Optimizing Java using D

[safety0ff via Digitalmars-d]

On Saturday, 21 June 2014 at 01:48:20 UTC, Peter Alexander wrote:
> On Saturday, 21 June 2014 at 01:07:20 UTC, safety0ff wrote:
>> std.algorithm.sort with SwapStrategy.unstable is considerably 
>> slower than his, whereas builtin sort is  abysmal.
>>
>> I find that generally using SwapStrategy.stable performs 
>> better.
>> This isn't universal though as there are cases where it is 
>> slower.
>
> wat
>
> Shouldn't unstable be faster, otherwise we might as well just 
> alias unstable = stable!?

Many times my data is the result of mapping sorted/partially 
sorted input to a function which does not randomize order.
This creates many partially sorted runs.

Timsort is used for stable sorting and it was designed for this 
type of input.