[EMAIL PROTECTED] wrote: >Hmm, yes I see your point, but derbyall already takes a long time, >doesn't it? > > > I think efforts to reduce the time derbyall takes should be focused on things that just make it run faster but not ever discourage addition of tests. For example running with relaxed durability ( -Dderby.system.durability=test) speeds things up quite a bit. Efforts to get multiple tests running against a single db seem like a good idea. Adding tests to existing tests instead of making new ones can be a way to economize too. Also ultimately if it got too long we could have a high coverage subset that would be typically run for checkin instead of derbyAll but that is more risky, but I tend to think that it is a really good idea to have a test for every bug fix where reasonable. Creating database objects when the default schema doesn't exist is an area where we could definitely use more testing as evidenced by this bug, DERBY-230 and I think others.
Kathey
