But I think we should aim to duplicate the behavior defined by the 
spec, not the idiosyncracies present in the RI.  The doc doesn't say 
anywhere that you should be able to register a null factory, and again, 
why iterate through a list full of nulls and duplicate entries?

Aaron

On Thu, 14 Aug 2003 [EMAIL PROTECTED] wrote:
> Personaly I agree completely, but I discovered this behaviour while
> writing the unittest against SUNs reference implementation. 
> I had to remove the null check from my original code to make the test
> failures go away. :)

Reply via email to