Data structures and stuff

[jack]

The premise

I have a tile-based computer adventure game. The current state is represented in memory as a 2d array of tiles representing the play area, each with one (or possibly more) objects on that tile. There is also an "undo" history, tracking recent actions, and the difference (objects moved, removed or added) so the previous or following state can be recreated.

In addition, each object remembers the most recent undo step which affected it, so you can click "undo" on an object and return it to the previous state (also undoing any following actions), not just undo one at a time until you rewind to the appropriate point.

I need to check the code, but as I remember, this is represented by each object having a pointer (well, the python equivalent) to one of the elements of the undo sequence. And when you redo or undo the pointers are updated to refer to the newly-correct object.

Now I'm unsure, but IIRC the undo steps refer to an object by coordinates in the play area, not a pointer to the object (in general, we assume the play area might store game objects as just ids or something, not as ab object in memory).

What happens when we want to save the game

We need to be able to save the game -- indeed, a modern game (especially one where actions aren't irreversible) should just save as it goes along, not require a separate load/save action to do so.

This means my instinctive layout above doesn't work. You can't save "a pointer". The best option is probably to use an index into the undo list which the undo list understands.

That can also cut out other possible bugs. If you have a pointer, it could be anywhere in memory. If you have an index into the undo list, you can choose to have the list check that the dereference is valid (and retain the option to turn those checks off if performance matters).

There's other possibilities but I think that's the best one. It is uncomfortably close to designing our own ORM -- we could alternatively have ALL objects represented by a unique id and index things by that instead (either via some global list of ids or only when we load from disk).

I run into this often when I'm game programming, the best way of 'referring' somehow to another game object -- by value or reference? by game coordinates or pointer to memory? But not in other types of programming. Does this experience ring a bell to anyone else?

But now I'm thinking...

This also reminds me of a problem I ran into when I started thinking about rust's memory models. If you have a class, that creates a bunch of other classes, and those classes want to have pointers to each other, there's no easy way of doing it iirc.

I think you need to rely on reference-counted pointers even though it feels like you shouldn't. That's not a problem in practice -- the "store an index" method above also has an indirection every time you need to access the object. But it feels like, you shouldn't need to. And a similar sort of "I want to refer to one of the classes which this big class is responsible for".

But I'm not sure if there's a way of combining these thoughts.

Comments

[jack]

Yeah, I was considering that, and I'm nervous about some aspects of it, but it also may be a very sensible approach (and better than anything to do with pointers :))

Some of this depends on how large your undo stack

I have a hidden motive here. I'm not sure if this is the game I'll end up making (if any), but I have this idea for an adventure game with an undo and a redo all the way through the game that can 'rebase' the redo on top of other actions.

So you can do things like:

Room 1: Player escape prison cell
Room 893: Player needs hacksaw to defeat final boss
Player: rewinds to room 1
Player: searches more thoroughly, finds hacksaw hidden in cake
Player: fast-forwards to final room
Room 893: Victory!

The point being, that sort of puzzle is staggeringly unfair if you can only progress linearly, and even in Lucas-arts style games, it often constrains the design a lot if you can't throw things off a cliff if you might use them later, or can't offend a character you need. But if you can try things out without penalty, you can create puzzles that interlock much more, with a philosophy of "anything can happen".

I may not actually do that, but I'd like the engine to support it :)

[simont]

an undo and a redo all the way through the game that can 'rebase' the redo on top of other actions.

Yes! And not just adventure games, either. I always used to want that feature in puzzle games of the XOR family (which also include the 'Enigma' game on my web page, and the spiritual sequel 'Chroma' by its original designer). Those too have the property that you get 700 moves into the game and then realise that if only you'd done something slightly differently in room 1 and then done the rest of your current solution unchanged, you could make progress.

I never got round to it, but I always quite wanted to try to build a rebasing UI for Enigma. I felt it might well be tractable in UI terms because it wouldn't make sense to rebase a move sequence from parent state A to state B unless the player character's location was the same between A and B, which would mean that when you selected a chunk of moves and tried to drag them, the UI would be able to narrow down to quite a small number of positions that it would even make sense to rebase them to, and mark them / auto-snap to them / generally make it easy to land on the right one.

(I think one reason I never got round to it is because those games depend critically on human-designed levels, and there's a finite supply of those, so once you've solved them all yourself, your motivation to design and implement an elaborate UI to make your own life easier suddenly vanishes...)

[jack]

Ah! Yes, that would make sense.

[green_knight]

(I'm currently working on a game engine, so these thoughts are super-relevant to me; I haven't reached the point where this is actually a thing I'm implementing, but I've thought about it. Had to go away and wrestle other things for a bit.)

One part of my solution to this problem is that I'm dividing the story into zones - and if you cannot get back from B to A, you'll need all of the Quests/QuestObjects before you can proceed.

In your case, I'm not sure that 'undo' is the right metaphor.

