Diamond-square algorithm producing incorrect results on web

[anissen]

The diamond-square sample takes a very long time (around 10 seconds) and then produces (something like) this on web (Chrome, OS X):

Running the same sample but using simply Math.random as the random function produces this interesting result:

Running the diamond-square sample on OS X natively produces this result:

Maybe some variables are uninitialized causing these platform differences?

[whuop]

Weird stuff! Will look into it very shortly. It definitely shouldn't do that on web.