Re: level editor?

[Peter S. <d98...@dt...>]

On Thu, Mar 21, 2002 at 10:28:46AM +0100, Marcus Lindblom wrote:
> 
> ocaml. It was a strict functional language, IIRC?

Mhm, almost as nice as Haskell, almost as efficient as C  =)


> > But what should it contain?
> > 
> >  * Object viewing and editing of parameters.
> 
> Parameters as in appearance, physics or both?

Thinking of both (the non-mesh part of) data/objects/<obj> and
data/objectdata/<obj>
I think the objectdata part should just be viewed as default values which
can be overridden in the scenario listings, that allows for a bit more
diversity..  It's simple to implement..
(that's what I meant with "instance-specific parameters" below)

> >  * Terrain viewing and editing of parameters.
> >  * Placing objects in the world for creation of "objectgroups",
> >    and editing of their (instance-specific) parameters.
> >  * Scenario creation/editing (this has no code-support yet since the 
> >    scenario is fixed, but this should change)
> A simple system for this shouldn't be hard to implement, if the
> object groups are tagged with ids.

They are tagged by their name ;)
And you can ask the objectmanager how many objects of each company is
alive in each group.

> Each 'scenario part' then has
> criterias for success and failure, which can simply be if a group
> is annhilated or sufficiently alive (at least X objects alive) and at
> destination. That should be enough for now. 

Yes, the current scenario-part can easily be ripped out from the code and 
expressed in data-files instead.
The question is how more complicated scenarios can be expressed: custom-code,
simple data, a small scripting-language, a big scripting-language, a
combination?

> Our AI-guys (Niklas, Henrik) had ideas about a scripting language, IIRC.
> How far did those ideas go?

I'm interested in scripting languages as well!  ;)

> I think we should support some OpenSource-modeler as well. Blender works,
> and exports WRML, for which there should be plenty of parsers around. 
> Or are there other modelling tools that are better suited for this?

Just accepting more data-formats doesn't seem to much of a problem.. 
At least not if the format is well-defined or if there's parsers available..


/Peter