pnikic-db commented on code in PR #55828:
URL: https://github.com/apache/spark/pull/55828#discussion_r3233592963
##########
sql/api/src/main/scala/org/apache/spark/sql/functions.scala:
##########
@@ -2531,6 +2531,63 @@ object functions {
// Window functions
//////////////////////////////////////////////////////////////////////////////////////////////
+ /**
+ * Window function: computes the differences between consecutive cumulative
counter values in a
+ * time series, thereby converting the counter from the cumulative to the
delta format.
+ *
+ * Gracefully handles counter resets by returning NULL. Counter resets are
detected when the
+ * counter value decreases.
+ *
+ * Use the PARTITION BY clause of the window to separate independent
counters. This is done by
+ * specifying all columns which uniquely identify a time series. These are
typically the counter
+ * name and any attributes tied to the counter.
+ *
+ * Use the ORDER BY clause of the window to order the observations by the
associated timestamp
+ * in ascending order.
+ *
+ * @param value
+ * A cumulative counter. Must be a numeric data type. Must be non-negative.
+ *
+ * @return
+ * The difference between the current and previous counter value within
the window partition,
+ * according to the order defined by the window's ORDER BY clause.
+ *
+ * @group window_funcs
+ * @since 4.2.0
+ */
+ def counter_diff(value: Column): Column = Column.fn("counter_diff", value)
+
+ /**
+ * Window function: returns the difference between consecutive cumulative
counter values in a
+ * time series. With this operation, the counter is converted from
cumulative to delta format.
+ *
+ * Gracefully handles counter resets by returning NULL. Counter resets are
detected when the
+ * counter value decreases, or when the start time advances between rows.
+ *
+ * Use the PARTITION BY clause of the window to separate independent
counters. This is done by
+ * specifying all columns which uniquely identify a time series. These are
typically the counter
+ * name and any attributes tied to the counter.
+ *
+ * Use the ORDER BY clause of the window to order the observations by the
associated timestamp
+ * in ascending order.
+ *
+ * @param value
+ * A cumulative counter. Must be a numeric data type. Must be non-negative.
+ *
+ * @param startTime
+ * An optional timestamp parameter which indicates when the counter was
last set to zero. Used
+ * to signal counter resets.
Review Comment:
Updated the description. The start time in the overload's description is no
longer marked as optional.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
