[Ecls-list] Accurately detecting FP exceptions

[Juan Jose Garcia Ripoll]

On Wed, 2005-10-26 at 11:21 +0200, Juan Jose Garcia Ripoll wrote:
> On Tue, 2005-10-25 at 16:39 +0200, Goffioul Michael wrote:
> > Aaah, OK, so the problem is something else. As stated in the MSDN documentation,
> > the exception is actually thrown on the next FP unit access, it is not thrown
> > immediately.
> 
> I have rewritten the floating-point number boxing routines
> (make_shortfloat and make_longfloat) so that on the x86 (Intel-32 bit)
> platforms additional FWAIT instructions are inserted. Since these
> routines are called after almost every floating point operation in the
> standard library, the exceptions will be detected and signaled at the
> right point.

A minor fix was left: RANDOM did not use make_{short/long}float() and
hence the floating point exceptions were not detected. I have fixed this
and now the PRINT*RANDOM tests are passed.

Juanjo