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

[Ian Wood]

IanWood1 wrote:

It might be nice to update the builders to ensure that output shape and the result shape match. It could create a constant when a dynamic value is needed in `outputShape` and use the static int value from the result when the result dim is static.

https://github.com/llvm/llvm-project/blob/31f31dd9d0460834f2ae67247d3886fec4931d6d/mlir/lib/Dialect/Tensor/IR/TensorOps.cpp#L1943-L1952

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