I reverted the changes to SimulationStateItem
and the templated copy functions. There are dummy functions now that simply output the default value based on the CPPType. This can be implemented…
Don't actually need them any more since i only look at requires_single
now, but will keep it in mind for later.
TBH i just added a bunch of common constructors without considering the details, just to get it working.
Yeah that works too. I just wanted to keep it simple, considering we have to do a bunch of type switches elsewhere anyway. Have to get the type idname
from the eNodeSocketDatatype
, then get…
I wonder if is_field_source has to be propagated
It's not obvious from the state struct what gets propagated where, but looking if i'm not mistaken:
requires_single
is propagated R->L to…
I'd be fine with only handling the UI part in this patch (so the field inferencing). These sockets can just be ignored during evaluation for now. I think it would be easier to solve that in a…
I found one correctness issue. The position input here should have a dot in the socket because it's currently a single value but can be a field.
Yeah, the sync_field_states
currently…
You're right, combining states first and then comparing is a lot shorter.