ForkOn 19/11/18 23:27, Jason Li wrote:
> The *self-contained test* which does not depend on STL:
>
> #include <iostream> int *getPtr(int& x) { std::cout << "getPtr(" << x <<
> ")\n"; // Artificial side effect return &x; } int main() { int zero = 0,
> one = 1; *getPtr(zero) = *getPtr(one); }
>
> // Similar tests can be done with other operators using this technique
>
> Expected stdout:
>
> getPtr(1)
> getPtr(0)
>
> MinGW (with -std=c++17) stdout:
>
> getPtr(0)
> getPtr(1)
FWIW, I can reproduce this with MinGW GCC-6.3.0 [*], as per your info:
> Version info:
>
> Operating system: Windows 10 Pro 1803, 17134.407
>
> gcc version 6.3.0 (MinGW.org GCC-6.3.0-1)
$ use mingw32-gcc-6.3.0
$ mingw32-g++ --std=c++17 foo.cc
$ ./a.exe ; rstty
getPtr(0)
getPtr(1)
However, with the *current* MinGW GCC-7.3.0 [*]:
$ use mingw32-gcc-7.3.0
$ mingw32-g++ --std=c++17 foo.cc
$ ./a.exe ; rstty
getPtr(1)
getPtr(0)
and with my latest experimental MinGW GCC-8.2.0 [*]:
$ use mingw32-gcc-8.2.0
$ mingw32-g++ --std=c++17 foo.cc
$ ./a.exe ; rstty
getPtr(1)
getPtr(0)
From the above, I deduce that this is an upstream GCC issue, resolved
from the release of GCC-7 onwards, (at least since GCC-7.3.0).
[*] All of my tests result from cross-compiling your sample code, and
running it under wine; the "rstty" is a local hack to work around a wine
bug, which trashes my TTY keymap configuration after running any wine
process (directly) in my console.
--
Regards,
Keith.
Public key available from keys.gnupg.net
Key fingerprint: C19E C018 1547 DE50 E1D4 8F53 C0AD 36C6 347E 5A3F
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: OpenPGP digital signature
URL: <https://lists.osdn.me/mailman/archives/mingw-users/attachments/20181120/459550cc/attachment.sig>