imported>ICHx at 03:03, 3 June 2025

2025-06-03T03:03:49Z

New page

{{Short description|Alternative implementation of the Python programming language}}
{{Distinguish|text=[[PyPI]], the Python official third-party software repository}}
{{Infobox software
| title =
| name = PyPy
| logo = Pypy-logo (2020).svg
| screenshot =
| caption =
| collapsible =
| author =
| developer =
| released = mid {{Start date and age|2007}}
| latest release version = {{LSR/wikidata}}
| latest preview version =
| latest preview date =
| programming language = RPython
| operating system = [[Cross-platform]]
| platform =
| size =
| language =
| genre = [[Python (programming language)|Python]] interpreter and compiler [[toolchain]]
| license = [[MIT License|MIT]]
}}

'''PyPy''' ({{IPAc-en|'|p|aɪ|p|aɪ}}) is an implementation of the [[Python (programming language)|Python]] [[programming language]].<ref>{{cite web|url=https://decisionstats.com/2015/11/29/interview-maciej-fijalkowski-pypy/|title=Interview Maciej Fijalkowski PyPy|date=29 November 2015 }}</ref> PyPy often runs faster than the standard [[implementation]] [[CPython]] because PyPy uses a [[just-in-time compilation|just-in-time compiler]].<ref>{{Cite web|url=https://speed.pypy.org/|title=PyPy Speed|website=speed.pypy.org|access-date=2019-12-01}}</ref> Most Python code runs well on PyPy except for code that depends on CPython extensions, which either does not work or incurs some overhead when run in PyPy.

PyPy itself is built using a technique known as meta-tracing, which is a mostly automatic transformation that takes an [[Interpreter (computing)|interpreter]] as input and produces a [[tracing just-in-time compilation|tracing just-in-time compiler]] as output. Since interpreters are usually easier to write than [[compiler]]s, but run slower, this technique can make it easier to produce efficient implementations of programming languages. PyPy's meta-tracing [[toolchain]] is called [[#RPython|RPython]].

PyPy officially supports Python 2.7 and 3.11<ref>{{cite web|url=https://pypy.org/index.html|title=PyPy|date=30 December 2024}}</ref> and has a few differences in implementations compared to CPython.<ref>{{cite web|url=https://doc.pypy.org/en/latest/cpython_differences.html|date=30 December 2024 |title=Differences between PyPy and CPython — PyPy documentation }}</ref>

==Details and motivation==
PyPy aims to provide a common translation and support framework for producing implementations of [[dynamic language]]s, emphasizing a clean separation between [[Programming language specification|language specification]] and implementation aspects. It also aims to provide a compliant, flexible and fast implementation of the Python programming language using the above [[Software framework|framework]] to enable new advanced features without having to encode low-level details into it.<ref name=mission-statement>{{ cite web | author = Samuele Pedroni | date = March 2007 | title = PyPy – Goals and Architecture Overview | url = http://codespeak.net/pypy/dist/pypy/doc/architecture.html#mission-statement | archive-url = https://web.archive.org/web/20120614024217/http://doc.pypy.org/en/latest/architecture.html#mission-statement | url-status = dead | archive-date = 2012-06-14 }}</ref><ref name=mission-statement-2>{{ cite web | access-date=11 October 2013 | title = PyPy – Goals and Architecture Overview – Mission Statement | url = http://doc.pypy.org/en/latest/architecture.html#mission-statement }}</ref>

===RPython===
The PyPy interpreter itself is written in a restricted subset of Python called RPython (Restricted Python).<ref>[https://doc.pypy.org/en/latest/coding-guide.html#our-runtime-interpreter-is-rpython Our runtime interpreter is "RPython"], Coding Guide – PyPy documentation</ref> RPython puts some constraints on the Python language such that a variable's type can be [[type inference|inferred]] at compile time.<ref>"It is a proper subset of Python, restricted in a way that enables easy analysis and efficient code generation", Ancona et al., 2007.</ref>

The PyPy project has developed a [[toolchain]] that analyzes RPython code and translates it into a form of [[Bytecode|byte code]], which can be lowered into [[C (programming language)|C]]. There used to be other [[Frontend and backend|backends]] in addition to [[C (programming language)|C]] ([[Java (programming language)|Java]], [[C Sharp (programming language)|C#]], and [[JavaScript]]), but those suffered from [[Software decay|bitrot]] and have been removed. Thus, the [[Recursion|recursive]] logo of PyPy is a [[Ouroboros|snake swallowing itself]] since the RPython is translated by a Python [[Interpreter (computing)|interpreter]]. The code can also be run untranslated for testing and analysis, which provides a nice [[Testbed|test-bed]] for research into dynamic languages.

It allows for pluggable [[Garbage collection (computer science)|garbage collectors]], as well as optionally enabling [[Stackless Python]] features. Finally, it includes a just-in-time (JIT) [[Generator (computer programming)|generator]] that builds a [[Just-in-time compilation|just-in-time compiler]] into the interpreter, given a few annotations in the [[Interpreter (computing)|interpreter]] [[source code]]. The generated JIT compiler is a [[tracing just-in-time compilation|tracing JIT]].<ref>{{cite conference |title=Tracing the Meta-Level: PyPy's Tracing JIT Compiler |doi=10.1145/1565824.1565827 |last1=Bolz |first1=Carl |last2=Cuni |first2=Antonio |last3=Fijalkowski |first3=Maciej |last4=Rigo |first4=Armin |conference=ICOOOLPS '09}}</ref>

RPython is now also used to write non-Python language implementations, such as Pixie.<ref>[https://medium.com/this-is-not-a-monad-tutorial/indie-languages-interview-pixie-and-timothy-baldridge-cadbc36418dc Timothy Balridge interview].</ref>

==Project status==
PyPy as of version 7.3.17 is compatible with two [[CPython]] versions: 2.7 and 3.10.<ref name=pypycompat>{{cite web |url=http://pypy.org/compat.html |title=PyPy – Python compatibility |access-date=2020-12-15 |website=pypy.org|date=28 December 2019 }}</ref><ref name=pypycompat36>{{cite web |url=https://www.pypy.org/posts/2021/10/pypy-v737-release.html |title=PyPy v7.3.7: bug-fix release of 3.7, 3.8 |access-date=2021-11-10 |website=pypy.org|date=25 October 2021 }}</ref> The first PyPy version compatible with CPython v3 is PyPy v2.3.1 (2014).<ref name=pypyblogpypy3>{{cite news |date=20 June 2014 |author=the PyPy team |title=PyPy3 2.3.1 – Fulcrum |publisher=PyPy blog |url=http://morepypy.blogspot.it/2014/06/pypy3-231-fulcrum.html }}</ref> The PyPy interpreter compatible with CPython v3 is also known as PyPy3.

PyPy has [[just-in-time compilation|JIT compilation]] support on 32-bit/64-bit [[x86]] and 32-bit/64-bit [[ARM architecture family|ARM]] processors.<ref name="whatsnew720">{{cite news |date=16 October 2019 |title=PyPy v7.2.0: release of 2.7, and 3.6 |publisher=[[pypy.org]] |url=https://pypy.readthedocs.io/en/latest/release-v7.2.0.html }}</ref> It is tested nightly on Windows, Linux, OpenBSD and Mac OS X. PyPy is able to run pure Python software that does not rely on implementation-specific features.<ref>{{Cite web|title = PyPy – Python compatibility | date = 28 December 2019| url = http://pypy.org/compat.html}}</ref>

There is a compatibility layer for CPython C API extensions called CPyExt, but it is incomplete and experimental. The preferred way of interfacing with [[C (programming language)|C]] [[shared libraries]] is through the built-in C [[foreign function interface]] (CFFI) or [[ctypes (language binding)|ctypes]] libraries.

==History==
PyPy is a followup to the [[Psyco]] project, a [[Just-in-time compilation|just-in-time]] [[run-time algorithm specialization|specializing compiler]] for Python, developed by Armin Rigo between 2002 and 2010. PyPy's aim is to have a just-in-time specializing compiler with scope, which was not available for Psyco.{{clarify|reason=what does "with scope" mean?|date=May 2013}} Initially, the RPython could also be compiled into [[Java bytecode]], [[Common Intermediate Language|CIL]] and [[JavaScript]], but these backends were removed due to lack of interest.

PyPy was initially a research and development-oriented project. Reaching a mature state of development and an official 1.0 release in mid-2007, its next focus was on releasing a production-ready version with more CPython compatibility. Many of PyPy's changes have been made during [[coding sprint]]s.

