The std::merge algorithm accepts two pairs of begin/end iterators, which denote the input ranges. These ranges must be sorted. The fifth parameter is an output iterator that accepts the incoming items during the merge.

There is also a variant called std::inplace_merge. This algorithm does the same as the other, but it does not need an output iterator because it works in place, as the name already suggests. It takes three parameters: a begin iterator, a middle iterator, and an end iterator. These iterators must all reference data in the same data structure. The middle iterator is at the same time the end iterator of the first range, and the begin iterator of the second range. This means that this algorithm handles a single range, which actually consists of two consecutive ranges, such as, for example, {A, C, B, D}. The first subrange is {A, C} and the second subrange is {B, D}. The std::inplace_merge algorithm can then merge both within the same data structure, which results in {A, B, C, D}.