Turns out the CCPP stub in directory stub is currently broken in main, following an earlier PR.
This PR fixes the problem in the ccpp_prebuild.py code by removing the hardcoded name of the CCPP data (type ccpp_t variable and replacing it with the actual name from the host model that gets passed to the group caps through the API and suite cap.
I took this opportunity to replace the ugly in-source build of the CCPP stub with an out-of-source build.
And while testing I discovered that the capgen unit tests are also broken ... simple fix in this PR included.
Note that each part of this PR is a separate commit so that individual parts can easily be identified cherry-picked, if necessary (thus this PR should be a merge commit, not a squash merge).
none conducted - someone needs to test this with the ufs-weather-model and the ccpp-scm (I expect no impact) @dustinswales @grantfirl @mkavulich is this something you can help with?
manual testing:
Turns out the CCPP stub in directory
stub
is currently broken inmain
, following an earlier PR.This PR fixes the problem in the
ccpp_prebuild.py
code by removing the hardcoded name of the CCPP data (typeccpp_t
variable and replacing it with the actual name from the host model that gets passed to the group caps through the API and suite cap.I took this opportunity to replace the ugly in-source build of the CCPP stub with an out-of-source build.
And while testing I discovered that the capgen unit tests are also broken ... simple fix in this PR included.
Note that each part of this PR is a separate commit so that individual parts can easily be identified cherry-picked, if necessary (thus this PR should be a merge commit, not a squash merge).
User interface changes?: No
Fixes: https://github.com/NCAR/ccpp-framework/issues/496 Fixes https://github.com/NCAR/ccpp-framework/issues/497
Testing: test removed: 0 unit tests:
tests
passtest
pass system tests: