Mercurial > dylan
view categorical/imperative.org @ 3:44d3dc936f6a
moved backup files
author | Robert McIntyre <rlm@mit.edu> |
---|---|
date | Fri, 28 Oct 2011 00:06:37 -0700 |
parents | b4de894a1e2e |
children |
line wrap: on
line source
1 #+TITLE: LAMBDA: The Ultimate Imperative2 #+AUTHOR: Guy Lewis Steele Jr. and Gerald Jay Sussman4 * Abstract6 We demonstrate how to model the following common programming constructs in7 terms of an applicative order language similar to LISP:8 - Simple Recursion9 - Iteration10 - Compound Statements and Expressions11 - GO TO and Assignment12 - Continuation—Passing13 - Escape Expressions14 - Fluid Variables15 - Call by Name, Call by Need, and Call by Reference16 The models require only (possibly self-referent) lambda application,17 conditionals, and (rarely) assignment. No complex data structures such as18 stacks are used. The models are transparent, involving only local syntactic19 transformations.20 Some of these models, such as those for GO TO and assignment, are already well21 known, and appear in the work of Landin, Reynolds, and others. The models for22 escape expressions, fluid variables, and call by need with side effects are23 new. This paper is partly tutorial in intent, gathering all the models24 together for purposes of context.25 This report describes research done at the Artificial Intelligence Laboratory26 of the Massachusetts Institute of Teehnology. Support for the laboratory's27 artificial intelligence research is provided in part by the Advanced Research28 Projects Agency of the Department of Defense under Office of Naval Research29 contract N000l4-75-C-0643.31 * Introduction33 We catalogue a number of common programming constructs. For each34 construct we examine "typical" usage in well-known programming languages, and35 then capture the essence of the semantics of the construct in terms of a36 common meta-language.37 The lambda calculus {Note Alonzowins} is often used as such a meta-38 language. Lambda calculus offers clean semantics, but it is clumsy because it39 was designed to be a minimal language rather than a convenient one. All40 lambda calculus "functions" must take exactly one "argument"; the only "data41 type" is lambda expressions; and the only "primitive operation‘ is variable‘42 substitution. While its utter simplicity makes lambda calculus ideal for43 logicians, it is too primitive for use by programmers. The meta-language we44 use is a programming language called SCHEME {Note Schemepaper) which is based45 on lambda calculus.46 SCHEME is a dialect of LISP. [McCarthy 62] It is an expression-47 oriented, applicative order, interpreter-based language which allows one to48 manipulate programs as data. It differs from most current dialects of LISP in49 that it closes all lambda expressions in the environment of their definition50 or declaration, rather than in the execution environment. {Note Closures}51 This preserves the substitution semantics of lambda calculus, and has the52 consequence that all variables are lexically scoped, as in ALGOL. [Naur 63]53 Another difference is that SCHEME is implemented in such a way that tail-54 recursions execute without net growth of the interpreter stack. {Note55 Schemenote} We have chosen to use LISP syntax rather than, say, ALGOL syntax56 because we want to treat programs as data for the purpose of describing57 transformations on the code. LISP supplies names for the parts of an58 executable expression and standard operators for constructing expressions and59 extracting their components. The use of LISP syntax makes the structure of60 such expressions manifest. We use ALGOL as an expository language, because it61 is familiar to many people, but ALGOL is not sufficiently powerful to express62 the necessary concepts; in particular, it does not allow functions to return63 functions as values. we are thus forced to use a dialect of LISP in many64 cases.65 We will consider various