Hey Russell!

Feb 12 17:22 05, Russell Tiedt wrote to Maurice Kinal:

 RT> As long as it does not generate msgs. faster than it creates MSGID's 
 RT> there 
 RT> should not be a problem, even if it wastes/skips a few here and 
 RT> there.

Personally I'd prefer it actually produced a meaningful ID that could be
employed further on down the line.  For dupechecking?  I doubt it is good
for that.  Too easy to give an actual duped message a whole new ID and it
will get past everyone.  You can count on that happening.  However the
reverse, a new message with a duped ID, is probably more common then any
one system is aware of.  For example I get non-dupe-dupes in my dupe file
from Janis that don't ever show up at all, nevermind in the dupe area, on
the node's system which has a different uplink.  I figure that is because
wherever these messages are originating aren't passing through too many
systems before Janis gets them and thus I recieve them from there, whereas
on the node there is at least one system between myself and the originating
system that is filtering them out as dupes based on the phoney-baloney,
flawed MSGID dupe "standard".  Given that then it seems to me
that I am only seeing the tip of the iceberg and that each so-called dupe I
do see probably represents a small fraction of what is really going on. 
Does that make sense?

 RT> I on occasion still catch a lot of dupes especially from Bob Seaborn, 
 RT> I would 
 RT> have expected his system to filter out all dupes but it does not seem 
 RT> so, must 
 RT> still check that sometime.

How many of those are REAL dupes?  Also, if they are REAL dupes check the
originating system, etc. and see if there is any commonality between all
the dupes. Also, also, how many REAL dupes are making it past your dupe
checking and what is the commonality of those?  I've noted some interesting
things following this ideology.  Here at Kumalockasun we started calling
stuff like that "pulling a Swindle", heh, heh.  Care to guess
why?

Life is good,
Maurice

--- Msged/LNX 6.1.2