Space Tyrant: A multiplayer network game for Linux

From:  http://librenix.com/?inode=6692



Since the last release of Space Tyrant, it has gained some actual game-like functionality. The new code can be downloaded from st2.c. Download it as well as the shell script you’ll need to compile it:makeit2.sh.

It’s now possible to connect to the game via telnet and to create an account, log in, and be issued a ship. Once you’re logged in, there is a universe to explore filled with ports for buying and selling goods and planets for scooping free goods. From those trading activities you can earn money, called microbots. Other than trading to earn more money, you only use your microbots to buy fighters -- which you can use to attack other players or the neutral fighters that guard some sectors.

The neutral fighters are only good for parking under while you’re not playing. They afford a little bit of free protection for your ship since no one can attack you until they first destroy the neutral fighters. Note that you can attack other players whether they are logged in or not and any fighters with their ship will automatically try to defend them.

Each player starts out with a specific supply of fuel, called antimatter. Each minute a small amount of additional antimatter is issued by a function called updatefuel(). If you’re not logged in, fuel just accumulates in your ship. This continuous allocation of fuel makes Space Tyrant a type of turn-based game. Unlike traditional turn-based games, however, you can play your fuel all at once or a little at a time and completely independently of how and when other players play.

You can also talk to other players on the radio. There is only one channel so everyone who is logged in hears everything that is said on the radio.

The game isn’t yet playable in any reliable sense, however, because it doesn’t yet back up and reload the data from disk files. The software creates an instance of the game when you run it and it remains running until you stop it or the system goes down. There is also no way to establish a time limit on a game.

That’s pretty much the extent of the functionality changes. Now, on to the code.

First, there is a login function. That function looks in the player database struct and, if the player name doesn’t already exist, it creates an account there. If the name does exist, it prompts for a password, matches it against the stored password, and logs you in if the two match. There is not yet a way to change your password.

Once a player is logged in, he is faced with a sector description and a ‘choice:’ prompt. Any character that the player types at this prompt is immediately acted on as if it were a command. There is a string called ‘commlist’, short for command list, containing the letters and characters that are used as commands. A function pointer array, ‘fp[]()’, is used to store the locations of the command functions. Another function called ‘commscan()’ looks up the command letter typed and returns an index into the fp[]() function pointer array. This combination of the commlist string, the commscan function, and the fp function pointer array constitute the command processing loop of the game, as shown below:
  userndx=commscan(
    tolower(
      threc[th].inbuf
        [threc[th].inptr][0]
    ),commlist
  );
  result=fp[userndx](th);

These lines are embedded in a loop where each user’s input thread, represented by the variable ‘th’, is examined for new input. The new input arrives in string called inbuf. Since each thread has several buffers, an index called inptr is used to keep track of which one is currently being processed.

And, as described above, commscan is used to extract the appropriate function index and place it in a variable called ‘userndx’. Then, userndx is used to index into the fp function pointer array and the thread index (th) is passed to the appropriate command-processing function.

There’s a small amount of misdirection in that first line but, once understood, it becomes trivial to add additional commands. Basically, you just need to replace the placeholder function, ‘nullrtn’, with your new function name adjacent to the command letter you select in the fp[]() definition list.

The new functions are discussed below.

makemap()
In keeping with the evolving game nature of this project, several actual game functions have been added. The first new function, makemap(), builds a 20,000-sector single-galaxy universe and populates it with objects. By changing the GAMESIZE constant, you can build a universe of arbitrary size, but make sure you don’t try to build a universe so big as to consume too much of your system’s memory. I’ve tested universes of up to 1,000,000 sectors, which seem to work just fine. The makemap() function randomly puts ports, planets, fleets of neutral fighters, and nebulas in various sectors throughout the universe, and interconnects the sectors with randomly-generated one-way ‘warps’. Note that planets and ports are randomly given varying productivity's and random initial inventories of our three commodities: Iron, Alcohol, and Hardware.

Note that makemap() builds each sectors array of six warps, sorts them into ascending order, and then looks for duplicate warps. If any duplicates are found, it decrements the loop-controlling variable and simply rebuilds that entire sector from scratch. The sorted warps are a convention that we will maintain throughout the project. Each new galaxy type that we add in the future will adhere to that convention and other functions (and users) will be allowed to assume that warps are in ascending order.

command()
We have added a function to simply list the implemented commands and a short line of description. For now, the function is attached to the command letters ‘?’ and ‘H’ via the function pointer array and the commlist array. This function is mostly useful to illustrate a design limitation. The output of any single sprintf buffer-building function cannot exceed the MAXLINE buffer size constant. This function produces a single buffer out output very near the current 511-byte limit and will soon have to be split to produce two buffers of output. Assuming, of course, that we don’t increase the MAXLINE buffer size.

jettison()
The ‘J’ command activates the jettison() function. It’s only purpose is to dump any cargo out of your cargo holds. It illustrates the method we use to let a function that requires multiple characters of input -- or simple confirmation -- to temporarily turn off command processing and send the next character of input back to it. Jettison requires confirmation so that you don’t accidentally dump your cargo just by hitting the J key. (Since all commands operate as ‘hot’ keys you do not have to hit the [Enter] key to activate a command -- each command immediately executes as soon as you press a key.)

Each command function has the ability to set the thread’s control variable to it’s own command letter. That way, the command processing loop can simply check ‘control’ pass the next buffer of input directly to whichever function is indicated.

warprtn()
jumprtn()
The warprtn and jumprtn functions process requests to move to another sector. warprtn() processes the commands 1, 2, 3, 4, 5, and 6, which represent a user’s request to move to the first through the sixth sector number in the warp list, respectively. jumprtn() processes requests to move to a randomly selected sector in the warp list via the ‘-’ (or ‘=’) key. jumprtn() also implements commands to move to the next larger sector, the next smaller sector, as well as the largest and smallest sector number via the ‘.’, ‘,’, ‘>’, and ‘game design and the programming model. The second article discusses the IO handling code and more of the details of the programming model.]

[Update, June 25, 2005: A Space Tyrant home page has been created as a central index to the various ST articles, links, and files.]

[Update, March 21, 2007: Space Tyrant now has a website of its own!. This site is new but growing and will be the quickest way to find new information and code on the Space Tyrant project.]