Re: std_limits.h os_defines.h

[David Edelsohn <dje at watson dot ibm dot com>]

>>>>> Benjamin Kosnik writes:

Ben> I think this patch is pretty weak. Some of the targets use LONG_MAX, some 
Ben> use stuff like __alpha, etc. I don't see any reason to not use LONG_MAX, 
Ben> as was discussed.

	Use of LONG_MAX discussed where?  Gaby appeared not to like
LONG_MAX because of the dependency on system header files (limits.h,
float.h).

	I checked in my patch because Mark asked me to in private email:

David>       Attached are the definitions that I have been able to deduce so
David> far.

Mark> Good; go ahead and check them in on both branch and mainline.  (They
Mark> cannot possibly hurt anything, and that way we'll be ready for the
Mark> new std_limits.h.)

	I thought that you and Mark and Gaby already were in agreement to
follow this approach of a generic std_limits.h using macros from
os_defines.h.

	All of this should have been settled months ago, before GCC 3.0
was released.  V3 should have been developed with a portable, robust
design for cross-compilation and multilibs from the very beginning.
Constructing std_limits.h during configure or build is a fatal flaw.

	Please direct your comments about this current design to Mark and
Gaby and leave me out of it.

Thanks, David