That's my point.Not really, as each calculation has to be done only once per missile (for homing missiles x times with x being the amount of line segments).
>There is no way it can fail.
A bold statement for a game with teleport spells. If a missile heads straight for a unit, and just misses a linear missile along the way, and then the unit blinks back to the other side of the linear missile, it will be chance whether or not the homing missile hits the linear missile, based on whether or not it recalculates right then.
It's all a bit silly. The linear method actually assumes that nothing interacts with missiles. I've made a map that speeds up/slows down/freezes/pauses missiles in AoE's before. That would screw up linear missiles trajectory paths completely. Furthermore, they were all homing missiles. There is no doubt this would be a periodical O(n^2) solution! It is checking each missile against each other missile (without checking the reverse, so yes, n(n-1), which is still O(n^2)). Worse yet, the calculations are far more intensive than just range checks, everything needs to be recalculated at once when a spell interacts with them (lag!) and it still isn't as accurate as simple range checks.
On the other hand, that's an amazing idea for a map that only needs to handle the collision of linear projectiles.