ahshahid commented on code in PR #45446:
URL: https://github.com/apache/spark/pull/45446#discussion_r1527235068
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/ColumnResolutionHelper.scala:
##########
@@ -477,6 +482,57 @@ trait ColumnResolutionHelper extends Logging with
DataTypeErrorsBase {
assert(q.children.length == 1)
q.children.head.output
},
+
+ resolveOnDatasetId = (datasetid: Long, name: String) => {
Review Comment:
@peter-toth I agree with your analysis. In the current PR, the approach I
had in mind, was to allow columns from only top Joining dataframes for
simplicity purposes . The reason for this thinking was :
1) It allows predictable behaviour and easier for user to comprehend the
outcome.
2) while resolving does not need to do deep traversals. In case of repeat
dataframes , if we reach the leaves, then dataset IDs most likely will clash,
so to resolve ambiguity we would have to resort to something like shortest
depth etc ( I still use depth but only 1 level deep).
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
