jrudolph commented on code in PR #1026:
URL: https://github.com/apache/incubator-pekko/pull/1026#discussion_r1467568190
##########
stream/src/main/scala/org/apache/pekko/stream/impl/fusing/CollectWhile.scala:
##########
@@ -43,18 +44,17 @@ private[pekko] final class CollectWhile[In, Out](pf:
PartialFunction[In, Out]) e
private lazy val decider =
inheritedAttributes.mandatoryAttribute[SupervisionStrategy].decider
import Collect.NotApplied
+ @nowarn("msg=Any")
override final def onPush(): Unit =
try {
// 1. `applyOrElse` is faster than (`pf.isDefinedAt` and then
`pf.apply`)
// 2. using reference comparing here instead of pattern matching can
generate less and quicker bytecode,
// eg: just a simple `IF_ACMPNE`, and you can find the same trick
in `Collect` operator.
// If you interest, you can check the associated PR for this
change and the
// current implementation of
`scala.collection.IterableOnceOps.collectFirst`.
- val result = pf.applyOrElse(grab(in), NotApplied)
- if (result.asInstanceOf[AnyRef] eq NotApplied) {
- completeStage()
- } else {
- push(out, result.asInstanceOf[Out])
+ pf.applyOrElse(grab(in), NotApplied) match {
+ case _: NotApplied.type => completeStage()
Review Comment:
Why not match on the singleton instance instead? In any case, it seems
uncommon to match on the singleton type.
It's not even easy to say if matching on the type or the reference is
supposed to be faster in general, and maybe the Scala compiler should make the
right choice. (On one hand, reference comparison is as fast as it gets, on the
other hand, getting the singleton instance reference is at least a static field
lookup which might not be faster than comparing types, would depend on the JIT
compiler).
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
