Tim Ward

I think the direction here is good. We probably need some more multi-release bundle generation tests before it can be merged though. One feature that seems likely to have issues...

> hi @reckart sorry for the omission in the example, I'll try to make it current. > > But yeah, it would be nice to improve the error messages. I've...

At this point I'm happy and think this is worth merging - @bjhargrave do you have anything to add?

> If I understand the code correctly, this is simply `EE.getCapabilityVersion().getMajor()` Yes, although the following isn't exactly readable... EE myEE = getResolveTargetEE(); manifest = jar.getManifest(myEE.getCapabilityVersion().getMajor()).get();

The following comment is wrong. The version should be 17 for the third check in the test. https://github.com/bndtools/bnd/compare/2d1a4c6ba66197105067b44814b5c4033f6f7a0d..9856c6cbda183da1a08ee1a8fd1dbaa77a76e090#diff-5ba9b2d89cbd966cfd1655fb0c9488599e851bc6ae563e20bfd63c9df4e0d93aR152

@bjhargrave - I think that this should be merged, can we go ahead?

With luck the Surefire team will have this fixed soon. I think it's affecting a lot of people!

> Why not prepopulate -runbundles in the initial archetype output? What would we populate it with? In the majority of cases the bundle in `-runrequires` will have dependencies and we...

> The bundle will not have any dependencies immediately after generation from the archetype. Even if the user starts adding basic "hello world" code to the bundle, they will not...

>I don't think there's a difference. The build will fail at the resolve stage, whether `-runbundles` is empty or non-empty-with-incorrect-content. There is a difference if you're in Eclipse. An empty...