Update validate_rig_output_ids.py

[antirotor]

Allowing different types of objects in out_SET of rig is not working because we are running getShape() function on them explicitly. This will return None for objects where shape is not defined, failing later in code. This is checking for this cases.

[BigRoy]

Just wondering about this validator, not necessarily the changes made to it. Does this do what you intend it to do?

Say you have two assets where the rigs are made as:

/rig/mesh/...
/rig/controls/...

Say both assets have a piece of geometry called: "body_GEO".
They both have: /rig/mesh/body_GEO however they should not! have identical cbId values. Nevertheless you are considering both due to the stripping of namespaces.

This seems to invalidate something that should actually be allowed.

What specific issue are we trying to solve?

---

Or wait... nevermind.
I think the assumption is made that in this particular scene (where the rig is published from) that the Artist has loaded the character's geometry, e.g. referenced - a single time. (Or at least it may be loaded more time, as long as they don't deviate in cbId values between them).

Maybe expand the validator's DOCSTRING that this validator can only capture the error if the original model is also present in the scene outside of the rig, and geometry has not been renamed.

---

By the way, note that this:

pc.ls(type="transform") + pc.ls(type="mesh")

Can be writtten as:

pc.ls(type=("transform", "mesh"))

[mkolar]

So you answered yourself :) Yes you are right that is the intention. It's actually I think almost unaltered from your original implementation in colorbleed config back when we started. It served us quite well till now. We have a client that uses bones as controllers quite often so some flexibility is needed.

That being said, I do think that we'll be going through most of these validations for pype 3.0 to make sure there isn't too much weirdness going on.