[Bug c++/25811] New: No failure creating a POD containing a const member, using new without a new-initializer.

[Bug c++/25811] New: No failure creating a POD containing a const member, using new without a new-initializer.

[fabien dot chene at laposte dot net]

$ cat podnew.cpp

struct Foo // POD
{
    double const d;
};

int main()
{
    Foo* foo = new Foo; // without the new-initializer
}

This simple code does not fail to compile.
According to the standard (5.3.4.15), I am not sure, but i think it should fail
to compile.

$ g++ podnew.cpp && echo "no failure"
no failure

I use gcc-4.0.2, but it is the same behaviour with 3.4.4.

$ g++ -v
Using built-in specs.
Target: i686-pc-linux-gnu
Configured with: /var/tmp/portage/gcc-4.0.2-r1/work/gcc-4.0.2/configure
--prefix=/usr --bindir=/usr/i686-pc-linux-gnu/gcc-bin/4.0.2
--includedir=/usr/lib/gcc/i686-pc-linux-gnu/4.0.2/include
--datadir=/usr/share/gcc-data/i686-pc-linux-gnu/4.0.2
--mandir=/usr/share/gcc-data/i686-pc-linux-gnu/4.0.2/man
--infodir=/usr/share/gcc-data/i686-pc-linux-gnu/4.0.2/info
--with-gxx-include-dir=/usr/lib/gcc/i686-pc-linux-gnu/4.0.2/include/g++-v4
--host=i686-pc-linux-gnu --build=i686-pc-linux-gnu --disable-altivec
--enable-nls --without-included-gettext --with-system-zlib --disable-checking
--disable-werror --disable-libunwind-exceptions --disable-multilib
--disable-libgcj --enable-languages=c,c++,f95 --enable-shared
--enable-threads=posix --enable-__cxa_atexit --enable-clocale=gnu
Thread model: posix
gcc version 4.0.2 (Gentoo 4.0.2-r1, pie-8.7.8)


-- 
           Summary: No failure creating a POD containing a const member,
                    using new without a new-initializer.
           Product: gcc
           Version: 4.0.2
            Status: UNCONFIRMED
          Severity: minor
          Priority: P3
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: fabien dot chene at laposte dot net
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25811

[Bug c++/25811] No failure creating a POD containing a const member, using new without a new-initializer.

[pinskia at gcc dot gnu dot org]

------- Comment #1 from pinskia at gcc dot gnu dot org  2006-01-16 20:06 -------
Confirmed, not a regression.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
           Keywords|                            |accepts-invalid
      Known to fail|                            |2.95.3 3.3.1 3.0.4 3.4.0
                   |                            |4.0.0 4.1.0 4.2.0
   Last reconfirmed|0000-00-00 00:00:00         |2006-01-16 20:06:02
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25811

[Bug c++/25811] No failure creating a POD containing a const member, using new without a new-initializer.

[fabien dot chene at gmail dot com]

------- Comment #2 from fabien dot chene at gmail dot com  2010-03-30 08:22 -------
is it still invalid in c++0X ?
5.3.4.15 has been revamped, and I no longer find a motif to reject such code.

I think the following code is also invalid, according to 8.5.6 (c++03) / 8.5.8
(c++0x):

struct A { int& i; };
void f () { new A; }

I have a patch to fix both issues.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25811

[Bug c++/25811] No failure creating a POD containing a const member, using new without a new-initializer.

[redi at gcc dot gnu dot org]

------- Comment #3 from redi at gcc dot gnu dot org  2010-03-30 09:16 -------
(In reply to comment #2)
> is it still invalid in c++0X ?

Yes.

> 5.3.4.15 has been revamped, and I no longer find a motif to reject such code.

