You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@kylin.apache.org by Yiming Liu <li...@gmail.com> on 2016/09/10 16:10:38 UTC

Could Integer encoding ignore the empty value?

Hi Kylin devs,

Currently, if some dimension is marked as Integer encoding, but there is
empty value(NULL and "" ) in that column, Kylin will report
NumberFormatException.

IntegerDimEnc Line 92, Long.parseLong()

Could we add some check before the parsing?

-- 
With Warm regards

Yiming Liu (刘一鸣)

Re: Could Integer encoding ignore the empty value?

Posted by Li Yang <li...@apache.org>.
I believe Hongbin has fixed the Integer encoding to support NULL recently.

Thanks YiMing for the reminder.

Yang

On Fri, Sep 16, 2016 at 11:59 PM, Luke Han <lu...@gmail.com> wrote:

> "" is not valid value for Integer, but NULL should be.
>
> Add type check for each dimension and each row will expand cube build time
> very much.
>
> I would like to suggest users to have some tools to profile data before
> build cube, normally, data profiling should belong to ETL part.
>
> Thanks.
>
>
> Best Regards!
> ---------------------
>
> Luke Han
>
> On Sun, Sep 11, 2016 at 12:10 AM, Yiming Liu <li...@gmail.com>
> wrote:
>
> > Hi Kylin devs,
> >
> > Currently, if some dimension is marked as Integer encoding, but there is
> > empty value(NULL and "" ) in that column, Kylin will report
> > NumberFormatException.
> >
> > IntegerDimEnc Line 92, Long.parseLong()
> >
> > Could we add some check before the parsing?
> >
> > --
> > With Warm regards
> >
> > Yiming Liu (刘一鸣)
> >
>

Re: Could Integer encoding ignore the empty value?

Posted by Luke Han <lu...@gmail.com>.
"" is not valid value for Integer, but NULL should be.

Add type check for each dimension and each row will expand cube build time
very much.

I would like to suggest users to have some tools to profile data before
build cube, normally, data profiling should belong to ETL part.

Thanks.


Best Regards!
---------------------

Luke Han

On Sun, Sep 11, 2016 at 12:10 AM, Yiming Liu <li...@gmail.com>
wrote:

> Hi Kylin devs,
>
> Currently, if some dimension is marked as Integer encoding, but there is
> empty value(NULL and "" ) in that column, Kylin will report
> NumberFormatException.
>
> IntegerDimEnc Line 92, Long.parseLong()
>
> Could we add some check before the parsing?
>
> --
> With Warm regards
>
> Yiming Liu (刘一鸣)
>