https://htmx.org logo
Join Discord
Powered by
# htmx-general
  • f
    Most of the compute will be running on the back end.
  • g
    Streaming simulation data from the server sounds really cool.
  • g
    Yes, then just some kind of visual updates streamed via WebSockets/SSE.
  • f
    Remember -- most of the graph will actually be running on the cloud somewhere.
  • f
    Yup.
  • f
    So it's just visual updates streamed over SSE ...
  • f
    Together with a little bit of interactivity sprinkled in.
  • g
    How frequent are your updates? That spinning turbine could be a lot of data.
  • g
    1/second?
  • g
    60/second?
  • f
    Application dependent ... but I think around 5 or 10/second is about the ballpark I'm shooting for.
  • g
    It's almost like Microsoft's xCloud -- streaming video games to your XBOX. 😄
  • f
    And video will probably be handled with a different mechanism.
  • f
    Yeah ...
  • f
    It's a lot like a game engine.
  • f
    The back end is a mix of Python and C++.
  • g
    It's a fascinating trade-off. How quickly could you download the results of the simulation, vs. doing some of the calculations on the client... because C++ should transpile into WebAssembly.
  • f
    C++ simulation kernels glued together with Python.
  • f
    You don't want to be doing the heavy lifting on the client.
  • f
    A lot of the 'simulation' is machine learning stuff ...
  • f
    So generating and consuming potentially Petabytes of data.
  • f
    But I want RAD tools to easily generate visualizations of what's going on in the browser.
  • f
    Sort of like a Jupyter notebook.
  • f
    Except not a notebook. A system model.
  • g
    Yup.
  • f
    In fact, some of the back end will be using multiple instances of (probably) Unity to generate synthetic imagery, and then using those synthetic images to train neural networks, and then evaluating the results, possibly by comparing with data coming back from an autonomous vehicle, and feeding that data back into the simulation to make it more representative of what's going on in the world.
  • g
    So, I know we're pretty far off of htmx's basic use cases. But whatever JS you need to run on your client browsers, I'd try to keep it as static as possible. Pass in arguments to call a 
    render()
    function every 1/10 of a second if you need to, but try to keep something you can depend on, instead of swapping out Javascript. I think you'll be asking for trouble if you head down that road. 🙂
  • g
    Wow. That's going to be very cool when you get it running. Promise to "show and tell" when it's ready?
  • f
    Yup!
  • f
    This is us, BTW:- http://xplain.systems/ -- just a placeholder page for the moment ... but I'll let you know when we (eventually) have something.
1...100101102...1146Latest