This page permanently redirects to gemini://gemini.ctrl-c.club/~nttp/toys/ripen/.

Ripen scripting engine

Welcome to Ripen, a stack-based scripting engine with highly readable syntax, that can be implemented with little code and effort in a language like C++ or Nim. It's informed by RetroForth and PostScript, but diverges from either in important ways, not least due to its string-threaded nature.

Ripen is based on principles explained in the article linked below:

=> More about tiny scripting engines

The point is having a single-file engine, easily copy-pasted into any application. For this reason, the default vocabulary is kept small. See below for more details, but for a quick start try:

	( Hello, world! ) type cr
	3 2 + 3 2 * < . cr

=> Language overview

News

• 20 July 2023: Version numbers were unified across ports, and compatibility improved.
• 2 March 2023: C++ port has been refreshed and relicensed.
• 1 March 2023: Nim port has been revived and relicensed; see below.
• 10 February 2022: Ripen was ported to C++ as "version 2.0".
• 20 December 2021: There's now an offshoot called Guava, by eli.
• 11 May 2021: Ripen (1.0) is used as the scripting language of Tipsy Turtle.

=> Guava | Tipsy Turtle

Beware that version 3.0 introduced breaking changes. Check the documentation.

Download

C++ port version 3.0 beta (2023-07-18); Boost Software License:

=> 64-bit Linux build (38K) | 32-bit Linux build (39K) | source code only (14K)

Nim port version 3.0 beta (2023-07-19); MIT License:

=> 64-bit Linux build (90K) | 32-bit Linux build (95K) | source code only (37K)

All archives include documentation + examples. See also:

=> old Python prototypes (13K); for educational purposes

Building and running

Ripen has multiple implementations; notes and/or build scripts are included with each source archive.

Binaries are built on a Debian 10 machine, and seem to have decent compatibility with other / older distributions. No promises.

Performance

The C++ port of Ripen is smaller but also slower than the Nim port. An informal test was performed with a modified version of the Rugg/Feldman benchmarks (#5), ported like this (yep, it's a one-liner):

	:nop ; { $i 2 / 3 * 4 + 5 - =a nop } 1000 times $a . cr

While timings for this are hard to get on a modern PC, Ripen appears to be faster than Python 2.7, and even Tcl. Of course, both are much more complex.

=> Rugg/Feldman benchmarks (Wikipedia page)

Differences between versions

In the Python version, everything goes on the stack: numbers, strings and lists / code blocks. In C++ and Nim, lists and strings go to their own memory areas. That requires a number of additions, like the ? sigil used to show the definition of a user-defined word, and .. to display the contents of the scratch pad.

The C++ port only defines included in the stand-alone interpreter.

Frequently Asked Questions

What do you have against parsing words?

Nothing at all! My very first Forth interpreter, that I made in 2009 for a tutorial, had parsing words and even quoted strings. This time I simply wanted to try something different after seeing RetroForth.

Does Ripen count as single-file anymore? The C++ code consists of two files.

The library proper is still just one file, ripen.hpp; the other one is a demo driving app so you can see that it works, and how to use it in your own projects.

Roadmap

Version 4.0(?)

Main goal: find ways to make the language more immediately useful.

• remove / sigil;
• deprecate clear word;
• reconsider true and false words;
• make aliases more powerful;
• add + and - sigils;

Support

Questions and feedback are welcome. All archives include contact information. See my homepage as well.

=> More small things | No Time To Play