[
https://issues.apache.org/jira/browse/HIVE-6769?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Eugene Koifman resolved HIVE-6769.
----------------------------------
Resolution: Won't Fix
no change required. On read the SerDe should provide the 'missing' column with
value NULL. This would be the same case as when a table is created with 5
columns and data written to it. Then 2 columns are added. Then a 'select *
form T' should still work.
> HCatStorer can initialize SerDe with less columns than table definition if
> PigSchema has less columns
> -----------------------------------------------------------------------------------------------------
>
> Key: HIVE-6769
> URL: https://issues.apache.org/jira/browse/HIVE-6769
> Project: Hive
> Issue Type: Bug
> Components: HCatalog
> Reporter: Jason Dere
>
> You should be able to see this in TestHCatStorer.testStoreFuncAllSimpleTypes.
> First a Hive table is created with 8 columns. Then HCatStorer is used to
> populate the table, but Pig relation only has data for the first 7 columns of
> the table. When HCat initializes the SerDe for writing the table, the schema
> only contains 7 columns rather than the 8 that the table should have. This
> test works without any errors, but gets an error if the table is changed to
> ORC storage (which can be fixed by HIVE-4975).
> HCat should initialize the Serde with the full number of columns in the
> table, and should fill in NULLs for any columns which are not provided by Pig
> relation.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
