srowen commented on a change in pull request #21632: [SPARK-19591][ML][MLlib]
Add sample weights to decision trees
URL: https://github.com/apache/spark/pull/21632#discussion_r247560135
##########
File path: project/MimaExcludes.scala
##########
@@ -241,7 +241,10 @@ object MimaExcludes {
// [SPARK-26216][SQL] Do not use case class as public API
(UserDefinedFunction)
ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.sql.expressions.UserDefinedFunction$"),
-
ProblemFilters.exclude[IncompatibleTemplateDefProblem]("org.apache.spark.sql.expressions.UserDefinedFunction")
+
ProblemFilters.exclude[IncompatibleTemplateDefProblem]("org.apache.spark.sql.expressions.UserDefinedFunction"),
+
+ // [SPARK-19591][ML] Add sample weights to decision trees
+
ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.mllib.tree.configuration.Strategy.this")
Review comment:
Hm, yeah, the problem is that both constructors define a bunch of optional
arguments, and this call doesn't specify those optional argument. So both
constructors apply. The 'fix' would be to make the backwards-compatible
constructor not take any optional arguments. It was already source-compatible
without this additional constructor, but would make it binary-compatible too.
I'm on the fence about it. It's not something apps would normally specify
directly, but it's also pretty trivial to write in this additional constructor.
If it works, I'd keep the backwards-compatible constructor. Little benefit,
very little cost.
----------------------------------------------------------------
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
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
