Re: header-munging

["Randall C. Gellens" <[e-mail]>]

On 13 Aug 1996 13:19:36 +0100, Tim Goodwin <[e-mail]> wrote:

>> It might be a good idea to extend SUBMIT so that the client can say
>> either SUBMIT or RELAY.  If the client says RELAY, that tells the MTA
>> not to muck with the message at all (except of course to add a
>> received header).
>
> So now every sender-SMTP that wants its messages relayed intact must:
> i) use ESMTP; ii) check the supported extensions to see if SUBMIT is
> among them; and iii) if it is, send RELAY with the MAIL command.
>
> This seems like a significant step in the wrong direction.

The current situation is that SMTP, which was designed for relaying, is
widely used for submission as well.  Some MTAs treat all messages as
submissions, and munge them, which is a bad thing when the message is
really being relayed.  Other MTAs treat all messages as relays.

It is far too late to argue that SMTP should not be used for submission.
All we can do, as I see it, is either (a) accept the status quo and do
nothing, or (b) try and improve the situation in an efficient way.

The SUBMIT proposal allows the client SMTP to inform the server if the
message is a submission.  If RELAY is added, the client can also inform
the server if it is being relayed.  This would permit the server to take
submit or relay action without needing to examine the message at all,
which I think is a good thing, as during relay an MTA should not care
about the text (except of course for adding a received header).

For clients which choose not to identify the mode of SMTP being used,
the MTA is free to either treat the message as a relay in all cases,
or to examine the text and decide if it is a submission or a relay.

Thus, clients which do not care what the MTA does are free to continue
using SMTP just as they now do.

Today, clients that do not want the server to muck with their messages
are powerless.  There is no way for them to inform the server that the
message is not a submission and should be relayed intact, per the
original intent and letter of the standard.  One could argue that an
MTA should never alter the message text (aside from adding a received
header of course), but the fact remains that SMTP is used for submission
and therefore some messages need to be altered in various ways, and some
MTAs are going to do that.

Clients are not forced to identify the message as submission or relay,
and MTAs are not forced to examine messages.  But the extension allows
an efficient way for clients to inform servers of the SMTP mode being
used.  How is that a step in the wrong direction?


--
|Randall Gellens             |                    [e-mail]|
|[phone]              | fax [phone] can add ,,,,,,,,6350|
|Opinions are personal;  facts are suspect;   I speak only for myself|