Reduce

Runs a query on each group that a Map formed, and merges the results.

Reduce(reducer, resultSelector) follows a Map(mapper, keySelector). It runs the reducer on each group’s stream, then passes the group and each result to the result selector. The selector’s first argument holds the group’s key as Key.

In the diagram, the mapper passes the input through, the key is Key, and the reducer counts the events alive in each group at each moment. The count for a is 1, then 2 while both a events are alive, then 1 again. The count for b is 1. The output arrives as start and end edges.

GroupApply(keySelector, apply, resultSelector) is built on these operators: it is Map(keySelector) followed by Reduce(apply, resultSelector), with a mapper that passes the input through. Reduce() with no arguments ends a Map that has no key selector, as on the Map page. Another overload reduces the groups of two Maps with the same key type together.

Known issue in this fork. After a Map with a key selector, Reduce() with no arguments throws a NullReferenceException when the query is subscribed, instead of refusing the query when it is built. Give that Reduce a reducer and a result selector.

See also: Map, GroupApply, Count (tumbling window)