nridge added a comment. Thanks for taking the time to do a systematic review!
In D127184#3577165 <https://reviews.llvm.org/D127184#3577165>, @falbrechtskirchinger wrote: > bits/mofunc_impl.h I see this included from `bits/move_only_function.h`, so I think `<functional>` would make sense for it. > bits/new_allocator.h I see this included from `<experimental/memory_resource>`, we could add that. (There are a few more implementation headers in `experimental/bits` which are included by standard headers in `<experimental/...>` that we could consider adding.) > bits/specfun.h I see this included from `<cmath>`. > I've seen `*.tcc` files being mapped and have identified the following > missing files: > > bits/regex.tcc > bits/regex_automaton.tcc > bits/regex_compiler.tcc > bits/regex_executor.tcc > bits/regex_scanner.tcc > bits/string_view.tcc > > Should these be added as well? I think we can skip these as they only contain definitions, and code completion should choose the file containing the declaration. ================ Comment at: clang-tools-extra/clangd/index/CanonicalIncludes.cpp:197 {"bits/basic_string.tcc", "<string>"}, + {"bits/boost_concept_check.h", "<numeric>"}, {"bits/char_traits.h", "<string>"}, ---------------- The choice of `<numeric>` is pretty random here, as is already the case for `bits/concept_check.h`. Given that these headers don't declare any standard symbols, only symbols which are pure libstdc++ implementation details, maybe we should just omit them from the list? Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D127184/new/ https://reviews.llvm.org/D127184 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits