You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Gengliang Wang (JIRA)" <ji...@apache.org> on 2018/04/22 13:53:00 UTC
[jira] [Updated] (SPARK-24045) Create base class for file data
source v2
[ https://issues.apache.org/jira/browse/SPARK-24045?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Gengliang Wang updated SPARK-24045:
-----------------------------------
Description:
From https://github.com/apache/spark/pull/20933, we can see that during File data source V2 migration, falling back to V1 implementation is necessary in following cases1. File datasource V2 might be implemented partially during migration.
E.g. if [[ReadSupport]] is implemented while [[WriteSupport]] is not,
write path should fall back to V1 implementation.
2. File datasource V2 implementations cause regression.
3. Catalog support is required, which is still under development for data source V2.
Create a base class FileDataSourceV2 and resolve the falling back problems.
was:
During File data source V2 migration, we need to fall back to V1 implementation in following cases:
1. File datasource V2 might be implemented partially during migration.
E.g. if [[ReadSupport]] is implemented while [[WriteSupport]] is not,
write path should fall back to V1 implementation.
2. File datasource V2 implementations cause regression.
3. Catalog support is required, which is still under development for data source V2.
Create a base class FileDataSourceV2 and resolve the falling back problems.
> Create base class for file data source v2
> -----------------------------------------
>
> Key: SPARK-24045
> URL: https://issues.apache.org/jira/browse/SPARK-24045
> Project: Spark
> Issue Type: Sub-task
> Components: SQL
> Affects Versions: 2.2.1
> Reporter: Gengliang Wang
> Priority: Major
>
> From https://github.com/apache/spark/pull/20933, we can see that during File data source V2 migration, falling back to V1 implementation is necessary in following cases1. File datasource V2 might be implemented partially during migration.
> E.g. if [[ReadSupport]] is implemented while [[WriteSupport]] is not,
> write path should fall back to V1 implementation.
> 2. File datasource V2 implementations cause regression.
> 3. Catalog support is required, which is still under development for data source V2.
>
> Create a base class FileDataSourceV2 and resolve the falling back problems.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org