Tom Lane wrote:
> Alvaro Herrera <alvherre@dcc.uchile.cl> writes:
> > On Mon, Dec 22, 2003 at 06:36:32PM -0500, Bruce Momjian wrote:
> >> Seems we have to test for __alpha and __alpha_. This applied patch
> >> makes that consistent.
>
> > Won't something like the following work?
>
> > #ifdef(__alpha)
> > #define __alpha__ 1
> > #endif
>
> It seems risky to me to define macros that are in the
> reserved-for-system-use namespace. Who knows what might break in the
> system headers if we did that?
>
> I'm not convinced that all of the changes Bruce made are needed, or even
> not likely to break things themselves. What if __alpha and __alpha__
> actually indicate slightly different platforms or OS releases? For
> example, we have *no* evidence to suggest that that NOFIXADE stuff in
> main.c is needed on platforms that don't define __alpha. I would tend
> to take an "if it ain't broke don't fix it" approach, especially on
> platforms we don't have handy to test.
The problem was that certain cases tested for __alpha__ and some
__alpha --- same with __sparc.
I think I might have gotten started with the these spinlock changes
because of an __alpha fix. I also verified with an alpha guy that we
need both in most cases, if not all. I remember trying to go with
__alpha__ and finding someone couldn't compile alpha after that.
--
Bruce Momjian | http://candle.pha.pa.us
pgman@candle.pha.pa.us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073
