sunhaibotb commented on a change in pull request #8409: [FLINK-12478] Decompose
monolithic run-loops in StreamTask implementa…
URL: https://github.com/apache/flink/pull/8409#discussion_r284733595
##########
File path:
flink-streaming-java/src/main/java/org/apache/flink/streaming/runtime/tasks/OneInputStreamTask.java
##########
@@ -98,13 +96,8 @@ public void init() throws Exception {
}
@Override
- protected void run() throws Exception {
- // cache processor reference on the stack, to make the code
more JIT friendly
- final StreamInputProcessor<IN> inputProcessor =
this.inputProcessor;
-
- while (running && inputProcessor.processInput()) {
- // all the work happens in the "processInput" method
- }
+ protected boolean performDefaultAction() throws Exception {
Review comment:
In theory, I don't think it's possible to get faster because I don't see the
optimization that might get faster in this change, at most it is the same as
before. Feeling some kind of interference makes benchmark inaccurate, or this
benchmarking can not accurately measure their performance. In my experience, on
the contrary, it will make one more virtual call than before (as @pnowojski
said), because JIT do not inline `performDefaultAction()` which has more than
one implementations. Of course, the accurate benchmark results should be the
final criterion.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
