Re: Connection tokens

• From: Roy T. Fielding <fielding@kiwi.ics.uci.edu>
• Date: Thu, 30 Jul 1998 11:54:22 -0700
• To: "J.P. Martin-Flatin" <martin-flatin@epfl.ch>
• Cc: http-wg@cuckoo.hpl.hp.com
• Message-Id: <9807301154.aa04641@paris.ics.uci.edu>

>I disagree: I don't think different implementations should be allowed to 
>use their own proprietary tokens. Allowing this would result in 
>interoperability issues.

Well, I completely disagree with that opinion.  The reason we added
Connection is because there was no way to indicate connection-only
extensions in HTTP.  We didn't need it for standard additions, except
for the obvious problem that all new standards look like extensions
to an already deployed implementation.

Allowing this is what solves the interoperability issue.

....Roy

Received on Friday, 31 July 1998 15:04:16 UTC