sammccall added inline comments.
================ Comment at: clang-tools-extra/clangd/ConfigProvider.cpp:39 + auto Buf = FS.getBufferForFile(Path); + // If stat() succeeds but we failed to read, retry once and cache failure. + if (!Buf) ---------------- kadircet wrote: > why do we want to cache failure case for missing files ? > > If it is truly missing and we didn't race with another process (likely > git-checkout), the next stat will fail and we'll return a no-op fragment. > If it was a race, we want to pick it up on the next run (and reading twice > might not have enough of a delay in between), so why not just cache with a > sentinel key in here too? Yeah I was worried about a file we can stat but chronically can't read (e.g. because permissions are wrong). But this is probably not worth worrying about. Changed no not cache as you suggest. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D82964/new/ https://reviews.llvm.org/D82964 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
