Extensions to SMTP

Julian Onions | 19 Aug 1992 09:43

Re: last call on smtp-8bittransport


> Hi Neil,
> 
> (Per repeated requests, I'm moving this to the smtp list.)
> 
> >   SIZE provides worthless information.  It specifies the message size.
> > But it is defined to specify it as an approximation.  What is an
> > approximation?  Is a 1000% error permitted?  I did suggest earlier, but
> > to no avail, that the data to SIZE include either
> > 
> > 	estimated size		error bound
> > or
> > 	lower size bound	upper size bound
> 
> SIZE is an upper bound, so it equals estimated size + error.  In addition,
> SIZE is given in kilobytes.  It's hard (for me) to imagine a case where
> the addition of headers will make a meaningful difference.  The case of
> a small message with an encyclopedia of headers doesn't really apply until
> their total exceeds 64K.  The client should have a pretty good idea of a
> reasonable upper bound on the message (including headers it adds).  The
> server should treat the estimate as correct.

Just a comment on this - something that I've found from implementing
PP. You can be quite easily fooled by the actual size of the message
and the envelope. One of the X.400 conformance tests is to see if an
implementation can handle 32767 recipients (I might say PP can given
sufficient VM - it takes a while though!). However this testing broke a
number of assumptions. If you have a 1K message with 32767 recipients
then a back-of-the-envelope calculation shows that you whilst the DATA
size is 1K (no one is going to type 32767 addresses into a To line -

(Continue reading)

headers

John C Klensin | 19 Aug 1992 12:40

Re: last call on smtp-8bittransport

Julian,
   Your example is quite interesting.  It is also a pretty interesting
argument for certain designs for some designs of mail sending and receiving
relative to others, one that a few people in the ietf-remmail discussion
should probably contemplate as they think about remote-system sending 
vs sending to a local server that does the network sending.
   But...
     SIZE applies, unless I forgot what I wrote, to message size, not
envelope size.  No help here.
     If you intend that your sender take London-residents and expand it
so that each recipient sees all of the others (the only place that 
message size, not envelope size, becomes relevant), your recipients are
going to be *very* unhappy, as is your network.
     And, if I recall, the minimum value for imposing "max number of
recipients" in RFC1123 and/or RFC821 is a lot lower than the 32K
of X.400.  If you exceed the listed limit by an order of magnitude or
more, you will break, or be rejected by, a lot of systems and make yourself
real unpopular.
       john