IRC Logs for #crux-devel Monday, 2007-06-11

*** pitillo has joined #crux-devel00:12
*** sepen has joined #crux-devel02:52
*** treach has joined #crux-devel06:52
*** treach has quit IRC07:11
*** treach has joined #crux-devel07:11
*** deus_ex has joined #crux-devel07:45
*** deus_ex is now known as pedja07:46
*** Romster has quit IRC08:56
*** Romster has joined #crux-devel08:59
aontilman: in case you haven't done this already, http://aon.iki.fi/misc/pkgutils-bsdtar.diff09:14
aon:)09:14
*** pedja has quit IRC09:42
tilmanaon: i didn't do the second chunk10:04
tilmanheh10:04
aonoh10:05
clbUpdate from opt: 11 Jun 14:57 - slim: fixed dependencies10:06
pitillonice... Sip was fast...10:09
*** pitillo has quit IRC11:52
*** pitillo has joined #crux-devel11:56
*** sepen has quit IRC12:04
*** prologic has quit IRC12:16
tilmanaon: in the 2nd chunk of the patch, we could drop the "c" argument12:31
tilmanerrr.12:31
tilmanthe "z"12:31
tilmanat least i think so12:31
aonoh, so it detects format from extension also when compressing?12:32
aongreat12:32
tilmanoh12:32
tilmani actually have no idea12:32
aonoh?12:32
tilmanit's true for the list (t) command at least i think12:32
tilmansorry, i'm half-asleep12:33
aonah12:33
aonokay12:33
aonyeah12:33
tilmanlong weekend messes with my sleep schedule12:33
aonprobably so12:33
tilmantilman@hammerfest [~/.tmp/pkgmk] > file openbox#3.4.0-2.pkg.tar.gz12:33
tilmanopenbox#3.4.0-2.pkg.tar.gz: POSIX tar archive12:33
tilmanso it doesn't work in create mode12:33
aonyep12:33
aonbtw, what's the 0 that appears with t as the third or so field?12:36
aoni don't recall it ever being anything else12:36
* tilman RTFM12:39
tilmanaon: fyi, the compression flags (z, j, y) are ignored in extract or list mode12:40
aonokay12:40
tilmani kind of thought it was a shortcut12:40
tilmanto prevent autodetection12:40
tilmanbut apparently it really is smarter than that12:41
aongit actually is quite nice12:44
aon:)12:44
tilmanerr, yes ;P12:44
tilman    bsdtar_strmode(entry, tmp);12:44
tilman    fprintf(out, "%s %d ", tmp, (int)(st->st_nlink));12:44
aoni have three branches of pkgutils and i still think i know what i'm doing12:44
aonah, okay12:44
tilmanthe mysterious number is the number of links to that file12:44
aonnifty12:45
tilmani kind of like the idea of having bsdtar in opt with crux 3.0 or something12:47
tilmanmv core/tar core/gnutar12:47
tilmanln -s whatever12:47
aonor just have it install as /bin/tar in both ports12:48
aonand echo 'ZOMG USE YOUR BRAIN FFS' > bsdtar/README12:48
tilman:>12:49
tilmancould they still be in core in that case?12:49
tilmani *think* theres an unwritten rule that core ports shouldn't fuck other core ports over12:49
aonwell, hmm12:49
tilmanthe mystery number seems to be wrong, btw12:49
tilmanor i don't know what n_links means12:49
aon20:47 < tilman> i kind of like the idea of having bsdtar in _opt_ ...12:49
tilmanbleh12:50
tilmanvery good point12:50
tilman:D12:50
aonnumber of hard links12:50
tilmanthat's what i guessed12:50
aonand that's wrong too?12:50
aon:s12:50
tilmanbut it's zero for everything in git12:50
tilmanand tehre's a lot of hard links in git12:50
tilman             nlink_t   st_nlink;   /* number of hard links */12:51
tilmanoh well12:51
aonhmm12:51
aon?rwxr-xr-x  0 root   root        0 Jun  4 21:44 usr/bin/git-pack-refs link to usr/bin/git12:52
aonbut it still says that there12:52
aoni've come across some fuckups in the bsd/gnu stat compatibility before12:52
aonwhy is it '?' and not 'l' :o12:52
tilmani would expect the entry for usr/bin/git have a n_link of 250 or so12:53
aonhmm, nevermind12:53
tilmanmmh12:53
tilmanthough12:53
tilmanmaybe it makes sense to have the same number for the linked files, hehe12:53
tilmani'm so useless today12:53
aondunno12:54
tilmanbut this smells like a bug12:54
* treach wonders about the benefits of bsdtar instead of gnutar..12:56
aonbeech[~]$ ./test12:57
aon7012:57
aonbeech[~]$ sed -n '/main/,$p' test.c12:57
aonint main(void)12:57
aon{ struct stat sbuf; stat("/usr/bin/git", &sbuf); printf("%d\n", sbuf.st_nlink); return(0);12:57
aon}12:57
aon;-|12:57
tilmantreach: gtar is three times as big a download as libarchive13:01
tilmanbsdtar auto-recognizes compression mode13:01
tilmanbsdtar tolerates minor errors from gzip ;)13:02
aonhttp://www.jwz.org/hacks/why-cooperation-with-rms-is-impossible.mp313:03
aondo you really have to ask? :D13:03
jjpko_o13:04
jjpkClearly not rms' strong points :D13:05
treachmh  "tar" is almost 2 MB.. seems right on the path to EXCESS. (now in the right window.. :o) )13:09
treachotoh, is it that enough of a deal to risk incompatibilities?13:10
aonwhich incompatibilities?13:10
treachgnu/bsd?13:10
aonare there any?13:10
tilmanwhat does the free software song have to do with impossible cooperation?13:10
tilmanright, nothing13:10
tilmanjwz sucks13:10
aondid freebsd use gtar before they did that13:11
tilmanhe was cool in 9513:11
aonor ustar?13:11
tilmanjust like esr13:11
tilman:P13:11
treachnot that I know, but I wouldn't be suprised if there were.13:11
treachsurprised*13:11
aoni spelled 'surprised' as 'suprised' for about ten years before anyone caught it13:12
aon:o13:12
treachfor instance tar -c blah > blah.tar doesn't work with bsdtar iirc.13:12
treachso some level of incompatibility exists.13:13
aonapparently it doesn't yeah13:13
aoneverything else seems to default to a tape device except gnu13:15
aonwhere everything else = the two implementations i cared to test13:15
aonso you need to do -cf -13:16
treachI'm not talking about "right or wrong", just for the sake of clarity.13:16
treachjust pointing out that some things might not like to get it's presumptions upset.13:17
aonindeed13:17
treachanyway, afk, bbiab.13:17
aontilman: the difference between them is that jwz doesn't claim to have written kernel stuff that he isn't credited for :)13:18
tilmanok, so he sucks less than esr13:20
jjpkeveryone sucks, some more and some less ;)13:22
*** pitillo has quit IRC15:11
*** sepen has joined #crux-devel15:18
*** deus_ex has joined #crux-devel17:17
*** deus_ex is now known as pedja17:18
*** treach has quit IRC18:05
*** treach has joined #crux-devel18:05
*** sepen has quit IRC18:31
*** treach has quit IRC18:51
*** prologic has joined #crux-devel23:06
*** treach has joined #crux-devel23:55

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