[Mlir-commits] [mlir] [mlir][ODS] Add `OptionalTypesMatchWith` and remove a custom assemblyFormat (PR #68876)

[Cullen Rhodes]

================
@@ -263,9 +265,8 @@ def Vector_ReductionOp :
                          "::mlir::arith::FastMathFlags::none">:$fastMathFlags)>
   ];
 
-  // TODO: Migrate to assemblyFormat once `AllTypesMatch` supports optional
-  // operands.
-  let hasCustomAssemblyFormat = 1;
+  let assemblyFormat = "$kind `,` $vector (`,` $acc^)? (`fastmath` $fastmath^)?"
----------------
c-rhodes wrote:

the current format of `fastmath<flag>` (vs `fastmath <flag>`) can be preserved with the suggestion. This is also consistent with other ops that use fastmath flags so I think it's worth preserving.

```suggestion
  let assemblyFormat = "$kind `,` $vector (`,` $acc^)? (`fastmath` `` $fastmath^)?"
```

https://github.com/llvm/llvm-project/pull/68876