Test with jdk 24 and 25 ea - #3949
Conversation
|
We'll keep this PR around and monitor the inferred spans support closer to release. |
JonasKunz
left a comment
There was a problem hiding this comment.
This PR has shown some test behaviour which has certainly been amongst the most confusing things I've encountered during my tenure at elastic so far. But hey, tests are finally green!
| <version.error_prone>2.2.0</version.error_prone> | ||
| <version.h2>1.4.196</version.h2> | ||
| <version.junit>5.9.1</version.junit> | ||
| <version.junit>5.12.2</version.junit> |
There was a problem hiding this comment.
JUnit had to be updated to include the newer java versions in @DisabledForJRE
|
|
||
|
|
||
| static synchronized void initAllButInstrumentation() { | ||
| ElasticApmAgent.reset(); // reset the AbstractInstrumentationTest beforeAll initialization |
There was a problem hiding this comment.
The JUnit upgrade made the lambda tests fail.
They apparently have been relieing on a bug where static @BeforeAll methods from subclasses would replace static @BeforeAll in parent methods. Now they both execute, causing the tracer to be initialized twice.
|
we lose inferred spans after 23? |
Yes, due to async-profiler 1.x not working correctly there. We will need to upgrade it to support it. |
What does this PR do?
Tests with Java 24 ea to ensure that the blocked bytebuddy-bump #3946 doesn't bite us.
Once we get notified by the
JdkVersionTestabout the release of Java 24, we can simply remove the-easuffix and drop java 23 from the list.