gmp 4.2 fails tests on G5

[Torbjorn Granlund]

howarth at bromo.msbb.uc.edu (Jack Howarth) writes:

      That is entirely the wrong conclusion. You would have the MacOS X port crippled
  in performance by removing -fast instead of simply appending an explicit -fPIC to
  insure that PIC code was generated? That makes no sense whatsoever except to punish
  MacOS X users for a slight glitch in their compiler. Lets try to be reasonable about
  this.

Separating options for parsing, code generation, ABI selection,
optimization, is critical.  The GCC project has always been very
careful about that.

I am not going to touch an option that confuses these areas.  It is
already really difficult to maintain GMP's configure.  This Apple
braindamage has already cost me a working day but I will not allow it
to cause more problems.

This thread has now wandered off topic for the bug reports list.
Please follow up privately, or to the gmp-discuss list.

-- 
Torbjörn