Coupling and repositories

[HendrikTolman-NOAA]

API style coupling as done with ESMF at NOAA would mostly require the capability of moving model input and output through the interfaces of the three main routines in the present design. Should this be part of the core libraries, or should the initialization, time stepping, and finalization routine become part of a ESMF-style coupling additional repository?

[JessicaMeixner-NOAA]

My comments here: #58 (comment) are also relevant for this issue.

I think it would be difficult to completely remove the nuopc driver used in the ESMF coupling from the code. That being said, being aware that WW4 will be used in coupled models, and making sure you put as much as possible in subroutines instead of in the main drivers - so that those subroutines are shared from the main standalone driver or a coupled driver, will be beneficial to everyone.

The ESMF, NUOPC, and other code however should not live in the WW4 repository itself and most testing might have to be done externally, but having the code itself as part of the repo will be extremely beneficial and make it easier for everyone to maintain.