Mastodawn

Siddhesh Poyarekar Mar 15

Hynek Schlawack

This is such an end of an era that fresh Python programmers can’t even fathom.

PyPy used to be our hope! No major Python conference that didn’t suggest that they’re gonna fix the GIL and make time go backwards. And yeah, it’s really fast! I suspect the money-backed focus on performance in CPython combined with the compat paper cuts PyPy always came with has sealed its fate. I‘ve watched its decline over the years so I’m not surprised, but damn.

https://github.com/numpy/numpy/issues/30416

Seth Larson Mar 8

@hynek I'm seeing mixed messages around the internet about the status of PyPy? I guess we'll wait for more clarity about what's happening.

Hynek Schlawack Mar 8

@sethmlarson Yeah but it definitely stands that their days of glory are over. Their latest supported version will be EOL in half a year

Hugo van Kemenade Mar 15

@hynek @sethmlarson There's a year and a half left for 3.11.
https://pypy.org/posts/2026/03/pypy-v7320-release.html

https://devguide.python.org/versions/

Although some scientific projects such as NumPy are dropping 3.11 around now:
https://numpy.org/neps/nep-0029-deprecation_policy.html

PyPy v7.3.21 release

PyPy v7.3.21: release of python 2.7, 3.11 The PyPy team is proud to release version 7.3.21 of PyPy after the previous release on July 4, 2025. This is a bug-fix release that also updates to Python 3.1

PyPy

@hynek wow. filed by mattip, even. dang. end of an era. sad but certainly understandable

Hynek Schlawack Mar 8

@glyph Yeah that’s the real burner.

Gina Peter Banyard Mar 8

@hynek I don't much about it, but this seems to be a similar story to PHP and Hacklang? In that when the "reference" implementation stops "sucking" (lack of a better word rn) people go back to it because the ecosystem is larger and less constrained.

Hynek Schlawack Mar 8

@Girgias the reasons might be the similar, but it’s probably different dynamics given that Hack is a Meta project they presumably still use? While they definitely have their users, my impression is that it was never even close to maintstream & mostly about the promises of an even brighter future b/c it was run by some of the smartest ppl alive. One of my memories is also that it definitely was an impediment to Python 3 adoption because they were 2-only for a long time so that played a role too

Gina Peter Banyard Mar 8

@hynek as far as I know Meta and Slack still use Hack yeah (although I've heard rumours that Meta is looking at moving some smaller projects back to PHP). It being an impediment to Python 3 sucks :/ Because yeah Hack was the thing that pushed PHP forward whereas what you're saying is that this might have been the case initially for PyPy it ended up being a hindrance?

Hynek Schlawack Mar 8

@Girgias yeah it definitely was. unlike Hack, it’s ambition wasn’t to improve Python – that was Python 3. Its ambition has always been to be a much faster drop-in replacement for CPython. At the same time they tackled adjacent performance and concurrency-related problems with mixed results. But their JIT is amazing and there’s a lot of legacy they’ll leave – including pytest.

Eric Matthes Mar 9

@hynek @Girgias I had no idea pytest came out of PyPy, and I also had no idea how long PyPy has been around!

@hynek PyPy's RPython powered the tech behind my PhD. @cfbolz Ideas and encouragement the concepts.

I'm sad now :(

Matěj Cepl 🇪🇺 🇨🇿 🇺🇦Mar 15

I am one of the leading maintainers of all #Python stuff for #openSUSE . One of the nightmares on a back burner is that out PyPy (https://src.opensuse.org/python-interpreters/pypy3) has been broken for years (actually it looks 10 years; patches are obviously very welcome!), and I have been just yesterday thinking, whether I shouldn't fix those packages with AI and make it work again. On the other hand, throughout those ten years, I haven’t got ONE complaint that it has been broken. And our users LOVE to complain!

pypy3

pypy3

openSUSE Gitea

Jesus Michał "Le Sigh" 🏔 (he)Mar 15

@hynek, for me it was always a fun project that tested whether packages aren't relying too deeply on CPython internals. It also helped to find a fair number of bugs in pure Python fallbacks of CPython stdlib that aren't normally used at all.

mirabilos Mar 15

@mgorny @hynek yeah damn, and with CPython being renamed to RustPython and ensloppified…

MicroPython perhaps?

@mirabilos @mgorny @hynek What am I missing? The last PyPy release seems to be 2 days old https://pypy.org/posts/2026/03/pypy-v7320-release.html

PyPy v7.3.21 release

PyPy v7.3.21: release of python 2.7, 3.11 The PyPy team is proud to release version 7.3.21 of PyPy after the previous release on July 4, 2025. This is a bug-fix release that also updates to Python 3.1

PyPy

mirabilos Mar 15

@kabel42 @mgorny @hynek PyPy implementing Python 3.11 only, which scientific-python extensions were actively recommended to drop last quarter (one early, their recommendation generator skips the last partial quarter of support).

byte::Arc<Eepy>

neobot_sign_beep

blobhaj_flag_nonbinary

@hynek rest in piss, i’ve wasted so much time dealing with PyPy back in uni