Page 4 of 4 FirstFirst ... 2 3 4
Results 31 to 36 of 36

Thread: anyone managed to use CreateHTTPRequest?

  1. #31
    Valve Developer
    Join Date
    Sep 2011
    Posts
    1,619
    We're totally cool with experiments with ML approaches, but for security reasons we only allow server-side bot scripting do be done with our lua interface. If we can maintain security, we're amenable to opening up the interface to better support ML back-ends.

  2. #32
    Basic Member
    Join Date
    Dec 2016
    Posts
    557
    Quote Originally Posted by ChrisC View Post
    We're totally cool with experiments with ML approaches, but for security reasons we only allow server-side bot scripting do be done with our lua interface. If we can maintain security, we're amenable to opening up the interface to better support ML back-ends.
    I think what you have is sufficient, b/c even if someone writes a good distributed RL algorithm and needs users at scale to help train it a webserver relay/proxy can be provided that that talks to the final server that aggregates all the test data and results. It is then on the people using that script if they want to partake and share their data with the group or not.

    I mean it wouldn't run straight out of the Workshop b/c it would require a separate setup of a webserver relay, but it's not hard (as I have shown) to create a python webserver which you could then do anything you want with. And the action of doing the step of setting up the webserver would suffice as consent I believe.

  3. #33
    Basic Member
    Join Date
    Dec 2016
    Posts
    557
    @ChrisC - thank you, confirming that CreateHTTPRequest() now properly uses the callback using the code I previously posted in this thread

    POST:
    Code:
    [VScript] {"data":{"MaxHealth": 640,"Health": 640}}
    REPLY:
    Code:
    [VScript] Request : table: 0x0022e168
    [VScript] Body : {"operation": "POST", "result": "success"}
    [VScript] StatusCode : 200

  4. #34
    Basic Member
    Join Date
    Dec 2016
    Posts
    557
    Also, I tested that I can specify a port of my choice (and not rely on the default port 80) and it works.

  5. #35
    Basic Member
    Join Date
    Dec 2016
    Posts
    557
    @ChrisC - thank you for including game/dkjson.lua

    I was dreading having to write my own JSON decoder for the replies from my webserver, low and behold you provided it.

    This is more of a PSA for other people that want to have JSON encoded bidirectional communication.

  6. #36
    Basic Member axetion's Avatar
    Join Date
    Jan 2017
    Posts
    4
    Quote Originally Posted by ChrisC View Post
    We're totally cool with experiments with ML approaches, but for security reasons we only allow server-side bot scripting do be done with our lua interface. If we can maintain security, we're amenable to opening up the interface to better support ML back-ends.
    It would be really nice if you could find a way to let us safely load Torch (and its NN framework). I'm currently nearing completion of a bot that uses it extensively, and it really sucks from both a technical and UX standpoint to have to use an HTTP server. Just for giggles I tried building it and linking it against libvscript which worked but obviously the sandbox doesn't let you then import it since it's a native library.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •