CCPP Framework Meeting Minutes 2020 03 26

Agenda

Issue 272: Framework should provide required variables to host model

• PR https://github.com/NCAR/ccpp-framework/pull/273
• in cap_gen's auto-generated code, the variable standard names are comments

NRL collaboration:

• need to get them involved more
• since they are using GFS_typedefs, it doesn't make sense for them to transition to cap_gen.py until NOAA does
• NOAA wants to move towards cap_gen.py, although this is not super urgent
• NRL will be at the next meeting (04/02/2020); will use Zoom for that meeting
• almost impossible for NOAA to receive money from the NAVY - look at option to receive funds at NCAR
• on Monday 03/30/2020, there will be an executive meeting at DTC on how to expand partnerships (including NAVY)
• from Jim Doyle: the money has to be spent by the receiving organization at the end of this calendar year (2020)
• status of ccpp_capgen.py: not yet ready to use by NOAA; Steve will ask Alex about specific features that they are interested in

ccpp-framework development:

• Python 3 compatibility for ccpp_prebuild.py coming into dtc/develop with RRTMGP; also going into the UFS public release branch for release 1.1.0
• timeline for transitioning to cap_gen.py at NOAA; needs several prerequisites on the FV3 side
  • for example the removal of the dynamic build
  • need blocking/non-blocking capability in cap_gen.py
  • need unit conversion code in cap_gen.py
  • the GFS_time_vary step will become a timestep_initial step
    • use of GFS_typedefs in physics should work, but they are not portable code (unless they are declared as types in the physics like the RRTMGP internal types are)
    • the RRTMGP commit comes with an improvement that allows ccpp to compile the typedefs and pass them back to the model (instead of the model compiling them)