Use 2023-09 simrel

[akurtakov]

As default repo in plugins that have such and in tests.

[github-actions]

Test Results

   561 files  ±0     561 suites  ±0   4h 11m 13s ⏱️ - 24m 57s
   363 tests ±0     356 ✔️  - 1    6 💤 ±0  0 ❌ ±0  1 🔥 +1 
1 089 runs  ±0  1 067 ✔️  - 3  19 💤 ±0  0 ❌ ±0  3 🔥 +3 

For more details on these errors, see this check.

Results for commit a5c1f3c. ± Comparison against base commit 03e790c.

♻️ This comment has been updated with latest results.

[laeubi]

@akurtakov what do you think about moving this to TychoConstants as it is also used in several places in Tycho (no test) code?

[laeubi]

do you know hy we have the four times / here? it seems to not harm but looks strange...

[akurtakov]

I would guess 2 // for the protocol and 2 more to escape them somewhere in the code.

[laeubi]

This looks suspicious... if we need escaping it should not be done in the constant itself ...

[akurtakov]

@laeubi Is there a chance you can help with the last failing test? I'm kind of lost on this one.

[laeubi]

I'll try to take a look tomorrow! This test is a bit special because it uses not the junit bundles from the target but the maven ones, as the test complains about missing junit maybe we need to explicitly include junit somehow explicitly, from the changes made this might previously be pulled in by hamcrest or something.

[laeubi]

By the way if I wanted to debug this I probably would enable the display of the bundles for this test project and see if anything looks suspicious (e.g. juni4 really missing or pointing not to the local maven repository).

[akurtakov]

The test itself is JUnit 5 so it should work just fine without junit.jar (aka 4.x), no?

[laeubi]

The test complains that the vintage engine requires junit4 ... if I rember correctly JDT defines JUNIT5 = JUNIT5 Vintage... maybe it would be great to have a pure JUNIT5 Container without any vintage/junit4...

[akurtakov]

I've tried by removing vintage engine from the junit5 bundles and the test works fine so what happens is that vintage engine fails and that way it even blocks other engines from looking up for tests.

[laeubi]

I have now created

• [junit] provide a pure JUnit5 classpath container eclipse-jdt/eclipse.jdt.ui#837

So we need to somehow convince Tycho to include junit4 when vintage engine is included, I'm just a bit confused why it is not select anymore automatically...

[laeubi]

I now extracted the easy part here:

• Introduce a public static constant for the current eclipse release #2862

[akurtakov]

@laeubi Do you agree with using

<configuration>
  <excludeJUnit5Engines>
    <engine>junit-vintage</engine>
  </excludeJUnit5Engines>
</configuration>

to workaround the issue for now?

[laeubi]

I have now tested it and at first everything seems okay but then surefire fails, I can fix the test by adding

<dependency>
		  <groupId>org.apache.servicemix.bundles</groupId>
		  <artifactId>org.apache.servicemix.bundles.junit</artifactId>
		  <version>4.13.2_1</version>
		  <scope>test</scope>
</dependency>

as an explicit dependency ... so it seems that something goes wrong maybe when injecting the dependencies even though I can't see how/why this happens need to debug a bit further.

[laeubi]

The junit dependency is currently injected as

 <dependency>
      <groupId>p2.eclipse.plugin</groupId>
      <artifactId>org.apache.servicemix.bundles.junit</artifactId>
      <version>4.13.2.1</version>
      <type>eclipse-plugin</type>
      <scope>system</scope>
      <systemPath>/tmp/file not yet available5847474459691235710.tmp</systemPath>
    </dependency>

so for some reason p2 things that the file is not (yet) there and then surefire can not use it of course ... will try to provide a fix for that...

[laeubi]

This should fix the issue we see here:

• Return the maven location for missing junit bundles #2867

[akurtakov]

Jenkins succeeded. Merging.