[
https://issues.apache.org/jira/browse/PHOENIX-2492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15043351#comment-15043351
]
ASF GitHub Bot commented on PHOENIX-2492:
-----------------------------------------
Github user jmahonin commented on the pull request:
https://github.com/apache/phoenix/pull/142#issuecomment-162212491
This is a good idea. There are actually a few implementations of DBWritable
in the codebase:
- PhoenixRecordWritable
- PhoenixPigDBWritable
- PhoenixIndexDBWritable
The code seems very similar across implementations, so we definitely should
have one base implementation that works well across the board and can be simply
extended, if need be. Having it in 'phoenix-core' is a bonus too.
I'd like to see the phoenix-spark 'PhoenixRecordWritable' renamed to maybe
'PhoenixSparkRecordWritable' and have it trimmed down to implement this new
class, but maybe that's better suited to a follow-up PR.
> Expose PhoenixRecordWritable outside of spark
> ---------------------------------------------
>
> Key: PHOENIX-2492
> URL: https://issues.apache.org/jira/browse/PHOENIX-2492
> Project: Phoenix
> Issue Type: Task
> Affects Versions: 4.6.0
> Reporter: Nick Dimiduk
> Assignee: Nick Dimiduk
> Priority: Minor
> Fix For: 4.7.0
>
>
> {{PhoenixOutputFormat}} provides an easy way to write to phoenix from MR
> jobs, or anything else that support's hadoop's {{OutputFormat}}
> implementations. It depends on an implementation of {{DBWritable}}. We
> actually provide an implementation of {{DBWritable}} that's specific to
> Phoenix, but it's packaged in {{phoenix-spark}}. Let's move this out into
> {{phoenix-core}} so downstreamers don't need to pull in the whole spark/scala
> workd to use it.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
