Date: 1 Jul 1996 16:22:36 -0000
Message-ID: <19960701162236.1811.qmail@koobera.math.uic.edu>
From: djb@koobera.math.uic.edu (D. J. Bernstein)
To: drums@cs.utk.edu
Subject: Re: Multiple 'To:" fields (was: Re: Microsoft's Messaging Contacts)

> Since this would invalidate existing conforming installations (unless you
> consider "readers that care about..." to be a loophole, I'd be a lot happier
> with a "SHOULD".  

The two of you are trying to go in completely opposite directions.

Here's the pro-writer direction:

   We're going to allow multiple To lines.

   Readers MUST interpret multiple To lines the same way as a single
   To line with all the addresses.
   
   Writers SHOULD NOT generate multiple To lines, since they were not
   assigned a meaning in RFC 822, and current readers treat them in
   several different ways; but this rule will change eventually.

Here's the pro-reader direction:

   We're not going to allow multiple To lines.
   
   Writers MUST NOT generate multiple To lines. They were not assigned a
   meaning in RFC 822, and their function can be served by a single To
   line with all the addresses.

There's no point in mixing these two directions. That would just mean 
the worst of both worlds for both sides.

I'd like to see more discussion of the practical impact of these two
positions. Which existing 822 parsers have trouble with multiple To
lines, and how hard would they be to change? Which existing generators
create multiple To lines, and how hard would they be to change?

---Dan