Attaching another intermediate patch - this one is pretty much finished.
I did not add it into a suite yet (nor have I run the test in a suite).
I need to do that, and remove the original and its canons from the derbyclientmats.
I also think this test needs an exception clause against running with CDC profile.
I also think maybe we need to have this forced in English locale, as we're checking on the contents of derby.log with an expected English string.
I think I addressed many of Dag's comments.
However, I tried the test with ibm 1.7 and it did not run the fixtures in a different sequence, so I wonder how this version runs with jdk 1.7. I can probably find one later.
Also, re the port number - I've used getNextAvailablePort, which ups the number every time it is used. I don't know exactly why - even if I just run this test - the junit framework thinks it's up to number 1532 - up 5 from 1527 - I assume under the covers we're cranking up the number...
I added 2 numbers to the maximum in TestConfiguration, although if the test runs with -Dderby.tests.basePort, we're only using getNextAvailablePort once.
Patch ready for review.