s7

s7 is a Scheme implementation intended as an extension language for other applications, primarily Snd and Common Music. It exists as just two files, s7.c and s7.h, that want only to disappear into someone else's source tree. There are no libraries, no run-time init files, and no configuration scripts. It can be built as a stand-alone interpreter (see below). s7test.scm is a regression test for s7. A tarball is available: ftp://ccrma-ftp.stanford.edu/pub/Lisp/s7.tar.gz.

s7 is the default extension language of Snd and sndlib (http://ccrma.stanford.edu/software/snd/), and Rick Taube's Common Music (commonmusic at sourceforge). There are X, Motif, Gtk, and openGL bindings in libxm in the Snd tarball, or at ftp://ccrma-ftp.stanford.edu/pub/Lisp/libxm.tar.gz. If you're running s7 in a context that has getenv, file-exists?, and system (Snd for example, or s7 with WITH_SYSTEM_EXTRAS set), you can use s7-slib-init.scm to gain easy access to slib. This init file is named "s7.init" in the slib distribution.

Although it is a descendant of tinyScheme, s7 is closest as a Scheme dialect to Guile 1.8. I believe it is compatible with r5rs: you can just ignore all the additions discussed in this file. It has continuations, ratios, complex numbers, macros, keywords, hash-tables, multiprecision arithmetic, generalized set!, and so on. It does not have syntax-rules or any of its friends, and it does not think there is any such thing as an inexact integer.

This file assumes you know about Scheme and all its problems, and want a quick tour of where s7 is different. (Well, it was quick once upon a time). The main difference: if it's in s7, it's a first-class citizen of s7, and that includes macros, environments, and syntactic forms.

I originally used a small font for scholia, but now I have to squint to read that tiny text, so less-than-vital commentaries are shown in the normal font, but indented and on a sort of brownish background.

Danger!

Men Working

in Trees

multiprecision arithmetic

All numeric types, integers, ratios, reals, and complex numbers are supported. The basic integer and real types are defined in s7.h, defaulting to long long int and double. pi is predefined, as are most-positive-fixnum and most-negative-fixnum. s7 can be built with multiprecision support for all types, using the gmp, mpfr, and mpc libraries (set WITH_GMP to 1 in s7.c). If multiprecision arithmetic is enabled, the following functions are included: bignum, bignum?, and bignum-precision. bignum-precision, which defaults to 128, sets the number of bits each float takes. pi automatically reflects the current bignum-precision:

> pi
3.141592653589793238462643383279502884195E0
> (bignum-precision)
128
> (set! (bignum-precision) 256)
256
> pi
3.141592653589793238462643383279502884197169399375105820974944592307816406286198E0

bignum? returns #t if its argument is a big number of some type; I use "bignum" for any big number, not just integers. To create a big number, either include enough digits to overflow the default types, or use the bignum function. Its argument is a string representing the desired number:

> (bignum "123456789123456789")
123456789123456789
> (bignum "1.123123123123123123123123123")
1.12312312312312312312312312300000000009E0

In the non-gmp case, if s7 is built using doubles (s7_Double in s7.h), the float "epsilon" is around (expt 2 -53), or about 1e-16. In the gmp case, it is around (expt 2 (- (bignum-precision))). So in the default case (precision = 128), using gmp:

> (= 1.0 (+ 1.0 (expt 2.0 -128)))
#t
> (= 1.0 (+ 1.0 (expt 2.0 -127)))
#f

and in the non-gmp case:

> (= 1.0 (+ 1.0 (expt 2 -53)))
#t
> (= 1.0 (+ 1.0 (expt 2 -52)))
#f

In the gmp case, integers and ratios are limited only by the size of memory, but reals are limited by bignum-precision. This means, for example, that

> (floor 1e56) ; bignum-precision is 128
99999999999999999999999999999999999999927942405962072064
> (set! (bignum-precision) 256)
256
> (floor 1e56)
100000000000000000000000000000000000000000000000000000000

The non-gmp case is similar, but it's easy to find the edge cases:

> (floor (+ 0.9999999995 (expt 2.0 23)))
8388609

math functions

s7 includes:

The random function can take any numeric argument, including 0. The following constants are predefined: pi, most-positive-fixnum, most-negative-fixnum. Other math-related differences between s7 and r5rs:

> (exact? 1.0)
#f
> (rational? 1.5)
#f
> (floor 1.4)
1
> (remainder 2.4 1)
0.4
> (modulo 1.4 1.0)
0.4
> (lcm 3/4 1/6)
3/2
> (log 8 2)
3
> (number->string 0.5 2)
"0.1"
> (string->number "0.1" 2)
0.5
> (rationalize 1.5)
3/2
> (make-rectangular 1/2 0)
1/2
> (logbit? 6 1) ; argument order, (logbit? int index), follows gmp, not CL
#t

The exponent itself is always in base 10; this follows gmp usage. Scheme normally uses "@" for its useless polar notation, but that means (string->number "1e1" 16) is ambiguous — is the "e" a digit or an exponent marker? In s7, "@" is an exponent marker.

> (string->number "1e9" 2)  ; (expt 2 9)
512.0
> (string->number "1e1" 12) ; "e" is not a digit in base 12
#f
> (string->number "1e1" 16) ; (+ (* 1 16 16) (* 14 16) 1)
481
> (string->number "1.2e1" 3); (* 3 (+ 1 2/3))
5.0

Should s7 predefine the numbers +inf.0, -inf.0, and nan.0? It doesn't currently, but you can get them via log or 1/0 (see below). But what is (/ 1.0 0.0)? s7 gives a "division by zero" error here, and also in (/ 1 0). Guile returns +inf.0 in the first case, which seems reasonable, but a "numerical overflow" error in the second. Slightly weirder is (expt 0.0 0+i). Currently s7 returns 0.0, Guile returns +nan.0+nan.0i, Clisp and sbcl throw an error. Everybody agrees that (expt 0 0) is 1, and Guile thinks that (expt 0.0 0.0) is 1.0. But (expt 0 0.0) and (expt 0.0 0) return different results in Guile (1 and 1.0), both are 0.0 in s7, the first is an error in Clisp, but the second returns 1, and so on — what a mess! This mess was made a lot worse than it needs to be when the IEEE decreed that 0.0 equals -0.0, so we can't tell them apart, but that they produce different results in nearly every use!

scheme@(guile-user)> (= -0.0 0.0)
#t
scheme@(guile-user)> (negative? -0.0)
#f
scheme@(guile-user)> (= (/ 1.0 0.0) (/ 1.0 -0.0))
#f
scheme@(guile-user)>  (< (/ 1.0 -0.0) -1e100 1e100 (/ 1.0 0.0))
#t

How can they be equal? In s7, the sign of -0.0 is ignored, and they really are equal. One other oddity: two floats can satisfy eq? and yet not be eqv?: (eq? nan.0 nan.0) might be #t (it is unspecified), but (eqv? nan.0 nan.0) is #f. The same problem afflicts memq and assq.

The random function takes a range and an optional state, and returns a number between zero and the range, of the same type as the range. It is perfectly reasonable to use a range of 0, in which case random returns 0. make-random-state creates a new random state from a seed. If no state is passed, random uses some default state initialized from the current time. random-state? returns #t if passed a random state object.

> (random 0)
0
> (random 1.0)
0.86331198514245
> (random 3/4)
654/1129
> (random 1+i)
0.86300308872748+0.83601002730848i
> (random -1.0)
-0.037691127513267
> (define r0 (make-random-state 1234))
r0
> (random 100 r0)
94
> (random 100 r0)
19
> (define r1 (make-random-state 1234))
r1
> (random 100 r1)
94
> (random 100 r1)
19

copy the random-state to save a spot in a random number sequence, or save the random-state as a list via random-state->list, then to restart from that point, apply make-random-state to that list.

I can't find the right tone for this section; this is the 400-th revision; I wish I were a better writer!

In some Schemes, "rational" means "could possibly be expressed equally well as a ratio: floats are approximations". In s7 it's: "is actually expressed (at the scheme level) as a ratio (or an integer of course)"; otherwise "rational?" is the same as "real?":

(not-s7-scheme)> (rational? (sqrt 2))
#t

That 1.0 is represented at the IEEE-float level as a sort of ratio does not mean it has to be a scheme ratio; the two notions are independent.

But that confusion is trivial compared to the completely nutty "inexact integer". As I understand it, "inexact" originally meant "floating point", and "exact" meant integer or ratio of integers. But words have a life of their own. 0.0 somehow became an "inexact" integer (although it can be represented exactly in floating point). +inf.0 must be an integer — its fractional part is explicitly zero! But +nan.0... And then there's:

(not-s7-scheme)> (integer? 9007199254740993.1)
#t

When does this matter? I often need to index into a vector, but the index is a float (a "real" in Scheme-speak: its fractional part can be non-zero). In one Scheme:

(not-s7-scheme)> (vector-ref #(0) (floor 0.1))
ERROR: Wrong type (expecting exact integer): 0.0   ; [why?  "it's probably a programmer mistake"!]

Not to worry, I'll use inexact->exact:

(not-s7-scheme)> (inexact->exact 0.1)
3602879701896397/36028797018963968                  ; [why? "floats are ratios"!]

So I end up using the verbose (floor (inexact->exact ...)) everywhere, and even then I have no guarantee that I'll get a legal vector index.

When I started work on s7, I thought perhaps "exact" could mean "is represented exactly in the computer". We'd have integers and ratios exact; reals and complex exact if they are exactly represented in the current floating point implementation. 0.0 and 0.5 might be exact if the printout isn't misleading, and 0.1 is inexact. "integer?" and friends would refer instead to the programmer's point of view. That is, if the programmer uses 1 or if the thing prints as 1, it is the integer 1, whereas 1.0 means floating point (not integer!). And to keep exactness in view, we'd have to monitor which operations introduce inexactness — a kind of interval arithmetic. But then what would inexact->exact do?

I have never seen any use made of the exact/inexact distinction — just wild flailing to try get around it. I think the whole idea is confused and useless, and leads to verbose and buggy code. If we discard it, we can maintain backwards compatibility via:

(define exact? rational?)
(define (inexact? x) (not (rational? x)))
(define inexact->exact rationalize) ; or floor
(define (exact->inexact x) (* x 1.0))

#i and #e are also useless because you can have any number after, for example, #b:

> #b1.1
1.5
> #b1e2
4.0
> #o17.5+i
15.625+1i

Speaking of #b and friends, what should (string->number "#xffff" 2) return?

(define (log-n-of n . ints)     ; return the bits on in exactly n of ints
  (let ((len (length ints)))
    (cond ((= len 0) (if (= n 0) -1 0))
	  ((= n 0)   (lognot (apply logior ints)))
	  ((= n len) (apply logand ints))
	  ((> n len) 0)
	  (#t 
	   (do ((1s 0)
		(prev ints)
		(i 0 (+ i 1)))
	       ((= i len) 1s)
	     (let ((cur (ints i)))
	       (if (= i 0)
		   (set! 1s (logior 1s (logand cur (apply log-n-of (- n 1) (cdr ints)))))
		   (let* ((mid (cdr prev))
			  (nxt (if (= i (- len 1)) () (cdr mid))))
		     (set! (cdr prev) nxt)  
		     (set! 1s (logior 1s (logand cur (apply log-n-of (- n 1) ints))))
		     (set! (cdr prev) mid)
		     (set! prev mid)))))))))

define*, lambda*

define* and lambda* are extensions of define and lambda that make it easier to deal with optional, keyword, and rest arguments. The syntax is very simple: every argument to define* has a default value and is automatically available as a keyword argument. The default value is either #f if unspecified, or given in a list whose first member is the argument name. The last argument can be preceded by :rest or a dot to indicate that all other trailing arguments should be packaged as a list under that argument's name. A trailing or rest argument's default value is '(). You can use :optional and :key, but they are ignored.

(define* (hi a (b 32) (c "hi")) (list a b c))

Here the argument "a" defaults to #f, "b" to 32, etc. When the function is called, the argument names are bound to their default values, then the function's current argument list is scanned. Any name that occurs as a keyword, :arg for example where the parameter name is arg, sets that argument's new value. Otherwise, as values occur, they are plugged into the actual argument list based on their position, counting a keyword/value pair as one argument. This is called an optional-key list in CLM. So, taking the function above as an example:

> (hi 1) 
(1 32 "hi")
> (hi :b 2 :a 3) 
(3 2 "hi")
> (hi 3 2 1) 
(3 2 1)

See s7test.scm for many examples.

(define* (make-parameter initial-value converter)
  (let ((value (if (procedure? converter) (converter initial-value) initial-value)))
    (lambda* ((val #<unspecified>))
      (if (not (eq? val #<unspecified>))
	  (set! value (if (procedure? converter) (converter val) val)))
      value)))

> (define hiho (make-parameter 12))
hiho
> (hiho)
12
> (hiho 32)
32
> (hiho)
32

If you want a version of define* that insists that any arguments before the keyword :optional are required:

(define-macro (define** declarations . forms)
  (let ((name (car declarations))
	(args (cdr declarations)))
    (define (position thing lst count)
      (if (or (null? lst)
	      (not (pair? (cdr lst)))) ; for dotted arg = "rest" arg
	  #f
	  (if (eq? thing (car lst))
	      count
	      (position thing (cdr lst) (+ count 1)))))
    (let ((required-args (position :optional args 0)))
      (if required-args
	  `(define* (,name . func-args)
	     (if (< (length func-args) ,required-args)
		 (error "~A requires ~D argument~A: ~A" 
			',name ,required-args 
                        (if (> ,required-args 1) "s" "") 
                        func-args)
		 (apply (lambda* ,args ,@forms) func-args)))
	  `(define* ,declarations ,@forms)))))

> (define** (hi a :optional (b 23)) (list a b))
hi
> (hi 1)
(1 23)
> (hi)
;hi requires 1 argument: ()

If a define* argument's default value is an expression, it is evaluated in the definition environment at the time of the procedure call:

(let ((c 1))
  (define* (a (b (+ c 1))) b)
  (set! c 2)
  (let ((c 123))
    (a))) ; (+ c 1) here is (+ 2 1) so this returns 3

Since the expression is not evaluated until the procedure is called, it is ok to use variables that are undefined at the definition point:

> (define* (a (b c)) b)
a
> c
;c: unbound variable
> (define c 123)
c
> (a)
123

To try to catch what I believe are usually mistakes, I added two error checks. One is triggered if you set the same parameter twice in the same call, and the other if an unknown keyword is encountered in the key position. These problems arise in a case such as

(define* (f (a 1) (b 2)) (list a b))

You could do any of the following by accident:

(f 1 :a 2)  ; what is a?
(f :b 1 2)  ; what is b?
(f :c 3)    ; did you really want a to be :c and b to be 3?

In the last case, to pass a keyword deliberately, either include the argument keyword: (f :a :c), or make the default value a keyword: (define* (f (a :c) ...)). To turn off this error check, add :allow-other-keys at the end of the parameter list.

s7's lambda* arglist handling is not the same as CL's lambda-list. First, you can have more than one :rest parameter:

> ((lambda* (:rest a :rest b) (map + a b)) 1 2 3 4 5) 
'(3 5 7 9)

and second, the rest parameter, if any, takes up an argument slot just like any other argument:

> ((lambda* ((b 3) :rest x (c 1)) (list b c x)) 32)
(32 1 ())

> ((lambda* ((b 3) :rest x (c 1)) (list b c x)) 1 2 3 4 5)
(1 3 (2 3 4 5))

CL would agree with the first case if we used &key for 'c', but would give an error in the second. Of course, the major difference is that s7 keyword arguments don't insist that the key be present. The :rest argument is needed in cases like these because we can't use an expression such as:

> ((lambda* ((a 3) . b c) (list a b c)) 1 2 3 4 5)
stray dot?

Yet another nit: the :rest argument is not considered a keyword argument, so

> (define* (f :rest a) a)
f
> (f :a 1)
(:a 1)

macros

define-macro, define-macro*, defmacro, defmacro*, macroexpand, gensym, and macro? implement the standard old-time macros.

> (define-macro (trace f)
    `(define ,f 
       (apply lambda 'args 
         `((format #t "(~A ~{~A~^ ~}) -> " ',,f args)
           (let ((val (apply ,,f args))) 
             (format #t "~A~%" val) 
             val)))))
> (trace abs)
abs
> (abs -1.5)
(abs -1.5) -> 1.5
1.5

macroexpand can help debug a macro:

> (define-macro (add-1 arg) `(+ 1 ,arg))
add-1
> (macroexpand (add-1 32))
(+ 1 32)

gensym returns a symbol that is guaranteed to be unused. It takes an optional string argument giving the new symbol name's prefix.

(defmacro pop! (sym)
  (let ((v (gensym)))
    `(let ((,v (car ,sym)))
       (set! ,sym (cdr ,sym))
       ,v)))

As in define*, the starred forms give optional and keyword arguments:

> (define-macro* (add-2 a (b 2)) `(+ ,a ,b))
add-2
> (add-2 1 3)
4
> (add-2 1)
3
> (add-2 :b 3 :a 1)
4

A bacro is a macro that expands its body and evaluates the result in the calling environment.

(define setf
  (let ((args (gensym)))
    (apply define-bacro
	   `((setf-1 . ,args)        
	     (if (not (null? ,args))
		 (begin
		   (apply set! (car ,args) (cadr ,args) ())
		   (apply setf (cddr ,args))))))
    setf-1))

Here define-bacro makes it possible to implement setf without any run-time consing or extra work — the second evaluation just sees #<unspecified> — so this is as fast as a function call. The setf argument is a gensym (created when setf is defined) so that its name does not shadow any existing variable. Bacros expand in the calling environment, and a normal argument name might shadow something in that environment while the bacro is begin expanded.

See s7test.scm for many examples including such perennial favorites as when, loop, dotimes, do*, enum, pushnew, and defstruct.

As the setf example shows, a macro is a first-class citizen of s7. You can pass it as a function argument, apply it to a list, return it from a function, call it recursively, and assign it to a variable. You can even set its procedure-setter!

> (define-macro (hi a) `(+ ,a 1))
hi
> (apply hi '(4))
5
> (define (fmac mac) (apply mac '(4)))
fmac
> (fmac hi)
5
> (define (fmac mac) (mac 4))
fmac
> (fmac hi)
5
> (define (make-mac)
    (define-macro (hi a) `(+ ,a 1))
    hi)
make-mac
> (let ((x (make-mac)))
    (x 2))
3
> (define-macro (ref v i) `(vector-ref ,v ,i))
ref
> (define-macro (set v i x) `(vector-set! ,v ,i ,x))
set
> (set! (procedure-setter ref) set)
#<macro>
> (let ((v (vector 1 2 3))) (set! (ref v 0) 32) v)
#(32 2 3)

Backquote (quasiquote) in s7 is almost trivial. Constants are unchanged, symbols are quoted, ",arg" becomes "arg", and ",@arg" becomes "(apply values arg)" — hooray for real multiple values! It's almost as easy to write the actual macro body as the backquoted version of it.

> (define-macro (hi a) `(+ 1 ,a))
hi
> (procedure-source hi)
(lambda ({defmac}-16) (apply (lambda (a) ({list} '+ 1 a)) (cdr {defmac}-16)))

;; so (define-macro (hi a) ({list} + 1 a)) is the same

> (define-macro (hi a) `(+ 1 ,@a))
hi
> (procedure-source hi)
(lambda ({defmac}-17) (apply (lambda (a) ({list} '+ 1 ({apply} {values} a))) (cdr {defmac}-17)))

;; same: (define-macro (hi a) ({list} + 1 ({apply} {values} a)))

> (define-macro (hi a) ``(+ 1 ,,a))
hi
> (procedure-source hi)
(lambda ({defmac}-18) (apply (lambda (a) ({list} '{list} ({list} 'quote '+) 1 a)) (cdr {defmac}-18)))

and so on. "{list}" is a special version of "list" to avoid name collisions and handle a few tricky details (similarly for "{values}" and "{apply}"). There is no unquote-splicing macro in s7; ",@(...)" becomes "(unquote ({apply} {values} ...))" at read-time. There shouldn't be any unquote either. In Scheme the reader turns ,x into (unquote x), so:

> (let (,'a) unquote)
a
> (let (, (lambda (x) (+ x 1))) ,,,,'3)
7

comma becomes a sort of symbol macro! I think I'll remove unquote; ,x and ,@x will still work as expected, but there will not be any "unquote" or "unquote-splicing" in the resultant source code. Just to make life difficult:

> (let (' 1) quote)
1

but that translation is so ingrained in lisp (it even works in Clisp) that I'm reluctant to change it. The two unquote names, on the other hand, seem unnecessary.

macro? returns #t if its argument is a macro or a symbol whose value is a macro. We can use it, and other macro-related stuff to make a version of macroexpand-all:

(define-macro (fully-expand form)
  (define (expand form)
    (if (pair? form)
	(if (macro? (car form))
	    (expand ((eval (procedure-source (car form))) form))
	    (cons (expand (car form))
		  (expand (cdr form))))
	form))
  (expand form))

> (define-macro (hi a) `(+ 1 ,a))
hi
> (define-macro (ha c) `(hi (+ ,c 1)))
ha
> (fully-expand (define (ho b) (+ 1 (ha b))))
ho
> (procedure-source ho)
(lambda (b) (+ 1 (+ 1 (+ b 1))))

fully-expand expands each macro it encounters by using the procedure-source of that macro, that is, the function that the macro definition expanded into:

(define-macro (hi a) `(+ ,a 1))

> (procedure-source hi)
(lambda ({defmac}-18) (apply (lambda (a) ({list} '+ a 1)) (cdr {defmac}-18)))

s7 macros are not hygienic. For example,

> (define-macro (mac b) 
    `(let ((a 12)) 
       (+ a ,b)))
mac
> (let ((a 1) (+ *)) (mac a))
144

This returns 144 because '+' has turned into '*', and 'a' is the internal 'a', not the argument 'a'. We get (* 12 12) where we might have expected (+ 12 1). Starting with the '+' problem, as long as the redefinition of '+' is local (that is, it happens after the macro definition), we can unquote the +:

> (define-macro (mac b) 
    `(let ((a 12)) 
       (,+ a ,b))) ; ,+ picks up the definition-time +
mac
> (let ((a 1) (+ *)) (mac a))
24                 ; (+ a a) where a is 12

But the unquote trick won't work if we have previously loaded some file that redefined '+' at the top-level (so at macro definition time, + is *, but we want the built-in +). Although this example is silly, the problem is real in Scheme because Scheme has no reserved words and only one name space.

> (define + *)
+
> (define (add a b) (+ a b))
add
> (add 2 3)
6
> (define (divide a b) (/ a b))
divide
> (divide 2 3)
2/3
> (set! / -) ; a bad idea — this turns off s7's optimizer
-
> (divide 2 3)
-1

Obviously macros are not the problem here. Since we might be loading code written by others, it's sometimes hard to tell what names that code depends on or redefines. We need a way to get the pristine (start-up, built-in) value of '+'. One long-winded way in s7 uses initial-environment:

> (define + *)
+
> (define (add a b) (with-environment (initial-environment) (+ a b)))
add
> (add 2 3)
5

But this is hard to read, and it's not inconceivable that we might want all three values of a symbol, the start-up value, the definition-time value, and the current value. The latter can be accessed with the bare symbol, the definition-time value with unquote (','), and the start-up value with either initial-environment or #_<name>. That is, #_+ is a reader macro for (with-environment (initial-environment) +).

> (define-macro (mac b) 
    `(#_let ((a 12)) 
       (#_+ a ,b))) ; #_+ and #_let are start-up values
mac
> (let ((a 1) (+ *)) (mac a))
24                 ; (+ a a) where a is 12 and + is the start-up +

;;; make + generic (there's a similar C-based example below)
> (define (+ . args) 
    (if (null? args) 0 
        (apply (if (number? (car args)) #_+ #_string-append) args)))
+
> (+ 1 2)
3
> (+ "hi" "ho")
"hiho"

#_<name> could be implemented via *#readers*:

(set! *#readers*
  (cons (cons #\_ (lambda (str)
		    (with-environment 
                      (initial-environment) 
                      (string->symbol (substring str 1)))))
	*#readers*))

So, now we have only the variable capture problem ('a' has been captured in the preceding examples). This is the only thing that the gigantic "hygienic macro" systems actually deal with: a microscopic problem that you'd think, from the hype, was up there with malaria and the national debt. gensym is the standard approach:

> (define-macro (mac b) 
    (let ((var (gensym))) 
      `(#_let ((,var 12))
         (#_+ ,var ,b))))
mac
> (let ((a 1) (+ *)) (mac a))
13

If gensym worries you because it's creating a symbol on each macro invocation, use a closure as in the bacro example above. gensym can make the macro code unreadable in more complex cases. It is possible to write a macro to do the gensymification for us, but in s7 it is easier to use the environment functions. The procedure-environment of a macro is the environment at its definition time, just like a function. By wrapping the macro body in with-environment, we make sure anything in that body reflects the definition environment, not the calling environment. We then augment that environment with the macro arguments getting their values from the call-time environment:

(define-macro (define-immaculo name-and-args . body)
  (let* ((gensyms (map (lambda (g) (gensym)) (cdr name-and-args)))
	 (args (cdr (copy name-and-args)))
	 (name (car name-and-args))
	 (set-args (map (lambda (a g) `(list ',g ,a)) args gensyms))
	 (get-args (map (lambda (a g) `(quote (cons ',a ,g))) args gensyms))
	 (blocked-args (map (lambda (a) `(,a ',a)) args))     
	 (new-body (list (apply let blocked-args body))))    ; unquote args
    `(define-macro ,name-and-args
       `(let ,(list ,@set-args)
          ,(list 'with-environment 
                 (append (list 'augment-environment) 
                         (list (list 'procedure-environment ,name)) 
                         (list ,@get-args))
                 ',@new-body)))))

> (define-immaculo (mac c d) `(let ((a 12) (b 3)) (+ a b ,c ,d)))
mac
> (let ((a 21) (b 10) (+ *)) (mac a b))
46     ; (+ 12 3 21 10) with the definition-time value of +
> (let ((a 21) (b 10) (+ *)) (mac a (+ b 10)))
136    ; (+ 12 3 21 (* 10 10)), (+ b 10) is treated as (* b 10)
       ;   because (+ b 10) refers to the '+' and 'b' in the calling environment

Here's a simpler version that does not try to import the definition environment. Better would be to use a gensym rather than "e" for the current-environment.

(define-macro (define-immaculo name-and-args . body)
  (let ((new-args (map (lambda (arg) 
                         `(list (quote ,arg) ,arg)) 
                       (cdr name-and-args)))
	(wrapped-args (map (lambda (arg) 
                             `(,arg (quote (with-environment e ,arg)))) 
                           (cdr name-and-args))))
    (let ((new-body (apply let wrapped-args body)))
      `(define-macro ,name-and-args
	 `(let ,(list ,@new-args)
	    (let ((e (current-environment)))
	      ,',new-body))))))

> (define-immaculo (mac b) 
    `(let ((a 1)) (+ a ,b)))
mac
> (let ((a 32)) (mac a))
33
> (procedure-source mac) ; I've edited the results to try to make the transformation clearer
(define-macro (mac b)
  `(let ((b ,b))
     (let ((e (current-environment)))	      
       (let ((a 1))     ; here is the original macro body
         (+ a (with-environment e b))))))

Apply let is very similar to eval:

> (apply let '((a 2) (b 3)) '((+ a b)))
5
> (eval '(+ a b) (environment '(a . 2) '(b . 3)))
5
> ((apply lambda '(a b) '((+ a b))) 2 3)
5

But we're using it here as a kind of "replace":

> (apply let '((a 2) (b 3)) '((list + a b))) ; a -> 2, b -> 3
(+ 2 3)

The redundant-looking double lists are for apply's benefit. We could use a trailing null instead (mimicking apply* in some ancient lisps):

> (apply let '((a 2) (b 3)) '(list + a b) ())
(+ 2 3)

There is another problem with macros: accidental loops. Take the following example; we're trying to write a macro that defines a function that returns its argument in a list statement.

> (define-macro (hang arg) `(define ,arg `(list (cdr ,,arg)))) ; obvious, no?
hang
> (macroexpand (hang (f a)))
(define #1=(f a) ({list} 'list ({list} 'cdr #1#)))
> (hang (f a))
f
> (procedure-source f)
(lambda #1=(a) ({list} 'list ({list} 'cdr (f . #1#))))
>(f 1)

And now we are hung — we've created a procedure with an infinite loop! This is surprisingly easy to do by accident. Here's one way out:

> (define-macro (hang arg) `(define ,arg `(list ,,@(cdr arg))))
hang
> (macroexpand (hang (f a)))
(define (f a) ({list} 'list a))
> (hang (f a))
f
> (f 1)
(list 1)

But we should end on a happy note. Here is Peter Seibel's wonderful once-only macro:

(defmacro once-only (names . body)
  (let ((gensyms (map (lambda (n) (gensym)) names)))
    `(let (,@(map (lambda (g) (list g '(gensym))) gensyms))
       `(let (,,@(map (lambda (g n) (list list g n)) gensyms names))
          ,(let (,@(map (lambda (n g) (list n g)) names gensyms))
             ,@body)))))

From the land of sparkling bacros:

(define once-only
  (let ((names (gensym))
	(body (gensym)))
    (symbol->value (apply define-bacro `((,(gensym) ,names . ,body)
      `(let (,@(map (lambda (name) `(,name ,(eval name))) ,names))
	 ,@,body))))))

procedure-setter

(procedure-setter proc)
(make-procedure-with-setter proc setter)

Each function (or macro!) can have an associated setter, much like defsetf in CL. As a convenience, there's also a way to associate the two functions under one name: make-procedure-with-setter. The setter is called when the procedure is the target of set! Its last argument is the value passed to set!:

> (procedure-setter cadr)
#f
> (set! (procedure-setter cadr) 
        (lambda (lst val) 
          (set! (car (cdr lst)) val)))
#<closure>
> (procedure-source (procedure-setter cadr))
(lambda (lst val) (set! (car (cdr lst)) val))
> (let ((lst (list 1 2 3))) 
    (set! (cadr lst) 4) 
    lst)
(1 4 3)

In some cases, the setter needs to be a macro:

> (set! (procedure-setter logbit?)
        (symbol->value 
          (define-macro (m var index on) ; here we want to set "var", so we need a macro
	    `(if ,on
	         (set! ,var (logior ,var (ash 1 ,index)))
	         (set! ,var (logand ,var (lognot (ash 1 ,index))))))))
#<macro>
> (define (mingle a b)
    (let ((r 0))
      (do ((i 0 (+ i 1)))
          ((= i 31) r)
        (set! (logbit? r (* 2 i)) (logbit? a i))
        (set! (logbit? r (+ (* 2 i) 1)) (logbit? b i)))))
mingle
> (mingle 6 3) ; the INTERCAL mingle operator?
30

Here is a pretty example of make-procedure-with-setter:

(define-macro (c?r path)
  ;; "path" is a list and "X" marks the spot in it that we are trying to access
  ;; (a (b ((c X)))) — anything after the X is ignored, other symbols are just placeholders
  ;; c?r returns a procedure-with-setter that gets/sets X

  (define (X-marks-the-spot accessor tree)
    (if (pair? tree)
	(or (X-marks-the-spot (cons 'car accessor) (car tree))
	    (X-marks-the-spot (cons 'cdr accessor) (cdr tree)))
	(if (eq? tree 'X) accessor #f)))

  (let ((body 'lst))
    (for-each
     (lambda (f)
       (set! body (list f body)))
     (reverse (X-marks-the-spot () path)))

    `(make-procedure-with-setter
      (lambda (lst) 
	,body)
      (lambda (lst val)
	(set! ,body val)))))

> ((c?r (a b (X))) '(1 2 (3 4) 5))
3

> (let ((lst (list 1 2 (list 3 4) 5))) 
   (set! ((c?r (a b (X))) lst) 32)
   lst)
(1 2 (32 4) 5)

> (procedure-source (c?r (a b (X))))
(lambda (lst) (car (car (cdr (cdr lst)))))

> ((c?r (a b . X)) '(1 2 (3 4) 5))
((3 4) 5)

> (let ((lst (list 1 2 (list 3 4) 5))) 
   (set! ((c?r (a b . X)) lst) '(32))
   lst)
(1 2 32)

> (procedure-source (c?r (a b . X)))
(lambda (lst) (cdr (cdr lst)))

> ((c?r (((((a (b (c (d (e X)))))))))) '(((((1 (2 (3 (4 (5 6)))))))))) 
6

> (let ((lst '(((((1 (2 (3 (4 (5 6))))))))))) 
    (set! ((c?r (((((a (b (c (d (e X)))))))))) lst) 32) 
    lst)
(((((1 (2 (3 (4 (5 32)))))))))

> (procedure-source (c?r (((((a (b (c (d (e X)))))))))))
(lambda (lst) (car (cdr (car (cdr (car (cdr (car (cdr (car (cdr (car (car (car (car lst)))))))))))))))

We can extend c?r into something incredibly useful! A goto implementation using circular lists:

(define-macro (define-with-goto name-and-args . body)
  ;; run through the body collecting label accessors, (label name)
  ;; run through getting goto positions, (goto name)
  ;; tie all the goto's to their respective labels (via set-cdr! essentially)
  
  (define (find-accessor type)
    (let ((labels ()))
      (define (gather-labels accessor tree)
	(if (pair? tree)
	    (if (equal? (car tree) type)
		(begin
		  (set! labels (cons (cons (cadr tree) 
					   (let ((body 'lst))
					     (for-each
					      (lambda (f)
						(set! body (list f body)))
					      (reverse (cdr accessor)))
					     (make-procedure-with-setter
					      (apply lambda '(lst) (list body))
					      (apply lambda '(lst val) `((set! ,body val))))))
				     labels))
		  (gather-labels (cons 'cdr accessor) (cdr tree)))
		(begin
		  (gather-labels (cons 'car accessor) (car tree))
		  (gather-labels (cons 'cdr accessor) (cdr tree))))))
      (gather-labels () body)
      labels))
	
  (let ((labels (find-accessor 'label))
	(gotos (find-accessor 'goto)))
    (if (not (null? gotos))
	(for-each
	 (lambda (goto)
	   (let* ((name (car goto))
		  (goto-accessor (cdr goto))
		  (label (assoc name labels))
		  (label-accessor (and label (cdr label))))
	     (if label-accessor
		 (set! (goto-accessor body) (label-accessor body))
		 (error 'bad-goto "can't find label: ~S" name))))
	   gotos))

    `(define ,name-and-args
       (let ((label (lambda (name) #f))
	     (goto (lambda (name) #f)))
	 ,@body))))

(define-with-goto (hi)
  (display "start ")
  (goto 'the-end)
  (display "oops")
  (label 'the-end)
  (display "all done"))

I wonder if it would be more consistent to use the name "procedure/setter" in place of Guile's "make-procedure-with-setter". Its syntax is closer to vector than make-vector, for example. Even better: "dilambda":

(define-macro (dilambda getter setter)
  `(make-procedure-with-setter
     (lambda ,@getter)
     (lambda ,@setter)))

(let ((a 32)) 
  (dilambda (() a) 
            ((b) (set! a b))))

"bilambda" would mix Latin and Greek since the Romans used "el", not "lambda", according to Wikipedia.

applicable objects, generalized set!, generic functions

A procedure with a setter can be viewed as one generalization of set!. Another treats objects as having predefined get and set functions. In s7 lists, strings, vectors, hash-tables, environments, and any cooperating C or Scheme-defined objects are both applicable and settable. newLisp calls this implicit indexing, Gauche implements it via object-apply, Guile via procedure-with-setter; CL's funcallable instance might be the same idea.

In (vector-ref #(1 2) 0), for example, vector-ref is just a type declaration. But in Scheme, type declarations are unnecessary, so we get exactly the same result from (#(1 2) 0). Similarly, (lst 1) is the same as (list-ref lst 1), and (set! (lst 1) 2) is the same as (list-set! lst 1 2). I like this syntax: the less noise, the better!

Well, maybe applicable strings look weird: ("hi" 1) is #\i, but worse, so is (cond (1 => "hi"))! Even though a string, list, or vector is "applicable", it is not currently considered to be a procedure, so (procedure? "hi") is #f. map and for-each, however, accept anything that apply can handle, so (map #(0 1) '(1 0)) is '(1 0). (On the first call to map in this case, you get the result of (#(0 1) 1) and so on). string->list and vector->list are (map values object). Their inverses are (and always have been) equally trivial.

The applicable object syntax makes it easy to write generic functions. For example, s7test.scm has implementations of Common Lisp's sequence functions. length, copy, reverse, fill!, map and for-each are generic in this sense (map always returns a list).

> (map (lambda (a b) (- a b)) (list 1 2) (vector 3 4))
(5 -3 9)
> (length "hi")
2

Here is a function that returns an iterator. It works for anything that accepts an integer argument (e.g. list, string, vector, ...):

(define (make-iterator obj)
  (let ((ctr 0))
    (lambda ()
      (and (< ctr (length obj))
	   (let ((val (obj ctr)))
	     (set! ctr (+ ctr 1))
	     val)))))

Here's a generic FFT:

(define* (cfft! data n (dir 1)) ; (complex data)
  (if (not n) (set! n (length data)))
  (do ((i 0 (+ i 1))
       (j 0))
      ((= i n))
    (if (> j i)
	(let ((temp (data j)))
	  (set! (data j) (data i))
	  (set! (data i) temp)))
    (let ((m (/ n 2)))
      (do () 
	  ((or (< m 2) (< j m)))
	(set! j (- j m))
	(set! m (/ m 2)))
      (set! j (+ j m))))
  (let ((ipow (floor (log n 2)))
	(prev 1))
    (do ((lg 0 (+ lg 1))
	 (mmax 2 (* mmax 2))
	 (pow (/ n 2) (/ pow 2))
	 (theta (make-rectangular 0.0 (* pi dir)) (* theta 0.5)))
	((= lg ipow))
      (let ((wpc (exp theta))
	    (wc 1.0))
	(do ((ii 0 (+ ii 1)))
	    ((= ii prev))
	  (do ((jj 0 (+ jj 1))
	       (i ii (+ i mmax))
	       (j (+ ii prev) (+ j mmax)))
	      ((>= jj pow))
	    (let ((tc (* wc (data j))))
	      (set! (data j) (- (data i) tc))
	      (set! (data i) (+ (data i) tc))))
	  (set! wc (* wc wpc)))
	(set! prev mmax))))
  data)

> (cfft! (list 0.0 1+i 0.0 0.0))
(1+1i -1+1i -1-1i 1-1i)
> (cfft! (vector 0.0 1+i 0.0 0.0))
#(1+1i -1+1i -1-1i 1-1i)

There is one place where list-set! and friends are not the same as set!: the former evaluate their first argument, but set! does not (with a quibble; see below):

> (let ((str "hi")) (string-set! (let () str) 1 #\a) str)
"ha"

> (let ((str "hi")) (set! (let () str) 1 #\a) str)
;((let () str) 1 #\a): too many arguments to set!

> (let ((str "hi")) (set! ((let () str) 1) #\a) str)
"ha"

> (let ((str "hi")) (set! (str 1) #\a) str)
"ha"

set! looks at its first argument to decide what to set. If it's a symbol, no problem. If it's a list, set! looks at its car to see if it is some object that has a setter. If the car is itself a list, set! evaluates the internal expression, and tries again. So the second case above is the only one that won't work. And of course:

> (let ((x (list 1 2))) 
    (set! ((((lambda () (list x))) 0) 0) 3) 
    x) 
(3 2)

By my count, around 20 of the Scheme built-in functions are already generic in the sense that they accept arguments of many types (leaving aside the numeric and type checking functions, take for example equal?, display, member, assoc, apply, eval, quasiquote, and values). s7 extends that list with map, for-each, reverse, and length, and adds a few others such as copy, fill!, sort!, object->string, and continuation?. newLisp takes a more radical approach than s7: it extends operators such as '>' to compare strings and lists, as well as numbers. In map and for-each, however, you can mix the argument types, so I'm not as attracted to making '>' generic; you can't, for example, (> "hi" 32.1), or even (> 1 0+i).

The make-iterator function above raises an interesting issue. What should the copy function do if passed a closure? To be consistent with copy as applied to a hash-table-iterator, we want a snapshot of the current state of the iteration:

(define (iterator->string iterator)
  (let ((look-ahead (copy iterator)))
    (string-append "#<iterator ... " 
		   (object->string (look-ahead)) " "
		   (object->string (look-ahead)) " "
		   (object->string (look-ahead))
		   " ...>")))

(let ((v (vector 0 1 2 3 4 5 6 7 8 9)))
  (let ((iterator (make-iterator v)))
    (let* ((vals (list (iterator) (iterator) (iterator)))
           (description (iterator->string iterator))
           (more-vals (list (iterator) (iterator))))
      (list vals description more-vals))))

which returns '((0 1 2) "#<iterator ... 3 4 5 ...>" (3 4)) But this consistency requires that we copy the function's environment chain (all its non-global "closed over" state), and each value in that chain:

(define (make-iterator-with-filter obj func)
  (let ((iterator (make-iterator obj)))
    (letrec ((filter (lambda ()
		       (or (func (iterator))
			   (filter)))))
      filter)))

(let ((v (vector 0 1 2 3 4 5 6 7 8 9)))
  (let ((iterator (make-iterator-with-filter v 
		    (lambda (val)
		      (if (number? val)
			  (and (even? val)
			       val)
			  :all-done)))))
    ...))

Now if we (copy iterator) in the let body above, but don't copy the internal iterator in the copy function, the two iterators step on each other because they share the same 'ctr' variable. If the copy function copies any closure it finds in the environment (so that the inner iterator is copied), we get an infinite loop because 'filter' is in its own closure (thanks to letrec). And if it copies all values in the environment, it copies the vector that is deliberately shared! Our motto is "hack now, think later", so we try to add circle checks and copy only closure environments, but now we have a big pile of queasy code that assumes we have thought of everything. Not likely. And copy has become incredibly wasteful; we might end up copying hundreds of local variables, when all we actually want is access to the 'ctr' and 'obj' variables in the current iterator's environment. I'm getting ahead of the story, but suffice it to say that environments are first class citizens of s7. We can think of the make-iterator function as defining an implicit class that contains the slots 'ctr' and 'obj', and that returns an instance of that class. So...

(define-macro (with-iterator obj . body) 
  `(with-environment (procedure-environment ,obj) ,@body))

;; just for fun:
(define (reset-iterator iter)
  (with-iterator iter (set! ctr 0)))	

(define (iterator-obj iter)
  (with-iterator iter obj))

(define iterator-ctr 
  (make-procedure-with-setter 
   (lambda (iter)
     (with-iterator iter ctr))
   (lambda (iter new-ctr)
     (with-environment 
      (augment-environment (procedure-environment iter) (cons 'new-ctr new-ctr))
      (set! ctr new-ctr)))))

(define (iterator-copy iter)
  (let ((new-iter (make-iterator (iterator-obj iter))))
    (set! (iterator-ctr new-iter) (iterator-ctr iter))
    new-iter))

After all that, the copy function simply returns its argument if passed a closure.

I don't know why this pleases me so much, but you can use make-iterator to define a function whose argument takes a different default value each time it is called:

> (let ((iter (make-iterator #(10 9 8 7 6 5))))
     (define* (func (val (iter))) 
       val)
     (list (func) (func 32) (func) (func)))
> (10 32 8 7)

multidimensional vectors

s7 supports vectors with any number of dimensions. It is here, in particular, that generalized set! shines. make-vector's second argument can be a list of dimensions, rather than an integer as in the one dimensional case:

(make-vector (list 2 3 4))
(make-vector '(2 3) 1.0)
(vector-dimensions (make-vector (list 2 3 4))) -> (2 3 4)

The second example includes the optional initial element. (vect i ...) or (vector-ref vect i ...) return the given element, and (set! (vect i ...) value) and (vector-set! vect i ... value) set it. vector-length (or just length) returns the total number of elements. vector-dimensions returns a list of the dimensions.

> (define v (make-vector '(2 3) 1.0))
#2D((1.0 1.0 1.0) (1.0 1.0 1.0))
> (set! (v 0 1) 2.0)
#2D((1.0 2.0 1.0) (1.0 1.0 1.0))
> (v 0 1)
2.0
> (vector-length v)
6

matrix multiplication:

(define (matrix-multiply A B)
  ;; assume square matrices and so on for simplicity
  (let* ((size (car (vector-dimensions A)))
	 (C (make-vector (list size size) 0)))
    (do ((i 0 (+ i 1)))
	((= i size) C)
      (do ((j 0 (+ j 1)))
	  ((= j size))
	(let ((sum 0))
	  (do ((k 0 (+ k 1)))
	      ((= k size))
	    (set! sum (+ sum (* (A i k) (B k j)))))
	  (set! (C i j) sum))))))

Conway's game of Life:

(define* (life (width 40) (height 40))
  (let ((state0 (make-vector (list width height) 0))
	(state1 (make-vector (list width height) 0)))

    ;; initialize with some random pattern
    (do ((x 0 (+ x 1)))
	((= x width))
      (do ((y 0 (+ y 1)))
	  ((= y height))
	(set! (state0 x y) (if (< (random 100) 15) 1 0))))

    (do () ()
      ;; show current state (using terminal escape sequences, borrowed from the Rosetta C code)
      (format *stderr* "~C[H" #\escape)           ; ESC H = tab set
      (do ((y 0 (+ y 1)))
	  ((= y height))
	(do ((x 0 (+ x 1)))
	    ((= x width))
	  (if (zero? (state0 x y))
	      (format *stderr* "  ")              ; ESC 07m below = inverse
	      (format *stderr* "~C[07m  ~C[m" #\escape #\escape)))
	(format *stderr* "~C[E" #\escape))        ; ESC E = next line

      ;; get the next state
      (do ((x 1 (+ x 1)))
	  ((= x (- width 1)))
	(do ((y 1 (+ y 1)))
	    ((= y (- height 1)))
	  (let ((n (+ (state0 (- x 1) (- y 1))
		      (state0    x    (- y 1))
		      (state0 (+ x 1) (- y 1))
		      (state0 (- x 1)    y)      
		      (state0 (+ x 1)    y)      
		      (state0 (- x 1) (+ y 1))
		      (state0    x    (+ y 1))
		      (state0 (+ x 1) (+ y 1)))))
	    (set! (state1 x y) 
		  (if (or (= n 3) 
			  (and (= n 2)
			       (not (zero? (state0 x y)))))
		      1 0)))))
      (do ((x 0 (+ x 1)))
	  ((= x width))
	(do ((y 0 (+ y 1)))
	    ((= y height))
	  (set! (state0 x y) (state1 x y)))))))

Multidimensional vector constant syntax is modelled after CL: #nd(...) or #nD(...) signals that the lists specify the elements of an 'n' dimensional vector: #2D((1 2 3) (4 5 6))

> (vector-ref #2D((1 2 3) (4 5 6)) 1 2)
6
> (matrix-multiply #2d((-1 0) (0 -1)) #2d((2 0) (-2 2)))
#2D((-2 0) (2 -2))

If any dimension has 0 length, you get an n-dimensional empty vector. It is not equal to a 1-dimensional empty vector.

> (make-vector '(10 0 3))
#3D()
> (equal? #() #3D())
#f

To save on costly parentheses, and make it easier to write generic multidimensional sequence functions, you can use this same syntax with lists.

> (let ((L '((1 2 3) (4 5 6))))
    (L 1 0))              ; same as (list-ref (list-ref L 1) 0) or ((L 1) 0)
4

> (let ((L '(((1 2 3) (4 5 6)) ((7 8 9) (10 11 12))))) 
    (set! (L 1 0 2) 32)   ; same as (list-set! (list-ref (list-ref L 1) 0) 2 32) which is unreadable!
    L)
(((1 2 3) (4 5 6)) ((7 8 32) (10 11 12)))

Or with vectors of vectors, of course:

> (let ((V '#(#(1 2 3) #(4 5 6)))) 
    (V 1 2))              ; same as (vector-ref (vector-ref V 1) 2) or ((V 1) 2)
6

> (let ((V #2d((1 2 3) (4 5 6))))
    (V 0))
#(1 2 3)

There's one difference between a vector-of-vectors and a multidimensional vector: in the latter case, you can't clobber one of the inner vectors.

> (let ((V '#(#(1 2 3) #(4 5 6)))) (set! (V 1) 32) V)
#(#(1 2 3) 32)

> (let ((V #2d((1 2 3) (4 5 6)))) (set! (V 1) 32) V)
;not enough args for vector-set!: (#2D((1 2 3) (4 5 6)) 1 32)

Using lists to display the inner vectors may not be optimal, especially when the elements are also lists:

#2D(((0) (0) ((0))) ((0) 0 ((0))))

The "#()" notation is no better (the elements can be vectors), and I'm not a fan of "[]" parentheses. Perhaps we could use different colors? Or different size parentheses?

#2D(((0) (0) ((0))) ((0) 0 ((0))))
#2D(((0) (0) ((0))) ((0) 0 ((0))))

I'm not sure how to handle vector->list and list->vector in the multidimensional case. Currently, vector->list flattens the vector, and list->vector always returns a one dimensional vector, so the two are not inverses.

> (vector->list #2d((1 2) (3 4)))
(1 2 3 4)             ; should this be '((1 2) (3 4)) or '(#(1 2) #(3 4))?
> (list->vector '(#(1 2) #(3 4))) ; what about '((1 2) (3 4))?
#(#(1 2) #(3 4))      

Perhaps I should add an optional number-of-dimensions argument? This also affects format and sort!:

> (format #f "~{~A~^ ~}" #2d((1 2) (3 4)))
"1 2 3 4"

> (sort! #2d((1 4) (3 2)) >) 
#2D((4 3) (2 1))

Another question: should we accept the multi-index syntax in a case such as:

(let ((v #("abc" "def"))) 
  (v 0 2))

My first thought was that the indices should all refer to the same type of object, so s7 would complain in a mixed case like that. If we can nest any applicable objects and apply the whole thing to an arbitrary list of indices, ambiguities arise:

((lambda (x) x) "hi" 0) 
((lambda (x) (lambda (y) (+ x y))) 1 2)

I think these should complain that the function got too many arguments, but from the implicit indexing point of view, they could be interpreted as:

(string-ref ((lambda (x) x) "hi") 0) ; i.e. (((lambda (x) x) "hi") 0)
(((lambda (x) (lambda (y) (+ x y))) 1) 2)

Add optional and rest arguments, and you can't tell who is supposed to take which arguments. Currently, you can mix types with implicit indices, but a function grabs all remaining indices. Trickier than I expected!

hash-tables

Any s7 object can be the key or the key's value. Each hash-table keeps track of the keys it contains, optimizing the search wherever possible. If you pass a table size that is not a power of 2, make-hash-table rounds it up to the next power of 2.

> (let ((ht (make-hash-table)))
    (set! (ht "hi") 123)
    (ht "hi"))
123

hash-table (the function) parallels (the functions) vector, list, and string. Its arguments are cons's containing key/value pairs. The result is a new hash-table with those values preinstalled: (hash-table '("hi" . 32) '("ho" . 1)). make-hash-table-iterator returns a function of no arguments (a "thunk"). Each time you call this function, it returns the next entry in the hash table. When it runs out of entries, it returns nil. hash-table-iterator? returns #t if passed one of these iterators.

Since hash-tables accept the same applicable-object syntax that vectors use, we can treat a hash-table as, for example, a sparse array:

> (define make-sparse-array make-hash-table)
make-sparse-array

> (let ((arr (make-sparse-array)))
   (set! (arr 1032) "1032")
   (set! (arr -23) "-23")
   (list (arr 1032) (arr -23)))
("1032" "-23")

map and for-each accept hash-table arguments. On each iteration, the map or for-each function is passed an entry, '(key . value), in whatever order the entries are encountered in the table. (Both use make-hash-table-iterator internally).

(define (hash-table->alist table)
  (map values table))

(define (merge-hash-tables . tables)
  (apply hash-table 
    (apply append 
      (map hash-table->alist tables))))

reverse of a hash-table returns a new table with the keys and values reversed. (fill! table ()) removes all entries from the hash-table.

Two hash-tables are equal if they have the same keys with the same values. This is independent of the table sizes, or the order in which the key/value pairs were added.

If the hash key is a float (a non-rational number), hash-table-ref uses morally-equal?. Otherwise, for example, you could use NaN as a key, but then never be able to access it!

(define-macro (define-memoized name&arg . body)
  (let ((arg (cadr name&arg))
	(memo (gensym "memo")))
    `(define ,(car name&arg)
       (let ((,memo (make-hash-table)))
	 (lambda (,arg)
	   (or (,memo ,arg)                             ; check for saved value
	       (set! (,memo ,arg) (begin ,@body)))))))) ; set! returns the new value

> (define-memoized (f1 abc) (+ abc 2))
> (f1 3)
5
> (environment->list (outer-environment (procedure-environment f1)))
(({memo}-16 . #<hash-table (3 . 5)>))

environments

An environment holds symbols and their values. The global environment, for example, holds all the variables that are defined at the top level. Environments are first class (and applicable) objects in s7.

(environment? obj)            #t if obj is an environment
(with-environment env . body) evaluate body in the environment env 
(outer-environment env)       the environment that encloses the environment env
(environment->list env)       return the environment bindings as a list of (symbol . value) cons's

(global-environment)          the top-level environment
(current-environment)         the currently visible variables and their values
(procedure-environment proc)  the environment at the time when proc was defined
(error-environment)           the environment at the point of the last error
(initial-environment)         the current environment, but all built-in functions have their original values

(environment . bindings)      make a new environment with the given bindings
(augment-environment env . bindings-or-envs)   add bindings to env
(augment-environment! env . bindings-or-env)

(open-environment env)        mark env as open (see below)
(open-environment? env)
(object-environment env)

(symbol->value sym (env (current-environment)))
(defined? sym (env (current-environment)))

(with-environment env . body) evaluates its body in the environment env. There are several examples of its use in the macro section. It is very close to let: (let env . body), but named let usurps that possibility, and eval: (eval form env). defined? returns #t if the symbol is defined in the environment, and symbol->value returns the value associated with it. To add or change a symbol's value, use augment-environment:

> (let ((a 1)) 
    (eval '(+ a b) 
           (augment-environment
             (current-environment) 
             (cons 'b 32)))) ; add 'b with the value 32 to this environment
33

augment-environment does not change the environment passed to it. It just prepends the new bindings, shadowing any old ones, as if you had called "let". To add the bindings directly to the environment, use augment-environment!. Both of these functions accept nil as the 'env' argument as shorthand for (global-environment). Both also accept other environments as well as individual bindings, adding all the argument's bindings to the new environment.

Augment-environment is a bad name, but I'm at a loss for a better one. If we view an environment as a disembodied let, akin to viewing a generic function as a disembodied lambda, these names come to mind:

environment?          let?, islet
global-environment    rootlet
initial-environment   unlet
with-environment      inlet 
outer-environment     outlet
augment-environment   sublet, make-let(?)
augment-environment!  varlet, let-append(?)
current-environment   valet,  make-begin(?)
error-environment     owlet, triplet, error-let(?)
procedure-environment runlet(?)
environment->list  let->list(?)
open-environment      open-let(?)
open-environment?
object-environment

It is possible in Scheme to redefine built-in functions such as car. To ensure that some code sees the original built-in function definitions, wrap it in (with-environment (initial-environment) ...):

> (let ((caar 123)) 
    (+ caar (with-environment (initial-environment) 
              (caar '((2) 3)))))
125

with-environment and initial-environment are constants, so you can use them in any context without worrying about whether they've been redefined. As mentioned in the macro section, #_<name> is a built-in reader macro for (with-environment (initial-environment) <name>), so for example, #_+ is the built-in + function, no matter what. (initial-environment) cleans up the current environment whenever it's called, so you can use it to revert the REPL.

I think these functions can implement the notions of libraries, separate namespaces, or modules. Here's one way: first the library writer just writes his library. The normal user simply loads it. The abnormal user worries about everything, so first he loads the library in a local let to make sure no bindings escape to pollute his code, and then he uses initial-environment to make sure that none of his bindings pollute the library code:

(let ()
  (with-environment (initial-environment)
    (load "any-library.scm" (current-environment)) 
    ;; by default load puts stuff in the global environment
    ...))

Now Abnormal User can do what he wants with the library entities. Say he wants to use "lognor" under the name "bitwise-not-or", and all the other functions are of no interest:

(begin
  (define bitwise-not-or #f)
  (let ()
    (with-environment (initial-environment)
      (load "any-library.scm" (current-environment))
      (set! bitwise-not-or lognor)))) ; lognor is presumably defined in "any-library.scm"

Or equivalently:

(augment-environment! (current-environment)
  (cons 'bitwise-not-or 
    (with-environment (initial-environment)
      (load "any-library.scm" (current-environment))
      lognor)))

Say he wants to make sure the library is cleanly loaded, but all its top-level bindings are imported into the current environment:

(apply augment-environment! (current-environment)
  (with-environment (initial-environment)
    (let ()
      (load "any-library.scm" (current-environment))
      (environment->list (current-environment)))))
      ;; these are the bindings introduced by loading the library

To do the same thing, but prepend "library:" to each name:

(apply augment-environment! (current-environment)
  (with-environment (initial-environment)
    (let ()
      (load "any-library.scm" (current-environment))
      (map (lambda (binding)
	     (cons (string->symbol 
		    (string-append "library:" (symbol->string (car binding))))
		   (cdr binding)))
	   (environment->list (current-environment))))))

That's all there is to it!

Well, almost, darn it. If the loaded library file sets (via set!) a global value such as abs, we need to put it back to its original form:

(define (safe-load file)
  (let ((e (with-environment          ; save the environment before loading
	       (initial-environment) 
	     (environment->list (current-environment)))))
    (load file (current-environment))
    (let ((new-e (with-environment    ; get the environment after loading
		     (initial-environment) 
		   (environment->list (current-environment)))))
      (for-each                       ; see if any built-in functions were stepped on
       (lambda (sym)
	 (if (not (assoc (car sym) e))
	     (begin
	       (format #t "~S clobbered ~A~%" file (car sym))
	       (apply set! (car sym) (list (cdr (assoc (car sym) new-e)))))))
       new-e))))

;; say libtest.scm has the line (set! abs odd?)

> (safe-load "libtest.scm")
"libtest.scm" clobbered abs
> (abs -2)
2

open-environment marks its argument, either an environment, a closure, or a c-object as open. I need better terminology here! An open object is one that the built-in s7 functions handle specially. If they encounter one in their argument list, they look in the object for their own name, and call that function if it exists. A bare-bones example:

> (abs (open-environment (environment (cons 'abs (lambda (x) 47)))))
47

In CLOS, we'd have to declare a class and a method, then call make-instance, and then we'd discover that it wouldn't work anyway. Here we have, in effect, an anonymous instance of an anonymous class.

Here is yet another object system. Each class and each instance is an environment. A class might be thought of as a template for instances, but there's actually no difference between them. To make an instance of a class, copy it. To inherit from another class, concatenate the environments. To access (get or set) a field or a method, use the implicit indexing syntax with the field or method name. To evaluate a form in the context of an instance (CL's with-slots), use with-environment. To run through all the fields, use map or for-each.

