Hi Pelle:
I've been running Fred Tydeman's C99 Test Suite available at
ftp://jump.net/pub/tybor/pkzip.zipand ran into non-compliance with the tflt2int.c which tests the following
* Assertion being tested:
*
* C99:
* F.4 Floating to integer conversion
* If the integral part of the floating value exceeds the range
* of the integer type, then the ''invalid'' floating-point exception
* is raised and the resulting value is unspecified.
*
* This test is for C99 with IEEE-754 floating-point support.
* Test conversions from floating-point types to integer types.
* Test with powers of 2.
* For numbers in range, checks for correct value and no exceptions.
* For numbers too big, checks for FE_INVALID exception.
* Also, check that no exceptions are cleared with a 2nd conversion.
*
You might want to check it out.
Robert Wishlaw