comments on IR2054

Here is what James Unterberger, who submitted the original IR has to say 
about the
proposed solution:


The analysis of 1) is not quite what I had in mind.


In the response labeled 1), I would prefer something along these lines.
(I was just getting at the idea that the "direction of the base type of
the formal", when said base type is an array type, is really "the
direction of the index subtype at the corresponding index position".)

     The direction of each of the index ranges of the formal generic or
     formal port is that of the corresponding index subtype of the base
     type of the formal.
     The high bound and low bound of the index ranges at each index
     position are respectively determined from the maximum value
     and minimum value of the indices given in the association elements
     corresponding to the formal.

In fact, what the analyzer is proposing, that the directions somehow
come from the directions of the indices given in the association elements
corresponding to the formal, is unworkable because there may not *be* any
directions (i.e., no slice-of-formal as the formal designator in any
of the association elements, e.g., they're all index names).