2006 PGD Annual: PGD Question

[jdarling]

LOL... I wouldn't swear at anyone about their coding standards. Honestly I'm not using anything from my library, I'm doing it all from scratch. I wanted to prove (to myself) that I could start from scratch and come up with something on a limited schedule. I must say that this has limited my engine greatly, but in some ways it has made it very simple to work with. I have taken ideas out of my previous experience though (don't know how I couldn't).

Pathfinding can be a VERY bit PITA! In fact it can cost so much time that it alone can cripple your game. So taking time to get it right is definately a necessity. I wouldn't actually use the same method I use in my maze stuff for a large game world (it would cripple the engine and the users PC would prolly puke), but the maze generation code works well for testing overhead and speed of pathfinding.

Have you put any thought into your AI yet? I've been boggleing mine, and still have yet to come up with a good solution that will fit the timeline .

[tux]

This deadline is a struggle I have to admit.

Glad I'm not the only one thinking that

[AthenaOfDelphi]

hehe..

I'm actually suprised how quickly our engine came together. I spent most of stage 1 messing around with code, sussing out the things I figured I'd need. I did plan on keeping my engine nicely oo, but time is against me, so its becoming one of those irritating projects where code is moving from one place to another to cater for feature requirements and such like. This pathfinding is a good example, I should have read about it before I got started... then I could have included costing data... as it is, I now have to modify my editor, shift onto version 5 of one part of my map data format and modify the engine to handle the new data (:? tool time).

That said, it will be worth it as the pathfinding is looking good considering. It didn't actually have that many bugs, they were mainly in one of the storage classes. I don't think its as fast as it could be, but I've not got the time at the moment to run it through the profiler and optimize it that will come later. That said, I've have no idea on the kind of performance I should expect. I'm running an Athlon 800, Win2K server. I have a testbed app (more tool time ;-) ). For benchmarking, it generates a random map 40 x 40 and trys to path from 0,0 to 19,19. Its taking about 18ms to calculate a path, regardless of whether it fails or not. Any comments on how that looks in terms of performance?

As for AI, I've got some ideas, nothing too fancy, just enough to make it nice :-)

tux... its bad. I swore I was going to go to bed early tonight to make up for only having 3 hours sleep last night and yet here I sit at nearly midnight, tinkering with my path finder :shock:

[jdarling]

Well, here is my comment on pathfinding . Its almost impossiable to benchmark using the standard methods. Sure you can do it, but then you run into the problem that on a slower system it runs like shit compared to your super box! Instead look at the number of instructions and stack depth required to perform a path calculation that is at least 1000 movements long. Now if you take that against your typical bsp search (0.25n if I remember properly) calculation of 250 and consider that you will still have to calculate a depth first search (500) and you come up with it taking 750 stages in an ideal search.

Now, to give you an idea, the path finding algo I pointed you at will perform as follows:
1st itteration it will take 1250 average steps to locate the path (not to ungodly bad, but not good)
2nd itteration (along all known and found on the inital) will result in 2 steps (ungodly in terms of speed).
2nd itteration with a new path that crosses an existing and known will result is average 250 (very good).

Reading over my own post it looks and sounds very confusing, but I hope you can gleam something out of it.

[savage]

I have a basic RPG setup where the player is dropped into a "World" that they can roam about. The different "Levels" are basically dungeions that the player reaches by exploring the map. Does this actually fit? Reading the rules state "Design a minimum of 2 levels to be played by optional selection or in sequence".

I'm not a judge, but doesn't the fact that the different levels of your game are reached by exploring and I assume defeating a boss, mean that you comply with the "in sequence" part of the rule?

[jdarling]

I'm not a judge, but doesn't the fact that the different levels of your game are reached by exploring and I assume defeating a boss, mean that you comply with the "in sequence" part of the rule?

Granted your not a judge, but your input is valued . I'm hoping that the exploration and unlocking is considered "in sequence", but one could judge it either way.