RE: updating RFC 2718 (Guidelines for new URL schemes)

> I  observe that there is an "school of thought" which posits that any 
> resource can and should be identified with an http URI. Without 
> addressing the merits of that argument, I would say the existence of 
> that argument, and the significant ignorance of that argument among 
> many important groups of implementers makes the 
> not-already-available-with-previously-registered-URI-schemes 
> criterion automatically controversial.
> Perhaps this is obvious.

I think you are mischaracterizing the "use HTTP" school; certainly
they don't advocate using HTTP instead of "mailto:", for example.
I think the position is rather narrower: the argument is that there
is little utility in assigned-name systems such as "urn:", and
those uses would be as well served by using "http:".

In any case, neither position is incompatible with the proposed
policy as written: use an existing scheme if you can (where
'http' is probably the most widely implemented existing scheme)
and invent a new scheme only if you must.

I think assertions of controversy should have first-hand
evidence: someone who actually believes that we _shouldn't_
request demonstrated, new utility for new URI schemes.

You should also note that these revised guidelines are proposed
for a context where URI schemes are not REQUIRED to meet them
in order to be registered; the judgment about whether the guidelines
are met is only used to put the URI scheme "above the line"
in the list of registered schemes. See minutes from the URIREV

I'm hoping Tony and Paul will come out with a draft of new
registration procedures soon.


Received on Wednesday, 8 September 2004 06:23:31 UTC