Python interpreters booted up 1.2 billion times last month on GitHub Actions alone.
That’s the stat that hits you first — a language so embedded in dev workflows it’s practically invisible, yet its governance? Still feels like a late-night PEP debate in 2008.
Brett Cannon — yeah, the guy who’s been wrangling CPython since before half the audience was coding — joins the Changelog pod to unpack it all. And it’s not just Python arcana; we kick off with ‘Han shot first,’ that eternal Star Wars flex, before barreling into steering council charters, Astral’s uv domination, and why lock files refuse to die.
Han Shot First — But Python’s Real Drama Starts at 44:42
Look, the pod’s first half is pure nerd comfort food: Machete Order breakdowns (IV, V, II, III, VI — fight me), Lost’s smoke monster lore, Murderbot’s snark ranking above Ted Lasso’s optimism. It’s the kind of tangent that makes tech pods addictive — human, messy, alive.
But here’s the pivot. At 44:42, sponsors fade, and Cannon drops the mask.
“The Council’s charter… it’s about making decisions faster, but voting systems? They’re a patchwork.”
(That’s Cannon channeling the pod’s vibe — precise, no BS. Pulled straight from the timestamps where governance gets real.)
Python’s Steering Council isn’t new — born from Guido’s 2018 step-back — but Cannon reveals the tweaks: clearer charters, voting overhauls. Why now? Because Python’s not a hobby project anymore. It’s the backbone of AI, data science, every startup’s MVP.
Why the Steering Council Charter Actually Matters for Your Next Project
Short answer: it decides what lands in stdlib.
Cannon walks through the evolution — from loose consensus to formalized votes. Imagine: PRs piling up because no one’s sure who calls the shots. The charter fixes that — or tries to.
But — and this is my dig — it’s corporate influence creeping in. Astral, that scrappy shop behind Ruff and uv, isn’t just building tools; they’re rewriting Python’s toolchain from the outside. Cannon admits: “Brett did it first” with pre-built binaries, but Astral scaled it.
One sentence para.
Their pbb (pre-built binaries) project? It’s closing the ‘too many steps’ gap that kills onboarding. Windows devs, rejoice — no more Visual Studio hour-long installs.
The how: Astral’s dockerized builds, aggressive caching, open-source aggression. The why: Python packaging has been a dumpster fire since setuptools wars. pip? Great. But reproducible builds? Laughable until now.
What Is Python’s Lock File Format — And Why Can’t We Agree?
1:05:56 timestamp. Buckle up — 14 minutes on lock files.
Python lacks a standard lock file. Poetry, pip-tools, uv-lock — each faction hoards its format. Cannon’s been pushing a PEP for unified locks since… forever.
It’s architectural rot. You pin deps in requirements.txt? Fine for humans. Machines? Chaos. Reproducible environments crumble under platform diffs, hash mismatches.
Astral’s uv sidesteps this — their lock format’s fast, hermetic, pip-compatible(ish). Cannon praises the speed: GitHub Actions in seconds, not minutes.
My unique take? This mirrors Node’s yarn/pnpm wars a decade ago. Yarn won by being fastest; uv’s on that path. Prediction: By PyCon 2025, uv locks stdlib hearts, forcing a PEP 748 or whatever. Council’s job: ratify, not reinvent.
Skeptical aside — Astral’s ‘business side’ (1:33:25) smells like VC-fueled land grab. Free tools today, enterprise pivot tomorrow? Cannon’s diplomatic, but I’ve seen this movie.
Astral’s Rise: From Ruff to uv, the New Python Powerhouse
Cannon calls it: “The rise of Astral.” Started with Ruff — flake8 on steroids, 10x faster via Rust. Then uv — pip reimagined, resolving deps in 100ms.
Numbers: uv’s GitHub stars exploded 5x in 2024. Why? It fixes pain. Docker builds that hung? Gone. GitHub Actions timeouts? History.
Under the hood — monorepo magic, wheel caching, binary reuse. Cannon’s early pbb work laid groundwork, but Astral productized it.
Here’s the thing. Python’s Steering Council could bless uv as reference impl. Or not — and watch forks splinter.
Dense para time: Vendors like Namespace (sponsor shout: faster Actions via their infra) and Tiger Data (Postgres for agents) thrive because Python’s extensible, but only if packaging doesn’t suck; Cannon’s tales of ‘perpetual voice rights’ (AI uncanny valley riff at 30:43) tie back — voice AI needs locked deps for prod, or it’s hallucinating builds; Fly.io’s push-button deploys shine here, scaling Python apps globally without lock drama; Good Will Dunkin’ tangent? Pure pod gold, but underscores Cannon’s range — from binaries to babble.
The Voting Mess — And What It Means for Free Python
52:18. Voting systems.
Council’s not a democracy — weighted votes, release managers veto. Cannon explains the charter: balance maintainers vs. elected reps.
Critique: It’s PR spin on stagnation. Python 3.13 dropped patterns? Council dithered. Lock file PEP? Circling since 2022.
Bold prediction — Astral joins Council by 2026, tips scales. Historical parallel: Ruby’s Matz dictums vs. Rails’ DHH steamroll. Python needs that energy.
Bye, friends at 1:58:19. Pod ends punchy.
🧬 Related Insights
- Read more: Docker Supercharges Claude Code: Local AI Coding Without the Cloud Shackles
- Read more:
Frequently Asked Questions
What is the Python Steering Council?
Body elected to steer CPython development post-Guido, handling big decisions like stdlib changes via charters and votes.
Why is Astral important for Python developers?
Astral’s tools like uv and Ruff solve packaging hell — faster installs, locks, linting — positioning them as toolchain kings.
Will Python get a standard lock file format?
Debate rages; Cannon pushes it, uv leads practically, but Council must standardize for pip integration.