(define-bacro* (define-class class-name inherited-classes (slots ()) (methods ()))
  ;; a bacro is needed so that the calling environment is accessible via outer-environment
  ;;   we could also use the begin/let shuffle, but it's too embarrassing
  `(let ((outer-env (outer-environment (current-environment)))
	 (new-methods ())
	 (new-slots ()))

    (for-each
     (lambda (class)
       ;; each class is a set of nested environments, the innermost (first in the list)
       ;;   holds the local slots which are copied each time an instance is created,
       ;;   the next holds the class slots (global to all instances, not copied);
       ;;   these hold the class name and other such info.  The remaining environments
       ;;   hold the methods, with the localmost method first.  So in this loop, we
       ;;   are gathering the local slots and all the methods of the inherited
       ;;   classes, and will splice them together below as a new class.

       (set! new-slots (append (environment->list class) new-slots))
       (do ((e (outer-environment (outer-environment class)) (outer-environment e)))
	   ((or (not (environment? e))
		(eq? e (global-environment))))
	 (set! new-methods (append (environment->list e) new-methods))))
     ,inherited-classes)

     (let ((remove-duplicates 
	    (lambda (lst)         ; if multiple local slots with same name, take the localmost
	      (letrec ((rem-dup
			(lambda (lst nlst)
			  (cond ((null? lst) nlst)
				((assq (caar lst) nlst) (rem-dup (cdr lst) nlst))
				(else (rem-dup (cdr lst) (cons (car lst) nlst)))))))
		(reverse (rem-dup lst ()))))))
       (set! new-slots 
	     (remove-duplicates
	      (append (map (lambda (slot)
			     (if (pair? slot)
				 (cons (car slot) (cadr slot))
				 (cons slot #f)))
			   ,slots)                    ; the incoming new slots, #f is the default value
		      new-slots))))                   ; the inherited slots

    (set! new-methods 
	  (append (map (lambda (method)
			 (if (pair? method)
			     (cons (car method) (cadr method))
			     (cons method #f)))
		       ,methods)                     ; the incoming new methods

		  ;; add an object->string method for this class (this is already a generic function).
		  (list (cons 'object->string (lambda* (obj (use-write #t))
				       (format #f "#<~A: ~{~A~^ ~}>" 
					       ',class-name
					       (map (lambda (slot)
						      (list (car slot) (cdr slot)))
						    (environment->list obj))))))
		  (reverse! new-methods)))           ; the inherited methods, shadowed automatically

    (let ((new-class (open-environment
                       (apply augment-environment           ; the local slots
		         (augment-environment               ; the global slots
		           (apply environment               ; the methods
			     (reverse new-methods))
		           (cons 'class-name ',class-name)  ; class-name slot
			   (cons 'inherited ,inherited-classes)
			   (cons 'inheritors ()))           ; classes that inherit from this class
		         new-slots))))

      (augment-environment! outer-env                  
        (cons ',class-name new-class)                       ; define the class as class-name in the calling environment

	;; define class-name? type check
	(cons (string->symbol (string-append (symbol->string ',class-name) "?"))
	      (lambda (obj)
		(and (environment? obj)
		     (eq? (obj 'class-name) ',class-name)))))

      (augment-environment! outer-env
        ;; define the make-instance function for this class.  
        ;;   Each slot is a keyword argument to the make function.
        (cons (string->symbol (string-append "make-" (symbol->string ',class-name)))
	      (apply lambda* (map (lambda (slot)
				    (if (pair? slot)
					(list (car slot) (cdr slot))
					(list slot #f)))
				  new-slots)
		     `((let ((new-obj (copy ,,class-name)))
			 ,@(map (lambda (slot)
				  `(set! (new-obj ',(car slot)) ,(car slot)))
				new-slots)
			 new-obj)))))

      ;; save inheritance info for this class for subsequent define-method
      (letrec ((add-inheritor (lambda (class)
				(for-each add-inheritor (class 'inherited))
				(if (not (memq new-class (class 'inheritors)))
				    (set! (class 'inheritors) (cons new-class (class 'inheritors)))))))
	(for-each add-inheritor ,inherited-classes))
    
      ',class-name)))


(define-macro (define-generic name)
  `(define ,name (lambda args (apply ((car args) ',name) args)))) ; (define (genfun any) ((any 'genfun) any))


(define-macro (define-slot-accessor name slot)
  `(define ,name (make-procedure-with-setter 
                   (lambda (obj) (obj ',slot)) 
		   (lambda (obj val) (set! (obj ',slot) val)))))


(define-bacro (define-method name-and-args . body)
  `(let* ((outer-env (outer-environment (current-environment)))
	  (method-name (car ',name-and-args))
	  (method-args (cdr ',name-and-args))
	  (object (caar method-args))
	  (class (symbol->value (cadar method-args)))
	  (old-method (class method-name))
	  (method (apply lambda* method-args ',body)))

     ;; define the method as a normal-looking function
     ;;   s7test.scm has define-method-with-next-method that implements call-next-method here
     ;;   it also has make-instance 
     (augment-environment! outer-env
       (cons method-name 
	     (apply lambda* method-args 
		    `(((,object ',method-name)
		       ,@(map (lambda (arg)
				(if (pair? arg) (car arg) arg))
			      method-args))))))
     
     ;; add the method to the class
     (augment-environment! (outer-environment (outer-environment class))
       (cons method-name method))

     ;; if there are inheritors, add it to them as well, but not if they have a shadowing version
     (for-each
      (lambda (inheritor) 
	(if (not (eq? (inheritor method-name) #<undefined>)) ; defined? goes to the global env
	    (if (eq? (inheritor method-name) old-method)
		(set! (inheritor method-name) method))
	    (augment-environment! (outer-environment (outer-environment inheritor))
   	      (cons method-name method))))
      (class 'inheritors))

     method-name))


(define (all-methods obj method)
  ;; for arbitrary method combinations: this returns a list of all the methods of a given name
  ;;   in obj's class and the classes it inherits from (see example below)
  (let* ((base-method (obj method))
	 (methods (if (procedure? base-method) (list base-method) ())))
    (for-each 
     (lambda (ancestor)
       (let ((next-method (ancestor method)))
	 (if (and (procedure? next-method)
		  (not (memq next-method methods)))
	     (set! methods (cons next-method methods)))))
     (obj 'inherited))
    (reverse methods)))


> (define-class class-1 () 
       '((a 1) (b 2)) 
       (list (list 'add (lambda (obj) 
                          (with-environment obj
                            (+ a b))))))
class-1
> (define v (make-class-1 :a 32))
v 
> (v 'a)                         ; to set the 'a slot, (set! (v 'a) 0) 
32
> (object->string v)             ; built-in functions are all generic
"#<class-1: (a 32) (b 2)>"       ;   so this uses the method defined in the class definition
> ((v 'add) v)
34
> (define-generic add)
add
> (add v)                        ; syntactic sugar for ((v 'add) v)
34
> (define-slot-accessor slot-a a) ; more sugar!
slot-a
> (slot-a v)                     ; same as (v 'a), set via (set! (slot-a v) 123)
32
> (map car v)                    ; map and for-each work with environments
(a b)                               ;   map cdr would return '(32 2) in this case
> (define-class class-2 (list class-1)
       '((c 3)) 
       (list (list 'multiply (lambda (obj) 
                               (with-environment obj 
                                 (* a b c))))))
class-2
> (define v2 (make-class-2 :a 32))
v2
> v2                            ; will use object->string
"#<class-2: (c 3) (a 32) (b 2)>"
> ((v2 'multiply) v2)
192
> (add v2)                      ; inherited from class-1
34
> (define-method (subtract (obj class-1)) (with-environment obj (- a b)))
subtract
> (subtract v2)  ; class-2 inherits from class-1 so it knows about subtract
30
> (define v1 (make-class-1))
v1
> (augment-environment! v1      ; change the add method just in this instance
       (cons 'add (lambda (obj) 
                    (with-environment obj
                      (+ 1 a (* 2 b))))))
#<environment>
> (add v1)
6
> (add v)                       ; other class-1 instances are not affected
34
> (define-class class-3 (list class-2) () 
    (list (list 'multiply (lambda (obj num) 
			    (* num 
			       ((class-2 'multiply) obj) ; method combination 
			       (add obj))))))
class-3
> ((class-3 'multiply) class-3 10)
180                               ; (* 10 (* 1 2 3) (+ 1 2))
> (all-methods v3 'multiply)
(#<closure> #<closure>)
> (for-each (lambda (p) (format #t "~A~%" (procedure-source p))) (all-methods v3 'multiply))
(lambda (obj num) (* num ((class-2 'multiply) obj) (add obj)))
(lambda (obj) (with-environment obj (* a b c)))


;; another obvious possibility: generic macros

with-environment (used briefly above) provides an even more striking simplification of syntax than implicit indexing or multidimensional vectors, and it is faster as well! See Snd's piano.scm for an extended example.

As shown in the preceding example, implicit indexing of a local environment does not search the global environment. Since initial-environment extends the current environment chain, it is considered a local environment:

> ((global-environment) 'abs)
abs
> (let () ((current-environment) 'abs))
#<undefined>
> ((initial-environment) 'abs) ; what a bad function name
#<undefined>
(define-bacro (value->symbol val)
   ;; (outer-environment (current-environment)) is the calling environment
  `(call-with-exit
    (lambda (return)
      (do ((e (current-environment) (outer-environment e))) ()
	(for-each 
	 (lambda (slot)
	   (if (equal? ,val (cdr slot))
	       (return (car slot))))
	 e)
	(if (eq? e (global-environment))
	    (return #f))))))

> (let ((a 1) (b "hi")) 
    (value->symbol "hi"))
b

Scheme needs a way to define functions that share a closure. Here's an idle thought:

(define f1 (let ((x 23))
	     (lambda (a)
	       (+ x a))))
(define (f2 b)
  (+ b (* 2 ((procedure-environment f1) 'x)))) ; import f1's x into f2

> (+ (f1 1) (f2 1))
71

That has now blossomed into...

open-environment replaces make-type. s7test.scm has a nearly backwards-compatible implementation of make-type. open-environment is still under development, but as an example, here is an implementation of float-vectors:

(begin
  (define float-vector? #f)
  (define make-float-vector #f)
  
  (let ((type (gensym))
	(->float (lambda (x)
		   (if (real? x)
		       (* x 1.0)
		       (error 'wrong-type-arg "float-vector new value is not a real: ~A" x)))))

    (augment-environment! (current-environment)
      (cons 'length (lambda (p) ((procedure-environment p) 'len)))
      (cons 'object->string (lambda (p) "#<float-vector>")))

    (set! make-float-vector
	  (lambda* (len (init 0.0)) 
	    (let ((obj (make-vector len (->float init))))
	      (open-environment
	        (make-procedure-with-setter
		  (lambda (i) (#_vector-ref obj i))
		  (lambda (i val) (#_vector-set! obj i (->float val))))))))
    
    (set! float-vector?
	  (lambda (p)
	    (and (procedure? p)
		 (eq? ((procedure-environment p) 'type) type))))))
  
> (define fv (make-float-vector 32))
fv
> fv
#<float-vector>
> (length fv)
32
> (set! (fv 0) 123)
123.0
> (fv 0)
123.0

If we add the following lines to the augment-environment! section above:

(cons 'vector? (lambda (p) #t))
(cons 'vector-length (lambda (p) ((procedure-environment p) 'len)))
(cons 'vector-dimensions (lambda (p) (list ((procedure-environment p) 'len))))
(cons 'vector-ref (lambda (p ind) (#_vector-ref ((procedure-environment p) 'obj) ind)))
(cons 'vector-set! (lambda (p ind val) (#_vector-set! ((procedure-environment p) 'obj) ind (->float val))))
(cons 'vector-fill! (lambda (p val) (#_vector-fill! ((procedure-environment p) 'obj) (->float val))))
(cons 'fill! (lambda (p val) (#_vector-fill! ((procedure-environment p) 'obj) (->float val))))
(cons 'vector->list (lambda (p) (#_vector->list ((procedure-environment p) 'obj))))
(cons 'reverse (lambda (p) (#_reverse ((procedure-environment p) 'obj))))

then our float-vectors are fully legitimate Scheme vectors. It is irrelevant to Scheme that they happen to be procedures; a vector is any value that behaves like a vector. Extend this to math ops, and we can treat float-vectors as matrices, so (* v1 v2) is matrix multiplication. Or add quaternions, gap-buffers as strings — I'd say "the possibilities are endless", but actually I can't think of any others right at the moment.

I can't resist adding at least some of the quaternion implementation. In the following code we add quaternions to some of the math functions (finally "number?" is not the same as "complex?"!), and include support for possible additional numeric types that might be added later. As in the earlier case, a number is whatever behaves like a number — we aren't bound by any previous classes, and we don't have to define separate methods for every possibility. The '+' function for instance, can take any mixture of numbers:

(define-class quaternion () 
  '((r 0) (i 0) (j 0) (k 0))
  (list (list 'number? (lambda (obj) #t))
	(list 'complex? (lambda (obj) #f))
	(list 'real? (lambda (obj) #f))
	(list 'integer? (lambda (obj) #f))
	(list 'rational? (lambda (obj) #f))

	(list '+ (lambda orig-args
		   (let add ((r ()) (i ()) (j ()) (k ()) (args orig-args))
		     (if (null? args)
			 (make-quaternion 
                           (apply + r) (apply + i) (apply + j) (apply + k)) 
			 (let ((n (car args)))
			   (cond
			    ((real? n)
			     (add (cons n r) i j k (cdr args)))
			    ((complex? n)
			     (add (cons (real-part n) r) (cons (imag-part n) i) j k (cdr args)))
			    ((quaternion? n)
			       (add (cons (n 'r) r) (cons (n 'i) i) 
                                    (cons (n 'j) j) (cons (n 'k) k) 
                                    (cdr args)))
			    ((open-environment? n) ; maybe we'll add octonions later!
			     (if (eq? n (car orig-args))
                                 ;; if we are not the first argument, and we don't know how to handle the
                                 ;; first argument, then we have to raise an error.  Otherwise, the two
                                 ;; methods will get into an infinite loop saying very politely
                                 ;;   "please, you go first".
				 (error 'missing-method "+ can't handle these arguments: ~A" args)
				 (apply (n '+) 
                                        (make-quaternion 
                                          (apply + r) (apply + i) (apply + j) (apply + k)) 
                                        (cdr args))))
			    (else (error 'wrong-type-arg "+ argument ~A is not a number" n))))))))
	))

> (let ((q1 (make-quaternion 1.0 1.0 0.0 0.0)))
    (+ 1 q1 2.5+i))
#<quaternion: (r 4.5) (i 2.0) (j 0.0) (k 0.0)>

multiple-values

In s7, multiple values are spliced directly into the caller's argument list.

> (+ (values 1 2 3) 4)
10
> (string-ref ((lambda () (values "abcd" 2))))
#\c
> ((lambda (a b) (+ a b)) ((lambda () (values 1 2))))
3
> (+ (call/cc (lambda (ret) (ret 1 2 3))) 4) ; call/cc has an implicit "values"
10
> ((lambda* ((a 1) (b 2)) (list a b)) (values :a 3))
(3 2)

(define-macro (call-with-values producer consumer) 
  `(,consumer (,producer)))

(define-macro (multiple-value-bind vars expr . body)
  `((lambda ,vars ,@body) ,expr))

(define (curry function . args)
  (if (null? args)
      function
      (lambda more-args
        (if (null? more-args)
            (apply function args)
            (function (apply values args) (apply values more-args))))))

There aren't that many real uses for multiple-values in Scheme. Nearly all can be replaced by a normal list. There are a couple of cases where multiple-values are handy. First, you can use "values" to return any number of values, including 0, from map's function application:

> (map (lambda (x) (if (odd? x) (values x (* x 20)) (values))) (list 1 2 3))
(1 20 3 60)

> (map values (list 1 2 3) (list 4 5 6))
(1 4 2 5 3 6)

(define (remove-if func lst) 
  (map (lambda (x) (if (func x) (values) x)) lst))

(define (pick-mappings func lst)          
  (map (lambda (x) (or (func x) (values))) lst))

(define (shuffle . args) 
  (apply map values args))

> (shuffle '(1 2 3) #(4 5 6) '(7 8 9))
(1 4 7 2 5 8 3 6 9)

Second, you can use multiple-values to turn off the short-circuit evaluation of 'or' and 'and'.

> (let ((x 1)) (and (values #f (let () (set! x 3) #f))) x)
3

But 'apply' has the same effect and is easier to read:

(define (and? . args) (apply and args))
(define (every? f . seqs) (apply and (apply map f seqs)))

More often you want to keep the short-circuiting, but add some action as 'and' or 'or' marches through its arguments:

(define-macro (every? function . args)
  `(and ,@(map (lambda (arg) `(,function ,arg)) args)))

(define-macro (and-let* vars . body)
  `(let () ;; bind vars, if any is #f stop, else evaluate body with those bindings
     (and ,@(map (lambda (var) `(begin (apply define ',var) ,(car var))) vars) 
          (begin ,@body))))

At the top-level, since there's nothing to splice into, you simply get your values back:

> (values 1 (list 1 2) (+ 3 4 5))
(values 1 (1 2) 12)

But this printout is just trying to be informative. There is no multiple-values object in s7. You can't (set! x (values 1 2)), for example. The values function tells s7 that its arguments should be handled in a special way, and the multiple-value indication goes away as soon as the arguments are spliced into some caller's arguments.

Internally, s7 uses (apply values ...) to implement unquote splicing (",@") in quasiquote.

In some Schemes, values behaves like CL's prog1:

(not s7)> (let ((x 1)) (cond ((values #f (set! x 2) #t) 3) (#t x)))
2
(not s7)> (if (values #f #t) 1 2)
2

But in s7 we're trying to implement real multiple values (else why have them at all?). There are many ways we could interpret (cond ((values ...))...) and (cond ((values...) => func)), but surely equivalent uses of "cond" and "if" should give the same result. Currently in s7, where a test is in progress, only (values #f) is the same as #f.

> (if (values #f #f) 1 2)            ; (values #f #f) is not #f
1
> (cond ((values #f #f) 1) (#t 2))
1
;;; but if we interpreted this as splicing in the values, we get an inconsistency:
> (cond (#f #f 1) (#t 2))
2

> (if (values #f) 1 2)
2
> (cond ((values #f) 1) (#t 2))
2

> (if (values) 1 2)
1
> (cond ((values) 1) (#t 2))
1
;;; this is consistent with (cond (1) (#t 2))

So "if" and "cond" agree, but it requires that in one case the "values" behavior is slightly weird. (or (values #f #f)) is #f, but that isn't inconsistent because "or" is not testing anything. We might choose to say that (if (values #f #f)...) is an error, but that would be hasty — our troubles have only begun. First, "cond" can omit the expressions that follow the test, unlike "if":

> (cond (3))
3

and even trickier, "cond" can pass the test value to a function:

> (cond (3 => +))
3

The various standards agree that in the "=>" case, the "fed to" function receives one argument, so

(not s7)> (cond ((values 1 2) => +))
1

If we were following the "splice immediately" model, this would be (cond (1 2 => +)) which is an error in some Schemes. So something has to give. My druthers is to make "values" work as consistently as possible, and hope that the one odd corner will not trip anyone. From that point of view, the "one arg" standard looks like a wasted opportunity. s7 handles it this way:

> (+ 1 (cond ((values 2 3))) 4)   ; trailing values are not ignored
10
> (cond ((values 1 2 3) => +))
6

Of course, it is unproblematic that the expression can itself involve multiple values:

> (+ (cond (#t (values 1 2))))
3

Now, what have I missed?

Since set! does not evaluate its first argument, and there is no setter for "values", (set! (values x) ...) is not the same as (set! x ...). (string-set! (values string) ...) works because string-set! does evaluate its first argument. ((values + 1 2) (values 3 4) 5) is 15, as anyone would expect.

I wrote all that and said "nobody needs multiple values!". In desperation,

(define (flatten lst)
  (define (flatten-1 lst)
    (cond ((null? lst) (values))
	  ((not (pair? lst)) lst)
	  (#t (values (flatten-1 (car lst))
		      (flatten-1 (cdr lst))))))
  (map values (list (flatten-1 lst))))
(define (boring-flatten x)
  (cond ((null? x) ())
        ((not (pair? x)) (list x))
        (#t (append (boring-flatten (car x))
		    (boring-flatten (cdr x))))))

Here's a much more entertaining version:

(define (flatten! lst) ; in-place flatten
  (let loop ((L lst))
    (if (pair? (car L))
	(let ((end (cdr L))
	      (p (car L)))
	  (set! (car L) (car p))
	  (set! (cdr L) (cdr p))
	  (set! (cdr (list-tail L (- (length p) 1))) end)
	  (loop L))
	(if (not (null? (cdr L)))
	    (if (null? (car L))
		(begin
		  (set! (car L) (cadr L))
		  (set! (cdr L) (cddr L))
		  (loop L))
		(loop (cdr L)))))
    (if (equal? lst '(()))
	()
	(let ((len (length lst)))
	  (if (null? (car (list-tail lst (- len 1))))
	      (set! (cdr (list-tail lst (- len 2))) ()))
	  lst))))

call-with-exit, with-baffle and continuation?

call-with-exit is call/cc without the ability to jump back into the original context, similar to "return" in C. This is cleaner than call/cc, and much faster.

(define-macro (block . body) 
  ;; borrowed loosely from CL — predefine "return" as an escape
  `(call-with-exit (lambda (return) ,@body)))

(define-macro (while test . body)
  ;; while loop with predefined break and continue
  `(call-with-exit
    (lambda (break) 
      (letrec ((continue (lambda () 
			   (if (let () ,test)
			       (begin 
				 (let () ,@body)
				 (continue))
			       (break)))))
	(continue)))))

(define (and-for-each func . args)
  ;; apply func to the first member of each arg, stopping if it returns #f
  (call-with-exit
   (lambda (quit)
     (apply for-each (lambda arglist
		       (if (not (apply func arglist))
			   (quit #<unspecified>))) 
	    args))))

(define (find-if f . args)  ; generic position-if is very similar
  (call-with-exit
   (lambda (return) 
     (apply for-each (lambda main-args 
		       (if (apply f main-args) 
			   (apply return main-args)))
	    args))))

> (find-if even? #(1 3 5 2))
2
> (* (find-if > #(1 3 5 2) '(2 2 2 3)))
6

The call-with-exit function's argument (the "continuation") is only valid within the call-with-exit function. In call/cc, you can save it, then call it later to jump back, but if you try that with call-with-exit (from outside the call-with-exit function's body), you'll get an error. This is similar to trying to read from a closed input port.

The other side, so to speak, of call-with-exit, is with-baffle. Both limit the scope of a continuation. Sometimes we need a normal call/cc, but want to make sure it is active only within a given block of code. Normally, if a continuation gets away, there's no telling when it might wreak havoc on us. Scheme code with call/cc becomes unpredictable, undebuggable, and completely unmaintainable. with-baffle blocks all that — no continuation can jump into its body:

(let ((what's-for-breakfast ())
      (bad-dog 'fido))        ; bad-dog wonders what's for breakfast?
  (with-baffle                ; the syntax is (with-baffle . body)         
   (set! what's-for-breakfast
	 (call/cc
	  (lambda (biscuit?)
	    (set! bad-dog biscuit?) ; bad-dog smells a biscuit!
	    (biscuit? 'biscuit!)))))
  (if (eq? what's-for-breakfast 'biscuit!) 
      (bad-dog 'biscuit!))     ; now, outside the baffled block, bad-dog wants that biscuit!
  what's-for-breakfast)        ;   but s7 says "No!": baffled! ("continuation can't jump into with-baffle")

continuation? returns #t if its argument is a continuation, as opposed to a normal procedure. I don't know why Scheme hasn't had this function from the very beginning, but it's needed if you want to write a continuable error handler. Here is a sketch of the situation:

(let ()
  (catch #t
	 (lambda ()
	   (let ((res (call/cc 
                        (lambda (ok) 
			  (error 'cerror "an error" ok)))))
	     (display res) (newline)))
	 (lambda args
	   (if (and (eq? (car args) 'cerror)
		    (continuation? (cadadr args)))
	       (begin
		 (display "continuing...")
		 ((cadadr args) 2)))
	   (display "oops"))))

continuing...2

In a more general case, the error handler is separate from the catch body, and needs a way to distinguish a real continuation from a simple procedure. Otherwise, it blithely announces that it is continuing from the point of the error, but then fails to do so.

format, object->string

s7's built-in format function is very close to that in srfi-48.

> (format #f "~A ~D ~F" 'hi 123 3.14)
"hi 123 3.140000"

The format directives (tilde chars) are:

~%        insert newline
~&        insert newline if preceding char was not newline
~~        insert tilde
~\n       (tilde followed by newline): trim white space
~{        begin iteration (take arguments from a list, string, vector, or any other applicable object)
~}        end iteration
~^        jump out of iteration
~*        ignore the current argument
~C        print as character
~P        insert 's' if current argument is not 1 or 1.0 (use ~@P for "ies" or "y")
~A        object->string as in display
~S        object->string as in write
~B        number->string in base 2
~O        number->string in base 8
~D        number->string in base 10
~X        number->string in base 16
~E        float to string, (format #f "~E" 100.1) -> "1.001000e+02", (%e in C)
~F        float to string, (format #f "~F" 100.1) -> "100.100000",   (%f in C)
~G        float to string, (format #f "~G" 100.1) -> "100.1",        (%g in C)
~T        insert spaces (padding)

The last eight take the usual numeric arguments to specify field width and precision.

(format #f ...) simply returns the formatted string; (format #t ...) also sends it to *stdout*. To send the string to *stderr* instead, (format *stderr* ...).

Floats can occur in any base, so:

> #xf.c
15.75

This also affects format. In most Schemes, (format #f "~X" 1.25) is an error. In CL, it is equivalent to using ~A which is perverse. But

> (number->string 1.25 16)
"1.4"

and there's no obvious way to get the same effect from format unless we accept floats in the "~X" case. So in s7,

> (format #f "~X" 21)
"15"
> (format #f "~X" 1.25)
"1.4"
> (format #f "~X" 1.25+i)
"1.4+1.0i"
> (format #f "~X" 21/4)
"15/4"

That is, the output choice matches the argument. A case that came up in the Guile mailing lists is: (format #f "~F" 1/3). s7 currently returns "1/3", but Clisp returns "0.33333334".

The curly bracket directive applies to anything you can map over, not just lists:

> (format #f "~{~C~^ ~}" "hiho")
"h i h o"
> (format #f "~{~{~C~^ ~}~^...~}" (list "hiho" "test"))
"h i h o...t e s t"

object->string returns the string representation of its argument:

> (object->string "hiho")
"\"hiho\""

> (format #f "~S" "hiho")
"\"hiho\""

I added object->string to s7 before deciding to include format. format excites a vague disquiet — why do we need this ancient unlispy thing? We can almost replace it with:

(define (objects->string . objects)
  (apply string-append (map (lambda (obj) (object->string obj #f)) objects)))

But how to handle lists (~{...~} in format), or columnized output (~T)? I wonder whether formatted string output still matters outside a REPL. Even in that context, a modern GUI leaves formatting decisions to a text or table widget.

(define-macro (string->objects str . objs)
  `(with-input-from-string ,str
     (lambda ()
       ,@(map (lambda (obj)
		`(set! ,obj (eval (read))))
	      objs))))

hooks

(make-hook . fields)           ; make a new hook
(hook-functions hook)          ; the hook's list of 'body' functions

A hook is a function created by make-hook, and called (normally from C) when something interesting happens. In GUI toolkits hooks are called callback-lists, in CL conditions, in other contexts watchpoints or signals. s7 itself has several hooks: *error-hook*, *unbound-variable-hook*, and *load-hook*. make-hook is:

(define (make-hook . args)
  (let ((init ())
	(body ())
	(end ()))
    (apply lambda* args
      '(let ((result #<unspecified>))
         (let ((e (current-environment)))
           (dynamic-wind 
	     (lambda () (for-each (lambda (f) (f e)) init))
	     (lambda () (for-each (lambda (f) (f e)) body) result)
	     (lambda () (for-each (lambda (f) (f e)) end)))))
       ())))

Each of the functions held by the hook ('f' above), takes one argument, the environment within the hook, and its return value is ignored. The function can use (with-environment arg ...) or (arg symbol) to access the arguments to the hook ('args'). The hook environment's result field is the returned value of the hook application. Each such environment has three lists, init, body, and end, and these lists of functions are called in dynamic-wind. Here's a sketch:

(define (add-to-hook hook function)
  (set! (hook-functions hook) (cons function (hook-functions hook))))

> (define h (make-hook '(a 32) 'b)) 
h
> (add-to-hook h (lambda (hook) 
                   (set! (hook 'result) 
                         (format #f "a: ~A, b: ~A" (hook 'a) (hook 'b)))))
(#<closure>)
> (h 1 2)
"a: 1, b: 2"
> (h)
"a: 32, b: #f"

To add an old-style hook function:

(add-to-hook hook (lambda (h) (set! (h 'result) (apply func (map h (h 'args))))))

In C, to make a hook:

hook = s7_eval_c_string("(make-hook '(a 32) 'b)");
s7_gc_protect(s7, hook);

And call it:

result = s7_call(s7, hook, s7_list(s7, 2, s7_make_integer(s7, 1), s7_make_integer(s7, 2)));
(define-macro (hook . body)
  `(let ((h (make-hook)))
     (set! (hook-functions h) (list (lambda (h) (set! (h 'result) (begin ,@body)))))
     h))

Hooks are split into three list of functions to reflect the roles that hook functions play. Some are "watchers" that simply want notification that some event has occurred (the 'init' list). Others want to deal with the event in some way, possibly returning a result, and possibly interacting with any other similar hook functions' results (the 'body' list). Still others are only interested in the decision reached by the main functions. We need the dynamic-wind to be sure that all the before and after functions are called no matter what happens during the handling of the main functions.

