pnowojski commented on a change in pull request #6299: [FLINK-9713][table][sql]
Support processing time versioned joins
URL: https://github.com/apache/flink/pull/6299#discussion_r208835295
##########
File path:
flink-libraries/flink-table/src/main/scala/org/apache/flink/table/plan/nodes/datastream/DataStreamJoin.scala
##########
@@ -103,27 +97,42 @@ class DataStreamJoin(
tableEnv: StreamTableEnvironment,
queryConfig: StreamQueryConfig): DataStream[CRow] = {
- val config = tableEnv.getConfig
- val returnType = schema.typeInfo
- val keyPairs = joinInfo.pairs().toList
+ validateKeyTypes()
- // get the equality keys
- val leftKeys = ArrayBuffer.empty[Int]
- val rightKeys = ArrayBuffer.empty[Int]
+ val leftDataStream =
Review comment:
I would be against that. I'm depending on this change in this PR, so it
defines strict order of those two commits. Either I wouldn't publish the actual
versioned joins PR after this commit is merged, or the versioned joins PR would
look exactly the same as it looks now. Splitting into two PRs adds a lot of
overhead (in case of rebases or applying review changes you need to keep
updating multiple PRs) and room for a lot of problems with PRs going out of
sync. During reviewing it's also annoying, because you see the same code twice
and reviewers (same or different ones) are often (almost always?) making half
of the comments in one PR and half in the other.
Besides, this is already in a separate commit so it can be reviewed
separately. Maybe instead of reviewing PR all at once try reviewing it commit
by commit?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
