IRC Logs for #crux-devel Sunday, 2015-08-09

frinnstI added a rsync job that copies all backups to my server btw, if anybody's wondering00:47
frinnstfrom crux.nu00:47
frinnstuses a wee bit cpu - 70% or so01:01
*** Feksclaus has quit IRC01:10
Romstershouldn't take that long after the first sync02:13
*** ________mavric has quit IRC02:41
*** ________mavric has joined #crux-devel02:42
*** heroux has quit IRC02:49
*** heroux has joined #crux-devel02:50
*** heroux has quit IRC03:03
*** heroux has joined #crux-devel03:03
*** heroux has quit IRC03:28
*** heroux has joined #crux-devel03:30
*** heroux has quit IRC03:45
*** heroux has joined #crux-devel03:46
teK___so you do not face an performance issues?10:54
teK___it would have been the first solution if it would have worked10:54
teK___remember? :)10:54
teK___but otherwhise: awesome. (I thought you polled from ./backup anyway)10:57
*** Feksclaus has joined #crux-devel12:42
frinnst20-30mbit12:56
teK___great12:56
*** Feksclaus has quit IRC13:35
frinnstAnnouncing ncurses 6.014:46
frinnstwee14:46
frinnstThis release is designed to be source-compatible with ncurses 5.0 through 5.9; providing a new application binary interface (ABI). Although the source can still be configured to support the ncurses 5 ABI, the intent of the release is to provide extensions which are generally useful, but binary-incompatible with ncurses 5:14:46
frinnsthttp://www.gnu.org/software/ncurses/14:47
teK___recompile all the things :O)14:53
*** Workster has quit IRC15:14
frinnstnah. pretty much only python15:29
frinnstvery surprising15:29
*** Workster has joined #crux-devel15:33
teK___you found out by manually testing?15:33
frinnstyeah15:42
frinnstcmake and a few other things broke too. but nothing major15:43
teK___you will have to follow revdep to be sure, which leads us to:15:45
teK___< teK___> recompile all the things :O)15:45
teK___right?15:45
frinnston my server only python reported errors in revdep15:49
frinnstand my desktop python, cmake and some other thingy15:49
frinnstRUN REVDEP YOURSELF IF YOU DONT BELIEVE ME!!!!!!!!!!!!!!!!!15:49
frinnstguess it didnt completely break abi15:50
frinnstfredrik@nibbler:~$ revdep15:52
frinnstpython15:52
frinnstalsa-utils15:52
frinnstcmake15:52
frinnstgst-plugins-good15:52
teK___it was more uncertainty than disbelief :)15:53
teK___swhat about bash? (mwahaha)15:54
teK___wait. bash still is a hard req. for 3.2. I'm taking back the laughter15:54
teK___here, bash is lined against libncurses.so.515:54
frinnstmwhaha15:54
frinnsthere too15:55
frinnstwtf15:55
teK___UH-OH15:55
frinnstfredrik@nibbler:~$ ls -lh /usr/lib/libncurses.so.515:55
frinnstlrwxrwxrwx 1 root root 12 okt  8  2013 /usr/lib/libncurses.so.5 -> libcurses.so15:55
teK___wat.15:56
frinnstindeed15:56
teK___I do that sometimtes manually to make old / statically linked cruft work15:57
frinnstwhy arent .so.5 removed?15:57
frinnstbut on two boxes?15:57
teK___% pkginfo -o usr/lib/libncurses.so.515:58
teK___pkginfo: no owner(s) found15:58
teK___% ls -lh /usr/lib/libncurses.so.515:58
teK___lrwxrwxrwx 1 root root 12 [2012/12/22] /usr/lib/libncurses.so.5 -> libcurses.so*15:58
teK___:P15:58
frinnstREMOVE ALL THE THINGS16:01
teK___:D16:02
frinnsthm, cleanish vm, libncurses*.so.5 still remain after updating16:07
frinnstwonder why pkgadd doesnt remove it16:07
teK___is it in the pkg db?16:07
teK___for me it wasnt16:07
frinnstno16:08
frinnstright16:08
frinnstclean 3.1 vm, only /lib/libncurses.so.516:08
teK___I just realized, that we probide even docker images but no VM images for a quick test driver16:09
teK___-er16:09
frinnstso after upgrading to ncurses 6, i get a new unowned libncurses.so.6 file under /usr/lib16:13
teK___oh my16:13
frinnstthey should only be in /lib/16:14
frinnsthm, perl 5.22 no longer provides s2p/psed16:54
frinnstbut our port creates a broken symlink for psed16:54
frinnstI wonder if ncurses might not respect --with-install-prefix=$PKG completely17:31
teK___was my first thoght, yeah17:32
frinnstIm not sure why we build ncurses as we do17:36
teK___consult git17:36
