Simple shell with JIT compilation
Update example in README, add to examples/
Allow assignments in bodies, allow @d = @s


browse  log 



You can also use your local clone with git send-email.

#Quack Shell


  • Concise & simple syntax for executing functions & programs
  • Easy to define and save functions & variables (REPL-esque)
  • Make accidental security vulnerabilities hard (see bash)
  • Efficient piping of data between processes


#!/usr/bin/env qsh

print Files in directory:

ls >
split < >
print @
for file in @; (
	file @file >type
	if @file; print ">" @file is @type

fn random >
	@ = $4

fn rand_either a b >r; (
	random >
	expr @ % 2 >
	printf %c @ >
	@r = @a
	if test @ -eq 1
		@r = @b

print random                      # Prints "random"
random >; print @                 # Calls the random function and prints the return value
rand_either foo bar r>; print @   # Prints either "foo" or "bar"

print 0000    # Prints 0000
print $0000   # Prints 0
print $xff    # Prints 255


Note: This specification is not yet complete and is subject to change. The shell does not implement all features yet either.

  • Separators: ;, \n
  • Scope: { a; b; ...; ret }
    • A scope counts as a single statement
    • The last value (ret) is returned
    • Exiting early is possible with return
  • Variables: @x = 5, print @x
  • Whitespace outside quotes is ignored.


  • Strings
  • Integers
  • Pipes
  • Arrays
  • Dictionaries (TODO)


All functions return an integer status code. Other return values can be extracted using pipes.


Note: none of this is implemented. Output is collected into a string.

The main mechanism for exchanging data with programs are streams: data is continuously written and read from a stream.

in qsh, streams and strings are identical from an abstract viewpoint. In practice they have different performance characteristics: while strings are stored entirely in memory, streams instead only buffer a portion of data at any time. This makes them ideal for exchanging a large amount of data.

Estabilishing a stream from one program to another is done through pipes. Each pipe has exactly one input and one output, i.e.:

# This is fine
read a.txt 1>
write b.txt 1<

# The stream from `read a.txt` is lost
read a.txt 1>
read b.txt 1>

# `write b.txt` will not write any data since `write a.txt` wrote all of it
write a.txt 1<
write b.txt 1<

A newly created pipe always has its input attached. If the input becomes disconnected, the pipe becomes unusable.

#Special variables

To keep scripts short, some variables automatically get a value assigned. These variables cannot be assigned to directly.

  • @?: the return value of a call.