RE: URI: Name or Network Location?

> I'm not asserting that. I never have. We need more than one URI
scheme.
> 
> I am, though, asserting that the goals of the info: URI scheme can
> be fully met by using http: URIs, and that the percieved/intended

I agree completely with Patrick here.  Just because someone writes a FAQ
asserting that "established convention does not work" does not make it
so.  Considering that the net result of this proposal is to violate the
basic principles and intent of URIs, one would expect the authors of the
proposal to have spent some time validating their beliefs with people
who have more experience.

> constraint against resolvability of info: URIs is quite simply
> an illusion.

It's ironic in a way, that the other class of people who invent
unnecessary URI schemes are people who claim that it's because they
*want* better resolvability within their agents; for example the "feed"
scheme for RSS feeds.

The justification offered in the FAQ seems to boil down to a fallacy --
since URLs and URNs *can* be used to locate things, then they by nature
are not suitable for *identifying* things, and therefore, the authors
must invent a new scheme for resource identification.  This is patently
false -- the primary purpose of URIs (and especially URNs) has always
been to identify, and resolvability is a secondary and non-critical
facet.  Furthermore, the DNS portion of the URI is as much about
delineating naming authority as it is about locating a port for GET
operations.

I really think that the proposal authors should have solicited input and
feedback much sooner, and perhaps could have avoided getting into a
situation where someone if defending what is fundamentally a bad
architecture. 

Received on Friday, 23 January 2004 12:35:37 UTC