IRC Logs for #crux-devel Wednesday, 2020-03-04

*** pedja has quit IRC00:38
*** SiFuh has quit IRC01:13
*** SiFuh has joined #crux-devel01:20
*** xor29ah has quit IRC04:03
*** xor29ah has joined #crux-devel04:03
*** xor29ah has quit IRC06:00
*** xor29ah has joined #crux-devel06:32
*** xor29ah has quit IRC07:59
*** tnut has joined #crux-devel10:01
tnutHi10:02
TimB_jaeger: sorry i didn't answer yesterday, i was tired and still busy with the updates i pushed. what do you think is the best way to proceed with this? I normally strive to have a .footprint that would only produce new files, because this is configurable for prt-get to ignore10:04
TimB_tnut: hello10:04
tnutHi TimB_10:05
*** xor29ah has joined #crux-devel10:21
TimB_http://dpaste.com/3CFPYJV does anybody have an idea on why i get these? MISSING is what my desktop produces, NEW is what my container produces. Both use umask=022..10:24
ryuoTimB_: are you sure about that?10:43
ryuoalso umask doesn't stop files from ending up with different permissions10:43
ryuoit just effects the implicit permissions on new files10:43
TimB_ryuo: yeah, i am sure, double checked10:44
ryuosomething could be modifying the umask explicitly10:44
ryuothough oddly why not in both10:44
TimB_on my desktop i use zsh, that's the only difference i can see10:44
TimB_however, bash -l and rebuilding will produce the same oddity10:45
ryuowait...10:45
ryuoit happens on a login shell?10:45
TimB_yes10:45
ryuocheck your profiles. login shells do more than a normal subshell.10:45
TimB_/etc/profile sets umask 02210:46
ryuoi forget, what permission bit is 2...10:46
ryuowrite?10:46
ryuohm.10:46
TimB_setting of write permission is prohibited (read and execute)10:47
ryuoTimB_: try setting the umask in the pkgmk and see if that makes a difference10:47
ryuoif it does then there's something fishy going on but pkgmk should probably set this already10:47
ryuoinstead of relying on the inherited one10:47
ryuoI mean10:48
ryuosetting it in the Pkgfile10:48
ryuoI try weird shit when i can't find the fault.10:48
TimB_it doesn't10:48
TimB_tried on both desktop and container10:48
ryuoso adding an explicit umask to the Pkgfile produces no difference?10:48
TimB_yeah10:49
ryuoweird. this is the only file in the footprint with10:50
ryuo077510:50
TimB_i'll be back in around an hour and a bit, i hope. thanks ryuo!10:51
*** tnut has quit IRC10:52
ryuoTimB_: could always add a hack to change the permissions manually10:52
frinnstfakeroot etc has been known to cause issues like that11:04
*** tnut has joined #crux-devel11:13
ryuoalso something to note is fakeroot isn't 100%11:27
ryuoit doesn't intercept static binaries11:27
*** xor29ah has quit IRC11:43
*** xor29ah has joined #crux-devel11:46
*** xor29ah has quit IRC11:52
*** xor29ah has joined #crux-devel11:56
TimB_fakeroot is a god hint, thanks frinnst12:02
*** pedja has joined #crux-devel12:06
TimB_indeed, it was fakeroot12:21
frinnstI dont think upstream even recommends building stuff with fakeroot12:38
frinnsti remember reading that somewhere12:38
TimB_mh, not sure12:38
TimB_i might dismiss fakeroot then.. if there is no possibility to make it work12:39
TimB_maybe the wiki should be adjusted to contain a warning about that?12:39
TimB_https://crux.nu/Wiki/FakerootPorts12:39
pedjawhile I used Crux, very few packages wouldn't build using fakeroot, if that's the issue here12:47
TimB_pedja: indeed, but still, as a maintainer i am interested in delivering the right footprint12:47
pedjalibvirt devs considered tests failing with fakeroot as a bug, and fixed that when I mentioned it :)12:49
pedjaTimB_, what's the issue you've been having? footprint mismatch fakeroot vs root build?12:51
TimB_pedja: exactly :)12:51
pedjathat's an upstream bug, then, since, afaik, Debian is using fakeroot to build packages12:52
pedjahow serious is the mismatch? wrong permissions?12:53
TimB_just permissions, all else will work fine12:53
pedjawhich package?12:54
TimB_sip5 and some others, i would have to check12:54
TimB_i noticed it again with sip5 yesterday and was reminded that i still wanted to take care of that12:54
pedjaI am guessing they are using autotools to build?12:55
TimB_sip5 is a python package12:56
TimB_but good to know, i will look out for those12:56
pedjaspeaking of python, any plans of purging python2 packages, leaving just the interpreter?12:58
frinnsti remember syslinux created some oddities ~5-10 years ago14:07
frinnsthttps://crux.nu/gitweb/?p=ports/opt.git;a=blobdiff;f=syslinux/Pkgfile;h=a21c6de84804627d36f5394555313a23d2bde225;hp=6c290fa6a2b7c0e6eaaeda626c05707b5731b3ab;hb=b2663b6ae9ee76a951b8ef0398f713652c8c8230;hpb=a930e1375e2419d56351088248512861c9f257d214:10
frinnst201114:10
frinnstmy memory is mighty!14:10
TimB_400 - Invalid hash parent base parameter ? :P14:11
TimB_found it though, interesting14:13
jaegerregarding xkeyboard-config, I'd just make it depend on libxslt15:14
TimB_or that :)15:14
jaegerI use fakeroot for everything, it's rare that it causes issues but does on occasion15:22
TimB_maybe i should just stop using it in the container to initially build..15:23
frinnsti use 'fakeroot pkgmk' as the makecommand in prt-get.conf :-)15:28
jaegersame15:29
TimB_i am not sure what to do. i mean the best thing would likely be to report the fault upstream?15:43
jaegerdoesn't hurt to try, probably15:49
TimB_done16:05
*** tnut has quit IRC16:11
*** tnut has joined #crux-devel19:00
*** tnut has quit IRC22:34
*** pedja has quit IRC23:12

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!