TheNote
Planet Python Note
GooglePlay AppStore

Planet Python

The website Planet Python is a planet site that aggregates Python-related content from various sources, including blogs, news sites, and other online publications. The website provides a one-stop destination for individuals to stay up-to-date with the latest developments in the Python programming world. The content on the site includes tutorials, news, project announcements, and discussions about various Python-related topics. Users can visit the site to stay informed about the Python community, new releases, conferences, and best practices in using the Python programming language. The website's purpose is to help promote and disseminate Python-related content, thereby contributing to the growth and development of the Python community.
Planet Python planetpython.org
RSS planetpython.org
RSS Hunter RSS Hunter • Aug 19, 2024

Thread Of Notes

Go Deh: Finding Flaws in Technical Documentation

The author developed a novel approach to documentation testing using AI. After updating project tutorials and documentation in Markdown files, they considered using AI to test their effectiveness. They instructed ChatGPT to create an isolated chat environment where it would read the documentation and attempt to use the described tool. The AI agreed, generating a detailed prompt for the new chat and monitoring its progress.The AI then reported back on documentation deficiencies it found. For instance, it noted the documentation lacked an explicit statement about the tool's intended Linux use, despite providing setup details for Linux and WSL. This led the AI to try running the tool in a Windows environment, highlighting an implicit assumption by the author that needed clarification.The author appreciates this method of having AI act as a new user, actively trying to complete tasks and reporting encountered issues. This goes beyond mere wording review, providing a practical test of the documentation's followability. While the AI may present invalid points, human judgment remains crucial in deciding what changes to implement.The author emphasizes that documentation must serve both human readers and AI assistants. With a growing number of users likely to employ AI for project interaction, documentation needs to be clear enough for direct human use and for AI to effectively assist users. This means considering how well documentation supports AI interaction, not just writing exclusively for AI.Ultimately, the author proposes testing documentation similarly to code, by assessing whether a user, human or AI, can successfully follow it to achieve a task. This innovative testing method offers a new perspective for improving documentation effectiveness on future projects.
http://paddy3118.blogspot.com/2026/10/finding-flaws-in-technical-documentation.html paddy3118.blogspot.com
CdXz5zHNQW_Tocx7RtYip.png
RSS Hunter RSS Hunter • Yesterday

LernerPython blog, from Reuven Lerner: Pandas offset aliases for resample: ME, MS, and other codes

Using resample in Python Pandas? You’ll want a list of “offset aliases,” the codes that let you specify units.Some are surprising, such as “ME” and “MS” for “month end” and “month start”.They’re at: https://pandas.pydata.org/docs/user_guide/timeseries.html#offset-aliases
https://lernerpython.com/2026/10/23/pandas-offset-aliases-for-resample-me-ms-and-other-codes/ lernerpython.com
CdXz5zHNQW_knVmIwe19q.png
RSS Hunter RSS Hunter • Oct 23

Daniel Roy Greenfeld: TIL: How to extract subtitles from video

Recently I've been making more and more videos for my youtube channel. I like videos with subtitles, and decided to add them to my own efforts. This is what I do:This command assumes you have uv and ffmpeg installed.uvx --python 3.14 --from openai-whisper whisper "name-of-video.mp4" --model small --language English --task transcribe --output_format srt --output_dir captions --fp16 FalseThe output is put in a captions directory, named after the .mp4 file that is its source.
https://daniel.feldroy.com/posts/til-2026-10-how-to-extract-subtitles-from-video daniel.feldroy.com
RSS Hunter RSS Hunter • Oct 22

Mariatta: Python Core Sprint 2026: Day 3

Day 3 of the Python Core Sprint, yesterday, was my last day in the room. Today I’m traveling to Seattle for SeaGL, where I’m one of the keynote speakers. With the bots settled on Cabotage and Heroku scaled to zero, I had the day for the smaller things that had been waiting on the follow-up list. A prompt generator for NEWS entries, and why I closed itOne of the open pull requests in blurb_it was #456, from An Long (aisk), opened back in July. It added a button to the “add blurb” page that generates a prompt. The contributor copies the prompt into ChatGPT, Claude, Gemini, or DeepSeek, and pastes the resulting NEWS entry back into blurb_it. The prompt includes links to the issue, the PR, the diff, and the changed files, plus the devguide’s guidance on how to write a NEWS entry.
https://mariatta.ca/posts/python-core-sprint-2026-day-3/ mariatta.ca
RSS Hunter RSS Hunter • Oct 22

Rodrigo Girão Serrão: Module profiling overview

Python 3.15 introduces a new standard library module named profiling. This module consolidates Python's built-in profiling tools under a single namespace. It provides access to two distinct profilers: Tachyon, a sampling profiler, and a tracing profiler. Tachyon, found in profiling.sampling, offers virtually zero overhead by using high-frequency sampling. The tracing profiler, located at profiling.tracing, deterministically profiles code by instrumenting it, though it incurs some overhead. Previously, Python had a deprecated pure-Python module called profile, which had limited practical utility. This old module is slated for removal in Python 3.17. The tracing profiler was formerly known as cProfile in Python 3.14 and earlier versions. This new profiling module aims to organize and modernize Python's profiling capabilities. The article serves as an introduction to these two key profilers.
https://mathspp.com/blog/module-profiling-overview mathspp.com
CdXz5zHNQW_FgX6KQYd0y.webp
RSS Hunter RSS Hunter • Oct 22

Tryton News: Information regarding eMudhra / emSign CA certificate security incident

Following Information for Python users regarding eMudhra / emSign CA certificate security incident - Announcements - Discussions on Python.org, we have updated the Docker images of Tryton because certifi is an indirect dependency of Tryton. It is recommended to update your deployment to use the updated images. If you deploy Tryton using a virtual environment, it must be updated following the instructions.1 post - 1 participantRead full topic
https://discuss.tryton.org/t/information-regarding-emudhra-emsign-ca-certificate-security-incident/9447 discuss.tryton.org
RSS Hunter RSS Hunter • Oct 22

Python Software Foundation: Mark Your Calendars! PSF Board Office Hours for the Year Ahead

The PSF Board will continue its Office Hour sessions on the PSF Discord until 2027. These sessions, which began in September 2024, facilitate community connection with the board. They have been crucial for discussions on the PSF Strategic Plan, elections, and the Grants Program. The office hours also allow the board and staff to understand community activities and challenges. Participants can engage in text-based live chats with PSF Board Directors and staff. Discussions primarily focus on Python, the PSF, and the community, with occasional dedicated topics like PyCon US. The PSF collaborates on promoting Python, supporting local communities, organizing PyCon US, and fostering diversity. They also address Code of Conduct issues and support regional communities through grants. Sessions occur in the #psf-board channel on the PSF Discord, with alternating times for global accessibility. Each session lasts one hour, with future dates beyond October 2027 to be announced. PSF board members and staff attend to listen, help, and provide support, and can follow up on issues.
https://pyfound.blogspot.com/2026/10/mark-your-calendars-psf-board-office.html pyfound.blogspot.com
RSS Hunter RSS Hunter • Oct 22

Zero-with-Dot (Oleg Żero): Text-to-Diagram for AI Systems: Designing a DSL an LLM Gets Right

This text introduces Gnomon, a solution to communication breakdowns in software development, where different stakeholders interpret system descriptions differently. The author, with diverse experience, highlights how terms like "fix" in "We are building an agent to fix invoices" lead to divergent understandings and project failures. The rise of large language models (LLMs) exacerbates this by generating plausible but potentially inaccurate architectural drawings.Gnomon addresses this by providing a small language, gnomocode, and a compiler that validates statements against an ontology. An LLM translates natural language into gnomocode, the compiler checks its validity, and the output can be visualized, queried, or converted back into prose. The compiler, written in Python, uses Lark, rdflib, and pySHACL.The motivation behind Gnomon is to bridge the gap between spoken words and actionable diagrams, ensuring consistency and accountability. It aims to work in both directions: from conversation to model and from code/diagrams to readable text. The forward direction, from conversation to model, is the primary focus of the article.Key requirements for Gnomon include unambiguous terms, native support for modern architectural elements (LLMs, memory, tools), readability for humans and models, multiple views from a single model, version control, and the ability to express uncertainty. The approach evolved from trying simpler methods like directly asking LLMs or using structured objects, which failed to address the complexity of architectural graphs and missing information.Existing notations like UML or C4 were also found insufficient due to their fixed symbols, lack of "maybe" statements, and limited semantic checking. The custom language with a compiler proved to be the most viable solution. The compiler provides localized errors and feedback to the LLM for correction.The core of Gnomon is its ontology, built upon existing RDF vocabularies like SKOS, PROV-O, and DPV. This kernel of 57 terms and twelve modules defines relationships and rules, ensuring consistency. The ontology dictates the lexicon, compiler error messages, prose generation, and diagram views.Gnomocode itself is designed for readability and precision, using a syntax that borrows from Markdown, Python, and graph query languages. It supports concepts like declarations, property assignments, actions, and flows. A crucial feature is the ability to express confidence levels for facts, allowing the system to identify information requiring further confirmation.The compiler, written in Python, uses Lark for parsing. Its grammar is generated directly from the ontology, meaning extending the language involves modifying data, not code. This modular design allows for flexible vocabulary expansion without altering core parsing rules.
https://zerowithdot.com/blog/text-to-diagram-for-ai-systems/ zerowithdot.com
CdXz5zHNQW_IUf3kDkWHC.png
RSS Hunter RSS Hunter • Oct 22

Peter Hoffmann: Loading 33,000 Peaks on Demand with a z10 Tile Grid

biwak.info developed a mobile-first map for Alpine huts, adding 33,437 named peaks with detailed information sourced from OpenStreetMap and Wikidata. To optimize for mobile devices with limited connectivity, the substantial 15.4 MB peaks data needed to be efficiently loaded. The solution was to segment the data into smaller, tile-based JSON files, mirroring the existing slippy map tile grid used by Leaflet. This allows the browser to fetch only the relevant peak data visible in the user's current map view.A fixed zoom level of 10 was chosen for the data grid, providing a balance between the number of network requests and the size of each downloaded file. At this level, an Alpine tile is roughly 27x27 km, and a typical phone screen at zoom 12 touches at most six tiles, with median tile sizes of 13 KB (3.6 KB gzipped).A Python script creates these tile files, along with a manifest of non-empty tiles, ensuring that only necessary files are requested. Each tile's JSON data is compact, using integer coordinates and a fields header to minimize file size. Peaks are ranked by prominence or isolation to control visibility at different map zoom levels, displaying only more significant peaks at lower zooms.The client-side JavaScript uses Leaflet's 'moveend' event to determine which tiles are needed, fetching them with a maximum of four parallel requests to prevent overload on slower connections. An LRU cache of 64 tiles manages downloaded data, dropping least recently used tiles if not in the current view. Peak data is rendered on a single canvas rather than using individual Leaflet markers, enhancing performance.Tapping on the map performs a nearest-neighbor search within loaded tiles to identify nearby peaks. The tile-based grid also naturally supports deep links, allowing stable URLs for specific peaks. Future enhancements include adding more categories of data like passes and viewpoints, and potentially migrating to vector tiles (MVT) for integration with mapping libraries like MapLibre. This system effectively handles a large dataset without a server or database, demonstrating the power of a simple file grid.
https://peter-hoffmann.com/2026/loading-peaks-on-demand-with-a-z10-tile-grid.html peter-hoffmann.com
CdXz5zHNQW_7Oo4LQLM3G.webp
RSS Hunter RSS Hunter • Oct 22

Jaime Buelta: Jev webinar video

I hosted a webinar on the 20th talking about the Jev model, which is a new AI decision model that works well in responding quickly to simple questions to a particular piece of information. This model has been talked about quite a lot recently when it was introduced, as it helps in a lot of workflows where it can be an alternative that’s faster and cheaper than LLMs. It has also spurred a lot of competing decision models, including of course Open Weights alternatives. You can see the recording of the webinar...
https://wrongsideofmemphis.com/2026/10/21/jev-webinar-video/ wrongsideofmemphis.com
RSS Hunter RSS Hunter • Oct 21

Python Engineering at Microsoft: Pylance + Pyrefly: bring your preferred type checker to Visual Studio Code

Microsoft is enhancing the Python developer experience by introducing the Type Server Protocol (TSP). This open standard, an extension of the Language Server Protocol, allows editors and tools to integrate type checkers as backend components. TSP decouples Pylance, the VS Code IDE intelligence layer, from its default type checker, Pyright. This separation enables developers to easily switch to alternative type checkers like Meta's Pyrefly. Pylance's core features, including completions and GitHub Copilot integration, remain consistent regardless of the chosen type checker. Microsoft has partnered with Meta to ensure Pyrefly's full TSP support, leveraging Meta's expertise in static type analysis and AI/ML libraries. This collaboration aims to foster interoperability within the Python ecosystem. Pyrefly, known for its high performance and CI compatibility, can now be seamlessly integrated with Pylance. This offers developers Pyrefly's type analysis engine combined with Pylance's AI-aware intelligence. Users can try this new experience by installing Pylance version 2026.4.1 or newer. Pylance automatically detects Pyrefly installations and prompts users to switch. Alternatively, users can manually select their preferred diagnostics source, including Pylance, Pylance + Pyright, or Pylance + Pyrefly, through VS Code settings. The status bar will then clearly indicate the active diagnostics source.
https://devblogs.microsoft.com/python/announcing-tsp/ devblogs.microsoft.com
CdXz5zHNQW_jwz7YqFtlb.png
RSS Hunter RSS Hunter • Oct 21

Rodrigo Girão Serrão: Tachyon: sampling profiling in Python

Tachyon is a new sampling profiler introduced in Python 3.15. It offers a command-line interface (CLI) for profiling scripts and attaching to running processes. Statistical profiling involves taking periodic snapshots of a program's state to estimate time spent in different code sections. Tachyon, as a sampling profiler, collects these snapshots to build its performance estimates. The accuracy of Tachyon's results increases with longer run times and higher sampling frequencies. The article aims to explain statistical profiling and demonstrate Tachyon's CLI usage. Tachyon is part of the profiling.sampling module in the standard library. This series celebrates Python 3.15 with daily explanations of new features. Subscribing to the email series provides clear examples for each feature. The provided text begins by introducing Tachyon and its core functionality.
https://mathspp.com/blog/tachyon-sampling-profiling-in-python mathspp.com
CdXz5zHNQW_hmfVV9UEzH.webp
RSS Hunter RSS Hunter • Oct 21

Mariatta: Python Core Sprint 2026: Day 2

The author spent Day 2 of the Python Core Sprint migrating GitHub bots from Heroku to the PSF's Cabotage. The migration process, announced for an hour-long downtime, involved switching over three bots: bedevere, blurb_it, and miss-islington. The actual downtime was minimal, with the first bot moved by 11:35 AM and the last by 12:08 PM, and the entire process was completed by 12:36 PM. This transition appeared to go very smoothly, as indicated by positive feedback.
https://mariatta.ca/posts/python-core-sprint-2026-day-2/ mariatta.ca
RSS Hunter RSS Hunter • Oct 21

PyCoder’s Weekly: Issue #757: Django Migrations, Functional Programming, MVC, and More (2026-10-20)

This issue of PyCoder's Weekly covers a range of Python topics. A tutorial explores Django ORM and migrations, detailing how to build, break, and fix them, including addressing performance issues in Django 6.1. Functional programming in Python is explained, discussing its paradigm and practical application. PropelAuth is highlighted for its security features, particularly for AI agent integrations. The Model-View-Controller (MVC) pattern in Python web apps is explored with a Flask example. Python 3.16.0 Alpha 1 is now available, and the Python Bytes podcast celebrated its 500th episode. The PyPy v8.0.1 release is also noted. A Claude Code course teaches building Python CLI apps without excessive ChatGPT reliance. OpenTeams reviews their 2026 PyTorch contributions, noting increased AI agent usage. Bayesian optimization for hyperparameter search is demonstrated with scikit-learn. The unique "+" operator in the Pendulum datetime library is examined. PyPI has updated its email reputation calculation, allowing Outlook addresses again. Parquet's advantages in data engineering are explained using Polars, DuckDB, and PyArrow. Gemini 4 Argon's capabilities are showcased with Python code generation. Project structure in Python is discussed, offering guidance on choosing between flat and src/ layouts. Comparisons are made between Docling and LlamaParse for PDF table extraction. Basic Python concepts like strings, data types, and first steps are covered in tutorials. Finally, a list of upcoming Python events is provided.
https://pycoders.com/issues/757 pycoders.com
CdXz5zHNQW_wpqLkNZyB4.png
RSS Hunter RSS Hunter • Oct 20

Python Bytes: #501 A Nod to Sir-Mix-Alot

Nod is a JupyterLab extension allowing users to pause Python programs and open a notebook at any specific line for debugging. This tool helps in examining variables and the call stack directly within a running program. Polars 2.0 has been released, featuring a default streaming engine for improved memory and speed. However, this new engine does not guarantee row order for certain operations, requiring explicit handling if order is critical. Polars 2.0 also enables out-of-core processing by default, allowing it to spill data to disk when memory is exhausted. The update introduces a native Map data type, enhancing performance and introducing new map manipulation functions. Django-cachalot is presented as a solution for caching ORM queries, automatically invalidating them when underlying data changes. It offers significant speedups for read-heavy applications but is less suitable for tables with frequent write operations. Progress is being made towards supporting asynchronous transactions in Django, aiming to improve how asyncio tasks interact within database transactions. Current proposals involve new designs that allow sibling tasks to continue writing even if a transaction rolls back. Jazzband, a significant project, is set to shut down, prompting exploration of alternative platforms. Treemd is highlighted as a useful TUI for viewing Markdown files. Discussions also touch upon Talk Python membership, visual topic guides, and recent Python Enhancement Proposals. An observation is made about the potential staleness of a free threaded Polars version on PyPI. Finally, a brief mention of a joke concludes the episode's diverse technical discussions.
https://pythonbytes.fm/episodes/show/501/a-nod-to-sir-mix-alot pythonbytes.fm
RSS Hunter RSS Hunter • Oct 20

