[GitHub] spark pull request #14299: [SPARK-16440][MLlib] Ensure broadcasted variables...

[AnthonyTruchet <gi...@git.apache.org>]

Github user AnthonyTruchet commented on a diff in the pull request:

    https://github.com/apache/spark/pull/14299#discussion_r103656554
  
    --- Diff: mllib/src/main/scala/org/apache/spark/mllib/feature/Word2Vec.scala ---
    @@ -314,6 +315,21 @@ class Word2Vec extends Serializable with Logging {
         val expTable = sc.broadcast(createExpTable())
         val bcVocab = sc.broadcast(vocab)
         val bcVocabHash = sc.broadcast(vocabHash)
    +    try {
    +      doFit(dataset, sc, expTable, bcVocab, bcVocabHash)
    +    }
    +    finally
    --- End diff --
    
    I already spotted and corrected a few instances of forgotten broadcasted variables. I've already reported and patched all I was aware off.
    
    I was suggesting introducing an ARM pattern and making broadcasted variable ARM aware (see comment above). This is by no way a general solution but it might help reduce the issue when the use of a broadcasted variable is compatibble with a syntactic scope. @thunterdb legitimately raised concern about introducing new dependencies (to scala-arm specifically). What is your insight wrt to this ?


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org