I'm also not happy with the game not offering manual saves, because sometimes I want to unroll to a certain save point. And if your game does not offer me an interface, I'm likely to find the location of the save file in the library and back it up manually and curse a lot. (Looking at you, Shadows of Mordor.)

But the other point is that you seem to know exactly where your game-breaking objects are located, so you could just take the character back to that place and let them search again, applying any other gains to their current status, and then continue the game from where they were - a 'revisit' rather than 'undo'.

But mainly, I wish people would stop putting breaking actions into their games, because playing 40h to find that you've made a mistake right at the start is a dealbreaker for me: not only will I despair, run around aimlessly and look under every rock, eventually google, and find the 'gotcha' the designers put in, I will put down the game, walk away, and tell everybody what a rotten piece of shyte that game is.

Basically, designers *literally* create the world, so when they print all the cards and then mock you that your hand sucks, I do not enjoy myself. If there's something I need to find, give me a clue. (I'm currently and for the foreseeable future playing ESO: Tamriel, and those markers are ridiculously obvious, and there's still a lot of gameplay and sneaking around trying in to find things.) I'd love to see more games where hints increase gradually: you can find it yourself, you get subtle and not-so-subtle hints, and if you made it through to room 893, there's a pedlar outside flogging hacksaws for exorbitant prices.

After a certain level of searching, I am no longer playing, I am googling solutions, and the more tedious you make it, the less I enjoy myself.

So... I'm not sure that the rewind would work for me. 'I will make you play this until you get it right' doesn't feel like gameplay to me.

At the same time I get the 'wanting to give players options' because that's one aspect of puzzle-type games that annoys me: you cannot improvise and use the items in your inventory in an innovative manner, you have to second-guess the designers. Which I find tedious.

[jack]

Oh, cool, I'm interested in what you're building -- I know it's not finished, but for what sort of game?

[green_knight]

The app is called WorldWalker and right now, you can create scenes, connect them, and move from one scene to the next.

Once upon a time (1984, IIRC), there was a Mac app called 'WorldBuilder'. It had two windows - an 2bit image where you could click on things, and a text window where you could type N/E/S/W and play twenty questions with the dictionary. You could equip two weapons and decide which one to swing when you - inevitably - encountered monsters, and have a really good time.

With the advent of PowerPCs, WorldBuilder no longer ran. At that point I'd used it for seven years, created dozens of apps with it, including exam preparation, and mourned its demise.

And ever since, I kept thinking 'I wish there was a tool like that'. Eventually, I realised I'd have to write it myself.

I'm keeping the image-and-text concept, I will let users choose from a menu of responses instead of making them guess keywords, but otherwise, I'm aiming for that kind of experience: scene-based, and story-driven.

Here's some concept art from the early days:



Almost any problem you have as a designer, I have squared: I not only need to think about what I want, but how it can be abused. And while I cannot stop anyone from being a dick to their players, I am working very hard at making good gameplay the default option, the easy path.

Ideally, I want this to replace WorldBuilder as a tool: a way for people without a lot of programming experience to make fun games; a way to think your way through a complex idea and go 'how would this flow' even if in the end you'll choose a different engine.

[jack]

Oh, that's cool!

[jack]

Agh, complicated reply. Basically, all of your rant is where I started, but my reaction was not "stories should never contain irreversible actions" but "how can they be incorporated into a game which is actually fun."

My idea was that always requiring that characters have an in-world way of reversing mistakes and finding objects has quite a lot of constraints on the plot -- it means that if the character ever moves from one place to another, or enters some dangerous situation, they either need to be required to have picked up all relevant objects and taken relevant actions before then, or have some way of leaving the current situation to do so now.

However, if you take the point of view that, near the end of the game, you can teleport back in space *and time* to do the thing you missed, and then teleport forward again to finish the climax, you have a lot more flexibility to add interesting interactions between objects and events. For clarity, I don't mean in-universe teleport, I just mean that the player can select a place (or equivalently, an earlier point in time) and do a thing, and then go back to the furthest point they reached.

To me that's the opposite of forcing them to replay it repeatedly, because they don't have to retake all the actions inbetween, they just click one button and go straight to the end of the game. But I'm aware, this is an idea I haven't really tested, so if people who have the same basic starting point as me are put off by it, I should reconsider if it's actually desirable.

What I like about it is that there's LOTS of puzzles that would be utterly unfair in a normal game, that work fine if the character has the option of exploring other areas to find a solution, even if those areas wouldn't logically be available to the character at that point in time. Random ideas like, you have a cell, you're breaking out, it strains credibility that if you do it wrong you can do it again without anyone getting suspicious, but that's what lots of games require, but this game wouldn't. And you have a bunch of actions you can take to do so, but the puzzle is getting them in the right order. Or it's often logical that you can *throw something away*. Most games prevent that (or you just lose if you throw away the necessary plot item). That means you can't have any puzzles that involve throwing something in the sea or whatever, unless the character immediately announces that doing so is relevant simply by letting you do so. This game you could easily do that, because you can undo that single action (either immediately, or rewind that one action individually much later on in the game).