Indeed adrian told me the same. Duplication implies more potential  
bugs. So people not willing to have traits
could just remove the classes and substitute the traits by the  
methods. It should be much easier to do than introducing
traits. Now we did not try that.

>> I think that if this eases traits adoption we could remove the
>> behavior refactorings and introduce code duplication
>> instead (ie flattened the traits) so that removing traits could be
>> done for people do not wanting them for x or y reasons.
>> Removing traits is much easier than introducing them because of
>> bootstrapping reasons.  We could also have the policy not
>> to use traits for kernel parts so that only applications use them.
>>
>> Stef
>
> I am not sure about the value of using code duplication - I mean, what
> plausible values of x and y are there? But other people can answer  
> that
> better than I. However, the non-Traits policy for kernel parts seems
> reasonable to me.