Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 74.28%. Comparing base (dee0d51) to head (1cacbf0).
Report is 28 commits behind head on main.

@@            Coverage Diff             @@
##             main    #1918      +/-   ##
==========================================
+ Coverage   74.11%   74.28%   +0.17%     
==========================================
  Files         445      448       +3     
  Lines       55454    55744     +290     
  Branches     9121     9190      +69     
==========================================
+ Hits        41101    41411     +310     
+ Misses      11262    11232      -30     
- Partials     3091     3101      +10     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

Should we apply the fix of usingmpicc/mpicxx for the post-merge tests? Otherwise, these are going to fail indefinitely, which doesn't seem all that useful.

Should we apply the fix of usingmpicc/mpicxx for the post-merge tests? Otherwise, these are going to fail indefinitely, which doesn't seem all that useful.

Fair. Reading the Debian message, I'm not too hopeful that there's any change in their approach.

I ended up moving the check into buildutils.py so we have a fallback if system_sundials=default; both resolution versions are now covered in "post-merge" tests.

PS: had to disable the f90 interface for MPI as there were some issues:

gfortran -o build/samples/f90/demo -pthread -Wl,-rpath=/usr/local/lib build/samples/f90/demo.o -Lbuild/lib -lcantera_fortran -lcantera -lsundials_cvodes -lsundials_idas -lsundials_nvecserial -lsundials_core -ldl -lhdf5_serial -lfmt -lyaml-cpp -lstdc++
/usr/bin/ld: build/lib/libcantera.a(CVodesIntegrator.os): undefined reference to symbol 'ompi_mpi_comm_null'
/usr/bin/ld: /lib/x86_64-linux-gnu/libmpi.so.40: error adding symbols: DSO missing from command line

which is the same as reported in #1843. Tried adding FORTRAN=mpifort but that wasn’t resolved correctly:

mpifort -o build/src/fortran/fct_interface.os -c -O3 -fPIC -Ibuild/src/fortran -Isrc/fortran -Isrc/fortran -module build/src/fortran src/fortran/fct_interface.f90
gfortran: error: unrecognized command-line option ‘-module’; did you mean ‘-Mmodules’?

PS: had to disable the f90 interface for MPI as there were some issues:

gfortran -o build/samples/f90/demo -pthread -Wl,-rpath=/usr/local/lib build/samples/f90/demo.o -Lbuild/lib -lcantera_fortran -lcantera -lsundials_cvodes -lsundials_idas -lsundials_nvecserial -lsundials_core -ldl -lhdf5_serial -lfmt -lyaml-cpp -lstdc++
/usr/bin/ld: build/lib/libcantera.a(CVodesIntegrator.os): undefined reference to symbol 'ompi_mpi_comm_null'
/usr/bin/ld: /lib/x86_64-linux-gnu/libmpi.so.40: error adding symbols: DSO missing from command line

which is the same as reported in #1843. Tried adding FORTRAN=mpifort but that wasn’t resolved correctly:

mpifort -o build/src/fortran/fct_interface.os -c -O3 -fPIC -Ibuild/src/fortran -Isrc/fortran -Isrc/fortran -module build/src/fortran src/fortran/fct_interface.f90
gfortran: error: unrecognized command-line option ‘-module’; did you mean ‘-Mmodules’?

Did you try FORTRAN=mpif90? I think mpifort might be assuming it's supposed to be an F77 compiler, where "modules" have no meaning. If that doesn't work, disabling this seems fine, too.

Did you try FORTRAN=mpif90? I think mpifort might be assuming it's supposed to be an F77 compiler, where "modules" have no meaning. If that doesn't work, disabling this seems fine, too.

It's something where you have to check what Fortran is wrapped by mpifort. I think I've got it ...

PS: confirmed that we just have to match gfortran syntax. The PR should be ready ...

After correcting the glitch @bryanwweber noticed, it looks like we're now getting - unrelated - crashes on macOS-13 / Python 13:

One runner crashed with:

[gw2] darwin -- Python 3.13.5 /Library/Frameworks/Python.framework/Versions/3.13/bin/python3
worker 'gw2' crashed while running 'test_reactor.py::TestWellStirredReactorIgnition::test_solve_steady[MoleReactor]'

and another one crashed with:

[gw2] darwin -- Python 3.13.5 /Library/Frameworks/Python.framework/Versions/3.13/bin/python3
worker 'gw2' crashed while running 'test_reactor.py::TestIdealGasMoleReactor::test_advance_with_limits'

The same happened during the merge of #1907, see https://github.com/Cantera/cantera/actions/runs/15888865696/job/44807273791

Another crashing runner is on #1911, see https://github.com/Cantera/cantera/actions/runs/15890473825/job/44811946486?pr=1911

Pinning Python 3.13 to 3.13.4 on macOS 13 resolves failures similar to those documented in #1916.