Parser structure (historical background question)

I'm curious how it came about that (at least for SAX) validation was 
incorporated into the parser. It'd seem to make a lot more sense to have 
it as a separate layer that works off a parse event stream. This type of 
approach would vastly increase the flexibility (allowing validation to 
be just another pipeline step) without substantially adding to the 
costs, as far as I can see.

Any comments?

  - Dennis