[
https://issues.apache.org/jira/browse/ORC-1251?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17583687#comment-17583687
]
Steve Loughran commented on ORC-1251:
-------------------------------------
bq. while keeping Apache Hadoop 2.7.3+ support.
whats the minimum java version ORC builds against? as if anyone really does
want to run orc 1.8.0 on a hadoop 2.7.3 system, it'd have to be built with java
7. if anyone is targeting java 11, they should be looking at hadoop 3.2.0 as a
baseline.
As for the hadoop version here, the new API will be in 3.3.5, though mukund has
promised he will add a bridging api in our WiP shim library
https://github.com/steveloughran/fs-api-shim which will use reflection to
invoke when present, and otherwise implement as the default vectored iO
implementation does: implement the vectored reads as a sequence of readFully
calls.
> Support VectoredIO in ORC
> -------------------------
>
> Key: ORC-1251
> URL: https://issues.apache.org/jira/browse/ORC-1251
> Project: ORC
> Issue Type: Improvement
> Reporter: Harshit Gupta
> Priority: Major
>
> ORC changes to support the new vectored read API.
>
> /cc : [~rbalamohan] , [~abstractdog] , [~mthakur]
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
