[
https://issues.apache.org/jira/browse/HIVE-4966?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13725882#comment-13725882
]
Harish Butani commented on HIVE-4966:
-------------------------------------
For my understanding, are you adding a new function collect_array or are you
enhancing collect_set to have a dedup=true/false option.
The signatures of collect_map and collect_set/array are different. So we have
to expose them as separate fns.
But open to sharing a single implementation. Makes sense. What specifically do
you have in mind?
> Introduce Collect_Map UDAF
> --------------------------
>
> Key: HIVE-4966
> URL: https://issues.apache.org/jira/browse/HIVE-4966
> Project: Hive
> Issue Type: Bug
> Reporter: Harish Butani
> Assignee: Harish Butani
>
> Similar to Collect_Set. For e.g. on a Txn table
> {noformat}
> Txn(customer, product, amt)
> select customer, collect_map(product, amt)
> from txn
> group by customer
> {noformat}
> Would give you an activity map for each customer.
> Other thoughts:
> - have explode do the inverse on maps just as it does for sets today.
> - introduce a table function that outputs each value as a column. So in the
> e.g. above you get an activity matrix instead of a map.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
