Re: IANA registry vs. URIs

[Danny Ayers <danny.ayers@xxxxxxxxx>]

On Tue, 19 Oct 2004 15:12:32 -0700 (PDT), Dare Obasanjo <kpako@xxxxxxxxx> wrote:
> 
> --- Danny Ayers <danny.ayers@xxxxxxxxx> wrote:
> >
> > > >Expecting an Atom reader to some how "learn" what
> > to
> > >do with an ad-hoc link simply given a URI is
> > venturing
> > >on AI.
> >
> > "learn"?
> > At it's most sophisticated I think we're talking
> > about the reader
> > downloading a plugin.
> > An Atom reader could fairly easily look up how to
> > interpret an ad hoc
> > link type, it might even involve behaviour. But I
> > think it would be a
> > stretch to call process/service identifiers AI.
> 
> And my predictable answer is if all an application has
> to do is key off of some URI then know where to go to
> download the code to learn how to consume that URI
> then there is no difference between "link based
> extensibility" and "extensiblity based on extension
> elements in a third party namespace".

There isn't all that much difference. The syntax restrictions of the
link construct do signal that the app should  treat this as a link
construct, and identify the endpoints (subject and object) of the link
relationship. But that could be in other ways.

> Of course, this ignores the feasibility of actually
> implementing such a mechanism.

That would depend on what is expected of the implementation.

> Without a clear definition of what is meant to be a
> link then there is no difference between using link
> elements for extensibility and using arbitrarily
> structured extension elements for extensibility
> besides an arbitrary syntactic restriction.

True, but  the definition could be anything from "a relationship
between the feed and the remote URI" onwards.

Cheers,
Danny.

-- 

http://dannyayers.com