Github user aljoscha commented on a diff in the pull request:
https://github.com/apache/flink/pull/2736#discussion_r86379858
--- Diff:
flink-streaming-java/src/main/java/org/apache/flink/streaming/runtime/operators/windowing/EvictingWindowOperator.java
---
@@ -273,32 +284,93 @@ public void onProcessingTime(InternalTimer<K, W>
timer) throws Exception {
return;
}
- TriggerResult triggerResult =
context.onProcessingTime(timer.getTimestamp());
- if (triggerResult.isFire()) {
- fire(context.window, contents);
- }
+ TriggerResult triggerResult =
context.onProcessingTime(timer.getTimestamp());
+ if (triggerResult.isFire()) {
+ fire(context.window, contents,
windowState);
+ }
if (triggerResult.isPurge() || (!windowAssigner.isEventTime()
&& isCleanupTime(context.window, timer.getTimestamp()))) {
cleanup(context.window, windowState, mergingWindows);
}
}
- private void fire(W window, Iterable<StreamRecord<IN>> contents) throws
Exception {
+ private void fire(W window, Iterable<StreamRecord<IN>> contents,
ListState<StreamRecord<IN>> windowState) throws Exception {
timestampedCollector.setAbsoluteTimestamp(window.maxTimestamp());
// Work around type system restrictions...
- int toEvict = evictor.evict((Iterable) contents,
Iterables.size(contents), context.window);
-
- FluentIterable<IN> projectedContents = FluentIterable
+ FluentIterable<TimestampedValue<IN>> recordsWithTimestamp =
FluentIterable
.from(contents)
- .skip(toEvict)
- .transform(new Function<StreamRecord<IN>, IN>() {
+ .transform(new Function<StreamRecord<IN>,
TimestampedValue<IN>>() {
+ @Override
+ public TimestampedValue<IN>
apply(StreamRecord<IN> input) {
+ return new
TimestampedValue<>(input.getValue(), input.getTimestamp());
+ }
+ });
+ evictorContext.evictBefore(recordsWithTimestamp,
Iterables.size(recordsWithTimestamp));
+
+ FluentIterable<IN> projectedContents = recordsWithTimestamp
+ .transform(new Function<TimestampedValue<IN>, IN>() {
@Override
- public IN apply(StreamRecord<IN> input) {
+ public IN apply(TimestampedValue<IN> input) {
return input.getValue();
}
});
+
userFunction.apply(context.key, context.window,
projectedContents, timestampedCollector);
+ evictorContext.evictAfter(recordsWithTimestamp,
Iterables.size(recordsWithTimestamp));
+
+
+ //work around to fix FLINK-4369, remove the evicted elements
from the windowState.
+ //this is inefficient, but there is no other way to remove
elements from ListState, which is an AppendingState.
+ windowState.clear();
+ for(TimestampedValue<IN> record : recordsWithTimestamp) {
+ if (record.getTimestamp() < 0) {
--- End diff --
I think this does not work properly in all cases because negative
timestamps are in fact also valid timestamps. You can look at `StreamRecord`
for how this should be solved in `TimestampedValue` and also for how the
copying between the two should work.
---
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.
---
