topN using a heap

[Timon Gehr via Digitalmars-d]

On 01/19/2016 12:51 AM, Ilya wrote:
> On Monday, 18 January 2016 at 23:49:36 UTC, Andrei Alexandrescu wrote:
>> On 1/18/16 6:44 PM, Ilya wrote:
>>> On Monday, 18 January 2016 at 23:27:19 UTC, Ivan Kazmenko wrote:
>>>> On Monday, 18 January 2016 at 23:18:03 UTC, Ilya wrote:
>>>>> A RNGs don't improve worst case. It only changes an permutation for
>>>>> worst case. --Ilya
>>>>
>>>> Still, use of RNG makes it impossible to construct the worst case
>>>> beforehand, once and for all.  In that sense, this is a regression.
>>>
>>> No, it is definitely possible, because RNGs are Pseudo-RNGs. --Ilya
>>
>> unpredictableSeed uses the system clock as a source of randomness, so
>> we're good there. -- Andrei
>
> Would this work for pure functions? --Ilya

Only if they accept the RNG as an additional argument.