On Nov 15, 2004, at 11:52 AM, Aleksandar Donev wrote:

> To finally answer a thread started long ago, J3 passed an interp at its
> last meeting...

Please note that J3 passing an interp does *NOT* constitute a final
answer.  That is but one step of interp processing.  Although it is a
very important step, it is rather far from a final one.  It isn't even
the final J3 step. Interps passed at a meeting must also pass a
subsequent letter ballot.  It is not unusual for interps that passed at
a meeting to fail a subsequent letter ballot after there has been more
time for reflection and for issues to be raised.  After passing the J3
letter ballot, the interp must then go to WG5 and pass there also.  It
also has happened quite a few times that interps fail at that stage.

You presumably know this, Aleksandar, but there are presumably readers
of this list that do not.  Therefore, I thought it important to clarify
the usage of "final".

> J3's ANSWER:
>
> Yes, this is allowed, since "whole" means that b is
> ASSOCIATED with all elements of an array that was ALLOCATEd.
> In the given example, ASSOCIATED(a,b) is .TRUE., therefore either
> a or b can be used to DEALLOCATE the memory.

Allow me to note that the 2 sentences in the answer seem to contradict
each other (which is among the kinds of reasons that sometimes cause
interps that pass at a meeting to be rejected later).  The first
sentence talks about all the elements being associated.  The second
sentence talks about the result of the ASSOCIATED intrinsic. The
wording seems to imply that these 2 conditions are the same, which is
incorrect (as was elaborated in one of the early f90 interps from long
ago).  This leaves it unclear in at least my mind which of the 2
different conditions is actually intended.

For those who might wonder how all the elements could be associated
without the associated intrinsic returning .true., consider the
situation after

  b => a(10:1:-1)

There is a 1-to-1 mapping between the elements of a and b.  Every
element of "a" is associated with an element of b and vise versa.
However, associated(a,b) returns .false.. If J3 forgot about that
distinction, then I predict that the interp will need reconsideration.
In any event, I'd think that the wording of the answer ought to be more
precise in order to avoid an interp request about how to interpret the
interpretation.

I'll not try to reopen argument about which answer should be given.
I'll just suggest that, whatever answer is given, it ought to be
written self-consistently.

--
Richard Maine                |  Good judgment comes from experience;
[log in to unmask]       |  experience comes from bad judgment.
                             |        -- Mark Twain