LÖVE

Try to stay on topic here >.>

ljdp wrote:Try to stay on topic here >.>

Ha! Good one.

Robin wrote:

ljdp wrote:Try to stay on topic here >.>

Ha! Good one.

Hehe.

In this version i've added port forwarding. Left click on a port (the red circles around the edges of a hexagon) to bring up a context menu (thanks to the lovely love frames which I'll be using to mock up a gui!), click the port forward button and then right click on another port in the system.
By chaining port forwards you can create express lanes as programs will immediately jump to the forwarded port, or if you know the enemy is coming you can sneakily forward a port so they bypass your system.

You can't cross existing port forwards and you can't forward backdoor ports, so there will always be at least one port into your system. Unless the grid generator makes an island system, but the only way I can think of checking if a system is an island is by pathfinding every system to every other system.

I'm debating wether the pathfinder should incorporate the forwarded ports. Once a path is found the port's could change making the path out of date so I'd have to keep generating a new path which would be expensive. The current implementation is when a program arrives at a system it checks each port and tries to choose the port that leads to the system closest to the goal.

I've also made the zooming smoother.

Nice!

One problem re: pathfinding is now if you have A<->B<->G<->Z and you forward A<->G, and you tell a program from A to go to G, it goes A->G->Z->G. I think it should at least check whether it's there yet, to prevent some really weird things.

Robin wrote:Nice!

One problem re: pathfinding is now if you have A<->B<->G<->Z and you forward A<->G, and you tell a program from A to go to G, it goes A->G->Z->G. I think it should at least check whether it's there yet, to prevent some really weird things.

Good tip, thanks.
I've decided to implement to core networking code now, I figure the earlier I get it done the easier it's going to be debugging it. I'm using a lockstep approach as that seems to be the defacto for RTS.

If you don't know what I mean by lockstep see this link (http://gamedevsoc.eusa.ed.ac.uk/2010/06 ... etworking/)
Basically, it becomes a simultaneous turn based game except the turns are around 100ms in length, actions are sent during the turn and once all clients know that everyone has completed their turn the actions are processed and we go to the next turn.