2014-10-29 19:14 GMT+01:00 Mark Thomas <ma...@apache.org>: > > Hmm. I withdraw that veto. This change has no impact on the failing > test. I need to investigate further. (It is odd though how we are now > seeing a ServletException where there should be an ISE). > > If there are tests which contradict these fixes, they'll need to be fixed IMO. I haven't run the TC testsuite yet with them, sorry.
Two of the changes should be rather logical though: - getResponse should ISE after complete, but the listener onComplete call should be part of that process - the started flag being still true for MUST_COMPLETE is rather gratuitous, but ... - the javadoc says Throws: ServletException - if the given clazz fails to be instantiated (so most exceptions should be wrapped) Could 15 wait for next week ? I could have more edgy fixes on the way. Rémy