Github user a-roberts commented on the issue:
https://github.com/apache/spark/pull/15713
Good point, I'll first see if we can build and pass the unit tests by
replacing CompactBuffer with ArrayBuffer[2] and we'd also repeat the profiling
(this would verify if groupByKey is still hot for PageRank).
I'll keep the CompactBufferSuite for our experiment and replace the
CompactBuffer references with ArrayBuffer to verify we can still perform all of
the operations that the tests exercise.
My concern is that in comment for the CompactBuffer class we see the
downsides of said ArrayBuffer (the 16 entry Object array we'd get, so we'd
indeed want a default two element ArrayBuffer that only grows if needed --
which sounds like what we currently have but encapsulated within this class so
we can override two operators and grow the backing array as required)
---
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 infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
