Do we need test cases?

Daniel Martin wrote:

>José Alves de Castro <[email protected]> writes:
>
>  
>
>>You're right.
>>
>>It is now my opinion that these quizzes should all have tests included :-)
>>    
>>
>
>But this quiz /did/ have a test script for the output posted:
>
> http://perl.plover.com/~alias/list.cgi?1:mss:2064
>
>I'll grant you, MJD didn't include the test himself, but it was the
>second message posted to the -discuss list after the quiz was posted.
>
>
>  
>

This does bring up the question of _should_ there be standard test cases 
posted with the QotW's?

Reasons For:

- With the apparent growing popularity of the list, published test cases 
would help ensure standardized inputs and outputs, thus lightening the 
load of the summarizer.

- It can help resolve many different ambiguities in the problem definition.

- Test Cases always help in the debugging process.


Reasons Against:

- For some problems (notably the expert levels), writing a suitable test 
case is on par with writing the problem. Witness the "center letters" of 
a couple weeks ago.

- Have such a strictly defined requirement could restrain creativity.

- The test case itself may itself be a spoiler, especially if it is a 
flexible one.


Personally, I could go either way. But I would be more inclined to 
volunteer as a summarizer if the test cases were in use.

What do others think?

-- Rod