We can also use a hook as a normal procecdure, but now the function lists make it easy to extend an existing function. The next example adds to a numerical function the possibility of calling it with a string:

> (define h (make-hook '(x 0.0)))
h
> (set! (hook-functions h) ; these are the 'body' functions
        (list (lambda (hook) 
                (set! (hook 'result) (+ (hook 'x) 1.0)))))
(#<closure>)
> (h 2.5)
3.5
> (set! ((procedure-environment h) 'init) 
        (list (lambda (hook) ; if x is a string, convert it to a number
                (if (string? (hook 'x)) 
                    (set! (hook 'x) (string->number (hook 'x)))))))
(#<closure>)
> (h "2.5")
3.5

procedure info

(procedure-name proc)
(procedure-source proc)
(procedure-arity proc) ; obsolete
(procedure-documentation proc)
(procedure-setter proc)
(procedure-environment proc)

(arity obj)
(aritable? obj num-args)

procedure-source, procedure-arity, procedure-setter, procedure-documentation, and help provide a look into a Scheme function. procedure-documentation returns the documentation string associated with a procedure: the initial string in the function's body. procedure-arity returns a list describing the argument list of a function: '(required-args optional-args rest-arg?). procedure-setter returns or sets the set function associated with a procedure (set-car! with car, for example). procedure-source returns the procedure source (a list). procedure-environment returns a procedure's environment. procedure-name returns the procedure's name.

(define (procedure-arglist f) (cadr (procedure-source f)))

In s7 lots of things are applicable, so procedure-arity is too restricted, and in its main use (checking that a function can be applied to some arguments), it is very clumsy. So, s7 has two generic functions: arity and aritable?. arity takes any object and returns either #f if it is not applicable, or a cons containing the minimum and maximum number of arguments acceptable. If the maximum reported is a really big number, that means any number of arguments is ok. aritable? takes two arguments, an object and an integer, and returns #t if the object can be applied to that many arguments.

> (define* (add-2 a (b 32)) "add-2 adds its 2 args" (+ a b))
add-2

> (procedure-documentation add-2)
"add-2 adds its 2 args"

> (procedure-arity add-2)
(0 2 #f)

> (procedure-source add-2)
(lambda* (a (b 32)) "add-2 adds its 2 args" (+ a b))

We can use procedure-environment and __func__ to write a function that tells us where the source is for a function:

(define (where-is func)
  (let ((addr (symbol->value '__func__ (procedure-environment func))))
    (if (not (pair? addr))
	"not found"
	(format #f "~A is at line ~D of ~A" (car addr) (caddr addr) (cadr addr)))))

> (where-is enveloped-mix)
"enveloped-mix is at line 68 of extensions.scm"

procedure-source returns the actual function source — more fun than a barrel of monkeys! If you modify the procedure source directly, it is safest to redefine the procedure so that everything in s7 knows about the change. Here'a an example that adds trace and local variable info for debugging:

(define-bacro (procedure-annotate proc) ; use "bacro" so we can annotate local functions
  (let ((orig (procedure-source proc)))

    (define (proc-walk source)
      (if (pair? source)
	  (if (or (eq? (car source) 'let)     ; if let or let*, show local variables
		  (eq? (car source) 'let*))
	      (if (symbol? (cadr source))
		  ;; (let name vars . body) -> (let name vars print-vars . body)
		  (append 
		   (list (car source)
			 (cadr source)
			 (caddr source)
			 `(format #t "    (let ~A (~{~A~^ ~}) ...)~%" 
                                  ,(cadr source) (environment->list (current-environment))))
		   (cdddr source))
		  ;; (let(*) vars . body) -> (let vars print-vars . body)
		  (append 
		   (list (car source)
			 (cadr source)
			 `(format #t "    (~A (~{~A~^ ~}) ...)~%" 
                                  ,(car source) (environment->list (current-environment))))
		   (cddr source)))
	      (cons (proc-walk (car source))
		    (proc-walk (cdr source))))
	  source))

    (let* ((new-body (proc-walk orig))
	   (result (gensym))
	   (new-source 
	    `(lambda ,(cadr orig)
	       (let ((,result #<undefined>))
		 (dynamic-wind
		     (lambda ()       ; upon entry, show procedure name and args
		       (format #t "(~A~{ ~A~})~%" 
                               ',proc 
                               (environment->list 
                                 (outer-environment 
                                   (outer-environment (current-environment))))))
		     (lambda ()
		       (set! ,result (,new-body ,@(cadr orig)))
		       ,result)
		     (lambda ()       ; at exit, show result
		       (if (eq? ,result #<undefined>)
			   (format #t "  ~A returns early~%")
			   (format #t "  ~A returns ~A~%" ',proc ,result))))))))

      `(set! ,proc (eval ,new-source)))))
			 
> (define (hi a) (let ((b 12)) (+ b a)))
hi
> (procedure-annotate hi)
#<closure>
> (let ((x 32)) (+ 1 (hi x)))
45
;; printing: 
(hi (a . 32))
    (let ((b . 12)) ...)
  hi returns 44

But maybe something less invasive is better. Here's a version of let that prints its bindings (this is borrowed from "nuntius" at reddit lisp):

(define-macro (print-let bindings . body)
  (let ((temp-symbol (gensym)))
    `(let ,(map (lambda (var/val)
		  `(,(car var/val) 
		    (let ((,temp-symbol ,(cadr var/val))) 
		      (format #t ";~S: ~S -> ~S~%" 
			      ',(car var/val) 
			      ',(cadr var/val) 
			      ,temp-symbol)
		      ,temp-symbol)))
		bindings)
       ,@body)))

;; or simpler:

(define-macro (print-let bindings . body)
  `(let ,bindings 
     (format #t "~{;~A~%~}" (current-environment))
     ,@body))	      

Since define* accepts multiple rest arguments, perhaps procedure-arity should return that number, rather than a boolean. I haven't run into a case where it matters. Another odd case: (procedure-arity (lambda* (:allow-other-keys) #f)). How should we indicate this in procedure-arity?

Speaking of define*, an argument's default value can be a function, so we can store argument-specific documentation in the documentation string:

(define (arg-info f n) 
  ((procedure-source f) 1 n 1 0 2)) ; get the documentation string of f's n-th argument

(define* (add-1 (arg ((lambda () "arg should be an integer" 0)))) 
  (+ arg 1))          ; the default value of arg is 0

> (add-1)
1
> (arg-info add-1 0)
"arg should be an integer"

This is a lot of boilerplate for a simple notion, so I implemented parameter docstrings such as: (define* (add-1 (arg 0 "arg should be an integer")) (+ arg 1)), but then decided not to go that way after all.

(define (for-each-subset func args)
  ;; form each subset of args, apply func to the subsets that fit its arity
  (define (subset source dest len)
    (if (null? source)
        (if (aritable? func len)             ; does this subset fit?
	    (apply func dest))
	(begin
	  (subset (cdr source) (cons (car source) dest) (+ len 1))
	  (subset (cdr source) dest len))))
  (subset args () 0))

eval

eval evaluates its argument, a list representing a piece of code. It takes an optional second argument, the environment in which the evaluation should take place. eval-string is similar, but its argument is a string.

> (eval '(+ 1 2))
3
> (eval-string "(+ 1 2)")
3

The environment argument is mainly useful in debugging. A breakpoint can be set, for example, then any input is evaluated in the environment of the break. Say we have the following code in ex.scm:

(define-macro (break)
  `(let ((break-env (current-environment))
	 (prompt (format #f "~%~A > " (if (defined? '__func__) __func__ "break"))))
     (call-with-exit
      (lambda (return)
	(do () ()                       ; our debugger's own REPL
	  (display prompt)              ; show where we stopped
	  (let ((str (read-line)))      ; read a line of input, :go -> exit the debugger
	    ;; the nil argument to read-line makes sure that we read C's stdin.  In any normal
	    ;;    program, we'd get the string from a text widget.
	    (if (> (length str) 0)
		(catch #t               ; try to handle typing mistakes etc
		       (lambda ()
			 (let ((val (eval-string str break-env)))
			   (if (eq? val :go)
			       (return))
			   (write val)))
		       (lambda args
			 (format #t "error: ~A" args))))))))))

;; now some random code that has a breakpoint
(define (a-function b)
  (let ((x 32))
    (do ((i 0 (+ i 1)))
	((= i 10))
      (if (= i 3)
	  (break)))
    x))

(a-function 123)
(display "done!") (newline)

Start up a REPL, and:

> (load "ex.scm")
(a-function "ex.scm" 26) > x    ; here we're in the debugger
32
(a-function "ex.scm" 26) > (+ b i)
126
(a-function "ex.scm" 26) > :go
done!

IO and other OS functions

Besides files, ports can also represent strings and functions. The string port functions are:

(with-output-to-string thunk)         ; open a string port as current-output-port, call thunk, return string
(with-input-from-string string thunk) ; open string as current-input-port, call thunk
(call-with-output-string proc)        ; open a string port, apply proc to it, return string
(call-with-input-string string proc)  ; open string as current-input-port, apply proc to it
(open-output-string)                  ; open a string output port
(get-output-string port)              ; return output accumulated in the string output port
(open-input-string string)            ; open a string input port reading string
> (let ((result #f) 
        (p (open-output-string)))
    (format p "this ~A ~C test ~D" "is" #\a 3)
    (set! result (get-output-string p))
    (close-output-port p)
    result)
"this is a test 3"

Other functions:

The variable *vector-print-length* sets the upper limit on how many vector elements are printed by object->string and format. The end-of-file object is #<eof>. When running s7 behind a GUI, you often want input to come from and output to go to arbitrary widgets. The function ports provide a way to redirect IO. See below for an example.

The default IO ports are *stdin*, *stdout*, and *stderr*. *stderr* is useful if you want to make sure output is flushed out immediately. The default output port is *stdout* which buffers output until a newline is seen. In most REPLs, the input port is set up by the REPL, so you need to use *stdin* if you want to read from the terminal instead:

> (read-char)
#<eof>
> (read-char *stdin*)
a          ; here s7 waited for me to type "a" in the terminal
#\a        ; this is the REPL reporting what read-char returned

binary-io.scm in the Snd package has functions that read and write integers and floats in both endian choices in a variety of sizes. Besides read-byte and write-byte, it uses integer-decode-float, and the various bitwise operators.

If the compile time switch WITH_SYSTEM_EXTRAS is 1, several additional OS-related and file-related functions are built-in. This is work in progress; currently this switch adds:

(directory? str)         ; return #t if str is the name of a directory
(file-exists? str)       ; return #t if str names an existing file
(delete-file str)        ; try to delete the file, return 0 is successful, else -1
(getenv var)             ; return the value of an environment variable: (getenv "HOME")
(directory->list dir)    ; return contents of directory as a list of strings (if HAVE_DIRENT_H)
(system command)         ; execute command

error handling

(error tag . info)           ; signal an error of type tag with addition information
(catch tag body err)         ; if error of type tag signalled in body (a thunk), call err with tag and info
(throw tag . info)           ; jump to corresponding catch

s7's error handling mimics that of Guile. An error is signalled via the error function, and can be trapped and dealt with via catch.

> (catch 'wrong-number-of-args
    (lambda ()     ; code protected by the catch
      (abs 1 2))
    (lambda args   ; the error handler
      (apply format #t (cadr args))))
"abs: too many arguments: (1 2)"

> (catch 'division-by-zero
    (lambda () (/ 1.0 0.0))
    (lambda args (string->number "inf.0")))
inf.0

(define-macro (catch-all . body)
  `(catch #t (lambda () ,@body) (lambda args args)))

catch has 3 arguments: a tag indicating what error to catch (#t = anything), the code, a thunk, that the catch is protecting, and the function to call if a matching error occurs during the evaluation of the thunk. The error handler takes a rest argument which will hold whatever the error function chooses to pass it. The error function itself takes at least 2 arguments, the error type, a symbol, and the error message. There may also be other arguments describing the error. The default action, in the absence of any catch, is to treat the message as a format control string, apply format to it and the other arguments, and send that info to the current-error-port.

Normally when reading a file, we have to check for #<eof>, but we can let s7 do that:

(define (copy-file infile outfile)
  (call-with-input-file infile
    (lambda (in)
      (call-with-output-file outfile
	(lambda (out)
	  (catch 'wrong-type-arg   ; s7 raises this error if write-char gets #<eof>
	    (lambda () 
	      (do () ()            ; read/write until #<eof>
		(write-char (read-char in) out)))
	    (lambda err 
	      outfile)))))))

catch is not limited to error handling:

(define (map-with-exit func . args)
  ;; map, but if early exit taken, return the accumulated partial result
  ;;   func takes escape thunk, then args
  (let* ((result ())
	 (escape-tag (gensym))
	 (escape (lambda () (throw escape-tag))))
    (catch escape-tag
      (lambda ()
	(let ((len (apply max (map length args))))
	  (do ((ctr 0 (+ ctr 1)))
	      ((= ctr len) (reverse result))      ; return the full result if no throw
	    (let ((val (apply func escape (map (lambda (x) (x ctr)) args))))
	      (set! result (cons val result))))))
      (lambda args
	(reverse result))))) ; if we catch escape-tag, return the partial result

(define (truncate-if func lst)
  (map-with-exit (lambda (escape x) (if (func x) (escape) x)) lst))

> (truncate-if even? #(1 3 5 -1 4 6 7 8))
(1 3 5 -1)

But this is less useful than map (it can't map over a hash-table for example), and is mostly reimplementing built-in code. Perhaps s7 should have an extension of map (and more usefully, for-each) that is patterned after dynamic-wind: (dynamic-for-each init-func main-func end-func . args) where init-func is called with one argument, the length of the shortest sequence argument (for-each and map know this in advance); main-func takes n arguments where n matches the number of sequences passed; and end-func is called even if a jump out of main-func occurs (like dynamic-wind in this regard). In the dynamic-map case, the end-func takes one argument, the current, possibly partial, result list. dynamic-for-each then could easily (but maybe not efficiently) implement generic functions such as ->list, ->vector, and ->string (converting any sequence into a sequence of some other type). map-with-exit would be

(define (map-with-exit func . args) 
  (let ((result ()))
    (call-with-exit
      (lambda (quit)
        (apply dynamic-map #f ; no init-func in this case
               (lambda main-args 
                 (apply func quit main-args)) 
               (lambda (res) 
                 (set! result res))
               args)))
    result))

With all the lambda boilerplate, nested catches are hard to read:

(catch #t
  (lambda ()
    (catch 'division-by-zero
      (lambda ()
	(catch 'wrong-type-arg
	  (lambda () 
	    (abs -1))
	  (lambda args (format #t "got a bad arg~%") -1)))
      (lambda args 0)))
  (lambda args 123))

Perhaps we need a macro:

(define-macro (catch-case clauses . body)
  (let ((base `(lambda () ,@body)))
    (map (lambda (clause)
	   (let ((tag (car clause)))
	     (set! base `(lambda () 
			   (catch ',(if (eq? tag 'else) #t tag)
			     ,base 
			     ,@(cdr clause))))))
	 clauses)
    (caddr base)))

;;; the code above becomes:
(catch-case ((wrong-type-arg   (lambda args (format #t "got a bad arg~%") -1))
	     (division-by-zero (lambda args 0))
	     (else             (lambda args 123)))
  (abs -1))

This is similar to r7rs scheme's "guard", but I don't want a pointless thunk for the body of the catch. Along the same lines:

(define (catch-if test func err)
  (catch #t
    func
    (lambda args
      (if (test (car args))
	  (apply err args)
	  (apply throw args))))) ; if not caught, re-raise the error

(define (catch-member lst func err)
  (catch-if (lambda (tag) (member tag lst)) func err))

When an error is encountered, and when s7 is interrupted via begin-hook, (error-environment) returns an environment that contains additional info about that error:

To find a variable's value at the point of the error: ((error-environment) var). To list all the local bindings from the error outward:

(do ((e (outer-environment (error-environment)) (outer-environment e))) 
    ((eq? e (global-environment))) 
  (format #t "~{~A ~}~%" e))

To evaluate the error handler in the environment of the error:

(let ((x 1))
  (catch #t
	 (lambda ()
	   (let ((y 2))
	     (error 'oops)))
	 (lambda args
	   (with-environment
	    (augment-environment (error-environment)
	     (cons 'args args))    ; add the error handler args
	    (list args x y)))))    ; we have access to 'y'

The hook *error-hook* provides a way to specialize error reporting. Its arguments are named 'type and 'data.

(set! (hook-functions *error-hook*) 
      (list (lambda (hook) 
              (apply format *stderr* (hook 'data)) 
              (newline *stderr*))))

There is a break macro defined in Snd (snd-xen.c) which allows you to stop at some point, then evaluate arbitrary expressions in that context.

If s7 encounters an unbound variable, it calls *unbound-variable-hook* before signalling an error. The offending symbol is named 'variable in the hook environment. In Snd, *unbound-variable-hook* is used to implement autoloading (see snd-help.c). If none of the *unbound-variable-hook* functions sets 'result' to something other than #<undefined>, s7 calls the error handler.

There are a lot of thoroughly disreputable ways to take advantage of *unbound-variable-hook*. Here are two. In the first, we can't ever remember how to get the error environment information when we're in the REPL, so we tie that action to the question-mark key. Whenever '?' is typed in the listener, the error info is typed out. That first moment of weakness leads inexorably to the second, dalliance with symbol-macros!

> (set! (hook-functions *unbound-variable-hook*)
      (append (hook-functions *unbound-variable-hook*) ; keep existing funcs like autoload
	      (list 
	       (lambda (hook) 
                 (if (memq (hook 'result) '(#<unspecified> #<undefined>))
	             (let ((sym (hook 'variable)))
		       (if (eq? sym '?)                       ; here '?' is the thing typed
		           (do ((e (outer-environment (error-environment)) (outer-environment e))) 
			       ((eq? e (global-environment)) 
                                (set! (hook 'result) #f))
		             (status-report (format #f "~%~A" (environment->list e)) -1))
                             ;; status-report is a Snd function that (in this case)
		             ;;   sends the data to the Snd listener.	       
                             )))))))
(#<closure> #<closure>)
> (let ((a "hi") (b 2)) (+ a b)) ; hit an error on purpose
;+ argument 1, "hi", is a string but should be a number
;    (+ a b)
> ?
((a . "hi") (b . 2))   ; ? prints the local variable info at the point of the error

> (set! (hook-functions *unbound-variable-hook*)
    (list (lambda (hook)
            (if (eq? (hook 'variable) 'hiho)
                (set! (hook 'result) (sin (random 1.0))))))) ; hiho -> sins of some number
(<closure>)
> hiho
0.46727567824396
> hiho
0.64985453979392

The s7-built-in catch tags are 'wrong-type-arg, 'syntax-error, 'read-error, 'out-of-memory, 'wrong-number-of-args, 'format-error, 'out-of-range, 'division-by-zero, 'io-error, and 'bignum-error. lint.scm in the Snd tarball checks for infelicities in s7 code (I think the C equivalent is now called splint).

define-constant, constant?, symbol-access

define-constant defines a constant and constant? returns #t if its argument is a constant. A constant in s7 is really constant: it can't be set or rebound.

> (define-constant var 32)
var

> (set! var 1)
;set!: can't alter immutable object: var

> (let ((var 1)) var)
;can't bind or set an immutable object: var, line 1

This has the possibly surprising side effect that previous uses of the constant name become constants:

(define (func a) (let ((cvar (+ a 1))) cvar))
(define-constant cvar 23)
(func 1)
;can't bind or set an immutable object: cvar

So, obviously, choose unique names for your constants, or don't use define-constant. A function can also be a constant.

Constants are very similar to things such as keywords (no set, always return itself as its value), variable trace (informative function upon set or keeping a history of past values), typed variables (restricting a variable's values or doing automatic conversions upon set), and notification upon set (either in Scheme or in C; I wanted this many years ago in Snd). The notification function is especially useful if you have a Scheme variable and want to reflect any change in its value immediately in C (see below). All of these cases modify the path between a symbol and its value. s7 gives you a handle on that path via symbol-access. (symbol-access symbol) returns that symbol's accessors, and (set! (symbol-access symbol) accessor-list) changes them. The symbol-accessor value is a list of three or more items, primarily the set, and bind functions, the second and third entries in the list respectively. These two functions take two arguments, the symbol in question and the value that it is about to be set or bound to. The variable is set or bound to the value they return. We could replace define-constant, and add local constants with:

(define constant-access 
  (list #f
	(lambda (symbol new-value) 
	  (format #t "can't change constant ~A's value to ~A" symbol new-value)
          'error)
	(lambda (symbol new-value) 
	  (format #t "can't bind constant ~A to a new value, ~A" symbol new-value)
          'error)))

(define-macro (define-constant symbol value)
  `(begin
     (define ,symbol ,value)
     (set! (symbol-access ',symbol) constant-access)
     ',symbol))

(define-macro (let-constant vars . body)
  (let ((varlist (map car vars)))
    `(let ,vars
       ,@(map (lambda (var)
		`(set! (symbol-access ',var) constant-access))
	      varlist)
       ,@body)))

In the next example, we restrict the values a variable can take to integers:

(define-macro (define-integer var value)
  `(begin
     (define ,var ,value)
     (set! (symbol-access ',var) 
	   (list #f
		 (lambda (symbol new-value)
		   (if (real? new-value)
		       (floor new-value) ; or min/max to restrict it to some range etc
                       (begin 
                         (format #t "~A can only take an integer value, not ~S" symbol new-value)
                         'error)))
		 #f))
     ',var))

> (define-integer int 123)
int
> (set! int 321.67)
321
> (set! int (list 1 2))
;int can only take an integer value, not (1 2)

Here are trace and untrace. We save the previous accessors in trace, restore them upon untrace, and in between, call the previous set accessor, if any, after reporting the set:

(define (trace var)
  (let* ((cur-access (symbol-access var))
	 (cur-set (and cur-access (cadr cur-access))))
    (set! (symbol-access var)
	  (list (and cur-access (car cur-access))
		(lambda (symbol new-value) 
		  (format #t "~A set to ~A~%" symbol new-value) 
		  (if cur-set 
		      (cur-set symbol new-value)
		      new-value))
		(and cur-access (caddr cur-access))
		cur-access)))) ; save the old version 

(define (untrace var)
  (if (and (symbol-access var)
	   (cdddr (symbol-access var)))
      (set! (symbol-access var) (cadddr (symbol-access var)))))

The first element of the symbol-accessor list is not used by s7. It is used for CL-style property lists by symbol-plist and friends in s7test.scm. I believe symbol-access is similar in spirit to a hooked variable in Ruby, or perhaps a tied variable in Perl (I know nothing about Perl).

miscellaneous stuff

*load-path* is a list of directories to search when loading a file. *load-hook* is a hook whose functions are called just before a file is loaded. The hook function argument, named 'name, is the filename. While loading, the port-filename and port-line-number of the current-input-port can tell you where you are in the file.

(set! (hook-functions *load-hook*)
       (list (lambda (hook) 
               (format #t "loading ~S...~%" (hook 'name)))))

Here's a *load-hook* function that adds the loaded file's directory to the *load-path* variable so that subsequent loads don't need to specify the directory:

(set! (hook-functions *load-hook*)
  (list (lambda (hook)
          (let* ((pos -1)
                 (filename (hook 'name))
	         (len (length filename)))
            (do ((i 0 (+ i 1)))
	        ((= i len))
	      (if (char=? (filename i) #\/)
	          (set! pos i)))
            (if (positive? pos)
	        (let ((directory-name (substring filename 0 pos)))
	          (if (not (member directory-name *load-path*))
		      (set! *load-path* (cons directory-name *load-path*)))))))))

As in Common Lisp, *features* is a list describing what is currently loaded into s7. You can check it with the provided? function, or add something to it with provide. In my version of Snd, at startup *features* is:

> *features*
(snd12 snd snd-s7 snd-motif gsl alsa xm clm4 clm sndlib gmp s7)
> (provided? 'gmp)
#t

Multi-line and in-line comments can be enclosed in #| and |#. (+ #| add |# 1 2).

Leaving aside this case and the booleans, #f and #t, you can specify your own handlers for tokens that start with "#". *#readers* is a list of pairs: (char . func). "char" refers to the first character after the sharp sign (#). "func" is a function of one argument, the string that follows the #-sign up to the next delimiter. "func" is called when #<char> is encountered. If it returns something other than #f, the #-expression is replaced with that value. Scheme has several predefined #-readers for cases such as #b1, #\a, #i123, and so on, but you can override these if you like. If the string passed in is not the complete #-expression, the function can use read-char or read to get the rest. Say we'd like #t<number> to interpret the number in base 12:

(set! *#readers* (cons (cons #\t (lambda (str) (string->number (substring str 1) 12))) *#readers*))

> #tb
11
> #t11.3
13.25

Or have #c(real imag) be read as a complex number:

(set! *#readers* (cons (cons #\c (lambda (str) (apply make-rectangular (read)))) *#readers*))

> #c(1 2)
1+2i

I use *#readers* primarily to implement a way to get the current line number and file name, along the lines of C's __LINE__ and __FILE__. port-line-number works if we're reading a file (during load for example), and (error-environment) has the same information if an error happens. But during Snd's auto-test sequence, there are many cases that aren't errors, and the file is no longer being loaded, but I need to know where something unexpected happened. So:

(set! *#readers* 
      (cons (cons #\_ (lambda (str)
			(if (string=? str "__line__")
			    (port-line-number)
			    (if (string=? str "__file__")
			        (port-filename)
			        #f))))
            *#readers*))

Here's a reader macro for read-time evaluation:

(set! *#readers*
  (cons (cons #\. (lambda (str)
		    (if (string=? str ".") (eval (read)) #f)))
	*#readers*))

> '(1 2 #.(* 3 4) 5)
(1 2 12 5)

To return no value from a reader, use (values).

> (set! *#readers* (cons (cons #\; (lambda (s) (read) (values))) *#readers*))
((#\; . #<closure>))
> (+ 1 #;(* 2 3) 4)
5

(make-list length (initial-element #f)) returns a list of 'length' elements defaulting to 'initial-element'.

reverse! is an in-place version of the built-in function reverse. That is, it modifies the list passed to it in the process of reversing its contents. list-set! sets a member of a list. sort! sorts a list or a vector using the function passed as its second argument to choose the new ordering.

> (sort! (list 3 4 8 2 0 1 5 9 7 6) <)
(0 1 2 3 4 5 6 7 8 9)

(define (mix-notelists . notelists)
  ;; assume the second parameter is the begin time in seconds (the first is the instrument name)
  (sort!
   (apply append notelists)
   (lambda (note1 note2)
     (< (cadr note1) (cadr note2)))))

> (mix-notelists '((fm-violin 0 1 440 .1)
		   (fm-violin 1 1 550 .1))
	         '((bird 0 .1 )
		   (bird .2 .1)
		   (bird 1.2 .3)
		   (bird .5 .5)))

((bird 0 0.1) (fm-violin 0 1 440 0.1) (bird 0.2 0.1) (bird 0.5 0.5) (fm-violin 1 1 550 0.1) (bird 1.2 0.3))

Despite the "!" in its name, sort! actually copies any list argument passed to it, but vectors are sorted in place.

Keywords exist mainly for define*'s benefit. The keyword functions are: keyword?, make-keyword, symbol->keyword, and keyword->symbol. A keyword is a symbol that starts or ends with a colon. The colon is considered to be a part of the symbol name. A keyword is a constant that evaluates to itself.

(symbol-table) returns the symbol table, a vector of lists of symbols. Here we scan the symbol table looking for any function that doesn't have documentation:

(let ((st (symbol-table)))
  (do ((i 0 (+ i 1))) 
      ((= i (length st)))
    (let ((lst (st i)))
      (for-each 
        (lambda (sym)
          (if (defined? sym)
              (let ((val (symbol->value sym)))
                (if (and (procedure? val)
                         (string=? "" (procedure-documentation val)))
                    (format #t "~A " sym)))))
        lst))))

But geez how boring can something be? Nobody cares about procedure-documentation! Here's a better example, an automatic software torture tester.

(let ((constants (list #f #t pi () 1 1.5 3/2 1.5+i)))

  (define (autotest func args args-left)
    (catch #t (lambda () (apply func args)) (lambda any #f))
    (if (> args-left 0)
	(for-each
	 (lambda (c)
	   (autotest func (cons c args) (- args-left 1)))
	 constants)))

  (let ((st (symbol-table)))
    (do ((i 0 (+ i 1))) 
	((= i (length st)))
      (let ((lst (st i)))
	(for-each 
	 (lambda (sym)
	   (if (defined? sym)
	       (let ((val (symbol->value sym)))
		 (if (procedure? val)
		     (let* ((arity (procedure-arity val))
			    (req (car arity))
			    (opt (cadr arity))
			    (rst (caddr arity)))
		       (if (or (> (+ opt req) 4)
		       	       (member (symbol->string sym) '("exit" "abort")))
			   (format #t ";skip ~A for now~%" sym) ; no time! no time!
			   (begin
			     (format #t ";whack on ~A...~%" sym)
			     (autotest val () (+ req opt 1 (if rst 1 0))))))))))
	 lst)))))

Equally useful, a profiler:

(define (profile expression)
  "(profile func) evaluates the function, then reports profiling information. \
The function takes one argument, the environment in which loads and evals should operate."

  (define calls (make-vector 4096 (cons 'unused -1)))
  (define call 0)

  (define (profile-1 n)
    (set! (cdr (calls n)) (+ (cdr (calls n)) 1)))
  
  (define (wrap-all)
    (let ((st (symbol-table))
	  (e (environment)))
      (do ((i 0 (+ i 1))) 
	  ((= i (length st)))
	(let ((lst (st i)))
	  (for-each 
	   (lambda (sym)
	     (if (and (defined? sym)
		      (not (constant? sym)))
		 (let ((val (symbol->value sym)))
		   (if (procedure? val)
		       (let ((new-val (apply lambda 'args `((profile-1 ,call) (apply ,val args)))))
			 (set! (calls call) (cons sym 0))
			 (set! call (+ call 1))
			 (if (procedure-setter val)
			     (set! (procedure-setter new-val) (procedure-setter val)))
			 (augment-environment! e (cons sym new-val)))))))
	   lst)))
      e))

  (expression (wrap-all))
  (sort! calls (lambda (a b) (> (cdr a) (cdr b))))
  (do ((i 0 (+ i 1)))
      ((= i call))
    (if (> (cdr (calls i)) 0)
	(format #t "~A: ~A~%" (car (calls i)) (cdr (calls i))))))

> (profile (lambda (e) 
             (load "lint.scm" e) 
             (eval '(lint "dsp.scm") e)))
;;; many lines of data print out

help tries to find information about its argument.

