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

    https://github.com/apache/spark/pull/15085#discussion_r78674369
  
    --- Diff: core/src/main/scala/org/apache/spark/storage/BlockManager.scala 
---
    @@ -857,9 +862,11 @@ private[spark] class BlockManager(
     
         val startTimeMs = System.currentTimeMillis
         var blockWasSuccessfullyStored: Boolean = false
    +    var exceptionWasThrown: Boolean = true
         val result: Option[T] = try {
           val res = putBody(putBlockInfo)
           blockWasSuccessfullyStored = res.isEmpty
    +      exceptionWasThrown = false
           res
         } finally {
           if (blockWasSuccessfullyStored) {
    --- End diff --
    
    One concern with using a `catch` here is handling of 
`InterruptedException`: if we use `case NonFatal(e)` that won't match 
`InterruptedException` and we'll miss out on cleanup following that. If we 
catch `Throwable`, on the other hand, then I think that we'll end up clearing 
the `isInterrupted` bit for `InterruptedException`s and it'll be awkward to 
match and re-set it when rethrowing. Therefore I'd like to keep the 
exception-handling case in the `finally` block with a simple check to see if we 
entered that block via an error case.
    
    Note that I've seen this same exception-handling idiom used in Java code, 
where code that catches and re-throws `Throwable` won't compile in older Java 
versions because of the checked exception-handling (I think that newer versions 
are a bit more permissive about throwing exceptions from a `catch` block). 


---
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 infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---

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

Reply via email to