Jørgen Løland a écrit, Le 23.06.2010 10:02:
> Guilhem Bichot wrote:
>> This commit is for next-mr-opt-backporting. Is it that this bug cannot
>> happen in next-mr-opt-team, because it requires materialization?
> The bug needs materialization turned off, and since subquery_mat_*.test
> has been rearranged into an .inc file included by _none.test and friends
> I figured we would save ourself some merge conflicts by using this tree.
Oystein has done this same rearrangement in next-mr-opt-team (where all
features are disabled): both trees have the same testcases, only certain
result files are different (but Oystein tried to minimize them). So IF
the bug has a chance to show up in next-mr-opt-team, it should be fixed
there, I suggest.
Mr. Guilhem Bichot <guilhem@stripped>
Sun Microsystems / MySQL / Optimizer team, Lead Software Engineer
www.sun.com / www.mysql.com