> (help 'caadar)
"(caadar lst) returns (car (car (cdr (car lst)))): (caadar '((1 (2 3)))) -> 2"

If the initial expression in a function body is a string constant, it is assumed to be a documentation string (accessible via help or procedure-documentation):

(define (add1 a)
  "(add1 a) adds 1 to its argument"
  (+ a 1))

> (help add1)
"(add1 a) adds 1 to its argument"

gc calls the garbage collector. (gc #f) turns off the GC, and (gc #t) turns it on.

(morally-equal? x y)

Say we want to check that two different computations came to the same result, and that result might involve circular structures. Will equal? be our friend?

> (equal? 2 2.0)
#f
> (let ((x +nan.0)) (equal? x x))
#f
> (equal? .1 1/10)
#f    
> (= .1 1/10)
#f
> (= 0.0 0+1e-300i)
#f

No! We need an equality check that ignores epsilonic differences in real and complex numbers, and knows that NaNs are equal for all practical purposes. Leaving aside numbers, closed ports are not equal, yet nothing can be done with them. #() is not equal to #2d(). And two closures are never equal, even if their arguments, environments, and bodies are equal. Since there might be circles, it is not easy to write a replacement for equal? in Scheme. So, in s7, if one thing is basically the same as some other thing, they satisfy the function morally-equal?.

> (morally-equal? 2 2.0)        ; would "equal!?" be a better name?
#t
> (morally-equal? 1/0 1/0)      ; NaN
#t
> (morally-equal? .1 1/10)
#t                              ; floating-point epsilon here is 1.0e-15 or thereabouts
> (morally-equal? 0.0 1e-300)
#t
> (morally-equal? 0.0 1e-14)
#f                              ; its not always #t!
> (morally-equal? (lambda () #f) (lambda () #f))
#t

I'd be happy to add *float-epsilon* or some such variable, if anyone needs it. Also, I can't decide how bignums should interact with morally-equal?. Currently, if a bignum is involved, either here or in a hash-table, s7 uses equal?.

Some other differences from r5rs:

Here are some changes I'd make to s7 if I didn't care about compatibility with other Schemes:

and perhaps:

There are several other less-than-ideal names. Perhaps s7 should use *pi*, *most-negative-fixnum*, *most-positive-fixnum* so that all the built-in variables and constants have the same kind of name. All the environment function names are bad — "environment" by itself is annoying. At least we can say s7 is environmentally sound! get-output-string should be current-output-string. "cond-expand" is bad — we're not expanding anything, and the macro's point is to make it easy to operate with the *features* list; perhaps "cond-provided"? write-char behaves like display, not write. symbol-access is not ideal. Better ideas are always welcome!

Schemes vary in their treatment of (). s7 considers it a constant that evaluates to itself, so you rarely need to quote it. (eq? () '()) is #t. This is consistent with, for example, (eq? #f '#f) which is also #t. The standard says "the empty list is a special object of its own type", so surely either choice is acceptable in that regard. One place where the quote matters is in a case statement; the selector is evaluated but the key is not:

> (case '() ((()) 2) (else 1))
2
> (case '() (('()) 2) (else 1)) ; (eqv? '() ''()) is #f
1
;;; which parallels #f (or a number such as 2 etc):
> (case '#f ((#f) 2) (else 1))
2
> (case '#f (('#f) 2) (else 1)) ; (eqv? '#f ''#f) is #f
1

Similarly, vector constants do not have to be quoted. A list constant is quoted to keep it from being evaluated, but #(1 2 3) is as unproblematic as "123" or 123.

Schemes also vary in handling trailing arguments: (* 0 "hi") in Guile returns 0, but s7 gives an error. (cond (1) (=>)) is 1 in both, and (or 1 2 . 3) is an error in Guile, and 1 in s7! Because it flushes trailing arguments, Guile returns 0 from (* 0 +inf.0), but I think it should return NaN.

And a harder one... How should s7 treat this: (string-set! "hiho" 1 #\z), or (vector-set! #(1 2 3) 1 32), or (list-set! '(1 2 3) 1 32)? Originally, in s7, the first two were errors, and the third was allowed, which doesn't make much sense. Guile and Common Lisp accept all three, but that leads to weird cases where we can reach into a function's body:

> (let ((x (lambda () '(1 2 3)))) (list-set! (x) 1 32) (x))
(1 32 3) ; s7, Guile

> (flet ((x () '(1 2 3))) (setf (nth 1 (x)) 32) (x))
(1 32 3) ; Clisp

> (let ((x (lambda () (list 1 2 3)))) (list-set! (x) 1 32) (x))
(1 2 3)

But it's possible to reach into a function's closure, even when the closed-over thing is a constant:

> (flet ((x () '(1 2 3))) (setf (nth 1 (x)) 32) (x))
(1 32 3)

> (let ((xx (let ((x '(1 2 3))) (lambda () x)))) (list-set! (xx) 1 32) (xx))
(1 32 3)

> (let ((xx (let ((x (list 1 2 3))) (lambda () x)))) (list-set! (xx) 1 32) (xx))
(1 32 3)

And it's possible to reach into a constant list via list-set! (or set-car! of course):

> (let* ((x '(1 2)) (y (list x)) (z (car y))) (list-set! z 1 32) (list x y z))
((1 32) ((1 32)) (1 32))

It would be a programmer's nightmare to have to keep track of which piece of a list is constant, and an implementor's nightmare to copy every list. set! in all its forms is used for its side-effects, so why should we try to put a fence around them? If we flush "immutable constant" because it is a ham-fisted, whack-it-with-a-shovel approach, the only real problem I can see is symbol->string. In CL, this is explicitly an error:

> (setf (elt (symbol-name 'xyz) 1) #\X)
*** - Attempt to modify a read-only string: "XYZ"

And in Guile:

> (string-set! (symbol->string 'symbol->string) 1 #\X)
ERROR: string is read-only: "symbol->string"

So both have a notion of immutable strings. I wonder what other Scheme programmers (not implementors!) want in this situation. Currently, there are no immutable list, string, or vector constants, and symbol->string returns a copy of the string. One simple way to ensure immutability is to use copy:

> (let ((x (lambda () (copy "hiho")))) (string-set! (x) 1 #\x) (x))
"hiho"

There is one pitfall here. s7 normally tries to optimize garbage collection by removing some constants from the heap. If that constant is a list or a vector, and you later set some member of it to something that needs GC protection, nobody in the heap points to it, so it is GC'd. Here is an example:

(define (bad-idea)
  (let ((lst '(1 2 3)))              ; or #(1 2 3) and vector-ref|set
    (let ((result (list-ref lst 1)))
     (list-set! lst 1 (* 2.0 16.6))
     (gc)
     result)))

Put this is a file, load it into the interpreter, then call (bad-idea) a few times. You can turn off the optimization in question by setting the variable *safety* to 1. *safety* defaults to 0. (Also, if *safety* is not 0, sort! is safe from infinite loops).

s7 handles circular lists and vectors and dotted lists with its customary aplomb. You can pass them to memq, or print them, for example; you can even evaluate them. The print syntax is borrowed from CL:

> (let ((lst (list 1 2 3))) 
    (set! (cdr (cdr (cdr lst))) lst) 
    lst)
#1=(1 2 3 . #1#)

> (let* ((x (cons 1 2)) 
         (y (cons 3 x))) 
    (list x y))
(#1=(1 . 2) (3 . #1#)) ; shared lists use the same syntax: '((1 . 2) (3 1 . 2)) spelled out

But should this syntax be readable as well? I'm inclined to say no because then it is part of the language, and it doesn't look like the rest of the language. (I think it's kind of ugly). Perhaps we could implement it via *#readers*.

Length returns +inf.0 if passed a circular list, and returns a negative number if passed a dotted list. In the dotted case, the absolute value of the length is the list length not counting the final cdr. (define (circular? lst) (infinite? (length lst))).

Here's an amusing use of circular lists:

(define (for-each-permutation func vals)
  ;; apply func to every permutation of vals: 
  ;;   (for-each-permutation (lambda args (format #t "~{~A~^ ~}~%" args)) '(1 2 3))
  (define (pinner cur nvals len)
    (if (= len 1)
        (apply func (cons (car nvals) cur))
        (do ((i 0 (+ i 1)))                       ; I suppose a named let would be more Schemish
            ((= i len))
          (let ((start nvals))
            (set! nvals (cdr nvals))
            (let ((cur1 (cons (car nvals) cur)))  ; add (car nvals) to our arg list
              (set! (cdr start) (cdr nvals))      ; splice out that element and 
              (pinner cur1 (cdr start) (- len 1)) ;   pass a smaller circle on down, "wheels within wheels"
              (set! (cdr start) nvals))))))       ; restore original circle
  (let ((len (length vals)))
    (set-cdr! (list-tail vals (- len 1)) vals)    ; make vals into a circle
    (pinner () vals len)
    (set-cdr! (list-tail vals (- len 1)) ())))    ; restore its original shape

s7 and Snd use "*" in a variable name, *features* for example, to indicate that the variable is predefined. It may occur unprotected in a macro, for example. The "*" doesn't mean that the variable is special in the CL sense of dynamic scope, but some clear marker is needed for a global variable so that the programmer doesn't accidentally step on it.

Although a variable name's first character is more restricted, currently only #\null, #\newline, #\tab, #\space, #\), #\(, #\", and #\; can't occur within the name. I did not originally include double-quote in this set, so wild stuff like (let ((nam""e 1)) nam""e) would work, but that means that '(1 ."hi") is parsed as a 1 and the symbol ."hi", and (string-set! x"hi") is an error. The first character should not be #\#, #\', #\`, #\,, #\:, or any of those mentioned above, and some characters can't occur by themselves. For example, "." is not a legal variable name, but ".." is. These weird symbols have to be printed sometimes:

> (list 1 (string->symbol (string #\; #\" #\\)) 2)
(1 ;"\ 2)            
> (list 1 (string->symbol (string #\.)) 2)
(1 . 2)

which is a mess. Guile prints the first as (1 #{\;\"\\}# 2). In CL and some Schemes:

[1]> (list 1 (intern (coerce (list #\; #\" #\\) 'string)) 2) ; thanks to Rob Warnock
(1 |;"\\| 2)        
[2]> (equalp 'A '|A|) ; in CL case matters here
T

This is clean, and has the weight of tradition behind it, but I think I'll use "symbol" instead:

> (list 1 (string->symbol (string #\; #\" #\\)) 2)
(1 (symbol ";\"\\") 2)       

This output is readable, and does not eat up perfectly good characters like vertical bar, but it means we can't easily use variable names like "| e t c |". We could allow a name to contain any characters if it starts and ends with "|", but then one vertical bar is trouble.

These symbols are not just an optimization of string comparison:

(define-macro (hi a) 
  (let ((funny-name (string->symbol (string #\;))))
    `(let ((,funny-name ,a)) (+ 1 ,funny-name))))

> (hi 2)
3
> (macroexpand (hi 2))
(let ((; 2)) (+ 1 ;))    ; for a good time, try (string #\") 

(define-macro (hi a) 
  (let ((funny-name (string->symbol "| e t c |")))
    `(let ((,funny-name ,a)) (+ 1 ,funny-name))))
> (hi 2)
3
> (macroexpand (hi 2))
(let ((| e t c | 2)) (+ 1 | e t c |))

> (let ((funny-name (string->symbol "| e t c |"))) ; now use it as a keyword arg to a function
    (apply define* `((func (,funny-name 32)) (+ ,funny-name 1)))
    ;; (procedure-source func) is (lambda* ((| e t c | 32)) (+ | e t c | 1))
    (apply func (list (symbol->keyword funny-name) 2)))
3

I hope that makes you as happy as it makes me!

The built-in syntactic forms, such as "begin", are almost first-class citizens.

> (let ((progn begin)) 
    (progn 
      (define x 1) 
      (set! x 3) 
      (+ x 4)))
7
> (let ((function lambda)) 
    ((function (a b) (list a b)) 3 4))
(3 4)
> (apply begin '((define x 3) (+ x 2)))
5
> ((lambda (n) (apply n '(((x 1)) (+ x 2)))) let)
3

(define-macro (symbol-set! var val) ; like CL's set
  `(apply set! ,var ',val ()))      ; trailing nil is just to make apply happy -- apply*?

(define-macro (progv vars vals . body)
 `(apply (apply lambda ,vars ',body) ,vals))

> (let ((s '(one two)) (v '(1 2))) (progv s v (+ one two)))
3

We can snap together program fragments ("look Ma, no macros!"):

(let* ((x 3) 
       (arg '(x)) 
       (body `((+ ,x x 1)))) 
  ((apply lambda arg body) 12)) ; "legolambda"?

(let ()
  (define (hi a) (+ a x))
  ((apply let '((x 32)) (list (procedure-source hi))) 12)) ; one function, many closures?

(let ((ctr -1))  ; (enum zero one two) but without using a macro
  (apply begin 
    (map (lambda (symbol) 
           (set! ctr (+ ctr 1)) 
           (list 'define symbol ctr)) ; e.g. '(define zero 0) 
         '(zero one two)))
  (+ zero one two))

If you apply define or define-macro, the returned value is a symbol, so to apply the new function or macro, you need to use either eval or symbol->value:

> ((symbol->value (apply define-macro '((m a) `(+ 1 ,a)))) 3)
4
> ((symbol->value (apply define '((hi a) (+ a 1)))) 3)
4

This gives us a way to make anonymous macros, just as lambda returns an anonymous function:

(define-macro (mu args . body)
  `(symbol->value (apply define-macro '((,(gensym) ,@args) ,@body))))

> ((mu (a) `(+ 1 ,a)) 3)
4
> (define-macro (glambda args) ; returns an anonymous macro that will return a function given a body
    `(symbol->value (define-macro (,(gensym) . body) 
                      `(lambda ,',args ,@body))))
glambda
> (let ((gf (glambda (a b))))  ; gf is now ready for any body that involves arguments 'a and 'b
    ((gf (+ a b)) 1 2))        ; apply (lambda (a b) (+ a b)) to '(1 2)
3

Currently, you can't set! a built-in syntactic keyword to some new value: (set! if 3). I hope this kind of thing is not actually very useful, but let me know if you need it. The issue is purely one of speed.

In s7, there is only one kind of begin statement, and it can contain both definitions and expressions. These are evaluated in the order in which they occur, and in the environment at the point of the evaluation. I think of it as being a little REPL. begin does not introduce a new frame in the current environment, so defines happen in the enclosing environment.

> (let ((y 2)) 
      (let ((x 1)) 
        (begin 
          (define x y)         ; x is 2 (this x is the same as the x in the let above it)
          (set! x (* x 2))     ; now it is 4
          (define y 123))      ; this is great, but it doesn't affect x 
         x))                   ; defines in begin are in the enclosing environment so
   4                           ;   we get 4

r7rs:

  (define vector-map map) 
  (define string-map map) 
  (define vector-for-each for-each) 
  (define string-for-each for-each) 
  (define copy-vector copy) 
  (define copy-list copy)
  (define (vector->string vect) (apply string (vector->list vect)))
  (define (string->vector str) (apply vector (string->list str)))

  (define char-foldcase char-downcase) 
  (define string-foldcase string-downcase)
  (define (string-downcase str) (apply string (map char-downcase str)))
  (define (string-upcase str) (apply string (map char-upcase str)))

  (define -inf.0 (real-part (log 0.0))) 
  (define +inf.0 (- -inf.0))    ; or (atanh 1.0)
  (define +nan.0 1/0)           ; 1/0 for NaN is just a convenience, not a religious dogma
  (define -nan.0 (- 1/0))
  (define (finite? n) (and (number? n) (not (nan? n)) (not (infinite? n))))

  (define exact-integer? integer?)	
  (define (exact-integer-sqrt i) (let ((sq (floor (sqrt i)))) (values sq (- i (* sq sq)))))
  (define inexact exact->inexact)
  (define exact inexact->exact)

  (define (port-open? p) (not (port-closed? p))) 
  (define (character-port? p) #t) 
  (define (binary-port? p) #t)
  (define (port? p) (or (input-port? p) (output-port? p)))

  (define read-u8 read-byte) 
  (define write-u8 write-byte) 
  (define u8-ready? char-ready?) 
  (define peek-u8 peek-char)

  (define blob? vector?)      ; "blob???"
  (define make-blob make-vector) 
  (define blob-length length) 
  (define blob-copy copy)
  (define blob-u8-ref vector-ref) 
  (define blob-u8-set! vector-set!)

  (define interaction-environment current-environment)
  (define-bacro (include file) `(load ,file (outer-environment (current-environment))))

  (set! *#readers* (cons (cons #\; (lambda (s) (read) (values))) *#readers*))

  (define-macro (when test . forms) `(if ,test (begin ,@forms)))
  (define-macro (unless test . forms) `(if (not ,test) (begin ,@forms)))

  (define-macro (define-values vars . body)
    `(apply begin (map (lambda (var val) `(define ,var ,val)) ',vars (list (begin ,@body)))))

  (define-macro (let*-values vars . body)
    `(let () 
       ,@(map (lambda (nvars . nbody)
	        `(apply define-values ',nvars ',@nbody))
	      (map car vars) (map cdr vars))
       ,@body))

  ;; floor/ and friends can't work as intended: they assume that multiple values 
  ;;   are not spliced.  The "division library" is a trivial, pointless micro-optimization.
  ;;   (define (floor-remainder x y) (- x (* y (floor (/ x y)))))
  ;; and why no euclidean-rationalize or exact-integer-expt?
  ;;   (imagine what will happen when r8rs stumbles on the zoo of continued fraction algorithms!)

  ;; get-environment-variable is a bad name: "environment" is already in use, and "get-"
  ;;   in any name should raise a red flag.

  ;; command-line is problematic: s7 has no access to the caller's "main" function, and
  ;;   outside Windows, there's no reasonable way to get these arguments.

s7 originally had multithreading support, but I removed it in August, 2011. It turned out to be less useful than I hoped, mainly because s7 threads shared the heap and therefore had to coordinate all cell allocations. It was faster and simpler to use multiple pthreads each running a separate s7 interpreter, rather than one s7 running multiple s7 threads. In CLM, there was also contention for access to the output stream. In GUI-related situations, as discussed at length below, threads were not useful mainly because the GUI toolkits are not thread safe, so we use begin_hook now instead. Last but not least, the effort to make the non-threaded s7 faster messed up parts of the threaded version. Rather than waste a lot of time fixing this, I chose to flush multithreading.

"Life", a poem.

(+(*(+))(*)(+(+)(+)(*)))
(((((lambda () (lambda () (lambda () (lambda () 1))))))))
(+ (((lambda () values)) 1 2 3))
(map apply (list map) (list map) (list (list *)) '((((1 2)) ((3 4 5)))))
(do ((do do do)) (do do do))
(*(*)(*) (+)(+) 1)

FFI examples

s7 exists only to serve as an extension of some other application, so it is primarily a foreign function interface. s7.h has lots of comments about the individual functions. Here I'll collect some complete examples. s7.c depends on the following compile-time flags:

HAVE_STDBOOL_H                 1 if you have stdbool.h
HAVE_SYS_PARAM_H               1 if you're running FreeBSD
HAVE_GETTIMEOFDAY              1 if you want timing info from the GC (default is 0)
HAVE_LSTAT                     1 if you want the load function to be able to tell a normal file from a directory
SIZEOF_VOID_P                  8 (default) or 4, but it's not a big deal (a minor optimization of a switch statement)

WITH_GMP                       1 if you want multiprecision arithmetic (requires gmp, mpfr, and mpc, default is 0)
WITH_COMPLEX                   1 if your compiler supports complex numbers
HAVE_COMPLEX_TRIG              1 if your math library has complex versions of the trig functions
WITH_AT_SIGN_AS_EXPONENT       1 if you want "@" to mark exponents (default is 1)
WITH_EXTRA_EXPONENT_MARKERS    1 if you want "d", "f", "l", and "s" in addition to "e" as exponent markers (default is 1)
                                   if someone defends these exponent markers, ask him to read 1l11+11l1i
WITH_SYSTEM_EXTRAS             1 if you want some additional OS-related functions built-in (default is 0)
S7_DISABLE_DEPRECATED          1 if you want to make sure you're not using any deprecated s7 stuff (default is 0)

See the comment at the start of s7.c for more information about these switches. s7.h defines the two main number types: s7_Int and s7_Double. The examples that follow show:

A simple listener

#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include "s7.h"

static s7_pointer our_exit(s7_scheme *sc, s7_pointer args) 
{
  /* all added functions have this form, args is a list, 
   *    s7_car(args) is the first arg, etc 
   */
  exit(1);
  return(s7_nil(sc)); /* never executed, but makes the compiler happier */
}

int main(int argc, char **argv)
{
  s7_scheme *s7;
  char buffer[512];
  char response[1024];

  s7 = s7_init();                     /* initialize the interpreter */
  s7_define_function(s7, "exit", our_exit, 0, 0, false, "(exit) exits the program");
                                      /* add the function "exit" to the interpreter.
                                       *   0, 0, false -> no required args,
				       *                  no optional args,
				       *                  no "rest" arg
				       */
  while (1)                           /* fire up a REPL */
    {
      fprintf(stdout, "\n> ");        /* prompt for input */
      fgets(buffer, 512, stdin);
      if ((buffer[0] != '\n') || 
	  (strlen(buffer) > 1))
	{                            
	  sprintf(response, "(write %s)", buffer);
	  s7_eval_c_string(s7, response); /* evaluate input and write the result */
	}
    }
}

/* make mus-config.h (it can be empty), then
 *
 *   gcc -c s7.c -I.
 *   gcc -o doc7 doc7.c s7.o -lm -I.
 *
 * run it:
 *
 *    doc7
 *    > (+ 1 2)
 *    3
 *    > (define (add1 x) (+ 1 x))
 *    add1
 *    > (add1 2)
 *    3
 *    > (exit)
 */

Since this reads stdin and writes stdout, it can be run as a Scheme subjob of emacs. One (inconvenient) way to do this is to set the emacs variable scheme-program-name to the name of the exectuable created above ("doc7"), then call the emacs function run-scheme: M-x eval-expression in emacs, followed by (setq scheme-program-name "doc7"), then M-x run-scheme, and you're talking to s7 in emacs. Of course, this connection can be customized indefinitely. See, for example, snd-inf.el in the Snd package.

To read stdin while working in a GUI-based program is trickier. In glib/gtk, you can use something like this:

static gboolean read_stdin(GIOChannel *source, GIOCondition condition, gpointer data)
{
  /* here read from g_io_channel_unix_get_fd(source) and call s7_eval_string */
  return(true);
}

/* ... during initialization ... */

GIOChannel *channel;
channel = g_io_channel_unix_new(STDIN_FILENO);  /* watch stdin */
stdin_id = g_io_add_watch_full(channel,         /* and call read_stdin above if input is noticed */
			       G_PRIORITY_DEFAULT, 
			       (GIOCondition)(G_IO_IN | G_IO_HUP | G_IO_ERR), 
			       read_stdin, NULL, NULL);
g_io_channel_unref(channel);

If we accidentally get into an infinite loop in our REPL, we have to exit the program. See signal handling and begin_hook to the rescue! below for two ways to fix this.

Speaking of emacs, I have these lines in my ~/.emacs file:

(put 'with-environment 'scheme-indent-function 1)
(put 'with-baffle 'scheme-indent-function 0)
(put 'catch 'scheme-indent-function 1)
(put 'augment-environment 'scheme-indent-function 1)

Define a function with arguments and a returned value, and a variable

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "s7.h"

static s7_pointer our_exit(s7_scheme *sc, s7_pointer args)
{
  exit(1);
  return(s7_nil(sc));
}

static s7_pointer add1(s7_scheme *sc, s7_pointer args)
{
  if (s7_is_integer(s7_car(args)))
    return(s7_make_integer(sc, 1 + s7_integer(s7_car(args))));
  return(s7_wrong_type_arg_error(sc, "add1", 1, s7_car(args), "an integer"));
}

int main(int argc, char **argv)
{
  s7_scheme *s7;
  char buffer[512];
  char response[1024];

  s7 = s7_init();                     /* initialize the interpreter */
  
  s7_define_function(s7, "exit", our_exit, 0, 0, false, "(exit) exits the program");
  s7_define_function(s7, "add1", add1, 1, 0, false, "(add1 int) adds 1 to int");
  s7_define_variable(s7, "my-pi", s7_make_real(s7, 3.14159265));

  while (1)                           /* fire up a "repl" */
    {
      fprintf(stdout, "\n> ");        /* prompt for input */
      fgets(buffer, 512, stdin);
      if ((buffer[0] != '\n') || 
	  (strlen(buffer) > 1))
	{                            
	  sprintf(response, "(write %s)", buffer);
	  s7_eval_c_string(s7, response); /* evaluate input and write the result */
	}
    }
}

/*    doc7
 *    > my-pi
 *    3.14159265
 *    > (+ 1 (add1 1))
 *    3
 *    > (exit)
 */

Call a Scheme-defined function from C, and get/set Scheme variable values in C

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "s7.h"

int main(int argc, char **argv)
{
  s7_scheme *s7;
  s7 = s7_init();

  s7_define_variable(s7, "an-integer", s7_make_integer(s7, 1));
  s7_eval_c_string(s7, "(define (add1 a) (+ a 1))");
  
  fprintf(stderr, "an-integer: %d\n", 
	  s7_integer(s7_name_to_value(s7, "an-integer")));

  s7_symbol_set_value(s7, s7_make_symbol(s7, "an-integer"), s7_make_integer(s7, 32));

  fprintf(stderr, "now an-integer: %d\n", 
	  s7_integer(s7_name_to_value(s7, "an-integer")));

  fprintf(stderr, "(add1 2): %d\n", 
	  s7_integer(s7_call(s7, 
			     s7_name_to_value(s7, "add1"), 
			     s7_cons(s7, s7_make_integer(s7, 2), s7_nil(s7)))));
}

/*
 *    doc7
 *    an-integer: 1
 *    now an-integer: 32
 *    (add1 2): 3
 */

C++ and Juce, from Rick Taube

int main(int argc, const char* argv[]) 
{ 
  initialiseJuce_NonGUI(); 

  s7_scheme *s7 = s7_init(); 
  if (!s7) 
    { 
      std::cout <<  "Can't start S7!\n"; 
      return -1; 
    } 

  s7_pointer val; 
  std::string str; 
  while (true) 
    { 
      std::cout << "\ns7> "; 
      std::getline(std::cin, str); 
      val = s7_eval_c_string(s7, str.c_str()); 
      std::cout << s7_object_to_c_string(s7, val); 
    } 

  free(s7); 
  std::cout << "Bye!\n"; 
  return 0; 
} 

Load sndlib into an s7 repl

#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <unistd.h>

/* assume we've configured and built sndlib, so it has created a mus-config.h file.
 * also assume we've built s7 with WITH_SYSTEM_EXTRAS set, so we have file-exists? and delete-file
 */

#include "mus-config.h"
#include "s7.h"
#include "xen.h"
#include "clm.h"
#include "clm2xen.h"

/* we need to redirect clm's mus_error calls to s7_error */

static void mus_error_to_s7(int type, char *msg)
{
  s7_error(s7,                               /* s7 is declared in xen.h */
	   s7_make_symbol(s7, "mus-error"),
	   s7_cons(s7, s7_make_string(s7, msg), s7_nil(s7)));
}

static s7_pointer our_exit(s7_scheme *sc, s7_pointer args)
{
  exit(1);
  return(s7_nil(sc));
}

int main(int argc, char **argv)
{
  char buffer[512];
  char response[1024];

  s7 = s7_init();                     /* initialize the interpreter */
  xen_initialize();                   /* initialize the xen stuff (hooks and the xen s7 FFI used by sndlib) */
  Init_sndlib();                      /* initialize sndlib with all the functions linked into s7 */  

  mus_error_set_handler(mus_error_to_s7); /* catch low-level errors and pass them to s7-error */
  s7_define_function(s7, "exit", our_exit, 0, 0, false, "(exit) exits the program");

  while (1)                           /* fire up a "repl" */
    {
      fprintf(stdout, "\n> ");        /* prompt for input */
      fgets(buffer, 512, stdin);

      if ((buffer[0] != '\n') || 
	  (strlen(buffer) > 1))
	{                            
	  sprintf(response, "(write %s)", buffer);
	  s7_eval_c_string(s7, response); /* evaluate input and write the result */
	}
    }
}

/* gcc -o doc7 doc7.c -lm -I. /usr/local/lib/libsndlib.a -lasound
 *
 *   (load "sndlib-ws.scm")
 *   (with-sound () (outa 10 .1))
 *   (load "v.scm")
 *   (with-sound () (fm-violin 0 .1 440 .1))
 *
 * you might also need -lgsl -lgslcblas
 */

Add a new Scheme type and a procedure with a setter

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "s7.h"

static s7_pointer our_exit(s7_scheme *sc, s7_pointer args)
{
  exit(1);
  return(s7_nil(sc));
}

/* define *listener-prompt* in scheme, add two accessors for C get/set */

static const char *listener_prompt(s7_scheme *sc)
{
  return(s7_string(s7_name_to_value(sc, "*listener-prompt*")));
}

static void set_listener_prompt(s7_scheme *sc, const char *new_prompt)
{
  s7_symbol_set_value(sc, s7_make_symbol(sc, "*listener-prompt*"), s7_make_string(sc, new_prompt));
}