complex programming language constructs and66 show how to model them in terms of only a few simple ones. As far as possible67 we will use only three control constructs from SCHEME: LAMBDA expressions, as68 in LISP, which are Just functions with lexically scoped free variables;69 LABELS, which allows declaration of mutually recursive procedures (Note70 Labelsdef}; and IF, a primitive conditional expression. For more complex71 modelling we will introduce an assignment primitive (ASET).i We will freely72 assume the existence of other comon primitives, such as arithmetic functions.73 The constructs we will examine are divided into four broad classes.74 The first is sfinph?Lo0pl; this contains simple recursions and iterations, and75 an introduction to the notion of continuations. The second is hmponflivo76 Connrucls; this includes compound statements, G0 T0, and simple variable77 assignments. The third is continuations, which encompasses the distinction between statements and expressions, escape operators (such as Landin's J-78 operator [Landin 65] and Reynold's escape expression [Reynolds 72]). and fluid79 (dynamically bound) variables. The fourth is Parameter Passing Mechanisms, such80 as ALGOL call—by-name and FORTRAN call-by-location.81 Some of the models presented here are already well-known, particularly82 those for G0 T0 and assignment. [McCarthy 60] [Landin 65] [Reynolds 72]83 Those for escape operators, fluid variables, and call-by-need with side84 effects are new.85 ** Simple Loops86 By /simple loops/ we mean constructs which enable programs to execute the87 same piece of code repeatedly in a controlled manner. Variables may be made88 to take on different values during each repetition, and the number of89 repetitions may depend on data given to the program.90 *** Simple Recursion91 One of the easiest ways to produce a looping control structure is to92 use a recursive function, one which calls itself to perform a subcomputation.93 For example, the familiar factorial function may be written recursively in94 ALGOL: '95 #+begin_src algol96 integer procedure fact(n) value n: integer n:97 fact := if n=0 then 1 else n*fact(n-1);98 #+end_src100 The invocation =fact(n)= computes the product of the integers from 1 to n using101 the identity n!=n(n-1)! (n>0). If $n$ is zero, 1 is returned; otherwise =fact=:102 calls itself recursively to compute $(n-1)!$ , then multiplies the result by $n$103 and returns it.105 This same function may be written in Clojure as follows:106 #+begin_src clojure107 (ns categorical.imperative)108 (defn fact[n]109 (if (= n 0) 1 (* n (fact (dec n))))110 )111 #+end_src113 #+results:114 : #'categorical.imperative/fact116 Clojure does not require an assignment to the ``variable'' fan: to return a value117 as ALGOL does. The IF primitive is the ALGOL if-then-else rendered in LISP118 syntax. Note that the arithmetic primitives are prefix operators in SCHEME.120 *** Iteration121 There are many other ways to compute factorial. One important way is122 through the use of /iteration/.123 A comon iterative construct is the DO loop. The most general form we124 have seen in any programming language is the MacLISP DO [Moon 74]. It permits125 the simultaneous initialization of any number of control variables and the126 simultaneous stepping of these variables by arbitrary functions at each127 iteration step. The loop is terminated by an arbitrary predicate, and an128 arbitrary value may be returned. The DO loop may have a body, a series of129 expressions executed for effect on each iteration. A version of the MacLISP130 DO construct has been adopted in SCHEME.132 The general form of a SCHEME DO is:133 #+begin_src nothing134 (DO ((<var1> (init1> <stepl>)135 ((var2> <init2> <step2>)136 (<varn> <intn> (stepn>))137 (<pred> (value>)138 (optional body>)139 #+end_src140 The semantics of this are that the variables are bound and initialized to the141 values of the <initi> expressions, which