Ask HN: What are you working on (September 2024)?

Working on an extension to my recently purchased outdoor AirGradient unit to add an atmospheric pressure sensor, a second temperature/humidity sensor, some connections to an external rain gauge and a way to power the unit without going through USB.

Many of the sensors and connections are small enough that I could have spun another PCB to replace the VOC/NOx module it comes with[1], but SGP41 ain't cheap & I wouldn't dare to desolder one from the existing module. So instead I'm going to try to use the extension I/O connector AG board has. Am currently waiting for my PCBs to arrive.

Speaking of PCBs. It is wonderful that it is possible to get 5 units of a prototype for a price of a coffee or two.

[1]: https://www.airgradient.com/shop/#!/SGP41-TVOC-NOx-Module/p/...

React Email Generator: https://reactemailgenerator.vercel.app

Just write a prompt and get the perfect email template for your use case.

Building a little CLI tool that stores API requests and spins up a small REST server that allows them to be pinged. This stemmed from being on a flight, not wanting to buy WiFi, or the WiFi being slow and I just want to build something around an API.

I called it gofaux: https://github.com/tjb/gofaux

I'm still pretty early in terms of figuring out if this approach is better for larger projects. I can confidently say it works great on small things - for small changes on small files the AST approach works pretty well. You can say things like "add a click listener to the button that calls a function to tally the user's score" to a simple game and it will do it. That is, less than a minute after typing the prompt, you will see the code update in the editor and see your preview-webview update with your changes applied.

However, I have noticed that the AST code quality heavily depends on how common it is in the training set. I think I will have to add documentation to it through RAG or something - because OpenAI's models that I'm using seem to have limited experience writing esprima for JavaScript for example.

So it's hit or miss. In some cases I do feel like I'm throwing stupid compute at solving small problems and it's unnecessary - however, as I work on the project, it is getting better and better at successfully making the modifications. Some of that is me improving the prompts, some of it is OpenAI improving the models themselves, and some of it is the infrastructure I'm building for the project itself.

I did notice a huge improvement when o1-mini released. It is dramatically better at writing the AST code than GPT-4o or 4o-mini. I haven't tried Claude 3.5 yet but I've been hearing it does an exceptional job at code writing - not sure about my AST requirements though!

I just picked up an old Sheldon Lathe. I got it for $200 and hauled it myself with a drop trailer. Its going to need a lot of work but its a beautiful old machine.