On Saturday, 16 May 2020 at 08:11:20 UTC, Jan Hönig wrote: > I am convinced that this works. I have another implementation, which should work: https://run.dlang.io/is/tfBgD0 It is not pretty. It is probably not fast, if the `ignore` is too large. I guess it is in O(n^2) if not O(mn^2), where m is the number of `same` structs.