On Monday 17 November 2003 17:08, Ulrich Mayring wrote:
> Niclas Hedhman wrote:
> > On Saturday 15 November 2003 05:41, Ulrich Mayring wrote:
> >>And I hate unit-testing, I'd rather prove my code correct ;-)
> >
 > You never miss an opportunity to point this out, LOL...
> >
> > Curious, how can you "prove" anything without running it?
>
> Very hard, but all I said was I'd rather do that than do unit-testing 

Prove the algorithm, but unit test the implementation - just to make 
sure you really implemented what you proved, and that your algorithm
works as expected.

For example, I may have proven that my algorithm will provide output o
for input i. But:

 + Did I implement it correctly (maybe some Java construct I used
doesn't
   work the way I think it does).

 + Was my proof correct?

 + Do I really want this algorithm? I.e. when run on the actual data it
   will process, do I get the output I want?

The prove versus unit-test Wrestlemania was fought when XP was new at
my university. The above is what I walked away with.

/LS


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to