This is more of a thought experiment than a request at this point, as it’s pretty involved, but I’d like to at least test the waters with the idea…
I’ve been thinking some more about properly controlling the Python environment our farm tasks are executed in (including Python version, interpreter, etc.) and as it stands, we would likely end up implementing all job types using managed Python interpreters to maintain the functionality and flexiblity we currently have (as opposed to executing code directly in the Deadline slave Python environment).
The sort of pipe-dream I have now is to be able to swap out the Python executable Deadline uses, so we could switch between Python versions if necessary, or even use mayapy or Nuke as the interpreter. With this in mind, one solution from an end-user/developer standpoint seems to be having all or part of the Deadline slave functionality rolled into a Python library to allow for this kind of thing. Now obviously this would require a transition away from a mono-based core (which I’m still holding out hope for), but besides that, what’s the immediate reaction to this kind of suggestion?
Thanks,
-Nathan