_migrated from Trac, where originally posted by **kohlhase** on 27-Jan-2008 5:19am_ if we have many types (and MathML3/OM3) acknowledge the existence of type systems, should we introduce new roles for types in CDs?