/* now add a new type, a struct named "dax" with two fields, a real "x" and a list "data" */
/*   since the data field is an s7 object, we'll need to mark it to protect it from the GC */

typedef struct {
  s7_Double x;
  s7_pointer data;
} dax;

static char *print_dax(s7_scheme *sc, void *val)
{
  char *data_str, *str;
  int data_str_len;
  dax *o = (dax *)val;
  data_str = s7_object_to_c_string(sc, o->data);
  data_str_len = strlen(data_str);
  str = (char *)calloc(data_str_len + 32, sizeof(char));
  snprintf(str, data_str_len + 32, "#<dax %.3f %s>", o->x, data_str);
  free(data_str);
  return(str);
}

static void free_dax(void *val)
{
  if (val) free(val);
}

static bool equal_dax(void *val1, void *val2)
{
  return(val1 == val2);
}

static void mark_dax(void *val)
{
  dax *o = (dax *)val;
  if (o) s7_mark_object(o->data);
}

static int dax_type_tag = 0;

static s7_pointer make_dax(s7_scheme *sc, s7_pointer args)
{
  dax *o;
  o = (dax *)malloc(sizeof(dax));
  o->x = s7_real(s7_car(args));
  if (s7_cdr(args) != s7_nil(sc))
    o->data = s7_car(s7_cdr(args));
  else o->data = s7_nil(sc);
  return(s7_make_object(sc, dax_type_tag, (void *)o));
}

static s7_pointer is_dax(s7_scheme *sc, s7_pointer args)
{
  return(s7_make_boolean(sc, 
			 s7_is_object(s7_car(args)) &&
			 s7_object_type(s7_car(args)) == dax_type_tag));
}

static s7_pointer dax_x(s7_scheme *sc, s7_pointer args)
{
  dax *o;
  o = (dax *)s7_object_value(s7_car(args));
  return(s7_make_real(sc, o->x));
}

static s7_pointer set_dax_x(s7_scheme *sc, s7_pointer args)
{
  dax *o;
  o = (dax *)s7_object_value(s7_car(args));
  o->x = s7_real(s7_car(s7_cdr(args)));
  return(s7_car(s7_cdr(args)));
}

static s7_pointer dax_data(s7_scheme *sc, s7_pointer args)
{
  dax *o;
  o = (dax *)s7_object_value(s7_car(args));
  return(o->data);
}

static s7_pointer set_dax_data(s7_scheme *sc, s7_pointer args)
{
  dax *o;
  o = (dax *)s7_object_value(s7_car(args));
  o->data = s7_car(s7_cdr(args));
  return(o->data);
}

int main(int argc, char **argv)
{
  s7_scheme *s7;
  char buffer[512];
  char response[1024];

  s7 = s7_init();
  
  s7_define_function(s7, "exit", our_exit, 0, 0, false, "(exit) exits the program");
  s7_define_variable(s7, "*listener-prompt*", s7_make_string(s7, ">"));

  dax_type_tag = s7_new_type("dax", print_dax, free_dax, equal_dax, mark_dax, NULL, NULL);
  s7_define_function(s7, "make-dax", make_dax, 2, 0, false, "(make-dax x data) makes a new dax");
  s7_define_function(s7, "dax?", is_dax, 1, 0, false, "(dax? anything) returns #t if its argument is a dax object");

  s7_define_variable(s7, "dax-x", 
                     s7_make_procedure_with_setter(s7, "dax-x", dax_x, 1, 0, set_dax_x, 2, 0, "dax x field"));

  s7_define_variable(s7, "dax-data", 
                     s7_make_procedure_with_setter(s7, "dax-data", dax_data, 1, 0, set_dax_data, 2, 0, "dax data field"));

  while (1)
    {
      fprintf(stdout, "\n%s ", listener_prompt(s7));
      fgets(buffer, 512, stdin);
      if ((buffer[0] != '\n') || 
	  (strlen(buffer) > 1))
	{                            
	  sprintf(response, "(write %s)", buffer);
	  s7_eval_c_string(s7, response); /* evaluate input and write the result */
	}
    }
}

/*
 *    > *listener-prompt*
 *    ">"
 *    > (set! *listener-prompt* ":")
 *    ":"
 *    : (define obj (make-dax 1.0 (list 1 2 3)))
 *    obj
 *    : obj
 *    #<dax 1.000 (1 2 3)>
 *    : (dax-x obj)
 *    1.0
 *    : (dax-data obj)
 *    (1 2 3)
 *    : (set! (dax-x obj) 123.0)
 *    123.0
 *    : obj
 *    #<dax 123.000 (1 2 3)>
 *    : (dax? obj)
 *    #t
 *    : (exit)
 */

Redirect output (and input) to a C procedure

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "s7.h"

static void my_print(s7_scheme *sc, char c, s7_pointer port)
{
  fprintf(stderr, "[%c] ", c);
}

static s7_pointer my_read(s7_scheme *sc, s7_read_t peek, s7_pointer port)
{
  return(s7_make_character(s7, fgetc(stdin)));
}

int main(int argc, char **argv)
{
  s7_scheme *s7;
  char buffer[512];
  char response[1024];

  s7 = s7_init();  

  s7_set_current_output_port(s7, s7_open_output_function(s7, my_print));
  s7_define_variable(s7, "io-port", s7_open_input_function(s7, my_read));

  while (1) 
    {
      fprintf(stdout, "\n> ");
      fgets(buffer, 512, stdin);
      if ((buffer[0] != '\n') || 
	  (strlen(buffer) > 1))
	{                            
	  sprintf(response, "(write %s)", buffer);
	  s7_eval_c_string(s7, response);
	}
    }
}

/* 
 *    > (+ 1 2)
 *    [3]
 *    > (display "hiho")
 *    [h] [i] [h] [o] [#] [<] [u] [n] [s] [p] [e] [c] [i] [f] [i] [e] [d] [>] 
 *    > (define (add1 x) (+ 1 x))
 *    [a] [d] [d] [1] 
 *    > (add1 123)
 *    [1] [2] [4] 
 *    > (read-char io-port)
 *    a                             ; here I typed "a" in the shell
 *    [#] [\] [a] 
 */

Extend a built-in operator ("+" in this case)

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "s7.h"

static s7_pointer old_add;           /* the original "+" function for non-string cases */
static s7_pointer old_string_append; /* same, for "string-append" */

static s7_pointer our_add(s7_scheme *sc, s7_pointer args)
{
  /* this will replace the built-in "+" operator, extending it to include strings:
   *   (+ "hi" "ho") -> "hiho" and  (+ 3 4) -> 7
   */
  if ((s7_is_pair(args)) &&
      (s7_is_string(s7_car(args))))
    return(s7_apply_function(sc, old_string_append, args));

  return(s7_apply_function(sc, old_add, args));
}

int main(int argc, char **argv)
{
  s7_scheme *s7;
  char buffer[512];
  char response[1024];

  s7 = s7_init();

  /* get built-in + and string-append */
  old_add = s7_name_to_value(s7, "+");      
  old_string_append = s7_name_to_value(s7, "string-append");

  /* redefine "+" */
  s7_define_function(s7, "+", our_add, 0, 0, true, "(+ ...) adds or appends its arguments");

  while (1)
    {
      fprintf(stdout, "\n> ");
      fgets(buffer, 512, stdin);
      if ((buffer[0] != '\n') || 
	  (strlen(buffer) > 1))
	{                            
	  sprintf(response, "(write %s)", buffer);
	  s7_eval_c_string(s7, response);
	}
    }
}

/* 
 *    > (+ 1 2)
 *    3
 *    > (+ "hi" "ho")
 *    "hiho"
 */

C-side define* (s7_define_function_star)

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "s7.h"

static s7_pointer plus(s7_scheme *sc, s7_pointer args)
{
  /* (define* (plus (red 32) blue) (+ (* 2 red) blue)) */
  return(s7_make_integer(sc, 2 * s7_integer(s7_car(args)) + s7_integer(s7_car(s7_cdr(args)))));
}

int main(int argc, char **argv)
{
  s7_scheme *s7;
  char buffer[512];
  char response[1024];

  s7 = s7_init();
  s7_define_function_star(s7, "plus", plus, "(red 32) blue", "an example of define* from C");

  while (1)
    {
      fprintf(stdout, "\n> ");
      fgets(buffer, 512, stdin);
      if ((buffer[0] != '\n') || 
	  (strlen(buffer) > 1))
	{                            
	  sprintf(response, "(write %s)", buffer);
	  s7_eval_c_string(s7, response);
	}
    }
}

/* 
 *    > (plus 2 3)
 *    7
 *    > (plus :blue 3)
 *    67
 *    > (plus :blue 1 :red 4)
 *    9
 *    > (plus 2 :blue 3)
 *    7
 *    > (plus :blue 3 :red 1)
 *    5
 */

C-side define-macro (s7_define_macro)

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "s7.h"

static s7_pointer plus(s7_scheme *sc, s7_pointer args)
{
  /* (define-macro (plus a b) `(+ ,a ,b)) */
  s7_pointer a, b;
  a = s7_car(args);
  b = s7_car(s7_cdr(args));
  return(s7_list(sc, 3, s7_make_symbol(sc, "+"),  a, b));
}

int main(int argc, char **argv)
{
  s7_scheme *s7;
  char buffer[512];
  char response[1024];

  s7 = s7_init();
  s7_define_macro(s7, "plus", plus, 2, 0, false, "plus adds its two arguments");

  while (1)
    {
      fprintf(stdout, "\n> ");
      fgets(buffer, 512, stdin);
      if ((buffer[0] != '\n') || 
	  (strlen(buffer) > 1))
	{                            
	  sprintf(response, "(write %s)", buffer);
	  s7_eval_c_string(s7, response);
	}
    }
}

/* 
 *    > (plus 2 3)
 *    5
 */

define a generic function in C

In scheme, a function becomes generic simply by (apply ((car args) 'func) args). To accomplish the same thing in C, we use s7_search_open_environment and s7_apply_function:

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "s7.h"

static s7_pointer plus(s7_scheme *sc, s7_pointer args)
{
  #define plus_help "(plus obj ...) applies obj's plus method to obj and any trailing arguments."
  s7_pointer obj;
  obj = s7_car(args);
  if (s7_is_open_environment(obj))          /* does obj have methods? */
    {
      s7_pointer method;                    /* does it have a 'plus method? */
      method = s7_search_open_environment(sc, s7_make_symbol(sc, "plus"), obj);
      if (s7_is_procedure(method))
	return(s7_apply_function(sc, method, args));
    }
  return(s7_f(sc));
}

int main(int argc, char **argv)
{
  s7_scheme *s7;
  s7 = s7_init();
  s7_define_function(s7, "plus", plus, 1, 0, true, plus_help);
  while (1)
    {
      char buffer[512];
      char response[1024];
      fprintf(stdout, "\n> ");
      fgets(buffer, 512, stdin);
      if ((buffer[0] != '\n') || 
	  (strlen(buffer) > 1))
	{                            
	  sprintf(response, "(write %s)", buffer);
	  s7_eval_c_string(s7, response);
	}
    }
}

/* gcc -c s7.c -I.
 * gcc -o ex15 ex15.c s7.o -I. -lm 
 *
 * > (plus 1 2)
 * #f
 * (define obj (open-environment 
 *               (environment
 *                 (cons 'plus (lambda args 
 *                               (apply + 1 (cdr args)))))))
 * obj
 * > (plus obj 2 3)
 * 6
 */

Signal handling and continuations

#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <signal.h>

#include "s7.h"

static s7_scheme *s7;
struct sigaction new_act, old_act;  
  
static void handle_sigint(int ignored)  
{  
  fprintf(stderr, "interrupted!\n");
  s7_symbol_set_value(s7, s7_make_symbol(s7, "*interrupt*"), s7_make_continuation(s7)); /* save where we were interrupted */
  sigaction(SIGINT, &new_act, NULL);  
  s7_quit(s7);                             /* get out of the eval loop if possible */
}  

static s7_pointer our_exit(s7_scheme *sc, s7_pointer args)
{ 
  /* this function is really needed if we are trapping C-C! */
  exit(1);
  return(s7_f(sc));
}

static s7_pointer our_sleep(s7_scheme *sc, s7_pointer args)
{
  /* slow down our infinite loop for demo purposes */
  sleep(1);
  return(s7_f(sc));
}

int main(int argc, char **argv)
{
  char buffer[512];
  char response[1024];

  s7 = s7_init();
  s7_define_function(s7, "exit", our_exit, 0, 0, false, "(exit) exits");
  s7_define_function(s7, "sleep", our_sleep, 0, 0, false, "(sleep) sleeps");
  s7_define_variable(s7, "*interrupt*", s7_f(s7)); 
  /* Scheme variable *interrupt* holds the continuation at the point of the interrupt */

  sigaction(SIGINT, NULL, &old_act);
  if (old_act.sa_handler != SIG_IGN)
    {
      memset(&new_act, 0, sizeof(new_act));  
      new_act.sa_handler = &handle_sigint;  
      sigaction(SIGINT, &new_act, NULL);  
    }

  while (1)
    {
      fprintf(stderr, "\n> ");
      fgets(buffer, 512, stdin);
      if ((buffer[0] != '\n') || 
	  (strlen(buffer) > 1))
	{                            
	  sprintf(response, "(write %s)", buffer);
	  s7_eval_c_string(s7, response);
	}
    }
}

/*
 *    > (do ((i 0 (+ i 1))) ((= i -1)) (format #t "~D " i) (sleep))
 *      ;;; now type C-C to break out of this loop
 *    0 1 2 ^Cinterrupted!
 *      ;;; call the continuation to continue from where we were interrupted
 *    > (*interrupt*)
 *    3 4 5 ^Cinterrupted!
 *    > *interrupt*
 *    #<continuation>
 *    > (+ 1 2)
 *    3
 */

Multidimensional vector element access

#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <stdarg.h>

#include "s7.h"

static s7_pointer multivector_ref(s7_scheme *sc, s7_pointer vector, int indices, ...)
{
  /* multivector_ref returns an element of a multidimensional vector */
  int ndims;
  ndims = s7_vector_rank(vector);

  if (ndims == indices)
    {
      va_list ap;
      s7_Int index = 0;
      va_start(ap, indices);

      if (ndims == 1)
	{
	  index = va_arg(ap, s7_Int);
	  va_end(ap);
	  return(s7_vector_ref(sc, vector, index));
	}
      else
	{
	  int i;
	  s7_pointer *elements;
	  s7_Int *offsets, *dimensions;

	  elements = s7_vector_elements(vector);
	  dimensions = s7_vector_dimensions(vector);
	  offsets = s7_vector_offsets(vector);

	  for (i = 0; i < indices; i++)
	    {
	      int ind;
	      ind = va_arg(ap, int);
	      if ((ind < 0) ||
		  (ind >= dimensions[i]))
		{
		  va_end(ap);
		  return(s7_out_of_range_error(sc, 
                                               "multivector_ref", i, 
                                               s7_make_integer(sc, ind), 
                                               "index should be between 0 and the dimension size"));
		}
	      index += (ind * offsets[i]);
	    }
	  va_end(ap);
	  return(elements[index]);
	}
    }
  return(s7_wrong_number_of_args_error(sc, 
                                       "multivector_ref: wrong number of indices: ~A", 
                                       s7_make_integer(sc, indices)));
}

int main(int argc, char **argv)
{
  char buffer[512];
  char response[1024];
  s7_scheme *s7;

  s7 = s7_init(); 
  s7_eval_c_string(s7, "(define vect (make-vector '(2 3 4) 0))");
  s7_eval_c_string(s7, "(set! (vect 1 1 1) 32)");

  fprintf(stdout, "vect[0,0,0]: %s, vect[1,1,1]: %s\n",
	  s7_object_to_c_string(s7, multivector_ref(s7, s7_name_to_value(s7, "vect"), 3, 0, 0, 0)),
	  s7_object_to_c_string(s7, multivector_ref(s7, s7_name_to_value(s7, "vect"), 3, 1, 1, 1)));
}

/* vect[0,0,0]: 0, vect[1,1,1]: 32
 */

Much later... I decided to add s7_vector_ref_n and s7_vector_set_n to s7.

Notification from Scheme that a given Scheme variable has been set

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "s7.h"

static s7_pointer my_exit(s7_scheme *sc, s7_pointer args) {exit(0);}

static s7_pointer scheme_set_notification(s7_scheme *sc, s7_pointer args)
{
  /* this function is called when the Scheme variable is set! */
  fprintf(stderr, "%s set to %s\n",
	  s7_object_to_c_string(sc, s7_car(args)),
	  s7_object_to_c_string(sc, s7_car(s7_cdr(args))));
  return(s7_car(s7_cdr(args)));
}

int main(int argc, char **argv)
{
  s7_scheme *s7;
  s7 = s7_init();  
  s7_define_function(s7, "exit", my_exit, 0, 0, false, "(exit) exits the program");

  s7_define_function(s7, "notify-C", scheme_set_notification, 2, 0, false, "called if notified-var is set!");
  s7_define_variable(s7, "notified-var", s7_make_integer(s7, 0));
  s7_symbol_set_access(s7,   /* set symbol-access of notified-var to (list #f notify-C #f) */
		       s7_make_symbol(s7, "notified-var"),
		       s7_list(s7, 3, s7_f(s7), s7_name_to_value(s7, "notify-C"), s7_f(s7)));

  if (argc == 2)
    {
      fprintf(stderr, "load %s\n", argv[1]);
      s7_load(s7, argv[1]);
    }
  else
    {
      char buffer[512];
      char response[1024];
      while (1) 
	{
	  fprintf(stdout, "\n> ");
	  fgets(buffer, 512, stdin);
	  
	  if ((buffer[0] != '\n') || 
	      (strlen(buffer) > 1))
	    {                            
	      sprintf(response, "(write %s)", buffer);
	      s7_eval_c_string(s7, response);
	    }
	}
    }
}

/*    > notified-var
 *    0
 *    > (set! notified-var 32)
 *    notified-var set to 32
 *    32
 */

Load C defined stuff into a separate namespace

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "s7.h"

static s7_pointer my_exit(s7_scheme *sc, s7_pointer args) {exit(0);}

static s7_pointer func1(s7_scheme *sc, s7_pointer args)
{
  return(s7_make_integer(sc, s7_integer(s7_car(args)) + 1));
}

int main(int argc, char **argv)
{
  s7_scheme *s7;
  s7_pointer new_env;

  s7 = s7_init();  
  s7_define_function(s7, "exit", my_exit, 0, 0, false, "(exit) exits the program");

  /* "func1" and "var1" will be placed in an anonymous environment,
   *   accessible from Scheme via the global variable "lib-exports"
   */
  
  new_env = s7_augment_environment(s7, s7_current_environment(s7), s7_nil(s7));
  /* make a private environment for func1 and var1 below (this is our "namespace") */
  s7_gc_protect(s7, new_env);

  s7_define(s7, new_env, 
	    s7_make_symbol(s7, "func1"),
	    s7_make_function(s7, "func1", func1, 1, 0, false, "func1 adds 1 to its argument"));
  
  s7_define(s7, new_env, s7_make_symbol(s7, "var1"), s7_make_integer(s7, 32));
  /* those two symbols are now defined in the new environment */

  /* add "lib-exports" to the global environment */
  s7_define_variable(s7, "lib-exports", s7_environment_to_list(s7, new_env));

  if (argc == 2)
    {
      fprintf(stderr, "load %s\n", argv[1]);
      s7_load(s7, argv[1]);
    }
  else
    {
      char buffer[512];
      char response[1024];
      while (1) 
	{
	  fprintf(stdout, "\n> ");
	  fgets(buffer, 512, stdin);
	  
	  if ((buffer[0] != '\n') || 
	      (strlen(buffer) > 1))
	    {                            
	      sprintf(response, "(write %s)", buffer);
	      s7_eval_c_string(s7, response);
	    }
	}
    }
}

/*     > func1
 *     ;func1: unbound variable, line 1
 *     > lib-exports
 *     ((var1 . 32) (func1 . func1))
 *     ;; so lib-exports has the C-defined names and values
 *     ;; we can use these directly:
 *
 *     > (define lib-env (apply augment-environment (current-environment) lib-exports))
 *     lib-env
 *     > (with-environment lib-env (func1 var1))
 *     33
 *
 *     ;; or rename them to prepend "lib:"
 *     > (define lib-env (apply augment-environment 
                                (current-environment) 
                                (map (lambda (binding) 
                                       (cons (string->symbol 
                                               (string-append "lib:" (symbol->string (car binding)))) 
                                             (cdr binding))) 
                                     lib-exports)))
 *     lib-env
 *     > (with-environment lib-env (lib:func1 lib:var1))
 *     33
 *
 *     ;;; now for convenience, place "func1" in the global environment under the name "func2"
 *     > (define func2 (cdadr lib-exports)) 
 *     func2
 *     > (func2 1)  
 *     2
 */

Handle scheme errors in C

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "s7.h"

static s7_pointer my_exit(s7_scheme *sc, s7_pointer args) {exit(0);}

static s7_pointer error_handler(s7_scheme *sc, s7_pointer args)
{
  /* put <<>> around the string so it's obvious who is producing what */

  fprintf(stdout, "<<%s>>", s7_string(s7_car(args)));
  return(s7_make_symbol(sc, "our-error"));
}


int main(int argc, char **argv)
{
  s7_scheme *s7;
  char buffer[512];
  char response[1024];
  bool with_error_hook = false;

  s7 = s7_init();  
  s7_define_function(s7, "exit", my_exit, 0, 0, false, "(exit) exits the program");
  s7_define_function(s7, "error-handler", error_handler, 1, 0, false, "our error handler");

  if (with_error_hook)
    s7_eval_c_string(s7, "(set! (hook-functions *error-hook*)                    \n\
                            (list (lambda (hook)                                 \n\
                                    (error-handler                               \n\
                                      (apply format #f (hook 'data))))))");

  while (1) 
    {
      fprintf(stdout, "\n> ");
      fgets(buffer, 512, stdin);
	  
      if ((buffer[0] != '\n') || 
	  (strlen(buffer) > 1))
	{                            
	  s7_pointer old_port, result;
	  int gc_loc = -1;
	  const char *errmsg = NULL;

	  /* trap error messages */
	  old_port = s7_set_current_error_port(s7, s7_open_output_string(s7));
	  if (old_port != s7_nil(s7))
	    gc_loc = s7_gc_protect(s7, old_port);

	  /* evaluate the input string */
	  result = s7_eval_c_string(s7, buffer);

	  /* print out the value wrapped in "{}" so we can tell it from other IO paths */
	  fprintf(stdout, "{%s}", s7_object_to_c_string(s7, result));

	  /* look for error messages */
	  errmsg = s7_get_output_string(s7, s7_current_error_port(s7));

	  /* if we got something, wrap it in "[]" */
	  if ((errmsg) && (*errmsg))
	    fprintf(stdout, "[%s]", errmsg); 

	  s7_close_output_port(s7, s7_current_error_port(s7));
	  s7_set_current_error_port(s7, old_port);
	  if (gc_loc != -1)
	    s7_gc_unprotect_at(s7, gc_loc);
	}
    }
}

/* 
 *   gcc -c s7.c -I. -g3
 *   gcc -o ex3 ex3.c s7.o -lm -I.
 *
 * if with_error_hook is false,
 *
 *   > (+ 1 2)
 *   {3}
 *   > (+ 1 #\c)
 *   {wrong-type-arg}[
 *   ;+ argument 2, #\c, is character but should be a number, line 1
 *   ]
 *
 * so s7 by default prepends ";" to the error message, and appends "\n",
 *   sending that to current-error-port, and the error type ('wrong-type-arg here)
 *   is returned.
 *
 * if with_error_hook is true,
 *
 *   > (+ 1 2)
 *   {3}
 *   > (+ 1 #\c)
 *   <<+ argument 2, #\c, is character but should be a number>>{our-error}
 *
 * so now the *error-hook* code handles both the error reporting and
 *   the value returned ('our-error in this case).
 */

Closures in C

We often have hooks or callback lists in Scheme, and would like to place a C-defined s7 function on such a list. If the C-side function does not have any state, we can just add its name to the list, but if it is effectively a closure, we have a problem. C itself does not provide closures, so the standard two-step is to include a void* pointer with the C function in a struct, then when that is called, pass the pointer to the function by hand. This obviously does not work if we want that function to be a member of a normal Scheme list of functions. So in the next example, we define a closure in C using s7_make_closure.

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "s7.h"

static s7_pointer closure_func(s7_scheme *sc, s7_pointer args)
{
  /* closure_func is the function portion of our closure.  It assumes its
   *   environment has an integer named "x".  The function also takes one argument,
   *   an integer we'll call it "y".
   */
  return(s7_make_integer(sc,                            /* return (+ y x) */
                         s7_integer(s7_car(args)) +     /*   this is y */
                         s7_integer(s7_name_to_value(sc, "x"))));
}

static s7_pointer define_closure(s7_scheme *sc, const char *name, s7_pointer func, s7_pointer x_value)
{
  /* make_closure creates a new closure with x_value as the local value of x,
   *   and func as the function.  It defines this in Scheme as "name".  
   *
   *   s7_make_closure's arguments are the closure args and body, and
   *   the closure's environment.  For the args, we'll use '(y), and
   *   the body is '(f y));  for the environment, we'll augment 
   *   the current environment with '((x . x_value) (f . func)).
   */
  s7_define(sc, 
	    s7_nil(sc),
	    s7_make_symbol(sc, name),
	    s7_make_closure(sc, 
			    s7_cons(sc,                         /* arg list: '(y) */
			            s7_make_symbol(sc, "y"), s7_nil(sc)), 
			    s7_cons(sc,                         /* body: '(f y) */
				    s7_cons(sc,                              
					    s7_make_symbol(sc, "f"),
					    s7_cons(sc, s7_make_symbol(sc, "y"), s7_nil(sc))),
				    s7_nil(sc)),
			    s7_augment_environment(sc, 
						   s7_current_environment(sc), 
						   s7_cons(sc, 
							   s7_cons(sc,  /* the local binding for "x" */
                                                                   s7_make_symbol(sc, "x"), 
                                                                   x_value), 
							   s7_cons(sc,  /*     and "f" */
								   s7_cons(sc, 
                                                                           s7_make_symbol(sc, "f"), 
                                                                           func), 
								   s7_nil(sc))))));
  return(s7_unspecified(sc));
}

static s7_pointer my_exit(s7_scheme *sc, s7_pointer args) {exit(0);}

int main(int argc, char **argv)
{
  char buffer[512];
  char response[1024];
  s7_pointer c_func;
  s7_scheme *s7;

  s7 = s7_init();  
  s7_define_function(s7, "exit", my_exit, 0, 0, false, "(exit) exits the program");

  c_func = s7_make_function(s7, "#<closure function>", closure_func, 1, 0, false, "function used by define_closure");

  define_closure(s7, "closure-1", c_func, s7_make_integer(s7, 32));  /* (let ((x 32)) (lambda (y) (+ y x))) */
  define_closure(s7, "closure-2", c_func, s7_make_integer(s7, 123)); /* (let ((x 123)) (lambda (y) (+ y x))) */

  while (1) 
    {
      fprintf(stdout, "\n> ");
      fgets(buffer, 512, stdin);
      if ((buffer[0] != '\n') || 
	  (strlen(buffer) > 1))
	{                            
	  sprintf(response, "(write %s)", buffer);
	  s7_eval_c_string(s7, response);
	}
    }
}

/*
 *   > (closure-1 3)
 *   35
 *   > (closure-2 3)
 *   126
 *   > (procedure-source closure-1)
 *   (lambda (y) (f y))
 *   > (environment->list (procedure-environment closure-1))
 *   ((f . #<closure function>) (x . 32))
 */

C and Scheme hooks

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "s7.h"

static s7_pointer my_exit(s7_scheme *sc, s7_pointer args) {exit(0);}

static s7_pointer my_hook_function(s7_scheme *sc, s7_pointer args)
{
  fprintf(stderr, "a is %s\n", s7_object_to_c_string(sc, s7_symbol_local_value(sc, s7_make_symbol(sc, "a"), s7_car(args))));
  return(s7_car(args));
}

int main(int argc, char **argv)
{
  s7_scheme *s7;
  char buffer[512];
  char response[1024];
  s7_pointer test_hook;

  s7 = s7_init();  
  s7_define_function(s7, "exit", my_exit, 0, 0, false, "(exit) exits the program");

  /* define test_hook in C, test-hook in Scheme, arguments are named a and b */
  test_hook = s7_eval_c_string(s7, "(make-hook 'a 'b)");
  s7_define_constant(s7, "test-hook", test_hook); 

  /* add my_hook_function to the test_hook function list */
  s7_hook_set_functions(s7, test_hook, 
			s7_cons(s7, 
				s7_make_function(s7, "my-hook-function", my_hook_function, 1, 0, false, "my hook-function"), 
				s7_hook_functions(s7, test_hook)));
  while (1) 
    {
      fprintf(stdout, "\n> ");
      fgets(buffer, 512, stdin);
	  
      if ((buffer[0] != '\n') || 
	  (strlen(buffer) > 1))
	{                            
	  sprintf(response, "(write %s)", buffer);
	  s7_eval_c_string(s7, response);
	}
    }
}

/* 
 *    > test-hook
 *    #<closure>
 *    
 *    > (hook-functions test-hook)
 *    (my-hook-function)
 *
 *    > (test-hook 1 2)
 *    a is 1
 *    #<unspecified>
 */

Load a shared library

We can use dlopen to load a shared library, and dlsym to initialize that library in our main program. The tricky part is to conjure up the right compiler and loader flags. First we define a module that defines a new s7 function, add-1 that we'll tie into s7 explicitly, and another function that we'll try to call by waving a wand.

