Hi Kurt,
+1 for making the Blink planner the default in 1.11.
One prerequisite that I see would be to resolve the remaining @Ignore
test cases in the code base.
One example that I recently spotted was here:
https://github.com/apache/flink/blob/master/flink-table/flink-table-planner-blink/src/test/scala/org/apache/flink/table/planner/plan/stream/table/TemporalTableJoinTest.scala
We should find a solution for those.
Apart from that the feature set is much bigger in 1.11 than in the
legacy planner and the APIs have settle after FLIP-65 and FLIP-95, which
make the Blink planner ready to be the default.
Regards,
Timo
On 01.04.20 03:15, Kurt Young wrote:
Hi Dev and User,
Blink planner for Table API & SQL is introduced in Flink 1.9 and already be
the default planner for
SQL client in Flink 1.10. And since we already decided not introducing any
new features to the
original Flink planner, it already lacked of so many great features that
the community has been working on, such as brand new type system, more DDL
support and more planner capabilities.
During this time, we've also received lots of great feedback from users who
were trying to use blink
planner, both positive and negative (like bugs). This is a good sign, it at
least shows more and more
users are starting to try out.
So I want to start this discussion more formally to talk about
replacing the default planner to blink.
Specifically, I want to gather feedbacks from dev and user about whether
blink planner already
cover the original planner's capabilities, what kind of issues you've ran
into when try out blink
planner and then make you fallback to original one. Since there is still a
month to go when feature
freeze, there's still enough time for community to further enhance blink
planner for this purpose.
Let me know what you think, especially if you want to report or complain
about something. Thanks
in advance.
Best,
Kurt
