I sort of did something similar for my Darts app for the iPhone, specifically for defining the throw model for the "AI" player.
I found it easiest to simulate millions of random dart throws parameterized by throw angle and velocity. I record the slot on the board (1-20, Bull) hit given the launch angles and force of the throw. I let this "simulation" run until I had 50 samples per board slot.
The simulation is run once ahead of time. The simulation results are bundled with the app.
When it becomes the AI's turn, it picks a slot to aim at based on the game variation and state of the game. Currently, it chooses rather poorly in Cricket.
Then it picks a random sample from the sample pool recorded from the aforementioned simulation. Based on the user-chosen competence level, it adds a small, random amount of error to the angle and velocity associated with the selected sample. Finally, the dart is launched using the calculated angles and force.
While the app is Objective C, the offline simulation was written in Python. The whole thing was a blast to write.
What is your take on CouchDB? After some initial investigation, I've all but written off CouchDB for anything moderately write-heavy due to its single writer at a time design. I'm by no means an expert on that however. Please tell me what I'm missing about CouchDB.
This reminds me of a funny Cracked skit about the 1979 game developer's conference wherein the birth of Frogger was reenacted. "Ok ok you are a truck driver and you try to run over frogs crossing the road". "No, no, you ARE the frog!" "Whoooa"
"Meanwhile, you can use your application to do the authenticating, and generate ids for already-authenticated clients from a large enough (> 2^128) keyspace for security."
How would you go about restricting access to channels to a specific user or users? My first take would be to issue an unguessable id upon successful authentication and use this id as the channel id -- a shared secret. Only those who should know the id do know it.
I don't see support for streaming; the connection to the long-polling client is closed after consuming. Are you planning on adding support for streaming new messages as they are published?
I found it easiest to simulate millions of random dart throws parameterized by throw angle and velocity. I record the slot on the board (1-20, Bull) hit given the launch angles and force of the throw. I let this "simulation" run until I had 50 samples per board slot.
The simulation is run once ahead of time. The simulation results are bundled with the app.
When it becomes the AI's turn, it picks a slot to aim at based on the game variation and state of the game. Currently, it chooses rather poorly in Cricket.
Then it picks a random sample from the sample pool recorded from the aforementioned simulation. Based on the user-chosen competence level, it adds a small, random amount of error to the angle and velocity associated with the selected sample. Finally, the dart is launched using the calculated angles and force.
While the app is Objective C, the offline simulation was written in Python. The whole thing was a blast to write.
Here's an old video of gameplay:
http://vimeo.com/2010432
Skip ahead to around 1:10 to see the AI take its turn.