Re: if statements

[James Hassett <hassett@planet8.tds-eagan.lmco.com>]

Ken Garlington <Ken.Garlington@computer.org> wrote:
> "Robert Dewar" <robert_dewar@my-deja.com> wrote in message
> news:8udce8$1qi$1@nnrp1.deja.com...
> 
> : I would write the above as:
> :
> :    procedure Conditional_Swap (A, B : in out Integer) is
> :    begin
> :       if A < B then
> :          declare
> :             Temp : constant Integer := A;
> :          begin
> :             A := B;
> :             B := Temp;
> :          end;
> :       end if;
> :    end Conditional_Swap;
> 
> This causes the *smaller* value to end up in A and the *larger* value to end
> up in B? Now I REALLY feel like a beginning student!
> 
> Here's the problem statement as I understand it:
> 
> : >   "Exercise 4.1: Write an if statement which will swap the values of
> : >    variables A and B if necessary so that the smaller value ends up in
> : >    A and the larger value ends up in B."
> 
> I always thought that a test such as "A < B" was true only if A was
> *already* smaller than B (in which case you shouldn't do a swap, right?).

This defect has been carried forward in a surprising number of
posts, so I'm glad Ken caught it.  I believe it originated with
Daniel Allex, who wrote

> procedure Swap ( A,    B : in out Integer ) is
>      Temp : Integer := 0;
>    begin
>      if A <= B then
>          Temp := A;
>          A := B;
>          B := Temp;
>      end if;
>    end;

When I saw this "solution" offered, I thought maybe Daniel had
deliberately introduced the bug to ensure that if the student
simply handed in the offered solution, it would be wrong, but if he
bothered to try to understand it, he would likely find and correct
the error.  I thought about posting a correction, but decided that
maybe the correction was best left as an exercise for the student.

I was surprised that Robert didn't catch this defect, but the
discussion has drifted far from the original problem posed, and
it is easy to lose sight of such (critical) details as which way
the test ought to go.  We rarely need to forgive Robert for
technical errors, so I'm willing to cut him some slack here.

Another problem (as noted by some already) is that the original
problem asked for an if statement, not a procedure, so we would
need to trim Robert's solution down to the if statement.  (The
solution offered by Daniel doesn't trim down as neatly, because
the Temp declaration gets lost.)

Otherwise, I certainly agree with Robert's critique of Daniel's
solution.

- Jim Hassett