Porting code example (Ruby to Elixir)

It’s not a problem for me, but If it’s possible wrote about when you will have a time for this project, so I do not need wait and I can better manage my time.

My English is not good, but I will try answer your questions:

escript - read this tutorial
rest is generated by mix new vexil command

It’s predefined color name. See bunt dependency github project.

No, when I wrote this code I had Grid struct (so I found another name for this module), but after some changes this struct was not needed. But now I think it’s much more cleaner, because Battleground generate grid and have methods for read and write to it. Of course I can send PR with changed module name if you wwant.

AI as a separated Erlang process will decide what to do.
It will ask Battleground module if it can do something and depends on results it will call Battleground to apply AI decision (for example because AI - here bot - see only cells in specific range, think about Battleground as game engine).

Note: color should be unique for all teams.
Database naming convention for example in simple blog we have:

1. Post with title, body and id
2. Comment with user_name, body and post_id.

So Team has color (like id) and Bot has team_color (like team_id).

I use Agent to store game data. In previous (not pushed) version I don’t have Agent and I was need to pass more arguments. For example see where I read map_last_index. In previous version I need to pass it in lots of functions. I think it’s much more easier to read when we read game data from Agent.

What do you exactly mean?
I added code to make it configurable, so you can easily setup game rules, for example Bot types and their count.
I think it’s much more easier to modify this code without completely rewrite “game engine”.
Do you know M.U.G.E.N? It was really configurable. Anybody could add new characters, backgrounds, music and lots more. Lots of fans from lots of communities used this to create own games with for example: characters from other game.
I think project like it (but much more simpler - your vexil) should be configurable too. You can add some exercises to implement new features without need to write new game.
Also I read your code in Ruby and my configuration and it’s implementation is based on setup method in Referee class.This method have statically selected data and I split it to configuration (config.exs) and implementation (Battleground and it’s utils). See how I place bots based on their number.
Do you agree with me?

start_edge from config.exs is based on your setup method.

# I translated it like:
# Defender count is: 3
# if the rest of the division bot count (from configuration) by 2 is 1
# place bot exactly at `start_edge` corner
place(team, Defender, 5, 5)
place(team, Defender, 5, rand(1..4)) # `start_edge` for `defender` is 5
place(team, Defender, rand(1..4), 5) # same here

corner is from Team struct and it’s randomly selected - see &Vexil.Battleground.TeamUtils.random_corner/1 method.

Note: in configuration I use “human” numbers (people count from 1) and translate it to List index (computers count from 0). So when start_edge is 5 I place bots on 4th index random cell (half on X edge and half on Y edge).

Do you have any more questions?

Sorry, very busy week here.

So far:

1. In mix.exs - why escript, build_embedded, start_permanent ?

2. In config.exs - why “coral” etc.?

3. Any reason Battleground shouldn’t be just Grid?

4. Uncertain how ai will work.

5. Are you distinguishing between color and team_color? Why?

6. Does Agent really buy us much here?

7. I don’t understand what you’ve done in battleground.ex – is the complexity really needed?

8. I don’t understand all the “edge/corner” stuff in bot_utils and elsewhere.

I don’t understand it all yet. I will look at it tonight after I leave work
and will ask questions.

Heh, finally my PR is accepted.
Before I begin work on AI I want to be sure:
Do you understand all from my code?
Whether the code is clear and intuitive for you?
I’m not sure how much implement AI could take for me, but I think I finish it by the end of next week.

I added PR with merge commit and now there are no conflicts with your master branch.
Any questions for my code? Something is hard to interpreting?

You can delete file that have conflict. In my commit this file is deleted.
Try:

git rm vexil/lib/referee.ex

and push changes in your shell and then try submit PR.

I had trouble merging somehow. Couldn’t figure out how to resolve conflict manually.

Not sure what state this is in now.

Next example:
In next version AI needs to calculate where to move. To do this AI give question to Battleground: “Can I move 3 times left and 1 time bottom?”. If true it responds: “Move bot from x,y 3 times left and 1 time bottom.” and Battleground should calculate new position (x and y). Fetch item from grid, remove it (replace with nil) and “move” (replace nil with item - struct - in new coords).

This seems reasonable. I will look at it further tonight. Thank you.

The main module (to manage grid) is Vexil.Battleground. It stores grid in Vexil.Agent and have functions to manipulate this grid. For example to store Bot or Flag structure. I wrote some function and split this module to some imports, so:
Vexil.Battleground imports: Vexil.Battleground.TeamUtils. TeamUtils (like a normal team) imports Vexil.Battleground.BotUtils and Vexil.Battleground.FlagUtils and both of them imports Vexil.Battleground.GridUtils.
Note: in your Ruby code Flag extends Bot, so it can move(!). I think it’s not a good to read your Ruby code, because Flag will not move at all.
In my code FlagUtils imports only functions for calculate position based on corner and spawn item. Ok, but why it’s all for Battleground? It’s simple. Bot and Flag do not need to calculate position from corner. I think my version is easiest to understand (if you know how import works).
More, flag do not think and bot do not see full grid, so they should not calculate coords. I know it’s not a bug, but it’s not easy to read - am I wrong?