Github user srowen commented on a diff in the pull request:
https://github.com/apache/spark/pull/11628#discussion_r55936088
--- Diff: core/src/main/scala/org/apache/spark/rdd/PipedRDD.scala ---
@@ -157,8 +167,16 @@ private[spark] class PipedRDD[T: ClassTag](
val lines = Source.fromInputStream(proc.getInputStream).getLines()
new Iterator[String] {
def next(): String = lines.next()
+
+ private def propagateChildThreadException(): Unit = {
--- End diff --
Yeah I should revise that -- it's fine for `hasNext` to report an exception
since it may have to do the read to figure out the answer. But `next` also
needs to be able to do this, even though that could be a pretty minor issue in
practice since `hasNext` is usually called before `next`. But if it isn't, I
guess this means the cleanup isn't called, which isn't great.
Yeah I see `getLines` doesn't block now, and see what it's trying to do, so
nevermind that part. I suggest this could be a little more straightforward if
maybe all of the cleanup logic (including exception handling were pulled out
into a private method called from `hasNext`, and then also check the return
value of `hasNext` from `next` (and throw `NoSuchElementException` if needed).
That's nice form and also solves the problem of `next` needing to do the
cleanup.
All that is not directly related to change you're making but I wonder if
it's worth cleaning up while you're modifying this anyway.
---
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 [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
