Matthew Elvey (FM) writes:
http://www.elvey.com/it/sieve/draft-elvey-refuse-sieve-01h.txt
I suggest adding a sentence saying that SMTP refusal may happen at the
RCPT TO or DATA stages, depending on what's convenient to the
implementation. (I suppose it might also happen at the MAIL FROM stage,
if a systemwide sieve does the refusal.)
The 550 result code is a fine thing, but shouldn't you also specify an
extended result code (1893/2034)? 5.7.1, perhaps?
Arnt