybody tried building chromium in 3.2?20:07
jaegernot I20:07
frinnstI had some build error at work but never looked into it20:07
nomiusI'll probably do it tonight20:49
pitillois just me? https://crux.nu/gitweb/?a=project_list;pf=ports23:28
pitillohttps://crux.nu/gitweb/?p=ports/opt.git;a=summary23:28
pitillojaeger: ping23:29
jaegerwhat's up? aside from the gitweb thing23:54
pitillojust that23:56
jaegerask teK__, it's probably his fault :D23:56
jaeger(I'm guessing it's related to the recent permissions problem)23:56

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