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "s7.h"

double a_function(double an_arg);
double a_function(double an_arg)
{
  return(an_arg + 1.0);
}

static s7_pointer add_1(s7_scheme *sc, s7_pointer args) 
{
  return(s7_make_integer(sc, s7_integer(s7_car(args)) + 1)); 
}

void init_ex(s7_scheme *sc);
void init_ex(s7_scheme *sc)  /* this needs to be globally accessible (not "static") */
{
  /* tell s7 about add-1, but leave a_function hidden */
  s7_define_function(sc, "add-1", add_1, 1, 0, false, "(add-1 x) adds 1 to x");
}

And here is our main program:

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "s7.h"

static s7_pointer my_exit(s7_scheme *sc, s7_pointer args) {exit(0);}

#include <dlfcn.h>

static void *library = NULL;

static s7_pointer try(s7_scheme *sc, s7_pointer args)
{
  /* try tries to call an arbitrary function in the shared library */
  void *func;
  func = dlsym(library, s7_string(s7_car(args)));
  if (func)
    {
      /* we'll assume double f(double) */
      typedef double (*dl_func)(double arg);
      return(s7_make_real(sc, ((dl_func)func)(s7_real(s7_cadr(args)))));
    }
  return(s7_error(sc, s7_make_symbol(sc, "can't find function"), 
		  s7_list(sc, 2, s7_make_string(sc, "loader error: ~S"), 
			         s7_make_string(sc, dlerror()))));
}

static s7_pointer cload(s7_scheme *sc, s7_pointer args)
{
  /* cload loads a shared library */
  #define CLOAD_HELP "(cload so-file-name) loads the module"
  library = dlopen(s7_string(s7_car(args)), RTLD_LAZY);
  if (library)
    {
      /* call our init func to define add-1 in s7 */
      void *init_func;
      init_func = dlsym(library, s7_string(s7_car(s7_cdr(args))));
      if (init_func)
	{
	  typedef void *(*dl_func)(s7_scheme *sc);
	  ((dl_func)init_func)(sc);  /* call the initialization function (init_ex above) */
	  return(s7_t(sc));
	}
    }
  return(s7_error(sc, s7_make_symbol(sc, "load-error"), 
		      s7_list(sc, 2, s7_make_string(sc, "loader error: ~S"), 
			             s7_make_string(sc, dlerror()))));
}

int main(int argc, char **argv)
{
  char buffer[512];
  char response[1024];
  s7_scheme *s7;

  s7 = s7_init();  

  s7_define_function(s7, "exit", my_exit, 0, 0, false, "(exit) exits the program");
  s7_define_function(s7, "cload", cload, 2, 0, false, CLOAD_HELP);
  s7_define_function(s7, "try", try, 2, 0, false, 
                         "(try name num) tries to call name in the shared library with the argument num.");

  while (1) 
    {
      fprintf(stdout, "\n> ");
      fgets(buffer, 512, stdin);
	  
      if ((buffer[0] != '\n') || 
	  (strlen(buffer) > 1))
	{                            
	  sprintf(response, "(write %s)", buffer);
	  s7_eval_c_string(s7, response);
	}
    }
}

/* Put the module in the file ex3a.c and the main program in ex3.c, then
 *
 * in Linux:
 *   gcc -c -fPIC ex3a.c
 *   gcc ex3a.o -shared -o ex3a.so
 *   gcc -c s7.c -I. -fPIC -shared
 *   gcc -o ex3 ex3.c s7.o -lm -ldl -I. -Wl,-export-dynamic
 *
 * in Mac OSX:
 *   gcc -c ex3a.c
 *   gcc ex3a.o -o ex3a.so -dynamic -bundle -undefined suppress -flat_namespace
 *   gcc -c s7.c -I. -dynamic -bundle -undefined suppress -flat_namespace
 *   gcc -o ex3 ex3.c s7.o -lm -ldl -I.
 *
 * and run it:
 *   ex3
 *   > (cload "/home/bil/snd-13/ex3a.so" "init_ex")
 *   #t
 *   > (add-1 2)
 *   3
 *   > (try "a_function" 2.5)
 *   3.5
 */

Bignums in C

Bignum support depends on gmp, mpfr, and mpc. In this example, we define "add-1" which adds 1 to any kind of number. The s7_big_* functions return the underlying gmp/mpfr/mpc pointer, so we have to copy that into a new number before adding.

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include <gmp.h>
#include <mpfr.h>
#include <mpc.h>

#include "s7.h"

static s7_pointer my_exit(s7_scheme *sc, s7_pointer args) {exit(0);}

static s7_pointer big_add_1(s7_scheme *sc, s7_pointer args)
{
  /* add 1 to either a normal number or a bignum */
  s7_pointer x;
  x = s7_car(args);
  if (s7_is_bignum(x))
    {
      s7_pointer n;
      if (s7_is_integer(x))
	{
	  mpz_t *big_n;
	  n = s7_make_big_integer(sc, s7_big_integer(x)); /* copy x */
	  big_n = s7_big_integer(n);                      /* get mpz_t pointer of copy */
	  mpz_add_ui(*big_n, *big_n, 1);                  /* add 1 to that */
	  return(n);                                      /* return the new bignum */
	}
      if (s7_is_ratio(x))
	{
	  mpq_t *big_q;
	  mpz_t num, den;
	  n = s7_make_big_ratio(sc, s7_big_ratio(x));
	  big_q = s7_big_ratio(n);
	  mpz_init_set(num, mpq_numref(*big_q));
	  mpz_init_set(den, mpq_denref(*big_q));
	  mpz_add(num, num, den);
	  mpq_set_num(*big_q, num);
	  mpz_clear(num);
	  mpz_clear(den);
	  return(n);
	}
      if (s7_is_real(x))
	{
	  mpfr_t *big_x;
	  n = s7_make_big_real(sc, s7_big_real(x));
	  big_x = s7_big_real(n);
	  mpfr_add_ui(*big_x, *big_x, 1, GMP_RNDN);
	  return(n);
	}
      /* x must be big complex */
      {
	mpc_t *big_z;
	n = s7_make_big_complex(sc, s7_big_complex(x));
	big_z = s7_big_complex(n);
	mpc_add_ui(*big_z, *big_z, 1, MPC_RNDNN);
	return(n);
      }
    }
  else
    {
      if (s7_is_integer(x))
	return(s7_make_integer(sc, 1 + s7_integer(x)));
      if (s7_is_rational(x))
	return(s7_make_ratio(sc, s7_numerator(x) + s7_denominator(x), s7_denominator(x)));
      if (s7_is_real(x))
	return(s7_make_real(sc, 1.0 + s7_real(x)));
      if (s7_is_complex(x))
	return(s7_make_complex(sc, 1.0 + s7_real_part(x), s7_imag_part(x)));
    }
  return(s7_wrong_type_arg_error(sc, "add-1", 0, x, "a number"));
}

int main(int argc, char **argv)
{
  s7_scheme *s7;
  char buffer[512];
  char response[1024];

  s7 = s7_init();  
  s7_define_function(s7, "exit", my_exit, 0, 0, false, "(exit) exits the program");
  s7_define_function(s7, "add-1", big_add_1, 1, 0, false, "(add-1 num) adds 1 to num");

  while (1) 
    {
      fprintf(stdout, "\n> ");
      fgets(buffer, 512, stdin);
      if ((buffer[0] != '\n') || 
	  (strlen(buffer) > 1))
	{                            
	  sprintf(response, "(write %s)", buffer);
	  s7_eval_c_string(s7, response);
	}
    }
}

/* 
 *   gcc -DWITH_GMP=1 -c s7.c -I. -O2 -g3
 *   gcc -DWITH_GMP=1 -o ex2 ex2.c s7.o -I. -O2 -lm -lgmp -lmpfr -lmpc
 *
 *   ex2
 *   > (add-1 1)   
 *   2
 *   > (add-1 2/3)
 *   5/3
 *   > (add-1 1.4) 
 *   2.4
 *   > (add-1 1.5+i)
 *   2.5+1i
 *   > (add-1 (bignum "3"))
 *   4          ; this is the bignum 4
 *   > (add-1 (bignum "3/4"))
 *   7/4
 *   > (add-1 (bignum "1.4"))
 *   2.399999999999999911182158029987476766109E0
 *   > (add-1 (bignum "1.5+i"))
 *   2.500E0+1.000E0i
 */

gtk-based repl

In this example, we use Gtk to make a window with a scrolled text widget, running s7 in a read-eval-print loop. I've tried to make this as short as possible; see Snd for a much more elaborate REPL. From s7's point of view, the only tricky part involves catching errors.

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

/* use Gtk to post a text widget as a REPL.
 */
#include <gtk/gtk.h>

#if HAVE_GTK_3
  #include <gdk/gdk.h>
  #define Return_Key GDK_KEY_Return
#else
  #include <gdk/gdkkeysyms.h>
  #define Return_Key GDK_Return
#endif

#include "s7.h"
static s7_pointer my_exit(s7_scheme *sc, s7_pointer args) {exit(0);}

#define S7_PROMPT "s7> "
#define S7_PROMPT_LENGTH 4

static GtkWidget* repl;                        /* the REPL text widget */
static GtkTextBuffer *repl_buf;                /* its text buffer */
static GtkTextTag *prompt_not_editable = NULL; /* a tag to make sure the prompt can't be erased */

static gint quit_repl(GtkWidget *w, GdkEvent *event, gpointer context)
{
  /* called when we click the 'x' window decoration */
  exit(0);
}

static void evaluate_expression(s7_scheme *sc, char *expression)
{
  /* evaluate expression, display result, catching and displaying any errors */
  if ((expression) &&
      (*expression))
    {
      if ((strlen(expression) > 1) || 
	  (expression[0] != '\n'))
	{
	  const char *errmsg;
	  int gc_loc = -1;
	  s7_pointer old_port, result;
	  GtkTextIter pos;

	  /* open a port to catch error info */
	  old_port = s7_set_current_error_port(sc, s7_open_output_string(sc));
	  if (old_port != s7_nil(sc))
	    gc_loc = s7_gc_protect(sc, old_port);
	  
	  result = s7_eval_c_string(sc, expression);
	  
	  errmsg = s7_get_output_string(sc, s7_current_error_port(sc));
	  /* if error, display it, else display result of evaluation */
	  gtk_text_buffer_get_end_iter(repl_buf, &pos);
	  
	  if ((errmsg) && (*errmsg))
	    gtk_text_buffer_insert(repl_buf, &pos, errmsg, strlen(errmsg));
	  else 
	    {
	      char *result_as_string;
	      result_as_string = s7_object_to_c_string(sc, result);
	      if (result_as_string)
		{
		  gtk_text_buffer_insert(repl_buf, &pos, "\n", 1);
		  gtk_text_buffer_insert(repl_buf, &pos, result_as_string, strlen(result_as_string));
		  free(result_as_string);
		}
	    }
	  
	  s7_close_output_port(sc, s7_current_error_port(sc));
	  s7_set_current_error_port(sc, old_port);
	  if (gc_loc != -1)
	    s7_gc_unprotect_at(sc, gc_loc);
	}
      g_free(expression);
    }
}

static char *get_current_expression(void)
{
  /* find the enclosing expression and return it.  This could be a lot smarter, but
   *    for now, I'll just look back to the previous prompt, then forward for the 
   *    next prompt or the buffer end.
   */
  GtkTextIter pos, previous, next, temp;
  GtkTextMark *m;

  m = gtk_text_buffer_get_insert(repl_buf);
  gtk_text_buffer_get_iter_at_mark(repl_buf, &pos, m);

  if (gtk_text_iter_backward_search(&pos, S7_PROMPT, 0, &temp, &previous, NULL))
    {
      /* previous now marks the end of the previous prompt */
      if (!gtk_text_iter_forward_search(&pos, S7_PROMPT, 0, &next, &temp, NULL))
	{
	  gtk_text_buffer_get_end_iter(repl_buf, &next);
	  /* next now marks the end of the buffer, so there's no complication */
	  return(gtk_text_buffer_get_text(repl_buf, &previous, &next, true));
	}
      /* here next marks the start of the next prompt, but that probably includes
       *   the result printout from an earlier evaluation.  s7_eval_c_string evaluates
       *   all the expressions in its string, returning the value of the last one,
       *   but we want the first.  We'll backup two '\n'.
       */
      gtk_text_iter_backward_search(&next, "\n", 0, &pos, &temp, NULL);     
      gtk_text_iter_backward_search(&pos, "\n", 0, &next, &temp, NULL);      
      return(gtk_text_buffer_get_text(repl_buf, &previous, &next, true));
    }
  return(NULL);
}

static gboolean repl_key_press(GtkWidget *w, GdkEventKey *event, gpointer scheme)
{
  /* called when we type anything in the text widget.
   *   'scheme' is our s7 interpreter.
   */
  guint key;
  key = event->keyval;
  if (key == Return_Key)
    {
      GtkTextIter pos;
      /* get enclosing expression, evaluate it, display result (or error), prompt for next */
      evaluate_expression((s7_scheme *)scheme, get_current_expression());

      /* prompt for next expression */
      gtk_text_buffer_get_end_iter(repl_buf, &pos);
      gtk_text_buffer_insert_with_tags(repl_buf, &pos, 
                                       "\n" S7_PROMPT, 1 + S7_PROMPT_LENGTH, prompt_not_editable, NULL);

      /* make sure the stuff we added is visible */
      gtk_text_buffer_place_cursor(repl_buf, &pos);
      gtk_text_view_scroll_mark_onscreen(GTK_TEXT_VIEW(repl), gtk_text_buffer_get_insert(repl_buf));

      /* tell Gtk that we already sent the '\n' */
      g_signal_stop_emission((gpointer)w, 
                             g_signal_lookup("key_press_event", G_OBJECT_TYPE((gpointer)w)), 0);
    }
  return(false);
}

int main(int argc, char **argv)
{
  s7_scheme *s7;
  GtkWidget *shell, *scrolled_window;
  GtkTextIter pos;

  s7 = s7_init();  
  s7_define_function(s7, "exit", my_exit, 0, 0, false, "(exit) exits the program");

  /* make a window with a scrolled text widget */
  gtk_init(&argc, &argv);
  shell = gtk_window_new(GTK_WINDOW_TOPLEVEL);
  g_signal_connect(G_OBJECT(shell), "delete_event", G_CALLBACK(quit_repl), NULL);

  scrolled_window = gtk_scrolled_window_new(NULL, NULL);
  gtk_scrolled_window_set_policy(GTK_SCROLLED_WINDOW(scrolled_window), 
                                 GTK_POLICY_AUTOMATIC, GTK_POLICY_AUTOMATIC);
  gtk_container_add(GTK_CONTAINER(shell), scrolled_window);

  repl = gtk_text_view_new();
  repl_buf = gtk_text_buffer_new(NULL);
  gtk_container_add(GTK_CONTAINER(scrolled_window), repl);

  gtk_text_view_set_buffer(GTK_TEXT_VIEW(repl), repl_buf);
  gtk_text_view_set_editable(GTK_TEXT_VIEW(repl), true);
  gtk_text_view_set_wrap_mode(GTK_TEXT_VIEW(repl), GTK_WRAP_NONE);
  gtk_text_view_set_cursor_visible(GTK_TEXT_VIEW(repl), true);
  gtk_text_view_set_left_margin(GTK_TEXT_VIEW(repl), 4);

  /* whenever a key is pressed, call repl_key_press */
  gtk_widget_set_events(repl, GDK_ALL_EVENTS_MASK);
  g_signal_connect(G_OBJECT(repl), "key_press_event", G_CALLBACK(repl_key_press), (void *)s7);

  gtk_widget_show(repl);
  gtk_widget_show(scrolled_window);
  gtk_widget_show(shell);

  /* start with a boldface '>' prompt that can't be stepped on */
  prompt_not_editable = gtk_text_buffer_create_tag(repl_buf, "prompt_not_editable", 
						   "editable", false, 
						   "weight", PANGO_WEIGHT_BOLD,
						   NULL);
  gtk_text_buffer_get_end_iter(repl_buf, &pos);
  gtk_text_buffer_insert_with_tags(repl_buf, &pos, 
                                   S7_PROMPT, S7_PROMPT_LENGTH, prompt_not_editable, NULL);

  /* make the initial window size reasonable */
  gdk_window_resize(gtk_widget_get_window(shell), 400, 200);
  gtk_main();
}

/* 
 *   gcc -c s7.c -I. 
 *   gcc -o listener listener.c s7.o -lm -I. <insert all the gtk flags and libraries here>
 */

scheme-based repl (using gtk)

In the same vein, we can use libxm's Gtk/s7 bindings to make the REPL in Scheme. Here is the same code, but now using libxm; first the C side, called gtkex.c:

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include <mus-config.h>
#include "xen.h"
void Init_libxg(void);

static s7_pointer my_exit(s7_scheme *sc, s7_pointer args) {exit(0);}

int main(int argc, char **argv)
{
  s7_scheme *s7;
  s7 = s7_init();             /* start s7 */
  s7_xen_initialize(s7);     /* start the xen connection to libxm */
  s7_define_function(s7, "exit", my_exit, 0, 0, false, "(exit) exits the program");
  Init_libxg();              /* load up all the glib/gdk/gtk/cairo/pango bindings in libxm */
  s7_load(s7, "gtkex.scm");  /* load and run our REPL (below) */ 
}

/*   libxm: configure --with-gtk, then make
 *   gcc -o gtkex gtkex.c libxg.so -lm -I.
 *   ;; you may need to include the libxm directory above
 */

And here is gtkex.scm. The main problem with this code is that Gtk is moving rapidly toward gtk3, and libxm is running along behind panting and complaining. There are several things that may change once Gtk settles down.

(gtk_init 0 #f)

(let ((shell (gtk_window_new GTK_WINDOW_TOPLEVEL))
      (s7-prompt "s7> ")
      (return-key (if (provided? 'gtk3) GDK_KEY_Return GDK_Return)))
  
  (g_signal_connect (G_OBJECT shell) "delete_event"
		    (lambda (window event data)
		      (gtk_main_quit)
		      (exit)))
  (g_signal_connect (G_OBJECT shell) "destroy" 
		    (lambda (window data)
		      (gtk_main_quit)
		      (exit)))
  
  (gtk_window_set_title (GTK_WINDOW shell) "s7")

  (let ((scrolled_window (gtk_scrolled_window_new #f #f)))
    
    (gtk_scrolled_window_set_policy (GTK_SCROLLED_WINDOW scrolled_window) 
                                    GTK_POLICY_AUTOMATIC GTK_POLICY_AUTOMATIC)
    (gtk_container_add (GTK_CONTAINER shell) scrolled_window)
    
    (let* ((repl (gtk_text_view_new))
	   (repl_buf (gtk_text_buffer_new #f))
	   (prompt_not_editable #f))

      (define (evaluate-expression expr)
	(let ((pos (GtkTextIter))
	      (result (catch #t
			     (lambda ()
			       (object->string (eval-string expr)))
			     (lambda args
			       (format #f "~A: ~S" (car args) (apply format #f (cadr args)))))))
	  (gtk_text_buffer_get_end_iter repl_buf pos)
	  (gtk_text_buffer_insert repl_buf pos "\n" 1)
	  (gtk_text_buffer_insert repl_buf pos result (length result))))

      (define (get-current-expression)
	(let ((m (gtk_text_buffer_get_insert repl_buf))
	      (pos (GtkTextIter))
	      (previous (GtkTextIter))
	      (next (GtkTextIter))
	      (temp (GtkTextIter)))
	  (gtk_text_buffer_get_iter_at_mark repl_buf pos m)
	  (if (gtk_text_iter_backward_search pos s7-prompt 0 temp previous #f)
	      (if (not (gtk_text_iter_forward_search pos s7-prompt 0 next temp #f))
		  (begin
		    (gtk_text_buffer_get_end_iter repl_buf next)
		    (gtk_text_buffer_get_text repl_buf previous next #t))
		  (begin
		    (gtk_text_iter_backward_search next "\n" 0 pos temp #f)
		    (gtk_text_iter_backward_search pos "\n" 0 next temp #f)
		    (gtk_text_buffer_get_text repl_buf previous next #t)))
	      "")))

      (define (repl-key-press w event data)
	(let ((key (gtk_event_keyval event)))
	  (if (equal? key return-key)
	      (let ((pos (GtkTextIter)))

		(evaluate-expression (get-current-expression))
		
		(gtk_text_buffer_get_end_iter repl_buf pos)
		(gtk_text_buffer_insert_with_tags repl_buf pos
						  (string-append (string #\newline) s7-prompt) 
						  (+ 1  (length s7-prompt))
						  (list prompt_not_editable))
		(gtk_text_buffer_place_cursor repl_buf pos)
		(gtk_text_view_scroll_mark_onscreen (GTK_TEXT_VIEW repl) 
						    (gtk_text_buffer_get_insert repl_buf))
		(g_signal_stop_emission (GPOINTER w)
					(g_signal_lookup "key_press_event" 
							 (G_OBJECT_TYPE (G_OBJECT w))) 
					0)))))
      
      (gtk_container_add (GTK_CONTAINER scrolled_window) repl)
      (gtk_text_view_set_buffer (GTK_TEXT_VIEW repl) repl_buf)
      (gtk_text_view_set_editable (GTK_TEXT_VIEW repl) #t)
      (gtk_text_view_set_wrap_mode (GTK_TEXT_VIEW repl) GTK_WRAP_NONE)
      (gtk_text_view_set_cursor_visible (GTK_TEXT_VIEW repl) #t)
      (gtk_text_view_set_left_margin (GTK_TEXT_VIEW repl) 4)
      
      (gtk_widget_set_events repl GDK_ALL_EVENTS_MASK)
      (g_signal_connect (G_OBJECT repl) "key_press_event" repl-key-press)

      (gtk_widget_show repl)
      (gtk_widget_show scrolled_window)
      (gtk_widget_show shell)
      
      (set! prompt_not_editable 
	    (gtk_text_buffer_create_tag repl_buf "prompt_not_editable" 
					(list "editable" 0 "weight" PANGO_WEIGHT_BOLD)))
      (let ((pos (GtkTextIter)))
	(gtk_text_buffer_get_end_iter repl_buf pos)
	(gtk_text_buffer_insert_with_tags repl_buf pos 
					  s7-prompt (length s7-prompt)
					  (list prompt_not_editable))
	(gdk_window_resize (gtk_widget_get_window shell) 400 200)
	(gtk_main)))))

;; build gtkex as above, then run it and it will load/run this code

s7 and pthreads

There is one major flaw in the two preceding REPLs; if s7 gets caught in an infinite loop (via (do () ()) for example), you have to kill the main program to stop it. We could use Unix interrupts as in the signal example earlier, but surely in a modern GUI-based program, we'd like to avoid such brute-force stuff. In particular, we want our interface to keep running while we clean up the s7 problem. So, in the next example, we use a separate thread for the s7 evaluation:

#include <stdlib.h>
#include <stdio.h>
#include <string.h>
/* changes from the Gtk/C REPL above are in red */
#include <pthread.h>
#include <gtk/gtk.h>

#if HAVE_GTK_3
  #include <gdk/gdk.h>
  #define Return_Key GDK_KEY_Return
  #define C_Key GDK_KEY_C
  #define c_Key GDK_KEY_c
#else
  #include <gdk/gdkkeysyms.h>
  #define Return_Key GDK_Return
  #define C_Key GDK_C
  #define c_Key GDK_c
#endif

#include "s7.h"
static s7_pointer my_exit(s7_scheme *sc, s7_pointer args) {exit(0);}

#define S7_PROMPT "s7> "
#define S7_PROMPT_LENGTH 4

static GtkWidget* repl;                        /* the REPL text widget */
static GtkTextBuffer *repl_buf;                /* its text buffer */
static GtkTextTag *prompt_not_editable = NULL; /* a tag to make sure the prompt can't be erased */
static GdkCursor *arrow, *spinner;

static gint quit_repl(GtkWidget *w, GdkEvent *event, gpointer context) {exit(0);}

static s7_scheme *s7;
static s7_pointer s7_result;
static bool just_quit = false, s7_is_running = false;
static pthread_t *s7_thread;

static void *run_s7_thread(void *obj)
{
  /* make sure we can stop this thread */
  pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, NULL);
  pthread_setcanceltype(PTHREAD_CANCEL_ASYNCHRONOUS, NULL);

  s7_result = s7_eval_c_string(s7, (const char *)obj);
  s7_is_running = false;
  return(NULL);
}

static void evaluate_expression(s7_scheme *sc, char *expression)
{
  /* evaluate expression, display result, catching and displaying any errors */
  if ((expression) &&
      (*expression))
    {
      if ((strlen(expression) > 1) || 
	  (expression[0] != '\n'))
	{
	  const char *errmsg = NULL;
	  int gc_loc;
	  s7_pointer old_port;
	  GtkTextIter pos;

	  /* open a port to catch error info */
	  old_port = s7_set_current_error_port(sc, s7_open_output_string(sc));
	  gc_loc = s7_gc_protect(sc, old_port);

	  /* evaluate the expression in a separate thread */
	  s7_thread = (pthread_t *)calloc(1, sizeof(pthread_t));
	  just_quit = false;
	  s7_is_running = true;
	  gdk_window_set_cursor(gtk_text_view_get_window(GTK_TEXT_VIEW(repl), GTK_TEXT_WINDOW_TEXT), spinner);

	  pthread_create(s7_thread, NULL, run_s7_thread, (void *)expression);

	  while (s7_is_running)
	    gtk_main_iteration();   /* make sure the GUI is responsive */

	  if (!just_quit)
	    {
	      pthread_join(*s7_thread, NULL);
	      free(s7_thread);
	      s7_thread = NULL;
	      gdk_window_set_cursor(gtk_text_view_get_window(GTK_TEXT_VIEW(repl), GTK_TEXT_WINDOW_TEXT), arrow);
	      errmsg = s7_get_output_string(sc, s7_current_error_port(sc));
	    }
	  else errmsg = "\ngot C-C!";

	  /* if error, display it, else display result of evaluation */
	  gtk_text_buffer_get_end_iter(repl_buf, &pos);
	  
	  if ((errmsg) && (*errmsg))
	    gtk_text_buffer_insert(repl_buf, &pos, errmsg, strlen(errmsg));
	  else 
	    {
	      char *result_as_string;
	      result_as_string = s7_object_to_c_string(sc, s7_result);
	      if (result_as_string)
		{
		  gtk_text_buffer_insert(repl_buf, &pos, "\n", 1);
		  gtk_text_buffer_insert(repl_buf, &pos, result_as_string, strlen(result_as_string));
		  free(result_as_string);
		}
	    }
	  s7_close_output_port(sc, s7_current_error_port(sc));
	  s7_set_current_error_port(sc, old_port);
	  s7_gc_unprotect_at(sc, gc_loc);
	}
      g_free(expression);
    }
}

static char *get_current_expression(void)
{
  GtkTextIter pos, previous, next, temp;
  GtkTextMark *m;

  m = gtk_text_buffer_get_insert(repl_buf);
  gtk_text_buffer_get_iter_at_mark(repl_buf, &pos, m);

  if (gtk_text_iter_backward_search(&pos, S7_PROMPT, 0, &temp, &previous, NULL))
    {
      if (!gtk_text_iter_forward_search(&pos, S7_PROMPT, 0, &next, &temp, NULL))
	{
	  gtk_text_buffer_get_end_iter(repl_buf, &next);
	  return(gtk_text_buffer_get_text(repl_buf, &previous, &next, true));
	}
      gtk_text_iter_backward_search(&next, "\n", 0, &pos, &temp, NULL);     
      gtk_text_iter_backward_search(&pos, "\n", 0, &next, &temp, NULL);      
      return(gtk_text_buffer_get_text(repl_buf, &previous, &next, true));
    }
  return(NULL);
}

static void prompt(GtkWidget *w)
{
  GtkTextIter pos;

  gtk_text_buffer_get_end_iter(repl_buf, &pos);
  gtk_text_buffer_insert_with_tags(repl_buf, &pos, "\n" S7_PROMPT, 1 + S7_PROMPT_LENGTH, prompt_not_editable, NULL);

  gtk_text_buffer_place_cursor(repl_buf, &pos);
  gtk_text_view_scroll_mark_onscreen(GTK_TEXT_VIEW(repl), gtk_text_buffer_get_insert(repl_buf));

  g_signal_stop_emission((gpointer)w, g_signal_lookup("key_press_event", G_OBJECT_TYPE((gpointer)w)), 0);
}

static gboolean repl_key_press(GtkWidget *w, GdkEventKey *event, gpointer scheme)
{
  /* called when you type anything in the text widget.
   */
  guint key, state;
  key = event->keyval;
  state = event->state;

  if (key == Return_Key)
    {
      if (!s7_is_running)
	evaluate_expression((s7_scheme *)scheme, get_current_expression());
      prompt(w);
    }

  /* if C-C typed, and s7 is running, we're trying to break out of a loop */
  if (((key == C_Key) || (key == c_Key)) &&
      ((state & GDK_CONTROL_MASK) != 0) &&
      (s7_is_running))
    {
      just_quit = true;
      pthread_cancel(*s7_thread);
      pthread_detach(*s7_thread);
      s7_is_running = false;
      s7_quit(s7);
      free(s7_thread);
      s7_thread = NULL;
      gdk_window_set_cursor(gtk_text_view_get_window(GTK_TEXT_VIEW(repl), GTK_TEXT_WINDOW_TEXT), arrow);
      prompt(w);
    }

  return(false);
}

int main(int argc, char **argv)
{
  GtkWidget *shell, *scrolled_window;
  GtkTextIter pos;

  s7 = s7_init();  
  s7_define_function(s7, "exit", my_exit, 0, 0, false, "(exit) exits the program");

  /* make a window with a scrolled text widget */
  gtk_init(&argc, &argv);
  
  spinner = gdk_cursor_new(GDK_WATCH);
  arrow = gdk_cursor_new(GDK_LEFT_PTR);
  
  shell = gtk_window_new(GTK_WINDOW_TOPLEVEL);
  g_signal_connect(G_OBJECT(shell), "delete_event", G_CALLBACK(quit_repl), NULL);

  scrolled_window = gtk_scrolled_window_new(NULL, NULL);
  gtk_scrolled_window_set_policy(GTK_SCROLLED_WINDOW(scrolled_window), GTK_POLICY_AUTOMATIC, GTK_POLICY_AUTOMATIC);
  gtk_container_add(GTK_CONTAINER(shell), scrolled_window);

  repl = gtk_text_view_new();
  repl_buf = gtk_text_buffer_new(NULL);
  gtk_container_add(GTK_CONTAINER(scrolled_window), repl);

  gtk_text_view_set_buffer(GTK_TEXT_VIEW(repl), repl_buf);
  gtk_text_view_set_editable(GTK_TEXT_VIEW(repl), true);
  gtk_text_view_set_wrap_mode(GTK_TEXT_VIEW(repl), GTK_WRAP_NONE);
  gtk_text_view_set_cursor_visible(GTK_TEXT_VIEW(repl), true);
  gtk_text_view_set_left_margin(GTK_TEXT_VIEW(repl), 4);

  /* whenever a key is pressed, call repl_key_press */
  gtk_widget_set_events(repl, GDK_ALL_EVENTS_MASK);
  g_signal_connect(G_OBJECT(repl), "key_press_event", G_CALLBACK(repl_key_press), (void *)s7);

  gtk_widget_show(repl);
  gtk_widget_show(scrolled_window);
  gtk_widget_show(shell);

  /* start with a boldface '>' prompt that can't be stepped on */
  prompt_not_editable = gtk_text_buffer_create_tag(repl_buf, "prompt_not_editable", 
						   "editable", false, 
						   "weight", PANGO_WEIGHT_BOLD,
						   NULL);
  gtk_text_buffer_get_end_iter(repl_buf, &pos);
  gtk_text_buffer_insert_with_tags(repl_buf, &pos, S7_PROMPT, S7_PROMPT_LENGTH, prompt_not_editable, NULL);

  /* make the initial window size reasonable */
  gdk_window_resize(gtk_widget_get_window(shell), 400, 200);
  gtk_main();
}

This code still has a flaw. Any GUI callback can call s7_call or s7_eval_c_string directly. In the code above, if the listener starts a long evaluation (or falls into an infinite loop), and we start idly poking around with the mouse, waiting for the evaluation to finish, a mouse or menu callback can trigger a call on the same s7 interpreter that is currently running in the listener. We are trying to evaluate two different expressions at the same time! What you get depends on where s7 notices the problem. If the evaluator senses that it is lost, it will raise some confusing error; otherwise the s7 stack will cause a segfault. We need the GUI to stay active so that we can interrupt computation by typing C-C. We can't sort through all the events dispatching only the C-C, because that confuses subsequent GUI actions, and the whole idea here is to keep the interface alive. We can put the listener's s7_eval_c_string in a separate s7 thread so that the multiple evaluations can be interleaved, but those evaluations can still make calls into Gtk or Motif, and unfortunately, neither GUI toolkit is thread-safe. If two threads invoke a GUI function, you get a segfault or some other bizarre screwup. In Gtk, we have to wrap gdk_threads_enter and gdk_threads_leave around any code that calls into Gtk. We'd set it up by adding

  g_thread_init(NULL);
  gdk_threads_init();
  gdk_threads_enter();

at the very start of our program, and gdk_threads_leave() at the very end. Then whenever an s7 call might involve Gtk, we wrap it within gdk_threads_enter() and gdk_threads_leave(). We can't just wrap the outer s7_eval_c_string call above, because that locks out the keyboard, and the whole point was to catch C-C while in the s7 evaluation. But in Snd, where both normal functions and gtk callbacks can trigger arbitrary Scheme code, it would be completely unreasonable to wrap these thread handlers around every bit of code that might touch the user interface (consider the xg module!). So...

begin-hook

Common Lisp has something called "evalhook" that makes it possible to insert your own function into eval. In s7, we have a "begin_hook" which sits at the opening of any begin block (implicit or explicit). Here are two REPLs, one for Gtk, and one for a bare terminal.

