add [openjdk11] to Travis cross build

[sullis]

No description provided.

[akka-ci]

Thank you for your pull request! After a quick sanity check one of the team will reply with 'OK TO TEST' to kick off our automated validation on Jenkins. This compiles the project, runs the tests, and checks for things like binary compatibility and source code formatting. When two team members have also manually reviewed and (perhaps after asking for some amendments) accepted your contribution, it should be good to be merged.

For more details about our contributing process, check out CONTRIBUTING.md - and feel free to ask!

[raboof]

OK TO TEST

[raboof]

I guess we can't merge this until #25449 is fixed

[akka-ci]

Test PASSed.

[johanandren]

Note that while useful, our travis build doesn't do much, where the actual ci is done is our own jenkins servers. We should probably update the JDK9/10 job to run on 11 instead given that is the LTS.

[chbatey]

We have an 11 build it just doesn't compile :) https://jenkins.akka.io:8498/job/akka-nightly-jdk11/

[He-Pin]

This is a must:)

[johanandren]

For reference it is blocked on a static forwarder methods conflict: scala/bug#11061

[johanandren]

Oh, wait, that has been fixed now, I'll try out 2.12.7 with Java 11 right away.

[johanandren]

Sorry for the noise, spoke too soon, 2.12.7 isn't out yet, so this should be fixed when it is.

We have a ticket tracking the compilation failure: #25449

Edit: which was already noted by Raboof earlier, I'll just shut up now.

[chbatey]

@sullis now that 2.12.7 is released want to add it to this PR and see if it works in travis for JDK11?

I've kicked off with 2.12.7 now that genjavadoc has been released

[raboof]

@sullis it would be great if you could rebase this one!

[akka-ci]

Test PASSed.

[johanandren]

Hmm, now it got this:

[error] Class jdk.internal.HotSpotIntrinsicCandidate not found - continuing with a stub.
[error] Class jdk.internal.HotSpotIntrinsicCandidate not found - continuing with a stub.
[error] Class jdk.internal.HotSpotIntrinsicCandidate not found - continuing with a stub.
[error] Class jdk.internal.HotSpotIntrinsicCandidate not found - continuing with a stub.
[error] Class jdk.internal.HotSpotIntrinsicCandidate not found - continuing with a stub.
[error] 5 errors found

Wasn't that what we saw in the community build? (@SethTisue ?)

[raboof]

Hm - might be interesting that it is failing for [info] Compiling 2 Scala sources and 1 Java source to /home/travis/build/akka/akka/akka-stream/target/CompileJdk9-classes - those classes should be compiled with -release 9

[SethTisue]

Wasn't that what we saw in the community build? (@SethTisue ?)

Um, probably, I have trouble keeping all 180 projects straight in my head :-)

but a recent run is https://scala-ci.typesafe.com/view/scala-2.12.x/job/scala-2.12.x-jdk11-integrate-community-build/57/ and it has:

[akka] [error] Failed tests:
[akka] [error] 	akka.io.TcpConnectionSpec
[akka] [error] 	akka.io.UdpConnectedIntegrationSpec

in our config file we have:

      // idk why, but prevents dbuild-only sun.misc.Unsafe compile errors;
      // see https://github.com/scala/community-builds/issues/757
      "set scalacOptions in Compile in actor --= Seq(\"-release\", \"8\")"

but anyway actually we're currently frozen at your v2.15.16 tag, I figured I'd wait for this PR to get resolved, and then try master.

[johanandren]

Late answer here for @SethTisue but, those test failures we know about but have yet to look into (#25733)

[patriknw]

What is the status here? Still waiting for some known issue?

[johanandren]

I think this should be working by now, just retriggered the build

[dwijnand]

Nope:

[info] Compiling 2 Scala sources and 1 Java source to /home/travis/build/akka/akka/akka-stream/target/CompileJdk9-classes ...
[error] Class jdk.internal.HotSpotIntrinsicCandidate not found - continuing with a stub.
[error] Class jdk.internal.HotSpotIntrinsicCandidate not found - continuing with a stub.
[error] Class jdk.internal.HotSpotIntrinsicCandidate not found - continuing with a stub.
[error] Class jdk.internal.HotSpotIntrinsicCandidate not found - continuing with a stub.
[error] Class jdk.internal.HotSpotIntrinsicCandidate not found - continuing with a stub.
[error] 5 errors found

[johanandren]

We have a nightly jenkins job on JDK 11 which compiles fine (https://jenkins.akka.io:8498/job/akka-nightly-jdk11/ ) so it is not a general compile-on-jdk-11-problem. So I guess it is something about the travis infra and finding JDK versions probably, because of the trix we do to compile the Flow APIs.

I don't think getting the travis job up is super important, even if it would be useful.

[SethTisue]

in the context of the community build, where we are building your v2.5.18 tag, I'm still seeing the "Class jdk.internal.HotSpotIntrinsicCandidate not found" errors. I'm still interested in resolving this somehow, since it's the top blocker (most downstream projects blocked) in the 2.12-on-JDK11 community build

[SethTisue]

I've been staring at this and fiddling with it all morning, to no avail... but then I finally had a hunch: maybe using -release 11 instead of -release 9 would fix it? trying it at #26078

as for what the cause might be, maybe it depends on whether the machine you're on has 9 installed in addition to 11? perhaps the Akka Jenkins has both 9 and 11 but Travis-CI has dropped 9 from the image we're using? (that wouldn't account for why I can reproduce the problem on my laptop inside dbuild but not outside, but there could be some separate cause for that.)

[SethTisue]

yeah so Travis went green over at #26078

that's good enough for me to get akka-stream green in the Scala 2.12-on-JDK11 community build, so I'd prefer to stop there and not to take this work any further myself. I suggest y'all follow through by dropping support for building on JDK 9 and 10 and make everything consistently 8 and 11

[patriknw]

I'm closing this because we have other jobs that are running full tests on JDK 11 and I don't see much value in adding build cycles to Travis build which doesn't do much. And we should use Adopt JDK as in #26170

[jrudolph]

We are now also seeing this but only on some machines.

I've been staring at this and fiddling with it all morning, to no avail... but then I finally had a hunch: maybe using -release 11 instead of -release 9 would fix it? trying it at #26078

Yes, that seems to work but we don't really know what the exact effect of changing that setting is. Is that explained somewhere, @SethTisue?

as for what the cause might be, maybe it depends on whether the machine you're on has 9 installed in addition to 11? perhaps the Akka Jenkins has both 9 and 11 but Travis-CI has dropped 9 from the image we're using? (that wouldn't account for why I can reproduce the problem on my laptop inside dbuild but not outside, but there could be some separate cause for that.)

We experimented with that and think that we can rule that out, as least it's not as simple as that.

[SethTisue]

Yes, that seems to work but we don't really know what the exact effect of changing that setting is. Is that explained somewhere, @SethTisue?

see scala/scala#6362

fwiw, I doubt it's worthwhile to get to the bottom of this, there's no reason to use -release 9 anymore that I can see when you could be using -release 11 instead; I think it's fine (and in fact, best) to pretend that JDK 9 and JDK 10 never existed, certainly that's what we're doing for Scala itself

[jrudolph]

Thanks, @SethTisue, I learned something there.

fwiw, I doubt it's worthwhile to get to the bottom of this, there's no reason to use -release 9 anymore that I can see when you could be using -release 11 instead; I think it's fine (and in fact, best) to pretend that JDK 9 and JDK 10 never existed, certainly that's what we're doing for Scala itself

Agreed.