[Mlir-commits] [mlir] [mlir][math] Fix the description of the semantics of math.clampf (PR #175012)

[Fabian Mora]

fabianmcg wrote:

> > > I wonder if we should also say that if min > max the result is poison?
> > 
> > 
> > I think we should, I would even argue that we should canonicalize to `ub.poison` if it's statically known, but that's introducing a dependent dialect.
> 
> We don't do it in arith either for the same reason.

We should add it there too, `ub` is likely one of the most fundamental dialects, and it's a lightweight dep, so I don't see why not, but that requires an RFC.

https://github.com/llvm/llvm-project/pull/175012