[
https://issues.apache.org/jira/browse/TINKERPOP-1111?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Marko A. Rodriguez closed TINKERPOP-1111.
-----------------------------------------
Resolution: Invalid
My assumption were wrong. The problem is that Spark's aggregators use
JavaSerialization (not the specified serialization). Internally, I've fixed
this by wrapping all aggregators in {{ObjectWritable}} which uses Gryo
serialization on {{writeObject}} and {{readObject}}.
> Gryo and nested objects not serializing using GryoSerializers
> -------------------------------------------------------------
>
> Key: TINKERPOP-1111
> URL: https://issues.apache.org/jira/browse/TINKERPOP-1111
> Project: TinkerPop
> Issue Type: Improvement
> Components: structure
> Affects Versions: 3.1.1-incubating
> Reporter: Marko A. Rodriguez
>
> If you have a {{HashMap}} filled with objects that are Kryo serializable
> (that is, the objects have a registered {{Serializer}}), it will try and
> serialize them as Java serialization objects.
> I believe this is because of {{add(Triplet.<Class, Function<Kryo,
> Serializer>, Integer>with(HashMap.class, null, 11));}} means use
> {{JavaSerializer}} and thus, Kryo serialization ends there. ?? .... :|
> I'm noticing this when I have a {{HashMap}} full of {{ComputerVertex}}
> objects which implement {{Vertex}}, but I'm get a "ComputerVertex not
> serializable exception" when it should be converted to a {{DetachedVertex}}
> and serialized. ???
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
