Deprecate global configuration variable like
Py_IgnoreEnvironmentFlag: the Py_InitializeFromConfig() API should be
instead.
Fix declaration of Py_GETENV(): use PyAPI_FUNC(), not PyAPI_DATA().
test_asyncio, test_logging, test_socket and test_socketserver now
create AF_UNIX domains in the current directory to no longer fail
with OSError("AF_UNIX path too long") if the temporary directory (the
TMPDIR environment variable) is too long.
Modify the following tests to use create_unix_domain_name():
* test_asyncio
* test_logging
* test_socket
* test_socketserver
test_asyncio.utils: remove unused time import.
Revert "bpo-23689: re module, fix memory leak when a match is terminated by a signal or memory allocation failure (GH-32283)"
This reverts commit 6e3eee5c11.
Manual fixups to increase the MAGIC number and to handle conflicts with
a couple of changes that landed after that.
Thanks for reviews by Ma Lin and Serhiy Storchaka.
regrtest now also implements checking for leaked temporary files and
directories when using -jN for N >= 2. Use tempfile.mkdtemp() to
create the temporary directory. Skip this check on WASI.
SharedMemory.unlink() uses the unregister() function from resource_tracker. Previously it was imported in the method, but this can fail if the method is called during interpreter shutdown, for example when unlink is part of a __del__() method.
Moving the import to the top of the file, means that the unregister() method is available during interpreter shutdown.
The register call in SharedMemory.__init__() can also use this imported resource_tracker.
GH-26658 introduced a regression in copy / pickle protocol for combined
`enum.Flag`s. `copy.copy(re.A | re.I)` would fail with
`AttributeError: ASCII|IGNORECASE`.
`enum.Flag` now has a `__reduce_ex__()` method that reduces flags by
combined value, not by combined name.
Align the docs for the following methods with the actual implementation:
- sqlite3.complete_statement()
- sqlite3.Connection.create_function()
- sqlite3.Connection.create_aggregate()
- sqlite3.Connection.set_progress_handler()
The fix involves using pysqlite_check_remaining_sql(), not only to check
for multiple statements, but now also to strip leading comments and
whitespace from SQL statements, so we can improve DML query detection.
pysqlite_check_remaining_sql() is renamed lstrip_sql(), to more
accurately reflect its function, and hardened to handle more SQL comment
corner cases.
When changing PyType_FromMetaclass recently (GH-93012, GH-93466, GH-28748)
I found a bunch of opportunities to improve the code. Here they are.
Fixes: #89546
Automerge-Triggered-By: GH:encukou
Suppress writing an XML declaration in open files in ElementTree.write()
with encoding='unicode' and xml_declaration=None.
If file patch is passed to ElementTree.write() with encoding='unicode',
always open a new file in UTF-8.
Classes ReferenceType, ProxyType and CallableProxyType have now correct
atrtributes __module__, __name__ and __qualname__.
It makes them (types, not instances) pickleable.
Fix the importlib.resources.as_file() context manager to remove the
temporary file if destroyed late during Python finalization: keep a
local reference to the os.remove() function. Patch by Victor Stinner.
Documentation for `pathlib` says:
> Spurious slashes and single dots are collapsed, but double dots ('..') are not, since this would change the meaning of a path in the face of symbolic links:
However, it omits that initial double slashes also aren't collapsed.
Later, in documentation of `PurePath.drive`, `PurePath.root`, and `PurePath.name` it mentions UNC but:
- this abbreviation says nothing to a person who is unaware about existence of UNC (Wikipedia doesn't help either by [giving a disambiguation page](https://en.wikipedia.org/wiki/UNC))
- it shows up only if a person needs to use a specific property or decides to fully learn what the module provides.
For context, see the BPO entry.
This checks the bases of of a type created using the FromSpec
API to inherit the bases metaclasses. The metaclass's alloc
function will be called as is done in `tp_new` for classes
created in Python.
