Cullen Jennings wrote:
On Jan 23, 2007, at 4:39 PM, John Levine wrote:
Remember that the point of the registry is primarily to prevent
namespace collisions.
I don't really agree with this - if this is the only point, you don't
need a registry - you only need to have people do something like prepend
the name with a domain name that the author of specification controls.
Or use a GUID. Or any number of things.
Cullen,
Registries come in many forms. Having a rule-based scheme, rather than a
registration-based scheme, is still a registry. There is nothing wrong a
rule-based scheme, if it does the job adequately. For most or all of the things
people define new IANA registries for, a rule-based scheme is not really
practical. To the extent that you disagree, try to apply your view to the
variety of existing registries that do not require standards status for new
entries. There are quite a few.
John's description of the function being served is exactly correct.
I believe you are conflating the administrative job of a registry with the
approval mechanism of a standards group. They need to be kept separate.
The reason we are using registries here is to define what the
standardization process is to make significant changes and update to
this protocol.
As I said, you are confusing things. The standardization process is for
standardizing things. A registry is for registering things. The latter has two
basic functions: 1) avoid collisions, 2) provide a means of finding the
registered entity.
The core question that we ought to be asking is the one that John raised: Is
there a fundamental damage in registering component mechanisms that are not
standardized? I believe the answer is not, and that the further answer is that
industry will provide ad-hoc mechanisms if we fail to provide an official one.
d/
--
Dave Crocker
Brandenburg InternetWorking
bbiw.net
_______________________________________________
NOTE WELL: This list operates according to
http://mipassoc.org/dkim/ietf-list-rules.html