In C++0x the object is default-initialized, which for a class type means the
default constructor is called.  In this code, the default constructor is
deleted, so the code will not compile. See 12.1/5

> I think the following code is also invalid, according to 8.5.6 (c++03) / 8.5.8
> (c++0x):
> 
> struct A { int& i; };
> void f () { new A; }

Not quite: in C++0x the program doesn't call for default-initialization of
A::i, it calls for default-initialization of A, which is invalid because the
default constructor is deleted.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25811

[Bug c++/25811] No failure creating a POD containing a const member, using new without a new-initializer.

[fabien dot chene at gmail dot com]

------- Comment #4 from fabien dot chene at gmail dot com  2010-03-30 10:07 -------
(In reply to comment #3)
> (In reply to comment #2)
> > is it still invalid in c++0X ?
> 
> Yes.
> 
> > 5.3.4.15 has been revamped, and I no longer find a motif to reject such code.
> 
> In C++0x the object is default-initialized, which for a class type means the
> default constructor is called.  In this code, the default constructor is
> deleted, so the code will not compile. See 12.1/5

OK thanks.

> > I think the following code is also invalid, according to 8.5.6 (c++03) / 8.5.8
> > (c++0x):
> > 
> > struct A { int& i; };
> > void f () { new A; }
> 
> Not quite: in C++0x the program doesn't call for default-initialization of
> A::i, it calls for default-initialization of A, which is invalid because the
> default constructor is deleted.

OK, but it appears that the C++0X part is not yet implemented in GCC -- the
constructor is not deleted in this case.
Nevertheless, can you confirm that it is valid C++03 ?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25811

[Bug c++/25811] No failure creating a POD containing a const member, using new without a new-initializer.

[fabien dot chene at gmail dot com]

------- Comment #5 from fabien dot chene at gmail dot com  2010-03-30 10:10 -------

> Nevertheless, can you confirm that it is valid C++03 ?

I mean invalid, sorry.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25811

[Bug c++/25811] No failure creating a POD containing a const member, using new without a new-initializer.

[redi at gcc dot gnu dot org]

------- Comment #6 from redi at gcc dot gnu dot org  2010-03-30 11:19 -------
(In reply to comment #5)
> > Nevertheless, can you confirm that it is valid C++03 ?
> 
> I mean invalid, sorry.

Yup :-)

It is invalid.  A is a non-POD class type, so 5.3.4/15 says the new-expression
without a new-initializer causes the object to be default-initialized, which
causes a default constructor to be implicitly-defined with an empty
mem-initializer list (12.1/7) which is ill-formed by 8.5/5 because the
reference member is not initialized.


In C++03 the cases of Foo and A are slightly different, "new Foo" is ill-formed
according to 5.3.4/15 and "new A" is ill-formed as described above.

In C++0x both "new Foo" and "new A" result in a call to a deleted constructor.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25811

[Bug c++/25811] No failure creating a POD containing a const member, using new without a new-initializer.

[fabien dot chene at gmail dot com]

------- Comment #7 from fabien dot chene at gmail dot com  2010-03-31 07:06 -------
(In reply to comment #6)
> (In reply to comment #5)
> > > Nevertheless, can you confirm that it is valid C++03 ?
> > 
> > I mean invalid, sorry.
> 
> Yup :-)
> 
> It is invalid.  A is a non-POD class type, so 5.3.4/15 says the new-expression
> without a new-initializer causes the object to be default-initialized, which
> causes a default constructor to be implicitly-defined with an empty
> mem-initializer list (12.1/7) which is ill-formed by 8.5/5 because the
> reference member is not initialized.
> 
> 
> In C++03 the cases of Foo and A are slightly different, "new Foo" is ill-formed
> according to 5.3.4/15 and "new A" is ill-formed as described above.

Thanks for the clarification !
Patch here for the non C++0X part: 
http://gcc.gnu.org/ml/gcc-patches/2010-03/msg01467.html

> In C++0x both "new Foo" and "new A" result in a call to a deleted constructor.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25811

[Bug c++/25811] No failure creating a POD containing a const member, using new without a new-initializer.

[redi at gcc dot gnu dot org]

------- Comment #8 from redi at gcc dot gnu dot org  2010-03-31 08:55 -------
(N.B. your email to gcc-patches gives the wrong PR number in the subject.)

