Le Dimanche 23 Février 2003 09:26, vous avez écrit :

Wouldn't this example too would allow for 2 B's? It may be messy (especially if you progress beyond two elements), but couldn't you just use a choice allowing for only one of an A-B group, an A or a B?

That was also my idea (see attached files), however it is prohibited for ambiguous content model: when the parser reach element A, he can't determine wether it is a single A element or part of AB group.

Reversing A and B order in group won't fix the problem, but anything else that would allows to distinguish them would. Gregor, could we have a bit more context ?