RE: First reactions to mandatory draft

From: Paul Leach (
Date: Tue, Jan 20 1998

Message-ID: <>
From: Paul Leach <>
To: Jeffrey Mogul <>, "'Scott Lawrence'" <>
Date: Tue, 20 Jan 1998 14:03:10 -0800
Subject: RE: First reactions to mandatory draft

> ----------
> From: 	Scott Lawrence[]
> Sent: 	Tuesday, January 20, 1998 1:19 PM
> To: 	Jeffrey Mogul
> Cc:
> Subject: 	Re: First reactions to mandatory draft
>   What I believe is needed is a way for the sender to say 'In order to
>   correctly interpret this message you must understand this set of
>   extentions'.  I don't think that the prefix mechanism is needed to
>   say that.
It is if the extensions are named by URL and can have header names that

To use the "Skidoo" example, if your server understands both "Skidoo"
extensions, then it should keep all headers of the form *-Skidoo, with the
prefix, until it gets a "Man:" header.

Suppose the headers were registered to guarantee no conflicts. Then they
would be "Skidoo1" and "Skidoo2", and the message might look like

GET /foo.html HTTP/1.1
Skidoo1: "aValue"
Skidoo2: foo="bar"
Man:, Skidoo1
Man:, Skidoo2

You'd still have to wait until you saw the Man: header to decide whether to
just discard the "Skidoo*" headers, or to return an error. I don't know
whether this qualifies as "backtracking" but its at least harder than the
current parsing regime.