You know what, I used to plan my leaving from home based on the timings at the station, but soon I realized that it is not worth it. It is not because trains are not sticking to the time table. Just randomly starting at your own comfort eliminates the anxiety that comes with planning. Your average wait time might increase to half of the interval between the trains, but that would be an increase of only a few minutes for mornings, in return for never bothering to check time again.
Thank you for the inspiration from your nice and simple real-time API. I made a pass a few years ago on digesting similar GTFS data and you've made me realise how much simpler it could be! :-)
Those are the projects I love and get inspired by. I love the execution and the level of detail, making it feel like a true miniature signage on the station.
Nice execution, I think you nailed the vibe. Nice find on that display, it’s awesome!
If you wanted to get rid of your middleware and maybe pick up some insight, one of the things that SOTA LLMs are really good at is translating code from one language into another.
The ESP has plenty of moxie to handle the API work, so you could try translating it for the ESP, then you could drop the weight of your middleware service. I use LLMs that way when I feel roadblocked (usually laziness more than anything lol) and I’m often surprised at how much I learn from the implementation.
Worse, predatory AI companies mean that vendors like DigiKey, Mouser, Farnell/Newark/Element14, and McMaster-Carr have to hide their sites behind anti-bot services. In practical terms that means you can expect to have to "click and hold" some stupid button for upwards of fifteen seconds just to access a page on the DigiKey site. Or maybe you'll just be flat out denied access to Farnell's catalog because you don't seem human enough.
Externalizing the costs of your cute little short cut tools has very real negative consequences for the maker community.
Cool project. Everyone else has made good comments. If I could add a little criticism I felt your "and then I added a computer because I didn't want to write the ESP32 code for interacting with the API" did substantially change the character of the project and felt a bit like a rug-pull vs the promise of the first sections of the post.
22 comments
[ 2.9 ms ] story [ 41.5 ms ] threadhttps://www.bart.gov/news/articles/2009/news20090309
You know what, I used to plan my leaving from home based on the timings at the station, but soon I realized that it is not worth it. It is not because trains are not sticking to the time table. Just randomly starting at your own comfort eliminates the anxiety that comes with planning. Your average wait time might increase to half of the interval between the trains, but that would be an increase of only a few minutes for mornings, in return for never bothering to check time again.
Well done and what a lovely spirit.
If you wanted to get rid of your middleware and maybe pick up some insight, one of the things that SOTA LLMs are really good at is translating code from one language into another.
The ESP has plenty of moxie to handle the API work, so you could try translating it for the ESP, then you could drop the weight of your middleware service. I use LLMs that way when I feel roadblocked (usually laziness more than anything lol) and I’m often surprised at how much I learn from the implementation.
Just an idea, it’s fine as it is.
Worse, predatory AI companies mean that vendors like DigiKey, Mouser, Farnell/Newark/Element14, and McMaster-Carr have to hide their sites behind anti-bot services. In practical terms that means you can expect to have to "click and hold" some stupid button for upwards of fifteen seconds just to access a page on the DigiKey site. Or maybe you'll just be flat out denied access to Farnell's catalog because you don't seem human enough.
Externalizing the costs of your cute little short cut tools has very real negative consequences for the maker community.
It runs on an ESP32-S3 using the government provided open data. https://opentransportdata.swiss
https://www.bart.gov/news/articles/2009/news20090309