In PR fortran/91513, Damian Rouson points out that the Fortran
does contain the words "impure variable". Damian and I spent
an afternoon together a few weeks ago where I gave Damian a
world wind view of how I see gcc/fortran and go about debugging
problems. Damian and I both have full schedules, but I was
hoping he would submit the patch to the list. My time of
fixing gfortran bugs is running out, so ...
OK to commit?
2019-09-29 Damian Rouson <dam...@sourceryinstitue.org>
PR fortran/91513
* resolve.c (resolve_ordinary_assign): Improved error message.
2019-09-29 Damian Rouson <dam...@sourceryinstitue.org>
PR fortran/91513
* gfortran.dg/impure_assignment_2.f90: Update dg-error regex.
--
Steve
Index: gcc/fortran/resolve.c
===================================================================
--- gcc/fortran/resolve.c (revision 276269)
+++ gcc/fortran/resolve.c (working copy)
@@ -10774,9 +10774,12 @@ resolve_ordinary_assign (gfc_code *code, gfc_namespace
"component in a PURE procedure",
&rhs->where);
else
- gfc_error ("The impure variable at %L is assigned to "
- "a derived type variable with a POINTER "
- "component in a PURE procedure (12.6)",
+ /* F2008, C1283 (4). */
+ gfc_error ("In a pure subprogram an INTENT(IN) dummy argument "
+ "shall not be used as the expr at %L of an intrinsic "
+ "assignment statement in which the variable is of a "
+ "derived type if the derived type has a pointer "
+ "component at any level of component selection.",
&rhs->where);
return rval;
}
Index: gcc/testsuite/gfortran.dg/impure_assignment_2.f90
===================================================================
--- gcc/testsuite/gfortran.dg/impure_assignment_2.f90 (revision 276269)
+++ gcc/testsuite/gfortran.dg/impure_assignment_2.f90 (working copy)
@@ -40,7 +40,7 @@ CONTAINS
PURE FUNCTION give_next3(node)
TYPE(node_type), intent(in) :: node
TYPE(node_type) :: give_next
- give_next = node ! { dg-error "impure variable" }
+ give_next = node ! { dg-error "pure subprogram" }
END FUNCTION
END MODULE pr20863
