[tei-council] FAND faces setback

[Syd Bauman]

> I see how introducing these [empty model.div1, model.div2, etc.]
> classes might help address the concerns of those who want to be
> able to tell from the name of an element what its hierarchic
> position is. I don't see how that helps with the fact that if the
> topmost classes (div0 or div1 as was) are empty you will wind up
> with a non-deterministic content model.
> The second problem seems to me more serious.

I don't think it helps at all in the sense that solving this problem
will still be a pain in the necks of the 'class' subcommittee members
who will try to make this work. But it does help in the sense that if
& when we have a solution, the user will get a system with the
solution in place up front. 

That is, in the current system the element is already present:
- users that want it have no work to do, but
- those FAND sympathizers who want to get rid of 'em have a lot of
  confusing work to do.

In the proposed system, the element is not present:
- users that want it have a small bit of almost trivial work to do,
  and
- those FAND sympathizers who don't want 'em have nothing to do.

Presuming the world is divided 50/50 between FAND sympathizers and
numbered-div-huggers, the proposed system involves a lot less user
effort (at the up-front cost of our effort).

And no, I don't have any proposal in mind for how we're going to make
non-deterministic content models work in the proposed system, but I
think it may be worth considering special-casing the DTD and XSD
generating code.