[GitHub] [geode] jhuynh1 commented on pull request #3196: GEODE-6404: work around possible sync issue with computeIfAbsent

["jhuynh1 (GitHub)" <gi...@apache.org>]

This was discovered in an adhoc test where we were unable to scale client queries.  The ticket has the stack from the thread dumps which helped identify the issue.
The cost to compute the first time will increase slightly as you stated, but if it's not called very often to get the benefit of a cache, is the performance degrade enough to 'matter'?  

I can limit this diff to only affect the two calls that I can see hampering our most common use cases, I just thought making them all the same would make it more consistent across the code base? 

A JMH benchmark can be written to hammer these methods directly, but that will just prove that the issues with computeIfAbsent  (in Pre Java 9) are blocking/slower when the value is expected to be present.

[ Full content available at: https://github.com/apache/geode/pull/3196 ]
This message was relayed via gitbox.apache.org for notifications@geode.apache.org