Repository navigation
Verify green after full ecosystem rollout - #75
Merged
Merged
Conversation
SCF#74 re-ran test_cmake_build against the newly-fixed nwchemex-scf 0.0.61, but still failed with the same "nwx::simde link interface contains nwx::chemist but the target was not found" error. Diagnosis: `pip install nwchemex` also transitively installs nwchemex-integrals, nwchemex-chemcache, and nwchemex-nux -- three more packages, never part of the original 6-repo rebuild list, that each bundle their own stale pre-rollout lib/cmake/simde (and other transitive deps') Config files. Any one of them installing after nwchemex-simde clobbers the correct config with the old broken one. All three have now been rebuilt against nwxcmake 0.2.3 and verified correct (NWChemEx/Integrals#174, NWChemEx/ChemCache#101, NWChemEx/NUX#15 -- nwchemex-integrals 0.0.68, nwchemex-chemcache 1.1.25, nwchemex-nux 0.0.22 all confirmed to declare find_dependency(chemist) and find_dependency(pluginplay) instead of the bare stub). This empty commit re-runs test_cmake_build to confirm the collision is finally gone.
The previous run raced PyPI: its jobs started ~70s after 0.2.4 published, before the index propagated, so they installed 0.2.3 and hit the same pybind11 duplicate-target error NWXCMake#89 fixes. pip resolves 0.2.4 now, so this run gets the fix.
NWXCMake#90 adds Development.Embed/Python::Python to test executables built by catch2_tests_from_dir, gated on BUILD_PYBIND11_BINDINGS. That supplies the libpython these C++ test exes were missing: ~622 undefined Py* at link on Linux, and on macOS a clean link that died at startup on "dyld: symbol not found in flat namespace '_PyBaseObject_Type'". Confirmed pip resolves 0.2.5 before pushing this time -- the previous re-trigger raced PyPI's index and silently reused the old version.
Smoke-tests the new macos-14/appleclang dev_matrix leg, which no CI job in the org has run yet. The Linux legs are still expected to fail until the ecosystem wheels are rebuilt on manylinux_2_28. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
All 11 ecosystem packages have been republished from manylinux_2_28 with the new libstdc++ ABI, and with NWChemEx/.github@620e081 so the macOS wheels no longer vendor a second OpenMPI. Verified on the published artifacts: the three symbols this leg could not resolve are now defined with new-ABI manglings -- _ZN3wtf4rtti7detail_10TypeHolder11m_registry_B5cxx11E _ZNK10pluginplay13ModuleManager2atERKNSt7__cxx1112basic_string... _ZN10pluginplay7utility13generate_uuidB5cxx11Ev The nwchemex metapackage itself could not be republished (PyPI rejects the now-compiled wheel at 124-132 MB against a 100 MB project limit, a failure predating this work), but 0.0.44 on PyPI is pure Python with no compiled artifacts, so it carries no ABI and its >= pins resolve the rebuilt wheels. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
All ecosystem packages have now been rebuilt with: - manylinux_2_28 / new libstdc++ ABI - no vendored OpenMPI in the macOS wheels (.github@620e081) - pybind11 3.1.0, internals v12, matching what a local build resolves - the TensorWrapper concept-mangling fix (TensorWrapper#260) Verified on the published wheels: weaklytypedfloat 0.0.33, parallelzone 0.1.46, pluginplay 1.0.67, tensorwrapper 0.0.115, chemist 1.3.52, simde 0.0.83, integrals 0.0.70, chemcache 1.1.27, nux 0.0.24, scf 0.0.63 all carry __pybind11_internals_v12. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Contributor
|
🚀 [bumpr] Bumped! |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
test_cmake_buildagainst the fixednwchemex-scf 0.0.61but still failed with the same "link interface of nwx::simde contains
nwx::chemist but the target was not found" error.
pip install nwchemexalso transitively installsnwchemex-integrals,nwchemex-chemcache, andnwchemex-nux--three more packages (never part of the original 6-repo rebuild list)
that each bundled their own stale, pre-rollout copy of
lib/cmake/simde(and other transitive deps'). Any one of theminstalling after
nwchemex-simdeclobbers the correct config withthe old broken stub.
correct: Rebuild against nwxcmake 0.2.3 Integrals#174 (0.0.68), Rebuild against nwxcmake 0.2.3 ChemCache#101
(1.1.25), Rebuild against nwxcmake 0.2.3 NUX#15 (0.0.22) -- each now declares
find_dependency(chemist)/find_dependency(pluginplay)instead ofthe bare stub.
test_cmake_buildand confirm the collisionis finally gone.
Test plan
test_cmake_buildgreen on all four legs (Linux+macOS x clang-18+gcc-14)🤖 Generated with Claude Code