[
https://issues.apache.org/jira/browse/CASSANDRA-18836?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17764771#comment-17764771
]
Maxim Muzafarov commented on CASSANDRA-18836:
---------------------------------------------
I have checked the sources and found {{BufferedChecksumIndexInput}} which is
used in several places, but still initialized to the CRC32 (not the CRC32C in
jdk11+). This class is a part of the Lucene library itself, so it is not easy
to make it to use CRC32C instead of the default CRC32.
It is out of the scope mentioned in the issue description since we only have to
deal with the IndexFileUtils. It should not be a big problem to write our own
extension of the ChecksumIndexInput that uses CRC32C checksum algorithm
considering the fact that we already have our SAICodecUtils. The question here
- do we want to have our own ChecksumIndexInput that uses crc32c or is it
enough to just fix the IndexFileUtils?
> Replace CRC32 w/ CRC32C in IndexFileUtils.ChecksummingWriter
> ------------------------------------------------------------
>
> Key: CASSANDRA-18836
> URL: https://issues.apache.org/jira/browse/CASSANDRA-18836
> Project: Cassandra
> Issue Type: Improvement
> Components: Feature/2i Index, Feature/SAI
> Reporter: Caleb Rackliffe
> Priority: Normal
>
> It seems that now we're on Java 11 for 5.0, there isn't much reason not to
> use CRC32C as a drop-in replacement for CRC32. SAI isn't even released, so
> has no binary compatibility entanglements, and this should be pretty
> straightforward.
> See https://github.com/apache/bookkeeper/pull/3309
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
