General host configuration fragment purge

[Zack Weinberg]

On Mon, Jan 14, 2002 at 09:23:29AM -0800, Stan Shebs wrote:

> > It'd be okay to stick some Darwin-specific annotation in [system.h].
> 
> The whole point of setting up the host define was to avoid
> cluttering up the source code.  Anyway, there is no builtin macro
> that distinguishes Apple compilers reliably, which is why the
> info has to come from config.gcc.

*sigh* Well, if that's what you have to do, then that's what you have
to do.

> Apple's compiler *is* 2.95, and identifies itself as such, so
> that heuristic won't work either.
> 
> I don't understand why you're so keen to replace something that
> works with something that's broken...  I'd fix it myself directly,
> but I want to get agreement here, so that my work isn't repeatedly
> undone.

It's the x-host fragments that I want to get rid of.  I changed the
xm_defines annotation because it seemed appropriate to put all the
Apple compiler workarounds in one place.  xm_defines isn't going to go
away, and I promise not to change it again once you put it back.

You still haven't said whether it suffices to put -no-cpp-precomp in
stage1_cflags, or whether we need to put the x-host fragment back too.

zw