Document interrogation

[gagern]

The library offers tools to interrogate the structure of the document, its nodes and attributes. It would be nice to expose this to the JavaScript world. It would be important, though, to make sure that this works nicely in the presence of memory reclamation.

• Every other object (like nodes) should contain an internal reference to the document to keep that from getting garbage collected.
• The document should have a version number. Every modification of the document should bump the version, invalidating all the nodes created prior to the modification. Similar to the fail-fast iterators in Java.

So we are looking at objects with about 3 internal fields: the wrapped node or attribute, the associated document and its generation at the time of creation.