From: Jens Maurer (Jens.Maurer_at_[hidden])
Date: 2001-05-22 16:31:15
Beman Dawes wrote:
>
> At 09:18 PM 5/21/2001, Greg Chicares wrote:
>
> > ... tests of Borland and GNU fp results with various switch settings ...
>
> So is the conclusion that limits_test.cpp is a standards conforming
> program, and that Borland, Metrowerks, and Microsoft compilers all have
> bugs in their FP processing?
I've refactored limits_test.cpp to check for numeric_limits<>::has_XXX
flags before actually assuming that e.g. QNaNs exist.
Also, having "is_iec559 == true" requires infinity and QNaN support,
as per the standard.
I now believe that my program doesn't have any assumptions not
covered by the C++ standard.
Jens Maurer