What do you plan on doing that you haven't done before?

ArmchairArmada on 2008/09/03 15:21

PyWeek seems like a good opportunity for experimentation. I'm planning on trying to make use of PyODE for some physics and collision detection. I've use some other physics libraries a little before (Bullet with C++), but I thought this might be a nice opportunity to see what PyODE has to offer.

I also thought a lot about how to create simple, flexible, scriptable, dynamic worlds. If I implement my idea the right way I should be able to very quickly and easily add new unique interactive gameplay elements on a whim. If I wish to change how an object in the world behaves it may even need only one word to be changed.

For example, in XML this might be:

<object id="check" type="variable">
	<property id="property" value="value" />
</object>

<object id="object" type="object">
	<action id="action">
		<if condition="check.property == value">
			<trigger object="object" action="action">
				<argument name="name" value="argument" />
			</trigger>
			<event actor="actor" target="target" event="event">
				<argument name="name" value="argument" />
			</event>
		</if>
	</action>
</object>

I'm still not entirely sure how I want this. This is actually getting more complicated than I thought it would be (the implementation might be difficult), so there may still be room for improvement or simplification. Maybe if I didn't try to use XML?

Anyhow, what do you plan on trying to do that you haven't attempted to do before?

(log in to comment)

Comments

simono on 2008/09/03 15:40:

Never Write a Framework (tm) come's to mind :)

But the idea is intriguing...

RB[0] on 2008/09/03 15:41:

Well, for a week, I would suggest you use python itself, and just make a core class, which then you can use and extend your objects, like this:
class (Object123(GameObject):
    def trigger(self, name, args):
      return getattr(self, name)(args)

    def move(self, args): #is a trigger...
      etc...

That would probably be faster to set up and use, but if you want to try something new, then go the xml approach, otherwise, I would suggest not *much* new, as you still need to be able to finish, and ODE can be a beast as it is - I would suggest you worry about that, and try xml next time...

As for us, we are not trying anything new, just trying to get what we tried last time to work, mainly networking :) LOL

lucio on 2008/09/03 16:27:

we will try some ai (reinforcement learning) and some skeletal animations. hope this works.

Chard on 2008/09/03 16:30:

@ArmchairArmada: We had a scriptable world last PyWeek, all credit to Martin and Adam but I think our scripting system was awesome. RB[0] is right, write it in Python, it'll be cleaner faster and more dynamic than anything you could write in a year, let alone a week. Python is after all a scripting language. :-)

Anyway you can check out what we did (generator-based coroutines incrementally advanced from a state-based script management system) by looking at Robot Underground from April.

I hope to write music for our team this PyWeek, which we've never done before. I'm not too set on it though, mostly I want to produce a great game.

ArmchairArmada on 2008/09/03 20:08:

Please disregard what I wrote. A few minutes after I made that post I realized that I was really over-engineering things. Implementation would have been time consuming and difficult and in the end I would not have had as flexible or powerful of a system as simple Python code could accomplish. I think I might stick with a more traditional use of classes instead.

nikolajbaer on 2008/09/03 21:19:

I under-engineer all my pyweek entries.. to me it is more about the iterative process.. hammering on a lump of metal until it shapes into something aesthetically pleasing. My pyweek 6 entry was essentially all in 1 py file of about 650 lines.. so gnarly.

Akake on 2008/09/06 19:56:

This'll be my first PyWeek entry, but I plan to try to make a non-arcade game for the first time.

My only previous Python-based game ( http://www.pygame.org/project/860/ ), as well as all of the games I have completed in Java, are arcade-style games.

I plan to change that this time around.

eugman on 2008/09/06 21:30:

Pyglet, Non euclidian 2d world, some new simple physics.

bjorn on 2008/09/07 16:30:

I had an idea for some 2d skeletal animation, but was actually hoping to write that as a library I could release before pyweek. That didn't happen; I didn't even start it. There's not going to be time for me to write that (I anticipate having maybe 2-4 hours to write a game - new babies are a large time sink). So the new thing I'm trying is cocos2d. I expect that'll be a huge time saver.