Dear Martin,
I fully agree that things get more involved for F_{omega,sub,rec}. I haven't
been clear enough in this repect.
Is it correct that the metric approach guarantees a unique fixpoint X for
every type equation
X = T(X,\vec{Y})
where T(X,\vec{Y}) is not simply a variable?? If so I would withdraw all
my complaints.
Thomas