XOM API for Processing XML with Java

[Randall R Schulz <rschulz AT sonic.net>]

Steve,

On Tuesday 31 October 2006 06:40, Steve Loughran wrote:
> On 31/10/06, Randall R Schulz <rschulz AT sonic.net> wrote:
> > Don't forget to add the "implements Iterable" clause to enable the
> > concise iteration syntax.
>
> I'm not so sure about that; I think that's a real design error there.
> Because an object can only have return one iterator, something like
> an element has to decide if it will be iterable over its child nodes,
> elements, attributes or namespaces.

It's probably just a compromise between simplicity and generality in the 
design of the Java 5 concise iteration syntax. You're forced by the 
current definition to pick a single iteration sequence for iteration 
over instances of a class that implements Iterable.

As you show below, you can still iterate over arbitrary iterators when 
there are distinct methods to retrieve them.

And if there is no single sequence associated with a given class, then 
I'd forgo the "implements Iterable" part and just implement the methods 
that return iterators. In that case, I would not name any 
method "iterator()." But if there's a method that deserves to be the 
exalted "iterator()," then I'd "implement Iterable" as well.


> I make my iterators implement iterable instead:
> public Iterator<Node> iterator() {
>         return this;
>     }
>
> This lets you
>
>  for(Element e:root.elementIterator() ) { ... }
> and
> for(Attribute e:root.attributeIterator() ) { ... }
>
> This will probably be considered an abuse of the interface come the
> next edition of Effective Java, but, well, tough.

I don't see the problem with that. There are surely well-designed 
classes out there that can yield more than one sequence for iteration.


> -steve

Randall Schulz