CCPP Framework Meeting Minutes 2020 08 27

Agenda

• transition to cap_gen.py
• ver="N" in suite definition file
• other business

Transition to cap_gen.py

• DTC timeline and resource chart (https://docs.google.com/spreadsheets/d/1XbSF1ejYcPMrVtZtk1OSiMX-vI5srLkwEUhyAirGr8k/edit?usp=sharing)
• things that need to happen for transitioning UFS to cap_gen.py (https://docs.google.com/document/d/1M3mCO3ESR9OJqLyUbQf-ahJtmfmGKilZlWaBTxsfMRk/edit?usp=sharing)
• CCPP dependencies PR is ready (waiting for our turn in the queue, a few weeks after Dom's vacation; need to keep updating)
  • PRs for ccpp_prebuild.py: https://github.com/ufs-community/ufs-weather-model/pull/180
  • issues: https://github.com/NCAR/ccpp-framework/issues/308 and https://github.com/NCAR/ccpp-framework/issues/318
• "active" attribute for GFS_typedefs.meta is ready (probably will be combined with the CCPP dependencies PR; need to keep updating)
  • PRs: https://github.com/ufs-community/ufs-weather-model/pull/182
• Xia working on assumed-size array declarations
  • several test failures and crashes
  • Dom needs to get started on this, use full compiler debug flags
• Grant is working on removing the GFS DDTs from the radiation routines
• Julie ran into an issue when using the new (?) cap_gen.py Fortran parser; what used to be a nice error message is now an cryptic Python exception (internal capgen error).
• Ligia and Grant started working on the consolidation of standard names; comparing standard names in GFS_typedefs.meta with CF convention, dictionary of standard names, ... - https://docs.google.com/spreadsheets/d/1kWDC7_2sagm-PTP_tXaGyI3O3sJFK8FY9FElQbb6V3o/edit#gid=613941307
  • need to get basic definitions of "bottom", "top", "lowest", "surface", ... right
  • when is it appropriate to have the name of a scheme in the standard name?
    • cleanup work (finding duplicates)
    • different units? can use unit conversion for this
    • laziness of developers; code reviewers need to be stricter
    • be as general as possible but as specific as necessary (i.e. use process names instead of scheme names?)
  • several namelist options in the UFS have scheme names, but there is no rule (prefix/suffix)

ver="N" and lib="ccppphys"" in suite definition file

• ver should be version and correspond to the version of the xml schema; allows for backward compatibility
• should not be updated all the time to match ccpp-framework version (only if xml schema changes)
• ccpp_prebuild.py does not use the xml schema version at all; cap_gen.py does, starting with xml schema version 1
  • run xml schema checker from cap_gen.py on FV3/ccpp/suites*.xml
  • replace ver="N" with version="1"
• remove lib="ccppphys" entirely; cap_gen.py should complain about this attribute
• add this to the "transition to cap_gen.py" tasks
• capgen runs xmllint against a versioned schema.

xmllint --schema <ccpp_framework>/schema/suite_v1_0.xsd --noout <root>/ccpp/suites/suite_FV3_GFS_v15.xml

Other business

• removal of IPD - meeting this Friday, but not yet accepted by EMC partners