Here are some comments on the containership paper.
First of all, great job! I think this lays out the requirements in an
objective and convincing fashion. I believe Lisa will be bringing this
topic to the attention of the LC SC shortly, and I have pointed her to
this paper to get her started. I think that we ideally need to engage
both groups together on this topic. (Lisa, do you think that this is
something you and I can arrange together?)
Regarding lists, I think it may be desirable for the model to add
list-grouping elements explicitly in at least some cases. First, it's
debatable whether this isn't a "semantic" construct, since it's
manifestly possible to define the semantics of each list. (It's a thin
line... :-) And certainly, in many cases where properties are
explicitly "common to all members of the list", it may very well be that
the properties are really properties "of the list"; these are obvious
cases of "semantic" list groupings.
The grouping section is excellent. It would be great if it could
conclude with a suggestion (or several) on how the modeling approach
could change to ensure proper coverage of these grouping opportunities.
(I fear it isn't trivial.)
Regarding extension and reuse, I would make the same comment as about
the semantics of lists. Who says these packages are not semantic?! :-)
The paper certainly makes it look easy to come up with properly
constructed names that follow our naming rules. If the question is
really "Are such groupings specific to the XML syntax binding or might
they have utility in other bindings?", I don't know the answer. But I
do know that groupings have been treated as "legitimate" semantic
constructs in every SGML and XML model I've ever done.
A minor nit: The next version of the paper should ideally use the new
template and should have numbered lines for ease of citation.
Eve
--
Eve Maler +1 781 442 3190
Sun Microsystems XML Technology Center eve.maler @ sun.com
