Oops yes. This is now fixed, the sentence I meant to write is: “In C++11, insertion adds elements at the upper bound of the range containing equivalent keys.”

Yes indeed, fixed now thanks.

A possible use case would be for a multi map that allows duplicates, but at some point we’d like to clean the multi map of the duplicate keys, for instance

Yes, indeed, I also noticed

If you are only considering keys (with potentially different associated values), then which key do you keep? With multi-map the ordering of equal keys is not defined (AFAIK). If the (key, value) pair is identical, then I see the value in removing them (but why where identical ones allowed to be inserted? – quicker to insert everything and remove later?). But if you are removing elements based only on key value, you’ll then end up with only 1 element for each key – so why use multi-map, why not just use map and report an error (or something) for duplicate keys??

multimap/multiset: [23.2.4/10] “The fundamental property of iterators of associative containers is that they iterate through the containers in the non-descending order of keys”
=> the standard ensures that the iteration is done in order and that then you can compare two consecutive elements.

unordered_multimap/set: [23.2.5/6] “In containers that support equivalent keys, elements with equivalent keys are adjacent to each other in the iteration order of the container.”
=> As you mentioned in the text, the standard ensures that equivalent keys will be seen as adjacent in iteration order.

1. first, that’s the fourth article in the serie.

2. “The other don’t have duplicates, by definition.” -> “the others” (with an s)

3. “Note that before C++11, we didn’t know which of the duplicates remain” -> “remains” (with an s)

4. “Here we make `it` point the second element of the container, and `previousIt` it to the first element.” -> “Here we make `it` point *to* the second element of the container, and `previousIt` -it- to the first element.” (“point to”, and remove an extra “it” after “previousIt”).

5. “To solve this we can provide default values for this policy, that would be correspond to the various cases.” -> remove “be”: “… that would correspond …”

6. “So we’ll have to make do with key_eq, and pass the keys to it ourselves.” -> remove “make”

I think the editing went slightly wrong here….