Sebastian Just has written about how to select an Ajax framework. He lists a number of questions that architects should answer in helping them narrow down the list of available frameworks, including community and corporate support, learning curve, and whether it fits the type of user the site has.
The questions he suggests include:
Structural Javascript enhancements? Javascript can become a maze if your team does not share a common development methodology . There are so many ways of doing the same thing ( like creating objects of accessing the DOM ) in JS, that you must formalize the way to do it. Such a framework may help via on-demand Javascript, packaging abilities, enhanced OOP...
Re-usability of your written components? Making use of it shall help you in re-using written components in your next projects.
Framework current documentation level? Often low for most projects, you must take care of it
Features you need? You may look at your projects needs ( or near future needs ) and see if the Framework can handle them. Some are GUI oriented, special-effects oriented, Javascript centrics, communication only oriented. None will completely cover your requirements. What is the complexity degree in making the framework evolve ? can you add your own features in the framework ? will you help committers ? Can you mashup multiple Frameworks ?
How long will it last? Most projects are companies own project spin-off. The Framework evolution depends on their ability to maintain and make the project evolve. A good user community is important as well as talking to developers when needed further functionalities. Taking care of sponsors and the framework current uses in production web sites, is a good start in guessing if the Framework will last years or will be stopped in six months.
In a discusion on Ajaxian some other issues are raised including testability and code quality.
An alternative approach to comparing frameworks was covered in December.