RE: How does topological sorting of kind variables really work?