Hi, I therefore cancel the vote for API 1.0.0-M25, there is a bug in the way we deal with the sort control.
Lucas finding is correct, it breaks server-integ. The reason is that the sort control grammar is incorrect : one transition (from orderingRule to reverseOrder ) is made mandatory, when teh reverseOrder default to False when missing. Adding an grammarEndAllowed in the associated action makes the test happy. I have fixed the issue in trunk, it's a bit late for me to cut a new release, will do that tomorrow. Many thanks Lucas, you have found a pb I missed because my test env is a mess. Bottom line, a few lessons for us : - always have a clean test env when cutting a release... (that's for me !) - there are some missing tests for the sortRequestControlTest class (typically, checking that a missing reverseOrder is not breaking the grammar) - I don't know how possibly the server-integ could have worked after the addition of the sort control... We should have detected the issue earlier. I'm quite sure that the commit http://svn.apache.org/r1630177 broke the tests. We have to be careful when we commit code, a full build is required (assuming of course the test env is correct ;-) Ok, guys, nothing really bad. Just a few lost hours... Thanks !
