auto firstDistinct(Range)(Range r, in size_t n)
if (isInputRange!Range) {
bool[ForeachType!Range] mySet;
return r.filter!((k) {
if (k in mySet)
return false;
mySet[k] = true;
return true;
}).take(n);
}
I think the standard library of Ada2012 has bounded containers,
so you can use them to stack-allocate a hash set that can contain
up to n items (plus some more free cells to keep its load factor
low enough).
In Rust there are fully safe stack-allocated closures. Combining
the two things it's possible to create a function like that
firstDistinct() with zero heap allocations, that is probably fast.
Bye,
bearophile