Thomas Enebo wrote:
> It is a good question why many impls all decided to copy this behavior, 
> but it is there and I am guessing the reason for supporting it is 
> "principle of least surprise" (and yes I find this somewhat ironic).

Yeah, the least surprising result is rather surprising, since I haven't 
thought of or been told any cases where dupes would be desirable. So I 
guess the consensus is that dupes need to be allowed/expected for {} and 
not otherwise.

- Charlie