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

Larry Masinter [mailto:LMM@acm.org] writes:

> The goal is to avoid duplicates in the real world.

agreed

> Assuring uniqueness in the registry doesn't assure uniqueness in the
world,  

agreed

> and can result in the registry being out of alignment with the real
world 
> in a way that cannot easily be repaired.

I see your point... A small number of these collisions now exist
currently?  Do we know anything about the problems that are caused, or
the inhibition of uptake that might have resulted?  Does the evidence
inform us as to how vigorously they might be discouraged?


> The optimum situation is one where the registry tracks the real world 
> but also reflects expert opinions about the real world.

Deployment of a duplicate URI scheme strikes me as evidence of either 
   - ignorance of an existing scheme or one developing in parallel 
     (easy to imagine and almost certainly the majority case)
   - willful defiance, or 
   - outright vandalism

Agree it is desirable to afford the mechanism to discover these early,
even the ones that duplicate the token of another scheme.  There are
messes in the real world and we have to cope with them.

But shouldn't we be working toward a system that avoids such name
collisions, rather than simply accepting them?  Is there any competent
information architect that would knowingly promulgate a protocol that
embraces name collisions?  I don't think so.  So, how might the registry
become a prominent resource that is consulted *first* rather than as an
afterthought?

> I think that it might be helpful for there to be an expert opinion
review 
> of a registration entry, even if expert opinion review isn't part of
"approval" 
> for entry into the resolution.

I'd be interested to hear more about the scope of such a review.

 
stu

Received on Thursday, 20 January 2005 21:47:05 UTC