> i'd favor the former, as well.  sig_atomic_t isn't necessarily an
> int... (in fact, in general, it seems that it should be a "long", at
> least on the present and near-future systems that we will be
> supporting).

Hmmm.

Can a signal be raised between the time a [34]86 SX processor writes
the first and second halves of a 32 bit int?  If so, sig_atomic_t
would have to be a "short".

	--jtc