Turing's built in whatdotcolor function is terrible. It gives the approximate color of a pixel instead of the exact red, green, and blue values. That means you can't apply special effects to photos. Bleh.
Therefore I rewrote a new whatdotcolor function. It works by taking a picture of the screen using the takepic function, which returns an integer array containing the raw picture information, and then directly fetching the red, green, and blue values from the array. Its syntax is the exact same as the original whatdotcolor, but instead of a color number, it returns a pixel class with R, G, and B values.
Here's a demo: http://dl.dropbox.com/u/17181264/better%20what%20dot%20color.zip. It first screenloads a picture, and then repeatedly inverts the picture over and over, using up a LOT of memory in the process.
Here it is:
| code: |
class _Color export ~.*Color, var r, var g, var b, use type Color : ^_Color var r, g, b : real := 0 proc use RGB.SetColor (255, r, g, b) end use end _Color fcn newColor (r, g, b : real) : Color var n : Color new n n -> r := r n -> g := g n -> b := b result n end newColor fcn tonyswhatdotcolor (x, y : int) : Color var bleh : array 1 .. sizepic (1, 1, 2, 2) of int for i : 1 .. upper (bleh) bleh (i) := 9999 end for takepic (x, y, x + 1, y + 1, bleh) var p, q: int var w, r, g, b : real if (bleh (19) ~= 9999) then p := bleh (19) else p := bleh (21) assert false end if b := bits (p, 24 .. 31) / 256 r := bits (p, 16 .. 23) / 256 g := bits (p, 8 .. 15) / 256 w := bits (p, 0 .. 7) / 256 result newColor (r, g, b) end tonyswhatdotcolor |
and here is the demo in action:
My main critisism would be that your variable naming needs work...