Thanks, Ray (and thanks for moving this off the Code4lib list).

So, I agree with the immediate problem.  I disagree about the
namespace thing (not because I think it's a valid identifier or not --
I don't care -- I just imagine it's been used as an identifier and
just like these here horses have left the stable, that means that mare
left with them), but that can be set aside for this discussion.

Here is my proposal for this, I'm not married to it, willing to accept
revisions anywhere, but it's pragmatic, simple and usable.

1) Register a vocabulary on the NSDL MetadataRegistry [1] for these
identifiers.  Include all of the variants, figure out an assertion to
help define the preferred identifier (there's no reason, of course, it
couldn't be the URI of the concept, although info URIs make this
trickier).  Relate NT/BT concepts to disambiguate our MODS mess (and

2) Make a site to solicit new proposals and a group from various
communities to vet the proposal and accept it into the registry or
reject it.  This could something as simple as a Google Site or
something to start.

3) Create a PURL to our registry that can work on either the prefLabel
or any of the associated identifiers.

4) Assuming the sun's over the yardarm somewhere, retire for a drink.
Call it a day.


On Tue, May 19, 2009 at 4:18 PM, Ray Denenberg
<[log in to unmask]> wrote:
> (Hopefully this�formats better.)
> �Ross Singer initiated� this discussion, first on code4lib, and
> subseqently��(at my suggestion) here and on OpenURL.
> Unfortunately it caught on only on� code4lib, and the thread there
> quickly���degenerated into incoherence.� I request that we discuss the issue
> here,��and/or OpenURL� (and not on the code4lib list)� and preferably on one
> list���only, and my preference would be here.
> �The issue, synthesizing Ross's message:
> �SRU has an XML� schema registry at
> �
> �OpenURL� has a registry of metadata formats:
> �
> it includes (but is not limited to) XML schemas.
> OpenURL and SRU are using different info URIs� to describe the same things.
> For example:
> info:srw/schema/1/marcxml-v1.1� and�� info:ofi/fmt:xml:xsd:MARC21
> orinfo:srw/schema/1/onix-v2.0�� info:ofi/fmt:xml:xsd:onix
> Is it possible/feasible to agree on� one or the other, or will (must) we
> continue to use the SRU-registry identifiers for SRU and the OpenURL
> registry� identifiers for OpenURL?
> From my point of view we would need to come to agreement (if any) on this
> fast (at least agreement in principle), because of a number of initiatives
> currently in progress, and if we can't then we would just keep using thedual
> systems.
> Comments please.
> (I really do strongly request that we resist the temptation to take this
> thread off-course, like making absurd suggestions that the identifer for a
> schema should be its namespace URI.� If you want to talk about that, please
> initiate a new thread.)
> --Ray