rorth wrote: > Thank you for the PR, imo this is a much better approach than the previous > one.
Thanks. The previous approach was just a minimal fix (hack?) to avoid breaking the build in a specific situation. This one (if completed) should handle all known issues with Python bindings testing. > As I mentioned before, I'm not familiar with the testing infrastructure, but > insofar as it works, this looks good to me. It will certainly need more testing in situations currently covered by `RUN_PYTHON_TESTS` that I known nothing about and probably cannot test myself. > Do note that the libclang CI should also be adapted: > `.github/workflows/libclang-python-tests.yml` The target it uses doesn't > exist anymore, and the paths that trigger the workflow should also be changed > to remove the old path and include `clang/tests/bindings/python`. You can now > also remove the trigger on `clang/CMakeLists.txt`, which should make this run > far less often on unrelated PRs. Ah, I see. I've made the obvious adjustments locally, but am at a bit of a loss how to handle `build_target` now. Doing this manually would be an invocation of `bin/llvm-lit clang/test --filter=bindings/python` or some such. Maybe a new implementation of the `check-clang-python` target needs to be introduced? Don't know if/how I can test this myself. https://github.com/llvm/llvm-project/pull/142948 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
