As an update to this, the next beta build will include a slider bar (labeled "Custom Latency Tolerance") that lets you tune how sensitive the ability queue is (including sliding it all the way to zero if you choose). This value allows you to control (in milliseconds) how long before the GCD finishes the client will let you queue your next ability.
The optimal setting will probably be right around what your ping time is. So, if your ping is 100 for example, you'd probably want to set it at 100 or slightly higher.
Does that mean that the plans to add in the ability to replace a previous command is out? or is that still a possibility?
That's still in too. It's just that while overriding is helpful it can still fail if your override gets to the server a little too late and your previously queued ability goes off anyways.
why cant you just make the client dynamically change this setting based on the ping rather than make the player choose it and probably get it wrong?
We plan to do this as the default value (when you aren't overriding it) in a future patch. There are some complexities that prevent us from being able to do it right now.
How are you going to address people using the TCPAckFrequency registry hack to "lower" their ping? Are you going to disable it in-game, similar to how you handled the nagle algorithm?
The server now enforces the actual global cooldown value, whereas in the past it allowed a .4 sec slush. So, instead of having a slush factor that let somebody cast a bit early, it now catches the request, waits until the cooldown has actually completed, then casts the spell for you (the queue'd ability) at the soonest possible time.
