Re: Prefixes fail-safe registry thoughts

Manu Sporny wrote:
> ...
>> Minimally, this issue should be addressed for @rel, because it affects
>> even non-RDFa use cases.
> 
> Do you think that this proposal could address your issue with @rel?

"My" issue with @rel is that I'd like to see a single interpretation 
across all HTML languages. For that, the conflict between URI and CURIE 
syntax would need to be resolved.

>> But that does change the rules for XHTML, doesn't it? Or are you saying
>> that XHTML != XHTML5 (== HTML5 serialized as XML)?
> 
> Not speaking as a member of the RDFa TF... just kicking around some ideas...
> 
> Yes, it does change the rules slightly, but shouldn't create any
> backwards-incompatible changes. That's what I should have said:
> 
> "The @prefix addition would not create any backwards-incompatible
> issues, nor would it adversely affect the processing rules. All of the
> 'XHTML+RDFa 1.0' documents would continue to work."

Yes, it's backwards compatible with existing valid XHTML+RDFa1.0 documents.

But (IMHO) XHTML+RDFa1.0 is already incompatible with other languages 
(see the long exchange over on the TAG mailing list from last weekend).

> ...
>> On the other hand, it will make the owners of purl.org extremely
>> unhappy, unless it doesn't happen frequently. (Did you talk to them?)
> 
> If the anti-indirection crowd is correct, it will happen quite
> frequently. If the pro-indirection crowd is correct, it will happen
> infrequently. We should plan for the worst.

In particular, as if opens up a simple DOS scenario against that server.

> I used purl.org as an example - any redirection service would do.  I
> haven't talked to the purl.org folks yet - don't know who to contact
> there. I also need to talk to them about http://purl.org/media# not
> resolving, but http://purl.org/media/# resolving. Do you know who I
> should talk to?

Nope, sorry.

Speaking of that: redirection is nice for delegating the authority, but 
of course is costly (it doubles the number of requests, and potentially 
the redirect will not be cached).

BR, Julian