This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: large_real_kind_1.f90 & huge()
- From: François-Xavier Coudert <fxcoudert at gmail dot com>
- To: Steve Ellcey <sje at cup dot hp dot com>
- Cc: fortran at gcc dot gnu dot org
- Date: Fri, 9 Sep 2005 08:15:10 +0200
- Subject: Re: large_real_kind_1.f90 & huge()
- Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=firknA2ItMcr8rzd3bDUi3Stw1jjrJDmMGM3bZXxn8mkhuuxyHArh301FeOlbPfdX0J/yBtRierPz4Xo5QoqmMiI5cD+YEAXrjr/BGQgKuooeQbrYt3+JYlOiaPQ5/9W6i/+UUTHdnCaMBB61qbLeKPsVMQJkjfNTUSxBP051T8=
- References: <200509081920.44889.paul@codesourcery.com> <200509082302.QAA19183@hpsje.cup.hp.com>
- Reply-to: fxcoudert at gmail dot com
Well, first, you should file a bug-report to HP-UX. isfinite() is
supposed (by C99) to be a macro that can work on long doubles.
Unfortunately, HP-UX is not the only system to ignore this
requirement: I wrote this TODO because I experienced the same
behaviour with Solaris 9 long doubles.
So, I guess we should have a test in acinclude.m4 that tests if
isfinite works (let's say, a test that calls isfinite() on a
LDBL_MAX). Then, in libgfortran.h:
#if (!defined(isfinite) || defined(__CYGWIN__))
#undef isfinite
#if defined(fpclassify)
#define isfinite(x) (fpclassify(x) != FP_NAN && fpclassify(x) != FP_INFINITE)
#else
#define isfinite(x) ((x) - (x) == 0)
#endif
#endif /* !defined(isfinite) */
use the poor man's isfinite() depending on the result of this test
(say HAVE_BROKEN_ISFINITE).
I don't have time now to work on this patch, but I will gladly review it.
FX