Presented at the 17-th EuroForth Conference (23-26 November 2001, Schloss Dagstuhl, Saarbruecken, Germany) by Reuben Thomas, on my behalf and with my thanks.
The language Joy is a purely functional programming language. Whereas all other functional programming languages are based on the application of functions to arguments, Joy is based on the composition of functions. All such functions take a stack as argument and produce a stack as value. Consequently much of Joy looks like ordinary postfix notation. However, in Joy a function can consume any number of parameters from the stack and leave any number of results on the stack. The concatenation of appropriate programs denotes the composition of the functions which the programs denote. One of the datatypes of Joy is that of quoted programs, of which lists are a special case. Some functions expect quoted programs on top of the stack and execute them in many different ways, effectively by dequoting. So, where other functional languages use abstraction and application, Joy uses quotation and combinators – functions which perform dequotation. As a result, there are no named formal parameters, no substitution of actual for formal parameters, and no environment of name-value pairs. Combinators in Joy behave much like functionals or higher order functions in other languages, they minimise the need for recursive and non-recursive definitions. Because there is no need for an environment, Joy has an exceptionally simple algebra, and its programs are easily manipulated by hand and by other programs. Many programs first construct another program which is then executed.
This paper is intended as an introduction for Forth programmers.
To add two integers, say 2 and 3, and to write their sum, you type the program:
2 3 +This is how it works internally: the first numeral causes the integer 2 to be pushed onto a stack. The second numeral causes the integer 3 to be pushed on top of that. Then the addition operator pops the two integers off the stack and pushes their sum, 5. So the notation looks like ordinary postfix. The Joy processor reads programs like the above until they are terminated by a period. Only then are they executed. In the default mode the item on the top of the stack (5 in the example) is then written to the output file, which normally is the screen.
To compute the square of an integer, it has to be multiplied by itself. To compute the square of the sum of two integers, the sum has to be multiplied by itself. Preferably this should be done without computing the sum twice. The following is a program to compute the square of the sum of 2 and 3:
2 3 + dup *After the sum of 2 and 3 has been computed, the stack just contains
the integer 5. The dup operator then pushes another copy of
the 5 onto the stack. Then the multiplication operator replaces the two
integers by their product, which is the square of 5. The square is then
written out as 25. Apart from the dup operator there are
several others for re-arranging the top of the stack. The
pop operator removes the top element, and the
swap operator interchanges the top two elements. There are
several other operators that do similar things. This idea is of course
familiar to Forth programmers.
Unlike Forth, Joy has a data type of lists. A list of integers is written inside square brackets. Just as integers can be added and otherwise manipulated, so lists can be manipulated in various ways. The following concatenates two lists:
[1 2 3] [4 5 6 7] concatThe two lists are first pushed onto the stack. Then the
concat operator pops them off the stack and pushes the list
[1 2 3 4 5 6 7] onto the stack.
Joy makes extensive use of combinators, even more so than most functional languages. In mainstream functional languages combinators take as arguments (lambda abstractions of) functions. In Joy combinators are like operators in that they expect something specific on top of the stack. Unlike operators they execute what they find on top of the stack, and this has to be the quotation of a program, enclosed in square brackets. One of these is a combinator for mapping elements of one list via a function to another list. Consider the program:
[1 2 3 4] [dup *] mapIt first pushes the list of integers and then it pushes the quoted
squaring program onto the stack. The map combinator then
removes the list and the quotation and constructs another list by
applying the program to each member of the given list. The result is the
list:
[1 4 9 16]which is left on top of the stack.
In definitions of new functions, as in Forth, no formal parameters are used, and hence there is no substitution of actual parameters for formal parameters. After the following definition:
square == dup *the symbol square can be used in place of
dup *. As in most other programming languages, definitions
may be recursive.
The data types of Joy are divided into simple and aggregate types.
The simple types comprise integers, floating point, characters
and the truth values. The aggregate types comprise sets, strings, lists
and files. Literals of any type cause a value of that type to be pushed
onto the stack. There they can be manipulated by the general stack
operations such as dup, pop and
swap and a few others, all familiar in Forth. Or they can
be manipulated by operators specific to their type.
Integers and floats are written in decimal notation. The usual binary operations and relations are provided. Operators are written after their operands. Binary operators remove two values from the top of the stack and replace them by the result. Unary operators are similar, except that they only remove one value. In addition to the integer type there are also characters and truth values (Boolean), both with the usual operations.
The aggregate types are the unordered type of sets and the ordered types of strings and lists. Aggregates can be built up, combined, taken apart and tested for membership. A set is an unordered collection of zero or more small integers, 0..31. Literals of type set are written inside curly braces, like this {3 7 21}. The usual set operations are available.
A string is an ordered sequence of zero or more characters. Literals of this type string are written inside double quotes, like this: “Hello world”.
A list is an ordered sequence of zero or more values of any type. Literals of type list are written inside square brackets, like this: [peter paul mary] or [42 true {2 5}]. Lists can contain mixtures of values of any type. In particular they can contain lists as members, so the type of lists is a recursive data type. The usual list operations are provided. Matrices are represented as lists of lists, and there is a small library for the usual matrix operations. To a large extent the operators on the structured types are identical. Lists are implemented using linked structures, and the stack itself is just a list. This makes it possible to treat the stack as a list and vice versa.
Lists are really just a special case of quoted programs. Lists only contain values of the various types, but quoted programs may contain other elements such as operators and some others that are explained below. A quotation can be treated as a passive data structure just like a list. For example:
[ + 20 * 10 4 - ]has size 6, its second and third elements are 20 and *, it can be reversed or it can be concatenated with other quotations. But passive quotations can also be made active by dequotation.
If the above quotation occurs in a program, then it results in the
quotation being pushed onto the stack – just as a list would be pushed.
There are many other ways in which that quotation could end up on top of
the stack, by being concatenated from its parts, by extraction from a
larger quotation, or by being read from the input. No matter how it got
to be on top of the stack, it can now be treated in two ways: passively
as a data structure, or actively as a program. The square brackets
prevented it from being treated actively. Without them the program would
have been executed: it would expect two integers which it would add,
then multiply the result by 20, and finally push 6, the difference
between 10 and 4. A combinator expects a quotation on the stack and
executes it in a way that is different for each combinator. One of the
simplest is the i combinator. Its effect is to execute a
single program on top of the stack, and nothing else. Syntactically
speaking, its effect is to remove the quoting square brackets and thus
to expose the quoted program for execution. Consequently the following
two programs are equivalent:
[ + 20 * 10 4 - ] i
+ 20 * 10 4 -The i combinator is mainly of theoretical significance,
but it is used occasionally. The many other combinators are essential
for programming in Joy.
Some combinators require that the stack contains values of certain
types. Many are analogues of higher order functions familiar from other
programming languages: map, filter and
fold. The map has already been explained.
Another combinator that expects an aggregate is the
filter combinator. The quoted program has to yield a truth
value. The result is a new aggregate of the same type containing those
elements of the original for which the quoted program yields true. For
example, the quoted program [’Z >] will yield truth for characters
whose numeric values is greater than that of Z. Hence it can be used to
remove upper case letters and blanks from a string. So the following
evaluates to “ohnmith”:
"John Smith" ['Z >] filterSometimes it is necessary to add or multiply or otherwise combine all
elements of an aggregate value. The fold combinator can do
just that. It requires three parameters: the aggregate to be folded, the
quoted value to be returned when the aggregate is empty, and the quoted
binary operation to be used to combine the elements. In some languages
the combinator is called reduce (because it turns the aggregate into a
single value), or insert (because it looks as though the binary
operation has been inserted between any two members). The following two
programs compute the sum of the members of a list and the sum of the
squares of the members of a list. They evaluate to 10 and 38,
respectively:
[2 5 3] 0 [+] fold
[2 5 3] 0 [dup * +] foldThe squaring function can be defined in just about any programming language. In infix notation the external definition might look similar to this:
square(x) == x * xInternally, and possibly externally too, the definition would be:
square == Lx : x * xwhere “Lx” is sometimes written “\x” or “lambda x” or “fn x”. The expression on the right side of the definition then is to be read as “the function of one argument x which yields the value x * x”. An expression like that is known as a lambda abstraction. Almost all programming languages use such a construct at least implicitly.
