Lots of new features here! We should also investigate the impact of the coming default language level change.

---------- Forwarded message ---------
From: Stefan Behnel <stefan_ml@behnel.de>
Date: Sun, Oct 14, 2018 at 11:33 AM
Subject: [Cython] Cython 0.29 final released
To: Cython-users <cython-users@googlegroups.com>, Cython-devel <cython-devel@python.org>

Dear Cythonistas,

after half a year of development, many community pull requests and a lot of
feedback and good ideas in online discussions and at conferences, I'm proud
to release Cython 0.29. This is a major feature release that comes with
many great improvements and several important bug fixes. See the long list
of changes below.



Given that Cython has been in critical production use all over the world
for several years, but never found the perfect time for a 1.0 version bump,
we designate this to become the last 0.x release of the project and decided
to skip the 1.0 release which the 0.x series has long represented anyway.

Planning has already started [1] for the next major release, titled "3.0".
It will finally switch the default Cython language level from Py2 to Py3,
to match what users expect from a Python compiler these days without
additional options or configuration.  Cython 2 code will continue to be
supported as before with the directive "language_level=2", although the
future default (called "3str", Py3 with str literals) is already available
in the 0.29 release, so users might prefer switching to that. We're happy
to hear your feedback on the proposed changes.

[1] https://github.com/cython/cython/milestone/58

Have fun,


0.29 (2018-10-14)

Features added

* PEP-489 multi-phase module initialisation has been enabled again.  Module
  reloads in other subinterpreters raise an exception to prevent corruption
  of the static module state.

* A set of ``mypy`` compatible PEP-484 declarations were added for Cython's
  C data types to integrate with static analysers in typed Python code.
  They are available in the ``Cython/Shadow.pyi`` module and describe the
  types in the special ``cython`` module that can be used for typing in
  Python code.  Original patch by Julian Gethmann. (Github issue #1965)

* Memoryviews are supported in PEP-484/526 style type declarations.
  (Github issue #2529)

* ``@cython.nogil`` is supported as a C-function decorator in Python code.
  (Github issue #2557)

* Raising exceptions from nogil code will automatically acquire the GIL,
  instead of requiring an explicit ``with gil`` block.

* C++ functions can now be declared as potentially raising both C++ and
  Python exceptions, so that Cython can handle both correctly.
  (Github issue #2615)

* ``cython.inline()`` supports a direct ``language_level`` keyword argument
  that was previously only available via a directive.

* A new language level name ``3str`` was added that mostly corresponds to
  language level 3, but keeps unprefixed string literals as type 'str' in
  both Py2 and Py3, and the builtin 'str' type unchanged.  This will become
  the default in the next Cython release and is meant to help user code
  a) transition more easily to this new default and b) migrate to Python 3
  source code semantics without making support for Python 2.x difficult.

* In CPython 3.6 and later, looking up globals in the module dict is almost
  as fast as looking up C globals.
  (Github issue #2313)

* For a Python subclass of an extension type, repeated method calls to
  non-overridden cpdef methods can avoid the attribute lookup in Py3.6+,
  which makes them 4x faster.  (Github issue #2313)

* (In-)equality comparisons of objects to integer literals are faster.
  (Github issue #2188)

* Some internal and 1-argument method calls are faster.

* Modules that cimport many external extension types from other Cython
  modules execute less import requests during module initialisation.

* Constant tuples and slices are deduplicated and only created once per
  module.  (Github issue #2292)

* The coverage plugin considers more C file extensions such as ``.cc`` and
  ``.cxx``.  (Github issue #2266)

* The ``cythonize`` command accepts compile time variable values (as set
  by ``DEF``) through the new ``-E`` option.
  Patch by Jerome Kieffer.  (Github issue #2315)

* ``pyximport`` can import from namespace packages.
  Patch by Prakhar Goel.  (Github issue #2294)

* Some missing numpy and CPython C-API declarations were added.
  Patch by John Kirkham. (Github issues #2523, #2520, #2537)

* Declarations for the ``pylifecycle`` C-API functions were added in a new
  .pxd file ``cpython.pylifecycle``.

* The Pythran support was updated to work with the latest Pythran 0.8.7.
  Original patch by Adrien Guinet.  (Github issue #2600)

* ``%a`` is included in the string formatting types that are optimised into
  f-strings.  In this case, it is also automatically mapped to ``%r`` in
  Python 2.x.

