[Bug fortran/48351] [OOP] Realloc on assignment fails if parent component is CLASS

["townsend at astro dot wisc.edu" <gcc-bugzilla at gcc dot gnu dot org>]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48351

Rich Townsend <townsend at astro dot wisc.edu> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |townsend at astro dot
                   |                            |wisc.edu

--- Comment #3 from Rich Townsend <townsend at astro dot wisc.edu> 2011-07-07 02:34:41 UTC ---
(In reply to comment #0)
> http://groups.google.com/group/comp.lang.fortran/browse_thread/thread/b7a36eba5ef7f68b
>  by Nasser M. Abbasi
> 
> In the following program "%u" is allocatable. For
> 
>         this%u = u
> 
> the LHS should be allocated, but this only happens if "this" is a TYPE and not
> a CLASS - but that should be completely unrelated to (re)alloc on assignment.
> 
> The program works with ifort 11.1.
> 
> 
> module foo
>     implicit none
>     type :: foo_t
> !      private
>       DOUBLE PRECISION , ALLOCATABLE :: u(:)
>       contains
>         PROCEDURE :: make  ! or procedure, pass, same effect
>     end type foo_t
> 
>     contains
>     subroutine make(this,u)
>         implicit none
>         CLASS(foo_t) :: this
>         DOUBLE PRECISION, intent(in) :: u(:)  ! must be CLASS
> !           allocate(this%u(size(u)))  ! Must allocate now, else crash
>         this%u = u
>     end subroutine make
>     end module foo
> 
> program main2
>  use foo
>  implicit none
>  TYPE(foo_t) :: o
>  DOUBLE PRECISION , ALLOCATABLE :: u(:)
> 
>  u=[1,2,3,4]
>  CALL o%make(u)
>  print *, o%u
> end program main2

I've run into what appears to be this bug with 4.7 (Mac OS 10.6). My sample
code is as follows:

module realloc_lhs_m

  implicit none

  type mytype
     real, allocatable :: a(:)
   contains
     procedure :: set_a
  end type mytype

contains

  subroutine set_a (this, a)
    class(mytype), intent(out) :: this
    real, intent(in)           :: a(:)
    this%a = a
  end subroutine set_a

end module realloc_lhs_m

program realloc_lhs

  use realloc_lhs_m
  implicit none

  real, allocatable :: a(:)
  type(mytype)      :: m

  a = [1.,2.,3.,4.,5.]

  call m%set_a(a)
  print *, m%a

end program realloc_lhs

This can be easily worked around -- but of course should nevertheless be fixed.

cheers,

Rich