How should I design a combat system that properly accounts for buffs, debuffs and control modifiers

AsCiiexe

I'm making a prototype of a top down combat system heavily inspired by that of MOBAs and ARPGs and have managed to make a few different basic attacks and some simple enemies.

The thing is when looking at most games in those genres there's usually a ton of different buffs/debuffs and modifiers that the characters can have with their own timed durations (sometimes permanent) and even custom effects. Are all of the possible modifiers built into the characters themselves and enabled/disabled in certain situations or is it a system in which the characters are blank slates where modifiers can be stapled into and the modifier itself decides how the character will be affected? Do I do that using yields or turn the modifiers into nodes that modify their parent they are attached to and delete themselves when the time runs out?

Right now my guess is that modifiers that limit the player controls like stuns, silences and roots should be built into the character itself and enabled/disabled when necessary while stat modifiers can be done dynamically attaching modules and externally changing the stat numbers of the character, but I'm not sure if this is the best way because if I want to make a special control limiting modifier like, for example "fear" which makes the character run away from another character for a set duration or "disarm" which only disables certain attacks but not all then I would have to mend all of those onto the character itself, would it be better if all of it were external nodes including the control limiting ones?

Sorry for the rambling, I'm mostly looking for opinions about what would be the best way to go about this without ending with a monolithic spaghetti code nightmare!

cybereality

You can make the modifiers and array on the character. I guess in OOP style, the Modifier would be a class, and have certain generic properties and functions (but you could sub-class it to make special versions). The Player has a variable called "modifiers" which is initially an empty array. When you create a modifier, you add it to the modifiers array. Every frame the player updates, you loop through all the modifiers and run their update() code, which can do whatever you want, and then return a state to the player. Then after this, you can use this returned state (probably a Dictionary of values like "speed_scale" = 0.5 to slow a player down). Then you would modify the players default state by the state that is returned. The modifiers themselves could have a "time_left" property in seconds and every update in the main loop you would minus delta from the time left and once that is less than zero, you delete the modifier.

AsCiiexe

Using a class system does sound like a good way to organise it, so in order for the dictionary results to make sense with the more "disruptive" modifiers like the fear example I guess they should be a function built into the character right? Like, stuns and silences can just be a boolean but the fear has to be a done as a function (or state in the state machine), or would there be a way to still keep that in the mod class?

cybereality

Yes, I would assume fear would be a state in the state machine, but maybe the modifier can have a state override and when that modifier is active it forces the player into a particular state. But it could also just be a boolean you check for, because I don't know if you need other states active (like can the user still move the player in a fear state or is it just AI controlled?).

Erich_L

From a programming perspective one of the most important things I'd say is just to be VERY strict on where your combat logic is. Keep it in ONE place.

On a hero, for example, have one function receiveAttack(attack) and in this one place count in all the buffs or debuffs the hero already has and compute the final damage taken. Resist the temptation to spread that logic around, like letting a simple minion look through the hero's buffs and armor to see what kind of damaging attack it's going to send over.

I've made this kind of mistake countless times. It's reason why I was drilled into using getters and setters for variables. When you change a variable, use the setter and getter, keep the logic there, and resist logic like: hey I'm a creep, how much armor do you have? ok, I'll do self.baseDamage - hero.Armor amount of damage hero.heath -= damage

AsCiiexe

I'm still learning a lot about programming and trying to develop the modifier class system was a bit messy so I have decided to use the timer nodes method at the end and for stuff like poison, health regen or other constantly shifting modifiers a timer node within a timer node that would determine the tickrate. For now it has worked pretty well but I guess in some future I should probably try more solid methods for it, thanks for your opinions anyways guys.

@Erich_L said: On a hero, for example, have one function receiveAttack(attack) and in this one place count in all the buffs or debuffs the hero already has and compute the final damage taken. Resist the temptation to spread that logic around, like letting a simple minion look through the hero's buffs and armor to see what kind of damaging attack it's going to send over.

I've made this kind of mistake countless times. It's reason why I was drilled into using getters and setters for variables. When you change a variable, use the setter and getter, keep the logic there, and resist logic like: hey I'm a creep, how much armor do you have? ok, I'll do self.baseDamage - hero.Armor amount of damage hero.heath -= damage

that sounds like sound advice, for now I do use a setget for the health and max health changing logic on all entities but I'll see what other places I can use it in. Now that I think about it I should probably setget the disruptive modifiers like "stun" in order to reset the attack speed of certain enemies like the sniper who otherwise is gonna have an instant shot as soon as it stun ends