[cfe-dev] is delete on abstract with non-virtal ever safe?

[David Blaikie]

On Mon, Jul 23, 2012 at 7:29 AM, Henry Miller <hank at millerfarm.com> wrote:
>
> I'm trying to settle an argument we are having at work.  Right now our
> code doesn't compile with clang because of a code like below.
>
> I understand (and agree with) the warning, however the owner of this
> code argues that only the default destructors are used, and they have
> procedures in place to ensure that no derived class will ever have
> members that need destruction.  So the question is he correct that the
> destructor for derived doesn't need to be called?

According to the letter of the standard, he is not correct:

5.3.5 [expr.delete] p3:

"if the static type of the object to be deleted is different from its
dynamic type, the static type shall be a base class of the dynamic
type of the object to be deleted and the
static type shall have a virtual destructor or the behavior is undefined."

> If so is there a way
> to shut clang up without disabling the warning (which has already found
> several real problems I've since corrected).  If not can someone point
> out why we need a virtual destructor in this case.
>
>     class base
>     {
>     public:
>       virtual void operator() () const = 0;
>     };
>
>     template<typename T>
>     class derived : public base
>     {
>     public:
>        derived(T& d) : data(d) {}
>        void operator() () const {data.DoSomething();}
>
>     private:
>        T& data; // note that the only data is a reference!
>     };
>
>     class myClass
>     {
>     public:
>        void DoSomething() {}
>     };
>
>     int main(int, char **)
>     {
>         myClass cl;
>         base* foo = new derived<myClass>(cl);
>         delete foo;
>     }
>
> clang++ ex.cpp
> ex.cpp:29:5: warning: delete called on 'base' that is abstract but has
>       non-virtual destructor [-Wdelete-non-virtual-dtor]
>     delete foo;
>
> Note, we are comparing to gcc 4.4.  I'm given to understand that gcc4.7
> also has this warning, but we don't have the ability to upgrade right
> now.
>
> --
>   Henry Miller
>   hank at millerfarm.com
> _______________________________________________
> cfe-dev mailing list
> cfe-dev at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev