Brian Hurt <bhurt@janestcapital.com> writes:
> Note that taking a signal on an FP exception is a horribly expensive
> proposition- we're talking about hundreds or thousands of clock cycles
> here. But it's probably worthwhile vr.s the cost of testing every
> floating point result, as generally FP exceptions will be rare (probably
> even more rare in database work than in general). So it's probably
> worthwhile.
I think we should probably stay away from relying on signals for this
on portability grounds. The cost of checking the results is small, and
will get smaller if we eliminate or simplify CheckFloat[48]Val as is
being discussed here.
regards, tom lane