>Hmm, this line of thought may be perpetuating what I see as one of
>the shortcomings of DTDs, namely that the DTD has to describe the
>whole document, i.e. a class of languages. What about partial
>validation/constraints? I think it's important that child-of-DTD
>support compond documents & partial validation. So in the terms above,
>maybe these things define sets of elements and attributes, rather
>than whole documents. -Tim
Also, in the realm of persistant data streams (i.e. TV), there is no way for
DTD to be updated by inlining DTD. I can not add new elements or change
containment rules once the stream has started without shutting down the
broadcast and restarting again.
I guess this sort of falls into the question of "is a document always
finite?". A document that takes an entire day to process is practically
infinite in my opinion.
Don Park
http://www.docuverse.com/personal/index.html