Zero-Width Markers for View-to-Model Traceability

Abstract

A common approach for producing graphical views from models is to transform them into textual representations that can then be rendered as diagrams using off-the-shelf diagram-as-code tools such as Mermaid, PlantUML, and Graphviz. Although this approach offers a practical and cost-effective solution for model visualisation, fine-grained traceability from generated views back to their corresponding model elements remains an open challenge. In this paper, we examine the mechanisms for fine-grained traceability provided by existing diagram-as-code tools and highlight their limitations. We then introduce a tool-agnostic technique for encoding, recovering, and navigating view-to-model traceability links using markers consisting of zero-width Unicode characters. Our evaluation against more than 25 diagram-as-code tools shows that the proposed approach is applicable to most of them and enables efficient fine-grained view-to-model traceability with minimal impact on model-to-view transformations.