You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@reef.apache.org by "Markus Weimer (JIRA)" <ji...@apache.org> on 2017/04/04 19:53:41 UTC
[jira] [Commented] (REEF-1765) Building a Parquet Reader for
Potential Integrations with Other ML Frameworks
[ https://issues.apache.org/jira/browse/REEF-1765?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15955706#comment-15955706 ]
Markus Weimer commented on REEF-1765:
-------------------------------------
From the draft PR, it looks like this adds the new Java code to the {{reef-io}} package, right? That is problematic, as it also needs to add another dependency to that package (parquet). Would it be feasible to move this code into its own sub-module?
> Building a Parquet Reader for Potential Integrations with Other ML Frameworks
> -----------------------------------------------------------------------------
>
> Key: REEF-1765
> URL: https://issues.apache.org/jira/browse/REEF-1765
> Project: REEF
> Issue Type: New Feature
> Components: REEF.NET IO
> Affects Versions: 0.16
> Reporter: Shouheng Yi
> Priority: Minor
> Labels: features
> Original Estimate: 336h
> Remaining Estimate: 336h
>
> Parquet file format is very common in some well-known frameworks like Hadoop and Spark. By enabling REEF to read parquet file, we could potentially integrate with those frameworks. Currently we want to only support data of non-nested types with a table-like property. This allows us to transform the data into formats like RDDs, etc.
> A draft of ParquetReader is provided here in a PR: https://github.com/apache/reef/pull/1283
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)