This repository has been archived on 2022-08-10. You can view files and clone it, but cannot push or open issues or pull requests.
chez-openbsd/csug/control.stex

663 lines
22 KiB
Text
Raw Normal View History

2022-07-29 15:12:07 +02:00
% Copyright 2005-2017 Cisco Systems, Inc.
%
% Licensed under the Apache License, Version 2.0 (the "License");
% you may not use this file except in compliance with the License.
% You may obtain a copy of the License at
%
% http://www.apache.org/licenses/LICENSE-2.0
%
% Unless required by applicable law or agreed to in writing, software
% distributed under the License is distributed on an "AS IS" BASIS,
% WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
% See the License for the specific language governing permissions and
% limitations under the License.
\chapter{Control Structures\label{CHPTCONTROL}}
This chapter describes {\ChezScheme} extensions to the set of standard
control structures.
See Chapter~\ref{TSPL:CHPTCONTROL} of {\TSPLFOUR} or the Revised$^6$ Report
on Scheme for a description of standard control structures.
\section{Conditionals}
%----------------------------------------------------------------------------
\entryheader
\formdef{exclusive-cond}{\categorysyntax}{(exclusive-cond \var{clause_1} \var{clause_2} \dots)}
\returns see below
\listlibraries
\endentryheader
\scheme{exclusive-cond} is a version of \scheme{cond}
(Section~\ref{TSPL:SECTCONDITIONALS} of {TSPLFOUR}) that differs
from \scheme{cond} in that the tests embedded within the clauses
are assumed to be exclusive in the sense that if one of the tests
is true, the others are not.
This allows the implementation to reorder clauses when profiling
information is available at expansion time (Section~\ref{SECTMISCPROFILE}).
The \scheme{(\var{test})} form of clause is not supported.
The order chosen when profiling information is available is based
on the relative numbers of times the RHS of each clause is executed,
and \scheme{(\var{test})} has no RHS.
\scheme{(\var{test} => values)} is equivalent, albeit less concise.
%----------------------------------------------------------------------------
\entryheader
\formdef{case}{\categorysyntax}{(case \var{expr_0} \var{clause_1} \var{clause_2} \dots)}
\returns see below
\listlibraries
\endentryheader
\noindent
Each clause but the last must take one of the forms:
\schemedisplay
((\var{key} \dots) \var{expr_1} \var{expr_2} \dots)
(\var{key} \var{expr_1} \var{expr_2} \dots)
\endschemedisplay
\noindent
where each \var{key} is a datum distinct from the other keys.
The last clause may be in the above form or it may be an
\index{\scheme{else}}\scheme{else} clause of the form
\schemedisplay
(else \var{expr_1} \var{expr_2} \dots)
\endschemedisplay
\var{expr_0} is evaluated and the result is compared
(using \scheme{equal?}) against the keys of each clause in order.
If a clause containing a matching key is found, the
expressions \scheme{\var{expr_1} \var{expr_2} \dots} are evaluated in sequence
and the values of the last expression are returned.
If none of the clauses contains a matching key and an \scheme{else} clause
is present, the expressions \scheme{\var{expr_1} \var{expr_2} \dots} of the
\scheme{else} clause are evaluated in sequence and the values of the last
expression are returned.
If none of the clauses contains a matching key and no \scheme{else} clause
is present, the value or values are unspecified.
The Revised$^6$ Report version of \scheme{case} does not support singleton
keys (the second of the first two clause forms above) and uses
\scheme{eqv?} rather than \scheme{equal?} as the comparison procedure.
Both versions are defined in terms of \scheme{exclusive-cond} so that
if profiling information is available at expansion time, the clauses will
be reordered to put those that are most frequently executed first.
\schemedisplay
(let ([ls '(ii iv)])
(case (car ls)
[i 1]
[ii 2]
[iii 3]
[(iiii iv) 4]
[else 'out-of-range])) ;=> 2
(define p
(lambda (x)
(case x
[("abc" "def") 'one]
[((a b c)) 'two]
[else #f])))
(p (string #\d #\e #\f)) ;=> one
(p '(a b c)) ;=> two
\endschemedisplay
%----------------------------------------------------------------------------
\noskipentryheader
\formdef{record-case}{\categorysyntax}{(record-case \var{expr} \var{clause_1} \var{clause_2} \dots)}
\returns see explanation
\listlibraries
\endnoskipentryheader
\noindent
\scheme{record-case} is a restricted form of \scheme{case} that supports the
destructuring of \index{records}\emph{records}, or \index{tagged lists}\emph{tagged lists}.
A record has as its first element a tag that determines what ``type''
of record it is; the remaining elements are the fields of the record.
Each clause but the last must take the form
\schemedisplay
((\var{key} \dots) \var{formals} \var{body_1} \var{body_2} \dots)
\endschemedisplay
\noindent
where each \var{key} is a datum distinct from the other keys.
The last clause may be in the above form or it may be an
\index{\scheme{else}}\scheme{else} clause of the form
\schemedisplay
(else \var{body_1} \var{body_2} \dots)
\endschemedisplay
\var{expr} must evaluate to a pair.
\var{expr} is evaluated and the car of its value is compared
(using \scheme{eqv?}) against the keys of each clause in order.
If a clause containing a matching key is found, the variables in
\var{formals} are bound to the remaining elements
of the list and the expressions
\scheme{\var{body_1} \var{body_2} \dots} are evaluated in sequence.
The value of the last expression is returned.
The effect is identical to the application of
\schemedisplay
(lambda \var{formals} \var{body_1} \var{body_2} \dots)
\endschemedisplay
\noindent
to the cdr of the list.
If none of the clauses contains a matching key and an \scheme{else} clause
is present, the expressions \scheme{\var{body_1} \var{body_2} \dots} of the
\scheme{else} clause are evaluated in sequence and the value of the last
expression is returned.
If none of the clauses contains a matching key and no \scheme{else} clause
is present, the value is unspecified.
\schemedisplay
(define calc
(lambda (x)
(record-case x
[(add) (x y) (+ x y)]
[(sub) (x y) (- x y)]
[(mul) (x y) (* x y)]
[(div) (x y) (/ x y)]
[else (assertion-violationf 'calc "invalid expression ~s" x)])))
(calc '(add 3 4)) ;=> 7
(calc '(div 3 4)) ;=> 3/4
\endschemedisplay
\section{Mapping and Folding}
%----------------------------------------------------------------------------
\noskipentryheader
\formdef{ormap}{\categoryprocedure}{(ormap \var{procedure} \var{list_1} \var{list_2} \dots)}
\returns see explanation
\listlibraries
\endnoskipentryheader
\noindent
\scheme{ormap} is identical to the Revised$^6$ Report \scheme{exists}.
%----------------------------------------------------------------------------
\entryheader
\formdef{andmap}{\categoryprocedure}{(andmap \var{procedure} \var{list_1} \var{list_2} \dots)}
\returns see explanation
\listlibraries
\endentryheader
\noindent
\scheme{andmap} is identical to the Revised$^6$ Report \scheme{for-all}.
\section{Continuations}
{\ChezScheme} supports one-shot continuations as well as the standard
multi-shot continuations obtainable via \scheme{call/cc}.
One-shot continuations are continuations that may be invoked at most
once, whether explicitly or implicitly.
They are obtained with \scheme{call/1cc}.
%----------------------------------------------------------------------------
\entryheader
\formdef{call/1cc}{\categoryprocedure}{(call/1cc \var{procedure})}
\returns see below
\listlibraries
\endentryheader
\noindent
\scheme{call/1cc} obtains its continuation and passes it to \var{procedure},
which should accept one argument.
The continuation itself is represented by a procedure.
This procedure normally takes one argument but may take an arbitrary
number of arguments depending upon whether the context of the call
to \scheme{call/1cc}
expects multiple return values or not.
When this procedure is applied to a value or values, it returns the values
to the continuation of the \scheme{call/1cc} application.
The continuation obtained by \scheme{call/1cc} is a
\index{one-shot continuations}``one-shot continuation.''
A one-shot continuation should not be returned to multiple times, either
by invoking the continuation or returning normally from \var{procedure} more
than once.
A one-shot continuation is ``promoted'' into a normal (multishot)
continuation, however, if it is
still active when a
normal continuation is obtained by \scheme{call/cc}.
After a one-shot continuation is promoted into a multishot continuation,
it behaves exactly as if it had been obtained via \scheme{call/cc}.
This allows \scheme{call/cc} and \scheme{call/1cc} to be used together
transparently in many applications.
