Github user srowen commented on a diff in the pull request:
https://github.com/apache/spark/pull/22032#discussion_r208425588
--- Diff: mllib/src/main/scala/org/apache/spark/ml/feature/LSH.scala ---
@@ -75,7 +75,7 @@ private[ml] abstract class LSHModel[T <: LSHModel[T]]
* The hash function of LSH, mapping an input feature vector to multiple
hash vectors.
* @return The mapping of LSH function.
*/
- protected[ml] val hashFunction: Vector => Array[Vector]
+ protected[ml] def hashFunction(elems: Vector): Array[Vector]
--- End diff --
This change does appear to resolve the issue by avoiding whatever is
happening in these two cases. This is at least reasonable, as it sort of
appears it's something to do with scala + Java 8 rather than Spark.
However I do wonder whether MiMa will allow this change. It is still
exposed as a function in the bytecode, so maybe. If not, will be a tough call
whether this experimental class is allowed to change.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
