SB 4.019 - Random "New" actions not reliable

[Rotareneg]

At least at the beginning of a scenario, actions depending on testing the "New" variable aren't entirely reliable. Attached is a scenario in which this occurs along with an AAR showing the problem.

The infantry and missile teams use jump-to-end routes testing "New" to pick random positions, but sometimes they don't jump and end up trying to run to the waypoint. The vehicles jump-to-end routes check the X1 variable instead and jump to their starting positions reliably. However, the waypoints have random routes that check "New" to determine which way the vehicles go. When the mission starts some vehicles won't take any of the routes and will just sit waiting until another vehicle comes along and triggers the routes to be checked again.

 

The scenario that is labeled "working" works around the issue by having the infantry jump based on fixed variables and has all vehicles spawn after a 2 second delay.

Tanks.7z

[Rotareneg]

Just made and attached a simpler test scenario. It is a bit random (not surprising since it's testing random numbers) so you might need to run it several times to see both broken behaviors (the jump not working, and the vehicles not taking the routes.)

 

Random_test.7z

Edited July 25, 2017 by Rotareneg

[DarkAngel]

Noob. Slap yourself!!!. 

Each route has its own new variable

[DarkAngel]

You need to make one of them conditioned so it will always work. You don't need to put the different numbers in them. You just need 0 < new <25. Butt you will always need one route which it will default to.

[Rotareneg]

Does that not work? I guess I assumed it checks all the routes on a waypoint with the same newly generated number, not just generating a new number for each individual test.

[Rotareneg]

Ok, I'll try that.

[DarkAngel]

no it generates a new number for each route check

 

[Ssnake]

That's why it's called XNew...

[DarkAngel]

One thing you might want to check too. When using a jump to end route which ends in a battle position it is best to make the route multi node and have the last leg in the direction you want the unit facing. This may have been fixed but it always used to be the case. Because instead of spawning at that BP they are Arriving at it.

[Rotareneg]

Ok, I got it working properly now. It seems the New variable only works the way I was expecting it to when a vehicle moves onto a waypoint, not when a vehicle starts at a waypoint or when checking for jump-to-end routes at the beginning of the planning phase. I changed the infantry to use fixed variables to pick their locations with jump-to-end routes and changed the vehicle jump-to-end routes to end at a waypoint with an unconditional route that goes to a conditional "hub" waypoint instead of jumping directly to them.

 

When a vehicle starts with multiple conditional routes that test the New variable, the sim appears to roll a new number for each check. If you have two routes, one being <50 and the other >= 50, the first route checked will have a 50% chance of being taken, the second route a 25% chance, and neither route the remaining 25%. On the other hand, when a vehicle moves onto a waypoint while the scenario is actually running it seems to generate a single New variable that is checked repeatedly against each conditional route at that moment, so that the above example has a 50% chance to take either route.

 

Attached is the fixed version.

Tanks_v3_M1A1.7z