[
https://issues.apache.org/jira/browse/SPARK-916?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Patrick Wendell resolved SPARK-916.
-----------------------------------
Resolution: Fixed
Turned out [~marmbrus] did all of this and more in SparkSQL (which btw also
works for nested types). So I'm gonna close this very old issue.
> Better Support for Flat/Tabular RDD's
> -------------------------------------
>
> Key: SPARK-916
> URL: https://issues.apache.org/jira/browse/SPARK-916
> Project: Spark
> Issue Type: Improvement
> Reporter: Patrick Cogan
>
> Many people use Spark to run analysis on flat datasets, where the RDD is
> composed records with a single set of non-nested fields. We could have better
> support for this use case in a variety of areas. Two of which are:
> 1. Allowing people to name individual fields and access them by name, rather
> than using tuple indices (see Scalding[1]). This avoids the mess that is {x
> => (x._3(), x._4())}
> 2. Support columnar in-memory storage.
> This is just an umbrella/brainstorming JIRA to see if other people have
> thoughts about this. Curious to hear feedback.
> [1] https://dev.twitter.com/blog/scalding
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