Co-authored-by: Petr Viktorin <encukou@gmail.com>
Co-authored-by: Erlend Egeberg Aasland <erlend.aasland@protonmail.com>
copy.copy() and copy.deepcopy() now always raise a TypeError if
__reduce__() returns a tuple with length 6 instead of silently ignore
the 6th item or produce incorrect result.
All install targets use the "all" target as synchronization point to
prevent race conditions with PGO builds. PGO builds use recursive make,
which can lead to two parallel `./python setup.py build` processes that
step on each others toes.
"test" targets now correctly compile PGO build in a clean repo.
Co-authored-by: Alex Waygood <Alex.Waygood@Gmail.com>
Co-authored-by: Adam Turner <9087854+AA-Turner@users.noreply.github.com>
Co-authored-by: Erlend E. Aasland <erlend.aasland@protonmail.com>
Note: This change is not effective on Microsoft Windows.
Cookies can store sensitive information and should therefore be protected
against unauthorized third parties. This is also described in issue #79096.
The filesystem permissions are currently set to 644, everyone can read the
file. This commit changes the permissions to 600, only the creater of the file
can read and modify it. This improves security, because it reduces the attack
surface. Now the attacker needs control of the user that created the cookie or
a ways to circumvent the filesystems permissions.
This change is backwards incompatible. Systems that rely on world-readable
cookies will breake. However, one could argue that those are misconfigured in
the first place.
Character ranges with upper bound less that lower bound (e.g. [c-a])
are now interpreted as empty ranges, for compatibility with other glob
pattern implementations. Previously it was re.error.
Add C++ overloads for _Py_CAST_impl() to handle 0/NULL. This will allow
C++ extensions that pass 0 or NULL to macros using _Py_CAST() to
continue to compile. Without this, you get an error like:
invalid ‘static_cast’ from type ‘int’ to type ‘_object*’
The modern way to use a NULL value in C++ is to use nullptr. However,
we want to not break extensions that do things the old way.
Co-authored-by: serge-sans-paille
When a `_PathParents` object has a drive or a root, the length of the
object is *one less* than than the length of `self._parts`, which resulted
in an off-by-one error when `path.parents[-n]` was fed through to
`self._parts[:-n - 1]`. In particular, `path.parents[-1]` was a malformed
path object with spooky properties.
This is addressed by adding `len(self)` to negative indices.
* bpo-42272: improve message/module warning filter docs
"The Warnings Filter" section of the warnings module documentation
describes the message and module filters as "a string containing a
regular expression". While that is true when they are arguments to the
filterwarnings function, it is not true when they appear in -W or
$PYTHONWARNINGS where they are matched literally (after stripping any
starting/ending whitespace). Update the documentation to note when they
are matched literally. Also clarify that module matches the
"fully-qualified module name", rather than "module name" which is
ambiguous.
skip news (since this is a doc fix)
Signed-off-by: Kevin Locke <kevin@kevinlocke.name>
* bpo-42272: remove bad submodule warning filter doc
The `error:::mymodule[.*]` example in the "Describing Warning Filters"
section of the warnings module documentation does not behave as the
comment describes. Since the module portion of the filter string is
interpreted literally, it would match a module with a fully-qualified
name that is literally `mymodule[.*]`.
Unfortunately, there is not a way to match '"module" and any subpackages
of "mymodule"' as documented, since the module part of a filter string
is matched literally. Instead, update the filter and comment to match
only "mymodule".
skip news (since this is a doc fix)
Signed-off-by: Kevin Locke <kevin@kevinlocke.name>
* bpo-42272: add warning filter doc changes to NEWS
Signed-off-by: Kevin Locke <kevin@kevinlocke.name>
This was added for bpo-40514 (gh-84694) to test out a per-interpreter GIL. However, it has since proven unnecessary to keep the experiment in the repo. (It can be done as a branch in a fork like normal.) So here we are removing:
* the configure option
* the macro
* the code enabled by the macro
Added a new stable API function ``PyType_FromMetaclass``, which mirrors
the behavior of ``PyType_FromModuleAndSpec`` except that it takes an
additional metaclass argument. This is, e.g., useful for language
binding tools that need to store additional information in the type
object.
Python now always use the ``%zu`` and ``%zd`` printf formats to
format a size_t or Py_ssize_t number. Building Python 3.12 requires a
C11 compiler, so these printf formats are now always supported.
* PyObject_Print() and _PyObject_Dump() now use the printf %zd format
to display an object reference count.
* Update PY_FORMAT_SIZE_T comment.
* Remove outdated notes about the %zd format in PyBytes_FromFormat()
and PyUnicode_FromFormat() documentations.
* configure no longer checks for the %zd format and no longer defines
PY_FORMAT_SIZE_T macro in pyconfig.h.
* pymacconfig.h no longer undefines PY_FORMAT_SIZE_T: macOS 10.4 is
no longer supported. Python 3.12 now requires macOS 10.6 (Snow
Leopard) or newer.
Fix __lltrace__ debug feature if the stdout encoding is not UTF-8.
If the stdout encoding is not UTF-8, the first call to
lltrace_resume_frame() indirectly sets lltrace to 0 when calling
unicode_check_encoding_errors() which calls
encodings.search_function().
Generally comparable perf for the "good" case where memchr doesn't
return any collisions (false matches on lower byte) but clearly faster
with collisions.
Also while there, clarify a few things about why we reduce the hash to 32 bits.
Co-authored-by: Eli Libman <eli@hyro.ai>
Co-authored-by: Yury Selivanov <yury@edgedb.com>
Co-authored-by: Łukasz Langa <lukasz@langa.pl>
At compile time, '+z' is already properly used with HP aCC, and shared
libraries are correctly linked with '+b'. The '-fPIC' switch can safely be
dropped.
The following sqlite3 features were deprecated in 3.10, scheduled for
removal in 3.12:
- sqlite3.OptimizedUnicode (gh-23163)
- sqlite3.enable_shared_cache (gh-24008)
Co-authored-by: Jelle Zijlstra <jelle.zijlstra@gmail.com>
Signed-off-by: Erlend E. Aasland <erlend.aasland@protonmail.com>
If Condition.notify() was interrupted just after it released the waiter lock,
but before removing it from the queue, the following calls of notify() failed
with RuntimeError: cannot release un-acquired lock.
Avoid mixing declarations and code in the C API to fix the compiler
warning: "ISO C90 forbids mixed declarations and code"
[-Werror=declaration-after-statement].
Remove the PyUnicode_InternImmortal() function and the
SSTATE_INTERNED_IMMORTAL macro.
The PyUnicode_InternImmortal() function is still exported in the
stable ABI. The function is removed from the API.
PyASCIIObject.state.interned size is now a single bit, rather than 2
bits.
Keep SSTATE_NOT_INTERNED and SSTATE_INTERNED_MORTAL macros for
backward compatibility, but no longer use them internally since the
interned member is now a single bit and so can only have two values
(interned or not interned).
Update stats of _PyUnicode_ClearInterned().
The `utc_to_seconds` call can fail, here's a minimal reproducer on
Linux:
TZ=UTC python -c "from datetime import *; datetime.fromtimestamp(253402300799 + 1)"
The old behavior still raised an error in a similar way, but only
because subsequent calculations happened to fail as well. Better to fail
fast.
This also refactors the tests to split out the `fromtimestamp` and
`utcfromtimestamp` tests, and to get us closer to the actual desired
limits of the functions. As part of this, we also changed the way we
detect platforms where the same limits don't necessarily apply (e.g.
Windows).
