I’ve been a die-hard Windows guy for a long time, but this year I switched to using a Mac so I could learn some iOS development. Since I made the switch, there are a few PC-only programs I miss. One of them is FlashDevelop – which I use(d) for all of my Flash/Actionscript work.
Flash is dying – that’s a fact. Not many new projects are being built in Actionscript these days, but I still get some Flash work trickling in now and then. I’ll also have a need to maintain or update older Flash projects for a few more years, so I need a good Actionscript IDE on my Mac. I tried running FlashDevelop on Parallels, but there were a few nagging problems with it that made it less than optimal. After doing some research on Actionscript IDEs for the Mac, I settled on Intellij IDEA. I was already using PHPStorm and I liked it a lot, so I thought I would give Intellij a try.
It’s been a great IDE for AS3 work. I like it almost as much as FlashDevelop, but the documentation is lacking a bit. Figuring out how to tweak the project settings was a hassle, so I made a video tutorial showing how to set up an Actionscript project in Intellij IDEA:
I strained my back last weekend (I’m getting old), so I’m spending a lot of time sitting down these days, recuperating. On the bright side, I’ve had time to start prototyping the Bouncing Babies clone I’m building for iOS. You can see my progress below:
I don't have any "levels" in the game yet. So far, it just feeds you one "baby" (yellow ball) after another until you get bored. I was just trying to test out the bouncing and the controls. I will add escalating difficulty in the next week or two.
Since this is a prototype, the code has almost no comments and the structure is a bit of a mess. Prototyping is meant to be quick and dirty. The idea is to get a working demo as fast as possible and not worry about the architecture. This helps define what's involved in building the real game. Ideally, you should throw out the prototype when you're done with it.
You'll need a copy of the Flash CS5 IDE to compile the demo. I'm comfortable coding in Actionscript, so I can build faster in Flash than any other method. I decided to sacrifice open-sourciness for speed in this case. I put all of the code in Actionscript class files, so you can dig through the code without Flash CS5 if you wish.
Realism and simulation are often unnecessary. In this game, I want the bounces to be equal distances apart. If the bounces were realistic, they wouldn't be equally spaced. But, I can make it feel right just by carefully adjusting the height and speed of each bounce. The game won't actually model reality, but a user won't notice because it feels right.
Along those lines, I wanted the bouncing on screen to look good. I know that the path of a ball flying through the air is given by a second-order (parabolic) equation:
y = Ax2 + Bx + C
So, I used this basic equation to specify each bounce. From linear algebra, I knew that I needed 3 equations to solve for my 3 unknowns (A, B, and C), which meant that I needed 3 points on each bounce. These were easily measured directly from the original game, but it takes about 5 minutes to do the algebra for each curve. There are 4 curves. What if I decided later that I wanted to tweak the height or width of the curves? So, I spent a little longer solving the equations with a set of hypothetical points: (a,b), (c,d), (e,f) and used them to create a helper file in Flash. It's in the GIT repo as arcmaker.fla. I can input any 3 points and it will calculate A, B, and C for me. It also draws the graph on top of a screenshot of the original game so I can see it. Last, it spits out some text that I can copy-paste into my game actionscript files. I probably spent a little over an hour on this. I used to resist coding helper scripts and level editors, thinking that it would be a waste of time, but it almost always pays off. If you think a helper script might be useful, go ahead and make it.
Another way of making the game feel right is to speed up the ball a bit on each bounce. When the ball moves at a constant horizontal speed, it actually feels like it's slowing down (because my bounce paths are too wide). I use a multiplier to speed up the ball a bit on each bounce. I created another helper for fine-tuning this. If you right-click on the Flash demo above and choose "Show Testing Console", you can change this multiplier. I have it set at a value that feels okay to me. a value of 1 will not change the speed at all - try it and you'll observe the "slowing down" effect. a value of 2 will double the speed on each bounce, 3 will triple it, etc. After updating the multiplier, right-click on the screen again and choose "Hide Testing Console" to play the game with the new speed settings. Tweak the multiplier value until the game feels right. It's amazing how such a small thing can make a huge difference in how the game feels. Again, this is another one of those helpers that is worth the time to code because I can quickly try different values and see what feels right.
The prototype is coming along nicely. I just have to code in my levels to ramp up the difficulty. Then, I can start figuring out how to port the game to iOS.
In case you missed it, the GIT repo is here. Clone it and you can follow my progress as I continue working.
...And you thought you'd never use algebra in "real life":
Yes, smarty pants, I know I could have solved the equations more easily with matrices, but my matrix algebra is super rusty, so I did it old-school.
I have been reading up on entity component game engine architecture a lot lately. Essentially, an entity component system uses composition to create objects and throws away old-school inheritance style programming. This is because game objects feature a lot of complex functionality that can create inheritance chain problems. I understand the concept and it makes a lot of sense, but there are very few nuts-and-bolts breakdowns of how to implement an entity component system.
Here’s a quick primer on Pixi.js:
It’s really comfortable for Flash/Actionscript developers. The folks who created Pixi.js are clearly very familiar with Flash and used a lot of Flash terminology in Pixi.js. It also feels like they worked hard to make it very easy to dig in and start creating cool stuff with minimal boilerplate code. I really appreciate that.
Pixi.js has a really clever renderer: It detects if your browser is WebGL enabled. If not, it falls back to an HTML5 canvas renderer. Not only is it cool that it handles this for you, but browsers with WebGL (Chrome, Safari) are hardware accelerated. This means that they can unload the rendering from the CPU to the GPU, making everything run much faster. Firefox also supposedly has WebGL capability, but from what I’ve read, the performance is questionable. The WebGL thing is going to be a big deal as it becomes more widespread. As usual, IE10 doesn’t support it, but it does have hardware accelerated canvas.
As a side note, the hardware acceleration is all based on your browser and hardware configuration. So, if you have a relatively new device and an updated browser, it just kicks in automatically. iPhones and iPads all support hardware acceleration and many Android devices do too. Obviously, how you handle older browsers is the challenge. In most cases, a less full-featured backup may need to be served up. In a few cases, the user may need to have hardware acceleration to run your game or app at a reasonable frame rate. A smart way to handle this is to check the frame rate of the game and alert a user if it drops below a minimum threshold.
So grab the Pixi.js code and start building some HTML5 goodness.