Re: Where did all the ...

From: Lemming <thiswillbounce@bumblbee.demon.co.uk>

Date: Sun, 04 Jul 2004 10:59:12 +0000

On Sat, 03 Jul 2004 17:45:47 +0200, Enno Rehling <enno@despammed.com>
wrote:

>Lemming wrote:
>> As an illustration of some of the problems I'm thinking about,
>> consider the question "what is a container"?
>
>I think you're over-generalizing. Do you want to make a game, or do you
>want to write an engine to make an arbitrary game? Adn even if it's the
>latter, you're still better off not gernealizing too much.

It is the latter.  But I think you are probably right, I'm trying to
make it too general.  I'm trying to support too many types of game
world in the same engine.

[snip]
>In the end we decided that items had very little data that changes, and
>for each kind of item had one object (in AO, we call them dynels, for
>dynamic elements). Character inventories are simple list of 3-tuples
>(itemtype, amount, ammunition). The latter entra is needed because
>weapons with their ammunition turned out to be the only massively used
>objets that have changing properties. If your system has item decay, you
>may want to add that, we decided not to have it.

Interesting point.  I hadn't considered the static aspect of most item
attributes.  It doesn't seem to allow much variation between items of
a particular type though.  Hmm, maybe this could be done by putting
most of those static attributes into the item type and have a minimal
number of attributes on the item itself (as you describe) but then
also storing a modifier or set of modifiers to indicate the variation
in the items quality or effectiveness.

>I'm not saying this is how you should model the items. In fact, we would
>like to change this again, but in a running game that's really not
>possible, and the gain wouldn't be all that great. But I'm saying that
>instead of going fully generalized, think about how small you can be and
>still get away with.

I'd be interested to learn what you would like to change it to, if
it's OK to share it.

>By the way, we still have the full-blown items - if you drop an item in
>the world, the 3-tuple gets transformed into a dynel item with a
>database row to represent it.
>
>As for your ammunition problem, you can think of it as a transformation,
>too. The bullet item gets transformed into a +1 increase on the
>ammunition counter of the weapon and vice versa. This means that the
>weapon can't automatically calculate its weight by calling
>container::weight() and summing up the weight of all contained items,
>but instead has to know about its ammunition property and multiply with
>the weight of the type - in short, you have to do some extra work for
>each of those special cases - but it's worth it, and with proper testing
>the extra code for well-chosen lack of generalization isn't a big issue.

Makes perfect sense, the only issue I might have with it is that it's
not restricted to weapons: a flashlight needs a battery; a car needs
petrol (and oil and water); and so on.  So rather than just modelling
weapons with charges, you end up with rather more item types which can
be "charged".  In the end, you might as well have everything capable
of "loading" something.  I accept though that charges as opposed to
contained items may well be the best way to go.  I'm still somewhat
surprised to find that something so apparently trivial can be so
difficult.

>My real advice though is to design the game you want to make first. What
>should people be able to do in your world, what's the important aspects,
>what's nice to have?

Once again, this seems like good advice.  Although in my case I'll end
up designing a handful of different games, at least that will allow me
to see what is the same across them all, and what is different.  That
should help identify the modules which would form the core, and those
which might be implemented as bolt-ons for specific games.

[snip]

Lemming
--
Curiosity *may* have killed Schrodinger's cat.

References

Re: Where did all the ...
lindahl@pbm.com (Greg Lindahl)
Sat, 03 Jul 2004 04:04:03 +0000
Re: Where did all the ...
Lemming <thiswillbounce@bumblbee.demon.co.uk>
Sat, 03 Jul 2004 14:06:31 +0000
Re: Where did all the ...
Enno Rehling <enno@despammed.com>
Sat, 03 Jul 2004 15:45:47 +0000
Where did all the ...
Lemming <thiswillbounce@bumblbee.demon.co.uk>
Wed, 30 Jun 2004 22:26:01 +0000
Re: Where did all the ...
Hugh Bayer <hugh@bigspider.net>
Thu, 01 Jul 2004 05:29:14 +0000
Re: Where did all the ...
Lemming <thiswillbounce@bumblbee.demon.co.uk>
Thu, 01 Jul 2004 22:33:19 +0000