From Daniel Roy Greenfeld, “TIL: Loading .env files with uv run”
We don’t need python-dotenv, use uv run with —env-file, and your env vars from .env get loaded.
Good to know, even though I’m all in on direnv to auto load .env
files. Also, handy to make Poe the Poet tasks that invoke uv
underneath the covers really explicit.
Link parkin, from Trey Hunner’s site: Python Articles on Cheat Sheets
A collection of the many Python cheat sheets within Python Morsels articles and screencasts.
I especially like the cheatsheet on the pathlib module:
I now use pathlib for nearly all file-related code in Python, especially when I need to construct or deconstruct file paths or ask questions of file paths.
I’d like to make the case for Python’s pathlib module… but first let’s look at a cheat sheet of common path operations.
Now that I have working search on this here blog I can ask questions like, “Have I ever mentioned amapiano music?” And as of this moment, the answer is “no”.
Let’s fix that.
Here’s the intro paragraph on amapiano from Wikipedia
Amapiano is a genre of music from South Africa that became popular in mid-2012 with an earlier regular occurrence on South African radio stations in the early 2000s. It is a hybrid of kwaito, deep house, gqom, jazz, soul, and lounge music characterized by synths and wide, percussive basslines. The word “amapiano” derives from the IsiZulu word for “pianos”.
I can’t pinpoint an exact moment, but it most likely was a few years ago coming out of the pandemic where I first bumped into the genre. YouTube tags me as streaming this video of TxC playing Boiler Room London four years ago. I’m pretty sure that was the first hit because I was also somewhat astounded that I’d sit through an hour long video of a dj session. Or at least have it on in the background. Also, TxC are definitely a hot look on stage. Last but not least that was likely my intro to Boiler Room which will get a whole post of its own someday.
I’m still a House and DnB guy in the main, but I’m always game to toss in an amapiano mix discovery on Apple Music, which does some curation and promotion of the form, or anything eye catching from Boiler Room. Since amapiano partially derives from House this makes complete sense. Connecting with the African continent is a cherry on top.
Highly recommended.
I enjoyed a relatively recent James Bennett, erm, broadside “Litestar is worth a look”. Broadside is probably too strong a term but gives you a sense of the tone. His post discusses why one should consider an alternative Python based HTTP serving engine, Litestar, as a productive modern framework. In particular, he got in a few healthy shots at a couple of my faves FastAPI and pydantic.
You save this as app.py, run with litestar run or hand it directly to the ASGI server of your choice, and it launches a web application. You go to /greet?name=Bob and it replies “Hi, Bob!”. Leave out the name parameter and it responds with an HTTP 400 telling you the name parameter is required.
So what. Big deal. The FastAPI Evangelism Strike Force will be along soon to bury you under rocket-ship emoji while explaining that FastAPI does the same thing but a million times better. And if you’re a Java person used to Spring, or a .NET person used to ASP.NET MVC, well, there’s nothing here that’s new to you; you’ve had this style of annotation/signature-driven framework for years (and in fact one thing I like about Litestar is how often it reminds me of the good parts of those frameworks). And did anyone tell you FastAPI does this, too! 🚀🚀🚀🚀🚀🚀🚀🚀🚀
But there are a lot of things that make Litestar stand out to me in the Python world. I’m going to pick out three to talk about today, and one of them is hiding in plain sight in that simple example application.
Here’s my summarization of his three points:
The entire blog post is well worth reading and reasonably argued. Litestar won’t immediately become the first thing I reach for when building an HTTP backend. However, Bennett succeeded in provoking me to at least consider exploring Litestar for some future projects just to understand the tradeoffs and the developer experience. Robust alternatives are always good to know about. His closing graf captures the intent and outcome.
I could go on for a lot longer listing things I like about Litestar, and probably wind up way too far into my own subjective preferences, but hopefully I’ve given you enough of a realistic taste of what it offers that, next time you’re about to build a Python web app, you might decide to reach for 💡⭐ to carry you to the moon 🚀🚀🚀.
Link parkin’, just because they’re so beautiful, zsa cards
The tagline is “A deck of inspiration and connection.” I’ve bought both the Original and Premium versions and have a deck within easy reach right on my desk. When I get stuck in a monotonous monthly status update Zoom call that doesn’t really need my participation, I just flip through and admire the cards, letting my mind wander.
As advertised, they are quite attractive, entrancing objects. Highly recommended.
In my previous full time gig, I did a bit of work implementing a set of APIs using FastAPI and deploying them into AWS. I needed to get a picture of API usage from external partners and AWS didn’t have anything to easily use straight out of the box. So I set about doing some dashboard development and initially thought about using Jupyter but decided to take a sidequest into marimo which had been popping up quite a bit on my podcast radar. Worked like a charm.
Also as part of my job, I built a little NLP model training platform on top of Coiled running in AWS. Highly recommend Coiled if you need to scale compute on AWS but have minimal in-house cloud and ops expertise and staffing. Especially coiled batch, which let us effectively use AWS GPU nodes. We were running super lean and didn’t have time to really drill down on all that AWS had to offer.
And now Coiled has illustrated running marimo notebooks on coiled.
My spider sense tells me marimo is having a moment and building towards escape velocity. It won’t dislodge Jupyter so much as provide a complement in the notebook ecosystem, similar to how polars complements pandas in the dataframe ecosystem.
Here’s some data points about marimo being on my radar. These are all from podcasts I subscribe to and where I consume episodes regularly:
A Listen Notes search would seem to confirm my intuition that marimo is making a push to increase visibility, possibly due to a round of venture funding at the end of last year. Podcasts aren’t the only place marimo has been popping up. The founder, Akshay Agrawal, did a PyCon US 2025 talk which is available on YouTube, and the tech features prominently in the TalkPython course, “LLM Building Blocks in Python”.
I’ve found Agrawal to be pretty engaging and thoughtful in all of these conversations. He seems to be coming from a pragmatic place of hard won experience. It’s giving me confidence that this project might have legs.
My limited experimentation with marimo has shown promising shoots although it’s a fast moving target. I really like the reactive execution design choice and the underlying usage of plain Python as the notebook storage format. They’re integrating agentic AI features, of course, but there’s interesting possibilities for agentic co-development of an interactive computational artifact along with the user. Probably worthy of doing some digging into the CS research literature for comps.
Here’s to marimo finding traction and enduring.
Link parkin’ this Stack Overflow solution for personal reference:
Homebrew v2.0 dropped all of the extra options that are not explicitly enabled in each formulae. So the —with options no longer work if you use the core Homebrew formulae.
Instead you can use a third-party repository (or “tap”) such as homebrew-ffmpeg. This tap was created in response to the removal of the options from the core formulae.
$ brew tap homebrew-ffmpeg/ffmpeg
$ brew install homebrew-ffmpeg/ffmpeg/ffmpeg --with-fdk-aac
# or
$ brew install homebrew-ffmpeg/ffmpeg/ffmpeg --with-fdk-aac --HEAD
I have a script that uses ffmpeg to convert WAV files to m4a
files for Apple Music. It needs the non-free Fraunhofer FDK AAC
to properly encode the data and write out in the correct file
format. It’s not often I use the script though and a skoosh of bitrot had
settled in to my installation with homebrew. This solution fixed
things up in a pinch.
What a momentous day!! According to my math, this blog is heading into its 17th year of existence. It all started way back when with a white plastic MacBook. That thing is still kicking as well 😲. An Intel Mac, I’ve had it happily running Ubuntu Linux for a few years. Despite only having 2 CPU cores and 4GB of RAM it still comes in handy for varied experiments.
But a downer as well. Today I got laid off from my current gig. I had a pretty good streak of picking my exits, but this came out of the blue. Won’t define me though. Just more time to really figure out what’s real and what’s BS in the agentic tool space.
Onwards to new adventures 🏴☠️ & 🥷!
Addendum: here’s the first post
Hola peeps! It’s been a minute. Time for an update.
Nothing major on the life front. In fact, a bit of summer relief from the tyranny of kid activity. I’ve actually been able to sleep in a few days. And work slogs on per usual.
On the personal tech front though, I’ve been spending some quality time seriously reworking my terminal lifestyle. The last couple of posts hint at what’s been going on. Those were just initial steps and much more buffing, waxing, polishing, and refinement were needed. Let’s dive in …
As previously mentioned, atuin has been something of a
godsend. Interacting with the bash command history is a complete
joy. There’s a minor configuration tweak, illustrated below, that I
want to mention in case it helps someone else out. I prioritized
session and directory history ahead of global for lookup. Tab
sprawl is the name of the game for me, each one like an individual
context. So session makes the right place to start even if it’s
initially empty. global is just a few C-r hits away if needed.
[search]
## The list of enabled filter modes, in order of priority.
## The "workspace" mode is skipped when not in a workspace or workspaces = false.
## Default filter mode can be overridden with the filter_mode setting.
# filters = [ "global", "host", "session", "workspace", "directory" ]
filters = [ "session", "directory", "global", "host", "workspace" ]
The above TOML goes in ~/.config/atuin/config.toml
Meanwhile, I spent some quality time working on direnv
configuration. Direnv is less of an immediate win because there’s some
effort needed to add it to existing projects and use it for initiating
new ones. Also, it works more as implicit magic than explicit
navigation. Trey Hunner provided a good starting point but I’m
molding his approach for my workflow. I’m working on a bash function
to properly setup .direnv for my pre-existing uv based Python projects. pyenv is going the way of the dodo. Viva la Frank
Wiles.
zoxide however, is going to take some getting used to. The directory teleportation mental model needs some burn in.
For the longest time, like a decade or more, I’ve been really irritated by the behavior of the bash C-r key binding. It’s supposed to be a reverse history search by default. It has some non-obvious behavior though if you decide you’ve gone too far in your history.
The fix to bash is probably straightforward and I could have got on with my life. But sometimes I’m a dope. Anyway, I’m retooling my shell life, prompted by this Frank Wiles’ post My CLI World
Enter atuin:
Atuin CLI
The magical shell history tool loved by developers worldwide. Sync your commands across machines, search everything instantly, and keep your data encrypted. Open source.
Yeah, atuin is what I needed. And being the dope I am, I’ve known about it for a while now. The cross-host synching seems cools, but the default synch server being someone else’s host put me off a little.
The initial experience is looking great and exceeding my expectations.
Yup. I really am a dope. Onwards to integrating direnv and zoxide.
Let it be known that I made my first foray into agentic AI coding, two days ago on July 6th, 2025. I worked with Claude Code to start prototyping a little tool to build m3u playlists. This will come in handy for making my local collection of music files exposable as part of my OwnTone project.
It wasn’t quite vibe coding, as I didn’t let Claude off the leash to make changes on its own. All of its requests were reviewed personally. At the same time, I haven’t written a line of code in the repo. In fact, I’ve barely looked at any of the source code. Meantime, my initial runs for building playlists look pretty good. To top it off, now a bunch of ideas for extending the tool are cooking in my head.
The experience hasn’t been mindblowingly life altering, but vaguely satisfying. This is a project I would have likely procrastinated on interminably. With an hour of effort, give or take, at least I’m started. Lots of boilerplate and mind numbing testing avoided. So there’s a path forward. And thoughts of many other deferred projects that could now be within reach. Credit to Andrew Ng, writing in The Batch, for the nudge to finally push out from the dock.
I’m completely sympathetic to all the folks who are apprehensive to pessimistic on where this is going and what it will ultimately cost. For me, guarded experimentation is the right path forward. It feels like there is some there there. Maybe (probably) not enough to match the huckster BS and snake oil, but possibly a useful normal technology. YMMV.
For the longest time, I’ve been dreaming of a hackable solution to drive my multiple Sonos speaker setup. A while back, I thought I could cobble something together based upon Music Player Daemon (mpd). Didn’t really have time to dig deep into setting things up. The other bit is that getting Sonos to pick up streams from mpd seemed a bit kludgey.
I kept plugging away doing background research and discovered OwnTone
OwnTone is a media server that lets you play audio sources such as local files, Spotify, pipe input or internet radio to AirPlay 1 and 2 receivers, Chromecast receivers, Roku Soundbridge, a browser or the server’s own sound system. Or you can listen to your music via any client that supports mp3 streaming.
You control the server via a web interface, Apple Remote, an Android remote (e.g. Retune), an MPD client, json API or DACP.
OwnTone also serves local files via the Digital Audio Access Protocol (DAAP) to iTunes (Windows), Apple Music (macOS) and Rhythmbox (Linux), and via the Roku Server Protocol (RSP) to Roku devices.
Did some holiday hunkering down and got an OwnTone server deployed on my homeLAN. Had some challenges doing a build under Linux on a mini-PC but I found a community package that worked out well. Turns out I had some Linuxbrew stuff causing conflicts. After that it was a little bit of firewall configuration aaaand …
The frontend is nice while OwnTone also provides an mpd facade that’s controllable with mpd clients like mpc. There’s also a good looking JSON API. Plus it supports LastFM integration although I’ll keep on with my soco-scribbler work. Even more beautiful is the fact that it’s reachable over my TailScale network. All in all, a side project hacker’s delight.
Some burn-in needs to happen with a full play list run, but this is promising. Given the protocols that OwnTone is designed to service (DAAP and AirPlay), it shouldn’t be a surprise it does an excellent job of finding my AirPlay 2 speakers as output targets. I also have a pair of Homepod Minis that are only lightly used. Maybe this will give them renewed life.
Today’s a good day to start a reread of Timothy Snyder’s On Tyranny. Might even complete it in one sitting.
Just have to throw a shout out to the 1Password CLI tool. It’s actually a really elegant way to plug your friendly neighborhood secrets manager into the terminal/shell life. And the integration with Apple’s Touch ID is well done. Highly recommended.
Simon Willison noted an AI related neologism “context engineering”, emerging on The Socials (TM).
The term context engineering has recently started to gain traction as a better alternative to prompt engineering. I like it. I think this one may have sticking power.
I used to scoff at the term prompt engineering. Its early connotation was one of messing around with text prompts in an ad hoc fashion and seeing what happens. Adding goofy stuff like “my grandmother’s life depends on you getting this answer correct.” Not much engineering involved.
Given the explosion in the maximum size of context windows and how they’re constructed impacts LLM performance, I’m now in the camp of prompt engineering is serious business. In my head, I was noodling with a term “context programming” to express the many ways that crafting contexts for LLMs have arisen in engineering practice. C. f. DSPy. There’s also a bit of science emerging around where and how models interact with large contexts.
If “context engineering” catches on, I’m good with it.
Link parkin’: platformdirs
platformdirs is a library to determine platform-specific system directories. This includes directories where to place cache files, user data, configuration, etc.
A handy module for use within other projects.
TIL Inspect
An open-source framework for large language model evaluations
Looks like a nice piece of open source kit from the UK government’s AI Security Institute
A big part of the day job is LLM evaluation so this is definitely of interest.
I forked sonos-lastfm just to noodle around with Sonos monitoring generally. Dubbed the project soco-scribbler as it’ll scribble down locally what’s been played for later usage. Goal one is to record plays to sqlite via sqlite-utils. Then onwards to reconciliation with last.fm assuming there’s a parallel scrobbler going, like the hit or miss Sonos plugin. Afterwards, who knows.
Forza.
Since I just put the chat link feature of Gemini, thought it might be a good idea to ask how long the links last. And who better to query then Gemini!
Public links to Gemini chats generally stay alive as long as the associated chat is saved in your Gemini Apps Activity. …
In summary, the lifespan of a public Gemini chat link is tied to how long you keep the original chat in your Gemini Apps Activity, unless you choose to delete the link specifically.
If I’d have figured this out earlier, I’d have used a Gemini link in my discussion of mixes and streaming platforms. Which reminds me to check in and see how to do this with ChatGPT and Claude.
Speaking of Beats In Space, this is a trove of mix history that I need to dive into:
Beats In Space is a radio show that started in 1999 on WNYU 89.1FM in New York City and is now broadcasting every week on Apple Music.
Beats In Space is brought to you by New York DJ Tim Sweeney.
Not only does Sweeney have a guest mix, often he also does a mix, and an interview with the guest dj:
I’m excited to share that all Beats In Space interviews are now available for FREE on Apple Podcasts! Whether you’re a long-time listener or just discovering the show, you can now dive into in-depth conversations with some of the most influential DJs and producers in electronic music—no paywall, just pure music talk. Subscribe and leave us a comment!
Great backgrounder at the fifteen year mark on Sweeney and the show. Now currently at twenty five plus years and counting 😮, with a only a single hiatus!
Previously I have spoken about the notion of a retrocast for spoken word podcasts, but the same could be done for music podcasts / mix series, tracing their evolution over time. I chatted up Gemini about this during some background research on Carlita and asked “Are there any such tools that look at longitudinal track trends, DJ relations and history, and/or track genealogy”. There’s a whole bunch more to the thread but this kicks off a lot of intellectual provocation.
You’re asking about a fascinating and cutting-edge area of AI in music, moving beyond simple track identification to deeper, more interconnected analysis. While the precise, all-in-one tools for “longitudinal track trends, DJ relations and history, and/or track genealogy” aren’t yet widely available as off-the-shelf consumer products, the underlying AI and research are certainly advancing in these directions.
Mixes, episodes, data integration and AI 🤔. Feels like a great rabbit hole to get stuck into.
I’ve recently fallen down a rabbit hole of adding DJ mix sets to my Apple Music library. Boiler Room and Fabric have been on the radar now for a year or two, with solid presences in the Apple Music service. For whatever reason, one day recently I started chasing the “More By …” and “You Might Also Like…” connections underneath a set’s tracklist and discovered:
The nice thing is that I’m coming across new to me DJs like Carlita, TSHA, Seth Troxler, Skream, Chase & Status, Sonny Fodera, Denis Sulta, and Kilimanjaro. Meantime there’s a heaping helping of old friends with mix sets I’ve never heard before like Josh Wink, Todd Terry, MK, Roger Sanchez, Armand van Helden, Lil’ Louie Vega, Masters at Work, Marshall Jefferson and Basement Jaxx.
Some of these sets are bananas in terms of length, including a 5 hour ride with Seth Troxler B2B Skream.
Out of curiosity I asked the major AI bots their opinion to compare and contrast Spotify with Apple Music on this front. Surely Spotify must be in the arena here? Apparently that’s not really the case. Here’s what Claude had to say:
Bottom Line
Apple Music is the clear winner for electronic DJ mixes, offering:
- Professional-grade content: Thousands of expertly crafted continuous DJ mixes
- Industry partnerships: Direct integration with major DJ software and hardware
- Regular updates: Monthly curated mixes and exclusive content from top DJs
- Variety: Comprehensive coverage of electronic music genres
Spotify falls short with mainly user-generated playlist collections rather than true DJ mixes, limited professional content, and no current support for DJ software integration.
For serious electronic music fans and DJs, Apple Music provides a significantly richer and more varied selection of actual DJ mixes compared to Spotify’s more basic offerings.
I need to do some further work to verify they’re all not hallucinating but this matches my general vibe. Full disclosure, I ditched a Spotify premium subscription a few years back since I got sucked into the Apple One bundle.
What really grabs me about Apple Music is the relation with curators, promoters, and labels such as The Warehouse Project, Boiler Room, Fabric, and Defected. Plus they have their own series Beats in Space and provide dj mix themed genres, although you have to dig a little. This provides some serendipitous discovery without needing to go completely algorithmic. Great folks like Boiler Room just putting up phenomenal events on Apple Music and the archives live on in perpetuity for someone like me to fall into.
Have to acknowledge the creator focused platforms Mixcloud and SoundCloud. They’re both great sources for sets directly from DJs. Bonus, they integrate nicely with the Sonos platform. They just don’t provide personal library of tracks that Apple Music does. He said grudgingly.
Now if only Apple would do something about their janky desktop app.
Happened to be looking at my profile on Last.fm and noticed the following
In case you can’t make out the text from the screen capture, there’s a “scrobbling since June 5 2005” hiding in there.
Hang on! I’ve been scrobbling for 20 years 😱 ?!
Just for calibration, that’s one year longer than The Site Formerly Known as Twitter has existed, and one year less than TheFacebook.
Some good ’ole Web 2.0 services just keep hanging in there.
Boy am I a dope.
I’ve been fiending for search capabilities on this here ’blog for
ages. Never got around to putting something together dreading grinding
through HTML and JavaScript coding. So I just limped along with
site:crossjam.net modifiers in search engines.
About this time last year, I posted about pagefind. pagefind is a JavaScript toolkit for creating embedded search indexes to go along with static server pages. Looked easy peasy to run and use. Lame old me still dragged his feet.
Finally for some Friday hacking, I decided to push through and see if I could get pagefind based search working in an experimental fashion.
After reacquainting myself with the pagefind docs, I had it working in an hour 🙄. Still have a few bits to polish, but once it’s working there will be an honest to gosh search page
Being a Systems guy, I don’t really know much about ML and AI. In my attempt to keep up, I’m trying to draft off the best that I know of. Andrew Ng is a legend, so I’m subscribed to his newsletter “The Batch”. The most recent edition, entitled Meet The New Breed of GenAI Application Engineers neatly summarizes a lot of what’s going on beyond the various echo chambers:
There’s a new breed of GenAI Application Engineers who can build more-powerful applications faster than was possible before, thanks to generative AI. Individuals who can play this role are highly sought-after by businesses, but the job description is still coming into focus. Let me describe their key skills, as well as the sorts of interview questions I use to identify them.
Skilled GenAI Application Engineers meet two primary criteria: (i) They are able to use the new AI building blocks to quickly build powerful applications. (ii) They are able to use AI assistance to carry out rapid engineering, building software systems in dramatically less time than was possible before. In addition, good product/design instincts are a significant bonus.
About two years ago, I would have been one to scoff at the notion of AI Engineers, thinking them as overblown prompt tweakers and API cobblers. I’ve come around because of two major shifts.
First, systematically, robustly, and safely building capabilities on top of LLMs really does require engineering. We’ve moved past the point of just bolting a chat window into an app and calling it a day. Compound AI systems (Go Bears!) with agentic components are a whole new ballgame.
Second, the shape and quality of system building is changing. The baseline velocity of software development is going up. Vibe coding may lead to crap applications but there are enough folks doing the required experimentation to lead to better outcomes. Still have concerns about the security, ops, and maintenance aspects but those become worthy avenues to provide value.
If you click through on Ng’s newsletter, that list of AI building blocks is daunting. Good taste on which of them to early adopt, lag, and deep dive will be another defining skill. As always, those who can discern the enduring qualities will be well positioned.
I have a Synology NAS and a few Sonos devices. Pulling these two nuggets of information together in case someone else needs to rework their setup. Previously I had all my music data on one machine, but since the NAS has plenty of space, it made sense to migrate the Apple Music library and then point Sonos in the right direction.
First you need to create a new library Apple Music
You can have more than one music library in Music. For example, you could have a library of holiday music that wouldn’t appear in Music the rest of the year. Or you could keep some of your music in a library on an external storage device.
Then these Sonos instructions illustrate how to get your old data over to the NAS library.
Choose in Apple Music under the File menu the ‘Import’ option. Select the folder where your music files are stored on your local drive. The Apple Music app will now copy all the files in the selected folder and subfolders to the NAS and update its music library. Bear in mind that all files will be copied to the NAS but not be moved, so the initial Apple Music library and local music folders will remain without change on your local drive.
This definitely can take a while. Unfortunately, Apple Music goes through a few different phases during import, only has minimal messages, and definitely no progress reporting. Oh well.
Link parkin’: OpenRouter
From the OpenRouter docs
OpenRouter provides a unified API that gives you access to hundreds of AI models through a single endpoint, while automatically handling fallbacks and selecting the most cost-effective options. Get started with just a few lines of code using your preferred SDK or framework.
I actually have an API key already, just need to get into some mischief with it.
Good podcast episode with the CEO of OpenRouter, Alex Atallah.
TIL, NASA has an API:
Welcome to the NASA API portal. The objective of this site is to make NASA data, including imagery, eminently accessible to application developers. This catalog focuses on broadly useful and user friendly APIs and does not hold every NASA API.
Requires obtaining an API key, which I assume is pretty straightforward. Wonder if its image search is good for querying Webb and Hubble telescope imagery.
Via the Pybites community forum
Link parkin’ a GitHub gist, because it leads to some fun pure local configuration and I can’t quite find anything similar in a web page or blog post.
The rub is that you can configure macOS DNS to direct resolution of domains by
doing a bit of file tweaking in /etc/resolver. For example, .test
by adding an entry in /etc/resolver/test that points to 127.0.0.1
for a DNS nameserver. Then if you install and
configure dnsmasqn to listen for DNS requests on your machine, it can take over the resolution of
a domain and point entries to services anywhere, but your machine in
particular. This includes wildcarding subdomains. A tricky bit is
closely reading the dnsmasq docs on the --address option and knowing
you can use that in a dnsmasq.conf file.
What’s this good for? I used it when working with Dokku which
allocates subdomains under a root domain for services it dynamically
builds and creates. Configured to use .test as
its root, all the services resolved nicely thanks to dnsmasq
pointing back to my laptop. This makes local experimentation and
demonstration of “push to deploy”, PaaS services easy, fun, and
portable. YMMV.
Of note, there’s a new(ish) PaaS on the block that I’d like to try out: coolify
Have to take a moment to recommend the MLX project.
MLX is a NumPy-like array framework designed for efficient and flexible machine learning on Apple silicon, brought to you by Apple machine learning research.
The Python API closely follows NumPy with a few exceptions. MLX also has a fully featured C++ API which closely follows the Python API.
But ultimately it’s the application of MLX in the mlx-lm CLI
tool that’s really impressive. I’ve been able to use mlx-lm to run
LLM model weights from Hugging Face in easy one liners. There’s a
great Hugging Face MLX Community that works exceptionally hard to
convert the latest open weight models into the MLX format. mlx-lm
also supports using GGUF files directly which blows open the space of
available models. Based on an HF overview article here’s an example:
$ uv tool install mlx-lm
$ uv tool install 'huggingface_hub[cli]'
#
# You'll need to
# 1) Accept terms and conditions for the Mistral model for your account on the HF site
# 2) Create an HF token in your account to login with from the cli
#
$ huggingface-cli login
...
$ mlx_lm.chat --model mistralai/Mistral-7B-Instruct-v0.2
... lots of downloading of model weights ...
[INFO] Starting chat session with mistralai/Mistral-7B-Instruct-v0.2.
The command list:
- 'q' to exit
- 'r' to reset the chat
- 'h' to display these commands
>> hello
Hello! How can I help you today? If you have any questions or need
assistance with something, feel free to ask. I'm here to help. If you
just want to chat or share some thoughts, that's fine too. Let me know
what's on your mind. :)
>> Who is your maker
I don't have a maker in the traditional sense, as I am an artificial
intelligence and don't have a physical body or a creator. I was
designed and developed by a team of engineers and computer
scientists. If you have any other questions or need help with
something, just let me know! :)
>>
>> /bye
Goodbye! It was nice chatting with you. If you have any other
questions or need assistance in the future, don't hesitate to
ask. Have a great day! :)
>> q
$
I have a 3 year old 24 Gb M2 MacBook Pro so I don’t get the best tokens per second output. Still impressive how easy it is to run a decent sized LLM with little to no development needed. There’s even a handy plugin for Simon Willison’s llm tool.
Messing around with LLMs is almost cookie cutter at this point.
Maybe later this year I should do a 100 Days of Fabric blog series. The idea would be to listen at least once to every mix from my collection of the Fabric series and post about it. Stretch goal would be to run these listens consecutively as a streak. Hard but doable.
From the data side, it would be a nice side quest to auto generate the skeleton of each post from the Discogs Data Dumps. Work with AI to develop the code including the embedding and styling for track lists. Maybe throw in an LLM generated summary and related artists section.
Another side quest would be to develop an app that could serve oEmbed content for track listings from slices of the Discogs Dataset.
Kagi, the alternative search engine and browser developer, turned three this week.
We have come a long way since introducing Kagi Search and Orion Browser as a public beta in 2022. What started as a quiet rebellion against ad-driven search engines tracking your every click became a complete dedication to building an alternative, humane web.
I signed up for Kagi Ultimate in October of last year after giving the free version a bit of a test drive and liking the results Or more accurately the distinctly customer centered UX vice Google Search. Can’t remember when it happened, but I also switched over to the Orion browser shortly thereafter, mainly to maximize Kagi Search usage but also for no-nonsense ad blocking. It’s also sort of neat that they have a search API.
Unlike many other subscriptions, when the bill for Kagi arrives I have give canceling no consideration. YMMV but I can recommend Kagi. Fingers crossed no rug pulls, third act betrayal reveals, and many more years of success.
Heck, if John Gruber is a paying customer that’s saying something.
P. S. podsearch.david-smith.org is surprisingly hard to find.
Link parkin’: podverse, from Matt D. Welsh Go Bears!
AI superpowers for your podcast.
Automatic episode transcripts, summaries, AI chat, and more. Take your podcast to the next level.
Welsh also open sourced the project and posted about it.
Podverse is a web app that brings AI superpowers to podcasts. Check out the live site at https://podverse.ai!
This project is meant as a demonstration of a full-stack webapp that leverages AI capabilities.
Of interest relative to the retrocast
Link parkin’: python-package-cookiecutter
There are many cookiecutter templates, but this one is mine and I’m sharing it with you. With it, you can quickly create a full-featured Python package designed to be managed with uv, a default typer command-line interface, optional settings using pydantic-settings and logging using my favorite logger, loguru. Best of all, testing, code quality checks, and publishing to PyPI are all baked in and ready to go.
I’ve been locked on to Simon Willison’s click-app cookiecutter,
mainly due to inertia, but this new one hits all my recent developer
buttons: uv, loguru, and typer. Also, TIL: Poe
the Poet and pydantic-settings
It’s a sort of classic Internet, spicy hot take, grenade in the hole rant (old Usenet heads will recognize the type), but there’s a lot in Thomas Ptacek’s My AI Skeptic Friends Are All Nuts that had me nodding my head. And I’m not even an AI hipster. (Although a bit of fly.io fanboy)
Tech execs are mandating LLM adoption. That’s bad strategy. But I get where they’re coming from.
Some of the smartest people I know share a bone-deep belief that AI is a fad — the next iteration of NFT mania. I’ve been reluctant to push back on them, because, well, they’re smarter than me. But their arguments are unserious, and worth confronting. Extraordinarily talented people are doing work that LLMs already do better, out of spite.
All progress on LLMs could halt today, and LLMs would remain the 2nd most important thing to happen over the course of my career.
Ptacek has the bona fides to back this up. To be sure, his focus is on coding and software development assistance, but that’s why I nodded my head so much. Can’t claim jumping in with both feet, doing a ton of vibe coding, and deploying to prod with my co-pilot. However, my gut tells me this change is on the order of impact that compilers had on programming. In 3-5 years we’ll be going, “did we really do it that way?”.
YMMV.
P.S. When I grow up, I want to write like Thomas Ptacek
Link parkin’
This script automatically scrobbles music playing on your Sonos speakers to Last.fm.
SoCo (Sonos Controller) is a Python library that allows you to control Sonos speakers programmatically.
I have four Sonos speakers and generally enjoy them. They make streaming audio from my ripped files and from Internet services easy and pleasant. The speakers were acquired a few years ago when Sonos supported last.fm scrobbling (tracing music plays).
Suffice it to say, Sonos has had a quite bad sequence of quarters from the software development perspective. Not much impact on me really. Apple Music playlist scanning has been crappy for a few years now, and since Sonos seems to rely on an outdated way to get access to that data, I’m not holding my breath. But then Sonos support for last.fm integration bit the dust.
So I’m looking forward to trying sonos-lastfm to see if it actually
works. Even if not, SoCo looks like fun just for messing around with
my speakers from Python. There’s also dreams of doing my own playlist
management and getting back to the nirvana of “album shuffle” that was
once an iTunes feature.
Here’s the kicker. My last.fm integration still seems to be working for some mysterious reason. I’m sure it’ll go poof the minute I attempt to reconfigure the speakers or tweak the service. And it has been occasionally missing some scrobbles if I don’t have the Sonos desktop app running on a machine somewhere. Whatevs.
I have an unfortunate addiction to Humble Bundle book bundles. It’s a convenient way to get a lot of tech ebook content at a reasonable price. Also, I can get graphic novels and an occasional speculative fiction package of interest, usually in DRM free formats and not through a global monopolist.
That said, for all the time I’ve been buying from Humble Bundle, I’ve only had the web interface to download my purchases. For bulk downloads the UX wasn’t great so I turned to the Humble Bundle Downloader extension for Firefox. Did the trick for quite a bit, but age and rot seem to be setting in. Plus the extension was the only reason to launch Firefox. But the downloader did have the nice feature that it would rename files to match titles listed on the download page, making the results much more human friendly.
Difficulties downloading a recent bundle put me on tilt and I cast about for a better approach.
Enter humble-cli, “The missing CLI for downloading your Humble
Bundle purchases”. humble-cli is, you guessed it, a command line
interface tool for interacting with your Humble Bundle purchases. It
is sweet. Here’s a quick sample of how it lists purchases
crossjam@burningchrome:~$ ~/.cargo/bin/humble-cli list
...
a7nAf3UH3Vru8d3n | Humble Tech Book Bundle: Computer Science the Fun Way by No Starch | 1.42 GiB | -
Ay3VN5rcSKrmt3d4 | Humble Tech Book Bundle: Data Engineering and Management by Pragmatic | 462.04 MiB | -
sRs4EtvxGf4hhDhb | Humble Tech Book Bundle: DevOps 2025 by O'Reilly | 348.66 MiB | -
bXPKrycBCSqbkhE7 | Humble Comics Bundle: Warren Ellis by Image Comics | 8.44 GiB | -
tDH8P68hkVWFnvNU | Humble Tech Book Bundle: Machine Learning, AI, and Bots by O'Reilly 2025 | 1.07 GiB | -
nT68upU5Kuhwxp2Y | Brian Michael Bendis' Jinxworld Bundle | 16.96 GiB | -
crossjam@burningchrome:~$
The tool is a Rust binary that can be easily installed via cargo
install. There’s a minor bit of finagling around in browser cookie
storage to go through an auth sequence which could be smoothed out,
but I’m not really complaining. Quite performant for downloading too
while having a similar file naming approach as the Firefox extension.
So far, so good. I’m looking forward to putting humble-cli through
further paces and reorganizing my collection.
NGC 2283. Just because:
Image Description: A spiral galaxy seen close up and almost face on. It is filled with puffy, patchy clouds of hot gas and dust. Red, orange and yellow colors indicate light emitted by different particles. The brightest colors are in the center and along the two spiral arms, which wind out from the center. Star clusters hide in the gas along the arms. A few large, bright white stars are prominent in the foreground, near to us.
As Simon Willison points out “Vibe coding is having a moment.”. Here’s a couple of related random thoughts. Just thinking out loud.
Does it make any sense for an open source project to craft issues that are highly amenable to vibe solutions? These could be a means of addressing low stakes work. Alternatively, a really gentle on ramp for new contributors.
Coding for generative art is a really low stakes form of development. Wonder how well vibe coding works for an environment like processing, Cinder, or openFrameworks.
Speaking of embedding and Flickr, here’s a quick demo using a galactic image of NGC 602 that I was hipped to by the Space Telescope Advent Calendar
© 2008-2025 C. Ross Jam. Built using Pelican. Theme based upon Giulio Fidente’s original svbhack, and slightly modified by crossjam.