Name: Jur Bartels
Afiliation: Eindhoven University of Technology
Abstract
Language composition is a very common approach used by language engineer used to combine features of two separate languages into one. In order to apply this approach however, the two languages need to exist within the same “world”. We can not arbitrarily compose languages where one exists as a grammar in the textual language workbench (LWB), and one exists as a model in the projectional LWB. In this project we bridge the textual and projectional LWBs by creating a process for transforming textual grammars into projectional models. Language composition will then be possible by “importing” a textual language into the projectional LWB, and applying the composition with other projectional languages there. There are three defined sub-goals. First of all, we need to be able to map the structure of source textual grammar rules to equivalent projectional models. This step also includes the problem of how information is to be exchanged between the LWBs. Secondly, we want to improve the usability of the imported language within the projectional LWB. There are differences between the textual and projectional LWBs which make languages designed for one less usable in the other. To remedy this we apply several heuristics to the imported language within the projectional LWB, which improve the “default” usability. Finally, after the language has been imported into the projectional LWB, we also want to be able to import programs written for the textual source grammar as projectional models. This removes the need to recreate already written programs within the projectional LWB manually. In the end, we aim to create a tool which allows for the bridging of both LWBs with the least amount of manual effort required.