I can make this change in a future patch, looking at threading more cohesively. For now, the variant that compiled:
They all fall off the end with the if / else clauses-- no return needed, I think.
Modified it to be more explicit that we're skipping the edge domain.
Ahh looks like you're right-- I was confused about has_value being from the optional vs being from the pxr class. Changed.
In this case no-- the conversions don't necessarily match up and it could cause false positives.
Empty Token was unused but defaultTime is not-- I removed empty_token.
Removed-- they were a holdover from when I was grabbing a return value and wanted to use the same name in each scope.
Now that it's an optional, there's a possibility that prim_varying is null, and attempting to access .has_value will cause a crash. I have to leave it like this to handle all cases.
idname
field to ModifierTypeInfo
Any movement on this? Apart from a Compose Matrix node and possibly adding local and world matrix outputs from the Object Info node, this feels done and would be incredibly useful.