Missing code implementation in the compiler

[laeubi]

I moved some code from one plugin to the other (but still require the new bundle with re-export) and now I get javadoc errors in JDT:

Missing code implementation in the compiler

Eclipse SDK

Version: 2024-12 (4.34)
Build id: I20241120-1800

[jukzi]

Minimal reproducer required. The Message is from org.eclipse.jdt.internal.compiler.problem.ProblemReporter.abortDueToInternalError(String, ASTNode) which is supposed to also give details about source location. Maybe that error handling needs to be improved for your case too.

[laeubi]

which is supposed to also give details about source location. Maybe that error handling needs to be improved for your case too.

The message is all I get, no further details or error logs.

[laeubi]

Minimal reproducer required.

I only can give the hint from the inital report, e.g. I have a javac reference in Bundle A

	/**
	 * See doc on {@link FileLocator#find(Bundle, IPath, Map)}
	 */

now I move the FileLocator to another required bundle then the error shows up.

[srikanth-sankaran]

Jay, can you take a look ? Thanks!

[jarthana]

Minimal reproducer required.

I only can give the hint from the inital report, e.g. I have a javac reference in Bundle A

You meant Javadoc reference, right?

[laeubi]

Yes javadoc of course :-)

[jarthana]

OK, I tried this within JDT. That is, moved some stuff from jdt.core (referenced via Javadoc in the same bundle) to compiler.batch, which is a required bundle for the former. Don't see the problem. There has to be something special about reported case.

[stephan-herrmann]

Minimal reproducer required. The Message is from org.eclipse.jdt.internal.compiler.problem.ProblemReporter.abortDueToInternalError(String, ASTNode) which is supposed to also give details about source location. Maybe that error handling needs to be improved for your case too.

More specifically: org.eclipse.jdt.internal.compiler.problem.ProblemReporter.needImplementation(ASTNode).

@jarthana when comparing invalidMethod() with javadocInvalidMethod() I could imagine we're hitting a case of ProblemReasons.NotAccessible which would be handled in the former method, but is not in the latter. I see some more problem reasons handled only in invalidMethod() but those look less relevant for javadoc, but ideally both methods would handle the same set of reasons?

[jarthana]

@stephan-herrmann Thanks, sounds plausible. But I am having a hard time coming up with a testcase, though. Any chance you already have one?