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

RE: Duplication of provisional URI namespace tokens in 2717/8-bis

From: Weibel,Stu <weibel@oclc.org>
Date: Wed, 19 Jan 2005 12:28:24 -0500
Message-ID: <8CC50D49B6828C4FBAB7DA1FCAB0526A2713D2@OAEXCH1SERVER.oa.oclc.org>
To: <uri@w3.org>

Dan Connolly writes:

> The U is long gone (witness mms: 
> http://esw.w3.org/topic/UriSchemes_2fmms)
> and people, being what they are, will
> only pay attention to IANA if it benefits them to do so.
> Currently, there is precious little benefit to pay attention to
IANA/IETF 
> when deploying a new URI scheme, compared to the cost.

The cost of consulting an IANA registry is what?

   - The social knowledge cost... Technology deployment is viral,
distributed, unruly, and often not well informed.  What incentive will
there be then, for the latests collection of exopthalmic 22 year olds to
know about an IANA URI?  Often, none, but the incentive grows large if
there is a need for broad scale interoperability (browsers that know
what to do with a particular flavor of URI, for example).

   - The cost of actually doing the lookup is negligible

The cost of NOT consulting a registry is evidenced in Dan's example of
duplicate scheme tokens in isolated pockets of the Internet.  The U is
*not* long gone, it is merely dinged up a bit.  This is evidence that
steps should be taken to assure an integral, well-managed Internet
namespace that can be relied upon to be U-compliant.


>>   Surely this cannot
>> be judged a good thing for Internet architecture?

> Yes, it can, if, on balance, it does more good than harm.

And the evidence that more harm would be done by a registry of
non-unique tokens as opposed to a registry of unique tokens is...?


Dan, I infer from your further remarks that you believe that a registry
that disallows duplicates will be ignored because of the strong branding
preference for wizzy: over wizzy2:?  You believe that people will prefer
the danger of name collisions for incapacitating their applications
rather than risk having to identify a different token?

Once again, I think we want to promote an architecture that makes it
easy to do the right thing (right being defined in a commonly-agreed set
of architectural principles, one of which is Thou Shalt Guard Against
Name Collisions Among Public Identifiers)
Received on Wednesday, 19 January 2005 17:28:53 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 13 January 2011 12:15:35 GMT