Are language workbenches dead?
Meinte Boersma
34

I don’t even think they smell funny. From my personal experience they are more alive then lets say 3 or 4 years ago. I think were they don’t shine as you mentioned is when you target people who would describe themselves as a developer. Because there they often don’t provide much more value than an internal DSL in Ruby, Groovy, etc. but suffer from all their problems and feel like an alien in the normal developers workflow.

I think for these use cases there are some ideas that gained traction in “normal” programming languages ecosystems. For instance Type providers in F#, design of Rosly or the Rust compiler which expose the AST to compilers plugins or even Babel which gives you the same for JavaScript. Sure theses tools aren’t fully mainstream yet but I think they solve the purpose of supporting software developers far better than a complete language workbench would do.

One aspect that you missed completely is that models provide other benefits than just “writing less code”. For instance the fact that you can use model checking to prove certain aspects. Proving the the state machine of a pacemaker is correct can provide huge value to the business. And these models if their abstractions are chosen good are independent of the technology that you use to “meterialise” them. Having your business rules that describes a tax law independant of your software architecture for instance.

For these more business domains many of the current LWBs still fell to IDE like and scare people away. Even MPS with its projectional editor (which business people often like in contrast to software developers) is a problem because of its IDE chrome, sidebar and tool windows. And ideally you would want to provide the language on the web instead of a standalone application.

While I first thought I would oppose your oppinion in this tropic, after writing this I essentially realised that we are both seeing the sample problem to some degree.

Let me conclude I think LWBs are happy and alive but should focus more on pleasing “the business” and should go to the web.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.