This reject the following valid program:

struct X {
  X() : c(0), r(c) {}
  const int c;
  int const& r;
};

int main()
{
  new X;
}

If there is a default constructor it will be called, and so the members will
not be uninitialized.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25811

[Bug c++/25811] No failure creating a POD containing a const member, using new without a new-initializer.

[fabien dot chene at gmail dot com]

------- Comment #9 from fabien dot chene at gmail dot com  2010-03-31 12:20 -------
(In reply to comment #8)
> (N.B. your email to gcc-patches gives the wrong PR number in the subject.)
> 
> This reject the following valid program:
> 
> struct X {
>   X() : c(0), r(c) {}
>   const int c;
>   int const& r;
> };
> 
> int main()
> {
>   new X;
> }
> 
> If there is a default constructor it will be called, and so the members will
> not be uninitialized.

OK, I've revised the patch to check for a user defined constructor.
I have also added your testcase, and others dealing with arrays.

http://gcc.gnu.org/ml/gcc-patches/2010-03/msg01486.html


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25811

[Bug c++/25811] No failure creating a POD containing a const member, using new without a new-initializer.

[redi at gcc dot gnu dot org]

------- Comment #10 from redi at gcc dot gnu dot org  2010-03-31 12:34 -------
Thanks. I can't test it now but that looks better.

Just below your new code I see this error:

    error ("uninitialized const in %<new%> of %q#T", elt_type);

Maybe your new diagnostics should also include 'new' like that.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25811

[Bug c++/25811] No failure creating a POD containing a const member, using new without a new-initializer.

[fabien dot chene at gmail dot com]

------- Comment #11 from fabien dot chene at gmail dot com  2010-03-31 13:46 -------
(In reply to comment #10)
> Thanks. I can't test it now but that looks better.
> 
> Just below your new code I see this error:
> 
>     error ("uninitialized const in %<new%> of %q#T", elt_type);
> 
> Maybe your new diagnostics should also include 'new' like that.

Would 'error ("uninitialized const member in %q#T using %<new%>", elt_type)'
suit you ?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25811

[Bug c++/25811] No failure creating a POD containing a const member, using new without a new-initializer.

[redi at gcc dot gnu dot org]

------- Comment #12 from redi at gcc dot gnu dot org  2010-03-31 13:52 -------
That works for me, although I should point out I can't approve the patch, so
it's just my opinion


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25811

[Bug c++/25811] No failure creating a POD containing a const member, using new without a new-initializer.

[jason at gcc dot gnu dot org]

------- Comment #13 from jason at gcc dot gnu dot org  2010-04-12 19:58 -------
Subject: Bug 25811

Author: jason
Date: Mon Apr 12 19:58:27 2010
New Revision: 158239

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=158239
Log:
        PR c++/25811
        * cp-tree.h (diagnose_uninitialized_cst_or_ref_member): Declare.
        * init.c (build_new_1): Check for uninitialized const members and
        uninitialized reference members, when using new without
        new-initializer. Call diagnose_uninitialized_cst_or_ref_member.
        (diagnose_uninitialized_cst_or_ref_member): Define, call
        diagnose_uninitialized_cst_or_ref_member_1.
        (diagnose_uninitialized_cst_or_ref_member_1): New function.

Added:
    trunk/gcc/testsuite/g++.dg/init/pr25811.C
Modified:
    trunk/gcc/cp/ChangeLog
    trunk/gcc/cp/cp-tree.h
    trunk/gcc/cp/init.c
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25811

[Bug c++/25811] No failure creating a POD containing a const member, using new without a new-initializer.

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25811

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
                 CC|gcc-bugs at gcc dot gnu.org |
         Resolution|                            |FIXED
      Known to fail|                            |

--- Comment #14 from Paolo Carlini <paolo.carlini at oracle dot com> 2012-05-24 22:36:47 UTC ---
So this fixed, I think already in 4.6.0.