Joel Spolsky posted an essay the other day drawing parallels between the optimization-in-assembly obsessed application developers of the late 80s and the optimization-of-download-and-compile-time obsessed web application developers of today. He proposes that someday soon someone (perhaps a “bratty Y combinator startup”) will come up with NewSDK, a rich SDK and language that does all sorts of fancy AJAXy things. Because they don’t have to convince anyone it’s a good idea, the brats will pay far more attention to functionality in NewSDK than they pay to performance, so when it first comes out the performance will be terrible. Existing AJAX apps (like everything at Google) will ignore NewSDK because of the performance problems, completely forgetting about Moore’s Law:
So far adoption of GWT has been fairly small. Three Rings is using it for Whirled. They use GWT to wrap the flash applet that does the heavy graphical lifting. They are able to call back and forth between the two so they appear to the user as one application. As the SDK improves over time, more people are bound to try it out, and it only takes one successful application for it to really take off.
All this kind of makes me wish I had a reason to write a major web app. As long as you survive, it’s good to be an early adopter of the Next Big SDK.