Re: Why float is called as 'float', not 'real'?

Brooks Moses wrote:

(snip on gfortran and multiple precision libraries)

FWIW, these are used for evaluating initialization expressions, since gfortran supports cross-compiling, and the host hardware may not have as much precision available as the target hardware has. For simplicity (and to make it more likely that bugs will be noticed), initialization expressions are evaluated with a GMP/MPFR-based representation using the same code regardless of whether one's actually doing a cross-compile or not.

So it only does a little more than double precision, and then rounds
as appropriate?

However, the compiled code produced by gfortran doesn't require GMP or MPFR in order to run.

So the old problem of compile time and run time evaluation being
different is likely to occur?

-- glen

.