Stephan Ewen created FLINK-17671:
------------------------------------
Summary: Simplify the ManuallyTriggeredScheduledExecutortriggering
logic
Key: FLINK-17671
URL: https://issues.apache.org/jira/browse/FLINK-17671
Project: Flink
Issue Type: Improvement
Components: Tests
Reporter: Stephan Ewen
Assignee: Stephan Ewen
Fix For: 1.11.0
Currently, the ManuallyTriggeredScheduledExecutor for testing of actor-style
components uses a complex dance to execute queued tasks. Even though the tasks
are synchronously executed, they get get delegated via ComplatableFuture, using
async supply of an execution stage, but with a synchronous executor injected
via a lambda.
Sounds complicated, and it is. And for no reason, as far as I can tell. It does
just as much (or as little) as directly calling {{run()}} on the enqueued task.
The advantage simply calling {{run()}} is that it is very easy to trace and for
the debugger to step into. For the complex dance above, it becomes rather
laborious and involved to try and find your way into the actual {{Runnable}}
execution.
This fix proposes to simplify the trigger() method to call {{run()}} directly.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
