You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@kylin.apache.org by dong wang <el...@gmail.com> on 2016/01/12 09:13:03 UTC

can we support adding mapping cube columns to hive table columns

sometimes, we have to change column names of the source hive table AFTER
the cube is built successfully, actually, in most cases, we may just change
the column name without changing the column type, but now, we have to
refresh the cube again when we change the column name of the source hive
table and sync the hive table to kylin meta. in common sense, we want that
the cube data may be avoid being calculated again since just changing the
column name.

Re: can we support adding mapping cube columns to hive table columns

Posted by Luke Han <lu...@gmail.com>.
Name, description, label both are dimension's attribution which should
allow to be changeable for such case.

But this feature should not considering slow change dimension case but also
should leave any interface for that.





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

Luke Han

On Tue, Jan 19, 2016 at 10:51 AM, Li Yang <li...@apache.org> wrote:

> Valid request. Creating new JIRA KYLIN-1332
> <https://issues.apache.org/jira/browse/KYLIN-1332>.
>
> On Tue, Jan 12, 2016 at 5:17 PM, yu feng <ol...@gmail.com> wrote:
>
> > My suggestion is adding this mapping by creating view, So, you can change
> > the column name in hive table, and recreate the view, that will do not
> have
> > any effect to cube. and need not sync hive table. this view only used at
> > the first step, so can recreate it in any time except doing this step.
> >
> > 2016-01-12 16:13 GMT+08:00 dong wang <el...@gmail.com>:
> >
> > > sometimes, we have to change column names of the source hive table
> AFTER
> > > the cube is built successfully, actually, in most cases, we may just
> > change
> > > the column name without changing the column type, but now, we have to
> > > refresh the cube again when we change the column name of the source
> hive
> > > table and sync the hive table to kylin meta. in common sense, we want
> > that
> > > the cube data may be avoid being calculated again since just changing
> the
> > > column name.
> > >
> >
>

Re: can we support adding mapping cube columns to hive table columns

Posted by Li Yang <li...@apache.org>.
Valid request. Creating new JIRA KYLIN-1332
<https://issues.apache.org/jira/browse/KYLIN-1332>.

On Tue, Jan 12, 2016 at 5:17 PM, yu feng <ol...@gmail.com> wrote:

> My suggestion is adding this mapping by creating view, So, you can change
> the column name in hive table, and recreate the view, that will do not have
> any effect to cube. and need not sync hive table. this view only used at
> the first step, so can recreate it in any time except doing this step.
>
> 2016-01-12 16:13 GMT+08:00 dong wang <el...@gmail.com>:
>
> > sometimes, we have to change column names of the source hive table AFTER
> > the cube is built successfully, actually, in most cases, we may just
> change
> > the column name without changing the column type, but now, we have to
> > refresh the cube again when we change the column name of the source hive
> > table and sync the hive table to kylin meta. in common sense, we want
> that
> > the cube data may be avoid being calculated again since just changing the
> > column name.
> >
>

Re: can we support adding mapping cube columns to hive table columns

Posted by yu feng <ol...@gmail.com>.
My suggestion is adding this mapping by creating view, So, you can change
the column name in hive table, and recreate the view, that will do not have
any effect to cube. and need not sync hive table. this view only used at
the first step, so can recreate it in any time except doing this step.

2016-01-12 16:13 GMT+08:00 dong wang <el...@gmail.com>:

> sometimes, we have to change column names of the source hive table AFTER
> the cube is built successfully, actually, in most cases, we may just change
> the column name without changing the column type, but now, we have to
> refresh the cube again when we change the column name of the source hive
> table and sync the hive table to kylin meta. in common sense, we want that
> the cube data may be avoid being calculated again since just changing the
> column name.
>