Domain specific language on automata: Design and Implementation problems

Even if Domain Specific Languages (DSL) are not well-known, they are quite used. Indeed, these languages, which are built especially for a domain or a problem, can be a powerful alternative to General Purpose Lan-guages (GPL) for their domains. Our problem is to manipulate automata easily and quickly in order to test algorithms, whether the automaton is the goal or the tool of the algorithm. A DSL seems to be a good solution. Two major points will be broached: • First, DSL in general will be discussed. What are their weaknesses, their interests, why and when using one instead of standard languages like C++. A methodology for building such a language will be exposed, illustrated by experiences given by MAL, our DSL concerning automata. • Next, MAL will be presented more precisely. Its syntax, its semantics were built in order to fit mathe-matical notations and way of thinking. It is one of the links between MAL and Vaucanson and other links will be shown. But design and implementation are two distinct parts, and if the language is a powerful one, its implementation is consequently harder. So difficulties will be exposed too