In NOCMODL, the voltage is (when vectorized) handled via an instance specific copy. While in CoreNEURON, this is never done, which why the field is named v_unused.
This PR always uses v_unused. The reason we can't (easily) use the CoreNEURON variation is because sometimes, e.g. CVODE we need to register callbacks, those callbacks have a fixed signature, we can't simply pass through the voltage as an argument, instead we need to retrieve it, e.g. from the instance data.
In NOCMODL, the voltage is (when
vectorized) handled via an instance specific copy. While in CoreNEURON, this is never done, which why the field is namedv_unused.This PR always uses
v_unused. The reason we can't (easily) use the CoreNEURON variation is because sometimes, e.g. CVODE we need to register callbacks, those callbacks have a fixed signature, we can't simply pass through the voltage as an argument, instead we need to retrieve it, e.g. from the instance data.