[
https://issues.apache.org/jira/browse/SOLR-9842?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15746922#comment-15746922
]
Mark Miller commented on SOLR-9842:
-----------------------------------
Yeah I guess AutoClosable is now preferred.
I had thought about deprecating finish, but I wasn't sure it was easy to have
the tolerant processor work correctly with a single final method. In the end it
seemed simpler to keep considering finish as containing logic that is part of
the request - failures are captured by the tolerant processor and close as just
for releasing resources - if releasing a resource fails, log an issue, but
don't consider the request as having been unsuccessful.
If you have an idea though, let me know.
> UpdateRequestProcessors have no way to guarantee the closing of resources
> used for a request.
> ---------------------------------------------------------------------------------------------
>
> Key: SOLR-9842
> URL: https://issues.apache.org/jira/browse/SOLR-9842
> Project: Solr
> Issue Type: Improvement
> Security Level: Public(Default Security Level. Issues are Public)
> Reporter: Mark Miller
> Assignee: Mark Miller
> Attachments: SOLR-9842.patch
>
>
> You cannot count on finish being called for an UpdateRequestProcessor because
> whether or not it's called in the face of errors is dependent on the other
> processors in the chain.
> We should make UpdateRequestProcessor closeable and implement this so that
> close is guaranteed to be called. I'm playing around with this in a patch for
> another issue.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
