XOM API for Processing XML with Java

[Elliotte Harold <elharo AT metalab.unc.edu>]

If anyone else missed it, this refers to 
https://lists.ibiblio.org/sympa/arc/xom-interest/2005-August/002515.html 
which for some reason didn't reach me the first time.

The reason there are both a copy constructor and a copy method is that 
the constructor is type-safe and the constructor isn't. i.e. it's 
because Java 1.4 and earlier don't support covariant return types. To be 
able to copy any node, I have to put the copy method in Node. However 
the copy method in Element then must return Node, not Element. The copy 
constructor however always gives the right type. It is a minor ugliness, 
I agree. It wouldn't be necessary if Java supported covariant return types.

The more serious issue is how to handle the copy method in your Element 
subclass. I think what you're missing is that shallowCopy is not meant 
to be invoked by user code, only overridden. The regular copy method 
calls shallowCopy. The purpose of shallowCopy is just to get the name 
and namespace right. So in your example you should override 
shallowCopy() which does not copy any attributes or children. It simply 
returns an empty element. The  you would invoke the copy method 
inherited from Element. This will call your shallowCopy() method and 
then add all the children and attributes. In other words, shallowCopy is 
used by copy. It does not replace copy.

Your copy method might work, but since it uses recursion it's vulnerable 
to stack overflow. The algorithm used in nu.xom.Element is non-recursive 
and not limited by the stack size.

--
Elliotte Rusty Harold  elharo AT metalab.unc.edu
XML in a Nutshell 3rd Edition Just Published!
http://www.cafeconleche.org/books/xian3/
http://www.amazon.com/exec/obidos/ISBN=0596007647/cafeaulaitA/ref=nosim