In other words: "Don't use any of the cool CSS 3 features that would make you want to code an app in HTML in the first place". If the only way to make it snappy is to draw it on a canvas and implement your own 'dirty regions only' drawing mechanism its feeling a bit like trying to make something look nice in Java!
I was thinking for a long time this was possible. Everyone always complained about performance.
But for these work arounds for performance, it is totally worth the extra 30% of your revenue you get back by distrubuting through the app store. Instant upgrades, works offline just like an app. This is the future.
Although, some have said, watch Apple closely with how fast they innovate mobile Safari as this type of application would bring developers away from primary apps - apps that Apple makes money from.
I think Apple thinks of HTML5 the way they think of DRM-free music. The more content there is, the more devices they will sell.
Making money from apps is all very well and good, but there more HTML5 apps there are, the better their devices look. And right now, there's way more money in selling iPads than there is in selling iPad apps.
Why would it be then that they push the "apps" part of it? Your comment suggests that for consumers and developers alike it is a PR stunt (a stunt that still generates revenue). But do you think Job intuitively knows that the future is not in apps as they exist today? But probably some locally-run hybrid?
The question isn't "what technology" because that always changes. I think the question is "how will we distribute it?"
Everyone wants this to be the reality, but right now it isn't. Thomas says "with almost complete desktop-like performance in most cases," but seems to say the exact opposite in the previous paragraph (let alone the rest of the article -- who needs drop shadows, right?). He's saying you have to hand code your app, because every framework runs poorly on the device. I'd like to know what, then, is meant by "most cases".
I didn't see anything to the tune of "every framework runs poorly", just that you don't need them since you're targeting a specific browser and much of the "help" you get from a framework is there to prevent cross-browser compatibility issues.
It's easy when you work in something like jquery a lot (myself included) to overlook how in certain cases it's unnecessary, especially in the mobile space where most interesting platforms are running a browser based on webkit.
How many sales do you lose because you can't sell it for $0.99 with a single tap, though? And you've gotta take on your own credit card processing handling. (Or you could do ads, though still it'd be easier on the App Store where iAds are practically a drag-and-drop component with no extra signup/relationship to handle.)
None of these are insurmountable, of course, and a web app certainly has advantages and is appropriate for lots of things. But for lots of apps, that 30% cut to the App Store sounds totally worth it.
Well for those that are packaged by PhoneGap and sold via the App Store, you're back to Apple taking 30%. But if you're also gonna distribute it over the web too, you've added on the time/money expense of handling credit cards and/or ads for the web-distributed version.
Are people selling PhoneGapped apps on the App Store that are essentially the same as browser-accessed apps they're selling independently?
The funny thing is that the app doesn't really work that great when you use the iPad in landscape mode. There is a big black space on the right and the bars cut off arbitrarily. Plus you can only see about half of the cities, and because the nearly all of the screen is a touch target, trying to scroll down to see the rest doesn't do anything. Illustrates the difficulties of coding an app even directly for the iPad.
OTOH, it's nice in portrait, although the last line of text is getting cut-off, probably because I have the bookmarks bar turned on.
To be fair, I imagine this is in part from the fact that it's only targeted for the iPad which means that one of the major benefits of using a framework, getting easy cross browser functionality, is kind of a moot point.
