# Data-pointer-object and unlimited polymorphic target **URL:** \ **Category:** Help\ **Created:** [December 18, 2024, 10:13am UTC](https://fortran-lang.discourse.group/t/data-pointer-object-and-unlimited-polymorphic-target/8983 "2024-12-18T10:13:54Z")\ **Posts on this page:** 12\ **Page:** 1
**Author:** ![Euler-37](https://avatars.discourse-cdn.com/v4/letter/e/b19c9b/32.png) [@Euler-37](https://fortran-lang.discourse.group/u/Euler-37)\ **Post date:** [December 18, 2024, 10:13am UTC](https://fortran-lang.discourse.group/t/data-pointer-object-and-unlimited-polymorphic-target/8983/1 "2024-12-18T10:13:55Z")
```fortran program main implicit none type pair sequence real::x integer::y end type pair type(pair),target::p type(pair),pointer::ptr_pair integer,target::i integer,pointer::i_ptr p=pair(1.0,2) ptr_pair=>view(p) print*,ptr_pair i=1 i_ptr => view(i)!WRONG print*,i_ptr contains function view(a)result(b) class(*),intent(in),target::a class(*),pointer::b b=>a end function view end program main ``` Compilation error ```sh app/example.f90:16:5: 16 | i_ptr => view(i) | 1 ~~~~~ Error: Data-pointer-object at (1) must be unlimited polymorphic, or of a type with the BIND or SEQUENCE attribute, to be compatible with an unlimited polymorphic target ``` Why intrinsic type pointer not permit point to polymorphic, but user defined type allowed. Is intrinsic type has `sequence` attribute? ---
**Author:** ![Arjen](https://avatars.discourse-cdn.com/v4/letter/a/b9bd4f/32.png) [@Arjen](https://fortran-lang.discourse.group/u/Arjen)\ **Post date:** [December 18, 2024, 10:25am UTC](https://fortran-lang.discourse.group/t/data-pointer-object-and-unlimited-polymorphic-target/8983/2 "2024-12-18T10:25:18Z")
I think the reason is that a derived type can be used as the parent of another type, whereas intrinsic types cannot. (At least that seems a plausible difference :).) It does not matter whether there are any actual procedures/methods connected to the type. ---
**Author:** ![ivanpribec](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/ivanpribec/32/3290_2.png) [@ivanpribec](https://fortran-lang.discourse.group/u/ivanpribec)\ **Post date:** [December 18, 2024, 10:57am UTC](https://fortran-lang.discourse.group/t/data-pointer-object-and-unlimited-polymorphic-target/8983/3 "2024-12-18T10:57:41Z")
Worth testing with different compilers. Flang: ```txt error: Semantic errors in /app/example.f90 /app/example.f90:13:5: error: function result type 'CLASS(*)' is not compatible with pointer type 'pair' ptr_pair=>view(p) ^^^^^^^^^^^^^^^^^ /app/example.f90:16:5: error: function result type 'CLASS(*)' is not compatible with pointer type 'INTEGER(4)' i_ptr => view(i)!WRONG ^^^^^^^^^^^^^^^^ Compiler returned: 1 ``` NAG: ```txt > nagfor example.f90 NAG Fortran Compiler Release 7.2(Shin-Urayasu) Build 7203 Error: example.f90, line 16: Pointer assignment of CLASS(*) target to INTEGER pointer [NAG Fortran Compiler error termination, 1 error] ``` Seems a bit like a dark corner of the language. It is probably worth looking into the J3 Fortran interpretation document to see what it says about a pointer to an unlimited polymorphic variable. ---
**Author:** ![Euler-37](https://avatars.discourse-cdn.com/v4/letter/e/b19c9b/32.png) [@Euler-37](https://fortran-lang.discourse.group/u/Euler-37)\ **Post date:** [December 18, 2024, 12:54pm UTC](https://fortran-lang.discourse.group/t/data-pointer-object-and-unlimited-polymorphic-target/8983/4 "2024-12-18T12:54:21Z")
As shown in [24-007.pdf](https://j3-fortran.org/doc/year/24/24-007.pdf), same as gfortran report ```plaintext C1017 (R1034) If data-target is unlimited polymorphic, data-pointer-object shall be unlimited polymorphic, or of a type with the BIND attribute or the SEQUENCE attribute. ``` ---
**Author:** ![everythingfunctional](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/everythingfunctional/32/176_2.png) [@everythingfunctional](https://fortran-lang.discourse.group/u/everythingfunctional)\ **Post date:** [December 18, 2024, 1:15pm UTC](https://fortran-lang.discourse.group/t/data-pointer-object-and-unlimited-polymorphic-target/8983/5 "2024-12-18T13:15:00Z")
> [@Euler-37](#): > > ```auto > or of a type with the BIND attribute or the SEQUENCE attribute. > > ``` I did not know about this, but omg, who thought this was a good idea? ---
**Author:** ![ivanpribec](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/ivanpribec/32/3290_2.png) [@ivanpribec](https://fortran-lang.discourse.group/u/ivanpribec)\ **Post date:** [December 18, 2024, 1:25pm UTC](https://fortran-lang.discourse.group/t/data-pointer-object-and-unlimited-polymorphic-target/8983/6 "2024-12-18T13:25:41Z")
Seems like it has something to do with alignment constraints which both C structs and sequence types have. It also suggests a workaround to get the desired result: ```auto type :: int_seq sequence integer :: i end type integer,target::i type(int_seq),pointer::i_ptr i=1 i_ptr => view(i) ! Now it works print*, i_ptr%i ``` It does look like a [loophole](https://fortran-lang.discourse.group/t/implicit-real-complex-conversion-in-fortran/7381/38) in the Wirth sense. ---
