The indexes are correct and, I believe, that they are actually related to the triangles themselves
0, 1, 2, 1, 3, 2 will not split the square into two triangles... it will simply not split the square at all.
1 and 2 are the common points that connected give a line that both triangles splitting the square have in common.
I wasn't talking about splitting it... just that second triangle that makes upper right half of the square is drawn clockwise if you put indexes 1,2,3, and the other is drawn counter-clock wise. So at the end you see half of the square upfront and half from behind?
Kova, you are correct. The reversed direction you found in the test is incorrect unless you were intending to have a flipped triangle or were drawing strips. If you want to correct the wiki (changing the indices to 0,1,2,1,3,2 as you proposed) that would be great.