gh-92898: Make _Py_Cast C++ version compatible with cast operator

[serge-sans-paille]

Use the adequate kind of cast based on the macro argument: const ref, cont
pointer, ref or pointer.

This should be compatible with C++ overloads.

Fix#92898

[serge-sans-paille]

cc @vstinner

[corona10]

Would you like to add NEWS.d about compatible?

[bedevere-bot]

A Python core developer has requested some changes be made to your pull request before we can consider merging it. If you could please address their requests along with any other requests in other reviews from core developers that would be appreciated.

Once you have made the requested changes, please leave a comment on this pull request containing the phrase I have made the requested changes; please review again. I will then notify any core developers who have left a review that you're ready for them to take another look at this pull request.

[serge-sans-paille]

@corona10 I'm not sure, there's already an entry

.. date: 2022-04-26-16-51-31 .. gh-issue: 91320 .. nonce: QDHmTv .. section: C API Fix C++ compiler warnings about "old-style cast" (``g++ -Wold-style-cast``) in the Python C API. Use C++ ``reinterpret_cast<>`` and ``static_cast<>`` casts when the Python C API is used in C++. Patch by Victor Stinner. 

And this is just patching that entry

[tacaswell]

I am now seeing errors compiling numpy with this branch:

 In file included from /home/tcaswell/.pybuild/bleeding/include/python3.12/Python.h:38, from numpy/core/include/numpy/npy_common.h:5, from numpy/core/include/numpy/npy_math.h:8, from numpy/core/src/npysort/selection.cpp:18: /home/tcaswell/.pybuild/bleeding/include/python3.12/pyport.h:31:1: error: template with C linkage 31 | template <typename type, typename expr_type> | ^~~~~~~~ numpy/core/include/numpy/npy_math.h:5:1: note: ‘extern "C"’ linkage started here 5 | extern "C"{| ^~~~~~~~~~ /home/tcaswell/.pybuild/bleeding/include/python3.12/pyport.h:36:1: error: template with C linkage 36 | template <typename type, typename expr_type> | ^~~~~~~~ numpy/core/include/numpy/npy_math.h:5:1: note: ‘extern "C"’ linkage started here 5 | extern "C"{| ^~~~~~~~~~ /home/tcaswell/.pybuild/bleeding/include/python3.12/pyport.h:41:1: error: template with C linkage 41 | template <typename type, typename expr_type> | ^~~~~~~~ numpy/core/include/numpy/npy_math.h:5:1: note: ‘extern "C"’ linkage started here 5 | extern "C"{| ^~~~~~~~~~ /home/tcaswell/.pybuild/bleeding/include/python3.12/pyport.h:46:1: error: template with C linkage 46 | template <typename type, typename expr_type> | ^~~~~~~~ numpy/core/include/numpy/npy_math.h:5:1: note: ‘extern "C"’ linkage started here 5 | extern "C"{| ^~~~~~~~~~ 

g++ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -fPIC -DNPY_INTERNAL_BUILD=1 -DHAVE_NPY_CONFIG_H=1 -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE=1 -D_LARGEFILE64_SOURCE=1 -DNO_ATLAS_INFO=1 -DHAVE_CBLAS -I/usr/local/include -I/home/tcaswell/.virtualenvs/bleeding/include -Ibuild/src.linux-x86_64-3.12/numpy/core/src/multiarray -Ibuild/src.linux-x86_64-3.12/numpy/core/src/common -Ibuild/src.linux-x86_64-3.12/numpy/core/src/umath -Inumpy/core/include -Ibuild/src.linux-x86_64-3.12/numpy/core/include/numpy -Ibuild/src.linux-x86_64-3.12/numpy/distutils/include -Ibuild/src.linux-x86_64-3.12/numpy/core/src/npysort -Inumpy/core/src/common -Inumpy/core/src -Inumpy/core -Inumpy/core/src/npymath -Inumpy/core/src/multiarray -Inumpy/core/src/umath -Inumpy/core/src/npysort -Inumpy/core/src/_simd -I/home/tcaswell/.virtualenvs/bleeding/include -I/home/tcaswell/.pybuild/bleeding/include/python3.12 -Ibuild/src.linux-x86_64-3.12/numpy/core/src/common -Ibuild/src.linux-x86_64-3.12/numpy/core/src/npymath -c numpy/core/src/npysort/selection.cpp -o build/temp.linux-x86_64-3.12/numpy/core/src/npysort/selection.o -MMD -MF build/temp.linux-x86_64-3.12/numpy/core/src/npysort/selection.o.d -msse -msse2 -msse3 -std=c++11 -D__STDC_VERSION__=0 -fno-exceptions -fno-rtti 

[serge-sans-paille]

@tacaswell indeed. I can work around this :-)

[tacaswell]

@serge-sans-paille I should note this is with numpy main, I have not gone back to see if released versions work or not.

[serge-sans-paille]

@tacaswell should be good now?

[tacaswell]

Through numpy, starting scipy. Will update if the whole environment builds cleanly (including greenlet main).

[tacaswell]

It built clean including master-branch greenlet 👍🏻

[miss-islington]

Thanks @serge-sans-paille for the PR, and @corona10 for merging it 🌮🎉.. I'm working now to backport this PR to: 3.11.
🐍🍒⛏🤖

[miss-islington]

Sorry @serge-sans-paille and @corona10, I had trouble checking out the 3.11 backport branch.
Please backport using cherry_picker on command line.
cherry_picker 5b71b519f966e1017c868ea2b27c61a5eac38c1f 3.11

[vstinner]

I created a follow-up PR: #93111

[miss-islington]

Thanks @serge-sans-paille for the PR, and @corona10 for merging it 🌮🎉.. I'm working now to backport this PR to: 3.11.
🐍🍒⛏🤖

[miss-islington]

Sorry, @serge-sans-paille and @corona10, I could not cleanly backport this to 3.11 due to a conflict.
Please backport using cherry_picker on command line.
cherry_picker 5b71b519f966e1017c868ea2b27c61a5eac38c1f 3.11