As all (relatively) new techniques, we are still in the process of learning when and how creating a DSLs is the best solution for a given software development scenario. Otherwise, we may end up following again the technology hype cycle (as already happened with UML; we are still paying the consequences) and fall trap of a DSL fever (again, same happened with the UML fever so let’s try not to make the same mistakes this time).
So, can I suggest you to read these four posts before starting the creation of a new DSL?
- Building DSLs: who, how and when
- Trapped
- Never design a language
- UML is a MDL (many-domains language)
DSLs can be a wonderful tool in your modeling arsenal but only when they’re properly used.
