tqchen opened a new pull request, #638: URL: https://github.com/apache/tvm-ffi/pull/638
## Summary - invert the header dependency so `object.h` depends on `type_traits.h` instead of the reverse - move object-dependent trait specializations into `object_type_traits.h` and move `Optional` traits to the end of `optional.h` - move `details::type_contains_v` into the shared portion of `type_traits.h` and add a low-level error-throw helper so top-level traits no longer depend on `object.h` ## Validation - `pre-commit run --all-files` - `cmake -G Ninja -S . -B build/cpp_tests -DTVM_FFI_BUILD_TESTS=ON -DCMAKE_BUILD_TYPE=RelWithDebInfo -DCMAKE_EXPORT_COMPILE_COMMANDS=ON -DCMAKE_C_COMPILER=/usr/bin/gcc -DCMAKE_CXX_COMPILER=/usr/bin/g++ -DCMAKE_C_COMPILER_LAUNCHER=/opt/rust/bin/sccache -DCMAKE_CXX_COMPILER_LAUNCHER=/opt/rust/bin/sccache` - `cmake --build build/cpp_tests --clean-first --config RelWithDebInfo --target tvm_ffi_tests` - `ctest -V -C RelWithDebInfo --test-dir build/cpp_tests --output-on-failure` - `TVM_FFI_GPU_BACKEND=cuda python3 -m pytest -vvs tests/python` ## Notes - The full Python suite passed with `TVM_FFI_GPU_BACKEND=cuda` because this container has ROCm tooling installed while the active PyTorch GPU stack is CUDA (`torch.version.hip is None`, `torch.cuda.is_available() == True`). The unpinned `test_load_inline.py` auto-detect path chose HIP on this machine and failed before the backend pin. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
