Austin Bingham

I had this mostly working but abandoned it for a later time. It won't require much actual coding, though some careful thought will be needed about things like cwd when...

See [python-anyconfig](https://github.com/ssato/python-anyconfig) as one possible tool to help here.

> I imagine the fix would be to return an empty list in _name_nodes if the node does not match one of the two supported scenarios. Yes, this does sound...

> Then we can run tests within current environment (environment where cosmic-ray was launched from). Why do we need to do this?

Ok, that's reasonable justification for this feature. I would, however, like to simplify the implementation. The use of metaclasses is a bit over-engineered. I think we could replace it and...

> Using plugin system, If there's really a need for many different kinds of workspaces, this is definitely my preference. It's how the rest of CR works, and it seems...

> It is very desirable is the base API is well defined. But it is not the case yet. Good, then we're in agreement :)

Can I close this in favor of defining a plugin system for workspaces?

> don't we need to serialize the nodes between every run? I don't think so. We'll rely on a coverage tool to tell us the correspondence between tests and code-under-test....

> It seems to me this could provide a huge performance win. In my thinking, this is one of the only ways to get mutation testing working in daily continuous...