Rodrigo Girão Serrão: DX improvements in Python 3.15

Python 3.15 introduces several developer experience enhancements, with a particular focus on improved error messages. One notable improvement is more helpful messages when encountering AttributeError exceptions. If an object has a member with the correct name located within a nested attribute, Python will now suggest the correct path. For instance, if o.value is incorrect but o.inner.value exists, the error message will point this out. Furthermore, when attempting to access methods of built-in types and encountering an AttributeError, Python will now check for common typos. If an obvious typo isn't detected, it will consult a table of frequently used methods. This aims to reduce debugging time by providing more direct guidance. The article is part of a series celebrating Python 3.15 features, with daily explanations and examples. Other improvements mentioned include the addition of more colors and the default use of UTF-8 encoding. These changes are designed to make Python development more efficient and user-friendly. The author reported a bug in the nested attribute error suggestion feature during the alpha phase, which was quickly addressed.
https://mathspp.com/blog/dx-improvements-in-python-315 mathspp.com
CdXz5zHNQW_saimMyYTTU.webp
RSS Hunter RSS Hunter • Oct 20

Mariatta: Python Core Sprint 2026: Day 1

This week I’m in San Francisco for the annual Python Core Sprint. This year the sprint is hosted by Brandt Bucher from OpenAI, at OpenAI’s office, and runs Monday to Friday, October 19 to 23. I’m only here for the first three days. On Thursday I fly out to Seattle for SeaGL, where I’m one of the keynote speakers :)I could not make it to last year’s sprint at Arm in Cambridge, UK, and it is unlikely I’ll be able to make it to next year’s European sprint either, so I’m glad this one is on the West Coast and within reach. The sprint is one of the very few times in a year when this many Python core team members are in the same room, focused on nothing but Python. It is still one of my favorite Python events of the year. Thank you to Brandt and OpenAI for hosting us!
https://mariatta.ca/posts/python-core-sprint-2026-day-1/ mariatta.ca
RSS Hunter RSS Hunter • Oct 20

Python Software Foundation: Wikis: Relaunched!

In February 2026, the Python Software Foundation (PSF) made its wikis read-only due to security risks and excessive bot traffic overwhelming outdated infrastructure. This abrupt decision caused disruption for community members who relied on and contributed to these resources. To address these issues, the PSF has launched new wikis hosted on GitHub Pages, offering a modern front-end and a permissions-based content management system. These new wikis are now accessible via provided links for the PSF, Python, and Jython wikis. The excessive automated traffic was primarily from AI bots attempting to scrape the internet, which overloaded the old servers. Additionally, some portions of the old wikis were running on unsupported Python 2, posing security vulnerabilities. The PSF acknowledges the frustration caused by the sudden retirement of the old wikis, especially for dedicated contributors. They aim for the new wikis to be a valuable and improved community resource, with contributions now managed through pull requests on GitHub. GitHub was chosen to streamline hosting and stewardship, aligning with existing PSF-hosted projects. The new permissions-based system via GitHub repositories allows anyone to request changes through pull requests. This move to a static site generated on GitHub mitigates security risks and protects against bot-driven overload. Contributions are encouraged by reviewing repository README files, and established contributors will be granted write access.
https://pyfound.blogspot.com/2026/10/wikis-relaunched.html pyfound.blogspot.com
RSS Hunter RSS Hunter • Oct 20

LernerPython blog, from Reuven Lerner: Pandas DatetimeIndex slicing: Fix KeyError with sort_index

Slice an unsorted datetime index in Python Pandas, and you’ll likely get:KeyError: ‘Value based partial slicing on non-monotonic DatetimeIndexes with non-existing keys is not allowed.’Why? If the dates are unsorted, where should Pandas start/stop?Solution: sort_index!
https://lernerpython.com/2026/10/20/pandas-datetimeindex-slicing-fix-keyerror-with-sort-index/ lernerpython.com
CdXz5zHNQW_rTcYH1V05z.png
RSS Hunter RSS Hunter • Oct 20

PyCon Ireland: Schedule for PyCon Ireland 2026

PyCon Ireland 2026 will take place at Chartered Accountants House in Dublin on Saturday, November 21st. The conference features over thirty talks across three tracks, at least four workshops, and two panels, alongside keynotes and lightning talks. A significant portion of the schedule is dedicated to Artificial Intelligence and Machine Learning topics. These include agentic AI, Bayesian modeling, ML benchmarking, data architecture, forecasting, ML monitoring, and neural networks. Beyond AI, sessions will cover asynchronous I/O, Python packaging, quantum programming, secure coding, and software archaeology. The organizers received over 170 submissions and thank all contributors for their proposals. Tickets are now available, with General tickets priced at €59, Student tickets at €35, and Corporate tickets at €159. All tickets are all-inclusive, providing access to all sessions, networking opportunities, breakfast, lunch, and refreshments. The schedule may see further additions, so attendees are encouraged to check for updates.
https://main--pycon-ireland-2026.netlify.app/blog/schedule-available/ main--pycon-ireland-2026.netlify.app
RSS Hunter RSS Hunter • Oct 20

Seth Michael Larson: Documenting certificate distribution pain points for the Python ecosystem

The eMudhra CA compromise highlighted systemic issues within the Python ecosystem's handling of certificate authorities. CPython relies on OpenSSL, which needs CA root certificates, and Python bundles these via the certifi package for platforms lacking native support. Historically, certifi was crucial, but Linux distributions now provide their own better-managed certificate stores. A significant problem is the lack of a default, active certificate revocation mechanism in OpenSSL-based Python installations. Upgrading certifi as a dependency causes cascading updates across the entire Python and pip ecosystem. Pip bundles certifi, meaning a compromise can necessitate urgent updates to both pip and Python itself. The Python Security Response Team must then coordinate releases across multiple Python branches. Pip is transitioning to use truststore by default, which would reduce its reliance on certifi, especially for Linux users. This incident required many other packaging tools and infrastructure components to act quickly. Alerting users to upgrade dependencies after patches are applied remains a major challenge. The author suggests that PyPI should cease being a certificate distribution channel, given that most platforms now offer adequate CA bundles. A shift towards relying more on system-provided certificate stores is proposed to alleviate the burden on Python and pip maintainers. The open question of effective certificate revocation for OpenSSL-based systems needs addressing to reduce future volunteer strain.
https://sethmlarson.dev/python-cert-dist-pain-points sethmlarson.dev
RSS Hunter RSS Hunter • Oct 20