* In August 2008, PyPy was able to run some popular Python libraries like [[Pylons project#Pylons Framework|Pylons]],<ref>{{Cite web|title = Running pylons on top of PyPy | date = 10 June 2008| url = http://morepypy.blogspot.com/2008/06/running-pylons-on-top-of-pypy.html}}</ref> [[Pyglet]],<ref>{{Cite web|title = Running Pyglet on top of PyPy | date = 20 February 2008| url = http://morepypy.blogspot.com/2008/02/running-pyglet-on-pypy.html}}</ref> [[Nevow]]<ref>{{Cite web| title=Running Nevow on top of PyPy| date = 20 June 2008| url = http://morepypy.blogspot.com/2008/06/running-nevow-on-top-of-pypy.html}}</ref> and [[Django (web framework)|Django]].<ref>{{Cite web| title=PyPy runs unmodified django 1.0 beta | date = 19 August 2008| url = http://morepypy.blogspot.com/2008/08/pypy-runs-unmodified-django-10-beta.html}}</ref>
* On 12 March 2010, PyPy 1.2 was released, focusing on speed. It included a working, though not yet stable, just-in-time compiler.<ref>{{Cite web| title=Introducing the PyPy 1.2 release | date =30 April 2011| url =https://morepypy.blogspot.dk/2011/04/pypy-15-released-catching-up.html}}</ref>
* On 30 April 2011, PyPy version 1.5 was released, which reached compatibility with CPython 2.7.<ref>{{Cite web| title= PyPy 1.5 Released: Catching Up | date =12 March 2010| url =https://morepypy.blogspot.dk/2010/03/introducing-pypy-12-release.html}}</ref>
* On 9 May 2013, PyPy 2.0 was released, which introduced alpha-quality support for JIT compilation on ARMv6 and ARMv7 JIT, and included CFFI in the standard library.<ref name=lwn2013>{{cite news |date=15 May 2013 |author=Jake Edge |title=A look at the PyPy 2.0 release |publisher=[[LWN.net]] |url=https://lwn.net/Articles/550427/ }}</ref><ref>{{Cite web| title= PyPy 2.0 – Einstein Sandwich | date = 9 May 2013| url = https://morepypy.blogspot.dk/2013/05/pypy-20-einstein-sandwich.html}}</ref>
* On 20 June 2014, PyPy3 was declared stable<ref name=pypyblogpypy3 /> and introduced compatibility with the more modern Python 3. It was released alongside PyPy 2.3.1 and bears the same version number.
* On 21 March 2017, the PyPy project released version 5.7 of both PyPy and PyPy3, with the latter introducing beta-quality support for Python 3.5.<ref>{{Cite web| title= PyPy2.7 and PyPy3.5 v5.7 – two in one release | date = 21 March 2017| url = https://morepypy.blogspot.dk/2017/03/pypy27-and-pypy35-v57-two-in-one-release.html}}</ref>
* On 26 April 2018, version 6.0 was released, with support for Python 2.7 and 3.5 (still beta-quality on Windows).<ref>{{Cite web| title= PyPy2.7 and PyPy3.5 v6.0 dual release | date = 26 April 2018| url = https://morepypy.blogspot.de/2018/04/pypy27-and-pypy35-v60-dual-release.html}}</ref>
* On 11 February 2019, version 7.0 was released, with support for Python 2.7 and 3.5.<ref>{{Cite web|last=Cuni|first=Antonio|date=2019-02-11|title=PyPy Status Blog: PyPy v7.0.0: triple release of 2.7, 3.5 and 3.6-alpha|url=https://morepypy.blogspot.com/2019/02/pypy-v700-triple-release-of-27-35-and.html|access-date=2020-08-17|website=PyPy Status Blog}}</ref>
* On 14 October 2019, version 7.2 was released, with support for Python 3.6.9.<ref>{{Cite web|last=Mattip|date=2019-10-14|title=PyPy Status Blog: PyPy v7.2 released|url=https://morepypy.blogspot.com/2019/10/pypy-v72-released.html|access-date=2020-08-17|website=PyPy Status Blog}}</ref>
* On 24 December 2019, version 7.3 was released, with support for Python 3.6.9.<ref>{{Cite web|last=Mattip|date=2019-12-24|title=PyPy Status Blog: PyPy v7.3.0 released|url=https://morepypy.blogspot.com/2019/12/pypy-730-released.html|website=PyPy Status Blog}}</ref>
* On 16 February 2020, the PyPy team announced the move of the source code hosting from [[Bitbucket]] to heptapod.net with the repositories of the CFFI (C [[Foreign Function Interface]]) project. A new logo and website design are also published. However, the author and the license of the new logo are unknown.<ref>{{Cite web|url=https://www.pypy.org/posts/2020/02/pypy-and-cffi-have-moved-to-heptapod-5791595152472747032.html|title=PyPy and CFFI have moved to Heptapod|date=16 February 2020}}</ref>
* On 29 December 2023, PyPy announced hosting has moved to [[GitHub]] and development will now be tracked with [[git]].<ref>{{Cite web|url=https://www.pypy.org/posts/2023/12/pypy-moved-to-git-github.html|title=PyPy has moved to Git, GitHub|date=29 December 2023}}</ref>

