That remark is helpful, although a bit thin. So with this, the
relaxation should only be for result = first1.

Am So., 14. Sept. 2025 um 20:44 Uhr schrieb Ell <ell_se_at_[hidden]>:

> On 9/14/25 21:12, Peter Neiss wrote:
> > Thank you. This is a great example. I am not even sure what the correct
> > set intersection definition for weak ordering should be!
> > If an element is in the intersection if it there exist elements a in A
> > and b in B with equiv(a,b) == true, then the answer should be {1, 2, 3,
> 4}
> > If an element has to be "the same value" and not only equivalent, the
> > answer is {2,3}
> > In math we could define the intersection on "equivalence <https://
> > dict.leo.org/german-english/equivalence> class <https://dict.leo.org/
> > german-english/class>es[math. sic]", not helpful here i believe.
> > Is it enough if we take one arbitrary represantant of "equivalence
> > <https://dict.leo.org/german-english/equivalence> class <https://
> > dict.leo.org/german-english/class>es[math. sic]", then {1}, {2}, {3} and
> > {4} would be a valid answer.
> > set_intersection might be broken for weakly ordering.
>
>
> set_intersection() already specifies the specific elements to take when
> there are multiple equivalent elements (basically, treat the inputs as
> multisets, and take the elements from the first range in order):
>
> https://eel.is/c++draft/set.intersection#6
>