- From: Roy T. Fielding <fielding@kiwi.ICS.UCI.EDU>
- Date: Wed, 26 Mar 1997 23:01:59 -0800
- To: Yaron Goland <yarong@microsoft.com>
- cc: w3c-dist-auth@w3.org
>Actually Jim, URIs are tokens and thus do not have spaces. As the HTTP
>spec requires spaces between all tokens (please see section 2.1 of RFC
>2068), your quotes are unnecessary.
Huh? URIs are not tokens and whitespace is not a reliable separator in
header fields. I must have said that a couple thousand times on http-wg,
and one was just last week. If you want to separate URIs from parameters
in a header field, you must surround the URI with <angle brackets> or
"double-quotes", since they are the only characters that will never
appear within a URI even if it is broken across more than one line.
>Also if the Link header is going to be dumped from the HTTP spec then
>there is no backwards compatibility to be worried about and we can
>format the header properly, which should be:
>Link = "Link" ":" Source Destination Type *(";" link-param)
If you are going to redefine the syntax, then
Link = "Link" ":" 1#( link-type *(";" link-param) )
link-type = token
link-param = ( "from" "=" <"> URI <"> )
| ( "to" "=" <"> URI <"> )
| ( token [ "=" ( token | quoted-string ) ] )
otherwise, the HTTP/1.1 spec defines it as
if rel=type
<first URI> is destination
anchor="URI" is source (current document if elided)
else if rev=type
<first URI> is source
anchor="URI" is destination (current document if elided)
I don't like that model, but it was inherited from HTML.
>Also, if the link header is going away then we can forget all the
>non-dav parameters. After all, no need to define the relationship with
>non-existent parameters. If it isn't in the HTTP 1.1 spec, it isn't our
>problem.
Extensibility is a WEBDAV problem.
....Roy
Received on Thursday, 27 March 1997 02:14:27 UTC