RE: Bridging XML/HTTP

Why not call it http:meta rather than html:meta?

Come up with a canonical DTD for the HTTP protocol & the http: namespace
would use elements & attributes (if any) from that.

<!-- canonical representation (not fully thought out...) -->
<http>
<Request Method='GET' Version='1.1' URI='a/b/c'>
 <Headers>
	<Authorization>BASIC ynz70f2=<Authorization>
	<If-Modified-Since>Sun, 06 Nov 1994 08:49:37 GMT</If-Modified-Since>
	<User-Agent>Mozilla etc</User-Agent>
	<Accept>text/plain</Accept>
 </Headers>
</Request>

<Response>
 <Headers>
	<Pragma>no-cache<Pragma>
	<Cache-Control>no-cache</Cache-Control>
	
<Location>http://www.w3.org/hypertext/WWW/NewLocation.html</Location>
	<Last-Modified>Tue, 15 Nov 1994 12:45:26 GMT</Last-Modified>
      <Expires>Thu, 01 Dec 1994 16:00:00 GMT</Expires>
	<Content-Type>image/gif</Content-Type>
 </Headers>
</Response>

</http>

<!-- some http aware/friendly representation of something -->
<foo>
 <http:Headers>
  <http:Pragma>no-cache</http:Pragma>
  <http:Expires>Thu, 01 Dec 1994 16:00:00 GMT</http:Expires>
 </http:Headers>

<ramble>
Now is the time for all good men to come to the aid of their...
</ramble>

</foo>




Mike D


> -----Original Message-----
> From: Mark Baker [mailto:mark.baker@Canada.Sun.COM]
> Sent: Friday, March 24, 2000 8:42 AM
> To: xml-dist-app@w3.org
> Cc: fielding@ics.uci.edu
> Subject: Bridging XML/HTTP
> 
> 
> The following is a proposal to attempt to solve the problem of how
> arbitrary XML documents can specify aspects of their transfer over
> HTTP.  Currently (AFAIK), the only widely deployed XML DTD that enjoys
> this capability is XHTML 1.0. It does this via the META tag,
> specifically the HTTP-EQUIV attribute. For example; 
> 
> <html> 
> <head><meta http-equiv="pragma" content="no-cache"/></head> 
> <body> ... </body> 
> </html> 
> 
> Rather than try and develop a new mechanism, I suggest we try to
> encourage DTD authors who need it, to reuse HTML.  Also, one
> mechanism is likely preferable to language specific ones, as it
> allows web servers to handle this generically for any text/xml or
> application/xml content.  For example, in responding to a HEAD (and
> per RFC 2616 sec 9.4, "SHOULD"), it becomes straightforward for the
> server to extract the HTTP headers from any content.  However, this
> does require a heuristic about where to find these tags since they
> won't be in the same place in the document;
> 
> - if the DTD has a place for metadata (eg. HTML HEAD, SVG METADATA)
>   - then the META tag will be added to its content model
> - otherwise
>   - the META tag will be added to the content model of the root
>     element
> 
> It should also be suggested that the METAs come "first" so as to
> minimize the work the server has to do to respond to a HEAD (in
> the case of large bodies).  In the case of an existing place
> for metadata, I would just suggest that the METAs SHOULD be placed
> as close to the "beginning" (in the serialized document sense, not
> a DOM sense) as possible.
> 
> For example, with SVG, we would add html:meta to the content
> model for metadata, so we could say;
> 
> <svg> 
> <metadata> 
> <html:meta xmlns:html="http://www.w3.org/1999/xhtml"
> html:http-equiv="pragma" html:content="no-cache"/> 
> </metadata> 
> </svg> 
> 
> Thoughts?
> 
> MB
>

Received on Monday, 27 March 2000 14:07:06 UTC