On 7/22/05, Karen Coyle <[log in to unmask]> wrote:
> So again I ask: is there a reason NOT to use elements
> only?
No
> Would it seem odd to folks who are accustomed to XML? Is it more
> difficult to define in a schema? I'm trying to understand why there
> might be resistance to this method given that we have identified at
> least two advantages (easier translation to RDF, and ability to encode
> items with spaces in leading & training positions, plus multiple spaces).
It's also easier to validate element-based structures in XML Schema
and DTDs than it is attribute based ones. Even RELAX NG -- which has
more powerful support for atttribute-based validation -- has certain
limits.
I have a hunch that the design focus on MODS and MADS is a function of
a) wanting maximum flexibility, and b) not quite understanding the
implications of the attribute-based approach (for enhanced validation,
as well as for RDF).
It *is* easier to just define a smaller number of elements, leave
attributes uncontrolled and/or optional, and not enforce any
constraints on their appearance (zero-or-more patterns).
What I worry about -- and I've said this a number of times over the
past year -- is that the limitations of MODS are being applied to
MADS, and with that will ensure they continuation for many years to
come, ultimately limiting the flexibility of both.
Bruce
|