Possible explaination why large_real_kind_form_io_2.f90 is failing on PPC/OSX

Dominique Dhumieres dominiq@lps.ens.fr
Mon Jan 15 06:24:00 GMT 2007


> I suspect that the 32nd digit (if printed) is greater that 5.

The result is:

1.00000000000000000000000000000002

so the last digit is non zero, but less than 5.

> Oh my that is ugly.

My goal was to have a simple way to build the weekly snapshot
with my knowledge of *nix.  And it works more or less (when the
snapshot is too broken, I just skip it, if there is a simple
patch, I let my machine compile for a other night).

Dominique



More information about the Fortran mailing list