I'm not a expert here, but i tend to use as a rule of thumb something like

If it's a value used to identify nodes, use as attributes
if you're going to manipulate that data, use as child nodes.

On 9/28/05, Soren Kuula <dongfang@xxxxxxxxxxx> wrote:
> Hi,
>
> One of the ideas of using XML at all, instead of just cooking your own
> small language, is that reliable and unambigous parsing comes for free.
>
> With
>
> >
> > <problem
> >    models="long list"
> >    software_releases="long list"
> > >
>
> you are throwing out some of the advantage -- you again have to parse
> manually, probably introducing a bug somewhere.
>
> > instead of
> >
> > <problem>
> >    <models>
> >       <model>...</model>    (many times)
> >    </models>
> >    <software_releases>
> >       <rel>...</rel>    (many times)
> >    </software_releases>
> > </problem>
>
> This is much nicer in that respect, and if you use XSL you'll probably
> have something working much sooner. That's a real performance
> consideration too :)
>
> Soren