Re: Flink SQL support array transform function

Hi Caizhi, I've tried to write UDF to support this function, but I found I can't pass the function parameter to udf because the data type of function is not supported. An exception throws in SQL validation. My UDF code: class ArrayTransformFunction extends ScalarFunction { def eval(a: Arra

Disable the chain of the Sink operator

Hello The current Sink operator will be split into two operations, Writer and Commiter. By default, they will be chained together and executed on the same thread. So sometimes when the commiter is very slow, it will block the data writer, causing back pressure. At present, FlinkSQL can be so

Re: Reusing the same OutputTag in multiple ProcessFunctions

Wow thank you so much! Good to know its not just me. At the end of my day yesterday, I started sniffing this out too. I think I effectively did the same thing as setting _j_typeinfo to None by manually recreating the _j_typeinfo in a new ('cloned') output tag: from pyflink.common.typeinfo impor

Re: Reusing the same OutputTag in multiple ProcessFunctions

Hi Andrew, I've found out that this's a bug brought by another bugfix FLINK-29681 , I've created an issue FLINK-31083 for this problem. You could temporarily set inner java type_info to None befo

Re: Watermark in global commit

Dear Jan, thank you for your answer! The logic that ensures consistency should already be implemented in the TwoPhaseCommitSink and the WithPostCommitTopology. So I would rather like to use these well-tested classes than implement my own logic for this. Best regards Tobias Hi, I'm not exper