[Mlir-commits] [mlir] [mlir][Affine] Handle null parent op in getAffineParallelInductionVarOwner (PR #142025)

Han-Chung Wang llvmlistbot at llvm.org
Thu May 29 14:04:26 PDT 2025


hanhanW wrote:

> I sympathize that it is hard to provide a repro, but we still need to be able to test behaviors upstream.

I can try, but I'm not sure if I'm able to do it. The other way to think about the change is that it has a safer check for the method. We already check if it is null in the above check, and it is adding an additional check.

I don't find direct test for this method in the first commit: https://github.com/llvm/llvm-project/commit/3e497a114730cebdf96f3556af081d876b6ac8b3 Maybe I can construct a test using c++, but it looks a little over-indexing on testing to me. 

https://github.com/llvm/llvm-project/pull/142025


More information about the Mlir-commits mailing list