The confusion comes from not understanding when Cvalue actually applies. A Cvalue expression is the final result after all conversions are done - not the individual pieces inside the expression. In your examples, operands like byte_val still go through normal arithmetic promotions just like they’re supposed to. Cvalue only kicks in for the entire expression once it’s in its final state where no more conversions should happen. Think of it as the expression’s “final form” rather than a rule that blocks intermediate conversions during evaluation. The rule basically says once you’ve got a properly converted result, don’t do extra unnecessary conversions on that final value. That’s why your examples work fine - those conversions are the standard ones needed to make the arithmetic operations actually work.