Very briefly, I think your full requirements cannot be met with what the present language has to offer. I’ll try to elaborate in a later post.
No, not at all. The full design that you are trying to implement is a perfectly legitimate use case.
In fact, it is one of the use cases that our Traits proposal was designed to make possible, and that I think cannot be (reasonably) coded using both the present language, and the committee’s proposed generics design (because it requires interoperability between true generic, and object-oriented programming, which both the present language and the committee’s generics lack).
I’ll provide an example (in a later post) of how it could be coded using the Fortran extensions that we propose in our document.