topN using a heap

[Ilya via Digitalmars-d]

On Monday, 18 January 2016 at 23:53:53 UTC, Timon Gehr wrote:
> On 01/19/2016 12:50 AM, Ilya wrote:
>> ...
>>
>> 1. Yes, probability of hitting the worst case repeatedly is is
>> practically zero. But RNGs do not change this probability.
>> 2. It is possible to build attack for our RNGs, because they 
>> are
>> Pseudo-RNGs.
>> --Ilya
>
> You also need to predict the seed. How do you do that?

We can not use unpredictable seed (like system clock) in pure 
functions. --Ilya