The DITA DTD files are highly modularized and while most XML processors have no problems dealing with nested parameter entities, in production environments this may introduce redundant complexity. In addition, because of the way DTD files are build, topic specializations contain elements from base types that cannot be used; for example, body element cannot be used if the root element is concept. One solution is to normalize the DTD files by resolving all parameter entities and removing their declarations, and removing comments and redundant elements.
Wow Jarno, thanks for doing this. Saves me a ton of time (yes, I’m one of those humans who also like to read the DTDs).
Have you by chance done this with the 1.2 DTDs?