Paint Attack

[TechSY730]

Have you guys figured out a better algorithm for the paint attack? I remember you mentioning that the algorithm used in the last SlimDX beta was an O(n*m) algorithm, which is of course impractical for 1000 units on 1000 unit type situations (as is quite possible in the late game).

[x4000]

We haven't been working on it as a priority yet; it's something we'll address 4.0.  We know how to send the data more efficiently, although it's still one of those features that I think is just begging for trouble.  But, hey.

[TechSY730]

No worries, it is a pretty minor thing. Can't you use some sort of clever "pooling and copying" of orders to help things out, like you do with target selection in FRD mode?

[keith.lamothe]

I already know the approach I want to use for it, I described it in the thread where I said the previous implementation wouldn't work; it's just a matter of having time.