← Back to news

Mixing Visual and Textual Code

Original article
arxiv.org|60 points|40 comments|by doppioandante|Jun 25, 2026

Mixing Visual and Textual Code

Authors:

  • Leif Andersen
  • Michael Ballantyne
  • Cameron Moy
  • Matthias Felleisen
  • Stephen Chang

📄 Document Metadata

AttributeDetail
arXiv Identifier2603.15855
Primary CategoryComputer Science →\rightarrow Programming Languages (cs.PL)
DOI10.48550/arXiv.2603.15855
Publication StatusForthcoming in JFP
Licenselicense icon

🎯 Abstract & Core Thesis

The authors argue that current mainstream programming languages are limited because they rely exclusively on linear text to represent complex, domain-specific geometric concepts.

The Goal: To develop hybrid languages that empower developers to build visual syntactic elements, allowing for a seamless blend of textual and visual syntax specifically tailored to a given application domain.

To be successful, this integration must meet several strict criteria:

  • Parity: Visual and textual syntaxes must be treated as equals.
  • Workflow Integration: The system must not disrupt the standard programming experience.
  • Composability: Visual extensions must be proper language extensions that work with other features.
  • Analysis: The ability to perform static reasoning about the program must be maintained.

🛠️ The Solution: Hybrid ClojureScript

The paper introduces Hybrid ClojureScript, the first language to implement these requirements. This system allows developers to:

  1. Define visual interactive syntax.
  2. Embed instances of this syntax directly within the program's source text.

How it behaves across environments:

  • Enhanced Hybrid IDEs: These render the embedded syntax as mini-GUIs for direct interaction.
  • Standard IDEs: These simply display the textual representation of the visual syntax.

Conceptual Model

The relationship between the source and the representation can be viewed as: Program=∑(Textual Code+Visual Constructs)\text{Program} = \sum (\text{Textual Code} + \text{Visual Constructs})

📝 Implementation Example (Conceptual)

While the paper discusses the framework, a conceptual implementation of a visual block in Hybrid ClojureScript might look like this in a standard editor:

(defn draw-scene []
  (let [circle (visual-syntax :circle {x 10 y 20 r 5})] 
    (render circle)))

In a hybrid IDE, the (visual-syntax ...) call would be replaced by an actual draggable circle.

📈 Paper Objectives & History

The authors utilize this work to:

  • Argue why such an extensibility mechanism is essential.
  • Prove that the design is adoptable.
  • Explore the requirements for porting this design to other programming languages.

Submission Timeline

  • v1: Submitted on Monday, 16 Mar 2026 by Stephen Chang.
  • v2: Revised on Friday, 27 Mar 2026.