On Nov 29, 2004, at 10:16 PM, Karen Coyle wrote:

> The problem with all of these solutions is that they break down
> because the underlying metadata can have variations that then affect
> the creation of the identifier, but in some cases the identifier is
> the same in different systems. The advantage is that the identifier
> can be derived from the metadata, therefore there is a chance that
> different systems will derive the same identifier (which isn't true
> with arbitrary IDs).

What do you think about my idea of a flexible system drawing on the
info uri thing?  E.g.:

        <citation><biblioref linkend="info:doi/10.1000.10/12345"/></citation>