[Mlir-commits] [mlir] [mlir] Enhance dimOp fold (PR #187286)

[llvmlistbot at llvm.org]

MaheshRavishankar wrote:

> Is the code you're pointing at running alongside canonicalization? If not then it's not clear to me how that would obsolete canonicalization patterns: they runs together until fixed-point, a separate pass does not compose in the same way.

These are legacy. These should not be in canonicalizations. The other pass runs all the dim resolution patterns in a fixed point.

https://github.com/llvm/llvm-project/pull/187286