It appears like pawn actions are going away and gameplay tasks (and the derived AI tasks) are going away. Not to be confused with behavior tree tasks, which are entirely different.
So gameplay tasks (and derived classes) are there to allow a sort of “scripting” of object logic control and resource usage. The idea seems reasonably good, it allows locking of resources such that abilities and such can be queried so that specific tasks can be enabled/disabled. So for instance, you don’t have the resource to “ride around on bike” unless you are actually ON a bike. So as you get on a bike, you’d add the ability to ride the bike. Then if you wanted something to ride, you’d add a task “ride to location” or whatever which would lock the ability resource and perform the move.
The general idea is that designers can simply script action by stringing together the correct series of tasks. But that AI (and other systems) could deal with the same systems the same way. Considering the AI entity as a set of resources that can be acquired is a pretty common pattern for AI guys.
Whats missing, well apart from lots of example functionality. Is some decent docs on exactly HOW this system is to be executed for example cases. More specifically, as is usual with Unreal Engine, I’d love to avoid the “you didn’t hook this function so now none of your code will work properly” that usually comes with undocumented UE4 systems.
Anyone actually gotten to grips with this task system yet? I’m currently using pawn actions and I’m about to add some ladder usage into my path following so that our AI guys can get off rooftops (or go to them to snipe) and I’m wondering if I should implement it as a AITask so I start moving my AI command functionality to them.
What say you guys?
.Zoom.