newt : yes, I'm planning to add speed (allowing you to choose how much time between two points), and resetting the behavior.

0plus1 : since it's using a Java .jar, you need to have a JS plugin that declare an applet, and dump the .jar in the same folder as your exported app, to access the functionnalities of the .jar (and hava Java installed, of course).

simpleex : your button can set a var inside index.html. And since the communication is bidirectional with CallJS, C2 can check the value of that var, and react accordingly.

It's a behavior which was getting dusty on my hard drive, so I cleaned it a bit, and I'm posting it here.

The idea is to set Spline Path for Sprite, making them moves along a spline with a smooth path.

<img src="http://dl.dropbox.com/u/1412774/SplinePathBehavior/demo.png" border="0">

The spline code is adapted from here : https://github.com/MmmCurry/jquery.crSpline. Instead of setting the (X,Y) position via CSS and jQuery.fx, I'm directly accessing the (X,Y) position of the Sprite.

The spline used is Catmull-Rom, allowing the trajectory to go through the points, instead of Bezier, where the control points and the trajectory points are different.

The demo : http://dl.dropbox.com/u/1412774/SplinePathBehavior/index.html

The capx : http://dl.dropbox.com/u/1412774/SplinePathBehavior/splinepathbehaviordemo.capx

The plugin : http://dl.dropbox.com/u/1412774/SplinePathBehavior/pode_splinepath.0.1.zip

Edit : the demo uses R0j0hound's Canvas, for the trail.

simpleex : one way to do that , while still using CallJS, is to have a variable in your page, and read the value of that variable with CallJs inside C2. With taht, you can react and start various reaction depending on the value of that var.

Spirantula: something that interest me is how do you define reaction time ?

From what point do you start counting ? Display of something ? Time between two touches ?

You still need to give credits to where it's due :)

Flash is fast, has lots of effects, and can access hardware devices without breaking the security models. And furthermore, they got audio right ! (For the moment, no browser vendor knows how to do that...)

Furthermore, as I discussed with lucid on IRC, I really Flash (from the 8 or 9 build) as a vector editor (even with no animation). Inkscape is of no use on Mac, and for the others, the tools/shortcuts/workflow is counter-intuitive...

From the comments of the video : he used a clever trick, nothing "HTML" specific. As Kiyoshi said, you need 2 canvases. The slow part with Canvas is drawImage(). So you need to move the viewport, know how much you moved by using two variables, and only draw new tiles in the white row or columns that has just been created. So the number of redraw is only linear, not exponential.

0plus1 : I completely agree with you on the fact "use the tool that fits the job". I use C2 and produce HTML5 games and app for client, in controlled environment, where I know what I do works. For the moment, it's doing what I want it to do, cheaper and faster (for my client) than other solutions, so for me, it's a win <img src="smileys/smiley1.gif" border="0" align="middle" />.

Anyway, I also do code in OpenCV and C/C++, and for the moment, nothing can beat that in pure speed...

0plus1 : I'm not claiming anything spectacular. If your javascript engine is using a JIT, then for code properly coded, you can have the same perfs as a native binary, because a jit'ted code page in memory is *directly* executable.

That's why, in various benchmarks, you can have Java code beating C/C++ code, or Python faster than C, for example. A codetree that has just been evaluated by a JIT can be more efficient than compiled-ahead code, with the compiler having no information of the context.

That's also why, on iOS, webapps can't use the various acceleration in Nitro. JITting means that you are flagging some memory page as "executable". That could be a potential security flaw for Apple (not if they properly designed their browser, but they are �ber strict about that kind of things...)

0plus1 : a shader is executed by the graphic card, so it's hardware accelerated.

And, by the way, interpreted can be faster than native, because the JIT has access to what the compiler don't : the execution context <img src="smileys/smiley2.gif" border="0" align="middle" />.

rexrainbow : ah yes ! It's because the template is copy/paste from another one I made. I'm going to change that in next release. Thanks !

Edit : I'm asking myself a question, because I'm not in front of C2. If you use an "object" plugin, are you allowed to have several of them, distinct from each other ? In fact, if you want to save several files, you need to have a "FileSaver" plugin for each file, to be sure that their content isn't mixed.