Doc: improve description of window function processing.

The previous wording talked about a "single pass over the data",
which can be read as promising more than intended (to wit, that only
one WindowAgg plan node will be used).  What we promise is only what
the SQL spec requires, namely that the data not get re-sorted between
window functions with compatible PARTITION BY/ORDER BY clauses.
Adjust the wording in hopes of making this clearer.

Reported-by: Christopher Inokuchi 
Author: Tom Lane 
Reviewed-by: David G. Johnston 
Discussion: https://postgr.es/m/CABde6B5va2wMsnM79u_x=n9KUgfKQje_pbLROEBmA9Ru5XWidw@mail.gmail.com
Backpatch-through: 13

diff --git a/doc/src/sgml/queries.sgml b/doc/src/sgml/queries.sgml
index 372cce1a487e35ce7974703e6c2fd7c98c222488..a326960ff4dfb714daa30644a94f1a1774601a8d 100644 (file)
--- a/doc/src/sgml/queries.sgml
+++ b/doc/src/sgml/queries.sgml
@@ -1461,10 +1461,10 @@ GROUP BY GROUPING SETS (
 
    
     When multiple window functions are used, all the window functions having
-    syntactically equivalent PARTITION BY and ORDER BY
-    clauses in their window definitions are guaranteed to be evaluated in a
-    single pass over the data. Therefore they will see the same sort ordering,
-    even if the ORDER BY does not uniquely determine an ordering.
+    equivalent PARTITION BY and ORDER BY
+    clauses in their window definitions are guaranteed to see the same
+    ordering of the input rows, even if the ORDER BY does
+    not uniquely determine the ordering.
     However, no guarantees are made about the evaluation of functions having
     different PARTITION BY or ORDER BY specifications.
     (In such cases a sort step is typically required between the passes of