[Top] [All Lists]

Re: per user post-data rejects, Processing after the end of DATA

2010-08-14 09:56:39

ned+ietf-smtp(_at_)mrochek(_dot_)com wrote:

So if they've decided it's spam, what's the problem with throwing it away?

The same problem as always: False positives. While I think the various
admonitions we have in various documents that try and discourage silently dropping messages are unrealistic, the fact remains that silent draps create reliability issues. Rejects address this issue and are preferable, as long as they don't create additional issues.



As a public note for the unaware, the new RFC 5321 for the first time ever, has "cracked opened the door" with new SMTP semantics allowing or provides a rationalization for discarding of mail under "reasonable" circumstances.

IMV, having new official RFC mail protocols specifically designed for this purpose is rational and reasonable, such as RFC 5617 (ASDP) with its DKIM=DISCARDABLE policy. As long as the ADSP domain publicly exposes this self-asserted policy, I don't think it contributes to false positives discarding of accepted but zero false positive proof for ADSP violating mail.

However, one implementation question is when a DATA level ADSP based (55x) rejection can also serve the same purpose for ADSP domains.

Currently this only seems to be a problem for DKIM resigners ignorant of errant or spoofed ADSP protected mail submissions. Murray's Mail List Manager (MLM) draft proposal hopes to provide, IMV, "sound protocol consistency" guidelines in this area for DKIM+ADSP compliant List Server and SMTP receiver developers to consider.


Hector Santos

<Prev in Thread] Current Thread [Next in Thread>