must all be evaluated in the142 environment outside the D0; then the predicate <pred> is evaluated in the new143 environment, and if TRUE, the (value) is evaluated and returned. Otherwise144 the (optional body) is evaluated, then each of the steppers <stepi> is145 evaluated in the current environment, all the variables made to have the146 results as their values, the predicate evaluated again, and so on.147 Using D0 loops in both ALGOL and SCHEME, we may express FACT by means148 of iteration.149 #+begin_src algol150 integer procedure fact(n); value n; integer n:151 begin152 integer m, ans;153 ans := 1;154 for m := n step -l until 0 do ans := m*ans;155 fact := ans;156 end;157 #+end_src159 #+begin_src clojure160 (in-ns 'categorical.imperative)161 (defn fact-do[n]162 )163 #+end_src165 Note that the SCHEME D0 loop in FACT has no body -- the stepping functions do166 all the work. The ALGOL DO loop has an assignment in its body; because an167 ALGOL DO loop can step only one variable, we need the assignment to step the168 the variable "manually'.169 In reality the SCHEME DO construct is not a primitive; it is a macro170 which expands into a function which performs the iteration by tail—recursion.171 Consider the following definition of FACT in SCHEME. Although it appears to172 be recursive, since it "calls itself", it is entirely equivalent to the DO173 loop given above, for it is the code that the DO macro expands into! It174 captures the essence of our intuitive notion of iteration, because execution175 of this program will not produce internal structures (e.g. stacks or variable176 bindings) which increase in size with the number of iteration steps.178 #+begin_src clojure179 (in-ns 'categorical.imperative)180 (defn fact-do-expand[n]181 (let [fact1182 (fn[m ans]183 (if (zero? m) ans184 (recur (dec m) (* m ans))))]185 (fact1 n 1)))186 #+end_src188 From this we can infer a general way to express iterations in SCHEME in189 a manner isomorphic to the HacLISP D0. The expansion of the general D0 loop190 #+begin_src nothing191 (DO ((<varl> <init1> <step1>)192 (<var2> (init2) <step2>)193 ...194 (<varn> <initn> <stepn>))195 (<pred> <va1ue>)196 <body>)197 #+end_src198 is this:199 #+begin_src nothing200 (let [doloop201 (fn [dummy <var1> <var2> ... <varn>)202 (if <pred> <value>203 (recur <body> <step1> <step2> ... <stepn>)))]205 (doloop nil <init1> <init2> ... <initn>))206 #+end_src207 The identifiers =doloop= and =dummy= are chosen so as not to conflict with any208 other identifiers in the program.209 Note that, unlike most implementations of D0, there are no side effects210 in the steppings of the iteration variables. D0 loops are usually modelled211 using assignment statements. For example:212 #+begin_src nothing213 for r :1 0 step b until c do <statement>;214 #+end_src216 can be modelled as follows: [Naur 63]217 #+begin_src nothing218 begin219 x := a;220 L: if (x-c)*sign(n) > 0 then go to Endloop;221 <statement>;222 x := x+b;223 go to L:224 Endloop:225 end;226 #+end_src227 Later we will see how such assignment statements can in general be228 modelled without using side effects.230 * Imperative Programming231 Lambda calculus (and related languages, such as ``pure LISP'') is often232 used for modelling the applicative constructs of programming languages.233 However, they are generally thought of as inappropriate for modelling234 imperative constructs. In this section we show how imperative constructs may235 be modelled by applicative SCHEME constructs.236 ** Compound Statements237 The simplest kind of imperative construct is the statement sequencer,238 for example the compound statement in ALGOL:239 #+begin_src algol240 begin241 S1;242 S2;243 end244 #+end_src246 This construct has two interesting properties:247 - (1) It performs statement S1 before S2, and so may be used for248 sequencing.249 - (2) S1 is useful only for its side effects. (In ALGOL, S2 must also250 be a statement, and so is also useful only for side effects, but251 other languages have compound expressions containing a statement252 followed by an expression.)254 The ALGOL compound statement may actually contain any number of statements,255 but such statements can be expressed as a series of nested two-statement256 compounds. That is:257 #+begin_src algol258 begin259 S1;260 S2;261 ...262 Sn-1;263 Sn;264 end265 #+end_src266 is equivalent to:268 #+begin_src algol269 begin270 S1;271 begin272 S2;273 begin274 ...276 begin277 Sn-1;278 Sn;279 end;280 end;281 end:282 end283 #+end_src285 It is not immediately apparent that this sequencing can be expressed in a286 purely applicative language. We can, however, take advantage of the implicit287 sequencing of applicative order evaluation. Thus, for example, we may write a288 two-statement sequence as follows:290 #+begin_src clojure291 ((fn[dummy] S2) S1)292 #+end_src294 where =dummy= is an identifier not used in S2. From this it is295 manifest that the value of S1 is ignored, and so is useful only for296 side effects. (Note that we did not claim that S1 is expressed in a297 purely applicative language, but only that the sequencing can be so298 expressed.) From now on we will use the form =(BLOCK S1 S2)= as an299 abbreviation for this expression, and =(BLOCK S1 S2...Sn-1 Sn)= as an300 abbreviation for302 #+begin_src algol303 (BLOCK S1 (BLOCK S2 (BLOCK ... (BLOCK Sn-1 Sn)...)))304 #+end_src306 ** 2.2. The G0 TO Statement308 A more general imperative structure is the compound statement with309 labels and G0 T0s. Consider the following code fragment due to310 Jacopini, taken from Knuth: [Knuth 74]311 #+begin_src algol312 begin313 L1: if B1 then go to L2; S1;314 if B2 then go to L2; S2;315 go to L1;316 L2: S3;317 #+end_src319 It is perhaps surprising that this piece of code can be /syntactically/320 transformed into a purely applicative style. For example, in SCHEME we321 could write:323 #+begin_src clojure324 (in-ns 'categorical.imperative)325 (let326 [L1 (fn[]327 (if B1 (L2) (do S1328 (if B2 (L2) (do S2 (L1))))))329 L2 (fn[] S3)]330 (L1))331 #+end_src333 As with the D0 loop, this transformation depends critically on SCHEME's334 treatment of tail-recursion and on lexical scoping of variables. The labels335 are names of functions of no arguments. In order to ‘go to" the labeled code,336 we merely call the function named by that label.338 ** 2.3. Simple Assignment339 Of course, all this sequencing of statements is useless unless the340 statements have side effects. An important side effect is assignment. For341 example, one often uses assignment to place intermediate results in a named342 location (i.e. a variable) so that they may be used more than once later343 without recomputing them:345 #+begin_src algol346 begin347 real a2, sqrtdisc;348 a2 := 2*a;349 sqrtdisc := sqrt(b^2 - 4*a*c);350 root1 := (- b + sqrtdisc) / a2;351 root2 := (- b - sqrtdisc) / a2;352 print(root1);353 print(root2);354 print(root1 + root2);355 end356 #+end_src358 It is well known that such naming of intermediate results may be accomplished359 by calling a function and binding the formal parameter variables to the360 results:362 #+begin_src clojure363 (fn [a2 sqrtdisc]364 ((fn[root1 root2]365 (do (println root1)366 (println root2)367 (println (+ root1 root2))))368 (/ (+ (- b) sqrtdisc) a2)369 (/ (- (- b) sqrtdisc) a2))371 (* 2 a)372 (sqrt (- (* b b) (* 4 a c))))373 #+end_src374 This technique can be extended to handle all simple variable assignments which375 appear as statements in blocks, even if arbitrary G0 T0 statements also appear376 in such blocks. {Note Mccarthywins}378 For example, here is a program which uses G0 TO statements in the form379 presented before; it determines the parity of a non-negative integer by380 counting it down until it reaches zero.382 #+begin_src algol383 begin384 L1: if a = 0 then begin parity := 0; go to L2; end;385 a := a - 1;386 if a = 0 then begin