[jira] [Commented] (SPARK-22248) spark marks all columns as null when its unable to parse single column

[ https://issues.apache.org/jira/browse/SPARK-22248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16202116#comment-16202116 ] Hyukjin Kwon commented on SPARK-22248: -- I think either way breaks existing support. JSON was started

[jira] [Commented] (SPARK-22248) spark marks all columns as null when its unable to parse single column

[ https://issues.apache.org/jira/browse/SPARK-22248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16201438#comment-16201438 ] Takeshi Yamamuro commented on SPARK-22248: -- yea, ok. Probably, you need to support both modes:

[jira] [Commented] (SPARK-22248) spark marks all columns as null when its unable to parse single column

[ https://issues.apache.org/jira/browse/SPARK-22248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16201433#comment-16201433 ] Gaurav Shah commented on SPARK-22248: - [~maropu] I am not sure on CSV, but on JSON we tokenize the

[jira] [Commented] (SPARK-22248) spark marks all columns as null when its unable to parse single column

[ https://issues.apache.org/jira/browse/SPARK-22248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16201309#comment-16201309 ] Takeshi Yamamuro commented on SPARK-22248: -- Once failed, I feel it is difficult to recover the