Andy says:
> I think this was suggested already on the mailing list.
> IMO, if Xpath is supported, it's a much more efficient encoding
> than what we have now. But I don't want to make Xpath mandatory.
To use Randy's terms there are two features:
(1) Specifying the path to the starting point in the tree.
(2) Specifying the set of instances within tree.
From a requirements perspective we need to do (1). I think we also need to do (2).
If we don't do XPath then how can we satisfy requirement (2)?
XPath seems like the obvious choice. And, as I have stated before, we *may* want to look at trimming XPath down, but that has its downsides as well.
> Andy