AaronBallman wrote: I think adding more attributes is a viable option, but it doesn't scale particularly well and I would not be surprised if it caused some confusion in practice.
> But maybe we can exclude functions where the attribute directly refers to a > parameter. This doesn't work for the example code in the issue directly. In my use case, there is a global permission object rather than an object that gets passed around to functions because I don't want to incur the cost of passing the struct. It contains no useful members, it's more of a tag type (C++ sense of tags, not the C sense) than anything, so it's not really like a mutex. However, maybe that's a logical fault with my design? https://github.com/llvm/llvm-project/pull/141432 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
