Data formats:
* Understand the password section
* Understand and document included objects
* Understand and document other file types (Data, Sheet, Agenda, etc)

Parser:
* Incorporate the replacement section. Stubs are already in place.
* Add additional file types.

Generators:
* HTML: Verify generated code, to see whether it is HTML-3.2 compatible
* HTML4: Fallback on normal HTML for TextEd files might be replaced with
         true code. But is it worth it?
* Add other targets (RTF would be very nice)

General:
* Out-of-memory segfaults (malloc fails are not caught). Better handling?
* Many common values are returned as int. Perhaps change this to long,
  for better behaviour on very large files on some architectures?
* It should be possible for users to catch parse errors with their own
  routines.
