🚀 go-pugleaf

In article <1992Sep26.153421.7698@cpio1.fr.mugnet.org> bernard@cpio1.fr.mugnet.org (Bernard Fouche) writes:
>[Making |time_t| a |double| could lose accuracy]

True, but if it can exactly represent all the values that a |long int| can,
then it's no worse than what we have now, right?

>[The solution is to use a Bignum library]

Unfortunately, in order to allow for implementations that have no clock, the
Standard requires that |time()| be able to return the constant |(time_t)-1|
to indicate failure.  Hence, |time_t| must be an arithmetic type.

Looks like the original implementation (when |time()| returned its value
*only* through its argument, because |long int| wasn't supported) was on the
right track after all.

Karl W. Z. Heuer (karl@ima.isc.com or uunet!ima!karl), The Walking Lint