Github user tillrohrmann commented on a diff in the pull request:
https://github.com/apache/flink/pull/5168#discussion_r157482818
--- Diff:
flink-runtime/src/main/java/org/apache/flink/runtime/dispatcher/Dispatcher.java
---
@@ -357,6 +362,28 @@ public void start() throws Exception {
return
CompletableFuture.completedFuture(jobManagerServices.blobServer.getPort());
}
+ @Override
+ public CompletableFuture<Either<Throwable,
SerializedJobExecutionResult>> getJobExecutionResult(
--- End diff --
I'm actually wondering why we are differentiating between execution success
and failure that much. Also in the failure case we should have a net runtime
and accumulator values. The only difference is that in the failure case we have
an additional `SerializedThrowable`.
We could for example have a common base class with a
`JobExecutionResultSuccess` and `JobExecutionResultFailure` specialization
(maybe also `JobExecutionResultCancelled` as a sub class of
`JobExecutionResultFailure`). What do you think?
---