teK___git log17:37
teK___or wait fro jue, I'm sure he knows17:37
frinnstyeah Im sure he does17:44
frinnstlooks like tilman did the change17:46
frinnstyep. when ncurses is built as root, it installs libncurses*.so.618:06
frinnstcreated a bug for it so i wont forget. too lazy to look into it more right now18:16
teK___upstream is supposed to fix this18:22
teK___[probably]18:22
*** Lukc has quit IRC18:43
frinnstgrr, very frustrating19:33
frinnstnot really a big deal but very ugly19:33
teK___so no vendor-error?19:40
frinnstpossibly19:59
frinnstasdf20:17
frinnststills installs crap with DESTDIR=$PKG install20:18
jaegersome manual makefile patching needed?20:19
frinnsthm. might be a pkgmk bug20:25
frinnst    for i in w ''; do20:25
frinnst        mv $PKG/usr/lib/libncurses$i.so.* $PKG/lib20:25
frinnst        ln -sf libncurses$i.so $PKG/usr/lib/libcurses$i.so20:25
frinnst        ln -sf ../../lib/libncurses$i.so.$version $PKG/usr/lib/libncurses$i.so20:25
frinnst    done20:25
frinnstln -sf libncurses$i.so $PKG/usr/lib/libcurses$i.so <-20:26
frinnstno, nevermind20:27
frinnstthat doesnt make sense20:27
frinnstif I comment out that line: ln -sf libncurses$i.so $PKG/usr/lib/libcurses$i.so20:32
frinnsti get this:20:32
frinnst=======> ERROR: Footprint mismatch found:20:32
frinnstMISSING   lrwxrwxrwx      root/root       usr/lib/libcurses.so -> libncurses.so20:32
frinnstMISSING   lrwxrwxrwx      root/root       usr/lib/libcursesw.so -> libncursesw.so20:32
frinnstNEW       lrwxrwxrwx      root/root       usr/lib/libcurses.so -> libncurses.so.6.020:32
frinnstand /usr/lib/libncurses.so.6 /usr/lib/libncursesw.so.6 are no longer installed when building as root20:33
frinnstsame with core/ppp20:37
frinnstcompare the footprint with whats installed under /usr/lib/pppd/$version20:38
frinnstI have 11 .so-files. only 9 listed in the footprint20:38
frinnstopenl2tp.so, ppol2tp.so20:40
teK___I can offer to look into it (with you?) tomorrow night :)20:40
frinnstive spammed my findings here: https://crux.nu/bugs/index.php?do=details&task_id=121020:46
frinnstmaybe its naive to rely on various projects doing the right thing and perhaps we need to enforce ports never write outside $PKG ?20:48
teK___something like a forcenorootfs binary?20:50
teK___(analogous to the inversly working fakeroot program)20:50
jaegershouldn't fakeroot solve this issue?20:50
frinnstyup20:52
teK___I am not sure. My take was, that it mimics euid==0?20:52
teK___fakeroot(1) says: fakeroot  doesn't  wrap open(), create(), etc.20:53
teK___it mentions libtricks20:53
jaegerit fakes it so that the process thinks you have euid 0, but it doesn't actually give you any elevated permissions20:53
frinnstwhen you build ppp under fakeroot:20:53
frinnstinstall -c -m 755 pppol2tp.so openl2tp.so /usr/lib/pppd/2.4.720:53
frinnstinstall: cannot remove '/usr/lib/pppd/2.4.7/pppol2tp.so': Permission denied20:53
frinnstinstall: cannot remove '/usr/lib/pppd/2.4.7/openl2tp.so': Permission denied20:53
jaegerwhich is why you have to create a pkgmk group or a user to build ports with fakeroot20:53
teK___yes. And no word about FS wrapping :)20:53
frinnstah20:54
teK___frinnst: of course, as you are not really root anymore20:54
jaegersounds to me like a broken makefile or something20:54
jaegerperhaps in an l2tp subdir?20:54
jaegerone that doesn't support DESTDIR20:54
frinnstyeah ppp is just borken20:54
frinnstncurses is however more wierd20:54
teK___I remember working around missing DESTDIR support in some ports20:54
frinnsttime to decompress. work tomorrow D;20:56
teK___good night :]20:57
jaegernight20:59
teK___I'm asking myself, if decompressing is meant to describe the process of getting to bed21:02
jaegerI take it as relaxing21:03
teK___he has been on vacation for at least a week. He tricked his relatives as to he was working the whle week. There's no reasons for slacking :>21:04
frinnstrelax a few hrs before bed obviously!21:09
teK___obviously!21:10
*** Feksclaus has joined #crux-devel23:24

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