iPhone Game Development: A Tale of Failure #

Having acquired an iPhone, I've been playing around with it to see just how much can be done with the "SDK" that is the MobileSafari browser. I thought it would be fun to see if I could make a simple game for it, since that's different enough from the kind of UI coding I do at work.

Most of the games for the iPhone that I've encountered thus far were of the puzzle/card variety, and there wasn't anything too exciting/button-mashing/mindless. It seemed like this was mainly caused by the limitations of the events that the browser exposes. With no real mousemove events, and with mousedown events actually firing when the cursor was lifted (i.e. at the same time as mouseup), it was hard to get anything to happen on the screen quickly in response to a user action. One avenue that I tried was to have the screen covered in narrow vertical <div>'s, and to listen for click events on each of them. Unfortunately, while this worked when each was clicked separately (piano key-style), it didn't work when the user dragged his finger from one to another, since this would switch the browser into panning mode, even if the page/viewport was set to be only 320 pixels wide (i.e. no scrolling could actually occur).

Since the viewport panning does not seem to expose any events, I thought that maybe there could be other ways of catching this user action. When scrolling quickly, the iPhone renders newly-visible areas with a checkerboard pattern before actually displaying their contents. I wondered if this meant that images contained in these areas would not be loaded until they were scrolled into view and rendered. If so, it would be possible to use a load event listener as a notification that the user had scrolled to a particular part of the screen. Unfortunately this didn't seem to be the case, images are loaded as soon as the document is displayed, even if they are off the screen.

I then remembered that the iPhone supports an additional gesture, a two-fingered scroll that can be used to scroll text-areas, elements with overflow: auto and possibly other nodes. It would appear that scroll events trigger for this scrolling motion, meaning that it's possible to detect and get continuous updates for this. Though having to use two fingers seemed a bit clumsy, this seemed pretty promising insofar as it allowed for quick responses to user input.

I had lofty goals for my iPhone game (I thought a vertical scrolling shooter was a good fit for the dragging input method and the vertical screen layout), but I decided that I should do something simple for my first attempt. I thought I would recreate aa racing game that I hard first played on my TI-82 calculator, where a car that can only be moved left or right moves at a constant vertical speed through a winding, narrowing road, with the objecting being to avoid hitting the sides. I thought this would map well to a horizontal scrolling motion that I could get with a scroll event handler as mentioned above.

After writing a simple random road generator and the animation code, I was about to add collision detection. Out of curiosity, I tried to move the "car" left/or right as road moved by, even without collision detection, to see if the speed was too high. To my surprise, the road stopped moving as soon as I put my two fingers on the screen. It appeared that intervals (set up with window.setInterval) do not fire when two-finger scrolling mode is engaged. I thought that maybe I could refresh the animation from the scroll event handler, but that only seems to fire when there is actual motion, thus smooth animation can't be ensured.

My failed attempt is visible here. Before pressing the "Start" button, use the two-fingered scrolling motion left and right to confirm the the "car" (the blue square) moves as expected. Then press "Start" and obverse the reasonably smooth animation. Unfortunately it all stops when placing two fingers on the screen again.

This is rather disappointing, since there doesn't seem to be any fundamental technical reason why this should be the case. We're not talking about Mac OS 9 here (there, due to the way UI events were processed, normal event loop execution would be suspended while a menu was held open or a scrollbar thumb was dragged). It's quite clear, given what native iPhone apps can do, that the platform has quite a lot of CPU and graphics leeway. It's therefore unfortunate just how constrained the MobileSafari browser is, since just a few additional hooks would allow much richer interactions with the user.

14 Comments

Having multiple "clicks" at the same time is a problem...
Does the iPhone support Flash? maybe it'll be easy to work with graphics and events from there, if supported.
Flash? I don't think so...

real shame that the event handling is so shoddy. I doubt this is gonna work, but what about a div with a bunch of display: inline type elements that all have onMouseEnter/onMouseLeave listeners on them? Basically your 'lots of divs' tactic except with a workaround to try and avoid going to panning mode? I don't have an iPhone so I can't test this theory.
A former colleague of mine who now works for Yahoo developed a great browser based game 'concept'. You should take a look at Scott Schiller's DHTML Arkanoid.

Although not for the iPod, at least not yet, it is a great example of a browser based game.

--ryanj
Support for flash is coming (everyone hopes) in the next update. Apparently there are some extra security issues that need to be considered for flash on this mobile device.

This will be our biggest/best bet for games on the iphone imo.
iPhone is sweet but like you said I'm using my comment with my iPhone I'm thinking it won't work with the finger control but then again apple can make anything possible
FWIW, I was able to get DHTML Arkanoid working on the iPhone yesterday - just had to throw in a stub element with onclick() to capture click events as a substitute for mousemove(), using the clientX property of the event just the same.
Why are you trying to use the screen? use the accelerometer..

wait.. why are you trying to make a stupid HTML game anyway?

just wait for flash, and ACTUAL 3rd party dev kits...

a web browser for games that arent flash?? you might as well jsut play with some playdoh.
"Apparently there are some extra security issues that need to be considered for flash on this mobile device."

How is this device different than the other flash-enabled smartphones? Other phones offered by ATT can run flash. The "security reasons" is a big fat lie. Same goes for a proper SDK.
weewar.com is a game that works on the iphone
I iHope for iPong ... something I'd expect to find in the gulagosphere!
The perfect iPhone game: NS-Tower. Just one control, and the iPhone has one giant button.
From what I understand, a 3rd party dev kit is going to come out in January, and if I got my source correct, the 20th.

Using the accelerometer would have rendered an experience similar to that of a Nintendo Wii game's motion controls, giving iPhone users something highly entertaining.
Wouldn't you want this to be a native game? Then you wouldn't be limited to what Safari gives you.

Of course that involves jailbreaking, etc... and limits distribution.
A new point of view based on the freshly released SDK
http://apptouch.blogspot.com/2008/07/iphone-good-plateform-for-game.html

Post a Comment