Re: No Go To's Forever!

[Robert Dewar <robert_dewar@my-deja.com>]

In article <yP9C4.1507$ZT3.32509@news.pacbell.net>,
  tmoran@bix.com wrote:
>   I personally find this clearer, even if a tad longer:
>    loop
>      J := D'first;
>      while J < D'last and then D(J) >= D(J+1) loop
>        J := J+1;
>      end loop;
>      exit when J >= D'last;
>      Swap (D (J), D (J + 1));
>    end loop;

Well that's a different algorithm. It tests J against D'Last
twice as often as my formulation. Yes, perhaps a clever
compiler will be able to rescue this, but algorithmically,
this simply is not equivalent.

Actually most compilers won't fix this, so you will find that
your version actually does run noticably slower that the
original.

Yes, code duplication is one way of getting around the goto
here, but again you are changing the algorithm.

Please tackle the harder problem: Take the algorithm *I*
presented, and remove the goto without changing the sequence
of computations, and see if that result is clearer.

P.S. I don't find your changed algorithm clearer, precisely
because of the peculiar duplicated check.



Sent via Deja.com http://www.deja.com/
Before you buy.