What we lose when switching from SGML to XML
============================================

This will be filled as I discover those items, so it's necessary
incomplete.

Many things listed at:
http://www.oreilly.com/catalog/docbook/chapter/book/appb.html#S-DOCBOOKXML

Things that _may_ be good to drop:

- tag ommission
- tag minimisation with other syntax (<tag/content/)

- enumerated attribute minimisation (<article Class="WhitePaper">
  vs. <article WhitePaper>)
 => when a DTD is subclassed for pure restriction, and an attr gets
    restricted from CDATA to an enumeration, allowing such
    minimisation looses the source compatibility we would get from the
    restriction.  (Dis)allowing this should be done on a per-attr
    basis.

Things are are we loose clarity from dropping:

- enumerated attribute minimisation (see above)
- tag minimisation with similar syntax (<tag>content</>)
- case insensitivity (eg. <refsynopsisdivinfo> vs <RefSynopsisDivInfo>)
- no factorisation of element content declaration (clarity of the DTD)


Things that may not be good to drop:

- subdocs


Things that definitely _are_not_ good to drop:

- attribute types: NUMBER, NUTOKEN (others ?)
- content inclusions and exclusions
