21 Comments

Summary:

Web apps saved to the home screen of iOS devices run less than half as fast as the same apps launched from the Safari browser and don’t have offline caching access and other features, according to a new report. Did Apple design this deliberately?

Apple-10BillionDownloads-Jan-2011

Web apps saved to the home screen of iOS devices run less than half as fast as the same apps launched from the Safari browser and don’t have offline caching access, according to a report by the Register. The site said web apps opened from the home screen aren’t able to take advantage of iOS’ updated Nitro JavaScript engine in 4.3, offline caching and synchronous mode for better-looking apps. While this could very well be just a bug that’s yet to be fixed, it raises conspiracy suspicions of some who wonder if Apple is not in any hurry to have web apps on the home screen compete with native apps, the sale of which results in a hefty cut for Apple.

This comes not long after Apple also required publishers to use its in-app purchase system for subscriptions, which has prompted a lot of grumbling from some developers and publishers because it forces them to give Apple a 30 percent cut of revenues. One way to get around that would be to build web applications, which can be saved as bookmarks on a home screen and can appear like a traditional native app. Apple has said it embraces both native and web applications built in HTML5 and other web standards. But if web apps launched from the home screen are slower or are crippled, it does give Apple’s native apps a competitive edge. That’s where concerns are popping up. A discussion on Hacker News includes suspicions that this is intentional.

“It makes web apps (that work with any platform) slower, while native apps are not penalized. That increases Apple’s revenue because it will encourage people to make native apps instead of web apps, which has a two-fold effect: one, they get 30 percent of the app’s revenue, and two, you have to buy an iPhone to use it,” said one commenter.

This, of course, assumes the performance problem is intentional. Again, this could be a bug, though the Register quotes a developer who alerted the Mobile Safari team to the situation and said Apple indicated the situation won’t be fixed. I’ve reached out to Apple and am waiting for a response. In the meantime, here are the details of the problem, according to the Register:

The Safari browser includes a new Nitro JavaScript engine, but that doesn’t appear to be available to apps launched from the home screen. That translates into web apps on the home screen running 2-2.5 times as slow as their Safari counterparts. Home screen web apps also can’t take advantage of web caching systems that allow an app to keep running even when offline. Web apps are also using an older “synchronous” mode of presenting content, instead of the new “asynchronous” mode which results in better screen rendering. The Register said that the issues also affect native apps that utilize Apple’s UIWebView API, basically native apps that are built in HTML5 and wrapped with a native wrapper. That means that native apps that lean heavily on web content can also see decreased performance.

If suspicions prove true and Apple is throttling the performance of some web apps, that would call into question its intentions behind supporting HTML5. Already, it has an advantage in promoting HTML5 and mobile web standards because they usually result in apps that are behind natively built apps. But the gap is closing, and some developers are looking at building more applications on the web, especially in light of Apple’s new subscription rules. I doubt Apple will leave this situation uncorrected, now that it’s come to light. If it wasn’t planning on fixing this quickly, it will now. It’s hard to see the argument be leaving this situation in place, except to boost downloads of native apps and hinder competition. Unlike the subscription rules, which can be interpreted as a benefit to consumers, Apple can’t really paint this in a positive light, so stay tuned for an update.

  1. Did you ask for a comment from Apple?

    Share
    1. Yeah, I reached out and am prepared to update once I hear back.

      Share
      1. did you actually try it yourself?

        i just tried it, and i got a slightly better sunspider score with the desktop launched one (4069ms) vs just going to the site (4128).. ???

        also.. why would you assume it’s deliberate vs. it just being a bug?

        Share
      2. Yeah sure you are. So you launched your National Inquirer styled headline filled with nonsense in the meantime. What a scoop. But hey you got yourself some page views and linkbait. So score.
        It would have taken you about 2 seconds to find out this is a known bug with the just released version of JS on iOS 4.3 and the difference is pretty minimal anyway.
        Time to take GigaOM out of my RSS feed. I’m tired of this.

