[NETBEANS-7069] Support Nashorn 15.x for JDK >= 15

[cz6ace]

Netbeans does support debugging of legacy Nashorn JavaScript engine bundled with OpenJDK < 15.0. It does not support the the most recent Nashorn 15.4 extracted from OpenJDK, that has different package name.

Change allows debugging both legacy Nashorn JavaScript and also extracted Nashorn with new org.openjdk.nashorn package name.

See issue #7069

[matthiasblaesing]

Thank you for working on this. I left a first set of comments inline. The bad part is, that it does not work for me.

This is the project I used to test: mavenproject3.zip

Baseline is (more or less) current master. I configure the maven project to use JDK 11 for execution and comment out the Nashorn dependency. Then I place a breakpoint in the test.js file, line 5. Then I invoke "Debug" from the context menu of the project. The IDE starts the programm and the breakpoint in the JS code is hit. This looks like the expected behavior.

I then reconfigured the maven project to run with JDK 21 and reenabled the Nashorn dependency (only out-of-jdk Nashorn is run). With that baseline does not break anywhere, programm just runs through.

Side information I noticed while testing was, that when both out-of-jdk Nashorn and in-jdk Nashorn are present (dependency commented in and running on JDK 11), the out-of-jdk version if preferred by ScriptEngineManager.

With this PR and the same setup I get several hits in java code:

But none in JS. I would first look inside JSJavaBreakpointsManager.java. One thing to keep in mind in that class is, that it might be called multiple times. NetBeans supports multiple debugging sessions in parallel and then engineAdded is called multiple times, without calls to engineRemoved in between.

[cz6ace]

Hi @matthiasblaesing ,
thank you for comments, I'll check. To be honest, I tried both scenarios with external application and I simply attached via socket to the jvm. And the debugger stopped in JDK, but following "step" got me into the javascript.

I'll try your project.

And yes, problem will be most probably in the JSJavaBreakpointsManager, this is tricky part, because one has to create breakpoints in advance and the debugger instance not available, so I created two sets of breakpoints. I need also put some explanation for potential reader.

[cz6ace]

Hi @matthiasblaesing ,

even though I was able to debug the javascript on my side, I also had those weird breakpoints you mentioned in the screenshots - and in fact, they revealed that I brought some regression(s). I'll work on it before the vacation ends.

Libor

[cz6ace]

Hi @matthiasblaesing,

with the last commit I can debug in both cases, I tried

• OpenJDK 11 (commented the Nashorn in the pom.xml)
• OpenJDK 17 with enabled Nashorn in the pom.xml
  First, it stops in the javascript, second, it does not stop in the Nashorn sources (your screenshots)

If you could please check on your side again, it will be appreciated.

My results:
JDK 11:

JDK 17:

[matthiasblaesing]

Thank you for the update. I confirm, that it works for me and I think we should get this in. @cz6ace would you please squash this, so that I can invoke a final CI/CD run. The current test failure is caused by CI/CD picking up commit messages, that are obviously not final.

[cz6ace]

Good to hear it works on your side too. I did the squash and rebased to the latest master to be sure there's no surprise.
Please execute the CI/CD, thx!

[matthiasblaesing]

Looks sane to me. Lets get this in.

[matthiasblaesing]

@cz6ace thank you for looking into this.