Nick Clifton writes: > Besides, just because it wouldn't be needed for naked > functions any more, there is no reason to suppose that individual > targets might not have other reasons for suppressing inlining. Perhaps. But complexity should alway be defered as long as possible.