Optimizations:

• No guard variables generated by the compiler for static/thread_local objects.
• Simpler, more optimal result<T> implementation.