Re: Need for RSET?

Ned Freed <NED@innosoft.com> Sat, 10 February 1996 19:21 UTC

Message-Id: <01I11OHGJM149QUQZX@INNOSOFT.COM>
Date: Sat, 10 Feb 1996 11:02:20 -0800 (PST)
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: Ned Freed <NED@innosoft.com>
To: RANDY@mpa15ab.mv.unisys.com
Cc: ietf-smtp@list.cren.net
Subject: Re: Need for RSET?
In-Reply-To: "Your message dated Fri, 09 Feb 1996 16:44 -0800" <ECLS441201AA24@MPA15AB>
MIME-version: 1.0
Content-type: TEXT/PLAIN; CHARSET=US-ASCII
Content-transfer-encoding: 7BIT

> I was under the impression from 821 that there is no need to send a RSET
> before a MAIL FROM when sending multiple messages; the MAIL FROM resets
> state and clears buffers.

> But RFC 1854 in 2.1 says "...That is, the RSET/MAIL FROM sequence
> necessary to initiate a new message transaction...."

It depends on the prior state. A RSET is absolutely necessary if there's a
prior transaction in progress. Without it a MAIL FROM is usually seen as a
command that's out of sequence -- in fact I don't know of an implementation
that doesn't work this way. And I find nothing in RFC 821 that says you could
count on a MAIL FROM to reset a transaction that's already in progress -- if it
were the case there would be no need to have a RSET command, would there?

> Is this correct?  It is the case that RSET is not technically required,
> but there are deployed SMTP servers which insist on it and so everyone
> sends it to be safe?

Sending it is a *really* good idea, as the are some systems that require
it on second and subsequent transactions within a session, even if you're
not aborting a transaction.

However, this is really a DRUMS issue and not something the pipelining document
should be addressing, so I'll change the wording the next time the document is
reissued so as not to imply that RSET is unconditionally necessary to initiate
a transaction in all cases.

				Ned