CCPP Framework Meeting Minutes 2021 01 26

Agenda

• capgen.py progress
• Transition to capgen.py
• Other business

capgen.py progress

• Working on PR addressing #346 (ignore missing optional arguments)
• Need to discuss requirements for host model to communicate threading information to CCPP suites
  • Suite needs horizontal_loop_begin and horizontal_loop_end
  • Chunked data turns into a flat array for suite-created data
  • UFS provides information on number of blocks, blocksizes, ... - information is scattered (some in host model, some in ccpp-framework)
    • what we have now might work as it is
  • Host model most provide these block details when using blocked data structures, and/or horizontal_loop_begin/horizontal_loop_end when using flat data structures
    • Possible issue with the and, we could end up with two (different) definitions of horizontal_loop_begin/horizontal_loop_end.
  • UFS only does "no blocks at all" (init/final) phases, or the same blocks throughout the run
  • CCPP has a requirement to be able to use different block sizes for different parts of the physics
• CCPP technical documentation - needs to be updated for capgen.py. Who is going to do that? And when?
  • Update as we transition the UFS to capgen.py (repository is NCAR ccpp-doc)

Transition to capgen.py

• Still working on the last IPD cleanup steps, and updating/simplifying the codebase
• Is it still realistic to target a transition to capgen.py by the end of February?
  • Plan on beginning the transition at the end of February - this would mean end of March
  • Need to consider Hurricane Supplemental projects that utilize capgen.py with the UFS (deliverable is end of June 2021)

Other business

• CCPP-MUSICA: invite developers from NCAR side to ccpp-framework meeting on February 16 (Ligia will do)