Re: [Uri-review] Request for review of "ab:" URI scheme

[Graham Klyne <[e-mail]>]

Barry,

My apologies, I've been a bit unwell of late and haven't got round to going over
your materials in detail.

This is mainly a "holding" message until I can properly look over that material, 
and to let you know I'm not ignoring your request.

On reflection, my general sense at this time is that an addrbook: URI scheme
might be an interesting and useful development, but I'm not yet seeing that the
requirements of Sieve alone justify introducing a new URI scheme.

I recognize this probably isn't what you want to hear.

I also recognize that you have a need in Sieve for a URI that can be used to 
denote a common notion of lists of addresses defined locally in some unspecified 
fashion.

Under the circumstances, I would agree that my earlier suggestion of using http: 
is problematic and confusing, as one might expect dereferencing of the URI to 
yield a list of addresses (or a list of entries containing addresses), which 
would not be satisfiable using the pre-existing http: dereference semantics.

Some alternative suggestions I have that might meet the immediate requirements 
of Sieve include:
- Use the existing URN namespace for IETF parameters (e.g. 
urn:ietf:param:sieve:addrbook:default).  I mentioned this previously, and 
recognize the concern mentioned by Alexey that this could be awkward for 
hand-written Sieve code, but feel that it is otherwise a technically reasonable 
approach.
- Register a new URN namespace for addrbook (or even just ab) - this could yield 
URIs like: urn:addrbook:default (or urn:ab:default)
- I think the about: scheme has extension points that might be used, yielding 
something like about:addrbook:default.  (I haven't researched this properly, so 
there could be problems here.)

These suggestions do not preclude the possibility of subsequent work to 
introduce an addrbook: scheme that serves the wider needs we've discussed.  But 
if there really does exist a desire to address those wider goals, I feel they'd 
be better addressed without being constrained by the timetable for your Sieve 
work.  While I see potential value for an addrbook: scheme, I would say the jury 
is still out regarding whether that value would justify the development of a new 
URI scheme.

#g
--



Barry Leiba wrote:
>> So, reinvention aside, I think what Barry sketches above is prima facie a
>> reasonable approach.  If there's any real advantage to an addrbook: scheme,
>> I think it is in part that it can abstract away from a particular protocol
>> or mechanism and leave just the essential capabilities exposed.
> 
> Graham, where are we on this?  Should we proceed with the Sieve
> document as it is, with the addrbook URI?  Or do we still need more
> discussion?  We're otherwise ready to send the document to the IESG.
> 
> Barry
> _______________________________________________
> Uri-review mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/uri-review
>