I've yet to try it but did poke around the site a bit. The only thing that I thought was a bit off was the fact that the automatic type conversion doesn't behave the way most languages do but instead converts the type of the object on the right to the type of the object on the left.
That just seems foolish since you might have some code which looks like this
float2 = int + float1
That just seems foolish since you might have some code which looks like this
float2 = int + float1
Are you sure about this? You're suggesting that the following two lines would store a different value in float2?
float2 = int + float1
float2 = float1 + int
It seems bizarre to convert to an int and then back to a float. Now if you said that the following two lines resulted in a different answer, I could see the logic.
float3 = float1 + float2
int = float1 + float2
I could see the floats being cast to integers before they are added. It's perhaps not ideal, but since the answer is going to be cast to an integer anyway, I could see that it might hold an advantage for the developer to make such assumptions.
But not the former.