Do we need to support FLIP-27: Refactor Source Interface

[spike-liu]

Split discussion from #38

[spike-liu]

https://cwiki.apache.org/confluence/display/FLINK/FLIP-27%3A+Refactor+Source+Interface

[spike-liu]

Here is my research:

1. The problems mentioned below currently are not problem for nebula connector.

2. Currently only message queue framework, such as Kafka/Pulsar, implemented this new API.

So far, from my point of view, there is no motivation for us to implement these new source API.

@Nicole00 what is your idea?

[Nicole00]

Sorry I missed this issue, I'll research it and reply later.

[Nicole00]

I agree with @spike-liu .
The main motivation of flip-27 is to facilitate the expansion of data sources and lower the threshold for everyone to contribute source code. Sense we have extended the Nebula data source, it seems there's no need to refactor the interface.

And the use requirement of checkpoint is not high, so I tend to see the support of FILP-27 as a future requirement.

[spike-liu]

Thanks for your reply. @Nicole00

Closed.