Github user KurtYoung commented on the issue:
https://github.com/apache/flink/pull/3520
Hi @fhueske @godfreyhe, thanks for the review, i addressed most all your
comments.
@fhueske Except for letting `TableSourceScan` be aware of whether filter
has been pushed down. I'm not sure to let the `TableSourceScan` has this kind
of information, i'd prefer to let them stay within the all kinds of
`TableSource`. One drawback to let `TableSourceScan` has such kind of
information is when we do the `TableSourceScan` copy, we need to take care all
these information, make sure they also be copied correctly. In the future, if
we add more extension to `TableSource` like we can push part of query down, we
will face this problem.
Regarding to the interface of `FilterableTableSource`, i agree with you
that the trait containing some logic is not friendly with java extensions. So i
removed the default implementation of `isFilterPushedDown`, the inherited class
should take care of this method. And regarding the `Tuple2` thing, how about we
pass in a mutable java list, and let table source to *pick out* expression from
it and return a copy of table source which contains these pushed down
predicates.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
