On Tue, Nov 25, 2008 at 20:08, Jonathan Rochkind <[log in to unmask]> wrote: > There may be reason to suggest or prefer or constrain what kinds of > identifiers can be used, for consistency. It's hard to write software that > can expect _any_ kind of identifier at all. Well, that depends on the premise for working with identifiers as opposed to just put them in there. For example, if your systems were designed with persistent identifiers (with simple policies and merge rules) it *doesn't actually matter* what the identifier itself is, you can still use them for identification precision. Of course, having a policy of guaranteed resolvement helps, too. > But if that _should_ be done, I think it should be done in the 'usage' end > (RDA itself, not that I believe RDA itself touches this with a ten foot > poll), not at the schema/vocabulary level that Karen is working on. I agree. May I suggest simple rules and policies at the RDA level, and leave resolving and usage to applications? For a nice introduction on where I'm trying to push things, see this PDF from OASIS ; http://www.oasis-open.org/committees/download.php/3050/pubsubj-pt1-1.02-cs.pdf Regards, Alex -- --------------------------------------------------------------------------- Project Wrangler, SOA, Information Alchemist, UX, RESTafarian, Topic Maps ------------------------------------------ http://shelter.nu/blog/ --------