Yes that seems right. Can you verify that Rail Revolve gives that result even if the rail is on the inside. That’s what I seem to be getting.
It’s worth mentioning that though a perimeter curve may be best for the sweep1 example above, I discovered with a quarter round that it requires an internal rail curve. Perhaps the rail must be on the deepest side of the section curve?
Maybe because, in both cases, you are sweeping along a rail aligned with the deepest part of the section curve?" In my sketch immediately above, the one that fails has the rail adjacent to the narrowest part of the section curve.
(Regardless, I’ve’ found that the test of sweeping success isn’t how it looks in wireframe, but what planes are actually found when displayed in a mode showing faces.)
In other words Pascal, both your attempts are the same.