Rodrigo Girão Serrão: Startup configuration files

Python 3.15 introduces startup configuration files with a .start extension to execute package entry points upon interpreter startup. These entry points function similarly to import statements, specifying a module followed by a colon and a callable function. For instance, error_links:install directs the interpreter to call the install function within the error_links module. This new feature builds upon the functionality of the standard library's site module, which automatically runs on startup. Historically, .pth files were used to extend Python's module search path and could also contain import lines. However, import lines in .pth files presented security risks. PEP 829 addressed this by introducing a more secure method for running custom startup code. Consequently, import lines in .pth files are now deprecated in favor of .start files in Python 3.15 and later. Python 3.15 offers a range of enhancements, including built-in types, lazy imports, improved typing, and more visual features. The "15 days of Python 3.15" series provides daily explanations of these new features.
https://mathspp.com/blog/startup-configuration-files mathspp.com
CdXz5zHNQW_88iLQz2pFM.webp
RSS Hunter RSS Hunter • Oct 19

Django Weblog: Quarterly contributor outreach

The Django Steering Council is seeking community contributions for Django 6.2. They are highlighting major features that require testing and review, such as CompositeField for multi-column subqueries and the DEFAULT_PK_FIELD setting for alternative primary keys like UUIDs. The council also points to a talk on testing projects against Django's main branch, which can be adapted to test specific feature branches. Smaller features also need attention, including adding QuerySet.executed_queries, partitioned cookies, support for the HTTP QUERY method, and improved CSRF debug messaging behind proxies.The New Features queue is also a focus, with ideas categorized by the type of input needed. Several ideas requiring technical vetting include customizing model Meta and migration generation, implementing modern CSRF protection, adding a timeout API for tasks, subresource integrity, and incorporating import maps. For those with limited time, community input on existing ideas is encouraged through emoji reactions, especially the 😕 emoji to indicate a review has been seen. Specific ideas needing community feedback include better hooks for password reset email customization, PR help for Git documentation, service URL parsing, standardizing task execution order, updating AuthenticationMiddleware, and supporting Unicode email addresses via SMTP. The council thanks the community and encourages participation through the Django Forum or Discord.
https://www.djangoproject.com/weblog/2026/oct/19/quarterly-contributor-outreach/ djangoproject.com
RSS Hunter RSS Hunter • Oct 19

Hugo van Kemenade: Releasing Python 3.15.0

The release of Python 3.15.0 was chronicled through a series of updates, beginning two weeks before the final release. Initial excitement built up, with inquiries about anticipated features. On the planned release day, a certificate issue caused a 503 error on crucial Python websites, delaying the process. Once the server issues were resolved, a flaky test and several release-blocker issues, including last-minute lazy import fixes, required attention. To properly incorporate these fixes, a surprise third release candidate, 3.15.0rc3, was announced and released, pushing the final release date back by a week. A dependency on unbackported tests caused a delay in merging one of the lazy import blockers. After the successful release of rc3, the team prepared for the final launch. On the second attempt at the final release, all blockers were cleared, and buildbots were green. However, a bug in the changelog tool, blurb, necessitated a quick fix, update, and backport. Following this, the release artifact build failed due to another blurb-related lockfile issue, which was promptly resolved. Installer builds for macOS and Windows were initiated, with the Windows process requiring manual approval for binary signing. Finally, Python 3.15.0 was released, boasting new features like lazy imports, frozendict, and a faster JIT. The post-release tasks included updating PEP 790, rebuilding documentation, and unlocking the 3.15 branch for further core team merges. The release manager then signaled the next alpha release for Python 3.16.
https://hugovk.dev/blog/2026/releasing-python-3-15-0/ hugovk.dev
CdXz5zHNQW_5t7QXRxkX2.png
RSS Hunter RSS Hunter • Oct 19

Python Software Foundation: Announcing Python Software Foundation Fellow Members for Q3 2026! 🎉

The Python Software Foundation (PSF) is delighted to announce its third cohort of PSF Fellows for 2026. This group, recognized for their outstanding contributions to the Python community, includes Aldo Caneo, Donghee Na, Paul Moore, and Takayuki Shimizukawa. These individuals are acknowledged for their leadership, support of the scientific Python community, maintenance of virtual communities and libraries, creation of educational materials, organization of events, and mentorship. Their efforts significantly contribute to Python's global accessibility. The PSF encourages recognition of Pythonistas worldwide and provides criteria for Fellow membership on its website. Nominations for Quarter 4 of 2026 are open until November 20th, 2026. Aspiring nominees should submit a description of their Python accomplishments and their email address to psf-fellow at python.org. Current PSF Fellows interested in assisting with nomination reviews are also invited to contact psf-fellow at python.org. The PSF expresses gratitude for the ongoing dedication of these fellows.
https://pyfound.blogspot.com/2026/10/announcing-python-software-foundation.html pyfound.blogspot.com
RSS Hunter RSS Hunter • Oct 19

Peter Hoffmann: Home Server Ops with a Read-Only Agent and pyinfra

A home server, initially managed manually, was transformed using a coding agent and declarative configuration. This agent has read-only access, preventing direct server modifications. All changes are now codified in pyinfra deploys within a git repository. The agent explores the server, writes Python code for desired state changes, and presents these changes for review. Crucially, the sudo password acts as a gatekeeper, requiring user intervention for any actual deployment. The repository structure includes agent rules, a server state inventory, pyinfra deploys, and configuration files. The agent's tasks include reading documentation, locating configurations, and writing systemd units. The user reviews dry runs and then executes the code for deployment. This workflow handles system upgrades, migrating services, and deploying custom applications. Mistakes are learned from and incorporated into agent rules for future sessions. While effective for personal data, read-only access requires consideration for machines holding sensitive third-party information.
https://peter-hoffmann.com/2026/home-server-with-a-read-only-agent-and-pyinfra.html peter-hoffmann.com
CdXz5zHNQW_zasqR31jDU.png
RSS Hunter RSS Hunter • Oct 19

The Python Coding Stack: 2. Who actually does the work, the model or the software?

The conversation explores the components of an AI agent, differentiating the Large Language Model (LLM) from the surrounding software. When an AI organizes photographs, the LLM doesn't directly access files on a computer; instead, it generates requests. This request is received by the "harness," which is the software managing the agent's operations. The harness then uses specific "tools" to perform actions like listing files or reading metadata.The harness sends the results from these tools back to the LLM, allowing it to process the information and decide on the next step. The LLM communicates its requests to the harness through structured "tool calls," which the harness interprets to invoke the correct tool. This interaction creates an agent loop where the model, harness, and tools repeatedly exchange information.The analogy of the model as the "brain," tools as "muscles," and the harness as the "coordinator" helps clarify their roles. However, the harness also enforces rules and restrictions, such as preventing unintended file modifications. The process concludes when the model generates a final response or the harness halts the operation due to errors or limits. The agent must also verify that the task, like copying files to specific folders, has been completed accurately and without altering originals.
https://www.thepythoncodingstack.com/p/2-who-actually-does-the-work-the thepythoncodingstack.com
CdXz5zHNQW_eyYaVLyYu6.jpeg
RSS Hunter RSS Hunter • Oct 17

EuroPython: Humans of EuroPython: Emma Cooke

This article highlights Emma Cooke's significant contributions to EuroPython 2026 as part of the Accessibility Team. Emma's involvement elevated the conference's approach to inclusivity, aiming to make it welcoming for all attendees. She described her volunteering experience as energizing, welcoming, and fulfilling. Having previously attended as an attendee and then as a volunteer, her role in 2026 on the organizing committee was her first in that capacity. Emma focused on making the conference physically accessible and supportive for neurodivergent individuals. Her responsibilities included ensuring quiet rooms, providing stim toys, and addressing various attendee needs like childcare and dietary requirements. A key achievement was introducing the Hidden Disabilities Sunflower scheme to EuroPython. This initiative allows individuals with non-visible disabilities to discreetly signal they might need understanding. Emma's perspective on conferences has fundamentally changed, fostering deep appreciation for behind-the-scenes work. She encourages others to volunteer to collaborate with passionate individuals, make a tangible impact, and gain unique insights.
https://blog.europython.eu/humans-of-ep-emma/ blog.europython.eu
CdXz5zHNQW_fdEJ2urIrT.png
RSS Hunter RSS Hunter • Oct 17

Talk Python Blog: Introducing Talk Python Membership

For years, I’ve wanted to offer something premium for podcast listeners. The obvious option was an ad-free feed. That never sat right with me. No ads is not much of a bonus. So over the last few months, I’ve been building something that goes beyond the podcast and beyond the courses. Today I’m thrilled to announce Talk Python Membership.In short, membership gets you community, live tech meetups, episode previews, and yes an ad-free feed.
https://talkpython.fm/blog/posts/introducing-talk-python-membership/ talkpython.fm
RSS Hunter RSS Hunter • Oct 17

Talk Python to Me: #566: State of the art in Python 2026

The state of Python tools in 2026 is being reshaped by advancements in AI, prompting a reevaluation of long-settled choices. A panel of experts including Calvin Hendryx-Parker, Pat Decker, Reuven Lerner, and Sydney Runkle discusses these shifts. Some established tools have seen performance improvements, while others have been entirely superseded. The conversation explores the current landscape and surprising replacements in the Python ecosystem. The episode features insights from individuals involved in DevOps, developer training, and product management for AI tools like LangChain. One panelist even admits to rarely using a traditional code editor anymore. This discussion highlights how rapidly evolving AI technology is influencing the selection and use of Python development tools. The episode aims to provide listeners with an updated perspective on the most effective Python tools available today. It delves into specific examples of tools and frameworks that are either new or have significantly improved. Ultimately, the panel offers guidance on navigating the modern Python development environment.
https://talkpython.fm/episodes/show/566/state-of-the-art-in-python-2026 talkpython.fm
RSS Hunter RSS Hunter • Oct 17

Artem Golubin: Python's binary wheels and supply chain attacks

Python is known for fast packages despite being a slow language. The fastest Python packages are written in compiled languages such as C, C++, or Rust, with Python acting only as a glue.Python users usually don't even notice that, because PyPI (Python's package index) provides precompiled binaries. They are called wheels, and when you run pip install, it prefers them over the source distribution. For example, each release of my packages ships up to 90 different wheels. You need a separate binary wheel for each Python version, platform, and architecture combination. The regular distribution only contains the source code.Without Python wheels, the distribution would be much harder. When wheels are not available, a user has to compile the code when installing a package. That requires the correct compiler and system dependencies.Since I've spent quite a while analyzing malicious packages on PyPI, this got me thinking about wheels.[......]
https://rushter.com/blog/python-binary-wheels/ rushter.com
RSS Hunter RSS Hunter • Oct 15

Rodrigo Girão Serrão: Unpacking inside comprehensions

Python 3.15 introduces the ability to use the asterisk operator for unpacking within list comprehensions. This new feature simplifies flattening nested lists, allowing for more concise code. For instance, a list of lists can be flattened by unpacking each sublist directly into the comprehension. Previously, achieving the same result required a nested list comprehension, which was more verbose. The unpacking syntax mirrors the structure of equivalent nested for loops. The unpacking operator works not only with list comprehensions but also with set comprehensions, dictionary comprehensions, and generator expressions. This article details the new syntax and provides alternatives for older Python versions. It is part of a series celebrating Python 3.15 features with clear explanations and examples. The goal is to make new functionalities accessible without needing to comb through official changelogs. Readers can subscribe to receive this free educational email series.
https://mathspp.com/blog/unpacking-inside-comprehensions mathspp.com
CdXz5zHNQW_X0sf33USpe.webp
RSS Hunter RSS Hunter • Oct 15

Bob Belderbos: My bookkeeping runs on code now. AI only reads the PDFs.

A freelancer found their paid bookkeeping tool too slow and lacking in essential reporting. They decided to build their own solution using Python, resulting in a small codebase with extensive testing. This custom system includes an invoice scanner with human oversight for data extraction. The process automates monthly bookkeeping tasks, generating an Excel report for their accountant. While tempted by AI, the freelancer identified only one step needing AI: extracting data from varied supplier PDFs. The rest of the process relies on robust code for arithmetic and validation. The system emphasizes a human-in-the-loop for critical decision points and data approval. Once approved, PDFs are renamed to serve as immutable records, eliminating the need for a database. The system's design prioritizes immutability and traceability, with all totals in the final report being formulas. This approach ensures easy verification by the accountant. The freelancer now maintains the system, but gains full control, reduced costs, and an improved workflow. Key takeaways include mapping workflows, using AI sparingly, coding rules, human oversight, and ensuring traceability.
https://belderbos.dev/blog/replaced-my-bookkeeping-tool-with-python/ belderbos.dev
CdXz5zHNQW_B1IdOrwP5A.png
RSS Hunter RSS Hunter • Oct 15

Rodrigo Girão Serrão: frozendict built-in

Python 3.15 introduces a new immutable dictionary type called frozendict. You create frozendicts using the built-in frozendict() constructor, not curly braces. This ensures that the dictionary's contents cannot be changed after creation. You can access elements within a frozendict using standard dictionary key access. Iterating through a frozendict's keys, values, or items is also supported. It is recommended to use frozendict for dictionaries that should not be modified. This immutability makes frozendicts hashable, a key characteristic. Understanding hashability is important for their use. The article series celebrates Python 3.15 features with daily explanations. The author plans to cover new features over 15 business days with examples.
https://mathspp.com/blog/pydonts/frozendict-builtin mathspp.com
CdXz5zHNQW_ybsC98iweG.webp
RSS Hunter RSS Hunter • Oct 14

PyPy: PyPy v8.0.1 release

The PyPy team has released version 8.0.1, updating PyPy on September 14, 2026. This release fixes buffer memory leaks and issues with tying RPython objects to C PyObject objects. The standard libraries for Python 3.11 and 3.12 were updated, along with vendored libexpat. Internally, pytest 3.10 is now used for testing.The release includes PyPy2.7, supporting Python 2.7.18+, and PyPy3.11, supporting Python 3.11.17, which will be its last release. PyPy3.12, supporting Python 3.12.15, is also included. All three interpreters are based on a shared codebase.Users are encouraged to update and can find download links on pypy.org. The team thanks donors and invites contributions to bug fixes, documentation, and JIT improvements. Library maintainers using C-extensions are urged to create CFFI versions for better PyPy performance, or utilize upcoming cp312-abi3 tag support.PyPy is a fast Python interpreter, a CPython replacement, known for its integrated tracing JIT compiler. Binary builds are provided for x86 machines (Linux, Mac OS, Windows) and 64-bit ARM (Linux, macOS). Other platforms like Windows 32-bit and various Linux architectures are supported, but without official binaries.The PyPy team extends gratitude to contributors and welcomes new members. They also invite developers of other dynamic languages to explore RPython. For more details, the full changelog is available.
https://www.pypy.org/posts/2026/10/pypy-v801-release.html pypy.org
RSS Hunter RSS Hunter • Oct 14

Python Bytes: #500 Episode 500!

The episode features a discussion with Brian, extra content, and a joke. Nick explores Jev, a "System 1" decision model that offers fast and cost-effective classification without token generation, positioning it between generative LLMs and traditional ML models. Brett announces the election of the inaugural Python Packaging Council (PPC), established via PEP 772, which is currently defining its scope and logistics. Calvin presents Daniel Lemire's benchmark of fastsimdjson for Python, demonstrating its ability to parse JSON data at tens of gigabytes per second, especially with multi-threading and lazy parsing. Michael discusses Miguel Grinberg's annual Python benchmark, showing that Python 3.15's experimental JIT build is significantly faster, though PyPy and Rust still outperform CPython in single-threaded tests. Carol introduces Satyrn AI, a community-driven initiative focused on transparent and trusted AI tools for Python developers. Michael also highlights ArtCraft's Rust-based, AI-built open-source alternatives to Adobe Creative Suite, emphasizing the changing landscape of software development and sustainability challenges. Calvin shares a link to llmfit.org for checking model compatibility and notes uv 0.13.0 now defaults to Python 3.15. Nick promotes the Black Python Devs' student ambassador program, and Brett announces the final release of Python 3.15.0. Finally, a joke about an "exceptional" Python developer who raises many exceptions concludes the episode.
https://pythonbytes.fm/episodes/show/500/episode-500 pythonbytes.fm
RSS Hunter RSS Hunter • Oct 14

Mark Dufour: CPython 3.15 (JIT) Performance Showdown

A performance comparison was conducted between CPython 3.15 and other Python implementations using eight example programs. The goal was to assess CPython's progress and compare it to alternatives like PyPy and Ahead-of-Time (AOT) transpilers. The benchmark results showed significant improvements in CPython 3.15 with its JIT compiler. Without the JIT, CPython 3.15 offered a modest speedup, but with the JIT enabled, performance increased substantially. This JIT improvement in 3.15 is considered a key takeaway, suggesting potential future convergence with PyPy's performance. PyPy 7.3.20 demonstrated a very impressive speedup, even with its broad Python feature support. AOT transpilers like TurboPython and Shed Skin, which operate on restricted Python subsets, achieved even greater speedups. TurboPython showed particularly strong performance, being the fastest for some programs, though it has some areas for improvement. Shed Skin also saw performance enhancements in its latest versions. The author acknowledges a slight bias towards Shed Skin in the results due to their involvement. Full detailed results for each program are available separately.
http://shed-skin.blogspot.com/2026/10/cpython-315-jit-performance-showdown.html shed-skin.blogspot.com
CdXz5zHNQW_ZYgwc6b5Y1.png
RSS Hunter RSS Hunter • Oct 14

PyCon Ireland: Call for sponsors PyCon Ireland 2026

PyCon Ireland 2026 will be a one-day, multi-track conference for Python enthusiasts, expected to draw around 500 attendees. The event will feature talks, workshops, tutorials, and an evening networking event. This conference offers a valuable opportunity for companies to connect with a diverse audience of software developers, data scientists, and entrepreneurs. Sponsorship packages range from Diamond at €7,000 to Silver at €2,500, with varying levels of brand visibility and included conference tickets. Diamond sponsors receive prime exhibition space, stage signage, and the option to present a talk. Platinum sponsors also get premium exhibition space and stage signage. Gold sponsors receive a poster exhibition, and Silver sponsors get basic website and email mentions. Special sponsorship options include sponsoring the bar and entertainment, lanyards, volunteers, or workshops, each with unique branding opportunities. Past sponsors include major tech companies like IBM, MongoDB, and Stripe. The event is organized by Python Ireland and will take place in Autumn 2026. Attendees receive access to all conference activities and provided meals.
https://main--pycon-ireland-2026.netlify.app/blog/call_for_sponsors/ main--pycon-ireland-2026.netlify.app
RSS Hunter RSS Hunter • Oct 14

PyCoder’s Weekly: Issue #756: 3.15 Released For Real This Time, Muse, and More (2026-10-13)

This newsletter highlights upcoming features and news relevant to Python developers. Python 3.15 is anticipated to introduce five exciting new features, with performance benchmarks showing its speed. CodeAF is presented as an efficient tool for AI-driven code problem-solving. A guide to Meta's AI agent, Muse, is offered for those interested in its capabilities and data handling. Python Enhancement Proposals (PEPs) are discussed as a way for developers to understand language changes. The newsletter also covers practical tutorials on various topics. These include generating images with OpenAI's API, leveraging Hugging Face Transformers for open-source AI, and optimizing flight paths with scikit-decide. Advanced topics like async transactions in Django, building async-aware profilers, and memory management in Python are explored. Additionally, several new projects and code repositories in the Python ecosystem are featured. The newsletter concludes with a diverse list of upcoming Python-related events and meetups.
https://pycoders.com/issues/756 pycoders.com
CdXz5zHNQW_xNle5WNziN.png
RSS Hunter RSS Hunter • Oct 13

Rodrigo Girão Serrão: Sentinel built-in

Python 3.15 introduces a new built-in called sentinel for creating placeholder values. This provides a more idiomatic approach for default arguments or return values when None is not suitable. The sentinel function takes a string argument for its name, which also serves as its default string representation. Optionally, a second named argument can specify a custom representation. It is good practice for the sentinel's name and the variable it's assigned to to match, as linters and type checkers may flag discrepancies. Sentinels often use uppercase names as they are typically unchanging constants. The primary use of sentinels is for comparison, checking if a value matches a specific placeholder. This is useful in functions that need to distinguish between a valid value and an intended absence of one. The "15 days of Python 3.15" series will explore new features with clear examples. An example scenario involves a function called "find" that utilizes sentinels.
https://mathspp.com/blog/pydonts/sentinel-builtin mathspp.com
CdXz5zHNQW_EoVfbilALi.webp
RSS Hunter RSS Hunter • Oct 13

Seth Michael Larson: Gaming’s all-digital future doesn’t include “Sonic the Hedgehog 3” and “Pikmin 2”

Sony has announced that all games released from 2028 onwards will be digital-only. This shift is expected to extend to Xbox, while Nintendo may continue offering physical releases for its first-party titles. An all-digital gaming future will subject platforms to the same constraints as streaming services like Netflix. Recent re-releases of Sonic the Hedgehog 3 and Pikmin 2 illustrate potential issues. Digital games function as licenses, requiring ongoing relationships between players and media owners. These relationships are affected by the media owner's licensing agreements for any included content. When a license expires or faces issues, developers must rectify it within the game. Games featuring music, like Rock Band, already demonstrate this with expiring licensing terms. Sonic the Hedgehog 3's soundtrack was altered in re-releases due to alleged uncredited involvement by Michael Jackson. Similarly, Pikmin 2's inclusion of real-life brands was replaced with fictional ones in its latest release. This trend suggests that future games may face creative limitations and challenges with ownership and preservation.
https://sethmlarson.dev/gaming-all-digital-future-and-creative-constraint sethmlarson.dev
CdXz5zHNQW_Iv3vUonWZa.jpeg
RSS Hunter RSS Hunter • Oct 13

Rodrigo Girão Serrão: Explicit lazy imports

Python 3.15 introduces explicit lazy imports to improve application startup times. These imports are indicated by the "lazy" keyword preceding the import statement. While normal imports load modules immediately, lazy imports defer module loading until the module is actually used. This prevents unnecessary overhead during the initial startup phase of a Python application. The article series aims to explain this new feature with clear examples. It will also cover how to write code that is compatible with older Python versions. The existing Python import system loads modules as soon as an import statement is encountered. This can lead to significant delays if many modules are imported at the beginning of a script. Lazy imports address this by only loading modules when they are explicitly accessed. This is particularly beneficial for modules that might not be used in every execution path of an application. The focus is on making Python applications start faster by optimizing the import process.
https://mathspp.com/blog/explicit-lazy-imports mathspp.com
CdXz5zHNQW_RgxhaSl0ry.webp
RSS Hunter RSS Hunter • Oct 12

LernerPython blog, from Reuven Lerner: Pandas groupby: Aggregating multiple columns with multiple functions

Does Python Pandas allow multiple data columns and multiple methods?Yes!df.groupby('passenger_count')[['trip_distance', 'total_amount']].agg(['mean', 'std'])The above returns a data frame. Its 4 columns are a 2-level multi-index with mean/std in the inner level.
https://lernerpython.com/2026/10/12/pandas-groupby-aggregating-multiple-columns-with-multiple-functions/ lernerpython.com
CdXz5zHNQW_SH2qLnYdGP.png
RSS Hunter RSS Hunter • Oct 12