CCPP Framework Meeting Minutes 2021 04 13

Agenda

• capgen.py progress
• Transition to capgen.py
• Process for creating a master cross-lab standard name resource
  • Do we want to have standard name aliases? What to do about wind?
• Other business
  • Cancel 2021-05-04 meeting?

capgen.py process

• Dom created a few issues when using Julie's version of the parser
  • Fortran parser must report error in dimension string correctly (https://github.com/NCAR/ccpp-framework/issues/362) - not urgent
  • Fortran/metadata comparison incorrectly reporting bug when comparing Fortran variable declarations and metadata for derived types (https://github.com/NCAR/ccpp-framework/issues/360) - slightly more important to fix
  • Fortran parser not recognizing nested subroutines/functions (https://github.com/NCAR/ccpp-framework/issues/359) - slightly more important to fix

Transition to capgen.py

• Dom is working on the PR that fixes issues with dimensions, intents etc using Julie's version of the parser
  • https://github.com/NCAR/ccpp-physics/pull/611/
• Weiwei is working on the Python script to update the standard names in the ccpp-physics metadata files based on the standard name spreadsheet (this will go to a separate branch); will use the same script to create PR to standard name repo

Process for creating a master cross-lab standard name resource

• Issue of wind: meridional_wind, zonal_wind, x_wind, y_wind, ...
  • Many models don't have meridional_wind or zonal_wind, and what are those over the poles?
  • x_wind, y_wind are more general
  • FV3 does convert winds to meridional/zonal winds before running physics
  • Do we want to have standard name aliases? Only if the meaning is identical, i.e. no translation necessary?
  • Let's wait and see what happens when we talk to the physics developers
    • Physics basically cares about the wind speed, not the direction (dycore cares)

Other business

• Chemistry in UFS prototype 7 (one of the candidates for GFS v17) requires calling chemistry between physics
  • As usual this is urgent and cannot wait for the capgen transition
  • Need to split physics group into surface_physics and atmosphere_physics (after GFS_surface_generic_post)
  • Need to introduce persistent GFS_suite_interstitial DDT and move interstitial variables over from GFS_interstitial DDT, as few as possible (to reduce memory footprint)
  • currently bruteforce approach, move everything: https://github.com/NCAR/ccpp-physics/pull/615 (includes two other PRs for easier merging later on)
• Rename "master" branch ... to "main" ?
  • NCAR is doing this for CAM, it is not very difficult and GitHub has resources to help with this process
  • It seems that there is no "forward" when trying to checkout the old name (unlike when moving repositories)
  • What happens to open PRs? See https://github.com/github/renaming
  • To consider: also rename variables such as "master" to "root" for MPI codes
• Use of DDTs in CCPP: do we need to clarify our rules/requirements for using DDTs, in particular when it comes to chemical constituents?
• Cancel May 4 meeting