This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: question on double to int conversion
- From: Andrew Haley <aph at redhat dot com>
- To: Richard Henderson <rth at redhat dot com>
- Cc: Jaeyoun Chung <jay at kldp dot org>, gcc at gcc dot gnu dot org
- Date: Sat, 22 May 2004 11:38:15 +0100
- Subject: Re: question on double to int conversion
- References: <871xletug0.fsf@jaylin.tmax.co.kr><20040521204332.GA11188@redhat.com>
Richard Henderson writes:
> On Fri, May 21, 2004 at 10:55:43AM +0900, Jaeyoun Chung wrote:
> > To be ieee 754 compliant, '%x' output should be 7fffffff and 80000000,
>
> Who says?
> ISO C says float->int conversion overflow is undefined.
Section 5.4 of IEEE 754, "Conversion Between Floating-Point and
Integer Formats", refers to Section 7.1, which says you get an Invalid
Operation exception for this case.
Section 7.3 says "when no trap occurs ... round towards zero carries
all overflows to the format's largest number with the sign of the
intermediate result".
However, ISO C, Section F.4 says of floating-point to integer overflow,
"the resulting value is unspecified."
Andrew.