[Mlir-commits] [mlir] [mlir] Add FP software implementation lowering pass: `arith-to-apfloat` (PR #166618)

[Maksim Levental]

================
@@ -60,6 +60,14 @@ mlir::FailureOr<std::pair<mlir::func::FuncOp, mlir::func::CallOp>>
 deduplicateArgsOfFuncOp(mlir::RewriterBase &rewriter, mlir::func::FuncOp funcOp,
                         mlir::ModuleOp moduleOp);
 
+/// Create a FuncOp with signature `resultTypes`(`paramTypes`)` and name `name`.
----------------
makslevental wrote:

I can't think of any other; I think these are basically mutually exclusive:

1. function with name/types doesn't exist (create decl)
2. function with name/types does exist (reuse decl)
3. function with name but not types exists (error)

Can you think of another failure mode?

https://github.com/llvm/llvm-project/pull/166618