Abstract

This dissertation addresses the need for easier construction and extension of language tools. Specifically, the construction and extension of so-called semantic editors is considered, that is, editors providing semantic services for code comprehension and manipulation. Editors like these are typically found in state-of-the-art development environments, where they have been developed by hand. The list of programming languages available today is extensive and, with the lively creation of new programming languages and the evolution of old languages, it keeps growing. Many of these languages would benefit from proper tool support. Unfortunately, the development of a semantic editor can be a time-consuming and error-prone endeavor, and too large an effort for most language communities. Given the complex nature of programming, and the huge benefits of good tool support, this lack of tools is problematic. In this dissertation, an attempt is made at narrowing the gap between generative solutions and how state-of-the-art editors are constructed today. A generative alternative for construction of textual semantic editors is explored with focus on how to specify extensible semantic editor services. Specifically, this dissertation shows how semantic services can be specified using a semantic formalism called reference attribute grammars (RAGs), and how these services can be made responsive enough for editing, and be provided also when the text in an editor is erroneous. Results presented in this dissertation have been found useful, both in industry and in academia, suggesting that the explored approach may help to reduce the effort of editor construction.