To: drums@cs.utk.edu
Subject: Re: case sensitivity 
In-Reply-To: Your message of "02 Jan 1997 08:12:40 -0000."
             <19970102081240.15601.qmail@koobera.math.uic.edu> 
Date: Thu, 02 Jan 1997 23:35:16 +1100
Message-Id: <8951.852208516@munnari.OZ.AU>
From: Robert Elz <kre@munnari.OZ.AU>

    Date:        2 Jan 1997 08:12:40 -0000
    From:        "D. J. Bernstein" <djb@koobera.math.uic.edu>
    Message-ID:  <19970102081240.15601.qmail@koobera.math.uic.edu>

    The record shows that one implementor generated a mixed-case Mail from.
    After he saw the damage he was causing, he switched to MAIL FROM.

Sendmail sends "MAIL From:" and "RCPT To:", and from what I can tell
always has.   I don't recall seeing any damage.

The (possibly old) copy I have of zmailer does the same thing.

They don't seem to have been all that much influenced by the
way the commands are written in 821.

I personally tend to send "Mail From:" as I think it looks nicer,
if I'm doing damage somewhere, I have yet to be informed of it.

While it is certainly true that if something in the spec is causing
problems, we can deprecate it, and eventually outlaw it.  But before
this is ever done there has to be real evidence that the thing to
be abandoned is truly useless, and really causes problems.  Finding
some buggy (and since corrected) implementation is not sufficient
for that.   If it were, all messages sent via SMTP would be required
to have a Message-ID: header - remember there was once an
implementation that bounced messages with no Message-ID header.

If we were designing a new protocol I would tend to agree that
case insensitivity was a waste of time, case insensitivity is to
make life easier for humans who tend to believe that a word is
still a Word however the characters are written.  Computers don't
need that, and generally prefer bytes to have just one possible
value.   But at this stage whatever benefit could possibly be
gained by a change to case sensitivity is so insignificant that
even discussing the change this way isn't worth the bother.

kre