faster splitter

[qznc via Digitalmars-d]

On Monday, 23 May 2016 at 11:52:35 UTC, Seb wrote:
> On Sunday, 22 May 2016 at 18:56:30 UTC, qznc wrote:
>> On Monday, 4 March 2013 at 19:11:17 UTC, Steven Schveighoffer 
>> wrote:
>>> [...]
>>
>>
>> Below is an implementation, which matches MySplitter with dmd, 
>> but not with ldc. More precisely:
>>
>> [...]
>
> have you thought about opening a PR to improve `splitter`?

Yes, but I'm not sure about the goals.

I also want to dig a little deeper. Being "sometimes faster" 
without some understand why and when feels unsatisfying.

Additionally, there is this weird special case for a 
bidirectional range, which just adds unnecessary overhead. Is 
"remove dead code" a good enough reason in itself for a PR?