P2996R3 drops "reflections of expressions", and makes it clear that what's reflected is a "result of evaluating an expression" - which can either be a value (if evaluated from a prvalue) or an object (if evaluated from an lvalue). As a bit of cleanup, it'd be nice to everywhere change the nomenclature from RK_const_value (and getReflectedConstValueExpr, etc) to RK_expr_result.
Still not clear whether storing this in a ConstantExpr was the smartest decision, but it doesn't seem to be getting in the way as of yet.
P2996R3 drops "reflections of expressions", and makes it clear that what's reflected is a "result of evaluating an expression" - which can either be a value (if evaluated from a prvalue) or an object (if evaluated from an lvalue). As a bit of cleanup, it'd be nice to everywhere change the nomenclature from
RK_const_value
(andgetReflectedConstValueExpr
, etc) toRK_expr_result
.Still not clear whether storing this in a
ConstantExpr
was the smartest decision, but it doesn't seem to be getting in the way as of yet.