All of the code for this demo, along with a few others, are available from GitHub, and for those unfamiliar with GitHub, here’s the Zip archive. While we’ve some time, let’s run through some key elements of this sample:

The demo:

If you can’t see the demo, for whatever reason, I’ve stuck an image of it below this paragraph.  There’s a search term box, a numeric stepper for the number of images, and a search button.  When you click ‘Get images’ it performs a search on the Flickr API, which returns a list of images.  I’m then using the StarRequests library to sequentially load the images from Flickr, and throwing them into Simple-Box2D to display them in a funky interactive way.

StarRequests

It’s probably more of a micro-framework at the moment, but StarRequests is designed to make it easy to perform a series of actions (requests) that can be queued or batch-executed in a way that’s portable between projects.  It’s designed to deal with common tasks, like loading a remote URL and extracting the BitmapData, while also abstracting their inputs and outputs to make it possible to mix request types in a single queue.  And queueing is important a lot of the time, because trying to do everything simultaneously often ends with snarl-ups and complex queue management.  With a properly constructed StarRequest class, the systems you so often write and rewrite for different projects are much more portable.

For this example I had to write a complete StarRequest class to load images from Flickr – and that’s all it does – but on the plus side I didn’t have to touch the queuing system, I now have a reusable Flickr image loader class that I can use again in the future, and I also know that elements, like the queuing system, are a solid base to work from.  There’s more tasks going on in this example that I could extract into StarRequests format, but that’s a task for another day.

Simple-Box2D

A little side-project of mine, Simple-Box2D is a set of classes that ease the learning curve between ActionScript and Box2D style code.  It’s in dire need of some new functions, cleanup and most of all, code commenting.  Sorry about that.  All this is doing is taking the final Bitmap objects returned from the StarRequest classes, and using their dimensions as the basis for new objects in the Box2D world.

Flickr

I’m making use of the excellent as3flickrlib classes to interact with Flickr, and then adding a bit of url stitching in the StarRequest classes to create the references to the actual image files (because the API doesn’t return that data).

… that’s all folks.  Comments appreciated – I know this isn’t much of a tutorial, I shall have to conspire with Mark and see if we can get one created.

http://projects.stroep.nl/ValueObjectGenerator/

What is a value object anyway?

For the uninitiated, a value object class is essentially a class with very little or no application logic inside it.

Instead of being a file that actually does something, it’s more of a class for just storing data.  What makes it special is that AS3 classes are not all dynamic, so you have to plan what data you need to store and how you’ll store it; and you can achieve a vast majority of that with value object classes.

If you’re coming from an AS2 or Flash IDE background you might think it’s a bit unnecessary or even frustrating, but when you plug value object classes into a development environment like Flex Builder you see where it comes in useful, when it can actually gives you useful autocomplete suggestions, and warns you if you’re passing the wrong type of data into your classes.

This is a very cool experiment – using a socket in AS3 to interface with Brain Computer Interface (BCI) hardware. Check it out at this url: http://seantheflexguy.com/blog/2009/12/30/neurosky-mindset-brain-computer-interface-actionscript-3-0-api/

C++ – or whatever Box2D is written in – is not, and it’s a little painful to setup and easy to break.  For my sanity as much as anyone else’s I’m working on a set of classes in AS3 that wrap around the Box2DAS3 classes, and provide you with a more familiar syntax for working with Box2D – objects, methods and utilities that makes it quicker to throw things together and don’t require you to rethink the way you work.

Well, that’s the eventual aim anyway.

Here’s the result of the first round of development – creating a Box2D world and adding objects, in about 6 lines of code.

var options:Box2DWorldOptions = new Box2DWorldOptions( 500, 280, 30, 9.8 );
options.setWorldEdges( true, true, true, true );
var world:Box2DWorld = Box2DUtils.createBoxedWorld( options );
world.debugDraw = true;
world.animateOnEnterFrame = true;
addChild( world );
for ( var i:int = 0; i < 30; i++ )
{
world.createCircle( 500 * Math.random(), 280 * Math.random(), 50 * Math.random());
}

And here’s the result:

Not at the point of getting any interaction, or anything more interesting yet, but I’ve create a utility class for creating Box2D worlds and a Box2DWorld class (as opposed to b2World, which it extends).  The aim of these classes is to condense about 40 – 60 lines of initialisation and basic scenarios into just a few lines, using default variables whilst still allowing for optional overrides.  So if you wanted a world with 0 gravity, that’s just a case of changing it at the start.

I’ll post the classes soon – just as soon as I decide the best way to do so.  It’ll all become more clear then, I hope.

Here’s the result:

It’s pretty.  I won’t even try to explain the idea behind it, but it’s basically an attempt at simple life simulation, with some AI governing the actions.  This time around I’ve built it with the aid of Box2D, so all the movement, impacts and interaction is achieved by punching some figures into the Box2D world, which makes the thing a helluva lot easier.

The big plus of “trying something new”? I now have a basic understanding of trigonometry!  I didn’t even have to Google anything!

You can name your loops!  Okay, so it’s not very exciting, but naming your loops will make things so much easier to organise and understand – I expect.  Check out the link to read all about it;

http://blogs.adobe.com/cantrell/archives/2009/12/labels_in_actionscript_3.html

For Animations & Tweening:

GreenSock Tweening Platform [link]
The name actually encompasses five different varieties of animation libraries, and is awesome at scripting individual and timeline animations.  Not just that, but it comes packed with a load of animation options that you didn’t know you wanted, until you see them.  Like realistic motion blur.  That one’s cool.  Especially once you’ve tried making it yourself in the past…

Game development

Collision Detection Kit [link]
Much like the GreenSock stuff, just how good this is really has to be seen to be believed.  The basic Actionscript hit-test tools were never really that great.  Sure they worked, but if you wanted anything more than a Boolean response, it was no good at all.  The Collision Detection Kit gives you enough information to do proper physics – working out things like what, where and how hard things hit.  It works with any display object – vectors, bitmaps, even video.

Box2D [link]
Box2D.  Runs blindingly fast, precisely calculates physics and interactions.  It’s a true physics engine, rather than others (like Collision Detection Kit) which are just imitating physics.  A little hard to get into because it’s all written differently to every other Actionscript example I’ve ever seen (“You must unlearn what you have learnt”) but once you get over that hurdle, it’s just so… powerful.

Documentation for Box2D isn’t very new user friendly, but Emanuele Feronato has a fantastic series of blog posts on Box2D which will prove invaluable.