RE: Hashed URIs

At 10:49 AM -0800 11/25/02, Joshua Allen wrote:
>  > 2) Appendix A will lead to lack of interoperability. You haven't
>>  shown why canonicalizing (that is, changing) the pre-hashed URI has
>>  any advantage.
>
>I assume canonicalizing would *help* interop.  The point after all is to
>have two URIs that are the same evaluate to the same.

Re-read Appendix A and explain how this will help. Section 4.3 says 
"A possible canonicalization algorithm for hierarchical URIs is given 
in Appendix A." Note the word "possible". Saying that this is done 
sometimes but not others will immediately lead to lack of interop.

>FWIW, I think that a hashed URI scheme is rather useful for semantic web
>applications.  Many semantic web applications already use
>canonicalization and hashes internally to provide faster query and
>comparison of URIs, and I think that the "one-way" nature of hashes will
>make them a popular choice for people who wish to provide metadata
>without necessarily revealing their web browsing habits.

I'm still confused. Who needs to provide metadata without revealing 
their web browsing habits?

>   For example:
>
>* publish a "1 to 5 rating" of a web page, so that anyone visiting the
>page can see all of the previous visitors' ratings

This is not useful unless you know what web page is being discussed.

>* publish a trust rating of a particular mailing list participant

This is not useful unless you know the participant

>* publish metadata associated with yourself in such a way that automated
>spam tools cannot use the metadata to target you

You can publish metadata about yourself without using your email address.

--Paul Hoffman, Director
--Internet Mail Consortium

Received on Monday, 25 November 2002 17:47:17 UTC