etimo / diamonds2 Goto Github PK

It seems like bot movement is slowed down when multiple bots are in play on all boards. The more bots, the slower the movement.

Nice to have both in-memory storage and persistent storage. Should be able to start app with either of the implementations.

When the files are moved the import paths not working if it´s imported relative project root.

Breaks:

import { IPosition } from "src/common/interfaces/position.interface";

Works:

import { IPosition } from "../common/interfaces/position.interface";

We need to host the game somewhere where we can have a persistent state. To prepare for spring competition.

Limit bot moves so it respects board settings

We have discussed this feature before but were unable to implement in old version due to race conditions. Lets continue discussing this feature here.

Proposed change:
If you move into another bot you take their diamonds and push them back to their base. This could also help bots from being stuck and blocked by other bots

Async queue:
http://caolan.github.io/async/v3/docs.html#queue

Calls to /boards could also return an list of enabled game features

If (when) we have more than 1 board it would be nice to have some kind of way of switching board in the frontend. Perhaps query the number of boards and links to switch current board?

This is running on a pretty fresh Windows host (with Linux containers):

Running development gives me:

backend_1   | yarn run v1.21.1
frontend_1  | yarn run v1.21.1
frontend_1  | $ react-scripts start
backend_1   | $ nodemon --config nodemon.json
frontend_1  | /bin/sh: react-scripts: not found
backend_1   | /bin/sh: nodemon: not found
backend_1   | error Command failed with exit code 127.
backend_1   | info Visit https://yarnpkg.com/en/docs/cli/run for documentation about this command.
frontend_1  | error Command failed with exit code 127.
frontend_1  | info Visit https://yarnpkg.com/en/docs/cli/run for documentation about this command.
diamonds2_backend_1 exited with code 127
diamonds2_frontend_1 exited with code 127

Running production gives me:

backend_1   | yarn run v1.21.1
backend_1   | $ yarn run typeorm migration:run
backend_1   | $ (rm ormconfig.json || :) && ts-node -r tsconfig-paths/register src/scripts/write-type-orm-config.ts
backend_1   | /bin/sh: ts-node: not found
backend_1   | error Command failed with exit code 127.
backend_1   | info Visit https://yarnpkg.com/en/docs/cli/run for documentation about this command.
backend_1   | error Command failed with exit code 127.
backend_1   | info Visit https://yarnpkg.com/en/docs/cli/run for documentation about this command.

So less errors from frontend (due to it running nginx), and another error from backend.

Gameobjects DiamondGameObject and DiamondButtonGameObject should be removed if bot interacts with DiamondButtonGameObject.

Fix python example bot so it works with the new api.

Split board config into:

• board config
• config for each provider

Add some form of continuous delivery so we get automatic deployments for master

Add frontend code to this repo. Also take care of merging PR from lukas and adapt the code to the new api structure.

Instead of showing all previous highscores in an all-time list sorted by the amount of diamonds, they could be split up into their individual seasons. They could also not show the amount of diamonds collected, instead just showing them in the correct order, and maybe only show the top 5 contestants. This would not scare away new people when they see very high scores from seasons that were very different than the one currently running.

First I join a board with email/name and play. Afterwards I join with my received token but receive 409 Bot already on board.

It is currently possible to move outside of the board.

If a bot is currently performing an operation on the board it should not be removed during the operation. Perhaps this could also be a problem during another bot's operation as well?

Should delete operation also be performed using the operation queue?

Migrate some (or all) of the old data to the new database. How it should be done is up for discussion

• backend/src/config/config.service.ts IS NOT TRANSCRIBED TO .JS
• It works in dev, but not in prod.

Production

backend_1   | [2020-02-07T17:12:13.918Z] DEBUG     AppModule dependencies initialized
backend_1   | [2020-02-07T17:12:15.577Z] ERROR     Unable to connect to the database. Retrying (1)...
backend_1   | [2020-02-07T17:12:20.140Z] ERROR     Unable to connect to the database. Retrying (2)...
backend_1   | [2020-02-07T17:12:24.738Z] ERROR     Unable to connect to the database. Retrying (3)...
backend_1   | [2020-02-07T17:12:29.357Z] ERROR     Unable to connect to the database. Retrying (4)...
backend_1   | [2020-02-07T17:12:33.986Z] ERROR     Unable to connect to the database. Retrying (5)...
backend_1   | [2020-02-07T17:12:38.609Z] ERROR     Unable to connect to the database. Retrying (6)...

Development

backend_1   | [2020-02-07T17:17:15.834Z] DEBUG     AppModule dependencies initialized
backend_1   | [2020-02-07T17:17:18.466Z] DEBUG     TypeOrmCoreModule dependencies initialized
backend_1   | [2020-02-07T17:17:18.527Z] DEBUG     Starting application on 5000
backend_1   | [2020-02-07T17:17:18.531Z] DEBUG     BotsController {/api/bots}:
backend_1   | [2020-02-07T17:17:18.533Z] DEBUG     Mapped {/, POST} route
backend_1   | [2020-02-07T17:17:18.534Z] DEBUG     Mapped {/:token, GET} route
backend_1   | [2020-02-07T17:17:18.534Z] DEBUG     BoardsController {/api/boards}:
backend_1   | [2020-02-07T17:17:18.534Z] DEBUG     Mapped {/, GET} route
backend_1   | [2020-02-07T17:17:18.535Z] DEBUG     Mapped {/:id, GET} route
backend_1   | [2020-02-07T17:17:18.535Z] DEBUG     Mapped {/:id/join, POST} route
backend_1   | [2020-02-07T17:17:18.536Z] DEBUG     Mapped {/:id/move, POST} route
backend_1   | [2020-02-07T17:17:18.536Z] DEBUG     HighscoresController {/api/highscores}:
backend_1   | [2020-02-07T17:17:18.536Z] DEBUG     Mapped {/, GET} route

Currently causes max callstack exceeded.

The API returns incorrect http errors in some places. These need to be handled so a user don't get unexpected errors.

Auto test, package and deploy to kubernetes

String -> number

In the previous version the teleporters moved every time the diamonds respawned. Which admittedly is not very good when the button to repspawn diamonds exists.

However maybe we could move them every 30, 60, 90 sec or something similar?

To prevent people from hogging bases and buttons

Ongoing task to further improve all jsdocs in code.

We dont attribute our icons anywhere. Dont know where they come from or if they are free. Perhaps replace icons with properly attributed ones?

As before, run two etimo bots that constantly play the game while no one else is playing.

Explaining config, functionality etc

Now we answer with a forbidden/rate limit when a move is performed too early (not letting a total of minimum required time pass). Before we kicked out the bot from the game in this case.

The difference is that if we kick out the bot they can't "try" and just hammer the api with requests until one succeeds, thus optimizing the latency handling more. Instead they have to work out when they can perform the move.

Discussions, suggestions?

Should we try having larger boards and possibly add more teleports? Pros/cons?

To prevent disk space from being eaten a lot when deploying.

We should have a place where we write about the rules so contenders can get started easily.

It looks like the bots can enter a position where another bot is located.

Game board takes to much vertical space on phones.

We don't need to lock simulatenous operations to different boards.

Perhaps wrap the board in a new class?