[JDEV] Message timestamps

David Waite mass at ufl.edu
Mon Oct 4 21:07:05 CDT 1999


> -----Original Message-----
> From: jdev-admin at jabber.org [mailto:jdev-admin at jabber.org]On Behalf Of
> Eric Bowersox

> Visual Basic?  (Not that *I'm* going to write any VB code, mind you, but I
> needed a counterexample...VB's date format is a floating-point number of
> days since January 1, 1900.  Dates before that time are indicated by
> negative numbers.  Time within a day is indicated by using the fractional
> part of the value.)

Yup, that is the windows DATE structure, just a double with the fractional
part indicating time of day, whole part indicating the date. Not defined to
be GMT, which has caused me quite a bit of headache in the past..

> The reason why I suggested ISO 8601, aside from the fact that I
> saw it used
> for this purpose in the XML-RPC specification, is that it strikes
> a balance
> between human readability and machine parsability.  Since all the
> fields are
> fixed-width, times are always GMT, and there are no issues of
> whitespace or
> alphabetic month/day-of-week names to consider, the format can be easily
> parsed (can anyone write code for parsing general RFC 822 dates
> as tight as
> code that would parse an ISO 8601 date?); however, it's also
> fairly easy for
> a human to look at it and see what it means, unlike a raw time_t value.

Yup, I think it is the best idea so far. I personally like raw time_t (as a
64 bit value), but only because I am an optimization freak.

*grin* perhaps if it is textual XML, we use ISO 8601, and binary XML we use
time_t?

-David Waite (who _WAS_ joking about Y10k, honest)





More information about the JDev mailing list