Re: [DISCUSS] (FLINK-2980) Add support for CUBE / ROLLUP / GROUPING SETS in Table API

2016-12-08 Thread Alexander Chermenin
Hi Fabian,

thanks for your comments.
I think that's good approach.
I will create new issue today and change my PR after that.

Regards,
Alexander

On Thu, Dec 8, 2016 at 11:40 PM, Fabian Hueske  wrote:

> Hi Alexander,
>
> thanks for the PR!
> It might take sometime until it is reviewed. We have received quite a few
> PRs for the Table API lately and there is kind of a review queue at the
> moment.
>
> The scope of the current issue (FLINK-2980) is to support grouping sets,
> cube, and rollup in the Table API while your PR adds support for SQL.
> I think it would be good if you could open a JIRA issue to support grouping
> sets, cube, and rollup in SQL and "retarget" your PR to the new issue
> (basically change the title + commit messages)
>
> For now, support in SQL is good enough, IMO. Adding feature to the Table
> API is blowing up the API quite a bit. I think we should first see if there
> is interest in this feature.
> The good message is that once we support it in SQL, we "only" need to add
> the API and validation part for the Table API.
>
> What do you think?
> Best, Fabian
>
>
> 2016-12-08 13:11 GMT+01:00 Alexander Chermenin :
>
> > Hi folks!
> >
> > I would like to discuss next question:
> > In PR #2965 (https://github.com/apache/flink/pull/2965) I have added
> > support for such operators as CUBE / ROLLUP / GROUPING SETS in SQL
> queries.
> > Should we do support for the same functionality in Table API? And if so,
> > how will it look?
> > Something like this `table.groupingBy("cube(...)").select("...")`, or
> > `table.cube(...).select(...)` or something else?
> >
> > Regards,
> > Alexander
> >
>


Re: [DISCUSS] (FLINK-2980) Add support for CUBE / ROLLUP / GROUPING SETS in Table API

2016-12-08 Thread Fabian Hueske
Hi Alexander,

thanks for the PR!
It might take sometime until it is reviewed. We have received quite a few
PRs for the Table API lately and there is kind of a review queue at the
moment.

The scope of the current issue (FLINK-2980) is to support grouping sets,
cube, and rollup in the Table API while your PR adds support for SQL.
I think it would be good if you could open a JIRA issue to support grouping
sets, cube, and rollup in SQL and "retarget" your PR to the new issue
(basically change the title + commit messages)

For now, support in SQL is good enough, IMO. Adding feature to the Table
API is blowing up the API quite a bit. I think we should first see if there
is interest in this feature.
The good message is that once we support it in SQL, we "only" need to add
the API and validation part for the Table API.

What do you think?
Best, Fabian


2016-12-08 13:11 GMT+01:00 Alexander Chermenin :

> Hi folks!
>
> I would like to discuss next question:
> In PR #2965 (https://github.com/apache/flink/pull/2965) I have added
> support for such operators as CUBE / ROLLUP / GROUPING SETS in SQL queries.
> Should we do support for the same functionality in Table API? And if so,
> how will it look?
> Something like this `table.groupingBy("cube(...)").select("...")`, or
> `table.cube(...).select(...)` or something else?
>
> Regards,
> Alexander
>


[DISCUSS] (FLINK-2980) Add support for CUBE / ROLLUP / GROUPING SETS in Table API

2016-12-08 Thread Alexander Chermenin
Hi folks!

I would like to discuss next question:
In PR #2965 (https://github.com/apache/flink/pull/2965) I have added
support for such operators as CUBE / ROLLUP / GROUPING SETS in SQL queries.
Should we do support for the same functionality in Table API? And if so,
how will it look?
Something like this `table.groupingBy("cube(...)").select("...")`, or
`table.cube(...).select(...)` or something else?

Regards,
Alexander