In message <[log in to unmask]>, Wendell
Piez <[log in to unmask]> writes
>I voted for #3 (IDREF). But then Francois observed what we'd missed, and
>unfortunately, Sebastian (below) is correct. If maintaining backwards
>compatibility is a priority, then I'm afraid they do have to be CDATA.
>Which makes me wonder about that requirement. Taken to the limit, it
>suggests that we could *never* have any new constraints. For an interchange
>DTD this is okay, but for new material (authoring/editorial DTD) it
>obviously has problems. Maybe a parameter entity with a switch, and a
>recommendation to switch it to IDREF for new documents?
That approach would get my vote. Surely TEI folks won't have a problem
with parameter entities ...
SGML/XML and Museum Information Consultancy
[log in to unmask]