(j3.2006) question about generic resolution

[Bill Long]

In f08 at [286:12.4.3.4.5p3] we allow (as a new feature) a dummy 
argument with the ALLOCATABLE attribute to be distinguishable from a 
dummy argument with the POINTER attribute.  Thus, I would conclude that 
this generic interface is OK:

   interface  sub

      subroutine suba (a)
          real,allocatable :: a(:)
      end subroutine suba

     subroutine subp (a)
         real,pointer,intent(in) :: a(:)
     end subroutine subp

   end interface sub

This extension made perfect sense in the f03 context of pointer dummy 
shall correspond to pointer actual.  However, at [296:12.5.2.7p2] we 
have another new feature that allows a pointer dummy to correspond to a 
non-pointer actual.  So, assuming the interface for sub above, how is 
the following resolved:

    real,allocatable,target :: actual(:)
    allocate(actual(100))

    call sub (actual)

Either of the specific interfaces appears to be viable as a callee. I 
assume that we intend that suba is called. However, I don't see anything 
in [303:12.5.5.2]  (resolving generics) that says this.   Did these two 
new features just fly by each other, or (hopefully) did I miss some part 
of the text that points out where the above example goes wrong, or is 
disambiguated.

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