Description

User API & Changelog headlines

• update psi4 and plugins for cmake v4

Dev notes & details

• fix up frontmatter versions/arch
• all at once for hdf5/h5py/adcc/einsums/ambit now (since they're now all built with hdf5 1.14.6)
• show ddx works with qcschema and make other ddx tests actually check answers
• fix up at least psi4 itself (not necessarily all deps) for cmake v4 (and release the practical cmake 3.29 pin I've had for a while)
• fix up plugins for cmake v4 (v2rdm_casscf got extra work)
• updated compiled addons in psi4Config and finished enabling integratorxx checks
• something about the conda env or cmake cache setup, but v2rdm build on Mac (I & S) was adamantly not detecting threads right Performing Test CMAKE_HAVE_LIBC_PTHREAD - Success even though it had earlier during the psi4-core build. maybe this is a little new b/c the unix c-f L2 packages have OMP now (wasn't finding that the same way either). Reproducible on laptop. Resistant to debugging, but moving the detection higher up the file (psi4Config -- really happens in find_package(L2)) fixed it, so running with that.
• we're not using modules, so tell cmake not to scan for them even if std is c++20 otherwise need add'l clang-scan-deps tool at buildtime
• bump optking to v0.3
• accommmodate some deps that haven't had cmake update in a while by forcing a later policy std on them (g2g, libxc, gauxc)
• after torsiondrive work to get a np v2 compatible release, we're now solving environments entirely with numpy v2. (TD was needed with qcf which we like to test with).
• openfermion suppressed in codedeps so that ppl solve for np v2 envs in general. The latest np v1 was build for py39/310/311/312 and np v2 py310/311/312/313 so we don't want openfermion as a choking the env sol'n
• On Windows, when libgcc 15 is present, import psi4 throws ImportError: DLL load failed while importing core: The specified procedure could not be found. It seems like a runtime problem, not a buildtime (that is, v15 ok to have around during compile). So for now, we're evading it. Thank goodness I had successive CI runs last week where one passed, then failed. Having eliminated code changes, it had to be env. I've no idea what's actually going wrong.
• also includes changes found for conda packaging cmake pin, v1.10 prep conda-forge/psi4-feedstock#39

Checklist

• Tests added for any new features
• All or relevant fraction of full tests run

Status

• Ready for review
• Ready for merge