5.2.1 fails 5 tests: SGI Indigo-2 IRIX 6.2 gcc -O2

[Quinn Dunkan]

On Fri, 21 Jul 2000 21:55:04 -0600, Andrew Dalke <dalke at acm.org> wrote:
>
>Eben Oldmixon wrote:
>>I particularly like the non-overflowing of maxint+1, although
>>'+-1' (!) is a close second.
>
>
>I recall Jeff Chang mentioned that Python on IRIX when compiled
>with gcc had sys.maxint give a value of -1.  The solution was either
>to upgrade (but upgrade the OS? gcc?) or a compiler flag.  Given
>the poor history coverage in DejaNews^H^H^H^H, I can't find his
>messages on the subject.

When I installed gcc 2.8.1 from freeware.sgi.com, it defined __LONG_MAX__ to
be a 64 bit long (which was wrong) unless you used -mno-long64 for gcc -E.  I
solved that problem by taking *long_max_spec out of the specs file.

Hope that helps...