sammccall added a comment. This would be pretty unfortunate for iteration times: in my workspace having just run `ninja check-clangd`, `ninja clang` has 1450 additional build steps remaining. (All of which would get frequently invalidated e.g. by changes to common LLVM headers).
`clangd/test/lit.cfg.py` mentions clang, our intent here is configure some common required variables etc, not depend clang itself. It's not very well-defined though :-( I'm pretty sure it used to work at some point, but now after a clean: llvm-lit: /usr/local/google/home/sammccall/src/llvm-project/llvm/utils/lit/lit/llvm/config.py:342: fatal: couldn't find 'clang' program, try setting CLANG in your environment Looking at the contents of llvm/config.py, it's getting the builtin includes directory in order to use it in lit substitutions like %clang_cc1. Which of course we don't use in clangd tests. I'm trying to remember exactly why we do need to use this module at all, but I suspect there was a real reason. I think we're going to need to refactor llvm/config.py a little to let us take the bits we want. I can have a dig into this soon... Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D90518/new/ https://reviews.llvm.org/D90518 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
