C, C++, Java suxx for a while
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Your computer science teacher gave a course about flow control structures
and you got a headache grasping the difference between "repeat-until" and
"while-endwhile"? In such case DON'T read this text or your brain will be
utterly damaged.

Let's consider the following programs:

  /******** for.c *********/          /******* while.c ********/
  #include <stdio.h>                  #include <stdio.h>
  int main(){                         int main(){
    char c;                             char c;
    for(c=1;c<10;c++){                  c=1;
      printf("%d",c);                   while(c<10){
    }                                     printf("%d",c);
  }                                       c++;
                                        }
                                      }

As you can see, they use different flow control structures. BASIC coders
will have problem with the for() statement, which is very different from
the "FOR c=1 TO 9 STEP 1" they are used to. And that's normal because:

      !!!!   IN C, THE FOR STATEMENT IS IN FACT A WHILE   !!!!

To prove this, i compiled both programs with different compilers, and they
gave EXACTLY THE SAME EXECUTABLE (Norcroft C v4.00 and GCC 2.95.2 pre3a on
RiscOS) or the same assembly source, except for some debug informations and
useless labels (Turbo C 2.01 on DOS, see Appendix).

                                     source
                                       \/
                                +------\/------+
                                | preprocessor <<< other source files
In usual C compilers, the       +------\/------+
compilation proceeds by                \/
successive steps, which are     +------\/------+
naively summarised on the       |   compiler   |
right (they depend on the       +------\/------+
compiler, and there are                \/
a lot more invisibles steps).   +------\/------+
                                |    linker    <<< other objects file
                                +------\/------+
                                       \/
                                   executable

During the processing step, all the "for(A;B;C) {...}" statements are
converted in "A;while(B) {... C}" statements. Thus the compiler only
has to cope with the while structure.

The problem is that a for structure can generally be easily optimised
in assembly language with a decrement+branch (except when the step or
bounds are variables), and i'm not that sure the compiler will optimise
it correctly.

So believe it, C, C++, Java suxx... ;p BASIC rulez!

Alain Brobecker


---- Appendix: for.asm, while.asm & differences ----
begin 644 8086asm.zip
M4$L#!!0  @ ( +A1KB@9\K]V5P$  "$#   ' !@ 1D]2+D%334%#% !!4D,Pa
M2?___Q#^@;XS         )52RT[#,! \[WY%;KD8*7T)*2<#K3@6T2(A(13%a
MCML&Y6'9#@U_C^VX55N0H*>U=V=GQKO.UHO7-6BQK45C@'T9$<F.526/XH?Ea
M?!'C_/%Y^?($6]5V$K+YW?J.9/>K%4*N=5<+X#K-' <I=#I@B3Z<T%= -(5&a
MWWH0BO:M*HY"KA)C0:'*F:B\"7O;AZN#ANZ!R*K_SF,+,;(S&G9!XWK_9^=4a
M\:\9975>-B!5RZ$1N4*0G=X!DPAU^VDCT?:H.P9:DG%(>A:CHC<F;T;O9(3Pa
M44O0NU:9B,Z03M,!F%?D HO V1Z"2-X'6$_:S48+$PVC3S4]0?"\JKPU3Y-)a
M539F8^NM!-X?(YU8S;+AE^:0SFR!6WL_7"?6=@5TBG2<(AT%S_:9[O&.UD4Ea
M#(09V7G*JWZ%/ELG% PFMSZ,DL3'Y*I-@>B-:B!,(/7;.K$# QJ\670Y_ 90a
M2P,$%  "  @ MU*N*%4*WL>S     @$   H & !214%$344N5%A404,4 $%2a
M0S!)____KKB"OC,         38^Q:L- $$1[?<6PE4W,@27<I'+ I$J9'SCYa
M5M:%U6W0WEG1W^>P7:B:XCV&F0OWY8:8!ITGGZ,FPS+RS AQ&&JF#)\"9I[Ta
MSJ'Y5HS^SJ#J.V\3@?_\-<L*BU,4/R,K:!FC\ ,?\%,L(RCRR!A41)>8;@UPa
M=+7T5_RUEIV[=WK[^D2_@@C8E52, \3W+/OJMENWI8=W[JB2;DM.+])2\V%8a
MM<"8#UCXN3GI]B=$S9QSS3]02P,$%  "  @ Q%&N*.>6)O=4 0  '0,   D a
M& !72$E,12Y!4TU!0Q0 05)#,$G___\'!H*^,P        "54DU/PS ,/=N_a
MHK=>@M1NDY!Z"K")XQ ;$A)"59-F6U$_HB1EY=^3I!EL PEV<A(_O_=L)U\Oa
MGM>@Q;81K0'V840D>U97/(KOEO-%C//[Q^73 VQ5UTO(YS?K&Y+?KE8(A=9]a
M(X#K+'<<I-39B"7Z<$*? =&6&GWI02C:=ZK\$G*9&$L*=<%$[4W8VSY<'314a
MCT16_7<>FXB1G="P,QI7^S\[QXI_S2AOBJH%J3H.K2@4@NSU#IA$:+IW&XFVa
M1]TST)),PJ-G,2IZ8?(J?24IPELC0>\Z92(Z03K+1F!1DS,L F=[""+%$& #a
MZ38;+4PTCC[3] C!B[KVUCQ-+E75FHW-=Q+X\!VKEI\;0SJQ1KBU]L-Q8BW7a
M0&=(IQG2-/BU+;K&':6+2A@(\[&SE!?]"'VR2B@93*]]2)/$Q^2B+8$8C&HAa
M=)_Y31W9@1$-WBRZ-_P$4$L! A0-%  "  @ N%&N*!GROW97 0  (0,   < a
M&          @         $9/4BY!4TU!0Q0 05)#,$G___\0_H&^,P      a
M  !02P$"% T4  ( " "W4JXH50K>Q[,    " 0  "@ 8         "    "4a
M 0  4D5!1$U%+E185$%#% !!4D,P2?___ZZX@KXS         %!+ 0(4#10 a
M @ ( ,11KBCGEB;W5 $  !T#   ) !@         (    (<"  !72$E,12Y!a
M4TU!0Q0 05)#,$G___\'!H*^,P        !02P4&      ,  P#L    &@0 a
#    a
 a
end