Python.h includes stdbool.h

[encukou]

BPO	46748
Nosy	@vstinner, @encukou, @ericsnowcurrently, @serhiy-storchaka, @brandtbucher, @erlend-aasland, @kumaraditya303
PRs	bpo-46748: Don't import <stdbool.h> in public headers #31553 bpo-46748: fix failing ctypes tests on buildbots #31600

^{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 2022-02-28.09:48:40.970>
created_at = <Date 2022-02-14.09:28:19.600>
labels = ['3.11']
title = 'Python.h includes stdbool.h'
updated_at = <Date 2022-02-28.15:02:34.507>
user = 'https://github.com/encukou'

bugs.python.org fields:

activity = <Date 2022-02-28.15:02:34.507>
actor = 'vstinner'
assignee = 'none'
closed = True
closed_date = <Date 2022-02-28.09:48:40.970>
closer = 'petr.viktorin'
components = []
creation = <Date 2022-02-14.09:28:19.600>
creator = 'petr.viktorin'
dependencies = []
files = []
hgrepos = []
issue_num = 46748
keywords = ['patch']
message_count = 11.0
messages = ['413216', '413217', '413220', '413235', '413239', '413970', '414053', '414121', '414165', '414183', '414200']
nosy_count = 7.0
nosy_names = ['vstinner', 'petr.viktorin', 'eric.snow', 'serhiy.storchaka', 'brandtbucher', 'erlendaasland', 'kumaraditya']
pr_nums = ['31553', '31600']
priority = 'normal'
resolution = None
stage = 'resolved'
status = 'closed'
superseder = None
type = None
url = 'https://bugs.python.org/issue46748'
versions = ['Python 3.11']

[encukou]

In main, cpython/pystate.h newly includes stdbool.h, providing a definition for bool that might be incompatible with other software.

See here: cmusphinx/sphinxbase#90

Eric, is this necessary? Would an old-school int do?
Or should we say it's 2022 already and everyone needs to use stdbool.hfore bools?

[serhiy-storchaka]

Is it compatible with C++?

[encukou]

Yes, stdbool.h is essentially a backport of C++'s bool type to C.

[ericsnowcurrently]

On Mon, Feb 14, 2022 at 2:28 AM Petr Viktorin report@bugs.python.org wrote:

Eric, is this necessary? Would an old-school int do?
Or should we say it's 2022 already and everyone needs to use stdbool.hfore bools?

I started using bool (stdbool.h) when I saw it specified in PEP-7
(https://www.python.org/dev/peps/pep-0007/#c-dialect). If it is a
problem then I think it should be answered relative to PEP-7. I'm not
sure what the best route is though. Personally, I'd argue that if
it's in C99 then it should probably be standard at this point. :)
However, I'm not opposed to going back to plain int. (And I know
there are definitely a number of old-school folks who prefer plain
int. :) )

[encukou]

It is in C99, but in an optional header.
IMO, including the header in the internals is perfectly OK, but opting everyone else in isn't very nice.
i.e. it would probably be OK to use _Bool, the actual C99 keyword that bool is defined as. But that's ugly enough to prefer int...

[encukou]

New changeset 2c228a7 by Petr Viktorin in branch 'main':
bpo-46748: Don't import <stdbool.h> in public headers (GH-31553)
2c228a7

[vstinner]

New changeset 2c228a7 by Petr Viktorin in branch 'main':
bpo-46748: Don't import <stdbool.h> in public headers (GH-31553)

It seems like this change broke ctypes on some architectures like s390x:

https://buildbot.python.org/all/#/builders/3/builds/1573

======================================================================
FAIL: test_frozentable (ctypes.test.test_values.PythonValuesTestCase) [phello_alias]
----------------------------------------------------------------------

Traceback (most recent call last):
  File "/home/dje/cpython-buildarea/3.x.edelsohn-fedora-z.clang/build/Lib/ctypes/test/test_values.py", line 86, in test_frozentable
    self.assertIsNone(spec.submodule_search_locations)
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: [] is not None

======================================================================
FAIL: test_frozentable (ctypes.test.test_values.PythonValuesTestCase) [phello]
----------------------------------------------------------------------

Traceback (most recent call last):
  File "/home/dje/cpython-buildarea/3.x.edelsohn-fedora-z.clang/build/Lib/ctypes/test/test_values.py", line 86, in test_frozentable
    self.assertIsNone(spec.submodule_search_locations)
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: ['/home/dje/cpython-buildarea/3.x.edelsohn-fedora-z.clang/build/Lib/__phello__'] is not None

======================================================================
FAIL: test_frozentable (ctypes.test.test_values.PythonValuesTestCase) [phello.ham]
----------------------------------------------------------------------

Traceback (most recent call last):
  File "/home/dje/cpython-buildarea/3.x.edelsohn-fedora-z.clang/build/Lib/ctypes/test/test_values.py", line 86, in test_frozentable
    self.assertIsNone(spec.submodule_search_locations)
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: ['/home/dje/cpython-buildarea/3.x.edelsohn-fedora-z.clang/build/Lib/__phello__/ham'] is not None

[vstinner]

New changeset e182c66 by Kumar Aditya in branch 'main':
bpo-46748: Fix ctypes test_frozentable() (GH-31600)
e182c66

[erlend-aasland]

Can we close this?

[encukou]

Thank you, Kumar & Victor, for fixing up the issue!

I meant to check the buildbots before closing the issue, but got side-tracked.

[vstinner]

Thank you, Kumar & Victor, for fixing up the issue!

You're welcome. I'm happy to see that you reduced the number of #include in the C API ;-) I made similar changes in bpo-45434.