For those interested in creating a client (and because the code isn’t yet ready for viewing!) here’s the current output of a test run on the server. Note that games, pieces and spells have a unique game ID, which is a letter followed by an incrementing numeric value. One of the current things I’m working on is cleaning up the code so all possible references are made to these UIDs, and not directly to the objects, as this will make things safer and more manageable within the engine at the expense of a little bit of performance (no doubt I’ll rewrite much of this at a later date so there are two ‘levels’ of code; the UID data-driven level that the users interact with, and the deeper object-driven stuff that processes things.

{
  "turn" : 1,
  "phase" : "select_spell",
  "interaction" : "all",
  "finished" : true,
  "actions" : [
    {
      "type" : "select_spell",
      "player" : "lewster32",
      "time" : 1262563476,
      "data" : {
        "spell" : "s0"
      }
    },
    {
      "type" : "select_spell",
      "player" : "bobster16",
      "time" : 1262563476,
      "data" : {
        "spell" : "s6"
      }
    }
  ]
}

{
  "turn" : 1,
  "phase" : "cast",
  "interaction" : "sequence",
  "finished" : true,
  "actions" : [
    {
      "type" : "cast",
      "player" : "lewster32",
      "time" : 1262563476,
      "data" : {
        "spell" : "s0",
        "target" : {
          "x" : 0,
          "y" : 1
        }
      }
    },
    {
      "type" : "spawn",
      "player" : "lewster32",
      "time" : 1262563476,
      "data" : {
        "x" : 0,
        "y" : 1,
        "occupant" : {
          "uid" : "p2",
          "owner" : "lewster32",
          "properties" : {
            "movement" : 1,
            "combat" : 7,
            "defense" : 8,
            "maneuver" : 6,
            "resistance" : 4
          }
        }
      }
    },
    {
      "type" : "succeeded",
      "player" : "lewster32",
      "time" : 1262563476,
      "data" : {
        "spell" : "s0",
        "target" : {
          "x" : 0,
          "y" : 1
        }
      }
    },
    {
      "type" : "end_turn",
      "player" : "lewster32",
      "time" : 1262563476
    },
    {
      "type" : "cast",
      "player" : "bobster16",
      "time" : 1262563476,
      "data" : {
        "spell" : "s6",
        "target" : {
          "x" : 11,
          "y" : 1
        }
      }
    },
    {
      "type" : "spawn",
      "player" : "bobster16",
      "time" : 1262563476,
      "data" : {
        "x" : 11,
        "y" : 1,
        "occupant" : {
          "uid" : "p3",
          "owner" : "bobster16",
          "properties" : {
            "movement" : 1,
            "combat" : 6,
            "defense" : 5,
            "maneuver" : 2,
            "resistance" : 4
          }
        }
      }
    },
    {
      "type" : "succeeded",
      "player" : "bobster16",
      "time" : 1262563476,
      "data" : {
        "spell" : "s6",
        "target" : {
          "x" : 11,
          "y" : 1
        }
      }
    },
    {
      "type" : "end_turn",
      "player" : "bobster16",
      "time" : 1262563476
    }
  ]
}

These are what every player will receive at the end of each phase. They contain each action performed by the players in the order that they performed them, and timestamped for reference and backup purposes. Players have relatively limited access to the game in terms of actions, just like in the original game. A player may select spells, cast spells, move and attack pieces and end their turn. From these actions however, more actions may occur, so a player casting a spell on tile x will set off a chain of actions, where the server rolls against the spell’s chance, then (if successful) performs the spell’s effect, which may be to spawn a new piece or whatever.

Soon I’ll be locking down the format of these outputs, and documenting the errors and notices a user receives directly. This will allow for anyone wishing to do so to begin planning or working on a client – and I sincerely hope people do!

• Creating games
• Joining and leaving games
• Creating and moving pieces on the board
• Extensive move validation (bounds checking, path checking, engagement checking etc)
• A customisable turn/phase system (this has been an absolute pain in the arse to implement – but it appears to be working as intended now)
• Full JSON messaging system (you talk to the server in JSON and it talks back to you in JSON, there’re no sockets, special protocols or proprietary formats)

The Google Code wiki is somewhat out of date (already!) as things slot into place during development, and I’ll update it as things become more solid. I’m currently using a private SVN, though as the project nears a working first version (and if there’s any interest) I’ll be happy to give access to the SVN – there’s a lot of experience and expertise out there, and I’d greatly appreciate a second opinion on aspects of the code!

It’s dawned on me that I’ve been very secretive even about the name of the project, so I think at the very least I owe you guys that – so here it is!

I think this is slightly more pleasing than ‘Chaos Enhanced Enhanced’!

I’d also like to talk about the plans regarding its launch and features, because I’m taking quite a different approach to the norm. Early on in the planning I wanted to make sure Archaos could be playable by as many people on as many different formats as possible. The general idea I had was that I’d write the game in haXe and then generate the various sets of code for servers and clients from that one source, and ‘et voila’ I’d have a cross-platform game. Since then, I’ve spent a lot of time at work playing with various web-based JSON APIs, and I’ve watched some fantastic projects appear (such as node.js, Persevere and CouchDB) and realised that, in fact, Archaos’s core could live on the web as a data-based API.

It makes a lot of sense when you think about it – you have a scalable server with a scalable database, written in Javascript (a language which I simply adore), using JSON as its end-to-end data format, accessible by anything with a web connection. Being turn-based, it doesn’t need to rely on any complicated comet solutions, and (and here’s the really exciting part) you can make your own client in whatever language, format or platform you like!

I’m really big on open source and open web, and so it’s really exciting for me to think that, although I will still be creating clients in Flash and (with Flash CS4’s leave) for iPhone, there’ll also be a well documented API out there that can be used by anyone. You can easily have pure HTML/Javascript clients, text-based clients to run in terminal windows and so on – and no matter which client you use, you’ll be able to play Archaos with anyone else.

As I said before, I’ll also be making the server and client source available (though the server source will likely need a closed development phase after launch so I can work out bugs, security issues and exploits before I make it available to all) so anyone with a Linux box can host their own server – though I’m not quite sure why anyone would want to do this at the moment, as a server will handle many hundreds, or possibly thousands of concurrent games with ease, and take care of matchmaking, login and so on. In fact this area of development will probably result in the separation of the login/matchmaking master server from the core gameplay servers.

So, to summarise:

• My current work focus has shifted away from a unified client/server architecture, and my time is now being devoted to creating a robust HTTP-based server for Archaos.
• In tandem to the server work I’ll have to develop a client for testing purposes, and so the first client may be a simple HTML/JS or Flash visual interface.
• Once these are at a satisfactory level, I will announce a beta period to work out bugs, and then shortly after that I will release full API documentation for the Archaos server (which will have its own snappy name, already got a few ideas) and allow interested parties to get the ball rolling on their own clients.
• Finally, I’ll finish and release my own client for Flash, and hopefully soon after for iPhone.

As you can probably tell, I’m fantastically excited by this whole project; indeed I can’t see why this hasn’t been done before – though no doubt it has, and I’ve just not looked hard enough… I’d love to hear your opinions on all of this – and I’d also love to find out if other people have attempted something similar!

Edit: Looks like someone has thought of this – in fact pretty much verbatim! http://web.archive.org/web/20031129175919/http://www.openchaos.org/