[Bug libstdc++/48771] New: [C++0x] is_literal_type incorrect for references to non-literal types

[Bug libstdc++/48771] New: [C++0x] is_literal_type incorrect for references to non-literal types

[daniel.kruegler at googlemail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48771

           Summary: [C++0x] is_literal_type incorrect for references to
                    non-literal types
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: libstdc++
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: daniel.kruegler@googlemail.com


gcc 4.7.0 20110423 (experimental) in C++0x mode rejects the following code:

//-----------
#include <type_traits>

struct NonLiteral {
  NonLiteral();
  ~NonLiteral();
};

static_assert(std::is_literal_type<NonLiteral&>::value, "Error"); // #1
static_assert(std::is_literal_type<NonLiteral&&>::value, "Error"); // #2
//-----------

because both assertions fire. This code should be accepted, because as of the
FDIS all reference types are literal types (similar to pointers).

[Bug c++/48771] [C++0x] is_literal_type incorrect for references to non-literal types

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48771

Paolo Carlini <paolo.carlini at oracle dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011.04.26 14:10:31
                 CC|                            |jason at gcc dot gnu.org
          Component|libstdc++                   |c++
     Ever Confirmed|0                           |1

--- Comment #1 from Paolo Carlini <paolo.carlini at oracle dot com> 2011-04-26 14:10:31 UTC ---
This is really a C++ front-end issue, not a library issue, because it just
boils down to __is_literal_type and in turn to the internal literal_type_p,
which apparently should be updated to reflect the FDIS concept.

[Bug c++/48771] [C++0x] is_literal_type incorrect for references to non-literal types

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48771

Paolo Carlini <paolo.carlini at oracle dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |paolo.carlini at oracle dot
                   |                            |com

--- Comment #2 from Paolo Carlini <paolo.carlini at oracle dot com> 2011-04-26 14:39:36 UTC ---
The below patchlet fixes the problem and passes the testsuite.

Jason, let me know if you consider it safe, for 4_6-branch too, in case.

Paolo.

//////////////

Index: testsuite/g++.dg/ext/is_literal_type1.C
===================================================================
--- testsuite/g++.dg/ext/is_literal_type1.C    (revision 0)
+++ testsuite/g++.dg/ext/is_literal_type1.C    (revision 0)
@@ -0,0 +1,11 @@
+// PR c++/48771
+// { dg-do compile }
+// { dg-options "-std=c++0x" }
+
+struct NonLiteral {
+  NonLiteral();
+  ~NonLiteral();
+};
+
+static_assert(__is_literal_type(NonLiteral&), "Error");
+static_assert(__is_literal_type(NonLiteral&&), "Error");
Index: cp/semantics.c
===================================================================
--- cp/semantics.c    (revision 172961)
+++ cp/semantics.c    (working copy)
@@ -5331,7 +5331,8 @@ float_const_decimal64_p (void)
 bool
 literal_type_p (tree t)
 {
-  if (SCALAR_TYPE_P (t))
+  if (SCALAR_TYPE_P (t)
+      || TREE_CODE (t) == REFERENCE_TYPE)
     return true;
   if (CLASS_TYPE_P (t))
     return CLASSTYPE_LITERAL_P (t);

[Bug c++/48771] [C++0x] is_literal_type incorrect for references to non-literal types

[jason at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48771

--- Comment #3 from Jason Merrill <jason at gcc dot gnu.org> 2011-04-27 03:58:16 UTC ---
Looks fine, but also remove the REFERENCE_TYPE check in
valid_type_in_constexpr_fundecl_p.

[Bug c++/48771] [C++0x] is_literal_type incorrect for references to non-literal types

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48771

Paolo Carlini <paolo.carlini at oracle dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
                 CC|paolo.carlini at oracle dot |
                   |com                         |
         AssignedTo|unassigned at gcc dot       |paolo.carlini at oracle dot
                   |gnu.org                     |com

--- Comment #4 from Paolo Carlini <paolo.carlini at oracle dot com> 2011-04-27 08:26:13 UTC ---
Ok, thanks, patch to the mailing list forthcoming.

I'm also considering removing valid_type_in_constexpr_fundecl_p altogether and
replacing the two calls with literal_type_p directly.

[Bug c++/48771] [C++0x] is_literal_type incorrect for references to non-literal types

[paolo at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48771

--- Comment #5 from paolo at gcc dot gnu.org <paolo at gcc dot gnu.org> 2011-04-28 09:21:25 UTC ---
Author: paolo
Date: Thu Apr 28 09:21:23 2011
New Revision: 173062

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=173062
Log:
/cp
2011-04-28  Paolo Carlini  <paolo.carlini@oracle.com>

    PR c++/48771
    * semantics.c (literal_type_p): Reference types are literal types,
    per the FDIS.
    (valid_type_in_constexpr_fundecl_p): Remove.
    (is_valid_constexpr_fn): Adjust.

/testsuite
2011-04-28  Paolo Carlini  <paolo.carlini@oracle.com>

    PR c++/48771
    * g++.dg/ext/is_literal_type1.C: New.

Added:
    trunk/gcc/testsuite/g++.dg/ext/is_literal_type1.C
Modified:
    trunk/gcc/cp/ChangeLog
    trunk/gcc/cp/semantics.c
    trunk/gcc/testsuite/ChangeLog

[Bug c++/48771] [C++0x] is_literal_type incorrect for references to non-literal types

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48771

Paolo Carlini <paolo.carlini at oracle dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.6.1

--- Comment #6 from Paolo Carlini <paolo.carlini at oracle dot com> 2011-04-28 09:23:50 UTC ---
Done on the mainline so far.

[Bug c++/48771] [C++0x] is_literal_type incorrect for references to non-literal types

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48771

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.6.1                       |---

[Bug c++/48771] [C++0x] is_literal_type incorrect for references to non-literal types

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48771

Paolo Carlini <paolo.carlini at oracle dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|---                         |4.7.0

--- Comment #7 from Paolo Carlini <paolo.carlini at oracle dot com> 2011-06-06 15:42:57 UTC ---
I guess we can be satisfied with having this fixed in mainline.