[
https://issues.apache.org/jira/browse/FLINK-10839?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Stefan Richter updated FLINK-10839:
-----------------------------------
Description: The implementation of {{PojoSerializer.duplicate()}} currently
ignores that {{subclassSerializerCache}} is a shared resource. We must always
return a new instance of the serializer, even if all the nested field
serializers are not stateful or else concurrency problems might occur if we
have cached subclass serializers. (was: The implementation of
{{PojoSerializer.duplicate()}} currently ignores that subclassSerializerCache
is a shared resource. We must always return a new instance of the serializer,
even if all the nested field serializers are not stateful or else concurrency
problems might occur if we have cached subclass serializers.)
> Fix implementation of PojoSerializer.duplicate() w.r.t. subclass serializer
> ---------------------------------------------------------------------------
>
> Key: FLINK-10839
> URL: https://issues.apache.org/jira/browse/FLINK-10839
> Project: Flink
> Issue Type: Bug
> Components: Type Serialization System
> Reporter: Stefan Richter
> Assignee: Stefan Richter
> Priority: Blocker
> Labels: pull-request-available
> Fix For: 1.6.3, 1.7.0
>
>
> The implementation of {{PojoSerializer.duplicate()}} currently ignores that
> {{subclassSerializerCache}} is a shared resource. We must always return a new
> instance of the serializer, even if all the nested field serializers are not
> stateful or else concurrency problems might occur if we have cached subclass
> serializers.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
