[Mlir-commits] [mlir] [mlir] Make [tensor|memref]::ExpandShapeOp verifier stricter. (PR #181020)

[Andrzej Warzyński]

================

----------------
banach-space wrote:

> So this means that if the shape is static in the output shape list it could be dynamic in the type of the result?

That's something we shouldn't allow. IIUC, this is similar to what we discussed here:
* https://discourse.llvm.org/t/tensor-ops-with-dynamic-sizes-which-behaviour-is-more-correct

Mixing static and dynamic should be a hard error.

https://github.com/llvm/llvm-project/pull/181020