[Bug c++/48872] New: [C++0x][noexcept] Placement-new problem with non-empty arguments

[Bug c++/48872] New: [C++0x][noexcept] Placement-new problem with non-empty arguments

[daniel.kruegler at googlemail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48872

           Summary: [C++0x][noexcept] Placement-new problem with non-empty
                    arguments
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: daniel.kruegler@googlemail.com
                CC: jason@redhat.com


gcc 4.7.0 20110430 (experimental) in C++0x mode rejects the following program:

//--------------
#include <new>

struct U {};

template<class T>
T&& create();

const bool b = noexcept(::new (((void*) 0)) U(create<U&&>()));
static_assert(b, "Ouch"); // #

int main() {}
//--------------

At the line marked with # the static_assert fires. The same problem occurs, if
the initializer of U is e.g. create<U&>() or create<const U&>(), but a program
with an empty initializer like

const bool b = noexcept(::new (((void*) 0)) U());

is accepted. In either case no operation which potentially throws an exception
is called, therefore the program should be accepted.

[Bug c++/48872] [C++0x][noexcept] Placement-new problem with non-empty arguments

[daniel.kruegler at googlemail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48872

--- Comment #1 from Daniel Krügler <daniel.kruegler at googlemail dot com> 2011-05-04 17:38:16 UTC ---
(In reply to comment #0)

<sigh> an error happended when I tryied to transform my original example (using
std::declval) to a form with a minimum of library support. So let me provide
this fixed example and please ignore the first example:

gcc 4.7.0 20110430 (experimental) in C++0x mode rejects the following program:

//---------------
#include <new>

template<class T>
T&& create() noexcept;

struct U {
  ~U() noexcept(false);
};

constexpr bool b = noexcept(::new ((void*) 0) U(create<U&&>()));

static_assert(b, "Ouch"); // #

int main() {}
//---------------

The compiler rejects the line marked with # because of an assertion failure.
This happens in all cases where an U is placement-constructed with arguments as
with create<U&>() or with create<const U&>(). The example is accpted when only
value-initialized, like this:

constexpr bool b = noexcept(::new ((void*) 0) U());

This example demonstrates that the compiler seems to create a temporary in the
affected cases, so this is just another aspect of #48873.

[Bug c++/48872] [C++0x][noexcept] Placement-new problem with non-empty arguments

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48872

--- Comment #2 from Paolo Carlini <paolo.carlini at oracle dot com> 2011-05-04 17:53:25 UTC ---
As I side remark, I'm not sure to understand why your testcases often include
an 'int main { }' line, I don't think are about link-time problems, no? If you
compile only what's the main for?

[Bug c++/48872] [C++0x][noexcept] Placement-new problem with non-empty arguments

[daniel.kruegler at googlemail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48872

--- Comment #3 from Daniel Krügler <daniel.kruegler at googlemail dot com> 2011-05-04 17:59:21 UTC ---
(In reply to comment #2)
> As I side remark, I'm not sure to understand why your testcases often include
> an 'int main { }' line, I don't think are about link-time problems, no? If you
> compile only what's the main for?

I will try to reduce that in the future for bug reports. I'm still in standard
mode when doing this to ensure that the program is well-formed and
well-defined. Partial programs seem ok here, if I understand you correctly.

[Bug c++/48872] [C++0x][noexcept] Placement-new problem with non-empty arguments

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48872

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

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

--- Comment #4 from Paolo Carlini <paolo.carlini at oracle dot com> 2011-05-04 18:27:21 UTC ---
To be clear, this is what I meant: do you want to test a compile-time,
link-time, or run-time problem (to name the most common categories)? Often the
former, I suppose, and indeed this is the way your testcases ended up being
used in the testsuite in the past in the majority of the cases. In that case no
need to add a main (somebody will certainly remove it anyway). For more
details:

  http://gcc.gnu.org/wiki/HowToPrepareATestcase

[Bug c++/48872] [C++0x][noexcept] Placement-new problem with non-empty arguments

[daniel.kruegler at googlemail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48872

--- Comment #5 from Daniel Krügler <daniel.kruegler at googlemail dot com> 2011-05-05 07:15:12 UTC ---
(In reply to comment #1)

More information by a different example: It seems that the problem case reacts
very sensitive to minor modifications of the class. The following provides a
slightly different version of the previous example:

//---------------
#include <new>

//#define ADD_CTOR

struct U2 {
#ifdef ADD_CTOR
  U2(int) noexcept;
#endif
  ~U2() noexcept(false);
};

template<class T>
T&& create() noexcept;

const bool b = noexcept(::new (((void*) 0)) U2(create<U2&&>()));
static_assert(b, "Ouch"); // #
//---------------

As written, the same static assert violation occurs as in the example provided
by comment 1. But if the line

//#define ADD_CTOR

is uncommented *without any further changes* the program becomes accepted.
Somehow the addition of the user-provided constructor - even though it is not
used here - influences the evaluation of the noexcept expression.

[Bug c++/48872] [C++0x][noexcept] Placement-new problem with non-empty arguments

[jason at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48872

Jason Merrill <jason at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
                 CC|                            |jason at gcc dot gnu.org
         Resolution|                            |DUPLICATE
   Target Milestone|---                         |4.7.0

--- Comment #6 from Jason Merrill <jason at gcc dot gnu.org> 2011-05-05 15:29:03 UTC ---
Resolved by the same fix as 48873.

*** This bug has been marked as a duplicate of bug 48873 ***