[Bug fortran/55758] New: LOGICAL and BIND(C): Reject kind=2/4/8/16 with -std=f2008, improve warning, switch to nonBOOLEAN_TYPE for those

["burnus at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55758

             Bug #: 55758
           Summary: LOGICAL and BIND(C): Reject kind=2/4/8/16 with
                    -std=f2008, improve warning, switch to nonBOOLEAN_TYPE
                    for those
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Keywords: accepts-invalid, diagnostic
          Severity: normal
          Priority: P3
         Component: fortran
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: burnus@gcc.gnu.org


See http://gcc.gnu.org/ml/fortran/2012-12/threads.html#00135

In the Fortran standard, only LOGICALs of kind C_BOOL interoperate with C;
namely, with C99's _Bool/bool.

In C, also integers - in particular "int" - are used in Boolean expressions,
allows for the whole range of integral values - all are "true" except for 0
which is "false".

gcc's _Bool and gfortran's LOGICAL all assume a binary state (BOOLEAN_TYPE)
which is either 0 or 1 such that .NOT. can be implemented by flipping a single
bit.

Hence, regarding an "int" as BOOLEAN_TYPE leads to .NOT.(-1) = -2.


Currently, LOGICAL(kind=C_INT) (c_int == 4) is accepted by gfortran and C
binding, but it might lead to wrong results with .NOT.


Expected:
* With -std=f95/f2003/f2008/f2008tr, only LOGICAL with kind=C_BOOL (C_BOOL ==
1) is accepted
* With -std=gnu, also the others (2,4,8,16) are accepted but a warning is
printed for them.
* Consider replacing internally BOOLEAN_TYPE by a signed-integer type LOGICAL
with kind /= C_BOOL in procedures with C binding