Do sensors correctly detect entry/exit of bodies that are "teleported" into/out of the sensor volume or that are initially fully inside?

[Avelina9X]

I big issue I had with PhysX were Trigger volumes often failing to detect bodies when their position is directly modified, and also for bodies which are initially fully inside the volume. This lead me to either require each body to perform a geometry overlap query against the equivalent trigger volume shape every frame, or alternatively run a scene overlap query of the trigger volume against the scene. This was often a large bottleneck and made it prohibitive to work with more than a handful of triggers.

Now that I'm working in Jolt I've still maintained these bad habits due to my distrust of triggers (or now Sensors) detecting bodies that may be arbitrarily teleported arounded. And while the performance is significantly better (I'm running queries every step against a broad phase layer with 1000s of sensors at only a few microseconds of cost) this is still overkill compared to using Sensors as intended.

So my question boils down to...

Do Sensors correctly detect added/removed contacts for bodies which may be teleported into/out of the sensor by their position being directly modified (for both dynamic and kinematic bodies)? And do Sensors correct detect added contacts for a body which started inside the sensor volume before the first physics step, regardless of the order of bodies added to the system?

[jrouwe]

Yes they should

[Avelina9X]

Every day that goes by I am become more stunned both by how well Jolt works, and how poorly PhysX seems to handle things.