boardgameio / boardgame.io

State Management and Multiplayer Networking for Turn-Based Games
https://boardgame.io
MIT License
9.93k stars 707 forks source link

transitioning between stages #447

Closed nicolodavis closed 4 years ago

nicolodavis commented 4 years ago

In the phases-overhaul branch, the only way to manipulate a player's stage is the setActivePlayers event.

For example, to set the current player's stage to 'A', you can do:

ctx.events.setActivePlayers({ currentPlayer: 'A' });

However, this has two problems:

  1. It clobbers the state in activePlayers (if some other player was in there, they would be removed).

  2. It cannot set the stage of the player that made the move (it only works on the "current player").

We could change the semantics of setActivePlayers to overcome these, or split it into multiple events that do different things.

delucis commented 4 years ago

IIRC isn't there the possibility to pass a value directly?

setActivePlayers({
  value: {
    ...ctx.activePlayers,
    [playerID]: 'stage'
  }
})

That said, obvious candidates would include:

(Side note: keep meaning to write a test for this, but I think I saw that setActivePlayers({ others: 'stage’ }) was preserving currentPlayer if it was already in activePlayers)

nicolodavis commented 4 years ago

Another idea that's been growing on me is to have a simple endStage event that's analogous to endPhase.

All this does is that it removes the player from activePlayers.

However, if the stage config contains a next, or if next is passed to endStage, then we just take the player to that stage instead. This is exactly how phases work and I think having the same concept for stages will reduce the cognitive overhead of having to grok different APIs.

I think if we arrange things so that we just have endTurn, endPhase, endStage and setActivePlayers, that would be a nice elegant API.

delucis commented 4 years ago

Now that I’m working on moveLimit (#452), I think setStage would probably also need a way to set a moveLimit, perhaps via a second options argument:

setStage('main', { moveLimit: 2 })

Otherwise users would only be able to set move limits inside setActivePlayers.

nicolodavis commented 4 years ago

Oh, I think that's perfectly ok.

Users can use setStage just to move between stages and setActivePlayers if they want the more advanced options.

On Wed, Sep 11, 2019, 12:00 AM Chris Swithinbank notifications@github.com wrote:

Now that I’m working on moveLimit (#452 https://github.com/nicolodavis/boardgame.io/pull/452), I think setStage would probably also need a way to set a moveLimit, perhaps via a second options argument:

setStage('main', { moveLimit: 2 })

Otherwise users would only be able to set move limits inside setActivePlayers.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/nicolodavis/boardgame.io/issues/447?email_source=notifications&email_token=AABZULRMNYDLDXGJ3WZLZA3QI7ABFA5CNFSM4IRLKFH2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD6LTY6A#issuecomment-530005112, or mute the thread https://github.com/notifications/unsubscribe-auth/AABZULXUFO7HN5EG4K4VX3LQI7ABFANCNFSM4IRLKFHQ .

nicolodavis commented 4 years ago

I think of setStage as a simple "get me to this other stage". I'd prefer to have it as minimal as possible. I'm open to adding more options to it in the future, but let's wait and see how people use the new API first.

setActivePlayers can be the swiss-army knife with all the fancy options.

delucis commented 4 years ago

@nicolodavis Totally understand.

I’ve finished work on moveLimit in #452. My concern was that with this implementation we reset move counts when calling setActivePlayers, so a scenario where one player’s stage changes but the others stay the same would not currently support move limits for that one player. Otherwise, you’d have to use setActivePlayers and do something like moveLimit - numMoves for each other player, which would be pretty complicated.

delucis commented 4 years ago

An example of just how complicated that would be:

const turn = {
  activePlayers: {
    all: true,
    moveLimit: 4,
  },

  moves: {
    move: (G, ctx) => {},

    changeStage: (G, ctx) => {
      ctx.events.setActivePlayers({
        value: {
          ...ctx.activePlayers,
          [ctx.playerID]: 'newStage',
        },
        moveLimit: {
          value: {
            ...Object.keys(ctx._activePlayersMoveLimit).reduce((obj, key) => {
              obj[key] =
                ctx._activePlayersMoveLimit[key] -
                ctx._activePlayersNumMoves[key];
              return obj;
            }, {}),
            [ctx.playerID]: 2,
          },
        },
      });
    },
  },
};

Whereas setStage could support the same scenario like this:

const turn = {
  activePlayers: {
    all: true,
    moveLimit: 4,
  },

  moves: {
    move: (G, ctx) => {},

    changeStage: (G, ctx) => {
      ctx.events.setStage('newStage', { moveLimit: 2 });
    },
  },
};
nicolodavis commented 4 years ago

OK, you've convinced me :)

Let's also support an additional argument in setStage that accepts an object that can carry things like moveLimit etc.