One-shot continuations may be more efficient for some applications than
multishot continuations.
See the paper ``Representing control in the presence of one-shot
continuations''~\cite{Bruggeman:oneshots} for more information about
one-shot continuations, including how they are implemented in
{\ChezScheme}.
The following examples highlight the similarities and differences
between one-shot and normal continuations.
\schemedisplay
(define prod
; compute the product of the elements of ls, bugging out
; with no multiplications if a zero element is found
(lambda (ls)
(lambda (k)
(if (null? ls)
1
(if (= (car ls) 0)
(k 0)
(* (car ls) ((prod (cdr ls)) k)))))))
(call/cc (prod '(1 2 3 4))) ;=> 24
(call/1cc (prod '(1 2 3 4))) ;=> 24
(call/cc (prod '(1 2 3 4 0))) ;=> 0
(call/1cc (prod '(1 2 3 4 0))) ;=> 0
(let ([k (call/cc (lambda (x) x))])
(k (lambda (x) 0))) ;=> 0
(let ([k (call/1cc (lambda (x) x))])
(k (lambda (x) 0))) ;=> \var{exception}
\endschemedisplay
%----------------------------------------------------------------------------
\entryheader\label{dynamic-wind}
\formdef{dynamic-wind}{\categoryprocedure}{(dynamic-wind \var{in} \var{body} \var{out})}
\formdef{dynamic-wind}{\categoryprocedure}{(dynamic-wind \var{critical?} \var{in} \var{body} \var{out})}
\returns values resulting from the application of \var{body}
\listlibraries
\endentryheader
The first form is identical to the Revised$^6$ Report \scheme{dynamic-wind}.
When the optional \var{critical?} argument is present and non-false,
the \var{in} thunk is invoked in a critical section along with the code
that records that the body has been entered, and the \var{out} thunk is
invoked in a critical section along with the code that records
that the body has been exited.
Extreme caution must be taken with this form of \scheme{dynamic-wind},
since an error or long-running computation can leave interrupts
and automatic garbage collection disabled.
\section{Engines\label{SECTENGINES}}
\index{engines}Engines are a high-level process abstraction supporting
\index{timed preemption}\emph{timed preemption}~\cite{Dybvig:engines,Haynes:abstracting}.
Engines may be used to simulate \index{multiprocessing}multiprocessing, implement operating
system kernels, and perform \index{nondeterministic computations}nondeterministic computations.
%----------------------------------------------------------------------------
\entryheader
\formdef{make-engine}{\categoryprocedure}{(make-engine \var{thunk})}
\returns an engine
\listlibraries
\endentryheader
An engine is created by passing a thunk (no argument procedure)
to \scheme{make-engine}.
The body of the thunk is the computation to be performed by the engine.
An engine itself is a procedure of three arguments:
\begin{description}
\item[\var{ticks}:]
\index{ticks@\var{ticks}|see{engines}}a positive integer that specifies
the amount of \emph{fuel} to be given
to the engine.
An engine executes until this fuel runs out or until its computation
finishes.
\item[\var{complete}:]
\index{complete@\var{complete}|see{engines}}a procedure of one or more
arguments that
specifies what to do if the computation finishes.
Its arguments are the amount of fuel left over and the
values produced by the computation.
\item[\var{expire}:]
\index{expire@\var{expire}|see{engines}}a procedure of one argument that
specifies what to do if the fuel runs
out before the computation finishes.
Its argument is a new engine capable of continuing the computation
from the point of interruption.
\end{description}
When an engine is applied to its arguments, it sets up a timer
to fire in \var{ticks} time units.
(See \scheme{set-timer} on page~\pageref{desc:set-timer}.)
If the engine computation completes before the timer expires, the
system invokes \var{complete}, passing
it the number of \var{ticks} left over and
the values produced by the computation.
If, on the other hand, the timer goes off before the engine computation
completes, the system creates a new engine from the continuation of
the interrupted computation and passes this engine to \var{expire}.
\var{complete} and \var{expire} are invoked in the continuation
of the engine invocation.
An implementation of engines is given
in Section~\ref{TSPL:SECTEXENGINES}.
of {\TSPLFOUR}.
Do not use the timer interrupt (see \index{\scheme{set-timer}}\scheme{set-timer}) and \index{engines}engines
at the same time, since engines are implemented in terms of the timer.
