Ronald Oussoren [Sun, 10 Dec 2023 11:38:49 +0000 (12:38 +0100)]
gh-109980: Fix test_tarfile_vs_tar on macOS (#112905)
On recentish macOS versions the system tar
command includes system metadata (ACLs, extended attributes
and resource forks) in the tar archive, which
shutil.make_archive will not do. This can cause
spurious test failures.
Barney Gale [Sat, 9 Dec 2023 15:07:40 +0000 (15:07 +0000)]
GH-110109: Move pathlib ABCs to new `pathlib._abc` module. (#112881)
Move `_PurePathBase` and `_PathBase` to a new `pathlib._abc` module, and
drop the underscores from the class names.
Tests are mostly left alone in this commit, but they'll be similarly split
in a subsequent commit.
The `pathlib._abc` module will be published as an independent PyPI package
(similar to how `zipfile._path` is published as `zipp`), to be refined
and stabilised prior to its possible addition to the standard library.
gh-112334: Regression test that vfork is used when expected. (#112734)
Regression test that vfork is used when expected by subprocess.
This is written integration test style, it uses strace if it is present and appears to work to find out what system call actually gets used in different scenarios.
Test coverage is added for the default behavior and that of each of the specific arguments that must disable the use of vfork. obviously not an entire test matrix, but it covers the most important aspects.
If there are ever issues with this test being flaky or failing on new platforms, rather than try and adapt it for all possible platforms, feel free to narrow the range it gets tested on when appropriate. That is not likely to reduce coverage.
Barney Gale [Fri, 8 Dec 2023 17:39:04 +0000 (17:39 +0000)]
GH-110109: Add `pathlib._PurePathBase` (#110670)
Add private `pathlib._PurePathBase` class: a private superclass of both `PurePath` and `_PathBase`. Unlike `PurePath`, it does not define any of these special methods: `__fspath__`, `__bytes__`, `__reduce__`, `__hash__`, `__eq__`, `__lt__`, `__le__`, `__gt__`, `__ge__`. Its initializer and path joining methods accept only strings, not os.PathLike objects more broadly.
This is important for supporting *virtual paths*: user subclasses of `_PathBase` that provide access to archive files, FTP servers, etc. In these classes, the above methods should be implemented by users only as appropriate, with due consideration for the hash/equality of any backing objects, such as file objects or sockets.
Barney Gale [Thu, 7 Dec 2023 21:45:40 +0000 (21:45 +0000)]
GH-112675: Move path joining tests into `test_posixpath` and `test_ntpath` (#112676)
In `test_pathlib`, the `check_drive_root_parts` test methods evaluated
both joining and parsing/normalisation of paths. This dates from a time
when pathlib implemented both functions itself, but nowadays path joining
is done with `posixpath.join()` and `ntpath.join()`.
This commit moves the joining-related test cases into `test_posixpath` and
`test_ntpath`.
Eric Snow [Thu, 7 Dec 2023 20:22:15 +0000 (13:22 -0700)]
gh-112826: Add a "What's New" Entry About _thread._is_main_interpreter (gh-112853)
As of gh-112661, the threading module expects the _thread module to have a _is_main_interpreter(), which is used in the internal threading._shutdown(). This change causes a problem for anyone that replaces the _thread module with a custom one (only if they don't provide _is_main_interpreter()). They need to be sure to add it for 3.13+, thus this PR is adding a note in "What's New".
This also forward-ports the "What's New" entry from 3.12 (gh-112850). Note that we do not also forward-port the fix in that PR. The fix is there only due to a regression from 3.12.0. There is no regression in 3.13+.
Sam Gross [Thu, 7 Dec 2023 19:33:40 +0000 (14:33 -0500)]
gh-111924: Use PyMutex for Runtime-global Locks. (gh-112207)
This replaces some usages of PyThread_type_lock with PyMutex, which does not require memory allocation to initialize.
This simplifies some of the runtime initialization and is also one step towards avoiding changing the default raw memory allocator during initialize/finalization, which can be non-thread-safe in some circumstances.
Sam Gross [Thu, 7 Dec 2023 19:11:45 +0000 (14:11 -0500)]
gh-112538: Add internal-only _PyThreadStateImpl "wrapper" for PyThreadState (gh-112560)
Every PyThreadState instance is now actually a _PyThreadStateImpl.
It is safe to cast from `PyThreadState*` to `_PyThreadStateImpl*` and back.
The _PyThreadStateImpl will contain fields that we do not want to expose
in the public C API.
Sam Gross [Thu, 7 Dec 2023 13:47:55 +0000 (08:47 -0500)]
gh-111962: Make dtoa thread-safe in `--disable-gil` builds. (#112049)
This updates `dtoa.c` to avoid using the Bigint free-list in --disable-gil builds and
to pre-computes the needed powers of 5 during interpreter initialization.
* gh-111962: Make dtoa thread-safe in `--disable-gil` builds.
This avoids using the Bigint free-list in `--disable-gil` builds
and pre-computes the needed powers of 5 during interpreter initialization.
* Fix size of cached powers of 5 array.
We need the powers of 5 up to 5**512 because we only jump straight to
underflow when the exponent is less than -512 (or larger than 308).
gh-112660: Do not clear arbitrary errors on import (GH-112661)
Previously arbitrary errors could be cleared during formatting error
messages for ImportError or AttributeError for modules. Now all
unexpected errors are reported.
