Re: Comments on "The use of Metadata in URIs"

FWIW, I like the Metadata draft very, very much.  I think it's on the 
short list of the most compelling and useful work that's come out of the 
Tag.  I learned from it and I think many other readers will too.

Mark Nottingham writes:

> And now for something completely different...
> As noted [1], I'm extremely pleased to see this draft, and am hopeful 
> that it will drive some interesting further work. A few comments;
> * I agree with the sentiment in the second editorial note; sections two 
> and three, while containing interesting information, don't justify 
> their length. I would advocate a severe editing down.

FWIW, I respectfully disagree with my esteemd colleague Mr. Nottingham.  I 
found the sections following the EdNote to be extremely valuable.  I thing 
they will alert less experienced readers to a number of subtleties that 
would otherwise be clear only to experts.  Were this a Recommendation I 
might argue as Mark does for sticking to a terse, normative core;  given 
that this is a Tag finding, I think the purpose should be to educate and 
clarify.  I like it better in long form.
> * I disagree with the third editorial note's suggestion; there's a lot 
> of misunderstanding of the Opacity axiom, and this document should try 
> to help disperse it.

Indeed.  See comments above. 

One suggestion on the Tag's draft:  I think you could be stronger in 
discouraging "observers" from inferring information from URI structure, 
even when there is a spec that licenses such inference.  The point is 
finally made strongly in the conclusions where it says:

"People and software using URIs assigned outside of their own authority 
should make as few inferences as possible about a resource based on its 
identity and inferences arising from delegated authority."

...but before that there is a lot of text telling you what you can do if 
you have read the spec for the URI scheme, etc.  I think the tone could be 
a little more:  if you think you need to look inside you're very probably 
wrong, or at least you're  buying into a significant compromise.  If that 
hasn't disuaded you, here are the rules for where you may look if 
necessary (e.g. what the scheme spec says), vs. what you should in all 
cases avoid (inference from filename extensions). 

It's just a matter of balance, I think all the right thoughts are in there 
somewhere.  Thanks again, and congratulations.  This is really nice work 

Noah Mendelsohn                              Voice: 1-617-693-4036
IBM Corporation                                Fax: 1-617-693-8676
One Rogers Street
Cambridge, MA 02142

Received on Wednesday, 13 August 2003 14:55:21 UTC