Re: MSGSIZE

[John C Klensin <KLENSIN@xxxxxxxxxxxxxxx>]

>If the message exceeds the permanent message size limitation for some
>recipients, but not for others, the receiver-SMTP should accept the message,
>but it must then issue a nondelivery report for those recipients for whom it
>could not deliver the message.

Keith,
  Excuse me for being either dense or just stuck in my thinking, but is
there some reason to not make this announcement early enough in SMTP
processing that recipient rejection can occur interactively, while the
connection is still open?
  The "non-delivery report" approach requires that mail messages go back
and, unless one adopts non-return of content, those messages in this
case are going to be pretty big, especially as compared to the
capacities of the hosts and links that are most likely to benefit from
message size announcement.
  While I recognize that we are doing more and more things in the style
of "accept message, send back non-delivery report later", and I wouldn't
dream of trying to roll back that tide, it seems to me to be unwise--and
actually, a quite dramatic change to the general tone of SMTP--to make
extensions that prohibit the more interactive way of behaving.
   --john