[
https://issues.apache.org/jira/browse/TINKERPOP3-640?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Marko A. Rodriguez closed TINKERPOP3-640.
-----------------------------------------
Resolution: Fixed
This is done. However, I'm scared of the implementation. There is now a
{{HadoopPoolsConfigurable}} interface. All the Input/OutputFormats implement
it. Thus, whenever data comes into HadoopGraph (or out), the HadoopPool is
initialized. The only problem I see is when a JVM crashes and restarts. Will
the InputFormat be reconfigured? .................... For Giraph its easy to
have the {{WorkerContext}} initlize the pools. For Spark, I have no simple
place to put the intialization!?
> GryoPool Not Getting Vendor Serializer Registrations
> ----------------------------------------------------
>
> Key: TINKERPOP3-640
> URL: https://issues.apache.org/jira/browse/TINKERPOP3-640
> Project: TinkerPop 3
> Issue Type: Improvement
> Components: io
> Reporter: stephen mallette
> Assignee: Marko A. Rodriguez
> Fix For: 3.0.0.GA
>
>
> {{GryoPool}} is currently creating {{GraphReader}} and {{GraphWriter}}
> interfaces with the standard builder:
> {{GryoReader/Writer.build().create()}}. Since they are not being created
> with {{g.io()}} they are not getting custom serializers registered. Need to
> figure out how to get those serializers from the source {{Graph}} registered
> to the builders.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
