Hi Arnt,
--On Friday, February 13, 2004 4:47 PM +0100 Arnt Gulbrandsen
<arnt(_at_)gulbrandsen(_dot_)priv(_dot_)no> wrote:
|> But SIEVE has to operate on the message data - there isn't enough
|> information before the DATA command to run SIEVE, unless you extend
|> SIEVE with tests for SMTP-Envelope data (e.g. a MAIL FROM test).
|
| As I remember the draft I read, a spamtest implementation can legally
| arrive at a conclusion without considering the body, so if the script's
| first rule is a spam test, the body need not be considered. (I agree that
| the body will almost always be considered.)
|
| Anyway, I think the possibility of e.g. a future MAIL FROM test is reason
| enough to _permit_ the refusal to happen at any stage of the SMTP/LTMP
| process.
If you are going to run a script before DATA then you are going to have to
limit the set of available sieve tests to just those that make sense - in
other words you need to define different 'profiles' of the sieve syntax
that are valid for different phases of the message delivery process. I
think it would be a mistake just to say that certain tests (e.g. for header
addresses) become no-ops when run before DATA. As such I think that a
document that defines such a profile would allow the refusal on MAIL FROM
or RCPT TO without needing to define that in refuse.
--
Cyrus Daboo