PyPy2.7 and PyPy3.5 v5.8 dual release¶
The PyPy team is proud to release both PyPy2.7 v5.8 (an interpreter supporting Python 2.7 syntax), and a beta-quality PyPy3.5 v5.8 (an interpreter for Python 3.5 syntax). The two releases are both based on much the same codebase, thus the dual release. Note that PyPy3.5 supports Linux 64bit only for now.
This new PyPy2.7 release includes the upstream stdlib version 2.7.13, and PyPy3.5 includes the upstream stdlib version 3.5.3.
We fixed critical bugs in the shadowstack rootfinder garbage collector strategy that crashed multithreaded programs and very rarely showed up even in single threaded programs.
We added native PyPy support to profile frames in the vmprof statistical profiler.
struct module functions
unpack* are now much faster,
especially on raw buffers and bytearrays. Microbenchmarks show a 2x to 10x
speedup. Thanks to Gambit Research for sponsoring this work.
This release adds (but disables by default) link-time optimization and profile guided optimization of the base interpreter, which may make unjitted code run faster. To use these, translate with appropriate options. Be aware of issues with gcc toolchains, though.
Please let us know if your use case is slow, we have ideas how to make things faster but need real-world examples (not micro-benchmarks) of problematic code.
Work sponsored by a Mozilla grant continues on PyPy3.5; numerous fixes from CPython were ported to PyPy and PEP 489 was fully implemented. Of course the bug fixes and performance enhancements mentioned above are part of both PyPy 2.7 and PyPy 3.5.
CFFI, which is part of the PyPy release, has been updated to an unreleased 1.10.1, improving an already great package for interfacing with C.
As always, this release fixed many other issues and bugs raised by the growing community of PyPy users. We strongly recommend updating.
You can download the v5.8 releases here:
We would like to thank our donors for the continued support of the PyPy project.
We would also like to thank our contributors and encourage new people to join the project. PyPy has many layers and we need help with all of them: PyPy and RPython documentation improvements, tweaking popular modules to run on pypy, or general help with making RPython’s JIT even better.
What is PyPy?¶
PyPy is a very compliant Python interpreter, almost a drop-in replacement for CPython 2.7 and CPython 3.5. It’s fast (PyPy and CPython 2.7.x performance comparison) due to its integrated tracing JIT compiler.
We also welcome developers of other dynamic languages to see what RPython can do for them.
The PyPy 2.7 release supports:
- x86 machines on most common operating systems (Linux 32/64 bits, Mac OS X 64 bits, Windows 32 bits, OpenBSD, FreeBSD)
- newer ARM hardware (ARMv6 or ARMv7, with VFPv3) running Linux,
- big- and little-endian variants of PPC64 running Linux,
- s390x running Linux
Highlights of the PyPy2.7, cpyext, and RPython changes (since 5.7 released March, 2017)¶
See also issues that were resolved
Note that these are also merged into PyPy 3.5
- New features and cleanups
- Implement PyModule_New, Py_GetRecursionLimit, Py_SetRecursionLimit,
Py_EnterRecursiveCall, Py_LeaveRecursiveCall, populate tp_descr_get and
add conversions of
__delitem__to appropriate C-API slots
- Fix for multiple inheritance in app-level for C-API defined classes
- Revert a change that removed tp_getattr (Part of the 5.7.1 bugfix release)
- Document more differences with CPython here
- Add native PyPy support to profile frames in vmprof
- Fix an issue with Exception order on failed import
- Fix for a corner case of __future__ imports
- Update packaged Windows zlib, sqlite, expat and OpenSSL to versions used by CPython
- Allow windows builds to use
jom.exefor compiling in parallel
itertools.groupby(), following CPython
- Backport changes from PyPy 3.5 to minimize the code differences
- Improve support for BSD using patches contributed by downstream
- Support profile-guided optimization, enabled with –profopt, , and
specify training data
- Implement PyModule_New, Py_GetRecursionLimit, Py_SetRecursionLimit, Py_EnterRecursiveCall, Py_LeaveRecursiveCall, populate tp_descr_get and tp_descr_set slots, add conversions of
- Bug Fixes
- Correctly handle dict.pop where the popping key is not the same type as the dict’s and pop is called with a default (Part of the 5.7.1 bugfix release)
- Improve our file’s universal newline .readline implementation for
- Tweak issue where ctype array
_basewas set on empty arrays, now it is closer to the implementation in CPython
- Fix critical bugs in shadowstack that crashed multithreaded programs and very rarely showed up even in single threaded programs
- Remove flaky fastpath function call from ctypes
- Support passing a buffersize of 0 to socket.getsockopt
- Avoid hash() returning -1 in cpyext
- Performance improvements:
- Tweaks made to improve performance by reducing the number of guards inserted in jitted code, based on feedback from users
- Add garbage collector memory pressure to some c-level allocations
- Speed up struck.pack, struck.pack_into
- Performance tweaks to round(x, n) for the case n == 0
- Improve zipfile performance by not doing repeated string concatenation
- RPython improvements
- Improve the default shadowstack garbage collector, fixing a crash with multithreaded code and other issues
- Make sure lstrip consumes the entire string
- Support posix_fallocate and posix_fadvise, expose them on PyPy3.5
- Test and fix for int_and() propagating wrong bounds
- Improve the generated machine code by tracking the (constant) value of r11 across intructions. This lets us avoid reloading r11 with another (apparently slowish) “movabs” instruction, replacing it with either nothing or a cheaper variant.
- Performance tweaks in the x86 JIT-generated machine code: rarely taken blocks are moved off-line. Also, the temporary register used to contain large constants is reused across instructions. This helps CPUs branch predictor
- Refactor rpython.rtyper.controllerentry to use use
- Refactor handling of buffers and memoryviews. Memoryviews will now be accepted in a few more places, e.g. in compile()
Highlights of the PyPy3.5 release (since 5.7 beta released March 2017)¶
- New features
- Implement main part of PEP 489 (multi-phase extension module initialization)
- Add docstrings to various modules and functions
- Adapt many CPython bug/feature fixes from CPython 3.5 to PyPy3.5
- Translation succeeds on Mac OS X, unfortunately our buildbot slave cannot be updated to the proper development versions of OpenSSL to properly package a release.
- Implement `` _SSLSocket.server_side``
- Do not silently ignore
_swappedbytes_in ctypes. We now raise a
- Implement and expose
- Bug Fixes
- Fix inconsistencies in the xml.etree.ElementTree.Element class, which on CPython is hidden by the C version from ‘_elementree’.
- OSError(None,None) is different from OSError()
- Get closer to supporting 32 bit windows, translation now succeeds and most lib-python/3/test runs
OrderedDict.__init__behave like CPython wrt. subclasses overridding
- Performance improvements:
- Use “<python> -m test” to run the CPython test suite, as documented by CPython, instead of our outdated regrverbose.py script
- Change _cffi_src/openssl/callbacks.py to stop relying on the CPython C API.
- Avoid importing the full locale module during _io initialization, CPython change fbbf8b160e8d
- Avoid freezing many app-level modules at translation, avoid importing many modules at startup
- Refactor buffers, which allows an optimization for
- The following features of Python 3.5 are not implemented yet in PyPy:
- PEP 442: Safe object finalization
Please update, and continue to help us make PyPy better.