>>>>> "RSW" == Randall S Winchester <rsw@Glue.umd.edu> writes:
RSW> Humm, Not sure what the high priority class is, but I would just as
RSW> soon run them at the same "Precedence=Bulk" and skip the extra search
RSW> (unless it is a per list flag that defaults to off).
A while ago someone wished to be able to set some users to receive the mail
first, before all other users. Hence the 'HIGH' class is processed and
delivered to first, before the 'EACH' class. Trivial to do, but probably
not a good idea from a performance standpoint. It will be zero cost with
some kinds of optimization, though.
RSW> This is interesting. So you just have TLB split the envelopes up with
RSW> an "assume ordered" flag, or just have mj2 do this.
This is one of the things that TLB supported (supports?); it also comes
with a tool called 'mungelist' that sorts a list. (It will strip comments
from it, too.)
Don't confuse TLB with the delivery modules in Mj2; they bear only a
passing resemblance. Think of the stuff in Mj2 a streamlined, second
generation of TLB.
RSW> I guess I really agree. I had bulk mailer sort down to the n-1 part of
RSW> the domain when the number of parts was greater then two. This helped
RSW> in a similar fashion as things tended to get grouped by domains and
RSW> not hosts. This works better with an MX world.
Well, what you really want is to just lop off the user part (s/^.*?\@//),
reverse the rest, and sort normally. This doesn't add too much additional
cost.
RSW> Note where MIT ends up. You can see though it would be better still to
RSW> catch wam.umd.edu && po1.wam.umd.edu. If the sorting is done before
RSW> delivery then it makes sence to spend more time doing it right...
Reverse domain sorting will place these two together. I think I'm missing
something, because I'm sure you know about sorting by reverse domain, but
then I don't get your part about sorting down to the n-1 part of the
domain.
[sendmail hacks deleted; I'm too tired now and will read tomorrow]
- J<
References:
|
|
