http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35831
--- Comment #9 from janus at gcc dot gnu.org 2011-09-11 10:06:03 UTC --- (In reply to comment #7) > Combining the three statements above, F08 clearly demands that the *shape* of > the argument should be the same (meaning that the bounds themselves may > differ). However, there is one more sentence which seems to restrict this further: "If a type parameter of an object or a bound of an array is not a constant expression, the exact dependence on the entities in the expression is a characteristic." Here, the standard talks explicitly about 'bounds of an array' and not just the 'shape'. Depending on how one reads this, it might mean two things: 1) for non-constant bounds, the bounds themselves have to match, not only the shape 2) for non-constant bounds, the shape expression (i.e. "upper bound minus lower bound") has to match, including the exact dependence on entities in the expression
