4.36 I-Build: I20250327-1800 - BUILD FAILED

[iloveeclipse]

https://ci.eclipse.org/releng/job/Builds/job/I-build-4.36/31/console

23:07:28  [INFO] Scanning for projects...
23:07:37  [ERROR] [ERROR] Some problems were encountered while processing the POMs:
23:07:37  [FATAL] Non-resolvable parent POM for org.eclipse.platform:org.eclipse.equinox.security.ui:1.4.500-SNAPSHOT: The following artifacts could not be resolved: org.eclipse.platform:bundles:pom:4.36.0-SNAPSHOT (absent): Could not find artifact org.eclipse.platform:bundles:pom:4.36.0-SNAPSHOT and 'parent.relativePath' points at wrong local POM @ 
23:07:37  [FATAL] Non-resolvable parent POM for org.eclipse.platform:org.eclipse.equinox.bidi.tests:1.4.300-SNAPSHOT: The following artifacts could not be resolved: org.eclipse.platform:bundles:pom:4.36.0-SNAPSHOT (absent): Could not find artifact org.eclipse.platform:bundles:pom:4.36.0-SNAPSHOT and 'parent.relativePath' points at wrong local POM @ 
23:07:37  [FATAL] Non-resolvable parent POM for org.eclipse.platform:org.eclipse.equinox.cm.test:1.2.0-SNAPSHOT: The following artifacts could not be resolved: org.eclipse.platform:bundles:pom:4.36.0-SNAPSHOT (absent): Could not find artifact org.eclipse.platform:bundles:pom:4.36.0-SNAPSHOT and 'parent.relativePath' points at wrong local POM @ 

[iloveeclipse]

@laeubi : I don't see any change in pom files neither in equinox nor in the platform or aggregator responsible for the build failure, could it be some tycho snapshot issue?

[akurtakov]

#2952 points that it's not Tycho regression.

[merks]

Looking at what has changed in all the repositories in the last day, these are the only things that look suspiciously related:

eclipse-platform/eclipse.platform.swt@91e2c7b
eclipse-platform/eclipse.platform.swt@ea7af55

[akurtakov]

It's definetely the SWT change. Resetting swt repo to ea7af55c9d61947f6bdaf79c1e4ee3697378bd60 unbreaks the build locally. My guess is that we end up with two different "pomless" modules with same gid/aid with these changes.
@HannesWell ^^ fyi.

[HeikoKlare]

I had to revert exactly the two commits mentioned by Ed to make the aggregator build run locally again.

@HannesWell do you see an easy and quick fix for this or should we revert the changes and reapply adapted ones once they are ready? Since we are approaching M1 and some platform PRs are waiting for an I-Build with recent SWT changes, it would just be nice if we can avoid having no I-Builds for several days 🙂

[HannesWell]

It's definetely the SWT change. Resetting swt repo to ea7af55c9d61947f6bdaf79c1e4ee3697378bd60 unbreaks the build locally. My guess is that we end up with two different "pomless" modules with same gid/aid with these changes.

Yes that's exactly the problem. Since eclipse-platform/eclipse.platform.swt#1957, we have to modules with id org.eclipse.platform:bundles one from SWT and the other one from P2.
I just created a change to individualize the name for SWT:

• Ensure unique artifactIds for pomless aggregation modules eclipse.platform.swt#1963

But for P2 I think we should do the same, just to be on the save side.