I think it would be OK - interpreter selection is a task, the interpreter is a product, and the resolve task uses that product.