parity := 1; go to L2; end;387 a := a - 1;388 go to L1;389 L2: print(parity);390 #+end_src392 This can be expressed in SCHEME:394 #+begin_src clojure395 (let396 [L1 (fn [a parity]397 (if (zero? a) (L2 a 0)398 (L3 (dec a) parity)))399 L3 (fn [a parity]400 (if (zero? a) (L2 a 1)401 (L1 (dec a) parity)))402 L2 (fn [a parity]403 (println parity))]404 (L1 a parity))405 #+end_src407 The trick is to pass the set of variables which may be altered as arguments to408 the label functions. {Note Flowgraph} It may be necessary to introduce new409 labels (such as L3) so that an assignment may be transformed into the binding410 for a function call. At worst, one may need as many labels as there are411 statements (not counting the eliminated assignment and GO TO statements).413 ** Compound Expressions '414 At this point we are almost in a position to model the most general415 form of compound statement. In LISP, this is called the 'PROG feature". In416 addition to statement sequencing and G0 T0 statements, one can return a /value/417 from a PROG by using the RETURN statement.419 Let us first consider the simplest compound statement, which in SCHEME420 we call BLOCK. Recall that421 =(BLOCK s1 sz)= is defined to be =((lambda (dummy) s2) s1)=423 Notice that this not only performs Sl before S2, but also returns the value of424 S2. Furthermore, we defined =(BLOCK S1 S2 ... Sn)= so that it returns the value425 of =Sn=. Thus BLOCK may be used as a compound expression, and models a LISP426 PROGN, which is a PROG with no G0 T0 statements and an implicit RETURN of the427 last ``statement'' (really an expression).429 Host LISP compilers compile D0 expressions by macro-expansion. We have430 already seen one way to do this in SCHEME using only variable binding. A more431 common technique is to expand the D0 into a PROG, using variable assignments432 instead of bindings. Thus the iterative factorial program:434 #+begin_src clojure435 (oarxnz FACT .436 (LAMaoA (n) .437 (D0 ((M N (- H 1))438 (Ans 1 (* M Ans)))439 ((- H 0) A"$))))440 #+end_src442 would expand into:444 #+begin_src clojure445 (DEFINE FACT446 . (LAMeoA (M)447 (PRO6 (M Ans)448 (sszro M n449 Ans 1) ~450 LP (tr (- M 0) (RETURN Ans))451 (ssero M (- n 1)452 Ans (' M Ans))453 (60 LP))))454 #+end_src456 where SSETQ is a simultaneous multiple assignment operator. (SSETQ is not a457 SCHEME (or LISP) primitive. It can be defined in terms of a single assignment458 operator, but we are more interested here in RETURN than in simultaneous459 assignment. The SSETQ's will all be removed anyway and modelled by lambda460 binding.) We can apply the same technique we used before to eliminate G0 T0461 statements and assignments from compound statements:463 #+begin_src clojure464 (DEFINE FACT465 (LAHBOA (I)466 (LABELS ((L1 (LAManA (M Ans)467 (LP n 1)))468 (LP (LAMaoA (M Ans)469 (IF (- M o) (nztunn Ans)470 (£2 H An$))))471 (L2 (LAMaoA (M An )472 (LP (- " 1) (' H flN$)))))473 (L1 NIL NlL))))474 #+end_src clojure476 We still haven't done anything about RETURN. Let's see...477 - ==> the value of (FACT 0) is the value of (Ll NIL NIL)478 - ==> which is the value of (LP 0 1)479 - ==> which is the value of (IF (= 0 0) (RETURN 1) (L2 0 1))480 - ==> which is the value of (RETURN 1)482 Notice that if RETURN were the /identity483 function/ (LAMBDA (X) X), we would get the right answer. This is in fact a484 general truth: if we Just replace a call to RETURN with its argument, then485 our old transformation on compound statements extends to general compound486 expressions, i.e. PROG.488 * Continuations489 Up to now we have thought of SCHEME's LAMBDA expressions as functions,490 and of a combination such as =(G (F X Y))= as meaning ``apply the function F to491 the values of X and Y, and return a value so that the function G can be492 applied and return a value ...'' But notice that we have seldom used LAMBDA493 expressions as functions. Rather, we have used them as control structures and494 environment modifiers. For example, consider the expression:495 =(BLOCK (PRINT 3) (PRINT 4))=497 This is defined to be an abbreviation for:498 =((LAMBDA (DUMMY) (PRINT 4)) (PRINT 3))=500 We do not care about the value of this BLOCK expression; it follows that we501 do not care about the value of the =(LAMBDA (DUMMY) ...)