PyCon US 2022 is underway in Salt Lake City this week, marking the return to in-person conferences after two years of virtual events. As someone who has attended more PyCons than I care to count, there’s something irreplaceable about the hallway track — those impromptu conversations that spark ideas no keynote can match. But beyond the social reunion, this year’s PyCon reflects a Python ecosystem that has matured enormously while somehow maintaining its welcoming character.
Python recently claimed the #1 spot on the TIOBE index, overtaking C for the first time. That’s not just a popularity contest metric — it reflects genuine adoption across data science, web development, DevOps tooling, and increasingly, systems programming adjacent work. The language I first picked up as a “scripting tool” decades ago has become the lingua franca of modern software development.
The Performance Story Is Getting Serious#
One of the most exciting developments in the Python world right now is the Faster CPython project, led by Mark Shannon and funded by Microsoft (with Guido van Rossum himself involved). The project’s goal is ambitious: make CPython 5x faster over several releases.
Python 3.11, expected later this year, already shows impressive gains. Early benchmarks suggest 10-60% speedups on various workloads compared to 3.10. That might not sound revolutionary to someone coming from compiled languages, but for the Python ecosystem, it’s transformative. Every web framework, every data pipeline, every CLI tool gets faster for free — no code changes required.
The approach is pragmatic too. Rather than trying to bolt on a JIT compiler in one massive effort, the team is making incremental improvements to the bytecode interpreter, specializing common operations, and laying groundwork for more aggressive optimizations in future releases. It’s the kind of engineering discipline that gives me confidence this will actually land.
Type Hints: From Optional Nicety to Essential Tool#
When PEP 484 introduced type hints back in Python 3.5, plenty of people in the community dismissed them. “If I wanted types, I’d write Java.” I’ll admit I was skeptical myself. But walking the PyCon halls this year, it’s clear that type hints have won. Not in the sense that everyone uses them — Python remains dynamically typed at runtime — but in the sense that serious projects increasingly treat them as essential.
The tooling ecosystem around types has exploded. mypy is the established player, but pyright (from Microsoft, powering Pylance in VS Code) has become incredibly capable. The experience of writing typed Python in a modern editor with Pylance is genuinely excellent — better than many statically typed languages, because you get the safety net without the ceremony.
What’s particularly interesting is how type hints are enabling new patterns. Libraries like Pydantic use type annotations for runtime data validation, FastAPI builds its entire request/response handling on them, and SQLModel combines them with SQLAlchemy for type-safe database access. Types have become a protocol for libraries to communicate with each other, which was always the real value proposition.
The Packaging Saga Continues#
If there’s one topic that reliably generates heated discussion at PyCon, it’s packaging. The Python packaging ecosystem has been a source of frustration for years — pip, setuptools, wheel, poetry, flit, pdm, hatch — the options are overwhelming and the “right” choice changes depending on who you ask.
This year, there’s cautious optimism. PEP 621 has standardized project metadata in pyproject.toml, which means tools can finally agree on the basics even if they differ in workflow. The new installer and build projects from PyPA are cleaning up the lower layers of the stack.
But honestly? I’ve been hearing “packaging is getting better” at PyCon for a decade. The fundamental challenge is that Python’s packaging story grew organically from a very different era, and the backwards compatibility constraints are enormous. Every improvement has to work with the millions of existing packages on PyPI. It’s an incredibly hard problem, and I respect the people working on it even when I’m cursing at my terminal trying to resolve dependency conflicts.
Python in the Cloud-Native World#
One trend that’s clearly visible at this year’s PyCon is Python’s growing role in cloud-native development. AWS Lambda, Google Cloud Functions, and Azure Functions all have first-class Python support. Tools like Pulumi let you define infrastructure in Python. Even Kubernetes operators are increasingly written in Python using frameworks like kopf.
This matters because it means Python developers don’t have to context-switch to another language for their infrastructure code. Your application code, your tests, your deployment scripts, your monitoring hooks — all Python. Whether that’s a good idea is debatable (I’ve seen some horrifying Python-based infrastructure code), but the option is there and people are using it.
My Take#
PyCon US 2022 feels like a celebration of a language that has found its stride. Python isn’t trying to be everything to everyone anymore — it knows what it’s good at and it’s getting better at those things. The performance improvements are real, the type system is maturing, and the ecosystem breadth is unmatched.
What strikes me most is the community. In an industry that can be tribal and exclusionary, PyCon remains remarkably welcoming. The hallway conversations I’ve had this week range from first-time programmers to CPython core developers, and everyone seems genuinely happy to be here.
If you’re not already writing Python, you probably should be — at least for some part of your toolkit. And if you are, it’s a great time to be in this ecosystem. Now if they could just fix packaging once and for all, we’d be golden.