(j3.2006) (SC22WG5.5177) [ukfortran] Draft result of ballot on Corrigendum 3

[Bill Long]

On 12/25/13 2:19 AM, Malcolm Cohen wrote:
> I certainly have no objection to rethinking the way that we describe an
> array-spec with a view to improving the exposition, quite the opposite.  But
> that would be for the next revision.  And it is certainly unclear at this point
> whether any real improvement could be made, or whether that would just be moving
> the complications out of the BNF and into the prose (as mentioned above, that is
> often a bad idea).

OK, I now see you were focused on BNF ambiguity while I was focused on 
visual ambiguity.  Hopefully I now get the concept behind the change:

OLD (two syntax terms):

  <assumed-size-spec>  is used to declare an assumed-size array.

<implied-shape-spec-list>  is used to declare an implied-shape array.

NEW (Three syntax terms):

<assumed-size-spec> is used to declare an assumed-size array of rank 2 
or larger.

<implied-shape-spec> is used to declare an implied-shape array of rank 2 
or larger.

<implied-shape-or-assumed-size-spec> is used for the rank 1 case of 
either assumed size or implied shape.

It is unfortunate that <assumed-size-spec> is no longer the syntax for 
declaring any assumed-size array, but I guess that is a price we pay for 
choosing this answer to the interp.

This is an improvement for BNF readers, but does not really do much for 
the programmer.  I think it would be helpful (at least in F2015, if not 
in the interp) to add a Note at the end of 5.3.8.6, at [96:31+] 
something like:

"NOTE 5.12a

For arrays of rank 1, a declaration like

    DIMENSION  ::  X(*)

or

    INTEGER :: X(*)

declares either a dummy array or a named constant array, depending on 
whether X appears as a dummy argument in a prior FUNCTION, SUBROUTINE, 
or ENTRY statement in the program unit, or X appears a named constant in 
a subsequent PARAMETER statement in the program unit.    The 
disambiguating FUNCTION, SUBROUTINE, ENTRY, or PARAMETER statement could 
be many lines separated from the declaration statements above.  As a 
result, such declaration statements could be visually confusing.  This 
confusion can be avoided if implied-shape arrays are always declared in 
a type declaration statement that contains the PARAMETER attribute, such as

    INTEGER,PARAMETER :: X(*) = [1,2,3]

"


Cheers,
Bill






-- 
Bill Long                                           longb at cray.com
Fortran Technical Support    &                 voice: 651-605-9024
Bioinformatics Software Development            fax:   651-605-9142
Cray Inc./Cray Plaza, Suite 210/380 Jackson St./St. Paul, MN 55101