faddison / jewb Goto Github PK

this is a click strategy. a click pattern is a sequential set of click coordinates to execute.

-columns
-rows
-spiral
-music beat
-etc...

the image illustrates a possible scenario where a race condition could occur. how does bejewelled handle this? will the order affect the future board state?

where do I use Strategy pattern and Factory pattern?

avoid clicking the same tile twice in a row and any tile directly adjacent or diagonal to that one. just generate new points until one outside of this dimension is found

if two tiles are clicked simultaneously and one of them depends on the second being in the original location to form a combo it will not occur because the second will be in a moving state. See figures below for example.

figure 1
x x x x x x x x
x x x x x x x x
x x x x x x x x
x x x x x x x x
x x x x x x x x
a b x x x x x x
b x x x x x x x
b x x x x x x x

In figure 1 above a combo can be made by swapping tiles (0,5) and (1,5). However, this requires that tiles (0,6) and (0,7) are stationary. the swap animation take between 300-600 ms. This limits how many moves can be made per second. Computationally, a move can be made in less than 1 ms.

In order to maximize the number of legitimate moves that can be made simultaneously (or pseudo simultaneously) all surrounding tiles that could possibly be affected by the swap must be found.

figure 2
x x x x x x x x
x x x o u x x x
x x x o u x x x
x o o a b u u x
x x x o u x x x
x x x o u x x x
x x x x x x x x
x x x x x x x x

Looking at figure 2, assume that a and b were just swapped. o and u show the possible combinations that could be formed from this swap (respectively). In order to minimize interference these areas should be avoided during the next swap. However, there's multiple degrees of interference to consider. First degree interference would mean trying to swap one of the affected tiles which would prevent a combo from being recognized. Second degree interference would be swapping another tile who's combo region intersected with this combo region. As shown below in figure 3.

figure 3
x x x x x x x d
x x x o uv v v c
x x x o u x x v
x o o a b u u v
x x x o u x x x
x x x o u x x x
x x x x x x x x
x x x x x x x x

Assuming c and d were just swapped immediately after a and b. Now one of their possible combo region intersect each other at (4,1) uv. What are the implications of this? One is that if a combo from the a and b swap forms which coincides with the c and d combo formation then we simply end up with a wasted move because the first combo already executed and the replacement tiles haven't been generated yet.

If this is the only consequence then a static click pattern can be successfully generated based on this algorithm. Since we are able to execute moves much quicker than the animation allows them to resolve eventually we will need to wait. But for now it will improve the bots legitimate moves per second significantly.

ie looking at adjacent tiles for possible combos instead of brute forcing every board state possible

relates to the previous issue of race conditions. I believe these issues may eventually resolve themselves due to the sequential nature of the event processing.

for now I must determine the order of operations when shifting columns down.