On 10/8/07, James M Snell <[EMAIL PROTECTED]> wrote: > I'm not convinced that we can call it buggy. The stream has to be fully > consumed before the parsed objects can be modified.
That seems kind of lame to me. The incremental parsing is an implementation detail, I don't see why a user of the API should need to know that it's happening, and if it does it certainly doesn't seem like they should have to work around it by doing something like calling clone on that node. That's such a hack. If anything it seems like modifying a node should force parsing to be completed up to that point in the stream. -garrett
