Re: Parameter entities vs. GI name groups

At 2:16 PM -0500 6/20/97, Eve L. Maler wrote:
>Having parameter entities for just (1) complete declarations (modules), (2)
>complete model groups (nearly equivalent to GI name groups), and (3)
>keywords for marked sections (switches) greatly impoverishes the set of
>useful PEs.  You don't have PEs for:
>
>4 Individual attribute specifications (common attributes)
>5 Sets of elements with which you build up repeatable-OR content models
>  ("classes" and "mixtures" in Jeanne's and my methodology)
>6 Other content model subgroups
>7 Other things that are less needful, but still useful
>
>If I can't have at least 1 through 5, I can't get away with doing
>high-quality, complex production DTDs.  If it's just too complicated to
>spec and build the capability for 4 and 5, which I'm willing to believe for
>V1.0, is it that useful to have 1 through 3?  I'm not sure.
>
>	Eve

Eve is correct, this proposal preserves compatibility at the expense of
utility. PEs are the universal bandaid for the generalizations that DTDs do
not let us express: commonalities of content model, attributes, context of
ocurrences, element classes and so forth. If XML is to be useful for
content on its own, we need these generalizations in our DTDs. We faced
this once and made the right decision in terms of power. We should not back
off now.

We already have HTML, we need something more than another delivery-only format.

  -- David

  -- David

_________________________________________
David Durand              dgd@cs.bu.edu  \  david@dynamicDiagrams.com
Boston University Computer Science        \  Sr. Analyst
http://www.cs.bu.edu/students/grads/dgd/   \  Dynamic Diagrams
--------------------------------------------\  http://dynamicDiagrams.com/
MAPA: mapping for the WWW                    \__________________________

Received on Monday, 23 June 1997 14:40:23 UTC