CCPP Framework Meeting Minutes 2020 01 09

Is a single _init routine enough to handle restart complexity?

• Do we need another phase to distinguish between coldstarts and warmstarts?
• Should physics manage restart information for the suite?
• Knowing whether a field is required as input for restarts (but not for coldstarts, for example) is valuable information

Transitioning to cap_gen.py with UFS - possible until end of June 2020? Stumbling blocks:

• handling of blocked data structure and accessing all blocks at once
• unit conversions
• build system
  • this should be easier now that we have an integrated cmake build
  • but uncertainty about the future of the dynamic build; keeping it would come with an increase in costs and maintenance, and we would loose information on the data flow (which fields are read/written etc.)
  • first step: get metadata information solely based on suite definition files and search directories
  • currently ccpp_prebuild.py does the opposite, it reads all available metadata and throws away whatever is not needed

Chemistry requirements update:

• runtime-configurable constituent arrays
• needs to be at the same time when parsing/processing namelist information
• could produce a CCPP metadata file (or a Fortran namelist file) with cap_gen.py

DTC needs to update the ccpp-framework frontpage (Readme.md)

What is the status with renaming ccpp-framework?