W3C home > Mailing lists > Public > uri@w3.org > November 2003

RE: non-IETF tree URI scheme draft

From: Hammond, Tony (ELSLON) <T.Hammond@elsevier.com>
Date: Wed, 12 Nov 2003 19:55:16 -0000
Message-ID: <54A600C436EA694581B93E4BD4D4788A06B73D7F@elslonexc004.eslo.co.uk>
To: 'Larry Masinter' <LMM@acm.org>, hardie@qualcomm.com
Cc: uri@w3.org

Well, this seems to be all good clean knockabout fun. :) But just to add a
couple of sober points:

	1. I think it's about time we decoupled DOI from NISO. These two
schemes ("doi" and "info") are entirely different initiatives coming from
separate directions and with entirely separate intents. Let's not confuse
them. (And BTW - NISO has plenty of other things to do than steer
prospective URI applicants through IETF registration processes.)

	2. URN syntax is /not/ IMO perhaps the best fit with DOI which makes
much of its multiple resolution capabilities. Without query/fragment
compenents I'm lost as to how to select one DOI service over another. And
the one single character that is guaranteed in any DOI string is the "/"
character - not suppported by URN. No - URN would seem to make a lousy fit
with DOI.

	3. And what's with this high store being placed in Google, which a)
is currently being talked up for public floatation, and b) only trawls the
surface (and non-subscription) Web anyway. As may have been previously
reported, working in terms of Google numbers then DOI knocks up
approximately 0.3% as regards deployed Web pages. That to my mind seems a
rather substantial number for curated resource identifiers which have
authority metadata records associated with them and are being continously


-----Original Message-----
From: Larry Masinter [mailto:LMM@acm.org]
Sent: 12 November 2003 19:19
To: hardie@qualcomm.com
Cc: uri@w3.org
Subject: RE: non-IETF tree URI scheme draft

I still stand by the advice that NISO should just
help organizations register their URN namespaces
(rather than creating a new registry) and that
DOI should use "urn:doi" since DOIs fit perfectly
as a URN namespace.

As far as organization names go, I wonder if it is
appropriate to suggest to IANA a heuristic:

If http://www.google.com/search?q=<orgname>
yields a web page of the organization asking for
the URI scheme, then the organization can have
<orgname> as its prefix.

Under this heuristic, the US National Information
Standards Organization rises above the National
Irish Safety Organization for "niso", but, since
the International DOI Foundation doesn't rise above
the International Diabetes Federation, perhaps
"doi-1" would be better than "idf-doi" (if IDF
insists on a private-URI rather than a registered

IANA would be free to choose whatever heuristics
they wanted, as far as judging the appropriateness
of a short ASCII string as an organization abbreviation,
but the above is what I use myself.


> > >  The IANA would then construct and assign a URI scheme of 
> the form:
> > > orgname-schemename.
> >
> >So 'info' would be 'niso-info'?  'doi' would be 'idf-doi'?
> >
> >Is the intention that if for example Mount Saint Mary's College came
up with
> >an 'info' scheme it wanted registered, them there would be a
> >scheme?
> >
> >--Ray
> That's the basic idea; niso-info: and idf-doi:.  This would allow for
> the gnome folks (who are apparently already using info: as well,
> according to a previous note) to use gnome-info: without collision,
> but wouldn't require any more complex decoration.
> Obviously, you still have the "two NISOs" problem, but that seems to
> be something that can be managed without adding extra complexity to
> avoid it.
> 			regards,
> 					Ted Hardi
Received on Wednesday, 12 November 2003 15:09:30 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:25:06 UTC