On Fri, 7 Nov 2003, Andrzej Filip wrote:
Ok, I misread what you were suggesting. I still really don't see a hugeNow if I get any message with a RCPT TO of "me+spam" it will be rejected if the first test is ignored, even if it shouldn't be.
Let me apply what I suggested to your example: first rule uses data not available at "RCPT TO:" stage => "envelope based" sieve script interpretation "quits" at this point doing nothing [ the script is going to be interpreted fully during LMTP delivery ]
advantage here, since the great majority of sieve scripts will be based
off of more than just header information.
It would require users to "optimize" their sieve scripts to take advantage of the suggested feature.
I disagree with you but I am not volunteering to implement cyrus part of it :) [ so my opinion is not important ]
In any case, it is unclear what could usefully be done with this, as
reject guarantees that an MDN will be sent, which SMTP error codes do not
do.
It will make remote host sending message to sendmail & cyrus pair responsible for generating the bounce message (in format beyond our control).
A discard action has a similar problem -- you still need to accept the message but later throw it out.
This is a tricky problem from sendmail perspective.
As I wrote incorporating sieve into map lookup results send back to sendmail is an option to investigate in some future version of cyrus "map protocol" server. [Just do not forget that some people think it makes some sense :) ]
-- Andrzej [pl>en: Andrew] Adam Filip http://www.polbox.com/a/anfi/ [EMAIL PROTECTED] [EMAIL PROTECTED] [former: [EMAIL PROTECTED]
