On Tue, 31 May 2005 09:46:57 PDT, David MacQuigg said:
I would accept as proof two messages, one sent directly to the receiver,
the other through the forwarder, both identical going in.
a) This tells you something between zero and diddly about any of the headers
that were acquired along the route through the forwarder - and we could *easily*
be talking 20-25 of them if the forwarder is a mailing list.
b) The fact that your test case shows up without anything reordered doesn't
prove that there's no misbehavior in other cases (or did you actually *test*
cases like "more than 900 local recipients *and* more than 100 recipients at
another host that happens to have us as a secondary MX forwarder" - had *that*
give some software some indigestion. Worst part was that it actually manifested
as a doubly-linked file in the message store - seems when a %3d rolled over in
the same message, the same filename got created for the 3rd and 1003rd
Description: PGP signature