And how reliable is the URL system? I like www.simonstl.com, but I
wouldn't count on it being there in ten years. A central registry, with
version control, and a mechanism for external resources to identify
themselves (like MIME content-headers) seems like a much more reliable
system. You'll still need to update your software periodically, to take
advantage of the latest toys, but at least you can always look in the same
place. Don't need to touch your documents or your schemas to accomodate
new types, either. Keep it out of the documents, in a place you can
control more easily than megabytes of elements.
Identifiers for internal pieces? How about standardized schemas? Sounds
good to me.
Simon St.Laurent
XML: A Primer / Cookies
Sharing Bandwidth (December)
Building XML Applications (January)
http://www.simonstl.com