[Bug c++/105606] New: [12 Regression] std::pair with nested struct and NSDMI

[Bug c++/105606] New: [12 Regression] std::pair with nested struct and NSDMI

[mrjoel at lixil dot net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105606

            Bug ID: 105606
           Summary: [12 Regression] std::pair with nested struct and NSDMI
           Product: gcc
           Version: 12.1.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: mrjoel at lixil dot net
  Target Milestone: ---

The following snippets fail to compile with GCC 12.1.0. The first relates more
to actual code usage while the second I believe to be a more minimal
reproducer.

This looks closely related to the examples and discussion in bug 96645 and yet
this case compiles in versions through 11, only failing to compile in 12.

It may also be closely related to (or have it as an underlying cause) bug
102199.

Snippet 1
=========

#include <map>
#include <utility>

struct Outer {
    struct Inner {
        int a{0};
    };

    std::pair<Inner, int> p;

    void f() {
        std::map<int, std::pair<Inner, int>> m;
        m[0] = p;
    }
};

Snippet 2
=========

#include <utility>

struct Outer {
    struct Inner {
        int a{0}; // removing initialization compiles
    };

    std::pair<Inner, int> p; // commenting out compiles
    void f() {
        [[maybe_unused]] std::pair<Inner, int> localp;
    }
};

[Bug c++/105606] [12 Regression] std::pair with nested struct and NSDMI

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105606

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

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

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Both examples compile just fine with the released GCC 12. You must be using a
pre-released version when testing.

*** This bug has been marked as a duplicate of bug 96645 ***

[Bug c++/105606] [12 Regression] std::pair with nested struct and NSDMI

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105606

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

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

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Wait you didn't mention you needed -std=c++20 to get the failure.

[Bug c++/105606] [12 Regression] std::pair with nested struct and NSDMI

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105606

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Note clang also rejects it for -std=c++20 and accepts it for -std=c++17.

Let me dig into the changes, I suspect this is another standard change.

[Bug c++/105606] [12 Regression] std::pair with nested struct and NSDMI and -std=c++20

[mrjoel at lixil dot net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105606

--- Comment #4 from Joel Johnson <mrjoel at lixil dot net> ---
> Wait you didn't mention you needed -std=c++20 to get the failure.

Ah indeed, my apologies!

I'm unable to reproduce the Clang failure using -std=c++20, at least in a quick
godbolt check at https://godbolt.org/z/3W4xeYdWs.

[Bug c++/105606] [12 Regression] std::pair with nested struct and NSDMI and -std=c++20

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105606

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

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

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
There was libstdc++ change to better support C++20.

*** This bug has been marked as a duplicate of bug 102199 ***

[Bug c++/105606] [12 Regression] std::pair with nested struct and NSDMI and -std=c++20

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105606

--- Comment #6 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Joel Johnson from comment #4)
> > Wait you didn't mention you needed -std=c++20 to get the failure.
> 
> Ah indeed, my apologies!
> 
> I'm unable to reproduce the Clang failure using -std=c++20, at least in a
> quick godbolt check at https://godbolt.org/z/3W4xeYdWs.

Try clang trunk. clang 14 in godbolt uses the libstdc++ headers from GCC
11.2.0.

[Bug c++/105606] [12 Regression] std::pair with nested struct and NSDMI and -std=c++20

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105606

--- Comment #7 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Jakub Jelinek <jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:3ba882c7b51ab1f14c62c748e989415834ccd9ce

commit r14-4293-g3ba882c7b51ab1f14c62c748e989415834ccd9ce
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Wed Sep 27 10:38:54 2023 +0200

    remove workaround for GCC 4.1-4.3 [PR105606]

    While looking into vec.h, I've noticed we still have a workaround for
    GCC 4.1-4.3 bugs.
    As we now use C++11 and thus need to be built by GCC 4.8 or later,
    I think this is now never used.

    2023-09-27  Jakub Jelinek  <jakub@redhat.com>

            PR c++/105606
            * system.h (BROKEN_VALUE_INITIALIZATION): Don't define.
            * vec.h (vec_default_construct): Remove BROKEN_VALUE_INITIALIZATION
            workaround.
            * function.cc (assign_parm_find_data_types): Likewise.