Re: RFC 1522 (for HTTP 1.1 Warnings)

On Thu, 12 Dec 1996, Martin J. Duerst wrote:
> On Wed, 11 Dec 1996, Erik van der Poel wrote:
> > > RFC 1522 is designed for 7-bit channels. If you have an 8-bit
> > > channel, there is no reason to use it.
> > 
> > No, 1522 was designed for something far more restrictive than a simple
> > 7-bit channel. It was designed for email headers (e.g. To:, Cc:, From:,
> > etc).
> > 
> > In these headers, characters such as comma, '<', '>', '(', ')' and so on
> > have a special meaning, and there is a *lot* of software out there that
> > does stuff with these characters. RFC 1522 has a lot of rules for this
> > type of thing.
> > 
> > So, even if you have an 8-bit email channel, RFC 1522 is needed.
> Erik - You are right in what concerns the use of RFC 1522 in email
> headers. However, for the problem I was discussing (the use of RFC
> 1522, and the strange exception for ISO-8859-1, in HTTP 1.1 warnings),
> these considerations are not really important. The only thing you
> may have to escape in a warning is a CRLF.

Not completely true - according to HTTP 1.1 syntax, the warn-text is
a quoted-string, i.e. at least <"> would also have to be escaped in
some way.

Another oddness of this RFC1522-encoding-in-HTTP-headers business is
that it violates the rules for using this encoding in MIME.  RFC 2047,
the new version of "MIME Part Three" which supersedes RFC1522, explicitly

   + An 'encoded-word' MUST NOT appear within a 'quoted-string'.

(And I understand this is not a new rule, it was already implicit
in RFC 1522 syntax.)


Follow-Ups: References: