Scheduled monthly dependency update for October by pyup-bot · Pull Request #167 · alimanfoo/pysamstats

pyup-bot · 2024-10-01T15:12:16Z

Update Cython from 0.29.23 to 3.0.11.

Changelog

3.0.11

===================

Features added
--------------

* The C++11 ``emplace*`` methods were added to ``libcpp.deque``.
Patch by Somin An.  (Github issue :issue:`6159`)

Bugs fixed
----------

* The exception check value of functions declared in pxd files was not always applied in 3.0.10.
(Github issue :issue:`6122`)

* A crash on exception deallocations was fixed.
(Github issue :issue:`6022`)

* A crash was fixed when assigning a zero-length slice to a memoryview.
Patch by Michael Man.  (Github issue :issue:`6227`)

* ``libcpp.optional.value()`` could crash if it raised a C++ exception.
Patch by Alexander Condello.  (Github issue :issue:`6190`)

* The return type of ``str()`` was mishandled, leading to crashes with ``language_level=3``.
(Github issue :issue:`6166`)

* ``bytes.startswith/endswith()`` failed for non-bytes substrings (e.g. ``bytearray``).
(Github issue :issue:`6168`)

* Fused ctuples crashed Cython.
(Github issue :issue:`6068`)

* A compiler crash was fixed when using extension types in fused types.
(Github issue :issue:`6204`)

* The module cleanup code was incorrect for globally defined memory view slices.
(Github issue :issue:`6276`)

* Some adaptations were made to enable compilation in Python 3.13.
(Github issues :issue:`5997`, :issue:`6182`, :issue:`6251`)

3.0.10

===================

Bugs fixed
----------

* Cython generated incorrect self-casts when directly calling final methods of subtypes.
Patch by Lisandro Dalcin.  (Github issue :issue:`2747`)

* Internal C names generated from C function signatures could become too long for MSVC.
(Github issue :issue:`6052`)

* The ``noexcept`` warnings could be misleading in some cases.
Patch by Gonzalo Tornaría.  (Github issue :issue:`6087`)

* The ``cython.ufunc`` implementation could generate incomplete C code.
(Github issue :issue:`6064`)

* The ``libcpp.complex`` declarations could result in incorrect C++ code.
Patch by Raffi Enficiaud.  (Github issue :issue:`6037`)

* Several tests were adapted to work with both NumPy 1.x and 2.0.
Patch by Matti Picus.  (Github issues :issue:`6076`, :issue:`6100`)

* C compiler warnings when the freelist implementation is disabled (e.g. on PyPy) were fixed.
It can now be disabled explicitly with the C macro guard ``CYTHON_USE_FREELISTS=0``.
(Github issue :issue:`6099`)

* Some C macro guards for feature flags were missing from the NOGIL Python configuration.

* Some recently added builtins were unconditionally looked up at module import time
(if used by user code) that weren&#x27;t available on all Python versions and could thus
fail the import.

* A performance hint regarding exported pxd declarations was improved.
(Github issue :issue:`6001`)

3.0.9

==================

Features added
--------------

* Assigning ``const`` values to non-const variables now issues a warning.
(Github issue :issue:`5639`)

* Using ``noexcept`` on a function returning Python objects now issues a warning.
(Github issue :issue:`5661`)

* Some C-API usage was updated for the upcoming CPython 3.13.
Patches by Victor Stinner et al.  (Github issues :issue:`6003`, :issue:`6020`)

* The deprecated ``Py_UNICODE`` type is no longer used, unless required by user code.
(Github issue :issue:`5982`)

* ``std::string.replace()`` declarations were added to libcpp.string.
Patch by Kieran Geary.  (Github issue :issue:`6037`)

Bugs fixed
----------

* Cython generates incorrect (but harmless) self-casts when directly calling
final methods of subtypes.  Lacking a better solution, the errors that recent
gcc versions produce have been silenced for the time being.
Original patch by Michał Górny.  (Github issue :issue:`2747`)

* Unused variable warnings about clineno were fixed when C lines in tracebacks are disabled.
(Github issue :issue:`6035`)

* Subclass deallocation of extern classes could crash if the base class uses GC.
Original patch by Jason Fried.  (Github issue :issue:`5971`)

* Type checks for Python ``memoryview`` could use an invalid C function.
Patch by Xenia Lu.  (Github issue :issue:`5988`)

* Calling final fused functions could generate invalid C code.
(Github issue :issue:`5989`)

* Declaring extern enums multiple times could generate invalid C code.
(Github issue :issue:`5905`)

* ``pyximport`` used relative paths incorrectly.
Patch by Stefano Rivera.  (Github issue :issue:`5957`)

* Running Cython with globbing characters (``[]*?``) in the module search path could fail.
Patch by eewanco.  (Github issue :issue:`5942`)

* Literal strings that include braces could change the C code indentation.

Other changes
-------------

* The &quot;enum class not importable&quot; warning is now only issued once per enum type.
(Github issue :issue:`5941`)

3.0.8

==================

Bugs fixed
----------

* Using ``const`` together with defined fused types could fail to compile.
(Github issue :issue:`5230`)

* A &quot;use after free&quot; bug was fixed in parallel sections.
(Github issue :issue:`5922`)

* Several types were not available as ``cython.*`` types in pure Python code.

* The generated code is now correct C89 again, removing some C++ style ``//`` comments
and C99-style declaration-after-code code ordering.  This is still relevant for some
ols C compilers, specifically ones that match old Python 2.7 installations.

3.0.7

==================

Bugs fixed
----------

* In the iterator of generator expressions, ``await`` and ``yield`` were not correctly analysed.
(Github issue :issue:`5851`)

* ``cpdef`` enums with the same name cimported from different modules could lead to
invalid C code.
(Github issue :issue:`5887`)

* Some declarations in ``cpython.unicode`` were fixed and extended.
(Github issue :issue:`5902`)

* Compiling fused types used in pxd files could crash Cython in Python 3.11+.
(Github issues :issue:`5894`,  :issue:`5588`)

* Source files with non-ASCII file names could crash Cython.
(Github issue :issue:`5873`)

* Includes all bug-fixes and features from the 0.29 maintenance branch
up to the :ref:`0.29.37` release.

3.0.6

==================

Features added
--------------

* Fused def function dispatch is a bit faster.

* Declarations for the ``wchar`` PyUnicode API were added.
(Github issue :issue:`5836`)

* The Python &quot;nogil&quot; fork is now also detected with the new ``Py_GIL_DISABLED`` macro.
Patch by Hugo van Kemenade.  (Github issue :issue:`5852`)

Bugs fixed
----------

* Comparing dataclasses could give different results than Python.
(Github issue :issue:`5857`)

* ``float(std::string)`` generated invalid C code.
(Github issue :issue:`5818`)

* Using ``cpdef`` functions with ``cimport_from_pyx`` failed.
(Github issue :issue:`5795`)

* A crash was fixed when string-formatting a Python value fails.
(Github issue :issue:`5787`)

* On item access, Cython could try the sequence protocol before the mapping protocol
in some cases if an object supports both.
(Github issue :issue:`5776`)

* A C compiler warning was resolved.
(Github issue :issue:`5794`)

* Complex numbers failed to compile in MSVC with C11.
Patch by Lysandros Nikolaou.  (Github issue :issue:`5809`)

* Some issues with the Limited API and with PyPy were resolved.
(Github issues :issue:`5695`, :issue:`5696`)

* A C++ issue in Python 3.13 was resolved.
(Github issue :issue:`5790`)

* Several directives are now also available (as no-ops) in Python code.
(Github issue :issue:`5803`)

* An error message was corrected.
Patch by Mads Ynddal.  (Github issue :issue:`5805`)

3.0.5

==================

Features added
--------------

* Preliminary support for CPython 3.13a1 was added to allow early testing.
(Github issue :issue:`5767`)

Bugs fixed
----------

* A compiler crash was fixed.
(Github issue :issue:`5771`)

* A typo in the ``always_allow_keywords`` directive for Python code was fixed.
Patch by lk-1984.  (Github issue :issue:`5772`)

* Some C compiler warnings were resolved.
Patch by Pierre Jolivet.  (Github issue :issue:`5780`)

3.0.4

==================

Features added
--------------

* A new compiler directive ``show_performance_hints`` was added to disable the
newly added performance hint output.
(Github issue :issue:`5748`)

Bugs fixed
----------

* cythonize` required ``distutils`` even for operations that did not build binaries.
(Github issue :issue:`5751`)

* A regression in 3.0.3 was fixed that prevented calling inline functions
from another inline function in ``.pxd`` files.
(Github issue :issue:`5748`)

* Some C compiler warnings were resolved.
Patch by Pierre Jolivet.  (Github issue :issue:`5756`)

3.0.3

==================

Features added
--------------

* More warnings were added to help users migrate and avoid bugs.
(Github issue :issue:`5650`)

* A warning-like category for performance hints was added that bypasses ``-Werror``.
(Github issue :issue:`5673`)

* FastGIL now uses standard ``thread_local`` in C++.
(Github issue :issue:`5640`)

* ``reference_wrapper`` was added to ``libcpp.functional``.
Patch by Vyas Ramasubramani.  (Github issue :issue:`5671`)

* The ``cythonize`` command now supports the ``--cplus`` option known from the ``cython`` command.
(Github issue :issue:`5736`)

Bugs fixed
----------

* Performance regressions where the GIL was needlessly acquired were fixed.
(Github issues :issue:`5670`, :issue:`5700`)

* A reference leak for exceptions in Python 3.12 was resolved.
Patch by Eric Johnson.  (Github issue :issue:`5724`)

* ``fastcall`` calls with keyword arguments generated incorrect C code.
(Github issue :issue:`5665`)

* Assigning the type converted result of a conditional (if-else) expression
to ``int`` or ``bool`` variables could lead to incorrect C code.
(Github issue :issue:`5731`)

* Early (unlikely) failures in Python function wrappers no longer set a
traceback in order to simplify the C code flow.  Being mostly memory
allocation errors, they probably would never have created a traceback anyway.
(Github issue :issue:`5681`)

* Relative cimports from packages with ``__init__.py`` files could fail.
(Github issue :issue:`5715`)

* Several issues with the Limited API support were resolved.
(Github issues :issue:`5641`, :issue:`5648`, :issue:`5689`)

* The code generated for special-casing both Cython functions and PyCFunctions was cleaned up
to avoid calling C-API functions that were not meant for the other type respectively.
This could previously trigger assertions in CPython debug builds and now also plays better
with the Limited API.
(Github issues :issue:`4804`, :issue:`5739`)

* Fix some C compiler warnings.
Patches by Ralf Gommers, Oleksandr Pavlyk, Sebastian Koslowski et al.
(Github issues :issue:`5651`, :issue:`5663`, :issue:`5668`, :issue:`5717`, :issue:`5726`, :issue:`5734`)

* Generating gdb debugging information failed when using generator expressions.
Patch by Oleksandr Pavlyk.  (Github issue :issue:`5552`)

* Passing a ``setuptools.Extension`` into ``cythonize()`` instead of a
``distutils.Extension`` could make it miss the matching extensions.

* ``cython -M`` needlessly required ``distutils``, which made it fail in Python 3.12.
(Github issue :issue:`5681`)

Other changes
-------------

* The visible deprecation warning for ``DEF`` was removed again since it proved
difficult for some users to migrate away from it.  The statement is still
meant to be removed at some point (and thus, like ``IF``, should not be
used in new code), but the time for sunset is probably not around the corner.
(Github issue :issue:`4310`)

* The ``np_pythran`` option raise a ``DeprecationWarning`` if it receives other values
than ``True`` and ``False``.  This will eventually be disallowed (in line with all
other boolean options).

3.0.2

==================

Bugs fixed
----------

* Using ``None`` as default value for arguments annotated as ``int`` could crash Cython.
(Github issue :issue:`5643`)

* Default values of fused types that include ``complex`` could generate invalid C code
with ``-DCYTHON_CCOMPLEX=0``.
(Github issue :issue:`5644`)

* Using C++ enum class types in extension type method signatures could generate invalid C code.
(Github issue :issue:`5637`)

3.0.1

==================

Features added
--------------

* The error messages regarding exception declarations were improved in order to give
better help about possible reasons and fixes.
(Github issue :issue:`5547`)

Bugs fixed
----------

* Memory view types in Python argument annotations no longer accept ``None``.  They now
require an explicit ``Optional[]`` or a ``None`` default value in order to allow ``None``
to be passed.  This was an oversight in the 3.0.0 release and is a BACKWARDS INCOMPATIBLE
change.  However, since it only applies to code using Python syntax, it probably only
applies to newly written code that was written for Cython 3.0 and can easily be adapted.
In most cases, we expect that this change will avoid bugs in user code rather than
produce problems.
(Github issue :issue:`5612`)

* ``nogil`` functions using parallel code could freeze when called with the GIL held.
(Github issues :issue:`5564`, :issue:`5573`)

* Relative cimports could end up searching globally and find the same package installed
elsewhere, potentially in another version.
(Github issue :issue:`5511`)

* Attribute lookups on known standard library modules could accidentally search
in the module namespace instead.
(Github issue :issue:`5536`)

* Using constructed C++ default arguments could generate invalid C++ code.
(Github issue :issue:`5553`)

* ``libcpp.memory.make_unique()`` was lacking C++ exception handling.
(Github issue :issue:`5560`)

* Some non-public and deprecated CAPI usages were replaced by public
(and thus more future proof) API code.

* Many issues with the Limited API support were resolved.
Patches by Lisandro Dalcin et al.
(Github issues :issue:`5549`, :issue:`5550`, :issue:`5556`, :issue:`5605`, :issue:`5617`)

* Some C compiler warnings were resolved.
Patches by Matti Picus et al.  (Github issues :issue:`5557`, :issue:`5555`)

* Large Python integers are now stored in hex instead of decimal strings to work around
security limits in Python and generally speed up their Python object creation.

* ``NULL`` could not be used as default for fused type pointer arguments.
(Github issue :issue:`5554`)

* C functions that return pointer types now return ``NULL`` as default exception value.
Previously, calling code wasn&#x27;t aware of this and always tested for raised exceptions.
(Github issue :issue:`5554`)

* Untyped literal default arguments in fused functions could generate invalid C code.
(Github issue :issue:`5614`)

* C variables declared as ``const`` could generate invalid C code when used in closures,
generator expressions, ctuples, etc.
(Github issues :issue:`5558`,  :issue:`5333`)

* Enums could not refer to previously defined enums in their definition.
(Github issue :issue:`5602`)

* The Python conversion code for anonymous C enums conflicted with regular int conversion.
(Github issue :issue:`5623`)

* Using memory views for property methods (and other special methods) could lead to
refcounting problems.
(Github issue :issue:`5571`)

* Star-imports could generate code that tried to assign to constant C macros like
``PY_SSIZE_T_MAX`` and ``PY_SSIZE_T_MIN``.
Patch by Philipp Wagner.  (Github issue :issue:`5562`)

* ``CYTHON_USE_TYPE_SPECS`` can now be (explicitly) enabled in PyPy.

* The template parameter &quot;delimeters&quot; in the Tempita ``Template`` class was corrected
to &quot;delimiters&quot;.  The old spelling is still available in the main template API but
now issues a ``DeprecationWarning``.
(Github issue :issue:`5608`)

* The ``cython --version`` output is now  less likely to reach both stdout and stderr.
Patch by Eli Schwartz.  (Github issue :issue:`5504`)

* The sdist was missing the `Shadow.pyi` stub file.

3.0.0

==========================

Features added
--------------

* Cython functions now use the `PEP-590`_ vectorcall protocol in Py3.7+.
Patch by Jeroen Demeyer.  (Github issue :issue:`2263`)

* Unicode identifiers are supported in Cython code (`PEP-3131`_).
Patch by David Woods.  (Github issue :issue:`2601`)

* Unicode module names and imports are supported.
Patch by David Woods.  (Github issue :issue:`3119`)

* Annotations are no longer parsed, keeping them as strings following `PEP-563`_.
Patch by David Woods.  (Github issue :issue:`3285`)

* Preliminary support for the CPython&#x27;s ``Py_LIMITED_API`` (stable ABI) is
available by setting the  ``CYTHON_LIMITED_API`` C macro.  Note that the
support is currently in an early stage and many features do not yet work.
You currently still have to define ``Py_LIMITED_API`` externally in order
to restrict the API usage.  This will change when the feature stabilises.
Patches by Eddie Elizondo and David Woods.  (Github issues :issue:`3223`,
:issue:`3311`, :issue:`3501`)

* The dispatch to fused functions is now linear in the number of arguments,
which makes it much faster, often 2x or more, and several times faster for
larger fused types with many specialisations.
Patch by will-ca.  (Github issue :issue:`1385`)

* ``with gil/nogil`` statements can be conditional based on compile-time
constants, e.g. fused type checks.
Patch by Noam Hershtig.  (Github issue :issue:`2579`)

* ``const`` can be used together with fused types.
Patch by Thomas Vincent.  (Github issue :issue:`1772`)

* Reimports of already imported modules are substantially faster.
(Github issue :issue:`2854`)

* Positional-only arguments are supported in Python functions (`PEP-570`_).
Patch by Josh Tobin.  (Github issue :issue:`2915`)

* The ``volatile`` C modifier is supported in Cython code.
Patch by Jeroen Demeyer.  (Github issue :issue:`1667`)

* ``cython.trashcan(True)`` can be used on an extension type to enable the
CPython :ref:`trashcan`. This allows deallocating deeply recursive objects
without overflowing the stack. Patch by Jeroen Demeyer.  (Github issue :issue:`2842`)

* Inlined properties can be defined for external extension types.
Patch by Matti Picus. (Github issue :issue:`2640`, redone later in :issue:`3571`)

* The ``str()`` builtin now calls ``PyObject_Str()`` instead of going
through a Python call.
Patch by William Ayd.  (Github issue :issue:`3279`)

* String concatenation can now happen in place if possible, by extending the
existing string rather than always creating a new one.
Patch by David Woods.  (Github issue :issue:`3453`)

* Multiplication of Python numbers with small constant integers is faster.
(Github issue :issue:`2808`)

* Some list copying is avoided internally when a new list needs to be created
but we already have a fresh one.
(Github issue :issue:`3494`)

* Extension types that do not need their own ``tp_new`` implementation (because
they have no object attributes etc.) directly inherit the implementation of
their parent type if possible.
(Github issue :issue:`1555`)

* The attributes ``gen.gi_frame`` and ``coro.cr_frame`` of Cython compiled
generators and coroutines now return an actual frame object for introspection.
(Github issue :issue:`2306`)

* Several declarations in ``cpython.*``, ``libc.*`` and ``libcpp.*`` were added.
Patches by Jeroen Demeyer, Matthew Edwards, Chris Gyurgyik, Jerome Kieffer
and Zackery Spytz.
(Github issues :issue:`3468`, :issue:`3332`, :issue:`3202`, :issue:`3188`,
:issue:`3179`, :issue:`2891`, :issue:`2826`, :issue:`2713`)

* Deprecated NumPy API usages were removed from ``numpy.pxd``.
Patch by Matti Picus.  (Github issue :issue:`3365`)

* ``cython.inline()`` now sets the ``NPY_NO_DEPRECATED_API=NPY_1_7_API_VERSION``
C macro automatically when ``numpy`` is imported in the code, to avoid C compiler
warnings about deprecated NumPy C-API usage.

* The builtin ``abs()`` function can now be used on C numbers in nogil code.
Patch by Elliott Sales de Andrade.  (Github issue :issue:`2748`)

* `PEP-479`_ (``generator_stop``) is now enabled by default with language level 3.
(Github issue :issue:`2580`)

* The ``cython.view.array`` type supports inheritance.
Patch by David Woods.  (Github issue :issue:`3413`)

* Code annotation accepts a new debugging argument ``--annotate-fullc`` that
will include the complete syntax highlighted C file in the HTML output.
(Github issue :issue:`2855`)

* ``--no-capture`` added to ``runtests.py`` to prevent stdout/stderr capturing
during srctree tests.
Patch by Matti Picus.  (Github issue :issue:`2701`)

* ``--no-docstrings`` option added to ``cythonize`` script.
Original patch by mo-han.  (Github issue :issue:`2889`)

* ``cygdb`` gives better error messages when it fails to initialise the
Python runtime support in gdb.
Patch by Volker Weissmann.  (Github issue :issue:`3489`)

* The Pythran ``shape`` attribute is supported.
Patch by Serge Guelton.  (Github issue :issue:`3307`)

Bugs fixed
----------

* The unicode methods ``.upper()``, ``.lower()`` and ``.title()`` were
incorrectly optimised for single character input values and only returned
the first character if multiple characters should have been returned.
They now use the original Python methods again.

* Fused argument types were not correctly handled in type annotations and
``cython.locals()``.
Patch by David Woods.  (Github issues :issue:`3391`, :issue:`3142`)

* Diverging from the usual behaviour, ``len(memoryview)``, ``len(char*)``
and ``len(Py_UNICODE*)`` returned an unsigned ``size_t`` value.  They now
return a signed ``Py_ssize_t``, like other usages of ``len()``.

* Nested dict literals in function call kwargs could incorrectly raise an
error about duplicate keyword arguments, which are allowed when passing
them from dict literals.
(Github issue :issue:`2963`)

* Item access (subscripting) with integer indices/keys always tried the
Sequence protocol before the Mapping protocol, which diverged from Python
semantics.  It now passes through the Mapping protocol first when supported.
(Github issue :issue:`1807`)

* Name lookups in class bodies no longer go through an attribute lookup.
Patch by Jeroen Demeyer.  (Github issue :issue:`3100`)

* Broadcast assignments to a multi-dimensional memory view slice could end
up in the wrong places when the underlying memory view is known to be
contiguous but the slice is not.
(Github issue :issue:`2941`)

* Pickling unbound methods of Python classes failed.
Patch by Pierre Glaser.  (Github issue :issue:`2972`)

* The ``Py_hash_t`` type failed to accept arbitrary &quot;index&quot; values.
(Github issue :issue:`2752`)

* The first function line number of functions with decorators pointed to the
signature line and not the first decorator line, as in Python.
Patch by Felix Kohlgrüber.  (Github issue :issue:`2536`)

* Constant integer expressions that used a negative exponent were evaluated
as integer 0 instead of the expected float value.
Patch by Kryštof Pilnáček.  (Github issue :issue:`2133`)

* The ``cython.declare()`` and ``cython.cast()`` functions could fail in pure mode.
Patch by Dmitry Shesterkin.  (Github issue :issue:`3244`)

* ``__doc__`` was not available inside of the class body during class creation.
(Github issue :issue:`1635`)

* Setting ``language_level=2`` in a file did not work if ``language_level=3``
was enabled globally before.
Patch by Jeroen Demeyer.  (Github issue :issue:`2791`)

* ``__init__.pyx`` files were not always considered as package indicators.
(Github issue :issue:`2665`)

* Compiling package ``__init__`` files could fail under Windows due to an
undefined export symbol.  (Github issue :issue:`2968`)

* A C compiler cast warning was resolved.
Patch by Michael Buesch.  (Github issue :issue:`2775`)

* Binding staticmethods of Cython functions were not behaving like Python methods.
Patch by Jeroen Demeyer.  (Github issue :issue:`3106`, :issue:`3102`)

* Memoryviews failed to compile when the ``cache_builtins`` feature was disabled.
Patch by David Woods.  (Github issue :issue:`3406`)

Other changes
-------------

* The default language level was changed to ``3str``, i.e. Python 3 semantics,
but with ``str`` literals (also in Python 2.7).  This is a backwards incompatible
change from the previous default of Python 2 semantics.  The previous behaviour
is available through the directive ``language_level=2``.
(Github issue :issue:`2565`)

* Cython no longer generates ``__qualname__`` attributes for classes in Python
2.x since they are problematic there and not correctly maintained for subclasses.
Patch by Jeroen Demeyer.  (Github issue :issue:`2772`)

* Source file fingerprinting now uses SHA-1 instead of MD5 since the latter
tends to be slower and less widely supported these days.
(Github issue :issue:`2790`)

* The long deprecated include files ``python_*``, ``stdio``, ``stdlib`` and
``stl`` in ``Cython/Includes/Deprecated/`` were removed.  Use the ``libc.*``
and ``cpython.*`` pxd modules instead.
Patch by Jeroen Demeyer.  (Github issue :issue:`2904`)

* The search order for include files was changed. Previously it was
``include_directories``, ``Cython/Includes``, ``sys.path``. Now it is
``include_directories``, ``sys.path``, ``Cython/Includes``. This was done to
allow third-party ``*.pxd`` files to override the ones in Cython.
Patch by Matti Picus.  (Github issue :issue:`2905`)

* The command line parser was rewritten and modernised using ``argparse``.
Patch by Egor Dranischnikow.  (Github issue :issue:`2952`, :issue:`3001`)

* Dotted filenames for qualified module names (``pkg.mod.pyx``) are deprecated.
Use the normal Python package directory layout instead.
(Github issue :issue:`2686`)

* Binary Linux wheels now follow the manylinux2010 standard.
Patch by Alexey Stepanov.  (Github issue :issue:`3355`)

* Support for Python 2.6 was removed.

.. _`PEP-560`: https://www.python.org/dev/peps/pep-0560
.. _`PEP-570`: https://www.python.org/dev/peps/pep-0570
.. _`PEP-487`: https://www.python.org/dev/peps/pep-0487
.. _`PEP-590`: https://www.python.org/dev/peps/pep-0590
.. _`PEP-3131`: https://www.python.org/dev/peps/pep-3131
.. _`PEP-563`: https://www.python.org/dev/peps/pep-0563
.. _`PEP-479`: https://www.python.org/dev/peps/pep-0479

0.29.37

====================

Bugs fixed
----------

* Fix a potential crash while cleaning up subtypes of externally imported extension
types when terminating Python.  This was introduced in Cython 0.29.35.

* Fix a ``complex`` related compile error on Windows.
(Github issue :issue:`5512`)

* Compiling fused types used in pxd files could crash Cython in Python 3.11+.
(Github issues :issue:`5894`, :issue:`5588`)

* ``cythonize`` failed to consider the ``CYTHON_FORCE_REGEN`` env variable.
Patch by Harmen Stoppels.  (Github issue :issue:`5712`)

0.29.36

====================

Bugs fixed
----------

* Async generators lost their return value in PyPy.
(Github issue :issue:`5465`)

* The outdated C macro ``_PyGC_FINALIZED()`` is no longer used in Py3.9+.

* The deprecated ``Py_OptimizeFlag`` is no longer used in Python 3.9+.
(Github issue :issue:`5343`)

* Using the global ``__debug__`` variable but not assertions could lead to compile errors.

* The broken HTML template support was removed from Tempita.
(Github issue :issue:`3309`)

0.29.35

====================

Bugs fixed
----------

* A garbage collection enabled subtype of a non-GC extension type could call into the
deallocation function of the super type with GC tracking enabled.  This could lead
to crashes during deallocation if GC was triggered on the type at the same time.
(Github issue :issue:`5432`)

* Some C compile failures and crashes in CPython 3.12 were resolved.

* ``except + nogil`` was syntactically not allowed.
``except +nogil`` (i.e. defining a C++ exception handling function called ``nogil``)
is now disallowed to prevent typos.
(Github issue :issue:`5430`)

* A C compile failure in PyPy 3.10 was resolved.
Patch by Matti Picus.  (Github issue :issue:`5408`)

* Cython modules now use PEP-489 multi-phase init by default in PyPy 3.9 and later.
Original patch by Matti Picus.  (Github issue :issue:`5413`)

* API header files generated by different Cython versions can now be included in the
same C file.
(Github issue :issue:`5383`)

* Function signatures containing a type like `tuple[()]` could not be printed.
Patch by Lisandro Dalcin.  (Github issue :issue:`5355`)

0.29.34

====================

Bugs fixed
----------

* A reference leak of the for-loop list/tuple iterable was resolved if the for-loop&#x27;s
``else:`` branch executes a ``break`` for an outer loop.
(Github issue :issue:`5347`)

* Some C compile failures in CPython 3.12 were resolved.

* Some old usages of the deprecated Python ``imp`` module were replaced with ``importlib``.
Patch by Matúš Valo.  (Github issue :issue:`5300`)

* Some issues with ``depfile`` generation were resolved.
Patches by Eli Schwartz.  (Github issues :issue:`5279`, :issue:`5291`)

0.29.33

====================

Features added
--------------

* The ``cythonize`` and ``cython`` commands have a new option ``-M`` / ``--depfile``
to generate ``.dep`` dependency files for the compilation unit.  This can be used
by external build tools to track these dependencies.
The ``cythonize`` option was already available in Cython :ref:`0.29.27`.
Patches by Evgeni Burovski and Eli Schwartz.  (Github issue :issue:`1214`)

Bugs fixed
----------

* ``const`` fused types could not be used with memory views.
Patch by Thomas Vincent.  (Github issue :issue:`1772`)

* ``wstr`` usage was removed in Python 3.12 and later (PEP-623).
(Github issue :issue:`5145`)

* A type check assertion for Cython functions failed in debug Python builds.
(Github issue :issue:`5031`)

* Fixed various compiler warnings.
Patches by Lisandro Dalcin et al.  (Github issues :issue:`4948`, :issue:`5086`)

* Fixed error when calculating complex powers of negative numbers.
(Github issue :issue:`5014`)

* Corrected a small mis-formatting of exception messages on Python 2.
(Github issue :issue:`5018`)

* The ``PyUnicode_AsUTF8AndSize()`` C-API function was missing from the CPython declarations.
(Github issue :issue:`5163`)

* A performance problem in the compiler was resolved when nesting conditional expressions.
(Github issue :issue:`5197`)

* Test suite problems with recent NumPy and CPython versions were resolved.
(Github issues :issue:`5183`, :issue:`5190`)

Other changes
-------------

* The undocumented, untested and apparently useless syntax
``from somemodule cimport class/struct/union somename`` was deprecated
in anticipation of its removal in Cython 3.  The type
modifier is not needed here and a plain ``cimport`` of the name will do.
(Github issue :issue:`4905`)

* Properly disable generation of descriptor docstrings on PyPy since they cause crashes.
It was previously disabled, but only accidentally via a typo.
Patch by Matti Picus.  (Github issue :issue:`5083`)

* The ``cpow`` directive of Cython 3.0 is available as a no-op.
(Github issue :issue:`5016`)

0.29.32

====================

Bugs fixed
----------

* Revert &quot;Using memoryview typed arguments in inner functions is now rejected as unsupported.&quot;
Patch by David Woods.  (Github issue :issue:`4798`)

* ``from module import *`` failed in 0.29.31 when using memoryviews.
Patch by David Woods.  (Github issue :issue:`4927`)

0.29.31

====================

Features added
--------------

* A new argument ``--module-name`` was added to the ``cython`` command to
provide the (one) exact target module name from the command line.
Patch by Matthew Brett and h-vetinari.  (Github issue :issue:`4906`)

* A new keyword ``noexcept`` was added for forward compatibility with Cython 3.
Patch by David Woods.  (Github issue :issue:`4903`)

Bugs fixed
----------

* Use ``importlib.util.find_spec()`` instead of the deprecated ``importlib.find_loader()``
function when setting up the package path at import-time.
Patch by Matti Picus.  (Github issue :issue:`4764`)

* Require the C compiler to support the two-arg form of ``va_start``
on Python 3.10 and higher.
Patch by Thomas Caswell.  (Github issue :issue:`4820`)

* Make ``fused_type`` subscriptable in Shadow.py.
Patch by Pfebrer.  (Github issue :issue:`4842`)

* Fix the incorrect code generation of the target type in ``bytearray`` loops.
Patch by Kenrick Everett.  (Github issue :issue:`4108`)

* Atomic refcounts for memoryviews were not used on some GCC versions by accident.
Patch by Sam Gross.  (Github issue :issue:`4915`)

* Silence some GCC ``-Wconversion`` warnings in C utility code.
Patch by Lisandro Dalcin.  (Github issue :issue:`4854`)

* Tuple multiplication was ignored in expressions such as ``[*(1,) * 2]``.
Patch by David Woods.  (Github issue :issue:`4864`)

* Calling ``append`` methods on extension types could fail to find the method
in some cases.
Patch by David Woods.  (Github issue :issue:`4828`)

* Ensure that object buffers (e.g. ``ndarray[object, ndim=1]``) containing
``NULL``  pointers are safe to use, returning ``None`` instead of the ``NULL``
pointer.
Patch by Sebastian Berg.  (Github issue :issue:`4859`)

* Using memoryview typed arguments in inner functions is now rejected as unsupported.
Patch by David Woods.  (Github issue :issue:`4798`)

* Compilation could fail on systems (e.g. FIPS) that block MD5 checksums at runtime.
(Github issue :issue:`4909`)

* Experimental adaptations for the CPython &quot;nogil&quot; fork was added.
Note that there is no official support for this in Cython 0.x.
Patch by Sam Gross.  (Github issue :issue:`4912`)

0.29.30

====================

Bugs fixed
----------

* The GIL handling changes in 0.29.29 introduced a regression where
objects could be deallocated without holding the GIL.
(Github issue :issue:`4796`)

0.29.29

====================

Features added
--------------

* Avoid acquiring the GIL at the end of nogil functions.
This change was backported in order to avoid generating wrong C code
that would trigger C compiler warnings with tracing support enabled.
Backport by Oleksandr Pavlyk.  (Github issue :issue:`4637`)

Bugs fixed
----------

* Function definitions in ``finally:`` clauses were not correctly generated.
Patch by David Woods.  (Github issue :issue:`4651`)

* A case where C-API functions could be called with a live exception set was fixed.
Patch by Jakub Kulík.  (Github issue :issue:`4722`)

* Pickles can now be exchanged again with those generated from Cython 3.0 modules.
(Github issue :issue:`4680`)

* Cython now correctly generates Python methods for both the provided regular and
reversed special numeric methods of extension types.
Patch by David Woods.  (Github issue :issue:`4750`)

* Calling unbound extension type methods without arguments could raise an
``IndexError`` instead of a ``TypeError``.
Patch by David Woods.  (Github issue :issue:`4779`)

* Calling unbound ``.__contains__()`` super class methods on some builtin base
types could trigger an infinite recursion.
Patch by David Woods.  (Github issue :issue:`4785`)

* The C union type in pure Python mode mishandled some field names.
Patch by Jordan Brière.  (Github issue :issue:`4727`)

* Allow users to overwrite the C macro ``_USE_MATH_DEFINES``.
Patch by Yuriy Chernyshov. (Github issue :issue:`4690`)

* Improved compatibility with CPython 3.10/11.
Patches by Thomas Caswell, David Woods.
(Github issues :issue:`4609`, :issue:`4667`, :issue:`4721`, :issue:`4730`, :issue:`4777`)

* Docstrings of descriptors are now provided in PyPy 7.3.9.
Patch by Matti Picus.  (Github issue :issue:`4701`)

0.29.28

====================

Bugs fixed
----------

* Due to backwards incompatible changes in CPython 3.11a4, the feature flags
``CYTHON_FAST_THREAD_STATE`` and ``CYTHON_USE_EXC_INFO_STACK`` are now disabled
in Python 3.11 and later.  They are enabled again in Cython 3.0.
Patch by David Woods.  (Github issue :issue:`4610`)

* A C compiler warning in older PyPy versions was resolved.
Patch by Matti Picus.  (Github issue :issue:`4236`)

0.29.27

====================

Features added
--------------

* The ``cythonize`` command has a new option ``-M`` to generate ``.dep`` dependency
files for the compilation unit.  This can be used by external build tools to track
these dependencies.
Patch by Evgeni Burovski.  (Github issue :issue:`1214`)

Bugs fixed
----------

* Compilation failures on PyPy were resolved.
Patches by Matti Picus.  (Github issues :issue:`4509`, :issue:`4517`)

* Calls to ``range()`` with more than three arguments did not fail.
Original patch by Max Bachmann.  (Github issue :issue:`4550`)

* Some C compiler warnings about missing type struct initialisers in Py3.10 were resolved.

* Cython no longer warns about using OpenMP 3.0 features since they are now
considered generally available.

0.29.26

====================

Bugs fixed
----------

* An incompatibility with CPython 3.11.0a3 was resolved.
(Github issue :issue:`4499`)

* The ``in`` operator failed on literal lists with starred expressions.
Patch by Arvind Natarajan.  (Github issue :issue:`3938`)

* A C compiler warning in PyPy about a missing struct field initialisation was resolved.

0.29.25

====================

Bugs fixed
----------

* Several incompatibilities with CPython 3.11 were resolved.
Patches by David Woods, Victor Stinner, Thomas Caswell.
(Github issues :issue:`4411`, :issue:`4414`, :issue:`4415`, :issue:`4416`, :issue:`4420`,
:issue:`4428`, :issue:`4473`, :issue:`4479`, :issue:`4480`)

* Some C compiler warnings were resolved.
Patches by Lisandro Dalcin and others.  (Github issue :issue:`4439`)

* C++ ``std::move()`` should only be used automatically in MSVC versions that support it.
Patch by Max Bachmann.  (Github issue :issue:`4191`)

* The ``Py_hash_t`` type failed to accept arbitrary &quot;index&quot; values.
(Github issue :issue:`2752`)

* Avoid copying unaligned 16-bit values since some platforms require them to be aligned.
Use memcpy() instead to let the C compiler decide how to do it.
(Github issue :issue:`4343`)

* Cython crashed on invalid truthiness tests on C++ types without ``operator bool``.
Patch by David Woods.  (Github issue :issue:`4348`)

* The declaration of ``PyUnicode_CompareWithASCIIString()`` in ``cpython.unicode`` was incorrect.
Patch by Max Bachmann.  (Github issue :issue:`4344`)

0.29.24

====================

Bugs fixed
----------

* Inline functions in pxd files that used memory views could lead to invalid
C code if the module that imported from them does not use memory views.
Patch by David Woods.  (Github issue :issue:`1415`)

* Several declarations in ``libcpp.string`` were added and corrected.
Patch by Janek Bevendorff.  (Github issue :issue:`4268`)

* Pickling unbound Cython compiled methods failed.
Patch by Pierre Glaser.  (Github issue :issue:`2972`)

* The tracing code was adapted to work with CPython 3.10.

* The optimised ``in`` operator failed on unicode strings in Py3.9 and later
that were constructed from an external ``wchar_t`` source.
Also, related C compiler warnings about deprecated C-API usage were resolved.
(Github issue :issue:`3925`)

* Some compiler crashes were resolved.
Patch by David Woods.  (Github issues :issue:`4214`, :issue:`2811`)

* An incorrect warning about &#x27;unused&#x27; generator expressions was removed.
(GIthub issue :issue:`1699`)

* The attributes ``gen.gi_frame`` and ``coro.cr_frame`` of Cython compiled
generators and coroutines now return an actual frame object for introspection,
instead of ``None``.
(Github issue :issue:`2306`)

Links

PyPI: https://pypi.org/project/cython
Changelog: https://data.safetycli.com/changelogs/cython/
Homepage: https://cython.org/

Update numpy from 1.21.0 to 2.1.1.

Changelog

2.1.1

discovered after the 2.1.0 release.

The Python versions supported by this release are 3.10-3.13.

Contributors

A total of 7 people contributed to this release. People with a \&quot;+\&quot; by
their names contributed a patch for the first time.

-   Andrew Nelson
-   Charles Harris
-   Mateusz Sokół
-   Maximilian Weigand +
-   Nathan Goldbaum
-   Pieter Eendebak
-   Sebastian Berg

Pull requests merged

A total of 10 pull requests were merged for this release.

-   [27236](https://github.com/numpy/numpy/pull/27236): REL: Prepare for the NumPy 2.1.0 release \[wheel build\]
-   [27252](https://github.com/numpy/numpy/pull/27252): MAINT: prepare 2.1.x for further development
-   [27259](https://github.com/numpy/numpy/pull/27259): BUG: revert unintended change in the return value of set_printoptions
-   [27266](https://github.com/numpy/numpy/pull/27266): BUG: fix reference counting bug in \_\_array_interface\_\_ implementation...
-   [27267](https://github.com/numpy/numpy/pull/27267): TST: Add regression test for missing descr in array-interface
-   [27276](https://github.com/numpy/numpy/pull/27276): BUG: Fix #27256 and 27257
-   [27278](https://github.com/numpy/numpy/pull/27278): BUG: Fix array_equal for numeric and non-numeric scalar types
-   [27287](https://github.com/numpy/numpy/pull/27287): MAINT: Update maintenance/2.1.x after the 2.0.2 release
-   [27303](https://github.com/numpy/numpy/pull/27303): BLD: cp311- macosx_arm64 wheels \[wheel build\]
-   [27304](https://github.com/numpy/numpy/pull/27304): BUG: f2py: better handle filtering of public/private subroutines

Checksums

MD5

 3053a97400db800b7377749e691eb39e  numpy-2.1.1-cp310-cp310-macosx_10_9_x86_64.whl
 84b752a2220dce7c96ff89eef4f4aec3  numpy-2.1.1-cp310-cp310-macosx_11_0_arm64.whl
 47ed4f704a64261f07ca24ef2e674524  numpy-2.1.1-cp310-cp310-macosx_14_0_arm64.whl
 b8a45caa870aee980c298053cf064d28  numpy-2.1.1-cp310-cp310-macosx_14_0_x86_64.whl
 e097ad5eee572b791b4a25eedad6df4a  numpy-2.1.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
 ae502c99315884cda7f0236a07c035c4  numpy-2.1.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
 841a859d975c55090c0b60b72aab93a3  numpy-2.1.1-cp310-cp310-musllinux_1_1_x86_64.whl
 d51be2b17f5b87aac64ab80fdfafc85e  numpy-2.1.1-cp310-cp310-musllinux_1_2_aarch64.whl
 1f8249bd725397c6233fe6a0e8ad18b1  numpy-2.1.1-cp310-cp310-win32.whl
 d38d6f06589c1ec104a6a31ff6035781  numpy-2.1.1-cp310-cp310-win_amd64.whl
 6a18fe3029aae00986975250313bf16f  numpy-2.1.1-cp311-cp311-macosx_10_9_x86_64.whl
 5b0b3aa01fbd0b5a8b0f354bb878351e  numpy-2.1.1-cp311-cp311-macosx_11_0_arm64.whl
 1c492dad399abe7b97274b4c6c12ae53  numpy-2.1.1-cp311-cp311-macosx_14_0_arm64.whl
 4d55d91e71b62eb5fa6561c606524f60  numpy-2.1.1-cp311-cp311-macosx_14_0_x86_64.whl
 88e99ecd063c178f25bc08d20792a9bf  numpy-2.1.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
 f3c8b0e4fb059b9219e8ec86d9fda861  numpy-2.1.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
 df632b5fed7eb78d39e7194d2475c19b  numpy-2.1.1-cp311-cp311-musllinux_1_1_x86_64.whl
 65499daccdb178d26e322d9f359cf146  numpy-2.1.1-cp311-cp311-musllinux_1_2_aarch64.whl
 eb97327fd7aa6027e2409d0dcca1129a  numpy-2.1.1-cp311-cp311-win32.whl
 9e4b05b38cbff22c2bdfead528b9d2bc  numpy-2.1.1-cp311-cp311-win_amd64.whl
 6b8a359bb865b5c624fd9ffc848393e1  numpy-2.1.1-cp312-cp312-macosx_10_9_x86_64.whl
 eaf8dce312efa2b0f17ad46612fb1681  numpy-2.1.1-cp312-cp312-macosx_11_0_arm64.whl
 c861ff048b336284fe7c0791b1a6b0b4  numpy-2.1.1-cp312-cp312-macosx_14_0_arm64.whl
 7e1befccfe729dc5d6c450a5fb6b801c  numpy-2.1.1-cp312-cp312-macosx_14_0_x86_64.whl
 ea0a401ef653a167221987a10cbef260  numpy-2.1.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
 97326ac792d26f2e536a519c82f2d6bc  numpy-2.1.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
 fdd2a82232c03d11bbc7cec0a8e01ab0  numpy-2.1.1-cp312-cp312-musllinux_1_1_x86_64.whl
 0d6716e9a7b2c0d6e5ace9c01b9bca01  numpy-2.1.1-cp312-cp312-musllinux_1_2_aarch64.whl
 ba589ed2a79c88187c3b8574ae72a1c7  numpy-2.1.1-cp312-cp312-win32.whl
 806ca7c1e2a2013b786edbb619f6da47  numpy-2.1.1-cp312-cp312-win_amd64.whl
 647665353e5af5884df4e51610990c22  numpy-2.1.1-cp313-cp313-macosx_10_13_x86_64.whl
 bfd3b3c5c4616ef99d917bd94d39114a  numpy-2.1.1-cp313-cp313-macosx_11_0_arm64.whl
 cb989095f9c74e3b32250a984390faeb  numpy-2.1.1-cp313-cp313-macosx_14_0_arm64.whl
 55ad7548e58f61b9a4f91749e36d237f  numpy-2.1.1-cp313-cp313-macosx_14_0_x86_64.whl
 5bc73d67dd1032524bfd36ef877b09e4  numpy-2.1.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
 c7dfb09db8284cb75296f708c3f77ea3  numpy-2.1.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
 7cf90ce1b844a97aeea1a5b8c71fb49b  numpy-2.1.1-cp313-cp313-musllinux_1_1_x86_64.whl
 6ec8baeac5f979a3b98017679d457bbc  numpy-2.1.1-cp313-cp313-musllinux_1_2_aarch64.whl
 1f198cb5210c76faae81359a83d58230  numpy-2.1.1-cp313-cp313-win32.whl
 1766258213ad41f7e36f2209ee6d2a30  numpy-2.1.1-cp313-cp313-win_amd64.whl
 f0a7a0456308dbeb739ad886f1632f16  numpy-2.1.1-cp313-cp313t-macosx_10_13_x86_64.whl
 302c9cf7b4aa695974500ee1935a92c9  numpy-2.1.1-cp313-cp313t-macosx_11_0_arm64.whl
 f4aa7d784992abb9bd9fe9db09c01c06  numpy-2.1.1-cp313-cp313t-macosx_14_0_arm64.whl
 3bb4ae9906499609769f1774438149a5  numpy-2.1.1-cp313-cp313t-macosx_14_0_x86_64.whl
 ff6b9e1993d3d540074736014b1d13af  numpy-2.1.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
 749489c091ee9c00abf1ad1ef822c3ca  numpy-2.1.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
 32d2daf4064031f365ced5036757ad8b  numpy-2.1.1-cp313-cp313t-musllinux_1_1_x86_64.whl
 603dfe4ef56c01e1fc0dcc9d5e3090ed  numpy-2.1.1-cp313-cp313t-musllinux_1_2_aarch64.whl
 70fa2d3b78633bb6061c90e17364f27f  numpy-2.1.1-pp310-pypy310_pp73-macosx_10_15_x86_64.whl
 9a430be5d14b689ed051eccc540dfbdc  numpy-2.1.1-pp310-pypy310_pp73-macosx_14_0_x86_64.whl
 7291ff124e471d32c03464da18ff108d  numpy-2.1.1-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
 e56ce141724af119c7c647a8705827a5  numpy-2.1.1-pp310-pypy310_pp73-win_amd64.whl
 f63b4750618bfa5490f10cae37fde998  numpy-2.1.1.tar.gz

SHA256

 c8a0e34993b510fc19b9a2ce7f31cb8e94ecf6e924a40c0c9dd4f62d0aac47d9  numpy-2.1.1-cp310-cp310-macosx_10_9_x86_64.whl
 7dd86dfaf7c900c0bbdcb8b16e2f6ddf1eb1fe39c6c8cca6e94844ed3152a8fd  numpy-2.1.1-cp310-cp310-macosx_11_0_arm64.whl
 5889dd24f03ca5a5b1e8a90a33b5a0846d8977565e4ae003a63d22ecddf6782f  numpy-2.1.1-cp310-cp310-macosx_14_0_arm64.whl
 59ca673ad11d4b84ceb385290ed0ebe60266e356641428c845b39cd9df6713ab  numpy-2.1.1-cp310-cp310-macosx_14_0_x86_64.whl
 13ce49a34c44b6de5241f0b38b07e44c1b2dcacd9e36c30f9c2fcb1bb5135db7  numpy-2.1.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
 913cc1d311060b1d409e609947fa1b9753701dac96e6581b58afc36b7ee35af6  numpy-2.1.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
 caf5d284ddea7462c32b8d4a6b8af030b6c9fd5332afb70e7414d7fdded4bfd0  numpy-2.1.1-cp310-cp310-musllinux_1_1_x86_64.whl
 57eb525e7c2a8fdee02d731f647146ff54ea8c973364f3b850069ffb42799647  numpy-2.1.1-cp310-cp310-musllinux_1_2_aarch64.whl
 9a8e06c7a980869ea67bbf551283bbed2856915f0a792dc32dd0f9dd2fb56728  numpy-2.1.1-cp310-cp310-win32.whl
 d10c39947a2d351d6d466b4ae83dad4c37cd6c3cdd6d5d0fa797da56f710a6ae  numpy-2.1.1-cp310-cp310-win_amd64.whl
 0d07841fd284718feffe7dd17a63a2e6c78679b2d386d3e82f44f0108c905550  numpy-2.1.1-cp311-cp311-macosx_10_9_x86_64.whl
 b5613cfeb1adfe791e8e681128f5f49f22f3fcaa942255a6124d58ca59d9528f  numpy-2.1.1-cp311-cp311-macosx_11_0_arm64.whl
 0b8cc2715a84b7c3b161f9ebbd942740aaed913584cae9cdc7f8ad5ad41943d0  numpy-2.1.1-cp311-cp311-macosx_14_0_arm64.whl
 b49742cdb85f1f81e4dc1b39dcf328244f4d8d1ded95dea725b316bd2cf18c95  numpy-2.1.1-cp311-cp311-macosx_14_0_x86_64.whl
 e8d5f8a8e3bc87334f025194c6193e408903d21ebaeb10952264943a985066ca  numpy-2.1.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
 d51fc141ddbe3f919e91a096ec739f49d686df8af254b2053ba21a910ae518bf  numpy-2.1.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
 98ce7fb5b8063cfdd86596b9c762bf2b5e35a2cdd7e967494ab78a1fa7f8b86e  numpy-2.1.1-cp311-cp311-musllinux_1_1_x86_64.whl
 24c2ad697bd8593887b019817ddd9974a7f429c14a5469d7fad413f28340a6d2  numpy-2.1.1-cp311-cp311-musllinux_1_2_aarch64.whl
 397bc5ce62d3fb73f304bec332171535c187e0643e176a6e9421a6e3eacef06d  numpy-2.1.1-cp311-cp311-win32.whl
 ae8ce252404cdd4de56dcfce8b11eac3c594a9c16c231d081fb705cf23bd4d9e  numpy-2.1.1-cp311-cp311-win_amd64.whl
 7c803b7934a7f59563db459292e6aa078bb38b7ab1446ca38dd138646a38203e  numpy-2.1.1-cp312-cp312-macosx_10_9_x86_64.whl
 6435c48250c12f001920f0751fe50c0348f5f240852cfddc5e2f97e007544cbe  numpy-2.1.1-cp312-cp312-macosx_11_0_arm64.whl
 3269c9eb8745e8d975980b3a7411a98976824e1fdef11f0aacf76147f662b15f  numpy-2.1.1-cp312-cp312-macosx_14_0_arm64.whl
 fac6e277a41163d27dfab5f4ec1f7a83fac94e170665a4a50191b545721c6521  numpy-2.1.1-cp312-cp312-macosx_14_0_x86_64.whl
 fcd8f556cdc8cfe35e70efb92463082b7f43dd7e547eb071ffc36abc0ca4699b  numpy-2.1.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
 d2b9cd92c8f8e7b313b80e93cedc12c0112088541dcedd9197b5dee3738c1201  numpy-2.1.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
 afd9c680df4de71cd58582b51e88a61feed4abcc7530bcd3d48483f20fc76f2a  numpy-2.1.1-cp312-cp312-musllinux_1_1_x86_64.whl
 8661c94e3aad18e1ea17a11f60f843a4933ccaf1a25a7c6a9182af70610b2313  numpy-2.1.1-cp312-cp312-musllinux_1_2_aarch64.whl
 950802d17a33c07cba7fd7c3dcfa7d64705509206be1606f196d179e539111ed  numpy-2.1.1-cp312-cp312-win32.whl
 3fc5eabfc720db95d68e6646e88f8b399bfedd235994016351b1d9e062c4b270  numpy-2.1.1-cp312-cp312-win_amd64.whl
 046356b19d7ad1890c751b99acad5e82dc4a02232013bd9a9a712fddf8eb60f5  numpy-2.1.1-cp313-cp313-macosx_10_13_x86_64.whl
 6e5a9cb2be39350ae6c8f79410744e80154df658d5bea06e06e0ac5bb75480d5  numpy-2.1.1-cp313-cp313-macosx_11_0_arm64.whl
 d4c57b68c8ef5e1ebf47238e99bf27657511ec3f071c465f6b1bccbef12d4136  numpy-2.1.1-cp313-cp313-macosx_14_0_arm64.whl
 8ae0fd135e0b157365ac7cc31fff27f07a5572bdfc38f9c2d43b2aff416cc8b0  numpy-2.1.1-cp313-cp313-macosx_14_0_x86_64.whl
 981707f6b31b59c0c24bcda52e5605f9701cb46da4b86c2e8023656ad3e833cb  numpy-2.1.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
 2ca4b53e1e0b279142113b8c5eb7d7a877e967c306edc34f3b58e9be12fda8df  numpy-2.1.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
 e097507396c0be4e547ff15b13dc3866f45f3680f789c1a1301b07dadd3fbc78  numpy-2.1.1-cp313-cp313-musllinux_1_1_x86_64.whl
 f7506387e191fe8cdb267f912469a3cccc538ab108471291636a96a54e599556  numpy-2.1.1-cp313-cp313-musllinux_1_2_aarch64.whl
 251105b7c42abe40e3a689881e1793370cc9724ad50d64b30b358bbb3a97553b  numpy-2.1.1-cp313-cp313-win32.whl
 f212d4f46b67ff604d11fff7cc62d36b3e8714edf68e44e9760e19be38c03eb0  numpy-2.1.1-cp313-cp313-win_amd64.whl
 920b0911bb2e4414c50e55bd658baeb78281a47feeb064ab40c2b66ecba85553  numpy-2.1.1-cp313-cp313t-macosx_10_13_x86_64.whl
 bab7c09454460a487e631ffc0c42057e3d8f2a9ddccd1e60c7bb8ed774992480  numpy-2.1.1-cp313-cp313t-macosx_11_0_arm64.whl
 cea427d1350f3fd0d2818ce7350095c1a2ee33e30961d2f0fef48576ddbbe90f  numpy-2.1.1-cp313-cp313t-macosx_14_0_arm64.whl
 e30356d530528a42eeba51420ae8bf6c6c09559051887196599d96ee5f536468  numpy-2.1.1-cp313-cp313t-macosx_14_0_x86_64.whl
 e8dfa9e94fc127c40979c3eacbae1e61fda4fe71d84869cc129e2721973231ef  numpy-2.1.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
 910b47a6d0635ec1bd53b88f86120a52bf56dcc27b51f18c7b4a2e2224c29f0f  numpy-2.1.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
 13cc11c00000848702322af4de0147ced365c81d66053a67c2e962a485b3717c  numpy-2.1.1-cp313-cp313t-musllinux_1_1_x86_64.whl
 53e27293b3a2b661c03f79aa51c3987492bd4641ef933e366e0f9f6c9bf257ec  numpy-2.1.1-cp313-cp313t-musllinux_1_2_aarch64.whl
 7be6a07520b88214ea85d8ac8b7d6d8a1839b0b5cb87412ac9f49fa934eb15d5  numpy-2.1.1-pp310-pypy310_pp73-macosx_10_15_x86_64.whl
 52ac2e48f5ad847cd43c4755520a2317f3380213493b9d8a4c5e37f3b87df504  numpy-2.1.1-pp310-pypy310_pp73-macosx_14_0_x86_64.whl
 50a95ca3560a6058d6ea91d4629a83a897ee27c00630aed9d933dff191f170cd  numpy-2.1.1-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
 99f4a9ee60eed1385a86e82288971a51e71df052ed0b2900ed30bc840c0f2e39  numpy-2.1.1-pp310-pypy310_pp73-win_amd64.whl
 d0cf7d55b1051387807405b3898efafa862997b4cba8aa5dbe657be794afeafd  numpy-2.1.1.tar.gz

2.1

3.13. This support was enabled by fixing a number of C thread-safety
issues in NumPy. Before NumPy 2.1, NumPy used a large number of C global
static variables to store runtime caches and other state. We have either
refactored to avoid the need for global state, converted the global
state to thread-local state, or added locking.

Support for free-threaded Python does not mean that NumPy is thread
safe. Read-only shared access to ndarray should be safe. NumPy exposes
shared mutable state and we have not added any locking to the array
object itself to serialize access to shared state. Care must be taken in
user code to avoid races if you would like to mutate the same array in
multiple threads. It is certainly possible to crash NumPy by mutating an
array simultaneously in multiple threads, for example by calling a ufunc
and the `resize` method simultaneously. For now our guidance is:
\&quot;don\&#x27;t do that\&quot;. In the future we would like to provide stronger
guarantees.

Object arrays in particular need special care, since the GIL previously
provided locking for object array access and no longer does. See
[Issue 27199](https://github.com/numpy/numpy/issues/27199) for more information about object
arrays in the free-threaded build.

If you are interested in free-threaded Python, for example because you
have a multiprocessing-based workflow that you are interested in running
with Python threads, we encourage testing and experimentation.

If you run into problems that you suspect are because of NumPy, please
[open an issue](https://github.com/numpy/numpy/issues/new/choose),
checking first if the bug also occurs in the \&quot;regular\&quot; non-free-threaded CPython 3.13 
build. Many threading bugs can also occur in code that releases
the GIL; disabling the GIL only makes it easier to hit threading bugs.

([gh-26157](https://github.com/numpy/numpy/issues/26157#issuecomment-2233864940))

`f2py` can generate freethreading-compatible C extensions

Pass `--freethreading-compatible` to the f2py CLI tool to produce a C
extension marked as compatible with the free threading CPython
interpreter. Doing so prevents the interpreter from re-enabling the GIL
at runtime when it imports the C extension. Note that `f2py` does not
analyze fortran code for thread safety, so you must verify that the
wrapped fortran code is thread safe before marking the extension as
compatible.

([gh-26981](https://github.com/numpy/numpy/pull/26981))

-   `numpy.reshape` and `numpy.ndarray.reshape` now support `shape` and
 `copy` arguments.

 ([gh-26292](https://github.com/numpy/numpy/pull/26292))

-   NumPy now supports DLPack v1, support for older versions will be
 deprecated in the future.

 ([gh-26501](https://github.com/numpy/numpy/pull/26501))

-   `numpy.asanyarray` now supports `copy` and `device` arguments,
 matching `numpy.asarray`.

 ([gh-26580](https://github.com/numpy/numpy/pull/26580))

-   `numpy.printoptions`, `numpy.get_printoptions`, and
 `numpy.set_printoptions` now support a new option, `override_repr`,
 for defining custom `repr(array)` behavior.

 ([gh-26611](https://github.com/numpy/numpy/pull/26611))

-   `numpy.cumulative_sum` and `numpy.cumulative_prod` were added as
 Array API compatible alternatives for `numpy.cumsum` and
 `numpy.cumprod`. The new functions can include a fixed initial
 (zeros for `sum` and ones for `prod`) in the result.

 ([gh-26724](https://github.com/numpy/numpy/pull/26724))

-   `numpy.clip` now supports `max` and `min` keyword arguments which
 are meant to replace `a_min` and `a_max`. Also, for `np.clip(a)` or
 `np.clip(a, None, None)` a copy of the input array will be returned
 instead of raising an error.

 ([gh-26724](https://github.com/numpy/numpy/pull/26724))

-   `numpy.astype` now supports `device` argument.

 ([gh-26724](https://github.com/numpy/numpy/pull/26724))


Improvements

`histogram` auto-binning now returns bin sizes \&gt;=1 for integer input data

For integer input data, bin sizes smaller than 1 result in spurious
empty bins. This is now avoided when the number of bins is computed
using one of the algorithms provided by `histogram_bin_edges`.

