Lists Home |
Date Index |
I don't think anybody would deliberately create
<x:foo bar="junk" x:bar="junque">
It's likely to appear in the same way as x:foo and y:foo -- the merging in
one document of two vocabularies that share terminology but give it
different usage (or have different contexts for using the same terminology).
Just as a single document, over its evolution and integration of several
subdocuments, may wind up using both x:foo and y:foo, a single element may
evolve to needing both x:bar and y:bar attributes. This is precisely what
XML Namespaces are supposed to enable us to deal with -- Namespaces provide
a path of least resistance and a consistent model for doing so. If it makes
sense for elements, why not for attributes?
This is not to say that given a chance to redesign the document or its
constituent subdocs, you wouldn't change one of the vocabularies or provide
a synonym for y:bar so you could avoid the confusion, or introduce some kind
of transformation in combining the subdocs, but to the extent that you're
locked in to existing vocabularies and have large volumes of documents to
deal with, the difference between bar and x:bar and y:bar might be important
It would be nice to be able to indicate that the difference doesn't matter
and be able to work as Simon suggests. This could be done as a feature of
the processing application (similar to the SAX features) or via a PI, or
some other means.