Such a definition would be used as in the following evaluation, where square is being applied (here using an explicit @ infix operator) to the argument 2:
square @ 2
(Lx : x * x) @ 2
2 * 2
4In the second line square is being replaced by its definition. This sets up an environment in which x = 2, in which the formal parameter x has been given the value of the actual parameter 2. An environment is a collection of associations of formal and actual parameters. In all but the simplest cases the environment will contain several such associations. In the third line this environment is used to replace the formal parameter x by the actual parameter 2. All this is completely hidden from the user when the first style of definition is used.
The two operations of lambda abstraction and application are complementary. They form the heart of the lambda calculus which underlies all the mainstream functional and imperative languages. Joy eliminates abstraction and application and replaces them by program quotation and function composition.
The previous definitions of the squaring function as a lambda abstraction might be written:
square == Lx : x x *
square == Lx : x dup *For the second definition the right hand side has to be read as “the stack function for stacks in which there is a top element x and which for stacks without that top element yields the same result as x dup *”. But one might also write without the explicit lambda abstraction as:
x square == x x *
x square == x dup *In the last of these both sides start with the formal paramater x, and otherwise there are no further occurrences of x on either side. Would it be possible for the two occurrences on the left and right to “cancel out”, so to speak? Yes indeed, and now the definition looks loke this:
square == dup *The mainstream imperative languages have a state of associations between assignable variables and their current values. The values are changed by assignments during the run of the program. These languages also have an environment of formal / actual parameter associations that are set up by calls of defined functions or procedures. Purely functional languages have no state. But the mainstream functional languages are based on the lambda calculus, and hence they have an environment. The purely functional language Joy has no state and no environment. The imperative language Forth has both.
As in other languages, definitions can be recursive in Joy. In the first line below is a recursive definition of the factorial function in one of many variants of conventional notation. In the second line is a recursive definition in Joy:
factorial(x) = if x = 0 then 1 else x * factorial(x - 1)
factorial == [0 =] [pop 1] [dup 1 - factorial *] ifteAgain the Joy version does not use a formal parameter x. It works like this: The right side of the definition pushes three quotations, called the if-part, the then-part and the else-part. Then the if-then-else combinator ifte sets these aside and exposes the numeric argument that is now on top of stack. From here on it behaves just like conditional expression in the conventional version.
If one wanted to compute the list of factorials of a given list, this could be done by:
[ factorial ] mapBut this relies on an external definition of factorial. It was necessary to give that definition explicitly because it is recursive. If one only wanted to compute factorials of lists of numbers, then it would be a minor nuisance to be forced to define factorial explicitly just because the definition is recursive.
A high proportion of recursively defined functions exhibit a very simple pattern: There is some test, the if-part, which determines whether the ground case obtains. If it does, then the non-recursive branch is executed, the base case of recursion. If it does not, then the recursive part is executed, including one or more recursive calls.
Joy has a useful device, the linrec combinator, which
allows computation of anonymous functions that might have been
defined recursively using a linear recursive pattern. Whereas the
ifte combinator requires three quoted parameters, the
linrec combinator requires four: an if-part, a then-part, a
rec1-part and a rec2-part. Recursion occurs between the two rec-parts.
For example, the factorial function could be computed by:
[null] [succ] [dup pred] [*] linrecThere is no need for a definition, the above program can be used directly. To compute the list of factorials of a given list of numbers the following can be used:
[ [null] [succ] [dup pred] [*] linrec ] mapIn many recursive definitions there are two recursive calls of the
function being defined. This is the pattern of binary
recursion, and it is used in the usual definitions of quicksort and
of the Fibonacci function. In analogy with the linrec
combinator for linear recursion, Joy has a binrec
combinator for binary recursion. The following will quicksort a
list whose members can be a mixture of anything except lists:
[small] [] [uncons [>] split] [swapd cons concat] binrecJoy programs are built from smaller programs by just two constructors: concatenation and quotation.
Concatenation is a binary constructor, and since it is associative it is best written in infix notation and hence no parentheses are required. Since concatenation is the only binary constructor of its kind, in Joy it is best written without an explicit symbol.
Quotation is a unary constructor which takes as its operand a program. In Joy the quotation of a program is written by enclosing it in square brackets. Ultimately all programs are built from atomic programs which do not have any parts.
