irclog2html for uclibc on 2002.08.30

14:09:32sjhillgreetings mjn3
14:11:47mjn3morning
14:12:59sjhillis spending ALL day on uclibc
14:13:52mjn3is spending most of the day looking at a zaurus, but he does have some uclibc stuff to do too
16:13:23mjn3morning
16:15:25anderseemorning
16:21:22sjhillafternoon :)
16:23:56andersee:-)
16:40:49sjhillandersee: got a minute?
16:43:24sjhill*sigh*
16:46:45mjn3sjhill: looks like he's been away for about 11 min
16:47:10andersee was last seen on #uclibc 0 seconds ago, saying: ibot: seen andersee? [Fri Aug 30 17:47:10 2002]
16:47:10anderseeibot: seen andersee?
16:48:00sjhillglibc has all these fscking undocumented functions
16:48:37sjhill'get_current_dir_name' is #if 0'd out...hmm
16:49:50mjn3glibc is awful about undocumented functions...
16:50:11sjhill'warn' 'errx' the list goes on
16:50:12mjn3gets annoyed with people who think this is uCglibc and is supposed to support them
16:50:42sjhillmjn3: i know, i'm trying to change the code to use normal documented C functions
16:51:02mjn3what files are you looking at?
16:51:19sjhillstuff from the Linux Test Project (LTP)
16:51:41sjhilloh, in glibc.... 'misc/err.c' the 'warn' function
16:52:41mjn3imho, you're wasting your time with LTP.  it might be useful to extract some tests from it, but what they seem to be testing is glibc compatibility... not, say, SUSv3 or something that isn't a "defacto" std
16:53:30sjhillmjn3: in actuality the problems i'm having are with their error/info printing functions :/
16:53:56mjn3i think some of that got added a while back... looking
16:54:42mjn3you're talking about the err.h stuff, right?
16:54:48sjhillyes
16:55:31mjn3ok.  i was thinking about the error.h stuff
16:56:27mjn3just checked and none of that appears to be SUSv3...
16:57:16sjhillmjn3: how can i find out if stuff is SUSv3 or not?
16:57:27mjn3they wouldn't be difficult to implement, if you needed them and the glibc stuff was too bloated
16:58:09mjn3http://www.opengroup.org/onlinepubs/007904975/toc.htm
16:59:45anderseemjn3: I was thinking it would be pretty easy to implement the err.h junk using inlines
17:00:13mjn3quite probably
17:00:16sjhillmjn3: thx a bunch, i'll use that as my excuse to hack up LTP
17:00:38mjn3no problem
17:01:17sjhillandersee: but is it really worth it :)
17:01:20mjn3andersee: i really think we need a "include glibc compatibility garbage" option
17:01:41sjhillshivers
17:01:41anderseemjn3: An excellent suggestion...
17:02:02anderseemjn3: So we could disable everything but SuSv3/C99 by default
17:02:06mjn3maybe sometime next month you'll have time to get the new config system in?
17:02:11anderseeyeah
17:02:16anderseemumbles
17:02:20mjn3i think that would be great
17:02:39mjn3doesn't speak mumble
17:02:54anderseeIts funny, I'd had the thing working for months
17:03:06sjhillhow will we handle Linux specific system calls...well i guess that doesn't matter
17:03:07anderseeBut I've just never been very happy with it.
17:03:51anderseeI really should split it into two parts...  Changing the config system and then, seperately, changing all the makefiles.
17:04:08sjhill*nod*
17:04:15anderseeRight now I have the two tasks totally coupled, but they need not be.
17:04:17mjn3i think it would be safe to include the syscalls.  i'm talking about things like the error.h, err.h, ... glibc extensions
17:06:54anderseemjn3: right.  All the things like asprintf, error, stpncpy, etc.
17:08:09mjn3yeah... the stdio stuff is even more fine-grained of course.
17:18:26mjn3andersee: of course, that would mean also hiding prototypes, typedefs, struct defs, etc in the glibc headers...
17:18:58mjn3andersee: uh... ignore that.. that's already taken care of by features isn't it..  duh...

Generated by irclog2html.pl by Jeff Waugh - find it at freshmeat.net! Modified by Tim Riker to work with infobot logs, split per channel and by date, etc.