Github user huawei-flink commented on the issue:
https://github.com/apache/flink/pull/3771
@fhueske I have created #3783 with just the code generation part. At least
the GROUP BY distinct can move ahead. I will close this PR and wait for the
merging of the Calcite fix.
---
If your p
Github user huawei-flink commented on the issue:
https://github.com/apache/flink/pull/3771
So, what do you want me to keep for this PR? just the code generation and
its test?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as wel
Github user fhueske commented on the issue:
https://github.com/apache/flink/pull/3771
If we merge this change together with #3764, we would check it with tests
that use distinct grouped window aggregates.
But you are right, it might in fact make sense to test the
`GeneratedAggreg
Github user huawei-flink commented on the issue:
https://github.com/apache/flink/pull/3771
@fhueske @haohui I have no problem removing the DIST() part, it is just not
possible to test it without. Shall I push just the code generation and
aggregates util changes?
---
If your project
Github user stefanobortoli commented on the issue:
https://github.com/apache/flink/pull/3771
@fhueske @rtudoran @shijinkui @sunjincheng121 I have create a new PR for
distinct in the code generator. Please have a look and let me know. I have
implemented and tested only for OverProcTime