        Share
  2. congratulations on achieving maximum click-throughs with minimum substance. this is a non-story. you found a bug and decided it was a conspiracy. why don’t you file your next report from a reactor in Japan…

    Share
  3. The Register’s report isn’t entirely accurate.

    The issue with web apps launched from home screen icons not using Nitro appears to only affect those that use Apple’s extensions to HTML which allow the app to go full screen, hiding Safari’s navigation bars. It doesn’t affect web apps run with navigation. That seems much more likely to be a bug to me than deliberate action.

    Share
    1. I think this could very well be a bug. It just doesn’t make sense for Apple to do that. So I expect some announcement soon.

      Share
      1. First you take your story from the register which is even less reliable than you and now you try to play innocent like you don’t know anything else. Glad to see I’m not the only one calling you out on your transparently hyperbolic ‘article’.

        Share
    2. Darwin, thanks Apple shill, Apple will be direct depositing your account soon. Keep up the good work of conducting personal attacks on anyone who criticizes Apple. Well done. More rewards will come your way, increase your attacks

      Share
  4. Didn’t I just read this article an hour ago (reportedly)?

    Share
  5. OMG the … erm … no web apps I use are not faster.

    I demand a refund.

    Share
  6. motionblurred Tuesday, March 15, 2011

    While I think this is just a bug, because Apple may have been in a rush to release 4.3 for the iPad, it may be necessary to call them out on this so they don’t behave like MS did with IE. It would be anti-competitive and it would have the EU and FCC on their backs while opening up larger issues with iTunes.

    Share
  7. Maybe Apple is special-casing the Sunspider benchmark.

    Up until last week Mobile Safari running on iOS was a bit of a dog in the Javascript performance department. Sunspider running on my iPad 1 and iOS 4.2 turned in scores of 10000ms. (In contrast the Motorola Xoom running Honeycomb turns in scores of 2000ms.) The new iOS 4.3 and the new iPad 2 magically match the Xoom’s performance of 2000ms – a 5x improvement. My iPad 1 running 4.3 turns in scores in 3300ms range – a 3x improvement. Quite an impressive performance boost.

    I was unable to reproduce the test on my iPad 1 running 4.3 using the official Sunspider page:
    http://www.webkit.org/perf/sunspider/sunspider.html ,
    but it is easily observable using a clone of the Sunspider test.
    http://www.mobilexweb.com/samples/nitro.html

    Using the latter URL launched from Mobile Safari I get 3300ms. Launched from the Home screen I get 8371ms.

    Share
    1. iOS 4.3 updates Safari to use the nitro javascript engine (same as desktop safari) and that is why you see the speed boost..

      Share
    2. It turns out that full-screen web apps launched from the Home screen run in Web.app rather than Safari.app. And Web.app lacks the special permissions entitlement to run Nitro as Safari does.

      What remains unknown is whether that is by design or because of a bug.

      Share
      1. yeah, but honestly… what is the difference? just run your app in safari.. i just opened a few web apps to see what people are doing and the big guys run their apps in Safari.. Google does, Yahoo does.. yahoo mail.. just fake full screen by scrolling down when the page opens.. puts a div at the top and make the whole rest of the page a scrolling panel so unless you scroll that div at the top you never even see the navigation bar…

        the more you look at this, the more you realize that it’s much to do about nothing..

        Share
  8. > That translates into web apps on the home screen running two to two and half
    > times as slow as their Safari counterparts.

    This may seem like a naive question but I don’t have an iOS device. Could you please give an example of one of these “web” apps on the home screen? I don’t understand what a “web app” is v.s. what a native app is specific to iOS (if an app is residing on the home screen of an iPad, for example, isn’t it by nature a native app?).

