Adobe Formally Enters Runtime Environment Market with 'Apollo'

One of the immediate dangers of operating a Web application outside of the browser is that it, by design, bypasses whatever security provisions are given to the browser, by the operating system as well as by third-party applications. This includes such things as group policy objects, which are the focus of much of the security model around Windows Vista. For this reason, Adobe is building a kind of "sandbox" model around Apollo, borrowing the phrase most commonly applied to Java.

But Adobe's early documentation states that Apollo applications under development run outside the sandbox, warning them of the potential implications - including the fact that, by default, developing applications run with full privileges. One other warning reads, "If an application uses data from a network source to determine a file name or write to a configuration file then that data needs to be validated to make sure that it is safe or that it comes from a trusted source." So built-in certification does not seem to be the order of the day for Apollo, at least not yet.

Despite these dangers, perhaps the greatest advantages developers realize from being able to deploy Web applications off of the browser, Downey suggested, has to do with the infrastructural requirements to run the application.

"Today, if eBay wants to redesign what their Web site looks like, or how you get at information, it's both a very expensive process and a dangerous process," he described, "because once you change how people interact with your application, you risk turning people away that are used to using it in a certain way. So if eBay wants to change how you get access to, say, their auctions, they would have to set up a beta version of their site that they invite users to, and start collecting usability feedback.

"The big downside of that is," Downey continued, "they have to invest pretty significantly in server-side infrastructure to power an alternative version of their Web site. They have to buy a server farm, hard drive space, memory, the whole thing, just to power a beta version of their site. However, with Apollo, they didn't have any of those issues because they were able to design an entirely new interface for accessing eBay's information, and they were able to push all of that out onto the end user's client machine, instead of having to build any infrastructure in the back end, because all of their services are exposed as Web services. So they're able to instantly tie into their existing services without disrupting any of their Web-based content."

Though Apollo is free to the user and optionally free to developers, like ActiveX long ago, it will require a new runtime platform - not .NET, not Java, but its own beast. Meanwhile, Microsoft's Web services architecture is moving toward an XML-based layout language called XAML, which addresses functionality built into the .NET 3.0 platform - specifically, Windows Presentation Foundation. While Windows XP users would need to download .NET Framework 3.0, Windows Vista users will already have it, saving them one more step. That gives XAML an automatic advantage by virtue of Microsoft's pre-distributed user base.

Does Adobe see Microsoft's OS-leveraging strategy - which has arguably worked well before - as a challenge? "The fact that Microsoft is showing and investing in technologies that do similar things to what Apollo enables you to do," Downey responded, "really validates that there's a huge market need for this type of application...The big difference is, first of all, with a WPF application, you do have to have .NET 3.0 or Windows Vista, which as we both know, is a very, very large download...and users are dependent upon the other option of just upgrading to Vista. That may happen within a certain length of time, but it took Windows XP five years to get over 50% market penetration. So far, Vista's not doing as well as XP did. So there's a factor of, how soon do you need to be able to address more than 10 or 20% of the market with your application?"

But Apollo's other advantage, which existing developers may come to appreciate, Downey believes, lies in the fact that Adobe's architecture stands opposed to Microsoft's now-typical campaign of using its pre-installed base as a kind of crowbar for its desktop application development tools, such as C# - a tool which may no longer be attractive in a world where app developers and Web developers live in two separate camps.

"Apollo is targeted at Web developers," said Adobe's Downey, "not at native application developers. So with Apollo, you're not learning any new languages. You're not learning XAML, you're not learning C#; you're using Web technologies that you already have an investment in. You already have code libraries, you already have the skills, and you already have the tools that you need to build these types of applications. That's where Apollo is unique."