[jira] [Commented] (HADOOP-18530) ChecksumFileSystem::readVectored might return byte buffers not positioned at 0

["Steve Loughran (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HADOOP-18530?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17634937#comment-17634937 ] 

Steve Loughran commented on HADOOP-18530:
-----------------------------------------

apparently the s3a connector gets this right, but not the vector support lib -it should be updating the position and limit of the source buffer before calling split()

* i want this in to 3.3.5 so its correct from the outset; avoids us having a release which doesn't work properly with orc and all the complications which follow
* + mention in fs specification and contract test validation

> ChecksumFileSystem::readVectored might return byte buffers not positioned at 0
> ------------------------------------------------------------------------------
>
>                 Key: HADOOP-18530
>                 URL: https://issues.apache.org/jira/browse/HADOOP-18530
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: fs
>    Affects Versions: 3.3.5
>            Reporter: Harshit Gupta
>            Priority: Blocker
>
> Checksystem::readVectored method returns the byte buffers that are not positioned at 0, which might be the underlying assumption for the readers like ORC.
>  
> cc:/ [~mthakur] [~rbalamohan] [~stevel@apache.org]



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: common-issues-help@hadoop.apache.org