Groping toward a solution
Back in February 2002, I published Ents, a dead-simple utility for converting entities to character references and vice-versa. It's trivial code, but it made it very easy for me to convert DocBook chapters into plain-enough XML for me to process directly with ordinary tools.
Unfortunately, by the time most parsers let me touch their results, the entity processing has already worked, or not worked. There are a few tools and techniques for dealing with this.
Elements as entities
Sean McGrath and a number of others recommend using empty elements rather than entity references, and using processing later to turn them into characters or whatever they should be. In many ways it's a less verbose XInclude, requiring more application configuration, but it's plainly a useful option in some cases.
Many parsers provide hooks for entity resolvers. Few people write them; fewer people publish them. They're a useful extension, but typically hard-wired, focused on a particular set of entities, and return an InputSource, not a simple result.
skippedEntity events, etc.
Applications can of course pick up the pieces if they want. (This doesn't always work for attributes!)
OASIS XML Catalogs are focused on URI resolution, getting results from (perhaps unavailable) SYSTEM identifiers and (otherwise opaque) PUBLIC identifiers. They're a useful patch for sharing processing assumptions across systems, though at the level of resource retrieval.
Previous Page <
> Next Page