[
https://issues.apache.org/jira/browse/RATIS-2049?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17829907#comment-17829907
]
Tsz-wo Sze commented on RATIS-2049:
-----------------------------------
!peerId.jpg!
The assumption from Sonar obviously is invalid in the code above since
RaftPeerId is immutable.
We may go through other code pointed out from Sonar to see if they are valid.
In my past experience, Sonar generates too many false positives.
> Ensure atomic safety for volatile object
> ----------------------------------------
>
> Key: RATIS-2049
> URL: https://issues.apache.org/jira/browse/RATIS-2049
> Project: Ratis
> Issue Type: Improvement
> Reporter: Xinyu Tan
> Assignee: Xinyu Tan
> Priority: Major
> Attachments: S3077.jpg, image-2024-03-22-17-00-53-814.png, peerId.jpg
>
>
> Currently, there are about 40 bugs on Sonar for
> [Ratis|[https://sonarcloud.io/summary/new_code?id=apache-ratis]], with more
> than half of these issues stemming from the use of volatile fields to
> decorate objects rather than primitive data types.
> Theoretically, volatile is indeed intended for use with primitive data types
> only, as decorating objects can lead to visibility issues. To avoid certain
> concurrency problems, using AtomicReference is recommended. However, it seems
> that for certain types of objects that are not modified, this usage does not
> result in bugs?
> I'm not sure what the Ratis community's stance is on this code style. If
> there are confirmed rules and some code that could be optimized, I'm willing
> to undertake these tasks to avoid potential concurrency issues and eliminate
> Sonar bugs.
> !image-2024-03-22-17-00-53-814.png!
> [~szetszwo] What's your opinion?
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
