https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115361
--- Comment #6 from Arthur O'Dwyer <arthur.j.odwyer at gmail dot com> --- Thanks, that looks much less noisy! (Assuming godbolt.org has updated already.) I now see this: // https://godbolt.org/z/WqT6hs8ed f3, f7, and f9 now all give -Wuninitialized at -O1 and higher (and false-negative at -O0). I continue to think that this is a confusing/wrong diagnostic, even though it does achieve the pragmatic effect of saying "hey *something* is wrong with this line." I also continue to think that the message could be marginally improved by spelling out the type of the '<anonymous>' object it's complaining about. If you're lucky, that's an object of type `int` (the temporary constructed from `c`). But I honestly don't know if you'll be lucky; it's unclear to me why GCC is giving the -Wuninitialized message at all, here. <source>:15:56: warning: '<anonymous>' is used uninitialized [-Wuninitialized] 15 | int f3(char c) { const int& x = GetKey().rc(c); return x; } // True positive | ^ <source>:15:44: note: '<anonymous>' was declared here 15 | int f3(char c) { const int& x = GetKey().rc(c); return x; } // True positive | ~~~~~~~~~~~^~~ (What happens here is that we convert the value of `char c` to a prvalue int; materialize that temporary; pass a reference to the temporary to `rc`; `rc` returns a reference to that same temporary; we bind `x` to that reference; and then the temporary is destroyed. `x` dangles.)
