On 03/03/12 08:41, Paul Richard Thomas wrote: > Dear Tobias, > > This is certainly OK for 4.8. > > I have a couple of remarks: > (i) The DTYPE_TYPE_MASK is 0x38 so that we saturated it a long time > since! At the moment it does not cause any problems because of the > extremely limited use of the dtype 'type'.
Hi, AFAIU, you're saying here that since there are very few uses of the type part of dtype, it's not a problem. But at the moment, BT_ASSUMED with value 11 bleeds into the size part of dtype. In gfc_get_dtype_rank_type, we just set the type directly without applying the mask: ... i = rank | (n << GFC_DTYPE_TYPE_SHIFT); ... In fact, the mask seems completely unused. > Whilst the array > descriptor revamp will eliminate such worries, we should be mindful of > this; and I think that if we have a mask that we're not respecting, we need to clearly document at the definiton how that's done and why that's not a problem. And ideally, add some asserts to detect when we break that allowed usage pattern. Thanks, - Tom
