Message-Id: <199601050601.BAA02630@wilma.cs.utk.edu>
From: Keith Moore <moore@cs.utk.edu>
To: Jacob Palme <jpalme@dsv.su.se>
cc: ietf-drums <drums@cs.utk.edu>, moore@cs.utk.edu
Subject: Re: The conservative and liberal commandment 
In-reply-to: Your message of "Fri, 05 Jan 1996 05:04:47 +0100."
             <Pine.SUN.3.91.960105045713.28505A-100000@ester.dsv.su.se> 
Date: Fri, 05 Jan 1996 01:00:58 -0500
Sender: moore@cs.utk.edu

> "Be liberal in what you accept and conservative in what you produce."
 
> Line length: Acccept 1000 character lines, but never send
> more than 80 character lines.
> 
> Quoting: Accept quoting in header fields, but never send
> out anything quoted except when you have to because you
> are repeating someone elses e-mail address which has quotes
> in it. Never allow your local users and mailing lists to
> have quoted characters.
> 
> Spaces: Since quoting is not allowed in what you produce,
> and since spaces must be quoted, a consequence is that
> spaces are not to be produced in addresses.
> 
> Re-writing of spaces: Replace spaces with ".", not with
> "_" or some other funny character.

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 (in almost
every case) because of limitations introduced by gateways 
between Internet mail and other mail systems.    (and it's 
worse -- there are at least a half dozen other hazards you 
need to be aware of when making up Internet mail addresses 
or mapping foreign addresses into Internet syntax).

I cannot fault the Internet mail standards (or the "be 
conservative/liberal" rule) for mail that bounces because
of bad address translation at a gateway.

Looking at it another way: we have this mess because Internet
mail is so successful that everyone else in the world has to 
adapt to it.

Of course, even if our old standards aren't to blame for 
the problem, doesn't mean that we shouldn't try to give good 
advice to implementors when rewriting them.  Advice as to
what constitutes conservative when sending and liberal 
when receiving certainly falls into this category.  The trick
is to make sure that such advice isn't confused with the
protocol specifications.

Keith