Cogmind 7DRL

A new 7DRL is born!

It was an intense week of toiling away in front of my computer (frequently with "my computer in front of me" actually, as I brought my laptop everywhere and even coded on the subway!). Thursday night was the worst--down at 3 AM, up at 6 AM! The long hours of coding were mostly fun, except for that nasty crash which threatened to derail the entire project and turned out to be a well-hidden null pointer dereference that took several hours and three debuggers to track down.

This was my first 7DRL, and I can say that working on it required some behavior decidedly against my style. I like to carefully write fairly safe code (because bug-hunting and refactoring are both a pain--might as well try your best to do it right the first time), but with so little time speed becomes such a priority that hacks are not just an occasional temporary fix, they become absolutely necessary!

The most important lesson learned this week was that if you want to do a roguelike right in only seven days, it better be damn simple. I was three weeks without a computer while on vacation in January-February, so there was a dangerous imbalance of time to plan compared to time to put those plans into action. I spent so long brainstorming the game and coming up with more and more ideas to expand on that the scope of the game was just too vast to be a feasible short-term project. Fortunately I was able to cut down the features list and still keep some semblance of the original game idea, but Cogmind would be a lot different if it hadn't dropped a big chunk of the intended features. Parts of the game that didn't make the cut:

• Fire/smoke/clouds
• Proximity triggered effects
• Programs and viruses
• Interactive terminals/machines
• Keys and multiple types of barriers to progress
• Terrain types and variable movement costs
• Environment props
• Part-specific special actions
• Throwing
• More sensor types
• Cave-ins
• (lots more stuff in the docs)

What was left? Still too much to do a perfect job on, so there are definitely some bugs lurking out there. There's even some sort of memory corruption made obvious by the screwed up application image (at least on my test systems). That said, it's still very playable, and actually a lot of fun.

So what features are there? Here's a more exhaustive description of what the game includes:

• Resource management
• Inventory and equipment system
• Item identification system
• Fog of war
• Several damage types each with unique properties/effects
• Simple yet dynamic combat system
• No XP (character development is automatic)
• Multiple AI behaviors
• Enemies with special abilities
• 12 classes of enemies
• 40 special abilities
• 400 items
• Permadeath
• Autosave/load on exit/startup
• Particle engine for animated ASCII effects (weapons/explosions etc.)
• Animated interface with sound effects
• Randomly generated maps
• Turn-based action
• Single player
• Message log
• Enemy units use the same ruleset as the player
• Score output to an external file

For in-depth coverage of gameplay, read the manual.

Before Cogmind, I was planning on doing a Battletech RL for 7DRL 2012 (and you can definitely see some Battletech influences in there: sci-fi robot combat, a "parts as equipment" theme, heat control as an important combat mechanic...), but a strict implementation of the Battletech rules would require a hex grid (as with any faithful battletech game), and 7DRL is not about rewriting basic algorithms, which is what I would've spent all my time on since I have no hex-based game engine. So I instead worked with what I had on hand. Making Cogmind as a 7DRL would've been absolutely impossible without the benefit of a great starting point: the X@COM codebase.

Think of Cogmind as a total conversion mod (and then some) of X@COM. Before the competition, I stripped X@COM of most of its X-COM-specific gameplay to create a generic roguelike engine on top of REX (my SDL game engine) called 7SHELL, so named because it is a great starting point for 7DRL development. The vast majority of the Cogmind code is directly reused from X@COM--even the Cogmind data files still contain tons of reusable X@COM data. Just by looking at Cogmind you can see it's obviously derived from X@COM (in fact, some of the things it includes are from still unreleased future versions of X@COM); even so, there were some fundamental changes that make it quite a different game:

• Team-based became single-player
• 3D became 2D
• Action point system became an "ADOM-esque" energy-based action mechanism
• (Then of course there's the completely different mechanics)

In the future I might fix up and expand Cogmind a little if the demand is there, but the hackish nature of the code will keep me from wanting to put too much more time into it. If it was going to be much bigger I'd probably restart from scratch. Also, now that the game is done, and looking back over the old design docs and list of cut features above, I have to say that Cogmind probably benefits from being a simpler game than originally planned. It already possesses a lot of dynamic gameplay under the current mechanics, and could easily become far too complicated with any more systems beyond the current build and rebuild theme where already you have to juggle resources, tactics, special abilities, and a huge number of parts.

The game seems to have turned out okay, although there wasn't enough time to spend on balancing the difficulty. It should be fairly easy to consistently beat the first several levels once you know how to play, but after that it can get difficult if you don't have enough gameplay experience. Now that I think of it, that's probably the way it should be... The game is winnable, but you've got to get good.

There will probably be some more postmortem posts about development soon.