IMHO it would still be a useful feature, just one that is not strictly needed. You are not able to to pass around such structs, but neither is that possible with the suggested `defer` statement. The only advantage of `defer` is that the destructor code is inline, rather than in a separate destructor function.
But you could gain reusability of headers to be also used in C++, not needing to reinvent the wheel with new issues (e.g. variable lifetime), and a whole lot of existing experience with RAII.
But you could gain reusability of headers to be also used in C++, not needing to reinvent the wheel with new issues (e.g. variable lifetime), and a whole lot of existing experience with RAII.