The DM tools have lots of limitations.  Be aware of the following problems
at least, for which there is the intention at least to remedy the
deficiencies.

The tools are case-sensitive in places where they shouldn't be.  They
discover "differences" in data sets:

o   when the syntax is case-independent

o   when attribute types have different representations (e.g. objectclass
        and objectClass)

The tools work better if DM data you generate uses the short form attribute
type names: e.g. cn instead of commonName.

The tools work better if you specify multi-valued attributes as a number of
separate attribute type-value pairs.  Prefer

  info=first bit of info
  info=second bit of info


over

  info=first bit of info & second bit of info

The tools could be optimised to attempt less changes.

Inherited attributes (with their multi-line syntax) half work - they cope
with "always" inherit but not "default".