**Author:** ![everythingfunctional](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/everythingfunctional/32/176_2.png) [@everythingfunctional](https://fortran-lang.discourse.group/u/everythingfunctional)\ **Post date:** [December 18, 2024, 2:44pm UTC](https://fortran-lang.discourse.group/t/data-pointer-object-and-unlimited-polymorphic-target/8983/7 "2024-12-18T14:44:11Z")
> [@ivanpribec](#): > > It does look like a [loophole](https://fortran-lang.discourse.group/t/implicit-real-complex-conversion-in-fortran/7381/38) in the Wirth sense. Exactly. Why on earth would you want something like this to be valid? ```auto type :: four_bytes sequence integer(c_int8) :: i1, i2, i3, i4 end type type(who_tf_knows), target :: thing type(four_bytes), pointer :: rep thing = ... ! lets just see what four of the bytes ! starting at the beginning of this object look like rep => view(thing) print *, rep ``` I’m guessing there is some other text somewhere that still says this is invalid, but the compiler really ought to yell loudly about it. ---
**Author:** ![ivanpribec](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/ivanpribec/32/3290_2.png) [@ivanpribec](https://fortran-lang.discourse.group/u/ivanpribec)\ **Post date:** [December 18, 2024, 3:05pm UTC](https://fortran-lang.discourse.group/t/data-pointer-object-and-unlimited-polymorphic-target/8983/8 "2024-12-18T15:05:26Z")
Maybe it is useful to write some dictionary-like object where the return type is polymorphic? But I don’t understand why intrinsic types aren’t supported, if `sequence` types are. It would be nice to hear from the standard-bearers on why they thought this feature is useful. ---
**Author:** ![everythingfunctional](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/everythingfunctional/32/176_2.png) [@everythingfunctional](https://fortran-lang.discourse.group/u/everythingfunctional)\ **Post date:** [December 18, 2024, 6:51pm UTC](https://fortran-lang.discourse.group/t/data-pointer-object-and-unlimited-polymorphic-target/8983/9 "2024-12-18T18:51:54Z")
> [@everythingfunctional](#): > > I’m guessing there is some other text somewhere that still says this is invalid, Yep, found it. 10.2.2.3 Data pointer assignment > If the pointer object is of a type with the BIND attribute or the SEQUENCE attribute, the dynamic type of the pointer target shall be that type. ---
**Author:** ![ivanpribec](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/ivanpribec/32/3290_2.png) [@ivanpribec](https://fortran-lang.discourse.group/u/ivanpribec)\ **Post date:** [December 19, 2024, 9:28am UTC](https://fortran-lang.discourse.group/t/data-pointer-object-and-unlimited-polymorphic-target/8983/10 "2024-12-19T09:28:53Z")
I think that is still interesting, because you can write a function that returns objects with different dynamic types: ```fortran module tmod implicit none type :: t_int integer :: val end type type :: t_real real :: val end type type :: t_logical sequence logical :: val = .false. end type integer, parameter :: tf_int = 1 integer, parameter :: tf_real = 2 integer, parameter :: tf_logical = 3 type(t_int), target :: a = t_int(42) type(t_real), target :: b = t_real(33.0) type(t_logical), target :: c contains function ret(iflag) integer :: iflag class(*), pointer :: ret select case(iflag) case(tf_int) ret => a case(tf_real) ret => b case(tf_logical) ret => c case default ret => null() end select end function end module program main use tmod integer :: i, iflag class(*), pointer :: r type(t_logical), pointer :: rl iflag = 1 do i = 1, 9 r => ret(iflag) if (.not. associated(r)) cycle if (iflag == tf_logical) then rl => r print *, rl%val else select type(r) type is (t_int) print *, r%val type is (t_real) print *, r%val end select end if iflag = modulo(i,3) + 1 end do end program ``` ---
**Author:** ![Euler-37](https://avatars.discourse-cdn.com/v4/letter/e/b19c9b/32.png) [@Euler-37](https://fortran-lang.discourse.group/u/Euler-37)\ **Post date:** [December 19, 2024, 9:31am UTC](https://fortran-lang.discourse.group/t/data-pointer-object-and-unlimited-polymorphic-target/8983/11 "2024-12-19T09:31:10Z")
By the way, I wish that fortran intrinsic function like `findloc` can be used for user defined type with `sequence` attribute.It just combined with static type,so it should have something like `Eq Trait` ---
**Author:** ![ivanpribec](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/ivanpribec/32/3290_2.png) [@ivanpribec](https://fortran-lang.discourse.group/u/ivanpribec)\ **Post date:** [December 19, 2024, 11:33am UTC](https://fortran-lang.discourse.group/t/data-pointer-object-and-unlimited-polymorphic-target/8983/12 "2024-12-19T11:33:39Z")
Can’t you just use component selection? ```auto type :: t_int sequence integer :: val end type type(t_int) :: a(100) a = t_int(100) a(42) = t_int(1) print *, find(a, t_int(1)) contains pure function find(a, val) result(res) type(t_int), intent(in) :: a(:), val integer :: res intrinsic :: findloc res = findloc(a%val,val%val,dim=1) end function end ```