If multiple threads concurrently imported `collections.abc`, some of the
threads might incorrectly see the "shim" `Lib/collections/abc.py` module
instead of the correct `Lib/_collections_abc.py` module. This affected
both the free threading build and the default GIL-enabled build.
* Remove `@suppress_immortalization` decorator
* Make suppression flag per-thread instead of per-interpreter
* Suppress immortalization in `eval()` to avoid refleaks in three tests
(test_datetime.test_roundtrip, test_logging.test_config8_ok, and
test_random.test_after_fork).
* frozenset() is constant, but not a singleton. When run multiple times,
the test could fail due to constant interning.
This replaces `_PyEval_BuiltinsFromGlobals` with
`_PyDict_LoadBuiltinsFromGlobals`, which returns a new reference
instead of a borrowed reference. Internally, the new function uses
per-thread reference counting when possible to avoid contention on the
refcount fields on the builtins module.
On Windows, `long` is a signed 32-bit integer so it can't represent
`0xffff_ffff` without overflow. Windows exit codes are unsigned 32-bit
integers, so if a child process exits with `-1`, it will be represented
as `0xffff_ffff`.
Also fix a number of other possible cases where `_Py_HandleSystemExit`
could return with an exception set, leading to a `SystemError` (or
fatal error in debug builds) later on during shutdown.
This fixes a crash when `gc.get_objects()` or `gc.get_referrers()` is
called during a GC in the free threading build.
Switch to `_PyObjectStack` to avoid corrupting the `struct worklist`
linked list maintained by the GC. Also, don't return objects that are frozen
(`gc.freeze()`) or in the process of being collected to more closely match
the behavior of the default build.
They used to be shared, before 3.12. Returning to sharing them resolves a failure on Py_TRACE_REFS builds.
Co-authored-by: Petr Viktorin <encukou@gmail.com>
* The parser no longer changes temporarily during parsing.
* Default values are not processed twice.
* Required mutually exclusive groups containing positional arguments are
now supported.
* The missing arguments report now includes the names of all required
optional and positional arguments.
* Unknown options can be intermixed with positional arguments in
parse_known_intermixed_args().
This is a follow up of GH-124974. Only Glibc needed a fix.
Now the returned value is a string consisting of semicolon-separated
symbols on all Posix platforms.
Previously, copying a super object returned a copy of the instance
invoking super(). Pickling a super object could pickle the instance
invoking super() or fail, depending on its type and protocol.
Now deep copying returns a new super object and pickling pickles the super
object. Shallow copying returns the same super object.
* Fix usage of PyStackRef_FromPyObjectSteal in CALL_TUPLE_1
This was missed in gh-124894
* Fix usage of PyStackRef_FromPyObjectSteal in _CALL_STR_1
This was missed in gh-124894
* Regenerate code
It looks like commit 43cf44ddcc
(gh-31501) accidentally moved the paragraph to the `tp_finalize`
section when the intent was to move it to the `tp_dealloc` section
(according to the commit message).
Also:
* Convert the paragraph to a warning.
* Apply the appropriate font style to `tp_dealloc`.
* Unlinkify the first mention of `tp_dealloc` since the paragraph is
already in the `tp_dealloc` section.
Dictionary watchers on an object's attributes dictionary
(`object.__dict__`) were not triggered when the managed dictionary used
the object's inline values.
Malcolm Smith
Sam Gross
Kerim Kabirov
Shantanu
Javad Shafique
Peter Bierma
partev
Mark Shannon
Marat Sharafutdinov
Jelle Zijlstra
Tian Gao
Petr Viktorin
Wulian
Eric Snow
Irit Katriel
Serhiy Storchaka
Jacob Walls
Mario Šaško
Ethan Furman
Mikhail Efimov
Adam Turner
Arjun Singh
Y5
mpage
Richard Hansen
Pablo Galindo Salgado
Lumír 'Frenzy' Balhar
ember91