Ok so for this test classloader check is the one I think. Avoid to use luck (I'd prefer an exception we found.size() == 2 but it would break your test).
For memories all tests using size (of ejbs, cdi beans etc...) to select an instance are just broken and breakable easily. We used it in OWB singleton just to not rewrite all our openejb-core tests but it is known as broken. I'll try to fix it tomorrow, since there is a test now and the fix is quite trivial it should be fast. Romain Manni-Bucau Twitter: @rmannibucau Blog: http://rmannibucau.wordpress.com/ LinkedIn: http://fr.linkedin.com/in/rmannibucau Github: https://github.com/rmannibucau 2014-09-23 20:29 GMT+02:00 Andy Gumbrecht <[email protected]>: > If there is something wrong with the test itself, and I really can't see > that there is right now, then we need to at least warn in the logs and offer > a solution/recommendation. > It was after all something that has only become visible through Java8 - I > wouldn't make them not deployable just for the sake of the test, it's what > highlights the issue. > > Andy. > > On 23/09/2014 19:59, Romain Manni-Bucau wrote: >> >> test enrich all apps? Shouldn't you declare some of them as not >> deployable? >> > > -- > Andy Gumbrecht > https://twitter.com/AndyGeeDe > http://www.tomitribe.com >
