Re: [xsl] XQuery/XPath 3.1: Node List to Node Set ("distinct nodes")

["Michael Kay mike@xxxxxxxxxxxx" <xsl-list-service@xxxxxxxxxxxxxxxxxxxxxx>]

> So given this expression:
>
> fold-left($nodes, (), function($a, $n) { $a, $n except $a })
>
> I understand this to be iterating over $nodes from left to right, applying
the function function($a, $n) to each node, where $a is the next node and $n
is the accumulated value (being the result returned by the function on each
invocation).
>
> The b$a, $n except $ab in the function body constructs a new sequence of
($a, $n), excluding $a if it is already in $n. This sequence is then passed as
the second parameter of the next invocation of the function. This has the
effect of preserving the order of input node list.
>

Almost but not quite. $a is the set of distinct nodes found so far, $n is the
node currently being examined. So ($n except $a) is the node currently being
examined if it isn't present in $a, and is the empty sequence otherwise.

As Dimitre says, ($n except $a) where $n is a singleton isn't immediately
intuitive - but intuition takes practice!

As it happens, Saxon probably does a better job of optimizing the more complex
expression $n[not(. intersect $a)] -- though it's still O(n*m).

Michael Kay
Saxonica