Ronald Oussoren [Thu, 7 Dec 2023 09:30:15 +0000 (10:30 +0100)]
gh-109981: Fix support.fd_count() on macOS 14 (#112797)
Use scanning "/dev/fd/" on macOS in support.fd_count(). That's both more efficient than scanning all possible file descriptors, and avoids crashing the interpreter when there are open "guarded" file descriptors.
"Guarded" file descriptors are a macOS feature where file descriptors used by system libraries are marked and cause hard crashes when used by "user" code.
Co-authored-by: Victor Stinner <vstinner@python.org>
gh-75666: Tkinter: "unbind(sequence, funcid)" now only unbinds "funcid" (GH-111322)
Previously, "widget.unbind(sequence, funcid)" destroyed the current binding
for "sequence", leaving "sequence" unbound, and deleted the "funcid"
command.
Now it removes only "funcid" from the binding for "sequence", keeping
other commands, and deletes the "funcid" command.
It leaves "sequence" unbound only if "funcid" was the last bound command.
Victor Stinner [Wed, 6 Dec 2023 14:09:22 +0000 (15:09 +0100)]
gh-111545: Add Py_HashPointer() function (#112096)
* Implement _Py_HashPointerRaw() as a static inline function.
* Add Py_HashPointer() tests to test_capi.test_hash.
* Keep _Py_HashPointer() function as an alias to Py_HashPointer().
gh-111178: Docs: fix `traverseproc`, `inquiry`, and `destructor` parameters in slot typedefs table (GH-112742)
In the slot typedefs table, the parameter of `destructor`
and the first parameter of `traverseproc` should both be
`PyObject *` rather than `void *`.
Same for `inquiry`.
Zackery Spytz [Tue, 5 Dec 2023 17:09:39 +0000 (09:09 -0800)]
bpo-35332: Handle os.close() errors in shutil.rmtree() (GH-23766)
* Ignore os.close() errors when ignore_errors is True.
* Pass os.close() errors to the error handler if specified.
* os.close no longer retried after error.
gh-94692: Only catch OSError in shutil.rmtree() (#112756)
Previously a symlink attack resistant version of shutil.rmtree() could ignore
or pass to the error handler arbitrary exception when invalid arguments
were provided.
Diego Russo [Tue, 5 Dec 2023 15:07:50 +0000 (15:07 +0000)]
gh-110190: Fix ctypes structs with array on Arm (#112604)
Set MAX_STRUCT_SIZE to 32 in stgdict.c when on Arm platforms.
This because on Arm platforms structs with at most 4 elements of any
floating point type values can be passed through registers. If the type
is double the maximum size of the struct is 32 bytes.
On x86-64 Linux, it's maximum 16 bytes hence we need to differentiate.
pan324 [Tue, 5 Dec 2023 08:11:44 +0000 (09:11 +0100)]
gh-82300: Add track parameter to multiprocessing.shared_memory (#110778)
Add a track parameter to shared memory to allow resource tracking via the side-launched resource tracker process to be disabled on platforms that use it (POSIX).
This allows people who do not want automated cleanup at process exit because they are using the shared memory with processes not participating in Python's resource tracking to use the shared_memory API.
Co-authored-by: Łukasz Langa <lukasz@langa.pl> Co-authored-by: Guido van Rossum <gvanrossum@gmail.com> Co-authored-by: Antoine Pitrou <pitrou@free.fr> Co-authored-by: Gregory P. Smith <greg@krypto.org>
gh-112736: Refactor del-safe symbol handling in subprocess (#112738)
Refactor delete-safe symbol handling in subprocess.
Only module globals are force-cleared during interpreter finalization, using a class reference instead of individually listing the constants everywhere is simpler.
gh-112334: Restore subprocess's use of `vfork()` & fix `extra_groups=[]` behavior (#112617)
Restore `subprocess`'s intended use of `vfork()` by default for performance on Linux;
also fixes the behavior of `extra_groups=[]` which was unintentionally broken in 3.12.0:
Fixed a performance regression in 3.12's :mod:`subprocess` on Linux where it
would no longer use the fast-path ``vfork()`` system call when it could have
due to a logic bug, instead falling back to the safe but slower ``fork()``.
Also fixed a security bug introduced in 3.12.0. If a value of ``extra_groups=[]``
was passed to :mod:`subprocess.Popen` or related APIs, the underlying
``setgroups(0, NULL)`` system call to clear the groups list would not be made
in the child process prior to ``exec()``.
The security issue was identified via code inspection in the process of
fixing the first bug. Thanks to @vain for the detailed report and
analysis in the initial bug on Github.
gh-105967: Work around a macOS bug, limit zlib C library crc32 API calls to 1gig (#112615)
Work around a macOS bug, limit zlib crc32 calls to 1GiB.
Without this, `zlib.crc32` and `binascii.crc32` could produce incorrect
results on multi-gigabyte inputs depending on the macOS version's Apple
supplied zlib implementation.
gh-65210: Add const qualifiers in PyArg_VaParseTupleAndKeywords() (GH-105958)
Change the declaration of the keywords parameter in functions
PyArg_ParseTupleAndKeywords() and PyArg_VaParseTupleAndKeywords() from `char **`
to `char * const *` in C and `const char * const *` in C++.
It makes these functions compatible with argument of type `const char * const *`,
`const char **` or `char * const *` in C++ and `char * const *` in C
without explicit type cast.