Re: Comments on Mathias blog post about contributing

On Sep 23, 2009, at 8:45 AM, Frank Schoenheit, Sun Microsystems  
Germany wrote:
> Hi Mathias,
>
>> Really, this part of the work is superfluous. IMHO tests that are  
>> known
>> to be broken in a particular milestone should be skipped in a CWS  
>> based
>> on it also.
>
> Don't think this is a good idea, since a test can be broken in  
> different
> ways. For instance, if your test checks 10 aspects, and one of them is
> broken in MWS, you still want to know if the other 9 are okay in your
> CWS. Otherwise, you'll notice a breakage in those 9 only when the one
> failure is fixed, and the whole test re-enabled in MWS.

I think this point is moot.  I strongly believe that tests that are  
regularly run on all CWS must also be run on the MWS, and the MWS must  
not be declared ready until all those test succeed.  Everything else  
is a waste of time.

(The link below is about a different scenario of running tests, if I  
understand correctly.)

-Stephan

> An interesting read (IMO), somewhat related to the topic:
> http://blog.ulf-wendel.de/?p=259 (sorry, German, but I know you  
> speak it
> pretty well :)
> My favoite quote, which I'd strongly agree to:
>
>  Deactivating a test, to reach "0 test failures", is wrong, since the
>  equality of a failure is also a sensible information.