Eclipse Juggernaut Reinvents Server-Side Development!

The Eclipse Juggernaut is about to make yet another land grab. After successfully conquering the rich client space ( that goes beyond Java development ) it is now setting its sites on an even richer and ambitious goal. That is, the Eclipse platform now intends to mandate how you define components on the Server-side.

Wolfgang Gehner wrote in January about the budding potential of the Equinox project as a basis for building server-side components using the Eclipse component model. Folks now want to take that idea to the next level by proposing new Eclipse projects called "Enterprise Component Platform", the "Rich AJAX Platform" and the "Rich Server Platform".

One can read the Enterprise proposal and easily conclude that it looks like a re-invention of the wheel. That is, it appears to be yet another model to compose server-side Java components. In fact, the proposal isn't even a bit shy when it states:

The initial project would not focus on compatibility with the existing enterprise component models such as EJB or currently developing such as SCA.

In otherwords, the hell with EJB 3.0 and the hell will lighweight containers like Spring, we have an even better model and it could just possibly put the Java development world on it head. We see re-invention all the time in the development world, how would this be any different? The question one needs to ask oneself are thess: "Is the Eclipse component model vastly superior to EJB 3.0 or Spring?" and "Is the Eclipse component modely vastly superior to JSF or Portlets?".

The second question is definitely easier to answer. Mashups and mixins are much easier to accomplish on the user interface side. The RAP proposal therefore has good odds on succeeding. However, "Integration at the Glass" isn't always enough. One would therefore need to take a leap of faith that the Eclipse component model applies equally well for business logic.

I wrote a piece once about "what Java needs to do better". My last recommendation was for a meatier web application stack. This sort of implies the need for an application level component model where complete applications can be stitched together in as a seamless solution. Billy Newport expounds on this problem more extensivley, he writes:

The more I think about it, the more I think that software companies need to make off the shelf applications or coarse components that can integrate even with competitors products without a lot of hassle/coding/understanding.

Unfortunately, the current Java component model specs (i.e. JavaBeans, EJB 3.0, JSF, Portlets) are seriously deficient in this area. They cover just a small subset of the larger problem.

Fortunately, the Eclipse component (a.k.a. plugin) model (a.k.a architecture) has aptly and convincingly demonsrated how one would acheive this seamless integration albeit in the client side. When you take this to the next level, that is on the server side (where incidentally that's where all the action is happening these days), you have the makings of an "Architection of Participation". This time however, it's a component model that goes beyond a low-level binding model. Surely, I would be dissapointed if this would not take a part in revolutionalizing Java server-side development.