kmm32(isa):keep flag

as:$ keep flag

keep flag, whether to decrement the stack pointers for
the arguments used in the instruction
e.g. (a--n) --> (a--a n)

this is useful in situations where we want to keep the
arguments on the stack to do something else with, e.g.
a lot of the times we'll do `du fb` or something. now
we can save bytes and just do `$fb`. yippee!

as.c

@@ -54,6 +54,7 @@ _ V pass1(S*t){_ B o=0;S x=*t;B c;c=*x.p;
   Q('-'==c,builtin(t))OR
   Q(';'==c,N(label->p,die("; at top-level"))label=label->p)OR
   Q('~'==c,++x.p,--x.l,o|=0x40,pass1(&x))OR
+  Q('$'==c,++x.p,--x.l,o|=0x80,pass1(&x))OR
   Q(2!=x.l||OPSZ==(o|=op(x.p)),tw(x),die(" unknown inst"))OR
   mb(mel++,o),o=0;}
 _ V pass2(V){Label*l;Ws a;Bs b;Vfor(Ref,refs,r,i,Q(l=lget(r.p,r.n),

d/kmm32

@@ -47,7 +47,8 @@ INSTRUCTIONS
     ||O: Opcode
     |F: Flip return and data stacks
     |   (eg 01000010 will put the next byte in memory onto the return stack)
-    U: Unused
+    K: Keep left hand operands
+       (eg (a--n) below becomes (a--a n))
 
 OPCODES
 

d/kmm32-as

@@ -42,6 +42,7 @@ SYNTAX
     resulting opcode:
 
       ~ - F flag (flip stack manipulations from data <-> return stacks)
+      $ - K flag (keep lhs on rhs)
 
   BUILTINS
     Colour of arguments denoted with first character.

kmm32.c

@@ -4,11 +4,12 @@
 #define DS ((W*)(mem+(MEMSZ-0x800)))
 #define RS ((W*)(mem+(MEMSZ-0x400)))
 #define SD (RS-DS)
+#include<stdio.h>
 
 B mem[MEMSZ];
 W*dp=DS,*rp=RS,ip;
 
-_ W pop(V){R*--dp;}
+//_ W pop(V){R*--dp;}
 _ V put(W i){*dp++=i;}
 _ V pur(W i){*rp++=i;}
 _ V swp(V){W*t;t=dp,dp=rp,rp=t;}
@@ -28,42 +29,50 @@ _ V pb(V){put(lob(ip)),ip+=1;}
 _ V ph(V){put(loh(ip)),ip+=2;}
 _ V pw(V){put(low(ip)),ip+=4;}
 _ V pr(V){put(ip+(Bs)lob(ip)),ip+=1;}
-_ V fb(V){put(lob(pop()));}
-_ V fh(V){put(loh(pop()));}
-_ V fw(V){put(low(pop()));}
-_ V mb(V){AB,stb(b,a);}
-_ V mh(V){AB,sth(b,a);}
-_ V mw(V){AB,stw(b,a);}
-_ V io(V){AB,ios[b].io(a);}
-_ V ii(V){put(ios[pop()].oi());}
-_ V ss(V){pur(pop());}
-_ V dr(V){--dp;}
-_ V sw(V){AB,put(b),put(a);}
-_ V du(V){put(*(dp-1));}
-_ V ov(V){put(*(dp-2));}
-_ V ad(V){AB,put(a+b);}
-_ V su(V){AB,put(a-b);}
-_ V mu(V){AB,put(a*b);}
-_ V di(V){AB,put(a/b),put(a%b);}
-_ V an(V){AB,put(a&b);}
-_ V or(V){AB,put(a|b);}
-_ V xr(V){AB,put(a^b);}
-_ V sl(V){AB,put(a<<b);}
-_ V sr(V){AB,put(a>>b);}
-_ V sa(V){AB,put((Ws)a>>b);}
-_ V eq(V){AB,put(a==b);}
-_ V lt(V){AB,put(a<b);}
-_ V gt(V){AB,put(a>b);}
-_ V no(V){put(!pop());}
-_ V ju(V){ip=pop();}
-_ V jc(V){AB;Q(a,ip=b)}
-_ V ca(V){pur(ip),ip=pop();}
-_ V cc(V){AB;Q(a,pur(ip),ip=b)}
+_ V fb(V){put(lob(a));}
+_ V fh(V){put(loh(a));}
+_ V fw(V){put(low(a));}
+_ V mb(V){stb(b,a);}
+_ V mh(V){sth(b,a);}
+_ V mw(V){stw(b,a);}
+_ V io(V){ios[b].io(a);}
+_ V ii(V){put(ios[a].oi());}
+_ V ss(V){pur(a);}
+_ V dr(V){(V)a;}
+_ V sw(V){put(b),put(a);}
+_ V du(V){put(a),put(a);}
+_ V ov(V){put(a),put(b),put(a);}
+_ V ad(V){put(a+b);}
+_ V su(V){put(a-b);}
+_ V mu(V){put(a*b);}
+_ V di(V){put(a/b),put(a%b);}
+_ V an(V){put(a&b);}
+_ V or(V){put(a|b);}
+_ V xr(V){put(a^b);}
+_ V sl(V){put(a<<b);}
+_ V sr(V){put(a>>b);}
+_ V sa(V){put((Ws)a>>b);}
+_ V eq(V){put(a==b);}
+_ V lt(V){put(a<b);}
+_ V gt(V){put(a>b);}
+_ V no(V){put(!a);}
+_ V ju(V){ip=a;}
+_ V jc(V){Q(a,ip=b)}
+_ V ca(V){pur(ip),ip=a;}
+_ V cc(V){Q(a,pur(ip),ip=b)}
 
 _ V(*ops[])(V)=
 /*0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f  10 11 12 13 14 15 16 17 18 19 1a 1b 1c 1d 1e 1f 20 21 22 23 24*/
 /*0  1  2  3  4  5  6  7  8  9  10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36*/
  {np,ex,pb,ph,pw,pr,fb,fh,fw,mb,mh,mw,io,ii,ss,dr,sw,du,ov,ad,su,mu,di,an,or,xr,sl,sr,sa,eq,lt,gt,no,ju,jc,ca,cc};
+_ B arity[]=
+ { 0, 0, 0, 0, 0, 0, 1, 1, 1, 2, 2, 2, 2, 1, 1, 1, 2, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 1, 1, 2, 1, 2};
 
-V emu(V){B m,f;while(ip<0x40000){Q((f=(m=mem[ip++])>>6)&1,swp());ops[0x3f&m]();Q(f&1,swp());}}
+_ V op(B n){B k,f,c,o;o=n&0x3f,f=n>>6&1,k=n>>7,c=arity[o];
+  Q(f,swp())
+  Q(0==c,                           ops[o]())OR
+  Q(1==c,a=dp[-1],         dp-=  !k,ops[o]())OR
+  Q(2==c,a=dp[-2],b=dp[-1],dp-=2*!k,ops[o]())
+  Q(f,swp())}
+V emu(V){while(ip<0x40000){op(mem[ip++]);}}
 V lod(B*c,W l){for(ip=0;ip<l;++ip)mem[ip+0x100]=c[ip];ip=0x100;}