PDA

View Full Version : Getting ready for the next SPK.


Skorche
2004.09.29, 07:42 PM
Today, I finally wrote down some of the ideas for SPK that I've been pondering. (read: I wasn't paying attention in class today)

Here are the major changes:
1) For now it will be focused on being 2D only.
2) The "model" in SPK will be an entirely different beast. For one, it won't necessarily contain it's own information about links or other kinds of constraints.
3) Collision processing will be simplified.
4) Other fairly obvious things that I never implemented before to make life easier.

The next SPK (new name maybe) Will be based on:
1) Collections: a list of particles (much like PK_model before without the links) That will contain:
a) Information about the particles (positions, mass, how many, etc)
b) Collision type

Also, unlike before it's in the plans from the beginning to be able to have multiple types, normal (linked particles), normal rotational (normal particles that track their rotation when colliding), and rigid (the whole collection is solid) collections.

2) Models: list of collections and rules to make them behave
Models in the new SPK will have their own particles at least, but will share as much information with other models through the use of templates. Collision information, relaxation instructions, and even initial state can be inherited from the template. So multiple models of similar objects can be created easily.

3) I'm also going to make some colliders for level data like tilemaps, convex spaces (marathon like), and terrain. Maybe more types, maybe less.

If anyone has any input about what more they want. Now would be the time to speak up.

aarku
2004.09.30, 11:00 AM
Is it still going to be in vanilla c? What's the license going to be?

-Jon

Skorche
2004.09.30, 07:50 PM
Considering that I've had quite a few people ask for it to stay in C, I'll either actually write it in C, or C++ with a C interface. C++ would be a bit easer because of the use of Classes and operator overloading, but It's really going to boil down to which is faster. Right now it looks like g++ is a lot smarter than me at making things fast, So I'll probably go with that.

As far as license. I'll probably go with something simple and plain like the BSD license.