[james anderson]
> what's stopping from from taking the xml-sdk of choice, locating the place
> where it's serializer determines the document definition's root, and
replacing
> that with the types collected from walking the document instance? that way
you
> get the types as defined - not as used.
>
Nothing's "stopping" this and it may be the way it ends up getting done. I
hoped that someone had already worked out some of the details, that's all.
As you said in a previous post, garbage collectors do something like it, but
I do not get the sense that writing a garbage collector is necessarily
trivial.
More to the point, the pruning will probably need to take some more human
input along the way, which sounds like several passes to me. I am not sure
how far the person involved wants to go in paring down the schema. I
suspect he wants to be fairly draconian about it.
Cheers,
Tom P
