Funge++

Funge++ is Befunge interpreter originally written by Conlan Wesson in Java for a Programming Language Concepts course with minimal functionality. It has since been converted to C++, supports all Funge-98 instructions and N-dimensional Funge. The Funge++ handprint is 0x464E2B2B (FN++).

Running

The Funge++ executable funge uses the first file argument to load the funge program. The first character of the file is an instruction at (0,0). Each subsequent character is an instruction at X+1. Both carriage returns and line feed (as well as CRLF) are treated as Y+1 and reset X to zero. Form feeds in the file represent Z+1 and reset X and Y to zero. Carriage returns and line feeds following a form feed are ignored (i.e FF may be own its own line).

See the man page for more information.

Higher Dimensions

Vertical Tabs in the file represent 4D+1 and reset X, Y, and Z to zero. Carriage returns and line feeds following a vertical tab are ignored. This is not part of the Funge-98 specification, but provides a simple way to create 4-dimensional funges.

Trefunge frunges can be loaded as funge-lib formatted .fl files.

Higher dimension funges can be loaded as BeQunge formatted .beq files. Loading a .beq file implies -lNFUN.

Features

Native Threads

The split instruction t can use native threads rather than the tick mechanism described in the Funge-98 specification. This mode can be enabled with -fthreads=native.

Auto-Dimensions

The dimensionality (up to 4D) is determined automatically based on the file contents. This can be overriden with the -std argument. Funge-lib formatted files are 3D. Higher dimensions can be determined automatically from BeQunge formatted files with or without the Dimensions directive.

Standard instructions which operate on vectors (g, p, x, ?), operate with vector lengths equal to the number of dimensions detected. This enables those instructions to function in a standardized way in higher-dimension funges. For example, in 4D mode, ? will change the delta to one of 8 possible directions, while x can move in all possible directions.

C-Style Strings

C-style strings can be enabled with -fstrings=c. In this mode, backslashes in strings are not pused to the stack, instead, it and the following character are interpretted as an escape sequence. This happens in one tick. In addition to special characters, this allows quotes and ANSI escape codes in strings.

64-bit

Funge++ stack and funge-space are 64-bit values. For that reason, FPDP numbers use a single stack cell, and LONG numbers are 128-bit.

Fish and Starfish

Funge++ also supports ><> (Fish) and *><> (Starfish). These can be selected using -std=fish or -std=sfish respectively. Alternatively, Funge++ will automatically detect the languge if a .fish or .sf file is loaded.

Errata

High/Low

The Funge-98 specification is inconsistent about the delta of the h and l instructions (https://github.com/catseye/Funge-98/issues/10). Funge++ uses thedefinition in the program flow section, that is h is "delta <- (0,0,1)" and l is "delta <- (0,0,-1)". This makes it compatible with BeQunge and Rc/Funge-98. The -finvert-hl argument flips this behavior.

Fingerprints

Funge++ supports the following fingerprints. These can either be loaded at runtime by the ( instruction, or by specifying the -l command line argument.

Funge++ also supports dynamic Funge fingerprints. When a fingerprint is loaded, Funge++ will search for funge-lib files in the fing directory before loading a built-in fingerprint. This allows users to override the built-in fingerprints. Instructions from dynamic Funges run with the same stack as the IP that called them.

BASE I/O for numbers in other bases.

BITW Bitwise Operators.

BOOL Boolean Operators.

CPLI Complex Integer Extension.

DBUG Debugger Control.

DIRF Directory Functions.

FING Operate on single fingerprint semantics.

FISH ><> and *><> commands.

FIXP Fixed point math functions.

FPDP Double precision floating point.

FPRT Formatted print.

FPSP Single precision floating point.