# Non-associated procedure pointer as optional argument

**URL:** <https://fortran-lang.discourse.group/t/non-associated-procedure-pointer-as-optional-argument/9936>\
**Category:** GNU\
**Created:** [July 17, 2025, 11:08am UTC](https://fortran-lang.discourse.group/t/non-associated-procedure-pointer-as-optional-argument/9936 "2025-07-17T11:08:25Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![FedericoPerini](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/federicoperini/32/1750_2.png) [@FedericoPerini](https://fortran-lang.discourse.group/u/FedericoPerini)\
**Post date:** [July 17, 2025, 11:08am UTC](https://fortran-lang.discourse.group/t/non-associated-procedure-pointer-as-optional-argument/9936/1 "2025-07-17T11:08:25Z")

</div>

Hello Fortran community,

I’ve stumbled upon an unexpected gfortran runtime error stop when using a non-associated procedure pointer as `optional` argument to a subroutine

```auto
module m

  abstract interface
    ! Just a function interface
    subroutine my_fun(x)
        real, intent(in) :: x
    end subroutine my_fun   
  end interface   

  contains

  ! Optional function argument
  subroutine with_fun(sub)
     procedure(my_fun), optional :: sub
     print *, present(sub)
  end subroutine   

end module m

program p
  use m

  procedure(my_fun), pointer :: ptr => null()

  call with_fun() ! no runtime error 
  call with_fun(sub=ptr) ! runtime error

end  

```

- The error is triggered usin `-fcheck=all`
- It is _not_ triggered when a _data_ pointer is passed rather than a procedure pointer.

I believe it is of course correct to check this when the procedure argument is not `optional`, but probably it should not be, when `optional` (this is what happens for variable pointers)

Do you think this issue may be submitted to the GNU bugzilla? IMHO this is a minor bug related to an overlooked check that may be worthwhile of lifting.

[Here](https://godbolt.org/z/3eboY6zGr) I’ve put an example that shows no error triggered on the data, error triggered on the `null()` procedure.

All gfortran versions seem to be affected by this issue, so most likely there is a Standard-related reason that check is still in place.

---

<div class="post-metadata">

**Author:** ![sblionel](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/sblionel/32/853_2.png) [@sblionel](https://fortran-lang.discourse.group/u/sblionel)\
**Post date:** [July 17, 2025, 7:12pm UTC](https://fortran-lang.discourse.group/t/non-associated-procedure-pointer-as-optional-argument/9936/2 "2025-07-17T19:12:59Z")

</div>

My reading of the standard suggests that no error should be issued. Here’s what the standard says:

> 15.5.2.13 Argument presence and restrictions on arguments not present  
> 14 1 A dummy argument or an entity that is host associated with a dummy argument is not present if the dummy  
> 15 argument  
> 16 • does not correspond to an actual argument,  
> 17 • corresponds to an actual argument that is not present, or  
> 18 • does not have the ALLOCATABLE or POINTER attribute, and corresponds to an actual argument that  
> 19 – has the ALLOCATABLE attribute and is not allocated, or  
> 20 – has the POINTER attribute and is disassociated;  
> 21 otherwise, it is present.

There is no distinction made between being a data pointer or a procedure pointer. In your example, the dummy does not have ALLOCATBLE nor POINTER, and is associated with an actual argument that is a disassociated pointer, therefore the dummy argument is not present. As all you do is call PRESENT on the dummy, that is permitted.

---

<div class="post-metadata">

**Author:** ![RonShepard](https://avatars.discourse-cdn.com/v4/letter/r/a3d4f5/32.png) [@RonShepard](https://fortran-lang.discourse.group/u/RonShepard)\
**Post date:** [July 17, 2025, 10:00pm UTC](https://fortran-lang.discourse.group/t/non-associated-procedure-pointer-as-optional-argument/9936/3 "2025-07-17T22:00:04Z")

</div>

> [@sblionel](#):
>
> 20 – has the POINTER attribute and is disassociated;

Does it make a difference if the pointer is disassociated with `ptr=>null()` or with `nullify(ptr)`?

---

<div class="post-metadata">

**Author:** ![sblionel](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/sblionel/32/853_2.png) [@sblionel](https://fortran-lang.discourse.group/u/sblionel)\
**Post date:** [July 17, 2025, 10:11pm UTC](https://fortran-lang.discourse.group/t/non-associated-procedure-pointer-as-optional-argument/9936/4 "2025-07-17T22:11:01Z")

</div>

No, these are the same.

FWIW, ifx, with -check all enabled, gives no error and displays F in both cases.

---

<div class="post-metadata">

**Author:** ![FedericoPerini](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/federicoperini/32/1750_2.png) [@FedericoPerini](https://fortran-lang.discourse.group/u/FedericoPerini)\
**Post date:** [July 18, 2025, 6:28am UTC](https://fortran-lang.discourse.group/t/non-associated-procedure-pointer-as-optional-argument/9936/5 "2025-07-18T06:28:06Z")

</div>

Thank you @sblionel @RonShepard for confirming the case with your comments! yes that was also my interpretation. I have filed a [ticket](https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121145) at the GNU Bugzilla.