hive-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Edward Capriolo <edlinuxg...@gmail.com>
Subject Re: UDF which takes entire row as arg
Date Thu, 08 Jul 2010 03:31:53 GMT
On Wed, Jul 7, 2010 at 8:44 PM, Pradeep Kamath <pradeepk@yahoo-inc.com> wrote:
> Hi Arvind,
>
>   Before I dig deeper with the patch on that jira, I was wondering if it is
> relevant that what I am trying is not an aggregate function (no group by
> involved). I am writing a row level UDF but would like all fields in the row
> to be presented to the udf. So my query literally is:
>
> “select myudf(*) from mytable;”
>
>
>
> Do you think the patch might address my use case even though my UDF is not a
> UDAF?
>
>
>
> Thanks,
>
> Pradeep
>
>
>
> ________________________________
>
> From: Arvind Prabhakar [mailto:arvind@cloudera.com]
> Sent: Wednesday, July 07, 2010 5:26 PM
> To: hive-user@hadoop.apache.org
> Subject: Re: UDF which takes entire row as arg
>
>
>
> Hi Pradeep,
>
>
>
> See HIVE-287. The most recent patch supplied with this Jira allows you to
> create UDAFs that support the  FUNCTION(STAR) syntax. You can apply this
> patch to your local working copy and see if that addresses your problem.
>
>
>
> Also feel free to ask questions regarding anything that this patch provides.
>
>
>
> Arvind
>
>
>
> On Wed, Jul 7, 2010 at 4:35 PM, Pradeep Kamath <pradeepk@yahoo-inc.com>
> wrote:
>
> Hi,
>
>   I wanted to implement a UDF which takes the entire row of input as
> argument. Unfortunately when I call it using select myudf(*) from my table,
> I get :
>
> FAILED: Parse Error: line 1:16 mismatched input '*' expecting ) in function
> specification
>
>
>
> Is this not allowed?
>
>
>
> Thanks,
>
> Pradeep
>
>

You could write a generic UDF since they accept arbitrary signature,
but you would have to pass each column specifically (no * support)

Mime
View raw message