Re: Fwd: about mpfr in gcc/gfortran

[Paul dot Zimmermann at loria dot fr (Paul Zimmermann)]

   From: Paul Brook <paul@codesourcery.com>
   Date: Mon, 14 Mar 2005 15:01:08 +0000

   Please send messages to the appropriate mailing lists, rather than direct mail 
   to a random selection of developers.

   Paul

   ----------  Forwarded Message  ----------

   I've had a look at the files using MPFR in the current cvs of gcc/fortran:

   ./gcc/doc/install.texi:
      Necessary to build the Fortran frontend.  It can be downloaded from
      @uref{http://www.mpfr.org/}.  It is also included in the current GMP
      release (4.1.3) when configured with @option{--enable-mpfr}.

   Better advise to use the latest mpfr version on mpfr.org (2.1.1 now),
   since that bundled with GMP is old, and not maintained any more by the
   GMP developers.

   Also make sure that when both versions are present (that in GMP and that
   with --with-mpfr-dir), the later is used.

      "@samp{--with-mpfr-dir=gmpbuilddir}"

   Should be "--with-mpfr-dir=mpfrbuilddir".

   ./gcc/fortran/arith.c
   - mpfr now has a mpfr_get_z() function (to replace gfc_mpfr_to_mpz)

   Also, I wonder why you set the default precision to 128 bits
   (file arith.c, function gfc_arith_init_1). Isn't it possible
   to set the precision to what is exactly wanted, to avoid
   double rounding problems?

   Regards,
   Paul Zimmermann
   MPFR developer

Sorry, what is the appropriate mailing list for this kind of message?

Paul Zimmermann