Improvements

[toptea]

In hindsight, here is a list of things I could have done differently but haven't, due to time constraints:

Level

• more comments/documentation in level.py. Currently the level generation code is a bit cryptic.
• use structured numpy arrays instead of keeping them seperate. That way, I can define all the tile data on one line.
• use a helper function that check adjacent cells in numpy arrays.
• fix eclipse shaped room code. Sometimes it goes out of bound and crash the game.

ECS

• find a better way to organise Esper's long self.world.get_components statement. Not sure if using generator functions to break up nested loop will help readability. Might be better if I ignore python's <80 character rule.
• when changing level, I need to keep the player and inventory items entity in esper's world object. The current code for this in SceneManager is messy.
• dublicate codes in MovePlayer and MoveEnemy processors.
• instead of having an inventory component on the player (a list of item entity ids), have a "location" component on the items instead. Since I removing the item's position (x, y) component to stop it from rendering, the code will be easier if I replace it with "location" component instead of dealing with inventory list on another entity. (Note: It's harder to deleting renderable component because it will loss item's character and colour information)

UI

• should have follow the tutorial and use Message/MessageLog class. Using collections.deque of tuples (message, colour) makes it a bit vague.
• refactor RenderPanel and RenderMenu to use tcod api.

Input

• replace tcod's input handler to something simplier. Hashable Key objects in dictionary is hacky.

Save/Load

• replace pickle to something better.