Back-of-the-envelope cost of extra data :-)

[Paul Eggert]

Arthur David Olson <olsona at lecserver.nci.nih.gov> writes:

> One way to reduce data file growth is to store just 5 bytes of time_t values
> rather than 8.

That sounds like progress, but can't the number of bytes for time_t (4
through 8) be in the header?  That way, we won't have a Y20K problem.

Another possibility is to use a varying-width representation, I
suppose.  This can be combined with the idea of using delta-T rather
than T, to yield fairly compact external representation.

Here's another idea: stick to the current representation unless the
input mentions time stamps past 2038.  That way, "zic" won't generate
new-format outputs for old-format inputs.