Simplify colon loop / dispatch; avoid word lookup for tick/draw

Substantial speed gains _and_ f_cexecute() now works for deferred words.
This is the first version of Jorth where I feel like I've gotten the VM
right enough that it can reasonably be called a proper Forth implementation.

jorth.c

@@ -1,4 +1,4 @@
-#define TRACE
+//#define TRACE
 
 #include <stdio.h>
 #include <sys/stat.h>
@@ -440,25 +440,8 @@ void f_cdef() { // func name --
   f_comma();
 }
 
-void f_docolon();
 void f_revlookup();
 
-// C code must always call a colon word through f_cexecute()
-void f_cexecute() {
-  cell oldW = W;
-  cell oldIP = IP;
-  cell retIP;
-  retIP.p = NULL;
-  W = TOP();
-  DROP(1);
-  if (W.p->f == f_docolon) {
-    RPUSH(retIP);
-  }
-  W.p->f();
-  W = oldW;
-  IP = oldIP;
-}
-
 #ifdef TRACE
 void f_traceon() {
   TRACING = 1;
@@ -466,11 +449,8 @@ void f_traceon() {
 void f_traceoff() {
   TRACING = 0;
 }
-#endif
 
 void f_colondispatch() {
-  cell codeptr;
-#ifdef TRACE
   static int printing = 0;
 
   if (TRACING && !printing) {
@@ -487,15 +467,11 @@ void f_colondispatch() {
     }
     printing = 0;
   }
-#endif
-  codeptr = *W.p;
-  if (codeptr.f == f_docolon) {
-    RPUSH(IP);
-    IP.p = W.p + 1;
-  } else {
-    codeptr.f();
-  }
+  W.p->f();
 }
+#else
+#define f_colondispatch() W.p->f()
+#endif
 
 void f_colonloop() {
   while (IP.p) {
@@ -505,16 +481,6 @@ void f_colonloop() {
   }
 }
 
-void f_docolon() {
-  IP.p = W.p + 1;
-  f_colonloop();
-}
-
-void f_dodeferred() {
-  W = *(W.p + 1);
-  f_colondispatch();
-}
-
 // this version of f_execute can be run from a colon word
 void f_execute() {
   W = TOP();
@@ -522,6 +488,27 @@ void f_execute() {
   f_colondispatch();
 }
 
+// C code must always call a colon word through f_cexecute()
+void f_cexecute() {
+  cell oldW = W;
+  cell oldIP = IP;
+  IP.p = NULL;
+  f_execute();
+  f_colonloop();
+  W = oldW;
+  IP = oldIP;
+}
+
+void f_docolon() {
+  RPUSH(IP);
+  IP.p = W.p + 1;
+}
+
+void f_dodeferred() {
+  W = *(W.p + 1);
+  f_colondispatch();
+}
+
 void f_lit_() {
   PUSHC(*IP.p);
   IP.p++;

testbed.c

@@ -271,6 +271,7 @@ void do_repl() {
 }
 
 int main(int argc) {
+  cell tick, draw;
   if (argc > 1) {
     do_repl();
     return 0;
@@ -278,14 +279,16 @@ int main(int argc) {
   ser_init(SER_COM2, BAUD_19200, SER_8N1);
   game_init();
   game_f_init();
+  tick = f_lookupcp("tick");
+  draw = f_lookupcp("draw");
 
   while (!keyIsDown(K_ESC)) {
     kbd_debounce();
     f_poll();
     f_taskloop();
-    f_runstring("tick");
+    f_execcp(tick);
     f_taskloop();
-    f_runstring("draw");
+    f_execcp(draw);
   }
 
   return 0;