ilya-biryukov added a comment. I'll leave the LGTM to Manuel. I don't have a strong opinion here, but here are some thoughts.
The approach taken seems to be the least-leaky that I can think of unless we want to change the return type of `CompilationDatabase::getCompileCommand` or add a new method that would explicitly be used to infer commands. Both options are rather tricky to implement. Yet, having the new field in the `CompileCommand` feels wrong, especially in combination with a fact that `CompilationDatabase` returns a vector of compile commands rather just a single command, i.e. after this patch implementations can technically return a combination of inferred and non-inferred commands for the same file, which might force clients to think about proper prioritization and filtering commands, something I'd rather keep hidden. Repository: rC Clang CHANGES SINCE LAST ACTION https://reviews.llvm.org/D60194/new/ https://reviews.llvm.org/D60194 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
