Should I implement a node or a deformer?

[Anonymous]

I'm writing a plugin where a cage drives the deformation of a polygonal mesh. Now I have seen this implemented both as a deformer, with the cage as the only input to the deformer node, and as a normal node with 2 inputs, the cage and the mesh to be deformed, and a single output, the deformed mesh. I'd like to hear your opinions on why people implement it either way, pros and cons. Thank you.

[Anonymous]

I think "Edit > Delete by Type > Non-Deformer History" could react in a different way, depending how you implement it.

Advantage of a node where you have two inputs (cage and mesh) could be in cases were you want to reuse the cage. The cage could be set as input for several nodes. But you could realize this also with a deformer (you can add your own attributes, connections, ...)

In my opinion, based on your short description, a deformer seems to be the right choice in your case.