RE: New URI registration draft; significant changed

Allow me to add my compliments to Larry, Ted, and Tony on the URI
Registration revision.  As a vocal critic of the previous version, I can
say that my concerns about token uniqueness are entirely resolved.

Larry has voiced concerns about making the review process more
deterministic, and I think this is likely to continue to be a problem,
but I am inclined to think that we will need some experience to fine
tune things.  As a current URI scheme supplicant, I'll certainly be
working through this process in my own mind and trying to understand
more clearly any nuances in the changes.

I would also like to see the advice about MIME versus URI articulated in
a persistant way.  I don't know whether it should be here, or perhaps in
a supporting document? [Best practice recommendations and guide to URI

Again, thanks to the authors of this important document.  It is a
significant contribution to the evolution of network infrastructure. 


Stuart Weibel
Senior Research Scientist
OCLC Research

-----Original Message-----
From: [] On Behalf Of Larry
Sent: Monday, February 21, 2005 10:29 PM
To: 'McDonald, Ira'
Subject: RE: New URI registration draft; significant changed

Thanks for the kudos. I think I waited until a little too close to the
deadline, so Ted didn't have a chance to see this version before I
shipped it. You might have any nits, but I do.

There are several formatting problems with the .txt version; not as much
of a problem with the .html. I just edited the XML and didn't have time
to proofread the results. These will get fixed up in the next version,
so I don't think you need to point them out :) (things like paragraph
separations being unclear).

Martin Duerst (2/8) pointed out that "" never took
off. I left the (optional) mailing list review to be I'm
still a little uneasy and wished we could have a different mailing list
name, even if the two were set up to automatically track (e.g., make a member of

Tim Berners-Lee suggested adding a pointer to the W3C Web Architecture
document. I did, but I'm not sure I gave it justice. I had thought of
incorporating both the reasons for always registering schemes, and also
the advice about when using a new MIME type is preferable to using a new
URI scheme. Next version.

Stu Weibel proposed several alternative methods of registration which
were interesting and worthy, but, after reading RFC 2434 (as suggested
by David Black) I thought we should try to stick to one of its
recommended.  I hope that the current draft addresses both the
"uniqueness" concerns (guaranteed, unless IESG approved), as well as the
"land grab" concerns (expert review will catch blatant cases.) Is it
credible that expert review against the limited criteria for
"provisional" registration can happen in a timely fashion?

I wondered about allowing a free-for-all private use "x-" designation,
or going back and proposing "vnd.organization.whatever" as a kind of
provisional registration that required even less review.

I'm not sure it's clear how this process affects the development of
internet drafts and RFCs that register URI schemes. If an individual or
working group has an Internet Draft with a new scheme definition, which
should it register it? Before anyone writes the first experimental
implementation, I suppose? How frequently should the registration be
updated? Is it possible to point to "the latest version of
draft-workingroup-stufwithscheme-XX"?  Should we point out that the
specification of the URI scheme MAY appear in a standards track
document, in which case the URI itself is standards track, and that the
registry itself doesn't track the standards dependence; e.g., a
standards track document which makes a normative use of URIs as protocol
elements and cares about the schemes should make a normative reference
to the URI scheme specification, which thus needs to have the
appropriate maturity level.  Is it necessary to say this?

I continue to be confused as to whether the IETF prefers "historic" or
"historical" as the right word for cruft.

Charles Lindsey gave a use case about VenderCo with their new wizzy:
URI scheme, finding an old defunct registration. I'm not sure the
document makes it clear how to ask for IESG approval. It's not a
"protocol action"; how does it get on the IESG agenda?

Dan Connolly sent a pointer to what I think is intended to be a list of
'known URI schemes', but the web site isn't responding at the moment.
I'm expecting we will want to populate the provisional registry quickly
with all known schemes, and encourage the actual owners to ask to update


Received on Tuesday, 22 February 2005 17:02:17 UTC