Ok, the test issue was an unrelated brain-fart

@Scott Marlow I ran the hibernate-core testsuite using both Javassist 3.18.2-GA and 3.20.0-GA. Both were fine in our tests. So as far as WildFly and what y'all want to use there, let me know and we'll sync up.

@Scott Marlow The reason I ask is that changing just the Hikari version as requested leads to problems running even the verrrrry simple test in hibernate-hikari. Its one of those strange cases though where I get the failure from the command line, but cannot reproduce from within IDE to debug it.

This is is slated for 5.1 which is the next major version, so I am ok with updating some dependency versions. I just don't want to end up running into the same issues we invariably seem to run into when upgrading Javassist. In fact I wonder if we should consider (moving forward) doing a shadow/shade of Javassist for Hibernate's use.

For the short term, I'll try updating Hikari (to 2.3.9) + Javassist (to both 3.18.2-GA and 3.20.0-GA) and see how that works.

WildFly 10 is also using javassist 3.18.1-GA. I expect that 3.18.2-GA will work for us and it has a little more Java 8 support.

We should also talk soon about javassist 3.20.x versus 3.18.3-GA. If we choose to use 3.18.3-GA, we need to do some work to create it. If 3.20.0-GA is just as compatible, we could consider using that (or even 3.21.0-GA which isn't out yet.)

Ugh. So in trying to set up this upgrade I notice they have upgraded their Javassist and that it no longer matches what we use in Hibernate 3.18.1 versus 3.18.2). I know that seems like it should be minor, but we have had major headaches in the past with even minor Javassist version changes.

@Scott Marlow Any thoughts on this one?

For 5.x I think we will need to stick with Hakari 2.3.9 (Java 6 compatibility)