Just a heads up that the past 3 days have been a bit of a problem for me to get the builds all done and ready for QE review.
I'd hoped to at least have something completed last night, but due to a change in the way Maven 3.5 handles dep resolution when building update sites, I've discovered a major flaw in the way we produce the aggregate update sites for JBT and devstudio.
The first manifestation of this was that the hibernate site was not publishing the new hibernate.bot.ui and hibernate.reddeer plugins, so the coretests site was not picking them up and the integration tests site was broken due to the missing dep.
So I fixed that, but then realized that we introduced a circular self-dependency in the coretests site by adding the new ITests profile [1] to the parent pom. This meant that when building the coretests site, the maven reactor would resolve against a previous coretests site build, rather than the local artifacts of the local build. That too has been fixed.
But then I discovered another problem w/ the way that the browsersim deployment was being done - the resulting update site contained all the latest plugins physically, but was NOT containing them via the p2 metadata; the result was that the JBT aggregate site was broken due to missing dependency: org.jboss.tools.cordovasim.eclipse 3.8.4.v20170530-1153 requires 'bundle org.jboss.tools.browsersim.eclipse 3.8.0
So, I've had to redo the way the browsersim standalone zip is built and deployed... that work is in progress but will hopefully allow me to resolve the missing deps and get a good JBT aggregate site build. And from there, we can rebuild devstudio installer & rpm, and finally get the bits staged later today.
Fingers crossed I don't find another problem as I head down this maze of twisty little passages, all alike [2].
--
Nick Boldt
Senior Software Engineer, RHCSA
Productization Lead :: JBoss Tools & Dev Studio
IM: @nickboldt / @nboldt / http://nick.divbyzero.com