Closed degski closed 4 years ago
Hi degski, thank you for this detail and for reading my essay! I will fix it.
I clarified this in the text. It turns out that I was not wrong, both gcc and clang detect the obvious case, but it is sufficient to wrap the throw in a nested function to get "rid" of the warning.
Nice write-up.
Clang-10 does not blindly trust the
noexcept
specifier and warns at compile-time. It also finds (warns on) destructors that (transitively) throw, while not being markednoexcept(false)
.d.