Subject: Summary: gcc -ansi -pedantic and long long
To: None <current@netbsd.org>
From: Matthieu Herrb <matthieu@laas.fr>
List: current
Date: 11/11/1994 17:13:10
I'd like to thank everybody who participated to this thread.
There where 37 messages, contributed by 17 persons. Here's a summary
of the discussion.

The flame war that began at the end of the discussion was promptly
stopped by an intervention of matthew green <mrg@mame.mu.OZ.AU>. 

Now for the interesting part:

8 of the participants think that something has to be done in gcc in
order to allow a clean compilation of header files using 64 bit ints
under '-pedantic'. 64 bits long are not the solution on 32 bit
machines. 

3 more  participants suggested (without trying) to use the
__extension__ keyword of gcc in the offending headers. Unfortunatly,
__extension__ works only in expressions. So I'm inclined to count them
in the count in favor of a gcc fix. It make 11.

2 participants suggested not to use -pedantic at all and one suggested
to keep it and ignore the warnings.

Ted Lemon <mellon@ipd.wellsfargo.com> offered to submit a patch to
gcc-2.4.5 to support __long_long__ as a type declarator. Perhaps
would be __quad__ more appropriate ?

He also stated that in the latest gcc code if a file is included with
#include <foo> (as opposed to #include "foo"), then long long
declarations within that file are allowed even if the -pedantic flag
is set. I checked that gcc 2.6.1 -ansi -pedantic still barfs on 'long
long' even in files included with #include <...>.

Let me end by saying that if XFree86 wants -pedantic it's to ensure
that it will compile with other compilers on other systems that are
more pedantic than gcc (which is used by most of the developpers)
without this option.

					Matthieu