3.0 PATCH RFC: Major Solaris 2 configuration cleanup

Rainer Orth ro@TechFak.Uni-Bielefeld.DE
Tue May 29 12:24:00 GMT 2001


Alexandre Oliva writes:

> > If this is in scope for 3.0, I'll continue working on this now
> 
> I'm afraid it isn't.  It certainly doesn't fix any outrageous bugs,
> and we shouldn't take the risk of introducing new ones with only two
> weekly snapshots remaining before 3.0 final.

I'd expected this, but thought to better ask :-)  And you're certainly
right about the risk: initial testing revealed several stupid errors, and
it's not a good idea to try to hash this out under extreme pressure.

> The second is better, because it lets the Makefiles and the developers
> see all dependencies (you don't have to go opening long chains of
> header files to find out whether a certain piece of configuration is
> in use, or why the heck it isn't).

Ok, I'll follow this route.  There's only one place where it might be
necessary to still have a mixture: sparc/sol2-sld-64.h defines
SPARC_BI_ARCH depending on AS_SPARC64_FLAG, and only then includes
sparc/sol2.h.  All other sol2 config files include some other file first
thing, overriding part of it in their bodies.

> > * i386/sol2gas.h and GAS_REJECTS_MINUS_S are gone.  This file supports a
> >   long-obsolete version of gas (2.9.5.0.12).  Since anyone can easily use
> >   the vendor as or upgrade to a newer binutils version (which might well be
> >   required for proper DWARF-2/EH support anyway), I don't think it's useful
> >   keeping this cruft.  If this should be considered desirable nonetheless,
> >   I suggest turning this into an autoconf test instead.
> 
> Let's keep it for 3.0 (fro the reasons above).  I agree it can be
> removed from 3.1.

Ok, fine.

	Rainer

-----------------------------------------------------------------------------
Rainer Orth, Faculty of Technology, Bielefeld University

Email: ro@TechFak.Uni-Bielefeld.DE



More information about the Gcc-patches mailing list