As part of refactoring the tests to hit this condition explicitly (even
though the user-facing behvior doesn't change in any way we plan to
guarantee), I noticed that there was a difference in the places that
`datetime.utcfromtimestamp` fails in the C and pure Python versions, which
was fixed by skipping the "probe for fold" logic for UTC specifically —
since UTC doesn't have any folds or gaps, we were never going to find a
fold value anyway. This should prevent some failures in the pure python
`utcfromtimestamp` method on timestamps close to 0001-01-01.
There are two separate news entries for this because one is a
potentially user-facing change, the other is an internal code
correctness change that, if anything, changes some error messages. The
two happen to be coupled because of the test refactoring, but they are
probably best thought of as independent changes.
Fixes GH-91581
Remove the token.h header file. There was never any public tokenizer
C API. The token.h header file was only designed to be used by Python
internals.
Move Include/token.h to Include/internal/pycore_token.h. Including
this header file now requires that the Py_BUILD_CORE macro is
defined. It no longer checks for the Py_LIMITED_API macro.
Rename functions:
* PyToken_OneChar() => _PyToken_OneChar()
* PyToken_TwoChars() => _PyToken_TwoChars()
* PyToken_ThreeChars() => _PyToken_ThreeChars()
ElementTree method write() and function tostring() now use the text file's
encoding ("UTF-8" if not available) instead of locale encoding in XML
declaration when encoding="unicode" is specified.
Lines beginning with ``?`` try to help understanding the given diff.
The output can be hard to understand when it contains whitespace characters, such as spaces, tabs or line breaks.
While previously only tabs were mentioned, now all are listed.
Automerge-Triggered-By: GH:rhettinger
We could try to remedy this by taking a slice, but we then run into an issue where the empty string will match altsep on POSIX. That rabbit hole could keep getting deeper.
A proper fix for the original issue involves making pathlib's path normalisation more configurable - in this case we want to retain trailing slashes, but in other we might want to preserve `./` prefixes, or elide `../` segments when we're sure we won't encounter symlinks.
This reverts commit ea2f5bcda1.
* Some handlers were wrongly described as text-encoding only, but actually they can also be used in text-decoding.
* Add more description to each handler.
* Add two REPL examples.
* Add indexes for Error Handler's name.
Co-authored-by: Kyle Stanley <aeros167@gmail.com>
Co-authored-by: Victor Stinner <vstinner@python.org>
Co-authored-by: Jelle Zijlstra <jelle.zijlstra@gmail.com>
Do not spawn ProcessPool workers on demand when they spawn via fork.
This avoids potential deadlocks in the child processes due to forking from
a multithreaded process.
Victor Stinner
Gregory P. Smith
Irit Katriel
Brandt Bucher
Serhiy Storchaka
samtygier
Christian Heimes
Steve Dower
Erlend Egeberg Aasland
Ken Jin
Mark Shannon
Petr Viktorin
Pablo Galindo Salgado
Dennis Sweeney
jackh-ncl
Oleg Iarygin
Barney Gale
Leo Trol
Sebastian Berg
Bluenix
Kalyan
Adrian Garcia Badaracco
Andreas Grommek
Hugo van Kemenade
Vinay Sajip
Pascal Wittmann
Davide Rizzo
AN Long
Illia Volochii
Neil Schemenauer
Dong-hee Na
Eric V. Smith
Éric
Kumar Aditya
Kevin Locke
Sergei Izmailov
georgically
oda-gitso
Eric Snow
Yury Selivanov
hfinucane
Wenzel Jakob
Éric
Miro Hrončok
Jordan Borean
goldsteinn
Sam Ezeh
Mori Bellamy
David Machaj
Nikita Sobolev
Michael Osipov
Zackery Spytz
David Foster
Ayappan Perumal
Inada Naoki
David CARLIER
Batuhan Taskaya
Paul Ganssle
ravi140222
Jelle Zijlstra
Raymond Hettinger
狂男风
Bar Harel
Jürgen Gmach
Cheryl Sabella
Mariusz Felisiak
Ma Lin