(j3.2006) Re J3 Fortran interp letter ballot #23 - due 25-Mar-2011

[Robert Corbett]

The following Fortran interpretations are being balloted:

Yes  No   Number     Title


-C-  ---  F03/0124   Definition is poorly defined
-Y-  ---  F08/0050   Ordering requirements on definition of
                       specification functions
-Y-  ---  F08/0051   Pure procedure arguments with VALUE
-Y-  ---  F08/0052   Private type-bound procedures
-Y-  ---  F08/0053   Restrictions on generic declarations, generic
                       resolution
-Y-  ---  F08/0054   Requirements for needing an explicit interface

---------------------------------------------------------------------

F03/0124  Definition is poorly defined

The new interpretation is a great improvement over the interpretation
offered in ballot #22.

While checking the assertion that the array and derived type cases
follow from the definitions in Clause 16.6.1, I found an apparent
oversight.  Consider the program

       PROGRAM MAIN
         TYPE T
           TYPE(T), ALLOCATABLE :: P
           INTEGER I
         END TYPE
         TYPE (T) :: X, Y
         X%I = 2
         Y = X
       END

This program is not standard conforming because the variable X is
referenced while it is undefined (see Clause 6.2, paragraph 2 and
Clause 16.6.1, paragraph 4).  Paragraph 4 of Clause 16.6.1 should
permit a variable with an unallocated allocatable component to be
defined.  The standard as written does not allow X or Y to be
defined.

Please note that I approve of the interpretation and edits provided.
I am merely observing that more work remains to be done.

Robert Corbett
representing Oracle America