[Top] [All Lists]

Re: New Sieve extension "refuse" proposal - draft-elvey-refuse-sieve-01h

2004-02-18 08:55:56

Tim Showalter wrote:

Alexey Melnikov wrote:

I don't remember now why the Seive RFC was written to use MDNs.

I do. Reject originally (early spec, long time ago) said generate a DSN, but MUAs can't do that--conceptually, anyway. An MUA can generate a good forgery, but it was inappropriate for it to do so.

The guarantee advertised by a DSN is supposed to be "this is what happened". The guarantee advertised by an MDN is "this is what happened, except that maybe then something else happened that negated this event". My memory of the spec is that it doesn't exactly allow outright lying, but you're allowed to state "deleted" and there is no guarantee that the message wasn't immediately undeleted, for instance.

So I guess the short version is that MDNs were chosen because we could get away with so much more.

Than my question is: is changing "reject" to generate DSNs in MTAs and MDN in MUAs acceptable? I personally have no problems with this, and I suspect most users will not probably care either.


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