Re: header-munging

On Tue, 06 Aug 1996 14:34:28 -0700, Michael D'Errico <michael.derrico <at> 
software.com> wrote:

> > > Comments on draft-gellens-smtp-submit-00.txt are appreciated.
> >
> > Instead of an SMTP extension, why not use a different TCP port?
>
> I won't comment on that, but....
>
> > [....] If the client can't produce [....] a correctly formatted
> > message, and the server won't accept incorrectly formatted
> > messages, the situation is hopeless.
>
> ....this is exactly why I don't like the SUBMIT idea either.

But this is rarely the case.  More commonly, the client produces a
message which is often good enough, but should be better (for example,
it lacks a Message-ID, and the Date is either missing or incomplete),
and the server either passes everything through as is, or attempts to
fix messages.  If it attempts to fix messages, it might do this for
every message, or for messages it guesses are being submitted.

The SUBMIT extension makes it very clear that the client is giving the
server permission to attempt any fix-ups that might be needed.  The
draft also lists tests which can be used for the server to guess at
message submission if SUBMIT isn't used.

Examples of clients which fail to produce Message-ID and/or proper Date
headers are everywhere, as are examples of servers which munge messages
(often all messages) in an attempt to fix them.

So the SUBMIT extension takes real-world situations and makes them more
deterministic.

--
|Randall Gellens             |                    randy <at> mv.unisys.com|
|(714) 380-6350              | fax (714)597-8053 can add ,,,,,,,,6350|
|Opinions are personal;  facts are suspect;   I speak only for myself|