The DFG originally used static predictions to drive speculation and optimization. But now the DFG has value profiles, rare case profiles, and the old JIT's structure stub infos and method call link infos to dynamic predictions, which have been found to do a better job of driving speculation and optimization. At this point, the static prediction code is just getting in the way and should be removed.
Created attachment 108619 [details]
Landed in r95930.