[Mailman-Developers] Re: way to minimize IO load with MTA supported VERP

Peter W peterw@usa.net
Thu, 6 Dec 2001 17:35:24 -0500


On Thu, Dec 06, 2001 at 04:59:50PM -0500, tneff@bigfoot.com wrote:

> If one was willing to extend SMTP again, it would be theoretically possible 
> to send one copy of a message to aol.com for 100 recipients with a form of 
> VERP, using an extended syntax like
> 
> 	MAIL FROM:mylist-owner@mydomain.com
> 	250 mylist-owner@mydomain.com... Sender ok
> 	RCPT TO:johnsmith@aol.com EPFX:johnsmith-aol.com-
> 	250 johnsmith@aol.com... Recipient ok

> and the receiving MTA would prepend the EPFX value to the envelope sender 
> for each recipient.

I like the basic idea a lot, but that doesn't look very backwards 
compatible, though. Why not something like
 	MAIL FROM: mylist-owner@mydomain.com
 	250 mylist-owner@mydomain.com... Sender ok
 	RCPT TO: johnsmith@aol.com 
 	250 johnsmith@aol.com... Recipient ok
        OVRD "MAIL FROM: johnsmith-aol.com-mylist-owner@mydomain.com"
        250 per-recipient override of "MAIL" ok
        DATA

i.e., a new SMTP command, not a change in existing rules. The sending MTA,
if it got an error in response to OVRD, could flag that SMTP connection as
being OVRD-incompatible, issue a NOOP, and procede to either give on message
efficiently w/o VERP, or several messsages with VERP using the old-style, 
bandwidth-hogging (current) technique. The NOOP-backout technique appears to 
be fully compatible with the SMTP RFC. Issuing a NOOP means the sender would 
not have to be concerened with whether the recipient MTA does not like 
messages sent with unrecognized control commands. 100% backwards compatible.

If VERP is the only real case where per-recipient overrides make sense, then 
it might make more sense to make a simple VERP command, e.g.
        VERP johnsmith-aol.com-mylist-owner@mydomain.com

-Peter

-- 
I am what I am 'cause I ain't what I used to be. - S Bruton & J Fleming