Date: Sun, 21 Jan 1996 10:22:46 +0100 (MET)
From: Jacob Palme <jpalme@dsv.su.se>
To: Keith Moore <moore@cs.utk.edu>
cc: ietf-drums <drums@cs.utk.edu>, moore@cs.utk.edu
Subject: Re: The conservative and liberal commandment 
In-Reply-To: <199601050601.BAA02630@wilma.cs.utk.edu>
Message-ID: <Pine.SUN.3.91.960121101424.28127F-100000@ester.dsv.su.se>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII

On Fri, 5 Jan 1996, Keith Moore wrote:

> Not to disagree with your advice, but I note that all of the 
> above "fixes" are necessary not because the standards
> are poorly specified, and not usually because native Internet 
> mail systems fail to meet the standards, but ...

That depends on what you think a standards document should
contain.

If your opinion is that the document is well specified (A)
if it specifies a standard which will work if everyone implements
it faithfully, then you are right.

If you think a standards document should specify (B) the protocol
which implementors should use in order to get well-working
systems, then the standard should specify two protocols,
one for what you generate (more limited) and another for
what you accept (wider).

(This message is not really in disagreement with what you
write, since you do write "Not to disagree with your advice".)

Formally, if a standard follows the first of the principle
A above, no statement "be conservative in what you send out
and liberal in what you accept" should occur in the standard,
since that statement is clearly according to principle B
and not principle A. Thus, if we accept that statement,
we should also specify in detail what we mean by "liberal"
and "conservative".

It is not at all obvious for all implementors that "be conservative"
means for example to avoid all quoted characters in addresses,
but does not mean to avoid using the character period (".")
in addresses! When we develop standards, we do look at
robustness and implementability, not only at something
which formally works if everyone follows the standard 100 %.

------------------------------------------------------------------------
Jacob Palme <jpalme@dsv.su.se> (Stockholm University and KTH)
for more info see URL: http://www.dsv.su.se/~jpalme