Remote

Michael Kennedy

@mkennedy@fosstodon.org

mastodon 4.7.3

A Python enthusiast and entrepreneur. Host of @talkpython@fosstodon.org and @pythonbytes@fosstodon.org podcasts, founder of Talk Python Training. #Python Software Foundation Fellow. Based in Portland, OR USA.

0 Followers

0 Following

20 Posts

Joined November 08, 2022

Talk Python:

@talkpython

Python Bytes:

@pythonbytes

Writing:

https://mkennedy.codes

Courses:

https://training.talkpython.fm

Michael Kennedy @mkennedy@fosstodon.org

· 6mo ago

New essay: Cutting Python Web App Memory Over 31%

I reduced Python web app memory by 3.2 GB using async workers, import isolation, the Raw+DC database pattern, and disk caching. Detailed before and after numbers for each technique.

https://mkennedy.codes/posts/cutting-python-web-app-memory-over-31-percent/

mkennedy.codes

Michael Kennedy @mkennedy@fosstodon.org

· 9mo ago

Python Numbers Every Programmer Should Know

New essay. Feedback welcome :) #python

https://mkennedy.codes/posts/python-numbers-every-programmer-should-know/

fosstodon.org

Fosstodon

Michael Kennedy @mkennedy@fosstodon.org

· 8mo ago

I'm thrilled to announce I just released Command Book App: A native macOS app for developers, data scientists, AI enthusiasts and more. https://commandbookapp.com

This is a tool I've been using lately to help build Talk Python, Talk Python Training, and many more applications.

It's a bit like advanced terminal commands or complex shell aliases, but hosted outside of your terminal. This leaves the terminal there for interactive commands, exploration, short actions. (1/2)

commandbookapp.com

Michael Kennedy @mkennedy@fosstodon.org

· 8mo ago

I just added an MCP server to @talkpython@fosstodon.org If you want to use the entire 10 year catalog of episodes, transcripts, guests, etc as part of your #AI / #LLM exploration. #python Just visit https://talkpython.fm/api/mcp/docs

fosstodon.org

Fosstodon

Michael Kennedy @mkennedy@fosstodon.org

· 7mo ago

Just added a new feature to Tallyman. Pass --image and you'll get a nice, sharable summary of your project.

https://mkennedy.codes/tools/tallyman/

mkennedy.codes

Michael Kennedy @mkennedy@fosstodon.org

· 8mo ago

New essay. This is an interesting one:

Raw+DC: The ORM pattern of 2026?
https://mkennedy.codes/posts/raw-dc-the-orm-pattern-of-2026/

Michael Kennedy's Thoughts on Technology

Raw+DC: The ORM pattern of 2026?

After 25+ years championing ORMs, I've switched to raw database queries paired with Python dataclasses. Better AI coding support, fewer aging dependencies, surprising performance gains, and type safety where it counts.

Michael Kennedy @mkennedy@fosstodon.org

· 6mo ago

Replying to

@hwine @wsha @pythonbytes @brianokken Thanks Hal. Well, I've seen that as well. Well, and I know that's why some projects require you to sign some sort of agreement passing over your copyright to the project even when it's open source. I seriously doubt Chardet, a simple encoding library, has such a setup though.

Michael Kennedy @mkennedy@fosstodon.org

· 6mo ago

Replying to

@wsha @pythonbytes @brianokken I think Dan does own the new copyright. Though there are issues of whether fully AI creations can actually be copyrighted. It's legally weird for sure and could have mega consequences beyond character encoding. :)

Michael Kennedy @mkennedy@fosstodon.org

· 7mo ago

Replying to

@mborus @talkpython Thanks. It really solves the problem nicely doesn’t it? I have it doing a ton of heavy lifting for the various Talk Python websites and couldn’t be happier.

Michael Kennedy @mkennedy@fosstodon.org

· 10mo ago

Replying to

@wsha @pythonbytes @brianokken Thanks wsha! I do think @brianokken said they were rewriting it in Go or whatever they are using but it wasn't emphasized. I didn't know about the status of MkDocs.

Michael Kennedy @mkennedy@fosstodon.org

· 11mo ago

Replying to

@arildsen @pythonbytes @brianokken Yes! Wouldn’t that be dope? I might prototype it. Like I was saying. I got a 7-8x speed up on a 10 core machine.

Michael Kennedy @mkennedy@fosstodon.org

· 11mo ago

Replying to

@arildsen @pythonbytes @brianokken Thanks! Glad you're enjoying the show. My thought was to use multiprocessing to cross the python <-> pythont boundary. Run under python, but multiprocessing burst into pythont sub processes.

Michael Kennedy @mkennedy@fosstodon.org

· 6mo ago

Replying to

@wsha @pythonbytes @brianokken Fair point about someone new to chardet assuming MIT and getting screwed. But the comment I was addressing was a company that currently used the LGPL license version and said the move to MIT harmed them. All they have to do is keep doing the same thing they are currently for LGPL and they are fine. (1/2)

Michael Kennedy @mkennedy@fosstodon.org

· 6mo ago

Replying to

@gumnos @pythonbytes Hey Tim. Thanks for the feedback. I just counted and of the last 8 episodes we had 25 main topics that we not AI and 6 that were. That's 20%. I feel like how central Python is to AI and how much AI is transforming the dev industry, 20% of the topics relating to AI is prety low-key and not "AI Bytes” (1/2)

Michael Kennedy @mkennedy@fosstodon.org

· 6mo ago

Replying to

@wsha @pythonbytes @brianokken Hi. Thanks. If the maintainer is not the owner, who is the owner?

Michael Kennedy @mkennedy@fosstodon.org

· 6mo ago

Replying to

@wsha “It depends on the agreement was between Mark and Dan” This is true and I don't really have visibility into it. But as you said, I don't think it matters in this case. He thew out that code and replaced it with newly created, newly licensed code.

Michael Kennedy @mkennedy@fosstodon.org

· 7mo ago

Replying to

@graham_knapp @brianokken @pythonbytes Haha! Truth.

Michael Kennedy @mkennedy@fosstodon.org

· 22mo ago

Replying to

@JoeRecursionJoe@fosstodon.org @talkpython@fosstodon.org @samuelcolvin@fosstodon.org Not until now. Thanks. :)

Michael Kennedy @mkennedy@fosstodon.org

· 4mo ago

Replying to

@wigging@fosstodon.org @talkpython@fosstodon.org Hey Gavin! Yes, I am very excited about that episode. I had to make it a private recording. In order to be the most genuine, the team wanted to review what we discussed. So it was super guarded public stream or record first, review, then publish a deeper look. I did the later. So maybe it'll be out on YouTube and podcast either this or next week.