([gh-12150](https://github.com/numpy/numpy/pull/12150))

`ndarray` shape-type parameter is now covariant and bound to `tuple[int, ...]`

Static typing for `ndarray` is a long-term effort that continues with
this change. It is a generic type with type parameters for the shape and
the data type. Previously, the shape type parameter could be any value.
This change restricts it to a tuple of ints, as one would expect from
using `ndarray.shape`. Further, the shape-type parameter has been
changed from invariant to covariant. This change also applies to the
subtypes of `ndarray`, e.g. `numpy.ma.MaskedArray`. See the
[typing docs](https://typing.readthedocs.io/en/latest/reference/generics.html#variance-of-generic-types)
for more information.

([gh-26081](https://github.com/numpy/numpy/pull/26081))

`np.quantile` with method `closest_observation` chooses nearest even order statistic

This changes the definition of nearest for border cases from the nearest
odd order statistic to nearest even order statistic. The numpy
implementation now matches other reference implementations.

([gh-26656](https://github.com/numpy/numpy/pull/26656))

`lapack_lite` is now thread safe

NumPy provides a minimal low-performance version of LAPACK named
`lapack_lite` that can be used if no BLAS/LAPACK system is detected at
build time.

Until now, `lapack_lite` was not thread safe. Single-threaded use cases
did not hit any issues, but running linear algebra operations in
multiple threads could lead to errors, incorrect results, or segfaults
due to data races.

We have added a global lock, serializing access to `lapack_lite` in
multiple threads.

([gh-26750](https://github.com/numpy/numpy/pull/26750))

The `numpy.printoptions` context manager is now thread and async-safe

In prior versions of NumPy, the printoptions were defined using a
combination of Python and C global variables. We have refactored so the
state is stored in a python `ContextVar`, making the context manager
thread and async-safe.

([gh-26846](https://github.com/numpy/numpy/pull/26846))

Type hinting `numpy.polynomial`

Starting from the 2.1 release, PEP 484 type annotations have been
included for the functions and convenience classes in `numpy.polynomial`
and its sub-packages.

([gh-26897](https://github.com/numpy/numpy/pull/26897))

Improved `numpy.dtypes` type hints

The type annotations for `numpy.dtypes` are now a better reflection of
the runtime: The `numpy.dtype` type-aliases have been replaced with
specialized `dtype` *subtypes*, and the previously missing annotations
for `numpy.dtypes.StringDType` have been added.

([gh-27008](https://github.com/numpy/numpy/pull/27008))

Performance improvements and changes

-   `numpy.save` now uses pickle protocol version 4 for saving arrays
 with object dtype, which allows for pickle objects larger than 4GB
 and improves saving speed by about 5% for large arrays.

 ([gh-26388](https://github.com/numpy/numpy/pull/26388))

-   OpenBLAS on x86_64 and i686 is built with fewer kernels. Based on
 benchmarking, there are 5 clusters of performance around these
 kernels: `PRESCOTT NEHALEM SANDYBRIDGE HASWELL SKYLAKEX`.

 ([gh-27147](https://github.com/numpy/numpy/pull/27147))

-   OpenBLAS on windows is linked without quadmath, simplifying
 licensing

 ([gh-27147](https://github.com/numpy/numpy/pull/27147))

-   Due to a regression in OpenBLAS on windows, the performance
 improvements when using multiple threads for OpenBLAS 0.3.26 were
 reverted.

 ([gh-27147](https://github.com/numpy/numpy/pull/27147))

`ma.cov` and `ma.corrcoef` are now significantly faster

The private function has been refactored along with `ma.cov` and
`ma.corrcoef`. They are now significantly faster, particularly on large,
masked arrays.

([gh-26285](https://github.com/numpy/numpy/pull/26285))

Changes

-   As `numpy.vecdot` is now a ufunc it has a less precise signature.
 This is due to the limitations of ufunc\&#x27;s typing stub.

 ([gh-26313](https://github.com/numpy/numpy/pull/26313))

-   `numpy.floor`, `numpy.ceil`, and `numpy.trunc` now won\&#x27;t perform
 casting to a floating dtype for integer and boolean dtype input
 arrays.

 ([gh-26766](https://github.com/numpy/numpy/pull/26766))

`ma.corrcoef` may return a slightly different result

A pairwise observation approach is currently used in `ma.corrcoef` to
calculate the standard deviations for each pair of variables. This has
been changed as it is being used to normalise the covariance, estimated
using `ma.cov`, which does not consider the observations for each
variable in a pairwise manner, rendering it unnecessary. The
normalisation has been replaced by the more appropriate standard
deviation for each variable, which significantly reduces the wall time,
but will return slightly different estimates of the correlation
coefficients in cases where the observations between a pair of variables
are not aligned. However, it will return the same estimates in all other
cases, including returning the same correlation matrix as `corrcoef`
when using a masked array with no masked values.

([gh-26285](https://github.com/numpy/numpy/pull/26285))

Cast-safety fixes in `copyto` and `full`

`copyto` now uses NEP 50 correctly and applies this to its cast safety.
Python integer to NumPy integer casts and Python float to NumPy float
casts are now considered \&quot;safe\&quot; even if assignment may fail or
precision may be lost. This means the following examples change
slightly:

-   `np.copyto(int8_arr, 1000)` previously performed an unsafe/same-kind cast
  of the Python integer. It will now always raise, to achieve an
  unsafe cast you must pass an array or NumPy scalar.

-   `np.copyto(uint8_arr, 1000, casting=&quot;safe&quot;)` will raise an
 OverflowError rather than a TypeError due to same-kind casting.

-   `np.copyto(float32_arr, 1e300, casting=&quot;safe&quot;)` will overflow to
 `inf` (float32 cannot hold `1e300`) rather raising a TypeError.

Further, only the dtype is used when assigning NumPy scalars (or 0-d
arrays), meaning that the following behaves differently:

-   `np.copyto(float32_arr, np.float64(3.0), casting=&quot;safe&quot;)` raises.
-   `np.coptyo(int8_arr, np.int64(100), casting=&quot;safe&quot;)` raises.
 Previously, NumPy checked whether the 100 fits the `int8_arr`.

This aligns `copyto`, `full`, and `full_like` with the correct NumPy 2
behavior.

([gh-27091](https://github.com/numpy/numpy/pull/27091))

Checksums

MD5

 2323404663c0b2a86362319d7526eb80  numpy-2.1.0-cp310-cp310-macosx_10_9_x86_64.whl
 3d4bca8d05eb1eba859e77ff8f91d843  numpy-2.1.0-cp310-cp310-macosx_11_0_arm64.whl
 9bd065f147dbf3f2d59ab57bff4f0074  numpy-2.1.0-cp310-cp310-macosx_14_0_arm64.whl
 47d177533511901cd6bf77f72cbd3d6e  numpy-2.1.0-cp310-cp310-macosx_14_0_x86_64.whl
 530b7f38f64216f1322b39bc50f36c0c  numpy-2.1.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
 d2a3161a10811a675a29a63e25636d83  numpy-2.1.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
 4e9fb20b080f7931791da71708740b83  numpy-2.1.0-cp310-cp310-musllinux_1_1_x86_64.whl

pyup-bot · 2024-11-01T15:12:33Z

Closing this in favor of #168

pyup-bot added 4 commits October 1, 2024 16:12

Update cython from 0.29.23 to 3.0.11

12269b5

Update numpy from 1.21.0 to 2.1.1

691a4f6

Update pysam from 0.16.0.1 to 0.22.1

e34edcd

Update tables from 3.6.1 to 3.10.1

af19c7d

pyup-bot mentioned this pull request Oct 1, 2024

Scheduled monthly dependency update for September #166

Closed

pyup-bot closed this Nov 1, 2024

alimanfoo deleted the pyup-scheduled-update-2024-10-01 branch November 1, 2024 15:12

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Scheduled monthly dependency update for October#167

Scheduled monthly dependency update for October#167
pyup-bot wants to merge 4 commits into
masterfrom
pyup-scheduled-update-2024-10-01

pyup-bot commented Oct 1, 2024

Uh oh!

pyup-bot commented Nov 1, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

pyup-bot commented Oct 1, 2024

Update Cython from 0.29.23 to 3.0.11.

3.0.11

3.0.10

3.0.9

3.0.8

3.0.7

3.0.6

3.0.5

3.0.4

3.0.3

3.0.2

3.0.1

3.0.0

0.29.37

0.29.36

0.29.35

0.29.34

0.29.33

0.29.32

0.29.31

0.29.30

0.29.29

0.29.28

0.29.27

0.29.26

0.29.25

0.29.24

Update numpy from 1.21.0 to 2.1.1.

2.1.1

2.1

Uh oh!

pyup-bot commented Nov 1, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant