[jdom-interest] partial parsing?

[philip.nelson at omniresources.com]

> could bloat up to 10 times its current size.  Now I know that memory
> prices are going down and I know that I can increase the 
> virtual memory
> partition a good bit on the system, but after a while, something is
> going to explode.

We can improve on this but not eliminate the problem.

> 
> What I was wondering is, is it possible to 'tweak' JDom so 
> that it does
> relatively lazy builds of trees?  Or will this require a 
> major rewrite?

We believe it will not require a major rewrite.  The architecture now allows
for different parsers and subclasses of the core elements so it should be
possible.  We have all wanted this from the beginning but it will not get
underway until 1.0 is released I think.

> 
> If its easy, I'd like to support this for the various users.  If its
> going to involve a great deal of work, then I'd like to start
> discouraging people on my end from mandating anything thats going to
> hammer on JDom that much right now.

For right now, pretty much all tree based apis are a problem.  Some do
better than others, depending on whether or not you actually have to visit
every node or not.