[Mlir-commits] [mlir] [MLIR] Support interrupting AffineExpr walks (PR #74792)

[Uday Bondhugula]

bondhugula wrote:

> Yes, something that shows we interrupt and propagate correctly, I saw we have « code coverage » with the utility, but not « path coverage » for the new behavior I think?

That's right - I think we'll need a test pass. I can think of a simple utility -- for eg. "check whether an affine expression has a modulo in it" and emit a diagnostic. There should be just one diagnostic emitted as a result even in the presence of multiple, etc.

https://github.com/llvm/llvm-project/pull/74792