Python Weekly (Issue 771 November 12 2026)

Author

Rahul Chaudhary
November 12, 2026

Welcome to issue 771 of Python Weekly. Let's get straight to the links this week.

Analytics on Live Data Without Leaving Postgres

When analytics on Postgres slows down, most teams add a second database. Then they manage pipelines, sync lag, and drift forever. TimescaleDB extends Postgres instead. Analytics run on live data, in the database you already have. No pipeline. No migration. No new query language.


News

Guido van Rossum pays tribute to Tim Peters, the legendary developer behind Timsort and The Zen of Python, who passed away in July 2026. Fellow Python contributors share memories of his extraordinary technical contributions, mentorship, humor, and lasting influence on the language.

PyPI has rolled out redesigned project pages with a new Security tab displaying package provenance, digital attestations, and warnings about suspicious changes in publishing workflows. The update also improves release file navigation, release history, and visibility into package security information.


Articles, Tutorials and Talks

Simon Willison used Claude Code to build a Python-powered system that triangulates sounds using microphones on phones, tablets, and laptops. Combining WebSockets, browser audio APIs, and signal processing, it tracked a mysterious beep to within 40cm and uncovered an old carbon monoxide detector.

AI agents translated parts of CPython 3.15 into 613,000 lines of Lean code, using formal proofs, differential testing, and independent reviews to verify interpreter behavior. The project reports 418 potential bugs, including roughly 55 cases involving crashes or memory corruption, while highlighting the challenges of making AI-generated proofs trustworthy.

So... Everyone and Their Agent Gets a Postgres Branch? (Even Your Worktrees)

This video demonstrates how to use Xata and ClickHouse to create isolated, anonymized Postgres database branches for AI agents to prevent production data leaks and accidental modifications. The process includes building a secure development workflow using Django, setting up real-time analytics, and managing scalable, ephemeral database environments for testing and migration experiments.

A three-part series explores the evolution of Python configuration management, why existing libraries fall short (Part 2), and introduces dature, a type-safe configuration library built around standard Python dataclasses (Part 3).

Explore Python's asyncio internals by tracing coroutine execution with pdb, from event loop initialization and task creation to scheduling, suspension, and resumption. Detailed debugger output and bytecode analysis reveal how await, futures, ready queues, and timers coordinate asynchronous execution.

Adam Johnson introduces django-launch-editor, a Django package that makes traceback file locations clickable, opening the relevant source code directly in your editor at the correct line and column. It supports multiple editors, requires minimal configuration, and adds no overhead in production.

Meta FAIR's e-means algorithm accelerates k-means clustering by 8–15x on CPUs and 10–45x on GPUs while achieving lower clustering error. Available through Faiss with Python support, it combines mini-batch updates, exponential moving averages, and automatic hyperparameter tuning.

Posit's new pymizu library uses shared memory to accelerate Python multiprocessing, reducing task round-trip overhead from 94.2 to 0.7 microseconds in benchmarks. It supports zero-copy data exchange between Python and R, integrates with concurrent.futures and asyncio, and enables efficient parallel processing.

Twisted creator Glyph Lefkowitz reflects on the framework's relevance nearly 30 years after its creation, despite Python's adoption of asyncio. He makes the case for understandable, self-hosted internet services built with integrated networking protocols, deterministic testing, and minimal infrastructure.

Building a secure sandbox for AI coding agents requires more than container isolation. Muhammad Raza implements six security layers in Python, tests them against sandbox escapes and other attacks, and examines why production environments often rely on microVMs.

This video identifies seven common mistakes developers make when using AI coding agents, such as relying on outdated patterns and failing to provide clear requirements. The host explains how to effectively guide these tools by implementing constraints, breaking work into smaller tasks, and maintaining responsibility for software quality.

This video introduces Pandera as a powerful tool for enforcing schema validation and ensuring data quality in Polars and Pandas workflows. It demonstrates how to implement both object-based and class-based schemas to catch data errors early in your processing pipelines.

Adopting Python 3.15's lazy imports: a 4× faster CLI, and where they didn't help

Adopting Python 3.15's lazy imports cut CLI startup time by 4x, but delivered much smaller improvements for a Textual TUI and MCP server. A detailed account of import profiling, safely deferring dependencies, avoiding subtle testing pitfalls, and measuring real-world performance gains.


Interesting Projects, Tools, and Libraries

PLSI is a Python-based multi-agent longevity research toolkit for Grok that processes health records, tracks biomarkers and aging metrics, evaluates scientific evidence, and generates personalized health recommendations with built-in safety checks.

A honeypot for supermarket refrigeration controllers.

Configure and automate Home Assistant with Claude, ChatGPT, and other AI assistants.

The High-Performance, Django-Native Application Server.

llama-modchip optimizes llama.cpp with custom quantization, CPU kernels, and speculative decoding to run large mixture-of-experts models locally without GPUs, including a 284B-parameter DeepSeek model on 64 GB of RAM.

Fast, async, fully-typed distributed task queue via Redis streams.

Minimal, provider-agnostic Python client for large language models, built on niquests.

PyLings brings Rustlings-style learning to Python with 55 hands-on coding exercises, automatic test feedback, hints, and solutions, covering everything from Python fundamentals to senior and staff-level interview challenges.

Open-source inference engine that runs language, vision, speech, image and video models from one container format, with a PyTorch and a Triton compute branch. Released for NVIDIA GPUs and CPU; Apple Silicon in progress.

Apache Arrow compute on the Apple silicon GPU, inside DataFusion, Polars and DuckDB.


Upcoming Events and Webinars

There will be a talk, From API to Insight: Building an End-to-End Data Pipeline with Python.

There will be following talk

  • Transformation Bias in Machine Learning

  • Beyond the Prompt: The Frontier of AI Coding

  • Agentic Validatation

There will be following talk

  • From Data & Pixels to Risk Context: Agentic Earth Observation for Explainable Climate Intelligence

  • Vectorized statistical distributions, built on the Polars engine


Our Other Newsletters

Programmer Weekly - A free weekly newsletter for programmers.

Founder Weekly - A free weekly newsletter for entrepreneurs featuring best curated content, must read articles, how to guides, tips and tricks, resources, events and more.