Dataflow programming - Wiki

Dataflow visual programming languages (DFVPLs) are related to pure functional programming languages (such as the typed lambda calculus) presented in point-free style (as in a cartesian closed category). In practical terms, there is some overlap with data provenance and workflow management.

Languages

Practical

LabVIEW / G language
Knime
- Control flow : support for loops and recursion
Blueprints : visual scripting for Unreal Engine 4

Research

Lucid
- An influential text-based dataflow language
Hume
- Integrated with Kissinger’s Quantomatic
- Uses box calculus, related to monoidal categories and cited in the Quantomatic paper (arxiv)
- Michaelson & Grov, 2012: Reasoning about multi-process systems with the box calculus (doi, pdf)
Skov (GitHub , HN )
- Based on concatenative , text-based language Factor (GitHub , tutorial )
- As of June 2017 , the “new Skov ” has tree-based visuals closer to textual syntax
- The “old Skov ” with graph-based visuals (boxes and wires) is closer to wiring diagrams and more interesting to me
Full Metal Jacket (HN )
- Functional, visual dataflow language, inspired by and implemented in Lisp

Literature

Informal

LtU discussions (1 ,2 )
- Comment on flow-based programming
- Comment on category theory and tension between data flow and control flow
HN discussion on pro’s/con’s of visual programming
MO question on graphical languages

General

Johnston et al, 2004: Advances in dataflow programming languages (doi, pdf)
Mosconi & Porta, 2000: Iteration constructs in dataflow visual programming languages (doi, pdf)
J. Paul Morrison, Flow-Based Programming, 2nd ed. (website )