I gave you the same answer last week on the other forum. Anyway… ::)
Oh, I thought you were referring to something else over there. You had mentioned ipfw and MAC based filtering causing multiple passes through the stack, and that confused me. So, are you saying that enabling bridging causes ipfw rules to be created and all traffic getting passed through ipfw AND pf even if there's only L3 rules involved?
If so, that'd be a bit more overhead than I thought!
Similar to your "What is wrong with bridging?" question, I would ask you "What is wrong with routing?"
In a general sense, or in the specific case I referenced?
For my specific, the issue with routing is that IPv4 network broadcasts aren't routed.