Hi Manish,

Please add impact area for block pruning and data query.

-Regards
Kumar Vishal

On Fri, Mar 10, 2017 at 12:56 PM, manish gupta <tomanishgupt...@gmail.com>
wrote:

> Hi Liang,
>
> Thanks for reviewing the document. Please find my answers inline.
>
> 1.For "change of data type" , whether only support from INT to BIGINT, or
> not ?
> - Yes as part of initial feature development we will support only
>  upcasting the datatype  INT to BIGINT and decimal datatype change from
> lower to higher precision.
>
> 2.Whether support adjust the order of columns for MDK , and make compaction
> to resort data as per the new order of columns , or not ?
> - Currently addition of column is supported only at the end. From my
> opinion once this is stable, we can further extend it to support reordering
> of columns to change the MDKEY order.
>
> Regards
> Manish Gupta
>
> On Fri, Mar 10, 2017 at 5:20 AM, Liang Chen <chenliang6...@gmail.com>
> wrote:
>
> > Hi
> >
> > Thanks for you started this discussion for alter table feature.
> >
> > A couple of comments:
> > 1.For "change of data type" , whether only support from INT to BIGINT, or
> > not ?
> > 2.Whether support adjust the order of columns for MDK , and make
> compaction
> > to resort data as per the new order of columns , or not ?
> >
> > Regards
> > Liang
> >
> > 2017-03-09 21:00 GMT+08:00 manish gupta <tomanishgupt...@gmail.com>:
> >
> > > Hi All,
> > >
> > > Currently in carbondata after a table is created once it becomes
> > immutable.
> > > Deletion or addition of column is not supported because of which the
> same
> > > table and data cannot be used again. To add more flexibility to the
> > > carbondata system alter table support needs to be added to carbondata
> > > system.
> > >
> > >
> > > Please refer the design document at below location and share your
> > > suggestions.
> > >
> > >
> > > https://drive.google.com/file/d/0B1DnrpMgGOu9dGxYMzVJUXEyZG8/
> > > view?usp=sharing
> > >
> > >
> > > Regards
> > >
> > > Manish Gupta
> > >
> >
> >
> >
> > --
> > Regards
> > Liang
> >
>

Reply via email to