Word on the street is that Adobe is gathering feedback for the release of AIR 2.0. I think AIR is and was a great addition to the Flash platform. There following are things I feel AIR really needs to succeed as it continues to mature:

• Framework embedded executeables (think projectors)
• A fully customizable AIR installer for one click installs
• The option to use a third party installers
• C level extensibility
• Less branding

When I run and app it feels the most “real” and “professional” to when I’m unaware of its underlying technology. The average user doesn’t care what their app is built with, just that it works well and looks nice. Sadly AIR currently lacks transparency and I can’t give this experience with Flash. I would have to build an application in a low level languages or a scripting languages where apps can be distributed with an embedded runtime and then use a scriptable installer to make per platform distributables. Visually none of these technologies are as advanced as Flash and Flex, and they certainly aren’t as rapid when it come to development. This is all the more reason why AIR needs to be able to produce “real” applications.

The desire of developers who aren’t hard-core native application developers is to make traditional desktop apps using technologies that they’re already familiar with. They want to use non-traditional means, but still compete in the same marketplace as traditional app developers. The issue is that apps that are created using these technologies often lack transparency and are constant reminder that they aren’t truly traditional apps. In the case of AIR, the air specific installation process makes a clear distinction between and AIR app and a standard app. As a user the AIR managed installation process is quite unlike that of any other application you’ve ever installed. This sort of thing makes it very difficult for the user to suspend his/her disbelief that this is actually a real application (much in the same way that Photoshop or Word are real application).

The next issue is runtime distribution. Download an popular app, Firefox for example. Run the installer. At any point during this process were you asked to download and install yet another file in order to run it? No, and this is the experience that people have come to expect when installing something. Consequently this is the experience AIR (and others) should emulate. Having to explicitly download a runtime and wait through a framework specific installation process in order to use an app and two more steps than most non RIA enthusiasts are willing to take.

The need for C-level extensibility in an application framework is a pretty easy one to make. Any and all intensive calculations are better off-loaded to low level languages. Extensibility also allows for custom functionality beyond what the framework provides.

I certainly don’t seek to discredit AIR in its current form (at least not entirely). One click web installs makes deploying desktop apps that provide out-of-browser access to online services dead simple, and its cool. After all, Flex is a RIA technology so it was only natural that the current model of AIR was RIA centric. My hopes, however, are that the next steps for AIR will be more in the direction of a full fledged application framework.

ian Air