Hi. I think you are probably referring to the master pom, at least the hudson Master-All build (
https://hudson.eclipse.org/papyrus-rt/job/Papyrus-RT-Master-All/configure) sets eclipse.release, although I am confused about that too, since I do not see any mention of the eclipse.release property in that pom or its immediate (sub)modules, and I don't see where to other poms (core, common, profile, tooling) are included in this. What am I missing here?
Anyway, regardless, since Celine's changes have not been merged to the master branch yet, we are still using our own separate codegen+rts pom and build, and in this one, the profile activation is on the PAPYRUS_UPDATE_SITE property, not on eclipse.release. You can see that in releng/codegen/pom.xml. This is where we define the eclipse.targetrelease property (which I think Celine added when she also added the codegen-specific target platforms) and the tycho configuration I mentioned. You are right that we need to unify these properties, but that was what Celine's unified build was all about, no?
Nevertheless, it seems to be working now, but it is ugly, because I had to add the following to the release target platform (in fact to all TPs):
I've sent an e-mail to Remi and Quentin who is supposed to work on ELK integration, but Remi is now on vacations and Quentin has not answered. So for now it will have to do.
But I'm still bewildered by the error I described in my original message. It doesn't make sense to me.
At least the CodegenRTS build works now, but I'm mystified with the fact that the codegen nightly build has been failing. Even more strange, I just triggered it manually and it succeeded. So there seems to be a problem building automatically at that particular time (11:30pm). Very weird. The console output suggests that the problem occurs when we run our tests and try to install the features.