Re: 2.3 URI Ambiguity

On Tue, 2003-12-02 at 12:51, Walden Mathews wrote:
[snipped text]
> Ian,
> 
> It's now clearer that you can use URI outside the scope of the
> Web, and that Web best practices don't apply there,
> necessarily.
> 
> I'm still seeing a few problems, though.
> 
> 1. Will the reader have enough information to know what
> is meant by "outside of Web architecture"?  I'm not sure I do,
> and likewise I'm not sure I see that the database application
> in question is "outside of Web architecture", unless I'm given
> more information.  It may be clearer to make the distinction
> as "URI used for identification tasks other than fetching
> representations via web protocol".  Is that the distinction?

By "inside" I pretty much mean "following relevant specifications."
So dereferencing is one operation you can carry out according
to specifications, but there are others (e.g., RDF).

> 2. I'd say "database keys" is still identification, 

Yes, but are you just using the URI as a string? Are you
following any Web specifications in using it in a database?

> and that
> there aren't any uses for URI except for identification, in
> one form or another.  So, I think you could remove the
> assertion that URI are "used in any number of roles..." without
> harm.

Ok, I'll think about that. Simpler is better.

> 3. The above text now takes a very strong position on the
> resolution of httpRange-14, but it need not. 

I didn't mean to! Whoops!


>  "In the Web
> architecture, mailto:nadia@example.com may identify a mailbox,
> not a person."

Ah, that's not an http URI, so I'm off the hook. The mailto
URI scheme spec allows me to assert that mailto URIs (of
a particular form; the TAG discussed this at length) identify
mailboxes.

> 4. The space given to explaining the exception seems
> disproportionate: only the last sentence hits the meat of the
> matter.  

I am happy to try to shrink the text; that was a first draft.

> And there, I think examples of poor practice are
> badly needed.  Were there some and I missed them?  I think
> if those examples were included, I'd be satisfied and would
> not need to look at examples from outside the architecture.

I welcome more examples. I am hoping to keep the text terse, however.
Thanks for your comments Walden,

 _ Ian
> 
> Thanks,
> Walden
-- 
Ian Jacobs (ij@w3.org)   http://www.w3.org/People/Jacobs
Tel:                     +1 718 260-9447