question

Hi,

It seems the type of a ternary expression a ? b : c is sometimes determined incorrectly. Consider the following four expressions:

true ? 1 : 0.5   // (1)
true ? 0.5 : 1   // (2)
false ? 1 : 0.5  // (3)
false ? 0.5 : 1  // (4)

I'd expect (1) and (4) to be equivalent and evaluate to 1, and (2) and (3) to be equivalent with a value of 0.5.

That expectation is mostly met, except by number (3): apparently ( false ? 1 : 0.5 ) == 0. It seems the type of the result is determined by the "true" branch: false ? 1.0 : 0.5 does indeed return 0.5.

Is this a bug or just weird? :-)