X-Git-Url: http://git.lukelau.me/?p=scheme.git;a=blobdiff_plain;f=typecheck.scm;h=821035230affddf4d813074316f5aecaeaa2efc5;hp=219b010202a7f7a0127847e8372ea7e064a05efb;hb=844dcd052c6f551d9936693c2b4c49cf920c7051;hpb=e561cf1ba9b2986a22cb08250247e402ac2a5871

diff --git a/typecheck.scm b/typecheck.scm
index 219b010..8210352 100644
--- a/typecheck.scm
+++ b/typecheck.scm
@@ -4,14 +4,13 @@
   (and (list? t) (eq? (car t) 'abs)))
 
 (define (tvar? t)
-  (and (not (list? t)) (not (concrete? t)) (symbol? t)))
+  (and (not (list? t))
+       (not (concrete? t))
+       (symbol? t)))
 
 (define (concrete? t)
-  (case t
-    ('int #t)
-    ('bool #t)
-    ('void #t)
-    (else #f)))
+  (and (symbol? t)
+       (char-upper-case? (string-ref (symbol->string t) 0))))
 
 (define (pretty-type t)
   (cond ((abs? t)
@@ -23,9 +22,22 @@
 	  (pretty-type (caddr t))))
 	(else (symbol->string t))))
 
+(define (pretty-constraints cs)
+  (string-append "{"
+		 (fold-left string-append
+			    ""
+			    (map (lambda (c)
+				   (string-append
+				    (pretty-type (car c))
+				    ": "
+				    (pretty-type (cdr c))
+				    ", "))
+				 cs))
+		 "}"))
+
 					; ('a, ('b, 'a))
 (define (env-lookup env n)
-  (if (null? env) (error #f "empty env")			; it's a type equality
+  (if (null? env) (error #f "empty env" env n)			; it's a type equality
       (if (eq? (caar env) n)
 	  (cdar env)
 	  (env-lookup (cdr env) n))))
@@ -69,48 +81,16 @@
 
 (define (builtin-type x)
   (case x
-    ('+ '(abs int (abs int int)))
-    ('- '(abs int (abs int int)))
-    ('* '(abs int (abs int int)))
-    ('! '(abs bool bool))
-    ('= '(abs int (abs int bool)))
-    ('bool->int '(abs bool int))
-    ('print '(abs string void))
-    (else #f)))
-
-; we typecheck the lambda calculus only (only single arg lambdas)
-(define (typecheck prog)
-  (define (check env x)
-    ;; (display "check: ")
-    ;; (display x)
-    ;; (display "\n\t")
-    ;; (display env)
-    ;; (newline)
-    (let
-	((res
-	  (case (ast-type x)
-	    ('int-literal (list '() 'int))
-	    ('bool-literal (list '() 'bool))
-	    ('string-literal (list '() 'string))
-	    ('builtin (list '() (builtin-type x)))
-
-	    ('if
-	     (let* ((cond-type-res (check env (cadr x)))
-		    (then-type-res (check env (caddr x)))
-		    (else-type-res (check env (cadddr x)))
-		    (then-eq-else-cs (~ (cadr then-type-res)
-					(cadr else-type-res)))
-		    (cs (consolidate
-			 (car then-type-res)
-			 (consolidate (car else-type-res)
-				      then-eq-else-cs)))
-		    (return-type (substitute cs (cadr then-type-res))))
-	       (when (not (eqv? (cadr cond-type-res) 'bool))
-		 (error #f "if condition isn't bool"))
-	       (list cs return-type)))
-	    
-	    ('var (list '() (env-lookup env x)))
-	    ('let
+    ('+ '(abs Int (abs Int Int)))
+    ('- '(abs Int (abs Int Int)))
+    ('* '(abs Int (abs Int Int)))
+    ('! '(abs Bool Bool))
+    ('= '(abs Int (abs Int Bool)))
+    ('bool->int '(abs Bool Int))
+    ('print '(abs String Void))
+    (else (error #f "Couldn't find type for builtin" x))))
+
+(define (check-let env x)
   					; takes in the current environment and a scc
 					; returns new environment with scc's types added in
   (let* ([components (reverse (sccs (graph (let-bindings x))))]
@@ -136,12 +116,13 @@
 		 [cs
 		  (fold-left
 		   (lambda (acc res c)
-				 (consolidate
-				  acc
-				  (consolidate (car res)
+		     (constraint-merge
+		      (constraint-merge
 					; unify with tvars from scc-env
 					; result ~ tvar
-					       (~ (cadr res) (env-lookup scc-env c)))))
+		       (~ (env-lookup scc-env c) (cadr res))
+		       (car res))				  
+		      acc))
 		   '() type-results comps)]
 					; substitute *only* the bindings in this scc
 		 [new-env
@@ -154,6 +135,38 @@
 	 [new-env (fold-left process-component env components)])
     (check new-env (last (let-body x)))))
 
+(define (check env x)
+  ;; (display "check: ")
+  ;; (display x)
+  ;; (display "\n\t")
+  ;; (display env)
+  ;; (newline)
+  (let
+      ((res
+	(case (ast-type x)
+	  ('int-literal (list '() 'Int))
+	  ('bool-literal (list '() 'Bool))
+	  ('string-literal (list '() 'String))
+	  ('builtin (list '() (builtin-type x)))
+
+	  ('if
+	   (let* ((cond-type-res (check env (cadr x)))
+		  (then-type-res (check env (caddr x)))
+		  (else-type-res (check env (cadddr x)))
+		  (then-eq-else-cs (~ (cadr then-type-res)
+				      (cadr else-type-res)))
+		  (cs (constraint-merge
+		       (car then-type-res)
+		       (constraint-merge (~ (cadr cond-type-res) 'Bool)
+					 (constraint-merge (car else-type-res)
+							   then-eq-else-cs))))
+		  (return-type (substitute cs (cadr then-type-res))))
+	     (list cs return-type)))
+	  
+	  ('var (list '() (env-lookup env x)))
+	  ('let (check-let env x))
+
+	  
 	  ('lambda
 	      (let* [(new-env (env-insert env (lambda-arg x) (fresh-tvar)))
 
@@ -167,6 +180,7 @@
 		;; (display "\n\t")
 		;; (display cs)
 		;; (display "\n\t")
+		;; (display (format "subd-env: ~a\n" subd-env))
 		;; (display resolved-arg-type)
 		;; (newline)
 		(list (car body-type-res)
@@ -182,7 +196,7 @@
 		      (other-func-type `(abs ,func-type ,return-type))
 		      (cs (~ func-type other-func-type))
 		      (resolved-return-type (substitute cs return-type))]
-		   (list cs resolved-return-type))
+		 (list cs resolved-return-type)))
 
 					; regular function
 	       (let* ((arg-type-res (check env (cadr x)))
@@ -192,12 +206,10 @@
 		      
 					; f ~ a -> t0
 		      (func-c (~
-				 func-type
-				 (list 'abs
-				       arg-type
-				       (fresh-tvar))))
-			(cs (consolidate
-			     (consolidate func-c (car arg-type-res))
+			       (substitute (car arg-type-res) func-type)
+			       `(abs ,arg-type ,(fresh-tvar))))
+		      (cs (constraint-merge
+			   (constraint-merge func-c (car arg-type-res))
 			   (car func-type-res)))
 		      
 		      (resolved-func-type (substitute cs func-type))
@@ -214,18 +226,26 @@
 		 (if (abs? resolved-func-type)
 		     (let ((return-type (substitute cs (caddr resolved-func-type))))
 		       (list cs return-type))
-		       (error #f "not a function"))))))))
+		     (error #f "not a function")))))))
     ;; (display "result of ")
     ;; (display x)
     ;; (display ":\n\t")
     ;; (display (pretty-type (cadr res)))
     ;; (display "\n\t[")
-      ;; (display (car res))
+    ;; (display (pretty-constraints (car res)))
     ;; (display "]\n")
     res))
-  (cadr (check '() (normalize prog))))
 
-					; returns a list of pairs of constraints
+					; we typecheck the lambda calculus only (only single arg lambdas)
+(define (typecheck prog)
+
+  (let ([init-env (flat-map data-tors (program-datas prog))])
+    (display init-env)
+    (newline)
+    (cadr (check init-env (normalize (program-body prog))))))
+
+  
+					; returns a list of constraints
 (define (~ a b)
   (let ([res (unify? a b)])
     (if res
@@ -235,116 +255,90 @@
 
 (define (unify? a b)
   (cond [(eq? a b) '()]
-	[(or (tvar? a) (tvar? b)) (list (list a b))]
+	[(tvar? a) (list (cons a b))]
+	[(tvar? b) (list (cons b a))]
 	[(and (abs? a) (abs? b))
 	 (let* [(arg-cs (unify? (cadr a) (cadr b)))
 		(body-cs (unify? (substitute arg-cs (caddr a))
 				 (substitute arg-cs (caddr b))))]
-	   (consolidate arg-cs body-cs))]
+	   (constraint-merge body-cs arg-cs))]
 	[else #f]))
 
-					; TODO: what's the most appropriate substitution?
-					; should all constraints just be limited to a pair?
-					; this is currently horrific and i don't know what im doing.
-					; should probably use ast-find here or during consolidation
-					; to detect substitutions more than one layer deep
-					; e.g. (abs t1 int) ~ (abs bool int)
-					; substituting these constraints with t1 should resolve t1 with bool
 (define (substitute cs t)
-					; gets the first concrete type
-					; otherwise returns the last type variable
-
-					; removes t itself from cs, to prevent infinite recursion
-  (define cs-without-t
-    (map (lambda (c)
-	   (filter (lambda (x) (not (eqv? t x))) c))
-	 cs))
-
-  (define (get-concrete c)
-    (let [(last (null? (cdr c)))]
-      (if (not (tvar? (car c)))
-	  (if (abs? (car c))
-	      (substitute cs-without-t (car c))
-	      (car c))
-	  (if last
-	      (car c)
-	      (get-concrete (cdr c))))))
-  
   (cond
-   ((abs? t) (list 'abs
-		   (substitute cs (cadr t))
-		   (substitute cs (caddr t))))
-   (else
-    (fold-left
-     (lambda (t c)
-       (if (member t c)
-	   (get-concrete c)
-	   t))
-     t cs))))
-
+   [(tvar? t)
+    (if (assoc t cs)
+	(cdr (assoc t cs))
+	t)]
+   [(abs? t) `(abs ,(substitute cs (cadr t))
+		   ,(substitute cs (caddr t)))]
+   [else t]))
+
+					; applies substitutions to all variables in environment
 (define (substitute-env cs env)
   (map (lambda (x) (cons (car x) (substitute cs (cdr x)))) env))
 
