How to transfer knowledge of a code performance to its dual?

Question

Suppose I have a binary linear block code $C$. If I know it's weight distribution, the MacWilliams identity will give the weight distribution of its dual $C^\perp$. Now suppose I have some other important data on $C$, in particular, let's say I have its minimum distance $d_{min}$, bit error probability $P_b$ and codeworde error probability, $P_w$; the last two for example would be functions of signal to noise ratio for an AWGN channel. How do I get $d_{min}(C^\perp)$,$P_b(C^\perp)$, $P_w(C^\perp)$ from $d_{min}(C)$,$P_b(C)$, $P_w(C)$?