> I suggest that instead of two string arguments, you use a single string
> argument containing an ISO 8601 style date with optional time:
>
> yyyy-mm-dd
> yyyy-mm-ddThh:mm:ss.ssss
>
> Note that T is a literal uppercase #\T, and that the fractional second
> is optional.  Most languages have libraries to parse this format.

That may be a good idea.

I put the date and time in different strings so it would be easy to get
only the date (which I believe people generally want).

You raise a good point about library support. And since the date is
always in YYYY-MM-DD format, (substring date/time 0 10) could always get
it from a combined date/time string even with no fancy parsing.

How would this work from awk and grep|sed? When the date is one string,
it works the same way parsing as any other string-valued property (i.e.
get the first double-quoted string from the list) - no extra parsing is
necessary. If the time were in the same string, the user would have to
limit it further with something like 'head -c 10'. Note that '-c' is not
in Posix.