search

PCMSolver / pcmsolver

An API for the Polarizable Continuum Model
http://pcmsolver.readthedocs.io/
GNU Lesser General Public License v3.0
32 stars 21 forks source link

Release/1.3.1 #200

Open foxtran opened 1 year ago

foxtran commented 1 year ago

Description

This PR updates release/1.3 branch for passing compilation with novel glibc. I have updated old Catch v1.11.0 to v2.13.10, then fixed the compilation of one of the tests.

Motivation and Context

Fixes #197

How Has This Been Tested?

cmake -H. -Bbuild -DCMAKE_BUILD_TYPE=Debug
cd build
make -j $(nproc)
ctest . --parallel $(nproc)

with gcc version 11.3.1 20220421 (Red Hat 11.3.1-2) (GCC) on Fedora Linux 35.

Unfortunately, I got the error described in #159.

Types of changes

Status

robertodr commented 1 year ago

Thanks for this PR! I cherry-picked #196 into it so it ends up in the release. I ended up having to deactivate the infamous green_spherical_diffuse test. With the same exact environment as in CI, I cannot reproduce the failure locally. I note that it fails only for the logarithmic dielectric profile. I want to figure out how to fix #198 as well and then cut the release.

foxtran commented 1 year ago

@robertodr, thanks!

Let me test this release before its official release. I got some configuration errors with the Intel compiler's specific builds.

robertodr commented 1 year ago

I haven't had access/used Intel compilers in a while, so I'd appreciate if you could double-check. I think the oneAPI series of compilers will work without problems though.

loriab commented 1 year ago

Please include #203 infrastucture changes in v1.3.1 .

If you want to test Intel compilers in CI, can get the icpx ones via https://github.com/evaleev/libint/blob/0eca57e412cd93c72c01fa0173698f9a082b475b/.github/workflows/cmake.yml#L290-L313 . I guess still use gnu for fortran?

foxtran commented 10 months ago

@robertodr, there is some issues with novel, LLVM-based Intel compiler. Fortunately, they did not come from this PR.