fabio-t / rlforj-alt Goto Github PK

It does not follow the same pattern of the others, it's not clear whether it works or not, and it should be less efficient than ShadowCasting - so we can safely purge it.

There's a lot of autoboxing. This can be avoided using primitive-friendly collections, like those in FastUtil.

Right now, Fov algorithms consider only a circle around the starting point, while Los algorithms consider a square. AStar considers a square, too.

Instead of trying to make this completely consistent, or to add yet more arguments to the contract, we should simply make this behaviour configurable.

• SQUARE fov implies manhattan distance

• CIRCLE fov implies "pseudo euclidean" distance (eg, floored and converted to int)

Must experiment with eg this

They are too big and confusing. For starters, a smaller radius would do just fine (even smaller for LoS).

Eventually it would be great to make a nicely coloured picture.

Right now, it still uses fabioticconi.. which breaks TravisCI.