bpo-46656: Remove Py_NO_NAN macro

[vstinner]

Building Python now requires support for floating point Not-a-Number
(NaN): remove the Py_NO_NAN macro.

https://bugs.python.org/issue46656

[vstinner]

Windows (x86) (pull_request) Failing after 14m — Windows (x86)

That's the urllib failure. I rebased my PR on the main branch to get the urllib test fix.

[mdickinson]

@vstinner This is a big change: historically, CPython has been relunctant to commit to requiring IEEE 754 floating-point. While requiring NaNs isn't exactly the same as requiring IEEE 754 floating-point, it's pretty close. I think this change should be discussed more widely before any change is made to the source.

[vstinner]

@vstinner This is a big change: historically, CPython has been relunctant to commit to requiring IEEE 754 floating-point. While requiring NaNs isn't exactly the same as requiring IEEE 754 floating-point, it's pretty close. I think this change should be discussed more widely before any change is made to the source.

Ok. I convert my PR to a draft to not merge it by mistake :-D

[vstinner]

Requiring IEEE 754 support is being discussed on python-dev: https://mail.python.org/archives/list/python-dev@python.org/thread/J5FSP6J4EITPY5C2UJI7HSL2GQCTCUWN/

[vstinner]

Following the python-dev discussion, I proposed a PR for PEP 7 to support C99 in the Python C API (and use a subset of C11 in Python internals): python/peps#2309

[vstinner]

Requiring IEEE 754 support is being discussed on python-dev: https://mail.python.org/archives/list/python-dev@python.org/thread/J5FSP6J4EITPY5C2UJI7HSL2GQCTCUWN/

The consensus is to require NAN and IEEE 754 to build Python 3.11. Moreover, Python 3.11 now requires a C11 compiler: #31557