Well to be fair, if they tried to stay ahead of the curve on all these new JDK releases they'd do nothing else ;) Anyway, glad you worked it out. On Wed, Jun 13, 2018 at 12:03 PM Sanne Grinovero <sanne(a)hibernate.org&gt; wrote:

On Sat, 16 Jun 2018 at 08:08, Emmanuel Bernard <emmanuel(a)hibernate.org&gt; wrote: No crystal ball, but so far we had quite some work to do for each JDK release; I thought I had highlighted that to you. Certainly Jigsaw had a big impact when it came in 9, but that work is far from over as barely any build tool / IDE / dependency management has caught up yet: stuff "boots" today but there's a long list of missing stuff. We're certainly far from getting benefits from it. Then there's all the tangential chaos: in 9 we had issues with the EE modules, such as the XML parsers; for that one the widely agreed on solution was to use options like "--add-modules java.xml.bind", yet now testing JDK11 that same module has been removed, so we need to re-think that one. Another example is Javadoc linting: options we need have been deprecated in 9, removed in 10. With this faster pace affecting deprecation + removal cycles one can't really be sure of what's going to be broken within a year. Doclet APIs are gone as well, some more JDK options and flags we used now fail to start, and some such options are hardcoded in e.g. Maven or Gradle extensions we used for other purposes. The issues are compounding each other: since build tools are barely able to keep up, some of our dependencies just don't keep up as they can't easily build / test; hitting issues first is quite time consuming and when reporting issues other communities aren't particularly eager to fix those. Not least sometimes we need to apply or experiment with JDK new flags or solutions which are presented as the appropriate way forward - like multi-release JARs - but build tools don't support them yet so you need complex build workarounds.. which spread issues to IDEs, CI, .. all needing updates, all breaking someone's workflow. In short, so far it has been breaking stuff at increased pace. I don't see it as too negative though, it's looking like overall speed is faster in all areas so maybe "speed to break" is proprortionate to overall speed increase - I hope thought that other communities we rely on can adjust to this too. That includes ourselves, I needed another HCANN release last weekend, some more metadata changes in ORM... list goes on and we haven't touched much on actually benefitting from the new version features. Thanks, Sanne