=. We are not using502 LAMBDA as a function at all.504 It is possible to write useful programs in terms of LAHBDA expressions505 in which we never care about the value of /any/ lambda expression. We have506 already demonstrated how one could represent any "FORTRAN" program in these507 terms: all one needs is PROG (with G0 and SETQ), and PRINT to get the answers508 out. The ultimate generalization of this imperative programing style is509 /continuation-passing/. (Note Churchwins} .511 ** Continuation-Passing Recursion512 Consider the following alternative definition of FACT. It has an extra513 argument, the continuation, which is a function to call with the answer, when514 we have it, rather than return a value516 #+begin_src algol.517 procedure Inc!(n, c); value n, c;518 integer n: procedure c(integer value);519 if n-=0 then c(l) else520 begin521 procedure l(!mp(d) value a: integer a;522 c(mm);523 Iacl(n-1, romp):524 end;525 #+end_src527 #+begin_src clojure528 (in-ns 'categorical.imperative)529 (defn fact-cps[n k]530 (if (zero? n) (k 1)531 (recur (dec n) (fn[a] (k (* n a))))))532 #+end_src clojure533 It is fairly clumsy to use this version of‘ FACT in ALGOL; it is necessary to534 do something like this:536 #+begin_src algol537 begin538 integer ann539 procedure :emp(x); value 2:; integer x;540 ans :1 x;541 ]'act(3. temp);542 comment Now the variable "am" has 6;543 end;544 #+end_src546 Procedure =fact= does not return a value, nor does =temp=; we must use a side547 effect to get the answer out.549 =FACT= is somewhat easier to use in SCHEME. We can call it like an550 ordinary function in SCHEME if we supply an identity function as the second551 argument. For example, =(FACT 3 (LAMBDA (X) X))= returns 6. Alternatively, we552 could write =(FACT 3 (LAHBDA (X) (PRINT X)))=; we do not care about the value553 of this, but about what gets printed. A third way to use the value is to554 write555 =(FACT 3 (LAMBDA (x) (SQRT x)))=556 instead of557 =(SQRT (FACT 3 (LAMBDA (x) x)))=559 In either of these cases we care about the value of the continuation given to560 FACT. Thus we care about the value of FACT if and only if we care about the561 value of its continuation!563 We can redefine other functions to take continuations in the same way.564 For example, suppose we had arithmetic primitives which took continuations; to565 prevent confusion, call the version of "+" which takes a continuation '++",566 etc. Instead of writing567 =(- (+ B Z)(* 4 A C))=568 we can write569 #+begin_src clojure570 (in-ns 'categorical.imperative)571 (defn enable-continuation "Makes f take a continuation as an additional argument."[f]572 (fn[& args] ((fn[k] (k (apply f (reverse (rest (reverse args)))))) (last args)) ))573 (def +& (enable-continuation +))574 (def *& (enable-continuation *))575 (def -& (enable-continuation -))578 (defn quadratic[a b c k]579 (*& b b580 (fn[x] (*& 4 a c581 (fn[y]582 (-& x y k))))))583 #+end_src585 where =k= is the continuation for the entire expression.587 This is an obscure way to write an algebraic expression, and we would588 not advise writing code this way in general, but continuation-passing brings589 out certain important features of the computation:590 - [1] The operations to be performed appear in the order in which they are591 performed. In fact, they /must/ be performed in this592 order. Continuation- passing removes the need for the rule about593 left-to-right argument evaluation{Note Evalorder)594 - [2] In the usual applicative expression there are two implicit595 temporary values: those of =(* B B)= and =(* 4 A C)=. The first of596 these values must