I don't like the idea of reducing  the size of the cyclic cache below the
daily threshold because  it can then no longer  guarantee that duplicates
are not posted. In the case of  an occasional duplicate due to a one-time
line failure right in the small time window where the protocols cause the
message to be resent yet the  receiving site has accepted delivery, there
will  be no  problem. But  when  a mailer  goes  crazy, it  is likely  to
generate duplicates  for several  messages, such that  you will  be under
attack from  a number of "waves"  and, with network delays,  the messages
will be interleaved. If you set the cache size to, say, 3, you may end up
delivering the duplicates. I would  much rather implement something based
on a  time stamp, but again  the target date  is 1.8b since that  is when
LSVXMAIL will be rewritten in PASCAL and  I don't want to write code that
is not critically needed and that I will need to trash 2 months later.
 
  Eric