floting-point => integral conversions and rounding

[Doug Gwyn (ACISD/MCSB)]

I'm just now catching up on my holiday mail backlog..
I think the other responses pretty much have covered this issue.
I note that IEEE 754 postdates Fortran and C specification for
truncation (*not* "rounding") when a f.p. value is assigned to
an integral object.  There are at least three sensible ways to
specify what should happen upon such a conversion, and each
specification seems to have picked the one that best suited its
clientele.  The language binding for an IEEE 754 feature need
not take any particular form (such as cast-like conversion), so
long as it is available in *some* form.