We now have started to put the TCK infrastructure in place:
* TCK porting package APIs to allow the TCK to request the RI do stuff
like create a SimpleBean from a class
* implementation of these APIs for RI
* Start TCK run and view output (testng format)
We're now running the TCK (all 5 tests which have been ported so far)
as part of WB testsuite - currently a single WB RI test is responsible
for kicking off the TCK run. If the TCK run fails, then that test will
fail. You won't see a merged testrun total on when you run mvn test,
however Hudson is correctly reporting the merged totals.
This perhaps isn't the ideal situation, and I'm thinking about how to
improve it (perhaps run the TCK as another module, rather than as part
of the RI testsuite, and keep that for RI unit tests).
David and Shane have also started work on porting the spec tests from
the RI unit test suite to the TCK.