The Eclipse Modeling infrastructure and tools have evolved to be a de-facto standard in the context of modeling in general: lots of important and interesting systems have been built with Eclipse technologies such as EMF, Xtext, and GEF.
However, over the last few years, two trends have developed. The first one concerns the web. Modeling tools, like all other tools, will have to run in the browser and the cloud. The viability for rich clients is in rapid decline. Second, the notion of “modeling” itself has evolved. In particular, more and more “modeling” tools are targeted at non-technical users. We have used modeling technologies successfully to develop business DSLs for insurance contracts, medical algorithms, or banking. Features near and dear to developers’ hearts, such as powerful IDEs, flexible version control, and the edit-build-run cycle are usually not valued by the non-technical users of these systems. The current Eclipse Modeling ecosystem offers no tools or technologies to convincingly address these two trends.
In the first half of this talk we will elaborate on these challenges. While the first one is probably undisputed, the second one needs more convincing: we will show examples from a couple of non-technical modeling systems to substantiate the claim.
The second half of the talk will outline a vision of where we think modeling needs to go. This vision will include technical aspects, such as: how can existing Eclipse Modeling technologies be ported to the browser, and what can we learn from non-Eclipse tools such as MPS. Just porting technologies to the browser is not enough though -- we will thus also outline a set of architectural and usability considerations for tools that will allow modeling to move forward over the next decade.