BJ>> We will need to address this long term.....  Using the
 BJ>> Unix time standard would allow finear granularity,

 WM> Only by half. Surely, if a two-second limit is a problem, a
 WM> one-second limit will still be a problem?

all the more reason to completely forego the use of a timestamp as a
duplicate detection mechanism... think about a multinode system where more
than one user just happens to post a message at the same time...

even when combined with CRCs, there is still trouble to be had... why? CRCs
are not unlimited... there is a very definite set of values and thus, there
is the possibility that a false duplicate will be found...

the easiest option is to use MSGID and simply use a serial number that
starts at x and increments x+1 for each message posted... the only problem
with this method is determining what digit to start x at because it is
possible that a system may need to be reinited and thus start with a new
counter...

of course, we also have to look to the buggy and flawed software that
regurgitates messages with modified headers and control lines... those that
strip, replace or add MSGID will mess up thie method but many would rather
read a true duplicate a second or third time than risk loosing a message
and having it never get read at all...

)\/(ark