RetroSaurus

June 7, 2026 · View on GitHub

A retro Windows 3.1-styled thesaurus and dictionary for your desktop — classic macOS Dictionary.app, reimagined with the chrome of 1992. It ships the Open English WordNet baked right into the binary, so it works fully offline with zero setup.

RetroSaurus showing the entry for "abandon"

Built in Rust on the Saudade toolkit, alongside its siblings RetroCalc and Git Journey.

Features

  • 135,969 words with definitions, examples, synonyms, antonyms and related words — every sense grouped by part of speech.
  • Live prefix search as you type, backed by a finite-state-transducer index.
  • Clickable cross-references — jump from a synonym straight to its entry, with Back / Forward history (browser style).
  • Fully offline — the whole word index (~15 MB) is embedded in the executable. No data files, no network, no setup.
  • Keyboard-driven and resizable, with authentic Win 3.1 chrome.

Install

cargo install retrosaurus

The first build downloads the WordNet source (~11 MB) once and compiles it into a compact index baked into the binary. See Building.

Run

retrosaurus

Type a word, pick a sense from the right-hand pane, and click any underlined word to follow it.

Keyboard

KeyAction
Ctrl+F / Ctrl+L / Alt+WFocus the word field & select all
/ Move the selection / scroll
Alt+←Back
Alt+→Forward
Ctrl+RRandom word
TabCycle search → list → definition
Ctrl+QQuit

Building from source

git clone https://github.com/roblillack/RetroSaurus
cd retrosaurus
cargo build --release

On the first build, build.rs fetches the Open English WordNet 2025 WN-LMF source from en-word.net (via curl/wget), verifies its checksum, and compiles it into $OUT_DIR/retrosaurus.dat, which the binary embeds. To build offline or in CI, point it at a local copy instead:

RETROSAURUS_WORDNET_XML=/path/to/english-wordnet-2025.xml.gz cargo build --release

Regenerate the README screenshot with:

cargo run --example screenshots

Word data & license

RetroSaurus's code is licensed under the MIT License.

The bundled word data is the Open English WordNet 2025, © the Open English WordNet community, distributed under the Creative Commons Attribution 4.0 License. RetroSaurus redistributes it unmodified in a compact binary index; the attribution also appears in the app's Help ▸ About dialog.