Re: [patch] Add toplevel configure check for MPFR

[Paul Brook <paul at codesourcery dot com>]

On Sunday 25 July 2004 17:50, Gabriel Dos Reis wrote:
> "Joseph S. Myers" <jsm@polyomino.org.uk> writes:
> | On Sun, 25 Jul 2004, Paul Brook wrote:
> | > Gfortran currently requires GMP (GNU Multiple Precision library) for
> | > compile-time arithmetic. In order to accurately match target arithmetic
> | > we also require MPFR. MPFR is part of GMP, although it is not always
> | > enabled.
> |
> | Note also that the GMP home page <http://www.swox.com/gmp/> says that "In
> | the next major GMP release, mpfr will not be included.  Users are
> | encouraged to download mpfr directly from that project's site.".  This
> | may only be of relevance to the documentation, not the configure tests,
> | though it would complicate importing GMP into GCC (i.e. we'd need to
> | import both GMP and MPFR) if we wish to use it outside the Fortran front
> | end (for more general constant folding of built-in functions).
>
> Why would we need it all at? (apart from Gfortran is using it at the
> moment?).

There has been some discussion about using it to replace and extend the 
existing target floating point arithmetic/constant folding in gcc.

http://gcc.gnu.org/ml/gcc-patches/2003-09/msg00874.html

Paul