I think it's a cool idea and most of you are forgetting it's visual, so it makes sense.

I see a way it could work, but it would take a lot of effort.

In my opinion you would need to have three sections for a spell (Also calling functions spells is a bad idea, maybe call them constructs) Cast, Continuation, Death for Initialise, Update and Destroy (Destroy would only be able to be done by the spell itself).

The truly hard part will be making sure that someone is limited based on level they are at. There will be people that could come up with something that would break your algorithm and allow them to make a super powerful spell. OR your algorithm will be really bad and no-one below like level 50 would be able to make anything. (That kind of makes sense, keep it to people who have achieved ancient-hood or something)

Two Worlds has a similar system. Each spell consists of a "Carrier" and "Effect". The Carrier may be, for example, a combination of the runes Fire and Arrow, and the effect may be "Fire" and "Area" which means that you will shoot a fire arrow that will explode upon hitting target. Basically, you put an element (Fire, Ice, Poison) and a type (Arrow, Area, Enchantment, Summoning) to both carrier and effect and that's how you create the spell.

Thanks for the responses! Please forgive the following ramble.

A Written Language. My example of Mana++ was an analogy. However, using a typed language for Mana++ could be a good place to start, or at least offered as an optional command line interface. I would prefer to use a syntax closer to spoken language than that of popular computer languages(ie C++). The notion here is to provide a sense of realism such as invoking a spell verbally as you read it aloud from a Grimoire. But, who knows, L33tspeak could be popular for casting spells in a Computer RPG.

For starters, Mana++ would be syntactically reversed in construct declaration and would compile constructs based on context of usage. Thus a construct may serve many different purposed based on context. I by default would use English (although i think latin would be more fun and challenging).

In Mana++, you name the construct followed by the Construct Type or Data type. The interpreter would parse the statement ignoring optional `inert fill` words like (the, a, of) and only acting on `active` keywords like `is` (=, ==). You could place `inert` filler words anywhere in the statement and the parser will ignore them.

The fireball Spell
Create the greatballoffire Rune from a Projectile
The greatballoffire's life is created from a timer that decrements by 30 Ticks per cycle
Cast Circle of throwgreatballoffire until it's collision occurs
move the greatballoffire's in the forward dirction of 1 meter until greatballoffire's collision occurs
if greatballoffire's collision occurs with a Enemy then Enemy's penalty recieves the degrade Spell
End throwgreatballoffire
The End of the fireball

void fireball(void)
{
Projectile* greatballoffire Rune = new Projectile;
greatballoffire.life = new timer(30 * 1);
Meter* movement = new Meter;

throwgreatballoffire:
greatballoffire.direction.forward = 1;
impact Target = greatballoffire.collision;
if (Target == Enemy)
{
}
if(collision == true) goto throwgreatballoffire;
}


A Visual Spell Language (with emphasis on creating the spell, not just the visual effects of the Spell). I've examined the Visual Logic Editing approaches used by Kodu, Kismet, Alice, Scratch. But what has really captivated me is the Little Big Planet's approach to Logic Circuits. What I admire about LBP's Visual Logic Circuit implementation is that complexity level is scalable. Simple circuits can be combined to create more complex ones. Additionally, LBP's visual representation is unlike any of the others described above. It makes a beautiful game out of creating mundane logic. That philosophy is on par with what I desire to deliver.

I've pondered over approaches to Visual Logic Editing. My current idea based on my experience in writing Script Systems and my plan is GUI-tize a stack-based Virtual Machine with a relatively small Visual Logic Instruction Set. The Logic Circuitry (Input/Output Connection) System that Kismet and LBP uses does appear to be very intuitive. The trick for me, will be to find a way to make Visual Logic Machine equally intuitive visually and for some reason I cannot get The Incredible Machine out of my mind.

I would have some kind of system where you can combine attributes of a spell into one spell as mentioned before like a recipe. As an example, you could have the following attributes:

Element: Fire, earth, water, spirit, air, poison
Trigger type: impact, timer, trap (enemy enters into the effect area), threshold (triggers when a certain threshold is met such as HP of target drops below a certain value)
Effect Area: scalable between 1-10
Strength: scalable between 1-10
Damage Type: slow bleed, instant damage

Each attribute has a certain cost of mana and posssibly cast time associated with it. So maybe impact costs hardly anything but threshold costs a lot. You could also not have certain attributes available until they reach a certain level.

