Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/1024#issuecomment-131577837
Is this really in need of optimization? It is written failsafe, where it
does not crash even if `state` does not handle null values properly. This is
good, because it helps prevent future bugs. The performance cost is minimal.
Concerning performance: This is an actor message send and processed
asynchronously by the JobManager. These code paths should never saturate the
CPU in any reasonable setup, and even the computation during the processing of
this message is dominated by other parts (cache misses during hash table
lookups). Checking more, this method is never even executed. It is there only
for the sake of testing / debugability.
Please double check whether statements really need fixing. Eliminating
checks for the sake of performance is really only necessary on the most
performance critical code paths (the inner loops of the runtime operators and
algorithms).
---
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.
---