===Funding===
PyPy was funded by the [[European Union]] being a [[Specific Targeted Research Project]]<ref>{{Cite web|title = EU Community Research and Development Information Service Entry | url = http://cordis.europa.eu/projects/rcn/74481_en.html }}</ref> between December 2004 and March 2007. In June 2008, PyPy announced funding being part of the [[Google]] Open Source programs and has agreed to focus on making PyPy more compatible with [[CPython]]. In 2009 Eurostars, a European Union funding agency specially focused on [[Small and medium enterprises|SMEs]],<ref>{{Cite web| title = Eurostars – Aim Higher | url = http://www.eurostars-eureka.eu/}}</ref> accepted a proposal from PyPy project members titled "PYJIT – a fast and flexible toolkit for dynamic programming languages based on PyPy". Eurostars funding lasted until August 2011.<ref>{{Cite web | title = Project Page on Eureka Network | url = http://www.eurekanetwork.org/project/-/id/4791 | access-date = 2011-10-17 | archive-date = 2012-04-03 | archive-url = https://web.archive.org/web/20120403075231/http://www.eurekanetwork.org/project/-/id/4791 | url-status = dead }}</ref>
At [[PyCon]] US 2011, the [[Python Software Foundation]] provided a $10,000 grant for PyPy to continue work on performance and compatibility with newer versions of the language.<ref>{{Cite web|title=A thank you to the PSF|date=22 March 2011|url=http://morepypy.blogspot.com/2011/03/thank-you-to-psf.html}}</ref>
The port to ARM architecture was sponsored in part by the [[Raspberry Pi Foundation]].<ref name=lwn2013 />

The PyPy project also accepts donations through its status blog pages.<ref>{{Cite web| title = PyPy Status Blog: Oh, and btw: PyPy gets funding through "Eurostars"| date = 10 December 2010| url = http://morepypy.blogspot.com/2010/12/oh-and-btw-pypy-gets-funding-through.html}}</ref> As of 2013, a variety of sub-projects had funding: Python 3 version compatibility, built-in optimized [[NumPy]] support for numerical calculations and [[software transactional memory]] support to allow better parallelism.<ref name=lwn2013 />

==See also==
{{Portal
|Free software
|Computer programming
}}
* [[Bootstrapping (compilers)]]
* [[Cython]]
* [[GraalVM]]
* [[Partial evaluation]]
* [[Psyco]]
* [[Self-hosting (compilers)|Self-hosting]]
* [[Self-interpreter]]
* [[Unladen Swallow]]

==Notes==
{{Reflist|30em}}

==References==
{{refbegin}}
* Davide Ancona, Massimo Ancona, Antonio Cuni, Nicholas D. Matsakis, 2007. [https://web.archive.org/web/20170706104906/ftp://ftp.disi.unige.it/pub/person/AnconaD/DLS08.pdf RPython: a Step Towards Reconciling Dynamically and Statically Typed OO Languages]. In Proc. Dynamic Language Symposium (DLS), 2007. ACM Press.
* Carl Friedrich Bolz, Antonio Cuni, Maciej Fijalkowski, 2009. [https://web.archive.org/web/20110716123213/http://codespeak.net/svn/pypy/extradoc/talk/icooolps2009/bolz-tracing-jit-final.pdf Tracing the meta-level: PyPy's Tracing JIT Compiler]. In Proc. ICOOOLPS, 2009. ACM Press.
* {{cite web | last = Corbet | first = Jonathan | title = A brief experiment with PyPy | publisher = [[LWN.net]] | date = 11 May 2011 | url = https://lwn.net/Articles/442268/ }}
* {{cite news | last = von Eitzen | first = Chris | title = PyPy 1.7 widens the performance "sweet spot" | newspaper = [[The H]] | publisher = [[Heinz Heise]] | date = 21 November 2011 | url = http://www.h-online.com/open/news/item/PyPy-1-7-widens-the-performance-sweet-spot-1382249.html }}
* {{cite web | last = Rose | first = John | title = A Day with PyPy | publisher = [[Oracle Corporation|Oracle]] developer blog | date = 2 December 2011 | url = http://blogs.oracle.com/jrose/entry/a_day_with_pypy }}
* {{cite web | title = Interview Maciej Fijalkowski pypy | date = 29 November 2015 | publisher = Decisionstats blog | url = http://decisionstats.com/2015/11/29/interview-maciej-fijalkowski-pypy/ }}
{{refend}}

==External links==
* {{Official website}}

{{Python (programming language)}}

[[Category:2007 software]]
[[Category:FP6 projects]]
[[Category:Free software programmed in Python]]
[[Category:Python (programming language) implementations]]
[[Category:Python (programming language) software|*]]
[[Category:Software using the MIT license]]
[[Category:Stack-based virtual machines]]

PyPy - Revision history

imported>ICHx at 03:03, 3 June 2025