[
Lists Home |
Date Index |
Thread Index
]
>can you simply check that the entity doesn't start a declaration and not
>finish it? This would be similar to the implementation of VC: Proper
>Declaration/PE Nesting [2]? It seems that the only two cases where the
>VC and WFC are not redundant is that the WFC adds the additional
>constraint that a PEReference (the actual ref, not the replacement
>value) cannot be started in one entity and ended in another and that a
>conditional section cannot be started in one entity and ended in another. *
It seems to me that the following would only violate the validity
constraint, since it doesn't have a PE reference in a DeclSep:
<!ENTITY % p "ANY> <!ATTLIST foo ">
<!ELEMENT foo %p;
bar CDATA #IMPLIED>
But I don't think I treat these cases differently in my parser.
Several of these constraints were added in errata to clean up things
that were not properly constrained in the first edition, and it would
not surprise me if this has led to redundancies.
-- Richard
|