You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@parquet.apache.org by "Junjie Chen (Jira)" <ji...@apache.org> on 2019/09/19 02:29:00 UTC

[jira] [Commented] (PARQUET-1657) [C++] Change Bloom filter implementation to use xxhash

    [ https://issues.apache.org/jira/browse/PARQUET-1657?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16932998#comment-16932998 ] 

Junjie Chen commented on PARQUET-1657:
--------------------------------------

Great, the Bloom filter thrift definition was agreed and It uses the same variant in arrow now (64-bit xxh3). It should be ok to remove murmur3 hash relative files as well as the unit test file. After releasing parquet-format 2.7.0, we will update the binary file in parquet-testing as well.

> [C++] Change Bloom filter implementation to use xxhash
> ------------------------------------------------------
>
>                 Key: PARQUET-1657
>                 URL: https://issues.apache.org/jira/browse/PARQUET-1657
>             Project: Parquet
>          Issue Type: Improvement
>          Components: parquet-cpp
>            Reporter: Wes McKinney
>            Priority: Major
>             Fix For: cpp-1.6.0
>
>
> I also strongly recommend doing away with the virtual function calls if possible. We have vendored xxhash in Apache Arrow so we should also remove the murmur3 code while we are at it



--
This message was sent by Atlassian Jira
(v8.3.4#803005)