Nicolas,
Aha! I found how a lattice can tell it comes from a "co" construction.
This can be used to fix all the ills I have been complaining about.
The cost of implementing the coweight lattice of type X as the
weight lattice of type X-dual, is that this little dual must be remembered
and taken in
Nicolas,
I think the classical lattice of an affine coroot lattice should be a
finite coroot lattice.
Coroot lattices should be paired with weight lattices. Yet the code
*prevents* me from
doing this. This is surely a bug.
I don't have a problem with code-sharing and using the "co" isomorphism