[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58876

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |diagnostic
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2013-10-30
          Component|c++                         |libstdc++
     Ever confirmed|0                           |1

--- Comment #1 from Jonathan Wakely <redi at gcc dot gnu.org> ---
This is because warnings are suppressed by default in system headers, and the
undefined delete operation occurs in a system header.  You get the warning if
you use -Wsystem-headers

I don't see an easy way to force the warning to always be emitted though.

[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58876

--- Comment #3 from Paolo Carlini <paolo.carlini at oracle dot com> ---
At some point Ian Taylor filed a Bugzilla about these issues, I think it's
still open. Not sure what we should do in this area...

[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58876

--- Comment #4 from Jonathan Wakely <redi at gcc dot gnu.org> ---
Yes, I should dig Ian's bug out and have another look. I'm planning to throw
some ideas around on the mailing list ...

[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58876

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED

[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58876
Bug 58876 depends on bug 61600, which changed state.

Bug 61600 Summary: #pragma GCC diagnostic pop  leaves warnings enabled
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61600

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |DUPLICATE

[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[mw_triad at users dot sourceforge.net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58876

Matthew Woehlke <mw_triad at users dot sourceforge.net> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mw_triad at users dot sourceforge.
                   |                            |net

--- Comment #5 from Matthew Woehlke <mw_triad at users dot sourceforge.net> ---
See also bug 64399, which proposes that a) the conversion itself should
generate a warning, and b) the presence of other virtual methods in A should
not be required for the warning to trip. (This could be achieved by something
like static_assert except to emit a warning, combined with
std::has_virtual_destructor, without otherwise having to fiddle with pragmas.)

Actually, this may be required for 'make_unique<A>(new B)' to warn, since the
conversion of a B* ('new B') to an A* (which is what is passed to make_unique /
unique_ptr::unique_ptr) should not warn. (IOW, unique_ptr / make_unique would
need overloads taking any pointer type and doing the conversion inside STL so
that std::has_virtual_destructor can be checked against the actual pointer
type.)

...or alternatively gcc would need to detect when a converted pointer is passed
to unique_ptr / make_unique, which seems like it would be harder.

[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58876

--- Comment #6 from Jonathan Wakely <redi at gcc dot gnu.org> ---
(In reply to Matthew Woehlke from comment #5)
> Actually, this may be required for 'make_unique<A>(new B)' to warn, since

That's not how make_unique works.

[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[mw_triad at users dot sourceforge.net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58876

--- Comment #7 from Matthew Woehlke <mw_triad at users dot sourceforge.net> ---
(In reply to Jonathan Wakely from comment #6)
> (In reply to Matthew Woehlke from comment #5)
> > Actually, this may be required for 'make_unique<A>(new B)' to warn, since
> 
> That's not how make_unique works.

...and I'm suggesting it *should* be. (How else are you going to warn? After
that executes, the pointer no longer knows that it really contains a B, unless
you teach the compiler some fancy extra tricks, which seems overly complicated.
Conversely, I feel that 'make_unique<A>(new B)' should warn if it's going to
result in failing to call B's dtor. I might even go so far as to say 'even if
the compiler can prove that B's dtor is trivial', though I'd be willing to
delegate that to a different and more pedantic warning.)

[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58876

--- Comment #8 from Jonathan Wakely <redi at gcc dot gnu.org> ---

No, really, that's not how make_unique works. You do not use 'new' with
make_unique, that's the whole point, so you would say make_unique<B>() to
create a B. Your motivating examples should be valid C++ of you want to
convince anyone, so maybe:

unique_ptr<A> p = make_unique<B>();

[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[mw_triad at users dot sourceforge.net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58876

--- Comment #9 from Matthew Woehlke <mw_triad at users dot sourceforge.net> ---
(In reply to Jonathan Wakely from comment #8)
> No, really, that's not how make_unique works. You do not use 'new' with
> make_unique, that's the whole point [...]

D'oh, sorry :-). Not sure what I was thinking. I think I meant
'unique_ptr<A>(new B)', like to the original example. That said...

> unique_ptr<A> p = make_unique<B>();

...this is also a good example that I would like to see warn. (I think this has
the same problems; the warning would need to trigger in the conversion
operator, otherwise the knowledge of the true type is lost by the time the
unique_ptr<A> is destroyed.)

[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[db0451 at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58876

DB <db0451 at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |db0451 at gmail dot com

--- Comment #10 from DB <db0451 at gmail dot com> ---
Sorry to pester, but is this likely to get anywhere, any time soon? I fixed a
bug in one of my projects recently by using AddressSanitizer, which this
would've caught if it warned on non-polymorphic deletion via unique_ptr<base>,
so it'd be nice to see.

I presume the same goes for std::shared_ptr, too.

[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58876

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |NEW
           Assignee|redi at gcc dot gnu.org            |unassigned at gcc dot gnu.org

[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58876

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           See Also|                            |https://gcc.gnu.org/bugzill
                   |                            |a/show_bug.cgi?id=64399

--- Comment #11 from Jonathan Wakely <redi at gcc dot gnu.org> ---
(In reply to DB from comment #10)
> Sorry to pester, but is this likely to get anywhere, any time soon?

No, probably not. Comment 2 doesn't work because -Wsystem-headers can't be
enabled and disabled using pragmas. It doesn't work like other warnings.

I don't think this can be fixed in the library.


> I presume the same goes for std::shared_ptr, too.

No, because shared_ptr type-erases the deleter and so deletes the dynamic type
even if the stored pointer is upcast to a different static type.

[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[harald at gigawatt dot nl]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58876

Harald van Dijk <harald at gigawatt dot nl> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |harald at gigawatt dot nl

--- Comment #12 from Harald van Dijk <harald at gigawatt dot nl> ---
(In reply to Jonathan Wakely from comment #11)
> No, probably not. Comment 2 doesn't work because -Wsystem-headers can't be
> enabled and disabled using pragmas. It doesn't work like other warnings.

However, the internal version of the #line directive, # [line number] [file
name] [flags] could be used to mark a region of a system header as non-system
header, which should achieve the same result, right? It might need a bit of
cleanup to be maintainable, but this seems to work as a proof of concept:

--- bits/unique_ptr.h
+++ bits/unique_ptr.h
@@ -82,7 +82,9 @@
                     "can't delete pointer to incomplete type");
        static_assert(sizeof(_Tp)>0,
                     "can't delete pointer to incomplete type");
+# 86 __FILE__
        delete __ptr;
+# 88 __FILE__ 3
       }
     };

[Bug libstdc++/58876] No non-virtual-dtor warning in std::unique_ptr

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58876

--- Comment #13 from Jonathan Wakely <redi at gcc dot gnu.org> ---
It's so crazy, it just might work.