[Bug c++/105752] New: Template function can access private member

[Bug c++/105752] New: Template function can access private member

[csaba_22 at yahoo dot co.uk]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105752

            Bug ID: 105752
           Summary: Template function can access private member
           Product: gcc
           Version: 12.1.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: csaba_22 at yahoo dot co.uk
  Target Milestone: ---

The following code

class Outer
{
private: //  !!!!!
    struct Inner
    {};
};

template<typename T>
struct Meow
{
    void purr() {
        Outer::Inner oi;
    }
};

int main()
{
    Meow<double> kitty;
    //kitty.purr();
}

is compiled by GCC from 4.1.2 to 12.1.0, but is rejected by all clang releases 
 available on godbolt.

https://godbolt.org/z/58Yca6dah

Also mentioned here:
https://stackoverflow.com/questions/28596242/accessing-private-inner-class-type-from-non-member-template-function

If the call to the function is uncommented, then GCC also emits an error.

Visual Studio 2013 update 4 also  emits an  error (allegedly), so this sounds
like a bug.

[Bug c++/105752] Template function can access private member

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105752

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
There might already been a dup of this. Gcc does not do access checking until
instantiation time (as you saw) even for non dependent things.

[Bug c++/105752] Template function can access private member

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105752

--- Comment #2 from Jonathan Wakely <redi at gcc dot gnu.org> ---
Which is permitted by the standard.

[Bug c++/105752] Template function can access private member

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105752

--- Comment #3 from Jonathan Wakely <redi at gcc dot gnu.org> ---
And if you never instantiate the function, who cares? It doesn't actually
access the private member, because you never call it. And if you try to, it
doesn't compile.

[Bug c++/105752] Template function can access private member

[csaba_22 at yahoo dot co.uk]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105752

--- Comment #4 from Csaba Ráduly <csaba_22 at yahoo dot co.uk> ---
Looks like there *was* a  bug, I just wasn't  able to properly reproduce it
initially:

#include <type_traits>
#include <utility>

class CB {
    struct DCB {};
};

struct  NMC1 : public CB {
    int meow() const { return __LINE__; }
};

struct CC {
    template <typename ... Args>
    int  meow(Args&&... args) {
        if constexpr(std::is_same_v<
decltype(std::declval<NMC1>().meow(std::forward<Args>(args)...)), CB::DCB>) {
            return __LINE__;
        }
        else {
            return __LINE__;
        }
    }
};

int main()
{
    CC  cc;
    return  cc.meow();
}

(accepted  by GCC 10, rejected by clang and GCC11).

https://godbolt.org/z/81rWdf6v8

This has been fixed  between GCC 10.3 and 11.1

[Bug c++/105752] Template function can access private member

[csaba_22 at yahoo dot co.uk]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105752

--- Comment #5 from Csaba Ráduly <csaba_22 at yahoo dot co.uk> ---
The fact that GCC 11 shows an error for the code in comment #4 suggests that
this was a bug which was fixed. Does it ring a bell to anyone? This bug should
probably marked as a duplicate, but of which other bug?

[Bug c++/105752] Template function can access private member

[ppalka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105752

Patrick Palka <ppalka at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |DUPLICATE
                 CC|                            |ppalka at gcc dot gnu.org

--- Comment #6 from Patrick Palka <ppalka at gcc dot gnu.org> ---
(In reply to Csaba Ráduly from comment #5)
> The fact that GCC 11 shows an error for the code in comment #4 suggests that
> this was a bug which was fixed. Does it ring a bell to anyone? This bug
> should probably marked as a duplicate, but of which other bug?

Looks like we started to correctly reject the comment #4 testcase after
r11-6800-g29853c653245c3.  Since the testcase is pretty similar to the one from
PR58993, I guess we can mark this a dup of that PR.

*** This bug has been marked as a duplicate of bug 58993 ***