On Tue, Oct 21, 2008 at 10:32 AM, Vaclav Stumbauer <[EMAIL PROTECTED]> wrote:
> My subjective view is that they're stupid only in the space of real
> world scenarios but not in the space of problem solving capability tests :-)

If they're not realistic, then what exactly are they supposed to be
testing? Why are they the best method of testing how you'll actually
perform on the job, rather than, say, stuff based on real life?

If you want to test problem solving, give a real problem to solve. Not
trick questions.

> They are short enough and have small solution space enough to be usefull
> in such tests.

They aren't useful though. That's the point. They are questions with
rote answers. All you're testing is the candidates ability to learn
rote answers. You might as well set a test with questions like "What's
6 times 12" if that's what you want to test.

> No offence  - what some useful and utile test chunks with similar
> properties based on real world scenarios would be?

Real, live example, which I had to do.

Assume 'int translate_float(const char* value, double* result)'
returns 0 on failure and converts a 'value' string into a double in
result on success:

int translate_latitude(const char* direction, const char* value,
double* result){
    /* Translates DDmm.mmmm{N|S} into {-}DD.DDDDDD.. */
    double mixed, minutes;
    int int_degrees;


    if (!direction || !direction[0] || !value || !value[0] || !result){
        /* either NULL pointers or NUL strings */
        *result = 0.0;
        return 0; /* Fail */
    }

    if (!translate_float(value, &mixed)){
        /* no conversion made */
        *result = 0.0;
        return 0; /* Fail */
    }

    int_degrees = (int)(mixed/100);
    minutes = (mixed - int_degrees*100)/60.0;

    *result = int_degrees + minutes;

    if ('S' == direction[0] || 's' == direction[0]){ /* toupper() was
generating a warning which I couldn't remove */
        (*result) *= -1;
    }

    return 1; /* success */
}

What tests/input would you put into this piece of code to test all
paths of execution?

> I'm asking just since my poor mind can't think of any such a solution
> right now, not to boost the quarrel.



-- 
PJH

http://shabbleland.myminicity.com/tra

Reply via email to