Closed Philippe-Cholet closed 10 months ago
@phimuemue So opt-in or opt-out?
If tomorrow we implement DoubleEndedIterator
on some existing iterator, then in both cases, we have to update the test here. Therefore both are not ideal so I think it's merely a preference, and I prefer functions/traits over macro when possible.
I sure would prefer to merge those two functions in whatever form that detect if it needs to be done or not (I mean, checking if each of them should have a double ended test was not really quick to do), here and for benchmarks. But I don't see a way to do it.
We don't have any
rfold/nth_back
specialization yet but it will come as I intend to dorfold
specializations alongsidefold
ones (benchmarks are ready too).Apart from
rciter
for which I did not add any specialization test, all our double ended iterators are included here.