moving ant into another ants previous location

[antatack]

I hope this is the right place to post

Say I have 3 ants all adjacent to each other. I want to move ant 1 away then ant 2 to where ant 1 was and ant 3 to where ant 2 was all in the same turn. Because the movement is order specific does the server make the moves in the order they are placed by my bot it seems like this is the case? If not then potentially some of my ants will die because they collide. I just wanted to be sure of what happened and how if anyone knows for sure.

[BearOff]

All next positions are computed before making decisions about collisions.

[antatack]

oops yes silly question then.

Thanks for the answer

[joexner]

All next positions are computed before making decisions about collisions.

So does that mean that 2 ants next to each other can swap positions in 1 turn without colliding?

[tmc]

Yes.

[DrClaes]

I'm working on an algorithm to resolve conflicts when several ants have orders to move into the same tile or each other's location; for the case described above, of two ants which would like to "swap places"; the actual switch is redundant, and it will speed up ant travel if they "bounce" off each other instead, swapping order queues.

E.g.

Ant 1 wants to move A->B->C

Ant 2 wants to move B->A->D

Orders as suggested above:
Ant 1: A->B
Ant 2: B->A

Orders after algorithm:
Ant 1: A->D
Ant 2: B->C

Anyone else doing this (and more sophisticated versions?)

[BearOff]

Good idea, I thought about this too, but as this is not so often case I postponed it - there are more important improvements.

[Parasprites]

I've thought about it, but I haven't gotten around to trying to implement it yet.

[tmc]

While I'll probably implement that just-in-case, ideally you would never assign tasks to ants like that because it means you're doing so sub-optimally. Of course that's a given if you're not reassigning/pathing for each ant every turn, which you probably can't get away with without some sophisticated pathfinding.