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


I disagree with Dan's argument below.

In the existing RFC 2717/2718, there is a provision for
'alternative trees' and URI schemes that included a hyphen
were definitely not IETF/IANA registered.

If in Dan's example below, the 'wizzy:' scheme was instead
'vendorco-wizzy:' the chance of collision would be orders
of magnitude less (especially if 'vendorco' was the DNS
domain name registered to VendorCo).

There is no demonstrable advantage at all to IANA registering
non-unique tokens (nor any precedent in any previous IANA-
maintained registry).

- Ira

Ira McDonald (Musician / Software Architect)
Blue Roof Music / High North Inc
PO Box 221  Grand Marais, MI  49839
phone: +1-906-494-2434

-----Original Message-----
From: []On Behalf Of Dan
Sent: Wednesday, January 19, 2005 11:45 AM
To: Weibel,Stu
Subject: Re: Duplication of provisional URI namespace tokens in

On Jan 19, 2005, at 3:47 PM, Weibel,Stu wrote:
> What I propose is, very simply, that IANA-registered tokens be unique.
> And yes, on a first come, first-served basis.  To do otherwise is to
> strip the U from URI under the imprimatur of IANA.

The U is long gone (witness mms:
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.

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

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

> Larry raised the point that:
>> The proposed registration rules are based on the fact
>> that it is possible to invent and deploy a URI scheme
>> without IANA and IESG approval.
> This may be the case, but shouldn't we be providing incentives to 
> reduce
> both the likelihood and impact of this happening?  Assuring that ALL
> IANA-registered URI scheme tokens are unique is a step in this
> direction.

I don't believe so. Consider VenderCo who has just released
WizBangTool that supports wizzy: URIs. Somebody files
a bug that says "your scheme isn't registered" so they follow
their nose to the registry, only to find that some long-defunct
sourceforge project registered wizzy: 5 years ago. With
unique registration, VendorCo's choices are:

  (a) change their software and register a wizzy2: uri scheme
  (b) ignore the process and go their merry way

Who benefits from (a)? Almost nobody. What are the odds
they'll choose (b)? Epsilon. What are the odds this case will
come up? I think it's quite likely.

Assuring that ALL IANA-registered URI scheme tokens are unique is more
likely to produce a useless, irrelevant registry than anything else that
I can see.

If we give VendorCo the option to register wizzy: along with the
sourceforge project, the community benefits from the ability
to look up wizzy: in the regsitry and contact VendorCo to encourage
them to participate in the rest of the standardization process.

Dan Connolly, W3C

Received on Wednesday, 19 January 2005 18:14:35 UTC