Re: [patch, fortran] Fix PR 33683, wrong gamma function

[Tobias Schlüter <tobias dot schlueter at physik dot uni-muenchen dot de>]

Thomas Koenig wrote:
> this fixes PR 33683 on the sytems (such as libc systems) where gamma()
> actually returns the logarithm of the gamma function.
>
> The test case risks failure on systems where there is no tgamma()
> function.  We probably need a separate tgamma function for
> these systems.  The current patch is an improvement over the old status,
> where we would return (horribly) wrong values silently.
>
> Regression-tested on i686-pc-linux-gnu, where we used to return the
> wrong values. OK for trunk?

Yes, definitely.  I think it would be a good thing to remove the gamma 
builtin from the middle-end, if it's not guaranteed to return the same 
value across platforms, but that's probably someone else's job.

Cheers,
- Tobi

> 2007-10-07  Thomas Koenig  <tkoenig@gcc.gnu.org>
>
> 	PR libfortran/33683
> 	* mathbuiltins.def (GAMMA):  Change function name to
> 	"tgamma" instad of "gamma".
>
> 2007-10-07  Thomas Koenig  <tkoenig@gcc.gnu.org>
>
> 	PR libfortran/33683
> 	* gamma_5.f90:  New test case