W3C home > Mailing lists > Public > uri@w3.org > February 2005

RE: Proposed wording (was: Duplication of provisional URI namespace tokens in 2717/8-bis)

From: Weibel,Stu <weibel@oclc.org>
Date: Fri, 4 Feb 2005 13:48:51 -0500
Message-ID: <8CC50D49B6828C4FBAB7DA1FCAB0526A271447@OAEXCH1SERVER.oa.oclc.org>
To: "Larry Masinter" <LMM@acm.org>
Cc: <uri@w3.org>

I agree that "MUST NOT" does not, in itself, protect.  The registry
should preclude duplication, either by brief review by a human or as a
feature of the registry code.

I am not proposing that provisional schemes be registerable without
review.  I am proposing that the level of review might be less rigorous
(consuming less time and expertise).

I will try, in a subsequent message, to articulate more clearly what I
think should happen, how it maps to current reality, and the benefits of
changing it

stu

-----Original Message-----
From: Larry Masinter [mailto:LMM@acm.org] 
Sent: Friday, February 04, 2005 1:17 PM
To: Weibel,Stu
Cc: uri@w3.org
Subject: RE: Proposed wording (was: Duplication of provisional URI
namespace tokens in 2717/8-bis)

> "SHOULD NOT" simply is not strong enough to allow organizations to 
> proceed with confidence in deploying business systems that cannot be 
> protected from ignorance or malice.

"MUST NOT" does not protect you against ignorance or malice, either.
Some other mechanisms must be employed to protect the registry from
misregistrations, for whatever reasons those registrations might have
been submitted.

> This is especially true given that in all this discussion no one has 
> advanced plausible arguments concerning the supposed duplicates in 
> existing URI scheme proposals and the consequences thereof.

Guaranteeing uniqueness to scheme names that have not passed any review
would make the registry itself a target for abuse.

It is at least plausible that URI scheme names might see an echo of the
abuse of top-level domains to first-come-first-serve allocation of
short, friendly names.

> There has yet to be a single unanswered objection to the proposed 
> requirement that ALL newly registered provisional tokens be unique.

There is some ambiguity about which "provisional" you're talking about,
hansen-provisional or weibel-provisional. I wrote:

# I think that your proposal is basically to take hansen-permanent # and
split it into weibel-permanent and weibel-provisional, # and map
hansen-provisional to weibel-vernacular.

if you agree with this analysis: I think it is important that the levels
that require review also assure uniqueness among those schemes that have
passed review; I also think that we should be very careful about what
guarantees for short names whose registration haven't passed any review.

Larry
Received on Friday, 4 February 2005 18:49:33 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:25:09 UTC