bpo-46541: Remove usage of _Py_IDENTIFIER from array module by corona10 · Pull Request #31376 · python/cpython

Raised RLIMIT_NOFILE: 256 -> 1024
0:00:00 load avg: 4.95 Run tests sequentially
0:00:00 load avg: 4.95 [1/1] test_array
beginning 6 repetitions
123456
......

== Tests result: SUCCESS ==

1 test OK.

Total duration: 3.3 sec
Tests result: SUCCESS

@ericsnowcurrently Should we create the separate bpo issue if module changes are too noisy?

This is an extension module so let's avoid using _Py_ID(). I see the module defines Py_BUILD_CORE_MODULE, but ideally we wouldn't. So let's not expand the reliance on internal API.

@ericsnowcurrently @erlend-aasland
PTAL

I try to remove the usage of Py_BUILD_CORE_BUILTIN, but things should be discussed before removing them.
So let's handle them as separate issues.

I try to remove the usage of Py_BUILD_CORE_BUILTIN, but things needed to be discussed before removing them. So let's handle them as separate issues.

Yeah, something like that should be handled separately, after any relevant discussion. Note that not everyone may agree with me that stdlib extensions should not use internal API. It may be worth asking on python-dev.

This is an extension module so let's avoid using _Py_ID(). I see the module defines Py_BUILD_CORE_MODULE, but ideally we wouldn't. So let's not expand the reliance on internal API.

FYI, relevant bpo's:

• https://bugs.python.org/issue43974
• https://bugs.python.org/issue41111

As far as I can remember, there has not been any discussion on python-dev or Discourse (yet). I'm pretty sure Victor is interested in starting such a discussion. He's off CPython this week, but we can hear with him sometime next week.

BTW, this PR looks good to me :)

FYI, relevant bpo's:

Thanks for the links!

As far as I can remember, there has not been any discussion on python-dev or Discourse (yet). I'm pretty sure Victor is interested in starting such a discussion. He's off CPython this week, but we can hear with him sometime next week.

Sounds good.