rorth wrote: > I still don't entirely understand how lit works exactly, but it seems to me > that completely removing the cmake target for this is not necessary, no? So > at least this wouldn't require the CI changes on top of the other changes. > Happy to hear your thoughts on this though @rorth
It occured to me that restoring a `clang-check-python` target might be the easiest solution: - I'd originally removed it because the python tests would now be run by regular `ninja check-clang`, not knowing about the CI and its use there. - Given how incredibly hard it seems to be to teach the CI to run just a single command (and me having no experience with GitHub Actions and seeing no need for them in the forseeable future), modifying the workflow doesn't seem doable to me. - Besides, while I'm by no means a `cmake` expert, I find my way around (and can test things locally). So I've reintroruced the target so the workflow should run unchanged, now only being `lit`-based. I've exluded it from `ninja check-all` for fear of it running twice then, once from `check-clang` and then by itself. I've retested this version on the same set of platforms as before (with the exception of Linux/sparc64 which is slow) and manually run `ninja check-clang-python`, all with the expected results. Let me know what you think. 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