One of the harder parts to me is how to visually represent the spell on the screen for every combination so that it accurately represents the spell.

I think the difference between what you're proposing and what you seem to be aiming for (i.e., fun!) is the difference between writing a compiler and writing a game. You want a spell creation system that is fun to use, but you seem to be intent on implementing that system by forcing the players to learn a non-trivial programming language.

Why not start from the fun side and work BACKWARDS? Someone else already touched on this. Enumerate the possible effects you want in your spell system. Then list some different methods/vehicles of producing those effects.

****************************************

Tossing out super simple ideas here...let's start with some EFFECTS:

Heal - adding health to a target

Harm - removing health from a target

****************************************

'K. Now we need some METHODS:

Create - causes to come into being

Remove - removes from existence

****************************************

And a VEHICLE (conveyance, pathway, etc.):

Project

****************************************

And so...

Create Heal - heals the caster
Create Harm - hurts the caster (derp!)

Remove Heal - cancels a heal on the caster (more below)
Remove Harm - cancels a harmful effect on the caster (more below)

Project (Create Heal) - heals the target
Project (Create Harm) - hurts the target (there we go!)

****************************************

Some basic assumptions: all spells are targeted on caster without the Project vehicle, and all spells while not necessarily instant cast are instant EFFECT (i.e., once casting is complete the effect takes place immediately).

With that said we could add a TIME MODIFIER:

Repeat - repeatedly casts the heal without further caster involvement

Said modifier would require increased initial investment of mana/power and could have diminishing effect.

****************************************

And so...

Create Heal Repeat - heals the caster over time
Project (Create Harm) Repeat - hurts the target over time

Now Remove Harm makes more sense, as well as Project (Remove Harm) and even Project (Remove Heal) to strip an enemy of a heal over time effect. And what about Project (Remove Heal) Repeat? Heh. Keep the target from getting any further heals for a bit. But how could you remove the (Remove Heal) Repeat? Lol! Perhaps we need another EFFECT, or maybe we could be loose and say Harm applies to any harmful spell.

We've now got 2 EFFECTS, 2 METHODS, 2 VEHICLES (self+project), and 1 TIME MODIFIER giving us a grand total of 16 possible spells. Yeah, some of the combo's are...weird, but I can envision situations where there could be a use (e.g., you're walking through a field of magical sleeping poppies, so you cast a Create Harm Repeat on yourself to keep from falling asleep).

Some of the removes are kinda' useless so we probably need another TIME MODIFIER:

Wait [Until/For] - until meaning not triggered until needed, meaning opposite effect; for meaning not triggered until same effect (examples below)

****************************************

There we go!

Now we can do...

Create Heal Wait [Until] - heals caster on next harmful effect received
Create Heal Wait [For] - heals caster on next heal effect received

Project Create Harm Wait [Until] - harms target when target receives next heal effect
Project Create Harm Wait [For] - harms target when target receives next harm effect

Remove Harm Wait [For] - removes next harmful effect on caster before it is applied!
Remove Harm Wait [Until] - would read Remove Harm Waiting Until [next harm is needed], which would mean confusingly enough when the next heal effect is received - cast this and the next heal effect you receive also triggers a remove harm on yourself which would be useful if you see you've been afflicted by any kind of Create Harm Wait

****************************************

Which brings me to some particulars. To make Waits worth casting I would think they would make the cast spell require LESS mana/power, since there is no immediate effect required. Conversely to balance this I would think the effect of a Wait spell should be less than a non-Wait spell when appropriate (e.g., a Project Create Harm may do 100 damage whereas when a Project Create Harm Wait [Until] fires off on a just healed target it only inflicts 75).

Anyway, those are just some random thoughts. And I didn't even add in ELEMENTAL modifiers.... (!)

Project Create [Fire] Harm
Project Create [Water] Heal Repeat Wait [Until]

Or FOCUS modifiers... (!!)

Area Project Create [Fire] Harm
Area Create Heal Wait [Until]

Every single one of the above items (modifiers, effects, vehicles, etc.) lends itself very well to any kind of graphical UI for spell creation. Each one could be assigned a different power cost and level of effect, and you could very easily allow the spell creator to "enhance" the effect by dumping more power into the spell. That way your generic level 1 Mage would only have say 10 mana to dump into a spell but your level 10 Mage might have 100. Ramp up the Harm effect, extend the Wait duration, or lengthen the Repeat. Increase the Area. Etc., etc.

Take care, and cool idea! Love to see it in a game some day.