The lack of power pellets is not a feature, but the result of a failure in time management on my part. :p
The turn-based version has its problems since every actor takes the same time to act, thus the player cannot outrun the ghosts, but the ghosts cannot catch up to them neither. I should experiment with different speeds like the original and see where that goes.
The AI pathfinding routine is incomplete and full of hacks and bugs, sometimes the pink ghosts fail to find a path and stop, and can end up blocking other ghosts since the code prevents them from walking onto one another, as well as from going the opposite direction they were taking, hence they stop.
I originally intended to, but ended up spending too much time on maze generation and too little on AI states and pathfinding, therefore I marked my entry as incomplete.
As of now, only the red ghost (chase the player) and the pink ghost (target a tile in front of the player's direction of travel) are somewhat implemented. Since I relied on rot.js' pathfinding tools, I could not really model the original ghost behaviors which often target a tile outside of the board.
I'm thinking of working on replicating the original logic after taking a little break to work on other projects.
I would like first to try and implement stuff from the original game that I couldn't yet, such as proper AI for the four different ghosts, solid collision detection, and power ups for the player to turn the table on their foes.
But I agree with you throwing in an original idea would be very nice!
It was a great experience for me to participate in the 7DRL, and I feel indeed more motivated to learn about algorithms, now that I can see how they could be applied in software.
