Re: RULE vs MGET

Patrick Stickler writes:
 > 
 > > A closer analogy would
 > > be if the HTTP spec mandated that URIs ending in .html should resolve
 > > to representations containing html.
 > 
 > Er... didn't I just say that?  ;-)
 > 

No - you said all html representations should be in URIs ending in
html. That would forbid other URIs resolving to references containing
html.

 > > Actually, my only real concern with this MGET stuff is that if it does
 > > become the standard way for an agent to retrieve descriptive metadata,
 > > the likelyhood of me personally being able to participate in the
 > > semantic web in the near future is vastly reduced. I just can't
 > > imagine web hosting providers providing URIQA enabled servers cheaply
 > > in the near future. The main benefit of the RULE approach for me is
 > > that I can participate today with my existing web account. - I suspect
 > > this also translates to a much faster uptake globally.
 > 
 > I appreciate your position. Adoption of URIQA is similar to adoption
 > of WebDAV. It requires the involvement of the web authority to a
 > greater or lesser degree.

Which is to my mind a very compelling reason for it *not* to become
the standard for something as important as information discovery on
the semantic web. You can't use MGET with existing web infrastructure,
and most people don't have the ability to change existing web
infrastructure.

In fact I'd speculate that probably > 99% of the web content in the
world is maintained by people without the ability to change the
infrastructure that serves it.

When you add that to the likelyhood that the sum total of SW agents in
the world probably currently numbers less than 100, I'd say that any
spec with a real chance of succeeding must be biased towards adding
complexity to the client in favour of modification of existing web
infrastructure.

Cheers,

Phil

Received on Wednesday, 10 March 2004 08:19:13 UTC