checkerror that is used to check error cases and conditions in protocol.tests does not fail gracefully if it does not receive the expected codepoint. This can cause problems as I noticed when working on
DERBY-959. Make changes so that the checkerror case in case of failures will fail and then proceed to the next testcase.
This is a trivial fix, but I am opening a separate jira for this and not fixing this as part of
DERBY-959 since I would want this fix to be backported to 10.1 at some point..