Re: [glob2-devel] scripting language

[Hein-Pieter van Braam]

On Tuesday 24 May 2005 15:08, Stephane Magnenat wrote:
> > SGSL is really nice for glob2, that wasn't the reason why I was
> > contemplating replacing it with something else, what I wanted to do is
> > have a complete script engine for every aspect of the game, and try and
> > move ALL the game logic to a script engine, that is why I was thinking
> > about LUA as it is EXTREMELY easy to implement for glob2, it would
> > require a lot of code changes, but it could very well make all the other
> > ideas on the roadmap a lot easier to do.
>
> Replacing the whole game logic with scripting is tricky from the
> performance point of view. Furthermore, that's a very big job, but if
> someone wants to do it, that's ok. Again, as for the GA stuff, I don't see
> this work on the main branch of glob2 for now, although I think such a
> projet could live in glob2's repository beside glob2.
> Anyway, a clear profiling result should be analysed before attempting such
> an endeavour, just to know if it is realistic or not. I know that war3 does
> it, but we have 10 times the number of units present in war3.
>
> Steph

I wasn't thinking about doing this for the 1.0 release, but, depending on just 
how easy it turns out to be, it might. (I know I don't have a very good track 
record of getting thing done in time(r) ;)) 

my motivation for this is also that it would make it a lot easier for other 
developers to join the project, as it is much easier to edit LUA scripts to 
change the game's behaviour than it is to hack Unit.cpp and friends :D
I am only beginning to understand how glob2 all fits together now, and it's 
very complex, as you already pointed out when I first started :)) I think 
that adding a scripting engine would greatly simplify the codebase, and make 
the glob2 platform much more versatile.

as for speed, from what I've read about lua benchmarks and the like, it seems 
that lua doesn't have to be something that slows us down, it might in fact be 
easier to write efficient code for the unit state machine in a scripting 
language than it is in C++ (but that is just a hunch)

basically, it'll come down to :"if I like it enough, I should just do it." 
probably ;)

I'd like some more input whether or not this is the way to go or not, if you 
and nuage are dead against it, I'm not going to attempt to do it.

(btw, did you have the time to look at my win32 build patches? I am still not 
sure if it is correct the way I did it)

thanks,

hp