Jeremy Boynes wrote: > Rick Hillegas wrote: > >> I would like to propose that we include JUnit as part of the required >> toolkit for Derby developers. Going forward, this will allow >> developers to write assertion-based tests. For the moment, this will >> not change the existing test harness. >> > > +1 with the caveat that it works in all the environments we need (and I > don't see yet why it won't). JUnit is a well understood harness used by > many projects (and support is actually built into Maven).
+1 [to the original proposal] I'm not so sure that Jeremy's caveat is valid, though I agree it is strongly desirable. The big question is really who is 'we' in the 'all environments we need'? I need to run Derby on IBM's z-series, but my guess is that Rick doesn't. I don't think any Derby patch/commit should require running tests on all platforms Derby has ever run on, especially as most people will not have access to them. I see it as Rick (and others) are providing new tests for Derby, that's a good thing, even if, as they provided them, they only worked on (say) Linux and Solaris. It seems better to have the ability to run these tests to improve/maintain the quality of Derby than not to have them. If I care about running the new tests on z-series, then that's my itch to scratch. I think the caveat is really that any such checkin should not cause a regression, namely derbyall does run on a number of platforms with its current set of tests. It would be a regression if a new testing scheme stopped my current ability to run the *existing* tests on z-series or any other platform where the tests run today. Of course it would be good to document which platforms Derby is running tests sucessfully on, seems like a good wiki page. Dan.
