Formal Proof of 'ought' implies 'can'

I have been looking into this problem since posting on ephphatha's blog philochristos.
I must admit I have found nothing satisfactory so far. Deontic logic seems beset with problems. Can anybody help?