https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109846
Bug ID: 109846
Summary: [rejects valid] Pointer-valued function reference
rejected as actual argument
Product: gcc
Version: 13.1.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: fortran
Assignee: unassigned at gcc dot gnu.org
Reporter: neil.n.carlson at gmail dot com
Target Milestone: ---
Gfortran rejects the following example.
module foo
type :: parameter_list
contains
procedure :: sublist
end type
contains
function sublist(this) result(slist)
class(parameter_list), intent(inout) :: this
class(parameter_list), pointer :: slist
allocate(slist)
end function
end module
program example
use foo
type(parameter_list) :: plist
call sub(plist%sublist())
contains
subroutine sub(plist)
type(parameter_list), intent(inout) :: plist
end subroutine
end program
With this error:
17 | call sub(plist%sublist())
| 1
Error: ‘sublist’ in variable definition context (actual argument to INTENT =
OUT/INOUT) at (1) is not a variable
It is accepted by both Intel OneAPI and NAG compilers. The sublist function
returns
a polymorphic pointer, which seems to be the source of the error. If it is
modified
to return a non-polymorphic pointer the example compiles without error.
Alternatively,
if the dummy argument intent is changed to intent(in) it also compiles without
error.
It is valid for a class(parameter_list) variable to be the actual argument for
a
type(parameter_list), intent(inout) dummy argument. So in light of 9.2/C902
(2018)
I think this is a gfortran bug.