It is really critical to have the ability to simulate test data for your programs. Some people rely on production data for their testing which has many undesirable ramifications. Having the ability to simulate data is a big advantage as you can create data that is not always encountered in testing. When you use production data, your testing is limited to what you have seen so far and no guarantee that your code can correctly handle all possible data conditions.

To simulate a complex set of data, you compose simple building blocks. Recently, I needed to test the performance of an Android app. While random lines and circles are easy enogh to generate, polygons with interior holes were a little tricky. Here is my approach to create random rectangles with a single interior hole. You can tweak some of the magic numbers to control the size of the polygon.

FOO #+BEGIN_SRC js /**

• Create a random rectangle around specified point.
• Steps:
• - set number of sides to a minimum of 3 and maximum of 12.
• - calculate theta (in radians) to 360/sides. We can imagine
• lines radiating at these angles from pt
• - for each side, pick a random point along the radiating lines
• - for the inner polygon, offset the above point.

*/ mk_{randompolygon}(pt) { var x = pt.longitude, y = pt.latitude; var sides = 3 + Math.floor(Math.random() * 12); var theta = 2 * Math.PI / sides; var pts = [], hole = []; var min = 0.00009; var max = 0.00002; for (var i = 0; i < sides; i++) { var rad = min + Math.random() * (max - min); var dx = rad * Math.sin(i*theta); var dy = rad * Math.cos(i*theta); pts.push({longitude: x + dx, latitude: y + dy}); dx = (1.3 * rad) * Math.sin(i*theta); dy = (1.3 * rad) * Math.cos(i*theta); hole.push({longitude: x + dx, latitude: y + dy}); } return [pts, hole.reverse()]; } #+END_SEC