be preserved over the computation of the second,597 whereas the second is used as soon as it is produced. These facts598 are /manifest/ in the appearance and use of the variable X and Y in599 the continuation-passing version.601 In short, the continuation-passing version specifies /exactly/ and602 explicitly what steps are necessary to compute the value of the603 expression. One can think of conventional functional application604 for value as being an abbreviation for the more explicit605 continuation-passing style. Alternatively, one can think of the606 interpreter as supplying to each function an implicit default607 continuation of one argument. This continuation will receive the608 value of the function as its argument, and then carry on the609 computation. In an interpreter this implicit continuation is610 represented by the control stack mechanism for function returns.611 Now consider what computational steps are implied by:613 =(LAMBDA (A B C ...) (F X Y Z ...))= when we "apply" the LAMBDA614 expression we have some values to apply it to; we let the names A,615 B, C ... refer to these values. We then determine the values of X,616 Y, Z ... and pass these values (along with "the buck",617 i.e. control!) to the lambda expression F (F is either a lambda618 expression or a name for one). Passing control to F is an619 unconditional transfer. (Note Jrsthack) {Note Hewitthack) Note that620 we want values from X, Y, Z, ... If these are simple expressions,621 such as variables, constants, or LAMBDA expressions, the evaluation622 process is trivial, in that no temporary storage is required. In623 pure continuation-passing style, all evaluations are trivial: no624 combination is nested within another, and therefore no ‘hidden625 temporaries" are required. But if X is a combination, say (G P Q),626 then we want to think of G as a function, because we want a value627 from it, and we will need an implicit temporary place to keep the628 result while evaluating Y and Z. (An interpreter usually keeps these629 temporary places in the control stack!) On the other hand, we do not630 necessarily need a value from F. This is what we mean by tail-631 recursion: F is called tail-recursively, whereas G is not. A better632 name for tail-recursion would be "tail-transfer", since no real633 recursion is implied. This is why we have made such a fuss about634 tail-recursion: it can be used for transfer of control without635 making any commitment about whether the expression expected to636 return a value. Thus it can be used to model statement-like control637 structures. Put another way, tail—recursion does not require a638 control stack as nested recursion does. In our models of iteration639 and imperative style all the LAMBDA expressions used for control (to640 simulate GO statements, for example) are called in tail-recursive641 fashion.643 ** Escape Expressions644 Reynolds [Reynolds 72] defines the construction645 = escape x in r =647 to evaluate the expression $r$ in an environment such that the variable $x$ is648 bound to an escape function. If the escape function is never applied, then649 the value of the escape expression is the value of $r$. If the escape function650 is applied to an argument $a$, however, then evaluation of $r$ is aborted and the651 escape expression returns $a$. {Note J-operator} (Reynolds points out that652 this definition is not quite accurate, since the escape function may be called653 even after the escape expression has returned a value; if this happens, it654 “returns again"!)656 As an example of the use of an escape expression, consider this657 procedure to compute the harmonic mean of an array of numbers. If any of the658 numbers is zero, we want the answer to be zero. We have a function hannaunl659 which will sum the reciprocals of numbers in an array, or call an escape660 function with zero if any of the numbers is zero. (The implementation shown661 here is awkward