[
https://issues.apache.org/jira/browse/PARQUET-160?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Alex Levenson resolved PARQUET-160.
-----------------------------------
Resolution: Fixed
Fix Version/s: parquet-mr_1.6.0
Issue resolved by pull request 98
[https://github.com/apache/incubator-parquet-mr/pull/98]
> Simplify CapacityByteArrayOutputStream
> --------------------------------------
>
> Key: PARQUET-160
> URL: https://issues.apache.org/jira/browse/PARQUET-160
> Project: Parquet
> Issue Type: Improvement
> Reporter: Alex Levenson
> Priority: Minor
> Fix For: parquet-mr_1.6.0
>
>
> There's a lot of complicated logic around trying to re-use byte buffers as
> well as allocating slabs before they are needed as well as shrinking /
> consolidating.
> After discussing with [~julienledem] we should simplify and not bother
> reusing the byte buffers, and not allocate buffers until they are needed. The
> hope is this will reduce the memory overhead and that the number of byte
> buffers should be small so they should not cause trouble for GC.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)