    Share
    1. “web apps” are “apps” that run in a browser.. when i comes right down to it they are really just web pages.. generally just with lots of javascript so they are more interactive and feel more like a real app.. e.g. google docs, or yahoo mail might be considered web apps..

      any link to any web page in the Safari web browser can be added as a icon on the home screen (aka desktop aka app tray) of the phone making it feel more like an app.. still opens in a browser though.. some just open full screen making the also feel more like apps… it seems when the open full screen there is a problem.. and you can also use open webpages within native apps..

      Share
    2. Hamranhansenhansen Tuesday, March 15, 2011

      No, all locally-installed apps are not native.

      iOS supports 2 kinds of locally-installed apps:

      • App Store: native, managed, CocoaTouch apps installed from Apple App Store (since mid-2008)
      • W3C: open standard cross-platform W3C HTML5 Web apps installed from any server in the world (since mid-2007)

      Both kinds of apps show icons side-by-side on the home screen, both run without a network connection (although a Web app can choose not to support this feature) and both are 100% supported. Contrary to the implications of this BS article and the BS article it is based on, the HTML5 application environment on iOS is the best in the industry. In spite of the fact that Apple gives away it’s WebKit HTML5 engine, other device makers just don’t bother to create a workable Web app platform with it. No, not even devices using Google Android.

      Some app developers choose to make App Store apps, some choose to make W3C apps, some choose to make both. For example, both Flickr and Facebook have both. Some users choose one or the other or a mix of both. This choice is a key feature of iOS that developers and users demanded in 2007 when iOS only supported W3C apps.

      W3C apps are not really “just Web pages.” The HTML5 spec was originally called “Web Applications 1.0″. It’s more accurate to say Web apps are what replaced Web pages as we changed from HTML4 to HTML5. A key difference is Web pages run their logic primarily on a server, while Web apps run their logic primarily on a client, enabling them to continue to work even without a server connection. This also enables the user interface to be more responsive, and for a server to support many more clients simultaneously without being overloaded. Any website can be added to your iOS device as a Web app, but only some will support offline operation right now.

      Go to Facebook on an iOS device, tap the share button in Safari and choose “Add to Home Screen” to install the HTML5 Facebook. Or do the same at Yahoo, Flickr, YouTube and many other Google apps. There are also some “app stores” with only HTML5 apps.

      Share
  9. Hamranhansenhansen Tuesday, March 15, 2011

    You know, it would be nice if somebody called out a non-Apple device for lousy W3C HTML5 Web app support. Even if everything you say here is true, iOS is still by far the best W3C app platform available. The biggest thing holding back a broader deployment of Web apps is they only run well on Apple platforms, so you might as well make a native iOS app and get all the benefits of App Store.

    Considering Apple has the best Web app platform, and has had the best Web app platform for 4 years with nobody even seeming to try to challenge them, I think they have earned the benefit of the doubt that this is part of the transition to Nitro and/or a bug. But there is a double standard with regards to Apple. Any conspiracy theory will do to show they are just like Microsoft. But Microsoft, for example, has no W3C support on its mobiles, and only today rolled out W3C support on the 2 least popular of its 3 PC systems, only 8 years after Apple.

    The Register is also not a good single source of any Apple-related story. They are an IT rag who are stuck in the Jurassic era of MSCE IT and who have a known bias against Apple. They are also in the UK, which sadly thinks Silicon Valley is located in Redmond, Washington.

    Share
    1. Hamranhansenhansen, thank you for clearing up the differences between an iOS home screen installed “native” app (Cocoa) and a Web (W3C standard) app that can also be installed on the home screen. The author of this article could have / should have had the technical background to be able to articulate it the way you did. Furthermore, I think its bad “journalism” on the part of GigaOm if they’re basing most of this “news report” on an article by primarily one source, the Register of the UK (which is definitely, as you correctly stated, rooted mentally in the MSCE IT era of last century and does seem to have an anti-Apple bias from the articles I’ve scanned there anecdotally in the past). The author of this article should have made the title different such as “Possible Apple Bug / Transition Problem in Web Apps” instead of using the words “Apple Sandbags”. Disappointing from GigaOm.

      Share

Comments have been disabled for this post