Dax

Dax is a very small, self-hosted Lisp for JavaScript. It provides a flexible compilation environment with an extensible reader, macros, and extensible special forms, but otherwise attempts to match the target runtime environment as closely as possible. You can get started by running npm start.

Introduction

Every piece of code in Dax is an expression, and expressions can be evaluated to give values. Dax has a few kinds of expressions that evaluate to themselves:

> 17
17
> -5e4
-5000
> true
true
> false
false

Strings are enclosed in quotation marks, and may contain newlines. Special characters are escaped using a backslash:

> "hi there"
"hi there"
> "one
two"
"one\ntwo"
> "a\"b"
"a\"b"

nil represents nothingness, and it isn't printed back at the command line:

> nil
>

Comments start with ; and continue through the rest of the line:

> 7 ; everyone's favourite
7

Lists contain other values, and are written by enclosing expressions in parentheses. Operators are called by placing them at the beginning of a list expression, and list values can be constructed using the list operator:

> (+ 10 2)
12
> (abs -10)
10
> (/ 128 2 2 2)
16
> (list 1 2 3 4)
(1 2 3 4)

Lists can contain values that are identified by their position, as well as values that are identified in the list by a name:

> (list 1 2 3)
(1 2 3)
> (list 1 2 a: 10 b: 20)
(1 2 b: 20 a: 10)

Note that values identified by name, known as keys, don't show up in any particular order.

Objects can be indexed using .dot notation or [index] notation:

> ((list 1 2 3) [0])
1
> ((list 1 2 3) .0)
1
> ((list 1 2 3) [+ 1 1])
3
> ((list a: 10 b: 20) .b)
20

A shortcut for a key whose value is true looks like this, called a flag:

> (list :yes)
(yes: true)

Variables

Variables are declared using define and define-global. Variables declared with define are available for use anywhere in subsequent expressions in the same scope, and define-global makes them globally available.

> (define-global zzz "ho")
> zzz
"ho"
> (do (define x 10) (+ x 9))
19

do evaluates multiple expressions, and itself evaluates to the value of the last expression.

Variables for a limited scope are introduced using let:

> (let (x 10 y 20)
    (+ x y))
30
> (let x 41 (+ x 1))
42
> (let x 1
    (let x 2
      (print x))
    (print x))
2
1

You can see that let accepts a list of names and values, called bindings, or it can work with a single binding. More than one expression can follow the bindings, which works like do:

> (let (x 10 y 20)
    (print x)
    (+ x y))
10
30
> (let x 9
    (print "hi")
    (let y (+ x 1) y))
hi
10

Assignment

Variables and list values can be updated using set, which evaluates to the value that it updated:

> (let x 10
    (set x 15))
15
> (let x 10
    (set x 20)
    (+ x 5))
25
> (let a (list 1 2 3)
    (set (at a 1) "b")
    a)
(1 "b" 3)
> (let a (list foo: 17)
    (set (a .foo) 19)
    a)
(foo: 19)

Conditionals

Conditional evaluation is done using an if expression. The value of an if expression is that of the branch whose condition evaluated to true:

> (if true 10 20)
10
> (if false 10 20)
20
> (+ (if false 10 20) 5)
25

if expressions can have any number of branches:

> (if true 10)
10
> (if false 10)
>
> (if false 1 false 2 false 3 true 10)
10
> (if false 1 false 2 false 3 10)
10
> (if true 9 (do (print 10) 11))
9
> (if false 9 (do (print 10) 11))
10
11

Comparing values is done using the = operator:

> (= 10 10)
true
> (if (= 10 10) "yes")
"yes"
> (if (= 10 "no") "yes")
>

Lists are values that have unique identity, so two separate lists that happen to contain the same values are not the same:

> (= (list 1 2 3) (list 1 2 3))
false

Functions

Functions in Dax are values, just like numbers and strings. Expressions that start with fn evaluate to functions:

> (fn () 10)
function

Functions can be called by placing them first in a list expression. The list that appears after the name fn identifies the function's parameters:

> (fn (a) (+ a 10))
function
> ((fn (a) (+ a 10)) 20)
30
> ((fn () 42))
42
> ((fn (a b) (+ a b)) 10 20)
30

Because functions are values, we can use variables to name them. The same rules apply when calling a function named by a variable:

> (let f (fn () 42)
    (f))
42
> (let plus (fn (a b) (+ a b))
    (plus 10 20))
30

The most common shortcut for defining functions is to use define and define-global in the following way:

> (define-global f (n)
    (* n 10))
> (f 3)
30
> (do (define f (n) (* n 10))
      (print (f 3))
      (print (f 4))
      (f 2.5))
30
40
25

A function's parameter list can contain both positional and key parameters, where the key's value is the name to bind:

> (let f (fn (a b: my-b) (+ a my-b))
    (f 13 b: 2))
15

If the key's value is true, the same name as the key is used to bind the parameter's value. This makes it easy to define named parameters with flags:

> (let f (fn (a b: true) (+ a b))
    (f 1 b: 2))
3
> (let f (fn (a :b) (+ a b)) ; use a flag
    (f 10 b: 20))
30

Parameters in Dax are always optional, and those without a supplied argument have the value nil:

> (let f (fn (a) a)
    (f))
>
> (let f (fn (:b) (if (= b nil) 10 20))
    (f a: 99))
10

Functions can also take a variable number of arguments by either specifying a single parameter instead of a list, or by using the rest key:

> (let f (fn xs (last xs))
    (f 1 2 3))
3
> (let f (fn (a rest: as) (+ a (last as)))
    (f 10 11 12 13))
23

Destructuring

Variables can be bound to values in a list at certain positions or key:

> (let ((a b c) (list 1 2 3))
    b)
2
> ((fn ((a b c)) c) (list 1 2 3))
3
> (let ((a b: my-b) (list 1 b: 2))
    my-b)
2
> ((fn ((a b: my-b)) (list a my-b)) (list 1 b: 2))
(1 2)
> (let ((a :b) (list 1 b: 2))
    b)
2

The rest key works with destructuring as it does with function parameters, which binds the remainder of the list:

> (let ((a rest: as) (list 1 2 3))
    (list a as))
(1 (2 3))
> (let ((a :rest) (list 1 2 3))
    (list a rest))
(1 (2 3))

Iteration

There are several iteration mechanisms in Dax. The simplest is a while loop:

> (let i 3
    (while (> i 0)
      (print (dec i))))
2
1
0

The shorthand for iterating from 0 to N is for:

> (for i 3
    (print i))
0
1
2

You can enumerate the keys and values of a list with each:

> (each (k v) (list 1 2 a: 10 b: 20)
    (print (cat k " " v)))
1 1
2 2
b 20
a 10
> (each v (list 1 2 a: 10 b: 20) ; values only
    (print v))
1
2
20
10
> (each (k (a b)) ; destructuring
      (list (list 10 20) bar: (list "a" "b"))
    (print (cat k " " a " " b)))
1 10 20
bar a b

each will bind keys and values in any order. If you want only the positional values of a list, you can enumerate them in order using step:

> (step x (list 1 2 3)
    (print x))
1
2
3
> (step (a b) (list (list 1 2) (list 10 20)) ; destructuring
    (print a)
    (print b))
1
2
10
20

Quotation

Expressions can be prevented from being evaluated using the quote operator:

> (quote (1 2 3))
(1 2 3)

Expressions that evaluate to themselves are unaffected by quotation:

> (quote 10)
10
> (quote false)
false

Quoting names and strings results in strings that would evaluate to the quoted expression:

> (quote a)
"a"
> (quote "hereanother")
"\"hereanother\""
> (quote "two\nlines")
"\"two\\nlines\""

This is also true for expressions inside lists:

> (quote (a b c))
("a" "b" "c")
> (quote (1 2 b: baz z: "frob"))
(1 2 b: "baz" z: "\"frob\"")

The shorthand for quotation is use a single quote:

> '17
17
> '(1 2 3)
(1 2 3)
> '(a b c)
("a" "b" "c")

Another way to write many strings is to simply quote their name:

> 'a
"a"
> (let x '(a: 10 b: 20)
    (x .a))
10

When you want to quote some parts of an expression, but want other parts to be evaluated, use quasiquote and unquote:

> (let x 10
    (quasiquote (1 5 (unquote x))))
(1 5 10)

The shorthand for quasiquotation is ` for quasiquote and , for unquote:

> (let x 10 `(1 5 ,x))
(1 5 10)

A different way to unquote expressions is unquote-splicing, which takes the values contained in a nested list and places them in the enclosing one:

> (let a '(1 2 3)
    (quasiquote (9 8 (unquote-splicing a))))
(9 8 1 2 3)

The shorthand for unquote-splicing is ,@:

> (let a '(1 2 3)
    `(9 8 ,@a))
(9 8 1 2 3)

Macros

Macros allow you to write functions that manipulate expressions before they have been evaluated. Macros take expressions as parameters and return an expression:

> (define-macro when (condition rest: body)
    `(if ,condition (do ,@body)))
(macro: function)
> (when true
    (print 'hi)
    (+ 10 20))
hi
30

Acknowledgements

Dax is a fork of Lumen.

Lumen was conceived and designed with Daniel Gackle, and many fixes and improvements have been contributed by Shawn Presser.