The GitHub Actions job "CI" on tvm-ffi.git/junrushao/2026-07-15/dtype-parsing-introspection has succeeded. Run started by GitHub user junrushao (triggered by junrushao).
Head commit for run: 8cef07679cedf4d9e0c7550a534c6dca42aa9b25 / Junru Shao <[email protected]> perf(dtype): avoid repeated classification allocations Architecture: - Reuse class-level type-code sets for Python integer and floating-point classification. - Remove unreachable returns after exhaustive C++ classification switches. Public Interfaces: - None; all constants remain private and existing signatures are unchanged. UI/UX: - None. Behavioral Changes: - Avoid constructing classification sets on every Python property access. - Preserve existing integer, floating-point, and fallback classification results. Docs: - None; public behavior and documented contracts are unchanged. Tests: - Executed: uv run --no-sync ruff check and ruff format --check on the touched Python files. - Executed: uv run --no-sync clang-format --dry-run --Werror include/tvm/ffi/dtype.h. - Executed: uv run --no-sync pytest -q tests/python/test_dtype.py; 77 passed. - Executed: rebuilt tvm_ffi_tests and ran --gtest_filter=*DType*:*DataType*; 9 passed. Untested Edge Cases: - The header cleanup was compiled with AppleClang only; GCC and MSVC were not run locally. Refs: https://github.com/apache/tvm-ffi/pull/668 Report URL: https://github.com/apache/tvm-ffi/actions/runs/29447256944 With regards, GitHub Actions via GitBox --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
