Revision 126185 introduced ISO C Binding to gfortran.
In that revision, a check for a conflict between a
derived type with the PRIVATE attribute and BIND(C) was
introduced. After checking the F2003, F2008, and F2018
standards, I cannot find this restriction. Thus, the
check is removed by the attached patch. Regression
checked on x86_64-*-freebsd. OK to commit?
2019-06-19 Steven G. Kargl <ka...@gcc.gnu.org>
PR fortran/86587
* symbol.c (verify_bind_c_derived_type): Remove erroneous error
checking for BIND(C) and PRIVATE attributes.
2019-06-19 Steven G. Kargl <ka...@gcc.gnu.org>
PR fortran/86587
* gfortran.dg/pr86587.f90: New test.
--
Steve
Index: gcc/fortran/symbol.c
===================================================================
--- gcc/fortran/symbol.c (revision 272481)
+++ gcc/fortran/symbol.c (working copy)
@@ -4529,16 +4529,6 @@ verify_bind_c_derived_type (gfc_symbol *derived_sym)
curr_comp = curr_comp->next;
} while (curr_comp != NULL);
-
- /* Make sure we don't have conflicts with the attributes. */
- if (derived_sym->attr.access == ACCESS_PRIVATE)
- {
- gfc_error ("Derived type %qs at %L cannot be declared with both "
- "PRIVATE and BIND(C) attributes", derived_sym->name,
- &(derived_sym->declared_at));
- retval = false;
- }
-
if (derived_sym->attr.sequence != 0)
{
gfc_error ("Derived type %qs at %L cannot have the SEQUENCE "
Index: gcc/testsuite/gfortran.dg/pr86587.f90
===================================================================
--- gcc/testsuite/gfortran.dg/pr86587.f90 (nonexistent)
+++ gcc/testsuite/gfortran.dg/pr86587.f90 (working copy)
@@ -0,0 +1,18 @@
+! { dg-do compile }
+! PR fortran/86587
+! Code contirubted by Valentin Clement <valentin.clement at env dot ethz dot ch>
+!
+module mod1
+ use iso_c_binding
+ type, bind(c), private :: mytype
+ integer(c_int) :: i1, i2
+ end type
+end module mod1
+
+module mod2
+ use iso_c_binding
+ private
+ type, bind(c) :: mytype
+ integer(c_int) :: i1, i2
+ end type
+end module mod2
