You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@flink.apache.org by Piotr Nowojski <pi...@data-artisans.com> on 2018/07/02 13:07:15 UTC
[TABLE][SQL] Enirchment/Time versioned joins support in Flink
Hi,
Together with Fabian, Timo and Stephan we were working on a proposal to add a support for time versioned joins in Flink SQL/Table API. The idea here is to support use cases, when user would like to join a stream of data with a table, that changes over time and where future updates to the table do not affect past/previously emitted results.
I have prepared a document describing such feature, with proposed nomenclature and syntax, as well with explained semantic of such join: https://docs.google.com/document/d/1KaAkPZjWFeu-ffrC9FhYuxE6CIKsatHTTxyrxSBR8Sk/edit?ts=5b34c416#heading=h.bpzgdkgvel0s <https://docs.google.com/document/d/1KaAkPZjWFeu-ffrC9FhYuxE6CIKsatHTTxyrxSBR8Sk/edit?ts=5b34c416#heading=h.bpzgdkgvel0s>
The document briefly covers also a case of enrichment join with static/bounded table.
If anyone has any comments/questions, please feel free to mark them in the document or in this thread.
Thanks, Piotrek
Re: [TABLE][SQL] Enirchment/Time versioned joins support in Flink
Posted by Stephan Ewen <se...@apache.org>.
+1 from my side as well.
Very excited to have this on the roadmap...
On Mon, Jul 2, 2018 at 5:41 PM, vino yang <ya...@gmail.com> wrote:
> +1, great work!
>
> 2018-07-02 22:57 GMT+08:00 Rong Rong <wa...@gmail.com>:
>
> > Huge +1!!!
> >
> > On Mon, Jul 2, 2018 at 6:07 AM Piotr Nowojski <pi...@data-artisans.com>
> > wrote:
> >
> > > Hi,
> > >
> > > Together with Fabian, Timo and Stephan we were working on a proposal to
> > > add a support for time versioned joins in Flink SQL/Table API. The idea
> > > here is to support use cases, when user would like to join a stream of
> > data
> > > with a table, that changes over time and where future updates to the
> > table
> > > do not affect past/previously emitted results.
> > >
> > > I have prepared a document describing such feature, with proposed
> > > nomenclature and syntax, as well with explained semantic of such join:
> > > https://docs.google.com/document/d/1KaAkPZjWFeu-
> > ffrC9FhYuxE6CIKsatHTTxyrxSBR8Sk/edit?ts=5b34c416#heading=h.bpzgdkgvel0s
> > > <
> > > https://docs.google.com/document/d/1KaAkPZjWFeu-
> > ffrC9FhYuxE6CIKsatHTTxyrxSBR8Sk/edit?ts=5b34c416#heading=h.bpzgdkgvel0s
> > > >
> > >
> > > The document briefly covers also a case of enrichment join with
> > > static/bounded table.
> > >
> > > If anyone has any comments/questions, please feel free to mark them in
> > the
> > > document or in this thread.
> > >
> > > Thanks, Piotrek
> >
>
Re: [TABLE][SQL] Enirchment/Time versioned joins support in Flink
Posted by vino yang <ya...@gmail.com>.
+1, great work!
2018-07-02 22:57 GMT+08:00 Rong Rong <wa...@gmail.com>:
> Huge +1!!!
>
> On Mon, Jul 2, 2018 at 6:07 AM Piotr Nowojski <pi...@data-artisans.com>
> wrote:
>
> > Hi,
> >
> > Together with Fabian, Timo and Stephan we were working on a proposal to
> > add a support for time versioned joins in Flink SQL/Table API. The idea
> > here is to support use cases, when user would like to join a stream of
> data
> > with a table, that changes over time and where future updates to the
> table
> > do not affect past/previously emitted results.
> >
> > I have prepared a document describing such feature, with proposed
> > nomenclature and syntax, as well with explained semantic of such join:
> > https://docs.google.com/document/d/1KaAkPZjWFeu-
> ffrC9FhYuxE6CIKsatHTTxyrxSBR8Sk/edit?ts=5b34c416#heading=h.bpzgdkgvel0s
> > <
> > https://docs.google.com/document/d/1KaAkPZjWFeu-
> ffrC9FhYuxE6CIKsatHTTxyrxSBR8Sk/edit?ts=5b34c416#heading=h.bpzgdkgvel0s
> > >
> >
> > The document briefly covers also a case of enrichment join with
> > static/bounded table.
> >
> > If anyone has any comments/questions, please feel free to mark them in
> the
> > document or in this thread.
> >
> > Thanks, Piotrek
>
Re: [TABLE][SQL] Enirchment/Time versioned joins support in Flink
Posted by Rong Rong <wa...@gmail.com>.
Huge +1!!!
On Mon, Jul 2, 2018 at 6:07 AM Piotr Nowojski <pi...@data-artisans.com>
wrote:
> Hi,
>
> Together with Fabian, Timo and Stephan we were working on a proposal to
> add a support for time versioned joins in Flink SQL/Table API. The idea
> here is to support use cases, when user would like to join a stream of data
> with a table, that changes over time and where future updates to the table
> do not affect past/previously emitted results.
>
> I have prepared a document describing such feature, with proposed
> nomenclature and syntax, as well with explained semantic of such join:
> https://docs.google.com/document/d/1KaAkPZjWFeu-ffrC9FhYuxE6CIKsatHTTxyrxSBR8Sk/edit?ts=5b34c416#heading=h.bpzgdkgvel0s
> <
> https://docs.google.com/document/d/1KaAkPZjWFeu-ffrC9FhYuxE6CIKsatHTTxyrxSBR8Sk/edit?ts=5b34c416#heading=h.bpzgdkgvel0s
> >
>
> The document briefly covers also a case of enrichment join with
> static/bounded table.
>
> If anyone has any comments/questions, please feel free to mark them in the
> document or in this thread.
>
> Thanks, Piotrek