/* terminal-based REPL, 
 *    an expansion of the read-eval-print loop program above.
 * type C-g to interrupt an evaluation.
 */
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <termios.h>
#include <signal.h>

#include "s7.h"
static s7_pointer my_exit(s7_scheme *sc, s7_pointer args) {exit(0);}

static struct termios save_buf, buf;

static void sigcatch(int n)
{
  /* put things back the way they were */
  tcsetattr(fileno(stdin), TCSAFLUSH, &save_buf);
  exit(0);
}

static char buffer[512];
static int type_ahead_point = 0;

static bool watch_for_c_g(s7_scheme *sc)
{
  char c;
  bool all_done = false;
  /* watch for C-g without blocking, save other chars as type-ahead */
  tcsetattr(fileno(stdin), TCSAFLUSH, &buf);

  if (read(fileno(stdin), &c, 1) == 1)
    {
      if (c == 7) /* C-g */
	{
	  all_done = true;
	  type_ahead_point = 0;
	}
      else buffer[type_ahead_point++] = c;
    }

  tcsetattr(fileno(stdin), TCSAFLUSH, &save_buf);
  return(all_done);
}

int main(int argc, char **argv)
{
  s7_scheme *s7;
  bool use_begin_hook;

  use_begin_hook = (tcgetattr(fileno(stdin), &save_buf) >= 0);
  if (use_begin_hook)
    {
      buf = save_buf;
      buf.c_lflag &= ~ICANON;
      buf.c_cc[VMIN] = 0;         /* if no chars waiting, just return */
      buf.c_cc[VTIME] = 0;

      signal(SIGINT, sigcatch);
      signal(SIGQUIT, sigcatch);
      signal(SIGTERM, sigcatch);
    }

  s7 = s7_init();  
  s7_define_function(s7, "exit", my_exit, 0, 0, false, "(exit) exits the program");

  if (argc == 2)
    {
      fprintf(stderr, "load %s\n", argv[1]);
      s7_load(s7, argv[1]);
    }
  else
    {
      char response[1024];
      while (1) 
	{
	  fprintf(stdout, "\n> ");
	  fgets((char *)(buffer + type_ahead_point), 512 - type_ahead_point, stdin);
	  type_ahead_point = 0;

	  if ((buffer[0] != '\n') || 
	      (strlen(buffer) > 1))
	    {                            
	      sprintf(response, "(write %s)", buffer);

	      if (use_begin_hook)
		s7_set_begin_hook(s7, watch_for_c_g);
	      s7_eval_c_string(s7, response);
	      if (use_begin_hook)
		s7_set_begin_hook(s7, NULL);
	    }
	}
    }
  if (use_begin_hook)
    tcsetattr(fileno(stdin), TCSAFLUSH, &save_buf);
}
/* Gtk-based REPL using s7_begin_hook 
 */
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <gtk/gtk.h>

#if HAVE_GTK_3
  #include <gdk/gdk.h>
  #define Return_Key GDK_KEY_Return
  #define C_Key GDK_KEY_C
  #define c_Key GDK_KEY_c
#else
  #include <gdk/gdkkeysyms.h>
  #define Return_Key GDK_Return
  #define C_Key GDK_C
  #define c_Key GDK_c
#endif

#include "s7.h"
static s7_pointer my_exit(s7_scheme *sc, s7_pointer args) {exit(0);}

#define S7_PROMPT "s7> "
#define S7_PROMPT_LENGTH 4
#define S7_INTERRUPTED_MESSAGE "\ns7 interrupted!"

static GtkWidget* repl;                        /* the REPL text widget */
static GtkTextBuffer *repl_buf;                /* its text buffer */
static GtkTextTag *prompt_not_editable = NULL; /* a tag to make sure the prompt can't be erased */
static GdkCursor *arrow, *spinner;

static gint quit_repl(GtkWidget *w, GdkEvent *event, gpointer context) {exit(0);}

static bool C_c_typed = false;

static bool begin_hook(s7_scheme *sc)
{
  C_c_typed = false;
  if (gtk_events_pending()) /* this is very slow — see snd-listener.c for a work-around */
    gtk_main_iteration();
  return(C_c_typed);
}

static void evaluate_expression(s7_scheme *sc, char *expression)
{
  /* evaluate expression, display result, catching and displaying any errors */
  if ((expression) &&
      (*expression))
    {
      if ((strlen(expression) > 1) || 
	  (expression[0] != '\n'))
	{
	  const char *errmsg = NULL;
	  int gc_loc;
	  s7_pointer old_port, result;
	  GtkTextIter pos;

	  old_port = s7_set_current_error_port(sc, s7_open_output_string(sc));
	  gc_loc = s7_gc_protect(sc, old_port);
	  gdk_window_set_cursor(gtk_text_view_get_window(GTK_TEXT_VIEW(repl), GTK_TEXT_WINDOW_TEXT), spinner);
	  
	  s7_set_begin_hook(sc, begin_hook);
	  result = s7_eval_c_string(sc, expression);
	  s7_set_begin_hook(sc, NULL);

	  gdk_window_set_cursor(gtk_text_view_get_window(GTK_TEXT_VIEW(repl), GTK_TEXT_WINDOW_TEXT), arrow);
	  gtk_text_buffer_get_end_iter(repl_buf, &pos);

	  if (C_c_typed)
	    {
	      gtk_text_buffer_insert(repl_buf, &pos, S7_INTERRUPTED_MESSAGE, strlen(S7_INTERRUPTED_MESSAGE));
	      C_c_typed = false;
	    }
	  else
	    {
	      errmsg = s7_get_output_string(sc, s7_current_error_port(sc));
	      if ((errmsg) && (*errmsg))
		gtk_text_buffer_insert(repl_buf, &pos, errmsg, strlen(errmsg));
	      else 
		{
		  char *result_as_string;
		  result_as_string = s7_object_to_c_string(sc, result);
		  if (result_as_string)
		    {
		      gtk_text_buffer_insert(repl_buf, &pos, "\n", 1);
		      gtk_text_buffer_insert(repl_buf, &pos, result_as_string, strlen(result_as_string));
		      free(result_as_string);
		    }
		}
	    }

	  s7_close_output_port(sc, s7_current_error_port(sc));
	  s7_set_current_error_port(sc, old_port);
	  s7_gc_unprotect_at(sc, gc_loc);
	}

      g_free(expression);
    }
}

static char *get_current_expression(void)
{
  GtkTextIter pos, previous, next, temp;
  GtkTextMark *m;
  m = gtk_text_buffer_get_insert(repl_buf);
  gtk_text_buffer_get_iter_at_mark(repl_buf, &pos, m);
  if (gtk_text_iter_backward_search(&pos, S7_PROMPT, 0, &temp, &previous, NULL))
    {
      if (!gtk_text_iter_forward_search(&pos, S7_PROMPT, 0, &next, &temp, NULL))
	{
	  gtk_text_buffer_get_end_iter(repl_buf, &next);
	  return(gtk_text_buffer_get_text(repl_buf, &previous, &next, true));
	}
      gtk_text_iter_backward_search(&next, "\n", 0, &pos, &temp, NULL);     
      gtk_text_iter_backward_search(&pos, "\n", 0, &next, &temp, NULL);      
      return(gtk_text_buffer_get_text(repl_buf, &previous, &next, true));
    }
  return(NULL);
}

static void prompt(GtkWidget *w)
{
  GtkTextIter pos;
  gtk_text_buffer_get_end_iter(repl_buf, &pos);
  gtk_text_buffer_insert_with_tags(repl_buf, &pos, "\n" S7_PROMPT, 1 + S7_PROMPT_LENGTH, prompt_not_editable, NULL);
  gtk_text_buffer_place_cursor(repl_buf, &pos);
  gtk_text_view_scroll_mark_onscreen(GTK_TEXT_VIEW(repl), gtk_text_buffer_get_insert(repl_buf));
  g_signal_stop_emission((gpointer)w, g_signal_lookup("key_press_event", G_OBJECT_TYPE((gpointer)w)), 0);
}

static gboolean repl_key_press(GtkWidget *w, GdkEventKey *event, gpointer scheme)
{
  /* called when you type anything in the text widget.
   */
  guint key, state;
  s7_scheme *sc;

  sc = (s7_scheme *)scheme;
  key = event->keyval;
  state = event->state;

  if (key == Return_Key)
    {
      if (s7_begin_hook(sc) == NULL)
	{
	  evaluate_expression(sc, get_current_expression());
	  prompt(w);
	}
    }

  if (((key == C_Key) || (key == c_Key)) &&
      ((state & GDK_CONTROL_MASK) != 0))
    C_c_typed = true;

  return(false);
}

int main(int argc, char **argv)
{
  GtkWidget *shell, *scrolled_window;
  GtkTextIter pos;
  s7_scheme *s7;

  s7 = s7_init();  
  s7_define_function(s7, "exit", my_exit, 0, 0, false, "(exit) exits the program");

  gtk_init(&argc, &argv);
  spinner = gdk_cursor_new(GDK_WATCH);
  arrow = gdk_cursor_new(GDK_LEFT_PTR);
  
  shell = gtk_window_new(GTK_WINDOW_TOPLEVEL);
  g_signal_connect(G_OBJECT(shell), "delete_event", G_CALLBACK(quit_repl), NULL);

  scrolled_window = gtk_scrolled_window_new(NULL, NULL);
  gtk_scrolled_window_set_policy(GTK_SCROLLED_WINDOW(scrolled_window), GTK_POLICY_AUTOMATIC, GTK_POLICY_AUTOMATIC);
  gtk_container_add(GTK_CONTAINER(shell), scrolled_window);

  repl = gtk_text_view_new();
  repl_buf = gtk_text_buffer_new(NULL);
  gtk_container_add(GTK_CONTAINER(scrolled_window), repl);

  gtk_text_view_set_buffer(GTK_TEXT_VIEW(repl), repl_buf);
  gtk_text_view_set_editable(GTK_TEXT_VIEW(repl), true);
  gtk_text_view_set_wrap_mode(GTK_TEXT_VIEW(repl), GTK_WRAP_NONE);
  gtk_text_view_set_cursor_visible(GTK_TEXT_VIEW(repl), true);
  gtk_text_view_set_left_margin(GTK_TEXT_VIEW(repl), 4);
  gtk_widget_set_events(repl, GDK_ALL_EVENTS_MASK);
  g_signal_connect(G_OBJECT(repl), "key_press_event", G_CALLBACK(repl_key_press), (void *)s7);

  gtk_widget_show(repl);
  gtk_widget_show(scrolled_window);
  gtk_widget_show(shell);

  prompt_not_editable = gtk_text_buffer_create_tag(repl_buf, "prompt_not_editable", 
						   "editable", false, 
						   "weight", PANGO_WEIGHT_BOLD,
						   NULL);
  gtk_text_buffer_get_end_iter(repl_buf, &pos);
  gtk_text_buffer_insert_with_tags(repl_buf, &pos, S7_PROMPT, S7_PROMPT_LENGTH, prompt_not_editable, NULL);
  gdk_window_resize(gtk_widget_get_window(shell), 400, 200);
  gtk_main();
}

In the Gtk case above, begin_hook itself checks the user-interface (via gtk_main_iteration), so a key_press event (or any other) can still get through, even if s7_eval_c_string falls into an infinite loop. If a GUI event triggers a call on s7, that call happens in the current thread at a protected point in the s7 evaluation, so the GUI toolkit is happy because there's only the current thread, and s7 is happy because it handles the interleaved expression at a point where it can't confuse the current evaluation. If the begin_hook function returns true (if C-C is typed), s7 itself calls s7_quit, interrupting the current evaluation. If return is typed, and s7 is running (begin_hook is not NULL), we ignore it. Otherwise return causes our REPL to send the current string to s7_eval_c_string. Upon returning from s7_eval_c_string, we check whether we were interrupted (C_c_typed is true), and if so, send a surprised message to the REPL. Otherwise we handle everything as in the previous REPLs.

If the s7 evaluation can't affect the GUI and the GUI can't call s7 itself, then you don't need to use begin_hook.

begin_hook provides a way to call arbitrary C or Scheme code at any time; the mind boggles! In the Snd listener, for example, you can ask for a stack trace to see if your long computation is making progress. Since the normal begin_hook function keeps the interface alive, a callback can save the current begin_hook function, insert its own function to do something useful, then that function can replace itself with the original:

static bool (*current_begin_hook)(s7_scheme *sc);        /* this will be the saved original function */

static bool stacktrace_begin_hook(s7_scheme *sc)         /* this is our one-shot replacement */
{
  fprintf(stderr, "%s\n", s7_string(s7_stacktrace(sc))); /* it prints local var info */
  s7_set_begin_hook(s7, current_begin_hook);             /*   then replaces itself with the original */
  return(false);
}

static void listener_stacktrace_callback(...)            /* the usual callback boilerplate here */
{
  current_begin_hook = s7_begin_hook(s7);                /* save current begin_hook function */
  if (current_begin_hook)                                /* s7 is running, so... */
    s7_set_begin_hook(s7, stacktrace_begin_hook);        /*   insert our stacktrace function in its place */
}

GSL

The next example provides access to the GSL functions gsl_sf_bessel_J0 and gsl_sf_bessel_J0_e. In the latter case, we return multiple values (in the Scheme sense): the J0 value and an error estimate. When using GSL, it is important to replace the default gsl error handler — it simply aborts the main program upon error!

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include <gsl/gsl_sf.h>
#include <gsl/gsl_errno.h>

#include "s7.h"
static s7_scheme *s7;

static void g_gsl_error(const char *reason, const char *file, int line, int gsl_errno)
{ 
  /* this is our replacement error handler */
  s7_error(s7, s7_make_symbol(s7, "gsl-error"),
	   s7_list(s7, 5, 
		   s7_make_string(s7, "GSL: ~A, ~A in ~A line ~A"),
		   s7_make_string(s7, gsl_strerror(gsl_errno)),
		   s7_make_string(s7, reason),
		   s7_make_string(s7, file),
		   s7_make_integer(s7, line)));
}

static s7_pointer g_J0(s7_scheme *sc, s7_pointer args)
{
  #define J0_help "(J0 x) returns the bessel function value J0(x)"
  s7_pointer x;
  x = s7_car(args);
  if (!s7_is_real(x))
    return(s7_wrong_type_arg_error(sc, "J0", 0, x, "a real number"));
  return(s7_make_real(sc, gsl_sf_bessel_J0(s7_number_to_real(x))));
}

static s7_pointer g_J0_e(s7_scheme *sc, s7_pointer args)
{
  #define J0_e_help "(J0_e x) returns the bessel function value J0(x) and an error estimate"
  s7_pointer x;
  double jx;
  gsl_sf_result r;     /* r.val and r.err are returned */
  int err;

  x = s7_car(args);
  if (!s7_is_real(x))
    return(s7_wrong_type_arg_error(sc, "J0", 0, x, "a real number"));

  err = gsl_sf_bessel_J0_e(s7_number_to_real(x), &r);
  if (err != GSL_SUCCESS)
    return(s7_error(sc, s7_make_symbol(sc, "gsl-error"),
		    s7_list(sc, 2,
			    s7_make_string(sc, "GSL: ~A"),
			    s7_make_string(sc, gsl_strerror(err)))));

  return(s7_values(sc, 2, s7_make_real(sc, r.val), s7_make_real(sc, r.err)));
}

static s7_pointer g_exit(s7_scheme *sc, s7_pointer args)
{
  #define exit_help "(exit) exits the program"
  exit(0);
}

int main(int argc, char **argv)
{
  gsl_set_error_handler(g_gsl_error); /* use our error handler */

  s7 = s7_init();  
  s7_define_function(s7, "exit", g_exit, 0, 0, false, exit_help);

  s7_define_function(s7, "J0", g_J0, 1, 0, false, J0_help);
  s7_define_function(s7, "J0_e", g_J0_e, 1, 0, false, J0_e_help);

  /* and now the usual repl */
  if (argc == 2)
    {
      fprintf(stderr, "load %s\n", argv[1]);
      s7_load(s7, argv[1]);
    }
  else
    {
      char buffer[512];
      char response[1024];
      while (1) 
	{
	  fprintf(stdout, "\n> ");
	  fgets(buffer, 512, stdin);
	  
	  if ((buffer[0] != '\n') || 
	      (strlen(buffer) > 1))
	    {                            
	      sprintf(response, "(write %s)", buffer);
	      s7_eval_c_string(s7, response);
	    }
	}
    }
}

/* 
 *  gcc -c s7.c -I.
 *  gcc -o ex14 ex14.c s7.o -I. -lm -lgsl -lgslcblas
 *
 *  > (J0 1.0)
 *  0.76519768655797
 *  > (J0 0.0)
 *  1.0
 *  > (list (J0_e 1.0))
 *  (0.76519768655797 6.7261301656723e-16)
 */
             

Here is a sketch of a connection to the GSL linear algebra functions, using objects that look like vectors in s7. An object returned by s7_make_object can have an environment, accessible in scheme via object-environment. It can hold whatever methods the object responds to. Below, we implement the basic generic functions via s7_new_type_x, and vector? via an open_environment. The rest of the vector functions could be added in that section.

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include <gsl/gsl_math.h>
#include <gsl/gsl_errno.h>
#include <gsl/gsl_linalg.h>

#include "s7.h"
static s7_scheme *s7;

static void g_gsl_error(const char *reason, const char *file, int line, int gsl_errno)
{
  s7_error(s7, s7_make_symbol(s7, "gsl-error"),
	   s7_list(s7, 5, 
		   s7_make_string(s7, "GSL: ~A, ~A in ~A line ~A"),
		   s7_make_string(s7, gsl_strerror(gsl_errno)),
		   s7_make_string(s7, reason),
		   s7_make_string(s7, file),
		   s7_make_integer(s7, line)));
}

static s7_pointer g_exit(s7_scheme *sc, s7_pointer args)
{
  #define exit_help "(exit) exits the program"
  exit(0);
}

typedef struct {
  size_t size;
  double *data;
} g_block;    

static int g_block_type = 0;
static s7_pointer g_block_methods;

static s7_pointer g_make_block(s7_scheme *sc, s7_pointer args)
{
  #define g_make_block_help "(make-block size) returns a new block of the given size"
  g_block *g;
  s7_pointer new_g;
  g = (g_block *)calloc(1, sizeof(g_block));
  g->size = (size_t)s7_integer(s7_car(args));
  g->data = (double *)calloc(g->size, sizeof(double));
  new_g = s7_make_object(sc, g_block_type, (void *)g);
  s7_object_set_environment(new_g, g_block_methods);
  s7_open_environment(new_g);
  return(new_g);
}

static s7_pointer g_to_block(s7_scheme *sc, s7_pointer args)
{
  #define g_block_help "(block ...) returns a block object with the arguments as its contents."
  s7_pointer p, b;
  size_t i, len;
  g_block *gb;
  len = s7_list_length(sc, args);
  b = g_make_block(sc, s7_cons(sc, s7_make_integer(sc, len), s7_nil(sc)));
  gb = (g_block *)s7_object_value(b);
  for (i = 0, p = args; i < len; i++, p = s7_cdr(p))
    gb->data[i] = s7_number_to_real(s7_car(p));
  return(b);
}

static char *g_block_display(s7_scheme *sc, void *value)
{
  return(strdup("#<block>"));
}

static void g_block_free(void *value)
{
  g_block *g = (g_block *)value;
  free(g->data);
  free(g);
}

static bool g_block_is_equal(void *val1, void *val2)
{
  return(val1 == val2);
}

static void g_block_mark(void *val)
{
  /* nothing to mark */
}

static s7_pointer g_block_ref(s7_scheme *sc, s7_pointer obj, s7_pointer args)
{
  g_block *g = (g_block *)s7_object_value(obj);
  size_t index;
  index = (size_t)s7_integer(s7_car(args));
  if ((index >= 0) && (index < g->size))
    return(s7_make_real(sc, g->data[index]));
  return(s7_out_of_range_error(sc, "block-ref", 2, s7_car(args), "should be less than block length"));
}

static s7_pointer g_block_set(s7_scheme *sc, s7_pointer obj, s7_pointer args)
{
  g_block *g = (g_block *)s7_object_value(obj);
  s7_Int index;
  index = s7_integer(s7_car(args));
  if ((index >= 0) && (index < g->size))
    {
      g->data[index] = s7_number_to_real(s7_cadr(args));
      return(s7_cadr(args));
    }
  return(s7_out_of_range_error(sc, "block-set", 2, s7_car(args), "should be less than block length"));
}

static s7_pointer g_block_length(s7_scheme *sc, s7_pointer obj)
{
  g_block *g = (g_block *)s7_object_value(obj);
  return(s7_make_integer(sc, g->size));
}

static s7_pointer g_block_copy(s7_scheme *sc, s7_pointer obj)
{
  g_block *g = (g_block *)s7_object_value(obj);
  g_block *g1;
  s7_pointer new_g;
  new_g = g_make_block(sc, s7_cons(sc, s7_make_integer(sc, g->size), s7_nil(sc)));
  g1 = (g_block *)s7_object_value(new_g);
  memcpy((void *)(g1->data), (void *)(g->data), g->size * sizeof(double));
  return(new_g);
}

static s7_pointer g_block_reverse(s7_scheme *sc, s7_pointer obj)
{
  size_t i, j;
  g_block *g = (g_block *)s7_object_value(obj);
  g_block *g1;
  s7_pointer new_g;
  new_g = g_make_block(sc, s7_cons(sc, s7_make_integer(sc, g->size), s7_nil(sc)));
  g1 = (g_block *)s7_object_value(new_g);
  for (i = 0, j = g->size - 1; i < g->size; i++, j--)
    g1->data[i] = g->data[j];
  return(new_g);
}

static s7_pointer g_block_fill(s7_scheme *sc, s7_pointer obj, s7_pointer args)
{
  size_t i;
  double fill_val;
  g_block *g = (g_block *)s7_object_value(obj);
  fill_val = s7_number_to_real(s7_car(args));
  for (i = 0; i < g->size; i++)
    g->data[i] = fill_val;
  return(obj);
}


/* this example comes from the GSL documentation: doc/examples/linalgu.c */

static s7_pointer g_solve(s7_scheme *sc, s7_pointer args)
{
  #define g_solve_help "(solve A b) solves Ax = b, A is a matrix, b a vector, solve returns the vector x."
  gsl_matrix_view m;
  gsl_vector_view b;
  gsl_vector *x; 
  gsl_permutation *p;
  int s;
  size_t i, len;
  s7_pointer result;
  g_block *gx, *gA, *gb;

  gA = (g_block *)s7_object_value(s7_car(args));
  gb = (g_block *)s7_object_value(s7_cadr(args));
  len = gb->size;

  m = gsl_matrix_view_array(gA->data, len, len);
  b = gsl_vector_view_array(gb->data, len);
  x = gsl_vector_alloc(len);
  p = gsl_permutation_alloc(len);

  gsl_linalg_LU_decomp (&m.matrix, p, &s);
  gsl_linalg_LU_solve (&m.matrix, p, &b.vector, x);

  result = g_make_block(sc, s7_cons(sc, s7_make_integer(sc, len), s7_nil(sc)));
  gx = (g_block *)s7_object_value(result);
  for (i = 0; i < len; i++)
    gx->data[i] = gsl_vector_get(x, i);

  gsl_permutation_free (p);
  gsl_vector_free(x);

  return(result);
}


int main(int argc, char **argv)
{
  gsl_set_error_handler(g_gsl_error);

  s7 = s7_init();  
  s7_define_function(s7, "exit", g_exit, 0, 0, false, exit_help);

  g_block_type = s7_new_type_x("#<block>", 
			       g_block_display, g_block_free, 
			       g_block_is_equal, g_block_mark,
			       g_block_ref, g_block_set, g_block_length, 
			       g_block_copy, g_block_reverse, g_block_fill);

  s7_define_function(s7, "make-block", g_make_block, 1, 0, false, g_make_block_help);
  s7_define_function(s7, "block", g_to_block, 0, 0, true, g_block_help);

  g_block_methods = s7_eval_c_string(s7, "(environment (cons 'vector? (lambda (p) #t)))");
  /* here load up all the methods ... */
  s7_gc_protect(s7, g_block_methods);

  s7_define_function(s7, "solve", g_solve, 2, 0, false, g_solve_help);

  if (argc == 2)
    {
      fprintf(stderr, "load %s\n", argv[1]);
      s7_load(s7, argv[1]);
    }
  else
    {
      char buffer[512];
      char response[1024];
      while (1) 
	{
	  fprintf(stdout, "\n> ");
	  fgets(buffer, 512, stdin);
	  
	  if ((buffer[0] != '\n') || 
	      (strlen(buffer) > 1))
	    {                            
	      sprintf(response, "(write %s)", buffer);
	      s7_eval_c_string(s7, response);
	    }
	}
    }
}

/* gcc -c s7.c -I.
 * gcc -o ex14 ex14.c s7.o -I. -lm -lgsl -lgslcblas
 *
 *  > (define g (make-block 32))
 *  g
 *  > (length g)
 *  32
 *  > (g 2)
 *  0.0
 *  > (set! (g 2) 1.5)
 *  1.5
 *  > (g 2)
 *  1.5
 *  > g         
 *  #<block>
 *  > (vector? g)
 *  #t
 *  > (define b (block 1.0 2.0 3.0 4.0))
 *  b
 *  > (define A (block 0.18 0.60 0.57 0.96  
 *                     0.41 0.24 0.99 0.58  
 *                     0.14 0.30 0.97 0.66  
 *                     0.51 0.13 0.19 0.85))
 *  A
 *  (define x (solve A b))
 *  > (format #f "~{~A~^ ~}" x)
 *  "-4.052050229574 -12.605611395907 1.6609116267088 8.6937669287952"
 */

I have toyed with adding a set of GSL bindings, along the lines of the current Motif/Gtk/OpenGL bindings in Snd, but I think the situation here is different. If you look at the examples in the GSL doc/examples directory, or snd-xen.c in Snd, it's evident that GSL users will want to write their own C code, then bind that into Scheme (blithely assuming they would rather use Scheme than Matlab!). If we provide GSL-level bindings, the Scheme code becomes a forest of opaque C pointers, and the (human) reader would have to know how the C stuff fits together. The other libraries work out better in this regard because they don't expose internal structs and types unnecessarily.