I know this has been asked before, but I want to approach it from a slightly different perspective.

I'm working on 2 games at the moment, a project I'm leading, and one my friend is. I'm thinking that it'd be advantageous in both to have a scripting language to control level layout/design, and cut scenes.

So I have a couple questions, the first would be which language to use. The ones I've looked at, but haven't tried to implement are:

Lua
Small
Object C (or something like that)
Spidermonkey (mozilla javascript c engine)
Java
Python
some homebrew language

I mention Lua because it's come up numerous times, but from what I've seen, the scripting language itself is hideous, almost basic-esque, and I really need a c or java style scripting language. If people are just recommending that one cause it's popular, that doesn't help me. I'd like to hear some pros/cons in comparison to the other languages.

Small is an interesting language I read about somewhere on this site, the fact that it has a compiler is a nice advantage (I really need a language that can work with compiled files so they're both smaller, faster, and slightly more difficult to modify). I started looking at the code examples, however, and it seems to have a number of rather odd syntaxes that are not from c, so I'm not sure how much I like it.

I don't know anything about object c.

Spidermonkey comes up because I love javascript, I think it's a great scripting language and the ability to extremely quickly design stuff in a very loose syntax sense is great. The main cons I see here are that it seems complicated to implement (although I haven't tried yet), and I haven't figured out if it has any kind of compiling system (it can compile at runtime, but I'm not sure about saving compiled bytecode).

Java, most likely this isn't a good choice, simply because it's so complex and has a lot of overhead. I might as well code the whole game (sans graphics) in java if I went that route. Plus it has to be compiled to use, and that makes dev time slower.

Python, don't know much about that, but I've look at the language a little when looking at jython.

And last of all, I could go all out and bug one of my friends to code me up a custom scripting language. Obviously this solves any possible problems I have, but takes longer, and perhaps is less reliable. There are certain advantages to using a knowing scripting language (though I'd probably try to get it based around javascript).

Ok, so that's part I. To decide on the language, it really depends on how I use it.

In a nutshell I need for sure some kind of OOP system, so I can have objects for enemies and call functions on them and whatnot. I don't want to be using global namespaces for all the functions, since that makes the scripts uglier and harder to manage.

The second part is how I should go about using the scripting language. I'd like to give level design quite a bit of control through scripting languages, such as enemy placement, setting up default parameters for them. Similar things could be used for cut-scenes, loading in characters, telling them where to move and displaying speech and whatnot.

But should I go so far as to have animation loading, playback, and enemy movement coded in scripting language? Obviously with a powerful enough scripting language it's not a problem of how to do this, or even repeated code, but whether the speed is there and if it's worth doing it that way.

I should note that these are both fairly big games, and will most likely be released as shareware of some sort, so licensing issues are also important. I'd favor the cheapest solution (not including own time).

So I think that about covers it, to sum up, which languages have you used and/or recommend, why, and how much do you use them for?

Thanks,

Marcello

-- for comments and lack of {}'s is pretty bad.

[edit]I was looking at http://www.lua.org/manual/5.0/manual.html#2 and I've decided I really can't stand that language, it's just too similar to languages I hate, and it's really stupid to use two different languages for the same problem (I have enough trouble switching between java, c++, and php as it is)

Marcello

Marcello, don't judge a language by its syntax (... unless it's perl). You said you liked Javascript, Lua is really very similar.

What else is there to judge it upon? But I'm not asking people to defend lua, which is all I'm getting here, I want to know the alternatives and for example, what makes lua better/worse than them.

Well designed clean code in any c style language is quite readable. I'd like to say that for all languages, but perl is one of probably many exceptions.

Surely you agree that learning a entirely new and different syntax just for a freaking scripting language (I mean, wtf is ::= or :== wtf... that's just messed up).

That's a silly question. I'm not gonna hardcode the levels, that means the level design can't be separated from the coding. That means more debugging, slower level design, and it requires someone who can code in c AND knows how to compile, link, etc. with the rest of the what could quite possibly be incomplete or in development source code. If it's all offloaded to a scripting language, it means more control, more separation, and frankly, cleaner code in the end.

Marcello

I can say that python has been used at least once in a professional game dev environment: disney's prime 3d Game Engine (now partly open sourced at http://www.etc.cmu.edu/panda3d) seems to use it heavily. So it's probably going to be viable in terms of efficiency and etc for whatever project you may be conceiving. And it's a very well developed and well liked language, and it's known for being pretty ("pseudo-code that compiles"). And it seems to have been designed around, and to support with tutorials and docs, methods of interacting with C/C++ code. Soo ... you may want to give it a closer look.

As a side note I would say that a script being "slightly more difficult to modify" is not exactly a good thing; imo the more accessible your game is to modder types the more likely it is to have a loving fanbase and etc. Can't see why this is such a bad thing. But if you're really worried about it you could try encrypting the files and naming them oddly, which (though obviously not perfect) should keep away idle prying while only slightly increasing load times.

Anyway, take my advice with a grain of salt, of course; I'm no expert

Well, I suppose you could always, in the final release, have the main level files encrypted but also include some mechanism for providing seperate, different level files / config files. Definitely if you're going to have an extensible program, it'll be worth it to consider adding a mod system

Anyway, as another plug for python, it does do the bytecode for fast loading thing, just in case you didn't know.

DIE!

Anyone who thinks "LuaBind" is a clean interface is at least as crazy as I am.

No, I'm just crazy

Hmm ... one advantage of a scripting langauge is that ideally you will be able to change the script code very quickly and while the engine is running (much less significant compile time pauses, etc) such that you can tweak scripts and such in real time, and quickly. I would guess you lose this in a .dll approach; the process of recompiling the scripts and such would probably slow the tweak process by a somewhat relevant amount.
As a side note, python has a debugger, and there do seem to be commercial, fully-featured ides for it, though I can't say I've tried one, unfortunately.

I don't know, once I finally get my VM done (ohy...) it wouldn't be hard to reload a script everytime its saved. Not a big deal if you just replace the code. There can be problems, but that should be expected.

Marcello, I thought you would have usd rhino in your java apps anyways....

I'm guessing that there are IDEs in the $50 price range as well, and I wouldn't be surprised to see one for free, either. I'm simply not using python enough yet to make it worth searching around. I didn't mean to imply that that was the only IDE for python with an graphical debugger; it was just meant as a demonstration that such things exist for the language.

Jesse Schelle (developer of Toontown Online, works at CMU's Entertainment Technology Center) mentioned exactly this kind of action in an IC speech (as part a freshmen CS course [IC] that is basically a lecture series meant to define CS through examples; features basically a researcher a week explaining what cool stuff they work on). So I think the technique is certainly feasible.
I do agree that there would be some issues with it, but still, the faster one can see what the code does, the faster one can iterate through the development cycle and such; it sounds like a nice feature to have.

Annnyway ... good luck, Marcello!