Dear Wiki user,

You have subscribed to a wiki page or wiki category on "Hadoop Wiki" for change 
notification.

The "Hive/PartitionedViews" page has been changed by JohnSichi.
http://wiki.apache.org/hadoop/Hive/PartitionedViews?action=diff&rev1=10&rev2=11

--------------------------------------------------

  
  Currently, changing a view definition requires dropping the view and 
recreating it.  This implies dropping and recreating all existing partitions as 
well, which could be very expensive.
  
- This implies that followup support for 
[[https://issues.apache.org/jira/browse/HIVE-1078|CREATE OR REPLACE]] is very 
important, and that it needs to preserve existing partitions (after validating 
that they are still compatible with the new view definition).
+ This implies that followup support for 
[[https://issues.apache.org/jira/browse/HIVE-1078|CREATE OR REPLACE VIEW]] is 
very important, and that it needs to preserve existing partitions (after 
validating that they are still compatible with the new view definition).
  
+ = Hook Information =
+ 
+ Although there is currently no connection between the view partition and 
underlying table partitions, Hive does provide dependency information as part 
of the hook invocation for ALTER VIEW ADD PARTITION.  It does this by compiling 
an internal query of the form
+ 
+ {{{
+ SELECT * FROM view_name
+ WHERE view_partition_col1 = 'val1' AND view_partition_col=2 = 'val2' ...
+ }}}
+ 
+ and then capturing the table/partition inputs for this query and passing them 
on to the ALTER VIEW ADD PARTITION hook results.
+ 
+ This allows applications to track the dependencies themselves.  In the 
future, Hive will automatically populate these depenencies into the metastore 
as part of [[https://issues.apache.org/jira/browse/HIVE-1073|HIVE-1073]].
+ 

Reply via email to