CCPP Framework Meeting Minutes 2025 07 10

Agenda

• GitHub issues/PRs
• New items
• Updates
• Other business
• notes

---

Attendees: Michael Kavulich, Michael Waxmonsky, Jordan Powers, Dom Heinzeller, Cheryl Craig, RhaeSung Kim, Dustin Swales, Courtney Peverley

GitHub issues, PRs, discussions

• Testing refactor #612

  • Looks like conflicts are resolved, just need reviews/re-reviews now?

• Host cap generation bug with long standard names Issue #664

Drafts/On Hold

• none

CCPP Documentation

• Remove rule that the SDF file name should be the same as the suite name #76
  • This has been brought up a few times, most recently in https://github.com/NCAR/ccpp-framework/issues/663

New items for discussion

• Issue backlog/triage
  • Framework needs to support restart capability #250
  • Framework should be able to insert timing calls #262
  • Add suite keyword to suite definition file? #275
  • Capgen inspection needs to provide more information #276
  • Standardized missing values from CCPP framework to use in CCPP physics #284
  • Set default values for run time options #290
  • To be continued...

Updates from last time

Previous notes

Other business

Meeting notes

Backlog Issues

• #275 - add suite keyword to suite definition file
  • Unclear what the described merge functionality would look like
  • Dom will tackle if he has the chance; otherwise it’ll fall to whichever host model wants it first (probably CAM-SIMA)
• #250 - framework needs to support restart capability
  • Dom - difficult problem; would be nice to have a way to tell the physics which fields it wants to save for restarts
  • Tabling for now (may readdress if we ever get around to a diagnostics object)
• #262 - timing calls
  • Host models can handle timing themselves (timing physics groups and inserting host-specific timing calls temporarily)
  • Dustin - can also break up the sdf more granulary to get more specific timings
• #276 - Capgen inspection needs to provide more information
  • Given existence of the constituents object, this issue is out of date
• #284 - standardized missing values
  • Dustin - similar to parameterized precision
  • Would be nice to unify/have missing values handled by the framework
  • Dom - use case = fractional land mass use missing values
    • Also PBL schemes with only parts of a vertical column active
    • Also cosp
• #290 - set default values for run time options
  • Probably difficult to implement
  • Defaults currently provided by the host (via namelist)
  • Optional “default” metadata attribute?
  • Each scheme has its own namelist?
    • CAM-SIMA currently doing this (handled by host)
  • May hurt portability if defaults are ingrained in the physics somehow
  • closing issue

Next meeting 2 weeks - 07/24