I think the goal is to try to leverage our existing stylesheet model's
constraints to build a directed editor, and/or to do a more tightly closed
edit/test/revise loop for an IDE.
However, I think most of our constraints are actually being applied in the
parser rather than in the model... and as I said I don't think there's any
promise that the model is a 1:1 mapping to XSLT source, or that it can be
manipulated by any code other than our own parser/optimizer, or that it
will remain stable enough from version to version that anyone should
consider coding against it.
I think the answer here is an editor equipped with its own understanding of
XSLT syntax... and since construction involves different issues than
validation/parsing (you need to be able to accept invalid documents while
they're under construction) I don't even think one could really leverage
our parser very well to assist in building that.
