Bug 51284 - [OOP] CLASS and VALUE attribute: No copy to temporary done
Summary: [OOP] CLASS and VALUE attribute: No copy to temporary done
Status: NEW
Alias: None
Product: gcc
Classification: Unclassified
Component: fortran (show other bugs)
Version: 4.7.0
: P3 normal
Target Milestone: ---
Assignee: Not yet assigned to anyone
URL:
Keywords: rejects-valid, wrong-code
Depends on:
Blocks:
 
Reported: 2011-11-23 16:35 UTC by Tobias Burnus
Modified: 2015-10-09 21:43 UTC (History)
1 user (show)

See Also:
Host:
Target:
Build:
Known to work:
Known to fail:
Last reconfirmed: 2015-10-09 00:00:00


Attachments
symbol.c part of the patch: Allow DIMENSION+VALUE with -std=f2008 (624 bytes, text/plain)
2011-12-09 21:20 UTC, Tobias Burnus
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Tobias Burnus 2011-11-23 16:35:07 UTC
The following program fails in several ways. As written, the program fails as "test" modifies the value of a%i - but it shouldn't:

"A present dummy argument with the VALUE attribute becomes argument associated with a definable anonymous data object whose initial value is the value of the actual argument." (12.5.2.3 Argument association).

If one uncomments "test2", it fails with:
  Error: VALUE attribute conflicts with DIMENSION attribute at (1)
However, in Fortran 2008 this restriction is not present:

C557 An entity with the VALUE attribute shall be a dummy data object
     that is not an assumed-size array or a coarray, and does not have
     a coarray ultimate component.
C558 An entity with the VALUE attribute shall not have the ALLOCATABLE,
     INTENT (INOUT), INTENT (OUT), POINTER, or VOLATILE attributes.

Additionally, with the current polymorphic array draft patch the
  b(:)%i  (in "b(:)%i = 8" and "any(b(:)%i /= 8")
causes an ICE (segfault). Doing a "do i=1,8; b(i)%i = 8; done" had a similar problem.

Finally, I think polymorphic arrays suffer from the same issue as scalar polymorphic arrays.


Note: A deep copy has to be done, which takes the allocatable components into account.


module m
  type t
    integer :: i
  end type t
contains
  subroutine test(x)
    class(t), VALUE :: x
    x%i = 5
  end subroutine test
!  subroutine test2(x)
!    class(t), VALUE :: x(:)
!    x(:)%i = 99
!  end subroutine test2
end module m

use m
class(t), allocatable :: a, b(:)
allocate ( t :: a, b(5))
a%i = 2
!b(:)%i = 8
call test(a)
!call test2(b)
if (a%i /= 2) call abort ()
!if (any(b(:)%i /= 8)) call abort ()
end
Comment 1 Tobias Burnus 2011-11-23 16:40:57 UTC
(In reply to comment #0)
> If one uncomments "test2", it fails with:
>   Error: VALUE attribute conflicts with DIMENSION attribute at (1)
> However, in Fortran 2008 this restriction is not present:
> 
> C558 An entity with the VALUE attribute shall not have the ALLOCATABLE,
>      INTENT (INOUT), INTENT (OUT), POINTER, or VOLATILE attributes.

In Fortran 2003:

"C527 (R501) If the VALUE attribute is specified, the PARAMETER, EXTERNAL, POINTER, ALLOCATABLE, DIMENSION, VOLATILE, INTENT(INOUT), or INTENT(OUT) attribute shall not be specified."

Note that the VALUE attribute is now allowed for EXTERNAL, e.g.
  subroutine foo(x)
    procedure(), VALUE :: x

As one may not modify "x" (only if it had a pointer attribute, but that's not allowed), it does not affect the code generation.
Comment 2 Tobias Burnus 2011-12-09 21:20:08 UTC
Created attachment 26036 [details]
symbol.c part of the patch: Allow DIMENSION+VALUE with -std=f2008

First patch: Allow DIMENSION with VALUE.

TODO:
- Handle dimension in trans*.c (currently, it will ICE)
- Handle the "copy in" for derived types and class (both scalar and array)
Comment 3 Tobias Burnus 2011-12-11 12:54:29 UTC
(In reply to comment #2)
> - Handle the "copy in" for derived types and class (both scalar and array)

For derived types: The arguments are passed by value thus it works. However, no copying is done for allocatable components. Note: Also "( )" (= INTRINSIC_PARENTHESES) does not work for DT w/ allocatable components.
Comment 4 Dominique d'Humieres 2015-10-09 21:43:03 UTC
Still present at r228660.