Re: status of Provisional URI registration in 2717/8-bis


When a similar model was discussed for message header fields [1], some 
people were concerned that the existence of a registry for headers that had 
not been subjected to review and community consensus-forming might be used 
as a back-door route to creating poorly-designed de-facto standards.  The 
name "provisional" was chosen there partly to make it clear that these were 
not definitions to which the community was committed;  in that sense, the 
implication of non-permanence is useful, even in the absence of any 
specific expectation of extinction.

In some cases, the intended purpose of the provisional registry was to 
document designs that have been published, so that the lessons of history 
are available for those who wish to learn from them.

So, in the case of rfc3864 [1], the registry record of fields labelled 
"provisional" is indeed potentially long-lived, but there is no guarantee 
that it will be so in any particular case.

Further (referring to your other message), the non-requirement of 
uniqueness in the provisional registry was specifically to allow 
alternative proposals to be recorded without creating a .com-style "land grab".

I'm not making any specific claim here that these criteria are appropriate 
for URI schemes, concerning which I think the bar for introduction should 
be at least a little higher than for message header fields.  (Also, there 
are fewer unregistered schemes that have escaped to breed in the wild.)



At 14:42 10/01/05 -0500, Weibel,Stu wrote:

>The introduction of two states for registered URI schemes (provisional
>and permanent) in  2717/8-bis [1] seems a very useful approach for
>providing a low-barrier mechanism for registration of new URI schemes
>while preserving a higher designation for those which have undergone the
>additional scrutiny of formal technical review.
>One can imagine different interpretations of this hierarchy, however,
>with important implications:
>In one scenario, registration of a new URI scheme would ALWAYS start
>with provisional status, after which one of several things can happen:
>     a.) a registrant takes no further steps beyond provisional
>registration; others may discover the registration, use it or not,
>replicate its functionality or not.  The registration remains active for
>as long as its instigators choose to sustain its documentation and
>specification, or it may simply fade into irrelevance, depending on
>uptake.  (only the completion of an appropriate registration template
>     b.) a registrant instigates formal technical review of a
>provisionally-registered scheme, which is, after due consideration,
>rejected according to standard procedural review.  The scheme remains a
>provisionally registered URI scheme, used (or not) by its advocates.
>The registration remains active for as long as its instigators choose to
>sustain its documentation and specification, or it may simply fade into
>irrelevance, depending on uptake.  (completion of an appropriate
>registration template and standards-track Internet Draft required; no
>requirement for maintaining ID in the case of rejection)
>     c.) a registrant instigates formal technical review of a
>provisionally-registered scheme, which is, after due consideration, is
>approved and designated a permanent URI scheme.  .  (completion of an
>appropriate registration template and standards-track Internet Draft
>required; approval transfers responsibility for maintenance of ID to
>Is this the intent of the authors?
>If so, it may be worth considering changing the name of the state from
>"provisional" to something like "Informational".  Provisional implies
>temporary, and it may well be that URIs which, for business or strategic
>reasons will never become part of the permanent URI array, may still be
>of importance to a limited set of stakeholders.  Similarly, "permanent"
>certainly conveys persistence, which is desirable, but implies that
>other calsses are not permanent, which is not necessarily logically
>true.  But the names of these things aside... the important thing is the
>underlying model for states.
>Is "provisional" potentially long-lived, or is it intended ONLY as a
>step to "permanent" or "rejected"?
>Stuart Weibel
>Senior Research Scientist
>OCLC Research

Graham Klyne
For email:

Received on Tuesday, 11 January 2005 13:59:23 UTC