From domain-specific languages to domain-specific environments

点击翻译按钮获取中文摘要

Software engineers use various tools and techniques to create programs. Generally, they rely on integrated development environments (IDEs) that offer a complete and unified set of concepts, techniques, and tools to cover the entire software development process. Nowadays, IDEs are no longer restricted to software practitioners: many are dedicated to specific domains of expertise, such as Max/MSP for sound engineers and AutoCAD for architects. These domain-specific IDEs provide modes of interactions and workflows adapted to their domain-specific users to develop models in their respective domain. To minimize the development effort of domain-specific IDEs, several generative techniques, such as model-driven engineering (MDE), can generate IDEs for domain-specific languages (DSLs). DSLs are small programming or modeling languages dedicated to a domain, such as a database model, to generate SQL databases. The benefit of DSLs is to tailor modeling languages to the domain of its users. MDE provides a sufficient level of automation to adapt the editor/canvas component of the IDE to the DSL, by constraining the user to build models specifically for that DSL. However, these techniques exclusively rely on the syntax of the language. Generated IDEs still suffer from a significant learning curve by users who need to adapt to the IDE, especially for domains where users are not programmers. Typically, the generated IDE is a customization of a general-purpose programming or modeling IDE. These customized generic IDEs impose some predefined procedures and interactions, familiar to programmers, to which the non-programming user must adapt. The goal of this research program is to change drastically how we produce domain-specific IDEs by making them adapt to the user of a DSL rather than forcing the user to adapt to the IDE. Our long-term vision is to allow users to program and model in a completely immersive environment that adapts to their needs, habits, and preferences. This pushes the boundaries of user interaction possibilities with the IDE. The working habits of the user become embedded in the IDE. This reduces the learning curve and error proneness in using the IDE, which ultimately improves the productivity of the user at his task. We propose to enhance the generation of domain-specific IDEs to take into account the pragmatics of the DSL, relying on an explicit specification of the user interaction with the IDE. Users are immersed in an IDE enriched with smart editors that minimize their cognitive effort by suggesting correct-by-construction programs/models through a new generation of IDEs. This allows users to specify the programs and models in representation styles that go beyond textual and graphical editing. We extend the interactions to other dimensions, such as multi-touch gestures, movements in space, and interactions with other hardware devices.