Remove PyEval_ReInitThreads() from the public C API

[vstinner]

BPO	36728
Nosy	@gpshead, @vstinner
PRs	bpo-36728: Remove PyEval_ReInitThreads() from C API #13241 bpo-36728: Remove PyEval_ReInitThreads documentation #13282

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2019-05-13.10:35:55.872>
created_at = <Date 2019-04-25.23:22:09.307>
labels = ['interpreter-core', '3.8']
title = 'Remove PyEval_ReInitThreads() from the public C API'
updated_at = <Date 2019-05-13.14:22:56.780>
user = 'https://github.com/vstinner'

bugs.python.org fields:

activity = <Date 2019-05-13.14:22:56.780>
actor = 'vstinner'
assignee = 'none'
closed = True
closed_date = <Date 2019-05-13.10:35:55.872>
closer = 'vstinner'
components = ['Interpreter Core']
creation = <Date 2019-04-25.23:22:09.307>
creator = 'vstinner'
dependencies = []
files = []
hgrepos = []
issue_num = 36728
keywords = ['patch']
message_count = 6.0
messages = ['340868', '342147', '342279', '342291', '342306', '342333']
nosy_count = 2.0
nosy_names = ['gregory.p.smith', 'vstinner']
pr_nums = ['13241', '13282']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = None
url = 'https://bugs.python.org/issue36728'
versions = ['Python 3.8']

[vstinner]

PyEval_ReInitThreads() is used internally by PyOS_AfterFork_Child(). I don't see the point of calling it directly. If you care of threads after fork, just call PyOS_AfterFork_Child(), no?

I propose to remove PyEval_ReInitThreads() from the public C API.

Problem: it's documented as a public function in High-level API of the Python Initialization API:
https://docs.python.org/dev/c-api/init.html#c.PyEval_ReInitThreads

On the Internet, I found a single project calling directly this function, but it's only in a test used to the test Python threading API:
https://github.com/DataDog/go-python3/blob/master/thread_test.go

func TestThreadInitialization(t *testing.T) {
Py_Initialize()
PyEval_InitThreads()

	assert.True(t, PyEval_ThreadsInitialized())

	PyEval_ReInitThreads()
}

I don't think that the PyEval_ReInitThreads() call here makes any sense.

[vstinner]

The function was added 19 years ago:

commit fee3a2d
Author: Guido van Rossum <guido@python.org>
Date: Sun Aug 27 17:34:07 2000 +0000

Charles Waldman's patch to reinitialize the interpreter lock after a
fork.  This solves the test_fork1 problem.  (ceval.c, signalmodule.c,
intrcheck.c)

SourceForge: [ Patch bpo-101226 ] make threading fork-safe

[gpshead]

fwiw - i agree that this can be removed. it has no reason to ever be called directly. i double checked in our massive internal codebase at work and found no calls other than python interpreters themselves.

[vstinner]

i double checked in our massive internal codebase at work and found no calls other than python interpreters themselves.

Thanks. It confirms what I understood using GitHub code search (see my initial message).

[vstinner]

New changeset d5d9e81 by Victor Stinner in branch 'master':
bpo-36728: Remove PyEval_ReInitThreads() from C API (GH-13241)
d5d9e81

[vstinner]

New changeset c1f7262 by Victor Stinner in branch 'master':
bpo-36728: Remove PyEval_ReInitThreads documentation (GH-13282)
c1f7262