RE: Unique handle generation

["Hollenbeck, Scott" <shollenbeck@verisign.com>]

Paul,

If we disallow identifier change, and require the identifier to reflect the
creating repository per your suggested rewording, I think we run the risk of
confusion over time.  After a repository change the unchanged identifier
will not reflect the new administrative repository; this seems counter to
some of the desires I've read over the last few days.

A basic question: do we feel that it's necessary for someone to look at an
identifier and know which repository it lives in, or would some kind of
resolver technology be OK to provide that information?

The eyeball approach leads to problems if we believe that objects can change
repositories over time, but identifiers shouldn't change over time.

The resolver approach helps manage the repository change problem, but then
identifiers may be less intuitive to someone who happens to be looking at
one.

<Scott/>

>-----Original Message-----
>From: Paul George [mailto:pgeorge@saraf.com]
>Sent: Monday, March 12, 2001 7:53 AM
>To: ietf-provreg@cafax.se
>Subject: RE: Unique handle generation
>
>
>Perhaps remove the second part of #4?  Why add auxilary info 
>to the handle?
>It would then be serving two purposes. So it would read:
>
>>4. An object handle MUST contain information that 
>unambiguously identifies
>>the object.
>
>But then is it duplicating #2.(2. An object handle MUST be 
>globally unique.)
>
>The alternative would be:
>
>4. An object handle MUST contain information that 
>unambiguously identifies
>both the object and the object's administrative repository at time of
>creation.