Frustration with RIA platforms
There has been a recent flurry of announcements around rich internet applications (RIA) recently, from Adobe (Apollo, open-sourcing the Flex SDK and Tamarin), Microsoft (Silverlight v1.1) and Sun (JavaFX). The battle against plain AJAX focuses on three main fronts: richer rendering capabilities, improved performance, and a comprehensive set of designer and developer tools.
But are they making much progress addressing the native challenges of the web environment?
This next generation of contenders to the RIA throne is impressive.
But I can't shake the feeling that this is more of the same. These platforms take solutions that have been figured out in the desktop world and incorporating them into the web world. You can now design, code and debug a really rich application in a web page, using state-of-the-art tools.
The release of these frameworks is no small feat, especially to achieve smooth integration across the different browsers.
But they do little to address the native problems of the web.
Here are some examples:
- integrating web sites and services,
- cross-domain mashups with private data,
- access control, identity (for users and entities) and charging for web services,
- user-driven website customization and extension, safe client-side composition.
In other words, the recent wave of RIA platforms make things look nicer and run faster, but we are still left off with the non-semantic markup, the page model, and the same-domain policy.
We have more glitter, more power and more tools, but limited innovation in support of data exchange (such as microformats or Live Clipboard), user customization (like greasemonkey), cross-domain integration (like the Windows Live contacts gadget cross-domain iframe hack or cross-domain XMLHttpRequest), or extensible browser services (for web application code to run longer than the page that loads it, for applications such as chat or P2P).