2008/11/18 Michael Kay <mike@xxxxxxxxxxxx>:

for $d in distinct-values($seq) return $d[count($seq[. eq $d]) ge $i]

They are both O(n^2).

$vSeq[index-of($vSeq,.)[$i]]

...would be O(n^2) for both best and worst cases - right?

With $vSeq being immutable, wouldn't the expression
index-of( $vSeq, $i) be memoized after the first evaluation?

Isn't that one of the advantages of the functional paradigm?
(Not that I'm qualified enough to know this - just asking.)