Re: ANNOUNCE Mika by Midoan New Release : Automated Test Data Generation for Ada

[info@midoan.com]

On May 26, 9:18 am, Maciej Sobczak <see.my.homep...@gmail.com> wrote:
> On 25 Maj, 22:26, i...@midoan.com wrote:
>
> > Mika [...] has no knowledge of the intended
> > behaviour of the code. It only automatises the generation of tests.
>
> These two sentences above seem to be conflicting to me, or at least
> they provoke a justified question on the actual value of such
> generated tests.
> Can you provide some examples?
>
> --
> Maciej Sobczak *www.msobczak.com*www.inspirel.com
>
> Database Access Library for Ada:www.inspirel.com/soci-ada

The most simple typical example is :

procedure swap(x, y : in out Integer) is
  t : Integer;
begin
  t := y;       --should be t := x;
  x := y;
  y := t;
end swap;

For this Mika generates the following test:

TEST NUMBER 1

CONSTRUCTED TEST
x = -3411276
y = -2256887

PREDICTED RESULTS
x = -2256887
y = -2256887

   100% branch coverage predicted.
1 tests generated (paths followed)
0 paths fully attempted but for which tests could not be generated

Mika generates tests, and can compile and run the tests but it cannot
answer the question: "did the tests pass or fail?". The fact that the
intended behaviour was a swap of the two arguments is obviously
completely unknown to Mika.

But if an oracle is available (e.g. during regression testing, or from
previous test runs) it is possible to perform back to back test runs
automatically using the tests generated: thus answering the question
did the test pass or fail... But of course the answer is completely
dependent on the integrity of the oracle.

More complex examples are available at http://www.midoan.com/mika.html

Hoping this clarify things a little.
Regards,
Midoan.