Tuple-Space Mapper: Design, Challenges and Goals

Abstract

Library-based and pre-compiled visualization tools incur many penalties that hinder the adoption of visualization as a technique for many applications. Libraries necessitate familiarity with the data structures and control flows that are incumbent in traditional programming, but not central to visualization. Task-specific visualization applications alleviate these needs, but induce users to move data between applications as their needs change. We propose a generative programming approach to visualization tool construction based on domain specific languages. We believe this approach will simplify the process of visualization creation, simplify integration with many tools (reducing error-prone conversion steps) and ease the modification of existing tools (reducing the pressure to introduce new tools as needs change). In this paper, we present an architecture for a such a tool, outline the technical requirements and challenges and argue for its relevance to improving the adoption of visualization.