[tycho-4.0.x] Adopt new o.e.e.p2.publisher.eclipse API for IMacOsBundleUrlType

[sratz]

Backport of #4246 + #4248

[laeubi]

@sratz I think we last release skipped this update due to some incompatibility for Tycho 4, so one needs probably other updates/bugfixes as well. I usually try to compare some things from main where the update was performed to see if something is missing or can be adapted.

[sratz]

@sratz I think we last release skipped this update due to some incompatibility for Tycho 4, so one needs probably other updates/bugfixes as well. I usually try to compare some things from main where the update was performed to see if something is missing or can be adapted.

I am wondering where to get started, though.

I don't have any idea what to make of the generic

org.apache.maven.it.VerificationException: Error in execution: [ERROR] [c8a6beaa-0a0d-4839-9687-0680bcd338fc][extension>org.eclipse.tycho:tycho-maven-plugin:4.0.9-SNAPSHOT] org.eclipse.equinox.p2.publisher.eclipse
at org.apache.maven.it.Verifier.verifyErrorFreeLog(Verifier.java:312)

error message that all tests are failing with :(

[laeubi]

Yes that's a bit annoying as it hides the exception.

The easiest in such case is to do:

• mvn clean install -T1C -DskipTests
• cd tycho-its
• mvn clean install -Pits -Dtest=<fqdn of test>

That way you can run one test and hopefully have access to everything (e.g. logs).

In this particular case, one can maybe better run

• cd tycho-its/projects/reactor.makeBehaviour
• mvn clean install -Dtycho-version=4.xx

to see the real error that is printed.

[github-actions]

Test Results

  597 files  ±0    597 suites  ±0   4h 14m 34s ⏱️ + 8m 30s
  418 tests ±0    411 ✅ +3   6 💤 ±0  0 ❌  - 2  1 🔥  - 1 
1 254 runs  ±0  1 234 ✅ +3  19 💤 ±0  0 ❌  - 2  1 🔥  - 1 

For more details on these errors, see this check.

Results for commit 01ec999. ± Comparison against base commit b30ba24.

♻️ This comment has been updated with latest results.

[sratz]

I updated all the P2 artifacts and their bouncycastle deps to be the same as in the main branch.

Looks pretty good and the one test failure on macOS seems unrelated.

Any concerns with this change?

[laeubi]

This seem to happen sometimes and as it is only for one OS I think we would be fine with that. You might want to rebase to be totally sure or just merge and see what happens :-)