[Mlir-commits] [mlir] [mlir][SCF] Apply inlining patterns with higher benefit (PR #190947)

[Mehdi Amini]

joker-eph wrote:

Is this the alternative we discussed here: https://github.com/llvm/llvm-project/pull/188986 ?

I think we should still be more defensive in `computeTiedSuccessorInputs` by having some sort of assertions to protect against the invalid usage.


https://github.com/llvm/llvm-project/pull/190947