[eesmith]

I've lost track of the thread, because I don't know what you are talking about. Who is saying what doesn't exist?

The lower_bound you pointed to takes start/end iterators defining the partially-ordered range to examine.

The linked-to essay and the Wikipedia take array + size.

These are different APIs.

The latter - which is what this thread is about, IMO - is easier to implement because the sizes are always non-negative.

[froh]

random access iterators provide exactly the pointer into array semantics, with a difference type and jumping to a location in the range.

https://en.cppreference.com/w/cpp/named_req/RandomAccessIter...

they were conceived for c++'98 and transmogrified to a trait for c++20

the run time in GPs lower_bound link is log(n) for random access iterators and liner for non-random access.

[eesmith]

Yes.

But how does that change my observation about the differences in the two APIs?