Re: RDF namespace conventions

> What do we do with the use cases which are prohibited, I wonder.

Your use cases showed a need to reference a schema from a document
instance.

For that you have the choice of schemaLocation, which is designed for
the purpose, and can be relied on to point to a schema relevant to the
document.

Or you can use the namespace URI.  Whatever decision happens about
relative URI references, the namespace URI might be using a uri scheme
that is not dereferenceable, or it might be using a scheme that may be
dereferenced, but returns a file not found error (this is probably the
most likely case) or it may return a file that is not a schema, or it
may return a different schema than the one required for the document
(you can't tell which schema is required from the namespace normally) or
it might, just might, return the right schema for your document.

What I haven't seen is any use case that dictates why the second choice
would ever be considered.

You have mentioned that people are `against URI' I do not detect that at
all but many many people are against using the namespace to locate
the schema. If you move or copy the schema file to some other place you
can  use a new schemaLocation, but you can't change the namespace URI
as then you change the names of all the elements in the file (and most
likely they no longer validate against the schema which would have
specified the original namespace as target.

David

Received on Sunday, 21 May 2000 19:23:15 UTC