pekko Deprecate `org.apache.pekko.dispatch.Futures`

I think this class was introduced when Akka still supported Java versions before Java 8, which introduced CompletionStage.

We should probably try and deprecate org.apache.pekko.dispatch.Futures and any Java APIs that still produce/expect Futures, and introduce ones that work with CompletionStage.

Jul 29 '24 15:07 raboof

I had a quick look at this and I think this Futures class is useful. See https://github.com/apache/pekko/pull/1418 for the PR that convinced that this change is not a good idea.

Jul 29 '24 19:07 pjfanning

I didn't intend to suggest Java users should use Future$.MODULE$.failed and such: what I was thinking is that Java users shouldn't need to touch the Scala Future at all, and we should make sure that where they currently must use Future, we provide equivalent APIs using the more Java-native CompletionStage.

Something along the lines of (sketch) https://github.com/apache/pekko/compare/main...raboof:pekko:no-scala-futures-in-java?expand=1

The Promise is useful and doesn't seem to have a direct counterpart in standard Java, but perhaps we should then introduce a Promise utility that works with CompletionStage / CompletableFuture as well.

Jul 29 '24 21:07 raboof

90% of the internal Pekko usages of org.apache.pekko.dispatch.Futures are in Java unit tests. I guess we could move the code to a test class and deprecate the org.apache.pekko.dispatch.Futures class. Some of the tests are linked in the Paradox documentation though.

FutureTimeoutSupport has one usage that could moved internal to that class.

Jul 29 '24 21:07 pjfanning