* New C macro ``CYTHON_HEX_VERSION`` to access Cython's version in the same
  style as ``PY_HEX_VERSION``.

* Constants in ``libc.math`` are now declared as ``const`` to simplify
  their handling.

* An additional ``check_size`` clause was added to the ``ctypedef class``
  name specification to allow suppressing warnings when importing modules
  with backwards-compatible ``PyTypeObject`` size changes.
  Patch by Matti Picus.  (Github issue #2627)

Bugs fixed

* The exception handling in generators and coroutines under CPython 3.7
  was adapted to the newly introduced exception stack.  Users of Cython
  0.28 who want to support Python 3.7 are encouraged to upgrade to 0.29
  to avoid potentially incorrect error reporting and tracebacks.
  (Github issue #1958)

* Crash when importing a module under Stackless Python that was built
  for CPython.  Patch by Anselm Kruis.  (Github issue #2534)

* 2-value slicing of typed sequences failed if the start or stop index
  was None.  Patch by Christian Gibson.  (Github issue #2508)

* Multiplied string literals lost their factor when they are part of
  another constant expression (e.g. 'x' * 10 + 'y' => 'xy').

* String formatting with the '%' operator didn't call the special
  ``__rmod__()`` method if the right side is a string subclass that
  implements it.  (Python issue 28598)

* The directive ``language_level=3`` did not apply to the first token
  in the source file.  (Github issue #2230)

* Overriding cpdef methods did not work in Python subclasses with slots.
  Note that this can have a performance impact on calls from Cython code.
  (Github issue #1771)

* Fix declarations of builtin or C types using strings in pure python mode.
  (Github issue #2046)

* Generator expressions and lambdas failed to compile in ``@cfunc``
  functions.  (Github issue #459)

* Global names with ``const`` types were not excluded from star-import
  assignments which could lead to invalid C code.
  (Github issue #2621)

* Several internal function signatures were fixed that lead to warnings
  in gcc-8.  (Github issue #2363)

* The numpy helper functions ``set_array_base()`` and ``get_array_base()``
  were adapted to the current numpy C-API recommendations.
  Patch by Matti Picus. (Github issue #2528)

* Some NumPy related code was updated to avoid deprecated API usage.
  Original patch by jbrockmendel.  (Github issue #2559)

* Several C++ STL declarations were extended and corrected.
  Patch by Valentin Valls. (Github issue #2207)

* C lines of the module init function were unconditionally not reported in
  exception stack traces.
  Patch by Jeroen Demeyer.  (Github issue #2492)

* When PEP-489 support is enabled, reloading the module overwrote any
  static module state. It now raises an exception instead, given that
  reloading is not actually supported.

* Object-returning, C++ exception throwing functions were not checking that
  the return value was non-null.
  Original patch by Matt Wozniski (Github Issue #2603)

* The source file encoding detection could get confused if the
  ``c_string_encoding`` directive appeared within the first two lines.
  (Github issue #2632)

* Cython generated modules no longer emit a warning during import when
  the size of the NumPy array type is larger than what was found at
  compile.  Instead, this is assumed to be a backwards compatible change
  on NumPy side.

Other changes

* Cython now emits a warning when no ``language_level`` (2, 3 or '3str')
  is set explicitly, neither as a ``cythonize()`` option nor as a compiler
  directive.  This is meant to prepare the transition of the default
  language level from currently Py2 to Py3, since that is what most new
  users will expect these days.  The future default will, however, not
  enforce unicode literals, because this has proven a major obstacle in
  the support for both Python 2.x and 3.x.  The next major release is
  intended to make this change, so that it will parse all code that does
  not request a specific language level as Python 3 code, but with ``str``
  literals.  The language level 2 will continue to be supported for an
  indefinite time.

* The documentation was restructured, cleaned up and examples are now
  tested. The NumPy tutorial was also rewritten to simplify the running
  example. Contributed by Gabriel de Marmiesse.  (Github issue #2245)

* Cython compiles less of its own modules at build time to reduce the
  installed package size to about half of its previous size.  This makes
  the compiler slightly slower, by about 5-7%.
cython-devel mailing list