GCC build failed with your patch on 2001-01-09T11:35:00Z.

[Alexandre Oliva]

On Jan  9, 2001, Richard Henderson <rth@redhat.com> wrote:

> On Tue, Jan 09, 2001 at 05:22:05PM -0200, Alexandre Oliva wrote:
>> And what is the correct definition maximum, natural host wide int?

> Don't be obtuse.  You know very well what he means.

I'm afraid I don't.  For example, we use `long' by default, even when
it's wider than `int', which I suppose would be the natural host wide
int.

Oh!  I think I get it!  It's the widest integer type that is natively
supported by the host, without having to go through auxiliary
functions to perform adds, subtracts, etc.  Right?

> But it is also not acceptable to slow down a native 32-bit ppc-linux 
> compiler by forcing it to do multi-word arithmetic when it is not
> necessary.

IMO, a 32-bit-only port should not have MAX_LONG_TYPE_SIZE set to 64.
This would fix the native 32-bit ppc-linux problem.

As for the fact that the 64-bit AIX port works without a 64-bit
HOST_WIDE_INT, I suggest one to try gcc.c-torture/execute/991014-1.c
and a few other tests whose names I don't recall at the moment, using
a GCC hosted on a 32-bit machine.

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me