Eclipse Platform e4 Just Getting Off the Ground
- By John K. Waters
The Eclipse Foundation offered attendees at last week's EclipseCon 2008 conference an early look at plans for the next generation of its Eclipse distribution and platform: Eclipse 4.0, better known as "e4."
What exactly is e4?
"The sound bite is that it's the next generation of the platform," said IBM's Mike Wilson, Eclipse Project Management Committee (PMC) member and leader of the Eclipse Platform and Incubator subprojects. "At this point, what that means is going to be defined by whoever gets involved in working on it. We're open to everything, as long as it's targeted at building a better platform and addressing the most pressing issues that are likely to impact the ongoing success of Eclipse."
The biggest issue on Wilson's Eclipse radar: the Web.
"I think it's very important for the Eclipse code base to move out to the Web, and get involved more in that space, because the world is changing," Wilson told his audience. "A lot of business software development is moving to Web-base UIs backed by services. We're also seeing the IDE world moving in that direction."
To get e4 the ball rolling, the Eclipse Platform team (led by Wilson) joined forces with the Eclipse Rich Ajax Platform (RAP) team, Wilson explained. RAP is an AJAX runtime based on the Eclipse Rich Client Platform, a framework for rich Internet application development. RAP allows developers to build rich Internet applications and programs entirely in Java, and to use the Eclipse plug-ins to modularize their applications.
Wilson was joined onstage by the leader of the RAP project, Jochen Krause, CEO of Innoopract. Also joining him were IBM's Steve Northover, PMC member and principal architect of the Standard Widget Toolkit (SWT), as well as Code 9 Founder Jeff McAffer, PMC member and leader of the Eclipse Equinox OSGi, Rich Client Platform and Orbit subprojects.
E4 is not due for two years, so these are early days for the project, said Eclipse Foundation Executive Director Mike Milinkovich. "It's new code, a new vision, and new opportunities," he said. "But it's just getting off the ground."
However, concerns exist among Eclipse community members that IBM, which released its once-proprietary framework to open source more than five years ago, has too much influence over e4. Last week, the Eclipse Foundation released a list of e4 committers: 17 currently work for IBM, one is a former IBM employee, and three work for German software company Innoopract. The majority of those submitting early code to the project are current or former IBM workers, and employees of Big Blue continue to make up the majority of contributors to the Eclipse platform.
Code 9's McAffer (a former IBM employee) addressed those concerns, which were exacerbated by an early code submission that made it look like IBM participants were taking over the e4 project. That code submission was not finished code, he said, but "intended to start the discussion" about where the e4 project would go.
Concerns about the relatively large amount of resources IBM commits to Eclipse projects are not new, said Milinkovich.
"It's a very mature, very complex code base," he said. "From a purely technical point of view, it's tough to get into it. Also, the development team from IBM has been working together for years, full time on this project. So you've got an extremely competent team with high standards and expectations of one other, and a complex and mature code base. Those two things together make it hard to get in."
Milinkovich added that "E4 gives us the opportunity to get off to a new start because it will be new code. It's an opportunity to get greater diversity into the community."
"It's actually very hard to get involved in that code base, and we really don't have enough community participation," Wilson said. "There's a lot of preexisting code written by a small team. We've gotten to the point now where that code has gotten baroque, and it's very difficult to surgically modify the code."
The PMC is planning to have a working concept prototype for next year's EclipseCon, Wilson said.
At this point, e4 is little more than a community gathering point where early code changes and ideas can be tracked. Current goals for the project include improvements to scripting, new tools to help Java programmers build plug-ins, an improved separation between the interface and data through a DOM, and possibly a scripting engine and support for Cascading Style Sheets. Plans also include a well-described set of services delivered using a RESTful architecture.
With all of the buzz around e4, Steve Northover sought to allay fears about support for the 3.x versions (particularly 3.4, Ganymede). He said that they would not be abandoned, and in fact, development would continue for about five more years, with many of the features developed for version 4.0 appearing in 3.x releases.
John K. Waters is a freelance journalist and author based in Palo Alto, CA.