[
https://issues.apache.org/jira/browse/SOLR-14700?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18083979#comment-18083979
]
David Smiley commented on SOLR-14700:
-------------------------------------
I strongly disagree here with the direction taken. IMO streaming expressions
should *insist* on a StreamContext for a deterministic / documented lifecycle
so that these many expressions needn't support multiple lifecycle
(context-less). Many streaming expressions having checks like -- {{if
(this.solrClientCache == null)}} then create one via {{new SolrClientCache()}}.
StreamContext should have insisted on a SolrClientCache ( SOLR-12718 ) and
lifecycle managed by the caller so that these expressions don't have tedious
code to deal with. There are security implications as well but I won't get
into that here.
> CloudSolrClient cannot be used without setting an explicit streamContext
> ------------------------------------------------------------------------
>
> Key: SOLR-14700
> URL: https://issues.apache.org/jira/browse/SOLR-14700
> Project: Solr
> Issue Type: Bug
> Components: clients - java
> Affects Versions: 8.6
> Reporter: Mads Bondo Dydensborg
> Priority: Minor
> Fix For: 8.7, 9.0
>
> Time Spent: 1h 10m
> Remaining Estimate: 0h
>
> A naive approach to using CloudSolrClient, such as on this page:
> [http://joelsolr.blogspot.com/2015/04/the-streaming-api-solrjio-basics.html]
> - will not work: opening the stream will throw an IOException with an
> embedded NullPointerException.
> This took me a while to figure out, and I think it would be good to either
> update the documentation or "fix" the constructor to set the needed
> streamContext as appropriate.
> I found this snippet of code in TuppleStream.java, that seems to "almost"
> work without a streamContext:
> {code:java}
> // line 127 in the sources to TuppleStream.java:
> public static List<String> getShards(String zkHost,
> String collection,
> StreamContext streamContext,
> SolrParams requestParams)
> ...
> // line 135
> if(streamContext != null) {
> shardsMap = (Map<String, List<String>>)streamContext.get("shards");
> }
> ...
> // line 144
> CloudSolrClient cloudSolrClient =
>
> Optional.ofNullable(streamContext.getSolrClientCache()).orElseGet(SolrClientCache::new).getCloudSolrClient(zkHost);
> {code}
> if streamContext is null in the last statement, a NullPointerException will
> be thrown and open'ening the stream will fail.
> This took me way too long time to figure out.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
