Github user brkyvz commented on a diff in the pull request:
https://github.com/apache/spark/pull/15949#discussion_r88941686
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/StreamExecution.scala
---
@@ -422,6 +432,7 @@ class StreamExecution(
val replacementMap = AttributeMap(replacements)
val triggerLogicalPlan = withNewSources transformAllExpressions {
case a: Attribute if replacementMap.contains(a) => replacementMap(a)
+ case t: CurrentTimestamp =>
CurrentBatchTimestamp(currentBatchTimestamp)
--- End diff --
@hvanhovell Timestamp is "non-deterministic" when it comes to streaming
aggregations. Therefore it shouldn't be pushed down stateful aggregations.
Otherwise the resulting table you get is different than what you asked for.
Example:
```
df.groupBy('time).agg(count("*")).where('time >=
current_timestamp().cast("long") - 30 * 60)
```
What you get:
All aggregates from all of time, but the aggregates older than 30 minutes
don't get updated. Those get filtered
What I wanted:
Just return me the last 30 minutes of aggregates. I don't care about
anything else older.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
