Re: coreutils 6.6 fails to compile on IRIX 5.3

[Jim Meyering]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Georg Schwarz <address@hidden> wrote:
> that was also my thought, but I would prefer to investigate only once
> the other things are properly solved so that any potential side
> effects can be ruled out.
>
>> get a GDB backtrace of the first core dump you find, to see more
>> details about it?
>
> I will try, but first I will have to build gdb (which might be a
> challenge, we'll see..).
> Since I would like to do so with a "clean" code, could you maybe
> provide me with the "processed" (automake etc.) current sources?

I've run "make dist" using the very latest of gnulib
and coreutils.  Here is the result:

  http://meyering.net/coreutils-6.7-dirty.tar.bz2

Here's its SHA256 checksum:
928bfff5da8b37f205c01c3ae3c6851d3c3c51a95efcc95809dabe5327397fcb  
coreutils-6.7-dirty.tar.bz2

>>> This is C99 syntax I think which the IDO cc does not like.
>>
>> You need to apply the patch in src/c99-to-c89.diff if you have an old
>> compiler that doesn't grok declarations after statements.
>
> why don't you integrate that for all compilers? It should not hold
> any penalties for newer ones, does it?
> I could imagine maintenance being easier without having any diffs
> included.

I have a strong preference for the newer syntax.
True, there is a little maintenance work in keeping
the patch up to date, but that is offset by the fact
that the c99 code is cleaner and more maintainable.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFFn8ud/dLerNMzy6ERArAKAKCwYc95IsYGQmVtJLHUZ5LkTglUxwCeJcaL
RubZW25+2H+pFhhd0wMN0L8=
=Oc0V
-----END PGP SIGNATURE-----