The semantics of Joy has to explain what the atomic programs mean, how the meaning of a concatenated program depends on the meaning of its parts, and what the meaning of a quoted program is. Moreover, it has to explain under what conditions it is possible to replace a part by an equivalent part while retaining the meaning of the whole program.
Joy programs denote functions which take one argument and yield one value. The argument and the value consist of several components. The principal component is a stack, and the other components are not needed here. Much of the detail of the semantics of Joy depends on specific properties of programs.
However, central to the semantics of Joy is the following, which also holds for the purely functional fragment of Forth:
The concatenation of two programs denotes the composition of the functions denoted by the two programs.
Function composition is associative, and hence denotation maps the associative syntactic operation of program concatenation onto the associative semantic operation of function composition.
One part of a concatenation may be replaced by another part denoting the same function while retaining the denotation of the whole concatenation.
One quoted program may be replaced by another denoting the same function only in a context where the quoted program will be dequoted by being executed. Such contexts are provided by the combinators of Joy. These denote functions which behave like higher order functions in other languages. There is no equivalent mechanism in Forth.
The above may be summarised as follows: Let P, Q1, Q2 and R be programs, and let C be a combinator. Then this principle holds:
IF Q1 == Q2
THEN P Q1 R == P Q2 R
AND [Q1] C == [Q2] CThe principle is the prime rule of inference for the algebra of Joy which deals with the equivalence of Joy programs, and hence with the identity of functions denoted by such programs. A few laws in the algebra can be expressed without combinators, but most require one or more combinators for their expression. Concatenation of Joy programs denote the composition of the functions which the concatenated parts denote. Hence if Q1 and Q2 are programs which denote the same function and P and R are other programs, then the two concatenations P Q1 R and P Q2 R also denote the same function. In other words, programs Q1 and Q2 can replace each other in concatenations. This can serve as a rule of inference for rewriting.
As premises one needs axioms such as in the first three lines below, and definitions such as in the fourth line:
(+) 2 3 + == 5
(dup) 5 dup == 5 5
(*) 5 5 * == 25
(def square) square == dup *A derivation using the above axioms and the definition looks like this:
2 3 + square
== 5 square (+)
== 5 dup * (def square)
== 5 5 * (dup)
== 25 (*)The comments in the right margin explain how a line was obtained from the previous line. The derivation shows that the expressions in the first line and the last line denote the same function, or that the function in the first line is identical with the function in the last line.
Consider the following equations in infix notation: The first says that multiplying a number x by 2 gives the same result as adding it to itself. The second says that the size of a reversed list is the same as the size of the original list:
2 * x = x + x size(reverse(x)) = size(x)In Joy the same equations would be written, without variables, like this:
2 * == dup + reverse size == sizeOther equivalences express algebraic properties of various
operations. For example, the predecessor pred of the
successor succ of a number is just the number itself. The
conjunction and of a truth value with itself gives just the
truth value. The less than relation < is the converse of the greater
than relation >. Inserting a number with cons into a
list of numbers and then taking the sum of that gives the same result as
first taking the sum of the list and then adding the other number.
In conventional notation these are expressed by:
pred(succ(x)) = x x and x = x
x < y = y > x sum(cons(x,y)) = x + sum(y)In Joy these can be expressed without variables, using the
identity function id:
succ pred == id dup and == id
< == swap > cons sum == sum +Some properties of operations have to be expressed by combinators.
One of these is the dip combinator which expects a program
on top of the stack and below that another value. It saves the value,
executes the program on the remainder of the stack and then restores the
saved value.
In the first example below, the dip combinator is used
to express the associativity of addition. Another combinator is the
app2unary2 combinator which expects a program on
top of the stack and below that two values. It applies the program to
the two values. In the second example below it expresses one of the De
Morgan laws. In the third example it expresses that the size of two
lists concatenated is the sum of the sizes of the two concatenands. The
last example uses both combinators to express that multiplication
distributes (from the right) over addition. (Note that the program
parameter for app2unary2 is first constructed
from the multiplicand and *.)
[+] dip + == + +
and not == [not] unary2 or (* app2 *)
concat size == [size] unary2 + (* app2 *)
[+] dip * == [*] cons unary2 + (* app2 *)The similarities and differences between Joy and Forth are striking and profound. They have been discussed in the mailing group which can be found at concatenative. The main Joy page is on Latrobe.