[
https://issues.apache.org/jira/browse/HADOOP-18748?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17725338#comment-17725338
]
ASF GitHub Bot commented on HADOOP-18748:
-----------------------------------------
alkis commented on PR #5685:
URL: https://github.com/apache/hadoop/pull/5685#issuecomment-1558976068
> you know, if we could speed up the basic .get() it can only be good. at
the same time, this is such a broadly used piece of code, we are scared of
doing anything risky to it.
>
> what can we do in terms of testing that all this is good? I'm thinking of
concurrency as well as everything else?
That's a good point. I have specifically avoided a larger rewrite to reduce
the risks in breaking the implementation. `Configuration` is used pervasively
in most tests and there are specific tests for its basic and deprecated keys
usages. In terms of concurrency I also avoided changing it to avoid risks - the
concurrency is the same as before.
Can you advice on how to apply this change in other branches as well such
that older versions can receive the benefits?
> Configuration.get is slow
> -------------------------
>
> Key: HADOOP-18748
> URL: https://issues.apache.org/jira/browse/HADOOP-18748
> Project: Hadoop Common
> Issue Type: Improvement
> Components: conf
> Affects Versions: 3.3.5
> Reporter: Alkis Evlogimenos
> Priority: Major
> Labels: pull-request-available
>
> `Configuration.get` is slow mainly because of the overhead of
> `handleDeprecation` and eager creation of `overlay` even when null.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
