Sam Gross [Sun, 13 Oct 2024 20:17:51 +0000 (16:17 -0400)]
gh-125243: Fix ZoneInfo data race in free threading build (#125281)
Lock `ZoneInfoType` to protect accesses to `ZONEINFO_STRONG_CACHE`.
Refactor the `tp_new` handler to use Argument Clinic so that we can just
use `@critical_section` annotations on the relevant functions.
Also use `PyDict_SetDefaultRef` instead of `PyDict_SetDefault` when
inserting into the `TIMEDELTA_CACHE`.
Barney Gale [Sun, 13 Oct 2024 17:46:10 +0000 (18:46 +0100)]
GH-125069: Fix inconsistent joining in `WindowsPath(PosixPath(...))` (#125156)
`PurePath.__init__()` incorrectly uses the `_raw_paths` of a given
`PurePath` object with a different flavour, even though the procedure to
join path segments can differ between flavours.
This change makes the `_raw_paths`-enabled deferred joining apply _only_
when the path flavours match.
In aeca373b3 (PR gh-12011, issue gh-71500), test_identify() was changed to expect different results on Darwin. Ned's fix was later adjusted by e52f9bee8. This workaround is only needed for some variants of Tk/Tcl on macOS, so we now allow both the workaround and the generic results for these tests.
Andrew Athan [Sun, 13 Oct 2024 07:22:05 +0000 (00:22 -0700)]
Trivial change: Update comments in activate about what running hash -r does (GH-125385)
Update comments about what running hash -r does
The old comment said "hash -r" forgets "past commands." However, the documentation for "hash" states that it forgets past locations. The old comment was, in my opinion, confusing. This is because it could be interpreted to mean it does something to the command history (HISTORY/HISTFILE etc) vs the cache of locations.
Ruoyu Zhong [Sun, 13 Oct 2024 00:53:28 +0000 (08:53 +0800)]
Doc: Fix suggested usage of `-X gil=0` in the glossary (#125366)
Currently, the "global interpreter lock" entry in the glossary mentions
that `-X gil 0` can be used to disable the GIL [1]. However, this is
invalid; the correct usage should be `-X gil=0`.
$ python -X gil 0 -c 'print("Hello, world")'
Fatal Python error: config_read_gil: PYTHON_GIL / -X gil must be "0" or "1"
Python runtime state: preinitialized
$ python -X gil=0 -c 'print("Hello, world")'
Hello, world
Stephen Rosen [Sat, 12 Oct 2024 20:21:55 +0000 (15:21 -0500)]
Prefer "similar" over "equivalent" in tutorial (#125343)
In the datastructures tutorial doc, some operations are described as
"equivalent to" others. This has led to some user-confusion -- at
least in the Discourse forums -- about cases in which the operations
differ.
This change doesn't systematically eliminate the word "equivalent"
from the tutorial. It just substitutes "similar to" in several cases
in which "equivalent to" could mislead users into expecting exact
equivalence.
Serhiy Storchaka [Sat, 12 Oct 2024 17:46:21 +0000 (20:46 +0300)]
gh-53203: Fix strptime() for %c and %x formats on many locales (GH-124946)
In some locales (like French or Hebrew) the full or abbreviated names of
the default month and weekday used in __calc_date_time can be part of
other name or constant part of the %c format. The month name can also
match %m with constant suffix (like in Japanese). So the code failed to
correctly distinguish formats %a, %A, %b, %B and %m.
Cycle all month and all days of the week to find the variable part
and distinguish %a from %A and %b from %B or %m.
Fixed locales for the following languges:
Arabic, Bislama, Breton, Bodo, Kashubian, Chuvash, Estonian, French, Irish,
Ge'ez, Gurajati, Manx Gaelic, Hebrew, Hindi, Chhattisgarhi, Haitian Kreyol,
Japanese, Kannada, Korean, Marathi, Malay, Norwegian, Nynorsk, Punjabi,
Rajasthani, Tok Pisin, Yoruba, Yue Chinese, Yau/Nungon and Chinese.
Serhiy Storchaka [Sat, 12 Oct 2024 13:04:17 +0000 (16:04 +0300)]
gh-85935: Check for nargs=0 for positional arguments in argparse (GH-124839)
Raise ValueError in add_argument() if either explicit nargs=0 or action
that does not consume arguments (like 'store_const' or 'store_true') is
specified for positional argument.
Serhiy Storchaka [Fri, 11 Oct 2024 08:43:29 +0000 (11:43 +0300)]
gh-61011: Fix inheritance of nested mutually exclusive groups in argparse (GH-125210)
Previously, all nested mutually exclusive groups lost their connection
to the group containing them and were displayed as belonging directly
to the parser.
