Compiler Bug? c_f_procpointer() not accepting procedure pointer in COMMON

[ereisch]

 SUBROUTINE PROCPTR_COMMON (C_FCN) BIND(C)
C
C
 USE, INTRINSIC :: ISO_C_BINDING
 IMPLICIT NONE
C
C
 TYPE(C_FUNPTR), VALUE, INTENT(IN) :: C_FCN
 PROCEDURE(), POINTER :: F_FCN
 COMMON/FOO/F_FCN
C
 CALL C_F_PROCPOINTER ( C_FCN, F_FCN )

 END

In the above code block, I would expect this to compile fine, but I get an error that F_FCN is an undefined type.  However, if you comment-out the COMMON line, it compiles fine.  Is this a compiler bug, or illegal use of Fortran?  No special compiler flags are needed to bring about the issue.

Using ifort version 19.1.3.304.

[Steve_Lionel]

Not valid Fortran. 

C8119 (R874) A common-block-object shall not be a dummy argument, a function result, an allocatable variable, a derived-type object with an ultimate component that is allocatable, a procedure pointer, an automatic data object, a variable with the BIND attribute, an unlimited polymorphic pointer, or a coarray.

The error message could be better, for sure. I would consider it a bug that the compiler did not give an error message on the COMMON statement.

[ereisch]

Hmm....I was afraid of that.  I'm trying to think of how to get a function pointer from one portion of my program to another (very distant) routine at a point much later, and am not having much luck.  I tried storing the type(c_funptr) in the common block, which seems to work, but the copy of the c_funptr from the routine that receives it as an argument to the common block variable doesn't seem to be happening correctly.  I tried both "a = b" and "a => b", but the former doesn't copy the contents and the latter is illegal (because it isn't a TARGET).

[Steve_Lionel]

Why not use a module variable? COMMON has outlived its usefulness nowadays.