For now the Simulation is a global singleton. This makes writing inputs for PHARE easy when using PHARE in "batch mode" since the user does not have to deal with objects, just writes constructors.
However if the user wants to manipulate several simulations/runs in one script or one notebook, it's inconvenient. Thus we should :
make sure we keep the convenience of the "batch mode" (no object handles to manipulate, only ctor)
make sure we can have multiple simulations in one python session
maybe hide the samrai hierarchy that we get from create_simulator. The user should not need that, so far it appears in client code because the wrangler needs it, but it's misleading because user generally wants to manipulate the python "PatchHierarchy" (for interacting with data) and not the SAMRAI one.
maybe rename Simulation into SimulationInputs? to avoid confusion with "Simulator"?
For now the
Simulation
is a global singleton. This makes writing inputs for PHARE easy when using PHARE in "batch mode" since the user does not have to deal with objects, just writes constructors.However if the user wants to manipulate several simulations/runs in one script or one notebook, it's inconvenient. Thus we should :