Thanks Guys, it’s just what I’ve looked for.
From: Dean Wampler [mailto:dean.wamp...@thinkbiganalytics.com]
Sent: יום ו 23 נובמבר 2012 15:41
To: user@hive.apache.org
Cc: Dima Datsenko
Subject: Re: Effecient partitions usage in join
A quick suggestion. If the format never changes, you might as
cord_date_iso >=
> DATE_SUB(currentisodate(), 30);
>
> ** **
>
> I’ve always had a preference for iso dates since they sort nicely:
> 2012-11-23 but you can obviously pick your own pattern.
>
> ** **
>
> ** **
>
> *From:* Dima Datsenko [mailto:di...@mic
cely: 2012-11-23
but you can obviously pick your own pattern.
From: Dima Datsenko [mailto:di...@microsoft.com]
Sent: Thursday, November 22, 2012 4:07 PM
To: Bennie Schut; user@hive.apache.org
Subject: RE: Effecient partitions usage in join
Hi Benny,
The udf solution sounds like a plan. Much better tha
@hive.apache.org
Cc: Dima Datsenko
Subject: RE: Effecient partitions usage in join
Unfortunately at the moment partition pruning is a bit limited in hive. When
hive creates the query plan it decides what partitions to use. So if you put
hardcoded list of partition_id items in the where clause it will
Unfortunately at the moment partition pruning is a bit limited in hive. When
hive creates the query plan it decides what partitions to use. So if you put
hardcoded list of partition_id items in the where clause it will know what to
do. In the case of a join (or a subquery) it would have to run t