-(define (consolidate x y)
-  (define (merge a b)
-    (cond ((null? a) b)
-	  ((null? b) a)
-	  (else (if (member (car b) a)
-		    (merge a (cdr b))
-		    (cons (car b) (merge a (cdr b)))))))
-  (define (overlap? a b)
-    (if (or (null? a) (null? b))
-	#f
-	(if (fold-left (lambda (acc v)
-			 (or acc (eq? v (car a))))
-		       #f b)
-	    #t
-	    (overlap? (cdr a) b))))
-
-  (cond ((null? y) x)
-	((null? x) y)
-	(else
-	 (let* ((a (car y))
-		(merged (fold-left
-			 (lambda (acc b)
-			   (if acc
-			       acc
-			       (if (overlap? a b)
-				   (cons (merge a b) b)
-				   #f)))
-			 #f x))
-		(removed (if merged
-			     (filter (lambda (b) (not (eq? b (cdr merged)))) x)
-			     x)))
-	   (if merged
-	       (consolidate removed (cons (car merged) (cdr y)))
-	       (consolidate (cons a x) (cdr y)))))))
-
-					; a1 -> a2 ~ a3 -> a4;
-					; a1 -> a2 !~ bool -> bool
-					; basically can the tvars be renamed
+					; composes constraints a onto b and merges, i.e. applies a to b
+					; a should be the "more important" constraints
+(define (constraint-merge a b)
+  (define (f cs constraint)
+    (cons (car constraint)
+	  (substitute cs (cdr constraint))))
+  
+  (define (most-concrete a b)
+    (cond
+     [(tvar? a) b]
+     [(tvar? b) a]
+     [(and (abs? a) (abs? b))
+      `(abs ,(most-concrete (cadr a) (cadr b))
+	    ,(most-concrete (caddr a) (caddr b)))]
+     [(abs? a) b]
+     [(abs? b) a]
+     [else a]))
+
+					; for any two constraints that clash, e.g. t1 ~ abs t2 t3
+					; and t1 ~ abs int t3
+					; prepend the most concrete version of the type to the
+					; list of constraints
+  (define (clashes)
+    (define (gen acc x)
+      (if (assoc (car x) a)
+	  (cons (cons (car x) (most-concrete (cdr (assoc (car x) a))
+					     (cdr x)))
+		acc)
+	  acc))
+    (fold-left gen '() b))
+
+  (define (union p q)
+    (append (filter (lambda (x) (not (assoc (car x) p)))
+		    q)
+	    p))
+  (append (clashes) (union a (map (lambda (z) (f a z)) b))))
+
+
+;; 					; a1 -> a2 ~ a3 -> a4;
+;; 					; a1 -> a2 !~ Bool -> Bool
+;; 					; basically can the tvars be renamed
 (define (types-equal? x y)
   (let ([cs (unify? x y)])
     (if (not cs) #f	
 	(let*
-	    ([test-kind
-	      (lambda (acc c)
-		(if (tvar? c) acc #f))]
-	     [test (lambda (acc c)
+	    ([test (lambda (acc c)
 		     (and acc
-			  (fold-left test-kind #t c) ; check only tvar substitutions
-			  (<= (length c) 2)))])	     ; check maximum 2 subs per equality group
+			  (tvar? (car c)) ; the only substitutions allowed are tvar -> tvar
+			  (tvar? (cdr c))))])
 	  (fold-left test #t cs)))))
 
 					; input: a list of binds ((x . y) (y . 3))
 					; returns: pair of verts, edges ((x y) . (x . y))
 (define (graph bs)
+  (define (go bs orig-bs)
     (define (find-refs prog)
       (ast-collect
        (lambda (x)
 	 (case (ast-type x)
 					; only count a reference if its a binding
-	 ['var (if (assoc x bs) (list x) '())]
+	   ['var (if (assoc x orig-bs) (list x) '())]
 	   [else '()]))
        prog))
     (if (null? bs)
@@ -358,10 +352,11 @@
 
 	       (rest (if (null? (cdr bs))
 			 (cons '() '())
-		       (graph (cdr bs))))
+			 (go (cdr bs) orig-bs)))
 	       (total-verts (cons vert (car rest)))
 	       (total-edges (append edges (cdr rest)))]
 	  (cons total-verts total-edges))))
+  (go bs bs))
 
 (define (successors graph v)
   (define (go v E)