Re: Recursive replacement

I think I miss understood your message first, so here's a second try...

> If however you replace the outer one first, when you come to replace the
> inner one, it's not there, so is that a merge conflict and an error, or
> do you just silently do nothing in which case commutativity is
> restored, and you get the same result as if you had done the operations
> in the other order.

The idea is that the query is first completely evaluated, so you have
update primitives that include the outer and the inner <a/>. If you then
replace the outer <a/>, the inner <a/> is no longer reachable within the
tree (it's a grandchild of outer <a/>, but that doesn't have a parent
anymore). But that doesn't matter for replacing it, at least not at the
moment.

Martin