use of legacy C library functions

[Rob Landley]

On Monday 05 April 2010 16:50:13 Denys Vlasenko wrote:
> On Monday 05 April 2010 20:24, Rob Landley wrote:
> > And I'm trying to figure out if this is a config issue or due to:
> >
> >   libc_hidden_def(fcntl64)
> >
> > In libc/sysdeps/linux/common/__syscall_fcntl64.c.  (Why "hidden"?  What's
> > the point in hiding it?  It's a libc export, right?)
>
> I had similar feelings. I tried to improve include/libc-symbols.h
> comments and structure so that it is not as confusing as it was.
>
> I also removed massive amounts of libc_hidden_proto's in .c files.
> (Need to do the same for libm, libutil etc.)
>
> There are a lot of places in uclibc which need cleaning up.
>
> Example: just now, I noticed "#undef _LIBC" statements in a few files.
> That's really not how it supposed to be. Compiling libc files
> as if they are not part of libc? That sounds paranoid, and bound
> to create problems. It already did:
>
> libc/misc/fnmatch/fnmatch.c
> ===========================
>
> /* unistd.h must be included with _LIBC defined: we need smallint */
> #include <unistd.h>
> #include <features.h>
> #ifdef __UCLIBC__
> # undef _LIBC
>
> At the very least, whoever added #ifdef __UCLIBC__ # undef _LIBC block
> had to explain why he did it.

Probably a better list to cc: this conversation to than the busybox one. :)

I'm vaguely poking at .31 trying to make it work with my build, but I've got a 
half-dozen things competing with it for attention tonight...

Rob
-- 
Latency is more important than throughput. It's that simple. - Linus Torvalds