GHC does not rewrite in FunTy, part II
@rae observed that GHC failed to rewrite kinds in FunTy
in #19677 (closed), and he fixed the issue in MR !2477 (closed) by ensuring that in the rewriter, when rewriting FunTy
, we also rewrite the kinds. See GHC.Tc.Solver.Rewrite.rewrite_one (FunTy ...)
.
However, there are other places in the compiler where we rewrite a FunTy
, but which have not been adapted:
-
GHC.Core.FamInstEnv.normalise_type
, -
GHC.Tc.Solver.Monad.breakTyVarCycle_maybe
.
It seems to me that we should be consistent and do the same thing in these other places.