[
https://issues.apache.org/jira/browse/SOLR-17461?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18042389#comment-18042389
]
David Smiley commented on SOLR-17461:
-------------------------------------
Sanjay, I sense maybe you had code excerpts or something that didn't print?
Hard to say.
This JIRA issue is mostly organizational / refactoring in nature. Thus
ClusterState would continue to benefit from the memory savings of the String
interner thing. I just don't want to see CoreContainer involved in it, and I
want no more than a line or so in ClusterState related to the matter. If you
can think of a solution within those constraints, maybe taking inspiration of
my vague proposal in the description, I'd like to see!
> (cleanup) Move ClusterState string "Interner" json parser to Utils
> ------------------------------------------------------------------
>
> Key: SOLR-17461
> URL: https://issues.apache.org/jira/browse/SOLR-17461
> Project: Solr
> Issue Type: Improvement
> Reporter: David Smiley
> Assignee: Noble Paul
> Priority: Minor
> Labels: newdev
>
> ClusterState.setStrInternerParser and its functionality is potentially of
> general utility and it's also a bit distracting in ClusterState. I also
> don't like CoreContainer being involved in initializing it (CoreContainer is
> doing too much!). It's doing it because ClusterState is in SolrJ without
> Caffeine being on the classpath. Instead, imagine a class in solr-core that
> implements this Function. Utils could then self-initialize via reflection
> detecting if that class is available, otherwise gracefully resorting to the
> non-intern mechanism. No touching CoreContainer which is way too busy doing
> many things.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
