[
https://issues.apache.org/jira/browse/HIVE-16222?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15941033#comment-15941033
]
Matt McCline commented on HIVE-16222:
-------------------------------------
[~sershe] I'd like to see some Q file cases showing an attempt to use the
org.apache.parquet.hadoop.ParquetInputFormat input file format and it not being
allowed.
And detail in the not met conditions showing why row-mode was not used...
{code}
enabled: false
enabledConditionsNotMet: hive.vectorized.use.row.serde.deserialize IS
true AND hive.vectorized.row.serde.inputformat.excludes
[org.apache.parquet.hadoop.ParquetInputFormat] NOT CONTAINS
org.apache.parquet.hadoop.ParquetInputFormat IS false
groupByVectorOutput: true
inputFileFormats: org.apache.parquet.hadoop.ParquetInputFormat
{code}
> add a setting to disable row.serde for specific formats; enable for others
> --------------------------------------------------------------------------
>
> Key: HIVE-16222
> URL: https://issues.apache.org/jira/browse/HIVE-16222
> Project: Hive
> Issue Type: Bug
> Reporter: Sergey Shelukhin
> Assignee: Sergey Shelukhin
> Attachments: HIVE-16222.01.patch, HIVE-16222.02.patch,
> HIVE-16222.03.patch, HIVE-16222.patch
>
>
> Per [~gopalv]
> {quote}
> row.serde = true ... breaks Parquet (they expect to get the same object back,
> which means you can't buffer 1024 rows).
> {quote}
> We want to enable this and vector.serde for text vectorization. Need to turn
> it off for specific formats.
>
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
