Handlers added to MappedEventStream get GCed

[stanch]

I’m on Android, but I don’t think it matters. What happens is that given a continuous stream of events stream,

• stream.foreach(_ ⇒ Log.d("S1", "working")) gets printed all the time, even after GC_EXPLICIT, but
• stream.map(_ + 1).foreach(_ ⇒ Log.d("S2", "working")) stops after first GC_FOR_ALLOC (see here for the meaning).

Strange thing is that Observer still keeps all four references (two Function1s, an EventSource and a MappedEventStream): I made a custom observer to debug this.

I am not sure what else could be garbage collected there. Any thoughts?

[stanch]

Could it be that eta-expanded method gets collected? https://github.com/dylemma/scala.frp/blob/master/src/main/scala/io/dylemma/frp/EventPipe.scala#L43

[dylemma]

Hmm.. I can't think of anything off the top of my head. I'll spend a bit of time and see if I can recreate the issue (I won't be on Android...).

[dylemma]

Oh you already did fix it with the pull request didn't you.

[stanch]

I did :)

[stanch]

Thanks! Do you mind to publish it?

[dylemma]

Working on that now. I have to go re-learn how to do it. Looks like it auto-closed when I did the merge. I'll re-open until I publish.

[stanch]

Great, thanks! Looking at Build.scala, I think you just need to run sbt publish and have your sonatype credentials somewhere.

[dylemma]

That's the easy part. IIRC there's some extra voodoo I have to do through sonatype to get it into Maven Central.. that's the part I don't remember how to do, but no worries. The internet helps ;)

[stanch]

Oh, I see... I’m going the easy way with bintray and https://github.com/softprops/bintray-sbt :)

[dylemma]

The release is up - I can see it at https://oss.sonatype.org/content/repositories/releases/io/dylemma/scala-frp_2.10/1.1/ and it should be sync'd to central within a few hours.

[stanch]

Thanks again! I’ll update my dependencies.