RE: [xml-dev] XML feature manifest/profile (was RE:: [xml-dev] Re : determining ID-ness in XML )

Do you have to understand what a feature set does in order to preserve
information related to that feature set? 

In other words, if there are rules for content manipulation that don't
require understanding beyond a syntax level, then there's a hope that
feature information will be preserved by processes that diregard its
semantics.

Ideally, one would hope to describe allowable syntax transformations in some
common formal language than any processor can use. Processes that associate
semantics with syntax could deal with such information at a deeper level;
processes that don't just have to follow the rules for content manipulation.