Co-authored-by: Danica J. Sutherland <djsutherland@users.noreply.github.com>
Sam Gross [Thu, 10 Oct 2024 20:21:29 +0000 (16:21 -0400)]
gh-125268: Use static string for "1e309" in AST (#125272)
When formatting the AST as a string, infinite values are replaced by
1e309, which evaluates to infinity. The initialization of this string
replacement was not thread-safe in the free threading build.
Petr Viktorin [Thu, 10 Oct 2024 14:27:52 +0000 (16:27 +0200)]
gh-124570: ctypes: Run some Structure tests on Union as well (GH-124976)
- Move some Structure tests to test_structunion; use a common base
test class + two subclasses to run them on Union too
- Remove test_union for now as it's redundant
Note: `test_simple_structs` & `test_simple_unions` are in the common
file because they share `formats`.
* Add definitions for "context", "current context", and "context
management protocol".
* Update related definitions to be consistent with the new
definitions.
* Restructure the documentation for the `contextvars.Context` class
to prepare for adding context manager support, and for consistency
with the definitions.
* Use `testcode` and `testoutput` to test the `Context.run` example.
* Expand the documentation for the `Py_CONTEXT_EVENT_ENTER` and
`Py_CONTEXT_EVENT_EXIT` events to clarify and to prepare for
planned changes.
Eric Snow [Wed, 9 Oct 2024 17:32:16 +0000 (11:32 -0600)]
gh-116510: Fix a Crash Due to Shared Immortal Interned Strings (gh-124865)
Fix a crash caused by immortal interned strings being shared between
sub-interpreters that use basic single-phase init. In that case, the string
can be used by an interpreter that outlives the interpreter that created and
interned it. For interpreters that share obmalloc state, also share the
interned dict with the main interpreter.
This is an un-revert of gh-124646 that then addresses the Py_TRACE_REFS
failures identified by gh-124785.
Each of the `LOAD_GLOBAL` specializations is implemented roughly as:
1. Load keys version.
2. Load cached keys version.
3. Deopt if (1) and (2) don't match.
4. Load keys.
5. Load cached index into keys.
6. Load object from (4) at offset from (5).
This is not thread-safe in free-threaded builds; the keys object may be replaced
in between steps (3) and (4).
This change refactors the specializations to avoid reloading the keys object and
instead pass the keys object from guards to be consumed by downstream uops.
Alex Waygood [Wed, 9 Oct 2024 10:53:57 +0000 (11:53 +0100)]
gh-101100: Consolidate documentation on `ModuleType` attributes (#124709)
Co-authored-by: Hugo van Kemenade <1324225+hugovk@users.noreply.github.com> Co-authored-by: Barry Warsaw <barry@python.org> Co-authored-by: Jelle Zijlstra <jelle.zijlstra@gmail.com>
gh-107562: make_ssl_certs.py: produce test certificates that expire far in the future by default (GH-107594)
This allows testing Y2038 with system time set to after that,
so that actual Y2038 issues can be exposed, and not masked
by expired certificate errors.
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
Victor Stinner [Wed, 9 Oct 2024 08:15:17 +0000 (10:15 +0200)]
gh-124502: Remove _PyUnicode_EQ() function (#125114)
* Replace unicode_compare_eq() with unicode_eq().
* Use unicode_eq() in setobject.c.
* Replace _PyUnicode_EQ() with _PyUnicode_Equal().
* Remove unicode_compare_eq() and _PyUnicode_EQ().
mpage [Tue, 8 Oct 2024 14:04:35 +0000 (07:04 -0700)]
gh-115999: Stop the world when invalidating function versions (#124997)
Stop the world when invalidating function versions
The tier1 interpreter specializes `CALL` instructions based on the values
of certain function attributes (e.g. `__code__`, `__defaults__`). The tier1
interpreter uses function versions to verify that the attributes of a function
during execution of a specialization match those seen during specialization.
A function's version is initialized in `MAKE_FUNCTION` and is invalidated when
any of the critical function attributes are changed. The tier1 interpreter stores
the function version in the inline cache during specialization. A guard is used by
the specialized instruction to verify that the version of the function on the operand
stack matches the cached version (and therefore has all of the expected attributes).
It is assumed that once the guard passes, all attributes will remain unchanged
while executing the rest of the specialized instruction.
Stopping the world when invalidating function versions ensures that all critical
function attributes will remain unchanged after the function version guard passes
in free-threaded builds. It's important to note that this is only true if the remainder
of the specialized instruction does not enter and exit a stop-the-world point.
We will stop the world the first time any of the following function attributes
are mutated: