cool. for those who are interested, you can actually use code to control the seq like this example in Glicol:

https://glicol.org/demo#minitechno

I really doubt this marketing approach is effective. Isn't this just shooting themselves in the foot? My actual experience with Cursor has been: their design is excellent and the UX is great—it handles frontend work reasonably well. But as soon as you go deeper, it becomes very prone to serious bugs. While the addition of Claude's new models has helped somewhat, the results are still not as good as Google's Antigravity (despite its poor UX and numerous bugs). What's worse, even with this much-hyped Claude model, you can easily blow through the $20 subscription limit in just a few days. Maybe they're betting on models becoming 10x better and 10x cheaper, but that seems unlikely to happen anytime soon.

Hitting my head into buggy apps made by these AI companies and seeing them all be amazed in parallel that skills/MCP would be necessary for real work has me pretty relaxed about ‘our jobs’.

OpenAIs business-model floundering, degenerating inline to ads soon (lol), shows what can be done with infini-LLM, infini-capital, and all the smarts & connections on Earth… broadly speaking, I think the geniuses at Google who invented a lot of this shizz understand it and were leveraging it appropriately before ChatGPT blew up.

We use mcp at work. Due to some typo the model ran absolutely random queries on our database most of the cases. We had initially kept ot open ended but after that, we wrote custom tools that took an input, gave an output and that was strictly mentioned in the prompt. Only then did it work fine.

I can't comment on this matter because I don't know the details. However, based on my personal experience consuming Adafruit products and their generous open-source approach, I personally trust Adafruit very much.

Meh wish it was that simple, this paints at least part of their leadership as completely unhinged: https://chaos.social/@North/115605819126197877

We're only probably seeing part of the whole mess though.

Damn that's a considerable body of work.

I'm rewriting glicol (https://glicol.org/) with no std, and embassy-rs + 2350 is my go-to choice. Highly recommand this stack if you're planning to start working with embedded systems in 2026.

Happy new year every one.

I was about to check out the GitHub activity summary like last year, but then I realized that a lot of things weren't actually pushed to GitHub this year. So I used Vibe-coded to take a look:

https://github.com/chaosprint/hindsight

I started this year with coding and finished it with coding. In between, I was preparing for and caring for my newborn.

Hope I can release something interesting in 2026.

If you don’t wanna use a daw on linux you are welcomed to try https://github.com/glicol/glicol-cli

Sure, as soon as you tell me how I'm supposed to record and mix a whole band with instruments with Glicol

Relevant to this discussion - my project Glicol (https://glicol.org) addresses this space. Currently working on a no_std rewrite, demo coming next year :)

Curious, would the rewrite allow for building on hardware platforms such as the Daisy?

Or maybe it is already possible, to be fair I haven't looked closely.

https://daisy.audio/hardware/, https://github.com/electro-smith/libDaisy

of course I already got a poc a while ago:

https://github.com/chaosprint/daisy-rust-playground

but for now my main mcu is rp2350

Awesome, will check it out!

I never use Arduino or Arduino IDE anyway; it's incredibly laggy for me, and I hate having these things in the cloud. I mainly use Pico and VS Code now.

Setting up the toolchain that's not Arduino IDE is a prohibitively high bar for a school child that wants to blink leds.

There is a version of Thonny[1] designed for use with the Pico that is great for education. Raspberry Pi have some good resources on getting started[2].

If your target audience is school kids, you really can't go past the micro:bit and Makecode[3].

1. https://thonny.org

2. https://projects.raspberrypi.org/en/projects/getting-started...

3. https://makecode.microbit.org

The Micro:bit Educational Foundation also make a web-based Python Editor at https://python.microbit.org which is designed to be a supportive introduction to text-based coding and physical computing with no installation, friendly error messages and device simulation

Yeah their IDE is basically unusable.

And yet that is the only thing, that they actually produce. Everything is else is from someone else.

I am endlessly thankful for the Arduino project as it was one of the major gateways to programming for me, but at the same time, I bought an Arduino R4 and have barely even used it. ESP32, Raspberry Pi, and even 8 bit Atmel chips get way more attention from me. I'm guessing that Renesas chip on the R4 won't be getting too much attention anymore.

And if you poke a bit around, not even that was really of their own making: http://arduinohistory.github.io/ . There also has been some discussion about that here on HN, e.g. https://news.ycombinator.com/item?id=46009209, which was less than a month ago.

Some simple thoughts:

I feel that one challenge of programming languages is how to remember these rules, formats, and keywords. Even if you're using familiar formats like YAML or JSON, how do you match keywords?

When developing Glicol (http://glicol.org/), I found that if it's based on an audio graph, all node inputs and outputs are all signals, which at least reduces the matching problems. The remaining challenge is ensuring that reference documentation is available at the minimal cost.

SWE's results were actually very close, but they used a poor marketing visualization. I know this isn't a research paper, but for Anthropic, I expect more.

They should've used an error rate instead of the pass rate. Then it'll get the same visual appeal without cheating.