Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!cbatt!cbosgd!ihnp4!inuxc!iuvax!cjl
From: cjl@iuvax.UUCP
Newsgroups: net.lang.c
Subject: Re: Pascal vs C, again (was: Pascals Ori
Message-ID: <9500032@iuvax.UUCP>
Date: Wed, 13-Aug-86 11:18:00 EDT
Article-I.D.: iuvax.9500032
Posted: Wed Aug 13 11:18:00 1986
Date-Received: Sat, 16-Aug-86 05:50:41 EDT
References: <7014@boring.UUCP>
Lines: 61
Nf-ID: #R:boring:-701400:iuvax:9500032:000:2051
Nf-From: iuvax!cjl    Aug 13 10:18:00 1986



  Let us see what Pascal would be if Dr. N. Wirth could redesign it again.
In his Modula-2, both conditional boolean operators and LOOP-EXIT-END are 
introduced. So with Modula-2 (Pascal-2 ??), we would write loops as :

   (1)  i:= 1;
        LOOP
          (* invariant : (i <= 1001) and (0 not in x[1..i-1]) *)
          IF i = 1001 THEN EXIT END;
          (* assert (i <= 1000) and (0 not in x[1..i-1]) *)
          IF x[i] = 0 THEN EXIT END;
          INC(i);
        END;
        IF i = 1001 THEN WriteString("not found."); WriteLn;
        ELSE WriteString("Zero at location "); WriteInt(i,1); WriteLn;
        END;

   (2)  i := 1;
        LOOP
          (* invariant : (i <= 1001) and (0 not in x[1..i-1]) *)
          IF (i =1001) OR (x[i] = 0) THEN EXIT END;
          INC(i);
        END;
        ..............

   (3)  i := 1;
        WHILE (i <= 1000) AND (x[i] <> 0) DO 
          INC(i);
        END;
        ................

   (4)  i:= 1;
        LOOP
          (* invariant : (i <= 1001) and (0 not in x[1..i-1]) *)
          IF i = 1001 THEN 
            WriteString("not found."); WriteLn; EXIT ;
          END;
          (* assert (i <= 1000) and (0 not in x[1..i-1]) *)
          IF x[i] = 0 THEN 
            WriteString("Zero at location "); WriteInt(i,1); WriteLn; EXIT;
          END;
          INC(i);
        END;

Wirth likes (3) because it tries to group and document the loop exit 
conditions AT one place(simpler). Personally I like (4) because it is 
easier to reason. (Ref Gries : The Science of Programming)
Note there is no statment between the two exit conditions and post-loop
clean-up follows corresponding exit condition more closely. That simplies
the program reasoning. 

  Apparently Wirth doesn't like to use boolean variables for controlling
loop-exit today. (Any other opinion ?)  Also arbitrary loop-break 
without trying to group them together would make post-loop assertion 
unnecessary complex. (Thus hard to read.)

C.J.Lo
Dept. of CIS, IUPUI
UUCP : ...!iuvax!cjl
ARPA : cjl@Indiana@CSNet-Relay