Help test a refactored Physics behavior

Hi all,

I've found a slightly updated version of Box2D that ought to be faster and reduce the garbage collection overhead. This should help reduce pauses or stuttering in complex physics-based games. I've done some of my own customisations and the final script after export should come out smaller too.

However since it's modified it might not work exactly the same as it used to. If you could give it a spin and give me any feedback on performance, GC pauses and functionality, that would be great... it worked fine with all the Physics demos I've made, but I don't want to slip a big change like this straight in to a new build in case it breaks lots of games.

Here it is - back up your old runtime.js for the Physics behavior then try pasting this one in:

http://dl.dropbox.com/u/15217362/physics-refactored2.zip

I'l give it a spin after midnight and report back after it.

Extra speed on the physics part is a real gem!

Thanx! <img src="smileys/smiley32.gif" border="0" align="middle" />

I tried it with my project, but unfortunetly it fails to load the game. Tried it with preview mode and exporting, in both Chrome and Firefox.

Chrome console is giving me this error message:

Uncaught ReferenceError: b2Vec2 is not defined

If you want me to try more things just tell me, I'll help in any way I can.

Xavier, what was the reported line number? Any chance you can share the .capx?

Im also getting the error.

<img src="http://www.ruimteschool.nl/SuperRingBall/Test/Test.jpg" border="0" />

Ah, is this one any better?

http://dl.dropbox.com/u/15217362/physics-refactored2.zip

Ah, is this one any better?

http://dl.dropbox.com/u/15217362/physics-refactored2.zip

Yes, this one does work. I will do more testing in a few hours with my laptop, because with my desktop it laways runs at fullspeed.

It runs!

Cant yet say that it has become more speedy, but then again im not using more then a few physics objects.

It works fine for what its suposed to do in my game.

I also run it on my laptop currently.

My test puts it at the exact same speed.

Well, it would only be faster when using a very large number of objects. The main difference should be that garbage collection pauses or stuttering are reduced in big games.

Yeah, tried it with 500 objects, got the same frame rate. :/

I noticed the size of the browser window seems to be slamming the framerate when physics is involved, even if the game is not set to fullscreen and there's one physics object in the layout. If the game window is 640x480 and the browser window is the same size, it's smooth. If the game window is 640x480 and the browser window is 1920x1080, the framerate is significantly lower. This seems to happen with both the old and new physics (it didn't affect the test I did). Does this happen for anyone else?

Arima, that's odd because the Physics object doesn't have any dependency on the size of the window, it's a pure logic engine. Are you sure the same thing doesn't happen without the Physics behavior? 1920x1080 resolution can max out the fill rate on some graphics cards and cause slowdown.

Ashley, Any chance of adding more Box2D features such as prismatic joints?

Yeah, I used the same .capx, removed the physics and 60 fps at browser window 1920x1080 (game window 640x480). Add physics to one object, 18 fps. Resize the browser window with physics to 640x480, 60 fps again. Weird. :/