[Mlir-commits] [mlir] [mlir][scf] Uplift `scf.while` to `scf.for` (PR #76108)

[Mehdi Amini]

joker-eph wrote:

> I'm not sure if making this a canonicalization is good idea. I can imagine lowering pipeline looking like scf.for -> scf.while -> cf and such canonicalization will actively fight against it.

I don't quite see the issue though: that just means that such lowering cannot run a canonicalization in the middle. That seems expected to me though, in particular if someone design a lowering with "in-dialect" intermediate steps (like scf.for -> scf.while).


https://github.com/llvm/llvm-project/pull/76108