> I see the elements in the document head being used by the HTTP server
> to generate HTTP headers. This way you can have a general purpose interface
> for agents like MOMspider which also works for non-HTML formats like GIF
> or plain text.
>
> How about the following for the HTML+ document HEAD:
>
> <!-- this element may often be implied by HTTP header info -->
> <!ELEMENT DATE - O EMPTY>
> <!ATTLIST DATE
> created CDATA #IMPLIED -- RFC 850 Date format --
> expires CDATA #IMPLIED -- RFC 850 Date format -->
>
> <!ELEMENT OWNER - O EMPTY>
> <!ATTLIST OWNER
> owner CDATA #IMPLIED -- name of owner --
> contact CDATA #IMPLIED -- email address as URL -->
>
> These two elements wouldn't cause any problems for existing browsers.
> The OWNER element can have either the owner or contact attributes or both.
> Sometimes it might be nice to specify the owner by a hypertext link to a
> page giving an autobiography, affiliation, contact details etc. The owner
> attribute is as you suggest and can be used as an indirection to look up
> the real owner's e-mail address by programs privy to that information.
> An alternative to OWNER is two new relationship types for LINK. The OWNER
> element seems marginally cleaner.
That is an excellent solution. However, I have a couple nits to pick.
I would prefer:
<!ATTLIST OWNER
name CDATA #IMPLIED -- name (or alias) of owner --
email CDATA #IMPLIED -- email address of owner --
href %URL; #IMPLIED -- URL for biographical info -->
as it more directly reflects what you describe. Obviously, it would be
easy for a server to translate this into
<LINK rev="OWNS" title="name (email)" href="href">
and thus follow Tim's suggestion as well.
....Roy Fielding ICS Grad Student, University of California, Irvine USA
(fielding@ics.uci.edu)