James Clark wrote:
> Hmm. So would you then split the inference rules for oneOrMore into two,
> one for the attribute case (which would have empty children), and one for
> the children case (which would have empty attributes)?
Yes.
James Clark wrote:
> > I think that the current restriction is extremely hard to understand.
> > Do you really think that the BNF in 7.1 is easy to learn?
>
> No, but I don't think users need to learn it, because it doesn't prevent a
> user doing anything that they might reasonably want to do.
At present, only the BNF prohibits (@a, a)*. I think that this is very
difficult for implementors.
Cheers,
Makoto
