mktime() failure vs 31-DEC-1969 23:59:59 GMT

[tomp at zk3.dec.com]

> Another possibility is to save away the original tm structure; if mktime
> returns -1, bump the tm_sec value by 1, call mktime again, and see if it
> returns zero.

This still doesn't tell me whether the -1 return from mktime() meant
that the time was not representable.  It does tell me whether it is
representable one second later.  If a particular implementation of
mktime() doesn't support times before the Epoch (and that seems okay
as far as the standards go), then a return of 0 resulting from bumping
the tm_sec up by 1 would lead me to the wrong conclusion concerning my
original call.

An errno would really be helpful here.  It's a shame the standards don't
define one.

- Tom