[
https://issues.apache.org/jira/browse/HIVE-637?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12731217#action_12731217
]
Zheng Shao commented on HIVE-637:
---------------------------------
The above approach does involve a single scan of the data which is unnecessary.
It shouldn't be a big concern right now because other parts of the process will
definitely take much longer time.
In the future (or the same transaction), we can modify LazySimpleSerDe to check
if the "column.delim" is an empty string or not. If it's an empty string, then
we will not scan the bytes array but return it directly in a Text.
> Add a simple way to create a blob table
> ---------------------------------------
>
> Key: HIVE-637
> URL: https://issues.apache.org/jira/browse/HIVE-637
> Project: Hadoop Hive
> Issue Type: New Feature
> Affects Versions: 0.3.0, 0.3.1
> Reporter: Zheng Shao
>
> A blob table has a single column of type string. We put all data from the row
> into that column.
> At present we are able to create blob table like this:
> {code}
> CREATE TABLE blobTable1 (row STRING)
> ROW FORMAT SERDE 'org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe'
> WITH SERDEPROPERTIES (
> 'serialization.last.column.takes.rest'='true'
> )
> STORED AS TEXTFILE;
> CREATE TABLE blobTable1 (row STRING)
> ROW FORMAT SERDE 'org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe'
> WITH SERDEPROPERTIES (
> 'serialization.last.column.takes.rest'='true'
> )
> STORED AS SEQUENCEFILE;
> {code}
> We should add a simpler way to create such a table, since it's pretty popular.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
