This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: 18/numeric_limits fail on solaris 2.8
- From: Paolo Carlini <pcarlini at unitus dot it>
- To: Paolo Carlini <pcarlini at unitus dot it>
- Cc: bkoz at redhat dot com, gdr at codesourcery dot com, libstdc++ at gcc dot gnu dot org
- Date: Mon, 01 Apr 2002 20:25:22 +0200
- Subject: Re: 18/numeric_limits fail on solaris 2.8
- References: <3CA8A4E8.7040205@unitus.it>
Paolo Carlini wrote:
> Anyway, in this kind of test it seems to me in order an abs(), like:
>
> VERIFY( abs(extrema_min - limits_min) < epsilon );
> VERIFY( abs(limits_min - extrema_min) < epsilon );
> VERIFY( abs(extrema_max / limits_max) < (1 + epsilon) );
> VERIFY( abs(limits_max / extrema_max) < (1 + epsilon) );
Sorry, I meant the abs only for the first two, like:
VERIFY( abs(extrema_min - limits_min) < epsilon );
VERIFY( abs(limits_min - extrema_min) < epsilon );
But the second two still seem not ok. Something like this should be needed:
VERIFY( (extrema_max / limits_max) < (1 + epsilon) &&
(extrema_max / limits_max) > (1 - epsilon));
VERIFY( (limits_max / extrema_max) < (1 + epsilon) &&
(limits_max / extrema_max) > (1 - epsilon));
P.