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

On Tue, 25 Jan 2005 14:47:30 +0000, Graham Klyne <GK@ninebynine.org> wrote:


> In the case of email header fields, it was recognized that duplicates  
> *did* occur (rarely) in the wild, and that the primary need was to be  
> able to document what did exist in order that designers could avoid  
> them.  Local experiments sometimes escape and become global de-facto  
> standards.  By prohibiting provisional registration of duplicates they  
> would be forced out of sight, with no clear way for designers of new  
> fields to avoid duplicates.

I am not so sure that the header registration scheme _does_ permit  
duplicates of any sort (except perhaps where both versions have the same  
intent and there is disagreement as to the exact syntax).

I think that, for both mail headers and scheme-names, any provisional  
registration SHOULD have the property that it is capable of being upgraded  
to the permanent registry in due course.

Of course that "SHOULD" has an RFC 2119 interpretation, which means that  
in exceptional circumstances (see RFC 2119 wording) it _can_ be violated.

But normally, I don't think you should allow a name to be registered that  
is already in the registry without requiring that the old name be first  
removed (or historicized or whatever). The anomaly HAS to be removed first  
by some means, and if the parties cannot agree, then ultimately the IESG  
gets to decide which remains.

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Tel: +44 161 436 6131 Fax: +44 161 436 6133   Web: http://www.cs.man.ac.uk/~chl
Email: chl@clerew.man.ac.uk      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9      Fingerprint: 73 6D C2 51 93 A0 01 E7 65 E8 64 7E 14 A4 AB A5

Received on Wednesday, 26 January 2005 03:12:52 UTC