8
points
ryry
Joined 593 karma
- Super helpful points - thank you!
Definitely gonna implement some more clustering. The zooming in/out being annoying bit I hadn't thought of, but now that it's mentioned I kind of see it myself too.
I actually started with a sidebar that would then highlight the element on the map, but got rid of it to focus more on the exploration aspect.
- Ah yes! So this is something I'm struggling with from a UX perspective (could use some feedback from anyone here).
News tied to a specific place gets pinned there (like an article about the Blue Jays being pinned at the Toronto Skydome). For stuff in cities, or larger areas like states, it tries to place it within the bounds of the state randomly, and if that fails, a random radius (which I think is what's gone wrong here).
The problem though is with the Toronto Star reporting on news in Jersey. The way I'm going about this right now is a filter I'm calling "Lens" which is just the country. Right now the only lens is Canada, but I want these to be filterable so you can easily see what other places are writing, or just get home grown news (but again, something I'm struggling with how to represent it).
- 8 points
- looks like it's at least possible! https://www.mta.maryland.gov/marc-tracker https://www.transit.land/feeds/f-brightline~trails~rt
I'll add them to the list and investigate; thanks!
- I did a bit of a deep dive on this over the weekend and left feeling more confused than when I started. From what I can gather, outside of CN's holiday freight train, most of the tracking is done by community members with SDR antennas and Raspberry Pis. They report to centralized servers (which I've yet to find), and the data indicates where trains are in signalling blocks. I'm the least knowledgeable person on trains, so I don't know if this is all accurate, but that's my best understanding.
I'd love to build some sort of service that takes this data, references a DB of signaling blocks, and establishes an estimated lat/lng - but that's a huge project of its own.
- thanks!
I did a quick write up about it here: https://rydercalmdown.com/projects/trains-fyi/
The hardest part was learning about GTFS-RT, which was a data format I wasn't familiar with until now.
- 503 points
- 5 points
- Thanks! A CLI tool is an interesting idea. I originally built something similar [as a python package](https://github.com/ryderdamen/fromchaos) to basically force LLMs to return data in a specific format, so I could probably easily adapt something.
What I like about the extension is that I can use it on any content that I can see, like that behind a login page.
- So there are definitely some features I'd like to add / improve upon: - json/yaml/etc validation - better pre-filtering before the openAI api call to save money - custom select (instead of the whole page) - adjustable models - the page doesn't disappear when you close it / data caching
but before those, I wanted to launch it to see if anyone would actually use this besides me. Let me know!
- 7 points
- 3 points
So it's pretty compute intensive - my machine also doesn't love it. I wanted to get the general gist of it out there before optimizing, but I'll try my best.
For the below the horizon bit - that was the only way I could get the map to fully render when at a low tilt angle. If I enabled the stop at the ground layer, it would stop rendering the map at close distances (not fully sure why), so it sits like this for now.
The enter an address thing is a great idea, and definitely possible! I'll add it to the road map!