Yu's suggestion is the right way, Kylin's join only support regular columns.

Hive view is very powerful for such case, even like to handle UDF.

Thanks.


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

Luke Han

On Fri, Nov 20, 2015 at 5:38 PM, Roy <[email protected]> wrote:

> Thanks for you suggestion.
>
> i will create view first ,next to process data.
>
> Best  Regards
>
> Roy
>
>
>
>
>
> At 2015-11-20 16:54:15, "yu feng" <[email protected]> wrote:
> >I don't think kylin can use expression in join, in our experience, we
> >always create view based on hive table when we meet the problem of source
> >metadata or data and it always works. for example, you can create a view
> >with remove issues datas and use the view as lookup table.
> >
> >hopefully it will be useful to you..
> >
> >2015-11-20 16:18 GMT+08:00 Roy <[email protected]>:
> >
> >> Hi All,
> >>
> >> when i was use kylin add lookup tables,i can add condition to
> >> tb1.field1=tb2.field2,but if i want field1=expression(like  and
> field2='0')
> >> ,need necessary write expression to cube filter?
> >>
> >> in look table  exitst some issues data, i need remove first.but i don't
> >> want to create the new tmp table, we konw in SQL left join sentence are
> >> before the where sentnece. any idea for this issues?
> >>
> >> Best Regards
> >>
> >>
> >>
> >>
> >>
> >>
>

Reply via email to