RE: Content constructors and sequences

> Mike, could you expand on this comment:
>   If a variable $v references such a node,
>   then when the node is added to a document, the value of
>   count($v/ancestor::*) is going to change.
>
>
>    .... My worries may be misplaced (the data model suggests
> that when a
>    node is added to a tree,
>
> ah maybe you already answered my question, but I'd still be interested
> to know what adding a node to a tree would mean (is this
> perhaps related
> to the xquery update question?

Yes, it is defined in the data model to work in terms of copying. So
conceptually, you create a sequence of child nodes, then you add these nodes
to a new parent element or document, which is done by creating copies of the
child nodes. So in principle, mutability shouldn't be a problem if you
follow this model: any variables are left pointing to the child node as it
was before you made the copy, and the value of count($v/ancestor::*) doesn't
change.

So another way of expressing my worries is, that if you actually have to
make a physical copy, and give each node a new identity, then it's going to
be horribly expensive. Perhaps there's an implementation solution based on
reference counting - you only need to actually copy a node if there are
variables that reference it - but I'd like to be convinced.

There are other problems with rootless nodes, for example what are the
in-scope namespaces for a parentless attribute node?

I'm not saying it doesn't work - just that the complexities haven't been
fully explored yet.

Mike Kay


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list