Python Weekly (Issue 719 October 23 2025)

Author

Rahul Chaudhary
October 23, 2025

In partnership with

Free email without sacrificing your privacy

Gmail is free, but you pay with your data. Proton Mail is different.

We don’t scan your messages. We don’t sell your behavior. We don’t follow you across the internet.

Proton Mail gives you full-featured, private email without surveillance or creepy profiling. It’s email that respects your time, your attention, and your boundaries.

Email doesn’t have to cost your privacy.


Articles, Tutorials and Talks

Learn how the Registry Pattern cleans up long if-elif chains by centralizing function registration in a dictionary. This design makes the code more extensible, decoupled, and easier to maintain, demonstrated with a Python CLI example using decorators and dynamic command loading.

This post tests Python 3.15’s proposed PEP 810 explicit lazy imports, which delay loading modules until first use to cut startup time.? Using the feature on author's CLI tool pypistats, he found it ran 2.92× faster (reducing startup from 104 ms to 36 ms), demonstrating how lazy imports can significantly speed up Python applications with large dependency graphs.

This series is a comprehensive workshop designed to teach SQLAlchemy from fundamental concepts through advanced data modeling and complex query crafting. By the end, it aims to equip learners with practical skills to confidently handle data challenges and build scalable, powerful applications using SQLAlchemy’s ORM and session features.

Graham Dumpleton explores how Python’s wrapt library must correctly handle immutable types when wrapping objects, highlighting hidden issues with in-place operators like += and @= when applied to proxies of immutables. He outlines a refined iadd strategy for ObjectProxy that preserves expected behaviour in version 2.0.0 of wrapt.

Recursive Language Models (RLMs) let language models recursively call themselves within an environment, like a Python REPL, to handle extremely long contexts without performance drop (context rot). They dynamically break down queries into smaller parts, delivering strong, cost-efficient results on big benchmarks and enabling scalable, interpretable reasoning beyond fixed context limits.

The free-threaded Python variant in 3.14 removes the Global Interpreter Lock (GIL), enabling true parallel multithreading for CPU-bound tasks. While it may have a modest performance cost on single-threaded code, it significantly improves memory efficiency and concurrency in web applications, simplifying deployment and boosting throughput, especially for ASGI and WSGI based services.​

A tutorial on building a computer use AI agent capable of executing multi-step tasks in a Bash shell, powered by the NVIDIA Nemotron Large Language Model. It covers creating the agent's brain, the Bash interface for safe command execution, and the agent loop, demonstrating how to build and deploy an autonomous assistant within an hour.

PyTorch Monarch is a distributed programming framework designed to simplify scaling AI workflows by enabling a single-controller model that orchestrates distributed resources like a single machine. It provides actor-based programming with scalable messaging, fault tolerance, and distributed tensor support, allowing seamless development, debugging, and efficient handling of large-scale training and reinforcement learning workloads across GPU clusters.


Interesting Projects, Tools, and Libraries

Full stack Python web framework to build websites and web apps with as little boilerplate as possible

Run any Python function on 1000 computers in 1 second.

An autonomous agent for deep financial research

A GitHub Action to report changes to uv.lock.

Know before you scrape. Analyze any website's anti-bot protections in seconds.

Blinter is a linter for Windows batch files. It provides comprehensive static analysis to identify syntax errors, security vulnerabilities, performance issues and style problems.

Transforms complex documents like PDFs into LLM-ready markdown/JSON for your Agentic workflows.

Open Source payments infrastructure for the 21st century

Self-hosted secure VM sandboxes for AI compute at scale.

Rust-powered API framework for Django achieving 60k+ RPS. Uses Actix Web for HTTP, PyO3 for Python bridging, msgspec for serialization. Decorator-based routing with built-in auth and middleware.

Override and customize Python packages without touching their code.

This is an in-progress project aiming at a Pythonic translation of TypeAgent KnowPro and a few related packages from TypeScript to Python.


New Releases

Django 6.0 beta 1, released on October 22, 2025, marks the second stage in the Django 6.0 release cycle, offering an opportunity to test upcoming changes with a focus on modern tools and thoughtful design. The final release is scheduled for December 3, 2025, and this beta is not for production use but encourages community testing and bug reporting to ensure stability.


Upcoming Events and Webinars

There will be following talks,

  • From RAG to Agentic System: lessons learnt from the transition

  • Chat with History – Bringing an archive to life using AI

There will be a talk, Getting Hired: From resume writing to accepting the offer.

There will be a talk, Theoretical AI Development with Ontology.

There will be a talk, Tabular Attention Models.

There will be a talk, Small and Smart AI Models.<br>

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.