because ALGOL requires that a function return its value by662 assignment.)663 #+begin_src algol664 begin665 real procedure Imrmsum(a, n. escfun)§ p666 real array at integer n; real procedure esciun(real);667 begin668 real sum;669 sum := 0;670 for i :1 0 until n-l do671 begin672 if a[i]=0 then cscfun(0);673 sum :1 sum ¢ I/a[i];674 enm675 harmsum SI sum;676 end: .677 real array b[0:99]:678 print(escape x in I00/hm-m.mm(b, 100, x));679 end680 #+end_src682 If harmsum exits normally, the number of elements is divided by the sum and683 printed. Otherwise, zero is returned from the escape expression and printed684 without the division ever occurring.685 This program can be written in SCHEME using the built-in escape686 operator =CATCH=:688 #+begin_src clojure689 (in-ns 'categorical.imperative)690 (defn harmonic-sum[coll escape]691 ((fn [i sum]692 (cond (= i (count coll) ) sum693 (zero? (nth coll i)) (escape 0)694 :else (recur (inc i) (+ sum (/ 1 (nth coll i))))))695 0 0))697 #+end_src699 This actually works, but elucidates very little about the nature of ESCAPE.700 We can eliminate the use of CATCH by using continuation-passing. Let us do701 for HARMSUM what we did earlier for FACT. Let it take an extra argument C,702 which is called as a function on the result.704 #+begin_src clojure705 (in-ns 'categorical.imperative)706 (defn harmonic-sum-escape[coll escape k]707 ((fn[i sum]708 (cond (= i (count coll)) (k sum)709 (zero? (nth coll i)) (escape 0)710 (recur (inc i) (+ sum (/ 1 (nth coll i))))))711 0 0))713 (let [B (range 0 100)714 after-the-catch println]715 (harmonic-sum-escape B after-the-catch (fn[y] (after-the-catch (/ 100 y)))))717 #+end_src719 Notice that if we use ESCFUN, then C does not get called. In this way the720 division is avoided. This example shows how ESCFUN may be considered to be an721 "alternate continuation".723 ** Dynamic Variable Scoping724 In this section we will consider the problem of dynamically scoped, or725 "fluid", variables. These do not exist in ALGOL, but are typical of many LISP726 implementations, ECL, and APL. He will see that fluid variables may be727 modelled in more than one way, and that one of these is closely related to728 continuation—pass1ng.730 *** Free (Global) Variables731 Consider the following program to compute square roots:733 #+begin_src clojure734 (defn sqrt[x tolerance]735 (736 (DEFINE soar737 (LAHBDA (x EPSILON)738 (Pace (ANS)739 (stro ANS 1.0)740 A (coup ((< (ADS (-s x (~s ANS ANS))) EPSILON)741 (nzruau ANS))) '742 (sero ANS (//s (+5 x (//s x ANS)) 2.0))743 (60 A)))) .744 This function takes two arguments: the radicand and the numerical tolerance745 for the approximation. Now suppose we want to write a program QUAD to compute746 solutions to a quadratic equation; p747 (perms QUAD748 (LAHDDA (A a c)749 ((LAHBDA (A2 sonmsc) '750 (LIST (/ (+ (- a) SQRTDISC) AZ)751 (/ (- (- B) SORTOISC) AZ)))752 (' 2 A)753 - (SORT (- (9 D 2) (' 4 A C)) (tolerance>))))754 #+end_src755 It is not clear what to write for (tolerance). One alternative is to pick756 some tolerance for use by QUAD and write it as a constant in the code. This757 is undesirable because it makes QUAD inflexible and hard to change. _Another758 is to make QUAD take an extra argument and pass it to SQRT:759 (DEFINE QUAD760 (LANODA (A 8 C EPSILON)761 (soar ... EPSILON) ...))762 This is undesirable because EPSILDN is not really part of the problem QUAD is763 supposed to solve, and we don't want the user to have to provide it.764 Furthermore, if QUAD were built into some larger function, and that into765 another, all these functions would have to pass EPSILON down as an extra766 argument. A third.possibi1ity would be to pass the SQRT function as an767 argument to QUAD (don't laugh!), the theory being to bind EPSILON at the768 appropriate level like this: A U‘ A769 (QUAD 3 A 5 (LAMBDA (X) (SORT X <toIerance>)))770 where we define QUAD as:771 (DEFINE QUAD772 (LAMBDA (A a c soar) ...))