Re: [XRI] Back to XRI

Hi David,

thing-described-by.org is a good example of the use of 303 redirects.

In HXRI and XRDS-Simple we do want "aware" applications to be able  
directly request the meta data about the "thing" that the URI is about.

XRDS-Simole as used in openID, oAuth, OpenSocial and other places,   
currently use the accept header to request the XRDS meta-data as  
opposed to HTML content.

For example https://ve7jtb.pip.verisignlabs.com/ with no accept header  
returns my personal identity page.
The same URL with an accept header specifying MIME media type,  
application/xrds+xml, returns XRDS meta-data.

For the HXRI proxy the request for the XRDS meta-data is accomplished  
by appending a query string to the URI.

One of the things we are looking for is the best way to indicate in a  
query about a "thing" that you want a particular sort of meta-data  
about the "thing",  XRDS document, RDF document,  or something else.

When the URI is about a "thing" in meet space what is the meaning of  
content negotiation?

One of the things that we need the most work on is how to perform what  
might be thought of meta-data content negotiation for URI that are  
about "things".

The best solution we have found at this point is to use Link Headers  
to indicate where the related meta-data can be found at distinct URI.
This is would be consistent with Mark Nottingham's draft  
recommendations: http://www.mnot.net/drafts/draft-nottingham-http-link-header-01.txt
This clearly requires an extra GET that some users are resistant to.

Perhaps that is just a limitation of http: that we will have to live  
with.

The XRDS-Simple community has raised the possibility of using OPTIONS  
and custom request headers as ways of doing meta-data content  
negotiation.

The XRI-TC is looking for input on this.
Our desire is that revisions to HXRI and XRDS-Simple are as compatible  
with AWWW as possible.

I know that people like David Booth have been thinking about this for  
some time.

Thanks
John Bradley


On 12-Sep-08, at 7:15 AM, Booth, David (HP Software - Boston) wrote:

>> From: John Bradley [mailto:john.bradley@wingaa.com]
>> [ . . . ]
>> The XRI TC originally selected the 302 redirect for
>> compatibility with pre http 1.1 browsers.
>> Changing to 303 redirects may brake some clients but it
>> unlikely to be a significant issue.
>>
>> I understand that some people will feel that using redirects
>> is inefficient,  however it seems the only way to communicate
>> the desired qualities of the identifier in http:
>
> Yes, there would be an extra round trip for an agent that is not  
> aware of the XRI http subscheme conventions.  But an agent that *is*  
> aware of the conventions could inspect the URI and safely skip the  
> extra round trip, in a manner similar to the optimization suggested  
> for thing-described-by.org:
> http://thing-described-by.org/#optimizing
> That seems like a reasonable trade-off to me: naive agents incur a  
> small penalty, but agents doing high volume have the option to  
> safely optimize.
>
>
> David Booth, Ph.D.
> HP Software
> +1 617 629 8881 office  |  dbooth@hp.com
> http://www.hp.com/go/software
>
> Statements made herein represent the views of the author and do not  
> necessarily represent the official views of HP unless explicitly so  
> stated.
>