> Matthew, the cse pass seems to have uncovered an x86 codegen bug. With cse on, > count-graphs compiles fine, but when run produces a seg fault. I think it is a > x86 codegen bug because it's fine when compiled -native false and it passes the > CPS type checker. I know that's not rock solid proof, but I think it's pretty > good indication. Very probable. I'll look into it.