The following example creates an engine from a trivial computation,
3, and gives the engine 10 ticks.
\schemedisplay
(define eng
(make-engine
(lambda () 3)))
(eng 10
(lambda (ticks value) value)
(lambda (x) x)) ;=> 3
\endschemedisplay
It is often useful to pass \scheme{list} as the \var{complete}
procedure to an engine, causing an engine that completes to return a
list whose first element is the ticks remaining and whose remaining elements
are the values returned by the computation.
\schemedisplay
(define eng
(make-engine
(lambda () 3)))
(eng 10
list
(lambda (x) x)) ;=> (9 3)
\endschemedisplay
\noindent
In the example above, the value is 3 and there are 9 ticks left over,
i.e., it takes one unit of fuel to evaluate 3.
(The fuel amounts given here are for illustration only.
Your mileage may vary.)
Typically, the engine computation does not finish in one try.
\index{\scheme{fibonacci}}The following example displays the use of an engine to
compute the 10th Fibonacci number in steps.
\schemedisplay
(define fibonacci
(lambda (n)
(let fib ([i n])
(cond
[(= i 0) 0]
[(= i 1) 1]
[else (+ (fib (- i 1))
(fib (- i 2)))]))))
(define eng
(make-engine
(lambda ()
(fibonacci 10))))
(eng 50
list
(lambda (new-eng)
(set! eng new-eng)
"expired")) ;=> "expired"
(eng 50
list
(lambda (new-eng)
(set! eng new-eng)
"expired")) ;=> "expired"
(eng 50
list
(lambda (new-eng)
(set! eng new-eng)
"expired")) ;=> "expired"
(eng 50
list
(lambda (new-eng)
(set! eng new-eng)
"expired")) ;=> (21 55)
\endschemedisplay
\noindent
Each time the engine's fuel runs out, the \var{expire} procedure assigns
\scheme{eng} to the new engine.
The entire computation requires four blocks of 50 ticks to complete; of the
last 50 it uses all but 21.
Thus, the total amount of fuel used is 179 ticks.
This leads to the following procedure, \scheme{mileage}, which ``times'' a
computation using engines:
\schemedisplay
(define mileage
(lambda (thunk)
(let loop ([eng (make-engine thunk)] [total-ticks 0])
(eng 50
(lambda (ticks . values)
(+ total-ticks (- 50 ticks)))
(lambda (new-eng)
(loop new-eng
(+ total-ticks 50)))))))
(mileage (lambda () (fibonacci 10))) ;=> 179
\endschemedisplay
\noindent
The choice of 50 for the number of ticks to use each time is
arbitrary, of course.
It might make more sense to pass a much larger number, say 10000,
in order to reduce the number of times the computation is interrupted.
The next procedure is similar to \scheme{mileage}, but it returns a list
of engines, one for each tick it takes to complete the computation.
Each of the engines in the list represents a ``snapshot'' of the
computation, analogous to a single frame of a moving picture.
\scheme{snapshot} might be useful for ``single stepping'' a computation.
\schemedisplay
(define snapshot
(lambda (thunk)
(let again ([eng (make-engine thunk)])
(cons eng
(eng 1 (lambda (t . v) '()) again)))))
\endschemedisplay
\noindent
The recursion embedded in this procedure is rather strange.
The complete procedure performs the base case, returning the empty
list, and the expire procedure performs the recursion.
The next procedure, \index{\scheme{round-robin}}\scheme{round-robin}, could be the basis for a simple
time-sharing \index{operating system}operating system.
\scheme{round-robin} maintains a queue of processes (a list of engines),
cycling through the queue in a \emph{round-robin} fashion, allowing each
process to run for a set amount of time.
\scheme{round-robin} returns a list of the values returned by the engine
computations in the order that the computations complete.
Each computation is assumed to produce exactly one value.
\schemedisplay
(define round-robin
(lambda (engs)
(if (null? engs)
'()
((car engs)
1
(lambda (ticks value)
(cons value (round-robin (cdr engs))))
(lambda (eng)
(round-robin
(append (cdr engs) (list eng))))))))
\endschemedisplay
\noindent
Since the amount of fuel supplied each time, one tick, is constant,
the effect of \scheme{round-robin} is to return a list of the values sorted
from the quickest to complete to the slowest to complete.
Thus, when we call \scheme{round-robin} on a list of engines, each computing
one of the Fibonacci numbers, the output list is sorted with the earlier
Fibonacci numbers first, regardless of the order of the input list.
\schemedisplay
(round-robin
(map (lambda (x)
(make-engine
(lambda ()
(fibonacci x))))
'(4 5 2 8 3 7 6 2))) ;=> (1 1 2 3 5 8 13 21)
\endschemedisplay
More interesting things can happen if the amount of fuel varies
each time through the loop.
\index{nondeterministic computations}In this case, the computation would
be nondeterministic, i.e., the results would vary from call to call.
The following syntactic form, \index{\scheme{por} (parallel-or)}\scheme{por} (parallel-or), returns the
first of its expressions to complete with a true value.
\scheme{por} is implemented with the procedure \scheme{first-true}, which is
similar to \scheme{round-robin} but quits when any of the engines
completes with a true value.
If all of the engines complete, but none with a true value,
\scheme{first-true} (and hence \scheme{por}) returns \scheme{#f}.
Also, although \scheme{first-true} passes a fixed amount of fuel to each
engine, it chooses the next engine to run at random, and is thus
nondeterministic.
\schemedisplay
(define-syntax por
(syntax-rules ()
[(_ x ...)
(first-true
(list (make-engine (lambda () x)) ...))]))
(define first-true
(let ([pick
(lambda (ls)
(list-ref ls (random (length ls))))])
(lambda (engs)
(if (null? engs)
#f
(let ([eng (pick engs)])
(eng 1
(lambda (ticks value)
(or value
(first-true
(remq eng engs))))
(lambda (new-eng)
(first-true
(cons new-eng
(remq eng engs))))))))))
\endschemedisplay
\noindent
The list of engines is maintained with \scheme{pick}, which randomly
chooses an element of the list, and \scheme{remq}, which removes the
chosen engine from the list.
Since \scheme{por} is nondeterministic, subsequent uses with the same
expressions may not return the same values.
\schemedisplay
(por 1 2 3) ;=> 2
(por 1 2 3) ;=> 3
(por 1 2 3) ;=> 2
(por 1 2 3) ;=> 1
\endschemedisplay
\noindent
Furthermore, even if one of the expressions is an infinite loop,
\scheme{por} still finishes as long as one of the other expressions
completes and returns a true value.
\schemedisplay
(por (let loop () (loop)) 2) ;=> 2
\endschemedisplay
\noindent
With \scheme{engine-return} and \scheme{engine-block}, it is possible to
terminate an engine explicitly.
\scheme{engine-return} causes the engine to complete, as if the
computation had finished.
Its arguments are passed to the \var{complete} procedure along with the
number of ticks remaining.
It is essentially a nonlocal exit from the engine.
Similarly, \scheme{engine-block} causes the engine to expire, as if the
timer had run out.
A new engine is made from the continuation of the call to \scheme{engine-block}
and passed to the \var{expire} procedure.
%----------------------------------------------------------------------------
\entryheader
\formdef{engine-block}{\categoryprocedure}{(engine-block)}
\returns does not return
\listlibraries
\endentryheader
\noindent
This causes a running engine to stop, create a new engine capable
of continuing the computation, and pass the new engine to the original
engine's third argument
(the expire procedure).
Any remaining fuel is forfeited.
\schemedisplay
(define eng
(make-engine
(lambda ()
(engine-block)
"completed")))
(eng 100
(lambda (ticks value) value)
(lambda (x)
(set! eng x)
"expired")) ;=> "expired"
(eng 100
(lambda (ticks value) value)
(lambda (x)
(set! eng x)
"expired")) ;=> "completed"
\endschemedisplay
%----------------------------------------------------------------------------
\entryheader
\formdef{engine-return}{\categoryprocedure}{(engine-return \var{obj} \dots)}
\returns does not return
\listlibraries
\endentryheader
\noindent
This causes a running engine to stop and pass control to the
engine's \var{complete} argument.
The first argument passed to the complete procedure is the amount of
fuel remaining, as usual, and
the remaining arguments are the objects \scheme{\var{obj} \dots}
passed to \scheme{engine-return}.
\schemedisplay
(define eng
(make-engine
(lambda ()
(reverse (engine-return 'a 'b 'c)))))
(eng 100
(lambda (ticks . values) values)
(lambda (new-eng) "expired")) ;=> (a b c)
\endschemedisplay